Skip to main content

Simulation of Artificial Photosynthesis

Machine learning methods to accelerate the simulation of Artificial Photosynthesis

 

Climate change and global warming is the most fundamental problem that mankind is facing nowadays. Regarding it, the reduction of the concentration of carbon dioxide (CO2) in the atmosphere is a key task in the fight against the global warming. Present work models the reaction of the photocatalytic reduction of CO2. The work should provide a better understanding of how CO2 can be catalysed using plasmonic nanoparticles and should provide a new method to study artificial photosynthesis and similar reactions useful in heterogeneous catalysis.

Photosynthesis is one of the most important reactions in nature that takes place in plants and converts the greenhouse gas CO2 into sugar molecules using sunlight energy. An artificial counterpart to this reaction could provide a solution to the increasing CO2 concentrations in the atmosphere and could provide a source for sustainable fuel. Plasmonic photocatalysis has recently emerged as a potential artificial counterpart to photosynthesis, but underlying mechanisms are little understood, partly because theoretical calculations are computationally infeasible. In this work, we aim to develop new methods based on artificial intelligence to investigate the plasmonic photocatalysis of CO2.

 

 

Kelvin-2 has many GPUs and allows for parallel computation on multiple GPUs. The project is conducted by Dr. Reinhard Maurer and Dr. Julia Westermayr, but several other members of the research group are also active on Kelvin-2. The computational resources allow us to test and validate different machine learning models that we develop. We were able to train newly developed machine learning methods that required a lot of memory and parallelization over multiple GPUs.

 

For further information:

Reinhard J. Maurer, Associate Professor, University of Warwick r.maurer@warwick.ac.uk

Twitter: @reinimau

Project funded by Austrian Science Fund (FWF, J-4522N), and UKRI Future Leaders Fellowship programme [MR/S016023/1]

 

Linked Publications :

 

  • Julia Westermayr, Shayantan Chaudhuri, Andreas Jeindl, Oliver T. Hofmann and Reinhard J. Maurer, "Long-range dispersion-inclusive machine learning potentials for structure search and optimization of hybrid organic–inorganic interfaces", Digital Discovery 2022; DOI 10.1039/D2DD00016D
  • Julia Westermayr and Reinhard J. Maurer, "Physically inspired deep learning of molecular excitations and photoemission spectra", Chemical Science 2021, 12, 10755; DOI 10.1039/D1SC01542G