Design of CO₂-philic molecular units with large language models

Area of Science:

• Chemical Sciences
• Materials Science
• Computational Chemistry

Background:

• Chemical sciences increasingly leverage artificial intelligence for molecular design.
• Developing efficient carbon capture technologies is crucial for climate change mitigation.
• Physisorption-based methods offer a promising avenue for CO2 capture.

Purpose of the Study:

• To explore the use of large language models (LLMs) for designing novel molecular structures.
• To enhance CO2 affinity in molecules for physisorption-based carbon capture.
• To investigate the synergy between AI-driven design and expert chemical knowledge.

Main Methods:

• Utilizing large language models (LLMs) to generate candidate molecular structures.
• Employing Density Functional Theory (DFT) for evaluating the CO2 affinity of generated molecules.
• Comparing AI-generated designs with established chemical principles and experimental data.

Main Results:

• LLMs successfully generated novel molecular structures with high CO2 affinity.
• DFT evaluation confirmed the efficacy of LLM-proposed physisorption agents.
• Identified emergent design strategies, including cooperative binding motifs, within LLM outputs.

Conclusions:

• LLMs are powerful tools for accelerating the design of advanced materials in chemical sciences.
• The integration of LLMs with computational chemistry enhances the discovery of effective carbon capture agents.
• AI-guided molecular design, combined with expert oversight, paves the way for innovative climate solutions.