Proximity field nanopatterning for large area three-dimensional photonic nanostructures: Forward and inverse problem modeling.

摘要

Production methods used in photonics and nanotechnology suffer from many limitations, hindering the ability to realize devices and restricting the actual number of applications. An ideal processing method should require low-cost equipment, be able to produce very fine details, and be scalable to process large area specimens in an acceptable amount of time. Proximity Field Nanopatterning (PnP) is a lithography method possessing these features. By using interference patterns produced by a two-dimensional phase mask, the technique is able to generate a submicron detailed exposure on a millimeter-size slab of light sensitive photopolymer. Exposure to light at certain wavelengths modify chemical properties of photopolymers at the exposed locations. In particular, solubilities of photopolymers in certain liquids are altered by exposure. In Proximity Field Nanopatterning, the photopolymer slab is developed like a photographic plate in such a liquid to reveal three-dimensional interference patterns from the phase mask.;While it is possible to use computer aided simulations to obtain the interference patterns produced by a mask with a certain pattern, the inverse problem of producing a mask for a desired interference pattern cannot be solved in the same way due to the intricacies of light interactions involved in producing the final interference pattern. An alternative technique is to iteratively optimize the phase mask so that the interference patterns obtained converge to the desired pattern. This work starts by elaborating on development and implementation of an integrated method which accomplishes the task by comparing results from a Finite Difference Time Domain method based simulation of a PnP experiment to a set of images from targeted structures. The comparison results, quantified through fuzzy image pattern recognition techniques, are interpreted by a fast gradient based optimizer, which provides corrections to the PnP phase mask for the next iteration.;Further improvement of the integrated method through automated learning by a fuzzy inference system is detailed next. Driven by a robust structural analysis technique parameterizing images from simulations and desired structures alike, the inference system learns the relationships between parameters of phase masks and resultant structures to provide invaluable initial guidance for the main optimizer. We undertake further theoretical studies of PnP to investigate its few limitations. Finally, we report the first customizable large-area production of a Penrose quasicrystal active in the infrared and visible wavelengths, complete with structural modeling and similarity analysis.