Corrosion behaviour of stainless steel reinforcement in concrete

3.1 Austenitic and ferritic SSs

Earlier studies were carried out on austenitic and ferritic grades of SS. From the late 1970s, the results of an extensive study on the behaviour of SS reinforcement in chloride-contaminated concrete were published (Brown et al., 1978; Flint & Cox, 1988; Treadaway et al., 1989; Cox & Oldfield, 1996). Tests in the above mentioned papers were performed on plain bars embedded in concrete prisms and beams made with 3.2% of mixed-in chlorides by mass of cement. Concrete beams, with a chloride content added in the mix up to 0.96% by mass of cement, were cracked before exposure to the aggressive environment. Concrete specimens were exposed to an industrial environment (Beckton, UK); furthermore, some cracked concrete beams, initially free of chlorides, were exposed to the splash zone of a marine environment (Langstone Harbour, UK). After 10 years of exposure, no corrosion was observed on austenitic SSs 302 (comparable to 304 but with a higher carbon content) and 316. Conversely, ferritic SSs 405 and 430 was shown to suffer pitting corrosion in high permeability concrete (w/c=0.75) with up to 1% chloride by mass of cement. Austenitic SSs did not show any sign of serious corrosion even after 22 years of exposure. Indeed, only slight pitting attacks occurred on 302 SS in concrete with 3.2% chloride by mass of cement (exposed to industrial atmosphere) (Cox & Oldfield, 1996).

Afterwards, several studies were carried out on 304 and 316 grades. The good corrosion resistance of austenitic SS was confirmed by a study on commercial ribbed bars of 304 and 316 austenitic SS bars in mortar (a sulphate-resistant portland cement was used) with up to 8% of mixed-in chlorides (Sorensen et al., 1990); no corrosion was observed in these specimens exposed outdoors for 5 months (Denmark). Further potentiostatic polarisation tests on the same SS at +200 mV/SCE (which can be considered representative of the maximum potential values reached in atmospheric exposure conditions) showed that corrosion could initiate at chloride contents of 3.5–5% by mass of cement and that the corrosion resistance of 316 was only marginally higher than that of 304.

A further study showed that fully passive conditions were maintained by plain bars of 304 and 316 austenitic SS embedded in concrete with mixed-in chlorides up to 3% by mass of portland cement and exposed in urban atmosphere (Milan, Italy) for 18 months (Pastore & Pedeferri, 1991).

Corrosion resistance of austenitic 316 Ti and ferritic 410 ribbed bars was also studied in alkaline and carbonated portland cement concrete contaminated with up to 2.5% chlorides by mass of cement (Nurnberger et al., 1993). Tests in cracked concrete, reinforced with austenitic and ferritic SS, initially free of chlorides and sprayed with chloride solution for 2.5 years, were also carried out (Nurnberger & Beul, 1999). Austenitic SS did not suffer pitting corrosion whether in alkaline or in carbonated concrete with up to 2.5% Cl⁻ and in cracked concrete specimens when a chlorides content of about 4.5% by mass of cement was reached at the steel surface. Conversely, ferritic SSs did not show pitting corrosion in alkaline concrete containing up to 2.5% Cl⁻ but were susceptible to pitting attack in carbonated concrete with 1% Cl⁻.

Tests in limestone-portland cement concrete contaminated with chlorides showed that bars of austenitic SS 304L and 316L did not undergo corrosion in alkaline and carbonated concrete with a chloride content up to 6% by cement mass, even after 8 months of exposure in very aggressive conditions simulating tropical climates (at 40°C and 95%–98% R.H.) (Bertolini et al., 2000). Also ferritic 430 steels were tested showing a lower corrosion resistance at 40°C compared to austenitic steels.

A further study showed that no corrosion occurred even at potential values up to +400 mV/SCE (potential values that can be reached only in the presence of external polarisation, as in the case of stray currents) on 304L and 316L in concrete with 5% of mixed-in chlorides. In carbonated concrete with 5% chlorides, a rather small increase in potential of 50 mV versus the free corrosion potential was enough to initiate pitting corrosion on 304L, showing that under this very aggressive condition (5% Cl⁻ + carbonation), 304L may be susceptible to pitting corrosion; for 316L, a higher anodic polarisation of 150 mV was required to initiate corrosion (Bertolini et al., 2002) (Figure 3).

Figure 3:

Results of potentiostatic polarisation tests on steel bars (φ 8 mm) embedded in alkaline (empty symbols) and carbonated (black symbols) concrete with 5% chloride by mass of cement (Bertolini et al., 2002).

In concrete with 2% and 4% mixed-in chlorides, 304 and 316, which were kept constantly wet using pads soaked in water, did not show initiation of pitting corrosion after approximately 2 years of exposure (García-Alonso et al., 2007a). Even in mortar with 2% and 5% mixed-in chlorides, the same rebars did not experience initiation of pitting corrosion after exposure to humidity cycles for about 3 years (García-Alonso et al., 2007b). No corrosion initiation was observed on 304L and 316L SSs during ponding tests, even with a chloride content at the steel surface of higher than 6% (Hartt et al., 2009).

Another study (Bertolini & Gastaldi, 2009) showed that corrosion initiation did not take place on ribbed bars of austenitic SS 304L and 316L, embedded in alkaline concrete with a content of chloride by mass of cement up to 8% and exposed at 20°C and 90% R.H. The passive condition was preserved also at temperatures higher than 40°C (Gastaldi & Bertolini, 2014).

In recent years, austenitic SS XM-28, with manganese added in place of nickel, has been proposed. Few works investigated its corrosion resistance, and it has been observed that it can be significantly affected by the differences in chemical composition (different producers vary the C, Ni and Mo content) and/or processing details (Hansson, 2016). According to Gastaldi et al. (2013), with potentiostatic polarisation tests at +200 mV/SCE for 24 h carried out on a set of 10 bars embedded in concrete with mixed-in chlorides in contents varying from 2% to 5%, Cl_th of XM-28 ranged between 2.3% and 5.5% by cement mass. According to Van Niejenhuis et al. (2016a), with potentiostatic polarisation tests at +200 mV/SCE for 4 days, corrosion occurred on three out of three bars embedded in concrete with 5% mixed-in chlorides and in two out of three bars embedded in concrete with 3% mixed-in chlorides. With potentiostatic polarisation tests at +100 mV/SCE for 4 days, corrosion did not occur on the three rebars in concrete with 3% mixed-in chlorides. In another study (Van Niejenhuis et al., 2016b), the XM-28 in sound concrete exhibited passive behaviour throughout more than 850 days’ exposure to salt brine containing 21% Cl⁻; however, no indication of chloride content at the bar depth at the end of exposure was reported, and hence, a minimum value for the critical chloride content cannot be discerned. Conversely, in concrete with cracks longitudinal to the bar, corrosion initiated after about 50 days.

The abovementioned studies show that traditional austenitic SS, i.e. of 316 and 304 grades, with and without molybdenum, have a high corrosion resistance in concrete. No corrosion was observed in alkaline concrete with chloride contents higher than 6% by mass of cement and even in carbonated concrete containing more than 4% chlorides by mass of cement. Even under aggressive exposure conditions simulating tropical climates, i.e. at 40°C and 95%–98% R.H., they remained passive. SS of 316 grade showed a better corrosion resistance than 304 only if anodic polarisation was applied. The more recent austenitic SS of grade XM-28 seems characterised by lower corrosion resistance in comparison with 304 and 316. Conversely, ferritic SSs showed poor corrosion resistance; pitting corrosion initiation was observed even with 1% chloride by cement mass.

3.2 Duplex SS

Austenitic-ferritic duplex SSs have also been proposed for use in concrete structures. The first study on this steel was performed in 1993 (Nurnberger et al., 1993); corrosion occurred neither in carbonated concrete with up to 2.5% Cl⁻ by cement mass nor in cracked alkaline concrete with a chloride content higher than 4.5% Cl⁻ by mass of cement.

The corrosion resistance of commercial ribbed bars of 22-05 SS embedded in concrete with 5% chlorides by mass of cement added in the mix was also analysed and compared to austenitic SS (Bertolini et al., 2002). Results showed that the corrosion rate of duplex SS was negligible in alkaline concrete in which either limestone-portland cement (CEM II/A-L) or ground granulated blast furnace slag cement (CEM III/B) was used, even during exposure at 40°C and 90%–95% R.H. Furthermore, corrosion initiated neither on bars in specimens in carbonated concrete with 5% chlorides by cement mass nor in alkaline concrete subjected to ponding with sodium chloride solution, although a chloride content higher than 6% by mass of cement was reached at the steel surface (Bertolini et al., 2001). Potentiostatic polarisation tests performed on these concrete specimens confirmed the outstanding corrosion resistance of 22-05 steel, showing that passivity was maintained even at potential values up to +400 mV/SCE both in alkaline and in carbonated concrete (Figure 3) (Bertolini et al., 2002). Conversely, as previously observed, austenitic SSs showed a decrease in corrosion resistance in carbonated concrete with 5% chloride when they were anodically polarised. A higher corrosion resistance of this SS, compared to 304L and 316L, in carbonated concrete and also in the presence of cracks was also observed by Islam et al. (2013), Bautista et al. (2015), and Van Niejenhuis et al. (2016b). This behaviour can be correlated to the high PREN of this SS grade.

The higher corrosion resistance of duplex SS 22-05 with respect to austenitic SSs was also confirmed by tests in alkaline (pH 12.6) and neutral (pH 7.5) solutions (Bertolini et al., 2002). Nevertheless, it was shown that the presence of welding oxide has a negative effect also on 22-05 SS, since it decreases the passivity range (Figure 4).

Figure 4:

Results of potentiostatic polarisation tests on SS bars (φ 8 mm) with simulated welding oxide in concrete with 5% Cl⁻ . White symbols show tests on annealed bars (Bertolini et al., 2002).

In recent years, the increase in the cost of the nickel has led manufacturers of SS reinforcing bars to develop low-nickel duplex SS rebars. Attention was initially focused on duplex SS of type 23-04 and, later on, on SS of type 21-01. Only few data are available in the literature regarding the corrosion resistance of these steels in concrete. Most of the published studies have been merely aimed at the evaluation of the pitting potential in alkaline solutions.

At the beginning of the 1990s, a study was carried out on 23-04 plain bars embedded in concrete with mixed-in chlorides up to 3% versus mass of cement and no corrosion was experienced during the 18 months of exposure in urban atmosphere (Pastore et al., 1991). Later on, tests were performed on the cross-section of bars of duplex austenitic-ferritic 23-04 and, for comparison, on austenitic 304, 304L, 316 and 316L in a solution simulating the concrete pore solution and in a solution with a pH of 12.6 at 20°C (Bertolini et al., 1996, 1998c). In potentiostatic polarisation tests, at +200 mV/SCE, a critical chloride content of 3.5% by mass in solution was found during tests at 40°C for 23-04 duplex SS, whilst for austenitic SS, corrosion took place at 4.5% by mass in solution.

Other authors (Bourgin et al., 2005) showed that in tests in alkaline solutions at pH 12 with 21 g/l of NaCl (about 1.3% Cl⁻ by mass of solution) and at a temperature of 50°C, neither duplex 23-04 nor 22-05 nor austenitic 304L and 316L SS bars suffered pitting corrosion during potentiodynamic polarisation tests.

In potentiostatic polarisation tests at +100 mV/SCE in alkaline solution (pH 13–13.5) at room temperature, 23-04 SS (in pickled condition) showed an average content of chlorides for corrosion initiation of 7.5% by mass of solution (Lysogorski & Hartt, 2006).

With potentiodynamic and potentiostatic polarisation tests at +200 mV/SCE in an alkaline solution (pH 12.6), at 20°C, the ribbed 23-04 rebars showed a pitting potential of +500/+600 mV/SCE with 5% chlorides versus mass of solution (in potentiodynamic tests) and the initiation of corrosion with chlorides content of 7.5%–8% versus mass of solutions (in potentiostatic tests) (Bertolini & Gastaldi, 2009, 2011).

In more recent years, few studies were carried out in concrete contaminated by chlorides, which, however, gave a different indication about the corrosion resistance of 23-04 duplex SS in comparison to tests in solution (Bertolini & Gastaldi, 2011). As a matter of fact, duplex 23-04 SSs showed pitting corrosion initiation even in concrete with 3% chlorides by cement mass (Bertolini & Gastaldi, 2009). Therefore, even though tests in solution suggested a corrosion resistance of 23-04 SS similar to that of 304L SS, tests in concrete showed a lower corrosion threshold for the former. This suggests that the behaviour of 23-04 SS in concrete needs to be further investigated.

According to Van Niejenhuis et al. (2016b), in sound concrete, 23-04 SS exhibited passive behaviour throughout the more than 850 days’ exposure to salt brine containing 21% Cl⁻ (no indication of the chloride content at the bar depth at the end of exposure is reported in the paper).

Even a cheaper duplex SS, i.e. 21-01, has been very recently proposed for reinforcing bars. Although the number of studies dealing with the corrosion resistance of this type of steel is rather modest, the available literature clearly shows that they have a chloride threshold even lower than that of 23-04 duplex SS. In potentiostatic polarisation tests at +200 mV/SCE in an alkaline solution (pH 12.6), the chloride content for corrosion initiation for 21-01 was lower than Cl⁻/OH⁻ ratio of 8 (corresponding to about 1% Cl⁻ versus mass of solution), whilst in the austenitic 316L, the Cl⁻/OH⁻ ratio was always higher than 20 (about 3% Cl⁻ versus mass of solution) (Hurley & Scully, 2005, 2006). Another study with potentiostatic polarisation tests in saturated calcium solutions showed that corrosion initiation on duplex 21-01 steel occurred with 3.5–6% chloride by mass of solution (Bertolini & Gastaldi, 2009) and showed that the corrosion resistance of this steel is lower than those of 23-04, 304LN and 316LN steels (Figure 5A).

Figure 5:

Chloride content for pitting corrosion initiation in potentiostatic polarisation tests at +200 mV/SCE in saturated Ca(OH)₂ solution at 20°C (A) and 40°C (B) carried out on two replicate specimens for each grade of SS (Bertolini & Gastaldi, 2009).

Others authors (Hartt et al., 2009) showed that in tests on concrete specimens (exposed in laboratory at room temperature) subjected to ponding, 21-01 steel bars showed a corrosion resistance lower than that of ferritic 410. The chloride threshold was estimated as the chloride level that led to the initiation of corrosion in 2% of the tested bars; a value of 1.33% chlorides by mass of cement was evaluated for 21-01 steel, which was even lower than the level of 1.63% chloride by cement mass that was obtained for ferritic 410.

In another study in chloride-bearing concrete exposed at 20°C and 90% R.H., pitting corrosion took place on 21-01 in concrete specimens with 3% chloride by cement mass (Bertolini & Gastaldi, 2009).

The previous results seem to be in contrast with the conclusions of some published works that claim that 21-01 SS has a corrosion resistance similar to that of 304L, 316L and 23-04 SS (Bergquist et al., 2005; Sederholm et al., 2009). The conclusion of these papers is unsubstantiated, as it is not based on the evaluation of the chloride threshold, but it is simply argued from the observation of a similar behaviour of the two grades of steel in mild exposure conditions (T=20°C).

In order to assess the effect of temperature on the corrosion resistance of 21-01 steel, potentiostatic polarisation tests at +200 mV/SCE were carried out in chloride-contaminated alkaline solutions (pH 12.6) at 40°C. Figure 5B shows that the chloride content for corrosion initiation on 21-01 SS decreases to about 2%, showing a reduction of about 60% compared to the values at 20°C (Figure 5A). Tests at a higher temperature were also carried out on SS bars embedded in concrete. Duplex 21-01 SS suffered pitting corrosion in concrete with 2.5% chloride by cement mass exposed to 40°C.

3.3 Final consideration

The previous analysis in the literature shows that the pitting resistance equivalent number (PREN=%Cr+3.3%Mo+16%–30%N), which is normally related to the ability of SS to resist pitting attack in neutral environments, may not be reliable in alkaline environments such as concrete. In particular, the presence of molybdenum seems to have a lower effect on corrosion resistance, as confirmed by the small differences observed between 304LN (without molybdenum) and 316LN; furthermore, nickel has been shown to give a positive effect on corrosion resistance especially in simulated tropical environment (Bertolini et al., 1996; Bertolini & Gastaldi, 2011; Elsener et al., 2011a,b). In carbonated concrete [or in the presence of cracks, as SSs are directly exposed to the atmosphere, as discussed by Van Niejenhuis et al. (2016b)], conversely, PREN index might still be a useful ranking index.

Furthermore, these results suggest that, depending also on the experimental techniques adopted by different researchers, a wide range of values of chloride threshold is reported. Nevertheless, the studies are in agreement to show that different grades of SSs have different chloride threshold for corrosion initiation. In general, a very good behaviour is reported for duplex 22-05 and austenitic 316L SSs, even in concrete with high chloride contamination (e.g. 5% chlorides by mass at the steel surface) and exposed to hot environments (e.g. 40°C or higher). These types of SSs therefore may be appropriate for use in harsh environments where high chloride contamination is associated to wet and hot exposure conditions. Austenitic 304L showed, in concrete, a slightly lower corrosion resistance than did 316L and 22-05. All other types of SS bars considered in this review were shown to have a lower corrosion resistance and therefore are likely to be appropriate only in milder environments. Anyway, the dissimilar testing procedures used in the different works make the definition of design values of the chloride threshold for the different grades of SS exposed to different environmental conditions difficult. In fact, only general information about the corrosion resistance of the austenitic and duplex SS are reported on technical reports available to the designer (The Concrete Society, 1998; The Concrete Society, 2008). In Figure 6, an attempt has been made to depict approximate values of the range of Cl_th for the different grades of SS rebars, based on previously reported data. Values reported in this figure should be assumed only as indicative values. In the figure, it can be observed that, although indicative, the range of Cl_th for 21-01 SS is between 1.33% and 3% by mass of cement and, for XM-28 SS, it is between 2.3% and 5% by mass of cement. For the other SSs, only a minimum value can be detected, considering that, in real cases, chloride contents higher than 5% by mass of cement are unlikely reached at the concrete cover depth of the steel surface and, hence, Figure 6 has been limited to this value (values suggested in literature exceeding 5% have been identified by a rightward arrow). The minimum values of Cl_th for austenitic 304 and 316 SS were around 3.5% by mass versus cement, detected under anodic polarisation (Sorensen et al., 1990), whilst the Cl_th minimum value for duplex 23-04 SS was 2.7% by mass of cement. For duplex 22-05 SS, values higher than 5% by mass of cement were detected in all the studies, and hence, it can be considered that Cl_th for this SS is higher than 5% by mass of cement.

Figure 6:

Approximate values of Cl_th for different grades of SS rebars obtained from the different studies (*=min. value detected under anodic polarisation by Sorensen et al., 1990).

3.4 Practical aspects

As far as practice at the construction site is concerned, it should be observed that compared with other types of corrosion-resistant rebars (such as epoxy coated or galvanised steel), corrosion resistance is a bulk property of SS. Therefore, the integrity of SS is unaffected if its surface is cut or damaged during handling. Obviously, this does not apply to clad bars (i.e. usual carbon steel bars clad with a thin layer of SS), which, in some cases, have been proposed as a cheaper alternative to solid SS bars (which have not been considered in this paper).

The chemical composition of the SS reinforcement guarantees weldability (mainly obtained by decreasing the carbon content). Nevertheless, welding is not recommended under site conditions unless adequate control is maintained; in fact, welding may have some negative consequences with regard to mechanical properties and corrosion resistance. Welding oxides and the mill scale, which negatively affect corrosion resistance, should be removed. As a matter of fact, the good corrosion resistance of austenitic grades may be affected by the presence of oxide produced at high temperature (i.e. mill scale or welding oxides). The influence of welding oxide scale on the surface of the austenitic SS was analysed by several authors. All of the published studies show that high-temperature oxides have a deleterious effect on the corrosion behaviour of the SS bars. For instance, it was shown that, in the presence of a welding oxide scale on the surface of austenitic 304 and 316 SS embedded in mortar, the chloride threshold decreased from up to 8% by cement mass to 5%–8% in the specimens in free corrosion condition (exposed outdoor) and from 3.5%–5% to 1%–2% in potentiostatic tests at +200 mV/SCE (Sorensen et al., 1990). In the study by Nurnberger et al. (1993), it was demonstrated that ferritic SS, which did not show pitting corrosion in alkaline concrete containing up to 2.5% Cl⁻, was susceptible to pitting attack in the presence of welding oxide in specimen with 1% Cl⁻. Pits were also observed on the welded bars of 316Ti SS bars embedded in cracked concrete specimens where a chloride content of about 4.5% by mass of cement was reached at the steel surface. These results show that the corrosion resistance of the SS bars is remarkably affected by the presence of high-temperature oxide scale. Therefore, it is good practice to fully remove both the mill scale and welding oxides from the surface of the steel.

The coefficient of thermal expansion of austenitic steels is higher (about 1.8×10⁻⁵°C⁻¹) than that of concrete and of the traditional carbon steel bars (about 10⁻⁵°C⁻¹); austenitic-ferritic steels are in an intermediate position. Although this may raise concern about differential expansion especially during fire, no cases of damage have been reported. Furthermore, the thermal conductivity of austenitic SS is much lower than that of carbon steel, and thus, the increase in temperature throughout the steel is delayed. Finally, SSs generally retain more of their room-temperature strength than carbon steel above temperatures of about 550°C and more of their stiffness than carbon steel across the whole temperature range (Gardner & Ng, 2006). Austenitic SSs are generally considered non-magnetic (although cold drawing can increase their magnetic permeability) (Bertolini et al., 2013).

In the past, concern has been expressed about the risk of galvanic corrosion of black steel induced by coupling with SS bars, in order to use the more expensive SS bars only where it is necessary to fulfil durability requirements. In principle, interaction of metals with different electrochemical behaviour could both promote corrosion initiation (e.g. due to anodic polarisation) and increase the corrosion rate (due to macrocell effects) in one of the coupled metals. Experimental tests (Bertolini et al., 1998a,b,c) showed that when both carbon steel and SS rebars are passive and embedded in aerated concrete, the macrocell does not produce appreciable effects, as both steels have almost the same free corrosion potential. Only when black steel undergoes corrosion (i.e. due to bad design or execution that does not prevent chloride penetration for the expected SL) that the macrocell current becomes significant. However, as shown in Figure 7, the increase in corrosion rate on carbon steel embedded in chloride-contaminated concrete due to galvanic coupling with austenitic SS (316L) is significantly lower than the increase brought about by coupling with passive black steel (Bertolini et al., 1998a,c; The Concrete Society, 1998). As corroding black steel is always coupled with passive areas of black steel, further coupling with SS minimises the increase of corrosion rate. By means of cathodic polarisation curves, it was shown that this behaviour is explained by higher overvoltage for the cathodic reaction of oxygen reduction on austenitic SS with respect to carbon steel (Figure 8).

Figure 7:

Macrocell current density exchanged between a corroding bar of carbon steel in 3% chloride contaminated concrete connected with a (parallel) passive bar of carbon steel in chloride-free concrete, 316L SS in chloride-free concrete and in 3% chloride-contaminated concrete (20°C, 95% R.H.). Results on SS bars with simulated welding scale are also reported (The Concrete Society, 1998).

Figure 8:

Cathodic polarisation curves in saturated Ca(OH)₂ solution (pH 12.6) of 316L SS compared with the curve of carbon steel. Tests on SSs with simulated welding scale are also shown (Bertolini et al., 1998b).

As a consequence of these findings, even in technical reports, it is stated that SS bars will minimise problems that could occur in neighbouring corroding and passive areas (The Concrete Society, 1998).

Furthermore, increased galvanic current can occur in the presence of welding scale on SS. Oxide scale produced at high temperatures increases the macrocell current density generated by SSs, to the same order of magnitude or even greater than that produced by coupling with carbon steel (Figure 7) (Bertolini et al., 1998a). This risk can be avoided by removing the welding scale, either mechanically or by pickling.

Further studies have confirmed that macrocell formed when duplex 22-05 SS was electrically connected with corroding carbon steel is lower with respect to those due to coupling with passive carbon steel. Nevertheless, a higher macrocell current density was generated with respect to those produced with an austenitic SS cathode. This behaviour can be explained with the different kinetics of oxygen reduction on duplex SS; in fact, duplex SS has shown a lower overvoltage for cathodic reaction with respect to that of austenitic SS (Gojkovic et al., 1998; Sagües et al., 2005). This suggests that cathodic reaction is less inhibited in steels with the presence of the ferritic phase together with the austenitic one. No data are available for duplex SS with lower nickel content (i.e. 23-04 and 21-01).