Liquid-metal embrittlement in refractory metal grain boundaries
||Liquid-metal embrittlement in refractory metal grain boundaries|
||SNIC Medium Compute|
||Jakob Blomqvist <email@example.com>|
||2021-02-25 – 2022-03-01|
The purpose of this project is to investigate the impact of liquid metal impurities (Ni, Sn and Li) on tungsten and molybdenum grain boundary strength by means of classical atomistic and DFT modelling. Liquid Ni, Sn and Li are potential liquid plasma facing materials for future fusion reactors whose main purpose is to shield the the inner walls from the deleterious plasma disruptions that can cause substantial damage to the structure. The problem is that such impurities may diffuse into the structural materials (tungsten and molybdenum alloys) and segregate at the grain boundaries at which they can weaken the grain boundary strength. Thus, we aim to investigate these issues by means of atomistic modelling. There are two objectives to this project (i) investigate impurities' impact on the grain boundary strength by means of DFT-modelling and (ii) generate semi-empirical second nearest neighbour modified embedded atom method (2NN-MEAM) potentials that can be used in classical molecular dynamics simulations for modelling such phenomena.
For the DFT-modelling we will consider highly symmetric tilt grain boundaries and use the approach outlined by Olsson and Blomqvist  to investigate the impurity segregation and agglomeration during loading, to investigate how impurity content varies when subjected to mode I the grain boundary separation. A typical job will use up to 160 cores. This is a computationally rather demanding venture that relies on evaluating the thermodynamic grand force potential for numerous impurity concentration and are estimated to require about 80 000 cpu-hours monthly on tetralith for a year.
For the fitting and testing of classical interatomic potentials and subsequent production MD modelling we will study the grain boundary wetting and how the grain boundary fracture toughness varies with impurity content. We will rely on classical molecular dynamics and grand canonical Monte Carlo modelling. This effort requires the modelling of slip and brittle mechanisms for large supercells. A typical job would require up to about 128 cores and we estimate a usage of about 40 000 core hours on tetralith.
In total, we estimate the monthly use to 120 000 core hours per month on tetralith.
 P. A. T. Olsson and J. Blomqvist, Computational Materials Science, Volume 139, Nov 2017, 368-378.