Early Strength-Promoting Mechanism of Inorganic Salts on Limestone-Calcined Clay Cement

: This study aims to report the early strength effect and hydration mechanisms of limestone-calcined clay cement (LC 3 ) with sodium carbonate, sodium sulfate and sodium chloride. The experimental results show that it is feasible to add three kinds of insoluble inorganic salts to improve the early strength of LC 3 through different promotion methods. In comparison to sodium sulfate, the strengthening effects of sodium carbonate and sodium chloride on early strength of LC 3 are more signiﬁcant. The hydration heat evolution, mercury intrusion porosity and a set of tests for microstructural characterization (XRD, FTIR and SEM) were utilized to better understand the enhancement mechanism of inorganic salts in LC 3 system. The mechanism by which sodium carbonate promotes the early strength of LC 3 is mainly the strengthening of the aluminate reaction and pozzolanic reaction of metakaolin. The mechanism by which sodium sulfate promotes the early strength of LC 3 is mainly the additional ettringite. The mechanism by which sodium chloride promotes the early strength of LC 3 is mainly the strengthening of the silicate reaction and the generation of Friedel’s salt by alumina from tricalcium aluminate and metakaolin


Introduction
Limestone calcined clay cement (LC 3 ) is a recently developed ternary cement system based on clinker, calcined clay (CC) and limestone (LS) [1], which can significantly reduce CO 2 emissions. Calcined clay and limestone (LC 2 ) as supplementary cementitious materials (SCMs) commonly used in cement can be used to replace clinker [2]. When they are added to the cement at a ratio of 2:1, a synergistic effect generated by CC and LS produces more carbo-aluminate phase, which refines the pore size, thus allowing LC 3 to achieve mechanical properties similar to OPC at a later stage with less clinker [3]. In addition, LC 3 also has excellent sulfate resistance, resistance to chloride ion penetration and an alkalisilica reaction (ASR) [4][5][6][7][8]. The consumption of portlandite by calcined clay reaction results in a low alkali environment, which is the main reason for its excellent ASR resistance [4,5]. Similarly, less available calcium ions in the secondary formation of gypsum and ettringite result in excellent sulfate resistance [8]. Furthermore, the ability of carbo-aluminate phase to bind chloride ions and the refinement of pore structure to reduce chloride ion transport are the main reasons for the chloride ion resistance of LC 3 [9].
Based on the above reasons, many researchers have widely studied LC 3 . S. Ferreiro et al. [10] investigated the workability and mechanical properties of two types (1:1 and 2:1) of calcined clay minerals in LC 3 . The results showed that the reaction speed of 1:1 calcined clay was faster than the reaction speed of 2:1 calcined clay. Blended cements containing 1:1 calcined clay require a higher water content than 2:1 calcined clay, so a superplasticizer was required to achieve the expected experimental results. Karen Scrivener et al. [9] investigated the effect of calcination temperature of clays on the properties of LC 3 . The results showed that the selection of the appropriate calcination temperature plays an important role in optimizing the properties of calcined clays. This is because early strength and reduce the water absorption of mortar. Meng Wu et al. [21] investigated the effect of sodium sulfate on the hydration and properties of lime-based low carbon cementitious materials. Sodium sulfate (3 wt%) could significantly improve the mechanical properties of cement, especially in the early stage. Yuanzhang Cao et al. [22] studied the effect of sodium chloride on the hydration of cement and found that the addition of chloride salts promoted the dissolution of minerals and increased the growth rate, resulting in improved mechanical properties.
The mechanism of action of dissoluble inorganic salt in the LC 3 system is more complex than that in other cements. On the one hand, dissoluble inorganic salt can affect the clinker part of LC 3 . On the other hand, they can also influence the pozzolanic reaction of LC 2 : carbonate ions can participate in the formation of CO 3 -AFm [23]; the addition of SO 3 can delay the aluminate reaction [24,25]; chloride ions added to the cement form Friedel's salt, existing competition or cooperation. For example, Zhenguo Shi et al. [26] found that adding metakaolin to silicate cement could improve the binding ability of chloride, and metakaolin could provide an additional source of aluminium, causing more Friedel's salts to form. Therefore, it is more esoteric to study the effect of dissoluble inorganic salt on LC 3 hydration. Presently, there are few comprehensive studies on the effect of dissoluble inorganic salts on the hydration of LC 3 . This study attempts to fill this gap. In addition, it must be pointed out in this study that adding soluble inorganic salts in the LC 3 system may reduce the durability of the cement, such as ASR and carbonation resistance [27,28]. However, current studies only focus on the enhancement effect and mechanism of inorganic salts on the early strength in LC 3 system, and the durability of LC 3 system has not been studied.
In this paper, the influence of sodium carbonate, sodium sulfate, and sodium chloride on the hydration of LC 3 cement was studied. LC 3 paste with different dissoluble inorganic salts was prepared to determine the compressive strength. The heat of hydration was measured by isothermal calorimetry. Phase analysis of the hydration products was carried out by X-ray diffraction, scanning electron microscopy and infrared spectroscopy. Finally, a comprehensive evaluation of the hydration process and hydration products of different dissoluble inorganic salts in LC 3 was performed. In view of both improving the early strength of LC 3 and promoting the application of seawater containing various inorganic salts in LC 3 concrete, it is significant to study the early strength effect and hydration mechanisms of LC 3 with inorganic salts.

Materials
The ordinary Portland cement (OPC) clinker used in this study was obtained from Zhejiang Xindu Cement Co., Ltd. in Jiaxing, China, and limestone (LS) and gypsum were acquired from Tonglu South Cement Co., Ltd., in Tonglu, Hangzhou, China. Clay was purchased from Shangxing Chemical Technology Co., Ltd., in Maoming, China. Calcined clay (CC) was obtained by calcination at 750 • C and holding for 1 h. Figure 1 shows the particle size distributions of the raw materials. Table 1 lists the chemical composition of the materials, as determined through X-ray fluorescence (XRF) analysis. The purity of experimental grade sodium carbonate, sodium sulfate and sodium chloride were 99.9%.

Mixture Preparation and Methods
The mix proportion of clinker, calcined clay, limestone and gypsum in LC 3 binder pastes was 50:30:15:5. The addition rate of sodium carbonate was 1.00% by solid mass. In order to maintain the additional Na 2 O eq values, the added sodium sulfate and sodium chloride solids were 1.34% and 1.10%, respectively. The mix proportions of the binder pastes are listed in Table 2.

Materials
The ordinary Portland cement (OPC) clinker used in this study was obtained from Zhejiang Xindu Cement Co., Ltd. in Jiaxing, China, and limestone (LS) and gypsum were acquired from Tonglu South Cement Co., Ltd., in Tonglu, Hangzhou, China. Clay was purchased from Shangxing Chemical Technology Co., Ltd., in Maoming, China. Calcined clay (CC) was obtained by calcination at 750 °C and holding for 1 h. Figure 1 shows the particle size distributions of the raw materials.  In addition, to maintain workability, 0.3 wt% superplasticizers were added to the samples during the mixing process with water [29]. A water to binder ratio (w/b) of 0.4 was used for all mixtures [30].
Before preparing the pastes, the corresponding proportions of sodium carbonate, sodium sulfate and sodium chloride were first placed in water. Then, an ultrasonic homogenizer was used to effectively scatter the materials.
The prepared paste was cast into a 20 × 20 × 20 mm steel mould and vibrated for 1 min to vent bubbles from the paste. Demoulding was performed after 24 h of steam curing at 20 ± 1 • C and 90% relative humidity. Then, the samples were cured in water until they reached certain ages. The compressive strength values used by the compression testing machine at loading speed of 2.4 kN/s show the averages from the six samples at 3, 7 and 28 days. In addition, samples damaged by press were first soaked in ethanol solution to stop hydration, placed in a vacuum box and dried for 24 h, and then used to conduct microstructural characterization tests. In each group of crushed paste, 4-5 block samples with sizes less than 1 × 1 × 1 cm were selected for pore structure analysis. The pore structure of fragmented samples was evaluated by mercury intrusion porosimetry (MIP, Micro Active Auto Pore V 9600). The measured pore size ranged from 5 nm to 0.8 mm. The relatively flat thin slice sample was cut out with a knife and treated with gold-plate and then used for the observation of surface morphology by Scanning electron microscopy (SEM, FEI FEG650). An accelerating voltage of 20 kV and a working distance of 10.3 mm were used. Since the mechanically ground powder may be decomposed by the ettringite phase and AFm-CO 3 phase in the LC 3 system, the broken samples were ground manually to about 45 µm for XRD and FTIR. The crystalline phases of the hydration products were characterized by employing an X-ray diffractometer (XRD, Bruker D8 advance) with a slow scan, rate of 2 • /min. The vibration of the chemical bond of the hydration products was measured by Fourier transform infrared spectroscopy (FTIR, Thermo Scientific Nicolet iS20) in the spectral range of 4000-5000 cm −1 . A TAM Air isothermal calorimeter was used to record the heat of hydration of the four tested groups in the first 72 h. A water to binder ratio (w/b) of 0.5 was used for all the mixtures. The experiment was conducted at 20 • C. Figure 2 shows the effect of inorganic salt on the compressive strength of LC 3 . The strengths of the LC 3 blank group were 35.4, 56.9 and 73.5 MPa after 3, 7 and 28 days, respectively. The addition of inorganic salt significantly increased the early strength of LC 3 . At 3 days, the addition of sodium carbonate, sodium sulfate, and sodium chloride increased the strength of LC 3 by 38%, 28% and 44%, respectively, and the improvement effect was obvious. At 7 days, the compressive strength of LC 3 -NaCl still increased by 17%, while the compressive strength of LC 3 -Na 2 CO 3 and LC 3 -Na 2 SO 4 remained basically unchanged compared with the control group. At 28 days, sodium carbonate and sodium sulfate showed a 15% and 10% reduction in the compressive strength of LC 3 , respectively, and only LC 3 -NaCl was similar to that of the control group. Obviously, the addition of three kinds of dissoluble inorganic salts can significantly improve the early strength, the mechanism of which will be discussed in the following article. powder may be decomposed by the ettringite phase and AFm-CO3 phase in the LC 3 system, the broken samples were ground manually to about 45 µm for XRD and FTIR. The crystalline phases of the hydration products were characterized by employing an X-ray diffractometer (XRD, Bruker D8 advance) with a slow scan, rate of 2°/min. The vibration of the chemical bond of the hydration products was measured by Fourier transform infrared spectroscopy (FTIR, Thermo Scientific Nicolet iS20) in the spectral range of 4000-5000 cm −1 . A TAM Air isothermal calorimeter was used to record the heat of hydration of the four tested groups in the first 72 h. A water to binder ratio (w/b) of 0.5 was used for all the mixtures. The experiment was conducted at 20 °C. Figure 2 shows the effect of inorganic salt on the compressive strength of LC 3 . The strengths of the LC 3 blank group were 35.4, 56.9 and 73.5 MPa after 3, 7 and 28 days, respectively. The addition of inorganic salt significantly increased the early strength of LC 3 . At 3 days, the addition of sodium carbonate, sodium sulfate, and sodium chloride increased the strength of LC 3 by 38%, 28% and 44%, respectively, and the improvement effect was obvious. At 7 days, the compressive strength of LC 3 -NaCl still increased by 17%, while the compressive strength of LC 3 -Na2CO3 and LC 3 -Na2SO4 remained basically unchanged compared with the control group. At 28 days, sodium carbonate and sodium sulfate showed a 15% and 10% reduction in the compressive strength of LC 3 , respectively, and only LC 3 -NaCl was similar to that of the control group. Obviously, the addition of three kinds of dissoluble inorganic salts can significantly improve the early strength, the mechanism of which will be discussed in the following article. However, in addition to LC 3 -NaCl, the strength of LC 3 -Na2CO3 and LC 3 -Na2SO4 decreased to varying degrees in the later stage, which may be because of the addition of sodium ions. Some previous studies [31,32] showed that adding alkali to cement reduces the later age strength achievement. On the one hand, it changed the atomic structure and density of calcium silicate hydrate (C-S-H), reducing the strength of the body [31]. On the other hand, it may be because sodium ions can combine with non-bridging oxygen in the saturated monomers to form Na-Onb connection and finally attach to tricalcium silicate, which indicates that the dissolution of ions in C3S is inhibited [32]. However, in addition to LC 3 -NaCl, the strength of LC 3 -Na 2 CO 3 and LC 3 -Na 2 SO 4 decreased to varying degrees in the later stage, which may be because of the addition of sodium ions. Some previous studies [31,32] showed that adding alkali to cement reduces the later age strength achievement. On the one hand, it changed the atomic structure and density of calcium silicate hydrate (C-S-H), reducing the strength of the body [31]. On the other hand, it may be because sodium ions can combine with non-bridging oxygen in the saturated monomers to form Na-O nb connection and finally attach to tricalcium silicate, which indicates that the dissolution of ions in C 3 S is inhibited [32].

Hydration Heat Evolution
The heat flow and cumulative heat of hydration measured per gram of binder are shown in Figure 3. Three major hydration peaks can be seen, numbered I, II and III. The initial dissolution peak of aluminate is not considered. The first hydration peak is the alite peak after the end of the induction period. The first hydration peak is the alite peak after the end of the induction period [24]. The heat flow and cumulative heat of hydration measured per gram of binder are shown in Figure 3. Three major hydration peaks can be seen, numbered I, II and III. The initial dissolution peak of aluminate is not considered. The first hydration peak is the alite peak after the end of the induction period. The first hydration peak is the alite peak after the end of the induction period [24]. Early dissolution peaks before the end of the induction period accounted for a small portion of the total hydration heat and could be excluded from the analysis. The alite (I) peak of LC 3 appeared at approximately 10.5 h, and the addition of sodium sulfate and sodium chloride accelerated the appearance of the main peak at approximately 8.5 h, significantly enhancing the main peak strength. The effect of sodium chloride was stronger than that of sodium sulfate in the I peak. The aluminate (II) peak of LC 3 appeared at approximately 16.5 h. The reaction of aluminate was accelerated by the addition of sodium carbonate, and, in LC 3 -Na2CO3, the aluminate (II) peak appeared at approximately 15 h and enhanced the peak strength. For LC 3 -Na2SO4 and LC 3 -NaCl, the aluminate (II) peak appeared earlier due to the accelerated hydration of silicate and the early depletion of solid gypsum. However, the peak strength of LC 3 -Na2SO4 is smaller than that of LC 3 -control. The III peak appeared at 30-60 h, and the peak strength was smaller in LC 3 -control and LC 3 -Na2SO4, which was clearly observed in LC 3 -Na2CO3 and LC 3 -NaCl.
For LC 3 -Na2CO3, carbonate ions, present in the solution due to the addition of sodium carbonate, react with solid gypsum and accelerate the precipitation of sulfate ions from gypsum, which are rapidly adsorbed on C-(A)-S-H, so that the aluminate (II) peak appears earlier. Zhang et al. [19] also found that adding Na2CO3 accelerated the cement hydration. In addition, the reaction between sodium carbonate and gypsum also generates nano-calcium carbonate, leaving Na + in the solution, which improves the alkalinity of the solution and enhances the solubility of metakaolin. A similar phenomenon was reported by Sreejith Krishnan [33]. Therefore, the third peak of LC 3 -Na2CO3 is considerably stronger than that of LC 3 -control. For LC 3 -Na2SO4, due to the addition of sulfate, the aluminate reaction will be strengthened in theory, but the silicate peak is strengthened, and the aluminate peak is slightly decreased. This may be due to the addition of soluble sulfate. The concentration of sulfate ions in the solution was high at the beginning, and the alumina from C3A or metakaolin reacted with it immediately, thus strengthening the silicate (I) peak, which also explains the decrease in the aluminate (II) peak. Therefore, Na2SO4 can promote ettringite formation and improve mechanical properties in LC 3 , which is in consistent with the literature [27,31]. For LC 3 -NaCl, similar to the OPC system, the addition of chloride Early dissolution peaks before the end of the induction period accounted for a small portion of the total hydration heat and could be excluded from the analysis. The alite (I) peak of LC 3 appeared at approximately 10.5 h, and the addition of sodium sulfate and sodium chloride accelerated the appearance of the main peak at approximately 8.5 h, significantly enhancing the main peak strength. The effect of sodium chloride was stronger than that of sodium sulfate in the I peak. The aluminate (II) peak of LC 3 appeared at approximately 16.5 h. The reaction of aluminate was accelerated by the addition of sodium carbonate, and, in LC 3 -Na 2 CO 3 , the aluminate (II) peak appeared at approximately 15 h and enhanced the peak strength. For LC 3 -Na 2 SO 4 and LC 3 -NaCl, the aluminate (II) peak appeared earlier due to the accelerated hydration of silicate and the early depletion of solid gypsum. However, the peak strength of LC 3 -Na 2 SO 4 is smaller than that of LC 3 -control. The III peak appeared at 30-60 h, and the peak strength was smaller in LC 3 -control and LC 3 -Na 2 SO 4 , which was clearly observed in LC 3 -Na 2 CO 3 and LC 3 -NaCl.
For LC 3 -Na 2 CO 3 , carbonate ions, present in the solution due to the addition of sodium carbonate, react with solid gypsum and accelerate the precipitation of sulfate ions from gypsum, which are rapidly adsorbed on C-(A)-S-H, so that the aluminate (II) peak appears earlier. Zhang et al. [19] also found that adding Na 2 CO 3 accelerated the cement hydration. In addition, the reaction between sodium carbonate and gypsum also generates nanocalcium carbonate, leaving Na + in the solution, which improves the alkalinity of the solution and enhances the solubility of metakaolin. A similar phenomenon was reported by Sreejith Krishnan [33]. Therefore, the third peak of LC 3 -Na 2 CO 3 is considerably stronger than that of LC 3 -control. For LC 3 -Na 2 SO 4 , due to the addition of sulfate, the aluminate reaction will be strengthened in theory, but the silicate peak is strengthened, and the aluminate peak is slightly decreased. This may be due to the addition of soluble sulfate. The concentration of sulfate ions in the solution was high at the beginning, and the alumina from C 3 A or metakaolin reacted with it immediately, thus strengthening the silicate (I) peak, which also explains the decrease in the aluminate (II) peak. Therefore, Na 2 SO 4 can promote ettringite formation and improve mechanical properties in LC 3 , which is in consistent with Sustainability 2023, 15, 5286 7 of 13 the literature [27,31]. For LC 3 -NaCl, similar to the OPC system, the addition of chloride ions can accelerate the dissolution of silicate minerals, thus increasing the reaction rate, so the I peak of LC 3 -NaCl is strengthened. The enhancement of the III peak is due to the volcanic ash reaction between the chloride ions in NaCl and alumina from tricalcium aluminate or metakaolin to form Cl-AFm, Friedel's salt, which enhances the peak strength. In addition, there is additional conversion of CO 3 -AFm to Cl-AFm. This was also confirmed by XRD of LC 3 -NaCl at 3 days. Figure 3b shows the cumulative hydration heat curve of the binder over 72 h. Compared with LC 3 -control, the cumulative heat release curves of LC 3 -Na 2 SO 4 and LC 3 -NaCl are always in a leading position after the induction period, and the same is true of the cumulative heat release of LC 3 -Na 2 CO 3 after the aluminate reaction. In general, the addition of three dissoluble inorganic salts to LC 3 improved the cumulative heat release and increased the hydration of the binder, which was also consistent with the improvement in the compressive strength of LC 3 at 3 days. Therefore, in Figure 2, LC 3 -Na 2 CO 3 , LC 3 -Na 2 SO 4 and LC 3 -NaCl have better compressive strength than the control group after 3 days.

XRD
In the LC 3 binder, the hydration phase can be characterized by XRD, and the strength of the phase characteristic peak can reflect the hydration degree of cement. In this section, XRD characterization was conducted on the binder after 3 days and 28 days, and the results are shown in Figure 4. The main hydration products of LC 3 were ettringite (AFt), portlandite (CH), hemicarbonate (Hc) and monocarbonate (Mc) [25,[34][35][36]]. The addition of sodium sulfate and sodium carbonate only affected the relative contents of each hydration product. It did not affect the phase assemblage of the hydrating paste. However, the addition of sodium chloride produces a new phase of Friedel's salt (Fs) [26,37]. Based on XRD characterization, the addition of inorganic salts can make LC 3 produce more hydration products in the early stage, so the sample with inorganic salts has a higher early strength. ions can accelerate the dissolution of silicate minerals, thus increasing the reaction rate, so the I peak of LC 3 -NaCl is strengthened. The enhancement of the III peak is due to the volcanic ash reaction between the chloride ions in NaCl and alumina from tricalcium aluminate or metakaolin to form Cl-AFm, Friedel's salt, which enhances the peak strength.
In addition, there is additional conversion of CO3-AFm to Cl-AFm. This was also confirmed by XRD of LC 3 -NaCl at 3 days. Figure 3b shows the cumulative hydration heat curve of the binder over 72 h. Compared with LC 3 -control, the cumulative heat release curves of LC 3 -Na2SO4 and LC 3 -NaCl are always in a leading position after the induction period, and the same is true of the cumulative heat release of LC 3 -Na2CO3 after the aluminate reaction. In general, the addition of three dissoluble inorganic salts to LC 3 improved the cumulative heat release and increased the hydration of the binder, which was also consistent with the improvement in the compressive strength of LC 3 at 3 days. Therefore, in Figure 2, LC 3 -Na2CO3, LC 3 -Na2SO4 and LC 3 -NaCl have better compressive strength than the control group after 3 days.

XRD
In the LC 3 binder, the hydration phase can be characterized by XRD, and the strength of the phase characteristic peak can reflect the hydration degree of cement. In this section, XRD characterization was conducted on the binder after 3 days and 28 days, and the results are shown in Figure 4. The main hydration products of LC 3 were ettringite (AFt), portlandite (CH), hemicarbonate (Hc) and monocarbonate (Mc) [25,[34][35][36]]. The addition of sodium sulfate and sodium carbonate only affected the relative contents of each hydration product. It did not affect the phase assemblage of the hydrating paste. However, the addition of sodium chloride produces a new phase of Friedel's salt (Fs) [26,37]. Based on XRD characterization, the addition of inorganic salts can make LC 3 produce more hydration products in the early stage, so the sample with inorganic salts has a higher early strength. At 3 days, compared with LC 3 -Control, LC 3 -Na2CO3 had a more prominent Hc diffraction peak and Mc diffraction peak, especially the diffraction peak of Mc, which was quite obvious, while the CH diffraction peak was weakened. This is because the carbonate ions in sodium carbonate dissolved in water and formed nanometre calcium carbonate with calcium ions in solution, which is more favourable to the pozzolanic reaction of metakaolin, consuming more CH and generating more CO3-AFm. This enhanced pozzolanic reaction phenomenon is confirmed in the hydration heat evolution of pastes.
For LC 3 -Na2SO4, the ettringite diffraction peak intensity was significantly higher than that of LC 3 , while the Mc diffraction peak intensity was weakened. This is mainly because the supplementary sulfate reacted with alumina from C3A or metakaolin at the beginning, At 3 days, compared with LC 3 -Control, LC 3 -Na 2 CO 3 had a more prominent Hc diffraction peak and Mc diffraction peak, especially the diffraction peak of Mc, which was quite obvious, while the CH diffraction peak was weakened. This is because the carbonate ions in sodium carbonate dissolved in water and formed nanometre calcium carbonate with calcium ions in solution, which is more favourable to the pozzolanic reaction of metakaolin, consuming more CH and generating more CO 3 -AFm. This enhanced pozzolanic reaction phenomenon is confirmed in the hydration heat evolution of pastes.
For LC 3 -Na 2 SO 4 , the ettringite diffraction peak intensity was significantly higher than that of LC 3 , while the Mc diffraction peak intensity was weakened. This is mainly because the supplementary sulfate reacted with alumina from C 3 A or metakaolin at the beginning, generating additional ettringite, which also reduced the alumina that was used to produce CO 3 -AFm. Thus, the Mc diffraction peak intensity was weakened. In LC 3 -NaCl mixtures, the ettringite diffraction peak intensity was enhanced, and the CO 3 -AFm diffraction peak almost disappeared. In addition, a new phase, Fs (2θ = 11.4), was observed [38]. The formation of Fs originated from two parts: one was directly generated, similar to the OPC system when sodium chloride was added to the binder [39], and the other was converted by Mc, according to Equation (1) [26]. This could also explain the greatly diminished Mc diffraction peak. Numerous studies have also reported this transition from Mc to Fs [26,37].
As shown in Figure 3b, the hydration product phases of LC 3 after 28 days only changed the relative content. As the hydration progressed, the ettringite diffraction peak was enhanced compared to that at 3 days. The diffraction peaks of Hc and Mc were particularly obvious. The portlandite diffraction peak of LC 3 was greatly diminished. LC 3 -Na 2 SO 4 and LC 3 -Na 2 CO 3 showed a lower CO 3 -AFm diffraction peak and a higher portlandite diffraction peak due to the slowing hydration rate in the later stages. The portlandite diffraction peak of LC 3 -Na 2 CO 3 was particularly high probably due to the early acceleration of the pozzolanic reaction of metakaolin, which led to the rapid refinement of voids. Because of the lack of capillary pores, the formation of Hc and Mc was inhibited, and the consumption of portlandite was reduced [24]. The refinement of the pore size of LC 3 -Na 2 CO 3 is discussed later in the characterization of MIP. Figure 5 shows the FTIR spectra of the samples in the scan range of 500-4000 cm −1 at 3 days and 28 days. The absorption peaks near 3642 cm −1 can be attributed to the vibrations of the OH bond of portlandite [40]. At 3 days, the absorption peak of CH of LC 3 -Na 2 CO 3 was smaller than that of the other groups with added inorganic salt because the enhanced pozzolanic reaction increased the CH consumption. However, the absorption peak of CH was also not significant in the LC 3 -control group, which may be due to sample carbonation during the experimental stage, leading to the consumption of CH [27]. This is also reflected in the later O-H of water transformation. The absorption peaks at 3443 and 1670 cm −1 represent the stretching and bending vibrations of the O-H bond of chemically combined water, respectively [40]. At 3 days, LC 3 -Control carbonation took place, and H-O-H at approximately 1670 cm −1 shifted to a smaller peak at 1650 cm −1 . The absorption peaks at 1433 and 878, 706 cm −1 are related to the asymmetric stretching vibration and bending vibration of C-O bonds, respectively [21]. The absorption peaks at 706 cm −1 can be attributed to limestone, where the peaks of LC 3 -Na 2 CO 3 and LC 3 -NaCl were more prominent because they produced nanometre calcium carbonate and left more unreacted limestone, respectively. The absorption peaks at 1433 cm −1 of LC 3 -Na 2 CO 3 were higher than those of the other samples due to additional Hc and Mc. The absorption peaks at 1110 cm −1 correspond to the SO 4 2− of ettringite [30]. The absorption band intensity of LC 3 -Na 2 SO 4 was strengthened, which was caused by the additional ettringite. The absorption band at 974 cm −1 corresponds to the in-plane stretching vibration of the Q2 tetrahedra in C-A-S-H [27]. As shown in Figure 5b, the largest difference between the 28 days FTIR and the 3 days FTIR of the samples was that the absorption peaks at 3443 cm −1 and 1443 cm −1 were enhanced. This may be due to the production of more C-A-S-H gel and carboaluminates, respectively, with hydration.

FTIR
Combined with FTIR and XRD analysis, we can see that the addition of inorganic salts promoted the generation of more hydration products. Meanwhile, the hydration heat evolution also indicated the intensification of cement hydration, so the sample with inorganic salts had better compressive strength after 3 days. Combined with FTIR and XRD analysis, we can see that the addition of inorganic salts promoted the generation of more hydration products. Meanwhile, the hydration heat evolution also indicated the intensification of cement hydration, so the sample with inorganic salts had better compressive strength after 3 days. Figure 6 shows the MIP results of LC 3 -Control, LC 3 -Na2CO3, LC 3 -Na2SO4 and LC 3 -NaCl paste at 3 days [27,41]. Figure 6a shows that at 3 days, the pore volume of all LC 3 samples with the addition of inorganic salts is smaller than the pore volume of LC 3 -control. According to the hydration heat evolution of the samples, the addition of three kinds of inorganic salts accelerates the hydration of LC 3 in different ways, resulting in the production of more hydration products such as C-(A)-S-H gel, CO3-AFm, ettringite and Friedel's salt. Additional hydration products can fill the pores and form a denser structure [34,[41][42][43]. Figure 6d further illustrates the porosity variation of the samples [44]. Adding Na2CO3, Na2SO4 and NaCl can reduce the MIP porosity from 31.55% to 25.46%, 25.77% and 25.39%, respectively, with a drop of approximately 20%. Because the smaller the porosity, the denser the structure, the greater the compressive strength of the sample, so the sample with inorganic salts has a higher early strength in Figure 2. Figure 6b,c show the change in the pore size of the sample in further detail. According to the pore size distribution curves, the peaks of the pore size distribution curves of the samples moved in the direction of small pore size after inorganic salts were added. The pore size percentages of different samples were analysed, and it was found that the percentage of pore sizes in the range of 20-50 nm and 50-100 nm was greatly reduced after the addition of inorganic salts, while the percentage of pore sizes below 20 nm was greatly increased. In particular, the pore sizes less than 10 nm in LC 3 -Na2CO3 and LC 3 -NaCl were 25.78% and 29.6%, respectively. This also demonstrates that additional CO3-AFm and Cl-AFm were produced, which greatly refined the pore size. The percentage of pore size less than 20 nm of LC 3 -Na2CO3 far exceeds that of the other groups, reaching approximately 73% of pore sizes less than 20 nm. A previous study [20] showed that Na2CO3 refine the pore size in cement. However, as described in the literature [24], rapid porosity refinement would affect the subsequent precipitation of Hc and Mc, which would have a detrimental effect on the later pozzolanic reaction of metakaolin.  Figure 6 shows the MIP results of LC 3 -Control, LC 3 -Na 2 CO 3 , LC 3 -Na 2 SO 4 and LC 3 -NaCl paste at 3 days [27,41]. Figure 6a shows that at 3 days, the pore volume of all LC 3 samples with the addition of inorganic salts is smaller than the pore volume of LC 3 -control. According to the hydration heat evolution of the samples, the addition of three kinds of inorganic salts accelerates the hydration of LC 3 in different ways, resulting in the production of more hydration products such as C-(A)-S-H gel, CO 3 -AFm, ettringite and Friedel's salt. Additional hydration products can fill the pores and form a denser structure [34,[41][42][43]. Figure 6d further illustrates the porosity variation of the samples [44]. Adding Na 2 CO 3 , Na 2 SO 4 and NaCl can reduce the MIP porosity from 31.55% to 25.46%, 25.77% and 25.39%, respectively, with a drop of approximately 20%. Because the smaller the porosity, the denser the structure, the greater the compressive strength of the sample, so the sample with inorganic salts has a higher early strength in Figure 2. Figure 6b,c show the change in the pore size of the sample in further detail. According to the pore size distribution curves, the peaks of the pore size distribution curves of the samples moved in the direction of small pore size after inorganic salts were added. The pore size percentages of different samples were analysed, and it was found that the percentage of pore sizes in the range of 20-50 nm and 50-100 nm was greatly reduced after the addition of inorganic salts, while the percentage of pore sizes below 20 nm was greatly increased. In particular, the pore sizes less than 10 nm in LC 3 -Na 2 CO 3 and LC 3 -NaCl were 25.78% and 29.6%, respectively. This also demonstrates that additional CO 3 -AFm and Cl-AFm were produced, which greatly refined the pore size. The percentage of pore size less than 20 nm of LC 3 -Na 2 CO 3 far exceeds that of the other groups, reaching approximately 73% of pore sizes less than 20 nm. A previous study [20] showed that Na 2 CO 3 refine the pore size in cement. However, as described in the literature [24], rapid porosity refinement would affect the subsequent precipitation of Hc and Mc, which would have a detrimental effect on the later pozzolanic reaction of metakaolin.

Microstructure Morphology
As shown in Figure 7, the microstructure of LC 3 samples is analysed using SEM. Clusters of C-(A)-S-H and hexagonal-platelet portlandite crystals can be seen in the image. In addition, ettringite is clearly seen growing from the C-(A)-S-H. Based on the heat evolution and MIP results, the hydration rate of LC 3 with inorganic salt is higher than that of LC 3 -control, and LC 3 with inorganic salt has a much denser structure. It makes the sample with inorganic salts have better macro performance in compressive strength.

Microstructure Morphology
As shown in Figure 7, the microstructure of LC 3 samples is analysed using SEM. Clusters of C-(A)-S-H and hexagonal-platelet portlandite crystals can be seen in the image. In addition, ettringite is clearly seen growing from the C-(A)-S-H. Based on the heat evolution and MIP results, the hydration rate of LC 3 with inorganic salt is higher than that of LC 3 -control, and LC 3 with inorganic salt has a much denser structure. It makes the sample with inorganic salts have better macro performance in compressive strength. As shown in Figure 7a, the LC 3 -control sample has many pores and unconsumed portlandite, which is a relatively loose structure. Figure 7b,d show that the structure is more compact, and the pore size is smaller. As shown in Figure 7b, the additional production of plate-like carboaluminates fills the pores. In addition, the portlandite crystal is significantly reduced because it is consumed. This is also demonstrated by XRD in Figure 4a. Grid-turned C-(A)-S-H can be observed in Figure 7d, which may be because chloride ions not only accelerate the formation of C-(A)-S-H but also change the morphology of C-(A)- As shown in Figure 7a, the LC 3 -control sample has many pores and unconsumed portlandite, which is a relatively loose structure. Figure 7b,d show that the structure is more compact, and the pore size is smaller. As shown in Figure 7b, the additional production of plate-like carboaluminates fills the pores. In addition, the portlandite crystal is significantly reduced because it is consumed. This is also demonstrated by XRD in Figure 4a. Gridturned C-(A)-S-H can be observed in Figure 7d, which may be because chloride ions not only accelerate the formation of C-(A)-S-H but also change the morphology of C-(A)-S-H [37,45]. In addition, the formation of a new phase of plate-like Friedel's salt was observed. In Figure 7c, a divergent morphology of C-(A)-S-H can be seen, which may be because the presence of sulfate and sodium ions act on C-(A)-S-H at the same time, prompting it to change its form [31]. In Figure 7c, many columnar ettringites are also found, probably because sodium sulfate is added to provide additional sulfate ions.

Conclusions
In this study, three different dissoluble inorganic salts were used to determine the early strength of LC 3 paste. The results show that it is feasible to add inorganic salts to improve the early strength of LC 3 , but it has an impact on later strength development. At 3 days, the most obvious effects on the early strength of LC 3 are produced by sodium carbonate and sodium chloride, which increase by 38% and 44%, respectively. The early strengthening mechanism of the hydration process was illustrated by hydration heat evolution, and hydration products were evaluated by XRD, FTIR, MIP and SEM. The main conclusions are as follows: (1) The carbonate ions in the solution react with solid gypsum, accelerating the precipitation of sulfate ions from gypsum, which are rapidly adsorbed on C-(A)-S-H so that the aluminate reaction strengthened. Then, the pozzolanic reaction occurs, improved by nano-calcium carbonate and an alkaline environment that enhances the solubility. (2) Supplementary sulfate ion from sodium sulfate reacted with alumina from C 3 A or metakaolin at the beginning, generating additional ettringite and promoting the early strength of LC 3 . (3) The sodium chloride reacts with alumina from tricalcium aluminate and metakaolin in the pozzolanic reaction, generating Friedel's salt, thus refining the pore size and enhancing the early strength of LC 3 . (4) At the early stage in LC 3 system with NaCl, the formation of carbo-aluminate phases had a competitive relationship with the formation of Friedel's salt. The pozzolanic reaction products of metakaolin in LC 3 were preferentially Friedel's salts, and the generated carbo-aluminate phases were also converted into Friedel's salt. At the later stage, the Friedel's salt phase and the carbo-aluminate phase together played a role in refining pore size and increasing strength, which was one of the reasons why LC 3 -NaCl still had a high strength after 28 days. Funding: This research received no external funding.
Data Availability Statement: All data, models and code generated or used during the study appear in the submitted article.