Ordinary Portland cement (OPC) is the main hydraulic construction material, the fabrication of which is an important source of carbon dioxide. Presently, alternative binders are attracting a lot of interest as a way to minimize their environmental impact. Calcium sulfoaluminate cements (CSAs) show massive potential for emissions reduction and possess unique properties. Nevertheless, there is an essential knowledge gap concerning the durability of CSA structures in terms of reinforcement corrosion. This work studies the corrosion behavior of plain carbon steel in media of fixed and variable composition that simulate the liquid phase of fresh CSA-based materials at different stages of binder hydration. The findings of electrochemical and nonelectrochemical techniques suggest that sulfate-induced corrosion of mild steel starts at early stages of hydration in pure CSA and in blended CSA-OPC binders. Once initiated, corrosion of reinforcement would continue regardless of favorable conditions found in the hydrated state in both systems, even though short-term repassivation was observed. In addition, the origin and possible solutions to tackle the corrosion problem are discussed.

Reinforced concrete is the main construction material in the world. It is relatively cheap, and it shows appropriate tensile strength and high compressive strength provided by steel reinforcement and concrete, respectively. Concrete itself is a composite made of aggregates embedded in a stone-like matrix called binder. The binder results from numerous chemical reactions of cement and is the crucial component that determines the mechanical, physical-chemical, and other service properties of reinforced concrete.1-3 

Today’s ever-increasing global annual production of ordinary Portland cement (OPC) exceeds 4 billion tons,1-,2,4-,5  being one of the largest contributors to global carbon dioxide emissions, estimated at ca. 6% to 8% annually.1,6  Therefore, cement producers are constantly seeking solutions to alleviate the environmental impact. Important strategies include the optimization of clinker production; improvement of grinding; introduction of alternative fuels and raw materials; and the use of blended cements, which are typically mixtures of OPC with supplementary cementitious materials, such as fly ash, blast furnace slag, or silica fume.1,4  Particularly, Portland clinker fraction in cement composition is reduced to 60% to 80% nowadays, compared to nearly 95% in 1980,1-,2,4,6  which decreased CO2 emissions by about 25%, compared to 1,000 kg of CO2, released per ton of OPC in the 1980s.1  As the production of blended cements can reduce CO2 emissions by 30% to 40% at most, alternative binders are considered.2,4  Such binders include alkali-activated materials7-9  in addition to a large family of nonhydraulic cements. However, binders possessing “chemistries” that are different from the conventional ones are not widely accepted so far, as they do not comply with the existing standards. To find compromises, research and development in the field of alternative binders have been focused on those that contain phases commonly found in Portland clinker (e.g., alite, belite, and celite), and their analogs, as well as combinations thereof. In this regard, high belite,1-,2,10-,11  calcium aluminate,1-,2  alinite,2,12,-13  and calcium sulfoaluminate cements (CSAs) deserve special attention. Providing unique characteristics that can be finely tuned for a specific application, and due to significant interest in emissions reduction, coupled with a lack of durability studies of steel-reinforced concrete, CSA is currently an important research topic.

Presently, around 1 million tons of CSA are produced worldwide annually.1,14  Various estimations suggest that only 220 kg to 600 kg of CO2 is released per ton of CSA.10,15-17  The principal components of CSA are tetracalcium trialuminate sulfate (3CaO·3Al2O3·CaSO4 or C4A3Ṡ)—10 wt% to 90 wt%, referred to as ye’elimite or Klein’s compound; belite (10 wt% to 60 wt%); and calcium sulfate (15 wt% to 30 wt%), being a part of CSA clinker or added later as anhydrite or gypsum, while other phases are found in trace quantities.1,-2  In a pure CSA binder, the main hydration products that provide mechanical properties are calcium trisulfoaluminate hydrate (ettringite, AFt) and hydrous alumina, in addition to calcium monosulfoaluminate hydrate (AFm), strätlingite, and hydrous ferric oxide in an insignificant amount.17-19  In blended CSA-OPC cements, calcium silica hydrate (C-S-H) is also a common hydration phase. Due to the presence of ettringite, CSA sare known foremost for their expansive behavior, determined by the ratio between ye’elimite and calcium sulfate.2,20,-21  In low sulfate formulations or in cements with relatively high aluminate content, the hydration of ye’elimite is fast, providing high early strength.20  Such concretes are dimensionally stable with no expansion evidenced,20  but they may experience lower sulfate resistance. On the opposite, when calcium sulfate is significant, ettringite continues to form at later stages after setting, and such binders exhibit expansion, used in shrinkage-compensated or self-stressing sulfate-resistant concretes.20  Shrinkage-compensating formulations are used when the prevention of shrinkage-induced cracking is critical (e.g., dams and piers, water tanks, etc.), while in self-stressing ones, larger expansion is used to increase stresses on the reinforcement, and, therefore, to produce thinner and stronger structures, as by mechanical prestressing.18,20,22  CSA with an adequate sulfate content is perfect for precast construction due to rapid hardening and high early strength.1,10,23  Exothermic hydration and high frost resistance of fresh CSA binders are attractive characteristics for cold weather construction.1  In addition, CSA binders are less prone to alkali-aggregate reactions due to lower alkali content, thus they are suitable for glass-fiber cement composites.1  However, lacking portlandite in the composition, they can be more susceptible to carbonation, which would have an adverse effect on the passivation of steel reinforcement, as it is proven for blended cements. Generally speaking, the long-term durability of CSA is not yet well documented.1 

The durability of concrete is paramount in sustainable construction practice because extending the life expectancy of structures is a very effective way to reduce their environmental impact. The durability of concrete, i.e., its resistance toward aggressive forces, is a collective measure, and the corrosion of steel reinforcement likely takes a central place. The corrosion resistance of steel reinforcement embedded in any type of concrete is defined mainly by two factors: (a) porosity and permeability and (b) chemical composition at every particular moment.3,24,-25 

Previously, the behavior of mild steel embedded in CSA- and OPC-based mortars was briefly studied by Kalogridis, et al.23  They reported 4, 5, and 7.5 times greater weight loss for steel in CSA mortars compared to OPC-based, when exposed to tap water, continuous and interrupted 3.5% NaCl, respectively.23  When tested in tap water, open-circuit potential (OCP) values as the early beginning were between –500 mVAg/AgCl and –400 mVAg/AgCl for steel embedded in CSA mortar, compared to –150 mV and 0 mV in OPC,23  suggesting that even in the absence of chlorides, carbon steel rebars embedded in CSA-based mortars suffered from corrosion, which likely is of an internal origin. On the other hand, porosimetry studies suggested that there should not be significant differences between the two cementitious systems in the long term.26  In particular, when tested in the identical conditions (w/c = 0.5, no additives) after 28 d of curing, the total porosity of CSA- and OPC-based pastes was estimated to be 52.5 mm3/g and 57.5 mm3/g, respectively,26  whereas at early hydration (12 h), CSA-based pastes demonstrated lower total porosity (65 mm3/g) compared to OPC (170 mm3/g).26  Moreover, other researchers observed lower levels of chloride ingress for CSA systems (1.8 to 2.2 times vs. OPC).23  These findings could only infer that CSA mortars and concrete should have better performance and higher durability, but they failed to explain the corrosion of carbon steel at early stages in CSA-based binders.

Therefore, the nature of hydrated CSA-binders seems to be the culprit for the observed corrosion of steel. Kalogridis, et al., proposed that higher susceptibility for reinforcement corrosion in CSA systems was provided by lower pH, compared to OPC mortars, estimated around 6,23  which did not agree with the findings of most other researchers, who observed pH of 11 to 13 for pure CSA18,27  and 13 to 13.6 for blended OPC-CSA16,28  materials. In our vision, sulfates are likely the cause of the observed corrosion of steel in fresh CSA binders as the cement contains a significant amount of sulfate-rich phases, estimated at 11.8 wt% to 22.5 wt% in terms of SO3 vs. 1.4 wt% to 3.7 wt% in OPC.1-,2,18,29-,30  The reported concentration of dissolved sulfates at the end of cement hydration is close for both CSA and OPC systems—2.5 mM (1 mM to 8 mM)18,27  and 2.3 mM (0.1 mM to 20 mM).29,31  However, at the beginning of hydration (0.5 h), traditional OPC-based grouts and mortars contain on average 5.5 times more dissolved sulfates than CSA analogs: 110 mM (9.7 mM to 199 mM)29,31  and 20 mM (17 mM to 31 mM),18,27  respectively. In addition, hydroxide concentration, observed at that stage, is 2.3 orders of magnitude lower for CSA compared to OPC cementitious systems, evidencing pH of 10.5 (10.1 to 11.3)18,27  and 12.8 (12.4 to 13.3).29,31  After 8 h of hydration, when the setting of CSA would have likely occurred, a pH of 12.5 (12.3 to 13.2) was reported,18,27  similar to the values for OPC.29,31  Unlike CSA-based cementitious systems, pH of OPC fresh mortars and grouts instantly rises to values of 12.3 to 12.7 instantly upon water addition, reaching 12.8 (12.4 to 13.3) after 30 min and attaining 13.6 (13.3 to 13.9) at the very end of hydration.29,31  Thus, regardless of higher values of dissolved sulfates, the corrosion of steel reinforcement is not reported in fresh OPC mortars and cements, unlike for CSA, as these higher values of dissolved sulfates observed are overbalanced by significantly higher values of hydroxide concentration (200 times). Blended CSA-OPC cementitious systems were reported to have higher pH values, when compared to CSA, determined by the presence of Portland clinker, but they were obviously lower than those observed in pure OPC binders. The concentration of dissolved sulfates, on the other hand, was reported to be higher than in both pure CSA and OPC binders, both in the beginning—213 mM (146 mM to 253 mM), and at the end—17 mM (10 mM to 35 mM) of hydration.28  In our opinion, it can be explained by the fact that in CSA-OPC cement sulfate-containing phases are deeper converted into soluble sulfates as they react with alkalis at higher pH, while in OPC these phases are in deficiency, and so they are practically completely dissolved. Concerning the subject of the study, the corrosion behavior of mild steel rebars has never been reported at all in mixed CSA-OPC systems, thus it needs elucidation.

Based on intensive analysis of the state-of-the-art, we assumed that the corrosion of mild steel in fresh CSA and CSA-OPC-based mortar and concrete in the initial stages of cement hydration is caused by a much higher sulfate-to-hydroxide ratio, when compared to OPC. It is believed that sulfate ions can adsorb on the steel surface, followed by the formation of soluble ferrous sulfates, causing thinning and subsequent destruction of passive oxide film, leading to the formation of stable corrosion pits in a similar manner to chloride ions.3,32-,33  It is also assumed that the deleterious action of sulfate ions is inversely proportional to the concentration of hydroxides in the pore solution, as they compete with sulfates to be adsorbed on the mild steel surface. As hydroxide ions stabilize passive oxide film, a critical ratio between sulfate and hydroxide ions, which separates corrosion and noncorrosion regions, should exist. The determination of such a critical ratio is a fundamental task of this research to confirm or disprove the assumptions made about the mechanism of steel corrosion in CSA-based mortars and concrete.

The purpose of this paper is to investigate the corrosion behavior of plain carbon steel reinforcement in simulating solutions of fixed composition that mimic the conditions encountered in pore solution at various stages of CSA and CSA-OPC cements hydration, by varying pH and sulfate concentration, in order to (a) understand the possible cause for carbon steel depassivation in CSA- and CSA-OPC-based binders and validating above proposed mechanism; and (b) define critical sulfate-to-hydroxide ratio, above which carbon steel is subjected to sulfate-induced corrosion. On the other hand, corrosion tests in simulating solutions of constant composition probably may not fully predict the behavior of steel rebars as hydration proceeds, accompanied by a gradual pH rise and decrease of dissolved sulfates. Therefore, additional dynamic corrosion testing of mild steel rebars immersed in variable simulating solutions, whose composition was altered in a controlled way to follow the evolution of key corrosion parameters (pH and dissolved sulfates) in the course of hydration of CSA and CSA-OPC binders, was performed to predict thoroughly the corrosion behavior of steel in CSA-based mortars and concrete. Both of these types of corrosion tests provide the comprehensive characterization of steel reinforcement behavior, which is required to propose a scientific solution that would allow avoiding or minimizing corrosion of plain steel, necessary for successful abundant application of CSA in the construction practice.

Low carbon steel CK45K (in wt%: C [0.35 to 0.45], Mn [0.6 to 0.9], Si [0.1 to 0.3], S [max. 0.05], P [max. 0.05], balance—Fe) was supplied by F. Ramada Aços e Indústrias, S.A. Rebars with a diameter of 1.2 cm diameter were cut perpendicularly to the rebar axis into pieces of 0.5±0.2 cm with a water-cooled circular sand saw, using 01014 resinoid cut-off wheel (Presi-Métallographie) and C180H cut-off machine (MetaServ), operating at 3,000 rpm. A copper wire was attached to the sample with silver paint SPI 5002 (Structure Probe Inc.) to ensure an electrical connection. Then the pieces were embedded into epoxy resin SP 106 (Gurit Ltd.), using acrylate tube of 2 mm wall thickness as a permanent mold, to form cylindrical samples of 2.0 cm in diameter. After complete epoxy hardening, the testing metal side was abraded and polished under water using silicon carbide papers up to 1000 grit size at 500 rpm (LaboPol-25, Struers Inc.), rinsed afterward with deionized water and ethanol, and dried with compressed air. To prevent crevice corrosion at the steel/epoxy interface and to define the exposed area (0.6 cm in diameter, 0.24 cm2), 3M™ Scotchrap™ 50 “all-weather corrosion protection” tape was applied over the polished samples. At least three identical samples were tested for each condition.

The first set of corrosion tests, called “static corrosion tests,” was performed in alkaline solutions of fixed composition containing sulfates. The samples were continuously immersed in 12 different aqueous solutions of 0.316 mM, 3.16 mM, and 31.6 mM NaOH (≥99%, 9356 Carl Roth) for, 12 h, 3 d, and 31 d, respectively, while sulfate concentration, provided by K2SO4 (≥99%, CN79 Carl Roth), was fixed at 0, 1 mM, 10 mM, and 200 mM to simulate the conditions encountered in the binders at various hydration stages of pure CSA or blended CSA-OPC cements.

The second set of corrosion tests, referred to as “dynamic corrosion tests” (DCTs), was in simulating solutions of variable composition. Though in reality, the changes in the course of binder hydration are continuous, the composition of the immersion solutions was altered in three discrete steps to cover the early, transitional, and late stages of hydration. In each step, the sample was removed from the previous immersion solution and placed into a new one immediately to avoid direct contact with air; and to prevent the change of subsequent solution composition, each following solution was refreshed three times after placing the sample. Three dynamic corrosion tests were proposed:
formula
formula
formula

The first dynamic test studies steel corrosion at constant sulfate concentration but changing pH, to understand if the corrosion rate would reduce upon alkalinization. Additionally, it mimicked the most adverse conditions reported for fully hydrated pure CSA. However, commonly dissolved sulfates are found at about 2.5 mM in a hydrated state, thus both pH and sulfate concentration were changed in the 2nd and 3rd dynamic corrosion tests, which differed from each other by the duration of each immersion step to reflect sluggish or rapid hydration, covering various CSA grades.

Autolab PGSTAT302N was used to perform OCP evolution monitoring and electrochemical impedance spectroscopy (EIS). Corrosion tests were performed at room temperature in a Faraday cage in a conventional three-electrode arrangement: a saturated calomel reference electrode, platinum wire as the counter electrode, and a mild steel sample as the working electrode. All potentials are recalculated to the standard hydrogen electrode (SHE). EIS spectra were obtained by applying single sinusoidal potential perturbations of 10 mVOCP (rms) amplitudes vs. OCP within the frequency range from 50 kHz to 5 mHz. The impedance plots were fitted using the ZView 3.5 a software using an equivalent circuit (Figure 1) composed of: Rs, solution resistance; CPEPF, passive film constant phase element; RPF, passive film resistance, which represents the resistance of solution inside pores of the oxide film (i.e., corrosion pits); CPEDL, electrical double layer constant phase element; and RF, faradaic resistance, which describes the ease of reduction-oxidation reactions involved in the corrosion process and provides the main contribution to polarization resistance, RP, inversely proportional to corrosion rate. Typically, values below 10 kΩ·cm2 are characteristic of active corrosion, while values above 1 MΩ·cm2 infer that a metal is passivated or immune to corrosion. Between 10 kΩ·cm2 and 1,000 kΩ·cm2 lies an intermediate region, in which localized corrosion is likely to occur. In addition, as found experimentally, the values above 50 MΩ·cm2 for an uncoated passivated metal cannot be precisely estimated by fitting (within the frequency range used in the experiment), thus being of a qualitative nature. Solution, passive film, and faradaic resistances are expressed per unit area in terms of Ω·cm2. To account for the inhomogeneity of the electrode surface and minor inseparable phenomena, CPEs were used instead of capacitors. They are described by the admittance/capacitance proportional pre-exponential parameter CPE-T, expressed per unit area as Ω1·sn·cm−2, equal to S·sn·cm−2 and F·sn–1·cm−2, where n is a dimensionless exponent factor, cited as CPE-P or α parameter. For an accurate estimation of corrosion parameters, CPE could have been recalculated to capacitance values,34  but we intentionally omitted that due to the limitations given by irreversible processes and to avoid overcomplication. The fitting of EIS data was done using the weighted linear least squares regression analysis method. Each data point weight was normalized by its magnitude, separately for real and imaginary parts, EIS spectra covered a wide range of impedances.35-41  In certain cases, only a fraction of the measured impedance data was analyzed, so the relative weighted sum of squares was used as a final criterion for the goodness of the fitting.
FIGURE 1.

Equivalent electrical circuit for EIS data fitting.

FIGURE 1.

Equivalent electrical circuit for EIS data fitting.

Close modal

The results of electrochemical studies were complemented with microscopy observations (Leica DMS 300, Achro 1.6 lens), in addition to scanning electron microscopy (SEM), obtained using a field emission gun (JEOL JSM-7001F) with energy dispersive x-ray spectroscopy (EDS) detector (Oxford INCA-250 HKL). EDS spectra were analyzed with respect to standard elements’ x-ray emission lines.42 

Corrosion Studies Performed in 0.316 mM NaOH Solutions of Fixed Composition

To elucidate the behavior of steel in pure CSA binders at early hydration stages, electrochemical tests were done in 0.316 mM NaOH solutions for 12 h of immersion. The photographs (Figure 2) of the exposed surfaces suggested that corrosion immediately started at pH 10.5 even in the presence of 1 mM of sulfates, while in the absence of sulfates steel remained protected. However, solutions of pH 10.5 are prone to pH drop due to carbonation, and the corrosion was observed even in the absence of sulfates at prolonged exposures. When the concentration of sulfates was low (i.e., 1 mM), carbon steel degradation occurred in the form of multiple pitting, which within 1 h evolved into a generalized form of corrosion, while in solutions containing 10 mM or 200 mM of K2SO4, corrosion followed a uniform pattern the early beginning. The elemental composition of the deposits formed on a mild steel surface and the corrosion products suspended in a 10 mM sulfate solution was determined by EDS. Based on the ratio between iron and oxygen, it can be inferred that the deposits on the steel surface were composed of Fe2O3 (40 at% of Fe) and Fe3O4 (43 at% of Fe), while the bulk rust was richer in oxygen, likely due to the presence of hydrated species such as FeOOH (33 at% of Fe). More importantly, EDS suggested that the corrosion products practically did not contain sulfur, likely suggesting the promoting role of sulfates in the corrosion process, rather than being a reactant.
FIGURE 2.

Optical observation of mild steel in 0.316 mM NaOH containing sulfates.

FIGURE 2.

Optical observation of mild steel in 0.316 mM NaOH containing sulfates.

Close modal

Carbon steel samples exposed to sulfate-free 0.316 mM NaOH solution showed a slow increase of OCP over time, but the values did not reach the typical for passive steel. In sulfate-containing solutions, initial OCP was 100 mV to 160 mV lower, and declined gradually, reaching values of between −420 mV and −460 mV at the end of the immersion period, suggesting severe corrosion of steel.

The corrosion behavior of mild steel was also studied by EIS (Figure 3), and important differences were noticed in the spectra in both the high- and low-frequency regions. In the high-frequency region, the response follows solution resistance, and the impedance modulus values decreased progressively with the addition of sulfates. In the middle-frequency range, the main contribution to EIS spectra comes from the oxide film, while at low frequencies, EIS response is dominated by electrical double layer and faradaic processes. Steel samples exposed to the blank solution evidenced high impedance values, being 2.5 to 3.5 orders of magnitude higher compared to those with sulfates. Fine differences between the samples were seen on Nyquist plots, which indicated increased corrosion activity with the increase of sulfate concentration. While steel samples immersed in sulfate-free solution showed increasing impedance values, accounting for strengthening protection, the low-frequency response for the samples tested in the presence of sulfates did not change notably with immersion time. The results of EIS fitting suggested faradaic resistance values, which are inversely proportional to the corrosion rate, increased for steel in sulfate-free solution until MΩ·cm2 level, indicating passivation of carbon steel, but remained nearly constant in the presence of 1 mM and 10 mM K2SO4, and decreased in 200 mM sulfate solution.
FIGURE 3.

EIS spectra of mild steel exposed to 0.316 mM NaOH solutions containing sulfates.

FIGURE 3.

EIS spectra of mild steel exposed to 0.316 mM NaOH solutions containing sulfates.

Close modal

Corrosion Studies Performed in 3.16 mM NaOH Solutions of Fixed Composition

Corrosion tests in alkaline 3.16 mM NaOH solutions (pH 11.5) are intended to simulate conditions in pure CSA between 2 h and 64 h of binder hydration and mixed CSA-OPC binders at early hydration stages.28  The visual inspection of the samples exposed to 3.16 mM NaOH solutions suggested that in the absence of sulfates, mild steel was not prone to corrosion. Contrarily, in solutions containing sulfates, corrosion has been evident since early times, progressively intensifying as the concentration of sulfates increased. Particularly, in 1 mM K2SO4, mild steel remained protected for almost 2 d, while at a concentration 10 times higher, corrosion was detected after several hours. In the 200 mM K2SO4 solution, corrosion of steel initiated almost instantly.

In sulfate-free and 1 mM K2SO4 3.16 mM NaOH, a gradual increase of OCP was observed. Both ones initially followed a similar trend, but in the presence of sulfates, corrosion of steel initiated, as confirmed by the sharp OCP drop from 66.5 mV to −331 mV between 52.4 h and 53.0 h of immersion, reaching −396 mV at the end of the test. Contrarily, samples exposed to 10 mM and 200 mM K2SO4 showed a gradual decrease in OCP values since the beginning and reached −471 mV and −501 mV at the end, respectively, as expected for corroding steel.

These results are corroborated by EIS spectra (Figure 4). Low-frequency impedance modulus values, |Z|5 mHz, proportional to Rp, increased over 2 d in 3.16 mM NaOH from 0.23 MΩ·cm2 to 1.36 MΩ·cm2. These values were higher than the ones (0.26 MΩ·cm2 to 1.27 MΩ·cm2) in the presence of 1 mM K2SO4. Nevertheless, for steel in 1 mM sulfate solution of pH 11.5, a sharp three-order drop to 1.9 kΩ·cm2 was observed after 53.5 h of immersion, reaching similar values found in 10 mM and 200 mM sulfate solutions. In the middle-frequency range, higher impedance values and phase angle values closer to −90° were observed for the more protected steel samples, while the samples exposed to higher content of sulfates showed both lower impedance and phase angle values. In the high-frequency region, the main contribution to EIS spectra came from the solution resistance and the addition of sulfates increased the conductivity of solutions. However, the tenfold increase in NaOH concentration attenuated that difference compared to solutions of pH 10.5. Then EIS spectra were fitted (Figure 5). Carbon steel was passivated in sulfate-free 3.16 mM NaOH, as evidenced by both RPF and RF values that exceeded MΩ·cm2 range. The samples tested in 1 mM K2SO4 followed a similar trend until localized oxide film rupture occurred and corrosion initiated after ca. 48 h of immersion, as reflected by the sharp drop in RPF and RF values by 4.8 and 3.7 orders of magnitude. In 10 mM and 200 mM K2SO4 corrosion started instantly. In particular, RPF and RF decreased as the concentration of sulfates increased, in the end being 178 Ω·cm2, 95 Ω·cm2, 15 Ω·cm2 (RPF) and 1,396 Ω·cm2, 1,215 Ω·cm2, and 838 Ω·cm2 (RF), respectively, for 1 mM, 10 mM, and 200 mM K2SO4.
FIGURE 4.

EIS spectra of mild steel exposed to 3.16 mM NaOH solutions containing sulfates.

FIGURE 4.

EIS spectra of mild steel exposed to 3.16 mM NaOH solutions containing sulfates.

Close modal
FIGURE 5.

The evolution of resistance values for mild steel in 3.16 mM NaOH solutions containing sulfates.

FIGURE 5.

The evolution of resistance values for mild steel in 3.16 mM NaOH solutions containing sulfates.

Close modal

Corrosion Studies Performed in 31.6 mM NaOH Solutions of Fixed Composition

Corrosion studies performed in 31.6 mM NaOH solutions containing sulfates (pH 12.5) aimed to simulate the interstitial pore environment in pure CSA binders in the end of cement hydration, as well as blended CSA-OPC-binders at the early and middle stages of hydration, respectively, for OPC-rich and OPC-poor grades. The appearance of samples immersed in 31.6 mM NaOH solutions suggested that steel easily tolerated small amounts of dissolved sulfates. In 1 mM K2SO4 it remained protected against corrosion till the end of immersion, and no visible differences were detected, similarly to a sulfate-free solution. When the concentration of sulfates was increased 10 times, there were no relevant corrosion effects in the beginning, but after 12 d corrosion was initiated. On the other hand, in 200 mM K2SO4 a corrosion spot was observed after an hour of immersion. For these two sets of samples, the corrosion started in the form of pitting but later evolved into uniform corrosion.

The OCP evolution, characterized by a gradual increase, revealed that steel was initially protected against corrosion in all 31.6 mM NaOH solutions. In 1 mM K2SO4 and sulfate-free 31.6 mM NaOH solutions steel remained passivated, and OCP reached 115 mVSHE within 2 d to 6 d, but in 10 mM and 200 mM K2SO4 corrosion started after 14 d to 20 d and 5 d to 10 d, respectively.

Bode phase angle plots indicated the existence of two overlapped time constants for all samples, similarly to solutions of pH 10.5 and 11.5. For the samples immersed in reference sulfate-free 31.6 mM NaOH, low-frequency impedance modulus values increased rapidly with time, reaching the MΩ·cm2 range within the first 24 h of immersion, and kept high throughout the testing period, thus indicating the passive condition of carbon steel. Middle- and high-frequency impedance values increased too due to oxide film development and slight carbonation of the solution, respectively. Similar behavior was observed for steel in 1 mM K2SO4 31.6 mM NaOH: the samples showed passive behavior throughout the testing period. The low-frequency impedance exceeded the MΩ·cm2 range, but the absolute values were slightly lower compared to the ones measured in the absence of sulfates. A very different behavior was observed in 10 mM and 200 mM sulfate solutions. In the presence of 10 mM K2SO4, in the beginning steel showed increasing values of low-frequency impedance, as observed for 1 mM and sulfate-free solutions. However, prolonged exposure induced corrosion, observed as the abrupt drop of |Z|5 mHz values after the 19th day. A similar trend was found for the impedance values in the middle-frequency range (1 Hz to 100 Hz). In 200 mM K2SO4, the loss of passivation occurred between the 5th and 7th days of immersion.

Key electrochemical parameters were extracted from the EIS data by fitting. The oxide film resistance increased gradually and exceeded 100 kΩ·cm2 for all samples after the 2nd day, suggesting that a protective film was formed on the steel surface in 31.6 mM NaOH (Figure 6). While for steel samples immersed in sulfate-free and 1 mM K2SO4 solutions the values further increased, but when the sulfate concentration exceeded 10 mM, localized rupture of the oxide layer occurred, forming corrosion pits, evidenced by the sharp drop in RPF values. Increased concentration of sulfates progressively decreased both time-to-corrosion onset and oxide film resistance. Particularly, the time-to-corrosion onset shortened from 19 d to 22 d to 5 d to 6 d, for 10 mM and 200 mM K2SO4, respectively; and RPF values at end of immersion in 0, 1 mM, 10 mM, and 200 mM K2SO4 were in average 9.4×106 Ω·cm2, 3.0×106 Ω·cm2, 30 Ω·cm2, and 3.5 Ω·cm2. For passivated steels, the estimation of RPF values was affected by error due to the marked capacitive response of the film, which largely overlapped with the low-frequency time constant.
FIGURE 6.

The evolution of resistance values for mild steel in 31.6 mM NaOH solutions containing sulfates.

FIGURE 6.

The evolution of resistance values for mild steel in 31.6 mM NaOH solutions containing sulfates.

Close modal

The capacitive response of the oxide film was also evaluated. Steel samples in the passive state showed stable low CPEPF-T values of ca. 20 µF·cm−2·sn–1. CPEPF-P values were also stable and approached 1, accounting for a near capacitive response. However, when corrosion activity intensified, the admittance of the CPEPF increased significantly, i.e. ca. 20 to 450 times. This was accompanied by a decrease in the CPEPF exponent. On the other hand, for steel exposed to sulfate-free and 1 mM sulfate 31.6 mM NaOH solutions, much lower CPEPF admittance values were observed until the end of the tests, as expected for steel in the passive state. However, the values of CPEPF-T for the reference blank sample at the end of immersion were slightly lower than for steel exposed to 1 mM K2SO4, suggesting that in the latter case oxide passive films contained more defects and eroded areas, thus being slightly less protective, compared to the ones formed in sulfate-free solution.

The faradaic resistance (Figure 6) showed a trend similar to RPF. At early stages of immersion for all samples, RF values increased rapidly from ca. 104 Ω·cm2 to 105 Ω·cm2, when the highest rate of passive film formation was expected, reaching stable RF values above 1 MΩ·cm2, when the initial formation of the passive film was completed. At this stage, the high RF, characteristic of passivated steels, corresponded to faradaic processes occurring at rates necessary to maintain uniformity of the passive film. In 10 mM K2SO4, RF, after reaching a maximum of 3.6 MΩ·cm2 on the 3rd day, decreased to stable values of 0.5 MΩ·cm2 to 1.2 MΩ·cm2, maintained for nearly 10 d, suggesting a higher passive corrosion rate, compared to the reference sample and steel in 1 mM K2SO4. After different periods of time and depending on the content of sulfate, passive film healing failed, followed by the formation of stable corrosion pits, evidenced by a sharp drop in RF values below 10 kΩ·cm2. At the end of the immersion, RF values were 8.6×108 Ω·cm2, 6.4×107 Ω·cm2, 2,100 Ω·cm2, and 980 Ω·cm2 (Figure 6), respectively, for the reference, 1 mM, 10 mM, and 200 mM K2SO4 solutions. Therefore, EIS fitting results suggested that mild steel was able to maintain the passive condition at pH 12.5, if the sulfate concentration was low, and did not exceed 1 mM. It is also inferred that the major negative impact of sulfates is on the integrity and protectiveness of the oxide film, but sulfates were not involved in corrosion as reactant, the increase in their concentration did not affect the corrosion rate substantially, as evidenced by close faradaic resistance values at the end of the immersion tests.

Similar conclusions were made concerning the evolution of the parameters of the electrical double-layer CPE. As corrosion initiated, the values of CPEDL-T increased proportionally to the area of exposed bare metal from 2 µF·cm−2·sn–1 to 20 µF·cm−2·sn–1, depending on the intensity of metastable pitting, to over 200 µF·cm−2·sn–1, finally reaching values as high as 1.26 mF·cm−2·sn–1 and 5.22 mF·cm−2·sn–1 on average at the end of corrosion tests for mild steel immersed in solutions of 10 mM and 200 mM K2SO4, respectively.

Though being not inherently related, metastable pitting was assessed by EIS at low frequencies: characteristic noise patterns were clearly observed in Lissajous curves (Figure 7) for all tested steel samples in the presence of sulfates before depassivation, linked with the dynamic process of rupture and consecutive repair of the passive film, while in the absence of sulfates, the noise was hardly seen. In particular, the intensity and frequency of noise peaks increased with sulfate concentration. The corrosion behavior in simulating sulfate solutions of pH 12.5 is concordant with the recent findings observed by electrochemical noise measurements reported in lime water (pH 12.6), in which 10 g/L Na2SO4 (7.0 mM) was sufficient for stable pitting of carbon steel.43 
FIGURE 7.

Noise in low-frequency Lissajous curves as an indication of metastable pitting.

FIGURE 7.

Noise in low-frequency Lissajous curves as an indication of metastable pitting.

Close modal

Dynamic Corrosion Tests

Nevertheless, the corrosion tests, performed in simulating solutions of a constant composition cannot fully predict the behavior of steel rebars in time as hydration proceeds, accompanied by pH rise and decrease of dissolved sulfates, and they lack to provide understanding of whether corrosion would continue, or repassivation would occur as conditions become progressively less aggressive in the course of CSA- and CSA-OPC-based binder hydration. Therefore, dynamic corrosion testing of mild steel rebars immersed in variable simulating solutions, whose composition was altered in a controlled way, aims to thoroughly predict the corrosion behavior of steel in CSA-based mortars and concrete.

OCP evolution in dynamic corrosion tests is shown in Figure 8 (only the last steps), plotted against the values observed in static corrosion tests exposed to identical solutions. the first two steps of dynamic tests were identical in terms of solution composition, but not of immersion duration, a similar evolution of electrochemical response was obviously observed, emphasizing the reliability of the obtained results. OCP evolutions in the first two steps of dynamic corrosion tests were similar to those observed in the static corrosion tests with some differences, such as a short-term increase of OCP at the beginning of each second step of the dynamic tests that was not observed in the static corrosion tests. On the other hand, major differences were observed in the third step of the dynamic corrosion test when compared to the static corrosion tests (Figure 8).
FIGURE 8.

OCP evolution of exposed mild steel in dynamic corrosion tests.

FIGURE 8.

OCP evolution of exposed mild steel in dynamic corrosion tests.

Close modal

In the first dynamic corrosion test, when exposed to 31.6 mM NaOH 10 mM K2SO4, freshly polished steel showed increasing OCP that reached as high as +254 mV, indicating the formation of the passive film protective for 19 d, but ruptured then, evidenced by the sharp decrease in OCP. However, in the dynamic corrosion test, steel that had previously suffered corrosion damage lost passivity much faster—just after 8.8 h of immersion, and the highest OCP value of just +43 mV was observed (Figure 8). In 31.6 mM NaOH 1 mM K2SO4 the difference in the behavior of pristine and precorroded steel was even more apparent. While freshly polished mild steel was protected throughout the testing period and no signs of corrosion were observed, steel samples with a corrosion “history” continued to degrade after a short-term increase in OCP. In particular, a sharp drop in OCP values to −330 mV occurred in the third step of the second dynamic corrosion test between the 2nd and 4th days, and an intensive fluctuation of OCP, related to the formation and healing of the corrosion pits, was observed until the end of the testing. In the third dynamic corrosion test, obviously, steel faced less damage due to the shorter duration of the first two steps, compared to the second one, and OCP of +250 mV to +350 mV was observed for 21 d in the third stage of testing. Nevertheless, by the 26th day of immersion in 1 mM sulfate solution of pH 12.5, OCP dropped to −320 mVSHE to −340 mVSHE, asserting corrosion re-initiation (Figure 8).

These findings were supported by impedance data (Figure 9). Based on the evolution of faradaic resistance, it is clear that all steel samples corroded in 10 mM K2SO4 solutions of pH 10.5 and 11.5, evidenced by low RF values estimated between 937 Ω·cm2 and 1,394 Ω·cm2, stable in time, except for a short increase to 1.7 kΩ·cm2, when 0.316 mM NaOH solution was replaced with 3.16 mM NaOH solutions. When pH was increased to 12.5, faradaic resistance rose sharply and mild steel samples entered the intermediate region (104 Ω·cm2 to 106 Ω·cm2) between the active corrosion region and the passivation region. Particularly, RF was 28.9 kΩ·cm2, 22.5 kΩ·cm2, and 43.5 kΩ·cm2 after 2 h of immersion in 31.6 mM NaOH for the 1st, 2nd, and 3rd dynamic corrosion tests, respectively. After that, RF continued to increase until corrosion re-initiated, evidenced by the drop in faradaic resistance values. Such decrease was observed between 9.8 h and 13.2 h in 10 mM K2SO4, between 2.0 d and 4.3 d in 1 mM K2SO4 in the 2nd dynamic test, and between 21 d and 26 d in the 3rd test. Furthermore, the third test was the only one in which the passivation region in fact was reached between 5th and 21th days of immersion, with RF of 1.4 MΩ·cm2 to 1.8 MΩ·cm2. At the end of immersion, the samples in all corrosion tests eventually approached a similar low RF (Figure 9), close to the values seen in previous steps of dynamic corrosion tests. This may again infer that neither sulfates nor hydroxides were directly involved in the corrosion reactions, and the corrosion process was likely limited by oxygen reduction, the availability of which should have been equal in all tests, solutions were in contact with air and an identical electrochemical cell configuration was used.
FIGURE 9.

The evolution of faradaic resistance in dynamic corrosion tests.

FIGURE 9.

The evolution of faradaic resistance in dynamic corrosion tests.

Close modal

The electrical double-layer constant phase element, CPEDL, is another corrosion indicator. In the absence of diffusion and for a perfectly smooth planar surface, CPEDL-P should approach 1, i.e., the behavior of a capacitor. However, actual CPEDL-P values were lower and decreased progressively with an increase in surface irregularity and strengthening of diffusion contribution, and at long exposures as steel was actively corroding, CPEDL-P values decreased. The CPEDL-T value, proportional to the total capacitance of the double layer, estimates the area of exposed bare metal surface, i.e., uncovered by passive oxide film. The evolution of CPEDL-T can serve as a sign for corrosion (re-)initiation, and the values help to assess corrosion severity. In 0.316 mM and 3.16 mM NaOH solutions, steel samples showed high CPEDL-T values, suggesting that large areas of steel were not protected. After placing mild steel samples to 31.6 mM NaOH, CPEDL-T sharply decreased 9 to 30 times, and the values continued to decline as the exposed area reduced due to the healing of corrosion pits, reaching minimum values of 23.7 µF·cm−2·sn–1, 60.3 µF·cm−2·sn–1, and 22.9 µF·cm−2·sn–1, respectively, for 1st, 2nd, and 3rd tests after 17 h, 39 h, and 116 h. Closer values of CPEDL-T before corrosion re-initiation were observed in the 1st and 3rd corrosion tests, compared to higher values in the second test, indicating that the duration of immersion in an aggressive environment, rather than the composition, i.e., the degree of its aggressiveness, was more important for the subsequent corrosion behavior of mild steel reinforcement in a passivating medium. Later, corrosion re-initiation led to an abrupt rise of CPEDL-T values, ultimately reaching 2.31 mF·cm−2·sn–1, 6.07 mF·cm−2·sn–1, and 8.02 mF·cm−2·sn–1, respectively, for 1st, 2nd, and 3rd dynamic tests.

The overall findings of dynamic corrosion tests suggested that the sulfate-induced corrosion of mild steel, once initiated, would proceed in favorable conditions, which have been proven by previous static corrosion tests, regardless of short-term repassivation. Thus, sulfate ions trigger the corrosion of steel reinforcement, but they are not required for corrosion to continue, being similar in action to chloride ions, as in particular studied with regard to electrochemical chloride removal as a rehabilitation method.44 

The overall results of the paper emphasized the destructive role of sulfates and the constructive role of hydroxide ions in the formation and stabilization of passive oxide film, and ultimately in the corrosion protection of carbon steel. The findings suggested that to ensure persistent passivation, the concentration of dissolved sulfates at any moment should not reach 10 mM at pH 12.5, and be lower than 1 mM when the pH is below 11.5.

The outcomes of corrosion tests for carbon steel in the presence of sulfate ions as a function of pH were combined in the form of a schematic sulfate—pH corrosion diagram (Figure 10). The diagram is divided into four corrosion domains, separated by lines 1 to 3. Line 1 is the experimentally determined boundary between sulfate—pH combinations where carbon steel corrosion is initiated instantly (domain A), and the one where it is delayed progressively with an increase in pH and/or decrease in sulfate concentration (domain B). This zone eventually passes into the region (domain C) where corrosion onset is so delayed that it is unlikely to be observed within a realistic timeframe (i.e., the region of improbable corrosion), followed by the passive domain D, as reflected by empirical lines 2 and 3 that define critical sulfate threshold levels of [SO42–] = [OH]/6 and [SO42–] = [OH]/30, accounting for optimistic and pessimistic corrosion scenarios, respectively.
FIGURE 10.

Sulfate-pH corrosion diagram of mild steel in relation to CSA-based binders.

FIGURE 10.

Sulfate-pH corrosion diagram of mild steel in relation to CSA-based binders.

Close modal

To explain the behavior of carbon steel in real structures, previously reported composition ranges for liquid phases of pure CSA,18,27  OPC,29,31  and blended CSA-OPC (30:70 wt%)28  binders after <2 h (respectively, regions E, G, and I) and >7 d (F, H, and J) of cement hydration were mapped on the sulfate-pH diagram (Figure 10). This suggests that the sulfate-induced corrosion of common mild carbon steel is very likely to start instantly in pure CSA (E) and most mixed CSA-OPC (I) binders at early stages of cement hydration, in both cases exceeding the thresholds. Moreover, as confirmed by the results of dynamic corrosion tests (reflected by arrows, Figure 10), sulfate-induced corrosion, once initiated, would progress, despite a short period of re-passivation observed, although the conditions found in a completely hydrated state for both cementitious systems, respectively, regions F and J, are favorable for passivation of carbon steel.

For further insight, the sulfate-pH diagram in Figure 10 can be divided into two regions by the composite line 6. Based on simple thermodynamic calculations, line 6 defines the highest concentration of dissolved sulfates achievable at a given pH in the liquid phase of a binder that contains gypsum (СaSO4·2H2O) as the most soluble sulfate phase. In particular, when pH is low, the concentration of dissolved sulfates is constant at 5.6 mM, defined by the solubility product of gypsum (3.14×10−5),45  the excess of which is in equilibrium with the aqueous phase—a condition corresponding to horizontal line 4. However, above a pH of 12.2, alkali hydroxides convert gypsum into calcium hydroxide, accompanied by the release of soluble sulfates. The concentration of sulfate ions increases proportionally with the concentration of hydroxide ions in the liquid phase, in accordance with the equilibrium between gypsum and calcium hydroxide (KSP(Ca(OH)2) = 5.5×10−6),45  corresponding to inclined line 5. Thus, the region below and to the right of line 6 denotes the conditions in which gypsum should be completely dissolved or transformed, and line 6 indicates the highest thermodynamically possible concentration of sulfates in the liquid phase of a binder in equilibrium with an excess of solid gypsum. The region above line 6 marks sulfate concentrations that could not be reached unless some other more soluble sulfate phase than gypsum is present.

The diagram clearly indicates that the reported composition domains of freshly mixed and hydrated OPC,29,31  as well as both hydrated pure CSA18,27  and mixed CSA-OPC28  binders are located at or below line 6. This is explained by the fact that sulfate ions can only be provided by gypsum or less soluble calcium trisulfoaluminate and monosulfoaluminate hydrate phases, and that there are no other more soluble sulfate phases in the composition of these binders, in accordance with ubiquitous literature data.17-19,29,-30  In contrast, the composition ranges of the liquid phase of freshly prepared pure CSA and mixed CSA-OPC binders are located above line 6, suggesting that such a high concentration of sulfates must have been provided by another sulfate phase, more soluble than gypsum. It is safe to assume that this phase is ye’elimite, the main component of CSA cement.1-,2  Therefore, this inference suggests that sulfate-induced corrosion of ordinary carbon steel originates from the elevated solubility of ye’elimite, being thereby a characteristic problem of all CSA-containing binders.

This sulfate-induced corrosion of mild steel, which should start in the early stages of binder hydration, would significantly reduce the service life of CSA- and CSA-OPC-based reinforced concrete structures. Therefore, for the successful application of CSA in construction, it is essential to eliminate this inherent source of corrosion and ensure that CSA reinforced concrete structures have a corrosion resistance comparable to that of OPC structures, where the sources of corrosion, such as chloride ingress, are only of external origin. Various approaches may be tentatively proposed to eliminate intrinsic sulfate-induced corrosion in fresh CSA-based concrete: (a) pH of the fresh concrete mix can be increased by alkaline additives or cement with high alkali content; (b) cathodic protection can be used; (c) corrosion inhibitors can be added to the fresh concrete mix; (d) conversion or organic barrier coatings can be applied on the surface of mild steel rebars; and (e) carbon steel can be replaced with stainless or galvanized steels or nonmetallic materials. While each of these approaches deserves dedicated consideration and research, at least the first one—the addition of alkalis—can be evaluated based on Figure 10. There, line 6 reflects the overall efficiency of alkaline additives: the corrosion of steel can be delayed if pH is increased above 11.5. However, any further increase above 12.2 should not bring any additional improvement, at that point gypsum gets converted into calcium hydroxide, accompanied by the equivalent release of sulfate ions, so that above that point their concentration and pH should increase simultaneously with no net beneficial effect. The presence of more soluble ye’elimite is expected to make corrosion control even more difficult. Solubility data of ye’elimite is not readily available, but it is expected that line 4 should shift to higher values of sulfate ions concentration, while line 5 should shift to lower pH values, and the resulting theoretical line 7 in Figure 10 would lie predominantly or entirely in the region of instant corrosion. Thus, alkali addition would make no improvement at all, regardless of the composition of any given freshly mixed CSA binder. Similarly, blended OPC-CSA formulations, which were proposed to solve the corrosion problem, given their higher alkali content and hence higher initial pH, indeed evidenced elevated values of sulfates (domain I, Figure 10), consistent with this prediction. Furthermore, the low susceptibility of CSA binder toward alkali-aggregate reactions, provided by its intrinsically lower pH, would be lost upon the addition of alkaline species. In summary, the increase in pH with alkaline additives or OPC blending would likely fail to prevent steel corrosion, so corrosion control approaches such as those listed above in (b) through (e) merit renewed attention.

  • Sulfate ions are detrimental to the passive oxide film that protects steel from corrosion, while hydroxides stabilize and repair it, providing increasing corrosion protection as pH increases.

  • In all simulating solutions in which corrosion of steel occurred, the ultimate corrosion rate did not depend on the composition of these solutions, and corrosion products did not contain sulfates, both suggesting that sulfate ions were not directly involved in corrosion reactions as reactants, thus they acted as corrosion activators.

  • At low concentrations, sulfates caused passive film thinning and induced metastable pitting, but when the concentration exceeded a certain critical level for a given pH, stable corrosion pits were formed and evolved, thereby initiating the corrosion of carbon steel reinforcement;

  • To ensure persistent passivation of mild steel, the concentration of hydroxide ions should exceed that of sulfates 6 to 30 times as in optimistic and pessimistic scenarios, respectively, which is a guide for the critical sulfate concentration for pH range of 10.5 to 12.5, and these thresholds can be respectively expressed as: [OH] ≥ 6·[SO42–] or [OH] ≥ 30·[SO42–], in M.

  • The results of the tests performed in the solutions of variable composition implied that a preceding corrosion event significantly undermined the subsequent resistance of steel to aggressive species.

  • Once initiated, sulfate-induced corrosion of mild steel progressed even in conditions favorable for passivation, as found in the solutions of fixed composition, despite a short period of corrosion rate reduction observed.

  • Corrosion damage that took place in aggressive solutions could not be fully healed in passivating media.

  • The extent of corrosion damage was proportional both to the duration of steel exposure to an adverse environment and its aggressiveness (directly to sulfate concentration and inversely to pH).

  • Considering the composition of binders reported elsewhere, sulfate-induced corrosion of mild steel is expected to start instantly in pure CSA and most blended CSA-OPC binders at the early stages of cement hydration, and would likely continue even when hydration reactions come to an end, though the conditions found in both fully hydrated cementitious systems are favorable for passivation.

  • It was tentatively concluded that sulfate-induced corrosion of ordinary plain carbon steel originated from the elevated solubility of ye’elimite, the main phase of CSA cement, therefore being a characteristic problem of all CSA-containing binders.

  • The addition of alkaline additives or OPC into CSA cement with the intention of increasing pH of the liquid phase at early hydration stages should bring no improvement to the corrosion state of carbon steel, the increase in pH is expected to cause an increase in dissolved sulfates concentration due to dissolution/transformation of ye’elimite.

  • In our future research, we aim to eliminate sulfate-induced corrosion at the early stages of hydration, thereby ensuring compatibility of CSA binders with carbon steel reinforcement as the requirement for successful application of CSA in construction practice, with the ultimate incentive for global carbon dioxide emission reduction.

Trade name.

The authors thank Filipovich Darya for her invaluable help in organizing raw and fitted EIS results into graphical form; acknowledge Fundação para a Ciência e Tecnologia (Portugal) for funding Ph.D. grant SFRH/BD/88016/2012 and Centro de Química Estructural (UID/QUI/00100/2013, UID/QUI/00100/2019, UIDB/00100/2020, UIDP/00100/2020, and LA/P/0056/2020); ITALY PROJECT of University of Bergamo and COST (CA15202) for financial support; Prof. António Castela (Instituto Politécnico de Setúbal) for providing carbon steel for the tests; Dr. Diógenes Carbonell Boix (Universidad Carlos III de Madrid) and Adrien Hoock (3M) for sending free samples of Scotchrap™ 50 “all-weather corrosion protection” tape.

1.
Aïtcin
P.C.
,
Binders for Durable and Sustainable Concrete
(
Oxon, United Kingdom
:
Taylor & Francis
,
2008
).
2.
Barnes
P.
,
Bensted
J.
,
Structure and Performance of Cements
, 2nd ed. (
Oxon, United Kingdom
:
Taylor & Francis
,
2002
).
3.
Bertolini
L.
,
Elsener
B.
,
Pedeferri
P.
,
Corrosion of Steel in Concrete: Prevention, Diagnosis, Repair
(
Hoboken, NJ: Wiley
,
2004
).
4.
Schneider
M.
,
Romer
M.
,
Tschudin
M.
,
Bolio
H.
,
Cem. Concr. Res.
41
(
2011
):
p
.
642
650
.
5.
van Oss
H.G.
, “
Background Facts and Issues Concerning Cement and Cement Data
,”
U.S. Department of the Interior, Open-File Report 2005-1152
,
2005
.
6.
Olivier
J.G.J.
,
Janssens-Maenhout
G.
,
Muntean
M.
,
Peters
J.A.H.W.
,
“Trends in Global CO2 Emissions: 2015 Report,” Institute for Environment and Sustainability of the European Commission’s Joint Research Centre
,
report no. 1803
,
2015
.
7.
Glukhovsky
V.D.
,
Gruntosilicates
(
Kiev, UkSSR
:
State Publishing House of Literature on Construction and Architecture of the Ukrainian SSR
,
1959
),
in Russian
.
8.
Provis
J.
,
van Deventer
J.
,
“Alkali Activated Materials: State-of-the-Art Report,” RILEM, TC 224-AAM
,
2013
.
9.
Turner
L.K.
,
Collins
F.G.
,
Constr. Build. Mater.
43
(
2013
):
p
.
125
130
.
10.
Quillin
K.
,
Cem. Concr. Res.
31
(
2001
):
p
.
1341
1349
.
11.
Santos
R.L.
,
Horta
R.B.
,
Pereira
J.
,
Nunes
T.G.
,
Rocha
P.
,
Lopes
J.N.C.
,
Colaço
R.
,
Cem. Concr. Res.
76
(
2015
):
p
.
212
221
.
12.
Ilyukhin
V.V.
,
Nevsky
N.N.
,
Bickbau
M.J.
,
Howie
R.A.
,
Nature
269
(
1977
):
p
.
397
398
.
13.
Bikbau
M. Ya.
,
Nudelman
B.I.
,
Alinite Cement
(
Moscow, USSR
:
Stroyizdat
,
1989
),
in Russian
.
14.
Glasser
F.P.
,
Zhang
L.
,
Cem. Concr. Res.
31
(
2001
):
p
.
1881
1886
.
15.
Arjunan
P.
,
Silsbee
M.R.
,
Della
M.R.
,
Cem. Concr. Res.
29
(
1999
):
p
.
1305
1311
.
16.
Ioannou
S.
,
Paine
K.
,
Reig
L.
,
Quillin
K.
,
Cem. Concr. Compos.
55
(
2015
):
p
.
196
204
.
17.
Liao
Y.S.
,
Wei
X.S.
,
Li
G.W.
,
Constr. Build. Mater.
25
(
2011
):
p
.
1572
1579
.
18.
Winnefeld
F.
,
Lothenbach
B.
,
Cem. Concr. Res.
40
(
2010
):
p
.
1239
1247
.
19.
Allevi
S.
,
Marchi
M.
,
Scotti
F.
,
Bertini
S.
,
Cosentino
C.
,
Mater. Struct.
49
(
2016
):
p
.
453
466
.
20.
Winnefeld
F.
,
Barlag
S.
,
J. Therm. Anal. Calorim.
101
(
2010
):
p
.
949
957
.
21.
Mehta
P.K.
,
Cem. Concr. Res.
3
(
1973
):
p
.
1
6
.
22.
Zhang
L.
,
Glasser
F.P.
,
Adv. Cem. Res.
14
(
2002
):
p
.
141
155
.
23.
Kalogridis
D.
,
Kostogloudis
G.C.
,
Ftikos
C.
,
Malami
C.
,
Cem. Concr. Res.
30
(
2000
):
p
.
1731
1740
.
24.
Page
C.L.
,
Treadaway
K.W.J.
,
Nature
297
(
1982
):
p
.
109
115
.
25.
Broomfield
J.P.
,
Corrosion of Steel in Concrete: Understanding, Investigation, and Repair
(
New York, NY: E & FN Spon
,
1997
).
26.
Bernardo
G.
,
Telesca
A.
,
Valenti
G.L.
,
Cem. Concr. Res.
36
(
2006
):
p
.
1042
1047
.
27.
Andac
M.
,
Glasser
F.P.
,
Adv. Cem. Res.
11
(
1999
):
p
.
23
26
.
28.
Pelletier-Chaignat
L.
,
Winnefeld
F.
,
Lothenbach
B.
,
Saout
G.L.
,
Müller
C.J.
,
Famy
C.
,
Cem. Concr. Compos.
33
(
2011
):
p
.
551
561
.
29.
Lothenbach
B.
,
Winnefeld
F.
,
Cem. Concr. Res.
36
(
2006
):
p
.
209
226
.
30.
Lothenbach
B.
,
Winnefeld
F.
,
Alder
C.
,
Wieland
E.
,
Lunk
P.
,
Cem. Concr. Res.
37
(
2007
):
p
.
483
491
.
31.
Vollpracht
A.
,
Lothenbach
B.
,
Snellings
R.
,
Haufe
J.
,
Mater. Struct.
49
(
2016
):
p
.
3341
3367
.
32.
Marcus
P.
,
Maurice
V.
,
Strehblow
H.-H.
,
Corros. Sci.
50
(
2008
):
p
.
2698
2704
.
33.
Dawson
J.L.
,
Ferreira
M.G.S.
,
Corros. Sci.
26
(
1986
):
p
.
1009
1026
.
34.
Brug
G.J.
,
van den Eeden
A.L.G.
,
Sluyters-Rehbach
M.
,
Sluyters
J.H.
,
J. Electroanal. Chem. Interf. Electrochem.
176
(
1984
):
p
.
275
295
.
35.
Armstrong
R.D.
,
Edmondson
K.
,
Electrochim. Acta
18
(
1973
):
p
.
937
943
.
36.
Hladky
K.
,
Callow
L.M.
,
Dawson
J.L.
,
Brit. Corros. J.
15
(
1980
):
p
.
20
25
.
37.
Lay
P.
,
Lawrence
P.F.
,
Wilkins
N.J.M.
,
Williams
D.E.
,
J. Appl. Electrochem.
15
(
1985
):
p
.
755
766
.
38.
Spinolo
G.
,
Chiodelli
G.
,
Magistris
A.
,
Tamburini
U.A.
,
J. Electrochem. Soc.
135
(
1988
):
p
.
1419
1424
.
39.
Macdonald
D.D.
,
Mckubre
M.C.H.
,
Urquidi-Macdonald
M.
,
Corrosion
44
(
1988
):
p
.
2
7
.
40.
Barsoukov
E.
,
Macdonald
J.R.
,
Spectroscopy
Impedance
.
Theory, Experiment, and Applications
(
Hoboken, NJ: John Wiley & Sons, Inc.
,
2005
).
41.
Morozov
Y.
,
Castela
A.S.
,
Dias
A.P.S.
,
Montemor
M.F.
,
Cem. Concr. Res.
47
(
2013
):
p
.
1
7
.
42.
Bearden
J.A.
,
Rev. Mod. Phys.
39
(
1967
):
p
.
78
124
.
43.
Permeh
S.
,
Lau
K.
,
Duncan
M.
,
Simmons
R.
,
Mater. Corros.
72
(
2021
):
p
.
1456
1467
.
44.
Miranda
J.M.
,
Cobo
A.
,
Otero
E.
,
González
J.A.
,
Cem. Concr. Res.
37
(
2007
):
p
.
596
603
.
45.
Dean
J.A.
,
Lange
N.A.
,
Lange’s Handbook of Chemistry
(
New York, NY
:
McGraw-Hill
,
1992
).