Ground granulated blast slag (GGBS) corresponding to Grade S105 was purchased from Shandong Kangjing New Materials Co., Ltd., China. Its density was 2800 kg/m³. Table 1 presents the chemical composition determined by X-ray fluorescence (XRF). MgO and KOH of analytical purity were purchased from Sinopharm Chemical Reagent Co., Ltd., China. Quartz sand with a size of less than 0.85 mm was used. Waterglass (27.3 wt% SiO₂ and 8.5 wt% Na₂O) was purchased from Yourui Refractory Material Co., Ltd., China. The water used was tap water.

Foundry sand was prepared in the laboratory by mixing quartz sand, waterglass, and KOH at a mass ratio of 100:50:2 in a mixer for 3 min (Wang et al., 2022). The mixture was then added to a ventilation mold and injected with CO₂ at a rate of 5 L/min for 5 min. The hardened mixture was, respectively, calcined at 100°C and 800°C for 1 h to simulate practical working conditions. Thereafter, the hardened mixtures were crushed into powder samples and were named SwFS₁ and SwFS₈ to indicate the subjected working temperatures.

The feasibility of SwFS as a source of fine aggregates to prepare MASM was examined with variable replacements (40%, 70%, and 100%) of quartz sand, as shown in Table 2. Before adding mixing water, solid materials (including GGBS, MgO, and aggregates) were well mixed in a Harbor mixer. Thereafter, water was carefully added to the mixer, and the mixture was further mixed for 3 min.

Table 3 presents the fluidity of mortars with and without SwFS. This shows that the incorporation of 40% or 70% SwFS₁ into MASM has resulted in slight increments in the fluidity of mortars, while 100% SwFS₁ reduced the fluidity. It is indicated that the fluidity of a mortar is reduced when more SwFS₁ is incorporated. This may be due to the soluble properties of the coated waterglass of SwFS₁, which could dissolve and initially work as a lubricant to improve fluidity. However, more waterglass would dissolve into the pore solution with the high content of SwFS₁, aiding in activating the slag, resulting in a faster early solidification of the mortar and the deterioration of fluidity. In contrast, these mortars with SwFS₈ are very similar and are lower than those of the control (Ctrl), indicating that the presence of SwFS₈ in MASM degrades the flow properties of MASM. This may result from high-temperature calcination causing the decomposition of the waterglass layer on the surface of SwFS₈, while reducing the moisture content of SwFS₈, resulting in the loss of mortar fluidity.

Figure 1 shows the compressive strengths of MASM mortars with and without SwFS. This clearly shows that the incorporation of different types of SwFS has resulted in different effects on the compressive strength of mortars. When SwFS₁ has been incorporated, the compressive strength of mortars significantly increases with its content. Despite that, the 3-day and 7-day compressive strengths of SwFS₁40 are slightly lower than that of Ctrl; the compressive strengths of mortars with SwFS₁ have been maximally increased by 74.1% at 3 days, 33.6% at 7 days, and 64.8% at 28 days. This could be due to the dissolved waterglass from SwFS₁ released into the surrounding paste, which is an effective activator for MASM by providing abundant [SiO₄]^4- (Wang et al., 2022). Moreover, the results also suggested that MgO may be better efficient for SwFS-based alkali-activated slag since MgO as an activator provides greater operational safety compared to SwFS-based NaOH-activated slag (Fang et al., 2023) and achieved higher compressive strength compared to SwFS-based Ca(OH)₂-activated slag (Wang et al., 2022).

In contrast, the compressive strength of SwFS₈-incorporated mortars displays a slight increase with the increased content of SwFS₈, while the 3-day and 7-day compressive strengths were reduced by 16.9–17.5% and 22.3–29.0%, respectively. This indicates that coated waterglass may not be dissolved and released into the surrounding paste in a similar way as SwFS₁ since SwFS₈ is less soluble (Wang et al., 2022). As a result, the structure of the interface of SwFS₈ could be converted into a sandwich structure of paste–waterglass coating–quartz sand, which would weaken the mechanical properties of mortars. In addition, the increased compressive strength of mortars with the content of SwFS₈ may suggest that some of the coated waterglass also dissolved into the surrounding paste in the lower contents.

Figure 2 shows the pore structure of MASM mortars. All mortars have similar distributions of pore sizes between 1–200 nm. For SwFS₁-added mortars, the peak heights are lower than those of the Ctrl. The height is increased when more SwFS₁ has been added, indicating the refinement in the pore structure of MASM mortars. This can be attributed to the dissolution of coated waterglass and the released [SiO₄]^4- into MASM pastes could facilitate the generation of extra hydration products, which not only fill the pores and cracks in the microstructure but also enhance the bonding efficiency of MASM. In addition, the profiles of the pore structure of SwFS₁-added mixtures are right-shifted with reference to Ctrl when incorporating more SwFS₁, implying that the presence of SwFS₁ increases the pore size of MASM due to the presence of the coating of soluble waterglass. When incorporating SwFS₈ into MASM, the peak heights of SwFS₈-added mixtures are increased with the addition of SwFS₈, indicating that the incorporation of SwFS₈ could introduce more pores to the microstructure of MASM due to the sandwich structure of the interface. Furthermore, the profiles of the pore structure of SwFS₈-added MASM mortars have been moved to the left side of Ctrl, indicating increased pore sizes.

Figure 3 presents the typical SEM images of 28-day MASM mortars. As can be found in Figure 3A, the quartz sand displays a very smooth surface in the microstructure of Ctrl with some cracks in the interface, indicating physical bonding. In contrast, when SwFS₁ has been incorporated, as shown in Figure 3B, the interface between SwFS₁ and the surrounding paste has been modified very much, with no cracks. The surface of SwFS₁ is rough and contains some fine particles, which may be the hydration products produced by the activation of dissolved waterglass. The extra hydration products can not only densify the microstructure of mortars but also contribute to an enhanced bonding effect between SwFS₁ and the surrounding paste. In addition, some flake hydration products were also found in this figure, which could be the produced Mg(OH)₂ through the hydration of MgO.

Figure 3C presents the SEM image of SwFS₈100. This shows that the surface of SwFS₈ is covered with fewer hydration products than that of Figure 3B. SwFS₈ is poorly bonded by MASM with obvious cracks in the interface. This may be due to the lower solubility of SwFS₈ since fewer hydration products can be produced through dissolved waterglass with slag. Furthermore, the presence of the hardened waterglass coating further changed the structure of the interface from a normal “aggregate–paste” to a sandwich-like interface “aggregate–waterglass–paste.” This is much related to the result of compressive strength.

Nanoindentation was carried out to further investigate the role of the waterglass coating of SwFS in the micromechanical properties of 28-day MASM. Figures 4A–C show the optical microscope images of the interface of Ctrl, SwFS₁100, and SwFS₈100, respectively. The difference between the samples is that SwFS₈100 has an obvious black stripe between SwFS₈ and MASM pastes, which could be related to the file epoxy resin granules in pores or cracks during the polishing process for nanoindentation samples. The indented area is a 7 × 7 lattice distribution, with a section of approximately 10 μm in the interface. The results of the corresponding indented area were obtained in contour mapping analysis, as shown in Figures 4D–F. Figures 4A, Dshow a strong interface between quartz sand and MASM pastes, where the elastic modulus is approximately 20–60 GPa. This similar interface is also found in SwFS₁100 (Figure 4E), and the size of this zone in the interface is wider and gentler than that in Ctrl, indicating some positive effects of waterglass on the microstructure of the interface. A popular explanation is the substantial accumulation of homogeneous hydration products in the interface, as [SiO₄]^4- dissolved from waterglass can rapidly react with Ca or Mg to generate more hydration products, especially in the interface. However, Ma et al. (2022) indicated that this rapid hydration reaction could form incompletely dissolved waterglass, which is attributed to dense precipitations (mainly calcium silicate) outside waterglass. In MASM, Ca²⁺ and Mg²⁺ may compete to bond with [SiO₄]^4- (Li et al., 2022). The hydration reaction may prefer the formation of magnesium silicate precipitates due to the presence of abundant MgO. This may change the composition of precipitation outside waterglass. As a result, the waterglass of SwFS₁ may be completely dissolved in MASM, and SwFS₁100 presented a gentle and continuous interface microstructure.

In contrast, the interface of SwFS₈100 is highly inhomogeneous. Some areas (<10 GPa) are located between SwFS₈ and MASM pastes, indicating that the hydration products did not fill the whole pores in the interface even at 28 days. Porous interfaces may have a negative effect on the performance of SwFS₈-added mortars. However, the waterglass of SwFS₈ is not entirely detrimental to the micromechanical properties at the interface. Since SwFS₈ was calcined at a high temperature, waterglass is transformed into a hardened shell of waterglass, which adheres to the outer surface of SwFS₈ (Subasri and Näfe, 2008). This hardened coating can also be observed near SwFS₈ in Figure 4F, and its elastic modulus is approximately 20–60 GPa.

Thereafter, deconvolution was carried out to identify hydration products of pastes around aggregates as shown in Figures 5A, C. Although Hydrated magnesium silicaluminate (M-A-S-H) is a hydration product presented in MASM, its content is much less than that of C-A-S-H (Kim et al., 2022). The coordination of Mg in M-A-S-H is easily replaced by Ca to generate new C-A-S-H (Li et al., 2022; Ma et al., 2022). Therefore, deconvolution results were only analyzed for the main hydration products (C-A-S-H gels) and un-hydrated GGBS. Three phases can be obtained from nanoindentation testing for Ctrl, SwFS₁100, and SwFS₈100, which are C-A-S-H, crystals of slag, and unreacted slag. In these figures, π represents the volume fraction of the phase, μ represents the mean value (in GPa), and σ represents the standard deviation (in GPa). When adding SwFS₁, the average elastic moduli of C-A-S-H of SwFS₁100 were significantly increased by 19.6%, which could explain the enhanced compressive strength of mortars. Nevertheless, SwFS₈100 displays a poor interface with a much thicker layer of crystals with average elastic moduli of approximately 40–50 GPa, which may be due to the hardened waterglass and the formed Mg(OH)₂, reflecting the complex structure of the interface.