Machine-Learning-Based Predictions of Polymer and Postconsumer Recycled Polymer Properties: A Comprehensive Review

Area of Science:

• Materials Science and Engineering
• Polymer Science
• Computational Chemistry

Background:

• Increasing demand for virgin and postconsumer recycled (PCR) polymers necessitates advanced design and characterization methods.
• Challenges in polymer informatics arise from the complex hierarchical structures of polymers.
• Sensor array technologies offer insights into polymer properties and compatibilization for recycling.

Purpose of the Study:

• To review the application of machine learning (ML) in polymer science, particularly for virgin and PCR polymers.
• To detail the fundamental steps and methodologies for implementing ML in polymer design and property prediction.
• To identify research gaps and challenges in using artificial intelligence for efficient PCR polymer discovery.

Main Methods:

• Review of existing literature on ML algorithms applied to polymer structures and properties.
• Discussion of key ML steps: fingerprinting, algorithm selection, database utilization, and representation strategies.
• Analysis of ML-based prediction of polymer material properties and exploration of sensor array technologies for characterization.

Main Results:

• ML algorithms are emerging as cost-effective, scalable solutions for understanding polymer physical and chemical structures.
• State-of-the-art reviews demonstrate the successful prediction of various polymer material properties using ML.
• The review provides a comprehensive overview of ML applications from basic principles to advanced property prediction.

Conclusions:

• ML holds significant potential for accelerating the discovery and development of novel polymers, including PCR materials.
• Further research is needed to address open questions and overcome challenges in applying AI to PCR polymer property prediction.
• Targeted ML approaches can lead to more efficient and customized polymer solutions for diverse applications.