1. Introduction
Self-leveling mortar (SLM) used as a floor material consist of cements, fine aggregates, mineral fillers, and chemical admixtures. When mixed with water, it can be leveled with high fluidity or a little auxiliary paving. Owing to the advantages of high fluidity, simple construction, good flatness, high compressive strength, and thin leveling layer, it is an ideal ground material for constructions in large supermarkets, shopping malls, parking lots, factory workshops, and warehouses [
1]. Conventional SLM usually uses Portland cement (PC) as a cementitious material. However, due to the low early strength and serious drying shrinkage of PC [
2], cracks and curls commonly appear at the corners and perimeter. Therefore, it is difficult for conventional SLM to meet the requirements of early strength and crack resistance if only PC is used. In order to achieve better performance, SLM with binary or ternary cementitious systems came into being. Binary or ternary cementitious systems refer to composites of PC, calcium aluminate cement (CAC) or sulfoaluminate cement (SAC), and calcium sulfate (CS).
For SLM with a binary cementitious system, a series of studies were carried out. Zhang et al. [
3] studied the influence of CAC on the performance of SLM (binary cementitious system: PC–CAC), and obtained the optimum dosage of CAC. By studying the properties of SLM (PC–SAC), Georgin et al. [
2], Pera et al. [
4], and Wang et al. [
5] found that SAC could reduce the dry shrinkage of SLM, as well as improve the flexural and compressive strength of mortar. The properties of sulfoaluminate cement-based SLM with CS were studied by Li et al. [
6,
7], who obtained that the formation of ettringite in the early term mainly contributes to strength development, while the formation of ettringite at a later hydration stage favors a decrease in late-term drying shrinkage.
With the development of research on SLM, scholars also carried out many researches on SLM with ternary cementitious systems [
8,
9,
10,
11,
12,
13,
14,
15]. SLM composed of CAC, ordinary PC (OPC), and CS had better hardening characteristics, faster drying, and better dimensional stability [
8,
9]. Zhang et al. [
10] studied the effect of CS type and dosage on the properties of SLM. The test results revealed that, compared to other CS, SLM prepared with anhydrite exhibited the highest initial flow value and setting time and the lowest fluidity loss. The dihydrate was favorable for the early strength. Mortar containing α-hemihydrate exhibited more expansion. The shrinkage compensating effect became more obvious with the increase in CS in SLM samples. According to research on the effect of curing temperature on the hydration of CAC–PC–CS blends [
11,
12], it was found that, compared to anhydrite, hemihydrate is more suitable to be applied in conditions of high temperature (40 °C). The study by David et al. [
13] found that the reaction mechanism of a ternary cementitious system (PC–CAC–CS) depends on the CAC and CS amount in the binders. Laure et al. [
14] studied the hydration mechanism and properties of mortar with a ternary system (OPC–CAC–CS). The results showed CSA clinker was mainly responsible for the early mechanical properties, while OPC played an important role at the later stages. Through the study of mechanical properties of SLM composed of a OPC–SAC–PG (phosphogypsum) ternary cementitious system [
15], it was found that the increase in compressive and flexural strength of mortar was not affected by PG, but the shrinkage, bond strength, and wear resistance were greatly affected.
It can be seen that cementitious materials are the important influencing factors for the performance of SLM. On the other hand, as an important component of SLM, the chemical admixtures (water reducer, retarder, early strength agent, and stabilizer) in the mortar also affect the performance of SLM [
8,
16,
17,
18,
19,
20,
21]. For example, the retarder can prolong the setting time and increase the fluidity of SLM. With an increase in retarder content, the strength of SLM increases first and then decreases.
RP is prepared by spray-drying or other process of polymer emulsion. When the powder mixes with water, the properties of the polymer emulsion are similar to those of the original emulsion [
22,
23]. As an important chemical admixture, although the amount of RP is very small, it has a great influence on the performance of mortar. Do [
24] found that the polymer modified PC-based SLM can be compared with the traditional floor facing material of thermosetting resin in practical applications. Sun et al. [
25] studied the influence of different types of RP on the performance of SLM. The results showed that RP can significantly improve the fluidity and wear resistance of self-leveling mortar. On the other hand, RP reduces the compressive strength of mortar, whereas it improves the flexural strength, the bond strength, the impact resistance, and the flexibility of mortar [
26,
27,
28,
29,
30,
31]. In addition, Schulze et al. [
23] also studied the long-term performance of the RP-modified PC mortar and concluded that, after 10 years, the microstructure of the polymer in the mortar still did not change, maintaining a stable bond, flexural, and compressive strength.
Therefore, RP is an indispensable additive component for excellent mortar as a floor material. However, if the dosage of RP is not in the appropriate range, it will increase the cost and be harmful to the performance of mortar [
16]. Therefore, it is valuable to conduct studies on the effect of its dosage on the performance of SLM. The existing researches primarily focused on the performance of the RP in SLM with binary systems. Few researches dealt with the influence of RP on the performance of SLM with ternary cementitious systems (OPC–SAC–CS). As the mortar is a complex system composed of a variety of materials, the requirements of material type and dosage are very strict. If the material type or dosage is slightly different, the performance will vary significantly.
Although there were many researches on the different kinds of SLM, the researches on SLM with ternary cementitious systems composed of OPC–SAC–CS are still limited, especially the effect of additives on its performance. As an important additive that affects the properties of materials, in order to obtain a better SLM with ternary cementitious system, it is necessary to study the effect of RP on the properties of mortar and determine a reasonable dosage. The current study experimentally investigated the influence of RP on the performance of SLM with a ternary cementitious system (OPC–SAC–CS). The investigation was conducted by testing the fluidity, bond strength, shrinkage rate, abrasion resistance, flexural strength, and compressive strength. How to build a complete and unified mortar performance evaluation method on the basis of the properties and data measured in the test is the key to determining the reasonable RP dosage. This paper introduced a method to determine the reasonable dosage of RP through an analytic hierarchy process (AHP). In this method, a hierarchy structure model for performance evaluation of mortar based on tests was first established, then the weighted values of fluidity, mechanical properties, and other evaluation indices were determined, and finally the comprehensive performance evaluation indices of mortar were calculated to determine the reasonable dosage of RP.
3. Results and Discussion
3.1. Fluidity
The initial fluidity and 20 min fluidity of SLM with 0%, 0.5%, 1.0%, 1.5%, 2.0%, 3.0%, and 4.0% RP were tested, as shown in
Table 4.
Table 4 shows that, in comparison to control formulation (R0), the formulations prepared with different dosages of RP showed different initial fluidity and 20 min fluidity. The initial fluidity and 20 min fluidity of SLM changed with the increase in the dosage of RP. The results showed a trend of gradually decreasing first, then increasing, and finally decreasing rapidly. When the dosage was less than 1.5%, the initial fluidity and 20 min fluidity of SLM gradually decreased with the increase in the dosage of RP. Moreover, when the dosage of RP was 1.5%, the fluidity was the lowest. The initial fluidity and 20 min fluidity of SLM with 1.5% RP were 129 mm and 125 mm, which were 9.2% and 8.8% lower than those of R0. This shows that the addition of RP reduced the fluidity of SLM within a certain range.
The fluidity increased gradually as the RP dosage increased from 1.5% to 3.0%. The fluidity of SLM with 3.0% RP was the highest. The initial fluidity and the 20 min fluidity were 143 mm and 139 mm. Compared with the control formulation, the improvements were 0.7% and 1.5%, respectively. The initial fluidity and 20 min fluidity of SLM with 4.0% RP were 9.1% and 10.1% lower than those of SLM with 3.0% RP. This indicated that, when the dosage exceeded a certain range, the fluidity was significantly reduced. Therefore, the dosage of RP should not be too high.
There are two main reasons why the fluidity of mortar changed with the dosage of redispersible powder. Firstly, when the RP is dissolved in water, some of the water needs to be consumed. Moreover, the viscosity of the RP after dissolving increases the cohesion of the slurry, which leads to a decrease in fluidity. On the other hand, due to the high viscosity of RP, the viscosity of SLM increases when RP is added to the mortar. The bubbles introduced in the mixing process and the gas generated in the cement mixing process cannot be discharged. The slurry density on the surface of mortar decreases, the adsorption capacity increases, and the surface area of cement particles increases, which further leads to the increase in slurry viscosity and the decrease in slurry fluidity. Secondly, RP accumulates on the surface of hydrated and unhydrated products, which reduces the internal friction of mortar. In addition, the surface active substances contained in the mortar produce a large number of tiny bubbles, which can be used as tiny balls when they move inside the mortar. Because of the sliding friction between the original cement particles, bubbles play a lubricating role between the particles, the sliding friction resistance in the mortar is greatly reduced, and the fluidity of the mortar increases.
The fluidity of mortar decreased gradually first, then increased, and finally decreased rapidly with the increase in the dosage of RP, due to a superposition of the above two effects. When the dosage was low, the thickening and viscosity of RP were dominant. With the increase in dosage, the fluidity decreased gradually. However, with the increase in dosage, the lubricating effect gradually exceeded that of the viscosity, and the fluidity of mortar gradually increased. However, the phenomenon of agglomeration appeared when too much RP was added, resulting in a rapid decline in fluidity.
3.2. Bond Strength
The bond strength results of SLM prepared with different dosages of RP are shown in
Figure 3.
As can be seen from
Figure 3, with the increase in the dosage of RP, the bond strength of SLM generally increased gradually. The bond strength of SLM without RP was 1.5 MPa. When the dosage of RP was 4.0%, the bond strength was the largest (2.5 MPa, 66.7% higher than that of SLM without RP).
The reason for the increase in bond strength is that RP is the powder obtained from the spray-drying of a polymer emulsion. The addition of RP promotes the formation of organic polymers and significantly improves the bond strength of mortar by forming films at the mortar interface and in the pores inside. On the other hand, the increase may have been because the flocculating network structure formed by the RP during hydration increased the toughness inside the mortar, thus improving the adhesive performance of the mortar. In this study, no image at the microscopic level was obtained; thus, the above reasons are presumed. However, the study of Qiao et al. [
33] could be used to support this presumption, where it was observed that polymers formed a continuous film between the sand and cement hydration products under a scanning electron microscope, which strengthened the bonding between sand and cement slurry, which was an important factor in improving the bonding strength of mortar.
3.3. Shrinkage Rate
The shrinkage rate of SLM prepared with different dosages of RP can be seen in
Figure 4.
As can be seen from
Figure 4, with the increase in the dosage of RP, the shrinkage rate of mortar significantly decreased. The shrinkage rate decreased gradually from 0.12% to 0.05% as the RP dosage increased from 0% to 4.0%. The results show that the RP had an obvious effect on reducing the shrinkage of SLM.
There are two possible reasons for the decrease in shrinkage rate of SLM. Firstly, under the action of polymer active groups, the water consumption required for cement hydration is greatly reduced, and the water utilization rate required for hydration is increased to some extent. After the excess free water evaporates in the environment, the probability of pore retention also decreases. Secondly, the polymer film formed by RP fills the large gap in cement slurry and forms an interwoven membrane network. This kind of spatial network structure also limits the shrinkage of mortar to a certain extent. The studies of Qiao et al. [
33] and Shi et al. [
34] could be used as support for the above reasons. Under a scanning electron microscope, Qiao et al. and Shi et al. observed that the polymers adhered to the surface of the cement gel or particle and formed a film, filling the pores, thus improving the density of the mortar.
Compared with the samples without RP, when the dosage of RP was 1.0%, 2.0%, 3.0%, and 4.0%, the shrinkage rate decreased by 25%, 41%, 50%, and 58%, respectively. It can be seen that, with the increase in the dosage of RP, the reduction effect of shrinkage rate increased continuously. However, the rate of decrease gradually lessened. For example, the rate of decrease was 25% as the RP dosage increased from 0% to 1.0%, whereas it was 8% as the RP dosage increased from 3.0% to 4.0%. This shows that, with the increase in the dosage of RP, the shrinkage rate of SLM gradually tended to be stable.
3.4. Abrasion Resistance
In order to clearly show the effect of RP on the abrasion resistance of SLM, we expressed the abrasion loss as a percentage. The abrasion loss of mortar without RP was recorded as 100%. The abrasion loss results of SLM prepared with different dosages of RP are shown in
Figure 5.
As can be seen from
Figure 5, with the increase in the dosage of RP, the abrasion loss of SLM first decreased rapidly and then gradually tended to be stable. When the dosage was 3.0%, the abrasion loss of mortar reached the minimum of 28.1%, which was 71.9% lower than that of SLM without RP. The abrasion loss of mortar decreased, which indicates that, within a certain dosage, RP can significantly improve the abrasion resistance of SLM.
The reason for the improvement of the abrasion resistance of SLM is that, after the addition of RP to the SLM, a layer of polymer film is formed on the surface of SLM during the hydration process, resulting in a smooth and hard surface. In the process of abrasion, the polymer film formed by RP retards the development of cracks, reduces the spalling of slurry, and reduces the number of abrasive particles, thus improving the abrasion resistance.
3.5. Flexural Strength and Compressive Strength
The results of the flexural strength and compressive strength tests of mortar prepared with different dosages of RP are shown in
Figure 6. This study tested the 1 day, 3 days, and 28 days flexural strength and compressive strength of SLM. The maximum range of the instrument used in the laboratory to measure the flexural strength was 12.4 MPa, and the flexural strength of the mortar in this experiment after 28 days exceeded this value. Therefore, the specific value of the flexural strength of mortar after 28 days was not measured. Although we did not measure the specific value of the 28 days flexural strength of mortar, it exceeded the specification of 10 MPa [
32] and, thus, met the requirements of the standard.
It can be seen from
Figure 6a that the 1 day and 3 days flexural strength of SLM basically decreased with an increase in the dosage of RP. The results show that the addition of RP reduced the flexural strength of SLM. At the age of 1 day, the flexural strength of SLM without RP was 5.8 MPa. When the dosages were 3.0% and 4.0%, the flexural strength was about 4.8 MPa, which was 18% lower than that of the mortar without RP. At the age of 3 days, the flexural strength of SLM without RP was 10.85 MPa. When the dosage was 4.0%, the mortar had the lowest flexural strength of 8.5 MPa, which was 22% lower than that of the mortar without RP. Furthermore, with the increase in curing age, the flexural strength of all seven groups of mortar with different dosages of RP increased significantly.
The reason for the decrease in flexural strength of mortar is that, because of the addition of RP, the surface adsorption and film formation of polymer particles inhibit the hydration of cement [
35].
It can be observed from
Figure 6b that the compressive strength of SLM at different ages showed different trends with the increase in RP. With the increase in RP, the compressive strength of SLM decreased gradually when the age was 1 day; when the age was 3 days, the compressive strength decreased slightly; when the age was 28 days, the compressive strength was basically greater than the compressive strength of SLM without RP. This indicates that the addition of RP reduced the early compressive strength of SLM and improved the later compressive strength.
At the age of 1 day, the compressive strength of SLM without RP was 26.27 MPa. When the dosage was 4.0%, the compressive strength was 12.85 MPa, which was 51% lower than that of the mortar without RP. This indicates that the RP significantly reduced the 1 day compressive strength of mortar. At the age of 3 days, the compressive strength of SLM without RP was 53.61 MPa. When the dosage was 4.0%, the compressive strength was 46.69 MPa, which was 13% lower than that of the mortar without RP. This shows that the RP had little influence on the 3 days compressive strength of mortar. At the age of 28 days, the compressive strength of the mortar after mixing with RP was slightly increased compared with that without RP, except when the dosage was 2.0%. The 28 days compressive strength of mortar without RP was 74.33 MPa. When the dosage of RP was 3.0%, the 28 days compressive strength of mortar was the highest, 81.15 MPa, which was 8.4% higher than that of the mortar without RP. This shows that RP increased the 28 days compressive strength of mortar, but the increase was relatively small.
RP reduced the early compressive strength of mortar, and the 1 day strength reduction was greater than the 3 days reduction. This may be due to the fact that, after the RP is mixed with mortar, on the one hand, the network structure formed by the polymer film and cement hydration products weakens the hydration process of cement particles [
35] and reduces the formation of gel of cement hydration products, thus reducing the compressive strength. On the other hand, the surfactant in the RP leads to the introduction of a large number of small bubbles in the mortar, resulting in a large number of small pores in the hardened mortar [
36], thus reducing the early strength. The reason for the increase in 28 days compressive strength of mortar may be that, with the formation of hydration products at a later stage and the formation of a flocculent film of RP, the compactness and compressive strength can be improved [
33,
34].
4. Analysis of Reasonable Dosage
The SLM used for ground materials must have good fluidity, strength, stability, cohesiveness, and abrasion resistance to meet the needs of the project. Therefore, in order to determine the reasonable dosage of RP, according to the experimental results, this study took fluidity, compressive strength, flexural strength, stability, cohesiveness, and abrasion resistance as evaluation indices, followed by establishing an evaluation model using the AHP method [
37,
38,
39]. By determining the weighted value of each index to evaluate the performance of mortar, the reasonable dosage of RP was analyzed.
The hierarchical structure model of performance evaluation for SLM is shown in
Figure 7. The plan layer denotes the dosage of RP, i.e., dosage (1), dosage (2), dosage (3), … dosage (
n).
According to the priority and importance of the performance evaluation indices, on the basis of the pairwise comparison of indices, the judgment matrix (A) was generated using Equation (4) according to the point scale (
Table 5) of AHP and combined with engineering experience, as shown in
Table 6. Each entry a
ij in the matrix was built by comparing the row element A
i with the column element A
j [
40].
A1, A2, A3, A4, A5, and A6 in the judgment matrix were used to represent the fluidity, compressive strength, flexural strength, stability, cohesiveness, and abrasion resistance of SLM. For the performance evaluation of mortar, because the evaluation indices were not independent variables and their priority was not absolute, it was difficult to achieve an accurate comparison among indices; thus, the judgment matrix was not unique.
To calculate the weighted value according to the judgment matrix, firstly, the maximum eigenvalue and corresponding eigenvector of matrix A were obtained. The eigenvector was normalized to the weighted value of each index. The maximum eigenvalue
λmax of matrix A was 6.16. The corresponding eigenvector W = (0.422, 0.196, 0.196, 0.089, 0.058, 0.039) was calculated through the AHP. Because the value of judgment matrix was given artificially and had subjectivity, the reliability of the judgment matrix was verified using a consistency check. The consistency index (
CI) was used to measure the consistency among pairwise comparisons according to the following equation:
where
λmax is the eigenvalue and
n is the number of indicators in the pairwise comparison matrix. Therefore, according to Equation (5),
CI was 0.032. The consistency ratio
CR was utilized to judge the consistency of the matrix according to Equation (6).
where
RI is the random index, given in
Table 7 according to Saaty [
41]. Generally, the upper limit of the satisfactory range of
CR is 0.1.
According to Equations (5) and (6) and
Table 7, the following values were obtained:
CI = (6.16–6)/(6–1) = 0.032;
CR =
CI/
RI = 0.032/1.24 = 0.026 < 0.1. Therefore, the value of W obtained above was acceptable; that is, the calculated relative importance was acceptable. Hence, the weighted values of fluidity, compressive strength, flexural strength, stability, cohesiveness, and wear resistance of mortar were 0.422, 0.196, 0.196, 0.089, 0.058, and 0.039 respectively. The sum of the weighted value of these six indices was 1, which means that the performance of the mortar was determined by these six indices.
The experimental results of performance indices of mortar with different dosages of RP are shown in
Table 8. In order to facilitate comparisons of the comprehensive evaluation indices, some original experimental data were simply processed, as shown in
Table 8. The compressive strength and flexural strength were expressed as 1 day test results; stability was expressed by shrinkage rate according to the following equation:
where
refers to the shrinkage rate of SLM when the dosage of RP is
i%. Cohesiveness was expressed by bond strength. Abrasion resistance was expressed by abrasion loss according to the following equation:
where
Fi refers to the abrasion loss of SLM when the dosage of RP is
i%.
The data listed in
Table 8 represent reference values for evaluating the performance of mortar and, thus, should not be neglected. If the dosage of RP was determined only by fluidity, the reasonable dosage would be 3.0%. If only based on the compressive strength, the compressive strength of SLM without RP was the highest. If only in terms of cohesiveness, the reasonable dosage of RP was 4.0%. Obviously, these evaluation results are contradictory and inconsistent. In order to evaluate the performance of mortar more reasonably and determine the reasonable dosage of RP, it is a better choice to consider the weighted value of each index.
The experimental data of six performance evaluation indices were processed in dimensionless form using Equation (9), and then the weighted average value of each evaluation index was calculated according to Equation (10), whereby the weighted average value was regarded as the comprehensive index of mortar performance evaluation, as shown in
Table 9.
where
Pi (
i = 1, 2, 3, 4, 5, 6) is the performance evaluation index,
mi is the dimensionless value of the performance evaluation index,
M is the comprehensive index of performance evaluation, and
ηi is the weighted value of the performance evaluation index. A higher value of the comprehensive index generally denotes the better performance of mortar. Thus, according to the comprehensive indices of SLM in
Table 9, the reasonable dosage of RP was found to be 3.0%.