Measurement of semiconductor structures based on spectral differences at different process steps

Abstract

Presented herein are methods and systems for measuring the value of one or more parameters of interest, including changes in the value of one or more parameters of interest, based on measured spectral differences. A trained spectral difference-based measurement model determines changes in the value of one or more parameters of interest based on spectral difference measurements measured before and after one or more process steps. In some examples, the spectral difference measurements are determined based on differences in measured intensity, differences in harmonic signal values, or differences in the values ​​of one or more Mueller matrix elements. The spectral difference measurements can be expressed as difference values, scalar values, or a set of coefficients of a function that fits the difference values. The spectral difference measurements can be determined based on weighting the spectral differences according to wavelength.

Description

[Technical Field]

[0001] This patent application claims priority under U.S. Provisional Patent Application No. 63 / 429,553, filed December 2, 2022, entitled "General Machine Learning Solution Based on Spectra Difference from Different Process Step Applied to Gate-all-around (GAA) Nanosheet, DRAM, 3D-Flash and 3D-DRAM Metrology," the subject matter of which is incorporated herein by reference in its entirety.

[0002] The described embodiments relate to metrology systems and methods, and more particularly to methods and systems for improved measurement of parameters that characterize semiconductor structures. [Background technology]

[0003] Generally, semiconductor devices, such as logic and memory devices, are fabricated by a series of process steps applied to a specimen. These process steps form the various features and structural levels of the semiconductor device. For example, lithography, among other semiconductor manufacturing processes, involves creating patterns on a semiconductor wafer. Further non-limiting examples of semiconductor manufacturing processes include chemical-mechanical polishing, etching, deposition, and ion implantation. Multiple semiconductor devices may be fabricated on a single semiconductor wafer, which may then be separated into individual semiconductor devices.

[0004] Metrology processes are used at various steps during the semiconductor manufacturing process to detect defects on wafers to improve yield. Optical metrology techniques enable higher throughput measurements without the risk of sample damage. Numerous optical 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, and other parameters of nanoscale structures.

[0005] A common approach to semiconductor manufacturing process control is to use metrology tools to measure structures fabricated on wafers after one or more critical process steps in the manufacturing process. Critical steps are typically performed more intensively, resulting in improved process control. However, process control in high-throughput manufacturing environments relies on adequate measurement sensitivity and computationally efficient analytical procedures at every process step.

[0006] In general, the semiconductor industry is struggling to produce smaller devices with increasingly complex structures and a growing variety of materials. Examples of devices exhibiting such complexity include gate-all-around (GAA) field-effect transistors (FETs), current dynamic random access memory (DRAM) structures, and current three-dimensional flash memory structures.

[0007] In one example, GaAs FETs fabricated using nanosheet fabrication techniques offer improved device performance and power savings. However, their nanoscale size and complex geometry make them challenging to fabricate. Nanosheet structures contain multiple material layers. The process for fabricating nanosheet structures begins with growing a superlattice of silicon and silicon germanium layers. These layers comprise the nanosheet base structure. Measuring the characteristics of each layer, such as film thickness, is essential to maintain control of the fabrication process.

[0008] Spectroscopic ellipsometry (SE) is an established optical measurement technique used to measure the physical and optical parameters of thin films. SE systems illuminate the structure under test with polarized light. Interaction between the illuminating light and the structure under test changes the polarization of the light reflected from the structure under test. This change in polarization is sensitive to film thickness and material properties. [Prior art documents] [Patent documents]

[0009] [Patent Document 1] U.S. Patent Application Publication No. 2018 / 0100796 Summary of the Invention [Problem to be solved by the invention]

[0010] SE is an indirect method for measuring the physical properties of the analyte being measured. Generally, physics-based measurement models are used to measure the raw measurement signal (e.g., α meas and β meas ) to determine the physical properties of the specimen. Conventional SE measurements on nanosheet structures are complicated by the complexity of the measurement model and the correlation of measurement data from multiple thin superlattice layers. Furthermore, SE measurements on nanoscale structures such as nanosheet structures, which are next-generation semiconductor structures, suffer from low sensitivity.

[0011] For example, nanosheet structures contain multiple extremely thin layers, and it is important to measure the thickness and material property changes of each layer within the multilayer stack at various points in the semiconductor manufacturing process flow. To monitor the process, changes in the multilayer structure are calculated by comparing the thickness and material properties before and after one or more process steps. For these differences to be meaningful, the thickness and material properties must be accurately measured both before and after one or more process steps. Accurate measurements require signal sensitivity and complex high-order models to attempt to de-correlate measurement system parameters from structural parameters. These models become too complex for practical solutions.

[0012] Metrology applications involving nanoscale structures, such as nanosheet structures, are challenged by the complexity of measurement models and practical limitations on signal sensitivity. Miniaturizing resolution requirements, multi-parameter correlations, and increasingly complex topographic structures compound this problem, creating undesirable gaps in the control of advanced manufacturing processes. Therefore, methods and systems for improved process control for advanced manufacturing processes are needed. [Means for solving the problem]

[0013] Presented herein are methods and systems for measuring values ​​of one or more parameters of interest, including changes in values ​​of the one or more parameters of interest due to one or more interfering process steps, based on measured spectral differences. A trained spectral difference-based measurement model determines changes in values ​​of the one or more parameters of interest based on differences in spectra measured before and after one or more process steps. In this way, structural changes caused by one or more process steps are measured based on changes in the measured spectra, rather than changes in values ​​of the parameters of interest.

[0014] In some embodiments, a spectral difference module determines a spectral difference measurement by determining the difference in measured intensities at each distinct set of wavelengths.

[0015] In some embodiments, a spectral difference module determines a spectral difference measurement by determining the difference between measurements of the harmonic signals at each distinct wavelength pair.

[0016] In some embodiments, a spectral difference module determines a spectral difference measurement by determining a difference between measurements of one or more Mueller matrix elements at each distinct wavelength pair.

[0017] In some embodiments, the spectral difference module determines the spectral difference measurement by determining a scalar value that indicates the difference in the measurement of the spectral signal between all measured wavelengths.

[0018] In some embodiments, the spectral difference measure is a set of coefficients that characterize a mathematical function that fits the spectral difference at each distinct set of wavelengths.

[0019] In some embodiments, the spectral difference measurements are determined based on a weighting of the spectral differences between the individual wavelength pairs as a function of wavelength. In these embodiments, one or more spectral difference values ​​are weighted differently than other spectral difference values ​​between the individual measured wavelength pairs. In this manner, the spectral difference measurements, whether expressed as difference values, scalar values, or a set of coefficients of a function that fits the difference values, are weighted based on wavelength.

[0020] Measurements of spectral differences between adjacent measurement points in the manufacturing process flow are communicated to a trained spectral difference-based measurement module, and the trained spectral difference-based measurement model is utilized to estimate values ​​of one or more parameters of interest, including changes in the values ​​of the one or more parameters of interest due to interfering process steps, between adjacent measurement points based on the spectral differences.

[0021] In some embodiments, a trained spectral difference-based measurement model is trained to receive spectral difference measurements between two particular measurement steps and generate spectral difference-based estimates of one or more parameters of interest. In these embodiments, different trained spectral difference-based measurement models are utilized to estimate values ​​of the one or more parameters of interest based on spectral difference measurements associated with different pairs of measurement steps.

[0022] In some other embodiments, the trained spectral difference-based measurement model is trained to receive multiple spectral difference measurements, each associated with a different pair of measurement steps, and to estimate values ​​of one or more parameters of interest based on each spectral difference. In these embodiments, the trained spectral difference-based measurement model is used to estimate values ​​of one or more parameters of interest based on the spectral difference measurements associated with the different pair of measurement steps.

[0023] In another aspect, the trained spectral difference-based measurement model is utilized to simulate changes in the value of one or more parameters of interest, and the simulated changes are utilized to adjust process control variables to improve process performance. In some embodiments, expected spectral difference data is simulated, for example, using a process and metrology simulator, or generated based on historical usage, user experience, etc. The expected spectral difference data is provided as input to the trained spectral difference-based measurement model. The trained spectral difference-based measurement model estimates expected values ​​of one or more parameters of interest based on the expected spectral difference data. In this manner, the trained spectral difference-based measurement model not only monitors the production of multilayer film stacks in-line, but also predicts production trends. Furthermore, one or more process variables, such as focus, dose, etch time, etc., are adjusted based on the expected values ​​of the parameters of interest. In this manner, the process is adjusted to improve process performance.

[0024] In another aspect, the spectral difference-based measurement model is trained based on design-of-experiment (DOE) spectral signals associated with different measurement steps in a manufacturing process flow. In some examples, the spectral difference-based measurement model is trained based on simulated or actual measurement data associated with measuring design-of-experiment (DOE) wafers.

[0025] Because the foregoing is a summary and thus necessarily contains simplifications, generalizations, and omissions of detail, those skilled in the art will appreciate that the summary is illustrative only and is not intended to be in any way limiting. 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 explanation of the drawings]

[0026] [Figure 1] FIG. 1 illustrates a system 100 for measuring a characteristic of an analyte, according to exemplary methods presented herein. [Figure 2] FIG. 1 illustrates a spectral difference based measurement engine 150 in one embodiment. [Figure 3] FIG. 1 illustrates a manufacturing process flow 175 including a series of manufacturing process steps and metrology steps in one embodiment. [Figure 4] FIG. 10 illustrates a manufacturing process flow 180 including a series of manufacturing process steps and metrology steps in another embodiment. [Figure 5] FIG. 1 illustrates a spectral difference based measurement model learning engine 160 in one embodiment. [Figure 6] 3 is a flowchart illustrating a method 300 for determining one or more parameter values ​​characterizing a structure of interest based on measured spectral differences, in one example. DETAILED DESCRIPTION OF THE INVENTION

[0027] Reference will now be made in detail to exemplary background and certain embodiments of the present invention, examples of which are illustrated in the accompanying drawings.

[0028] Presented herein are methods and systems for measuring values ​​of one or more parameters of interest, including changes in values ​​of the one or more parameters of interest due to one or more interfering process steps, based on measured spectral differences. A trained spectral difference-based measurement model determines changes in values ​​of the one or more parameters of interest based on differences in spectra measured before and after one or more process steps. In this way, structural changes caused by one or more process steps are measured based on changes in the measured spectra, rather than changes in estimated values ​​of the parameters of interest.

[0029] Generally, the spectral signal measured at any particular process step is sensitive to fundamental geometric and material characteristics of the structure being measured as well as external factors such as the current environment, sample state, etc. The size of the measurement model required to extract the value of the parameter of interest from the spectral signal at any particular process step is relatively large because the model must take into account the external factors.

[0030] On the other hand, the spectral difference signal significantly reduces sensitivity to external factors and highlights changes in the measured structure due to interfering process steps. The spectral difference signal contains the information needed to isolate and measure changes in key features with greater sensitivity than would otherwise be possible based on the underlying measurement signal. Therefore, a measurement model based on spectral difference more accurately predicts changes in the measured structure. Furthermore, the size of a measurement model operating on the spectral difference signal is significantly smaller than a measurement model operating directly on the spectral signal. This is because the spectral difference signal is more sensitive to changes in the measured structure due to interfering process steps. Therefore, the size of the measurement data required to train the spectral difference measurement model is also smaller than that of a traditional measurement model. Thus, the spectral difference-based measurement model increases computational efficiency and enables high-throughput metrology suitable for in-line measurements within high-volume semiconductor manufacturing process flows.

[0031] FIG. 1 illustrates a system 100 for measuring characteristics of a specimen according to an exemplary method presented herein. As illustrated in FIG. 1, the system 100 can be used to perform spectroscopic ellipsometry measurements on one or more structures on the specimen 101. In this embodiment, the system 100 can include a spectroscopic ellipsometer equipped with an illuminator 102 and a spectrometer 104. The illuminator 102 of the system 100 is configured to generate and direct illumination in a selected wavelength range (e.g., 150 to 2500 nm) onto structures disposed on the surface of the specimen 101. Meanwhile, the spectrometer 104 is configured to receive illumination reflected from the surface of the specimen 101. Note further that the light generated by the illuminator 102 is polarized using a polarization state generator 107, thereby generating a polarized illumination beam 106. The illumination light reflected by the structures disposed on the specimen 101 passes through a polarization state analyzer 109 toward the spectrometer 104. The illumination received by the spectrometer 104 in the collected beam 108 is analyzed for polarization state, allowing the spectrometer to perform spectral analysis on the illumination passing through the analyzer. One or more computing systems 130 are communicatively coupled to the spectrometer 104. The measured spectrum 111 is communicated to the computing system 130 for structural analysis.

[0032] As shown in FIG. 1 , system 100 includes a single measurement technique (i.e., SE). However, in general, system 100 can include any number of different measurement techniques. By way of non-limiting example, system 100 can be configured as 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, or any combination thereof. Furthermore, in general, measurement data collected by different measurement techniques and analyzed according to the methods described herein can be collected from multiple tools rather than a single tool integrating multiple techniques.

[0033] In some embodiments, the computing system 130 is configured as a spectral difference based measurement engine 150. As shown in Figure 2, the spectral difference based measurement engine 150 includes a spectral difference module 151 and a trained spectral difference based measurement module 152. 1…M S MEAS 153 are collected from M different measurement points in the manufacturing process flow by a measurement system, for example, metrology system 100. Each different measurement point is separated from an adjacent measurement point by one or more manufacturing process steps in the manufacturing process flow.

[0034] The measured spectrum may include any suitable indicia of the spectral response of the structure being measured. By way of non-limiting example, the measured spectral signal may include conventional formulations such as harmonic signals (Ψ,Δ), (α,β), etc., one or more Mueller matrix elements, an intensity profile, or any other suitable representation of the spectral response of the structure being measured.

[0035] As shown in Figure 2, the measured spectrum 1…M S MEAS 153 is communicated to the spectral difference module 151. The spectral difference module calculates a measurement value indicative of the spectral difference between any two measurement points based on the measured spectra at the two measurement points. In some examples, the spectral difference module 151 calculates a measurement value indicative of the spectral difference between adjacent measurement points in the process flow.

[0036] 3 shows a manufacturing process flow 175 that includes a series of manufacturing process steps, including a lithography step 196, an etching step 197, another lithography step 198, and another etching step 199. Each process step physically transforms the structure to be measured. At each metrology step, metrology tool 100 is utilized to measure the spectral response of the measurement object on the wafer in its current physical state and communicate the measured spectral signal to spectral difference module 151.

[0037] An input wafer at a particular physical state in the manufacturing process is measured by a metrology tool, such as metrology tool 100, in metrology step 181 before lithography step 196, metrology step 182 before etch step 197, metrology step 183 before lithography step 198, metrology step 184 before etch step 199, and metrology step 185 after etch step 199.

[0038] Based on measured spectral signals 186 and 187 associated with measurement steps 181 and 182, respectively, spectral difference module 151 determines a measure of spectral difference 191. Based on measured spectral signals 187 and 188 associated with measurement steps 182 and 183, respectively, spectral difference module 151 determines a measure of spectral difference 192. Based on measured spectral signals 188 and 189 associated with measurement steps 183 and 184, respectively, spectral difference module 151 determines a measure of spectral difference 193. Based on measured spectral signals 189 and 190 associated with measurement steps 184 and 185, respectively, spectral difference module 151 determines a measure of spectral difference 194.

[0039] In some embodiments, the spectral difference module 151 determines the spectral difference measure by determining the difference in measured intensities at each of the individual wavelength pairs. In these examples, the spectral difference measure is an intensity difference value set. In some embodiments, the spectral difference module 151 determines the spectral difference measure by determining the difference in measurements of harmonic signals, such as (Ψ,Δ) values ​​or (α,β) values, at each of the individual wavelength pairs. In these examples, the spectral difference measure is a harmonic difference value set. In some embodiments, the spectral difference module 151 determines the spectral difference measure by determining the difference in measurements of one or more Mueller matrix elements at each of the individual wavelength pairs. In these examples, the spectral difference measure is a Mueller matrix difference value set.

[0040] In some embodiments, the spectral difference module 151 determines the spectral difference measure by determining a scalar value indicative of the difference in the measurement of the spectral signal between all measured wavelengths. In some examples, a root mean square error (RSME) measure of the spectral signal difference between all measured wavelengths is utilized. In general, any suitable scalar metric indicative of the difference between sets of measured signals is contemplated within the scope of this patent document.

[0041] In some embodiments, the spectral difference measurements are a set of coefficients that characterize a mathematical function that fits the spectral differences at each distinct wavelength set. In some embodiments, the spectral difference module 151 fits a mathematical function, such as a polynomial function, to the set of spectral differences determined at each distinct wavelength set. In these embodiments, the coefficients of the mathematical function are a measure of the spectral difference between the two sets of measured spectral signals under consideration.

[0042] In some embodiments, the spectral difference measurements are determined based on a weighting of the spectral differences between the individual wavelength pairs as a function of wavelength. In these embodiments, one or more spectral difference values ​​are weighted differently than other spectral difference values ​​between the individual measured wavelength pairs. In this manner, the spectral difference measurements, whether expressed as difference values, scalar values, or a set of coefficients of a function that fits the difference values, are weighted based on wavelength. In some embodiments, the shorter the wavelength, the greater the weighting, or vice versa.

[0043] Measurement of the spectral difference between adjacent measurement points in the manufacturing process flow, as shown in Figure 2 1…M-1 S-DIFF MEAS 154 is communicated to the learned spectral difference based measurement module 152. Based on the spectral differences, one or more parameters of interest, including changes in the one or more parameters of interest due to interfering process steps between adjacent measurement points, are determined. 1…M-1 POI MEASA learned spectral difference based measurement model is utilized to estimate the value of 155. The values ​​of one or more parameters of interest are communicated to a memory, such as memory 132, for example.

[0044] In some embodiments, a trained spectral difference-based measurement model is trained to receive spectral difference measurements between two particular measurement steps and generate estimates of one or more parameters of interest based on the spectral differences. In these embodiments, different trained spectral difference-based measurement models are utilized to estimate values ​​of the one or more parameters of interest based on the spectral difference measurements associated with different pairs of measurement steps.

[0045] 3 shows trained spectral difference-based measurement models 200 through 203. Each spectral difference-based measurement model is trained based on the spectral differences associated with a different pair of measurement steps. Thus, each trained spectral difference-based measurement model operates on the spectral difference measurements associated with a particular pair of measurement steps in the manufacturing process flow.

[0046] 2 , trained spectral difference-based measurement model 200 receives measurements of the spectral difference between measured spectral signals 186 and 187 associated with measurement steps 181 and 182 and generates values ​​204 of one or more parameters of interest associated with the structure measured in measurement steps 181 and 182. Similarly, trained spectral difference-based measurement model 201 receives measurements of the spectral difference between measured spectral signals 187 and 188 associated with measurement steps 182 and 183 and generates values ​​205 of one or more parameters of interest associated with the structure measured in measurement steps 182 and 183. Trained spectral difference-based measurement model 202 receives measurements of the spectral difference between measured spectral signals 188 and 189 associated with measurement steps 183 and 184 and generates values ​​206 of one or more parameters of interest associated with the structure measured in measurement steps 183 and 184. The trained spectral difference based measurement model 203 receives measurements of the spectral difference between the measured spectral signals 189 and 190 associated with the measurement steps 184 and 185 and generates values ​​207 of one or more parameters of interest associated with the structure measured in the measurement steps 184 and 185.

[0047] In some other embodiments, the trained spectral difference-based measurement model is trained to receive multiple measurements of spectral differences, each associated with a different pair of measurement steps, and to estimate values ​​of one or more parameters of interest based on each spectral difference. In these embodiments, the trained spectral difference-based measurement model is used to estimate values ​​of one or more parameters of interest based on the measurements of spectral differences associated with the different pair of measurement steps.

[0048] FIG. 4 illustrates a manufacturing process flow 180 including the sequence of manufacturing process steps shown in FIG. 3 in another embodiment. Like-numbered elements shown in FIG. 4 and referenced in FIG. 3 are similar. As shown in FIG. 4, a spectral difference-based measurement model 210 has been trained based on spectral differences associated with multiple different pairs of measurement steps, namely, measurement steps 181 and 182, measurement steps 182 and 183, measurement steps 183 and 184, and measurement steps 184 and 185. In the embodiment shown in FIG. 4, the trained spectral difference-based measurement model 210 is provided with the measured values ​​of spectral differences 191 through 194 as inputs. In response, the trained spectral difference-based measurement model 210 estimates values ​​211 of one or more parameters of interest based on the measured values ​​of spectral differences associated with the different pairs of measurement steps in the manufacturing process flow. The estimated values ​​211 are communicated to a memory, such as memory 132.

[0049] In some other embodiments, the trained spectral difference-based measurement model is a machine learning-based measurement model trained to estimate values ​​of structural parameters of interest from spectral difference measurements. Exemplary machine learning-based models include linear models, polynomial models, response surface models, decision tree models, random forest models, support vector machine models, neural network models, deep learning models, genetic algorithm-based models, or other types of models.

[0050] In another aspect, the trained spectral difference-based measurement model is used to simulate changes in the value of one or more parameters of interest, and the simulated changes are used to adjust process control variables to improve process performance. In some embodiments, expected spectral difference data is generated, for example, using a process and metrology simulator, or based on past usage examples, user experience, etc. The expected spectral difference data is provided as input to the trained spectral difference-based measurement model. The trained spectral difference-based measurement model estimates expected values ​​of one or more parameters of interest based on the expected spectral difference data. In this manner, the trained spectral difference-based measurement model not only monitors the production of multilayer film stacks in-line, but also predicts production trends. Furthermore, one or more process variables, such as focus, dose, etch time, etc., are adjusted based on the expected values ​​of the parameters of interest. In this manner, the process is adjusted to improve process performance.

[0051] In another aspect, a spectral difference based measurement model is trained based on design of experiments (DOE) spectral signals associated with different measurement steps in a manufacturing process flow.

[0052] In some examples, a measurement model based on spectral differences is trained based on simulated or actual measurement data associated with measuring design-of-experiment (DOE) wafers. Machine learning, feature extraction, and other techniques are used to train a direct input-output model (i.e., transfer function) relating the DOE spectral differences to corresponding reference measurements of the parameter of interest. In some embodiments, the training measurement object set includes substantially identical objects, i.e., the objects differ from each other by process variables. In some embodiments, process variables that affect the parameter of interest are intentionally amplified for model training.

[0053] In some examples, the DOE spectral signals and corresponding parameter values ​​of interest associated with one or more process steps are generated synthetically, i.e., by simulation. For example, a process simulator such as Positive Resist Optical Lithography (PROLITH) simulation software available from KLA-Tencor Corporation of Milpitas, California, USA can be utilized. Generally, any process modeling technique or tool is contemplated within the scope of this patent document (e.g., Coventor simulation software available from Coventor, Inc. of Cary, North Carolina, USA).

[0054] In some examples, the DOE spectral signal includes two ellipsometric parameters (Ψ, Δ) between spectral ranges acquired at different measurement locations, but in general, the measurement data can be any measurement data that is indicative of structural or topographical properties of structures patterned on the surface of a semiconductor wafer.

[0055] For model training, metrology data can be acquired from any location with known perturbations in design parameters, such as structure or process parameters. These locations can be, for example, within a scribe line, on a device, or at other locations on a wafer where, for example, lithography exposure conditions or reticle design features vary within a range of values. In another example, metrology data can be acquired from different device locations (e.g., a location with dense features and a location with isolated features, or locations with two different CDs on a mask). Typically, metrology data is acquired from different locations with perturbations in a known manner. The perturbations can be known from mask data, Equipment Data Acquisition (EDA) data, process data, etc.

[0056] The value of the parameter(s) of interest is known at each of a plurality of measurement locations.

[0057] In some examples, in preferred embodiments, a systematic set of variables is realized in the fabrication of actual DOE wafers. The DOE wafers are then measured to generate raw measurement data. The fabricated wafers contain systematic errors that cannot be easily modeled by simulation. For example, the effects of underlayers are more accurately captured by measuring the actual wafers. The effects of underlayers can be de-correlated from the measurement response by varying fabrication process parameters, such as focus and exposure variables, for fixed underlayer conditions. In another example, the effects of underlayers can be reduced by acquiring multiple data sets from features with varying topography of the overlayer and constant underlayer conditions. In one example, the overlayer can have a periodic structure and the underlayer can be aperiodic.

[0058] The measurement location can be selected to increase the measurement sensitivity. In one example, measurements performed at the line ends are most sensitive to changes in focus. In general, measurements should be made at the structures that are most sensitive to changes in the parameter being measured.

[0059] In some other examples, the DOE spectral signals and corresponding parameter values ​​of interest associated with one or more process steps are measured from an actual DOE wafer. The DOE spectral measurement data includes actual spectral measurements at multiple measurement steps in a manufacturing process flow. The corresponding DOE values ​​of the parameter(s) of interest characterizing the measurement object are measured by a reference measurement system. The reference measurement system is a reliable metrology system, such as a scanning electron microscope (SEM), a tunneling electron microscope (TEM), an atomic force microscope (AFM), or an X-ray measurement system, such as a small-angle X-ray scatterometer (SAXS) or an X-ray fluorescence (XRF) system, that can accurately measure the parameter values. However, reference measurement systems typically lack the capability to operate as inline metrology systems due to, for example, low throughput or high measurement uncertainty when measuring individual positions.

[0060] In some embodiments, the process variables and corresponding parameter variables are organized in a design of experiments (DOE) pattern on the surface of a semiconductor wafer (e.g., a DOE wafer). In this manner, the measurement system interrogates various locations on the wafer surface corresponding to different process and corresponding structural parameter values.

[0061] In general, measurement data related to any known variation of a process parameter (eg, lithographic focus, exposure, or other local or global parameter), a structural parameter, or both, may be considered.

[0062] In some examples, one or more features of the spectral measurement data are extracted by reducing the magnitude of the measurement data, and although this reduction is optional, if performed, a spectral difference-based measurement model is trained based at least in part on the one or more extracted features.

[0063] In general, the magnitude of the measurement data can be reduced by a number of known methods, including principal component analysis, non-linear principal component analysis, selection of individual signals from the measurement data of the second quantity, and filtering of the measurement data of the second quantity.

[0064] In some examples, the measurement data is analyzed using principal component analysis (PCA), nonlinear PCA, kernel PCA, independent component analysis (ICA), fast Fourier transform analysis (FFT), discrete cosine transform analysis (DCT), or a combination of these techniques to extract features that most strongly reflect changes in structural parameters present at different measurement locations. In some other examples, signal filtering techniques may be applied to extract signal data that most strongly reflect changes in parameters present at different measurement locations. In some other examples, individual signals that most strongly reflect changes in parameters present at different measurement locations may be selected from multiple signals present in the measurement data. Extracting features from the measurement data and processing the information to reduce the size of the data subject to subsequent analysis is preferred, but not strictly required.

[0065] 5 illustrates one embodiment of a spectral difference-based measurement model learning engine 160. In some embodiments, the computing system 130 is configured as the spectral difference-based measurement model learning engine 160 described herein. As shown in FIG. 5, the spectral difference-based measurement model learning engine 160 includes a spectral difference module 161, a machine learning module 162, and an error estimation module 163.

[0066] A training data set containing DOE spectra from M different measurement points in a manufacturing process flow at N sample wafer locations 1…M S 1…N DOE 165 is communicated to the spectral difference module 161. Each different measurement point is separated from adjacent measurement points by one or more manufacturing process steps in the manufacturing process flow.

[0067] In the example shown in FIG. 5, the spectral difference module 161 measures the spectral difference between adjacent measurement points in the process flow. 1…M-1 S-DIFF 1…N DOE 166. Thus, DOE measurement data at M different measurement points in the manufacturing process flow results in M-1 measurements of spectral differences associated with each of the N sample locations.

[0068] Spectral Difference 1…M-1 S-DIFF 1…N DOE The DOE measurements, indicative of 166, are provided as input to machine learning module 162. In some examples, the spectral difference based measurement model is a neural network model. As shown in FIG. 5, machine learning module 162 generates a neural network model based on the data set. 1…M-1 S-DIFF 1…N DOE 166. The output of the neural network model is an estimate of each parameter of interest associated with each pair of measurement points in the process flow.1…M-1 POI * 167. Estimated value 1…M-1 POI * 167 is communicated to the error estimation module 208. The error estimation module 208 calculates the estimate of the parameter of interest determined by the neural network model. 1…M-1 POI * 167 to the reliable values ​​of the corresponding parameters of interest. 1…M-1 POI DOE 168. The error estimation module 208 updates the neural network weights 169 to minimize a function (e.g., a quadratic error function, a linear error function, or any other suitable difference function) characterizing the difference between the determined and known values ​​of the parameter of interest. The updated neural network weights 169 are communicated to the machine learning module 162. The machine learning module 162 updates the neural network model with the updated neural network weights for the next iteration of the training process. The iterations continue until the function characterizing the difference between the determined and known values ​​of the parameter of interest is minimized. The resulting trained spectral difference-based measurement model 170 is communicated to a memory (e.g., memory 132).

[0069] As shown in FIG. 5, the spectral difference based measurement model learning engine 160 learns the values ​​of the parameters of interest. 1…M-1 POI DOE 168 from a reference source 164. The reference source 164 is a trusted metrology system, simulator, or both, utilized to generate the DOE parameter value sets, as previously described herein.

[0070] 6 illustrates a method 300 suitable for implementation by a metrology system, such as metrology system 100 shown in FIG. 1 of the present invention. It is recognized that, in one aspect, the data processing blocks of method 300 may be performed via pre-programmed algorithms executed by one or more processors of computing system 130, or any other general-purpose computing system. It is recognized that the specific structural aspects of metrology system 100 herein should be construed as illustrative only, and not limiting.

[0071] In block 301, a constant amount of illumination light is provided to one or more measurement targets disposed on a wafer for each of a plurality of different measurement instances in a semiconductor manufacturing process flow, each of the different measurement instances being separated by at least one semiconductor process step in the semiconductor manufacturing process flow.

[0072] At block 302, an amount of light from one or more measurement targets is detected in response to an amount of illumination light.

[0073] In block 303, a spectral measurement signal is generated in response to the amount of light detected at each of a plurality of different measurement instances in the semiconductor manufacturing process flow.

[0074] At block 304, a measurement indicative of a spectral difference between the spectral measurement signals associated with two different measurement instances of the plurality of different measurement instances is determined.

[0075] In block 305, values ​​of structural parameters of interest characterizing one or more measurement objects are estimated based on a trained spectral difference-based measurement model that functionally relates measurements indicative of spectral differences to values ​​of the structural parameters of interest.

[0076] In some examples, using measurement data associated with multiple targets for model building, training, and measurement eliminates or significantly reduces the influence of underlying layers on the measurement results. In one example, measurement signals from two targets are subtracted to eliminate or significantly reduce the influence of underlying layers on each measurement result. Using measurement data associated with multiple targets increases the sample and process information embedded in the model. Specifically, using training data including measurements of multiple different targets at one or more measurement locations enables more accurate measurements.

[0077] In one example, a measurement model is generated from spectral measurements of a DOE wafer for both isolated and dense objects. The measurement model is then trained based on the spectral measurement data and known structural parameter values. The resulting trained measurement model is then used to calculate structural parameter values ​​for both isolated and dense objects on the sample wafer. In this way, each parameter has its own trained model that calculates the parameter value from measured spectra (or extracted features) associated with both isolated and dense objects.

[0078] In another further aspect, measurement data obtained from measurements performed using a combination of multiple different measurement techniques is collected for model building, training, and measurement. Using measurement data associated with multiple different measurement techniques increases the sample and process information embedded in the model, enabling more accurate measurements. Measurement data can be obtained from measurements performed using any combination of multiple different measurement techniques. In this manner, measurements can be performed at different measurement locations using multiple different measurement techniques to improve the measurement information available for characterizing a semiconductor structure.

[0079] Generally, within the scope of this patent document, any measurement technique or a combination of two or more measurement techniques may be considered. Exemplary measurement techniques include, but are not limited to, spectroscopic ellipsometry, including Mueller matrix ellipsometry, spectroscopic reflectometry, spectroscopic scatterometry, scatterometry overlay, beam profile reflectometry, both angle-resolved and polarization-resolved, beam profile ellipsometry, single or multiple discrete wavelength ellipsometry, transmission small-angle X-ray scatterometry (TSAXS), small-angle X-ray scattering (SAXS), grazing-incidence small-angle X-ray scattering (GISAXS), wide-angle X-ray scattering (WAXS), X-ray reflectivity (XRR), X-ray diffraction (XRD), grazing-incidence X-ray diffraction (GIXRD), high-resolution X-ray diffraction (HRXRD), X-ray photoelectron spectroscopy (XPS), X-ray fluorescence (XRF), grazing-incidence X-ray fluorescence (GIXRF), low-energy electron-induced X-ray emission scatterometry (LEXES), X-ray tomography, and X-ray ellipsometry. Generally, any metrology technique applicable to characterizing semiconductor structures may be considered, including image-based metrology techniques. Further sensor options include electrical sensors, such as non-contact capacitance / voltage or current / voltage sensors, that bias the device and detect the resulting bias with an optical sensor (or vice versa), or with auxiliary optical techniques, such as XRD, XRF, XPS, LEXES, SAXS, or pump-probe techniques. In one embodiment, a two-dimensional beam profile reflectometer (pupil imager) may be used to collect both angle-resolved and / or multispectral data with a small spot size. A UV Linnik interferometer may also be used as a Mueller matrix spectral pupil imager.

[0080] In another example, the methods and systems described herein can be applied to overlay metrology. Particularly relevant to overlay metrology is diffraction grating metrology. The purpose of overlay metrology is to determine the offset between different lithography exposure steps. Performing overlay metrology on devices is difficult due to the small size of structures on the devices and the generally small overlay values.

[0081] For example, the pitch of typical scribe-line overlay metrology structures varies from 200 nanometers to 2,000 nanometers. However, the pitch of overlay metrology structures on devices is typically 100 nanometers or less. Also, in a nominal production environment, device overlay is only a small fraction of the periodicity of the device structures. In contrast, proxy metrology structures used in scatterometry overlay are frequently offset by larger values, such as one-quarter of the pitch. This increases the sensitivity of the overlay signal.

[0082] Under these conditions, overlay metrology is performed with a sensor architecture that is sufficiently sensitive to small misalignments and small pitch overlay. The methods and systems described herein can be utilized to obtain overlay-sensitive measurement signals based on on-device structures, proxy structures, or both.

[0083] After acquisition, the measurement signals are analyzed to determine overlay errors based on variations in the measurement signals. In a further aspect, the spectrally or angle-resolved data is analyzed using PCA, and an overlay model is trained to determine overlay based on principal components detected in the measurement signals. In one example, the overlay model is a neural network model. Therefore, the overlay model is not a parametric model and is therefore less susceptible to errors introduced by inaccurate modeling assumptions.

[0084] In some embodiments, the overlay metrology model is trained based on measurements of dedicated metrology structures that are substantially identical to the device features but with larger offsets. This can help overcome sensitivity issues. These offsets can be introduced by fixed design offsets introduced between the features of the two layers being measured during reticle design. Offsets can also be introduced by lithography exposure misalignment. By using multiple offset targets (e.g., pitch / 4 and -pitch / 4), the overlay error can be extracted more efficiently from the compressed signal (e.g., PCA signal). Furthermore, the influence of the underlying layer can also be mitigated.

[0085] It should be appreciated that the various steps described throughout this disclosure may be performed by a single computing system 130 or by multiple computing systems 130. Furthermore, different subsystems of system 100, such as spectroscopic ellipsometer 104, may include computing systems suitable for performing at least some of the steps described herein. Accordingly, the above description should not be construed as limiting the invention, but merely as illustrative. Furthermore, one or more computing systems 130 may be configured to perform any other step(s) of any method embodiment described herein.

[0086] Additionally, computing system 130 may be communicatively coupled to spectrometer 104 in any manner known in the art. For example, one or more computing systems 130 may be coupled to a computing system associated with spectrometer 104. In another example, spectrometer 104 may be directly controlled by a single computing system coupled to computing system 130.

[0087] The computing system 130 of the metrology system 100 may be configured to receive and / or acquire data or information from a subsystem of the system (e.g., spectrometer 104, etc.) or one or more process tools 120 by way of a transmission medium, which may include wired and / or wireless portions. In this manner, the transmission medium may act as a data link between the computing system 130 and other systems or subsystems of the system 100.

[0088] The computing system 130 of the measurement system 100 may be configured to receive and / or acquire data or information (e.g., measurement results, modeling inputs, modeling results, etc.) from other systems via 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 computing system 130 and other systems (e.g., memory onboard the measurement system 100, external memory, a reference measurement source, or other external systems). 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 the data link. For example, spectral results acquired using the spectrometer 104 may be stored in a persistent or semi-persistent memory device (e.g., the memory 132 or an external memory). In this regard, the spectral results may be imported from the onboard memory or from an external memory system. Additionally, the computing system 130 may transmit data to other systems via the transmission medium. For example, the measurement model or structural parameter values ​​140 determined by the computing system 130 may be communicated to and stored in external memory. In this regard, the measurement results may be exported to another system.

[0089] Computing system 130 may include, but is not limited to, a personal computing system, a mainframe computing system, a workstation, an image 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 having one or more processors that executes instructions from a memory medium.

[0090] Program instructions 134 for implementing methods as 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 over bus 133. Program instructions 134 are stored on 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 magnetic tape.

[0091] In some embodiments, the illumination light and the light collected from the illuminated measurement site include multiple different wavelengths. In some embodiments, light is collected from the illuminated measurement site at multiple different collection angles. Detecting light at multiple wavelengths and collection angles improves measurement sensitivity to critical dimensions (e.g., CD). In some embodiments, light is collected from the illuminated measurement site at multiple different azimuthal angles. These out-of-plane measurements can also improve measurement sensitivity to critical dimensions. In some embodiments, the collection of optical measurement data is optimized for a particular set of system settings, such as, for example, a spectroscopic or angle-resolved system, one or more azimuthal angles, one or more wavelengths, and any combination thereof.

[0092] In some examples, the model building, training, and measurement methods described herein are implemented as components of a SpectraShape® Optical Critical Dimension Metrology System available from KLA-Tencor Corporation, Milpitas, Calif. In this manner, a ready-to-use model is created immediately after a DOE wafer spectrum is collected by the system.

[0093] In some other examples, the model building and training methods described herein are performed offline by a computing system running AcuShape® software, available from KLA-Tencor Corporation of Milpitas, Calif. The resulting trained model may be incorporated as an element of an AcuShape® library that can be accessed by metrology systems that perform measurements.

[0094] In general, the methods and systems for performing semiconductor metrology presented herein can be applied directly to actual device structures or to dedicated metrology targets (e.g., proxy structures) located within the die or within the scribe line.

[0095] In yet another aspect, the measurement techniques described herein may 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 structural parameters determined using the methods described herein may be communicated to a lithography tool to adjust the lithography system to achieve a desired output. Similarly, etching parameters (e.g., etch time, diffusivity, etc.) or deposition parameters (e.g., time, concentration, etc.) may be included in the measurement model to provide active feedback to an etch tool or a deposition tool, respectively.

[0096] In general, the systems and methods described herein can be implemented as part of a dedicated metrology tool or can be implemented as part of a process tool (e.g., a lithography tool, an etch tool, etc.).

[0097] The term "critical dimension" as used herein includes any critical dimension of a structure (e.g., bottom critical dimension, middle 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 displacement between two or more structures (e.g., the overlay displacement between overlapping grating structures, etc.). The structures may include three-dimensional structures, patterned structures, overlay structures, etc.

[0098] The terms "critical dimension application" or "critical dimension measurement application" as used herein include any critical dimension measurement.

[0099] The term "metrology system" as used herein includes any system utilized at least in part to characterize a specimen in any manner, including metrology applications such as critical dimension metrology, overlay metrology, focus / dose metrology, and composition metrology. However, such terminology does not limit the scope of the term "metrology system" as used herein. Additionally, the metrology system 100 may be configured for measurement 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 simultaneous data from one or more platforms), as well as any other metrology or inspection tool that benefits from calibration of system parameters based on critical dimension data.

[0100] Various embodiments are described herein for semiconductor processing systems (e.g., inspection systems or lithography systems) that may be used to process specimens. As used herein, the term "specimen" refers to a wafer, reticle, or any other specimen that may be processed (e.g., printed or inspected for defects) by means known in the art.

[0101] As used herein, the term "wafer" generally refers to a substrate made of semiconductor or non-semiconductor material. Examples include, but are not limited to, monocrystalline silicon, gallium arsenide, and indium phosphide. Such substrates may commonly exist and / or be processed within a semiconductor manufacturing facility. 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 repeatable pattern features.

[0102] A "reticle" may refer to a reticle at any stage in the reticle manufacturing process or to a finished reticle that may or may not be sold for use in a semiconductor manufacturing facility. A reticle or "mask" is generally defined as a substantially transparent substrate having substantially translucent regions formed thereon, configured in a pattern. The substrate may comprise, for example, a glass material such as amorphous SiO2. The reticle may be placed over a resist-covered wafer during the exposure step of the lithography process so that the pattern on the reticle can be transferred to the resist.

[0103] One or more layers formed on a wafer may be patterned or unpatterned. For example, a wafer may include multiple dies, each with repeatable pattern features. The formation and processing of such layers of material may ultimately 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 encompass a wafer on which any type of device known in the art may be fabricated.

[0104] In one or more exemplary embodiments, the functions described may be implemented in hardware, software, firmware, or any combination thereof. If implemented in software, the functions may be stored on or transmitted over 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. Storage media may be any available medium that can be accessed by a general-purpose or special-purpose computer. By way of example, and not limitation, such computer-readable media may include RAM, ROM, EEPROM, CD-ROM or other optical disk storage, magnetic disk storage or other magnetic storage devices, or any other medium that can be used to carry or store desired program code means in the form of instructions or data structures and that can be accessed by a general-purpose or special-purpose computer, or a general-purpose or special-purpose processor. Furthermore, any connection is properly termed a computer-readable medium. For example, if software is transmitted from a website, server, or other 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 medium. 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 disks typically reproduce data magnetically while discs reproduce data optically with lasers. Combinations of the above are also included within the scope of computer-readable media.

[0105] Although, for instructional purposes, several specific embodiments are described above, the teachings of this patent document have general applicability and are not limited to the specific embodiments described above. Accordingly, various modifications, adaptations, and combinations of various features of the described embodiments can be made without departing from the scope of the invention as defined in the claims.

Claims

1. 1. A measurement system comprising: a spectroscopic measurement subsystem, the spectroscopic measurement subsystem comprising: an illuminator configured to provide a quantity of illumination light to one or more metrology targets disposed on a wafer for each of a plurality of different measurement instances in a semiconductor manufacturing process flow, wherein each of the different measurement instances is separated by at least one semiconductor process step in the semiconductor manufacturing process flow; and a detector configured to detect an amount of light from the one or more measurement targets in response to the amount of illumination light, and to generate a spectral measurement signal in response to the amount of light detected at each of the plurality of different measurement instances in the semiconductor manufacturing process flow. a spectroscopic measurement subsystem comprising:

1. A computing system comprising: receiving the spectral measurement signals associated with each of the plurality of different measurement instances; determining a measurement value indicative of a spectral difference between spectral measurement signals associated with two different measurement instances of the plurality of different measurement instances; Estimating values ​​of structural parameters of interest characterizing the one or more measurement objects based on a trained spectral difference-based measurement model that functionally relates the measurements indicative of the spectral differences to the values ​​of the structural parameters of interest. a computing system configured to: A measurement system comprising:

2. The metrology system of claim 1 , wherein the measurement indicative of the spectral difference is a difference in measured intensity at each distinct pair of measured wavelengths.

3. 10. The metrology system of claim 1, wherein the measurement indicative of the spectral difference is a difference in measured harmonic signal values ​​at each distinct pair of measured wavelengths.

4. The metrology system of claim 1 , wherein the measurement indicative of the spectral difference is a difference in measurements of one or more Mueller matrix elements at each distinct pair of measured wavelengths.

5. The metrology system of claim 1 , wherein the measurement indicative of the spectral difference is a scalar value indicative of a difference in measurement of the spectral measurement signals at each distinct pair of measured wavelengths.

6. 6. The metrology system of claim 5, wherein the scalar value is a root mean square error (RSME) measurement of the difference in measurements of the spectral measurement signal at each of the distinct measured wavelength pairs.

7. The metrology system of claim 1 , wherein the measurements indicative of the spectral differences are a set of coefficients characterizing a mathematical function that fits the spectral differences at each distinct set of wavelengths.

8. The metrology system of claim 1 , wherein the measurement indicative of a spectral difference is determined based on a weighting of the spectral difference between spectral measurement signals as a function of wavelength between distinct wavelength pairs.

9. The metrology system of claim 1 , wherein the two different measurement instances of the plurality of different measurement instances are adjacent measurement instances in the semiconductor manufacturing process flow.

10. 10. The metrology system of claim 1, wherein the computing system is further configured to train the spectral difference based measurement model with simulated measurement data, actual measurement data associated with a design of experiments (DOE) wafer, or a combination thereof.

11. The metrology system of claim 1 , wherein the illuminator and the detector comprise one of a spectroscopic ellipsometer, a spectroscopic reflectometer, a spectroscopic scatterometer, a beam profile reflectometer, and a beam profile ellipsometer.

12. providing a constant amount of illumination light to one or more measurement targets disposed on a wafer for each of a plurality of different measurement instances in a semiconductor manufacturing process flow, each of the different measurement instances being separated by at least one semiconductor process step in the semiconductor manufacturing process flow; detecting a certain amount of light from the one or more measurement targets in response to the certain amount of illumination light; generating a spectral measurement signal responsive to the amount of light detected at each of the plurality of different measurement instances in the semiconductor manufacturing process flow; determining a measurement value indicative of a spectral difference between spectral measurement signals associated with two different measurement instances of the plurality of different measurement instances; estimating values ​​of structural parameters of interest characterizing the one or more measurement objects based on a trained spectral difference-based measurement model that functionally relates the measurements indicative of the spectral differences to the values ​​of the structural parameters of interest; A method comprising:

13. 13. The method of claim 12, wherein the measurement indicative of the spectral difference is a difference in measured intensity at each of a respective set of measured wavelengths, a difference in measured harmonic signal values ​​at each of a respective set of measured wavelengths, or a difference in measured values ​​of one or more Mueller matrix elements at each of a respective set of measured wavelengths.

14. The method of claim 12 , wherein the measurement indicative of the spectral difference is a scalar value indicative of the difference in measurements of the spectral measurement signals at each distinct pair of measured wavelengths.

15. The method of claim 12 , wherein the measurements indicative of the spectral differences are a set of coefficients that characterize a mathematical function that fits the spectral differences at each distinct set of wavelengths.

16. The method of claim 12 , wherein the measurement indicative of a spectral difference is determined based on a weighting of the spectral difference between spectral measurement signals as a function of wavelength between distinct wavelength pairs.

17. The method of claim 12 , wherein the two different measurement instances of the plurality of different measurement instances are adjacent measurement instances in the semiconductor manufacturing process flow.

18. 13. The method of claim 12, further comprising training the spectral difference based measurement model with simulated measurement data, actual measurement data associated with a design of experiments (DOE) wafer, or a combination thereof.

19. 1. A measurement system comprising: a spectroscopic measurement subsystem, the spectroscopic measurement subsystem comprising: an illuminator configured to provide a quantity of illumination light to one or more metrology targets disposed on a wafer for each of a plurality of different measurement instances in a semiconductor manufacturing process flow, wherein each of the different measurement instances is separated by at least one semiconductor process step in the semiconductor manufacturing process flow; and a detector configured to detect an amount of light from the one or more measurement targets in response to the amount of illumination light, and to generate a spectral measurement signal in response to the amount of light detected at each of the plurality of different measurement instances in the semiconductor manufacturing process flow. a spectroscopic measurement subsystem comprising: A non-transitory computer-readable medium that, when executed by one or more processors, causes the one or more processors to: receiving the spectral measurement signals associated with each of the plurality of different measurement instances; determining a measurement indicative of a spectral difference between spectral measurement signals associated with two different measurement instances of the plurality of different measurement instances; and estimating values ​​of structural parameters of interest characterizing the one or more measurement objects based on a trained spectral difference-based measurement model that functionally relates the measurements indicative of the spectral differences to the values ​​of the structural parameters of interest. a computer-readable medium storing instructions; A measurement system comprising:

20. 20. The metrology system of claim 19, wherein the two different measurement instances of the plurality of different measurement instances are adjacent measurement instances in the semiconductor manufacturing process flow.

Images