Multiple signal sources for hybrid metrology using physical modeling and machine learning

Abstract

Physical modeling and machine learning modeling are combined to analyze signals from multiple data sources including metrology data acquired from different tool sets or at different process steps and data related to processing equipment, such as sensor data, process parameters, advanced process control (APC) parameters, context data, etc. At least one physical model is generated and used to analyze the metrology signal from the metrology tool to extract measurements of critical and non-critical parameters of the structure on the sample. At least one machine learning model is constructed and trained to predict a parameter of interest based on the extracted measurements and additional data, including raw measurement signals, reference data and / or experimental design (DOE) data, and data from a different set of tools or the same tool as used for the physical modeling.

Description

[0001] CROSS-REFERENCE TO RELATED APPLICATIONS

[0002] This application claims priority to and the benefit of U.S. Provisional Application No. 63 / 355,053, filed on June 23, 2022, entitled “METROLOGY SOLUTIONS FOR GATE-ALL-AROUND TRANSISTORS,” U.S. Provisional Application No. 63 / 498,475, filed on April 26, 2023, entitled “MULTIPLE SOURCES OF SIGNALS FOR HYBRID METROLOGY USING PHYSICAL MODELING AND MACHINE LEARNING,” and U.S. Non-Provisional Application No. 18 / 339,973, filed on June 22, 2023, entitled “MULTIPLE SOURCES OF SIGNALS FOR HYBRID METROLOGY USING PHYSICAL MODELING AND MACHINE LEARNING,” all of which are assigned to the assignee of this application and are incorporated herein by reference in their entirety. Technical Field

[0003] The subject matter described herein relates generally to metrology, and more specifically to modeling and measuring structures using multiple data sources and a combination of physical modeling and machine learning. Background Art

[0004] The semiconductor and other similar industries often use metrology methods such as optical metrology or X-ray metrology to perform non-contact evaluation of samples during processing. For example, with optical metrology, the sample being measured is illuminated with light, for example, at a single wavelength or multiple wavelengths. After the light interacts with the sample, the resulting light is detected and analyzed to determine one or more characteristics of the sample.

[0005] The analysis typically includes modeling the structure under test. The model can be generated based on the physical properties of the structure (such as the material and nominal parameters of the structure, such as film thickness, optical properties of the material, line width and spacer width, etc.), and is therefore sometimes referred to as a physical model. One or more parameters of the model can be changed, and predicted data for each parameter change can be calculated based on the model, for example, using Rigorous Coupled Wave Analysis (RCWA) or other similar techniques. For example, in a nonlinear regression process, the measured data can be compared with the predicted data for each parameter change until a good fit is achieved between the predicted data and the measured data, at which point the fitted parameters are considered to be accurate representations of the parameters of the structure under test. However, modeling can be time-consuming and computationally intensive, and is expensive, especially for small complex features. Summary of the invention

[0006] Combine physical modeling and machine learning modeling to analyze signals from multiple data sources of hybrid metrology and ecosystems. Signals from data sources include metrology data that can be obtained from different tool groups or at different process steps and additional data related to processing equipment, such as sensor data, process parameters, advanced process control (Advanced Process Control, APC) parameters, contextual data, etc. The predictive ability of machine learning is improved by data mining and data fusion of signals from the multiple data sources. Generate at least one physical model and use it to analyze metrology signals from one or more metrology tools to extract measurement results of key parameters and non-key parameters of structures on samples. In addition, build and train at least one machine learning model to predict parameters of interest based on the extracted measurement results and additional data. The input data for the machine learning model includes, for example, additional data, such as raw measurement signals, reference data and / or experimental design (design of experiment, DOE) data used by one or more physical models, and data from different tool groups or the same tool used for physical modeling, such as process parameters, advanced process control (APC) parameters, contextual data, and sensor data from production equipment.

[0007] In one specific implementation, a method for characterizing a structure on a sample includes: obtaining a measurement signal for the structure on the sample from a first metrology device; and extracting a measurement result for the structure on the sample from a first physical model based on the measurement signal. The method also includes using a machine learning model to determine a parameter of interest for the structure on the sample based on the measurement result extracted from the first physical model. The machine learning model may further determine the parameter of interest based on at least one of the following: data from a measurement signal that is not used to extract the measurement result from the first physical model, the measurement signal from the first metrology device; a second measurement signal for the structure on the sample obtained from a second metrology device; a process parameter used to generate the structure on the sample; an advanced process control (APC) parameter used to generate the structure on the sample; context data for the structure on the sample; and sensor data from production equipment used to generate the structure on the sample.

[0008] In one specific implementation, a computer system configured to characterize a structure on a sample includes at least one processor, wherein the at least one processor is configured to obtain a measurement signal for the structure on the sample from a first metrology device, and extract a measurement result for the structure on the sample from a first physical model based on the measurement signal. The at least one processor is further configured to determine a parameter of interest for the structure on the sample based on the measurement result extracted from the first physical model using a machine learning model. The machine learning model may further determine the parameter of interest based on at least one of: data from a measurement signal that is not used to extract the measurement result from the first physical model, the measurement signal from the first metrology device; a second measurement signal for the structure on the sample obtained from a second metrology device; a process parameter for generating the structure on the sample; an advanced process control (APC) parameter for generating the structure on the sample; context data for the structure on the sample; and sensor data from production equipment used to generate the structure on the sample.

[0009] In one specific implementation, a system configured for characterizing a structure on a sample includes: a component for obtaining a measurement signal for the structure on the sample from a first metrology device; and a component for extracting a measurement result for the structure on the sample from a first physical model based on the measurement signal. The system also includes a component for determining a parameter of interest for the structure on the sample based on the measurement result extracted from the first physical model using a machine learning model. The machine learning model may further determine the parameter of interest based on at least one of the following: data from a measurement signal that is not used to extract the measurement result from the first physical model, the measurement signal from the first metrology device; a second measurement signal for the structure on the sample obtained from a second metrology device; a process parameter for generating the structure on the sample; an advanced process control (APC) parameter for generating the structure on the sample; context data for the structure on the sample; and sensor data from production equipment used to generate the structure on the sample.

[0010] In one specific implementation, a method for characterizing a structure on a sample includes: obtaining a pre-processing step measurement signal for the structure on the sample from a metrology device at a pre-processing step; and obtaining a post-processing step measurement signal for the structure on the sample from the metrology device at a post-processing step. The method also includes: extracting a post-processing measurement result for the structure on the sample from a post-processing physics model based on the post-processing step measurement signal; and generating pre-processing step data based on at least the pre-processing step measurement signal. The method also includes determining a parameter of interest of the structure on the sample based on the post-processing measurement result extracted from the post-processing physics model and the pre-processing step data using a machine learning model.

[0011] In one specific implementation, a computer system configured to characterize a structure on a sample includes at least one processor, wherein the at least one processor is configured to obtain a pre-processing step measurement signal for the structure on the sample from a metrology device at a pre-processing step, and obtain a post-processing step measurement signal for the structure on the sample from the metrology device at a post-processing step. The at least one processor is further configured to extract a post-processing measurement result for the structure on the sample from a post-processing physical model based on the post-processing step measurement signal, and generate pre-processing step data based on at least the pre-processing step measurement signal. The at least one processor is further configured to determine a parameter of interest of the structure on the sample based on the post-processing measurement result extracted from the post-processing physical model and the pre-processing step data using a machine learning model.

[0012] In one specific implementation, a system configured for characterizing a structure on a sample includes: a component for obtaining a pre-processing step measurement signal for the structure on the sample from a metrology device at a pre-processing step; and a component for obtaining a post-processing step measurement signal for the structure on the sample from the metrology device at a post-processing step. The system also includes: a component for extracting a post-processing measurement result for the structure on the sample from a post-processing physical model based on the post-processing step measurement signal; and a component for generating pre-processing step data based on at least the pre-processing step measurement signal. The system also includes a component for determining a parameter of interest of the structure on the sample based on the post-processing measurement result extracted from the post-processing physical model and the pre-processing step data using a machine learning model.

[0013] In one specific implementation, a method for characterizing a structure on a sample includes: obtaining a measurement signal for one or more reference samples of the structure from a first metrology device; and generating a first physical model to extract measurement results for the structure on the sample, wherein the first physical model is generated based on the measurement signal for the one or more reference samples from the first metrology device. The method also includes generating a machine learning model to predict a parameter of interest of the structure on the sample. The machine learning model is generated based on the measurement results extracted by the first physical model, and at least one of reference data and experimental design information. The machine learning model may be further generated based on at least one of the following: data from a measurement signal that is not used to generate the first physical model, the measurement signal being from the first metrology device; a second measurement signal for the one or more reference samples obtained from a second metrology device; a process parameter for generating the one or more reference samples; an advanced process control (APC) parameter for generating the one or more reference samples; context data for the one or more reference samples; and sensor data from production equipment used to generate the one or more reference samples.

[0014] In one specific implementation, a computer system configured to characterize a structure on a sample includes at least one processor, wherein the at least one processor is configured to obtain a measurement signal for one or more reference samples of the structure from a first metrology device, and generate a first physical model to extract measurement results for the structure on the sample, wherein the first physical model is generated based on the measurement signal for the one or more reference samples from the first metrology device. The at least one processor is further configured to generate a machine learning model to predict a parameter of interest of the structure on the sample. The machine learning model is generated based on the measurement results extracted by the first physical model, and at least one of reference data and experimental design information. The machine learning model may be further generated based on at least one of the following: data from a measurement signal not used to generate the first physical model, the measurement signal from the first metrology device; a second measurement signal for the one or more reference samples obtained from a second metrology device; a process parameter used to generate the one or more reference samples; an advanced process control (APC) parameter used to generate the one or more reference samples; context data for the one or more reference samples; and sensor data from production equipment used to generate the one or more reference samples.

[0015] In one specific implementation, a system configured for characterizing a structure on a sample includes: a component for obtaining a measurement signal for one or more reference samples of the structure from a first metrology device; and a component for generating a first physical model to extract measurement results for the structure on the sample, wherein the first physical model is generated based on the measurement signal for the one or more reference samples from the first metrology device. The system also includes a component for generating a machine learning model to predict a parameter of interest of the structure on the sample. The machine learning model is generated based on the measurement results extracted by the first physical model, and at least one of reference data and experimental design information. The machine learning model may be further generated based on at least one of the following: data from a measurement signal that is not used to generate the first physical model, the measurement signal being from the first metrology device; a second measurement signal for the one or more reference samples obtained from a second metrology device; a process parameter for generating the one or more reference samples; an advanced process control (APC) parameter for generating the one or more reference samples; context data for the one or more reference samples; and sensor data from production equipment used to generate the one or more reference samples.

[0016] In one specific implementation, a method for characterizing a structure on a sample includes: obtaining a pre-processing step measurement signal for one or more reference samples of the structure from a metrology device at a pre-processing step; and obtaining a post-processing step measurement signal for the one or more reference samples of the structure from the metrology device at a post-processing step. The method also includes generating a post-processing physics model to extract post-processing measurement results for the structure on the reference sample, wherein the post-processing physics model is generated based on the post-processing step measurement signal; and generating pre-processing step data based at least on the pre-processing step measurement signal. The method also includes generating a machine learning model to predict a parameter of interest of the structure on the sample. The machine learning model is generated based on the post-processing measurement results extracted by the post-processing physics model, and at least one of reference data and experimental design information and the pre-processing step data.

[0017] In one specific implementation, a computer system configured to characterize a structure on a sample includes at least one processor, wherein the at least one processor is configured to obtain a pre-processing step measurement signal for one or more reference samples of the structure from a metrology device at a pre-processing step; and obtain a post-processing step measurement signal for the one or more reference samples of the structure from the metrology device at a post-processing step. The at least one processor is further configured to generate a post-processing physics model to extract post-processing measurement results for the structure on the reference sample, wherein the post-processing physics model is generated based on the post-processing step measurement signal; and generate pre-processing step data based at least on the pre-processing step measurement signal. The at least one processor is further configured to generate a machine learning model to predict a parameter of interest of the structure on the sample. The machine learning model is generated based on the post-processing measurement results extracted by the post-processing physics model, and at least one of reference data and experimental design information and the pre-processing step data.

[0018] In one specific implementation, a system configured for characterizing a structure on a sample includes: a component for obtaining a pre-processing step measurement signal for one or more reference samples of the structure from a metrology device at a pre-processing step; and a component for obtaining a post-processing step measurement signal for the one or more reference samples of the structure from the metrology device at a post-processing step. The system also includes: a component for generating a post-processing physics model to extract post-processing measurement results for the structure on the reference sample, wherein the post-processing physics model is generated based on the post-processing step measurement signal; and a component for generating pre-processing step data based at least on the pre-processing step measurement signal. The system also includes a component for generating a machine learning model to predict a parameter of interest of the structure on the sample. The machine learning model is generated based on the post-processing measurement results extracted by the post-processing physics model, and at least one of reference data and experimental design information and the pre-processing step data. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 Illustrated is a schematic diagram of a metrology apparatus that may be used to characterize samples as discussed herein.

[0020] Figure 2 A workflow for offline recipe creation according to a first example scenario is illustrated, wherein signals are collected from multiple data sources including different tools and / or sources.

[0021] Figure 3 A workflow for online measurement according to a first example scenario is illustrated, wherein signals are collected from multiple data sources including different tools and / or sources.

[0022] Figure 4 A workflow for offline recipe creation according to a second example scenario is illustrated, where signals are collected from multiple data sources including different manufacturing process steps.

[0023] Figure 5 A workflow for in-line measurement according to a second example scenario is illustrated, where signals are collected from multiple data sources including different manufacturing process steps.

[0024] Figures 6 to 9 A flow chart depicting a method for characterizing structures on a sample is illustrated. DETAILED DESCRIPTION

[0025] During the manufacture of semiconductor devices and similar equipment, it is often necessary to monitor the manufacturing process by non-destructively measuring these devices. One type of metrology that can be used to non-destructively measure samples during processing is optical metrology, which can use a single wavelength or multiple wavelengths and can include, for example, ellipsometry, reflectometry, Fourier transform infrared spectroscopy (FTIR), etc. Other types of metrology can also be used, including X-ray metrology, photoacoustic metrology, electron beam (E-beam) metrology, etc.

[0026] Optical metrology (such as thin film metrology and optical critical dimension (OCD) metrology) and other types of metrology sometimes use physical modeling techniques to generate predicted data for a sample, which is compared with the measured data from the sample. Using physical modeling techniques, a sample model including key parameters and non-key parameters is generated. The model can be based on the nominal parameters of the sample and can include one or more variable parameters, such as layer thickness, line width, space width, sidewall angle, material properties, etc., which can vary within a desired range, for example, depending on the process parameters used to manufacture the sample being measured. The model can also include parameters related to the tool set, such as the characteristics of the optical system used by the metrology equipment. The predicted data can be calculated based on the parameters of the physical model (including changes in the variable parameters) and the characteristics of the metrology equipment using analytical or semi-analytical methods (such as effective medium theory (EMT), finite difference time domain (FDTD), transfer matrix method (TMM), Fourier modal method (FMM) / rigid coupled wave analysis (RCWA), finite element method (FEM), etc.). For example, in the nonlinear regression process, the measured data obtained from the sample by the metrological equipment are compared with the predicted data for different parameter changes until the best fit is achieved, at which point the values ​​of the fitted parameters are considered to be an accurate representation of the sample parameters.

[0027] Typically, modeling requires that preliminary structural and material information about the sample is known in order to generate an accurate representative model of the sample, which includes one or more variable parameters. For example, the preliminary structural information and preliminary material information of the sample can include a physical description of the structure type and the sample, which has nominal values ​​of various parameters, such as layer thickness, line width, interval width, sidewall angle, material properties, etc., and the range in which these parameters can vary. The model may further include one or more parameters that are immutable, i.e., parameters that are not expected to change significantly in the sample during manufacturing. The variable parameters of the model are adjusted, and prediction data can be generated in real time during the nonlinear regression process, or a library can be generated in advance. Therefore, modeling applies physical constraints in the analysis, and therefore provides a high level of fidelity for the measurement results. However, due to the physical calculations necessary to generate prediction data, modeling has a high computational cost. For example, modeling a complex 3D structure suffers from a slow solution time (TTS), and due to the difficulty in fitting the data for the complex structure, the modeling accuracy may be reduced.

[0028] Another technique that can be used to generate prediction data for a sample based on measurement data obtained from the sample by a metrology device is machine learning. Machine learning algorithms that can be used for metrology may include, for example, but are not limited to, linear regression, neural networks, deep learning, convolutional neural networks (CNNs), ensemble methods, support vector machines (SVMs), random forests, etc., or a combination of multiple models in sequential and / or parallel modes. Machine learning does not require a physical model of the sample. Instead, reference data (e.g., measurement data obtained by a metrology device from one or more reference samples) and values ​​of structural parameters of interest are obtained, and the reference data is used to generate and train a machine learning model. The machine learning model is automatically trained using reference data and known values ​​of structural parameters to find relevant data features and learn the inherent relationship and connection between input features and output features in order to make decisions and predictions for new data. The benefits of using machine learning are fast solution time (TTS) and minimal demand for computing resources. However, machine learning requires a large amount of reference data, which is expensive and time-consuming to obtain. In the absence of a large amount of reference data, the machine learning model may suffer from overfitting due to the lack of physical constraints.

[0029] As semiconductor devices continue to shrink, metrology budgets become tighter. In addition, complex 3D structures are more frequently adopted to achieve continuous device scaling. Advances in semiconductor technology, such as the use of complex 3D structures, present additional challenges to metrology due to increased modeling complexity and parameter dependencies as well as reduced sensitivity. For example, the signal from a single metrology tool or source may not have sufficient sensitivity to accurately measure the parameters of interest for semiconductor process quality control. Ultimately, there may not be a single metrology tool that can handle all metrology requirements for most advanced semiconductor devices.

[0030] As discussed herein, by using data collected from multiple data sources (e.g., from multiple tool groups and / or process steps) and using additional data related to samples collected from metrology and / or production equipment (e.g., sensor data), computationally efficient data analysis methods can fuse the multiple data sources and produce more accurate and consistent measurements than can be provided by any individual data source. The analysis methods can be flexible to accommodate data of a variety of different natures, while making the most of existing well-developed techniques, such as physical modeling or machine learning, for each type of data source, and synergizing the strengths of the individual metrology techniques.

[0031] As discussed herein, physical modeling and machine learning are combined to analyze multiple data sources for hybrid metrology and ecosystems. The methods described herein generate predictive capabilities through data mining and data fusion from multiple data sources (e.g., multiple metrology tool groups, sample data from multiple process steps, metrology equipment parameters, and production equipment parameters). For example, at least one physical model can be used to analyze metrology signals from one or more metrology tools (such as spectroscopic ellipsometers, spectroscopic reflectometers, X-ray metrology, photoacoustic metrology, Fourier transform infrared spectroscopy (FTIR), electron beam metrology, etc.) to extract measurement results of key parameters and non-key parameters of samples. In addition, at least one machine learning model can be constructed and trained to predict parameters of interest. The machine learning model can use input data from one or more of the following: measurement results (key parameters and non-key parameters) from one or more physical models; raw signals and optionally mismatches for the one or more physical models; data sources from different tool groups or the same tool but not included in physical modeling; process parameters, advanced process control (APC) parameters, contextual data; and sensor data from production equipment. Online measurement of the sample uses one or more physical models and a trained machine learning model to predict parameters of interest of the sample based on data acquired from multiple data sources.

[0032] The proposed technique can be used to combine and analyze multiple data sources in an efficient and flexible way, thereby synergizing physical modeling and machine learning with controllable computational cost and software and modeling complexity, thus providing the most feasible solution with manageable time to solution (TTS) and improved final results and overall metrology performance. The method is also general and can be applied to the measurement of any device, OCD, thin film or other type of target.

[0033] For example, Figure 1 A schematic diagram of a metrology apparatus 100 that can be used to characterize structures on a sample is illustrated, as described herein. The metrology apparatus 100 can be configured to perform one or more types of measurements of the sample 103, such as, for example, spectroscopic reflectance measurements, spectroscopic ellipsometry measurements (including Miller matrix ellipsometry measurements), spectroscopic scatterometry measurements, overlap scatterometry measurements, interferometry measurements, photoacoustic metrology, electron beam metrology, X-ray metrology, FTIR measurements, etc. The metrology apparatus 100 can include, for example, a first metrology tool 101 and a second metrology tool 170, but can include additional metrology tools, or can be coupled to receive sample data measured by separate metrology tools. It should be understood that the metrology apparatus 100 is illustrated as one example configuration of a metrology apparatus, and other configurations and other metrology apparatuses can be used if desired.

[0034] The metrology apparatus 100 includes an oblique incidence metrology tool 101, which includes a light source 110 that generates light 102. For example, the light 102 may be UV visible light having a wavelength, for example, between 200 nm and 1000 nm. The light 102 generated by the light source 110 may include a range of wavelengths, i.e., a continuous range or a plurality of discrete wavelengths, or may be a single wavelength. The metrology apparatus 100 includes focusing optics 120 and 130 that focus and receive the light and direct the light so as to be obliquely incident on the top surface of the sample 103. The optics 120, 130 may be refractive, reflective, or a combination thereof, and may be an objective lens.

[0035] The reflected light may be focused by lens 114 and received by detector 150. Detector 150 may be a conventional charge coupled device (CCD), photodiode array, CMOS, or similar type of detector. If broadband light is used, detector 150 may be, for example, a spectrometer, and detector 150 may, for example, generate a spectral signal as a function of wavelength. A spectrometer may be used to disperse the full spectrum of received light into spectral components across the detector pixel array. One or more polarization elements may be located in the beam path of metrology device 100. For example, metrology device 100 may include one or both (or none) of one or more polarization elements 104 in the beam path before sample 103 and a polarization element (analyzer) 112 in the beam path after sample 103, and may include one or more additional elements 105a and 105b, such as compensators or photoelastic modulators, which may be before, after, or before and after sample 103. By using a spectroscopic ellipsometer using a dual rotating compensator between polarization elements 104 and 112 and the sample, a full Miller matrix may be measured.

[0036] The metrology apparatus 100 may include or may be coupled to additional metrology apparatus. For example, as illustrated, the metrology apparatus 100 may include a second normal incidence metrology tool 170. For example, the second metrology tool 170 may be configured for use in spectral reflectance measurements, spectral scatterometry measurements, overlap scatterometry measurements, interferometry measurements, electron beam metrology, X-ray metrology, FTIR measurements, etc. In some implementations, the metrology apparatus 100 may include additional tools, such as a third (or more) metrology tool. In some implementations, the additional metrology tools may be separate from the metrology apparatus 100.

[0037] The metrology apparatus 100 also includes at least one computing system 160 configured to characterize one or more parameters of the sample 103 using the methods described herein. The at least one computing system 160 is coupled to the first metrology tool 101 (e.g., the detector 150) and the second metrology tool 170 and any additional metrology tools (if present) to receive metrology data acquired during measurement of the structure of the sample 103. The acquisition of data may be performed during pre-processing manufacturing steps as well as during post-processing manufacturing steps. The at least one computing system 160 may be, for example, a workstation, a personal computer, a central processing unit, or other suitable computer system, or multiple systems.

[0038] It should be understood that at least one computing system 160 may be a single computer system or multiple separate or linked computer systems, which may be interchangeably referred to herein as computing system 160 or at least one computing system 160. Computing system 160 may be included in or connected to or otherwise associated with metrology device 100 and any additional metrology tools. Different subsystems of metrology device 100 may each include a computing system configured to perform steps associated with the associated subsystem. For example, computing system 160 may control the positioning of sample 103, for example, by controlling the movement of stage 109 coupled to the chuck. Stage 109 may, for example, be capable of horizontal movement in Cartesian (i.e., X and Y) coordinates or polar (i.e., R and θ) coordinates or some combination of the two. The stage may also be capable of vertical movement along the Z coordinate. Computing system 160 may further control the operation of chuck 108 to hold or release sample 103. The computing system 160 may further control or monitor the rotation of one or more polarization elements 104, 112 or additional elements 105a, 105b, etc.

[0039] The computing system 160 may be communicatively coupled to the detector 150 in the first metrology tool 101 and the detector in the second metrology tool 170 (if present) in any manner known in the art. For example, at least one computing system 160 may be coupled to a separate computing system associated with the detector 150. The computing system 160 may be configured to receive and / or obtain metrology data from the detector 150 and the controller polarization elements 104, 112 and the additional elements 105a, 105b, etc. and the components of the second metrology tool 170 via a transmission medium that may include wired and / or wireless portions, for example. Thus, the transmission medium may serve as a data link between the computing system 160 and other subsystems of the metrology device 100. The computing system 160 may be further configured to receive and / or obtain additional information about the sample and one or more subsystems of the first metrology tool 101 and the production equipment, for example, from a user interface (UI) 168 or via a transmission medium that may include wired and / or wireless portions.

[0040] The computing system 160 includes at least one processor 162 with a memory 164, and a UI 168 communicatively coupled via a bus 161. The memory 164 or other non-transitory computer-usable storage medium includes computer-readable program code 166 embodied therein and can be used by the computing system 160 to cause at least one computing system 160 to control the metering device 100 and perform functions including the techniques and analyses described herein. For example, as illustrated, the memory 164 may include instructions for causing the processor 162 to perform both modeling and machine learning, and in some implementations, feedforward and / or feedback may be employed, as discussed herein. In view of this disclosure, one of ordinary skill in the art may implement the data structures and software codes described in this detailed description for automatically implementing one or more actions, and these data structures and software codes are stored on a computer-usable storage medium such as the memory 164, which may be any device or medium that can store code and / or data for use by a computer system such as the computing system 160. Computer usable storage media may be, but are not limited to, read-only memory, random access memory, magnetic and optical storage devices such as disk drives, magnetic tape, etc. In addition, the functions described herein may be embodied in whole or in part within the circuitry of an application specific integrated circuit (ASIC) or a programmable logic device (PLD), and these functions may be embodied in a computer-understandable descriptor language that may be used to create an ASIC or PLD that operates as described herein.

[0041] The computing system 160, for example, may be configured to obtain data of a reference sample and data related to the sample (such as reference data and / or DOE data) and data related to the metrology tool and / or processing equipment (such as process parameters, advanced parameter control (APC) parameters, contextual data, and sensor data from production equipment) from multiple data sources (including from one or both of the metrology tools 101 and 170) and any desired additional metrology tools. The computing system 160 may be configured to generate one or more physical models (models 164pm) of the sample based on measurement data from one or more reference samples and optionally additional information related to the sample and / or processing equipment, and to generate and train one or more machine learning models (ML 164ml) of the sample based on measurement results extracted from the one or more physical models and data, as discussed herein. In some embodiments, different computing systems and / or different metrology devices may be used to acquire metrology data and additional information from training samples and generate one or more physical models (model 164pm) and / or generate and train one or more machine learning models (ML 164ml), and the resulting physical models and / or trained machine learning models (or portions thereof) may be provided to the computing system 160, for example, via computer readable program code 166 on a non-transitory computer-usable storage medium (such as memory 164).

[0042] The computing system 160 may additionally or alternatively be used to obtain data from a test sample from a plurality of data sources. The data may be of the same type used to generate the physical model and to generate and train the machine learning model described above, and the test sample has the same structure as the reference sample. The computing system 160 may be configured to use data from a plurality of sources and one or more physical models (models 164pm) and one or more trained machine learning models (ML 164ml) to determine one or more parameters of interest for the sample, as discussed herein.

[0043] The results from the analysis of the data may be reported, for example, stored in a memory 164 associated with the sample 103 and / or indicated to a user via a UI 168, an alarm, or other output device. In addition, the results from the analysis may be reported and fed forward or fed back to the processing equipment to adjust the appropriate manufacturing steps to compensate for any detected differences in the manufacturing process. For example, the computing system 160 may include a communication port 169, which may be any type of communication connection, such as a communication connection to the Internet or any other computer network. The communication port 169 may be used to receive instructions for programming the computing system 160 to perform any one or more of the functions described herein and / or to output, for example, a signal with measurement results and / or instructions to another system, such as an external process tool, in a feed forward or feedback process to adjust a process parameter associated with a manufacturing process step of the sample based on the measurement results.

[0044] As discussed herein, to characterize a sample, (1) at least one physics-based model is constructed to analyze metrology signals from one or more tools (such as spectroscopic ellipsometry (SE), spectroscopic reflectometry (SR), X-ray, electron beam, photoacoustic data, Fourier transform infrared spectroscopy (FTIR), etc.) and from one or more sources to extract measurements of key parameters and non-key parameters. Additionally, (2) at least one machine learning model is constructed and trained to predict parameters of interest. The machine learning model may take as input one or more of the following data: a) measurements (key parameters and non-key parameters) from the physical model of 1); b) raw signals and optionally mismatches from the physical model of (1); data sources from different tool groups or the same tool in (1) but not included in the physical modeling; process parameters, APC parameters, contextual data; and sensor data from production equipment. Additionally, (3) the created and trained physical model and machine learning model may be used to perform in-line measurements of the sample to predict parameters of interest based on data from multiple data sources.

[0045] For example, Figure 2 A workflow 200 for offline recipe creation (e.g., generating one or more physical models and one or more machine learning models) according to a first example scenario is illustrated, wherein data is collected from multiple data sources (e.g., different tools and / or sources). Figure 2 In FIG. 2 , solid black arrows indicate processes used in workflow 200 , dashed black arrows indicate processes that are optional but at least one exists, and dashed grey arrows indicate optional processes.

[0046] As illustrated, measurement signals 202 from one or more reference samples are collected from a first data source or tool (Source 1). The measurement signals may be collected from any desired metrology equipment such as Figure 1) or collects measurement signals 202 from any other desired type of metrology equipment.

[0047] Additionally, data is acquired from one or more additional data sources. For example, in some implementations, measurement signals 204 and 206 from one or more reference samples may be collected from one or more additional sources or tools, such as illustrated as a second source or tool (Source 2) and a third source or tool (Source 3). For example, measurement signals 204 and 206 from one or more reference samples may be collected from one or more additional sources or tools, such as illustrated as a second source or tool (Source 2) and a third source or tool (Source 3). Figure 1Additional measured signals 204 may be collected from metrology tools 170 shown in Figure 101 or from any other desired type of metrology equipment, and measurement signals 206 may be collected from metrology equipment different from source 1 and source 2 (such as different types of measurements from either of metrology tools 101 or 170 or from any other desired type of metrology equipment). Additional data 208 related to the reference samples may be collected and used as training data for one or more machine learning models 222, as illustrated by the box arrows. The additional data 208 may include, for example, reference data and DOE data for the samples. For example, the reference data may be measurement signals acquired from one or more reference samples by a metrology equipment and values ​​of structural parameters of interest, typically provided by a CD-AFM (atomic force microscope), a CD-SEM (scanning electron microscope), or a TEM (transmission electron microscope). For example, the DOE data may be measurement data from a set of reference samples treated with intentionally introduced skew conditions such that the structural parameters of interest are varied by the skew processing conditions having a known pattern. Reference data and / or DOE data can be used as training data sets to train machine learning models to find relevant data features and learn the inherent relationships and connections between input features and output features in order to make decisions and predictions on new data. In some specific implementations, the additional data 208 related to the reference sample may further include wafer conditions, accuracy, tool matching data, etc. For example, accuracy data is data from the same target repeatedly measured from the same tool instance at multiple times. Accuracy metrics are another metrology key performance indicators (KPIs) that indicate the consistency of measurement results from multiple runs of the same sample. Tool matching data is, for example, measurement data of the same target from multiple instances of tools of the same tool type. Tool matching metrics are another metric KPI that indicates the consistency of measurement results from different tools of the same type from the same sample. Measurement accuracy (assessed by matching with reference values ​​provided from CD-AFM, CD-SEM, TEM, etc. and / or consistency with DOE conditions), accuracy, and tool matching are typical metrology KPIs. If accuracy and tool matching data are provided, physical modeling or machine learning models can be optimized to not only closely match reference values, but also predict consistent results for the same sample using measurement signals from multiple runs of the same tool or from different tools of the same type.

[0048] In addition, in some specific implementations, the additional data signal 209 can be used as an input to the physical model or an input feature to the machine learning model. For example, additional data signals 209 that can be associated with sources (e.g., source 1, source 2, and source 3) such as process parameters, advanced process control (APC) parameters, contextual data, and sensor data from production equipment can be obtained. By way of example, some process control parameters (e.g., substrate temperature and chemical concentration for wet etching) can affect the etching rate (the speed at which material is removed from the surface of the wafer), and the etching rate is one of the important factors in determining the etching depth and CD profile. Some of these parameters (such as temperature) are measured by sensors from production equipment. Other parameters (such as etching time, etching chamber name) are user-controlled parameters. The name of the etching chamber is an example of contextual data. Since each etching chamber has its own characteristic distribution of etching profiles across the wafer, knowing this information can help machine learning predict the correct wafer map. An example of an APC parameter is an atomic force microscope (AFM) result that is measured from the same sample in different processing steps, which contains relevant information, such as non-critical parameters of the structure of interest. Adding non-critical parameters as machine learning input features can help improve the robustness of machine learning in predicting critical parameters. Adding all of these relevant parameters as machine learning input features can provide additional information that helps determine the structural parameters of interest that are controlled by these process parameters and conditions.

[0049] Measurement signals and data from multiple data sources can be used to generate one or more physical models. For example, as illustrated by the black solid arrow, the measurement signal 202 from the first source (source 1) can be used to generate a first physical model 212 of the sample. For example, a physical model of the sample is created based on the known geometry, nominal values, and materials of the structure. The measurement signal 202 can be used to generate the first physical model 212 by providing data from which the measurement results are extracted, and the first physical model 212 can be adjusted and optimized so that the calculated signal is well fitted to the measured signal and a good match between the extracted measurement results and the known parameters of the reference sample is achieved. In some specific implementations, additional data can be used to help generate the first physical model 212. For example, as illustrated by the gray dashed arrow, additional data 208 (such as reference data and / or DOE) and optionally wafer condition, precision, and tool matching data can also be used to help generate the first physical model 212. In addition, as illustrated by the gray dashed arrow, the data signal 209 can be used to help generate the first physical model 212. In another example, as illustrated by the gray dashed arrow, the measurement signal 204 from the second source (Source 2) can be used to help generate the first physical model 212 of the sample. In some implementations, both the additional data 208 and the measurement signal 204 can be used to help generate the first physical model 212.

[0050] In some implementations, multiple physical models can be generated. For example, as illustrated by the gray dashed arrow and the gray dashed box, a second physical model 214 can be generated based on the measurement signal 204 from the second source (source 2). In some implementations, additional data can be used to generate the second physical model 214. For example, as illustrated by the gray dashed arrow, additional data 208 (such as reference data and / or DOE) and optionally wafer condition, precision and tool matching data can also be used to help generate the second physical model 214. In another example, as illustrated by the gray dashed arrow, the measurement signal 206 from the third source (source 3) can be used to help generate the second physical model 214 of the sample. In some implementations, both the additional data 208 and the measurement signal 206 can be used to help generate the second physical model 214. In addition, as illustrated by the gray dashed arrow, the data signal 209 can be used to help generate the second physical model 214. In addition, multiple physical models can be optimized independently or together. For example, in some implementations, as illustrated by the gray dashed line, the first physical model 212 and the second physical model 214 can be linked so that at least some parameters can be coupled across the physical models 212 and 214, and the combined parameter space can be searched to fit the measurement signals from one or more data sources. The first physical model 212 and optionally the second physical model 214 can be configured to provide a goodness of fit 223 of the physical modeling.

[0051] One or more machine learning models 222 are built and trained using multiple data sources to predict parameters of interest 225. Machine learning measurement indicators 227 may be developed and reported along with goodness of fit 223 from physical modeling to indicate the quality of measurements from a formulation of physical modeling and machine learning collaboration. As illustrated by the solid black arrows, the machine learning model 222 is built using measurements extracted by the first physical model 212 as input features. As indicated by the dashed black arrows, the input features of the machine learning model 222 may additionally include at least one of measurement signals 204 from one or more reference samples collected from a second source (source 2), measurement signals 206 from one or more reference samples collected from a third source (source 3), additional data signals 209, measurements extracted by the second physical model 214, or any combination thereof. In some implementations, as illustrated by the dashed grey arrows, the input features of the machine learning model 222 may optionally include measurement signals 202 from one or more reference samples collected from a first source (source 1). In some implementations, the input features from the measurement signal 202 may include data from the measurement signal that is not used to generate the first physical model 212, such as at least one data channel or at least one data block. For example, in general, a data channel may be a measurement subsystem defined by at least one of an energy source such as a light source, an optical path guided by an optical component, a detector, or any combination thereof, and a data block may be a subset of wavelength (e.g., as used in spectroscopic metrology), frequency (e.g., as used in frequency-resolved metrology), angle (e.g., as used in angle-resolved metrology), time span (e.g., as used in time-resolved metrology), or any combination of the above from the complete data set provided by the data channel. For example, the first metrology device may collect a normal incidence signal and an oblique incidence spectroscopic ellipsometer (SE) signal. The SE signal may be used to generate the first physical model 212, but the normal incidence signal may not be used because it may be difficult to fit the normal incidence signal. Therefore, in addition to the physical modeling results generated from different data channels (e.g., SE signals), the normal incidence signal may be a data channel of data used as an input feature of the machine learning model 222. In another example, the same data channel may be divided into multiple data blocks (e.g., signals from different wavelength ranges), and some data blocks may be difficult to fit using physical modeling but may be used as data for input features of the machine learning model 222.

[0052] The machine learning model 222 is trained with at least a portion of the data 208, such as reference data and / or DOE, and optionally with wafer condition, precision, and tool matching data. The data 208 is training data and is used for offline training. For example, the reference data can be a set of signals (e.g., including any of the measurement results from the first physical model 212, the measurement signal 204, the measurement signal 206, the additional data signal 209, and the measurement signal 202) with labels (e.g., values ​​of key parameters provided by other metrology systems such as CD-SEM, TEM, CD-AFM). During the training of the machine learning model 222, the set of signals from the reference data is used as machine learning input features, and based on these input features, the machine learning model 222 predicts the key parameters. The machine learning model 222 is trained to learn and make predictions of key parameters that match the labels of the reference data. The DOE from data 208 is a set of signals measured from a reference sample processed with intentionally introduced skew conditions (e.g., including any one of the measurement results from the first physical model 212, the measurement signal 204, the measurement signal 206, the additional data signal 209, and the measurement signal 202). During machine learning training, the machine learning model 222 takes the signals from the DOE data as input features and predicts the key parameters. The machine learning model 222 is trained so that the predicted key parameter values ​​follow the expected skew pattern based on the process skew conditions. The precision data from data 208 is the measurement signal from the same sample but on multiple runs from the same metrology tool (e.g., including any one of the measurement results from the first physical model 212, the measurement signal 204, the measurement signal 206, the additional data signal 209, and the measurement signal 202). Similarly, tool matching data from data 208 is a signal from the same sample but measured from different instances of the same type of metrology tool (e.g., including any of the measurement results from the first physical model 212, the measurement signal 204, the measurement signal 206, the additional data signal 209, and the measurement signal 202). The machine learning model 222 takes the accuracy and tool matching data as input features and makes predictions. The machine learning model 222 is trained so that the predicted values ​​of the key parameters are consistent for signals measured from the same sample but from different runs or different tools. The machine learning model 222 can be trained so that if all of this data is provided during training, all criteria, matching to reference values, DOE skew conditions, high accuracy, and consistent tool matching are met simultaneously.

[0053] For example, Figure 3 A workflow 300 for online measurement according to a first example scenario is illustrated, for example, for characterizing a sample based on one or more physical models and one or more machine learning models, where signals are collected from multiple data sources (e.g., different tools and / or sources). Figure 2 One or more physical models and one or more machine learning models are generated as discussed above. Figure 3 In FIG. 3 , solid black arrows indicate processes used in workflow 300 , dashed black arrows indicate processes that are optional but at least one exists, and dashed grey arrows indicate optional processes.

[0054] As illustrated, a measurement signal 302 from a sample is collected from a first data source or tool (Source 1). The measurement signal 302 may be collected from any desired metrology equipment, such as Figure 1 1) or from any other desired type of metrology equipment and can be collected from the same Figure 2 The same metering device or the same type of metering device in source 1 is collected.

[0055] Additionally, data is obtained from one or more additional data sources. For example, in some implementations, measurement signals 304 and 306 can be collected from one or more additional sources or tools, which are shown, for example, as a second source or tool (Source 2) and a third source or tool (Source 3). For example, additional measurement signal 304 can be collected from a metrology device other than Source 1 (such as, Figure 1 170) or from any other desired type of metrology equipment and can be collected from the same Figure 2 The measurement signals 306 may be collected from the same metrology device or the same type of metrology device as source 2 in source 1. The measurement signals 306 may be collected from metrology devices different from source 1 and source 2 (such as different types of measurements from either metrology tool 101 or 170 or from any other desired type of metrology device), and may be collected from the same metrology device or the same type of metrology device as used for source 2. Figure 2 In addition, in some implementations, additional data signals 309 may be obtained, such as process parameters, APC parameters, context data, and sensor data from production equipment, which may be related to the sources (e.g., source 1, source 2, and source 3).

[0056] Signals and data from multiple data sources can be used to extract measurements from one or more physical models. For example, as illustrated by the solid black arrow, a measurement signal 302 from a first source (Source 1) can be used to extract measurements for a sample from a first physical model 312, which can be used with Figure 2312. In some implementations, additional data can be used to help extract measurements from the first physical model 312. For example, as illustrated by the gray dashed arrow, a measurement signal 304 from a second source (Source 2) can be used to help extract measurements from the first physical model 312 of the sample. Additionally, as illustrated by the gray dashed arrow, an additional data signal 309 can be used to help extract measurements for the sample from the first physical model 312.

[0057] In some implementations, multiple physical models can be used to extract measurement results for the sample. For example, as illustrated by the gray dashed arrow and the gray dashed box, the second physical model 314 can be used to extract measurement results for the sample based on the measurement signal 304 from the second source (source 2). Figure 2 314. In some implementations, additional data may be used to help extract measurements from the second physical model 314. For example, as illustrated by the gray dashed arrow, a measurement signal 306 from a third source (source 3) may be used to help extract measurements for a sample from the second physical model 314. In addition, as illustrated by the gray dashed arrow, an additional data signal 309 may be used to help extract measurements for a sample from the second physical model 314. In addition, multiple physical models may be optimized independently or together. For example, in some implementations, as illustrated by the gray dashed line, the first physical model 312 and the second physical model 314 may be linked so that at least some parameters may be coupled across the physical models 312 and 314, and the combined parameter space may be searched to fit the measurement signals from one or more data sources. The first physical model 312 and optionally the second physical model 314 may be configured to provide a goodness of fit 323 for physical modeling.

[0058] One or more trained machine learning models 322 are used to predict parameters of interest 325 based on multiple data sources. Machine learning measurement indicators 327 and goodness of fit 323 from physical modeling can be reported to indicate the quality of the measurements from the collaborative formulation of physical modeling and machine learning. For example, the trained machine learning model 322 can be used with Figure 2The trained machine learning model 322 is the same as the machine learning model 222 of the first physical model 312. As illustrated by the solid black arrow, the trained machine learning model 322 uses the measurement results extracted by the first physical model 312 as input features. As indicated by the dashed black arrow, the trained machine learning model 322 may also use input features that include at least one of the following: a measurement signal 304 from a sample collected from a second source (source 2), a measurement signal 306 from a sample collected from a third source (source 3), an additional data signal 309, a measurement result extracted by the second physical model 314 based on the additional signal 304 and / or 306 and optionally based on the additional data signal 309, or any combination thereof. In some specific implementations, as illustrated by the dashed gray arrow, the trained machine learning model 322 may optionally also use input features including the measurement signal 302 from a sample collected from the first source (source 1). In some implementations, the machine learning input features from the measurement signal 302 may include data from the measurement signal that is not used to extract the measurement result from the first physical model 312, such as at least one data channel or at least one data block, such as reference Figure 2 discussed.

[0059] For example, Figure 4 A workflow 400 for offline recipe creation (e.g., generating one or more physical models and one or more machine learning models) according to a second example scenario is illustrated, where signals are collected from multiple data sources (e.g., different manufacturing process steps). Figure 4 In FIG. 4 , solid black arrows indicate processes used in workflow 400 , dashed black arrows indicate processes that are optional but for which at least one exists, and dashed grey arrows indicate optional processes.

[0060] As illustrated, post-processing step measurement signals 402 are measured from one or more reference samples from a metrology device. For example, the reference sample may be an OCD target pad or a semiconductor device, and the post-processing step measurement signals 402 are obtained after completing the desired manufacturing steps of the sample. The post-processing step measurement signals 402 may be obtained from any desired metrology device (e.g., Figure 1 ) or collect post-processing step measurement signals 402 from any other desired type of metrology equipment.

[0061] Additionally, a pre-processing step measurement signal 402 from one or more reference samples is measured using a metrology device (e.g., the same or different metrology device used to acquire the post-processing step measurement signal 404) and used to generate pre-processing step data. For example, the pre-processing step measurement signal 404 is obtained prior to completing a desired manufacturing step for the sample. In some implementations, the post-processing step measurement signal 402 and the pre-processing step measurement signal 404 may be combined (e.g., combined by addition, subtraction, multiplication, or division) to form a pre-conditioning signal 405. Additionally, data 408 related to the reference samples may be collected, such as reference data for the samples, a design of experiments (DOE). In some implementations, additional data 408 related to the reference samples may also include wafer condition, accuracy, tool matching data, and the like. Additionally, data may be obtained from other sources, such as from a second measurement pad 406, from a fault detection pad 409, or any combination thereof. Although Figure 2 and Figure 3 The first example scenario in emphasizes multiple data sources collected from different metrology equipment, but the second example scenario (for example) illustrates that the multiple data sources can come from different measurement pads or the same pad at different process steps. Different measurement pads can be measured from the same or different metrology equipment. The pre-processing step measurement signal 404 and the post-processing step measurement signal 402 can be measured on a designed OCD target or device. For example, the second measurement pad 406 refers to a pre-processing step measurement and / or post-processing step measurement from a measurement pad that is not measured for the pre-processing step measurement signal 404 and the post-processing step measurement signal 402. For example, if the pre-processing step measurement signal 404 and the post-processing step measurement signal 402 are measured on the OCD target, the second measurement pad 406 can refer to an auxiliary signal from a device pad, or vice versa.

[0062] Signals and data from multiple data sources can be used to generate one or more physical models. For example, as illustrated by the solid black arrow, a post-processing step measurement signal 402 from a metrology device can be used to generate a post-processing physical model 412 of a sample. In some implementations, additional data can be used to help generate the post-processing physical model 412. For example, as illustrated by the dashed gray arrow, additional data 408 (such as reference data and / or DOE) and optionally wafer condition, precision and tool matching data can also be used to help generate the post-processing physical model 412. In another example, as illustrated by the dashed gray arrow, a pre-conditioning signal 405 can be used to help generate a post-processing physical model 412 of a sample. In another example, as illustrated by the dashed gray arrow, a signal from a second measurement pad 406 can be used to help generate a post-processing physical model 412 of a sample. In another example, as illustrated by the dashed gray arrow, a signal from a fault detection pad 409 can be used to help generate a post-processing physical model 412 of a sample. In some implementations, all or any combination of the data 408 and signals from different measurement pads (eg, the second measurement pad 406 and / or the fault detection pad 409 ) can be used to help generate the post-processing physics model 412 .

[0063] In some implementations, multiple physical models can be generated. For example, as illustrated by the gray dashed arrow and the gray dashed box, a pre-processing physical model 414 can be generated based on the pre-processing step measurement signal 404 from the metrology equipment. In some implementations, additional data can be used to generate the pre-processing physical model 414. For example, as illustrated by the gray dashed arrow, additional data 408 (such as reference data and / or DOE) and optionally wafer condition, precision and tool matching data can also be used to help generate the pre-processing physical model 414. In another example, as illustrated by the gray dashed arrow, the signal from the second measurement pad 406 can be used to help generate the pre-processing physical model 414 of the sample. In another example, as illustrated by the gray dashed arrow, the signal from the fault detection pad 409 can be used to help generate the pre-processing physical model 414 of the sample. In some implementations, all or any combination of data 408 and the signals from the second measurement pad 406 and the fault detection pad 409 can be used to help generate the pre-processing physical model 414. In addition, multiple physical models can be optimized independently or together. For example, in some implementations, as illustrated by the gray dashed line, the post-processing physical model 412 and the pre-processing physical model 414 can be linked so that at least some parameters can be coupled across the post-processing physical model 412 and the pre-processing physical model 414, and the combined parameter space can be searched to fit the measurement signals from one or more data sources. The post-processing physical model 412 and optionally the pre-processing physical model 414 can be configured to provide a goodness of fit 423 of the physical modeling.

[0064] One or more machine learning models 422 are built and trained using multiple data sources to predict parameters of interest 425. Machine learning measurement indicators 427 can be developed and reported along with goodness of fit 423 from physical modeling to indicate the quality of measurements from the formulation of physical modeling and machine learning collaboration. As illustrated by the solid black arrows, the machine learning model 422 is built using post-processing measurements extracted by the post-processing physical model 412 as input features. As indicated by the dashed black arrows, the input features of the machine learning model 422 also include pre-processing step data generated based on the pre-processing step measurement signal 404. The pre-processing step data can be generated based on the pre-processing step measurement signal 404 in a variety of ways. For example, Figure 4 As illustrated, the pre-processing step data can be generated from the pre-processing step measurement signal 404 in three different ways labeled 1, 2, and 3, wherein at least one of (1), (2), or (3) or any combination thereof is used. As illustrated by label 1 for the pre-processing step measurement signal 404, the pre-processing step data can be generated by combining the pre-processing step measurement signal 404 with the post-processing step measurement signal 402 to form a pre-conditioned signal 405. Figure 4 As described in , in some implementations, if a pre-conditioned signal 405 is generated, the pre-conditioned signal 405 may be (A) provided to the post-processing physical model 412 and the machine learning model 422 constructed based at least in part on the post-processing measurements extracted by the post-processing physical model 412, or (B) the pre-conditioned signal 405 is provided to the machine learning model 422 and the machine learning model 422 is constructed based at least in part on the pre-conditioned signal 405. In addition, as Figure 4 As further described in , in some implementations, at least one of (A) or (B) can be used with workflow 400. As illustrated by label 2 for pre-processing step measurement signal 404, pre-processing step data can be generated by providing pre-processing step measurement signal 404 to pre-processing physical model 414, and machine learning model 422 can be constructed based at least in part on pre-processing measurement results extracted by pre-processing physical model 414. As illustrated by label 3 for pre-processing step measurement signal 404, pre-processing step data can be generated by providing pre-processing step measurement signal 404 to machine learning model 422, and machine learning model 422 can be constructed based at least in part on pre-processing step measurement signal 404.

[0065] In addition, as indicated by the black dashed arrow, the machine learning model 422 is constructed using additional data including at least one of the following: pre-processing step data (i.e., at least one of (1), (2), or (3) of the pre-processing step measurement signal 404, or any combination thereof), a signal from the second measurement pad 406, and a signal from the fault detection pad 409, or any combination thereof. In some specific implementations, as illustrated by the gray dashed arrow, the machine learning model 422 may optionally be further constructed using the post-processing step measurement signal 402, the pre-conditioned signal 405, the measurement results extracted by the pre-processing physical model 414, or some combination thereof.

[0066] The machine learning model 422 is trained with at least a portion of the data 408, such as reference data and / or DOE, and optionally with wafer condition, accuracy, and tool matching data.

[0067] For example, Figure 5 A workflow 500 for online measurement according to a second example scenario is illustrated, for example, for characterizing a sample based on one or more physical models and one or more machine learning models, where signals are collected from multiple data sources (e.g., different manufacturing process steps). Figure 4 One or more physical models and one or more machine learning models are generated as discussed above. Figure 5 In FIG. 5 , solid black arrows indicate processes used in workflow 500 , dashed black arrows indicate processes that are optional but at least one exists, and dashed grey arrows indicate optional processes.

[0068] As illustrated, a post-processing step measurement signal 502 is collected from a sample from a metrology device. For example, the sample may be an OCD target pad or a semiconductor device, and the post-processing step measurement signal 502 is obtained after completing the desired manufacturing steps of the sample. The post-processing step measurement signal 502 may be collected from any desired metrology device, such as, Figure 1 1) or from any other desired type of metrology equipment and may be collected from any device used to obtain Figure 4 The post-processing step in measurement signal 402 is collected by the same metrology device or the same type of metrology device.

[0069] Additionally, a pre-processing step measurement signal 504 is collected from the sample using a metrology device, such as the same or different metrology device used to obtain the post-processing step measurement signal 502 and the same metrology device used to obtain the post-processing step measurement signal 504. Figure 4The pre-processing step measurement signal 504 is the same metrology device or the same type of metrology device as the pre-processing step measurement signal 404 in the sample. The pre-processing step measurement signal 504 is used to generate the pre-processing step data. For example, the pre-processing step measurement signal 504 is obtained before the required manufacturing steps of the sample are completed. In some specific implementations, the post-processing step measurement signal 502 and the pre-processing step measurement signal 504 can be combined (for example, combined by addition, subtraction, multiplication or division) to form a pre-conditioning signal 505. In addition, data can be obtained from other sources, such as from a second measurement pad 506, from a fault detection pad 509, or any combination thereof. The pre-processing step measurement signal 504 and the post-processing step measurement signal 502 can be measured on a designed OCD target or device. For example, the second measurement pad 506 refers to a pre-processing step measurement and / or a post-processing step measurement from a measurement pad that is not measured for the pre-processing step measurement signal 504 and the post-processing step measurement signal 502. For example, if the pre-processing step measurement signal 504 and the post-processing step measurement signal 502 are measured on an OCD target, the second measurement pad 506 may refer to an auxiliary signal from a device pad, or vice versa.

[0070] Signals and data from multiple data sources can be used to extract measurements from one or more physical models. For example, as illustrated by the solid black arrow, post-processing step measurement signal 502 can be used to extract measurements for the sample from post-processing physical model 512, which can be used with Figure 4 4. In some implementations, additional data may be used to help extract measurements from the post-processing physical model 512. For example, as illustrated by the gray dashed arrow, the pre-conditioned signal 505 may be used to help extract measurements from the post-processing physical model 512 of the sample. In another example, as illustrated by the gray dashed arrow, the signal from the second measurement pad 506 may be used to help extract measurements from the post-processing physical model 512 of the sample. In another example, as illustrated by the gray dashed arrow, the signal from the fault detection pad 509 may be used to help extract measurements from the post-processing physical model 512 of the sample. In some implementations, all or any combination of the signals from the second pad 506 and the fault detection pad 509 may be used to help extract measurements from the post-processing physical model 512 of the sample.

[0071] In some implementations, multiple physical models can be used to extract measurement results for a sample. For example, as illustrated by the gray dashed arrow and gray dashed box, the pre-processing physical model 514 can be used to extract measurement results for the sample based on the pre-processing step measurement signal 504. The pre-processing physical model 514 can be used with Figure 44. In some implementations, additional data may be used to help extract measurements from the preprocessing physical model 514. For example, as illustrated by the gray dashed arrow, the signal from the second measurement pad 506 may be used to help extract measurements from the preprocessing physical model 514 of the sample. In another example, as illustrated by the gray dashed arrow, the signal from the fault detection pad 509 may be used to help extract measurements from the preprocessing physical model 514 of the sample. In some implementations, all or any combination of the signals from the second pad 506 and the fault detection pad 509 may be used to help extract measurements from the preprocessing physical model 514 of the sample. In addition, multiple physical models may be optimized independently or together. For example, in some implementations, as illustrated by the gray dashed line, the post-processing physical model 512 and the preprocessing physical model 514 may be linked so that at least some parameters may be coupled across the post-processing physical model 512 and the preprocessing physical model 514, and the combined parameter space may be searched to fit the measurement signals from one or more data sources. Post-processing physics model 512 and optionally pre-processing physics model 514 may be configured to provide goodness of fit 523 of the physics modeling.

[0072] One or more trained machine learning models 522 are used to predict parameters of interest 525 based on multiple data sources. Machine learning measurement indicators 527 can be developed and reported along with goodness of fit 523 from physical modeling to indicate the measurement quality of the recipe from physical modeling and machine learning collaboration. As illustrated by the solid black arrows, the trained machine learning model 522 uses the post-processing measurements extracted by the post-processing physical model 512 as input data, as well as pre-processing step data generated based on the pre-processing step measurement signal 504.

[0073] The pre-processing step data can be generated in a variety of ways based on the pre-processing step measurement signal 504. For example, Figure 5 As illustrated, the pre-processing step data can be generated from the pre-processing step measurement signal 504 in three different ways labeled 1, 2, and 3, wherein at least one of (1), (2), or (3) or any combination thereof is used. As illustrated by label 1 for the pre-processing step measurement signal 504, the pre-processing step data can be generated by combining the pre-processing step measurement signal 504 with the post-processing step measurement signal 502 to form a pre-conditioned signal 505. Figure 5 As described in , in some implementations, if a pre-conditioned signal 505 is generated, the pre-conditioned signal 505 may be (A) provided to the post-processing physical model 512 and the trained machine learning model 522 receives input data in the form of post-processing measurements extracted by the post-processing physical model 512, or (B) the pre-conditioned signal 505 is provided to the trained machine learning model 522 as input data. In addition, as Figure 5 As further described in , in some implementations, at least one of (A) or (B) can be used with the workflow 500. As illustrated by the label 2 for the pre-processing step measurement signal 504, the pre-processing step data can be generated by providing the pre-processing step measurement signal 504 to the pre-processing physics model 514, and the trained machine learning model 522 uses the measurement results extracted by the pre-processing physics model 514 as input data. As illustrated by the label 3 for the pre-processing step measurement signal 504, the pre-processing step data can be generated by providing the pre-processing step measurement signal 504 as input data to the trained machine learning model 522.

[0074] In some implementations, as illustrated by the dashed grey arrow, the trained machine learning model 522 may optionally also use input data of the measurement signal 502 including a post-processing step.

[0075] In some implementations, primary data (e.g., measurement signals used for physical modeling and in some implementations for machine learning models) and auxiliary data (e.g., supplemental data used for machine learning models and in some implementations for physical modeling) may originate from different tool groups, or may originate from the same tool group but from different data channels, or may originate from the same tool group and the same data channels but from different wavelength ranges, time spans, etc. Different data sources may collect data from the same sample site (e.g., on an OCD target or device) of the same wafer from the same process step or from different process steps. Different data sources may collect data from different sample sites of the same wafer from the same or different process steps, for example, when the underlying structure has related parameters, so that analyzing the combined data can improve the overall performance. As illustrated, at least one physical model may be created to analyze the measurement signals from at least one data source. In addition, if more than one physical model is used, multiple physical models may be optimized independently or together, for example, the physical models may be linked so that at least some parameters may be coupled across the physical models, and the combined parameter space may be searched to fit the measurement signals from one or more data sources. The primary data and the auxiliary data may be of different natures, for example, some of the data may be metrology data collected from a tool set, while other data may be sensor data from process equipment, or wafer processing process equipment such as gas flow rates, APC parameters, or contextual data such as specific processing tools. In addition, feature engineering and signal preprocessing may be applied before providing data from all sources to a machine learning model for training and prediction. Machine learning algorithms may include, for example, but are not limited to, linear regression, neural networks, deep learning, convolutional neural networks (CNNs), ensemble methods, support vector machines (SVMs), random forests, etc., or a combination of multiple models in sequential and / or parallel modes.

[0076] The illustrated workflow effectively combines the use of various measurement techniques and multiple data sources through collaborative physical modeling and machine learning to produce more available information than that provided by a single measurement technique or a single data source. Physical modeling can be performed using previously well-established modeling solutions using the required measurement equipment, and the physical modeling results can be combined with other data that is difficult or impossible to model (which can be called auxiliary data) for machine learning training and prediction. Therefore, the resulting process provides a feasible solution with the advantages of both physical modeling and machine learning, while controlling the computational cost, enabling acceptable TTS for production, and easy implementation and use in practice. In addition, the predictive power can be increased by using data (such as process parameters and sensor data from production equipment), which is combined with metrological data through data mining and data fusion as discussed in this article. The proposed method flexibly adapts to various signals of different properties, while using existing well-developed algorithms to the greatest extent for each type of data source. In addition, the method discussed in this article has a universal application and can be applied to the measurement of any device, OCD or film or other types of targets, for example.

[0077] Figure 6 An illustrative flow chart depicting an example method 600 for characterizing a structure on a sample according to some implementations is shown. In some implementations, the example method 600 may be performed by at least one memory (such as memory 164) configured to store measurement signals, measurement results, one or more physical models, one or more machine learning models, and parameters of interest of the structure, and coupled to one or more processors, such as Figure 1 Implementation in the computing system 160 Figure 3 1. Processor 162 of workflow 300 illustrated in FIG.

[0078] The one or more processors may obtain a measurement signal for a structure on a sample from a first metrology device (602). For example, the measurement signal for the structure on the sample may be obtained by Figure 1 The measurement signal for the structure on the sample can be, for example, Figure 3 The means for obtaining the measurement signal for the structure on the sample from the first metrology device may be, for example, Figure 1 The metering device 100 shown in FIG. Figure 1 At least one memory 164 and at least one processor 162 in the computing system 160 shown in FIG.

[0079] The one or more processors may extract measurement results for the structure on the sample from the first physical model based on the measurement signal (604). For example, the first physical model may be Figure 3 The first physical model 312 shown in FIG. The means for extracting the measurement results for the structure on the sample from the first physical model based on the measurement signal can be, for example, at least one processor 162, which is configured to, for example, extract the measurement results for the structure on the sample based on the measurement signal from a non-transitory computer usable storage medium (such as Figure 1 The one or more physics models are implemented by instructions of the model 164pm of the computer readable program code 166 on the memory 164 shown in FIG.

[0080] The one or more processors may use a machine learning model to determine a parameter of interest for the structure on the sample based on measurements extracted from the first physical model and further based on at least one of: data from a measurement signal not used to extract measurements from the first physical model, the measurement signal from the first metrology device; a second measurement signal for the structure on the sample obtained from a second metrology device; process parameters used to generate the structure on the sample; advanced process control (APC) parameters used to generate the structure on the sample; context data for the structure on the sample; and sensor data from production equipment used to generate the structure on the sample (606). The machine learning model may, for example, be a data set received from a processor. Figure 3 In addition, a second measurement signal for the structure on the sample obtained from the second metrology device may be the measurement signal 304, and a process parameter for generating the structure on the sample, an APC parameter for generating the structure on the sample, context data for the structure on the sample, and sensor data from a production equipment for generating the structure on the sample may be the sensor data. Figure 3 The component for determining the parameter of interest of the structure on the sample based on the measurement results extracted from the first physical model and further based on at least one of the following using the machine learning model can be, for example, configured to be based on, for example, a non-transitory computer usable storage medium (such as Figure 1 At least one processor 162 is configured to implement one or more physical models by executing instructions of a model 164ml of a computer readable program code 166 on a memory 164) shown in FIG. 1 to implement: data from a measurement signal not used to extract a measurement result from the first physical model, the measurement signal being from a first metrology device; a second measurement signal for a structure on a sample obtained from a second metrology device; process parameters for generating the structure on the sample; advanced process control (APC) parameters for generating the structure on the sample; contextual data for the structure on the sample; and sensor data from production equipment used to generate the structure on the sample.

[0081] In some implementations, the data from the measurement signal can be one of: at least one data channel, which can be a measurement subsystem defined by at least one of an energy source (such as a light source), an optical path guided by an optical component, a detector, or a combination thereof; and at least one data block, which can be, for example, a subset of the wavelength, frequency, angle, time span, or any combination of the above from the complete data set provided by the at least one data channel, for example, with reference to a measurement subsystem defined by at least one of an energy source (such as a light source), an optical path guided by an optical component, a detector, or a combination thereof. Figure 3 The measurement signal 302 is provided to the data of the machine learning model 322 as discussed above.

[0082] In some implementations, the machine learning model may be generated based on measurement results of one or more reference samples of the structure extracted by the first physical model, and at least one of reference data and experimental design information, e.g. Figure 2 208 to the machine learning model. The machine learning model may be further generated based on at least one of: data from a measurement signal not used to generate the first physical model, a second measurement signal for one or more reference samples obtained from a second metrology device, process parameters used to generate the one or more reference samples, APC parameters used to generate the one or more reference samples, contextual data for the one or more reference samples, and sensor data from production equipment used to generate the one or more reference samples, such as Figure 2 The self-measurement signal 204 and the additional data signal 209 to the machine learning model 222 are illustrated by the black dashed line.

[0083] In some implementations, the measurement results for the structure on the sample can be further extracted from the first physical model based on a second measurement signal for the structure on the sample from a second metrology device, e.g. Figure 3 This is illustrated by the dashed grey line from the measured signal 304 to the first physical model 312 shown in FIG.

[0084] In some implementations, first measurements of the structure on the sample may be extracted from the first physical model based on at least one of process parameters, APC parameters, context data, and sensor data from production equipment, such as Figure 3 This is illustrated by the grey dashed line from the additional data signal 309 to the first physical model 312 in FIG.

[0085] In some implementations, the one or more processors may further extract a second measurement result for the structure on the sample from the second physical model based on a second measurement signal from the second metrology device, and the machine learning model further determines a parameter of interest of the structure on the sample based on the second measurement result extracted from the second physical model, e.g. Figure 3 304 and a gray dashed line from the measurement signal 304 to the second physical model 314 and a gray dashed line from the second physical model 314 to the trained machine learning model 322. For example, in some implementations, a second measurement result for the structure on the sample is extracted from the second physical model based on a third measurement signal for the structure on the sample from a third metrology device, such as Figure 3 306 to the second physical model 314. For example, in some implementations, a second measurement result for the structure on the sample can be further extracted from the second physical model based on at least one of process parameters, APC parameters, context data, and sensor data from production equipment, such as Figure 3 The means for extracting a second measurement result for the structure on the sample from the second physical model based on the second measurement signal from the second metrology device can be, for example, configured to be based on, for example, a non-transitory computer usable storage medium such as Figure 1 At least one processor 162 is configured to implement one or more physical models using instructions of a model 164pm of a computer readable program code 166 on a memory 164) shown in FIG. 1 , wherein the machine learning model further determines a parameter of interest of a structure on the sample based on a second measurement result extracted from a second physical model.

[0086] In some implementations, the machine learning model determines a parameter of interest for the structure on the sample further based on a second measurement signal from a second metrology device and further based on a third measurement signal for the structure on the sample from a third metrology device, e.g. Figure 3 , as illustrated by the black dashed lines from the measured signal 304 and the measured signal 306 to the trained machine learning model 322.

[0087] Figure 7 An illustrative flow chart depicting an example method 700 for characterizing a structure on a sample according to some implementations is shown. In some implementations, the example method 700 can be performed by at least one memory (such as memory 164) configured to store measurement signals, measurement results, one or more physical models, one or more machine learning models, and parameters of interest of the structure, and coupled to one or more processors, such as Figure 1Implementation in the computing system 160 Figure 5 Processor 162 of workflow 500 illustrated in FIG.

[0088] The one or more processors may obtain a pre-processing step metrology signal for a structure on a sample from a metrology device at a pre-processing step (702). For example, the pre-processing step measurement signal may be obtained by Figure 1 The measurement signal of the preprocessing step can be, for example, Figure 5 The means for obtaining a pre-processing step metrology signal for a structure on a sample from a metrology device at a pre-processing step may be, for example, Figure 1 The metering device 100 shown in FIG. Figure 1 At least one memory 164 and at least one processor 162 in the computing system 160 shown in FIG.

[0089] The one or more processors may obtain a post-processing step measurement signal for a structure on the sample from a metrology device at a post-processing step (704). For example, the post-processing step measurement signal may be obtained by Figure 1 The measurement signal for the post-processing step of the structure on the sample can be, for example, Figure 5 The post-processing step measurement signal 502 shown in FIG. The means for obtaining the post-processing step measurement signal for the structure on the sample from the metrology device at the post-processing step may be, for example Figure 1 The metering device 100 shown in FIG. Figure 1 At least one memory 164 and at least one processor 162 in the computing system 160 shown in FIG.

[0090] The one or more processors may extract post-processing measurement results for the sample from the post-processing physics model based on the post-processing step measurement signal (706). For example, the post-processing physics model may be Figure 5 The post-processing physical model 512 shown in FIG. The means for extracting the post-processing measurement results for the sample from the post-processing physical model based on the post-processing step measurement signal can be, for example, at least one processor 162, which is configured to, for example, based on a non-transitory computer usable storage medium (such as Figure 1 The one or more physics models are implemented by instructions of a model 164pm of a computer readable program code 166 on a memory 164 shown in FIG.

[0091] The one or more processors may generate pre-processing step data based at least on the pre-processing step measurement signal (708). For example, the pre-processing step data generated based at least on the pre-processing step measurement signal may be any pre-processing step data from Figure 5The means for generating the pre-processing step data based at least on the pre-processing step measurement signal may be, for example, Figure 1 The metering device 100 shown in FIG. Figure 1 At least one memory 164 and at least one processor 162 in the computing system 160 shown in FIG.

[0092] The one or more processors may use a machine learning model to determine a parameter of interest for the sample based on the post-processing measurements extracted from the post-processing physics model and the pre-processing step data (710). The trained machine learning model may, for example, be a machine learning model that receives the post-processing measurements extracted from the post-processing physics model 512 and the pre-processing step data (e.g., from Figure 5 1, 2, and 3 of the pre-processing step measurement signal 504 shown in FIG. 5 ). The means for determining the parameter of interest of the sample based on the post-processing measurement results extracted from the post-processing physics model and the pre-processing step data using the machine learning model can be, for example, at least one processor 162, which is configured to, for example, be based on a non-transitory computer usable storage medium (such as Figure 1 The instructions of the model 164ml of the computer readable program code 166 on the memory 164) shown in Figure 1 are used to implement one or more physical models.

[0093] In some implementations, the machine learning model can further determine a parameter of interest of the structure on the sample based on at least one of the preprocessing step measurement signal, the second measurement signal obtained from the measurement pad, and the third measurement signal obtained from the fault detection pad, e.g. Figure 5 504, the signals from the second measurement pad 506 and the fault detection pad 509 to the machine learning model 522. The pre-processing step measurement signal, the second measurement signal obtained from the measurement pad and the third measurement signal obtained from the fault detection pad may originate from different measurement pads, or from the same pad at different process steps, and may be measured from the same or different metrology equipment.

[0094] In some implementations, the post-processing measurement results are further extracted from the post-processing physical model based on at least one of the second measurement signal from the measurement pad and the third measurement signal from the fault detection pad, such as illustrated by the gray dashed arrows from the second measurement pad 506 and the fault detection pad 509 to the post-processing physical model 512.

[0095] In some implementations, the pre-processing step data may include a pre-conditioned signal generated based on a combination of the pre-processing step measurement signal and the post-processing step measurement signal, e.g. Figure 5This is illustrated by the pre-conditioned signal 505 shown in and the grey dashed line from the pre-conditioned signal 505 to the machine learning model 522.

[0096] In some implementations, the one or more processors may further generate a pre-conditioned signal based on a combination of a pre-processing step measurement signal and a post-processing step measurement signal, wherein a post-processing measurement result is extracted from a post-processing physical model further based on the pre-conditioned signal, e.g. Figure 5 , and the grey dashed line from the pre-conditioned signal 505 to the post-processing physical model 512. The means for generating the pre-conditioned signal based on the combination of the pre-processing step measurement signal and the post-processing step measurement signal may be, for example Figure 1 The metering device 100 shown in FIG. Figure 1 At least one memory 164 and at least one processor 162 of the computing system 160 shown in , wherein the post-processing measurement results are extracted from the post-processing physical model further based on the pre-conditioned signal.

[0097] In some implementations, the one or more processors may further extract preprocessing measurements from the preprocessing physical model based on the preprocessing step measurement signals, wherein the preprocessing step data includes the preprocessing measurements extracted from the preprocessing physical model, e.g. Figure 5 , and the gray dashed line from the preprocessing step measurement signal 504 to the preprocessing physical model 514, and the gray dashed line from the preprocessing physical model 514 to the machine learning model 522. The means for extracting the preprocessing measurement results from the preprocessing physical model based on the preprocessing step measurement signal can be, for example, at least one processor 162, which is configured to, for example, be based on a non-transitory computer usable storage medium (such as Figure 1 The instructions of the model 164pm of the computer readable program code 166 on the memory 164 shown in FIG. 1 are used to implement one or more physical models, wherein the pre-processing step data includes pre-processed measurement results extracted from the pre-processed physical models.

[0098] For example, in some implementations, the preprocessed measurement result is extracted from the preprocessed physical model based further on at least one of a second measurement signal obtained from the measurement pad and a third measurement signal obtained from the fault detection pad, e.g. Figure 5 This is illustrated by the dashed grey line from the second measurement pad 506 to the pre-processing physical model 514 and the dashed grey line from the fault detection pad 509 to the pre-processing physical model 514 shown in FIG.

[0099] In some implementations, the pre-processing step data may include a pre-processing step measurement signal, such as Figure 5This is illustrated by the dashed black line from the pre-processing step measurement signal 504 to the machine learning model 522 shown in FIG.

[0100] Figure 8 An illustrative flowchart depicting an example method 800 for characterizing a structure on a sample according to some implementations is shown. In some implementations, the example method 800 can be performed by at least one memory (such as memory 164) configured to store measurement signals, measurement results, one or more physical models, one or more machine learning models, and parameters of interest for the structure, and coupled to one or more processors, such as Figure 1 Implementation in the computing system 160 Figure 2 1. Processor 162 of workflow 200 illustrated in FIG.

[0101] The one or more processors may obtain measurement signals for one or more reference samples of the structure from a first metrology device (802). For example, the measurement signals for the one or more reference samples may be obtained by Figure 1 The measurement signal of one or more reference samples can be, for example, Figure 2 The means for obtaining the measurement signals for one or more reference samples of the structure from the first metrology device may be, for example, Figure 1 The metering device 100 shown in FIG. Figure 1 At least one memory 164 and at least one processor 162 in the computing system 160 shown in FIG.

[0102] The one or more processors may generate a first physical model to extract measurement results for the structure on the sample, wherein the first physical model is generated based on the measurement signal for one or more reference samples from the first metrology device (804). For example, the first physical model generated based on the measurement signal for one or more reference samples from the first metrology device may be Figure 2 The first physical model 212 shown in FIG. 1 is used to generate the first physical model to extract the measurement results for the structure on the sample. The component for generating the first physical model to extract the measurement results for the structure on the sample can be, for example, at least one processor 162, which is configured to generate a first physical model based on a non-transitory computer usable storage medium (such as Figure 1 The one or more physics models are implemented by instructions of a model 164pm of a computer readable program code 166 on a memory 164 shown in FIG. 1 , wherein a first physics model is generated based on a measurement signal for the one or more reference samples from the first metrology device.

[0103] The one or more processors may generate a machine learning model to predict a parameter of interest of a structure on a sample, wherein the machine learning model is generated based on measurements extracted by the first physical model, and at least one of reference data and experimental design information, and further based on at least one of: data from a measurement signal not used to generate the first physical model, the measurement signal being from a first metrology device; a second measurement signal for one or more reference samples obtained from a second metrology device; process parameters used to generate the one or more reference samples; advanced process control (APC) parameters used to generate the one or more reference samples; context data for the one or more reference samples; and sensor data from production equipment used to generate the one or more reference samples (806). The machine learning model for predicting a parameter of interest of a structure on a sample may be, for example, based on measurements extracted by the first physical model 212 and at least one of reference data and experimental design information. Figure 2 The machine learning model 222 is generated by using at least one of the reference data and the experimental design information in the additional data 208 shown in . In addition, the data from the measurement signal can be at least one data channel or at least one data block from the first metrology device that is not used by the first physical model 212, the second measurement signal for one or more reference samples obtained from the second metrology device can be the measurement signal 204, and the process parameters for generating the one or more reference samples, the APC parameters for generating the one or more reference samples, the context data for the one or more reference samples, and the sensor data from the production equipment can be Figure 2 The additional data signal 209 shown in FIG. 1 is used to generate a machine learning model to predict the parameter of interest of the structure on the sample. The component for generating a machine learning model to predict the parameter of interest of the structure on the sample can be, for example, at least one processor 162, which is configured to, for example, based on a non-transitory computer usable storage medium (such as Figure 1 The machine learning model is generated based on the instructions of the model 164ml of the computer readable program code 166 on the memory 164 in the memory 164 to implement one or more physical models, wherein the machine learning model is generated based on the measurement results extracted by the first physical model, and at least one of the reference data and the experimental design information and further based on at least one of the following: data from a measurement signal not used to generate the first physical model, the measurement signal coming from the first metrology device; a second measurement signal for one or more reference samples obtained from the second metrology device; process parameters for generating the one or more reference samples; advanced process control (APC) parameters for generating the one or more reference samples; contextual data for the one or more reference samples; and sensor data from the production equipment used to generate the one or more reference samples.

[0104] In some implementations, the data from the measurement signal can be one of: at least one data channel, which can be a measurement subsystem defined by at least one of an energy source (such as a light source), an optical path guided by an optical component, a detector, or a combination thereof; and at least one data block, which can be, for example, a subset of the wavelength, frequency, angle, time span, or any combination of the above from the complete data set provided by the at least one data channel, for example, with reference to a measurement subsystem defined by at least one of an energy source (such as a light source), an optical path guided by an optical component, a detector, or a combination thereof. Figure 2 The measurement signal 202 is provided to the data of the machine learning model 222 as discussed above.

[0105] In some implementations, the first physical model may be further generated based on a second measurement signal for one or more reference samples from a second metrology device, e.g. Figure 2 This is illustrated by the dashed grey line from the measured signal 204 to the first physical model 212 shown in FIG.

[0106] In some implementations, the first physical model can be further generated based on at least one of process parameters, APC parameters, context data, and sensor data from production equipment, such as Figure 2 This is illustrated by the grey dashed line from the additional data signal 209 to the first physical model 212 in FIG.

[0107] In some implementations, the one or more processors may further generate a second physical model to extract second measurements for the structure on the sample, wherein the second physical model is generated based on a second measurement signal for one or more reference samples from a second metrology device, and the machine learning model may be further generated based on the second measurements extracted by the second physical model, e.g. Figure 2 , and a gray dashed line from the measurement signal 204 to the second physical model 214 and a gray dashed line from the second physical model 214 to the machine learning model 222. For example, in some implementations, the second physical model is further generated based on a third measurement signal for one or more reference samples from a third metrology device, such as Figure 2 206 to the second physical model 214. For example, in some implementations, the second physical model is further generated based on at least one of process parameters, APC parameters, context data, and sensor data from production equipment, such as Figure 2 209 to the second physical model 214. The means for generating the second physical model to extract the second measurement result for the structure on the sample can be, for example, at least one processor 162, which is configured to generate the second physical model based on, for example, a non-transitory computer usable storage medium such as Figure 1 The machine learning model may be further generated based on second measurement results extracted by the second physical model.

[0108] In some implementations, the machine learning model can be generated further based on a second measurement signal from a second metrology device and further based on a third measurement signal for one or more reference samples from a third metrology device, e.g. Figure 2 exemplified by the black dashed lines from the self-measurement signal 204 and the measurement signal 206 to the machine learning model 222.

[0109] Fig. 9 An illustrative flowchart depicting an example method 900 for characterizing a structure on a sample according to some implementations is shown. In some implementations, the example method 900 can be performed by at least one memory (such as memory 164) configured to store measurement signals, measurement results, one or more physical models, one or more machine learning models, and parameters of interest for the structure and coupled to one or more processors, such as Figure 1 Implementation in the computing system 160 Figure 4 Processor 162 of workflow 400 illustrated in FIG.

[0110] The one or more processors may obtain, at a pre-processing step, a pre-processing step measurement signal for one or more reference samples of the structure from a metrology device (902). For example, the pre-processing step measurement signal for the one or more reference samples may be obtained by Figure 1 The measurement signal can be obtained by the metrology device 100 shown in FIG. Figure 4 The means for obtaining the pre-processing step measurement signals for one or more reference samples of the structure from the metrology device at the pre-processing step may be, for example, Figure 1 The metering device 100 shown in FIG. Figure 1 At least one memory 164 and at least one processor 162 in the computing system 160 shown in FIG.

[0111] The one or more processors may obtain a post-processing step measurement signal for one or more reference samples from a metrology device at a post-processing step (904). For example, the post-processing step measurement signal for one or more reference samples may be obtained by Figure 1 The measurement signal for the post-processing step of one or more reference samples can be, for example, Figure 4 The post-processing step measurement signal 402 is shown in FIG. The means for obtaining the post-processing step measurement signal for one or more reference samples from the metrology device at the post-processing step may be, for example Figure 1 The metering device 100 shown in FIG. Figure 1 At least one memory 164 and at least one processor 162 in the computing system 160 shown in FIG.

[0112] The one or more processors may generate a post-processing physics model to extract post-processing measurement results for one or more reference samples, wherein the post-processing physics model is generated based on the post-processing step measurement signal (906). For example, the post-processing physics model generated based on the post-processing step measurement signal may be Figure 4 The post-processing physical model 412 shown in FIG. The means for generating the post-processing physical model to extract the post-processing measurement results for one or more reference samples can be, for example, at least one processor 162, which is configured to generate a post-processing physical model based on, for example, a non-transitory computer usable storage medium (such as Figure 1 The one or more physics models are implemented by instructions of a model 164pm of a computer readable program code 166 on a memory 164 shown in FIG. 1 , wherein the post-processing physics model is generated based on a post-processing step measurement signal.

[0113] The one or more processors may generate pre-processing step data based at least on the pre-processing step measurement signal (908). For example, the pre-processing step data generated based at least on the pre-processing step measurement signal may be any pre-processing step data from Figure 4 The means for generating the pre-processing step data based at least on the pre-processing step measurement signal may be, for example, Figure 1 The metering device 100 shown in FIG. Figure 1 At least one memory 164 and at least one processor 162 in the computing system 160 shown in FIG.

[0114] The one or more processors may generate a machine learning model to predict a parameter of interest of a structure on the sample, wherein the machine learning model is generated based on post-processing measurements extracted by the post-processing physics model, and at least one of reference data and experimental design information and pre-processing step data (910). The machine learning model may be, for example, based on post-processing measurements extracted by the post-processing physics model 412, and Figure 4 At least one of the reference data and the experimental design information in the additional data 408 shown in FIG. 4 and the pre-processing step data (e.g., from Figure 4The machine learning model 422 is generated by using any of the labels 1, 2, and 3 of the measurement signal 404 in the preprocessing step shown in FIG. The component for generating the machine learning model to predict the parameter of interest of the structure on the sample can be, for example, at least one processor 162, which is configured to, for example, be based on a non-transitory computer usable storage medium (such as Figure 1 The machine learning model is configured to implement one or more physical models by means of instructions of a model 164ml of a computer readable program code 166 on a memory 164) shown in FIG. 1 , wherein the machine learning model is generated based on post-processing measurement results extracted by the post-processing physical model, and at least one of reference data and experimental design information and pre-processing step data.

[0115] In some implementations, the machine learning model may be further generated based on at least one of the preprocessing step measurement signal, the second measurement signal obtained from the measurement pad, and the third measurement signal obtained from the fault detection pad, e.g. Figure 4 406, and from the fault detection pad 409 to the machine learning model 422. The pre-processing step measurement signal, the second measurement signal obtained from the measurement pad, and the third measurement signal obtained from the fault detection pad may originate from different measurement pads, or from the same pad at different process steps, and may be measured from the same or different metrology equipment.

[0116] In some implementations, the post-processing physical model may be further generated based on at least one of the second measurement signal from the measurement pad and the third measurement signal from the fault detection pad, e.g. Figure 4 This is illustrated by the dashed grey arrows from the second measurement pad 406 and the fault detection pad 409 to the post-processing physical model 412 shown in FIG.

[0117] In some implementations, the pre-processing step data may include a pre-conditioned signal generated based on a combination of the pre-processing step measurement signal and the post-processing step measurement signal, e.g. Figure 4 , as illustrated by the pre-conditioned signal 405 shown in , and the grey dashed line from the pre-conditioned signal 405 to the machine learning model 422 .

[0118] In some implementations, the one or more processors may further generate a pre-conditioned signal based on a combination of a pre-processing step measurement signal and a post-processing step measurement signal, wherein the post-processing physical model is further generated based on the pre-conditioned signal, e.g. Figure 4 405 and the grey dashed line from the pre-conditioned signal 405 to the post-processing physical model 412. The means for generating the pre-conditioned signal based on the combination of the pre-processing step measurement signal and the post-processing step measurement signal may be, for example Figure 1The metering device 100 shown in FIG. Figure 1 At least one memory 164 and at least one processor 162 of the computing system 160 shown in , wherein the post-processing physical model is further generated based on the pre-conditioned signal.

[0119] In some implementations, the one or more processors may further generate a preprocessing physics model to extract preprocessing measurements for the sample, wherein the preprocessing physics model is generated based on preprocessing step measurement signals for the one or more reference samples, and the preprocessing step data includes preprocessing measurements extracted from the preprocessing physics model, e.g. Figure 4 4 and the gray dashed line from the preprocessing step measurement signal 404 to the preprocessing physical model 414 and the gray dashed line from the preprocessing physical model 414 to the machine learning model 422. The component for generating the preprocessing physical model to extract the preprocessing measurement results for the sample can be, for example, at least one processor 162, which is configured to, for example, generate a preprocessing physical model based on a non-transitory computer usable storage medium (such as Figure 1 The computer program code 166 of the memory 164 shown in the figure is used to implement one or more physical models by instructions of the model 164pm of the computer readable program code 166 on the memory 164) shown in the figure, wherein the preprocessing physical model is generated based on the preprocessing step measurement signals for one or more reference samples, and the preprocessing step data includes preprocessing measurement results extracted from the preprocessing physical model.

[0120] For example, in some implementations, the preprocessing physical model may be further generated based on at least one of a second measurement signal obtained from the measurement pad and a third measurement signal obtained from the fault detection pad, e.g. Figure 4 This is illustrated by the dashed grey line from the second measurement pad 406 to the pre-processing physical model 414 and the dashed grey line from the fault detection pad 409 to the pre-processing physical model 414 shown in FIG.

[0121] In some implementations, the pre-processing step data may include a pre-processing step measurement signal, such as Figure 4 This is illustrated by the dashed black line from the pre-processing step measured signal 404 to the machine learning model 422 shown in FIG.

[0122] The above description is intended to be illustrative rather than limiting. For example, the above examples (or one or more aspects thereof) may be used in combination with each other. Other specific implementations may be used, such as those of ordinary skill in the art after reading the above description. In addition, various features may be grouped together, and less than all features of a specific disclosed specific implementation may be used. Therefore, the following aspects are thus incorporated into the above description as examples or specific implementations, wherein each aspect is independently used as a separate specific implementation, and it is contemplated that such specific implementations may be combined with each other in various combinations or permutations. Therefore, the spirit and scope of the appended claims should not be limited to the foregoing description.

Claims

1. A method for characterizing a structure on a sample, the method comprising: obtaining a measurement signal for the structure on the sample from a first metrology device; extracting measurement results for the structure on the sample from a first physical model based on the measurement signal; as well as determining, using a machine learning model, a parameter of interest of the structure on the sample based on the measurements extracted from the first physical model and further based on at least one of: data from a measurement signal not used to extract the measurements from the first physical model, the measurement signal from the first metrology device; a second measurement signal obtained from a second metrology device for the structure on the sample; process parameters for generating the structure on the sample; advanced process control (APC) parameters used to generate the structure on the sample; context data for the structure on the sample; and sensor data from production equipment used to generate the structure on the sample.

2. A method according to claim 1, wherein the data from the measurement signal comprises one of: at least one data channel, the at least one data channel comprising a measurement subsystem defined by at least one of a light source, an optical path guided by optical components, a detector, or a combination thereof; and at least one data block, the at least one data block comprising a subset of wavelength, frequency, angle, time span, or any combination thereof from a complete data set provided by the at least one data channel.

3. The method of claim 1 , wherein the machine learning model is generated based on measurement results of one or more reference samples of the structure extracted by the first physical model, and at least one of reference data and experimental design information, and at least one of: data from a measurement signal not used to generate the first physical model, a second measurement signal for the one or more reference samples obtained from the second metrology device, process parameters used to generate the one or more reference samples, APC parameters used to generate the one or more reference samples, contextual data for the one or more reference samples, and sensor data from production equipment used to generate the one or more reference samples. 4 . The method of claim 1 , wherein the measurement result for the structure on the sample is extracted from the first physical model further based on the second measurement signal for the structure on the sample from the second metrology device.

5. The method of claim 1, wherein the measurements extracted from the first physical model for the structure on the sample are further based on at least one of the process parameters, the APC parameters, the contextual data, and the sensor data from production equipment.

6. The method of claim 1 , further comprising extracting a second measurement result for the structure on the sample from a second physical model based on the second measurement signal from the second metrology device, wherein the machine learning model further determines the parameter of interest of the structure on the sample based on the second measurement result extracted from the second physical model. 7 . The method of claim 6 , wherein the second measurement result for the structure on the sample is extracted from the second physical model further based on a third measurement signal for the structure on the sample from a third metrology device.

8. The method of claim 6, wherein the second measurement result for the structure on the sample is extracted from the second physical model further based on at least one of the process parameter, the APC parameter, the context data, and the sensor data from production equipment.

9. The method of claim 1 , wherein the machine learning model determines the parameter of interest for the structure on the sample further based on the second measurement signal from the second metrology device and further based on a third measurement signal for the structure on the sample from a third metrology device.

10. A computer system configured to characterize a structure on a sample, the computer system comprising: at least one memory configured to store the measurement signal, the measurement result, the first physical model, the machine learning model, and the parameter of interest of the structure; and at least one processor coupled to the at least one memory, wherein the at least one processor is configured to: obtaining a measurement signal for the structure on the sample from a first metrology device; extracting measurement results for the structure on the sample from a first physical model based on the measurement signal; as well as determining, using a machine learning model, a parameter of interest of the structure on the sample based on the measurements extracted from the first physical model and further based on at least one of: data from a measurement signal not used to extract the measurements from the first physical model, the measurement signal from the first metrology device; a second measurement signal obtained from a second metrology device for the structure on the sample; process parameters for generating the structure on the sample; advanced process control (APC) parameters used to generate the structure on the sample; context data for the structure on the sample; and sensor data from production equipment used to generate the structure on the sample.

11. A computer system according to claim 10, wherein the data from the measurement signal comprises one of: at least one data channel, the at least one data channel comprising a measurement subsystem defined by at least one of a light source, an optical path guided by optical components, a detector, or a combination thereof; and at least one data block, the at least one data block comprising a subset of wavelength, frequency, angle, time span, or any combination thereof from a complete data set provided by the at least one data channel.

12. The computer system of claim 10 , wherein the machine learning model is generated based on measurement results of one or more reference samples of the structure extracted by the first physical model, and at least one of reference data and experimental design information, and at least one of: data from a measurement signal not used to generate the first physical model, a second measurement signal for the one or more reference samples obtained from the second metrology device, process parameters used to generate the one or more reference samples, APC parameters used to generate the one or more reference samples, contextual data for the one or more reference samples, and sensor data from production equipment used to generate the one or more reference samples.

13. The computer system of claim 10, wherein the measurement result for the structure on the sample is extracted from the first physical model further based on the second measurement signal for the structure on the sample from the second metrology device.

14. The computer system of claim 10, wherein the measurements extracted from the first physical model for the structure on the sample are further based on at least one of the process parameters, the APC parameters, the contextual data, and the sensor data from production equipment.

15. The computer system of claim 10, wherein the at least one processor is further configured to extract a second measurement result for the structure on the sample from a second physical model based on the second measurement signal from the second metrology device, wherein the machine learning model further determines the parameter of interest of the structure on the sample based on the second measurement result extracted from the second physical model. 16 . The computer system of claim 15 , wherein the second measurement result for the structure on the sample is extracted from the second physical model further based on a third measurement signal for the structure on the sample from a third metrology device.

17. The computer system of claim 15, wherein the second measurement for the structure on the sample is extracted from the second physical model based further on at least one of the process parameter, the APC parameter, the context data, and the sensor data from production equipment.

18. The computer system of claim 10, wherein the machine learning model determines the parameter of interest for the structure on the sample further based on the second measurement signal from the second metrology device and further based on a third measurement signal for the structure on the sample from a third metrology device.

19. A system configured for characterizing a structure on a sample, the system comprising: means for obtaining a measurement signal for said structure on said sample from a first metrology device; means for extracting a measurement result for the structure on the sample from a first physical model based on the measurement signal; and means for determining, using a machine learning model, a parameter of interest of the structure on the sample based on the measurements extracted from the first physical model and further based on at least one of: data from a measurement signal not used to extract the measurements from the first physical model, the measurement signal from the first metrology device; a second measurement signal obtained from a second metrology device for the structure on the sample; process parameters for generating the structure on the sample; advanced process control (APC) parameters used to generate the structure on the sample; context data for the structure on the sample; and sensor data from production equipment used to generate the structure on the sample.

20. The system of claim 19, further comprising a component for extracting a second measurement result for the structure on the sample from a second physical model based on the second measurement signal from the second metrology device, wherein the machine learning model further determines the parameter of interest of the structure on the sample based on the second measurement result extracted from the second physical model.

Images