Networks of materials: a complexity approach to material synthesizability

Network science can provide a new arsenal of methods for materials informatics to address some of the daunting challenges in the field of materials.[1,2] One such challenge we address in this talk is the laboratory synthesis of computationally designed materials. The art of synthesis itself is too complex to be treated with an all-encompassing, quantitative approach as yet, because it involves not only the modeling of the energy landscape of competing phases or the kinetics of transformations, but also abstract factors, such as experience of scientists, state-of-the-art in experimental methods, resources and more. We present one of the first examples of “networks of materials” in the form of thermodynamic equilibria derived from high-throughput computational phase diagrams and analyze the local and global properties of this network.[1] We discuss how synthesis probabilities of yet-to-be-made, novel inorganic materials can be predicted from the dynamics of this network via machine-learning, connecting high-throughput computational design and experiments.
[1] M. Aykol, V. I. Hegde, S. Suram, L. Hung, P. Herring, C. Wolverton, J. S. Hummelshøj, arXiv:1806.05772.
[2] V. I. Hegde, M. Aykol, S. Kirklin, C. Wolverton, arXiv:1808.10869.