Abstract
Structure characterization and classification is frequently based on local environment information of all or selected atomic sites in the crystal structure. Therefore, reliable and robust procedures to find coordinated neighbors and to evaluate the resulting coordination pattern (e.g., tetrahedral, square planar) are critically important for both traditional and machine learning approaches that aim to exploit site or structure information for predicting materials properties. Here, we introduce new local structure order parameters (LoStOPs) that are specifically designed to rapidly detect highly symmetric local coordination environments (e.g., Platonic solids such as a tetrahedron or an octahedron) as well as less symmetric ones (e.g., Johnson solids such as a square pyramid). Furthermore, we introduce a Monte Carlo optimization approach to ensure that the different LoStOPs are comparable with each other. We then apply the new local environment descriptors to define site and structure fingerprints and to measure similarity between 61 known coordination environments and 40 commonly studied crystal structures, respectively. After extensive testing and optimization, we determine the most accurate structure similarity assessment procedure to compute all 2.45 billion structure similarities between each pair of the ≈70 000 materials that are currently present in the Materials Project database.
Highlights
Crystal structure databases[1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16] play an increasingly important role in materials science, chemistry, and related elds
We introduce a Monte Carlo optimization approach to ensure that the different local structure order parameters (LoStOPs) are comparable with each other
The steady increase is linked to continuously increasing computing power and memory storage, and it has fostered the creation of many different crystallographic databases that catalog existing materials such as the Cambridge Crystallographic Data Centre (CCDC) in 1965,1 the Inorganic Crystal Structure Database (ICSD) in 1983,2,3 CRYSTMET in 1993,4,5 Pauling File in 2002,6,7 the Crystallography Open Database (COD) in 20038, and the Pearson's Crystal Data (PCD) in 2007.10 computational databases, which mainly use crystallographic databases as a source, and fully hypothetical structure databases are currently being created more and more: the Predicted
Summary
Crystal structure databases[1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16] play an increasingly important role in materials science, chemistry, and related elds. As computational resources still continue to grow[18] and to become more omnipresent and accessible, the computational chemistry, physics, and materials science communities have focused their efforts more and more on automation tools for materials database analysis and on employing statistical and machine learning (SML)[19,20] to help expedite materials discoveries and chemical innovations.[21] This includes, for example, predicting properties (e.g., formation energies, crystal structure dimensionalities, phase diagrams, band gaps, elastic moduli, ionic conductivity) of diverse materials from classes and families such as AX binary compounds,[22] M2AX ternary phases,[23] delafossite and related layered phases of composition ABX2,24 conventional[25] and double perovskite halides (or elpasolites),[26] zeolites[27,28] and other silicates,[29] and other inorganic materials[30,31,32,33,34,35,36] as well as polymers;[37] indicating possible synthesis approaches by screening and predicting synthesis parameters and reactions of inorganic materials,[38,39] metal–organic frameworks,[40] and organic molecules;[41,42] generating interatomic potentials;[43,44,45,46] and expediting ab initio[47,48,49,50] calculations.[51,52,53,54,55]. Because of the plethora of ways to classify structures, de ning and automatically nding prototype structures is currently a very active research area.[60–62] In particular, the usage of coordination number and pattern has culminated in a larger current effort of the community to leverage ngerprinting.[58,63–66] This is the process of combining crucial information about the structure and/or its constituting local environments around each atom into a vector that represents the structure as a whole and includes, for example: a twodimensional ngerprint based on simulated diffraction patterns;[57] the Coulomb matrix;[67] a many-body tensor representation;[68] deep tensor neural networks;[45] Voronoi tessellation;[69] radial distribution functions with[70] and without[71] incorporating partial atomic charges; and local environmentbased crystal ngerprints.[72]
Sign-in/Register to view highlights & summary Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
