Magnesia cements have attracted increasing attention as low-carbon and functional binders because of their diverse reaction pathways, rapid hardening characteristics, and potential applications in rapid repair, fire-resistant boards, waste immobilization, carbon sequestration, and specialized engineering environments. Typical magnesia cement systems include magnesium silicate hydrate cement (M-S-H), magnesium phosphate cement (MPC), magnesium oxysulfate cement (MOS), magnesium oxychloride cement (MOC), and reactive magnesia cement (RMC). These systems involve complex reactions among MgO, soluble salts, siliceous or aluminosilicate components, water, and CO₂, leading to the formation of various hydration and carbonation products, such as M-S-H/M-(A)-S-H gels, 3Mg(OH)₂·MgCl₂·8H₂O, 5Mg(OH)₂·MgCl₂·8H₂O, 3Mg(OH)₂·MgSO₄·8H₂O, 5Mg(OH)₂·MgSO₄·7H₂O, and KMgPO₄·6H₂O. Because these products are often highly sensitive to mix proportion, temperature, pore solution chemistry, relative humidity, CO₂ concentration, and aggressive ions, it remains difficult to fully elucidate their phase assemblage evolution and long-term durability using experimental characterization alone. Thermodynamic modeling provides an effective approach to predict stable phase assemblages, pore solution composition, saturation indices, and phase stability boundaries under defined boundary conditions and has thus become an important tool for understanding the hydration, modification, degradation, and carbonation mechanisms of magnesia cements.

This review summarizes recent progress in thermodynamic modeling of phase assemblage evolution in magnesia cements. First, the basic principles of thermodynamic modeling are introduced, including the role of ion activity, saturation index, chemical equilibrium, Gibbs free energy minimization, and mass action law. Particular attention is paid to the selection of activity coefficient models, because magnesia cement systems, especially MOC, MOS, and salt-exposed systems, commonly contain highly concentrated electrolyte solutions. Second, this review collects and summarizes the thermodynamic databases currently available for magnesia cements and further discusses their main limitations. The reviewed databases cover M-S-H, MPC, MOS, MOC, and carbonation-related systems, including poorly crystalline M-S-H/M-(A)-S-H gels, magnesium phosphate hydrates, basic magnesium sulfate phases, magnesium oxychloride hydrate phases, and magnesium carbonate hydrates. These data provide the basis for thermodynamic modeling of hydration, modification, salt attack, leaching, and carbonation processes. Nevertheless, important gaps remain in the existing databases. For M-S-H-based systems, thermodynamic data are still insufficient to describe the continuous variation of Mg/Si and Al/Si ratios in M-S-H and M-(A)-S-H gels. For MPC, the thermodynamic data for intermediate phosphate hydrates and retarder-related complexes remain incomplete. For MOS and MOC, more reliable thermodynamic parameters are needed for magnesium sulfate/chloride phases, especially under variable temperature and high-salinity conditions. In addition, the incorporation of silica fume, fly ash, slag, metakaolin, and other industrial by-products introduces Si, Al, Ca, and Fe into magnesium-based systems, potentially leading to M-S-H, M-(A)-S-H, C-S-H, hydrotalcite-like phases, calcium phosphate phases, and other secondary products; however, thermodynamic data for these multicomponent reaction products are still insufficient. For carbonation thermodynamic modeling, further improvement is needed in the thermodynamic data of metastable magnesium carbonate hydrate phases, as well as in the ion interaction parameters for multicomponent Mg-Cl-SO₄-Na-Ca-K-CO₂-H₂O systems. Therefore, future database development should move from the compilation of individual phase data toward internally consistent multicomponent databases that can better represent the complex reaction environments of magnesia cements.

Third, the applications of thermodynamic modeling in different magnesia cement systems are reviewed. For M-S-H, modeling studies show that the Mg/Si ratio strongly controls the equilibrium among M-S-H gel, Mg(OH)₂, and SiO₂. Low Mg/Si systems tend to form silica-rich M-S-H with better chemical stability, whereas high Mg/Si systems are more likely to generate Mg(OH)₂, which may reduce resistance to salt attack and carbonation. The introduction of metakaolin, sodium carbonate, or other aluminosilicate components can further induce M-(A)-S-H gel, hydrotalcite-like phases, Al(OH)₃, and zeolite-type phases. For MPC, thermodynamic modeling reveals that the Mg/PO₄ ratio, pH, phosphate type, and temperature govern the transition from acidic magnesium phosphate phases to intermediate phases and ultimately to more stable struvite-type products. Modeling has also been applied to assess the effects of wollastonite, carbonation, chloride attack, sulfate attack, and seawater exposure on MPC phase assemblages. For MOS, modeling studies indicate that n_MgO/n_MgSO4 and n_H2O/n_MgSO4 ratios strongly affect the formation of 5-1-7, which is closely associated with compressive strength and water resistance. Chemical modifiers, especially citric acid and tartrate species, can alter Mg²⁺ activity and promote the formation of 5-1-7. For MOC, thermodynamic modeling has been widely used to clarify the stability of P3, P5, and Mg(OH)₂ as functions of n_MgO/n_MgCl2, n_H2O/n_MgCl2, temperature, modifiers, carbonation, chloride attack, and sulfate attack. These studies explain why P5 can provide high early strength but tends to decompose into Mg(OH)₂ under leaching or dilution conditions. Finally, carbonation modeling of RMC and other magnesia cements is discussed. The phase composition of magnesium carbonate hydrates is highly sensitive to CO₂ concentration, humidity, temperature, and reaction time. Because full-equilibrium calculations tend to overpredict thermodynamically stable phases such as MgCO₃ or Mg₅(CO₃)₄(OH)₂·4H₂O, metastable equilibrium modeling is often performed by suppressing selected stable phases to better reproduce experimentally observed intermediate products. This approach has been applied to RMC, M-S-H, MPC, and MOC carbonation, revealing the importance of kinetic limitations and environmental boundary conditions.

Overall, thermodynamic modeling has provided important insights into the hydration, modification, salt attack, leaching, carbonation, and durability of magnesia cements. Nevertheless, several limitations remain, including incomplete thermodynamic data for key hydrates and carbonates, insufficient ion interaction parameters for high-salinity multicomponent systems, simplified descriptions of low-crystalline gels and metastable phases, and limited coupling with kinetic processes. Future studies should further improve thermodynamic databases, validate predictions using XRD, TG, NMR, FTIR, SEM/EDS, ICP-OES, and pore solution analysis, and integrate thermodynamic modeling with kinetic, transport, humidity-controlled, and performance models. Such advances will enhance the predictive capability of thermodynamic modeling and support the rational design, durability assessment, and low-carbon applications of magnesia cements.