9th International Conference on Body Area Networks

Research Article

Modeling LSPR nano-particles by using neural networks

  • @INPROCEEDINGS{10.4108/icst.bodynets.2014.257112,
        author={Daryoush Mortazavi and Abbas Kouzani and Ladislau Matekovits},
        title={Modeling LSPR nano-particles by using neural networks},
        proceedings={9th International Conference on Body Area Networks},
        keywords={artificial neural network optimization localized surface plasmon resonance numerical method},
  • Daryoush Mortazavi
    Abbas Kouzani
    Ladislau Matekovits
    Year: 2014
    Modeling LSPR nano-particles by using neural networks
    DOI: 10.4108/icst.bodynets.2014.257112
Daryoush Mortazavi1, Abbas Kouzani1, Ladislau Matekovits2,*
  • 1: School of Engineering, Deakin University, Geelong, Victoria 3216, Australia
  • 2: Dept. of Electronics and Telecomunications, Politecnico di Torino, C.so Duca degli Abruzzi 24, 10129 - Torino, Italy
*Contact email: ladislau.matekovits@polito.it


Localized surface plasmon resonance (LSPR) biosensors represent a relatively new and hot research topic in biosensing applications. Since the fabrication of LSPR biosensors is time consuming and costly, providing a mathematical model that can predict the LSPR characteristics before any fabrication is on edge. Implementing such a model for the LSPR devices, and then optimally designing the LSPR geometrical parameters for a particular surface enhanced Raman Scattering (SERS) biosensor function is the concept that has not been explored yet. In this paper, a multi layered artificial neural network (ANN) is proposed which produces a mathematical model representing the characteristics of LSPR devices as a function of their physical dimensions for a specific shape of nano-particles. Such a model can be used to identify a LSPR structure that is appropriate for a biosensing application requiring specific LSPR characteristics. The numerical electromagnetic modeling approach of the finite difference time domain (FDTD) method, and the analytical method of electrostatic eigenmode are used to implement the proposed model.