Medicine, information and communication technologies, as well as the fight against climate change require continuous exploration of new inorganic materials. The range of accessible materials is intimately linked to their methods of synthesis, which often involve procedures with a significant environmental impact, motivating the search for organic solvent-free and, if possible, low-temperature processes. The search for alternative synthesis methods and new ways of triggering chemical reactions is opening up new avenues for the production of new solids and materials. We will focus on syntheses in molten salts, which have demonstrated there ability to provide new materials, for example for the production of H2 from water.
While promising, these alternative methods are not widespread enough to be integrated into robotic platforms. What’s more, they often require complex manipulations. So how can we speed up the search for materials using techniques that are difficult to automate?
Our goal is to combine original in situ experimental techniques on synchrotron facilities, with artificial intelligence (AI) applied to data processing on the one hand, and to AI-assisted prediction of solids on the other hand, to accelerate the search for materials.
The MADNESS project draws on PEPR DIADEM‘s infrastructure to provide a methodology for accelerating materials discovery via molten salt synthesis and, more generally, via other synthesis methods unsuitable for automation. In this project, AI assists experimental work, in screening, predicting, anticipating and realizing new functional materials. MADNESS relies on a close link between inorganic synthesis (LCMCP, PI: D. Portehault), advanced characterization (SOLEIL), AI-assisted data processing (IMN) and AI-assisted crystal structure prediction (LINK), so that predicted materials can be realized experimentally.