Alexandria Burger1,Shuang Tang1,Andrew Lemieszewski1
State University of New York Polytechnic Institute1
Alexandria Burger1,Shuang Tang1,Andrew Lemieszewski1
State University of New York Polytechnic Institute1
The trend of novel electronic materials development is transitting from thin film based chips into smaller scaled materials systems, such as nanowires, nanotubes and nanoribbons. Among many materials candidates, bismuth antimony nanowires are promising ultra-high mobility, low power consumption, and a great richness of electronic phases and band-edge configurations. When the alloy materials system is made into nanowires, the symmetry between the three L-points and the six H-points in the first Brillouin zone can be either kept or broken, providing much larger flexibility than silicon in chip and circuit designs.<br/><br/>Semi-metals, direct-band-gap semiconductors, and indirect-band-gap semiconductors can coexist in one system. One bottle-neck in materials design is accurately predicting the electronic phase and the band structure. For bulk and pure materials, the band structure can be well described using first principle calculations. However, when the materials are alloyed and made into nanostructures, the calculational time that is needed may not be feasible with the current computer power. In the recent decade, pattern recognition techniques and other artificial intelegence have been widely developed and utilized in research of condensed matter physics and materials science.<br/><br/>In this present work, we will use multiple pattern recognition tools, including the support vectors, the trees, the bagging method, as well as the artificial neural networks, to realize the accurate and convenient describing of electronic phases and band structure of bismuth antimony nanowires. Different values of alloy composition, growth orientation, and wire diameter will be discusses. We will also discuss how the size of training pool will affect the predicting accuracy when different tools are used.