Estimating chamber component conditions using substrate measurements

Abstract

A method includes processing a substrate in a process chamber according to a recipe, the substrate comprising at least one of a film or a feature after processing. The method further includes generating a profile map of the first substrate. The method further includes processing data from the profile map using a first model, the first model outputting at least one of an estimated mesa condition of a substrate support for the process chamber, an estimated lift pin location condition of the substrate support, an estimated seal band condition of the substrate support, or an estimated process kit ring condition for a process kit ring for the process chamber. The method further includes outputting a notification as a result of the processing.

Description

[Technical field]

[0001] Embodiments of the present disclosure relate generally to detecting the condition of chamber parts in a processing chamber, and in particular to generating maps and / or numerical profiling of a substrate processed using the chamber and estimating the condition of one or more chamber parts of the chamber based on the maps and / or numerical profiling of the substrate. [Background technology]

[0002] Substrate processing can include a series of processes that fabricate electrical circuits in semiconductors according to a circuit design. These processes can be performed in a series of process chambers. Successful operation of modern semiconductor manufacturing facilities can be aimed at facilitating the movement of a steady flow of wafers from one chamber to another in the process of forming electrical circuits in the wafers to form products. In the process of performing many substrate processes, the conditions of the processing chambers can be changed, which can result in the processed substrates failing to meet the target conditions and results.

[0003] A process chamber includes many chamber parts that over time can wear and / or become damaged, resulting in a deterioration in the quality of products fabricated using the process chamber and / or the scrapping of substrates processed using the process chamber.

[0004] In addition, the new components may not be compliant with the specification. Summary of the Invention

[0005] 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 embodiments 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 presented later.

[0006] In an exemplary embodiment, a method includes processing a first substrate in a process chamber of a substrate processing system according to a recipe, the first substrate comprising at least one of a film or a feature after processing. The method further includes generating a profile map of at least one of the film or the feature on the first substrate using a substrate measurement system of the substrate processing system. The method further includes processing data from the profile map using a first model, the first model outputting at least one of an estimated mesa condition of the substrate support for the process chamber, an estimated lift pin location condition of the substrate support, an estimated seal band condition of the substrate support, or an estimated process kit ring condition for the process chamber. The method further includes outputting a notification of at least one of the estimated mesa condition of the substrate support, the estimated lift pin location condition, the estimated process kit ring condition, or the estimated seal band condition as a result of the processing.

[0007] In an exemplary embodiment, a computer-readable medium includes instructions that, when executed by a processing device, cause the processing device to perform operations. The operations include receiving a profile map of at least one of a film or feature on a first substrate measured using a substrate measurement system after the first substrate is processed in a process chamber. The operations further include processing data from the profile map using a first trained machine learning model, the first trained machine learning model outputting at least one of an estimated mesa condition of a substrate support for the process chamber, an estimated lift pin location condition of the substrate support, an estimated seal band condition of the substrate support, or an estimated process kit ring condition for a process kit ring for the process chamber. The operations further include outputting a notification of at least one of the estimated mesa condition of the substrate support, the estimated lift pin location condition, the estimated process kit ring condition, or the estimated seal band condition as a result of the processing.

[0008] In an exemplary embodiment, a computer-readable medium includes instructions that, when executed by a processing device, cause the processing device to perform operations. The operations include receiving a first profile map of at least one of a film or feature on a first substrate, the first profile map being generated by a substrate measurement system after the first substrate is processed by a process chamber according to the recipe for a first time. The operations further include receiving a second profile map of at least one of a film or feature on a second substrate, the second profile map being generated by a substrate measurement system after the second substrate is processed by a process chamber according to the recipe for a second time. The operations further include comparing the first profile map and the second profile map. The operations further include determining, based on a result of the comparison, at least one of a) an estimated mesa condition for one or more mesas of a substrate support in the process chamber, b) an estimated lift pin location condition for one or more lift pins of the substrate support, or c) an estimated seal band condition for a seal band of the substrate support.

[0009] The present disclosure is illustrated by way of example, and not by way of limitation, in the figures of the accompanying drawings in which like references indicate similar elements. It should be noted that various references in this disclosure to "an embodiment" or "one embodiment" do not necessarily refer to the same embodiment, but such references do mean at least one. [Brief description of the drawings]

[0010] [Figure 1] FIG. 1 illustrates an exemplary computer system architecture in accordance with aspects of the present disclosure. [Figure 2A] FIG. 1 illustrates a schematic top view of an exemplary manufacturing system according to aspects of the present disclosure. [Figure 2B] 2B is a schematic cross-sectional side view of a substrate measurement system included within the example manufacturing system of FIG. 2A in accordance with an embodiment of the present disclosure. [Figure 2C]1 is a schematic cross-sectional side view of a substrate measurement subsystem according to an aspect of the present disclosure. [Diagram 3] FIG. 1 illustrates an exemplary system architecture for chamber part condition prediction of a processing chamber in accordance with aspects of the present disclosure. [Figure 4] FIG. 1 illustrates a model training workflow and a model application workflow for chamber component condition determination in accordance with one embodiment. [Diagram 5] 1 is a flowchart of a method for generating a training dataset for training a machine learning model according to an aspect of the present disclosure. [Figure 6] 1 illustrates a flow diagram of a method for training a machine learning model to determine a condition of a chamber component according to an embodiment. [Figure 7] 1 illustrates a flow diagram of a method for determining chamber conditions of one or more chamber parts using one or more trained machine learning models, according to an aspect of the present disclosure. [Figure 8] 1 illustrates a flow diagram of a method for determining chamber conditions of one or more chamber parts using one or more trained machine learning models, according to an aspect of the present disclosure. [Figure 9] 1 is a flow diagram of a method for determining chamber conditions of one or more chamber parts based on measurements of substrates processed by a process chamber over time, according to an aspect of the present disclosure. [Figure 10] 1 is a flow diagram of a method for determining a chamber condition of one or more chamber parts based on a comparison of data and a curve, according to an aspect of the present disclosure. [Figure 11A] 1 is a flow diagram of a method for performing a quality assessment of a process chamber according to an aspect of the present disclosure. [Figure 11B] 1 is a flow diagram of a method for determining how a chamber part was installed in a process chamber, according to an aspect of the present disclosure. [Figure 12] 1 is a diagram of a profile map (eg, heat map) of a processed substrate according to an embodiment of the present disclosure. [Figure 13]1 is a profile map (eg, heat map) of a portion of a processed substrate according to an embodiment of the present disclosure. [Figure 14] 1 is a diagrammatic representation of a machine, an exemplary form of a computing device, capable of executing a set of instructions to cause the machine to perform any one or more of the techniques discussed herein. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0011] Embodiments of the present disclosure are directed to systems and methods for evaluating the condition of one or more chamber parts in a process chamber, such as a process chamber for semiconductor processing. The process results of a manufacturing process depend on many factors, including the process recipe, tool parameter settings (including, for example, chamber parameter settings and settings for parameters of one or more tools external to the chamber), and chamber part conditions. For example, the process results may vary across the surface of the substrate based on the condition of the showerhead, the condition of the substrate support supporting the substrate, the condition of the chamber liner, the condition of pumps and / or valves, etc. Any change over time in the substrate support, such as, for example, an electrostatic chuck, clamps, vacuum chucks, heaters, supports including pockets with lips at the edge of the support, and / or substrate supports including one or more embedded features (e.g., heaters, cold plates, electrical elements, etc.), slowly impacts the substrate results, such as by temperature changes across the surface of the processed substrate. Changes in the substrate support (e.g., chuck) may additionally or alternatively negatively impact the chucking and / or dechucking of the substrate, lift pin regions of the substrate, seal band regions around the periphery of the substrate, etc. The embodiments provide a new mechanism for evaluating the condition of chamber parts. The embodiments can be used, for example, to indicate slow aging or degradation of the substrate support and / or other chamber parts. The embodiments can be used to determine whether a chamber part, such as a process kit ring, was installed correctly and / or how the chamber part was installed (e.g., the placement of the chamber part) after a maintenance event or upon new installation. The embodiments can be used to perform a quality evaluation of the process chamber. Information from the evaluation of the chamber parts can be used to determine whether to adjust one or more settings for the process chamber, whether to adjust the process recipe, whether and / or when to perform maintenance on the process chamber (e.g., by repairing and / or replacing one or more chamber parts of the process chamber), etc.

[0012] In an embodiment, in a process chamber, a process is performed to deposit or etch a film, layer, and / or one or more features (e.g., measurable features) on a surface of a substrate. The substrate may be a bare substrate or a test substrate that does not include a product. Alternatively, the substrate may be a product substrate. The film may be a blanket film deposited on a surface of the substrate. The film may be a stack of layers, which may be patterned or unpatterned. In some embodiments, the film includes a stack of patterned layers having features such as trenches, structures, devices, etc. Such features may include or have critical dimensions that can be measured in an embodiment (e.g., width of a trench at one or more depths of the trench). The measured features may include features distributed across the surface of the substrate. The features may include, for example, mesas, dots, structures, valleys, walls, lines, grooves, fiducials, critical dimensions, etc. In an embodiment, the film is a polymer film and / or the features are polymer features.

[0013] After the film and / or feature is deposited or etched on the substrate, the substrate can be removed from the process chamber and placed into a substrate measurement system. The substrate measurement system can be, for example, a reflected light measurement system or other measurement system that measures the thickness of the film and / or feature at multiple locations on the substrate. The thickness and / or feature information can be used to generate a thickness profile map for the substrate. Alternatively, one or more other profile maps of the substrate (e.g., of optical constants, roughness, grain count, etc.) can be generated from other measurement data. In one example, where the substrate includes a stack of patterned layers, the profile map can include a critical dimension profile map. Models (e.g., trained machine learning models, physics-based models, statistical models, etc.) and / or image processing can then be used to process the film and / or feature thickness information (e.g., thickness profile maps), or other film and / or feature information (e.g., other profile maps, such as optical constant profile maps, particle number profile maps, critical dimension profile maps, etc.) to identify variations in one or more film properties and determine information about one or more chamber parts of the process chamber based on the identified property variations (e.g., film thickness variations, critical dimension variations, etc.). Embodiments are discussed with reference to measuring a film on a substrate. However, it should be understood that the embodiments also apply to measuring one or more features (e.g., critical dimensions) on a substrate in addition to or instead of measuring a film on a substrate.

[0014] When measuring product substrates (e.g., product wafers) to determine the profile map of the substrate, measurements of the product substrates can be performed frequently because the process chamber uptime (the time the process chamber is used to process product) is not affected by frequent measurements. Alternatively, when processing and measuring test substrates to determine the profile map of the test substrate, measurements may be performed less frequently because product is generally not being processed during the processing of the test substrate, resulting in a reduced chamber uptime.

[0015] In one example, the substrate support surface includes a number of features corresponding to chucking and dechucking (or clamping and unclamping) of the substrate, temperature control, backside helium control, and lift pin control. The substrate support features can include, for example, a series of mesas and / or channels on the top surface of the substrate support, which provide areas that can be filled with a thermally conductive gas, such as helium, during processing to facilitate thermal contact between the supported substrate and the substrate support while supporting the substrate. The substrate support features can further include a seal band on the periphery of the top surface of the substrate support, which seals the thermally conductive gas in the region between the bottom surface of the substrate and the top surface of the substrate support. The substrate support features can further include lift pin holes that accommodate lift pins that lift the substrate onto and off the substrate support, and a seal band around the lift pin holes. The substrate support features can further include a dielectric layer across the surface of the substrate support. The features of the substrate support may further include bonds between parts of the substrate support (e.g., between a ceramic puck and a cold plate, or between ceramic plates of an electrostatic chuck). Due to the use of the substrate support in the process chamber, one or more of the surface features of the substrate support may begin to wear, may break, or may fail. In addition, bonds may also fail, causing delamination between layers or bodies of the substrate support. In another example, recesses or valleys in the surface of the substrate support, planarity of the surface of the substrate support, roughness across the surface of the substrate support, concentricity of one or more circular features of the substrate support, etc. may affect a substrate being processed while supported by the substrate support. The embodiments provide a mechanism for detecting erosion and / or degradation of various features of the substrate support and / or detecting one or more properties of the substrate support (e.g., surface roughness, concentricity of circular elements, recesses or grooves, planarity, etc.) without adding any additional measurement devices to the process chamber housing the substrate support.Additionally, embodiments can identify erosion of a process kit ring around a substrate support and can be used to determine if and / or when to replace the process kit ring.

[0016] In embodiments, the systems and methods described herein can also be used to evaluate the condition of chamber parts other than the substrate support. For example, the systems and methods can be used to evaluate the condition of one or more characteristics of other chamber parts, such as the hole size of the gas delivery holes of the showerhead, the estimated gas delivery (e.g., flow rate and / or total delivered gas) to each of multiple regions of the gas distribution device (e.g., gas distribution plate or showerhead), the erosion of a protective layer coating on the surface of the gas distribution device, the gas or plasma delivery rate from various parts of the gas distribution device, the erosion of the process kit ring, etc. According to embodiments of the present disclosure, the condition of many other types of chamber parts can also be evaluated. The condition of the chamber parts can be used to perform diagnostics of the process chamber, improve the design of one or more parts of the process chamber, determine when to perform maintenance on and / or replace one or more parts of the process chamber, and so on.

[0017] In some embodiments, the methods and systems discussed herein can be used to detect floating, worn, and / or eroded chamber parts over time. Additionally, the methods and systems can be used to evaluate whether new parts meet specifications. In one example, when a part is switched out during planned maintenance, the process described herein can be run to qualify the new part by running tests to qualify it as being within specified values.

[0018] Thus, embodiments add new detection capabilities to process chambers without increasing costs to those process chambers. In embodiments, these new detection capabilities may be leveraged to predict reduced production rates, reduced product quality, increased amounts of product wasted, part failures, flaws in chamber part designs, and the like. In embodiments, these new detection capabilities may be further used to determine when to perform maintenance on one or more chamber parts of a process chamber, resulting in reduced unplanned downtime, reduced product wastage, reduced chamber failures, and the like. In embodiments, these new detection capabilities may be further used to adjust one or more process recipes, adjust settings and / or positions of one or more chamber parts, perform closed-loop control of the process, and the like.

[0019] FIG. 1 illustrates an exemplary computer system architecture 100 according to aspects of the disclosure. The computer system architecture 100 includes a client device 120, a manufacturing device 122, a substrate metrology system 126, a prediction server 112 (e.g., for generating prediction data, providing model adaptation, using a knowledge base, etc.), and a data store 150. The prediction server 112 can be part of a prediction system 110. The prediction system 110 can further include server machines 170 and 180. In some embodiments, the computer system architecture 100 can include or be part of a manufacturing system for processing substrates, such as the manufacturing system 200 of FIG. 2A. In additional or alternative embodiments, the computer system architecture 100 can include or be part of a chamber component condition prediction system (e.g., for evaluating the condition of one or more chamber components in a process chamber). Further details regarding the chamber component condition prediction system are provided with respect to FIGS. 3-4.

[0020] Components of client devices 120, manufacturing equipment 122, substrate metrology system 126, prediction system 110, and / or data store 150 may be coupled to one another via network 140. In some embodiments, network 140 is a public network that provides client devices 120 with access to prediction server 112, data store 150, and other publicly available computing devices. In some embodiments, network 140 is a private network that provides client devices 120 with access to manufacturing equipment 122, substrate metrology system 126, data store 150, and / or other privately available computing devices. Network 140 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.

[0021] The client devices 120 may include computing devices such as personal computers (PCs), laptops, mobile phones, smartphones, tablet computers, netbook computers, network-connected televisions ("smart TVs"), network-connected media players (e.g., Blu-ray players), set-top boxes, over-the-top (OTT) streaming devices, operator boxes, and the like.

[0022] The manufacturing equipment 122 can fabricate products according to a recipe. In some embodiments, the manufacturing equipment 122 can include or be part of a manufacturing system that includes one or more stations (e.g., process chambers, transfer chambers, load locks, factory interfaces, etc.) configured to perform various operations on substrates.

[0023] The substrate measurement system 126 may be a component of a manufacturing system that may be used to measure substrates before and / or after they are processed in one or more process chambers. The substrate measurement system 126 may be configured to generate optical emission spectroscopy data, reflected light measurement data, and / or other metrology data. The substrate measurement system 126 may include one or more components configured to collect and / or generate measurement data associated with one or more portions of a profile of a surface of a substrate after the substrate is removed from a process chamber.

[0024] In some embodiments, the substrate metrology system 126 can be configured to generate metrology data associated with substrates processed by other manufacturing equipment 122. The metrology data can include one or more values ​​of film property data (e.g., wafer spatial film properties such as thickness), dimensions (e.g., thickness, height, etc.), dielectric constants, dopant concentrations, densities, defects, etc. The metrology data can be of finished or semi-finished products, or test substrates such as blanket wafers. Some embodiments are discussed with reference to using reflected light metrology data and thickness profile maps to determine the condition of chamber components. However, it should be understood that the principles and embodiments described herein with respect to reflected light metrology data and thickness profile maps also apply to other types of metrology data. For example, measurements can be made of particle counts, optical constants of coatings, surface roughness of coatings, material composition of coatings, etc. Such measurements can be made for many regions on the substrate and can be used to generate profile maps of particle counts, optical constants, surface roughness, material composition, etc. across the measured substrate.

[0025] The substrate measurement system 126 can be configured to generate metrology data associated with a substrate before or after a substrate process. The substrate measurement system 126 can be integrated with a station of a manufacturing system, including the manufacturing equipment 122. In some embodiments, the substrate measurement system 126 can be coupled to or be a part of a station (e.g., a process chamber, a transfer chamber, etc.) of a process tool that is maintained in a vacuum environment. Such a substrate measurement system 126 can be referred to as an on-board metrology instrument. Thus, the substrate can be measured by the substrate measurement system 126 while the substrate is in a vacuum environment. For example, the substrate measurement system can be mounted in a transfer chamber that is maintained in a vacuum. For example, after a substrate process (e.g., an etch process, a deposition process, etc.) is performed on the substrate, metrology data can be generated for the processed substrate by the substrate measurement system 126 without the processed substrate being removed from the vacuum environment. Such a substrate measurement system 126 can be referred to as an on-board metrology instrument. In other or similar embodiments, the substrate measurement system 126 can be coupled to or be a part of a manufacturing system that is not maintained in a vacuum environment (e.g., a factory interface module, etc.). In some embodiments, the substrate measurement system is mounted in a process chamber or tool cluster that is not maintained under vacuum, such as a chemical mechanical polishing (CMP) process chamber.

[0026] As an alternative to the substrate measurement system 126 being included within the manufacturing system (e.g., mounted to a factory interface or transfer chamber), the substrate measurement system 126 may be a device separate from (i.e., external to) the manufacturing equipment 122. For example, the substrate measurement system 126 may be a stand-alone piece of equipment that is not coupled to any station of the manufacturing equipment 122. To obtain measurements for a substrate using the detached substrate measurement system 126, a user of the manufacturing system (e.g., a technician, an operator) may remove a substrate processed on the manufacturing equipment 122 from the manufacturing equipment 122 and transfer it to the substrate measurement system 126 for measurement. In some embodiments, the substrate measurement system 126 may transfer metrology data generated for the substrate over the network 140 to a client device 120 coupled to the substrate measurement system 126 (e.g., for presentation to a manufacturing user, such as an operator or technician). In other or similar embodiments, a manufacturing system user may obtain metrology data for a substrate from the substrate measurement system 126 and provide the metrology data to a computer system architecture via a graphical user interface (GUI) of the client device 120.

[0027] The data store 150 can be a memory (e.g., random access memory), a driver (e.g., a hard driver, a flash driver), a database system, or another type of component or device capable of storing data. The data store 150 can include multiple storage components (e.g., multiple drivers or multiple databases) that can span multiple computing devices (e.g., multiple server computers). The data store 150 can store profile maps (e.g., generated from reflected light measurement data, spectral data, etc.), such as film thickness profile maps and / or other substrate profile maps. The film thickness profile maps and / or other substrate profile maps can include historical maps and / or current maps.

[0028] One or more portions of data store 150 can be configured to store data that is not accessible to users of the manufacturing system. In some embodiments, all data stored in data store 150 can be inaccessible to manufacturing system users. In other or similar embodiments, a portion of the data stored in data store 150 is inaccessible to users, while another portion of the data stored in data store 150 is accessible to users. In some embodiments, the inaccessible data stored in data store 150 is encrypted using an encryption mechanism unknown to the users (e.g., the data is encrypted using a private encryption key). In other or similar embodiments, data store 150 can include multiple data stores, where data that is inaccessible to users is stored in a first data store and data that is accessible to users is stored in a second data store.

[0029] In some embodiments, prediction system 110 includes a server machine 170 and a server machine 180. Server machine 170 includes a training set generator 172 capable of generating a training data set (e.g., a set of data inputs and a set of target outputs) for training, certifying, and / or testing a machine learning model 190 or a set of machine learning models 190. Some operations of training set generator 172 are described in more detail below with respect to FIG. 5. In some embodiments, training set generator 172 can partition the training data into a training set, a validation set, and a test set.

[0030] The server machine 180 may include a training engine 182. An engine 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 enable training of one machine learning model 190 or a set of machine learning models 190. The machine learning model 190 may refer to a model artifact resulting from the training engine 182 using training data including training inputs and corresponding target outputs (correct responses to each training input). The training engine 182 may discover patterns in the training data that map the training inputs to the target outputs (predicted responses) and provide a machine learning model 190 that captures these patterns. The machine learning model 190 may include a linear regression model, a partial least squares regression model, a Gaussian regression model, a random forest model, a support vector machine model, a neural network, a ridge regression model, etc.

[0031] The training engine 182 may also be enabled to certify the trained machine learning models 190 using a corresponding set of features of the validation set from the training set generator 172. In some embodiments, the training engine 182 may assign a performance grade to each of the set of trained machine learning models 190. The performance grade may correspond to the accuracy of the respective trained model, the speed of the respective model, and / or the efficiency of the respective model. The training engine 182 may select a trained machine learning model 190 having a performance grade that satisfies a performance criterion to be used by the prediction engine 114, according to embodiments described herein. Further details regarding the training engine 182 are provided with respect to FIG. 6.

[0032] The prediction server 112 includes a prediction engine 114 that can provide data from the substrate metrology system 126 (e.g., a film thickness profile map) as input to a trained machine learning model 190 and execute the trained model 190 on the input to obtain one or more outputs. In some embodiments, the trained model 190 executed by the prediction engine 114 is selected by the training engine 182 as having a performance grade that meets the performance criteria. As will be further described with respect to FIG. 7, in some embodiments, the prediction engine 114 processes the input data using the model 190 to evaluate one or more conditions of one or more chamber parts for the process chamber.

[0033] It should be noted that in some other embodiments, the functionality of server machines 170 and 180 and prediction server 112 may be provided by more or fewer machines. For example, in some embodiments, server machines 170 and 180 may be integrated into a single machine, while in some other or similar embodiments, server machines 170 and 180 and prediction server 112 may be integrated into a single machine. In general, functionality described in one embodiment as being performed by server machine 170, server machine 180, and / or prediction server 112 may also be performed on client device 120. In addition, functionality attributed to a particular component may also be performed by different or multiple components operating together.

[0034] 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 that are aggregated as a group of administrators may be considered a "user."

[0035] 2A is a schematic top view of an exemplary manufacturing system 200 according to aspects of the present disclosure. The manufacturing system 200 can perform one or more processes on a substrate 202. The substrate 202, according to aspects of the present disclosure, can be any suitably rigid, dimensionally fixed, planar object suitable for fabricating electronic devices or circuit components thereon, such as, for example, a silicon-containing disk or wafer, a patterned wafer, a glass plate, etc. In some embodiments, the manufacturing system 200 can include or be part of a computer system architecture 110 according to an embodiment described with respect to FIG.

[0036] The manufacturing system 200 can include a process tool 204 and a factory interface 206 coupled to the process tool 204. The process tool 204 can include a housing 208 having a transfer chamber 210 therein. The transfer chamber 210 can include one or more processing chambers (also referred to as process chambers) 214, 216, 218 arranged around and coupled to the transfer chamber 210. The processing chambers 214, 216, 218 can be coupled to the transfer chamber 210 through respective ports, such as slit valves. The transfer chamber 210 can also include a transfer chamber robot 212 configured to transfer the substrate 202 between the process chambers 214, 216, 218, the load lock 220, and the like. The transfer chamber robot 212 can include one or more arms, each arm including one or more end effectors at the end of each arm. The end effectors can be configured to handle a particular object, such as a wafer.

[0037] In some embodiments, the transfer chamber 210 can also include metrology equipment, such as a substrate metrology system 126 mounted to the transfer chamber 210. The substrate metrology system 126 can be configured to generate metrology data associated with the substrate 202 before or after substrate processing while the substrate is maintained in the vacuum environment. As shown in FIG. 2, the substrate metrology system 126 can be mounted to or disposed within the transfer chamber 210. When the substrate metrology system 126 is disposed within or coupled to the transfer chamber 210, metrology data associated with the substrate 202 can be generated without the substrate 202 being removed from the vacuum environment (e.g., transferred to the factory interface 206).

[0038] The process chambers 214, 216, 218 can be adapted to perform any number of processes on the substrate 202. The same or different substrate processes can be performed in each processing chamber 214, 216, 218. Substrate processes can include atomic layer deposition (ALD), physical vapor deposition (PVD), chemical vapor deposition (CVD), etching, annealing, hardening, pre-cleaning, metal or metal oxide removal, polishing, wet processing, etc. Other processes can also be performed on the substrates herein.

[0039] The load lock 220 may also be coupled to the housing 208 and the transfer chamber 210. The load lock 220 may be configured to interface and be coupled to the transfer chamber 210 on one side and to the factory interface 206 on the other side. In some embodiments, the load lock 220 may have an environmentally controlled atmosphere that may be changed from a vacuum environment (where substrates may be transferred to and from the transfer chamber 210) to an inert gas environment at atmospheric pressure (or near atmospheric pressure) (where substrates may be transferred to and from the factory interface 206).

[0040] The factory interface 206 may be any suitable enclosure, such as a front-end equipment module (EFEM). The factory interface 206 may be configured to receive substrates 202 from substrate carriers 222 (e.g., front-opening integrated pods (FOUPs)) docked to various load ports 224 of the factory interface 206. A factory interface robot 226 (shown in dashed lines) may be configured to transfer substrates 202 between the substrate carriers (also referred to as containers) 222 and the load locks 220. In other and / or similar embodiments, the factory interface 206 may be configured to receive replacement parts from replacement part storage containers 222.

[0041] In some embodiments, the manufacturing system 200 may include a substrate metrology system 126 mounted to the factory interface 206. The substrate metrology system 126 mounted to the factory interface may be configured to generate metrology data associated with the substrate 202 before the substrate 202 is placed into the vacuum environment (e.g., transferred to the load lock 220) and / or after the substrate 202 is removed from the vacuum environment (e.g., removed from the load lock 220).

[0042] The manufacturing system 200 may also be connected to a client device (e.g., client device 120 of FIG. 1 ) configured to provide information about the manufacturing system 200 to a user (e.g., an operator). In some embodiments, the client device may provide information to a user of the manufacturing system 200 via one or more graphical user interfaces (GUIs). For example, the client device may provide information about one or more chamber condition metrics of the processing chambers 214, 216, 218 (e.g., while performing a substrate process) via the GUI.

[0043] The manufacturing system 200 may also include or be coupled to a system controller 228. The system controller 228 may be and / or may include a computing device such as a personal computer, a server computer, a programmable logic controller (PLC), a microcontroller, etc. The system controller 228 may include one or more processing devices, which may be general-purpose processing devices such as a microprocessor, a central processing unit, etc. More specifically, the processing device may be a complex instruction set computing (CISC) microprocessor, a reduced instruction set computing (RISC) microprocessor, a very long instruction word (VLIW) microprocessor, or a processor implementing other instruction sets or a processor implementing a combination of instruction sets. The processing device may also be one or more special-purpose processing devices such as an application specific integrated circuit (ASIC), a field programmable gate array (FPGA), a digital signal processor (DSP), a network processor, etc. The system controller 228 may include a data storage device (e.g., one or more disk drivers and / or solid state drivers), a main memory, a static memory, a network interface, and / or other components. The system controller 228 may execute instructions to perform any one or more of the techniques and / or embodiments described herein. In some embodiments, the system controller 228 may execute instructions to perform one or more operations in the manufacturing system 200 in accordance with a process recipe. These instructions may be stored in a computer-readable storage medium, which may include a main memory, a static memory, a secondary storage, and / or a processing device during execution of the instructions.

[0044] In some embodiments, the system controller 228 may receive data from the substrate measurement system 126 based on measurements of the substrates processed by the process chambers 214, 216, 218. The data received by the system controller 228 may include spectral data, reflected light measurement data, and / or other data for all or a portion of the substrate 202. The data received from the substrate measurement system 126 may be stored in a data store 250. The data store 250 may be included as a component within the system controller 228 or may be a separate component from the system controller 228. In some embodiments, the data store 250 may be or may include a portion of the data store 150, as described with respect to FIG. 1.

[0045] 2B illustrates one embodiment of a substrate measurement system 251 that can be used to measure a processed substrate. The substrate measurement system 251 can be an integrated measurement and / or imaging system (e.g., an integrated reflected light measurement (IR) system) configured to measure film properties (e.g., thickness, etc.) across a surface of the substrate 264 after the substrate 264 has been processed in a processing chamber. Reflectance measurement is a measurement technique that uses measured changes in light reflected from an object to determine the geometric and / or material properties of the object. A reflectance spectrometer measures the intensity of the reflected light over a range of wavelengths. For dielectric films, these intensity variations can be used to determine the thickness of the film.

[0046] Process results including film thickness can be monitored across one or more substrates for etching and deposition processes, for example, using the substrate measurement system 251. An integrated measurement and / or imaging system can be used to measure the surface of the substrate 264 while the substrate 264 is still in the device manufacturing system. In an embodiment, the substrate measurement system 251 can correspond to the substrate measurement system 126. The substrate measurement system 251 can be connected to a factory interface or a transfer chamber. Alternatively, the substrate measurement system 251 can be located inside the factory interface or the transfer chamber. The substrate measurement system 251 can also be a stand-alone system that is not connected to a manufacturing system. The substrate measurement system 251 can be mechanically isolated from the factory interface and the external environment to protect the substrate measurement system 251 from external vibrations. In some embodiments, the substrate measurement system 251 and the components included therein can provide analytical measurements (e.g., thickness measurements) that can provide a uniformity profile across the surface of the substrate 264, referred to herein as a profile map. A computing device can process the data from the substrate measurement system 251 and provide feedback to a user. The substrate metrology system 251 can be an assembly capable of measuring film thickness and / or other film properties such as optical constants, particle count, roughness, etc., over a portion of a substrate or over the entire substrate after the substrate has been processed in the chamber. The results of the measurements can be used to determine when to perform maintenance on the process chamber, when to perform further tests on the substrate, when to flag the substrate as out of specification, etc.

[0047] When the substrate 264 is lowered onto the substrate support 256 (e.g., a chuck) and secured to the substrate support 256, the center of the substrate 264 may be offset from the center of the chuck. A processing device of the substrate measurement system 251 may determine one or more coordinate transformations between the center of the substrate 264 and the center of the chuck 256 (the center of the chuck corresponds to the axis of rotation about which the chuck rotates) and apply the one or more coordinate transformations to correct for the offset, as described in more detail below.

[0048] The substrate measurement system 251 may include a rotary actuator 252 and a linear actuator 254. The rotary actuator 252 may be a motor, a rotary actuator (e.g., an electric rotary actuator), etc. The linear actuator 254 may be an electric linear actuator and may convert rotational motion in a motor to linear or linear motion along an axis. The substrate measurement system 251 may include a substrate support 256, a camera 208, a sensor 260, and a processing device 262.

[0049] The substrate support 256 can be a vacuum chuck, an electrostatic chuck, a magnetic chuck, a mechanical chuck (e.g., a four-jaw chuck, a three-jaw chuck, an edge / ring clamp chuck, etc.), or other type of chuck. The substrate support 256 can secure a substrate 264 (e.g., a wafer). The rotational actuator 252 can rotate the substrate support 256 about a first axis 253. The rotational actuator 252 can be controlled by a servo controller and / or a servo motor, which can enable precise control of the rotational position, velocity, and / or acceleration of the rotational actuator and thus the substrate support 256. The linear actuator 254 can move the substrate support 256 linearly along a second axis 255. The linear actuator 254 can be controlled by a servo controller and / or a servo motor 272, which can enable precise control of the linear position, velocity, and acceleration of the linear actuator 254 and thus the substrate support 256.

[0050] The camera 258 can be disposed on the substrate support 256 and can generate one or more images of a substrate 264 held by the substrate support 256. The camera 258 can be an optical camera, an infrared camera, or any other suitable type of camera. The sensor 260 can also be disposed on the chuck 256 and can measure at least one target location on the substrate at a time (e.g., can generate reflected light or other measurements of the target locations). The camera 258 and the sensor 260 can be fixed in a stationary position on the substrate measurement system 251, and the chuck 256 can be moved with an rθ motion by the rotary actuator 252 and the linear actuator 254.

[0051] In some embodiments, the processing device 262 can determine that the substrate 264 is not center aligned on the chuck 256 based on one or more images of the substrate 264 generated by the camera 258. The substrate 264 may not be center aligned on the substrate support 256 when initially placed on the substrate support 256. A robot blade 270 can place the substrate 264 on a transfer station 268 (e.g., a set of lift pins). The substrate support 256 can be moved in a first direction along a second axis 255 such that the chuck 256 is placed on the transfer station 268. The transfer station 268 can be located on (or can be a set of lift pins), and the lift mechanism 266 can move the transfer station 268 up and down in a vertical direction (orthogonal to the second axis 255 and parallel to the first axis 253). The substrate 264 can be received by the substrate support 256 while the substrate support 256 is placed on the transfer station 268. The substrate 264 may not be centered on the substrate support 256. The substrate support 256 may be moved in a second direction along the second axis 255 until the edge of the substrate 264 is detected by the sensor 260 to be at the target position.

[0052] The substrate support may be rotated 360 degrees and images may be generated during the rotation of the substrate support. One or more of the measurements and / or images may be obtained with the chuck at a variety of different θ values ​​and the detected location of the edge may vary. The change in the detected edge may indicate that the substrate (which may be a circular substrate) is off-center. Additionally, the determined change in the detected edge may be used to calculate the amount of offset.

[0053] In one embodiment, the parameters (r, θ) determine the offset of the substrate relative to the stage. These parameters allow the motion system to generate forward and inverse transformations that convert the (r, θ) coordinates of the stage to the (r, θ) coordinates of the substrate. The motion system can then calculate the trajectory in space of the substrate while sending commands to move the motors attached to the substrate support 256. In one embodiment, the motion system runs real-time control software connected (e.g., through an EtherCAT network) to the motion drivers of the linear and rotary actuators, so that any trajectory in space can be calculated. The controller 262 can calculate the corrected trajectory and transmit commanded positions to the motion drivers in real time (e.g., at a rate of 1 kHz).

[0054] In some embodiments, rotation of the substrate support 256 by the rotational actuator 252 for measurement of the target position causes an offset between the field of view of the sensor 260 and the target position on the substrate 264 due to the substrate 264 not being centered on the chuck 256. In this case, the linear actuator 254 can move the substrate support 256 linearly along a second axis to correct the offset. The sensor 260 can then measure the target position on the substrate 264. After measurements of all target points on the substrate have been measured, the processing device 262 can determine a uniformity profile across the surface of the substrate 264 based on the measurements.

[0055] In some embodiments, the processing device 262 can determine one or more coordinate transformations between the center of the substrate support 256 (corresponding to the first axis 253 about which the chuck 256 rotates) and the center of the substrate 264 that are applied during rotation of the substrate support 256 to correct for the offset.

[0056] 2C is a schematic cross-sectional side view of a substrate measurement subsystem 282 according to aspects of the disclosure. The substrate measurement subsystem 282 may be configured to obtain measurements on one or more portions of a substrate, such as the substrate 202 of FIG. 2A, before or after processing of the substrate 202 in a processing chamber. In an embodiment, the substrate measurement subsystem 282 may correspond to the substrate measurement system 126 of FIG. 2A. The substrate measurement subsystem 282 may obtain measurements on a portion of the substrate 202 by generating data associated with the portion of the substrate 202. In some embodiments, the substrate measurement subsystem 282 may be configured to generate spectral data, positional data, and / or other property data associated with the substrate 202.

[0057] The substrate measurement subsystem 282 can be configured to generate one or more types of data for the substrate, including spectral data, positional data, substrate property data, and the like. The substrate measurement subsystem 282 can generate data for the substrate in response to a request to obtain one or more measurements for the substrate before or after the substrate is processed in the manufacturing system. The substrate measurement subsystem 282 can include one or more components that facilitate the generation of data for the substrate. For example, the substrate measurement subsystem can include a spectral sensing component for sensing one or more spectra from a portion of the substrate to generate spectral data for the substrate. In some embodiments, the spectral sensing component can be an interchangeable component and can be configurable based on the type of process performed in the manufacturing system or the target type of measurement to be obtained in the substrate measurement subsystem. For example, one or more pieces of the spectral sensing component can be interchangeable in the substrate measurement subsystem to enable collection of reflectometry spectral data, ellipsometry spectral data, hyperspectral imaging data, chemical imaging (e.g., X-ray photoelectron spectroscopy (XPS), energy dispersive X-ray spectroscopy (EDX), X-ray fluorescence (XRF), and the like) data, and the like.

[0058] The substrate measurement subsystem 282 may include a controller 283 configured to execute one or more instructions to generate data associated with a portion of the substrate 202. The substrate measurement subsystem 282 may include a substrate sensing component 284 configured to detect when the substrate 202 has been transferred to the substrate measurement subsystem 282. The substrate sensing component 284 may include any component configured to detect when the substrate 202 has been transferred to the substrate measurement subsystem 282. For example, the substrate sensing component 284 may include an optical sensing component that transmits a light beam across an entrance to the substrate measurement subsystem 282. The substrate sensing component 284 may detect that the substrate 202 has been transferred to the substrate measurement subsystem 282 in response to the substrate 202 breaking the light beam transmitted across the entrance to the substrate measurement subsystem 282, once the substrate 202 is positioned within the substrate measurement subsystem 282. In response to detecting that the substrate 202 has been transferred to the substrate measurement subsystem 282, the substrate sensing component 284 may transmit an indication to the controller 283 indicating that the substrate 202 has been transferred to the substrate measurement subsystem 282.

[0059] In some embodiments, the substrate sensing component 284 can be further configured to detect identification information associated with the substrate 202. In some embodiments, the substrate 202 can be embedded in a substrate carrier (not shown) when the substrate 202 is transferred to the substrate measurement subsystem 282. The substrate carrier can include one or more registration features that enable identification of the substrate 202. For example, the optical sensing component of the substrate sensing component 284 can detect that the substrate 202 embedded in the substrate carrier breaks a light beam transmitted across an entrance to the substrate measurement subsystem 282. The optical sensing component can further detect one or more registration features included on the substrate carrier. In response to detecting the one or more registration features, the optical sensing component can generate an optical signature associated with the one or more registration features. The substrate sensing component 284 can transmit the optical signature generated by the optical sensing component to the controller 283 along with an indication that the substrate has been placed in the substrate measurement subsystem 282. In response to receiving the optical signature from the sensing component 214, the controller 283 can analyze the optical signature to determine identification information associated with the substrate 202. The identification information associated with the substrate 202 may include an identifier for the substrate 202, an identifier for a process for the substrate 202 (eg, a batch number or process run number), an identifier for a certain type of substrate 202 (eg, a wafer, etc.), and the like.

[0060] The substrate measurement subsystem 282 may include one or more components configured to determine a position and / or orientation of the substrate 202 within the substrate measurement subsystem 282. The position and / or orientation of the substrate 202 may be determined based on an identification of a reference location of the substrate 202. The reference location may be a portion of the substrate 202 that includes an identification feature associated with a unique portion of the substrate 202. For example, the substrate 202 may have a reference tag embedded in a center of the substrate 202. In another example, the substrate 202 may have one or more structural features on a surface of the substrate 202 that are included in a center of the substrate 202. The controller 283 may determine an identification feature associated with a unique portion of the substrate 202 based on the determined identification information for the substrate 202. For example, in response to determining that the substrate 202 is a wafer, the controller 283 may determine one or more identification features generally included in a portion of a wafer.

[0061] The controller 283 can identify a reference location relative to the substrate 202 using one or more camera components 285 configured to capture image data for the substrate 202. The camera components 285 can generate image data for one or more portions of the substrate 202 and transmit the image data to the controller 283. The controller 283 can analyze the image data to identify an identifying feature associated with the reference location relative to the substrate 202. The controller 283 can further determine a position and / or orientation of the substrate 202 shown in the image data based on the identified identifying feature of the substrate 202. The controller 283 can determine a position and / or orientation of the substrate 202 based on the identified identifying feature of the substrate 202 and the determined position and / or orientation of the substrate 202 shown in the image data.

[0062] In response to determining the position and / or orientation of the substrate 202, the controller 283 can generate position data associated with one or more portions of the substrate 202. In some embodiments, the position data can include one or more coordinates (e.g., Cartesian coordinates, polar coordinates, etc.), each coordinate associated with a portion of the substrate 202, and each coordinate determined based on a distance from a reference location relative to the substrate 202. For example, in response to determining the position and / or orientation of the substrate 202, the controller 283 can generate first position data associated with a portion of the substrate 202 that includes the reference location, the first position data including a Cartesian coordinate of (0,0). The controller 283 can generate second position data associated with a second portion of the substrate 202 relative to the reference location. For example, a portion of the substrate 202 located approximately 2 nanometers (nm) due east of the reference location can be assigned a Cartesian coordinate of (0,1). In another example, a portion of the substrate 202 located 5 nm due north of the reference location can be assigned a Cartesian coordinate of (1,0).

[0063] The controller 283 can determine one or more portions of the substrate 202 to measure based on the position data determined for the substrate 202. In some embodiments, the controller 283 can receive one or more operations of a process recipe associated with the substrate 202. In such embodiments, the controller 283 can further determine one or more portions of the substrate 202 to measure based on one or more operations of the process recipe. For example, the controller 283 can receive an indication that an etching process has been performed on the substrate 202 if several structural features have been etched onto the surface of the substrate 202. As a result, the controller 283 can determine one or more structural features to measure and expected locations of such features at various portions of the substrate 202.

[0064] The substrate measurement subsystem 282 can include one or more measurement components for measuring the substrate 202. In some embodiments, the substrate measurement subsystem 282 can include one or more spectral sensing components 220 configured to generate spectral data for one or more portions of the substrate 202. As previously discussed, the spectral data can correspond to the intensity (i.e., strength or amount of energy) of the detected energy waves for each wavelength of the detected waves.

[0065] In one embodiment, the reflected energy waves received by the substrate measurement subsystem 282 may include multiple wavelengths. Each reflected energy wave may be associated with a different portion of the substrate 202. In some embodiments, an intensity may be measured for each reflected energy wave received by the substrate measurement subsystem 282. The intensity may be measured for each wavelength of the reflected energy waves received by the substrate measurement subsystem 282. The association between the intensity and the wavelength may be the basis for forming the spectral data. In some embodiments, one or more wavelengths may be associated with intensity values ​​that are outside of an expected range of intensity values. In such an embodiment, an intensity value that is outside of an expected range of intensity values ​​may be an indication that a defect exists in a portion of the substrate 202.

[0066] The measurement components for measuring the substrate 202 may also include non-spectral sensing components configured to collect and generate non-spectral data. For example, the measurement components may include eddy current sensors or capacitive sensors. Although some embodiments of this description may refer to collecting and using spectral data for the substrate 202, embodiments of this description may also be applicable to non-spectral data collected for the substrate 202.

[0067] The spectral sensing component 287 can be configured to detect energy waves reflected from a portion of the substrate 202 and generate spectral data associated with the detected waves. The spectral sensing component 287 can include a wave generator 288 and a reflected wave receiver 224. In some embodiments, the wave generator 288 can be a light wave generator configured to generate a light beam toward the portion of the substrate 202. In such embodiments, the reflected wave receiver 289 can be configured to receive the light beam reflected from the portion of the substrate 202. The wave generator 288 can be configured to generate an energy flow 289 (e.g., a light beam) and transmit the energy flow 289 toward the portion of the substrate 202. The reflected energy wave 290 can be reflected from the portion of the substrate 202 and can be received by the reflected wave receiver 291. Although FIG. 2C shows a single energy wave reflected from the surface of the substrate 202, multiple energy waves can be reflected from the surface of the substrate 202 and can be received by the reflected wave receiver 291.

[0068] In response to the reflected wave receiver 291 receiving the reflected energy waves 290 from the portion of the substrate 202, the spectral sensing component 287 can measure the wavelength of each wave included in the reflected energy waves 289. The spectral sensing component 287 can further measure the intensity of each measured wavelength. In response to measuring each wavelength and each wavelength intensity, the spectral sensing component 287 can generate spectral data for the portion of the substrate 202. The spectral sensing component 287 can transmit the generated spectral data to the controller 283. In response to receiving the generated spectral data, the controller 283 can generate a mapping between the received spectral data and positional data for the measured portion of the substrate 202.

[0069] The substrate measurement subsystem 282 can be configured to generate a specific type of spectral data based on the type of measurement to be obtained by the substrate measurement subsystem 282. In some embodiments, the spectral sensing component 287 can be a first spectral sensing component configured to generate one type of spectral data. For example, the spectral sensing component 287 can be configured to generate reflectometry spectral data, ellipsometry spectral data, hyperspectral imaging data, chemical imaging data, thermal spectral data, or conductance spectral data. In such an embodiment, the first spectral sensing component can be removed from the substrate measurement subsystem 282 and replaced with a second spectral sensing component configured to generate a different type of spectral data (e.g., reflectometry spectral data, ellipsometry spectral data, hyperspectral imaging data, or chemical imaging data).

[0070] The controller 283 can determine a type of data (i.e., spectral data, non-spectral data) to be generated for the substrate 202 based on a type of measurement to be obtained for one or more portions of the substrate 202. In some embodiments, the controller 283 can determine the type or types of measurements based on notifications received from the system controller 228 of FIG. 2A. In other or similar embodiments, the controller 283 can determine the type or types of measurements based on instructions to generate measurements for a portion of the substrate 202. In response to determining the type or types of measurements to be obtained, the controller 283 can determine a type of data to be generated for the substrate 202. For example, the controller 283 can determine that spectral data should be generated for the substrate 202 and that the second spectral sensing component is the optimal sensing component to obtain the determined type of measurement for one or more portions of the substrate 202. In response to determining that the second sensing component is the optimal sensing component, the controller 283 may transmit a notification to the system controller indicating that the first spectral sensing component should be replaced with the second spectral sensing component and that the second spectral sensing component should be used to obtain one or more types of measurements on one or more portions of the substrate 202. The system controller 128 may transmit the notification to a client device connected to the manufacturing system, and the client device may provide the notification to a user (e.g., an operator) of the manufacturing system via a GUI.

[0071] In other or similar embodiments, the spectral sensing component 287 can be configured to generate multiple types of spectral data. In such embodiments, the controller 283 can cause the spectral sensing component 287 to generate a specific type of spectral data based on a type of measurement to be obtained for one or more portions of the substrate 202, according to the embodiments described above. In response to determining the type of measurement to be obtained, the controller 283 can determine that a first type of spectral data should be generated by the spectral sensing component 287. Based on the determination that the first type of spectral data should be generated by the spectral sensing component 287, the controller 283 can cause the spectral sensing component 287 to generate the first type of spectral data for one or more portions of the substrate 202.

[0072] As discussed above, the controller 283 can determine one or more portions of the substrate 202 for measurement with the substrate measurement subsystem 282. In some embodiments, one or more measurement components, such as the spectral sensing component 287, can be stationary components within the substrate measurement subsystem 282. In such embodiments, the substrate measurement subsystem 282 can include one or more position components 295 configured to modify a position and / or orientation of the substrate 202 relative to the spectral sensing component 287. In some embodiments, the position components 295 can be configured to translate the substrate 202 relative to the spectral sensing component 287 along a first axis and / or a second axis. In other or similar embodiments, the position components 295 can be configured to rotate the substrate 202 relative to the spectral sensing component 287 about a third axis.

[0073] Once the spectral sensing component 287 generates spectral data for one or more portions of the substrate 202, the position component 295 can modify a position and / or orientation of the substrate 202 according to the one or more determined portions to be measured for the substrate 202. For example, before the spectral sensing component 287 generates spectral data for the substrate 202, the position component 295 can position the substrate 202 at Cartesian coordinate (0,0), and the spectral sensing component 287 can generate first spectral data for the substrate 202 at the Cartesian coordinate (0,0). In response to the spectral sensing component 287 generating the first spectral data for the substrate 202 at the Cartesian coordinate (0,0), the positioning component 240 can translate the substrate 202 along a first axis such that the spectral sensing component 287 is configured to generate second spectral data for the substrate 202 at the Cartesian coordinate (0,1). In response to the spectral sensing component 287 generating the second spectral data for the substrate 202 at Cartesian coordinate (0,1), the controller 283 can rotate the substrate 202 along a second axis such that the spectral sensing component 287 is configured to generate third spectral data for the substrate 202 at Cartesian coordinate (1,1). This process can be performed multiple times until spectral data has been generated for each determined portion of the substrate 202.

[0074] In some embodiments, the surface of the substrate 202 may include one or more layers 297 of material. The one or more layers 297 may include an etching material, a photoresist material, a mask material, a deposition material, etc. In some embodiments, the one or more layers 297 may include an etching material to be etched according to an etching process performed in a processing chamber. In such embodiments, spectral data may be collected for one or more portions of an unetched etching material of the layer 212 deposited on the substrate 202 according to previously disclosed embodiments. In other or similar embodiments, the one or more layers 297 may include an etching material that has already been etched by an etching process in a processing chamber. In such embodiments, one or more structural features (e.g., lines, posts, openings, etc.) may be etched into the one or more layers 297 of the substrate 202. In such embodiments, spectral data may be collected for one or more structural features etched into the one or more layers 297 of the substrate 202.

[0075] In some embodiments, the substrate measurement subsystem 282 may include one or more additional sensors configured to capture additional data for the substrate 202. For example, the substrate measurement subsystem 282 may include additional sensors configured to determine a thickness of the substrate 202, a thickness of a film deposited on a surface of the substrate 202, etc. Each sensor may be configured to transmit the captured data to the controller 283.

[0076] In response to receiving at least one of the spectral data, positional data, or characteristic data for the substrate 202, the controller 283 can transmit the received data to the system controller 128 for processing and analysis according to embodiments described herein.

[0077] In some embodiments, the substrate measurement subsystem 282 includes one or more image capture devices 299 coupled to the controller 283, such as a camera (e.g., including a complementary metal oxide semiconductor (CMOS) sensor or a charge-coupled device (CCD) sensor). The image capture devices 299 can generate images (e.g., two-dimensional (2D) color images, infrared (IR) images, near-infrared images, etc.). In embodiments, these images can be processed along with the spectral data generated by the spectral sensing components 287 by one or more trained machine learning models to make a determination regarding one or more chamber components.

[0078] 3 illustrates an exemplary system architecture 300 for predicting or assessing chamber component conditions of a processing chamber in accordance with aspects of the disclosure. In some embodiments, system architecture 300 may include or be a part of one or more components of computer architecture 100 and / or manufacturing system 200. System architecture 300 may include one or more components of manufacturing equipment 122 (e.g., substrate measurement system 126), a server machine 320, and a server machine 350.

[0079] As previously described, the manufacturing equipment 122 may perform operations according to a recipe or over a period of time to create a product. The manufacturing equipment 122 may include a process chamber 310 configured to perform a substrate process on a substrate according to a substrate process recipe. In some embodiments, the recipe may be a blanket wafer recipe that deposits a film on a test substrate. In some embodiments, the process chamber 310 may be any of the process chambers 214, 218, 218 described with respect to FIG. 2. The manufacturing equipment 122 may also include a substrate metrology system 126, as described herein.

[0080] The manufacturing equipment 122 can be coupled to a server machine 320. The server machine 320 can include a processing device 322 and / or a data store 332. In some embodiments, the processing device 322 can be configured to execute one or more instructions to perform operations at the manufacturing equipment 122. For example, the processing device 322 can include or be part of the system controller 228 described with respect to FIG. 2. In some embodiments, the data store 332 can include or be part of the data store 150 and / or the data store 250.

[0081] The processing device 322 can be configured to receive data from one or more components of the manufacturing equipment 122 (i.e., over a network). For example, the processing device 322 can receive film thickness profile data (e.g., wafer map) 336 collected by the substrate measurement system 126 after the substrate is processed in a process chamber. In another example, the processing device 322 can receive metrology data collected by other metrology equipment before and / or after a substrate process for that substrate. The metrology data can include metrology measurements generated for the substrate by integrated metrology equipment. In some embodiments, the processing device 322 can store the received spectral data, film thickness profile data, and / or the received metrology data in a data store 332.

[0082] The processing device 352 may include a chamber part condition engine 330. The chamber part condition engine 330 in the processing device 322 may be configured to determine a chamber part condition of one or more chamber parts in the process chamber 310 that processed the measured substrate. The chamber part condition engine 330 may determine one or more chamber part condition metrics for one or more chamber parts of the process chamber 310 that processed the substrate from the film thickness profile data 336. In some embodiments, the chamber part condition metric may include a set of values ​​(e.g., a vector, a matrix, etc.) that indicates correlation of a particular data combination, correlation, pattern, and / or relationship present in the sensor data. For example, the chamber part condition metric may include a feature vector that includes binary values ​​that indicate the presence or absence of a particular feature in the data.

[0083] The chamber part condition metric can be compared to known patterns and / or combinations of chamber part metrics (e.g., target chamber part metrics). The target chamber part metrics can be associated with new or unworn chamber parts. In response to determining that the chamber part condition metric meets one or more chamber part condition thresholds (e.g., a condition for excessive wear of an electrostatic chuck mesa, or seal band, or lift pin hole, a condition for a missing mesa, a condition for a recess or valley in the surface of the substrate support, a condition for concentricity of a circular element on the surface of the substrate support, a condition for planarity of the surface of the substrate support, etc.), the chamber part condition engine 330 can flag or schedule the process chamber for maintenance and / or provide an alert to a user. The chamber part condition threshold can be a specified combination of values ​​indicated by the chamber part condition metrics.

[0084] In some embodiments, a curve or other numerical or non-numerical inference is generated based on one or more chamber part metrics of a healthy or standard chamber part over time. The curve or other numerical or non-numerical inference may represent, for example, standard wear of a seal band, mesa, and / or lift pin location of an electrostatic chuck or other substrate support over time. The current chamber part condition metric may be compared to the curve or other numerical or non-numerical inference. The number of times the chamber part has been used to process substrates and / or the number of substrates the chamber part has been used to process may be known. Using this information and the current chamber part condition metric, the current chamber condition metric may be plotted against the curve or other numerical or non-numerical inference (e.g., to determine where the current chamber part metric falls into the curve or other numerical or non-numerical inference). If the current chamber condition metric is within a threshold distance from the curve, it may be determined that the chamber part condition is as expected. If the current chamber condition metric deviates from the curve or other numerical or non-numerical inference by more than a threshold distance, it may be determined that the chamber part condition is not as expected (e.g., the chamber part is wearing out faster than normal for one or more areas or components of the chamber part).

[0085] As shown in FIG. 3, the processing device 322, in some embodiments, can include a training set generator 324 and / or a training engine 326. In some embodiments, the training set generator 324 can correspond to the training set generator 172 described with respect to FIG. 1, and / or the training engine 326 can correspond to the training engine 182. The training set generator 324 can be configured to generate a training set 340 for training a machine learning model 334 or a set of machine learning models 334. For example, the training set generator 324 can generate training inputs based on historical film thickness profile data 336. The film thickness profile data 336 (e.g., a film profile map) can be associated with a label that identifies a state of one or more features or regions of one or more chamber parts (e.g., an amount of wear or erosion of a seal band in one or more regions, an amount of wear or erosion to one or more mesas, etc.). In some embodiments, the training set generator 324 can retrieve the historical film thickness profile data 336 (e.g., film thickness profile data) from a data store 332 to generate the training inputs. The training set generator 324 can generate target outputs indicative of chamber part conditions (e.g., chamber part condition metrics) for the training inputs based on the historical film profile data 336. The training set generator 324 can include the generated training inputs and the generated target outputs in a training set 340. The training set 340 can further include an indicator of the age of one or more components (e.g., which may be expressed as the number of hours the component was used to process substrates and / or the number of substrates processed using the chamber part) for each historical film profile. Further details regarding generating the training set 340 are provided with respect to FIG. 5.

[0086] The training engine 326 can be configured to train, certify, and / or test the machine learning model 334 or sets of machine learning models 334. The training engine 326 can provide a training set 340 to train the machine learning model 334 and store the trained machine learning model 334 in the data store 332. In some embodiments, the training engine 326 can certify the trained machine learning model 334 using a validation set 342. The validation set 342 can include the film profile data 336 and associated chamber part condition metrics. The training set generator 324 and / or the training engine 326 can generate the validation set 342 based on the historical film profile data 336. In some embodiments, the validation set 342 can include historical film profile data 336 that is different from the historical film profile data 336 included in the training set 340.

[0087] The training engine 326 can provide the historical film profile data 336 as input to the trained machine learning model 334 and can extract chamber part condition metrics for the chamber parts of the process chamber from one or more outputs of the trained model 334. The input can further include the age of one or more components and / or one or more images of the substrate used to generate the historical film profile data 236. The training engine 326 can assign a performance score to the trained model 334 based on the accuracy of the chamber part condition metrics for the chamber parts of the process chamber. The training engine 326 can select the trained model 334 to be used to assess the condition of one or more chamber parts of the process chamber based on the film profile data for the substrates processed by the process chamber.

[0088] As previously discussed, the training set generator 324 and / or the training engine 326 may, in some embodiments, be components of a processing device 322 residing in the server 320. In additional or alternative embodiments, the training set generator 324 and / or the training engine 326 may also be components of a processing device 352 residing in the server 350. The server 350 may include or be part of a computing system separate from the manufacturing system 200. As previously discussed, the server 320 may, in some embodiments, include or be part of the system controller 228 described with respect to FIG. 2A. In such an embodiment, the server 350 may include or be part of a computing system coupled to the system controller 228 (i.e., via a network) but separate from the system controller 228. For example, a user of the manufacturing system 200 may be provided with access to data stored in one or more portions of the data store 332 or one or more processes executed on the processing device 322. However, a user of the manufacturing system 200 may not be provided with access to any of the data stored in one or more portions of the data store 354 or to any of the processes executed on the processing device 352 .

[0089] The processing device 352 can be configured to execute the training set generator 324 and / or the training engine 326, similar to the processing device 322. In some embodiments, the server 350 can be coupled to the manufacturing equipment 122 and / or the in-line metrology equipment 130 via a network. Thus, the processing device 352 can obtain film thickness profile data 336 and chamber part condition metrics corresponding to the film profile data 336 to be used by the training set generator 324 and / or the training engine 326 to generate the training set 340 and the validation set 342, according to the embodiments described with respect to the processing device 322. In other or similar embodiments, the server 350 is not coupled to the manufacturing equipment 122 and / or the external metrology equipment 130. Accordingly, the processing device 352 can obtain the film thickness profile data 336 from the processing device 322.

[0090] A training set generator 324 in the processing device 352 can generate the training set 340 according to the previously described embodiments. A training engine 326 in the processing device 352 can train and / or validate the machine learning model 334 according to the previously described embodiments. In some embodiments, the server 350 can be coupled to other manufacturing equipment and / or other server machines different from the manufacturing equipment 122 and / or the server machine 320. The processing device 352 can obtain film profile data 336 from the other manufacturing equipment and / or server machines according to the embodiments described herein. In some embodiments, the training set 340 and / or the validation set 342 can be generated based on the film thickness profile data 336 obtained for substrates processed in the process chamber 310 as well as other film profile data obtained for other substrates processed in process chambers in other manufacturing systems.

[0091] In response to the training engine 326 selecting the trained model 334 to be used, the processing device 352 may transmit the trained model 334 to the processing device 322. The chamber component condition engine 330 may use the trained model 334 to provide a chamber component condition assessment, as described above.

[0092] 4 illustrates a model training workflow 405 and a model application workflow 417 for chamber component condition assessment according to one embodiment. The model training workflow 405 and the model application workflow 417 may be performed by processing logic executed by a processor of a computing device. One or more of these workflows 405, 417 may be performed, for example, by one or more machine learning models implemented on the processing device and / or other software and / or firmware executing on the processing device.

[0093] A model training workflow 405 is for training one or more machine learning models (e.g., deep learning models) to determine one or more conditions of one or more features and / or regions of one or more chamber parts. A model application workflow 417 is for applying the one or more trained machine learning models to perform chamber part condition assessments. Each of the profile maps 412 can be associated with one or more labels indicating, for example, the erosion state and / or damage of the seal band, the condition of one or more mesas (e.g., the erosion state and / or presence of the mesa), the dielectric thickness of a region of the substrate support, the condition of the seal band at and / or around the lift pin holes (e.g., to identify lift pin hot spots), damage or aging of other components, etc.

[0094] Various machine learning outputs are described herein. A particular number and arrangement of machine learning models are described and illustrated. However, it should be understood that the number and type of machine learning models used, as well as the arrangement of such machine learning models, can be modified to achieve the same or similar end result. Thus, the arrangement of the machine learning models described and illustrated is merely an example and should not be construed as limiting.

[0095] In an embodiment, one or more machine learning models are trained to perform one or more chamber component condition estimation tasks. Each task may be performed by a separate machine learning model. Alternatively, a single machine learning model may perform each of the tasks or a portion of the tasks. For example, a first machine learning model may be trained to detect a first condition of a first chamber component, and a second machine learning model may be trained to detect a second condition of the first chamber component. In another example, a first machine learning model may be trained to detect one or more conditions of a first chamber component, and a second machine learning model may be trained to detect one or more conditions of a second chamber component. Additionally or alternatively, different machine learning models may be trained to perform different combinations of tasks. In one example, one or several machine learning models may be trained, and the trained machine learning (ML) model is a single shared neural network with multiple shared layers and multiple higher-level separate output layers, each of which outputs a different prediction, classification, identification, etc. For example, a first higher level output layer can determine chamber part conditions for a first type of chamber part (e.g., an electrostatic chuck), and a second higher level output layer can determine chamber part conditions for a second type of chamber part (e.g., a showerhead).

[0096] One type of machine learning model that can be used to perform some or all of the above tasks is an artificial neural network, such as a deep neural network. An artificial neural network generally includes a feature representation component with a classifier or regression layer that maps features to a target output space. A convolutional neural network (CNN), for example, hosts multiple layers of convolutional filters. Pooling can be performed to address nonlinearities at the lower layers, and a multi-layer perceptron is usually attached on top to map the upper layer features extracted by the convolutional layers to a decision (e.g., classification output). Deep learning is a type of machine learning algorithm that uses a cascade of multiple layers of nonlinear processing units for feature extraction and transformation. Each successive layer uses the output from the previous layer as input. Deep neural networks can be trained in a supervised (e.g., classification) and / or unsupervised (e.g., pattern analysis) manner. Deep neural networks include a hierarchy of multiple layers, with different layers learning different levels of representations corresponding to different levels of abstraction. In deep learning, each level learns to transform its input data into a slightly more abstract and complex representation. In particular, the deep learning process can learn which features naturally fall best into which levels. The "deep" in "deep learning" refers to the number of layers through which data is transformed. More precisely, deep learning systems have substantial Contribution Allocation Path (CAP) depth. A CAP is a chain of transformations from input to output. A CAP describes the potentially causal connections between inputs and outputs. For forward neural networks, the CAP depth can be the depth of the network, which can be the number of hidden layers plus one. For recurrent neural networks, where a signal can propagate through a layer more than once, the CAP depth is potentially unlimited.

[0097] Training a neural network can be accomplished in a supervised learning manner, which involves feeding a training data set of labeled inputs through the network, observing its output, defining an error (by measuring the difference between the output and the label value), and adjusting the mass of the network across all of its layers and nodes to minimize the error, using techniques such as deep gradient descent and backpropagation. In many applications, repeating this process across many labeled inputs in the training data set results in a network that can provide the correct output when presented with inputs different from those present in the training data set.

[0098] For the model training workflow 405, a training data set including hundreds, thousands, tens of thousands, hundreds of thousands, or more profile maps 412 of one substrate should be used to form the training data set. The data may include, for example, uniformity profiles determined using a given number of measurements, each associated with a particular target location. This data may be processed to generate one or more training data sets 436 for training one or more machine learning models. Training data items in the training data set 436 may include the profile map 412, labels indicating chamber part conditions of the chamber parts used to process the substrates measured to generate the profile maps, one or more images of the substrate, and / or age information for one or more chamber parts used to process the substrates.

[0099] To accomplish the training, the processing logic inputs the training data set 436 to one or more untrained machine learning models. Before inputting the first input to the machine learning model, the machine learning model can be initialized. The processing logic trains the untrained machine learning model based on the training data set to generate one or more trained machine learning models that perform the various operations described above. The training can be performed by inputting input data, such as one or more profile maps 412 (e.g., thickness profile map, spectral profile map, roughness profile map, particle count profile map, optical constant profile map, etc.), component images and / or age information, one at a time into the machine learning.

[0100] The machine learning model processes this input to generate an output. An artificial neural network includes an input layer consisting of values ​​in the data points. The next layer is called the hidden layer, and the nodes in the hidden layer each receive one or more of the input values. Each node includes parameters (e.g., masses) to apply to the input values. Thus, each node essentially inputs the input values ​​into a multivariate function (e.g., a nonlinear mathematical transformation) to result in an output value. The next layer can be another hidden layer or an output layer. In either case, the nodes in the next layer receive output values ​​from the nodes in the previous layer, and each node applies a mass to those values ​​and then generates its own output value. This can be done at each layer. The last layer is the output layer, and there is one node for each class, prediction, and / or output that the machine learning model can result in.

[0101] In response, the output may include one or more predictions or inferences (e.g., an estimate of a chamber part condition for one or more regions or features of one or more chamber parts of the process chamber in which the measured substrate was processed). The processing logic may compare the output estimated chamber part condition to the label chamber part condition. The processing logic may determine an error (i.e., a classification error) based on a difference between the estimated chamber part condition and the target chamber part condition. The processing logic adjusts the mass of one or more nodes in the machine learning model based on the error. An error term or delta may be determined for each node in the artificial neural network. Based on the error, the artificial neural network adjusts one or more of its parameters (masses for one or more inputs of the node) for one or more of its nodes. The parameters may be updated in a backpropagation manner, such that the nodes in the highest layer are updated first, followed by the nodes in the next layer, and so on. The artificial neural network includes multiple layers of "neurons", each layer receiving as input values ​​from the neurons in the previous layer. The parameters for each neuron include a mass associated with a value received from each of the neurons in the previous layer. Accordingly, adjusting the parameters may include adjusting masses assigned to each of the inputs to one or more neurons in one or more layers within the artificial neural network.

[0102] After the model parameters are optimized, model validation can be performed to determine if the model has been improved and to determine the current accuracy of the deep learning model. After one or more training rounds, the processing logic can determine if a stopping criterion has been met. The stopping criterion can be a target accuracy level, a target number of processed images from the training data set, a target amount of change to the parameters with respect to one or more previous data points, combinations thereof, and / or other criteria. In one embodiment, the stopping criterion is met when at least a minimum number of data points have been processed and at least a threshold accuracy has been achieved. The threshold accuracy can be, for example, 70%, 40%, or 90% accuracy. In one embodiment, the stopping criterion is met when the accuracy of the machine learning model stops improving. If the stopping criterion is not met, further training is performed. If the stopping criterion is met, the training can be completed. After the machine learning model is trained, the model can be tested using a reserved portion of the training data set. After one or more trained machine learning models 438 are generated, they can be stored in the model storage 445 and can be added to the chamber part condition engine 330.

[0103] According to an embodiment, for the model application workflow 417, the input data 462 can be input to one or more chamber part condition determiners 467, each of which can include a trained neural network, a statistical model, a physics-based model, and the like. Additionally or alternatively, the one or more chamber part condition determiners 467 can apply image processing algorithms to determine the chamber part condition. The input data can include a profile map generated based on measurements of a substrate processed using the process chamber (e.g., of a polymer layer on a substrate measured using an integrated reflected light measurement device or substrate measurement system, of a stack of patterned layers, and the like). The process chamber can include new and / or used components (e.g., new and / or used substrate supports, process kit rings, gas delivery plates, and the like). The input data may further include one or more images of the substrate and / or age information (e.g., number of hours of use, number of substrates processed) for one or more chamber parts of the process chamber used to process the substrate.

[0104] Based on the input data 462, the chamber part condition determiner 467 outputs one or more estimated chamber part conditions 469. The estimated chamber part conditions can include chamber part conditions for one or more chamber parts (which may be new or used), such as an electrostatic chuck or other substrate support, a gas delivery plate, a showerhead, another gas delivery device, a process kit ring, etc. Multiple chamber part conditions can be determined for a single chamber part. For example, a mesa condition, a seal band condition, and a lift pin location condition (e.g., lift pin location relative to where a substrate is placed in a process chamber) can all be determined relative to an electrostatic chuck or other substrate support.

[0105] The estimated mesa condition can include an estimate of an amount of erosion for one or more mesas of the substrate support, an estimate of missing mesas (e.g., new or used substrate support), an estimate of the mesa pattern (e.g., new or used substrate support), an estimate of the extent to which the estimated mesa pattern deviates from the target mesa pattern (e.g., new or used substrate support), and the like. The estimated seal band condition can include an estimate of erosion for one or more portions of the seal band of the substrate support, an estimate of the concentricity of the seal band, and the like. The estimated lift pin location condition can include an estimate of hot spots at one or more locations of the substrate support, an estimate of the lift pin location, an estimate of deviation of the estimated lift pin location from the target lift pin location, and the like. The process kit ring can surround the substrate support and / or can be disposed below the substrate support. The process kit ring condition estimate for the process kit ring can include an estimate of erosion for the process kit ring, an estimate of defects in the process kit ring (e.g., the amount the process kit ring is out of round), an estimate of missing process kit rings, and the like.

[0106] In some embodiments, the chamber component condition determiner 467 includes one or more models that receive input data 462 (e.g., a profile map of a substrate processed by the process chamber) and output chamber component conditions 469 and / or suggested adjustments to one or more chamber components of the process chamber that compensate for the current chamber component conditions 469. These models can be or include trained machine learning models (e.g., convolutional neural networks), tables that relate profile map data to recipes and / or equipment settings, physics-based models that relate profile map data to recipes and / or equipment settings (e.g., digital twins), and the like. These models can output recommended adjustments to one or more parameters of a recipe used to process a substrate associated with a profile map input to the models. Additionally or alternatively, these models can output recommended settings for one or more parameters of a recipe used to process a substrate associated with a profile map input to the models. Additionally or alternatively, these models can output recommended settings (e.g., chamber parameter settings) for one or more chamber components used to process a substrate associated with a profile map input to the models. Additionally or alternatively, these models can output recommended settings for parameters of one or more tools (tool parameter settings), which can include chamber parameter settings and / or settings for tools external to the process chamber. Such tools external to the process chamber can be connected to multiple different process chambers. Examples of external tools include abatement, gas supplies, vacuum pumps, etc. Such settings can be tool parameter settings (e.g., equipment settings) that are not associated with any particular recipe, such as the amount of power to deliver to a heating element to achieve a target temperature, valve settings to achieve a target flow rate, etc.Examples of strategy settings that can be adjusted or controlled based on the output of the model include one or more plasma power settings, one or more temperature set points (e.g., for various sections of a multi-section heater or chuck), one or more process step durations, one or more gas delivery flow rates, pressures, etc.

[0107] The action determiner 472 can determine one or more actions 470 to take based on the chamber component condition 469. In one embodiment, the action determiner 472 compares the chamber component condition estimates to one or more chamber component condition thresholds or other criteria. If one or more of the chamber component condition estimates meet or exceed the chamber component condition thresholds or other criteria, the action determiner 472 can determine that preventative maintenance is recommended and can output a recommendation or notification to schedule maintenance and / or replacement of one or more chamber components of the process chamber. Alternatively, the action determiner 472 can output one or more design change recommendations to modify the chamber components (e.g., to mitigate lift pin hot spots). Alternatively or additionally, the action determiner 472 can output one or more suggestions for remedy adjustments, setting adjustments for one or more chamber components, etc.

[0108] In some embodiments, the action determiner 472 automatically schedules maintenance of the process chamber based on the chamber part condition 469 meeting one or more maintenance criteria. For example, the chamber part condition can include an estimate of the amount of erosion or wear on the seal band of the substrate support, on the mesas for one or more regions of the substrate support, and / or around the lift pin holes of the substrate support. The chamber part condition can further include mesas present or missing. The chamber part condition can further include a dielectric layer thickness of the substrate support (e.g., electrostatic chuck). The chamber part condition can further include a bond layer delamination condition. The amount of erosion or wear for each region or feature can be compared to an erosion threshold. If the erosion exceeds the erosion threshold, maintenance can be recommended or initiated. Additionally, if one or more mesas are missing, a replacement of a puck for the electrostatic chuck can be recommended or initiated. Additionally, if one or more regions of the electrostatic chuck are determined to have a dielectric layer less than a target dielectric layer thickness, a replacement of a puck of the electrostatic chuck can be recommended or initiated.

[0109] In some embodiments, the action determiner 472 can output one or more remedy adjustments (e.g., suggestions for one or more remedy adjustments) and / or adjustments to settings or positions of one or more chamber components to compensate for the chamber component condition estimates. Such adjustments can be made automatically by the action determiner 472 in some embodiments. For example, if an edge seal band of a substrate support is eroding (e.g., an estimated seal band erosion value falls within a value range associated with a particular amount of erosion), a chucking voltage can be increased (e.g., to an amount associated with that value range) to maintain substrate chucking. In another example, if a mesa is worn (e.g., an estimated mesa erosion value falls within a value range associated with a particular amount of erosion), a backside cooling gas pressure can be reduced (e.g., to an amount associated with that value range) to compensate for the worn mesa. In another example, if process kit erosion is detected (e.g., an estimated process kit ring erosion value falls within a value range associated with a particular amount of erosion), a process kit ring height can be adjusted (e.g., to an amount associated with that value range) to compensate. This allows for closed-loop control of the process recipe based on the sensed chamber component conditions.

[0110] In some embodiments, the action determiner 472 determines one or more actions to perform based on the proposed recipe set points or parameters, equipment set points, etc. output by the chamber component condition determiner 467. Examples of actions that may be performed include adjusting set points of the chamber components, adjusting parameter values ​​for one or more steps of the process recipe being executed on the process chamber, etc. In response, in embodiments, the chamber component condition engine 330 can perform closed loop process control, forward process control, feedback process control, etc.

[0111] In one embodiment, the action determiner 472 includes a quality evaluator that can perform one or more tests or checks to determine the quality of one or more chamber parts. The quality evaluator can determine whether the chamber part condition satisfies one or more lift pin criteria, one or more seal band criteria, one or more mesa criteria, one or more layer delamination criteria, one or more process kit ring criteria, etc. For example, the seal band criteria can include a concentricity to substrate support threshold. If the detected concentricity to substrate support does not meet the concentricity threshold, the seal band criterion may not be met. In another example, the mesa criteria can include a mesa presence criterion. If any mesa is detected to be missing, the mesa presence criterion may not be met. In another example, a mesa pattern can be determined and compared to a target mesa pattern. If the determined mesa pattern deviates from the target mesa pattern by more than a threshold amount, the mesa pattern criterion may not be met. In another example, the presence of a process kit ring can be detected. If the process kit ring is determined to be missing, the process kit ring presence criterion may not be met. In another example, a process kit ring, mesa, and / or seal band erosion value can be determined and compared to a process kit ring, mesa, and / or seal band erosion threshold. If any of the detected erosion amounts exceed the respective erosion threshold, the erosion criteria may not be met. In another example, a delamination amount (indicative of the amount of delamination between layers of the substrate support) can be determined for the substrate support. If the delamination amount exceeds the delamination threshold, the delamination criteria may not be met. If any of these quality criteria are not met, the action determiner 472 can take appropriate action.

[0112] In one embodiment, the action determiner 472 determines a quality metric value for the process chamber and / or one or more chamber parts of the process chamber. The quality metric value can be based on one or more chamber part conditions, whether one or more chamber parts passed their respective tests, and / or the test score of the chamber part. The quality metric value can indicate the quality of the process chamber and / or the individual chamber parts. In one embodiment, the action determiner 472 can determine whether a chamber part (e.g., a substrate support) causes a reduction in product quality based on the quality metric value and / or the chamber part condition estimates (e.g., an estimated mesa condition, an estimated lift pin condition, an estimated seal band condition, etc.). If the chamber part causes a reduction in product quality (e.g., an increased device failure rate), the processing logic can determine that maintenance should be performed to replace or repair the chamber part.

[0113] In one embodiment, the action determiner 472 compares the chamber component condition to one or more previous chamber component conditions for the chamber component of the chamber. Based on such comparison, the action determiner 472 can determine a change and / or rate of change of one or more chamber component conditions (e.g., erosion rate for one or more regions of the substrate support). Based on the change or rate of change, the action determiner 472 can determine one or more actions to perform. In one embodiment, based on the rate of change of the chamber component and the current state of the chamber component, processing logic can determine when to perform maintenance on the chamber component.

[0114] In one embodiment, the action determiner 472 compares the chamber part condition to a curve. The curve may be a curve of chamber part condition over time for one or more chamber parts of a process chamber. Based on where the chamber part condition falls on the curve, the action determiner 472 can determine one or more actions to take.

[0115] 5 is a flow chart of a method 500 for generating a training data set for training a machine learning model to perform a chamber component condition assessment, according to an aspect of the disclosure. The method 500 is performed by processing logic, which may include hardware (circuitry, dedicated logic, etc.), software (such as running on a general-purpose computer system or a dedicated machine), firmware, or some combination thereof. In an embodiment, the method 500 may be performed by a computer system, such as the computer system architecture 100 of FIG. 1. In other or similar embodiments, one or more operations of the method 500 may be performed by one or more other machines not shown. In some aspects, one or more operations of the method 500 may be performed by the training set generator 324 of the server machine 320 or the server machine 350 described with respect to FIG. 3.

[0116] At block 510, processing logic initializes a training set T to an empty set (e.g., {}). At block 512, processing logic obtains substrate surface data associated with a substrate processed in a process chamber of the manufacturing system (e.g., reflected light measurement data of a surface of a film on the substrate, such as a film thickness profile map or a wafer map).

[0117] At block 514, processing logic obtains chamber component condition information for the process chamber that processed the substrate. As mentioned above, the chamber component condition information can include erosion information regarding one or more features of the substrate support, information regarding the condition of one or more features of the showerhead (e.g., gas delivery rate and / or amount to one or more regions of the showerhead), etc.

[0118] At block 516, processing logic generates training inputs based on the sensor data acquired for the substrate at block 512. In some embodiments, the training inputs may include a normalized set of sensor data (e.g., including surface reflectometer data as described herein).

[0119] At block 518, processing logic may generate target outputs based on the chamber component condition information obtained at block 514. The target outputs may correspond to chamber component condition metrics (data indicative of the condition of characteristics for one or more chamber components) for the process chamber in which the substrate was processed.

[0120] At block 520, processing logic generates an input / output mapping. The input / output mapping refers to training inputs that include or are based on data for a substrate, and target outputs for the training inputs, where the target outputs identify chamber part conditions, and the training inputs are associated with (or mapped to) the target outputs. At block 522, processing logic adds the input / output mapping to a training set T.

[0121] At block 524, the processing logic determines whether the training set T includes a sufficient amount of training data to train the machine learning model. Note that in some embodiments, the training set T may be determined to be sufficient based solely on the number of input / output mappings in the training set, whereas in some other embodiments, the training set T may be determined to be sufficient based on one or more other criteria (e.g., the degree of diversity of the training examples, etc.) in addition to or instead of the number of input / output mappings. In response to determining that the training set T includes a sufficient amount of training data to train the machine learning model, the processing logic provides the training set T for training the machine learning model. In response to determining that the training set does not include a sufficient amount of training data to train the machine learning model, the method 500 returns to block 512.

[0122] At block 526, the processing logic provides a training set T for training the machine learning model. In some embodiments, the training set T is provided to the training engine 326 of the server machine 320 and / or the server machine 350 to perform training. In the case of a neural network, for example, input values ​​of a given input / output mapping (e.g., spectral data and / or chamber data for a previous substrate) are input to the neural network, and output values ​​of the input / output mapping are stored in the output node of the neural network. Then, according to a learning algorithm (e.g., backpropagation, etc.), the connection mass in the neural network is adjusted, and the procedure is repeated for other input / output mappings in the training set T. After block 526, the machine learning model 190 can be used to provide chamber part conditions (e.g., chamber condition metrics) for chamber parts of the process chamber.

[0123] 6 is a flow diagram illustrating an embodiment of a method 600 for training a machine learning model to estimate a condition of one or more chamber parts of a process chamber. The method 600 is performed by processing logic that may include hardware (circuitry, dedicated logic, etc.), software (such as running on a general-purpose computer system or a dedicated machine), firmware, or some combination thereof. In an embodiment, the method 600 may be performed by a computer system such as the computer system architecture 100 of FIG. 1. In other or similar embodiments, one or more operations of the method 600 may be performed by one or more other machines not shown. In some aspects, one or more operations of the method 600 may be performed by the training engine 326 for the server machine 320 or the server machine 350 described with respect to FIG. 3.

[0124] At block 602 of method 600, processing logic assembles a training data set that may include data from multiple film profile maps (e.g., a film thickness profile map indicating film thickness of a polymer film for multiple locations on a substrate). Each data item in the training data set may include one or more labels. Data items in the training data set may include input level (e.g., image level) labels indicating the presence or absence of one or more chamber part conditions without identifying coordinates associated with the chamber part conditions. For example, some data items may include a label for a worn seal band, some data items may include a label for a worn mesa, and some data items may include a label for a non-adverse chamber part condition. Data items in the training data set may also include other labels, such as coordinate level (e.g., pixel level) classifications of chamber conditions. In embodiments, each data item includes a feature vector that includes a thickness value for each of multiple substrate coordinates. In some embodiments, each data item includes a wafer map, which may be an image of a wafer (e.g., a heat map of the wafer). Colors in the heat map may indicate film thickness or other parameter values. Alternatively, an actual thickness value can be used for each of many coordinates on the surface of the substrate (e.g., wafer). The labels can be or can include a mask that includes coordinate level labels, and each coordinate can be associated with one or more labels, such as an erosion value, a pass or fail value, one or more specific chamber part condition classifications, etc.

[0125] At block 604, data items from the training data set are input to an untrained machine learning model. At block 606, the machine learning model is trained to generate a trained machine learning model that classifies or estimates one or more chamber part conditions for one or more chamber parts of the process chamber based on the training data set. The machine learning model can also be trained to output one or more other types of predictions, coordinate level classifications, decisions, etc.

[0126] In one embodiment, at block 610, training data item inputs are input to the machine learning model. The inputs may include data from a film profile map (e.g., a wafer map) that indicates one or more film properties (e.g., thickness, optical constants, grain count, roughness, material properties, etc.) across the substrate. The data may be input as an image or feature vector, in an embodiment. At block 612, the machine learning model processes the inputs to generate an output. The output may include one or more chamber part condition classifications (e.g., worn seal bands, worn mesas, worn lift pin holes, worn showerhead gas delivery holes, etc.). The output may additionally or alternatively include one or more coordinate level chamber part condition classifications. For example, the output may indicate a wear or erosion value for each coordinate of the substrate. The output, in some embodiments, may indicate a wear or erosion value for one or more regions of a substrate support that supported the substrate during processing. For example, the output may indicate an amount of erosion or wear for each coordinate of the substrate support (e.g., an electrostatic chuck). The output may include, for each chamber part condition class that the machine learning model is trained to identify, a probability that the chamber part condition class is present.

[0127] At block 614, processing logic compares the output probabilities and / or values ​​of the chamber part conditions to labels of one or more chamber part conditions associated with the input. At block 616, processing logic determines an error based on a difference between the output and the label. At block 618, processing logic adjusts the mass of one or more nodes in the machine learning model based on the error.

[0128] At block 620, processing logic determines whether a stopping criterion has been met. If the stopping criterion has not been met, the method returns to block 610 where another training data item is input to the machine learning model. If the stopping criterion has been met, the method proceeds to block 625 where training of the machine learning model is completed.

[0129] In one embodiment, multiple different ML models are trained to identify different types of chamber part conditions and / or to identify chamber part conditions for different types of chamber parts.

[0130] In one embodiment, one or more ML models are trained to be applied across multiple process chambers, which may be of the same type or model. The trained ML models can then be further tuned for use with a particular instance of a process chamber. Further tuning can be performed by using additional training data items including film thickness profile maps of substrates processed by the process chamber. Such tuning can account for chamber mismatches between chambers and / or specific hardware process kits of some process chambers. Additionally, in some embodiments, further training is performed to tune the ML models for a process chamber after maintenance on the process chamber and / or one or more changes to the hardware of the process chamber.

[0131] 7 is a flow diagram of a method 700 for determining chamber conditions for one or more chamber components using one or more trained machine learning models, according to aspects of the disclosure. The method 700 is performed by processing logic, which may include hardware (circuitry, dedicated logic, etc.), software (such as running on a general-purpose computer system or a dedicated machine), firmware, or some combination thereof. In one embodiment, the method 700 may be performed by a computer system, such as the computer system architecture 100 of FIG. 1. In other or similar embodiments, one or more operations of the method 700 may be performed by one or more other machines not shown. In some aspects, one or more operations of the method 700 may be performed by the chamber component condition engine 330 of the server machine 320 described with respect to FIG. 3.

[0132] At block 710 of method 700, a process chamber of a substrate processing system (also referred to as a fabrication system) processes a substrate. For example, the process chamber may perform a film deposition process to deposit a film on the substrate or an etching process to partially remove a film on the substrate. In some embodiments, the film is a polymer. In other embodiments, the film is a ceramic (e.g., a metal oxide).

[0133] At block 712, one or more robots transfer the substrate from the process chamber to a substrate measurement system. If a substrate measurement system is connected to or contained within the transfer chamber, the transfer chamber robot can remove the substrate from the process chamber and insert the substrate into the substrate measurement system. If a substrate measurement system is connected to or contained within the factory interface, the transfer chamber robot can remove the substrate from the process chamber and place the substrate into a load lock. The factory interface robot can then remove the substrate from the load lock and insert the substrate into the substrate measurement system. The substrate measurement system can be any of the substrate measurement systems described above, such as an integrated reflected light measurement (IR) device.

[0134] At block 714, the substrate measurement system generates measurements of many points on the surface of the substrate. Each point may have a unique set of coordinates. At block 716, the substrate measurement system and / or the computing device may generate one or more profile maps of the film on the substrate based on the measurements of the substrate measurement system. The profile map may be or may include an image, where each pixel in the image corresponds to a coordinate on the substrate. Each pixel may have an intensity value that corresponds to a measurement (e.g., a thickness value) at the substrate coordinate associated with the pixel. The profile may be or may include a feature vector, where each entry in the feature vector is associated with a substrate coordinate, and each entry may have a value that represents a value of the film at the substrate coordinate.

[0135] At block 718, the computing device processes the one or more profile maps (e.g., thickness profile map, grain map, optical constant map, roughness map, etc.) using the one or more trained machine learning models. Each of the trained machine learning models outputs one or more chamber part condition estimations for one or more chamber parts of the process chamber. The output may include image-level or component-level chamber part estimations. For example, the output may indicate that electrostatic chuck wear has been detected or that showerhead wear has been detected. The output may include classifications of the one or more chamber part conditions. For example, the output may indicate that seal band erosion, mesa erosion, dielectric layer erosion, lift pin hole erosion, etc. have been detected. The output may provide values ​​for the one or more chamber part conditions, such as an amount of seal band erosion, an amount of mesa erosion, an amount of lift pin hole erosion, etc. In some embodiments, the output includes a coordinate-level or area-level classification of the chamber part condition for the chamber part. For example, the output may indicate, for various coordinates on the substrate support, erosion values, bond erosion values, dielectric thickness, planarity of the surface of the substrate support, concentricity of features on the surface of the substrate support, roughness of the surface of the substrate support, the presence and / or size of valleys or grooves in the surface of the substrate support, and / or chamber condition classifications identified for those coordinates, which may enable a user to determine which areas on the chamber parts are worn out or problematic.

[0136] In one embodiment, instead of or in addition to using a trained ML model, one or more computer vision algorithms are used to process the profile map. In one embodiment, a gold standard or reference film profile map associated with a healthy process chamber part is compared to the generated profile map. Based on this comparison, differences between the reference film profile map and the generated film profile map can be determined. Different types of differences can be associated with different chamber part condition classifications. For example, the reference film profile map can include a representation of a circle or ridge representing a mesa, and the absence of a similar circle or ridge in the generated film profile map can indicate that the electrostatic chuck is missing a mesa. The comparison can be a full image comparison between the reference film profile map and the generated film profile map. Alternatively, the comparison can be made between a patch or region of the generated film profile map and a patch or region of the reference film profile map (e.g., in the case of feature mapping).

[0137] In one embodiment, at block 720, processing logic of the computing device determines, based on the output of the machine learning model, a probability that the chamber part (or multiple chamber parts) will cause reduced product quality, product waste, processing delays, etc. In one embodiment, at block 722, processing logic estimates a failure time of one or more chamber parts of the process chamber based at least in part on the output of the machine learning model (e.g., based on the chamber part condition estimation).

[0138] At block 724, processing logic determines one or more actions to perform based at least in part on the chamber component condition estimate and / or the probability that the chamber component will cause a reduction in product quality. These actions may include generating a notification, recommending maintenance, scheduling maintenance, recommending or scheduling a part replacement for the one or more chamber components, recommending or scheduling a repair for the one or more chamber components, etc. In one embodiment, processing logic determines whether maintenance should be performed on one or more chamber components of the process chamber at block 726. This determination may then be output in a notification. In one embodiment, processing logic determines when to perform maintenance on the one or more chamber components based on the estimated failure time.

[0139] In one embodiment, the processing logic generates an image of the chamber part and outputs the image to a graphical user interface. The image can show various locations of the chamber part and label one or more regions with a chamber part condition classification. For example, the image can include one or more visualizations showing worn or missing mesas, areas of worn seal bands, areas of the surface of the electrostatic chuck having a dielectric thickness less than a threshold thickness, etc. The generated image can show estimated dielectric layer thickness and / or other characteristics across the surface of the electrostatic chuck, such as mesa and / or seal band thickness across the surface of the electrostatic chuck. If hot spots are identified around lift pin holes, such hot spots can be shown in the image. Thus, a graphical display of the electrostatic chuck can be provided that shows diagnostic information about the electrostatic chuck. From the image, a user can distinguish lift pin location, mesa location and / or thickness, seal band location and / or thickness, dielectric layer thickness, bond thickness, etc. The image can also identify cracks, heat distribution, etc.

[0140] Accordingly, in an embodiment, method 700 can be implemented to perform health diagnostics and / or anomaly detection on one or more chamber components, such as an electrostatic chuck. The embodiment can identify subtle signs of problems in one or more chamber components and provide steps to be taken to address the issues before they affect production.

[0141] 8 is a flow diagram of a method 800 for determining conditions for one or more chamber parts using one or more models (e.g., one or more trained machine learning models) according to aspects of the disclosure. The method 800 is performed by processing logic that may include hardware (circuitry, dedicated logic, etc.), software (such as running on a general-purpose computer system or a dedicated machine), firmware, or some combination thereof. In one embodiment, the method 800 may be performed by a computer system such as the computer system architecture 100 of FIG. 1. In other or similar embodiments, one or more operations of the method 800 may be performed by one or more other machines not shown. In some aspects, one or more operations of the method 800 may be performed by the chamber part condition engine 330 of the server machine 320 described with respect to FIG. 3.

[0142] At block 810 of method 800, a process chamber of a substrate processing system (also referred to as a manufacturing system) processes a substrate according to a process recipe (e.g., using an etch process or a deposition process). The process can result in a film (e.g., a polymer film) having particular film properties, such as film thickness, dielectric constant, density, etc. Depending on the process being performed, one or more film properties may be highly dependent on the particular conditions to be evaluated. For example, the film thickness for some deposition or etch processes may be temperature dependent. Accordingly, regions of the substrate having a thicker film may be associated with a greater or lesser temperature during processing, depending on the process. Thus, based on the thickness of the film at different points on the substrate, inferences can be made about the temperature of the substrate at those different points during the process. In another example, the film thickness at a point may depend on the amount of plasma that reaches the substrate at that point or the energy of the plasma. Thus, based on the thickness of the film at the different points, inferences can be made about the RF field affecting the plasma at those different points. The process recipe used to process the substrate may be a recipe designed to cause the greatest change in a film property, such as film thickness, with minimal change in the parameters of interest (e.g., temperature, RF field strength, plasma energy level, etc.).

[0143] At block 812, one or more robots can transfer the substrate from the process chamber to a substrate measurement system. If a substrate measurement system is connected to or included within the transfer chamber, the transfer chamber robot can remove the substrate from the process chamber and insert the substrate into the substrate measurement system. If a substrate measurement system is connected to or included within the factory interface, the transfer chamber robot can remove the substrate from the process chamber and place the substrate into a load lock. The factory interface robot can then remove the substrate from the load lock and insert the substrate into the substrate measurement system. The substrate measurement system can be any of the substrate measurement systems described above, such as an integrated reflected light measurement (IR) device.

[0144] At block 814, the substrate measurement system generates measurements of many locations on the surface of the substrate. Each location may have a unique set of coordinates. At block 816, the substrate measurement system and / or the computing device may generate one or more profile maps of the film on the substrate based on the measurements of the substrate measurement system. The profile map may be or may include an image, where each pixel in the image corresponds to a coordinate on the substrate. Each pixel may have an intensity value corresponding to a measurement (e.g., a thickness value) at the substrate coordinate associated with the pixel. The profile may be or may include a feature vector, where each entry in the feature vector is associated with a substrate coordinate, where each entry may have a value representing a value of the film at the substrate coordinate. In one embodiment, each value of the film corresponds to a condition during the execution of a process on the substrate that is measured. For example, each film thickness measurement may correspond to a temperature of the substrate during processing. In one embodiment, the profile map and information about the correlation between values ​​of the film property and conditions during processing (e.g., a function relating film thickness to temperature) are used to generate a new profile map, where each location of the profile map includes a value that represents the conditions during processing at that location (e.g., each value represents the temperature at that location during processing).

[0145] In one embodiment, the substrate measurement device further generates one or more images of the substrate, which may include, for example, 2D or 3D color images, IR images, near IR images, ultraviolet images, etc.

[0146] At block 818, the computing device processes one or more profile maps (e.g., a thickness profile map, a grain map, an optical constant map, a roughness map, a temperature map, etc.) using one or more trained machine learning models. In some embodiments, input to the machine learning models may include data from the one or more profile maps as well as data from one or more images of the substrate. Each of the trained machine learning models outputs one or more chamber part condition estimates for one or more chamber parts of the process chamber. The output may include image-level or component-level chamber part estimates.

[0147] In one embodiment, the output includes a profile map of the condition of the substrate during processing (e.g., a temperature map of the substrate during processing). Alternatively, such a profile map of the condition of the substrate during processing may have already been generated (or input into a trained machine learning model) in block 816 without a machine learning user.

[0148] In some cases, the output of the trained machine learning model includes information regarding the condition of one or more chamber components. In some embodiments, the information can include information regarding one or more components of the chamber components. In embodiments, the output can include estimated mesa conditions, lift pin location conditions, seal band conditions, and / or layer delamination conditions for the electrostatic chuck or other substrate support. For example, the output can include an indication of the detected mesa and its estimated wear height and / or amount, an indication of the estimated wear height and / or amount of the seal band region, an indication of the estimated location of lift pin hot spots (points that had a temperature exceeding a temperature threshold during processing), an indication of estimated areas where the electrostatic chuck plates are delaminating and / or an indication of the estimated amount of layer delamination in those areas, etc.

[0149] In one embodiment, the one or more profile maps are input into a trained machine learning model along with one or more images of the substrate. In some cases, the addition of images can improve the accuracy of the output of the trained machine learning model.

[0150] At block 820, the processing logic outputs a notification of one or more conditions of the chamber parts (e.g., the substrate support). This may include outputting 824 estimated mesa conditions (e.g., missing mesas, mesas with excessive wear exceeding a wear threshold, etc.), outputting 826 estimated lift pin conditions (e.g., estimated lift pin hot spots, etc.), outputting 828 estimated seal band conditions (e.g., spots on the seal band with excessive wear exceeding a wear threshold, etc.), outputting 828 estimated delamination conditions (e.g., areas of delamination and / or the extent or level of delamination in those areas, etc.), etc. In an embodiment, this output may be provided via a graphical user interface (GUI). In some embodiments, the output may be a message sent to a device (e.g., a user's mobile device) via text message, email, etc.

[0151] 9 is a flow diagram of a method 900 for determining a chamber condition of one or more chamber parts based on measurements of substrates processed by a process chamber over time, according to aspects of the disclosure. The method 900 is performed by processing logic, which may include hardware (circuitry, dedicated logic, etc.), software (such as running on a general-purpose computer system or a dedicated machine), firmware, or some combination thereof. In one embodiment, the method 900 may be performed by a computer system, such as the computer system architecture 100 of FIG. 1. In other or similar embodiments, one or more operations of the method 900 may be performed by one or more other machines not shown. In some aspects, one or more operations of the method 900 may be performed by the chamber part condition engine 330 of the server machine 320 described with respect to FIG. 3.

[0152] At block 910 of the method 900, a process chamber of a substrate processing system initially processes a first substrate according to a process recipe.

[0153] At block 916, the substrate measurement system may generate measurements of many locations on a surface of the substrate and may generate one or more first profile maps of the substrate from the multiple measurements. The first profile map may be or include, for example, a first thickness map, a first grain map, a first temperature map, a first dielectric constant map, a first optical constant map, a first roughness map, etc.

[0154] At block 918, the computing device may process the one or more first profile maps using one or more trained machine learning models. Each of the trained machine learning models outputs one or more first chamber part condition estimates for one or more chamber parts of the process chamber. The first output may include image-level or component-level chamber part estimates. In one embodiment, the first output includes a profile map of the condition of the substrate being processed (e.g., a temperature map of the substrate being processed). Alternatively, such a profile map of the condition of the substrate being processed may have already been generated (input to the trained machine learning models) at block 916 without the use of machine learning.

[0155] In some cases, a first output of the trained machine learning model includes information regarding the condition of one or more chamber components. In some embodiments, the information can include information regarding one or more components of the chamber components. In embodiments, the output can include estimated mesa conditions, lift pin location conditions, seal band conditions, and / or delamination conditions for an electrostatic chuck or other substrate support.

[0156] At block 920, a process chamber of the substrate processing system processes the second substrate a second time according to the process recipe.

[0157] At block 924, the substrate measurement system may generate measurements of many locations on a surface of the second substrate and may generate one or more second profile maps of the second substrate from the multiple measurements. The second profile maps may be or include, for example, a second thickness map, a second grain map, a second temperature map, a second dielectric constant map, a second optical constant map, a second roughness map, etc.

[0158] At block 928, the computing device may process the one or more second profile maps using one or more trained machine learning models. Each of the trained machine learning models may output one or more second chamber part condition estimates for one or more chamber parts of the process chamber. The second outputs may include image-level or component-level chamber part estimates.

[0159] In one embodiment, the second output includes a profile map of the condition of the substrate during processing (e.g., a temperature map of the substrate during processing). Alternatively, such a profile map of the condition of the substrate during processing may have already been generated in block 924 without the use of machine learning (and may have been input into a trained machine learning model).

[0160] In some cases, a second output of the trained machine learning model includes information regarding the condition of one or more chamber components. In some embodiments, the information can include information regarding one or more of the chamber components. In embodiments, the output can include estimated mesa conditions, lift pin location conditions, seal band conditions, and / or delamination conditions for an electrostatic chuck or other substrate support.

[0161] At block 930, the processing logic at least one of a) compares the second profile map to the first profile map, or b) compares the second machine learning output to the first machine learning output. Based on this comparison, the processing logic can determine the change in one or more chamber parts over time and / or the rate of change of one or more chamber parts. In some embodiments, prior to processing the second substrate, a number of additional substrates are processed at other times, the substrates are measured, profile maps of the additional substrates are generated, and the profile maps are processed by the trained machine learning model to generate additional outputs. The second output and / or profile map can be compared to the additional output and / or profile map, as well as the first output and / or profile map. This provides additional data points from which changes in erosion rate, flotation, wear rate, and / or other chamber component conditions over time can be determined. For example, the processing logic can determine whether the wear rate on one or more mesas is accelerating.

[0162] At block 932, processing logic makes one or more determinations regarding the chamber parts of the process chamber based on the comparison made in block 930. The determination can include, for example, one or more of an erosion rate of one or more chamber parts (block 920), a change in the erosion rate of one or more chamber parts (block 922), whether it is time to replace any chamber parts (block 924), a remaining life of one or more chamber parts (block 926), an expected yield of devices resulting from the second substrate (block 928), and / or future substrates to be processed using the process chamber, etc.

[0163] In one example, the substrate support may be replaced when the seal band thickness reaches a particular threshold. Based on the comparison performed in block 932, the processing logic may determine how fast the seal band on the substrate support is eroding. This erosion rate may be applied to the currently determined thickness of the seal band to extrapolate to the future years of use of the substrate support or the time when the seal band thickness will be below the threshold thickness. The process chamber may then be scheduled for maintenance such that the substrate support is replaced at the determined further years of use of the substrate support or the time when the seal band thickness is predicted to be below the threshold thickness.

[0164] 10 is a flow diagram of a method 1000 for determining the condition of one or more chamber components based on a comparison of data and a curve, according to aspects of the disclosure. The method 1000 is performed by processing logic, which may include hardware (circuitry, dedicated logic, etc.), software (such as executed on a general-purpose computer system or a dedicated machine), firmware, or some combination thereof. In an embodiment, the method 1000 may be performed by a computer system, such as the computer system architecture 100 of FIG. 1. In other or similar embodiments, one or more operations of the method 900 may be performed by one or more other machines not shown. In some aspects, one or more operations of the method 1000 may be performed by the chamber component condition engine 330 of the server machine 320 described with respect to FIG. 3.

[0165] At block 1010 of method 1000, a process chamber of a substrate processing system initially processes a substrate according to a process recipe. The processing logic may initially determine the age of the chamber parts. The age may be measured using exact hours of use (e.g., 100 days of active use or 1000 hours of use of processing substrates), number of substrates or lots processed (multiple groups of substrates processed together), or other metrics. The age may be determined for a single chamber part of interest (e.g., a substrate support or electrostatic chuck) or for multiple chamber parts (e.g., a substrate support and a process kit ring).

[0166] At block 1016, the substrate measurement system may generate measurements of many locations on a surface of the substrate and may generate one or more first profile maps of the substrate from the multiple measurements. The first profile map may be or include, for example, a first thickness map, a first grain map, a first temperature map, a first dielectric constant map, a first optical constant map, a first roughness map, etc.

[0167] At block 1018, the computing device may process the one or more first profile maps using one or more trained machine learning models. Each of the trained machine learning models may output one or more first chamber part condition estimates for one or more chamber parts of the process chamber. The output may include image-level or component-level chamber part estimates. In one embodiment, the output includes a profile map of the condition of the substrate being processed (e.g., a temperature map of the substrate being processed). Alternatively, such a profile map of the condition of the substrate being processed may have already been generated (and input to the trained machine learning models) at block 1016 without the use of machine learning.

[0168] In some cases, the output of the trained machine learning model includes information regarding the condition of one or more chamber components. In some embodiments, the information can include information regarding one or more components of the chamber components. In embodiments, the output can include estimated mesa conditions, lift pin location conditions, seal band conditions, and / or delamination conditions for an electrostatic chuck or other substrate support. The output can additionally or alternatively include estimated process kit ring conditions, showerhead conditions, etc.

[0169] At block 1020, the processing logic compares at least one of the profile map or the trained machine learning model output to a curve to determine where the data falls on the curve. The curve may have been generated based on measurements and / or machine learning model output of substrates processed by an ideal or golden process chamber over time, or may have been generated based on measurements and / or machine learning model output of substrates processed in multiple process chambers of the same type and / or performing the same recipe over time. The curve may indicate a range of expected chamber part conditions for the chamber parts for any given age of the chamber parts of a process chamber of that type and / or performing that recipe. The curve may include a single value with upper and / or lower deviation thresholds (e.g., standard deviation) for each age value.

[0170] At block 1024, based on the comparison, processing logic may determine whether the data deviates from the curve by more than a deviation threshold (e.g., outside an upper or lower deviation threshold). At block 1028, processing logic may generate an alert in response to determining that the data deviates from the curve by more than the deviation threshold. In an embodiment, processing logic may plot the current chamber component condition values ​​against the curve on a display to show a user how the chamber component compares to the curve.

[0171] In an embodiment, the method 1000 can be performed periodically on substrates processed by the process chamber and the results can be plotted against a curve. This can show a user how the chamber component conditions are changing over time compared to how they are expected to change over time. For example, if a particular side of the process kit ring and / or electrostatic chuck is eroding faster than expected, this can indicate that these components are out of alignment.

[0172] 11A is a flow diagram of a method 1100 of performing a quality assessment of a process chamber according to aspects of the disclosure. The method 1100 is performed by processing logic that may include hardware (circuitry, dedicated logic, etc.), software (such as executed on a general-purpose computer system or dedicated machine), firmware, or some combination thereof. In one embodiment, the method 1100 may be performed by a computer system such as the computer system architecture 100 of FIG. 1. In other or similar embodiments, one or more operations of the method 1100 may be performed by one or more other machines not shown. In some aspects, one or more operations of the method 1100 may be performed by the chamber part condition engine 330 of the server machine 320 described with respect to FIG. 3.

[0173] At block 1110 of the method 1100, a process chamber of a substrate processing system processes a first substrate according to a first process recipe. The first process recipe can be designed to maximize the variation of a film property to be measured at block 1116 based on the variation of the first chamber component conditions (e.g., RF field, temperature, etc.) during processing.

[0174] At block 1116, the substrate measurement system may generate measurements of many locations on a surface of the substrate and may generate one or more first profile maps of the substrate from the multiple measurements. The first profile map may be or include, for example, a first thickness map, a first grain map, a first temperature map, a first dielectric constant map, a first optical constant map, a first roughness map, etc.

[0175] At block 1118, the computing device may process the one or more first profile maps using one or more trained machine learning models. Each of the trained machine learning models outputs one or more first chamber part condition estimations for one or more chamber parts of the process chamber. The first output may include image-level or component-level chamber part estimations. In one embodiment, the first output includes information regarding the first condition of the one or more chamber parts. In some embodiments, the information may include information regarding one or more parts of the chamber part.

[0176] A process chamber of the substrate processing system may process a second substrate according to a second process recipe at block 1120. The second process recipe may be designed to maximize the variation of a film property to be measured at block 1122 based on the variation of second chamber component conditions (e.g., RF field, temperature, etc.) during processing.

[0177] At block 1122, the substrate measurement system may generate measurements of many locations on a surface of the second substrate and may generate one or more second profile maps of the second substrate from the multiple measurements. The second profile maps may be or include, for example, a second thickness map, a second grain map, a second temperature map, a second dielectric constant map, a second optical constant map, a second roughness map, etc.

[0178] At block 1124, the computing device may process the one or more second profile maps using one or more additional trained machine learning models (which may be different from the machine learning model used at block 1118). Each of the additional trained machine learning models outputs one or more second chamber part condition estimates for one or more chamber parts of the process chamber. The second output may include image-level or component-level chamber part estimates. In one example, the first recipe used to process the first substrate may generate a film that is sensitive to temperature variations and may be used to determine lift pin hot spots, electrostatic chuck delamination, mesa erosion, missing mesas, and / or seal band erosion. The second recipe used to process the second substrate may generate a film that is sensitive to RF field variations and / or plasma variations and may be used to determine process kit ring placement and / or process kit ring erosion.

[0179] At block 1130, processing logic performs a quality assessment of the process chamber based on at least one of the profile map, or the output of the one or more trained machine learning models, and / or the one or more additional trained machine learning models. Performing the quality assessment may include performing multiple tests on the chamber part condition estimation and / or comparing the chamber part estimation to multiple criteria. Different criteria may be associated with different chamber part condition tests.

[0180] In one embodiment, at block 1132, processing logic determines whether one or more lift pin criteria are met. The lift pin criteria may include a detected temperature at the lift pin location below a temperature threshold. In one embodiment, at block 1134, processing logic determines whether one or more mesa criteria are met. For example, the mesa criteria may not be met if a missing mesa is identified and / or if a mesa having an erosion amount greater than an erosion threshold is detected and / or if a threshold number of mesas having an erosion amount greater than an erosion threshold (e.g., a thickness less than a thickness threshold) are detected. In one embodiment, at block 1136, processing logic determines whether one or more seal band criteria are met. The seal band criteria may not be met if a chip or crack is detected in the seal band and / or if a location on the seal band having an erosion amount greater than an erosion threshold (e.g., a thickness less than a thickness threshold) is detected. In one embodiment, at block 1138, processing logic determines whether one or more delamination criteria are met. If areas of delamination are detected between plates or layers of the substrate support or electrostatic chuck, the layer delamination criteria may not be met. In one embodiment, at block 1140, processing logic determines whether one or more process kit ring criteria are met. The process kit ring criteria may not be met if the process kit ring is not centered about the substrate support (e.g., the process kit ring is off center by more than a threshold amount) or if a threshold amount of erosion is detected for the process kit ring.

[0181] At block 1150, processing logic generates a quality assessment report based on the results of the quality assessment performed at block 1130. The quality assessment report may indicate whether the process chamber passed the quality assessment and / or whether any particular chamber part within the process chamber passed the quality assessment. If any chamber part fails to meet the quality assessment, processing logic may output an indication of why the quality assessment failed for that chamber part (e.g., what criteria were not met) and / or the extent to which the chamber part failed to meet the criteria.

[0182] Method 1100 may be performed periodically, in embodiments, and / or as part of a recertification procedure after a maintenance event. If all tests and / or criteria of the qualification are successfully passed, recertification of the process chamber may be confirmed.

[0183] 11B is a flow diagram of a method 1150 for determining how a chamber part is installed in a process chamber according to aspects of the disclosure. The method 1150 is performed by processing logic that may include hardware (circuitry, dedicated logic, etc.), software (such as running on a general-purpose computer system or a dedicated machine), firmware, or some combination thereof. In an embodiment, the method 1150 may be performed by a computer system such as the computer system architecture 100 of FIG. 1. In other or similar embodiments, one or more operations of the method 1150 may be performed by one or more other machines not shown. In some aspects, one or more operations of the method 1150 may be performed by the chamber part condition engine 330 of the server machine 320 described with respect to FIG. 3.

[0184] At block 1152 of the method 1150, a process chamber of a substrate processing system processes a substrate in the process chamber according to a first process recipe.

[0185] At block 1154, the substrate measurement system may generate measurements of many locations on a surface of the substrate and may generate one or more first profile maps of the substrate from the multiple measurements. The first profile map may be or include, for example, a first thickness map, a first grain map, a first temperature map, a first dielectric constant map, a first optical constant map, a first roughness map, etc.

[0186] At block 1156, the computing device may process the one or more first profile maps using one or more trained machine learning models. Each of the trained machine learning models outputs one or more chamber part condition estimations for one or more chamber parts of the process chamber. The output may include image-level or component-level chamber part estimations. In some embodiments, the output may include information about one or more parts of the chamber part.

[0187] At block 1160, the processing logic determines one or more installation parameters of the chamber component based on the output. In some cases, the output of the machine learning model indicates a location or point of installation (e.g., an offset of the process kit ring from a center of the electrostatic chuck around which the process kit ring is disposed, etc.). Other installation parameters that can be detected include, for example, the degree to which one or more fixtures or bolts are tightened.

[0188] At block 1162, processing logic can determine whether the chamber component was correctly installed based on the determined installation parameters for the chamber component. In one example, the process kit ring criteria can include a threshold offset from center. If the process kit ring has an offset that is greater than the threshold offset, it can be determined that the process kit ring was not correctly installed.

[0189] At block 1164, the processing logic can adjust a recipe to correspond to the determined installation parameters for future process runs. For example, the RF field and / or temperature can be adjusted at one or more regions of the substrate support based on the offset of the process kit ring. Additionally or alternatively, the processing logic can output a notification to reinstall the chamber parts. In some embodiments, the processing logic automatically schedules maintenance for the process chamber. In some cases, the process chamber may already be offline for maintenance, and method 1150 can be performed as part of a qualification procedure. If the chamber parts were not correctly installed, further maintenance can be performed (e.g., by reinstalling the chamber parts, etc.) before requalifying the process chamber.

[0190] FIG. 12 is a profile map 1200 of a processed substrate according to an embodiment of the disclosure. The profile map 1200 is a heat map showing the temperature at different locations on a substrate during processing in a process chamber. The temperature can be based on the thickness at different locations of a film deposited or etched during processing. A key 1202 is provided to show how to interpret the profile map 1200. As shown, hot spots 1205 were provided at each lift pin location, which had a higher temperature during processing. Also shown, there was a hot spot 1210 around the center of the substrate. In addition, a region 1215 on the right side of the substrate was warmer than other regions around the substrate, which may indicate, for example, delamination on the right side of the substrate support.

[0191] FIG. 13 is a profile map 1300 of a portion of a processed substrate, according to an embodiment of the disclosure. The profile map 1300 is a heat map showing the temperature at different locations on the substrate during processing in a process chamber. The temperature can be based on the thickness at different locations of the film deposited or etched during processing. It is noted that other profile maps can represent non-thermal properties such as RF field strength, plasma strength or amount, film thickness, etc. A key 1302 is provided to show how to interpret the profile map 1300. As shown, the area around the substrate 1310 was significantly cooler than the rest of the substrate during processing. In addition, a number of localized spots 1305 of reduced temperature are shown, each of which represents a mesa of the substrate support. The points where the mesa contacted the substrate may have improved thermal contact and therefore may have a lower temperature due to cooling by the substrate support during processing. Based on such a profile map, the processing logic can determine whether any mesas are missing (e.g., based on identifying that there are no localized cold spots where a mesa would be expected). Such information can be extracted by image processing algorithms or trained machine learning models to determine, for example, whether any mesas are missing from the electrostatic chuck.

[0192] FIG. 14 illustrates a diagrammatic representation of a machine in the exemplary form of a computing device 1400 capable of executing a set of instructions to cause the machine to perform any one or more of the techniques discussed herein. In alternative embodiments, the machine may be connected (e.g., networked) to other machines in a local area network (LAN), an intranet, an extranet, or the Internet. The machine may operate in the capacity of a server or a client machine in a client-server network environment, or as a peer machine in a peer-to-peer (or distributed) network environment. The machine may be a personal computer (PC), a tablet computer, a set-top box (STB), a personal digital assistant (PDA), a cellular phone, a web appliance, a server, a network router, a switch or bridge, or any machine capable of executing (sequentially or otherwise) a set of instructions that specify actions to be taken by the machine. Further, although only a single machine is shown, the term "machine" should also be construed to include any group of machines (e.g., computers) that individually or jointly execute a set (or sets) of instructions to perform any one or more of the techniques discussed herein. In an embodiment, computing device 1400 may correspond to one or more of server machine 170, server machine 180, prediction server 112, system controller 228, server machine 320, or server machine 350, as described herein.

[0193] The exemplary computing device 1400 includes a processing device 1402, a main memory 1404 (e.g., read only memory (ROM), flash memory, dynamic random access memory (DRAM), such as synchronous DRAM (SDRAM), etc.), a static memory 1406 (e.g., flash memory, static random access memory (SRAM), etc.), and a secondary memory (e.g., a data storage device 1428), which communicate with each other via a bus 1408.

[0194] The processing device 1402 may represent one or more general-purpose processors, such as a microprocessor, a central processing unit, etc. More specifically, the processing device 1402 may be a complex instruction set computing (CISC) microprocessor, a reduced instruction set computing (RISC) microprocessor, a very long instruction word (VLIW) microprocessor, a processor implementing other instruction sets, or a processor implementing a combination of instruction sets. The processing device 1402 may also be one or more special-purpose processing devices, such as an application specific integrated circuit (ASIC), a field programmable gate array (FPGA), a digital signal processor (DSP), a network processor, etc. The processing device 1402 may also be or include a system on a chip (SoC), a programmable logic controller (PLC), or other type of processing device. The processing device 1402 is configured to execute processing logic for performing the operations and steps discussed herein.

[0195] The computing device 1400 may further include a network interface device 1422 for communicating with a network 1464. The computing device 1400 may also include a video display unit 1410 (e.g., a liquid crystal display (LCD) or a cathode ray tube (CRT)), an alphanumeric input device 1412 (e.g., a keyboard), a cursor control device 1414 (e.g., a mouse), and a signal generation device 1420 (e.g., a speaker).

[0196] The data storage device 1428 may include a machine-readable storage medium (or more specifically, a non-transitory computer-readable storage medium) 1424 having stored thereon one or more sets of instructions 1426 that implement any one or more of the techniques or functions described herein. For example, the instructions 1426 may include instructions for the chamber component condition engine 330. A non-transitory storage medium refers to a storage medium other than a carrier wave. The instructions 1426 may also reside, completely or at least partially, within the main memory 1404 and / or within the processing device 1402 during execution thereof by the computing device 1400, the main memory 1404, and the processing device 1402, which also constitute computer-readable storage media.

[0197] While in the exemplary embodiment, computer-readable storage medium 1424 is shown as a single medium, the term "computer-readable storage medium" should be interpreted 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 instructions. The term "computer-readable storage medium" should also be interpreted to include any medium capable of storing or encoding a set of instructions for execution by a machine that causes the machine to perform any one or more of the techniques of this disclosure. Thus, the term "computer-readable storage medium" should be interpreted to include, but is not limited to, solid-state memories and optical and magnetic media.

[0198] The above description describes numerous specific details, such as examples of specific systems, components, methods, etc., to provide a good understanding of some embodiments of the present disclosure. However, it will be apparent to those skilled in the art that at least some embodiments of the present disclosure can be implemented without these specific details. In other instances, well-known components or methods have not been described in detail or have been presented in a simple block diagram format to avoid unnecessarily obscuring the present disclosure. Thus, the specific details described are merely exemplary. It is contemplated that certain embodiments may vary from these exemplary details and still be within the scope of the present disclosure.

[0199] Throughout this specification, a reference to "one embodiment" or "an embodiment" means that a particular feature, structure, or characteristic described in connection with that embodiment is included in at least one embodiment. Thus, the appearances of "in one embodiment" or "in an embodiment" in various locations throughout this specification do not necessarily all refer to the same embodiment. In addition, the term "or" is intended to mean an inclusive "or" rather than an exclusive "or". When the term "about" or "approximately" is used herein, it is intended to mean that the nominal value presented is accurate to within ±10%.

[0200] Although the method operations herein are illustrated and described in a particular order, the order of the method operations may be changed, such that certain operations may be performed in reverse order, and certain operations may be performed at least in part concurrently with other operations. In alternative embodiments, separate operation instructions or sub-operations may be performed intermittently and / or alternately.

[0201] It is to be understood that the above description is intended to be illustrative, and not limiting. Many other embodiments will be apparent to those skilled in the art upon reading and understanding the above description. The scope of the present disclosure should therefore be determined with reference to the appended claims, along with the full scope of equivalents to which such claims are entitled.

Claims

1. Processing a substrate in a process chamber of a substrate processing system, the substrate comprising at least one of a film or a feature after the processing; generating a profile map of at least one of the film or the feature on the substrate using a substrate metrology system of the substrate processing system; processing data from the profile map using a first trained machine learning model, the first trained machine learning model outputting a first chamber part condition estimate for a first chamber part of the process chamber; determining whether to perform maintenance on the first chamber part of the process chamber based at least in part on the first chamber part condition estimate; The method includes:

2. 2. The method of claim 1 , wherein the first chamber part comprises a substrate support, and the first chamber part estimation includes at least one of an estimated mesa erosion condition of the substrate support, an estimated seal band erosion condition of the substrate support, or an estimated lift pin location erosion condition of the substrate support.

3. the estimated mesa erosion condition includes an estimate of an amount of erosion for one or more mesas of the substrate support; the estimated seal band erosion condition includes an estimate of erosion for one or more portions of the seal band of the substrate support; the estimated lift pin site erosion condition includes an estimate of erosion for each lift pin site of the substrate support; The method of claim 2.

4. 2. The method of claim 1 , wherein the first chamber part comprises a substrate support, and the first chamber part estimation includes at least one of an estimation of recesses in a surface of the substrate support, an estimation of roughness across the surface of the substrate support, an estimation of planarity of the surface of the substrate support, or an estimation of concentricity of one or more circular elements of the substrate support.

5. The method of claim 1 , wherein the profile map of the film comprises a thickness profile map of the film.

6. 2. The method of claim 1, wherein the first chamber part comprises a showerhead, and the first chamber part estimation includes an estimated gas delivery to each of a plurality of regions of the showerhead.

7. determining a probability that the first chamber part will cause a reduction in product quality based on the first chamber part estimation; determining whether to perform the maintenance on the first chamber part of the process chamber based on the probability that the first chamber part will cause the reduction in product quality; The method of claim 1 further comprising:

8. estimating a fault time of the first chamber component based at least in part on the first chamber component condition estimate; and determining when to perform the maintenance on the first chamber part of the process chamber based on the estimated downtime; The method of claim 1 further comprising:

9. processing the data from the profile map using a second trained machine learning model, the second trained machine learning model outputting a chamber part condition estimate for a second chamber part of the process chamber; determining whether to perform maintenance on the second chamber part of the process chamber based at least in part on the chamber part condition estimate for the second chamber part; The method of claim 1 further comprising:

10. training a first machine learning model to produce the first trained machine learning model, wherein the first machine learning model is trained using data from a plurality of process chambers that share a common process chamber type. The method of claim 1 further comprising:

11. adjusting the first trained machine learning model using additional data from the process chamber prior to processing the data from the profile map. The method of claim 10 further comprising:

12. The method of claim 1 , wherein the substrate is a blanket wafer.

13. a process chamber for processing the substrate, the substrate comprising at least one of the film or feature after processing; one or more robots for moving the substrate from the process chamber to a substrate measurement system; a substrate measurement system for measuring at least one of the film or features on the substrate and generating a profile map of the film or at least one of the features; a computing device, the computing device comprising: processing data from the profile map using a first trained machine learning model, the first trained machine learning model outputting a first chamber part condition estimate for a first chamber part of the process chamber; and determining whether to perform maintenance on the first chamber part of the process chamber based at least in part on the first chamber part condition estimate. Substrate processing system.

14. 14. The substrate processing system of claim 13, wherein the first chamber part comprises a chuck, the profile map comprises a thickness profile map, and the first chamber part estimation comprises at least one of an estimated mesa erosion condition of the chuck, an estimated seal band erosion condition of the chuck, or an estimated lift pin site erosion condition of the chuck.

15. the estimated mesa erosion condition includes an estimate of an amount of erosion for one or more mesas of the chuck; the estimated seal band erosion condition includes an estimate of erosion for one or more portions of the seal band of the chuck; the estimated lift pin site erosion condition includes an estimate of erosion for each lift pin site of the chuck. The substrate processing system of claim 14.

16. 14. The substrate processing system of claim 13, wherein the first chamber part comprises a showerhead, and the first chamber part estimation comprises an estimated gas delivery to each of a plurality of regions of the showerhead.

17. the computing device further comprising: determining a probability that the first chamber part will cause a reduction in product quality based on the first chamber part estimation; and determining whether to perform the maintenance on the first chamber part of the process chamber based on the probability that the first chamber part will cause the reduction in product quality.

18. the computing device further comprising: estimating a fault time of the first chamber component based at least in part on the first chamber component condition estimate; and and determining when to perform the maintenance on the first chamber part of the process chamber based on the estimated downtime.

19. the computing device further comprising: processing the data from the profile map using a second trained machine learning model, the second trained machine learning model outputting a chamber part condition estimate for a second chamber part of the process chamber; and determining whether to perform maintenance on the second chamber part of the process chamber based at least in part on the chamber part condition estimate for the second chamber part.

20. 1. A computer-readable medium comprising instructions that, when executed by a processing device, cause the processing device to perform an operation, the operation comprising: receiving a thickness profile map of a film on a substrate, the substrate being processed in a process chamber of a substrate processing system, the thickness profile map being generated by a substrate metrology system of the substrate processing system after the substrate is processed in the process chamber; processing data from the thickness profile map using a first trained machine learning model, the first trained machine learning model outputting a first chamber part condition estimate for a first chamber part of the process chamber; and determining whether to perform maintenance on the first chamber part of the process chamber based at least in part on the first chamber part condition estimate.

21. processing a first substrate in a process chamber of a substrate processing system according to a recipe, the first substrate comprising at least one of a film or a feature after the processing; generating a profile map of at least one of the film or the feature on the first substrate using a substrate metrology system of the substrate processing system; processing data from the profile map using a first model, the first model outputting at least one of an estimated mesa condition of a substrate support for the process chamber, an estimated lift pin location condition of the substrate support, an estimated seal band condition of the substrate support, or an estimated process kit ring condition for a process kit ring for the process chamber; outputting a notification of at least one of the estimated mesa condition, the estimated lift pin location condition, the estimated process kit ring condition, or the estimated seal band condition of the substrate support as a result of the processing; The method includes:

22. determining whether to perform maintenance on the substrate support based at least in part on at least one of the estimated mesa condition, the estimated lift pin location condition, or the estimated seal band condition; 22. The method of claim 21 further comprising:

23. 22. The method of claim 21 , wherein the first model outputs the estimated mesa condition, the estimated mesa condition including at least one of an estimate of an amount of erosion for one or more mesas of the substrate support, an estimate of missing mesas, an estimate of a mesa pattern, or an estimate of the extent to which the estimated mesa pattern deviates from a target mesa pattern.

24. 22. The method of claim 21 , wherein the first model outputs the estimated seal band condition, the estimated seal band condition including at least one of an estimate of erosion for one or more portions of the seal band of the substrate support, or an estimate of concentricity of the seal band.

25. 22. The method of claim 21 , wherein the first model outputs the estimated lift pin location conditions, the estimated lift pin location conditions including at least one of an estimate of hot spots at one or more locations of the substrate support, an estimate of a lift pin location, or an estimate of deviation of the estimated lift pin location from a target lift pin location.

26. the first model outputs the process kit ring condition estimate for the process kit ring, the process kit ring at least one of a) surrounding the substrate support, or b) underlying the substrate support, and the estimated process kit ring condition includes at least one of an estimate of erosion on the process kit ring, an estimate of defects in the process kit ring, or an estimate of a missing process kit ring, and the method further comprises: The method of claim 21 , further comprising determining whether to replace the process kit ring based on the process kit ring estimation.

27. processing a second substrate in the process chamber, the second substrate including at least one of a second film or a second feature after the processing; generating a second profile map of at least one of the second film or the second feature on the second substrate using the substrate metrology system of the substrate processing system; processing data from the second profile map using the model, wherein the model outputs at least one of a second estimated mesa condition of the substrate support, a second estimated lift pin location condition of the substrate support, a second estimated seal band condition of the substrate support, or a second estimated process kit ring condition of the process kit ring; and further comprising at least one of: a) comparing the second estimated mesa condition to the estimated mesa condition; b) comparing the second estimated lift pin site condition to the estimated lift pin site condition; c) comparing the second estimated seal band condition to the estimated seal band condition; or d) comparing the second estimated process kit ring condition to the estimated process kit ring condition.

22. The method of claim 21.

28. determining an erosion rate for at least one of: a) one or more mesas of the substrate support, b) one or more lift pin locations of the substrate support, c) one or more seal band portions of the substrate support, or d) the process kit ring based on results of the comparison.

28. The method of claim 27, further comprising:

29. Based on the results of the comparison, a) whether it is time to replace a chamber part of the process chamber; b) the remaining life of the chamber component; or c) the expected yield performance of devices produced from said second substrate; 30. The method of claim 27, further comprising determining at least one of:

30. determining a quality of the substrate support based on at least one of the estimated mesa condition, the estimated lift pin location condition relative to where the substrate was placed in the process chamber, or the estimated seal band condition.

22. The method of claim 21 further comprising:

31. determining whether one or more mesas are damaged, missing, or improperly positioned based on the estimated mesa condition; determining that the substrate support did not pass a quality test in light of the one or more mesas being broken, missing, or improperly positioned; 31. The method of claim 30, further comprising:

32. Determining at least one of: a) when to replace the process kit ring based on the estimated process kit ring condition; or b) a location of the process kit ring about the substrate support.

22. The method of claim 21 further comprising:

33. 22. The method of claim 21 , wherein the substrate support comprises a first component, a second component, and a bond between the first component and the second component, and the first model or the second model further outputs a bond condition estimate for the bond indicating a degree of failure of the bond and whether delamination of the second component from the first component has occurred.

34. The method of claim 21 , wherein the profile map of the film comprises a thickness profile map of the film.

35. determining a probability that the substrate support will cause a reduction in product quality based on at least one of the estimated mesa condition, the estimated lift pin location condition, or the estimated seal band condition of the substrate support; determining whether to perform maintenance on the substrate support based on the probability that the substrate support will cause the reduction in product quality; and 22. The method of claim 21 further comprising:

36. estimating a failure time of the substrate support based at least in part on at least one of the estimated mesa condition, the estimated lift pin location condition, the estimated seal band condition, or the estimated process kit ring condition; and determining when to perform maintenance on at least one of the substrate support or the process kit ring based on the estimated downtime; 22. The method of claim 21 further comprising:

37. the first model is a first trained machine learning model, and the method further comprises: training a first machine learning model to produce the first trained machine learning model, wherein the first machine learning model is trained using data from a plurality of process chambers that share a common process chamber type.

22. The method of claim 21 further comprising:

38. the first model is a first trained machine learning model, and the method further comprises: adjusting the first trained machine learning model using additional data from the process chamber prior to processing the data from the profile map.

22. The method of claim 21 further comprising:

39. generating one or more images of the first substrate using an imaging device; inputting the one or more images into the first model together with the profile map; 22. The method of claim 21 further comprising:

40. determining one or more adjustments to at least one of a process recipe or tool parameter settings to compensate for at least one of the estimated mesa condition, the estimated lift pin location condition, the estimated seal band condition, or the estimated process kit ring condition.

22. The method of claim 21 further comprising:

41. 22. The method of claim 21, wherein at least one of the substrate support or the process kit ring is a new component that has not yet been used to process a product substrate.

42. 1. A computer-readable medium comprising instructions that, when executed by a processing device, cause the processing device to perform an operation, the operation comprising: receiving a profile map of at least one of a film or feature on a first substrate measured using a substrate measurement system after the first substrate is processed in a process chamber; processing data from the profile map using a first trained machine learning model, the first trained machine learning model outputting at least one of an estimated mesa condition of a substrate support for the process chamber, an estimated lift pin location condition of the substrate support, an estimated seal band condition of the substrate support, or an estimated process kit ring condition for a process kit ring for the process chamber; outputting a notification of at least one of the estimated mesa condition, the estimated lift pin location condition, the estimated process kit ring condition, or the estimated seal band condition of the substrate support as a result of the processing; 13. A computer readable medium comprising:

43. The operation, 43. The computer readable medium of claim 42, further comprising determining whether to perform maintenance on the substrate support based at least in part on at least one of the estimated mesa condition, the estimated lift pin location condition, the estimated seal band condition, or the estimated process kit ring condition.

44. The operation, 43. The computer readable medium of claim 42, further comprising determining one or more adjustments to a process recipe to compensate for at least one of the estimated mesa condition, the estimated lift pin location condition, the estimated seal band condition, or the estimated process kit ring condition.

45. 1. A computer-readable medium comprising instructions that, when executed by a processing device, cause the processing device to perform an operation, the operation comprising: receiving a first profile map of at least one of a film or feature on a first substrate, the first profile map being generated by a substrate metrology system after the first substrate is initially processed by a process chamber according to a recipe; receiving a second profile map of at least one of a film or feature on a second substrate, the second profile map being generated by the substrate metrology system after the second substrate is processed by the process chamber according to the recipe for a second time; comparing the second profile map to the first profile map; and determining at least one of: a) an estimated mesa condition for one or more mesas of a substrate support in the process chamber, b) an estimated lift pin location condition for one or more lift pins of the substrate support, or c) an estimated seal band condition for a seal band of the substrate support based on results of the comparison.

Images