N. S. Berke1, M. C. Hicks2
1Tourney Consulting Group, Kalamazoo, Michigan, USA 2Grace Construction Products, Cambridge, Massachusetts, USA
Steel reinforced concrete is one of the most durable and cost-effective construction materials, but it can suffer in high chloride environments from corrosion due to chloride-induced breakdown of the normal passive layers protecting the steel. One way of protecting embedded steel reinforcement from chloride induced corrosion is by the addition of corrosion inhibiting admixtures. The most widely used corrosion-inhibiting admixture is calcium nitrite, due to its excellent inhibitor properties and its benign effect on concrete properties. One advantage to calcium nitrite is that its protective mechanism is well defined. In this paper, data are presented that show the levels of chloride to which given levels of calcium nitrite will protect. Furthermore, it will be shown that once corrosion infiltrates, the rates are lower with calcium nitrite present. Finally, it is demonstrated how these results can be used by the design engineer in an integrated durability model to produce reinforced concrete structures with durabilities in excess of 50-100 years.
Concrete, Durability, Corrosion inhibitors, Calcium nitrite, Design life modelling.
To predict the corrosion service life of a structure, it is necessary to be able to predict the rate of chloride ingress to the steel reinforcing, the cover over the steel and the level of chloride necessary to start corrosion. Corrosion inhibitors that increase the chloride threshold level for corrosion initiation provide a major advantage in that the engineer can use the extensive data on protection levels provided to design for service life. Using historical chloride exposure levels for a given concrete quality and known concrete cover over the reinforcement steel, the engineer is able to estimate the time for reaching the chloride level for corrosion initiation. Thus, an inhibitor that increases the threshold value of chloride will extend the time before the onset of corrosion and provide an increase in service life.
In this paper, we present a methodology for predicting an increased threshold for chloride induced corrosion, which can then be utilised with models that address the issue of chloride ingress into concrete over time, to predict service life extension through the use of calcium nitrite (inorganic corrosion inhibitor).
1.1 Corrosion Principles
A brief review of the corrosion behavior of steel in concrete is useful in understanding how the protection tables for calcium nitrite corrosion inhibitor have been developed. Even in the absence of chloride or carbonation, steel in concrete is always demonstrating a small trickle of background corrosion current. This trickle of current will subside to very low levels as the steel approaches the passive state caused by the high pH (>12.5) environment existing in concrete. Over the course of a few months, this highly alkaline environment builds up a protective oxide layer on the steel. Even in the passive state there is always a small corrosion current present from the process of maintaining the protective passive oxide. This results in no significant rust buildup or expansive processes, therefore, the steel will remain in this condition for centuries with no loss in performance.
When chloride enters the concrete and reaches the steel, the normal passivity is disrupted, and active pitting corrosion initiates. Pitting is a very localised form of corrosion and the corrosion rates at the pits are orders of magnitude higher than the passive rate. However, since only a small portion of the steel is initially pitting, the actual corrosion rates measured over the entire exposed area of steel will initially show only a 4-10 fold increase. As corrosion occurs in a pit, the local pH becomes lower and a more active corrosion process occurs in which more of the surface is corroding, and the measured corrosion rates increase. In other words, the entire corrosion process is accelerated.
Anodic inhibitors, such as calcium nitrite, strengthen the process of producing a passive layer on the steel, and thereby raise the concentration of chloride at which the passive corrosion process turns into an active or pitting process. A higher concentration of nitrite allows the passive layer to strengthen and resist a higher level of chloride. This is the threshold chloride content for that calcium nitrite dosage. This effect is well documented by Rozenfeld in his book on inhibitors . Numerous researchers outside our laboratory have also documented that nitrite increases the chloride content at which pitting and active corrosion initiate for steel in concrete and concrete-type environments [2-12]. As will be explained in detail later, even though these references clearly show an increase in threshold values for active corrosion initiation, the precise valued needs to be determined from longer-term chloride ingress experiments, which accurately reflect the exposure conditions of field concrete.
Admixed chloride tests are useful to show that calcium nitrite is an inhibitor, but as pointed out by Rozenfeld , could require more inhibitor than what would be needed if the inhibitor was present before exposure to chloride. Therefore, protection levels are based upon ingressed chloride in better quality concretes to better represent the slow ingress of chloride in the field.
1.2 Development of Calcium Nitrite to Chloride Content Protection Tables
Prior to the early 1980s, our studies in concrete used electrode (half-cell) potentials to indicate the performance of calcium nitrite . Chloride analyses were performed and active corrosion was considered not to be occurring until corrosion until electrode potentials were more negative than -350 mV vs. copper-copper sulfate electrode (CSE). This was based upon an earlier ASTM C 876 test method version replaced in 1991, and reflected the state of the art at that time. Original nitrite-to-chloride protection levels were based upon these methods.
In the 1980s, extensive work by us and others made it clear that the
electrode potential measured in ASTM C 876 is very much affected by
the moisture content in the concrete. For water-saturated concretes,
potentials more negative than
-350 mV vs. CSE are not necessarily indicative of corrosion [15, 16]. On the other hand, in test regimes where severe drying occurs, corrosion has been reported at more positive potentials in the -240 mV vs. CSE range . This is either due to more oxygen availability or more likely to junction potential errors that can cause the measured potential to be as much as 200mV more positive . As an example, via autopsy in our laboratory, we found that no corrosion was occurring in one of the decks in Ref.  even though the electrode potentials taken before the autopsy were at
-473 mV vs. CSE. Since the decks were ponded daily and always wet, this is not unexpected in light of current knowledge. Thus, there is no single electrode potential value, which uniformly defines the limit between passive and active corrosion behavior.
Accordingly, based upon internal observation of this effect, in the mid-1980s a reevaluation of the procedure for determining the chloride-to-nitrite ratio for corrosion initiation was started, and we chose to use only direct autopsy results in future laboratory work in support of corrosion protection dosage recommendations, Table 1.
It should be strongly emphasised that confirmation of the correctness of Table 1 was carried out on this basis. As will be shown later, the time for autopsy was determined by using electrochemical measurements of the corrosion rate. The autopsies involve a physical examination of the reinforcing steel and measurement of the average chloride content at the steel level. If corrosion is observed, the chloride content at the exact corrosion location is determined provided the sample permits such analysis. Determining the chloride content at the exact corrosion site is critical since the chloride content is not constant due to the inhomogeneity of the concrete. Comparisons between chloride and nitrite contents are based on the calcium nitrite dosage added to the concrete.
A paper by Rosenberg and Gaidis  used electrode (half-cell) potentials to determine the chloride-to-nitrite protection ratio. The high scatter in their Fig. 3 reflects the poor correlation of electrode potential to actual chloride content. Since this work was done well prior to 1985, it does not take into account the effects of the concrete moisture on potential readings, and does not reflect the methods we use today to validate our dosage table. However, further studies have been undertaken to confirm the original dosage conclusions of the Rosenberg and Gaidis paper. Recently, both Feliu et al  and Weyers et al  have shown that the electrode potential is a poor indication of absolute corrosion activity.
2. Results of field and laboratory chloride/nitrite measurements
Fig. 1 shows the ingressed chloride contents at the reinforcing level for many concretes with different addition rates of calcium nitrite, different concrete qualities, and various covers, as reported in numerous references [5, 7, 19-29]. The line represents the dosage recommendations given in Table 1. Open symbols represent those cases where the steel was not corroding, as noted by visual analysis of the steel, and the solid symbols are where actual corrosion damage was found on the steel. Specimens are from both laboratory studies and field sites, and represent work conducted at our laboratory and field sites as well as by others. For the seven data points from field sites, corrosion rate or potential measurements and physical appearance of the concrete were used to assess performance, since the structures could not be sacrificed for complete autopsy. These data are presented in Fig. 2.
It can be seen that by far the greatest number of points follow the prediction of the Table 1, and that in many cases concretes expected to be corroding due to chloride content in excess of the predicted protection level are nevertheless not corroding.
In Fig. 1, there are only two cases where corroding points are very slightly below the protection line from Table 1. They are both from laboratory studies using a low cover, short curing, and a harsh drying cycle exposure regime. These specimens had 25 mm of cover and the chloride values were obtained at that depth. Since these data were obtained in an outside lab, we were not able to sample and perform chloride analysis of the concrete at the actual points of corrosion. However, autopsy of a similar concrete specimen from that laboratory showed as much as a factor of three increase in chloride contents at individualised points vs. the average value at 25 mm.
This was due to a high chloride content right next to a 19mm coarse aggregate particle. Had actual chloride analyses been performed next to the corrosion sites in these samples, the ratios would most likely have been well above the recommended protection line in Fig. 1. These data are included for completeness despite the less than precise knowledge of the chloride content at the corroding sites. It should also be noted that these concretes were of relatively low quality (0.5 water-to-cement ratio [w/c] and low cover), and thus would not meet American Concrete Institute (ACI) recommendations for cover and quality in corrosive environments . Finally, very rapid ingress of chloride as occurred in these concretes can lead to misleading conclusions, since the chloride can reach the steel before the full passive barrier is established. Overall, the data collected from laboratory and field structure studies for the last 20 years clearly support the protection levels for ingressed chloride given in Table 1.
As shown above, electrode potentials are not directly related to the
corrosion rate. The lack of oxygen in
water-saturated concrete results in more negative corrosion potentials. Drier concrete can result in junction potentials which cause the corrosion potential to be more positive , and which accounts for some observations of corrosion occurring at potentials more positive than -350 mV vs. CSE. Thus, statements about corrosion activity made solely on the basis of electrode potentials are unwarranted in light of current knowledge. The current version of ASTM C 876 has moved the interpretation of electrode potentials into the non-mandatory appendix; this reflects the lack of consensus regarding their use in predicting corrosion activity.
On the other hand, when properly interpreted, corrosion rate measurements using polarisation resistance or electrochemical impedance can be a good indication of corrosion activity in the laboratory [31-33], where the area of steel affected in the test is known and results can be verified by autopsies. Macrocell measurements (current between a bar in chloride containing concrete and concrete without chloride) can be misleading because corrosion could be occurring due to local anodic sites. In any event, we use these tests only in developmental research and as an indication that an autopsy should be performed. A few examples of how these were used in determining autopsy times are given below.
Figs. 3 and 4 provide laboratory corrosion rate data for several of the specimens used to develop the chloride threshold levels. They are given to illustrate how corrosion rate measurements can be used to determine autopsy times, and to show that the corrosion rate techniques give a good representation of the corrosion activity.
Fig. 3 gives corrosion rate measurements on concrete lollipops (rebar in concrete cylinder exposed to NaCl) before removal for autopsy, from data in Ref. . The corrosion rate is expressed in terms of the inverse of the polarisation resistance, Rp. Values above 25 µS/cm2 are indicative of severe corrosion [21, 31]. The corresponding chloride to nitrite ratio at the time of autopsy and autopsy results are also given for each point. These data clearly show that corrosion rate measurements are a good method to zero in on when to autopsy. The highly corroding specimens were produced with high w/c ratios to accelerate chloride ingress and in some cases the rapid ingress of chloride might provide an over accelerated test as noted in the previous section.
Fig. 4 shows modified ASTM G109 type minibeam specimens tested for 5 years from data in Ref. . The chloride-to-nitrite levels were determined after 5 years and autopsy data showed no corrosion on the bars, in good agreement with the macrocell currents. Since calcium nitrite does not significantly change resistivity between the top and bottom mats, the macrocell technique is useful in determining corrosion activity. The corrosion rate is measured in µA, and integration over time is the total macrocell corrosion and is expressed in units of current x time (µA x months).
In field structures, the area of steel polarized in a corrosion test is not exactly known, so corrosion rate tests are best used as a guide as to where cores or autopsies are to be performed. Electrode potential contours can be used to identify sites for autopsies as noted in ASTM C 876. However, in marine structures, potential values in the tidal zone and below can be misleading, as noted in the literature [15, 16]. Table 2 provides some corrosion rate data on beams that were exposed in a power plant cooling tower using brackish water and returned to the laboratory for polarisation resistance measurements of corrosion rate . The data show that corrosion rate measurements clearly identified corrosion in control specimens and that these relatively early age specimens with low chloride-to-nitrite ratio were not corroding.
2.1 Corrosion Rates after Corrosion Initiation
In addition to increasing the threshold value for corrosion initiation, several laboratory tests show that calcium nitrite reduces corrosion rates after corrosion initiation as shown in Figs. 5 and 6. After corrosion has occurred, corrosion rates are less for concretes containing calcium nitrite as seen by the lower slopes in the integrated total corrosion curves. Fig. 7 shows that a reduction in chloride levels, made possible by the addition of two dosages of silica fume (different dosage rates reduce chloride diffusion to the level needed for a given service life), can increase time to corrosion initiation, but not reduce the rate once corrosion starts. A combined increase in chloride threshold levels and lower corrosion rates after initiation should lead to significantly improved times to repair in reinforced concrete structures subjected to chlorides.
A study in the Strategic Highway Research Program indicated that there was about 2.5-5 years from severe corrosion initiation to repair in US highway bridges . Reduced corrosion rates after initiation could extend these times, and more laboratory and fieldwork is in progress to quantify the potential improvements. Thus, a conservative approach for now would be to base future time to repairs on the time to corrosion initiation.
2.2 Life-Cycle Modelling
Considerable work has been conducted in the area of life-cycle modelling using Ficks’ second law of diffusion to obtain an effective diffusion coefficient for chloride in concrete [24, 35-40]. Corrosion is typically considered to start at approximately 0.9 kg/m3 of chloride (i.e. ~0.04% by weight of concrete, assuming a concrete bulk density of 2400 kg/m3) at the reinforcement level and, as noted earlier, repairs might occur within 2.5-5 years of initiation. In this section, chloride ingress is modelled for typical structures that would be found in a marine environment. The benefits of calcium nitrite in increasing chloride threshold values are highlighted. Note that with good cover and low w/c ratios, cracks under approximately 0.2-0.3 mm play a minor role in the overall corrosion [41, 42].
Effective diffusion coefficients (Deff) as a function of temperature are given for two good quality concretes in Table 3.
These represent a typical 0.4 w/c concrete with ordinary Portland cement and a similar concrete produced with silica fume, fly ash, or granulated ground blast furnace slag (mineral additions). In general, silica fume is the most efficient of the three in reducing diffusion coefficients, at an early age, and its addition rates would be lower for an equivalent diffusion coefficient. Corrections for temperature in this case were made by converting data at 22˚C (295 K) by using the following equation :
Where k=5450 and D2 is the effective diffusion coefficient at temperature T2 and D1 is the effective diffusion coefficient at T1 which is 295 K in this case. Chloride ingress tests on the particular concrete designs of interest will provide more accurate estimates of Deff.
Fig. 8 illustrates the estimated chloride concentrations 75 mm in
from each face at the corner of a square pile in the splash-tidal
zone at an average yearly temperature of 19˚C, which is similar to
that of the water in Melbourne, Australia.
A two-dimensional solution to Fick’s second law was used, since chloride is entering from two faces. Even though the concrete cover is high, and both concretes are of good quality, chloride contents are in excess of 1 kg/m3 in only 8 years for the ordinary Portland cement at 0.4 w/c and 18 years when mineral additions (supplementary cementitious materials, SCMs) are added. Thus, reducing permeability, while clearly beneficial, is insufficient alone to reach design lives in excess of 50 years.
A conservative design life of 50 years can be obtained by adding calcium nitrite corrosion inhibitor. The 0.4 w/c concrete will have an estimated chloride content of 9.4 kg/m3 at 50 years and the addition of mineral admixtures will reduce the chloride levels to 5.2 kg/m3.
Thus, from Table 1, 30 or 15 l/m3 of 30% calcium nitrite would be needed depending upon the concrete chosen to extend severe corrosion initiation times to 50 years. Furthermore, there is a clear advantage in the combination of calcium nitrite with lower permeability concrete as can be seen by the reduced dosage needed. To extend corrosion initiation to 100 years, 25 l/m3 of 30% calcium nitrite can be used with the low permeability concrete with mineral additions (SCMs).
The above example involved relatively high average temperatures. Fig. 9 shows that in environments that are colder, significantly less chloride reaches the steel. However, chloride will still surpass 1 kg/m3 for the concrete with SCMs within 33 years and supplemental protection such as using calcium nitrite would be needed to reach longer design lives.
Marine walls in the splash-tidal zone have a reduced ingress of chloride as they are subjected to ingress in only one-dimension. Fig. 10 shows the estimated profiles for the same quality of concretes as in Fig. 8 in the same marine environment. At a cover of 65 mm, the times to chloride reaching 1 kg/m3 are the same as in the square pile case with 75 mm of cover; however, there is considerably less chloride at 50 and 100 years. For protection to corrosion initiation at 100 years, the 0.4 w/c needs approximately 18 1/m3 of 30% calcium nitrite.
Reinforced concrete that is not directly in the splash-tidal zone still needs to be protected from airborne chlorides. This is illustrated in Fig. 11 which shows the estimated chloride content at 40 mm for a wall subjected to a mild chloride build-up from airborne chlorides of 0.2 kg/m3 occurring after 90 years. Chloride contents of 1 kg/m3 are exceeded at 22 and 33 years for concretes at 0.4 w/c and 0.4 w/c with SCMs respectively. The addition of approximately 12 l/m3 will extend the time to corrosion initiation to approximately 65 years for the lower permeability concrete.
Recent advances in modelling include the development of mechanistic programs that address the true transport properties of ions through concretes and the chemical reactions that occur in the cementitious matrix. One such program, STADIUM® does this and accounts for the movement of ions in non-saturated concretes . In addition, probabilistic tools can now be applied to STADIUM and the other simpler models described above . Figure 12 shows how the probability for corrosion to occur is predicted by using STADIUM. Design engineers and owners can designate the time to a given amount of corrosion initiation or damage. Typically, a value of 10% is used for critical elements. Elements that are easily repaired with minor disruption to operations could have a threshold of 20% or higher. Note that the previous curves were all at the 50% level.
Based upon extensive work performed on the corrosion of steel in concrete with and without calcium nitrite corrosion inhibitors, the following conclusions can be made:
- The 30% calcium nitrite solution protection levels given in Table 1 are appropriately but not excessively conservative. The method used today to support the table values is straightforward examination and analysis of concrete samples.
- In no test where actual chloride content at the corroding sites has been measured have dosages, predicted by the protection table, failed to give the required protection.
- Arguments questioning the table based on interpretation from electrode potential measurements are not justified, based on the known lack of direct correlation between potential and corrosion activity, and based on the fact that the table is supported by direct measurement.
- Review of test data should be carefully performed with special consideration to the conditions used in the actual testing. Rapid ingress of chloride as occurs in some testing protocols can lead to misleading conclusions, since the chloride can reach the steel before the full passive barrier is established [46, 47].
- Corrosion rate data on laboratory specimens are in good agreement with autopsy results, but are not a substitute for verification by autopsy.
- The data collected from laboratory and field structure studies for the last 20 years clearly support the protection levels for ingressed chloride given in Table 1.
- Calcium nitrite does not increase corrosion rates after chloride protection values are surpassed, and in contrast often lowers them.
- A major advantage to the use of calcium nitrite corrosion inhibitor is that the engineer can use rational procedures based on chloride exposure, concrete quality and quantity of calcium nitrite to design for service life on the basis of expected chloride-to-nitrite ratios.
The authors could not have written this paper without the extensive work of numerous technicians over the past 30 years. They wish to thank L. Roberts for his valuable input and suggestions, and the management at Grace Construction Products for supporting these studies.
 Rosenfeld IL. Corrosion inhibitors. New York: McGraw-Hill Inc.; 1981. P. 327.
 Andrade C et al. Cem Concr Aggregates 1986; CCAGDP 8(2):110-6.
 Hope B, Ip AKC. ACI Mater J 1989:602-8.
 Tomosawa F et al. Niho Architecture Society, October 1987.
 Tomasawa F et al. RILEM Durability Symposium. 1990. P. 382-91.
 Virmani UP et al. Report No. FHWA/RD-83/012. Federal Highway Administration, Washington, DC, 1982. P. 71.
 Wiss Janney Elstner Associates, Inc. Report No. 850498, Skokie, IL, September, 1987.
 Briesemann D. Civil Engineering Department, Munich Technical University, Germany.
 Craig RJ, Wood LE. Highway Research Record No. 3328, 1970. P. 77.
 Ramirez E, Gonzalez JA. BR Corros J 1993;28(2):121-4.
 Laboratory performance of corrosion control methods. Interim Report, Res. Program Div. of Planning, South Dakota Department of Transportation, Pierre, South Dakota, March 1984.
 Hinatsu JT et al. J Appl Electrochem 1990;20:841-7.
 Gaidis JM, Rosenberg AM, Salek J. ASTM STP 713. Philadelphia: ASTM; 1980. p. 64.
 Gaidis JM, Rosenberg AM. Cem Concr Aggregates 1987;9(1);30-3.
 Sharp JV et al. ACI SP-109. American Concrete Institute, Detroit, MI, 1988. P. 105-25.
 Krauss PD et al. ASTM STP 1276. American Society for Testing and Materials, 1996.
 Pfeifer DW et al. Report No. FHWA/RD-86/193. Fed. Highway Adm. Washington, DC, 1987.
 Bennett JE et al. Corrosion 92, Paper No. 191. NACE International, Houston, TX, 1992.
 Feliu S et al. ASTM STP 1276. American Society for Testing and Materials, 1996.
 Weyers RE et al. ASTM STP 1276. American Society for Testing and Materials, 1996.
 Berke NS. Corrosion 87, Paper No. 132. NACE International, Houston, TX, 1987.
 Berke NS, Rosenberg A. Transportation Research Record 1211, TRB, Washington, DC, 1989.
 Berke NS et al. 12th Int. Corrosion Congress, Houston, TX, 1993. P. 3271-7.
 Berke NS, Hicks MC. ASTM STP 1137. American Society for Testing and Materials, 1992. P. 207-31.
 Berke NS et al. Corrosion 89, Paper No. 386. NACE International, Houston, TX, 1989.
 Berke NS, Hicks MC. Concrete durability in the Arabian Gulf. 1995. P. 119-32.
 Berke NS et al. Corrosion 1993;49(11):934-43.
 Berke NS et al. ACI SP-151. American Concrete Institute, 1994. P. 43-72.
 Berke NS et al. ASTM STP 1276. American Society for Testing and Materials, 1996.
 ACI-318. Building code requirements for reinforced concrete. ACI manual of concrete practice. Part 3, Use of concrete in buildings—Design, specifications, and related topics. American Concrete Institute, Detroit, MI, USA.
 Berke NS, Hicks MC. ASTM STP 1000. American Society for Testing and Materials, 1990. P. 425-40.
 Gonzales JA et al. Corrosion 85, Paper No. 257. NACE International, Houston, TX, 1989.
 Sagues AA. Corrosion 89, Paper No. 25. NACE International, Houston, TX, 1989.
 Strategic Highway Research program. Report No. SHRP-S-668, National Research Council, Washington, DC, 1994.
 Bamforth PB. Concrete across borders. Odense, Denmark, June, 1994.
 Bamforth PB. U.K. Corrosion & Eurocorr 94, November, 1994.
 Sagues AA et al. Corrosion of reinforcement in concrete construction, SCI, 1996. p. 105-14.
 Liu Y, Weyers RE. Corrosion of reinforcement in concrete construction. SCI, 1996. p. 88-104.
 Brown RD. Durability Building Mater 1982;1(2):113-25.
 Berke NS, Hicks MC. Corrosion 1994;50(3):234-9.
 Raupach M. Corrosion of reinforcement in concrete construction. SCI, 1996. p. 13-23.
 Beeby AW. Concr Int 1983;5(2):35-68.
 Page CL, Short NR, El Tarra A. Cem Concr Res 1991;21:581-8.
 Marchand J, Snyder KA, Beaudoin JJ. Cem and Concr Res 2005;35:141-153.
 Bentz EC. ACI Materials Journal, 2003;100(5):391-397.
 Sun W, Shi JJ, Jiang, JY. or.nsfc.gov.cn/bitstream/00001903-5/408622/1/1000009402869 2012.
 Cabrini M, Fontana F, Lorenzi S, Pastore T, Pellegrini, S. J of Chemistry 2015:2015, ID521507:10 pp.
6. Author Details
Dr. N Berke, currently at Tourney Consulting Group, LLC, has been actively involved in the durability of materials since receiving his corrosion related Ph.D. in 1980 in Metallurgical Engineering at the University of Illinois at Urbana-Champaign. He is Vice President of Research, which is primarily focused on corrosion and durability of construction materials and service life engineering.
Dr. Berke has received several awards over the years including the ACI Jean-Claude Roumain Innovation in Concrete Award, ASTM Frank E. Richart Award, and the ASTM Francis L. LaQue Award. He is a Fellow of ACI, ASTM, and NACE International.
M Hicks worked at Grace Construction Products primarily on the corrosion performance of concretes with calcium nitrite and other additives. She has since retired.