AI machine learning predicts heat capacities of MOFs
Machine learning artificial intelligence has been shown to predict the heat capacities of metal-organic frameworks
Metal-organic frameworks (MOFs) are a class of materials that contain nano-sized pores. These pores give MOFs record-breaking internal surface areas, which make them extremely versatile for a number of applications.
These include but are not limited to: separating petrochemicals and gases; mimicking DNA; producing hydrogen; and removing heavy metals, fluoride anions and even gold from water.
MOFs are the focus of Professor Berend Smit’s research in the School of Basic Sciences at the Swiss Federal Institute of Technology, Lausanne (EPFL), where his group employs machine learning to make breakthroughs in the discovery, design, and even categorisation of the ever-increasing MOFs that currently flood chemical databases.
See also: Intelligent microscope developed for detecting rare biological events
In a new study, Smit and his colleagues have developed a machine-learning model that predicts the heat capacity of MOFs.
He said: “This is about very classical thermodynamics; how much energy is needed to heat up a material by one degree?
“Until now, all engineering calculations have assumed that all MOFs have the same heat capacity, for the simple reason that there is hardly any data available.”
Seyed Mohamad Moosavi, a postdoc in Smit’s group, added: “If there is no data, how can one make a machine-learning model? That looks impossible!”
The answer is the most innovative aspect of the work: a machine-learning model that predicts how the local chemical environment changes the vibrations of each atom in an MOF molecule.
Smit explained: “These vibrations can be related to the heat capacity.
“Before, a very expensive quantum calculation would give us a single heat capacity for a single material, but now we get up to 200 data points on these vibrations.
“So, by doing 200 expensive calculations, we had 40,000 data points to train the model on how these vibrations depend on their chemical environment.”
The researchers then tested their model against experimental data as a real-life check.
Smit stated: “The results were surprisingly poor until we realised that those experiments had been done with MOFs that had solvent in their pores.
“So, we re-synthesised some MOFs and carefully removed the synthesis solvent – measured their heat capacity – and the results were in very good agreement with our model’s predictions!”
Moosavi advised: “Our research showcases how Artificial Intelligence (AI) can accelerate solving multi-scale problems.
“AI empowers us to think about our problems in a new way and even sometimes tackle them.”
To demonstrate the real-world impact of the work, engineers at Heriot-Watt University simulated the MOFs performance in a carbon capture plant.
Smit said: “We used quantum molecular simulations, machine learning, and chemical engineering in process simulations.
“The results showed that with correct heat capacity values of MOFs the overall energy cost of the carbon capture process can be much lower than we originally assumed.
“Our work is a true multi-scale effort, with a huge impact on the techno-economic viability of currently considered solutions to tackle climate change.”
The research is published in Nature Materials.
Image: Metal organic frameworks capturing CO2 from flue gasses. © S.M. Moosavi/ EPFL.