A comparison between uniform and galvanic corrosion rate in carbonation induced corrosion of reinforced concrete

. Carbonation-induced corrosion is usually considered in uniform or generalized corrosion. This is true for small laboratory samples but not necessary the case for real reinforced concrete structures built with multilayers reinforcement. Multi-layers reinforcement could lead to a galvanic corrosion between first de-passivated layer closed to the concrete surface and deeper layers still passivated. The present paper compares the uniform corrosion density and the galvanic corrosion density for concrete samples made with two different type of cements. CEM III cement despite their higher susceptibility to carbonation show a significant lower galvanic corrosion density, which would lead to a significant increase in the service life of RC structures, exposed to carbonation.


Introduction
Reinforced concrete is popular as a building material all over the world.Unfortunately, it can be deteriorated by corrosion after a while, which can lead to a premature reduction of its service life.Carbon dioxide is one of the aggressive factors involved in the corrosion of concrete structures [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14].It penetrates in gaseous form and reacts with water in the concrete pore space, then with the hydrates of cement paste.This leads to a pH decrease in the concrete porosity from 12.5-13, which can protect the embedded steel against corrosion, to a value below 9, which results in the depassivation of steel bars [12] [15] [16] [17] [18].Recently, concern about carbonationinduced corrosion has increased due to the need to lower emissions released to the environment [18] [19] [20].However, a change of clinker content to reduce CO2 emission could lead to concrete exhibiting lower resistance against the carbonation process.Among the binder substitutes available, slag is a supplementary cementitious material that is currently in favour [3] [8] [12].This paper compares the carbonation induced corrosion occurring when cement with high clinker content is used and when a high percentage of the clinker is replaced by ground granulated blastfurnace slag.Numerous researchers are of the opinion that carbonation-induced corrosion occurs in uniform or generalized corrosion [6] [12] [21] [22] [23] [24] [25] [26].However, there are several alternative opinions.According to [3], in real structures, corrosion caused by carbonation may be localized.In [27], Castel and Nasser indicate that galvanized corrosion in carbonated concrete structures seems higher than uniform corrosion.This phenomenon is also described by François et al. in [15].As schematically shown in Figure 1, when the carbonation front reaches the first layer of rebars, there is a strong galvanic coupling between the other layers of the reinforcement, including the stirrups, which are still in un-carbonated concrete that is quite saturated due to the distance to the concrete surface.The corrosion of the first layer then results from a coupling between a uniform corrosion and a galvanic corrosion.Unlike the corrosion products formed during chloride-induced corrosion, those of carbonation-induced corrosion are not fluid, so the pressure on the concrete cover will be high.It could lead to the premature appearance of corrosion-induced cracks, which are inaesthetic, may reduce the bond between reinforcing steel and concrete, and may lead to delamination and thus to safety problems for the public.In the present paper, the corrosion in carbonated concrete will be studied from the points of view of both galvanized corrosion and uniform corrosion.It is important to better understand the relative proportions of galvanic corrosion and uniform corrosion in the early process of corrosion in real carbonated reinforced concrete structures.By encouraging hydroxide production in the cathodic area, the galvanic part of the corrosion current could also reduce the kinetics of carbonation.This point is not discussed in this paper.

Materials and method 2.1 Materials
Three concrete formulations (B1-15L-049, B3-041 and B3-15L-051) were used in this study.Their compositions are given in Table 2.These 3 concretes are studied in a French Project call Perfdub (https://www.perfdub.fr/)which aims to study concrete performance according to durability indicators.It corresponds to industrial concrete compositions used on site, but the W/B ratio is almost the same (i.e.0.5) for the CEM I concrete (B1-15L-049) and one CEM III concrete (B3-15L-051).The 2 nd CEM III concrete allows to study the influence of a lower W/B ratio (0.41).The steel bars used in this study were cold rolled ribbed steel bars with diameters of 6 mm and 12 mm, manufactured by Fimurex, France.No treatment was applied to the rebar and the presence of a non-uniform layer of mill scale on the surface was observed.This surface state, also called "As Received", is one of the most popular types for steel bars.

Characterization tests
Table 3 presents the results of several characterization tests which were performed on these 3 concretes.Water porosity tests were carried out according to standard NF P18-459.It was found that the water porosity of concrete B3-15L-051 at 90 days was slightly higher than that of B1-15L-049 at 28 days and that the porosity of B3-041 was much lower than that of B1-15L-049 and B3-15L-051 due to the lower w/b ratio of this formulation.
The average values of electrical resistivity of all the mixes with the coefficient of variation (CV) were tested in the French Perfdub project [40].It can be seen that slag cement formulations (B3-041 and B3-15L-051) present higher resistivities than the concrete without slag (B1-15L-049), B3-041 having a higher resistivity than B3-15L-051 due to its lower w/b ratio.

Preparation of samples
The samples were based on the test protocol with separation of the cathode and anode derived by Chalhoub et al. [28].However, the anodes were involved in the process of carbonation instead of chloride contamination.

Cathode samples
Cylindrical cathodes and a reinforced concrete (RC) wall were used in the research.
The cylindrical cathode samples had a diameter, Ø, of 110 mm, a height, h, of 220 mm and a cover of 5.2 cm (Figure 2).One Fe 500 ribbed steel bar with 6 mm diameter was embedded at the centre of the cathode.To verify the influence of the cathode/anode ratio, 3 lengths of steel bar were utilized: 10 mm, 50 mm and 160 mm.After casting and demolding, the cathode samples were placed in a wet curing room (relative humidity = 95%).Cylindrical cathodes allowed cathode/anode ratios of 1, 5 and 16 to be tested.
The RC wall had dimensions 75x20x100 cm and contained 10 horizontal Ø 12 mm bars 70 cm in length and 8 vertical Ø 12 mm bars having lengths of 102 cm and 5 cm emerging from the concrete (Figure 4).The steel bar network was completely electrically disconnected but could be electrically connected externally thanks to a connection box.During the measurement of the galvanic current, the anode was immersed in NaOH solution contained in a PVC pipe firmly sealed outside the wall.Reinforced concrete wall was used to test cathode/anode ratios (C/A) from 750 to 2200.The RC wall was made with ordinary concrete: CEM I cement and w/b ratio of 0.55. the electrical resistivity is 110 .m at 90d.The RC wall has a different mix composition because it is used only to test high C/A ratios.

Anode samples
The anode samples were cylindrical with dimensions Ø= 33 mm and h= 70 mm, having one steel bar with a diameter of 6 mm and a height of 10 mm embedded at the centre (Figure 3).These samples were carbonated before being utilized in the following tests.

Anode carbonation process
After 8 weeks of wet curing, the anodes were stored in an oven at controlled temperature of 45°C and constant relative humidity of 25% during 14 days, followed by 7 days at 20°C and 65% relative humidity.The contamination of the anodes with carbon dioxide was carried out in a 65% relative humid chamber with 50% of CO2.The choice of 50% of CO2 was done in accordance to NF XP P18-458 standard, this method was used in the first part of French Perdub project.The percentage of CO2 is very high and it could have an effect on the length of steel concrete interface carbonated from a crack as shown by Ghantous et al. [41].
Nevertheless, here the goal is to fully carbonated the anode sample, and not studied the progress of the carbonation front.Then, results will allow to compare the corrosion rates of the 3 concretes.
The carbonation depth was measured with phenolphthalein solution.After spraying with this solution, if the concrete is observed to be purple, its pH is higher than 9.The pH is from 8 to 9 when the concrete becomes pink and is below 8 when no change of color is observed.The aim was to obtain complete carbonation of the anode samples.For each formulation, 3 control specimens (without reinforcements) 40 mm in diameter and 70 mm in height were extracted from specimens 110 mm in diameter and 220 mm in length by coring.These test pieces were placed under the same conditions of attack (carbonation chamber) as those intended to reveal the progress of the carbonation front.
The complete carbonation of the anode samples was quite long especially for the B3-041 concrete with the lower W/B ratio.The end of the carbonation process was decided at 100d, 200d and 250d respectively for the 3 concretes B1-15L-049, B3-15L-051 and B3-041.

Corrosion process
After complete carbonation of the anode samples, checked according to the control samples, they were placed in tap water for 48 hours to induce full capillary suction to reach the re-bar and lead to the onset of corrosion.The corrosion onset was checked by both a potential drop greater than 200 mV and the existence of galvanic current when connected to a cathode.

Measurement of current
The methods chosen for the assessment of corrosion under uniform current and galvanized current were Polarization resistance and a Zero resistance ammeter (ZRA), respectively.All the electrochemical experiments were performed at a constant temperature of 20°C in a controlled room.

Measurement of uniform current
The measurement of general current was based on RILEM TC 154-EMC [29].This is a non-destructive electrochemical method to obtain the corrosion current density.The corrosion rate was calculated by means of the Stern-Geary formula (Equation 1): (1) The value of 26 mV is recommended for constant B in cases where steel is in an active state [30].For the anode used in the test, the dimension of the steel is very small and the concrete totally carbonated.Corrosion can be considered uniform and the corrosion current can therefore be calculated by means of Equation 2: where S is the surface area of the steel embedded in the anode.With the anodes used in the test, S= 1.88 cm 2 .
The LPR technique gives an "apparent" response including both the polarization resistance Rp and the Ohmic resistance Re.The calculation of Re was achieved by a potentiostatic step of 30 mV to calculate the instantaneous response corresponding to the Ohmic behavior (see Figure 5a and 5b).

Measurement of galvanic current
The anode and cathode samples were connected by a Potentiostat controlled by EC lab Software using the electrochemical technique to measure the galvanized current between them (figure 6).The measurement of the current was set up in 24 hours.The anodes were totally immersed in water during the test.
Cathode permutation was used to study the impact of cathode/anode ratio by changing the size of the cathode.Cylindrical cathode corresponding to a ratio of 1, 5 and 16 were used.In addition, the anode samples were coupled a concrete wall to obtained larger C/A ratio (varying from 750 to 2200).The approach used to measure the resistivity of carbonated sample consisted in using the electrical resistance obtained during the potentiostatic tests of steel reinforced specimens (from the instantaneous responses at the polarization step such as shown in Figure 5a) and converting it into electrical resistivity by inverse numerical modeling.In other words, a numerical model was involved to assess the specimen geometrical factor, linking resistance and resistivity.The protocol is defined in Chalhoub et al. [45].

Electrical resistivity of carbonated samples
Table 5 presents the average value of electrical resistivity of the 3 concretes after full carbonation, with the coefficient of variation (CV) and the carbonation duration of the samples when tested.It can be seen that CEM III formulations (B3-041 and B3-15L-051) present higher resistivities that the CEM I concrete (B1-15L-049), B3-041 having the higher resistivity than B3-15L-051 due to its lower w/b ratio.The resistivity of the 3 concretes after carbonation is about 3 times higher than non-carbonated concretes.In both cases, the measurement was achieved on saturated materials.The results is in agreement with previous research [38][39].

Uniform corrosion current
The corrosion current densities measured according to LPR tests for the 3 different concretes are shown in figure 7a, 7b, 7c.For each concrete, four samples are tested (S-1 to S-4; or S-2 to S-5 for B3-041 concrete).A significant dispersion can be seen between the different anodes of the same concrete group.The difference in terms of uniform current does not change very much from one concrete composition to another.Despite a significant difference in resistivity for the 3 concretes, the uniform current densities are in the same order of magnitude about 1.5 µA/cm 2 .The porosities of the 3 concretes B1-15L-049, B3-15L-051 and B3-041 are respectively 16.9%, 16.7% and 12.1%.The absence of effect of porosity on the uniform corrosion current is probably linked to the high moisture environment involved in the test protocol [35].

Galvanic current
The galvanic corrosion current, expressed as galvanic current density by using the anode surface area, between the 4 anodes of each group connected to a cathode of the same group are shown in figures 8a, 8b and 8c.The cathode used is the one with a C/A ratio of 16.It can be seen that the differences in galvanic current density between the different anodes of the same group of concretes are not very large but, on the contrary, there is a large difference in terms of galvanic current density between the different concrete compositions.The effect of concrete resistivity in the case of galvanic corrosion is thus highlighted.When dealing with galvanic corrosion process, the effect of resistivity of concrete is linked to both anode and cathode.At anode location the resistivity is higher than at cathode location.Nevertheless, for a given concrete, a higher resistivity at anode corresponds also to a higher resistivity at cathode.As a result, the galvanic current in case of CEM III concretes (B3-041 and B3-15L-051) is significantly lower than CEM I concrete (B1-15L-049), B3-041 having the lower galvanic current density than B3-15L-051 due to its lower w/b ratio.

Galvanic current in relation to the Cathode to Anode ratio (C/A)
When dealing with carbonation-induced corrosion, the Cathode to Anode ratio (C/A) is not so high as in the case of chloride-induced corrosion, because a significant surface area of reinforcing steel is depassivated by the motion of the carbonation front.In the case of two layers of reinforcing steel, the ratio could be estimated to have a value of 1.In the case of a multi-layer structure or structure with a very dense reinforcement layout, such as can be found in a nuclear power plant, the ratio could be higher -but probably less than 10.It should be noticed that in case of pre-cracked concrete (load-induced cracks or others causes), the carbonation front could reach quickly the steel bar at crack tip and then induced corrosion initiation with a high C/A ratio.Nevertheless, to compare the effect of the C/A ratio on the galvanic current density with that of chlorideinduced corrosion as shown by Chalhoub et al. [28], the B3-041 anodes were connected to different cylindrical cathodes with C/A ratios varying from 1 to 16, and then to a reinforced concrete wall with a C/A ratio varying from 750 to 2200.The same CEM I concrete cathode as used by Chalhoub et al. [28] was connected to the different anodes of the 3 groups of concrete.This experiment does not correspond to a practical case since, on-site, the anode and cathode are obviously made with the same concrete.Because the tendency was the same, it was chosen to only show the B3-041 results.It gives information on the control process due to the resistivity of the cathode and the effect of C/A ratio.

Fig. 9b. Galvanic current density vs C/A ratio up to 2200
From the results plotted on figure 9a, it can be seen that the galvanic current for a C/A ratio of 16 is varying between 4 and 5.5 µA/cm 2 , with an average value of 4.75 µA/cm 2 , which could be compare with figure 8b where the average value is 0.425 µA/cm 2 .As a result, by changing the concrete at the cathode from CEMIII concrete (B3-041) (resistivity of 718 .m) to the CEM I concrete from the wall (resistivity of 110 .m), the galvanic current measured is 10 times higher.It confirms that the galvanic corrosion process is on ohmic control.The effect of C/A ratio on galvanic current is shown in figures 9a and 9b.For low C/A ratios, it seems that the current density is quite proportional to the ratio but, for high C/A ratios, the behavior tends to become asymptotic.These results are similar to those found by Chalhoub et al. [28] when dealing with chloride-induced corrosion.Effect of C/A is well-known for macrocell current in chloride-induced corrosion as shown by Warkus and Raupach [42], but usually not considered in carbonation-induced corrosion.Because the increase in electrical resistivity reduces the mobilizable Cathode Anode distance [43] [44], a high resistivity concrete can limit the galvanic process in case of structures with dense reinforcement lay-out.

Influence of resistivity of concrete on corrosion current
For uniform corrosion process, the resistivity involved is the one after carbonation which is higher than the initial one.Figure 10 shows the effect of the resistivity of concrete and the uniform current density measured both after carbonation.Resistivity was measured in saturated conditions corresponding to the corrosion test.
Resistivities of the 3 carbonated concretes B1-15L-049, B3-15L-051 and B3-041 were respectively 200, 800 and 2000 .m as shown in Table 4.The results plotted in Figure 10 do not show a clear influence of concrete resistivity on uniform current.This result is quite logical since, in a uniform corrosion process, anodic and cathodic sites are at the same location, so neither anodic nor cathodic current is supposed to be influenced by the concrete resistivity.This indicates that the uniform corrosion rate of steel in carbonated concrete is not under ohmic control as found by Stefanoni et al. [35].
Nevertheless, many studies have shown a negative correlation between corrosion rates and the resistivity of concrete for carbonation induced corrosion

Fig. 10. Influence of resistivity on the uniform current density
For galvanic corrosion process, both resistivities before and after carbonation are involved in cathodic and anodic process respectively.Nevertheless, it was decided to plot the galvanic current density as a function of actual resistivity of the anode during the measurement (after carbonation).
Figure 11 shows the effect of resistivity of concrete on the galvanic current density measured both after carbonation.The results plotted in Figure 11 show a huge effect of resistivity of the concrete on the galvanic current.Here again, this result is quite logical since the distance between the anode site and the cathode site is affected by the ionic current flowing in the concrete, which is strongly dependent on the resistivity.It should be noticed that the cathodic process corresponds to a noncarbonated concrete where the resistivity is lower than the one used in the Figure 11.Nevertheless, the use of the resistivity of sound concrete in abscissa of Figure 11 will not change the tendency because the change in resistivity before and after carbonation are almost the same for the 3 concretes tested.The use of CEMIII cement in concrete, which significantly increases the resistivity of the concrete, is thus shown to reduce the galvanic part of the corrosion current in real structures with multi-layer reinforcements.Indeed, Figure 11 shows that galvanic current density for B1-15L-049 concrete is about 5 times higher than B3-15L-051 concrete for a quite identical w/b ratio of respectively 0.49 and 0.51 resulting in a quite identical water porosity of respectively 16.9%, 16.7%.This result clearly shows an impact of the cement type on the corrosion rate of carbonated concrete.Moreover, when the two CEM III concretes (B3-15L-051 and B3-041) are compared, the corrosion current density is significantly lower for the B3-041, which has the lowest resistivity due to a lower w/b ratio.This result clearly confirms the impact of w/b ratio on the galvanic current density.
The main object of the study was to investigate the two aspects of corrosion involved in the carbonation of real multi-layer reinforced concrete structures: uniform corrosion and galvanic corrosion.Three different industrial concretes actually used on site were tested: a CEM I and two CEM III.CEM III concrete are nowadays the most common concretes currently used for reinforced concrete structures.Indeed, using CEM III allows the emission of CO2 into the atmosphere to be reduced and also leads to higher resistivity concretes, which are considered to be more durable.Nevertheless, the resistance of CEM III to carbonation is currently considered to be lowered when the amount of clinker is reduced [36], so the use of CEM III in high-carbonate environments could be questionable.It should be reminded that the carbonation process is not studied in this paper, only the corrosion rate is checked after full carbonation of the steel concrete interface.From this study, it is possible to highlight some points of discussion: The possibility of carbonation-induced corrosion should be considered with both uniform and localized or galvanized corrosion, rather than with only uniform corrosion.Indeed, real reinforced concrete structures use multi-layers reinforcement framework which lead to the co-existence of passived and depassivated layers during the corrosion process.

Corrosion rate
The uniform corrosion rate found in this study are about 1.5 µA/cm 2 .For the results of this limited study, the uniform corrosion rate is quite constant whatever the type of cement used and whatever the w/b ratio.The galvanic corrosion rate found in this study are varying between 15.5 µA/cm 2 (average value for B1-15L-049), 3.3 µA/cm 2 (average value for B3-15L-051) and 0.5 µA/cm 2 (average value for B3-41).Nevertheless, it corresponds to a high C/A ratio (16).In practice, in case of a situation with two layers of steel reinforcement, where one is in carbonated and one not, the C/A ratio will be reduced (C/A =1), then the galvanic corrosion rate will be reduced.Nevertheless, in practice, there is also the possibility to have high C/A ratio: for example, for corrosion in pre-cracked concrete (loadinduced cracks, shrinkage cracks etc..) where the carbonation front reach quickly the rebar at crack tip.Some local lack of concrete cover could also lead to high C/A ratio since the carbonation front reach only a small area of steel bar at the cover defect.In case of high C/A ratio, the galvanic corrosion rate is significant and could not be neglected.As a result, the use of high resistivity concrete lead to reduce the corrosion rate, by both ohmic effect and also because the mobilizable cathode surface decrease with the increase in resistivity.It is generally considered that corrosion rate decreases with an increase of the resistivity [35].This global trend is confirmed by the results of this study by assuming that carbonation-induced corrosion is the sum of a uniform corrosion process and a galvanic corrosion process.

Concrete resistivity before and after carbonation
Carbonation on concrete leads to a significant increase in resistivity of about 3 times.For the 3 concretes tested the increase in resistivity seems more important when initial resistivity is lower.Such result was also found in [38].Uniform corrosion rate does not depend on resistivity.For galvanic corrosion process, the cathodic reaction occurs in sound concrete zone (non-carbonated), then the resistivity value which need to be considered is the one before carbonation.It is the case when dealing with durability indicator for concrete, such as in the French Perfdub project: indeed, the concrete resistivity corresponds to the value of sound concrete before carbonation.

Corrosion environment
In this study, the Relative Humidity RH% at anode correspond to saturated conditions.Such saturated conditions could be found on-site, when for example a RC wall or RC floor initially exposed to dry environment during many years is exposed to water leakage due for example to a damage in roof waterproofing.Moreover, as shown in a review by Stefanoni et al. [35], the contact with water is the most aggressive conditions in term of carbonation-induced corrosion.

Conclusion
The paper compares two aspects of corrosion process in case of carbonation-induced corrosion on RC structures: the uniform corrosion process and the galvanic corrosion process which exists in case of multi-layers of steel where only one layer is in carbonated concrete.
-The uniform corrosion rate does not depend of the concrete resistivity.(resistivity in this paper, varies from 69 to 718 Ω.m) -The galvanic corrosion rate is highly dependent of the resistivity.
-The galvanic corrosion rate is highly dependent of the C/A ratio.
-On-site corrosion rate, which is the sum of the uniform part and the galvanic part, is then likely to depend of the resistivity of concrete.
-The galvanic corrosion rate is highly dependent of the C/A ratio.In practice, C/A ratio could be limited or important.In the case of local defects or presence of cracks, the galvanic part of corrosion rate could be the most important contribution to the total corrosion rate.
-the use of CEM III concrete, which corresponds to high resistivity concrete, leads to a significant limitation in the galvanic corrosion rate on-site.

Fig. 1 .
Fig. 1.Schematic Illustration of galvanic corrosion due to carbonation: when the carbonation front reaches the first layer of the reinforcement layout, the other layers of steel bars, which are still passive and located in the un-carbonated zone act as a cathode.

Fig. 11 .
Fig. 11.Influence of resistivity on the galvanic current density (for a C/A ratio of 16)

Table 1 .
Silicate and aluminate compositions of cements, and GGBS slag compositions.

Table 1
presents the silicate, aluminate and GGBS compositions and of the cements used in concrete formulations.

Table 2 .
Composition of concrete formulations.

Table 3 .
Characterization tests results of the concrete formulations.