Brazilian Portland cement type CP II F-40 was used (composed of 11‒25% in mass of limestone filler and 75‒89% of clinker, with a small amount of calcium sulfate), whose mechanical strength requirement is a minimum of 40 MPa of compressive strength at the age of 28 days (similar to the European CEM II/A-L), together with micro filler, filler, fine aggregate, crystalline chemical admixture and water. The chemical and physical properties of the cement used are given in Table 1.

The filler and micro filler used were waste from crushed granitic rocks. Figure 1 shows the granulometric distribution curves of these two materials. D50 particles of 26 μm and 62 μm were observed for the micro filler and filler, respectively. The granite sand was used as fine aggregate (Table 2 and Figure 2). It was observed that the sand used was almost entirely within the “usable zone”, according to Brazilian Standard NBR 7211: 2009 (63). The crystalline admixture utilized was a crystallizing powder admixture. This was a gray powder with a pH value of 12 ± 1.0, equivalent sodium oxide less than or equal to 3%, and a volume density of approximately 750 kg/m³. The recommended content is between 1% and 2% of the weight of the binder. Besides that, this crystalline chemical admixture is usually composed of a mixture of cements, amino alcohols, and other binding materials (64).

To perform the tests, it was decided to mold cylindrical samples, adapting the method proposed by NBR 7182: 2016 (65) for soil compaction testing. The cylinder used in molding was a small cylinder (Proctor) 10 cm in diameter and 12.7 cm in height, with a volume of 997.5 cm³.

The purpose of this adaptation was to simulate press molding, because the press was unable to mold the cylindrical samples required for the tests. A previous study (41) showed that the density obtained through molding in the press for the manufacture of modular blocks (the same used to obtain the water content adopted in the present work) approached the density obtained manually, using the cylinders described with normal energy. In this research, the density of each mixture was also measured.

After that, the components of the micro concrete were placed in a receptacle in the following order: sand, filler, micro filler, cement and admixture. They were then pre-mixed manually until well homogenized and placed in a planetary mixer with a capacity of 20 liters. The mixing was carried out by adding water with a mixer rotation of 125 rpm for 5 minutes.

The molding of the samples was performed in 3 layers with 26 strokes per layer with a proctor socket of 2.5 kg, falling from a height of 30.5 cm, equivalent to normal compaction energy (65). The specimens used in the tests were molded, adapting the proposed method for soil compaction with normal energy. After compacting the three layers, the specimens were leveled, removing excess concrete using a metal ruler. After extraction, the specimens were labeled and wrapped in four layers of plastic film.

After the demolding, because of technical problems in the moist room at the laboratory, the curing was performed for 12 weeks (84 days) in the following way: all the specimens were labeled and wrapped in four layers of plastic film in the laboratory at 25±2 °C and 60± 5% of relative humidity. After 84 days of ageing, two exposure conditions were conducted: firstly, immersed in lime-saturated water and then kept for an additional 67 days until the age of 151 days, which was named the “saturated” condition. Secondly, they were kept dry (wrapped in four layers of plastic film). This last condition was named “sealed”. These procedures were adopted to evaluate the action of the crystalline admixture in the presence of water.

The compressive strength tests and water absorption tests were performed on cylindrical specimens with dimensions of 10 cm x 12.7 cm (diameter x height). The former test was performed according to the Brazilian standard NBR 5739: 2018 (66) at the age of 28 days and 154 days on the sealed specimens and saturated specimens.

The absorption of water and void index in the hardened micro concrete were investigated at the age of 154 days, following Brazilian standard NBR 9778: 2009 (67). Three specimens were tested for each condition of the study.

For a good interpretation of the results, the values obtained for the compressive strength, water absorption and voids index were submitted to the analysis of variance (ANOVA) test, using Statistica software version 10.0 from StatSoft®.

In addition, factorial experiments can allow general conclusions, which are more interesting than the isolated analysis of each variable. Dixon tests were therefore performed to analyze outliers and it was possible to observe the influence of the variables: crystalline chemical admixture, water content, exposure conditions (sealed or saturated) and age of the tests, as well as the interaction of these variables.

To analyze the microstructure of these different materials, images were taken with a scanning electron microscope Jeol JSM-6610 from LABMIC laboratory (High Resolution Microscopy Multiuser Laboratory). Amplifications from 100x to 5000x were used for comparing the cement matrices, assessing their homogeneity, the presence of fissures, and hydrated products, among other aspects.

The micro concrete samples were reduced in a concrete cutter and manually fractured using tools. They were dried in a laboratory oven until they were ready to be preconditioned. After drying, the fractured surface was coated with gold. These tests were run on samples of more than 1170 days and comparisons between the micro concretes studied were made.

The averages of the results obtained for density and compressive strength are shown in Table 3. Figure 3 presents the average and the standard deviation of compressive strength at the age of 28 days (orange color) and 154 days with sealed (blue color) and saturated (grey color) micro concrete specimens, with (A) and without (N) crystalline chemical admixtures for 7% and 11% of water content.

From Table 3 and Figure 3, it is observed that the densities of the concretes with 11% water content were considerably higher than those with 7%, as expected. Comparing the densities with and without crystalline admixtures for the same water content, the values were the same or the differences were negligible. This same observation was validated for the densities of the micro concretes at 28 and 154 days with and without crystalline admixture and in the water content studied.

In relation to the compressive strength, the micro concretes with 11% water content had higher values of compression than the micro concretes with 7% water content at all ages and conditions studied (about 33%). Thus, the greater amount of water in the micro concrete with 11% water content caused both greater compaction of the mixtures and greater hydration of the cement. As a result, the compressive strength increased. The effect of the age of these micro concretes is negligible or small.

Analyzing the micro concretes with 11% of water content, there was a tendency to a reduced compressive strength when the crystalline admixture was used. This trend occurred for both ages: 28 days (reduction of 12%) and 154 days, in both conditions of exposure (about 6%). As for the micro concretes with 7% water content, there was an inverse behavior, which meant an increase in compressive strength when the admixture was added (from 6% to 12%, depending on the age and exposure condition). This tendency of behavior with the two water content levels of the micro concretes can be related to the interaction between the filler effect and the influence of the greater surface area of the fines included in the mixtures with crystalline admixture. However, a more accurate analysis can be done through consideration of the statistical results. Table 4 shows the analysis of variance (ANOVA) of the compressive strength results.

Table 4 shows that the model adopted was significant, with a coefficient of determination of the model (R²) of 0.75. This means that 75% of the total variation of the data is explained by the model. The analysis of variances showed that the only statistically significant individual effect was that of water content at a 95% confidence level. This expresses that this single variable influences the compressive strength of the micro concretes, with the formation of two statistical groups of water contents: one with 11% and other with 7%. The crystalline admixture used did not influence the compressive strength results. Additionally, only the double interaction between the water content and the admixture was significant, with the water content having more influence.

In the literature, there is no agreement relating to the effect of the crystalline admixture on the compressive strength of micro concrete, considering that there was no pre-cracking in self-healing studies. Pazderca and Hájková (23) evaluated the influence of two different crystalline admixtures on the compressive strength at 28 days and concluded that the result was practically the same as for the concrete without admixture. By contrast, in the research developed by Hassani et al. (25), the crystalline admixture used significantly influenced the compressive strength at the age of 28 days, with rates of increase of 11% and 21% and a rate of decrease of 24%, depending on the type of cement used. The cement type had a greater influence than the type of admixture studied. The concrete with admixture studied by Reddy and Ravitheja (68) achieved an increase of 11% compared to the reference concrete, which was attributed to the filling of the voids by C-S-H and calcite (one of the morphologies of calcium carbonate). In another study, Al-Rashed and Al-Jabari (8) observed improvements in the compressive strength of different concretes with multi-crystallization, mainly at the ages of 28 and 56 days.

Regarding the effect of exposure at different ages, Weng and Cheng (24) tested different levels of crystalline admixtures (0, 3, 5 and 7%). They observed that the compressive strength of the concretes studied after water exposure was considerably higher than those after air curing exposure at the ages of 7, 28 and 56 days. The increase of crystalline admixture content showed an increase of compressive strength. For example, with 3% of this admixture, there were increases of 14%, 20% and 22% at the ages of 7, 28 and 56 days, respectively, in comparison to those without the admixture.

The water absorption and void index results are summarized in Table 5. Dixon’s test was performed and no data were excluded. As mentioned before, this test was performed only for the age of 154 days.

Similar to the compressive strength, the effect of the water content of the micro concretes was clear in this test. By increasing the water content from 7% to 11%, the average value of the total water absorption decreased by 56% and the void index by 47%.

Figures 4 and 5 show the average and standard deviation results of the water absorption and void index of all the studied micro concrete mixes.

As concerns the type of exposure conditions, it was observed that the saturated condition promoted a tendency of reducing both the total water absorption and the void index in the mixes studied. However, this trend was greater for N11 (5.4% for both the water absorption and void index). In the A11 micro concrete with the crystalline admixture, these reductions were 4.1% and 4.6%, respectively. For the mixture with 7% water content, the greatest reduction trend was for the mix without admixture (N7), with 3.9% for both indexes. For the micro concrete with admixture (A7), this trend was much lower, with values of 0.9% and 1.5%, respectively.

Related to the effect of the crystalline admixture, for the 11% water content, it was observed that both the water absorption and the void index showed no considerable reduction. However, for the water content of 7%, this admixture caused a tendency to reduce the total water absorption and the void index in both cases (sealed and saturated samples). This behavior was consistent with an increase in compressive strength obtained, although, in that case, such an increase could not be confirmed by the statistical analysis performed.

Tables 6 and 7 present the ANOVA results for the absorption and void index, which were analyzed together because they were obtained from same test.

The ANOVA results evaluated the influence of three independent variables on the absorption and void index: water content, crystalline admixture and exposure conditions. All three factors were significant, and the interaction between the water content and crystalline admixture was equal in the compressive strength. When comparing the Fcal values, considering that Ftab has the same value, it was observed that the most significant variable, much more than the others, was the water content, followed by the exposure conditions, the interaction between the water content and crystalline admixture, and finally the crystalline admixture.

By the grouping of means, using the Duncan test, the mixes were divided into four distinct groups, shown in Table 8. Across the groups, the water content and conditions of exposure were more predominant than the other variables analyzed. The crystalline admixture, although significant, did not interfere much in the division of groups (see Table 8).

Different from the results obtained in this research, a study by Petrucci and Hastenpflug (30) indicated that 0.8% content of crystalline admixture reduced the total absorption (by immersion) and the void index at each wetting and drying cycle (four in total), from a 3.1% water absorption rate to 0.3% for conventional concrete. In comparison with the reference concrete (without admixture), the reduction was 90.5% for the water absorption rate and 90.2% for the void index.

In the research developed by Hassani et al. (25), also with conventional concrete, the crystalline admixture used did not influence significantly the void ratio (ANOVA results), calculated by the ASTM C642-13 standard, presenting a maximum reduction of 10% for a given type of concrete.

To connect the results obtained with the microstructure of the micro concretes evaluated, the SEM images are shown in Figures 6 and 7. It was thus possible to observe the influence of the water content and the type of exposure condition on the appearance of the micro concretes.

Analyzing Figure 6 showing the micro concretes without admixture, it is noted that the micro concrete specimens with 7% water content (N7) were more heterogeneous and porous than the mixtures with 11% (N11), including the presence of a crack (in the saturated condition). Considering the exposure conditions (saturated or sealed), the appearance was similar: the concrete surfaces had the same aspects. This corroborates the compressive strength and water absorption results, which indicated a better performance of the micro concretes with 11% water content and no difference between the types of exposure conditions.

Considering the water content, the same behavior seen in Figure 6 was observed in Figure 7, especially for the sealed condition. Comparing the exposure type, for 7% water content with the crystalline admixture the saturated condition promoted a more compact and homogeneous cement matrix, corroborating the small improvement in the compressive strength and water absorption results. This could indicate that water is necessary for adequate performance of the crystalline admixture.

To analyze the effect of the crystalline admixture, Figure 8 shows some SEM images of mixtures with 7% water content in the saturated and sealed conditions.

Figure 8 presents cementitious matrices with no great differences between them. The presence of the crystalline admixture seems not to significantly alter the appearance of these micro concrete samples, but has a tendency to produce slightly more porous micro concretes, especially for the saturated condition. The use of wet curing (saturated) produced, apparently, less homogeneous materials. But, as seen in the macro properties presented before, the exposure condition showed no significant differences in the results. For 11% water content, related trends were observed with similar micro concretes studied, analyzing the same aspects.