Finding the needle in the haystack: Materials discovery and design through computational ab initio high-throughput screening

Hautier, Geoffroy

doi:10.1016/j.commatsci.2019.02.040

Cited by 34 publications

(17 citation statements)

References 51 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Any high-throughput effort requires a highly automated framework for launching, monitoring, parsing, and storing a large number of calculations on many structures. Recording explicitly, ideally in an easily queryable format, the provenance of the resulting data allows for fully reproducible results [40,[92][93][94][95]. To achieve automation and explicit storage of the provenance, we leverage the Automated Interactive Infrastructure and Database for Computational Science (AiiDA) materials informatics platform, developed by Pizzi et al [39] The novelty of AiiDA in the field of materials informatics is that every calculation is stored as a node in a graph, with input data forming incoming nodes, and output data stored as outcoming nodes, that can again be input to a different calculation.…”

Section: Automation and Provenancementioning

confidence: 99%

High-throughput computational screening for solid-state Li-ion conductors

Kahle

Marcolongo

Marzari

2020

Energy Environ. Sci.

124

106

View full text Add to dashboard Cite

We present a computational screening of experimental structural repositories for fast Li-ion conductors, with the goal of finding new candidate materials for application as solid-state electrolytes in next-generation batteries. We start from ∼1400 unique Licontaining materials, of which ∼900 are insulators at the level of density-functional theory. For those, we calculate the diffusion coefficient in a highly automated fashion, using extensive molecular dynamics simulations on a potential energy surface (the recently published pinball model) fitted on first-principles forces. The ∼130 most promising candidates are studied with full first-principles molecular dynamics, first at high temperature and then more extensively for the 78 most promising candidates. The results of the first-principles simulations of the candidate solid-state electrolytes found are discussed in detail.of superionic conductors is the recent discovery of Li-argyrodites [22], with the general formula Li 7 PS 5 X (X=Cl, Br, I) or Li 7 PS 6 (sulphur can be replaced with oxygen, but this reduces the ionic conductivity [23]). Third, Li-containing NASICONs (sodium superionic conductors) are phosphates with the structural formula Li 1+6x X 4+2−x Y 3+x (PO

show abstract

Section: Automation and Provenancementioning

confidence: 99%

High-throughput computational screening for solid-state Li-ion conductors

Kahle

Marcolongo

Marzari

2020

Energy Environ. Sci.

124

106

View full text Add to dashboard Cite

show abstract

“…7(a). A screening funnel is a series of criteria that a material must pass in order to be considered for a certain application [63]. Here, we consider an ab initio computational screening procedure using DFT that can be used by other researchers while searching for new hole-selective contacts.…”

Section: Prospective: Exploring New Hole-selective Contacts In Shjmentioning

confidence: 99%

Evaluating Materials Design Parameters of Hole-Selective Contacts for Silicon Heterojunction Solar Cells

Woods‐Robinson¹,

Fioretti²,

Haschke³

et al. 2021

IEEE J. Photovoltaics

View full text Add to dashboard Cite

Silicon heterojunction (SHJ) solar cell efficiencies are limited by parasitic absorption from the hydrogenated amorphous silicon (a-Si:H) front contact, but this may be mitigated by selecting an alternative carrier selective contact material with a wider band gap. When choosing such a material as the hole-selective contact ("p-layer"), the alignment of the material's valence band edge energy (E VB ) with that of crystalline silicon (c-Si) is an important criterion, but several other material parameters can also influence the band bending at the contact interface. In this article, we simulate an (n)c-Si/(i)a-Si:H/p-layer interface to explore the influence of six materials parameters in a variable p-layer on the SHJ performance. We find a strong influence on the fill factor (FF) from thickness, doping, and E VB ,and on V OC from the interfacial defect density; notably, optimal E VB is ∼0.1 eV higher than the valence band edge energy of a-Si:H. Multiparameter sensitivity analyses demonstrate how performance is simultaneously influenced by E VB and doping; thus, both parameters should be optimized alongside one another. To assess the influence of these parameters experimentally, we grow p-type NiO x as a test-case p-layer, which shows that FFs decrease

show abstract

“…2,3 These high throughput techniques have been effectively deployed in a tiered screening strategy wherein high-speed methods that may sacrifice some accuracy and/or precision can effectively down-select materials that merit detailed attention from more resource-intensive methods. [4][5][6] The recent advent of machine learning prediction of materials properties has introduced the possibility of even higher throughput primary screening due to the minuscule expense of making a prediction for a candidate material using an already-trained model. [7][8][9][10] Toward this vision, we introduce the materials to spectrum (Mat2Spec) framework for predicting spectral properties of crystalline materials, demonstrated herein for the prediction of the ab initio phonon and electronic densities of state.…”

Section: Introductionmentioning

confidence: 99%

Density of States Prediction for Materials Discovery via Contrastive Learning from Probabilistic Embeddings

Kong,

Ricci,

Guevarra

et al. 2021

Preprint

View full text Add to dashboard Cite

Machine learning for materials discovery has largely focused on predicting an individual scalar rather than multiple related properties, where spectral properties are an important example. Fundamental spectral properties include the phonon density of states (phDOS) and the electronic density of states (eDOS), which individually or collectively are the origins of a breadth of materials observables and functions. Building upon the success of graph attention networks for encoding crystalline materials, we introduce a probabilistic embedding generator specifically tailored to the prediction of spectral properties. Coupled with supervised contrastive learning, our materials-to-spectrum (Mat2Spec) model outperforms state-of-the-art methods for predicting ab initio phDOS and eDOS for crystalline materials. We demonstrate Mat2Spec's ability to identify eDOS gaps below the Fermi energy, validating predictions with ab initio calculations and thereby discovering candidate thermoelectrics and transparent conductors. Mat2Spec is an exemplar framework for predicting spectral properties of materials via strategically incorporated machine learning techniques.

show abstract

Finding the needle in the haystack: Materials discovery and design through computational ab initio high-throughput screening

Cited by 34 publications

References 51 publications

High-throughput computational screening for solid-state Li-ion conductors

High-throughput computational screening for solid-state Li-ion conductors

Evaluating Materials Design Parameters of Hole-Selective Contacts for Silicon Heterojunction Solar Cells

Density of States Prediction for Materials Discovery via Contrastive Learning from Probabilistic Embeddings

Contact Info

Product

Resources

About