On the basis of different types of experiments, we develop implicitly the model of surfaceenhanced Raman scattering (SERS) of adsorbates on metal s u r f a c g . The long-range enhancement by resonances of the macroscopic laser and Stokes field is separated quantitatively from the metal eledmn-mediated resonance Raman effect. The latter mechanism proceeds b y increased electron-photon coupling at an atomically rough surface and by temporary charge transfer to orbitals of the adsorbates. This model can account for the chemical spedficity and vibrational selectivity of SERS and (partly) for the SERS specificity of the various metals.

A new method of exciting nonradiative surface plasma waves (SPW) on smooth surfaces, causing also a new phenomena in total reflexion, is described. Since the phase velocity of the SPW at a metal-vacuum surface is smaller than the velocity of light in vacuum, these waves cannot be excited by light striking the surface, provided that this is perfectly smooth. However, if a prism is brought near to the metal vacuum-interface, the SPW can be excited optically by the evanescent wave present in total reflection. The excitation is seen as a strong decrease in reflection for the transverse magnetic light and for a special angle of incidence. The method allows of an accurate evaluation of the dispersion of these waves. The experimental results on a silver-vacuum surface are compared with the theory of metal optics and are found to agree within the errors of the optical constants.

The model of surface-enhanced Raman scattering (SERS) by time-dependent evolution in the intermediate anionic state of the adsorbate is analogous to intramolecular Franck-Condon resonance Raman scattering.For adsorbates with a p * state, the residence time of some femtoseconds (10 −15 s) in the anionic state leads to a separation of electron (e) and hole (h), which quenches SERS at a smooth surface. At so-called SERS-active sites, the residence time of the hole is enhanced and therefore there is no final e-h pair and the excitation of only a molecular vibration leads to SERS. In contrast, for molecules with only high-energy s * states, the residence time in the anionic state is <1 fs (analogous to the impulse mechanism in electron scattering), and the creation of e-h pairs is less likely. This leads to first-layer electronic Raman scattering, especially by C-H stretch vibrations with an average enhancement of about 30-40-fold.

Proposed single molecule surface-enhanced Raman scattering (SM-SERS) mechanisms are discussed and problems in SM-SERS of biological molecules are pointed out. It is unlikely that the single molecule signals can be explained exclusively by electromagnetic (EM) 'hot spots' in fractal structures or in narrow gaps. Care must be taken to recognize and avoid the signal of sp 2 -carbon contamination, the origins of which are chemical impurities and/or photophysical reactions of the chosen adsorbate. As already pointed out by Käll and co-workers, relatively large globular proteins fit only in wider gaps, and therefore the EM enhancement is several orders smaller than often reported enhancements in SM-SERS. SM-SERS of haemoglobin probably shows signals from the Fe-porphyrin constituent among other unknown signals. It cannot be excluded that the haemoglobin is denatured and the haem group is situated in a smaller gap.

No abstract