Structure identification at extreme conditions

We are investigating a series of low-crystallinity, amorphous uranium oxides relevant for long term storage of spent nuclear fuel. One compound has stoichiometry of ~UO_3.5 (a-UO₃), which appears to be x-ray amorphous. To understand the a-UO₃ structure, we couple density functional theory with a genetic algorithm structure prediction mechanism to identify stable structures of non-stoichiometric uranium oxides. The lowest energy predicted structure with stoichiometry UO_3.5 contains a peroxide bridge, compatible with a study of comparable material using neutron pair-distribution function techniques. We attempted to crystallize a-UO₃ using a diamond anvil cell and Raman spectroscopy up to 25 GPa and observed highly anharmonic responses to pressure, but no long range order. Computational studies on the low-energy structure from the genetic algorithm show commensurate responses to the dynamical observations, along with anomalous changes in local bonding character, suggesting an indirect pathway for material identification.