Surface Diffusion and Ordering Processes Exploited for Directed Self-Assembly Using Amorphous Semiconductors

University Of Illinois At Urbana-Champaign, Urbana IL

Investigators

Abstract

This Nanotechnology research project seeks to develop a science base connected with surface diffusion and surface ordering of amorphous materials, with the ultimate goal of developing a new method for directed surface self-assembly on the nanoscale using amorphous semiconducting materials. Diffusion over amorphous and other energetically heterogeneous surfaces plays a role in sintering of ceramics and in reflow processes and memory device fabrication in microelectronics. There presently exists virtually no literature for diffusion on amorphous surfaces, and very little that specifically addresses continuously distributed energetic heterogeneity on highly defected crystalline surfaces. Such heterogeneity should lead to values of the diffusivity D that differ significantly from those measured for well-defined crystalline surfaces, however. For amorphous materials, fabrication processes can be devised based upon promoting or inhibiting surface ordering driven by surface diffusion. Example applications include fabrication of memory device electrodes, efficient solar cells, and amorphous ceramics. The rates of ordering cannot be followed experimentally by conventional techniques, rendering creation of process models difficult. The first goal of this project is to develop a science base for surface diffusion and ordering on amorphous and other energetically heterogeneous surfaces, using the combined expertise of the two laboratories. An experimental method will be developed to quantify the distribution function describing surface diffusion on energetically heterogeneous surfaces, using amorphous silicon and the ceramic titanium dioxide as paradigm cases. Experiments will also confirm that optical illumination can be used to drive surface diffusion non-thermally. With respect to surface ordering, fluctuation microscopy will be adapted to the study of near-surface regions by implementation on a scanning transmission electron microscope, with spatial resolution as small as 0.8 nm. Furthermore, the method will be extended to the study of binary compounds like titanium dioxide; up to now application has been restricted to single-element systems. The second goal of this project is to develop a new surface self-assembly method at the 10-200 nm length scale using amorphous semiconducting materials containing controlled amounts and size distributions of subcritical nuclei. Patterned optical or electron beam exposure should yield a spatially varying surface mass flux that, when performed at an annealing temperature just at the cusp of crystallization, provides the extra nudge to crystallize subcritical nuclei in regions dictated by the light flux. The full-fledged crystallites should then grow by surface diffusion and Ostwald ripening until the desired fraction of the film has accreted onto the original nuclei. Demonstrations will focus on amorphous silicon and titanium dioxide. Some computational aspects of the proposed work will be incorporated into a new interdisciplinary laboratory course for undergraduates focusing on nano-materials synthesis. The core goals of elucidating diffusion phenomena on amorphous surfaces, extending the capabilities of fluctuation microscopy, and demonstrating a version of the self-assembly method will remain in place. The downward revision in budget will impact the proposed work in several ways involving the scope of the individual tasks outlined. 1. Reduce the accuracy with which we can determine the locus of conditions giving optimal medium range order in amorphous silicon. 2. Limit the scope of studies for controlling medium range order via variations in hydrogen addition, ion energy and flux, and related parameters. 3. Limit the implementation of the directed self-assembly method to optical means, rather than including an electron-beam. 4. Limit the accuracy with which distribution functions describing surface diffusion can be obtained. (The distribution functions are more accurate with more data.)