Research on Shrinkage in Lithium Slag Geopolymer Mortar: Effects of Mix Proportions and a Shrinkage Prediction Model

Lithium slag (LS), a solid waste generated during lithium smelting, exhibits significant potential for geopolymer preparation. However, the high shrinkage of lithium slag geopolymer mortar (LSGM) severely restricts its engineering application. Currently, research on the effects of mix proportions (GBFS-LS mass ratio, water-binder ratio, and binder-sand ratio) on LSGM's shrinkage, and the correlation between shrinkage behavior and microstructures (pore structure and thermal behavior), remains insufficient. Additionally, there is a lack of targeted shrinkage prediction models for LSGM. To address these research gaps, this study systematically investigates the shrinkage characteristics of LSGM and develops a modified prediction model. Thermogravimetric analysis-differential thermal gravimetric analysis (TG-DTG) results show that a lower GBFS-LS ratio promotes the formation of dense sodium-alumino-silicate hydrate (N-A-S-H) gels. Meanwhile, mercury intrusion porosimetry (MIP) tests demonstrate that optimizing the water-binder ratio and binder-sand ratio refines the pore structure of LSGM, where the average pore size is reduced from 53.5 nm at a GBFS-LS ratio of 3 to 28.75 nm at a GBFS-LS ratio of 1.5.Quantitatively; compared with the group with a GBFS-LS ratio of 3, the 90-day shrinkage strain of the group with a GBFS-LS ratio of 1.5 decreases by 25.8%. When the water-binder ratio decreases from 0.57 to 0.27, the 90-day shrinkage strain reduces by 36.7%; in contrast, increasing the binder-sand ratio from 0.21 to 0.39 leads to a 39.8% increase in 90-day shrinkage strain. Based on the experimental data and the fundamental framework of the American Concrete Institute (ACI) model, this study introduces mix proportion influence coefficients and constructs a novel shrinkage prediction model tailored to LSGM. The coefficient of determination (R²) of the proposed model exceeds 0.98. This model provides a reliable quantitative tool for the mix proportion optimization and engineering application of LSGM.