Cement and Concrete Research 39 (2009) 11221138

Contents lists available at ScienceDirect

Cement and Concrete Research

Critical chloride content in reinforced concrete A review

NTNU Norwegian University of Science and Technology, Department of Structural Engineering, Richard Birkelandsvei 1A, N-7491 Trondheim, Norway ETH Zurich, Institute for Building Materials (IfB), ETH Hnggerberg, CH-8093 Zurich, Switzerland Norwegian Public Roads Administration, N-0033 Oslo, Norway

a r t i c l e

i n f o

a b s t r a c t
Chloride induced corrosion as the major cause for degradation of reinforced concrete has been the subject of great research efforts over the last fty years. The present literature review summarises the state of the art by presenting the concept of the critical chloride content, discussing inuencing factors, and assessing available measurement techniques. A large number of published chloride threshold values together with the respective experimental details are collected. While today's experience is mostly based on Portland cement, more modern studies with non-traditional binders often reported contradictory results. The present literature evaluation highlights the strong need for a practice-related test method, and, in this regard, focuses especially on experimental procedures by discussing advantages and drawbacks of methods and setups. It clearly emerges that many of the setups used to determine critical chloride contents are not suited to give realistic results. 2009 Elsevier Ltd. All rights reserved.

Article history: Received 15 October 2008 Accepted 13 August 2009 Keywords: Corrosion Durability Chloride Reinforcement

Contents 1. 2. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Concept of critical chloride content . . . . . . . . . . . . . . . . . . 2.1. Denitions . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2. Expression of Ccrit . . . . . . . . . . . . . . . . . . . . . . . Critical chloride contents in the literature . . . . . . . . . . . . . . . Inuencing parameters . . . . . . . . . . . . . . . . . . . . . . . . 4.1. pH of the pore solution . . . . . . . . . . . . . . . . . . . . 4.2. Steel potential . . . . . . . . . . . . . . . . . . . . . . . . . 4.3. Steelconcrete interface . . . . . . . . . . . . . . . . . . . . 4.4. Binder type . . . . . . . . . . . . . . . . . . . . . . . . . . 4.4.1. Chloride binding capacity . . . . . . . . . . . . . . . 4.4.2. Mineral admixtures . . . . . . . . . . . . . . . . . . 4.5. Surface condition of the steel . . . . . . . . . . . . . . . . . . 4.6. Inuence of other factors . . . . . . . . . . . . . . . . . . . . 4.6.1. Moisture and oxygen content . . . . . . . . . . . . . 4.6.2. Degree of hydration and w/b ratio . . . . . . . . . . . 4.6.3. Type of chloride salt and chloride source . . . . . . . . 4.7. Overall tendencies in the literature . . . . . . . . . . . . . . . 4.8. Summary of inuencing parameters . . . . . . . . . . . . . . Experimental setups . . . . . . . . . . . . . . . . . . . . . . . . . 5.1. General considerations regarding the determination of Ccrit values 5.2. Methods for detecting active corrosion . . . . . . . . . . . . . 5.2.1. Steel potential . . . . . . . . . . . . . . . . . . . . 5.2.2. Linear polarisation resistance measurements (LPR) . . . 5.2.3. Electrochemical impedance spectroscopy (EIS) . . . . . 5.2.4. Weight loss . . . . . . . . . . . . . . . . . . . . . . 5.2.5. Other methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1123 1123 1123 1124 1126 1128 1128 1129 1129 1130 1130 1130 1130 1130 1130 1131 1131 1131 1131 1131 1131 1133 1133 1133 1133 1133 1133

3. 4.

5.

Corresponding author. Tel.: +47 735 94 538. E-mail address: [email protected] (U. Angst). 0008-8846/$ see front matter 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.cemconres.2009.08.006

U. Angst et al. / Cement and Concrete Research 39 (2009) 11221138

1123

Methods for determining the total chloride content Methods for determining the free chloride content 5.4.1. Pore solution expression . . . . . . . . 5.4.2. Leaching techniques . . . . . . . . . . 5.4.3. Ion selective electrodes . . . . . . . . . 5.5. Experimental setups in the literature . . . . . . . 6. Evaluation with regard to chloride binding . . . . . . . 7. Conclusions . . . . . . . . . . . . . . . . . . . . . . Acknowledgements . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . .

5.3. 5.4.

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

1134 1134 1134 1134 1134 1134 1135 1135 1136 1136

1. Introduction After it was recognised in the second half of the last century that chloride may induce steel corrosion in reinforced concrete structures, great research efforts have been made in this regard: over the last fty years, a considerable amount of papers has been published presenting values for critical chloride content (Ccrit) in reinforced concrete [1 53]. Considering marine exposure conditions and the extensive use of de-icing salts in many countries, chloride induced corrosion is one of the most common causes of degradation of reinforced concrete structures. Both for the design of new structures and for condition assessment of existing structures, knowledge of reliable Ccrit values is important as the remaining service life is often considered as the time required to reach the chloride threshold value at the depth of the reinforcement. In probabilistic service life modelling, Ccrit has been identied to be one of the most decisive input parameters [54,55]. Despite the multitude of studies undertaken, many aspects of chloride induced reinforcement corrosion in concrete are still incompletely understood and no general agreement on a Ccrit value has been achieved. Results reported in the literature scatter over a large range [5658]. This is not only the result of different denitions, measuring techniques and testing conditions, but also owing to the stochastic nature and complexity of initiation of pitting corrosion. Thus, often conservative values are nowadays used as critical chloride content: In European countries as well as in North America it has become common practice to limit the tolerable chloride content to or around 0.4% by weight of cement [59]. In probabilistic modelling the critical chloride content is a stochastic variable as e.g. in the b model code for service life design [60], where Ccrit is dened by a betadistribution with a lower boundary of 0.2% chloride by weight of cement and a mean value of 0.6% by weight of cement. Although there is a strong need for reliable Ccrit values, an accepted or standardised test method to measure critical chloride does at present not exist. The present review summarises the state of the art regarding critical chloride content in reinforced concrete. It is not only aimed at collecting Ccrit values reported in the literature, but also all the relevant details about experimental procedures are collated. The data is analysed with regard to factors that have an inuence on Ccrit, thereby focussing on experimental setups and measurement techniques. By highlighting advantages and drawbacks of experimental parameters, a basis for developing a test setup will be provided. In addition, the literature evaluation will reveal certain aspects of chloride induced reinforcement corrosion in concrete, that are currently not well understood and where researchers working in this eld need to focus on. Only carbon steel is considered, although a limited number of publications on corrosion resistant reinforcement in connection with critical chloride content can be found in the literature, e.g. [19,6163]. Minor parts of this review were presented in [64]. 2. Concept of critical chloride content 2.1. Denitions Reinforcement corrosion in non-carbonated, alkaline concrete can only start once the chloride content at the steel surface has reached a

certain threshold value [65]. In the literature, this value is often referred to as critical chloride content or chloride threshold value. In the present work, both terms or simply the abbreviation Ccrit are used. Two different ways of dening Ccrit are common [22,66]: From a scientic point of view, the critical chloride content can be dened as the chloride content required for depassivation of the steel (Denition 1), whereas from a practical engineering point of view Ccrit is usually the chloride content associated with visible or acceptable deterioration of the reinforced concrete structure (Denition 2). It has to be emphasized that the two denitions are related to different phenomena: the depassivation-criterion in Denition 1 only considers the initiation stage, whereas in the case of Denition 2 with visible or acceptable deterioration as a criterion, also the propagation stage is included. As a result, the two denitions lead to different Ccrit values. Fig. 1 illustrates this by combining Tuutti's corrosion model [67] with an assumed curve representing the chloride concentration at the steel reinforcement vs. time. The gure clearly shows that using the practical denition leads to higher Ccrit values. It is important to understand that this is only the result of a longer time passing until the chloride content is determined. The rate at which corrosion proceeds has a large inuence on when this is done and thus greatly affects the chloride threshold value when applying this denition. Denition 1 is more precise, since it expresses the chloride content that is directly related to depassivation. In Denition 2, the chloride

Fig. 1. Denitions for chloride thresholds (based on Tuutti's model [67]).

1124 Table 1 Possibilities to express the critical chloride content. Aggressive species Total chloride Inhibitive property Weight of cement/binder Weight of concrete Acid neutralisation capacity Weight of cement/binder Weight of concrete OH concentration

U. Angst et al. / Cement and Concrete Research 39 (2009) 11221138

2.2. Expression of Ccrit

Free chloride Free chloride ion concentration

content associated with an acceptable degree of corrosion has no theoretical background: the amount of chloride that is measured at that time has nothing to do with the degree of corrosion or the corrosion rate. Also the term acceptable degree is imprecise and thus Denition 2 results in a larger scatter of Ccrit values. In the literature, these two denitions are often mixed up. Care has thus to be taken when comparing and evaluating results reported by different researchers.

The critical chloride content is most commonly expressed as total chloride content relative to the weight of the cement. The main reason for this is the fact that the measurement of total chloride content is relatively simple and well documented in standards [68,69]. Since the quantication of the binder content in hardened concrete can be difcult, it is sometimes preferred to express Ccrit as total chloride content relative to the weight of the concrete. By assuming that the bound chlorides are completely removed from the pore solution and present no risk for pitting initiation, it would make sense to only consider the free chlorides. Thus, chloride threshold values are also expressed by the use of free chloride contents, either related to the weight of cement or concrete or as concentration in the pore solution. Another commonly used form to express critical chloride thresholds relates the chloride ion activity to the pH of the pore solution, i.e. Ccrit is presented in terms of (Cl)/(OH) or just Cl/OH. Publications often cited in this regard are those by Hausmann [6] and Gouda [8], but chloride threshold values in terms of Cl/OH have

Table 2 Published Ccrit values under outdoor exposure conditions or from real structures. Ccrit Total Cl (%bw) 0.2 1.4a 0.25 1.5 0.1 0.19 0.96 1.96 0.7 0.2 0.65 0.4 1.5 0.2 0.4 0.72 0.4 1.3 0.1 1.96 Experimental details Chloride cation (Na) (Na) Ca, (Na) Ca Seawater " Seawater Na Seawater Na Chloride introduction CAP + DIF CAP + DIF MIX MIX CAP + DIF " DIF DIF CAP + DIF MIX, CAP, DIF Specimen (w/b) C (NR) C (NR) C (0.45) C (0.6, 0.75) C (0.32 0.68) " C (0.3 0.75) C (NR) C (NR) C (0.4, 0.6) Cement type NR NR OPC OPC OPC FA SRCP, FA, SF, GGBS NR NR OPC Steel condition NR (RIB; AR) NR (RIB; AR) P SB, CL NR " RIB; AR NR NR (RIB; AR) NR Corrosion detection E E LPR, EIS, VI, WL WL WL " E, (LPR, GP) MC VI LPR, E Ccrit def. 12 12 1 2 12 " 12 1 12 1 1975 1984 1987 1989 1996 " 1998 2000 2001 2004 Stratfull et al. [10] Vassie [12] Hope and Ip [16] Treadaway et al. [19] Thomas et al. [23,31] " Sandberg [36] Zimmermann [39] Fluge [41] Morris et al. [45,48] min max Year Reference

Table 3 Published Ccrit values obtained from experiments with the steel directly immersed in solution (laboratory conditions). Ccrit Free Cl (mol/l) 0.170.34 0.020.13 1.0 0.6 0.57 0.48 0.29 0.27 0.30 0.6%b 0.3%b 4.9 0.250.8 0.14 0.28 0.42 0.056 Cl/OH Experimental details Chloride cation Na Na Na Na Na " " " " Na " Na Na, Ca Na " " Na Na Na " " Na " Steel condition NR ABR, CL ABR, CL SM, P, CL SM, P, CL " " " " SM, P, CL " P P, CL ABR, CL RIB, CL P CL SB; CL AR (MIL); SB; PR " " CL " Corrosion detection E GP GP E, (VI) GP, E, (VI) " " " " GP " LPR, E LPR, E PDP " " PC E, MC E, EIS " " PDP " Remarks In cement extract Solution aerated, pH 12.6 Solution stirred with nitrogen, pH 13 Solution aerated pH 11.613.2 pH 11.8 pH 12.1 pH 12.6 pH 13.0 pH 13.3 OPS mortar suspension GGBS mortar suspension Solution aerated, pH 13.8 pH 11.6413.22 Sat. Ca(OH)2 " " pH 13.5 Solution aerated, pH 13.213.5 pH 12.6 pH 13.3 pH 13.6 pH 12.5 pH 13.9 1955 1962 1965 1967 1970 " " " " 1970 " 1988 1990 1996 " " 1998 2000 2001 " " 2004 " Bird, presented in [3] Rajagopalan et al. [4] Venu et al. [5] Hausmann [6] Gouda [8], Diamond [14]a " " " " Gouda and Halaka [9] " Yonezawa et al. [18] Goi and Andrade [20] Mammoliti et al. [29] " " Breit [34] Zimmermann et al. [37,39] Li and Sags [40] " " Moreno et al. [49] " min max Year Reference

0.0056 0.28 0.00560.42

0.26 0.71.7 0.010.04 0.20.8 1.02.5 0.178 0.313 0.014.9

Diamond evaluated the results by Gouda, that did not report specic Ccrit values, but a linear relationship between pH and the logarithm of the chloride concentration. It is not clearly stated in the article whether this percentage is related to the sample weight, cement weight or water in the mix.

U. Angst et al. / Cement and Concrete Research 39 (2009) 11221138 Table 4 Published Ccrit values obtained from experiments with the steel embedded in cement based material (laboratory conditions). Ccrit Total Cl Free Cl (%bw) 0.32 0.571.09 > 0.4 0.45%a 0.15%a 0.40.8 0.250.5 >0.69 >0.31 0.10.19 745 0.20.68 0.482.02 1.52.5 0.42.0 0.51.8 320 Cl/OH Experimental details Chloride cation Ca Ca, Al Ca Na " Na Ca Na, Ca " Ca, (Na) Na Na, (Ca) NR Na " Na 2.56 Na Na " Na Na Chloride introduction MIX MIX MIX MIX " MIX MIX MIX " MIX MIX, DIF DIF Specimen (w/b) C (NR) C (0.7) M (0.45) C (0.6) " C (0.4) M (0.5) CEP " C (0.45) M (0.5) M (0.4 0.6) Cement type NR NR OPC OPC GGBS OPC OPC OPC GGBS OPC OPC OPC, FA, SF, SRPC, RHPC OPC, FA, SF, GGBS, SRPC OPC, SRPC " OPC, SF, FA Steel condition ABR, CL NR SM SM; P, CL " RIB; CL SB ABR, CL " P P Corrosion detection GDP, VI PDP, VI VI GP " LPR, (GDP) EIS, VI LPR " LPR, EIS, VI, WL LPR, E Remarks Submerged Submerged No corrosion observed 1959 Kaesche [1] 1959 Bumel [2] 1969 Richartz [7] Year Reference

1125

1970 Gouda and Halaka [9] " " Exposure to air 1980 Locke and Siman [11] Submerged/ 60% RH 1986 Elsener and Bhni [13] 1986 Andrade and Page [15] " " 1987 Hope and Ip [16] 1988 Yonezawa et al. [18]

Wet/dry cycles Submerged (but aerated) PC at 0 V SCE

MIX, CAP + DIF C (0.4 0.6) CAP + DIF MIX CAP + DIF, MIX CAP + DIF, MIX MIX, DIF " CAP + DIF DIF MIX CAP + DIF CAP + DIF " CAP + DIF MIG CAP + DIF MIG MIX " MIX, CAP + DIF CAP + DIF MIG CEP, C (0.5) " M (0.4 0.6)

SM, (RIB); PC, E, LPR CL NR MC

0.363.22 mol/l 0.141.83 mol/l

1990 Hansson and Srensen [21] Macro-cell 1990 Schiessl and Raupach [22] SM; SB, CL LPR, E, WL 1991 Lambert, Page et al. [24,25] " " " " " CL LPR Exposure to air 1992 Pettersson [26] RIB SM " NR LPR MC " E PC, VI LPR, E E, MC PC " LPR, E " LPR, E LPR E, VI " LPR, E PC LPR LPR, E, PDP, VI LPR, E " Exposure to air 1995 Pettersson [27] Exposure to air 1996 Schiessl and Breit [30] " " " 1997 Elsener et al. [32] Submerged 100% RH 1998 Breit [35] 2000 Alonso et al. [38] 2000 Zimmermann et al. [37,39] 2002 Alonso et al. [43] " 2002 " 2002 2003 2003 " Castellote et al. [42] " de Rincn et al. [44] Trejo and Pillai [46] Oh et al. [47]

M, C (0.3 0.75) OPC, SF C (0.5 0.7) " M (0.75) M (0.5 0.6) M (0.5) M (0.6) M (0.5) " M (0.37) " M (0.58) M (0.5) C (0.35 0.55) " C (0.4, 0.6) C (0.45) M (0.5) C (0.5) C (0.6) " OPC GGBS, FA OPC

0.51.0 1.01.5 0.440.65 mol/l 0.250.75 0.1 mol/l

1.243.08 0.391.16%bw 1.173.98 Na, (Ca) 0.251.25 0.0450.55 mol/l Na 0.735 1.08.34 0.62 0.42 0.51%bw 1.04.0%bw 0.36 mol/l 0.33 mol/l 0.4%bw 1.67 0.3 Na 1.720 1.5 2.0 "

OPC, SF, FA, SM SRPC, GGBS OPC RIB, SM OPC NR OPC, SRPC, FA " SRPC " OPC, FA, SF OPC OPC, FA, GGBS SRPC OPC OPC OPC OPC OPC SF RIB; MIL " RIB " NR CL (SM) " NR SM MIL P, PR, MIL, PP SM, RIB; AR, SB "

0.040.24 0.680.97 0.070.13%bw 0.45 0.41.3 0.520.75 0.050.15 0.40.8%bw 1.12.0 0.61.2 0.10%bw

Na " Na 0.090.62 NR 0.160.26 Na 0.27 " Na, seawater Na NR

E > 0.2 V SCE E < 0.2 V SCE 95% RH " Exposure to air Submerged 95% RH

Seawater CAP + DIF Na " MIX "

" " " Submerged; 2004 Morris et al. [45,48] exposure to air PC at 0 V SCE 2005 Nygaard and Geiker [50] Submerged 2005 Trejo and Monteiro [51] Exposure to air 2006 Mohammed and Hamada [52] Exposure to air 2008 Manera et al. [53] " " " min max

0.048.34 0.0453.22 mol/l 0.0945 0.071.16%bw

It is not clearly stated in the article whether this percentage is related to the sample weight, cement weight or water in the mix.

already been mentioned earlier, e.g. by Venu et al. [5]. Whereas Hausmann [6] suggested a constant value for the Cl/OH ratio, Gouda [8] found a linear relation between pH and the logarithm of the chloride concentration with a slope of 0.83, which yields that the ratio (Cl)0.83/(OH) is a constant. This relation implies that the inhibiting effect of the hydroxide ions becomes stronger with increasing pH. Li

and Sags [40] also reported that the Cl/OH threshold ratio increases with higher pH. Although Cl/OH threshold ratios are apparently not constant values, this form is often considered as the most accurate way to express critical chloride contents. Glass and Buenfeld [70], however, argue that this is not supported by analysis of available data in

1126

U. Angst et al. / Cement and Concrete Research 39 (2009) 11221138

literature. From their analysis of reported Ccrit values it was concluded that presenting critical chloride thresholds is best done in the form of total chloride by weight of cement. They mainly argued that the inhibitive properties of the concrete cannot be expressed only by the OH concentration in the pore solution, since also a lot of other factors such as the alkaline reserves of the concrete (buffer capacity) and the condition of the steelconcrete interface affect Ccrit. Moreover, they point out that if only the free chlorides are taken into account, the corrosion risk presented by bound chlorides which might be released as a consequence of several factors is ignored. Also Page and Havdahl [71] wrote that the Cl/OH ratio is not a reliable index. In connection with experiments involving silica fume they found that addition of silica fume increases the Cl/OH ratio in the pore solution (as a result of less chloride binding and a lower pH). On the other hand, it also leads to a denser microstructure, which not only slows down chloride ingress, but also reduces the oxygen content and thus depresses the steel potential. These effects might compensate for the negative effects on the pore solution chemistry, and thus a higher Cl/OH ratio in the pore solution must not necessarily be accompanied by a higher risk of corrosion initiation. Glass et al. proposed to express Ccrit as the total chloride content relative to the resistance presented by the concrete to a fall in pH (acid neutralisation capacity) [70,72,73]. The amount of acid (moles of H+) required to reduce the pH of suspensions of ground concrete samples to a certain value is thereby used to quantify the buffer capacity of the concrete and represents its inhibitive properties. Published critical chloride contents for various binder types have been evaluated with regard to this, and rather consistent values for Ccrit have been found: in terms of the ratio of total chlorides (moles of acid soluble chloride) to the acid neutralisation capacity (moles of H+), simply written as Cl/H+, the same value of ca. 0.01 has been obtained for all binder types considered [72]. Whereas this method of expressing and determining Ccrit values still needs further research, it might be promising in the sense that it includes both free and bound chlorides as the aggressive species and relates them to a decisive property of the hydration products. From a practical point of view, it is important to bear in mind that it is difcult or even impossible to measure the free chloride content or the pH of the concrete pore solution. Accurate values are thus often unknown. It is comparatively simple to determine the total chloride and to relate it to the weight of concrete or cement. Table 1 sums up the main forms of expression of chloride contents in concrete and shows how they reect the aggressive species and inhibitive properties of the concrete. 3. Critical chloride contents in the literature Numerous publications in connection with Ccrit can be found in the literature. Occasionally, Ccrit values have also been reported based on calculations and results from others: For example in Refs. [17,28], the authors determined the pH after pore liquid expression, and used Cl/ OH ratios according to Hausmann [6] and Gouda [8] to estimate the tolerable chloride concentration in the pore solution; a method to detect depassivation, however, was not included in the experiments and thus the results provide no new ndings with regard to the chloride threshold. In Ref. [74], on the other hand, depassivation was experimentally detected, but the corresponding chloride content was estimated with a theoretical diffusion model and not determined experimentally. Results of such studies are not considered in the present literature evaluation, since the experimental setups are not complete. With reference to critical chloride content, also standards and regulations are occasionally cited, as they often present limits on the tolerable chloride content in concrete, for instance the European standard EN 206-1 that restricts the chloride content to 0.2 0.4% chloride by mass of binder for reinforced concrete and 0.1 0.2% for prestressed concrete [75]. These limits, however, are not proper

chloride threshold values, but rather practical guidelines for the production of fresh concrete. Tables 24 summarise all the publications considered in the present evaluation. Since these Ccrit values were obtained from a variety of experimental setups, the corresponding details have to be considered when comparing them. The regarding information is also given in Tables 24, with abbreviations according to Table 5. For results from eld studies (Table 2) it is not always clear whether the corrosion state is close to depassivation or corresponds to a later stage in the propagation period; the numbers given in column 8 indicate which denition (according to Section 2.1) is more likely to be true. The reported Ccrit values in Table 2 are derived from investigations of bridge decks and coastal structures [10,12,39,41] or samples under outdoor exposure conditions [16,19,31,36,48]. In the case of laboratory studies, both with the steel directly immersed in alkaline solutions (Table 3) and when embedded in a cement based material (Table 4), corrosion was mostly detected close to depassivation due to the application of electrochemical techniques, and thus the values correspond more to Denition 1 (an exception from this is the work by Richartz [7]). In some publications [7,15,76] only lower boundaries for Ccrit were reported, since no corrosion activity was detected at the time of measurement of chloride content. Figs. 2 and 3 show reported Ccrit values expressed in the form of total chloride by binder weight and in terms of Cl/OH ratios, respectively. Also these diagrams summarise relevant experimental data. Together with Tables 24 they will be the basis for the detailed literature evaluation in the following sections. From a global perspective, a large overall scatter is apparent: The reported values range from 0.04 to 8.34% total chloride by weight of cement and by this over two orders of magnitude; in terms of Cl/ OH ratios even from 0.01 to 45 and by this over three orders of

Table 5 Abbreviations used in Tables 24 and Figs. 2, 3, and 6. Abbreviation CAP DIF MIG MIX C M CEP OPC FA SF GGBS SRPC RHPC RIB SM SB ABR P AR MIL PR PP CL LPR E EIS MC WL PC GP PDP GDP VI NR %bw Meaning Chloride introduced by capillary suction diffusion migration Added to the mix Concrete Mortar Cement paste Ordinary Portland cement Fly ash Silica fume Ground granulated blast furnace slag Sulphate resistant Portland cement Rapid hardening Portland cement Ribbed steel bars Smooth steel bars Sandblasted Abraded Polished As-received Mill-scaled Pre-rusted Pre-passivated Cleaned/degreased Linear polarisation resistance Potential Electrochemical impedance spectroscopy Macro-cell current Weight loss Potentiostatic control/polarisation Galvanostatic polarisation Potentiodynamic polarisation Galvanodynamic polarisation Visual inspection Not reported Total chloride by weight of binder

U. Angst et al. / Cement and Concrete Research 39 (2009) 11221138

1127

Fig. 2. Reported Ccrit values presented in terms of total chloride per weight of binder (abbreviations according to Table 5).

1128

U. Angst et al. / Cement and Concrete Research 39 (2009) 11221138

Fig. 3. Reported Ccrit values presented in terms of Cl/OH ratio; two separate plots with different ordinates due to large range (abbreviations according to Table 5).

magnitude. In another publication [76], though not directly comparable to the studies reviewed here since it dealt with the corrosion behaviour of steel bres in concrete, no corrosion was found at Cl/ OH ratios as high as 320. The large scatter in the literature is related to different experimental procedures and the numerous parameters that affect chloride induced corrosion in concrete; this will be discussed in the following sections. 4. Inuencing parameters From an electrochemical point of view, it is the potential of the steel, Ecorr, relative to the pitting potential, Epit, that determines whether corrosion will start or not. The pitting potential depends on both environmental inuences (chloride content) and on properties of the metal such as the degree of alloying (e.g. stainless steel). The open circuit potential of the passive steel, on the other hand, only depends on the environment (pH and oxygen content). Whereas parts of the steel electrode are in contact with the concrete pore liquid, others might be covered with hydration products and thus to a certain extent be shielded from aggressive species in solution. The critical chloride content in concrete is thus not only a matter of pure electrochemistry, but also of physical protection of the steel electrode. Numerous parameters affect the value of Ccrit and many of them are interrelated [77,78]: Steelconcrete interface Concentration of hydroxide ions in the pore solution (pH) Electrochemical potential of the steel Binder type

Surface condition of the steel Moisture content of the concrete Oxygen availability at the steel surface w/b ratio Electrical resistivity of the concrete Degree of hydration Chemical composition of the steel Temperature Chloride source (mixed-in initially or penetrated into hardened concrete) Type of cation accompanying the chloride ion Presence of other species, e.g. inhibiting substances. It has been suggested to consider the condition of the steel concrete interface as the most dominating inuencing factor [57], together with the pH of the concrete pore solution and the steel potential [78]. The variety of factors involved indicates that the concept of critical chloride content faces some difculties regarding a unique chloride threshold value applicable to a wide range of structures. 4.1. pH of the pore solution The presence of Portlandite is the reason for high pH values of ca. 12.5 in the pore solution, whereas NaOH and KOH can increase it to values above 13.5. The formation of a cement rich layer at the steel concrete interface stabilises the high pH and contributes to the passivity of steel in concrete [79]. The pH of the pore liquid initially mainly depends on the type of binder, but can be affected at later stages as the

U. Angst et al. / Cement and Concrete Research 39 (2009) 11221138

1129

result of carbonation, leaching, proceeding hydration, etc. The inhibiting effect of hydroxide ions against chloride induced corrosion as a major factor inuencing chloride threshold values was early recognised [5,6,8]. The suggestion to present Ccrit in terms of Cl/OH ratios reects this, since in this form the only inhibiting effect is ascribed to the presence of hydroxide ions. It is in this regard often referred to the work by Gouda [8] who experimentally found increasing Cl/OH ratios with increasing pH, or Hausmann [6] who suggested the famous Cl/OH ratio of 0.6 based on probability considerations. Fig. 4 illustrates that this value is more a lower boundary than a clear threshold. 4.2. Steel potential The presence of chloride ions at the steel surface modies the anodic polarisation curve, primarily by shifting the pitting potential, E pit , to more negative values (from A to B in Fig. 5). The electrochemical potential of steel at a certain location in a structure is thus of great importance: If the corrosion potential, Ecorr, is more positive than Epit, pitting corrosion takes place (B), otherwise the inuence of chloride is negligible (A). The chloride threshold is thus higher for steel with a more negative potential. This was experimentally conrmed in [43], where it was found that the chloride threshold is independent of the potential for values higher than 200 mV SCE, whereas for more cathodic potentials the chloride threshold increases with decreasing potential. For structures exposed to the atmosphere, the potential of the reinforcement is usually between +100 and 200 mV SCE, whereas for submerged structures it is in the range 400 to 500 mV SCE [78] and consequently higher chloride concentrations can be tolerated. Apart from environmental factors, the steel might be cathodically polarised by an external current, which also alters the critical chloride content (e.g. cathodic protection). 4.3. Steelconcrete interface The importance of the steelconcrete interface for Ccrit was already mentioned in a publication by Bumel as early as in 1959 [2]. The author presented a schematic sketch of the interfacial zone and pointed out the protecting character of calcium hydroxide directly surrounding the steel. It was also mentioned that treating the reinforcement with cement slurry prior to embedding can improve the corrosion resistance. Investigations of the interface between steel and concrete by scanning electron microscopy (SEM) reported by Page have revealed the presence of a dense, lime-rich layer of hydration products on the

Fig. 5. Inuence of chlorides on the anodic polarisation curve, leading to pitting corrosion in situation (B).

Fig. 4. Results from Hausmann (Table 1 in Ref. [6]) showing the probability of corrosion (number of corroding rods in a set of 12) vs. the Cl/OH ratio.

steel surface [79]. The author postulated that these solid hydration products might buffer the pH in the pore solution, for instance in the case of a local fall in pH in the vicinity of a pit. Furthermore, it was suggested that the presence of solid material on the steel surface may also act as a physical barrier and restrict the charge transfer reactions (both cathodic and anodic) in certain areas. Experimental work has also shown that the interfacial zone rich in Portlandite limits the diffusion of chloride more effectively than the concrete away from the interface [80]. Several studies conrmed the presence of higher amounts of calcium hydroxide in interfacial zones compared to bulk paste, such as Refs. [8183]. Contradicting this, no differences between the hydration products in the interfacial zone and those in the bulk cement paste were observed in Ref. [84]. It has been reported by Yonezawa et al. [18] that the chloride threshold level was signicantly lower when the formation of this lime-rich layer on the steel surface was restricted. Page et al. [24,25] reported Cl/OH threshold ratios of at least 3 for steel bars embedded in OPC concrete that had been exposed to external chlorides. These Ccrit values are much higher than those obtained by Hausmann [6] and Gouda [8], a fact that was attributed to the presence of a dense layer of hydration products at the steelconcrete interface in comparison with a steelsolution interface. In addition to microscopic characteristics, also macroscopic air voids and cracks at the steelconcrete interface affect the onset of corrosion. Macroscopic voids can be the result of incomplete compaction or low workability. Also the orientation of the rebars with respect to the casting direction may lead to gaps between the steel surface and the concrete [83,85] and even the ribs of reinforcement steel may favour the presence of voids. In this context, it was reported that corrosion preferably occurred at the corners or indents of the proling [21]; Alonso et al. [38] found a higher susceptibility to corrosion for ribbed steel bars in comparison to smooth bars. In a recent study [86] reinforced concrete beams that had been exposed to salty spray for 14 and 17 years under static load (three-point exion) were evaluated. It was found that despite rather high total chloride contents (ca. 1.5 2.2% by weight of cement) only little corrosion occurred in parts that were subjected to compressive stress, whereas for tensile reinforcement corrosion occurred preferably in the areas most loaded with tension. This observation was explained by mechanical degradation of the steelconcrete interface.

1130

U. Angst et al. / Cement and Concrete Research 39 (2009) 11221138

The condition of the interfacial zone appears to have a major inuence on the critical chloride content, both on a microscopic and a macroscopic level. However, since it is difcult to measure the amount of entrapped air voids and characterise microscopic defects, the condition of the interface is a property that cannot be quantied. 4.4. Binder type The type of binder inuences corrosion initiation by determining the amount of chloride that is available in the pore solution as a result of chloride binding and by affecting the pH of the pore solution. Moreover, certain binders might increase the electrical resistivity of the concrete or improve the characteristics of the steelconcrete interface by forming a denser microstructure. 4.4.1. Chloride binding capacity The capacity of the hydration products to bind chlorides affects the critical chloride content when expressed in terms of total chlorides. Since bound and free chlorides are suggested to be connected by a chemical equilibrium, also bound chloride presents a corrosion risk by acting as a reservoir of chloride that might dissolve at altered conditions [8789]. The degree of chloride binding depends on many factors [90] among which some are related to the binder type: The content of C3A and C4AF in the cement is a main parameter since the formation of Friedel's salt between chloride and the AFm phases removes considerable amounts of chloride from the pore solution (chemical binding); sulphate resistant Portland cements (SRPC) have thus lower chloride binding capacities [25,47]. Chloride can also be removed from the pore solution due to adsorption to the hydration products (physical binding). Tang and Nilsson [91] reported that the capacity of chloride binding strongly depends on the amount of CSH gel in the concrete, regardless of w/c ratio and the amount of aggregates. Binders containing mineral admixtures such as silica fume (SF), pulverized y ash (FA) or ground granulated blast furnace slag (GGBS) enhance the formation of more gel thereby offering larger surface areas available for adsorption. In addition, y ash and slag react to calcium aluminate hydrates which also might form Friedel's salt [90]. The mechanism of binding by sorption signicantly contributes to chloride binding and might be more important than traditionally assumed [89]. 4.4.2. Mineral admixtures Silica fume. In comparison with ordinary Portland cement (OPC), lower critical chloride contents have been reported for SF containing cement [27,53]; in Ref. [71] slightly higher corrosion rates in SF containing cement were reported. The chloride binding capacity of SF containing cements was reported to be lower compared to OPC [89,92,93]. The partial replacement of OPC with silica fume reduces the amount of aluminate phases and thereby the ability of the cement to bind chloride. But since the addition of SF also leads to a renement of the pores, the effect of physical adsorption is more pronounced in SF containing binders. However, it was reported that CSH produced by the pozzolanic reaction may have lower chloride sorption than C SH obtained by hydration of OPC [89]. Owing to the pozzolanic reaction, the alkalinity of the pore solution signicantly decreases with increasing addition of silica fume [17]. On the one hand, this affects the chloride binding capacity, since solubility of Friedel's salt increases as the pH of the pore solution decreases [92] (it was observed that the Cl/OH ratio in the pore solution increased considerably with increasing addition of SF [71,92]), on the other hand, the passive state of the steel will be less stable at a lower pH. Fly ash. Thomas [31] found lower Ccrit values for reinforced concrete specimens after exposure to a marine environment for up to four years when they contained FA; the tolerable chloride content

decreased with increasing substitution of OPC with FA. Also Oh et al. [47] measured lower chloride thresholds with increasing addition of FA. In contrast, Schiessl and Breit [30] reported higher chloride threshold values for concrete containing FA, and Alonso et al. [43] did not nd signicant differences in Ccrit when replacing cement with FA. Fly ash improves the chloride binding capacity of the binder. This can be attributed to both more efcient chemical binding due to higher proportions of active alumina often present in FA [93,94] and better physical adsorption of chloride as the result of more gel produced in the course of hydration [95]. On the other hand, use of FA lowers the pH of the pore liquid [17,96]. In one study, the reduction of pH appeared to be more pronounced than the improved chloride binding capacity and an increased Cl/OH ratio in the pore solution was found when FA was added to the binder [97]. Blast furnace slag. Gouda and Halaka [9] reported lower chloride threshold values for slag containing concrete specimens in comparison with OPC specimens, whereas Schiessl and Breit [30] found the opposite and Oh et al. [47] did not nd a signicant effect. The use of GGBS increases the chloride binding capacity due to improved chemical and physical binding [93,98,99]. However, it also decreases the pH of the pore solution: a pH of 12.8 in case of 40% GGBS and 12.4 in case of 60% GGBS replacement in OPC was reported (the age of the samples is unfortunately not given in the paper) [100]. 4.5. Surface condition of the steel In many investigations in the laboratory, the reinforcing steel is prepared prior to testing, e.g. by sandblasting or polishing, whereas in practice the reinforcement is used as-received and might be prerusted or covered with mill scale. It was shown that the condition of the steel surface has a signicant effect on the critical chloride threshold: Mohammed and Hamada [52] investigated steel bars with various surface conditions such as mill-scaled, polished, brown- and black-rusted and steel bars that were covered with cement paste (prepassivated) before casting. The chloride threshold values were sequenced as pre-passivated > black-rusted > polished > brown-rusted > mill-scaled. Also Mammoliti et al. [29] reported higher Ccrit for polished steel surfaces compared to ground or as-received samples. Li and Sags [40] immersed steel bars with different surface conditions in alkaline solutions containing chloride. They investigated steel bars in as-received conditions (mill-scaled), sandblasted and pre-rusted and found higher chloride threshold levels for sandblasted bars, although the corrosion rate of the sandblasted steel was higher once corrosion was initiated. Also Manera et al. [53] reported a higher critical chloride content for sandblasted steel bars in comparison with steel bars in as-received condition. Mahallati and Saremi [101] found that the presence of mill scale retards the formation and protective characteristics of the passive layer. Gonzales et al. [102] reported that passivation is delayed or even inhibited if reinforcement steel is considerably pre-rusted. It has to be noticed however, that the investigated steel had been pre-rusted by exposure to seawater and the rust layer thus contained chloride. Certainly, the presence of chloride affects the process of passivation. 4.6. Inuence of other factors 4.6.1. Moisture and oxygen content Both water and oxygen are required for the corrosion process, thus a lack of one of them is sufcient to inhibit corrosion. Since the moisture content regulates the availability of water and oxygen at the steel surface, it can be regarded as a global, environmental inuencing factor for Ccrit. In the case of water saturated concrete as well as in rather dry concrete, higher chloride concentrations are required to initiate corrosion; the situation most favourable for corrosion initiation is in the range 9095% RH [66] or concrete exposed to wetting/drying cycles

U. Angst et al. / Cement and Concrete Research 39 (2009) 11221138

1131

[103]. The amount of water in the concrete pores also affects the distribution between free and bound chlorides and thus determines the concentration of free chloride in the pore liquid. 4.6.2. Degree of hydration and w/b ratio The w/b ratio as well as the degree of hydration has an effect on the porosity of the paste and by this the availability of moisture and oxygen at the reinforcement. It was shown that the critical chloride content increases with decreasing w/b ratio [21,27,74] (compare Fig. 3). Both the porosity and the moisture content are reected by the electrical resistivity, a parameter that has been found to empirically correlate with chloride threshold values in a recent study [48]. 4.6.3. Type of chloride salt and chloride source Several studies have conrmed that calcium chloride leads to more chloride binding than sodium chloride, e.g. [15,93,104]. If Ccrit is expressed in terms of total chlorides, this theoretically results in higher threshold values. Nevertheless, it was reported that CaCl2 has a much more corrosive effect than NaCl or KCl [15,105]. Care has to be taken, however, when evaluating studies with regard to this: When chloride is added to the mix it acts as an accelerator resulting in a coarser pore structure, which, as was shown earlier, certainly will affect the corrosion behaviour. If the chloride source is seawater, less chloride might be bound due to the inherent sulphates that block some of the adsorption sites [90]. 4.7. Overall tendencies in the literature

ribbed and smooth bars, or for conditions such as sandblasted, asreceived, polished, etc (Fig. 2). None of these parameters are strong enough to dominate the others. However, when studying Fig. 3 it is apparent that Ccrit values obtained from experiments dealing with steel embedded in mortar or concrete exhibit a clear global tendency towards higher Ccrit values compared with solution experiments. This conrms the importance of the steelconcrete interface as a major inuencing parameter. The pH was also mentioned as one of the governing inuencing factors and experimental evidence for this is in Fig. 3 apparent from the results by e.g. Li and Sags [40] or Moreno et al. [49]. However, the effect is only visible when looking at results derived from a certain experimental setup, i.e. within a certain study, and not when evaluating the results from different authors as a whole. In this case, the effect of the pH is overshadowed by other parameters. As stated above, this is true for most of the inuencing factors. 4.8. Summary of inuencing parameters The above discussed interrelations are summarised in Table 6, which is based on a similar compilation in Ref. [57]; in the present article, it was extended to various forms to express Ccrit and the references were updated. Note that the effect of certain factors is not the same when different forms for expressing the threshold value are chosen. 5. Experimental setups 5.1. General considerations regarding the determination of Ccrit values

The values in Fig. 2 were grouped according to the initial free corrosion potential. In many studies, these potentials were higher than 200 mV SCE, where, according to [43], the potential is expected to have little inuence on Ccrit. Unfortunately, the potential was not always reported, but in most of the concerning publications, the exposure conditions indicate that the steel potential was presumably in the same range. In two publications, the steel potential was below 200 mV SCE and signicantly higher chloride threshold values were reported. In Fig. 6, these results were excluded and the scatter of reported Ccrit values was plotted for some selected parameters. It is apparent that there is no signicant difference between mortar and concrete, apart from a somewhat larger range in the case of concrete, but this might simply be owing to more results being available for concrete than for mortar. When analysing different binder types, much more publications reported values for OPC and thus the range is slightly larger compared with SF, FA and GGBS; apart from that, no clear tendencies can be found. In fact, the same is true when examining other characteristics: Ccrit values scatter over just as large ranges for

Apart from a suggestion for a method to determine Ccrit in the eld [107], an accepted testing procedure does at present not exist.

Table 6 Effect of inuencing factors on the critical chloride content. Factor Effect on critical chloride content Total Cl % cem wt Steel condition Defects at steelconcrete interface Polishing, sandblasting Steel potential (> 200 mV SCE) Steel potential (< 200 mV SCE) O Cl/OH ratio Free Cl References

[18,21,38,86] [29,40,52,53] [43] [43]

Concrete and binder properties w/b ratio Chloride binding pH Electrical resistivity SF FA Ob GGBS Ob SRPC (low C3A + C4AF content) External factors Moisture in rather dry concrete Moisture in nearly saturated concrete Moisture variations Oxygen availability Temperature Fig. 6. Scatter of chloride threshold values in the literature sorted by selected parameters and when excluding values obtained at potentials < 200 mV SCE. The numbers above the bars indicate the frequency of occurrence in the literature. Abbreviations according to Table 5.

O Ob O
c

O Ob O
c

[21,27,74]
a

[5,6,8,40] [48] [27,53] [30,31,43,47] [9,30,47] [106]

() ()

() ()

[66] [66,79] [103] [6] [28]

() indicates an increase (decrease) in threshold level with an increase of the concerning factor; O means no inuence on Ccrit. a According to theoretical considerations. b Contradictory results reported in the literature. c No results reported.

1132

U. Angst et al. / Cement and Concrete Research 39 (2009) 11221138

Experimental setups are thus developed individually. To measure Ccrit values in the laboratory or on a real structure, such a setup has to include the following [64]: 1. A steel electrode embedded in a cement based material (cement paste, mortar, concrete) or immersed in a solution that simulates the concrete. 2. Chloride ions at the steel surface. 3. A method to detect depassivation of the steel (Denition 1) or for determining if the degree of corrosion has reached the acceptable limit (Denition 2). 4. A method to quantify the chloride content. Many options exist in order to design an experiment that fulls these four requirements, as schematically depicted in Fig. 7 [64]. The most convenient setup is a steel electrode immersed in an alkaline solution, where both the chloride concentration and the pH can

easily and rapidly be changed and accurately quantied. It is, however, generally agreed that alkaline solutions are not suitable to model real concrete in the laboratory for many reasons. When using cement paste, mortar or concrete, the introduction of chloride into hardened samples becomes more time-consuming; in addition, measuring the concentration actually present at the steel surface is more difcult and laborious. One option is to add chlorides directly to the mix, a common method to simulate the use of chloride bearing aggregates or the addition of CaCl2 as an accelerator as it was done in the past. All these methods normally differ from reality and have thus advantages and disadvantages when modelling the practical situation in the laboratory as summarised in Table 7. For instance, adding chloride to the mix is fast and leads to a homogeneously distributed and thus well-dened total chloride content in the sample; on the other hand, the steel might not passivate if the chloride concentration is too high, more chloride might be bound than in the case of natural chloride penetration into hardened concrete

Fig. 7. Experimental possibilities to study critical chloride content (from [64], reprinted with permission).

U. Angst et al. / Cement and Concrete Research 39 (2009) 11221138 Table 7 Techniques to introduce chloride into mortar or concrete samples. Chloride introduction technique Mixed-in Advantages Disadvantages

1133

Fast Homogeneously distributed Total chloride content known

Initial formation of passive layer questionable Homogeneously distributed Different chloride binding Porosity affected Not practice-related (anymore) Very time-consuming Sample usually water saturated (limited oxygen availability) Time-consuming

Pure diffusion

Afnity to practice

Capillary suction Afnity to and diffusion practice Faster than pure Drying coarsens the pore structure diffusion Migration Fast Electrical eld Migration of OH ions changes the pH in the sample Sample water saturated (limited oxygen availability)

5.2.2. Linear polarisation resistance measurements (LPR) The most accurate technique to detect depassivation of the steel is the measurement of the linear polarisation resistance, which is inversely proportional with the corrosion current as described by the SternGeary-equation [108]. The technique is non-destructive and allows the determination of the instantaneous corrosion rate. As criterion for active corrosion, an averaged sustained corrosion rate higher than 0.1 0.2 A/cm2 has been suggested for concrete containing substantial moisture and oxygen [109]. One of the reasons for this value is the empirical observation that the measured corrosion current in solution and mortar studies was never below 0.1 A/cm2 when rust was observed [110]. One has to be aware, however, that the measured corrosion rate is an average value over the exposed steel area and that the local current density inside a pit is signicantly higher. Consequently, the measured value depends strongly on the number and the size of the pits. To avoid such interpretation difculties, it might be preferred to observe a change in polarisation resistance (or corrosion rate) over time rather than an absolute value. 5.2.3. Electrochemical impedance spectroscopy (EIS) AC impedance spectroscopy is performed by imposing an external sinusoidal voltage signal of small amplitude over a range of frequencies. The impedance spectra, as derived from the voltage signals and the current responses, are interpreted on the basis of equivalent electrical circuits, which allow determination of e.g. the charge transfer resistance of the process, a parameter that, in many cases, can be equated to the polarisation resistance [111]. By applying the SternGeary-equation and assuming Tafel slopes the corrosion rate can then be calculated. 5.2.4. Weight loss Gravimetric loss of reinforcement can be determined by weighing steel bars prior to embedding and after removing them from concrete. To detect the mass difference, it is required that a signicant amount of corrosion has already taken place. This method is thus not suitable to identify the time of depassivation and leads to Ccrit values according to Denition 2. Prior to weighing, the bars have to be cleaned to remove any concrete particles. This procedure might result in an additional weight loss. Thomas [31] suggested to apply the cleaning procedure also to control rebars to nd this processing weight loss. 5.2.5. Other methods The corrosion rate can be determined by measuring the macro-cell current between the electrode of interest and an auxiliary electrode acting as cathode as e.g. in Refs. [22,30,37,39]. In some cases the working electrode is polarised and held constant at a certain potential during the experiments (potentiostatic control) [21,34,43,50]; the current required for this polarisation is monitored until an increase indicates active corrosion. Depassivation can also be detected by means of galvanostatic polarisation: A xed anodic current is applied and the potential response is monitored vs. time. The measured curve gives information on the corrosive or inhibitive character of a certain environment. Though rarely, this method has been used to determine Ccrit values [4,5,8,9]. It has to be noted that controlling the steel potentiostatically (or galvanostatically) does not provide realistic conditions. Under the externally applied current, the entire steel surface exhibits the same potential, which is in the case of potentiostatic control determined by an innitely strong cathode. No differential aeration cells are formed, although in reality, these have to provide a current strong enough for pitting corrosion initiation [112]. In addition, a critical parameter with this method is the selection of the potential to which the working electrode is polarised (or in the case of galvanostatic polarisation the current density). It might be interesting to note in this context that Ccrit values obtained from test setups using

and also the porosity of the sample might be affected. While a homogeneous chloride distribution is benecial for the quantication of the concentration at the steel surface, it is not with regard to modelling a practice-related situation: In contrast to reality, no differential concentration cells are formed along the steel surface, a fact that is certainly important for the onset of pitting corrosion. Other techniques to introduce chloride are pure diffusion and acceleration by capillary suction (wetting/drying cycles) or an electrical eld (migration). Whereas capillary suction and diffusion have a high afnity to practice, migration techniques differ strongly from reality. The most adverse conguration would be to use the rebar of interest directly as anode in order to attract chloride ions, but even if another electrode is applied for this purpose, the presence of the electric eld introduces several uncertainties. It is not only chloride ions that migrate, but rather all ions present in the pore solution including those that are generated at anode and cathode. This changes the composition of the pore solution and thus affects critical parameters such as the pH. In fact, Fig. 6 shows that Ccrit values obtained by the use of migration are signicantly lower compared with results from other studies. It should be noted, however, that this observation is only based on two publications [46,51]. Not only the way chloride is introduced, but also the type of salt has an effect on the outcome of experimental work; the same is true for the selection of a certain type of rebar (ribbed or smooth) and its surface condition. These points must be taken into account carefully when designing experiments. 5.2. Methods for detecting active corrosion 5.2.1. Steel potential Since actively corroding steel has a signicantly more negative potential than passive steel in concrete, onset of corrosion can be detected by potential measurements. In this case, a certain shift in potential or an absolute value can be used as criterion for depassivation. Potential measurements are common in eld investigations [10,12,36], but are also often used in laboratory research aimed at nding Ccrit values, as e.g. in Refs. [6,8,9]. It has to be kept in mind that the measured potential is not only dependent on the corrosion state of the steel, but also of other factors such as oxygen availability or iR drops; a low potential value does not necessarily mean signicant corrosion. Thus, a clear potential drop (change from passive to active condition) rather than an absolute value should be considered.

1134

U. Angst et al. / Cement and Concrete Research 39 (2009) 11221138

potentiostatic control are generally lower compared with those obtained on rebars without external polarisation (Fig. 6). It is of course also possible to measure complete polarisation curves, or only the anodic part, in order to determine pitting potentials. Such E icurves can be measured by either stepwise impressing a xed current until a certain potential is reached (galvanodynamic polarisation), or by controlling the potential and measuring the corresponding current (potentiodynamic polarisation). Finally, if the concrete is split and the steel exposed, visual inspection of the steel surface can help to identify depassivation with the appearance of rust on the steel surface. The accuracy of the method is low, since it is not possible to know how much time has passed between depassivation and visual observation of rust. Moreover, the appearance of rust spots may take some time and, once present, does not necessarily mean signicant and sustained corrosion activity. Visual inspection to detect depassivation was mostly used in early work on corrosion initiation [7] and in eld studies [41]; sometimes it was applied in combination with other techniques, e.g. in [6,13,16]. All of the techniques mentioned have in common that a certain amount of corrosion is necessary to detect depassivation. Depassivation is not considered as an instant event, but rather as a period of time during which the depassivation process takes place from the rst defect until active corrosion is established [77]. It is impossible to identify the very start, e.g. the rst local defect in the passive lm. Electrochemical techniques come much closer to this moment than visual inspection and weight loss measurements. This should be kept in mind when considering the denitions of Ccrit as described in Section 2.1. 5.3. Methods for determining the total chloride content The analysis of total chloride is frequently applied in practice and well documented in standards, such as [68,69]. The total chloride content in concrete is usually determined by analysing cores drilled from hardened concrete, which are cut into slices of a certain thickness to obtain a chloride prole. The sample is then crushed, powdered and homogenised and subsequently dissolved in dilute nitric acid. The chloride concentration in the extraction solution can be determined by several techniques such as titration, use of ion selective electrodes or spectrophotometric methods. Dhir et al. [113] compared various methods to determine the total chloride content in concrete on OPC concrete specimens with mixed-in chloride. They found that the quantity of chloride extracted depends on the strength of the nitric acid and dissolution time. The results also suggested that the acid extraction technique cannot completely dissolve all the chlorides from the powder samples; values in the range 70 90% of the true content were reported. A more expensive but very accurate way is to determine the total chloride content in concrete powder samples by X-ray uorescence spectrometry (XRF) [113]. In a recent round robin test [114] involving 30 laboratories methods for the analysis of total chloride were compared and a good reproducibility was found for all of the methods. 5.4. Methods for determining the free chloride content

concentration in the exposure solution [89,116]. It has also to be kept in mind that the pore solution expression technique results in an average value of the concrete volume under investigation; in case of high concentration gradients in the pore solution this leads to inaccurate results. 5.4.2. Leaching techniques Leaching techniques are based on mixing crushed or ground samples with a solvent and measuring the amount of chloride passing into solution. Distilled water is usually used as solvent. Alternatives such as methanol or ethyl alcohol have been investigated but both proved to be extremely ineffective at leaching out the free chlorides; measured concentrations were in the range of 5 10% of the free chloride concentration obtained from pore solution expression [116,119]. The amount of leached chloride appears to depend on the time during which the sample is in contact with the leaching media and on temperature. Arya et al. [119] investigated different extraction procedures in this regard and compared the results to pore solution expression (considered as true value). At chloride concentrations up to 1% by weight of cement (mixed-in) much higher chloride concentrations were measured compared to pore water expression, whereas in the range from 1.5 to 2% chloride by weight of cement the leached chloride concentration was lower than the true value. It was concluded that none of the investigated methods was sufciently accurate over the range of chloride contents tested, but that the most accurate procedure can be selected if the total chloride content is known. In a later study [117] it was found that not only the total chloride content but also other parameters such as the cement type and source of chloride (mixed-in or external) have to be known to select an accurate leaching procedure and that thus the leaching technique is not practical for determining the free chloride content. Castellote et al. [120] presented a method based on an alkaline solvent to extract the free chloride. Good agreement with pore water expression was found. However, rather high chloride contents have been investigated (up to 6% by weight of sample) and it might be questionable if the technique is also accurate at lower and more practical concentrations. The chloride concentrations obtained from leaching methods are sometimes referred to as water soluble chloride (if water is the solvent) and often considered to be equal to free chloride. Contrary to Europe, water soluble chloride analysis is a standard method used in North America [121]: In the Soxhlet extraction technique boiling water is used to extract chloride from concrete chips. 5.4.3. Ion selective electrodes The free chloride content in the pore solution can also be measured by the use of ion selective electrodes embedded in the concrete [32,122,123]. A major advantage of this technique is that it measures non-destructively and thus allows continuous monitoring of the chloride concentration in the concrete pore solution. Moreover, results are obtained highly localised, whereas the above mentioned techniques require comparatively large samples volumes for the analysis and thus result in average values. 5.5. Experimental setups in the literature

5.4.1. Pore solution expression The technique of expressing pore water from cement paste or mortar is well established and has been used by many researchers, e.g. [115,116]. It is probably the most accepted method to determine the free chloride content in the pore solution. However, the application is limited when mortar with lower w/c ratios, coarse aggregate particles or rather dry specimens are investigated [117]. It has been noted that under the pressure loosely bound chloride might be released and thus result in an overestimation of the free chloride content [118]. Also other studies reported that the chloride concentration in the liquid expressed from samples that had been immersed in solutions was higher than the

The percentages given in Fig. 7 represent the proportion of experiments, in which the corresponding conguration was chosen. For example, in 25% of the experiments evaluated in the present literature review, smooth steel was used, while in 15% the bars were ribbed; in 60% no details with regard to this were given in the publication. Note that experiments rather than publications have been counted (meaning that one publication might contain several different experiments) and that the sum of the percentages can be higher than 100%, as e.g. in the case of corrosion detection techniques, since in some experiments more than only one technique was applied.

U. Angst et al. / Cement and Concrete Research 39 (2009) 11221138

1135

In addition, one has to be aware that an evaluation of this kind cannot be accurate as in some studies hundreds of samples have been tested, while other reported results from very limited amounts of samples. Nevertheless, it gives an impression of preferably used setups. Concerning the steel electrode, often smooth steel with a prepared surface was investigated: in nearly 40% of the experiments the steel was cleaned and degreased and in ca. 30% polishing or abrading was used to prepare the steel surface; in only 14% the steel was reported to have been investigated in as-received or mill-scaled condition. Without doubt this affects the characteristics of the steelconcrete interface, one of the major inuencing factors for Ccrit as was shown earlier. In ca. 20% the steel was investigated directly immersed in synthetic pore solution; also this does not give a realistic interface, but it allows comparatively simple measurements of hydroxide and free chloride ion concentrations and fast experimental programmes. As in the case of steel embedded in mortar or concrete, chloride introduction can be time-consuming, in nearly 50% of the evaluated experiments chloride has been added directly to the fresh mix; in the case of hardened samples, a combination of capillary suction and diffusion appears to be the most often used technique to introduce chloride. In more than 60% of the experiments Ccrit was reported in terms of total chloride relative to the binder weight, but when excluding the experiments with the steel electrode directly immersed in solution, this gure is as high as 84%. Free chloride concentrations or Cl/OH ratios have only been determined in a minority of experiments involving hardened samples, and, in addition, these were often porous mortar rather than concrete samples. Reliable Ccrit values in terms of both total and free chloride contents for dense concrete are lacking. A wide variety of testing setups is benecial when exploring the issue of Ccrit from different perspectives; in fact, varying parameters is even required when specically investigating certain inuences as e.g. the surface condition of the steel. On the other hand, different test setups make comparison of the results difcult. With regard to a practical and relatively fast test it is also favoured to simplify reality and e.g. use mixed-in chloride; in order to obtain reproducible results the surface might be polished or sandblasted. However, the results of such studies cannot be carried over into practice. 6. Evaluation with regard to chloride binding In the literature review in Ref. [70] Glass and Buenfeld evaluated data by Lambert et al. [25] and pointed out that Cl/OH threshold ratios span a larger range (from 3 to 20) in comparison with corresponding total chloride contents (from 1.5 to 2.5%). More publications are now available reporting Ccrit values in terms of total and free chloride or Cl/OH ratios and they conrm this nding without exception as illustrated in Fig. 8. For example, Pettersson [26] reported total chloride threshold values from 0.9 to 1.4% by weight of binder (for OPC) and corresponding free chloride threshold concentrations from 36 to 73.9 g/l. The lowest total chloride content (0.9%) is 64% of the highest value (1.4%), whereas the lowest free chloride content (36 g/l) is 49% of the highest value (73.9 g/l). In Fig. 8 these percentages are plotted on a reverse y-axis where the height of the bars indicates the range of reported values, i.e. the higher the bar the larger the span. Also when looking at Ccrit values reported in the literature as a whole (Tables 24), values in terms of total chloride per binder weight span a smaller range than free chloride concentrations or Cl/OH threshold ratios. It was suggested in Ref. [70] that this might be explained by the non-linear relationship between bound (total) and free chlorides: any scatter in bound chloride is amplied when expressed as a free chloride content. In addition, when assuming that only the free chloride is relevant for corrosion initiation, the following is expected: forms to express Ccrit that ignore bound chloride, such as Cl/OH ratios or free chloride by mass of cement, should theoretically result in a smaller range of values. In the

Fig. 8. Range of reported Ccrit values for both mortar and concrete expressed in different forms.

available data, however, the opposite was found. Based on this observation, Glass and Buenfeld [70] questioned the importance of chloride binding and postulated that bound chloride plays a more important role in corrosion initiation than generally assumed. The additional data evaluated in the present review appears to support this observation: the same was found even when different techniques to measure the free chloride content were applied (as indicated above the bars in Fig. 8). In the light of the uncertainties associated with measurement techniques for the free chloride concentration as discussed in Section 5.4, one should, however, also consider the reliability and accuracy of the reported free chloride concentrations. All of the procedures used to quantify free chloride contents are likely to lead to a scatter in the results, which could at least partly explain the larger range of reported free chloride concentrations compared with total chloride contents. 7. Conclusions From the present literature review, the following major conclusions are drawn: 1. The critical chloride content in reinforced concrete can and has been studied by many different experimental setups. When evaluating the entirety of reported results in the literature, it appears that certain parameters inherent to the test procedure (such as the application of an electric eld to accelerate chloride ingress or potentiostatic control of the rebar) can have a more dominant inuence on the result than the parameters under investigation (e.g. binder type). Many of the used experimental setups are not suited to give realistic results. In addition, comparison of the reported values is difcult owing to differences in procedures. On this basis, it is not possible to select a reliable range of chloride threshold values and thus the current practice of condition assessment as well as service life design cannot be improved. 2. Consultant engineers thus still base their decisions on long-term experience from existing structures, of which most were built with ordinary Portland cement. However, more and more non-traditional cements containing pozzolanic materials or other additions are used. The effect of some of these materials namely SF, FA, and GGBS has been studied several times, but the results were often

1136

U. Angst et al. / Cement and Concrete Research 39 (2009) 11221138 [17] K. Byfors, Inuence of silica fume and yash on chloride diffusion and pH values in cement paste, Cement and Concrete Research 17 (1987) 115130. [18] T. Yonezawa, V. Ashworth, R.P.M. Procter, Pore solution composition and chloride effects on the corrosion of steel in concrete, Corrosion 44 (1988) 489499. [19] K.W.J. Treadaway, R.N. Cox, B.L. Brown, Durability of corrosion resisting steels in concrete, Proceedings of the Institution of Civil Engineers Part 1 (86) (1989) 305331. [20] S. Goi, C. Andrade, Synthetic concrete pore solution chemistry and rebar corrosion rate in the presence of chlorides, Cement and Concrete Research 20 (1990) 525539. [21] C.M. Hansson, B. Srensen, The threshold concentration of chloride in concrete for the initiation of reinforcement corrosion, in: N.S. Berke, V. Chaker, D. Whiting (Eds.), Corrosion rates of steel in concrete, ASTM STP 1065, 1990, pp. 316. [22] P. Schiessl, M. Raupach, Inuence of concrete composition and microclimate on the critical chloride content in concrete, Proc. 3rd Int. Symp. Corrosion of Reinforcement in Concrete , Elsevier Applied Science, Wishaw, UK, 1990, pp. 4958. [23] M.D.A. Thomas, J.D. Matthews, C.A. Haynes, Chloride diffusion and reinforcement corrosion in marine exposed concrete containing pulverized-fuel ash, Proc. 3rd Int. Symp. Corrosion of Reinforcement in Concrete, Elsevier Applied Science, Wishaw, UK, 1990, pp. 198212. [24] C.L. Page, P. Lambert, P.R.W. Vassie, Investigations of reinforcement corrosion. 1. The pore electrolyte phase in chloride-contaminated concrete, Materials and Structures 24 (1991) 243252. [25] P. Lambert, C.L. Page, P.R.W. Vassie, Investigations of reinforcement corrosion. 2. Electrochemical monitoring of steel in chloride-contaminated concrete, Materials and Structures 24 (1991) 351358. [26] K. Pettersson, Corrosion threshold value and corrosion rate in reinforced concrete, CBI report 2:92, Swedish Cement and Concrete Research Institute, 1992. [27] K. Pettersson, Chloride threshold value and the corrosion rate in reinforced concrete, Proc. of the Nordic Seminar, Lund, 1995, pp. 257266. [28] S.E. Hussain, A. Rasheeduzzafar, A. Al-Musallam, A.S. Al-Gahtani, Factors affecting threshold chloride for reinforcement corrosion in concrete, Cement and Concrete Research 25 (1995) 15431555. [29] L.T. Mammoliti, L.C. Brown, C.M. Hansson, B.B. Hope, The inuence of surface nish of reinforcing steel and pH of the test solution on the chloride threshold concentration for corrosion initiation in synthetic pore solutions, Cement and Concrete Research 26 (1996) 545550. [30] P. Schiessl, W. Breit, Local repair measures at concrete structures damaged by reinforcement corrosion aspects of durability, Proc. 4th Int. Symp. Corrosion of Reinforcement in Concrete Construction, The Royal Society of Chemistry, Cambridge, 1996, pp. 525534. [31] M. Thomas, Chloride threshold in marine concrete, Cement and Concrete Research 26 (1996) 513519. [32] B. Elsener, L. Zimmermann, D. Flckiger, D. Brchler, H. Bhni, Chloride penetration non destructive determination of the free chloride content in mortar and concrete, Proc. RILEM Int. Workshop Chloride penetration into concrete, Paris, 1997. [33] P. Sandberg, K. Pettersson, H.E. Srensen, H. Arup, Critical chloride concentrations for the onset of active reinforcement corrosion, Proc. RILEM Int. Workshop Chloride penetration into concrete Paris, 1997. [34] W. Breit, Critical chloride content investigations of steel in alkaline chloride solutions, Materials and Corrosion 49 (1998) 539550. [35] W. Breit, Kritischer korrosionsauslsender Chloridgehalt Neuere Untersuchungsergebnisse (Teil 2), Beton 8 (1998) 511. [36] P. Sandberg, Chloride initiated reinforcement corrosion in marine concrete, Dissertation (Report TVBM-1015), Lund Institute of Technology, Division of Building Materials, Lund University, 1998. [37] L. Zimmermann, B. Elsener, H. Bhni, Critical factors for the initiation of rebar corrosion, Proc. EUROCORR '99, European Federation of Corrosion, Aachen, Germany, 1999. [38] C. Alonso, C. Andrade, M. Castellote, P. Castro, Chloride threshold values to depassivate reinforcing bars embedded in a standardized OPC mortar, Cement and Concrete Research 30 (2000) 10471055. [39] L. Zimmermann, Korrosionsinitiierender Chloridgehalt von Stahl in Beton, Dissertation, ETH Nr. 13870, ETH Zrich, 2000. [40] L. Li, A.A. Sags, Chloride corrosion threshold of reinforcing steel in alkaline solutions open-circuit immersion tests, Corrosion 57 (2001) 1928. [41] F. Fluge, Marine chlorides a probabilistic approach to derive provisions for EN 206-1, Third Workshop on Service Life Design of Concrete Structures From Theory to Standardisation, DuraNet, Troms, 2001, pp. 6383. [42] M. Castellote, C. Andrade, C. Alonso, Accelerated simultaneous determination of the chloride depassivation threshold and of the non-stationary diffusion coefcient values, Corrosion Science 44 (2002) 24092424. [43] C. Alonso, M. Castellote, C. Andrade, Chloride threshold dependence of pitting potential of reinforcements, Electrochimica Acta 47 (2002) 34693481. [44] O. de Rincn, Y. Hernndez, R. Fernndez, M. Morales, K. Inciarte, Comparison between chloride ion threshold and electrochemical measurements for reinforcement corrosion, Proc. 3rd RILEM Int. Workshop Testing and Modelling the Chloride Ingress into Concrete, Madrid, 2002. [45] W. Morris, A. Vico, M. Vazquez, S.R. de Sanchez, Corrosion of reinforcing steel evaluated by means of concrete resistivity measurements, Corrosion Science 44 (2002) 8199. [46] D. Trejo, R.G. Pillai, Accelerated chloride threshold testing: Part I ASTM A 615 and A 706 reinforcement, ACI Materials Journal 100 (2003) 519527.

contradictory and/or cannot be transferred to real structures due to unrealistic testing conditions. Regarding reinforcement corrosion, the behaviour of the mentioned pozzolanas, but also many other upcoming cementing materials, is completely unknown. 3. In the light of Conclusions 1 and 2, there is a strong need for a practice-related test method. For a procedure that realistically models corrosion of reinforced concrete structures exposed to seawater or deicing salt spray, the authors suggest the following experimental parameters: The rebar has to be ribbed and in asreceived condition, and must be embedded in concrete (or at least mortar). Chloride has to be introduced by a combination of capillary suction and diffusion and must not be added to the fresh mix. Depassivation is best detected by electrochemical measurements in order to stick to a precise denition of Ccrit (Denition 1). In this regard measurement of steel potential, linear polarisation resistance, or electrochemical impedance spectroscopy are appropriate methods; potentio- or galvanostatic (or -dynamic) polarisation is not suitable. 4. The most important inuencing factors on Ccrit are the steel concrete interface and the steel potential, since these are the only parameters standing out when comparing the reported results as a whole. All the other parameters are interrelated and overlap each other in such a way that no overall trends are visible. 5. More research is needed regarding the importance of chloride binding. For this purpose, however, reliable methods to measure the free chlorides are required. At present, these techniques are associated with uncertainties and might at least partly be responsible for the inconsistency between the general assumption that bound chlorides are rendered harmless and experimental results in the literature. Acknowledgements The authors acknowledge the support of COIN (www.sintef.no/ coin). References
[1] H. Kaesche, Die Prfung der Korrosionsgefhrdung von Stahlarmierungen durch Betonzusatzmittel, Zement-Kalk-Gips 7 (1959) 289294. [2] A. Bumel, Die Auswirkung von Betonzusatzmitteln auf das Korrosionsverhalten von Stahl in Beton, Zement-Kalk-Gips 7 (1959) 294305. [3] D.A. Lewis, W.J. Copenhagen, Corrosion of reinforcing steel in concrete in marine atmospheres, Corrosion 15 (1959) 382388. [4] K.S. Rajagopalan, K. Venu, K. Balakrishnan, Anodic polarization studies in neutral and alkaline solutions containing corrosion inhibitors. I. NaOHNaCl system, Journal of the Electrochemical Society 109 (1962) 8187. [5] K. Venu, K. Balakrishnan, K.S. Rajagopalan, A potentiokinetic polarization study of the behaviour of steel in NaOHNaCl system, Corrosion Science 5 (1965) 5969. [6] D.A. Hausmann, Steel corrosion in concrete. How does it occur? Materials Protection 6 (1967) 1923. [7] W. Richartz, Die Bindung von Chlorid bei der Zementerhrtung, Zement-KalkGips 10 (1969) 447456. [8] V.K. Gouda, Corrosion and corrosion inhibition of reinforcing steel. I. Immersed in alkaline solutions, British Corrosion Journal 5 (1970) 198203. [9] V.K. Gouda, W.Y. Halaka, Corrosion and corrosion inhibition of reinforcing steel, II. Embedded in concrete, British Corrosion Journal 5 (1970) 204208. [10] R.F. Stratfull, W.J. Jurkovich, D.L. Spellman, Corrosion testing of bridge decks, Transportation Research Record 539 (1975) 5059. [11] C.E. Locke, A. Siman, Electrochemistry of reinforcing steel in salt-contaminated concrete, in: D.E. Tonini, J.M. Gaidis (Eds.), Corrosion of Reinforcing Steel in Concrete, ASTM STP 713, American Society for Testing and Materials, 1980, pp. 316. [12] P. Vassie, Reinforcement corrosion and the durability of concrete bridges, Proceedings of the Institution of Civil Engineers Part 1 (76) (1984) 713723. [13] B. Elsener, H. Bhni, Corrosion of steel in mortar studied by impedance measurements, Electrochemical Methods in Corrosion Research, in: M. Duprat (Ed.), Materials Science Forum 8, 1986, pp. 363372. [14] S. Diamond, Chloride concentrations in concrete pore solutions resulting from calcium and sodium chloride admixtures, Cement, Concrete, and Aggregates 8 (1986) 97102. [15] C. Andrade, C.L. Page, Pore solution chemistry and corrosion in hydrated cement systems containing chloride salts: a study of cation specic effects, British Corrosion Journal 21 (1986) 4953. [16] B.B. Hope, A.K.C. Ip, Chloride corrosion threshold in concrete, ACI Materials Journal (JulyAugust 1987) 306314.

U. Angst et al. / Cement and Concrete Research 39 (2009) 11221138 [47] B.H. Oh, S.Y. Jang, Y.S. Shin, Experimental investigation of the threshold chloride concentration for corrosion initiation in reinforced concrete structures, Magazine of Concrete Research 55 (2003) 117124. [48] W. Morris, A. Vico, M. Vzquez, Chloride induced corrosion of reinforcing steel evaluated by concrete resistivity measurements, Electrochimica Acta 49 (2004) 44474453. [49] M. Moreno, W. Morris, M.G. Alvarez, G.S. Duff, Corrosion of reinforcing steel in simulated concrete pore solutions: effect of carbonation and chloride content, Corrosion Science 46 (2004) 26812699. [50] P.V. Nygaard, M.R. Geiker, A method for measuring the chloride threshold level required to initiate reinforcement corrosion in concrete, Materials and Structures 38 (2005) 489494. [51] D. Trejo, P.J. Monteiro, Corrosion performance of conventional (ASTM A615) and low-alloy (ASTM A706) reinforcing bars embedded in concrete and exposed to chloride environments, Cement and Concrete Research 35 (2005) 562571. [52] T.U. Mohammed, H. Hamada, Corrosion of steel bars in concrete with various steel surface conditions, ACI Materials Journal 103 (2006) 233242. [53] M. Manera, . Vennesland, L. Bertolini, Chloride threshold for rebar corrosion in concrete with addition of silica fume, Corrosion Science 50 (2008) 554560. [54] J. Gulikers, Considerations on the reliability of service life predictions using a probabilistic approach, Journal de Physique IV (136) (2006) 233241. [55] J. Gulikers, Probabilistic service life modelling of concrete structures: improvement or unrealistic? CONSEC'07, Proceedings of the 5th International Conference on Concrete Under Severe Conditions, Environment and Loading, Tours, France, 2007, pp. 891902. [56] K.Y. Ann, H.-W. Song, Chloride threshold level for corrosion of steel in concrete, Corrosion Science 49 (2007) 41134133. [57] G.K. Glass, N.R. Buenfeld, Chloride threshold levels for corrosion induced deterioration of steel in concrete, Proc. RILEM Int. Workshop Chloride Penetration into Concrete, Paris, 1997, p. 429. [58] W. Breit, Kritischer korrosionsauslsender Chloridgehalt Sachstand (Teil 1), Beton 7 (1998) 442. [59] RILEM, Draft recommendation for repair strategies for concrete structures damaged by reinforcement corrosion, Materials and Structures 27 (1994) 415436. [60] Model code for service life design, b Bulletin No. 34, 2006. [61] L. Bertolini, F. Bolzoni, T. Pastore, P. Pedeferri, Behaviour of stainless steel in simulated concrete pore solution, British Corrosion Journal 31 (1996) 218222. [62] D. Trejo, R.G. Pillai, Accelerated chloride threshold testing Part II: corrosionresistant reinforcement, ACI Materials Journal 101 (2004) 5764. [63] H. Yu, W.H. Hartt, Effects of reinforcement and coarse aggregates on chloride ingress into concrete and time-to-corrosion: Part 1 spatial chloride distribution and implications, Corrosion 63 (2007) 843849. [64] U. Angst, . Vennesland, Critical chloride content in concrete state of the art, in: Concrete Repair, Rehabilitation and Retrotting II. Proceedings of the 2nd Int. Conf. on Concrete Repair, Rehabilitation and Retrotting, Cape Town, South Africa, CRC Press/Balkema, The Netherlands, 2008, p. 149. [65] C.L. Page, K.W.J. Treadaway, Aspects of the electrochemistry of steel in concrete, Nature 297 (1982) 109114. [66] P. Schiessl, S. Lay, Inuence of concrete composition, in: H. Bhni (Ed.), Corrosion in Reinforced Concrete Structures, Woodhead Publishing Limited, Cambridge, 2005, pp. 91134. [67] K. Tuutti, Corrosion of Steel in Concrete, Swedish Cement and Concrete Research Institute, 1982. [68] ASTM C-1152. Standard test method for acid-soluble chloride in mortar and concrete, American Society for Testing and Materials. [69] C. Andrade, M. Castellote, Recommendation of RILEM TC 178-TMC: Testing and modelling chloride penetration in concrete: analysis of total chloride content in concrete, Materials and Structures 35 (2002) 583585. [70] G.K. Glass, N.R. Buenfeld, The presentation of the chloride threshold level for corrosion of steel in concrete, Corrosion Science 39 (1997) 10011013. [71] C.L. Page, J. Havdahl, Electrochemical monitoring of corrosion of steel in microsilica cement pastes, Materials and Structures 18 (1985) 4147. [72] G. Sergi, G.K. Glass, A method of ranking the aggressive nature of chloride contaminated concrete, Corrosion Science 42 (2000) 20432049. [73] G.K. Glass, B. Reddy, N.R. Buenfeld, Corrosion inhibition in concrete arising from its acid neutralisation capacity, Corrosion Science 42 (2000) 15871598. [74] O. Poupard, A. At-Mokhtar, P. Dumargue, Corrosion by chlorides in reinforced concrete: determination of chloride concentration threshold by impedance spectroscopy, Cement and Concrete Research 34 (2004) 9911000. [75] European Standard EN 206-1, Concrete Part 1: Specication, Performance, Production and Conformity, European Committee for Standardisation, 2000. [76] P.S. Mangat, K. Gurusamy, Corrosion resistance of steel bres in concrete under marine exposure, Cement and Concrete Research 18 (1988) 4454. [77] R. Cigna, C. Andrade, U. Nrnberger, R. Polder, R. Weydert, E. Seitz (Eds.), COST 521: Final Report Corrosion of Steel in Reinforced Concrete Structures, Luxembourg, 2002. [78] L. Bertolini, B. Elsener, P. Pedeferri, R. Polder, Corrosion of Steel in Concrete, WILEY VCH, 2004. [79] C.L. Page, Mechanism of corrosion protection in reinforced concrete marine structures, Nature 258 (1975) 514515. [80] C.L. Page, N.R. Short, A. El Tarras, Diffusion of chloride ions in hardened cement pastes, Cement and Concrete Research 11 (1981) 395406. [81] M.N. Al Khalaf, C.L. Page, Steel/mortar interfaces: microstructural features and mode of failure, Cement and Concrete Research 9 (1979) 197207. [82] L. Yue, H. Shuguang, The microstructure of the interfacial transition zone between steel and cement paste, Cement and Concrete Research 31 (2001) 385388.

1137

[83] A.T. Horne, I.G. Richardson, R.M.D. Brydson, Quantitative analysis of the microstructure of interfaces in steel reinforced concrete, Cement and Concrete Research 37 (2007) 16131623. [84] G.K. Glass, R. Yang, T. Dickhaus, N.R. Buenfeld, Backscattered electron imaging of the steelconcrete interface, Corrosion Science 43 (2001) 605610. [85] T.U. Mohammed, H. Hamada, A discussion of the paper Chloride threshold values to depassivate reinforcing bars embedded in a standardized OPC mortar by C. Alonso, C. Andrade, M. Castellote and P. Castro, Cement and Concrete Research 31 (2001) 835838. [86] A. Castel, T. Vidal, R. Franois, G. Arliguie, Inuence of steelconcrete interface quality on reinforcement corrosion induced by chlorides, Magazine of Concrete Research 55 (2003) 151159. [87] G.K. Glass, B. Reddy, N.R. Buenfeld, The participation of bound chloride in passive lm breakdown on steel in concrete, Corrosion Science 42 (2000) 20132021. [88] B. Reddy, G.K. Glass, P.J. Lim, N.R. Buenfeld, On the corrosion risk presented by chloride bound in concrete, Cement & Concrete Composites 24 (2002) 15. [89] C.K. Larsen, Chloride binding in concrete, Dr. Ing. Thesis, Report No 1998:101, Norwegian University of Science and Technology, NTNU, 1998. [90] H. Justnes, A review of chloride binding in cementitious systems, Nordic Concrete Research, Publication No. 21. 1/98, Nordic Concrete Federation, Norsk Betongforening, Oslo, 1998, pp. 4863. [91] L. Tang, L.-O. Nilsson, Chloride binding capacity and binding isotherms of OPC pastes and mortars, Cement and Concrete Research 23 (1993) 247253. [92] C.L. Page, . Vennesland, Pore solution composition and chloride binding capacity of silica fume-cement pastes, Materials and Structures 19 (1983) 1925. [93] C. Arya, N.R. Buenfeld, J.B. Newman, Factors inuencing chloride-binding in concrete, Cement and Concrete Research 20 (1990) 291300. [94] R.K. Dhir, M.R. Jones, Development of chloride-resisting concrete using y ash, Fuel 78 (1999) 137142. [95] O.A. Kayyali, M.N. Haque, The Cl/OH ratio in chloride-contaminated concrete a most important criterion, Magazine of Concrete Research 47 (1995) 235242. [96] S. Diamond, Effects of two danish yashes on alkali contents of pore solutions of cement-yash pastes, Cement and Concrete Research 11 (1981) 383394. [97] M. Kawamura, O.A. Kayyali, M.N. Haque, Effects of yash on pore solution composition in calcium and sodium chloride-bearing mortars, Cement and Concrete Research 18 (1988) 763773. [98] R.K. Dhir, M.A.K. El-Mohr, T.D. Dyer, Chloride binding in GGBS concrete, Cement and Concrete Research 26 (1996) 17671773. [99] R. Luo, Y.B. Cai, C.Y. Wang, X.M. Huang, Study of chloride binding and diffusion in GGBS concrete, Cement and Concrete Research 33 (2003) 17. [100] A. Cheng, R. Huang, J.-K. Wu, C.-H. Chen, Inuence of GGBS on durability and corrosion behavior of reinforced concrete, Materials Chemistry and Physics 93 (2005) 404411. [101] E. Mahallati, M. Saremi, An assessment on the mill scale effects on the electrochemical characteristics of steel bars in concrete under DC-polarization, Cement and Concrete Research 36 (2006) 13241329. [102] J.A. Gonzlez, E. Ramrez, A. Bautista, S. Feliu, The behaviour of pre-rusted steel in concrete, Cement and Concrete Research 26 (1996) 501511. [103] K. Tuutti, Effect of cement type and different additions on service life, in: Proc. Int. Conf. "Concrete 2000", Dundee, Scotland, UK, E & FN Spon, Chapman & Hall, London, 1993, pp. 12851295. [104] J. Tritthart, Chloride binding in cement. II. The inuence of the hydroxide concentration in the pore solution of hardened cement paste on chloride binding, Cement and Concrete Research 19 (1989) 683691. [105] C.M. Hansson, T. Frolund, J.B. Markussen, The effect of chloride cation type on the corrosion of steel in concrete by chloride salts, Cement and Concrete Research 15 (1985) 6573. [106] C.L. Page, N.R. Short, W.R. Holden, The inuence of different cements on chlorideinduced corrosion of reinforcing steel, Cement and Concrete Research 16 (1986) 7986. [107] . Vennesland, C. Andrade, M.A. Climent, R. Polder, J. Gulikers, Recommendation of RILEM TC 178-TMC: Testing and modeling chloride penetration in concrete: Method of eld determination of critical chloride content which causes corrosion of reinfocement, draft. [108] M. Stern, A.L. Geary, Electrochemical polarization. I. A theoretical analysis of the shape of polarization curves, Journal of the Electrochemical Society 104 (1957) 5663. [109] C. Andrade, C. Alonso, J. Gulikers, R. Polder, R. Cigna, . Vennesland, M. Salta, A. Raharinaivo, B. Elsener, Recommendation of RILEM TC 154-EMC: Test methods for on-site corrosion rate measurement of steel reinforcement in concrete by means of the polarization resistance method, Materials and Structures 37 (2004) 623643. [110] C. Alonso, C. Andrade, M. Castellote, P. Castro, Reply to the discussion by T.U. Mohammed and H. Hamada of the paper Chloride threshold values to depassivate reinforcing bars embedded in a standardized OPC mortar, Cement and Concrete Research 31 (2001) 839840. [111] B.B. Hope, J.A. Page, A.K.C. Ip, Corrosion rates of steel in concrete, Cement and Concrete Research 16 (1986) 771781. [112] J.A. Gonzlez, E. Otero, S. Feliu, W. Lpez, Initial steps of corrosion in the steel/Ca (OH)2 + Cl system: the role of heterogeneities on the steel surface and oxygen supply, Cement and Concrete Research 23 (1993) 3340. [113] R.K. Dhir, M.R. Jones, H.E.H. Ahmed, Determination of total and soluble chlorides in concrete, Cement and Concrete Research 20 (1990) 579590. [114] M. Castellote, C. Andrade, RoundRobin test on chloride analysis in concrete Part I: analysis of total chloride content, Materials and Structures 34 (2001) 532556.

1138

U. Angst et al. / Cement and Concrete Research 39 (2009) 11221138 [120] M. Castellote, C. Alonso, C. Andrade, P. Castro, M. Echeverra, Alkaline leaching method for the determination of the chloride content in the aqueous phase of hardened cementitious materials, Cement and Concrete Research 31 (2001) 233238. [121] AASHTO T260, Standard method of test for sampling and testing for chloride ion in concrete and concrete raw materials, American Association of State Highway and Transportation Ofcials. [122] B. Elsener, L. Zimmermann, H. Bhni, Non destructive determination of the free chloride content in cement based materials, Materials and Corrosion 54 (2003) 440446. [123] U. Angst, B. Elsener, C.K. Larsen, . Vennesland, Potentiometric determination of the chloride ion activity in cement based materials, submitted to Journal of Applied Electrochemistry.

[115] R.S. Barneyback, S. Diamond, Expression and analysis of pore uids from hardened cement pastes and mortars, Cement and Concrete Research 11 (1981) 279285. [116] J. Tritthart, Chloride binding in cement. I. Investigations to determine the composition of porewater in hardened cement, Cement and Concrete Research 19 (1989) 586594. [117] C. Arya, An assessment of four methods of determining the free chloride content of concrete, Materials and Structures 23 (1990) 319330. [118] G.K. Glass, Y. Wang, N.R. Buenfeld, An investigation of experimental methods used to determine free and total chloride contents, Cement and Concrete Research 26 (1996) 14431449. [119] C. Arya, N.R. Buenfeld, J.B. Newman, Assessment of simple methods of determining the free chloride ion content of cement paste, Cement and Concrete Research 17 (1987) 907918.