Collaborative efforts to develop centralized databases have become common in some fields. To be useful, databases need appropriate support and to be annotated using standardized approaches. Whether time and resources are well spent on these tasks depends on the value gained from exploring vast data repositories.
Three papers in Nature this week show that materials science is reaping the rewards of such investment (see T. Zhang et al., M. G. Vergniory et al. and F. Tang et al.). The teams developed algorithms to scan through tens of thousands of non-magnetic materials catalogued in crystal-structure databases and, astonishingly, found that around one-quarter could be considered ‘topological’ — harbouring unusual states at their surfaces or edges that are caused by the geometry of their electronic structures.
The unusual properties of topological systems offer new possibilities for materials engineering, including the design of energy-efficient transistors and circuits. Yet only a few topological materials have been identified.
The findings in Nature are theoretical and, as physicist Alex Zunger points out in a Comment, many such materials might be difficult to synthesize, or could turn out not to have the predicted properties when tested experimentally. Even if researchers need to curb their enthusiasm down the line, the unimagined scale of this discovery was possible only because of the existence of large crystallographic databases.
Just as data sharing has facilitated these theoretical predictions, it will also benefit future materials discovery and engineering. But to achieve this, researchers will need to systematically release all underlying data. Topological catalogues are still in the early stages of development, so this community would do well not to miss the opportunity to push for widespread and standardized sharing of experimental data.
Nature 566, 425 (2019)