Numerical modeling of a near-field scanning optical system.

Abstract

A near-field scanning optical (NFO) system utilizes a subwavelength sized aperture to illuminate a sample. The aperture raster scans the sample. During the scan, the aperture is held in proximity to the sample. At each sampling point, the integrated far-zone energy distribution is stored. This collection of data is used to generate an image of the sample's surface. The main advantage of NFO systems is their very high spatial resolution. In this dissertation a hybrid finite-difference-time-domain (FDTD)/angular spectrum code is used to study the electromagnetic and imaging properties of a NFO scanning system. In addition, a finite-difference thermal (FD-thermal) code is used to calculate the thermal properties of a NFO system. Various aperture/sample geometries are studied numerically using both TE and TM polarization within a two-dimensional metallic waveguide that forms the aperture. The spatial properties of the electric field emitted by the aperture with no sample present are greatly influenced by the polarization. In particular, the electric field with TM polarization exhibits sharp peaks near the corners of the aperture, while the field with TE polarization is smooth and peaked at the center of the aperture. For both polarizations, the electric field remains collimated for a distance comparable to the aperture size. The electric field for both polarizations is altered when a dielectric sample is placed in proximity to the aperture. It is shown that the most representative image of the sample's topography is obtained using TE polarization and the resulting total far-zone energy as the sampled data. It is also shown that simpler scalar methods do not accurately predict the imaging behavior of a NFO system. Under certain circumstances the relationship between the sample's topography and the detected image is nearly linear. Under these conditions a system transfer function is calculated. Using the transfer function, it is shown that the spatial resolution of a NFO system is on the order of the aperture size plus the aperture to sample spacing. Interestingly, the transfer function is object dependent. Post image equalization techniques are shown to increase system resolution. When a metallic sample is imaged, the object/image relationship is more complex than with a pure dielectric sample. In the metallic sample, signal enhancement is observed over sharp topographic features. In addition, the optical power that is absorbed in a metallic sample is converted to heat that flows throughout the sample. Thermal transfer between the tip and the sample is shown to play a smaller role in sample heating. It is shown that a wider thermal profile in the sample is obtained with TM polarization than with TE polarization. This is important in areas such as optical data storage, where an elliptically shaped data mark is predicted.