Pure calcium aluminate cement (SECAR 71) was supplied by the Sinosteel Luoyang Institute of Refractories Research (Luoyang, China), the chemical composition of which is listed in Table 1. Analytical-grade SHMP was purchased from Luoyang Hexin Refractories Co., Ltd (Luoyang, China).

Based on the residual concentration method, the concentrations of P element determined by ICAP can be used to estimate the amounts of SHMP in the filtrate and adsorbed onto the surfaces of cement particles. At the beginning of the experiment, the cement particles sank to the bottom, and the adsorption amount of SHMP on its surface was very little, almost none. The concentration of p element was measured and compared after standing for 5 min and 30 min, and it was found that there was no change. Therefore, stirring was adopted to enhance its adsorption capacity. Figure 1 shows these two amounts as a function of total addition of SHMP and stirring time, revealing that the amounts of SHMP adsorbed onto the surfaces of cement particles increased almost linearly with the total addition of SHMP (Figure 1c,d). Figure 2 further illustrates the change of the adsorption ratio of SHMP (i.e., the ratio of the amount of SHMP adsorbed onto cement particles to the total addition of SHMP) with the total addition of SHMP and stirring time. When 0.05% SHMP was added, the adsorption ratio after 5 min stirring reached 99.6%. However, it decreased to 89.2% upon increasing the SHMP addition to 0.4%. Furthermore, comparison of Figure 2a,b reveals that, for a given addition of SHMP, the adsorption ratio after 30 min stirring was greater than that after 5 min stirring, indicating that the stirring assisted the adsorption of SHMP onto the cement particles, as the stirring time increased from 5 min to 30 min, the adsorption capacity was enhanced.

From Figure 1a,b, it also can be seen that only minor amounts of P element were detected in the filtrates. This might be due to one of the following reasons: (1) the chelation of SHMP with Ca2+ in the solution resulted in insoluble phosphates, and (2) the chelation of SHMP with Ca2+ in the solution resulted in soluble complexes which were adsorbed onto the surfaces of cement particles, and re-precipitated with cement particles. Previous studies [28,29,30] showed that the phosphates generated from the strong chelation of SHMP with Ca2+ were water soluble, so the first case could be ruled out. Therefore, it can be considered that the complexes formed from the chelation of SHMP with Ca2+ (see Reaction (1) in Section 3.2 below) were adsorbed onto the surfaces of cement particles, leaving only a little P element in the filtrates.

Sodium Hexametaphosphate

Figure 3 illustrates the change of Ca2+ concentration in the filtrate with the total addition of SHMP. After 5 min stirring, the Ca2+ concentration in the case of no SHMP was 497.3 mg/mL, but it decreased to 186.4 mg/mL upon increasing the SHMP addition to 0.2%. The Ca2+ concentration became almost constant as the total addition of SHMP was increased to >0.2%. Comparison of Figure 3a,b reveals that the Ca2+ concentration decreased with increasing the stirring time. The above results indicated that the addition of SHMP inhibited the dissolution of Ca2+, which was additionally assisted by stirring, and with the increase of stirring time, the inhibitory effect is better. This was consistent with that found from the adsorption tests presented above (Section 3.1), and could be similarly explained. In the presence of SHMP, it combined with Ca2+ to form water-soluble [Ca2(PO3)6]2− ions (Reaction (1)) [30,31,32], which were adsorbed onto the surfaces of cement particles and subsequently re-precipitated with them, resulting in the reduced Ca2+ concentration in the filtrates. The stirring promoted the chelation of SHMP with Ca2+ and thus the whole process stated above. According to Figure 3, the Ca2+ concentration reached the minimum upon addition of about 0.2% SHMP, indicating the best water reducing effect.

Figure 4 shows the ζ-potential of cement particles as a function of the addition amount of SHMP. ζ-potential in the case of without SHMP was 8.5 mV. However, it changed to −14.8 and −20.2 mV upon adding, respectively, 0.1% and 0.2% SHMP. Upon further increasing the SHMP addition from 0.2% to 0.4%, the ζ-potential almost did not change. These results and their significance can be discussed as follows.

As indicated by Reaction (1) and mentioned above, when SHMP was added, water soluble [Ca2(PO3)6]2− ions were formed due to its chelation with Ca2+, which were subsequently adsorbed onto the surfaces of cement particles, resulting in a reduced concentration of P element in the filtrate. The adsorption of [Ca2(PO3)6]2− ions on the surfaces of cement particles led to in the change of ζ-potential from positive to negative. With increasing the amount of SHMP from 0 to 0.2%, more and more [Ca2(PO3)6]2− ions were formed and accumulated onto the surfaces of cement particles, which led to the significant increase in the absolute value of ζ-potential. When the SHMP addition was >0.2%, the surfaces of cement particles became “saturated” with complexes [Ca2(PO3)6]2− ions, and the negative electric strength of cement particles reached the maximum value; thus, the electrostatic repulsion between the cement particles became the largest. Consequently, further increasing the amount of SHMP to above 0.2% did not lead to any obvious change in the ζ-potential value. The great increase in the ζ-potential value with the SHMP addition implied that the dispersion of cement particles could be significantly improved. When the concentration of SHMP is greater than 0.2%, the dispersibility of PCAC is not further improved. SHMP contains sodium ions, so when the concentration of SHMP is increased, the concentration of Na+ in the solution will increase accordingly. Keita Irisawa and others mentioned that, in the formulation of refractory castables, the main refractory raw materials, cement and sodium salt additives all contain soluble sodium, which is easy to form into low-melting sodium salt, which is easy to collapse at high temperatures, which affects the refractoriness of the castable and other high-temperature performance properties [24]. The purpose of this research is to improve the performance of PCAC by adding SHMP, so that it can better act as a binder in the castable. Based on Figure 3 and that discussed above, the optimal addition of SHMP for achieving the best dispersion effect was around 0.2%. The excessive addition of SHMP beyond 0.2% would not make further improvement in the dispersibility of PCAC. On the contrary, due to the increase in the concentration of Na+, it will ultimately affect the refractoriness and other high-temperature properties of castable products.

Figure 5 demonstrates the relationship between shear stress and shear rate in the cases of cement-water slurries added with different amounts of SHMP [33]. According to the rheological theory, all the slurries belonged to non-Newtonian fluids. As can be seen from Figure 5, the shear stress generally increased with increasing the shear rate. In the case of without SHMP, the shear stress decreased rapidly as the shear rate increased from 0.1 to 16 s−1, which was probably caused by the destruction of the flocculated structures due to the rotor rotation. Furthermore, at a given shear rate, the shear stress generally decreased with increasing the addition amount of SHMP, suggesting that SHMP was very effective in r reducing/avoiding the agglomeration of fine particles and the generation of flocculation structure.

Figure 6 further displays viscosity values of cement-water slurries added with different amounts of SHMP, versus shear rates. The viscosity in general decreased with increasing the shear rate. Furthermore, at a given shear rate, it decreased with increasing the addition amount of SHMP, showing the shear-thinning behavior in most cases (except for in the case of addition of 0.4% SHMP, where the viscosity, changed very little with increasing the shear rate). The decrease in viscosity could be attributed to the destruction of the flocculated structures. The results shown in Figure 5 and Figure 6 indicated that the addition of appropriate amounts of SHMP (around 0.2% in this work) could make a significant improvement in the rheological properties of cement-water systems, which is believed to be beneficial to the improvements in rheological properties and workability of a PCAC-bonded castable.

As is well documented, the hydration of calcium aluminate cement involves three main steps: dissolution, nucleation and setting. Upon combination with water, Ca2+ and Al(OH)−44− are released from the cement. When their concentrations reach a critical level, hydration products will start to nucleate and then precipitate [34,35]. These hydration steps are closely related to temperature, CAH10 is the main hydration products at temperatures less than 20 °C and C2AH8 and AH3 are the main hydration products above 20 °C; however, at a temperature above 35 °C (the test temperature in this work was at about 37 °C), CA and CA2 in the cement react with water according to Reactions (2) and (3) [35,36,37].

However, these hydration reactions were significantly affected by the addition of SHMP, as will be described and discussed below.

Figure 7 illustrates the effect of SHMP addition on the hydration extent of PCAC (after 1 d hydration), revealing that 3CaO·Al2O3·6H2O (C3AH6) was formed as the main hydration product, and its diffraction peaks in the sample without SHMP (Figure 5a) were higher than in the samples added with different amounts of SHMP (Figure 7b–e). In the latter, Al4(PO4)3(OH)3·9H2O (Vantasselite) was also detected [38], and it decreased with increasing the addition amount of SHMP. The results of XRD can only judge the type of substance, not the specific quantity, and the integrated area of the peak line can represent the relative content. In several samples, except for the different concentration of SHMP, other conditions are the same. We compared the peak-line integrated areas of C3AH6, CA, CA2 and other substances in different samples to determine the relative change of their contents in different samples. With increasing the addition amount of SHMP, C3AH6 decreased whereas both CA and CA2 increased. Figure 8 further gives SEM images of the samples after 1 d hydration. Granular C3AH6 was formed in the hydrated microstructure, but its amount decreased after increasing the addition amount of SHMP, especially when the addition was ≥0.1%. Combination of the results shown in Figure 7 and Figure 8 reveals that the addition of SHMP inhibited the hydration processes of CA and CA2. This can be explained as follows:

As mentioned earlier in this paper, [Ca2(PO3)6]2− ions formed by the chelation of SHMP with Ca2+ were adsorbed onto the surfaces of cement particles, which reduced the contact area between the cement particles and water, leading to the significant delay in the hydration of CA and CA2. Moreover, the nucleation and precipitation of hydration product requires sufficiently high concentrations of Ca2+ and Al(OH)−44− [14]. However, this was not the case when SHMP was added. As discussed above, the Ca2+ concentration was very low when SHMP was present. Therefore, the nucleation and precipitation of hydration products would be inhibited. For the above reasons, when SHMP was used as a dispersing agent for calcium aluminate cement, its addition amount should be carefully controlled. Otherwise, its excessive addition could lead to an extended hydration time of calcium aluminate cement, and significant degradation in the high-temperature properties.