M3GNet Accelerates Discovery Of Quaternary Oxides
Researchers systematically assess a universal machine-learning interatomic potential (M3GNet) to accelerate crystal structure prediction in quaternary oxides, exploring Sr-Li-Al-O and Ba-Y-Al-O. They rediscover known phases absent from training and predict seven new thermodynamically and dynamically stable compounds, including a new Sr2LiAlO4 polymorph. The study finds PBE stability predictions need cross-validation with SCAN and RPA, and CSP search algorithms remain bottlenecks.
Scoring Rationale
Finds new, stable quaternary oxides via M3GNet-enabled CSP, but preprint evidence and PBE reliance limit immediate confidence.
Practice interview problems based on real data
1,500+ SQL & Python problems across 15 industry datasets — the exact type of data you work with.
Try 250 free problems
