MIT team used graph neural network to accelerate the screening of amorphous polymer electrolytes and promote the development of next-generation lithium battery technology

Polymer electrolytes are promising candidates for the next generation of lithium-ion battery technology . Large scale screening of polymer electrolytes , Subject to molecular dynamics in amorphous systems （MD） The huge cost of simulation ： The amorphous structure of polymer requires repeated sampling to reduce noise , Slow relaxation requires a long simulation time to converge .

ad locum , MIT researchers , Use multi task graph neural network to speed up screening , The network from a lot of noise 、 Unconverged short MD Data and a small amount of convergence are long MD Learning from data . They achieved 4 Accurate prediction of different convergence characteristics , And screened 6247 Space for polymers , This space is several orders of magnitude larger than previous computational studies .

Besides , Researchers extracted several design principles for polymer electrolytes , And provide an open data set for the community . Studies have shown that , This method can be applied to complex simulation 、 A wide range of material discovery problems for amorphous materials .

The research 「 Accelerating amorphous polymer electrolyte screening by learning to reduce errors in molecular dynamics simulated properties」 entitled , On 2022 year 6 month 14 Published on 《 Nature Communications》.

Thesis link ： https://www.nature.com/articles/s41467-022-30994-1