Multiple solution solving in plasmon sensing by deep learning: determination of a layer refractive index and...

Plasmons have received intensive attention owing to their significant spectrum shift in environmental sensing. Experimentally, the same spectral shifts may be caused by different combinations of structural parameters of a plasmonic nanoparticle. This multi-parameter problem cannot be solved by just a single feature analysis, but requires using the full scattering spectrum containing all features of the parameters. In this Letter, a deep learning method for solving multi-parameter problems is proposed based on the layer refractive index (n) and layer thickness (d) sensing of different nanorods and nanospheres. The full scattering spectrum can be theoretically simulated, precisely predicted using a well-trained deep learning method, and experimentally obtained using a homemade dark-field microscope. An error analysis of the simulation and experimental results indicates that this method is a potential way to determine n and d and further solve multi-parameter in plasmon sensing.