*3.3. Microstructure Observation under SEM*

The microstructure observation revealed the different behavior of each nanoparticle in different curing conditions, concerning their potential contribution to self-healing. Representative SEM images of the systems are displayed in Figure 6, and the elemental analysis (EDS) of the samples is given in Table 5.

The cement pastes presented a typical dense structure, where few large pores were observed. In the case of water immersion, ettringite formation was observed inside the pores, as elemental analysis confirmed with the presence of Al/S = 1/2 (Rw, Sp.2, Table 5). In NLw samples, more CaCO3 is formed in relation to NLc. Nevertheless, this difference does not seem to affect the compressive strength or porosity. In samples with combined nanoparticles, the CaCO3 is much higher in the water immersion regime in relation to the cycle's conditions. Also, in this case, the CaCO3 formation at 28 days does not affect either compressive strength or porosity. On the edge of the pore, low silicon content was recorded, while the Ca/Si ratio as 11/1 indicates an area where calcite content prevails. Similarly, in the case of wetting-drying cycles, the edges of the pores presented high calcium content. A small number of deposits inside the pores had a Ca/Si ratio equal to 2/1.

The action of NC offered different results depending on the curing conditions. In water immersion, the NC-cement pastes presented healed pores with newly formed material that had its micro-porosity and a "sponge"-like structure. The additional micro-porosity created assisted in recording high open porosity, as shown in Figure 3a. This material has a high quantity of silicon and aluminum, according to NLw, Sp.2, Table 5. The previous open pores can hardly be recognized, as they have been filled with this rich in Si/Al material. This fact also explains the strength recorded for NLw samples at 28 days. In the absence of NC, the pores of cement pastes did not present such behavior, so it is believed that NC has contributed to these formations.

When NC-cement pastes were cured under wetting-drying cycles, the diameter of the pores was found smaller and this was recorded as low open porosity in Figure 3a. In this case, the pores were not filled, though some formations were started to develop. The elemental analysis showed high calcium content of these deposits (NLc, Sp.1 and Sp.3, Table 5).

The results are reversed when NS and NC are combined in cement pastes. The microstructure of the pastes treated in water immersion was dense. Small and few pores were present in the structure. This fact explains and the low porosity values recorded for these compositions. Indications of newly crystallized material were observed in the pores. In the case of NSLw, precipitation of newly formed crystals was observed within the pores but did not manage to fill the pores. The edges of the pores were hardly observed due to the material formed that had a ratio of Ca/Si/Mg approximately equal to 4/0.6/1. The indications of NS-NC-cement pastes' microstructure were very promising for self-healing capacity under wetting-drying cycles, probably, due to the initial size of the pores formed into the structure. In this case, the open pores were filled with newly formed material. The micro-porosity of this material was also observed, and the morphology was rougher than the neat cement. This material had a Ca/Si ratio of 3/1, which might have played an essential role in the density and the morphology of the formed material. This proportion could be an indication of calcium silicate hydrated compounds that compose the new material.

**Figure 6.** SEM-micrographs of samples cured; (**a**) in water immersion and; (**b**) under wetting- drying cycles, at 28 days.
