Carbonation of hybrid concrete with high blast furnace slag content and its impact on structural steel corrosion

The aim of this research was to study the carbonation resistance of a blast furnace slag concrete (80% GBFS/20%OPC), with and without alkaline activation, and its influence on the corrosion of structural reinforcement. An OPC-based concrete produced under the same specifications was used as a reference material. To do this, the material was subjected to an accelerated carbonation process under controlled conditions (65% relative humidity, 1% CO2, 25°C). The half-cell potential (Ecorr), linear polarization resistance (LPR) tests showed that both concretes based on GBFS led to depassivation of the reinforcing steel at approximately 99 days, which is the time required for full carbonation of the evaluated concretes.


INTRODUCTION
Most of the deterioration processes of reinforced concrete are related to carbonation and the presence of chlorides that generate corrosion processes in the reinforcing steel. Specifically, concrete carbonation is a natural phenomenon that consists of the diffusion and dissolution of CO 2 in the pores of the concrete and the subsequent reaction or attack on portlandite (Ca(OH) 2 ) and tobermorite (C-S-H), which generates calcium carbonate. This reaction causes a decreased alkalinity (pH) in pore solution, which results in a loss of passivation in the steel, leading to the initiation and subsequent spread of corrosion (1)(2)(3)(4)(5)(6).
To produce cements that are more environmentally friendly and durable and have better mechanical performance, studies of cements with added supplementary materials, such as fly ash, blast furnace slag, silica fumes, metakaolin (MK) and spent fluid catalytic cracking, have been conducted. However, it has been concluded that ordinary Portland cement (OPC) is more resistant to carbonation than cements with high amounts of these materials (4,5,(7)(8)(9)(10).
Alkali-activated, geopolymer and hybrid cements are other environmentally friendly cements that include between 70 and 100% blast furnace slag (GBFS) or fly ash as a cementitious material. These types of materials generally produce structures that are less permeable and have higher mechanical strengths; however, several studies (7,(11)(12)(13)(14)(15) have reported that Alkali-Activated Slag Concretes are more susceptible to carbonation, which they attribute to the small or non-existent amount of Ca(OH) 2 available for neutralization by the CO 2 diffused throughout the structure, and consequently, the attack on the C-S-H progresses faster and results in a greater loss of strength (7,13).
Borges et al. (9) cured samples containing GBFS at 60°C and obtained denser structures (with more C-S-H gel) in addition to a decrease in the CO 2 diffusion in the matrix, which increased the durability; however, the authors mentioned in their study that the presence of CO 2 in alkali-activated mixtures also affects C-S-H by causing the degradation of gels by decalcification. Bernal et al. (16) confirmed the increased susceptibility of alkali-activated slag concretes with respect to concretes produced only with Portland cement; however, when comparing concretes containing more cementitious material (300, 400 and 500 kg/m 3 ), they observed that this susceptibility decreases, which can be attributed to lower permeability or the direct relationship between the mechanical strength and the probability of carbonation, which has also been mentioned by other researchers (1). It should be noted that several factors could influence the degradation process in alkali-activated cements. Bernal et al. (17) assessed the impact of the activator solution modulus (Ms) and the incorporation of MK on the carbonation resistance of an alkali-activated blast furnace slag and found that susceptibility to carbonation is higher when the Ms is low. However, this behaviour is reversed when MK is added due to the formation of secondary silicoaluminate phases. By studying how the type of activator influences the carbonation resistance, Puertas et al. (15) found that the resistance to carbonation depends strongly on the type of activator used; therefore, by comparing sodium hydroxide with sodium silicate or waterglass (Wg) they showed that the samples activated with Wg are more susceptible.
This study determined the carbonation resistance of a blast furnace slag concrete (80% GBFS/20% OPC) activated with a mixture of sodium silicate and sodium hydroxide and assesses its influence on the corrosion of the structural reinforcement. The results were compared with those obtained in the same blended concrete without alkaline activation. An OPC-based concrete produced under the same specifications was used as a reference material.

MATERIALS AND METHODS
This study used granulated blast furnace slag (GBFS) and a Portland cement (OPC) from Colombia (cement type GU according to ASTM C1157) as cementitious materials. Their chemical compositions and physical characteristics are shown in Table 1. It is important to note that the use of limestone added cement causes high loss on ignition (LOI). Diffractograms of the raw materials are shown in Figure 1. Three types of concrete are prepared, a concrete control based on OPC 100%, a reference material 80%GBFS+20%OPC named CE, and an alkali-activated concrete containing the same proportion of blast furnace slag named HB. A mixture of sodium hydroxide and sodium silicate was used as activator (Ms: 1; %Na 2 O: 5% with respect of GBFS). Table 2 shows the proportions of the materials used in this study.
Cylindrical concrete specimens (76 mm in diameter and 152 mm in height) with and without reinforcement were made; reinforcing steel (6.35 mm of diameter) was placed in the centre of a cylinder with an exposed area of 1000 mm 2 inside the concrete ( Figure 2). As shown in Table 2, the slump is held between 50 and 100 mm to promote medium or plastic consistency in the resulting concrete; this value is ideal for concrete placement by hand and for reinforced sections. The HB samples are cured for 28 days in a humidity chamber, and the reference (CE) and control (OPC) concrete samples are cured in water; subsequently, both are subjected to accelerated carbonation in a chamber under controlled conditions (65% relative humidity (RH); 1% CO 2 , 25°C). Before being placed in the carbonation chamber, the specimens were preconditioned by drying for 6 hours at room temperature (25°C, 80%H.R.), and a coat of impermeable paint was applied to the top and bottom surfaces of the cylinders. This treatment allows to lead the entrance of CO 2 . For the unreinforced concretes, the carbonation front is located by making cross cuts every 8 days for evaluation with phenolphthalein. For the steel reinforced concretes, the half-cell potential (Ecorr) according to ASTM C876 standard (18) and the linear polarization resistance (LPR) following the procedure of the ASTM G59 (19) standard were measured every 30 days. These electrochemical tests were performed using an Autolab PGSTAT128N Potenciostat/Galvanostat instrument ( Figure 2). Ag/AgCl was used as the reference electrode and stainless steel as counterelectrode. The calculation of the corrosion current density was carried out by applying the Stern-Geary equation. These electrochemical measurements were also carried out on the same concretes immersed in water, environment used as a comparison to the accelerated carbonation environment. Pastes with the same proportions of cementitious material and activator were prepared to monitor the progress of the hydration reactions using X-ray diffraction.  Blast furnace slag Type I cement 5  10  15  20  25  30  35  40  45  50 55 Figure 2. Setup for the electrochemical measurements of reinforced concretes exposed to carbonation. Figure 3 shows diffractograms of the HB and CE (OPC/GBFS 20%/80%) pastes after 28 days of curing. The diffractograms showed the presence of some crystalline components of the raw materials ( Figure 1) and of the reaction products. The presence of quartz, calcite, aragonite and C-S-H can be observed in both types of concrete, and in particular, hydrated gehlenite (C 2 ASH 8 /C 4 AH 13 ) is found in the hybrid samples; this is a hydration product characteristic of slags that are alkali activated using sodium silicate and sodium hydroxide (20) and is associated with the absence of portlandite in these samples (21). Some researchers have noted that due to the reduced amounts of OPC in the mixtures (20%), it is possible for portlandite (Ca(OH) 2 ) not to be found (22,23); conversely, other studies have shown that it is possible for a small amount to participate in the reaction with the GBFS or the silicate present as an activator to generate more C-S-H (24,25). Table 3 shows the physical-mechanical properties of the CE and HB samples evaluated and compares them to those of a 100% OPC concrete. It is observed that the compressive strength of the CE after 28 days of curing is approximately 50% less than that of the HB and 100% OPC concretes. This behaviour is in agreement with the results of different studies that have shown that the strengths of concretes with high contents (30% -90%) of blast furnace slag replacing Portland cement after 28 days of curing are lower than those of the 100% OPC control samples, which is due to the relatively low reaction rate of the slag (26)(27)(28)(29)(30)(31)(32)(33)(34)(35). Therefore, it is concluded that gradually increasing the amount of slag in the concrete causes a reduction on the compressive strength at early ages. Regarding absorption and porosity, note that the differences in the total absorption and the porosity are directly related to the reported strength. However, the capillary absorption coefficient (K), resistance to water penetration (m) and effective porosity indicate that both the CE and the alkaliactivated HB are less permeable than the 100% OPC concrete; this is attributable to greater refinement in the pore structure, behaviour that coincides with the research carried out by Rodríguez et al. (36), these results indicate the presence of a more dense and resistant structure. Figure 4 shows the progress of the carbonation front of the HB and CE samples, which is compared with the behaviour of a 100% OPC concrete produced with the same mixture proportions. The samples were subjected to accelerated carbonation conditions (65% RH, 1% CO 2 and 25°C) after 28 days of curing.

Carbonation front
In general, all the concretes show gradual carbonation ( Figure 5 through Figure 7); however, the HB and CE samples became completely carbonated in a shorter period (99 days) compared to the 100% OPC concrete, which showed a carbonation depth of 13 mm at same age of exposure. Therefore, the high susceptibility to accelerated carbonation of the HB and CE concretes has been demonstrated, and it should be noted that although the HB sample generally has a greater compressive strength and lower permeability than the 100% OPC and CE concretes, its susceptibility to carbonation is higher.  Some authors (7,9,37,38) have suggested that the greater susceptibility of the carbonation process is proportional to the amount of blast furnace slag and independent of the decreased porosity caused by the added material, especially in concretes containing 70% or more of GBFS, although their compressive strength remains high. This greater susceptibility is also associated with the decalcification of the C-S-H gel (11,15,39), which affects the mechanical properties, as shown in Table 4, in which the mechanical strengths of the concretes are compared before and after the samples are subjected to the accelerated carbonation process; it can be seen that after exposure to CO 2 for 99 days, compared to those samples cured in the absence of CO 2 , the compressive strength of these samples decreases drastically. The residual strength of CE and HB, after 99 days of accelerated exposure to CO2, was similar (3.35 and 3.60 MPa, respectively). However, it is to be noted that due to the higher initial resistance of HB (34.04 MPa), which is approximately two times higher than the corresponding from CE to 28 days of normal curing, the loss of compressive strength calculated was superior (89%). On the contrary, in the case of OPC concrete, the compressive strength to the same age of exposure presented an increase, which is related to the lower carbonation depth (Figure 4). Similar behaviours have been found by Backharev et al., Bernal et al. (17,40,41), who noted that concretes containing alkaliactivated GBFS exhibited higher rates of carbonation even though they are more alkaline than 100% OPC concretes and blended concretes with pozzolan. The higher alkalinity of alkali-activated concretes is due to the activators, in this case, to a mixture of sodium silicate and sodium hydroxide (with a pH of greater than 13.5), whereas the pH of the OPC paste is between 12.6 and 13.5 (11,42). Some authors (16,37,43,44) have noted that the adverse behaviour of the alkaline-activated concretes towards carbonation can be controlled and reduced by using a larger proportion of cementitious material, a lower water/ cementitious material ratio and a higher Ms.
Due to the large number of factors that affect the results, it is difficult to establish a correlation between the exposure times under accelerated and natural conditions. Several authors (45)(46)(47)(48) have suggested that there is a relationship between natural exposure (K N ) and exposure to accelerated conditions (Kc) that can be expressed using [1] K K C N , where C represents the CO 2 concentration of the accelerated environment and N represents the CO 2       concentration of the natural environment; for the purpose of analysing this condition, an environment with a CO 2 concentration of 0.1% or [N=0.1] was assumed, and the values of K N were calculated (Table 5). In this case, C corresponds to the value used in the accelerated environment (1% CO 2 ) and K C to the value obtained during the test. It should be noted that these expressions have been developed for Portland cement concrete.
Using the values of K N found for each of the concretes evaluated, the time required for complete carbonation (t C ) can be estimated in an environment containing 0.1% CO 2 , as shown in Table 5. Note that the time required for the 100% OPC concrete is greater than 50 years, which is the estimated time required for a medium-strength Portland cement concrete, whereas the HB and CE concretes require only 10 years; this demonstrates the increased susceptibility to carbonation of Portland concretes with high percentages of GBFS and hybrid alkali-activated concretes based on 80% GBFS in urban and industrial environments containing CO 2 emissions at concentrations of 0.1%. However, as explained by Duffó et al. (49), for a good concrete, the values of K are between 0.25 and 1; based on this and as Table  5 shows, the HB and the concrete CE (80% slag-20% cement) can be considered also good, as is the 100% OPC concrete subjected to natural carbonation. Figure 8 shows the half-cell potential (Ecorr) over time for the steel embedded in the HB and CE concretes in an accelerated carbonation environment (65% RH, 1% CO 2 and 25°C). In general, both concretes exhibit similar behaviour. In accordance with ASTM C876 [18] and (50), during the first 95 days of exposure, Ecorr is in a zone in which corrosion may or may not occur (between -0.10 and -0.25 V vs. Ag/AgCl). After 120 days of exposure, the values of Ecorr for the concretes decrease drastically until they reach the level at which the probability of corrosion is 90%. These results coincide with those obtained in the carbonation front test, which showed complete carbonation of the concretes at the same exposure age (Figures 4 through Figure 7). After this age, Ecorr values of both concretes, hybrid and reference, show a variable behaviour, with potential values between -0.25 and -0.60 V vs. Ag / AgCl until approximately 700 days in an accelerated carbonation chamber, thus remaining in the zone of 90% probability that corrosion will occur. It is observed, from the 425 days of exposure in samples of reinforced hybrid concrete submerged in water, as well as in those exposed to accelerated carbonation, a similar behaviour regarding the corrosion potential.  Carbonation of hybrid concrete with high blast furnace slag content and its impact on structural steel corrosion • 7 Figure 9 shows the values of the corrosion current density (icorr) of the steel embedded in the concretes evaluated, calculated from measurements of LPR using the Stern-Geary equation. All the concretes showed high icorr above 2 µA/cm 2 . These results are strongly correlated with those obtained in the corrosion potential test (Figure 8), in which both concretes are likely to become corroded. The steels extracted from the different concretes to confirm that the corrosion has occurred (Figure 10). The results of the electrochemical tests are associated with the results of Aperador et al. (51,52). These authors evaluated alkali-activated concretes containing GBFS (100%) and attributed the high carbonation rate, among other factors, to micro-cracks produced by the contraction of these materials during the drying process. These micro-cracks facilitate the entry of CO 2 into the structure; this has also been explained by other researchers (14,39). Additionally, Bernal et al. (53) noted that the advancement of the carbonation process contributes to increases in the porosity, and consequently, this phenomenon contributes to the acceleration of the corrosion process. Alcaide et al. (54) studied the effect of carbon fibre on alkali-activated slag mortars reinforcing steel corrosion and showed that the maximum corrosion density reached in the carbonation process is 5.7 µA/cm 2 , after around 50 days this value flattened to 2.5 µA/cm 2 , this is because to the interaction between the CO 2 and C-S-H gel and additionally to the presence of carbonaceous materials which modify the reaction kinetics. However, it is possible to increase the useful life of these alkaliactivated reinforced concretes and to decrease their susceptibility to carbonation by appropriately controlling the design of the concrete, especially the proportion of cementitious material and the type and proportion of the alkaline activator (55-57).

CONCLUSIONS
This study evaluated the susceptibility to carbonation of a blast furnace slag concrete (80% GBFS/20%OPC), with and without alkaline activation, named HB and CE respectively. The concretes were exposed to accelerated carbonation conditions (65% RH, 1% CO 2 and 25°C) and the decrease in compressive strength and the carbonation front were measured. A concrete based on OPC was used as reference. The susceptibility of these concretes to corrosion was also assessed using the corrosion potential and the linear polarization technique. From the results obtained, the following is concluded: • HB and CE concretes showed a greater susceptibility to accelerated carbonation than Portland cement concrete. • Under accelerated carbonation conditions, CE (80%GBFS/20%OPC) lost up to 79% of its initial strength, and HB lost 89% of its initial strength at 99 days of exposition. It is to be pointed out that the initial compressive strength of HB is twice as high as that of CE, but the strength after the exposition were similar. • Complete carbonation (100%) of both concretes (CE y HB) was observed at 99 days of CO 2 exposure. However, the OPC concrete barely reached 35,3% of carbonation depth at same age. • When the actual time for complete carbonation of the evaluated concretes under aggressive environmental conditions (0.1% CO 2 ) was estimated, it was found that the blended Portland concrete (80%GBFS/20%OPC) required 9 years, the HB (alkali-activated concrete 80%GBFS/20%OPC) required 10 years and the Portland cement concrete required approximately 58 years. • The corrosive process of the steel reinforcement in the concrete with the addition of GBFS 80%, CE and HB began after 99 days of exposure to an accelerated carbonation environment, and corrosion current density greater than 1 µA/cm 2 could be observed.