Optimizing SERS from Gold Nanoparticle Clusters: Addressing the Near Field by an Embedded Chain Plasmon Model

We study experimentally and theoretically the optimization of surface enhanced Raman scattering (SERS) from nanoplasmonic clusters of gold nanoparticles separated by a fixed subnanometer gap. To maximize the enhancement we discuss how the optimal cluster size is influenced by the constituent nanoparticle size and illumination wavelength. We find good qualitative agreement between the experimental SERS from nanoparticle clusters and a simple composite model that describes the response of the full cluster as arising from a composition of linear nanochains. For fixed illumination wavelengths encountered experimentally it is best to choose a cluster size that supports its lowest energy resonance near this wavelength. Our chain simulations indicate the existence of an optimal length also when the illumination laser is continuously tuned to the frequency that maximizes the signal. We explain the optimal length under these illumination conditions with a simple model that accounts explicitly for radiative losses.