Carbon steel electrodes within chloride induced reinforcement corrosion in cracked concrete: Approximation to deterministic system response

Research on chloride‐induced corrosion in cracked, reinforced concrete is of steady economic and scientific importance. The ever‐growing amount of reinforced concrete structures in need for maintenance and preservation calls for suitable and reliable methods to assess the current state. The present study demonstrates a specimen design that allows authentic corrosion monitoring, including solely carbon mild steel as working and counter electrodes, singularly cracked concrete as electrolytes and a conservative dismantling of the anodic steel‐concrete interface (SCI) at the end of 180 days of monitoring. The electrochemical features of the evolved specimen narrow down to the textbook behavior of the investigated electrodes according to the electro potential series of metals. The most prominent results are the chloride profiles at depth of rebar along the SCI, the build‐up of driving potential, and the decisive contribution arising from the anodic equilibrium potential. Accordingly, future repair methods should always include observation of the anodic equilibrium potential and their effectiveness is clearly addressed in the course of that value.

concrete has been of steadily increasing economic and scientific importance. [4]Due to the latent progress of damaging effects during chloride-induced rebar corrosion and its catastrophic consequences for buildings and infrastructures, all possible scenarios have to be investigated parametrically.Additionally, the complexity of constructively conditioned cracks in reinforced concrete, which do shorten the time of corrosion initiation and thereby accelerate the chloride-induced loss of rebar cross section, must be taken into account. [5]Cracks display an accelerated pathway for those deleterious chlorides and their features have been discussed appropriately: crack width, [6,7] crack depth, [8] micro cracks, [9] numerical modeling of crack propagation, [10] numerical modeling of corrosion initiation, [11,12] holistic approaches onto corrosion in cracked concrete, [13,14] and reviews. [15]s many monitoring methods and concepts have been developed abundantly and numerous studies on mechanisms of chloride-induced corrosion were performed, a clear tendency of critical parameters is still highly debated.Complex problems arising from investigating corrosion activity in cracked concrete are numerous but plenty of discussions have already built up a solid base of crucial parameters.
The first and probably most important in terms of corrosion determinant is the steel-concrete interface (SCI). [16,17]Without hierarchical judgment the prominence of the SCI is followed by the polarization resistance on the anodic site, macrocell, [18,19] and microcell corrosion processes, [20,21] electrolytic resistance of concrete [22,23] and effects arising in long-term corrosion monitoring. [1,24,25]or some reason, little has been reported about the formation and evolution of driving potential ΔE, [3] although it builds up a major part in macrocell corrosion modeling (see Section 2).
It can be stated that the processes on and along the SCI are of crucial importance. [16,17]Because of that, the present study demonstrates a specimen design which allows authentic corrosion monitoring concepts including solely steel as working and counter electrodes, singularly cracked concrete as electrolyte and finally a conservative uncovering of the SCI in addition with chloride profiling along that specific area.Furthermore, the present study is part of a major research project to evaluate the effectiveness of the patch repair "increasing the electrical resistivity by coating" (MR 8.3), but within this report, we stick to the evolution of the specimen setup, built up of driving potential and correlation with chloride profiling.Chloride in chloride-induced corrosion is obviously the predominant factor in damage initiation and damage assessment.28] Thereby, the concentrations are mostly given in mass percent per mass of cement (M.-%/c.)and mandatorily determined closely below the depth of rebar.To give a few key facts: According to Schnell and Raupach the critical chloride content c crit to enable corrosion ranges at 0.5 M.-%/c., [29] many of the real concrete structures and representative, scientific specimens range at 1.0 M.-%/c.after a chloride exposure equivalent to one winter period. [1,2]Additionally, it is clearly stated by Wunderle et al. that conservative repair principles like application of a surface protection system will most probably fail if c Cl is above 1.5 M.-%/c and monitoring is imperative in any case. [1]Within the framework of the present study, precisely this borderline case is simulated and the underlying processes are uncovered.

| ELECTRO/-CHEMICAL PREREQUISITES FOR REBAR CORROSION
To gain all possible electrochemical processes at the surface of the reinforcement, a three-electrode-setup is mandatory.Next to the anodic and cathodic conceived rebar, working as corrosion sensors, a MnO 2 reference electrode, referenced to Ag/AgCl, completes the present three-electrode-setup.From then on, the prominent model of macrocell corrosion, introduced by Schwenk, [30] is applicable: Whereby I macro is the macrocell corrosion current measured in [µA], ΔE is the driving potential measured in [mV], E 0 is the resting potential measured in [mV], R P is the polarization resistance measured in [Ω] and R E is the electrolytic resistance of concrete measured in [Ω].Unfortunately, the terms for electrode potential are used contrary to each other in the countless literature references.For the sake of clarity, we assume the "corrosion potential" to be the value of the whole, short-circuited reinforcement against the reference electrode.This value is comparable to potential values from half-cell potential mapping, [31][32][33][34][35] with the difference of measuring the half-cell potential within the concrete and not on its surface.The term "resting potential" is used synonymously to "equilibrium potential" and "fully depolarized open circuit potential," whereby the term "free corrosion potential" describes the transient value of potential between closed circuit and resting potential of each decoupled electrode-like anode or cathode.The indices A and C describe the affiliation of each parameter to the different electrodes anode and cathode.The theoretical value of each depolarized electrode, hence the "equilibrium potential" arises from the Nernst equation.However, as we will not determine the concentrations of oxidized iron ions Fe 2+ and reduced water ions OH − , we stick to the empirical measure.
] we dispensed on further upgrades of the model given in Equation (1) for the time being.

| Methodological approach
This study is the result of several attempts to successfully reproduce the natural formation of corrosion in cracked reinforced concrete according to corrosion parameters of real reinforced concrete structures as shown in the work of Wunderle et al. [1] Natural formation is supposed to emphasize the approach of corrosion initiation via chloride solution impingement onto the cracked surface at suitable chloride solution concentrations (1.5% NaCl [3] ) and without any further potential or galvanic polarization.The specimen design is based on the works of Hiemer et al. [3] and Kessler et al., [2] who focused on the influence of bending cracks and is now adopted on transverse cracks with the major advantage of crack proof and conservative SCI-exposure at the end of the investigation period.

| Specimen setup
Figure 1 depicts the evolution of the herein presented specimen setup over the last few years and its outlook.Additionally, the concretization as well as instrumentation features, material properties, and experimental timeline are shown.
Following the previous studies [2,3] the concrete casting, the measurement routine and exposure conditions were analogously carried out on a set of 10 transversely cracked specimens (version I in Figure 1).The thereby applied working principle is mostly valid for both versions I & II and is summarized in the following.Major differences within the versions are depicted apparently but we start with the explanation of version I: The anodic conceived rebar with 10 mm in diameter of the type B500B, manufactured according to Tempcore® procedure [38] is positioned closely below the glass fibers (Schöck-Combar®), centered in the specimen and has a concrete cover of 40 mm respectively.The total length of the anode is 200 mm, its marginal regions are covered by highly alkaline mortar and a shrinkage tube on top of the mortar to reduce the possibility of cathodic processes on the anodic conceived steel.Both electrodes, anodic bars, and cathodic baskets are electrically contacted via soldering eyes.The static structure of version I consists of four glass fiber rebars, which are symmetrically oriented in the c-axis of the specimen.That glass fiber rebar was supposed to ensure crack confinement without any electrochemical contribution to the formation of a galvanic cell.
Material categories, concrete casting according to DIN EN 12350, part 5, [39] part 6, [40] and part 7 [41] as well as DIN 1045-2 [42] 28 days of posttreatment, and permanent "dry" storage at 20°C and 85% relative humidity as well as the transverse crack formation were performed identically for version I and II.The median compressive strength of the separately casted concrete cubes with 150 mm edge length according to DIN EN 12390, part 1, [43] part 2, [44] and part 3 [45] ranged closely above 50 MPa.The cracks were formed by the application of splitting compressive load onto the center of the specimen, until the desired crack width of 0.3 mm was reached at the specimen's surface.
The failure of the version I approach is explained in Section 4. Because of this failure, we updated the specimen design to its final version II: In the second turn of this study, finally a series of eight transverse cracked, reinforced concrete bars were investigated.The splitting-induced crack had a superficial width of 0.3 mm (see Figure 1) and was caught by just a single Tempcore® steel bar B500B, [38] d = 8 mm as expected.This bar works as the conceived anode and has a concrete cover of 40 mm respectively.The total length of the anode is 150 mm.The electric contact to both the anodic conceived bar and the cathodic conceived cages is guaranteed via a noble steel threaded rod with 3 mm in diameter and a cable lug on that rod (see Figure 1).The marginal regions of the steel-rebar-rod-junction were sealed with a Teflon disc to avoid crevice corrosion at the junction. [46]The marginal regions of the glass-fiber-rebar rod junction were just glued.Statically this glued part of the junction is very weak, but as it lies in the faraway of the desired transverse crack, no problems should occur and a dismantling at the end of the investigation is very easy.If bending cracks are desired, the glued glass fiber rod junction needs further fortification.Furthermore, this rebar-rod-junction procedure convinces with the combination of electrical contact and accurate geometric positioning: The glass fiber near end is attached to the far end of the rod, the center part of the glass fiber is fixed to the cathodic cages and the fiber's far end is fixed in the wall of the framework.Thereby, a precision of ±3 mm of concrete cover is possible and no other spacers are needed in the whole framework (see e in Figure 1).
In contrast to the first approach (I in Figure 2), the cathodic cages of the final version (II in Figure 2) consist out of 24 manually welded steel bars per cage of the same steel type as the conceived anode.This welding procedure of the cathodic cages is on the one hand of extreme high effort, but on the other hand many very important conditions like electrical continuity, rebar spacing, mineralogical congruence within the steel electrodes, authenticity to cathodic regions in real structures and precise positioning within the framework is ensured permanently.
In both versions, the cathodic cages are positioned at the sides of the reinforced concrete bar to ensure symmetry of an occasional galvanic cell following the work of Warkus et al. [19] and have a constructively caused concrete cover of 10 mm.To round up the threeelectrode setup, consisting of the isolated anode and the cathodic baskets, a MnO 2 -reference electrode by Protec-tor® is positioned just below the later SCI of the anodic steel.The electric connectivity of the pair of cathodic cages is permanently given externally.As soon as the crack initiation is complete, the protruding parts of the glass fibers were cut and the specimens were circumferentially sealed with aluminum laminated butyl rubber tape.This circumferential sealing prohibits further lateral atmospheric ingress.
From then on the continuous monitoring of electrochemical parameters shown in Equation ( 1) started.The macrocell corrosion current between anode and cathode thus the coupled cages was recorded via an Almemo® DCshunt ZA9601FS1 once per hour.The potential values were measured once per week with a Gamry®1000potentiostat, whereby the first measure was the corrosion potential E corr at closed circuit.Subsequently, the circuit was opened, the transient potential was measured at 100 Hz for 30 s and left open for 24 h.After this day of depolarization the two equilibrium potentials E 0,A and E 0,C were recorded.Additionally, the linear polarization resistances of anode and cathode as well as the electrolytic resistance via a two-point-probing was measured between anode and cathode.For the sake of clarity, we will focus on the course of static potential values and skip the plethora of resistance values.The course of the resistances will be shown as soon as the total research project mentioned in the introduction is completed.
Each measurement routine within the 3 months of chloride exposure concluded with an impingement of 75 mL in version I and 30 mL in version II of 1.5% NaCl solution onto the confined basin above the crack.
After 3 months of weekly chloride impingement, the surface of the samples was coated with a surface protection system OS 11b by Sika®.The monitoring routine was maintained for the following 3 months.After this last period of sample monitoring, the anodic conceived rebars were dismantled whereby the original SCI was conserved.This procedure allowed chloride profiling perpendicular to the crack surface, along the decisive SCI.The exact method of profiling and chloride concentration determination is explained in the following section.

| Chloride content via drill dust sampling
As already indicated in Section 3.2 we performed the concrete casting via DIN EN 12390-11, [47] which is equivalent to the concrete, mortar, and cement-based testing method on chloride migration coefficient D RCM from non-steady-state migration experiments described in NT BUILD 492. [48]This allows for cross-referencing the diffusion coefficients resulting from sound concrete D RCM in contrast to the apparent diffusion coefficient along the SCI abbreviated with D SCI .The calculation of D SCI followed the well-known adaption of Fick's law of diffusion explained in fib Bulletin No. 34. [49]ue to the very low degree of reinforcement in the cracked region, the specimen setup of version II easily enables to dismantle the SCI on the anodic site at the end of the 180 days of corrosion monitoring.After destructive splitting, the outer regions of the sample the cathodic baskets fell apart and the remaining center of the specimen was divided parallel to the anodic rebar direction.From then on, we grinded the outcropped SCI along the former direction of the anodic rebar, starting at the crack surface and ending at 45 mm depth away from the crack (see Figure 2).
To enable a drill dust sampling, we grinded along the SCI with a diamond grinding pin possessing a length of five and a diameter of 10 mm.The grinding proceeded in steps of 5 mm corresponding to the height of the pin.The aim was to grind until the mass of one gram of drill dust was collected.A funnel below the grinded SCI guaranteed the collection of the grinded dust.
Chloride analysis within the drill dust samples was determined via argentometric titration according to Springenschmid et al. [50] As we gained multiple parts either left or right, above or below the crack, of conserved SCI we sampled two to three SCIs per specimen to increase the statistic value of chloride concentrations.With the present concrete composition, shown in Figure 1, the conversion factor from chloride per mass of concrete to chloride per mass of cement of 7.7 could be derived.

| General
In the following sections, the focus lies on the results of version II specimens.For the sake of clarity, we would already like to point out that version I did not provide authentic results with regard to real reinforced concrete structures.In version II, macrocell corrosion occurred in 50% of the specimens.Additionally, this percentage of corrosion activity was ensured via visual inspection after the dismantling.In this case, data yielding from passive samples will be depicted in gray and data yielding from active samples will be depicted in specimen-specific colors.

| Crack patterns
Although the crack widths held the desired 0.3 mm at the top and the bottom of the specimens, the interior crack pattern needed to be checked.The aim was, to prove, the cracks hit the rebars at the desired position.Figure 3 shows the center region of both specimen versions I and II and the resulting interior crack pattern of a resonated specimen under UV-light.
The resulting, interior crack pattern delivered a very good reason for strongly scattering system response and the sensitivity to the later-on mentioned driving potential onto chloride exposure in version I: The superficially applied 75 mL of 1.5% NaCl-solution had the possibility to reach either the cathodic conceived cages, or the anodic conceived rebar.This prevented the reproducible expression of an otherwise logical, positive driving potential and explains the possibility for negative values (see Section 4.3).As the cause for the undesired crack pattern, we assumed the degree of total reinforcement in the vicinity of the crack to be too high (compare Figure 1).The cracks followed a statically strong axis along the glass fiber rebar up to the cathodic conceived cages.To avoid any of those failures happening again, the final specimen version II was evolved.Within version II a very sharp crack with a width of 0.3 mm throughout the total height established and only scattered along coarse aggregates.This sharp crack is authentic to real concrete structures [51] and allows reproducible formation of confined anodic formation.

| Course and buildup of driving potential
After the version I approach of electrochemical corrosion monitoring and weekly chloride exposure, only very few parametric correlations or signs of reproducibility according to the concept established within 180 days of observation.Figure 4 shows a comparison of the potential trends within version I and II.
The driving potential of mostly all specimens from version I showed no response onto chloride contamination and more than half of the specimens showed negative values.This led to the assumption of severe misconception.Within a framework of an extensive sensitivity analysis onto all parameters of the macrocell The resulting interior crack pattern is shown in greyscale under UV-light after application of fluorescent resin.corrosion model (Equation 1) within the failed version I, the cathodic and the anodic equilibrium potentials equally represented more than 99% of influence onto corrosion initiation.We concluded, that the interaction of the two equilibrium potentials and their mergence to the so-called driving potential bear the reason for the misconception.
The driving potential values of version II show several features to be emphasized: • all values are positive as desired, • all specimens react equally to the first impingement with a jump of about 100 mV, • all specimens show an increase of driving potential during chloride exposure and a decrease after the surface coating as well as • a general increase in scatter of values with time.
Their maximum values occur only within macrocell corrosion activity (see Figure 5) and range at 350 mV, which is also evaluated in the previous study. [2]The present results allow for a holistic evaluation of the current and future series of the same setup.
The following figures solely focus on the potential values of version II and ensure an in-depth insight into the measurement of potential values with respect to chloride induced corrosion of steel in cracked concrete.The potential values are thereby converted into values against a standard calomel electrode (Ag/AgCl).To allow for further reference, we added the 0-V value against a standard hydrogen electrode (SHE).
As soon as a macrocell corrosion current was detected (I macro ≥ 1 µA) at the start of each measurement routine, the corresponding specimen in Figure 5 is depicted in a specific color (four out of eight specimens).This enables us to trace corrosion activity with respect to time and ambient conditions.Starting with Figure 5a and the cathodic resting potential E 0,C we see, that those values do not change drastically whether the sample shows corrosion activity or not.Nevertheless, a slight increase with time can be seen, which could be addressed to ongoing passivation of the cathodic conceived rebar. [52]The blue line in Figure 5b which shows an almost straight line of cumulative frequency also depicts this continuity and congruence.In contrast to that, the anodic resting potential strongly varies with time and state of corrosion.Similar to the driving potential values in Figure 4, the anodic resting potentials record a jump of about 100 mV after the first impingement heading toward lower values.The difference of E 0,C and E 0,A according to Equation (1) shows that the driving potential ΔE is exclusively built up by the course of E 0,A .The corrosion samples (active hence colored) of E 0,A range at 350 mV versus Ag/AgCl at the beginning of their active state and after few weeks all those values increase towards a more noble state.However, if repassivation occurs this runs only slowly but steadily.This trend is not necessarily linked to the condition of ongoing or terminated chloride exposure.Thus, it can be assumed that the initiation of corrosion activity in the case of transverse cracks depends on chloride exposure.However, the progress of corrosion activity is not solely determined by chloride exposure.The arising question of whether this phenomenon is applicable onto bending cracks is the core of a running study.Additionally, Figure 5a,b shows the 0-V-line versus a standard  [53] Other authors like Becker et al. [54] state that, solely measuring open circuit potential for a description of corrosion activity is not deterministic and strongly dependent on other parameters like, for example, ambient humidity.However, we state that as soon as this 0-V-line is crossed, the specimens of carbon mild steel in cracked concrete show macrocell corrosion activity.In the context of building preservation, we would like to point out that the measure of anodic resting potential is indispensable.
To conclude with the observation of potential values we address some electrochemical features to the corrosion potential E corr .This value corresponds to half-cell potential measurements or potential mapping respectively with the single difference of measuring the reinforcement potential value within the concrete and not on its surface.As soon as no corrosion activity is present, the values of E corr are congruent with E 0,C .This

| Chloride distribution along the SCI
Finally, we want to observe, if the above-mentioned courses of potential values and corrosion activity can be linked to the chloride concentration and distribution at the steel-crack-intersection along the SCI at the end of 180 days of monitoring.Figure 6 shows the assemblage of 18 chloride profiles as indicated in Figure 2, whereby the specimen-specific colors refer to the colors in Figure 5a.
From the total chloride profiles depicted in Figure 6, one can see that all specimens bear chloride concentrations above the critical threshold value of 0.5 (M.-%/c.)according to Schnell and Raupach [29] at low depth values and thereby in the vicinity of the crack.This applies for samples that showed corrosion activity as well as for those which did not.In general, the concentration values fit similar studies performed by Osterminski et al. [55] However, another example of the well-known difficulties in deterministic correlation of corrosion activity and chloride concentration is provided. [27,28]But one can still conclude, that in the present scope of simulating the borderline case of repair effectiveness via surface coating, no specimen except one value clearly surpasses 1.0 (M.-%/c.).Accordingly, within the present circumstances the effectiveness of the repair principle is suggested by Breit et al. [56] and Wunderle et al. [1] The occurrence of macrocell corrosion activity and the imaginable synergetic influence of anodic attraction onto the negatively charged chlorides cannot be confirmed.The diffusion coefficient D RCM , resulted in a value of 9.81 × 10 −12 m 2 s −1 with a standard deviation of 0.15 × 10 −12 m 2 s −1 .In contrast to that, the diffusion coefficients D SCI by fitting the present chloride profiles via the fib model code No. 34 [49] range at 2.14 × 10 −11 m 2 s −1 with a standard deviation of 1.51 × 10 −11 m 2 s −1 (see excerpt in Figure 6).This means that in this case the diffusion of chlorides is 2-3 times faster along the SCI than along pathways in sound concrete.Nevertheless, as anticipated in Figure 2, corrosion products and corrosion pits did only form at the steel-crack intersection and solely within the active samples.

| CONCLUSIONS
The following conclusions can be drawn from the abovementioned methods and results: • The potential values of version II follow the theory of terminated chloride induced corrosion in cracked concrete but their chloride profiles via drill dust sampling cannot confirm this trend deterministically.• Within the presented specimen design (version II), the parametric precision of chloride induced rebar corrosion in cracked concrete narrows down to textbook behavior of the investigated steel electrodes according to the electro potential series of metals.34. [49]As soon as macrocell corrosion was detected within the whole period, the profile is depicted in color.[Color figure can be viewed at wileyonlinelibrary.com] • All of the above-mentioned electrochemical parameters do influence each other, whereby certain values of potential change significantly when corrosion changes from passive to active state.In contrast to that, during repassivation the potentials change slowly but steadily.• The strong impact from formation of anodic iron dissolution onto half-cell potential measurements is confirmed.

F
I G U R E 1 2D-Sketches of the preliminary specimen history.3D-Sketches demonstrate the cathodic conceived mild steel baskets (linked to blue cable), a centered anodic conceived mild steel rebar (linked to red cable) with sealed margins (transparent gray in case of version I) and a MnO 2reference electrode (linked to black cable).Four (in version I) and two (in version II) glass fiber rebars (Schöck-Combar®) (yellow) ensure geometric alignment.The rebar-rod-junction allows geometric and electric confinement.Additionally, the rebar and concrete recipe as well as the crack initiation and the overall empirical timeline are shown.[Color figure can be viewed at wileyonlinelibrary.com]

F I G U R E 2
Dismantled set of steel-concrete interface (SCI)-bearing parts of the initial specimen displaying the sampling of chlorides along the SCI perpendicular to the crack.Sampling started at the crack surface facing into the primarily sound section of SCI.In the case of macro cell corrosion activity rust products formed at the SCI-crack interface.[Color figure can be viewed at wileyonlinelibrary.com]

F I G U R E 4
Course of the driving potential of both versions I and II.[Color figure can be viewed at wileyonlinelibrary.com] hydrogen electrode, which in this study can be found to separate passive rebar from active ones.We wanted to show this line, as it helps to display that almost every sample color versus gray, except three passive and seven active values out of more than 200 values, of anodic resting potential is clearly linked to this definition of noble or base metals.The red line in Figure 5b also indicates this segregation.Above and below the intersection of the red line with the dotted 0-V-line a change in the development of the curve can be observed, which may indicate a new distribution of resting potentials for active specimens.The breakdown onto a potential series of metals using steel electrodes in cracked, chlorideexposed concrete is new.The absolute values of E 0,A correlate very well with for example the works of Garcia et al. on chloride concentration dependent open circuit potential.

F I G U R E 5
Course of the anodic resting potential E 0,A (letter A in a); the cathodic resting potential E 0,C (letter C in a) as well as the corrosion potential E corr (letter E in a) and their response onto weekly chloride exposure and surface coating.As soon as macrocell corrosion current is measured, the sampled specimen is depicted in color (a).Display of all three potential values on the basis of their cumulative frequency (b).[Color figure can be viewed at wileyonlinelibrary.com] congruence is logic, as both cathodic and anodic conceived rebars are of the identical material.But as soon as corrosion activity is present, E corr decreases by about 100 mV and is sort of attracted by the corresponding value of E 0,A .The green line in Figure5bconfirms this observation by two standalone deviations above and below the amount of corrosion active samples.

F I G U R E 6
Chloride concentration in (M.-%/c.)starting at the steel-crack intersection and ending at 45 mm depth of the original sound steel-concrete interface (SCI).The threshold value of critical chloride content is displayed via the red line.The zoomed excerpt shows the fitted profiles according to fib model code No.
Beck et al. and Osterminski et al. and Warkus et al. extended this model by an uncertainty factor "microcell corrosion current."But as microcell currents were negligible in here and effective electrode surfaces are still indeterminable,