Nanometrology with structured light and artificial intelligence

Abstract

This thesis addresses the fundamental challenge of overcoming the diffraction limit in optical metrology, aiming to achieve high-precision, sub-wavelength measurements of the position and size of nanostructures. Traditional optical imaging methods are constrained by the diffraction limit, which restricts the resolution achievable with conventional techniques. To address this limitation, this research explores the use of structured light and artificial intelligence (AI) to develop novel optical nanometrology methods that enable non-contact, real-time, and highprecision measurements of nanoscale objects.
The primary focus of this research is the development of a set of metrology techniques leveraging structured light—specifically superoscillatory, Laguerre-Gaussian (LG), and Hermite-Gaussian (HG) beams—along with machine learning to analyze scattered/reflected patterns and retrieve high-resolution information about the object. These innovations lead to precise measurements of the position and size of sub-wavelength objects beyond the diffraction limit.
Key achievements of this work include the theoretical and experimental demonstration of 1D positional nanometrology, where a single sub-wavelength slit was measured with a random statistical error of 7.2 nm (λ/110). The research extends to 3D measurements, where a 100 nm nanoparticle was located with a precision of 5.3 nm (λ/91), far exceeding the conventional diffraction limit. Additionally, a novel 2D size metrology technique was developed, achieving an experimental precision of 18 nm (λ/27), with simulations predicting a potential precision of 6 nm (λ/81).
The integration of AI into these optical techniques significantly enhances the accuracy, efficiency, and scope of nanometrology, enabling highly detailed measurements without the need for fluorescent labeling or complex sample preparation. This approach has the potential to revolutionize fields such as semiconductor manufacturing, biomedicine, intelligent manufacturing, quality control, and advanced materials characterization, where precise and non-invasive nanoscale measurements are essential.

More information

Identifiers

ORCID for Bruce (Jun-Yu) Ou: orcid.org/0000-0001-8028-6130

Catalogue record

Export record

Altmetrics