Computational properties of one-dimensional materials

In the work leading to this thesis, we conducted a systematic investigation of the properties of around 2000 one-dimensional materials using the density functional theory (DFT) method. Our aim was to accurately calculate the structural, thermodynamic, electronic, magnetic, and optical properties of these materials. To discover one-dimensional materials, we perform an extensive screening of experimental databases with structural information. This screening entails a thorough examination of the atomic structure to distinguish low-dimensional materials. The materials that pass the screening form the fundamental components of the database and which we referred to as the ”core” materials. Then we expand the database by substituting atoms with comparable chemical properties to generate novel materials, which we referred to as the ”shell” materials. Additionally, we employ a generative neural network model that trains on the database of core and shell materials to create new structures, and we also systematically calculated their properties. Some of those materials could be obtained by chemical element substitution in the training materials, but we also produced completely new classes of materials. We refer to those materials as the ”Machine learning generated” materials.
We thoroughly investigated the global (thermodynamic) and local (dynamic) stability of the structures to confirm their physical soundness, and the results showed hundreds of novel, very stable one-dimensional materials that may potentially be synthesized. Then having access to a comprehensive structured database allows us to explore the connections between material structure and their properties. By utilizing this approach, we can for example make precise predictions about the energies above the convex hull of 1D materials with a given composition of chemical elements. Moreover, we can explore the connections between two different properties, like thermodynamic and dynamic stability. This exploration would perhaps give us the opportunity to rapidly assess the stability of many thousands of newly generated materials using machine learning.
One-dimensional materials offer a plethora of exciting possibilities, three of which we describe in particular. The first is the potential for Majorana-bound states, which we identified by making a computational screening of materials that may exhibit this phenomenon. The second is the light absorption of 1D materials which are used in photovoltaics and photocatalysis to convert visible light into energy. The goal is to make them more efficient, stable, and cost-effective. However, defects can also affect the absorption process, which has been studied in detail for one-dimensional Selenium. Finally, the ability of one-dimensional materials to bind to gas atoms is an area that has yet to be fully explored due to the challenges presented by the small size of the atoms involved and the specific equipment and techniques required to examine these interactions at the atomic level. Despite these difficulties, chemisorbed atoms on one-dimensional materials could be an exciting platform for the discovery of new compounds with useful chemical properties. In this work, we have investigated chemisorption on nine different one-dimensional materials.