Effect of Nanosilica on Impermeability of Cement-Fly Ash System

Surface protection has been accepted as an eﬀective way to improve the durability of concrete. In this study, nanosilica (NS) was used to improve the impermeability of cement-ﬂy ash system and this kind of material was expected to be applied as surface protection material (SPM) for concrete. Binders composed of 70% cement and 30% ﬂy ash (FA) were designed and nanosilica (NS, 0–4% of the binder) was added. Pore structure of the paste samples was evaluated by MIP and the fractal dimension of the pore structure was also discussed. Hydrates were investigated by XRD, SEM, and TG; the microstructure of hydrates was analyzed with SEM-EDS. The results showed that in the C-FA-NS system, NS accelerated the whole hydration of the cement-FA system. Cement hydration was accelerated by adding NS, and probably, the pozzolanic reaction of FA was slightly hastened because NS not only consumed calcium hydroxide by the pozzolanic reaction to induce the cement hydration but also acted as nucleation seed to induce the formation of C-S-H gel. NS obviously reﬁned the pore structure, increased the complexity of the pore structure, and improved the microstructure, thereby signiﬁcantly improving the impermeability of the cement-FA system. This kind of materials would be expected to be used as SPM; the interface performance between SPM and matrix, such as shrinkage and bond strength, and how to cast it onto the surface of matrix should be carefully considered.


Introduction
Concrete has been widely employed in civil constructions in the past decades, and the main reason is due to its excellent performance, especially long life serving [1,2]. However, reinforced-concrete structures inevitably suffer from damages in harsh environment, such as in the presence of chloride ions, carbon dioxide, and sulphate, resulting in shortening its life [3][4][5][6]. e generally accepted reason for those is that concrete is a porous material with microstructural defect, and the harmful ions could penetrate the surface into the inside of concrete structure and cause corrosion, finally leading to crack [7,8]. For example, sulphate was able to penetrate the surface into the inside and react with hydrates to produce ettringite or gypsum, associating with a volume expansion. e mechanical performance could be deteriorated, and under load, the cracks occurred; thereafter, sulphate ions were easier to go into the inside of concrete structure along the cracks and the more serious corrosion would happen [9][10][11]. Chloride ions could easily permeate into the inside of the matrix and migrate to the surface of steels and then cause steel corrosion probably [12,13]; finally, volume expansion happened and the concrete structure was destroyed. In the process of various external erosions, one precondition is that corrosion ions need to penetrate through the surface and go into the inside of the concrete. Accordingly, surface protection has been accepted as one of the most popular way to promote the concrete durability in actual engineering process [14].
Coating a film on the surface of concrete is one of the main methods for surface protection [15]. Two kinds of surface protection materials (SPM) were reported in the literature. One was the penetrative protective agent based on organic materials, with intention to block the surface pore structure; the other was the inorganic materials, such as mortar or cement paste, to isolate the matrix from erosion ions in outside environment. Silanes [16], ethyl silicate [17,18], and isobutyl-triethoxy-silane [19] were used as typical penetrative protective agents for the resistance of carbonation and erosion, and a silica-based hybrid nanocomposite was reported for surface treatment to reduce the water absorption and gas permeability [20]. One mechanism behind the enhanced impermeability was that the chemicals could go through the capillary and form an inorganic hydrophobic film along the capillary as a result of reacting with hydration products; the other was that these hydrates in the capillary structure would block the transportation of water and gas [21]. Furthermore, another kind of SPM was coated as a layer on the surface of the matrix. Zhang reported that cement mortar modified with nanosilica (NS) and silica fume could be used as SPM to enhance the impermeability [22]. e reason was revealed that NS and silica fume not only consumed CH to form the dense C-S-H but also exerted the excellent filling effect, thereby declining the porosity, increasing density, improving the microstructure including interfacial transition zone (ITZ), and enhancing the mechanical performance [23][24][25][26]. Additionally, the inorganic materials generally performed better antiaging property and also showed excellent ability to augment the impermeability by coating on the surface [27,28].
In the literature, incorporating mineral admixtures, such as fly ash and ground granulated blast furnace slag (GGBS), could not only exhibit economic and environmental benefits [29][30][31] but also greatly improve the durability of cementitious materials [32,33]. In this paper, the cement-fly ash-NS system was designed and used as a SPM. 30% fly ash (FA) and NS (0-4%) were added with the intention to improve the impermeability. e chloride diffusion coefficient of SPM was determined to illustrate the effect of NS addition on the impermeability of SPM under investigation. Furthermore, the properties of the pore structure, microstructure, and hydration products of SPM were investigated to reveal the mechanism of its high impermeability. e pore structure was discussed by compressive strength, water absorption, and MIP; microstructure was investigated with SEM-EDS. Hydrates were evaluated with XRD, TG-DSC, NMR, and SEM. e findings would expect to be useful for the design of surface protection materials.

Nanosilica.
e average particle diameter of powdered nanosilica (NS, supplied by Sinopharm Chemical Reagent Co., Ltd.) used in this study was about 15-20 nm, which was indicated from the TEM image, as shown in Figure 2.

Preparation of Samples.
e mortar samples were prepared according to the mixture proportion, as shown in Table 2. Because NS showed negative effect on the workability of C-FA system [34], the same fluidity level (160-180 mm) was controlled by adding PCE. NO.0 mortar was deemed as the reference sample. Cement and fly ash was firstly mixed, and then the NS suspension, dispersed by PCE [35] and water, was added. After that, sand was added within 30 s, followed by continuous stirring for 90 s, and then SPM was prepared. After the samples were cured in the standard condition (>95% R.H. and 20 ± 1°C) for 24 h, they were demoulded and further cured till 28 d. ese mortars were also cast in the model with ϕ100 mm and 50 mm height and cured under the same condition till 28 d. In accordance with Chinese standard GB/T 50082-2009, rapid chloride migration (RCM) was used for the measurement of chloride diffusion coefficient. e sand used in experiments was Standard Sands (GSB 08-1337-2017, supplied by Xiamen aisio standard sand Co., Ltd.).
ese fresh mortars were also cast in the cubic moulds (40 × 40 × 40 mm). After being cured for 7 d, 28 d, and 90 d under standard condition, the measurements of compressive strength, water absorption, and microstructure were conducted. e cement pastes were also prepared, as shown in Table 2, and cured under the standard condition for 28 d.
e samples were broken into small pieces and immersed into the anhydrous ethanol for 48 h to stop its hydration. ese small pieces were prepared for the measurement of SEM and pore structure. Furthermore, the broken pieces were grinded into powder to pass a 45 μm sieve, being ready for the analysis of XRD, TG, and NMR.

Chloride Penetrability.
Mortar samples were put into a vacuum container for 3 hours, followed by being immersed into saturated Ca(OH) 2 solution for 18 ± 2 h. 0.3 mol/L NaOH and 10 wt% NaCl were used as cathode solution and anode solution. External electrical potential (30 ± 0.2 V) was employed to the upper and lower section of the samples. After the measurement, the samples were carefully cleaned and then split into two semicylinders along the axial direction. Finally, 0.1 mol/L silver nitrate solution was sprayed on the cross section, and the chloride penetration depth of the specimen was measured at 10 different locations. ree samples were used for each test group. e D RCM was calculated as follows:  Advances in Civil Engineering

Water Absorption.
e water absorption of cement mortars cured for 28 d was tested according to GB/T 17671. e specimen was dried in an oven with temperature of 80 ± 2°C for 48 h and then removed and cooled to room temperature. After that the mass of the sample (G 0 ) was weighted and then immersed into water with the temperature of 20 ± 2°C for 48 h; thereafter, the sample was taken out of the water, and the water on the surface was cleaned; the mass of the sample (G 1 ) was weighted again. Water absorption (equation (2)) was calculated as follows: where W A , water absorption (%); G 0 , the mass of the specimen after drying (g); and G 1 , the mass of the specimen after adsorbing water (g).

Phase
Analysis. e measurement of XRD was conducted with X-ray Diffractometer (D/Max-RB, made by Rigaku, Japan) with Cu (Kα) radiation and a current of 40 mA at 40 kV, at a speed of 2θ 4°/min and a step of 0.02°within the range from 5 to 70°. e data was processed with Jade 5.0. e measurement of TG-DSC was carried out with the instrument of STA449c/3/G (NETZSCH, Germany) under an atmosphere of flowing air from room temperature to 1000°C at a heating rate of 10°C/min. e accurate content of calcium hydroxide (CH) was calculated from the weight loss ranging from 400-500°C [36,37].
Morphology characterization of hydrates in the paste sample was conducted with Field Emission Scanning Electron Microscope (FE-SEM, QUANTA FEG 450).

NMR.
e hydration products were characterized with 29 Si MAS NMR. It was reported that six peaks were found in NMR spectrum of hydrated cement-FA paste [12,38,39]; Q 1 and Q 2 represented the chain-end Si-O tetrahedrons and middle-chain Si-O tetrahedrons in hydration products; Q 2 (1Al) denoted the middle-chain groups where one of the adjacent tetrahedral sites was occupied by Al 4+ ; Q 0 , Q 3 , and Q 4 represented the Si-O tetrahedrons in unhydrated cement minerals and Si-O tetrahedrons in FA, respectively. A Bruker Advance III400 spectrometer was adopted to carry out 29 Si-NMR (solid-state nuclear magnetic resonance), operating at 79.5 MHz, with rotation frequency of 5 kHz and the delay time of 10 s. Tetramethylsilane was used as 29 Si standard. e data were processed with commercial solid-state NMR software package. During the deconvolution of 29 Si NMR spectra, the peak was fitted and Gaussian function was used to constrain the peak shapes. e main chain length (MCL, equation (3)) of C-S-H gel and the ratio of Si in C-S-H substituted by Al (equation (4)) were obtained as follows [40][41][42]:   Advances in Civil Engineering 3 Al Reaction degree of cement is shown in equation (5). Reaction degree of FA in the C-FA system is exhibited in equation (6). Because Q 3 and Q 4 are also presented in NS, it is difficult to calculate the hydration degree of FA and NS, respectively. However, the total reaction degree of FA and NS could be calculated, as shown in equation (7). It was reported that most of NS was depleted at the early age [43]. Based on this, A (FA+NS) could reflect the pozzolanic reaction of FA: where I(Q 0 ), I(Q 1 ), I(Q 2 ), and I[Q 2 (Al)] denoted the intensities of signals Q 0 , Q 1 , Q 2 , and Q 2 (Al) in hydrated paste, respectively; I 0 (Q 0 ), I 0 (Q 3 ), and I 0 (Q 4 ) represented the intensities of signals Q 0 , Q 3 , and Q 4 in unhydrated mixture.
2.2.6. Pore Structure. Mercury intrusion porosimetry (Auto Pore IV 9510, America) was used for pore structure characterization. e calculation of pore distribution assumes that the pores are spherical in shape, with the surface tension of mercury of 480 mN/m and contact angle of 140°. During the measurement of MIP, external pressure on mercury was equal to the surface energy increased by pores and the correlation appeared in where W is the surface energy of the pore; P is the mercury pressure; V is the intruded mercury volume; c L is the surface tension of mercury; θ is the contact angle of mercury with the surface of the pore; and S is the surface area of the pore. In the literature, the relationship between the accumulated intrusion surface energy (W n ) and the accumulated mercury intrusion surface (Q n ) (shown in equation (9)) was proposed by Zhang and Li [44]. Based on this, the accumulated mercury volume is V n and the smallest pore radius is r n ; then, W n and Q n are expressed as equations (10) and (11): where C is a constant; P i and V i stand for the mercury pressure and intruded pore volume at step i; and D stands for the fractal dimension of the pore surface [45][46][47][48]. erefore, equation (9) is expressed as equation (12). e fractal dimension of the pore surface can be obtained by the slope of the curve log (V 1/3 n /r n ) versus log (W n /r 2 n ). Greater than 0.99 was controlled in the correlation coefficient: 2.2.7. Analysis of Microstructure. e mortar samples were dried under vacuum and coated with a golden film, and then, a scanning electron microscope (SEM-EDS, JSM-5610LV, Japan) equipped with an energy dispersive spectrometer (EDS, made by EDAX Inc., USA) was employed to observe the microstructure. Based on EDS, different phases were reflected from different colors, and the element content (C) in each phase and the ratio of each phase (A) in the whole image was read directly from the EDS data. Based on these results, the average Ca/Si ratio (equation (13)) and the Ca/Si ratio of each phase (equation (14)) were calculated as follows [49]: where C n−Ca and C n−Si mean Ca and Si content in n phase, and this phase occupied A n ratio in this SEM image.

Results and Discussion
3.1. Chloride Penetrability. Chloride diffusion coefficient (D RCM ) of the C-FA-NS system was examined, and the results are shown in Figure 3. From the figure, it was seen  [8,22]. It is generally believed that the impermeability of cement-based material is related to the pore structure and microstructure [50]. Improvement in the microstructure and refining of the pore structure would significantly augment the impermeability. ese would be further discussed in the following text.

Pore Structure.
e pore structure was investigated in terms of water absorption, MIP, and compressive strength and shown as follows.

Water Absorption of C-FA-NS System.
Water absorption of the C-FA system with various dosages of NS cured for 28 d are shown in Figure 4. It can be seen that the water absorption was reduced with the increasing dosage of NS. Addition of 0.5% NS reduced the water absorption by 5.45% and increase in dosage to 4.0% further reduced it by 21.82%, in comparison with the reference (without NS). As reported in the literature [51,52], water absorption reflected the surface pore structure of the mortar. Generally, greater water absorption represented more amount of open pore which was related to the impermeability. Based on this, it was inferred that NS in the C-FA system significantly improved the surface pore structure, which was responsible for the enhanced impermeability.

MIP.
e pore structure was characterized with MIP, and the results are shown in Figure 5 and Table 3. From the figure, it was seen that addition of NS could obviously refine the pore structure of the C-FA system, and the most probable aperture (MPA) was notably reduced by adding NS, in agreement with the results in the literature [53,54]. In Table 3, it was found that NS reduced the porosity of C-FA from 0.1808 mL/g in reference (C-FA) to 0.1448 mL/g and 0.1304 mL/g in sample with 2.0% NS and 4.0% NS, respectively. Additionally, NS notably increased the pore volume with size smaller than 20 nm, while significantly reducing the pore volume with size bigger than 200 nm. ese results implied that NS, with nanoscale particles, was able to significantly refine the pore structure [55][56][57]. Similar results concerning nanoparticles were also reported in the literature [58][59][60][61][62]. e microporous structure was related to transport properties of cement paste, which could be reflected by fractal dimension. In this study, fractal characteristics of the pore structure was calculated from MIP data [45,48,63], and the results are posted in Figure 6. It was observed that two fractal areas were separated by a transition area. It was reported that Ds-a of the macrofractal region represented the packing patterns of the hydrated binder particles, while the Ds-i of the differential region mainly showed the microstructure of the C-S-H gel [45,46]. 2% and 4% NS increased the Ds-a of the samples from 2.006 in reference to 2.066 and 2.329 and increased Ds-i from 2.685 in reference ese phenomena indicated that the addition of NS disordered the transport tunnel and made the pore structure more tortuous at the age of 28 d, which could reduce the transport properties of the hardened paste.

Compressive Strength.
Compressive strength was used to offer the supplementary evidence to illustrate the pore structure [64,65]. e compressive strength of the C-FA-NS system cured for 7 d, 28 d, and 90 d was examined, and the results are shown in Figure 7. With the increasing dosage of NS, 7 d strength and 28 d strength were increased. Especially at the age of 7 d, 2.0% NS increased the strength from 21.0 MPa in reference to 30 MPa, with an increase by 43%; the increase rate is by 11% for 28 d, while in the case of 90 d, the NS containing sample showed lower strength than the reference sample. As reported in the literature, the increased strength at the early age was related to the refined pore structure and the accelerated hydration process, while the reduced strength at the later age was due to the coating effect resulting from pozzolanic reaction of NS on the surface of FA [66].
Based on discussion about the water absorption, MIP, and compressive strength, it can be concluded that NS added in the C-FA system could significantly refine the pore structure of the system. With increase in curing age, the hardened structure was densified, showing compacter structure, which was inferred conducive to impermeability of the system [67].  Furthermore, analysis about the microstructure of these areas was carried out, and Ca, Si, and phase distribution in different phases was displayed in Table 4. It can be found that the Ca/Si ratio of the red color was as low as 0.02, indicating that the red color represented the sand. Furthermore, as shown in Table 4, the average Ca/Si ratio in image was calculated. In the C-FA (reference) system, the average Ca/Si ratio was 1.86; 2.0% NS reduced it to 1.52, and increase in dosage to 4.0% further reduced it to 1.43. In fact, the Ca/Si ratio in hydrates reflected the CH content in C-S-H, and generally, higher Ca/Si ratio meant greater amount of CH. Based on this, the reduced Ca/Si ratio by adding NS could be explained from the fact that NS with nanosize entered these areas near the aggregates and reacted with CH to reduce the Ca/Si ratio. e whole microstructure including the microstructure of ITZ was improved significantly, with obvious contribution to the strength.

Phase Analysis.
In order to discuss the effect of NS on hydration properties of the C-FA system, the samples were investigated with XRD, TG, and NMR.

XRD.
e peak intensity of CH in the XRD pattern could reflect the relative content of CH in hydrates, which illustrated the hydration process of the cementitious system [69]. In the C-FA-NS system, cement hydration produced CH, and the pozzolanic reaction of FA and NS consumed CH to reduce the amount of CH in hydrates [70]. e paste cured for 28 d was characterized with XRD, and the results are shown in Figure 9. From the figure, CH, ettringite (AFt), C 3 S, portlandite, and mullite were seen clearly. It was found that NS significantly reduced the peak intensity of CH and slightly enhanced the peak intensity of AFt, while obvious change of the peak intensity of C 3 S and mullite was not observed. It was confirmed that the reaction between NS and CH happened. It was deduced that the consumption of CH by this reaction could induce cement hydration. Obviously, the consumption of CH by NS was dominated, and therefore, the reduced CH peak intensity by adding NS was observed.

TG-DTG.
To further confirm the content of CH in hydrates, TG-DTG was employed. As shown in Figure 10, the peaks ranging from 50-200°C and 400-500°C were seen clearly. e former represented the dehydration of C-S-H and AFt, and the later denoted the decomposition of CH. Table 5 illustrates the weight loss calculated by TG.
From the table, it was found that NS addition increased the weight loss occurring at 50-200°C. In the C-FA system, the weight loss within the temperature range of 50-200°C was 7.46%; 2.0% NS increased it to 8.29%; increase in dosage to 4.0% further increased it to 9.05%. is result indicated that NS accelerated the formation of hydration products such as AFt or C-S-H gel. Furthermore, the weight loss at 400-500°C was also reduced from 3.21% in C-FA to 2.67% in C-FA-2.0% NS system and 2.04% in C-FA-4.0% NS system. Based on this, the CH content was calculated; in the C-FA system, the CH content was 7.85%, and 2.0% and 4.0% NS reduced it to 6.53% and 4.99%. It was clear that NS reduced the content of CH in hydrates. e results agreed with the results in XRD analysis. e mechanism behind was revealed as follows: NS would react with CH to produce the C-S-H gel, and the consumption of CH would induce the cement hydration to generate more CH. NS with nanoparticles could also act as nucleation seed to induce the hydration of cement to produce hydrates [37,71,72]. In this case, cement hydration was hastened in the following age.

NMR.
NMR was used to analyze the hydration products. e NMR pattern obtained from paste hydrated for 28 d are shown in Figure 11. In the process of hydration, Q 0 was due to the unhydrated cement; Q 3 and Q 4 were attributed to unhydrated FA and NS. As shown in Figure 11, the intensity of Q 0 , Q 3 , and Q 4 in raw materials reduced, while that for Q 1 , Q 2 , and Q 2 (1Al) in hydrates increased obviously, indicating the hydration of cement and FA as well as the formation of C-S-H gel. As shown in Table 6, in the C-FA system, the reaction ratio of cement was 58.68%; 2.0% NS increased the reaction ratio of cement to 64.91%. Furthermore, it was noted that the total reaction degree of FA and NS could be calculated as 14.77%; the reaction degree of FA could not be directly obtained. As the value of A (FA+NS) was 4.74% higher than that of A FA , it is most likely that hydration of FA was slightly promoted in the C-FA-NS system in comparison with C-FA. NS also increased the MCL value from 3.94 in the C-FA system to 4.40 in 2% NS system and 4.48 in 4.0% NS system, indicating that NS accelerated the polycondensation of C-S-H gel in the system. e Al/Si ratio was also increased by addition of NS, indicating that NS accelerated the dissolution of aluminate Advances in Civil Engineering from FA into the liquid phase to take part in formation of C-S-H gel [73][74][75]. Based on this, it was confirmed that the cement hydration was promoted; and probably, pozzolanic hydration of FA was slightly accelerated by adding NS.

SEM.
Microstructure of hydrates at the age of 28 d was characterized by SEM, and the SEM images are shown in Figure 12. From Figure 12(a), C-S-H gel and FA were seen clearly, and a small amount of hydrates on the surface of FA was found. In Figure 12(b), a large amount of hydration products on the surface of FA was found clearly; fly ash was encased by these hydration products, and the same phenomenon was also found in the literature [76]; these hydrates were attributed to the pozzolanic reaction of FA and NS. As reported in the literature [66], in the C-FA-NS system, a coating on the surface of FA could be formed by pozzolanic reaction of NS and FA, and it would possibly slow down the pozzolanic reaction of FA if the coating was better formed. Probably, this was one reason why NS reduced the strength of C-FA at the age of 90 d. In comparison with Figure 12(b), fly ash was encased by more amount of hydrates in Figure 12(c), and this indicated that more amount of hydrates on the surface of FA particles was produced in the C-FA system with 4.0% NS. It was deduced that NS would significantly improve the interface between FA particles and cement particles, and this could contribute to compressive strength and impermeability.
Based on the discussion above, it was concluded that NS accelerated cement hydration, and most likely, it slightly hastened the pozzolanic reaction of FA. e amount of CH in hydrates was reduced due to the consumption of pozzolanic reaction of FA and NS; the amount of C-S-H gel was increased because of the accelerated hydration of cement and pozzolanic reaction of NS.

Mechanism.
In the C-FA system, NS was able to significantly increase the impermeability, and the reason was revealed as follows: firstly, NS with nanosize could exert the filling effect to block capillary pore. Because of high reaction activity, it could efficiently take part in pozzolanic reaction and also act as nucleation seed to induce the cement hydration; accordingly, the formation of C-S-H gel with nanosize was facilitated, which could also fill or block the capillary pores greatly to improve the impermeability   [77,78]. Furthermore, NS induced the hydration of cement by consuming CH and acting as nucleation seed, and the hydration degree of the system was promoted. is significantly improved the microstructure, such as microstructure of C-S-H gel as well as the interface between FA and C-S-H gel, thereby enhancing impermeability. e pore structure        Figure 12: SEM images of C-FA hydrated for 28 d.

Cement particle Hydration products
Fly ash NS 10 Advances in Civil Engineering was refined and the microstructure was improved, with great contribution to the impermeability, as shown in Figure 13.

Conclusion
(1) In the C-FA-NS system, NS obviously refined the pore structure, increased the complexity of the pore structure, and improved the microstructure, and the reason was not only due to the filling effect of NS but also because of a large amount of hydrates formed to improve the microstructure. (2) NS accelerated the whole hydration of the cement-FA system. NS not only consumed calcium hydroxide by the pozzolanic reaction to induce the cement hydration but also acted as nucleation seed to induce the formation of C-S-H gel, thereby accelerating the cement hydration. Probably, it also slightly accelerated the pozzolanic reaction of FA. (3) NS significantly improved the impermeability of the cement-FA system, and this kind of materials would be expected to be used as surface protection material. However, the interface performance between SPM and matrix, such as shrinkage and bond strength, and how to cast SPM onto the surface of the matrix should be carefully considered.

Data Availability
e data used to support the findings of this study are available from the corresponding author upon request.

Conflicts of Interest
e authors declared that there were no conflicts of interest regarding the publication of this paper.