A combined DFT and machine learning investigation of structural, mechanical, and thermal behavior in magnesium oxide

Magnesium oxide (MgO) is a class of ceramic materials that is used in refractory systems, thermal management devices, electronic substrates, and high-temperature applications because of its thermal stability and mechanical strength. In this work, using DFT and ML, we have studied the structural, mechanical, and thermal properties of MgO in the rock salt crystal structure. First-principles calculations based on the GGA-PBE functional have been performed to obtain the equilibrium lattice constant, bulk modulus, elastic constants, elastic moduli, Debye temperature, thermal conductivity, and other thermodynamic parameters. The optimum structure has an equilibrium lattice constant of 4.214 Å and bulk modulus of 160.8 GPa. This proves the ionic bonding and excellent incompressibility of MgO. The elastic constants have been calculated in accordance with Born stability, and they show mechanical stability at ambient conditions. A high Debye temperature of 742 K and a melting temperature above 3000 K indicate the material's refractory nature. In order to speed up property prediction, machine learning models were trained on data generated from DFT and then used to predict the same properties - structural, mechanical, and thermal. We see that ML predictions are almost identical to DFT predictions, with errors being less than 2.5%. The similarity between the two models is encouraging and emphasizes how machine learning can predict at first principles very accurately and effectively, and at much lower cost. The present work provides an efficient DFT-ML model for materials screening and offers insights into MgO for engineering and energy applications.