Machine-learning driven analysis of an Aluminum-based Plasmonic sensor in the Near-infrared region

---

Abstract

We present a multilayer SPR-based sensor capable of detecting analyte-induced changes in refractive index. The design incorporates an Aluminum-based prism with calcium fluoride, Silicon as the dielectric layer, and Fluorinated Graphene as the biorecognition element. The sensor is analyzed across the near-infrared region using the transfer matrix method, evaluating key parameters such as sensitivity (°/RIU) and figure of merit (1/RIU). Various machine learning regression models are applied to the curated dataset, compared using R² scores, and feature importance is investigated to provide new design insights. Testing on unseen inputs demonstrates that ML-based approaches can efficiently optimize sensor performance compared to conventional analytical methods.

---

Keywords

Aluminium, plasmonic, machine learning, sensors, XgBoost, Random Forest, ANN