Developing a Machine Learning Model to Identify Novel Compounds in the Fight Against Global Warming

Of all the greenhouse gases, carbon dioxide is the largest contributor to global warming. If we don’t take action by the year 2100, according to the Intergovernmental Panel on Climate Change, the average temperature of our world will rise by about 34 degrees Fahrenheit. Find effective ways to capture and store carbon monoxide₂ It’s been a challenge for researchers and industries focused on combating global warming, and Amir Parati Varimani is working to change that.

“Machine learning models hold promise for discovering new compounds or chemicals to combat global warming,” explains Barati Varimani, assistant professor of mechanical engineering at Carnegie Mellon University. “Machine learning models can achieve accurate and efficient virtual carbon dioxide screening₂ Candidates for storage and may generate preferred compounds that did not exist before.”

Barati Farimani has made a breakthrough by using machine learning to identify ionic liquid molecules. Ionic liquids (ILs) are families of molten salt that remain in a liquid state at room temperature, have high chemical stability and a high carbon dioxide content₂ solubility, making them ideal candidates for CO₂ storage. The mixture of ions largely determines the properties of ILs. However, such aggregation possibilities of cations and anions make it very difficult to exhaust the design space of ILs for effective CO.₂ Storage through traditional experiences.

Machine learning is often used in drug discovery to create so-called molecular fingerprints along with graph neural networks (GNNs) that treat molecules as graphs and use a matrix to identify molecular bonds and related properties. For the first time, Barati Farimani has developed both fingerprint-based ML models and GNNs capable of predicting CO₂ absorption in ionic liquids.

“Our GNN method achieves outstanding CO prediction accuracy₂ “Solubility in ionic liquids,” says Barati Varimani. “Unlike previous ML methods that relied on handcrafted features, GNN directly learns features from molecular diagrams.”

Understanding how machine learning models make decisions is just as important as the molecular properties that define them. This interpretation provides researchers with insight into how molecule structure affects the properties of ionic liquids from a data-driven perspective. For example, Barati Farmimani’s team has found that molecular fragments that physically interact with carbon dioxide₂ Less important than those that have a chemical reaction. In addition, those with less hydrogen bound to nitrogen could be more apt at formalizing a stable chemical reaction with carbon dioxide₂.

These results have been published in ACS Sustainable Chemistry and Engineeringwill enable researchers to advise on the design of novel and efficient carbon dioxide ionic liquids₂ storage in the future.