[SY-D4] Big-data insights into solute-GB segregation
Solute-grain boundary (GB) interaction plays a critical role in the evolution and stabilization of grain structure and thus strongly impacts final material properties. Due to the large number of possible grain boundary configurations, there are a many inequivalent sites solutes can be incorporated at. Combined with the fact that there are many possible segregating chemical species in modern alloys, concepts to derive general trends in segregation are still missing. Using classical molecular statics, we perform high-throughput calculations of 1.4 million segregation energies for six solutes to 38 different boundaries in Al. The size of this data set is sufficient to apply machine learning techniques for building predictive models capable of predicting segregation energy to new GBs. It also provides useful insights into trends in the atomistic mechanisms controlling segregation behaviour. We show that the resulting segregation energy distributions can be interpreted analogously to electronic density of states and provides a useful perspective to consider solute concentration enrichment at the GB and GB embrittlement.