Impact of Substrate on Tip-enhanced Raman Spectroscopy --- A Comparison of Frequency Domain Simulations and Graphene Measurements

Tip-enhanced Raman spectroscopy (TERS) has reached nanometer spatial resolution for measurements performed at ambient conditions and sub-nanometer resolution at ultra high vacuum. Super-resolution (beyond the tip apex diameter) TERS has been obtained, mostly in the gap mode configuration, where a conductive substrate localizes the electric fields. Here we present experimental and theoretical TERS to explore the field distribution responsible for spectral enhancement. We use gold tips of $40\pm 10 \ \text{nm}$ apex diameter to measure TERS on graphene, a spatially delocalized two-dimensional sample, sitting on different substrates: (i) glass, (ii) a thin layer of gold and (iii) a surface covered with $12\ \text{nm}$ diameter gold spheres, for which $6\ \text{nm}$ resolution is achieved at ambient conditions. The super-resolution is due to the field configuration resulting from the coupled tip-sample-substrate system, exhibiting a non-trivial spatial surface distribution. The field distribution and the symmetry selection rules are different for non-gap vs. gap mode configurations. This influences the overall enhancement which depends on the Raman mode symmetry and substrate structure.