Mitigating the Drying Shrinkage and Autogenous Shrinkage of Alkali-Activated Slag by NaAlO2

The shrinkage of alkali-activated slag (AAS) is obviously higher than ordinary Portland cement, which limited its application in engineering. In this study, the effects of NaAlO2 in mitigating drying shrinkage and autogenous shrinkage of AAS were studied. To further understand the shrinkage mechanism, the hydration products and microstructures were studied by X-ray diffraction, scanning electron microscopy and nitrogen adsorption approaches. As the partial substitution rate of NaAlO2 for Na2SiO3 increased, the drying shrinkage and autogenous shrinkage reduced significantly. The addition of NaAlO2 could slow down the rate of hydration reaction and reduce the porosity, change the pore diameter and the composition of generated paste and cause more hydrotalcite and tetranatrolite generated—which contributed to reduced shrinkage. Additionally, raising the Na2O content rate caused obvious differences in drying shrinkage and autogenous shrinkage. As the Na2O content elevated, the drying shrinkage decreased and autogenous shrinkage increased. A high Na2O content would cause complete hydration reactions and provoke high autogenous shrinkage. However, incomplete hydration reactions left more water in the paste, and the evaporated water dramatically influenced drying shrinkage. The results indicate that addition of NaAlO2 could greatly mitigate the drying shrinkage and autogenous shrinkage of AAS.


Introduction
Alkali-activated slag (AAS)-replaced ordinary Portland cement as a new binding material could be considered as possibility [1]. Realized by compounded blast furnace slag and alkaline activator was obtained the binding materials which is similar to cement. AAS could effectively reduce carbon dioxide emissions and energy consumption, while also effectively used industrial solid waste [2,3]. The AAS had good performance in early compressive strength, high temperature resistance and chemical corrosion resistance [4,5]. However, unsolved problems were existed, such as higher drying shrinkage and autogenous shrinkage [6,7].
Previous research has demonstrated that the shrinkage rate of AAS was significantly higher than that of ordinary Portland cement. The study by Lee et al. [8] provided that the AAS autogenous shrinkage and drying shrinkage at 28 d was about four and six times to that of ordinary Portland cement, respectively. Atiş et al. [9] reported that the AAS mortar drying shrinkage prepared with Na 2 SiO 3 as an alkaline activator was three to six times to that of ordinary Portland cement mortar. Drying shrinkage of cementitious materials has already been confirmed that mainly due to the capillary stress, disjoining pressure and surface tension. Cartwright et al. [10] believed that the capillary stress was caused by the smaller pore diameter, which led to the AAS shrinkage. In addition, the C-A-S-H in the AAS hydration product was considered to be the main reason for the large shrinkage. The alkali metal cations in the reaction solution entered the C-A-S-H gel, changed structures and led to collapse and redistributed upon drying [11,12]. Four different NaAlO 2 substitution rates (for 20%, 40%, 60% and 80%), five different Na 2 O contents (for 3%, 5%, 7%, 9% and 11%, which substitution rate fixed 20%) and a non-substituted control group were tested. The prepared mix proportions of AAS are shown in Table 2. For example, the code "NA0.2NS0. 8-5" represents the alkaline activator with NaAlO 2 substitution rates of 20% and Na 2 SiO 3 Materials 2020, 13, 3499 3 of 13 content of 80% and Na 2 O contents 5%. In order to reduce the influence of dosage to the test error, the slag was fixed to 2000 g and the water-slag ratio was kept constant at 0.35.

Drying Shrinkage
The drying shrinkage test was on the basis of JC/T 603-2004 [20]. All prepared paste were poured into prismatic molds (h = 25 mm, w = 25 mm, l = 280 mm). The sample was cured and measured in the curing room (20 ± 2 • C and RH > 50%). The test start time was recorded from demolding done and the measurement accuracy was 0.001 mm. The drying shrinkage was calculated according to Equation (1). Three samples were prepared for each code and the drying shrinkage was the mean value of three samples and recorded once per day until 28 d.
S n was the drying shrinkage rate of the test at the age of n d and the unit was %. L 0 was the initial reading and the unit was mm. L n was the measurement of the test at the age of n d and the unit was mm. The effective length of the fixed value and the unit was 280 mm.

Autogenous Shrinkage
An autogenous shrinkage test was under the standard ASTM C1697-09 [21]. This solution could effectively avoid moisture loss and minimize the constraints on volume changes. All prepared pastes were poured into a corrugated tube of 420 mm (Ø28.5 mm). The loaded samples were placed horizontally on a corrugated plastic board to avoid length changes and damage. The samples were cured and measured in the curing room (20 ± 2 • C and RH > 50%). The start time of the test was recorded from the time of adding water to mix. The measurement accuracy was 0.001 mm. The autogenous shrinkage was calculated according to Equation (2). Three samples were prepared for each code, and the drying shrinkage was the mean value of three samples and recorded once per day until 28 d.
µ n was the autogenous shrinkage rate of the test at the age of n d, and the unit was µε. L n was the measurement of the test at the age of n d, and the unit was mm. L 0 was the initial reading and the unit was mm.

Instrumental Techniques
In order to better understand the shrinkage performance of AAS with different mix ratios, scanning electron microscopy (SEM, JEOL-6360LV, Tokyo, Japan), X-ray diffraction (XRD, Empyrean-X, Amsterdam, The Netherlands) and Brunauer-Emmett-Teller (BET, Micromeritics ASAP 2020 instrument, Pullman, WA, USA) were used to analyze microstructures, hydration products and pore structures. The specimens from the crushed samples were poured into alcohol for 24 h to discontinue hydration. The specimens were removed from the alcohol and dried in a 60-• C oven for use before testing. Figure 1 plots the effects of different NaAlO 2 substitution rates on drying shrinkage. Drying shrinkage mainly occurred in the first 7 d and tended to increase steadily and slowly after 7 d. At 7 d, the drying shrinkage of the NaAlO 2 substitution rate of 20%, 40%, 60% and 80% was 93%, 84%, 64% and 54% of control group (NS 1.0-5), respectively. The substitution of NaAlO 2 for Na 2 SiO 3 had a significant effect on mitigate shrinkage. Substitution rate of NaAlO 2 has a significantly negative correlation with drying shrinkage. The main reason was that the rate of drying shrinkage was clearly related to the hydration reaction degree of the paste [22]. Na 2 SiO 3 had stronger alkali activating ability. Neto et al. [23] noted that more Na 2 SiO 3 content in the alkaline activator caused clear high drying shrinkage. Meanwhile, Ben et al. [17] reported that more soluble Al in the paste could delay the alkali-activated hydration reaction, which could contribute to mitigate drying shrinkage. Empyrean-X, Amsterdam, The Netherlands) and Brunauer-Emmett-Teller (BET, Micromeritics ASAP 2020 instrument, Pullman, WA, USA) were used to analyze microstructures, hydration products and pore structures. The specimens from the crushed samples were poured into alcohol for 24 h to discontinue hydration. The specimens were removed from the alcohol and dried in a 60-°C oven for use before testing. Figure 1 plots the effects of different NaAlO2 substitution rates on drying shrinkage. Drying shrinkage mainly occurred in the first 7 d and tended to increase steadily and slowly after 7 d. At 7 d, the drying shrinkage of the NaAlO2 substitution rate of 20%, 40%, 60% and 80% was 93%, 84%, 64% and 54% of control group (NS 1.0-5), respectively. The substitution of NaAlO2 for Na2SiO3 had a significant effect on mitigate shrinkage. Substitution rate of NaAlO2 has a significantly negative correlation with drying shrinkage. The main reason was that the rate of drying shrinkage was clearly related to the hydration reaction degree of the paste [22]. Na2SiO3 had stronger alkali activating ability. Neto et al. [23] noted that more Na2SiO3 content in the alkaline activator caused clear high drying shrinkage. Meanwhile, Ben et al. [17] reported that more soluble Al in the paste could delay the alkali-activated hydration reaction, which could contribute to mitigate drying shrinkage.  Figure 2 shows the effects of different Na2O contents on drying shrinkage at fixed 20% NaAlO2 substitution rate. The drying shrinkage was also obvious during the first 7 d. At 7 d, Na2O content of 5%, 7%, 9% and 11%, the drying shrinkage was 83%, 73%, 56% and 30% of Na2O content of 3%, respectively. The Na2O content had a clear effect on drying shrinkage: a high Na2O content led to low drying shrinkage. As result of low Na2O content with low OHions concentration, the paste could not complete hydration reaction and residual moisture evaporation led to high drying shrinkage [24]. Meng Wu et al. [25] also suggested that amorphous gels lost moisture caused greater drying  Figure 2 shows the effects of different Na 2 O contents on drying shrinkage at fixed 20% NaAlO 2 substitution rate. The drying shrinkage was also obvious during the first 7 d. At 7 d, Na 2 O content of 5%, 7%, 9% and 11%, the drying shrinkage was 83%, 73%, 56% and 30% of Na 2 O content of 3%, respectively. The Na 2 O content had a clear effect on drying shrinkage: a high Na 2 O content led to low drying shrinkage. As result of low Na 2 O content with low OHions concentration, the paste could not complete hydration reaction and residual moisture evaporation led to high drying shrinkage [24]. Meng Wu et al. [25] also suggested that amorphous gels lost moisture caused greater drying shrinkage. In general, the mechanism of AAS drying shrinkage was complicated and depended on different NaAlO 2 substitution rates and Na 2 O contents.

Drying Shrinkage
Materials 2020, 13, x FOR PEER REVIEW 5 of 13 shrinkage. In general, the mechanism of AAS drying shrinkage was complicated and depended on different NaAlO2 substitution rates and Na2O contents.  Figure 3 shows the effects of different NaAlO2 substitution rates on autogenous shrinkage. Used bellows to test autogenous shrinkage could effectively prevent the loss of water and minimize the constraints on volume changes. Autogenous shrinkage had similar characteristics to drying shrinkage and great influence to mitigate autogenous shrinkage by the NaAlO2 substitution rate increased and also was obvious shrank during the first 7 d. Moreover, the autogenous shrinkage of alkali-activated slag used alkaline activator with Na2SiO3 alone was still very large, same to findings from by Kumarappa et al. [26]. The sample with substitution rate at 80%, the drying shrinkage at 7 d and 28 d were 73% and 72% of the control group (NS 1.0-5), respectively. Figure 4 plots the effects of different Na2O contents on autogenous shrinkage at fixed 20% NaAlO2 substitution rates. By elevating the Na2O content, the autogenous shrinkage increased, consistent with the Ma et al. [27] found. However, the results were contrary to the drying shrinkage test results. The autogenous shrinkage happened without the exchange of external substances and the water from paste was not migrated. Hence, high Na2O content promoted completed the hydration reaction of the paste, which in turn caused greater autogenous shrinkage [28]. The maximum autogenous shrinkage group (NA0.2NS0.8-11) was 1.4 times and 1.5 times that of the minimum autogenous shrinkage group (NA0.2NS0.8-3) at 7 d and 28 d, respectively. Then, the phenomenon of autogenous shrinkage with different Na2O contents would be consistent with the degree of hydration reaction. Neto et al. [23] also believed that drying shrinkage due to the water evaporation from the paste was more significant than autogenous shrinking.  Figure 3 shows the effects of different NaAlO 2 substitution rates on autogenous shrinkage. Used bellows to test autogenous shrinkage could effectively prevent the loss of water and minimize the constraints on volume changes. Autogenous shrinkage had similar characteristics to drying shrinkage and great influence to mitigate autogenous shrinkage by the NaAlO 2 substitution rate increased and also was obvious shrank during the first 7 d. Moreover, the autogenous shrinkage of alkali-activated slag used alkaline activator with Na 2 SiO 3 alone was still very large, same to findings from by Kumarappa et al. [26]. The sample with substitution rate at 80%, the drying shrinkage at 7 d and 28 d were 73% and 72% of the control group (NS 1.0-5), respectively. Figure 4 plots the effects of different Na 2 O contents on autogenous shrinkage at fixed 20% NaAlO 2 substitution rates. By elevating the Na 2 O content, the autogenous shrinkage increased, consistent with the Ma et al. [27] found. However, the results were contrary to the drying shrinkage test results. The autogenous shrinkage happened without the exchange of external substances and the water from paste was not migrated. Hence, high Na 2 O content promoted completed the hydration reaction of the paste, which in turn caused greater autogenous shrinkage [28]. The maximum autogenous shrinkage group (NA0.2NS0.8-11) was 1.4 times and 1.5 times that of the minimum autogenous shrinkage group (NA0.2NS0.8-3) at 7 d and 28 d, respectively. Then, the phenomenon of autogenous shrinkage with different Na 2 O contents would be consistent with the degree of hydration reaction. Neto et al. [23] also believed that drying shrinkage due to the water evaporation from the paste was more significant than autogenous shrinking. Materials 2020, 13, x FOR PEER REVIEW 6 of 13

XRD
The composition of the hydration products was analyzed by XRD. The generation of different hydration products had significant impacts on the pore structure and shrinkage properties. By checking the results of drying shrinkage and autogenous shrinkage, we found that drying shrinkage

XRD
The composition of the hydration products was analyzed by XRD. The generation of different hydration products had significant impacts on the pore structure and shrinkage properties. By checking the results of drying shrinkage and autogenous shrinkage, we found that drying shrinkage

XRD
The composition of the hydration products was analyzed by XRD. The generation of different hydration products had significant impacts on the pore structure and shrinkage properties. By checking the results of drying shrinkage and autogenous shrinkage, we found that drying shrinkage and autogenous shrinkage before 7 d were more severe than that of after 7 d. Therefore, the characteristics of hydration products at 7 d could explain the different shrinkage changes. Figure 5 shows the XRD image of raw slag and different NaAlO 2 substitution rates at 7 d. The characteristic phenomenon at approximately 30 • of all AAS was a feature corresponding to big quantity of amorphous products which were C-(A)-S-H gels [29]. Calcite was found in all pastes and was generated by sample carbonization during preparation and curing time [30]. However, hydrotalcite and tetranatrolite were also major products found in this research. Hydrotalcite had micro-expansion in a certain degree; tetranatrolite was a zeolite phase structure, which was beneficial to mitigate paste shrinkage [31]. At the same time, as the substitution rate of NaAlO 2 increased, the amount of hydrotalcite and tetranatrolite also increased, which generated different shrinkage mitigation effects among all the AAS-and was consistent with the results of the drying shrinkage test.
Materials 2020, 13, x FOR PEER REVIEW 7 of 13 and autogenous shrinkage before 7 d were more severe than that of after 7 d. Therefore, the characteristics of hydration products at 7 d could explain the different shrinkage changes. Figure 5 shows the XRD image of raw slag and different NaAlO2 substitution rates at 7 d. The characteristic phenomenon at approximately 30° of all AAS was a feature corresponding to big quantity of amorphous products which were C-(A)-S-H gels [29]. Calcite was found in all pastes and was generated by sample carbonization during preparation and curing time [30]. However, hydrotalcite and tetranatrolite were also major products found in this research. Hydrotalcite had micro-expansion in a certain degree; tetranatrolite was a zeolite phase structure, which was beneficial to mitigate paste shrinkage [31]. At the same time, as the substitution rate of NaAlO2 increased, the amount of hydrotalcite and tetranatrolite also increased, which generated different shrinkage mitigation effects among all the AAS-and was consistent with the results of the drying shrinkage test.

SEM
The microscopic structure of AAS with different NaAlO2 substitution rates at 7 d and 28 d were analyzed by SEM. The progress of hydration reaction was investigated by studying the microscopic morphology. The density of hydration products and the changes of voids could be retrieved directly from the SEM, which contributed to better understanding of the mechanism of hydration. Figure 6 shows the microscopic structure of AAS with the different NaAlO2 substitution rates at 7 d and 28 d. The specimen NS1.0-5 was significantly denser than that of NA0.2NS0.8-5 and NA0.8NS0.2-5. The slag was almost completely dissolved and generated more C-A-S-H gels and less voids at 7 d. The Na2SiO3 was conducive to the progress of the hydration reaction and could make the hydration reaction more complete. This showed that NaAlO2 partially substituted Na2SiO3 as a composite alkaline activator, inhibiting the progress of hydration reaction, delaying and reducing the formation of hydration products of AAS, which supports the conclusions of Ben et al. [31].
The phenomenon of microstructure at 28 d was exactly opposite of that at 7 d. The specimen NA0.8NS0.2-5 generated denser structure (for mainly C-A-S-H gels) at 28 d, and the microstructure was denser than that of NS1.0-5. This indicates that NA0.8NS0.2-5 had a more complete hydration reaction at 28d. Moreover, the NaAlO2 substitution rates difference had significant impact on the formation of 28 d hydration products. This may be due to the addition of NaAlO2 having an inhibitory

SEM
The microscopic structure of AAS with different NaAlO 2 substitution rates at 7 d and 28 d were analyzed by SEM. The progress of hydration reaction was investigated by studying the microscopic morphology. The density of hydration products and the changes of voids could be retrieved directly from the SEM, which contributed to better understanding of the mechanism of hydration. Figure 6 shows the microscopic structure of AAS with the different NaAlO 2 substitution rates at 7 d and 28 d. The specimen NS1.0-5 was significantly denser than that of NA0.2NS0.8-5 and NA0.8NS0.2-5. The slag was almost completely dissolved and generated more C-A-S-H gels and less voids at 7 d. The Na 2 SiO 3 was conducive to the progress of the hydration reaction and could make the hydration reaction more complete. This showed that NaAlO 2 partially substituted Na 2 SiO 3 as a composite alkaline activator, inhibiting the progress of hydration reaction, delaying and reducing the formation of hydration products of AAS, which supports the conclusions of Ben et al. [31].
The phenomenon of microstructure at 28 d was exactly opposite of that at 7 d. The specimen NA0.8NS0.2-5 generated denser structure (for mainly C-A-S-H gels) at 28 d, and the microstructure was denser than that of NS1.0-5. This indicates that NA0.8NS0.2-5 had a more complete hydration reaction at 28 d. Moreover, the NaAlO 2 substitution rates difference had significant impact on the formation of 28 d hydration products. This may be due to the addition of NaAlO 2 having an inhibitory effect on the hydration reaction in the early phase, but a motivating effect on the reaction in the later phase, which supports the research of Huang et al. [32]. The NA0.8NS0.2-5 clearly had completed hydration reaction to generate more gels where micropores mainly existed. Compared with other samples, NA0.8NS0.2-5 had the most amount of gels, which meant a higher micropore-to-volume ratio was excited, which corresponded to the BET results.
Materials 2020, 13, x FOR PEER REVIEW 8 of 13 effect on the hydration reaction in the early phase, but a motivating effect on the reaction in the later phase, which supports the research of Huang et al. [32]. The NA0.8NS0.2-5 clearly had completed hydration reaction to generate more gels where micropores mainly existed. Compared with other samples, NA0.8NS0.2-5 had the most amount of gels, which meant a higher micropore-to-volume ratio was excited, which corresponded to the BET results.

BET
Many studies found that the volume proportions of pore structure have critical impact on the shrinkage properties of AAS [33,34]. The pore size was determined to measure by the Brunauer-Emmett-Teller (BET) theory. The specimens at 28 d were tested to observe the volume proportion of AAS pore structure. The pore sizes were in the range of 1.7-300 nm. According to Zhongwei Wu, a Chinese academic, the pore structure of cement-based materials be classified as micropores (d < 20 nm), mesopores (20 nm < d < 50 nm) and macropores (d > 50 nm). The volume proportions of pore structure distribution of AAS showed in Table 3. The pore-size distributions of AAS with different NaAlO2 substitution rates are shown in Figure  7, mainly concentrated between 1.7-20 nm. Based on Figure 7, a higher NaAlO2 substitution rate resulted to lower total proportion of porosity and higher micropores volume. The reduction of total porosity was beneficial to mitigate drying shrinkage. When the NaAlO2 substitution rates were 20% and 80%, the total porosities were 56% and 51% of the control group (NS1.0-5), respectively, which

BET
Many studies found that the volume proportions of pore structure have critical impact on the shrinkage properties of AAS [33,34]. The pore size was determined to measure by the Brunauer-Emmett-Teller (BET) theory. The specimens at 28 d were tested to observe the volume proportion of AAS pore structure. The pore sizes were in the range of 1.7-300 nm. According to Zhongwei Wu, a Chinese academic, the pore structure of cement-based materials be classified as micropores (d < 20 nm), mesopores (20 nm < d < 50 nm) and macropores (d > 50 nm). The volume proportions of pore structure distribution of AAS showed in Table 3. The pore-size distributions of AAS with different NaAlO 2 substitution rates are shown in Figure 7, mainly concentrated between 1.7-20 nm. Based on Figure 7, a higher NaAlO 2 substitution rate resulted to lower total proportion of porosity and higher micropores volume. The reduction of total porosity was beneficial to mitigate drying shrinkage. When the NaAlO 2 substitution rates were 20% and 80%, the total porosities were 56% and 51% of the control group (NS1.0-5), respectively, which was consistent with the drying shrinkage and SEM results. On the other hand, the adjustment in pore-size distribution also had significant effects on the drying shrinkage. The substitution rate increased with higher micropores volume proportion, which was the reason the drying shrinkage reduced. The highest substitution rate corresponded to the largest micropores volume proportion of 61.67%, which also corresponded with the smallest drying shrinkage. This indicates that the micropore volume proportion had a significant effect on drying shrinkage. Compare this to the discussion by Chen et al. [35], namely, that that pore structure also has significant effect on drying shrinkage: a higher micropore volume proportion can result in a greater drying shrinkage, which was contrary to the phenomenon found in this study. The reason for the difference in Chen's conclusion was that the hydration products had no changes. However, in this study, the addition of NaAlO 2 changed the hydration products. Based on the findings from the XRD results, more hydrotalcite and tetranatrolite were generated. Nevertheless, the micropores only existed in the gels, so the micropores volume proportion increased may correspond to the increased in the amount of hydrotalcite and tetranatrolite [36]. As discussed in this article, both hydrotalcite and tetranatrolite contributed to mitigate drying shrinkage and generated lower drying shrinkage with more micropores volume proportion.
Materials 2020, 13, x FOR PEER REVIEW 9 of 13 was consistent with the drying shrinkage and SEM results. On the other hand, the adjustment in poresize distribution also had significant effects on the drying shrinkage. The substitution rate increased with higher micropores volume proportion, which was the reason the drying shrinkage reduced. The highest substitution rate corresponded to the largest micropores volume proportion of 61.67%, which also corresponded with the smallest drying shrinkage. This indicates that the micropore volume proportion had a significant effect on drying shrinkage. Compare this to the discussion by Chen et al. [35], namely, that that pore structure also has significant effect on drying shrinkage: a higher micropore volume proportion can result in a greater drying shrinkage, which was contrary to the phenomenon found in this study. The reason for the difference in Chen's conclusion was that the hydration products had no changes. However, in this study, the addition of NaAlO2 changed the hydration products. Based on the findings from the XRD results, more hydrotalcite and tetranatrolite were generated. Nevertheless, the micropores only existed in the gels, so the micropores volume proportion increased may correspond to the increased in the amount of hydrotalcite and tetranatrolite [36]. As discussed in this article, both hydrotalcite and tetranatrolite contributed to mitigate drying shrinkage and generated lower drying shrinkage with more micropores volume proportion.
The pore-size distributions of AAS with different Na 2 O contents are shown in Figure 8. As Na 2 O content increased, the total porosity and micropores volume proportion decreased. The content of Na 2 O content increased from 3 to 11%. The micropore volume proportions decreased from 75.92% to 16.24%. A higher Na 2 O content led to more complete reaction, and the paste became denser and the total porosity decreased, which supports the research of Chindaprasirt et al. [37]. At this point, the largest micropores volume proportion corresponded to the maximum drying shrinkage-the same as concluded by Yang et al. [38]. The micropores of diameter 1.25-25 nm were the main reason for AAS drying shrinkage. However, contrary to the conclusions obtained by the different substitution rates above, the major reason was that the NaAlO 2 substitution rate was fixed and changed the Na 2 O content could not change the composition of the generated phase. More micropore volume proportions in the same generated paste led to greater drying shrinkage. The pore-size distributions of AAS with different Na2O contents are shown in Figure 8. As Na2O content increased, the total porosity and micropores volume proportion decreased. The content of Na2O content increased from 3 to 11%. The micropore volume proportions decreased from 75.92% to 16.24%. A higher Na2O content led to more complete reaction, and the paste became denser and the total porosity decreased, which supports the research of Chindaprasirt et al. [37]. At this point, the largest micropores volume proportion corresponded to the maximum drying shrinkage-the same as concluded by Yang et al. [38]. The micropores of diameter 1.25-25 nm were the main reason for AAS drying shrinkage. However, contrary to the conclusions obtained by the different substitution rates above, the major reason was that the NaAlO2 substitution rate was fixed and changed the Na2O content could not change the composition of the generated phase. More micropore volume proportions in the same generated paste led to greater drying shrinkage.
Furthermore, as the Na2O content elevated, the critical pore size became smaller, which corresponded to the lower shrinkage and was consistent with the results of the different NaAlO2 substitution rates.

Conclusions
In this study, the mitigation function of alkali-activated slag-drying shrinkage and autogenous shrinkage mixed with NaAlO2 was studied. The mechanism and reason of mitigating shrinkage were discussed. According to the research in this article, the conclusions are as follows: Furthermore, as the Na 2 O content elevated, the critical pore size became smaller, which corresponded to the lower shrinkage and was consistent with the results of the different NaAlO 2 substitution rates.

Conclusions
In this study, the mitigation function of alkali-activated slag-drying shrinkage and autogenous shrinkage mixed with NaAlO 2 was studied. The mechanism and reason of mitigating shrinkage were discussed. According to the research in this article, the conclusions are as follows: 1.
Through NaAlO 2 addition, the drying shrinkage and autogenous shrinkage could be clearly mitigated. In addition, the solid alkaline activator provided a convenient way of preparation and curing and reduced damage from alkaline corrosion; 2.
The addition of NaAlO 2 slowed the early hydration reaction, while made the later hydration reaction more complete, reducing the total porosity, changing the pore-size distribution and having a significant impact on the shrinkage performance; 3.
Added NaAlO 2 could generate hydrotalcite and tetranatrolite, and the generation increased with increased added NaAlO 2. Moreover, the change of the generated phase composition clearly mitigated shrinkage of AAS; 4.
For AAS with different Na 2 O contents, the evaporation from the paste played a major role in drying shrinkage and the degree of hydration reaction mainly contributed to autogenous shrinkage.