Chemistry Seminar - İlker Demiroğlu

In order to understand the properties of materials, their atomic configurations must be known. One of the simplest to describe but difficult to solve problem in computational chemistry is the determination of atomic configuration. Any computational approach to study a chemical system requires a mathematical model to calculate the energy of the system as a function of its configuration. The success of the approach will depend on the quality of the mathematical model used. For smaller chemical systems studied, quantum mechanical (QM) approaches are appropriate and feasible. However, some of the chemical or biological processes require higher time scales and thus, the energy function might be subjected to over huge numbers of energy evaluations. Large numbers of evaluations are also required for using global optimization techniques to search the configurational energy landscapes. Machine learning type potentials are recently very promising to accurately describe such systems. In this talk several examples of structure and property prediction for different nanostructured systems will be given for different scales ranging from 0D nanoclusters to 3D porous nanocrystals. The talk will thus cover the applications of several computational tools and models such as force fields, density functional theory, machine learning, global optimization and molecular dynamics approaches to specific systems. The subjects covered in the talk will be: (I) Effect of dimensionality and polymorphism on the properties of metal oxides, (II) Estimation of catalytic properties of bimetallic nanoclusters, (III) Alkali metal intercalation properties of novel 2D materials, and (IV) Machine learning type gaussian approximation potentials to estimate thermal expansion properties of 2D materials.