Shape and size dependence of dipolar plasmonic resonance of nanoparticles (Q2274023)

This paper addresses the electromagnetic scattering by a compact and smooth body with permittivity \(\epsilon_c\), which is assumed constant, frequency-dependent and different from the constant background permittivity \(\epsilon_m\). The body occupies a domain \(D\subset\mathbb{R}^3\) whose characteristic size is controlled by a scalar parameter \(\delta\). The incident wave is given by the monochromatic field \(\mathbf{E}^{i}\) whose operating frequency is denoted by \(\omega\). The scattered field can then be retrieved by solving Maxwell's equations complemented with standard radiation conditions. The authors aim to describe solutions near resonance frequencies, i.e. \(\omega\) for which the system is nearly singular; the resonance being enabled by the fact that \(\Re(\epsilon_c)<0\). To that end they formulate the system in terms of a volume integral equation, thereby relating resonances to solutions of the following non-linear eigenvalue problem \[ \mathcal{T}_{\tilde{D}}^{\delta\omega}\left[\mathbf{E}\right] = \frac{\epsilon_m}{\epsilon_m-\epsilon_c(\omega)}\mathbf{E}, \] where \(\omega\) is the eigenvalue, \(\mathbf{E}\) is the eigenfunction, \(\tilde{D}\) is the domain \(D\) re-scaled by \(\delta\), and \(\mathcal{T}_{\tilde{D}}^{\delta\omega}\) is a singular integral operator also known as the magnetization integral operator. Under the extra hypothesis of \(\delta\omega\ll1\) (justified in the plasmonic setting), the authors carry-out an eigenvalue perturbation analysis from the static operator \(\mathcal{T}_{\tilde{D}}^{0}\) whose spectral properties are shown to be directly linked to the well-known Neumann-Poincare operator. Using these results, Ammari and Millien provide a simple formula for estimating the resonance shift induced by retardation and radiation effects of the full-wave (dynamic) regime. The paper also contains explicit computations for the case where \(D\) is an ellipsoid. These computations involve the use of the so-called \(\mathbf{L}\) dyadic, which is widely used in the physics literature, but often incorrectly derived; thus, a correct derivation is also provided.