Electrochemical Method for Studying Localized Corrosion beneath Disbonded Coatings under Cathodic Protection

A 100 steel (UNS no. G10350) electrodes WBE^36–39 was used as the basis for the DAS. All the 1.6 mm diameter wires were closely packed into a 10 by 10 array of 18.5 by 18.5 mm and mounted in epoxy resin avoiding electrical contact between them. Before each test, the sensor surface was progressively abraded with 240, 600 and 1200 grit silicon carbide paper and cleaned with acetone followed by ethanol. To simulate the disbonded coating, a polymethyl methacrylate (PMMA) cover with an incorporated rubber seal was used. Fig. 1 illustrates the geometry of the WBE and the simulated disbonded coating. The crevice gap between the PMMA cover and the array's surface was fixed at 0.25 mm by introducing spacers at the areas indicated in Fig. 1. The PMMA cover and rubber seal were cleaned with ethanol and mounted over the WBE epoxy body using four AISI 316 stainless steel screws. In all cases, two rows of electrodes were directly exposed to the bulk solution outside of the crevice (simulating a coating holiday), and the eight remaining rows were under the PMMA cover (simulating a disbonded area). Fig. 2 illustrates the basic geometry of the electrochemical cell and the electrical connections used to measure current density maps under CP. The 2.5L cell was constructed from polyethylene and glass, with a flange connection for quick installation of the DAS. All potentials were measured against a Silver/Silver Chloride/Saturated KCl reference electrode. A Luggin capillary was used to minimize IR drops. The capillary tip was located 4 mm away from the section of the electrode array not covered by the crevice. Two 4.5 mm diameter graphite rods immersed to a depth of 120 mm in the solution were used as counter electrodes. The cell lid was designed to maintain a constant distance between the DAS, the reference and the counter electrodes in all tests. All tests were performed at a controlled room temperature of 22 ± 2°C.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Test solutions were prepared from analytical grade reagents and ultrapure water. Solutions were purged with air overnight to ensure oxygen saturation and were left open to air during the tests. The DAS was exposed for 50h in 0.1M NaCl, with initial pH values around 6.2, and phosphate buffer solutions of pH 7.8. In NaCl solutions, the DAS was tested at open circuit potential (OCP) and at −730 mV_Ag/AgCl. Although the selected NaCl concentration is too high to be considered representative of a typical soil, the objective of its selection is to maximize corrosion in order to assess the performance of the DAS in accelerated tests. The phosphate buffer solution was prepared by adding 227 mL of 1M K₂HPO₄ solution and 23 mL of 1M KH₂PO₄ to 5L of ultrapure water. Tests performed in pH 7.8 buffer solution, aiming to reduce the likelihood of steel passivation under CP, were also performed. In this case the DAS was tested at a less negative CP potential of −680 mV_Ag/AgCl.

Current densities across the sensor surface were mapped at regular intervals during all immersion tests. These current densities were measured using a 100 channel multiplexer (Fig. 2). The multiplexer was programmed to maintain 99 electrodes connected to the WE1 terminal while leaving the remaining electrode connected to the WE2 terminal. After a certain sample time, the electrode connected to WE2 was changed following a predetermined scanning sequence that swept the whole array. To avoid CP potential fluctuations when switching from one electrode to another, a make-before-break switching sequence was followed. CP was controlled by a Bio-logic VMP3 potentiostat using a typical three electrode setup where WE1 was used as the working electrode. A Gill AC potentiostat, used as zero resistance ammeter (ZRA), was interconnected between WE1 and WE2 to measure currents and, simultaneously, achieve the same electrical potential in all electrodes. Ground loops between both potentiostats were avoided as a result of the floating ground capability of the potentiostat acting as ZRA. The current sampling rate was 100 Hz in all cases. For all three experimental conditions, current densities were mapped every 10 min measuring the currents for 1 second at each electrode. At the end of each test, the bulk solution was removed from the cell and the pH of the solution trapped between the PMMA cover and the array was measured with strips of pH indicator paper.

Once the DAS was detached from the electrochemical cell, the PMMA cover was removed, the electrode array surface was cleaned with water followed by acetone and ethanol, and photographed. Corrosion products were then removed using ASTM G1-03 C.3.1 solution (1L HCl sp gr 1.19, 20g Sb₂O₃, 50g SnCl₂).^4.7 In all cases, the array was exposed to ASTM G1-03 solution for less than a minute before being rinsed with water followed by ethanol. After removing the corrosion products, the surface of the electrode array was photographed again. Due to the large size of the electrode array, surface profilometry measurements could not be performed directly on it. Therefore, surface replicas of the electrode array were obtained using a Struers repliset-F5 kit, and profilometry measurements were performed over the surface replicas instead.

Optical surface profilometry measurements of each electrode in the array were performed over the surface replicas using a Wyko NT9100 interferometry profilometer. The surface profilometry results were processed to calculate the negative volume (volume below the mainline). An additional surface replica of the electrode array in "as polished" condition was also measured to identify the negative volume associated with the surface roughness. This roughness negative volume was subtracted from all measurements and the result was divided by the area of each electrode in order to express the results in average thickness loss.

An ad-hoc (MATLAB R2012b, MathWorks, Inc., Massachusetts, USA) script was used to process the raw data obtained from the ZRA. This script segmented the data corresponding to each electrode and current map based on the switcher scanning sequence and rate. These data were then used to calculate the average current value at each electrode. Current densities were calculated based on the area of a single electrode. Each electrode current density value was arranged in a matrix following the electrode address. Current density surface maps were generated using linear interpolation of the 10 by 10 data matrix.

In order to study the cathodic reactions, additional tests were performed incorporating a 1 mm diameter by 10 mm long Pt electrode at the crevice mouth. Potentiodynamic polarization curves at a scanning rate of 10 mV/min where obtained for each of the experimental conditions after 25 h of exposure. For the experimental conditions where the DAS was maintained under potentiostatic CP, two channels of a Bio-logic VMP3 potentiostat were used as bi-potentiostat to maintain the DAS potential while polarizing the Pt electrode.

Consider an electrochemically coupled electrode array (e.g. WBE,^36–41 CMAS⁴² or any other^10,15,23) exposed to an electrolyte at OCP. In order to satisfy the principle of conservation of charge, the summation of the net current measured at all the electrodes must equal zero. Discriminating the anodic component ((i_a)_j) from the cathodic component ((i_c)_j) at each j-electrode, the conservation of charge for a n-electrodes array can be expressed by Eq. 1.

Since the electrodes in the array are electrochemically coupled, they can interact with each other. In other words, Eq. 1 does not necessarily imply that the anodic and cathodic current components at each electrode must be equal (Eq. 2).

In fact, if Eq. 2 were the only way to satisfy Eq. 1, the net currents measured at each electrode would be zero and the electrode array would not be capable of providing information regarding current distributions. Although Eq. 2 satisfies the principle of conservation of charge, it only represents the particular cases where the corrosion cells at each electrode operate isolated from each other. The results obtained when the DAS was exposed at OCP agree with this theory. While the total net current flowing in or out the array was constantly zero (Fig. 10), the current distribution maps (Fig. 4) indicated predominantly cathodic and anodic areas across the array surface.

When the electrode array is being cathodically polarized, independent of whether under potentiostatic or galvanostatic control, the net current resulting from all cathodic and anodic reactions on the array, must be cathodic. This condition is mathematically expressed in Eq. 3.

Similarly to the OCP case, the possible electrochemical interaction between electrodes in the array implies that (Eq. 4) is a sufficient but not necessary condition to satisfy (Eq. 3).

In other words, despite the whole array being cathodically polarized, some electrodes could still present a net anodic current. Whether a particular electrode will present a net anodic or cathodic current depends upon the balance between anodic and cathodic reactions occurring at its surface. The DAS takes advantage of this, by controlling the distribution of cathodic reactions which allows the detection of anodic currents at the simulated disbonded coating area.

As shown in Fig. 3, under both experimental conditions where CP was applied, oxygen reduction was the main cathodic reaction. However, it is known from previous experimental^12,13,18 and theoretical analyses^14,17,22,32 that oxygen concentration is not constant across a crevice formed beneath disbonded coatings. According to the Chin³² and Song^17,22 models, dissolved oxygen should be totally depleted under the disbonded coatings at a distance equivalent to four crevice gaps in the steady state. This suggests that along disbonded coatings, there are two distinctive scenarios based on the accessibility to oxygen. A schematic Evan's Diagram illustrating the corrosion in the bulk solution case, representative of the area outside the crevice, is presented in Fig. 11a. In this figure, the anodic curve (in dashed lines) is a generic representation. For the discussion to follow, the only value of relevance in this curve is the anodic current density at the applied CP potential (i_a). Outside the crevice, even though E_CP is less negative than the iron dissolution equilibrium potential and metal dissolution indeed occurred, the recoded current density values at this location were rarely anodic. This is because the difference between small anodic currents (i_a) and the larger cathodic currents (i_lim.1), results in net cathodic currents being measured at these electrodes (i_M1). This not only explains why the sensor was unable to detect anodic currents outside the crevice, but also why no anodic currents were detected deep inside the crevice at the beginning of the test when sufficient oxygen still existed, i.e., even though anodic reactions still occur, the cathodic reactions dominate. However, deep inside the crevice, as the reactions proceed and the oxygen concentration decreases below a certain level, the situation changes considerably. This condition is represented by an Evans Diagram shown in Fig. 11b. In an extreme case, this lower oxygen limiting current (i_lim.2) could reach negligible values in the steady state. Consequently, the difference between i_a and i_lim.2 would be i_M2 (almost equal to i_a) providing a way to evaluate anodic currents within the crevice and thus the localized corrosion rates. Depending on the IR drop produced along the crevice, the local potential can change (due to cathodic shielding), but since the method does not require knowledge of the local potentials, this does not affect the measurement process.

Zoom In Zoom Out Reset image size

The accumulated charge of the anodic current densities that were measured throughout the course of an experiment were used to calculate the metal loss distributions by means of Faraday's Law. Fig. 12 presents the results of these calculations for all experimental conditions and the same surface profilometry results shown in Figs. 5c, 6c and 7c re-scaled to allow an easier comparison with the electrochemical results.

Zoom In Zoom Out Reset image size

When comparing the metal loss distribution from Faraday's law found for the OCP case (Fig. 12a) with the corresponding surface profilometry results (Fig. 12d), a high correlation in the distribution of the attack was found inside and outside the crevice area. On the other hand, the metal losses measured electrochemically for the test performed under CP in NaCl solution (Fig. 12b) did not show an obvious correlation with the corresponding surface profilometry results (Fig. 12e). Despite the fact that most of the corrosion damage was observed outside the crevice, no corrosion was detected electrochemically at this location. Inside the crevice, although a certain correlation was observed between the corrosion pattern measured electrochemically (Fig. 12b) and the corrosion products found on the array surface (Fig. 9a), the corresponding surface profilometry results (Fig. 12e) did not present the same pattern. This is likely due to the very small metal losses produced under these experimental conations and the uncertainties associated with surface profilometry measurements. When the DAS was exposed in pH buffer solution at a less negative CP potential, where significantly more corrosion was produced within the crevice, a better correlation in metal loss distribution was obtained between electrochemically calculated (Fig. 12c) and profilometry results (Fig 12f) at that location. Outside the crevice, in the same way as for the NaCl case under CP, no corrosion was detected electrochemically although the most severe corrosion attack was found at this area. Outside the crevice, the predominance of cathodic currents drives Faraday's Law based calculations to falsely indicate that almost no corrosion is taking place. This important limitation on the direct application of Faraday's Law to current density maps recorded under CP is addressed in the discussion below.

Based on the previously described theoretical framework and our recently published data analysis method,⁴⁶ a technique to estimate the corrosion damage outside the crevice can also be developed. Fig. 13 considers the case where cathodic reactions are greater than anodic reactions (as in Fig. 11a), but takes into account two different electrodes with the same cathodic but different anodic behavior. Differences in anodic behavior between electrodes in the same row can be expected due to the complex microstructure of commercial carbon steels and the very heterogeneous conditions developed under disbonded coatings. In Fig. 13 the currents measured by both electrodes are cathodic (i_M1 and i_M1^'), but present different magnitudes. In fact, the larger the cathodic current measured, the lower the anodic current at that electrode. Consequently, if the largest cathodic current measured is taken as a 'baseline', then the difference between this value and the current measured on any other electrode, with the same access to oxygen, could be used as an estimation of the anodic component at this second electrode (i_est = i_M1-i_M1^'). I_est is considered the corrected anodic current value for that second electrode.

Zoom In Zoom Out Reset image size

A critical step in the implementation of this method to the DAS, is to determine which electrodes are comparable. As mentioned before, this method is only applicable between electrodes with the same cathodic reaction rates. This is not generally the case across the sensors surface, but taking advantage of the geometry of oxygen diffusion along this uniform gap crevice, it is acceptable to assume that oxygen has the same access to all the electrodes in the same row. As shown in Fig. 3, in all experimental conditions the main cathodic reaction was diffusion controlled. Under these conditions, the oxygen limiting current would only depend on the oxygen accessibility and not on variables such as pH and potential, providing a stable baseline for comparison. An exception needs to be considered for the electrodes at the extreme ends of each row, where a small gap left between the rubber seal and array surface would have provided an extra solution volume with additional oxygen, producing the abnormally large cathodic currents observed in Figs. 5 and 6.

Nonetheless, this method implicitly assumes, by taking the electrode with the largest cathodic current as the baseline, that the anodic current component at this electrode is zero. Although in the general case this assumption is not always satisfied, it can be demonstrated (demonstration presented in Appendix) that for a row of electrodes presenting a cathodic or zero total net current, the method always produces an equally or more accurate corrosion estimation than the direct application of Faraday's Law to the anodic current density values measured by the array. However, when the total net current flowing through the analyzed row of electrodes is anodic, the 'corrected currents' method could lead to a larger underestimation of corrosion than that produced by Faraday's Law.

Despite the fact that the DAS was tested under CP and at OCP conditions, cases of anodic total net current for a given row of electrodes were commonly observed. In order to avoid the cases where corrected currents would lead to a larger underestimation of corrosion compared with the direct application of Faraday's Law, the corrected currents calculation method compares the corrected currents with the anodic values at each electrode as follows.

First, for each current map, the largest cathodic current measured for each electrode row (except those at the extreme ends of each row) are identified as 'baseline' values. Then, the differences between the currents registered at other electrodes and the baseline corresponding to its row are computed as the preliminary corrected currents, for that current map. These corrected currents are then compared against the measured current at each electrode. If the value estimated by the corrected currents method is lower than the anodic value measured by the electrode, the value estimated by the corrected currents is replaced by the measured value. Finally, the accumulated charge from all current maps is used to calculate the corrosion damage based on Faraday's Law.

The metal loss maps calculated with the 'corrected currents' method are presented in Figs. 12g to 12i. In the OCP case (Fig. 12g), the method outperformed the direct application of Faraday's Law (Fig. 12a), presenting an improved correlation with the corrosion pattern observed over the electrode array in the first row of electrodes (Fig. 7). The observed differences between Figs. 12a and 12g indicate that this method may successfully estimate the anodic component at electrodes where cathodic currents are dominant. When this calculation method was applied to the current maps obtained under CP (Figs. 12h and 12i), the resultant metal loss maps showed a good correlation with the corrosion patterns observed outside the crevice (Figs. 8 and 9) in both cases. Within the crevice, the differences between metal losses predicted by this method and the direct application of Faraday's Law (Figs. 12a to 12c), were smaller. This suggests that cathodic current components at this location were small.

The proposed working principle of the DAS not only provides a reasonable explanation of the experimental observations, but also the opportunity to anticipate some of its limitations. For instance, one may expect that anodic currents component can only be directly measured under a disbonded coating when other cathodic reactions, besides oxygen reduction, are negligible. This may not generally be the case, oxygen and other oxidizing species could permeate through the coating, or if the system does not allow the pH to increase, water reduction rate could become significant at any location within the crevice. The corrected currents method could still be applicable in the cases where oxidizing species permeate through the coating because the transport of species would still be described as a two dimensional (2D) diffusion problem (simple geometry crevices), and the resultant cathodic reactions should still occur at the same rate over all electrodes in the same row. Due to the crevice geometry, similar local potentials could be expected for electrodes in the same row, and this could provide the basis to extend the corrected currents method to most activation controlled reactions. However, the case for water reduction would be different. Because not only is this reaction pH dependant, but also it produces H₂ bubbles that would affect the local potential of each electrode within the same row in an unpredictable way. Thus, the corrected currents method would not be applicable in this case.

The current distributions measured by the electrode array when H₂ evolves at a significant rate could still assist in the estimation of corrosion. At locations near the crevice opening, where CP effectively polarizes the steel, the large cathodic currents would most likely conceal any anodic current component and corrosion would be undetected. However, sufficiently deep into the crevice, where CP is completely shielded, the overall cathodic reaction rate must be equal to the overall anodic reaction rate. At this location, the electrode array would be able to measure the imbalance between the anodic and cathodic current components at each electrode. The evaluation of these hypotheses requires an extensive redesign of our experimental setup and will be subject of further research.

The installation of the sensors presented in this work would be similar to that of the CP coupons commonly used to monitor OFF-potentials.^48,49 The DAS would be permanently buried next to the pipeline at areas of high corrosion susceptibility and an electrical connection between the pipeline and the WE1 terminal (Fig. 2) would be required to maintain the sensor under CP.