City Research Online

Evaluation of Photonic Characteristics of Plasmonic Integrated Metallic Nanoparticles with the help of Artificial Neural Network Parameterisation

Verma, S. (2023). Evaluation of Photonic Characteristics of Plasmonic Integrated Metallic Nanoparticles with the help of Artificial Neural Network Parameterisation. (Unpublished Doctoral thesis, City, University of London)


Plasmonic Nanostructures with its highly Localised Surface Plasmon Resonance (LSPRs), have opened up excellent opportunities for molecular biosensing applications. This PhD work studies a novel elliptical shaped gold nano antenna array surface as a sensing platform for Refractive Index (RI) diagnostics by using the finite element method (FEM) of COMSOL Multiphysics package. In this work, initially various computational approaches for characterising nanoantennas are benchmarked. Then, effect of various nano antenna parameters are optimised to achieve a high sensitivity and uniformity of the sensor chips. It has been shown that nanoantenna array with major axes (a) = 100 nm, minor axes (b) = 10 nm, height (h) = 40 nm and separation gap (g) = 10 nm with unit cell period of 400x200 nm yielding a very high sensitivity of 526-530 nm/RIU, FWHM = 110 nm and FOM = 8.1. Next, a hybrid coupled nano-structured antenna with stacked multilayer gold and Lithium Tantalate (LiTaO3) or Aluminum Oxide (Al2O3) is designed. A 10 layers of gold (Au) and Lithium tantalate (LiTaO3) or Aluminum oxide (Al2O3) with h1 = h2 = 10 nm exhibits very high bulk sensitivity (S) of 730 and 660 nm/RIU, respectively with major axis, (a) = 100 nm, minor axis, (b) = 10 nm, separation gap (g) = 10 nm, and height, (h) = 100 nm, which is a significant increase in its sensitivity (S). This innovative novel plasmonic hybrid nanostructures provide a framework for developing plasmonic nanostructures for use in various sensing applications. Additionally, as an alternative to the use of computationally expensive FEM, use of multi layer perception (MLP) deep learning method is developed with the help of Pytorch and scikt learn frameworks. The training of MLP model has been carried out with the help geometrical data as a input layer and predicted the sensitivity, Full-Width Half-Maximun (FWHM), Figure of Merit (FOM), plasmonic wavelength and the spectral patterns of reflection, transmission and absorption spectra. The efficacy and reliability of the design strategy are confirmed through conventional FEM validations and evaluation shows that over 95 % accuracy and 40 time faster computational cost as compare to the conventional FEM method.

Publication Type: Thesis (Doctoral)
Subjects: Q Science > QA Mathematics > QA75 Electronic computers. Computer science
Q Science > QC Physics
Departments: School of Science & Technology
School of Science & Technology > School of Science & Technology Doctoral Theses
Doctoral Theses
[thumbnail of Verma thesis 2023 PDF-A.pdf]
Text - Accepted Version
Download (18MB) | Preview


Add to AnyAdd to TwitterAdd to FacebookAdd to LinkedinAdd to PinterestAdd to Email


Downloads per month over past year

View more statistics

Actions (login required)

Admin Login Admin Login