Introduction and application of a new approach for model-based optical bidirectional measurements

Accurate measurements of micro- and nanoscale features in optical microscopy demand comprehensive modelling approaches. In this study, we introduce an enhanced evaluation method, utilizing rigorous simulations based on a finite element method algorithm within an advanced Bayesian optimization framework. We provide an in-depth explanation of the measurement process, including the dimension reduction techniques applied to the acquired measurement data. Additionally, we employ Hopkins’ approximation or also referred to as local Hopkins’ methods for an efficient microscopic image simulation, resulting in a significant reduction of the computing time. We applied this method to measure the linewidths of six different chrome lines, nominally 300 nm–1000 nm wide, on a glass substrate. Our results show an excellent agreement with previous investigations conducted using various measurement systems, including atomic force microscopy, scanning electron microscopy, and optical microscopy in combination with different measurement evaluation techniques.