SNIC
SUPR
SNIC SUPR
Analyzing and predicting new hydrogen-rich materials
Dnr:

SNIC 2017/1-473

Type:

SNAC Medium

Principal Investigator:

Ulrich Häussermann

Affiliation:

Stockholms universitet

Start Date:

2018-01-01

End Date:

2019-01-01

Primary Classification:

10403: Materials Chemistry

Secondary Classification:

10404: Inorganic Chemistry

Webpage:

Allocation

Abstract

This project is meant to support experimental efforts in the synthesis of new hydrogen-rich materials using high pressure techniques. In particular, the experimental program targets ternary and quaternary hydrides A(A’) – E – H , A(A’) – T – H and RE – T – H where A = alkali and/or alkaline earth metal, T = transition metal, RE = rare earth metal, E = Si, P, S. We search for new high-temperature superconductors by investigating formation reactions (e.g. x AHn + y E + z H2) and high-pressure, high temperature phase diagrams of multinary hydrides using in-situ diffraction techniques at synchrotron beamlines. We target p,T conditions suitable for upscaling in large volume high pressure devices (5 – 15 GPa, 600 – 1300 K) and are especially interested in identifying ambient conditions recovery pathways for new hydrogen rich materials. Our experimental efforts were fueled by the discovery of a critical temperature as high as 203 K at 150 GPa for H3S (Drozdov et al. (2015) Nature, 525, 73). However, this material is not recoverable and the pressure conditions for superconductivity are far too extreme for any technological significance. The large compositional and structural flexibility of multinary hydrides allows optimization of superconducting properties and provides the possibility for synthesis at comparatively lower pressures (e.g. K. Puhakainen, D. Benson, J. Nylen, S. Konar, K. Leinenweber, E. Stoyanov, U. Häussermann Octahedral hypervalent SiH62- species from high pressure synthesis. Angew. Chem. Int. Ed. 51 (2012), 3156). Our in situ investigations at synchrotron facilities provide diffraction patterns from which the arrangement of metal atoms in the structures of obtained new hydride materials can be deduced. We will apply computational methods to also assess the amount of hydrogen atoms and their distribution. In particular we plan to use the VASP package at Kebnekaise (HPC2N), which has, according to our experience, almost linear scaling with respect to increase of the cores per job. We will use the binary and ternary metal atom arrangements suggested by experiment and predict H concentrations and H atom distributions by combining state-of-the-art ab initio methods and the evolutionary algorithm. We expect to perform a large amount of average size jobs limited by 12-24 hours and typically 2 nodes (56 processors) in order to calculate total energies for numerous H atom configurations involved in the evolutionary algorithm, which we use for the structural predictions. In total we plan to start about 150 jobs per month that requires approximately 200,000 hours. Approximate memory requests for each job are 1-2 Gb/node.