Semiconductor Metrology with Robust In-Line Tool Matching

Abstract

A method and system are presented herein for improved monitoring of tool drift and inter-tool matching across a large fleet of measurement systems used to measure semiconductor structures. One or more quality control (QC) wafers are measured by each of the fleet of measurement systems. Values ​​of system variables are extracted from QC measurement data associated with each measurement system using a trained QC encoder. The extracted values ​​of the system variables are used to calibrate corresponding measurement models used by each measurement tool to characterize structures to be measured having unknown values ​​of one or more parameters of interest. Accurate inter-tool matching across the fleet of calibrated measurement systems is achieved by extracting values ​​of the system variables from measurement data collected from the same set of QC wafers. Tool health is monitored based on changes in values ​​of the system variables extracted from measurements made at various times.

Description

[Technical field]

[0001] The described embodiments relate to metrology systems and methods, and more particularly, to methods and systems for improved measurement of semiconductor structures. [Background technology]

[0002] Semiconductor devices, such as logic and memory devices, are typically fabricated by a series of processing steps applied to a specimen. The various features and multiple structural levels of a semiconductor device are formed by these processing steps. For example, lithography is one semiconductor manufacturing process that involves, among other things, creating patterns on a semiconductor wafer. Further examples of semiconductor manufacturing processes include, but are not limited to, chemical mechanical polishing, etching, deposition, and ion implantation. Multiple semiconductor devices can be fabricated on a single semiconductor wafer and then separated into individual semiconductor devices.

[0003] Metrology processes are used at various steps during the semiconductor manufacturing process to detect defects on the wafer and improve yield. Optical and X-ray based metrology techniques offer the potential for high throughput without the risk of destroying the sample. Many metrology-based techniques, including implementations of scatterometry, reflectometry, and ellipsometry and associated analysis algorithms, are commonly used to characterize the critical dimensions, film thickness, composition, overlay, and other parameters of nanoscale structures.

[0004] The performance, integration, and reliability of semiconductor devices have continuously improved over time due to improved process resolution and more complex device structures. The improved process resolution allows for a reduction in the minimum critical size of the structures to be fabricated. The process resolution is primarily determined by the wavelength of the light source used in the fabrication process. The latest extreme ultraviolet lithography (EUV) light sources produce wavelengths of 13.5 nanometers, which allows for the fabrication of structural features smaller than 32 nanometers. In addition, more complex device structures, such as FinFET and vertical NAND structures, are being developed to improve overall performance, energy costs, integration, and reliability.

[0005] As devices (e.g., logic and memory devices) move to smaller nanometer-scale dimensions, characterization becomes more difficult. Devices incorporating complex three-dimensional geometries and materials with diverse physical properties contribute to the difficulty of characterization. Metrology systems are generally required to measure devices at more process steps with greater precision.

[0006] In addition to accurate device characterization, measurement consistency across different measurement applications and metrology systems tasked with the same measurement objectives is also important. If measurement consistency is compromised in a manufacturing environment, consistency between processed semiconductor wafers will be lost and yields will decrease to unacceptable levels. Matching measurement results across applications and across multiple systems (i.e., tool-to-tool matching) ensures that measurements on the same wafer in the same application will produce the same results.

[0007] Each metrology tool in a fleet has systematic errors even if the hardware configuration of each metrology tool in the fleet is properly calibrated. These systematic errors result in offsets in the measurement results between different tools in the fleet.

[0008] In some examples, to compensate for these systematic errors, an offset value assigned to each metrology tool is added to the measurement results associated with each tool, after which the measurement results monitored within a statistical process control (SPC) system are consistent across a fleet of metrology tools.

[0009] Conventionally, the offset calibration value associated with each tool is determined from measurements of a set of dedicated quality control (QC) wafers measured by each of the metrology tools in the family. The offset calibration value for each tool is estimated based on the raw measurement data. By minimizing the effect of tool mismatch, the process variation captured by measurements made by the metrology tool family is effectively magnified.

[0010] Unfortunately, in many metrology applications, the offset calibration values ​​determined solely from QC wafer measurements and applied to the measurement results only take into account a limited subset of all system-to-system differences, so in practice, system variations between metrology tools are not adequately captured by the application of simple offset calibration values.

[0011] Many metrology techniques are indirect methods of measuring the physical properties of a sample of interest. In most cases, the raw measurement signal cannot be used to directly determine the physical properties of the sample. Instead, some kind of metrology model can be iteratively solved to match the raw measurement signal. Once solved, the metrology model is used to estimate values ​​for one or more parameters of interest.

[0012] In some examples, a trained machine learning based measurement model is used to directly estimate values ​​of the parameters of interest based on the raw measurement data. In these examples, the machine learning based measurement model takes the raw measurement signals as model inputs and produces values ​​of the parameters of interest as model outputs.

[0013] To generate useful estimates of the parameters of interest for a particular measurement application, the machine learning-based measurement model must be trained. Typically, the training of the model is based on raw measurement signals collected from samples with known values ​​of the parameters of interest (i.e., design of experiments (DOE) data). Traditionally, the machine learning-based measurement model is trained by a regression process (e.g., ordinary least squares regression). The machine learning-based measurement model is parameterized by several weight parameters. The values ​​of the weight parameters are iteratively adjusted to minimize the difference between the known reference value of the parameter of interest and the value of the parameter of interest estimated by the machine learning-based measurement model based on the measured raw measurement signals.

[0014] In some examples, each machine learning based metrology model associated with a metrology tool in the family of metrology tools is trained using layer-specific measurement data from multiple tools in the family to reduce the measurement sensitivity to systematic errors associated with any particular tool in the family of tools.

[0015] Unfortunately, in many applications, training a machine learning based measurement model associated with a particular metrology tool based on measurement data associated with multiple tools in a fleet degrades measurement performance due to correlation between the critical parameters being measured and the system parameters, which is particularly problematic for measurement applications with low sensitivity to critical parameters where the tool-to-tool difference is greater than the measurement sensitivity associated with the critical parameter.

[0016] In some examples, each machine learning based measurement model associated with a metrology tool in the family of metrology tools is trained using synthetically generated data having perturbations in system model parameters to reduce the measurement sensitivity to systematic errors associated with any particular tool in the family of tools.

[0017] Unfortunately, training a machine learning-based measurement model associated with a particular metrology tool based on synthetically generated data using system parameter perturbations takes into account only a limited subset of all differences between tools. In addition, in many applications, training a machine learning-based measurement model associated with a particular metrology tool based on synthetically generated data using system parameter perturbations degrades measurement performance due to correlations between the critical parameters to be measured and the system parameters. Again, this is particularly problematic for measurement applications with low sensitivity to critical parameters, where the differences between tools are greater than the measurement sensitivity associated with the critical parameters. [Prior art documents] [Patent documents]

[0018] [Patent Document 1] US Patent Application Publication No. 2007 / 0229855 Summary of the Invention [Problem to be solved by the invention]

[0019] As metrology systems have evolved to measure devices at more process steps with greater precision, the inter-tool calibration process has become more complex and less efficient. Improved methods and tools for reducing the time and costs associated with matching measurement results across a fleet of metrology tools are desirable. [Means for solving the problem]

[0020] Presented herein are methods and systems for improved monitoring of tool drift and inter-tool matching across a large fleet of measurement systems used to measure semiconductor structures. One or more quality control (QC) wafers are measured by each of the fleet of measurement systems. Values ​​of system variables are extracted from QC measurement data associated with each measurement system using a trained QC encoder. The extracted values ​​of the system variables are used to adjust corresponding measurement models used by each measurement tool to characterize structures of interest having unknown values ​​of one or more parameters of interest.

[0021] By extracting the values ​​of system variables directly from the measurement data collected by each metrology system from the same set of QC wafers, tool variations are decoupled from any wafer-to-wafer variations. In this way, precise tool-to-tool matching across a fleet of tuned metrology systems is achieved.

[0022] In a further aspect, the extracted values ​​of the system variables provide a monitor of tool variation over time. The health of each measurement system (e.g., drift, matching, etc.) can be monitored over time by measuring the same set of QC wafers at different times with the same trained QC encoder. In some embodiments, preventive maintenance actions are initiated based on changes in the values ​​of one or more system variables. In one example, a preventive maintenance action is initiated if the difference in values ​​of one or more system variables associated with different measurement instances exceeds a predetermined threshold.

[0023] In another embodiment, the trained layer encoder functions are trained as an autoencoder. Since the same measurement signals are provided as input and output of the autoencoder, information indicative of both structural and system variations is compressed into a small number of variables.

[0024] In another embodiment, the trained QC encoder function is trained as an autoencoder. Because the same QC measurement signals are provided as input and output of the autoencoder, the information indicative of the system variations is compressed into a smaller number of variables. In addition, because the QC measurement signals are collected from multiple tools but from the same QC wafer, the system variables capture the tool-to-tool variations without the wafer variations. In this way, the trained QC encoder is an encoder of the system variables rather than the wafer variables.

[0025] In another aspect, the metrology tools comprising the group of matched metrology tools may include various types of metrology tools.

[0026] In another further embodiment, multiple metrology tools (eg, measurement subsystems) are used to perform a single hybrid measurement.

[0027] The above is a summary and, as such, necessarily contains simplifications, generalizations, and omissions of details. Thus, those skilled in the art will appreciate that the summary is illustrative only and is not intended to be limiting in any way. Other aspects, inventive features, and advantages of the devices and / or processes described herein will become apparent in the non-limiting detailed description set forth herein. [Brief description of the drawings]

[0028] [Figure 1] FIG. 1 illustrates an embodiment of a small angle X-ray scatterometry (SAXS) metrology tool for measuring properties of a sample according to exemplary methods presented herein. [Diagram 2] FIG. 2 illustrates an exemplary quality control (QC) tuned measurement engine in one embodiment. [Diagram 3] FIG. 2 illustrates a layer encoder training engine in one embodiment. [Figure 4] FIG. 2 illustrates a QC encoder training engine in one embodiment. [Diagram 5]FIG. 1 illustrates an exemplary quality control (QC) tuned measurement model training engine in one embodiment. [Figure 6] 2 shows a flowchart of a method 200 for estimating values ​​of one or more parameters of interest using a trained QC-tuned measurement model in at least one novel embodiment. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0029] Reference will now be made in detail to background examples and certain embodiments of the invention, examples of which are illustrated in the accompanying drawings.

[0030] A method and system are presented herein for improved monitoring of tool drift and inter-tool matching across a large fleet of metrology systems used to measure semiconductor structures. As described herein, one or more quality control (QC) wafers are measured by each of the fleet of metrology systems. The QC wafers have structures similar to those fabricated on the in-line wafers that are measured by the fleet of metrology systems. Values ​​of system variables are extracted from QC measurement data associated with each metrology system of the fleet using a trained QC encoder. The extracted values ​​of the system variables associated with each metrology tool are used as inputs to tune a measurement model used by the same metrology tools to characterize structures of interest having unknown values ​​of one or more parameters of interest.

[0031] By extracting the values ​​of the system variables directly from the measurement data collected by each metrology system from the same set of QC wafers, the tool variations are decoupled from any wafer-to-wafer variations that exist between in-line wafers. Thus, the values ​​of the system variables decorrelate the tool variations from the variations of the structures being measured on the wafers. By extracting the decorrelated system variations from the structures being measured, accurate tool-to-tool matching across a fleet of coordinated metrology systems is achieved, even when different in-line wafers are measured by each metrology tool in the fleet.

[0032] Additionally, by extracting values ​​of the system variables directly from the measurement data collected from the QC wafers by each metrology system using the same trained QC encoder, tool variations that are not otherwise captured in the metrology model associated with a particular metrology tool are captured. As a result, the system variables are decorrelated from any calibration parameters associated with the metrology model.

[0033] In a further embodiment, the extracted values ​​of the system variables provide a monitor of tool variation over time. By measuring the same set of QC wafers at different times with the same trained QC encoders, the health of each measurement system can be monitored over time (e.g., drift, matching, etc.).

[0034] FIGURE 1 illustrates an embodiment of a SAXS metrology tool 100 for measuring properties of a sample according to exemplary methods presented herein. As shown in FIGURE 1, the system 100 can be used to perform X-ray scatterometry measurements of structures within a measurement area 102 of a sample 101 disposed on a sample positioning system 140. In some embodiments, the measurement area 102 has a spot size of 80 micrometers or less. In some embodiments, the measurement area 102 has a spot size of 50 micrometers or less. In some embodiments, the measurement area 102 has a spot size of 40 micrometers or less.

[0035] In the embodiment shown, the SAXS metrology tool 100 includes an X-ray radiation source 110 that generates X-ray radiation suitable for X-ray scatterometry measurements. In some embodiments, the X-ray radiation source 110 produces wavelengths between 0.01 nanometers and 1 nanometer. The X-ray radiation source 110 produces an X-ray beam 117 that is incident on the examination area 102 of the sample 101.

[0036] In general, any suitable high brightness X-ray radiation source capable of generating high brightness X-rays at flux levels sufficient to enable high throughput in-line metrology may be contemplated for providing X-ray radiation for the X-ray scatterometry measurements described herein, in some embodiments the X-ray source includes a tunable monochromator that enables the X-ray source to deliver X-ray radiation at different selectable wavelengths.

[0037] In some embodiments, one or more X-ray sources emitting at photon energies greater than 15 keV are used to ensure that the X-ray source provides light at a wavelength sufficient to penetrate the entire device and the wafer substrate. As non-limiting examples, any of particle acceleration sources, liquid anode sources, rotating anode sources, fixed solid anode sources, microfocus sources, microfocus rotating anode sources, and inverse Compton sources may be utilized as the X-ray source 110. In one example, an inverse Compton source available from Lyncean Technologies, Inc., Palo Alto, California (USA) may be contemplated. Inverse Compton sources have the added advantage that X-rays can be generated over a range of photon energies, thereby allowing the X-ray source to deliver X-ray radiation at different selectable wavelengths. Exemplary X-ray sources include electron beam sources configured to bombard a solid or liquid target to stimulate X-ray radiation. In some embodiments, the X-ray irradiation source 110 is a liquid metal-based X-ray irradiation system. A jet of liquid metal containing one or more elements is generated. As non-limiting examples, the liquid metal jet includes any of aluminum, gallium, indium, tin, thallium, and bismuth. An electron beam source (e.g., an electron gun) generates a stream of electrons that are directed by electron-optical elements to the liquid metal jet. Suitable electron-optical elements for focusing the electron beam and directing the beam to the liquid metal jet include electromagnets, permanent magnets, and combinations of electromagnets and permanent magnets. The coincidence of the liquid metal jet and the electron stream generates an X-ray beam 117 that is incident on the inspection region 102 of the specimen 101. In this manner, the liquid metal jet generates X-ray lines that correspond to its constituent elements. In one embodiment, the liquid metal jet includes an alloy of gallium and indium.

[0038] A method and system for producing high brightness liquid metal x-ray radiation is described in U.S. Pat. No. 7,929,667, issued to KLA-Tencor Corp. on Apr. 19, 2011, which is incorporated herein by reference in its entirety.

[0039] In one embodiment, the incident X-ray beam 117 is at the Indium kα line at 24.2 keV. The X-ray beam is collimated with a divergence of less than 1 milliradian using multi-layer X-ray optics for X-ray scatterometry measurements.

[0040] In some embodiments, the X-ray scattering measurements described herein are achieved without a screen located between the X-ray source and the sample being measured. In these embodiments, the measured intensity of the beam over a range of angles of incidence, multiple wavelengths, or a combination of both, provides sufficient information to resolve a distribution map (i.e., an image) of the desired property (e.g., complex refractive index, electron density, or absorptivity) of the material of the measured structure. However, in some other examples, to improve collimation of the X-ray beam, a pinhole or another aperture is located on another opaque screen located between the X-ray source and the sample being measured. The intensity of the diffraction pattern is measured at several positions of the aperture. In some other embodiments, a screen with a pseudorandom pattern of apertures is used and the diffraction pattern is measured at multiple screens. These approaches can also be designed to provide additional information to resolve the three-dimensional distribution of the desired property of the material of the measured structure.

[0041] In some embodiments, the profile of the incident X-ray beam is controlled by two or more apertures, slits, or a combination thereof, hi further embodiments, the apertures, slits, or both are configured to rotate in coordination with the sample orientation to optimize the incident beam profile for each incidence angle, azimuthal angle, or both.

[0042] As shown in FIG. 1 , the X-ray optics 115 form and direct the incident X-ray beam 117 to the sample 101. In some examples, the X-ray optics 115 includes an X-ray monochromator to monochromatize the X-ray beam incident on the sample 101. In one example, a crystal monochromator, such as a Loxley-Tanner-Bowen monochromator, is used to monochromatize the beam of X-ray radiation. In some examples, the X-ray optics 115 uses multi-layer X-ray optics to collimate or focus the X-ray beam 117 onto the inspection region 102 of the sample 101 with a divergence of less than 1 milliradian. In some embodiments, the X-ray optics 115 include one or more X-ray collimating mirrors, X-ray apertures, X-ray beam stops, X-ray refractive optics, diffractive optics such as zone plates, specular X-ray optics such as grazing incidence ellipsoidal mirrors, polycapillary optics such as hollow capillary X-ray waveguides, multi-layer optics or systems, or any combination thereof. Further details are described in U.S. Patent Application Publication No. 2015 / 0110249, the contents of which are incorporated herein by reference in their entirety.

[0043] Typically, the focal plane of the illumination optical system is optimized for each measurement application. In this manner, system 100 can be configured to place the focal plane at different depths within the sample depending on the measurement application.

[0044] The X-ray detector 116 collects X-ray radiation 125 scattered from the sample 101 and generates an output signal 126 indicative of a characteristic of the sample 101 that is sensitive to the incident X-ray radiation. In some embodiments, the scattered X-rays 125 are collected by the X-ray detector 116 while a sample positioning system 140 positions and orients the sample 101 to generate angularly resolved scattered X-rays.

[0045] The measurement system 100 has a high dynamic range (e.g., 10 5The full-beam x-ray scatterometry technique includes one or more photon-counting detectors with a photon-counting angle (e.g., greater than 100 nm) and a thick, highly absorbing crystalline substrate that absorbs the straight-through beam (i.e., the zeroth order beam) without damage and with minimal parasitic backscatter. Exemplary detector materials suitable for full-beam x-ray scatterometry include cadmium telluride (CdTe), germanium (Ge), and gallium arsenide (GaAs) crystals, among others. In some embodiments, the detector material is selected to provide high conversion efficiency over a narrow energy band corresponding to the energy source.

[0046] In some embodiments, a single photon counting detector detects the location and number of detected photons. In some embodiments, a dual threshold detector is used to improve the SNR.

[0047] The X-ray detector 116 resolves one or more X-ray photon energies to generate a signal for each energy component of the X-ray that is indicative of a characteristic of the sample. In some embodiments, the X-ray detector 116 includes any of the following: a CCD array, a microchannel plate, a photodiode array, a microstrip proportional counter, a gas-filled proportional counter, a scintillator, or a phosphor.

[0048] In this way, X-ray photon interactions within the detector are identified by energy in addition to pixel location and count number. In some embodiments, X-ray photon interactions are identified by comparing the energy of the X-ray photon interactions to a predefined upper threshold and a predefined lower threshold. In one embodiment, this information is communicated via output signal 126 to computer system 130 for further processing and storage.

[0049] High aspect ratio vertically fabricated structures diffract a collimated x-ray beam into diffraction orders. Each diffraction order is translated in a specific, predictable direction. The angular spacing of the diffraction orders is inversely proportional to the lattice constant of the sample divided by the wavelength. The diffraction orders are detected by a detector array placed some distance from the wafer. Each pixel of the detector outputs a signal indicative of the number of photons that strike the pixel.

[0050] As shown in FIG. 1, the SAXS metrology system 100 includes a computing system 130 that is used to acquire a signal 126 generated by the detector 116 and determine a characteristic of a structure of interest based at least in part on the acquired signal. FIG. 2 illustrates an exemplary embodiment of a quality-controlled conditioned metrology engine 150 that estimates values ​​of one or more geometric parameters, compositional parameters, or both that characterize a complex semiconductor structure based on X-ray scatterometry measurement data associated with the complex semiconductor structure. In some embodiments, the X-ray scatterometry data 126 is collected from a wafer by an X-ray scatter metrology system, such as the metrology system 100 shown in FIG. 1. Furthermore, in some embodiments, the computing system 130 is configured as a quality-controlled conditioned metrology engine 150 as described herein.

[0051] FIG. 2 illustrates an exemplary quality control tuned measurement engine 150 in one embodiment. As shown in FIG. 2, the quality control tuned measurement engine 150 includes a trained quality control encoder module 152, a trained measurement model module 153, and, optionally, a trained layer encoder module 151. The quality control tuned measurement engine 150 estimates values ​​of target parameters 158 that characterize a structure to be measured, for example, a semiconductor structure to be measured by an X-ray based metrology system. In some embodiments, the target parameters include target geometric parameters, target process parameters, target electrical parameters, target dispersion parameters, etc. As non-limiting examples, the target parameters can be overlay dimensions, critical dimensions, lithography focus, lithography dose, etch exposure time, etc.

[0052] As shown in FIG. 2, the trained layer encoder module 151 receives a measurement signal S i IN-LINE Receive 155. S i IN-LINE 155 is a measurement signal associated with the measurement of an in-line wafer by the i-th metrology tool of the group of metrology tools. i IN-LINE 155 is a measurement signal related to the in-line wafer measurements by metrology system 100 shown in FIG. 1. The trained layer encoder module 151 also receives the in-line measurement signal, S i IN-LINE 155, one or more latent variables, V I IN-LINE It contains a trained layer encoder function that determines a set of 156 values. I IN-LINE156 is communicated to trained measurement model module 153. The latent variables are defined by the trained latent encoder function and are the same for all similar in-line wafers measured by the metrology tools in the fleet of metrology tools, e.g., all SAXS metrology tools in the fleet of metrology tools. In this way, the latent variables defined by the trained latent encoder function capture both system variations across the fleet of metrology tools and structural variations across the in-line wafers.

[0053] In general, the trained layer encoder module 151 is optional. The trained layer encoder module 151 is used to compress a measurement data set into a relatively small number of large variables. In general, however, the trained layer encoder module 151 may not be used and a trained measurement model may be used to compress the compressed signal, V I IN-LINE 156, but not the measured signal, S i IN-LINE It may also work directly on 155.

[0054] As shown in FIG. 2, the trained QC encoder module 152 receives a measurement signal, S i QC Receive 154. S i QC 154 is a measurement signal associated with the measurement of one or more QC wafers by the i-th metrology tool of the group of metrology tools. i QC 154 is a measurement signal associated with measurements of one or more QC wafers by metrology system 100 shown in FIG. 1. The trained QC encoder module 152 generates the QC measurement signal, S i QC One or more system variables, SV, associated with the i-th measurement tool from 154 I QC Contains a trained QC encoder function that determines a set of 157 values. SV I QC157 is communicated to the trained metrology model module 153. The system variables are defined by the trained QC encoder function and are the same for all similar metrology tools in the fleet of metrology tools, e.g., all SAXS metrology tools in the fleet of metrology tools. In this way, the system variables defined by the trained QC encoder function capture the system variation across the fleet of metrology tools, i.e., the system variables correspond to tool-to-tool variation.

[0055] One or more system variables, SV I QC The values ​​of 157 are specific to the ith metrology tool and represent the contribution of the ith metrology tool to the system variation across the fleet of metrology tools captured by each system variable. Periodically, the one or more QC wafers are measured again by the ith metrology tool, e.g., every 12 hours, every 24 hours, etc. The trained QC encoder function determines an updated set of values ​​for the one or more system variables and communicates the updated set of values ​​for the one or more system variables to the trained measurement model module 153. In this manner, the trained measurement model is adjusted with the values ​​of the one or more system variables that represent the metrology tool at its current conditions.

[0056] As shown in FIG. 2, the trained measurement model module 153 trains one or more system variables, SVs, associated with the i-th measurement system. I QC a set of 157 values ​​and one or more latent variables, V, associated with the in-line wafer measurements by the i-th metrology system I IN-LINE 156 as input. The trained measurement model of the trained measurement model module 153 receives as input both the set of values ​​of SV I QC 157 and V I IN-LINE156, values ​​of one or more parameters 158 (e.g., CD, SWA, HT, etc.) of the subject are determined. The values ​​of the parameters 158 of the subject are stored in a memory, e.g., memory 132. In some embodiments, the trained measurement model is a machine learning based measurement model (e.g., a linear model, a neural network model such as a multi-layer perceptron model, a convolutional network model, etc.).

[0057] In another embodiment, changes in the values ​​of one or more system variables are used to monitor the health of the tool. As shown in FIG. 2, one or more system variables, SV I QC The values ​​of 157 are stored in a memory, such as memory 132. Each time one or more QC wafers are measured by the ith metrology tool, one or more system variables, SV I QC 157 are stored in a memory, such as memory 132. A change in the value of the one or more system variables is indicative of the health of the tool, e.g., tool drift, tool component failure, etc. In some embodiments, a preventive maintenance action is initiated based on the change in the value of the one or more system variables. In one example, a preventive maintenance action is initiated if a difference in the values ​​of the one or more system variables associated with different measurement instances exceeds a predetermined threshold.

[0058] In another embodiment, the trained layer encoder functions are trained as an autoencoder as shown in Figure 3. Because the same measurement signals are provided as input and output of the autoencoder, information indicative of both structural and system variations is compressed into a small number of variables.

[0059] As shown in FIG. 3, the layer encoder training engine 160 includes a layer encoder training module 161. The layer encoder training module 161 receives an indication 162 of a desired dimension of the parameter space provided as output from the trained layer encoder. For example, the dimension of the parameter space may be an integer in the range of 1 to 100. In addition, the layer encoder training module 161 receives a measurement signal, S, associated with measurements of one or more nominally identical in-line wafers by N different metrology tools of a family of metrology tools. 1:N IN-LINE 163. The layer encoder training module 161 receives a set of measurement signals, S 1:N IN-LINE 163 to a set of values ​​for each parameter in a parameter space having a desired dimension, and 1:N IN-LINE The training is done by simultaneously training a decoder that maps a set of measurement signals, S 1:N IN-LINE 163 and the corresponding decoded measurement signal *S determined by the decoder from the parameters determined by the autoencoder. 1:N IN-LINE The trained encoder model 164 is stored in a memory, for example the memory 132.

[0060] In another embodiment, the trained QC encoder function is trained as an autoencoder as shown in FIG. 4. Because the same QC measurement signals are provided as input and output of the autoencoder, the information indicative of the system variations is compressed into a smaller number of variables. In addition, because the QC measurement signals are collected from multiple tools but from the same QC wafer, the system variables capture the tool-to-tool variations without the wafer variations. In this way, the trained QC encoder is an encoder of the system variables rather than the wafer variables.

[0061] As shown in FIG. 4, the QC encoder training engine 170 includes a QC encoder training module 171. The QC encoder training module 171 receives an indication 172 of a desired dimension of a system parameter space provided as output from a trained QC encoder. For example, the dimension of the parameter space may be an integer in the range of 1 to 100. In addition, the QC encoder training module 171 receives a measurement signal, S, associated with measurements of one or more QC wafers by each of N different metrology tools in a group of metrology tools. 1:N QC 173. The QC encoder training module 171 receives a set of measurement signals, S 1:N QC 173 to a set of values ​​for each parameter in a parameter space having a desired dimension, and 1:N QC The training is done by simultaneously training a decoder that maps a set of measurement signals, S 1:N QC 173 and the corresponding decoded measurement signal *S determined by the decoder from the parameters determined by the autoencoder. 1:N QC Iterate until the difference between is minimized. The trained QC encoder model 174 is stored in a memory, for example the memory 132.

[0062] The encoder function may be determined as an autoencoder, but in general any suitable method for training the encoders described herein, e.g., contrastive learning, principal component analysis, etc., is contemplated within the scope of this patent document.

[0063] In general, a trained QC encoder model is a function that maps measurement signals to specific values ​​of system variables. The trained QC encoder model receives measurement signals collected from the same QC wafer used to train the QC encoder function. As explained above, the trained QC encoder model receives measurement signals collected from the QC wafer by a specific metrology system and determines values ​​of the system variables associated with that specific metrology system at the time of measurement of the QC wafer. Furthermore, the QC wafer is repeatedly measured by the same metrology system at different time points and updated values ​​of the system variables are determined using the same trained QC encoder model. The updated values ​​of the system variables are provided as inputs to the trained metrology model associated with the same specific metrology system.

[0064] In another embodiment, a machine learning based QC-tuned metrology model is trained based on measurements of QC wafers and DOE wafers.

[0065] As shown in FIG. 5, the quality control tuned measurement model training engine 180 includes a trained QC encoder module 182, a QC tuned measurement model training module 183, and optionally, a trained layer encoder module 181.

[0066] As shown in FIG. 5, the trained layer encoder module 181 receives a measurement signal, S i DOE Receive 185. S i DOE 185 is a measurement signal associated with the measurement of a Design of Experiments (DOE) wafer by the ith metrology tool of the group of metrology tools. A DOE wafer is a wafer having known values ​​of one or more parameters of interest, such as CD, SWA, H, etc. In one example, S i DOE 185 is a measurement signal associated with measuring a DOE wafer with the metrology system 100 shown in FIG. 1. The trained layer encoder module 151 converts the DOE measurement signal, S i DOE185, one or more latent variables, V I DOE It contains a trained layer encoder function that determines a set of 187 values. I DOE 187 is communicated to the QC tuned metrology model training module 183. The latent variables are defined by the trained latent encoder function and are the same latent variables for all similar DOE and in-line wafers measured by metrology tools in the fleet of metrology tools, e.g., all SAXS metrology tools in the fleet of metrology tools. In this way, the latent variables defined by the trained latent encoder function capture both system variations across the fleet of metrology tools and structural variations across the DOE wafers.

[0067] In general, the trained layer encoder module 151 is optional. The trained encoder module 151 is used to compress a measurement data set into a relatively small number of large variables. In general, however, the trained layer encoder module 151 may not be used, and the QC-tuned measurement model training module 183 may compress the compressed signal, V I DOE 187, but not the measured signal, S i DOE It may also work directly against 185.

[0068] As shown in FIG. 5, the trained QC encoder module 182 receives a measurement signal, S i QC Receive 184. S i QC 184 is a measurement signal associated with the measurement of one or more QC wafers by the i-th metrology tool of the group of metrology tools. i QC 184 is a measurement signal associated with one or more QC wafer measurements by metrology system 100 shown in FIG. 1. Trained QC encoder module 182 generates the QC measurement signal, S i QC One or more system variables, SV, associated with the i-th measurement tool from 184 IQC Contains a trained QC encoder function that determines a set of 188 values. SV I QC 188 is communicated to the QC tuned metrology model training module 183. The system variables are defined by the trained QC encoder function and are the same for all similar metrology tools in the fleet of metrology tools, e.g., all SAXS metrology tools in the fleet of metrology tools. In this way, the system variables defined by the trained QC encoder function capture the system variation across the fleet of metrology tools, i.e., the system variables correspond to the tool-to-tool variation. To this end, one or more system variables, SV I QC The set of 188 values ​​captures the contribution of the i-th metrology tool to the variation across the set of metrology tools for each system variable.

[0069] As shown in FIG. 5, the QC-tuned measurement model training module 183 trains one or more system variables, SVs, associated with the i-th measurement system. I QC A set of 188 values ​​and one or more latent variables, V, associated with measurements of one or more DOE wafers by the i-th metrology system. I DOE In addition, the QC-tuned metrology model training module 183 receives both the set of values ​​of one or more parameters of interest, POIs, that characterize the measured structures on the one or more DOE wafers. DOE 186 known values ​​are received.

[0070] The QC-tuned measurement model training module 183 trains the latent variables V I DOE 187 and the corresponding values ​​of one or more system variables associated with the i-th measurement system, SV I QC 188 and based on known values ​​POI DOE186. During training, the values ​​of the system variables are provided as known inputs to the model training process and the trained model, so that the QC-tuned metrology model is optimized to be insensitive to the system variations associated with the i-th metrology tool. In this sense, the system variations associated with the i-th metrology tool are fed forward to the trained metrology model, effectively decoupling it from the wafer variations during model training and subsequent application of the trained metrology model. The trained QC-tuned metrology model 189 is stored in a memory, such as memory 132. In some embodiments, the trained metrology model is a machine learning based metrology model (e.g., a linear model, a neural network model such as a multi-layer perceptron model, a convolutional network model, etc.).

[0071] The QC-tuned measurement model 189 is trained by a regression process (e.g., ordinary least squares regression). The machine learning-based measurement model is parameterized by several weight parameters. The target parameters, POI DOE 186 known reference values ​​and the measurement signal, S i DOE The latent variable corresponding to 185, V I DOE Based on the value of 187, the values ​​of the weighting parameters are iteratively adjusted to minimize the difference between the values ​​of the target parameters estimated by the QC-tuned measurement model.

[0072] Typically, the QC-adjusted metrology models 189 are trained specifically for measurements made by a particular type of metrology tool measuring a particular layer of an in-line wafer, such that a different QC-adjusted metrology model 189 is trained for each layer of an in-line wafer being measured, but the same QC-adjusted metrology model is used by all metrology tools of the same type (e.g., SAXS, spectroscopic ellipsometry (SE), etc.) that measure the same layer of an in-line wafer.

[0073] In another embodiment, as described herein, the metrology tools comprising a group of matched metrology tools can include metrology tools of various types. For example, a group of five metrology tools can include three spectroscopic ellipsometry (SE) metrology tools and two SAXS metrology tools. In this example, the same QC wafer or set of QC wafers is measured by all five tools in the group of metrology tools, but the QC-adjusted measurement models used by all the SE tools are trained separately from the QC-adjusted measurement models used by all the SAXS tools.

[0074] In another further embodiment, multiple metrology tools (e.g., measurement subsystems) are used to perform a single hybrid measurement. In one example, the measurement signals associated with each metrology tool are concatenated and provided as input to a trained encoder and measurement model, as described above. Similarly, training of the QC-tuned measurement model used to perform the hybrid measurement is also performed using concatenated measurement signals from the multiple tools.

[0075] Typically, as described herein, QC-adjusted measurements may be performed by multiple metrology tools of the same type, multiple types of metrology tools, or any combination thereof. By way of non-limiting example, types of metrology tools contemplated herein include spectroscopic ellipsometers, spectroscopic reflectometers, soft x-ray reflectometers, small angle x-ray scatterometers, imaging systems, hyperspectral imaging systems, scatterometry overlay metrology systems, and the like.

[0076] As shown in FIG. 1, system 100 includes a single measurement technique (i.e., SAXS). In general, however, system 100 can include any number of different measurement techniques. As non-limiting examples, system 100 can be configured as a reflective small angle X-ray scatterometer, a soft X-ray reflectometer, a spectroscopic ellipsometer (including Mueller matrix ellipsometry), a spectroscopic reflectometer, a spectroscopic scatterometer, an overlay scatterometer, an angle-resolved beam profile reflectometer, a polarization-resolved beam profile reflectometer, a beam profile reflectometer, a beam profile ellipsometer, any single or multiple wavelength ellipsometer, a hyperspectral imaging system, or any combination thereof. Furthermore, in general, the measurement data collected by the different measurement techniques and analyzed according to the methods described herein can be collected from multiple tools, a single tool integrating multiple techniques, or a combination thereof.

[0077] In further embodiments, the system 100 can include one or more computing systems 130 that are used to perform measurements of the structures and estimate values ​​of the parameters of interest according to methods described herein. The one or more computing systems 130 can be communicatively coupled to the detector 116. In one aspect, the one or more computing systems 130 are configured to receive measurement data 126 associated with measurements of the structures to be measured (e.g., structures disposed on the sample 101).

[0078] In yet another aspect, the measurements described herein can be used to provide active feedback to a process tool (e.g., a lithography tool, an etch tool, a deposition tool, etc.). For example, values ​​of the measured parameters determined based on the measurement methods described herein can be communicated to an etch tool to adjust the etch time to achieve a desired etch depth. Similarly, an etch parameter (e.g., etch time, diffusivity, etc.) or a deposition parameter (e.g., time, concentration, etc.) can be included in a measurement model to provide active feedback to an etch tool or a deposition tool, respectively. In some examples, modifications of the process parameters determined based on the measured device parameter values ​​can be communicated to the process tool. In one embodiment, the computing system 130 determines values ​​of one or more parameters of interest. Furthermore, the computing system 130 communicates a control command to a process controller based on the determined values ​​of the one or more parameters of interest. The control command causes the process controller to change a state of the process (e.g., stop the etch process, change the diffusivity, etc.). In one example, the control command causes the process controller to adjust the focus of the lithography system, the exposure dose of the lithography system, or both. In another example, the control command may cause the process controller to change the etch rate to improve the measured wafer uniformity of the CD parameter.

[0079] In some examples, the metrology model is implemented as an element of a SpectraShape® optical critical dimension metrology system available from KLA-Tencor Corporation, Milpitas, California, USA In this manner, the model is created and ready for use immediately after a spectrum is collected by the system.

[0080] In some other examples, the measurement model is performed offline, for example by a computing system implementing AcuShape® software available from KLA-Tencor Corporation, Milpitas, California, USA. The resulting trained model can be incorporated as an element of an AcuShape® library that is accessible by the metrology system that performs the measurements.

[0081] 6 illustrates a method 200 for estimating values ​​of one or more parameters of interest using a trained QC-tuned measurement model in at least one novel embodiment. The method 200 is suitable for implementation by a metrology system, such as the metrology system 100 illustrated in FIG. 1 of the present invention. It is recognized that in one embodiment, the data processing blocks of the method 200 can be performed by a pre-programmed algorithm executed by one or more processors of the computing system 130, or by any other general-purpose computing system. It is recognized that the specific structural aspects of the metrology system 100 herein should be construed as illustrative only, and not limiting.

[0082] In block 201, a first set of measurement signals indicative of measurements of one or more quality control (QC) structures disposed on one or more QC semiconductor wafers is generated by a first semiconductor metrology system of a group of semiconductor metrology systems.

[0083] At block 202, a second set of measurement signals indicative of measurements of one or more structures disposed on a first in-line semiconductor wafer by a first semiconductor metrology system is generated.

[0084] In block 203, a first set of values ​​of one or more system variables associated with the first semiconductor measurement system is determined from the first set of measurement signals using the trained QC encoder function.

[0085] In block 204, values ​​of one or more parameters of interest characterizing a structure disposed on a first in-line semiconductor wafer are estimated based on the first set of values ​​of the one or more system variables and the second set of measurement signals using the trained QC-tuned measurement model.

[0086] In a further embodiment, the system 100 includes one or more computing systems 130 that are used to perform measurements on the semiconductor structures based on the measurement data according to the methods described herein. The one or more computing systems 130 can be communicatively coupled to one or more detectors, active optical elements, process controllers, etc.

[0087] It should be appreciated that one or more steps described throughout this disclosure may be performed by a single computer system 130 or, alternatively, multiple computer systems 130. Additionally, different subsystems of system 100 may include computer systems suitable for performing at least some of the steps described herein. Thus, the foregoing description should not be construed as a limitation on the present invention, but merely as an example.

[0088] Additionally, computer system 130 may be communicatively coupled to other elements of the metrology system in any manner known in the art. For example, one or more of computing systems 130 may be coupled to a computing system associated with a detector. In another example, the detector may be directly controlled by a single computer system coupled to computer system 130.

[0089] The computer system 130 of the system 100 may be configured to receive and / or acquire data or information from subsystems of the system (e.g., detectors, etc.) by way of a transmission medium that may include wired and / or wireless portions. In this manner, the transmission medium may serve as a data link between the computer system 130 and other subsystems of the system 100.

[0090] The computer system 130 of the system 100 may be configured to receive and / or acquire data or information (e.g., measurement results, modeling inputs, modeling results, reference measurement results, etc.) from other systems by a transmission medium, which may include wired and / or wireless portions. In this manner, the transmission medium may serve as a data link between the computer system 130 and the other system (e.g., the memory-equipped system 100, an external memory, or another external system). For example, the computing system 130 may be configured to receive measurement data from a storage medium (i.e., the memory 132 or an external memory) via a data link. For example, the measurement results obtained using the detectors described herein may be stored in a permanent or semi-permanent memory device (e.g., the memory 132 or an external memory). In this regard, the measurement results may be imported from an on-board memory or an external memory system. Additionally, the computer system 130 may send data to other systems via the transmission medium. For example, the measurement models or estimated parameter values ​​determined by the computer system 130 may be communicated and stored in an external memory. In this regard, the measurement results may be exported to another system.

[0091] Computing system 130 may include, but is not limited to, a personal computer system, a mainframe computer system, a workstation, a graphics computer, a parallel processor, or any other device known in the art. In general, the term "computing system" may be broadly defined to encompass any device with one or more processors that execute instructions from a memory medium.

[0092] Program instructions 134 implementing methods such as those described herein may be transmitted over a transmission medium such as a wire, cable, or wireless transmission link. For example, as shown in Figure 1, program instructions 134 stored in memory 132 are transmitted to processor 131 via bus 133. Program instructions 134 are stored in a computer-readable medium (e.g., memory 132). Exemplary computer-readable media include read-only memory, random-access memory, a magnetic or optical disk, or a magnetic tape.

[0093] As described herein, the term "critical dimension" includes any critical dimension of a structure (e.g., bottom critical dimension, mid critical dimension, top critical dimension, sidewall angle, grating height, etc.), the critical dimension between any two or more structures (e.g., the distance between two structures), and the misregistration between two or more structures (e.g., the overlay misregistration between overlaid grating structures, etc.). The structures can include three-dimensional structures, patterned structures, overlay structures, etc.

[0094] As described herein, the terms "critical dimension application" or "critical dimension measurement application" include any critical dimension measurement.

[0095] As described herein, the term "metrology system" includes any system used at least in part to characterize a specimen in any aspect including measurement applications such as critical dimension metrology, overlay metrology, focus / exposure metrology, and composition metrology. However, such terms for technology do not limit the scope of the term "metrology system" as described herein. In addition, the system 100 may be configured for measurements of patterned and / or unpatterned wafers. The metrology system may be configured as an LED inspection tool, an edge inspection tool, a backside inspection tool, a macro inspection tool, or a multi-mode inspection tool (including data from one or more platforms simultaneously), and any other metrology or inspection tool that would benefit from the techniques described herein.

[0096] Various embodiments are described herein for a semiconductor metrology system, which may be used to measure a specimen within any semiconductor processing tool (e.g., an inspection system or a lithography system). The term "specimen" is used herein to refer to a wafer, a reticle, or any other sample that may be processed (e.g., printed or inspected for defects) by means known in the art.

[0097] As used herein, the term "wafer" generally refers to a substrate formed from a semiconductor or non-semiconductor material. Examples include, but are not limited to, monocrystalline silicon, gallium arsenide, and indium phosphide. Such substrates may be commonly found and / or processed within semiconductor manufacturing facilities. In some cases, a wafer may include only a substrate (i.e., a bare wafer). Alternatively, a wafer may include one or more layers of different materials formed on a substrate. One or more layers formed on a wafer may be "patterned" or "unpatterned." For example, a wafer may include multiple dies having repeating pattern features.

[0098] A "reticle" may be a reticle at any stage in a reticle manufacturing process or a finished reticle that may or may not be released for use in a semiconductor manufacturing facility. A reticle or "mask" is generally defined as a substantially transparent substrate having substantially opaque areas formed thereon and arranged in a pattern. The substrate may comprise, for example, a glass material such as amorphous SiO2. The reticle may be disposed on a resist-coated wafer during an exposure step of a lithography process, thereby transferring the pattern on the reticle to the resist.

[0099] One or more layers formed on a wafer may be patterned or unpatterned. For example, a wafer may include multiple dies, each having repeating pattern features. Formation and processing of such layers of material may result in a completed device. Many different types of devices may be formed on a wafer, and the term "wafer" as used herein is intended to include a wafer having any type of device fabricated thereon known in the art.

[0100] In one or more exemplary embodiments, the functions described may be implemented in hardware, software, firmware, or any combination thereof. If the functions are implemented in software, the functions may be stored on or transmitted as one or more instructions or code on a computer-readable medium. Computer-readable media includes both computer storage media and communication media, including any medium that facilitates transfer of a computer program from one place to another. A storage medium may be any available medium that may be accessed by a general purpose or special purpose computer. By way of example, and not limitation, such computer-readable media may comprise RAM, ROM, EEPROM, CD-ROM or other optical disk storage, magnetic disk storage or other magnetic storage, or any other medium that may be used to convey or store desired program code means in the form of instructions or data structures and may be accessed by a general purpose or special purpose computer, or a general purpose or special purpose processor. Also, any connection is properly referred to as a computer-readable medium. For example, if the software is transmitted from a website, server or another remote source using coaxial cable, fiber optic cable, twisted pair, digital subscriber line (DSL), or wireless technologies such as infrared, radio and microwave, the coaxial cable, fiber optic cable, twisted pair, DSL, or wireless technologies such as infrared, radio and microwave are included within the definition of media. As used herein, disk and disc include compact discs (CDs), laser discs, optical discs, digital versatile discs (DVDs), floppy disks, and Blu-ray discs, where a disk typically reproduces data magnetically, while a disc reproduces data optically with a laser. Combinations of the above should also be included within the scope of computer-readable media.

[0101] Although certain specific embodiments are described above for instructional purposes, the teachings of this patent document have general applicability and are not limited to the specific embodiments described above. Thus, various modifications, adaptations and combinations of the various features of the described embodiments may be made without departing from the scope of the invention as set forth in the claims.

Claims

1. 1. A method comprising: irradiating one or more quality control (QC) structures disposed on one or more QC semiconductor wafers with a first amount of illumination radiation generated by an illumination source of a first semiconductor metrology system of the group of semiconductor metrology systems, the first amount of illumination radiation being characterized by one or more physical properties; detecting a first amount of collected radiation from the one or more QC structures in response to the first amount of illumination radiation at a detector, the first amount of collected radiation depending on the one or more physical properties of the first amount of illumination radiation and one or more physical properties of the one or more QC structures; irradiating one or more structures disposed on one or more in-line semiconductor wafers with a second amount of illumination radiation generated by the illumination source of the first semiconductor metrology system of the group of semiconductor metrology systems, the second amount of illumination radiation characterized by the one or more physical properties; detecting a second amount of collected radiation from the one or more structures in response to the second amount of illumination radiation at the detector, the second amount of collected radiation depending on the one or more physical properties of the second amount of illumination radiation and one or more physical properties of the one or more structures; generating a first set of measurement signals indicative of the first amount of collected radiation detected; generating a second set of measurement signals indicative of the second amount of collected radiation detected; determining a first set of values ​​of one or more system variables associated with the first semiconductor measurement system from the first set of measurement signals using a trained QC encoder function; using the trained QC-tuned measurement model to estimate values ​​of one or more parameters of interest characterizing a structure disposed on a first in-line semiconductor wafer based on the first set of values ​​of the one or more system variables and the second set of measurement signals; A method comprising:

2. 10. The method of claim 1, further comprising: generating a third set of measurement signals indicative of measurements of one or more Design of Experiment (DOE) structures disposed on one or more DOE semiconductor wafers by the first semiconductor measurement system of the group of semiconductor measurement systems, wherein values ​​of one or more parameters of interest characterizing the one or more DOE structures disposed on the one or more DOE semiconductor wafers are known; training the QC-tuned measurement model based on the first set of values ​​of the one or more system variables and the third set of measurement signals; The method further comprising:

3. 3. The method of claim 2, wherein the trained QC-tuned model is a machine learning-based model.

4. 10. The method of claim 1, generating a third set of measurement signals indicative of measurements of the one or more quality control (QC) structures disposed on the one or more QC semiconductor wafers by each semiconductor metrology system of the group of semiconductor metrology systems; training the QC encoder function based on the third set of measurement signals; The method further comprising:

5. 5. The method of claim 4, wherein the training of the QC encoder function involves any of an autoencoder, a principal component analysis, and contrastive learning.

6. 10. The method of claim 1, generating, by the first semiconductor metrology system of the group of semiconductor metrology systems, a third set of measurement signals indicative of measurements of the one or more quality control (QC) structures disposed on the one or more QC semiconductor wafers at a time subsequent to the generating of the first set of measurement signals; determining a second set of values ​​of the one or more system variables associated with the first semiconductor measurement system from the third set of measurement signals using the trained QC encoder function; The method further comprising:

7. 7. The method of claim 6, generating a fourth set of measurement signals indicative of measurements of one or more structures disposed on a second in-line semiconductor wafer by the first semiconductor metrology system; using the trained QC-tuned measurement model to estimate values ​​of one or more parameters of interest characterizing the structure disposed on the second in-line semiconductor wafer based on the second set of values ​​of the one or more system variables and the fourth set of measurement signals; The method further comprising:

8. 7. The method of claim 6, estimating a difference between the first set of values ​​of the one or more system variables and the second set of values ​​of one or more of the system variables.

9. 9. The method of claim 8, The method, further comprising taking a preventive maintenance action if the difference exceeds a predetermined threshold.

10. 10. The method of claim 1, wherein each of the semiconductor metrology systems in the group of semiconductor metrology systems is any of a spectroscopic ellipsometer, a spectroscopic reflectometer, a soft x-ray reflectometer, a small angle x-ray scatterometer, an imaging system, a hyperspectral imaging system, and a scatterometry overlay metrology system.

11. 10. The method of claim 1, the first set of measurement signals includes measurement data associated with measurements of the one or more QC structures by a plurality of semiconductor measurement systems of the group of semiconductor measurement systems, and the second set of measurement signals includes measurement data associated with measurements of the one or more structures by the plurality of semiconductor measurement systems of the group of semiconductor measurement systems.

12. 1. A system comprising: an illumination source configured to generate a first amount of illumination radiation characterized by one or more physical properties and to irradiate one or more quality control (QC) structures disposed on one or more QC semiconductor wafers with the first amount of illumination radiation, and further configured to generate a second amount of illumination radiation characterized by the one or more physical properties and to irradiate one or more structures disposed on one or more in-line semiconductor wafers; a first detector configured to detect a first amount of collected radiation from the one or more QC structures in response to the first amount of illumination radiation, the first amount of collected radiation being responsive to the one or more physical properties of the first amount of illumination radiation and one or more physical properties of the one or more QC structures, and further configured to generate a first set of measurement signals indicative of the detected first amount of collected radiation; and further configured to detect a second amount of collected radiation from the one or more structures in response to the second amount of illumination radiation, the second amount being responsive to the one or more physical properties of the second amount of illumination radiation and one or more physical properties of the one or more structures, and further configured to generate a second set of measurement signals indicative of the detected second amount of collected radiation; a first semiconductor measurement system of a group of semiconductor measurement systems comprising: one or more computer systems, determining a first set of values ​​of one or more system variables associated with the first semiconductor measurement system from the first set of measurement signals using a trained QC encoder function; using a trained QC-tuned measurement model to estimate values ​​of one or more parameters of interest characterizing the structure disposed on the first in-line semiconductor wafer based on the first set of values ​​of the one or more system variables and the second set of measurement signals; one or more computer systems configured to: A system comprising:

13. 13. The system of claim 12, wherein the one or more computer systems: receiving a third set of measurement signals indicative of measurements of one or more Design of Experiment (DOE) structures disposed on one or more DOE semiconductor wafers by the first semiconductor measurement system of the group of semiconductor measurement systems, wherein values ​​of one or more parameters of interest characterizing the one or more DOE structures disposed on the one or more DOE semiconductor wafers are known; training the QC-tuned measurement model based on the first set of values ​​of the one or more system variables and the third set of measurement signals; The system, further configured to:

14. 13. The system of claim 12, wherein the one or more computer systems: generating a third set of measurement signals indicative of measurements of the one or more quality control (QC) structures disposed on the one or more QC semiconductor wafers by each semiconductor metrology system of the group of semiconductor metrology systems; training the QC encoder function based on the third set of measurement signals; The system, further configured to:

15. 13. The system of claim 12, wherein the one or more computer systems: receiving, at a time subsequent to the generating of the first set of measurement signals, by the first semiconductor measurement system of the group of semiconductor measurement systems, a third set of measurement signals indicative of measurements of the one or more quality control (QC) structures disposed on the one or more QC semiconductor wafers; determining a second set of values ​​of the one or more system variables associated with the first semiconductor measurement system from the third set of measurement signals using the trained QC encoder function; The system, further configured to:

16. 16. The system of claim 15, wherein the one or more computer systems: receiving a fourth set of measurement signals indicative of measurements of one or more structures disposed on a second in-line semiconductor wafer by the first semiconductor metrology system; using the trained QC-tuned measurement model to estimate values ​​of one or more parameters of interest characterizing the structure disposed on the second in-line semiconductor wafer based on the second set of values ​​of the one or more system variables and the fourth set of measurement signals; The system, further configured to:

17. 16. The system of claim 15, wherein the one or more computer systems:

10. The system, further configured to estimate a difference between the first set of values ​​of the one or more system variables and the second set of values ​​of one or more of the system variables.

18. 18. The system of claim 17, wherein if the difference exceeds a predetermined threshold, a preventative maintenance action is taken.

19. 13. The system of claim 12, wherein each of the semiconductor metrology systems in the group of semiconductor metrology systems is any of a spectroscopic ellipsometer, a spectroscopic reflectometer, a soft x-ray reflectometer, a small angle x-ray scatterometer, an imaging system, a hyperspectral imaging system, and a scatterometry overlay metrology system.

20. 1. A system comprising: an illumination source configured to generate a first amount of illumination radiation characterized by one or more physical properties and to irradiate one or more quality control (QC) structures disposed on one or more QC semiconductor wafers with the first amount of illumination radiation, and further configured to generate a second amount of illumination radiation characterized by the one or more physical properties and to irradiate one or more structures disposed on one or more in-line semiconductor wafers; a first detector configured to detect a first amount of collected radiation from the one or more QC structures in response to the first amount of illumination radiation, the first amount of collected radiation being responsive to the one or more physical properties of the first amount of illumination radiation and one or more physical properties of the one or more QC structures, and further configured to generate a first set of measurement signals indicative of the detected first amount of collected radiation; and further configured to detect a second amount of collected radiation from the one or more structures in response to the second amount of illumination radiation, the second amount being responsive to the one or more physical properties of the second amount of illumination radiation and one or more physical properties of the one or more structures, and further configured to generate a second set of measurement signals indicative of the detected second amount of collected radiation; a first semiconductor measurement system of a group of semiconductor measurement systems comprising: A non-transitory computer-readable medium storing computer-readable instructions that, when executed by one or more processors, cause a computing system to: determining a first set of values ​​of one or more system variables associated with the first semiconductor measurement system from the first set of measurement signals using a trained QC encoder function; using a trained QC-tuned measurement model to estimate values ​​of one or more parameters of interest characterizing the structure disposed on the first in-line semiconductor wafer based on the first set of values ​​of the one or more system variables and the second set of measurement signals; a non-transitory computer-readable medium for causing A system comprising:

Images