Mortars were prepared with a natural hydraulic lime NHL3.5 (1) in the form of powder, whose commercial name is i.design CALIX 3.5, produced (and supplied) by the FYM-Italcementi enterprise (Malaga, Spain, www.fym.es/ES).

Two natural aggregates with continuous grading were used: a standardized siliceous sand (SA) with a grading of 0.1 < φ < 2 mm (10) of the Eduardo Torroja Institute (CSIC, Madrid, Spain) and a crushing calcareous aggregate (CA, 0 < φ < 6 mm) from the province of Granada.

The major and trace elements of the lime and the aggregate were analysed by means of a S4 Pioneer-Bruker X-ray fluorescence (XRF) spectrometer with wavelength dispersion, equipped with Rh X-ray tube (60 kV, 150 mA) and LIF200/PET/OVO-55 crystals. Samples were ground and the powder dispersed in KBr pellets before the analysis. The identification and semi-quantification of mineral phases was performed by X-ray diffraction (XRD, disoriented powder method), using a Panalytical X′Pert PRO MPD diffractometer, with automatic loader. Experimental conditions were: 45 kV voltage and 40 mA current intensity; CuKα radiation (λ≈1.5405 Ȧ); 3 to 70° 2θ explored area; 0.01° 2θ/s goniometer speed. Mineral phase identification was carried out using the X-Powder software package (11).

Table 1 shows the chemical and mineralogical composition of the hydraulic lime (NHL3.5) and the calcareous aggregate (CA). The natural hydraulic lime is mainly composed of portlandite and calcium (CaSiO₃) and di-calcium (Ca₂SiO₄, belite) silicates. Also a small amount of calcite was detected in the lime, as a result of a slight carbonation of portlandite, probably occurred due to air exposure during lime storing and preparation. The mineral composition of the calcareous aggregate (mainly calcite and, in minor proportion, quartz and dolomite) is typical of the limestone extracted from the Sierra Elvira in Granada (12). It is fundamental to control the quantity of dolomite present in the aggregate, since this mineral phase can react with sulphates, forming soluble salts, harmful for the mortar and the other materials in the building (13). The chemical elements identified agree with the described mineralogy, being CaO the main oxide, in a quantity higher than 60% by weight. The siliceous aggregate (SA) has a SiO₂ content ≥98% (10).

Mortar samples were prepared following the UNE-EN 1015-2 (14) standard, with some variations detailed hereinafter. Two types of natural hydraulic lime mortar were produced with siliceous standard sand (NHL-SA) and calcareous crushing aggregate (NHL-CA). These mortars were prepared with a 1:3 binder-to-aggregate dosage by volume, being this one of the most used (15, 16), whilst the amount of adding water was chosen according to the mortar consistency. Mortar binder-to-aggregate and water-to-binder dosages, flow values (17), penetration, trapped air, water shrinkage and bulk density (18) in the fresh state are shown in Table 2. After mixing (14), mortars were casted in standardized moulds (40×40×160 mm) and cured for 7 days in the mould and the following 21 days out of the mould (15), under controlled temperature (T = 20±5 °C) and relative humidity (RH = 60±5% and 90±5%).

The mineralogy of mortars, determined by means of XRD, and the percentage of each phase, calculated by means of XRD and TGA, are indicated in Table 3. The type of aggregate used influences the mineral composition. Approximately 80% of quartz was detected in NHL-SA mortars and a slightly higher quantity of calcite (∼86%) in NHL-CA mortars, in addition to a small quantity of dolomite (Table 3) due to the presence of this phase in the limestone from which the CA aggregate proceeds (Table 1). Moreover, part of the calcite found in NHL-CA mortars proceeds not only from the aggregate but also from the carbonation of portlandite.

Although the two types of mortar were prepared with the same dosage by volume (1:3), converted in weight proportion, this corresponds to a lime-to-aggregate dosage of 1:6 in NHL-CA mortar and of 1:3 in the NHL-SA one (Table 2). This means that the initial amount of lime is approximately 14% wt. in NHL-CA and 25% wt. in NHL-SA on the total mass of samples, respectively. As a consequence, for each mortar type, half of this amount corresponds to the content of portlandite and the other half to the content of calcium silicates (each phase is present as a 7% in NHL-CA and a 13% in NHL-SA mortars, considering the XRD results, Table 1). Taking into account these initial amounts in the two types of mortars, it is possible to estimate towards which process (carbonation or hydration) and in which conditions of relative humidity the two mortar types are more reactive, calculating the percentage of portlandite and calcium silicates that have reacted. In hydraulic lime mortars, one part of the lime (the calcium hydroxide) reacts with the atmospheric CO₂ in the carbonation process that leads to the formation of calcite, whilst the other part (the calcium silicates) reacts with water, leading to the formation of hydrated phases. A bigger consumption of Ca(OH)₂ with respect to the calcium silicates indicates a higher reactivity of the matrix towards the carbonation process and, vice versa, towards hydration.

The approximate amounts of portlandite (CH) and calcium silicates (CS) that have reacted in the two types of mortars cured at 60% and 90% of relative humidity, calculated by means of TGA, are indicated in Table 3. Notwithstanding, it is important to take into account that, during the hydration of calcium silicates, calcium hydroxide is formed and this phase can either increase the portlandite content or the calcite one (if it carbonates). This percentage of newly-formed calcium hydroxide has not been taken into consideration in the quantitative analysis, due to the difficulty of determining it exactly. As a consequence, the amount of portlandite (CH) in Table 3 has been given approximately, as a range of values.

In NHL-SA mortars, the external zone is more reactive towards the carbonation process compared to the interior of samples, in both mortars cured at 60% and 90% of RH. In NHL-CA mortars, instead, carbonation occurred with similar velocity in both zones (almost the same amount of portlandite reacted in the core and the surface of mortar). Generally, carbonation is slower at 90% of relative humidity (a lower percentage of portlandite reacts) and this difference is evident especially in NHL-CA mortars.

It is worth considering that the quantification of hydraulic phases (calcium and di-calcium silicates and hydrated calcium silicates) by means of XRD is not accurate, since the amount of these phases is often lower than the resolution limit of this technique (3–5% (19)). For this reason, it is advisable to use complementary techniques for their determination. The formation of hydrated phases and, consequently, the transformation of calcium silicates, was evaluated comparing the height of the peaks obtained from the thermal decomposition of the calcium silicate hydrates (C-S-H), which occurs between 110 and 225 °C (20) in the thermogravimetric curves (Figures 1 and 2). In these curves, the differential variation of weight due to the thermal decomposition of C-S-H (between 110 and 225 °C), portlandite (CH, between 300 and 600 °C) and calcite (CC, between 550 and 955 °C), was represented as a function of temperature. The TG curves reflect the differences in the quantity of the three detected phases (C-S-H, CH and CC) depending on the type of aggregate (calcareous or siliceous), the analysed zone (exterior or interior) and the relative humidity (60% or 90%).

It is worth highlighting that the TG curves are also useful to understand the contribution of the newly-formed Ca(OH)₂, since its content is proportional to the signal of the C-S-H. In all mortars cured at 60% of RH this contribution is very similar both between the external and internal part of each mortar and between the two mortar types. In mortars cured at 90% of RH, instead, a larger peak corresponding to the C-S-H phases is detected in the internal sample of NHL-SA mortar. This may indicate a slightly greater content of newly-formed Ca(OH)₂. The peak corresponding to the dehydroxilation of Ca(OH)₂, in fact, is larger in this internal sample compared to the external one but it is not possible to know whether this is an effect of only the greater hydration or also a lower carbonation. However, if more calcium hydroxide was formed due to the CS hydration, this has not significantly contributed to the increase of calcium carbonate content in the mortar (as observed from the height of the peak corresponding to the CaCO₃ decomposition in NHL-SA_90 IN, Figure 2).

In mortars cured at 60% (Figure 1), the height of the peak of C-S-H is very similar in both types of mortars, as well as in the external and internal zones, which indicates that a similar quantity of hydrated phases formed. The height of the same peak is larger in mortars cured at 90% of RH (Figure 2), as hydration reactions are more favoured at higher relative humidity. Notwithstanding, small differences in the height of the peak of C-S-H were found between the two types of mortars. The internal zone of NHL-SA mortars cured at 90% of RH has a larger amount of hydrated phases compared to the external one (Figure 2). This is due to a higher humidity (water content) in the interior of samples, which favours the development of hydrated phases more than in the exterior, where water evaporates easier (due to exposure to air). In NHL-CA mortars cured at the same relative humidity, instead, the C-S-H amount is the same in the exterior and the interior of samples (Figure 2).

From the observations on the amount of portlandite, calcium silicates and calcium silicate hydrates in the two types of mortars, at different relative humidity conditions, it is possible to state that:

the carbonation process is faster at 60% than at 90% of RH, whilst the contrary is observed for the hydration reaction;

The morphology of the formed mineral phases and the texture of mortars were observed by means of FESEM.

In NHL-SA mortars a larger amount of hydrated phases is present (needle-like C-S-H, Figure 3a–c) compared to NHL-CA mortars (Figure 3d–f), especially in those cured at 60% of RH. In the latter, calcite crystals with different size and morphology were observed (Figure 3f and g). It is possible to recognize nanometric (with size smaller than 500 nm, Figure 3e and f) and micrometric (1–3 µm, Figure 3f) calcite crystals with rhombohedral habit. In NHL-CA mortars, it is possible to see also portlandite crystals, untransformed or in process of transformation, and needle-like crystals of C-S-H (Figure 3f). Figure 3 also shows the EDX analysis carried out in some areas of the observed samples. The detection of Si in the first spectrum (Figure 3c) has to be attributed mainly to the quartz grain of the siliceous aggregate located just below the C-S-H needles.

The matrix of NHL-SA mortars presents a higher porosity compared to that of NHL-CA, at all values of RH, as observed by FESEM (Figure 3a) and determined by means of MIP (P_o ∼27%, Table 4). NHL-CA mortars present a compact texture (Figure 3d) and a porosity of approximately 25% (P_o, Table 4). In NHL-SA mortars, instead, the increase of relative humidity leads to a slight decrease in the total porosity (around 2% less), due to the formation of a larger quantity of hydrated phases in the internal zone of this mortar.

The real and bulk densities are slightly higher in NHL-CA mortars (Table 4), due to the different density of the minerals constituting the two types of mortars (D_{r calcite}=2.71 g/cm³; D_{r dolomite}=2.84 g/cm³; D_{r quartz}=2.62 g/cm³).

In the pore size distribution curves (Figure 4) two families of pores are observed: the main one, including pores with radii comprised between 0.05 and 3 µm, and the second smaller one, with pores of radii comprised between 0.005 and 0.05 µm. Both types of pores are mostly present in the matrix of mortars. Their volume is higher in NHL-SA mortars (Figure 4a), and this agrees with the higher values of open total porosity measured in this mortar compared to the NHL-CA one (Table 4). This is due, on the one hand, to the presence of a more porous matrix (Figure 3a) and, on the other hand, to the formation of a larger amount of C-S-H, which create a higher volume of micropores of small size (Figure 3b).

In mortars cured at 90% of RH, both the pore size (the main peak has moved between 0.1 and 1 µm, Figure 4) and the pore volume are reduced, due to the formation of a larger amount of hydrated phases, as observed by means of FESEM and determined by TGA (Table 3).

Despite the higher porosity of the matrix of NHL-SA mortar, a better adhesion exists between the matrix and the aggregate grains (Figure 5a), with respect to NHL-CA mortar (Figure 5b). The porosity in the contact zone between the matrix of NHL-CA and the calcareous grains of biggest size (6 mm) can be observed in the highlighted zone in Figure 5b; this contact is more evident than in NHL-SA mortar. According to MIP analysis, those pores have a radius comprised between 3 and 20 µm (Figure 4) and their volume, indeed, is higher in NHL-CA mortar samples (Figure 4b).

Values of flexural and compressive strength of the two mortar types cured at 60% and 90% of relative humidity for 28 days are presented in Table 4. NHL-SA mortars present higher values of compressive strength and similar values of flexural strength compared to NHL-CA mortars. The higher compressive strength of the former is especially related to the higher hardness of the quartz crystals and to the better adhesion between matrix and aggregate, as observed by FESEM (Figure 5).

As expected, due to the larger formation of hydrated phases and the lower porosity, mortars cured at 90% of RH present the highest values of R_c, as already observed by other authors (21). However, the measured R_c values do not reach 3 MPa, value that is under the expected one in mortars made with NHL3.5 and exposed to 90% of relative humidity (22). This is due to the fact that the mechanical properties of a mortar do not depend only on the class of the lime (in this case 3.5) but also on other factors, such as aggregate type, water-to-binder and binder-to-aggregate dosages, among others. In particular, mortars prepared with calcareous aggregate do not reach the expected values of mechanical strength due to the porosity of the aggregate, which increases the water demand during mixing (see the water-to-binder dosage in Table 2).

After the flexural rupture, phenolphthalein was sprayed on the broken surfaces to observe the progress of carbonation in mortars. The carbonation edge was approximately 1 mm in both types of mortars cured at 60% of RH, whilst those cured at 90% of RH did not present a well-defined carbonation edge (Table 4, Figure 6). The hydrated phases formed have an alkaline pH, therefore phenolphthalein keeps its pink colour and it is not possible to distinguish them clearly, although these phases are present in the mortar, as observed by means of FESEM. This demonstrates that hydration of hydraulic lime is the predominant process at 90% of RH and in the interior of samples (which shows a pink colour, Figure 6), whilst carbonation is more favoured at lower relative humidity and in the exterior of samples (which shows a neutral colour, Figure 6).

The way in which samples break during compressive rupture is characteristics because, under a compressive stress perpendicular to the compaction plane, the first layers that come off the mortar are the most superficial ones (Figure 7). This indicates that there exists a discontinuity in the compressive strength between the external zone, more carbonated, and the internal one, richer in hydrated phases (23, 24). NHL-CA mortars are less resistant to compression compared to NHL-SA ones, to such extent that the pressure exerted gives place to a partial breaking down of samples, especially in the contact zone between the matrix and the biggest aggregate grains (Figure 7).

Mortars cured at 90% of relative humidity present higher values of ultrasound velocity than those cured at 60% of RH (Table 4). The former, indeed, have lower porosity, as indicated by MIP analyses (Table 4). It is worth mentioning that in NHL-SA mortars it was possible to measure the propagation velocity of ultrasounds along the longest edge (16 cm) of the samples, parallel to the compaction plane. The velocity measured in this direction was 608 m/s, very low compared to the other measured values (between 1400 and 2050 m/s, Table 4). In the rest of samples, instead, this measurement did not provide any value, since the velocity of waves was under the detection limit of the transducers used. The fact that velocity along the longest edge was measured only in NHL-SA samples and that this mortar presents the highest V_p values, once again depends on the better adhesion between matrix and aggregate with respect to NHL-CA mortar. The velocity of propagation of ultrasounds through a mortar decreases and the spatial attenuation increases considerably when the wave propagates from an aggregate grain to the matrix, due to differences of mineralogy and, especially, porosity (25). Moreover, if the wave finds a void between the grain and the matrix its velocity decreases more, since it is travelling from a solid medium of higher density (the aggregate grain) through a gaseous medium (the pore), and finally, to a solid medium of lower density (the matrix), as it occurs in NHL-CA mortars.

The chromatic parameters L*, a*, b*, W and Y of mortars are presented in Table 4. The mortars with higher lightness are NHL-CA cured at 60% and NHL-SA cured at 90% of RH (in both, L* ∼83). NHL-CA mortars become slightly darker at a condition of high relative humidity (L* y W at 90% of RH, Table 4). The presence in this mortar of calcareous aggregate grains of dark grey colour gives a darker appearance to samples, compared to NHL-SA. In the latter, the smoother rounded texture of the siliceous grains and the higher reflectivity of the quartz crystals produce a higher lightness in the mortar. a* and b* values indicate a slight tendency of the mortar colour toward yellow, especially in NHL-CA, as also the yellow index indicates (Y, Table 4).