Predicting synthesizability of crystalline materials via deep learning

Ali Davariashtiyani,Zahra Kadkhodaie,Sara Kadkhodaei

doi:10.1038/s43246-021-00219-x

Ali Davariashtiyani, Zahra Kadkhodaie + Show 1 more

Open Access

https://doi.org/10.1038/s43246-021-00219-x

Copy DOI

Abstract

Predicting the synthesizability of hypothetical crystals is challenging because of the wide range of parameters that govern materials synthesis. Yet, exploring the exponentially large space of novel crystals for any future application demands an accurate predictive capability for synthesis likelihood to avoid a haphazard trial-and-error. Typically, benchmarks of synthesizability are defined based on the energy of crystal structures. Here, we take an alternative approach to select features of synthesizability from the latent information embedded in crystalline materials. We represent the atomic structure of crystalline materials by three-dimensional pixel-wise images that are color-coded by their chemical attributes. The image representation of crystals enables the use of a convolutional encoder to learn the features of synthesizability hidden in structural and chemical arrangements of crystalline materials. Based on the presented model, we can accurately classify materials into synthesizable crystals versus crystal anomalies across a broad range of crystal structure types and chemical compositions. We illustrate the usefulness of the model by predicting the synthesizability of hypothetical crystals for battery electrode and thermoelectric applications.

Highlights

Predicting the synthesizability of hypothetical crystals is challenging because of the wide range of parameters that govern materials synthesis
The basis of this framework is that the zerotemperature enthalpy of the amorphous phase provides an accurate upper bound for the Gibbs energy of synthesizable crystals at any temperature due to the inevitably larger entropy of the amorphous solid compared to ordered crystals
We present a deep-learning model that can predict the synthesizability of hypothetical crystalline materials

Summary

Introduction

Predicting the synthesizability of hypothetical crystals is challenging because of the wide range of parameters that govern materials synthesis. In a later study[7], the energy of the amorphous solid (or super-cooled liquid state) with a given chemical composition was used as the limit on the energy scale useful for establishing a necessary condition for synthesis. Aykol et al.[8] modeled the free energy convex hull in the composition space that encompasses the chronological discovery timeline of each composition by an evolving network, where the nodes encode the convex hull and the edges encode the circumstantial factors They utilized their model to predict the likelihood of successful experimental synthesis of hypothetical materials. The machine learning framework used in this study can be extended to serve as a predictive tool for the synthesizability likelihood across a wide range of crystalline materials, from elemental, ionic, and covalent crystals to complex molecular crystals

Methods

Results

Conclusion