The effect of two types of modified Mg-Al hydrotalcites on reinforcement corrosion in cement mortar

Cement and Concrete Research 100 (2017) 186–202

Contents lists available at ScienceDirect

Cement and Concrete Research journal homepage: www.elsevier.com/locate/cemconres

The effect of two types of modified Mg-Al hydrotalcites on reinforcement corrosion in cement mortar

MARK

Zhengxian Yanga,b,c,⁎, Rob Polderc,d, J.M.C. Mole, Carmen Andradef a

College of Civil Engineering, Fuzhou University, Xueyuan Road 2, University Town, 350116 Fuzhou, China Department of Civil and Environmental Engineering, Washington State University, P.O. Box 642910, Pullman, WA 99164-2910, United States c Section of Materials and Environment, Faculty of Civil Engineering and Geosciences, Delft University of Technology, P.O. Box 5048, 2600 GA Delft, The Netherlands d TNO Structural Reliability, P.O. Box 49, 2600 AA Delft, The Netherlands e Department of Materials Science and Engineering, Delft University of Technology, Mekelweg 2, 2628 CD Delft, The Netherlands f Institute “Eduardo Torroja” of Construction Science, Spanish National Research Council (IETcc-CSIC), Serrano Galvache 4, 28033 Madrid, Spain b

A R T I C L E I N F O

A B S T R A C T

Keywords: Modified hydrotalcites Durability (C) Corrosion (C) Chloride (D) Mortar (E)

Two modified Mg-Al hydrotalcites (MHTs), (MHT-pAB and MHT-NO2) were incorporated into mortar (with different w/c ratios) in two different ways: (1) as one of the mixing components in bulk mortar; (2) as part of cement paste coating of the reinforcing steel. Accelerated chloride migration, cyclic wetting-drying and diffusion tests were performed to investigate their effect on reinforcement corrosion. The results indicated that MHTs could be promising alternatives for preventing chloride-induced corrosion when an appropriate dosage is adopted and applied in a proper way, particularly, replacing 5% mass of cement by MHT-pAB in bulk mortar or as a coating of reinforcing steel (MHT-pAB/MHT-NO2 to replace 20% mass of cement). The effect of MHT-pAB on time-to-corrosion initiation (TTC) of reinforcing steel was estimated using the DuraCrete model. It was found that the incorporation of 5% MHT-pAB in bulk mortar led to a more than double TTC relative to reference mortar without MHTs.

1. Introduction Corrosion of reinforcing steel is a major culprit to durability and serviceability of concrete structures. This problem is highly relevant for civil engineering structures in the transport sector, such as bridges, tunnels, harbour quays and parking structures. The dominant aggressive influence is the chloride load from de-icing salts or sea water, penetrating the concrete and destroying the natural passivation of the steel [1]. The direct and indirect costs of corrosion are substantial, as it entails additional repair, rehabilitation, and monitoring activities to ensure the safety, functionality and aesthetics of concrete structures and components. Presently available corrosion preventive measures are either too costly or technically too complicated [2–5]. Stainless steel reinforcement is 5 to 10 times more expensive than reinforcing (carbon) steel [2–4]. Cathodic prevention (CPre) and protection (CP) may be effective but both are a special niche expertise and are thus not applied on a wide scale [5,6]. In addition, maintaining CPre or CP systems involves significant cost, mainly for semi-annual control measurements [6]. Coatings on the concrete surface normally do not last long enough (10–20 years), which causes a maintenance cycle of its own [2]. Corrosion inhibitors seem to be attractive owing to their low cost and the

⁎

ease of application [7–9]. However, there are conflicting opinions about their reliability in terms of long-term effectiveness [8]; some are toxic, such as nitrites [10]. A promising solution to overcome this problem is the immobilization of desired inhibitors within the molecular structure of a host compound. The immobilized inhibitor then can be slowly released in a controlled way by an external stimulus (e.g. chloride ions) and therefore provide a relatively long-term corrosion protection. Hydrotalcite is one representative of a large mineral group of Layered Double Hydroxides (LDHs). They can be represented by a x+ − x− general formula [MII1−x MIII [(Anx/n )] ·mH2O, where MII and x (OH)2] MIII are di- and trivalent metals respectively, and An − is an interlayer charge-balancing anion with valence n. The x value is in the range 0.20–0.33. Fig. 1 schematically shows a typical structure of hydrotalcite-like compounds and more related information can be found elsewhere [11,12]. Owing to the unique fine tuneable molecular structure and high anion exchange capacity, hydrotalcites exhibit a great potential to be tailor-made as an active component of concrete for the immobilization of a desired inhibitor. Hydrotalcites have been found in hydrated slag cements, which are known to be able to bind more chloride ions than pure Portland cements [13–15]. The existence of hydrotalcite-like phases such as Friedel's salt or its iron analogue

Corresponding author at: College of Civil Engineering, Fuzhou University, Xueyuan Road 2, University Town, 350116 Fuzhou, China. E-mail address: [email protected] (Z. Yang).

http://dx.doi.org/10.1016/j.cemconres.2017.06.004 Received 28 November 2015; Received in revised form 10 May 2017; Accepted 16 June 2017 0008-8846/ © 2017 Elsevier Ltd. All rights reserved.

Cement and Concrete Research 100 (2017) 186–202

Z. Yang et al.

Fig. 1. Schematic representation of a typical hydrotalcite molecular structure with intercalated anions and water molecules.

both internal reservoir of corrosion inhibitors and a trap for chlorides. This paper presents the effect of two types of modified Mg-Al hydrotalcites on reinforcement corrosion in cement mortar. The primary objective is to illustrate the effect of the two MHTs on: a) the critical chloride concentration (also known as chloride threshold, CT) responsible for corrosion initiation of the reinforcing steel and b) on chloride penetration in mortar and their possible correlations.

and/or Kuzel's salt are believed to contribute to chloride binding and thus enhance the corrosion resistance of reinforced concrete [16]. Tatematsu et al. [17] synthesized a calcium-aluminium based hydrotalcite and added it into cement mixtures as a salt adsorbent. Their results showed that once the admixed material contacts with chloride ions, it could adsorb them and release the intercalated nitrite anions simultaneously to further protect concrete from corrosion. More recently, Kayali et al. [18] demonstrated that hydrotalcite which comprises 54% of the crystalline phases in pure hardened ground granulated blast furnace slag (GGBFS) paste activated with 4 M NaOH is the main hydration product responsible for the remarkable improvement in chloride binding by concrete containing GGBFS. Within the modified hydrotalcites (MHTs) family, a class of materials with emerging importance is that constituted by MHTs intercalated with organic species. Raki et al. [19] demonstrated the potential of MHTs as suitable hosts for intercalation of organic functional admixtures with the long-term view of controlling their release rate in concrete. Although many organic compounds have been tested as corrosion inhibitors in reinforced concrete, more concerns are being raised regarding their impact on the environment. Increasing awareness of the health and ecological risks attracts much attention to amino acid-based inhibitors because they are environmentally friendly [20,21], relatively cheap and easy to produce with higher purity [22]. Our preliminary study [23,24] has shown that certain organic species (in particular amino acids) with known inhibitive properties could be intercalated and ion exchange occurs between free chloride ions in the simulated concrete pore solution and anions intercalated in MHTs, thus reducing the free chloride concentration. The release of the intercalated inhibitive anions is trigged by arrival of chloride ions. The inhibiting effect from the released inhibitors increases the chloride threshold level for corrosion initiation and/or reduces the subsequent corrosion rate. Less aggressive electrochemical potentials have been observed in chloride-containing simulated concrete pore solution with MHTs as compared to solutions without MHTs [23]. As shown in Fig. 2, a distinctive feature of MHTs relative to the other corrosion inhibiting approaches is the dual-role working mechanism: capturing aggressive chlorides and simultaneously releasing the intercalated inhibitive anions to further protect the reinforcing steel from corrosion [11,25]. Therefore, once applied in concrete, the MHTs can be envisioned as

2. Experimental 2.1. Materials CEM I 42.5 N cement, CEN-Standard sand (particle size 0–2 mm) and deionized water were used for preparing mortar. Reinforcing steel was low‑carbon steel (B500A) bars with a nominal diameter of 8 mm. The two MHTs used in this research were MHT-pAB and MHT-NO2. They were synthesized by the modification of carbonate hydrotalcite (MHT-CO3, Mg/Al atomic ratio 2:1) with sodium p-aminobenzoate (−pAB) and sodium nitrite (−NO2) through a calcination-rehydration procedure [26]. Elemental analysis showed that 11% NO2− and 32% -pAB by mass of corresponding MHT were intercalated into their molecular structures. The relevant electrode materials are: AISI 304 type stainless steel mesh with a wire thickness of 0.5 mm and mesh width of 1.6 mm, platinized titanium mesh with a wire thickness of 1.2 mm and mesh width of about 2 mm, copper plate with a thickness of 0.8 mm, and REF401 type (Radiometer Analytical) Saturated Calomel Electrode (SCE). 2.2. Sample preparation 2.2.1. Pre-treatment of the reinforcing steel bars Steel bars were cut into 120 mm long pieces from a longer bar. The newly cut ends were slightly ground to remove sharp edges. Before being embedded in mortar, the bars were pretreated to remove any rust to get a uniform surface condition by using a newly prepared chemical cleaner solution (1:1 diluted HCl + 3 g/l urotropine). 2.2.2. Mortar with embedded reinforcing steel MHT was incorporated in two different ways: (1) as one of the 187

Cement and Concrete Research 100 (2017) 186–202

Z. Yang et al.

Fig. 2. Dual-role mechanism of MHTs in reinforced concrete exposed to chloride ions: capturing Cl- and releasing inhibitors to protect the reinforcing steel in parallel.

be found in our previous study [27], in which the MHT's effect on the properties of fresh and hardened mortar was also tested. They were then demolded and moved to a fog room. The specimens were taken out of the curing room at 28 days. After wiping off excess water, a PVCpond (about 50 mm in depth) was glued to the top surface of the specimens. All vertical surfaces of the mortar specimens were carefully coated with silicone sealant to prevent water evaporation during the test. When the silicone sealant had completely hardened after 24 h, the specimens were transferred to a plastic tray, in which tap water was added in order to keep the bottom surface of the specimens in a wet condition. The PVC-ponds of the specimens were filled with a saturated Ca(OH)2 solution and covered by plastic film for 3 days to prevent drying. Then the Ca(OH)2 solution was removed and the PVC-ponds were rinsed using tap water before being filled by the chloride solution. The chloride solution used was a mixed solution of 0.6 M NaCl and 0.4 M CuCl2 for accelerated chloride migration test and 16.5% NaCl for cyclic wetting-drying and diffusion test. The pond volume was 200 ml and the exposed area is 44 cm2 as illustrated in Fig. 3.

mixing components in mortar; (2) as part of cement paste coating of the reinforcing steel. For (1), The mortar samples were prepared with a constant (cement + MHT) content, a constant water-to-(cement + MHT) mass ratio of 0.50, a constant sand-to-(cement + MHT) mass ratio of 3 and a MHT-to-(cement + MHT) mass ratio of 0%, 5%, 10%, respectively. In fact, MHT was used to replace cement. This means the water-cement (w/c) ratio was 0.50, 0.53 and 0.56 respectively for the three replacement levels. For (2), a cement paste was pre-mixed with water-to-(cement + MHT) mass ratio of 0.4 and MHT to replace 20% cement mass resulting in a w/c ratio of 0.5. Then the cement paste was uniformly applied over the surface of the steel bars with approx. 1.5–2.0 mm thickness. The coated bars were subsequently stored in a fog room and cured for 24 h before being embedded into mortar. As shown in Fig. 3, mortar specimens were prisms of 40 × 40 × 110 mm3 in which a pretreated 120 mm long steel bar was embedded at a cover depth of 10 mm. The bar was isolated from the air/mortar interface using insulating heat shrink tubing. The geometric length and surface area of the steel bar that were exposed to mortar were 60 mm and 1560 mm2 respectively. Eight groups of mortar specimens embedded with reinforcing steel bars were prepared as shown in Table 1. The mortar prisms were cast in the molds and sealed by plastic film for 24 h in the normal lab condition. The detailed mixing procedure can

2.3. Anti-corrosion performance evaluation The anti-corrosion performance of two MHTs was evaluated by exposure of mortar specimens to chloride solution through three test

Fig. 3. Schematic illustration of reinforced mortar specimen.

188

Cement and Concrete Research 100 (2017) 186–202

Z. Yang et al.

Table 1 Preparation of eight groups of mortar specimens with/without addition of MHTs. Sample group

Ref.

Ref. (coating)

Remarks

Reference sample Steel bars coated with without addition pure paste; no MHTs in bulk mortar of MHTs

MHT-pAB (coating) MHT-NO2 (coating) MHT-pAB (5%) MHT-pAB (10%) Steel bars coated with paste containing MHTs; no MHTs in bulk mortar

MHT-NO2 (5%) MHT-NO2 (10%)

MHTs incorporated in bulk mortar to replace 5% or 10% mass of cement respectively; no coating on steel bars.

Fig. 4. Experimental setup for anti-corrosion performance evaluation of MHTs (Note: specimens under accelerated migration test were connected to a power supply; no power supply connection for cyclic wetting-drying and diffusion test).

catholyte (ponding solution) and thus prevent the negative effect of OH– on chloride migration. Normally, the main cathodic reaction under the applied electrical field is the conversion of water into hydrogen gas and OH– (Eq. (1)), although the oxygen reduction may also occur (Eq. (2)):

regimes based on open circuit potential (OCP) and linear polarization resistance (LPR) measurements: 1. An accelerated chloride migration test 2. A cyclic wetting-drying test 3. A diffusion test

2H2 O + 2e− → H2 + 2OH−

(1)

Unless otherwise stated, all potentials used in this paper are referred to Saturated Calomel Electrode (SCE). Fig. 4 shows the experimental setup for the three series of tests. The specimens were set on a stiff plastic grid (about 20 mm height) in a plastic tray. Between specimens and the plastic grid a sponge cloth and a stainless steel mesh were positioned. A sufficient amount of distilled water was poured into the plastic tray to maintain moist condition of the sponge cloth throughout the tests. Once all the relevant electrodes were set in place, the PVCpond of the specimen was covered by parafilm to prevent evaporation of the chloride solution. A more detailed description for each of the three test regimes is presented in the following sections.

O2 + 2H2 O + 4e− → 4OH−

(2)

2.3.1. Accelerated chloride migration test 2.3.1.1. The principle of accelerated chloride migration test. The principle of accelerated chloride migration test is to apply an external electrical potential across the mortar specimen forcing the chloride ions to migrate through the specimen. In this study, a modified accelerated migration test based on Andrade's integral corrosion test method [28–31] was employed. Fig. 5A schematically illustrates the electrode arrangement and cross-sectional view of a mortar specimen. The copper plate acting as cathode was submerged in chloride solution, while a stainless steel mesh acting as anode was placed underneath the specimen. The stainless steel mesh was also used as the counter electrode (CE) when LPR measurement was performed. A watersaturated sponge was sandwiched between the stainless steel mesh and the specimen to ensure good electrical contact. A potential drop of 6 V was applied between copper cathode and stainless steel anode in order to accelerate the migration of the chloride ions through mortar towards the embedded rebar. The ponding solution was a mixed solution of 0.6 M NaCl and 0.4 M CuCl2 [31,32]. The choice of CuCl2 solution as one of the chloride sources combined with the copper plate as cathode (Cu/CuCl2) aims to avoid the production of OH– in the

2.3.1.2. OCP and LPR measurements. OCP and LPR measurements were carried out using a Solartron analytical SI1287 potentiostat coupled with an eight-channel 1281 multiplexer. As can be seen from Fig. 5, a two-reference electrode system was adopted for OCP measurement. RE1, i.e., Saturated Calomel Electrode (SCE) was used for manually measuring the OCP, while the platinized titanium mesh (RE2) was used as another reference electrode for automatically measuring the OCP through the custom designed computer program. A conventional threeelectrode electrochemical system was employed for LPR measurements with the steel bar as working electrode (WE), stainless steel mesh as the counter electrode (CE) and platinized titanium mesh (RE2) as the reference electrode. When a certain voltage is applied to a reinforced mortar/concrete specimen, polarization of the reinforcing steel occurs even if the steel is not connected to the power source. This polarization may in turn cause the passive film on the reinforcing steel to become unstable and vulnerable to be destroyed upon the arrival of chloride ions. In order to minimize such a polarization effect, some pretests were performed by applying 6 V potential to a reinforced mortar specimen. Fig. 6 shows a typical potential evolution curve of the rebar by switching on/off the

–

The production of OH in the catholyte would result in a decrease of the transference number of Cl− in the mortar, which consequently slows down its penetration rate and even further possibly influences the chloride threshold for initiating corrosion [29,33]. When Cu/CuCl2 is used, copper reduction (Eq. (3)) replaces OH– production as the main cathodic reaction due to the higher reduction potential of copper ions relative to that of water [34].

Cu2 + + 2e− → Cu

189

(3)

Cement and Concrete Research 100 (2017) 186–202

Z. Yang et al.

Fig. 5. Schematic cross-sectional view of the experimental setup for accelerated chloride migration test (A) and for cyclic wetting-drying and diffusion test (B). WE (working electrode): Reinforcing steel bar; CE (counter electrode): Stainless steel mesh and also used as the Anode with the copper plate as the Cathode; RE1 (reference electrode 1): Saturated calomel electrode; RE2 (reference electrode 2): Platinized titanium mesh. Once corrosion was detected, a 30 mm slice was cut off from the left side of the prism for chloride analysis (A).

OCP might lead to misunderstanding in some cases either because the potential shift may not be detected or because it may not really account for a significant activity. Such uncertainties promote the need for use of the corrosion current as an additional parameter. The corrosion current (Icorr) can be calculated via polarization resistance (Rp) obtained by LPR measurement according to the Stern-Geary equation (Eq. (4))

power. As can be seen, when the power is switched on, a potential difference between the rebar and SCE reference electrode is measured. This is due to the polarization and the Ohmic drop between the electrodes. However, when the power is switched off, the potential drops back and recovers the initial value in about 45–90 min, which suggests that the potential difference measured with “power on” status is mainly related to the Ohmic drop. The “power off” polarization is < 100 mV (see potential difference in 90 min in embedded part of Fig. 6), which is considered not to significantly change the steel behavior with regard to corrosion initiation. This pretest suggested that the OCP and LPR measurements could be done in a near-natural condition with limited polarization effects after switching off the power for 45–90 min. In the real experiments, the OCP was manually recorded versus SCE, two or three times per day after a 45–90 min waiting period as well as automatically recorded versus RE2 three time per day exactly after waiting for 90 min. The LPR was measured once per day upon finishing the first OCP measurement, 90 min after switching off the power. Prior studies have shown that the OCP (i.e., corrosion potential, Ecorr) can provide qualitative information on the probability of corrosion [3,35,36] and is usually employed to indicate the occurrence of depassivation (i.e., corrosion initiation). However, the indication from

I corr = B Rp

(4)

where B is the proportionality constant and a typical value of 26 mV is normally used for active corrosion and 52 mV for passive steel in concrete [37]. Icorr is often expressed as corrosion current density (icorr) when the surface area of the steel is taken into account. In this study, an observable shift in the evolution of Ecorr as well as icorr over time were set as criteria to identify the moment of the depassivation of the embedded steel bars. As a quantitative measure for shifting criteria, the empirical boundary values of Ecorr = − 350 mV (SCE) and icorr = 0.1 μA/cm2 were adopted [30,38]. This is: if Ecorr is found to be more negative than −350 mV and/or icorr higher than 0.1 μA/cm2, it can be considered that corrosion has been initiated and active corrosion is developing. For OCP measurement using the platinized titanium mesh (Ti, RE2) as the reference electrode, however no well-established 190

Cement and Concrete Research 100 (2017) 186–202

Z. Yang et al.

Fig. 6. Typical potential evolution of the rebar embedded in mortar by switching on/off the power supply at 6.0 V.

groups: (1) corroding specimens; (2) non-corroding specimens. For corroding specimens, once depassivation was detected, the chloride content at rebar depth was analyzed to investigate the effect of MHTs on the CT. For non-corroding specimens, the chloride content at three depths (see sampling below) was analyzed and chloride profiles were obtained accordingly.

boundary value of Ecorr is applicable and the results were viewed as additional information to OCPs measured by SCE. Once depassivation was detected, the specimens were disassembled from the experimental setup. Then, different procedures were followed for MHT in bulk application and coating application. For specimens in which MHT was incorporated as a mixing component (bulk), a 30 mm slice was cut off from the left side of mortar prism (see Fig. 5A and Fig. 8A later). This 30 mm mortar slice was used for chloride threshold (CT) analysis as will be discussed below in Section 2.4. The newly cut surface of the mortar prism was immediately sealed using Bison® silicone kit and the pond on the top of the prism was re-glued in which chloride solution was refilled. The mortar prism was then placed back in the experimental setup in order to continuously monitor the corrosion process after depassivation without applying a potential until around 30 days. For specimens in which MHT was applied as coating of the rebar, two specimens were broken upon depassivation from which CT at the rebar depth was analyzed (see Section 2.4). The other three or four specimens were left for continued monitoring of the corrosion process after depassivation without applying a potential.

2.4.1. Sampling Depending on the way that MHT was applied, different sampling methods were employed. As shown in Fig. 7A, for specimens in which MHT was incorporated in mortar bulk (App. #1, bulk), a 30 mm slice was cut off from the mortar prism for CT analysis (for corroding specimens) and chloride profiling (for non-corroding specimens). This 30 mm slice was thought to have the same chloride penetration depth as that in the main part of the mortar prism [40]. In this way, the main part of the prism could be kept for continued monitoring the corrosion process after corrosion initiation. For CT analysis (Fig. 7A), a small amount of mortar powder (about 2 g) at the rebar depth (7.5–12.5 mm) was collected by dry drilling (drill diameter: 5 mm) from the 15 mm thick part of the 30 mm slice that was originally close to the rebar. Considering that in the “coating” specimens the material composition of mortar at the interfacial zone of the embedded steel bar (containing MHTs) is different from the rest of the mortar (without MHTs), the specimens in which MHTs were applied as coating of the rebar (App. #2, coating) were broken upon depassivation and the mortar powder (about 2 g) was drilled directly from the interfacial zone of the rebar surface at the depth of 7.5–12.5 mm (Fig. 7B). To analyze the chloride profile of the non-corroding specimens, the same sampling methods with respect to the 30 mm slice (App. #1, bulk) and the broken specimens (App. #2, coating) were used, while mortar powders were collected at three depths (about 2 g for each) below the exposed surface: 2–7 mm (top part), 7.5–12.5 mm (middle part) and 27.5–32.5 mm (bottom part).

2.3.2. Cyclic wetting-drying and diffusion test The cyclic wetting-drying test was performed by one day of wetting by chloride solution followed by six days of drying under normal laboratory conditions. The OCP/LPR measurements were conducted during the wetting cycles. For the diffusion test, the OCP/LPR were measured on a weekly basis. A schematic illustration of the electrode arrangement and cross-sectional view of the reinforced mortar under the cyclic wetting-drying and diffusion test is given in Fig. 5B. The ponding solution used for these two tests was 16.5% NaCl solution in accordance with NordTest method NT Build 443 [39]. The depassivation of the embedded steel under the wetting-drying cycles and diffusion condition was determined following the same criteria as used for the accelerated migration test. Once depassivation was detected, the same handling procedure for the mortar specimens as in the accelerated migration test was implemented. The CT for corrosion initiation was subsequently analyzed in a similar fashion. In addition, chloride profiles of non-corroding specimens from diffusion test were analyzed after the 30 weeks' test period.

2.4.2. Analysis The total (acid-soluble) chloride content of the mortar powders were analyzed photometrically using a Spectroquant® NOVA 60 photometer with a chloride test kit following the procedure described elsewhere [27]. The cement content was calculated according to recommended methods [41,42] by subtracting the weight of insoluble matter from the initial weight of the oven-dried mortar powder and correcting for hydration water assuming 18% of the acid soluble

2.4. Chloride analysis Mortar specimens for chloride analysis were divided into two 191

Cement and Concrete Research 100 (2017) 186–202

Z. Yang et al.

Fig. 7. Schematic illustration of sampling area for chloride threshold (CT) analysis in mortar prism with an embedded steel bar: (A) MHTs incorporated in the bulk mortar; (B) MHTs incorporated in the coating on the rebar.

from the two figures, the OCP evolution obtained from the two different reference electrodes follows the same trend, which suggests that the use of platinized titanium mesh as a reference electrode is feasible in this research. It is interesting to note that in the later stage of the test, after the specimen was disconnected from the power for several days, the OCPs of all the specimens except for MHT-NO2(10%) started to increase and went on to become more positive than the empirical passive-active “boundary” value of Ecorr = − 350 mV. This may indicate that the steel bars are experiencing repassivation. The icorr evolution obtained by LPR measurements as shown in Fig. 10 further confirmed the observation from the OCP measurements. The mortar samples were broken after finishing the accelerated migration test and the occurrence of corrosion pits was verified by visual inspection. Fig. 11 presents examples of pitting corrosion found on the steel bars. More and severe corrosion pits were found on the steel

cementitious mass was hydration water in addition to subtraction of the weight of MHTs from the total acid soluble mass.

3. Results and discussion 3.1. Accelerated chloride migration test The OCP (i.e., Ecorr) evolution of steel bars obtained versus SCE and Titanium reference electrode during the accelerated chloride migration test are shown respectively in Figs. 8 and 9, where each value has been averaged from the results of three to four parallel specimens. The time scale of the x-axis in the two figures is the test running time which includes the “power on” time and the “waiting” time (i.e., “power off” time) before depassivation and the continued time after depassivation when the specimens were disconnected from the power. As can be seen

Fig. 8. The OCP/Ecorr (average of 3–4 specimens) evolution obtained using SCE as reference electrode for rebars embedded in mortar specimens in the accelerated chloride migration test (Note: the time scale of the x-axis is the real test running time which includes the “power on” time and the “waiting time” in the early stage before depassivation was detected and the continued time after depassivation in the later stage when the specimens were disconnected from the power supply).

192

Cement and Concrete Research 100 (2017) 186–202

Z. Yang et al.

Fig. 9. The OCP/Ecorr (average of 3–4 specimens) evolution obtained using the platinized titanium mesh as reference electrode for rebars embedded in mortar specimens in the accelerated chloride migration test (Note: the time scale of the x-axis is the real test running time which includes the “power on” time and the “waiting time” in the early stage before depassivation was detected and the continued time after depassivation in the later stage when the specimens were disconnected from the power supply).

same bulk w/c gave almost the same value. On the other hand, both MHT-pAB(coating) and MHT-NO2(coating) showed higher td values than Ref. while they have higher w/c. The chloride transport can be related to the combined effects of the w/c, the amount of cementitious phases as well as the added MHTs which are available to react with chloride hindering its ingress. Consequently, these observations may indicate the important role of MHT's active chloride binding capacity in slowing down the chloride ingress. The lower td values of the specimens incorporated with a higher percentage of MHTs, i.e., MHT-pAB(10%) and MHT-NO2(10%) were likely due to a higher chloride migration coefficient resulting from a higher w/c ratio. Mortars with MHT added while increasing the w/c also have a higher porosity as described elsewhere [27]. Fig. 14 shows chloride threshold (CT) values which were averaged from two or three specimens and are expressed in total chloride content by weight of cement. As can be seen from Fig. 14, compared to the Ref., the CT of MHT specimens was increased to a higher level, being 67% (1.55 vs 0.93), the highest percentage of increase for MHT-pAB (coating) and 28% (1.19 vs 0.93) the lowest for MHT-NO2(5%). Considering the proposed dual function working mechanism of MHT (see Fig. 2), the increased CT level resulting from the incorporation of MHTs could be ascribed to the combined effect of active chloride binding of MHTs and the inhibiting effect of the simultaneously released inhibitors. In addition, if MHTs incorporated in the same dosage (i.e. 5%, 10% in bulk and 20% in coating), specimens incorporated with MHTpAB exhibited higher CT values than those with MHT-NO2, likely owing

bar from MHT-NO2(10%) samples which correlate to a longer period of active corrosion as suggested by its higher icorr for a longer time (Fig. 10). Commonly, hydroxyl ions exhibit higher mobility than other ions (in particular, Cl− in this case) in aqueous solution [43]. Therefore, it can be imagined that without external driving forces (electrical potential in this case), for the same corrosion pit during the same time range, more hydroxyl ions (from pore solution) than chloride ions can enter into the pit. The higher mobility of hydroxyl ions also results in a reduction of the mobility of chloride ions at these sites [29,33,44,45]. Consequently, a high local alkalinity and high OH−/Cl− ratio could be reached in the pit and that may subsequently result in repassivation (occurred in the later stage of the test). In addition, for specimens with MHTs, the inhibitors (i.e., −pAB and -NO2) that were released from MHTs via the chloride exchange reaction could also enter into the pit and contribute to the repassivation. For MHT-NO2(10%), the higher porosity (tested in [27]) may be the dominating factor controlling the chloride migration and/or diffusion process and consequently caused the severe pitting which did not allow for repassivation. Figs. 12 and 13 respectively give the representative curves of Ecorr and icorr evolution over the first 120 “power on” testing hours (here the “waiting time” during the test has been subtracted and is not included in the x-axis) during which the depassivation was detected. Table 2 gives the average depassivation time obtained from OCP/LPR measurements. As can be seen from Table 2, compared to Ref., MHT-pAB (5%) and MHT-NO2(5%) presented a relatively higher time-to-depassivation (td) and higher w/c, while Ref.(coating), which has the

Fig. 10. The corrosion current density (average of 3–4 specimens) evolution for rebars embedded in mortar specimens in the accelerated chloride migration test (Note: the time scale of the x-axis is the real test running time which includes the “power on” time and the “waiting time” in the early stage before the depassivation and later the “power off” time after depassivation had been detected).

193

Cement and Concrete Research 100 (2017) 186–202

Z. Yang et al.

mortar, was found to increase the CT as well. In addition to the higher chloride binding capacity resulting from the higher added amount of MHTs, the relatively higher concentration of released inhibitors (i.e, −pAB and -NO2) is considered to have contributed to raise the CT level. For the Ref.(coating), in which the rebar was coated with a layer of cement paste, the formation of a more dense and cement rich layer at the steel-paste interface relative to the steel-mortar interface of Ref. is believed to be the main reason for the increase of CT [41,44,46], but the increase percentage of CT being 7% (0.93 vs 1.00) is not as high as those of MHT specimens.

3.2. Cyclic wetting-drying test The recorded evolution of Ecorr and icorr are given respectively in Figs. 15 and 16. It should be noticed that since the Ecorr evolution obtained from the platinized Ti reference electrode followed the same trend as that of SCE, as observed in the accelerated chloride migration test (see Figs. 8 and 9), the Ecorr evolution results via platinized Ti reference electrode were not included in the following discussion. As can be seen from Figs. 15 and 16, only few of the rebars embedded in these eight groups of mortar specimens showed active corrosion (Ecorr ˂ −350 mV and icorr > 0.1 μA/cm2) at any point during the 30 weekly wetting-drying cycles. Specifically, depassivation had been detected in the 26th week for one out of the three Ref. specimens, in the 27th week for two out of five Ref.(coating) specimens (one shown and the other one was broken for CT analysis, so data are not shown in the figure), in the 22nd week for one out of three MHT-pAB(10%) specimens, in the 29th week for one out of the four MHT-NO2(5%) specimens and in the 18th and 20th week for two out of the four MHTNO2(10%) specimens. The rebars embedded in MHT-pAB(5%), MHTpAB(coating) and MHT-NO2(coating) mortar specimens maintained their passive state and no active corrosion was detected. Table 3 gives an overview of the depassivation time and probability of corrosion of the rebars. The probability of corrosion is the percentage of corroding specimens out of total number of specimens in each group. The following order of the time to depassivation can be established in terms of the first appearance of active corrosion (Table 3): MHT−NO2(10%) < MHT−pAB(10%) < Ref. ≤ Ref. (coating) < MHT−NO2(5%) < MHT−pAB(5%) ~ MHT−NO2(coating) ~ MHT−pAB(coating) (no corrosion detected) In general, the order of the time to depassivation is consistent with that obtained from accelerated chloride migration test except for MHTpAB(10%) which performed similarly as Ref. and Ref.(coating) in the accelerated migration test (Table 2). As explained in Section 3.1, the anti-corrosion behavior of MHT specimens can be explained in terms of

Fig. 11. Examples of the pits found on bars (Ref., top, and MHT-NO2(10%), bottom) after the accelerated chloride migration test.

to the higher chloride binding/exchange capacity of the MHT-pAB than MHT-NO2, which in turn resulted in less amount of -NO2 released from MHT compared to the -pAB. Therefore, a lower amount of bound chloride as well as a lower amount of released inhibitor in the case of MHT-NO2 may be responsible for the lower CT values compared to mixes with MHT-pAB. Moreover, regardless of the two application ways, a higher mixing percentage of both MHTs, that is to say 10% relative to 5% in mortar and 20% in the coating relative to 10% in

Fig. 12. The Ecorr/OCP evolution for rebars embedded in mortar specimens during the early stage of the accelerated chloride migration test; the “waiting time/power off” has been deducted and is not included in the time scale of the xaxis.

194

Cement and Concrete Research 100 (2017) 186–202

Z. Yang et al.

Fig. 13. The corrosion current evolution for rebars embedded in mortar specimens during the early stage of the accelerated chloride migration test (Note: the “waiting time/power off” has been deducted and is not included in the time scale of the x-axis).

measured as 0%. Cs is often used as one of the major parameters to estimate the chloride binding capacity of cementitious materials with higher values implying higher chloride binding [48,49]. The resulting Cs and Dapp values are presented in Table 4. As can be seen from Table 4, all mortar with MHTs in their bulk had higher Cs than Ref. mortar, which confirms the important role of MHTs in terms of higher chloride binding capacity. However, 10% replacement gave lower Cs and higher Dapp values than 5% replacement. Considering the higher amount of addition of MHTs could have bound more chlorides and the w/c is higher for 10% replacement than for 5% replacement (Table 2), this suggests a higher w/c ratio, with less binding per unit volume of mortar, could have counteracted the higher chloride binding capacity of MHTs. In any case, Cs with MHT is always higher than without MHT. Additionally, it has to be noted that MHT-pAB(5%) presented the highest Cs value along with the lowest Dapp value relative to all the other specimens, indicating a more active chloride binding as well as a stronger resistance to chloride penetration, despite a higher w/c (0.53 vs 0.50) than the reference mortar. The coating specimens, as expected, showed similar values of both Cs and Dapp as that of Ref., presumably due to the same material composition in bulk mortar, hence almost the same pathway for chloride transport before arriving at rebar depth. In this regard, the coating seems to have more influence on raising the chloride threshold against chloride penetration.

the combined effects of the chloride binding capacity and the inhibiting effect from the released inhibitors of MHTs, which both promote longer time to corrosion initiation, while the higher w/c for 5% and 10% MHT additions promotes faster chloride transport [47] and reduces the time to corrosion initiation. CT values for corroding specimens are presented in Fig. 17. As found in the accelerated chloride migration test, the incorporation of MHTs increased the CT level independent of whether the corrosion was detected earlier (in the cases of MHT-NO2(10%) and MHT-pAB(10%)) or later (in the case of MHT-NO2(5%)) than Ref. In particular, the CT values of MHT-pAB(10%), MHT-NO2(10%) and MHT-NO2(5%) were increased respectively by 67% (3.18 vs 1.90), 64% (3.12 vs 1.90), and 36% (2.59 vs 1.90) relative to Ref. In addition, the CT was also found increased due to the application of the cement paste coating, i.e., Ref (coating), but the percentage of increase being 15% (2.18 vs 1.90) relative to Ref. is less than those of specimens with MHT in the bulk mortar. 3.3. Diffusion test Base on the results obtained from OCP/LPR measurements (data not shown), no active corrosion (Ecorr ˂ −350 mV and icorr > 0.1 μA/cm2) had been detected in any of the specimens during the 30 weeks' diffusion testing. All the specimens were broken after 30 weeks for visual inspection in order to confirm absence of corrosion. Fig. 18 shows the average chloride penetration profiles of the mortar specimens obtained from the diffusion test. It is to be noted that variation between specimens were small (indicated by error bars). The measured chloride profiles were fitted to Eq. (5) to yield surface chloride contents (Cs) and non-steady-state diffusion coefficients (i.e., the apparent chloride transport coefficient, Dapp) from the error function solution of Fick's second law,

x ⎛ ⎞ C (x, t) = Cs − (Cs − Ci ) erf ⎜ 4Dapp t ⎟ ⎝ ⎠

3.4. The applied test methodology and the effect on chloride diffusion/ migration and chloride threshold Based on the results and discussion above, the modified accelerated migration test provides an alternative means of classifying the effect of new corrosion mitigating admixtures in a relatively short term. In addition to information obtained on chloride threshold (CT) as well as the time to corrosion initiation, the corrosion process after corrosion initiation can also be continuously monitored. Therefore, it gives more insight in terms of reinforcement corrosion than testing only chloride penetration. Compared to the accelerated migration test, the diffusion test, on the other hand, is closer to real-world scenarios in normal circumstances. However, it is very time consuming, often requiring months or years to obtain the results [50]. It has to be noted that the

(5)

where C (x, t) is the chloride concentration at depth x below the exposed surface after time t. The initial chloride concentration Ci was

Table 2 The time needed to depassivate the reinforcing steel in the accelerated chloride migration test and bulk w/c ratios (numbers in the parentheses are statistical variation). Sample group

Ref.

Ref. (coating)

MHT-pAB (5%)

MHT-pAB (10%)

MHT-pAB (coating)

MHT-NO2 (5%)

MHT-NO2 (10%)

MHT-NO2 (coating)

Depassivation time (h) by OCP Depassivation time (h) by LPR w/c (bulk)

72 ( ± 3) 80 ( ± 7) 0.50

73 ( ± 3) 78 ( ± 8) 0.50

87 ( ± 3) 101 ( ± 8) 0.53

80 ( ± 15) 78 ( ± 8) 0.56

100 ( ± 10) 118 ( ± 8) 0.50

80 ( ± 7) 90 ( ± 8) 0.53

65 ( ± 8) 84 ( ± 14) 0.56

87 ( ± 3) 92 ( ± 8) 0.50

195

Cement and Concrete Research 100 (2017) 186–202

Z. Yang et al.

Fig. 14. Measured chloride threshold values from the accelerated chloride migration test.

the chloride contents at the rebar depth after 30 weeks' test are lower than those for the non-corroding specimens from wetting-drying test, but the differences are small. This finding, to a certain degree, is in agreement with the work of Arya et al. [54], who found that after the first cycle of wetting-drying, chloride penetration away from the nearsurface zone of concrete is largely diffusion controlled and the increase of chloride penetration reduces with an increasing number of wettingdrying cycles. Furthermore, Arya et al. [54] found that diffusion coefficients after 24 weeks total immersion and 24 wetting-drying cycles are broadly similar.

diffusion coefficients obtained from diffusion tests may possibly decrease with exposure/testing time because of factors such as continuing cement hydration, chloride exchange reactions involved in MHT specimens and precipitation of insoluble salts [51]. Regarding the cyclic wetting-drying test, which may be considered more natural than the accelerated migration test, it turned out to be relatively slow since corrosion was only detected in few specimens during the 30 weeks testing period. The CT can be dependent on a number of factors determined by the application methods and mixing dosage of MHTs, as well as the properties of MHTs such as the type and chloride binding capacity and inhibiting effect from the simultaneously released inhibitors. Varying the test method could also influence the CT values. Rearranging Fig. 14 and Fig. 17, comparing the influence of test methods on CT (particularly corroding specimens) results into Fig. 19, in which the chloride contents at rebar depth of the non-corroding specimens from both wettingdrying and diffusion test after 30 weeks were also included. As can be seen, for all corroding specimens, the CT values obtained from wettingdrying test were found to be more than twice as high as those obtained from the accelerated migration test. A few possible explanations for this observation are explored as following. Firstly, in the accelerated migration test, some of the negatively charged inhibitive ions (i.e., −pAB and −NO2) released via the chloride exchange reactions could also migrate away from the rebar surface under the electric field before interacting with the passive film. Secondly, the released inhibitive ions which are located in the vicinity of the rebar surface in more natural conditions (i.e., wetting-drying and diffusion) can be retained and have more chance and space exerting their inhibiting effect to a larger degree. In this respect, the CT may be increased markedly in conditions of wetting-drying and diffusion versus electrically accelerated migration. Thirdly, the higher transport rate of chloride ions as well as released anionic inhibitor ions caused by the electrical field may consequently result in less binding [52,53] and less inhibition than under the wettingdrying and diffusion conditions. The chloride contents of the non-corroding specimens from 30 weeks wetting-drying test were found to be lower than those of the corroding specimens (if any), but the differences in most cases are not significant. In the cases of MHT-pAB(10%) and MHT-NO2(10%), some specimens however presented a bit higher chloride contents than the CT values detected earlier from corroding ones. This is understandable since the non-corroding specimens experienced more wetting-drying cycles and the longer testing time may account for the increase of the chloride content at the same depth in mortar (i.e., rebar depth). For the diffusion test (no corrosion detected for any of the tested specimens),

3.5. Effect of MHT on time to corrosion initiation As discussed above, the incorporation of the two MHTs, in particular the MHT-pAB (5%), bulk application and the coating application for both MHT-pAB and MHT-NO2 resulted in an increased chloride threshold (CT) and prolonged corrosion initiation compared to the reference specimens. From environmentally friendly point of view (−NO2 is toxic), only MHT-pAB, in both applications, i.e., MHT-pAB(5% bulk) and MHT-pAB(coating), will be explored in the following discussion. Aiming to give a picture of the effect of MHT on time to corrosion initiation of concrete structures in chloride contaminated environment, a simplification of the DuraCrete [55,56] modeling approach was adopted, by which only deterministic calculations were made. 3.5.1. The DuraCrete model The DuraCrete model [55] for chloride induced corrosion is based on the concept of chloride penetration by diffusion and initiation of corrosion when the chloride content at the surface of the reinforcing steel has reached a critical “threshold” value. Corrosion initiation due to chloride ingress is specified as the limit state for service life design. According to the DuraCrete model, the time-development of chloride profiles can be approximated as Eq. (6):

x ⎞ C (x , t ) = Cs − (Cs − Ci ) erf ⎛⎜ ⎟ 4 kD (t ) t ⎠ ⎝

(6)

where C(x, t), Cs, Ci, x and t have the same meaning and unit as used in Eq. (5). k is the environmental coefficient by which D(t) is multiplied to obtain the chloride diffusivity in a real structure. D(t) is the chloride diffusion coefficient, which is a function of time since the rate of chloride penetration into the mortar/concrete decreases with time. This is due to factors such as continuing cement hydration as described in Section 3.4. The time dependency of D(t) is given by Eq. (7): 196

Cement and Concrete Research 100 (2017) 186–202

Z. Yang et al.

Fig. 15. The Ecorr/OCP evolution for rebars embedded in mortar specimens during 30 weekly cyclic wetting-drying test.

D (t ) = D0 (t0 t )n

(7)

the diffusion test (Table 4). The reference time t0 is counted as 105 days which is half-way of the 30 weeks exposure. The ageing coefficient n is set at 0.30 referring to the DuraCrete specified value in “submerged” conditions for Portland cement concrete [55]. By inserting all the above mentioned values into Eqs. (6) and (7), Figs. 20 and 21 give chloride profiles in mortar specimens with/without MHT-pAB. Fig. 20 was calculated for a given cover depth of 50 mm and Fig. 21 for a given time-to-corrosion initiation (TTC) of 50 years when CT was reached at the cover depth. Table 5 gives an overview of the results calculated using the DuraCrete model. As can be seen from Table 5, with a given cover depth of 50 mm, TTC is significantly

where D0 is the diffusion coefficient at reference time t0 and n is the ageing coefficient (0 < n < 1). 3.5.2. Application of the model Following the methodology described above, the parameters in Eqs. (6) and (7) were defined based on the experimental results obtained in this study. For example, Cs values from the diffusion test (Table 4) were taken for each mix. Ci was measured as 0%. The chloride threshold (CT) values were taken from the migration tests (Fig. 14). The k-values were set to be 1.0 for simplicity. D0 is the corresponding value of Dapp from 197

Cement and Concrete Research 100 (2017) 186–202

Z. Yang et al.

Fig. 16. The corrosion current density (icorr) evolution for rebars embedded in mortar specimens during 30 weekly cyclic wetting-drying test.

Table 3 Depassivation time and probability of corrosion of the steel bars embedded in mortar specimens during 30 weekly wetting-drying cycles (“-” means no depassivation detected at 30 weeks). Sample group

Ref.

Ref. (coating)

MHT-pAB (5%)

MHT-pAB (10%)

MHT-pAB (coating)

MHT-NO2 (5%)

MHT-NO2 (10%)

MHT-NO2 (coating)

Depassivation time (week) Probability of corrosion (%)

26 33

27 40

– 0

22 33

– 0

29 25

18, 20 50

– 0

198

Cement and Concrete Research 100 (2017) 186–202

Z. Yang et al.

Fig. 17. Measured chloride threshold values for corroding specimens at the depassivation time during wetting-drying cyclic test.

Fig. 18. Measured chloride profiles in 16.5% NaCl solution after 30 weeks' diffusion test.

Table 4 Surface chloride content and chloride diffusion coefficient obtained by fitting the error function equation (Eq. (5)) to the measured chloride profiles after 30 weeks' diffusion test. Sample group

Ref.

Ref. (coating)

MHT-pAB (5%)

MHT-pAB (10%)

MHT-pAB (coating)

MHT-NO2 (5%)

MHT-NO2 (10%)

MHT-NO2 (coating)

Cs (% by mass of cement) Dapp (× 10− 12 m2/s)

5.3( ± 0.1) 3.0( ± 0.04)

5.3( ± 0.01) 3.0( ± 0.04)

8.1( ± 0.02) 1.6( ± 0.01)

7.0( ± 0.04) 4.0( ± 0.1)

5.3( ± 0.1) 3.1( ± 0.1)

6.9( ± 0.01) 2.8( ± 0.01)

6.5( ± 0.1) 4.5( ± 0.1)

5.4( ± 0.01) 3.0( ± 0.01)

Fig. 19. Chloride contents at rebar depth of corroding specimens from the migration test (C_M) and the cyclic wetting-drying test (C_W/D) and of non-corroding specimens from the cyclic wetting-drying (NC_W/D) and the diffusion test (NC_D) after 30 weeks.

199

Cement and Concrete Research 100 (2017) 186–202

Z. Yang et al.

Fig. 20. Chloride profiles in mortar with/without MHTpAB with a given 50.0 mm cover depth of rebar at their respective time-to-corrosion initiation (TTC, as shown in the key).

Fig. 21. Chloride profiles in mortar with/without MHTpAB at a given time-to-corrosion initiation of 50.0 years with their respective cover depth (lines and arrows show chloride threshold (CT) values of MHT-pAB(5%) and Ref. at cover depth).

time cost (coated steel needs (at least) one day of curing before casting can take place), the incorporation of MHT-pAB in the bulk instead of as a coating on the steel seems more attractive from a practical point of view. However, if further improvement is desired, a combination of the two application ways may be worth considering. Finally, it needs to be pointed out that the calculations are deterministic and consequently yield mean outcomes, i.e., TTC or cover depth in our cases. In consequence, the probability in terms of corrosion initiation at that point in time and space is 50%. Such a high probability however would be unacceptable for service life design in practice. To address this issue, various DuraCrete reports and later documents [47,57,58] provide information on how to obtain lower than 50%

increased due to the incorporation of the MHT-pAB in both applications. The highest increase which was a more than double TTC (60 years vs. 29 years) with respect to reference mortar was obtained by incorporation of 5% MHT-pAB in the bulk. Second best was 20% MHTpAB in cement paste as a coating on the rebar. On the other hand, if TTC is fixed at 50 years, the lowest cover depth (i.e., 47 mm) can be required by incorporation of 5% MHT-pAB in the bulk, which is 13 mm less than the reference case (60 mm cover required). In addition, it was found that the performance improvement due to the incorporation of MHT-pAB in the two different ways (i.e., in the bulk mortar or as a coating on the steel) is almost the same (TTC: 60 years vs. 57 years or cover depth: 47 mm vs. 48 mm). Considering the ease of execution and

Table 5 Input data and calculated results of time-to-corrosion initiation (TTC) at cover depth 50 mm and cover depth for TTC = 50 years based on DuraCrete model [55] for mortar specimens with/without MHT-pAB; calculation results rounded to nearest mm or year. Sample group

Ref.

Ref.(coating)

MHT-pAB(5%)

MHT-pAB(coating)

Cs (% by mass of cement) D0 (× 10− 12 m2/s) CT (% by mass of cement) k t0 = 105 days (year) n TTC (year, cover depth = 50 mm) Cover depth (mm, TTC = 50 year)

5.3 3.0 0.93 1.00 0.29 0.30 29 60

5.3 3.0 1.00 1.00 0.29 0.30 31 59

8.1 1.6 1.24 1.00 0.29 0.30 60 47

5.4 3.0 1.55 1.00 0.29 0.30 57 48

200

Cement and Concrete Research 100 (2017) 186–202

Z. Yang et al.

institute (M2i) (www.m2i.nl). Dr. Hartmut Fischer at TNO Materials Performance is acknowledged for his contribution in previous stages of this project. Sasol Germany GmbH is acknowledged for providing the carbonate hydrotalcites PURAL® MG 63 HT that was used in this work.

probabilities. In our study, the DuraCrete model for chloride transport was mainly used for comparing “options”. Further detailed probabilistic calculations were considered out of the present scope. 4. Conclusions

References Two Mg-Al hydrotalcites modified with intercalated inhibitive ions, para-aminobenzoate (-pAB) and nitrite (-NO2), (i.e. MHT-pAB and MHT-NO2) were incorporated into Portland cement mortar specimens with embedded reinforcing steel in two different ways: (1) as one of the mixing components in bulk mortar at two dosage levels replacing 5% and 10% mass of cement; the w/c ratio was 0.53 and 0.56 for the two replacement levels, respectively and was 0.5 for the reference mortar; (2) as part of cement paste coating of the reinforcing steel to replace 20% mass of cement. The w/c ratio was 0.5 in the paste coating (0.40 for water / (cement + MHT)). The anti-corrosion performance of the two MHTs was investigated by three test regimes including electrically accelerated chloride migration, cyclic wetting-drying and diffusion test. The results obtained from the accelerated chloride migration test revealed that both the incorporation of MHT-pAB to replace 5% cement weight in mortar and MHTs (for both MHT-pAB and MHT-NO2) applied as 20% of a cement paste coating of the reinforcing steel produced a notably extended time to corrosion initiation. This is despite the fact that the mortars with MHTs had a higher w/c, which would normally be thought to reduce the time to corrosion initiation (by increasing the chloride diffusion coefficient). The chloride threshold (CT) of MHTs specimens, which was expressed in total chloride percent by weight of cement, was found to increase to a higher level. In addition, it was found that specimens with MHT-pAB exhibited higher CT values than those with MHT-NO2. Considering the effect of increased w/c for mortar with MHTs, this interpretation of the effects of MHT is conservative. Only few specimens showed active corrosion during 30 weekly wetting-drying cycles and the CT values of the corroding specimens obtained from wetting-drying test were found to be more than twice as high as those from the accelerated migration test. For the corroding specimens, the following order of time to depassivation was established based on the first appearance of active corrosion: MHT−NO2(10%) < MHT−pAB(10%) < Ref. ≤ Ref. (coating) < MHT−NO2(5%) < MHT−pAB(5%) ~ MHT−NO2(coating) ~ MHT−pAB(coating) (no corrosion detected) No corrosion was detected after 30 weeks of diffusion testing. Chloride profile analysis of the non-corroding specimens from diffusion test revealed that MHT-pAB(5%) presented the highest Cs value along with the lowest Dapp value relative to all other specimens indicating a more active chloride binding and a stronger resistance to chloride penetration. The effects of MHT-pAB on time-to-corrosion initiation (TTC) of reinforcing steel in concrete in chloride contaminated environment was estimated based on the DuraCrete chloride transport model. It was found that the incorporation of 5% MHT-pAB in bulk mortar led to a more than double TTC with respect to reference mortar without MHTs. Under the experimental conditions of this study (including a higher w/ c), results for 10% MHT-pAB do not support this conclusion. In summary, experimental results presented in this paper served to indicate that the MHTs could be promising alternatives for preventing chloride-induced corrosion in concrete. An appropriate mixing dosage should be adopted and applied in a proper way, i.e., either incorporation of a small amount (in particular, MHT-pAB to replace 5% mass of cement) in the bulk mortar or as part of a cement paste coating of the reinforcing steel (with 20% MHT-pAB or MHT-NO2 by mass of cement).

[1] L. Bertolini, B. Elsener, P. Pedeferri, E. Redaelli, R.B. Polder, Corrosion of Steel in Concrete: Prevention, Diagnosis, Repair, Wiley-VCH Verlag GmbH & Co, Weinheim, KGaA, 2013. [2] R. Cigna, C. Andrade, U. Nürnberger, R. Polder, R. Weydert, E. Seitz, COST 521: Corrosion of Steel in Reinforced Concrete Structures-Final Report, European communities EUR20599, Luxembourg, 2002. [3] B. Elsener, D. Addari, S. Coray, A. Rossi, Stainless steel reinforcing bars–reason for their high pitting corrosion resistance, Mater. Corros. 62 (2011) 111–119. [4] C. Monticelli, M. Criado, S. Fajardo, J. Bastidas, M. Abbottoni, A. Balbo, Corrosion behaviour of a low Ni austenitic stainless steel in carbonated chloride-polluted alkali-activated fly ash mortar, Cem. Concr. Res. 55 (2014) 49–58. [5] P. Pedeferri, Cathodic protection and cathodic prevention, Constr. Build. Mater. 10 (1996) 391–402. [6] R.B. Polder, G. Leegwater, D. Worm, W. Courage, Service life and life cycle cost modelling of cathodic protection systems for concrete structures, Cem. Concr. Compos. 47 (2014) 69–74. [7] M. Ormellese, M. Berra, F. Bolzoni, T. Pastore, Corrosion inhibitors for chlorides induced corrosion in reinforced concrete structures, Cem. Concr. Res. 36 (2006) 536–547. [8] B. Elsener, Corrosion Inhibitors for Steel in Concrete: State of the Art Report, Woodhead Pub Limited, 2001. [9] M. Paredes, A. Carvallo, R. Kessler, Y. Virmani, A. Sagüés, Corrosion Inhibitors in Concrete Second Interim Report (Second Interim Report No.FL/DOT/SMO/10-531), Florida Department of Transportation, 2010. [10] A. Rosenberg, J. Gaidis, The mechanism of nitrite inhibition of chloride attack on reinforcing steel in alkaline aqueous environments, Mater. Perform. 18 (1979). [11] Z. Yang, H. Fischer, R. Polder, Modified hydrotalcites as a new emerging class of smart additive of reinforced concrete for anticorrosion applications: a literature review, Mater. Corros. 64 (2013) 1066–1074. [12] V. Rives, Layered Double Hydroxides: Present and Future, Nova Publishers, 2001. [13] R. Dhir, M. El-Mohr, T. Dyer, Chloride binding in GGBS concrete, Cem. Concr. Res. 26 (1996) 1767–1773. [14] C. Arya, Y. Xu, Effect of cement type on chloride binding and corrosion of steel in concrete, Cem. Concr. Res. 25 (1995) 893–902. [15] S.-D. Wang, K.L. Scrivener, Hydration products of alkali activated slag cement, Cem. Concr. Res. 25 (1995) 561–571. [16] U. Birnin-Yauri, F. Glasser, Friedel's salt, Ca2 Al (OH)6 (Cl, OH)·2H2O: its solid solutions and their role in chloride binding, Cem. Concr. Res. 28 (1998) 1713–1723. [17] H. Tatematsu, T. Sasaki, Repair materials system for chloride-induced corrosion of reinforcing bars, Cem. Concr. Compos. 25 (2003) 123–129. [18] O. Kayali, M.S.H. Khan, M. Sharfuddin Ahmed, The role of hydrotalcite in chloride binding and corrosion protection in concretes with ground granulated blast furnace slag, Cem. Concr. Compos. 34 (2012) 936–945. [19] L. Raki, J. Beaudoin, L. Mitchell, Layered double hydroxide-like materials: nanocomposites for use in concrete, Cem. Concr. Res. 34 (2004) 1717–1724. [20] M. Yadav, S. Kumar, L. Gope, Experimental and theoretical study on amino acid derivatives as eco-friendly corrosion inhibitor on mild steel in hydrochloric acid solution, J. Adhes. Sci. Technol. 28 (2014) 1072–1089. [21] M. Ormellese, L. Lazzari, S. Goidanich, G. Fumagalli, A. Brenna, A study of organic substances as inhibitors for chloride-induced corrosion in concrete, Corros. Sci. 51 (2009) 2959–2968. [22] H. Ashassi-Sorkhabi, M. Majidi, K. Seyyedi, Investigation of inhibition effect of some amino acids against steel corrosion in HCl solution, Appl. Surf. Sci. 225 (2004) 176–185. [23] Z. Yang, H. Fischer, J. Cerezo, J. Mol, R. Polder, Aminobenzoate modified MgAl hydrotalcites as a novel smart additive of reinforced concrete for anticorrosion applications, Constr. Build. Mater. 47 (2013) 1436–1443. [24] Z. Yang, H. Fischer, R. Polder, Synthesis and characterization of modified hydrotalcites and their ion exchange characteristics in chloride-rich simulated concrete pore solution, Cem. Concr. Compos. 47 (2014) 87–93. [25] J. Tedim, S.K. Poznyak, A. Kuznetsova, D. Raps, T. Hack, M.L. Zheludkevich, M.G. Ferreira, Enhancement of active corrosion protection via combination of inhibitor-loaded nanocontainers, ACS Appl. Mater. Interfaces 2 (2010) 1528–1535. [26] Z. Yang, H. Fischer, J. Cerezo, J.M.C. Mol, R. Polder, Modified hydrotalcites for improved corrosion protection of reinforcing steel in concrete - preparation, characterization, and assessment in alkaline chloride solution, Mater. Corros. 67 (2016) 721–738. [27] Z. Yang, H. Fischer, R. Polder, Laboratory investigation of the influence of two types of modified hydrotalcites on chloride ingress into cement mortar, Cem. Concr. Compos. 58 (2015) 105–113. [28] C. Andrade, N. Rebolledo, Accelerated evaluation of corrosion inhibition by means of the integral corrosion test, Internation Conference on Concrete Repair, Rehabilitation and Retrofitting III, CRC Press, Cape Town, South Africa, 2012, pp. 364–368. [29] M. Castellote, C. Andrade, C. Alonso, Accelerated simultaneous determination of the chloride depassivation threshold and of the non-stationary diffusion coefficient values, Corros. Sci. 44 (2002) 2409–2424.

Acknowledgements The research was carried out under project number M81.609337 in the framework of the Research Program of the Materials innovation 201

Cement and Concrete Research 100 (2017) 186–202

Z. Yang et al.

[45] C. Page, N. Short, A. El Tarras, Diffusion of chloride ions in hardened cement pastes, Cem. Concr. Res. 11 (1981) 395–406. [46] C. Page, K. Treadaway, Aspects of the electrochemistry of steel in concrete, Nature 297 (1982) 109–115. [47] G. Wegen, R.B. Polder, K.V. Breugel, Guideline for service life design of structural concrete: a performance based approach with regard to chloride induced corrosion, Heron 57 (2012) 153–168. [48] G. Glass, N. Buenfeld, The influence of chloride binding on the chloride induced corrosion risk in reinforced concrete, Corros. Sci. 42 (2000) 329–344. [49] K. Ann, J. Ahn, J. Ryou, The importance of chloride content at the concrete surface in assessing the time to corrosion of steel in concrete structures, Constr. Build. Mater. 23 (2009) 239–245. [50] J. Pacheco, R.B. Polder, Critical chloride concentrations in reinforced concrete specimens with ordinary Portland and blast furnace slag cement, Heron 61 (2016) 99–119. [51] P. Phurkhao, M. Kassir, Note on chloride-induced corrosion of reinforced concrete bridge decks, J. Eng. Mech. 131 (2005) 97–99. [52] C. Andrade, M. Castellote, C. Alonso, C. González, Non-steady-state chloride diffusion coefficients obtained from migration and natural diffusion tests. Part I: Comparison between several methods of calculation, Mater. Struct. 33 (2000) 21–28. [53] M. Castellote, C. Andrade, C. Alonso, Chloride-binding isotherms in concrete submitted to non-steady-state migration experiments, Cem. Concr. Res. 29 (1999) 1799–1806. [54] C. Arya, P. Vassie, S. Bioubakhsh, Modelling chloride penetration in concrete subjected to cyclic wetting and drying, Mag. Concr. Res. 66 (2014) 364–376. [55] DuraCrete-Probabilistic Performance Based Durability Design of Concrete Structures, Final Technical Report, Document BE95-1347/R17. The European Union-Brite EuRam III, CUR, Gouda, 2000. [56] T. Siemes, C. Edvardsen, Duracrete: service life design for concrete structures, in: M.A. Lacasse, D.J. Vanier (Eds.), Proc. of 8DCMC, 8th International Conference on Durability of Building Materials and Components, NRC Research Press, Ottawa, 1999, pp. 1343–1356. [57] Fédération Internationale du Béton, Model code for service life design, 34 fib Bull., Lausanne, Switzerland, 2006. [58] R.B. Polder, M.R. De Rooij, Durability of marine concrete structures: field investigations and modelling, Heron 50 (2005) 133–153.

[30] C. Andrade, R. Buják, Effects of some mineral additions to Portland cement on reinforcement corrosion, Cem. Concr. Res. 53 (2013) 59–67. [31] U.N.E. Spanish Standard, 83992–2 EX. Durability of concrete. Test methods. Chloride penetration tests on concrete, Part 2: Integral Accelerated Method, AENOR, Madrid, Spain, 2012. [32] M. Castellote, C. Andrade, C. Alonso, Measurement of the steady and non-steadystate chloride diffusion coefficients in a migration test by means of monitoring the conductivity in the anolyte chamber. Comparison with natural diffusion tests, Cem. Concr. Res. 31 (2001) 1411–1420. [33] M. Castellote, C. Andrade, C. Alonso, Modelling of the processes during steady-state migration tests: quantification of transference numbers, Mater. Struct. 32 (1999) 180–186. [34] A.J. Bard, R. Parsons, J. Jordan, Standard Potentials in Aqueous Solution, Marcel Dekker, New York, 1985. [35] A. Poursaee, C. Hansson, Reinforcing steel passivation in mortar and pore solution, Cem. Concr. Res. 37 (2007) 1127–1133. [36] D. Baweja, H. Roper, V. Sirivivatnanon, Relationships between anodic polarisation and corrosion of steel in concrete, Cem. Concr. Res. 23 (1993) 1418–1430. [37] C. Andrade, J. Gonzalez, Quantitative measurements of corrosion rate of reinforcing steels embedded in concrete using polarization resistance measurements, Mater. Corros. 29 (1978) 515–519. [38] C. Alonso, C. Andrade, M. Castellote, P. Castro, Chloride threshold values to depassivate reinforcing bars embedded in a standardized OPC mortar, Cem. Concr. Res. 30 (2000) 1047–1055. [39] NTBuild443, Concrete Hardened: Accelerated Chloride Penetration, NordTest, Espoo, 1995. [40] R.B. Polder, W.H. Peelen, Characterisation of chloride transport and reinforcement corrosion in concrete under cyclic wetting and drying by electrical resistivity, Cem. Concr. Compos. 24 (2002) 427–435. [41] G.K. Glass, N.R. Buenfeld, Chloride threshold levels for corrosion induced deterioration of steel, in: L. Nilsson, J. Ollivier (Eds.), 1st RILEM International Workshop on Chloride Penetration Into Concrete, Rilem Publications SARL, 1995, pp. 429–440. [42] M. Castellote, C. Andrade, Round-Robin test on chloride analysis in concrete—part I: analysis of total chloride content, Mater. Struct. 34 (2001) 532–549. [43] P. Atkins, Physical Chemistry, 5th ed., Oxford University Press, Oxford, 1994. [44] T. Yonezawa, V. Ashworth, R. Procter, Pore solution composition and chloride effects on the corrosion of steel in concrete, Corrosion 44 (1988) 489–499.

202