Understanding the basic mechanisms that determine microstructure changes in neutron irradiated steels is vital for a safe lifetime management of existing nuclear reactors and a safe design of future nuclear options. Low-alloyed ferritic steels containing Cu, Ni, Mn and Si as principal solute atoms are used as structural materials for current reactor vessels. The microstructural evolution under irradiation in alloys is decided by the interplay between defect formation and thermodynamic driving forces, together determining the appearance of phase transformations (precipitation, segregation, …) and favouring or delaying the nucleation and growth of point-defect clusters, their diffusion and their mutual recombination or removal at sinks. A reliable description of the production, evolution and accumulation of radiation damage must therefore start from the atomic level and requires being able to describe multicomponent systems for timescales ranging from few picoseconds to years. This goal demands firstly the fabrication of interatomic potentials for alloys that must be both consistent with the thermodynamic properties of the system and capable of reproducing correctly the characteristic solute-point defect interactions, versus ab initio or experimental data. Secondly the performance of extensive molecular dynamics (MD) simulations, to grasp the main mechanisms of defect production, diffusion, mutual interaction, and interaction with solute atoms and impurities. Thirdly, the development of simulation tools capable of describing the microstructure evolution beyond the timeframe and lengthscale of MD, while reproducing as much as possible the atomic-level origin of the mechanisms governing the evolution of the system, including phase changes.
In this presentation the results of recent efforts made in this direction in the case of Fe-Cu, Fe-Cr and Fe-Ni alloys, as basic model alloys for the description of steels of technological relevance, are highlighted. In particular, advanced techniques to fit interatomic potentials consistent with thermodynamics are proposed and the results of their application to the mentioned alloys are presented. Next, the development of advanced methods, based on the use of artificial intelligence, to improve both the physical reliability and the computational efficiency of kinetic Monte Carlo codes for the study of point-defect clustering and phase changes beyond the scale of MD, is reported. These recent progresses bear the promise of being able, in the near future, of producing reliable tools for the description of the microstructure evolution of realistic model alloys under irradiation.