Unlocking the Potential of Machine Learning in Materials Chemistry

New global machine learning force fields (MLFFs) will be trained for several hundred atoms, including transition metals, without resorting to uncontrolled approximations.

The developed MLFFs and statistical analysis kernels will be used to address problems in computational heterogeneous catalysis.

Catalyst design principles will be established and lead to discovery of highly efficient catalysts and solving pressing issues for a sustainable future.

The new MLFFs will be also extended to investigation of the vast materials space and chemistry of metal-organic chemical vapour deposition and molecular beam epitaxy growth of crystalline 2D layers required in the design of complex 2D semiconductor electronics.