Ai-driven polymer design platform for targeted gas permeability and selectivity

Overview

A cloud-based polymer informatics platform, PolymRize, leverages artificial intelligence and machine learning to predict and design polymers with specific gas permeability and selectivity properties. By combining a large available database of polymer permeability values with proprietary fingerprinting schemas and deep neural networks, the platform can recommend new-to-the-world polymers that meet defined property targets. This approach accelerates materials development, reducing the time and cost associated with traditional experimental trial-and-error methods.

The technology has been validated through the design of polymer dielectric materials for capacitive energy storage that outperformed industry benchmarks by a factor of two at elevated temperatures. The same methodology is being applied to develop biodegradable polymers for food packaging and other sustainable applications.

Technical specifications

Key capabilities:

• Predictive modeling: AI algorithms trained on extensive polymer-property datasets can instantly predict permeability and selectivity for new polymer candidates
• Generative design: State-of-the-art machine learning and generative models recommend polymers meeting specific property goals for experimental validation
• Comprehensive database: Houses a large collection of permeability values for multiple gas molecules across diverse polymers, including biodegradable chemistries
• Custom model building: Allows R&D teams to process proprietary data alongside pre-trained AI models
• Virtual synthesis and property prediction: Enables researchers to simulate and evaluate polymer performance before committing to laboratory synthesis
• Conversational AI assistant: AskPOLY enables natural-language interaction with the platform to streamline researcher workflows

Applications include:

• Biodegradable food packaging polymers with precise oxygen permeability
• Polymer dielectrics for capacitive energy storage in electric vehicles
• Sustainable bio-based polymers such as PHAs and PLAs
• General-purpose polymer design and formulation optimization across electronics, energy, and packaging sectors

Technology readiness level

The underlying AI-driven polymer design methodology has been repeatedly validated by the Ramprasad Research Group at Georgia Tech. Recently designed polymer dielectric materials for capacitive energy storage have demonstrated performance exceeding the industry state-of-the-art by a factor of two at twice the operating temperature, and these materials are currently undergoing scale-up.

Future validation efforts for gas permeability applications include expanding the polymer dataset, improving predictive models for oxygen permeability, enhancing generative design capabilities for biodegradable polymers, and conducting iterative design-test-refine cycles with industrial partners. The platform is commercially available and has demonstrated practical utility through collaborations with partners including CJ Biomaterials for sustainable polymer assessment.