Abstract: Methods and systems for estimating values of parameters of interest from X-ray scatterometry measurements with reduced computational effort are described herein. Values of parameters of interest are estimated by regression using a trained, machine learning (ML) based electromagnetic (EM) response model. A training data set includes sets of Design Of Experiments (DOE) values of parameters of interest and corresponding DOE values of a plurality of electromagnetic response metrics. In some examples, values of parameters of interest are determined from measured images based on regression using a sequence of trained ML based electromagnetic response models. In some examples, input values employed to train the ML based EM response model are scaled based on model output variation.

Abstract: Methods and systems for optimizing a semiconductor measurement recipe that is robust to variations of hardware modeling parameters and geometric modeling errors are described herein. Robust measurement recipe optimization minimizes a cost function including one or more regularization terms that constrain the process space, and thus, significantly reduces the computational effort required to optimize a measurement recipe. This reduces overall process time and improves wafer throughput. In some examples, optimization is performed based on measurement data associated with multiple instances of a semiconductor structure; each instance characterized a different value of one or more geometric parameters of interest. In some examples, the search for optimized measurement recipes is limited to the discrete set of measurement system parameter values associated with the available measurement data set. In this manner, the performance of a particular measurement recipe is validated using existing measurement data.

Abstract: Methods and systems for estimating values of parameters of interest from X-ray scatterometry measurements with reduced computational effort are described herein. Values of parameters of interest are estimated by regression using a trained, machine learning (ML) based electromagnetic (EM) response model. A training data set includes sets of Design Of Experiments (DOE) values of parameters of interest and corresponding DOE values of a plurality of electromagnetic response metrics. In some examples, values of parameters of interest are determined from measured images based on regression using a sequence of trained ML based electromagnetic response models. In some examples, input values employed to train the ML based EM response model are scaled based on model output variation.