Accelerating preventive maintenance recovery and recipe optimization using machine learning-based algorithms
Machine learning models predict and adjust chamber conditions in semiconductor manufacturing, addressing inefficiencies in preventive maintenance by enabling real-time monitoring and rapid recovery, thus enhancing productivity and consistency.
Patent Information
- Application Number
- JP2024529351
- Authority / Receiving Office
- JP · JP
- Patent Type
- Applications
- Current Assignee / Owner
- Priority Date
- 2021-11-23
- Filing Date
- 2022-11-18
- Publication Date
- 2024-11-08
AI Technical Summary
Existing methods for determining chamber conditions in semiconductor manufacturing facilities after preventive maintenance are inefficient, leading to unpredictable processing results and prolonged downtime due to the need for extensive seasoning procedures and reliance on offline test samples.
Employing machine learning models trained on historical sensor data, including optical emission spectra, to predict and adjust chamber conditions in real-time, enabling in-situ monitoring and rapid recovery of processing chambers.
Facilitates consistent processing results by reducing downtime and improving productivity through real-time chamber condition prediction and parameter adjustments, minimizing the need for excessive seasoning and offline testing.
Smart Images

Figure 2024541406000001_ABST
Abstract
Description
[Technical field]
[0001] FIELD OF THE DISCLOSURE Embodiments of the present disclosure relate generally to predicting chamber conditions in a manufacturing system. In particular, the present disclosure relates to determining chamber conditions, identifying chamber recovery states, and / or updating process parameters.
[0002] Substrate processing may include a series of operations that fabricate electrical circuits in semiconductors (e.g., silicon wafers) according to a circuit design. These operations may be performed in a series of chambers. Successful operation of a modern semiconductor manufacturing facility may aim to facilitate a steady flow of wafers moving from one chamber to another during the formation of electrical circuits in the wafers. In the process of performing many substrate processes, the conditions in the processing chambers may change, resulting in processed substrates that do not meet the desired conditions and results.
[0003] One such substrate processing may include plasma etching, which is a process that transfers a pattern in a layer of mask material to another layer (e.g., a layer of conductive or dielectric material) beneath the mask by removing the layer of material from the wafer surface. Such processes inevitably produce various types of etch by-products (e.g., silicon oxide, organic polymers, etc.) depending on the layer material and the etch chemistry. Some of the by-products deposit, for example, on the interior surfaces of the chamber in which the plasma etch process is performed. The deposition of by-products can affect etch performance by depositing particles (e.g., flakes) on the substrate or by reacting with the plasma to affect the process results.
[0004] To mitigate the effects of etch by-products, preventive maintenance (e.g., chamber cleaning, etc.) can be used to periodically remove deposits from the chamber walls. An example of preventive maintenance can include removing the chamber from production and introducing a cleaning plasma (e.g., CF4+O2 plasma, etc.) into the chamber to clean silicon oxide deposited during silicon etching. This plasma reacts with the deposited material, and products of this reaction are exhausted from the chamber. However, it has been observed that after such chamber cleaning, if the chamber walls are clean, the chamber becomes unsuitable for etching immediate production wafers. Chamber seasoning is the procedure of etching a series of substrates (e.g., blank silicon wafers) to restore chamber conditions suitable for production substrate processing. After chamber seasoning, a thin layer of silicon oxide can coat the chamber walls. The chamber then returns to etching product wafers until the next round of chamber cleaning and seasoning is performed. Preventive maintenance can also include removing dirt and / or deposits by physical methods (e.g., wiping one or more surfaces of the processing chamber). Overview
[0005] The following is a simplified summary of the disclosure in order to provide a basic understanding of some aspects of the disclosure. This summary is not an extensive overview of the disclosure. It is not intended to identify key or critical elements of the disclosure, nor is it intended to identify the scope or claims of particular implementations of the disclosure. Its sole purpose is to present some concepts of the disclosure in a simplified form as a prelude to the more detailed description that is presented below.
[0006] In an exemplary embodiment, a method includes a processing device receiving sensor data including chamber data indicative of a condition of an environment of a processing chamber processing a substrate according to a set of processing parameters of a current process. The chamber data further includes spectral data indicative of optical emission spectroscopy (OES) measurements of a plasma disposed in the processing chamber. The processing chamber processes the substrate according to the set of processing parameters of the current process. The method further includes using the sensor data as an input to a machine learning model. The method further includes obtaining one or more outputs of the machine learning model. The one or more outputs are indicative of one or more chamber condition metrics. The method further includes determining a recovery state of the processing chamber based on the one or more chamber condition metrics, the recovery state being associated with a chamber recovery process (e.g., a chamber seasoning procedure) performed after a preventive maintenance procedure. The method further includes altering a performance of the processing chamber based on the recovery state of the processing chamber.
[0007] In an exemplary embodiment, a method is provided for training a machine learning model to determine a state of a processing chamber in a chamber recovery procedure. The processing chamber processes a current substrate according to a current process. The method includes generating training data for the machine learning model. Generating the training data includes identifying a first training input having historical sensor data including historical chamber data. The historical chamber data is indicative of a state of an environment of a second processing chamber processing a previous substrate according to a previous process. The sensor data further includes historical spectral data indicative of optical emission spectroscopy (OES) measurements of a previous plasma disposed in the second processing chamber processing a previous substrate according to a previous process. Generating the training data further includes identifying a first target output for the first training input. The first target output includes historical process result data having process result measurements of a previous substrate processed using the second processing chamber according to a previous process. The method further includes providing training data for training the machine learning model based on a set of training inputs including the first training input and a set of target outputs including the first target output.
[0008] In an exemplary embodiment, a non-transitory computer readable medium includes instructions that, when executed by a processing device, cause the processing device to perform actions. The actions that are performed include receiving sensor data including chamber data indicative of a condition of an environment of a processing chamber that processes a substrate according to a set of processing parameters of a current process. The chamber data further includes spectral data indicative of optical emission spectroscopy (OES) measurements of a plasma disposed in the processing chamber. The processing chamber processes the substrate according to the set of processing parameters of the current process. The actions further include using the sensor data as an input to a machine learning model. The method further includes obtaining one or more outputs of the machine learning model. The one or more outputs are indicative of one or more chamber condition metrics. The actions further include determining a recovery state of the processing chamber based on the one or more chamber condition metrics, the recovery state being associated with a chamber recovery process that is performed following a preventive maintenance procedure. The actions further include modifying a performance of the processing chamber based on the recovery state of the processing chamber. [Brief description of the drawings]
[0009] The present disclosure is illustrated by way of example, and not by way of limitation, in the accompanying drawings in which like reference numerals indicate like elements. It should be noted that different references to "an" or "one" embodiment in the present disclosure do not necessarily refer to the same embodiment, and such references mean at least one. [Figure 1] 1 illustrates an exemplary system architecture according to an aspect of the present disclosure. [Diagram 2] FIG. 1 illustrates a top schematic view of an exemplary manufacturing system according to aspects of the present disclosure. [Diagram 3] 1 illustrates an exemplary system architecture for predicting chamber conditions of a processing chamber in accordance with an aspect of the present disclosure. [Figure 4] 1 is a flowchart of a method for training a machine learning model according to an aspect of the present disclosure. [Diagram 5]2 is a cross-sectional schematic side view of a chamber condition apparatus in accordance with an aspect of the present invention; [Figure 6] 1 is a flowchart of a method for training and / or updating a machine learning model for predicting chamber conditions of a processing chamber in accordance with an aspect of the present disclosure. [Figure 7] 1 is a flowchart of a method for training and / or updating a machine learning model for predicting chamber conditions of a processing chamber in accordance with an aspect of the present disclosure. [Figure 8] 1 is a flowchart of a method for predicting chamber states of a processing chamber processing a current chamber using a machine learning model, according to an aspect of the present disclosure. [Figure 9] 1 is a flowchart of a method for selecting a machine learning model for estimating a type of metrology measurement, according to an aspect of the present disclosure. [Figure 10] 1 shows a diagrammatic representation of a machine, an exemplary form of a computing device capable of executing a set of instructions for causing the machine to perform any one or more of the methods described herein. Detailed Description of the Embodiments
[0010] The disclosed embodiments relate to a spectral measurement system for enhancing spectral data collection and predicting chamber condition metrics using machine learning. The process results of a manufacturing process depend on many factors, one of which is the process recipe and / or one or more chamber parameter settings. Another factor that frequently occurs is a process shift due to preventive maintenance (PM) or other events, which causes the process results to vary even if the recipe is the same. A newly serviced chamber (post-PM) or cold chamber often has a "first wafer effect", where the process results of a substrate processed using a newly serviced or cold chamber deviate from the process results of a substrate processed in a chamber that has not been recently serviced. The effects of PM can be ameliorated by performing one or several seasoning wafer runs to season the chamber to a "normal" operating condition. In some cases and in some applications, a post-PM chamber takes days / weeks to reach the target "normal" operating condition.
[0011] Preventive maintenance procedures (e.g., chamber cleaning) are often used as part of a chamber recovery process to restore processing chamber conditions to conditions suitable for entering a substrate processing production mode (e.g., processing a large volume of substrates). Recovery procedures are often used to prepare the chamber for production mode (e.g., to "warm up" the chamber) after a preventive maintenance procedure. A common recovery procedure used in the past is process chamber seasoning. Chamber seasoning is a procedure in which a series of substrates (e.g., blank silicon wafers) are processed to restore (e.g., coat the chamber walls) chamber conditions suitable for production substrate processing (e.g., substrates processed in the chamber have process results that meet a desired threshold criterion). After chamber seasoning, the chamber can be operated in production mode for a period of time until another round of preventive maintenance and further chamber seasoning is performed or recovery of the processing chamber conditions is recommended.
[0012] Penalties for "abnormal" chamber conditions may include the inability of substrate processing to produce substrates that meet target process results (e.g., film depth, critical dimensions on the wafer, sidewall angles, etc.). For example, a process recipe may indicate a processing time, and an "abnormal" condition would prevent the substrate from meeting the target process result within the indicated recipe time. Chamber anomalies may cause process / etch results to deviate from the target process results across an entire batch of substrates. For example, abnormal chamber conditions may cause the etch pattern of processed substrates to deviate from the target etch pattern. Failure to achieve the target etch pattern on the surface of the substrate may reduce the yield of integrated circuit (IC) chips ultimately manufactured from the substrate.
[0013] Knowing the chamber's current chamber state (e.g., chamber condition metrics) allows a manufacturing system to adaptively modify process parameters to achieve the target process results. However, identifying the chamber state can be challenging. This difficulty is due in part to subtle differences between "normal" and "abnormal" states of chamber parameters. For example, optical emission spectra (OES) may differ slightly after a chamber recently underwent preventive maintenance (PM). However, predicting the impact of these subtle differences on process results remains complex.
[0014] Traditionally, a series of substrates are processed in a processing chamber after a preventive maintenance procedure. Following the processing of the series of substrates, a test sample is processed and evaluated. The test sample may include substrates that are processed in the processing chamber and evaluated (e.g., the process results are measured) to determine whether the test sample includes process results that meet a threshold condition. If the condition is not met, additional seasoning substrates may be processed and other test samples may be processed to determine whether the chamber has fully recovered. This process may be repeated until the chamber condition is met or a threshold amount of seasoning wafers have been used. In traditional methods of determining chamber condition as described above, an initial amount of seasoning wafers is used, but an unnecessary amount of seasoning wafers (e.g., "over-seasoning" the chamber) may be required when in fact a smaller number is required to restore the chamber to an operable state to perform substrate manufacturing. Furthermore, traditional methods may require a delay in returning the chamber to a manufacturing mode pending the measurement and evaluation of the test samples necessary to determine whether the chamber has sufficiently recovered. The time taken for the recovery procedure results in lost productivity because the tool or chamber is down (e.g., not operating in a production mode). Furthermore, conventional methods do not allow for in-situ monitoring of the chamber condition but rely on test samples and await evaluation of the test samples during the seasoning process.
[0015] Aspects and embodiments of the present disclosure address these and other deficiencies of the prior art by utilizing machine learning to identify chamber condition metrics that are associated with the processing results of a substrate processed in the chamber. The application of machine learning and deep learning provides a method to identify unique features in data and build a correlation model between the data and the chamber label. Using the trained model, the processing chamber can provide in situ spectral data that can be used to predict and control real-time chamber conditions. This can be used, for example, to achieve consistent processing results even when chamber conditions vary. The chamber conditions can be mapped to the state of the processing chamber in a chamber recovery process. For example, preventive maintenance can be performed on the chamber and the chamber can undergo a recovery process (e.g., a chamber seasoning procedure) to return the chamber to a consistent operating state. Different stages of the recovery process can result in different chamber conditions, and the chamber conditions can be used to indicate at which stage of the recovery process the processing chamber is currently operating.
[0016] Aspects and embodiments of the present disclosure address these and other shortcomings of existing technologies by providing methods and systems of various embodiments that estimate chamber conditions without relying on test sample inspection results, provide quantitative information on chamber health and recovery after preventive maintenance procedures, and enable simulation of chamber conditions through artificial parameter adjustments. In some embodiments, the present disclosure enables health checks and preventative measures during chamber seasoning sequences and early detection of abnormal chamber conditions before sample testing. One or more of these aspects and / or embodiments can reduce downtime of processing tools (such as processing chambers) and increase productivity of manufacturing equipment. Additionally, chamber sensors or optical emission spectroscopy (OES) can be used to modify process parameters based on ML models. Modifying process parameters allows the chamber to return to production much more quickly without waiting for full seasoning and full validation of chamber conditions through metrology measurements.
[0017] Optical emission spectra (OES), reflectance spectra (e.g., endpoint detection interferometry (IEP), etc.), and chamber sensor data (e.g., temperature, pressure, radio frequency (RF) power, plasma state, vacuum state, etc.) can be indicators of chamber state. The chamber state changes after PM, when cold, and even after the chamber is seasoned. These changes cause process results (e.g., etch results, etc.) to change even if the same recipe is used. Various process results (e.g., yield, electrical measurements, metrology data, etc.) can be processed to identify appropriate labels for training a model. The chamber state labels (e.g., chamber state metrics) with associated OES and chamber sensor data can be used as training data for a machine learning model. In some embodiments, the machine learning type can include neural networks and / or can include logistic regression and multi-layer perceptrons. After the model is trained, validated, and tested, it can be applied to a prediction engine. The engine takes the in situ OES spectra and sensor data and predicts the chamber state. It can further revise / suggest optimal chamber parameters for current and / or future wafer processing.
[0018] Processing results (e.g., process yield, electrical measurements, metrology data, etc.) of substrates (e.g., from a production lot) can be used to examine the model's performance. If the results are not satisfactory, the chamber can be inspected, tested, and improved. Further training can be performed if the results are not satisfactory or if new chamber conditions are identified / discovered in a new lot run (e.g., to address more complete chamber conditions).
[0019] In an exemplary embodiment, the method includes a processing device receiving sensor data including chamber data indicative of a condition of an environment of a processing chamber that processes a substrate according to a set of processing parameters of a current process. Further, the chamber data includes spectral data indicative of an optical emission spectroscopy (OES) measurement of a plasma disposed in the processing chamber. The processing chamber processes a substrate according to the set of processing parameters of the current process. Further, the method includes using the sensor data as an input to a machine learning model. Further, the method includes obtaining one or more outputs of the machine learning model. The one or more outputs are indicative of one or more chamber condition metrics. Further, the method includes determining a recovery state of the processing chamber based on the one or more chamber condition metrics, the recovery state being associated with a chamber recovery process performed after a preventive maintenance procedure. Further, the method includes modifying a performance of the processing chamber based on the recovery state of the processing chamber. For example, the processing parameters of the current substrate or future substrates can be processed or further processed according to the modification of the processing chamber.
[0020] In an exemplary embodiment, a method for training a machine learning model for determining a state of a processing chamber in a chamber recovery procedure is provided. The method includes generating training data for the machine learning model. Generating the training data includes identifying a first training input having historical sensor data including historical chamber data. The historical chamber data is indicative of a state of an environment of a second processing chamber that processed a previous substrate according to a previous process. Additionally, the sensor data also includes historical spectral data indicative of optical emission spectroscopy (OES) measurements of a previous plasma disposed in the second processing chamber that processed the previous substrate according to a previous process. Additionally, generating the training data includes identifying a first target output for the first training input. The first target output includes historical process result data including process result measurements of a previous substrate processed using the second processing chamber according to a previous process. Additionally, the method includes providing training data for training the machine learning model based on a set of training inputs including the first training input and a set of target outputs including the first target output.
[0021] In some embodiments, the trained machine learning model receives new inputs having new sensor data, the new sensor data having new chamber data indicative of a new condition of a new environment in a new processing chamber processing a new substrate according to a new process, and new spectral data indicative of an optical emission spectroscopy (OES) measurement of a new plasma disposed in the new processing chamber processing the new substrate according to the new process, based on the new inputs, and generates new outputs indicative of chamber condition metrics corresponding to a recovery status associated with a chamber recovery process performed following a preventive maintenance procedure.
[0022] In an exemplary embodiment, a non-transitory computer readable medium includes instructions that, when executed by a processing device, cause the processing device to perform an action. The action that is performed includes receiving sensor data including chamber data indicative of a condition of an environment of a processing chamber that processes a substrate according to a set of processing parameters of a current process. For example, the processing chamber may be processing one or more seasoning wafers as part of a chamber recovery process. Further, the chamber data includes spectral data indicative of an optical emission spectroscopy (OES) measurement of a plasma disposed within the processing chamber. The processing chamber processes a substrate according to the set of processing parameters of the current process. Further, the action includes using the sensor data as an input to a machine learning model. Further, the method includes obtaining one or more outputs of the machine learning model. The one or more outputs are indicative of one or more chamber condition metrics. Further, the action includes determining a recovery state of the processing chamber based on the one or more chamber condition metrics, the recovery state being associated with a chamber recovery process (e.g., a chamber seasoning procedure) performed after a preventive maintenance procedure. Further, the action includes modifying a performance of the processing chamber based on the recovery state of the processing chamber. For example, the modification can be associated with a current substrate process, such as processing a current substrate according to updated processing parameters. The process parameters may include temperature, gas flow rates, gas pressure, power, electrical bias, etc. In another example, the modification may include stopping the current substrate processing. In another example, the modification may include modifying a process recipe used to process future substrates.
[0023] FIG. 1 illustrates an exemplary system architecture 100 according to an aspect of the disclosure. The system architecture 100 includes a client device 120, a manufacturing tool 122, a metrology tool 132, a prediction server 112 (e.g., for generating prediction data, providing model applications, using a knowledge base, etc.), and a data store 150. The prediction server 112 may be part of a prediction system 110. Furthermore, the prediction system 110 may include server machines 170 and 180. In some embodiments, the system architecture 100 may include or be part of a manufacturing system for processing substrates (e.g., such as the manufacturing system 200 of FIG. 2 ). In additional or alternative embodiments, the system architecture 100 may include or be part of a chamber state prediction system (e.g., for determining a chamber recovery state from a PM procedure). Further details regarding the chamber state prediction system are provided in connection with FIG. 3 .
[0024] The components of the client devices 120, the manufacturing equipment 122, the metrology equipment 132, the prediction system 110, and / or the data store 150 may be interconnected via a network 140. In some embodiments, the network 140 is a public network that provides the client devices 120 with access to the prediction server 112, the data store 150, and other publicly available computing devices. In some embodiments, the network 140 is a private network that provides the client devices 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 of a wide area network (WAN), a local area network (LAN), a wired network (e.g., an Ethernet network), a wireless network (e.g., an 802.11 network or a Wi-Fi network), a cellular network (e.g., a Long Term Evolution (LTE) network), a router, a hub, a switch, a server computer, a cloud computing network, and / or combinations thereof.
[0025] Client devices 120 include computing devices such as personal computers (PCs), laptops, mobile phones, smartphones, tablet computers, netbook computers, network-connected televisions ("smart TVs"), network-connected media players (e.g., Blu-ray players), set-top boxes, over-the-top (OTT) streaming devices, operator boxes, and the like.
[0026] The manufacturing tool 122 may run according to a recipe or over a period of time to produce a product. In some embodiments, the manufacturing tool 122 may include or be part of a process tool that includes one or more stations (e.g., processing chambers, transfer chambers, load locks, etc.) configured to perform different functions on the substrate. In some embodiments, the manufacturing tool 122 may further include a chamber state device 124 configured to collect data used to determine chamber states (e.g., chamber state metrics) of the processing chambers that perform processes on the substrate in the manufacturing tool 122. Chamber states may refer to the state of the processing chambers in a recovery process (e.g., a chamber seasoning procedure) after performing a PM procedure. The chamber state device 124 may include one or more components configured to collect and / or generate spectral data (e.g., OES or reflectometry spectra) associated with one or more portions of a profile of a surface of a substrate during substrate processing. Spectral data refers to data associated with the intensity (i.e., the strength or amount of energy) of the energy waves at each wavelength of the detected waves.
[0027] In some embodiments, the chamber state device 124 may include a fiber optic bundle and collimator assembly configured to direct incident light from a light source to a surface of the 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 chamber state device 124 may generate spectral data of a substrate profile based on the reflected light transmitted to the light detection component and / or spectral data associated with OES of the plasma generated in the processing chamber. In other or similar embodiments, the chamber state device 124 may include any sensor configured to generate spectral data associated with a substrate profile. Such sensors may include reflectometry sensors, ellipsometry sensors, thermal spectrum sensors, capacitance sensors, etc. Further details regarding the manufacturing tool 122 and the chamber state device 124 are described in conjunction with FIG. 2 and FIG. 5, respectively.
[0028] In some embodiments, one or more stations of the manufacturing equipment 122 may include sensors configured to generate and / or collect sensor data related to the manufacturing equipment 122. The sensor data may include one or more values of temperature (e.g., heater temperature), space (SP), pressure, high frequency radio frequency (HFRF), electrostatic chuck (ESC) voltage, current, flow rate, power, voltage, etc. The sensor data may be associated with or indicative of a manufacturing parameter, such as a hardware parameter, such as a setting or component (size, type, etc.) of the manufacturing equipment 122, or a processing parameter of the manufacturing equipment 122. The sensor data may be provided while the manufacturing equipment 122 is performing substrate processing. The sensor data may vary from substrate to substrate.
[0029] In some embodiments, the manufacturing tool 122 may include a metrology tool 126. The metrology tool 126 may be configured to generate metrology data related to substrates processed by the manufacturing tool 122. The metrology data may include one or more values of film property data (e.g., wafer spatial film properties), dimensions (e.g., thickness, height, etc.), dielectric constant, dopant concentration, density, defects, etc. In some embodiments, the metrology data may further include one or more values of surface profile property data (e.g., etch rate, etch rate uniformity, critical dimension of one or more features included in the surface of the substrate, uniformity of critical dimension across the surface of the substrate, edge placement error, etc.). The metrology data may be for finished or semi-finished products. The metrology data may vary from substrate to substrate.
[0030] The metrology tool 126 may be configured to generate metrology data related to a substrate before or after substrate processing. The metrology tool 126 may be integrated with a station of a process tool of the manufacturing tool 122. In some embodiments, the metrology tool 126 may be coupled to or part of a station of a process tool (e.g., a processing chamber, a transfer chamber, etc.) that is maintained in a vacuum environment. Such a metrology tool 126 is referred to as an integrated metrology tool 128. Thus, the substrate may be measured by the integrated metrology tool 128 while in a vacuum environment. For example, after performing a substrate process (e.g., an etch process, a deposition process, etc.) on the substrate, metrology data of the processed substrate may be generated by the integrated metrology tool 128 without removing the processed substrate from the vacuum environment. In other or similar embodiments, the metrology tool 126 may be coupled to or part of a process tool station (e.g., a factory interface module, etc.) that is not maintained in a vacuum environment. Such a metrology tool 126 is referred to as an in-line metrology tool 130. Thus, the substrate may be measured by the in-line metrology tool 130 outside of a vacuum environment.
[0031] In addition to or instead of the metrology tool 126, the system architecture 100 may include a metrology tool 132. The metrology tool 132 may include a metrology measurement device that is separate (i.e., external) from the fabrication tool 122. For example, the metrology tool 132 may be a stand-alone instrument that is not connected to any station of the fabrication tool 122. To obtain measurements of a substrate using the metrology tool 132, a user (e.g., an engineer, an operator) of the fabrication system may remove a substrate processed in the fabrication tool 122 from the fabrication tool 122 and transport the substrate to the metrology tool 132 for measurement. In some embodiments, the metrology tool 132 may transport metrology data generated for the substrate to a client device 120 coupled to the metrology tool 132 via a network 140 (e.g., for presentation to a fabrication user, such as an operator or engineer). In other or similar embodiments, a user of the fabrication system may obtain metrology data of the substrate from the metrology tool 132 and provide the metrology data to the system architecture via a graphical user interface (GUI) of the client device 120.
[0032] The data store 150 may be a memory (e.g., random access memory), a drive (e.g., hard drive, flash drive), a database system, or other type of component or device capable of storing data. The data store 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 store 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 previous substrates processed on the manufacturing tool 122 or other manufacturing tools coupled to the data store 150 via the network 140) and / or current spectra (spectral data generated for the current substrate being processed on the manufacturing tool 122). The current spectral data may be the data from which the predictive data is generated. In some embodiments, the metrology data may include historical metrology data (e.g., metrology measurements of previous substrates processed on the manufacturing tool 122 or other manufacturing tools). The data store 150 may also store context data associated with substrates being processed in the manufacturing system (eg, recipe name, recipe step number, preventive maintenance indicators, operator, etc.).
[0033] One or more portions of data store 150 can be configured to store data that is not accessible to users of the manufacturing system. In some embodiments, all data stored in data store 150 can be inaccessible to users of the manufacturing system. In other or similar embodiments, some of the data stored in data store 150 is inaccessible to users, while other portions of the data stored in data store 150 are accessible to users. In some embodiments, the inaccessible data stored in data store 150 is encrypted using an encryption mechanism unknown to the users (e.g., the data is encrypted using a secret encryption key). In other or similar embodiments, data store 150 can include multiple data stores, where data that is not accessible to users is stored in a first data store and data that is accessible to users is stored in a second data store.
[0034] In some embodiments, prediction system 110 includes a server machine 170 and a server machine 180. Server machine 170 includes a training set generator 172 that can generate a training data set (e.g., a set of data inputs and a set of target outputs) for training, validating, and / or testing a machine learning model 190 or a set of machine learning models 190. Some operations of training set generator 172 are described in more detail below with reference to FIG. 4. In some embodiments, training set generator 172 can split the training data into a training set, a validation set, and a test set.
[0035] The server machine 180 may include a training engine 182. An engine may refer to hardware (e.g., circuitry, dedicated logic, programmable logic, microcode, processing device, etc.), software (e.g., instructions executed on a processing device, general-purpose computer system, dedicated machine, etc.), firmware, microcode, or a combination thereof. The training engine 182 may train a machine learning model 190 or a set of machine learning models 190. A machine learning model 190 refers to a model product created by the training engine 182 using training data including training inputs and corresponding target outputs (correct answers for each training input). The training engine 182 finds patterns in the training data that map training inputs to target outputs (predicted answers) and provides a machine learning model 190 that captures these patterns. The machine learning model 190 may include a linear regression model, a partial least squares regression model, a Gaussian regression model, a random forest model, a support vector machine model, a neural network, a ridge regression model, etc.
[0036] The training engine 182 may also validate the trained machine learning models 190 using the corresponding feature sets 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 each trained model, the speed of each model, and / or the efficiency of each model. The training engine 182 may select trained machine learning models 190 having a performance rating that meets the performance criteria used by the prediction engine 114 according to embodiments described herein. Further details regarding the training engine 182 are provided in connection with FIG. 9.
[0037] The prediction server 112 includes a prediction engine 114 that can provide spectral data for a portion of a current substrate being processed on the manufacturing tool 122 as input to a trained machine learning model 190 and execute the trained model 190 on the input to obtain one or more outputs. In some embodiments, the trained model 190 executed by the prediction engine 114 is selected by the training engine 182 as having a performance evaluation that meets a performance criteria, as described above. As will be further described in connection with FIG. 8, 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 determine the status (e.g., chamber condition metrics) of the processing chamber of the manufacturing tool 122.
[0038] The confidence data may include or indicate a level of confidence that the chamber condition metric corresponds to one or more characteristics of the processing chamber associated with the current spectral data. In one example, the confidence level is a real number between 0 and 1, where 0 indicates no confidence that the chamber condition metric corresponds to one or more characteristics of the processing chamber associated with the current spectral data, and 1 indicates absolute confidence that the chamber condition metric corresponds to one or more characteristics of the processing chamber associated with the current spectral data. In some embodiments, the chamber condition prediction system may provide chamber condition metrics and determine measured metrology values for processing chambers processing substrates in the manufacturing system 122 using the prediction system 110 instead of using the in-line metrology 130, the integrated metrology 130, and / or the external metrology 132. The chamber condition prediction system may determine the condition of a processing chamber associated with a recovery procedure (e.g., chamber seasoning) performed after a PM procedure (e.g., chamber cleaning) according to embodiments provided herein.
[0039] It should be noted that in some other embodiments, the functionality of server machines 170 and 180, as well as 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, functionality described in one embodiment as being performed by server machine 170, server machine 180, and / or prediction server 112 may also be performed on client device 120. Furthermore, functionality attributed to a particular component may also be performed by different or multiple components operating in conjunction. Further details regarding the grouping of functionality of server machines 170, 180 and prediction server 112 are provided in conjunction with FIG. 3.
[0040] In embodiments, a "user" may be represented as a single individual. However, other embodiments of the present disclosure encompass a "user" that is an entity controlled by multiple users and / or automated sources. For example, a set of individual users organized as a group of administrators may be considered a "user."
[0041] 2 is a top schematic view of an example of a manufacturing system 200 according to aspects of the invention. The manufacturing system 200 can perform one or more processes on a substrate 202. The substrate 202 can be any suitable rigid, fixed-dimension, planar article (e.g., a silicon-containing disk or wafer, a patterned wafer, a glass plate, etc.) suitable for manufacturing electronic devices or circuit components according to aspects of the invention. In some embodiments, the manufacturing system 200 can include or be part of the system architecture 100 according to the embodiments described in connection with FIG.
[0042] 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 with 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 210. The processing chambers 214, 216, 218 may be coupled to the transfer chamber 210 via respective ports (e.g., slit valves, etc.). The transfer chamber 210 may also include a transfer chamber robot 212 configured to transfer the substrate 202 between the processing chambers 214, 216, 218, the load lock 220, etc. The transfer chamber robot 212 may include one or more arms, each arm including one or more end effectors at the end of each arm. The end effectors may be configured to process a particular object, such as a wafer.
[0043] In some embodiments, the transfer chamber 210 may also include a metrology device, such as the integrated metrology device 128 described in connection with FIG. 1. The integrated metrology device 128 may be configured to generate metrology data related to the substrate 202 before or during substrate processing while the substrate is maintained in a vacuum environment. As shown in FIG. 2, the integrated metrology device 128 may be located within the transfer chamber 210. In other or similar embodiments, the integrated metrology device 128 may be coupled to the transfer chamber 210. When the integrated metrology device 128 is located within or coupled to the transfer chamber 210, metrology data related to the substrate 202 may be generated without removing the substrate 202 from the vacuum environment (e.g., transferred to the factory interface 206).
[0044] The processing chambers 214, 216, 218 may be adapted to perform any number of processes on the substrate 202. The same or different substrate processes may be performed in each processing chamber 214, 216, 218. Substrate processes may include atomic layer deposition (ALD), physical vapor deposition (PVD), chemical vapor deposition (CVD), etching, annealing, hardening, pre-cleaning, metal or metal oxide removal, and the like. Other processes may also be performed on the substrate. In some embodiments, a chamber state device, such as the chamber state device 124 described in connection with FIG. 1, may be coupled to or disposed within the processing chambers 214, 216, 218. The chamber state device 124 may be configured to collect spectral data. The spectral data may include OES data related to the plasma in the processing chambers 214, 216, 218 and reflectance data for a profile of a surface of the substrate during substrate processing. A processing device coupled to the chamber state device 124 (e.g., system controller 228) can determine chamber state metrics (e.g., indicative of whether the chamber is operating normally or abnormally) of the processing chamber 214, 216, 218 performing the etch process based on the collected spectral data. In some embodiments, one or more components of the chamber state device 124 can include components described in connection with FIG. 5 (e.g., fiber optic bundles, collimator assemblies, etc.). In other or similar embodiments, the chamber state device 124 can include one or more sensors disposed inside or outside the processing chamber 214, 216, 218 and configured to collect spectral data of a portion of the substrate 202 and / or an environment within the processing chamber 214, 216, 218 before, after, or during substrate processing.
[0045] The load lock 220 may also be coupled to the housing 208 and the transfer chamber 210. The load lock 220 may be configured to interface and couple to the transfer chamber 210 on one side and the factory interface 206 on the other side. The load lock 220 may have an environmentally controlled atmosphere that may be changed from a vacuum environment (where substrates may be transferred to and from the transfer chamber 210) to an atmospheric (or near atmospheric) inert gas environment (where substrates may be transferred to and from the factory interface 206), in some embodiments.
[0046] The factory interface 206 may be any suitable enclosure, such as a front-end equipment module (EFEM). The factory interface 206 may be configured to receive substrates 202 from substrate carriers 222 (e.g., front-opening unified pods (FOUPs)) docked to various load ports 224 of the factory interface 206. A factory interface robot 226 (shown in dotted lines) may be configured to transport substrates 202 between the substrate carriers (also referred to as containers) 222 and the load locks 220. In other and / or similar embodiments, the factory interface 206 may be configured to receive replacement parts from a replacement part storage container 222.
[0047] In some embodiments, the manufacturing system 200 may include a metrology tool configured to generate metrology data related to the substrate 202 outside of a vacuum environment. For example, as shown in FIG. 2, the integrated metrology tool 128 may be coupled to a process chamber (e.g., process chambers 214, 216, and / or 218). The integrated metrology tool 128 may be configured to generate metrology data related to the substrate 202 before the substrate 202 is placed in the vacuum environment (e.g., transferred to a load lock 220) and / or after the substrate 202 is removed from the vacuum environment (e.g., removed from the load lock 220). Although FIG. 2 shows the in-line metrology tool 130 coupled to the factory interface 206, the in-line metrology tool 130 may be coupled to any portion of the process tool 204 that is outside of a vacuum environment (e.g., coupled to a load lock 220, etc.).
[0048] The manufacturing system 200 may also be connected to a client device (e.g., client device 120 of FIG. 1 ) configured to provide information about the manufacturing system 200 to a user (e.g., an operator). In some embodiments, the client device may provide information to a user of the manufacturing system 200 via one or more graphical user interfaces (GUIs). For example, the client device may provide information about one or more chamber state metrics of the processing chambers 214, 216, 218 (e.g., during substrate processing) via the GUI.
[0049] The manufacturing system 200 may also include or be coupled to a system controller 228. The system controller 228 may be and / or may include a computing device such as a personal computer, a server computer, a programmable logic controller (PLC), a microcontroller, etc. The system controller 228 may include one or more processing devices, which may be general-purpose processing devices such as a microprocessor, a central processing unit, etc. More specifically, the processing devices may be complex instruction set computing (CISC) microprocessors, reduced instruction set computing (RISC) microprocessors, very long instruction word (VLIW) microprocessors, or processors that execute other instruction sets or combinations of instruction sets. The processing devices may also be one or more special-purpose processing devices such as application specific integrated circuits (ASICs), field programmable gate arrays (FPGAs), digital signal processors (DSPs), network processors, etc. The system controller 228 may include a data storage device (e.g., one or more disk drives and / or solid state drives), 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 methods and / or embodiments described herein. In some embodiments, the system controller 228 may execute instructions to perform one or more operations in the manufacturing system 300 according to a process recipe. The instructions may be stored in a computer-readable storage medium, which may include a main memory, a static memory, a secondary storage device, and / or a processing device during execution of the instructions.
[0050] In some embodiments, the system controller 228 may receive data from sensors or other components (e.g., the chamber state device 124) included on or within various portions of the manufacturing system 200 (e.g., the processing chambers 214, 216, 218, the transfer chamber 210, the load lock 220, etc.). The data received by the system controller 228 may include spectral and / or non-spectral data relating to portions of the substrate 202. For purposes of this description, the system controller 228 is described as receiving data from the chamber state device 124 coupled to or disposed within the processing chambers 214, 216, 218. However, the system controller 228 may receive data from any portion of the manufacturing system 200 and use the data received from that portion in accordance with embodiments described herein. In one illustrative example, the system controller 228 may receive spectral data from the chamber state device 124 coupled to the processing chambers 214, 216, 218 before, after, or during substrate processing in the processing chambers 214, 216, 218. Data received from the chamber state device 124 or other sensors of the manufacturing system 200 may be stored in a data store 250. The data store 250 may be included as a component within the system controller 228 or may be a separate component from the system controller 228. In some embodiments, the data store 250 may be included as a component within the system controller 228. 250 may be part of or include a portion of the data store 150, as described in connection with FIG.
[0051] 3 illustrates an exemplary system architecture 300 for chamber state prediction of a processing chamber (e.g., performing substrate processing) in accordance with aspects of the present invention. In some embodiments, the chamber state prediction system 300 may include or be part of one or more components of the system architecture 100 and / or the manufacturing system 200. The chamber state prediction system 300 may include one or more components of the manufacturing tool 122 (e.g., the chamber state tool 124), the metrology tool 130, the server machine 320, and the server machine 350.
[0052] As previously described, the manufacturing tool 122 may produce a product or perform a process over a period of time according to a recipe. The manufacturing tool 122 may include a process chamber 310 configured to perform substrate processing of a substrate according to a substrate processing recipe. In some embodiments, the process chamber 310 may be any of the process chambers 214, 218, 218 described in connection with FIG. 2. The manufacturing tool 122 may also include a chamber state device 124 as described herein. The chamber state device 124 may be coupled to or disposed within the process chamber 310 and may collect spectral data used to detect an endpoint of a step of a substrate processing recipe. In some embodiments, the manufacturing tool 122 may also include an integrated metrology device 128 as described herein. The integrated metrology device 128 may be configured to generate metrology data related to the substrate before or after substrate processing is completed.
[0053] The manufacturing equipment 122 can be connected to a server machine 320. The server machine 320 can include a processing device 322 and / or a data store 332. In some embodiments, the processing device 322 can be configured to execute one or more instructions to perform operations at the manufacturing equipment 122. For example, the processing device 322 can include or be part of the system controller 228 described in FIG. 2. In some embodiments, the data store 332 can include or be part of the data store 150 and / or the data store 250.
[0054] The processing device 322 can be configured to receive data from one or more components of the manufacturing tool 122 (i.e., over a network). For example, the processing device 322 can receive spectral data 336 (e.g., OES data of a plasma disposed in the chamber, optical reflectance measurements corresponding to a reflection pattern of light reflected from a surface of a substrate disposed in the processing chamber) collected by the chamber state tool 124 during substrate processing of a substrate in the processing chamber 310. In another example, the processing device 322 can receive metrology data 338 collected by the integrated metrology tool 128 before and / or after substrate processing of a substrate. The metrology data 338 can include metrology measurements generated for the substrate by the integrated metrology tool 128. In some embodiments, the processing device 322 can store the received spectral data and / or the received metrology data 338 in the data store 332.
[0055] In some embodiments, the processing device 322 can receive metrology data 338 from other metrology tools. For example, in some embodiments, the server 320 can be connected (i.e., over a network) to the in-line metrology tool 130. The substrate is removed from the processing chamber 210 and transferred to the in-line metrology tool 130 as described herein. The in-line metrology tool 130 can generate metrology data 338 for the substrate and transmit the generated metrology data 338 to the processing device 322 over a network. In another example, the substrate can be removed from the manufacturing tool 122 and transferred to an external metrology tool, such as the external metrology tool 132 described in connection with FIG. 1 . A client device, such as the client device 120, can be connected (i.e., over a network) to the server 320. In some embodiments, a user of the manufacturing tool 122 can obtain metrology data 338 using the external metrology tool 132 and provide the obtained metrology data 338 via a GUI of the client device 120. The client device 120 can transmit the metrology data 338 over a network. In additional or alternative embodiments, the external metrology device 132 can be connected to the server 320 via a network, with the external metrology device 132 transmitting the metrology data 338 directly to the processing device 352 .
[0056] The processing device 352 may include a prediction engine 328 and a chamber condition engine 330. The prediction engine 328 may be configured to provide metrology measurements based on spectral data 336 collected for a substrate during substrate processing. For example, the prediction engine 328 may provide the spectral data 336 collected for a current process being performed on a current substrate in the process chamber 310 as an input to the trained machine learning model 334. The prediction engine 328 may obtain, as an output of the machine learning model 334, predicted metrology data 338 that includes an indication of metrology measurements corresponding to the current substrate. In some embodiments, the prediction engine 328 may correspond to the prediction engine 114 described in connection with FIG. 1 .
[0057] The chamber metric engine 330 of the processing device 322 can be configured to determine a chamber state of the process chamber 310 performing substrate processing. The chamber state engine 330 can obtain or determine one or more chamber state metrics of the process chamber 310 processing the current substrate from the prediction engine 328. The chamber state metrics can include a selection of values associated with a combination, feature, or pattern identified as an input to the prediction engine 328. For example, a first value can represent a particular spectral data and sensor data combination at a given time. In other examples, other values can be associated with a gradient of the combination of one or more variables determined and / or identified by the prediction engine 328. In some embodiments, the chamber state metric can include a set of values (e.g., vectors, matrices, etc.) indicative of correlations of particular data combinations, correlations, patterns, and / or relationships present in the sensor data. For example, the chamber state metric can include a feature vector that includes binary values indicating the presence or absence of a particular feature in the data.
[0058] The chamber condition metrics may be compared to known patterns and / or combinations of chamber condition metrics (e.g., target chamber condition metrics). The target chamber condition metrics may be associated with one or more stages of a recovery process (e.g., a seasoning procedure). In response to determining that the chamber condition metrics meet one or more chamber condition thresholds (e.g., conditions associated with a recovery process for the processing chamber), the chamber state engine 330 may modify the performance of the processing chamber 310 (e.g., by generating instructions including commands to modify process parameters). The chamber condition thresholds may be a particular combination of values indicated by the chamber condition metrics. For example, the chamber state engine 330 may determine an update to a minimum process recipe associated with the received sensor data based on determining that one or more chamber conditions exist in the processing chamber based on the identified patterns, features, correlations, and relationships between the chamber condition metrics received by the prediction engine 328. For example, the modification may be associated with a current substrate processing, such as processing the current substrate according to the updated process parameters. In another example, the modification may include stopping the current substrate processing. In another example, the modification may include modifying a process recipe used to process future substrates. The processing device 322 may send instructions to the fabrication tool 122 to modify the identified performance of the processing chamber. For example, the instructions may instruct the processing chamber to continue processing substrates according to an updated set of processing parameters. In another example, the instructions may include stopping substrate processing in the processing chamber. In some embodiments, the target chamber metric is associated with a stage of a PM recovery procedure. More details related to predicting chamber conditions for a processing chamber that processes substrates according to a current process are provided in FIG. 7.
[0059] As shown in FIG. 3, in some embodiments, the processing device 322 can include a training set generator 324 and / or a training engine 326. In some embodiments, the training set generator 324 can correspond to the training set generator 172 described in connection with FIG. 1, and / or the training engine 326 can correspond to the training engine 182. The training set generator 324 can be configured to generate a training set 340 for training the machine learning model 334 or the 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 the historical spectral data 336 from the data store 332 to generate the training inputs. The training set generator 324 can generate target outputs indicative of chamber conditions (e.g., chamber condition metrics) 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 inline metrology tool 130, the integrated metrology tool 128, or the external metrology tool 132. Training set generator 324 can include the generated training inputs and the generated target outputs in training set 340. Details regarding the generation of training set 340 are provided in connection with FIG.
[0060] The training engine 326 can be configured to train, validate, and / or test the machine learning model 334 or the set of machine learning models 334. The training engine 326 can provide a training set 340 for training the machine learning model 334 and storing the trained machine learning model in the data store 332. In some embodiments, the training engine 326 can use a validation set 342 to validate the trained machine learning model 334. The validation set 342 can include spectral data 336 and chamber condition metrics corresponding to metrology data 338 acquired for a previous substrate (i.e., processed in the process chamber 310 or another process chamber). The training set generator 324 and / or the training engine 326 can generate the validation set 342 based on the historical spectral data 336 and historical metrology data 338 acquired for a previous substrate. In some embodiments, the validation set 342 can include historical spectral data 336 and historical metrology data 338 that are different from the historical spectral data 336 and historical metrology data 338 included in the training set 340.
[0061] The training engine 326 can provide the spectral data 336 (and in some embodiments the sensor data) of the previous substrate as input to the trained machine learning model 334 and can extract chamber condition metrics of the process chamber processing the previous substrate from one or more outputs of the trained model 334. The training engine 326 can assign a performance score to the trained model 334 based on the accuracy of the chamber condition metrics of the process chamber processing the previous substrate and taking into account the measured chamber conditions of the metrology data 338 of the previous substrate included in the validation set 342. In response to determining that the performance score meets the performance score criteria (e.g., exceeds a performance score threshold), the training engine 326 can select the trained model 334 to be used to provide future metrology measurements for future substrates processed in the process chamber 310. Further details regarding the selection of the trained model 334 are provided in connection with FIG. 9.
[0062] As mentioned above, in some embodiments, the training set generator 324 and / or the training engine 326 may be components of the processing device 322 of the server 320. In additional or alternative embodiments, the training set generator 324 and / or the training engine 326 may be components of the processing device 352 of the server 350. The server 350 may include or be part of a computing system separate from the manufacturing system 200. As mentioned above, in some embodiments, the server 320 may include or be part of the system controller 228 described in connection with FIG. 2. In such an embodiment, the server 350 may include or be part of a computing system that is 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 have access to data stored in one or more portions of the data store 332 or one or more processes executed on the processing device 322. However, a user of the manufacturing system 200 may not have access to data stored in one or more portions of the data store 354 or processes executed on the processing device 352.
[0063] The processing device 352 can be configured to execute the training set generator 324 and / or the training engine 326 in a manner similar to the processing device 322. In some embodiments, the server 350 can be coupled to the manufacturing tools 122 and / or the in-line metrology tools 130 via a network. Thus, the processing device 352 can obtain the spectral data 336 and chamber condition metrics corresponding to the metrology data 338 used by the training set generator 324 and / or the training engine 326, and can generate the training set 340 and validation set 342 according to the embodiments described with respect to the processing device 322. In other or similar embodiments, the server 350 is not coupled to the manufacturing tools 122 and / or the external metrology tools 132. Thus, the processing device 352 can obtain the spectral data 336 and / or the metrology data 338 from the processing device 322. For example, the processing device 322 can receive the spectral data 336 from the chamber condition tool 124, as described above. The processing device 322 can transmit the received spectral data 336 to the processing device 352 (i.e., over a network). In some embodiments, the processing device 352 can store the spectral data 336 in a 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 described above, the processing device 322 can receive metrology data 338 from the in-line metrology tool 130, the integrated metrology tool 128, and / or the external metrology tool 132. The processing device 352 can transmit the received metrology data 338 from the processing device 322, and in some embodiments, the metrology data 338 can be stored in the data store 354.
[0064] The training set generator 324 of the processing device 352 can generate the training set 340 according to the above-described embodiments. The training engine 326 of the processing device 352 can train and / or validate the machine learning model 334 according to the above-described embodiments. In some embodiments, the server 350 can be coupled to other manufacturing tools and / or other server machines different from the manufacturing tool 122 and / or the server machine 320. In accordance with the embodiments described herein, the processing device 352 can obtain the spectral data 336 and / or metrology data 338 from the other manufacturing tools and / or server machines. In some embodiments, the training set 340 and / or validation set 342 can be generated based on the spectral data 336 and metrology data 338 acquired for substrates processed in the process chamber 310 and other spectral data and metrology data acquired for other substrates processed in process chambers of other manufacturing systems.
[0065] Once the training engine 326 has selected the trained model 334 to apply to future spectral data for future substrates in the processing chamber 310, the processing device 352 may send the trained model 334 to the processing device 322. The prediction engine 328 may use the trained model 334 to provide metrology measurements for future substrates in the processing chamber 310, as described above.
[0066] 4 is a flow chart of a method 400 for training a machine learning model according to an aspect of the disclosure. The method 400 is performed by processing logic that may include hardware (circuitry, dedicated logic, etc.), software (such as running on a general-purpose computer system or a dedicated machine), firmware, or a combination thereof. In one embodiment, the method 400 may be performed by a computer system such as the system architecture 100 of FIG. 1. In other or similar embodiments, one or more operations of the method 400 may be performed by one or more other machines not shown. In some aspects, one or more operations of the method 400 may be performed by the training set generator 324 of the server machine 320 or the server machine 350 described in connection with FIG. 3.
[0067] At block 410, processing logic initializes a training set T to an empty set (e.g., {}). At block 412, processing logic acquires spectral and / or chamber data associated with a substrate being processed in a processing chamber of the manufacturing system. In some embodiments, the spectral and / or chamber data may be received from one or more sensors located within or coupled to the processing chamber. In other or similar embodiments, the spectral data may be received from a chamber state device located within or coupled to the processing chamber.
[0068] Returning to FIG. 4 , in block 414, processing logic obtains metrology data for the substrate. As previously described, the metrology data may include one or more values of film property data (e.g., wafer spatial film properties), dimensions (e.g., thickness, height, etc.), dielectric constant, dopant concentration, density, defects, etc. In some embodiments, the metrology data may further include one or more surface profile property data (e.g., etch rate, etch rate uniformity, critical dimension of one or more features included in the surface of the substrate, uniformity of critical dimension across the surface of the substrate, edge placement error, etc.). In some embodiments, the metrology measurements may be received from a metrology tool (e.g., in-line metrology tool 130, integrated metrology tool 128, etc.) of the fabrication system 200. For example, after substrate processing is completed, the substrate may be transferred to the in-line metrology tool 130 or integrated metrology tool 130 of the fabrication system 200. The in-line metrology device 130 or the integrated metrology device 128 may generate metrology data related to the substrate and transmit the metrology data over a network to a computing device containing processing logic (e.g., server machine 320, server machine 350, etc.) In other or similar embodiments, metrology measurements may be received from a metrology device separate from the manufacturing system 200 (e.g., external metrology device 132) as described herein.
[0069] At block 416, processing logic generates training inputs based on the spectral data acquired for the substrate at block 412. In some embodiments, the training inputs may include a set of normalized spectral data (e.g., including both OES data and substrate surface reflectance data as described herein) generated based on the spectral data acquired for the processing device and the substrate. The set of normalized spectral data may include one or more spectral features corresponding to a particular type of metrology measurement. The spectral features may be based on a combination of optimal optical emission measurements and optical reflectance spectral measurements. Further details regarding the generation of the training inputs are described in connection with FIG. 6. At block 418, processing logic may generate target outputs based on the metrology data acquired for the substrate at block 414. The target outputs may correspond to chamber condition metrics (data indicative of a pattern of metrology measurements) corresponding to metrology measurements associated with the substrate. For example, at block 414, processing logic may acquire metrology data indicative of a film thickness at one or more portions of a surface of the substrate after an etch process. The thickness may be indicative of a chamber condition (e.g., a chamber that has not fully recovered from a recovery process or a cold chamber). Processing logic may generate a target output corresponding to the identified chamber condition (e.g., abnormal chamber condition, normal chamber condition, not fully restored, not fully seasoned, etc.).
[0070] At block 420, processing logic generates an input / output mapping. The input / output mapping references training inputs that include or are based on data for the substrate and target outputs for the training inputs. The target outputs specify metrology measurements for the substrate and the training inputs are related (or mapped) to the target outputs. At block 422, processing logic adds the input / output mapping to a training set T.
[0071] At block 424, the processing logic determines whether the training set T includes a sufficient amount of training data to train the machine learning model. It should be noted that in some embodiments, the sufficiency of the training set T can be determined solely 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 the training examples, etc.) in addition to or instead of the number of input / output mappings. Upon determining that the training set T includes a sufficient amount of training data to train the machine learning model, the processing logic provides the training set T for training the machine learning model. In response to a determination that the training set does not include a sufficient amount of training data to train the machine learning model, the method 400 returns to block 412.
[0072] At block 426, the processing logic provides a training set T for training the machine learning model. In some embodiments, the training set T is provided to the training engine 326 of the server machine 320 and / or the server machine 350 to perform the training. For example, in the case of a neural network, input values of a given input / output mapping (e.g., spectral data and / or chamber data of a previous substrate) are input to the neural network, and output values of the input / output mapping are stored at output nodes of the neural network. The connection weights in the neural network are then adjusted according to a learning algorithm (e.g., backpropagation, etc.), and the procedure is repeated for other input / output mappings in the training set T. After block 426, the machine learning model 190 can be used to provide chamber conditions (e.g., chamber condition metrics) for future processing chambers that will process future substrates (e.g., according to method 800 of FIG. 8, described below).
[0073] 5 is a cross-sectional schematic side view of the chamber state device 124 coupled to a processing chamber 310, according to an aspect of the invention. In some embodiments, the processing chamber 310 can be used for processes in which a corrosive plasma environment is provided. For example, the processing chamber 310 can be a plasma etcher or plasma etch reactor, a plasma cleaner, or the like. In other or similar embodiments, the processing chamber 310 can be used for processes in which a non-corrosive environment is provided. For example, the processing 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 processing chamber 310 includes a chamber body 502, a lid 530, and / or a showerhead (not shown) that surrounds the interior volume 506. Generally, the chamber body 502 includes a sidewall 508 and a bottom 510. The showerhead can include a showerhead base and a showerhead gas distribution plate. The lid 530 and / or the showerhead can be supported on the sidewall 508 of the chamber body 502. The lid 530 (or showerhead) can be opened to allow access to the interior volume 506 of the processing chamber 310 and can be closed to seal the processing chamber 310. A gas panel (not shown) can be coupled to the processing chamber 310 to supply process gases and / or cleaning gases to the interior volume 506 via the lid 530 and nozzles (e.g., via openings in the showerhead or lid and nozzles) and / or the showerhead. An exhaust port 526 can be defined in the chamber body 502, and the interior volume 506 can be coupled to a pumping system 528. The pumping system 528 may include one or more pumps and a throttle valve used to evacuate and regulate the pressure in the interior volume 506 of the processing chamber 310. A substrate support assembly 548 is disposed within the interior volume 506, the lid 530, and / or the showerhead. The substrate support assembly 548 holds a substrate, such as the substrate 202 of FIG. 2, during processing. In one embodiment, the substrate support assembly 548 includes a pedestal 552 that supports an electrostatic chuck 550.
[0075] The chamber state device 124 can be configured to optically monitor the environment of the internal volume 506 during substrate processing of the substrate 202. In some embodiments, the chamber state device 124 can be mechanically coupled to the chamber body 502 and optically interfaced (i.e., via an optical interface 570) with the environment of the internal volume 506. The chamber state device 124 can include a collimator assembly 554, a fiber optic bundle 556, an optical component 562, a processing device 576, and, in some embodiments, a polarizer component 586. As shown in FIG. 5, the collimator assembly 554 can be coupled to a chamber interface 570. In some embodiments, the chamber interface 570 can be an orifice, a converging or diverging lens, a transparent slab, or any other device or material capable of conveying light between the collimator assembly 554 and the environment of the internal volume 506. Although FIG. 5 shows the chamber interface 570 as being embedded within the lid 530, the chamber interface 570 may be embedded within or coupled to any portion of the processing chamber 310 (e.g., the sidewalls 508, the bottom 510, etc.).
[0076] A first end of the fiber optic bundle 556 can be coupled to the collimator assembly 554, and a second end of the fiber optic bundle 556 can be coupled to the optical component 562. The optical component 562 can include a light source 564 configured to generate light. Here, "light" refers to electromagnetic radiation in any spectral range, including visible light, far and near infrared (IR), far and near ultraviolet (UV), etc. Additionally, "light" can include unpolarized (e.g., natural light), linearly polarized, circularly or elliptically polarized light, partially polarized light, focused light, diverging light, collimated light, etc. In some embodiments, the light source 564 can 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 can include a broadband light source. The light source 564 can include multiple component light sources, such as multiple narrowband light sources (used together to generate a broadband light input), in some embodiments. The light source 564 can include additional optical elements (i.e., filters, absorbers, polarizers, etc.) to control the spectral distribution and / or polarization of the light.
[0077] 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 emitting optical fibers 558 of the light beam 556. Upon receiving the input light via the emitting optical 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 pass through one or more optical elements of the collimator assembly 554, such as lenses, reflectors, filters, apertures, etc. In some embodiments, the spatial characteristics 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 within a wide range of wavelengths λ of the input light and, therefore, the various spectral components included in the light beam 572. In some embodiments, the collimator assembly 554 can include one or more achromatic lenses. Thus, the light beam 572 generated by the collimator assembly 554 can be an achromatic light beam.
[0078] As shown in FIG. 5, in some embodiments, the collimator assembly 554 can include a polarizer component 568. The polarizer component 568 is configured to polarize the unpolarized light (e.g., natural light) generated by the light source 564. For example, the polarizer component 568 can convert the unpolarized input light to linearly polarized, circularly polarized, or elliptically polarized light. It should be noted that while FIG. 5 illustrates the polarizer component 568 as part of the collimator assembly 554, the polarizer component 568 can be coupled to any portion of the chamber condition device 124 that passes the input light to the optical interface 570. For example, the polarizer component 568 can be connected to the outlet of the light source 564, the outlet of one or more emission optical fibers 558, or between the collimator assembly and the optical interface 570.
[0079] The collimator assembly 554 can direct the light beam 572 to 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 off the surface of the substrate 202 as a reflected light beam 574 and received by the collimator assembly 554. One or more receiving optical fibers 560 of the light beam 556 can transmit the reflected light beam 574 to a photodetector 566 of the optical component 562. The photodetector 566 can include one or more of a spectrometer, a spectrometer, a diffraction grating, a mirror, a lens, a photodiode, and other devices. The photodetector 566 can determine, alone or in conjunction with a processing device 576, one or more optical responses associated with the surface of the substrate 202 based on the reflected light beam 574. For example, the photodetector 566 and / or the processing device 576 can determine a reflectance R(l), a refractive index n(l), or other optical quantities 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 substrate 202 in terms of polarization dependence of reflectance, rotation angle of the plane of polarization upon reflection, emission intensity, etc. Spectral data as described in connection with this application can refer to data corresponding to the optical response of the reflected light 574 and / or optical properties of the substrate 202 derived from the optical response of the reflected light 574.
[0080] The chamber condition device 124 can include an optical sensor 582 that captures plasma emissions of the plasma disposed within the interior volume 506. The optical sensor 582 is coupled to an optical fiber 584 that transmits optical signals corresponding to the plasma emissions to a photodetector 566. The photodetector 566 can include an optical emission spectroscopy device (OES). The OES analyzes the optical signal received from the optical fiber 584 to identify emission peaks and patterns within the signal and identifies particular emission peaks as corresponding to energy transitions of particular elements. In some embodiments, information characterizing the spectrum and / or emission peaks can be displayed and / or manipulated by the OES. In some of these and other embodiments, the spectral data can include emission peak information. The spectral data can be conveyed to a processing device 576 for further processing.
[0081] The chamber status device 124 may include one or more sensors 512 configured to generate and / or collect sensor data related to the process chamber 310. The sensor data may include one or more values of temperature (e.g., heater temperature), spacing (SP), pressure, high frequency radio frequency (HFRF), electrostatic chuck (ESC) voltage, current, flow rate, power, voltage, etc. The sensor data may be related to or indicative of manufacturing parameters, such as hardware parameters, such as settings or components (size, type, etc.) of the manufacturing equipment, or processing parameters of the manufacturing equipment. The sensor data may be provided while the processing equipment 310 is performing substrate processing. The sensor data may be different for each substrate. The sensor data may be provided to a processing device 576.
[0082] In some embodiments, the processing device 576 can be included as part of a system controller (e.g., system controller 228) for a fabrication system that includes the processing chamber 310. In such embodiments, the processing device 576 can store the spectral data generated for the substrate 202 in a data store (e.g., data store 250, 332, 354, etc.) coupled to the processing device 576. In other or similar embodiments, the processing device 576 can be a processing component that is separate from the system controller, but coupled to the system controller via a network. The processing device 576 can transmit the generated spectral data to the system controller for storage in a data store for each of the fabrication systems.
[0083] 6 is a flow chart of a method 600 for training and / or updating a machine learning model for predicting chamber conditions of a processing chamber according to aspects of the disclosure. The method 600 is performed by processing logic, which may include hardware (circuitry, dedicated logic, etc.), software (such as running on a general-purpose computer system or a dedicated machine), firmware, or any combination thereof. In one embodiment, the method 600 may be performed by a computer system, such as the system architecture 100 of FIG. 1. In other or similar embodiments, one or more operations of the method 600 may be performed by one or more other machines not shown. In some aspects, one or more operations of the method 600 may be performed by the chamber condition engine 330 of the server machine 320 described in connection with FIG. 3.
[0084] At block 602, the processing logic performs data collection. The data collection may include receiving sensor data including chamber data and spectral data (e.g., OES and reflectance measurement data), measurements, and data labels (process yield, electrical measurements, etc.) from the manufacturing system. The data may be associated with one or more process chambers processing one or more substrates. The sensor data may include data acquired by the chamber state device 124, the integrated metrology device 128, and / or the in-line metrology device 130. In some embodiments, the spectral data may include optical emission spectroscopy (OES) measurements of a plasma disposed in the process chamber. In some embodiments, the spectral data may include optical reflectance spectral measurements corresponding to a reflectance pattern of light reflected from a surface of a substrate disposed in the process chamber. In some embodiments, the spectral and chamber data may be received from a chamber state device, such as the chamber state device 124, or other sensors disposed in or coupled to the process chamber, according to the embodiments described above.
[0085] At block 604, the processing logic performs data pre-processing, including determining one or more algorithms of machine learning to be trained using the sensor data, size, and rate threshold requirements. The processing logic generates normalized sensor data including indices of spectral features. The processing logic may perform functions and / or methods to extract features and / or generate synthetic / processed data related to the received sensor data. For example, the processing logic may perform feature extraction of the spectral data to identify spectral features. The spectral features may correspond to identified correlations, patterns, and / or anomalies in the spectral data. In other examples, the processing logic may perform feature extraction using a combination of the spectral data to determine whether a criterion is met. The processing logic may analyze multiple data points of related parameters to determine whether a rapid change in the parameter occurred during a portion of the substrate processing. In some embodiments, the processing logic performs normalization across different sensor data (e.g., spectral data and / or chamber data) associated with different processing chamber conditions. Normalization may include processing the input sensor data (e.g., spectral data and / or chamber data) so that it is similarly identifiable across different chambers and sensors used to acquire the data.
[0086] In some embodiments, the processing logic performs one or more of a data extrapolation procedure and / or a data interpolation procedure to obtain additional data other than the received measured data points. The processing logic may predict the spectral data at a time different from the measured values. For example, the processing logic may identify one or more features, patterns, and / or relationships between individual data points of the spectral data and use an interpolation procedure to determine estimates of spectral data points occurring at points between data points at which the measured spectral data (e.g., OES measurements) were captured (e.g., different time intervals or times at which no measurements were taken). In other examples, the processing logic may identify one or more features, patterns, and / or relationships between individual data points of the spectral data and use an extrapolation procedure to determine spectral data points occurring outside of a time window corresponding to the measured data points of the spectral data (e.g., spectral data predictions associated with instances before and / or after acquisition of the spectral data).
[0087] The processing logic may divide the received sensor data into a training set and / or a validation set. Further, the processing logic may classify the data into groups and assign individual data combinations to one or both of the training set and / or validation set.
[0088] At block 606, the processing logic performs model training for a set of one or more machine learning algorithms. Each of the set of machine learning models can correspond to a different machine learning model type. For example, each of the set of machine learning models can 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, a logistic regression type algorithm, a multi-layered perceptual algorithm, a recurrent neural network (RNN), etc. Performing the training of the set of machine learning models can be performed according to the embodiments described in connection with method 400 of FIG.
[0089] In some embodiments, the trained machine learning model receives new inputs having new sensor data having new chamber data indicative of a new condition of a new environment in a new processing chamber processing a new substrate according to a new process and new spectral data indicative of an optical emission spectroscopy (OES) measurement of a new plasma disposed in the new processing chamber processing a new substrate according to a new process generating new outputs based on the new inputs, the new outputs indicative of chamber condition metrics corresponding to a recovery condition associated with a chamber recovery process performed after a preventive maintenance procedure.
[0090] At block 608, processing logic performs validation of one or more trained machine learning models. Results and statistics from both the training and validation data sets are compared, and an overall "best" model is selected. Model selection may be determined according to method 900 of FIG. 9. Validation may include determining the accuracy of each trained machine learning model based on a set of corresponding features of each training set. Processing logic may discard trained machine learning models that do not meet an accuracy threshold. Processing logic may determine a trained machine learning model that has the highest accuracy among all of the trained machine learning models based on the test (and optionally validation) set.
[0091] At block 610, one or more models are installed on a server (e.g., an on-tool server). In one embodiment, the one or more selected models receive sensor data (e.g., spectral data and chamber data) of a new substrate and predict one or more chamber condition metrics corresponding to a prediction of a process result. The process result prediction may include one or more values indicative of an estimate of a process result (e.g., film thickness, critical dimension, sidewall angle, etc.) of the substrate corresponding to the received sensor data. The chamber condition metrics may include a selection of values each associated with a combination, feature, or pattern identified in the input data to the prediction engine 328. For example, a first value may be indicative of a certain spectral data and sensor data combination at a given time. In other examples, other values may be associated with a gradient of a combination of one or more variables determined and / or identified by the prediction engine 328. In some embodiments, the chamber condition metric may include a set of values (e.g., a vector, a matrix, etc.) indicative of a correlation of a particular data combination, correlation, pattern, and / or relationship present in the sensor data. For example, the chamber condition metric may include a feature vector including binary values indicative of the presence or absence of a particular feature in the data. The chamber condition metrics can be used to modify the performance of the processing chamber. For example, the chamber conditions can be utilized to update the processing parameters of a process recipe. The update can be based on a difference between the determined chamber condition metrics and a target chamber condition metric associated with the process recipe. In another example, the processing logic can display a notification on a graphical user interface (GUI). The notification can indicate a change to be made by the processing chamber (or more generally, the manufacturing tool). For example, the processing logic can cause a currently processed substrate or a new substrate to be processed according to the updated set of processing parameters. In another example, the processing logic can cause substrate processing in the processing chamber to be stopped.
[0092] At block 612, the processing logic performs in situ data collection of the live substrate processing environment. The collected data may include spectral data, metrology data, and data labels (e.g., process results such as yield, metrology data, electrical data, etc.). The data may be associated with one or more process chambers currently processing a substrate. The sensor data may include data acquired by the chamber state device 124, the integrated metrology device 128, and / or the in-line metrology device 130. In some embodiments, the spectral data may include optical emission spectroscopy (OES) measurements of a plasma disposed in the process chamber. In some embodiments, the spectral data may include optical reflectance spectrum measurements corresponding to a reflectance pattern of light reflected from a surface of a substrate disposed in the process chamber. In some embodiments, the spectral and chamber data may be received from a chamber state device such as the chamber state device 124 or other sensors disposed in or coupled to the process chamber according to the embodiments described above.
[0093] At block 614, processing logic determines whether there is an anomaly in the received in situ data collected in association with the substrate processing based on a comparison of the received data to the historical sensor data. Processing logic determines whether the metrology and process result labels are appropriate (e.g., meet a threshold performance evaluation) based on a comparison of the metrology data to the historical metrology data and associated labels. If processing logic determines there is no process anomaly, it proceeds along the "no" path to block 610 to continue processing additional sensor data and process result data associated with the current process chamber and substrate, or one or more additional process chambers and / or one or more additional substrates. If processing logic determines there is an anomaly in the received data set, it proceeds along the "yes" branch to block 616.
[0094] In some embodiments, the processing logic inputs the received data (e.g., sensor data, spectral data, chamber data, process result data) into a model (e.g., a statistical model). The processing logic receives one or more outputs from the statistical model. The one or more outputs may indicate a confidence level that a temporally associated data point of the chamber data, spectral data, and / or process result data accurately indicates a process chamber state (e.g., cold chamber, recovered chamber, failed chamber). The statistical model may be generated using regression between historical chamber data, historical spectral data, and / or historical process result data. The processing logic may determine that the confidence level meets a threshold and determine whether an anomaly exists in the received data.
[0095] In some embodiments, the statistical model is generated using statistical process control (SPC) analysis to determine control limits for the received data and to identify data as more or less reliable based on these control limits. In some embodiments, the statistical model is associated with univariate and / or multivariate data analysis (e.g., of historical data). For example, the statistical model can be used to analyze various parameters and determine patterns and correlations through statistical processes (range, minimum, maximum, quartiles, variance, standard deviation, etc.). In other examples, relationships between multiple variables (e.g., chamber data and spectral data) can be identified using regression analysis, path analysis, factor analysis, multivariate statistical process control (MCSPC) and / or multivariate analysis of variance (MANOVA).
[0096] At block 616, a chamber inspection is performed. The chamber inspection may include inspection of individual components of the manufacturing system. Processing logic may display a notification in the GUI indicating errors, data anomalies, and / or corrective actions to be taken. In some embodiments, processing logic initiates the creation, training, validation, and / or selection of one or more new models to perform the actions described herein. In some embodiments, processing logic may update the currently used model to correct the detected anomalies. For example, a user may load an old model specifying the type of mode, settings / parameters, and old data used to create the model. Processing logic may use the old model as a starting place (e.g., initial guess) to train an updated model. Additional data may be added to the system and the model may be further trained and / or validated to generate an updated model.
[0097] 7 is a flow chart of a method 700 for training and / or updating a machine learning model to predict chamber states of a processing chamber according to an aspect of the disclosure. The method 700 is performed by processing logic, which may include hardware (circuitry, dedicated logic, etc.), software (such as running on a general-purpose computer system or a dedicated machine), firmware, or a combination thereof. In one embodiment, the method 700 may be performed by a computer system, such as the system architecture 100 of FIG. 1. In other or similar embodiments, one or more operations of the method 700 may be performed by one or more other machines not shown. In some aspects, one or more operations of the method 700 may be performed by the chamber state engine 330 of the server machine 320 described in connection with FIG. 3.
[0098] In block 710, processing logic receives sensor data (e.g., spectral data, chamber data, etc.) associated with a current substrate being processed according to a current process in a processing chamber of the manufacturing system 200 (e.g., real-time in-situ data). In some embodiments, the processing logic may receive the current spectral data from the chamber state machine 124 or other sensors disposed in or coupled to the processing chamber performing the current process, as described above. In some embodiments, the processing logic may receive the current sensor data at a particular time period of the current process. For example, the chamber state machine 124 may be configured to collect sensor data for the current substrate at particular intervals (e.g., once per second) during substrate processing. The chamber state machine 124 collects the sensor data at each interval of substrate processing and transmits the spectral data to the server machine 320, where the spectral data is received by processing logic of the chamber state engine 330.
[0099] At block 712, the processing logic generates normalized sensor data including indices of the spectral features (e.g., data pre-processing). The processing logic may execute functions and / or methods to extract features and / or generate composite / processed data related to the received sensor data. For example, the processing logic may perform feature extraction on the spectral data to identify the spectral features. In another example, the processing logic may perform feature extraction using a combination of the spectral data to determine if a criterion is met. In some embodiments, the processing logic may perform principal component analysis (PCA) to select the most significant features of the spectral data. Principal component analysis refers to analyzing a set of points in real coordinate space and performing a basis change on the set of points. In some embodiments, the set of spectral features includes wavelength ranges of the detected light that correspond to individual chamber condition metrics. In such an embodiment, the processing logic may identify specific wavelengths based on the analysis of the spectral data. For example, the processing logic may provide data related to the structure of the substrate (e.g., CD, thickness, material properties, SWA, etc.) as input to a waveform analysis model and extract one or more outputs. The processing logic may determine, based on the one or more outputs, that a particular range of the optical spectrum corresponds to a particular type of metrology measurement, and that wavelengths X and Y fall within 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 the particular wavelengths that correspond to the particular type of metrology measurement, the processing logic may extract a set of spectral data corresponding to wavelengths X and Y from the normalized spectral data.
[0100] In some embodiments, the set of spectral features includes spectral trends or patterns present in the spectral data at one or more wavelengths corresponding to a particular type of metrology measurement corresponding to a particular chamber condition. The processing logic can perform one or more analytical operations (e.g., rigorous coupled wave analysis (RCWA)) on the normalized spectral data at one or more wavelengths to identify spectral trends or patterns corresponding to a particular type of metrology measurement corresponding to the chamber condition. In some embodiments, the processing logic can 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 processing logic can 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. The normalization can include processing the input sensor data (e.g., the spectral data and / or the chamber data) so that it is similarly identifiable across various chambers and sensors used to acquire the data.
[0101] At block 714, processing logic determines whether there is an anomaly in the received in situ data collected in association with the substrate processing. The anomaly may include identifying one or more data points having a combination of values (e.g., spectral values, chamber parameter values, etc.) that do not correlate with a historical combination of associated values. The processing logic may determine that the received data meets a threshold condition (e.g., a threshold specification). The threshold condition may include data that matches (e.g., correlates) with the historical data (e.g., has a correlation coefficient exceeding the threshold condition). The processing logic may determine that no processing anomaly exists and proceed along the "no" path to block 720 to continue further processing the received data. The processing logic may determine the presence of an anomaly in the received data set and proceed along the "yes" branch to block 716.
[0102] In some embodiments, the processing logic inputs the received data (e.g., sensor data, spectral data, chamber data) into a model (e.g., a statistical model). The processing logic receives one or more outputs from the model. The one or more outputs may indicate a confidence level that the temporally associated data points of the chamber data, spectral data, and / or process result data accurately indicate the condition of the process chamber. The statistical model may be generated using regression between historical chamber data, historical spectral data, and / or historical process result data. The processing logic determines whether the confidence level meets a threshold determination to determine the presence or absence of an anomaly in the received data.
[0103] In some embodiments, the statistical model is generated using statistical process control (SPC) analysis to determine control limits for the received data and determine whether the data is reliable or not based on these control limits. In some embodiments, the statistical model is associated with univariate and / or multivariate data analysis (e.g., of historical data). For example, the statistical model can be used to analyze various parameters and determine patterns and correlations through statistical processes (e.g., range, minimum, maximum, quartiles, variance, standard deviation, etc.). In other examples, regression analysis, path analysis, factor analysis, multivariate statistical process control (MCSPC), and / or multivariate analysis of variance (MANOVA) can be used to ascertain relationships between multiple variables (e.g., chamber data and spectral data).
[0104] In some embodiments, the processing logic determines the data based on whether the substrate criteria are met. The substrate criteria may include one or more of a film thickness requirement, a process uniformity requirement, a critical dimension specification, an SWA specification, etc. In some embodiments, the processing logic may determine that the substrate criteria are met by determining that the received spectral data corresponds to spectral data associated with a particular type of substrate processing and / or a particular type of substrate. For example, the processing logic may retrieve (e.g., from the data store 150) previously collected spectral data 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) specific to a type of substrate and / or a type of process, or a particular step or time period of a type of process. The processing 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) each of the spectral data features included in the previously collected spectral data.
[0105] At block 716, corrective actions are taken. In some embodiments, a chamber inspection is performed to identify one or more defects in the manufacturing equipment and determine corrective actions to repair the defects. The processing logic may display a notification on the GUI indicating errors, data anomalies, and / or corrective actions to be taken, etc. In some embodiments, the processing logic initiates the creation, training, validation, and / or selection of one or more new models to perform the actions described herein. In some embodiments, the processing logic may update the currently used model to correct the detected anomalies. For example, a user may load an old model that specifies the type of mode, settings / parameters, and old data used to create the model. The processing logic may use the old model as a starting place (e.g., initial guess) to train an updated model. Additional data may be added to the system to further train and / or validate the model to generate an updated model. The processing logic continues to update with new data and further evaluates the system after performing corrective actions with the new data.
[0106] At block 720, the processing logic provides the sensor data to be used as input to the trained machine learning model. In some embodiments, the processing logic can provide the spectral data to the prediction engine 328, which can enable the prediction engine 328 to perform one or more operations of the method 800 described in connection with FIG.
[0107] At block 722, processing logic obtains chamber state metrics extracted from one or more outputs of the trained machine learning model. In some embodiments, processing logic may receive an indication of the chamber state metrics from the prediction engine 328 obtained according to the method 800. At block 724, processing logic determines whether the chamber state metric criteria are met. In some embodiments, processing logic may determine whether the chamber state metric criteria are met by determining whether a substrate processed in the chamber under current chamber conditions meets a process result criteria associated with the current process. The extracted chamber state metric may correspond to a recovery state of the processing chamber associated with a recovery process (e.g., chamber seasoning) after a PM procedure.
[0108] In response to processing logic determining that the chamber condition metric criteria is satisfied, the method 700 proceeds to block 728. At block 728, processing logic continues the current process on the current substrate. In some embodiments, processing logic may continue the current process by sending an instruction to the system controller 228 or a local controller of the processing chamber to continue the current process. In other or similar embodiments, processing logic may continue the current process without generating and sending an instruction. Processing logic may send the sensor data (e.g., spectral data, chamber data, etc.) and extracted metrology measurements to the training set generator 324 for use as additional training data, as described above.
[0109] In response to the processing logic determining that the chamber condition metric criteria is not met, the method 700 proceeds to block 726. At block 726, the processing logic generates instructions to modify the performance of the manufacturing equipment. In some embodiments, the processing logic transmits instructions to terminate a current process in the manufacturing system. In some embodiments, the processing logic transmits instructions to a system controller (e.g., system controller 228), which may cause the system controller to terminate a current process. In other or similar embodiments, the processing logic transmits instructions to a local processing device of the processing chamber, which may cause the processing chamber to terminate a current process. In additional or alternative embodiments, the processing logic transmits the spectral data and the extracted metrology measurements to a training set generator 324 (i.e., server machine 320 or server machine 350) for use as additional training data. In some embodiments, the processing logic transmits instructions to update one or more process parameters associated with the manufacturing equipment. The instructions may include instructions to further process the current substrate according to the updated process parameters.
[0110] Returning to block 724, in some embodiments, the processing logic may further determine whether the chamber condition metric criteria is met by determining whether the chamber condition metric is within a range of predicted chamber condition metric values associated with the duration of the current process. The predicted chamber condition metric range may include a set of values predicted to be associated with the current substrate for the current duration of the current process. In some embodiments, the processing logic may determine 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) in light of the extracted chamber condition metric values. In-line or integrated metrology data may be used to validate the accuracy of the trained machine learning model. For example, the model may output that the chamber is operating under normal operating conditions, but the associated metrology data processed under the expected normal operating conditions may not meet the process result requirements (e.g., thickness requirements, critical dimension requirements, SWA requirements, etc.). Thus, the processing logic may send a notification to the client device indicating that the machine learning model needs to be retrained. In some embodiments, the processing logic may send a notification to the training set generator 324 and / or the training engine 326 to retrain the machine learning model according to embodiments described herein.
[0111] At block 730, processing logic proceeds to the next iteration of the measurement substrate and / or process recipe to perform method 700 using data related to further processing of the current substrate and / or processing of a next substrate and / or using a next process recipe.
[0112] 8 is a flow chart of a method 800 for predicting chamber conditions for a process chamber processing a current chamber using a machine learning model, according to an aspect of the disclosure. The method 800 is performed by processing logic, which may include hardware (circuitry, dedicated logic, etc.), software (such as running on a general-purpose computer system or a dedicated machine), firmware, or a combination thereof. In one embodiment, the method 800 may be performed by a computer system, such as the system architecture 100 of FIG. 1. In other or similar embodiments, one or more operations of the method 800 may be performed by one or more other machines not shown. In some aspects, one or more operations of the method 800 may be performed by the prediction engine 328 of the server machine 320 described in connection with FIG. 3.
[0113] At block 810, processing logic receives spectral data and chamber data associated with a process chamber for processing a substrate. In some embodiments, the spectral data may include optical emission spectroscopy (OES) measurements of a plasma disposed in the process chamber. In some embodiments, the spectral data may include optical reflectance spectral measurements corresponding to a reflectance pattern of light reflected from a surface of a substrate disposed in the process chamber. In some embodiments, the spectral and chamber data may be received from a chamber state device, such as chamber state device 124 or other sensors disposed in or coupled to the process chamber, in accordance with previously described embodiments.
[0114] At block 812, the processing logic generates a normalized sensor data including a representation of the spectral features. The processing logic may perform functions and / or methods to extract features and / or generate composite / processed data related to the received sensor data. For example, the processing logic may perform feature extraction on the spectral data to identify the spectral features. In other examples, the processing logic may perform feature extraction using a combination of the spectral data to determine if a criterion is met. The processing logic may analyze multiple data points of a relevant parameter to determine if a rapid change occurred during a portion of the substrate processing. In some embodiments, the processing logic performs normalization across different sensor data (e.g., spectral data and / or chamber data) associated with different processing chamber conditions. The normalization may include processing the input sensor data (e.g., spectral data and / or chamber data) so that it is similarly identifiable across different chambers and sensors used to acquire the data.
[0115] At block 814, processing logic provides the normalized sensor data as input to a trained machine learning model. In some embodiments, the trained machine learning model can correspond to the machine learning model 334 described in connection with FIG. 3. In some embodiments, the training engine 326 selects the machine learning model 334 for use by the prediction engine 328 according to an embodiment described in connection with FIG. 9 below. At block 816, processing logic obtains one or more outputs of the machine learning model. At block 818, processing logic extracts from the one or more outputs (i) one or more chamber conditions associated with the sensor data, and (2) a chamber condition metric that identifies an indication of a level of confidence that each of the one or more chamber conditions corresponds to a condition of the process chamber. In one example, the confidence level is a real number between 0 and 1. 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 chamber condition metrics may not equal 1.
[0116] In some embodiments, processing logic may use the chamber state metrics to determine a recovery state of the processing chamber corresponding to a recovery procedure performed after execution of a preventive maintenance (PM) procedure. In some embodiments, a processing state of the processing chamber is identified as being associated with the chamber state metric if a confidence level of the chamber state metric satisfies a threshold condition. In response to determining that the confidence level exceeds a confidence threshold level, processing logic may determine that a confidence level of the metrology measurement satisfies the threshold condition. Processing logic may provide the chamber state metrics to the chamber state engine 330 according to an embodiment described in connection with FIG. 10 .
[0117] 9 is a flow chart of a method 900 for selecting a machine learning model for estimating a type of metrology measurement, according to an aspect of the disclosure. The method 900 is performed by processing logic, which may include hardware (circuitry, dedicated logic, etc.), software (such as running on a general-purpose computer system or a dedicated machine), firmware, or any combination thereof. In one embodiment, the method 900 may be performed by a computer system, such as the system architecture 100 of FIG. 1. In other or similar embodiments, one or more operations of the method 900 may be performed by one or more other machines not shown. In some aspects, one or more operations of the method 900 may be performed by the training engine 326 of the server machine 320 or 350 described in connection with FIG. 3.
[0118] At block 910, processing 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, processing logic may receive the training data and validation data from a training set generator 324 of server machine 320 or 350. The training set generator 324 may generate the training data according to the embodiment described in connection with method 400 of FIG. 4. In other or similar embodiments, processing logic may receive the training data from the training set generator 324 and generate the validation data based on the received training data. As described in connection with FIG. 3, the training data may correspond to the training set 340 and the validation data may correspond to data included in the validation set 342.
[0119] At block 912, the processing logic trains each of the set of machine learning models using the received training data. At block 914, the processing logic performs one or more test operations on each of the set of machine learning models using the validation data. As described above, the validation set 342 can include sensor data (e.g., chamber data and spectral data) and chamber conditions corresponding to metrology data of a processing chamber that processed a previous substrate that is different from the sensor data (e.g., spectral data and chamber data) and chamber conditions corresponding to metrology data included in the training data. To perform the one or more test operations, the processing logic can provide the sensor data of the validation set 342 as input to each of the set of trained machine learning models and can obtain one or more outputs of each of the trained models. The processing logic can extract chamber condition metrics from the obtained one or more outputs according to embodiments provided herein.
[0120] In some embodiments, the trained machine learning model receives new inputs including new sensor data having new chamber data indicative of a new condition of a new environment in a new processing chamber processing a new substrate according to a new process, and new spectral data indicative of an optical emission spectroscopy (OES) measurement of a new plasma disposed in the new processing chamber processing the new substrate according to the new process based on the new inputs, the new outputs indicative of chamber condition metrics corresponding to a recovery status associated with a chamber recovery process performed following a preventive maintenance procedure.
[0121] At block 916, the processing 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 in relation to block 914. In some embodiments, the processing logic may assign a performance rating to each of the machine learning models based on the accuracy scores determined for the machine learning models. The processing logic may determine the accuracy score based on a difference between a chamber condition metric extracted from an output of each machine learning model and an actual chamber condition of the processing chamber operation. For example, as described in relation to block 914, the processing logic may provide sensor data from the validation set 342 as an input to each machine learning model and extract a chamber condition metric from the output of the model. The processing logic may compare the extracted chamber condition metric to the actual chamber condition (e.g., by comparing metrology results of the wafer processing under both conditions) and the provided sensor data of the validation set 342. The processing logic may assign an accuracy score to the model based on a difference between the extracted value and the actual value of the validation set 342. For example, the processing logic may assign a high accuracy score to each model if the difference between the extracted value and the actual value generated by the model is small. Similarly, processing logic may assign a lower accuracy score to each model if the difference is large.
[0122] In additional or alternative embodiments, the processing logic may further assign a performance rating to each machine learning model based on the speed score determined for the machine learning model. In some embodiments, the processing logic may determine the speed score based on an amount of time between the processing logic providing sensor data as input to the model and one or more outputs of the model being obtained. In other or similar embodiments, the processing logic may determine the speed score based on an amount of time between the processing logic providing spectral data as input to the model and the processing logic extracting metrology measurements from the one or more outputs that it has obtained. In one example, the processing logic may assign a speed score to each model if the amount of time between the processing logic providing sensor data as input to the model and the model providing one or more outputs (or the processing logic extracting metrology measurements) is small.
[0123] In some embodiments, the processing logic can assign a performance rating to each machine learning model based on the efficiency score determined for the machine learning model. As described in connection with FIG. 3, in some embodiments, the training engine 356 can be included in a server machine 350 separate from the server machine 320. In such an embodiment, the training engine 356 can train each of the set of machine learning models on the server machine 350. The training engine 356 can perform one or more test operations described in connection with block 914 to determine which of the set of machine learning models to send to the server machine 320 for use in the prediction engine 328. The processing logic can determine an efficiency score for each machine learning model based on an overall system efficiency (e.g., for a manufacturing system) for sending each machine learning model from the server machine 350 to the server machine 320 and / or initializing each machine learning model on the server machine 320. In some embodiments, the processing logic may determine the efficiency score based on the amount of memory used to store each trained machine learning model, the amount of memory available in the data store 332 of the server machine 320, the amount of network bandwidth available to transmit each trained machine learning model to the server machine 320, etc.
[0124] At block 918, processing logic determines whether the performance criteria are met based on the performance rating assigned to each machine learning model. In some embodiments, processing logic may determine whether the performance criteria are met by determining whether the performance rating assigned to each machine learning model (i.e., determined based on the accuracy score, and in some embodiments, determined based on the speed score and / or efficiency score) exceeds a threshold performance score. In other or similar embodiments, processing logic may determine whether the performance criteria are met by determining whether the accuracy score of each model exceeds a threshold score and whether the overall performance rating of each model (i.e., determined based on the accuracy score and the speed score and / or efficiency score) exceeds a threshold performance score. In some embodiments, multiple trained machine learning models may be associated with an accuracy score and / or performance rating that each meets the threshold score and / or threshold rating. In such embodiments, processing logic may determine each model of the multiple trained models that is associated with the highest accuracy score and / or performance rating that meets the performance criteria. In some embodiments, the processing logic may determine that no trained machine learning models of the set of machine learning models are associated with an accuracy score and / or performance rating that meets a threshold score and / or threshold rating, respectively. In such embodiments, the processing logic may determine that no models meet the performance criteria.
[0125] In response to determining that the performance criteria are met, processing logic proceeds to block 920. At block 920, processing logic selects respective machine learning models to be applied to future sensor data collected for future substrates processed according to the future substrate process. As described above, in some embodiments, the training engine 326 can be included in the prediction engine 328 of the server machine 320. In such embodiments, in response to selecting the respective machine learning models to be applied to the future spectral data, processing logic can store in the data store 332 the respective models and / or an indication that the respective models will be used by the prediction engine 328. In other embodiments, the training engine 326 can be included in the server machine 350. In such embodiments, in response to the training engine 326 selecting the respective machine learning models to be applied to the future sensor data, the server machine 350 can transmit the respective machine learning models to the server 320 for storage in the data store 150.
[0126] In response to determining that the performance criteria is not met, processing logic proceeds to block 922. At block 922, processing logic receives additional training data for further training the set of machine learning models. The processing logic may receive the additional training data from the training set generator 324 according to the embodiments described above. If the performance criteria is not met, after the data is received by block 922, the system may send the data to block 910 and begin flowing through another iteration of method 900.
[0127] FIG. 10 illustrates a diagrammatic representation of a machine in the exemplary form of a computing device 1000 capable of executing a set of instructions to cause the machine to perform one or more of the methodologies described herein. In alternative embodiments, the machine may be connected (e.g., networked) to other machines in a local area network (LAN), an intranet, an extranet, or the Internet. The machine may operate in a capacity as a server or client machine in a client-server network environment, or as a peer machine in a peer-to-peer (or distributed) network environment. The machine may be a personal computer (PC), a tablet computer, a set-top box (STB), a personal digital assistant (PDA), a mobile phone, a web appliance, a server, a network router, switch or bridge, or any machine capable of executing a set of instructions (sequential or otherwise) that specify actions to be performed by the machine. Furthermore, although only a single machine is illustrated, the term "machine" is intended to include a collection of machines (e.g., computers) that individually or jointly execute a set of instructions (or multiple sets) to perform any one or more of the methodologies described herein. In an embodiment, computing device 1000 may correspond to one or more of server machine 170, server machine 180, prediction server 112, system controller 228, server machine 320, or server machine 350, as described herein.
[0128] The exemplary computing device 1000 includes a processing device 1002, a main memory 1004 (e.g., read only memory (ROM), flash memory, dynamic random access memory (DRAM), such as synchronous DRAM (SDDRAM), etc.), a static memory 1006 (e.g., flash memory, static random access memory (RAM), etc.), and a secondary memory (e.g., a data storage device 1028), which communicate with each other via a bus 1008.
[0129] The processing device 1002 may represent one or more general-purpose processors, such as a microprocessor, a central processing unit, or the like. More specifically, the processing device 1002 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 that executes other instruction sets, or a processor that implements a combination of instruction sets. The processing device 1002 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, or the like. The processing device 1002 may also be or include a system on a chip (SoC), a programmable logic controller (PLC), or other type of processing device. The processing device 1002 is configured to execute processing logic for performing the operations and processes described herein.
[0130] Additionally, computing device 1000 may include a network interface device 1022 for communicating with a network 1064. Computing device 1000 may also include a video display unit 1010 (e.g., a liquid crystal display (LCD) or a cathode ray tube (CRT)), an alphanumeric input device 1012 (e.g., a keyboard), a cursor control device 1014 (e.g., a mouse), and a signal generation device 1020 (e.g., a speaker).
[0131] The data storage device 1028 may include a machine-readable storage medium (or, more specifically, a non-transitory computer-readable storage medium) 1024 on which is stored one or more sets of instructions 1026 that embody one or more of the methods or functions described. Here, non-transitory storage medium refers to a storage medium other than a carrier wave. The instructions 1026 may also reside, completely or at least partially, within the main memory 1004 and / or within the processing device 1002 during execution by the computing device 1000, with the main memory 1004 and the processing device 1002 also constituting computer-readable storage media.
[0132] Although the computer-readable storage medium 1024 is shown as a single medium in the exemplary embodiment, the term "computer-readable storage medium" is intended to include a single medium or multiple media (e.g., a centralized or distributed database and / or associated caches and servers) that store one or more sets of instructions. The term "computer-readable storage medium" is also intended to include any medium that can store or encode a set of instructions that are executed by a machine and cause the machine to perform one or more methods of the present disclosure. Thus, the term "computer-readable storage medium" is intended to include, but is not limited to, solid-state memory, optical media, and magnetic media.
[0133] In the preceding description, numerous specific details have been set forth, such as examples of specific system components, methods, and the like, to provide a thorough understanding of some embodiments of the present disclosure. However, it will be apparent to those skilled in the art that at least some embodiments of the present disclosure can be practiced without these specific details. In other instances, well-known components or methods have not been described in detail or have been presented in simple block diagram form to avoid unnecessarily obscuring the present disclosure. Thus, the specific details described are merely illustrative. A particular implementation may differ from these example details and still be considered to be within the scope of the present disclosure.
[0134] 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 appearance of the phrase "in one embodiment" or "in an embodiment" in various places throughout this specification do not necessarily all refer 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 is intended to mean that the nominal value presented is accurate to within ±10%.
[0135] Although the operations herein are illustrated and described in a particular order, the order of operations of each method may be changed, certain operations may be performed in reverse order, and certain operations may be performed at least in part concurrently with other operations. In other embodiments, instructions or sub-instructions of separate operations may be performed intermittently and / or alternatingly.
[0136] It is understood that the above description is intended to be illustrative and not limiting. Many other embodiments will become apparent to those skilled in the art upon reading and understanding the above description. The scope of the disclosure should therefore be determined with reference to the appended claims, along with the full scope of equivalents to which such claims are entitled.
Claims
1. 1. A method comprising: receiving, by the processing device, (i) chamber data indicative of an environmental condition of a processing chamber in which a substrate is processed according to a set of processing parameters for a current process; and (ii) sensor data comprising spectral data indicative of optical emission spectroscopy (OES) measurements of a plasma disposed within the processing chamber in which a substrate is processed according to the set of processing parameters for the current process; using, by a processing device, the sensor data as input to a machine learning model; obtaining, with a processing device, one or more outputs of the machine learning model, the one or more outputs indicative of one or more chamber condition metrics; determining, by the processing device, a recovery state of the processing chamber based on one or more chamber condition metrics, the recovery state corresponding to a chamber recovery procedure performed after a preventive maintenance procedure; A method comprising altering, by a processing device, performance of a processing chamber based on a recovery state of the processing chamber.
2. 2. The method of claim 1, further comprising: determining, by the processing device, an update to at least one process parameter of the set of process parameters based on one or more chamber condition metrics to generate an updated set of process parameters, wherein a change in performance of the processing chamber is based on the update to the at least one process parameter of the set of process parameters.
3. The step of modifying the performance of the processing chamber includes: Processing the substrate or a new substrate in accordance with the updated set of processing parameters; and / or stopping substrate processing in the processing chamber; 3. The method of claim 2, further comprising transmitting a first command that causes at least one of the following:
4. 10. The method of claim 1, further comprising displaying a notification on a graphical user interface (GUI) indicating a change in performance of the processing chamber.
5. The method of claim 1 , wherein the change in performance corresponds to a chamber seasoning procedure.
6. The method of claim 1 , wherein the spectral data corresponds to light reflectance spectral measurements corresponding to a reflectance pattern of light reflected from a surface of a substrate disposed within the processing chamber.
7. determining one or more spectral features based on a combination of the optical emission spectral measurements and the optical reflectance spectral measurements to generate feature data; The method of claim 6 , comprising using the feature data as input to a machine learning model.
8. using the sensor data as an input to a statistical model; receiving one or more outputs from the statistical model, the one or more outputs indicating a confidence level that the temporally associated data points of the chamber data and the spectral data accurately indicate the condition of the process chamber, the statistical model being generated using a regression between the historical chamber data and the historical spectral data; The method of claim 1 , further comprising determining whether a confidence level meets a threshold condition.
9. The method of claim 1 , wherein both receiving the sensor data and altering the performance of the processing chamber occur while the processing chamber is processing a substrate according to a set of processing parameters.
10. 1. A method for training a machine learning model to determine a state of a processing chamber in a chamber recovery procedure, the processing chamber processing a current substrate according to a current process, the method comprising: Generating training data for a machine learning model, identifying a first training input having historical sensor data including: i) historical chamber data indicative of an environmental condition of a second processing chamber processing a previous substrate according to a previous process; and ii) historical spectral data indicative of optical emission spectroscopy (OES) measurements of a previous plasma disposed in the second processing chamber processing a previous substrate according to a previous process; generating training data for the machine learning model, the training data including identifying a first target output for a first training input, the first target output including historical process result data having process result measurements of a previous substrate processed using the second process chamber according to a previous process; Providing training data for training the machine learning model based on (i) a set of training inputs comprising a first training input, and (ii) a set of target outputs comprising a first target output, A method comprising: providing training data, the trained machine learning model receiving new input having i) new chamber data indicative of a new condition of a new environment of a new processing chamber processing a new substrate according to a new process, and ii) new sensor data including new spectral data indicative of optical emission spectroscopy (OES) measurements of a new plasma disposed in the new processing chamber processing the new substrate according to the new process, and generating new output based on the new input, the new output indicative of a chamber condition metric corresponding to a recovery condition associated with a chamber recovery process performed after a preventive maintenance procedure.
11. Historical spectral data is The method of claim 10, further comprising optical reflectance spectrum measurements corresponding to a reflection pattern of light reflected from a surface of a previous substrate disposed in the second processing chamber.
12. The method of claim 11 , further comprising determining one or more spectral features based on a combination of optical emission spectral measurements and optical reflectance spectral measurements to generate feature data, the training data including the feature data.
13. The method of claim 10 , further comprising performing a data extrapolation procedure using the historical spectral data to generate optical emission spectrum (OES) estimates corresponding to time instances occurring before or after the OES measurement.
14. The method of claim 10 , further comprising performing a data interpolation procedure using the historical spectral data to generate optical emission spectroscopy (OES) estimates for one or more time instances occurring between pairs of OES measurements.
15. The method of claim 10 , wherein each training input in the set of training inputs is mapped to a target output in the set of target outputs.
16. The method of claim 10 , wherein the trained machine learning model comprises at least one of a logistic regression type algorithm, a multi-layer perception algorithm, or a recurrent neural network (RNN).
17. A non-transitory computer-readable medium that, when executed by a processing device, receiving, by the processing device, (i) chamber data indicative of an environmental condition of a processing chamber in which the substrate is processed according to the set of processing parameters of the current process; and (ii) sensor data including spectral data indicative of optical emission spectroscopy (OES) measurements of a plasma disposed within the processing chamber in which the substrate is processed according to the set of processing parameters of the current process; The processing device causes the sensor data to be used as an input to a machine learning model; obtaining, with the processing device, one or more outputs of the machine learning model, the one or more outputs indicative of one or more chamber condition metrics; determining, by the processing device, a recovery state of the processing chamber based on the one or more chamber condition metrics, the recovery state corresponding to a chamber recovery procedure performed following a preventive maintenance procedure; A computer-readable medium comprising instructions for causing a processing device to alter performance of a processing chamber based on a recovery condition of the processing chamber.
18. A non-transitory computer-readable medium that, when executed by a processing device, 20. The computer-readable medium of claim 17, comprising instructions for causing the processing device to determine an update to at least one processing parameter of the set of processing parameters based on the one or more chamber condition metrics and generate an updated set of processing parameters, wherein a change in performance of the processing chamber is based on the update to the at least one processing parameter of the set of processing parameters.
19. The computer-readable medium of claim 17 , wherein the change in performance corresponds to a chamber seasoning procedure.
20. A non-transitory computer-readable medium that, when executed by a processing device, Use the sensor data as input to a statistical model, receiving one or more outputs from the statistical model, the one or more outputs indicating a confidence level that the temporally associated data points of the chamber data and the spectral data accurately indicate the condition of the process chamber, the statistical model being generated using a regression between the historical chamber data and the historical spectral data; 20. The computer readable medium of claim 17 comprising instructions for determining whether a confidence level meets a threshold condition.
