2016 Volume 56 Issue 11 Pages 2029-2036
The degradation of epoxy coating for ballast tanks was examined by electrochemical impedance spectroscopy (EIS). EIS spectra were measured in LiCl solutions using approximately 200-µm-thick epoxy-coated steel in the temperature range 30–70°C. To clarify the effect of water activity (aw), three different LiCl concentrations were used: 0.01 M (aw = 1), 5 M (aw = 0.7), and 10 M (aw = 0.3). The EIS spectra in 0.01 M LiCl solution revealed capacitive behavior over the whole 1 kHz–1 mHz frequency range at 30°C, whereas distinct resistive behavior arose in the low-frequency region above 40°C. Increasing temperature and water activity slightly increased the epoxy film capacitance (Cf) and drastically decreased the film resistance (Rf). The long-term monitoring of the coating degradation by EIS was also performed under thermal cycling in the 0.01 M LiCl solution, from which it was found that Rf decreased and Cf increased gradually with increasing cycle number. Observation by scanning electron microscopy after 250 thermal cycles revealed that several defects (voids and cracks) that could act as water absorption sites formed inside the coating and at the coating/steel interface. The epoxy coating was removed from the steel substrate and a trace amount of Fe3O4 was confirmed to have formed over the whole surface of the steel substrate. EIS is extremely useful for monitoring the degradation of thick polymer coatings under thermal cycling.
Commercial ships require good corrosion protection since they are exposed to severe corrosive environments. Organic coatings and cathodic protection are major tools for preventing the corrosion of ships. The water ballast tank (WBT) is one of the compartments that might be exposed to the most corrosive environment in a commercial cargo ship. The WBT is equipped to make an empty ship stable by filling it with sea water, and sea water in the WBT is discharged overboard during cargo loading. Thus, the WBT materials are subjected to both immersion in sea water and high relative humidity. The coated steels employed in the WBT are severely degraded at the back side of the upper deck and the back side of the heated liquid tank, which are exposed to higher temperatures.1) In addition, cathodic protection by a sacrificial anode is not available for the back side of the upper deck because it is not immersed in the ballast water during service periods. The WBT is also a part of the ship structure and plays an important role in the strength of the ship. Therefore, evaluating the protective performance of the WBT coating is necessary for maintaining structural safety.
Current inspection of WBT coatings usually depends on visual observation. This is a partially quantitative evaluation method and only surface degradation, such as blisters, surface rust, and cracks, can be found. To improve the maintenance of WBT coating systems, a quantitative evaluation method is necessary for the earlier stages of degradation. EIS is a quantitative method that may be used to evaluate the protective performance of coatings. EIS was employed to study WBT-coated steel in a research project called SR-201, which was conducted by the Shipbuilding Research Association of Japan and ran from the 1980s to 1990.2) Since an increase of tanδ was observed, especially at low frequency, when blisters appeared, it was concluded at that time that tanδ is suitable for the evaluation of the corrosion performance of WBT coatings. tanδ is given by (2πfRC)−1 for a circuit comprising a resistance R and a capacitance C connected in parallel, where f is the frequency. An increase of the tanδ indicates a decrease of the coating resistance.
Significant changes in WBT coating regulation occurred in recent years. The use of tar epoxy coatings was prohibited because of the harmful nature of coal tar, and epoxy coatings were employed as a substitute for tar epoxy since then. In addition, the Performance Standard for Protective Coatings (PSPC) was adopted by the International Maritime Organization (IMO) in 2006.3) This standard regularized many factors concerning WBT coatings, in particular the thickness of the WBT coating increased relative to that in the 1980s according to the PSPC. The electrical resistance of WBT coating systems became significantly higher as a result of these regulation changes.
EIS spectra were principally interpreted with an equivalent circuit comprising the parallel connection of a resistance, Rf, and a capacitance, Cf. In order to express the variation of the impedance across the surface,4) constant phase elements (CPE) are often used instead of pure capacitances. The protective performance of the coating is usually evaluated by the Rf value, since Rf is assumed to relate to the barrier properties of the coating. However, for thick coatings such as heavy-duty coatings, Rf may be difficult to measure by EIS because of its high value, even if delamination of the coating and corrosion of the underlying steel have initiated. However, some reports of the application of EIS to the evaluation of thick coatings exist.5,6,7) Hattori et al.5) carried out a long-term immersion test on a heavy-duty coating in NaCl solution and reported that Rf was decreased by the onset of blisters and surface rust. Westing et al.6) tested thick epoxy coatings in NaCl solution and reported that Rf remained high (>100 MΩ cm2), but the CPE parameters changed at the initiation of corrosion of the underlying steel. These different results might result from the difference in the degree of coating degradation, or from the difference in the properties of the coating materials. To establish EIS as a quantitative evaluation method, it is necessary to clarify the correlation between degradation of heavy-duty coatings and EIS characteristics.
Accelerated corrosion tests such as immersion tests, salt spray tests (SST), and cyclic corrosion tests (CCT) are generally employed in the laboratory for polymer-coated steels. Thermal cycling tests were first employed by Bierwagen et al.8) to accelerate the degradation of coating systems. This method was expected to correlate well with natural exposure because coatings experience temperature cycles in actual usage environments. Valentinelli et al.7) also applied a “thermal cycling” test to thick coating systems and evaluated the coating performance by EIS. They reported that phase shifs of EIS spectra undergo a small reduction at room temperature but showed a significant loss at low frequency in the temperature range 55–85°C, and that a reversible change of the impedance with thermal cycling was observed for high-performance coatings, whereas irreversible changes occurred for less protective coatings. They evaluated the coating performance by the irreversible increase in the coating capacitance.
WBT coatings are exposed to various temperature cycles, mainly due to solar heat. Thus, degradation of WBT coatings should be evaluated under thermal cycling. In the present study, the applicability of EIS as an evaluation method for the degradation of WBT coatings was investigated. First, EIS was measured at various temperatures to clarify the dependence of the EIS spectra on temperature. The effect of water activity, aw, on the EIS spectra was also studied because WBT coatings are exposed to different relative humidities. Second, WBT coatings were exposed to thermal cycling and the EIS spectra were monitored to clarify the relation between the EIS spectra and coating degradation.
Epoxy-coated mild steel panels were used in this study. The substrate panels were prepared from JIS G3101 SS400 steel plate, the chemical composition of which is shown in Table 1. The dimensions of the panel were 200 mm × 100 mm × 3.2 mm. Prior to painting, the steel surface was prepared by blast processing to grade Sa2.5 of ISO 8501-1 and roughness degree 30–75 μm. A commercial two-pack epoxy-type paint, which contained talc and mica as body pigments, was applied to the panel. Spray coating was carried out two times in order to minimize initial defects such as pinholes. The average total dry film thickness in the tested area was 187 μm, although the target was 200 μm. The back surface and edges were also coated with the same paint. The coated panel was dried for more than 1 month before the corrosion tests.
| C | Si | Mn | P | S | Fe |
|---|---|---|---|---|---|
| 0.07 | 0.01 | 0.53 | 0.022 | 0.003 | Bal. |
Figure 1 shows the schematic of experimental setup for EIS measurements of the coated steels. A glass cell was assembled on the test sample surface with a silicone rubber gasket. The effective surface area was 43 cm2. A two-electrode cell configuration was employed using a Pt counter electrode in order to measure extreme high impedance of the thick coated steel. In order to widely vary the water activity aw, test solutions of various LiCl concentrations were employed: 0.01 M (aw = 1), 5 M (aw = 0.7), and 10 M (aw = 0.3). Their water activities correspond to relative humidities of 100%, 70%, and 30%, respectively.9) EIS measurements were carried out at various temperatures. The coated steel sample was first kept at 70°C for more than 18 h to saturate the coating with water, and then held at 30°C. The temperature was increased stepwise from 30 to 70°C in 10°C increments. The sample was kept for more than 2 h at each temperature to establish equilibrium conditions. EIS was measured at each temperature with a lock-in amplifier (LA, NF Corp. LI5640). The AC input voltage was 45–100 mV RMS and the frequency range was 1 kHz to 1 mHz. The reliability of the LA was confirmed using a film capacitor with a capacitance of 220 pF, the same order of magnitude as that of the employed epoxy coating. The measurement was carried out in the order of 0.01, 10, and 5 M LiCl solutions using the same coated steel sample, before finally measuring in 0.01 M LiCl solution again to determine the extent of degradation of the coating due the increases in temperature and chloride concentration.
Schematic of the EIS measurement of the coated steel panel.
Thermal cycling tests3,7,8,10,11,12) that simulated the WBT environments were conducted to evaluate the coating performance. In the PSPC-approved test,3) coated test samples are exposed to two different conditions: (1) alternate high humidity conditions at 50 and 20°C with sea water splashes that simulate the back side of the upper deck (wave tank test) and (2) dry conditions at 70°C that simulate the region between the heated fuel oil tank and the double bottom. In the present study, in order to continuously measure the impedance of the coated steel, the coated steel was exposed to thermal cycling in 0.01 M LiCl solution. Alternate temperatures of 70°C for 20.5 h and 25°C for 3.5 h were adopted for the thermal cycling conditions. As shown in Fig. 2 schematically, EIS spectra were measured at an interval of several dozen cycles in the test solution at 25 and 70°C using the cell shown in Fig. 1. During the cycling, the coated sample was kept in electrical contact with the Pt counter electrode (surface area: approximately 6 cm2). Thus, the test sample was kept under anodic polarization during the thermal cycling.
Schematic drawing of EIS measurement procedure during thermal cycling.
After 250 thermal cycles, the coated steel was removed from the chamber, and the surface and cross section of coating were observed by optical microscopy and scanning electron microscopy (SEM). As mentioned in the experimental result section, the whole surface of the underlying steel was uniformly covered with iron rust, although high impedance values were still obtained. In order to estimate the average corrosion rate of the underlying steel, the amount of iron corrosion product present under the coating was collected and analyzed in the following way. A 5 cm2 piece was cut out of the test panel and then the coating was removed from the substrate. The corrosion product on the peeled coating was dissolved by 24 h immersion in 20% HCl solution. On the other hand, the corrosion product on the substrate was dissolved ultrasonically and chemically by 5 min immersion in 5% HCl + 0.3% inhibitor (IBIT 700AS of Asahi Chemical Co., Ltd). The collected corrosion product from the substrate was completely dissolved by addition of 20% HCl. After that, the amount of collected Fe ions was analyzed quantitatively with an atomic absorption spectrometer, AAS (AA-6800 of Shimadzu Corporation, 248.3 nm absorbance in air–C2H2 flame).
To clarify the EIS behavior of the coated steel, EIS measurements were performed at various temperatures. Figure 3 shows Bode plots for the coated steel in (a) 0.01 M, (b) 5 M, and (c) 10 M LiCl solutions. It can be clearly seen that the magnitude of the impedance (|z|) decreased upon increasing the temperature over the entire frequency range. The decrease was most significant at low frequency. The EIS spectra recorded for 0.01 M LiCl solution showed capacitive behavior over almost the whole frequency at 30°C, although the phase shift slightly decreased with decreasing frequency in the low-frequency regime (<10 Hz). Meanwhile, the resistive behavior appeared clearly above 40°C at low frequency. In 5 M and 10 M LiCl solution, resistive behavior was observed at higher temperatures, above 50°C and 60°C, respectively. Comparing |z| at the same frequency, |z| increased with increasing LiCl concentration. We will later show that this change in impedance can be attributed to differences in water activities aw, which take values of 1, 0.7, and 0.3 for 0.01 M, 5 M, and 10 M LiCl, respectively.9) Thus, the resistive behavior more clearly appeared at higher temperatures and higher water activities (lower chloride concentration).
Bode plots for the coated test panel in 0.01 M, 5 M, and 10 M LiCl solutions at various temperatures.
Figure 3 also shows Bode plots for the first and second EIS runs in 0.01 M LiCl. Both runs exhibit almost the same values over the whole frequency range. The resistive behavior observed in the temperature range 40–70°C in the first run completely disappeared at 30°C in the second run. The EIS change was reversible during heat cycling, indicating that the coating degradation due to a single thermal cycle is negligibly small.
3.2. Equivalent CircuitThe EIS spectra of thin polymer-coated steels can be explained by an equivalent circuit comprising a film resistance Rf and a film capacitance Cf connected in parallel.13) In the present study, as shown in Fig. 4(a), Cf was replaced with a CPE for the curve fitting. The fitting results using the equivalent circuit shown in Fig. 4(a) at 40 and 70°C in 0.01 M LiCl are shown in Fig. 5, where the symbols and dotted lines indicate the experimental data and the fit results, respectively. It can be seen that the EIS results obtained at 70°C were well fitted. However, the EIS results at 40°C were not successfully fitted, especially the phase shift in the low-frequency regime. Curve fitting was also performed with an equivalent circuit containing multiple time constants, as shown in Fig. 4(b). The series combination of parallel Rf(n)|Cf(n) elements (Voigt elements) means that heterogeneity in the direction perpendicular to the surface of the film coating causes the dispersion of time constants. In-plane heterogeneity that can be expressed by an equivalent circuit comprising the parallel combination of parallel Rf(n)|Cf(n) elements should be also considered, but in this case, the equivalent circuit was rewritten using a single time constant for Rf|Cf, where 1/Rf and Cf are given by sums of 1/Rf(1) + 1/Rf(2) + ··· + 1/Rf(n) and Cf(1) + Cf(2) + ··· + Cf(n), respectively. Finally, the equivalent circuit representing the in-plane heterogeneity reduced to that shown in Fig. 4(a). Thus, curve fitting using Fig. 4(b) with five time constants (n = 5) was performed because the results did not significantly differ for n > 5. The fitting result (solid line) for n = 5 is shown in Fig. 5. During this curve fitting, all CPE parameters (CPE1 to CPE5) were fixed to the same value (CPE1 = ··· = CPE5) to reduce the number of free parameters in the equivalent circuit since the dispersion of CPE(n) was smaller than that of Rf(n). As can be seen in Fig. 5, the fitting result was much better than for n = 1, especially in the low-frequency regime. The obtained parameters are presented in Table 2. CPE values are only tabulated for EIS runs between 30 and 40°C at aw = 0.7, and between 30 and 50°C at aw = 0.3, because the phase shifts for these measurements were very noisy at low frequency and a decrease in the phase shift was not observed. The film resistances (Rf(1) to Rf(5)) were widely distributed in the depth direction of the coating, ranging from 107 to 1012 Ω cm2. This might be caused by defects (pinholes, voids, and cracks) initially introduced into the epoxy film, or by the heterogeneity of water absorption sites, as discussed in the next session.
Equivalent circuit models for coated steel: (a) single time constant Rf | Cf and (b) multi-time constants (Rf(n) | Cf(n)).
Experimental Bode plots and curve-fitting results using the equivalent circuit in Fig. 4(b) for n = 1 and 5.
;%0A%09%09%09newWindow.document.open();%0A%09%09%09newWindow.document.write('<img src=%22./Graphics/56_2029_tbl_2.jpg%22>');%0A%09%09%09newWindow.document.close();%0A%09%09)
|
In the curve fitting, CPE-T and CPE-p were employed as the CPE parameters. CPE-T is an index of the capacitance, but it has a little different dimension. Thus, the film capacitance Cf was calculated from the impedance Z1kHz at 1 kHz using the equation sin(θ1kHz)/(2π103|z|1kHz). The Cf values are tabulated in Table 2. The magnitudes were similar to the obtained CPE-T values because the CPE-p values were close to unity. CPE-p is an index of the deviation from ideal capacitive behavior. CPE-T is equal to capacitance when CPE-p is unity.
Arrhenius plots of the film resistance Rf and film capacitance Cf at three different water activities are shown in Fig. 6. Increasing the temperature significantly decreased Rf and slightly increased Cf. The apparent activation energy of Rf, EA(Rf), was 187 kJ·mol−1 at aw =1230 kJ·mol−1 at aw = 0.7, and 224 kJ·mol−1 at aw = 0.3. The EA(Rf) values are similar to those reported in the other studies.14,15) The apparent activation energy EA(Cf) for Cf was 5.9 kJ·mol−1 at aw = 1, 3.9 kJ·mol−1 at aw = 0.7, and 2.3 kJ·mol−1 at aw = 0.3. The EA(Cf) values are much smaller than the EA(Rf) values. The solution resistance Rs did not appear until 1 kHz in the present study, but the apparent activation energies EA(s) of 36 kJ·mol−1 for pure water and 14 kJ·mol−1 for 1 mol/kg NaCl16) are also much smaller than those for Rf. The water activity effects on Rf and Cf at various temperatures are shown in Fig. 7. Increasing the water activity also decreased Rf and increased Cf. The increase of the Cf can be attributed to the increasing amount of water absorbed in the epoxy coating film because the dielectric constant of water (~80 in the bulk) is much larger than that of the epoxy resin (less than 10). Gravimetric measurements also confirm that water uptake indeed increases with increasing temperature.17)
Arrhenius plots of film resistance Rf and capacitance Cf of the coated test panel as a function of water activity aw.
Plots of film resistance Rf and capacitance Cf of the coated test panel versus water activity.
The water uptake volume fraction Xv of the coating was estimated using Brasher & Kingsbury’s Eq. (1).18)
| (1) |
Plots of water volume fraction in the coating versus temperature.
From the above results, the decrease of Rf and the increase of Cf with increasing temperature and water activity can be attributed to an increase in the amount of water absorbed by the film, although the variation of Rf was more remarkable (Fig. 6). The difference can be described as follows.
Schematic diagrams explaining the effects of temperature and water activity on Rf are shown in Fig. 9. Water molecules (open circles in Fig. 9) are fixed at hydrophilic groups such as OH groups and C–O groups in the epoxy. In the vicinity of these groups, water molecules (closed circles in Fig. 9) will be clustered. At low temperature and low water activity (bottom left in Fig. 9), very small numbers of water molecules will be present in the clusters. These clusters will be isolated from each other, leading to high Rf. On the other hand, at high temperature and high water activity (top right in Fig. 9), large water clusters could be formed. In addition, the free volume (dotted circle in Fig. 9) in the epoxy becomes larger due to the thermal motion of the polymer,19) and water clusters might be formed in the larger free volume, leading to the formation of continuous water clusters in the epoxy. These continuous clusters drastically decrease the Rf. From the drastic change in Rf and the slight change in Cf caused by heating, relatively small amounts of water in the epoxy may form continuous water clusters.
Schematic diagrams explaining the effect of temperature and water activity on the polymer film resistance.
During the thermal cycle test, EIS measurements were made on the coated sample in a 0.01 M LiCl test solution at 25 and 70°C; the results are shown in Fig. 10. The impedance at both temperatures gradually decreased during the cycling. The Rf value from curve fitting and the Cf value calculated from Z1kHz are plotted versus cycle number in Fig. 11. The gradual decrease in Rf and the gradual increase in Cf due to increased water absorption in the coating were clearly detected, indicating the gradual degradation of the epoxy coating. In addition, Fig. 11 reveals a good correlation between the Rf decrease and the Cf increase, meaning that coating degradation can be evaluated by monitoring not only the film resistance but also the film capacitance. For steel with thick polymer coatings, such as heavy-duty coatings, it may be difficult to measure Rf, even after the onset of coating degradation and underlying steel corrosion, because Rf for heavy-duty coatings very often appears at lower frequencies than those available in EIS measurements. Monitoring of Cf may, therefore, be more suitable for evaluating the degradation of heavy-duty coatings, because Cf can be determined at intermediate frequencies (e.g., 1 kHz).
Bode plots of the coated test panel in 0.01 M LiCl solution during thermal cycles at 70 and 25°C.
Plots of film resistance Rf and film capacitance Cf at 25 and 70°C versus the thermal cycle number.
Figure 12 shows optical observations of the coating surface after 250 thermal cycles. Many small blisters were observed all over the surface. Figure 13 shows the result of attenuated total reflection Fourier transform infrared spectroscopy (FTIR-ATR) analysis of the coating surface. The magnitude of absorbance peaks assigned to epoxy did not change greatly after 250 thermal cycles, indicating that no chemical change of the coating occurred. SEM observations of a cross section of the coating are shown in Fig. 14, where (a) was taken before the corrosion test, and (b) and (c) were after 250 thermal cycles, revealing significant damage to the coating. There were many cracks in coating layer and areas where delamination from the steel substrate had occurred (Fig. 14(b)). These damaged sites would act as water absorption sites, thus, causing the observed changes in Rf and Cf. Under the blisters on the coating surface (Fig. 12), voids were formed, but not at the coating/steel boundary (Fig. 14(c)). Blisters formed during the employed thermal cycling test were clearly different from so-called “cathodic blisters” where the coating is delaminated from the steel substrate.
Photographs of the coating surface at different magnifications after 250 thermal cycles.
FTIR-ATR spectra of the coating surface: (a) before the tests and (b) after 250 thermal cycles.
SEM-secondary electron images (SEI) of cross sections of coated test panels: (a) before the tests, and (b) and (c) after 250 thermal cycles.
An adhesion test of the coating was also carried out by the pull-off method after two weeks of drying at room temperature. Adhesion maintained a relatively high value (approximately 5 MPa). The entire surface of the underlying steel was uniformly covered with a black corrosion product (Fig. 15). The main corrosion product was detected to be Fe3O4 by X-ray diffraction (XRD) (MiniFlex600 of Rigaku Corporation). Underlying steel corrosion usually progresses separately for anodic and cathodic parts. Corrosion takes place at the anode, and oxygen reduction occurs under cathodic blisters. In the employed thermal cycling test, the underlying steel corroded uniformly. The characteristic degradation morphology may be attributed to the thermal cycling or anodic polarization due to galvanic coupling with the Pt wire counter electrode; the reason is not clear yet.
Photos of corroded steel substrate (left) and peeled-off coating (right) after an adhesion test. The test panel was exposed to 250 thermal cycles.
The corrosion products were attached to both surfaces of the peeled-off coating and substrate steel. All of the corrosion products were collected in the manner described in the experimental section and the Fe content was determined with an atomic absorption analyzer. The average corrosion depth of the underlying steel estimated from the amount of corrosion product was about 3 μm, which corresponds to an average corrosion rate of approximately 4 μm/y (icorr = 0.3 μAcm−2). This corrosion rate is sufficiently small compared with that of carbon steel (about 100 μm/y) without a coating in 0.01 M LiCl solution. It means coating degradation after 250 thermal cycles remained its initial stage. The charge transfer resistance Rct for the corrosion reaction of the steel substrate was estimated to be of the order of 105 Ω cm2 using the Stern-Geary equation20) (icorr = k/Rct, where k is a proportionality constant), assuming k = 0.017 V.21) The obtained Rct values of ~105 are much smaller than Rf in Fig. 11. This means that it was difficult to measure Rct (corrosion rate) by EIS, since the underlying steel corrosion progressed at a very slow rate because Rct should be in series with Rf. During the 250 thermal cycles, EIS afforded information on the degradation of the coating through Rf and Cf (Fig. 11), but EIS did not provide the corrosion rate of the underlying steel. The corrosion rate may be measured after the coating is severely degraded and the barrier effect is lost.
Approximately 200-μm-thick epoxy-coated steel for use in ballast tanks was investigated using EIS in LiCl solutions of various water activities at various temperatures. The corrosion test was carried out under thermal cycling, and the film capacitance Cf and film resistance Rf were monitored. The following conclusions were drawn.
(1) Multiple series-connected parallel Rf|Cf elements provided a good fit to the EIS data for the thick coating. The dispersion of time constants results from the Rf heterogeneity in the depth direction of the coating film.
(2) Increasing temperature and water activity slightly increased Cf and drastically decreased Rf. These changes can be explained by the increase in the absorbed amount of water and the formation of continuous water clusters in the coating.
(3) Changes in Cf are well correlated with changes in Rf due to the coating degradation during thermal cycling. Cf is a more suitable parameter for monitoring the initial stages of degradation of thick polymer coatings, because Cf is more easily determined than Rf by EIS when Rf is very large.
(4) The estimated interfacial charge transfer resistance of the corrosion reaction was much smaller than the measured Rf. Thus, EIS is a very powerful method for evaluating the degradation of thick polymer coatings, but it does not provide the corrosion rate of the steel substrate in the initial stages of coating degradation.
The authors are grateful to Chugoku Marine Paints, LTD. for the preparation of test panels. This work was supported by JSPS KAKENHI Grant Number 25820427.
