DFT-assisted machine learning for polyester membrane design in textile wastewater recovery applications

Area of Science:

• Materials Science
• Chemical Engineering
• Environmental Science

Background:

• Resource recovery from textile wastewater is crucial for sustainability.
• Polyester membranes show promise for dye recovery but face challenges in performance optimization.
• Complex relationships exist between polyester membrane properties and their recovery efficiency.

Purpose of the Study:

• To develop a machine learning model integrating DFT descriptors and fabrication parameters for polyester membrane design.
• To facilitate the generative design of high-performance polyester membranes for textile wastewater treatment.
• To identify optimal membrane compositions and operating conditions for enhanced resource recovery.

Main Methods:

• Utilized density functional theory (DFT) to generate descriptors for polyester materials.
• Developed machine learning models to predict membrane permeance and separation performance.
• Employed Bayesian optimization to explore a large chemical space for optimal membrane design.

Main Results:

• The machine learning model accurately predicted polyester membrane performance.
• Fabrication parameters were key for permeance, while DFT descriptors influenced dye/salt rejection.
• Identified optimal membrane designs that surpassed existing performance benchmarks.

Conclusions:

• A DFT-assisted machine learning approach enables efficient inverse design of polyester membranes.
• This method significantly improves dye and salt recovery from textile wastewater.
• The developed model offers a universal platform for advancing polyester membrane technology.