Evaluating organic coating performance by EIS: Correlation between long‐term EIS measurements and corrosion of the metal substrate

In this study, the possibility of obtaining a corrosion indicator from a time series of electrochemical impedance spectra acquired during immersion of intact organic coatings was inspected. To generate a wide distribution in performance, coatings with different pigment volume concentrations were applied on AA2024‐T3 and cold‐rolled mild steel and tested in four different electrolytes and for different exposure times. The relative ranking of anticorrosion performance was based on the visual appearance of the metal substrates after stripping the coating and it correlated well with the estimation of the corrosion charge estimated from the time series of electrochemical impedance spectroscopy spectra.


| INTRODUCTION
Organic coatings are widely used to protect metals from corrosion and extend the operational life of artefacts and structures.Eventually, most organic coatings will fail due to environmental exposure allowing corrosion of the underlying metal surface.There is a wide range of mechanisms that lead to failure, and since they often act in combination, the overall in-service coating behaviour is not completely understood. [1]From an industrial point of view, assessing coating performance before deployment is critically important.However, no single corrosion test exists that can be considered a completely reliable indicator of in-service performance.
Most of the common industrial tests rely on exposure of the specimen to an aggressive environment in purpose-built exposure cabinets, [2,3] by immersion in a specific solution [4] or outdoor conditions, and evaluating the surface during or after a set exposure times.These tests have the advantage of being easy to perform, but the evaluation tends to only rely on the final appearance of the specimen.An alternative to the typical industrial tests involves coupling the exposure to the aggressive environment with periodic electrochemical measurements [5] since the latter can provide quantitative and, in some cases, mechanistic information on the coating behaviour.8][9][10][11][12] EIS can in fact be applied on bare metals, scribed coatings and to study through coating corrosion, since information on various aspects of the coating, including water uptake and presence of defects, can be obtained.Typically, for an intact coating during the early stages of exposure, the EIS spectrum is mainly dominated by a capacitive response.At a later stage, water uptake occurs and the coating resistance can be resolved.Finally, when a thorough coating defect is developed, a significant shape variation compared to the initial spectra is observed.
EIS spectra can be interpreted using the equivalent circuits proposed by Beaunier et al. [13] to provide quantitative information on the faradaic process, double-layer capacitance and coating resistance and capacitance. [7]EIS is also used to monitor coated systems during accelerated tests involving, for example, the increase of temperature of the electrolyte [14] or scribing the surface of the coating, exposing a portion of the metal surface [15] to an aggressive environment or environmental cycle.Compared to other corrosion test results, the interpretation of EIS data is relatively complex and requires well-trained personnel to select equivalent circuits, perform fitting and interpret the outcome. [16]As a result, a clear evaluation of the coating performance and relation with the underlying metal surface appearance is difficult to establish from EIS data.Although previous work aimed to relate the coating failure with time evolution of the impedance modulus for a given environment [17] was proven effective, further work is needed to relate EIS spectra to the performance of organic coatings regardless of the exposure time and test environment.Therefore, this work explores the possibility to relate the metal surface appearance underneath the organic coating at the end of an immersion corrosion test to one single indicator obtained from EIS data that is independent of the exposure time and environment.To study this, a relatively large data set comprised of approximately 500 EIS spectra was generated by periodically measuring the response of selected combinations of model epoxy-amine coated systems and test environment, generating 38 different conditions in total.Each condition represents a particular combination of metal substrate, pigment volume concentration (PVC) and aggressive environment.

| Metal panels
Bare AA2024-T3 (5.0 × 7.0 cm, 0.8 mm thickness) aluminium alloy panels were provided by Arconic (identified as 'AA' hereon).Before coating application, AA panels were wet abraded with Scotch-Brite™ 7447 PRO pads and the commercial solvent cleaning agent 98068, provided by AkzoNobel, was used to remove surface contamination.Cold-rolled mild steel panels (10 × 20 cm, 0.5 mm thickness) were obtained from Van Raak Staal (indicated as 'steel' hereon).They were first degreased using Sikkens solvent and dried in air.Sanding was performed at P180 grit using 10% wt/ wt alabastine/water as lubricant followed by a rinsing step with distilled water.The surface was then immediately dried using a paper towel to avoid corrosion initiation.

| Preparation of organic model coatings
Model epoxy-amine coatings were prepared by varying the total PVC from 0% to 40% to obtain a wide variety of behaviours in terms of water permeability, rusting and blistering. [18]The experimental matrix is reported in Figure 1.The PVC was in all cases kept below the critical pigment volume concentration, 66% (CPVC), calculated using the following equation [18] : In this equation, i represents a pigment, N is the total number of pigments, V i is the volume fraction of the pigment, ρ i its density, OA i is the gram of binder per gram of pigment and ρ L is the binder density (1.16 g/cm 3 ).

| 157
The formulations are summarised in Table 1. [19]In all formulations, the ratio of titanium dioxide to barium sulphate is kept constant at approximately 0.9.

| Sample preparation
The coatings were prepared by mixing the raw materials as follows: Component A was prepared by adding the ingredients under stirring in a 370 mL glass jar.After the addition of 350 g Zirconox® pearls (1.7-2.4 mm), the pigments were dispersed to a fineness of grind lower than 25 μm by shaking on a Skandex® paint shaker.After 20 min shaking, the pearls were separated from the coating.Component B and C were mixed separately and added to Component A just before the application.The coating formulations were applied on both aluminium and steel panels with a high-volume low-pressure spray gun.After the application at 23°C and 55% relative humidity, the coatings were cured at ambient environment for at least 7 days.The dry film thickness of the coatings after drying was between 25 and 40 μm.

| EIS measurements
EIS was used to assess the coating conditions as a function of time.Due to the requirement of measuring a large number of specimens for an extended period of time, an in-house developed setup (Figure 2) was used to minimise intervention during testing.The setup comprised of a LattePanda microcomputer running an in-house developed LabVIEW programme, which controlled both an Ivium Compactstat potentiostat and a 10 channels inhouse developed multiplexer board.Each channel of the multiplexer was connected to an individual electrochemical cell (three electrodes configuration), containing a coated specimen connected to the working and reference 2 terminal, an Ag/AgCl reference electrode connected to the reference 1 terminal and titanium foil connected to the auxiliary terminal.Although the employed setup allowed to measure a large number of specimens, it is susceptible to some noise, especially when the impedance measured is higher than 10 8 Ωcm 2 .However, this limitation was not considered to be a substantial issue for the purpose of obtaining an indicative performance ranking on large sample batches.Before EIS testing, all the specimens were masked with a mixture of beeswax and colophony with a mass ratio of 3:1, to leave an area of 9 cm 2 of coated metal exposed to the electrolyte.The masking was applied such as the thickness of the masking layer was in large excess compared to the thickness of the coating under study (typically 500 μm vs. 25-40 μm), such as the capacitive contribution associated to the masking was minimised.
EIS measurements were performed by immersing the intact coatings (nominally free of optical defects) in four different test electrolytes: (i) 5 wt% NaCl (pH 6.40), (ii) 5 wt% NaCl + 0.1 M HCl (pH 1.09), (iii) 5 wt% NaCl + 0.1 M NaOH (pH 12.96) and (iv) 0.5 g/L NaCl + 3.5 g/L (NH 4 ) 2 SO 4 (pH 6.84, hereon indicated as 'Prohesion electrolyte') (Figure 1).The test duration varied from approximatively 1000 h when using 5 wt% NaCl and prohesion electrolytes, to approximately 168 h for the other two more aggressive electrolytes (low and high pH).Based on preliminary testing, the exposure time was chosen such as, for each test electrolyte, a range of behaviours could be obtained, including significantly corroded specimens, slightly corroded specimens and relatively noncorroded specimens.Further, the selected times are comparable to those applied during neutral salt spray and prohesion tests following the standard practices and water reagents as described in ASTM B117-19 and ASTM D1193-06 accordingly.All the coated samples were tested only once in each environment since the specific effects of the electrolyte and PVC are not of primary interest in this study.The occasional outlier, possibly due to a pre-existing defect in the coating, is not considered an issue with regard to the identification of a suitable corrosion indicator (CI).The reason is that if a particular specimen has a pre-existing defect, it generally results in more severe corrosion compared to defect-free coatings.Such an outlier simply represents an example of a coating performance.Instead, obtaining a large number of varied EIS responses for extended periods of time provides a wider window on the variability of coating performance, and hence allows us to evaluate the accuracy and reliability of having a CI calculated from EIS spectra.
Measurements were performed on each individual specimen at 24 h time intervals.The open-circuit potential (OCP) was recorded for 3 min before each EIS measurement.After 3 min of OCP, an AC perturbation was applied with an amplitude of 0.1 V [20] in a frequency range between 10 5 and 10 −3 Hz recording 10 points per decade.The relatively large amplitude was selected to improve the signal-to-noise ratio, which at lower amplitudes was low due to the relatively high intact coating resistance and the presence of the 10-channel multiplexer board between the potentiostat and the cells.Given the high impedance of the intact coatings in the region of 10 8 Ωcm 2 , an amplitude of 0.1 V produces a low-frequency AC current with an amplitude in the order of 10 −9 A cm −2 .Such relatively high amplitude is not expected to induce a significant acceleration to the corrosion process under the intact coating; even when a continuous anodic current of 10 −9 A cm −2 is assumed, which is a large overestimation.This is due to the fact that the AC signal is not always anodic and EIS is measured only for approximatively 1 h/day.Hence, Faraday's law of electrolysis indicates that the material oxidised in 1 month would be of the order of 1.2 nm, which is negligible.Thus, although the relatively large amplitude might introduce some acceleration of corrosion once a macroscopic through-coating defect is well developed, it is unlikely to significantly affect the coating degradation process that results in the formation of such defect.

| Surface evaluation
Following EIS testing in the various electrolytes, optical images of the coated metals were acquired immediately after the test to visually assess the condition of the organic coatings and the metal surface.Subsequently, a commercial noncorrosive paint stripper (Strypit) was used to strip the coatings and observe the underlying metal.The stripping product was applied using a brush.After 30 min the excess paint stripper was removed with paper and the coating was stripped using pressuresensitive tape.When this procedure did not result in complete stripping, it was repeated until needed.Images were then acquired following stripping, to visually assess the corrosion of the underlying metal.ImageJ software was used to analyse the images of all the specimens after coating stripping.The images were binarized to calculate the amount of corroded area over the total exposed area.During image analysis, the dark local spots or dark regions developed against the lighter metal background were considered as corrosion during the analysis of the corroded area.

| EIS
Figure 3 presents the first set (i.e., the first 6 h) of EIS spectra acquired for AA and steel coated with organic coatings with different PVC and immersed in the four selected electrolytes.For coatings without pigments (Figure 3a,b), the responses obtained are closely similar for both the AA and steel substrates since high-impedance modulus >10 8 Ωcm 2 are recorded.
Increasing the PVC to 10%, the behaviour is generally the same considering the capacitance and lowfrequency impedance modulus (Figure 3c,d); whereas for high PVC concentrations (Figure 3e,f), significant differences regarding the modulus are occasionally revealed, indicating that the high PVC systems are more likely to have not optically visible initial defects [21] whose effects dominate the EIS response from an early stage.The quality of the spectra deteriorates at a low frequency for impedance modulus above 10 9 Ωcm 2 and the values are somewhat scattered.As a result, some error is expected when estimating the low-frequency impedance modulus from the spectra.However, discriminating between extremely high values of low-frequency impedance modulus is not a primary concern, since the corrosion rate of the metal under such high-impedance coating is negligible.In some other cases (in Figure 3d steel with 10% PVC in the basic and acid environment, in Figure 3e AA with 30% PVC in acid environment and in Figure 4f steel with 30% PVC in acid environment) the highly aggressive environment results in changes to the system which occur within the same time frame of a measurement, and this produces an artefact evident as a decrease in the impedance modulus with decreasing frequency.The evolution of the coating systems in time covers a wide range of behaviours, mostly depending on the substrate and pigment concentration.Figure 4, for example, shows a comparison of the coating behaviours at approximately 168 h of immersion in electrolyte.The coatings without pigments maintain in all cases a low-frequency impedance modulus above 10 6 Ωcm 2 (Figure 4a,b) and little variation in capacitance both for AA and steel.In general, higher impedance values are measured in the case of AA compared to steel (Figure 4a), and lower impedance values are measured in acidic environments compared to the rest of the chosen environments (Figure 4b).The drop in impedance modulus and the variation of capacitance becomes more evident when considering slightly (Figure 4c,d) or significantly (Figure 4e,f) higher PVC, respectively 10% and 30%.
Some characteristic examples of the time evolution of the impedance response for high PVC systems (equal or higher than 30%) are presented in Figure 5.All systems show a significant change in their EIS response, evident as a drop in the low-frequency impedance modulus and the appearance of a plateau in the impedance modulus at medium frequencies (generally from 100 to 1 Hz), associated with a second-time constant.This is reflected on the surface appearance of the specimens in Figure 6, which shows coating failure in the form of blistering (blue circles).Thus, the presence of a second-time constant in the EIS spectra can be associated with the formation of a coating defect.On the other hand, systems with low PVC (equal or less than 10%) display generally a stable EIS response throughout the whole duration of the test, (Figure 7 reports some examples of this characteristic behaviour), with only a slight increase in the coating capacitance and decrease in impedance modulus.These systems generally appear not corroded or only slightly corroded (blue circles) at the end of the test (Figure 8).Other differences in coating behaviour were observed depending on the substrate.As an example, Figure 9 shows EIS spectra measured on AA and steel coated with a 0% PVC coating immediately after immersion in 5 wt% NaCl and prohesion electrolyte (Figure 9a,b) and after 48 h (Figure 9c,d).The two metals show very similar initial responses, due to the application of the same coating (Figure 9a,b).After 48 h (Figure 9c,d), the high-frequency part of the spectra still overlaps, indicating that the systems have probably a similar capacitance, but the low-frequency impedance modulus is significantly different.AA maintains a high modulus throughout the immersion time, whereas a substantial drop is observed for steel.This was noted in 5 wt % NaCl (Figure 9a,c) and in prohesion electrolyte (Figure 9b,d), but it is also valid for 10% PVC coatings and for the other two electrolytes (5 wt% NaCl + 0.1 M NaOH and 5 wt% NaCl and 0.1 M HCl), showing that this behaviour is independent of the active or passive conditions associated with the nominal electrolyte's pH.In summary, the systems studied display a wide variety of behaviours, mainly depending on the PVC, the substrate and the test environment.

| Visual appearance after testing
Figure 10 shows the specimens after the test in 5 wt% NaCl and prohesion electrolyte before (a) and after (b) coating stripping.It can be seen that before stripping (Figure 10a), blisters and regions of damaged coating are evident in some specimens, especially when PVC is equal to or higher than 30% (red frame in Figure 10a,b).Evidence of corrosion before stripping is rarer when the PVC is less than 30%, and the surface of the coatings appears intact.However, after stripping (Figure 10b), more systems display corrosion that had developed underneath the coating (black frame in Figure 10b) visible as red corrosion products or dark local points on the lighter metal surface.Finally, some specimens, especially with low PVC, showed an optically clean or only slightly corroded metal surface, indicating that the coating was able to protect the underneath metal surface effectively from corrosion (green frame in Figure 10b).
The same procedure was applied for the shorter tests in 5 wt% NaCl and 0.1 M HCl and 5 wt% NaCl and 0.1 M NaOH (Figure 11).In the case of 5 wt% NaCl and HCl, a clear distinction in the protective properties of the coating can be seen with respect to the PVC percentage: above 10%, all samples present a big blister covering the whole exposed area (Figure 11a) and severe corrosion signs are observed underneath the coating (Figure 11b).Corrosion appears less severe in the case of 5 wt% NaCl and 0.1 M NaOH electrolyte also for AA, suggesting that the coating was effective in protecting the metal even under such alkaline conditions.Generally, low PVC coatings are associated with better metal surface appearance in both short and long tests; whereas, the systems with PVC higher than 20% have shown severe corrosion signs in many cases.
Figure 12 displays the appearance of the steel (Figure 12a) and AA (Figure 12b) surfaces after the coating has been stripped.The images are arranged in order of the percentage of surface corroded area, as determined by the analysis performed using ImageJ software.The binarized images are reported in Figure 13.The least corroded substrates are labelled with a lower number and the most corroded with a higher number.This ranking is based on the corrosion area of the metal surface following image analysis, regardless of the test electrolyte, the duration of the test, or the coating formulation used.
For this reason, all the images of the specimens after the tests are reported.Minor local corrosion points are highlighted in blue to guide the reader.In the case of steel (Figure 12a and Table 2), signs of corrosion are evident even on the best specimens, considering the metal surface.As corrosion becomes more severe, it is associated with an increasing number of corrosion sites (from number 4 to number 8), or with a more general corrosion affecting the whole surface (number 9).In the case of AA (Figure 12b and Table 3), the surface of the first five specimens appears similar, and no differences could be observed.For this reason, all of them were labelled as first.Corrosion signs become clearly visible from the specimen in sixth position.Single small corrosion sites are evident from specimens 6 to 9, and more severe corrosion is visible from specimen 10 onwards.

| Correlation between EIS and corrosion on metal substrate
Tables 2 and 3 report in the second column the ranking based on the images of Figure 12 for steel and aluminium alloy, respectively.The tables also report the values of the impedance modulus spectra acquired shortly after immersion (i.e., 0 h) and at the end of the exposure time (i.e., end of the experiment time), measured at 10 −3 Hz.For both cases, the impedance values are ranked from high to low accordingly.A CI is calculated using the time integral of the reciprocal of the low-frequency impedance modulus: where CI has the units of Ω −1 cm −2 s.In Equation ( 2), i is the single measurement, N is the total number of measurements, Z | | i LF is the low-frequency impedance modulus normalised by area (Ωcm 2 ) at measurement i, and t Δ i is the time interval between measurement i and the previous one.The rationale behind this parameter is explained in detail in the discussion section.Figure 14 compares graphically the relative ranking based on the appearance of the metal after stripping (horizontal axis), used as a reference ranking, and the rankings based on impedance modulus at 10 −3 Hz acquired at 0 h (vertical axis, Figure 14a) and at the end of the selected test time (vertical axis, Figure 14b).The red line only represents the ideal scenario in which the visual and EIS-based ranking coincide.In other words, the relative ranking obtained from a single EIS measurement (first in Figure 14a, or last in Figure 14b) corresponds to the relative ranking determined by visual inspection after stripping and ImageJ analysis.From Figure it is evident that a correlation between image-based and EIS ranking exists, but it is relatively poor since there are many points where the ranking from EIS measurement is significantly different from the ranking from surface images.Comparing the two, the ranking obtained from the last measurement has a better correlation to the surface appearance than the ranking based on the first measurement.
Figure 15 shows in a similar graph the comparison between the ranking based on the images (horizontal axis) and the ranking based on the time integral of the reciprocal of the impedance modulus at low frequency (vertical axis).It can be observed that the correlation between the two rankings is much improved compared to the two previous cases.The comparison between the CI ranking and the image ranking in Figure 12 shows an average of 2.4 incorrect positions.On the other hand, the rankings based on the last and first impedance modulus have an average error of 3.8 and 4.3 positions, respectively.These calculations were done by finding the differences between the considered ranking and the image ranking for each specimen, squaring those differences, summing them, and then dividing by the total number of specimens.The square root of this result gives the average difference between the ranking based on images and the considered ranking.

| DISCUSSION
The intact coatings used in this set of tests show different behaviours depending on the amount of PVC present, the substrate and the test electrolyte.For very short test times, systems with low PVC coatings (0% and 10%, Figure 3a,b) show a similar first EIS response regardless of the metal substrate (Figure 9a,b) and electrolyte.Coatings with high PVC tend to display a  more differentiated response from the early stages and to corrode earlier; regardless of the test conditions, the impedance response is often characterised by the presence of a second-time constant.Optical evaluation of the coated surfaces and the underlying metal at the end of testing suggests that such a second-time constant is likely to be associated with the presence of coating defects, responsible for the corrosion process initiation.This behaviour can be due to the fact that high volume of produces a considerable reduction of the coating impedance, due to the possible presence of water penetration pathways, resulting in increased likelihood of corrosion initiation. [22,23]Coatings with low PVC generally do not show clear signs of localised failure, possibly because these coatings do not commonly offer preferential paths for electrolyte penetration.As a result, the corrosion behaviour is determined by progressive penetration of the electrolyte into the polymeric matrix followed by interactions taking place at the metal/coating interface.
Having identified, for these specific systems, the general trends in terms of corrosion behaviour as a function of metal substrate, PVC and electrolyte, it is useful to consider the possibility of obtaining a single parameter from a time series of EIS spectra, which can account for the relative ranking between specimens regardless of the test conditions and duration.[26] However, with this procedure is difficult to intuitively account for the effect of exposure time on corrosion, which is of main importance, together with corrosion rate.To identify such a CI, it is useful to consider three main cases that might unfold during exposure.These cases cover three possibilities, even if some of them might or might not frequently occur depending on the coating performance and ions transport in the polymer. [27]These cases involve: (1) an intact coating well attached to the metal substrate with no defect, (2) the development of cathodic and anodic regions within a defect under relatively intact coating separating the inner environment from the outer environment (Figure 16a) and (3) anodic and cathodic regions within a defect (including a blister) directly connected to the outer environment due to a macroscopic through coating defect (Figure 16b).
The first case, representing an intact coating well bonded to the metal substrate, is typical of the initial stages of the test or of a good coating.In this scenario, the coating resistance can be estimated by EIS, as it has been commonly employed to measure the barrier property of an intact coating. [14,28,29]In this case, high values of coating resistance are typically associated with good anticorrosion performance.So, qualitatively, the corrosion rate is likely to be related to the reciprocal of the coating resistance.When anodic and cathodic regions are both established within a blister under an intact coating (Figure 16a), the corrosion current might not be necessarily related to the coating resistance, and EIS would likely underestimate corrosion.Finally, in the last case (Figure 16b), corrosion cells are established within the defect and there is an electrolyte path from the metal surface to the electrolyte.In this case, the charge transfer resistance at the metal surface is likely to dominate the resistive component of the EIS response and the corrosion current is again proportional to the reciprocal of the low-frequency impedance modulus.The first case described is likely to be representative of a coating that has been in contact with the environment for a time significantly shorter than the time required to induce significant deterioration.Vice versa, the third case (Figure 16b) is representative of a damaged coating and established substrate corrosion.The second case is likely to evolve into the third.
From this discussion, it appears that; (1) when a coating is intact, the corrosion charge, and hence the corrosion damage is very small and it is likely to be proportional to the time integral of the reciprocal of the impedance modulus, which is very high; (2) when corrosion is well established due to a through coating damage, the corrosion charge is likely to be again proportional to the time integral of the reciprocal of the impedance, although with a different proportionality coefficient, (3) it is likely that the case described in Figure 16a, evolves relatively rapidly in a trough coating defect.Considering the coatings and pigments studied, it seems appropriate to select the time integral of the reciprocal of the low-frequency impedance modulus as a single CI that can approximatively incorporate all the corrosion history of a particular specimen.From the discussion above and the results presented in Figures 14  and 15, obtained testing a relatively wide variety of systems and environments, it is evident that the time integral of the inverse of the low-frequency impedance modulus demonstrates a stronger correlation with the final appearance of the metal under the coating rather than a single impedance value.This was proven to be valid regardless of the duration of the test and of the composition of the test electrolyte; because it considers the evolution of the system and not only its initial or final conditions.By incorporating the degradation history, the proposed provides a more comprehensive evaluation of the metal degradation.Thus, this procedure suggests that acquiring a series of EIS spectra as a function of time and using the low-frequency impedance modulus and calculate the time integral of its reciprocal to use a numerical CI that accounts for the coating degradation history during a specific test and allows direct comparison of test results obtained from different environments, test times and coatings formulations.

| CONCLUSIONS
In this work, AA and cold-rolled mild steel protected by organic coatings with varying PVC were exposed to selected aggressive environments for different periods of time.Two main behaviours in EIS responses depending on the PVC were observed.For low PVC coatings, the EIS response is generally similar in all environments.Low PVC systems generally maintained a high impedance indicative of an intact and corrosion resistance coating whereas high PVC systems showed lower values and a second-time constant and thus were less protective.
Despite the variation in the protective performance of the organic coatings observed in the different tests, it appears that a more accurate performance ranking can be achieved by considering the time integral of Z 1 | | LF during the corrosion test rather than relying only on the single value of impedance at the beginning or at the end of the test.This is because the proposed indicator keeps into account the exposure time, hence the degradation history.When evaluating a specimen, the ranking based on the proposed CI demonstrates a stronger correlation with the metal compared to solely relying on the initial and final impedance modulus.Additionally, integrating the time effect might be beneficial to compare specimens that have been tested for different durations of exposure.

F I G U R E 1
Experimental matrix.[Color figure can be viewed at wileyonlinelibrary.com]BONGIORNO ET AL.

F
I G U R E 3 EIS spectra measured at the early stages (<6 h) of the tests: AA (a, c, e) and steel (b, d, f) supporting coatings with different PVC.EIS, electrochemical impedance spectroscopy; PVC, pigment volume concentration.[Color figure can be viewed at wileyonlinelibrary.com]

F I G U R E 6 7
Visual appearance of the coating and of the metal surface of the specimens analysed in Figure 4: AA with 30% PVC in 5% NaCl (a, b), steel with 40% PVC in 5%NaCl (c, d), AA with 40% PVC in prohesion electrolyte (e, f) and steel with 40% PVC in prohesion electrolyte (h, j).PVC, pigment volume concentration.[Color figure can be viewed at wileyonlinelibrary.com]Time evolution of impedance spectra measured from specimens with low PVC coatings.PVC, pigment volume concentration.[Color figure can be viewed at wileyonlinelibrary.com]

8 F
Visual appearance of the coating and of the metal surface of the specimens analysed in Figure 6: AA with 0% PVC in 5% NaCl (a, b), AA with 10% PVC in 5% NaCl (c, d), AA with 0% PVC in prohesion electrolyte (e, f) and steel with 0% PVC in prohesion electrolyte (h, j).PVC, pigment volume concentration.[Color figure can be viewed at wileyonlinelibrary.com]I G U R E 9 EIS measurement on AA and steel for 0% PVC coating in 5% NaCl (a, c) and prohesion electrolyte (b, d) at initial time (a, b) and after 48 h (c, d).EIS, electrochemical impedance spectroscopy; PVC, pigment volume concentration.[Color figure can be viewed at wileyonlinelibrary.com]

F
I G U R E 10 Samples surface appearance after the test in 5% NaCl and prohesion electrolyte (a) and after coating stripping (b).PVC, pigment volume concentration.[Color figure can be viewed at wileyonlinelibrary.com]

F
I G U R E 11 Samples surface appearance after the test in 5% NaCl and 0.1 M NaOH and 5% NaCl and 0.1 M HCl electrolyte (a) and after coating stripping (b).PVC, pigment volume concentration.[Color figure can be viewed at wileyonlinelibrary.com]

F
I G U R E 12 Steel (a) and AA (b) specimens ranked according to ImageJ binarization analysis.[Color figure can be viewed at wileyonlinelibrary.com]

F I G U R E 13
Binarized images and of steel (a) and AA (b) specimens.[Color figure can be viewed at wileyonlinelibrary.com]

F
I G U R E 14 Correlation between ranking based on visual appearance and ranking based on low-frequency impedance modulus measured at initial (a) and final time (b) of exposure.[Color figure can be viewed at wileyonlinelibrary.com]F I G U R E 15 Correlation between ImageJ binarization ranking and corrosion indicator.[Color figure can be viewed at wileyonlinelibrary.com] Rankings of steel specimens.Rankings of AA specimens.
T A B L E 2