Single wafer orientation tool induced offset removal

Abstract

A metrology system may include a spectral metrology subsystem and a controller including one or more processors configured to execute program instructions, the program instructions are configured to cause the one or more processors to: generate, by the spectral metrology subsystem, a tool induced offset (TIS) model of a training sample, including: receiving training data from metrology measurements of the training sample, the training data comprises spectral data associated with at least one off-diagonal Mueller matrix element resulting from one or more first measurements of the training sample at a first azimuth angle and one or more second measurements of the training sample at a second azimuth angle, deriving superimposed spectral data and TIS spectral data from the training data, decomposing the superimposed spectral data and the TIS spectral data, and inferring a TIS signature of the training sample; and removing the TIS signature from the test sample.

Description

[0001] Cross-reference of related applications

[0002] This application claims the right to U.S. Provisional Application No. 63 / 545,796, filed October 26, 2023, pursuant to 35 USC § 119(e), the entire contents of which are incorporated herein by reference. Technical Field

[0003] This disclosure generally relates to superimposed metering, and more specifically, to removing metering subsystem errors from metering measurements. Background Technology

[0004] OVL (Output Volume) metrology techniques continuously require improvement to keep pace with the increasing circuit density in semiconductor lithography. For example, post-development inspection (ADI) OVL control requires a relatively high number of inspection sites per wafer to achieve higher-order model correction. The increasing use of per-exposure correction (CPE) methods by customers further drives the demand for measurements of more in-field targets. In response to the increased sampling requirements resulting from increased circuit density, spectral ellipsometry has been developed, particularly for measuring OVL in post-etch inspection (AEI) apparatuses.

[0005] A challenge faced by spectroscopic ellipsometers is that their asymmetric architecture contributes to tool-induced offsets (TIS) that must be removed or cleared from the reported OVL. One method for clearing TIS from the reported OVL is by averaging measurements on both wafer orientations and then subtracting the result from the measured OVL. However, performing this process for each bit is time-consuming, leading to increased costs. Therefore, it is desirable to provide a system and method for removing TIS from OVL measurements without requiring measurements on both wafer orientations. Summary of the Invention

[0006] According to one or more embodiments of this disclosure, a spectrometric system is disclosed. In one illustrative embodiment, the spectrometric system includes: a spectrometric subsystem; and a controller communicatively coupled to the spectrometric subsystem, the controller including one or more processors configured to execute program instructions. In another illustrative embodiment, the instructions are configured to cause the one or more processors to: generate a tool-induced offset (TIS) model of a training sample via the spectrometric subsystem, comprising: receiving training data from metrological measurements of the training sample, the training data including spectral data associated with at least one off-diagonal Mueller matrix element generated by one or more first measurements of the training sample at a first azimuth angle and one or more second measurements of the training sample at a second azimuth angle; deriving superimposed spectral data and TIS spectral data from the training data; decomposing the superimposed spectral data and the TIS spectral data; and inferring a TIS signature of the training sample based on the decomposition of the training data and the superimposed spectral data and the TIS spectral data. In another illustrative embodiment, the instructions are configured to cause one or more processors to remove the TIS signature from the test sample, comprising: receiving spectral data of a single azimuth measurement of the test sample; and removing the TIS signature from the single azimuth measurement of the test sample based on the TIS model.

[0007] According to one or more embodiments of this disclosure, a method for removing a TIS signature from a metrological measurement is disclosed. In one illustrative embodiment, the method includes generating a TIS model of a training sample via a spectrometrology subsystem, comprising: receiving training data from metrological measurements of the training sample, the training data including spectral data associated with at least one off-diagonal Mueller matrix element generated by one or more first measurements of the training sample at a first azimuth angle and one or more second measurements of the training sample at a second azimuth angle; deriving superimposed spectral data and TIS spectral data from the training data; decomposing the superimposed spectral data and the TIS spectral data; and inferring a TIS signature of the training sample based on the decomposition of the training data, the superimposed spectral data, and the TIS spectral data; and removing the TIS signature from a test sample comprises: receiving spectral data of a single azimuth measurement of the test sample. In another illustrative embodiment, the method includes: removing the TIS signature from the single azimuth measurement of the test sample based on the TIS model.

[0008] It should be understood that both the foregoing general description and the following detailed description are exemplary and illustrative only and are not intended to limit the disclosure as claimed. The accompanying drawings, which are incorporated in and form a part of this specification, illustrate embodiments of the disclosure and, together with the general description, serve to explain the principles of the disclosure. Attached Figure Description

[0009] Those skilled in the art can better understand the many advantages of this disclosure by referring to the accompanying drawings.

[0010] Figure 1A A conceptual diagram illustrating a semiconductor device metering system according to one or more embodiments of the present disclosure.

[0011] Figure 1B A conceptual diagram illustrating a metering subsystem according to one or more embodiments of the present disclosure is provided.

[0012] Figure 2A The description depicts simulated M-wavelength shifts in a test sample within the wavelength range according to one or more embodiments of this disclosure. 20 -M 02 The measured curve.

[0013] Figure 2B The description depicts a superimposed simulated M-wavelength array of test samples within the wavelength range according to one or more embodiments of this disclosure. 20 -M 02 The measured curve.

[0014] Figure 3 The illustration depicts a process flowchart of a method for removing a signature from a metrological measurement (TIS) according to one or more embodiments of the present disclosure. Detailed Implementation

[0015] Reference will now be made in detail to the disclosed subject matter, which is illustrated in the accompanying drawings. This disclosure has been specifically shown and described with respect to certain embodiments and their specific features. The embodiments set forth herein are to be considered illustrative rather than restrictive. Those skilled in the art will readily understand that various changes and modifications in form and detail may be made without departing from the spirit and scope of this disclosure.

[0016] Embodiments of this disclosure pertain to systems and methods for removing TIS signatures from OVL measurements of test samples.

[0017] For the purposes of this disclosure, the term Overlay (OVL) is used to describe the relative positions of features on a sample fabricated by two or more photolithographic patterning steps. For example, a multilayer device may include features patterned on multiple sample layers using different photolithographic steps for each layer, where the alignment of features between layers must typically be tightly controlled to ensure proper performance of the resulting device.

[0018] Some OVL metrology techniques determine OVL measurements by illuminating a portion of a test sample and collecting data associated with the symmetry and / or overlap characteristics of the test sample. However, such techniques are sensitive to measurement inhomogeneities, which can manifest as errors in OVL measurements and are commonly referred to as tool-induced offset (TIS) errors. It should be understood that the examples and descriptions throughout this disclosure related to specific applications of OVL metrology are provided for illustrative purposes only and should not be construed as limiting the scope of this disclosure.

[0019] As used herein, a TIS signature refers to a set of TIS values ​​that vary based on the measurement wavelength of the training samples. TIS values ​​can vary in response to process variations in the samples. Specifically, the effect of process variations on TIS values ​​in a sample can be site-specific, and a TIS model containing information about the impact of site-specific process variations on TIS values ​​can be generated. During test sample measurements to determine process variations, a TIS signature can be inferred from the TIS model, which is then used to correct site-specific TIS for those known process variations. Therefore, when process variations in a test sample are determined during runtime, a customized TIS signature can be generated to remove TIS from single-angle measurements of the sample.

[0020] The generation of the TIS model may involve illuminating training samples at first and second azimuth angles and receiving spectral data from the detection of reflections from the training samples. The spectral data may include data from one or more off-diagonal Mueller matrix element pairs. Off-diagonal Mueller matrix element pairs corresponding to TIS signal values ​​or specific training process variations (e.g., layer thickness) within the training samples can then be used to train the model. Off-diagonal Mueller matrix elements that are selectively sensitive to TIS and selectively insensitive to OVL can also be used to train the model. Using the spectra of off-diagonal Mueller matrix elements for training provides more efficient TIS removal from superimposed measurements compared to relying solely on TIS values.

[0021] The embodiments of this disclosure are particularly advantageous for current TIS removal methods. For example, a simple method for determining TIS from a test sample is to average the OVL measurements in two opposite azimuth directions (0° and 180°), and then add the second measurement to the first measurement, as shown in Equation 1 below:

[0022] Equation 1.

[0023] The OVL after TIS removal can also be determined by averaging the superposition (OVL) measurements in two opposite azimuth directions and then subtracting the second measurement from the first measurement, as shown in Equation 2 below:

[0024] Equation 2.

[0025] However, this method requires measuring test samples in two azimuth directions for each measurement group, significantly increasing measurement time. However, when using the method described in this disclosure, once the model is trained, test samples only need to be measured in one azimuth direction instead of two opposite azimuth directions, thus reducing the total measurement time. Furthermore, the embodiments of this disclosure are advantageous because they can be implemented with minimal modifications to existing platform optics and software while providing an increase in the number of samples measured per unit time.

[0026] Figures 1A to 3 This invention describes a system and method for removing TIS signatures from test samples according to one or more embodiments of the present disclosure.

[0027] Figure 1A The following describes a conceptual diagram of a semiconductor device manufacturing system 100 according to one or more embodiments of the present disclosure. In one embodiment, system 100 includes a metrology subsystem 102 configured to characterize one or more properties of a test sample 104. In another embodiment, system 100 includes a controller 106 communicatively coupled to the metrology subsystem 102. In yet another embodiment, controller 106 includes one or more processors 108 configured to execute program instructions held on memory medium 110 or memory. For example, one or more processors 108 of controller 106 may generate (e.g., train) a TIS model of a training sample by first receiving training data from metrology subsystem 102 (e.g., from a detector within metrology subsystem 102), deriving OVL spectral data and TIS spectral data from the training data, and decomposing the derived OVL spectral data and TIS spectral data. Based on the decomposed OVL spectral data and TIS spectral data, one or more processors 108 of controller 106 may remove a TIS signature (e.g., a TIS signature inferred from the trained TIS model) from measurements of the test sample 104. For example, one or more processors 108 of controller 106 may receive spectral data of a single azimuth measurement of a test sample, and then remove the TIS signature from the single azimuth measurement of the test sample based on a TIS model.

[0028] Figure 1BThis illustration depicts a conceptual diagram of a metrology subsystem according to one or more embodiments of the present disclosure. In a general sense, metrology subsystem 102 can illuminate sample 104 with at least one illumination beam and collect at least one measurement signal from sample 104 in response to said illumination beam. The illumination beam may include, but is not limited to, any wavelength or wavelength range of optical beams (e.g., a light beam). In this way, metrology subsystem 102 can operate as an optical metrology subsystem. The metrology subsystem may be configured to include, but is not limited to, a spectroscopic ellipsometer (SE), an SE having multiple illumination angles, an SE measuring Mueller matrix elements (e.g., using a rotation compensator), a single-wavelength ellipsometer, a beam distribution ellipsometer (angle-resolved ellipsometer), a beam distribution reflectometer (angle-resolved reflectometer), a broadband reflectance spectrometer (spectral reflectometer), a single-wavelength reflectometer, an angle-resolved reflectometer, an imaging system, or a scatterometer (e.g., a speckle analyzer). The wavelength of the optical system may vary from about 120 nm to 3 micrometers. For non-ellipsometer systems, the collected signal may be polarization-resolved or unpolarized.

[0029] In one embodiment, the metrology subsystem 102 includes at least one illumination source 112 for generating illumination (e.g., one or more illumination beams 114). The illumination source 112 can provide (but is not required to provide) an illumination beam 114 having a wavelength in the range of about 100 nm to about 2,000 nm. In this regard, the illumination source 112 can generate illumination having wavelengths in any spectral range, including, but not limited to, vacuum ultraviolet wavelengths, extreme ultraviolet wavelengths, visible wavelengths, or infrared wavelengths. For example, the illumination source 112 can provide an illumination beam spanning ultraviolet wavelengths (e.g., but not limited to 160 nm to 1000 nm). By another example, the illumination source 112 can provide an illumination beam spanning both ultraviolet and visible wavelengths (e.g., but not limited to 150 nm to 900 nm). By yet another example, the illumination source 112 can provide an illumination beam spanning ultraviolet wavelengths (e.g., but not limited to 150 nm to 450 nm). Furthermore, the metrology subsystem 102 can include any number of illumination sources 112 for providing illumination in any one or more spectral ranges.

[0030] In another embodiment, illumination source 112 provides a tunable illumination source (e.g., one or more tunable lasers). By another example, illumination source 112 may comprise a broadband illumination source coupled to a tunable filter.

[0031] The illumination source 112 may further provide an illumination beam 114 having any time distribution. For example, the illumination beam 114 may have a continuous time distribution, a modulated time distribution, or a pulsed time distribution.

[0032] In another embodiment, the metrology subsystem 102 includes an illumination subsystem 116 (e.g., an illumination path) for guiding illumination (e.g., one or more illumination beams 114) from the illumination source 112 to the test sample 104, and a collection subsystem 118 (e.g., a collection path) for collecting radiation (e.g., light) emitted from the sample 104. The illumination subsystem 116 may include one or more illumination beam adjustment elements 120 suitable for modifying and / or adjusting the illumination. For example, one or more illumination beam adjustment elements 120 may include, but are not limited to, illumination aperture stops, illumination field stops, one or more polarizers, one or more compensators, one or more filters, one or more beam splitters, one or more diffusers, one or more homogenizers, one or more apodizers, one or more beam shapers, one or more mirrors, or one or more lenses. For example, one or more illumination beam adjustment elements 120 may include one or more focusing mirrors that reflect light and focus the illumination beam onto a point on the test sample.

[0033] In another embodiment, the illumination subsystem 116 may utilize a focusing assembly 122 to focus illumination from one or more illumination sources 112 onto a test sample 104 mounted on a sample stage 124. For example, the focusing assembly 122 may include one or more optical elements with non-zero optical power. In another embodiment, the collection subsystem 118 may include a collection assembly 126 for collecting radiation from the test sample 104. For example, the collection assembly 126 may include one or more optical elements with non-zero optical power.

[0034] In another embodiment, the metrology subsystem 102 includes a detector 128 configured to capture light (e.g., sample light 130) emitted from the test sample 104 via the collection subsystem 118. For example, the detector 128 may receive radiation reflected or scattered from the test sample 104 (e.g., via specular reflection, diffuse reflection). By another example, the detector 128 may receive radiation generated by the test sample 104 (e.g., luminescence associated with absorption of the illumination beam 114).

[0035] Detector 128 may comprise any type of optical detector known in the art suitable for measuring illumination received from test sample 104. For example, detector 128 may comprise, but is not limited to, photodiode arrays (PDAs), charge-coupled devices (CCDs), complementary metal-oxide-semiconductor (CMOS) detectors, time-delay integration (TDI) detectors, photomultiplier tubes (PMTs), or avalanche photodiodes (APDs). Furthermore, detector 128 may comprise any type of sensor having any geometry, including, but not limited to, planar or line sensors. In another embodiment, detector 128 may comprise a spectral detector suitable for identifying the wavelength of radiation emitted from test sample 104.

[0036] For measurement techniques involving the collection of spectral data (such as spectroreflectometers or spectral ellipsometers), it is desirable to generate continuous spectral data within the spectral range of interest. For example, metrology subsystem 102 may include a dispersive element (such as a prism or grating) for spatially dispersing light from a superimposed target onto one or more detectors 128 to capture spectral measurements.

[0037] In another embodiment, the metrology subsystem 102 may include a plurality of detectors 128 for facilitating a plurality of metrological measurements of the metrology subsystem 102. In this regard, Figure 1B The metrology subsystem 102 described herein can perform multiple simultaneous metrological measurements.

[0038] The collection subsystem 118 may further include any number of collection beam adjustment elements 132 for guiding and / or modifying the illumination collected by the collection assembly 126, including, but not limited to, collection aperture stops, collection field stops, one or more mirrors, one or more lenses, one or more filters, one or more polarizers, or one or more compensators. For example, one or more collection beam adjustment elements 132 may include one or more focusing mirrors for guiding light to the detector 128. The collection beam adjustment element 132 and the illumination beam adjustment element 120 may both be referred to as beam adjustment elements 120 and 132.

[0039] The metrology subsystem 102, configured as a spectroelliptometer as described herein, can illuminate the test sample 104 at any selected angle of incidence (AOI) and at any selected azimuth angle from 0 degrees to 360 degrees. Furthermore, the spectroelliptometer can provide a range of measurements in various combinations of angle of incidence and azimuth.

[0040] Figure 2A and 2B The simulated M-wavelengths of TIS and OVL in the wavelength range of training samples with process variations, according to one or more embodiments of the present disclosure, are described respectively. 20 -M 02 The measured curves are shown in graphs 200 and 202. Asymmetric spectra were simulated for different superposition values. Each asymmetric spectrum was defined as the difference between spectral measurements performed at zero azimuth and 180-degree azimuth.

[0041] In this embodiment, the asymmetry of the measurement is derived from the difference between one or more off-diagonal Mueller elements. The Mueller matrix (M) is a 4×4 matrix describing the measured sample and can be written as:

[0042] .

[0043] Although the diagonal elements of the Mueller matrix (M 00 M 11 M 22 M33 This describes the intensity and fundamental polarization changes of the measured sample, but the off-diagonal elements contain six element pairs (M). 10 and M 01 M 20 and M 02 M 30 and M 03 M 21 and M 12 M 31 and M 31 and M 32 and M 23 The Mueller matrix describes the interaction between different polarization states. Each of the six Mueller matrix element pairs can be determined from spectral data collected from measurements of the samples. The increasing difference between off-diagonal Mueller matrix element pairs corresponds to an increasing asymmetry between measurements and can be considered an indicator of asymmetry.

[0044] Once the Mueller matrix element values ​​are collected, the Mueller matrix element pair difference (pMΔ) can be calculated. The focus is on Mueller matrix element pairs with pMΔ values ​​associated with the TIS magnitude. Also of interest are Mueller matrix element pairs with pMΔ values ​​associated with TIS magnitudes that vary according to process variations. As used herein, process variations refer to variations in sample properties within a sample that can affect the TIS at each point (TIS 3σ) and also the difference in OVL at each point (OVL 3σ). Process variations may include, but are not limited to, layer thickness, sample tilt, length, critical dimension, height, sidewall angle, film dispersion, and / or sample or sample element width. Detection and measurement of process variations are important for determining sample quality and predicting the functionality of the resulting semiconductor device.

[0045] The measurement can be performed in the same manner or with the same measurement and / or measurement process variations used to determine the element values ​​of the Mueller matrix. Alternatively, it can be performed via another measurement or measurement modality and / or measurement process variations. For example, measurements that result in the determination of training samples for off-diagonal Mueller matrix elements can also be used to determine layer thickness. In another instance, layer thickness measurements can be provided by another instrument, such as via a scanning electron microscope (SEM) device.

[0046] In an embodiment, a pMΔ value associated with a process variation (PV) already determined to be substantially related to an increase in TIS is also evaluated to determine whether said value is substantially uncorrelated with OVL. For example, since the OVL and TIS spectra are caused by different physical asymmetries, it is expected that at least one of the pMΔ values ​​of the six off-diagonal Mueller element pairs is uncorrelated between the TIS and OVL spectra. For example, and as Figures 2A to 2B As shown in the figure, when calculating TIS, the M of the spectrum varies along the process of the sample. 20 -M 02The value differs from that when calculating the superposition. For example, although Figure 2A M in 20 -M 02 The TIS spectrum shows peaks at approximately 365 nm and 440 nm, but Figure 2A M in 20 -M 02 The OVL spectrum exhibits a peak band from 190 nm to approximately 265 nm. A TIS signature can then be generated, focusing on the removal of TIS, where the difference between the TIS and OVL values ​​is significant. By determining the correlation between process variations and TIS, measurements of these variations (e.g., layer thickness) can then be used, along with the TIS signature, to more accurately determine the magnitude of the TIS correction to be applied.

[0047] One or more off-diagonal Mueller matrix element pairs that have been identified as having low OVL sensitivity and high TIS sensitivity can then be used for training and testing purposes. For example, training and testing may include one or more off-diagonal Mueller matrix element pairs, two or more off-diagonal Mueller matrix element pairs, three or more off-diagonal Mueller matrix element pairs, four or more off-diagonal Mueller matrix element pairs, five or more off-diagonal Mueller matrix element pairs, or may include all six off-diagonal Mueller matrix element pairs.

[0048] In an embodiment, spectral measurements of training samples are used to train a TIS model capable of recognizing or inferring a TIS signature generated by metrology subsystem 102. Once the TIS signature is recognized, a formulation for operating metrology subsystem 102 is generated, instructing it to measure and derive superimposed spectra from test sample 104 and remove the TIS signature from the derived spectra (e.g., from a single azimuth measurement). For example, and referring to Figures 2A to 2B Because TIS and OVL spectra differ at different wavelengths (e.g., varying depending on process variations), the TIS model can generate a TIS signature, which can be used to selectively remove TIS to correct errors in OVL measurements. Therefore, the TIS signature is not a general or static control for removing TIS, but rather a custom signature generated for each sample during runtime.

[0049] TIS models can be generated or trained through one or more of the following: mathematical modeling, statistical modeling (e.g., general linear modeling or time series modeling), simulation modeling (e.g., system dynamics modeling or Monte Carlo simulation), empirical modeling, heuristic modeling, and artificial intelligence / machine learning modeling.

[0050] Figure 3This illustration depicts a process flow diagram of a method 300 for removing a signature from a metrological measurement (TIS) according to one or more embodiments of this disclosure. One or more steps of method 300 may be performed by a semiconductor device manufacturing system 100 and / or a metrology subsystem 102 as described herein. For example, one or more steps of the method may be performed by one or more of the spectral subsystems and tools described herein (e.g., a spectroscopic ellipsometer).

[0051] In embodiments, according to one or more embodiments of this disclosure, method 300 includes the step of generating (e.g., training) a TIS model from a spectrometry subsystem. In an embodiment, generating the TIS model includes a step 302 of receiving training data from metrological measurements of the training samples, the training data including spectral data associated with at least one off-diagonal Mueller matrix element generated by one or more first measurements of the training samples at a first azimuth angle and one or more second measurements of the training samples at a second azimuth angle. For example, metrology subsystem 102 may be used to collect spectra from both the first and second azimuth angles (e.g., 0° and 180°). Up to six off-diagonal Mueller matrix pairs, wherein one or more off-diagonal Mueller matrix pairs are associated with the TIS, may be determined based on the spectra generated under certain illumination conditions. The one or more off-diagonal Mueller matrix pairs associated with the TIS are then included as part of the training data received by one or more processors 108. For example, the values ​​of all six off-diagonal Mueller matrix pairs may be included in the training data.

[0052] In an embodiment, generating the TIS model further includes step 304 of deriving superimposed spectral data and TIS spectral data from the training data. For example, OVL spectral data can be derived from Equation 2 for one or more of the six off-diagonal Mueller matrix pairs. In another instance, TIS spectral data can be derived from Equation 1 for one or more of the six off-diagonal Mueller matrix pairs. In the example where the correlation between the off-diagonal Mueller matrix pairs and TIS is unknown, the spectra of OVL and TIS can be derived for all six off-diagonal Mueller element pairs. In the example where the correlation between the off-diagonal Mueller matrix pairs and TIS is known, uncorrelated off-diagonal Mueller matrix pairs can be excluded from the calculation.

[0053] In an embodiment, generating a TIS model further includes step 306 of decomposing the superimposed spectral data and the TIS spectral data. For example, generating a TIS model may include performing spectral decomposition, also known as eigenvalue decomposition, which decomposes the Mueller matrix into its normal form using eigenvectors and eigenvalues. It is known that the Mueller matrix can be decomposed into multiple factors using several methods.

[0054] In an embodiment, generating a TIS model further includes step 308 of inferring or identifying a TIS signature of a training sample based on measurement data, superimposed spectral data, and / or a decomposition of TIS spectral data. For example, the TIS model can be trained via a machine learning method that takes training data (e.g., factors generated from the decomposition of OVL spectral data and TIS spectral data) and processes the data to generate a TIS signature that can be used to remove TIS from a test sample.

[0055] In one embodiment, method 300 includes a step for removing a TIS signature from test sample 104. In this embodiment, removing the TIS signature from test sample 104 includes a step 310 of receiving spectral data of a single azimuth angle of the test sample and a step 312 of removing the TIS signature from the spectral data of that single azimuth angle of the test sample. For example, the OVL spectrum of the test sample can measure a single azimuth angle (e.g., 0°). The TIS can then be removed from the OVL spectrum by subtracting the inferred TIS signature generated from the trained model. The inferred TIS signature provides an estimate of the TIS without requiring data from a second azimuth angle (e.g., 0°), thereby reducing the measurement time for the test sample.

[0056] In an embodiment, one or more off-diagonal Mueller matrix pairs included with the training data may include off-diagonal Mueller matrix pairs having pMΔ values ​​that are substantially correlated with changes in the magnitude of TIS. Off-diagonal Mueller matrix pairs that do not have pMΔ values ​​substantially correlated with TIS can be excluded from the training data.

[0057] In an embodiment, one or more off-diagonal Mueller matrix pairs included with the training data may comprise off-diagonal Mueller matrix pairs having pMΔ values ​​substantially correlated with changes in the magnitude of TIS, which in turn are correlated with changes in the magnitude of process variations (e.g., layer thickness or one or more process variations as described herein). For example, different locations in a sample (e.g., a wafer) may be characterized by different layer thicknesses. Once the first and second azimuth measurements of the training samples are completed, it can be determined that one or more of the different layer thicknesses are substantially correlated with TIS, and one or more of the six off-diagonal Mueller matrix pairs derived at these locations may also exhibit a significant correlation with TIS and are therefore included in the training data. During the measurement of the test samples, the spectra generated from these off-diagonal Mueller elements can then be used to determine the magnitude of the TIS signature to be removed from a single azimuth measurement of the test samples.

[0058] Although all six off-diagonal Mueller matrix pairs can be included in the training data, off-diagonal Mueller matrix pairs that do not have a pMΔ value substantially correlated with changes in the magnitude of process variations (e.g., which may also correlate with changes in the magnitude of TIS) can be excluded from the training data. Process variations measured on the samples can be determined from data collected during the first and second azimuth measurements or using other instruments, as described above. The system and method's ability to consider the correlation between the TIS signature and process variations is particularly advantageous than other methods that rely solely on modeling TIS values ​​without considering differences in process variations (e.g., layer thickness) along the samples. Considering the off-diagonal Mueller elemental spectra used for its process variation sensitivity provides an independent indication of the process variation-induced TIS3σ, which provides better information for removing TIS from OVL. In other words, by considering process variations in method 300, method 300 then utilizes process variation-sensitive off-diagonal Mueller matrix elements to reduce TIS3σ.

[0059] In embodiments, one or more off-diagonal Mueller matrix pairs associated with TIS may further include off-diagonal Mueller matrix pairs having pMΔ values ​​that are substantially correlated with changes in the magnitude of TIS and substantially uncorrelated or insensitive to changes in the magnitude of OVL. For example, once the first and second azimuth measurements of the training samples are completed, one or more process variations may be identified as being associated with both relatively large changes in the magnitude of TIS and relatively small changes in the magnitude of OVL indicating that the TIS signature should be removed from OVL. Although all six off-diagonal Mueller matrix pairs may be included in the training data, off-diagonal Mueller matrix pairs that do not have pMΔ values ​​that are substantially correlated with changes in the magnitude of TIS and substantially uncorrelated or insensitive to changes in the magnitude of OVL may be excluded from the training set. By using off-diagonal Mueller matrix elements with low OVL sensitivity but large TIS sensitivity, the method may utilize off-diagonal Mueller matrix elements that are not directly used in OVL derivation. This comparison strategy can be further limited to using off-diagonal Mueller matrix elements that are generally related to both TIS and process variations.

[0060] As used herein, "substantially correlated" can refer to a pMΔ value that shows consistency and / or a substantial correlation with the magnitude of TIS within a wavelength or wavelength band. For example, and as... Figure 2A As shown, when calculating TIS, the pMΔ value (e.g., the dimensionless value corresponding to the dimensionless asymmetry) deviates substantially from zero at approximately 365 nm (e.g., approximately -0.02) and 440 nm (e.g., approximately 0.04). By deviating substantially from zero at 440 nm, the off-diagonal Mueller matrix elements pair with M... 20 -M 02 It can be considered to be largely related to TIS at the wavelength mentioned.

[0061] In the embodiments, "substantially correlated" with the TIS spectrum may mean that the pMΔ value is ±0.01 or more from zero, ±0.02 or more from zero, ±0.04 or more from zero, ±0.08 or more from zero, ±0.12 or more from zero, ±0.24 or more from zero, ±0.48 or more from zero, ±0.96 or more from zero, or ±1.92 or more from zero.

[0062] As used herein, "substantially uncorrelated" can refer to a pMΔ value that does not exhibit consistency and / or a substantial correlation with the magnitude of OVL within a wavelength or wavelength band. For example, and as... Figure 2B As shown, when calculating OVL, the pMΔ value does not deviate significantly from zero in the wavelength band of approximately 270 nm to 440 nm, where the magnitude of the pMΔ value is less than 0.1. By not deviating significantly from zero in said bandwidth, the off-diagonal Mueller matrix elements pair with M... 20 -M 02 It can be considered to be substantially uncorrelated with OVL at the said bandwidth. "Substantially uncorrelated with OVL spectrum" can mean that the pMΔ value of OVL is ±0.25 or less from zero, ±0.50 or less from zero, ±1.0 or less from zero, ±2.0 or less from zero, ±0.40 or less from zero, or ±8.0 or less from zero.

[0063] In an embodiment, one or more off-diagonal Mueller matrix pairs associated with TIS may be included in the TIS pMΔ value (pMΔ). TIS ) and OVL pMΔ value (pMΔ OVL Off-diagonal Mueller matrix pairs with a relatively or generally high ratio between them. For example, referring to Figures 2 and 3, when measuring the sample at 440 nm, for a ratio (pMΔ) of approximately 0.44. TIS :pMΔ OVL ), pMΔ TIS It is approximately 0.04, while pMΔ OVL It is approximately 0.09. In contrast, at 230 nm, the ratio (pMΔ) is approximately 0.003. TIS :pMΔ OVL ), pMΔ TIS It is approximately -0.001, while pMΔ OVL It is approximately 0.3, a difference of a hundredfold. Therefore, one or more off-diagonal Mueller matrix pairs used to train the model can be selected for the wavelength or frequency band of interest in pMΔ. TIS With pMΔ OVL The off-diagonal Mueller matrix pairs that have an increasing ratio between them are used to determine pMΔ. TIS :pMΔ OVLA fairly high ratio may include ratios greater than 0.01, 0.5, 0.1, 0.5 or 1.0.

[0064] The training data used in step 302 for training the TIS model can contain any type of data generated by any type of superposition metrology subsystem. Similarly, the TIS signature generated in step 302 can provide superposition measurements based on the device superposition data as input using any suitable superposition recipe. Any type of machine language modeling technique can be used to determine the TIS signature of the training samples, including but not limited to supervised machine learning techniques, unsupervised machine learning techniques, model-assisted methods, model-free methods, reference-based methods, self-calibrating recipes, or combinations thereof.

[0065] In embodiments, training can be analyzed using various data fitting and optimization techniques (e.g., but not limited to machine learning algorithms such as machine learning libraries, linear machine learning models, neural networks, convolutional networks, or support vector machines (SVM)), dimensionality reduction algorithms (e.g., PCA (principal component analysis), ICA (independent component analysis), or LLE (locally linear embedding)), fast order reduction models, regression, sparse representations (e.g., Fourier transform or wavelet transform), Kalman filters, or algorithms) to facilitate matching from the same or different tool types. Furthermore, statistical model-based econometrics are generally described in U.S. Patent No. 10,101,670 (the entire contents of which are incorporated herein by reference). By another example, device-stacked data can be analyzed using algorithms that do not involve modeling, optimization, and / or fitting, such as patterned wafer characterization generally described in U.S. Patent Publication No. 2015 / 0046121, the entire contents of which are incorporated herein by reference.

[0066] By way of another example, the superposition formulation may include modeling or simulating the optical interaction between the illumination beam 114 and the sample 104 using various techniques, including but not limited to rigorous coupled-wave analysis (RCWA), finite element method (FEM) analysis, method of moments, surface integration, volume integration, or finite-difference time-domain (FDTD) techniques. Furthermore, a geometry engine, a process modeling engine, or a combination thereof may be used to model or parameterize the device target. The use of process modeling is substantially described in U.S. Patent Publication No. 2014 / 0172394, the entire contents of which are incorporated herein by reference.

[0067] In this embodiment, the training data includes beam conditioning data describing the influence of one or more beam conditioning elements 120, 132 on metrological measurements. For example, in a spectroscopic ellipsometer, beam conditioning elements 120, 132, such as those on a focusing mirror, can affect the TIS signature, particularly specific components of the beam conditioning elements 120, 132, such as coatings (e.g., focusing mirror coatings). Parameters or other data known to affect the TIS signature from these beam conditioning elements 120, 132 can be added to the training data to increase model accuracy.

[0068] In embodiments, one or more of the parameters of one or more adjustment elements 120, 132 can be changed to increase or improve the spectral separability between TIS spectral values ​​and OVL spectral values. Spectral values ​​may include, but are not limited to, spectral shape, Mueller matrix elements, pMΔ, main spectral region, or average spectral value (e.g., within a wavelength or wavelength band). The one or more parameters to be changed may include, but are not limited to, mirror coatings. For example, the coating of the focusing mirror of the metrology subsystem 102 may be modified or optimized to increase the separability between TIS spectral values ​​and OVL spectral values. While eliminating TIS by modifying hardware (e.g., adjustment elements 120, 132) may be a long-term goal of spectrometry, increasing or optimizing TIS separability (e.g., by changing the coating parameters of the focusing mirror) appears to be a simpler goal that is easier to achieve.

[0069] One or more processors 108 of controller 106 may comprise any or more processing elements known in the art. In this sense, one or more processors 108 may comprise any microprocessor-type device configured to execute software algorithms and / or instructions. In embodiments, one or more processors 108 may comprise a desktop computer, host computer system, workstation, graphics computer, parallel processor, or other computer system (e.g., a networked computer) configured to execute programs configured to operate system 100, as described throughout this disclosure. It should be appreciated that the steps described throughout this disclosure may be performed by a single computer system or alternatively by multiple computer systems. Generally, the term “processor” may be broadly defined to encompass any device having one or more processing elements that execute program instructions from non-transitory memory media 110. Furthermore, different subsystems of system 100 (e.g., metering subsystem 102) may include processors or logic elements suitable for performing at least a portion of the steps described throughout this disclosure.

[0070] Memory media 110 may comprise any memory medium known in the art suitable for storing program instructions (e.g., instructions configured to cause one or more processors 108 to perform various functions) executable by one or more associated processors 108. For example, memory media 110 may comprise, but is not limited to, read-only memory, random access memory, magnetic or optical storage devices (e.g., disks), magnetic tape, or solid-state drives. In embodiments, memory media 110 is configured to store one or more results from the metering subsystem and / or the outputs of various data processing steps described herein. It should be further noted that memory media 110 may be housed together with one or more processors 108 within a common controller housing. In alternative embodiments, memory media 110 may be remotely located relative to the physical location of the processors and controller 106. For example, one or more processors 108 of controller 106 may access remote memory (e.g., a server) accessible via a network (e.g., the Internet or an internal network).

[0071] Further attention should be paid to, although Figure 1A The controller 106 is depicted as being separate from the metering subsystem 102, but this configuration of system 100 is not a limitation of the scope of this disclosure, but is provided for illustrative purposes only. For example, the controller 106 may be incorporated into the controller of the metering subsystem 102.

[0072] All methods described herein may include storing the results of one or more steps of the method embodiments in a memory medium. The results may include any of the results described herein and may be stored in any manner known in the art. The memory medium may include any memory medium described herein or any other suitable memory medium known in the art. After storing the results, the results may be accessed in the memory medium and used by any of the methods or system embodiments described herein, formatted for display to a user, used by another software module, method, or system, etc. Furthermore, the results may be stored "permanently," "semi-permanently," temporarily, or for a period of time. For example, the memory medium may be random access memory (RAM), and the results may not necessarily be stored indefinitely in the memory medium.

[0073] Furthermore, each of the embodiments of the above methods may include any other steps of any other method described herein. Additionally, each of the embodiments of the above methods may be performed by any of the systems described herein.

[0074] Those skilled in the art will recognize that it is common practice within the art to describe devices and / or processes in the manner set forth herein and subsequently integrate such described devices and / or processes into data processing systems using engineering practice. That is, at least a portion of the devices and / or processes described herein can be integrated into a data processing system through a reasonable amount of experimentation. Those skilled in the art will recognize that a typical data processing system generally includes one or more of the following: a system unit housing, a video display device, memory (e.g., volatile and non-volatile memory), a processor (e.g., a microprocessor and a digital signal processor), computational physics (e.g., an operating system, drivers, a graphical user interface, and applications), one or more interactive devices (e.g., a touchpad or touchscreen), and / or a control system including feedback loops and control motors (e.g., feedback for sensing position and / or speed; control motors for moving and / or adjusting components and / or quantities). Typical data processing systems can be implemented using any suitable commercially available components (e.g., the elements typically found in data computing / communication and / or network computing / communication systems). In addition, one or more processors 108 can provide feedforward information to downstream processing tools and subsystems to adjust the process conditions of the substrate in question as it moves along the semiconductor production line.

[0075] The topics described herein sometimes refer to different components contained within or connected to other components. It should be understood that such depicted architectures are merely exemplary, and in fact, many other architectures can be implemented to achieve the same functionality. Conceptually, any arrangement of components used to achieve the same functionality is effectively “associated” to achieve the desired functionality. Therefore, any two components combined herein to achieve a particular functionality can be considered “associated” with each other to achieve the desired functionality, regardless of the architecture or intermediate components. Similarly, any two such associated components can also be considered “operably connected” or “operably coupled” with each other to achieve the desired functionality, and any two components that can be suchly associated can also be considered “operably coupled” with each other to achieve the desired functionality. Specific examples of operable coupling include, but are not limited to, physically mateable and / or physically interacting components and / or wirelessly interacting and / or logically interacting components.

[0076] Those skilled in the art will understand that, in general, the terms used herein, and especially in the appended claims (e.g., the body of the appended claims), are intended to be “open-ended” terms (e.g., the term “comprising” should be interpreted as “comprising but not limited to,” the term “having” should be interpreted as “having at least,” etc.). Those skilled in the art will further understand that if a particular number of claims is to be introduced, then this intention must be explicitly stated in the claims, and if such a statement is not present, then this intention does not exist. For example, as an aid to understanding, the appended claims may contain the introductory phrases “at least one” and “one or more” to introduce the claims. However, the use of such phrases should not be construed as implying that the introduction of a claims by the indefinite article “a” limits any particular claim containing such an introductory claim to an invention containing only one such claim, even if the same claim contains the introductory phrases “one or more” or “at least one” and indefinite articles such as “a” (e.g., “a” should generally be interpreted as meaning “at least one” or “one or more”); the same applies to the use of definite articles used to introduce the claims. Furthermore, even if a specific number of claims is explicitly stated, those skilled in the art will recognize that this statement should generally be interpreted as meaning at least a certain number of claims (e.g., a bare statement of "two claims" without other modifiers generally means at least two claims or two or more claims). Additionally, in examples where conventions such as "at least one of A, B, and C, etc." are used, this construction is generally intended to have the meaning commonly understood by those skilled in the art (e.g., "a system having at least one of A, B, and C" would include, but is not limited to, systems having only A, only B, only C, both A and B, both A and C, both B and C, and / or both A, B, and C, etc.). In examples where conventions such as "at least one of A, B, or C, etc." are used, this construction is generally intended to have the meaning commonly understood by those skilled in the art (e.g., "a system having at least one of A, B, or C" would include, but is not limited to, systems having only A, only B, only C, both A and B, both A and C, both B and C, and / or both A, B, and C, etc.). Those skilled in the art will further understand that virtually any extractive and / or phrase presenting two or more alternatives, whether in the detailed description, claims, or drawings, should be understood to consider the possibility of including one, either, or both of the terms. For example, the phrase "A or B" should be understood to include the possibility of including "A" or "B" or "A and B".

[0077] While specific aspects of the subject matter of this disclosure have been shown and described herein, those skilled in the art will understand that changes and modifications can be made based on the teachings herein without departing from the subject matter and its broader aspects, and therefore the appended claims are intended to cover all such changes and modifications that fall within the true spirit and scope of the subject matter described herein. Furthermore, it should be understood that this disclosure is defined by the appended claims.

Claims

1. A spectrometric system, comprising: Spectrometric subsystem; and A controller communicatively coupled to the spectrometer subsystem, the controller comprising one or more processors configured to execute program instructions configured to cause the one or more processors to: The tool-induced bias (TIS) model used to generate training samples via the spectrometric subsystem includes: Training data is received from the metrological measurements of the training sample, the training data including spectral data associated with at least one off-diagonal Mueller matrix element generated by one or more first measurements of the training sample at a first azimuth angle and one or more second measurements of the training sample at a second azimuth angle; Derive the superimposed spectral data and TIS spectral data from the training data; Decompose the superimposed spectral data and the TIS spectral data; and Based on the decomposition of the training data, the superimposed spectral data, and the TIS spectral data, the TIS signature of the training sample is inferred; and Removing the TIS signature from the measurement of the test sample includes: Receive spectral data from a single azimuth angle measurement of the test sample; and The TIS signature is removed from the spectral data measured from the single azimuth angle of the test sample based on the TIS model.

2. The spectral metrology system according to claim 1, further comprising: From the training data, determine one or more off-diagonal Mueller matrix elements having spectra that are related to the magnitudes of process variation measurements of the training samples and the TIS spectral data, wherein the TIS signature is based on the spectra from the one or more off-diagonal Mueller matrix elements.

3. The spectral metrology system according to claim 2, wherein the process variation measurement is derived from the metrological measurement.

4. The spectrometer system according to claim 2, wherein the process variation measurement is derived from another metrological measurement.

5. The spectral metrology system according to claim 1, wherein the first azimuth angle and the second azimuth angle are separated by 180 degrees.

6. The spectrometric system according to claim 2, wherein the one or more off-diagonal Mueller matrix elements are substantially correlated with the values ​​of the TIS spectral data and substantially uncorrelated with the values ​​of the superimposed spectral data.

7. The spectrometer system according to claim 2, wherein the process variation measurement includes sample thickness.

8. The spectrometer system according to claim 2, wherein the process variation measurement includes sample tilting.

9. The spectral metrology system according to claim 1, wherein the spectral metrology subsystem comprises at least one of a spectral ellipsometer, a spectral reflectometer, a single-wavelength ellipsometer, a single-wavelength reflectometer, an angle-resolved ellipsometer, or an angle-resolved reflectometer.

10. The spectrometric system of claim 1, wherein the TIS model comprises a trained machine learning model.

11. The spectrometric system of claim 1, wherein the spectrometric subsystem comprises one or more beam adjustment elements, wherein the training data comprises beam adjustment data describing the influence of the one or more beam adjustment elements on the metrological measurement.

12. The spectrometer system of claim 11, wherein the one or more beam adjustment elements include a focusing mirror.

13. The spectrometer system of claim 12, wherein the beam adjustment data includes parameters of the coating of the focusing mirror.

14. A method for removing tool-induced offset (TIS) signatures from metrological measurements, comprising: The TIS model, which generates training samples through a spectrometric subsystem, includes: Training data is received from the metrological measurements of the training sample, the training data including spectral data associated with at least one off-diagonal Mueller matrix element generated by one or more first measurements of the training sample at a first azimuth angle and one or more second measurements of the training sample at a second azimuth angle; Derive the superimposed spectral data and TIS spectral data from the training data; Decompose the superimposed spectral data and the TIS spectral data; and Based on the decomposition of the training data, the superimposed spectral data, and the TIS spectral data, the TIS signature of the training sample is inferred; and Removing the TIS signature from the measurement of the test sample includes: Receive spectral data from a single azimuth angle measurement of the test sample; and The TIS signature is removed from the spectral data of the single azimuth angle measured from the test sample based on the TIS model.

15. The method of claim 14, further comprising: From the training data, determine one or more off-diagonal Mueller matrix elements having spectra that are related to the magnitudes of process variation measurements of the training samples and the TIS spectral data, wherein the TIS signature is based on the spectra from the one or more off-diagonal Mueller matrix elements.

16. The method of claim 14, wherein the first azimuth angle and the second azimuth angle are separated by 180 degrees.

17. The method of claim 15, wherein the one or more off-diagonal Mueller matrix elements are substantially correlated with the magnitudes of the TIS spectral data and substantially uncorrelated with the magnitudes of the superimposed spectral data.

18. The method of claim 15, wherein the process variation measurement includes sample thickness.

19. The method of claim 15, wherein the process variation measurement includes sample tilting.

20. The method of claim 15, wherein the TIS method is performed by at least one of a spectral ellipsometer, a spectral reflectometer, a single-wavelength ellipsometer, a single-wavelength reflectometer, an angle-resolved ellipsometer, or an angle-resolved reflectometer.

21. The method of claim 15, wherein the TIS model comprises a trained machine learning model.

22. The method of claim 14, wherein the spectrometry subsystem comprises one or more beam adjustment elements, wherein the training data comprises beam adjustment data describing the effect of the one or more beam adjustment elements on the metrological measurement.

23. The method of claim 22, wherein the one or more beam adjustment elements include a focusing mirror.

24. The method of claim 23, wherein the beam adjustment data includes parameters of the coating of the focusing mirror.

25. The method of claim 22, further comprising changing one or more parameters of the one or more beam adjustment elements to increase the spectral separability between the TIS spectral values ​​and the superimposed spectral values.

26. The method of claim 25, wherein the one or more beam adjustment elements include a focusing mirror.

27. The method of claim 26, wherein the one or more parameters include a focusing mirror coating.

Images