Network analysis of synthesizable materials discovery

Abstract

Assessing the synthesizability of inorganic materials is a grand challenge for accelerating their discovery using computations. Synthesis of a material is a complex process that depends not only on its thermodynamic stability with respect to others, but also on factors from kinetics, to advances in synthesis techniques, to the availability of precursors. This complexity makes the development of a general theory or first-principles approach to synthesizability currently impractical. Here we show how an alternative pathway to predicting synthesizability emerges from the dynamics of the materials stability network: a scale-free network constructed by combining the convex free-energy surface of inorganic materials computed by high-throughput density functional theory and their experimental discovery timelines extracted from citations. The time-evolution of the underlying network properties allows us to use machine-learning to predict the likelihood that hypothetical, computer-generated materials will be amenable to successful experimental synthesis.

Machine Learning Approaches for Thermoelectric Materials Research

“…Iwasaki et al trained four machine learning methods to predict the thermopower of spin‐driven thermoelectric phenomena, and due to its much higher data modeling ability, the NN achieved the best accuracy . Aykol et al proposed to reinforce the integration of material informatics theory, but experiment emphasized that it was challenging but critical for materials research . Kolb et al developed PROPerty Prophet (PROPhet), a framework based on a fully connected neural network, to predict the material properties.…”

“…The materials data and informative landscape are the most important sources of the machine learning method . For example, Aykol et al utilized abundant data from the open quantum materials database (OQMD) to establish new predictive methods in combination with machine learning because it is currently impractical to develop a first‐principles method of synthesizability . Katsura et al developed a web system, namely, Starrydata2, to speed up the collection of experimental data from published papers .…”

MaterialNet: A web-based graph explorer for materials science data

“…• MaterialNet has been architected to support different types of data. The current deployment at http://maps.matr.io/ includes the materials stability network (Aykol, Hegde, et al, 2019), a text co-occurrence network extracted from MatScholar Weston et al, 2019), and a materials similarity network (Ward et al, 2017), but the tool can be easily extended to display any other type of material network as well. • A more powerful and flexible search mode, featuring a domain-specific language tailored to searching materials databases, will extend the researcher's ability to find materials with very specific properties or ranges of properties.…”

“…With recent advances in computational power and automation of simulation techniques, material structure and property databases have emerged (Curtarolo et al, 2012;Jain et al, 2013;Kirklin et al, 2015), allowing a more data-driven approach to carrying out materials research. Recent studies have demonstrated that representing these databases as material networks can enable extraction of new materials knowledge (Hegde, Aykol, Kirklin, & Wolverton, 2018;Isayev et al, 2015) or help tackle challenges like predictive synthesis (Aykol, Hegde, et al, 2019) that require relational information between materials. Materials databases have become very popular because they enable their users to do rapid prototyping by searching near globally for figures of merit for their target application.…”

“…MaterialNet is an open-source web application designed to explore the topology of such material networks, while also displaying information about each material, highlighting its immediate neighborhood within the graph, and offering several auxiliary tools to help drill down into the details contained within the dataset. Such graph-encoded datasets for a collection of materials can become large and complex: for example, the materials stability network exemplified in this paper contains on the order of 20,000 materials, with on the order of 200,000 links between them (Aykol, Hegde, et al, 2019). Large graphs of this sort demand interactive visualization, empowering materials researchers to explore the data, a key user requirement highlighted by field experts .…”