Ring Repeating Unit: An Upgraded Structure Representation of Linear Condensation Polymers for Property Prediction.

Area of Science:

• Polymer Science
• Materials Informatics
• Computational Chemistry

Background:

• Accurate polymer structure representation is vital for predictive modeling and data-driven design.
• Conventional monomer and repeating unit (RU) methods struggle with unique representation of heterochain polymers due to complex monomer combinations.
• Existing methods may lead to non-unique predictions, hindering reliable quantitative structure-property relationship (QSPR) modeling.

Purpose of the Study:

• To introduce a novel Ring Repeating Unit (RRU) method for unique polymer structure representation.
• To develop and validate an RRU-based QSPR model for predicting the glass transition temperature (Tg) of polyimides (PIs).
• To demonstrate the superiority of the RRU method over traditional RU-based approaches for complex polymer systems.

Main Methods:

• Development of the Ring Repeating Unit (RRU) method for unique polymer structure encoding.
• Creation of an RRU-based Quantitative Structure-Property Relationship (QSPR) model.
• Prediction of glass transition temperature (Tg) for a large dataset of polyimides (PIs).
• Rigorous model validation using internal, external, and Y-randomization techniques.

Main Results:

• The RRU method provides a unique and effective representation for diverse polymer structures, including heterochain polymers.
• The RRU-based QSPR model accurately predicted the glass transition temperature (Tg) of polyimides with deterministic values.
• Comparison with an RU-based model highlighted the non-unique prediction issues of conventional methods, underscoring the RRU's necessity.
• Comprehensive validation confirmed the robustness and reliability of the RRU-based QSPR model.

Conclusions:

• The proposed Ring Repeating Unit (RRU) method offers a significant advancement in polymer structure representation for data-driven approaches.
• RRU-based modeling enables accurate and reliable prediction of polymer properties, such as glass transition temperature (Tg).
• This novel representation is crucial for developing sophisticated QSPR models and advancing polymer design through data-centric strategies.