Computational prediction of novel MAB phases

The synthesis procedure of any materials system is often considered a challenging task if performed without any prior knowledge. Theoretical models may thus be used as an external input and guide experimental efforts toward novel exotic materials which are most likely to be synthesizable. The aim of this work is to apply theoretical models and develop frameworks for reliable predictions of thermodynamically stable materials. The material in focus herein is the family of atomic layered boride-based materials referred to as MAB phases. The ground state energy of a material system may be obtained by applying firstprincipal calculations, such as density functional theory (DFT), which has thoroughly been used throughout this thesis. However, performing modern state-of-the-art quantum mechanical calculations, in general, relies on a pre-defined crystal structure which may be constructed based on an a priori known structure or obtained through the use of crystal structure prediction models. In this work, both approaches are explored. We herein perform a thermodynamical screening study to predict novel stable ternary boron-based materials by considering M2AB2, M3AB4, M4AB6, MAB and M4AB4 compositions in orthorhombic and hexagonal symmetries with inspiration from experimentally synthesized MAB phases. The considered atomic elements are M = Sc, Y, Ti, Zr, Hf, V, Nb, Ta, Cr, Mo, W, Mn, Fe, Co, A = Al, Ga, In, and B is boron. Among the considered compounds, seven experimentally synthesized phases are verified as stable, and we predict the three hypothetical phases to be stable - Hf2InB2, Zr2InB2, and Mo4AlB4. Additionally, 23 phases of varying symmetries and compositions are predicted as close to stable or to be metastable. However, the assumption of assigning initial crystal structures based on neighbouring compounds may drastically limit the outcome of a screening study. State-of-the-art techniques to generate low energy crystal structures within the considered material phase space is thus explored. More specifically, the Mo-Sc-Al-B system is studied along the ternary joints of (MoxSc1-x)2AlB2 where 0 < x < 1 by using the cluster expansion (CE) and the crystal structure prediction (CSP) codes, CLEASE and USPEX, in analogy. Previous attempts to study the Mo-Sc-Al-B system has been limited by only considering either hexagonal or orthorhombic symmetries. We challenge such approaches by covering larger portions of the phase space efficiently by combining CSP and CE frameworks. The Mo4/3Sc2/3AlB2 (R ̅3m) phase, previously referred to as i-MAB, is verified stable in addition to Mo2/3Sc4/3AlB2 (R3). The suggested approach of combining CE and CSP frameworks for investigating multi-component systems consists of initially performing CSP searches on the systems of smaller order constituting the system in focus. In the pseudo-ternary (MoxSc1-x)2AlB2 system, this refers to performing CSP searches on the ternary Mo2AlB2 and Sc2AlB2 systems. In addition, we also consider the structures of experimentally known phases with similar compositions. The complete set of structures obtained either from CSP or public databases, was later used to design CE models where mixing tendencies in addition to stability determined which model to further study. The predicted low-energy structures of the CE model were relaxed and used as seed structures within a complete CSP search covering the (MoxSc1-x)2AlB2 system for 0 < x < 1. We demonstrate that the use of seed structures, obtained from CE models, efficiently improved the search for low-energy structures within a multi-component system. The suggested approach is yet to be tested on any other system but is applicable to any alternative multi-component system