Large-Scale Construction and Analysis of Amorphous Porous Polymer Network Materials

Area of Science:

• Materials Science
• Computational Materials Science
• Polymer Chemistry

Background:

• Data-driven methods are vital in materials science for discovering new materials.
• Amorphous materials, like porous polymer networks (PPNs), are underrepresented in materials databases due to structural complexity.
• Accurate modeling of disordered structures is essential for understanding material properties.

Purpose of the Study:

• To construct the largest database of amorphous porous polymer networks (PPNs) to date.
• To highlight the importance of considering structural disorder in PPNs for accurate property prediction.
• To develop machine learning models for predicting PPN properties from monomer structures.

Main Methods:

• Generated a database of 10,237 porous polymer networks (PPNs) using self-assembly simulations.
• Analyzed structural differences compared to existing databases, emphasizing amorphous characteristics.
• Trained machine learning models on the constructed PPN database.

Main Results:

• Created the largest database of amorphous PPNs, accounting for structural disorder.
• Demonstrated that structural disorder significantly impacts the chemical behaviors of PPNs.
• Machine learning models successfully predicted macroscopic PPN properties from monomer atomic structures.

Conclusions:

• Accurate characterization of amorphous PPNs requires explicit consideration of their disordered structures.
• Predicting properties of novel PPNs is feasible using monomer structures, bypassing extensive simulations.
• This work facilitates accelerated discovery and design of amorphous porous materials.