Step-change materials discoveries don't come from extrapolating past ones. Models get you to the right place to look. Then you look for the unexpected on purpose. This is "guided serendipity".
Just published with Prof. Mas Subramanian in Chem. Mater.:
doi.org/10.1021/acs.chemmate…
Matthew McDermott retweeted
Joining the A-Lab conversation: linkedin.com/feed/update/urn…
Matthew McDermott retweeted
👉 Major AI progress out of @BerkeleyLab with Google
To transform the future of superconductors, power systems, electric vehicles, and other clean energy tech — we need new materials.
A new algorithm to predict them + an AI-driven robotic lab to make them aims to do just that.
Researchers use our Materials Project to develop new materials for future tech - think better batteries or solar cells. With new #materials calculated by @GoogleDeepMind, they’ll have even more data to work with. @Nature #AI #FutureTech
The latest ⬇️
newscenter.lbl.gov/2023/11/2…
Matthew McDermott retweeted
In A-lab, we discover new materials autonomously and fast.
Excited to see our paper published on Nature today!
Really honored to be a member of A-lab team @cedergroup. Thanks to @NJSzymanski and @YanHelenZENG for leading the project!
nature.com/articles/s41586-0…
Matthew McDermott retweeted
Excited to share our work in @Nature on the A-Lab, an AI-driven robotic platform that automates materials synthesis. It discovered 41 new compounds in 17 days of closed-loop experiments! Thanks to @cedergroup and @YanHelenZENG for leading this effort!
nature.com/articles/s41586-0…
Matthew McDermott retweeted
Our method to automate the optimization of solid-state synthesis is out now in @NatureComms! It uses DFT data but also learns from experiments to find the best precursors for a given material. We hope this can help guide future autonomous platforms 🤖 😃
nature.com/articles/s41467-0…
Oh man someone really messed up the mastered audio on the new 1989 re-release. Buzzing and clicking sounds all over it. Very disappointing :(
5
486
Matthew McDermott retweeted
Very proud of this work - congratulations Matt ! A crowning achievement to your PhD 🎉 🎊
Our article on assessing the selectivity of solid-state reactions for improved synthesis planning is available now in @ACSCentSci! Feeling excited as we get one step closer to achieving the "holy grail" of predictive materials synthesis. @KPatBerkeley
pubs.acs.org/doi/10.1021/acs…
Our article on assessing the selectivity of solid-state reactions for improved synthesis planning is available now in @ACSCentSci! Feeling excited as we get one step closer to achieving the "holy grail" of predictive materials synthesis. @KPatBerkeley
pubs.acs.org/doi/10.1021/acs…
This work was a large collaboration involving much of the great team in @EfrcGenesis. Special thanks to the @csu_chemistry folks: Brennan McBride, Corlyn Regier, Thinh Tran, and @jrnchem!
Someone vandalized a speed limit sign in Berkeley (on Marin Ave) and I have to be honest I laughed for way too long when I saw it 😂
12
458
Predicting how to synthesize new materials is tricky. Getting a pure phase is even harder (we saw this w/ lk99). To address this challenge, we came up with a framework & tool for coming up with more optimal solid-state synthesis recipes! Preprint:
arxiv.org/abs/2308.11816
Stated geometrically: adding another element means adding another dimension in energy-composition phase space. That gives so many more ways you can find convex hull “binary slices” that yield the target while shortcutting around undesired competing phases
1
154
I’m really happy to have worked with the amazing people on this project. Thank you especially to Brennan McBride, @jrnchem, and @KPatBerkeley who all were somehow okay with me attempting to be an experimentalist after four years of only coding 😬😎
1
145
My catalytic converter was stolen a few weeks ago, and now my rental car replacement was broken into. What are the odds my new rental replacing my old rental also gets vandalized? 😛
1
4
430
Matthew McDermott retweeted
Excited to share our recent work in @Nature_NPJ, where we show that machine learning can be used to "drive" XRD measurements and identify phases in real time!
Thanks to @ChrisJBartel @YanHelenZENG @delmouha
nature.com/articles/s41524-0…
Never thought I’d see snow in Berkeley, CA @NWSBayArea (Vollmer Peak)








