2017 Volume 58 Issue 12 Pages 1687-1694
The early stages of the degradation of thick epoxy-coated steels used for water ballast tanks were monitored by electrochemical impedance spectroscopy (EIS). 200-μm-thick epoxy-coated steels were exposed to accelerated corrosion tests containing humid air at 70℃ for 246 days before immersion in a NaCl solution at 70℃ for 79 days. They were removed from the corrosion test chamber during the test so that EIS measurements could be performed in order to determine the coating resistance (Rf) and the coating capacitance (Cf). In addition, the water volume fraction (XV) and the diffusion coefficient (DH2O) in the coating were determined from the Cf vs. t1/2 curves during the water desorption process. These parameters were compared with the degradation morphology of the coated steel after a pull-off test. Only “physical aging” of coating was observed for first 80 days. After 190th day, Rf decreasing, Cf and XV increasing started. Underlying corrosion also initiated between 80th and 190th day. On the other hand DH2O did not indicate the clear change associated with the degradation. The changes in Rf, Cf, and XV were closely correlated with the progress of the degradation of the coated steels.
The structures of steel ships are covered with organic coatings for corrosion protection. The internal area of water ballast tanks (WBTs), which are used to stabilize ships by filling seawater, is one of the most corrosive environments in commercial cargo ships. In particular, there is considerable WBT coating degradation at the inner side of the upper decks or walls adjacent to the heated liquid tank as they are exposed to high temperatures1). Epoxy coatings are usually used for the corrosion protection of WBT structures, and these coatings do not contain any sacrificial pigments or inhibitors. The anti-corrosion performance of the ship's protective coating is supported by its barrier property against the permeation of oxygen, water, and salt. Therefore, inspecting the coating degradation with respect to its barrier property is important in preventing a structure of steel ship from underlying corrosion.
Electrochemical impedance spectroscopy (EIS) is used as a nondestructive and quantitative method to evaluate this degradation. Coating degradation has been frequently evaluated by the decrease in coating resistance2) (Rf), which is related to the coating barrier property. However, it might be difficult to measure Rf even after the onset of coating degradation and underlying steel corrosion, particularly for thick coatings (heavy-duty coatings). In our previous study on a 200-μm-thick epoxy coating3), the coating capacitance (Cf) clearly changed due to the increase of water uptake in coatings in the degradation process. This is because the dielectric constant of water (≈80) is considerably larger than that of the dry coating (usually <10). Several other studies4,5) also proposed the evaluation method using Cf or a constant phase element (CPE) that is frequently used instead of Cf in the curve fitting of EIS spectra. Water uptake is frequently estimated using the Brasher–Kingsbury equation6) as the water volume fraction (XV) in coating.
Water diffusion behavior through the coating might be important in evaluating the coating degradation. The diffusion coefficient of water (DH2O) was estimated by EIS7–13). There are a few studies9–11) on the relation between DH2O and the coating degradation. It was reported that the coating degradation due to UV irradiation increased DH2O by approximately 2.5 times9). Conversely, corrosion tests by wet and dry using the NaCl and ionic liquids showed a decrease in Rf and blisters were observed; however, DH2O did not change significantly10). The change in DH2O may depend on the degradation process. Only Hinderliter et al.11) have examined epoxy-coated Al alloys in the thermal cycling from room temperature (RT) up to 70℃ for approximately 40 days.
In this study, water uptake parameters such as XV and DH2O of coated steel, were monitored during an accelerated corrosion test by EIS in order to examine the feasibility of these parameters as an index for the coating degradation of WBT.
Sand-blasted mild steel panels (JIS G3101 SS400, 200 mm × 100 mm × 3.2 mm) were spray coated with a commercial paint of two-packs epoxy-type containing talc and mica as body pigments so that the total thickness of the dry coating was within 200 ± 50 μm. They were coated twice in order to minimize initial defects like pinholes. The inner epoxy coating had a yellow color pigment, whereas the outer epoxy coating had a gray color pigment. Coated panels were dried for more than one month at RT. Three test panels, TP0, TPA, and TPB were prepared in the same manner. The average dry coating thicknesses of TP0, TPA, and TPB were measured to be 211, 201, and 198 μm, respectively, by the electromagnetic thickness gauge (Elcometer 456). These panels were used for three following experiments. Each experiment was carried out by using only one test panel, respectively. TP0 was not exposed to the corrosion test so as to be used to investigate Rf, Cf and the water uptake parameters (XV and DH2O) of the intact test panel by EIS measurement. TPA and TPB were exposed to the corrosion test. TPA was employed for measuring Rf, Cf and the water uptake parameters of the degraded test panel by EIS measurement. TPB was used to measure the adhesion strength by the pull-off test and confirm the occurrence of the coating delamination and underlying steel corrosion.
2.2 Corrosion testThe degradation of the WBT coating was examined at 70℃ by using TPA and TPB. To simulate a high-temperature environment in ballast tanks, test panels were placed in a chamber, which is schematically shown in Fig. 1. The bottom of the chamber was filled with tap water. The water was heated using an electric heater so that the air-phase temperature was maintained at 70℃. The water temperature was maintained at 70℃–75℃, thus the relative humidity (RH) of the air phase was estimated to be nearly 100%. The testing temperature was close to the maximum one in the approval test of the performance standards for protective coatings applied in WBT14).
Schematic of the corrosion test chamber.
The accelerated corrosion test comprised two stages. First, the panels were exposed to air in the chamber described above at 70℃ and 100%RH for the first 246 days. Subsequently, they were immersed in a 0.05-M NaCl solution at 70℃ for the next 79 days (from 247th to 325th day) so as to introduce further coating degradation. TPA and TPB were removed from the corrosion environment on the 80th, 190th, 246th, and 325th days to perform each experiment.
2.3 EIS measurementsThe EIS measurements of the coated steel were carried out using a two-electrode configuration comprising the coated steel panel (working electrode) and a Pt wire counter electrode. In the measurement, an acrylic cylindrical vessel with an effective area of 49 cm2 was attached to the coating surface using a silicone rubber O-ring and filled with electrolytic solutions. A lock-in amplifier (LI5640, NF Corp.) was used for the measurement. All EIS measurements were performed in a Faraday cage at RT to reduce exogenous noise. The frequency and amplitude were 1 kHz–1 mHz with 100 mVrms for the EIS spectrum measurement and only 1 kHz with 50 mVrms for the Cf monitoring described later, respectively.
2.4 Water uptake and diffusion coefficientWater adsorption and desorption were performed to determine XV and DH2O. Dilute and concentrated LiCl solutions were employed to control water uptake of coating because water activity (aw) can be widely varied by the concentration. In this study, water was absorbed in a 0.01-M LiCl solution and desorbed in a 10-M LiCl solution. According to the literature, the aw is 1 for the 0.01-M LiCl solution and 0.3 for the 10-M LiCl solution15).
The time change of XV is given by Fick's second law:
| \[ \frac{\partial X_{\rm V}}{\partial t} = D_{\rm H2O} \frac{\partial^{2} X_{\rm V}}{\partial x^{2}}, \] | (1) |
| \[ \begin{split} X_{\rm V} (t) & = X_{\rm V0} + (X_{\rm Vi} - X_{\rm V0}) \frac{8}{\pi^{2}}\\ &\quad \times \sum_{n=0}^{\infty} \left\{ \frac{1}{(2n + 1)^{2}} \exp \left( - \frac{(2n + 1)^{2}\pi^2 D_{\rm H2O}}{4L^{2}}t \right) \right\}, \end{split} \] | (2) |
The relation between Cf and XV is given by the Brasher–Kingsbury equation6):
| \[ X_{\rm V} = 100 \times \frac{\log(C_{\rm f}/C_{\rm f0})}{\log \ 80}, \] | (3) |
Bode plots for TPA at aw = 1 saturation before the corrosion test and after the 80th, 190th, 246th, and 325th days of exposure to the corrosion test.
| \[ C_{\rm f} = \sin \theta_{\rm 1kHz} (2\pi f|z|_{\rm 1kHz})^{-1}, \] | (4) |
XV and DH2O can be determined in either of the water absorption or desorption process. If the water absorption process is employed, a coating must be completely dried in advance. Conversely, if the water desorption process is used, a water-absorbed coating to be saturated must be employed. These processes are time-consuming, particularly for thick coatings. In this study, XV and DH2O were determined in the process of water desorption just after the corrosion test to save time since water absorption was nearly completed in the tested corrosion environments.
In the measurement, the acrylic vessel on TP0 and TPA was firstly filled with 0.01-M LiCl solution (aw = 1) and the Cf was monitored until the Cf became apparently constant value, Cf-sat (Cf saturated in 0.01-M LiCl solution). The EIS spectrum of this condition was also obtained to determine Rf saturated in 0.01 M-LiCl solution (Rf-sat). After that, the solution in the acrylic vessel was replaced by a 10-M LiCl solution (aw = 0.3) for water desorption. Cf at each time, Cf(t) was monitored at approximately 3 min interval.
The monitored Cf(t)s were analyzed using eq. (5), which had been developed from eq. (2) and (3) in the water desorption process.
| \[ \begin{split} {\rm Log}(C_{\rm f}(t)) & = {\rm Log}(C_{\rm f0}) + {\rm Log}(C_{\rm f \text{-} Sat}/C_{\rm f0}) \frac{8}{\pi^{2}}\\ &\quad \times \sum_{n = 0}^{\infty} \left\{ \frac{1}{(2n {+} 1)^{2}} \exp \left( -\frac{(2n {+} 1)^{2} \pi^{2} D_{\rm H2O}}{4L^{2}}t \right) \right\}, \end{split} \] | (5) |
In particular, TPA was removed from the chamber on the 0th (before the corrosion test), 80th, 190th, 246th, and 325th days of the corrosion test. After Cf became Cf-sat (approximately 500 h for 0th and within 120 h for 80th, 190th 246th and 325th day in the corrosion test) in 0.01-M LiCl solution. Cf(t) was monitored in 10-M LiCl solution immersion for more than 130 h.
The Rf, Cf, XV-sat and DH2O for TP0 was also evaluated. TP0 was immersed in a 0.01-M LiCl solution for a longer time (approximately 1,000 h) because the TP0 was not exposed to the corrosion test (i.e., it was completely dry). Cf(t) of TP0 was monitored in 10-M LiCl for approximately 450 h.
In the corrosion test, the change of parameters from EIS measurement seemed to be caused by not only degradation but also “physical aging” due to heating in 70℃. To confirm that such changes were attributed to the physical aging, the influence of heating on Rf-sat, Cf-sat, XV-sat and DH2O was examined by using TP0. After the EIS measurement described above, TP0 was exposed to dry air at 70℃ for 100 h and subsequently immersed in distilled water at 70℃ for 24 h for water saturation. Then, it was immersed in a 0.01-M LiCl solution at RT for approximately 150 h to confirm the saturation of water in the coating. Rf-sat, Cf-sat, XV-sat and DH2O were determined by the same procedure in 0.01-M and 10-M LiCl solutions for the EIS spectrum measurement and the Cf(t) monitoring, respectively.
2.5 Adhesion testAdhesion between the coating and substrate was examined by a pull-off method using the TPB panel. Before each test, TPB was dried for more than 2 weeks17) at RT to eliminate the influence of absorbed water in coating. An aluminum alloy cylindrical dolly (diameter = 20 mm) was attached to the TPB surface with instant adhesives, and the dolly was mechanically pulled off. The adhesion strength and peeled-off surfaces (interfaces of the coating/dolly, coating/steel, and inner/outer coatings) were examined. The delaminated area affected by the pull-off test was recoated with the same epoxy paint and dried. The partially recoated TPB was carried back to the chamber, and the corrosion test was continued. In the next pull-off test, another part of the coating surface was examined.
Figure 2 shows the Bode plots of TPA (exposed sample) measured under a water-saturated condition in a 0.01-M LiCl solution (aw = 1). The EIS spectra until 246 days indicated a typical capacitive behavior of the phase shift θ being nearly −90° in the frequency region higher than 10 Hz and a gradual decrease of θ below 10 Hz, illustrating the contribution of the resistive components. Even the intact coating (0 day) indicated similar behavior. Some study18,19) described that the distribution of Rf in the vertical direction to coating surface mainly caused such time dispersion in the spectra. Rf distribution will be due to the inhomogeneity of coating. These studies eventually reduced EIS model to a single parallel circuit of Rf/CPEf. However, a single Rf/CPEf model was not enough to represent EIS spectra obtained in our study. More degree of freedom will be required since the spectra were obtained wide frequency range like 1 kHz–1 mHz. As previously reported3), the EIS spectra of the thick-coated steel at an early stage of degradation can be represented by a multi-time constant circuit, as shown in Fig. 3(a). In this equivalent circuit, the Rf/CPEf parallel combinations are distributed in the vertical direction to the coating surface. In addition, they could be distributed in the horizontal direction, as shown in Fig. 3(b). However, the multi-parallel combinations are rewritten by an equivalent circuit of one time constant. Thus, the deviation from the capacitive behavior in the low-frequency range will be explained by the equivalent circuit of the multi-series combinations (Fig. 3(a)).
Multi-time constant equivalent circuit model for coated steel with (a) a distribution in the depth direction and (b) a distribution in the horizontal direction.
The symbols and solid lines in Fig. 2 indicate the experimental data and the curve-fitting results shown in Fig. 3 (a). The fitting was attempted by the equivalent circuits of two to eight series combinations of Rf/CPEf, where it was assumed that CPEf was constant and only Rf was distributed. This is because Rf more significantly changes with water uptake and temperature3) than CPEf. The EIS spectra were found to be well fitted by five or more series combinations. Thus, in this study, the fitting was performed using five series combinations, Rf1/CPEf1–Rf5/CPEf5, and the total coating resistance Rf-sat was defined as the sum of Rf1–Rf5.
The EIS spectra for TPA before testing and after the 80th, 190th, and 246th days of exposure to the corrosion test were well fitted by the circuit of Fig. 3(a). On the other hand, the circuit did not provide good fitting for 325th days. This is attributed to the formation of a macroscopic path through the coating and the delamination due to degradation. The pull-off test also indicated the occurrence of the coating delamination and underlying steel corrosion. The details of the pull-off test are described in the adhesion test subsection of the Results and Discussion section. For 325 days, the transmission line type equivalent circuit20) was adopted for curve fitting. The schematic of the cross-section of the delaminated coated steel and one-dimensional, distributed equivalent circuit is shown in Fig. 4. Interfacial impedances such as R*ct (charge transfer resistance) and C*dl (double layer capacitance) will appear because Rf decreased largely by the macroscopic water path through coating. R*s corresponds to the solution resistance of thin water layer beneath coating.
Transmission line type equivalent circuit used for the electrochemical impedance spectroscopy (EIS) spectrum of TPA on the 325th day of exposure to the corrosion test.
The obtained fitting parameters of the equivalent circuits are tabulated in Table 1, where CPEf-sat-T and CPEf-sat-p are a radix parameter and power parameter of CPE for water-saturated coating in 0.01-M LiCl solution, respectively. Commercial software (Scribner Associates, Inc., ZView) was used for fitting. Rf-sat can be obtained with a small standard variation (Error: 2%–7%) for all measurements. CPEf-sat–T corresponds to Cf-sat when CPEf-sat–p is unity. The obtained value of CPEf-sat–T is slightly larger than the Cf-sat value calculated from the impedance at 1 kHz (Table 3) that is used for measuring XV-sat and DH2O because the CPEf-sat–p values (0.94–0.95) are slightly smaller than unity. However, they indicated a similar trend for the changes in the exposure time. CPEf-sat–p will not change with exposure time, as shown in Table 1.
| Exposure time | Rf-sat | CPEf-sat-T | CPEf-sat-p |
|---|---|---|---|
| (day) | × 1011 Ω･cm2 | × 10−11(Ωcm2)−1･sp | |
| 0 | 10 | 4.2 | 0.94 |
| 80 | 35 | 4.3 | 0.95 |
| 190 | 21 | 4.7 | 0.95 |
| 246 | 9.9 | 5.1 | 0.95 |
| 325 | 0.084 | 7.7 | 0.94 |
| Cf-sat | Cf0 | XV-sat | DH2O |
|---|---|---|---|
| × 10−11 F･cm−2 | × 10−11 F･cm−2 | vol% | × 10−9 cm2･s−1 |
| 2.5 | 1.9 | 5.9 | 1.0 |
| Exposure time | Cf-sat | Cf0 | XV-sat | DH2O |
|---|---|---|---|---|
| (day) | × 10−11 F･cm−2 | × 10−11 F･cm−2 | vol% | × 10−9 cm−2･s−1 |
| 0 | 2.6 | 2.0 | 5.6 | 1.3 |
| 80 | 2.7 | 2.3 | 4.3 | 1.1 |
| 190 | 3.1 | 2.2 | 8.2 | 0.70 |
| 246 | 3.2 | 2.2 | 8.7 | 1.0 |
| 325 | 4.6 | 2.4 | 15 | 1.2 |
Figure 5 shows a plot of Cf(t) vs. t1/2 for TP0 (intact sample) in the water desorption process in a 10-M LiCl solution (aw = 0.3) at RT. The solid and dotted lines indicate the measured Cf(t) and the fitting result by eq. (5), respectively. Cf(t) slightly fluctuated being synchronized with considerably small temperature changes within ±1℃ (25.7–26.8℃). Thus, the Cf(t) was compensated for by an equation obtained from the correlation between the saturated Cf (Cf-sat and Cf0) and the temperature in order to fit eq. (5) to measured Cf(t). The compensated Cf(t) values (corresponding to Cf at average temperature during water desorption process) are plotted in Fig. 5. It is evident that water diffusion through the coating obeys Fick's law. XV-sat and DH2O were obtained to be 5.9% and 1.0 × 10−9 cm2s−1 by using eq. (5), respectively. All parameters obtained are tabulated in Table 2. The obtained DH2O was close to that obtained for epoxy resin (0.1–9 × 10−9 cm2s−1), as reported by the gravimetric12,13,21) and EIS7,8,10–13) methods.
Plots of Cf obtained from TP0 against the square root of time in the water desorption process in a 10-M LiCl solution (aw = 0.3).
In the corrosion test, any visible sign of coating degradation without slight color changes were not observed on both the TPA and TPB surfaces after the first stage of the corrosion test when they had been exposed to air at 70℃ and 100%RH for 246 days. Even at the end of the second stage of the corrosion test where they had been immersed in a 0.05-M NaCl solution at 70℃ for 79 days (from the 247th to 325th day of the corrosion test), as shown in Fig. 6, the distinct coating degradation was not confirmed with by sight, although considerably small blisters were observed with an optical microscope. In the cross-sectional view of the coating, several cracks were found, as shown in Fig. 7.
Optical image of TPA coating surface after an exposure of 325 days to the corrosion test.
SEM-secondary electron images of the cross-section of TPA after an exposure of 325 days to the corrosion test. Arrows in the figure indicates cracks observed after the corrosion test.
Figure 8 shows the Cf(t) changes of TPA (exposed sample) before the corrosion test and after the 80th, 190th, 246th, and 325th days of exposure to the test. The solid lines indicate the fitting results. It is obvious that the water diffusion behavior for TPA before the corrosion test and after an exposure of 80 days obeys Fick's law. On the other hand, deviation from Fick's law was observed after the 190th, 246th, and 325th days and was particularly pronounced after 325 days. The deviation will be attributed to the coating degradation. XV-sat and DH2O were determined from the solid lines shown in Fig. 8, although a large deviation is observed for 325 days. They are tabulated in Table 3. XV-sat can be more accurately determined because Cf(t) becomes constant at the last stage of the Cf monitoring. On the other hand, some uncertainty for the DH2O values on the 190th, 246th, and 325th days is expected because of the deviation from Fick's law.
Plots of Cf obtained from TPA against the square root of time in the water desorption process in a 10-M LiCl solution (aw = 0.3) before the corrosion test and after the 80th, 190th, 246th, and 325th days of exposure to the corrosion test.
Photos of TPB (exposed sample) after the pull-off test are shown in Fig. 9 along with the adhesion strength. For an exposure of 80 days, the adhesion strength remained a higher value than 5 MPa, which is one of the criteria for degradation in the testing standatds14,17,22). However, the strength decreased to 4.0 MPa on the 190th day. It further decreased to 2.0 MPa on the 246th day. It can be observed that the adhesion between the coating and substrate steel was weakened due to exposure to the corrosion test, particularly after 190 days.
Optical images of peeled-off surface on TPB by pull-off tests after 80th, 190th, 246th, and 325th days of exposure to the corrosion test with adhesion strengths and illustrations explaining peeled-off section.
On the peeled-off surface in Fig. 9, there are four different areas in the photos, which vary in color: black (corrosion product), dark gray (substrate steel), light gray (outer coating), and yellow (inner coating). These areas are schematically illustrated in Fig. 9. The black color indicates an area where the underlying steel corrosion occurred, and the remaining colors indicate that no corrosion has occurred. Adhesion between the coating and the substrate steel was completely lost in the black area and is weakened in the dark gray area.
The underlying corrosion was not found after 80th days, and it was clearly observed after the 190th, 246th, and 325th days. It is obvious that steel corrosion commenced between the 80th and 190th days of the corrosion test. It is evident from the photos for the 190th and 246th days that the weakest adhesion area (dark gray) surrounded the corrosion area (black), and the dark gray area significantly increased between the 190th and 246th days, although the corroded area hardly changed. The dark gray area may act as the cathode and the black area may act as the anode. The underlying corrosion may initiate at defect and/or delaminated sites due to the coating degradation. Subsequently, the cathode area may widen around the anode as the degradation progresses. Since the coated steel was exposed to air at 70℃ and 100%RH until the 246th day, the cathode may be limited to the surrounding area of the anode. However, at the later stage (from the 247th to 325th day), it was immersed in a 0.05-M NaCl solution at 70℃. Thus, for 325 days, the entire surface can act as the cathode, leading to the underlying steel corrosion nearly over the entire steel surface.
3.4 Correlation between the parameters and degradation of coated steelThe changes in (a) adhesion strength of TPB, (b) Rf-sat, (c) Cf-sat, (d) XV-sat, and (e) DH2O of TPA during the corrosion test are summarized in Fig. 10. As mentioned earlier, it is expected from the adhesion strength that the coating will work well as a barrier layer at least until the 80th day. It was also confirmed from the surface morphology in the pull-off test that no distinct underlying steel corrosion occurred for an exposure of 80 days (Fig. 9). On the other hand, Rf-sat for 80 days increased 3.5 times higher than that before the test and XV-sat decreased from 5.6 vol% to 4.3 vol%. These changes were opposite to those expected by the coating degradation. This increase in Rf-sat and decrease in XV-sat may be attributed to the “physical aging” of the coating by heating. Hinderliter et al.11) reported that DH2O decreases by heating up to 70℃ and that such a change is caused by “physical aging” due to the rearrangement of polymer chain. In our study, DH2O decreased from 1.3 to 1.1 × 10−9 cm2·s−1 after an 80 days exposure to the corrosion test. Physical aging will lead to an increase in Rf-sat and decrease in XV-sat due to the reduction in free volume23). Figure 11 shows the Bode diagrams of TP0 (intact sample) before and after the heating in dry air and subsequent water immersion for 124 h at 70℃ measured in a 0.01-M LiCl solution. The influence of heating appeared especially at lower frequency region. The impedance and phase shift were slightly increased by the heating. Curve fitting was attempted to obtain Rf-sat value using the equivalent circuit of Fig. 3(a). The fitting results are presented by solid lines in Fig. 11. The obtained Rf-sat (29 × 1011 Ωcm2) was 3.5 times larger than that before heating (8.4 × 1011 Ωcm2). XV-sat and DH2O of TP0 was obtained from the Cf monitoring in the subsequent water desorption (10-M LiCl) process by using eq. (5), and it was decreased from 5.9 vol% to 4.4 vol% and from 1.0 × 10−9 to 0.79 × 10−9 cm2s−1 after heating, respectively. These changes of TP0 due to heating for 124 h at 70℃ were nearly identical to those of TPA after an exposure of 80 days to the corrosion test (100%RH air at 70℃). Thus, the changes in Rf-sat and Cf-sat on the 80th day will be attributed to physical aging, and the changes due to coating degradation will be negligibly small until the 80th day of the corrosion test.
Plots of (a) adhesion force, P of TPB by pull-off test, (b) coating resistance Rf-sat and (c) coating capacitance Cf-sat (c) at aw = 1 saturation, (d) water volume fraction XV-sat and (e) diffusion coefficient of water DH2O of TPA before corrosion test and after the 80th, 190th, 246th, 325th days of exposure to the corrosion test.
Bode plots for TP0 at aw = 1 saturation before and after 70℃ heating of 124 h.
The decrease in Rf-sat and increase in Cf-sat were clearly observed after 190 days, where a trace amount of corrosion product was detected at the interface of the coating/steel (Fig. 9). The adhesion strength also decreased to less than the criterion value (5 MPa) at the 190th day. These changes of Rf-sat and Cf-sat continued from the 190th to 325th day. At end of the corrosion test, The Rf-sat value for 325 days was approximately 400 times smaller than that for 80 days, and the Cf-sat value was 1.7 times larger than that for 80 days. However, the Rf-sat value still maintained approximately 1010 Ωcm2 at 325 days as shown in Table 1. In the case of the thick-coated steels, despite a sufficiently large Rf-sat value, the underlying steel corrosion, although a considerably small amount, commences.
The XV-sat value was 5.6 vol% before the corrosion test, and it decreased to 4.3 vol% on the 80th day. This decrease is attributed to physical aging as discussed earlier. The XV-sat value increased twofold at the 190th day (8.2 vol%) in comparison with the 80th day (4.3 vol%), and it further increased with time as 8.7 vol% at the 246th day before finally becoming 15 vol% on the 325th day. DH2O value was 1.3 × 10−9 cm2s−1 before the corrosion test, and physical aging will also decrease DH2O to 1.0 × 10−9 cm2s−1 on the 80th day. However, DH2O after 190th day did not indicate the clear change with the exposure time.
The decrease of Rf-sat and the increases of Cf-sat and XV-sat were well correlated to the progress of the degradation of the coated steel. Thus, these parameters are extremely useful for evaluating the early stages of the degradation of steels coated with thick polymer films. Among them, Rf-sat exhibited the greatest change by the degradation, indicating that it could be the most efficient parameter for evaluating the degradation in a Faraday cage in a laboratory. However, in actual WBTs, it is quite difficult to measure Rf-sat appearing in the low-frequency range due to exogenous noise. In addition, to obtain an accurate Rf-sat, curve fitting using the equivalent circuit is necessary at the early stage of the degradation. Conversely, although the change in Cf-sat is smaller than that of Rf-sat, it is quite easy to measure Cf-sat because it can be determined from the impedance at a certain frequency (1 kHz employed in this study) in the high-frequency region without any interference from exogenous noise.
The early stages of the degradation of the 200 μm thick epoxy-coated steel for water ballast tanks were monitored by EIS. In addition, the water diffusion process in the coating was examined. The obtained parameters: Rf-sat, Cf-sat, XV-sat, and DH2O were compared with the degradation morphology during the corrosion test, which comprised exposure to air at 70℃ and 100%RH for 246 days prior to a subsequent immersion in a 0.05-M NaCl solution at 70℃ for 79 days. The following conclusions were drawn:
(1) The EIS spectra for the intact and slightly degraded coatings can be expressed using an equivalent circuit comprising multi-series combinations of parallel Rf/CPEf distributed in the depth direction of the coating. The coated steel rapidly deteriorated by its subsequent immersion. The EIS spectrum for the deteriorated coated steel can be explained by a transmission line type equivalent circuit.
(2) DH2O was obtained to be 0.7–1.3 × 10−9 cm2s−1. The DH2O did not change in the coating degradation process. The XV-sat was obtained to be 4.3 vol%–5.9 vol% for the intact coating, and it increased to 8.7 vol% on the 246th day and 15 vol% during its subsequent immersion. The parameter XV-sat increases as the deterioration progresses.
(3) The degradation of the coated steel is closely correlated with the changes in Rf-sat and Cf-sat. These parameters are considerably effective in monitoring the early stages of the degradation before the appearance of visible blisters. Especially, the Cf-sat is a more convenient parameter to evaluate the degradation in actual WBTs.
The authors are grateful to Chugoku Marine Paints, LTD. for the preparation of test panels. This study was supported by JSPS KAKENHI Grant Number 25820427.
