3.2. Adsorption/Desorption Isotherms of Dry Mortar References (Mix 1)
According to the International Union of Pure and Applied Chemistry (IUPAC), a classification system for adsorption isotherms can be categorized into six types based on the isotherm shape [
37].
This type of adsorption is a type of physisorption, which is caused by the affinity of the nitrogen gas with the surface of each sample. A material is characterized by its adsorption isotherm curve, that is to say, the quantity of gas adsorbed, as a function of the ratio P/P
0, see
Figure 3. The quantity of adsorbed gas intruded depends on the ratio P/P
0, on the temperature, on the morphology of the solid/gas interface.
The analysis of the adsorption/desorption isotherms of the reference samples show that their configuration are close to types IV and V (
Figure 3). According to the classification of IUPAC, these two types are associated with capillary condensation taking place in mesopores. Considering the classification of the type of hysteresis loop, they are close to types H
3 and H
4 [
38]. The H
3 type hysteresis loop can be attributed to capillary condensation taking place in a non-rigid texture but is not characteristic of a specific porosity. This type of loop is, often observed with materials in powder form, or fine aggregates [
39]. The H
4 hysteresis loop is often observed with materials with micropores in sheets, between which capillary condensation can occur. The isotherm with a hysteretic loop is obtained with mesoporous and/or microporous adsorbents in which capillary condensation occurs. During the desorption if the adsorbed gas condensed by capillarity is trapped, a hysteresis between the sorption and desorption curve is then observed. This effect is observed for all the reference samples.
According to
Figure 3, the reference samples prepared with the w/b ratio = 0.5 show completely similar hysteresis for all energy levels and they have the same trends. The isotherm curve of samples in the same series but with the w/b ratio = 0.6, it presents a slightly different hysteresis in terms of height, which shows the volume absorbed.
The specific surface measurements on the reference materials (
Table 6) clearly show that the increase in the w/b ratio from 0.5 to 0.6 leads to a clear increase in the specific surface from 9.63 to 10.59 m
2/g for the lowest compaction energy (400 kN·m/m
3) and from 9.35 to 11.53 m
2/g for the highest compaction energy (1000 kN·m/m
3). The results show that the effect of compaction on the evolution of the adsorption capacity of the tested mixtures remains visible but negligible compared to the influence of the w/b ratio. The contribution of compaction energy is thus negligible compared to the contribution of hydration.
3.3. Porosity Distribution of Dry Mortar References (Mix 1)
The pore size distribution is derived from the BJH model. The most common classification is defined as the measured pore sizes that range from <2 nm to 300 nm, the microporosity refers to pores smaller than 2 nm, the mesoporosity indicates pores from 2 nm to 50 nm, and macroporosity refer to pores with sizes from 50 nm to 300 nm. According to Baroghel-Bouny [
40], the dominant pore size in a current cement-based hardened paste is approximately 1.7 nm (micropore), which is associated with the pore size of the C-S-H gel, in fact these pores occur during hydration of gels (C-S-H and CH). On the other hand, it also depends on the W/C ratio (water/cement), the type of cement and the duration of cure. For mortars and concretes which contain aggregates, the pores are larger corresponding to sizes of mesopores and macropores.
Figure 4 shows the influence of the water ratio on the modification of the pore size distribution of the reference sample. In Mix 1 with the ratio w/b = 0.5, they have a monomodal pore size distribution with a peak around 30 nm (0.03 µm) which corresponds to mesopores according to the classification above mentioned. According to Feldman [
41], this size has 36% porosity. Feldman has investigated the percentage of porosity of cement-based pastes with different w/b ratio in terms of pore size.
On the other hand, the rise in the w/b ratio = 0.6 generate a multimodal distribution. At least, two populations of pore sizes are present in these samples.
A difference is clearly observed between the 0.5 and 0.6 ratio curves. This difference is illustrated in
Figure 4 with arrows. The dominant size remains more or less in the same order of magnitude. For these two formulations with different amounts of water and whatever the level of compaction energy, the pore distribution curves deviate completely, this phenomenon is clearly illustrated in
Figure 4.
This phenomenon can be explained by the difference in the specific surface related to these formulations which is presented on
Table 6. The formulation prepared with the ratio w/b = 0.6 has a specific surface area greater than that of 0.5. Baroghel-Bouny [
40] has shown that when the water/ cement ratio increases in cementitious pastes, the specific surface area also increases. The author observed that there is a linear relationship between the specific surface area and the degree of hydration. Therefore, in our mixes, when the amount of water increases, the hydration reaction is higher and, in fact, the pores appear to be more present in the structure. In
Table 7, the average pore size is presented according to the BJH method at the time of nitrogen gas adsorption and desorption. It is shown that the pore size of mixtures with a ratio of 0.6 is slightly larger than that observed with the ratio 0.5. This is in agreement with the curves in
Figure 4.
3.4. Adsorption/Desorption Isotherms of Activated Dry Mortars (Mix 2, Mix 3 et Mix 4)
This section present the test results of the samples activated by different activators. These samples are mortars prepared with three types of activators: Na
2CO
3, Na
2O
3Si and CaCl
2. The isothermal curves thus obtained are presented in
Figure 5,
Figure 6 and
Figure 7. The isotherms relating to the samples with the addition of sodium carbonate activator show a clear difference in the volume adsorbed.
As presented in
Figure 5, when 5% to 10% Na
2CO
3 is added, there is a decrease in the volume of N
2 adsorbed from 28 cm
3/g (
Figure 3) to 20 to 23 cm
3/g, respectively. The increase of 5% of Na
2CO
3 leads to a slight increase in the adsorbed volume. For the same formulation and implementation, the increase in the w/b ratio from 0.5 to 0.6 leads to a significant increase in the adsorbed N
2 volume. The increase is 12 cm
3/g for 5% and 10 cm
3/g for 10% Na
2CO
3. However, the hysteresis volume seems to be little impacted by the addition.
This phenomenon is well-explained in the above
Figure 3 for references samples. The “ink bottle” effect is more present for samples prepared with a higher amount of water, see
Figure 5.
This mechanism of interaction is a little different for samples activated by sodium metasilicate. When 6% Na203Si is added for the WSA+GGBBS mixture and a w/b = 0.5 to 0.6, then, an increase in N2 adsorbed volume is observed with very little impact on the hysteresis evolution. This indicates that the kinetics of N2 restitution is very close.
If we compare the mixtures containing WSA+GGBS with 3% and 6% Na2O3Si, for the same w/b:0.5 or 0.6 ratio, increasing the concentration of Na2O3Si leads to a significant decrease in the volume of hysteresis and the volume of adsorbed N2.
A 3% addition of Na2O3Si results in an increase in the volume of the hysteresis compared to the reference mixture Mix 1. Increasing the concentration to 6% leads to a significant decrease in the volume of the hysteresis and the volume of adsorbed N2. This decrease indicates a certain ease of N2 to leave the porous matrix. The amount of adsorbed N2 also decreased drastically. This decrease indicates a decrease in the accessible network volume in the porous matrix.
This information is also coherent with the specific surface area values presented in (see
Table 8). The specific surface area values decrease with the raise of Na
2O
3Si concentration whatever the ratio w/b selected.
Figure 7 shows the curves with very little hysteresis obtained for the samples with the addition of calcium chloride as activator for a ratio w/b = 0.5 (Mix 4).
The increase of 1% CaCl2 leads to a decrease in the volume of adsorbed N2 from 28 to 17 cm3/g. The volume of hysteresis seems to be little impacted. On the other hand, the intensity of the adsorbed N2 volume is six times higher than the volumes measured on the raw materials.
These two curves have a very low adsorption/desorption hysteresis loop compared to the samples activated by the two previous activators. Their specific surface areas are also low according to the data in
Table 8.
All isotherms of Mix 2, Mix 3 and Mix 4 exhibit hysteresis loops at P/P0 > 0.5; this is related to the capillary condensation in the interparticle voids. The linear section corresponding to P/P0 = 0.05–0.2 is due to the sorption of nitrogen on the surface of the mesopores and the external surface of particles, mesopores will be filled at around P/P0 = 0.3–0.4. Then, the second linear section at P/P0 = 0.4–0.05 is due to the sorption of nitrogen on the external surface of particles.
In summary, when 5% to 10% Na2CO3 is added, a decrease in the volume of adsorbed N2 is observed. For the similar formulation and implementation, the increase in the w/b ratio from 0.5 to 0.6 leads to a significant increase in the volume of N2 adsorbed. The increase is 12 cm3/g for 5% and 10 cm3/g for 10% Na2CO3. The volume of hysteresis seems to be little impacted by the addition.
When 6% Na203Si is added, for the mixture of WSA+GGBBS and a w/b ratio = 0.5 to 0.6, then, an increase in adsorbed volume is observed with very little impact on the evolution of hysteresis. For the same ratio w/b:0.5 or 0.6, increasing the concentration of Na2O3Si leads to a significant decrease in the volume of hysteresis and the volume of N2 adsorbed. Increasing 1% CaCl2 results in a decrease in the volume of adsorbed N2 from 28 to 17 cm3/g. Hysteresis seems to be little impacted.
3.5. Porosity Distribution of Activated Dry Mortars (Mix 2, Mix 3 and Mix4)
Figure 8 shows the curves for samples activated with three types of activators and the two w/b ratios. All the curves show a size distribution of the multimodal pores. In
Figure 8a, we observe four pore families for the w/b = 0.6 ratio regardless of the Na
2CO
3 dosage and three pore populations for the w/b = 0.5 ratio. However, for all cases, the dominant pore size is between 0.03 and 0.04 µm; these are mesopores.
According to the total volumes of N
2 adsorbed by the pores (see
Table 9), the sample containing 5% Na
2CO
3 with the ratio of w/b = 0.5 has a low value of adsorption (0.0304 cm
3/g) compared to the other samples in this series with addition of Na
2CO
3. This is in good agreement with the related curve of this sample where the area under the curve is less compared to the others (see
Figure 8a, curve in blue). The area of the curve is representative of the volume of voids in the structure.
According to
Figure 8a, addition of 5% Na
2CO
3 results in a reduction in pore volume 0.20 >
p > 0.035 micron. Increasing the w/b ratio by 0.1 results in an increase in pore volumes in the same range of 0.02 cm
3/g. The behavior of samples treated with Na
2O
3Si is particularly remarkable (
Figure 8b). The addition of the Na
2O
3Si allows for a shift in the peaks of the distribution curve (0.034 micron) towards larger-pore families (0.15 micron). At last, the addition of 1% more of CaCl
2, allows one to lower the pore volume by 0.01 cm
3/g.
To establish a link between the mechanical strength and the adsorption capacity of each sample, it is necessary to relate the values of the BET specific surface area and the values of the difference in volume of adsorption/desorption of the pores. In this series of samples, the one with 5% Na2CO3 and ratio of w/b = 0.6, shows a large difference between the adsorption volume and the pore desorption volume of 0.0201 cm3/ g. Moreover, this sample has a high specific surface area. This suggests a very small pore size and the possibility of gas/liquid retention in the pores, and consequently a high mechanical resistance. This has been demonstrated by measuring the highest compressive strength of this series, where a value of 6.10 MPa was obtained.
In
Figure 8b, the pore distribution graphs for the samples activated by Na
2O
3Si are plotted. The dominant pore size for the dosage of 6% is larger than that with 3%; this size corresponds to 0.1 and 0.03 µm, respectively. The sample activated by 3% Na
2O
3Si has a higher adsorption volume than the others (0.0550 cm
3/g) which corresponds to the volume of voids of material at the time of adsorption. This sample has a fairly high specific surface area, and, this results in a high resistance (compressive strength of 9.02 MPa has been measured). One can conclude that there exist several small pores.
On
Figure 8c, a distribution in size of the multimodal pores is also observed, for which at least three populations of size of the pores are present. The sample activated by 3% CaCl
2 shows a very low adsorption volume which is related to the volume of voids in the material. From the small specific surface area of this sample and the small difference in adsorption/desorption volume, it can be concluded that there are pores that are not necessarily small, but they are in small quantities. In addition, this sample has the highest mechanical strength in this sample series.
The impact of microstructure evolution on UCS evolution is different depending on the nature of each additive.
The addition of Na2CO3 increases the pore volume. The increase in pore volume is accompanied by a drop in the compressive strength UCS. The UCS value is relatively higher (6.10 MPa) with a w/b of 0.6 compared to 0.5 but an average drop of 3 MPa is recorded on the tested formulations.
The addition of Na2O3Si decreases the pore size. For an w/b ratio of 0.5, a decrease in UCS of 2 MPa is observed. An inverse mechanism occurs with a w/b of 0.6. The gain in UCS is 1.5 MPa.
For CaCl2, the highest compressive strength is recorded among the different activators used. The addition of CaCl2 seems to close the pores and act positively on the UCS value. A gain of 1 MPa is recorded.
Several factors can contribute to the modification of hysteresis. The presence of trapped gas and its ability to exit are the most important; however, the contact angle and geometry of the pores also play a non-negligible role. The hysteresis phenomenon is due to the irregular distribution of the size of the pores, which are generally voids of variable shape interconnected by smaller passages and also the “ink bottle” effect retains an amount of fluid/gas not negligible in the pores during desorption. This effect is all the more important as the variation in section in the pore is large. This is the reason why adsorption/desorption isotherms in the form of hysteresis are observed, especially for cementitious materials [
42]. The gas retention capacity of the reference samples with the ratio w/b = 0.6 occurs to be higher than those of the ratio 0.5; the loop is larger than those observed with the ratio 0.5.
This observation is also confirmed by the specific surface area values determined for each sample (see
Table 6). The larger the specific surface area, the more the demand for the quantity of water increases per sample. The values observed are around 11 m
2/g. Then, the more the number of fluid molecules increases, the more the risk of being trapped in the pores increases; thus, the isothermal effect with hysteresis is more present.
The main contribution of the present research work deals with the microstructure modification as the first product of hydration of the binder. The contrasting contribution of microstructure development on the mechanical strength of the material is observed. The different microstructure pattern generated by each mixture and its correlation with the specimen UCS value are intimately linked to the nature of additives and to the hydration mechanism. The mechanical strength of each hydration product depends also to the mechanical properties of the pores structure generated. Then, it could be an interesting issue to provide further information on the elastic properties of the pore matrix generated by each additive: binder and activators.