Materials cartography: A forward-looking perspective on materials representation and devising better maps

Abstract

Machine learning (ML) is gaining popularity as a tool for materials scientists to accelerate computation, automate data analysis, and predict materials properties. The representation of input material features is critical to the accuracy, interpretability, and generalizability of data-driven models for scientific research. In this Perspective, we discuss a few central challenges faced by ML practitioners in developing meaningful representations, including handling the complexity of real-world industry-relevant materials, combining theory and experimental data sources, and describing scientific phenomena across timescales and length scales. We present several promising directions for future research: devising representations of varied experimental conditions and observations, the need to find ways to integrate machine learning into laboratory practices, and making multi-scale informatics toolkits to bridge the gaps between atoms, materials, and devices.

Autonomous laboratories for accelerated materials discovery: a community survey and practical insights

“…This type of tighter integration is already being demonstrated in predictive synthesis. 25–30 There is still a critical gap between simulation and experiment, largely due to differences in materials representation, 31 but automated labs that integrate theory and experiment enable the creation of new multimodal datasets and models that can aid in the construction of a large experimental knowledge graph 32 and ultimately improve our fundamental understanding of materials.…”

Surface Segregation Studies in Ternary Noble Metal Alloys: Comparing DFT and Machine Learning with Experimental Data

“…[3,4] However, even as the accuracy of machine learned DFT surrogate models improves, important challenges remain in applying these models to understanding or supplementing real-world data to achieve rational materials design. [5] One challenge for accurate materials simulations is the need to model disorder, which is computationally costly and can stymie materials discovery efforts that assume materials are ordered crystals. [6] A challenge more specific to ab initio catalyst design is the need to understand surface stability.…”

Physics-Informed Design of Hybrid Pulse Power Characterization Tests for Rechargeable Batteries

“…Lithium-ion batteries, owing to their high power and energy densities, have become ubiquitous energy storage devices for portable electronic devices. 1,2 A key metric in the design of Li-ion battery materials is rate capability for discharge, [3][4][5] but there is a complex, material-dependent trade-off between increased cycling rates and reduced battery lifetime, strongly correlated with power fade in electrodes. 6 Capacity fade and internal degradation resulting from long-term use of Li-ion batteries must be rapidly and accurately quantified in order to improve their performance, reliability, and safety [7][8][9][10] and inform second-use and end-of-life decisions.…”