Self-healing corrosion protection by phosphate-doped enamel coatings on steel in simulated cement pore fluid

Xiaoming Cheng1,a

1Department of Materials Science and Engineering

2Department of Civil, Architectural, and Environmental Engineering

3Department of Mechanical & Aerospace Engineering

Missouri University of Science and Technology, Rolla, MO 65409

axcpn@mst.edu;

ABSTRACT

Phosphate was incorporated into enamel coatings by doping sodium borosilicate glasses with 7 mol% P2O5. Corrosion resistance of phosphate-doped (7P) and base enamels (0P) in simulated cement pore fluid (Lawrence solution) was evaluated, with the focus of studying the effect of artificial damage on the protection property of the coating. It reveals that after three days immersion in Lawrence solution (LS) 7P coating developed a precipitate layer on the defect, whereas 0P shows initialization of pitting corrosion in the exposed substrate at the defect. The precipitate layer was characterized by Micro-Raman and X-ray diffraction, suggesting formation of hydroxyapatite. Linear polarization of damaged coatings shows active corrosion for 0P through all immersed times. However, there is a passivation gradually developed for 7P coatings with increasing immersion time. Electrochemical impedance spectroscopy was also used to study the corrosion resistance of damaged coating. By equivalent circuit simulation, it reveals that the charge transfer resistance of 7P coating increased as a function of immersion time and is two orders of magnitude higher than 0P coating after three day immersion. It suggests that 7P coating exhibit active corrosion protection, which is not only protect the substrate with a nonconductive barrier but provide continued protection to the exposed substrate after partial damage of the coating. The self-healing property is believed to be attributed to the Ca-P precipitates which impede the movement of

electrolyte and corrosive species, slowing the corrosion rate.

1. Introduction

In view of coatings for corrosion protection of metallic substrates, organic coating, metallic coating and porcelain enamels are the most common coatings used in practice. To achieve long-term corrosion protection, the following requirements must be met for the coatings: a) high mechanical resistance and adhesion, especially during transport and installation; b) chemical stability under service conditions (aging); c) sufficiently low permeability for corrosive component under service conditions; d) sufficient stability under electrochemical influences, especially with electrochemical protection measures. Organic coatings show varying degree of solubility and permeability for components of the corrosive medium. In addition, organic coatings in general contain many polar groups that promote adhesion or pretreatment on substrate (conversion coatings) to achieve necessary bond strength against peeling. Once defect is created on organic coatings, however, permeation of oxygen and corrosive species through the coatings initialize the cathodic reaction and complete the electrical cell, while the exposed substrate at the defect oxidized as the anodic reaction, which leads to disbonding of the coating and catastrophic corrosion. With regard to metallic coating, it has the advantage of providing electrochemical protection by preferential oxidizing the metal coating instead of substrate metal due to galvanic corrosion. The intermetallic layer developed

1

during the heat treatment between the metallic coating and substrate promotes the bonding and prevents delamination. However, as sacrificial coatings, metal coatings will ultimately corrode away and exposing the underneath substrate and also direct electrical contact between uncoated and coated parts has to be avoided for the same reason.

Porcelain enamel displays an exceptional ability to withstand harsh environments. It is now widely used on ovens, clothes washers, and bathtubs. However, not all porcelain technology is aimed at the appliance industry. Vitreous enamel has been applied on reinforcing steel rebar to protect the steel from corrosive environment. Recent studies conducted at Missouri S&T [1] compared the corrosion resistance of pure enamel on steel rebar with ECR. It was found that pure enamel coatings were significantly outperformed by the intact epoxy coating. However, in the presence of defects on coatings, pitted corrosion was initiated at the location of damaged pure enamel coatings but restrained locally due to well-adhered glassy layers on rebar surface. As a result, epoxy coatings exhibited significantly degraded corrosion performance with local damage due to the well-known under-film corrosion mechanism. Porcelain enamel coating holds great promise in corrosion protection and extending the service life of reinforcing steel in concrete structures. The superior corrosion resistance can only be maintained when the enamel coating remains intact. Coatings lose the protection property when their thickness is significantly reduced or cracks/defects form. Therefore, active corrosion protection based on self-healing of defects in coatings is a vital issue for development of new advanced corrosion protection coating system.

Considering the exceptional corrosion resistance of vitreous enamel coating on metal surface and the loss of this protection property with the presence of defects, the present work is intended to investigate the corrosion resistance of phosphate-doped enamel coatings on steel rebar in simulated concrete pore solution, using electrochemical methods in an attempt to shed light on active corrosion protection of inorganic enamel coating systems. At the same time, it is very important to understand how enamels dissolve under such environments and to understand

the role of phosphate incorporated in the glass to provide active protection of the steel.

2. Experimental

2.1. Glass preparationTable 1 shows the composition of a typical, commercially-available borosilicate enamel that was used on reinforcing steel [2]. This composition has the thermal properties (coefficient of thermal expansion CTE, softening temperature Ts) to form a mechanically strong bond to steel at a relatively low (~800 °C) temperature. For the present study, compositions from Na2O ∙ B2O3 ∙ SiO2 systems were prepared, with and without P2O5. Table 2 shows the compositions of glasses in this study.

Table 1 Chemical composition of a commercial borosilicate enamel [2].

Oxides Concentration (mol%)SiO2 49B2O3 19Na2O 17K2O 2CaO 0CaF2 4Al2O3 3ZrO2 3MnO2 1NiO 1CoO 1Total 100

Table 2 Nominal glass compositions xP2O5 ∙ (100-x)(25Na2O ∙ 25B2O3 ∙ 50SiO2) in mol%.

SBNxP Na2O B2O3 SiO2 P2O5

0P (x=0) 25 25 50 -7P (x=7) 23.25 23.25 46.5 7

The sodium borosilicate and sodium phospho-borosilicate glasses were prepared using reagent grade SiO2, H3BO3, Na2CO3 and (NaPO3)n. The initial batches were melted in alumina crucibles for 1 h at 1200 °C, and then the melts were quenched in a steel mold. The resulting glasses were annealed at around 530 °C for 3 hr.

2.2. Enamel preparationGlass frits were crushed using shatter box to

reduce the particle size to less than m. To create

2

a solid suspension, 20 g of glass powders were mixed with 150 ml of ethanol and the mixture was stirred at 450 rpm with a stirring bar. Discs of 1.7 cm in diameter and 5 mm in thickness were cut from steel rebar and polished to 600 grit. These discs were loaded by a copper wire basket into the suspension. When the stirring stopped, glass powders were deposited on steel surface uniformly by gravity. The deposition set-up is shown in Fig. 1.

During the firing process, coated steel pieces were slowly heated at 10 °C/min to 200°C and held for 1hr to burn off the ethanol. Then the temperature was ramped up to 800°C at 40 °C/min and held for 10min before the samples were cooled down to room temperature with fast cooling. Figure. 2 shows the appearance of the coating before and after firing.

Figure 1 Deposition set-up to apply thin film of glass powders uniformly on steel surface. The steel discs were

loaded in the middle of copper wire basket and immersed in the glass-ethanol suspension.

Figure 2 Enamel (0P) coated steel samples before (a) and after (b) firing.

2.3. Immersion testsIn order to evaluate the corrosion resistance of

these coatings through electrochemical methods, the coatings were deliberately damaged using a drill with 1/16 inch drill bit to expose the substrate. The area of

defects is between 0.006 cm2 – 0.02 cm2. The damaged coating was immersed in simulated pore water (composition shown in Table. 3) for 3 days. All tests were carried out on both phosphate-doped (7P) and phosphate-free enamels (0P). Samples immersed for various times would be pulled out of solution to examine by optical microscope and corrosion resistance by electrochemical tests.

Table 3 Composition of Lawrence solution (LS).

Conc. (mol/L) KOH NaOH Ca(OH)2 CaCl2

pH at 25°C

LS [3] 0.061 0.022 0.0065 -- 12.95

2.4. Electrochemical testsIn the electrochemical tests, 3.5 wt.% NaCl

solution was used as the test electrolyte and a typical three-electrode set-up was used, including a 25.4 mm x 25.4 mm x 0.254 mm platinum sheet as the counter electrode, a saturated calomel electrode (SCE) as a reference electrode, and the testing sample as the working electrode. Three electrodes were connected to a Gamry, Reference 600 potentiostat/galvanostat for data acquisition. Open circuit potentials (Eocp) were stabilized for 2500 seconds immediately after the sample was immersed in the electrolyte. Electrochemical impedance spectroscopy (EIS) was conducted at five points per decade around Eocp with a sinusoidal potential wave of 10 mV in amplitude and frequency ranging from 0.01 Hz to 106 Hz. The sample was subsequently tested using potentiodynamic polarization method from 300 mV below Eocp to 1500 mV above Eocp with a scanning rate of 1 mV/s. Triplicate samples for each condition were tested and the representative results are shown below.

3. Results and Discussion

3.1. Immersion and Optical ObservationOptical microscope (OM) images of the

damaged samples before and after immersion in LS are shown in Fig. 3. P-free enamel (0P) shows signs of corrosion of exposed steel after immersing in LS for 3 days, whereas P-doped enamel (7P) forms white film of precipitates above the defect and no corrosion products of steel were observed beneath the layer.

3

3 cm 3 cm

Figure 3 OM images of 0P enamel coated on steel discs: as damaged (a), after immersion LS for 3 days (b); 7P enamel coatings as damaged (c), after immersion in LS for 3 days (d), where red dashed circle indicates the defect location.

3.2. Characterization of precipitatesEnergy dispersive spectroscopy (EDS) showed a

Ca-rich phase in the precipitates formed on enamel surface (shown in Tab. 4). Micro-Raman spectroscopy and X-ray diffraction (XRD) results (shown in Fig. 4) indicate that this phase is a Ca-P rich phase which is most likely the hydroxyapatite (HA).

Table 4 Elemental composition of enamel bulk (probe #1) and the precipitate layer (probe #2) from EDS result.

Atomic % O Na Si P Cl CaProbe #1 48 9 34 10 -- --Probe #2 26 -- 8 19 0.5 46

3.3. Potentiodynamic polarization test Potentiodynamic polarization test (PD) is a good

means to compare different coatings by accelerating the corrosion process. Representative PD data for each condition are shown in Fig. 5.

Corrosion current for both 0P and 7P remain very close with 7P slightly higher. However, a passive region was observed with 7P (600mV for sample immersed for 3d and 400mV for 2d

200 400 600 800 1000 1200 1400 1600 1800 2000

Raman shift (cm-1)

Hydroxyapatite-Fisher

0 10 20 30 40 50 60 70 80

o

oo

oo

o

o

o

o

o

2

o

o Hydroxyapatite

Figure 4 Micro-Raman spectroscopy (top) and XRD (bottom) on precipitaes of 7P enamel coatings immersed in LS for 3 days.

4

(a) 0P-as damaged (b) 0P-LS-3d

(d) 7P-LS-3d(d) 7P-as damaged

rust

10-10 10-9 10-8 10-7 10-6 10-5 10-4 10-3 10-2 10-1 100-1.5

-1.0

-0.5

0.0

0.5

1.0

0P-as defect 0P-def-LS1d 0P-def-LS2d 0P-def-LS3d

E vs

.SC

E (V

)

I (Amps/cm2)

10-10 10-9 10-8 10-7 10-6 10-5 10-4 10-3 10-2 10-1 100-1.5

-1.0

-0.5

0.0

0.5

1.0

E vs

.SC

E (V

)

I (Amps/cm2)

7P-as defect 7P-def-LS1d 7P-def-LS2d 7P-def-LS3d

Figure 5 Potentiodynamic polarization of enamel coated steel discs immersed in simulated pore water for different amount of time: (a) phosphorus-free enamel 0P; (b) 7 mol% P-doped enamel coating 7P.

Table 5. Capacitance values obtained from experimental data for undamaged enamels

0P enamel (F cm-2) 7P enamel (F cm-2)f = 999.95 Hz,r = 1.81×106 Ω

8.78 × 10-11 f = 999.95 Hz,r = 1.74×106 Ω

9.14 × 10-11

f = 100.00 Hz,r = 1.66×107 Ω

9.61 × 10-11 f = 100.00 Hz,r = 1.51×107 Ω

1.05 × 10-10

f = 9.9987 Hz,r = 1.41×108 Ω

1.13 × 10-10 f = 9.9987 Hz,r = 9.99×107 Ω

1.59 × 10-10

immersion) but not with 0P (active corrosion). Therefore, higher potential is required for 7P enamel coating to break the passive layer, where the potential plateau is right after the passive region. This potential increased with increasing immersion time in simulated pore water.

Therefore, the protection resulted from the passivation could be related with the reaction products between P-doped enamel and Lawrence solution.

3.4. Electrochemical impedance spectroscopy Electrochemical impedance spectroscopy (EIS)

has been widely used to understand the corrosion mechanisms especially in coating system [4, 5].

3.4.1. Characterization of intact enamel coatings

Firstly, undamaged vitreous enamel coatings are characterized by electrochemical measurement. Figure 6 shows the EIS Bode plots obtained on 0P and 7P enamel coatings, respectively. The responses from both coatings are characterized by a straight line

with a slope of -0.96, and -0.94 for 0P and 7P, respectively, through the frequency range between 105 to 10 Hz. The phase angle describes an angle near 90° at the corresponding frequencies, revealing a purely capacitive behavior of enamel coatings [5, 6].

Such responses confirm that both enamels behave as an insulating and protective coating on steel. The capacitance values associated with coatings are determined based on the equation:

C= 12 π f i r i

(1)

where fi and ri are coordinates of any point on Bode modulus line [7]. When the slope is -1, capacitance obtained is the same irrespective of which point it is calculated. Whereas, when the slope slightly differs from -1, the capacitance value obtained may vary with different frequencies. Table 5 shows capacitances determined from three different points on 0P and 7P Bode plot, respectively. In each case all of values are of the order of 10-10 Fcm-2, suggesting a very low capacitance that is typical of non-conductive coatings. 0P and 7P enamel coatings

5

Active corrosion

Passive region

exhibit the similar capacitive property as black enamel reported elsewhere [5]. It also reveals that in lower frequency (10-2 ~ 10 Hz) horizontal section

. Figure 6. EIS Bode plot for undamaged vitreous enamel coatings: (A) 0P, (B) 7P.

Figure 7 Experimental measured (msd) and model calculated (cal) Bode plots for enamels as defected and immersed in LS for 3 days: (A) 0P, (B) 7P.

was observed on 7P enamel which indicates resistive control of the response. Such response is associated with metallic substrate which may be exposed due to the pores and defects in the coating

However, after the coating was intentionally damaged and immersed in simulated pore water the impedance response differed significantly to that of the intact coating, as will be discussed in greater detail below.

3.4.2. Deterioration of damaged enamel coating under immersion condition

Figure 7 shows Bode curves for as defected coating and after immersion in LS for 3 days. For 0P enamels, the response differs drastically to that described above and almost remains constant through 3 days immersion (Fig. 7-A). Impedance modulus characterizes a slope of -0.69 followed by a horizontal section, which indicates a resistive behavior under charge transfer control. /Z/10 mHz

reaches a value of 105 Ω, five orders of magnitude lower than that of undamaged sample. Bode phase curve appears a wide maximum around 10 Hz.

With regards to 7P enamels, similar responses in Bode impedance curve except for 3d /Z/10 mHz reaches a value of 106 Ω versus 4.95×104 Ω for as defected 7P

6

enamel. In the phase curve, the phase angle maximum for 3d sample shifts to lower frequency than both as defected 7P and 0P samples. Bode phase curve for as defected 7P describes two maximum at 1 and 102 Hz, respectively. In the literature, the presence of the phase angle maximum in the frequency range (0.010-10 Hz) is associated with a corrosion process that occurs in the metallic substrate, while the maximum located in the range (102-104 Hz) is related with the response of the pores and defects in the coatings [8]. Therefore, the impedance response of as defected 7P samples may be comprised by both the exposed substrate due to the damage and pores and defects present within the coating.

In such cases, it is possible to determine the capacitance value associated with the system using the expression by Walter [7]:

C= 12 π f θmax

rθmax

(2)

where fmax is the frequency at which the bode phase angle reaches its maximum and rmax is the value of impedance modulus at that corresponding frequency. This expression fits the situation better than Eq. (1) since the slope greatly differs from -1. Thus, capacitance values obtained, shown in Table 6, are of the order of 10-6 F cm-2. This is a very high value to be considered for a protective coating. However, it falls in the range of values associated with an electrochemical double layer, indicating electrochemical double layer was obtained in both damaged and immersion cases.

Table 6 Capacitance values obtained for artificially damaged coating and defected coatings immersed in

LS for 3days.

fmax (Hz) rmax (Ω) 0P (F cm-2)as defect 7.8939 3000.3 6.72 × 10-6

LS-3d 9.9987 2424.4 6.57 × 10-6

fmax (Hz) rmax (Ω) 7P (F cm-2)as defect 1.9949 5947.7 1.34 × 10-5

LS-3d 1.0001 33563.5 4.74 × 10-6

3.4.3. Corrosion evaluation by equivalent circuit modeling

The impedance responses can be studied in greater details by means of equivalent circuit

simulation. The equivalent circuit used in this study is that broadly described in the literature for defective coating [5, 6, 9, 10]. Figure 8. shows the equivalent circuit model along with schematic representation of coating systems for different conditions. Intact coating behaves as a capacitor which isolates the substrate from the electrolyte only allowing corrosion occurs under diffusive control (Fig. 8-A, RQ). When defect was created in the coating either by artificial defect or degradation of coating, corrosion happens under charge transfer control with the R as the charge transfer resistance (Fig. 8-B, R(QR)). The capacitor is represented by a constant phase element Q, which is defined by the parameters Yo and n to take into account the nonhomogeneity of the coating surface. When n = 1, Q is a pure capacitor and the Nyquist curve describes a semicircle. However, most of the time the value of n is less than unity, reflected by a depression in the arc as a consequence of the lack of homogeneity caused by deterioration of the coating.

Figure 7 shows the experimental data for as defect coating and coating immersed in LS for 3 days, together with simulated results based on R(QR) model. Even though the model does not produce the best concordance between the experimental and simulated results for some samples (especially as defect-7P), the relative error associated with each element comprising the circuit are always below 5% (shown in Table 7). The two maximums in phase angle of 7P-as defect strongly indicates two time constants, which corresponds to the case in Fig. 8-C (R(Q(R(QR))). Applying this simulation to the experimental data yields excellent concordance. However, the error associated with some of the elements is very high (58%), which leads to distrust of their values and validity of the model.

7

Figure 8. Equivalent circuit model used for numerical simulation of EIS data. (A) intact coating; (B) coating

presenting defect and pathways for electrolyte uptake; (C) development of anodic activity at the steel/coating interface

[6].

Figure 9 EIS curves for enamels vs. immersion time in LS: (a) 0P enamel; (b) 7P enamel.

Table 7. Values obtained for each element parameters for 7P-as defect sample based on R(RQ) model.

Rs Yo n RValues 123.3 2.58 ×

10-50.684 5.75 ×

104

error % 2.42 3.15 0.94 4.00

Therefore, R(QR) model was used to simulate all experimental data of defected coatings and the parameters extracted from the model were used to quantitatively study the protective property of the coating system. Figure 9 shows the corrosion evolution of coatings that have been artificially damaged and immersed in LS for various time spans (as defect, 1d, 2d and 3d). The capacitance (Yo)

associated with the constant phase element and charge transfer resistance (R) were plotted against immersion time, as shown in Fig. 10. The overall capacitance values are at the order of 10 -5, which are too high to be considered as protective coatings due to the relative large area of exposed substrate by artificial defect. Although there is only one time constant (arc) appear drawn (Fig. 9), there may be more time constants involved. The time constant associated with the coating would be located at higher frequencies and would not be in the spectrum due to the limitations of the equipment [5, 8]. Therefore, in this study, the EIS responses correspond to the attack occurring in the metallic substrate at the artificial defect and the capacitance is associated with an electrical double layer at the

8

R(QR)

RQ

R(Q(R(QR)))

interface of the substrate and the electrolyte. Capacitance values for as defect 0P and 7P coatings stay very close (1.66 × 10-5 F cm-2 vs. 1.24 × 10-5 F cm-2), in agreement with the capacitance values calculated earlier (Table 6). As immersion time elapses, Yo slightly decreases for both coatings with 0P having values 2 times higher than 7P. Generally, coating systems with lower capacitance and higher resistance always shows slower corrosion rate and higher corrosion resistance.

Charge transfer resistances for as defect 0P and 7P coatings are identical (8.3 × 104 Ω cm2). During immersion, the resistance of 0P coating stays around 105 Ω cm2. However, resistance of 7P coating increases as a function of immersion time in LS.

After 3 days immersion, the resistance of 7P coating is two orders of magnitude higher than that of 0P, suggesting slower corrosion rate and higher corrosion resistance of defected 7P enamel coatings after immersion.

It reveals that base enamel coatings (0P) can protect the substrate metal when they are intact as a nonconductive barrier. It implies the protection mechanism persist until defects appear in the coating. Such defected coatings will allow the electrolyte and corrosive species reach the substrate and corrosion starts to develop, as can be seen in the significant drop of impedance at lower frequency and five orders of magnitude increase in capacitance. The corrosion

Figure 10 Corrosion evolution for defected 0P and 7P enamel coatings: (A) capacitance Yo associated with the constant phase element, (B) charge transfer resistance R, as a function of immersion time in LS

resistance of the coating does not change with further immersion in LS and the corrosion is mainly due to attacking the metallic substrate exposed at the defect. However, phosphate-doped enamel coatings (7P) exhibit enhanced corrosion resistance during immersion with presence of defect, indicating active corrosion protection. Active corrosion protection implies not only mechanical covering of the protected surface with a dense barrier coating (the case of intact coating), but also provides self-healing properties which allow continued protection even after partial damage of the coating [11]. The active corrosion protection shown in 7P coatings is most likely related to the Ca-P precipitates forming on the defect, which is not observed in 0P coating. The self-healing is the consequence of the release of the PO4

3- from the

enamel coating, promoted by alkalinity solution followed by fast precipitation of phosphate inhibitor with calcium ions present in the surrounding environment. The precipitates effectively impede the movement of electrolyte and ionic species, increasing the charge transfer resistance.

4. Conclusions

It can be concluded that the protective property of both 0P and 7P enamel coatings diminishes after the coating is damaged. The undamaged coatings behave as a barrier layer which isolates the metallic substrate from the aggressive medium. However, after the artificial defect was created the system’s responses changes drastically as a result of the

9

exposed substrate metal at the defect. Linear polarization of the damaged coating shows active corrosion of the exposed substrate, whereas the phosphate-doped coating exhibit wide passivation range on corrosion potential (600 mV) for samples after 3 day immersion in LS, which indicates after immersion 7P enamel coating passivates the defect and slows the corrosion rate.

Unlike the polarization technique, by using EIS it is possible not only to evaluate the surface state of the coating, in situ, without the need to induce the polarization of the systems and damage the material applying direct current, but also to establish hypothesis about the mechanism responsible for the corrosion process and carry out simulations which allow to evaluate the validity of the hypothesis. In this study, the response of defected coating is attributed to the attacking in the metallic substrate, which is described by a phase angle maximum at lower frequency. Consequently the corrosion process changes from diffusive control to resistive control, which is characterized by a horizontal section in the Bode impedance spectrum. The corrosion of defected coatings was evaluated quantitatively by capacitance and charge transfer resistance extracted from model simulation. It reveals that the corrosion resistance of 0P enamel is independent on immersion time and it almost stays identical after immersion. However, the corrosion resistance of 7P enamel increased with immersion time and reaches a value of two orders of magnitude higher than as that of defect 7P and 0P enamels. The increased corrosion resistance of defected 7P after immersion suggests active corrosion protection which is a continued protection of the substrate even after partial damage of the coating.

Either the passivation or active corrosion protection could be attributed to the self-healing property of the phosphate-doped coating. Phosphate released from the glass fast precipitates with the calcium ion present in LS, which is enhanced by the high alkalinity nature of the cement environment. The precipitates form a layer on top of the defect as a barrier to the electrolyte medium and impeding the movement of the corrosive species. Whereas the base enamel creates direct pathways for electrolyte to reach the metallic substrate after the coating is damaged. Therefore, phosphate-doped enamel

coating outperforms the base enamel by providing active corrosion protection and passivates the defect.

5. ACKNOWLEDGMENTS

The authors gratefully acknowledge the financial support provided by Intelligent Systems Center and the U.S. National Science Foundation under Award No. CMMI-0900159.

6. REFERENCES

1. Tang, F., et al., Corrosion resistance and mechanism of steel rebar coated with three types of enamel. Corrosion Science, 2012. 59(0): p. 157-168.

2. K.M. Fyles, P.S., Alkali resistant glass fibers for cement reinforcement, P.B.L. (GB), Editor 1982.

3. Kriker, A., et al., Durability of date palm fibres and their use as reinforcement in hot dry climates. Cement and Concrete Composites, 2008. 30(7): p. 639-648.

4. Zheludkevich, M.L., et al., On the application of electrochemical impedance spectroscopy to study the self-healing properties of protective coatings. Electrochemistry Communications, 2007. 9(10): p. 2622-2628.

5. Conde, A. and J.J. De Damborenea, Electrochemical impedance spectroscopy for studying the degradation of enamel coatings. Corrosion Science, 2002. 44(7): p. 1555-1567.

6. Trabelsi, W., et al., The use of pre-treatments based on doped silane solutions for improved corrosion resistance of galvanised steel substrates. Surface and Coatings Technology, 2006. 200(14-15): p. 4240-4250.

7. Walter, G.W., A review of impedance plot methods used for corrosion performance analysis of painted metals. Corrosion Science, 1986. 26(9): p. 681-703.

8. Kouloumbi, N., et al., Evaluation of the behaviour of particulate polymeric coatings in a corrosive environment. Influence of the concentration of metal particles. Progress in Organic Coatings, 1996. 28(2): p. 117-124.

10

9. Feliu, S., J.C. Galván, and M. Morcillo, The charge transfer reaction in Nyquist diagrams of painted steel. Corrosion Science, 1990. 30(10): p. 989-998.

10. Trabelsi, W., et al., Electrochemical assessment of the self-healing properties of Ce-doped silane solutions for the pre-treatment of galvanised steel substrates. Progress in Organic Coatings, 2005. 54(4): p. 276-284.

11. Lamaka, S.V., et al., Nanoporous titania interlayer as reservoir of corrosion inhibitors for coatings with self-healing ability. Progress in Organic Coatings, 2007. 58(2–3): p. 127-135.

11