Chemical Raman Enhancement of Organic Adsorbates on Metal Surfaces

Using first-principles theory and experiments, chemical contributions to surface-enhanced Raman spectroscopy for a well-studied organic molecule, benzene thiol, chemisorbed on planar Au(111) surfaces are explained and quantified. Density functional theory calculations of the static Raman tensor demonstrate a strong mode-dependent modification of benzene thiol Raman spectra by Au substrates. Raman active modes with the largest enhancements result from stronger contributions from Au to their electron-vibron coupling, as quantified through a deformation potential. A straightforward and general analysis is introduced to extract chemical enhancement from experiments for specific vibrational modes; measured values are in excellent agreement with our calculations.

Figure 1

(A) Calculated Raman spectra of BT (from $R_{zz}$ components only) for four different binding geometries: adatom, hollow, $\mathrm{\text{adatom}}+\mathrm{H}$, and bridge. Inset shows our calculations of the orientationally-averaged gas-phase Raman spectrum of benzene thiol. All spectra are broadened with a Lorentzian of $10{\mathrm{cm}}^{-1}$ width. (B) Binding geometries. Our supercell consists of 5 atomic layers of Au stacked along [111] with 16 atoms per layer, with 30 Å of vacuum. The in-plane lattice parameters are kept fixed to their computed Au fcc bulk value of 4.17 Å.

Figure 2

(A) Computed isosurfaces (at $1.65\times {10}^{-4}e/{\AA }^3$) of charge density induced by freezing-in small amplitudes (0.1 Å) of the 1100 and $3151{\mathrm{cm}}^{-1}$ modes. For both modes, gold atoms are stationary, and charge rearrangement in the substrate is induced by the molecule. (B) Partial densities of states projected on the molecule near the Au Fermi energy for three different amplitudes.

Figure 3

(a) Partial densities of states projected on the molecule for different binding sites shown about the Au Fermi energy. (b) Difference $R_{zz}^{\mathrm{Adatom}}-R_{zz}^{\mathrm{Adatom}+\mathrm{H}}$ plotted versus $\partial {\omega }_{\mathrm{H}}/\partial Q_n$. (c) The deviation $\delta$ of these modes from linear trend in (b) plotted versus their contribution to the lattice susceptibility, which is proportional to the square of the ratio of the mode dynamical charge and its frequency.

Figure 4

(a) Au SERS substrate used in our measurements with BT. (b) Raman and SERS BT spectra. (c) Calculated relative enhancements versus experimental relative enhancements for adatom, bridge, and hollow sites.