Machine Learning-Assisted Rapid Development of High Performance Flexible Lead-Free Radiation Shielding Gels
The escalating use of ionizing radiation in medical and industrial applications necessitates lead-free, flexible, and sustainable shielding materials. Current development relies on empirical trial-and-error, which is inefficient. This study introduces a machine learning-assisted Monte Carlo simulation strategy for rapid optimization of metal filler compositions for X-ray attenuation across 40–120 kV. Guided by this AI-driven approach, polyvinyl alcohol (PVA)-based gels containing uniformly dispersed Bi/W/Gd2O3 nanoparticles were developed, forming within 1 minute at -20°C using a PVA-DMSO/H2O co-solvent system. The optimized gel with 50 wt% metal loading exhibits exceptional mechanical properties: tensile strength of 1.76 MPa, toughness of 6.3 MJ m−3, and elongation of 600%. It achieves >98% X-ray shielding efficiency at 5 mm thickness, outperforming lead composites at 120 kV. The physically cross-linked network provides recyclability and anti-freezing capability, retaining flexibility at -50°C. This work establishes a data-driven paradigm for designing high-performance radiation-shielding materials, demonstrating AI's potential to accelerate materials discovery and enable scalable fabrication of eco-friendly protective systems.