Assistant Professor of Physics
Search for Stable Crystal Structures with an Evolutionary Algorithm: Prediction and Discovery of a New Superconductor
March 17, 2014
Academic Building A G008, 5:00 PM
Recent advances in compound prediction methods have changed the way materials science is conducted. High-throughput and targeted searches powered by machine learning and evolutionary algorithms can considerably reduce the experimental cost in the development of high-performance materials. First theory-guided discoveries have included thermoelectric, catalytic, and battery materials. However, design of superconductors ‘from scratch’ has remained elusive due to the difficulty of finding new candidate materials that are both superconducting and synthesizable. I will describe our computational work that led to identification of a previously unknown stable FeB4 superconductor via a combination of high-throughput and evolutionary searches [1-3]. Calculation of FeB4 properties indicated that its superconductivity would be defined by phonon-mediated (aka conventional) mechanism which was unexpected for a material containing iron. My collaborators have recently synthesized the material and confirmed its predicted structural and superconducting properties . In addition, the material has been shown to be exceptionally hard. The discovery makes FeB4 the first superconductor designed fully in silico .
 A. N. Kolmogorov et al., Phys. Rev. Lett., 105, 217003 (2010)
 A. F. Bialon et al., Appl. Phys. Lett., 98 081901 (2011)
 A. N. Kolmogorov, Module for Ab Initio Structure Evolution (2009) http://maise-guide.org
 H. Gou et al. Phys. Rev. Lett., 111, 157002 (2013)
 F. Ronnig and J. L. Sarrao, Physics 6, 109 (2013) http://physics.aps.org/
Dr. Alexey Kolmogorov earned a M.S. in Physics from Moscow Institute and Physics and Technology in 1995. He was awarded a Landau Scholarship to continue his research in experimental Condensed Matter Physics at Kapitza Institue for Physical Problems. Having entered the Ph.D. program at Penn State University, Dr. Kolmogorov started working in computational Physics and did most of his studies on carbon nanomaterials. He completed the graduate program in 2004 with a Ph.D. in computational Condensed Matter Physics and a minor in Computer Science. The following postdoctoral work on the development of new materials from first principles was done in Departments of Materials, at Duke University (2004-2007) and the University of Oxford (2007-2008). In 2008, Dr. Kolmogorov was awarded a prestigious EPSRC Career Acceleration Fellowship in the UK. As a Senior Research Fellow at the University of Oxford, he led the research on the high-throughput identification of novel materials for superconductivity and hydrogen storage applications. In 2012 Dr. Kolmogorov joined the Department of Physics at Binghamton as an Assistant Professor of Physics and has expanded his research to the development of catalysts and battery materials. A major part of his research is dedicated to the adaptation of bio-inspired methods, such as evolutionary algorithms and neural networks, to Condensed Matter Physics problems.
Reading will be posted to the EvoS blackboard group. Anyone with a Binghamton University email address can request to be added to the blackboard group by emailing EvoS[at]binghamton[dot]edu.
EchoCenter Course Portal