Substrate process endpoint detection using machine learning

Abstract

Methods and systems for detecting an endpoint of a substrate process are provided. A set of machine learning models is trained to provide metrology measurement values associated with a particular type of metrology measurement of a substrate based on spectral data collected for the substrate. In view of a performance rating associated with the particular type of metrology measurement, a corresponding machine learning model is selected for application to future spectral data collected during a future substrate process for a future substrate. Current spectral data is collected during a current process of a current substrate and provided as input to a respective machine learning model. An indication of a respective metrology measurement corresponding to the current substrate is extracted from one or more outputs of the trained machine learning model. In response to determining that the respective metrology measurement satisfies a metrology measurement criterion, an indication is generated that includes a command to terminate the current process.

Description

This application is a divisional application with a filing date of June 9, 2022, application number 202280041007.0, and titled “Substrate Process Endpoint Detection Using Machine Learning”. Technical Field

[0001] Embodiments of the present disclosure relate generally to manufacturing systems and, more particularly, to substrate process endpoint detection using machine learning. Background Art

[0002] Manufacturing microelectronics and integrated circuit devices generally involves performing a large number of operations on semiconductors, dielectrics, and conductive substrates. In some cases, single crystals, semiconductor films, found coatings, and many other materials are used in electronic device manufacturing and other practical applications. As the atoms of the selected type are added to the substrate (e.g., via deposition) or removed from the substrate (e.g., via etching), efficient and accurate endpoint monitoring techniques and systems become valuable. Insufficient processing (e.g., insufficient deposition, insufficient etching) and over-processing (e.g., excessive deposition, over-processing) may result in substandard and malfunctioning devices. Therefore, an optical control system that allows real-time monitoring of the various stages of device manufacturing can significantly improve the quality of the product, and considering the ever-increasing requirements for high-quality semiconductor devices, this optical control system is particularly useful. Summary of the invention

[0003] Some of the embodiments encompass a method for training a machine learning model to provide metrology measurements for a current substrate being processed according to a current process at a first manufacturing system. The method includes obtaining a set of historical spectral data associated with a previous substrate processed according to a previous process at a second manufacturing system. The set of historical spectral data includes historical spectral data collected for the previous substrate during each step in a set of steps of the previous process. The method further includes generating a regularized set of historical spectral data based on the set of historical spectral data. The regularized set of historical spectral data is generated based on a difference between first historical spectral data collected during an initial step in the set of steps of the previous process and second historical spectral data collected during one or more subsequent steps in the set of steps. The method further includes generating training data for the machine learning model. Generating the training data includes generating a training input, the training input including a subset of historical spectral data extracted from the regularized set of historical spectral data. The subset of historical spectral data includes an indication of one or more historical spectral features associated with a particular type of metrology measurement. Generating the training data further includes generating a target output for the training input. The target output includes a historical metrology measurement value of the previous substrate, the historical metrology measurement value being associated with the particular type of metrology measurement. The method further includes: providing the training data regarding the following items to train the machine learning model: (i) a set training input including the training input, and (ii) a set of target outputs including the target output.

[0004] In some embodiments, a system includes a memory for storing a trained machine learning model, and a processing device coupled to the memory. The processing device is used to receive a set of spectral data associated with a current substrate being processed according to a current process at a manufacturing system. The received set of spectral data is associated with a current step of the current process performed on the current substrate. The processing device is further used to generate a regularized set of spectral data based on the received set of spectral data. The regularized set of spectral data is generated based on the difference between spectral data collected during the current step of the current process and spectral data collected during a previous step of the current process. The processing device is further used to provide a subset of spectral data extracted from the regularized set of spectral data as an input to the trained machine learning model. The subset of spectral data includes an indication of one or more spectral features corresponding to a particular type of metrology measurement. The processing device is further used to obtain one or more outputs of the trained machine learning model. The processing device is further configured to extract, from the one or more outputs, metrology measurement data identifying one or more metrology measurement values ​​associated with the particular type of metrology measurement and an indication of a confidence level that each of the one or more metrology measurement values ​​corresponds to the current substrate, the one or more metrology measurement values ​​being obtained for a previous substrate processed according to a previous process at the manufacturing system.

[0005] In some embodiments, a non-transitory computer-readable storage medium includes instructions that, when executed by a processing device, cause the processing device to receive a set of spectral data associated with a current substrate being processed at a manufacturing system according to a current process. The received set of spectral data is associated with a current step of the current process performed on the current substrate. The processing device is further used to generate a regularized set of spectral data based on the received set of spectral data. The regularized set of spectral data is generated based on the difference between spectral data collected during the current step of the current process and spectral data collected during a previous step of the current process. The processing device is further used to provide a subset of spectral data extracted from the regularized set of spectral data as input to the trained machine learning model. The subset of spectral data includes an indication of one or more spectral features corresponding to a particular type of metrology measurement. The processing device is further used to obtain one or more outputs of the trained machine learning model. The processing device is further configured to extract, from the one or more outputs, metrology measurement data identifying one or more metrology measurement values ​​associated with the particular type of metrology measurement and an indication of a confidence level that each of the one or more metrology measurement values ​​corresponds to the current substrate, the one or more metrology measurement values ​​being obtained for a previous substrate processed according to a previous process at the manufacturing system.

[0006] In some embodiments, a method is provided, the method comprising: providing training data to train each of a set of machine learning models to provide a metrology measurement value associated with a particular type of metrology measurement of a substrate based on spectral data collected during a substrate process performed on the substrate. Each of the set of machine learning models is associated with a different type in a set of machine learning model types. The method further comprises: assigning a performance rating to each of the set of machine learning models based on the accuracy of the provided value of the metrology measurement provided by the corresponding machine learning model in view of the measured value of the metrology measurement, the measured value being generated based on historical metrology data collected by the metrology equipment for a previous substrate in a previous set of substrates. The method further comprises: selecting the corresponding machine learning model to be applied to future spectral data, the future spectral data being collected during a future substrate process performed on a future substrate, in view of the performance rating of each of the set of machine learning models.

[0007] In some embodiments, a system includes a memory and a processing device coupled to the memory. The processing device is used to obtain a machine learning model, the machine learning model being trained to provide a current value of a metrology measurement based on current spectral data collected during a current process of a current substrate performed at a manufacturing system. The trained machine learning model is associated with a performance rating, and the performance rating satisfies a performance criterion associated with a type of the metrology measurement. The processing device is further used to provide the current spectral data as an input to the trained machine learning model. The processing device is further used to extract an indication of a corresponding metrology measurement value corresponding to the current substrate from one or more outputs of the trained machine learning model. The corresponding metrology measurement value is associated with the type of the metrology measurement. The processing device is further used to generate an instruction in response to determining that the corresponding metrology measurement value satisfies the metrology measurement criterion associated with the current process, the instruction comprising a command to terminate the current process at the manufacturing system.

[0008] In some embodiments, a non-transitory computer-readable storage medium comprising instructions comprises instructions that, when executed by a processing device, cause the processing device to: provide training data to train each of a set of machine learning models to provide a metrology measurement value associated with a particular type of metrology measurement of a substrate based on spectral data collected during a substrate process performed on the substrate. Each of the set of machine learning models is associated with a different type in a set of machine learning model types. The processing device is further configured to assign a performance rating to each of the set of machine learning models based on an accuracy of the provided value of the metrology measurement provided by the corresponding machine learning model in view of the measured value of the metrology measurement. The measured value is generated based on historical metrology data collected by the metrology equipment for a previous substrate in a previous set of substrates. The processing device is further configured to select the corresponding machine learning model to be applied to future spectral data, the future spectral data being collected during a future substrate process performed on a future substrate, in view of the performance rating of each of the set of machine learning models. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] The present disclosure is illustrated by way of example and not limitation in the diagrams of the accompanying drawings in which like reference numerals represent similar elements. It should be noted that different references to "one" or "an" embodiment in the present disclosure do not necessarily refer to the same embodiment, and such references refer to at least one.

[0010] Figure 1 An illustrative computer system architecture in accordance with aspects of the present disclosure is depicted.

[0011] Figure 2is a schematic top view of an example manufacturing system according to aspects of the present disclosure.

[0012] Figure 3 An illustrative system architecture for endpoint detection of a substrate process in accordance with aspects of the present disclosure is depicted.

[0013] Figure 4 is a flowchart of a method for training a machine learning model according to aspects of the present disclosure.

[0014] Figure 5 is a schematic cross-sectional side view of an endpoint detection setup according to aspects of the present disclosure.

[0015] Figure 6 is a flowchart of a method for obtaining data for use with a machine learning model according to aspects of the present disclosure.

[0016] Figure 7A-7F Obtaining data for use with a machine learning model according to aspects of the present disclosure is described.

[0017] Figure 8 is a flow chart of a method for estimating a metrology value of a profile of a substrate using a machine learning model according to aspects of the present disclosure.

[0018] Fig. 9 is a flow chart of a method for selecting a machine learning model for estimating a certain type of metrology measurement according to aspects of the present disclosure.

[0019] Fig.10 is a flow chart of a method for detecting an endpoint of a substrate process using a machine learning model according to aspects of the present disclosure.

[0020] Fig.11 A block diagram of an illustrative computer system that operates in accordance with one or more aspects of the present disclosure is depicted. DETAILED DESCRIPTION

[0021] Embodiments of the present disclosure relate to an endpoint detection system for enhancing spectral data collection. A substrate process (e.g., a deposition process, an etching process, etc.) can be performed on a substrate at a process chamber of a manufacturing system. The endpoint of a substrate process refers to a point in the process at which the profile of the substrate corresponds to (i.e., matches or substantially matches) a target substrate profile. For example, a mask including a pattern of a target substrate profile can be used during an etching process of a substrate (such as a silicon wafer). The mask can be placed on the surface of the wafer and exposed to a reactive (e.g., wet or dry etching) environment to remove portions of the substrate that are not protected by the mask. The endpoint of an etching process refers to a point in the etching process at which the profile of the substrate corresponds to the pattern of the target substrate profile provided by the mask.

[0022] Deviations from the process sequence may result in changes in the rate and / or uniformity of the substrate process. For example, changes in the etching environment or differences in the photomask pattern may result in changes in the rate and uniformity of the etching, including changes across the entire substrate surface and between etching processes for multiple substrates. Tracking and responding to such changes involves precise and adjustable optical endpoint systems that are capable of collecting accurate and large amounts of optical response data that characterize the substrate surface at various time periods during the substrate process. The ever-shrinking size of microelectronic devices, the increasingly complex design of photomasks, and the increasing demands for device uniformity have further driven accuracy goals. Existing endpoint detection systems are generally unable to meet this increased technical demand.

[0023] Aspects and embodiments of the present disclosure address this and other shortcomings of conventional techniques by methods and systems for using machine learning to perform endpoint detection of a substrate process. Endpoint detection equipment may be coupled to or disposed together with a process chamber and may be configured to collect spectral data of a substrate during a substrate process. Spectral data refers to data associated with the intensity of energy waves (i.e., the intensity or amount of energy) detected for each wavelength of light reflected from the surface of the substrate. The endpoint detection equipment may transmit the collected spectral data to a processing device of a manufacturing system including the process chamber (e.g., a system controller of the manufacturing system). The processing device may generate regularized spectral data based on the spectral data collected by the endpoint detection equipment, and subsets of spectral data may be extracted from the regularized spectral data, the subsets comprising spectral features associated with a particular type of metrology measurement (e.g., critical dimensions, etc.).

[0024] A processing device may provide a subset of the spectral data as input to a machine learning model, the machine learning model being trained to provide metrological measurement values ​​associated with the particular type of metrological measurement of the substrate over the time period in which the spectral data was collected. A machine learning model may be selected from a set of trained machine learning models to be applied to the spectral data of the substrate, each trained machine learning model being associated with a different machine learning model type. A machine learning model may be selected as a trained machine learning model, the performance rating of the trained machine learning model satisfying performance criteria regarding the particular type of metrological measurement. In some embodiments, the machine learning model is trained and selected for use by a system controller. In other or similar embodiments, the machine learning model is trained and / or selected for use by another processing device separate from the system controller of the manufacturing system (e.g., located at an offline server that is not accessible to users of the manufacturing system).

[0025] The processing device may obtain one or more outputs of the trained machine learning model and extract metrology measurements provided for the substrate from the one or more outputs. The processing device may determine whether the provided metrology measurements meet the measurement criteria (i.e., approximately correspond to target metrology measurements for the substrate process). If the provided metrology measurements meet the measurement criteria, then this may indicate that an endpoint of the substrate process has been reached (i.e., the current profile of the substrate corresponds to the target substrate profile). In response to determining that the measurement criteria are met, the processing device may terminate the substrate process at the process chamber. For example, the processing device may generate instructions to end the substrate process. In response to determining that the measurement criteria are not met, the processing device may allow the substrate process to continue at the process chamber (e.g., the processing device may not generate instructions for the process chamber).

[0026] Aspects of the present disclosure address the deficiencies of conventional techniques by providing systems and methods for training and using machine learning models to detect the endpoint of a substrate process. An endpoint detection device coupled to a process chamber can collect spectral data of a substrate at a specific frequency during a substrate process. The collected spectral data can be provided as an input to a trained machine learning model that provides a real-time estimate of a specific type of metrological measurement of a substrate, which is used to determine whether the endpoint of the substrate process has been reached. The frequency collection of spectral data and the estimation of metrological measurements allow frequency endpoint monitoring at the process chamber, which can prevent the possibility of under-processing (e.g., under-etching, under-deposition, etc.) or over-processing (e.g., over-etching, over-deposition, etc.) of a substrate. As a result, the number of defects occurring within the manufacturing system will be reduced, thereby improving overall system efficiency and overall system yield.

[0027] A particular type of metrology measurement for a substrate may be selected (e.g., by an operator, engineer, etc.) as an endpoint metric that indicates an endpoint of a substrate process with a higher accuracy than other endpoint metrics associated with other types of metrology measurements. Because the machine learning model is trained and selected from a collection of different types of machine learning models as a model having a performance rating that satisfies performance criteria associated with the particular type of metrology measurement, the metrology measurements provided for the substrate are associated with a higher accuracy than other metrology measurements provided by a conventional metrology estimation system. Thus, the metrology measurements provided by the selected machine learning model are associated with a high accuracy rating, which improves the accuracy of endpoint detection based on the provided metrology measurements, further preventing the possibility of under-processing or over-processing of the substrate.

[0028] In addition, as described above, the machine learning model can be trained and selected from a collection of machine learning models at a processing device separate from the system controller of the manufacturing system. This greatly reduces the amount of system resources consumed by the system controller, increases overall system efficiency, and reduces overall system latency.

[0029] The endpoint detection system of the present disclosure provides enhanced spectral data collection that cannot be achieved by traditional endpoint detection systems. The optical beam of the endpoint detection system of the present disclosure enables multiple incident light signals to be transmitted from the light source component to the substrate surface, and enables multiple reflected light signals to be transmitted from the substrate surface to the photodetector without using additional equipment (e.g., beam splitter) that may reduce the power of the transmission signal. Therefore, the optical beam collects the reflected light more efficiently, reduces the signal loss of the reflected light beam, and ensures that the overall amplitude of the reflected light signal is greater than that of the traditional optical fiber cable. The collimator assembly of the endpoint detection system generates an incident light beam that has a uniform spatial distribution over a wide range of wavelengths. For example, the width of the incident light beam can be the same for the spectral component of the 250nm of the beam and the spectral component of the 750nm of the beam. In addition, even if the beam alignment is different between different manufacturing systems, the width of the incident light beam can also be the same. The enhanced uniformity ensures that the measurement of the optical response of the target portion of the substrate surface in the process chamber is more accurate, thereby providing more accurate data, which allows the state of the substrate profile to be more accurately determined during the substrate process.

[0030] Figure 1An illustrative computer system architecture 100 according to aspects of the present disclosure is depicted. The computer system architecture 100 includes a client device 120, manufacturing equipment 122, metrology equipment 132, a prediction server 112 (e.g., for generating prediction data, for providing model adaptation, for 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 a manufacturing system (such as a computer system) for processing a substrate. Figure 2 In an additional or alternative embodiment, the computer system architecture 100 may include an endpoint detection system for detecting an endpoint of a substrate process or may be a part of an endpoint detection system. Figure 3 Further details on the endpoint detection system are provided.

[0031] Components of the client device 120, the manufacturing equipment 122, the metrology equipment 132, the prediction system 110, and / or the data store 150 may be coupled to each other via a network 140. In some embodiments, the network 140 is a public network that provides the client device 120 with access to the prediction server 112, the data store 140, and other publicly available computing devices. In some embodiments, the network 140 is a private network that provides the client device 120 with access to the manufacturing equipment 122, the metrology equipment 132, the data store 150, and other privately available computing devices. The 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.

[0032] 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, etc.

[0033] The manufacturing equipment 122 can produce products according to a recipe or perform operations over a period of time to produce products. In some embodiments, the manufacturing equipment 122 may include a process tool or may be a part of a process tool, and the process tool includes one or more stations (e.g., process chambers, transfer chambers, load locks, etc.) configured to perform different functions on the substrate. In some embodiments, the manufacturing equipment 122 may further include an endpoint detection equipment 124, which is configured to collect data for detecting the endpoint of the process performed on the substrate at the manufacturing equipment 122. The endpoint of the substrate process refers to a point in the process at which the profile of the substrate corresponds to (i.e., matches or substantially matches) the target profile. The endpoint detection equipment 124 may include one or more components that are configured to collect and / or generate spectral data associated with one or more portions of the profile of the surface of the substrate during the substrate process. Spectral data refers to data associated with the intensity of the wave detected for each wavelength of the detected energy wave (i.e., the intensity or amount of energy).

[0034] In some embodiments, the endpoint detection equipment 124 may include a fiber bundle and a collimator assembly that are configured to direct incident light from a light source to the surface of a substrate and transmit reflected light from the substrate surface to a light detection component. A processing device (e.g., a system controller of a process tool) coupled to the endpoint detection equipment 124 may generate spectral data of a substrate profile based on the reflected light transmitted to the light detection component. In other or similar embodiments, the endpoint detection equipment 124 may include any sensor configured to generate spectral data associated with a substrate profile. Such sensors may include reflectance measurement sensors, elliptical polarization sensors, thermal spectral sensors, capacitive sensors, etc. Respectively Figure 2 and 5 Additional details regarding manufacturing equipment 122 and endpoint detection equipment 124 are described.

[0035] In some embodiments, one or more stations of the manufacturing equipment 122 may include sensors configured to generate and / or collect sensor data associated with the manufacturing equipment 122. The sensor data may include values ​​of one or more of the following: temperature (e.g., heater temperature), spacing (SP), pressure, high frequency radio frequency (HFRF), voltage of an electrostatic chuck (ESC), current, flow, power, voltage, etc. The sensor data may be associated with or indicate manufacturing parameters, such as hardware parameters, such as settings or components (e.g., size, type, etc.) of the manufacturing equipment 122, or process parameters of the manufacturing equipment 122. The sensor data may be provided while the manufacturing equipment 122 is performing a substrate process. The sensor data for each substrate may be different.

[0036] In some embodiments, the manufacturing equipment 122 may include a metrology equipment 126. The metrology equipment 126 may be configured to generate metrology data associated with a substrate processed by the manufacturing equipment 122. The metrology data may include values ​​of one or more of the following: film property data (e.g., wafer space film properties), dimensions (e.g., thickness, height, etc.), dielectric constants, dopant concentrations, densities, defects, etc. In some embodiments, the metrology data may further include values ​​of one or more surface profile property data (e.g., etching rate, etching rate uniformity, critical dimensions of one or more features included on the surface of the substrate, critical dimension uniformity across the substrate surface, edge placement errors, etc.). The metrology data may be data for a finished product or a semi-finished product. The metrology data for each substrate may be different.

[0037] The metrology equipment 126 may be configured to generate metrology data associated with the substrate before or after the substrate process. The metrology equipment 126 may be integrated with a station of a process tool of the manufacturing equipment 122. In some embodiments, the metrology equipment 126 may be coupled to a station of a process tool or may be a part of a station of a process tool, and the station is maintained in a vacuum environment (e.g., a process chamber, a transfer chamber, etc.). Such metrology equipment 126 is referred to as an integrated metrology equipment 128. Therefore, when the substrate is in a vacuum environment, the substrate may be measured by the integrated metrology equipment 128. For example, after a substrate process (e.g., an etching process, a deposition process, etc.) is performed on the substrate, the integrated metrology equipment 128 may generate metrology data for the processed substrate without removing the processed substrate from the vacuum environment. In other or similar embodiments, the metrology equipment 126 may be coupled to a process tool station (e.g., a factory interface module, etc.) that is not maintained in a vacuum environment or may be a part of a process tool station. Such metrology equipment 126 is referred to as an inline metrology equipment 130. Therefore, the substrate is measured by the inline metrology equipment 130 outside the vacuum environment.

[0038] In addition to or in lieu of the metrology equipment 126, the computer system architecture 100 may include a metrology equipment 132. The metrology equipment 132 may include metrology measurement devices that are separate from (i.e., external to) the manufacturing equipment 122. For example, the metrology equipment 132 may be a stand-alone device that is not coupled to any station of the manufacturing equipment 122. To obtain measurements of a substrate using the metrology equipment 132, a user of the manufacturing system (e.g., an engineer, an operator) may cause a substrate processed at the manufacturing equipment 124 to be removed from the manufacturing equipment 122 and transferred to the metrology equipment 132 for measurement. In some embodiments, the metrology equipment 132 may transmit metrology data generated for the substrate to a client device 120 coupled to the metrology equipment 132 via the network 140 (e.g., for presentation to a manufacturing user (such as an operator or engineer)). In other or similar embodiments, the manufacturing system user may obtain metrology data for the substrate from the metrology equipment 132 and may provide the metrology data to the computer system architecture via a graphical user interface (GUI) of the client device 120.

[0039] The data storage 150 may be a memory (e.g., random access memory), a drive (e.g., a hard drive, a flash drive), a database system, or another type of component or assembly capable of storing data. The data storage 150 may include multiple storage components (e.g., multiple drives or multiple databases) that may span multiple computing devices (e.g., multiple server computers). The data storage 150 may store spectral data, non-spectral data (e.g., sensor data), metrology data, predictive data, and the like. The spectral data may include historical spectral data (e.g., spectral data generated for a previous substrate processed at the manufacturing equipment 122 or at other manufacturing equipment coupled to the data storage 150 via the network 140) and / or current spectral data (spectral data generated for a current substrate being processed at the manufacturing equipment 122). The current spectral data may be data for which predictive data is generated. In some embodiments, the metrology data may include historical metrology data (e.g., metrology measurements for a previous substrate processed at the manufacturing equipment 122 or at other manufacturing equipment). The data storage 150 may also store contextual data associated with the substrate being processed at the manufacturing system (e.g., recipe name, recipe step number, preventive maintenance indicator, operator, etc.).

[0040] One or more portions of the data store 150 may be configured to store data that is not accessible to users of the manufacturing system. In some embodiments, all of the data stored at the data store 150 may be inaccessible to users of the manufacturing system. In other or similar embodiments, a portion of the data stored at the data store 150 is not accessible to the user, while another portion of the data stored at the data store 150 is accessible to the user. In some embodiments, the inaccessible data stored at the data store 150 is encrypted using an encryption mechanism that is unknown to the user (e.g., the data is encrypted using a private encryption key). In other or similar embodiments, the data store 150 may include multiple data stores, wherein data that is not accessible to the user is stored in a first data store, and data that is accessible to the user is stored in a second data store.

[0041] In some embodiments, prediction system 110 includes server machine 170 and server machine 180. Server machine 170 includes training set generator 172, which is capable of generating a training data set (e.g., a data input set and a target output set) to train, validate and / or test machine learning model 190 or a set of machine learning models 190. Figure 4 Some operations of the training set generator 172 are described in detail. In some embodiments, the training set generator 172 may divide the training data into a training set, a validation set, and a test set.

[0042] 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 (such as instructions running on a processing device, a general-purpose computer system, or a dedicated machine), firmware, microcode, or a combination of the above. The training engine 182 may be capable of training a machine learning model 190 or a collection of machine learning models 190. The machine learning model 190 may refer to a model artifact generated by the training engine 182 using training data, the training data including data inputs and corresponding target outputs (correct answers for the corresponding training inputs). The training engine 182 may look for patterns in the training data that map training inputs to target outputs (answers to be predicted) 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, and the like.

[0043] The training engine 182 may also be able to validate the trained machine learning models 190 using a corresponding feature set of the validation set from the training set generator 172. In some embodiments, the training engine 182 may assign a performance rating to each of the set of trained machine learning models 190. The performance rating may correspond to the accuracy of the corresponding trained model, the speed of the corresponding model, and / or the efficiency of the corresponding model. According to the embodiments described herein, the training engine 182 may select a trained machine learning model 190 whose performance rating satisfies the performance criteria to be used by the prediction component 114. Fig. 9 Further details regarding training engine 182 are provided.

[0044] The prediction server 112 includes a prediction engine 114 that is capable of providing spectral data of a portion of a current substrate being processed at the manufacturing equipment 122 as an input to a trained machine learning model 190 and running the trained model 190 on the input to obtain one or more outputs. In some embodiments, the trained model 190 run by the prediction engine 114 is selected by the training engine 182 as having a performance rating that satisfies the performance criteria, as described above. Figure 8 As further described, in some embodiments, the prediction engine 114 can also extract data from the output of the trained machine learning model 190 and use the confidence data to provide a metrology measurement for the portion of the substrate.

[0045] The confidence data may include or indicate a confidence level that the metrology value corresponds to one or more properties of the substrate associated with the current spectral data. In one example, the confidence level is a real number between 0 and 1, where 0 indicates that there is no confidence that the metrology value corresponds to one or more properties of the substrate associated with the current spectral data, and 1 indicates that there is absolute confidence that the metrology value corresponds to one or more properties of the substrate associated with the current spectral data. In some embodiments, the endpoint detection system may use the prediction system 110 to provide metrology values ​​for substrates being processed at the manufacturing system, rather than using the in-line metrology equipment 128, the integrated metrology equipment 130, and / or the external metrology equipment 132 to determine the measured metrology values. According to the embodiments provided herein, the endpoint detection system may determine whether the endpoint of the substrate process has been reached based on the metrology values ​​provided for the substrate.

[0046] 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 a greater or lesser number of 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, functions described as being performed by server machine 170, server machine 180, and / or prediction server 112 in one embodiment may also be performed on client device 120. Furthermore, functionality attributed to a particular component may also be performed by a different component or components operating together. Figure 3 Further details regarding the functional grouping of server machines 170, 180 and prediction server 112 are provided.

[0047] In an embodiment, 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 collection of individual users united as a group of administrators may be considered a "user."

[0048] Figure 2 2 is a schematic top view of an example manufacturing system 200. The manufacturing system 200 can perform one or more processes on a substrate 202. According to aspects of the present disclosure, the substrate 202 can be any suitable rigid, fixed-size, planar article, such as, for example, a silicon-containing disk or wafer, a patterned wafer, a glass plate, etc., which is suitable for manufacturing electronic devices or circuit components thereon. Figure 1 In the described embodiments, in some embodiments, manufacturing system 200 may include computer system architecture 110 or may be a portion of computer system architecture 110 .

[0049] The manufacturing system 200 may include a process tool 204 and a factory interface 206 coupled to the process tool 204. The process tool 204 may include a housing 208 having a transfer chamber 210 therein. The transfer chamber 210 may include one or more processing chambers (also referred to as process chambers) 214, 216, 218 disposed around and coupled to the transfer chamber. The processing chambers 214, 216, 218 may be coupled to the transfer chamber 210 through corresponding ports (such as slit valves, etc.). The transfer chamber 210 may also include a transfer chamber robot 212, which is configured to transfer the substrate 202 between the process chambers 214, 216, 218, the load lock 220, etc. The transfer chamber robot 212 may include one or more arms, wherein each arm includes one or more end effectors at the end of each arm. The end effectors may be configured to carry specific objects, such as wafers.

[0050] In some embodiments, the transfer chamber 210 may also include metrology equipment, such as Figure 1 The integrated metrology equipment 128 described herein can be configured to generate metrology data associated with the substrate 202 before or during a substrate process when the substrate is maintained in a vacuum environment. Figure 2 As shown, the integrated metrology equipment 128 can be disposed within the transfer chamber 210. In other or similar embodiments, the integrated metrology equipment 128 can be coupled to the transfer chamber 210. Because the integrated metrology equipment 128 is disposed within or coupled to the transfer chamber 210, metrology data associated with the substrate 202 can be generated without removing the substrate 202 from the vacuum environment (e.g., transferring to the factory interface 206).

[0051] The process chambers 214, 216, 218 may be adapted to implement any number of processes on the substrate 202. The same or different substrate processes may be performed in each of the process chambers 214, 216, 218. The substrate processes may include atomic layer deposition (ALD), physical vapor deposition (PVD), chemical vapor deposition (CVD), etching, annealing, curing, pre-cleaning, metal or metal oxide removal, etc. Other processes may be implemented on the substrate therein. In some embodiments, endpoint detection equipment (such as for Figure 1 The endpoint detection equipment 124 described herein can be coupled to or disposed within the process chambers 214, 216, 218. The endpoint detection equipment 124 can be configured to collect spectral data of a profile of a surface of a substrate during a substrate process. A processing device (e.g., a system controller 228) coupled to the endpoint detection equipment 124 can determine whether an endpoint of an etching process has been reached (i.e., whether a target profile has been etched onto the surface of the substrate) based on the collected spectral data. In some embodiments, the one or more components of the endpoint detection equipment 124 can include information about Figure 5 Components described (e.g., fiber bundles, collimator assemblies, etc.) In other or similar embodiments, the endpoint detection equipment 124 may include one or more sensors disposed within or outside the process chamber 214, 216, 218 and configured to collect spectral data of a portion of the substrate 202 and / or the environment within the process chamber 214, 216, 218 before, after, or during substrate processing.

[0052] 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 dock and couple with the transfer chamber 210 on one side and dock and couple with the factory interface 206 on the other side. In some embodiments, the load lock 220 may have an environmentally controlled atmosphere that may change from a vacuum environment (wherein substrates may be transferred to or from the transfer chamber 210) to an inert gas environment at atmospheric pressure (or near atmospheric pressure) (wherein substrates may be transferred to and from the factory interface 206).

[0053] The factory interface 206 may be any suitable housing, such as an equipment front end module (EFEM). The factory interface 206 may be configured to receive substrates 202 from substrate carriers 222 (e.g., front opening unit (FOUP)) docked at various load ports 224 of the factory interface 206. A factory interface robot 226 (shown in phantom) may be configured to transfer substrates 202 between substrate carriers (also referred to as containers) 222 and load locks 220. In other and / or similar embodiments, the factory interface 206 may be configured to receive replacement parts from replacement parts storage containers 222.

[0054] In some embodiments, the manufacturing system 200 may include metrology equipment configured to generate metrology data associated with the substrate 202 outside of a vacuum environment. Figure 2 As shown, the integrated metrology equipment 128 can be coupled to the process chamber (e.g., the process chambers 214, 216, and / or 218). The integrated metrology equipment 128 can be configured to generate metrology data associated with the substrate 202 before the substrate 202 is placed in 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). It should be noted that although Figure 2 The inline metrology equipment 130 is depicted coupled to the factory interface 206 , but the inline metrology equipment 130 may also be coupled to any portion of the process tool 204 that is outside of a vacuum environment (eg, coupled to a load lock 220 , etc.).

[0055] The manufacturing system 200 can also communicate with a client device (e.g., Figure 1 The client device 120 is connected to a client device 120 configured to provide information to a user (e.g., an operator) about the manufacturing system 200. In some embodiments, the client device can provide information to a user of the manufacturing system 200 via one or more graphical user interfaces (GUIs). For example, the client device can provide information about an endpoint of a substrate process performed at a process chamber 214, 216, 218 via a GUI.

[0056] The manufacturing system 200 may also include a system controller 228 or be coupled to the 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 that implements other instruction sets or a processor that implements 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 drives and / or solid-state hard disks), 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 methodologies and / or embodiments described herein. In some embodiments, the system controller 228 may execute instructions to perform one or more operations according to the process recipe at the manufacturing system 300. The instructions may be stored on a computer-readable storage medium, which may include main memory, static memory, secondary memory, and / or a processing device (during execution of the instructions).

[0057] In some embodiments, the system controller 228 can receive data from sensors or other components (e.g., endpoint detection equipment 124) included on or within various parts of the manufacturing system 300 (e.g., processing chambers 314, 316, 318, transfer chamber 310, load lock 320, etc.). The data received by the system controller 228 can include spectral data and non-spectral data for a portion of the substrate 302. For the purposes of this description, the system controller 228 is described as receiving data from an endpoint detection equipment 124 coupled to or disposed within the process chambers 214, 216, 218. However, the system controller 228 can also receive data from any part of the manufacturing system 200, and the data received from the part can be used according to the embodiments described herein. In an illustrative example, the system controller 228 can receive spectral data from an endpoint detection equipment 124 coupled to the process chambers 214, 216, 218 before, after, or during a substrate process at the processing chambers 214, 216, 218. Data received from endpoint detection equipment 124 or other sensors at manufacturing system 200 may be stored in data store 250. Data store 250 may be included as a component within system controller 228, or may be a separate component from system controller 228. In some embodiments, data store 250 may be or include information about Figure 1 A portion of data store 150 is depicted.

[0058] Figure 3 An illustrative system architecture 300 for endpoint detection of a substrate process according to aspects of the present disclosure is depicted. In some embodiments, endpoint detection system 300 may include or may be a part of computer architecture 100 and / or manufacturing system 200, one or more components. Endpoint detection system 300 may include one or more components (e.g., endpoint detection equipment 126) of manufacturing equipment 124, metrology equipment 130, server machine 320, and server machine 350.

[0059] As previously described, the manufacturing equipment 122 may produce a product or perform a job over a period of time according to a recipe. 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 process chamber 310 may be a Figure 2The manufacturing equipment 122 may also include an endpoint detection device 124, as described herein. The endpoint detection device 124 may be coupled to the process chamber 310 or disposed within the process chamber 310 and may collect spectral data for detecting an endpoint of a step of a substrate process recipe. In some embodiments, the manufacturing equipment 122 may also include an integrated metrology device 128, as described herein. The integrated metrology device 128 may be configured to generate metrology data associated with the substrate before or after the substrate process is completed.

[0060] Manufacturing equipment 122 may be coupled to server machine 320. Server machine 320 may include processing device 322 and / or data storage 332. In some embodiments, processing device 322 may be configured to execute one or more instructions to perform operations at manufacturing equipment 122. For example, processing device 322 may include instructions for performing operations on the server machine 320. Figure 2 System controller 228 is depicted or may be a part of system controller 228. In some embodiments, data store 332 may include data store 150 and / or data store 250 or may be a part of data store 150 and / or data store 250.

[0061] The processing device 322 may be configured to receive data (i.e., via a network) from one or more components of the manufacturing equipment 122. For example, the processing device 322 may receive spectral data 336 collected during a substrate process of a substrate performed by the endpoint detection equipment 124 at the process chamber 310. In another example, the processing device 322 may receive metrology data 338 collected by the integrated metrology equipment 128 before and / or after the substrate process of the substrate. The metrology data 338 may include metrology measurements generated by the integrated metrology equipment 128 for the substrate. In some embodiments, the processing device 322 may store the received spectral data and / or the received metrology data 338 at the data storage 332.

[0062] In some embodiments, the processing device 322 can receive metrology data 338 from other metrology equipment. For example, in some embodiments, the server 320 can be coupled to the inline metrology equipment 130 (i.e., via a network). As described herein, the substrate can be removed from the process chamber 210 and transferred to the inline metrology equipment 130. The inline metrology equipment 130 can generate metrology data 338 for the substrate and transmit the generated metrology data 338 to the processing device 322 via the network. In another example, the substrate can be removed from the manufacturing equipment 122 and can be transferred to an external metrology equipment, such as for example, a substrate. Figure 1The external metrology equipment 132 described. A client device (e.g., client device 120) can be coupled to server 320 (i.e., via a network). In some embodiments, a user of manufacturing equipment 122 can obtain metrology data 338 using external metrology equipment 132, and the obtained metrology data 338 can be provided via a GUI of client device 120. Client device 120 can transmit metrology data 338 via the network. In additional or alternative embodiments, external metrology equipment 132 can be coupled to server 320 via a network, and external metrology equipment 132 can transmit metrology data 338 directly to processing device 332.

[0063] The processing device 322 may include a prediction engine 328 and a process endpoint engine 330. The prediction engine 328 may be configured to provide a value of a metrology measurement based on spectral data 336 collected for a substrate during a substrate process. For example, the prediction engine 328 may provide the spectral data 336 collected for a current process performed on a current substrate at the process chamber 310 as an input to the trained machine learning model 334. The prediction engine 328 may obtain metrology data 338 as an output of the machine learning model 334, the metrology data 338 including an indication of a metrology measurement value corresponding to the current substrate. In some embodiments, the prediction engine 328 may be associated with a machine learning model 334 that is trained on the current substrate. Figure 1 The prediction engine 114 described corresponds to .

[0064] like Figure 3 As shown, in some embodiments, the processing device 322 may include a training set generator 324 and / or a training engine 326. In some embodiments, the training set generator 324 may correspond to the training set generator 172, and / or the training engine 326 may correspond to the training set generator 172. Figure 1 The described training engine 182 corresponds. The training set generator 324 can be configured to generate a training set 340 to train the 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 spectral data 336 associated with previous substrates. In some embodiments, the training set generator 324 can retrieve historical spectral data 336 from the data store 332 to generate training inputs. The training set generator 324 can generate target outputs for the training inputs based on historical metrology data 338 obtained for previous substrates. As described above, the historical metrology data 338 can be generated by the in-line metrology equipment 128, the integrated metrology equipment 130, or the external metrology equipment 132. The training set generator 324 can include the generated training inputs and the generated target outputs in the training set 340. About Figure 4 Further details regarding generating training set 340 are provided.

[0065] The training engine 326 may be configured to train, validate, and / or test the machine learning model 334 or a set of machine learning models 334. The training engine 326 may provide a training set 340 to train the machine learning model 334, and store the trained machine learning model 334 at the data store 332. In some embodiments, the training engine 326 may use a validation set 342 to validate the trained machine learning model 334. The validation set 342 may include spectral data 336 and metrology data 338 obtained for a previous substrate (i.e., a previous substrate processed at the process chamber 310 or at another process chamber). The training set generator 324 and / or the training engine 326 may generate the validation set 342 based on historical spectral data 336 and historical metrology data 338 obtained for the previous substrate. In some embodiments, the validation set 342 may include historical spectral data 336 and historical metrology data 338 different from the historical spectral data 336 and historical metrology data 338 included in the training set 340.

[0066] The training engine 326 may provide spectral data 336 of previous substrates as input to the trained machine learning model 334, and may extract metrology measurements of the previous substrates from one or more outputs of the trained model 334. The training engine 336 may assign a performance score to the trained model 334 based on the accuracy of the metrology measurements of the previous substrates in view of the measured metrology values ​​of the metrology data 338 of the previous substrates included in the validation set 342. The training engine 336 may select the trained model 334 for use in providing future metrology measurements for future substrates processed at the process chamber 310 in response to determining that the performance score satisfies a performance score criterion (e.g., exceeds a performance score threshold). Fig. 9 Further details regarding selecting trained model 334 are provided.

[0067] As previously discussed, in some embodiments, the training set generator 324 and / or the training engine 326 may be a component of a processing device 322 at the server 320. In additional or alternative embodiments, the training set generator 324 and / or the training engine 326 may be a component of a processing device 352 at the server 350. The server 350 may include a computing system separate from the manufacturing system 200 or may be part of a computing system separate from the manufacturing system 200. As previously discussed, in some embodiments, the server 320 may include information about Figure 2The system controller 228 is depicted or may be a part of the system controller 228. In such embodiments, the server 350 may include or may be a 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 at one or more portions of the data store 332 or one or more processes performed at the processing device 322. However, a user of the manufacturing system 200 may be provided with access to any data stored at one or more portions of the data store 354 or any process performed at the processing device 352.

[0068] The processing device 352 can be configured to execute the training set generator 324 and / or the training engine 326 in a similar manner as 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. Therefore, according to the embodiments described with respect to the processing device 322, the processing device 352 can obtain the spectral data 336 and the metrology data 338 for use by the training set generator 324 and / or the training engine 326 to generate the training set 340 and the validation set 342. In other or similar embodiments, the server 350 is not coupled to the manufacturing equipment 122 and / or the external metrology equipment 130. Therefore, the processing device 352 can obtain the spectral data 336 and / or the metrology data 338 from the processing device 322. For example, as previously described, the processing device 322 can receive the spectral data 336 from the endpoint detection equipment 124. The processing device 322 can transmit the received spectral data 336 to the processing device 352 (i.e., via the network). In some embodiments, the processing device 352 can store the spectral data 336 at the data store 354. In some embodiments, the processing device 322 can similarly transmit metrology data 338 obtained for the substrate to the processing device 352. For example, as previously described, the processing device 322 can receive the metrology data 338 from the inline metrology equipment 128, the integrated metrology equipment 130, and / or the external metrology equipment 132. The processing device 352 can transmit the metrology data 338 received from the processing device 322 and, in some embodiments, store the metrology data 338 at the data store 354.

[0069] The training set generator 324 at the processing device 352 may generate the training set 340 according to the previously described embodiments. According to the previously described embodiments, the training engine 326 at the processing device 352 may train and / or validate the machine learning model 340. In some embodiments, the server 350 may be coupled to other manufacturing equipment and / or other server machines that are different from the manufacturing equipment 122 and / or the server machine 320. According to embodiments described herein, the processing device 352 may obtain the spectral data 336 and / or the metrology data 338 from the other manufacturing equipment and / or the server machine. In some embodiments, the training set 340 and / or the validation set 342 may be generated based on the spectral data 336 and the metrology data 338 obtained for the substrate processed at the process chamber 310 and other spectral data and metrology data obtained for other substrates processed at the process chambers at other manufacturing systems.

[0070] In response to the training engine 326 selecting the trained model 334 to be applied to future spectral data for future substrates at the process chamber 310, the processing device 352 can transmit the trained model 334 to the processing device 322. As previously described, the prediction engine 328 can use the trained model 334 to provide metrology measurements for future substrates at the process chamber 310.

[0071] The process endpoint engine 330 at the processing device 322 can be configured to detect the endpoint of a process performed on a substrate at the process chamber 310. The process endpoint engine 330 can obtain the metrology measurement value of the current substrate at the process chamber 310 from the prediction engine 328. In response to determining that the metrology measurement value satisfies the metrology measurement criteria (e.g., corresponds to the target metrology measurement value), the process endpoint engine 330 can generate an instruction including a command to terminate the current process (or the current step of the current process) at the process chamber 310. The processing device 322 can transmit an instruction to the manufacturing equipment 122 so that the current process of the current substrate is terminated. In response to determining that the metrology measurement criteria are not met, the process endpoint engine 330 can allow the current process to continue on the current substrate (e.g., without generating an instruction to terminate the process). About Fig.10 Further details associated with detecting the endpoint of the current process are provided.

[0072] Figure 4 is a flow chart of a method 400 for training a machine learning model according to aspects of the present disclosure. The method 400 is performed by process logic, which may include hardware (circuitry, dedicated logic, etc.), software (such as software running on a general-purpose computer system or a dedicated machine), firmware, or some combination of the above. In one embodiment, the method 400 may be performed by a computer system (such as Figure 1In other or similar embodiments, one or more operations of method 400 may be performed by one or more other machines not depicted in the figure. In some aspects, one or more operations of method 400 may be performed by a computer system architecture 100. Figure 3 The training set generator 324 of the described server machine 320 or server machine 350 executes.

[0073] At block 410, the process logic initializes the training set T to an empty set (e.g., {}). At block 412, the process logic obtains spectral data associated with a substrate processed at a process chamber of a manufacturing system. In some embodiments, the spectral data may be received from one or more sensors disposed within or coupled to the process chamber. In other or similar embodiments, the spectral data may be received from an endpoint detection device disposed within or coupled to the process chamber. Figure 5 1 is a schematic cross-sectional side view of an endpoint detection device 124 coupled to a process chamber 310 according to aspects of the present disclosure. In some embodiments, the process chamber 310 can be used for processes in which a corrosive plasma environment is provided. For example, the process chamber 310 can be used in a chamber of a plasma etcher or a plasma etching reactor, a plasma cleaner, etc. In other or similar embodiments, the process chamber 310 can be used for processes in which a non-corrosive environment is provided. For example, the process chamber 310 can be used as a chemical vapor deposition (CVD) chamber, a physical vapor deposition (PVD) chamber, an atomic layer deposition (ALD) chamber, an ion assisted deposition (IAD) chamber, and other types of processing chambers.

[0074] Briefly, the process chamber 310 includes a chamber body 502 and a lid 530 and / or a showerhead (not shown) that enclose an interior volume 506. The chamber body 502 generally includes a sidewall 508 and a bottom 510. The showerhead may include a showerhead base and a showerhead gas distribution plate. The lid 530 and / or showerhead may be supported on the sidewall 508 of the chamber body 502. The lid 530 (or showerhead) may be opened to allow access to the interior volume 506 of the process chamber 510 and may provide a seal for the process chamber 510 when closed. A gas panel (not shown) may be coupled to the process chamber 510 to provide process and / or cleaning gases to the interior volume 506 through the lid 530 and nozzles (e.g., through holes in the showerhead or lid and nozzles) and / or showerheads. An exhaust port 526 may be defined in the chamber body 502 and may couple the interior volume 506 to a pump system 528. The pump system 528 may include one or more pumps and throttle valves for evacuating and regulating the pressure of the interior volume 506 of the process chamber 510. The substrate support assembly 548 is disposed in the lid 530 and / or showerhead of the interior volume 506. The substrate support assembly 548 holds a substrate (such as a Figure 2In one embodiment, the substrate support assembly 548 includes a base 552 that supports an electrostatic chuck 550.

[0075] The endpoint detection equipment 124 can be configured to optically monitor the environment of the interior volume 506 during substrate processing of the substrate 202. In some embodiments, the endpoint detection equipment 124 can be mechanically coupled to the chamber body 502 and optically interfaced with the environment of the interior volume 506 (i.e., interfaced via an optical interface 570). The endpoint detection equipment 124 can include a collimator assembly 554, an optical fiber bundle 556, an optical component 562, a processing device 576, and in some embodiments, a polarizer component 586. Figure 5 As shown, the collimator assembly 554 can be coupled to the chamber interface 570. In some embodiments, the chamber interface 570 can be an orifice, a focusing or diverging lens, a transparent plate, or any other device or material capable of transmitting light between the collimator assembly 554 and the environment of the internal volume 506. It should be noted that although Figure 5 The chamber interface 570 is depicted as being embedded within the lid 530 , but the chamber interface 570 may also be embedded within or coupled to any portion of the process chamber 310 (eg, the sidewall 508 , the bottom 510 , etc.).

[0076] The first end of the optical fiber bundle 556 can be coupled to the collimator assembly 554, and the second end of the optical fiber bundle 556 can be coupled to the optical component 562. The optical component 562 may include a light source 564 configured to generate light. In this article, "light" refers to electromagnetic radiation in any spectral range (including visible light, far infrared and near infrared (IR), far ultraviolet and near ultraviolet (UV), etc.). "Light" may further include unpolarized (e.g., natural) light, linearly, circularly, or elliptically polarized light, partially polarized light, focused light, divergent light, collimated light, etc. In some embodiments, the light source 564 may include a narrowband light source, such as a light emitting diode (LED), a laser, a light bulb, etc. In other or similar embodiments, the light source 564 may include a broadband light source. The light source 564 may include a light source of more than one component, such as in some embodiments, multiple narrowband light sources produce (when used together) a broadband light input. The light source 564 may include additional optical components (i.e., filters, absorbers, polarizers, etc.) to control the spectral distribution and / or polarization of the light.

[0077] The light generated by the light source 564 (referred to herein as input light) can be transmitted to the collimator assembly 554 via one or more launch fibers 558 of the optical bundle 556. In response to receiving the input light via the launch fibers 558, the collimator assembly 554 can be configured to convert the input light into a light beam 572. For example, the input light can be transmitted via one or more optical components (such as lenses, reflectors, filters, holes, etc.) of the collimator assembly 554. In some embodiments, the spatial properties of the light beam generated by the collimator assembly 554 can be the same for multiple spectral components of the light beam 572. For example, the diameter of the light beam 572 can be the same over a wide range of wavelengths λ of various spectral components contained in the input light (and therefore contained in the light beam 572). In some embodiments, the collimator assembly 554 can include one or more achromatic lenses. Therefore, the light beam 572 generated by the collimator assembly 554 can be an achromatic beam.

[0078] like Figure 5 As shown, in some embodiments, the collimator assembly 554 can include a polarizer component 568. The polarizer component 568 is configured to polarize unpolarized (e.g., natural) light generated by the light source 564. For example, the polarizer component 568 can convert unpolarized input light into linearly, circularly, or elliptically polarized light. It should be noted that although Figure 5 The polarizer component 568 is shown as part of the collimator assembly 554, but the polarizer component 568 may be coupled to any portion of the endpoint detection equipment 124 that passes input light to the optical interface 570. For example, the polarizer component 568 may be coupled to the outlet of the light source 564, to the outlet of the one or more launch optical fibers 558, between the collimator assembly and the optical interface 570, etc.

[0079] The collimator 572 can direct the light beam 572 toward the surface of the substrate 202 disposed on the substrate support assembly 548 via the optical interface 570. The light beam 572 can be reflected from the surface of the substrate 202 as a reflected light beam 574, which is received by the collimator assembly 554. One or more receiving optical fibers 560 of the optical bundle 556 can transmit the reflected light beam 574 to the light detector 566 of the optical component 562. The light detector 566 can include one or more spectrographs, spectrometers, diffraction gratings, mirrors, lenses, photodiodes, and other devices. The light detector 566, alone or in conjunction with the processing device 576, can determine one or more optical responses associated with the surface of the substrate 202 based on the reflected light beam 574. For example, the light detector 566 and / or the processing device 576 can determine the reflectivity R(λ), the refractive index n(λ), or any other optical quantity that can be used to characterize the substrate 202 based on the reflected light 574. In some embodiments, the optical response can be used to characterize the polarization dependence of the reflectivity, the rotation angle of the polarization plane upon reflection, the luminous intensity, etc. for the substrate 202. Spectral data described herein may refer to data corresponding to an optical response of reflected light 574 and / or an optical characteristic of substrate 202 derived from the optical response of reflected light 574 .

[0080] In some embodiments, the processing device 576 may be included as part of a system controller (e.g., system controller 228) of a manufacturing system including the process chamber 310. In such embodiments, the processing device 576 may store the spectral data generated for the substrate 202 at a data store (e.g., data store 250, 332, 352, etc.) coupled to the processing device 576. In other or similar embodiments, the processing device 576 may be a processing component separate from the system controller but coupled to the system controller via a network. The processing device 576 may transmit the generated spectral data to the system controller for storage at a corresponding data store of the manufacturing system.

[0081] Back to Figure 4At block 414, the process logic obtains metrology data for the substrate. As previously described, the metrology data may include values ​​for one or more of the following: film property data (e.g., wafer space film properties), dimensions (e.g., thickness, height, etc.), dielectric constants, dopant concentrations, densities, defects, etc. In some embodiments, the metrology data may further include values ​​for one or more surface profile property data (e.g., etch rate, etch rate uniformity, critical dimensions of one or more features included on the surface of the substrate, critical dimension uniformity across the substrate surface, edge placement error, etc.). In some embodiments, metrology measurements may be received from metrology equipment (e.g., inline metrology equipment 126, integrated metrology equipment 128, etc.) of the manufacturing system 200. For example, after the substrate process is completed, the substrate may be transferred to the inline metrology equipment 128 and the integrated metrology equipment 130 of the manufacturing system 200. The inline metrology equipment 128 or the integrated metrology equipment 130 may generate metrology data associated with the substrate, and may transmit the metrology data to a computing device (e.g., server machine 320, server machine 350, etc.) including the process logic via a network. In other or similar embodiments, as described herein, metrology measurements can be received from metrology equipment separate from manufacturing system 200 (eg, external metrology equipment 132 ).

[0082] At block 416, the process logic generates a training input based on the spectral data obtained for the substrate at block 412. In some embodiments, the training input may include a regularized spectral data set generated based on the spectral data obtained for the substrate. The regularized spectral data set may include one or more spectral features that correspond to a particular type of metrology measurement. Figure 6 Additional details about generating training inputs are provided. At block 418, the process logic may generate a target output based on the metrology data obtained for the substrate at block 414. The target output may correspond to a value of a metrology measurement associated with the substrate. For example, at block 414, the process logic may obtain metrology data indicating a thickness of a film at one or more portions of a surface of the substrate after an etching process. The process logic may generate a target output corresponding to an etch rate and / or an etch rate uniformity based on the obtained metrology data and process data for the etching process.

[0083] At block 420, the process logic generates an input / output map. The input / output map refers to a training input that includes or is based on the data of the substrate and a target output of the training input, wherein the target output identifies a metrology measurement value of the substrate, and wherein the training input is associated with (or mapped to) the target output. At block 422, the process logic adds the input / output map to the training set T.

[0084] At block 424, the process logic determines whether the training set T includes an amount of training data sufficient to train the machine learning model. It should be noted that in some embodiments, the sufficiency of the training set T can be determined simply based on the number of input / output mappings in the training set, while in some other embodiments, the sufficiency of the training set T can be determined based on one or more other criteria (e.g., a measure of diversity of training examples, etc.) in addition to or in lieu of the number of input / output mappings. In response to determining that the training set T includes an amount of training data sufficient to train the machine learning model, the process logic provides the training set T to train the machine learning model. In response to determining that the training set does not include an amount of training data sufficient to train the machine learning model, the method 400 returns to block 412.

[0085] At box 426, the process logic provides a training set T to train 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 the training. In the case of a neural network, for example, input values ​​for a given input / output mapping (e.g., spectral data for a previous substrate) are input to the neural network, and the output values ​​of the input / output mapping are stored in the output nodes of the neural network. The connection weights in the neural network are then adjusted according to a learning algorithm (e.g., back propagation, etc.), and the procedure is repeated for other input / output mappings in the training set T. After box 426, the machine learning model 190 can be used to provide metrology values ​​for future substrates processed at the manufacturing system (e.g., according to the method described below). Figure 8 Method 800).

[0086] Figure 6 6 is a flow chart of a method 600 for obtaining data for use with a machine learning model (e.g., machine learning model 334) according to aspects of the present disclosure. The method 600 is performed by process logic, which may include hardware (circuitry, dedicated logic, etc.), software (such as software running on a general-purpose computer system or a dedicated machine), firmware, or some combination of the above. In one embodiment, the method 600 may be performed by a computer system (such as Figure 1 In other or similar embodiments, one or more operations of method 600 may be performed by one or more other machines not depicted in the figure. In some aspects, one or more operations of method 600 may be performed by a computer system architecture 100. Figure 3 The method 600 may be performed by the training set generator 324 of the server machine 320 or the server machine 350 described above. Additionally or alternatively, one or more operations of the method 600 may be performed by the prediction engine 328 of the server machine 350. In one example, according to the Figure 4In the embodiment described in method 400, the training set generator 324 may perform the operations described in one or more of blocks 610-624 to generate training inputs. Figure 8 As described in method 800 of , the prediction engine 328 may perform the operations described with respect to one or more of blocks 620 - 624 to obtain a subset of spectral data to provide as input to a trained machine learning model.

[0087] At block 610, the process logic may obtain spectral data associated with the substrate. In some embodiments, the spectral data may correspond to spectral data obtained by the process logic of the training set generator 324 as described with respect to block 412 of the method 400. In other or similar embodiments, the spectral data may correspond to spectral data obtained by the process logic of the prediction engine 328, as described herein. Fig. 7A Spectral data collected for a substrate during substrate processing according to embodiments described herein is illustrated. Fig. 7A As shown, the spectral data may correspond to the amplitude of a light wave having a specific wavelength of the light signal, which is detected by a sensor or endpoint detection equipment disposed in or coupled to the process chamber during a corresponding time period of the substrate process. For example, the spectral data may correspond to the amplitude of wavelengths X, Y, and Z detected by the endpoint detection equipment 124 during the initial (e.g., at time 0 seconds), the middle (e.g., at time 1 second to time N-1 seconds), and the final (e.g., at time N seconds) time periods of the etching process of the substrate.

[0088] Back to Figure 6 In some embodiments, the process logic may also obtain metrology data associated with the substrate. The process logic may obtain the metrology data according to embodiments described with respect to block 414 of method 400 .

[0089] At block 612, the process logic may obtain data associated with a rate of a substrate process performed on the substrate. In some embodiments, the substrate process may be an etching process, and the rate data may be an etching rate of the etching process. In other or similar embodiments, the substrate process may be a deposition process, and the rate data may be a deposition rate of the deposition process. It should be noted that the substrate process may be another process other than an etching process and / or a deposition process, and the process logic may obtain corresponding rate data associated with the corresponding substrate process.

[0090] In some embodiments, the process logic may obtain rate data based on contextual data and / or metrology data of the substrate. For example, the obtained metrology data may indicate the thickness of a film deposited on a surface of the substrate before and / or after a substrate process. Figure 7BExample rate data for a substrate process according to embodiments described herein are illustrated. Data point 702 indicates the thickness of a film deposited on the surface of a substrate before an etching process. Data point 704 indicates the thickness of the film after the substrate process. According to embodiments described herein, process logic may obtain data point 704 based on metrology data collected for the substrate after the etching process. In some embodiments, process logic may obtain data point 702 based on metrology data collected for the substrate before the etching process. In other or similar embodiments, process logic may obtain data point 702 based on data that is not generated by metrology equipment before the etching process. For example, a user of a manufacturing system may provide data associated with the thickness of a film (e.g., from a substrate specification) before a substrate process via a client device coupled to the manufacturing system. In another example, metrology equipment may generate metrology data for another substrate before another etching process is performed. Data point 702 may indicate the thickness of a film obtained based on metrology data collected for another substrate.

[0091] In some embodiments, the process logic may obtain an indication of the length of the substrate process based on contextual data associated with the substrate (e.g., stored at the data store 150). For example, the process logic may retrieve data from the data store 150 indicating that an etching process was initiated on the substrate at an initial time period of 0 seconds, and the etching process was completed at a final time period of approximately N seconds. Therefore, the process logic may determine that the length of the etching process is approximately N seconds.

[0092] The process logic may determine the rate of the process based on the difference between the thickness of the film before and after the substrate process and the length of the substrate process. For example, the process logic may determine that data point 702 is The film thickness corresponds to , while data point 704 corresponds to The film thickness corresponds to (i.e., the film thickness etched away during the etching process) Based on the obtained context data, the process logic may determine that the length of the etching process is 8 seconds. Therefore, the process logic may determine that the etching rate 706 of the etching process is approximately

[0093] It should be noted that in some embodiments, the process logic may determine the rate of the substrate process based on other techniques. In some embodiments, the process logic may obtain the rate data of the substrate process based on data provided by a user of the manufacturing system. For example, a user of the manufacturing system may provide an etching rate of an etching process via a client device coupled to the manufacturing system. In other or similar embodiments, the process logic determines the rate of the substrate process based on process recipe data associated with the substrate. In another example, the process logic may obtain etching process recipe data from the data storage 150, and determine the etching rate of the etching process based on the amount of etching plasma generated at the process chamber during one or more time periods of the etching process. In some embodiments, the process logic may also determine whether the process rate is constant over the duration of the substrate process. For example, a user may provide an indication that the etching rate of the etching process is constant via a client device. Alternatively, a user may provide an indication that the etching rate of the etching process is not constant, and an indication of the etching rate at each instant during the etching process may be provided. In another example, according to the previously described embodiments, the process logic may determine whether the process rate is constant based on the process recipe data. If the process rate is not constant, the process rate may also be determined based on process recipe data (eg, based on the amount of etching plasma generated at the process chamber at each instant in time at the process chamber).

[0094] At block 614, the process logic may generate synthetic metrology data for the substrate based on the rate data obtained for the substrate. In some embodiments, the synthetic metrology data for the substrate may correspond to an estimated thickness of a film on a surface of the substrate for one or more intermediate time periods between an initial time period of a substrate process and a final time period of the substrate process. In some embodiments, the process logic may generate synthetic metrology data for each intermediate time period during which spectral data was generated for the substrate during the substrate process. Figure 7C The synthetic data generated for a substrate according to the embodiments described herein is illustrated. Figure 7B As described above, the process logic may determine (e.g., obtain from contextual data of the etching process) an etch rate 706 for the etching process based on data point 702, data point 704, and the length of the etching process. The process logic may determine the thickness of the film at other time periods during the etching process based on the determined etch rate 706. Based on the example provided previously, the process logic may determine that the length of the etching process is approximately 8 seconds and the etch rate 706 is approximately Therefore, the process logic can estimate that the film thickness is about 1 second after the etching process starts. About 2 seconds after the etching process starts The composite metrology data may correspond to an estimated film thickness for each of a set of time periods between the initial time period and the final time period.

[0095] At block 616, the process logic may map the acquired spectral data to acquired metrology data and synthetic metrology data for the substrate. Fig.7D As shown, the process logic may map the metrology data obtained at the initial and final time periods of the substrate process to the spectral data collected during the corresponding time periods of the substrate process. For example, the process logic may map the initial thickness of the film (i.e., data point 702) at the initial period of the etching process to the amplitude of each wavelength detected by the endpoint detection equipment 124 during the initial period. Additionally or alternatively, the process logic may map the generated synthetic metrology data associated with each corresponding intermediate time period of the etching process to the amplitude of each detected wavelength during the corresponding intermediate time period.

[0096] At box 618, the process logic may generate normalized spectral data for the substrate. Regularized spectral data may refer to spectral data that is adjusted to account for differences naturally caused by data collected for substrate processes performed at different instants (e.g., processes performed during the morning shift and processes performed during the afternoon or evening shift). In some embodiments, the process logic may generate normalized spectral data by selecting a set of spectral data as baseline spectral data. In some embodiments, the baseline spectral data may be spectral data collected during an initial period of the substrate process. The process logic may determine the amplitude of one or more wavelengths of the detected optical signal based on the selected set of spectral data. In some embodiments, the process logic may determine the amplitude of each wavelength of the detected optical signal during the initial period of the substrate process. In other or similar embodiments, the process logic may determine the amplitude of a specific wavelength of the detected optical signal. For example, the process logic may determine one or more specific wavelengths of the detected signal that correspond to specific metrology measurements of the metrology data of the substrate.

[0097] In some embodiments, the process logic may use a wave analysis model (such as a rigorous coupled wave analysis (RCWA) model) to determine the one or more wavelengths corresponding to a particular metrology measurement. The wave analysis model may be a theoretical model that is trained to provide spectral wavelengths (i.e., theoretical spectral wavelengths) corresponding to metrology measurement values ​​associated with a particular type of metrology measurement, which are more accurate than other spectral wavelengths. For example, the wave analysis model may be trained to identify a range of spectral wavelengths. Considering the variation of critical dimensions across the surface of the substrate, some wavelengths in the spectral wavelength range may exhibit different behaviors from other wavelengths in the wavelength range. Such spectral wavelengths may indicate critical dimension measurements with higher accuracy than other spectral wavelengths of the optical signal detected by the endpoint detection equipment 200. The process logic may provide one or more portions of the obtained spectral data as inputs to the wave analysis model, and may obtain one or more outputs of the wave analysis model. The process logic may extract a spectral wavelength range corresponding to a particular metrology measurement from the one or more outputs. In some embodiments, the spectral wavelength range may correspond to a particular portion of the spectrum. For example, the spectral wavelength range may correspond to a particular color (e.g., green) of the visible light spectrum. In additional or alternative embodiments, the spectral wavelength range may correspond to a particular wavelength of the spectrum. In response to extracting the spectral wavelength ranges, the process logic may select baseline spectral data for each of the spectral wavelength ranges to include in the set of spectral data.

[0098] In response to determining the amplitude of the one or more wavelengths of the detected optical signal in the initial time period, the process logic can determine the amplitude of the one or more wavelengths for each subsequent time period, and the spectral data is collected in the subsequent time period during the substrate process. As described above, the endpoint detection equipment 124 can collect spectral data of the substrate in one or more intermediate time periods (e.g., from time 1 second to time N-1 seconds) and a final time period (e.g., at time N seconds) of the etching process. Therefore, the process logic can determine the amplitude of each of the one or more wavelengths of the detected optical signal in each intermediate time period and the final time period.

[0099] In some embodiments, the process logic may generate normalized spectral data for the substrate based on the difference between the amplitude of the one or more wavelengths in the initial time period of the etching process and the amplitude of the one or more wavelengths in each subsequent time period (i.e., in each intermediate time period and the final time period). In an illustrative example, the process logic may determine that the amplitude of wavelength X in the initial time period is about 10 units, while the amplitude of wavelength X in the first intermediate time period is about 11 units. The normalized spectral data may include a first normalized spectral data point indicating that the normalized amplitude of wavelength X in the first intermediate time period is about 1 unit (e.g., the difference between the amplitude of wavelength X in the initial time period and the amplitude of wavelength X in the second time period). The process logic may also determine that wavelength X is about 12 units in the second intermediate time period. Therefore, the normalized spectral data may include a second normalized spectral data point indicating that the normalized amplitude of wavelength X in the second intermediate time period is about 2 units. According to the above example, the process logic may generate normalized spectral data for each intermediate time period and the final time period.

[0100] In other or similar embodiments, the process logic may generate normalized spectral data for the substrate based on a ratio between the amplitude of the one or more wavelengths in each subsequent time period of the etching process and the amplitude of the one or more wavelengths in the initial time period. For example, the process logic may determine that the amplitude of wavelength X in the initial time period is about 5 units, while the amplitude of wavelength X in the first intermediate time period is about 10 units. The normalized spectral data may include a first normalized spectral data point indicating that the normalized amplitude of wavelength X in the first intermediate time period is about 2 units (e.g., the ratio between the amplitude of wavelength X in the second time period and the amplitude of wavelength X in the first time period).

[0101] Fig.7D Further illustrated is example normalized spectral data mapped to acquired and synthesized metrology data for a substrate process. As described above, the normalized spectral data does not account for any variations present in the raw spectral data collected by the endpoint detection equipment 124 (e.g., process chamber differences, process run time differences, etc.). Fig.7D is described in Fig. 7ACompared to each line of the original spectral data representing the corresponding wavelength, each line of the normalized spectral data representing the corresponding wavelength is smooth. As described above, in some embodiments, the process logic can generate the normalized spectral data based on the difference between the amplitude of the one or more wavelengths in the initial time period and the amplitude of the one or more wavelengths in each subsequent time period. In such an embodiment, at time 0, the amplitude of each wavelength is approximately 0 units. In other or similar embodiments, the process logic can generate the normalized spectral data based on the ratio between the amplitude of the one or more wavelengths in each subsequent time period and the amplitude of the one or more wavelengths in the initial time period. In such an embodiment, at time 0, the amplitude of each wavelength is approximately 1 unit.

[0102] Back to Figure 6 At block 620, the process logic may extract a set of spectral data including one or more spectral features from the regularized spectral data, the spectral features corresponding to a particular type of metrology measurement. In some embodiments, the process logic may perform a principal component analysis (PCA) to select the most important features of the spectral data. Principal component analysis refers to an analysis of a set of points in a real coordinate space to perform a change in the basis of the set of points. In some embodiments, the set of spectral features includes a wavelength range of the detected light corresponding to a particular type of metrology measurement. In such embodiments, the process logic may identify a particular wavelength based on the output of the wave analysis model described above. For example, the process logic may provide data associated with the structure of the substrate (e.g., CD, thickness, material properties, SWA, etc.) as input to the wave analysis model and extract one or more outputs. The process logic may determine, based on the one or more outputs, that a particular spectral range corresponds to a particular type of metrology measurement, and that wavelengths X and Y are included in the particular range. In additional or alternative embodiments, the output of the wave analysis model may indicate that wavelengths X and Y correspond to a particular type of metrology measurement. In response to identifying a particular wavelength corresponding to a particular type of metrology measurement, the process logic may extract a set of spectral data corresponding to wavelengths X and Y from the regularized spectral data.

[0103] In some embodiments, the spectral feature set includes spectral trends or patterns present in spectral data at one or more wavelengths corresponding to a particular type of metrology measurement. The process logic may perform one or more analysis operations (e.g., RCWA, two-dimensional scanning operations, etc.) on the normalized spectral data at the one or more wavelengths to identify the spectral trends or patterns corresponding to the particular type of metrology measurement. In some embodiments, the process logic may identify a portion of the normalized spectral data at a particular wavelength that is associated with a particular type of metrology measurement. For example, the process logic may identify that the normalized spectral data at wavelength Y between the initial time period and the first intermediate time period is associated with a critical dimension measurement. Fig. 7EThe data range 710 indicates such identified normalized spectral data. In other or similar embodiments, the process logic may identify portions of the normalized spectral data that indicate spectral trends or patterns common to each wavelength associated with a particular type of metrology measurement. For example, the process logic may identify spectral trends common to wavelength X and wavelength Y. Fig. 7E Data ranges 712, 714, and 716 indicate such identified normalized spectral data.

[0104] In other or similar embodiments, the process logic may identify a portion of the regularized spectral data corresponding to a specific type of measurement value based on the results of a two-dimensional scan of a calibration substrate. A calibration substrate may refer to a substrate including one or more profile features corresponding to a target profile feature, the target profile feature being associated with a specific substrate process. The calibration substrate may be placed at a scanning device (e.g., a metrology device or other device) configured to scan the surface of the substrate and collect spectral data associated with the topography of the substrate profile. The collected spectral data may include a set of spectral data, each of which is associated with a specific portion or region of the substrate surface. The process logic may obtain data of the calibration substrate, and may identify a portion or region of the calibration substrate based on the set of spectral data, the portion or region including a profile pattern that is different from a profile pattern at other portions of the substrate surface. The process logic may identify a spectral trend of the regularized spectral data of wavelengths X, Y, and / or Z, the regularized spectral data corresponding to the spectral data of the different profile patterns collected during the two-dimensional scan of the calibration substrate.

[0105] Back to Figure 6 At block 622, the process logic may identify a subset of spectral data associated with a specific time period of the substrate process from the extracted set of spectral data. In some embodiments, the specific time period of the substrate process may include a final time period of the substrate process and / or one or more intermediate time periods of the substrate process adjacent to the final time period. In other or similar embodiments, the specific time period of the substrate process may include a time period that is closest to a current time period of the current substrate process. Figure 7F A subset 718 of spectral data associated with a final time period of an etch process (ie, time N) and an adjacent intermediate time period of the etch process (ie, time N-1) is illustrated. The process logic may identify the subset 718 based on the data range 710-716 extracted from the normalized spectral data.

[0106] As described above, one or more operations of method 600 may be performed by training set generator 324. For example, training set generator 324 may perform operations 610-622 to obtain subset 718. Figure 4In the embodiment provided, subset 718 may be included in the training input generated for training machine learning model 334. In other or similar embodiments, one or more operations of method 600 may be performed by prediction engine 328. For example, prediction engine 328 may perform operations 610 and 618-622 to obtain subset 718. Figure 8 In the described embodiment, subset 718 can be provided as input to the trained machine learning model 334.

[0107] Figure 8 8 is a flow chart of a method 800 for estimating a metrology value of a profile of a substrate using a machine learning model according to aspects of the present disclosure. The method 800 is performed by process logic, which may include hardware (circuitry, dedicated logic, etc.), software (such as software 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 Figure 1 In other or similar embodiments, one or more operations of method 800 may be performed by one or more other machines not depicted in the figure. In some aspects, one or more operations of method 800 may be performed by a computer system architecture 100. Figure 3 The prediction engine 328 of the server machine 320 is depicted as executing.

[0108] At block 810, process logic receives spectral data associated with a substrate being processed at a process chamber of a manufacturing system. In some embodiments, the spectral data may be received from an endpoint detection device (such as endpoint detection device 124) or other sensor disposed within or coupled to the process chamber in accordance with previously described embodiments. At block 812, one or more subsets of the normalized spectral data are obtained that include indications of spectral features corresponding to a particular type of metrology measurement. In some embodiments, the process logic may perform a process for determining the spectral features of the substrate being processed at a process chamber of a manufacturing system. Figure 6 The process logic may further include performing one or more operations (e.g., operations 618-622) described in method 600 to obtain the one or more subsets of the regularized spectral data. At block 816, the process logic provides the extracted one or more subsets of the regularized spectral data as input to the trained machine learning model. In some embodiments, the trained machine learning model may be associated with a machine learning model. Figure 3 The described machine learning model 334 corresponds to. In some embodiments, according to the following Fig. 9In the described embodiment, the training engine 326 selects the machine learning model 334 for use by the prediction engine 328. At block 818, the process logic obtains one or more outputs of the machine learning model. At block 820, the process logic extracts metrology measurement data from the one or more outputs, the metrology measurement data identifying: (i) one or more metrology measurement values ​​associated with a particular type of metrology measurement, and (2) an indication of a confidence level that each of the one or more metrology measurement values ​​corresponds to a profile of the substrate. In one example, the confidence level is a real number between 0 and 1 (inclusive). It should be noted that in some embodiments, the confidence level does not correspond to a probability. For example, the sum of the confidence levels of all metrology measurement values ​​may not be equal to 1.

[0109] In some embodiments, the process logic may use the metrology measurement data to provide a metrology measurement value for a substrate being processed at the manufacturing system. In some embodiments, if a confidence level of the metrology measurement value satisfies a threshold condition, the substrate is identified as being associated with the metrology measurement value. The process logic may determine that the confidence level of the metrology measurement value satisfies the threshold condition in response to determining that the confidence level exceeds a threshold level of confidence. Fig.10 In the described embodiment, the process logic may provide metrology measurements to the process endpoint engine 330 .

[0110] Fig. 9 is a flow chart of a method 900 for selecting a machine learning model for estimating a certain type of metrology measurement value according to aspects of the present disclosure. The method 900 is performed by process logic, which may include hardware (circuitry, dedicated logic, etc.), software (such as software running on a general-purpose computer system or a dedicated machine), firmware, or some combination of the above. In one embodiment, the method 900 may be performed by a computer system (such as Figure 1 In other or similar embodiments, one or more operations of method 900 may be performed by one or more other machines not depicted in the figure. In some aspects, one or more operations of method 900 may be performed by a computer system architecture 100. Figure 3 The training engine 326 of the server machine 320 or 350 is depicted for execution.

[0111] At block 910, the process logic receives training data and / or validation data for a set of machine learning models. Each of the set of machine learning models may correspond to a different machine learning model type. For example, each of the set of machine learning models may correspond to 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. In some embodiments, the process logic may receive the training data and validation data from a training set generator 324 of a server machine 320 or 350. The training set generator 324 may generate a training set based on the training data and validation data. Figure 4 In other or similar embodiments, the process logic may receive training data from the training set generator 324 and may generate validation data based on the received training data. The training data may be associated with Figure 3 The described training set 340 corresponds, and the validation data may correspond to the data included in the validation set 342 .

[0112] At box 912, the process logic uses the received training data to train each of the set of machine learning models. At box 914, the process logic uses the verification data to perform one or more test operations on each of the set of machine learning models. As described above, the verification set 342 may include spectral data and metrology data of previous substrates processed at the process chamber, which are different from the spectral data and metrology data included in the training data. In order to perform the one or more test operations, the process logic may provide the spectral data of the verification set 342 as input to each of the trained set of machine learning models, and one or more outputs of each of the trained models may be obtained. According to the embodiments provided herein, the process logic may extract metrology measurements from the one or more outputs obtained.

[0113] At block 916, the process logic assigns a performance rating to each of the set of machine learning models based on the results of the one or more test operations performed with respect to block 914. In some embodiments, the process logic may assign a performance rating to the corresponding machine learning model based on the accuracy score determined for the machine learning model. The process logic may determine the accuracy score based on the difference between the metrology measurement value extracted from the output of the corresponding machine learning model and the actual metrology measurement value of the substrate being below a threshold difference value. For example, as described with respect to block 914, the process logic may provide spectral data according to the validation set 342 as input to the corresponding machine learning model, and may extract metrology values ​​from the output of the model. The process logic may compare the extracted metrology value with the actual metrology measurement value associated with the provided spectral data of the validation set 342. The process logic may assign an accuracy score to the model based on the difference between the extracted value and the actual value of the validation set 342. For example, if the difference between the extracted value generated by the model and the actual value is small, then the process logic may assign a high accuracy score to the corresponding model. Similarly, if the difference is large, then the process logic may assign a low accuracy score to the corresponding model.

[0114] In additional or alternative embodiments, the process logic may further assign a performance rating to the corresponding machine learning model based on the speed score determined for the machine learning model. In some embodiments, the process logic may determine the speed score based on the amount of time it takes to obtain the one or more outputs of the model after the process logic provides the spectral data as an input to the model. In other or similar embodiments, the process logic may determine the speed score based on the amount of time it takes the process logic to extract metrological measurements from the one or more obtained outputs after the process logic provides the spectral data as an input to the model. In one example, if the amount of time it takes for the model to provide the one or more outputs (or for the process logic to extract metrological measurements) after the process logic provides the spectral data as an input to the model is small, then the process logic may assign a high speed score to the corresponding model.

[0115] In some embodiments, the process logic may assign a performance rating to a corresponding machine learning model based on the efficiency score determined for the machine learning model. Figure 3As described, in some embodiments, the training engine 356 may be included at a server machine 350 that is separate from the server machine 320. In such embodiments, the training engine 356 may train each of the set of machine learning models at the server machine 350. The training engine 356 may perform the one or more test operations described with respect to block 914 to determine which of the set of machine learning models to transfer to the server machine 320 for use by the prediction engine 328. The process logic may determine an efficiency score for the respective machine learning model based on an overall system efficiency (e.g., an overall system efficiency of a manufacturing system) with respect to transferring the respective machine learning model from the server machine 350 to the server machine 320 and / or initializing the respective machine learning model at the server machine 320. In some embodiments, the process logic may determine the efficiency score based on an amount of memory used to store the respective trained machine learning model, an amount of memory available at the data store 332 of the server machine 320, an amount of network bandwidth available for transferring the respective trained machine learning model to the server machine 320, and the like.

[0116] At block 918, the process logic determines whether the performance criteria are met based on the performance rating assigned to each of the machine learning models. In some embodiments, the process logic may determine whether the performance criteria are met by determining whether the assigned performance rating (i.e., a performance rating determined based on the accuracy score, and in some embodiments, also based on the speed score and / or efficiency score) exceeds a threshold performance score. In other or similar embodiments, the process logic may determine whether the performance criteria are met by determining whether the accuracy score of the corresponding model exceeds the threshold score and whether the overall performance rating of the corresponding model (i.e., an overall performance rating determined based on the accuracy score and the speed score and / or efficiency score) exceeds the threshold performance score. In some embodiments, more than one trained machine learning model may be associated with an accuracy score and / or performance rating that satisfies the threshold score and / or threshold rating, respectively. In such an embodiment, the process logic may determine that the corresponding model of the more than one trained model is associated with the highest accuracy score and / or performance rating that satisfies the performance criteria. In some embodiments, the process logic may determine that none of the trained machine learning models in the set of machine learning models are associated with an accuracy score and / or performance rating that satisfies the threshold score and / or threshold rating, respectively. In such an embodiment, the process logic may determine that no model meets the performance criteria.

[0117] In response to determining that the performance criteria are satisfied, the process logic proceeds to block 920. At block 920, the process logic selects a corresponding machine learning model to be applied to future spectral data collected for future substrates processed according to the future substrate process. As described above, in some embodiments, the training engine 326 may be included with the prediction engine 328 at the server machine 320. In such embodiments, in response to selecting a corresponding machine learning model to apply to the future spectral data, the process logic may store the corresponding model and / or an indication that the corresponding model is to be used by the prediction engine 328 at the data storage 332. In other embodiments, the training engine 326 may be included at the server machine 350. In such embodiments, in response to the training engine 326 selecting a corresponding machine learning model to be applied to the future spectral data, the server machine 350 may transmit the corresponding machine learning model to the server 320 for storage at the data storage 150.

[0118] In response to determining that the performance criteria are not met, the process logic proceeds to block 922. At block 922, the process logic receives additional training data for further training the machine learning model set. According to the previously described embodiments, the process logic may receive additional training data from the training set generator 324.

[0119] Fig.10 is a flow chart of a method 1000 for detecting an endpoint of a substrate process using a machine learning model according to aspects of the present disclosure. The method 1000 is performed by process logic, which may include hardware (circuitry, dedicated logic, etc.), software (such as software running on a general purpose computer system or a dedicated machine), firmware, or some combination thereof. In one embodiment, the method 1000 may be performed by a computer system (such as Figure 1 In other or similar embodiments, one or more operations of method 1000 may be performed by one or more other machines not depicted in the figure. In some aspects, one or more operations of method 1000 may be performed by a computer system architecture 100 of the present invention. Figure 3 The process endpoint engine 330 of the server machine 320 is depicted as executing.

[0120] At box 1010, process logic receives spectral data associated with the current substrate processed according to the current process at the manufacturing system 200. In some embodiments, as previously described, the process logic can receive the current spectral data from the endpoint detection equipment 124 or other sensors disposed in the process chamber performing the current process or coupled to the process chamber. In some embodiments, the process logic can receive the current spectral data at a specific time period of the current process. For example, the endpoint detection equipment 124 can be configured to collect spectral data for the current substrate at specific intervals (e.g., once per second) during the substrate process. The endpoint detection equipment 124 can collect spectral data at corresponding intervals of the substrate process and transmit the spectral data to the server machine 320, where the spectral data is received by the process logic of the process endpoint engine 330.

[0121] At block 1012, the process logic determines whether the substrate criteria are satisfied. In some embodiments, the process logic may determine that the substrate criteria are satisfied by determining that the received spectral data corresponds to spectral data associated with a particular type of substrate process and / or a particular type of substrate. For example, the process logic may retrieve (e.g., from data storage 150) spectral data previously collected for a particular type of substrate processed according to a particular type of process. The previously collected spectral data may include one or more spectral data features (e.g., spectral signatures) that are unique to the type of substrate and / or the type of process or a particular step or time period of the type of process. The process logic may determine whether the received spectral data for the current substrate includes one or more spectral data features that correspond to (i.e., are approximately equal to) corresponding spectral data features included in the previously collected spectral data.

[0122] In response to determining that the current spectral data signature does not correspond to the spectral signature of previously collected data, the process logic may determine that the substrate criteria are not met. The determination that the current spectral data signature does not correspond to the spectral signature of previously collected data may indicate that the current substrate is not associated with an expected substrate type for a substrate process, or that a current process or step of a current process does not correspond to an expected type of the current process or a current step. Therefore, in response to the process logic determining that the substrate criteria are not met, the method 1000 proceeds to block 1014, where the process logic transmits an alert to a client device of the manufacturing system 200. The alert may include a notification that the current substrate may not correspond to an expected type of the current substrate and / or that a current process or step of a process may not correspond to an expected type or step of a process.

[0123] In response to determining that the current spectral data features correspond to spectral features of previously collected data, the process logic may determine that the substrate criteria are met. In response to the process logic determining that the substrate criteria are met, the method 1000 proceeds to block 1016. At block 1016, the process logic provides the spectral data for use as input to the trained machine learning model. In some embodiments, the process logic may provide the spectral data to the prediction engine 328, which may cause the prediction engine 328 to perform a prediction on the substrate. Figure 8 One or more operations of method 800 are described.

[0124] At block 1018, the process logic obtains metrology measurements extracted from one or more outputs of the trained machine learning model. In some embodiments, the process logic may receive an indication of the metrology measurements obtained according to method 800 from the prediction engine 328. At block 1020, the process logic determines whether a metrology measurement criterion is satisfied. In some embodiments, the process logic may determine whether the metrology measurement criterion is satisfied by determining whether the extracted metrology measurement corresponds to a metrology measurement associated with an endpoint of the current process. If the difference between the extracted value and the endpoint value is below a difference threshold, then the extracted metrology measurement may correspond to a metrology measurement associated with an endpoint of the current process.

[0125] In response to the process logic determining that the metrology measurement criteria are satisfied, the method 1000 proceeds to block 1022. At block 1022, the process logic generates an instruction to terminate the current process at the manufacturing system. In some embodiments, the process logic may transmit an instruction to a system controller (e.g., system controller 228), which causes the system controller to terminate the current process. In other or similar embodiments, the process logic may transmit an instruction to a local processing device of a process chamber, which causes the process chamber to terminate the current process. In additional or alternative embodiments, the process logic may transmit the spectral data and the extracted metrology measurements to a training set generator 324 (i.e., a training set generator at server machine 320 or server machine 350) for use as additional training data.

[0126] In response to the process logic determining that the metrology measurement criteria are not met, the method 1000 proceeds to block 1024. At block 1024, the process logic continues the current process for the current substrate. In some embodiments, the process logic may continue the current process by transmitting an instruction to the system controller 228 or a local controller of the process chamber to continue the current process. In other or similar embodiments, the process logic may continue the current process by not generating and transmitting an instruction. As described above, the process logic may transmit the spectral data and the extracted metrology measurements to the training set generator 324 for use as additional training data.

[0127] Returning to box 1020, in some embodiments, the process logic may further determine whether the metrology measurement criteria are met by determining whether the extracted metrology measurement value falls within an expected metrology measurement value range associated with the time period of the current process. The expected metrology measurement value range may include a set of values ​​expected to be associated with the current substrate in the current time period of the current process. For example, if the expected (i.e., expected in view of the process data of the current substrate process and the previous substrate process) time period is the final time period of the current substrate process, then the expected metrology measurement value range may include the endpoint metrology value and the adjacent expected value (e.g., the endpoint value ±X). In response to determining that the expected metrology measurement value does not fall within the expected metrology measurement value range, the process logic may transmit an alert to a client device of the manufacturing system as described above. In some embodiments, the process logic may determine in view of the extracted metrology measurement value that the accuracy of the trained machine learning model no longer meets the accuracy criteria (i.e., the overall accuracy of the trained machine learning model is below a threshold overall accuracy). Therefore, the process logic may transmit a notification to the client device indicating that the machine learning model is to be retrained. In some embodiments, the process logic may transmit a notification to the training set generator 324 and / or the training engine 326 to retrain the machine learning model in accordance with embodiments described herein.

[0128] Fig.11 A diagrammatic representation of a machine in the example form of a computing device 1100 is depicted, within which a set of instructions for causing the machine to perform any one or more of the methodologies discussed herein may be executed. 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 as a server or client machine in a client and server network environment, or as a peer machine in a 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 a bridge, or any machine capable of executing a set of instructions (in sequence or otherwise) that specifies actions to be taken by the machine. Further, although only a single machine is shown, the term "machine" should also be considered to include a collection of any machines (e.g., computers) that execute an instruction set (or multiple instruction sets) individually or collectively to perform any one or more of the methodologies discussed herein. In embodiments, computing device 1100 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 described herein.

[0129] The example computing device 1100 includes a processing device 1102, a main memory 1104 (e.g., read-only memory (ROM), flash memory, dynamic random access memory (DRAM) such as synchronous dynamic random access memory (SDRAM), etc.), a static memory 1106 (e.g., flash memory, static random access memory (SRAM), etc.), and a secondary memory (e.g., a data storage device 1128) that communicate with each other via a bus 1108.

[0130] The processing device 1102 may represent one or more general purpose processors such as a microprocessor, a central processing unit, etc. More specifically, the processing device 1102 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 1102 may also be one or more special 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 1102 may also be or include a system on a chip (SoC), a programmable logic controller (PLC), or other types of processing devices. The processing device 1102 is configured to execute processing logic to perform the operations and steps discussed herein.

[0131] The computing device 1100 may further include a network interface device 1122 for communicating with a network 1164. The computing device 1100 may also include a video display unit 1110 (e.g., a liquid crystal display (LCD) or a cathode ray tube (CRT)), an alphanumeric input device 1112 (e.g., a keyboard), a cursor control device 1114 (e.g., a mouse), and a signal generating device 1120 (e.g., a speaker).

[0132] The data storage device 1128 may include a machine-readable storage medium (or more specifically, a non-transitory computer-readable storage medium) 1124 on which is stored one or more instruction sets 1126 that implement any one or more of the methodologies or functions described herein. Non-transitory storage media refers to storage media other than carrier waves. The instructions 1126 may also reside completely or at least partially in the main memory 1104 and / or the processing device 1102 during the execution of the instructions by the computer device 1100, and the main memory 1104 and the processing device 1102 also constitute computer-readable storage media.

[0133] Although the computer-readable storage medium 1124 is shown as a single medium in the example embodiment, the term "computer-readable storage medium" should also be considered to include a single medium or multiple media (such as a centralized or distributed database and / or associated caches and servers) storing the one or more instruction sets. The term "computer-readable storage medium" should also be considered to include any medium capable of storing or encoding an instruction set for execution by a machine and causing the machine to perform any one or more of the methodologies of the present disclosure. Therefore, the term "computer-readable storage medium" should be considered to include, but is not limited to, solid-state memories and optical and magnetic media.

[0134] The foregoing description sets forth many specific details such as examples of specific systems, components, methods, etc., so that several embodiments of the present disclosure are well understood. However, it will be appreciated by those skilled in the art that at least some embodiments of the present disclosure may be implemented without these specific details. In other cases, well-known components or methods are not described in detail, or are presented in a simple block diagram format to avoid unnecessarily obscuring the present disclosure. Therefore, the specific details set forth are merely exemplary. Specific implementations may differ from these exemplary details and are still considered to be within the scope of the present disclosure.

[0135] References throughout this specification to "one embodiment" or "an embodiment" mean that a particular feature, structure, or characteristic described in connection with the embodiment is included in at least one embodiment. Thus, the phrases "in one embodiment" or "in an embodiment" appearing in various places throughout this specification are not necessarily all referring to the same embodiment. Furthermore, the term "or" is intended to mean an inclusive "or" rather than an exclusive "or". When the term "about" or "approximately" is used herein, this term is intended to mean that the nominal value presented is accurate to within ±10%.

[0136] Although the operations of the methods herein are shown and described in a particular order, the order of operations of each method may also be changed so that certain operations may be performed in reverse order so that certain operations may be performed at least partially in parallel with other operations. In another embodiment, instructions or sub-operations of different operations may be performed in an intermittent and / or alternating manner.

[0137] It should be understood that the above description is intended to be illustrative, rather than restrictive. After reading and understanding the above description, those skilled in the art will find many other embodiments. Therefore, the scope of this disclosure will be determined with reference to the entire scope of the equivalents given by the accompanying claims and such claims.

Claims

1. A method for substrate process detection using machine learning, the method comprising: identifying a plurality of machine learning models, each of the machine learning models trained to predict metrology measurements based on spectral data collected during a substrate process performed on the substrate; assigning a performance rating to each of the plurality of machine learning models based on an accuracy of the metrology measurements provided by the corresponding machine learning model; as well as Given the performance rating of each of the plurality of machine learning models, selecting the corresponding machine learning model to be applied to future spectral data collected during a future substrate process performed for a future substrate.

2. The method of claim 1, wherein each of the plurality of machine learning models is associated with a different type of machine learning model in a set of machine learning types.

3. The method of claim 1, further comprising: The accuracy of the metrology measurement value provided by the corresponding machine learning model is determined in view of an actual measured value of the metrology measurement, the actual measured value generated based on historical metrology data collected by metrology equipment for a previous substrate in a previous set of substrates.

4. The method of claim 3, wherein determining the accuracy of the metrology measurements provided by the corresponding machine learning model comprises: obtaining a validation data set for the set of previous substrates, the validation data set comprising (i) historical spectral data collected during previous substrate processes performed on other previous substrates in the set of previous substrates, and (ii) the actual measured values ​​of the metrology measurements; providing the historical spectral data as input to each of the plurality of machine learning models; extracting the metrology measurements from one or more outputs of the corresponding machine learning models; as well as A difference between the extracted metrological measurement and the actually measured value of the metrological measurement is determined.

5. The method of claim 1, further comprising: receiving historical spectral data collected during a previous substrate process performed on another previous substrate in the set of previous substrates; obtaining historical metrology measurements for the previous substrate, the historical metrology measurements generated based on additional historical metrology data collected by the metrology equipment for the previous substrate; as well as Generating the training data to train each of the plurality of machine learning models, wherein generating the training data comprises: generating a training input including the historical spectral data; as well as generating a target output for the training input, wherein the target output comprises the historical metrology measurements for the previous substrate, And wherein providing the training data to train each of the multiple machine learning models includes providing: (i) a training input set including the training inputs; and (ii) a target output set including the target outputs.

6. The method of claim 5, wherein the historical spectral data is received from at least one of a system controller of a manufacturing system that performed the previous substrate process, or an endpoint detection system of the manufacturing system.

7. The method of claim 5, wherein the future process is to be performed on the future substrate at a first manufacturing system and the previous process is performed on the previous substrate at a second manufacturing system, the second manufacturing system being different from the first manufacturing system.

8. The method of claim 1 , wherein the performance rating of each of the plurality of machine learning models is further assigned based on at least one of: a rate at which the corresponding machine learning model provides extracted metrology measurements, or an amount of time to transmit the corresponding machine learning model to a processing device of a manufacturing system configured to perform the future substrate process for the future substrate.

9. A system, comprising: Memory; as well as A set of one or more processing devices, coupled to the memory, the set of one or more processing devices being used for: receiving current spectral data associated with a current substrate being processed in a manufacturing system; identifying a machine learning model trained to predict a value of a metrology measurement based on given spectral data associated with the substrate, wherein the machine learning model is associated with a performance rating that satisfies a performance criterion associated with the metrology measurement; providing the current spectral data associated with the current substrate as input to the machine learning model; as well as Given the current spectral data, current values ​​of the metrology measurements for the current substrate are extracted from one or more outputs of the machine learning model.

10. The system of claim 9, wherein the set of one or more processing devices is further configured to: In response to determining that the corresponding metrology measurement values ​​satisfy metrology measurement criteria associated with the current process, instructions are generated, the instructions including a command to terminate the current process at the manufacturing system.

11. The system of claim 10, wherein to determine that the corresponding metrological measurement value satisfies the metrological measurement criterion, the set of one or more processing devices is configured to: The corresponding metrology measurement is determined to correspond to a target metrology measurement associated with the current process for the current substrate.

12. The system of claim 9, wherein the set of one or more processing devices is part of a system controller of a manufacturing system that performs the current process on the substrate, and wherein to identify the machine learning model, the set of one or more processing devices is used to: The machine learning model is received from one or more processing devices of a computing system coupled to the system controller.

13. The system of claim 9, wherein to identify the machine learning model, the set of one or more processing devices is configured to: Training data is provided to train the machine learning model to provide the current value of the metrology measurement.

14. The system of claim 9, wherein the set of one or more processing devices is further configured to: obtaining the current spectral data collected during the current process of the current substrate; and In response to determining that the current spectral data does not correspond to target spectral data associated with the current substrate, an error notification is transmitted to a client device coupled to the manufacturing system.

15. A non-transitory computer readable medium comprising instructions that, when executed by a set of one or more processing devices, cause the set of one or more processing devices to: obtaining first spectral data collected for a first substrate during a first substrate process performed in a processing chamber; identifying metrology data comprising one or more actual measured values ​​of metrology measurements associated with the first substrate during an endpoint of a first substrate process; generating composite metrology data comprising one or more composite values ​​of the measured metrology during the first substrate process based on the first spectral data; extracting one or more spectral features from the first spectral data based on the identified metrology data and the generated synthetic metrology data in view of the type associated with the metrology measurement, the one or more spectral features corresponding to an endpoint of the substrate process; as well as An endpoint of a second substrate process performed on the second substrate is determined based on the second spectral data obtained for the second substrate and the extracted one or more spectral features.

16. The non-transitory computer readable medium of claim 15, wherein to generate the synthetic metrology data, the set of one or more processing devices is configured to: identifying, from the obtained first spectral data, one or more first spectral data items corresponding to an endpoint of the first substrate process and one or more second spectral data items corresponding to a time period before the endpoint of the first substrate process; and Based on the one or more actually measured values ​​of the metrology measurement and the difference between the one or more first spectral data items and the one or more second spectral data items, a composite value of the metrology measurement during the time period is calculated.

17. The non-transitory computer readable medium of claim 15, wherein the set of one or more processing devices is further configured to: The extracted one or more spectral features are provided for training a machine learning model to predict an endpoint of a current substrate process being performed on the current substrate.

18. The non-transitory computer readable medium of claim 17, wherein to provide the extracted one or more spectral features for training the machine learning model, the set of one or more processing devices is further configured to: The extracted one or more spectral features are transmitted to a computing device associated with a training process of the machine learning model.

19. The non-transitory computer readable medium of claim 17, wherein to provide the extracted one or more spectral features for training the machine learning model, the set of one or more processing devices is further configured to: generating a first mapping between the extracted one or more spectral features and an indication that the extracted one or more spectral features are associated with the endpoint of the first substrate process; generating a second mapping between one or more additional spectral features of the first spectral data and an indication that the one or more additional spectral features are not associated with an endpoint of a first substrate process; and A training data set of the machine learning model is updated to include the generated first mapping and the generated second mapping.

20. The non-transitory computer readable medium of claim 17, wherein to determine an endpoint of a second substrate process performed on a second substrate, the set of one or more processing devices is further configured to: providing the second spectral data as input to the machine learning model; and Obtaining one or more outputs of the machine learning model, wherein: The one or more outputs include an indication of a confidence that a current operation of the second substrate process corresponds to the endpoint of the second substrate process.

Images