Does the chemical contribution have a secondary role in SERS?

Abstract

It is an established understanding that the electromagnetic contribution (plasmon-mediated enhancement of a laser and scattered local electromagnetic fields) is the main actor in surface enhanced Raman scattering (SERS), with the so-called chemical (molecule-related) contribution assuming only, if any, a supporting role. The conclusion of our comprehensive experimental resonant study of a broad range of nanosphere lithography based metallic substrates, with covalently attached 4-mercaptobenzoic acid monolayers used as a probe (standard molecules that are non-resonant in solution), is that this accepted understanding needs to be revised. We present a detailed resonant SERS study of metal-film-over-nanosphere (MFON) substrates that is done by both scanning the laser wavelength and tuning the plasmon response through the nanosphere diameter, which is varied from 500 to 900 nm. Far and local field properties are characterized through measures of optical reflectivity and SERS efficiency, respectively, and are supported by numerical simulations. We demonstrate that SERS intensity depends indeed on the electromagnetic mechanism, determined by the plasmonic response of the system, but we observe that it is also strongly defined by a chemical resonant contribution related to a metal-to-ligand electronic transition of the covalently bound probe molecule. Optimum amplification occurs when the plasmon modes intersect with the ligand-to-metal chemical resonance, contributing synergically both mechanisms together. Quite notably, however, the largest SERS signal is observed when the laser is tuned with the metal-to-ligand transition, and typically does not follow the wavelength dependence of the plasmon modes when varying the nanosphere size. The same general trend is observed for other nanosphere lithography based substrates, including sphere segment void cavities and hexagonally ordered triangular nanoparticles, using either Ag or Au as the plasmonic metal, and also with a commercial substrate (Klarite). Interestingly, this extensive comparative investigation shows in addition that MFONsubstrates are significantly better than these other studied plasmonic substrates in terms ofRaman intensity and homogeneity. We conclude that a deep understanding of both electromagnetic and chemical mechanisms is necessary to fully exploit these substrates for analytical applications.