High performance computing (HPC) resources allow us to perform high resolution FDTD (Finite-Difference Time-Domain) simulations reaching a space discretization in the order of 0.1 nm for a 3D domain. A parallel message passing interface (MPI) FDTD in-house code has been run on an IBM BlueGene/Q supercomputer [1]. Three nanoantennas (nanosphere, dipole and bow-tie) of gold and silver have been studied. The dispersion modelling is based on the Drude equation with the two critical points correction [2]. The extinction, scattering and absorption coefficients have been calculated varying the mesh size. The classical staircasing approach is compared with a sub-cell one, where the material parameters are assigned for each electric field component of the Yee cell. The results show which discretization approach is better for each of the nanostructures and in which conditions round-off errors appear.
High resolution parallel FDTD simulation of plasmonic nanostructures
Vaccari, Alessandro;
2014-01-01
Abstract
High performance computing (HPC) resources allow us to perform high resolution FDTD (Finite-Difference Time-Domain) simulations reaching a space discretization in the order of 0.1 nm for a 3D domain. A parallel message passing interface (MPI) FDTD in-house code has been run on an IBM BlueGene/Q supercomputer [1]. Three nanoantennas (nanosphere, dipole and bow-tie) of gold and silver have been studied. The dispersion modelling is based on the Drude equation with the two critical points correction [2]. The extinction, scattering and absorption coefficients have been calculated varying the mesh size. The classical staircasing approach is compared with a sub-cell one, where the material parameters are assigned for each electric field component of the Yee cell. The results show which discretization approach is better for each of the nanostructures and in which conditions round-off errors appear.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.