AI-Driven Materials Development for Sustainable Rare Earth Element Recovery Using Electrodialysis

• Applicant: Professor Dr. Patrick Rinke
• Subject Area: Physical Chemistry of Molecules, Liquids and Interfaces, Biophysical Chemistry ; Physical Chemistry of Solids and Surfaces, Material Characterisation
• Term: since 2026
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 572984715

The global demand for rare earth elements (REEs) - vital for renewable energy, electronics, and defense - is rising rapidly, putting pressure on supply chains. Some REEs may face depletion by 2040 - 2050. Conventional recovery methods like solvent extraction and ion exchange are costly, energy-intensive, and difficult to scale. Electrodialysis (ED) offers a selective and continuous alternative, but its adoption is limited by high capital and operational costs, mainly due to expensive membranes and energy use. The AI-REE project (AI-Driven Materials Development for Sustainable Rare Earth Element Recovery Using Electrodialysis) proposes a novel solution by integrating graphene-based membranes (GBMs) with artificial intelligence to improve REE recovery efficiency and affordability. It combines experimental membrane development with AI-guided optimization. The project focuses on creating ultrathin GBMs (0.2 - 1 µm) functionalized with rare-earth-ion-selective agents (REISAs), using a new photocatalytic interfacial polymerization (PCIP) method. This technique aims to cut membrane production costs by threefold and reduce fabrication time by 50%. To enhance membrane performance, Bayesian optimization (BO) is used to explore variables such as feed concentration, current density, membrane thickness, and REISA composition. The goal is to achieve over 99% selectivity and long-term stability, while minimizing experimental iterations through BO-based active learning. From a techno-economic standpoint, the project targets hybrid GBM-polymeric membranes that reduce capital costs by 1.5× and operational costs by 5× compared to conventional ED and solvent extraction. These improvements stem from using lower-cost materials and enabling more energy-efficient ion transport. The 36-month project is divided into three phases. Phase 1 (Months 1–12) establishes baseline performance with single-REISA GBMs achieving over 90% selectivity. Phase 2 (Months 13–24) optimizes mixed-REISA formulations to exceed 95% selectivity using multi-objective BO. Phase 3 (Months 25–36) scales the technology to lab-scale systems and extracts structure-property insights for industrial application. By combining advanced materials science with AI, AI-REE addresses key challenges in REE recovery and offers a sustainable path to securing critical materials for green technologies. Led by Prof. Ang (membrane design, ED) and Prof. Rinke (AI-driven materials science), the project builds on prior work in hybrid ED systems and machine learning optimization. Expected outcomes include open-access AI tools, high-performance membranes, and a scalable, cost-effective framework for REE recycling.

• DFG Programme: Research Grants
• International Connection: Singapore
• Partner Organisation: Agency for Science, Technology and Research (A*STAR)
• Cooperation Partner: Professor Dr. Edison Huixiang Ang, Ph.D.