| Details: |
Interfacing crystal structure prediction (CSP) algorithms with ab-initio/force field
methods has provide an avenue for materials discovery by just knowing their chemical
composition/species for diverse applications namely high Tc-superconductors, superhard,
thermoelectric materials and many more. Universal Structure Predictor: Evolutionary
Xstallography (USPEX) based on the evolutionary approach is one of the most successful CSP
algorithm [1] for structure prediction. USPEX is an open source with multiple functionalities, very
user friendly and widely used across the globe. In the first part of my talk, I will describe the
methodology, its recent developments [2], and showcase few studies for prediction of hidden
martensite phase transition in MClF (M = Sr, Ba and Pb) [3] and polymorphs of Urea under high
pressure.
In the second part of my talk, I will discuss on designing materials with an extreme lattice
thermal conductivity (kl) through phonon engineering for thermal energy applications. For
example, ultralow kl is crucial to achieve high figure of merit (ZT) for thermoelectric materials.
Several strategies have been proposed to achieve ultralow kl, among them, the interplay of lone
pair, layered structure, mass contrast between constituent elements in a material provide a pathway
for engineering anharmonicity, bonding heterogeneity, flat and soft phonon bands to increase
phonon-phonon scattering channels to suppress kl [4-6]. I will discuss layered materials consisting
of a lone pair cation with mass contrast [6] which aid to design ultralow kl materials in combination
with ab-initio molecular dynamics (AIMD) simulations, Boltzmann transport theory and
temperature dependent effective potential (TDEP) methods. Later, I will introduce the recent
developments of training on-the-fly machine learning force fields (MLFFs) using AIMD
simulations [7]. The MLFFs allow us to perform MD simulations for longer time and length scales
within classical MD simulation time close to ab-initio accuracy. This would enable us to predict
materials and their properties at finite temperatures, which remains as a longstanding problem in
computational materials science.
1. Oganov, A. R. et al, J. Chem. Phys. 2006, 124, 244704; Lyakhov, A. O. et al, Comput. Phys. Commun. 2013,
184, 1172−1182; Oganov, A. R. et al, Acc. Chem. Res. 2011, 44, 227−237.
2. Allahyari, Z. et al. npj Comput Mater, 2020, 6, 55.
3. Yedukondalu N. et al. 2019, Inorg. Chem. 58, 5886-5899; Lavanya, K. et al, J. Phys. Chem. C, 2021,
125, 31, 17261-17270.
4. Yedukondalu, N.; et al, ACS Applied Mater. & Inter., 2022, 14, 40738-40748.
5. Yedukondalu, N.; et al, arxiv:2205.07091v2.
6. Rakesh Roshan, S. C. et al, ACS Appl. Energy Mater., 2022, 5, 882-890.
7. Liu, P. et al, Phys. Rev. B, 2022, 105, L060102. |