MOF discovery has long run on manual synthesis and educated guessing. The variables that decide whether a crystal even forms, which phase wins out, and how fast it gets there are many, and testing them one by one at the bench is slow. So most of that parameter space stays unexplored, and the rules behind rare earth MOF crystallization stay hidden. The team set out to change that by putting automation on the "make" side and pairing it with characterization that could keep up.
- Built a high-throughput MOF discovery workflow that pairs automated solvothermal synthesis with scalable characterization, so many conditions run in parallel instead of one at a time.
- Used Chemspeed automation to run the synthesis side reproducibly across a wide grid of metals, ligands, and reaction conditions, mapping structure, kinetics, and selectivity together.
- Applied high-throughput powder X-ray diffraction and fast pattern matching to sort each product against known MOF structures.
- Identified five previously unreported rare earth MOFs (NU-2501 to NU-2505), confirmed by single-crystal X-ray diffraction.
- Turned scattered results into phase diagrams that show which crystallization outcomes to expect, and why.
When the make step runs itself on a Chemspeed platform, the crystallization landscape stops being a guessing game and starts being a map.
Authors and institutions: Madeleine A. Gaidimas, Gyu-Hee Kim, Zi-Ming Ye, Julian S. Magdalenski, Nathaniel M. Barker, Abhijoy Mandal, Kourosh Darvish, Kent O. Kirlikovali, Varinia Bernales, Alán Aspuru-Guzik, Christos D. Malliakas, and Omar K. Farha.
Northwestern University (USA) and University of Toronto (Canada).
Published in
Chemical Science (Royal Society of Chemistry), 2026.
Read the full paper: https://doi.org/10.1039/D5SC09992G
All rights to the referenced publication belong to the original authors and the publisher. This post is a short, independent summary shared to highlight research where our customers used Chemspeed automation. No text, figures, or data from the original article are reproduced here. Product and company names are the property of their respective owners. For full methods, results, and licensing terms, please refer to the original publication via the link above.
