Traditional and Novel Approaches in Ångstchemistry: A Step into Postmodernism

• Author / Creator

  Oliynyk, Anton

• This thesis focuses on the experimental and theoretical study of various rare-earth transition-metal germanides that contain three or four components. Ternary and quaternary germanides were synthesized through various methods, including direct reaction of the elements, arc-melting, and flux growth. Several new series – RE4M2InGe4, RE4MInGe4, RE3M2Ge3, RE2+xMnGe2+y – were identified which contain characteristic structural motifs in common. These motifs can then be regarded as design elements to derive new structures. The crystal structures of these compounds contain many types of building units, leading to diverse physical properties (electrical and thermal transport, magnetic properties) amenable to a broad array of materials applications. For example, the complex structures of some germanides give rise to surprisingly low thermal conductivity, suitable for thermoelectric materials, and the interaction of f-electrons on rare-earth atoms with d-electrons on transition-metal atoms generates interesting magnetic behaviour. To guide experimental attempts to thinking “outside of the box,” data-driven machine-learning tools have been applied to identify new germanides, and to predict their likelihood to display favourable thermoelectric properties. A recommendation engine was first tested on a previously known intermetallic compound, Gd12Co5Bi, which was suggested to be a counterintuitive candidate for thermoelectric materials. Property measurements on this compound revealed promising performance and suggested that this may be a member of a new class of thermoelectric materials. This approach was then extended to find new germanides with low thermal conductivity (<10 mW/K), which is highly unusual for intermetallic compounds. The total thermal conductivities vary from ~20 to <2 mW/K, which are unprecedentedly low for unoptimized and metallic compounds (cf., typical values for thermal conductivity for metals are ~100 mW/K). Prediction of crystal structures and physical properties is still an unsolved problem. To address this challenge, the machine-learning strategies described above have been complemented with statistical methods (principal component analysis, support vector machines) and applied to simple binary intermetallics as well as more complex ones (Heusler phases and equiatomic ternary phases). This approach may greatly accelerate the search for new materials and minimizes the risks in exploratory synthesis.

• Subjects / Keywords

  • X-ray diffraction
  • Germanides
  • Crystallography
  • Solid state chemistry
  • Machine learning

• Graduation date

  Fall 2016

• Type of Item

  Thesis

• Degree

  Doctor of Philosophy

• DOI

  https://doi.org/10.7939/R3HX1648B

• License

  This thesis is made available by the University of Alberta Libraries with permission of the copyright owner solely for non-commercial purposes. This thesis, or any portion thereof, may not otherwise be copied or reproduced without the written consent of the copyright owner, except to the extent permitted by Canadian copyright law.