A previous study (47) comprises a research program designed to assess the coarse RCA effect on RAC resistance to severe sulfate exposure. With this purpose, four concretes were designed with durability criteria (low w/c and appropriate air entrainment) and different coarse RCA contents. Then, prisms were moulded and half-buried in sulfate-rich soil. These exposure conditions simulate the exposure of ground-level structural elements, where concrete is in contact with sulfate-laden soils and variable moisture and temperature conditions.

Table 1 summarizes the most relevant results of this previous study. In the fresh state, similar concrete slump and a slight variation in air-entrained content with the increment in RCA content is observed. In the hardened state, similar compressive strength for all concretes is observed. Regarding the RACs performance against ESA, after 200 days and 3000 days of exposure, a negligible effect of the RCA content is observed, either by weight variation and dynamic modulus of elasticity. Visual inspection confirms these experimental results and suggests that the new cement matrix of RAC was logically able to mitigate any possible adverse effect of RCA particles embedded in it.

At the same time that prisms for rich-sulfate soil exposure were casted, other prisms were moulded to be immersed in sulfate solution, i.e., in a saturated condition. After 10 years of exposure these specimens were evaluated to assess the RCA influence on solely external sulfate attack process. This paper presents an extended profile assessment of the ESA on these RACs by multiple instrumental analyses.

The used binder was a moderately sulfate resisting blended Portland cement, according to IRAM 50001:2010 (48), containing 12% w/w blast furnace slag and 18% w/w limestone filler. This cement is comparable in composition to a CEM II/B-M (L-S) 42.5 N according to EN 197-1 (49) produced with a sulfate resisting clinker. Natural aggregates were two river siliceous sands, a fine sand (FS) and a coarse sand (CS). Also, two crushed granite aggregates (CG) with maximum sizes of 19 and 25 mm were used. Recycled concrete aggregate (RCA) was obtained from crushed conventional concretes with variable and unknown characteristics and the only common features being that, all of them corresponded to structural concrete (with compressive strength above 21 MPa) and contained crushed granite as coarse aggregate. The crushing of concretes was carried out by two size lab-scale sized jaw crushers. The maximum size of RCA was 25 mm. Particle size distribution was determined following ASTM C136 (50). To obtain coarse and fine aggregates that satisfy the limits established by ASTM C33 (51), FS and CS where combined in proportions of 20%-80%, respectively, of the total fine aggregate, and CG-19 and CG-25 in proportions of 20%-80%, respectively, of the total coarse aggregate. Figure 2 shows the particle size distributions of the resulting fine and coarse natural aggregates, RCA, and limits established by ASTM C33. In Table 2, evaluated properties of all aggregates are presented, including maximum size and fineness modulus, density in saturated surface-dry condition and water absorption capacity (ASTM C127) (52), content of material finer than 75 µm (ASTM C117) (53), and Los Angeles abrasion loss (ASTM C131) (54).

With the aim of knowing the effects of different contents of coarse RCA on ESA resistance, concrete mixes were designed with 0, 50, 75 and 100% of replacement v/v of NA by RCA. In addition, to comply with durability criteria for sulfate-laden environments, and determine any detrimental effect of the replacement of NCA by RCA on the performance of concrete, mixes were designed with a water/cement ratio of 0,35 and air-entrainment admixture was used. Table 3 presents the proportions of mixes, where the designation of each concrete mix corresponds with its RCA content. To avoid a variation in the effective water/cement ratio of fresh concrete, RCA was pre-saturated in water and then superficially dried. Thus, RCA were used in saturated and surface dry condition (SSD).

For sulfate testing, 75 × 150 × 250 mm prisms were cast. To evaluate transport properties of RACs, 75 × 75 x 300 mm prisms for sorptivity (according to IRAM 1871 (55)), and 200 mm cubes for water penetration under pressure tests (according to IRAM 1554 (56), equivalent to EN 12390-8 (57)), were also cast. All samples were cured for 28 days in a humid chamber. Before testing, 75 × 75 × 300 mm prisms were sawed to obtain 50 × 75 × 75 mm prismatic specimens to evaluate the sorptivity of concrete. Moreover, density, porosity, and water absorption were evaluated with the remaining parts of the specimens, following ASTM C642 (58).

Prismatic specimens of 75 × 150 × 250 mm were covered with waterproof paint on all faces excepting the casting face of 75 × 250 mm. These specimens were submerged in 50 g/l sodium sulfate solution maintained at 23±3°C, with a solution/concrete volume ratio of 4 and without periodic solution refreshing. After 10 years of exposure, sulfate ingress was assessed in portions of the specimens by collecting powder samples by dry progressive profiling with a diamond tool, at increasing depths from the exposure face of 0-3 mm, 3-6 mm, 6-10 mm, 10-15 mm, 15-20 mm, and 20-30 mm. The powder samples were collected simultaneously as the tool was advancing. The sampling area in all cases was larger than 20 cm². Then, RAC powder samples were homogenized by grinding them with mortar and pestle up to all the material passed through a 75 µm sieve.

Chemical analysis (CA), X-ray fluorescence (XRF), and TGA/DTA were performed on fractions of the powder samples for obtaining the chemical and mineralogical compositions at different depths. Chemical compositions of the powder samples were determined by wet analysis and X-ray fluorescence (XRF). For wet analysis, the sample was dissolved with hydrochloric acid, and the insoluble residue was eliminated by filtration. Then, sulfate was precipitated as barium sulfate after the addition of barium chloride solution. The precipitate was collected by filtration, calcined, and weighed to compute sulfate contents expressed as SO₃. XRF analyses were made with an Ametek Spectro IQ II. TGA/DTAs were conducted in a previously ground and homogenised samples of 0.05 g, at a heating rate of 10°C/min (59), up to 1050°C under an inert atmosphere of N₂.

To analyse the damage in the specimens, 75 mm × 150 mm slices were taken for microscopic observations. The sampling direction was such that the whole cracking pattern from the surface to the inner layers could be observed. These samples were impregnated with fluorescent resin (epoxy resin with sodium fluorescein) and polished to obtain thin slices. Images were taken at 1.5, 4.5, 8, and 25 mm, covering an area of 2.7 × 14 mm on each depth.

Table 4 presents physical properties of RACs. In accordance with the literature, water absorption and porosity of RAC increase with a higher RCA content (7,10,13,14). This effect is connected with the porosity of the attached mortar in RCA particles. Density of RACs decreases with RCA content, due to the higher porosity of recycled aggregates than natural aggregates.

Transport properties are also presented in Table 4. Sorptivity values increase with the RCA content. This suggests that the sorptivity coefficient is affected by all the pore range of RAC resulting from the summation of the attached mortar and the new cement matrix. However sorptivity have a sligth slope from R-75 to R-100, which suggest that the effect of RCA content could be overlaped by differences in the new cement matrix. Compressive strength and air-entrained content values for R-100 concrete suggest a relative denser matrix than other RACs.

Conversely, the average water penetration under pressure was relatively low for all concretes (below 30 mm, which is the maximum limit value accepted for concrete exposed to a high sulfate-laden environment by Argentine Regulations (60)), with a slight decrease of the water penetration when the RCA content increased. In this case, the higher porosity of RAC caused by the attached mortar on RCA particles showed no adverse effect on the water penetrability under pressure. These results may appear inconsistent with the overall porosity values of each concrete mix. However, considering that only coarse RCA was used, the influence of this constituent on the porosity of the most external layer of 15-20 mm is normally limited, and this seems to be reflected by the results of water penetration under pressure.

Figure 3 shows the pictures of specimen 75 × 150 mm cross-section, with the ingress face on the top for each picture. After 10 years of exposure in high concentration sodium sulfate solution, all specimens seems to have a good state, with no visual signs of deterioration, even for the specimens with 100% v/v RCA content. This is in accordance with the results of the previous study (47), where no detrimental effect of RCA content on the performance of recycled concretes specimens semi- buried in sulfate soil was observed.

The evaluation of the damage in RACs is presented in Figure 4, corresponding to the images from epifluorescent optical microscopy taken from the four concrete specimens. From left to right, each image shows the layers from the surface to the bulk, centered at 1.5, 4.5, 8 and 25 mm from the surface. A common micro-cracking pattern is observed for all mixes at all depths, with a morphology non-related to ESA deterioration mechanisms reported in bibliography. This low level of micro-cracking can be connected with the affectation of samples due to drying or the preparation of thin slices.

More importantly, cracking development is notorious for the first depth for R-0 and R-100 and it is present over the entire field of these images. The morphology of these cracks, which develop edging natural aggregate particles, is parallel to the surface and consistent with damage due to ESA reported in bibliography (61–63). For R-50, small cracks are observed only at the top of the image, whereas for R-75 no cracking that could be associated with ESA was detected. These results suggest that, although it is possible that sulfate penetrated into the concrete matrix and precipitated as ESA products, the developed internal stresses were not enough to exceed the RACs tensile strength levels, with exception of the first layers. Thereby, a good performance of all RACs against ESA regardless of the RCA content is observed. With small disparities, microscopic observations show a better behaviour of R-75, followed by R-50, and the worse performance corresponding to R-0 and R-100 concretes. This ambiguous effect of RCA on RAC resistance has been previously reported in bibliography (31,34,38).

Results of wet CA, XRF, and average values between both are presented in Figure 5 for each concrete mix. The reported SO₃ contents are the summation of the sulfate contained in the constituting cement and attached mortar in RCA, the sulfate contained in ESA products, and the free sulfate in the pore solution at the moment of sampling (which precipitated as salt crystals when specimens were dried). The second depth (3 to 6 mm) has a higher sulfate content than the surface layer (0 to 3 mm) for all concretes. This outcome is in agreement with other studies on ESA mechanisms (20,26,63), which is usually explained by the presence of a densification zone in which a large amount of ESA products precipitate and reduce porosity. In Figure 5, a slight increment in the sulfate content at the second depth with increasing RCA content (from 1.63% for R-0 to 2.18% for R-100) is noted, but this increment is not proportional to the RCA content (2.10% and 1.93% for R-50 and R-75 respectively) (Table 5). Although each pair of full penetration profiles determined by both analytical techniques are consistent, these profiles are not in relation with the RCA content in concrete. The total amount of incoming sulfate can be computed as the integral of the SO₃ contents over the depth. The results in Table 5 show that R-0 and R50 have the smallest amount of incoming sulfate, followed by R-100 with an intermediate value, while R-75 has the highest content (about a 30% more SO₃ content than control concrete).

Figure 6 presents the results of thermogravimetry (TG). Each subfigure presents the samples at different depths for each concrete mix, referenced with increasing numbers from 1 to 6 for the increasing depths (0-3 mm, 3-6 mm, 6-10 mm, 10-15 mm, 15-20 mm, 20-30 mm). The weight loss at a certain temperature is certainly originated by the summation of the contributions of more than one compound, but the presence of different compounds can be inferred from the identification of peaks in the DTG pattern which show characteristic weight loss temperature for different compounds (64). Therefore, the observed peaks in the derivative TG patterns (DTG) indicate the presence of free water, ettringite, gypsum, portlandite, calcite, and decomposition of sulfates, at approximate temperatures of 50, 100, 120, 420, 650, and over 750°C, respectively. These compounds are indicated in Figure 6 for identification purposes only and quantitative analysis was done only for calcite as explained below. The decomposition of sulfates usually takes place at much higher temperatures, but the mix with compounds like alumina or silica seems to reduce this temperature quite significantly (65).

Figure 7 presents the profiles for the ratios between the calcium contained in calcite (determined by integration of the corresponding peaks in the DTG patterns applying the tangent method for discounting the contribution of other phases (64)), and the total calcium content (determined by XRF). All samples show noticeable calcite contents as all mixes were prepared with a limestone blended cement. Additional calcite is provided by partially carbonated cement paste contained in RCA, and, therefore, R-50, R-75, and R-100 show an increased calcite content in relation to R-0. However, the most external layers of all series contained even higher calcite contents and also depletion of portlandite. This is indicative of the carbonation of these samples. Portlandite is also consumed with the formation of gypsum, which is also mostly detected for the outer layers. The portlandite depletion and the reduction in pH due to carbonation reduce the stability of C-S-H and contribute to the progression of its decalcification (66–68). The significant amount of calcite noted for all concretes in the most external depth is higher than the potential maximum amount corresponding to the full carbonation of portlandite. After complete portlandite depletion to form calcite, gypsum, and ettringite, additional Ca⁺⁺ ions are made available for calcite formation by progressive decalcification of the C-S-H phase. This process is sometimes regarded as carbonation of C-S-H, but the contribution of sulfate to the progress of the process should also be taken into account.

Regarding ESA products, on the one hand, ettringite was identified in all concretes at all depths. This situation evidences that sulfate ions penetrated into the concrete matrix and precipitated in combination with C₃A hydrates. The precipitation of products and the associated internal stresses seem to be insufficient to overcome the tensile strength of RACs. R-0 and R-75 present the highest amounts of ettringite on surface layers. On the other hand, gypsum appears only near the surface, except in the case of R-50 for which significant amounts of gypsum were found for all layers. At first, it is expected that ettringite would evolve into gypsum, as the latter is a more thermodynamically stable phase at high SO₃ concentrations. However, ettringite is possibly stabilized due to a high concentration of sodium (69,70), so both phases are found simultaneously even after long periods of exposure and after the development of cracking. Finally, some free sulfate is suggested by thermogravimetry in some samples, as precipitates of mirabilite or thenardite during the drying process.

Although all concretes mixes showed an acceptable performance (little development of cracking) and small differences regarding the amounts of bound sulfate, the mineralogical profiles show that the ESA mechanism was not the same for all the specimens. For R-0, ettringite is observed over the entire profile, whereas gypsum is present only near the surface, in coincidence with the cracking detected by microscopy. For R-50, gypsum is observed throughout the profile, which suggests a process of conversion of ettringite into gypsum. As no significant cracking is observed in this concrete, it is possible that the porosity of the matrix allowed the precipitation of formation products with no generation of internal stresses. In the case of R-75, only a small amount of gypsum is observed near the surface and ettringite in all depths. R-75 is actually more porous than R-50, but the results from the sulfate penetration profile revealed a higher apparent diffusivity in R-50 than in R75, and, therefore, more significant precipitation of gypsum is expected in the former. In these two concretes, precipitation of formation products is confirmed, but this is apparently insufficient for the development of cracking due to a low internal restriction to deformation with respect to control concrete. For R-100, ettringite is observed at all the analyzed depths in superposition with significant contents of free sulfate. In this case, cracking seems to allow faster sulfate penetration, which was not bound as gypsum at the moment of sampling. The reason for this situation is not totally clear. The thermogravimetry patterns also show a bump near 300 °C, which can be attributed to the presence of strätlingite (which is similar to AFm but with a specific silica content). The differences in the mineralogical outcome of the analyses describe slightly different mechanisms in the deterioration process by ESA occurring in each concrete, which can be connected with the different internal restriction of concretes in each mix.

The R-0 mix had the lowest porosity among the studied mixes. Cracking and elevated gypsum content near the surface was consistently detected. The SO₃ content reaching the bulk is lower than in the other mixes in connection with its reduced pore structure. The higher internal restriction to deform makes this concrete more prone to the development of internal stresses and cracks than the other series, for the same amount of reaction products being formed. The observation of significant cracking confirms this situation, and partially explains certain acceleration in the sulfate penetration rate and high contents of ettringite in depth. The portlandite contents at high depths are consistent with stable hydration products despite the exposure to sulfate (third zone from the surface in Figure 1).

For the R-50 and R-75, results are mostly in relation with the resulting porosity of each concrete. The increased porosity due to the inclusion of RCA is reflected in higher SO₃ contents near the surface, but the situation becomes more complex for the deeper layers. Even though R-50 shows lower SO₃ contents in internal layers, a high content of gypsum was also found. Conversely, R-75 shows higher contents of SO₃ in the inner layers than R-50 and a high content of ettringite near the surface. The presence of unaltered portlandite in the depth confirms a reduction in the transport rate as a consequence of precipitation of formation products. The porosity of the material seems to show an optimal balance between the sulfate penetration rate and the product formation rate. This is in agreement with other studies where the formation of ESA products seems to have a beneficial effect in some mixes, where porosity is being reduced and internal stresses does not reach the tensile strength of concrete (34,45). With additional ingress of sulfate, it is expected that internal stresses will continue increasing and cracking will progress. Whereas this situation seems to be the case for R-100, which demonstrated a much faster sulfate penetration rate, R-50 apparently needed less incoming sulfate for development of cracking due to its comparatively low porosity and greater internal restriction to deform.

With the highest content of RCA, R-100 shows significant development of cracking at the surface, similarly to R-0 in the opposite situation regarding RCA content. In this concrete, significant content of ettringite at middle depths and high contents of gypsum and calcite near the surface were found. Significant amounts of free sulfate were also found, suggesting fast ingress of sulfate due to the formation of cracks. The benefit of the lower internal restriction due to the porosity of R-100 seems to be less important than the increase in the sulfate penetration rate. As large amounts of calcite are also found, there seems to be a competition for the consumption of portlandite, and being the sulfate the last arriving, the precipitation of gypsum would only be possible by decalcification of C-S-H, which is a less direct process.

Moreover, the balance of the effects of increased porosity, decreased internal restriction of concrete to deform, and increased sulfate penetration rate, could depends on the particular features of the test, either mix properties (e.g. porosity of aggregate, w/c of the new matrix) or exposure conditions (e.g. moisture conditions, sulfate concentration, etc.). For example, the complexity of the impact of highly porous aggregates is exemplified by the work of Lee et al (30), where mortar containing variable contents of fine recycled aggregate was exposed in high concentration sodium sulfate solution. Replacements ratios of 25 and 50% resulted in lower expansion than the one of control mortar, while 75 and 100% replacement ratios resulted in more substantial expansion than in the control mortar. In this case, 50% was the optimal fraction of recycled aggregate to minimize bars deformation under that ESA conditions.

Regarding the transport properties assessed in RACs, Figure 8 shows the relationship between the incoming sulfate and both transport properties assessed in RACs (sorptivity and depth of water penetration under pressure). Sorptivity seems to correspond better with the performance of RACs against this condition of ESA than the water penetration depth. As mentioned above, sorptivity seems to be related to the porosity of the new cement matrix besides the RCA content.