Two sets of 15 beams, manufactured with a geometry defined by a square cross section with sides of 150mm and a length of 600mm, were subjected to accelerated contamination by sulfate and chloride ions. Each set corresponded to a different mix as characterized in Table 1 (concretes designated as C2 and C3), but both batches were made using the same cement type, as before. The introduction of 35% more fly ashes and 37.5% more cement, per cubic meter, in the mix of C3 versus that of C2, led to the reverse decrease of 20% of sand as customary in the industry of construction, leading to a concrete C3 with higher strength and lower porosity (7.7% vs. 10.6% as detailed below) than those found for C2.

Each beam was reinforced with 2 longitudinal rebars. The diameters differed from beam to beam as shown in Table 2 that refers only to the contaminated beams, 6 more having been built for control, two per each bar diameter. To further study the influence of cracking in the contamination process, some of the beams were pre-cracked before the beginning of the contamination protocol. This was achieved loading those beams with a concentrated force at mid-span that generated the cracking bending moment. For each concrete, and for each rebar size, five beams were cast (one reference beam not to be contaminated, two non-cracked and two pre-cracked to be subjected to contamination). The concrete cover clearance of the longitudinal reinforcement was 20mm.

Concrete cubes (150mm side) were also cast along with the beams and were used to assess the concrete strength. The results for the concrete compressive strength (f_ccm) at 28 days were 46.8MPa for C2, and 50.7 MPa for C3. The values for C2 and C3 were tested again after 850 days and the compressive strength found was higher, averaging, respectively, 64.3MPa and 70.5MPa.

The tensile strength properties of the reinforcement steel bars ϕ6, ϕ8 and ϕ10 were obtained from tensile tests and the values found for the yield strength (f_y) were, respectively, 523 MPa, 550 MPa and 542 MPa. For the ultimate strength (f_t), in the same order, 618 MPa, 697 MPa and 669 MPa.

The contamination by sulfate and chloride ions was started after concrete was cured for 28 days at 20 °C and 90% RH (23). A small acrylic container was installed having the bottom on the tensile side of the beam, as shown in Figure 2. Inside the container a thin sponge was positioned under a stainless steel mesh, and connected to a copper conductor. After assemblage, the container was partially filled with a solution of sodium sulfate and sodium chloride at 5% and 3% by weight, respectively.

At mid-height and mid-length of each of the beams to be contaminated, a grid of A304 stainless steel was placed, connected to a copper wire linked to an external circuit, as sketched also in Figure 2. Both conductors were connected to a DC voltage source (Sorensen, XHR 40–25). A current density between 2 and 4mA/cm² was applied between the stainless steel meshes to generate a constant electric field that accelerated contamination of the concrete by the ions contained in the solution. The contamination process was conducted during a 9 week period on the beams cast with the concrete mix of smaller strength (C2) and 12 weeks for the beams made with the stronger concrete (C3). The reason for the different times adopted was the difference of porosity between C2 and C3. Every 3 or 4 weeks, concrete powder samples were obtained by drilling 10mm to 20mm deep into the beams, the distances bring measured from the top surfaces. The determination of the sulfates percentage in the concrete powder followed (24, 25).

The level of alteration on the concrete microstructure and any eventual formation of major mineralogical phases were studied via X-ray diffraction analysis and mercury intrusion porosimetry performed both in plain and contaminated concrete.

The amounts of sulfate present in the samples obtained at depths of 20mm and 40mm, after 1000h of contamination in the SO₂ chamber, were respectively 0.58% and 0.40%. At 2000h the values changed to 1.20% and 0.64%. The SO₂ beams contamination is expressed by: ion sulfate massconcrete mass percentage.

Despite the 2000 h of contamination, the percentage of sulfates found in the concrete is very small. This small contamination values achieved in the cubes, only 1.2% (maximum value), shows how the natural process of contamination can be slow for a concrete of good composition.

The 28 days average compressive strength of the reference concrete cubes, i.e. those without contamination (C1), was 29.3MPa. The cubes subjected to 1000h of SO₂ contamination showed a mean strength of 32.0MPa, whereas at 2000h the value was 31.0MPa, i.e. it remained approximately constant. The failure of the reference samples, unlike the case of the contaminated samples, was accompanied by the projection of numerous debris, evidencing that the contaminated samples became less brittle. It is suggested that crystals, resulting from the reaction between the cementitious materials and SO₂, partially filled concrete pores, and their continued formation and growth induced internal stresses and cracking. Expectable decrease of strength ensued as the contamination proceeded. The surfaces of contaminated cubes also showed a yellowish color due to chemical alteration as previously described in the literature (27).

The contamination in the SO₂ chamber proved to be insufficient to cause significant effects in the flexural behavior. This was perhaps because the sulfate contamination occurred only on the surface as a result of the short time of actual aging due to technical limitations of the chamber which did not allow sustained application of higher rates of contamination. In addition, these initial concrete specimens were conceived essentially to verify the effectiveness of the process to accelerate the corrosion of the steel and the diameter of the reinforcement bars versus the geometry of the beams was not representative of what would be required for them as structural members, i.e. the reinforcement was excessive. Despite that, flexural tests were performed after contamination and the results obtained in two beams subjected to 1000h and 2000h of contamination, respectively, were compared. The values obtained indicated higher failure load after 2000h (39 versus 32 kN), a fact difficult to interpret since the compressive strength of concrete, as previously seen, was not altered.

The electric resistance of the wires, connections and steel mesh was considered negligible when compared to the electric resistance of the concrete electrolyte and the electric resistance of some specimens produced by possible electrode reactions in the presence of oxygen. This electric resistance was calculated according to the first law of Ohm, R=U/I, where U is the electric potential and I is the current intensity in the circuit. Figure 4 illustrates representative cases of the electrical resistance evolution during the time of contamination of the beams.

The total porosities calculated by Hg intrusion and the corresponding pores size distribution are shown in Figure 5. As expected, concrete C2 is considerably more porous than C3. This difference was reflected in the overall concrete electrical resistance and had implications on the beams contamination through differential ionic current flows. Despite the different porosities, the pore size distribution in each mortar is quite similar.

Along the process of contamination the beams cast with C3 concrete showed consistently higher electrical resistance values than those of the beams cast with C2, despite the influence on the ion transport process of the electric field applied, i.e. the initial concrete porosity controled the ionic current flow, see Figure 4.

In general, as the contamination process advanced the electrical resistance of the system tended to increase. This might be due to the production of substances from the electrodes reaction in the presence of oxygen, and to the reaction of sulfate ions with the tricalcium aluminate (3CaO.Al₂O₃) present in the cement. Reaction of the sulfate ions with tri-calcium aluminate, first, formed calcium sulfo-aluminates and then ettringite, and the resulting new crystals partially filled the pores, further reducing the solution uptake and explaining the increase of the electrical resistivity in the beams, as previously found by Cunha (26).

The larger electric resistance of concrete caused by the decrease of porosity, also decreased the diffusion or mass transport process. However, the speed of the corrosive process was no longer controlled by diffusion but by the presence of free chloride ions near the rebar. This change in the corrosive mechanism can explain the corrosion detected in the rebar.

After a few weeks of contamination, the beams showed oxides leaching at mid-level of the lateral faces. The accumulation of corrosion products, salt crystallization and weakness due to the introduction of the mesh created parallel cracks in the zone of the mesh. The mesh and ensuing parallel cracks, though located at the region of lowest stresses, required careful attention when changes of mechanical strength were analyzed.

Concrete-mesh adhesion was deteriorated by the presence of corrosion products between the mesh and the steel bars. This loss of adhesion between the mesh and the concrete did not occur in non-pre-cracked beams.

The chemical and microstructural damage of concrete due to multiple deterioration mechanisms associated with the combined action of sulfate and chloride involves numerous chemical phenomena and resultant products. Such complexity required thorough analyses and, despite not being a central target of this research, an effort was placed to identify major chemical/microstructural changes caused by the contamination protocol. In order to measure the changes on the microstructure, small pieces of the cement matrix were obtained both from the reference and contaminated concrete beams (non-cracked). Both concrete mortars evidenced a negligible change in porosity due to the contamination, especially C3. However, through the pore size distributions, it was possible to detect important differences in the microstructures after the contamination. Generically, the volume of pores with diameters below 0.025µm (25nm) decreased, whereas, the volume of pores with diameters ranging between 0.025µm and 0.200µm increased. The likely cause is that crystallized salts and/or reaction products between the contaminants and the cement components remained in the system. Another reason is the appearance of micro-cracks due to salts crystallization. Both contributed to significant modifications on the concrete microstructure that affected the transport phenomena.

To investigate the eventual development of new mineral forms during the contamination, the cement paste was also analysed by x-ray diffraction, Figure 6. In all cases, for both concrete types and in pre-cracked and non-cracked beams, gypsum and halite were identified. The major peaks identified in all samples corresponded to quartz and calcite whose presence is attributed to the concrete aggregates (limestone and siliceous sand). In all the contaminated specimens, additional peaks appeared that were attributed to gypsum and halite. These new mineral forms are a consequence of concrete contamination by sulfates and chlorides, respectively. While the occurrence of halite did not indicate any reaction between the cement components, the presence of gypsum confirmed chemical reactions between sulfates and cement alkaline compounds. It was not possible to identify ettringite and thaumasite, probably because the contamination did not last long enough to allow the generation of significant quantities.

Several studies (28, 29) have indicated that chloride ions diffuse much faster than sulfate ions into hardened cement paste. Oberholster reported that the diffusion of chloride ions in concrete is generally 10 to 100 times faster than that of sulfate ions (28). It was thought reasonable to accept that the same happened when applying a constant electric field to push these ions into the concrete.

The presence of sulfate ions in the chloride solution did not influence the initiation time of the chloride-induced reinforcement corrosion, but the rate of corrosion increased with increasing sulfate concentration. That occurred because the tricalcium aluminate in the concrete also binds to sulfate ions, and because the amount of free chloride also intensified the corrosive process.

Accordingly the fast corroding process verified in the rebars is attributed to the high concentration of free chloride ions carried by the applied electric field, and not to the presence of sulfate ions.

Figure 7 displays the evolution in time of the quantity of sulfates in the concrete. The data represent the mean values of the gain of mass of sulfates found for each type of beam.

At the end of the contamination process of the specimens C2 and C3, for 9 and 12 weeks, respectively, the results obtained varied between 0.04g and 0.06g sulfate per 5g of concrete. That matches the percentages of sulfate contamination between 0.8% and 1.2% found in the concrete mass.

Dispersion of the observed results was expected due to the heterogeneity of concrete, reflected in the contents of powder samples taken at different horizontal plane locations and depths. Nonetheless, some conclusions can be advanced from the mean values found.