Free-photo; pixabay.com; CC0 Seyed Mohamad Moosavi, Arunraj Chidambaram, Leopold Talirz, Maciej Haranczyk, Kyriakos C. Stylianou & Berend Smit
Graphic showing how the three components of the framework - synthesis, optimization, and machine learning - interact.
Researchers from the lab of NCCR MARVEL's deputy director Berend Smit and colleagues have developed a methodology for collecting the lessons learned from partially failed trials and incorrect hypotheses -- the experiments that didn't work.
The researchers used machine learning to capture chemical intuition -- which they defined as the collection of unwritten guidelines chemists use to find the right synthesis conditions -- from a set of (partially) failed attempts to synthesize a metal-organic framework.
Since these experiments are usually unreported, they reconstructed a typical track of failed experiments in the successful search for the optimal synthesis conditions for yielding a certain MOF with the highest surface area reported to date. They go on to show how important quantifying this chemical intuition is in the synthesis of novel materials.