CatTSunami: Accelerating Transition State Energy Calculations with Pretrained Graph Neural Networks
Brook Wander, Muhammed Shuaibi, John R. Kitchin, Zachary W. Ulissi, C. Lawrence Zitnick
Abstract
Direct access to transition state energies at low computational cost unlocks the possibility of accelerating catalyst discovery. We show that the top performing graph neural network potential trained on the OC20 data set, a related but different task, is able to find transition states energetically similar (within 0.1 eV) to density functional theory (DFT) 91% of the time with a 28× speedup. This speaks to the generalizability of the models, having never been explicitly trained on reactions, the machine learned potential approximates the potential energy surface well enough to be performant for this auxiliary task. We introduce the Open Catalyst 2020 Nudged Elastic Band (OC20NEB) data set, which is made of 932 DFT nudged elastic band calculations, to benchmark machine learned model performance on transition state energies. To demonstrate the efficacy of this approach, we for the first time explicitly treated a large reaction network with 61 intermediates and 174 dissociation reactions at DFT resolution (40 meV). To find low energy transition states we densely enumerate many possible NEBs. Using DFT this would have taken 52 GPU years. With ML we realized a 1500× speedup for dense enumerations, using just 12 GPU days of compute. Similar searches for complete reaction networks could become routine using the approach presented here. Finally, we constructed an ammonia synthesis activity volcano and systematically found lower energy configurations of the transition states and intermediates on six stepped unary surfaces than had previously been reported. This scalable approach offers a more complete treatment of configurational space to improve and accelerate catalyst discovery.