Growth chamber smart aging

Abstract

A method of analyzing completion of aging of a semiconductor processing chamber may include training a model using aging cycle characteristic data obtained from an existing semiconductor processing chamber. The supervised learning process may mark the characteristic data based on expert-determined identification aging completion, and may optionally mark the characteristic data based on chamber opening event information or preventive maintenance information. The trained model may be used to characterize another chamber during aging to determine whether aging is complete and / or when an aging cycle may be performed or how long an aging cycle may be performed or how many aging cycles may be performed prior to completion of aging.

Description

[0001] Cross-references to related applications

[0002] This application claims the benefit and priority of U.S. nonprovisional application No. 18 / 361,326, filed July 28, 2023, entitled “GROWTH CHAMBER SMARTSEASONING”, the entire contents of which are incorporated herein by reference for all purposes. Technical Field

[0003] This disclosure broadly describes techniques for completing the aging of semiconductor processing chambers after preventive maintenance (PM) in a reduced amount of time, compared to conventional techniques used for aging over fixed periods. More specifically, this disclosure describes systems and methods for intelligent aging of post-PM semiconductor processing chambers by monitoring chamber characteristics during aging cycles to determine when aging is complete, enabling the chambers to be suitable for wafer monitoring and manufacturing. Background Technology

[0004] Epitaxial deposition can include any type of crystal growth or deposition of a new crystalline layer performed in a known orientation relative to a seed layer. Many modern semiconductor devices utilize epitaxial growth on silicon substrate wafers. For example, providing a silicon-based precursor to a deposition chamber at temperatures ranging from 300°C to 1200°C allows for the eventual formation of an incremental layer of film lattice-aligned with the underlying seed layer. Conventional epitaxial film deposition typically involves a baking operation that can occur at temperatures greater than or about 700°C or higher. Epitaxial growth is being extended to different device structures and can be used for selective deposition or growth on specific surfaces. Summary of the Invention

[0005] As processing technologies advance, temperature ranges shrink, and newer structures are produced, epitaxial film process productivity becomes increasingly important and challenging to achieve, making improvements in recovery time after preventative maintenance all the more valuable.

[0006] In a first aspect, a method for aging a semiconductor processing chamber, such as after a chamber opening event or preventative maintenance event, is disclosed. For example, the method of this aspect may include performing an aging cycle, monitoring characteristics of the semiconductor processing chamber, and using a trained model to determine the completion of aging of the semiconductor processing chamber based on chamber signals and historical data in a library. In some examples, the method of this aspect includes performing an aging cycle using the semiconductor processing chamber or causing the semiconductor processing chamber to perform an aging cycle, determining characteristics of the semiconductor processing chamber during the aging cycle, determining based on the characteristics that aging of the semiconductor processing chamber is incomplete, causing the semiconductor processing chamber to perform an additional aging cycle, determining additional characteristics of the semiconductor processing chamber during the additional aging cycle, determining based on the additional characteristics that aging of the semiconductor processing chamber is complete, and generating an indicator identifying the completion of aging of the semiconductor processing chamber. In some examples, a model (such as a trained machine learning model) may be used to evaluate the characteristics to determine whether aging is complete. For example, the method in this aspect may include, or further include, providing characteristics to a trained machine learning model configured to receive characteristics of a semiconductor processing chamber as input and provide output relating to the completion of aging of the semiconductor processing chamber. Output may be received from the trained machine learning model, and the output may indicate or be used to determine whether aging of the semiconductor processing chamber is complete. The process of repeatedly performing an aging cycle, measuring or monitoring the characteristics of the semiconductor processing chamber during the aging cycle, providing the characteristics to the model, and determining whether aging is complete based on the model output may be repeated one or more times as needed until aging is determined to be complete.

[0007] In some non-limiting examples, the output from the model may indicate the completion of aging in the semiconductor processing chamber, the expected time for the completion of aging in the semiconductor processing chamber, the expected duration of aging in the semiconductor processing chamber, the expected number of additional aging cycles for the completion of aging in the semiconductor processing chamber, or any combination thereof. In some non-limiting examples, the method of this aspect may generate a separate indicator that identifies one or more of the expected time for the completion of aging in the semiconductor processing chamber, the expected duration of aging in the semiconductor processing chamber, the expected number of additional aging cycles for the completion of aging in the semiconductor processing chamber, or any combination thereof. Such an indicator may be generated after it has been determined that aging in the semiconductor processing chamber has not been completed, and such an indicator may be used to convey information about the aging process to a user.

[0008] In some examples, the aging cycle may include controlling the semiconductor processing chamber in any order to establish etch conditions within the semiconductor processing chamber or controlling the semiconductor processing chamber to establish epitaxial growth conditions within the semiconductor processing chamber, or both. In any example, the execution of the aging cycle may include controlling the semiconductor processing chamber to establish etch conditions within the semiconductor processing chamber or controlling the semiconductor processing chamber to establish epitaxial growth conditions within the semiconductor processing chamber, or more of these methods. In some examples, the semiconductor processing chamber may be controlled to establish etch conditions and epitaxial growth conditions in any order for the aging cycle.

[0009] In any example, the characteristics of the semiconductor processing chamber may be monitored, measured, or otherwise determined during the execution of an aging cycle, and / or used by a machine learning model to evaluate the completion of the aging process. In some non-limiting examples, the characteristics may include one or more of the following: temperature within the semiconductor processing chamber during an aging cycle; thickness of the epitaxial layer formed within the semiconductor processing chamber; power delivered to the heater associated with the semiconductor processing chamber; power setting of the heater associated with the semiconductor processing chamber; growth rate within the semiconductor processing chamber during an aging cycle; etch rate within the semiconductor processing chamber during an aging cycle; optical conditions within the semiconductor processing chamber during an aging cycle; pressure within the semiconductor processing chamber during an aging cycle; gas composition within the semiconductor processing chamber during an aging cycle; flow rate entering or leaving the semiconductor processing chamber during an aging cycle; changes in any of these or any combination thereof; or various physical or structural parameters optionally associated with the semiconductor processing chamber, such as the geometric details of the semiconductor processing chamber or any component thereof, or the design specifications for the semiconductor processing chamber or any component thereof. In some non-limiting examples, characteristics may include, correspond to, or be derived from sensor data from or from one or more optical sensors, heat sources, thermal sensors, piezoelectric sensors, or the like. Optionally, characteristics may include or correspond to virtual sensor data derived from one or more of the following: temperature within the semiconductor processing chamber during aging cycles, thickness of the epitaxial layer formed within the semiconductor processing chamber, power delivered to a heater associated with the semiconductor processing chamber, power setting of the heater associated with the semiconductor processing chamber, growth rate within the semiconductor processing chamber during aging cycles, etch rate within the semiconductor processing chamber during aging cycles, optical conditions within the semiconductor processing chamber during aging cycles, pressure within the semiconductor processing chamber during aging cycles, gas composition within the semiconductor processing chamber during aging cycles, flow rates entering or leaving the semiconductor processing chamber during aging cycles, changes in any of these or any combination thereof, or various physical or structural parameters optionally associated with the semiconductor processing chamber, such as geometric details of the semiconductor processing chamber or any component thereof, or design specifications for the semiconductor processing chamber or any component thereof.

[0010] The method in this aspect may include, or further include, training a machine learning model. For example, the machine learning model may be trained to generate output before features are provided. In an example, training the machine learning model includes receiving training features from multiple executions of an aging cycle performed through one or more semiconductor processing chambers (such as one or more reference semiconductor processing chambers), and generating training data based on the training features. In some examples, the method in this aspect includes receiving input of tagging information corresponding to the multiple executions of the aging cycle. In an example, training data is generated using tagging information identifying the completion of aging in one or more semiconductor processing chambers. Example tagging information may include or correspond to information indicating the aging cycle that determines the completion of aging, such as information that may be determined based on expert or user analysis. Example tagging information may include or correspond to information about chamber opening events or preventative maintenance events associated with multiple aging cycles, such as information indicating the duration of a chamber opening, information indicating changes to components within the semiconductor processing chamber, information indicating the preventative maintenance procedures or protocols used (e.g., dry air purification, exposure to ambient air, exposure to water), or the like. Once the training data is ready, the method in this aspect may further include executing a supervised learning algorithm to train the machine learning model using the training data. In this way, machine learning models are particularly well-suited for generating output information based on input characteristics to provide meaningful analysis of the aging process.

[0011] In some examples, training characteristics can be derived from or obtained using the same semiconductor processing chamber undergoing aging, such as by utilizing different processing assemblies and / or devices within the same semiconductor processing chamber. For instance, training characteristics may correspond to characteristics obtained during a previous aging process in the semiconductor processing chamber.

[0012] Optionally, the training of the machine learning model can be continuously updated during each completed aging process. In some examples, this approach may include performing one or more additional aging cycles on the semiconductor processing chamber or causing the semiconductor processing chamber to perform one or more additional aging cycles after determining the completion of aging, such as to obtain additional characteristic information of the semiconductor processing chamber in which aging has been completed for use in confirming completion and / or updating the training of the machine learning model. In this case, determining the completion of aging of the semiconductor processing chamber may include comparing the characteristics of the semiconductor processing chamber with previously determined characteristics of the semiconductor processing chamber, such as the characteristics of the semiconductor processing chamber obtained when the semiconductor processing chamber is in an aging state.

[0013] Optionally, the user can control the completion determined by a trained machine learning model and manually trigger the execution of one or more additional aging cycles using the semiconductor processing chamber or cause the semiconductor processing chamber to execute one or more additional aging cycles after the completion of aging in the semiconductor processing chamber is determined. In some cases, this can be used to confirm the completion of aging before using the semiconductor processing chamber in a production environment where the processing of semiconductor components will take place. In some cases, this approach may include receiving input indicating that an opening chamber event or preventative maintenance event will occur, and in response, executing one or more additional aging cycles or causing the semiconductor processing chamber to execute one or more additional aging cycles and determining the characteristics of the semiconductor processing chamber during the one or more additional aging cycles. In this way, historical characteristic data can be obtained before the opening chamber event or preventative maintenance event, which can be used to determine whether aging is complete after the opening chamber event or preventative maintenance event.

[0014] In any approach in this regard, the machine learning model may perform regression and / or classification analyses, such as using various characteristics determined during one or more aging cycles. For example, regression analysis may analyze measured characteristic information or virtual sensor data, for example, to predict aging behavior, which may be informed by training data and / or labeling information. Similarly, classification analysis may analyze characteristics determined during one or more aging cycles, optionally classifying the aging process together with labeling information (such as identifying chamber opening events or preventative maintenance events) for meaningful comparison with appropriate training data. Regardless of whether regression analysis, classification analysis, or both or other analytical techniques are used, the machine learning model may be used to identify or plot true positives in prior characteristic data associated with known aging processes in the training dataset to build high confidence in predicting aging completion in the machine learning model. In any approach in this regard, the machine learning model may compare the characteristics of an aging cycle with corresponding characteristics of the immediately preceding aging cycle to assess the completion of aging in the semiconductor processing chamber. Some examples include machine learning models that determine the characteristic changes between the current cycle and the immediately preceding aging cycle and assess the completion of aging in the semiconductor processing chamber based on whether the characteristic changes are below a threshold. In some cases, the threshold can be predetermined by a trained machine learning model. Optionally, the threshold can be adjusted, such as by receiving input corresponding to a user-defined threshold or threshold overshoot.

[0015] In another aspect, this document provides systems, such as semiconductor processing chambers and systems for operating semiconductor processing chambers. For example, a system of this aspect may be configured or adapted to age a semiconductor processing chamber and automatically determine when the aging of the semiconductor processing chamber is complete, such as based on sensor data or other characteristics measured or monitored during the aging process. In some examples, a system of this aspect may be configured, adapted, or programmed to perform any of the methods described herein. In some examples, a system of this aspect may include one or more processors and one or more non-transitory computer-readable storage devices comprising instructions that, when executed by one or more processors, cause one or more processors to perform operations. In some examples, the operation may include causing a semiconductor processing chamber to perform an aging cycle; determining the characteristics of the semiconductor processing chamber during the aging cycle; providing the characteristics to a trained machine learning model (the trained machine learning model is configured to receive the characteristics of the semiconductor processing chamber as input and provide output related to the completion of aging of the semiconductor processing chamber); receiving the output from the trained machine learning model; determining based on the output that aging of the semiconductor processing chamber is not complete; causing the semiconductor processing chamber to perform an additional aging cycle; continuing to determine the characteristics of the semiconductor processing chamber during the additional aging cycle; providing the characteristics of the semiconductor processing chamber during the additional aging cycle to the trained machine learning model; receiving additional output from the trained machine learning model; determining based on the additional output that aging of the semiconductor processing chamber is complete; and generating an indicator identifying the completion of aging of the semiconductor processing chamber. In some examples, the system of this aspect may further include one or more sensors that communicate data with one or more processors to determine the characteristics of the semiconductor processing chamber during the aging cycle. Non-limiting examples of sensors include optical sensors, heat sources, thermal sensors, piezoelectric sensors, or the like. It will be understood that a heat source can be used as a sensor because the power consumption or power output of the heat source can be measured by the operation of the heat source or during the operation of the heat source.

[0016] In the systems and methods described herein, features may optionally include or correspond to time series of data obtained from one or more sensors. Optionally, features may include or correspond to a comparison of a first time series of data obtained from one or more sensors during a first aging cycle with a second time series of data obtained from one or more sensors during a second aging cycle immediately preceding the first aging cycle.

[0017] In any example, the semiconductor processing chamber may include one or more of a quartz dome, a flow module, or a base. Optionally, the semiconductor processing chamber may include any other suitable components, such as one or more sensors, a flow module, a heat source, a light source, a window, or the like. In some examples, the system described herein may include or further include one or more sensors positioned to measure the surface characteristics of one or more of a quartz dome, a flow module, a base, or a substrate within the semiconductor processing chamber.

[0018] In another aspect, this document describes a computer program product. Example computer program products may include, for example, a non-transitory computer-readable medium comprising instructions that, when executed by one or more processors, cause one or more processors to perform operations. Without limitation, operations may correspond to any of the methods described herein. For example, operations may include causing a semiconductor processing chamber to perform an aging cycle, determining characteristics of the semiconductor processing chamber during the aging cycle, providing the characteristics to a trained machine learning model configured to receive the characteristics of the semiconductor processing chamber as input and provide output relating to the completion of aging of the semiconductor processing chamber, receiving output from the trained machine learning model, determining based on the output that aging of the semiconductor processing chamber is not complete, causing the semiconductor processing chamber to perform an additional aging cycle, continuing to determine the characteristics of the semiconductor processing chamber during the additional aging cycle, providing the characteristics of the semiconductor processing chamber during the additional aging cycle to the trained machine learning model, receiving additional output from the trained machine learning model, determining based on the additional output that aging of the semiconductor processing chamber is complete, and generating an indicator identifying the completion of aging of the semiconductor processing chamber.

[0019] As discussed elsewhere herein, a trained machine learning model can be trained to model the aging of a semiconductor processing chamber following a chamber opening event or preventative maintenance event associated with the semiconductor processing chamber. Optionally, the methods, systems, or computer program products described herein may include or be configured to perform operations including obtaining user input corresponding to one or more of the following: an identification of the duration of the chamber opening event or preventative maintenance event; changes to components of the semiconductor processing chamber during the chamber opening event or preventative maintenance event; environmental conditions during the chamber opening event or preventative maintenance event; maintenance procedures or protocols applied to the semiconductor processing chamber during the chamber opening event or preventative maintenance event; or a severity rating for the chamber opening event or preventative maintenance event. In examples, the trained machine learning model may be further configured to use the user input when generating output relating to the completion of the aging of the semiconductor processing chamber. Non-limiting examples include information about the duration of the chamber opening, information about changes to components within the semiconductor processing chamber, and information about the preventative maintenance procedures or protocols used (e.g., dry air purification, exposure to ambient air, exposure to water). This information can be useful in or through trained machine learning models to assess the progress of aging processes and provide insights into the rate of aging progress, since aging processes can correspond to physical processes that passivate surface components (such as through the removal or desorption of water) during cyclic exposure to etch and growth conditions, and various information about chamber opening events or preventative maintenance events can indicate that a larger amount of water may be incorporated into the semiconductor processing chamber during chamber opening events or routine maintenance events, potentially leading to a longer aging process.

[0020] Optionally, the output from the trained machine learning model may include or correspond to an indication of the completion or incompleteness of aging in the semiconductor processing chamber and / or one or more variance values ​​of one or more characteristics or environmental variables. One or more variance values ​​may be determined, for example, by comparing the characteristics of the semiconductor processing chamber during a first aging cycle with the characteristics of the semiconductor processing chamber during a second aging cycle immediately preceding the first aging cycle. In some examples, determining the completion of aging in the semiconductor processing chamber based on the output from the trained machine learning model may include deriving a variance value between the output of the first aging cycle and the output of the second aging cycle immediately preceding the first aging cycle, and determining that the variance value is less than a target variance value. In some examples, an input corresponding to a tolerance of the model may be received, indicating an allowable deviation from the target variance value, which may correspond to user overriding, allowing the user to control the reduction or increase of the number of aging cycles in operation.

[0021] Not wishing to be bound by any particular theory, this document may discuss the basic principles relating to the invention and the beliefs or understandings of the operations or parameters described herein. It will be appreciated that, regardless of the ultimate correctness of any mechanical interpretation or assumption, the described examples and embodiments may still be operable and useful. Attached Figure Description

[0022] A further understanding of the nature and advantages of various embodiments and examples can be achieved by referring to the remainder of the specification and the accompanying drawings, in which the same reference numerals are used in several drawings to refer to similar parts. In some cases, sub-reference numerals are associated with reference numerals to indicate one of a plurality of similar parts. When reference is made to reference numerals without specifying existing sub-reference numerals, it is intended to refer to all such plurality of similar parts.

[0023] Figure 1 A schematic side cross-sectional view of a semiconductor processing chamber according to some embodiments is shown.

[0024] Figure 2 A flowchart is shown of a method for training a machine learning model to characterize the completion of aging in a semiconductor processing chamber, according to some embodiments.

[0025] Figure 3 A block diagram illustrating how semiconductor processing chamber sensor characteristic data can be collected and processed to generate virtual sensor data according to some embodiments is shown.

[0026] Figure 4 A graph showing time-series data measured during aging cycles through a semiconductor processing chamber, according to some embodiments, is presented.

[0027] Figure 5 A bar graph showing virtual sensor data used by a machine learning model during aging in a semiconductor processing chamber is illustrated according to some embodiments.

[0028] Figure 6A , Figure 6B , Figure 6C and Figure 6D Additional bar graphs are shown, according to some embodiments, displaying virtual sensor data used by a machine learning model during the aging of a semiconductor processing chamber subjected to different opening events or preventative maintenance event conditions.

[0029] Figure 7 A flowchart is shown, according to some embodiments, for training a model to predict the completion of semiconductor processing chamber aging.

[0030] Figure 8 A flowchart of a method for aging a semiconductor processing chamber according to some embodiments is shown.

[0031] Figure 9 A flowchart is shown illustrating the use of a trained machine learning model to evaluate the completion of aging in a semiconductor processing chamber, according to some embodiments.

[0032] Figure 10 An exemplary computer system in which various embodiments may be implemented is shown. Detailed Implementation

[0033] Thermochemical vapor deposition (CVD) is one of the hottest processes in semiconductor manufacturing. During the deposition process, precursor gases can flow into the semiconductor processing chamber at temperatures ranging from approximately 300°C to approximately 1200°C during deposition, etching, and / or chamber cleaning processes. The temperature setpoint can vary throughout the process depending on the processing formulation, and the temperature within the semiconductor processing chamber can rise / fall between different setpoints based on the chamber's thermal properties and as different gases flow into or out of the chamber. The rate and quality of epitaxial growth or etching can depend not only on the absolute temperature and gas mixture but also on the consistent behavior of the semiconductor processing chamber. Broadly speaking, the epitaxial growth and / or etching performance within the semiconductor processing chamber can vary based on the chamber's processing history. For optimal performance, the semiconductor processing chamber should exhibit very stable conditions between different epitaxial growth or etching cycles.

[0034] At a more detailed level, the epitaxial growth and etching behavior within a semiconductor processing chamber can be altered by the presence of water or impurities (e.g., halide salts) within or on the surfaces of the chamber or its components. Semiconductor processing chambers may be periodically opened or subjected to preventative maintenance within production or research facilities, allowing contaminants or other alterations to the surfaces. For example, exposure to the atmosphere can introduce water into the semiconductor processing chamber, and changes to the processing equipment during preventative maintenance or troubleshooting can introduce water or other impurities, or alter the balance or presence of passivated surfaces. Different chamber opening events or preventative maintenance events can expose semiconductor processing chambers to varying environmental conditions (e.g., relative humidity) for varying durations or alter various components, leading to instability in their condition.

[0035] To restore performance after a chamber opening or preventative maintenance event, semiconductor processing chambers can undergo a cyclic process, commonly referred to herein as "aging," in which the temperature and gas mixture within the chamber are repeatedly circulated, such as cycling between etch conditions and epitaxial growth conditions. Aging can be used to remove moisture and / or impurities introduced or absorbed into or on the semiconductor processing chamber, to establish passivated surfaces of the semiconductor processing chamber or components therein, or otherwise return the semiconductor processing chamber to conditions prior to the chamber opening or preventative maintenance event. Because varying amounts of aging cycles may be required to return to conditions prior to the chamber opening or preventative maintenance event, depending on the duration of the event, environmental conditions during the event, changes to internal components during the event, or the like, a common protocol for aging is to subject the semiconductor processing chamber to a large, fixed number of aging cycles, which is sufficient to establish stable conditions after any chamber opening or preventative maintenance event.

[0036] However, in many cases, the large, fixed number of aging cycles in conventional aging processes can be excessive for establishing stable conditions, as some semiconductor processing chambers may require fewer aging cycles after some chamber opening events or preventative maintenance events. Since each aging cycle takes time, subjecting a semiconductor processing chamber to excessive aging cycles can reduce available production time. This disclosure overcomes this and other problems by establishing techniques for monitoring conditions within a semiconductor processing chamber to determine when sufficient aging has been achieved and / or when the semiconductor processing chamber has returned to or has established stable conditions suitable for production. Furthermore, aspects described herein include using trained machine learning models to evaluate monitored conditions to automatically determine when aging is complete or to provide feedback to the user regarding the expected duration or number of aging cycles required to return the semiconductor processing chamber to stable conditions.

[0037] Figure 1 A schematic side cross-sectional view of an example semiconductor processing chamber 100 according to some embodiments is shown. The semiconductor processing chamber 100 may represent a deposition chamber. In some embodiments, the semiconductor processing chamber 100 may include or comprise an epitaxial deposition or growth chamber or an etching chamber. The semiconductor processing chamber 100 may be used to grow or process an epitaxial film on a substrate 102. The semiconductor processing chamber 100 may generate a precursor flow across the top surface 150 of the substrate 102. Throughout this disclosure, the epitaxial deposition chamber may be used as a specific example of a semiconductor processing chamber. However, the embodiments described herein are equally applicable to any semiconductor processing chamber. Thus, any specific reference to an epitaxial deposition chamber may be considered, and more generally, to any semiconductor processing chamber.

[0038] Semiconductor processing chamber 100 may include an upper body 156, a lower body 148 disposed below the upper body 156, and a flow module 112 disposed between the upper body 156 and the lower body 148. The upper body 156, the flow module 112, and the lower body 148 may form the chamber body. A substrate support 106, an upper window 108, a lower window 110 (such as a lower dome), a plurality of upper heat sources 141, and / or a plurality of lower heat sources 143 may be disposed within the chamber body. As shown, a controller 120 may communicate with the semiconductor processing chamber 100 and may be used to control processes and operations, such as those described herein. The controller 120 and the semiconductor processing chamber 100 may be part of a larger substrate processing system or tool platform.

[0039] The semiconductor processing chamber 100 may also include a plurality of upper heat sources 141 for heating the top portion of the semiconductor processing chamber 100 and a plurality of lower heat sources 143 for heating the bottom portion of the semiconductor processing chamber 100. The plurality of upper heat sources 141 may be disposed between the upper window 108 and the cover 154. The plurality of upper heat sources 141 may form part of an upper heating module 155. The plurality of lower heat sources 143 may be disposed between the lower window 110 and the chamber bottom plate 152. The plurality of lower heat sources 143 form part of a lower heating module 145. Figure 1 In the illustrated embodiment, heat sources 141 and 143 may be heat lamps. Other heat sources, such as resistance heaters, light-emitting diodes (LEDs), and / or lasers, may also be used without limitation. Heat sources 141 and 143 may be coupled to reflector 175. Reflector 175 may be configured to redirect heat energy from heat sources 141 and 143 toward processing volume 136.

[0040] The upper window 108 may be an upper dome and at least partially formed of an energy-transmitting material (such as quartz). Therefore, the upper window 108 may also be referred to as a dome or a quartz dome. The lower window 110 is a lower dome and at least partially formed of an energy-transmitting material (such as quartz). The lower window 110 may also be simply referred to as a dome or a quartz dome. The upper window 108 may include a recessed or flat first surface 111 (in... Figure 1 In the illustrated embodiment, the first surface 111 is recessed. The upper window 108 includes a raised second surface 113. The second surface 113 may face the substrate support 106. The upper window 108 may include an inner portion 122 and / or an outer portion 124. The first surface 111 and the second surface 113 may form at least a portion of the inner portion 122. The inner portion 122 may be transparent, and the outer portion 124 may be opaque. The outer portion 124 may be received at least partially in one or more sidewalls of the semiconductor processing chamber 100 (such as in the flow module 112).

[0041] Processing volume 136 and cleanup volume 138 may be positioned between upper window 108 and lower window 110. Processing volume 136 and cleanup volume 138 may be portions of an internal volume at least partially defined by upper window 108, lower window 110 and one or more pads 163. Upper window 108 may at least partially define processing volume 136.

[0042] The internal volume may include a substrate support 106 disposed therein. The substrate support 106 may be disposed between an upper window 108 and a lower window 110. The substrate support 106 may include a support surface 123 supporting a substrate 102. The substrate support 106 may also be referred to as a base, pressure plate, or platform. The substrate support 106 may include a top surface on which the substrate 102 is disposed. The substrate support 106 may also be attached to a shaft 118. The shaft 118 may be connected to a motion assembly 121. The motion assembly 121 may include one or more actuators and / or adjustment devices providing movement and / or adjustment of the shaft 118 and / or the substrate support 106 within the processing volume 136.

[0043] The substrate support 106 may include a lifting rod hole 107 disposed therein. The size of the lifting rod hole 107 may be adjusted to receive a lifting rod 132 for lifting the substrate 102 from the substrate support 106 before or after performing a deposition process. When the substrate support 106 is lowered from a processing position to a transfer position, the lifting rod 132 may rest on a lifting rod stop 134. The lifting rod stop 134 may be coupled to a second shaft 104. A substrate (such as substrate 102) may be transferred through a transfer gate 137 (such as a slit valve) into and / or out of the internal volume of the semiconductor processing chamber 100. When the transfer gate 137 is open, a transfer device (on which the substrate is supported) may extend through the transfer gate 137 into the internal volume, such that the lifting rod 132 can lift the substrate from the transfer device and place the substrate on the substrate support 106 for processing. After processing, the lifting rod 132 can lift the substrate from the substrate support 106 and place the substrate on the transfer device, and the transfer device can retract through the open transfer door 137 to remove the substrate from the semiconductor processing chamber 100.

[0044] The flow module 112 may include a plurality of gas inlets 114, a plurality of purge gas inlets 164, and one or more gas outlets 116. The plurality of gas inlets 114 and the plurality of purge gas inlets 164 may be disposed on the sides of the flow module 112 opposite to the one or more gas outlets 116. One or more flow guides 117 may be disposed below the plurality of gas inlets 114 and / or one or more gas outlets 116. The one or more flow guides may include, for example, one or more preheating rings. The one or more flow guides 117 may be disposed above the purge gas inlets 164. One or more gaskets 163 may be disposed on the inner surface of the flow module 112 and may protect the flow module 112 from the reactive gases used during deposition and / or cleaning operations. The gas inlets 114 and the purge gas inlets 164 may each be positioned such that the gas flows parallel to the top surface 150 of the substrate 102 disposed within the processing volume 136. The gas inlets 114 may be fluidly connected to one or more processing gas sources 151 and one or more cleaning gas sources 153. A purified gas inlet 164 may be fluidly connected to one or more purified gas sources 162 and / or one or more clean gas sources 153. One or more gas outlets 116 may be fluidly connected to an exhaust pump 157. One or more process gases supplied using one or more process gas sources 151 may include one or more reactive gases (such as gases containing silicon (Si), phosphorus (P), and / or germanium (Ge)) and / or one or more carrier gases (such as nitrogen (N2) and / or hydrogen (H2)). One or more purified gases supplied using one or more purified gas sources 162 may include one or more inert gases (such as argon (Ar), helium (He), hydrogen (H2), and / or nitrogen (N2)). One or more clean gases supplied using one or more clean gas sources 153 may include one or more gases containing hydrogen (H) and / or chlorine (Cl). In some embodiments, one or more process gases may include silicon phosphide (SiP) and / or phosphine (PH3), and one or more clean gases may include hydrochloric acid (HCl). It will be understood that these example gases are non-limiting and other process gases, purified gases, or clean gases may be used.

[0045] One or more gas exhaust outlets 116 may be further connected to or include an exhaust system 178. The exhaust system 178 may fluidly connect one or more gas exhaust outlets 116 and an exhaust pump 157. The exhaust system 178 may facilitate controlled deposition of layers on the substrate 102. The exhaust system 178 may be disposed on the opposite side of the semiconductor processing chamber 100 relative to the flow module 112.

[0046] Controller 120 may include a central processing unit (CPU), memory containing instructions, and support circuitry for the CPU. For example, controller 120 may include one or more processors. The one or more processors may be distributed via a network or wireless connection among local controllers of semiconductor processing chamber 100, tool servers on tools or platforms operating multiple different types of semiconductor processing chambers, and / or cloud-based or facility-based servers. One or more non-transitory computer-readable media may store instructions that cause one or more processors of controller 120 to perform the operations described herein. The computer-readable media and one or more processors may collectively constitute controller 120 along with other components, which are not necessarily limited to a single computer system. Rather, controller 120 may be distributed across different locations among several different computer systems. In some embodiments, controller 120 may be communicatively coupled to a dedicated controller, and controller 120 may function as a central controller. Alternatively, any of the operations performed or described herein may be distributed between controller 120 and other computing systems. For example, processing operations, model training, and / or model storage may be performed via controller 120, via another non-residential competing system, via a cloud-based computing system, or via any combination of these or other systems. Some of these processing operations can also be distributed to individual sensors or control circuits within the chamber.

[0047] Examples of computer systems that can be used to implement at least a portion of controller 120 are described below. Figure 10 The details are described below. Operating parameters (pressure of the processed gas, flow rate of the processed gas, and / or rotational position of the processing unit) and operations may be stored as software routines in a computer-readable medium, which are executed or invoked to transform controller 120 into a dedicated controller to control the operation of the various chambers / modules described herein. Controller 120 is configured to perform any of the operations described herein. The various operations described herein may be performed automatically using controller 120, or may be performed automatically and / or manually using certain operations performed by the user.

[0048] Controller 120 can control various items directly or via other computers and / or controllers. For example, instructions executed by controller 120 can cause semiconductor processing chamber 100 to perform operations in the recipe, such as gas flow, temperature increase, substrate reception, etc. Controller 120 can be configured to control rotational position, heating, and gas flow through semiconductor processing chamber 100 by providing outputs to heat sources 141, 143, gas flow, and motion assembly 121. Control may include control of upper heat source 141, lower heat source 143, process gas source 151, purging gas source 162, motion assembly 121, and exhaust pump 157.

[0049] Controller 120 can be configured to adjust the output to control based on sensor readings, a system model, and stored readings and calculations. Controller 120 may include embedded software and compensation algorithms for calibrating measurement results. Controller 120 may include one or more machine learning and / or artificial intelligence algorithms that analyze sensor readings, estimate optimized parameters for deposition, etching, cleaning, and / or scavenging operations, or evaluate performance or stability within the semiconductor processing chamber 100, such as for determining or estimating when or whether an aging process is complete. The one or more machine learning and / or artificial intelligence algorithms may use, for example, regression models (such as linear regression models) or clustering techniques. The algorithms may be unsupervised or supervised.

[0050] Cover 154 may include multiple sensors disposed therein or on it, such as those for measuring characteristics within the semiconductor processing chamber 100, such as measuring temperature within the semiconductor processing chamber, monitoring optical conditions within the semiconductor processing chamber 100, or monitoring etching or growth conditions, rates, or characteristics within the semiconductor processing chamber. For example, a central temperature sensor 172 may be disposed on cover 154 and configured to measure temperature at or near the central portion of the semiconductor processing chamber 100. More specifically, the central temperature sensor 172 may be configured to measure the temperature of the upper window 108 or the quartz dome. The central temperature sensor 172 may also be configured to measure the temperature of the top surface 150 of the substrate 102 and / or the substrate support 106. One or more temperature sensors 173 may also be distributed around the periphery of the semiconductor processing chamber 100. These temperature sensors 173 may be configured to measure specific locations around the periphery of the semiconductor processing chamber 100. For example, these temperature sensors 173 may point to the periphery of the upper window 108, to internal components such as one or more gaskets 163, to the periphery of the substrate support 106 or the base, etc. Despite Figure 1Not explicitly illustrated, the semiconductor processing chamber 100 may include other sensors distributed throughout the semiconductor processing chamber 100. For example, a temperature sensor may be disposed in the lower heating module 145 and positioned to measure the temperature at the periphery and / or center of the bottom of the substrate support 106. As another example, one or more optical sensors may be disposed in the cover 154 and positioned to measure the optical properties of the substrate support 106, the upper window 108, the lower window 110, or the like. In some examples, one or more piezoelectric or piezoresistive sensors may be positioned within the semiconductor processing chamber 100, such as in or on the substrate support 106 or the shaft 118, to characterize mass changes or rates of change during processing. In some examples, the upper heat source 141 and the lower heat source 143 may be used as sensors or operated as sensors, as the power consumption of the upper heat source 141 and the lower heat source 143 may be useful characteristics that can provide insight into the stability within the semiconductor processing chamber 100.

[0051] More typically, multiple sensors may be specifically targeted at individual components of the semiconductor processing chamber 100. For example, a central temperature sensor 172 may be specifically assigned to measure the temperature at the center of the upper window 108 or the quartz dome. Temperature or optical sensors may be specifically oriented and assigned to measure the temperature or optical properties below the base or substrate support 106 and / or to measure the temperature or optical properties of the substrate or the top of the base or substrate support 106. Temperature or optical sensors may be specifically assigned and oriented toward the liner in the semiconductor processing chamber 100. These sensor assignments are provided by way of example only and are not intended to be limiting. Figure 1 Any of the components depicted or described above may be assigned a specific sensor configured to measure characteristics of that component (e.g., temperature, optics, etc.). As will be described below, this allows individual sensor data to be used to analyze characteristics or changes in characteristics associated with a specific component, or to assess stability within the semiconductor processing chamber 100 on a general or more granular basis. This may also allow the controller to identify specific components that could cause the semiconductor processing chamber 100 to malfunction.

[0052] The temperature sensor can be implemented using any type of temperature sensor. For example, some embodiments may use a pyrometer. Some embodiments may use a thermocouple. Some embodiments may use a resistance temperature detector (RTD). Some embodiments may use a semiconductor-based temperature sensor. Some embodiments may use a temperature scanning system that includes a single sensor (e.g., a pyrometer) and is configured to scan the temperature measurements throughout the chamber along a predefined path through the upper / lower window. More generally, the temperature sensor can be implemented using any device configured to provide an output that varies in a manner dependent on the temperature within the semiconductor processing chamber 100.

[0053] The sensor can be used to sample conditions (e.g., temperature or optical properties) inside the semiconductor processing chamber 100 during different operations performed through the semiconductor processing chamber 100. For example, the sensor can be used to acquire measurement data that can be used to train a model.

[0054] Figure 2 A flowchart of a method 200 for training a machine learning model to characterize aging of a semiconductor processing chamber or evaluate the completion of an aging process, according to some embodiments, is shown. The method can be executed by a controller, including any of the controllers and / or computer systems described herein. The semiconductor processing chamber may include an epitaxial deposition chamber or an epitaxial chamber.

[0055] At block 205, method 200 may include receiving training characteristics from multiple executions of a plurality of aging cycles performed through one or more semiconductor processing chambers. Example training characteristics may include time-series data obtained from one or more sensors within the semiconductor processing chamber, or virtual sensor data derived from time-series data. In some embodiments, the model may be trained for a type of semiconductor processing chamber (e.g., an epitaxial chamber), and the training characteristics may be received from a plurality of different semiconductor processing chambers of that type (e.g., epitaxial chambers) or from a specific semiconductor processing chamber. Alternatively, some embodiments may train the model for a specific semiconductor processing chamber, and the training characteristic data may be received from multiple process executions on the same semiconductor processing chamber. It will be understood that temperature data is used only as an example of a characteristic that can be used to evaluate stability within the semiconductor processing chamber during aging and to determine the completion of aging. The embodiments and examples described herein may also use any other characteristic of the chamber instead of temperature. For example, in any embodiment, temperature measurements may be replaced by and combined with other parameters such as optical measurements, heater power, growth rate, etch rate, or the like. Thus, temperature data is used only as an example and is not intended to be limiting. Any embodiment specifically involving temperature data may alternatively use any other semiconductor processing chamber characteristics without limitation.

[0056] In some cases, training characteristics may include processed characteristics corresponding to virtual sensors, such as one or more sensor datasets being combined or processed to generate a dataset that can function similarly to a single sensor dataset. In some cases, sensor data acquired during subsequent aging cycles in a single chamber can provide virtual sensor data that can establish characteristic changes between subsequent aging cycles. As an example, in the case of obtaining time series of temperature sensor measurements for different aging cycles, analysis can be performed on the time series to determine the variance or other measure of the degree to which the time series of one aging cycle differs from that of the immediately preceding aging cycle. In some cases, as aging nears completion, the time series of temperature or other sensor data may change by a small amount between aging cycles; therefore, determining the changes in the time series data can provide a useful virtual sensor signal that can be used to assess the progress of aging.

[0057] At block 210, method 200 may also include receiving marker information indicating the completion of aging in one or more semiconductor processing chambers during multiple aging cycles. The marker information may be received as user input in some cases and may determine when an expert or other user determines when aging is complete based on training characteristics. That is, the marker may identify the specific aging cycle at which aging is complete or the number of aging cycles subsequently completed. In some cases, the marker information may specify a threshold, such as variance, for virtual sensor measurements below which aging is automatically considered complete, such as when one or more virtual sensor measurements in a row are below the threshold. In some examples, additional marker information may be provided in addition to indicators of the specific aging cycle at which aging is complete or the number of aging cycles subsequently completed. For example, the marker information may also include chamber open event data or preventative maintenance event data to establish information about the conditions the semiconductor processing chamber undergoes prior to the aging process. For example, information about environmental conditions (e.g., humidity, temperature) or duration during a chamber open event may be used to mark training characteristics. Similarly, information about specific preventative maintenance performed on a semiconductor processing chamber (e.g., indications of changes to specific parts or cleaning processes) can be used to mark training characteristics.

[0058] Method 200 may also include generating training data at box 215 from training features labeled with aging completion information, chamber opening event data, and / or preventative maintenance data, and performing a supervised learning algorithm at box 220 to train the model using the training data. In some cases, the labeled information can be used by the model to develop formulas characterizing how the training features change with aging cycles, so that predictions about when aging is complete can be evaluated based on a small number of aging cycles and optionally chamber opening event data and / or preventative maintenance data.

[0059] Figure 3 A block diagram illustrates how virtual sensor data 312 (such as variance data, exemplified) can be generated from characteristics measured during aging, according to some embodiments. The semiconductor processing chamber 302 may include a plurality of temperature sensors 304, a plurality of optical sensors 306, and / or a plurality of piezoelectric sensors 308, such as in… Figure 1 As shown in the text or referenced above Figure 1As described, during the aging cycle performed through the semiconductor processing chamber 302, temperature sensor 304, optical sensor 306, and / or piezoelectric sensor 308 can record sensor data measurements in real time throughout the process. Sensor sampling stage 314 can generate time-series data 310 as a series of continuous measurements from temperature sensor 304, optical sensor 306, and / or piezoelectric sensor 308. The time-series data 310 can then be processed by conversion stage 316, which converts the time-series data 310 into virtual sensor data 312.

[0060] For example, an aging process performed through semiconductor processing chamber 302 may include multiple aging cycles. Each cycle may be associated with a specific set of processing conditions, such as epitaxial growth conditions or etching conditions. In some cases, a single cycle may include two sub-cycles (e.g., a growth condition sub-cycle and an etching condition sub-cycle). During a single cycle or sub-cycle, the characteristics of semiconductor processing chamber 302 may be monitored by temperature sensor 304, optical sensor 306, and / or piezoelectric sensor 308. For example, specific processing conditions may include heating the semiconductor processing chamber to a specific temperature setpoint, but the temperature may vary over time, such as when the temperature reaches equilibrium. In some cases, chemical reactions may occur during etching or epitaxial growth, which may affect the temperature or the rate at which the temperature reaches equilibrium. In some examples, generating dummy sensor data may compare time-series data of one cycle or sub-cycle with time-series data of another cycle or sub-cycle (e.g., the cycle or sub-cycle immediately preceding it), such as to derive the variation (e.g., variance) of the time-series data between cycles or sub-cycles.

[0061] Figure 4 Examples of time-series data of temperature measurements at two different locations within a semiconductor processing chamber during several aging cycles, according to some embodiments, are provided. Although Figure 4 The diagram illustrates only the data from the first and every tenth cycle up to the fiftieth aging cycle. Only a limited amount of data is shown to provide clarity and avoid obscuring other details. In practice, time-series data of this property can be obtained for each aging cycle.

[0062] Time series data 310 ( Figure 4 (Some examples are provided) can be combined at conversion stage 316 to generate virtual sensor data 312. As described above, examples of virtual sensor data that can be used to evaluate the aging progress of a semiconductor processing chamber can be a comparative evaluation of sensor data for a specific aging cycle with sensor data for the immediately preceding aging cycle. For example, aging cycle 2 ( Figure 4Temperature time-series data (not shown in the diagram) can be directly compared to temperature time-series data of aging cycle 1. In some examples, differences or ratios of the temperature time-series data can be calculated, which can generate a time series of difference or ratio data that can be used as virtual sensor data 312. In some cases, the time series of difference or ratio data can be combined into a single value, such as the sum or average of the difference or ratio data, which provides a single value characterizing how time-series data and specific characteristics change between aging cycles. In other examples, variance (e.g., the root mean square deviation from the mean) can be determined. In some examples, this single value can be used as virtual sensor data 312. In some cases, multiple different comparisons of time-series data from different sensors can be obtained in this way and further combined, such as according to various weighting factors, to generate virtual sensor data 312. It will be understood that these are merely a few examples of how virtual sensor data 312 can be generated and are not intended to be limiting.

[0063] Figure 5 Example bar chart 500 is provided to illustrate the changes in virtual sensor data information as aging cycles (provided here as the variance between subsequent cycles of the first 50 cycles). Figure 5 The diagram also shows a threshold of 510 that indicates the variance value; below this value, aging can be considered complete. Figure 5 The diagram also shows mark 512, indicating that aging is completed at approximately aging cycle 23 or 24.

[0064] In such Figure 5 When the virtual sensor data shown is used as training data, the user can provide input corresponding to marker 512, which determines when aging is considered complete, such as at aging cycle 23 or 24. For example, the position of marker 512 can be input by the user. Additionally or alternatively, the user can provide input corresponding to threshold 510 as marker information to determine the virtual sensor value that determines the completion of aging, such as once one or more (e.g., 2, 3, 4, 5, 6, etc.) virtual sensor values ​​are below threshold 510.

[0065] Figure 6A , Figure 6B , Figure 6C and Figure 6D Additional example bar charts are provided to illustrate how virtual sensor data changes with different aging cycles. Figure 6A , Figure 6B , Figure 6C and Figure 6D Threshold 610 and marking information 612 are also provided. Here, additional marking information 614 also includes information related to chamber opening events or preventative maintenance events.

[0066] In some cases, similar to Figure 4 , Figure 5 or Figures 6A to 6D As shown, sensor data or virtual sensor data can be acquired during the aging process and displayed in real time on an interface (e.g., a display) as data is collected or processed during aging, such as to allow the user to monitor the progress of aging. Optionally, the interface can be used to obtain marker information as input, such as during a training process, thereby allowing the user to receive positioning indicators, selections from drop-down menus, or text input.

[0067] Return to Figure 2 The training method in flowchart 200 may also include generating training data from training features that utilize aging to complete information labeling. Figure 7 A flowchart 700 for training a model to predict aging completion is shown according to some embodiments and examples. Training characteristic data 705 can be labeled using corresponding labeled data 710 for each semiconductor processing chamber. For example, the semiconductor processing chamber may use one or more sensors to perform the aging process and collect characteristic data 705, and the labeled data 710 may be determined, for example, via user input. Here, the labeled data 710 may correspond to or indicate the number of aging cycles required to determine aging completion, such as based on determination by a user or expert. In some cases, the labeled data 710 indicating the number of aging cycles required to determine aging completion may be automatically determined, for example, based on one or more characteristic data 705 that satisfy or reach a threshold. In some examples, the number of aging cycles that require the threshold to be satisfied before determining automatic aging completion may be adjustable or provided as labeled information. In some examples, the threshold may be adjustable or provided as labeled information.

[0068] Additionally, the chamber opening or preventative maintenance event information 715 may be determined, for example, via user input or derived from one or more data storage systems, and optionally used as additional labeling information for generating training data or training a machine learning model. As described above, the chamber opening or preventative maintenance event information 715 can be used to correlate with the number of aging cycles prior to aging completion, since various physical processes associated with the chamber opening or preventative maintenance event can affect the number of cycles prior to aging completion. Making this chamber opening or preventative maintenance event available as labeling information can be used to improve model training.

[0069] Characteristic data 705, labeled using one or both of the labeled data 710 and / or chamber opening or preventative maintenance event information 715, can be considered as the training set 720 for training the model. Although Figure 7 Only a single training set 720 is shown in the figure. It should be understood that multiple training sets can be provided to train algorithms for models from multiple chambers or multiple processes from the same chamber.

[0070] exist Figure 2 In the training method of flowchart 200, the method may also include executing a supervised learning algorithm to train the model using training data. Figure 7 The training phase 725 is illustrated as potentially including the execution of a training algorithm 730 on a model 735. The model 735 may include a machine learning model trained to model the aging of a semiconductor processing chamber (e.g., an epitaxial chamber) following a chamber opening event or preventative maintenance event. The model 735 may take into account the stability of surfaces within the semiconductor processing chamber, and the training algorithm 730 may receive characteristic data from component-specific sensors throughout the semiconductor chamber. For example, as the training algorithm 730 adjusts the weights or parameters of the model 735, these weights or parameters may be trained such that the model 735 approximates the passivation behavior of individual components within the semiconductor processing chamber during epitaxial growth or etching conditions. The training phase 725 may also be trained as a classification model that uses the characteristic data and chamber opening or preventative maintenance event information to output a classification based on the degree of matching between the input data and the training data. For example, the model 735 may be implemented using a neural network or convolutional learning network to classify the input, although the model is not intended to be limited to neural networks. Any type of machine learning model can be used, including data pattern models, physical models, hybrid models, and so on.

[0071] After training phase 725 is completed, the process can generate a trained model 740. The trained model 740 can now be used to predict whether aging of the semiconductor processing chamber is complete or to predict additional information about aging completion immediately after each aging cycle or sub-cycle is completed. Additionally, as new training data becomes available, the trained model 740 can be continuously retrained and / or refined over time. For example, some implementations may periodically collect characteristic data from existing semiconductor processing chambers during aging and request or obtain confirmation from the user whether a cycle is complete for labeling purposes. This information can be combined to form newly labeled training data, and training phase 725 can be repeated to continuously or as needed refine the trained model 740.

[0072] Figure 8A flowchart of a method 800 for aging a semiconductor processing chamber, according to several examples, is shown. This method can be performed using a trained model 740 as described above. For example, several different scenarios may benefit from using the trained model 740 to characterize the aging of a semiconductor processing chamber and predict whether and / or when the aging process will be completed. For instance, when a new semiconductor processing chamber is added to a manufacturing facility, the ability of the new chamber to have stable conditions for controlled processing (e.g., etching or epitaxial growth) may be of interest. The trained model 740 can be used to determine how long it may take to complete the aging of the new semiconductor processing chamber so that it can be used in production. Similarly, when an existing semiconductor processing chamber is used and opened or preventative maintenance is performed on it, the trained model 740 can be used to determine how long it may take to complete the aging of the semiconductor processing chamber so that it can be used in production again. For example, characteristic data can be obtained during one or more aging cycles and provided as input to the trained model 735. The output of the trained model 735 can then indicate whether aging is complete, how many aging cycles are expected before aging is complete, and / or the duration before aging is complete.

[0073] The method may include performing an aging cycle or sub-cycle using a semiconductor processing chamber or causing the semiconductor processing chamber to perform an aging cycle or sub-cycle (805), such as processes that change or cause changes in temperature, pressure, or gas composition mixture within the chamber during the process, such as for establishing etching conditions or epitaxial growth conditions, for example. This process may be related to the above... Figures 3 to 5 The process described herein and used to generate various characteristic data is the same. For example, the process may include sequentially establishing etch and growth conditions within a semiconductor processing chamber, which may passivate surfaces therein and optionally result in the removal of unwanted impurities from the semiconductor processing chamber. Aging cycles may occur in the absence of a semiconductor substrate (e.g., a wafer) within the semiconductor processing chamber and may result in the repeated growth and / or etching of materials on the surfaces of other components within the semiconductor processing chamber, and the construction of a passivation layer that can be used to provide stable conditions within the semiconductor processing chamber during etch or growth conditions.

[0074] The method may also include determining characteristics of the semiconductor processing chamber during aging cycles, such as temperature measurements, optical measurements, and piezoelectric measurements (810). As described above, temperature, optical, or piezoelectric sensors may be associated with individual components and / or locations within the semiconductor processing chamber. These sensors can record various physical parameters during an aging cycle or sub-cycle as a time-series data set. The method may optionally include deriving virtual sensor data from the measured characteristics. As described above, virtual sensor data can be derived by comparing time-series data from a given aging cycle with time-series data from immediately preceding aging cycles, such as measurements used to provide variability of various characteristics from aging cycle to aging cycle.

[0075] The method may additionally include providing characteristic data to a trained model (815). As described above, the trained model may be configured to receive characteristic data as input and provide one or more outputs relating to the completion of aging. The outputs may indicate whether the aging process is complete, or may provide information about the expected completion of the aging process. Figure 9 A flowchart 900 is shown illustrating the use of a trained model 904 to predict the completion of aging in a semiconductor processing chamber according to some embodiments. As described in detail above, training data with multiple labels may be provided from one or more reference semiconductor processing chambers 902 to train or retrain the trained model 904. When the semiconductor processing chamber 906 undergoes a chamber opening event or a preventative maintenance event, the process described above may be performed on the semiconductor processing chamber 906 to perform aging and determine when aging is complete, without having to undergo excessive aging cycles beyond those required to ensure adequate aging. During each aging cycle, characteristic data 908 is obtained and provided as input to the trained model 904.

[0076] Some embodiments may also provide chamber opening event or preventative maintenance event information 910 as supplementary input to the trained model 904. This information may be provided supplementally and used by the trained model 904 to more accurately determine the completion of aging or estimate the amount of time or number of aging cycles required until aging is complete. For example, opening event or preventative maintenance event information 910 can increase the accuracy of the trained model 904 because this additional information can provide a more sophisticated understanding of the amount of aging cycles that may be required to properly passivate surfaces and / or remove impurities that may have been introduced during the chamber opening or preventative maintenance event.

[0077] Return to Figure 8 Method 800 may also include receiving output from a trained model that provides aging completion information generated based on the provided features (820). Reference Figure 9The output 912 of the trained model 904 may include a binary (e.g., 0 / 1 or true / false) and / or scalar value indicating whether aging is complete, a variance level, a deviation from a threshold representing completion, the number of additional aging cycles expected before aging is complete, the expected duration before aging is complete, or the expected time for aging to complete.

[0078] Method 800 may further include determining, based on the output, whether aging of the semiconductor processing chamber is complete (825). In some examples, the output may directly represent whether aging is complete, such as in the form of a binary value (yes / no). In other examples, the output may indicate the number of additional aging cycles expected before aging is complete, such as in the form of an integer value; the integer value may be zero if aging is complete, and the integer value may be one or more than one if aging is not complete.

[0079] Even if the completion of aging is not determined, method 800 may return to operation 805, in which the semiconductor processing chamber may perform an additional aging cycle, during which characteristics are measured (810), the additional characteristics are provided to a trained machine learning model (815), and an output is received from the trained machine learning model based on the additional characteristics (820), and then the completion of aging is evaluated again (825). Once the completion of aging is determined, method 800 may then proceed to generating an indicator indicating the completion of aging (830).

[0080] Furthermore, refer to Figure 9 Indicators indicating the completion of aging, or indicators related to the completion of aging, may optionally be generated at any time during the aging process. Indicators may provide notification of whether aging is complete, the expected time or number of additional aging cycles prior to completion, the expected duration of the aging process, or a timer countdown, which may optionally be updated in real time as additional characteristic data becomes available and the aging model is evaluated. Optionally, indicators may provide characteristic data, such as in the form of time-series data, which can provide this information to experts or other users, such as for troubleshooting or other informational purposes. In some examples, indicators may provide virtual sensor data, such as variance data, thereby displaying aging progress in a form more easily understood by experts or other users.

[0081] It should be understood that Figure 2 and Figure 8 The specific steps illustrated provide a particular method for predicting the aging performance of semiconductor processing chambers based on training and using models according to various embodiments. Other sequences of steps may also be performed according to alternative embodiments. For example, alternative embodiments may perform the steps outlined above in a different order. Furthermore, Figure 2 and Figure 8The individual steps shown may include multiple sub-steps, which can be adapted to various sequences of execution of the individual steps. Furthermore, additional steps may be added or removed depending on the specific application. Many variations, modifications, and alternatives also fall within the scope of this disclosure.

[0082] Each of the methods described herein can be implemented by a computer system. Each step of these methods can be performed automatically by a computer system and / or can be provided using inputs / outputs involving a user. For example, a user can provide inputs for each step of the method, and each of these inputs can provide a specific output in response to a request for such input, wherein the output is generated by the computer system. Each input can be received in response to a corresponding requested output. Furthermore, inputs can be received from a user, received as a data stream from another computer system, retrieved from a memory location, retrieved on a network, requested from a network service, and / or the like. Similarly, outputs can be provided to a user, provided to another computer system as a data stream, stored in a memory location, sent on a network, provided to a network service, and / or the like. In short, each step of the methods described herein can be performed by a computer system and can involve any number of inputs, outputs, and / or requests to and from the computer system, which may or may not involve a user. Those steps that do not involve a user can be considered to be performed automatically by the computer system without human intervention. Therefore, it will be understood that, in view of this disclosure, each step of each method described herein may be modified to include inputs and outputs to and from the user, or may be performed automatically by a computer system without human intervention, wherein any determination is made by a processor. Furthermore, some embodiments of each of the methods described herein may be implemented as a set of instructions stored on a tangible, non-transitory storage medium to form a tangible software product.

[0083] Figure 10 An exemplary computer system 1000 in which various embodiments may be implemented is shown. System 1000 can be used to implement any of the computer systems described above. As shown, computer system 1000 includes a processing unit 1004 that communicates with several peripheral subsystems via a bus subsystem 1002. These peripheral subsystems may include a processing acceleration unit 1006, an I / O subsystem 1008, a storage subsystem 1018, and a communication subsystem 1024. Storage subsystem 1018 includes a tangible computer-readable storage medium 1022 and system memory 1010.

[0084] Bus subsystem 1002 provides a mechanism for enabling the various components and subsystems of computer system 1000 to communicate with each other as intended. Although bus subsystem 1002 is schematically illustrated as a single bus, alternative embodiments of the bus subsystem may utilize multiple buses. Bus subsystem 1002 may be any of several types of bus architectures, including memory buses or memory controllers, peripheral buses, or local buses using any of various bus architectures. For example, such architectures may include Industry Standard Architecture (ISA) buses, Micro Channel Architecture (MCA) buses, Enhanced ISA (EISA) buses, Video Electronics Standards Association (VESA) local buses, and Peripheral Component Interconnect (PCI) buses, which may be implemented as Mezzanine buses manufactured according to the IEEE P1386.1 standard.

[0085] A processing unit 1004, which may be implemented as one or more integrated circuits (e.g., conventional microprocessors or microcontrollers), controls the operation of a computer system 1000. One or more processors may be included in the processing unit 1004. These processors may include single-core or multi-core processors. In some embodiments, the processing unit 1004 may be implemented as one or more independent or sub-processing units 1032 and / or 1034, wherein each processing unit includes a single-core or multi-core processor. In other embodiments, the processing unit 1004 may also be implemented as a quad-core processing unit formed by integrating two dual-core processors into a single chip.

[0086] In various embodiments, processing unit 1004 can execute various programs in response to program code and can maintain multiple concurrently executing programs or processes. At any given time, some or all of the program code to be executed can reside in processor 1004 and / or storage subsystem 1018. Through suitable programming, processor 1004 can provide the various functionalities described above. Computer system 1000 may additionally include processing acceleration unit 1006, which may include digital signal processor (DSP), dedicated processor, and / or the like.

[0087] The I / O subsystem 1008 may include user interface input devices and user interface output devices. User interface input devices may include keyboards, pointing devices (such as mice or trackballs), touchpads or touchscreens integrated into displays, scroll wheels, click wheels, dial pads, buttons, switches, keypads, audio input devices with voice command recognition systems, microphones, and other types of input devices. Additionally, user interface input devices may include voice identification sensing devices that enable users to interact with a voice identification system via voice commands. User interface input devices may also include, but are not limited to, three-dimensional (3D) mice, joysticks or pointing sticks, gaming keyboards and drawing tablets, and audio / visual devices such as speakers, digital cameras, digital portable cameras, portable media players, webcams, image scanners, fingerprint scanners, barcode readers, 3D scanners, 3D printers, laser rangefinders, and eye-tracking devices. Additionally, user interface input devices may include, for example, medical imaging input devices such as computational tomography, magnetic resonance imaging, positional emission tomography, and medical ultrasound examination devices. User interface input devices may also include, for example, audio input devices such as MIDI keyboards, digital musical instruments, augmented reality (AR) input / output devices, virtual reality (VR) input / output devices, and / or the like.

[0088] User interface output devices may include display subsystems, indicator lights, or non-visual displays such as audio output devices. Display subsystems may be cathode ray tubes (CRTs), flat panel devices (such as those using liquid crystal displays (LCDs) or plasma displays), projection devices, touch screens, and the like. Generally, the term "output device" is intended to include all possible types of means and mechanisms for outputting information from computer system 1100 to a user or other computer. For example, user interface output devices may include, but are not limited to, various display devices that visually convey text, graphics, and audio / video information, such as monitors, printers, speakers, headsets, automatic navigation systems, plotters, voice output devices, and modems.

[0089] Computer system 1000 may include a storage subsystem 1018, which includes software elements, denoted as those currently located within system memory 1010. System memory 1010 may store program instructions that are loadable and executable on processing unit 1004, as well as data generated during the execution of these programs.

[0090] Depending on the configuration and type of the computer system 1000, the system memory 1010 may be volatile (such as random access memory (RAM)) and / or non-volatile (such as read-only memory (ROM), flash memory, etc.). RAM typically contains data and / or program modules that can be immediately accessed by the processing unit 1004 and / or are currently being operated and executed by the processing unit. In some embodiments, the system memory 1010 may include various different types of memory, such as static random access memory (SRAM) or dynamic random access memory (DRAM). In some embodiments, a basic input / output system (BIOS) containing basic routines, such as those that facilitate the transfer of information between elements within the computer system 1000 during startup, may typically be stored in ROM. For example, and without limitation, system memory 1010 also includes application 1012 (which may include client applications, web browsers, middleware applications, relational database management systems (RDBMS), etc.), program data 1014, and operating system 1016. For example, operating system 1016 may include various versions of Microsoft Windows®, Apple Macintosh®, and / or Linux operating systems, various commercially available UNIX® or similar UNIX operating systems (including but not limited to various GNU / Linux operating systems, Google Chrome® OS, and the like) and / or mobile operating systems such as iOS, Windows® Phone, Android® OS, BlackBerry® 10 OS, and Palm® OS.

[0091] Storage subsystem 1018 may also provide a tangible computer-readable storage medium for storing basic programming and data constructions that provide functionality according to some embodiments. Software (programs, code modules, instructions) that provides the functionality described above when executed by a processor may be stored in storage subsystem 1018. These software modules or instructions may be executed by processing unit 1004. Storage subsystem 1018 may also provide a repository for storing data used according to some embodiments.

[0092] Storage subsystem 1018 may also include computer-readable storage medium reader 1020, which may be further connected to computer-readable storage medium 1022. Together and optionally, in conjunction with system memory 1010, computer-readable storage medium 1022 may collectively represent remote, local, fixed and / or removable storage devices plus storage media for temporarily and / or more permanently containing, storing, transmitting and retrieving computer-readable information.

[0093] The computer-readable storage medium 1022 containing code or portions thereof may also include any suitable medium, including storage and communication media, such as, but not limited to, volatile and non-volatile, removable and non-removable media implemented in any method or technology for storing and / or transmitting information. This may include tangible computer-readable storage media such as RAM, ROM, electronically erasable programmable ROM (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile disk (DVD) or other optical storage, magnetic tape, magnetic tape, disk storage or other magnetic storage devices, or other tangible computer-readable media. This may also include non-tangible computer-readable media, such as data signals, data transmissions, or any other medium that can be used to transmit desired information and can be accessed by the computing system 1000.

[0094] For example, computer-readable storage medium 1022 may include a hard disk drive that reads or writes to a non-removable non-volatile magnetic medium, a disk drive that reads or writes to a removable non-volatile disk, and an optical disc drive that reads or writes to a removable non-volatile optical disc (such as a CD-ROM, DVD, and Blu-ray® disc, or other optical media). Computer-readable storage medium 1022 may include, but is not limited to, Zip® drives, flash memory cards, universal serial bus (USB) flash drives, secure digital (SD) cards, DVD discs, digital video tapes, and the like. Computer-readable storage medium 1022 may also include: solid-state drives (SSDs) based on non-volatile memory, such as flash-based SSDs, enterprise flash drives, solid-state ROMs, and the like; SSDs based on volatile memory, such as solid-state RAM, dynamic RAM, static RAM, DRAM-based SSDs, magnetoresistive RAM (MRAM) SSDs; and hybrid SSDs using a combination of DRAM and flash-based SSDs. The disk drive and its associated computer-readable medium provide non-volatile memory for computer-readable instructions, data structures, program modules, and other data for the computer system 1000.

[0095] The communication subsystem 1024 provides an interface to other computer systems and networks. The communication subsystem 1024 serves as an interface for receiving data from and sending data to other systems from computer system 1000. For example, the communication subsystem 1024 may enable computer system 1000 to connect to one or more devices via the Internet. In some embodiments, the communication subsystem 1024 may include radio frequency (RF) transceiver components (e.g., using cellular phone technology, advanced data network technologies such as 3G, 4G, or EDGE (Global Evolution Enhanced Data Rate)), WiFi (IEEE 802.11 family of standards or other mobile communication technologies or any combination thereof), global positioning system (GPS) receiver components, and / or other components for accessing wireless voice and / or data networks. In some embodiments, in addition to or instead of a wireless interface, the communication subsystem 1024 may provide wired network connectivity (e.g., Ethernet).

[0096] In some embodiments, the communication subsystem 1024 may also receive input communications in the form of structured and / or unstructured data feeds 1026, event streams 1028, event updates 1030, and the like on behalf of one or more users who may use the computer system 1000.

[0097] For example, the communication subsystem 1024 may be configured to receive data feeds 1026 or network feeds (such as Rich Site Summary (RSS) feeds) and / or real-time updates from one or more third-party information sources from users of social networks and / or other communication services in real time.

[0098] Furthermore, the communication subsystem 1024 can also be configured to receive data in the form of continuous data streams, which may include event streams 1028 of real-time events and / or event updates 1030 that may be continuous or boundless without a definite end. Examples of applications that generate continuous data may include, for example, sensor data applications, financial instruments, network performance measurement tools (e.g., network monitoring and traffic management applications), point-and-click flow analysis tools, vehicle traffic monitoring, and the like.

[0099] The communication subsystem 1024 can also be configured to output structured and / or unstructured data feeds 1026, event streams 1028, event updates 1030, and the like to one or more databases that can communicate with one or more streaming data source computers coupled to the computer system 1000.

[0100] The computer system 1000 can be one of a variety of types, including handheld portable devices (e.g., iPhone® cellular phones, iPad® computing tablets, PDAs), wearable devices, PCs, workstations, mainframes, public information kiosks, server racks, or any other data processing system.

[0101] Due to the constantly evolving nature of computers and networks, the description of the computer system 1000 depicted in the figures is intended only as a concrete example. Many other configurations with more or fewer components than the system depicted in the figures are possible. For example, custom hardware may also be used and / or specific elements may be implemented in hardware, firmware, software (including small applications), or a combination thereof. Additionally, connections to other computing devices, such as network input / output devices, may be employed. Other ways and / or methods of implementing the various embodiments should be apparent based on the disclosure and teachings provided herein.

[0102] As used herein, the terms “about,” “approximately,” or “substantially” may be interpreted as within the scope that a person skilled in the art would expect in light of the specification.

[0103] In the foregoing description, numerous specific details have been set forth for purposes of explanation in order to provide a thorough understanding of the various embodiments. However, it will be apparent, however, that some embodiments may be practiced without some of these specific details. In other examples, well-known structures and devices are illustrated in block diagram form.

[0104] The foregoing description provides exemplary embodiments only and is not intended to limit the scope, applicability, or configuration of this disclosure. Rather, the foregoing description of various embodiments will provide a disclosure that can be implemented for carrying out at least one embodiment. It should be understood that various changes may be made to the function and arrangement of elements without departing from the spirit and scope of some embodiments set forth in the appended claims.

[0105] Specific details are set forth in the foregoing description to provide a thorough understanding of this disclosure. However, it will be understood that embodiments may be practiced without these specific details. For example, circuits, systems, networks, processes, and other components may be shown as components in the form of block diagrams so as not to obscure the embodiments with unnecessary details. In other examples, well-known circuits, processes, algorithms, structures, and techniques may be illustrated without unnecessary details in order to avoid obscuring the embodiments.

[0106] Furthermore, it should be noted that individual embodiments may have been described as processes depicted as flowcharts, flow diagrams, data flow diagrams, structural diagrams, or block diagrams. While flowcharts may describe operations as a sequential process, many operations may be performed in parallel or concurrently. Additionally, the order of operations can be rearranged. A process terminates upon completion of its operations, but may also have additional steps not included in the figures. A process may correspond to a method, function, procedure, subroutine, subroutine, etc. When a process corresponds to a function, its termination may correspond to the function returning to the calling function or the main function.

[0107] The term "computer-readable medium" includes, but is not limited to, portable or fixed storage devices, optical storage devices, wireless channels, and various other media capable of storing, containing, or carrying instructions and / or data. A code segment or machine-executable instruction may represent a procedure, function, subroutine, program, routine, subroutine, module, software package, class, or any combination of instructions, data structures, or program statements. A code segment can be coupled to another code segment or hardware circuit by passing and / or receiving information, data, arguments, parameters, or memory contents. Information, arguments, parameters, data, etc., can be passed, forwarded, or transmitted via any suitable means, including memory sharing, message passing, token passing, network transmission, etc.

[0108] Furthermore, the embodiments may be implemented by hardware, software, firmware, middleware, microcode, hardware description languages, or any combination thereof. When implemented as software, firmware, middleware, or microcode, the program code or code segments used to perform the necessary tasks may be stored in a machine-readable medium. The processor may execute these necessary tasks.

[0109] In the foregoing specification, features are described with reference to specific embodiments thereof, but it should be understood that not all embodiments are limited thereto. Various features and aspects of some embodiments may be used individually or in combination. Furthermore, without departing from the broader spirit and scope of the specification, embodiments may be used in any number of environments and applications beyond those described herein. Therefore, the specification and drawings are to be regarded as illustrative rather than restrictive.

[0110] Additionally, for illustrative purposes, the methods are described in a specific order. It should be understood that in alternative embodiments, the methods may be performed in a different order than described. It should also be understood that the methods described above may be executed by hardware components, or may be embodied by a sequence of machine-executable instructions that can be used to cause a machine (such as a general-purpose or special-purpose processor, or logic circuitry programmed with instructions) to execute the methods. These machine-executable instructions may be stored on one or more machine-readable media (such as CD-ROMs or other types of optical discs, floppy diskettes, ROMs, RAMs, EPROMs, EEPROMs, magnetic or optical cards, flash memory, or other types of machine-readable media suitable for storing electronic instructions). Alternatively, the methods may be executed by a combination of hardware and software.

Claims

1. A method for aging a semiconductor processing chamber, the method comprising: Aging cycles are performed using semiconductor processing chambers; Determine the characteristics of the semiconductor processing chamber during the aging cycle; The characteristics are provided to a trained machine learning model, which is configured to receive the characteristics of the semiconductor processing chamber as input and provide an output related to the completion of aging of the semiconductor processing chamber. The output is received from the trained machine learning model based on the aforementioned characteristics; Based on the output, it is determined that the aging of the semiconductor processing chamber is not complete, wherein the output indicates the expected time for the completion of the aging of the semiconductor processing chamber, the expected duration of the aging of the semiconductor processing chamber, or the expected number of additional aging cycles for the completion of the aging of the semiconductor processing chamber. Additional aging cycles are performed using the semiconductor processing chamber; Determine additional characteristics of the semiconductor processing chamber during the additional aging cycle; The characteristics of the semiconductor processing chamber during the additional aging cycle are provided to the trained machine learning model; Receive additional output from the trained machine learning model; The completion of aging of the semiconductor processing chamber is determined based on the additional output; as well as Generate an indicator that marks the completion of aging in the semiconductor processing chamber.

2. The method of claim 1, wherein the method further comprises: Generate an indicator that identifies one or more of the following: the expected time for the completion of aging in the semiconductor processing chamber, the expected duration of aging in the semiconductor processing chamber, or the expected number of additional aging cycles for the completion of aging in the semiconductor processing chamber.

3. The method of claim 1, further comprising, after determining, based on the output, that the aging of the semiconductor processing chamber is incomplete: Generate an indicator that identifies the expected time for the completion of aging in the semiconductor processing chamber, the expected duration of aging in the semiconductor processing chamber, or the expected number of additional aging cycles for the completion of aging in the semiconductor processing chamber.

4. The method of claim 1, wherein the execution of the aging cycle comprises: Control the semiconductor processing chamber to establish etching conditions within the semiconductor processing chamber; as well as The semiconductor processing chamber is controlled to establish epitaxial growth conditions within the semiconductor processing chamber.

5. The method of claim 1, wherein the characteristic comprises one or more of the following: temperature within the semiconductor processing chamber during the aging cycle, thickness of the epitaxial layer formed within the semiconductor processing chamber, power delivered to a heater associated with the semiconductor processing chamber, power setting of the heater associated with the semiconductor processing chamber, growth rate within the semiconductor processing chamber during the aging cycle, etch rate within the semiconductor processing chamber during the aging cycle, optical conditions within the semiconductor processing chamber during the aging cycle, pressure within the semiconductor processing chamber during the aging cycle, gas composition within the semiconductor processing chamber during the aging cycle, flow rate entering or leaving the semiconductor processing chamber during the aging cycle, change in any of these, physical or structural parameters associated with the semiconductor processing chamber, or any combination thereof.

6. The method of claim 1, further comprising, before providing the feature to the machine learning model, training the machine learning model, wherein training the machine learning model comprises: Receive training characteristics from multiple executions of aging cycles performed through one or more semiconductor processing chambers; Training data is generated based on the training characteristics, wherein the training data is generated using marker information that identifies the completion of aging of the one or more semiconductor processing chambers; as well as A supervised learning algorithm is executed to train the machine learning model using the training data.

7. The method of claim 6, wherein: The one or more semiconductor processing chambers and the semiconductor processing chambers are the same chambers; as well as Determining the completion of aging of the semiconductor processing chamber includes comparing the characteristics of the semiconductor processing chamber with previously determined characteristics of the semiconductor processing chamber.

8. The method of claim 1, wherein the machine learning model compares the characteristics of the aging cycle with the corresponding characteristics of the immediately preceding aging cycle to assess the completion of aging in the semiconductor processing chamber.

9. A system comprising: One or more processors; as well as One or more non-transitory computer-readable storage devices include instructions that, when executed by the one or more processors, cause the one or more processors to perform operations including: This causes the semiconductor processing chamber to undergo an aging cycle; Determine the characteristics of the semiconductor processing chamber during the aging cycle; The characteristics are provided to a trained machine learning model, which is configured to receive the characteristics of the semiconductor processing chamber as input and provide an output related to the completion of aging of the semiconductor processing chamber. Receive the output from the trained machine learning model; Based on the output, it is determined that the aging of the semiconductor processing chamber is not complete, wherein the output indicates the expected time for the completion of the aging of the semiconductor processing chamber, the expected duration of the aging of the semiconductor processing chamber, or the expected number of additional aging cycles for the completion of the aging of the semiconductor processing chamber. This causes the semiconductor processing chamber to perform an additional aging cycle; Continue to determine the characteristics of the semiconductor processing chamber during the additional aging cycle; The characteristics of the semiconductor processing chamber during the additional aging cycle are provided to the trained machine learning model; Receive additional output from the trained machine learning model; The completion of aging of the semiconductor processing chamber is determined based on the additional output; as well as Generate an indicator that marks the completion of aging in the semiconductor processing chamber.

10. The system of claim 9, further comprising one or more sensors that communicate with the one or more processors to determine the characteristics of the semiconductor processing chamber during an aging cycle.

11. The system of claim 10, wherein: The characteristic includes a time series of data obtained from the one or more sensors; or The feature includes a comparison of a first time series of data obtained from the one or more sensors during a first aging cycle with a second time series of data obtained from the one or more sensors during a second aging cycle immediately preceding the first aging cycle.

12. The system of claim 9, wherein the aging cycle comprises: Control the semiconductor processing chamber to establish etching conditions within the semiconductor processing chamber; as well as The semiconductor processing chamber is controlled to establish epitaxial growth conditions within the semiconductor processing chamber.

13. The system of claim 9, wherein the semiconductor processing chamber comprises one or more of a quartz dome, a flow module, or a base.

14. The system of claim 13, further comprising one or more sensors positioned to measure the characteristics of the surface of one or more of the quartz dome, the flow module, the base, or the substrate within the semiconductor processing chamber.

15. A non-transitory computer-readable medium comprising instructions that, when executed by one or more processors, cause the one or more processors to perform operations including: This causes the semiconductor processing chamber to undergo an aging cycle; Determine the characteristics of the semiconductor processing chamber during the aging cycle; The characteristics are provided to a trained machine learning model, which is configured to receive the characteristics of the semiconductor processing chamber as input and provide an output related to the completion of aging of the semiconductor processing chamber. Receive the output from the trained machine learning model; Based on the output, it is determined that the aging of the semiconductor processing chamber is not complete, wherein the output indicates the expected time for the completion of the aging of the semiconductor processing chamber, the expected duration of the aging of the semiconductor processing chamber, or the expected number of additional aging cycles for the completion of the aging of the semiconductor processing chamber. This causes the semiconductor processing chamber to perform an additional aging cycle; Continue to determine the characteristics of the semiconductor processing chamber during the additional aging cycle; The characteristics of the semiconductor processing chamber during the additional aging cycle are provided to the trained machine learning model; Receive additional output from the trained machine learning model; The completion of aging of the semiconductor processing chamber is determined based on the additional output; as well as Generate an indicator that marks the completion of aging in the semiconductor processing chamber.

16. The non-transitory computer-readable medium of claim 15, wherein the trained machine learning model is trained to model the aging of the semiconductor processing chamber following a chamber opening event or preventative maintenance event associated with the semiconductor processing chamber.

17. The non-transitory computer-readable medium of claim 16, wherein the operation further comprises obtaining user input corresponding to one or more of the following: an identifier of the duration of the chamber opening event or the preventive maintenance event, a change of a component of the semiconductor processing chamber during the chamber opening event or the preventive maintenance event, environmental conditions during the chamber opening event or the preventive maintenance event, a maintenance procedure or protocol applied to the semiconductor processing chamber during the chamber opening event or the preventive maintenance event, or a severity rating for the chamber opening event or the preventive maintenance event, wherein the trained machine learning model is further configured to use the user input when generating the output relating to the completion of aging of the semiconductor processing chamber.

18. The non-transitory computer-readable medium of claim 15, wherein the output from the trained machine learning model includes an indication of the completion or incomplete aging of the semiconductor processing chamber and one or more variance values ​​determined by comparing the characteristics of the semiconductor processing chamber during a first aging cycle with the characteristics of the semiconductor processing chamber during a second aging cycle immediately preceding the first aging cycle.

19. The non-transitory computer-readable medium of claim 15, wherein determining the completion of aging of the semiconductor processing chamber based on the additional output comprises: Derive the variance between the output and the additional output; as well as Determine that the variance value is less than the target variance value.

20. The non-transitory computer-readable medium of claim 19, wherein the operation further includes receiving an input corresponding to a tolerance of the trained machine learning model, the tolerance indicating an allowable deviation from the target variance value.

Images