Hydrogen Traps in defective M7C3 structures
Title: Hydrogen Traps in defective M7C3 structures
SNIC Project: SNIC 2022/5-514
Project Type: SNIC Medium Compute
Principal Investigator: Christina Bjerkén <christina.bjerken@mah.se>
Affiliation: Malmö universitet
Duration: 2022-11-01 – 2023-11-01
Classification: 10304 20599


Hydrogen embrittlement is a common problem for a series of metals leading to decrease in ductility causing crack formation leading to catastrophic failure of the material. The mechanisms of hydrogen embrittlement are complex and still not fully understood. (Barrera O et al. (2018) “Understanding and mitigating hydrogen embrittlement of steels: a review of experimental, modelling and design progress from atomistic to continuum.” J Mater Sci 53:6251-6290.) The purpose of this project is to understand the effect of defects, such as vacancies, substitutions and Frenkel defects, on hydrogen diffusion in heptametal carbide ferrite interfaces. Plane wave density functional theory (PW-DFT) calculations using VASP and OpenMX will be used to calculate binding energies of hydrogen in different solution sites within the heptametal carbide lattices. The calculations will focus on the carbides M_7 C_3 with M=Cr,Mn,Fe and an orthorhomic Pnma structure. Zero-Point Energies (ZPE) have to be calculated for hydrogen containing structures. Stable interface structures with ferrite are to be created. Hydrogen diffusion energy barriers will be calculated using the nudge elastic band method (NEB). A more detailed literature overview can be found in Järnkontorets report TO41-44 which is part of the HySteel project. VASP and OpenMX are DFT codes (MPI) to be used in this project. Benchmark calculations have been performed on a similar cluster. They both scale well up to 144 cores. Large systems (large supercells to tackle compositional variations) are to be addressed which requires extensive computational resources. Magnetism must be considered, which increases the computational efforts. Hence, typical job settings (chained execution) are as follows: 144 cores, up to 120 core hours/job, and 3 GB/core. Only short-term data storage is required since all files will be stored and post-processed locally.