Effect of AC interference on failure mechanism of zinc-rich epoxy coatings in alkaline environment

In this paper, the failure mechanism of Zn-rich epoxy coating under the AC interference in alkaline environment was revealed based on EIS tests. Using X80 steel as matrix, the coated X80 sample was formed by hand-brushing coating with the thickness of 25 ± 5 μm determined by five-point test. The EIS curves of the coated X80 sample in 3 wt% NaOH solution with immersion time was studied under (no) AC interference, and the failure evolution mechanism of Zn-rich epoxy coating was established through the fitting parameters. The results showed that during the curing process of the Zn-rich epoxy coating, Zn reacted with O2 in the air to generate ZnO, which coated the surface of Zn particle to form the ZnO-Zn structure. In alkaline environment, ZnO dissolved to form Zn(OH)2, namely the activation process of Zn particles, which was inhibited by AC interference. Furthermore, the activated Zn particles reacted to form Zn(OH)2 in alkaline environment, that is, the electrochemical reaction process of activated Zn particles, which was significantly promoted by AC interference. In conclusion, the non-conductive Zn(OH)2 generated in alkaline environment under AC interference, together with un-activated ZnO, can not only isolate the electrical connection between activated Zn particles, but also cut off the cathodic protection between Zn and Fe. Meanwhile, the matrix Fe was passivated in alkaline environment, which weakened the wet binding force between Fe and epoxy coating, resulting in coating stripping.


Introduction
Epoxy coating with the high crosslinking density was widely used for anti-corrosion of metals. However, due to its single structure and low life, the performance of pure epoxy coating cannot meet the requirements of modern industry. Therefore, the Zn-rich epoxy coating was one of the commonly used as anticorrosive coatings [1].
Compared with pure epoxy coating, the Zn content in the Zn-rich coating can reach more than 95% [2]. On the one hand, the Zn particles and its corrosion products [3] can effectively fill the free volume due to the evaporation of curing agent to block the diffusion path of corrosive medium. On the other hand, the cathodic protection at coating/metal interface was formed to protect the matrix (Fe) [4,5]. However, the disadvantages of the Zn-rich epoxy coating were also obvious. The Zn particles will reduce the physical protection performance compared with the pure epoxy coating. For example, the insulation resistivity of pure epoxy coating was on the order of 10 6 and on the order of 10 5 when it failed. However, the insulation resistivity of Zn-rich epoxy coating was on the order of 10 5 and only on the order of 10 2 when it failed [6]. Moreover, more than 30% of Zn particles were not utilized, which were easy to cause environmental pollution [7].
Nowadays, the proposals of new coatings such as grapheme or carbon nanotubes zinc-rich coatings [8,9] had provided new ideas and methods for improving the utilization rate of zinc particles. Cui et al reviewed the graphene in different coatings [10] and found that the grapheme can significantly promote the electric connection of more than 80% of Zn particles in the coating, which enhanced the anti-corrosion performance of Zn-rich coating. But it should be noted that the reported researches had paid more attention to determining optimal Zn content in different environments and improving the utilization rate of Zn particles in the coating. However, on the one hand, the reaction process of Zn as an amphoteric metal in the alkaline atmosphere had not been clarified. On the other hand, due to the construction of the electrification project, the effect of (alternating current) AC interference on buried pipeline was inevitable, under which the adaptability of Zn-rich epoxy coatings to AC interference had not been reported. Therefore, in this paper, the failure mechanism of Zn-rich epoxy coating under (no) AC interference was investigated by the EIS curves.

Experimental settings
In this paper, the X80 steel with the size of 25 × 25 × 2 mm 3 was adopted as the metal base, and the main components were shown in table 1. In order to carry out EIS (electrochemical impedance spectroscopy) tests, the X80 steel was welded by a wire on one of the surfaces, and then sealed by epoxy resin, leaving only 25 × 25 mm 2 of exposed area. Use 600#−1200# sandpaper successively to polish the exposed surface to be mirror-like, and dry it after cleaning.
The experiment coatings were DGEBA epoxy coatings produced by Anhui Wuhu. The weight ratio of epoxy resin to curing agent (modified ethylenediamine) was 10:1 and the density was 1.28 g cm −3 . The coating was applied on the surface of the X80 steel by hand brush and cured for 7 days at room temperature (ASTM D609 standard). The thickness of the coating on X80 steel was determined in 25 ± 5 μm by QNIX8500 gauge through five-point sampling method [11] (ASTM D823 and ASTM D1005 standards).
The 3 wt% NaOH solution was prepared by deionized water and analytical pure NaOH (ASTM D2248 standard), and the experiment temperature was set at 20°C in the constant temperature-humidity device. The EIS curves of samples with immersion time were obtained, and the ZSimpwin software was adopted for the data analysis.
Based on the traditional three-electrode electrochemical system, an AC source was added, as shown in figure 1, in which the coated X80 sample and the Pt electrode were respectively used as the positive and negative poles of the AC source. In the three-electrode electrochemical test system, the working electrode (WE) was the coated X80 sample, the auxiliary electrode (AE) was Pt electrode, and the reference electrode (RE) was saturated calomel electrode (SCE). During the electrochemical test, the AC interference source was disconnected, and the EIS test time was 5 min, which had a negligible impact on the immersion experiment [12,13]. The electrochemical workstation was PARSTAT 2273, the frequency range of EIS curves was 10 5 -10 −2 Hz, and the amplitude of AC sinusoidal signal was ±10 mV. The inductance and capacitance were adopted to eliminate interference between the electrochemical station and AC source. Figure 2 showed the EIS curves of Zn-rich coating in 3 wt% NaOH solution under no AC interference within the experiment time of t = 200 h, and figure 4 showed the fitting results.

EIS curves under no AC interference
When the experiment time was in t = 0-50 min, Nyquist diagram was composed of two capacitive arcs, and there were two peaks at (frequency) f = 10 5 Hz representing the electrochemical reaction process, and f = 23.95 Hz on behalf of solution diffusion. Therefore, the equivalent electric circuit was chosen as R s (Q c R c )(C dl R ct ), as shown in figure 3(a), in which R s was the solution resistance, Q c was the constant phase angle element including Q-Y 0 (coating capacitance) and n (dispersion index), R c was the coating resistance, C dl was the double-layer capacitance, and R ct was the charge-transfer resistance.
In this stage, the coating capacitance (Q c ) and resistance (R c ) increased, indicating that the electrochemical reaction of Zn/ZnO (formed during the curing in the air) → Zn(OH) 2 behaved by the increasing double-layer capacitance (C dl ) and decreasing charger-transfer resistance (R ct ) was the key factor affecting the coating properties in this stage. This was because, on the one hand, the Zn(OH) 2 and the diffused solution increased the dielectric constant of the coating, which led to the increase of the coating capacitance. On the other hand, Zn(OH) 2 increased the resistance of the coating by blocking the diffusion channel, so as to that there was an insignificant peak at f = 10 2 Hz, while the phase angle (j) was small and changed little in j = 10°−30°at f = 10 −2 −10 2 Hz, presenting that physical shielding effect of the coating on solution diffusion [14,15]. However, the physical shielding effect of Zn-rich coating was far lower than that of epoxy coating, mainly because the electrochemical reaction of Zn was significantly enhanced [16]. When the experiment time was in t = 1-200 h, corrosion products were diffused in the coating, thus the Warburg diffusion characteristics appeared in Nyquist diagram. Meanwhile, an electrochemical reaction occurred when the solution reached the coating/metal interface. Therefore, the equivalent electric circuit was chosen as R s (Q c (R c W))(C dl R ct ), as shown in figure 3(b), in which W was the Warburg diffusion impedance.
This stage can be divided into two periods based on the results of coating resistance and charge transfer resistance. Within the experiment time of t = 3-50 h, the coating resistance and charge transfer resistance decreased rapidly, in which the electrochemical reaction of Zn/ZnO was in full taking place and then Zn(OH) 2 diffusion in the coating was inhibited. Therefore, the max phase angle (j max ) locating at f = 10 −2 Hz in logf-j decreased gradually from j max = 47.7°to j max = 23.3°, and moreover, the logf-j curves at t = 13 h and t = 50 h coincided within f = 10 −2 −10°Hz. During the experiment time of t = 50-200 h, the coating resistance and charge transfer resistance decreased slightly, indicating that the reaction process of zinc particles and the diffusion process of corrosion products were basically completed at this time, represented by that the max phase angle at f = 10 5 Hz decreased from j max = 56.3°at t = 1 h to j max = 15.3°at t = 200 h. Furthermore, the phase angle at f = 10 −2 Hz representing the mass diffusion increased from j max = 23.3°to j max = 42.0°, indicating that the Zn-rich coating had lost its physical shielding effect, and Zn(OH) 2 diffused into the solution, leading to the appearance of Warburg diffusion impedance in EIS curves [17,18]. Under this circumstance, the electrochemical reaction process at the interface was dominant. However, in alkaline environment, the metal matrix was easy to form a Fe 3 O 4 passivation layer was easy to be formed on the surface of the metal matrix to prevent the further occurrence of the electrochemical reaction [17], and the passivation layer also easily made the coating stripping [19].
During the later stage of experiment, the solution diffused again because the coating had lost the physical shielding effect, but the charge-transfer process in high frequency gradually disappeared due to the passivation of the metal matrix surface. For example, when the experiment time reached t = 191 h, there was only a capacitive arc in Nyquist diagram, indicating the degradation process of the Zn-rich coating. At this time, the   coating capacitance and the coating resistance were basically consistent with the parameters of the passivation film of metal in alkaline environment [17]. Figure 5 showed EIS curves of Zn-rich coating under AC potential of 7 V with time in 3 wt% NaOH solution, and figure 6 showed the fitting results.

EIS curves under AC potential of 7 V
Under the condition of AC potential of 7 V, the Nyquist diagram at t = 30 min was similar to that under no AC interference, showing the characteristics of double capacitive arcs with a weak high-frequency capacitive feature. Therefore, the equivalent electric circuit was chosen as R s (Q c R c )(C dl R ct ). In logf-j curve, the max phase angle (j max = 66.23°) occurred at f = 0.16 Hz instead of the high frequency, indicating that the solution diffusion was dominated at 30 min. Compared with the corresponding results under no AC interference, it can be found that the ZnO activation was inhibited by applied AC potential.
When the experiment time reached t = 1 h, a smaller phase angle peak (j max = 22.78°) appeared at f = 672.34 Hz, indicating the electrochemical reaction of Zn particles, and disappeared at t = 2 h. Subsequently, the mass diffusion characteristic in Nyquist diagram occurred. As a result, the equivalent electric circuit of R s (Q c (R c W))(C dl R ct ) was chosen. As the experiment lasted to t = 3 h, there was only a single capacitive arc presented in Nyquist diagram. Moreover, this feature exited for a long time to 200 h, so the equivalent circuit was chosen as R s (Q c R c ), as shown in figure 3(c). This was because the activation process of ZnO was inhibited under AC interference, so the solution can quickly reach the surface of X80 steel to lead to passivation [20][21][22][23][24], proved by that the max phase angle decreased from j max = 46.22°at t = 3 h to j max = 23.49°at t = 49 h. Meanwhile, for activated zinc particles, AC interference can promote rapid electrochemical reaction to form Zn(OH) 2 . Under this circumstance, there was little conductive Zn particle, so it was similar to the characteristics of pure epoxy coating.
However, in the fourth stage (t = 49-200 h) of the experiment, the Zre and Zim of the Nyquist diagram increased significantly, and the max phase angle and its frequency range increased significantly, indicating that the densification of the coating increased. At the same time, the EIS curves at t = 129 h and t = 200 h basically coincided, indicating that the system did not change much at this stage. This may be because AC interference inhibited the activation of Zn particles, while OH − promoted the crosslinking of organic epoxy coating. Therefore, the coating capacitance and resistance continued to increase during the whole experiment time.  Figure 7 showed the mechanism of Zn particles in the coating and figure 8 presented the images of Zn-rich coating in alkaline solution under (no) AC interference.

Failure mechanism of zinc-rich coating
As mentioned above, the main effects of Zn particles in coatings included electrochemical consumption of corrosive media, blocking of diffusion paths by corrosion products, and cathodic protection at the interface, as shown in figure 7. During the curing process of the Zn-rich coating, the Zn particles in the coating form hard and  dense zinc oxide (Zn→ZnO), which blocked the electrical connection between the zinc particles due to its nonconductive property.
In near-neutral solution, due to the presence of H 2 O, O 2 and Cl − et al, ZnO gradually turned into loose corrosion products of ZnCl 2 and Zn(OH) 2 , as shown in figure 7(a). In theory, the Zn particles were exposed, and then were electrically connected with each other through the solution, as shown in figure 7(b), which can on the one hand provide the more effective cathodic protection of Zn as anode and Fe as cathode at the interface, as shown in figure 7(c), and on the other hand block the diffusion path of solution by Zn(OH) 2 .
However, in alkaline environment, ZnO can dissolve more quickly (ZnO → Zn(OH) 2 ), and furthermore, the electrochemical reaction of activated Zn particles quickly occurred to form Zn(OH) 2 , under which the accumulation of Zn(OH) 2 can easily cut off the electrical connection of activated Zn particles. When AC potential was applied, the activation process of ZnO was inhibited and the electrochemical reaction of activated Zn particles was accelerated. Under this circumstance, compared with the image under no AC interference as shown in figure 8(a), more corrosion products can be produced faster under AC interference as shown in figure 8(b) (ASTM D714 standard), which greatly reduced the cathodic protection of the Zinc-rich coating.

Conclusions
In this paper, the electrochemical behavior of Zn particles in epoxy coating in alkaline environment and its effect on coating failure were analyzed by electrochemical method, and the effect of AC potential was considered. The following conclusions were drawn.
(1) In alkaline environment, ZnO in the coating was activated more quickly, and then the activated Zn particles reacted to form Zn(OH) 2 covering the surface of Zn particles, which blocked the electrical connection between activated Zn particles. Meanwhile, the alkaline solution reached the metal surface and the passivation occurred. (2) Under the AC potential, the activation process of ZnO was inhibited, while the reaction process of activated Zn particles was promoted. Under this circumstance, the Zn-rich coating was more likely to lose its electrochemical effect.

Data availability statement
All data that support the findings of this study are included within the article (and any supplementary files).