The sulfate source in calcium sulfoaluminate (CSA) cement is an important factor that determines its hydration characteristics, and the properties of CSA cement can be controlled via adjusting the molar ratio of calcium sulfate to ye'elimite (generally referred to as the "M-value"). Sulphosilicate cement is a novel cementitious material developed via incorporating ternesite (C₅S₂$) and free-calcium sulfate (f-C$) into the mineral composition of CSA cement. The calculation of the M-value must comprehensively account for their contents in the clinker due to the presence of SO₃ in C₅S₂$ and f-C$. This study optimized the conventional calculation formula for the M-value and systematically investigated its influence on the compressive strength and expansion behavior of sulphosilicate cement. In addition, the mechanism underlying the effect of anhydrite dosage on cement hydration was also elucidated through the evolution of hydration products.

In this experiment, sulphosilicate clinker was synthesized with limestone, fly ash, phosphogypsum, and bauxite as raw materials. The raw materials were calcined at 1250 ℃ for 1 h, followed by rapid air quenching to obtain the clinker. Five groups of sulphosilicate cement with M-values of 0, 0.9, 1.8, 2.7, and 3.6 were designed. Mortar specimens were prepared to evaluate compressive strength and expansion rate, while paste specimens were prepared for analyzing hydration product characteristics at specific ages. The hydration heat release was measured by an isothermal calorimeter. The phase compositions and microstructure of the hydration products were analyzed by X-ray diffractometer, thermal behavior analyzer and field-emission environmental scanning electron microscope. The pore structure was performed by a mercury intrusion porosimeter.

Among the various sulphoaluminate cement groups, the M_1.8 specimen demonstrates the maximum compressive strength of 43.9 MPa after 3-d hydration. As hydration progresses, the compressive strength of the M_2.7 and M_3.6 groups shows a significant growth, with the M_3.6 group reaching 56.7 MPa at 14 d. However, the M_3.6 specimen exhibits a notable strength retrogression, which, is decreased by 4.3 MPa, after 28-d hydration. In contrast, the compressive strength of the M_0.9 group increases markedly from 14 d to 28 d, with an improvement of 12.3 MPa.

For specimens with the M-value below 1, the early hydration products at 3 d are dominated by AFm phases with insufficient AFt content, resulting in an initial shrinkage. As hydration progresses, the dissolution of SO₃ from anhydrite and C₅S₂$ promotes AFt formation, leading to a measurable expansion after 7 d. At low M-values, AFt contributes to pore refinement, while the reaction of C₅S₂$ and C₂S in later stages generates C-(A)-S-H gel and stratlingite (C₂ASH₈), refining the microstructure of hydration products and enhancing long-term strength. In contrast, specimens with higher M-values (i.e., 1.8, 2.7, and 3.6) exhibit expansion as early as 3 d. Increased anhydrite content and prolonged hydration amplify the crystallographic stress exerted by AFt crystals on the surrounding matrix, causing coarsening of pores, and elevated total porosity, and resulting in strength retrogression during 14-28 d of curing.

This study was to propose a formula for calculating the gypsum coefficient M-value in sulphosilicate cement, i.e., C_G= 0.13 × M × (Y - A×2.24 - T×1.27)/S_g, establishing the relationship between gypsum dosage and cement performance, while elucidating the influence of M-values on the mechanical properties and hydration mechanisms. The experimental results demonstrated that within the optimal M-range of 0.9-1.8, cement could achieve continuous strength development for all curing ages. Higher M-values (2.7-3.6) enhanced mid-term strength but induced strength regression at 28 d. Increasing M-values effectively suppressed early-age shrinkage and induced progressive expansion, and expansion rates were positively correlated to M-values at all hydration stages. Hydration products primarily consisted of AFt, AH₃, and C-(A)-S-H gel. At M₀ and M_0.9, AFm phases formed during early hydration, followed by C₂ASH₈ in later stages. Conversely, M-values of ≥1.8 accelerated C₄A₃$ hydration but inhibited both C₅S₂$ dissolution and C₂ASH₈ formation. Microstructurally, M_0.9 specimens exhibited pore refinement and reduced porosity through gel-phase densification, enhancing strength development, whereas M_3.6 samples suffered from coarse pore proliferation due to excessive anhydrite content promoting oversized AFt crystals, ultimately degrading 28-d strength. At the M-value of 0.9-1.8, the system achieved balanced high strength and minimal volumetric deformation, making it suitable for general engineering applications. Elevating the M-value to 1.8-2.7 induced a controlled expansion behavior, enabling targeted use in shrinkage compensation scenarios. However, at the M value of 3.6, delayed strength retrogression and microstructural deterioration could occur, thus establishing 3.6 as a critical upper limit for M-values.