Machine Learning-Driven Optimization of Magnetoplasmonic Biosensors

Abstract

The Bernhard Gross Polymer Group at the São Carlos Institute of Physics (IFSC/USP) has been developing biosensors for many years, for which materials and architectures are empirically selected based on previous experimental data. This research project aims to design sensitive optical sensors, taking advantage of the synergistic interaction of localized surface plasmon resonance (LSPR) and magneto-optical effects, with optimization based on machine learning (ML). By using LSPR on ferromagnets, ie magnetoplasmonic nanoparticles, enhanced light-matter interactions can be achieved, which is important for sensing. The integration of magnetoplasmonic effects expands the capabilities of plasmonic systems, allowing new sensor functionalities. Combining ML techniques such as neural networks and deep learning with plasmonics-based detection technologies may revolutionize optical and photonic devices. ML optimization can improve sensor design, increase sensitivity, and allow one to extract valuable information from datasets. The optical sensors foreseen in the project can be used in health, environmental monitoring and industrial applications. The expected results include the design of sensors with exceptional sensitivity and a high Figure of Merit, based on optimization using ML algorithms.