Investigation of the near fields of sputtered Au thin films, used for surface enhanced R aman spectroscopy, using the AFM and DDA

Abstract

Surface enhanced Raman spectroscopy (SERS) is a powerful technique that uses metal nanostructures (Au, Ag, Cu) to gain in the best case single molecule sensitivity [1], because these nanostructures give rise to a huge enhancement of the Raman signal of molecules in their immediate vicinity. Most of the enhancement is attributed to the enhancement of the electric field near the surface of the metal, because the Raman signal in this case scales approximately proportional to the fourth power of the electric filed strength [2]. Therefore a good knowledge of both the geometry of the metal nanostructure and the electric near field it causes is crucial for the understanding of a SERS‐substrate. A combination of AFM and the discrete dipole approximation (DDA) is a promising approach for a better understanding of solid SERS substrates. Using the AFM the geometry of a nanostructure on a flat substrate or a structured surface can be measured very accurately and the DDA is a well‐established method for solving scattering problems for arbitrary shapes, which makes little assumptions about the sample other than the target geometry [3]. Furthermore, using the FFT‐accelerated DDA a very large number of dipoles (>10 6 ) can be simulated on a desktop computer [4], allowing for a discretization of comparatively large scatters on a nm scale. We have written a Matlab program that uses the DDA to calculate the near field from a given AFM image of a nanostructure composed of a single material. The incident field and dielectric constant of the material can be chosen arbitrarily. We have tested the program against Mie theory (fig. 1) and shown that the error of the electric field strength is reasonably small two discretization steps away from the surface (fig.2). The average error of the electric field strength two discretization steps away from the surface is smaller than 5 % (fig.2), with the maximum error being smaller than 10 % (fig.1; apart from the strip around x=0, where the relative error appears large due to |E Mie |≈0, but the absolute error is actually very small). This program is used to simulate SERS substrates, produced by sputtering thin Au‐films on glass slides (fig. 3). These substrates give rise to small but regular (across the substrate) enhancement with enhancement factors of the order of 10 2 – 10 3 . Our goal is to demonstrate the feasibility of using AFM based DDA simulations to study near field enhancement, by predicting the enhancement factors of several substrates.