AI-AI-OH: Machine Learning in Catalysis

AI – Artificial Intelligence – is on the cusp of transforming everything we do. Machine learning is one aspect of that transformation.

In Machine Learning In Catalysis, John R. Kitchin (Nature Catalysis, 2018), presents an excellent overview of machine learning in catalysis, especially as it pertains to such industrial processes as syngas processing.

In his words …

“Catalysis is a complex, multidimensional and multiscale field of research. Machine learning is helping to build better models, understand catalysis research and generate new knowledge about catalysis.”

He goes on to say that DFT has been the workhorse …

“Density functional theory (DFT) has been a workhorse of first-principles based simulations in catalysis. A key limitation of DFT is the computational cost of the calculations. In many scenarios it is desirable to run thousands or hundreds of thousands of calculations, for example, screening, free energy calculations, or for Monte Carlo or molecular dynamics simulations. DFT is simply not practical for this.”

As wonderful as it is, DFT is not able to handle massive data …

Machine learning to the rescue …

“An area of growing significance is the use of machine learning to develop atomistic potentials that are learned from DFT calculations.”
“Machine-learned potentials can be used for reactive dynamics on catalyst surfaces, which allows one to probe reaction trajectories at realistic temperatures. Shakouri and co-workers used a neural network potential to model nitrogen dissociation on Ru(0001) that included the surface coupling of surface phonon modes with adsorbate vibrational modes. The machine-learned neural network enabled them to use molecular dynamics with DFT accuracy to simulate the full dynamical dissociation of N2 on Ru(0001). This is important because the reaction probabilities are low, requiring long simulation times and a large number of trajectories. With the accelerated calculations they were able to run the required length of simulations, and obtain good agreement with experimental results.”

Even for those of you deeply immersed in AI research, Kitchen’s article is worth a read. For one thing, it can help to see things from a bird’s eye view … because who knows where your research will take you next?

The article can be particularly useful in helping younger colleagues broaden their perspective on the AI field.
///////

TIP: While you will have to pay for a downloadable copy of the full article, you can access the full text for online viewing only at: https://rdcu.be/LGrM

///////
Jean Steinhardt served as Librarian, Aramco Services Co., Engineering Division, for 13 years. He now heads Jean Steinhardt Consulting LLC, producing the same high quality research that he performed for Aramco.

Follow Jean’s blog at:http://desulf.blogspot.com/ for continuing tips on effective online research
Email Jean at research@jeansteinhardtconsulting.com with questions on research, training, or anything else
Visit Jean’s Web site at http://www.jeansteinhardtconsulting.com/ to see examples of the services we can provide