Systems and methods for autonomous process control and optimization of semiconductor equipment using optical interferometry and reflectometry - Patents.com

Abstract

At least one laser sensor and controller are embedded in the substrate processing system in communication with a remote big data and machine learning server that transmits and receives data to and from the fleet of substrate processing systems for autonomous process control and optimization. The laser sensor is positioned proximate to a region of the substrate processing system and configured to capture first data from at least one of the edge-coupled rings and semiconductor substrates being transported from a processing chamber to the region / from the region to the processing chamber. The controller is configured to receive the first data from the laser sensor, process the first data to generate second data, transmit the second data to a remote server via the network, receive third data from the remote server via the network in response to transmitting the second data to the remote server, and operate the substrate processing system based on the third data for process optimization.

Description

[Technical field]

[0001] [CROSS REFERENCE TO RELATED APPLICATIONS] This application claims the benefit of U.S. Provisional Application No. 62 / 905,738, filed September 25, 2019. The entire disclosures of the above-referenced applications are incorporated herein by reference.

[0002] The present disclosure relates generally to substrate processing systems, and more particularly to systems and methods for autonomous process optimization and matching of semiconductor equipment using optical interferometry and reflectometry. [Background technology]

[0003] The background description provided herein is intended to provide a general overview of the contents of the present disclosure. Work by the currently named inventors within the scope of what is described in this Background section, as well as aspects of the description that may not otherwise be regarded as prior art at the time of filing, are not admitted, expressly or impliedly, as prior art against the present disclosure.

[0004] A substrate processing system typically includes multiple processing chambers (also called process modules) that perform deposition, etching, and other processing of substrates such as semiconductor wafers. Examples of processes that can be performed on a substrate include, but are not limited to, plasma-enhanced chemical vapor deposition (PECVD) processes, chemically enhanced plasma vapor deposition (CEPVD) processes, and sputtering physical vapor deposition (PVD) processes. Additional examples of processes that can be performed on a substrate include, but are not limited to, etching (e.g., chemical etching, plasma etching, reactive ion etching, etc.) and cleaning processes.

[0005] During processing, a substrate is placed on a substrate support, such as a pedestal, electrostatic chuck (ESC), etc., in a processing chamber of a substrate processing system. During deposition, a gas mixture containing one or more precursors is introduced into the processing chamber and a plasma is struck to activate a chemical reaction. During etching, a gas mixture containing an etching gas is introduced into the processing chamber and a plasma is struck to activate a chemical reaction. A computer-controlled robot typically transfers the substrates from one processing chamber to another in the order in which the substrates are processed. Summary of the Invention

[0006] A system and method are disclosed that includes at least one laser sensor and a controller embedded in a substrate processing system that communicates with a remote big data and machine learning server that transmits and receives data to and from a fleet of substrate processing systems for autonomous process control and process optimization. The laser sensor is positioned proximate to a region of the substrate processing system. The laser sensor is configured to capture first data from at least one of the edge bond rings and a semiconductor substrate being transported from a processing chamber to / from the region. The controller is configured to receive the first data from the laser sensor, process the first data to generate second data, transmit the second data to a remote server over the network, and in response to sending the second data to the remote server, receive third data from the remote server over the network and operate the substrate processing system based on the third data for process optimization.

[0007] In another feature, the controller is configured to determine a change in at least one geometric parameter of the edge coupled ring and the semiconductor substrate based on the first data, the second data being indicative of the change in the geometric parameter.

[0008] In another feature, the controller is configured to measure a change in thickness of at least one of the edge coupled ring and the semiconductor substrate based on the first data. The second data shows the change in thickness.

[0009] In other features, the controller is configured to measure a change in flatness of the edge bond ring based on the first data at least one location. The change in flatness is indicative of an etch rate uniformity of a process used to process a substrate in a processing chamber using the edge bond ring. The second data is indicative of the change in flatness. The controller is configured to adjust at least one parameter of the processing chamber and the process to control the etch rate based on the third data. The parameters include a temperature distribution in a substrate support assembly of the processing chamber in which the substrate is processed. The parameters include a height of the edge bond ring above a substrate support assembly of the processing chamber in which the substrate is processed. The parameters include recipe parameters selected from the group consisting of gas flow, pressure, upper electrode temperature, substrate support temperature distribution, bias voltage, etch / deposition rate, and etch / deposition uniformity.

[0010] In other features, the controller is configured to measure a change in a surface roughness of the edge bond ring based on the first data. The surface roughness is indicative of a variation in a quality of a process used to clean the processing chamber. The second data is indicative of the change in the surface roughness. The controller is configured to adjust parameters for controlling a recipe of the process based on the third data. The parameters include one or more of a duration and a frequency of the cleaning process.

[0011] In other features, the controller is configured to determine a profile of an inner diameter of the edge bond ring based on the first data, the profile being indicative of a malfunction of a component or assembly of the processing chamber, the second data being indicative of the profile, and the controller is configured to generate a message for maintaining the component or assembly based on the third data.

[0012] In still other features, the system includes a receiver and a controller. The receiver is configured to receive, via a network, first data captured by a laser sensor from at least one of an edge coupled ring and a semiconductor substrate in a substrate processing system. The controller is configured to process the first data to generate second data, input the second data into a model trained using machine learning, and in response to receiving the second data, output third data useful for operating the substrate processing system, and transmit the third data to the substrate processing system via the network.

[0013] In another feature, the third data includes adjustments for at least one of a processing chamber configuration of the substrate processing system and a process performed in the processing chamber.

[0014] In another feature, the controller is configured to determine a change in a geometric parameter of the edge coupled ring based on the first data, the second data being indicative of the change in the geometric parameter.

[0015] In another feature, the controller is configured to determine a change in thickness or weight of at least one of the edge bond ring and the semiconductor substrate based on the first data, the second data being indicative of the change in thickness or weight.

[0016] In other features, the controller is configured to measure a flatness of the edge bond ring based on the first data. A change in flatness at different locations is indicative of an etch rate uniformity of a process used to process the substrate in a processing chamber of the substrate processing system. The second data is indicative of the flatness at different locations of the edge bond ring. The third data includes adjustments to at least one parameter of a processing chamber configuration and a process to optimize the etch rate based on the second data. The parameters include a temperature distribution in a substrate support assembly of the processing chamber in which the substrate is processed. The parameters include a height of the edge bond ring above a substrate support assembly of the processing chamber in which the substrate is processed. The parameters include recipe parameters selected from the group consisting of gas flow, pressure, upper electrode temperature, substrate support temperature distribution, bias voltage, etch / deposition rate, and etch / deposition uniformity.

[0017] In other features, the controller is configured to measure a surface roughness of the edge bond ring based on the first data. The surface roughness is indicative of a variation in a quality of a process used to clean a processing chamber of the substrate processing system. The second data is indicative of a change in the surface roughness. The third data includes adjustments to parameters for controlling a recipe of the cleaning process based on the second data. The parameters include one or more of a duration and a frequency of the process.

[0018] In other features, the controller is configured to determine a profile of an inner diameter of the edge bond ring based on the first data, the profile being indicative of a malfunction of an assembly of a processing chamber of the substrate processing system, the second data being indicative of the profile, and the third data including a message for maintaining the assembly based on the second data.

[0019] In still other features, a substrate processing system includes a laser sensor and a controller. The laser sensor is disposed in the substrate processing system. The laser sensor is configured to capture first and second data from a component used in a processing chamber of the substrate processing system at a first and second time, respectively. The component includes at least one of a semiconductor substrate and an edge-coupled ring used with the semiconductor substrate. The component is used in the processing chamber during a process performed in the processing chamber between the first time and the second time. The controller is configured to receive the first and second data from the laser sensor and measure a change in a geometric parameter of the component based on the first and second data. The controller is configured to transmit the measured change to a remote server over a network. The remote server includes a model trained to recommend an adjustment to at least one parameter of the process and the processing chamber based on the measured change. The controller is configured to receive the adjustment to the parameter from the remote server over the network and implement the adjustment to at least one parameter of the process and the processing chamber.

[0020] In other features, the first data and the second data include measurements taken at a point, along a line, or across an area of ​​the component.

[0021] In other aspects, the geometric parameters include the thickness, flatness, or surface roughness of the component.

[0022] In other features, the controller is configured to determine a non-uniformity of at least one of an etch rate of the process and a deposition rate of the process based on the measured changes, the adjustment minimizing the non-uniformity of at least one of an etch rate of the process and a deposition rate of the process.

[0023] In other features, the controller is configured to first adjust the process parameters and then adjust the processing chamber parameters.

[0024] In other features, the controller is configured to first adjust the processing chamber parameters and then adjust the process parameters.

[0025] In other features, the process parameters are selected from the group consisting of gas flow, pressure, upper electrode temperature, substrate support temperature distribution, bias voltage, etch / deposition rate, and etch / deposition uniformity.

[0026] In other features, the processing chamber parameter is selected from the group consisting of a height of an edge bond ring above a substrate support assembly in the processing chamber, a slope of the edge bond ring, a gap between an upper electrode and a lower electrode of the processing chamber, an alignment between the upper electrode and the lower electrode, and a slope between the upper electrode and the lower electrode.

[0027] In other features, the measured changes indicate a variation in the quality of a process used to clean the processing chamber, and adjustments to parameters of the process minimize the variation, the parameters including one or more of a duration and a frequency of the process.

[0028] In other features, the controller is configured to determine a profile of an inner diameter of the edge bond ring based on the first data and the second data, the profile being indicative of a malfunction of an assembly of the processing chamber, the controller is configured to transmit the profile to a remote server, receive instructions from the remote server based on the profile for maintaining the assembly, and generate a message for maintaining the assembly based on the instructions.

[0029] In still other features, the system comprises a processor and a memory including instructions that, when executed by the processor, configure the processor to receive first and second data captured by a laser sensor from a component used in a processing chamber of a substrate processing system at a first and second time, respectively. The component includes at least one of a semiconductor substrate and an edge bond ring used with the semiconductor substrate. The component is used in the processing chamber during a process performed in the processing chamber between the first and second times. The instructions cause the processor to measure a change in a geometric parameter of the component based on the first and second data. The instructions cause the processor to input the measured change into a model trained to recommend an adjustment to at least one parameter of the process and the processing chamber based on the measured change. The instructions cause the processor to receive the adjustment to the parameter from the model and to implement the adjustment to at least one parameter of the process and the processing chamber.

[0030] In other features, the first data and the second data include measurements taken at a point, along a line, or across an area of ​​the component.

[0031] In other aspects, the geometric parameters include the thickness, flatness, or surface roughness of the component.

[0032] In other features, the instructions further configure the processor to determine a non-uniformity of at least one of an etch rate of the process and a deposition rate of the process based on the measured variations, the adjusting minimizing the non-uniformity of at least one of an etch rate of the process and a deposition rate of the process.

[0033] In other features, the instructions further configure the processor to first adjust the process parameters and then adjust the processing chamber parameters.

[0034] In other features, the instructions further configure the processor to first adjust the parameters of the processing chamber and then adjust the parameters of the process.

[0035] In other features, the instructions further configure the processor to select parameters of the process from the group consisting of gas flow, pressure, upper electrode temperature, substrate support temperature distribution, bias voltage, etch / deposition rate, and etch / deposition uniformity.

[0036] In other features, the instructions further configure the processor to select parameters of the process from the group consisting of a height of the edge bond ring above a substrate support assembly in the processing chamber, a slope of the edge bond ring, a gap between an upper electrode and a lower electrode of the processing chamber, an alignment between the upper electrode and the lower electrode, and a slope between the upper electrode and the lower electrode.

[0037] In other features, the measured changes indicate a variation in the quality of a process used to clean the processing chamber, and adjustments to parameters of the process minimize the variation, the parameters including one or more of a duration and a frequency of the process.

[0038] In other features, the instructions further configure the processor to determine a profile of an inner diameter of the edge bond ring based on the first data and the second data, the profile being indicative of a malfunction of an assembly of the processing chamber, the instructions further configure the processor to send the profile to a model, receive instructions from the model based on the profile for maintaining the assembly, and generate a message for maintaining the assembly based on the instructions.

[0039] Other areas of applicability of the present disclosure will become apparent from the detailed description, claims, and drawings. The detailed description and specific examples are for purposes of illustration only and are not intended to limit the scope of the present disclosure. [Brief description of the drawings]

[0040] The present disclosure will become more fully understood from the detailed description and the accompanying drawings, wherein:

[0041] [Figure 1] FIG. 1 illustrates an example of a substrate processing tool.

[0042] [Diagram 2] FIG. 2 is a diagram illustrating an exemplary arrangement of substrate processing tools in a fabrication facility.

[0043] [Figure 3A] FIG. 3A illustrates various exemplary configurations of a substrate processing tool. [Figure 3B] FIG. 3B illustrates various exemplary configurations of substrate processing tools. [Figure 3C] FIG. 3C illustrates various exemplary configurations of substrate processing tools. [Figure 3D] FIG. 3D illustrates various exemplary configurations of substrate processing tools. [Figure 3E] FIG. 3E illustrates various exemplary configurations of substrate processing tools. [Figure 3F] FIG. 3F illustrates various exemplary configurations of substrate processing tools. [Figure 3G] FIG. 3G illustrates various exemplary configurations of substrate processing tools. [Figure 3H] FIG. 3H illustrates various exemplary configurations of substrate processing tools. [Figure 3I] FIG. 3I illustrates various exemplary configurations of substrate processing tools.

[0044] [Figure 4] FIG. 4 illustrates another example of a substrate processing tool.

[0045] [Figure 5A] FIG. 5A is a plan view of various exemplary configurations of a substrate processing tool. [Figure 5B]FIG. 5B is a plan view of various exemplary configurations of substrate processing tools. [Figure 5C] 5A-5C are plan views of various exemplary configurations of substrate processing tools. [Figure 5D] FIG. 5D is a plan view of various exemplary configurations of substrate processing tools.

[0046] [Figure 6A] FIG. 6A is a plan view of a further exemplary configuration of a substrate processing tool. [Figure 6B] FIG. 6B is a plan view of a further exemplary configuration of a substrate processing tool. [Figure 6C] FIG. 6C is a plan view of a further exemplary configuration of a substrate processing tool.

[0047] [Figure 7A] FIG. 7A is a functional block diagram of a substrate processing system including various processing chambers. [Figure 7B] FIG. 7B is a functional block diagram of a substrate processing system including various processing chambers. [Figure 7C] FIG. 7C is a functional block diagram of a substrate processing system including various processing chambers.

[0048] [Figure 7D] FIG. 7D illustrates an example of an edge bond ring in a processing chamber.

[0049] [Figure 8A] FIG. 8A illustrates a high level architecture of a measurement system for performing autonomous process optimization and matching on semiconductor equipment using optical interferometry and reflectometry in accordance with the present disclosure.

[0050] [Figure 8B] FIG. 8B illustrates a client-server architecture for implementing the systems and methods of the present disclosure. [Figure 8C]FIG. 8C illustrates a client-server architecture for implementing the systems and methods of the present disclosure. [Figure 8D] FIG. 8D illustrates a client-server architecture for implementing the systems and methods of the present disclosure.

[0051] [Figure 9A] FIG. 9A is a diagram illustrating an exemplary arrangement of laser sensors for use in the measurement system of the present disclosure.

[0052] [Figure 9B] FIG. 9B illustrates an example of thickness changes of the edge bond ring and wafer due to etching and deposition processes. [Figure 9C] FIG. 9C illustrates an example of thickness changes of the edge bond ring and wafer due to etching and deposition processes. [Figure 9D] FIG. 9D shows an example of thickness changes of the edge bond ring and wafer due to etching and deposition processes. [Figure 9E] FIG. 9E illustrates an example of thickness changes of the edge bond ring and wafer due to etching and deposition processes. [Figure 9F] FIG. 9F illustrates an example of thickness changes of the edge bond ring and wafer due to etching and deposition processes. [Figure 9G] FIG. 9G illustrates an example of thickness changes of the edge bond ring and wafer due to etching and deposition processes.

[0053] [Figure 10A] FIG. 10A is a diagram illustrating an example of a laser sensor used with 1D / 2D reflectometry and 3D interferometry in the measurement system of the present disclosure. [Figure 10B] FIG. 10B is a diagram illustrating an example of a laser sensor used with 1D / 2D reflectometry and 3D interferometry in the measurement system of the present disclosure. [Figure 10C]FIG. 10C is a diagram illustrating an example of a laser sensor used with 1D / 2D reflectometry and 3D interferometry in the measurement system of the present disclosure.

[0054] [Figure 11] FIG. 11 is a block diagram of a measurement system implemented using the client-server architecture shown in FIGS. 8B-8D and using the laser sensor shown in FIGS. 9-10B.

[0055] [Figure 12] FIG. 12 is a flowchart of a method for training various models using machine learning based on data captured by a measurement system from various tools.

[0056] [Figure 13] FIG. 13 is a flowchart of a method implemented in a tool using a model trained based on data captured during manufacturing.

[0057] [Figure 14A] FIG. 14A is a flow chart of an additional model-based method implemented to improve tool operation using a measurement system. [Figure 14B] FIG. 14B is a flow chart of an additional model-based method implemented to improve tool operation using the measurement system. [Figure 14C] FIG. 14C is a flow chart of an additional model-based method implemented to improve tool operation using the measurement system. [Figure 14D] FIG. 14D is a flow chart of an additional model-based method implemented to improve tool operation using the measurement system. [Figure 14E] FIG. 14E is a flow chart of an additional model-based method implemented to improve tool operation using the measurement system. [Figure 14F]FIG. 14F is a flow chart of an additional model-based method implemented to improve tool operation using the measurement system. [Figure 14G] FIG. 14G is a flow chart of an additional model-based method implemented to improve tool operation using the measurement system. [Figure 14H] FIG. 14H is a flow chart of an additional model-based method implemented to improve tool operation using the measurement system. [Figure 14I] FIG. 14I is a flow chart of an additional model-based method implemented to improve tool operation using the measurement system.

[0058] [Figure 15A] FIG. 15A is a flow chart illustrating an example method for generating, training, and validating a model. [Figure 15B] FIG. 15B is a flowchart illustrating an example method for generating, training, and validating a model.

[0059] [Figure 16A] FIG. 16A is a graphical representation of an exemplary neural network used to generate the model. [Figure 16B] FIG. 16B is a graphical representation of an exemplary neural network used to generate the model.

[0060] In the drawings, reference numbers may be reused to refer to similar and / or identical elements. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0061] An edge coupling ring (an example of which is shown and described with reference to FIG. 7D below) is used with a substrate, such as a semiconductor wafer (hereafter wafer), during processing in a process module (PM). The edge coupling ring allows for better control of the plasma field in the PM during the etch process to achieve optimized on-wafer process results. The edge coupling ring can be automatically transferred from the tool's equipment front-end module (EFEM) to the tool's PM via a wafer handling robot in a vacuum transfer module (VTM). Due to differences in the conditions of the PM within a tool or between different tools, on-wafer process results tend to vary. To achieve inter-PM consistency (i.e., to achieve uniformity of on-wafer process results in PMs within a tool or between different tools), techniques are needed to quantify process characteristics (etch rate, etch uniformity, etc.) between different tools, monitor drift over time, record changes after maintenance (e.g., wet cleaning, etc.), and create procedures to automatically adjust parameters (e.g., recipe, ESC temperature, gas flow, duration, pressure, etc.) of the etch process in the PM to compensate for such changes.

[0062] Brand new (i.e., unused) edge bond rings have a non-uniform thickness in the radial direction. During processing performed in the PM, the edge bond rings erode and their thickness decreases. The thickness reduction at the inner diameter (ID) of the edge bond ring, close to the center of the edge bond ring, is typically greater than the thickness reduction at a point close to the outer diameter (OD) of the edge bond ring. Thus, one of the objectives of this disclosure is to determine the change in the thickness profile of the edge bond ring across the radial direction before and after processing. The thickness measurement can be performed, for example, using one or more laser sensors located adjacent to the airlock chamber of the tool. (Throughout this disclosure, the phrases "a laser sensor" and "the laser sensor" are used to mean one or more laser sensors.) However, the laser sensor does not have to be mounted in the airlock. A dedicated station can be designed in the VTM or EFEM to perform the measurements. The motivation for placing the laser sensor on the platform rather than in the PM is that the harsh environments in which PMs operate (plasma, high temperature, vacuum, corrosion, etc.) are very challenging and may damage the laser sensor, thereby increasing the cost of implementing such sensing technology within a PM.

[0063] Currently, no technique exists to measure the thickness change between a brand new edge bonded ring entering a PM and the same edge bonded ring leaving the PM at the end of its life or any time prior to the end of its life (e.g., daily) and automatically adjust the PM's process parameters (e.g., etch / deposition rate, etch / deposition uniformity, etc.) for process optimization and inter-PM matching. This disclosure proposes a technique that enables automated measurements of inbound / outbound (new / used) edge bonded rings (and pre-processed / post-processed wafers) and associated information processing infrastructure for autonomous process optimization and inter-PM matching. Using this technique, the edge bonded ring can be removed from the PM via robot and transported to an airlock midway through its life. After thickness profile measurements are made, the edge bonded ring can be returned to the PM and adjustments to system parameters can be made to optimize process results.

[0064] Additionally, the same techniques described for measuring the change in thickness profile of the edge bond ring can also be applied to wafers. For example, in an etching application, optical reflectometry can be used to measure the thickness of the wafer before and after processing to determine the thickness of the etched material. The thickness of the etched material can then be compared to the expected thickness of the etched material. Based on the difference between the actual thickness of the etched material and the expected thickness, the process parameters (e.g., etch / deposition rate, etch / deposition uniformity, etc.) can be adjusted to optimize the process results for the PM and the entire fleet. Examples of parameters that can be adjusted in the PM include, but are not limited to, pressure, gas flow, ESC temperature, top electrode temperature, process duration, bias voltage, etc. For a typical etching application, the thickness of the etched material can be 0.1-10 μm. A laser sensor can measure the thickness of the etched material with repeatability down to 0.1 μm via optical reflectometry and interferometry.

[0065] Additionally, the same techniques can be applied to deposition tools to measure the thickness of films deposited on wafers and adjust process parameters (e.g., etch / deposition rate, etch / deposition uniformity, etc.) based thereon to optimize process results for PMs and the entire fleet. Thus, throughout this disclosure, systems and methods are often described with reference only to edge bond rings, for example, but it should be understood that the teachings of this disclosure apply equally to wafers processed using etch and deposition processes, regardless of specific reference to wafers, unless otherwise stated. For example, some measurements, such as ID measurements, apply only to edge bond rings and not to wafers.

[0066] According to the present disclosure, laser sensors using optical reflectometry and interferometry are placed, for example, in the transfer module (TM) at locations where new / used edge bond rings and wafers pass (e.g., airlock, aligner, robot arm, etc.). This placement allows for accurate measurement of the change in thickness profile in the edge bond ring used (thickness reduction) and processed wafers (thickness reduction in etch process and thickness increase in deposition process). By knowing the usage history of the edge bond ring (e.g., number of RF hours the edge bond ring was used in the PM, process parameters, chamber type, etc.) and its thickness reduction, the process characteristics of the PM (e.g., etch / deposition rate, etch / deposition uniformity, etc.) can be determined and the process in the PM can be optimized. The reduction in thickness of the etched wafer can be used to adjust (optimize) the parameters of the etch process in the PM. The increase in thickness of the wafer used in the deposition process can be used to adjust (optimize) the parameters of the deposition process in the PM. Both optical reflectometry sensors operating in 1D (at a point) or 2D (across a line) as well as interferometric sensors operating in 3D (across a surface) can be used to perform height / thickness measurements as described with reference to Figures 9A-10C below.

[0067] In simple terms, a laser sensor measures changes in geometric parameters such as thickness of the edge bond ring (or wafer) and sends them to the tool's system computer. The system computer pre-processes the data and sends the pre-processed / post-processed data to the host / cloud (e.g., in a data center). The host / cloud contains the big data infrastructure and stores all the input data. The host / cloud determines the PM's process characteristics (e.g., etch / deposition rate, uniformity, etc.) based on the thickness changes between the incoming / outgoing edge bond rings (and wafers). The host / cloud trains machine learning models (e.g., based on neural networks, tensorflow, etc.) using data from similar PMs running the same recipe periodically (e.g., daily or weekly) using the data science infrastructure. The trained models are used with the input data to recommend adjustments to PM process parameters (e.g., etch / deposition rate, etch / deposition uniformity, etc.) to achieve autonomous process optimization and inter-PM alignment in a closed-loop manner. The system controller automatically adjusts (optimizes) the PM process parameters (eg, etch / deposition rate, etch / deposition uniformity, etc.) to achieve autonomous process optimization and inter-PM matching in a closed-loop manner.

[0068] Thus, the present disclosure relates to a system and method for capturing data from an edge bond ring and a wafer when they are temporarily stationed while being transported (e.g., in an airlock) to or from a PM. The captured data is processed and geometric properties of the edge ring and wafer are calculated via feature extraction. The data is captured and processed for a period of time to generate training data that is used to train various models using machine learning. The trained models are then used to quantify edge bond ring wear and wafer thickness changes over time and automatically adjust (optimize) process parameters (e.g., etch / deposition rate, etch / deposition uniformity, etc.) for PM-to-PM matching to improve system throughput, process yield, and on-wafer process quality as described below.

[0069] In addition, the system and method can be optionally complemented with various sensors, algorithms, and feedback mechanisms typically used to control the tool and recommend measures to improve throughput, yield, and on-wafer process quality. The system and method continue to learn from data captured from the edge-bonded ring and wafers continuously during use or production. The system and method also automates the performance of some preventative and corrective tasks that typically require human intervention (e.g., PM cleaning and signature-based guided troubleshooting described below). The system and method monitors fleet-wide tool operation in a semiconductor fab and provides feedback for autonomous control of the tools in a closed-loop manner.

[0070] The measurement system of the present disclosure comprises a detector and / or a sensor, such as a laser sensor (shown and described with reference to Figures 9A-10C below), and further comprises data processing algorithms, perception algorithms, sensor fusion algorithms, models trained using machine learning, and closed-loop control algorithms. Throughout this disclosure, a laser sensor is used only as an example of a detector / sensor, and it is understood that any other suitable detector / sensor may be used instead or in addition. These and other features of the present disclosure are described in detail below.

[0071] composition The present disclosure is organized as follows. First, numerous examples of substrate processing tools, their configurations, and processing chambers used therein are shown and described with reference to Figures 1-7D. These figures and corresponding descriptions are provided to explain the various processes that may be performed on different wafers, as well as the various paths traveled by the wafer and associated edge bond rings between different chambers and tools. Additionally, these figures are provided to illustrate the complexity, variety, and breadth of data that is typically collected from various sensors in chambers and tools. The data used to diagnose, correct, and prevent problems, as well as to perform various adjustments, calibrations, and preventive maintenance procedures in chambers and tools, can be used in conjunction with the systems and methods of the present disclosure.

[0072] The systems and methods of the present disclosure can operate standalone or in combination with these sensors and the data collected therefrom, which can significantly improve the diagnostic, corrective, and preventative procedures performed in chambers and tools. These figures provide insight into the operation of chambers and tools that is helpful in understanding the range of operations performed by the systems and methods of the present disclosure. Additionally, these figures are helpful in understanding the various locations where the laser sensors of the measurement systems of the present disclosure can be installed on different tools.

[0073] FIG. 8A shows a high level architecture of a measurement system according to the present disclosure. FIG. 8B-FIG. 8D show a client-server architecture that can be used to implement the system and method of the present disclosure. FIG. 9A shows examples of locations within a tool where one or more detectors, such as a laser sensor of the measurement system of the present disclosure, can be located. FIG. 9B-FIG. 9G show examples of edge bond rings and wafers where measurements are taken before and after processing according to the system and method of the present disclosure. FIG. 10A-FIG. 10C show examples of 1D, 2D, and 3D laser sensors. FIG. 11 shows an example of a measurement system implemented using the client-server architecture of FIG. 8B-FIG. 8D and using one or more laser sensors as shown in FIG. 9A-FIG. 10C.

[0074] FIG. 12 shows an example of a method for training various models using machine learning based on data captured by a metrology system from various tools. FIG. 13 shows an example of a method implemented using models trained based on data captured in real time (i.e., on the fly or when the tool is being used to process wafers during production). FIGs. 14A-14I show additional examples of model-based methods implemented to improve tool operation by optimizing fleet-wide process and PM alignment using the systems and methods of the present disclosure. FIGs. 15A-16B show examples of methods for training models using machine learning.

[0075] Tool examples 1 illustrates a top-down view of an example of a substrate processing tool 100. The substrate processing tool 100 includes multiple process modules 104. By way of example only, each of the process modules 104 may be configured to perform one or more respective processes on a substrate. A substrate to be processed is loaded into the substrate processing tool 100 via a port of a loading station of an equipment front-end module (EFEM) 108 and then transferred to one or more of the process modules 104. For example, a substrate may be loaded into each of the process modules 104 in succession.

[0076] FIG. 2 illustrates an example arrangement 200 of a fabrication chamber 204 including multiple substrate processing tools 208, such as substrate processing tool 100. FIGS. 3A-3I illustrate various examples of configurations in which multiple substrate processing tools 208, such as substrate processing tool 100, may be arranged. In these examples, the wafers move through various paths. Accordingly, the configuration and operation of the robots used to transport the wafers through these paths varies. Additionally, in these examples, various sensors are used to sense various aspects of the robots and wafers. Additionally, one or more detectors, such as laser sensors of a measurement system, may be mounted anywhere in these examples, as will be described in more detail with reference to FIGS. 9-11.

[0077] 3A illustrates a first exemplary configuration 300 including a first substrate processing tool 304 and a second substrate processing tool 308 in accordance with the principles of the present disclosure. The first substrate processing tool 304 and the second substrate processing tool 308 are sequentially arranged and connected by a transfer stage 312 that is under vacuum. As shown, the transfer stage 312 includes a pivoting transfer mechanism configured to transfer substrates between a vacuum transfer module (VTM) 316 of the first substrate processing tool 304 and a VTM 320 of the second substrate processing tool 308. In other examples, the transfer stage 312 may include other suitable transfer mechanisms, such as a linear transfer mechanism.

[0078] By way of example only, a first robot (not shown) of the VTM 316 may place a substrate on the support 324 located at a first position, the support 324 is pivoted to a second position, and a second robot (not shown) of the VTM 320 retrieves the substrate from the support 324 at the second position. In some examples, the second substrate processing tool 308 may include a storage buffer 328 configured to store one or more substrates between processing stages. The transfer mechanism may also be stacked to provide two or more transfer systems between the substrate processing tools 308 and 304. The transfer stage 324 may also have multiple slots to transport or buffer multiple substrates at a time. In the configuration 300, the first substrate processing tool 304 and the second substrate processing tool 308 are configured to share a single front-end equipment module (EFEM) 332.

[0079] 3B illustrates a second exemplary configuration 400 including a first substrate processing tool 404 and a second substrate processing tool 408 arranged sequentially and connected by a transfer stage 412. Configuration 400 is similar to configuration 300 of FIG. 3A, except that in configuration 400, the EFEM is eliminated. Thus, a substrate may be loaded directly into the first substrate processing tool 408 via an airlock loading station 416 (e.g., using a vacuum wafer carrier, a storage or transport carrier such as a Front Opening Unified Pod (FOUP), or other suitable mechanism). The placement of laser and other sensors in this configuration may correspondingly differ from configuration 300.

[0080] 3C illustrates a third exemplary configuration 500 including a substrate processing tool 504. Configuration 500 eliminates the EFEM, uses only a single loading station 508, and allows for a larger number (e.g., seven) of process modules 512. At the loading station 508, substrates can be loaded directly into the second substrate processing tool 408 via an airlock loading station 416 (e.g., using a vacuum wafer carrier, a storage or transport pod such as a Front Opening Unified Pod (FOUP), or other suitable mechanism). Thus, the placement of laser sensors and other sensors in this configuration may correspondingly differ from configurations 300 and 400.

[0081] 3D illustrates a fourth exemplary configuration 600 including a first substrate processing tool 604 and a second substrate processing tool 608 sharing a single EFEM 612. More specifically, the first substrate processing tool 604 and the second substrate processing tool 608 are connected to respective ends of the EFEM 612 via respective loading stations 616 and 620. The loading stations 616 and 620 may each have a stacked configuration that may include laser sensors and other suitable sensors.

[0082] 3E illustrates a fifth exemplary configuration 700 including a first substrate processing tool 704 and a second substrate processing tool 708 sharing a single EFEM 712. The first substrate processing tool 704 and the second substrate processing tool 708 are connected to respective ends of the EFEM 712 via respective loading stations 716 and 720. The loading stations 716 and 720 may each have a stacked configuration that may include laser sensors and other suitable sensors.

[0083] FIG. 3F illustrates a sixth exemplary configuration 800 including one or more rows of sequentially arranged substrate processing tools 804, 808, etc. In the configuration 800, each row may include three or more substrate processing tools connected via a respective transfer stage 812. The transfer stage 812 may include a pivoting transfer mechanism, a linear transfer mechanism, etc. A first EFEM 816 is provided at a first end of the row of substrate processing tools 804, 808, and a second EFEM 820 is provided at a second end of the row of substrate processing tools 804, 808. For example, a substrate may be loaded and processed in the first EFEM 816, sequentially transferred through various process modules of the substrate processing tools 804, 808, and then unloaded / recovered from the second EFEM 820. In some examples, the transfer mechanisms in the transfer stage 812 may be stacked vertically to provide two or more transfer systems between adjacent substrate processing tools. The transfer stage 812 may also have multiple slots to move or buffer multiple substrates at a time. As can be appreciated, the placement of laser sensors and other sensors in configuration 800 may vary significantly from that in other configurations.

[0084] 3G illustrates a seventh exemplary configuration 900 including a substrate processing tool 904. In configuration 900, the substrate processing tool 904 includes, for example, eight process modules 908 and eliminates both the EFEM and any external loading station. Instead, one or more transport carriers (e.g., vacuum wafer carriers) 912 are provided in an interior 916 of the substrate processing tool 904. For example, the carriers 912 can be transported from above the substrate processing tool 904 using an automated transport system, such as an automated material handling system (AMHS). A robot 920 retrieves the substrate from the carrier 912 and transfers the substrate to the process module 908. As can be appreciated, the placement of laser and other sensors in configuration 900 can differ from that in other configurations.

[0085] 3H shows an eighth exemplary configuration 1000 including a substrate processing tool 1004 having multiple process modules 1008. The substrate processing tool 1004 includes a linear VTM 1012 and a robot 1016 configured to transfer substrates between the EFEM 1020 and the process modules 1008. The VTM 1012 is configured to adjust the linear position of the robot 1016 relative to the process modules 1008 (i.e., adjusts the end-to-end position of the robot 1016 relative to the VTM 1012).

[0086] 3I illustrates a ninth exemplary configuration 1100 that includes a substrate processing tool 1104. The substrate processing tool 1104 includes a cluster arrangement of process modules 1108, and an optional end process module 1112. The process modules 1108 share a single EFEM 1116. Again, the arrangement of laser and other sensors in configurations 1000 and 1100 may differ from other configurations.

[0087] In some examples, any of the processing tools described herein can implement a loading station having a stacked configuration. For example, loading stations 508, 716, 720, etc., as shown in Figures 3C and 3E, can implement a stacked configuration. In other words, in a stacked configuration, the loading station can include two or more vertically stacked loading stations. In some examples, the stacked configuration can also include one or more loading stations and vertically stacked process modules or chambers (such as integrated critical strip (ICS) chambers). Additional laser sensors and other sensors can be used in these stacked configurations.

[0088] Adding a tool example 4 shows a top-down view of yet another example of a substrate processing tool 1150. The substrate processing tool 1150 includes multiple process modules 1154. For example, each of the process modules 1154 can be configured to perform one or more respective processes on a substrate. The substrate to be processed is loaded into the substrate processing tool 1150 via a port of a loading station of an atmospheric-to-vacuum (ATV) transfer module, such as an equipment front-end module (EFEM) 1158, and then transferred to one or more of the process modules 1154. For example, a transfer robot 1162 is arranged to transfer the substrate from the loading station 1166 to an airlock or load lock 1170, and a vacuum transfer robot 1178 of a vacuum transfer module 1174 is arranged to transfer the substrate from the load lock 1170 to the various process modules 1154.

[0089] For example, an equipment front-end module (EFEM) of a substrate processing tool may include one or more transfer robots for transferring substrates between the EFEM and a load lock located between the EFEM and a vacuum transfer module (VTM). The interior volume of the EFEM must be sufficient to accommodate the transfer robots. Thus, the load lock is typically located outside the footprint of the equipment front-end module (EFEM) between the EFEM and the VTM. In some examples, the EFEM may include a transfer robot having a configuration that allows the airlock to be located at least partially within the EFEM. The fabrication chamber 204 shown in FIG. 2 may include multiple substrate processing tools 1150.

[0090] 5A-6C show various examples of configurations in which multiple substrate processing tools, such as substrate processing tool 1150, may be arranged. In these examples, the wafers move through various paths. Thus, the configuration and operation of the robots used to transport the wafers through these paths varies. Furthermore, in these examples, various sensors are used to sense various aspects of the robots and wafers. Additionally, one or more detectors, such as laser sensors of a measurement system, may be mounted anywhere in these examples, as will be described in more detail with reference to FIGS. 9A-11.

[0091] 5A-5D show plan views of an exemplary configuration of a first substrate processing tool 1200-1, a second substrate processing tool 1200-2, and a third substrate processing tool 1200-3 (collectively, substrate processing tools 1200). Similar to the substrate processing tool 1150, each of the substrate processing tools 1200 includes a modified front end equipment module (EFEM) 1204 configured to house at least a portion of a load lock 1208. In other words, instead of being located outside the EFEM 1204 in the gap between the EFEM 1204 and the vacuum transfer module (VTM) 1212, the load lock 208 extends into the interior of the EFEM 1204.

[0092] Thus, the EFEM 1204 can be located closer to the VTM 1212, reducing the overall footprint and increasing the pitch of the substrate processing tool 1200. For example, the transfer robot 1216 of the EFEM 1204 is located closer to the loading station 1220 on the front wall (e.g., first side) than the back wall 1224 (e.g., second side) of the EFEM 1204 to provide space for the load lock 1208 to extend into the interior of the EFEM 1204. In some examples, the load lock 1208 may be configured as shown in the alternative arrangement of the substrate processing tool 1200-3 in FIG. 5D. By way of example only, the loading station 1220 may accommodate a front opening unified pod (FOUP).

[0093] As shown, the substrate processing tool 1200 includes six process modules 1228. However, other configurations of the substrate processing tool 1200 may include more than seven process modules 228. For example, the length of the VTM 1212 may be extended to accommodate the additional process modules 1228. Similarly, the VTM 1212 may include vacuum transfer robots 1232 having various configurations. For example, the substrate processing tool 1200-1 includes three vacuum transfer robots 1232 and the substrate processing tool 1200-2 includes two vacuum transfer robots 1232. In the substrate processing tools 1200-1 and 1200-3, the robots 1232 are aligned with a central longitudinal axis of the VTM 1212.

[0094] Conversely, the substrate processing tool 1200-3 includes a single vacuum transfer robot 1232 that is positioned off-center (i.e., shifted right or left toward the process module 1228) relative to the central longitudinal axis of the 1VTM 212. In other words, the primary pivot point of the robot 1232 is off-center. Although shown as having one or two arms, each of the robots 1216 and 1232 may have a configuration that includes one, two, or more arms. In some examples, the robot 1232 may include two end effectors 1234 on each arm, as shown in Figures 5C and 5D.

[0095] The substrate processing tool 1200 may include one or more storage buffers 236 configured to store one or more substrates between processing stages. In some examples, a storage buffer 1240 may be located within the VTM 1212. In some examples, one or more of the storage buffers 1236 may be replaced by a process module or other component.

[0096] In some examples, one or more of the EFEM 1204, the load locks 1208, the VTMs 1212, and the process modules 1228 may have a stacked configuration. For example, each of the process modules 1228 may correspond to two process modules 1228 in a vertically stacked configuration (i.e., one process module 1228 is disposed above / below the other), the VTMs 1212 may correspond to two VTMs 1212 in a vertically stacked configuration, each of the load locks 1208 may correspond to two load locks 1208 in a vertically stacked configuration, and each of the loading stations 1220 may correspond to two loading stations 1220 in a vertically stacked configuration. The height of the EFEM 1204 may be increased to allow the robot 1216 to be raised or lowered to different levels within the EFEM 1204 and to access multiple levels of the loading stations 1220 and the load locks 1208. As can be appreciated, the placement of the laser and other sensors in these configurations may vary accordingly.

[0097] 6A-6C show plan views of an exemplary configuration of another substrate processing tool 1600. The substrate processing tool 1600 includes a modified front-end equipment module (EFEM) 1604 configured to accommodate at least a portion of one or more load locks 1608. In other words, instead of being located completely outside the EFEM 1604 in a gap between the EFEM 1604 and a vacuum transfer module (VTM) 1612, the load locks 1608 extend inside the EFEM 1604. Thus, the EFEM 1604 can be located closer to the VTM 1612, reducing the overall footprint and increasing the pitch of the multiple substrate processing tools 1600.

[0098] As shown, the substrate processing tool 1600 includes ten process modules 1616. However, other configurations of the substrate processing tool 1600 may include more than ten process modules 1616. For example, the length of the VTM 1612 may be extended to accommodate the additional process modules 1616. Similarly, the VTM 1612 may include one or more vacuum transfer robots 1620 (e.g., transfer robots 1620-1, 1620-2, 1620-3, 1620-4, and 1620-5) having various configurations. As shown, the transfer robot 1620 includes one arm 1624 having three arm segments 1628 and one end effector 1632 in each of the configurations. In other configurations, the transfer robot 1620 may include one, two, or more arms 1624. In some examples, the robot 1620 may include two of the end effectors 1632 on each arm 1624.

[0099] 6A, the substrate processing tool 1600 includes a single vacuum transfer robot 1620-1 that is positioned off-center (i.e., shifted right or left toward the process module 1616) relative to the central longitudinal axis of the VTM 1612. In other words, the primary pivot point of the robot 1620-1 is off-center. The robot 1620-1 is positioned and configured to access each of the ten process modules 1616 and the load locks 1608. In a configuration in which the substrate processing tool 1600 includes a storage buffer 1636 and / or a storage buffer 1640, the robot 1620-1 is also configured to access the storage buffers 1636 / 1640.

[0100] 6B and 6C, the substrate processing tool 1600 includes two vacuum transfer robots 1620-2 and 1620-3 or 1620-4 and 1620-5, respectively, positioned off-center (i.e., shifted right or left toward the process modules 1616) relative to a central longitudinal axis of the VTM 1612. The robots 1620-2 and 1620-4 are positioned and configured to access selected ones of the ten process modules 1616 and load locks 608. Conversely, the robots 1620-3 and 1620-5 are positioned and configured to access others of the ten process modules 1616. In a configuration in which the substrate processing tool 1600 includes a storage buffer 1636 and / or a storage buffer 1640, robots 1620-3 and 1620-5 may also be configured to access the storage buffer 1636, while both robots 1620-2 and 1620-3 of FIG. 6B and both robots 1620-4 and 1620-5 of FIG. 6A are configured to access the storage buffer 1640.

[0101] For example, as shown in Figure 6B, robot 1620-2 is aligned with each of the process modules 1616 (e.g., about its horizontal axis) and robot 1620-3 is centered between adjacent ones of the process modules 1616. Conversely, as shown in Figure 6C, robots 1620-4 and 1620-5 are each aligned with a respective one of the process modules 1616. As can be appreciated, the placement of laser and other sensors in these configurations can vary significantly.

[0102] Chamber Example 7A-7C show different examples of process modules (PMs) that can be used in the tools shown in FIGS. 1-6C. These process chambers can also include one or more cameras (not shown) mounted on the outside of the PM's window and used to monitor the interior of the PM. These cameras are some of the other sensors typically used to sense the various operations of the PM and the tool. A measurement system works in conjunction with these other sensors and cameras, as described below with reference to FIGS. 9-11. The controllers described in FIGS. 7A-7C can exchange data with the measurement system and can operate the respective PMs based on the data received from the measurement system, as described below with reference to FIGS. 9A-11.

[0103] 7A illustrates an example of a substrate processing system 1700 including a processing chamber 1702. Although the example is described in the context of plasma-enhanced chemical vapor deposition (PECVD), the teachings of the present disclosure can be applied to other types of substrate processing, such as atomic layer deposition (ALD), plasma-enhanced ALD (PEALD), CVD, or other processes including etch processes. The system 1700 includes a processing chamber 1702 that surrounds the other components of the system 1700 and contains an RF plasma (if used). The processing chamber 1702 includes an upper electrode 1704 and an electrostatic chuck (ESC) 1706 or other substrate support. During operation, a substrate 1708 is positioned on the ESC 1706.

[0104] For example, the upper electrode 1704 can include a gas distribution device 1710, such as a showerhead, for introducing and distributing process gases. The gas distribution device 1710 can include a stem portion including one end connected to the top surface of the processing chamber 1702. The base portion of the showerhead is generally cylindrical and extends radially outward from the opposite end of the stem portion at a location spaced from the top surface of the processing chamber 1702. The substrate-facing surface or faceplate of the showerhead base portion includes a number of holes through which vaporized precursors, process gases, or purge gases flow. Alternatively, the upper electrode 1704 can include a conductive plate, and the process gases can be introduced in another manner.

[0105] The ESC 1706 includes a base plate 1712 that acts as a bottom electrode. The base plate 1712 supports a heating plate 1714, which may correspond to a ceramic multi-zone heating plate. A thermal resistance layer 1716 may be disposed between the heating plate 1714 and the base plate 1712. The base plate 1712 may include one or more channels 1718 for flowing coolant through the base plate 1712.

[0106] If a plasma is used, the RF generating system 1720 generates and outputs an RF voltage to one of the upper electrode 1704 and the lower electrode (e.g., the base plate 1712 of the ESC 1706). The other of the upper electrode 1704 and the base plate 1712 can be DC grounded, AC grounded, or floating. For example, the RF generating system 1720 may include an RF generator 1722 that generates RF power that is supplied to the upper electrode 1704 or the base plate 1712 by a matching and distribution network 1724. In other examples, the plasma may be generated inductively or remotely.

[0107] Gas delivery system 1730 includes one or more gas sources 1732-1, 1732-2, ..., and 1732-N (collectively gas sources 1732), where N is an integer greater than zero. Gas sources 1732 are connected to manifold 1740 by valves 1734-1, 1734-2, ..., and 1734-N (collectively valves 1734) and mass flow controllers 1736-1, 1736-2, ..., and 1736-N (collectively mass flow controllers 1736). Vapor delivery system 1742 supplies vaporized precursors to manifold 1740, which is connected to process chamber 1702, or to another manifold (not shown). The output of manifold 1740 is provided to process chamber 1702.

[0108] A temperature controller 1750 may be connected to a plurality of thermal control elements (TCEs) 1752 disposed on the heating plate 1714. The temperature controller 1750 may be used to control the plurality of TCEs 1752 to control the temperature of the ESC 1706 and the substrate 1708. The temperature controller 1750 may be in communication with a coolant assembly 1754 to control the flow of coolant through the channels 1718. For example, the coolant assembly 1754 may include a coolant pump, a reservoir, and one or more temperature sensors (not shown). The temperature controller 1750 operates the coolant assembly 1754 to selectively flow coolant through the channels 1718 to cool the ESC 1706. A valve 1756 and a pump 1758 may be used to evacuate the coolant from the processing chamber 1702. A system controller 1760 controls the components of the system 1700.

[0109] FIG. 7B illustrates another example of a substrate processing system 1800. The substrate processing system 1800 includes a coil driver circuit 1811. In some examples, the coil driver circuit 1811 includes an RF source 1812, a pulse circuit 1814, and an adjustment circuit (i.e., matching circuit) 1813. The pulse circuit 1814 controls a transformer coupled plasma (TCP) envelope of the RF signal generated by the RF source 1812 and varies the duty cycle of the TCP envelope between 1% and 99% during operation. As can be appreciated, the pulse circuit 1814 and the RF source 1812 can be combined or separate.

[0110] The conditioning circuitry 1813 may be directly connected to the induction coil 1816. Although the substrate processing system 1810 uses a single coil, some substrate processing systems may use multiple coils (e.g., inner and outer coils). The conditioning circuitry 1813 conditions the output of the RF source 1812 to a desired frequency and / or a desired phase to match the impedance of the coil 1816.

[0111] The dielectric window 1824 is disposed along an upper side of the processing chamber 1828. The processing chamber 1828 includes a substrate support (or pedestal) 1832 that supports a substrate 1834. The substrate support 1832 may include an electrostatic chuck (ESC), or a mechanical chuck, or other type of chuck. Process gases are supplied to the processing chamber 1828, and a plasma 1840 is generated inside the processing chamber 1828. The plasma 1840 etches the exposed surface of the substrate 1834. An RF source 1850, a pulse circuit 1851, and a bias matching circuit 1852 can be used to bias the substrate support 1832 during operation to control ion energy.

[0112] A gas delivery system 1856 can be used to supply a process gas mixture to the processing chamber 1828. The gas delivery system 1856 can include process and inert gas sources 1857, a gas metering system 1858, such as valves and mass flow controllers, and a manifold 1859. A gas injector 1863 can be positioned at the center of the dielectric window 1824 and is used to inject the gas mixture from the gas delivery system 1856 into the processing chamber 1828. Additionally or alternatively, the gas mixture can be injected from a side of the processing chamber 1828.

[0113] A heater / cooler 1864 can be used to heat / cool the substrate support 1832 to a predetermined temperature. An exhaust system 1865 includes a valve 1866 and a pump 1867 to control the pressure within the processing chamber and / or to remove coolant from the processing chamber 1828 by purging or evacuation.

[0114] A controller 1854 can be used to control the etching process. The controller 1854 monitors system parameters and controls the delivery of gas mixtures, striking, maintaining, and extinguishing the plasma, removing coolant, supplying cooling gas, etc. Additionally, as described below, the controller 1854 can control various aspects of the coil driver circuitry 1810, the RF source 1850, and the bias matching circuitry 1852, etc.

[0115] 7C shows a processing chamber 1900 for etching a layer of a substrate. The processing chamber 1900 includes a lower chamber region 1902 and an upper chamber region 1904. The lower chamber region 1902 is defined by a chamber sidewall surface 1908, a chamber bottom surface 1910, and a lower surface of a gas distribution device 1914.

[0116] The upper chamber region 1904 is defined by a top surface of the gas distribution device 1914 and an inner surface of a dome 1918. In some examples, the dome 1918 rests on a first annular support 1921. In some examples, the first annular support 1921 includes one or more spaced holes 1923 for delivering process gas to the upper chamber region 1904. In some examples, the process gas is delivered by the one or more spaced holes 1923 upward at an acute angle relative to a plane containing the gas distribution device 1914, although other angles / directions can be used. In some examples, a gas flow channel 1934 in the first annular support 1921 delivers gas to the one or more spaced holes 1923.

[0117] The first annular support 1921 can rest on a second annular support 1925 that defines one or more spaced apart holes 1927 for delivering process gas from gas flow passages 1929 to the lower chamber region 1902. In some examples, holes 1931 in the gas distribution device 1914 align with holes 1927. In other examples, the gas distribution device 1914 has a smaller diameter and holes 1931 are not needed. In some examples, the process gas is delivered downward toward the substrate 1926 by one or more spaced apart holes 1927 at an acute angle relative to a plane containing the gas distribution device 1914, although other angles / directions can be used. In other examples, the upper chamber region 1904 is cylindrical with a flat top surface and one or more flat inductive coils can be used. In still other examples, a single chamber can be used with a spacer located between the showerhead and the substrate support.

[0118] A substrate support 1922 is disposed in the lower chamber region 1904. In some examples, the substrate support 1922 includes an electrostatic chuck (ESC), although other types of substrate supports can be used. A substrate 1926 is disposed on an upper surface of the substrate support 1922 during etching. In some examples, the temperature of the substrate 1926 may be controlled by a heater plate 1930, an optional cooling plate with fluid channels, and one or more sensors (not shown), although any other suitable substrate support temperature control system may be used.

[0119] In some examples, the gas distribution device 1914 includes a showerhead (e.g., a plate 1928 having a plurality of spaced apart holes 1927). The plurality of spaced apart holes 1927 extend from an upper surface of the plate 1928 to a lower surface of the plate 1928. In some examples, the spaced apart holes 1927 have a diameter in the range of 0.4 inches to 0.75 inches, and the showerhead is made of a conductive material, such as aluminum, or a non-conductive material, such as ceramic, with embedded electrodes made of a conductive material.

[0120] One or more induction coils 1940 are disposed about an outer portion of the dome 1918. When energized, the one or more induction coils 1940 generate an electromagnetic field inside the dome 1918. In some examples, an upper coil and a lower coil are used. A gas injector 1942 injects one or more gas mixtures from a gas delivery system 1950-1.

[0121] In some examples, gas delivery system 1950-1 includes one or more gas sources 1952, one or more valves 1954, one or more mass flow controllers (MFCs) 1956, and mixing manifold 158, although other types of gas delivery systems may be used. A gas splitter (not shown) may be used to vary the flow rate of the gas mixture. Another gas delivery system 1950-2 may be used to supply an etching gas or etching gas mixture to gas flow channels 1929 and / or 1934 (in addition to or instead of the etching gas from gas injector 1942).

[0122] In some examples, the gas injector 1942 includes a central injection location that directs gas downward and one or more side injection locations that inject gas at an angle to the downward direction. In some examples, the gas delivery system 1950-1 delivers a first portion of the gas mixture at a first flow rate to the central injection location of the gas injector 1942 and a second portion of the gas mixture at a second flow rate to the side injection location of the gas injector 1942. In other examples, a different gas mixture is delivered by the gas injector 1942. In some examples, the gas delivery system 1950-1 delivers a regulated gas to the gas flow channels 1929 and 1934 and / or to other locations within the processing chamber, as described below.

[0123] A plasma generator 1970 can be used to generate RF power that is output to the one or more inductive coils 1940. The plasma 1990 is generated in the upper chamber region 1904. In some examples, the plasma generator 1970 includes an RF generator 1972 and a matching network 1974. The matching network 1974 matches the impedance of the RF generator 1972 to the impedance of the one or more inductive coils 1940. In some examples, the gas distribution device 1914 is connected to a reference potential, such as ground. A valve 1978 and a pump 1980 can be used to control the pressure in the lower and upper chamber regions 1902, 1904 and to evacuate the coolant.

[0124] A controller 1976 communicates with gas delivery systems 1950-1 and 1950-2, valves 1978, pumps 1980, and plasma generator 1970 to control the flow of process gases, purge gases, RF plasma, and chamber pressure. In some examples, the plasma is maintained inside the dome 1918 by one or more inductive coils 1940. One or more gas mixtures are introduced from the top of the chamber using gas injectors 1942 (and / or holes 1923), and the plasma is confined within the dome 1918 using gas distribution device 1914.

[0125] Confining the plasma within the dome 1918 allows for volumetric recombination of plasma species to release the desired etchant species through the gas distribution device 1914. In some instances, there is no RF bias applied to the substrate 1926. As a result, there is no active sheath over the substrate 1926 and ions are not hitting the substrate with finite energy. A certain amount of ions diffuse out of the plasma region through the gas distribution device 1914. However, the amount of plasma that diffuses is an order of magnitude less than the plasma located inside the dome 1918. Most ions in the plasma are lost through volumetric recombination at high pressures. Surface recombination losses at the top surface of the gas distribution device 1914 also reduce the ion density below the gas distribution device 1914.

[0126] In another example, an RF bias generator 1984 is provided and includes an RF generator 1986 and a matching network 1988. The RF bias can be used to generate a plasma between the gas distribution device 1914 and the substrate support or to generate a self-bias on the substrate 1926 to attract ions. A controller 1976 can be used to control the RF bias.

[0127] An edge bond ring can be used to adjust the etch rate and / or etch profile of the plasma near the outer radial edge of the substrate. The edge bond ring is typically located on a pedestal around the outer radial edge of the substrate. The process conditions at the outer radial edge of the substrate can be modified by changing the position of the edge bond ring, the shape or profile of the inner edge of the edge bond ring, the height of the edge bond ring relative to the top surface of the substrate, the material of the edge bond ring, etc.

[0128] FIG. 7D illustrates an example of an edge bond ring 1870 surrounding a pedestal 1871. The edge bond ring 1870 may include a single portion or two or more portions. In the example shown, the edge bond ring 1870 includes a first annular portion 1872 disposed radially outward of a substrate 1873. A second annular portion 1874 is located radially inward from the first annular portion 1872 below the substrate 1873. A third annular portion 1876 is disposed below the first annular portion 1872. In use, a plasma 1875 is directed to the substrate 1873 to etch exposed portions of the substrate 1873. The edge bond ring 1870 is positioned to help shape the plasma such that uniform etching of the substrate 1873 occurs. After the edge bond ring 1870 has been used, the top surface of the radially inner portion of the edge bond ring 1870 may exhibit erosion (e.g., at 1878). As a result, the plasma 1875 may tend to etch the radially outer edge of the substrate 1873 at a faster rate than it etches the radially inner portion of the substrate 1873, which may result in non-uniform etching of the substrate 1873 near the radially outer edge of the substrate 1873. In some implementations, the edge coupling ring is located on three or more lift pins to gradually transport the ring up as the top surface of the ring erodes, maintaining an optimal value for the height of the edge coupling ring above the ESC for the life of the ring.

[0129] One or more portions of the edge bond ring 1870 can be moved vertically and / or horizontally relative to the substrate or pedestal 1871. This movement changes the edge-coupled effect of the plasma 1875 on the substrate 1873 during etching or other substrate processing without the need to open the processing chamber. The actuator 1880 can be positioned at various locations to move one or more portions of the edge bond ring 1870 relative to the substrate 1873. By way of example only, the actuator 1880 can be positioned between the first annular portion 1872 and the third annular portion 1876 of the edge bond ring 1870. In some examples, the actuator 1880 can include a piezoelectric actuator, a stepper motor, a pneumatic drive, or other suitable actuator. In some examples, one, two, three, or four or more actuators are used. In other examples, multiple actuators can be uniformly positioned around the edge bond ring 1870. The actuator can be positioned inside or outside the processing chamber.

[0130] The actuator 1880 is used to move one or more portions of the edge bond ring 1870 to change the position of one or more portions of the edge bond ring 1870. For example, the actuator 1880 can be used to move the first annular portion 1872 of the edge bond ring 1870. In this example, the actuator 1880 moves the first annular portion 1872 of the edge bond ring 1870 in an upward or vertical direction, such that an edge of the first annular portion 1872 of the edge bond ring 1870 is higher relative to the radially outer edge of the substrate 1873. As a result, the etch uniformity near the radially outer edge of the substrate 1873 is improved.

[0131] The actuator 1880 can move in other directions, such as horizontally, diagonally, etc. Horizontal movement of portions of the edge bond ring 1870 can be implemented to center the edge bond effect with respect to the substrate 1873. For example, the actuator 1880 can be positioned radially outward of the edge bond ring 1870. In addition, the actuator 1880 can move in the vertical (or up / down) direction as well as the horizontal (or left / right) direction. Horizontal repositioning can be used if the etching of the substrate indicates a horizontal offset of the edge bond ring 1870 with respect to the substrate. The horizontal offset can be corrected without opening the processing chamber. Similarly, tilting of the edge bond ring 1870 can be implemented by actuating some of the actuators differently than others to correct or create a left / right asymmetry.

[0132] Measurement System As can be understood from the above description, there are various types of processes performed on various types of wafers. The wafers are processed in process modules located in tools having different configurations. Depending on the process being performed and the configuration of the tool, the wafer and associated edge bond ring are often transported (i.e., transported) into and out of the chamber. Various types of data are collected from the chambers and tools using sensors located at various locations within these chambers and tools. These sensors can include cameras located around the processing chamber to observe and provide additional data regarding various components within the processing chamber.

[0133] Additional cameras can be placed elsewhere in the tool (e.g., above the airlock) and can be used to measure various parameters of the edge bonded ring. However, determining the profile (3D model) of the edge bonded ring along the X, Y, and Z axes typically requires at least two cameras placed at the locations where the new / used edge bonded ring passes (e.g., airlock, aligner, robot arm, etc.). To obtain a profile with an accuracy of one micron or less, the cameras need to be very high resolution and mounted on high-precision machined brackets with high-quality lenses. Such camera systems are bulky and expensive. Furthermore, in order to use the cameras, the lighting needs to be positioned in a consistent and repeatable manner. Furthermore, the computer vision algorithms used to process the images captured by these cameras can be very complex and computationally expensive, yet still have some degree of uncertainty. All these issues make the use of cameras very difficult.

[0134] Instead, the present disclosure provides systems and methods for determining the profile of the edge bond ring and wafer along the X, Y, and Z axes (e.g., radial and tangential edge ring thickness) using one or more laser (optical) sensors. Laser sensors are smaller in size than high-resolution cameras. Laser sensors can measure the profile with an accuracy of 0.1 microns or less. Laser sensors are relatively less expensive than high-resolution cameras and lenses and can provide higher measurement accuracy. Laser sensors do not require the additional illumination arrangements that cameras require. Furthermore, algorithms for processing the data captured by the laser sensors are robust, repeatable, and easy to implement.

[0135] The laser sensor can measure a wide range of variables (geometric parameters) of the edge bond ring and the wafer. For example, edge bond ring variables can include, but are not limited to, thickness, width, outer / inner diameter, bow / flatness, 2D / 3D shape, and surface roughness. For example, wafer variables can include, but are not limited to, thickness and bow / flatness. Changes in any of these parameters, including changes in thickness of new / used edge bond rings, are tools to quantify process characteristics (etch rate, etch uniformity, etc.) in the processing module. Similarly, changes in wafer thickness, etc. are indicators of process characteristics (etch / deposition rate, etch / deposition uniformity, etc.) in the processing module. Measurements of these changes across a fleet of tools can be used for autonomous process optimization and PM alignment by adjusting (optimizing) process recipe parameters and controlling PM configurations (edge ​​bond ring height above ESC, top / bottom electrode gap / alignment / tilt, etc.) and auto-cleaning recipes / frequency in an automated, closed-loop manner.

[0136] By measuring the flatness of the edge bonded ring used, non-uniform etch rates across the tangential direction can be identified. This information can be used to adjust PM parameters (e.g., spatially dependent temperature distribution of ESC, adding artificial non-uniformity to edge bonded ring lift pin height adjustment, etc.) to achieve PM matching, autonomous process optimization, uniform on-wafer results, etc. PM control parameters can be determined using machine learning algorithms running on distributed computing systems that monitor factory operations across the fleet (described below).

[0137] There are multiple causes of suboptimal process results including PM mismatch, as well as misalignment of the top / bottom electrode assemblies, chamber component tolerances and assembly tolerance stack-up variations, chamber subsystem malfunctions (valves, actuators, sensors, heating / cooling mechanisms, etc.), and variations in process by-product accumulation on the chamber (quality of the chamber cleaning process). Any of these factors can alter the uniformity of the plasma field in the PM and affect the process results. Parameters such as radial and tangential thickness variations, top surface slope, inner diameter (ID), and surface roughness of pre- / post-treatment of edge coupling rings, which can be extracted from 1D / 2D / 3D laser sensor data via reflectometry and interferometry, may indicate the above PM issues. Once a problem is detected, corrective actions can be identified on the remote server and implemented on the tool.

[0138] Throughout this disclosure, references to autonomous process optimization and PM matching should be understood to refer to adjusting (optimizing) process recipe parameters in an automated, closed-loop manner, controlling PM configuration (height of edge bond ring above ESC, top / bottom electrode gap / alignment / tilt, etc.), and automated cleaning recipe / frequency. Additionally, throughout this disclosure, references to PM matching should be understood as correcting multiple sources of PM mismatch and suboptimal process results including mismatch of top / bottom electrode assemblies, variations in chamber part tolerances and assembly tolerance stack-up, malfunctions of chamber subsystems (valves, actuators, sensors, heating / cooling mechanisms, etc.), and variations in accumulation of process by-products on the chamber (quality of chamber cleaning process).

[0139] Corrective actions are implemented on the cleaning process as well as other parameters. Non-limiting examples are described below. Examples of corrective actions include adjustments of process parameters (e.g., gas flow, pressure, top electrode temperature, ESC spatially dependent temperature distribution, bias voltage, etch / deposition rate, etch / deposition uniformity, etc.), top / bottom electrode gap and alignment / tilt, edge ring height above ESC, cleaning process recipe and frequency, etc. Examples of cleaning mechanisms include waferless automatic cleaning (WAC), cover wafer area cleaning (CWAC), etc. Collecting measurements of these parameters across the fleet can help adjust process parameters (e.g., etch / deposition rate, etch / deposition uniformity, etc.) as well as WAC / CWAC parameters, duration, frequency, etc. in a closed loop manner. Changes in ID edge shape / profile of pre- / post-processing of edge bond rings can be indicators of PM issues, which allows for guided troubleshooting and automatic prompts for PM maintenance. Different thickness profiles near the ID edge determined based on measurements collected over time can help indicate malfunction of a particular component / subsystem. Similarly, wafer measurements can also be used to evaluate process characteristic variations across the wafer, as well as cleaning quality.

[0140] Furthermore, the cleaning processes used in different chambers can vary widely based on the type of processing (e.g., etch, deposition, etc.) performed in those chambers. For example, some chambers may need to be cleaned after processing every wafer, while others may need to be cleaned after processing multiple wafers. Furthermore, the parameters of the cleaning process recipes may also vary. All these factors, such as the type and frequency of the cleaning process recipes, can be correlated with the edge bond ring and wafer measurements. The cleaning process can be automated using models trained based on the correlations. For example, in use, based on the edge bond ring and / or wafer measurements, the trained models can select the WAC recipe and frequency of cleaning. Furthermore, PM performance (e.g., etch rate, etch uniformity, autoclean, etc.) can also be evaluated using measurements performed continuously, and PM performance parameters can be fine-tuned on the fly based on these evaluations to achieve autonomous process optimization and PM alignment.

[0141] Thus, the measurement system not only evaluates the performance of the cleaning process, but also evaluates on-wafer process results (e.g., etch rate and deposition rate uniformity). Furthermore, based on measurements of the edge bond ring and / or wafer, the measurement system can use one or more trained models to autonomously adjust process parameters (gas flow, pressure, top electrode temperature, ESC spatially dependent temperature distribution, bias voltage, etc.), top / bottom electrode gap and alignment / tilt, edge ring height above the ESC, as described below with reference to FIG. 9A et seq. Throughout this disclosure, the phrase "etch rate" as used with respect to the edge bond ring should be understood as "effective etch rate" and as "etch rate" when used with respect to the wafer.

[0142] To implement these systems and methods according to the present disclosure, a distributed computing system consisting of a big data and machine learning platform (described below) is used. Most machine learning applications require an initial stage of data collection and data labeling for use in model training. Data labeling during this initial period (which may be, for example, six months to several years) can be automated or manual. After the models are trained using machine learning, the trained models are used in real time and their accuracy is improved by continuous training based on new data continuously collected during use (i.e., during production).

[0143] The geometric parameters of the edge bond ring and the wafer can be measured using one or more laser sensors located on the lid or cover of the airlock chamber, for example, as follows: For example, the change in the thickness profile of the edge bond ring or the wafer can be measured using a laser sensor located on top of the airlock chamber, as follows: Prior to using the edge bond ring or the wafer in the PM, the laser sensor transmits a laser beam to the top surface of the edge bond ring or the wafer while the edge bond ring or the wafer is placed in the airlock chamber. The laser sensor senses the reflection of the laser beam from the top surface of the edge bond ring or the wafer. Optical interferometry is performed based on the travel time of the laser beam from the light source (emitter) in the laser sensor to the top surface of the edge bond ring or the wafer and back to the sensing element of the laser sensor, and the distance between the laser sensor and the top surface of the edge bond ring or the wafer is calculated. After the edge bond ring has been used in the PM for several RF hours (or the wafer has been processed), the same procedure is performed to obtain the distance between the laser sensor and the top surface of the edge bond ring or the wafer. The difference between the two distance measurements represents the change in thickness of the edge bond ring or the wafer. Further discussion regarding measurements performed using reflectometry and interferometry follows with reference to the description of the laser sensor shown in Figures 10A-10C.

[0144] The edge coupling rings can be made of different materials. The wafers can also include multiple layers of different membrane materials. In particular, the composition of the top membrane layers of the wafer material can change after etching and deposition. Some of these materials can be transparent to some frequencies of light, while others can be opaque to some other frequencies of light. The sensor can include emitters that can emit and sense electromagnetic waves of different frequencies and perform measurements of edge coupling rings made of different materials or wafers that include different materials of membrane. These frequencies can be controlled during the measurements. For example, these frequencies can be changed depending on the material of the edge coupling ring or the wafer being measured. Alternatively, white light composed of a wide range of frequency components can be used as the emitter.

[0145] The overall data flow of the measurement system is as follows: The laser sensors send the sensed data from the edge coupled ring or wafer to the system controller. The system controller can process the data and send pre-processed / post-processed data to the host / cloud (e.g., data center). For autonomous process optimization and inter-PM alignment, measurements are performed on the edge coupled ring or wafer of several tools and a fleet of tools. Data from several tools and a fleet of tools is collected at the host / cloud (e.g., data center). The model is trained based on the data collected at the host / cloud (e.g., data center) using machine learning. In use, the model receives data from the tools and sends responses to the tools in real time or near real time based on the received data to optimize the performance of the tools.

[0146] If the etch rate of the PM varies and is spatially dependent, different regions of the edge bond ring may erode at different rates. Erosion of the edge bond ring can be detected by measuring the flatness of the edge bond ring as follows. For example, depending on the change in etch rate in the PM, location A, which can be a point / spot (1D), line (2D), or area (3D) on the edge bond ring, may erode faster or slower than point B on the edge bond ring. This can be detected by measuring the thickness and slope of the edge bond ring at points A and B. One way to measure the thickness of the edge bond ring at points A and B is to mount the edge bond ring in an airlock and perform a thickness measurement at point A. The edge bond ring can then be rotated (e.g., using an aligner in the EFEM) or moved via robot under the sensor and a thickness measurement can be performed at point B. If the thickness measurements are different at points A and B, then the etch rate is spatially dependent in the PM. The difference in the thickness measurements at points A and B can quantify the amount that the etch rate is spatially dependent in the PM.

[0147] Using this information, adjustments can be made within the process chamber to uniform the etch rate throughout the process chamber. For example, based on the information, the spatial temperature profile of the chuck can be altered by controlling a matrix heater within the chuck, the height / tilt of the edge bond ring can be altered by controlling the pins used to adjust the height / tilt of the edge bond ring, and the alignment / tilt of the top / bottom electrode assemblies can also be adjusted.

[0148] Similar procedures can be used to measure the change in wafer thickness before and after processing in the PM, and the measurements can be used to determine the etch rate / uniformity or deposition rate / uniformity and adjust the recipe parameters of the process performed in the PM and to adjust the process results by adjusting the temperature profile of the chuck in the PM.

[0149] Two or more laser sensors can be used, for example, in an airlock chamber. The uncertainty of the measurements can decrease in direct proportion to the number of sensors used to collect data from the edge-coupled ring. However, increasing the number of sensors can also increase costs, data processing complexity, and maintenance issues. Instead, the use of a small number of sensors (e.g., two) in combination with robust data processing can provide sufficient accuracy.

[0150] A laser sensor may include a transmitting element for emitting electromagnetic (EM) waves (e.g., light) and a receiving element for measuring the EM waves reflected from the surface of an object (e.g., an edge ring or a wafer). The transmitting element may include one or more arrays of emitters, and the receiving element may include one or more arrays of receivers. For point (1D) measurements, only one emitter may be used to transmit the EM waves (e.g., light) to a point on the edge-coupled ring or wafer. For line (2D) measurements, a single or line emitter may be used to transmit light to a portion of the edge-coupled ring or wafer. To perform area (3D) measurements, an array of one or more emitters may be used to transmit light to an area of ​​the edge-coupled ring or wafer. Alternatively, a sensor with a broadband EM emitter (e.g., a light source) and multiple receivers may be used to perform these measurements. One or more lenses and mirrors may be used to direct the generated / received EM beams in parallel paths toward a desired location on the object or sensing element. Examples of laser sensors are shown and described below with reference to Figures 10A-C.

[0151] Over time, as more and more data is collected and analyzed, signatures or patterns of erosion of the edge bond ring, and thickness changes of the wafer can be recorded and correlated with corrective actions / adjustments required for the PM. This can greatly simplify process optimization and enable guided troubleshooting and predictive maintenance. For example, data collected from the edge bond ring or wafer using a laser sensor can be converted into a signature (feature vector) and sent to a trained model. The trained model can compare the received data signatures to signatures and associated corrective actions stored in a database. Based on the comparison, the trained model can return an output indicating the PM operating conditions, process recipe, and modifications required for maintenance (e.g., auto cleaning, valve maintenance, lift pin calibration, top / bottom electrode gap adjustment, etc.) required for a particular assembly or subassembly within the PM. In some cases, the corrective action may be an adjustment of the process parameters of the process performed in the PM, a change in the temperature profile of the chuck, the recipe used for the automatic cleaning (WAC / CWAC), or an adjustment of the height / tilt of the edge bond ring of the PM, which may be performed automatically based on the signature comparison in a closed-loop manner.

[0152] Thus, the measurement system of the present disclosure includes a hierarchical strategy for addressing problems and providing solutions to the problems. For example, each time data is collected from an edge-coupled ring or wafer and sent to the model in the form of a data signature, the model can perform a signature comparison and indicate multiple solutions or corrective actions that can be implemented in sequence. Then, in the tool, the system controller can determine if and which action has already been implemented and implement the next action in the sequence based on the determination.

[0153] For example, in response to detecting a particular thickness signature pattern for an edge bond ring, a suggestion from the trained model may be to implement the following actions in the order shown (by way of example only): first change the process parameters and / or change the spatially dependent temperature profile of the chuck, then change the height / tilt of the edge bond ring if the temperature profile has already been changed, and finally replace the edge bond ring if the height / tilt of the edge bond ring has already been changed and cannot be changed any further.

[0154] In response to detecting a particular signature pattern for the wafer, a suggestion from the trained model may be to implement the following actions in the order shown (by way of example only): first modify the spatially dependent temperature profile of the chuck, modify the process recipe parameters if the temperature profile has already been modified, and modify one or more items related to the edge bond ring as described above.

[0155] In addition, the weight of the edge bond ring or wafer is also monitored and can be used as an indication of the amount of erosion of the edge bond ring or wafer, as well as the amount of material etched / deposited on the wafer. For example, sensors for sensing the weight of the edge bond ring and / or wafer can be installed, for example, in the airlock chamber embedded in the wafer fingers (or other suitable locations in the tool). Alternatively, sensor fusion algorithms such as Bayesian or Kalman filtering methods can be used to combine measurements from laser sensors with the weight measurements of the edge bond ring and wafer to reduce measurement uncertainty.

[0156] Using these systems and methods of the present disclosure, wafers can be monitored before and after processing, edge bond rings can be monitored periodically (e.g., daily), PM parameters can be adjusted, and / or recipes within PM can be fine-tuned (optimized). These and other features of the systems and methods of the present disclosure are now described in further detail.

[0157] A wide range of sensor hardware can be used to measure different geometric parameters of the edge bonded ring and the wafer. Data collected from the edge bonded ring and the wafer by the laser sensors can be used to verify if the process in the PM is running correctly, if the process and / or any components in the PM / tool ​​need to be changed / adjusted in any way, if preventive maintenance (e.g., PM cleaning) needs to be performed, or if any components in the PM / tool ​​need to be serviced or replaced, etc.

[0158] In summary, the measurement system of the present disclosure comprises one or more detectors, such as laser sensors, mounted at one or more locations within the tool (e.g., above the airlock of the tool, on a robot arm, on an aligner, etc.). These laser sensors capture data from the edge bond ring and wafer as they are transported between different process modules of the tool. For example, these laser sensors capture data from the incoming / outgoing edge bond ring and wafer. This allows for near in-situ inspection of the edge bond ring and wafer, enabling inter-PM matching, autonomous process optimization, and predictive maintenance. The inspection is referred to as near in-situ because the inspection is not performed while the edge bond ring or wafer is seated in the PM of the tool, but is nevertheless performed when the edge bond ring or wafer is transported from the PM to the airlock of the tool (i.e., near and simultaneously in the vicinity of the PM).

[0159] The data collected by these laser sensors can be processed for feature extraction by a processor embedded in the laser sensor, or on a printed circuit board external to the laser sensor, or on a system computer. The extracted features or raw data are sent to and collected in a fleet-wide distributed computing and communication system capable of storing big data coming in from various tools in the fab. The measurement system uses machine learning to build and train models and correlates the data to achieve autonomous process optimization and inter-PM alignment, implementing near real-time closed-loop control of the tools across the fleet. The measurement system also enables autonomous operation and control of the semiconductor tools across the fleet by automating various maintenance, troubleshooting, and tool control tasks.

[0160] For example, a measurement system can be used to monitor changes in edge bond ring thickness and edge profile (inner diameter, radial and tangential gradients, etc.) at the airlock and determine process characteristics (e.g., etch rate, etch uniformity, etc.) in the process module based on changes between the edge erosion profiles of the incoming / outgoing edge bond ring. This data can be used to automatically adjust the process module recipe, the height of the edge bond ring above the chuck via the lift pins, and / or adjust the temperature profile of the chuck by controlling one or more heaters in the chuck for autonomous process optimization and inter-tool process module matching.

[0161] environment The following is a simple example of a distributed computing environment in which the disclosed systems and methods may be implemented. Throughout the description, references to terms such as server, client device, application, etc. are for illustrative purposes only. The terms server and client device should be broadly understood to represent a computing device with one or more processors and memory configured to execute machine-readable instructions. The terms application and computer program should be broadly understood to represent machine-readable instructions executable by a computing device.

[0162] FIG. 8A illustrates a high level architecture of a measurement system 1770 according to the present disclosure. The measurement system 1770 is described in detail with reference to FIG. 9A onward. The measurement system 1770 comprises a number of fleets 1772-1, 1772-2, ..., and 1772-N (collectively fleets 1772) communicatively connected to a distributed computing system 1774. For example, the fleet of tools may be similar to the arrangement of tools in a fabrication facility illustrated in FIG. 2. Each fleet 1772 comprises one or more detectors (e.g., laser sensors) 1780, a data processor 1782, a perception algorithm 1784, a controller 1786, a robot 1788, an actuator 1790, and a recipe 1792.

[0163] An overview of the measurement system 1770 is provided below. Details of the measurement system 1770 are described below with reference to FIG. 9A onwards. Briefly, the detector 1780 includes a laser sensor that captures data from edge bond rings and wafers in the tools in the fleet 1772. The data processor 1784 processes the captured data. For example, the data processor 1784 can perform operations such as noise filtering, feature detection and extraction, etc.

[0164] The distributed computing system 1774 comprises multiple servers (see, e.g., FIGS. 8B-8D and 11). The distributed computing system 1774 trains various models that can be used to automatically perform certain tasks on the fleet 1772. The distributed computing system 1774 trains the various models based on data collected from the fleet 1172 over time. The distributed computing system 1774 trains the various models using perception, sensor fusion, and machine learning algorithms.

[0165] In use, during operation of the fleet 1772, the captured data is sent to the trained models in the distributed computing system 1774. In response, a controller 1786 in the fleet 1772 receives as input the output of the trained models in real time and automatically performs actions such as optimizing the processes and recipes used on the tools in the fleet 1772 and PM alignment by performing adjustments, cleaning, maintenance, guided troubleshooting, etc. of the PMs and tools components in the fleet 1772, as described in further detail below with reference to FIG.

[0166] 8B illustrates a simplified example of a distributed computing system 2000. The distributed computing system 2000 includes a distributed communication system 2010, one or more client devices 2020-1, 2020-2, ..., and 2020-M (collectively, client devices 2020), and one or more servers 2030-1, 2030-2, ..., and 2030-N (collectively, servers 2030). M and N are integers equal to or greater than 1. The distributed communication system 2010 may include a local area network (LAN), a wide area network (WAN) such as the Internet, or other types of networks. The client devices 2020 and the servers 2030 may be located in different geographic locations and may communicate with each other via the distributed communication system 2010. The client devices 2020 and the servers 2030 connect to the distributed communication system 2010 using wireless and / or wired connections.

[0167] The client device 2020 may include one or more tools, a system computer that controls the tool, a PM, and a controller that controls the PM. The client device 2020 may also include a smartphone, a personal digital assistant (PDA), a tablet, a laptop computer, a personal computer (PC), etc., used by an operator of the tool. The server 2030 may provide multiple services to the client device 2020. For example, the server 2030 may run software and machine learning applications developed by one or more vendors. The server 2030 may host multiple databases that are relied upon by the software applications in providing services to users of the client device 2020. The server 2030 and databases may be hosted in the cloud, on-premise, or both.

[0168] In some examples, one or more of the client devices 2020 or the servers 2030 execute one or more applications that process data captured by a laser sensor installed on the tool. The applications also train one or more models based on the captured data using machine learning techniques. In addition, the applications receive and analyze data from various other sensors in the tool and PM. The applications use the models to analyze the data collected from the laser sensor and other sensors to perform various functions such as optimization of processes performed in the PM, PM alignment, and diagnostics, as well as guided troubleshooting of root causes of problems in the tool and PM, predictive maintenance, etc. The applications may be implemented as Software as a Service (SaaS).

[0169] 8C illustrates a simplified example of a client device 2120-1. Client device 2120-1 may typically include one or more central processing units (CPUs), one or more graphical processing units (GPUs), and one or more tensor processing units (TPUs) (collectively referred to as processors 2150), one or more input devices 2152 (e.g., a keypad, touchpad, mouse, touch screen, detectors or sensors such as a camera, etc.), a display subsystem 2154 including a display 2156, a network interface 2158, memory 2160, and bulk storage 2162.

[0170] The network interface 2158 connects the client device 2120-1 to the distributed computing system 2000 via the distributed communications system 2110. For example, the network interface 2158 may include a wired interface (e.g., an Ethernet, EtherCAT, or RS-485 interface) and / or a wireless interface (e.g., Wi-Fi, Bluetooth, Near Field Communication (NFC), or other wireless interface). The memory 2160 may include volatile or non-volatile memory, cache, or other types of memory. The bulk storage 2162 may include flash memory, magnetic hard disk drives (HDDs), and other bulk storage devices.

[0171] The processor 2150 of the client device 2120-1 executes an operating system (OS) 2164 and one or more client applications 2166. The client applications 2166 include applications that access the server 2130 via the distributed communication system 2110. The client applications 2166 may include applications executed by a system computer that controls the tool. The client applications 2166 may also include applications that process data captured by a laser sensor installed on the tool and applications that execute perceptron algorithms.

[0172] 8D illustrates a simplified example of a server 2130-1. Server 2130-1 typically includes one or more CPU / GPU / TPU or processors 2170, a network interface 2178, memory 2180, and bulk storage 2182. In some implementations, server 2130-1 may be a general-purpose server and may include one or more input devices 2172 (e.g., a keypad, touchpad, mouse, etc.) and a display subsystem 2174 including a display 2176.

[0173] The network interface 2178 connects the server 2130-1 to the distributed communications system 2110. For example, the network interface 2178 may include a wired interface (e.g., an Ethernet or EtherCAT interface) and / or a wireless interface (e.g., Wi-Fi, Bluetooth, Near Field Communication (NFC), or other wireless interface). The memory 2180 may include volatile or non-volatile memory, a cache, or other types of memory. The bulk storage 2182 may include flash memory, one or more magnetic hard disk drives (HDDs), or other bulk storage devices.

[0174] The processor 2170 of the server 2130-1 executes one or more operating systems (OS) 2184 and one or more server applications 2186, which may be housed in a virtual machine hypervisor or a containerized architecture with shared memory. The bulk storage 2182 may store one or more databases 188 that store data structures used by the server applications 2186 to perform their respective functions. The server applications 2186 may include applications that process data captured by laser sensors installed on the tool, applications that run perceptron algorithms, applications that use machine learning techniques to train one or more models based on the captured data, and sensor fusion applications that receive and analyze data from various other sensors in the tool and PM.

[0175] In some situations, the data captured by the laser sensor alone may not be sufficient to develop accurate models for autonomous process optimization, PM alignment, predictive maintenance, and guided troubleshooting. Using sensor fusion applications such as Bayesian or Kalman filtering, data from other sensors in the tool can be used in combination with the data captured by the laser sensor to determine with a high degree of certainty the changes required to the process recipe, PM configuration, and details of the cause or root cause of the problem. The server application 2186 may include applications that use the models to analyze the data collected from the laser sensor and other sensors to optimize the process performed in the PM, control the PM configuration, perform PM alignment, determine the root cause of problems in the tool and PM for guided troubleshooting, and use these data as feedback for closed-loop autonomous control of the tool, as described in more detail below.

[0176] Block Diagram FIG. 9A shows an example of a laser sensor installed in the airlock chamber of a tool. By way of example only, the laser sensor 2200 is placed in an observation port in the airlock lid on the airlock chamber 2202. The laser sensor 2200 does not have to be mounted in the airlock chamber 2202, other existing stations on the tool can be used for sensor installation. A dedicated station can be designed in the VTM or EFEM to perform the measurements. It is understood that one or more laser sensors can be placed on new / used edge bond rings and any other locations where the wafer passes (e.g., airlock, aligner, robot arm, VTM, etc.).

[0177] FIG. 9B shows a cross section of a brand new edge bonded ring. Only a half portion of the edge bonded ring is shown to illustrate the thickness variation. FIG. 9C shows a cross section of a used edge bonded ring where the thickness has changed due to erosion from processing. Again, only a half portion of the edge bonded ring is shown to illustrate the thickness variation. The thickness reduction along the radial direction is non-uniform. As shown, more thickness reduction can occur at the inner diameter (ID) of the edge bonded ring than at the outer diameter (OD).

[0178] FIG. 9D shows the wafer before etching. FIG. 9E shows the reduction in thickness of the wafer after etching. The thickness reduction may be non-uniform. FIG. 9F shows the wafer before deposition. FIG. 9G shows the increase in thickness of the wafer after deposition. The thickness increase may be non-uniform. The thickness changes shown in FIGS. 9B-9G can be measured using a laser sensor as follows.

[0179] 9A shows a schematic diagram of the components located within the airlock chamber 2202. Airlock fingers 2300, 2302 include finger pads 2304, 2306, respectively. A wafer (currently shown) is transported by the robot into the airlock chamber 2202 and placed on finger pad 2304. An end effector 2310 of the robot has a pad 2312. An edge bond ring 2314 is transported by the robot into the airlock chamber 2202 and placed on pad 2316 located on the finger 2300.

[0180] A laser sensor 2200 located on the airlock lid emits (i.e., transmits) one or more electromagnetic waves (light / laser beams) onto the edge bonded ring 2314 and receives (i.e., senses or detects) the waves reflected from the edge bonded ring 2314 or the wafer (not shown). In addition to sensing the top surface of the edge bonded ring 2314 or the wafer, the laser sensor 2200 can sense other portions of the edge bonded ring 2314 or the wafer. For example, the laser sensor 2200 can sense the outer diameter (OD) of the edge bonded ring 2314 or the wafer, the inner diameter (ID) of the edge bonded ring 2314, the surface roughness of the edge bonded ring, etc. Multiple laser sensors can be used to sense the geometric parameters of the edge bonded ring 2314 and the wafer in the X, Y, and Z dimensions. Additionally, the frequency of the electromagnetic waves emitted from the sensor 2200 can be controlled and altered depending on the materials of the edge bonded ring 2314 and the wafer.

[0181] Based on measurements of the edge bond ring 2314 and the wafer before and after processing (e.g., etching or deposition), changes in geometric parameters of the edge bond ring 2314 and the wafer can be determined, as shown in Figures 9B-9G. Based on the changes, processes within the PM can be optimized, and PM matching can be performed fleet-wide, as described in more detail below with reference to Figures 12 et seq.

[0182] 10A-10C show a non-limiting example of a sensor 2200. FIG. 10A shows a laser sensor 2201 suitable for 1D (point / spot) height measurement of targets such as edge-coupled rings and wafers using optical reflectometry. The laser sensor 2201 comprises an emitter 2202 and a receiver (also called a detector) 2204. The emitter 2202 may comprise various optical components such as lenses, mirrors, and an intensity controller (not shown). The optical components can be used to lightly focus the laser spot while eliminating beam irregularities, allowing measurement of fine targets when the beam spot size is constant. The receiver 2204 comprises lenses and filters to minimize the effects of optical distortion, as well as an array of sensing elements. The intensity controller improves the resolution of the emission time, which increases the sensitivity in adjusting the light intensity. The intensity controller can shift the laser emission time and the laser power at the target reflectivity. The measurement principle is based on triangulation. Given the known relative positions of the emitter and detector, the position of the target can be calculated by determining the location of the reflected beam spot at the detector.

[0183] A translucent object causes diffuse reflection beneath the surface as the laser beam from the laser sensor 2201 penetrates the interior of the object and the received light wave pattern gently spreads. The laser sensor 2201 can perform detection by counteracting the effect of the expanded wave pattern. For transparent objects, the laser sensor 2201 optimizes the laser intensity by sensing and adjusting the reflected light of each layer of the transparent object. High accuracy is achieved because the measurement is not affected by the reflectivity of each layer. For metal surfaces, the laser sensor 2201 minimizes the effect of roughness of the rough target surface. Surfaces that may appear flat often contain microscopic protrusions and depressions. This microscopic surface roughness can often cause measurement errors with conventional sensors. In contrast, by using a wide beam spot, the laser sensor 2201 can average out the effect of uneven surfaces and obtain stable measurements of rough targets.

[0184] FIG. 10B shows a laser sensor 2210 suitable for 2D (line) height profile measurements across a line using optical reflectometry. The laser sensor 2210 includes an emitter 2212 and a receiver 2214, which may be configured with one or more arrays of receiving sensing elements. The sensing elements may be made of Charged Coupled Device (CCD) sensors or Complementary Metal Oxide Semiconductor (CMOS) sensors to convert light into electronic signals. The laser sensor 2210 can measure height data across a line on objects such as edge-coupled rings and wafers. Typically, multiple sensors or sensor movements are required for measurements such as height difference and warpage. Instead, the laser sensor 2210 can perform these measurements using a 2D laser. A sharp line beam is formed on the optical receiving element by focusing a short-wavelength laser with a 2D lens. The receiving light density of the laser can be increased to ensure a higher level of receiving light intensity. This achieves stable and accurate measurements on all types of objects including edge-coupled rings and wafers.

[0185] The laser sensor 2210 can use dual polarization capability to distinguish and cancel multiple reflections of the light beam that act as obstacles to the measurement. This capability is particularly beneficial in measuring materials with complex shapes and complex areas, such as edge-coupled rings and wafers.

[0186] FIG. 10C shows a laser sensor 2220 suitable for 3D height measurement over an area of ​​an object using white light interferometry. The light emitted from the light source of the laser sensor 2220 is split into two by a beam splitter. One beam reflects off the target (both incident and reflected beams are shown at 2222) and the other reflects off the reference mirror of the laser sensor 2220. Both beams then hit the optical receiving element of the laser sensor 2220 as interfering light waves. These interfering light waves provide the highest level of interference when their optical path lengths match. The optical unit including these components of the laser sensor 2220 moves up and down to obtain contrasting images. For each measurement point of these images, the laser sensor 2220 reads the position of the optical unit at the highest level of interference. This information is used to determine the distance to the target. Thus, the laser sensor 2220 accurately captures the 3D shape of objects such as edge-coupled rings and wafers.

[0187] The laser sensor 2220 uses white light interferometry to overcome problems that impede the accuracy of profile measurements, such as differences in target materials and colors, as well as areas with dead zones that typically cannot be measured. The laser sensor 2220 emits light in a wide dynamic frequency range, which allows targets with different reflectivities to be measured simultaneously from a single captured image. The process is not affected by internal reflections, so accurate shapes can be captured for targets made of semi-transparent materials. Traditional triangulation methods can block light returning to the sensor, causing dead zones that cannot be measured. The laser sensor 2220 avoids this limitation using the techniques described above.

[0188] In general, the laser sensor 2200 can sense and provide data that can be processed to determine overall shape measurements such as thickness, shape, contour, flatness, and surface roughness analysis of the edge bond ring 2314 and thickness of the wafer. For example, by combining measurements of the top and side surfaces of the edge bond ring 2314, the data sensed by the laser sensor 2200 after processing can provide measurements of the undercut and overall thickness of the edge bond ring 2314. Similar techniques can also be used with wafers.

[0189] The light-emitting element of a laser sensor can be a semiconductor laser. The laser light is focused through a light-emitting lens and projected onto the object, at which point a portion of the light beam reflected from the object generates a light spot on a position sensing device (PSD).

[0190] Some of the receiving elements use linear image sensors rather than PSDs. PSDs can only obtain information about the central position based on the amount of light in the entire light spot. On the other hand, light-emitting elements with linear image sensors detect the amount of light received by each cell. Therefore, even if there are variations in the amount of light in the spot due to influences from the object's surface, accurate detection can be performed using the peak position of the light intensity. This significantly reduces errors due to the influence of variations in the object's surface.

[0191] Below is an overview of the various functions of the laser sensor 2200. The laser sensor 2200 can measure various parameters of the edge bond ring 2314 and the wafer as follows: The thickness and / or width of targets such as the edge bond ring 2314 and the wafer can be measured in a number of ways. Two-sensor systems are common, but measurements with one sensor are possible. In confocal reflectometry, thickness measurements are performed using two sensors to sandwich the target. With a transparent target, thickness can be measured directly with a single sensor on either side. In triangulation reflectometry, thickness measurements are performed using two sensors to sandwich the target. With a transparent target, thickness can be measured directly by exposure from only one side. In specular interferometry, two-sensor systems are common, but measurements with one sensor are possible if the position of one side is pre-determined or the target is transparent. Thickness measurements can also be performed by measuring the height of the target as a step height relative to a reference plane.

[0192] Height can also be measured by detecting the position of the top surface of the object. For objects with an unfixed bottom surface, the shape is measured directly by using two sensors or by projecting a laser on a horizontal line. In the light cut method, the position of the top surface of the object is detected. The ability to perform non-contact 2D profile measurements makes it possible to perform step measurements with respect to a reference surface on the target. In the confocal reflection method, the position of the top surface of the object is detected. Thus, stable measurements are possible even if the target moves. In the optical through-beam method, by capturing the shadow of the target's shape, the sensor can detect any position. Simultaneous measurements of multiple points are also possible.

[0193] The warpage or flatness is generally measured by detecting a target, such as the edge bond ring 2314, at three points. Measurements can also be made by moving and scanning the target under a point-type sensor, or by measuring the shape directly by projecting a laser on a horizontal line. In the light-cut method, the shape is measured directly by exposing the target to a line laser. The warpage or flatness is then measured using the measured waveform data. 3D shape measurements are also possible by moving the target during the scan. In the confocal reflectance method, the use of ultra-small and lightweight sensors allows multiple heads to be installed in parallel. By capturing height data at multiple locations on the target, the warpage or flatness can be calculated. In the triangulation reflectance method, the height data of any three points is used to measure the warpage or flatness after scanning the target with a sensor.

[0194] The laser sensor 2200 can measure the shape of a target by scanning over the target with a single point optical sensor or by using a laser line to generate a 2D profile or 3D height profiler of the surface. In the confocal reflectance method, a profile measurement can be performed using a spot laser to scan the target. The target can be measured regardless of material or color. In the light cut method, the cross-sectional shape of the target is instantly measured using a line laser. Laser sensors can detect the height, width, and cross-sectional area of ​​objects such as edge bonded rings and wafers.

[0195] The laser sensor 2200 can also measure the shape of a target by scanning the target to measure its 2D shape, or by using a 2D line laser to directly measure the shape of the object. In the light cut method, a line laser is used to continuously scan the cross-sectional profile. The waveform data is combined to create a three-dimensional profile. In the 3D interferometry method, height data of multiple (hundreds or thousands) points is instantly acquired for the measurement area (see, for example, FIG. 10C). By utilizing the white light interference principle, it is possible to perform micron-level high-precision measurements that are not affected by material type or color and have no blind spots. In the 3D pattern projection, multiple stripe patterns can be projected at high speed in patterns from two directions. The light reflected from the target is analyzed in real time to create a 3D shape image of objects such as edge-coupled rings and wafers.

[0196] Figure 11 illustrates a measurement system 2400 that processes data captured by one or more laser sensors 2402 mounted on a tool 2404 (e.g., any of the tools shown in Figures 1-6C). For example, although only one tool is shown, the measurement system 2400 may include multiple tools similar to tool 2404. The laser sensor 2402, which may be any of the laser sensors 2200 shown in Figures 10A-10C, senses data from an edge coupling ring 2406 (e.g., edge coupling ring 2314) and the wafer, as described with reference to Figures 9A-9G above.

[0197] The data processor 2408 processes the data captured by the laser sensor 2402. For example, the data processor 2408 may clean up and / or format the data. For example, the data processor 2408 may perform operations such as noise filtering, feature extraction, etc. The data processor 2408 may be located on the laser sensor 2402 or may be external to the laser sensor 2402 (e.g., on a PCB).

[0198] The system computer 2410 controls the tool 2404. In one embodiment, raw data from the laser sensor 2402 may be sent to a remote data center 2412 for processing and / or analysis. Alternatively, in another embodiment, the raw data may be processed by a data processor 2408, which may be located on the laser sensor 2402 or on a PCB outside the laser sensor 2402, and only a portion of the analyzed data may be sent to the system computer 2410 and / or the remote data center 2412. The system computer 2410 may send the raw data to the remote data center 2412 in real time or in batches via the distributed communication system 2010 shown in Figures 8B-8D. In addition, the system computer 2410 may also send data collected by other sensors in the tool 2404 and by sensors in the PM of the tool 2404 to the remote data center 2412 for training models, as described below with reference to Figures 12 et seq.

[0199] For example, the remote data center 2412 may include multiple servers 2030 as shown in Figures 8B-8D. At the remote data center 2412, several models are trained based on data collected from multiple tools, such as tool 2404, using machine learning. The training and functioning of the models are described in detail below. Using the trained models received from the remote data center 2412, the system computer 2410 operates the tool 2404 based on new data collected by the laser sensor 2402 and other sensors during use (manufacturing), as described below.

[0200] Briefly, the system computer 2410 performs various actions recommended by the trained model on the tool 2404. For example, the actions include measuring geometric parameters (e.g., thickness, width, OD / ID, warpage / flatness, 2D / 3D shape, surface roughness, etc.) of the edge bond ring and wafer, and based on the measurements, performing automatic adjustment (optimization) of process parameters in PM, PM alignment, recommending preventive maintenance (e.g., PM auto-clean), guided troubleshooting actions such as repair / replacement of parts / subassemblies, etc.

[0201] For example, the system computer 2410 can measure parameters of the edge bond ring to determine the effective etch rate in the PM (and can measure parameters of the wafer to determine the etch / deposition rate), monitor drift over time, and use this data to autonomously optimize the process in the PM, achieve PM matching, and perform preventative / corrective actions to compensate for such drift. For example, the system computer 2410 can fine-tune the height of the edge bond ring above the chuck by adjusting the lift pins. For example, the system computer 2410 can determine the optimal duration during which wet clean, waferless automatic clean (WAC), and cover wafer area clean (CWAC) procedures can be used.

[0202] Additionally, the system computer 2410 can implement various adaptations recommended by the trained model. The system computer 2410 can automatically adjust the recipe used in the PM based on the reduction in thickness of the incoming / outgoing edge bond ring and based on the change in thickness of the wafer. The system computer 2410 can determine the effective etch rate in the PM based on the detected change in inner diameter (i.e., change in edge erosion profile) or the thickness of the incoming / outgoing edge bond ring. The system computer 2410 can determine the effective etch rate or deposition rate in the PM based on the detected change in thickness of the pre-processed / post-processed wafer. The system computer 2410 can automatically adjust the recipe for PM matching and autonomous process control and optimization. These and other functions implemented by the measurement system 2400 based on the data captured by the laser sensor 2402 in a closed-loop manner will now be described in detail.

[0203] The measurement system 2400 can assist with PM matching. PMs are periodically cleaned after a certain number of wafers have been processed. However, different wafer types may have different materials that outgas into the PM. In addition, different processes performed on wafers in the PM may leave different by-products in the PM that need to be cleaned. Using the same cleaning recipe for all incoming wafer types may not properly clean the chamber to process the next wafer. The measurement system 2400 can leverage historical data on PM configuration, cleaning processes performed on the PM, and how it affected process results, learn from the data, and assist the PM in selecting the appropriate recipe to automatically clean the PM, so that the PM is ready to process the next incoming wafer.

[0204] The measurement system 2400 can measure effective etch and deposition rates at different tools, monitor drift over time, record changes after maintenance (e.g., wet clean, CWAC), and create procedures to automatically adjust the parameters of the etch and deposition process in the PM (e.g., recipe, ESC temperature, gas flow, duration, etc.) to correct such changes. The measurement system 2400 can measure the thickness change between a brand new edge bond ring going into the PM and the same edge bond ring (and / or thickness change of pre-processed / post-processed wafers) leaving the PM at the end of its life or any time in between, and automatically adjust the PM's process parameters (e.g., etch / deposition rate, etch / deposition uniformity, etc.) for autonomous process optimization and inter-PM matching. The laser sensor 2402 precisely measures the thickness reduction of the used edge bond ring or the thickness change of the processed wafer. By knowing the usage history of the edge bond ring (e.g., number of RF hours the edge bond ring was used in the PM, process recipe, chamber configuration) and its thickness reduction, the effective erosion rate of the PM can be determined. Similar techniques can also be applied to the etch and deposition processing of wafers. Thus, the metrology system 2400 can automate measurements of inbound / outbound (new / used) edge bonded rings and wafers, and can automate process optimization and inter-PM matching.

[0205] Not only does the measurement system 2400 capture data from edge bond rings and wafers within areas of the tool such as the airlock, but the measurement system 2400 can work in conjunction with other data collected by many other systems located in the tool 2404 (e.g., sensors, cameras associated with the PM, etc.). For the measurement system 2400 to be reliable, many factors such as the orientation and leveling of the laser sensor 2402 need to be relatively constant. To allow for variations in these factors, which may be unavoidable, the data collected by the measurement system 2400 may be normalized using data collected by these other systems to minimize or eliminate the effects of discrepancies in these factors. In-situ calibration of the laser sensor may be performed where measurements are performed on objects with known heights and a correction curve is created and applied to measurements during normal operation.

[0206] The location of the laser sensor 2402 can vary and does not have to be restricted to the airlock. For example, a dedicated station can be designed in the VTM or EFEM to perform the measurements described below. The laser sensor 2402 can be located at other locations in the tool 2404. Additionally, the laser sensor 2402 can be located on tools throughout the fleet of tools. This can expand the database of captured data, improving the training of models built in the remote data center 2412 based on the captured data and making the models more robust. A robust model can perform the functions of the measurement system 2400 with increased statistical confidence, which can improve the overall performance of the PM, the tools, and the fleet of tools.

[0207] The tool 2404 can be optimized based on the data accumulated by the measurement system 2400. For example, the number of cleaning cycles of the PM in the tool 2404 can be optimized. For example, cleaning of the PM in the tool 2404 may be recommended by another mechanism in the tool 2404 but postponed or delayed based on the data captured by the laser sensor 2402. Notably, the data is immediately available and analyzed in real time or near real time by the measurement system 2400. Thus, the optimization of the tool 2404 can be performed on the fly. Using the measurement system 2400, a model can be trained in the remote data center 2412 to receive the data captured by the laser sensor 2402 as input and provide an output that controls when cleaning can be performed on the PM. It can be seen that this procedure fully automates PM cleaning via WAC or CWAC in a closed-loop manner.

[0208] The measurement system 2400 can analyze the edge bond ring and the wafer as follows: The edge bond ring is typically transferred from the EFEM to the airlock, to the PM where it is mounted around the ESC. The edge bond ring has a particular thickness, inner diameter, and thickness profile across the radius. These and other parameters of the edge bond ring are observed by capturing data from the edge bond ring before it is transferred to the PM. In the PM, the edge bond ring is exposed to the process performed on the wafer. As a result, one or more parameters of the edge bond ring may change. The edge bond ring is designed for use for a particular number of RF hours. Similar edge bond rings can be used in PMs performing similar processes for similar periods of time (i.e., number of RF hours).

[0209] Then, once the edge bond rings are fully used (or daily without waiting for the end of their life), they are transferred from the tool to a load lock via a robot. Data from these edge bond rings exiting the PM are captured and geometric parameters are measured and correlated by the measurement system 2400. For example, one edge bond ring may show more variation than the other. The variation of one edge bond ring is analyzed, optionally together with other data (e.g., from other sensors in the tool 2404), to autonomously optimize the process and achieve PM alignment, and optionally identify the root cause of the variation or control the operating conditions of the PM to achieve optimal on-wafer process quality.

[0210] For example, based on an analysis of changes indicating that one of the PMs is behaving differently (i.e., having a different effect on the edge coupling ring), but the same process is performed on that PM for the same number of RF hours as the other PMs, the recipe of the process performed on that PM may be adjusted using a model trained by the measurement system 2400. Alternatively, the positioning (height, tilt angle, etc.) of the edge coupling ring on that PM may be adjusted using a model trained by the measurement system 2400. A similar procedure may be used on a wafer to measure thickness changes before and after etch / deposition to achieve autonomous process optimization and PM matching.

[0211] Using the measurement system 2400, different models can be trained to perform these different functions. Once fully trained, these models can receive data from the edge bond ring and wafer as inputs, as well as all other on-tool sensors, and can provide outputs such as adjustments to be made to the recipe, the height of the lift pins of the edge bond ring, the spatially dependent temperature profile of the chuck, etc. Continuous training based on data acquired during use (i.e., manufacturing) can refine (i.e., improve accuracy) the output of these models. It can be seen that such a system enables autonomous control of the tool for autonomous process optimization and PM matching by measuring the geometric parameters of the edge bond ring and wafer as described above.

[0212] In summary, several edge-coupled rings allow better control of the plasma field in the etch PM to achieve optimized on-wafer process results. These edge-coupled rings can be automatically transferred from the EFEM to the PM via a transfer module (TM), which makes it possible to achieve continuous production without the need to open the PM for long periods of time (e.g., up to a year). However, due to differences in PM conditions within a tool or across different tools, the on-wafer results are not always similar.

[0213] To achieve inter-PM matching and optimize process results, the measurement system 2400 provides techniques to measure effective etch or deposition rates across different tools, monitor drift over time, record changes after maintenance, and create procedures to automatically adjust PM etch and deposition process parameters (e.g., recipe, ESC temperature, gas flow, duration, bias voltage, etch / deposition rate, etch / deposition uniformity, etc.) to compensate for such changes. The measurement system 2400 enables automated measurement of inbound / outbound (new / used) edge-coupled ring parameters and pre-processed / post-processed wafer parameters, as well as corresponding systems for autonomous process optimization and inter-PM matching.

[0214] Generally, several control parameters of the tool can be monitored (observed) using the measurement system 2400. Based on the observed data for these parameters, the measurement system 2400 can train a model to verify whether the automatic process control loops used in the tool 2404 are operating properly. The measurement system 2400 can use the trained model to automatically adjust the process parameters (e.g., etch / deposition rate, etch / deposition uniformity, etc.) for PM matching.

[0215] Flowcharts and Models First, Figures 12 and 13 show broad and general examples of various methods implemented by the server 2030 and the system computer 2410 shown in Figure 11. Next, Figures 14A to 14I show specific and detailed examples of various methods implemented by the server 2030 and the system computer 2410 shown in Figure 11.

[0216] First, Figure 12 illustrates broadly and generally the methodology implemented by the server 2030 to train various models using perception, sensor fusion, and machine learning algorithms based on data captured by the measurement system 2400 from various tools. Then, Figure 13 illustrates broadly and generally the methodology implemented by the server 2030 and system computer 2410 using the trained models in real time (i.e., on the fly, or as the tool is being used to process wafers during production). Details of model generation and training are described with reference to Figures 15A-16B.

[0217] FIG. 12 illustrates a method 2500 for capturing data from edge coupled rings and wafers from various tools and using perception, sensor fusion, and machine learning algorithms to train various models based on the captured data. At 2502, the method 2500 captures data from edge coupled rings and wafers at locations such as load locks within the tool. At 2504, the method 2500 processes the captured data. For example, the captured data may be cleaned up, formatted, and converted into signatures that indicate patterns in the captured data. For example, the method 2500 may perform operations such as noise filtering, geometric transformations, feature extraction, and the like. Additionally, features and signatures may be recognized and extracted from the captured data that may be used to build feature vectors to determine the erosion profile of the edge coupled rings, determine the thickness change of the wafer, determine effective etch / deposition rates in PM, autonomously optimize the process in PM, and for PM alignment.

[0218] At 2506, method 2500 uses perception, sensor fusion, and machine learning algorithms to train various models to perform specific tasks based on data collected by sensors and laser sensors in the PM over time (i.e., by using these data as inputs to the models). For example, the tasks may include, but are not limited to, determining edge coupling ring erosion profiles, determining wafer thickness changes, determining effective PM etch / deposition rates, tuning (optimizing) PM recipes, determining WAC recipes and frequencies, etc. Method 2500 establishes correlations between the performance of these tasks and the signatures and feature vectors generated based on the collected data and stores them in a database. Thus, method 2500 trains models to determine edge coupling ring erosion profiles, determine wafer thickness changes, determine PM etch / deposition rates, autonomously optimize PM processes, and perform PM matching based on signatures and features. At 2508, method 2500 continues to improve (ie, further train) these models based on data collected and processed using the continually trained models.

[0219] FIG. 13 illustrates a method 2600 for capturing data during manufacturing from edge bond rings and wafers from various tools and using trained models to determine erosion profiles of edge bond rings, determine thickness changes of wafers, determine effective etch / deposition rates in PM, autonomously optimize processes in PM, PM matching, and the like. At 2602, the method 2600 captures data during manufacturing from edge bond rings and wafers at locations such as load locks. At 2604, the method 2600 processes the captured data. For example, the captured data may be cleaned up, formatted, and converted into signatures indicative of patterns in the captured data. For example, the method 2600 may perform operations such as noise filtering, geometric transformations, feature extraction, and the like. Additionally, features may be recognized and extracted from the captured data that can be used to determine erosion profiles of edge bond rings, determine thickness changes of wafers, determine etch / deposition rates in PM, autonomously optimize processes in PM, and for PM matching.

[0220] At 2606, using the various trained models, method 2600 performs various tasks based on the collected data. For example, method 2600 sends signatures and features and data from the PM to the trained models and receives commands / data from the trained models that the tool uses to automatically perform various tasks. For example, the tasks may include, but are not limited to, determining the erosion profile of the edge bond ring, determining the thickness change of the wafer, determining the etch / deposition rate in the PM, autonomous optimization of the process in the PM, PM matching, etc. At 2608, method 2600 updates (i.e., further trains) these trained models based on data collected and processed using the trained models during manufacturing.

[0221] Additional Models 14A-14I illustrate various additional examples of methods implemented by server 2030 and system computer 2410 using models trained based on data captured by laser sensor 2402 in near real-time (i.e., on the fly or as the tool is being used to process wafers during production). The models used in these methods can be trained using measurement system 2400 and can be used to implement the operations described below. These are non-limiting examples. Additional use cases are contemplated.

[0222] In these examples, the training of the models using data from the PM and data captured by the laser sensor 2402 to perform the described functions using machine learning algorithms including perception, sensor fusion, and continuous updating of the trained models is similar to the training of the models described above with reference to FIG. 12. Thus, the training of the models used in the following methods is estimated and not described for brevity. In use, similar to the trained models described with reference to FIG. 13, these trained models receive data from the laser sensor 2402 as input, and the trained models output commands / data to tools to automatically perform actions such as autonomously adjusting process parameters (e.g., etch / deposition rate, etch / deposition uniformity, etc.), performing PM matching, assisting in repair / replacement of parts or subassemblies, etc.

[0223] FIG. 14A illustrates a method 2700 for optimizing process parameters (e.g., etch / deposition rate, etch / deposition uniformity, etc.) and controlling PM configuration across a fleet of tools using models trained using perception, sensor fusion, and machine learning algorithms. At 2702, method 2700 captures and processes data regarding thickness variations of wafers and edge bond rings in PMs across the fleet of tools. At 2704, based on the thickness variation data, method 2700 uses the trained model to determine variations in etch / deposition process characteristics (process rate, process uniformity, etc.) in PMs across the fleet of tools. At 2706, based on the variations in process characteristics, method 2700 receives data from the trained model to automatically optimize process parameters (e.g., etch / deposition rate, etch / deposition uniformity, etc.) and control PM configuration across the fleet of PMs. Thus, method 2700 automatically optimizes process parameters (e.g., etch / deposition rate, etch / deposition uniformity, etc.) and controls PM configuration across the fleet of PMs in a closed-loop manner.

[0224] FIG. 14B illustrates a method 2710 for measuring edge bond ring flatness and wafer thickness variation using models trained using perception, sensor fusion, and machine learning algorithms, and performing PM matching based on the measured edge bond ring flatness and wafer thickness variation. At the start, the wafer / ring is presumed to be located in the airlock. The ring may be new and is sent to the PM for the first time and measured as described in the steps below. Alternatively, the ring may be in use for a period of time and is sent to the PM for further use according to the steps below. The wafer may be unprocessed and is sent to the PM for processing and measured as described in the steps below. Alternatively, the wafer may be partially processed and is sent to the PM for further processing following the steps below.

[0225] At 2712, method 2710 determines whether the edge bonded ring / wafer is transported from the airlock to the PM. At 2714, if the edge bonded ring / wafer is transported from the airlock to the PM, method 2710 collects data from the edge bonded ring / wafer using a laser sensor before the edge bonded ring / wafer is transported in a pre-processed state to the PM. At 2718, method 2710 determines geometric parameters (e.g., flatness, thickness, surface roughness, etc.) of the edge bonded ring / wafer based on the data collected by the laser sensor. At 2720, method 2710 transports the edge bonded ring / wafer from the airlock to the PM.

[0226] At 2722, method 2710 determines whether it is time to inspect the edge bond ring again. For example, inspection of the ring may be performed daily after the edge bond ring has been used for a predetermined number of RF hours or when triggered by an error event. For example, inspection of the wafer may be performed after each process (etch / deposition). At 2724, if it is time to inspect the edge bond ring / wafer, method 2710 transports the edge bond ring / wafer from the PM to the airlock. At 2726, method 2710 collects data from the edge bond ring / wafer using a laser sensor. At 2730, method 2710 determines geometric parameters (e.g., flatness, thickness, etc.) of the edge bond ring / wafer based on the data collected by the laser sensor.

[0227] At 2732, method 2710 determines a thickness change of the edge bonded ring / wafer based on the geometric parameter values ​​determined at 2718 and 2730. At 2734, method 2710 sends the thickness change to the trained model from which a determination of the process characteristics of the PM is received. At 2736, method 2710 automatically adjusts one or more process parameters of the PM (e.g., etch / deposition rate, etch / deposition uniformity, etc.) and / or PM configuration (e.g., top / bottom electrode gap / alignment / tilt, ring height above ESC) based on the thickness change and the determination of the process characteristics (erosion / deposition rate, uniformity, etc.). For example, method 2710 can automatically adjust these parameters in a closed-loop manner based on the values ​​of the parameters received from the trained model based on the thickness change data.

[0228] FIG. 14C illustrates a method 2740 for using a model trained using perception, sensor fusion, and machine learning algorithms to determine changes in pre-processing / post-processing geometric parameters (e.g., thickness, surface roughness, and ID edge shape / profile) of edge bonded rings across a fleet of tools and adjust PM operations based on surface roughness and ID edge shape / profile. At 2742, method 2740 uses laser sensors to collect data from edge bonded rings across the fleet. At 2744, method 2740 uses the trained model to determine changes in pre-processing / post-processing geometric parameters (e.g., thickness, ID, surface roughness) of edge bonded rings. This can be thickness measurements at points / spots (1D), across lines (2D), or over areas (3D). Surface roughness can only be calculated from 3D measurements and can be estimated via 2D measurements. At 2748, method 2740 uses the trained model to determine the ID edge shape / profile of the edge bonded rings. At 2750, method 2740 adjusts PM operations using the trained model. For example, method 2740 adjusts the automatic cleaning process (WAC or CWAC) in PM based on data from surface roughness and edge bond ring ID edge shape / profile, as well as other PM sensors using the trained model. Similar methods can be used to determine the shape and profile of the wafer, which can be used to adjust PM operations.

[0229] FIG. 14D illustrates a method 2760 for guided troubleshooting of PM issues in tools based on ID edge shape / profile or top surface slope profile of edge coupled rings using models trained using perception, sensor fusion, and machine learning algorithms. At 2762, method 2760 uses laser sensors to collect data from edge coupled rings across the fleet. At 2764, method 2760 uses the trained model to determine changes in pre-processing / post-processing geometric parameters of edge coupled rings (e.g., thickness, ID, surface roughness). This can be thickness measurements at points / spots (1D), across lines (2D), or over areas (3D). Surface roughness can only be calculated from 3D measurements and can be estimated via 2D measurements. At 2766, method 2760 uses the trained model to determine the ID edge shape / profile of the edge coupled ring. The ID edge shape / profile can be stored as a signature of the edge coupled ring in a database. The signatures are correlated to performance and / or malfunction of parts and / or subassemblies of the PM, and thus a particular signature may be indicative of a particular performing and / or malfunctioning part and / or subassembly in the PM.

[0230] At 2750, method 2760 sends to the trained model a signature of the edge coupled ring generated based on data captured from the edge coupled ring using a laser sensor. Method 2760 receives instructions to identify the problematic part or subassembly of the PM along with instructions for maintenance / replacement of the problematic part or subassembly. Additionally or alternatively, in some circumstances, method 2760 receives data from the trained model that automatically adjusts the operating parameters of the PM in a closed loop manner. For example, the data received from the trained model can be used to adjust the height of the edge coupled ring, adjust the temperature profile of the ESC, change the flow rate of the process gas, etc. If all of these options have already been performed, the data can indicate a final option to repair or replace components of the PM. Thus, method 2760 can automatically implement corrective actions and assist in guided troubleshooting problems with PM in a hierarchical manner. Similar procedures can be used to troubleshoot PM problems based on wafer profile measurements based on thickness changes, and descriptions are omitted for brevity.

[0231] FIG. 14E illustrates a method 2770 for adjusting a PM cleaning process based on data collected by laser sensors from edge-coupled rings / wafers using models trained using perception, sensor fusion, and machine learning algorithms. At 2772, the method 2770 uses laser sensors to collect data from edge-coupled rings / wafers across the fleet. At 2774, the method 2770 automatically selects a recipe type and frequency to clean PMs based on data collected from edge-coupled rings / wafers using the models trained in a closed-loop manner. At 2776, the method 2770 automatically selects values ​​for parameters for PM operation using the models trained in a closed-loop manner.

[0232] FIG. 14F illustrates a method 2780 for adjusting PM configuration (e.g., automatic cleaning recipe, top / bottom electrode gap and alignment / tilt, edge ring height above ESC, etc.) based on data collected by a laser sensor from an edge coupled ring / wafer using a model trained using perception, sensor fusion, and machine learning algorithms. At 2782, method 2780 collects data from the edge coupled ring / wafer using a laser sensor. At 2784, method 2780 determines PM performance based on the collected data (e.g., ring condition, drift in wafer thickness reduction (etch rate), or uniformity of wafer thickness change, etc.). For example, sensor data collected after a previous cleaning procedure is saved in a database as a signature of the edge coupled ring / wafer. Thus, correlations between the signature and the quality of the cleaning procedure performed on the PM can be stored in the database. The currently collected data is converted into a signature (feature vector) that is input to the trained model. The trained model compares the input signature to the signatures stored in the database.

[0233] At 2786, based on the comparison, the trained model returns an evaluation on the PM performance. At 2788, based on the comparison and / or evaluation, the trained model also provides parameters for fine-tuning the PM configuration (e.g., auto-clean recipe, top / bottom electrode gap and alignment / tilt, edge ring height above ESC, etc.). Thus, method 2780 automatically fine-tunes the PM configuration in a closed-loop manner. When subsequently performed, method 2780 verifies whether the fine-tuning has actually improved PM performance. The results of the validation are used as feedback to continually improve the trained model.

[0234] FIG. 14G illustrates a method 2790 for determining etch rate variations in a PM using a model trained using perception, sensor fusion, and machine learning algorithms and adjusting operational parameters in the PM based on data collected by a laser sensor from an edge bond ring used in the PM. At 2792, the method 2790 transports an edge bond ring from the PM to an airlock and collects data from location A on the edge bond ring using a laser sensor. At 2794, the method 2790 uses the trained model to determine a thickness of the edge bond ring at location A. At 2796, the method 2790 rotates the edge bond ring or scans the edge bond ring via a robot under the laser sensor. At 2798, the method 2790 collects data from location B on the edge bond ring using a laser sensor. At 2800, the method 2790 determines a thickness of the edge bond ring at location B.

[0235] At 2802, method 2790 determines an effective etch rate variation based on the thickness at locations A and B. For example, method 2790 sends the thickness at locations A and B as inputs to a trained model, which quantifies the etch rate non-uniformity at the PM. In addition, the trained model returns as output values ​​of one or more operational parameters of the PM to be adjusted. At 2804, method 2790 automatically adjusts one or more process parameters (e.g., etch / deposition rate, etch / deposition uniformity, etc.) and / or operational parameters of the PM based on the measured etch rate non-uniformity in a closed-loop manner. A similar method can be used with wafers for etch and deposition processing, where etch / deposition rate and uniformity variations are ascertained based on the wafer thickness change and one or more process parameters (e.g., etch / deposition rate, etch / deposition uniformity, etc.) and / or operational parameters of the PM are automatically adjusted based on the etch / deposition rate variations in a closed-loop manner.

[0236] FIG. 14H illustrates a method 2810 for guided troubleshooting and correcting issues in a PM based on data collected by a laser sensor from an edge-coupled ring / wafer used in the PM using a model trained using perception, sensor fusion, and machine learning algorithms. At 2812, the method 2810 transports the edge-coupled ring / wafer from the PM to an airlock and collects data from the edge-coupled ring / wafer using a laser sensor. At 2814, the method 2810 sends the collected data to the trained model. For example, the method 2810 converts the collected data into a signature (feature vector) and sends the signature as an input to the trained model. The trained model returns one or more multiple outputs by comparing the signature to various signatures that correlate with properly operating and malfunctioning parts and subassemblies of the PM and preventive / corrective actions. For example, one output may be for gas flow rate, another output for ESC temperature distribution, another output for PM pressure, etc. For example, the output may include commands for preventive / corrective actions to be automatically performed at the PM, or instructions indicating a part or subassembly repair or replacement to be performed at the PM or tool. At 2816, method 2810 receives data from the trained model. At 2818, method 2810 automatically performs the action indicated by the output to adjust PM configuration and / or process parameters in a closed-loop manner, or prompt to perform a repair or replacement.

[0237] FIG. 14I illustrates a method 2820 for monitoring the weight of the edge bond ring / wafer and adjusting the process parameters (e.g., etch / deposition rate, etch / deposition uniformity, etc.) and / or operational parameters of the PM based on the weight changes of the edge bond ring / wafer pre-processing / post-processing. For example, the weight of the edge bond ring / wafer may be monitored using a weight sensor and used along with data collected by a laser sensor from the edge bond ring / wafer as input to a model. The process parameters (e.g., etch / deposition rate, etch / deposition uniformity, etc.) and / or operational parameters of the PM may be adjusted based on the observed weight changes using a model trained using perception, sensor fusion, and machine learning algorithms.

[0238] At 2822, method 2820 transports the edge bonded ring / wafer from the PM to an airlock and collects data from the edge bonded ring / wafer using a weight sensor or a laser sensor. For example, if a weight sensor is used, method 2820 can obtain the weight of the edge bonded ring directly from the weight sensor. Thus, at 2824, method 2820 determines the weight of the edge bonded ring / wafer.

[0239] At 2826, method 2820 determines whether the weight of the edge bonded ring / wafer has changed, and if so, by how much. For example, method 2820 sends data indicating the weight of the edge bonded ring / wafer as an input to the trained model. The trained model determines the change in weight of the edge bonded ring / wafer and returns an output based on the comparison. For example, a database referenced by the trained model may store a correlation between the weight or weight change and a process parameter adjustment or preventive / corrective action to be performed at the PM based on the weight or weight change. For example, the output may include a command for a preventive / corrective action to be automatically performed at the PM. For example, the preventive / corrective action may include changing an operating parameter of the PM and / or changing a parameter of a process, including a cleaning process, used at the PM. At 2828, if the weight of the edge bonded ring / wafer has changed, method 2820 receives the data output by the trained model, and method 2820 automatically performs the action indicated by the output in a closed-loop manner.

[0240] In some implementations, the measurement system 2400 can send raw data collected by the laser sensor from the tool to a remote data center, and all of the data processing and model training can be performed at the data center. Additionally, data processing and recommendations by the trained models during use of the tool can be performed at the data center. In such implementations, the measurement system 2400 can be deployed at the data center as software as a service (SaaS).

[0241] Training and validating the model Figures 15A and 15B show example methods for generating, training, and validating models such as those described with reference to Figures 12A-14I above. These methods are implemented by one or more applications executed by server 2030, which are referred to as control in the method descriptions that follow.

[0242] FIG. 15A illustrates a method 2900 for generating a model. At 2902, the control selects a machine learning model (e.g., a supervised or unsupervised classification or regression model, e.g., a deep neural network, tensorflow, etc.) to define and build the model. The machine learning model does not have to be a neural network. Some examples of machine learning algorithms that can be used include K-nearest neighbors, support vector machines, K-means clustering, deep neural networks, linear or nonlinear regression, etc. For example, the control can define the model by selecting a network topology for the neural network.

[0243] An example of a neural network is shown in FIG. 16A. For example, the selection of the network topology includes selecting the number of inputs, the number of neurons, the number of hidden layers, and the number of outputs. FIG. 16A and FIG. 16B and the corresponding description below describe the model generation in further detail. At 2904, the control trains the model. FIG. 15B and the corresponding description below describe the model training in further detail. At 2906, the control validates the model using any of the well-known validation methods such as K-fold cross-validation, leave-one-out cross-validation, re-permutation, bootstrap, random subsampling, etc. At 2908, the control iteratively fine-tunes the model parameters in a closed-loop manner.

[0244] FIG. 15B illustrates a method 2950 for training a model. At 2952, control runs the model against training data (e.g., data from a laser sensor collected over several months). At 2954, control checks the output predicted by the model against the labeled training data. At 2956, control adjusts the model's parameters and / or network topology to generate a better match between the model's predictions and the actual data (e.g., minimize the error between the model's predictions and the training data). At 2958, control determines whether the model meets some predetermined training criteria. If the model does not meet the predetermined training criteria, control returns to 2952. If the model meets the predetermined training criteria, control ends.

[0245] 16A and 16B show examples of neural networks used to generate models such as those described above with reference to FIGS. 12A-14I using machine learning techniques. Machine learning is a methodology used to devise complex models and algorithms useful for predictions (e.g., predictions for adjusting PM recipe parameters, scheduling services, and replacing components as described above). Models generated using machine learning such as those described above with reference to FIGS. 12A-14I can yield reliable and repeatable decisions and outcomes and reveal hidden insights by learning from historical relationships and trends in the data.

[0246] The goal of using a neural network based model and training the model using machine learning as described above with reference to Figures 12A-15B is to directly predict the dependent variables without casting the relationships between the variables into a mathematical form. A neural network model includes a large number of virtual neurons operating in parallel and arranged in layers. The first layer is the input layer, which receives the input data. Each successive layer modifies the outputs from the previous layer and sends them to the next layer. The last layer is the output layer, which generates the output of the system.

[0247] FIG. 16A shows an example of a fully connected neural network where each neuron in a given layer is connected to each neuron in the next layer. In the input layer, each input node is associated with a numerical value, which can be any real number. In each layer, each connection leaving an input node has a weight associated with it, which can be any real number (see FIG. 16B). In the input layer, the number of neurons is equal to the number of features in the dataset. The output layer can have multiple continuous or discrete outputs.

[0248] Please note that FIG. 16A shows only one non-limiting example of a neural network. The layers between the input layer and the output layer are called hidden layers. The number of hidden layers can be one or more. For example, a deep neural network usually has many hidden layers. Furthermore, although only one output neuron is shown, for example, a multi-layer deep neural network can have a larger number of output neurons.

[0249] The number of neurons can be optimized. At the beginning of training, the network configuration is likely to have an excess of nodes. Some of the nodes that do not noticeably affect the network performance can be removed from the network during training. For example, nodes whose weights approach zero after training can be removed (this process is called pruning). The number of neurons and layers should be chosen appropriately to avoid underfitting (not being able to adequately capture the signals in the dataset) or overfitting (there is insufficient information to train all the neurons and the network performs well on the training dataset but not on the test dataset).

[0250] Various methods and criteria can be used to measure the performance of a model. For example, the root mean square error (RMSE) measures the average distance between the observed values ​​and the model predictions. The coefficient of determination (R 2) measures the correlation (not accuracy) between observed and predicted outcomes. If there is a large variance in the data, the method may be unreliable. Other performance measures include irreducible noise, model bias, and model variance. High model bias for a model indicates that the model is unable to capture the true relationship between predictors and outcomes. Model variance can indicate if the model is not stable (small perturbations in the data significantly change the model fit).

[0251] The foregoing description is merely exemplary in nature and is not intended to limit the present disclosure, its application, or use in any way. The broad teachings of the present disclosure can be implemented in various forms. Thus, although the present disclosure includes specific examples, the true scope of the present disclosure should not be limited to such examples, since other modifications will become apparent upon review of the drawings, specification, and claims below. It should be understood that one or more steps in a method may be performed in different orders (or simultaneously) without changing the principles of the present disclosure.

[0252] Additionally, while each embodiment is described above as having certain features, any one or more of these features described with respect to any embodiment of the present disclosure can be implemented in other embodiments and / or combined with any features of the other embodiments (even if such combination is not explicitly described). In other words, the described embodiments are not mutually exclusive, and it is within the scope of the present disclosure to substitute one or more embodiments for one another.

[0253] Spatial and functional relationships between elements (e.g., modules, circuit elements, semiconductor layers, etc.) are described using various terms such as "connected," "engaged," "coupled," "adjacent," "next to," "on," "above," "below," and "disposed." Also, when a relationship between a first element and a second element is described in the above disclosure, unless expressly described as "direct," the relationship may be a direct relationship where there are no other intervening elements between the first element and the second element, but may also be an indirect relationship where there are one or more intervening elements (spatial or functional) between the first element and the second element.

[0254] As used herein, the phrase at least one of A, B, and C should be interpreted in the sense of a logical (A or B or C) using a non-exclusive logical OR, and not as "at least one of A, at least one of B, and at least one of C."

[0255] In some implementations, the controller is part of a system, such as may be part of the examples described above. Such systems may include semiconductor processing equipment, including one or more processing tools, one or more chambers, one or more processing platforms, and / or specific processing components (wafer pedestals, gas flow systems, etc.). These systems may be integrated with electronics for controlling system operation before, during, and after processing of a semiconductor wafer or substrate.

[0256] Such electronics may be referred to as a "controller" and may control various components or subcomponents of one or more systems. The controller may be programmed to control any of the processes disclosed herein depending on the processing requirements and / or type of system. Such processes may include delivery of process gases, temperature settings (e.g., heating and / or cooling), pressure settings, vacuum settings, power settings, radio frequency (RF) generator settings, RF matching circuit settings, frequency settings, flow rate settings, fluid delivery settings, position and motion settings, wafer loading and unloading from the tool, and wafer loading and unloading from other transport tools and / or load locks connected or interfaced with the particular system.

[0257] Broadly, a controller may be defined as an electronic device having various integrated circuits, logic, memory, and / or software that receives instructions, issues instructions, controls operations, enables cleaning operations, enables endpoint measurements, etc. Integrated circuits may include chips in the form of firmware that store program instructions, digital signal processors (DSPs), chips defined as application specific integrated circuits (ASICs), and / or one or more microprocessors, i.e., microcontrollers, that execute program instructions (e.g., software).

[0258] Program instructions may be instructions communicated to the controller in the form of various individual settings (or program files) that define operational parameters for performing a particular process on or for a semiconductor wafer or for a system. The operational parameters may, in some embodiments, be part of a recipe defined by a process engineer for implementing one or more processing steps in the fabrication of one or more layers, materials, metals, oxides, silicon, silicon dioxide, surfaces, circuits, and / or wafer dies.

[0259] The controller, in some embodiments, may be part of a computer that is integrated or coupled with the system or otherwise networked to the system, or may be coupled to such a computer, or a combination thereof. For example, the controller may be in the "cloud" or all or part of a fab host computer system. This allows for remote access of wafer processing. The computer may allow remote access to the system to monitor the current progress of a fabrication operation, review the history of past fabrication operations, review trends or performance criteria from multiple fabrication operations, change parameters of a current process, set up processing steps following the current process, or initiate a new process.

[0260] In some examples, a remote computer (e.g., a server) can provide a process recipe to the system over a network. Such a network may include a local network or the Internet. The remote computer may include a user interface that allows entry or programming of parameters and / or settings, which are then communicated from the remote computer to the system. In some examples, the controller receives instructions in the form of data. Such data identifies parameters for each processing step that is performed during one or more operations. It should be understood that the parameters may be specific to the type of process being performed and the type of tool that the controller is configured to interface with or control.

[0261] Thus, as discussed above, the controller may be distributed, for example, by having one or more individual controllers networked together and working together toward a common purpose (such as the processes and controls described herein). An example of a distributed controller for such purposes would include one or more integrated circuits on the chamber in communication with one or more integrated circuits located remotely (e.g., at the platform level or as part of a remote computer) and coupled to control the processes in the chamber.

[0262] Exemplary systems may include, but are not limited to, a plasma etch chamber or module, a deposition chamber or module, a spin rinse chamber or module, a metal plating chamber or module, a cleaning chamber or module, a bevel edge etch chamber or module, a physical vapor deposition (PVD) chamber or module, a chemical vapor deposition (CVD) chamber or module, an atomic layer deposition (ALD) chamber or module, an atomic layer etch (ALE) chamber or module, an ion implantation chamber or module, a tracking chamber or module, and any other semiconductor processing system that may be associated with or used in the fabrication and / or manufacturing of semiconductor wafers.

[0263] As described above, depending on the process step or steps being performed by the tool, the controller may communicate with one or more other tool circuits or modules, other tool components, cluster tools, other tool interfaces, adjacent tools, nearby tools, tools located throughout the factory, a main computer, another controller, or tools used for material transport to and from tool locations and / or load ports within a semiconductor manufacturing factory. The present disclosure includes the following application examples: [Application example 1] 1. A substrate processing system, comprising: a laser sensor disposed in the substrate processing system, the laser sensor comprising: Capturing first and second data from a component employed in a process chamber of the substrate processing system at first and second times, respectively. It is configured as follows: the component includes at least one of a semiconductor substrate and an edge-coupled ring for use with the semiconductor substrate; The component is used in the processing chamber during a process performed in the processing chamber between the first time and the second time. A laser sensor; A controller, receiving the first data and the second data from the laser sensor; determining a change in a geometric parameter of the component based on the first data and the second data; transmitting the measured changes over a network to a remote server, the remote server including a trained model to recommend adjustments to at least one parameter of the process and the processing chamber based on the measured changes; receiving the adjustments to the parameters from the remote server over the network; performing said adjustment to said at least one parameter of said process and said processing chamber; With a controller configured like this: A substrate processing system comprising: [Application example 2] The substrate processing system according to Application Example 1, A substrate processing system, wherein the first data and the second data include measurements taken at a point, along a line, or across an area of ​​the component. [Application example 3] The substrate processing system according to Application Example 1, The geometric parameters include a thickness, a flatness, or a surface roughness of the component. [Application example 4] The substrate processing system according to Application Example 1, The controller is configured to determine a non-uniformity of at least one of an etch rate of the process and a deposition rate of the process based on the measured changes, and the adjustment minimizes the non-uniformity of the at least one of the etch rate of the process and the deposition rate of the process. [Application example 5] The substrate processing system according to Application Example 1, The controller is configured to first adjust the parameters of the process and then adjust the parameters of the processing chamber. [Application Example 6] The substrate processing system according to Application Example 1, The substrate processing system, wherein the controller is configured to first adjust the parameters of the processing chamber and then adjust the parameters of the process. [Application Example 7] The substrate processing system according to Application Example 1, The parameters of the process are selected from the group consisting of gas flow, pressure, upper electrode temperature, substrate support temperature distribution, bias voltage, etch / deposition rate, and etch / deposition uniformity. [Application Example 8] The substrate processing system according to Application Example 1, 11. A substrate processing system according to claim 10, wherein the parameters of the processing chamber are selected from the group consisting of a height of the edge bond ring above a substrate support assembly in the processing chamber, a slope of the edge bond ring, a gap between an upper electrode and a lower electrode of the processing chamber, an alignment between the upper electrode and the lower electrode, and a slope between the upper electrode and the lower electrode. [Application Example 9] The substrate processing system according to Application Example 1, A substrate processing system, wherein the measured changes indicate a variation in quality of the process used to clean the processing chamber, and wherein the adjustments to the parameters of the process minimize the variation, the parameters including one or more of a duration and a frequency of the process. [Example 10] The substrate processing system according to Application Example 1, The controller: determining a profile of an inner diameter of the edge bond ring based on the first data and the second data, the profile being indicative of a malfunction of an assembly of the processing chamber; transmitting said profile to said remote server; receiving instructions for maintaining the assembly from the remote server based on the profile; Generate a message for maintaining the assembly based on the instruction. The substrate processing system is configured as follows. [Application Example 11] 1. A system comprising: A processor; When executed by the processor, receiving, at first and second times, respectively, first and second data captured by a laser sensor from a component employed in a process chamber of a substrate processing system, the component including at least one of a semiconductor substrate and an edge bond ring employed with the semiconductor substrate, the component employed in the process chamber during a process performed in the process chamber between the first time and the second time; determining a change in a geometric parameter of the component based on the first data and the second data; inputting the measured changes into a model trained to recommend adjustments to at least one parameter of the process and the processing chamber based on the measured changes; receiving the adjustments to the parameters from the model; performing said adjustment to said at least one parameter of said process and said processing chamber; a memory containing instructions to configure said processor to A system comprising: [Application Example 12] The system according to Application Example 11, The system, wherein the first data and the second data include measurements taken at a point, along a line, or across an area of ​​the component. [Application Example 13] The system according to Application Example 11, The geometric parameters include a thickness, flatness, or surface roughness of the component. [Example 14] The system according to Application Example 11, The instructions further configure the processor to determine a non-uniformity of at least one of an etch rate of the process and a deposition rate of the process based on the measured changes, and the adjusting minimizes the non-uniformity of the at least one of the etch rate of the process and the deposition rate of the process. [Example 15] The system according to Application Example 11, The instructions further configure the processor to first adjust the parameters of the process and then adjust the parameters of the processing chamber. [Example 16] The system according to Application Example 11, The instructions further configure the processor to first adjust the parameter of the processing chamber and then adjust the parameter of the process. [Application Example 17] The system according to Application Example 11, The instructions further configure the processor to select the parameters of the process from the group consisting of gas flow, pressure, upper electrode temperature, substrate support temperature distribution, bias voltage, etch / deposition rate, and etch / deposition uniformity. [Example 18] The system according to Application Example 11, The instructions further configure the processor to select the parameters of the process from the group consisting of a height of the edge bond ring above a substrate support assembly in the processing chamber, a slope of the edge bond ring, a gap between an upper electrode and a lower electrode of the processing chamber, an alignment between the upper electrode and the lower electrode, and a slope between the upper electrode and the lower electrode. [Example 19] The system according to Application Example 11, The system, wherein the measured changes indicate a variation in the quality of the process used to clean the processing chamber, and the adjustments to the parameters of the process minimize the variation, the parameters including one or more of a duration and frequency of the process. [Example 20] The system according to Application Example 11, The instruction: determining a profile of an inner diameter of the edge bond ring based on the first data and the second data, the profile being indicative of a malfunction of an assembly of the processing chamber; Sending the profile to the model; receiving instructions for maintaining the assembly from the model based on the profile; Generate a message for maintaining the assembly based on the instruction. and further configuring the processor to:

Claims

1. 1. A substrate processing system, comprising: a laser sensor disposed in the substrate processing system, the laser sensor comprising: Capturing first and second data from a component employed in a processing chamber of the substrate processing system at a first and second time, respectively. It is configured as follows: the component includes at least one of a semiconductor substrate and an edge-coupled ring for use with the semiconductor substrate; The component is used in the processing chamber during a process performed in the processing chamber between the first time and the second time. A laser sensor; A controller, receiving the first data and the second data from the laser sensor; determining a change in a geometric parameter of the component based on the first data and the second data; transmitting the measured changes over a network to a remote server, the remote server including a trained model to recommend adjustments to at least one parameter of the process and the processing chamber based on the measured changes; receiving the adjustments to the parameters from the remote server over the network; Sequentially performing said adjustments to said at least one parameter of said process and said processing chamber. With a controller configured like this: A substrate processing system comprising:

2. 2. The substrate processing system of claim 1, The substrate processing system, wherein the first data and the second data include measurements taken at a point, along a line, or across an area of ​​the component.

3. 2. The substrate processing system of claim 1, The geometric parameters include a thickness, a flatness, or a surface roughness of the component.

4. 2. The substrate processing system of claim 1, The controller is configured to determine a non-uniformity of at least one of an etch rate of the process and a deposition rate of the process based on the measured changes, and the adjustment minimizes the non-uniformity of the at least one of the etch rate of the process and the deposition rate of the process.

5. 2. The substrate processing system of claim 1, The controller is configured to first adjust the parameters of the process and then adjust the parameters of the processing chamber.

6. 2. The substrate processing system of claim 1, The substrate processing system, wherein the controller is configured to first adjust the parameters of the processing chamber and then adjust the parameters of the process.

7. 2. The substrate processing system of claim 1, The parameters of the process are selected from the group consisting of gas flow, pressure, upper electrode temperature, substrate support temperature distribution, bias voltage, etch / deposition rate, and etch / deposition uniformity.

8. 2. The substrate processing system of claim 1, 11. A substrate processing system according to claim 10, wherein the parameters of the processing chamber are selected from the group consisting of a height of the edge bond ring above a substrate support assembly in the processing chamber, a slope of the edge bond ring, a gap between an upper electrode and a lower electrode of the processing chamber, an alignment between the upper electrode and the lower electrode, and a slope between the upper electrode and the lower electrode.

9. 2. The substrate processing system of claim 1, A substrate processing system, wherein the measured changes indicate a variation in quality of the process used to clean the processing chamber, and the adjustments to the parameters of the process minimize the variation, the parameters including one or more of a duration and a frequency of the process.

10. 2. The substrate processing system of claim 1, The controller: determining a profile of an inner diameter of the edge bond ring based on the first data and the second data, the profile being indicative of a malfunction of an assembly of the processing chamber; transmitting said profile to said remote server; receiving instructions for maintaining the assembly from the remote server based on the profile; Generate a message for maintaining the assembly based on the instruction. The substrate processing system is configured as follows.

11. 1. A system comprising: A processor; When executed by the processor, receiving, at first and second times, respectively, first and second data captured by a laser sensor from a component used in a process chamber of a substrate processing system, said component including at least one of a semiconductor substrate and an edge bond ring used with said semiconductor substrate, said component used in said process chamber during a process performed in said process chamber between said first and second times; determining a change in a geometric parameter of the component based on the first data and the second data; inputting the measured changes into a model trained to recommend adjustments to at least one parameter of the process and the processing chamber based on the measured changes; receiving the adjustments to the parameters from the model; Sequentially performing said adjustments to said at least one parameter of said process and said processing chamber. a memory containing instructions for configuring said processor to A system comprising:

12. 12. The system of claim 11, The system, wherein the first data and the second data include measurements taken at a point, along a line, or across an area of ​​the component.

13. 12. The system of claim 11, The geometric parameters include a thickness, flatness, or surface roughness of the component.

14. 12. The system of claim 11, The instructions further configure the processor to determine a non-uniformity of at least one of an etch rate of the process and a deposition rate of the process based on the measured changes, and the adjusting minimizes the non-uniformity of the at least one of the etch rate of the process and the deposition rate of the process.

15. 12. The system of claim 11, The instructions further configure the processor to first adjust the parameters of the process and then adjust the parameters of the processing chamber.

16. 12. The system of claim 11, The instructions further configure the processor to first adjust the parameter of the processing chamber and then adjust the parameter of the process.

17. 12. The system of claim 11, The instructions further configure the processor to select the parameters of the process from the group consisting of gas flow, pressure, upper electrode temperature, substrate support temperature distribution, bias voltage, etch / deposition rate, and etch / deposition uniformity.

18. 12. The system of claim 11, The instructions further configure the processor to select the parameters of the process from the group consisting of a height of the edge bond ring above a substrate support assembly in the processing chamber, a slope of the edge bond ring, a gap between an upper electrode and a lower electrode of the processing chamber, an alignment between the upper electrode and the lower electrode, and a slope between the upper electrode and the lower electrode.

19. 12. The system of claim 11, The system, wherein the measured changes indicate a variation in the quality of the process used to clean the processing chamber, and the adjustments to the parameters of the process minimize the variation, the parameters including one or more of a duration and frequency of the process.

20. 12. The system of claim 11, The instruction: determining a profile of an inner diameter of the edge bond ring based on the first data and the second data, the profile being indicative of a malfunction of an assembly of the processing chamber; Sending the profile to the model; receiving instructions for maintaining the assembly from the model based on the profile; Generate a message for maintaining the assembly based on the instruction. and further configuring the processor to:

Images