Abstract
The information society has an ever-increasing need for computational resources, as emerging technologies such as artificial intelligence become increasingly important and transform the world we live in. The growing energy demands of the data centers supporting these technologies make it necessary to explore new computational paradigms and materials that could reduce power consumption. Furthermore, existing semiconductor technologies for memory and logic are approaching their physical limits in terms of miniaturization.
The magnetic properties of materials around us have fascinated mankind for millennia. In recent decades, progress has been made in utilizing the electron’s spin degree of freedom, enabling new technologies for memory and logic with the potential to reduce energy consumption to a fraction of that of conventional charge-based technologies. Antiferromagnetic materials, in particular, offer properties that are beneficial for the miniaturization of electronic components and high-speed operation. Their negligible stray fields allow components to be packed more closely together, while their magnetic structures are generally robust against external magnetic fields. The challenge, however, remains how to control their magnetic structures.
Strain engineering has proven effective in controlling functional properties in perovskite oxide-based electronic materials. This thesis demonstrates how the same approach can be used to modify existing properties and induce new ones in less-explored materials. Using computational modeling, it investigates strain-induced magnetic phases and structural changes in two materials systems: the antiperovskite Mn3GaN, known for its noncollinear antiferromagnetic order and investigated for its potential in spin-based electronics, and the semiconductor CuFeS2, known for its thermoelectric properties and its antiferromagnetic order at room temperature.
The results demonstrate that strain engineering can induce new magnetic phases in these materials, as well as new ferroic properties, such as piezomagnetism and piezoelectricity, that are not allowed by the symmetry of their naturally occurring phases. These findings expand the range of possible mechanisms for implementing more energy-efficient computational paradigms in the coming decades.