Improving the Protection Performance of Waterborne Coatings with a Corrosion Inhibitor Encapsulated in Polyaniline-Modified Halloysite Nanotubes

Abstract

Organic coatings provide an effective way to improve the corrosion resistance of metals. Traditional organic varnishes, however, either contain highly polluting or toxic components or lack self-healing ability. In this article, we report a feasible method of preparing polyaniline-modified halloysite nanotubes (PANI@HNTs). They were loaded with a corrosion inhibitor, benzotriazole (BTA), and were tested as multifunctional anticorrosion additives for environmentally friendly epoxy waterborne coatings. The PANI@HNTs were formed via the in situ polymerization of aniline in the presence of halloysites. The BTA loading was then carried out and reached up to 14.5 wt.%. The BTA retention ability of the PANI@HNTs was significantly improved in comparison to that of pure HNT. Electrochemical impedance spectroscopy (EIS) tests of the coatings immersed in a 3.5 wt.% NaCl solution showed that the barrier and corrosion inhibition effects were enhanced by two to four orders of magnitude with the incorporation of BTA-loaded PANI@HNTs. The salt spray tests on artificially scratched coatings revealed that the surfaces protected by varnishes doped with the BTA-loaded PANI@HNTs exhibited the lowest degree of corrosion compared to the control samples, illustrating the self-healing potential of the modified coatings.

1. Introduction

Metal corrosion is an inevitable natural phenomenon causing substantial economic loss each year. To prevent this process, many anticorrosion technologies have been developed. Among these, organic coatings over metal surfaces offer an efficient and economical solution to this problem [1,2]. Traditional coatings are organic solvent-borne, which makes them harmful to the environment due to their high content of volatile organic compounds (VOCs). Currently, low-VOC-containing waterborne coatings are becoming more prevalent [3,4]. However, the protection effect of waterborne coatings is generally lower than their organic solvent-borne counterparts. In most cases, this is due to the presence of hydrophilic channels to the metal surface, formed by the surfactants used to stabilize the latex particles [5]. The addition of nanomaterials to water-based coatings is an effective way of blocking micro- and nano-pores, thus enhancing the barrier effect [6]. Various nanomaterials have been used, including graphene derivatives [7], metal–organic frameworks (MOFs) [8], and silica-based nanoparticles [9,10], to name a few. Polyaniline (PANI)-based nanomaterials might offer another promising protection pathway due to their relatively easy synthesis and intrinsic anticorrosion properties [11]. PANI nanofibers [12], PANI-modified graphene [13], and PANI-doped polymeric spheres [14,15] have been demonstrated to improve corrosion protection.

While organic coatings initially provide excellent corrosion protection, they lack the sufficient scratch resistance to withstand mechanical and/or environmental damage. The quick diffusion of water and other corrosive ions would cause an instant corrosion reaction in the damaged zone. If the exposed area is not repaired in a timely manner, the corrosion reaction tends to spread quickly and ultimately might lead to a loss of function [16]. Traditionally, chromate corrosion inhibitors containing Cr(VI) have been shown to be capable of forming a protective oxide layer at damaged sites. However, the usage of highly toxic Cr(VI) compounds was greatly reduced over the last several decades and completely banned by the EU in 2017. Propelled by the development of environmentally friendly solutions of “self-healing” coatings, various strategies have been investigated and proven to be effective, such as embedding polymerizable agents into coatings and using polymers bearing reversible bonds as binders [17].

The addition of micro/nanoparticles to organic corrosion-inhibiting reagents is a promising approach to incorporating “self-healing” or “self-repair” functionality in these coatings. In this strategy, corrosion inhibitors leak from capsules to form a passivating layer over the exposed metal surfaces in damaged areas and “heal” the “wound”. Furthermore, organic corrosion inhibitors often contain N–H, S–H, or other reactive groups suitable for epoxy binders. The chemical linkage formed by the inhibitor with the epoxy groups during the curing stage could eliminate the functionality of the inhibitor and reduce the crosslinking density of the epoxy coating. The incorporation of organic corrosion inhibitors into nanoparticles prevents their reaction to an epoxy binder. Extensive studies on various micro/nano containers have been reported, such as mesoporous SiO₂ [10], clays [18], Mn₂O₃ [19], zirconia [20], and carbon nanotubes [21]. Halloysite is a type of naturally occurring aluminosilicate with a hollow tubular structure [22]. It has been used for the encapsulation of small organic corrosion inhibitors [23]. A common way to manipulate the release rate of guest molecules is to cover the halloysite nanotubes (HNTs) with polyelectrolytes via layer by layer self-assembly (LbL) [24,25]. Although this strategy is effective, its preparation is time consuming and not economically advantageous. In situ polymerization is an efficient way of modifying the outer surface of HNTs, forming a porous three-dimensional structure, often found in sensor applications [26]. These recent findings inspired us to extend their applications into the area of corrosion protection.

In distinction to earlier research, the main aim of this article was to combine the good corrosion inhibition properties of PANI with the broad binding/release capabilities of HNTs into a multifunctional microcapsule for waterborne epoxy coatings with an improved corrosion resistance. The PANI-modified HNTs were prepared via the emulsion-free in situ polymerization of aniline. A set of scanning electronic microscopy (SEM), energy-dispersive X-ray (EDX)and Fourier transform infrared (FT IR) spectroscopy techniques were used to characterize the structure and composition of microcapsules and evaluate their loading capacity and release profile. Electrochemical impedance spectroscopy (EIS) and salt spray tests revealed a notable improvement in the corrosion resistance and propensity for self-healing.

2. Materials and Methods

2.1. Materials

Aniline, benzotriazole (BTA), and 1M HCl were purchased from Sinopharm Chemical Co. (Shanghai, China). Ammonium persulfate (APS), (NH₄)₂S₂O₈, was acquired from J&K Scientific Co. (Beijing, China). Halloysite (HNTs), Al₂Si₂O₅(OH)₄·2H₂O, was purchased from Sigma Aldrich (Shanghai, China). Waterborne epoxy resin WEP 804 and the curing reagent Wh-5 were received from Yueyang Baling Chemical company (Yueyang, China). Flash rust inhibitor Halox Flash-X 330 was obtained from Shanghai Pengpan Chemical Corporation (Shanghai, China). All the chemicals were used without any further treatment. The steel plates Q235 (0.14%–0.22% C, 0.30%–0.65% Mn, 0.30% Si, 0.08% S, 0.045% P, and major Fe) employed in the coating experiments were initially polished with 400 mesh sandpaper and cleaned with acetone before the coating application.

2.2. Preparation of Polyaniline-Modified HNT (PANI@HNT)

Three types of PANI@HNTs were prepared with different “aniline: HNT” weight ratios of 0.5, 1, and 2. PANI@HNT-0.5, with a weight ratio of aniline to HNT of 0.5, was prepared according to the literature [27]. Briefly, 1 g of HNT was suspended in 200 mL of 1M HCl, followed by the addition of 0.5 g of aniline. The suspension was magnetically stirred at room temperature for 1 h before being cooled in an ice bath, then it was further stirred for another hour. Then, 2.3 g of APS was dissolved in 30 mL of 1 M HCl and added to the cold suspension over a period of 0.5 h. The mixture was stirred in the ice bath for 1 h and allowed to warm up to room temperature. After stirring overnight, the reaction mixture was centrifuged to collect the crude product. The product was washed five times with portions of pure water and dried at 40 °C under vacuum. The product was obtained as a dark green powder. PANI@HNT-1 and PANI@HNT-2, with different weight ratios of aniline to HNT, were prepared in a similar manner. The purified PANI@HNTs were characterized with SEM, EDX, and FITR.

2.3. Scanning Electron Microscopy (SEM)

The SEM analyses were performed on a Zeiss Supra 55 Field Emission Scanning Electron Microscope (FESEM) (Zeiss China, Shanghai, China) at an accelerating voltage of 10 KeV. The specimen preparation protocol was as follows: the HNT and modified HNT samples were placed onto a conductive tape and subjected to the SEM analysis directly. The waterborne epoxy coatings were sputter coated with a thin layer of Au before the SEM observation. For the analysis of the coating’s cross-section, the samples were immersed in liquid nitrogen, fractured, and coated.

2.4. Fourier Transform Infrared Spectroscopy (FT IR)

The spectroscopic characterization of the PANI@HNTs and their precursor was carried out on a Nicolet iS10 spectrometer (Thermo Fisher Scientific, Shanghai, China). The specimens were analyzed at room temperature in KBr pellets.

2.5. BTA Loading into PANI@HNTs

The method for loading BTA into the nanotubes was adopted from reports by other groups with minor modifications [25]. The PANI@HNT (0.5 g) was dispersed in a 250 mL saturated aqueous solution of BTA (20 g/L) and stirred at room temperature overnight. The suspension was placed under vacuum with a circulating water pump at 60 °C for 4 h. Then, the vacuum was released, and the suspension was kept for an additional 2 h at the same temperature to facilitate the diffusion of the inhibitor into the nanotubes. Vacuum was applied again for 4 h and then paused for 2 h until the suspension turned into a thick slurry. The slurry was stored at room temperature overnight for the entrapped BTA to fully crystallize. Then, the slurry was quickly washed with acetone and thoroughly rinsed several times with water to remove the excess BTA. The slurry was dried at 40 °C under vacuum to obtain the final product.

2.6. Determination of the Loading Ability of PANI@HNT

The measurement of the loading capacity was adopted from the literature [25]. The PANI@HNTs loaded with BTA were suspended in water, sonicated for 20 min, and then centrifuged to extract the BTA into the supernatant. The above process was repeated twice, and the supernatants were combined. The quantity of the extracted BTA was determined from the absorbance intensity at 260 nm using a Thermo Fisher Evolution 220 UV-Vis spectrophotometer (Thermo Fisher Scientific, Shanghai, China). The calibration curve is presented in Figure S1. The loading percentage of the BTA is calculated using the following equation:

$$\mathrm{BTA} \mathit{loading}\%=\frac{\mathit{weight} \mathit{of} \mathit{loaded} \mathrm{BTA}}{\mathit{weight} \mathit{of} \mathit{PANT}@\mathit{HNT} \mathit{loaded} \mathit{with} \mathrm{BTA}}\times 100$$

(1)

2.7. Controlled Release of Corrosion Inhibitor

The release assay was conducted on ~30 mg of PANI@HNT-0.5, 1, or 2 in 5 mL of deionized water with a pH in the range of 6.5–7.2. The procedure was adopted from the literature [24,25]. The suspension was continuously stirred during the release experiment to keep the nanotubes well dispersed. To determine the amount of released BTA into the solution, 500 μL of the suspension was taken at certain time intervals, diluted to 2 mL, and filtered through a 0.22 μm syringe filter before the absorbance testing. In total, 500 μL of fresh deionized water was added to the suspension after the sample was taken to keep the total volume constant. The control experiments with bulk BTA crystals and the as-received BTA-loaded HNTs were carried out in a similar fashion. Each release test was conducted twice. The percentage of accumulative release was calculated as the average of the tests.

2.8. Coating Preparation and Characterization

The weight percentages of PANI@HNT-0.5, 1, or 2 in the dry coating in this report were set at 4, 8, and 12%. The additives were suspended in water and then added to the waterborne epoxy under vigorous stirring. After complete dispersion was visually achieved (in ~10 min), the anti-flash rust and curing agents were added. The coating was then applied onto the Q235 plates by the wire-wound metering rod (120 μm) in two layers. The coatings were further cured at room temperature overnight and then in the heated oven at 75 °C for 6 h. The total thickness was measured in the range of 180–200 μm. The micro morphology of the coatings was observed using FESEM. The coated samples were sputter coated with a thin layer of gold before observation. A thermogravimetric analysis (TGA) of the coating was conducted on a Mettler Toledo TGA/DSC 3+ instrument (Mettler Toledo China, Shanghai, China) at a heating rate of 10 °C∙min⁻¹ under a nitrogen atmosphere. The water droplet contact angle tests on the coatings were recorded by a Shanghai JCY goniometer (Shanghai Fangrui Instrument, Shanghai, China). Differential Scanning Calorimetry (DSC) was performed on a TA Q20 instrument (TA China, Shanghai, China) by heating the sample from −40 to 200 °C at 10 °C∙min⁻¹ under a nitrogen atmosphere. The adhesion of the coating to the substrate was evaluated using the cross-cut test according to the ASTM D 3359 standard [28].

2.9. Electrochemical Impedance Spectroscopy (EIS)

The EIS experiments were conducted on a Princeton Applied Research PARSTAT 2273 workstation (Ametek, Beijing, China) using a three-electrodes cell consisting of a 2 × 2 cm² platinum foil as a counter electrode, a saturated calomel electrode (SCE) as a reference electrode, and a test coating applied on a piece of Q235 steel as a working electrode. The working electrode was waxed, leaving a 1 × 1 cm² area of the test coating. An Open circuit potential (OCP) vs. time test was performed before every single EIS test lasting from 10 to 20 min. When the fluctuation of the OCP did not exceed 3 mV, the EIS test was started. The EIS tests were performed two to three times. The 10 mV sinusoidal perturbation around the OCP was performed with 10 mV peak to peak. A total of 64 points were recorded in the frequency range from 10⁵ to 0.01 Hz. The Kramer–Kronig relations were performed in the ZSimpWin software (ZSimpWin 3.60) to validate the data.

2.10. Salt Spray Test

The salt spray tests on the coating panels with specifically made scratches were carried out in accordance with the ASTM B-117 standard [29]. The salt spray chamber remained constant at 37 °C and a 5 wt.% NaCl solution with a pH of 6.5–7.5 was used as the corrosion medium.

3. Results and Discussion

3.1. Synthesis and Characterization of PANI@HNT

The preparation of the PANI-modified HNTs was performed using the in situ polymerization of aniline on the outer surface of the HNT. According to the literature, the outer surface of HNT is negatively charged at a pH of <2 [30]. The positively charged anilinium chloride was adsorbed onto the outer surface of the HNT and polymerized at a low temperature in the presence of ammonium persulfate as an oxidant. The weight ratios of aniline to HNT were 0.5, 1, and 2 in the polymerizations performed in this study. The morphology comparison of the resulting PANI@HNT-0.5, PANI@HNT-1, PANI@HNT-2, as-received HNT, and PANI polymerized in the absence of HNT is presented in Figure 1 and Figure S2.

The SEM images show that, while the overall shape of the HNTs remained almost unchanged, they increased in thickness. For example, the PANI@HNT-0.5 diameter was 45 ± 13 nm vs. the pure HNT with a diameter of 40 ± 12 nm. The lengths of the tubes ranged from 20 nm to half a micron. When the weight ratio of the aniline monomer to HNT increased, the resulting average diameters of PANI@HNT-1 and PANI@HNT-2 rose to 80 and 97 nm, respectively. The roughness of the outer surface of PANI@HNT was notably more pronounced compared to the relatively smooth HNTs. These are clear indications that PANI was successfully grown on the HNT. The template polymer synthesis preserved the characteristic tubular shape of the HNT. In comparison, the PANI particles prepared under the same conditions, but in the absence of HNT, exhibited irregular morphology and a wide range of size distributions (Figure S2). The vacuum-dried PANI@HNTs could be easily dispersed in water by magnetic stirring, which would facilitate their usage in waterborne coatings. Additional proof for the formation of PANI layers on the HNT peripheral surfaces was provided by the EDX analyses, as shown in Figure 1. With the growth of PANI on the surface, the element contents of aluminum, silicon, and oxygen from the HNT were diminished, while the signals of carbon, sulfur, and chlorine constituting the PANI emeraldine salts increased. It should be noted that the EDX results depended strongly on the depth distribution of the detected elements on the sample surfaces due to the limited penetration ability of the electron beam. Thus, the detection of the HNT elements became more difficult as the layer of the PANI thickened.

The FT IR spectra of the HNT and PANI@HNTs are presented in Figure 2. In the HNT spectrum, the absorbance bands at 3700 and 3620 cm⁻¹ were attributed to Al₂–OH stretching. The bands at 1030 and 910 cm⁻¹ were associated with Si–O stretching and Al₂–OH bending, respectively. These bands were still present after coating with PANI, as marked in the spectra of PANI@HNT-0.5. In comparison to the neat HNT, the new bands of the PANI@HNT-0.5 were caused by the polyaniline adsorption. The band at 3460 cm⁻¹ was assigned to the stretching of N-H. The adsorption band of the C=C stretching vibration in the quinoid ring was observed at 1580 cm⁻¹, while the one in the benzoic ring appeared at 1496 cm⁻¹. The C–N stretching vibration of the benzoic ring was observed at 1300 cm⁻¹. The absorption band of the C=N stretching vibration of the quinoid ring was found at 1130 cm⁻¹ [31,32]. The FTIR spectra of PANI@HNT-1 and -2 exhibited similar characteristic peaks, yet the bands corresponding to Al₂–OH and Si–O became weaker as the PANI layer grew thicker. This was consistent with the EDX results.

3.2. Corrosion Inhibitor Loading and Release

BTA, a common corrosion inhibitor [33], was loaded into the lumen of the HNT via diffusion from its concentrated aqueous solution and deposition after solvent evaporation. The binding capacities of the HNT and modified HNTs are presented in Figure 3. The capacity of the neat HNT was estimated as 4.68%, which was quite similar to that of PANI@HNT-0.5. By increasing the PANI layer thickness, the loading capacity was significantly improved, with PANI@HNT-1 and PANI@HNT-2 binding 12.53% and up to 15.94%, respectively. It was observed that the PANI could also absorb a sizable amount of BTA. According to recent studies, PANI in situ polymerized on HNT has a more porous structure than its neat form [26]. This phenomenon could be more pronounced when the PANI layer grows. This could have contributed to the increment in the BTA loading capacity as the PANI layer became thicker.

The BTA release profiles from the HNT and PANI@HNTs are presented in Figure 3b. Since PANI@HNT-2 had the highest loading capacity, it was used in the release rate tests. It was obvious that the BTA diffusion from the neat HNT and PANI-modified HNTs was much slower than that from its neat form. The release profiles of the HNT and PANI@HNTs had two stages: the initial burst release and the later steady release. The burst release in the first two hours was attributed to the quick diffusion of the exogenously absorbed BTA. The slow and steady release was from the BTA loaded inside the lumen of the HNT. Since the PANI itself could absorb a certain amount of BTA, PANI@HNT-2 released more BTA in the burst release stage.

Electrochemical impedance spectroscopy (EIS) tests of clean steel sheets in 3.5 wt.% NaCl with and without PANI@HNT-2-BTA were performed to examine their corrosion inhibition effect, using the strategy from the literature [34]. The Bode plots for steel panels immersed in solutions for different time spans are shown in Figure S3. Usually, the impedance modulus at a low frequency (Z_{f = 0.01 Hz}) is employed to evaluate the ability to suppress the current between the anode and cathode. The higher the Z_{f = 0.01 Hz}, the better the corrosion suppression. After 24 h of immersion, Z_{f = 0.01 Hz} approached 3.6 kΩ∙cm² in the presence of PANI@HNT-2-BTA (Figure S3a), while the Z_{f = 0.01 Hz} of the bare steel was only 0.39 kΩ cm² (Figure S3b). This difference indicated that the corrosion of steel could be inhibited in the presence of PANI@HNT-2-BTA. The impedance modulus of PANI@HNT-2-BTA grew continuously over time and levelled off after 8 h, while the bare steel exhibited no considerable time difference during the immersion. The corrosion inhibition can be attributed to the protective film formation due to the adsorption of released BTA onto the exposed metal surface.

3.3. Coating Preparation and Physical Properties

The loading of PANI@HNT-2-BTA was performed by simply mixing the modified nanotubes with waterborne epoxy resins. The mixtures were then coated onto Q235 steel panels. The secondary amino group in the emeraldine base was active towards the epoxy groups, which could improve the compatibility of the nanotubes and the resin. On the other hand, an “overdose” of PANI could possibly affect the curing of the epoxy resin. Thus, the dosage of the PANI@HNT-2-BTA was a reasonable study target for finding the optimal recipe. Pictures of the coatings prepared by mixing the epoxy resin with 1%–12% PANI@HNT-2 (calculation based on dry film weight percentage) are presented in Figure 4. At a 1% dosage (Figure 4b), the cover was insufficient. If the content of PANI@HNT-2-BTA was higher than 3%, the coatings had a satisfactory cover efficiency. The coatings, however, started to lose glossiness as the dosage became higher. In particular, when the PANI@HNT-2 content was at 12% (Figure 4f), visible defects such as pinholes and overall roughness were spotted on the surface of the coating. It also took longer for the coating to harden. This phenomenon has also been reported in other studies with different types of PANI nanoparticles and is attributed to the role of the secondary amino group [35]. Despite its low reactivity, this secondary amino group could react with the oxirane rings in the epoxy resin and increase its stiffness. We postulate that too much PANI would encumber the curing of the epoxy resin, leading to defects in the film. Considering the fact that the high PANI content could further deteriorate the quality of the epoxy resin coating, a dosage greater than 12% was not attempted in this research. The adhesion strength of these doped coatings together with the pure epoxy coating was evaluated using the cross-cut test. It was found that all these coatings had a decent adhesion strength, with classification at 4B according to ASTM D 3359 (see detail in Table S1).

The SEM images of the coating surfaces with the PANI@HNT-2 dosage ranging from 3 to 12% are displayed in Figure 5. The surfaces of these coatings were relatively smooth and defect free, except the one at 12%. The surface of the 12% coating had cracks and wrinkles, and nanotubes were occasionally spotted, suggesting a possible overdosage of additives, as seen in Figure 5g. The SEM images of the cross-sections of these coatings are also exhibited in Figure 5, showing the increasing appearance of randomly oriented PANY@HNTs. At relatively low dosages (3% and 5%), the distribution of the nanotubes throughout the coating was uniform and without obvious agglomeration (Figure 5b,d). When the nanotube content increased (8% and 12%), nanotube clustering was observed (encircled area in Figure 5f,h). This kind of agglomeration could potentially impair the compactness of the coating, leading to a decrease in the corrosion protection efficiency. The water droplet contact angles of these coatings were also evaluated. It was found that the wettability of the varnishes was not grossly affected by the addition of the hydrophilic nanotubes (Figure S4). This implied that the nanotubes were predominantly embedded in the bulk of the coating, with a negligible effect on the surface properties.

The thermal stability of the coatings was evaluated using TGA and DSC, and the results are shown in Figure 6 and Figure S5. The derivative thermogravimetry (DTG) curve of the coatings without filler presented in Figure 6a reveals two major weight loss peaks resulting from the loss of residual solvent and the degradation of the epoxy framework. In comparison, the coatings doped with PANI@HNT had their weight loss peaks shift to a slightly higher temperature and this weight loss was also reduced. The introduction of the thermally stable filler improved the overall thermal durability of the coatings. The glass transition temperature (T_g) of the coatings was measured using DSC (Figure 6b). The doping of PANI@HNT into the coatings increased the T_g of the epoxy coatings. The T_g grew linearly as more PANI@HNT was used (Figure S6). It is possible that PANI@HNT was covalently connected to the epoxy frame via the reaction of the secondary amino group with the oxirane rings of the epoxy. These stiff nanotubes restricted the movement of the polymer chains, thus leading to the increment in T_g and the toughness of the coating.

3.4. EIS Corrosion Protection Evaluation of Intact Coatings

To find the optimal PANI@HNT-2-BTA dosage, sample panels with different PANI@HNT-2-BTA amounts were immersed in a 3.5 wt.% NaCl solution and EIS was performed to evaluate the corrosion resistance. To simplify the test and save time, the experiment was carried out at 50 °C to accelerate the diffusion of the aggressive electrolytes into the coatings, thus escalating the coating failure [36]. A higher impedance modulus of the coating at a low frequency was the sign of a lower extent of corrosion (i.e., an overall better corrosion protection by the coating) [37,38]. As shown in Figure 7a, the coating with 5% PANI@HNT-2-BTA provided the best corrosion protection at Z_{f = 0.01Hz} (5.01 × 10⁶ Ω∙cm²), followed by the 3% and 8% coatings at the same frequency (2.52 × 10⁶ and 3.45 × 10⁵ Ω∙cm², respectively). The coating with 12% PANI@HNT-2-BTA and the pure epoxy varnish had extremely low impedance moduli, indicating a poor protective ability. This finding was not surprising in view of the surface defects observed with the 12% coatings (Figure 5 and Figure 6). The Bode phase curve presented different degrees of coating failure, as shown in Figure S7. The lower the frequency of the inflection point, the higher the degree of the protection loss. From the Nyquist plots (Figure 7b), it can be seen that more than one time constant appeared in all circumstances, suggesting that all the coatings suffered from corrosion at different degrees [39]. The semicircular arc of all the coatings decreased in the order of 5% > 3% > 8% > 0% > 12%. A larger semicircle diameter and higher impedance modulus at a low frequency generally signified a superior corrosion resistance feature [40,41]. The EIS indicated that the 5% doped coating showed the best corrosion resistance. The numerical EIS data analysis confirmed the graphical observations. The EIS results were fitted into an electrochemical equivalent circuit containing two-time constants and a Warburg element, as shown in Figure 8a. R_s, R_p, and R_ct represent the resistance of the solution, the electrolyte seepage into the coating pores, and the charge transfer resistance of the double layer at the interface between the electrolytes and the metal substrate, respectively. Constant phase-angle elements (CPE) were used to simulate the capacitance of the coating and the interface double layer. The presence of the Warburg elements signified the corrosion reaction of the metal substrate caused by the diffusion of corrosive ions in the coatings. The electrochemical impedance parameters are listed in

Table 1. Electrochemical impedance parameters of coatings immersed in 3.5 wt.% NaCl after heating.

The Amount of PANI@HNT-2-BTA Doped in the Coatings (%)	Cp		Rp (Ω∙cm2)	Cdl		Rct (Ω∙cm2)	W	Chi Square (×10−3)
	Y0 (F∙cm−2)	n		Y0 (F∙cm−2)	n		(Ω∙cm2)
0	5.19 × 10−5	0.70	1.16 × 104	3.12 × 10−4	0.62	5.32 × 104	2.57 × 104	1.29
3	3.72 × 10−9	0.98	7.46 × 105	1.26 × 10−6	0.61	2.23 × 106	1.04 × 104	4.35
5	5.00 × 10−9	0.79	8.46 × 105	5.01 × 10−7	0.65	2.83 × 106	1.96 × 105	4.86
8	3.62 × 10−6	0.80	2.19 × 103	2.06 × 10−7	0.53	1.81 × 105	7.50 × 105	0.51
12	6.76 × 10−5	0.98	3.10 × 103	4.44 × 10−5	0.61	2.33 × 104	2.12 × 106	0.58

C_p—coating CPE, R_p—coating resistance, C_dl—CPE of the double layer, R_ct—charge transfer resistance, Y₀—admittance magnitude of the capacitor, and n—CPE component power index. Electrochemical equivalent circuit model used: Figure 8a.

. A high R_p was a sign of a low degree of electrolyte penetration into the coating and a high barrier effect. R_ct estimated the electron loss from the metal substrate during the electrochemical reaction. A large R_ct hinted at a low degree of corrosive reaction. It was obvious that the 5% PANI@HNT-2-BTA-doped coating had the highest R_p and R_ct, and therefore offered the best overall corrosion protection. Therefore, this type of coating was used in the subsequent studies. Epoxy coatings overdosed with PANI@HNT-2-BTA probably had nanotubes agglomeration, as discussed in the previous section. The pinholes and other defects could greatly impair the compactness of the coating layer, resulting in quick failure. The admittance magnitude of the capacitor Y₀ could also be used to quantify the corrosion performance of the coating. A low value of the Y₀ of Cp indicated less conductive electrolytes absorbed into the coating, and a low value of the Y₀ of Cdl suggested a slower corrosion rate at the metal surface. The 3% and 5% doped coatings carried the lowest coating capacitance (Y₀ of Cp is at ~10⁻⁹ F∙cm²), while the 5% and 8% doped coatings showed the slowest corrosion rate (Y₀ of Cdl is at ~10⁻⁷ F∙cm²). In agreement with the above discussion about the resistance elements, the 5% doped coating exhibited the best corrosion protection performance in our tests.

To evaluate the corrosion protection performance of the coatings in the usual service environment, long-term EIS tests were conducted at room temperature with the EIS data exhibited in Figure 9. The impedance modulus in the Bode plot, together with the radius of the arch in the Nyquist plot, decreased in the following order: PANI@HNT-2-BTA coating > PANI@HNT coating > PANI-doped coating > original waterborne epoxy varnish. The Z_{f = 0.01Hz} of the PANI@HNT-2-BTA coating remained higher than 10⁸ Ω∙cm² throughout the entire time test (Figure 9a), while the Z_{f = 0.01Hz} of the waterborne varnish dropped from 10⁸ after 1 day to 10⁵ Ω∙cm² after 50 days (Figure 9g). The coatings doped only with PANI@HNT and PANI also had an improved corrosion protection compared to the coating without any filler. The PANI@HNT coating exhibited a better performance than the PANI. The Z_{f = 0.01Hz} in the Bode plot and radius in the Nyquist plot decreased slower in the EIS test (Figure 9c–f). As mentioned in Section 3.1, the PANI prepared in this research was agglomerate and the overall particle size was larger than that of the PANI@HNT. The smaller surface area and poorer dispersion of the PANI could be attributed to the deteriorated protection. The electro equivalent circuit in Figure 8b was used to fit the acquired EIS data. The electrochemical impedance parameters are listed in

Table 2. Electrochemical impedance parameters of coatings immersed in 3.5 wt.% NaCl at room temperature and different times.

Sample	Immersion Time (Days)	Cp		Rp (Ω∙cm2)	Cdl		Rct (Ω∙cm2)	Chi Square (×10−3)
		Y0 (F∙cm−2)	n		Y0 (F∙cm−2)	n
PANI@HNT-2-BTA-doped coating	1	4.64 × 10−10	0.97	3.74 × 108	8.37 × 10−10	0.78	1.23 × 109	8.94
	15	3.73 × 10−10	0.99	1.14 × 108	8.17 × 10−10	0.69	5.26 × 108	3.22
	30	3.13 × 10−10	0.95	1.04 × 108	1.83 × 10−9	0.73	2.70 × 108	2.67
	50	2.98 × 10−10	0.98	7.06 × 107	1.45 × 10−9	0.68	3.96 × 108	1.92
PANI@HNT-2-doped coating	1	4.99 × 10−10	0.98	9.41 × 108	1.15 × 10−9	0.39	2.45 × 109	6.27
	15	3.91 × 10−10	0.97	3.10 × 108	1.98 × 10−9	0.64	1.33 × 108	3.38
	30	3.93 × 10−10	0.98	2.90 × 108	1.11 × 10−8	0.65	9.33 × 107	3.61
	50	3.43 × 10−10	0.98	7.74 × 107	1.98 × 10−8	0.78	6.96 × 107	0.91
PANI-doped coating	1	1.9 × 10−10	0.99	4.96 × 108	7.95 × 10−10	0.63	2.86 × 108	2.37
	15	2.54 × 10−10	0.99	6.76 × 107	6.9 × 10−10	0.66	1.86 × 108	1.91
	30	2.81 × 10−10	0.99	4.14 × 106	1.91 × 10−9	0.61	4.05 × 107	0.21
	50	3.49 × 10−10	0.97	3.65 × 105	1.47 × 10−7	0.67	1.15 × 106	0.32
Waterborne epoxy varnish	1	2.73 × 10−10	1	5.34 × 108	9.18 × 10−10	0.69	3.80 × 108	8.22
	15	3.63 × 10−10	0.97	3.92 × 107	1.26 × 10−9	0.38	4.59 × 108	4.39
	30	4.30 × 10−10	0.97	3.80 × 106	1.15 × 10−8	0.64	6.98 × 106	1.21
	50	5.18 × 10−10	0.96	2.85 × 105	5.65 × 10−7	0.73	2.92 × 105	0.78

C_p—coating CPE, R_p—coating resistance, C_dl—CPE of the double layer, R_ct—charge transfer resistance, Y₀—admittance magnitude of the capacitor, and n—CPE component power index. Electrochemical equivalent circuit model used: Figure 8b.

. After 50 days, the resistance R_p of the PANI@HNT-2-BTA coating was 1–2 orders of magnitude higher than that of the original varnish, implying that an extended barrier effect could be achieved by the incorporation of halloysite nanotubes. The R_ct was also up to four orders of magnitude. The Y₀ of C_dl of the PANI@HNT-2-BTA coating was at the lowest value (one magnitude lower than the PANI@HNT-2 coating and two magnitudes lower than the waterborne epoxy vanish), indicating the lowest degree of corrosion reaction at the metal surface. The inhibition of the corrosion reaction might be attributed not only to the better shielding effect of the coating, but also to the corrosion reaction suppression from the filler added [42]. Based on the literature [13], PANI@HNT-2-BTA is supposed to act in two ways: (a) the adsorption of BTA leachate onto the metal surface, causing the formation of a BTA complex layer and (b) a PANI-mediated redox reaction, promoting the formation of an Fe₂O₃ passivation layer on the metal substrate. These two effects may cause the synergistic anticorrosion protection of the metal surface. A similar phenomenon was also reported for a nanocomposite composed of PANI and mercaptobenzothiazole (MBT).

3.5. EIS Self-Healing Evaluation of Damaged Coatings

To evaluate their ability to inhibit the corrosion reaction in a damaged area, the PANI@HNT-2-BTA coatings were applied onto the Q235 plates and artificially scratched. The specimens were microscopically examined to verify that the metal surfaces were fully exposed. The EIS of the scratched coating doped with PANI@HNT-2-BTA after immersion in 3.5 wt.% NaCl for up to 48 h is presented in Figure 10 and compared with the coatings mixed with PANI@HNT-2 and the original waterborne epoxy varnish as control. At the initial immersion stage, the coating doped with PANI@HNT-2-BTA exhibited a low impedance modulus. As the immersion continued, the impedance modulus increased. It is logical to assume that this was affected by the released BTA that formed a passivation layer onto the metal substrate (Figure 10a). Z_{f = 0.01Hz} increased to 44 kΩ cm² after 48 h immersion, which was almost two orders of magnitude higher than the initial value of 0.66 kΩ∙cm². In the Bode phase plot (Figure 10b), the phase peak shifted to a higher frequency (from 0.26 Hz to 230 Hz) at the prolonged immersion, suggesting that the corrosion was increasingly inhibited [43]. The coating doped with PANI@HNT-2 was also capable of suppressing corrosion near the scratched area, but the effect was limited. The Z_{f = 0.01Hz} experienced a slight boost from the original 0.76 kΩ cm² to 3.5 kΩ∙cm² after 24 h immersion (Figure 10d). Unsubstantial phase peak shifts were also observed in the Bode phase plot (Figure 10e). The Bode plots of the scratched epoxy varnish coatings are shown in Figure 10g–i. The increment in the impedance modulus in the high-frequency region was mainly caused by the formation of less conductive corrosion products accumulated at the scratched area. The peak in the Bode phase plot shifted to an even lower-frequency region, indicating that the corrosion resistance was further reduced.

To elucidate the electrochemical process in the scratched coatings immersed in saline, the EIS data were fitted into the electrochemical equivalent circuits, as presented in Figure 11. The R_s, R_ox, R_b, and R_ct are defined as the solution resistance, oxide layer resistance, BTA complex layer resistance, and charge transfer resistance of the electrolytes–metal double-layer interface, respectively. C_ox, C_b, and C_dl are the oxide layer capacitor, BTA complex layer capacitor, and the double-layer capacitor at the metal/electrolyte solution interface. For simplicity, the electrochemical impedance parameters with resistances only are presented in

Table 3. Electrochemical impedance parameters of scratched coatings immersed in 3.5 wt.% NaCl at room temperature.

	Immersion Duration	Rox (Ω∙cm2)	Rct (Ω∙cm2)	Rb (Ω∙cm2)	Circuit Model
Waterborne epoxy varnish coating	0 h	517	599	-	a
	8 h	529	858	-
	24 h	388	3316	-
	48 h	254	4378	-
Coating doped with PANI@HNT-2	0 h	529	858	-	a
	8 h	3260	16,190	-
	24 h	810	5760	-	b
	48 h	650	5354	-
Coating doped with PANI@HNT-2-BTA	0 h	582	729	-	a
	8 h	42	5181	3210	c
	24 h	118	35,178	4245
	48 h	141	33,290	6181

R_ox—oxide layer resistance, R_ct—charge transfer resistance, R_b—BTA complex layer resistance. Circuit Model: Figure 11.

. The full set of parameters is listed in Table S2. The oxide layer resistance (R_ox) was measurable due to the formation of metal oxides at the scratched area. It decreased with the passage of the test time, suggesting that the oxide layer was not sufficiently dense to prevent further corrosion. R_ct, an important indicator for the corrosion rate, increased in the waterborne vanish due to the accumulation of corrosion products in the scratched area. The inhibition of the corrosion reaction by these products was, however, very limited. The redox-active PANI could catalyze the formation of passivating Fe₂O₃. This was evidenced by the significant increase in the R_ct of the PANI@HNT-2-doped coating after 8 h of immersion, reaching 1.6 × 10⁴ Ω∙cm² (

Table 4. Neutral salt spray test on scratched coatings.

Sample	Salt Spray Test Duration
	0 h	50 h	100 h	150 h
Waterborne epoxy varnish
Coating doped with PANI@HNT-2
Coating doped with PANI@HNT-2-BTA

). The R_ct values dropped lower as the immersion continued. It is possible that the passive oxide layer formed with PANI assistance deteriorated over time [44,45]. In contrast, R_ct together with R_b increased steadily in the PANI@HNT-2-BTA-doped coatings during the immersion test. The barrier layer formed by the BTA-ferrous/ferric ion complexes contributed to the R_b and R_ct increments [46]. Thus, after 48 h of immersion, the R_ct was eight times higher than the same varnish parameter. It was evident that the newly formed BTA complex served as a healing agent, covering the exposed metal surface and preventing its corrosion.

3.6. Salt Spray Test on the Scratched Coating

The scratched coating doped with PANI@HNT-2-BTA was subjected to a neutral salt spray test and compared with a coating doped with PANI@HNT and a waterborne epoxy varnish. The digital images of the sample panels during the test are shown in Table 4. Consistent with the EIS results, the PANI@HNT-2-BTA coating demonstrated the best corrosion inhibition, showing no obvious signs of corrosion. It was noticed that the scratches became narrower and less identifiable, suggesting the “self-healing” propensity of the coating. The coating infused with PANI@HNT-2 only started to corrode in the inscribed area after 100 h and corrosion became more apparent after 150 h. This coating demonstrated a certain limited ability to suppress the corrosion in the damaged area. Finally, the waterborne epoxy varnish bearing no corrosion inhibition additives started to blister after 50 h and the corrosion area began to spread after 100 h. After 150 h, severe corrosion and coat delamination were observed, evidencing the failure of the epoxy varnish.

Based on the EIS and salt spray tests, a tentative illustration of the corrosion inhibition mechanism is presented in Scheme 1. When a damage or defect appeared in the coating, the corrosive Cl⁻ ion promoted the oxidation of Fe to Fe²⁺ [47,48]. At the same time, the large surface area and intrinsic hydrophilicity of the HNTs could facilitate water molecule absorption, thus triggering the release of the hydrophilic BTA from the nanotubes [49]. Simultaneously, PANI could facilitate the oxidation of Fe²⁺ to Fe³⁺ [50]. The released BTA adhered onto the exposed metal surface and chelated the ferrous/ferric ions, forming a passivating layer composed of a BTA-ferrous/ferric ion complex which blocked the further interaction of the corrosive media with the metal substrate. This newly formed layer composed of [Fe²⁺(BTA)₂]_n, [Fe²⁺(Cl⁻)_mBTA]_n, or other types of [Fe³⁺]_m[BTA]_n [51], filled the void in the existing epoxy coating and “healed” the wound.

4. Conclusions

In summary, PANI-modified HNTs were prepared via polymerization in situ. PANI@HNT-2 exhibited the best loading capacity and retention towards BTA as a guest molecule. The EIS analysis with immersion tests performed on Q235 steel plates coated with PANI@HNT-2-BTA-loaded varnishes demonstrated their positive corrosion inhibition effect compared to the control experiments with common water-born epoxy coatings. The EIS examination of the results from the accelerated corrosion tests revealed that the coating with 5% PANI@HNT-2-BTA was the most effective. An overdose of additives negatively affected the corrosion protection of the coatings. The long-term EIS tests confirmed that the corrosion protection performance was improved by adding BTA-loaded PANI@HNT-2 to the standard epoxy coatings. The tests also demonstrated the synergistic effect of BTA and PANI, promoting the formation of passivating film(s) over the exposed area. Overall, we developed a facile strategy for the preparation of ready-to-use pigment to improve the corrosion protection of water-born epoxy coatings and apply its “self-healing” mechanism to fight against external damage. Future studies will be focused on the precise control of the release kinetics of the organic inhibitors by extending the library of the polymers wrapping the HNT and encapsulated inhibitors. Optimizing the coating system elements such as binders, additives, and other fillers also needs to be considered to maximize the corrosion suppression.