Performance Optimization of Radiation-Shielding Serpentine Concrete with Slurry-Coated Aggregates Using the Response Surface Methodology

Abstract

With the rapid development of nuclear energy and related infrastructure, there is an urgent demand for concrete that combines mechanical strength with radiation-shielding capability. Although serpentine aggregate is advantageous for neutron attenuation due to its high crystalline water content, its inherent porosity, low strength, and weak interfacial bonding constrain its practical application. In this study, a cement-based composite paste was used to modify serpentine aggregates. Five composite pastes incorporating fly ash (FA), ground granulated blast-furnace slag (GGBFS), silica fume (SF), calcium bentonite (Ca-Bent), and calcined white kaolin (CWK) were evaluated. Preliminary screening identified the GGBFS-modified slurry as the most effective; it exhibited a 25.2% increase in compressive strength over the control group while maintaining a high crystalline water content of 11.12%, indicating excellent neutron shielding potential. A Box–Behnken design was employed to optimize three factors—concrete water–binder ratio (A), coating water–binder ratio (B), and GGBFS dosage in coating (C)—for their effects on mechanical and shielding performance. The response surface models for crystalline water content and compressive strength were both significant. Factor B had the greatest impact on both responses, while the quadratic term of Factor C showed a notable nonlinear effect on strength. Analysis of interaction terms showed the AB interaction primarily governed water retention, while the BC interaction was key to strength. Regarding γ -ray shielding, the half-value layer (HVL) was negatively correlated with density, indicating that attenuation is primarily governed by material compactness. Finally, scanning electron microscopy (SEM) analysis confirmed the effect of the coating water–binder ratio on the calcium silicate hydrate (C─S─H) gel microstructure, aligning with the model’s predictions.