Machine and deep learning techniques for predicting ecological efficiency in substrate processing

Abstract

In some embodiments, a method includes receiving a process recipe, the process recipe including process recipe setpoint data. The method further includes inputting the process recipe into one or more trained machine learning models, the models outputting predicted environmental resource usage data indicative of environmental resource consumption associated with processing a substrate in a processing chamber according to the process recipe. The method further includes outputting a suggestion related to the process recipe based at least in part on the predicted environmental resource usage data.

Description

Technical Field

[0001] This specification generally relates to the environmental impact of manufacturing equipment, such as semiconductor manufacturing equipment. More specifically, this specification relates to machine and deep learning techniques for predicting eco-efficiency in substrate processing. Background Art

[0002] The continuous demand for electronic components has led to an increasing demand for semiconductor wafers. The growth of manufacturing to produce such wafers has caused serious damage to the environment in the form of resource utilization and generation of environmentally harmful waste. Therefore, the demand for more environmentally friendly and environmentally responsible wafer manufacturing methods and general manufacturing methods has increased. Given that wafer processing is energy-intensive, it is valuable to decouple the growth of the semiconductor industry from its environmental impact. The growing demand for chips and increasing chip complexity have increased resource consumption that has an impact on the environment. In addition, the increase in chip complexity has increased the difficulty of determining an eco-efficient substrate process recipe. Summary of the Invention

[0003] The following is a brief summary of the present disclosure to provide a basic understanding of certain aspects of the present disclosure. This summary is not an extensive review of the present disclosure. This summary is not intended to delineate any scope of any particular embodiment of the present disclosure or any scope of the claims. Its sole purpose is to present some concepts of the present disclosure in a simplified form as a prelude to the more detailed description that follows.

[0004] In some embodiments, a method includes receiving a process recipe that includes process recipe set point data. The method further includes inputting the process recipe into one or more trained machine learning models that output predicted environmental resource usage data that indicates environmental resource consumption associated with processing a substrate in a processing chamber according to the process recipe. The method further includes outputting a recommendation associated with the process recipe at least in part based on the predicted environmental resource usage data.

[0005] In some embodiments, a system includes one or more processing chambers configured to process substrates. One or more of the chambers include a plurality of sensors. The system further includes a system controller that controls the one or more processing chambers. The system controller will receive a process recipe that includes process recipe set point data. The system controller further inputs the process recipe into one or more trained machine learning models that output predicted environmental resource usage data that indicates environmental resource consumption associated with processing a substrate in the processing chamber according to the process recipe. The system controller further outputs a recommendation associated with the process recipe at least in part based on the predicted environmental resource usage data.

[0006] In some embodiments, a non-transitory machine-readable storage medium includes instructions that, when executed by a processing device, cause the processing device to train a first machine learning model to form a trained first machine learning model. The trained first machine learning model is trained to output predicted measurement data based on a process recipe input into the trained first machine learning model. The processing device further uses training data to train a second machine learning model to form a trained second machine learning model, the training data including the predicted measurement data output from the trained first machine learning model. The trained second machine learning model is trained to output predicted environmental resource usage data that indicates environmental resource consumption associated with processing a substrate in a processing chamber according to a first process recipe input into the trained second machine learning model. BRIEF DESCRIPTION OF THE DRAWINGS

[0007] Aspects and embodiments of the present disclosure can be more fully understood from the following detailed description and the accompanying drawings, which are intended to illustrate aspects and embodiments by way of example and not limitation.

[0008] Figure 1 FIG. 9 is a top view schematic diagram of an exemplary manufacturing system according to one embodiment.

[0009] Figure 2A FIG. 13 is a block diagram of a logical view of an exemplary eco-efficiency platform according to one embodiment.

[0010] Figure 2B FIG. 17 is a simplified block diagram of a logical view of an exemplary eco-efficiency prediction platform according to some embodiments of the present disclosure.

[0011] Figure 3 FIG. 21 is a block diagram showing an exemplary system architecture in which embodiments of the present disclosure can operate.

[0012] Figure 4 FIG. 25 depicts an exemplary digital replica according to some embodiments of the present disclosure.

[0013] Figure 5 FIG. 29 is an exemplary illustration of a process parameter value window according to some embodiments of the present disclosure.

[0014] Figure 6 FIG. 33 is a flowchart of a method for generating a training data set for training a machine learning model according to an aspect of the present disclosure.

[0015] Figure 7 FIG. 37 shows a flowchart of a method for training a machine learning model to determine a predicted cooling parameter value according to an aspect of the present disclosure.

[0016] Figure 8AFlowchart of a method for obtaining recommendations for processing a substrate, according to some embodiments of the present disclosure.

[0017] Figure 8B Flowchart of a method for obtaining predicted process recipe setpoint data, according to some embodiments of the present disclosure.

[0018] Figure 8C Flowchart of a method for obtaining predicted environmental resource usage data, according to some embodiments of the present disclosure.

[0019] Figure 9A Shows a graph of predicted environmental resource consumption data relative to observed environmental resource consumption, according to some embodiments of the present disclosure.

[0020] Figure 9B Shows a graph of predicted or actual temporal environmental resource consumption data, according to some embodiments of the present disclosure.

[0021] Figure 10 Block diagram of an exemplary computing device operating in accordance with one or more aspects of the present disclosure. Detailed Description

[0022] Eco-efficiency characterization is a complex technique for determining different levels of inputs (e.g., resources, utilization rates, etc.) associated with a particular manufacturing tool during tool use. Eco-efficiency characterization is used to determine how changing inputs affect the eco-efficiency of a manufacturing tool. Eco-efficiency characterization and / or eco-efficiency prediction can be beneficial during the development of a manufacturing tool, helping to develop a manufacturing tool that maximizes eco-efficiency per unit (or per unit time) and minimizes harmful environmental impacts. After tool development, considering the specific parameters under which the tool operates, eco-efficiency characterization is also beneficial for tuning the per-unit eco-efficiency characteristics of the tool and / or the process recipe when the tool is operating.

[0023] The embodiments described herein provide a system for predicting and optimizing the eco-efficiency of a substrate process recipe during the design, development, and implementation of the process recipe. In some embodiments, the methods disclosed herein can assist engineers in developing, optimizing, and / or operating processes that meet materials engineering design and eco-efficiency specifications. In some embodiments, sensor data and / or models are utilized to provide predictions of the eco-efficiency of multiple manufacturing systems, individual process chambers of such manufacturing systems, and / or specific process recipes executed in the individual process chambers. Additionally, the methods described herein can optimize process recipes to improve eco-efficiency while maintaining the processed substrate target. For example, by using the output of a predictive model as an indicator of environmental resource usage, a process recipe can be selected and / or optimized to reduce the consumption of environmental resources while still meeting the set substrate target results. According to some embodiments described herein, the optimization and / or selection of a process recipe for improved eco-efficiency can be completed prior to the actual implementation of the process recipe. For example, by using historical data, models can be developed and utilized to determine the eco-efficiency of a process recipe being developed. In some embodiments, the eco-efficiency of several process recipes can be compared, and the process recipe with the greatest eco-efficiency that still meets the manufacturing goals can be selected. Thus, the eco-efficiency and / or environmental impact of a substrate process recipe can be predicted and / or improved without physical testing or empirical results.

[0024] In some embodiments, an eco-efficiency prediction platform (e.g., software of a system controller) can receive a process (e.g., process recipe set point data) and sensor data to form an eco-efficiency prediction. The process recipe and / or sensor data can be input into one or more models, such as one or more trained machine learning models, physics-based models (e.g., digital twins), and / or one or more additional models. In some embodiments, the process recipe is determined by a first model (e.g., a first predictive model, a trained machine learning model, etc.) based on the processed substrate target input into the model. For example, a user (e.g., an engineer, a technician, etc.) can input the target processing result (e.g., of the processed substrate) into the model. The first model can be trained to output possible process recipes for processing the substrate, where the output process recipes each meet the target processing result. Since there may be many ways to achieve the target result (e.g., many recipes can produce a substrate that meets the target), the first model can output multiple different process recipes, each of which meets the target result.

[0025] Each output recipe output by the model can be input into a second model (e.g., a second prediction model, a trained second machine learning model, etc.), which is configured to predict an input process recipe related to eco - efficiency. The second model can output predicted eco - efficiency data corresponding to each process recipe. The predicted eco - efficiency values can include predicted environmental resource usage data indicating environmental resource consumption (e.g., consumption of chemicals, gases, electricity, water, etc.). As used herein, the environmental resource usage data can include data on the following: consumption of resources and / or chemicals, environmental impacts of the resources and / or chemicals that have been used / consumed, energy consumption, and / or environmental impacts of the consumed energy. In some instances, the predicted data includes time - series data indicating power consumption and / or gas flow related to the process recipe. Each predicted eco - efficiency data can be analyzed and / or compared to determine the most eco - efficient process recipe (e.g., the process recipe that consumes the least resources).

[0026] In some embodiments, based on the eco - efficiency data corresponding to the process recipe, a recommendation for processing the substrate is output. The recommendation can indicate that a specific process recipe will be implemented to process the substrate to meet the process objectives. In some embodiments, the recommendation can include modifications to one or more process recipes and / or one or more additional objectives and / or constraints of the process recipe to increase their respective eco - efficiencies. The recommendation can be input into the first model (to predict the process recipe), and the first model can output a further predicted or recommended process recipe. These further process recipes can be processed by the second model to determine the resource consumption and / or eco - efficiency values of the further process recipes. Given the eco - efficiency values associated with the further recipes, the recipes can be analyzed again to provide further recommendations. This process can be repeated so that the models working together can converge on the most eco - efficient process recipe for meeting the product objectives.

[0027] In some embodiments, eco - efficiency is calculated based on units. Generally, unit eco - efficiency is not considered during the manufacturing tool and / or process recipe development process. Additionally, when the tool is in use (e.g., when the tool is used for substrate production), characterizing unit eco - efficiency to adjust the settings of the manufacturing tool or process recipe can be a cumbersome and complex process. Moreover, in previous solutions, special eco - efficiency training was used for personnel, professional engineers, and analysts to perform eco - efficiency characterization analysis. Embodiments of the present disclosure provide improved methods, systems, and software for unit - based eco - efficiency characterization. These methods, systems, and software can be used by individuals who have not received special eco - efficiency training.

[0028] In some embodiments, eco-efficiency characterization and / or prediction can be performed through software tools at all stages of the manufacturing equipment lifecycle, including during the design time and the operating phase of the manufacturing equipment. Eco-efficiency can include the amount of environmental resources consumed per unit of equipment production (e.g., electrical energy, water, gas, chemicals, etc.), such as per wafer or per manufactured component. Eco-efficiency can also be characterized as the amount of environmental impact generated per unit of equipment production (e.g., CO2 emissions, heavy metal waste, etc.).

[0029] Eco-efficiency can be more precisely characterized per unit of analysis, where the unit is any measurable quantity (e.g., substrate, die, area (cm 2 )), time period, component, etc.) that the manufacturing tool operates on. Eco-efficiency based on the "unit" can allow for the accurate determination of the resource usage rate and environmental impact per unit of output and can be easily manipulated as a value metric. For example, it can be determined that a particular manufacturing tool has an electrical energy eco-efficiency rating of 1.0 to 2.0 kWh per substrate pass (in other embodiments, the eco-efficiency rating per substrate pass can be less than 0.5 kWh, up to 20 kWh, or even greater than 20 kWh), indicating that each substrate processed by the manufacturing tool can use, for example, 1.0 to 2.0 kWh of electrical energy per substrate processed. In other embodiments, various other amounts of electrical energy can be used. Determining eco-efficiency on a per-substrate-pass basis can allow for easy comparison with other manufacturing tools that have different annual electrical energy consumption values due to differences in the annual substrate throughput. In one embodiment, the eco-efficiency per component can also be determined by dividing the eco-efficiency characterization per substrate by the number of components per wafer.

[0030] Eco-efficiency characterization or calculation can be performed on the manufacturing equipment and / or process recipe during operation. The manufacturing equipment can access real-time variables, such as equipment utilization and utility usage data from a first sensor and a second sensor on the manufacturing equipment (the second sensors are external sensors and not components of the manufacturing equipment); and use the real-time variables in one or more eco-efficiency models. Given the current operating conditions of the manufacturing equipment, the manufacturing equipment can fine-tune the settings on the equipment to maximize eco-efficiency. Similarly, sensor data (e.g., from the first sensor and / or the second sensor) can be input into a model (e.g., a trained machine learning model, a deep learning model, etc.) together with process recipe data (e.g., process recipe setpoint data) for the model to predict eco-efficiency data corresponding to the process recipe. In some embodiments, the sensor data is input into the model to inform the model of physical constraints (e.g., imposing physical constraints on the model to form a physics-informed model, etc.).

[0031] In some embodiments, modifications to a manufacturing process (e.g., a process or a subset of processes, process recipe operations, etc.) can be determined based on environmental resource usage data or eco-efficiency characterization and / or prediction. For example, based on the predicted environmental resource usage data of multiple process recipes output from a model, one or more modifications to the process recipe parameters (e.g., setpoints) of a specific recipe can be determined. Modifications to the process recipe parameters can be related to improving the eco-efficiency of a selected set of manufacturing processes (e.g., reducing environmental resource consumption and / or environmental impact).

[0032] In some embodiments, eco-efficiency is based on resource consumption, such as energy consumption, chemical consumption (e.g., gases such as hydrogen, nitrogen, chemicals for etching or depositing thin films, and / or liquids that can be evaporated, atomized, or converted to a gaseous state through a bubbler, syringe, or nebulizer), clean dry air (CDA), and / or water consumption (e.g., process cooling water (PCW), de-ionized water (DIW), and ultrapure water (UPW)). However, in some embodiments, eco-efficiency is based on life cycle data of components associated with manufacturing equipment. For example, the environmental resource consumption and / or environmental impact associated with eco-efficiency characterization can be related to the replacement or maintenance procedures of consumable parts of manufacturing equipment. However, it should be understood that such embodiments are also applicable to the consumption of chemicals in other states, such as liquid chemicals. Any embodiments discussed herein regarding gas consumption are equally applicable to the consumption of other types of chemicals, such as liquids.

[0033] As described above, in some embodiments, an eco-efficiency prediction platform predicts the environmental resource usage of a processing chamber performing a manufacturing process according to a process recipe based on predicted process recipe data, sensor data, and / or substrate process targets. By predicting eco-efficiency based on process targets and / or sensor data (e.g., sensor data used to inform the model), the predicted resource amount for a process in a processing chamber can be more accurately determined. In some embodiments, the improved accuracy of an eco-efficiency prediction platform using such data can lead to better process development and lower overall resource consumption.

[0034] Figure 1FIG. 100 is a top view schematic of an exemplary processing system 100 (also referred to herein as a manufacturing system) according to one embodiment. In some embodiments, the processing system 100 may be an electronic processing system configured to perform one or more processes on a substrate 102. In some embodiments, the processing system 100 may be an electronic component manufacturing system. The substrate 102 may be any suitable rigid, fixed-size planar article, such as a silicon-containing wafer or disk, a patterned wafer, a glass plate, or the like, suitable for manufacturing electronic components or circuit parts thereon. In some embodiments, the processing system 100 is a semiconductor processing system. Alternatively, the processing system 100 may be configured to process other types of devices, such as display devices.

[0035] The processing system 100 includes a processing tool 104 (e.g., a host) and a factory interface 106 coupled to the processing tool 104. The processing tool 104 includes a housing 108 having a transfer chamber 110 therein. The transfer chamber 110 includes one or more processing chambers (also referred to as processing chambers) 114, 116, 118, which are disposed around and coupled to the transfer chamber. The processing chambers 114, 116, 118 may be coupled to the transfer chamber 110 through respective ports (such as slit valves, etc.). The processing chambers 114, 116, 118 may be configured to process substrates.

[0036] The processing chambers 114, 116, 118 may be adapted to perform any number of processes on the substrate 102. The same or different substrate processing may be performed in each of the processing chambers 114, 116, 118. Examples of substrate processes include atomic layer deposition (ALD), physical vapor deposition (PVD), chemical vapor deposition (CVD), etching, annealing, curing, pre-cleaning, metal or metal oxide removal, or similar processes. In one example, a PVD process is performed in one or two of the processing chambers 114, an etching process is performed in one or two of the processing chambers 116, and an annealing process is performed in one or two of the processing chambers 118. Other processes may be performed on the substrates in the processing chambers. The processing chambers 114, 116, 118 may each include a substrate support assembly. The substrate support assembly may be configured to hold the substrate in place during the performance of the substrate process.

[0037] The transfer chamber 110 further includes a transfer chamber robot 112. The transfer chamber robot 112 may include one or more arms, where each arm includes one or more end effectors located at the end of the arm. The end effectors may be configured to transfer a particular object, such as a wafer. In some embodiments, the transfer chamber robot 112 is a selective compliance assembly robot arm (SCARA) robot, such as a 2-link SCARA robot, a 3-link SCARA robot, a 4-link SCARA robot, etc.

[0038] The load lock 120 may also be coupled to the housing 108 and the transfer chamber 110. The load lock 120 may be configured to interface and couple to the transfer chamber 110 on one side and to the factory interface 106 on the other side. In some embodiments, the load lock 120 may have an environmentally controlled atmosphere that changes from a vacuum environment (where substrates are transferred in and out of the transfer chamber 110) to an atmospheric pressure inert gas environment or a near-atmospheric pressure inert gas environment (where substrates are transferred in and out of the factory interface 106). In some embodiments, the load lock 120 is a stacked load lock having a pair of upper inner chambers and a pair of lower inner chambers located at different vertical levels (e.g., one above the other). In some embodiments, the pair of upper inner chambers are configured to receive processed substrates from the transfer chamber 110 for removal from the processing tool 104, while the pair of lower inner chambers are configured to receive substrates from the factory interface 106 for processing in the processing tool 104. In some embodiments, the load lock 120 is configured to perform substrate processes (e.g., etching or pre-cleaning) on one or more substrates 102 contained therein.

[0039] The factory interface 106 can be any suitable housing, such as an Equipment Front End Module (EFEM). The factory interface 106 can be configured to receive a substrate 102 from a substrate carrier 122 (e.g., a Front Opening Unified Pod (FOUP)) docked at respective load ports 124 of the factory interface 106. A factory interface robot 126 (shown in dashed lines) can be configured to transfer the substrate 102 between the substrate carrier 122 (also referred to as a container) and the load lock 120. In other and / or similar embodiments, the factory interface 106 is configured to receive replacement parts from a replacement part storage container 123. The factory interface robot 126 can include one or more robotic arms and can be or include a SCARA robot. In some embodiments, the factory interface robot 126 has more links and / or more degrees of freedom than the transfer chamber robot 112. The factory interface robot 126 can include end effectors located at the end of each robotic arm. The end effectors can be configured to pick up and transfer specific objects, such as wafers. Alternatively or additionally, the end effectors can be configured to transfer objects such as process kit rings.

[0040] Any known type of robot can be used for the factory interface robot 126. The transfer can be performed in any order or direction. In some embodiments, the factory interface 106 can be maintained in a non-reactive gas environment, such as a slightly positive pressure (by using, for example, nitrogen as the non-reactive gas).

[0041] The processing system 100 may also include a system controller 128. The system controller 128 may be and / or include a computing device, such as a personal computer, a server computer, a programmable logic controller (PLC), a microcontroller, etc. The system controller 128 may include one or more processing devices, which may be general-purpose processing devices, such as a microprocessor, a central processing unit, or the like. More specifically, the processing device may be a complex instruction set computing (CISC) microprocessor, a reduced instruction set computing (RISC) microprocessor, a very long instruction word (VLIW) microprocessor, or a processor implementing other instruction sets or a combination of instruction sets. The processing device may also be one or more dedicated processing devices, such as an application specific integrated circuit (ASIC), a field programmable gate array (FPGA), a digital signal processor (DSP), a network processor, etc. The system controller 128 may include data storage devices (e.g., one or more disk drives and / or solid state drives), main memory, static memory, a network interface, and / or other components. The system controller 128 may execute instructions to perform any one or more of the methods and / or embodiments described herein. The instructions may be stored on a computer-readable storage medium, which may include main memory, static memory, auxiliary storage devices, and / or the processing device (during instruction execution). In an embodiment, the system controller 128 executes instructions such that the system controller performs Figure 8A and 8B the method. The system controller 128 may also be configured to allow a human operator to input and display data, operation commands, etc.

[0042] In some embodiments, system controller 128 includes an eco-efficiency module 129, which can be an area server executing on the system controller 128 of the processing system 100. The eco-efficiency module 129 can be responsible for processing first sensor data generated by sensors of one or more of the processing chambers 114, 116, 118 and second sensor data from additional sensors 140, 142, 144 external to the processing chambers 114, 116, 118. The first sensor data can be generated by sensors integrated with the processing chambers 114, 116, 118. Such sensors can include, for example, temperature sensors, power sensors, current sensors, pressure sensors, concentration sensors, etc. The first sensor data output by the integrated sensors of the processing chambers 114, 116, 118 can include measurements of current, voltage, power, flow rate (e.g., the flow rate of one or more gases, CDA, water, etc.), pressure, concentration (e.g., the concentration of one or more gases), velocity (e.g., the velocity of one or more moving parts, gases, etc.), acceleration (e.g., the acceleration of one or more moving parts, gases, etc.), or temperature (e.g., the temperature of the substrate being processed, different locations in the processing chamber, etc.). In one embodiment, each chamber includes between about 20 and about 100 sensors.

[0043] To obtain additional data that is generally not obtainable by the integrated sensors of the processing chambers 114, 116, 118, one or more external sensors 140, 142, 144, 152 can be attached to the processing chambers 114, 116, 118, and / or attached to feed into and / or out of the processing chambers 114, 116, 118, and / or attached to sub-components (e.g., such as pumps and / or abatement systems) that operate to achieve the benefits of the processing chambers 114, 116, 118. In one embodiment, each processing chamber includes about 3 to 6 external sensors attached to the processing chamber, subsystems associated with the processing chamber, and / or inputs / outputs to / from the processing chamber. The second sensor data output by the external sensors 140, 142, 144, 152 can include, for example, current, flow rate, temperature, eddy current, concentration, vibration, voltage, or power factor. Examples of external sensors 140, 142, 144, 152 that can be used include chuck sensors (also known as current chucks) that measure alternating current or direct current, chuck sensors that measure voltage, and chuck sensors that measure leakage current. Other examples of external sensors are vibration sensors, temperature sensors, ultrasonic sensors (e.g., ultrasonic flow sensors), accelerometers (i.e., acceleration sensors), etc.

[0044] In the illustrated example, the abatement system 130, the gas delivery system 134, the water system 132, and / or the CDA system 136 may provide environmental resources to the processing chambers 114, 116, 118, and / or other components of the processing system 100 (e.g., to the transfer chamber, the factory interface, the load lock, etc.). In an embodiment, the abatement system 130 performs abatement of residual gases, reactants, and / or outputs related to the processes performed on the processing chambers 114, 116, 118. For example, the abatement system 130 may combust the residual gases and / or reactants to ensure that the gases and / or reactants do not pose an environmental risk. Additionally, in an embodiment, one or more pumps may be attached to and / or operate on behalf of one or more of the processing chambers 114, 116, 118. For clarity, the external sensors 140, 142, 144, 152 are shown simplified with respect to a single processing chamber 116. However, it should be understood that similar external sensors may be attached to additional processing chambers, and / or to the lines entering and exiting such additional processing chambers, and / or to subsystems associated with such additional processing chambers.

[0045] In certain embodiments, the external sensors 140, 142, 144, 152 may be Internet of Things sensors. In some embodiments, the external sensors include a power source such as a battery. In some embodiments, the external sensors are wired sensors plugged into a power source such as an AC power outlet. In some embodiments, the external sensors do not include a power source but receive sufficient power to operate based on environmental conditions. For example, sensors that detect voltage, power, and / or current may be wirelessly powered by such power or current (e.g., obtain energy through the current flowing through a wire, and the sensor is clamped onto the wire).

[0046] In one embodiment, the external sensors 140, 142, 144, 152 are sensors with an embedded system. An embedded system is a class of computing device that is embedded as a component of one device into another device. The external sensors 140, 142, 144, 152 typically also include other hardware, electrical, and / or mechanical components that may interface with the embedded system. The embedded system is typically configured to handle a specific task or set of tasks, and for this purpose, the embedded system may be optimized (e.g., to generate and / or send measurements). Thus, compared to a general-purpose computing device, the embedded system may have minimal cost and size.

[0047] The embedded systems may each include a communication module (not shown) that enables the embedded systems (and thus the external sensors 140, 142, 144, 152) to connect to a LAN, a hub 150, and / or a wireless carrier network (e.g., a network implemented by using various data processing devices, communication towers, etc.). The communication module may be configured to manage security, manage communication sessions, manage access control, manage communication with external devices, etc.

[0048] In one embodiment, the communication modules of the external sensors 140, 142, 144, 152 are configured to communicate by using Wi-Fi®. Alternatively, the communication module may be configured to communicate by using Bluetooth®, Zigbee®, Internet Protocol version 6 over Low-Power Wireless Personal Area Networks (6LowPAN), power line communication (PLC), Ethernet (e.g., 10 megabytes (Mb), 100 megabytes, and / or 1 gigabyte (Gb) Ethernet), or other communication protocols. If the communication module is configured to communicate with a wireless carrier network, the communication module may communicate by using Global Systems for Mobile Communication (GSM), Code-Division Multiple Access (CDMA), Universal Mobile Telecommunication System (UMTS), 3GPP Long Term Evaluation (LTE), Worldwide Interoperability for Microwave Access (WiMAX), or any other second-generation wireless telephone technology (2G), third-generation wireless telephone technology (3G), fourth-generation wireless telephone technology (4G), or other wireless telephone technologies.

[0049] In one embodiment, the communication module is configured to communicate with hub 150. For example, hub 150 can be a Wi-Fi router or other types of routers, switches, or hubs. Hub 150 can be configured to communicate with the communication modules of each of the external sensors 140, 142, 144, 152 and send the measurement values received from the external sensors 140, 142, 144, 152 to the system controller 128. In one embodiment, hub 150 has a wired connection (e.g., Ethernet connection, parallel connection, serial connection, Modbus connection, etc.) to the system controller 128 and sends the measurement values to the system controller 128 through the wired connection. In one embodiment, hub 150 is connected to one or more external sensors through a wired connection.

[0050] In some embodiments, hub 150 is connected to a network device, and the network device is connected to a local area network (LAN). The system controller 128 and the network device can each be connected to the LAN through a wireless connection and can be wirelessly connected to each other through the LAN. The external sensors 140, 142, 144, 152 may not support any communication types supported by the network device. For example, external sensor 140 can support Zigbee, while external sensor 142 can support Bluetooth. To enable such devices to connect to the LAN, hub 150 can act as a gateway device that connects to a network device (not shown) through one of the connection types supported by the network device (e.g., through Ethernet or Wi-Fi). The gateway device can additionally support other communication protocols, such as Zigbee, PLC, and / or Bluetooth, and can perform conversions between the supported communication protocols.

[0051] The system controller 128 can be connected to a wide area network (WAN). The WAN can be a private WAN (e.g., an intranet) or a public WAN such as the Internet, or can include a combination of private and public networks. In an embodiment, the system controller 128 can be connected to a LAN that can include routers and / or modems (e.g., cable modems, direct serial link (DSL) modems, Worldwide Interoperability for Microwave Access (WiMAX®) modems, long term evolution (LTE®) modems, etc.), and these modems provide a connection to the WAN.

[0052] A wide area network may include or be connected to one or more server computing devices (not shown). The server computing devices may include physical machines and / or virtual machines hosted by physical machines. The physical machines may be rack servers, desktop computers, or other computing devices. In one embodiment, the server computing devices include virtual machines managed and provided by a cloud provider system. Each virtual machine provided by the cloud service provider may be hosted on a physical machine configured as part of the cloud. Such physical machines are typically located in data centers. The cloud provider system and the cloud may be provided as an infrastructure as a service (IaaS) layer. An example of such a cloud is Amazon®'s Elastic Compute Cloud (EC2®).

[0053] The server computing devices may host one or more services, which may be network-based services and / or cloud services (e.g., network-based services hosted in a cloud computing platform). The services may maintain a communication session (e.g., via a continuous or intermittent connection) with the system controller 128 and / or system controllers of other manufacturing systems at the same location (e.g., in a manufacturing facility or a wafer fab) and / or at different locations. Alternatively, the service may periodically establish a communication session with the system controller. Through the communication session with the system controller 128, the service may receive status updates from the eco-efficiency module 129 running on the system controller 128. The service may aggregate data and may provide a graphical user interface (GUI) that can be accessed via any device connected to the wide area network (e.g., a mobile phone, a tablet computer, a portable computer, a desktop computer, etc.).

[0054] The eco-efficiency module 129 executing on the system controller 128 may process first sensor data from integrated sensors of one or more processing chambers 114, 116, 118 and second sensor data from external sensors 140, 142, 144, 152 to determine environmental resource usage data reflecting environmental resource consumption, such as water consumption, gas consumption, power consumption, etc. Operations that may be performed by the eco-efficiency module 129 are described with reference to the remaining figures below.

[0055] In some embodiments, the eco - efficiency module can predict the eco - efficiency of a process recipe running in one of the processing chambers 114, 116, 118. In some instances, by utilizing one or more machine - learning models, the eco - efficiency module 129 can predict the environmental resource consumption of a process operation. By using multiple process recipes as inputs, the eco - efficiency module 129 can determine the environmental resource consumption of each recipe. In some embodiments, the system controller can output a recommendation for substrate processing after comparing the environmental resource consumption associated with executing each recipe. The recommendation can be the recipe with the highest eco - efficiency to be executed. In some embodiments, the recommendation can include a modification to one of the process recipes to make the process recipe more eco - efficient. In some instances, the eco - efficiency module 129 can update the process recipe based on the modification included in the recommendation.

[0056] Figure 2A FIG. is a block diagram showing a logical view of an exemplary eco - efficiency platform 200A according to one embodiment. In an embodiment, the eco - efficiency platform 200A can be executed on the system controller 201. In one embodiment, the system controller 201 corresponds to Figure 1 the system controller 128, and the eco - efficiency platform 200A is provided by Figure 1 the eco - efficiency module 129.

[0057] The eco - efficiency platform 200 can receive first sensor data 270 from the tool sensor 202. In some embodiments, the tool sensor 202 can be Figure 1 the integrated sensor of the processing chambers 114, 116, 118. The eco - efficiency platform 200 can additionally receive second sensor data 272 from the hub 206, where the hub 206 receives the second sensor data from one or more external sensors 204. In some embodiments, the external sensors 204 can correspond to Figure 1 the external sensors 140, 142, 144, 152. In some embodiments, the hub 206 provides the second sensor data to the server 207, and the server 207 can be executed on one or more computing devices (e.g., in a cloud environment). The server 207 (e.g., an Internet of Things platform) can aggregate the second sensor data into aggregated second sensor data 274 and can send the aggregated second sensor data 274 to the eco - efficiency platform 200. Such aggregated second sensor data 274 can be provided to the eco - efficiency platform 200 in place of or together with the second sensor data 272.

[0058] In some embodiments, historical data 208 (e.g., historical sensor data) may be stored in a data storage such as a database. In some embodiments, such historical data 208 may additionally be provided to the eco-efficiency platform 200. In some embodiments, the historical data 208 may be used to train one or more machine learning models to predict eco-efficiency data, as described herein.

[0059] At block 230, the eco-efficiency platform 200 collects the first sensor data 270, the second sensor data 272, the aggregated second sensor data 274, and / or the historical data 208. At block 232, the eco-efficiency platform 200 may preprocess some or all of the received data. The preprocessing may include normalizing the data, changing the data units, adding timestamps to the data, synchronizing the data based on the timestamps, adding labels to the data, and the like.

[0060] At block 234, the eco-efficiency platform 200 performs data processing on the received data (e.g., the first sensor data 270 and the second sensor data 272). This may include inputting the data into one or more data processing algorithms or functions, inputting the data into one or more physics-based models (e.g., such as digital twins), inputting the data into one or more trained machine learning models, and the like. At block 236, outputs are generated by one or more models, data processing algorithms, functions, and the like. The outputs may include physical conditions and / or environmental resource usage data related to the manufacturing process performed on the processing chamber. The outputs may be stored in a regional data storage such as the database 210.

[0061] A client computing device that executes a network client 220 or other client application including a graphical user interface (GUI) 222 or other type of user interface may interface with the eco-efficiency platform 200. The network client 220 may send a request 212 to the eco-efficiency platform 200 and receive a response 214. The request 212 may include, for example, a request for environmental resource usage data for one or more processing chambers, a request for a manufacturing system including multiple processing chambers, a request for a recipe executed on a processing chamber, and the like. The requests may include requests to present the environmental resource usage data in charts, tables, and the like.

[0062] In some embodiments, the eco - efficiency platform 200 of multiple system controllers 201 interfaces with a remote computing device 250 (e.g., via a wide - area network). The remote computing device 250 may include a remote server that aggregates data from multiple eco - efficiency platforms and stores the aggregated data in a data storage such as a database 255. A network client 220 (or other client application) may interface with the remote server of the computing device 250 to access environmental resource usage data of multiple manufacturing systems in a fab, environmental resource usage data of multiple fabs, etc.

[0063] Figure 2B To simplify the block diagram, a logical view of an exemplary eco - efficiency prediction platform according to some embodiments of the present disclosure is shown. In some embodiments, the eco - efficiency prediction platform 200B may execute on a system controller (e.g., system controller 201). Alternatively, the eco - efficiency prediction platform 200B may execute on a server computer, which may execute in a cloud environment or may not execute in a cloud environment. In one embodiment, the eco - efficiency prediction platform 200B is provided by Figure 1 the eco - efficiency module 129.

[0064] The eco - efficiency prediction platform 200B may receive one or more process targets 278 (e.g., a set of process targets) from a substrate processing tool 268 or other sources. In one embodiment, a user (e.g., a technician) inputs one or more process targets through a user interface (e.g., a graphical user interface) of the eco - efficiency prediction platform 200B. In some embodiments, the one or more process targets 278 include target substrate processing data indicating the target substrate conditions of the processed substrate. For example, the process target 278 may indicate the target post - processing substrate results and / or target substrate specifications. In some instances, the process target 278 may indicate that the processed substrate will have one or more features (e.g., etched features, deposited features, coating features, film thickness, etc.). The substrate processing tool 268 may receive the process target through user input (e.g., through the GUI).

[0065] In some embodiments, a process target 278 is input into a process model 262. The process model 262 can be a model such as a physics-based model, a statistical model, a trained machine learning model (e.g., one or more trained machine learning models), or a hybrid model (e.g., a combination of one or more model types). For example, the process model 262 can be a physics-informed trained machine learning model. In some embodiments, the process model 262 is trained to output a plurality of process recipes (e.g., process recipes 280(1) to 280(n)) based on the input process target. In some embodiments, the process model 262 represents a substrate manufacturing process. The process model 262 can be trained using historical data 208, which includes historical process targets, historical process recipes, and / or historical eco-efficiency data. The process model 262 can be trained using training input data that includes historical target substrate process data corresponding to process targets for different substrate process operations. Historical target substrate process data can be collected over time when processing substrates and / or receiving new process targets. The process model 262 can be trained using training target data that includes historical process recipes (e.g., historical process recipe setpoint data) corresponding to the historical target substrate process data. For example, the process model 262 can be trained using process recipe data (e.g., training target output) and corresponding process target data (e.g., training input) to be achieved by the process recipe. Historical process recipes can be collected over time during and / or prior to performing new substrate process operations.

[0066] The output of the process model 262 is a plurality of process recipes 280(1) to 280(n). In some instances, the process model 262 outputs n process recipes. Each process recipe 280 can indicate setpoints (e.g., process recipe setpoints, control knob setpoints, etc.) for one or more process recipe operations. In some embodiments, each of the process recipes 280(1) to 280(n), when executed, produces a processed substrate that meets the process target 278 (e.g., meets the target specifications, etc.). Each of the process recipes 280(1) to 280(n) can have different process setpoints, such as different temperatures, gas flow rates, gas delivery times, pressures, etc. Additionally, each of the process recipes 280(1) to 280(n) can use different amounts of environmental resources, such as process gases and / or electricity. Thus, each of the process recipes 280(1) to 280(n) can have a different eco-efficiency.

[0067] In some embodiments, process recipes 280(1) to 280(n) are input into one or more chamber models 264. The chamber models 264 can receive and / or perform eco-efficiency predictions serially or in parallel based on the process recipes 280(1) to 280(n). In some embodiments, multiple chamber models 264 are used, where some chamber models receive process recipes and the outputs of other chamber models and generate outputs based on such inputs. In some embodiments, the multiple chamber models are "daisy-chained", where one or more first models in the chain can output predictions of readings and / or resource consumption that have a direct and easily understandable correlation with sensor readings and / or the recipe settings of the process recipe. The correspondence of subsequent models with the recipe settings and / or sensor readings may be less easy or less direct. However, there may be a correspondence between the first readings / resource consumption output by the first model and the readings / resource consumption output by the subsequent second model. The chamber models 264 can each be a model such as a physics-based model, a statistical model, a trained machine learning model (e.g., one or more trained machine learning models), or a hybrid model (e.g., a combination of one or more model types). For example, the chamber model 264 can be a physics-informed trained machine learning model.

[0068] The chamber model 264 can be a model representing a processing chamber. For example, the chamber model 264 can be a digital twin of the processing chamber. In some embodiments, the chamber model 264 is trained to output predicted environmental data (e.g., eco-efficiency data 282(1) to 282(n)) corresponding to the input process recipe. The chamber model 264 can be trained with historical data 208. The chamber model 264 can be trained with training input data including historical process recipe data collected over time. For example, the chamber model 264 can be trained using process recipe data corresponding to process recipe operations performed in the corresponding processing chamber. In some embodiments, the chamber model 264 includes two or more trained machine learning models. In some instances, a first machine learning model is trained by using historical process recipe data and historical eco-efficiency data. The first machine learning model can be trained to output predicted measurements (e.g., predicted sensor measurements such as temperature, power, flow rate, and / or other data related to environmental resource consumption) based on the input process recipe. The output (e.g., predicted measurements) from the first machine learning model, historical process recipes, and / or historical eco-efficiency data can be used to train a second machine learning model. The second machine learning model can be trained to output predicted eco-efficiency data based on the input process recipe.

[0069] In some embodiments, the chamber model 264 is trained using further training input data that includes sensor data related to substrate processing (e.g., historical sensor data) received from sensors of the corresponding processing chamber. In some embodiments, a second machine learning model of the chamber model 264 is trained using predicted sensor data output by a first machine learning model of the chamber model 264. The sensor data can include the first sensor data and / or the second sensor data as described herein. For example, the chamber model 264 can be trained based on data that includes measurements of: current, voltage, power, flow rate (e.g., the flow rate of one or more gases, CDA, water, etc.), pressure, concentration (e.g., the concentration of one or more gases), velocity (e.g., the velocity of one or more moving parts, gases, etc.), acceleration (e.g., the acceleration of one or more moving parts, gases, etc.), or temperature (e.g., the temperature of the substrate being processed, different locations in the processing chamber, etc.). In some embodiments, the sensor data is collected over time and stored in a database for subsequent training of the chamber model 264. In some embodiments, the sensor data is used to inform the chamber model 264. For example, by providing historical sensor data to the chamber model 264, the chamber model 264 can become a physically informed trained machine learning model. Informing the chamber model 264 can provide constraints to the chamber model 264, thus increasing the model accuracy.

[0070] In some embodiments, predicted eco-efficiency data 282 is output from the chamber model 264. The eco-efficiency data 282 can indicate the environmental resource usage (e.g., consumption) of the process recipe 280. In some instances, the eco-efficiency data 282 includes predicted time-series data (e.g., energy usage over time, gas consumption over time, etc.) that reflects the predicted behavior of the processing chamber. Figure 9BAn example of predicted timing resource consumption data is shown. In some examples, the eco-efficiency data 282 indicates predicted environmental resource consumption timing data associated with substrate processing over time, such as predicted power consumption, predicted gas consumption, predicted water consumption, etc. For each process recipe 280 input into the chamber model 264, a corresponding set of eco-efficiency data 282 is output. For example, based on the process recipe 280(1) input into the chamber model 264, the chamber model 264 outputs the eco-efficiency data 282(1). Similarly, based on the process recipe 280(n) input into the chamber model 264, the chamber model 264 outputs the eco-efficiency data 282(n). The chamber model 264 can output each set of eco-efficiency data 282(1) to 282(n) serially or in parallel. In some embodiments, the data analyzer 266 receives the eco-efficiency data 282. The data analyzer 266 can perform data analysis operations on the eco-efficiency data 282. In some examples, the data analyzer 266 can perform a comparison of each set of eco-efficiency data 282 to determine the corresponding process recipe 280 with the highest eco-efficiency.

[0071] In some embodiments, the data analyzer 266 outputs a recommendation 284 to the substrate processing tool 268. The recommendation 284 can be related to processing a substrate in a processing chamber according to one of the process recipes 280(1) to 280(n). For example, based on (e.g., in response to, etc.) the eco-efficiency data 282(2) indicating that the process recipe 280(2) is the process recipe with the highest eco-efficiency among the process recipes 280(1) to 280(n), the data analyzer 266 can recommend to the substrate processing tool 268 that the process recipe 280(2) should be implemented to process the substrate to meet the process goal 278. In some embodiments, the recommendation 284 includes a modification of one of the process recipes 280 and / or a modification of the process goal 278 to increase the eco-efficiency of the process recipe. In some embodiments, the modification includes one or more additional goals, one or more process recipe constraints for the process recipe (e.g., maximum temperature, minimum temperature, etc.). In one example, the data analyzer 266 can determine that, in order to increase the eco-efficiency of one of the process recipes 280, the previously predicted process recipe should be changed. The data analyzer 266 can indicate this change to the substrate processing tool 268 through the recommendation 284. In some embodiments, the modification of the process recipe is to form a modified process recipe. Compared with the unmodified process recipe, the modified process recipe can have reduced environmental resource consumption (e.g., higher eco-efficiency). In some embodiments, the data analyzer 266 utilizes one or more trained machine learning models that are trained to output the recommendation 284 based on the input eco-efficiency data 282.

[0072] In some embodiments, recommendation 284 is received by process model 262. Process model 262 may use recommendation 284 to predict more process recipes 280(1) to 280(n) that will have lower resource consumption and / or meet one or more updated process goals and / or newly added constraints.

[0073] In some examples, process model 262 is further trained using training inputs that include historical recommendations 284. Process model 262 may use recommendations 284 indicating modifications to process recipes to output more predicted process recipes 280(1) to 280(n) that are more eco-efficient than previously predicted process recipes. Thus, an iterative loop may be established. For each loop, process goals 278 may be updated based on recommendation 284, process model 262 may output updated process recipes based on the updated process goals, chamber model 264 may output updated eco-efficiency data based on the updated process recipes, data analyzer 266 may output updated recommendations based on the updated eco-efficiency data, and so on. In some examples, process model 262 may output more efficient process recipes 280 indicated by eco-efficiency data 282. Data analyzer 266 may then output recommendation 284 for further modifying the determined most eco-efficient process recipe and / or modifying process goals 278 to further improve the eco-efficiency of one or more process recipes. Substrate processing tool 268 may generate substrate processing in a processing chamber based on process goals 278 and / or recommendation 284. For example, substrate processing tool 268 may initialize substrate processing by using the process indicated by recommendation 284 to meet process goals 278.

[0074] Figure 3 A block diagram of an exemplary system architecture 300 in which embodiments of the present disclosure may operate. As Figure 3As shown, the system architecture 300 includes a manufacturing system 302, a data storage 312, a server 320, a client device 350, and / or a machine learning system 370. The machine learning system 370 can be a part of the server 320. In some embodiments, one or more components of the machine learning system 370 can be fully or partially integrated into the client device 350. The manufacturing system 302, the data storage 312, the server 320, the client device 350, and the machine learning system 370 can each be hosted by one or more computing devices, such as server computers, desktop computers, portable computers, tablet computers, notebook computers, personal digital assistants (PDAs), mobile communication devices, cellular phones, handheld computers, augmented reality (AR) displays and / or headsets, virtual reality (VR) displays and / or headsets, mixed reality (MR) displays and / or headsets, or similar computing devices. As used herein, a server can refer to a server, but can also include edge computing devices, in-room servers, the cloud, etc.

[0075] The manufacturing system 302, the data storage 312, the server 320, the client device 350, and the machine learning system 370 can be coupled to each other via a network (e.g., for performing the methods described herein). In some embodiments, the network 360 is a private network that provides access to each other and other privately available computing devices to each element of the system architecture 300. The network 360 can include one or more wide area networks (WANs), local area networks (LANs), wired networks (e.g., Ethernet networks), wireless networks (e.g., 802.11 networks or Wi-Fi networks), cellular networks (e.g., Long Term Evolution (LTE) networks), cloud networks, cloud services, routers, hubs, switches, server computers, and / or any combination of the foregoing. Optionally or additionally, any element of the system architecture 300 can be integrated or otherwise coupled together without using the network 360.

[0076] The client device 350 can be or include any personal computer (PC), portable computer, mobile phone, tablet computer, laptop computer, Internet-connected television (“smart TV”), Internet-connected media player (such as a Blu-ray player), set-top box, over-the-top (OTT) streaming device, carrier box, etc. The client device 350 can include a browser 352, an application 354, and / or other tools described and executed by other systems of the system architecture 300. In some embodiments, the client device 350 is capable of accessing the manufacturing system 302, the data storage 312, the server 320, and / or the machine learning system 370, and transmitting (e.g., sending and / or receiving) indications of predicted eco-efficiency, including one or more predicted environmental resource consumptions (e.g., environmental resource consumption) and / or predicted environmental impacts, and / or inputs and outputs of various processing tools (such as the component integration tool 322, the digital replica tool 324, the optimization tool 326, the recipe builder tool 328, the resource consumption tool 330, etc.) at various processing stages of the system architecture 300, as described herein.

[0077] As Figure 3 shown, the manufacturing system 302 includes machine equipment 304, a system controller 306, a process recipe 308, and sensors 310. The machine equipment 304 can be any combination of ion implanters, etch reactors (e.g., processing chambers), photolithography devices, deposition devices (e.g., for performing chemical vapor deposition (CVD), physical vapor deposition (PVD), ion-assisted deposition (IAD), etc.), or any other combination of manufacturing devices.

[0078] The process recipe 308, also referred to as a manufacturing recipe or manufacturing process instructions, includes the sequencing of machine operations in the process implementation, which, when applied in the specified order, produces a manufactured sample (e.g., a substrate having predetermined target characteristics or meeting predetermined target specifications). In some embodiments, the process recipe is stored in the data storage, or alternatively or additionally, it can be stored in a manner that generates a data sheet indicating the operations in the manufacturing process. Each operation can be associated with known environmental resource usage data. Alternatively or additionally, each process operation can be associated with parameters indicating the physical conditions of the process operation (e.g., target pressure, temperature, exhaust, energy throughput, etc.).

[0079] The system controller 306 may include software and / or hardware components capable of executing the operations of the process recipe 308. The system controller 306 may monitor the process through sensors 310. The sensors 310 may measure process parameters to determine whether process standards (e.g., target process standards) are met. The process standards may be related to a process parameter value window. The sensors 310 may include various sensors that can be used to measure (explicitly measure, or act as a metric) the consumption related to substrate processing (e.g., power, current, etc.). The sensors 310 may include physical sensors, integrated sensors that are components of the processing chamber, external sensors, Internet-of-Things (IoT), and / or virtual sensors (e.g., non-physical sensors but sensors based on virtual measurement values that are based on models estimating parameter values), etc.

[0080] Additionally or alternatively, the system controller 306 may monitor the eco-efficiency by measuring the resource consumption of various process operations (e.g., waste discharge, energy consumption, process ingredient consumption, etc.). In some embodiments, the system controller 306 determines the eco-efficiency of the associated machinery 304. The system controller 306 may also adjust the settings related to the manufacturing equipment 304 based on a predicted and / or determined eco-efficiency model (e.g., including determined and / or predicted modifications to the process recipe 308) to optimize the eco-efficiency of the equipment 304 according to the current manufacturing conditions.

[0081] In one embodiment, the system controller 306 may include a main memory (e.g., read-only memory (ROM), flash memory, dynamic random access memory (DRAM), static random access memory (SRAM), etc.), and / or an auxiliary memory (e.g., a data storage device such as a disk drive) (e.g., data storage 312 or cloud data). The main memory and / or the auxiliary memory may store instructions for performing various types of manufacturing processes (e.g., the process recipe 308).

[0082] In one embodiment, the system controller 306 may determine an actual ecological efficiency characterization associated with the manufacturing device 304 based on first utility usage data associated with the manufacturing device 304 and first utilization data associated with the manufacturing device 304. For example, the first utility usage data and the first utilization data may be determined by the system controller 306. In another embodiment, the first utility usage data and the first utilization data are received from an external source (e.g., server 320, cloud service, and / or cloud data storage). The system controller 306 may compare the actual ecological efficiency characterization with a first ecological efficiency characterization (e.g., a first estimated ecological efficiency characterization) and / or a predicted ecological efficiency associated with the manufacturing device 304. The ecological efficiency characterization may be different from the actual value associated with the operating manufacturing device 304 when different usage and utilization data values are used to calculate the first ecological efficiency characterization.

[0083] In one embodiment, the system controller 306 may determine that the predicted ecological efficiency characterization is more efficient than the actual ecological efficiency characterization, indicating that it may be possible to adjust the settings of the manufacturing device 304 to better optimize the ecological efficiency of the manufacturing device 304. In some embodiments, the manufacturing device 304 may control and / or adjust sub-component settings to better optimize the ecological efficiency.

[0084] The system controller 306 may also determine that the actual usage data or the actual utilization data is different from the predicted usage data and utilization data associated with the first ecological efficiency characterization. This may be the case when nominal or estimated data values are used to determine the first ecological efficiency characterization and different actual recorded data values are used when the manufacturing device 304 is operating. In such a case, adjustment of one or more settings associated with the manufacturing device 304 may be beneficial to optimize the ecological efficiency of the manufacturing device.

[0085] The data storage 312 can be a memory (e.g., random access memory), a drive (e.g., hard disk drive, flash drive), a database system, or another type of component or device capable of storing data, such as a memory provided by a cloud server and / or a processor. The data storage 312 can store one or more historical sensor data. The data storage 312 can store one or more eco-efficiency data 314 (e.g., including historical, predicted, and / or current eco-efficiency data), sensor and process recipe data 316 (e.g., including historical, predicted, and / or current sensor and process recipe data 316), modification and optimization data (e.g., including historical, predicted, and / or current modification and optimization data 318), and digital replica data 319. The sensor and process recipe data 316 can include various process operations, process parameter windows, alternative process operations, process queue instructions, etc. for performing multiple processes on overlapping manufacturing equipment. The sensor and process recipe data 316 can be linked or otherwise associated with the eco-efficiency data 314 to track and / or predict the eco-efficiency of various process operations, recipes, etc. The modification and optimization data 318 can include historical modifications made to previous process recipes (including individual process operations, or the coordination of multiple process recipes), and the associated eco-efficiency changes resulting from such modifications.

[0086] The eco-efficiency data 314 can include various resource consumptions used in the eco-efficiency characterization and / or prediction. In one embodiment, the eco-efficiency data 314 incorporates water usage, emissions, electrical energy usage, and one or more of any combination of the foregoing related to substrate processing. In some embodiments, the eco-efficiency data 314 can include other categories of resource consumption, such as gas usage, heavy metal usage, and eutrophication potential.

[0087] The digital replica data 319 can include data related to digital replicas. The digital replica data 319 can include data related to digital twins. As used herein, a digital twin can include a digital replica of a physical asset such as the manufacturing equipment 304. The digital twin includes the characteristics of the physical asset at each stage of the manufacturing process, where the characteristics include but are not limited to axis dimensions, weight characteristics, material characteristics (e.g., density, surface roughness), electrical characteristics (e.g., conductivity), optical characteristics (e.g., reflectivity), etc.

[0088] As described above, the digital replica may include a physics-based model of one or more physical assets of the substrate manufacturing system. The digital replica data 319 may encapsulate relationships, parameters, specifications, etc. related to one or more aspects of the physics-based model. For example, the physics-based model may indicate the relationship between the size and geometry of the substrate processing chamber and the environmental resource consumption. The physics-based model may indicate the relationship between the type of purge gas used within the substrate manufacturing system and the environmental resource consumption.

[0089] The server 320 may include a component integration tool 322, a digital replica tool 324, an optimization tool 326, a recipe builder tool 328, a resource consumption tool 330, and / or a discovery tool. The component integration tool 322 may determine the cumulative consumption of each device (e.g., each individual manufacturing equipment). As described herein, the various tools of the server 320 may communicate data with each other to perform each corresponding function.

[0090] The component integration tool 322 may receive manufacturing data (e.g., recipes, recipe selections, manufacturing equipment, inter-recipe and intra-recipe processes, etc.) and perform an eco-efficiency prediction analysis on different parts of the data. In some embodiments, the component integration tool 322 may determine an eco-efficiency characterization across multiple process operations from a single process recipe. For example, the component integration tool 322 may determine an eco-efficiency prediction for all operations in the substrate manufacturing process from start to finish. For example, each manufacturing operation may include one or more manufacturing operations (e.g., hundreds of manufacturing operations), each with an eco-efficiency prediction and a collective eco-efficiency prediction at the same time. In another example, a process selection set may be used to predict the eco-efficiency of a subset of manufacturing process operations.

[0091] In another embodiment, the component integration tool 322 may perform an eco-efficiency prediction on inter-recipe processes. For example, the eco-efficiency prediction may be related to a manufacturing device (e.g., a manufacturing device of the manufacturing system 302) that performs multiple different process operations from multiple different manufacturing processes (e.g., process recipe 308). In another example, the sequencing of the various process operations (e.g., intra-recipe or inter-recipe) may affect the overall eco-efficiency. The component integration tool 322 may perform an overall eco-efficiency prediction on the manufacturing device system and / or the process sequence. For example, the component integration tool 322 may perform an eco-efficiency comparison between sub-components (e.g., multiple processing chambers) that perform similar functions.

[0092] In an illustrative example, each process operation can be performed by a processing chamber, such as epitaxial deposition or etching. Each of these operations is performed using a process recipe. There may be many different process recipes to perform a process such as epitaxial deposition. For example, a process recipe can include multiple operations such as: 1) purging the chamber; 2) pumping; 3) gas flow; 4) heating the chamber, etc. These operations can be associated with one or more process recipes.

[0093] In another embodiment, the component integration tool 322 can perform an eco-efficiency prediction that includes the eco-efficiency of auxiliary equipment. Auxiliary equipment may include equipment that is not directly used in manufacturing but helps to perform various process recipes. For example, auxiliary equipment can include a substrate transfer system designed to move wafers between various manufacturing apparatuses. In another example, auxiliary equipment can include heat sinks, common exhaust ports, power delivery systems, etc. The component integration tool 322 can consider the resource consumption of the auxiliary devices and combine the auxiliary device resource consumption with the manufacturing resource consumption to predict the resource consumption of a process recipe (e.g., a subset or the entire recipe) or a recipe combination (e.g., multiple subsets or multiple entire recipes).

[0094] In another embodiment, the component integration tool 322 can perform an eco-efficiency prediction that considers a series of processes or recipes. For example, performing process operation A followed by process operation B results in a first resource consumption, while performing process operation B followed by process operation A results in a second resource consumption different from the first resource consumption. The component integration tool 322 integrates the eco-efficiencies of multiple pieces of machinery and / or process operations and considers the process operation sequence of a process recipe (e.g., a subset or the entire recipe) or a recipe combination (e.g., multiple subsets or multiple entire recipes).

[0095] In some embodiments, each process operation has a different manufacturing device. For example, a film on a wafer can have multiple layers. A first machine can perform a first operation (e.g., deposition), a second machine can perform a second operation (e.g., etching), a third machine can perform a third operation (e.g., deposition), etc. The component integration tool 322 can instruct a resource consumption tracker to track multiple processing operations on multiple machines to generate a data storage report. As previously mentioned, the consumption report can be plotted for process recipe selection, including the entire wafer lifetime.

[0096] In some embodiments, the component integration tool 322 can perform an environmental resource consumption comparison between chambers. The component integration tool can utilize a digital replica tool 324 to provide one or more physical data that indicate the rationale for the eco-efficiency difference between two chambers.

[0097] The digital replica tool 324 receives manufacturing data from the manufacturing system 302 and / or the client device 350 and generates a digital replica related to the manufacturing data. The manufacturing data may include a selection of machine equipment 304 and process operations of the process recipe 308. The digital replica tool 324 generates a digital twin of the physical system architecture or virtual input system of the manufacturing system (e.g., generated by the user on the client device 350).

[0098] The digital replica generated by the digital replica tool 324 may include one of a physical model, a statistical model, and / or a hybrid model. The physical model may include physics-based constraints and control algorithms that are designed to estimate the physical conditions of the input manufacturing data (e.g., exhaust temperature, power delivery requirements, and / or other conditions indicating the physical environment related to environmental resource consumption). For example, the user may generate a process recipe on the client device 350. The process recipe may include the parameters of the process or recipe and instructions for using the machine equipment in a certain way. The digital replica tool 324 may obtain the manufacturing data and determine the physical constraints of the system (e.g., operating temperature, pressure, exhaust parameters, etc.). For example, the physical model may identify the physical conditions of the system based on the hardware configuration of the chamber (e.g., using equipment material of type A or using equipment material of type B) and / or the recipe parameters. In another example, the physical conditions may be determined according to the relevant machine equipment components that affect the heat loss of water, air, and / or heating ventilation, and air conditioning (HVAC) equipment. The digital replica tool 324 may cooperate with other tools (e.g., the component integration tool 322 and / or the resource consumption tool 330) to specify the eco-efficiency prediction related to the received manufacturing data. It should be noted that the digital replica tool 324 can predict the eco-efficiency of the manufacturing process and the selection of manufacturing equipment without receiving empirical data from the execution of the process recipe by the manufacturing equipment 304. Therefore, the digital replica of the manufacturing equipment can be used to predict the eco-efficiency of the equipment design and / or the process recipe without actually constructing a specific equipment design or running a specific process recipe.

[0099] In some embodiments, the digital replica tool 324 may operate in conjunction with a digital twin. As used herein, a digital twin is a digital replica of a physical asset such as a manufactured part. The digital twin includes the physical asset characteristics of each stage of the manufacturing process, where such characteristics include, but are not limited to, axis dimensions, weight characteristics, material characteristics (e.g., density, surface roughness), electrical characteristics (e.g., conductivity), optical characteristics (e.g., reflectivity), and other characteristics.

[0100] In some embodiments, the physical models used by the digital replica tool 324 may include fluid flow modeling, gas flow and / or consumption modeling, chemistry-based modeling, heat transfer modeling, electrical power consumption modeling, plasma modeling, and the like.

[0101] In some embodiments, the digital replica tool 324 may employ statistical modeling to predict the eco-efficiency of manufacturing data. Statistical models can be used to process manufacturing data based on historical eco-efficiency data of previous processes (e.g., eco-efficiency data 314), by using statistical operations to validate, predict, and / or transform the manufacturing data. In some embodiments, a statistical model is generated by using statistical process control (SPC) analysis to determine the control limits of the data and to identify the reliability level of the data based on those control limits. In some embodiments, the statistical model is related to univariate and / or multivariate data analysis. For example, various parameters can be analyzed by using a statistical model to determine patterns and correlations through statistical processes (e.g., range, minimum, maximum, quartiles, variance, standard deviation, etc.). In another example, the relationships between multiple variables can be determined by using regression analysis, path analysis, factor analysis, multivariate statistical process control (MCSPC), and / or multivariate analysis of variance (MANOVA).

[0102] The optimization tool 326 can receive a selection of process recipes 308 and machine equipment 304 and identify modifications to the selection to improve eco-efficiency (e.g., reduce resource consumption, resource cost consumption, and / or environmental impact (e.g., gaseous or particulate species entering the atmosphere)). The optimization tool 326 can use one or more machine learning models in combination (e.g., model 390 of the machine learning system 370). In some instances, a first machine learning model can receive a target substrate output related to the target result of the processed substrate as input. A second machine learning model can receive a selection of process recipes as input (e.g., the output from the first machine learning model) and determine eco-efficiency data for each recipe corresponding to the selection of process recipes. In some embodiments, the machine learning model can determine one or more modifications to the selection that, when executed by the manufacturing system 302, improve the overall eco-efficiency of the selection. In some embodiments, the machine learning model can use a digital replica tool to generate synthetic manufacturing data for training. Alternatively or additionally, the machine learning model can use historical data (e.g., eco-efficiency data 314, sensor and process recipe data 316, and / or modification and optimization data 318) to train the machine learning model.

[0103] Modifications identified by the optimization tool 326 can include changing process operations, changing process sequences, changing parameters executed by machine equipment, changing the interaction between a first process recipe and a second process recipe (e.g., sequence, simultaneous operation, delay time, etc.), and so on. In some embodiments, the optimization tool 326 can send instructions to the manufacturing system 302 to directly execute the optimization. However, in other embodiments, the optimization tool can display the modifications on a graphical user interface (GUI) for an operator to take action. For example, the digital replica tool 324 can send one or more modifications to the client device 350 for display in the browser 352 and / or application 354.

[0104] In some embodiments, the optimization tool 326 can adjust the hyperparameters of the digital twin model generated by the digital replica tool 324. As will be discussed in the embodiments below, the optimization tool 326 can combine reinforcement learning and / or deep learning by running simulation modifications on the digital replica and evaluating the predicted eco-efficiency results output from the digital replica.

[0105] In some embodiments, the optimization tool 326 may perform ecological efficiency prediction and optimization that prioritizes one or more types of environmental resources. For example, as previously described, the ecological efficiency prediction may be based on various predicted resource consumptions such as water usage, gas usage, energy usage, etc. The optimization tool 326 may perform an optimization that prioritizes a first resource consumption (e.g., water usage) over a second resource consumption (e.g., gas usage). In some embodiments, the optimization tool 326 may perform an optimization that uses a weighted priority system. For example, when optimizing ecological efficiency and / or identifying ecological efficiency modifications to a process, one or more resource consumptions may be assigned a weight that may indicate the optimization priority of the associated unit resource consumption.

[0106] The recipe builder tool 328 may receive a selection of manufacturing processes and / or machine equipment and incrementally and dynamically predict ecological efficiency after each addition, deletion, and / or modification to the virtual manufacturing process and / or equipment selection. The recipe builder tool 328 may use other tools (e.g., the component integration tool 322, the digital replica tool 324, the optimization tool 326, and the resource consumption tool 330) to dynamically update the determined ecological efficiency when updating the manufacturing recipe. For example, a user may create a manufacturing recipe. The recipe builder tool 328 may output the current ecological efficiency of the current iteration of the process recipe. The recipe builder tool 328 may receive a modification to the current iteration of the updated process recipe. The recipe builder tool 328 may output an updated ecological efficiency prediction. In some embodiments, the recipe builder tool 328 uses one or more models to predict a substrate process recipe that meets threshold criteria. (See Figure 2B and related descriptions).

[0107] In some embodiments, the recipe builder tool 328 and the optimization tool 326 may be used to identify one or more recipes that have a higher ecological efficiency than other recipes. For example, the recipe builder tool 328 may cause or otherwise present on a GUI (e.g., the client device 350) one or more (e.g., the top three) recipes with the highest energy efficiency associated with the processing tool. The recipe builder tool 328 may provide details using the digital replica tool 324 that indicate the rationale for why one or more high-energy efficiency recipes can perform with a corresponding high ecological efficiency.

[0108] The resource consumption tool 330 can track various resource consumptions (e.g., predicted resource consumptions). For example, as previously described, the eco-efficiency prediction can be based on a broader range of resources such as energy consumption, gas emissions, water usage, etc. However, the resource consumption tool 330 can more specifically track the predicted resource consumptions. In some embodiments, the resource consumption tool 330 receives a process recipe and / or a selected set of manufacturing equipment. The resource consumption tool 330 can determine the life cycle data of components associated with the selected set of manufacturing equipment and / or the process recipe. For example, manufacturing equipment wears out with use and, in some cases, requires corrective actions such as replacing and / or repairing components. Such corrective actions are also associated with the predicted environmental consumption (e.g., the predicted resource consumption to perform the corrective action). The resource consumption tool 330 can track the component life data separately and provide the unit environmental resource consumption and / or environmental impact based on the expected future corrective actions to be performed.

[0109] In some embodiments, the environmental resource consumption can be predicted, monitored, tracked, and / or determined for various faults. In some embodiments, the resource consumption tool 330 can predict the resource consumption based on a selected process recipe. In some embodiments, the resource consumption tool 330 can perform real-time monitoring of energy, gas, and water consumptions. The resource consumption tool 330 can determine the chamber-level consumption, including calculating the total electrical power consumption, gas consumption, and water consumption of the chamber (e.g., per wafer, per day, per week, per year, etc.). The resource consumption tool 330 can determine the tool-level consumption, including determining the total electrical power consumption, gas consumption, and water consumption of the tool (e.g., per day, per week, per year, etc.). The resource consumption tool 330 can determine the individual gas consumption, including determining the components of the individual gas consumption (e.g., per wafer, per day, per week, per year, etc.). The resource consumption tool 330 can generate a standard report including the energy, gas, and water consumptions at the chamber and tool levels.

[0110] In some embodiments, the resource consumption tool 330 can predict the total electrical power consumption, gas consumption, and water consumption of all sub-fab components (e.g., per day, per week, per year, etc.). The resource consumption tool can determine the recipe-level consumption, including the total electrical power consumption, gas consumption, and water consumption for any recipe run on a chamber / or tool. The resource consumption tool can determine the component-level consumption, including the components of the energy consumption of all energy-consuming components within the chamber. The resource consumption tool 330 can perform customized reports on demand, including determining the information on demand and the eco-efficiency report on demand. The resource consumption tool 330 can perform comparisons between the energy consumptions of different recipes and / or time points, including quantifying the energy savings and the energy-saving opportunities through the use of recipe optimization (e.g., using the optimization tool 326).

[0111] The exploration tool 332 can communicate with the digital replica tool 324 to determine the effects of one or more updates to the manufacturing equipment 304. The exploration tool 332 can utilize the digital replica tool 324 to generate a digital replica that includes a digital reproduction of a substrate manufacturing system (e.g., the manufacturing equipment 304). The exploration tool can receive updates to the manufacturing equipment and allow a user to explore various alternative arrangements of the equipment in use, equipment configurations, process parameters related to equipment performance, etc. The exploration tool 332 can use the resource consumption tool 330 to determine environmental resource usage data that corresponds to performing one or more processing procedures by the substrate manufacturing system incorporating the updates as described herein. The environmental resource usage data can be provided for display on a graphical user interface (GUI) (e.g., on the client device 350).

[0112] In some embodiments, the environmental resource usage data determined and / or predicted by other tools of the server can include predicted environmental resource consumption and / or predicted environmental impact related to either a replacement procedure or a maintenance procedure of a consumable portion of the first manufacturing equipment. In some embodiments, the optimization tool 326 can determine modifications to the manufacturing process, and such modifications can include performing corrective actions related to components of the machine equipment (e.g., the machine equipment 304).

[0113] The exploration tool 332 can perform a cost - of - ownership analysis related to the manufacturing system. The cost - of - ownership analysis can include a comprehensive analysis of the interaction of the manufacturing system to calculate the total cost of owning and / or operating the system. The exploration tool 332 can calculate the cost for a consumer to perform a specific manufacturing procedure. The exploration tool 332 can determine the wafer cost, the cost corresponding to the gases used by the system, the cost related to the tools in use (e.g., lifetime degradation data), and the electricity used for the manufacturing system to perform one or more process procedures. The cost of ownership can be calculated per unit (e.g., per wafer).

[0114] In some embodiments, the machine learning system 370 further includes servo machine 372, servo machine 380, and / or servo machine 392. The servo machine 372 includes a data set generator 374 that is capable of generating a data set (e.g., a set of data inputs and a set of target outputs) to train, validate, and / or test the machine learning model 390.

[0115] The servo machine 380 includes a training engine 382, a validation engine 384, and / or a test engine 386. An engine (e.g., the training engine 382, the validation engine 384, and / or the test engine 386) can refer to hardware (e.g., circuitry, dedicated logic, programmable logic, microcode, a processing device, etc.), software (such as instructions running on a processing device, a general computer system, or a dedicated machine), firmware, microcode, or a combination of the foregoing. The training engine 382 is capable of training the machine learning model 390 by using one or more sets of features related to the training set from the data set generator 374. The training engine 382 can generate one or more trained machine learning models 390, where each trained machine learning model 390 can be trained based on a different set of features of the training set and / or a different set of labels of the training set. For example, the first trained machine learning model has been trained by using the resource consumption data output by the digital replica tool 324, the second trained machine learning model has been trained by using historical eco-efficiency data (e.g., the eco-efficiency data 314), etc.

[0116] The validation engine 384 is capable of validating the trained machine learning model 390 by using the validation set from the data set generator 374. The test engine 386 is capable of testing the trained machine learning model 390 by using the test set from the data set generator 374.

[0117] The machine learning model 390 can refer to one or more trained machine learning models created by the training engine 382 using a training set that includes data inputs and, in some embodiments, includes corresponding target outputs (e.g., the correct answers for each training input). Patterns in the data set can be found that cluster the data inputs and / or map the data inputs to the target outputs (correct answers), and the mapping is provided to the machine learning model 390 and / or the machine learning model 390 learns to capture the mapping of the patterns. The machine learning model 390 can include an artificial neural network, a deep neural network, a convolutional neural network, a recurrent neural network (e.g., a long short term memory (LSTM) network, a convLSTM network, etc.), and / or other types of neural networks. The machine learning model 390 can additionally or alternatively include other types of machine learning models, such as those using one or more of linear regression, Gaussian regression, random forest, support vector machine, etc.

[0118] One machine learning model that can be used to perform some or all of the above tasks is an artificial neural network, such as a deep neural network. An artificial neural network generally includes a feature representation component that has a classifier or a regression layer that maps features to a desired output space. For example, a convolutional neural network (CNN) hosts multiple layers of convolutional filters. Pooling is performed at lower layers, and non-linearity can be processed. Generally, multiple layers of perceptrons are added above the lower layers to map the top-level features extracted by the convolutional layer to a decision (e.g., a classification output). Deep learning is a class of machine learning algorithms that use cascades of multiple non-linear processing units for feature extraction and transformation. Each subsequent layer uses the output from the previous layer as input. Deep neural networks can learn in a supervised (e.g., classification) and / or unsupervised (e.g., pattern analysis) manner. Deep neural networks include hierarchical layers, where different layers learn different levels of representation corresponding to different levels of abstraction. In deep learning, each level learns to transform its input data into a slightly more abstract and composite representation. For example, in an image recognition application, the original input can be a pixel matrix; the first representation layer can extract pixels and encode edges; the second layer can form and encode the arrangement of edges; the third layer can encode higher-level shapes (e.g., teeth, lips, gums, etc.); and the fourth layer can recognize the scanned object. It is worth noting that the deep learning process can learn by itself which features are best placed at which level. The "depth" in "deep learning" refers to the number of layers through which the data transformation passes. More precisely, a deep learning system has a relatively large credit assignment path (CAP) depth. The CAP is a chain of transformations from input to output. The CAP describes the potential causal relationship between the input and the output. For a feedforward neural network, the depth of the CAP can be the depth of the network and can be the number of hidden layers plus one. For a recurrent neural network where the signal can propagate through the layers more than once, the CAP depth may be unlimited.

[0119] The training of a machine learning model can be roughly divided into supervised learning and unsupervised learning. Both techniques for training a machine learning model can be used in an implementation. In one implementation, the training of a neural network can be achieved in a supervised learning manner, which involves feeding a training data set composed of labeled inputs through the network, observing its output, defining an error (by measuring the difference between the output and the label value), and using techniques such as deep gradient descent and backpropagation to tune the weights across all layers and nodes of the network so that the error is minimized. In many applications, repeating this process for many labeled inputs in the training data set results in a network that can produce the correct output when an input different from the inputs present in the training data set appears. In a high-dimensional setting, such as large images, this generalization can be achieved when a sufficiently large and diverse training data set is available.

[0120] For model training, a training data set containing hundreds, thousands, tens of thousands, hundreds of thousands, or more data inputs should be used to form the training data set. In an embodiment, up to thousands, tens of thousands, hundreds of thousands, or millions of historical data (e.g., historical data of the processes and resource consumption-related tags performed in the processing chamber) can be used to form the training data set, where each case can include various tags of one or more types of useful information. Each case can include, for example, data showing the processing chamber, recipe, various resource utilization rates, etc. This data can be processed to generate one or more training data sets for training one or more machine learning models. The machine learning model can be trained, for example, based on the input processing chamber, recipe, and / or process target information to predict the process recipe, estimate resource consumption and / or ecological efficiency, propose modifications to the recipe and / or processing chamber, etc. Such a trained machine learning model can be added to the ecological efficiency dashboard and can be applied to provide detailed information on resource consumption and ecological efficiency, and methods for reducing resource consumption and / or improving ecological efficiency before, during, and / or after performing a process on the processing chamber.

[0121] The processing logic can collect a training data set including historical process run information with one or more associated tags (e.g., tags of resource consumption, ecological efficiency values, suggestions for improving process recipe parameters, process recipe parameters, etc.). The training data set can be augmented additionally or alternatively. Training of large-scale neural networks generally uses tens of thousands of inputs, which are not easily obtained in many real-world applications. Data augmentation can be used to artificially increase the effective sample size.

[0122] To implement the training, the processing logic inputs the training data set into one or more untrained machine learning models. The machine learning model can be initialized before the first input is input into the machine learning model. The processing logic trains the untrained machine learning model based on the training data set to generate one or more trained machine learning models that perform the various operations described above.

[0123] Training can be performed by inputting one or more of the data inputs into the machine learning model one at a time. Each input can include data from the historical process running in the training data items of the training dataset. The machine learning model processes the input to produce an output. An artificial neural network includes an input layer, which consists of values in the data points (e.g., intensity values and / or height values of pixels in a height map). The next layer is called the hidden layer, and the nodes of the hidden layer each receive one or more of the input values. Each node contains parameters (e.g., weights) applied to the input values. Thus, each node essentially inputs the input values into a multivariate function (e.g., a non-linear mathematical transformation) to produce an output value. The next layer can be another hidden layer or the output layer. In either case, the nodes of the next layer receive the output values from the nodes of the previous layer, and each node applies weights to these values and then produces its own output value. This action can be performed at each layer. The last layer is the output layer, where there is a node for each class, prediction, and / or output that the machine learning model can produce. Thus, the output can include predictions or estimated resource consumption of one or more resources and can include an eco-efficiency value, and so on.

[0124] Subsequently, the processing logic can compare the produced output with the known labels included in the training data items. The processing logic determines an error (i.e., classification error) based on the difference between the output and the provided labels. The processing logic adjusts the weights of one or more nodes in the machine learning model based on the error. An error term or increment can be determined for each node in the artificial neural network. Based on the error, the artificial neural network adjusts one or more of its parameters (weights of one or more inputs of the node) for one or more of its nodes. The parameters can be updated in a backpropagation manner, so that the nodes of the highest layer are updated first, followed by the nodes of the next layer, and so on. The artificial neural network contains multiple layers of "neurons", where each layer receives the values from the neurons of the previous layer as inputs. The parameters of each neuron include weights, which are related to the values received from each neuron of the previous layer. Thus, adjusting the parameters can include adjusting the weights assigned to each input of one or more neurons in one or more layers of the artificial neural network.

[0125] Once the model parameters have been optimized, model validation can be performed to determine whether the model has been improved and to determine the current accuracy of the deep learning model. After one or more rounds of training, the processing logic can determine whether a stopping criterion is met. The stopping criterion can be a target accuracy level, a target number of processed images from the training data set, a target change in the parameters on one or more previous data points, a combination of the foregoing, and / or other criteria. In one embodiment, the stopping criterion is met when at least a minimum number of data points have been processed and at least a threshold accuracy has been achieved. The threshold accuracy can be, for example, an accuracy of 70%, 80%, or 90%. In one embodiment, the stopping criterion is met if the accuracy of the machine learning model has stopped improving. If the stopping criterion is not met, further training is performed. If the stopping criterion has been met, training can be completed. Once the machine learning model has been trained, the held-out portion of the training data set can be used to test the model.

[0126] The modification identification component 394 can provide current data to the trained machine learning model 390 and can run the trained machine learning model 390 on the input to obtain one or more outputs. The modification identification component 394 is capable of making determinations and / or performing operations based on the output of the trained machine learning model 390. The ML model output can include confidence data that indicates the confidence level of the ML model output (e.g., modifying and optimizing parameters) corresponding to the modification, which when applied, improves the overall eco-efficiency of a selected set of manufacturing processes and / or manufacturing equipment. In some embodiments, the modification identification component 394 can perform process recipe modifications based on the ML model output. The modification identification component 394 can provide the ML model output to one or more tools of the server 320.

[0127] The confidence data can include or indicate the confidence level that the ML model output is correct (e.g., the ML model output corresponds to a known label associated with a training data item). In one example, the confidence level is a real number between 0 and 1 (inclusive of 0 and 1), where 0 indicates no confidence that the ML model output is correct and 1 indicates absolute confidence that the ML model output is correct. In response to the confidence data indicating that the confidence levels of a predetermined number of instances (e.g., instance percentage, instance frequency, total number of instances, etc.) are below a threshold level, the server 320 can cause the trained machine learning model 390 to be retrained.

[0128] For purposes of illustration and not limitation, aspects of the present disclosure describe training a machine learning model by using process target data and / or process recipe data and inputting a current selection of manufacturing processes and / or manufacturing equipment into the trained machine learning model to determine a machine learning model output (predicted eco-efficiency data based on process targets, such as predicted resource consumption, etc.). In other embodiments, a heuristic model or a rule-based model is used to determine the output (e.g., without using a trained machine learning model).

[0129] In some embodiments, the functionality of the manufacturing system 302, the client device 350, the machine learning system 370, the data storage 312, and / or the server 320 may be provided by a smaller number of machines. For example, in some embodiments, the servo machines 372 and 380 may be integrated into a single machine, and in some other embodiments, the servo machines 372, the servo machine 380, and the servo machine 392 may be integrated into a single machine. In some embodiments, the server 320, the manufacturing system 302, and the client device 350 may be integrated into a single machine.

[0130] Generally, functions described in one embodiment as being performed by the manufacturing system 302, the client device 350, and / or the machine learning system 370 may also be performed on the server 320 in other embodiments (where applicable). Additionally, functionality attributed to a particular component may be performed by different or multiple components operating together. For example, in some embodiments, the server 320 may receive manufacturing data and perform machine learning operations. In another example, the client device 350 may perform manufacturing data processing based on an output from a trained machine learning model.

[0131] Furthermore, the functionality of a particular component may be performed by different or multiple components operating together. One or more of the server 320, the manufacturing system 302, or the machine learning system 370 may be accessed as a service that is provided to other systems or devices through an appropriate application programming interface (API).

[0132] In an embodiment, a "user" may represent a single individual. However, other embodiments of the present disclosure include a "user" as an entity controlled by multiple users and / or automated sources. For example, a group of individual users united as a group of administrators may be considered a "user".

[0133] Figure 4Illustrates an exemplary digital replica according to some embodiments of the present disclosure. The digital replica 400 may include a digital twin of a manufacturing system selection set and may include, for example, a digital reproduction of a manufacturing system that includes the same chambers, valves, gas delivery pipelines, materials, chamber components, etc. The digital replica 400 may receive manufacturing equipment process data as input, which may include first sensor data 402 to 404 output by integrated sensors of a processing chamber and second sensor data 406 to 408 output by external sensors of non-processing chamber components. The input may further include a process recipe and / or output physical conditions of the manufacturing system that includes the processing chamber. In some embodiments, the digital replica 400 includes a physics-based model that may incorporate various physical relationships, such as thermodynamics, fluid dynamics, energy conservation, gas laws, mechanical systems, energy conservation, transport, and conveyance, etc. The digital replica 400 processes the input data and generates an output 410. The output may include one or more physical conditions of the processing chamber and / or other systems or devices. The output may additionally or alternatively include environmental resource usage data.

[0134] In one example, the digital replica 400 may receive a first gas flow rate of a first gas, a second gas flow rate of a second gas, and a third gas flow rate of a third gas and a first process recipe as input. The digital replica may use the physics-based model to estimate the amount of energy leaving the chamber through the gas flow. For example, the model determines the exhaust temperature and the total energy flow through the exhaust device. In another example, the same digital replica 400 may output predicted eco-efficiency data, such as predicted environmental resource consumption. The digital replica may identify relevant resource consumption and identify recommended optimizations to improve energy conservation.

[0135] In some embodiments, the digital replica 400 may determine the exhaust of one or more gas panels or gas boxes that contain gases used at one or more locations throughout the manufacturing system. For example, each gas box may use a dedicated exhaust device with a negative pressure to effectively exhaust the gas, such as in the case of a gas pipeline leak or a more general malfunction (e.g., to prevent toxins from entering the manufacturing equipment or an undesirable location in the manufacturing system). The digital replica may be part of a digital twin that utilizes information about the possible types and volumes of gases in the gas box and determines the adjustment of the exhaust flow rate required to properly dispose of the gas (e.g., evacuate the leak). The exhaust flow rate may be determined based on the exhaust pressure and flow rate. The exhaust flow rate may include determining relevant parameters to optimize eco-efficiency while maintaining minimum safety thresholds and / or standards.

[0136] In some embodiments, the digital replica 400 may indicate exhaust temperature and total energy flow through the exhaust based on heating within the process chamber. For example, during a substrate processing procedure, the process chamber may include one or more processing devices, such as a substrate pedestal. Excess heat within the chamber may be reduced via exhaust. The operation of the pedestal may be varied to reduce heat loss through the exhaust. Several methods have been reported to control heat transfer in heat transfer components, such as a pedestal for supporting a substrate that includes heating and cooling elements, the cooling elements removing excess heat by circulating a cooling medium (such as a gaseous or liquid coolant) within the pedestal or between the substrate and the pedestal. When the substrate temperature rises above a set range during the process, the heating elements are turned off and the cooling elements are activated to remove the excess heat, thereby controlling the temperature. One or more parameters associated with the process may be used as inputs to the digital replica 400 to determine how much excess heat is lost through the exhaust device.

[0137] In some embodiments, the digital replica 400 may indicate energy flow and / or chemicals, including lost precursors or reaction by-products discharged from a scrubber system. For example, gaseous effluent streams from the manufacture of electronic materials, components, products, solar cells, and memory articles (hereinafter referred to as "electronic components") may involve decomposition products of a wide variety of chemical compounds, organic compounds, oxidants, photoresists, and other reagents, as well as other gases and suspended particles. Ideally, such gases and suspended particles may be removed from the effluent stream before the effluent stream is discharged from the processing facility to the atmosphere.

[0138] The effluent stream to be scrubbed may include species generated by an electronic component manufacturing process and / or species delivered to an electronic component manufacturing process and transported through the process chamber without undergoing chemical change. As used herein, the term "electronic manufacturing process" is intended to be construed broadly to include any and all processing and unit operations in electronic component manufacturing, and all disposal or processing operations involving materials used or produced in an electronic component and / or LCD manufacturing facility, and all operations performed in the context of an electronic component and / or LCD manufacturing facility not involving active manufacturing (examples include conditioning of processing equipment, purging of chemical delivery lines during operation preparation, etch cleaning of processing tool chambers, scrubbing of toxic or harmful gases from effluents generated by an electronic component and / or LCD manufacturing facility, etc.).

[0139] In some embodiments, the digital replica 400 accounts for the exhaust flow of the leaked gas, or as part of a cleaning procedure. For example, the gas can be periodically purged from the manufacturing asset to increase the lifespan of the asset, improve the performance of the product, or prepare the product for different functions that are ready and assigned to be executed. The digital replica 400 can determine the environmental consumption (e.g., energy consumption, gas consumption) associated with performing this purification procedure. For example, the digital replica 400 can indicate the energy and / or gas consumption for purging the system (e.g., constantly providing an air stream to the system to maintain dynamic gas movement within the system). The digital replica 400 can indicate how the energy and / or gas consumption changes by adjusting one or more gas flow rates (e.g., purge gas) within the processing system.

[0140] In some embodiments, the digital replica 400 can utilize a process recipe and determine which gases are entering the processing chamber, what reactions are occurring on the substrate placed within the processing chamber, how the gases are utilized in the substrate reactions, etc. The digital replica 400 can further determine what gases remain and the amounts of such gases after the reactions occur on the substrate surface. The digital replica 400 can further determine the amounts and types of gases lost through abatement. The digital replica 400 can further determine what the final by-products of the abatement are and the overall environmental impact of the final by-products.

[0141] In some embodiments, one or more substrate processing procedures may require a consistent gas inflow and / or outflow to the processing chamber to process the substrate to meet the target processing result conditions. The substrate processing system can perform a stable gas flow procedure by executing one or more flow-to-outlet-to-flow-to-chamber conversions to reduce the transient gas flow caused by opening / closing the gas flow to the chamber. For example, a first gas flow can be initialized and exhausted, and once the gas flow has stabilized, a stable gas flow can be provided to the processing chamber by directing the exhausted air to the processing chamber. Due to this process, the digital replica 400 can determine the gas consumption (e.g., the gas lost through exhaust). For example, the digital replica can identify the transition time and the amount of gas lost through venting during the transient period of initializing or terminating the gas flow. The digital replica can determine the optimization of the transition between gas venting and introducing gas into the chamber. Optimizing the transition time can reduce the gas loss through exhaust while identifying the time when the gas reaches a steady state. In some embodiments, the transition rhythm of the gas flow can be determined based on the process result requirements. For example, the gas flow transition time can be determined (e.g., optimized) to include flow rates that do not negatively impact the processing results within the corresponding processing chamber.

[0142] The digital replica 400 can be used to predict eco-efficiency data related to one or more operating states of the physical assets of a manufacturing system. In one example, the digital replica 400 can receive data related to one or more operating states of the physical assets of a manufacturing system. For example, the digital replica 400 can receive power reduction data, sleep mode data, shared operating mode data, and / or process recipe data indicating one or more processing procedures performed by the manufacturing system represented by the digital replica 400.

[0143] Energy savings can be achieved when one or more physical assets operate in various operating states during operating time and idle time. For example, during different operations of a manufacturing process, various components of a sub-fab facility may not be necessary and can therefore be placed in a sleep, idle, hibernate, or off state, depending on how soon those components may be needed. Examples of power-saving low-power states include the idle state, the sleep state, and the hibernate state. The main differences between the three power-saving states lie in the duration and the energy consumption. Deeper levels of idle mode energy savings, such as sleep or hibernate, require longer periods of time to recover from the energy-saving mode to achieve full-load production without affecting the quality or yield of the manufacturing process. Depending on the degree of deviation from the BKM chamber conditions related to the power-saving states of the sub-fab facility and the processing chamber, it may take seconds, minutes, or hours to restore the processing chamber and the associated sub-fab facility to the temperature and pressure of the best known method (BKM). The idle state typically lasts for seconds, the sleep state typically lasts for minutes, and the hibernate state typically lasts for hours.

[0144] The digital replica 400 can identify one or more operating / power states of the physical assets of a manufacturing system and determine the effect of using that power state in a given context (e.g., system hardware architecture, subsystem hardware architecture, processing one or more process recipes, performing certain predefined processes, etc.). For example, the digital replica 400 can be part of a digital twin that determines the effect of such a power state and determines the context of idle or full power or modulation before actually implementing a power adjustment to the manufacturing system.

[0145] Based on the operating requirements, the processing tool and the associated manufacturing system can have a variety of different power configurations. For example, after a manufacturing operation is completed, there may be a power configuration in which the processing tool is in the "off" state when various gas flow and abatement systems are operating at full capacity to perform a shutdown operation. For the purposes of this application, the term "low-power configuration" refers to any state in which one or more controllers instruct one or more components of the processing tool and / or the sub-fab of the manufacturing system to operate in a power-saving mode, such as different energy consumption levels during a specific process recipe operation, or a non-production idle operation mode, such as the idle, sleep, and hibernate states described above, or the off state.

[0146] In some embodiments, one or more support assets may provide support functions for more than one other entity asset. For example, the pumping of two processing chambers may be performed by a single pump. Alternating the operation of the support asset between two entity assets can reduce energy and overall environmental costs.

[0147] The digital replica 400 may predict environmental resource consumption data associated with operating one or more entity assets in one or more corresponding operating modes. The digital replica 400 may provide suggestions for reducing environmental consumption costs by suggesting that one or more entity assets utilize reduced power states, sleep mode states, hibernation states, and / or shared operating mode data during periods when the corresponding tool is in an idle state or when the demand for the entity asset is low.

[0148] In another example, the digital replica 400 may be configured to determine eco-efficiency data related to performing preventative maintenance (PM) and / or cleaning of the entity assets of a manufacturing system. The digital replica 400 may receive purge gas data, cleaning process data, preventative maintenance data, chamber recovery data, and / or process recipes to determine environmental resource consumption.

[0149] Substrate processing may include a series of processes for fabricating circuits in a semiconductor (e.g., a silicon wafer) according to a circuit design. These processes may be performed in a series of chambers. The successful operation of a modern semiconductor manufacturing facility may aim to facilitate a steady flow of wafers from one chamber to another during the process of forming circuits in the wafers. In processes that execute many substrate programs, the conditions of the processing chambers may degrade, resulting in the processed substrates not meeting the desired conditions or processing results (e.g., critical dimensions, processing uniformity, thickness dimensions, etc.).

[0150] The cleaning process data may indicate one or more parameters related to the cleaning process, such as cleaning duration, frequency, and / or etchant flow rate. The cleaning process may utilize certain environmental resources, such as cleaning materials, precursors, etchants, and / or other substances used to perform the cleaning procedure. For example, to ensure that the processing results of subsequent substrates meet threshold conditions (e.g., processing uniformity, critical dimensions, etc.), the cleaning procedure may be performed at a certain rhythm or frequency (e.g., after processing a large number of wafers). The cleaning frequency of the processing chamber can be adjusted (e.g., optimized) to identify the cleaning frequency at which the substrates processed by the chamber operating at that cleaning frequency schedule still meet the threshold conditions (e.g., minimum processing result requirements). For example, a multi-wafer cleaning procedure may be implemented to save environmental resources such as cleaning materials, precursors, etchants, and / or other substances used to perform the cleaning procedure. The digital replica 400 may receive the cleaning data and determine cleaning optimizations, such as updates to the cleaning duration, frequency, amount of cleaning agent used, etc.

[0151] Preventive maintenance data indicates one or more of the type, frequency, duration, etc. of one or more preventive maintenance procedures associated with one or more physical assets of a manufacturing system. Preventive maintenance procedures (e.g., chamber cleaning) are typically used as part of a chamber restoration process to return the state of a processing chamber to a state suitable for entering a substrate processing production mode (e.g., mass processing of substrates). A restoration procedure is typically used after a preventive maintenance procedure to prepare the chamber for production mode (e.g., “preheat” the chamber).

[0152] Chamber restoration data indicates one or more of the type, frequency, duration, etc. of one or more chamber restoration procedures associated with one or more physical assets of a manufacturing system. A conventional common restoration procedure is to dry the processing chamber. Chamber drying is a procedure that includes processing a series of substrates (e.g., blank silicon wafers) to restore chamber conditions suitable for a substrate processing process (e.g., coating the chamber walls) (e.g., the substrates processed in the chamber have process results that meet desired threshold criteria). After chamber drying, the chamber can be operated in production mode for a period of time until another round of preventive maintenance and further chamber drying is required, or until it is recommended to otherwise restore the state of the processing chamber.

[0153] Purge gas data may indicate the type, quantity, frequency flow rate, cleaning duration of the purge gas. A digital replica can determine the effect of changing one or more operating parameters associated with the purge gas used. For example, the digital replica 400 can predict an update to environmental resource consumption based on a purge procedure that switches to using an alternative purge gas type such as H2, N2, clean dry air (CDA), etc.

[0154] In another example, the digital replica 400 can be configured to determine eco-efficiency data associated with one or more operating states of a physical asset of a manufacturing system. The digital replica 400 can receive coolant loop configuration data, process chilled water (PCW) data, ambient air data, and / or process recipes, and determine environmental resource consumption data therefrom.

[0155] Processing chambers used in substrate processing typically include many internal components that are repeatedly heated and cooled during and after processing. In some cases, for example, when routine service or maintenance is required after processing in a processing chamber, the components are cooled to approximately room temperature. In temperature-controlled components, such as a processing chamber showerhead having coolant channels, to cool the component from its normal operating temperature (e.g., approximately 90 degrees Celsius), the heat source heating the component can be turned off, and coolant is passed through the coolant channels to extract heat from the component.

[0156] Coolant loop configuration data indicates one or more geometries of one or more coolant loops configured to extract heat from one or more physical assets of a manufacturing system. The one or more coolant loops may operate in parallel such that multiple loops cool a common area of the physical assets. The one or more coolant loops may cool multiple physical assets in series with each other. Process chilled water (PCW) data indicates one or more parameters of a coolant substance, such as the type, flow rate, temperature of the coolant (e.g., process chilled water (PCW)). The digital replica may include a heat flow model indicating where heat can be transferred in an environment of a manufacturing system utilizing one or more coolant loops. The digital replica 400 may identify modifications to physical assets (e.g., chambers, chamber walls, chamber systems) that direct heat to the cooling loops and the associated eco-efficiency savings by directing heat to the cooling loops. When PCW modulation occurs, the digital replica 400 may further determine the effect of the processing result. PCW modulation may involve changing the flow rate within the cooling loop to alter heat exchange within the physical assets of the manufacturing system.

[0157] Figure 5 FIG. is an example diagram of operating parameter limits 500 for a manufacturing process operation according to some embodiments of the present disclosure. Various manufacturing process operations may include operating parameter limits 500 that indicate a process parameter window 510 or a set of values (e.g., value combinations) for a set of corresponding parameters when a threshold condition (e.g., minimum quality condition, target condition, etc.) is met to obtain a result. For example, the process parameter window 510 may include a first parameter 502 (e.g., the first flow rate of a first gas) and a second parameter 504 (e.g., the temperature of the gas). To perform a manufacturing process and meet a threshold condition (e.g., minimum quality standard, statistical process control (SPC) limits, specification limits, substrate process target, etc.), a process parameter value window 510 is determined that may identify value combinations of parameters that make it possible for a product to meet the threshold condition. As Figure 5 shown, the process parameter window 510 includes a lower limit 506A and an upper limit 506A of the first parameter 502, and a lower limit 508B and an upper limit 508A of the second parameter.

[0158] Optimizations identified by a manufacturing process system (e.g., by using suggestions 284, data analyzer 266, etc.) may include determining an eco-optimized process parameter window 512 within the process parameter window 510 that enables a manufacturing operation to consume fewer resources compared to process parameter values outside the eco-optimized process parameter window 512.

[0159] Note thatFigure 5 Illustrated are a simplified process parameter window 510 and an eco-optimized process parameter window 512 that depend only on two parameters 502, 504. Both the process parameter window 510 and the eco-optimized process parameter window 512 form simple rectangles. The process parameter window may include more than two parameters and may include more different parameter dependencies. For example, non-linear, physics-based, statistical, and / or empirical relationships between the parameters may result in non-linear process parameter windows and eco-optimized process parameter windows.

[0160] Figure 6 is a flowchart of a method 600 for generating a training dataset for training a machine learning model to perform cooling parameter evaluation, according to aspects of the present disclosure. Method 600 is performed by processing logic that may include hardware (circuitry, dedicated logic, etc.), software (such as software running on a general-purpose computer system or a dedicated machine), firmware, or some combination of the foregoing. In one implementation, method 600 may be performed by a computer system, such as Figure 3 the computer system architecture 300. In other or similar implementations, one or more operations of method 600 may be performed by one or more other machines not shown in the figures. In some aspects, one or more operations of method 600 may be performed by the dataset generator 374 of the machine learning system 370 referred to Figure 3 described.

[0161] For simplicity of illustration, method 600 is shown and described as a series of operations. However, the operations in accordance with the present disclosure may occur in various orders and / or concurrently with other operations not presented and described herein. Additionally, not all of the operations shown may be performed to implement method 600 in accordance with the disclosed objectives. Further, those skilled in the art will understand and appreciate that method 600 may alternatively be represented as a series of related states via a state diagram or events.

[0162] At block 610, the processing logic initializes the training set T to an empty set (e.g., {}). At block 612, the processing logic obtains substrate process recipe data related to processing a substrate in a processing chamber of a manufacturing system (e.g., process recipe setpoint data, process knob setpoint data, process pressure setpoint data, process temperature setpoint data, etc.). The process recipe data may include and / or consist of historical process recipe data (e.g., recipe data collected over time). In some implementations, the processing logic further obtains sensor data (e.g., temperature sensor data, pressure sensor data, energy sensor data, etc.) and / or predicted sensor data (e.g., data output from one or more additional models) related to processing the substrate in the processing chamber according to the process recipe and / or according to other process recipes.

[0163] At block 614, the processing logic obtains environmental resource usage information. The environmental resource usage information may include information related to the consumption of resources such as chemical precursors, gases, water, energy, etc. The environmental resource usage information may include and / or consist of historical environmental resource usage information (e.g., historical consumption data, etc.).

[0164] At block 616, the processing logic generates training inputs based on the process recipe data and / or sensor data obtained at block 612. In some embodiments, the training inputs may include a set of normalized recipe data.

[0165] At block 618, the processing logic may generate a target output based on the environmental resource usage information obtained at block 614. The target output may correspond to an environmental resource usage metric (data indicating resource consumption) for the process recipe being executed in the processing chamber.

[0166] At block 620, the processing logic generates an input / output map. The input / output map refers to the training inputs that include or are based on the process recipe data, and the target outputs of the training inputs, where the target outputs identify the predicted environmental resource consumption, and where the training inputs are related to (or mapped to) the target outputs. At block 622, the processing logic adds the input / output map to the training set T.

[0167] At block 624, the processing logic determines whether the training set T includes sufficient training data to train a machine learning model. It should be noted that in some embodiments, the sufficiency of the training set T may be determined solely based on the number of input / output maps in the training set, while in some other embodiments, in addition to or instead of the number of input / output maps, the sufficiency of the training set T may be determined based on one or more other criteria (e.g., a diversity metric of the training instances, etc.). In response to determining that the training set T includes sufficient training data to train a machine learning model, the processing logic provides the training set T to train the machine learning model. In response to determining that the training set does not include sufficient training data and the machine learning model cannot be trained, method 600 returns to block 612.

[0168] At block 626, processing logic provides a training set T to train a machine learning model. In some embodiments, the training set T is provided to a training engine 382 and / or a servo machine 392 of the machine learning system 370 to perform training. In the case of a neural network, for example, input values of an input / output map (e.g., recipe data and / or cooling parameter data) are input into the neural network, and output values of the input / output map are stored in output nodes of the neural network. Subsequently, connection weights, layers, and / or hyperparameters in the neural network are adjusted according to a learning algorithm (e.g., backpropagation, etc.), and this procedure is repeated for other input / output maps in the training set T. After block 626, the machine learning model 390 can be used to provide predicted environmental resource usage (e.g., predicted data indicating resource consumption) for process recipe operations performed in a processing chamber.

[0169] Figure 7 FIG. is a flow chart that illustrates an embodiment of a method 700 for training a machine learning model to estimate cooling parameter values of a process recipe executed in a processing chamber according to aspects of the present disclosure. The method 700 is performed by processing logic that may include hardware (circuitry, dedicated logic, etc.), software (such as software running on a general-purpose computer system or a dedicated machine), firmware, or some combination of the foregoing. In one embodiment, the method 700 may be performed by a computer system, such as Figure 3 the computer system architecture 300. In other or similar embodiments, one or more operations of the method 700 may be performed by one or more other machines not shown in the figures. In some aspects, one or more operations of the method 700 may be performed by a training engine 382 of the machine learning system 370 referred to in Figure 3 the description.

[0170] For simplicity of illustration, the method 700 is shown and described as a series of operations. However, the operations according to the present disclosure may occur in various orders and / or concurrently with other operations not presented and described herein. Additionally, not all operations shown may be performed to implement the method 700 according to the disclosed objectives. Further, those skilled in the art will understand and appreciate that the method 700 may alternatively be represented as a series of related states via a state diagram or events.

[0171] At block 702 of method 700, processing logic collects a training data set, which may include data from multiple substrate process recipes (e.g., process recipe set points, process control knob set points, etc.). The training data set may further include sensor data related to the performance of the substrate process recipes. Each data item in the training data set may include one or more labels. The data items in the training data set may include input level labels indicating the use of environmental resources related to the substrate process recipes. For example, some data items may include labels for the use of resources related to the process recipe (e.g., the use of one or more resources such as chemical precursors, gases, energy, etc.).

[0172] At block 704, data items from the training data set are input into an untrained machine learning model. At block 706, the machine learning model is trained based on the training data set to produce a trained machine learning model that estimates environmental resource use (e.g., resource consumption, etc.) for processing a substrate in a processing chamber according to a process recipe. The machine learning model may also be trained to output one or more other types of predictions, classifications, decisions, etc.

[0173] In one embodiment, at block 710, the input of the training data item is input into the machine learning model. The input may include substrate process recipe data (e.g., substrate process recipe) indicating one or more process recipe set points. In some embodiments, the data may be input as a feature vector. At block 712, the machine learning model processes the input to produce an output. The output may include environmental resource use (e.g., the consumption of one or more resources, etc.). The environmental resource use may be the estimated environmental resource use for processing a substrate according to the process recipe.

[0174] At block 714, the processing logic compares the predicted environmental resource use data of the output with the known environmental resource use related to the input. At block 716, the processing logic determines an error based on the difference between the output and the known environmental resource use. At block 718, the processing logic adjusts the weights of one or more nodes, one or more layers in the machine learning model, and / or one or more hyperparameters of the machine learning model based on the error.

[0175] At block 720, the processing logic determines whether a stop criterion is met. If the stop criterion is not met, the method returns to block 710 and another training data item is input into the machine learning model. If the stop criterion is met, the method proceeds to block 725 and the training of the machine learning model is completed.

[0176] In one embodiment, one or more ML models are trained to be applied to multiple processing chambers, which may be processing chambers of the same type or model. The trained ML model can then be further tuned for specific situations of the processing chamber. Further tuning can be performed by using additional training data items including substrate process recipes that can be executed in the processing chambers under discussion. Such tuning can address chamber mismatches between the chambers and / or specific hardware processing kits of some of the processing chambers. Additionally, in some embodiments, after maintenance of the processing chamber and / or one or more changes to the processing chamber hardware, further training is performed to tune the ML model of the processing chamber.

[0177] Figure 8A FIG. 800A is a flow chart of a method 800A for obtaining recommendations for processing substrates, according to some embodiments of the present disclosure. Method 800A is performed by processing logic that may include hardware (circuitry, special logic, etc.), software (such as software running on a general-purpose computer system or a special-purpose machine), firmware, or some combination of the foregoing. In one embodiment, method 800A may be performed by a computer system, such as Figure 3 computer system architecture 300. In other or similar embodiments, one or more other machines not shown in the figures may perform one or more operations of method 800A. In some embodiments, one or more operations of method 800A may be performed by the eco-efficiency module 129 described with reference to Figure 1 FIG. In some aspects, one or more operations of method 800A may be performed by one or more components of the server 320 described with reference to Figure 3 FIG.

[0178] For simplicity of illustration, method 800A is shown and described as a series of operations. However, the operations in accordance with the present disclosure may occur in various orders and / or concurrently with other operations not presented and described herein. Additionally, not all of the operations shown may be performed to implement method 800A in accordance with the disclosed objectives. Further, those skilled in the art will understand and appreciate that method 800A may alternatively be represented as a series of related states via a state diagram or events.

[0179] At block 802, the processing logic receives a process recipe that includes process recipe setpoint data. The process recipe can be used to process substrates in a processing chamber of a manufacturing system. In some embodiments, the processing logic receives multiple process recipes, each process recipe including recipe setpoint data. For example, a first set of process recipe setpoint data can indicate the setpoints of a first process recipe, while a second set of process recipe setpoint data can indicate the setpoints of a second process recipe. The processing logic can receive the first set and the second set of data. Similarly, a third set of process recipe setpoint data can indicate the setpoints of a third process recipe. The processing logic can receive the first set, the second set, and the third set. In some embodiments, the process recipe setpoint data includes predicted setpoint data output from a model (e.g., Figure 2B process model 262), which is configured to predict recipe setpoint data based on input process goals. The recipe setpoint data can indicate one or more process recipes that can be executed (e.g., in a processing chamber) to process substrates that meet the process goals.

[0180] The processing logic optionally receives sensor data related to substrate processing in the processing chamber. In some embodiments, the sensor data indicates conditions (such as temperature, pressure, precursor flow rate, gas flow rate, etc.) in the processing chamber during substrate processing. In some instances, the sensor data indicates conditions within the operating range of the processing chamber. The sensor data can be related to processing a first substrate according to a first process recipe, a second substrate according to a second process recipe, and / or a third substrate according to a third process recipe, etc. In some embodiments, the sensor data indicates the physical boundaries of the processing chamber conditions and / or the process recipes being executed in the processing chamber. For example, the sensor data can indicate the normal range (e.g., normal temperature range, normal pressure range, etc.) of the processing chamber conditions during substrate processing according to one or more process recipes.

[0181] At block 804, the processing logic inputs the process recipe received at block 802 into one or more machine learning models. In some embodiments, the one or more machine learning models are trained to predict eco-efficiency data (e.g., Figure 2B chamber model 264). In some embodiments, the one or more machine learning models are trained with training input data that includes historical process recipe data (e.g., recipe setpoint data, recipe setpoints, etc.) and training target output data that includes historical environmental resource usage data (e.g., resource consumption data, etc.).

[0182] In some embodiments, one or more machine learning models include a "chain" of machine learning models. For example, a first machine learning model can be trained to output first prediction data, and the first prediction data can be used to train a second machine learning model to output second prediction data. In some embodiments, the first machine learning model is trained with training input data including historical process recipes and training target output data including historical sensor data (e.g., historical sensor measurement data related to substrate processing in a processing chamber). One or more process recipes can be input into the first machine learning model to obtain predicted measurement values (e.g., predicted measurement data, predicted sensor measurement data, etc.) corresponding to the input process recipe at block 805A.

[0183] In some embodiments, the second machine learning model is trained with the predicted measurement values output by the first machine learning model. The second machine learning model can further be trained with training input data including historical process recipes and training output data including historical eco-efficiency data. The second machine learning model can be trained to output predicted eco-efficiency data (e.g., predicted environmental resource usage data). At block 805B, one or more process recipes and predicted measurement values output from the first machine learning model can be input into the second machine learning model to determine the predicted environmental resource usage data.

[0184] In some embodiments, at block 804, a plurality of process recipes are input into one or more machine learning models, each process recipe including a corresponding set of recipe set point data.

[0185] In some embodiments, one or more trained machine learning models are trained to output predicted environmental resource usage data. The predicted environmental resource usage data may indicate the environmental resource consumption associated with processing a substrate in a processing chamber according to a process recipe. For example, the predicted environmental resource usage data may indicate the predicted consumption of one or more resources used to process a substrate according to a process recipe. In some embodiments, the predicted environmental resource usage data indicates the consumption of a particular resource (e.g., chemical precursor, water, etc.) and / or the consumption of multiple resources. In some embodiments, the predicted environmental resource usage data includes multiple sets of environmental resource usage data. For example, one or more trained machine learning models may output a set of environmental resource usage data for each corresponding process recipe input into the one or more models. In such an instance, a set of environmental resource usage data may indicate that a corresponding process recipe is more eco-efficient than another process recipe. Specifically, the set of environmental resource usage data may indicate that the corresponding process recipe uses fewer resources to execute the recipe compared to other recipes input into the one or more models. In some instances, one or more models output predicted first environmental resource usage data corresponding to a first process recipe and predicted second environmental resource usage data corresponding to a second process recipe. In some embodiments, the environmental resource usage data may include time-series data from which resource consumption can be determined.

[0186] At block 808, the processing logic determines a recommendation associated with processing a substrate according to a process recipe based on a comparison of the predicted first environmental resource usage data and the predicted second environmental usage data. In some embodiments, the processing logic compares the predicted resource consumption associated with a process recipe (e.g., indicated by the first environmental resource usage data) and the predicted resource consumption associated with another process recipe (e.g., indicated by the second environmental resource usage data). The predicted resource consumption may indicate that one process recipe is more eco-efficient than another. This can be determined through comparison. In some embodiments, the processing logic compares multiple sets of environmental resource usage data to determine the process recipe with the highest eco-efficiency. For example, the processing logic may determine the process recipe with the highest eco-efficiency from a subset of multiple process recipes based on the corresponding predicted (e.g., through model prediction) resource consumption.

[0187] In some embodiments, the recommendation indicates that the most ecologically efficient process recipe will be implemented for processing the substrate to meet process goals (e.g., post - processed substrate goals, etc.). For example, the recommendation can be a selection of the predicted most ecologically efficient process recipes from a plurality of process recipes (e.g., receiving its set - point data at block 802). In some embodiments, the recommendation indicates a modification to the process recipe to make the process recipe more ecologically efficient. For example, the recommendation can indicate a change in the recipe set - point to reduce the resource consumption of the process recipe. In some instances, the processing logic can use predicted resource consumption data corresponding to other process recipes to determine the modification. The recommendation can optimize the ecological efficiency of the process recipe by changing the recipe set - point to more closely match another process recipe with higher predicted ecological efficiency (e.g., indicated by lower predicted resource consumption).

[0188] At block 810, the processing logic outputs a recommendation related to the process recipe. In some embodiments, the recommendation is output to the system controller for implementation in substrate processing. For example, the system controller can implement the process recipe indicated by the recommendation for processing the substrate in the processing chamber (e.g., the most ecologically efficient process recipe generated after comparison). In another example, the system controller can modify the process recipe according to the recommendation to form a more ecologically efficient process recipe and / or increase the ecological efficiency of the process recipe.

[0189] Figure 8B is a flowchart of method 800B for obtaining predicted process recipe set - point data according to some embodiments of the present disclosure. Method 800B is executed by processing logic that can include hardware (circuitry, dedicated logic, etc.), software (such as software running on a general - purpose computer system or a dedicated machine), firmware, or some combination of the foregoing. In one embodiment, method 800B can be executed by a computer system, such as Figure 3 the computer system architecture 300. In other or similar embodiments, one or more other machines (not shown in the figures) can execute one or more operations of method 800B. In some embodiments, one or more operations of method 800B can be executed by the ecological efficiency module 129 as described with reference to Figure 1 In some aspects, one or more operations of method 800B can be executed by one or more components of the server 320 as described with reference to Figure 3 In some embodiments, method 800B is executed in conjunction with method 800A.

[0190] For simplicity of explanation, method 800B is illustrated and described as a series of operations. However, the operations in accordance with the present disclosure can be performed in various orders and / or concurrently with other operations not presented and described herein. Additionally, not all of the operations shown may be performed to implement method 800B in accordance with the disclosed objectives. Further, those skilled in the art will understand and appreciate that method 800B can alternatively be represented as a series of related states via a state diagram or events.

[0191] At block 822, the processing logic receives target data, the target data including target substrate conditions related to a processed substrate. In some embodiments, the target conditions indicate one or more characteristics of the target processed substrate (e.g., surface characteristics, coatings, etc.). In some instances, the target substrate process data indicates the specifications of the processed substrate. The specifications can indicate thresholds for an acceptable processed substrate.

[0192] At block 824, the processing logic inputs the target conditions into a model (e.g., Figure 2B process model 262). The model can include one or more additional models in addition to the models described in reference to Figure 8A method 800A. In some embodiments, the model is an additional trained machine learning model. The additional machine learning model can be trained (to form the additional trained machine learning model) using training inputs including historical process target data and training target output data, the historical process target data including historical target conditions and the training target output data including historical process recipes (e.g., including historical process recipe set point data). In some instances, the historical target conditions include a plurality of historical target conditions. The historical conditions can indicate one or more historical characteristics of the historical target processed substrate. In some instances, the historical target conditions indicate the historical specifications of the historical processed substrate. The historical specifications can indicate historical thresholds for an acceptable historical processed substrate. In some embodiments, the additional machine learning model is trained to output a predicted process recipe related to the process target input into the model. For example, a process target can be input into the additional trained machine learning model and one or more predicted process recipes can be output from the model, the process recipes producing a processed substrate that meets the process target.

[0193] At block 826, the processing logic receives a first process recipe and a second process recipe as outputs of the model. The first process recipe and / or the second process recipe may each correspond to the target substrate process data received at block 822. In some embodiments, the model outputs additional process recipes (e.g., additional process recipe setpoint data sets). The process recipes output by the model, when executed, may each produce a substrate that meets the target substrate conditions. For example, the model may output a first process recipe and a second process recipe. When a substrate is processed in a processing chamber according to the first process recipe or the second process recipe, the processed substrate will meet the target conditions indicated by the target conditions. In some embodiments, one or more of the process recipes received at block 826 correspond to the process recipes received at block 802 in method 800A.

[0194] Figure 8C FIG. 800C is a flow diagram of a method 800C for obtaining predicted process recipe setpoint data according to some embodiments of the present disclosure. Method 800C is performed by processing logic that may include hardware (circuitry, dedicated logic, etc.), software (such as software running on a general-purpose computer system or a dedicated machine), firmware, or some combination of the foregoing. In one embodiment, method 800C may be performed by a computer system, such as Figure 3 computer system architecture 300. In other or similar embodiments, one or more other machines not shown in the figures may perform one or more operations of method 800C. In some embodiments, one or more operations of method 800C may be performed by an eco-efficiency module 129 as referred to in Figure 1 FIG. 800C. In some aspects, one or more operations of method 800C may be performed by one or more components of a server 320 as referred to in Figure 3 FIG. 800C. In some embodiments, method 800C is performed in conjunction with method 800A.

[0195] For simplicity of illustration, method 800C is shown and described as a series of operations. However, the operations in accordance with the present disclosure may occur in various orders and / or concurrently with other operations not presented and described herein. Additionally, not all of the operations shown may be performed to implement method 800C in accordance with the disclosed objectives. Further, those skilled in the art will understand and appreciate that method 800C may alternatively be represented as a series of interrelated states via a state diagram or events.

[0196] At block 832, processing logic trains a first machine learning model that is trained to output predicted measurement data based on a process recipe input to the first machine learning model. In some embodiments, the first machine learning model is the first model in a “chain” of machine learning models. The first machine learning model may be trained using training input data that includes historical process recipes (e.g., historical process recipe setpoint data). The first machine learning model may be further trained using training target output data that includes historical measurement data. The historical measurement data may include sensor data collected during processing of substrates in one or more processing chambers. In some instances, the historical measurement data may include measurements of current, voltage, power, flow rate, pressure, concentration, velocity, acceleration, and / or temperature. Similarly, the predicted measurement data output by the first machine learning model may include predicted measurements of current, voltage, power, flow rate, pressure, concentration, velocity, acceleration, and / or temperature. In some embodiments, the predicted measurements include predicted time series data of the measurements.

[0197] At block 834, processing logic trains a second machine learning model to output predicted environmental resource usage data (e.g., predicted eco-efficiency data). In some embodiments, the second machine learning model is the second model in a “chain” of machine learning models. In some embodiments, the second machine learning model is trained using training input data that includes predicted measurement data output from the first machine learning model. In some embodiments, the second machine learning model is trained using predicted time series measurements output from the first machine learning model. The training input data may further include the historical process recipes used to train the first machine learning model. In some embodiments, the second machine learning model is trained using training target output data that includes historical environmental resource usage data (e.g., historical eco-efficiency data). By using the output of the first machine learning model to train the second machine learning model, the accuracy of the predicted environmental resource data output by the second machine learning model may be increased. In some instances, using the intermediate output (e.g., predicted measurement data) from the first machine learning model to train the second machine learning model may provide higher accuracy for the final output (e.g., predicted environmental resource usage data) from the second machine learning model compared to predicting the final output using a single model. In some embodiments, the first machine learning model represents the behavior of a processing chamber that closely tracks changes in process recipe setpoints, while the second machine learning model additionally represents the behavior of a processing chamber that does not closely track changes in process recipe setpoints.

[0198] Depending on the situation, in some embodiments, the processing logic trains a third machine learning model to output further predicted environmental resource usage data (e.g., further predicted eco-efficiency data). In some embodiments, the third machine learning model is the third model in a "chain" of machine learning models. In some embodiments, the third machine learning model is trained using training input data that includes the predicted measurement data output from the first machine learning model. In some embodiments, the third machine learning model is trained using the predicted time-series measurements output from the first machine learning model. The training input data may further include the historical process recipes used to train the first machine learning model. In some embodiments, the third machine learning model is trained using training target output data that includes historical environmental resource usage data (e.g., historical eco-efficiency data) and the predicted environmental resource usage data output from the second machine learning model. Training the third machine learning model by using the output of the first machine learning model and / or the output of the second machine learning model can increase the accuracy of the predicted environmental resource data output by the third machine learning model (e.g., the further predicted environmental resource data output by the third machine learning model may be more accurate than the predicted environmental resource data output by the second machine learning model).

[0199] In block 836, the processing logic inputs a process recipe (e.g., data indicative of the process recipe, such as process recipe set points) into the trained second machine learning model. The trained second machine learning model can predict environmental resource usage data based on (e.g., corresponding to) the process recipe.

[0200] In block 838, the processing logic receives the predicted environmental resource usage data output from the second machine learning model. In some embodiments, the environmental resource usage data indicates the environmental resource consumption associated with processing a substrate according to the process recipe. In some embodiments, the predicted environmental resource usage data is time-series data indicative of resource consumption over time. For example, the second machine learning model can predict the power consumption of one or more components (e.g., heaters, etc.) of a processing chamber when processing a substrate according to the recipe input into the second machine learning model.

[0201] Figure 9A A graph of predicted environmental resource consumption data versus observed environmental resource consumption is shown in accordance with some embodiments of the present disclosure. Figure 9AThe graph shown may show predicted environmental resource consumption from a regression model (e.g., a trained machine learning model using one or more regression methods). In some embodiments, the predicted environmental resource consumption for a particular process recipe is within a threshold bounded by an upper limit 912 and a lower limit 914. Data points 908 (several of which are shown) may represent a comparison of predicted resource consumption (e.g., via one or more trained machine learning models as described herein) with actual observed resource consumption. In cases where the predicted resource consumption matches the actual resource consumption, the data points will lie on the dashed line 910, meaning that the predicted resource consumption is equal to the actual resource consumption. In cases where the data points are within the threshold bounded by 914 and 912, one or more of the machine learning models described herein may have sufficient accuracy to predict environmental resource consumption. In some embodiments, even when executed in the same processing chamber, a particular process recipe may vary in performance. Thus, the predicted environmental resource consumption may correspond to a possible average environmental resource consumption.

[0202] Figure 9B A graph showing predicted and actual temporal environmental resource consumption data 950 is shown in accordance with some embodiments of the present disclosure. The solid line 952 may represent actual resource consumption data, and the dashed line 954 may represent predicted resource consumption data. In some embodiments, the predicted temporal environmental resource consumption data corresponding to the dashed line 954 is output from a trained machine learning model. In some embodiments, the trained machine learning model (e.g., one or more trained machine learning models, multiple “daisy chained” trained machine learning models, etc.) is trained with actual resource consumption data, such as represented by the solid line 952. In some embodiments, physical constraints are used to inform the trained machine learning model. Additional machine learning models may predict additional resource consumption data based on the predicted temporal environmental resource consumption data represented by the dashed line 954. The consumption data may be predicted based on various inputs, such as substrate targets, process recipes, and / or historical training data. In some embodiments, eco-efficiency may be determined by using the temporal environmental resource consumption data 950. For example, in cases where energy consumption is represented in the data 950, the total energy and / or power consumption over time for a process recipe (e.g., area under the curve, etc.) may be calculated, and eco-efficiency data may be determined based on the energy and / or power consumption.

[0203] Figure 10 A block diagram of an exemplary computing device operating in accordance with one or more aspects of the present disclosure is depicted. In various illustrative examples, the various components of the computing device 1000 may represent the various components of the system controller 128, the computing device 250, the device executing the network client 220, etc.

[0204] The exemplary computing device 1000 may be connected to other computer devices in a LAN, an intranet, an extranet, and / or the Internet (e.g., by using cloud environments, cloud technologies, and / or edge computing). The computing device 1000 may operate in the capacity of a server in a client-server network environment. The computing device 1000 may be a personal computer, a set-top box (STB), a server, a network router, a switch, or a bridge, or any device capable of executing a set of instructions (sequentially or in other orders) that specify actions to be taken by that device. Additionally, although only a single exemplary computing device is shown, the term "computer" should also be understood to include any collection of computers that individually or jointly execute a set (or multiple sets) of instructions to perform any one or more of the methods discussed herein.

[0205] The exemplary computing device 1000 may include a processing device 1002 (also referred to as a processor or CPU), a main memory 1004 (e.g., read-only memory (ROM), flash memory, dynamic random access memory (DRAM), such as synchronous DRAM (SDRAM), etc.), a static memory 1006 (e.g., flash memory, static random access memory (SRAM), etc.), and auxiliary memory (e.g., data storage device 1018), and each of the foregoing may communicate with each other via a bus 1030.

[0206] The processing device 1002 represents one or more general-purpose processing devices, such as a microprocessor, a central processing unit, etc. More specifically, the processing device 1002 can be a complex instruction set computing (CISC) microprocessor, a reduced instruction set computing (RISC) microprocessor, a very long instruction word (VLIW) microprocessor, a processor implementing other instruction sets, or a processor implementing a combination of instruction sets. The processing device 1002 can also be one or more dedicated processing devices, such as an application specific integrated circuit (ASIC), a field programmable gate array (FPGA), a digital signal processor (DSP), a network processor, etc. According to one or more aspects of the present disclosure, the processing device 1002 can be configured to execute instructions implementing Figures 6 to 8B the methods 600 to 800B shown.

[0207] The exemplary computing device 1000 can further include a network interface device 1008, which can be communicatively coupled to the network 1020. The exemplary computing device 1000 can further include a video display 1010 (e.g., a liquid crystal display (LCD), a touch screen, or a cathode ray tube (CRT)), an alphanumeric input device 1012 (e.g., a keyboard), a cursor control device 1014 (e.g., a mouse), and a sound signal generating device 1016 (e.g., a speaker).

[0208] The data storage device 1018 can include a machine-readable storage medium (or more specifically, a non-transitory machine-readable storage medium) 1028, on which one or more sets of executable instructions 1022 are stored. For example, the data memory can be a physical memory in a computer room or remote, such as a cloud storage environment. According to one or more aspects of the present disclosure, the executable instructions 1022 can include executable instructions related to executing Figure 8A the method 800A and / or Figure 8B the method 800B. In one embodiment, the instructions 1022 include instructions for Figure 1 the eco-efficiency module 129.

[0209] During execution by the exemplary computing device 1000, the executable instructions 1022 may also be located, in whole or in part, within the main memory 1004 and / or the processing device 1002, which also constitute a computer-readable storage medium. The executable instructions 1022 may further be sent or received over a network via the network interface device 1008.

[0210] Although the computer-readable storage medium 1028 is shown as a single medium in Figure 10 FIG., the term "computer-readable storage medium" should be understood to include a single medium or multiple media (e.g., a centralized or distributed database, and / or associated caches and servers) that store one or more sets of operational instructions. The term "computer-readable storage medium" should also be understood to include any medium that is capable of storing or encoding a set of instructions executable by a machine, such that the machine performs any one or more of the methods described herein. Thus, the term "computer-readable storage medium" should be understood to include, but is not limited to, solid-state memory and optical and magnetic media.

[0211] Certain portions of the detailed descriptions above are presented in terms of algorithms and symbolic representations of operations on data bits within a computer memory. These algorithmic descriptions and representations are the means used by those skilled in the data processing arts to most effectively convey the substance of their work to others skilled in the art. An algorithm is here, and generally, conceived to be a self-consistent sequence of operations leading to a desired result. These operations require physical manipulations of physical quantities. Usually, though not necessarily, these quantities take the form of electrical or magnetic signals capable of being stored, transferred, combined, compared, and otherwise manipulated. For the sake of common usage, it is sometimes convenient to refer to these signals as bits, values, elements, symbols, characters, terms, numbers, or the like.

[0212] It should be borne in mind, however, that all of these and similar terms are to be associated with the appropriate physical quantities and are merely convenient labels applied to these quantities. Unless specifically stated otherwise, it will be apparent from the following discussion that throughout the description, discussions using terms such as "providing," "determining," "storing," "adjusting," "causing," "receiving," "comparing," "creating," "stopping," "loading," "copying," "throwing," "replacing," "executing," "outputting," etc., refer to the actions and processes of a computer system or similar electronic computing device that manipulates and transforms data represented as physical (electronic) quantities within the computer system registers and memories into other data similarly represented as physical quantities within the computer system memories or registers or other such information storage, transmission, or display devices.

[0213] Examples of the present disclosure also relate to apparatuses for performing the methods described herein. The apparatus may be specially constructed for the required purposes, or it may be a general-purpose computer system selectively programmed by a computer program stored in the computer system. Such a computer program may be stored in a computer-readable storage medium, such as, but not limited to, any type of disk, including optical disks, compact disc read only memory (CD-ROM), and magneto-optical disks, read-only memory (ROM), random access memory (RAM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), magnetic disk storage media, optical storage media, flash memory components, other types of machine-accessible storage media, or any type of medium suitable for storing electronic instructions, each coupled to the computer system bus.

[0214] The methods and displays presented herein are not inherently related to any particular computer or other apparatus. Various general-purpose systems may be used with programs in accordance with the teachings herein, or it may prove convenient to construct more specialized apparatuses to perform the required method operations. The structure required for various such systems will be set forth in the description below. In addition, the scope of the present disclosure is not limited to any particular programming language. It should be understood that a variety of programming languages may be used to implement the teachings of the present disclosure.

[0215] It should be understood that the foregoing description is intended to be illustrative, not restrictive. Many other embodiments will be apparent to those of skill in the art upon reading and understanding the above description. Although the present disclosure describes specific examples, it will be recognized that the systems and methods of the present disclosure are not limited to the examples described herein, but may be implemented with modifications within the scope of the appended claims. Accordingly, the specification and drawings are to be regarded as illustrative rather than restrictive. Thus, the scope of the present disclosure should be determined with reference to the full scope of the appended claims and their equivalents.

Claims

1. A method, the method comprising: Receiving a first process recipe including first process recipe set point data; Inputting the first process recipe into one or more trained machine learning models, the machine learning models outputting predicted first environmental resource usage data, the predicted first environmental resource usage data indicating a first environmental resource consumption associated with processing a substrate in a processing chamber according to the first process recipe; and Outputting a recommendation associated with the first process recipe at least in part based on the predicted first environmental resource usage data.

2. The method of claim 1, the method further comprising: Determining the recommendation based on a comparison of the predicted first environmental resource usage data with predicted second environmental resource usage data, wherein the predicted second environmental resource usage data indicates a second environmental resource consumption associated with processing the substrate in the processing chamber according to a second process recipe.

3. The method of claim 2, the method further comprising: Receiving target data including target substrate conditions of a processed substrate; Inputting the target data into one or more additional models; and Receiving the first process recipe and the second process recipe as outputs from the one or more additional models.

4. The method of claim 3, wherein the one or more additional models include a trained second machine learning model.

5. The method of claim 3, the method further comprising: Predicting, by a first additional model among the one or more additional models, one or more first measurements corresponding to the first process recipe; and Predicting, by a second additional model among the one or more additional models, one or more second measurements based on the first process recipe and the one or more first measurements output from the first additional model.

6. The method of claim 5, wherein the one or more first measurements and the one or more second measurements include predicted measurements of at least one of current, voltage, power, flow rate, pressure, concentration, speed, acceleration, or temperature.

7. The method of claim 1, wherein the predicted first environmental resource usage data includes predicted time series data related to a predicted behavior of the processing chamber during execution of the first process recipe.

8. The method of claim 1, wherein the recommendation includes a modification to the first process recipe to form a modified first process recipe, and wherein processing the substrate according to the modified first process recipe has a reduced environmental resource consumption compared to processing the substrate according to the first process recipe.

9. The method of claim 1, wherein the environmental resource usage data includes time series data of at least one of energy consumption, gas consumption, or water consumption related to substrate processing in the processing chamber.

10. A system, the system comprising: One or more processing chambers configured to process substrates, the one or more processing chambers including a plurality of sensors; and A system controller for controlling the one or more processing chambers, wherein the system controller is configured to: Receive a first process recipe including first process recipe set point data; Input the first process recipe into one or more trained machine learning models, which output predicted first environmental resource usage data indicative of a first environmental resource consumption associated with processing a substrate in a first processing chamber according to the first process recipe; and Output a recommendation associated with the first process recipe based at least in part on the predicted first environmental resource usage data.

11. The system of claim 10, wherein the system controller is further configured to: Determine the recommendation based on a comparison of the predicted first environmental resource usage data and predicted second environmental resource usage data, wherein the predicted second environmental resource usage data indicates a second environmental resource consumption associated with processing the substrate in the first processing chamber according to a second process recipe.

12. The system of claim 11, wherein the system controller is further configured to: Receive target data including target substrate conditions of the processed substrate; Input the target data into one or more additional models; and Receive the first process recipe and the second process recipe as outputs from the one or more additional models.

13. The system of claim 12, wherein the one or more additional models include a trained second machine learning model.

14. The system of claim 12, wherein the system controller is further configured to: Predict, via a first additional model of the one or more additional models, one or more first measurements corresponding to the first process recipe; and Predict, via a second additional model of the one or more additional models, one or more second measurements based on the first process recipe and the one or more first measurements output from the first additional model.

15. The system of claim 10, wherein the recommendation includes a modification to the first process recipe to form a modified first process recipe, and wherein processing the substrate according to the modified first process recipe has a reduced environmental resource consumption compared to processing the substrate according to the first process recipe.

16. A non-transitory machine-readable storage medium including instructions that, when executed by a processing device, cause the processing device to: Train a first machine learning model to form a trained first machine learning model, wherein the trained first machine learning model is trained to output predicted measurement data based on a process recipe input into the trained first machine learning model; and Train a second machine learning model using training data that includes the predicted measurement data output from the trained first machine learning model to form a trained second machine learning model, where the trained second machine learning model is trained to output predicted first environmental resource usage data, and the predicted first environmental resource usage data indicates environmental resource consumption associated with processing a substrate in a processing chamber according to the process recipe input into the trained second machine learning model.

17. The non-transitory machine-readable storage medium of claim 16, wherein the processing device is further configured to: Train a third machine learning model using training data that includes the predicted measurement data output from the trained first machine learning model and the predicted first environmental resource usage data output from the second machine learning model to form a third trained machine learning model, where the third machine learning model is trained to output predicted second environmental resource usage data, and the predicted second environmental resource usage data can indicate the environmental resource consumption associated with processing a substrate in a processing chamber according to the process recipe input into the third trained machine learning model.

18. The non-transitory machine-readable storage medium of claim 16, wherein the processing device is further configured to: Train an additional machine learning model using training input data that includes historical process target data and training target output data that includes historical process recipes to form an additional trained machine learning model, where the additional trained machine learning model is trained to output one or more predicted process recipes related to the process target input into the additional trained machine learning model.

19. The non-transitory machine-readable storage medium of claim 16, wherein the processing device is further configured to: Receive measurement data related to a plurality of process recipes, where the measurement data includes at least one measurement value of current, voltage, power, flow rate, pressure, concentration, speed, acceleration, or temperature; Receive environmental resource usage data corresponding to the plurality of process recipes, where the environmental resource usage data indicates environmental resource consumption associated with the plurality of process recipes; and Train one or more of the first machine learning model or the second machine learning model using one or more of the measurement data or the environmental resource usage data.

20. The non-transitory machine-readable storage medium of claim 16, wherein the processing device is further configured to: Receive a first process recipe that includes first process recipe setpoint data; Input the first process recipe into the trained second machine learning model; And Output a recommendation related to the first process recipe based at least in part on the predicted first environmental resource usage data associated with the first process recipe.

Images