Artificial Neural Network Modelling for Optimizing the Optical Parameters of Plasmonic Paired Nanostructures

Area of Science:

• Optics and Material Science
• Computational Physics

Background:

• Artificial Neural Networks (ANNs) are increasingly used in Machine Learning (ML) for complex data analysis.
• ANNs offer time-efficient solutions popular in physics, optics, and material science.

Purpose of the Study:

• To present a novel, computationally efficient Artificial Neural Network (ANN) approach for designing and optimizing electromagnetic plasmonic nanostructures.
• To predict key performance metrics of nanostructures, including sensitivity (S), Full Width Half Maximum (FWHM), Figure of Merit (FOM), and Plasmonic Wavelength (PW).

Main Methods:

• Developed a computational model using the Finite Element Method (FEM) to generate a dataset.
• Optimized the number of hidden layers and neurons in the ANN for improved ML model efficiency.
• Utilized the trained ANN to predict nanostructure performance based on input parameters (Major axis 'a', Minor axis 'b', separation gap 'g').

Main Results:

• The ANN model accurately predicted sensitivity, FWHM, FOM, and plasmonic wavelength.
• Compared predicted results with simulated data, demonstrating a low error margin.
• The ANN approach showed superior performance over direct numerical simulation methods for output prediction.

Conclusions:

• The proposed ANN-based method provides a computationally efficient alternative for designing and optimizing plasmonic nanostructures.
• This AI-driven approach accelerates the prediction of nanostructure performance, aiding material science research.