Scott, D.; (2008) The discovery of new functional oxides using combinatorial techniques and advanced data mining algorithms. Doctoral thesis , UCL (University College London).

Preview

PDF 15214.pdf Download (1MB)

Abstract

Electroceramic materials research is a wide ranging field driven by device applications. For many years, the demand for new materials was addressed largely through serial processing and analysis of samples often similar in composition to those already characterised. The Functional Oxide Discovery project (FOXD) is a combinatorial materials discovery project combining high-throughput synthesis and characterisation with advanced data mining to develop novel materials. Dielectric ceramics are of interest for use in telecommunications equipment; oxygen ion conductors are examined for use in fuel cell cathodes. Both applications are subject to ever increasing industry demands and materials designs capable of meeting the stringent requirements are urgently required. The London University Search Instrument (LUSI) is a combinatorial robot employed for materials synthesis. Ceramic samples are produced automatically using an ink-jet printer which mixes and prints inks onto alumina slides. The slides are transferred to a furnace for sintering and transported to other locations for analysis. Production and analysis data are stored in the project database. The database forms a valuable resource detailing the progress of the project and forming a basis for data mining. Materials design is a two stage process. The first stage, forward prediction, is accomplished using an artificial neural network, a Baconian, inductive technique. In a second stage, the artificial neural network is inverted using a genetic algorithm. The artificial neural network prediction, stoichiometry and prediction reliability form objectives for the genetic algorithm which results in a selection of materials designs. The full potential of this approach is realised through the manufacture and characterisation of the materials. The resulting data improves the prediction algorithms, permitting iterative improvement to the designs and the discovery of completely new materials.

Download activity - last month

Export as CSVXMLJSON

Download activity - last 12 months

Export as CSVJSONXML

Downloads by country - last 12 months

Export as CSVJSONXML

Archive Staff Only

View Item