On the advancement of high-yield hollow core anti-resonant optical fibres for the transmission of broadband UV-visible light

Abstract

Hollow core Anti‐Resonant optical Fibres (ARFs) confine light within an air‐filled core by virtue of their microstructure rather than total internal reflection. Such fibres have recently been demonstrated with great success in the near infra-red (near-IR) wavelength range, however, ultra-violet (UV)-visible wavelength guidance has not yet received as much attention. Development of near-IR guiding ARFs has mainly been driven by high value applications such as telecoms. There are many existing and emerging applications that would benefit from ARFs that guide in the UV-visible, including sensing, quantum computing, quantum communication, high-value manufacture, and high precision manufacture. As light guidance in the air-filled core of an ARF has the potential to overcome many of the issues encountered when UV-visible light is guided in a traditional solid-core fibre, such as increased Rayleigh scattering, the requirement of a smaller core (compared to near-IR guidance), and the potential for photo-darkening, it makes UV-visible guiding ARFs highly desirable. This thesis is concerned with the design and fabrication of ARFs with thin ( Based on this result, a new approach to thin-membraned fibre fabrication is developed and reported, whereby the traditional two stage ("stack and draw") process that has been used to fabricate ARFs is extended to three stages (“stack, draw, and draw”), such that the inherent weakness with scaling to lower draw yield - the need to draw very small diameter fibre preforms, can be resolved. This new approach is explored numerically identifying several new constraints and potential fibre designs. The work is then experimentally validated by the fabrication of multiple fibres with a single 350 m long fibre band, with ∼167 nm thick membranes, being selected for detailed analysis. As the design space for ARFs is potentially extremely large, identification of geometries that may offer superior optical performance can be highly challenging. Simulation is the obvious choice but, due to the large parameter space, simulation via traditional parametric sweeps is computationally expensive; especially at shorter (UV-visible) wavelengths. Gaussian Process Regression (GPR), a machine learning technique, is demonstrated to be an efficient method to map this large ARF design space. Combined with Bayesian Optimization (BO), a sequential model‐based optimization routine is developed. Using this approach, a large parameter space is explored and a potentially "optimal" fibre design for broadband UV-visible light transmission is proposed; expected to be capable of ∼7 dB/km loss at 273 nm. The SARS-CoV-2 pandemic influenced the research presented in this thesis, leading to modifications of the original planned objectives. Instead of participating directly with optical fibre fabrication, my involvement focused on assistance via simulations.

More information

Identifiers

ORCID for Gregory Johannes Norman Jackson: orcid.org/0009-0006-5506-2323

ORCID for Francesco Poletti: orcid.org/0000-0002-1000-3083

ORCID for Gregory Jasion: orcid.org/0000-0001-5030-6479

Catalogue record

Export record