Optical Properties of Capped Metallic Nanostructures, Grown on Silicon

McAlinden, Niall and McGilp, John (2011) Optical Properties of Capped Metallic Nanostructures, Grown on Silicon. PhD thesis, Trinity College Dublin.

[thumbnail of Optical Properties of Capped Metallic Nanostructures, Grown on Silicon]
Preview
PDF. Filename: ThesisA4.pdf
Accepted Author Manuscript
License: Unspecified

Download (27MB)| Preview

Abstract

Reflectance anisotropy spectroscopy (RAS) is a linear optical technique that measures the difference in the reflectance of two orthogonal polarisations at normal incidence. It achieves surface and interface sensitivity when the bulk material, such as a cubic semiconductor, is optically isotropic. The penetration depth of optical radiation allows RAS to probe buried interfaces. RAS has been used to probe various one-dimensional (1-D) structures grown on vicinal Si(111) surfaces under ultra-high vacuum (UHV) conditions. The RAS system response was extended into the IR, where important optical transitions occur, for both a photoelastic modulated system and a rotating sample system using a tuneable IR laser. RAS spectra of single domain Si(111)-5x2-Au, Si(557)-Au and Si(775)-Au structures showed large minima in the region around 2 eV and, in the case of Si(111)-5x2-Au a large maximum below 1 eV. The monolayer (ML) coverage of Au required for the Si(111)-5x2-Au surface reconstruction has been extracted from the RAS response. Using the well known coverage of Au required for the Si(557)-Au reconstruction, the Au deposition rate was accurately calibrated. By analysis of the coverage required for several Si(111) vicinal off-cuts, taking into account the different step densities, a coverage for a "pure" Si(111)-5x2-Au surface was calculated. A value of 0.59 ML +/- 0.08ML was found, in agreement with the recent work. The value supports a new three chain model for the Si(111)-5x2-Au surface reconstruction. Upon deposition of small amounts of Si adatoms on the Si(111)-5x2-Au surface and subsequent annealing, the RAS spectra changed dramatically, as the adatom decorated "5x4" reconstruction was formed. Temperature dependent studies allowed 100% and a 0% adatom filled sites RAS spectra to be extracted. These spectra will be particularly useful for comparison with future ab initio optical response calculations. The optical signatures from this surface could prove to be very interesting in the study of defect induced charge density waves. A strong optical anisotropy was also seen on the Si(775)-Au and Si(557)-Au surfaces. The RAS spectra showed a minimum around 2 eV but the maximum above 1 eV, seen on the Si(111)-5x2-Au surface was not present. A possible explanation is that the chain structures on these narrower terraces are more sensitive to the presence of kinks. The average length of the Au chains is expected to be significantly shorter on the Si(775)-Au and Si(557)-Au surfaces, as the kinks will terminate the Au chains more efficiently than on the lower angle offcuts, used for the Si(111)-5x2-Au studies. The RAS response from Si(557) shows two peaks related to surface modified bulk states at 3.4 eV and 4.25 eV, and a surface state at 1.2 eV. The RAS signal was compared with preliminary ab initio optical response calculations. Reasonable results were found for a bulk terminated and relaxed Si(557) surface. However, the structure is known to consist of a triple step structure of approximately (112) orientation and the large terrace of the Si(111)-7x7 reconstruction. Calculations of the RAS spectra from Si(112) did not reproduce the features seen experimentally. Elongated Pb islands with lengths of up to 430 nm and widths of 60 nm were grown on Si(557)-Au and their RAS spectra were recorded. The wires showed a strong RAS signal with a negative peak at 1.1 eV and a positive peak at 0.47 eV. The Pb islands could be capped with a-Si and their reflection anisotropy was retained, with both peaks shifted to the IR. The results showed that capping with a-Si was largely successful. The modelling of the RAS response was less successful. A nanoantenna approach gave reasonable values of the length of the Pb islands, using the wavelength of the maximum. Other models, which attempted to predict the line shape, could reproduce either the minimum or maximum accurately but not both. However, these models neglect both quadrupolar effects and dipole-dipole interactions between islands. The sensitivity of the RAS response to the detail of the island structure indicates that RAS could be a powerful probe of plasmonic structures if a suitable theoretical model can be developed.