Effects of High-Volume Ground Slag Powder on the Properties of High-Strength Concrete under Different Curing Conditions

Massive high-strength concrete structures tend to have a high risk of cracking. Ground slag powder (GSP), a sustainable and green industrial waste, is suitable for high-strength concrete. We carried out an experimental study of the effects of GSP with a specific surface area of 659 m2/kg on the hydration, pore structure, compressive strength and chloride ion penetrability resistance of high-strength concrete. Results show that adding 25% GSP increases the adiabatic temperature rise of high-strength concrete, whereas adding 45% GSP decreases the initial temperature rise. Incorporating GSP refines the pore structure to the greatest extent and improves the compressive strength and chloride ion penetrability resistance of high-strength concrete, which is more obvious under early temperature-matching curing conditions. Increasing curing temperature has a more obvious impact on the pozzolanic reaction of GSP than cement hydration. From a comprehensive perspective, GSP has potential applications in the cleaner production of green high-strength concrete.


Introduction
Currently, high-rise buildings have become increasingly widespread in China due to their advantages in space, stability and unique design [1,2]. They typically symbolize the landscape architecture and construction of a city, such as the Shanghai Tower (632 m), the Shenzhen Ping'an Finance Centre (599 m) and the China Zun in Beijing (528 m). The foundation structures of high-rise buildings with heavy loads are extremely deep and wide, which is typical for massive high-strength concrete structures [1]. During the hydration process, substantial hydration heat is generated in massive high-strength concrete, resulting in a high internal temperature rise because of its slow heat dissipation [3][4][5][6]. After hardening, large tensile stresses are formed due to restrained thermal and autogenous shrinkage deformations, which are the main driving forces of cracking in concrete [7].
Using supplementary cementitious materials (SCMs) to lower early heat and attendant volume changes is the most common preventative method [8][9][10][11][12]. The application of SCMs in concrete also has positive effects on workability, pumpability, strength and permeability resistance to chemical attacks [13][14][15][16][17][18][19][20][21]. Meanwhile, SCMs, as mineral admixtures to replace cement in high-strength concrete, reduce the carbon footprint in cement and concrete production and are conducive to sustainable development due to the conservation of natural resources [22][23][24]. Slag, as one of the most suitable SCMs, has been extensively identified and used to directly replace cement, minimizing cracking in massive concrete applications. Slag is a non-metallic residual generated from blast furnaces when iron is extracted from its ore [25,26]. Molten slag, comprising mostly silicates and alumina, is swiftly quenched with abundant water [27]. The rapid cooling method results in amorphous phases of slag (nearly  Table 2 shows the mix proportions of the high-strength concrete. The total amount of binder was 550 kg/m 3 . A water/binder ratio of 0.28 and a sand ratio of 0.43 were selected. Plain cement concrete was regarded as the reference sample (sample C). Two substitution rates of GSP (25% and 45% by mass) were used, corresponding to sample S25 and sample S45. Cubic concrete samples with side lengths of 100 mm were prepared. Each sample had a total of 36 square concrete specimens. The paste had the same water/binder ratio and substitution rates as the concrete. Fresh pastes were cast into plastic tubes and cured under the same curing conditions as the concrete.

Curing Conditions and Test Methods
In this study, two curing conditions for the concretes and pastes were set-the standard curing condition (symbol S) and the temperature-matching curing condition (symbol M). Thus, sample SS25 represents S25 concrete cured under the standard curing condition, and sample MS45 represents S45 concrete cured under the temperature-matching curing condition. The standard curing condition required a constant temperature (20 • C ± 2 • C) and relative humidity (>95%). The temperature-matching curing condition needed to be adjusted according to the adiabatic temperature rise curve of the concrete. The adiabatic temperature rise curve of high-strength concrete for the initial 7 d was determined by a temperature measuring instrument (50 L) with an accuracy of ±0.1 • C.
The compressive strength and chloride ion penetrability resistance of concrete for each concrete mixture were obtained from an average of three specimens. The pastes were prepared for the tests of pore structure, Ca(OH) 2 (CH) content, and non-evaporable water content. First, the hardened paste was broken into small pieces (less than 5 mm). Then, the pieces were soaked in ethanol (Tongguang fine chemicals company, Beijing, China) for 24 h at test ages. Finally, all pieces were dried in the oven at 110 • C. For tests of CH content and non-evaporable content water content, dried pieces were further ground into ultrafine powder. The pore structure of hardened paste at ages of 28 d and 90 d was measured with a mercury intrusion porometer (MIP, AUTOPORE II 9220 manufactured by Micromeritics, America) with a maximum mercury intrusion pressure of 300 MPa. The CH content was determined by thermogravimetric (TG) and derivative thermogravimetric (DTG) analyses. TG and DTG curves were obtained using Instrument TGA 3+ (METTLER TOLEDO, Switzerland) in an N 2 atmosphere from 30 • C to 900 • C at 14 d, 28 d and 90 d. The non-evaporable water content W n was determined using the following Equation (1): where m 1 is the dried mass of hardened paste at 110 • C, m 2 is the mass of hardened paste after heating at 1050 • C, α represents the replacement rate of GSP, and LOI C and LOI S represent the loss on ignition of cement and GSP, respectively.

Adiabatic Temperature Rise
The adiabatic temperature rise curves of plain cement concrete and GSP concrete for the initial 7 d are depicted in Figure 1. It is obvious that the growth trends of the three temperature curves are similar. The temperature rose sharply before 24 h and remained stable after 150 h. The final temperature rises of samples C, S25 and S45 were 56.56 • C, 60.58 • C and 54.07 • C, respectively. This illustrates that the incorporation of 25% GSP increases the adiabatic temperature rise by 4.02 • C and incorporating 45% GSP decreases the adiabatic temperature rise by 2.49 • C. Note that sample S25 exhibited the maximum temperature rise. This indicates that the promoting effect of GSP on cement hydration exceeds the negative effect due to cement reduction. The promoting effects derived from two main contributing factors [34]. One is the heterogenous nucleation effect [34]. Compared to cement particles, GSP has finer particles, of which the specific surface area is 659 m 2 /kg. Finer slag particles can serve as heterogeneous nucleation and crystallization sites of C-S-H gel and CH, thus improving the degree of cement hydration. The other is related to the higher reactivity of slag, which participates in pozzolanic reactions at an early stage, thus promoting cement hydration. The drop in the adiabatic temperature rise is attributed to a significant reduction in cement content. 90 d. The non-evaporable water content Wn was determined using the following Equation (1): where m1 is the dried mass of hardened paste at 110 °C, m2 is the mass of hardened paste after heating at 1050 °C, α represents the replacement rate of GSP, and LOIC and LOIS represent the loss on ignition of cement and GSP, respectively.

Adiabatic Temperature Rise
The adiabatic temperature rise curves of plain cement concrete and GSP concrete for the initial 7 d are depicted in Figure 1. It is obvious that the growth trends of the three temperature curves are similar. The temperature rose sharply before 24 h and remained stable after 150 h. The final temperature rises of samples C, S25 and S45 were 56.56 °C, 60.58 °C and 54.07 °C, respectively. This illustrates that the incorporation of 25% GSP increases the adiabatic temperature rise by 4.02 °C and incorporating 45% GSP decreases the adiabatic temperature rise by 2.49 °C. Note that sample S25 exhibited the maximum temperature rise. This indicates that the promoting effect of GSP on cement hydration exceeds the negative effect due to cement reduction. The promoting effects derived from two main contributing factors [34]. One is the heterogenous nucleation effect [34]. Compared to cement particles, GSP has finer particles, of which the specific surface area is 659 m 2 /kg. Finer slag particles can serve as heterogeneous nucleation and crystallization sites of C-S-H gel and CH, thus improving the degree of cement hydration. The other is related to the higher reactivity of slag, which participates in pozzolanic reactions at an early stage, thus promoting cement hydration. The drop in the adiabatic temperature rise is attributed to a significant reduction in cement content.

Compressive Strength
The compressive strengths of plain cement concrete and GSP concrete under standard curing conditions are shown in Figure 2a. It can be easily observed that the compressive strength of concrete slightly increased with GSP at all ages. The compressive strengths of concrete at ages of 7 d, 28 d, 90 d and 180 d were more than 70 MPa, approximately 80 MPa, 85 MPa and 90 MPa, respectively. Compared to the 7 d compressive strength, the growth rates of the strength at the ages of 28 d, 90 d and 180 d were approximately 12%, 20% and 24%, respectively. Thus, the growth rates of the compressive strengths of plain cement concrete and GSP concrete showed little difference at the same ages under standard curing conditions. Compared to the compressive strengths of plain cement concrete, the growth rates in strength due to the addition of GSP were calculated and are presented in Figure 2b. The growth rate of sample SS25 at all ages was relatively low, at no more than 2%. The growth rate of sample SS45 was higher than that of sample SS25 at all ages. In particular, the 28 d growth rate of sample SS45 reached 6%. The growth rate is related to pore structure and hydration products. Compared to sample SS25, sample SS45 has a higher substitution rate and the presence of GSP with finer particles has a more positive influence on early hydration, resulting in a higher growth rate.
sive strength of concrete slightly increased with GSP at all ages. The compressive strengths of concrete at ages of 7 d, 28 d, 90 d and 180 d were more than 70 MPa, approximately 80 MPa, 85 MPa and 90 MPa, respectively. Compared to the 7 d compressive strength, the growth rates of the strength at the ages of 28 d, 90 d and 180 d were approximately 12%, 20% and 24%, respectively. Thus, the growth rates of the compressive strengths of plain cement concrete and GSP concrete showed little difference at the same ages under standard curing conditions. Compared to the compressive strengths of plain cement concrete, the growth rates in strength due to the addition of GSP were calculated and are presented in Figure 2b. The growth rate of sample SS25 at all ages was relatively low, at no more than 2%. The growth rate of sample SS45 was higher than that of sample SS25 at all ages. In particular, the 28 d growth rate of sample SS45 reached 6%. The growth rate is related to pore structure and hydration products. Compared to sample SS25, sample SS45 has a higher substitution rate and the presence of GSP with finer particles has a more positive influence on early hydration, resulting in a higher growth rate. The compressive strength and growth rate of compressive strength under temperature-matching curing conditions are presented in Figure 3a,b, respectively. The compressive strength of high-strength concrete significantly increases with GSP at all ages under temperature-matching curing conditions, which is different from the trend under standard curing conditions. Compared to the 7 d compressive strength, the growth rates of the strength at 28 d and 90 d were approximately 11% and 21%, respectively. However, the growth rates of 180 d compressive strength were approximately 22%, 28% and 30%, respectively. The growth rate of GSP concrete at 180 d was higher than that of plain cement concrete. Meanwhile, compared to the compressive strength of plain cement concrete, the growth rates of strength at different ages due to the addition of GSP under temperaturematching curing conditions (Figure 3b) were higher than those under standard curing temperatures ( Figure 2b). In addition, the growth rate increased with GSP. In particular, the 180 d growth rates were the highest. This result indicates that the temperature-matching curing conditions have a more positive effect on the development of the late compressive strength of GSP concrete. Elevated temperatures promote the pozzolanic reaction of The compressive strength and growth rate of compressive strength under temperaturematching curing conditions are presented in Figure 3a,b, respectively. The compressive strength of high-strength concrete significantly increases with GSP at all ages under temperature-matching curing conditions, which is different from the trend under standard curing conditions. Compared to the 7 d compressive strength, the growth rates of the strength at 28 d and 90 d were approximately 11% and 21%, respectively. However, the growth rates of 180 d compressive strength were approximately 22%, 28% and 30%, respectively. The growth rate of GSP concrete at 180 d was higher than that of plain cement concrete. Meanwhile, compared to the compressive strength of plain cement concrete, the growth rates of strength at different ages due to the addition of GSP under temperaturematching curing conditions (Figure 3b) were higher than those under standard curing temperatures ( Figure 2b). In addition, the growth rate increased with GSP. In particular, the 180 d growth rates were the highest. This result indicates that the temperature-matching curing conditions have a more positive effect on the development of the late compressive strength of GSP concrete. Elevated temperatures promote the pozzolanic reaction of GSP. The pozzolanic reaction of GSP consumes CH and forms C-S-H gel, improving the density of the interfacial transition zone between the cement and aggregates [40,41]. Furthermore, C-S-H gel plays a key role in mechanical performance. When GSP is added to the cementitious system, Al3+ is released from the slag and finally forms a C-(A)-S-H gel, leading to an increase in the Al/Si molar ratio and a decrease in the Ca/Si molar ratio [42,43]. C-S-H with higher Al/Si and lower Ca/Si ratios has a higher bonding capacity and thus improves the compressive strength [40][41][42][43][44]. thermore, C-S-H gel plays a key role in mechanical performance. When GSP is added to the cementitious system, Al3+ is released from the slag and finally forms a C-(A)-S-H gel, leading to an increase in the Al/Si molar ratio and a decrease in the Ca/Si molar ratio [42,43]. C-S-H with higher Al/Si and lower Ca/Si ratios has a higher bonding capacity and thus improves the compressive strength [40][41][42][43][44].

Chloride Ion Penetrability Resistance
The chloride ion penetrability resistance of concrete at ages of 28 d and 180 d under standard curing conditions are shown in Figure 4. It can be seen in Figure 4 that the chloride ion penetrability grades of sample SC were ''moderate" and "low" at 28 d and 180 d, respectively. However, the penetrability grades of samples SS25 and SS45 fell to the "low" level and the "very low" level at the two ages. Therefore, substitution with GSP can improve the chloride ion penetrability resistance of high-strength concrete, and the effect increases with increasing GSP. This is because the filling effect of grinding slag fills the pore structure of concrete, and the pozzolanic reaction consumes CH in the transition zone, resulting in more C-S-H gel, which refines the pore structure.

Chloride Ion Penetrability Resistance
The chloride ion penetrability resistance of concrete at ages of 28 d and 180 d under standard curing conditions are shown in Figure 4. It can be seen in Figure 4 that the chloride ion penetrability grades of sample SC were "moderate" and "low" at 28 d and 180 d, respectively. However, the penetrability grades of samples SS25 and SS45 fell to the "low" level and the "very low" level at the two ages. Therefore, substitution with GSP can improve the chloride ion penetrability resistance of high-strength concrete, and the effect increases with increasing GSP. This is because the filling effect of grinding slag fills the pore structure of concrete, and the pozzolanic reaction consumes CH in the transition zone, resulting in more C-S-H gel, which refines the pore structure.  The chloride ion penetrability resistance of concrete at 28 d and 180 d under temperature-matching curing conditions are presented in Figure 5. Significantly, as the age increases, the chloride ion penetration resistance of concrete did not change. The chloride ion penetrability grades of sample MC were ''moderate" at the two ages. The penetrability grades of both samples MS25 and MS45 fell to the "very low" level at the same time. The GSP content has little effect on the chloride ion penetration resistance of concrete with further hydration. This is because the pozzolanic reaction of GSP mainly occurred at an The chloride ion penetrability resistance of concrete at 28 d and 180 d under temperaturematching curing conditions are presented in Figure 5. Significantly, as the age increases, the chloride ion penetration resistance of concrete did not change. The chloride ion pene-trability grades of sample MC were "moderate" at the two ages. The penetrability grades of both samples MS25 and MS45 fell to the "very low" level at the same time. The GSP content has little effect on the chloride ion penetration resistance of concrete with further hydration. This is because the pozzolanic reaction of GSP mainly occurred at an early age and increasing the early curing temperature promoted the reaction of GSP, which had an adverse effect on the late reaction. In terms of the chloride ion penetration resistance, combined with the results under standard curing conditions, increasing the curing temperature has a greater influence on high-strength concrete mixed with 25% GSP. The chloride ion penetrability resistance of concrete at 28 d and 180 d under temperature-matching curing conditions are presented in Figure 5. Significantly, as the age increases, the chloride ion penetration resistance of concrete did not change. The chloride ion penetrability grades of sample MC were ''moderate" at the two ages. The penetrability grades of both samples MS25 and MS45 fell to the "very low" level at the same time. The GSP content has little effect on the chloride ion penetration resistance of concrete with further hydration. This is because the pozzolanic reaction of GSP mainly occurred at an early age and increasing the early curing temperature promoted the reaction of GSP, which had an adverse effect on the late reaction. In terms of the chloride ion penetration resistance, combined with the results under standard curing conditions, increasing the curing temperature has a greater influence on high-strength concrete mixed with 25% GSP.

Pore Structure
The differential pore size distributions of different hardened pastes under standard curing conditions are presented in Figure 6. As shown in Figure 6, the most likely pore sizes of samples SC, SS25 and SS45 at 28 d were 62.1 nm, 30.2 nm and 13.94 nm, respectively. With further hydration, the pore structure of the hardened paste becomes dense. The most likely pore sizes of samples SC, SS25 and SS45 at 28 d were 40.7 nm, 5.42 nm and 3.39 nm, respectively. The most likely pore size obviously decreased with GSP due to its positive effects on pore structure. Pores in hardened paste can be divided into four types, depending on their diameters and functions, according to Mehta's opinion: >100 nm, 50-100 nm, 4.5-50 nm and <4.5 nm [45,46]. Pores larger than 100 nm (also called harmful pores) have adverse effects on the development of mechanical strength and the permeability of hardened matrices [36]. Thus, the detailed pore size distribution of the hardened paste is presented in Figure 7. Figure 7a shows that the porosities of samples SC, SS25 and SS45 at 28 d were 21.52%, 20.81% and 17.29%, respectively. The porosities at 90 d, shown in Figure 7b, were 16.77%, 16.79% and 19.09%, respectively. Compared to the plain cement paste (sample SC), the total porosity and the volume of harmful pores in hardened paste at 28 d were significantly reduced due to the addition of GSP. However, the total porosity in hardened paste containing 45% GSP (sample SS45) was the largest at 28 d due to more harmless pores (<4.5 nm). in Figure 7b, were 16.77%, 16.79% and 19.09%, respectively. Compared to the plain cement paste (sample SC), the total porosity and the volume of harmful pores in hardened paste at 28 d were significantly reduced due to the addition of GSP. However, the total porosity in hardened paste containing 45% GSP (sample SS45) was the largest at 28 d due to more harmless pores (<4.5 nm).  The differential pore size distributions of different hardened pastes under temperature-matching curing conditions are presented in Figure 8. The most likely pore sizes of samples MC, MS25 and MS45 at 28 d were 24.3 nm, 49.0 nm and 5.4 nm, respectively. With further hydration, the pore structure of the hardened paste becomes denser. The most likely pore sizes of samples MC, MS25 and MS45 at 90 d were 17.1 nm, 5.7 nm and 5.0 nm, respectively. The detailed pore size distribution of the hardened paste is presented in Figure 9. Figure 9a shows that the porosities of samples MC, MS25 and MS45 at 28 d were 24.74%, 17.09% and 16.23%, respectively. The porosities at 90 d, shown in Figure 9b, were 22.20%, 16.35% and 11.31%, respectively. Compared to the plain cement paste (sample MC), the total porosity and the volume of harmful pores in hardened paste containing 25% GSP were significantly reduced. Sample MS25 had lower total porosity than sample MC. However, the volumes of harmful pores of both pastes showed little difference. The detailed pore size distribution of the hardened paste is presented in Figure 9. Figure 9a shows that the porosities of samples MC, MS25 and MS45 at 28 d were 24.74%, 17.09% and 16.23%, respectively. The porosities at 90 d, shown in Figure 9b, were 22.20%, 16.35% and 11.31%, respectively. Compared to the plain cement paste (sample MC), the total porosity and the volume of harmful pores in hardened paste containing 25% GSP were significantly reduced. Sample MS25 had lower total porosity than sample MC. However, the volumes of harmful pores of both pastes showed little difference. ure 9a shows that the porosities of samples MC, MS25 and MS45 at 28 d were 24.74%, 17.09% and 16.23%, respectively. The porosities at 90 d, shown in Figure 9b, were 22.20%, 16.35% and 11.31%, respectively. Compared to the plain cement paste (sample MC), the total porosity and the volume of harmful pores in hardened paste containing 25% GSP were significantly reduced. Sample MS25 had lower total porosity than sample MC. However, the volumes of harmful pores of both pastes showed little difference.

CH Content
The main hydration products of cement after 110 °C are CH and C-S-H gel. TG and DTG analyses can express the mass loss resulting from the decomposition of the hydration products. TG and DTG curves of hardened pastes under standard curing conditions at different ages are presented in Figure 10. It is obvious that the mass loss of sample SC was lower than that of samples SS25 and SS45 at the same ages, especially at early ages (14 d and 28 d). However, the gap between samples SS25 and SS45 was small. As seen from the DTG curves, there are two distinct peaks. One peak represents the dehydration of the C-S-H gel before 200 °C, and the other is associated with the decomposition of CH within a temperature range of 400 °C-500 °C [46,47]. It is worth noting that a smaller peak can be found at approximately 800 °C, especially at 90 d. This peak is related to the decomposition of CaCO3 originating from the carbonation of CH. The total CH content was calculated based on the mass losses, corresponding to the decomposition of CH and CaCO3.

CH Content
The main hydration products of cement after 110 • C are CH and C-S-H gel. TG and DTG analyses can express the mass loss resulting from the decomposition of the hydration products. TG and DTG curves of hardened pastes under standard curing conditions at different ages are presented in Figure 10. It is obvious that the mass loss of sample SC was lower than that of samples SS25 and SS45 at the same ages, especially at early ages (14 d and 28 d). However, the gap between samples SS25 and SS45 was small. As seen from the DTG curves, there are two distinct peaks. One peak represents the dehydration of the C-S-H gel before 200 • C, and the other is associated with the decomposition of CH within a temperature range of 400 • C-500 • C [46,47]. It is worth noting that a smaller peak can be found at approximately 800 • C, especially at 90 d. This peak is related to the decomposition of CaCO 3 originating from the carbonation of CH. The total CH content was calculated based on the mass losses, corresponding to the decomposition of CH and CaCO 3 .
S-H gel before 200 °C, and the other is associated with the decomposition of CH within a temperature range of 400 °C-500 °C [46,47]. It is worth noting that a smaller peak can be found at approximately 800 °C, especially at 90 d. This peak is related to the decomposition of CaCO3 originating from the carbonation of CH. The total CH content was calculated based on the mass losses, corresponding to the decomposition of CH and CaCO3. Therefore, the CH content of hardened paste at different ages under standard curing conditions is shown in Figure 11. It is clear that the CH content declined as the incorporation of GSP increased at early and late ages, which can be primarily attributed to the reduction in the cement content. Furthermore, unlike cement, GSP does not produce CH, but consumes a certain amount of CH due to its pozzolanic reaction. In order to better explain the influence of GSP on the cement system, the CH contents of plain cement systems were calculated using reduction factors of 0.75 (25% GSP) and 0.55 (45% GSP), respectively. The calculation results are also marked in Figure 11. At 14 d, compared to those of the cement after the calculation with reduction factors of 0.75 and 0.55, the CH contents of samples SS25 and SS45 were significantly higher. This indicates that the promoting effects of GSP on the cementitious system exceeded the pozzolanic effect of GSP. These promoting effects on cement hydration can be attributed to the dilution effect and the nucleation effect, which forms a higher effective water/cement ratio and a larger growth space for hydration products. At an early age, hydration of cement in the cementitious system is dominant. At 28 d, the CH contents in the cement paste after calculation and the corresponding hardened paste containing GSP showed little difference. However, with further Therefore, the CH content of hardened paste at different ages under standard curing conditions is shown in Figure 11. It is clear that the CH content declined as the incorporation of GSP increased at early and late ages, which can be primarily attributed to the reduction in the cement content. Furthermore, unlike cement, GSP does not produce CH, but consumes a certain amount of CH due to its pozzolanic reaction. In order to better explain the influence of GSP on the cement system, the CH contents of plain cement systems were calculated using reduction factors of 0.75 (25% GSP) and 0.55 (45% GSP), respectively. The calculation results are also marked in Figure 11. At 14 d, compared to those of the cement after the calculation with reduction factors of 0.75 and 0.55, the CH contents of samples SS25 and SS45 were significantly higher. This indicates that the promoting effects of GSP on the cementitious system exceeded the pozzolanic effect of GSP. These promoting effects on cement hydration can be attributed to the dilution effect and the nucleation effect, which forms a higher effective water/cement ratio and a larger growth space for hydration products. At an early age, hydration of cement in the cementitious system is dominant. At 28 d, the CH contents in the cement paste after calculation and the corresponding hardened paste containing GSP showed little difference. However, with further hydration, the pozzolanic reaction of GSP consumed more CH, leading to a lower CH content in composite systems at 90 d. The TG and DTG curves of hardened pastes at different ages under temperaturematching curing conditions are presented in Figure 12. As shown in Figure 12, the trend in mass losses was obviously different from that observed under standard curing conditions ( Figure 10). The gaps between samples MC, MS25 and MS45 were relatively small at all ages. There are also two endothermic peaks on the DTG curves at approximately 100 °C and 450 °C, corresponding to the sequential decomposition of C-S-H gel and CH crystals. Another absorption peak is located at approximately 650 °C, which represents the decomposition of CaCO3.
(a) (b) Figure 11. CH content of hardened pastes under standard curing conditions.
The TG and DTG curves of hardened pastes at different ages under temperaturematching curing conditions are presented in Figure 12. As shown in Figure 12, the trend in mass losses was obviously different from that observed under standard curing conditions ( Figure 10). The gaps between samples MC, MS25 and MS45 were relatively small at all ages. There are also two endothermic peaks on the DTG curves at approximately 100 • C and 450 • C, corresponding to the sequential decomposition of C-S-H gel and CH crystals. Another absorption peak is located at approximately 650 • C, which represents the decomposition of CaCO 3 .
The CH contents of different hardened pastes, including measured values and normalization marks, under temperature-matching curing conditions are indicated in Figure 13. At 14 d, the CH contents in the cement paste after calculation and the corresponding hardened paste containing GSP showed little difference. This indicates that the CH content remained balanced between production from cement hydration and consumption due to the pozzolanic reaction of GSP. Currently, the promoting effect of GSP on the cementitious system can compensate for its negative effect on CH content due to the pozzolanic effect. At 28 d, compared to those of the cement after calculation with reduction factors of 0.75 and 0.55, the CH contents of samples SS25 and SS45 were significantly lower. At 90 d, the gap in CH contents between the cement paste and hardened paste containing GSP became wider. This indicates that the pozzolanic effect of GSP proves its importance with further hydration. Compared to that observed under standard curing conditions, the gap in the CH content at 14 d was obviously smaller under temperature-matching curing condition. Elevated curing temperature has a greater impact on the pozzolanic reaction of GSP, consuming more CH. tions ( Figure 10). The gaps between samples MC, MS25 and MS45 were relatively small at all ages. There are also two endothermic peaks on the DTG curves at approximately 100 °C and 450 °C, corresponding to the sequential decomposition of C-S-H gel and CH crystals. Another absorption peak is located at approximately 650 °C, which represents the decomposition of CaCO3. The CH contents of different hardened pastes, including measured values and normalization marks, under temperature-matching curing conditions are indicated in Figure  13. At 14 d, the CH contents in the cement paste after calculation and the corresponding hardened paste containing GSP showed little difference. This indicates that the CH content remained balanced between production from cement hydration and consumption due to the pozzolanic reaction of GSP. Currently, the promoting effect of GSP on the cementitious system can compensate for its negative effect on CH content due to the pozzolanic effect. At 28 d, compared to those of the cement after calculation with reduction factors of 0.75 and 0.55, the CH contents of samples SS25 and SS45 were significantly lower. At 90 d, the gap in CH contents between the cement paste and hardened paste containing GSP became wider. This indicates that the pozzolanic effect of GSP proves its importance with further hydration. Compared to that observed under standard curing conditions, the gap in the CH content at 14 d was obviously smaller under temperaturematching curing condition. Elevated curing temperature has a greater impact on the pozzolanic reaction of GSP, consuming more CH.

Non-Evaporable Water Content
The non-evaporable water contents of hardened paste under two curing conditions are presented in Figure 14a,b, respectively. The lowest non-evaporable water content of hardened paste exceeded 12% under both curing conditions. Under standard curing condition, the non-evaporable water content of hardened paste containing GSP was slightly higher than that of sample SC. In particular, hardened paste containing 25% GSP showed the highest non-evaporable water content. This indicates that the promoting effects, including the dilution effect and the nucleation effect of GSP on the cement hydration, were obvious at early and late ages. The results agree with the results of the adiabatic temperature rise of concrete ( Figure 1). In theory, the hydration of Portland cement per gram produces 0.25 g of non-evaporable water, but the pozzolanic reaction of slag per gram generates 0.30 g of non-evaporable water [48]. Meanwhile, increasing the fineness of active powder has a more positive effect on the pozzolanic reaction [48]. Therefore, the faster reaction of GSP results with a higher temperature rise rate was observed for sample SS25 (Figure 1), along with the higher non-evaporable water content of sample SS25. However, the gap in non-evaporable water contents between samples SC and SS45 was relatively small, resulting from a sharp reduction in cement content. Combined with the results of compressive strength under two curing conditions, sample SS45 had the highest compressive strength. The compressive strength of concrete depends not only on the amount of hydration products, it is also closely related to the pore structure. Sample SS45 had the lowest total porosity and volume of harmful pores, which resulted in the highest compressive strength.

Non-Evaporable Water Content
The non-evaporable water contents of hardened paste under two curing conditions are presented in Figure 14a,b, respectively. The lowest non-evaporable water content of hardened paste exceeded 12% under both curing conditions. Under standard curing condition, the non-evaporable water content of hardened paste containing GSP was slightly higher than that of sample SC. In particular, hardened paste containing 25% GSP showed the highest non-evaporable water content. This indicates that the promoting effects, including the dilution effect and the nucleation effect of GSP on the cement hydration, were obvious at early and late ages. The results agree with the results of the adiabatic temperature rise of concrete ( Figure 1). In theory, the hydration of Portland cement per gram produces 0.25 g of non-evaporable water, but the pozzolanic reaction of slag per gram generates 0.30 g of non-evaporable water [48]. Meanwhile, increasing the fineness of active powder has a more positive effect on the pozzolanic reaction [48]. Therefore, the faster reaction of GSP results with a higher temperature rise rate was observed for sample SS25 (Figure 1), along with the higher non-evaporable water content of sample SS25. However, the gap in non-evaporable water contents between samples SC and SS45 was relatively small, resulting from a sharp reduction in cement content. Combined with the results of compressive strength under two curing conditions, sample SS45 had the highest compressive strength. The compressive strength of concrete depends not only on the amount of hydration products, it is also closely related to the pore structure. Sample SS45 had the lowest total porosity and volume of harmful pores, which resulted in the highest compressive strength. Under temperature-matching curing conditions, the trend of the non-evaporable water content was similar to that under standard curing conditions (Figure 14a). However, the non-evaporable water content of hardened paste became higher than that under standard curing conditions. Increasing the early curing temperature evidently improves the hydration rate of the plain cement and composite binder. Meanwhile, the highest nonevaporable water content was observed for sample SS45 at 90 d. Replacing cement with GSP markedly increased the late non-evaporable water content under temperaturematching curing conditions. It is worth noting that the non-evaporable water content of Under temperature-matching curing conditions, the trend of the non-evaporable water content was similar to that under standard curing conditions (Figure 14a). However, the non-evaporable water content of hardened paste became higher than that under standard curing conditions. Increasing the early curing temperature evidently improves the hydra-tion rate of the plain cement and composite binder. Meanwhile, the highest non-evaporable water content was observed for sample SS45 at 90 d. Replacing cement with GSP markedly increased the late non-evaporable water content under temperature-matching curing conditions. It is worth noting that the non-evaporable water content of hardened paste containing GSP under two curing conditions reached or exceeded the content of cement paste after 14 d of curing. A previous study on ordinary slag showed that the same non-evaporable water content with replacement rates of 50%-70% can be achieved after 60 d of curing [28,49]. This indicates that the fineness of slag plays an important role in cement hydration and the pozzolanic reaction of mineral admixtures.

Conclusions
(1) Adding 25% GSP increases the adiabatic temperature rise of high-strength concrete due to the promoting effects on cement hydration, whereas adding 45% GSP decreases the adiabatic temperature rise, which can be attributed to a reduction in cement content. (2) Compared to the compressive strength of plain cement concrete, the growth rates of strength at different ages due to the addition of GSP under temperature-matching curing conditions are higher than those under standard curing temperature. Temperaturematching curing conditions have a positive effect on the development of the late compressive strength of GSP concrete. (3) Compared to plain cement systems, cementitious material systems containing GSP tend to have lower total porosity and a lower volume of harmful pores. The dense pore structure of the GSP system leads to better chloride ion penetrability resistance of the concrete, which is more distinct under early temperature-matching curing conditions. Increasing the curing temperature has a greater influence on high-strength concrete mixed with 25% GSP. (4) The improvement effects of incorporating GSP on the late non-evaporable water content of hardened paste are significant under early temperature matching curing conditions. However, the effect of elevated temperatures on the early non-evaporable water content are limited. Increasing early curing temperatures has a more obvious effect on the pozzolanic reaction of GSP than the hydration of cement.