Bright light particles spinning around each other with chemical symbol surrounding it

Computational Materials Discovery

Developing new techniques for the simulation of next-generation materials
Simulating new materials from a single atom to fully functioning devices using quantum computers, multiscale simulation, artificial intelligence and machine learning.

Computational Materials Discovery

Accurate computer simulations underpin mature technologies: airplanes, bridges, and smartphones are designed using precise computational models of the real world. By contrast, much materials discovery is driven by trial and error.

We envision a world where it is possible to accurately simulate any material, from single atoms to functioning devices. That ability would revolutionise materials discovery both by better explaining properties of existing materials and by proposing new materials for particular applications, from catalysts and photovoltaics, to batteries and superconductors. 

The Grand Challenge in Computational Materials Discovery contains three themes, each addressing a major challenge in computational materials science:

Theme 1: Quantum computing to model tricky quantum effects

Matter is fundamentally quantum mechanical, and accurately capturing quantum effects—which play functional roles in many materials—can be exponentially difficult on ordinary computers. Materials science will be the killer app for quantum computation because of the ease of simulating quantum effects on quantum computers. We are putting this idea into practice on existing, small-scale quantum computers here at the University of Sydney.

Theme 2: Multiscale modelling to span disparate length scales

There are too many atoms in macroscopic objects for each to be simulated in full detail. We are developing new approaches for connecting simulations at different levels of complexity—from the subatomic to the macroscopic—to show how the function of materials, such as photovoltaics or heterogeneous catalysts, emerges from interactions across vastly different length scales.

Theme 3: Machine learning to sift through vast chemical space

Even if we could model any material, the space of all possible materials for a particular application is enormous, making it impossible to simulate them all and choose the best. We are using the latest in artificial intelligence to make sense of the wealth of data accumulated through our simulations, and training machine-learning models to identify promising new candidates from the vast space of possibilities.

Ivan Kassal

Associate Professor
  • School of Chemistry F11

Lamiae Azizi

Senior Lecturer
  • Carslaw F07