Multiwalled carbon nanotubes (MWNT-30 type nanotubes with average diameter of 40 nm, supplied by Shenzhen Nanotech Port Co. Ltd., China) were dispersed by ultrasonication in the mixture of aqueous ethanol and acetic acid (by the volume mixtures of 10:1:3) twice for 30 min. To ensure enhanced dispersity as well as stability of the dispersions and to facilitate exfoliation, poly(4-ammonium styrenesulfonic acid, 30%, Aldrich) and sodium dodecyl sulfate (98%, Aldrich) were used and are later referenced as PSS and SDS, respectively.

For preparation of the nano-size particles, pyrrole (98%, Aldrich) was dissolved in water-dispersed sol of fumed alumina (AluC, BET surface: 100 m² g^− 1, Evonik Industries AG, Germany) and the colloid was stirred intensively for 2 h. In case of the alumina-supported PPy type particles, the procedures of polymerization and deposition were carried out in aqueous ethanol (at a content of 13% by volume) dispersed alumina sol. Initial pyrrole concentrations were 1.94 × 10^− 2 and 1.192 × 10^− 1 M, whereas alumina contents were 2.44 and 4.76 wt.% solely to the alumina and the nanotube containing dispersions, respectively. PSS content and SDS concentration of the nanotube containing mixture were 2.15 wt.% and 2.43 × 10^− 3 M. For oxidative polymerization of pyrrole, the solution of iron(III) nitrate (97%, Fluka) was added to the vigorously stirred sol to obtain an iron(III)-pyrrole molar ratio of 8 × 10^− 1. When the mixture became homogeneous, 1 M nitric acid solution (~ 1 ml) was added immediately to set the pH to 3. Then the mixture was stirred slowly for 6 h then let stagnant for 16 h. Suspensions were filtrated and the solid was washed 12 times (to remove SDS), then particles were dried and ground. Composition of the nano-size particles is summarized in Table 6.1.

6.2.1.2 Preparation of paint coatings

Paint coatings were layered on standard low-carbon, cold-rolled steel panels (RS type CRS, roughness; 25-65 micro-inches, complying with ASTM A1008.1010, A-109 and QQS-698 standards) used as received from Q-Lab Ltd. Particles were dispersed by 20 min mortar grinding and milling in dissolved epoxy (at a polymer concentration of 50 wt.%).

The organic solvent was composed of 80:10:10 volume % of xylene (Fluka), 1-methoxy-2-propanol (Fluka) and 2-butanon (Aldrich). Suspensions of the particles were mixed with component A—stabilized zinc-rich masterbatch (HZO Farbenzinkstaub, Norzinco GmbH) in epoxy (Epoxidharz CHS141, bisphenol-A epoxy resin, Prochema). Component B—poly(amido amine) cross-linking agent (Durepoxy H15VP, USNER) was added to the suspensions, stirred to homogenize and diluted to get desired wet concentration. Primer coatings were roll-blade casted on steel panels in 90 μm wet thickness (35 ± 5 μm dry thickness) and let cure for a week at room temperature. Top-coatings were layered on hardened epoxy in an overall wet thickness of 120 μm (80 ± 5 μm dry thickness) with Macrynal SM 2810/75BAC hydroxyacrylic resin (CYTEC Industries Inc.) cured with aliphatic polyisocyanate resin: hexamethylene diisocyanate (Desmodur® N75MPA/X, Bayer Material-Science LLC) which was dissolved in the mixture of n-butyl acetate, xylene, and isobutanol (2:2:1). Coatings were let dry and cure for three weeks at room temperature. Scribes were made by a blade as X-cuts. Composition of the paint coatings is detailed in Table 6.2.

Cyclic voltammetry is able to reveal characteristic differences and minor changes in electrochemical activity of the intrinsically conducting polymers, including their derivatives and composites. In addition, due to the mechanism of mediated electron transfer reaction, anodic corrosion inhibition function of the particle samples is strongly influenced by redox activity of PPy and conductivity of the nano-size particles regardless of its free-standing or supported form, applied as a stand-alone film or in composites—paint coatings. Therefore, particles were voltammetry tested, and the most representative scanning results are shown in Figure 6.1.

Moderate electroactivity of the PPy was noticed during investigation of the p1. The redox activity related capacitive current density well exceeded the double-layer capacitance and adsorption resulted background current measured with bare platinum electrode. The first capacitive current response and consecutive scanning gave similar curves with the location of oxidative peak potentials at − 0.1 V, but much lower oxidative-reductive charges, indicating good electrochemical reversibility. The low and progressively decreasing current transients by sequential scanning and the low charging efficiency (Table 6.1) are due to the very low electrical conductivity of p1, containing the undoped semiconducting type polymer as a thin superficial layer on the alumina at the contents (~ 4 wt.%) of just around the percolation threshold in the sample.⁶⁸

In case of the p2 powder sample, orders of magnitude greater than anodic- cathodic capacitive current transients were measured not just as first scanning but as consecutive curves, which reflect far higher electrical conductivity. Oxidation peak potentials were easily recognized at 0.0 V, but reduction peak potentials could only be scarcely identified at − 0.35 V, meaning noticeable electrochemical reversibility of the PPy content. The voltammogram is less affected by the lower ohmic potential drop during the measurement as a consequence of high conductivity of the sample. This is due to the geometrical, statistical percolation of the MWCNTs and so the result was electrical percolation in the p2 at MWCNT content of 14 wt.%, whereas the onset of electrical percolation was found at 9 wt.% nanotube contents. The transient profile and the scale of capacitive currents with coulombic efficiency of 2.1 ± 3 × 10^− 1 at low relative amount of PPy manifest altogether mediocre electrical conductivity of the partially doped PPy, supported partly by the PSS modified nanotubes. Shift of the peak potentials compared to neat and alumina supported PPy is partly attributed to a greater extent of conjugation and less electron deficient state of the PPy because of the high relative amount of the nanotubes and effective electron transfer from them. However, hindered charge mobility through the network is connected to the sterical structure of PPy deformed by PSS, causing limited conjugation in the polymer segments. This is because the low amount of homogeneously dispersed nanotubes was modified with large amount of immobile PSS complexed PPy film, and the polyelectrolyte limits the extent of conjugation in PPy by distortion and conformational adjustments leading to energy barrier to slow down electron conduction.⁶⁹

6.3.1.2 Fourier-transform infrared spectroscopy

In Figure 6.2 FTIR spectra of the nano-size particles are compared to each other and the free-standing PPy samples prepared in water (PPy1) and aqueous ethanol (PPy2). In the PPy1 spectrum the characteristic bands of PPy are found at 1528 and 1467 cm^− 1 corresponding to the νCC and νCN stretching vibrational bands of the pyrrole ring (arrows), respectively, while the bands at 1285 and 1142 cm^− 1 belong to the CH deformations.^70,71 The weak, but well-defined band at 1093 cm^− 1 corresponds to the in-plane deformation mode of NH₂⁺ of protonated PPy backbone.⁷² In case of the PPy2, a doublet at 1554 and 1531 cm^− 1 in the CC stretching region was identified, which was attributed to the combination of intra-ring CC and inter-ring CC stretching vibrations.⁷³ Furthermore, slight blue-shift of the peaks suggests a more compact 3D structure of the PPy2. As for the alumina supported PPy (p1), spectral features of the polymer film changed. The new bands at 1634 and 1340 cm^− 1 are assigned to the deformation vibration modes of protonated species, NH₂⁺ of doped PPy. The broad, intensive band around 1120 cm^− 1 corresponding to the CH in-plane deformation vibrations suggest a compact 3D structure of the thin polymer film. The blue-shift of CC/CC inter and intra-ring vibration (arrow) can also be explained by enhanced electron delocalization in the PPy segments.⁷⁴

In the spectrum of MWCNT incorporated particles (p2), the main CC/CC band of PPy inter- and intra-rings around 1550 cm^− 1 (arrow) is slightly blue-shifted compared to the one of the neat polymer (around 1530 cm^− 1).⁴⁶ Supramolecular assembly is partly accounted for the effect, leading to more efficient, enhanced electron delocalization along the PPy network. However, suppressed CH deformations at 1294 and 1047 cm^− 1 suggest a coplanar (characteristic) structure of the PPy segments throughout the backbone. Blue-shifted bands (at 1575 and 1220 cm^− 1) agree with the deprotonated state of the PPy,⁷⁵ resulting in moderate electrical conductivity but good electroactivity based on high redox activity and reversibility of the polymer. Typical thin PPy film structure is indicated by the low intensity bands of CH in plane bending and CN stretching at 1038 and 1173 cm^− 1, respectively. As to the PPy/PSS complexes, lower PPy bands intensities were detected and partial incorporation of the PSS into PPy modified nanotubes-alumina was evidenced by the PSS bands in the spectrum. Despite the masking phenomenon of SO bands at ~ 1220 cm^− 1, a well-defined νCN stretching band was detected at 1180 cm^− 1, which is in connection with the PPy conductivity mechanism undergoing via electron hopping by bipolarons near NC bonds.⁷⁶ No sign of oxidative degradation, that is, the presence of carbonyl bands in the spectrum, was detected.

6.3.1.3 Transmission electron microscopy

In Figure 6.3a–f, TEM images of the nano-size particles are presented. Figure 6.3a depicts the loosely aggregated flocs of the alumina-supported PPy particles at submicron- scale, interacting through moderate spatial density of interlinks at micron-scale. The highly and evenly dispersed particles exhibited markedly extended interparticle interactions, infinite association, and 3D cluster formation of the particles (p1). By taking observation at higher magnification (Figure 6.3b), the reason for the well-dispersed and associated distribution of the particles became obvious. PPy modification of alumina leads to a minor proportion of coalescencelike aggregation of the particles (arrowed by A), the majority of bridging type—loosely aggregated particles (arrowed by B) and some fraction of almost individually dispersed particulates with thin PPy film coatings. The latter ones were core-shell type particles (B and C) that were merged to give globularly structured multiphaselike structures (A) found in moderate proportion around the size range of 200 nm with weak tendency toward agglomeration. The complete coverage with thin PPy film could be well observed (Figure 6.3c), and size distribution of the aggregates showed smaller contribution around 200 nm (A), predominant fraction at 100 nm (B), and much smaller proportion below 100 nm (C).

As it is shown in Figure 6.3d, the nanotube incorporated sample indicated much lower dispersity, more pronounced aggregation (A) of particles (p2), and loosely assembled arrangement on micron-scale (B) with finite size of aggregated particle-floc interactions (C). The PPy modified alumina was partly strongly (Figure 6.3e, A) and less coherently (B and C) aggregated with the nanotubes, but not evenly distributed along the MWCNTs (B and C). However, the nanotubes were almost evenly dispersed and their filaments supported microgel flocs with extended 3D interconnection of the particulates with low spatial density. PPy deposition resulted in reversible aggregation and some degree of coalescence of the alumina (Figure 6.3f, A) around the MWCNTs (B) but all of the alumina and nanotube support was completely covered with thin polymer film (C). Large-scale aggregation of the p2 compared to the p1 is due to the codeposition of PPy/PSS complexes causing a double-flocculent effect. The p2 sample is regarded as a nonuniform, inhomogeneous particle composed of anisotropic filaments and nano-size isotropic particles in the form of randomly distributed submicron voluminous island-like arrangements. Although cluster topologies and dispersibility of the p2 was far inferior to the p1, its electroactivity allows effective improvement of galvanic function of the ZRPs with zinc contents lower than critical PVC, while the p1 type particulates can only facilitate inhibition of the sacrificial metallic pigments and may enhance the barrier nature of the paint coatings.

6.3.1.4 Rheology characterization of dispersion of the nano-size particles

Dynamic measurement results are presented in Figures 6.4a and b. Strong viscoelastic response of dispersion of the p1 (D1) is revealed by the high storage and the lower loss moduli over a wide-frequency domain (Figure 6.4a). Pseudo-solid flow behavior is associated with the medium spatial density interlinked, coherent microstructure of the sol. The elastic-viscous transition with an intersection at ~ 3 Hz reflects moderate stress-induced deterioration of the 3D associated colloid system. The steeply increasing loss factor also indicates such a quasi-solid and viscous transition. Considerable ramp in the complex viscosity is attributed to the large extent of interparticle interactions of the highly dispersed filler phase. The plateau between 10 and 100 Hz in the loss factor supports moderately high strain rate yielded deterioration in the colloid structure.

The amplitude sweep test revealed two stages of stress-activated attrition in the macro- and microgel structure of the D1 along with elastic-viscous transition signified by intersecting storage and loss moduli at 10 Pa (Figure 6.4b). The former is connected to the less extensively interlinked floc system and moderately coherent association of the particles within the microgel flocs, whereas the latter is most probably in accordance with the stress amplitude (critical stress level) induced collapse of the coherent pseudo-solid gel system. However, decreasing complex viscosity confirms effective shear-induced reorientation, disruption of 3D particle associations.

As for the D2, the wide angular frequency range constant ramp of low storage and somewhat higher loss moduli besides the high phase angle in the loss factor over wide range mean a viscous flow characteristic with the absence of effective 3D particle associations and coherent microstructure of the colloid system. Nonetheless, the intersection of moduli at 50 Hz and the sharply decreasing phase angle unanimously reveal a transition from sol to gel behavior that is associated with the straining caused structure build-up, loose and low spatial density interaction of the microgel flocs in the dispersion. This is confirmed by the stress sweep test that indicated out-of-range intersection of ~ 0.4 Pa.

The lack of hydrogen bonding between alumina and the dispersant led to inefficient interfacial interactions. So, hydrated alumina repels much of the alkyd matrix, since insignificant rheology control of unmodified alumina is connected to the weak van der Waals forces between the solid and the vehicle. Thus, Newtonian flow of dissolved vehicle and its dispersion with bare alumina gave constant complex viscosity with wide range linear ramp of storage and loss moduli associated with rapid relaxation processes. Greater loss and lower storage moduli evidenced viscous flow characteristic due to the lack of association of unmodified alumina and coherent microstructure of the colloid system. High surface/volume ratio and dispersity of the alumina led to increased hydrodynamic resistance, greater volume exclusion effect of the nano-size filler. Therefore, remarkable interfacial forces in the D1 are obviously related to the thin film PPy content interacting effectively with the lipophile matrix because of the match of solubility parameters.⁷⁷ Enhanced interfacial attractive forces between the liquid phase-dissolved polymer matrix and the highly dispersed particles are responsible for the high mobility restriction of the polymer segments and the effective load transfer related massive flow resistance.

Constant rate and stress load results are shown in Figure 6.5a and b, whereas creep test data are presented Figure 6.5c. In case of the visco-elastic type D1, the high strain rate domain located large hysteresis and the different slope of shear-thinning effect are connected to the various strength and scale of particle associations and rearrangements (Figure 6.5a). The high forward (fw.) and backward (bw.) apparent dynamic viscosities at all strain rates leading to considerable thixotropy index mean moderate reorientation rate of the particles and slow recovery of the colloid structure. Besides the large thixotropy, continuous shear-thinning is connected to increasing deterioration and disruption of the interlinked flocs with subsequent gradual floc attrition-activation of bridging and coalescence aggregated or flocculated particles. Nevertheless, all interparticle interactions seemed to diminish with durable straining at high rates, leading to complete disruption of the microgel floc system which was indicated by the asymptotic hydrodynamic flow resistances at the heist rate (the same was measure to the alumina dispersion). The high yield stress (Figure 6.5b) and the remarkable creep resistance at low stress loading (Figure 6.5c) suggest a highly interlinked floc system, well-interacted microstructure of the highly dispersed p1 particles (featuring high, effective volume fraction).

As for the rather visco-plastic type D2, the high shear-thinning in a pseudo-plastic viscous way suggests fast rotational orientation of the nanotube filaments with low hindrance to relax and build up (Figure 6.5a). Moderate thixotropic nature (28,350 Pa s^− 1) and the lower yield stress (6.5 Pa, Table 6.3) also support the idea of rapid relaxation and fast reorientation of the nanotube-loaded p2 particles. Apparent viscosities were high despite volume fraction of the nanotubes with an average aspect ratio of 100 were under the rheology percolation threshold at volume fraction of 7 × 10^− 3,⁷⁸ in the semidilute regime of the bare MWCNTs. Furthermore, fast rupture of floc interactions (slow recovery) and the gradual minor attrition of anisotropic filaments (faster recovery) are concluded along with a mediocre particle dispersity and the resulted less effective load transfer, low stress threshold-resistance to yield to flow and creep (Figure 6.5c). In dispersions of the p2 type particles with lower relative amount of the nanotubes (between 8.5 and 14 wt.%), no effect of infinite cluster formation, geometrical (statistical) or kinetic percolation of the nanotubes and their filamentous particles but high thixotropicity was experienced by all measurement methods. So, electrical percolation of the p2 by its higher relative MWCNT content is suggested to transfer into statistical and kinetic percolation threshold measured by rheological methods.

OCP and impedance data acquired during the immersion test are presented in Figure 6.6a and b–f with assessment of the samples summarized in Table 6.4. Based on the evaluation of impedance data, fitting parameters of discrete elements of the most probable electrical circuits are given in Figures 6.7 and 6.8. The traditional type zinc-rich Z coated steel panel exhibited mixed potentials below − 0.9 V over the first half of the test (~ 3000 h), which means stable and viable galvanic function of the primer, cathodic polarization of the steel surface, and thermodynamic immunization protection of the steel substrate against corrosion. However, this should partly mean inefficient corrosion inhibition of the substrate by the interfacially precipitating corrosion products. This long-term stage of active protection was allowed by the highly porous primer and the intense anodic action of the zinc. Weak barrier nature of the Z is reflected by the small and decreasing impedance moduli besides the high low-frequency phase-angle minimum and breakpoint frequencies shifting toward the high-frequency range (Figure 6.6b), which is a consequence of inefficient pore-blocking effect. After an initial deactivation at 552 h, continuous drops in the impedance moduli correlated with the slowly anodic shifting OCPs. Uniform surface polarization, partially diminishing galvanic function of the Z became obvious after 3000 h, whereas evolving some mechanical blockage and recoveries of the barrier nature could be identified at 552, 4800, and 6100 h of immersion. Intense sacrificial and self-corrosion of part of the zinc pigments (around 30%) was appointed by cross-sectional SEM observations, finding corroded phase at grain boundaries. Thus, intense sacrificial action of zinc led to high rate of oxygen reduction reactions and obviously the evolution of strongly alkaline milieu especially at the interface. So, the primer severely delaminated relating to oxidative degradation,^79,80 and saponification of the binder, no macroscopic corroded spots were detected on the steel surface. Interestingly, some degree of cathodic protection was even available at 6000 h, obviously in inhomogeneous lateral distribution as corrosion potential of the system remained around − 0.8 V, which is considered to be a threshold to provide enough anodic current output for cathodic immunization of steel substrates (under fairly protecting barrier coatings) even in performance demanding circumstances.⁸¹

The OCPs between 0.2 and 0.3 V and the high impedance moduli (Figure 6.6c) signify markedly slow initial activation by the electrolyte, evidencing a very efficient barrier character of the H1 lasting for around 1500 h. These data suggest significant resistance control of the coating (unfortunately masking the corrosion potential) besides a remarkable shrinkage in the specific anodic area of zinc. Although some activation of the primer and its zinc content with a highly restrained galvanic action are thought to take place based on the OCP having shifted to around − 0.3 V at the time of 2000 h, it should not be considered as a capacitive feature of the coating. Accordingly, the impedance spectra revealed ignorable changes in the inbound mass transport regime, that is, electrolyte infiltration processes. The wide range of nonideal capacitive phase response is attributed to the nonperfect geometry of the measuring cell with its inhomogeneous current distribution and the uneven coating surface both at the primer/top coat and the top coat/solution interface. After a minor activation at 2000 h, durable passive protection by the H1 coating seemed to recover at ~ 2400 h of the immersion and lasted for over 3000 h, owing probably partly to the very low amount of accumulation of corrosion products. Remarkable anodic shift of the OCPs to around 0.4 V and the considerable resistance control (above 10¹⁰ Ω cm² by the coating) mean very low rate of oxygen reduction reaction undergoing at the steel substrate acting as an oxygen electrode as it was given by Evans.⁸² It was pointed out that corrosion potentials and oxygen reduction on passive films of metal oxides and thermal oxide electrodes occur at 0.4 V and in galvanic coupled cases at 0.2 V versus SCE.⁸³ In comparison with the initial state of the primer, it means notably small active anodic areas and low rate of zinc dissolution. However, the firm barrier nature of the H1 seemed to deteriorate at 5000 h when OCPs turned to change to the cathodic direction and decreasing low-frequency impedance was accompanied with increasing low-frequency phase-angle minimum. According to the robust passive character, no impact of the corrosion processes such as delamination, blistering, and steel rusting were experienced.

Similar function of the H2 coating was noticed based on the slightly fluctuating but continuously increasing corrosion potentials and the durable highly resistive behavior (Figure 6.6d). A slightly altered protection characteristic of the H2 hybrid (with some similarities partly to the Z and the H1) is recognized both in the corrosion potential pattern and impedance data as a function of time. The measurement of more anodic OCPs at ~ 1000, 3000, and 4000 h all coincided well with the increasing capacitive nature of the coating, giving bit higher impedance amplitudes and a lower phase minimum in the low-frequency domain. Steady changing tendency of the OCPs toward the more anodic range supports the idea of progressively restrained anodic action of zinc (decreasing active surface) and galvanic function of the coating. Impedance spectra revealed coating activation at 1752 and 2520 h by the increasing low-frequency phase-angle even though impedance moduli remained practically unaffected, suggesting a minor proportion access by the penetrating electrolyte-slightly increasing volume and area of pores in the primer. In comparison with the H1, the more frequent activation and recovery of the coating can probably be attributed to the short-term pore-filling action by the smaller amount of accumulating corroded species in a bulk phase of the primer which should be hardly leachable. In addition, as almost the same strongly anodic shifted corrosion potentials were measured with lower resistance control in the case of H2 compared to H1 in the end (as a convergence), which obviously suggest more effective corrosion inhibition at the steel surface by the precipitating and insulating corrosion products and which should also mean the absence of evolution of alkaline milieu in the interface with an altogether low rate of oxygen reduction reaction. As a result, no corrosion spots on the steel surface could be detected (Table 6.4). Nevertheless, some oxidative deterioration might as well happen owing to complete delamination of the coating. This cathodic delamination phenomenon is in agreement with the interfacial electron transfer reactions undergoing even under firm barrier coatings as it was experienced to neat epoxy-top coated system. So, the highly dispersed PPy-coated alumina particles based primers provided efficient passive protection, long-term mechanical blockage besides the lack of evidence of cathodic protecting functionality in immersion tests.

In case of the H3 loaded with nanotube incorporated particles (Figure 6.6e), the OCP and impedance changes evidenced efficient barrier characteristics and some cathodic function in the early activation up to ~ 400 h and the latter steady periods from 2500 h. Considerable passive protection is well reflected by the high impedance amplitudes over the entire test along with the low phase minimum (between − 75° and 55°) either in the early and the late period of the test or the location of breakpoint frequencies in the low-frequency domain (over ~ 1 Hz) measured in the medium and late periods. The somewhat decreasing impedance moduli up to ~ 1300 h are related to very fast activation and blockage of the coating accompanied with anodic shifting corrosion potentials. As electrical percolation of the coating relies on the medium dispersed p2 particles (with far nonpercolating micron-size isotropic zinc), the nanotube supported filaments provide interconnections between both the zinc grains and the steel substrate. The insulation at spanning clusters seemed to slowly diminish as OCPs turned changing in steps toward the cathodic range at 1500, 2500, and 5500 h, ending in a suppressed cathodic potential of − 0.7 V. All these stages were followed with minor recoveries in the barrier function, which were highlighted by the more capacitive coating nature at ~ 1300, 2600, and 6100 h besides the slight anodic changes in the OCPs. The most efficient pore-blocking action was identified at around 2600 h probably resulting in improved cathodic function of the coating. This means some shrinkage in the active anodic surface of the sacrificial pigments in bulk phase of the primer but remaining electrical connection at the steel/coating region causing relatively high oxidative degradation of the epoxy binder in the interface (see Section 6.3.2.3).

Similarly to the Z, the H4 coating led to the evolution of well stabilized corrosion potentials initially at − 0.4 V until ~ 2000 h, − 0.5 V up to 3500 h and − 0.6 V till ~ 4000 h. Based on the corrosion potential changes, moderate variations in the coating state are thought to take place at 500 and 4000 h. From 4000 h to the end of immersion test, OCPs remained − 0.4 and − 0.5 V. Impedance data indicated recoveries in the barrier function; a short-lived evolution at ~ 550 h and a prolonged one from 3000 h lasting to the end (Figure 6.6f). The first is attributed to a minor coating deactivation by the pore-filling corrosion products, which was followed by moderate extent of electrolyte infiltration over the next 2500 h. Rest of the term from 3500 h was recognized as partial inhibition and passive protection by the precipitation of corrosion products botat the coating/steel interface and bulk phase of the primer. The stable OCPs around − 0.5 V suggest moderate electrical interaction of the zinc facilitated by the nano-size particles, resulting in restrained galvanic function and some degree of cathodic protection. The small second time-constant at 0.1 Hz until ~ 550 h manifests redox reactions taking place at continuously decreasing proportion of the steel surface. Both the H3 and H4 coatings exhibited very good condition without blistering. No cohesion or adhesion loss of the H3 was experienced, whereas only minor de-adhesion of the H4 was noticed, hinting on the low rate (or short-term) and small extent (inhomogeneous) distribution of interfacial electron transfer reactions leading to partial cathodic delamination failure. Interestingly, no sign of precipitated zinc corrosion products at grain boundaries was found in cross-section of the H3 and H4 hybrid primers by SEM observations.

By analyzing impedance data based on the applied modeling circuits, pore resistance of the Z decreased until 4500 h in a similar way to the charge-transfer resistance with an absolute minimum around 5000 h (Figure 6.7a and b). Coating capacitance was found slightly affected but mostly following coating resistance changes in a reversed manner as a function time in accordance with our expectation. This was parallel with the continuously decreasing resistance of the finite diffusion element, indicating less hindered mass transport through the coating (represented in the low-frequency domain) whose effect on the impedance spectra could be identified up to 3500 h. Halved exponent of the Warburg element and by several orders of magnitudes greater diffusion coefficient between around 1000 and 1300 h are interpreted as more hindered mass transport of the electrolyte, which is typical porous diffusion through coatings. Interestingly, it did not translate into much greater coating capacitance, but found coincided with markedly increasing double-layer capacitance identified with local and absolute maximums around 1000 and 4000 h, respectively. So, lower scale of dispersion in the Warburg impedance describing finite diffusion should obviously result from the inhomogeneous spatial distribution of mass transport in the primer, partial dissolution and transport of corrosion products in the interface at higher pH (alkaline milieu because of the reduction reactions concentrated on the steel interface) and some precipitation taking place in bulk and outer region of the primer in a neutral or slightly acidic milieu.

Intermittent uptick in the pore resistance around 400, 2000, and 5800 h (measured as greater low-frequency impedances) concur with minor enhancement of the coating barrier nature by the corrosion products which lasted for short-term as a consequence of leaching, that is, outbound mass transport leading to faster depletion of the primer. In addition, double-layer capacitance tended to turn to local minimums as coating resistance became greater, which agrees with the previously mentioned hypothesis. Corrosion inhibition of steel by the precipitating corrosion products is supposed to happen and insulate partially the primer from its substrate between 3000 and 4500 h of exposure as OCPs manifested nonviable or incomplete cathodic polarization of the interface (positive shifted corrosion potentials) besides the enhanced coating activation by the electrolyte. This is in line with the easily detectable swelling spots over the coating and the large amount of corrosion products in cross-section and at the steel surface detected by scanning electron microscope and X-ray photoelectron spectroscopy. Whilst the double-layer capacitance seemed to be well decreasing in the later stage, minor increase of the coating and charge-transfer resistances can be explained by a more stabilized recovery in mechanical blockage and lower rate of electron transfer reactions, especially oxygen reduction reaction in the coating/steel interface. Fading inhibition by the piles of accumulated corrosion products did not entail in ensuing evolving appreciable cathodic immunization. In fact, diminishing galvanic protection of the substrate is connected to partial fading of electrical percolation as a result of accumulation of pore-filling species inside the bulk primer and at steel interface. On the one hand, the several orders of magnitude higher charge-transfer resistance compared to bare steel under similar conditions (R_ct ~ 3.5 × 10³ Ω cm^− 2)⁸⁴ points out valuable corrosion inhibition efficiency throughout the entire test. On the other hand, exponent of the double-layer capacitance at 600 at 5800 h suggests clear impact of the mass transport regime in the coating/steel interface and near bulk region of the primer as parameters of the constant phase element can partially be transferred to parameters of the Warburg element.

In case of the hybrid samples, only one parallel circuit was used for fitting (Figure 6.8a–e). The high barrier type H1 and H3 exhibited enhanced passive protection between 500 and 3200 h while the H2 and H4 showed decreasing efficiency of pore blocking over the same period. In addition, difference between the generally higher pore resistance of the nanotube free H1 and H2 coatings (~ 10¹¹-10¹² Ω cm^− 2) compared to the MWCNT incorporating H3 and H4 counterparts (~ 10⁸-10¹⁰ Ω cm^− 2) was evidenced. Nevertheless, the H1, H2 and H3 coatings are regarded as highly cross-linked, homogeneous I type coatings, whereas the H4 based on its coating resistive behavior is respected as a lower cross-linked, inhomogeneous D type coating. So, greater zinc content with lower spatial density of cross-linking in the primer led to lower pore resistance (despite structure of the filler phase enclosed by the binder), whereas incorporation of increasing amount of surface modified nanotubes leads to the same effect on a much larger scale even though smaller relative amounts of particles and nanotubes were loaded in both the primer and the binder at higher zinc contents (Table 6.2). The profiles of capacitance changes, more exactly the constant changes of the constant phase elements, by time were found similar to the similar formulations, the H1 and H3 as well as the H2 and H4 (Figure 6.8b). Furthermore, pore resistance increases in most cases coincided with decreasing coating capacitance. Despite the lower particle content, constant phase element of the H2 remained consistently above that of the H1, which was comparable to the difference between particle contents of the primers, and some convergence to the end as resistance of the coatings showed intersection around 6000 h. This may be explained by the greater zinc content with its increased initial interfacial area and the effective interaction with the highly dispersed particles increasing porosity of the H2. Active zinc surface and porosity would surely decrease by time entailing in more positive OCPs as coating resistance was slightly affected in the meantime. Interestingly, coating capacitance of the less porous H3 was markedly greater than that of the H4, indicating an opposite tendency compared to the H1 and H2 hybrids. This can be attributed to increased content of the nano-size particles and their geometrically closely arranged aggregates in the form of percolating filaments in the H3. No double-layer capacitance type component could be assigned to the hybrid samples, so corrosion inhibition with interfacially deposited corrosion products can be surmised by comparing pore resistance, coating capacitance, and OCPs. So, remarkable inhibition function and slight cathodic surface polarizing ability of the H4 are suggested, but no such valuable action should be connected to other hybrids as their pore resistances were far higher. Exponent of the constant phase element was close to ideal capacitive behavior except for the H2 manifesting the highest porosity in the early test stage (Figure 6.8c), which became similar to the other coatings in the second half of the test period.

Compared to the electrically percolating H1, H2, and H3 coatings (on different scales), the H4 was both electrically percolating and electrolytically permeating as a consequence of combination of high zinc and specific particle (p2) contents. In comparison with the H3, coating capacity constant of the H4 decreased by the scale expected by the lower PPy and alumina contents in the binder even though its pattern over time was similar to the H2 featuring the same coating formulation. Permeable initial structure of the H4 formulation is reflected by the need of additional circuit elements giving with at least a second time constant in the lower frequency domain (Figure 6.8d and e). Although charge-transfer resistance assigned to the H4 remained at ~ 10⁷ Ω cm^− 2 between 1400 and 4000 h, its later stage magnitudes of ~ 10⁸ Ω cm^− 2 became appreciable connecting to more pronounced corrosion inhibition at the steel surface in comparison with the Z. Double-layer capacitance related constant phase element was more steady and at least an order of magnitude lower (Figure 6.8d) than that was measured to the Z (Figure 6.7a). In partial connection, the greater double-layer capacitance between 1000 and 3200 h correlated well with lower pore resistance of the H4. The exponent of the constant phase element describing double-layer capacitance was around 0.7 lower than the Z by 0.15 on average, which means highly porous interfacial range between the coating and the steel surface (Figure 6.8d). In case of the Z, such an exponent in the last stage of the test is connected to the considerable amount of corrosion products detected on the steel surface obviously affecting surface porosity and the mass transport processes through the porous layer coupled to its outer part in the bulk phase of the primer. This exponent of the H4 became ideal capacitive (~ 1) at 4000 h and accompanied with increased pore and charge-transfer resistance along with lower double-layer capacitance and anodic shifted OCPs, suggesting efficient pore sealing in the interface and the primer. The semi-infinite diffusion modeling type Warburg element was used for fitting mass transport related impedance contributions in the H4. Although its resistance was comparable to the Z, it showed opposite tendency to change from 1000 h. The very low exponent ~ 0.15 (Figure 6.8e) is partly attributed to the highly heterogeneous deterioration at the steel surface (as it was described with a number of unit cells to thick coatings by Mansfeld et al.⁸⁵) and it probably means long spatial range of porous diffusion processes which are partly reflected by exponents of the constant phase elements associated with the coating and double-layer capacitance. However, in comparison with the Z, concentrating in the coating range near the steel surface, much higher dispersion of the Warburg impedance should be accounted as a result of higher pH gradient in the coating/steel interface and much hindered mass transport in the primer; insignificant dissolution and transport of the alkaline dissolved corrosion products and a large proportion of precipitation in the inner phase of the primer under a neutral or slightly acidic milieu.

Impedance data of the H1, H2, and H3 coatings showed low phase angle minimums over the low-frequency domain reflecting the absence of coating breakdown in accordance with the evaluation routines, degradation monitoring, and prediction of coating failures.⁸⁶ This is corroborated by the impedance/frequency slope of − 1 over the 10²-10⁴ Hz range obtained to all coating samples, which is another evaluation routine of coating degradation.⁸⁷ As for the ennobled OCPs measured on the hybrid samples in immersion tests (pronounced resistance control similarly to neat, un-pigmented epoxy), the observations can be explained by the effect of interfacial oxygen reduction reactions undergoing under electrolyte deprived, dry condition as a consequence of highly cross-linked, least penetrated coatings. According to literature data,⁸² oxygen reduction occurs around 0.4 V and (if galvanic coupled) at 0.2 V versus SCE on passive metal oxide electrodes and thermal oxide electrodes, respectively. This is in agreement with Evans;⁸¹ freshly exposed steel substrates with epoxy coatings feature excellent barrier function, exhibiting OCPs of around 0-0.25 V when polymer coated steel reacts as an oxygen reduction electrode. In this potential range, the ratio of corroding area can be lower than 10^− 3% of the overall geometrical surface. As the proportion of the anodic area increases, potential of the coated steel must decrease converging to around − 0.65 V. So, the systems with very good barrier coatings experience considerable potential drops when surface area of the corroding metal substrate is small, since increasing potential is a useful degradation monitor. Because potential of pure zinc in seawater is approx. − 1.050 V¹¹ iron must have acted as cathode in vicinity of the sacrificial pigments in all cases. According to the prediction of Evans,⁸¹ 99.99% of the coating coverage is located in area of the Evans diagram under the current of ~ 10^− 11 A cm^− 2 over any OCP measured for the hybrids. On the other hand, voltage drops were defined in the range of 10^− 3-10^− 2 V to the Z and the H4, showing increasing and steady tendencies over measured periods, respectively. In case of the H1, H2, and H3, much greater voltage drops were calculated around 10¹ and 10⁰-10^− 1 at current densities of ~ 50 pA cm², exhibiting overall decreasing tendencies. So, hybrid samples can be termed as high barrier coatings in agreement with engineering classification based on DC current measurement.⁸⁸

To estimate dry effective relative permittivity-initial capacity of the ZRPs, the exponential mixture rule known as the Lichteneker-Rother model, which was successfully for epoxy matrix based systems of nano-size alumina⁸⁹ and zinc oxide,⁹⁰ was utilized⁹¹ to calculate dielectric constants of the primers composed of epoxy, nano-size alumina, PPy, CNTs, and zinc oxide. In accordance with the presumption and method of the Brasher-Kingsbury relationship,⁹² thickness of the coatings was assumed to remain constant during immersion tests even though polymers swell by moisture uptake. Estimated water uptake of the epoxy binder and the coatings of the samples using the impedance data defined CPE related capacities is as follows; the Z 22-33% and 10-14%, the H1 4-12% and 3-9%, the H2 9-19% and 6-14%, the H3 7-14% and 5-10%, the H4 2-10% and 1-6%, respectively. Furthermore, according to the analysis of a nested pair model based breakpoint frequency evaluation method,^93,94 to make a prediction on relative increase in the defective area besides the standard classification evaluation procedure,⁹⁵ breakpoint frequencies crossing at − 45° in the phase diagram of the Bode represented impedance data and estimated relative increase in the redox active area on the steel surface associated partly with delamination of the coatings were used to appraise coating/steel samples over immersion test periods, which are summarized in the following: the Z over 10⁰-10² Hz with the increase of up to 5 × 10⁴ (10^− 5-10^− 3% overall), the H1 over 10^− 3-10^− 2 Hz with the increase remaining below of 2 × 10¹ (lower than 10^− 6% overall), the H2 over 10^− 2-10^− 1 Hz with the increase of up to 10² (lower than 10^− 6% overall), the H3 over 10^− 3-10^− 1 Hz with the increase of up to 5 × 10² (lower than 10^− 5% overall), the H4 over 10⁰-10² Hz with the increase of just below the 10⁵ times (10^− 5-10^− 2% overall). The latter is also referred to as surface ratio of the pores, percentage of total cell area with damage or defect area percentage of the coatings, which proved to describe well hybrid ZRPs but underestimate the traditional ZRP type Z. This is probably because evaluation requires proportional validity of the breakpoint frequency to the redox active surface area of the metallic substrate by satisfying the following requirements; (1) breakpoint frequency is independent of coating thickness so it is proportional to disbonded area, (2) the area of corroding metal is the same as the pore area, and (3) dielectric constant of the coating is constant over the corrosion propagation period. The second and the third assumption cannot be complied with by the Z but may well be met by the hybrid ZRPs after some days of exposure (at least constant or steady state of the pore and coating surface ratios).

6.3.2.2 Glow-discharge optical-emission spectroscopy

Depth profile data of the main elements detected in primers of the H2 and the H4 are given in Figures 6.9 and 6.10. In line with the firm barrier nature of the coatings indicated by the impedance spectroscopy results, none of the coating elements especially zinc and oxygen was found depleted or enriched in entire cross-section of the primer after the immersion test. In comparison with carbon and zinc as the main elements in atomic % in bulk phase of the primers, no species of the electrolyte solution were detected in greater relative quantities after propagation but most of the elements such as aluminum, sodium, and potassium was found proportional to the zinc content obviously because they could be its impurities. This should mean very low ionic permeability-conductivity of the zinc-rich hybrid coatings including the H4 even though impedance data of the latter suggested porous nature of the coating with pore resistance between 10⁷ and 10⁸ Ω cm².

6.3.2.3 X-ray photoelectron spectroscopy

XPS detected regions acquired from the surface of steel panels of the immersion tested Z and H1-H3 samples are presented in Figure 6.11a–d, where intensities of the peaks correlate with the element contents. Relative quantities of the detected elements and components are considered to be more adequate for qualitative analysis and the data are given in Table 6.5, whereas proportion of the main elements and components related to the result of corrosion processes are summarized in Table 6.6 to have overview on the function and characteristic of the coatings.

Table 6.5

Ratio of Elements Detected by XPS on the Surface of Immersion Tested Paint Coated Steel Panels

Coated Samples	Elements (Region)	Binding Energy (eV)	Chemical State (Relative Proportion)	Content (Atomic%)
Z	Fe 2p	710.9	FexOy (thin Fe(III) rich phase)	7.5
		706.8	Fe(O)	0.08
	Zn 2p3/2	1022.2	ZnO	4.5
		1024.0	Zn(OH)2	0.14
		1019.6	Zn(O)	0.1
	O 1s	530.2	FexOy	18.1
		531.8	OH, OC	19.7
		533.4	OOC	4.2
	C 1s	285.0	CC	25.3
		286.6	CO	5.5
		288.6	CO	4.0
		289.7	CO(O)	1.1
	Na 1s	1072.1	Na2O, NaCl, Na2CO3 or phosphates	1.7
	K 2p3/2	292.8	Halogenide and phosphate	0.3
H1	Fe 2p	711.5	FexOy (thin Fe(III) rich phase)	14.5
	Zn 2p	1021.7	ZnO	2.8
		1022.9	Zn(OH)2	0.8
		1019.6	Zn(O)	0.3
	C 1s	285.0	CC	27.4
		286.4	OH	5.6
		287.9	CO	1.3
		289.0	CO(O)	3.2
	O 1s	531.5	OC, HOFe and Zn	14.7
		530.0	OZn and Fe	21.5
		533.0	OOC	3.9
	Na 1s	1071.5	Na2O, NaCl, Na2CO3 or phosphates	0.8
H2	Fe 2p	711.1	FexOy (thin Fe(III) rich phase)	16.9
	Zn 2p	1021.7	ZnO	4.4
		1022.9	Zn(OH)2	0.2
	C 1s	285.0	CC	25.5
		286.4	OH	5.5
		288.7	CO(O)	3.5
	O 1s	531.5	OC, HOFe and Zn	15.3
		530.0	OZn and Fe	25.0
		533.0	OOC	3.7
	Na 1s	1072.1	Na2O, NaCl, Na2CO3 or phosphates	0.2
H3	Fe 2p	710.7	FexOy (thin Fe(III) rich phase)	5.3
		713.4	Fe(OH)x (partial)	1.6
	Zn 2p	1022.0	ZnO and Zn(OH)2	8.1
	C 1s	284.8	CC	16.6
		286.3	OH	9.7
		287.2	CO	2.3
		289.0	CO(O)	2.5
	O 1s	531.2	OC, HOFe and Zn	16.8
		530.4	OZn and Fe	10
		529.8	OOC	10.4
	Na 1s	1072.1	Na2O, NaCl, Na2CO3 or phosphates	1.6
	K 2p3/2	292.8	Halogenide and phosphate	0.4

Iron was detected as a mixture of the 2⁺ and 3⁺ valence states in the Z and found predominantly in its 3⁺ valence state in the H1-H3 besides occasionally a small contribution of metallic iron which feature obviously came from the metallic phase under the thin oxide film (Figure 6.11a). Oxidized iron phase compared to traces of the binder (carbon) was found in much greater relative proportion in the H1 and H2 (Table 6.5) than in the Z and H3 comparable to each other. In case of the Z even severe delamination and de-adhesion was observed, since the enhanced amount of carbonaceous remnants can be connected to oxidative degradation of the epoxy, which is in accordance with prolonged intense galvanic function of the primer, interfacial electron transfer reactions, and oxygen reduction reactions. Some metallic zinc could be detected in the Z and the H1 but the majority of the zinc oxide was found as dominant phase with inferior contribution of the hydroxide species (Figure 6.11b). Very similar to what was found in connection with the carbonaceous matter, a relative amount of zinc oxide precipitated in the interface (with respect to iron-oxide) showed high ratios in the H3 and the Z and much lower in the H1 and H2 (Table 6.6). This is a sensitive measure of overall rate (over the entire test period) of electrically coupled bimetallic corrosion, that is, galvanic function of the primers, which confirmed intense action of the Z and the H3, and lower rates in the nanotube-free hybrid ZRPs. Although repeated measurements exhibited much lower zinc corrosion in steel/coating interface of the H3 formulation, such a ratio of corroded species generally exceeded the level found in the H1 and H2 type coatings on average (as a consequence of resistance control), which signifies impact of the electrically well percolating structure of the filler phase and the available corrosion preventing the ability afforded to the steel substrate. It is also closely connected to the anodic current output of the zinc-rich primer system and its utilization efficiency for cathodic protection of the steel substrate. In relation to the corrosion inhibition effect by the interfacially accumulating zinc corrosion products, this feature was obviously more efficient in the hybrids than in the Z where the generated alkaline milieu and high porosity of the primer allowed less efficient accumulation of the precipitated corroded species at the interfaces as suggested by the measured corrosion potentials and evaluated resistance and capacities of the coatings. Oxygen containing species indicated varied composition as a result of high relative amounts of oxides/hydroxides of iron and zinc. Regarding the oxygen region of the spectra of the H3, chemical state of zinc could not be resolved to components but corroded iron species were managed to assign to oxides and hydroxides. The facts that no iron corrosion spots were visually observed on the steel surface protected by the hybrids and the Fe 2p spectra showed the same characteristics as the native iron oxides formed under ambient conditions.⁹⁶ This supports the presumption that most of the steel surface became oxidized before the immersion test and the hybrid coating must have provided viable corrosion prevention over the test period.

The overall oxidative conversion-degradation of the epoxy binder was almost the same in all investigated samples (Table 6.6). However, the ratio of oxidized carbon components indicated a characteristic deviation in the H3 in the form of more pronounced hydroxyl type degradation (rather than carbonyl or carboxyl type) when compared to other samples (Figure 6.11c). This obviously means that enhanced electron transfer reactions between the primer and steel substrate should have taken place, whereas catalytic effect of the MWCNTs toward ORR¹⁰ was not suppressed by the semiconducting PPy film.

According to the tendency of galvanic function of the coatings and oxidative degradation of the organic binder (besides the consequence of the various cross-linking degree of the binder depending on the relative amount of solid inclusions), considerable amount of sodium was found both in the Z and the H3 along with potassium (clearly above the detection limit and in greater ratio to sodium than in the electrolyte) (Figure 6.11d) but low amounts were found in the H1 and H2 close to the detection limit without any sign of potassium.

6.3.2.4 Salt-spray chamber test

Evaluation of the samples after salt spray testing is summarized in Table 6.7, whereas photographs of the samples are shown in Figure 6.12. Test results of the H1 type coatings pointed out firm barrier function, considerable resistance against delamination and prevention of iron rust formation around the scribed areas. The scale and later stage deterioration of the coating suggested good protection. The H2 exhibited an altogether much better protection characteristic with very good active corrosion prevention around the scribed areas and resistance against delamination. Even buried areas seemed to be in a better preserved state with a lower scale of coating degradation, which supports the notion of quite good semiactive/passive protection nature. Only some proportion of the buried areas were scattered with a small size of blisters but only the larger ones, about half of them in number, were related to iron rust formation. These are the result of optimal hybrid formulation; high zinc and the low specific particle contents, leading to low porosity and electrolytic conductivity besides good utilization of the low anodic current output for steel protection. Robust protection nature is emphasized by the almost uniform way of small scale corrosion around scribes, lower degree of staining and the later stage of coating deterioration along with the very low migration progress of delaminating fronts.

Both the nanotube-loaded H3 and H4 samples exhibited very good condition with the lowest scale of blistering over intact areas along with good and excellent resistance against delamination around scribed areas, respectively. Lower susceptibility to blistering over buried areas is in accordance with the directly non-interconnected or moderately interconnected structure of fillers (measured as threshold of rheological and electrical percolation by the nano-size particles), allowing to afford dominating passive protection feature of the coatings. Nevertheless, a notable difference was identified in their corrosion prevention functionality; the rate of staining, amount of iron rust formation was far the best in the H4, especially in the intersection and an altogether good condition of the H3 was noticed.

Due to the high zinc PVC, coating porosity and cross-linking density of the binder, the Z afforded only moderate resistance against blistering, but viable corrosion prevention at scratched areas. It also provided an acceptable degree of protection over the intact surface against rusting and blistering, but the latter was clearly inferior to most of the hybrids. This is the consequence of an increased rate of coating deterioration partly because of the accumulation and subsequent leaching of the massive amount of corrosion products, which was noticed to occur after 40 days of exposure.