Terahertz time-domain spectroscopy for characterisation of photoconductive layers and non-linear soft matter systems

Abstract

This thesis explores techniques based on machine learning and hybrid metamaterial-liquid crystal systems to enhance terahertz time-domain spectroscopy (THz-TDS) for the probing of complex optical properties and study of non-linear phenomena. First, the use of artificial neural networks to extract the complex conductivity from THz spectra of large-area graphene monolayers is proposed and demonstrated. By combining synthetic and experimentally acquired data into the network training sets, improved performance is demonstrated on out-of-distribution data when compared to traditional techniques and singularly trained networks. Second, non-linear optical behaviour of nematic liquid crystals integrated with two dimensional metasurfaces is investigated. A robust theoretical framework of all-optical switching induced by the electric field of a conventional low-power oscillator-based THz-TDS system is described. This analysis aims to provide a robust theoretical framework to recent observations of anomalous resonance shifts in liquid crystal-loaded metamaterials when subjected to terahertz illumination. The final part of this thesis describes the first direct experimental evidence of liquid crystal reorientation driven by polarised THz radiation. This effect was observed using a purpose built THz-pump optical-probe setup. The reorientation, facilitated by resonant field amplification of a planar metasurface, demonstrates a path toward strong optical nonlinearity in the terahertz regime and the development of novel all-optical active THz devices.

More information

Identifiers

ORCID for Benjamin Beddoes: orcid.org/0000-0003-1577-5362

ORCID for Vasileios Apostolopoulos: orcid.org/0000-0003-3733-2191

Catalogue record

Export record