EUROPEAN CONGRESS AND EXHIBITION ON ADVANCED MATERIALS AND PROCESSES - EUROMAT 2021

Multi-principal element alloys (MPEAs) consist of four or more elements alloyed in approximately equal fractions and often crystallize in a simple crystal lattice. In many cases, their mechanical properties like structural stability or ductility exceed that of common modern alloys. Usually they contain transition or refractory metals whose bonds are dominated by their d-electrons. Up to now, investigations into low density light metal MPEAs have been rare due to the complex binding modes of their constituents.

We use both a Cluster Expansion approach, augmented by stochastic prescreening steps, and neural network based pair potentials to scan the large configuration space of the Mg-Al-Cu-Zn system for stable phases. The training data was generated using density functional theory calculations implemented in the VASP code. We present an analysis of the strengths and limitations of the respective techniques with respect to their accuracy and ability to predict structure stabilities and physical parameters like hardness. In conjunction with experiments employing magnetron sputtering, we find that while the introduction of Al into the brittle MgZn$_{2}$ hexagonal Laves phase leads to phase separation and does not improve the mechanical properties of the alloy, the addition of Cu inhibits this process and leads to the formation of a highly stable cubic phase. We find that further increasing the Cu concentration leads to higher hardness of the samples, which is also reflected in an increase of the calculated bulk modulus. Furthermore, we show how the combination of modelling and experimental scanning techniques can reveal insights into the phase diagram of such complex multicomponent alloys.