Chemically directed structure evolution for crystal structure prediction

Sharp, Paul M.; Dyer, Matthew S.; Darling, George R.; Claridge, John B.; Rosseinsky, Matthew J.

doi:10.1039/d0cp02206c

Cited by 14 publications

(18 citation statements)

References 46 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We employ Crystal Structure Prediction (CSP) for 244 different compositions in the Li-Sn-S-Cl phase field (cyan tripods in Fig. 2b ) with the complementary algorithms of basin-hopping ChemDASH 22 , 39 , which explores the energy landscape, with starting ionic configurations based on hcp and rhombohedral lattices and evolutionary XtalOpt 40 exploring random and mutated (hybrid) ionic configurations drawn from a population of structures, spanning in total up to 1200 structures for each composition to maximize coverage of possible ionic arrangements. Both methods are coupled with VASP 41 for energy computation at the DFT level of accuracy (cf.…”

Section: Resultsmentioning

confidence: 99%

“…In CSP with ChemDASH 39 , for each composition, the structure was initialized with anions (S 2− , Cl 1− ) located on a 2 × 2 × 2 or 3 × 2 × 2 grid in a close-packed arrangement and cations (Li 1+ , Sn 4+ ) occupying the interstitial sites. Up to 600 Li-Sn-vacancy and S-Cl atomic swaps were performed from initial structures and optimized geometrically with VASP to produce candidate structures for each composition.…”

Section: Methodsmentioning

confidence: 99%

See 1 more Smart Citation

Element selection for crystalline inorganic solid discovery guided by unsupervised machine learning of experimentally explored chemistry

et al. 2021

Self Cite

View full text Add to dashboard Cite

The selection of the elements to combine delimits the possible outcomes of synthetic chemistry because it determines the range of compositions and structures, and thus properties, that can arise. For example, in the solid state, the elemental components of a phase field will determine the likelihood of finding a new crystalline material. Researchers make these choices based on their understanding of chemical structure and bonding. Extensive data are available on those element combinations that produce synthetically isolable materials, but it is difficult to assimilate the scale of this information to guide selection from the diversity of potential new chemistries. Here, we show that unsupervised machine learning captures the complex patterns of similarity between element combinations that afford reported crystalline inorganic materials. This model guides prioritisation of quaternary phase fields containing two anions for synthetic exploration to identify lithium solid electrolytes in a collaborative workflow that leads to the discovery of Li3.3SnS3.3Cl0.7. The interstitial site occupancy combination in this defect stuffed wurtzite enables a low-barrier ion transport pathway in hexagonal close-packing.

show abstract

Section: Resultsmentioning

confidence: 99%

Section: Methodsmentioning

confidence: 99%

Element selection for crystalline inorganic solid discovery guided by unsupervised machine learning of experimentally explored chemistry

et al. 2021

Self Cite

View full text Add to dashboard Cite

show abstract

“…It has been shown that aligning the initial molecular dynamics velocities along the directions of soft phonon modes helps accelerate the search. , The minima hopping method has been used extensively to predict the structures of materials under pressure such as superconducting S x Se (1– x ) H 3 phases, binary intermetallics that are immiscible at 1 atm, and structural candidates for cold compressed graphite . Minima hopping should not be confused with the similarly named basin hopping method, which has been applied extensively to finite clusters and water-ice, and it is under active development for more complex systems. − …”

Section: Computational Detailsmentioning

confidence: 99%

The XtalOpt Evolutionary Algorithm for Crystal Structure Prediction

et al. 2020

View full text Add to dashboard Cite

Significant progress has been made in the field of a priori crystal structure prediction, with a number of recent remarkable success stories. Herein, we briefly outline the methods that have been developed for finding the global minimum structure and interesting local minima without the need for experimental information. Focus is placed on describing the XtalOpt evolutionary algorithm (EA) developed in our group toward this end. XtalOpt is published under well-known open-source licenses, and the EA searches can be analyzed via the Avogadro chemical editor and visualizer. We describe new algorithmic developments that have made it possible to predict the structures of ever-more complex crystalline lattices. Benchmark tests, which clearly illustrate how the new developments improve the success rate and accelerate the discovery of the global minimum structure, are performed. Finally, we describe how XtalOpt has been employed to predict novel ternary hydrides that have the propensity for high-temperature superconductivity under pressure.

show abstract

“…10,12 To improve the sampling efficiency, a variety of strategies have been proposed such as exploiting symmetry 15,16 and pseudosymmetry, 10 smart variation operators, clustering, 17 machine-learning interatomic potentials with active learning, 18 and designing chemically based swapping operators. 19 However, the scalability of these ab initio approaches remains an unsolved issue.…”

Section: Introductionmentioning

confidence: 99%

“…In the past 10 years, CSP algorithms based on evolutionary algorithms and particle swarm optimization have led to a series of new materials discoveries. ,, However, these ab initio free energy-based global search algorithms have a major challenge that limits their success to relatively simple crystals (mostly binary materials or compounds with less than 20 atoms in the unit cell , ) because of their dependence on the costly DFT calculations of free energies for sampled structures. With a limited DFT calculation budget, it is a key issue to efficiently sample the atom configurations. , To improve the sampling efficiency, a variety of strategies have been proposed such as exploiting symmetry , and pseudosymmetry, smart variation operators, clustering, machine-learning interatomic potentials with active learning, and designing chemically based swapping operators . However, the scalability of these ab initio approaches remains an unsolved issue.…”

Section: Introductionmentioning

confidence: 99%

Distance Matrix-Based Crystal Structure Prediction Using Evolutionary Algorithms

Yang

Siriwardane

2020

J. Phys. Chem. A

View full text Add to dashboard Cite

Crystal structure prediction (CSP) for inorganic materials is one of the central and most challenging problems in materials science and computational chemistry. This problem can be formulated as a global optimization problem in which global search algorithms such as genetic algorithms (GAs) and particle swarm optimization have been combined with first-principles free-energy calculations to predict crystal structures given only the material composition or a chemical system. These DFT-based ab initio CSP algorithms are computationally demanding and can usually be used only to predict crystal structures of relatively small systems. The vast coordinate space and the expensive DFT free-energy calculations limit their inefficiency and scalability. On the other hand, a similar structure prediction problem has been intensively investigated in parallel in the protein structure prediction (PSP) community of bioinformatics, in which the dominating predictors are knowledge-based approaches including homology modeling and threading that exploit known protein structures. Surprisingly, the CSP field has mainly focused on ab initio approaches in the past decade. Inspired by the knowledge-rich PSP approaches, herein, we explore whether known geometric constraints such as the pairwise atomic distances of a target crystal material can help predict/reconstruct its structure given its space group and lattice information. We propose DMCrystal, a GA-based crystal structure reconstruction algorithm based on predicted pairwise atomic distances. Based on extensive experiments, we show that the predicted distance matrix can dramatically help reconstruct the crystal structure and usually achieves much better performance than that of CMCrystal, an atomic contact map-based CSP algorithm. This implies that the knowledge of atomic interaction information learned from the existing materials can be used to significantly improve the CSP in terms of both speed and quality.

show abstract

Chemically directed structure evolution for crystal structure prediction

Abstract: The chemically directed structure evolution method uses chemical models to quantify the environment of atoms and vacancy sites in a crystal structure with that information used to inform how to modify the structure for crystal structure prediction.

Cited by 14 publications

References 46 publications

Element selection for crystalline inorganic solid discovery guided by unsupervised machine learning of experimentally explored chemistry

Element selection for crystalline inorganic solid discovery guided by unsupervised machine learning of experimentally explored chemistry

The XtalOpt Evolutionary Algorithm for Crystal Structure Prediction

Distance Matrix-Based Crystal Structure Prediction Using Evolutionary Algorithms

Contact Info

Product

Resources

About