Key points are not available for this paper at this time.
In the late 1970s, signal intensity in Raman spectroscopy was found to be enormously enhanced, by a factor of 10(6) and more recently by as much as 10(14), when an analyte was placed in the vicinity of a metal nanoparticle (particularly Ag). The underlying source of this huge increase in signal in surface-enhanced Raman scattering (SERS) spectroscopy has since been characterized by considerable controversy. Three possible contributions to the enhancement factor have been identified: (i) the surface plasmon resonance in the metal nanoparticle, (ii) a charge-transfer resonance involving transfer of electrons between the molecule and the conduction band of the metal, and (iii) resonances within the molecule itself. These three components are often treated as independently contributing to the overall effect, with the implication that by properly choosing the experimental parameters, one or more can be ignored. Although varying experimental conditions can influence the relative degree to which each resonance influences the total enhancement, higher enhancements can often be obtained by combining two or more resonances. Each resonance has a somewhat different effect on the appearance of the resulting Raman spectrum, and it is necessary to invoke one or more of these resonances to completely describe a particular experiment. However, it is impossible to completely describe all observations of the SERS phenomenon without consideration of all three of these contributions. Furthermore, the relative enhancements of individual spectral lines, and therefore the appearance of the spectrum, depend crucially on the exact extent to which each resonance makes a contribution. In this Account, by examining breakdowns in the Born-Oppenheimer approximation, we have used Herzberg-Teller coupling to derive a single expression for SERS, which includes contributions from all three resonances. Moreover, we show that these three types of resonances are intimately linked by Herzberg-Teller vibronic coupling terms and cannot be considered separately. We also examine the differences between SERS and normal Raman spectra. Because of the various resonant contributions, SERS spectra vary with excitation wavelength considerably more than normal Raman spectra. The relative contributions of totally symmetric and non-totally symmetric lines are also quite different; these differences are due to several effects. The orientation of the molecule with respect to the surface and the inclusion of the metal Fermi level in the list of contributors to the accessible states of the molecule-metal system have a strong influence on the observed changes in the Raman spectrum.
Lombardi et al. (Fri,) studied this question.