Abstract: Methods and systems for improved monitoring of tool drift and tool-to-tool matching across large fleets of measurement systems employed to measure semiconductor structures are presented herein. One or more Quality Control (QC) wafers are measured by each of a fleet of measurement systems. Values of system variables are extracted from the QC measurement data associated with each measurement system using a trained QC encoder. The extracted values of the system variables are employed to condition the corresponding measurement model employed by each measurement tool to characterize structures under measurement having unknown values of one or more parameters of interest. Accurate tool-to-tool matching across a fleet of conditioned measurement systems is achieved by extracting values of system variables from measurement data collected from the same set of QC wafers. Tool health is monitored based on changes in values of system variables extracted from measurements performed at different times.

Abstract: Methods and systems for generating optimized geometric models of semiconductor structures parameterized by a set of variables in a latent mathematical space are presented herein. Reference shape profiles characterize the shape of a semiconductor structure of interest over a process space. A set of observable geometric variables describing the reference shape profiles is transformed to a set of latent variables. The number of latent variables is smaller than the number of observable geometric variables, thus the dimension of the parameter space employed to characterize the structure of interest is reduced. This dramatically reduces the mathematical dimension of the measurement problem to be solved. As a result, measurement model solutions involving regression are more robust, and training of machine learning based measurement models is simplified.

Abstract: Methods and systems for measuring semiconductor structures based on a trained scanning conditional measurement model are described herein. A scanning conditional model is trained based on Design Of Experiments (DOE) measurement data associated with known values of one or more parameters of interest and a set of perturbed values of the one or more parameters of interest. The trained conditional model minimizes the output of an error function characterizing the error between the known values of the perturbed values of the one or more parameters of interest for the given DOE measurement data. During inference, an error value associated with each candidate value of one or more parameters of interest is determined by the trained scanning conditional measurement model. The estimated value of the parameter of interest is the candidate value of the parameter of interest associated with the minimum error value.

Abstract: Methods and systems for improved regularization associated with tomographically resolved image based measurements of semiconductor structures are presented herein. The regularizations described herein are based on measurement data and parameterization of a constrained voxel model that captures known process variations. The constrained voxel model is determined based on simplified geometric models, process models, or both, characterizing the structure under measurement. A constrained voxel model has dramatically fewer degrees of freedom compared to an unconstrained voxel model. The value associated with each voxel of the constrained voxel model depends on a relatively small number of independent variables. Selection of the independent variables is informed by knowledge of the structure and the underlying fabrication process. Regularization based on a constrained voxel model enables faster convergence and a more accurate reconstruction of the measured structure with less computational effort.

Abstract: Methods and systems for measurement of wafer tilt and overlay are described herein. In some embodiments, the measurements are based on the value of an asymmetry response metric and known wafer statistics. Spectral measurements are performed at two different azimuth angles, preferably separated by one hundred eighty degrees. A sub-range of wavelengths is selected with significant signal sensitivity to wafer tilt or overlay. An asymmetry response metric is determined based on a difference between the spectral signals measured at the two different azimuth angles within the selected sub-range of wavelengths. The value of the asymmetry response metric is mapped to an estimated value of wafer tilt or overlay. In some other embodiments, the measurement of wafer tilt or overlay is based on a trained measurement model. Training data may be programmed or determined based on one or more asymmetry response metrics at two different azimuth angles.

Abstract: Methods and systems for estimating a value of a quality metric indicative of one or more performance characteristics of a semiconductor measurement are presented herein. The value of the quality metric is normalized to ensure applicability across a broad range of measurement scenarios. In some embodiments, a value of a quality metric is determined for each measurement sample during measurement inference. In some embodiments, a trained quality metric model is employed to determine the uncertainty of defect classification. In some embodiments, a trained quality metric model is employed to determine the uncertainty of estimated parameters of interest, such as geometric, dispersion, process, and electrical parameters. In some examples, a quality metric is employed as a filter to detect measurement outliers. In some other examples, a quality metric is employed as a trigger to adjust a semiconductor process.

Abstract: Methods and systems for training and implementing metrology recipes based on performance metrics employed to quantitatively characterize the measurement performance of a metrology system in a particular measurement application. Performance metrics are employed to regularize the optimization process employed during measurement model training, model-based regression, or both. For example, the known distributions associated with important measurement performance metrics such as measurement precision, wafer mean, etc., are specifically employed to regularize the optimization that drives measurement model training. In a further aspect, a trained measurement model is employed to estimate values of parameters of interest based on measurements of structures having unknown values of one or more parameters of interest. In a further aspect, trained measurement model performance is validated with test data using error budget analysis.

Abstract: Methods and systems for measurements of semiconductor structures based on a trained parameter conditioned measurement model are described herein. The shape of a measured structure is characterized by a geometric model parameterized by one or more conditioning parameters and one or more non-conditioning parameters. A trained parameter conditioned measurement model predicts a set of values of each non-conditioning parameter based on measurement data and a corresponding set of predetermined values for each conditioning parameter. In this manner, the trained parameter conditioned measurement model predicts the shape of a measured structure. Although a parameter conditioned measurement model is trained at discrete geometric points of a structure, the trained model predicts values of non-conditioning parameters for any corresponding conditioning parameter value.

Abstract: Methods and systems for estimating values of process parameters based on measurements of structures fabricated on a product wafer are presented herein. Exemplary process parameters include lithography dosage and exposure and lithography scanner aberrations. A measurement model is employed to estimate process parameter values from measurements of structures fabricated on a wafer by a particular fabrication process. The measurement model includes process parameters and geometric parameters of structures under measurement. In some embodiments, a model based regression of both a process model and a metrology model is employed to arrive at estimates of at least one process parameter value based on measurements of a fabricated structure. In some embodiments, a trained measurement model is employed to directly estimate process parameter values based on measurements of structures.

Abstract: Methods and systems for training and implementing metrology recipes while dynamically controlling the convergence trajectories of multiple performance objectives are described herein. Performance metrics are employed to regularize the optimization process employed during measurement model training, model-based regression, or both. Weighting values associated with each of the performance objectives in the loss function of the model optimization are dynamically controlled during model training. In this manner, convergence of each performance objective and the tradeoff between multiple performance objectives of the loss function is controlled to arrive at a trained measurement model in a stable, balanced manner. A trained measurement model is employed to estimate values of parameters of interest based on measurements of structures having unknown values of one or more parameters of interest.

Abstract: Methods and systems for estimating a value of a quality metric indicative of one or more performance characteristics of a semiconductor measurement are presented herein. The value of the quality metric is normalized to ensure applicability across a broad range of measurement scenarios. In some embodiments, a value of a quality metric is determined for each measurement sample during measurement inference. In some embodiments, a trained quality metric model is employed to determine the uncertainty of defect classification. In some embodiments, a trained quality metric model is employed to determine the uncertainty of estimated parameters of interest, such as geometric, dispersion, process, and electrical parameters. In some examples, a quality metric is employed as a filter to detect measurement outliers. In some other examples, a quality metric is employed as a trigger to adjust a semiconductor process.

Abstract: Methods and systems for combining information present in measured images of semiconductor wafers with additional measurements of particular structures within the measured images are presented herein. In one aspect, an image-based signal response metrology (SRM) model is trained based on measured images and corresponding reference measurements of particular structures within each image. The trained, image-based SRM model is then used to calculate values of one or more parameters of interest directly from measured image data collected from other wafers. In another aspect, a measurement signal synthesis model is trained based on measured images and corresponding measurement signals generated by measurements of particular structures within each image by a non-imaging measurement technique.

Abstract: Methods and systems for training and implementing metrology recipes based on performance metrics employed to quantitatively characterize the measurement performance of a metrology system in a particular measurement application. Performance metrics are employed to regularize the optimization process employed during measurement model training, model-based regression, or both. For example, the known distributions associated with important measurement performance metrics such as measurement precision, wafer mean, etc., are specifically employed to regularize the optimization that drives measurement model training. In a further aspect, a trained measurement model is employed to estimate values of parameters of interest based on measurements of structures having unknown values of one or more parameters of interest. In a further aspect, trained measurement model performance is validated with test data using error budget analysis.

Abstract: Methods and systems for improved regularization associated with tomographically resolved image based measurements of semiconductor structures are presented herein. The regularizations described herein are based on measurement data and parameterization of a constrained voxel model that captures known process variations. The constrained voxel model is determined based on simplified geometric models, process models, or both, characterizing the structure under measurement. A constrained voxel model has dramatically fewer degrees of freedom compared to an unconstrained voxel model. The value associated with each voxel of the constrained voxel model depends on a relatively small number of independent variables. Selection of the independent variables is informed by knowledge of the structure and the underlying fabrication process. Regularization based on a constrained voxel model enables faster convergence and a more accurate reconstruction of the measured structure with less computational effort.

Abstract: Disclosed are apparatus and methods for determining process or structure parameters for semiconductor structures. A plurality of optical signals is acquired from one or more targets located in a plurality of fields on a semiconductor wafer. The fields are associated with different process parameters for fabricating the one or more targets, and the acquired optical signals contain information regarding a parameter of interest (POI) for a top structure and information regarding one or more underlayer parameters for one or more underlayers formed below such top structure. A feature extraction model is generated to extract a plurality of feature signals from such acquired optical signals so that the feature signals contain information for the POI and exclude information for the underlayer parameters. A POI value for each top structure of each field is determined based on the feature signals extracted by the feature extraction model.

Abstract: Methods and systems for performing semiconductor measurements based on hyperspectral imaging are presented herein. A hyperspectral imaging system images a wafer over a large field of view with high pixel density over a broad range of wavelengths. Image signals collected from a measurement area are detected at a number of pixels. The detected image signals from each pixel are spectrally analyzed separately. In some embodiments, the illumination and collection optics of a hyperspectral imaging system include fiber optical elements to direct illumination light from the illumination source to the measurement area on the surface of the specimen under measurement and fiber optical elements to image the measurement area. In another aspect, a fiber optics collector includes an image pixel mapper that couples a two dimensional array of collection fiber optical elements into a one dimensional array of pixels at the spectrometer and the hyperspectral detector.

Abstract: Methods and systems for performing measurements based on a measurement model integrating a metrology-based target model with a process-based target model. Systems employing integrated measurement models may be used to measure structural and material characteristics of one or more targets and may also be used to measure process parameter values. A process-based target model may be integrated with a metrology-based target model in a number of different ways. In some examples, constraints on ranges of values of metrology model parameters are determined based on the process-based target model. In some other examples, the integrated measurement model includes the metrology-based target model constrained by the process-based target model. In some other examples, one or more metrology model parameters are expressed in terms of other metrology model parameters based on the process model. In some other examples, process parameters are substituted into the metrology model.

Abstract: Methods and systems for robust overlay error measurement based on a trained measurement model are described herein. The measurement model is trained from raw scatterometry data collected from Design of Experiments (DOE) wafers by a scatterometry based overlay metrology system. Each measurement site includes one or more metrology targets fabricated with programmed overlay variations and known process variations. Each measurement site is measured with known metrology system variations. In this manner, the measurement model is trained to separate actual overlay from process variations and metrology system variations which affect the overlay measurement. As a result, an estimate of actual overlay by the trained measurement model is robust to process variations and metrology system variations. The measurement model is trained based on scatterometry data collected from the same metrology system used to perform measurements. Thus, the measurement model is not sensitive to systematic errors, aysmmetries, etc.

Abstract: Methods and systems for performing semiconductor measurements based on hyperspectral imaging are presented herein. A hyperspectral imaging system images a wafer over a large field of view with high pixel density over a broad range of wavelengths. Image signals collected from a measurement area are detected at a number of pixels. The detected image signals from each pixel are spectrally analyzed separately. In some embodiments, the illumination and collection optics of a hyperspectral imaging system include fiber optical elements to direct illumination light from the illumination source to the measurement area on the surface of the specimen under measurement and fiber optical elements to image the measurement area. In another aspect, a fiber optics collector includes an image pixel mapper that couples a two dimensional array of collection fiber optical elements into a one dimensional array of pixels at the spectrometer and the hyperspectral detector.

Abstract: Methods and systems for evaluating the performance of multiple patterning processes are presented. Patterned structures are measured and one or more parameter values characterizing geometric errors induced by the multiple patterning process are determined. In some examples, a single patterned target and a multiple patterned target are measured, the collected data fit to a combined measurement model, and the value of a structural parameter indicative of a geometric error induced by the multiple patterning process is determined based on the fit. In some other examples, light having a diffraction order different from zero is collected and analyzed to determine the value of a structural parameter that is indicative of a geometric error induced by a multiple patterning process. In some embodiments, a single diffraction order different from zero is collected. In some examples, a metrology target is designed to enhance light diffracted at an order different from zero.