1.1. Materials

For this research, following materials were used: Portland cement, fine and coarse aggregates, nanosilica, silica fume and superplasticizer additive.

An early high strength cement (CPV ARI) was used. Results of physical and chemical characterization are presented in Table 1.

Physical characterization of fine and coarse aggregates is shown in Table 2 and granulometric distribution curves, obtained based on NBR NM 248 (^{ABNT, 2003}), are presented in Figures 1 and 2, respectively.

Figure 1 Particle size distribution of fine aggregate. 

Fine aggregate had fineness modulus equal to 1.33 and maximum characteristic size equal to 0.6 mm. Although fine aggregate has a particle size distribution outside the usable zone established by NBR 7211 (^{ABNT, 2009}), the same standard allows its use, whereas previous dosage studies prove its applicability. Therefore, and because it is a material commercialized in the region of study development, its use was maintained.

Figure 2 Particle size distribution of coarse aggregate. 

A nanosilica dispersed in a polycarboxylate superplasticizer liquid additive was used and the recommended content ranges from 0.5 to 3.0% relative to binder mass. Their characteristics are presented in Table 3. Table 4 shows physical and chemical characteristics of silica fume.

In reference concrete, without silica fume and nanosilica incorporation, was used a superplasticizer additive (SP) based on polycarboxylic ether, MasterGlenium® Sky 150, which specifications are presented in Table 5.

1.2. Concrete mix proportioning

Nanosilica influence was studied on 1:1.6:1.6 concrete (cement: fine aggregate: coarse aggregate) and water/binder ratio (w/b) equal to 0.4.

As already mentioned, the nanosilica used in this work is dispersed in superplasticizer additive, which limited the contents to be used. As concretes workability, obtained according to NBR 13276 (^{ABNT, 2016}), increase with increasing nanosilica content, the limits (0.5% to 3.0%) recommended were not possible to being explored. SP additive used in reference concrete also has their content defined based on mixture workability.

Therefore, nanosilica contents between 0.1 and 0.7%, isolated and with silica fume, were added to concrete as a partial cement replacement (10% in volume).

Concretes were identified by a C-N-S type nomenclature, where N and S represent nanosilica and silica fume contents used, respectively and reference concrete is identified as REF. Concretes compositions and their flow spread are shown in Table 6.

Total mixing time of materials was 7 minutes, according to following order of placement: fine aggregate, 90% water, cement, additive SP, 10% water and coarse aggregate. In concrete containing silica fume, this material was added to mixture with cement.

After mixing, cylindrical specimens of 50 mm diameter x 100 mm in length were molded. After 24 h, specimens were demoulded and taken to humid chamber, where they remained until tests date.

1.3. Tests

The nanosilica and silica fume influence was evaluated in mechanical and physical properties of concretes, by means of mechanical resistance and water absorption by capillarity tests, and their durability against the chloride attack.

3.1 Mechanical strength

The mean values of results obtained from compressive strength at the ages of 1, 7 and 28 days and 28-day tensile strength by diametrical compression tests are presented in Figures 3 and 4, respectively.

Figure 3 Compressive strength of concretes. 

Results indicated that 0.5% nanosilica addition to concretes (C-0.5-0) induce a 9.6% and 7.1% increase in 7-day and 28-day compressive strength compared to reference concrete (REF), respectively.

When combining nanosilica with silica fume, it is noticed that in early ages REF reached a higher compressive strength. However, at 28 days, the concretes C-0.5-10 and C-0.7-10 reached 76.0 MPa and 77.7 MPa, respectively, while REF reached 70.2 MPa. The greatest resistance gain at 28 days, and even at 90 days, was also observed by other researchers (^{Senff et al., 2010}; ^{Joshagani; Moeini, 2017}).

Evidences of a synergistic effect between these additions was not observed, since when adding silica in the mixture, for the same content of nanosilica (C-0.5-0 and C-0.5-10) there were no compressive strength gains.

Differently from that observed by ^{Li et al. (2017a)}, who found that addition of 2% powder nanosilica caused a 17% increase in 28-day compressive strength, whereas at 10% silica fume the increase was 11%. By combining two additions, 2% nanosilica + 10% silica fume, the increase was 48%, evidencing the existence of a synergistic effect between the additions.

Regarding tensile strength by diametrical compression (Figure 4), results indicated that additions use did not induce significant improvements, especially when considering the standard deviations. However, ^{Ganesh et al. (2016)} observed gains of 17% and 24% in 28-day tensile strength of concrete, incorporating 1% and 2% nanosilica solution, respectively.

Figure 4 28-day tensile strength by diametrical compression of concretes. 

It should be emphasized that the nanosilica used in this work is dispersed in superplasticizer and it makes the study of its incorporating at higher levels difficult, including the recommended levels. Moreover, the real content of silica present in material is not known, making it impossible to determine its percentage in relation to binder.

3.2 Water absorption by capillarity

The results obtained in water absorption by capillarity test made possible to plot curves of capillary absorption versus square root of time for each studied concrete, during 72 hours of test (Figure 5).

Figure 5 28-day capillary water absorption of concretes. 

Note that isolated nanosilica addition in the analyzed contents did not improve the water absorption by capillarity, since C-0.1-0, C-0.3-0 and C-0.5-0 presented higher capillary absorption (between 0.652 g/cm² and 0.722 g/cm²) than the REF (0.587 g/cm²). However, by raising the nanosilica content to 0.7% and adding silica fume (C-0.7-10), it was possible to reduce capillary absorption to 0.395 g/cm².

These results indicate that silica fume, with its filer and pozzolanic effects, was able to refine cementitious matrix microstructure. On the other hand, the nanosilica dispersed in superplasticizer does not have physical and chemical effects expected.

Other researchers studied the nanosilica effects on physical properties of cementitious materials and obtained positive results with reduction of sorptivity, absorption and voids index (^{Li et al., 2017b}, Ganesh et al. 20116, ^{Joshaghani and Moeini, 2017}).

Should be commented that in the cited studies, the nanosilica were in powder or solution form. Therefore, it is possible that the dispersion of nanosilica in superplasticizer additive impaired the positive results in this work.

Figure 6 shows maximum capillary rises observed in concretes after 72 hours of test. It was observed that the concretes that had lower capillary absorption index (C-0.5-10 and C-0.7-10) also presented lower ascending height, corroborating the assertion that the silica fume was efficient in microstructure matrix refinement, unlike to concretes with nanosilica only (C-0.1-0, C-0.3-0 and C-0.5-0), which presented a maximum capillary rise higher than the REF.

Figure 6 Fractured specimens after water absorption by capillarity test: a) C-0.1-0; b) C-0.3-0; c) C-0.5-0; d) C-0.5-10; e) C-0.7-10 e f) REF. 

Besides filler effect, responsible for filling voids between cement grains and aggregates, the pozzolanic effect of silica fume also contributes to microstructure densification, due to additional C-S-H formation.

3.3 Chloride ions penetration

The depth of chloride penetration in concretes was determined after 8 and 16 weekly cycles of drying and wetting in NaCl solution. Results are shown in Table 7 and the specimens submitted to the colorimetric test are shown in Figures 7 and 8.

Figure 7 Fractured specimens after colorimetric test - 8 weeks: a) C-0.1-0; b) C-0.3-0; c) C-0.5-0; d) C-0.5-10; e) C-0.7-10; f) REF. 

Figure 8 Fractured specimens after colorimetric test - 16 weeks: a) C-0.1-0; b) C-0.3-0; c) C-0.5-0; d) C-0.5-10; e) C-0.7-10; f) REF. 

Overall, nanosilica addition did not prevent chlorides penetration into concrete. After 8 weeks, the depth of penetration on C-0.1-0, C-0.3-0 and C-0.5-0 concretes were similar to reference concrete (REF). Already after 16 weeks, REF presented a lower value than those containing additions.

On the other hand, concretes with combined addition of nanosilica and silica fume presented reduction in depth of penetration. The values obtained after 16 weeks for concretes C-0.5-10 and C-0.7-10 were 1.6 mm and 2.9 mm smaller than REF, respectively. These results were expected since they were the same concretes that obtained lower capillary absorption index, as discussed previously.

^{Silva (2006)} realized the same tests on high performance concretes with 5 and 10% silica fume, and after 16 cycles drying and wetting cycles, obtained a reduction in the depth of penetration of 1.9 mm and 2.7 mm, respectively, compared to the reference concrete, which had a penetration front of 5.7 mm.

Regarding the nanosilica addition, although the present work has not obtained satisfactory results, some studies indicate that 3% and 6% colloidal nanosilica addition in mortars could reduce 70% and 77% 28-day chlorides migration, respectively, in comparison to reference mortars (^{Joshaghani; Moeini, 2017}).

Likewise, ^{Ganesh et al. (2016)} studied the nanosilica solution on concrete’s durability, by means of chloride migration test. The authors observed that, at 28 days, 2% nanosilica addition increased the concrete resistance to aggressive ions penetration. These positive results were attributed to filler effect of nanosilica and its ability to fill the voids present in concrete, thus reducing chlorides penetration.