We present a comprehensive computational study on the optimization of the size of gold nanorods for single-molecule plasmonic sensing in terms of optical refractive index sensitivity. We construct an experimentally relevant model of single-molecule–single-nanoparticle sensor based on spherically capped gold nanorods, tip-specific functionalization and passivation layers, and biotin-streptavidin affinity system. We introduce a universal figure of merit for the sensitivity, termed contrast-to-noise ratio (CNR), which relates the change of measurable signal caused by the discrete molecule binding events to the inherent measurement noise. We investigate three distinct sensing modalities relying on direct spectral measurements, monitoring of scattering intensity at fixed wavelength and photothermal effect. By considering a shot-noise-limited performance of an experimental setup, we demonstrate the existence of an optimum nanorod size providing the highest sensitivity for each sensing technique. The optimization at constant illumination intensity (i.e., low-power applications) yields similar values of approximately 20 × 80 nm2 for each considered sensing technique. Second, we investigate the impact of geometrical and material parameters of the molecule and the functionalization layer on the sensitivity. Finally, we discuss the variable illumination intensity for each nanorod size with the steady-state temperature increase as its limiting factor (i.e., high-power applications).

The optimum size of gold nanorods for single-molecule plasmonic sensing in scattering and absorption modalities is determined by computing ratio of measurable signal to the inherent measurement noise (contrast-to-noise ratio) as a prime figure-of-merit.