Jordan Moshcovitis University of Melbourne, School of Physics
This thesis investigates the design and optimization of nano-textured diamond moth-eye surfaces for high-transmission anti-reflective coatings in the near-infrared regime. By combining computational electromagnetic modeling with machine learning techniques, the work explores how subwavelength nanostructures on diamond substrates can suppress Fresnel reflections across broadband spectral ranges — and how fabrication defects impact optical performance.
- Metasurface design — subwavelength moth-eye geometries on single-crystal diamond
- Computational physics — rigorous coupled-wave analysis (RCWA) and finite-element electromagnetic simulations
- Machine learning for photonics — surrogate models and optimization of nanostructure parameters
- Anti-reflective coatings — broadband near-IR transmission enhancement
- Fabrication tolerance — quantifying the impact of structural defects on optical performance
The research employed physics-based electromagnetic solvers to generate large parametric datasets of nanostructure geometries and their optical responses. Machine learning models were then trained on these datasets to accelerate the inverse design process — mapping desired spectral properties back to optimal geometric parameters. This combination of physical simulation and data-driven modeling demonstrates how computational and ML methods can jointly tackle complex photonic design problems.
The full thesis is available in this repository as a PDF.
All rights reserved. This work is shared for academic reference only.