Inﬂuence of Polyvinyl Alcohol Powder on the Mechanical Performance and Volume Stability of Sulfoaluminate–Portland Cement Composite

: Cement quick repair materials generally have the defects of brittleness and early shrinkage. The modiﬁcation of composite cement mortar by water-soluble glue powder—polyvinyl alcohol (PVA)—was studied in the condition of settled water cement ratio. At 90 days, the maximum ﬂexural strength of mortar was 15.4 MPa, which was 1.3 times of the group without PVA and there was no strength shrinkage phenomenon. Through the analysis of the porosity, PVA improved the total porosity; high porosity could play a better role of buffering and absorbing the impact stress, effectively improving the toughness of mortar. At 28 days, the maximum impact resistance was 1.72 J/cm 2 , which was 1.75 times of the group without PVA. In addition, the increased total porosity also greatly absorbed the shrinkage stress at the early stage, and reduced the mortar self-shrinkage. When the PVA content was 1%, the volume self-shrinkage of mortar decreased from the initial 3360 µ m · m − 1 to 700 µ m · m − 1 . According to the analysis of hydration heat, the addition of PVA effectively reduced the early hydration heat release rate and alleviated the concentrated heat release phenomenon to a certain extent. In this paper, the hydration reaction of composite cement and the action mechanism of PVA in composite cement were studied by means of X-ray diffraction, infrared spectrum and hydration heat analysis.


Introduction
Concrete is the most widely used construction material in the world. Compared with other kinds of repair materials, cement-based repair materials have a great advantage in cost. In addition, high strength and durability are also important reasons for the wide application of cement-based repair materials, but the lack of toughness is also a major defect of cement-based materials. Although the durability of cement concrete was better, during the long-term service the cement concrete material was inevitably appearing aged, damaged and had other issues. The cost of rebuilding, especially under the condition of damage was not serious, but if a deaf ear to damage location was likely to develop into more serious damage, this would affect the normal use of concrete material [1]. Therefore, it was necessary to repair the damaged cement concrete in time.
Conventional Portland cement was widely used to repair the concrete structures. As a result of the relatively slow hydration process, the repaired concrete structures, such as bridges, pavements, etc., needed to wait for a long period of 14 days or even 28 days to reopen. This is not acceptable for some urgent situations. Therefore, rapid setting and rapid strength development are important properties for the repair materials. If conventional Portland cement was used for repair, the hydration was slow, the maintenance time was 5-7 days, the early strength was low, the requirements for construction equipment and maintenance after repair were high, and the time consuming and laborious effect polyvinyl alcohol powder (PVA) was produced in Anhui WanWei huashan new materials company. Chemical composition and mineral composition of PC and SAC were tested by XRF and XRD as shown in Table 1 and Figure 1. It was observed that the main minerals of PC were alite(C 3 S), belite(C 2 S), and calcite, and SAC were Ye'elimite and gypsum. Portland cement (PC.II52.5) was produced in Jiang nan xiao ye tian company, sulfoaluminate cement (R·42.5) was produced in Yi cheng Anda special cement company (SAC), polyvinyl alcohol powder (PVA) was produced in Anhui WanWei huashan new materials company. Chemical composition and mineral composition of PC and SAC were tested by XRF and XRD as shown in Table 1 and Figure 1. It was observed that the main minerals of PC were alite(C3S), belite(C2S), and calcite, and SAC were Ye'elimite and gypsum.

Preparation of Specimens
The composite cement contained PC and SAC, as shown in Table 2, SAC was replaced by 30% of PC; water-to-cement ratio (w/c) was chosen as 0.23 (ensure the mortar to achieve self-leveling), cement-to-sand ratio was 1/1 (the standard sand consistent with ISO 679); content of PVA was 0-2% of cement. Preparation of mortars was in accordance with Chinese standard GB/T 175-2007, mortars were casted into molds and cured in the standard 20 ± 1 °C. After demolded, mortars were cured in air condition and temperature was 20 ± 1 °C, relative humidity was 50%.
Preparation and curing of pastes were the same as mortars. At the specific curing age, pastes were crushed into small pieces and then were immersed into ethanol to stop hydration. After being dried in a vacuum at 50 °C for 3 days, small pieces of pastes were used for the test of SEM and MIP. Furthermore, small pieces were also grinded to pass through 80 μm sieve for the test of hydrates, such as XRD and TG.

Preparation of Specimens
The composite cement contained PC and SAC, as shown in Table 2, SAC was replaced by 30% of PC; water-to-cement ratio (w/c) was chosen as 0.23 (ensure the mortar to achieve self-leveling), cement-to-sand ratio was 1/1 (the standard sand consistent with ISO 679); content of PVA was 0-2% of cement. Preparation of mortars was in accordance with Chinese standard GB/T 175-2007, mortars were casted into molds and cured in the standard 20 ± 1 • C. After demolded, mortars were cured in air condition and temperature was 20 ± 1 • C, relative humidity was 50%. Preparation and curing of pastes were the same as mortars. At the specific curing age, pastes were crushed into small pieces and then were immersed into ethanol to stop hydration. After being dried in a vacuum at 50 • C for 3 days, small pieces of pastes were used for the test of SEM and MIP. Furthermore, small pieces were also grinded to pass through 80 µm sieve for the test of hydrates, such as XRD and TG. According to the strength test of GB/T 17671, mortars with the dimension of 40 × 40 × 160 mm 3 were used for the measurement of compressive strength at the age of 4 h, 1 day, 28 days, 90 days. Compressive strength and flexural strength was tested by compressive machine (AEC-201), the displacement speed was set up to (2400 ± 200) N/S and (50 ± 10) N/S.

Volumetric Stability
A German Micro-Epsilon displacement sensor was used to measure the early volumetric shrinkage (or expansion) of mortar. Thread pipe length was 42 cm, outer diameter was 28 mm, and inner diameter was 20 mm. The mortar was added into the threaded pipe and fully vibrated to avoid the pipe cavity, the threaded pipe was sealed and placed in the inner diameter of 60 mm pipe, solidifying one end to ensure the linear contraction (expansion) of the threaded pipe. During the test, it was strictly forbidden to touch the threaded pipe and laser equipment, and the ambient temperature should be controlled at 20 ± 1 • C.
The calculation of mortar volume deformation is shown in Equation (1).
where ∆l is the shrinkage of threaded tube (µm), L is the length of threaded pipe (m), ω is the shrinkage of mortar in microns per meter (µm/m).

Impact Resistance
Impact resistance was a way to evaluate the toughness of materials. Mortars with the dimension of 300 × 300 × 20 mm 3 were used for the measurement of impact resistance at the age of 1 day and 28 days.
According to the Chinese standard the impact resistance test method for ceramics was used to test the impact resistance of mortar (GB/T 38494-2020). When the iron ball falls on the surface of mortar and produces visible cracks with the origin of the drop point, the impact resistance energy was regarded as the initial impact resistance energy of mortar. The greater the impact resistance, the better the toughness of mortar. The second impact of the falling ball on mortar must be avoided in the test process.
The calculation of impact strength is shown in Equations (2) and (3).
where S is the impact strength of mortar(J/cm 2 ), I is the impact energy (J), t is the thickness at the point of impact (cm), m is the quality of the iron ball (kg), g is the gravity (m/s 2 ), H is the distance between iron ball and mortar plate (m).

Hydration Heat
Hydration heat of pastes was tested by an isothermal calorimeter at 20 • C. The proportion of pastes was the same as in Table 2. The data were automatically recorded by the apparatus for 3 days.

Hydrates Assemblages
(1) X-ray diffraction (XRD): The phase composition of the hydration products was analyzed by X-ray diffractometer. The speed was 10 • /min, and the scanning range was 5~80 • . The equipment was manufactured by Rigaku company of Japan with a power of 3 kw.
(2) Thermogravimetric (TG): The hydration products were analyzed with a comprehensive thermal analyzer. In nitrogen atmosphere, the temperature rose from room temperature to 900 • C at a heating rate of 20 • C/min. The equipment was manufactured by Netzsch company of Germany and the model is STA449C. (2) Surface fractal characteristics: Emission scanning electron microscopy (SEM) was used to observe the microstructure of the samples at hydration ages of 4 h and 28 days. The equipment was manufactured by JEOL company of Japan and the model is S-5900.

Flexural and Compressive Properties and Flexural Compressive Ratio
The flexural strength and compressive strength of composite cement were tested to 90 days, and the results are shown in Figure 2a However, the change in the compressive strength of mortar was just opposite the change in the flexural strength, which was the same as the research results of many scholars. The compressive strength decreases with the increase in the gathering ratio [26][27][28], because the strength of the polymer was less than the strength of the cement [29]. At 4 h, the compressive strength of P2 was decreased from 56.99 MPa to 45.56 MPa, compared with D0, which decreased by 20.1%. At 28 days, the compressive strength of P2 was decreased from 90.86 MPa to 83.59 MPa, compared with D0, which decreased by 8%. At 90 days, the compressive strength of P2 was decreased from 108.25 MPa to 92.72 MPa, compared with the group D0, which decreased by 14.3%. The flexural strength was increased with the increase in the PVA content, and the compressive strength with the increase in the PVA content was lower, which made the flexural compression ratio of mortar increase with the PVA content, as shown in Figure 3. The higher the PVA content, the greater the mortar flexural compression ratio; the greater the mortar flexural compression ratio could reflect the change in the mortar toughness, and the greater the flexural compression ratio of mortar, the better the toughness [29,30]. This was enough to regard that PVA powder could effectively improve the toughness of composite cement. 2021, 11, x FOR PEER REVIEW 6 of 18 mortar, the better the toughness [29,30]. This was enough to regard that PVA powder could effectively improve the toughness of composite cement.

Volume Deformation and Fluidity
Because of the rapid hydration rate of cement quick repair material, resulting in large early shrinkage, this study was aimed at testing the early autogenous shrinkage of com-

Volume Deformation and Fluidity
Because of the rapid hydration rate of cement quick repair material, resulting in large early shrinkage, this study was aimed at testing the early autogenous shrinkage of composite cement. Before the mortar reached the initial setting state, the fluidity was large and the plasticity was strong. Therefore, the volume deformation before the initial setting belonged to the plastic deformation and would not cause the cracking of mortar.
Shown in Figure 4, when the mortar reached the initial setting we began counting. The main component of composite cement was SAC, and its hydration product AFt had an expansion effect, besides the rapid hydration of composite cement, which produced a large amount of hydration heat. So, before the emergence of a large expansion process there was a quick shrinkage. The expansion part contains both the thermal expansion of mortar and hydration products, the expansion of the threaded pipe cooling to room temperature in the process, the hydration heat of thermal expansion gradually disappearing, and, finally, the only expansion effect was the hydration of composite cement. Therefore, all of the curves in Figure 4 show, first, the contraction, then expansion, then contraction, and finally stability. As shown in Figure 4, the addition of PVA effectively improved the expansion value of the composite cement. With the addition of PVA increasing from 0 to 2%, the shrinkage values of mortar were 3360 µm·m −1 , 2800 µm·m −1 , 700 µm·m −1 , 1020 µm·m −1 , and 1170 µm·m −1 . It implied that the addition of PVA could effectively reduce the self-shrinkage of mortar, and the self-shrinkage value was the minimum when the content of PVA was 1%.

Impact Resistance
There were the following three ways to evaluate the toughness of materials: lateral deformation, flexural compression ratio, and impact resistance. Lateral deformation was applicable to materials with high toughness, flexural compression ratio was applicable to materials with high brittleness, and impact resistance was applicable to both high toughness materials and brittleness materials [31].
The initial impact resistance of composite cement at 1 day and 28 days was studied, as shown in Figure 5. The addition of PVA could greatly improve the impact resistance of mortar. At 1 day, the impact resistances of the D0 and experimental groups were 0.49 J/cm 2 , 1.10 J/cm 2 , 0.98 J/cm 2 , 1.35 J/cm 2 , 1.23 J/cm 2 , respectively, and the impact resistances of each of the experimental groups were 200%, 224%, 251% and 276%, respectively, of that in D0. The addition of PVA could effectively improve the toughness of the mortar, and it worked at the early age. When curing to 28 days, the impact resistance of each group increased with the increase in mortar strength, and the impact resistance of P2 was increased by 75.5%, compared with that of D0. The law of the impact resistance curve of mortar corresponded to the flexural compression ratio of the corresponding age. PVA had good toughness and good solubility in water; after mixing with cement materials, it could form the three dimensional space structure, which cement and PVA package of each other

Impact Resistance
There were the following three ways to evaluate the toughness of materials: lateral deformation, flexural compression ratio, and impact resistance. Lateral deformation was applicable to materials with high toughness, flexural compression ratio was applicable to materials with high brittleness, and impact resistance was applicable to both high toughness materials and brittleness materials [31].
The initial impact resistance of composite cement at 1 day and 28 days was studied, as shown in Figure 5. The addition of PVA could greatly improve the impact resistance of mortar. At 1 day, the impact resistances of the D0 and experimental groups were 0.49 J/cm 2 , 1.10 J/cm 2 , 0.98 J/cm 2 , 1.35 J/cm 2 , 1.23 J/cm 2 , respectively, and the impact resistances of each of the experimental groups were 200%, 224%, 251% and 276%, respectively, of that in D0. The addition of PVA could effectively improve the toughness of the mortar, and it worked at the early age. When curing to 28 days, the impact resistance of each group increased with the increase in mortar strength, and the impact resistance of P2 was increased by 75.5%, compared with that of D0. The law of the impact resistance curve of mortar corresponded to the flexural compression ratio of the corresponding age. PVA had good toughness and good solubility in water; after mixing with cement materials, it could form the three dimensional space structure, which cement and PVA package of each other [14,24,25].

Hydration Heat of Composite Cement
The early hydration process of composite cement could be explained by the determination of hydration heat. The main body of composite cement in this study was SAC, so the hydration process of the composite cement was dominated by SAC [10]. Firstly, SAC had the hydration reaction to produce ettringite and hydrated alumina hydroxide, while giving of a large amount of heat. On the one hand, the increase in temperature promotes the hydration of more anhydrous calcium sulphoaluminate, and C3A and C3S; on the other hand, the ettringite generated first also acted as a nucleating seed to further promote the hydration of cement particles [35,36]. Then, the hydration product CH of C3S reacted with hydrated alumina hydroxide in the case of excess gypsum to further generate ettringite [9,37], and finally formed a mutually filled structure with ettringite as skeleton and CSH and cement particles. Some polymer powder contains a carboxyl group, which is a kind of highly active functional group. It could chelate with Ca 2+ to generate calcium carboxylate; cement and chelate were connected by chemical bonds so as to improve the toughness of cement [32][33][34]. The essence of the chelation reaction was an acid-based neutralization reaction. The hydroxyl group in PVA also has certain reactivity. However, if PVA reacted with Ca(OH) 2 to form alcohol calcium, the basic chemical principle of strong base to weak base would be violated. Therefore, PVA would not react in composite cement. From Figure 6, the difference in the infrared absorption peaks between unhydrated D0 and D0 was 2800-4800 cm −1 , which was the absorption peak of -OH, while the main product of D0 was AFt, which contains bound water. The infrared absorption peaks of PVA, unhydrated P2, and P2 were also different in the range of 2800 to 4800 cm −1 . It could be concluded, from the above results, that PVA improved the toughness of mortar only by physical action.

Hydration Heat of Composite Cement
The early hydration process of composite cement could be explained by the determination of hydration heat. The main body of composite cement in this study was SAC, so the hydration process of the composite cement was dominated by SAC [10]. Firstly, SAC had the hydration reaction to produce ettringite and hydrated alumina hydroxide, while giving of a large amount of heat. On the one hand, the increase in temperature promotes the hydration of more anhydrous calcium sulphoaluminate, and C 3 A and C 3 S; on the other hand, the ettringite generated first also acted as a nucleating seed to further promote the hydration of cement particles [35,36]. Then, the hydration product CH of C 3 S reacted with hydrated alumina hydroxide in the case of excess gypsum to further generate ettringite [9,37], and finally formed a mutually filled structure with ettringite as skeleton and CSH and cement particles.

Hydration Heat of Composite Cement
The early hydration process of composite cement could be explained by the determination of hydration heat. The main body of composite cement in this study was SAC, so the hydration process of the composite cement was dominated by SAC [10]. Firstly, SAC had the hydration reaction to produce ettringite and hydrated alumina hydroxide, while giving of a large amount of heat. On the one hand, the increase in temperature promotes the hydration of more anhydrous calcium sulphoaluminate, and C3A and C3S; on the other hand, the ettringite generated first also acted as a nucleating seed to further promote the hydration of cement particles [35,36]. Then, the hydration product CH of C3S reacted with hydrated alumina hydroxide in the case of excess gypsum to further generate ettringite [9,37], and finally formed a mutually filled structure with ettringite as skeleton and CSH and cement particles.
The hydration heat release rate and total heat release of composite cement are shown in Figure 7 and Figure 8. As shown in Figure 7, the heat release of hydration was mainly The hydration heat release rate and total heat release of composite cement are shown in Figures 7 and 8. As shown in Figure 7, the heat release of hydration was mainly concentrated in the early stage, the heat release rate was closed to 0 after 5 h, and there was no heat release peak. Composite cement early hydration was mainly composed of two exothermic peaks. The first exothermic peak was explained by cement particles and exothermic water solubility, and there was nearly no difference between the experimental groups and the D0 group, all of them were finished in a very short time. The second heat peak was sharp, mainly explained by the hydration of the C 3 S and calcium sulfoaluminate cement, and the hydration rate of the D0 group was far greater than the other experimental groups, which was due to the PVA that encapsulated the cement particles to delay the hydration of the cement [24,25]. As shown in Figure 8, due to the retarding effect of PVA, the total amount of heat in the D0 group before 3.2 h was slightly greater than that in the experimental group, while the total amount of heat in all components in 1d would be flat, indicating that the PVA would not have too much adverse effect on the amount of heat release. concentrated in the early stage, the heat release rate was closed to 0 after 5 h, and there was no heat release peak. Composite cement early hydration was mainly composed of two exothermic peaks. The first exothermic peak was explained by cement particles and exothermic water solubility, and there was nearly no difference between the experimental groups and the D0 group, all of them were finished in a very short time. The second heat peak was sharp, mainly explained by the hydration of the C3S and calcium sulfoaluminate cement, and the hydration rate of the D0 group was far greater than the other experimental groups, which was due to the PVA that encapsulated the cement particles to delay the hydration of the cement [24,25]. As shown in Figure 8, due to the retarding effect of PVA, the total amount of heat in the D0 group before 3.2 h was slightly greater than that in the experimental group, while the total amount of heat in all components in 1d would be flat, indicating that the PVA would not have too much adverse effect on the amount of heat release.  The hydration heat release rate and total heat release of composite cement are shown in Figure 7 and Figure 8. As shown in Figure 7, the heat release of hydration was mainly concentrated in the early stage, the heat release rate was closed to 0 after 5 h, and there was no heat release peak. Composite cement early hydration was mainly composed of two exothermic peaks. The first exothermic peak was explained by cement particles and exothermic water solubility, and there was nearly no difference between the experimental groups and the D0 group, all of them were finished in a very short time. The second heat peak was sharp, mainly explained by the hydration of the C3S and calcium sulfoaluminate cement, and the hydration rate of the D0 group was far greater than the other experimental groups, which was due to the PVA that encapsulated the cement particles to delay the hydration of the cement [24,25]. As shown in Figure 8, due to the retarding effect of PVA, the total amount of heat in the D0 group before 3.2 h was slightly greater than that in the experimental group, while the total amount of heat in all components in 1d would be flat, indicating that the PVA would not have too much adverse effect on the amount of heat release.

Hydrates Analysis
(1) XRD analysis: XRD was an effective method to analyze hydration products. Figure 9 shows an XRD diagram of composite cement at 4 h, the detected substances were ettringite and Ye'elimite, gypsum, and calcium silicate, which have not yet been hydrated. At 28 days, there was still existing gypsum, in Figure 10, which indicated that gypsum was excessive. Therefore, it could be considered that the hydration products of the composite cement were basically AFt, which provided strength for the mortar. In the case of excess gypsum, the hydration reaction of composite cement was shown as Equation (4) Firstly, Ye'elimite started to hydrate to form AFt and aluminum glue, and then calcium silicate reacted with hydration to form CSH gel and calcium hydroxide. Because gypsum was excessive, calcium hydroxide, gypsum and aluminum glue reacted to generate more AFt. From Figure 9 and Figure 10, these were not the diffraction peaks of

Hydrates Analysis
(1) XRD analysis: XRD was an effective method to analyze hydration products. Figure 9 shows an XRD diagram of composite cement at 4 h, the detected substances were ettringite and Ye'elimite, gypsum, and calcium silicate, which have not yet been hydrated. At 28 days, there was still existing gypsum, in Figure 10, which indicated that gypsum was excessive. Therefore, it could be considered that the hydration products of the composite cement were basically AFt, which provided strength for the mortar. In the case of excess gypsum, the hydration reaction of composite cement was shown as Equations (4)-(6) [9,37]. XRD was an effective method to analyze hydration products. Figure 9 shows an XRD diagram of composite cement at 4 h, the detected substances were ettringite and Ye'elimite, gypsum, and calcium silicate, which have not yet been hydrated. At 28 days, there was still existing gypsum, in Figure 10, which indicated that gypsum was excessive. Therefore, it could be considered that the hydration products of the composite cement were basically AFt, which provided strength for the mortar. In the case of excess gypsum, the hydration reaction of composite cement was shown as Equation (4), Equation (5) and Equation (6) [9,37].
Firstly, Ye'elimite started to hydrate to form AFt and aluminum glue, and then calcium silicate reacted with hydration to form CSH gel and calcium hydroxide. Because gypsum was excessive, calcium hydroxide, gypsum and aluminum glue reacted to generate more AFt. From Figure 9 and Figure 10, these were not the diffraction peaks of calcium hydroxide, indicating the secondary reaction of all calcium hydroxide generated by hydration. By the way, the AFt generated first became the skeleton attached to the CSH gel, and the two cement particles and their hydration products filled each other, making the mortar denser and obtain a higher strength [9,37].  (2) TG-DSC analysis: TG-DSC was used to analyze the decomposition process of hydration products. According to Figure 11, these had two obvious endothermic peaks, which belong to the hydration products of D0, P1 and P2. From 80 °C to 160 °C was the endothermic peak of AFt losing bound water. The endothermic peak of 250 °C was mainly Al(OH)3 (gel) decomposition. The endothermic peak of 700 °C was mainly caused by the decomposition of calcium carbonate, and around the temperature of 400 °C there were no apparent endothermic peaks; combined with Figure 9, it could be inferred that there was no calcium hydroxide in the hydration products. According to Figure 11 and Figure 12, the total amount of ettringite generated by each component at different ages could be calculated [38], as shown in Table 3. It could be inferred that PVA had no effect on the production of AFt.  Firstly, Ye'elimite started to hydrate to form AFt and aluminum glue, and then calcium silicate reacted with hydration to form CSH gel and calcium hydroxide. Because gypsum was excessive, calcium hydroxide, gypsum and aluminum glue reacted to generate more AFt. From Figures 9 and 10, these were not the diffraction peaks of calcium hydroxide, indicating the secondary reaction of all calcium hydroxide generated by hydration. By the way, the AFt generated first became the skeleton attached to the CSH gel, and the two cement particles and their hydration products filled each other, making the mortar denser and obtain a higher strength [9,37].
(2) TG-DSC analysis: TG-DSC was used to analyze the decomposition process of hydration products. According to Figure 11, these had two obvious endothermic peaks, which belong to the hydration products of D0, P1 and P2. From 80 • C to 160 • C was the endothermic peak of AFt losing bound water. The endothermic peak of 250 • C was mainly Al(OH) 3 (gel) decomposition. The endothermic peak of 700 • C was mainly caused by the decomposition of calcium carbonate, and around the temperature of 400 • C there were no apparent endothermic peaks; combined with Figure 9, it could be inferred that there was no calcium hydroxide in the hydration products. According to Figures 11 and 12, the total amount of ettringite generated by each component at different ages could be calculated [38], as shown in Table 3. It could be inferred that PVA had no effect on the production of AFt. decomposition. The endothermic peak of 700 °C was mainly caused by the decomposition of calcium carbonate, and around the temperature of 400 °C there were no apparent endothermic peaks; combined with Figure 9, it could be inferred that there was no calcium hydroxide in the hydration products. According to Figure 11 and Figure 12, the total amount of ettringite generated by each component at different ages could be calculated [38], as shown in Table 3. It could be inferred that PVA had no effect on the production of AFt. Figure 11. DSC curves of composite cement in N2 atmosphere.   Figure 11. DSC curves of composite cement in N 2 atmosphere.
decomposition. The endothermic peak of 700 °C was mainly caused by the decomposition of calcium carbonate, and around the temperature of 400 °C there were no apparent endothermic peaks; combined with Figure 9, it could be inferred that there was no calcium hydroxide in the hydration products. According to Figure 11 and Figure 12, the total amount of ettringite generated by each component at different ages could be calculated [38], as shown in Table 3. It could be inferred that PVA had no effect on the production of AFt. Figure 11. DSC curves of composite cement in N2 atmosphere.   (3) MIP analysis: Porosity and the distribution of pore size was an effective method to characterize the evolution of the microstructure. Figure 13a,b, are pore distributions of each sample at 4 h and 28 days. The total porosity of the sample was P2 > P1 > D0, because the PVA had the effect on air intake. The greater the amount of mixing, the greater the amount of air intake. A pore size larger than 100 µm would have a negative effect on the strength of the mortar. The relationship of harmful pore content in the sample was P1 > D0, P2 > D0. The addition of PVA increased the proportion of harmful holes in the mortar, and the compressive strength of the mortar was lower than D0. With the growth of curing age, the total porosity of each sample decreases, while the compressive strength of mortar increases. (3) SEM Analysis: The microstructure of D0 and P2 are shown in Figure 14a,b, respectively. As shown in Figure 14a,b, the two groups of samples were closely stacked, the needle and the CSH gel were interwoven-they provide the strength for the cement. However, in Figure 14b, the membrane structure of the polymer was not found, because the hydration products and the polymer films formed a staggered spatial structure. The microstructure of D0 and P2 are shown in Figure 15a,b, respectively, after corrosion by hydrochloric acid. As shown in Figure 15a, AFt was distributed everywhere and the density of structure was decreased. In Figure 15b, the addition of PVA greatly changed the microstructure of the cement. AFt closely grew together and, after the corrosion of hydrochloric acid, the surface was uneven, which was greatly different from Figure 15a. (3) SEM Analysis: The microstructure of D0 and P2 are shown in Figure 14a,b, respectively. As shown in Figure 14a,b, the two groups of samples were closely stacked, the needle and the CSH gel were interwoven-they provide the strength for the cement. However, in Figure 14b, the membrane structure of the polymer was not found, because the hydration products and the polymer films formed a staggered spatial structure. The microstructure of D0 and P2 are shown in Figure 15a,b, respectively, after corrosion by hydrochloric acid. As shown in Figure 15a, AFt was distributed everywhere and the density of structure was decreased. In Figure 15b, the addition of PVA greatly changed the microstructure of the cement. AFt closely grew together and, after the corrosion of hydrochloric acid, the surface was uneven, which was greatly different from Figure 15a

Discussion
According to the research results, the addition of PVA could effectively improve the flexural resistance of mortar. At 90 days, 2% PVA content increased the flexural strength of mortar by 31.7%, and this improvement effect was the most significant. However, PVA had the opposite effect on the compressive properties of mortar. Therefore, the flexural compression ratio of mortar at any age was increased with the increase in PVA. The flexural compression ratio could represent the toughness of mortar to some extent [29][30][31], while the impact resistance of mortar could better reflect the toughness of mortar. The impact strength of mortar was increased with the increase in PVA. At 1d, the impact strength of mortar with 2% PVA content was 2.7 times that of the group without PVA. At 28 days, the impact strength of mortar with 2% PVA content was 1.7 times that of the group without PVA, which shows a very significant improvement effect. Due to the physical action of PVA, the toughness of mortar was improved. In physics, the amount of PVA in mortar internal filtration film, wrapped with cement particles, and the hydration products, form the mutual crisscross in the three-dimensional mesh structure [14,24,25]. The toughness of the polymer was much higher than the cement material, thus enhancing the toughness of the mortar.
According to the change in the TG-DSC curve to calculate the amount of ettringite at 4 h and 28 days, as shown in Table 3, there were no significant differences among the total ettringite. The generation of the ettringite was not affected by PVA, so presumably the

Discussion
According to the research results, the addition of PVA could effectively improve the flexural resistance of mortar. At 90 days, 2% PVA content increased the flexural strength of mortar by 31.7%, and this improvement effect was the most significant. However, PVA had the opposite effect on the compressive properties of mortar. Therefore, the flexural compression ratio of mortar at any age was increased with the increase in PVA. The flexural compression ratio could represent the toughness of mortar to some extent [29][30][31], while the impact resistance of mortar could better reflect the toughness of mortar. The impact strength of mortar was increased with the increase in PVA. At 1d, the impact strength of mortar with 2% PVA content was 2.7 times that of the group without PVA. At 28 days, the impact strength of mortar with 2% PVA content was 1.7 times that of the group without PVA, which shows a very significant improvement effect. Due to the physical action of PVA, the toughness of mortar was improved. In physics, the amount of PVA in mortar internal filtration film, wrapped with cement particles, and the hydration products, form the mutual crisscross in the three-dimensional mesh structure [14,24,25]. The toughness of the polymer was much higher than the cement material, thus enhancing the toughness of the mortar.
According to the change in the TG-DSC curve to calculate the amount of ettringite at 4 h and 28 days, as shown in Table 3, there were no significant differences among the total ettringite. The generation of the ettringite was not affected by PVA, so presumably the

Discussion
According to the research results, the addition of PVA could effectively improve the flexural resistance of mortar. At 90 days, 2% PVA content increased the flexural strength of mortar by 31.7%, and this improvement effect was the most significant. However, PVA had the opposite effect on the compressive properties of mortar. Therefore, the flexural compression ratio of mortar at any age was increased with the increase in PVA. The flexural compression ratio could represent the toughness of mortar to some extent [29][30][31], while the impact resistance of mortar could better reflect the toughness of mortar. The impact strength of mortar was increased with the increase in PVA. At 1d, the impact strength of mortar with 2% PVA content was 2.7 times that of the group without PVA. At 28 days, the impact strength of mortar with 2% PVA content was 1.7 times that of the group without PVA, which shows a very significant improvement effect. Due to the physical action of PVA, the toughness of mortar was improved. In physics, the amount of PVA in mortar internal filtration film, wrapped with cement particles, and the hydration products, form the mutual crisscross in the three-dimensional mesh structure [14,24,25]. The toughness of the polymer was much higher than the cement material, thus enhancing the toughness of the mortar.
According to the change in the TG-DSC curve to calculate the amount of ettringite at 4 h and 28 days, as shown in Table 3, there were no significant differences among the total ettringite. The generation of the ettringite was not affected by PVA, so presumably the volume expansion value caused by ettringite was nearly equal, and PVA mixed and greatly reduced the volume of mortar autogenous shrinkage, especially P1; its autogenous shrinkage was reduced by 4.8 times than D0. Through the calculation of MIP data, the total porosity and pore size distribution data of cement were obtained, as shown in Table 4. The thickening of PVA and air suction effect increased the gas content of mortar and made it difficult for bubbles to overflow, so finally improved the total porosity of mortar. Of course, the increase in harmful pores had a negative effect on the strength of the mortar, but the increase in total porosity also absorbed a part of the shrinkage stress of mortar [39,40] and reduced the volume shrinkage of mortar. As shown in Figure 7, the PVA of cement particles parcel role effectively delayed the mortar early hydration reaction. After the initial setting, the hydration heat release rate of the group without PVA increased rapidly, while the peak time was nearly the same as the other experimental groups, but the heat release rate was greater than the other experimental groups, which also led to the rapid shrinkage.

Conclusions
(1) At 90 days, the flexural strength of mortar with 2% PVA content was 15.4 MPa, which was increased by 31.7%, compared with that without PVA; (2) PVA powder can significantly improve the toughness of mortar. At 90 days, the flexural compression ratio of mortar with 2% PVA content was 0.166, which was 1.5 times that of the group without PVA. In Figure 6, the impact strength of mortar with 2% PVA content at 28 days was 1.72 J/cm 2 , which was 1.75 times that of the group without PVA. The modification of cement by PVA was only by physical, without a chemical reaction; (3) PVA can significantly reduce the volume shrinkage of mortar. The self-shrinkage of mortar with 1% PVA content was 700 µm·m −1 , which was 4.8 times lower than the group without PVA.

Data Availability Statement:
The data that supports the findings in this study are available from the corresponding author upon reasonable request.