Raman spectroscopy of single molecules and nanostructures

The rich structural information provided by the vibrational Raman response, compared to molecular fluorescence, makes Raman spectroscopy highly desriable for nanoscale analysis. In this study we showed that the simple geometry of the sharp metallic tip located above molecules adsorbed on planar metal surfaces result in sufficient field enhancement to allow for single molecule Raman spectroscopy with the additional advantage provided by the scanning probe microscope:

Probing the tip-scattered and -enhanced Raman response applying a side-on confocal epi-illumination and detection geometry. The plasmon resonance of the tip is selected to provide strongly localized and enhanced pump (νi) and Stokes Raman fields (νs) at the tip apex.

The following figure shows the tip-scattered Raman signal during tip approach of the planar gold surface covered with ∼ 1 ML of malachite green. The enhancement is confined to a tip-sample spacing of just several nanometers and correlated with the apex radius of the tip:

Tip-enhanced vibrational Raman response of ∼ 1 ML Malachite Green on a flat Au surface (a) and (b). From the tip-sample distance dependence (a) the enhanced Raman signal is seen to be spatially confined to a length scale of ≤ 15 nm, i.e., correlated with the tip apex radius. Comparison of tip-enhanced Raman response (b) with a corresponding far-field micro-Raman spectrum (c) of the same molecular surface layer and the normal mode analysis based on a DFT calculation. The simultaneous near-field enhancement of the fluroescence largely emerges from the Au-tip apex itself as evident from the control experiment without the molecules.

The characteristic differences between the tip-scattered and -enhanced Raman response and and the far-field spectrum is the result of the strong optical field localization, and related to different selection rules due to the field-gradient.

In our experiment we expect the enhancement to be purely electromagnetic in origin. From a direct comparision of the tip-scattered response with the far-field signal of the same surface molecules we can estimate a Raman enhancement of up to 5×109 compared to the free molecule response in our experiment. Together with the sensitivity of our detection system this suffices for probing the single molecule Raman response:

Time series of successive tip-scattered Raman spectra for a sub-monolager surface coverage of Malachite Green. The spectral diffusion observed is characteristic from single molecule emission. From the statistics (not shown) distinct peaks can be observed in the histogram of the Raman peak intensities, suggesting that the molecules are diffusing in and out of the probe volume.

The coupling of the axial plasmon resonance of the tip to the underlying substrate accross the nanometer tip-sample spacing gives rise to the strong lateral field confinement and related to a local field enhancement about one order of mangitude larger compared to the corresponding values for free standing tip:

Calculated local field distribution and enhancement E/E0 for a Au hyperbolic model tip with apex radius r = 10 nm, free standing (a) versus near the Au surface at a distance of d = 10 nm (b). Scale bar = 10 nm.

This tip-sample coupling provides the necessary local field enhancement of order 70 - 130 deduced from the Raman experiment and necessary to observe a single molecule response.

With the demonstrated potential of TERS to probe molecules at the single emitter level, it is highly desirable to extend the technique to probe other classes of materials such as crystalline nanostructures. By taking advantage of the intrinsic symmetry selectivity of the Raman response in combination with the polarization-selective enhancement of the tip we are applying TERS to identify the crystallographic orientation and -domains in nanostructures.

In Raman scattering, the scattering response is given by I = |es R ei|2, where R is the Raman tensor and es and ei are the scattered and incident light polarizations, respectively. Each normal mode oscillation (phonon in crystalline structures) will have a corresponding Raman tensor, generally containing few nonzero elements. By appropriate selection of the incident and scattered polarization directions, specific phonon modes can be isolated and studied. Conversely, with a priori knowledge of the phonon mode frequencies and Raman tensors, we can determine the crystallographic orientation of a nanostructure.

a) Shear-force topography of a crystalline BaTiO3 nanorod on a Au substrate (1.28 x 1.28 µm). b) Spectrally integrated TERS signal from the same surface region, showing strong optical contrast on top of the nanorod, as well as above highly localized substrate regions. c) Raman spectra acquired on a BaTiO3 rod (blue) and on Au substrate (black).
The peaks are identified (blue vertical lines) as E1 TO mode at 510 cm-1, a second order peak at 520 cm-1 and a combination of the A1 LO and E1 LO modes at 727 cm-1 and 715 cm-1, respectively. A cross section of the region of high enhancement (blue) and corresponding topography (black) on the rod is shown in panel d), taken as an average over 3 adjacent pixels along the straight dashed line in panels a) and b). The strong rise in optical signal on the left is correlated with a ~3 nm height variation, unlike the sharp decrease in optical response on the right which has no topographic correspondent; this indicates the presence of a crystalline nanodomain.

To top