Cathodic delamination: Quantification of ionic transport rates along coating–steel interfaces

Progress in Organic Coatings 68 (2010) 70–78

Contents lists available at ScienceDirect

Progress in Organic Coatings journal homepage: www.elsevier.com/locate/porgcoat

Cathodic delamination: Quantification of ionic transport rates along coating–steel interfaces夽 P.A. Sørensen a , K. Dam-Johansen a , C.E. Weinell b , S. Kiil a,∗ a b

Department of Chemical and Biochemical Engineering, Technical University of Denmark, Building 229, DK-2800 Kgs., Lyngby, Denmark Hempel A/S, Lundtoftevej 150 DK-2800 Kgs., Lyngby, Denmark

a r t i c l e

i n f o

Article history: Received 2 June 2009 Received in revised form 4 August 2009 Accepted 20 August 2009 Keywords: Corrosion Disbonding Disbondment Accelerated testing Protective coatings

a b s t r a c t So-called cathodic delamination is one of the major modes of failure for organic coatings immersed in electrolyte solutions (e.g. seawater). Cathodic delamination occurs as a result of the electrochemical reactions, which takes place on a corroding steel surface. This means that reactants must continuously be transported from the bulk solution to the cathodic areas. The transport of sodium ions from a defect in the coating to the cathodic areas is generally considered the rate-determining step for cathodic delamination because the transport of oxygen and water through the coating is sufficient for the corrosion processes. In this work, a novel practical method, which allows direct estimation of the apparent diffusion coefficient of sodium ions in the ultrathin aqueous layer at the coating–steel interface, is described. The apparent diffusion coefficients estimated are of similar magnitude as previously reported values and show an acceptable repeatability. The method was used to obtain the apparent diffusion coefficients of sodium ions in the coating–steel interface for three commercial inert-pigmented epoxy coatings. The delamination rates predicted using the apparent interfacial diffusion coefficients and Fick’s second law, under the assumption of a transport-controlled mechanism, show qualitative agreement with the observed delamination rates in 0.5 M sodium chloride. This confirms that the rate-determining step of cathodic delamination is the transport of sodium ions along the coating–steel interface. © 2009 Elsevier B.V. All rights reserved.

1. Introduction Organic barrier coatings are extensively used to protect immersed steel structures, such as ballast tanks, ship hulls, wind turbines and bridges, from corrosion [1]. When a part of the coating system is damaged or mechanically removed, electrochemical reactions will occur at the defect site because the bare steel is in direct contact with the surrounding environment. At the defect, a rapid dissolution of the solid steel is possible, whereas the dissolution of solid steel is strongly inhibited below the adherent coating. Seawater and other reactants within the defect will slowly diffuse into the intact coating–steel interface, where the ions will serve as charge carriers and thereby enable oxygen to be reduced below the coating on the oxidized steel surface [2]. The reduction of oxygen on steel surfaces is based on a complex mechanism consisting of many steps in which electrons are transferred from the oxide

DOI of original article:10.1016/j.porgcoat.2009.10.011. 夽 The Publisher regrets that this article was previously published in Progress in Organic Coatings, 67 (2010) 107–115, for citation purposes please use the original publication details. ∗ Corresponding author. Tel.: +45 45252827; fax: +45 45882258. E-mail address: [email protected] (S. Kiil). 0300-9440/$ – see front matter © 2009 Elsevier B.V. All rights reserved. doi:10.1016/j.porgcoat.2009.08.018

surface to the oxygen species. Several reactive intermediate products, such as superoxide and hydroxy radicals may be formed in this process [3–5]. However, the main product is hydroxyl ions. In order to preserve the local charge neutrality, cations must therefore be transported from the defect to the cathodic areas to neutralize the charge of the hydroxyl ions generated by the cathodic reaction [6]. The presence of positively charged alkali metal ions, such as sodium and potassium ions, and negatively charged hydroxyl ions results in a highly alkaline environment within the thin aqueous layer at the coating–steel interface and pH values up to 14 have been reported [7–9]. The reactive intermediate products and the alkaline environment will adversely affect the adhesion between the anticorrosive coating and the steel, resulting in partly or complete delamination of the coating as shown in Fig. 1. The mechanism of cathodic delamination of organic coatings on steel surfaces has been studied extensively for several decades [6,10–16]. Experimental and theoretic studies have shown that the amount of water and oxygen consumed by the cathodic reaction underneath a delaminated coating is less than the amount of water and oxygen that diffuses through a typical anticorrosive coating [2,17,18]. This means that the transport of oxygen and water cannot control the rate of cathodic delamination. The rate-determining step of cathodic delamination of inert-pigmented epoxy coatings appears to be the transport of positively charged

P.A. Sørensen et al. / Progress in Organic Coatings 68 (2010) 70–78

71

Fig. 1. Idealized sketch of cathodic delamination. Solid iron is transformed into ferrous iron at the defect and oxygen is reduced to hydroxyl ions at the cathodic sites underneath the delaminated coating. Sodium ions must migrate along the coating–steel interface to neutralize the charge of the hydroxyl ions. The presence of a highly alkaline environment and reactive intermediate products weakens the adhesion between the coating and the substrate, resulting in delamination of the coating.

hydrated ions to the cathodic areas. Several possible routes for the cations have been suggested [6,14,15]. However, cathodic delamination of well-formulated epoxy coatings appear to be controlled by the interfacial transport of hydrated cations in a thin aqueous layer at the coating–steel interface from the defect to the delamination front in most naturally encountered environments [18–21]. Several studies have considered the effects of the type and concentration of the ions in the electrolyte solution on the delamination rate. The type of anion appears to have little influence on the rate of delamination [10,11,13] whereas the delamination rate is nearly proportional with the mobility of the alkali metal cations. Furthermore, the delamination rate is proportional with the concentration of the cations [11,13]. The delamination of coated metal in solutions of divalent cations has been reported to be very slow or not occurring at all, which has been attributed to the low solubility of the various hydroxides [10,22,23]. The typical apparent reaction order of 0.5 [24] for cathodic delamination (corresponding to a process obeying the laws of Fick) supports the hypothesis of a process controlled by the interfacial transport of hydrated cations. Despite the fact that the transport of sodium ions along the coating–steel interface appears to be the rate-determining step for cathodic delamination, attempts to quantify the interfacial rates of transport are scarce [25–28]. The interfacial diffusion coefficients of various cations have recently been estimated by application of the Scanning Kelvin Probe technique [27]. Other authors [25] apply the product of the diffusion coefficient and the coating–steel interface height as a measure for the interfacial transport rate. The application of such a setup is limited by the fact that the height of the coating–steel interface is not readily estimated or in any way well defined. Consequently, a reliable estimate of the interfacial diffusion coefficient is not readily achieved. Other reported values for diffusion coefficients of the hydrated cations along the coating–steel interface have been estimated using mathematical models of cathodic delamination by fitting the predicted delamination rate to the observed delamination rate by varying the diffusion coefficient [25,29,30]. However, there may be significant uncertainties in the diffusion coefficients estimated from mathematical models of cathodic delamination because the detailed mechanism for cathodic delamination is not completely understood [31]. As a consequence of this, new methods capable of quantifying the transport of cations along the coating–steel interface may provide important knowledge about the detailed mechanism of cathodic delamination and assist in the development of novel anticorrosive coatings. Quantification of the interfacial transport may also be a useful input to the present mathematical models of cathodic delamination [32,33], which are restricted by the lack of information on ionic interfacial diffusion. Hence, a simple method capable of estimating the interfacial diffusion coefficients of the cations along the coating–steel interface is needed. This article addresses the development of a novel practical method, which can be applied for estimation of the diffusion coefficients of hydrated sodium ions along a coating–steel interface.

2. Theory The so-called time-lag method has proven to be an effective method for obtaining the diffusion and permeability coefficients of ideal systems [34]. Since its origins in 1920, interest in the time-lag method has expanded because of its value in characterising simple permeation processes and also complex systems of diffusion with simultaneous adsorption and surface diffusion. The method has historically been applied to diffusion and adsorption in porous membranes and diffusion in polymer membranes and has recently been applied for estimation of the ionic permeability of intact coatings [35]. The simple nature of the time-lag method enables transport parameters to be directly obtained from experimental data when the diffusion coefficient of the species considered does not change with time. Hence, avoiding the intensive mathematical treatment required by other techniques. The basis of time-lag methods is a concentration step perturbation, which is performed in one chamber of a two-chamber permeation cell, which is divided by a semi-permeable membrane. The time dependent diffusion response may be followed by continuous monitoring of the concentration in the non-pertubated chamber. This procedure is known as the time-lag method, because there is a time-lag between the perturbation and the onset of the change in the concentration on the other side of the semi-permeable membrane. During a time-lag experiment, the concentration in the pertubated chamber remains almost constant and increases slightly in the non-pertubated chamber. This means the concentration gradient is approximately constant. After some time, the mass transfer between the two chambers stabilizes and steady state conditions are achieved. A typical time-lag response for a semi-permeable membrane is shown in Fig. 2. The permeability coefficient can be obtained from the steady state part of the experiment, whereas the time-lag is related to the diffusion of the species considered across the semi-permeable membrane. Reliable quantification of the interfacial transport of sodium ions by application of the time-lag method, requires that a given model interface separating the two solutions bears a high degree of resemblance to a real interface between an organic coating and its substrate. Furthermore, the experimental time needed for estimation of the interfacial diffusion coefficient should be minimized. Here, it is proposed that the coating–steel model interface can be established by insertion of nails consisting of thoroughly cleaned cold rolled low-alloy carbon steel into the liquid coating and that the experimental time can be controlled by adjusting the thickness of the wet coating. The principle of preparing a coating–steel interface by insertion of nails into a liquid coating is illustrated in Fig. 3. 3. Mathematical model The mathematical model is analyzed under the simplifying assumption that only diffusion determines the transport of cations

72

P.A. Sørensen et al. / Progress in Organic Coatings 68 (2010) 70–78

Fig. 4. Cross-section view of simplified coating sample without nails separating the sodium chloride solution and the distilled water.  = 0 is the boundary between the concentration perturbed chamber and the coating sample and  = 0 is the boundary between the coating sample and the ion free solution.

in these chambers. The concentration of sodium ions in the pertubated chamber is ci,u , while no sodium ions are present in the non-pertubated chamber. The initial condition is given by assuming that the coating sample is free of penetrant at the start of the experiment. Fig. 2. Idealized time-lag response for a concentration perturbation experiment. The dotted line illustrates the time-lag,  (time required to obtain a constant rate of diffusion) resulting from diffusion through a semi-permeable membrane.

along the coating–steel interface. This is a simplifying assumption because anions are excluded from the all transport processes during cathodic delamination, suggesting that the electrical field crated by the corrosion process controls the ionic transport along the coating–steel interface. The effect from the corrosion process and diffusion may be lumped into an apparent diffusion coefficient. This means that the formal diffusion coefficient derived in mathematical analysis may be considered as an apparent diffusion coefficient, which has been modified by the ionic mobility and an electrical field gradient [11]. Thus, Fick’s second law can be used to describe the transient one-dimensional diffusion of ions along in the ultrathin aqueous layer at the coating–steel interface. ∂ci ∂ =− ∂t ∂



−Di

∂ci ∂



(1)

where ci is the concentration of component i, t the diffusion time,  the distance from the reference point and Di the diffusion coefficient of the species considered. The boundary conditions may be identified by considering the simplified schematic view of a coating specimen separating the two solutions is shown in Fig. 4. Furthermore, the concentration gradient across the coating sample can be assumed to be constant because the concentrations in the pertubated and non-pertubated chambers only changes slightly with time. This means that the boundary conditions are given by concentrations of sodium ions

c(, 0) = 0

(2)

c(0, t) = ci,u

(3)

c(, t) = 0

(4)

The rate at which the ions emerges from the coating sample to the non-pertubated chamber is given by D(dc/d)|=0 . The total amount of ions transported to the non-pertubated chamber can be obtained by integration of equation describing the ions emerges from the coating sample to the non-pertubated chamber with respect to the time, t. The full analytical solution to Fick’s second law in this specific case, is given by [36] 2  (−1)2 Dt 1 = i2 − − 2 exp 6   n2 ∞

Qi,t ci,u

n=1



−Di n2 2 t 2

 (5)

where Qi,t is the amount of ions transported across the sample,  the thickness of the membrane, ci,u the initial concentration on the feed side and n is an integer. Low order approximations of the analytical solution to Fick’s law will converge rapidly towards the full analytical solution for the conditions relevant for this study (e.g. n = 3). The exponential term can be neglected when t → ∞. This means that the analytical solution to Fick’s second law may be approximated by the straight line given in Eq. (6) Qi,t =

Di ci,u 



t−

2 6Di



(6)

By isolating t in Eq. (3), it can be seen that extrapolation of the linear part of a plot of Qi,t /ci,u versus time to intercept with the abscissa yields a time-lag, which is related to the diffusion coefficient by Eq. (7) =

2 6Di

(7)

4. Experimental

Fig. 3. Principle sketch of coating–steel interface prepared by insertion of nails into the liquid coating.

The design of the two-chamber diffusion cell, illustrated in Fig. 5, used for estimation of the diffusion coefficient of hydrated sodium ions in the coating–steel interface is similar to the design previously used by Carneiro et al. [35] for measuring the permeability of intact coatings towards chloride ions. The upper chamber of the diffusion cell is designed to contain an aqueous solution of sodium chloride while the lower chamber contains distilled water. The upper chamber has a volume of 210 cm3 . This is nearly three times greater than the volume of the

P.A. Sørensen et al. / Progress in Organic Coatings 68 (2010) 70–78

73

Table 1 Coating composition (wt%) of liquid paint. Compound

Amide cured epoxy resin Plasticizer Pigments Extenders Solvents Additives

Fig. 5. Cross-section scheme of diffusion cell and coating sample with nails.

lower chamber, which has a volume of 75 cm3 . Thus, it is reasonable to assume that the ionic concentration in the upper chamber remains constant throughout the duration of the experiment. The two chambers are separated by a coating sample supported by 77.0 ± 3.7 ␮m thick Kraft paper with the coating surface downwards. Kraft paper is manufactured from wood pulp by a modified sulphate pulping process. Kraft paper is cheep, well-known for its high strength and does not provide any significant resistance to mass transfer (see Section 6) and has previously been used to support coatings in permeation measurements [35]. In order to prevent development of osmotic pressure gradients, both the upper and lower solutions are in contact with the atmosphere through small holes in the upper part of the permeation cell. During the experiment, both sodium and chloride ions will diffuse from the upper chamber to the lower chamber while water will diffuse from the lower chamber to the upper chamber. Therefore, a small reservoir with distilled water is connected to the lower chamber to maintain a constant volume in the lower chamber. The upper chamber must have a free volume, which can accommodate the amount of water transported across the coating sample to allow transport of water from the lower to the upper chamber. A magnetic bar is placed in the lower chamber to homogenize the lower solution. The concentration of sodium ions in the lower chamber is continuously monitored by a sodium-selective polymer membrane electrode from Metrohm with a measuring range of 5 × 10−6 M to 1 M. A double-junction Ag/AgCl reference electrode from Metrohm was inserted in the cell and placed at a fixed position. 5. Experimental procedure and preparation of coating samples The coating samples were produced by application of a liquid epoxy coating on a sheet of Kraft paper by an automatic coating applicator – Coatmaster 509 MC-III. Subsequently, the Kraft paper with the liquid coating was transferred to an easily penetrable support consisting of polystyrene. Cold rolled low-alloy carbon steel nails with a diameter of 1.2 mm were manually pushed into the coating and the Kraft paper and to some extent the polystyrene. Prior to insertion into the liquid coating, the nails were thoroughly cleaned with a water based alkaline detergent – Hempel Light Clean, rinsed with ultra pure water, abrasive grinded with 3 M abrasive paper P360, rinsed with ultra pure water and immersed in ethanol to reduce the amount of residual water on the nails. This procedure removes contaminants and prevents significant corrosion of the cleaned nails, which must be stored carefully prior to usage.

Coatings I

II

III

30.7 7.9 7.9 33.5 18.6 1.5

30.9 8.0 6.9 33.8 18.9 1.6

26.3 6.9 5.1 37.7 21.3 2.7

The roughness of the surface of the nails was measured by a Handysurf E-35A from Zeiss. The measured arithmetic mean of maximum peak-to-valley height of five adjoining single sampling lengths, Rz , was 0.52 ± 0.20 ␮m. This indicates that the roughness of the nails will not influence the magnitude of the estimated diffusion coefficients significantly [19]. Hence, the diffusion coefficients estimated will not have to be corrected by a tortuosity factor. The coating samples were allowed to cure for 7 days before they were gently removed from the polystyrene support. Using a pair of sharp scissors, circular test specimens with a diameter of 9 cm were cut from the samples. The dry film thickness of the coating specimens were measured using a magnetic gauge instrument – Elcometer 355 Top, which was calibrated prior to initialization of measurements. The coating specimens were allowed to cure for another 3 weeks at ambient temperature before they were used in the permeation cell. The coating sample supported by the Kraft paper was placed in the diffusion cell, which must be tightly closed. A magnetic bar was placed in the lower chamber before distilled water was fed to the lower chamber using a syringe. The amount of water used for filling the lower chamber was estimated by weighing the syringe prior and after injection of water. The sodium-selective polymer membrane electrode was then inserted and the permeation cell was placed horizontally so that the upper chamber could be filled with a known volume of 2.5 M sodium chloride solution. The reference electrode and the sodium-selective electrode was connected to a 781 pH/Ion Meter from Metrohm with a measuring accuracy of ±0.4%. The unit was controlled by a computer, which was connected to the pH/ion Meter by a RS 232 interface. Verification support for validation software from Metrohm was used to control the entire unit and for data acquisition. The entire setup was placed in a thermostatically controlled room with a temperature between 20 and 23 ◦ C. The time-lag method was used for estimation of the interfacial diffusion coefficient of sodium ions for three commercial coatings. The applied coatings were all inert-pigmented, based on diglycidyl ether of bisphenol A and cured by an amide-adduct. Information about composition of coatings is given in Table 1. Microscopic cross-cut investigation of the coating samples showed that the penetration of the applied coatings into the Kraft paper was negligible. Estimations of the interfacial diffusion coefficients were performed three times with different coating specimens for evaluation of the precision. The potential difference between the reference electrode and the sodium-selective polymer electrode in the lower chamber was continuously recorded and converted into concentration data by a calibration curve. An example of the concentration history of sodium ions in the lower chamber of the permeation cell is shown in Fig. 6. 6. Results and discussion The penetration of electrolytes through epoxy coatings is a very slow process compared to the diffusion of electrolytes along the coating–steel interface. The reported diffusion coefficients for

74

P.A. Sørensen et al. / Progress in Organic Coatings 68 (2010) 70–78

Fig. 6. Example of sodium ion concentration history in lower chamber for commercial magnesium silicate and titanium dioxide pigmented epoxy-amide coating.

sodium ions in the coating–steel interface is several orders of magnitude greater than the diffusion coefficient of sodium ions in bulk epoxy coatings (see Table 2). This suggests that ionic transport through the bulk epoxy coating will not interfere with the measurements of the interfacial diffusion coefficients and therefore may be neglected. It is also clear, that there is a significant scatter in the values of the reported diffusion coefficients for sodium ions at the coating–steel interfaces. This may be partly explained by the fact that the applied coating systems have different properties and that some of diffusion coefficients reported has been indirectly estimated from delamination rates. Furthermore, neither of the references addresses important parameters such as the temperature and the tortuosity of the coating–steel interface, which are known to influence to magnitude of the observed diffusion coefficients. 7. Verification of fundamental principles There are several issues, which must be addressed before, it has been verified that a change in the concentration of sodium ions in the lower chamber is due to transport along the coating–steel interface. Most commercial coatings contain extenders or pigments with impurities of sodium ions, which may leach during testing. The leaching of sodium ions from the different types of coating specimens supported by Kraft paper was investigated by immersion of the coating specimens in ultra pure water in sealed containers for 3 months. Analysis of the concentration of sodium ions by a sodiumselective electrode at the start and termination of the experiment showed that the concentration of sodium ions in the ultra pure water remained very low throughout the measuring period and could be neglected. Hence, the leaching of sodium ions from the coating specimens can be neglected.

Diffusion measurements on coating specimens without inserted nails with dry film thickness identical with the coating specimens containing nails showed that no sodium ions penetrated the bulk coating during the duration of the experiments. This means that the penetration of sodium ions through the bulk epoxy coatings in the regions without nails can be neglected and that the sodium ions in the non-pertubated chamber the must have been transported along the coating–steel interface. Measurements performed with nails of various lengths showed that the apparent diffusion coefficients estimated were not affected by the area of the nails, which were in contact with the 2.5 M solution of sodium chloride. This shows that the transport of ions along the coating–steel interface was not altered by a lager corroding area. Thus, the small differences in the corroding areas amongst the samples will not interfere with the estimation of the apparent diffusion coefficients. Verification of the negligible resistance towards mass transfer provided by the Kraft paper was found by measuring the diffusion coefficient of sodium ions through intact Kraft paper by the timelag method. The estimated value of 2.1 × 10−11 ± 1.3 × 10−11 m2 s−1 is significantly greater than the reported values for the diffusion coefficient of sodium ions along the coating–steel interface. Furthermore, the value is similar magnitude as the diffusion coefficient of sodium ions at both infinite dilution (1.334 × 10−11 m2 s−1 [41]) and concentrations up to 2.25 M (1.11 × 10−11 m2 s−1 [42]). This confirms that the resistance provided by the Kraft paper can be neglected. The rate of which the sodium ions emerge from the coating specimens must be directly proportional to the area (e.g. perimeter of the nails) available for diffusion according to Fick’s first law. Scaling of the interfacial transport rates by changing the area available for diffusion may therefore be used to see whether the diffusion of sodium ions from the upper chamber to the lower chamber obey the laws of Fick. This is because the interactions leading to adhesion between the coatings and the nails should not depend on the total perimeter provided that the nails are placed sufficiently far apart. Here, it is implied that the concentration gradient across the coating sample is constant (e.g. thickness of the coating sample is constant). The total perimeter of the nails in the coating samples was varied by changing the number of nails in the coating specimens. Fig. 7 shows that the state rates of transport of sodium ions from the upper chamber to the lower chamber are directly proportional to the total perimeter of the nails inserted in the coating. This means that the laws of Fick are obeyed and that the total perimeter of the nails may be scaled without affecting the apparent rate at which the hydrated sodium ions are transported along the coating–steel interface. 8. Estimation of interfacial length In addition to the total perimeter of the nails inserted into the coating, the rate of transport across the coating specimen is proportional to the concentration gradient across the coating specimen

Table 2 Representative interfacial and bulk epoxy diffusion coefficients of Na+ and Cl− for various pigmented and non-pigmented epoxy coatings. Reference

Glass and Smith [37] Glass and Smith [37] Glass and Smith [38] Hu et al. [39,40] Wapner et al. [27] Hernandez et al. [30] Leng et al. [11] a

Coating system

Epoxy-polyamide Pigmented epoxy-polyamide Pigmented epoxy Pigmented epoxy-polyamide Heat-curing two-component epoxy Amine-modified epoxy Polyamidoamines-epoxy-ester

Interfacial diffusion coefficients [m2 s−1 ]

Bulk polymer diffusion coefficients [m2 s−1 ] Na+

Cl−

Na+

3.2 × 10−15 9.2 × 10−17 8.7 × 10−17 – – – –

4.7 × 10−16 6.3 × 10−17 5.8 × 10−17 4.6 × 10−16 – – –

– – – – 1.5 × 10−11 2.25 × 10−14 a 6.4 × 10−11 a

The values have been estimated by fitting a simulated delamination rate to a measured delamination rate using mathematical models.

P.A. Sørensen et al. / Progress in Organic Coatings 68 (2010) 70–78

75

Fig. 7. Rate of transport of sodium ions across the coating sample versus the total perimeter length of the nails inserted into the coating. The thickness of the coating is identical for all coating samples and the vertical lines show the experimental uncertainty on the measured rate of transport of sodium ions along the coating–steel interface.

(e.g. the thickness of the coating in proximity of the nails). Liquid anticorrosive coatings have low surface tensions and viscosities because they are designed for maximal wetting of the substrate. This means that the coating will creep up along edges and that the length of the coating–steel interface is greater than the actual thickness of the dry coating as shown in Fig. 8. Measurements on several different coatings showed the distance a coating will creep up along a nail and thereby exact length of the coating–steel interface is a characteristic for the specific type of coating and therefore has to be evaluated for each coating sample. Optically microscopic investigations showed that the thickness of the coating which has crept up along the nails may be of similar order of magnitude as the interfacial distance along the coating–steel interface. These observations confirm that it is reasonable to assume that the diffusion of sodium ions through the bulk coating to the coating–steel interface does not affect the measurements. This is because the time required for diffusion of sodium ions through the bulk coating is significantly larger than the time of diffusion of sodium ions along the coating–steel interface. The length of the coating–steel interface was estimated by microscopically investigations of the uncorred region of the nails that had been placed in the coating sample because the uncorred distance was identical to the distance of the nails covered with coat-

Fig. 9. Microscopic analysis of nails after exposure. The vertical lines show the length of the coating–steel interface.

ing prior to exposure. As illustrated in Fig. 9, the border between the corroded and un-corroded steel is distinct. 9. Apparent diffusivity of sodium ions and cathodic delamination The apparent diffusion coefficients of sodium ions at the coating–steel interface estimated, which are presented in Table 3 demonstrate that the method has an acceptable repeatability. Furthermore, apparent interfacial diffusion coefficients of sodium ions are of similar magnitude as the values tabulated in Table 2. The apparent interfacial diffusion coefficients estimated are about 3 orders of magnitude smaller than the diffusion coefficient of sodium ions in a bulk electrolyte at infinite dilution [41]. On the other hand, the apparent interfacial diffusion coefficients are much lager than the diffusion coefficients of ions in bulk epoxy coatings. This shows that the estimated apparent diffusion coefficients are within a reliable range for the diffusion of hydrated ions in a case where the motion of ions is limited to the free volumes at the coating–steel interface. Thus, the coating–steel interface is a region where hydrated cations can be transported fast compared to transport in the bulk coating. In the absence of an oxygen rich atmosphere or a galvanic cell, both sodium and chloride ions diffuse into the coating–steel interface. However, once a galvanic cell is established only positively charged ions are transported along the coating–steel interface, where they will serve as charge carriers [11,43]. This is because hydrated cations are attracted by the excess of negative charge in the coating–steel interface. Increased local interfacial concentrations of hydroxyl ions resulting from the reduction of oxygen will thus accelerate the cations transport due to the presence of electrostatic field. The selective ion transport on corroding steel surfaces can be assigned to the electrostatic field resulting from the corrosion process (e.g. potential gradient between the anode and cathode). Experimental results suggest that, the diffusion of cations Table 3 Apparent interfacial diffusion coefficients of sodium ions at the coating–steel interface for three commercial epoxy coatings. Coating

Fig. 8. Microscopic analysis of nail with attached coating. The length of the coating–steel interface is significantly greater than the thickness of the coating because the liquid coating creeps up along edges.

I II III

Interfacial diffusion coefficient [m2 s−1 ] 1.4 × 10−13 ± 0.1 × 10−13 2.2 × 10−13 ± 0.2 × 10−13 3.0 × 10−13 ± 0.2 × 10−13

76

P.A. Sørensen et al. / Progress in Organic Coatings 68 (2010) 70–78

Fig. 12. Chemical structure of N-[3-(trimethoxysilyl)propyl]ethylenediamine.

Fig. 10. R relationship between the extent cathodic delamination rate of inert-pigmented coatings and apparent interfacial diffusion coefficient of the coating–steel interface towards sodium ions. The vertical lines illustrate the experimental uncertainty related to the delamination rate while the horizontal lines show the experimental uncertainty on the estimation of the interfacial diffusion coefficient.

appears may be insignificant in comparison with contribution from migration (transport of ions under the influence of a potential field) [44]. The transport of sodium ions is principally described by the Nernst–Planck equation, which takes the effects of both diffusion and migration into account. However, the Nernst–Planck equation and diffusion coefficients estimated by the time-lag method cannot be used in combination for a quantitatively description of the transport of sodium ions along coating–steel interfaces. This is because effects from migration and diffusion have not been measured individually but have been lumped into an apparent diffusion coefficient. Regardless of the transport mechanism, the apparent interfacial diffusion coefficients estimated may be used as a measure for the delamination rate when the rate-determining step in the process of cathodic delamination is the transport of cations along the coating–steel interface Ji = −Di,e

 dc

i

d

+

RT d c zi F i d

 (8)

where Ji is the flux, Di,e the effective interfacial diffusion coefficient at infinite dilution, R the universal gas constant, T the absolute temperature, F Faradays constant, zi the charge of the considered specie and  the electrical potential created by the ionic species. The true interfacial diffusion coefficient of sodium ions along the coating–steel interface can be estimated when the steel is not polarized (e.g. no corrosion occurs). This type of measurements could be performed with the described experimental setup by purging the upper and lower solutions with an inert gas (e.g. nitrogen) and placing the entire diffusion cell in an inert atmosphere. The subsequent treatment of the obtained data is already described in Section 2. The importance of interfacial transport of sodium ions for cathodic delamination was confirmed by comparison of the apparent interfacial diffusion coefficients with measured of the

delamination rates of the applied coatings. The extent of cathodic delamination of the applied coatings on cold rolled low-alloy carbon steel was removing the delaminated coating after exposure to a 0.5 M aerated solution of sodium chloride. Fig. 9 shows a direct proportionality between the estimated apparent interfacial diffusion coefficients of sodium ions in the coating–steel interface and the observed delamination rate. This demonstrates that the delamination rates predicted from Fick’s second law using the interfacial diffusion coefficients, under the assumption of a transport-controlled mechanism, show qualitative agreement with the observed delamination rates in 0.5 M sodium chloride (e.g. similar conditions as seawater). This confirms that the cathodic delamination of inert-pigmented coatings, under the conditions studied, is controlled by the interfacial transport of sodium ions from the defect to the delamination front (Fig. 10). 10. Influence of adhesion promoters on diffusivity and cathodic delamination The adhesion of organic coatings to steel is weak due to the nature of the steel surface, which mainly consisted of oxidized steel. Epoxy coatings interact with the oxidized steel surface through acid–base interactions, typically hydrogen bonds. Superior adhesion can be achieved by the formation of covalent bonds between the coating and the steel surface. One way of improving the adhesion between the organic coatings and the steel is to stabilize their connection by addition of coupling agents (e.g. adhesion promoters) [45,46]. Adhesion promoters are often reported to improve adhesion and reduce interfacial de-adhesion failures between the coating and the steel during exposure to corrosive environments [47–50]. Silanes react with water in aqueous solutions to form hydrolyzed silanes, which will react with the surface of the substrate. The individual bonds will subsequently polymerize and build up layers outward from the substrate with the organic functionality oriented towards the coating [51]. This process is shown in Fig. 11. Amongst the group of adhesion promoters amino and epoxy functional silanes have found the most widespread usage for epoxy coatings because of their high degree of compatibility with the binder. The effect of an amino functional silane on the rate of cathodic delamination and interfacial transport of sodium ions was investigated by addition of 2 wt% N-[3(trimethoxysilyl)propyl]ethylenediamine, see Fig. 12, to coating III. The addition of the amino functional silane to the coating reduced the rate of delamination significantly. However, the addition of amino functional silane had insignificant influence on the apparent interfacial diffusion coefficient estimated. Therefore, the

Fig. 11. Hydrolysis of organic functional silane and subsequent polymerization of the silane on a metal substrate.

P.A. Sørensen et al. / Progress in Organic Coatings 68 (2010) 70–78

interfacial transport of hydrated sodium ions does not seem to be the rate-determining step of the cathodic delamination in the silane modified coating system. The improved durability of the silane coating compared to the unmodified coating system is probably a result of the improved stability of the coating–steel interface due to the dense silane modified coating–steel interface. It has been suggested that the stable coating–steel interface will lead to a more diffuse double layer structure when hydrated sodium ions are present at the interface [27,52]. The diffuse double layer will reduce the rate of all electrochemical reactions at the coating–steel interface. In combination with the blocking of active adsorption sites for oxygen on the oxidized steel surface by adsorbed amino-silanes, which bind to the oxide surface via silanol groups, this leads to the significant slow down of cathodic delamination. 11. Prospective and limitations From a practical point of view, the method may be used to predict the extent of cathodic delamination of organic coatings on metal surfaces. Although the experimental procedure was established using cold rolled low-alloy carbon steel, the cold rolled low-alloy carbon steel may be substituted by other metals which experience cathodic delamination. The method can easily be applied for estimation of the diffusion coefficients of various other cations, such as calcium and potassium ions, at the coating–steel interface by replacement of the sodium-selective electrode by a suitable ion selective electrode. However, the experimental procedure is limited to predict the delamination rate for inert-pigmented coatings. This is because the delamination process of non-inertpigmented coatings may involve chemical reactions, which the mathematical model for diffusion along the coating–steel interface does not consider. An example of such a coating could be aluminum pigments coatings. In comparison with the Scanning Kelvin Probe technique, which measures the surface work function difference between conducting, coated or semi-conducting materials and a metallic probe, this new experimental design requires larger measurement times. Furthermore, the work function can be directly correlated to the surface condition which means that the Scanning Kelvin Probe can provide information regarding interfacial phenomena such as delamination. However, the perturbation method does not rely on costly equipment, the experimental procedure for the preparation of coating samples is simple and the method requires little analysis of the collected data. Therefore, the perturbation method is a sturdy alternative to the Scanning Kelvin Probe when the interfacial diffusion coefficient is to be estimated. 12. Conclusions A novel practical method, which allows direct estimation of the apparent diffusion coefficients of sodium ions along the coating–steel interface, was successfully designed. The setup was applied in the characterization of the coating–steel interface of three commercial epoxy coatings and the estimated diffusion coefficients of sodium ions along the coating–steel interface show an acceptable repeatability. The apparent diffusion coefficients estimated are of similar magnitude as previous estimated values for the apparent diffusion coefficients of sodium ions along the coating–steel interface. The delamination rates predicted using the apparent interfacial diffusion coefficients and Fick’s second law, under the assumption of a transport-controlled mechanism, show qualitative agreement with the observed delamination rates in 0.5 M sodium chloride. This confirms that the cathodic delamination of inert-pigmented coatings, under the conditions studied,

77

is controlled by the interfacial transport of sodium ions from the defect to the delamination front. Nomenclature ci concentration of component i in bulk solution (mol m−3 ) ci,u concentration of component i in the upper chamber (mol m−3 ) Di molecular diffusivity of component i in the coating–steel interface (m2 s−1 ) apparent effective diffusivity of component i in the De,i coating–steel interface (m2 s−1 ) F Faradays constant (J V−1 mol−1 ) Ji flux of component i (mol m2 s−1 )  length of coating–steel interface (m) position of interface between coating and lower solution 0 (m) n integer R universal gas constant (J K−1 mol−1 ) Qi,t amount of component i transported across the coating sample (mol) t time (s) T absolute temperature (K) zi charge number of component i Greek letters  time-lag (s)  electrical potential (V) Acknowledgements Financial support by J.C. Hempel’s Foundation and The Technical University of Denmark is gratefully acknowledged. The authors are grateful to Fie Wilbek and Solveig Thorsteinsson for their assistance in some of the experiments. References [1] P.A. Sørensen, S. Kiil, K. Dam-Johansen, C.E. Weinell, J. Coat. Technol. Res. 6 (2009) 135. [2] H. Leidheiser, D.J. Mills, W. Bilder, Proc. Electrochem. Soc. 87 (1987) 23. [3] D. Gervasio, I. Song, J.H. Payer, J. Appl. Electrochem. 28 (1998) 979. [4] H.S. Wroblowa, J. Electroanal. Chem. 339 (1992) 31. [5] H. Wroblowa, S. Qaderi, J. Electroanal. Chem. 279 (1990) 231. [6] H. Leidheiser, W. Wang, L. Ingetoft, Prog. Org. Coat. 11 (1983) 19. [7] E.L. Koehler, Corrosion 40 (1984) 5. [8] W. Funke, Ind. Eng. Chem. Process. Res. Dev. 24 (1985) 343. [9] J. Parks, H. Leidheiser, Ind. Eng. Chem. Process. Res. Dev. 25 (1986) 1. [10] H. Leidheiser, W. Wang, J. Coat. Technol. 53 (1981) 77. [11] A. Leng, H. Streckel, M. Stratmann, Corros. Sci. 41 (1999) 579. [12] A. Leng, H. Streckel, M. Stratmann, Corros. Sci. 41 (1999) 599. [13] F. Deflorian, S. Rossi, Electrochim. Acta 51 (2006) 1736. [14] U. Steinsmo, J.I. Skar, Corros. Sci. 50 (1994) 934. [15] J.I. Skar, U. Steinsmo, Corros. Sci. 35 (1993) 1385. [16] O.O. Knudsen, J.I. Skar, Cathodic Disbonding of Epoxy Coatings—Effect of Test Parameters, NACE, 2008. [17] S.B. Lyon, L. Philippe, E. Tsuousoglou, Trans. Inst. Met. Finish. 84 (2006) 23. [18] M. Stratmann, R. Feser, A. Leng, Electrochim. Acta 39 (1993) 1207. [19] P.A. Sørensen, S. Kiil, K. Dam-Johansen, C.E. Weinell, Prog. Org. Coat. 64 (2009) 142. [20] T. Nguyen, D. Bentz, E. Byrd, J. Coat. Technol. 66 (1994) 39. [21] T. Nguyen, E. Byrd, D. Bentz, J. Adhes. 48 (1995) 169. [22] K.N. Strafford, P.K. Datta, C.G. Googan, Coatings and Surface Treatment for Corrosion and Wear Resistance, Ellis Horwood Ltd., Chichester, 1984. [23] M.W. Kendig, H. Leidheiser, Corrosion Protection by Organic Coatings, The Electrochemical Society, Pennington, 1987. [24] W. Furbeth, M. Stratmann, Corros. Sci. 43 (2001) 229. [25] J.M. Pommersheim, T. Nguyen, Z. Zhang, J. Adhes. Sci. Technol. 9 (1995) 935. [26] J.M. Pommersheim, T. Nguyen, Z. Zhang, J.B. Hubbard, Prog. Org. Coat. 25 (1994) 23. [27] K. Wapner, M. Stratmann, G. Grundmeier, Electrochim. Acta 51 (2006) 3303. [28] T. Nguyen, J.M. Pommersheim, Polym. Inorg. Interfaces 304 (1993) 15. [29] A. Leng, H. Streckel, M. Stratmann, Corros. Sci. 41 (1999) 547. [30] M.A. Hernandez, F. Galliano, D. Landolt, Corros. Sci. 46 (2004) 2281. [31] T. Nguyen, T.B. Hubbard, J.M. Pommersheim, J. Coat. Technol. 68 (1996) 45.

78

P.A. Sørensen et al. / Progress in Organic Coatings 68 (2010) 70–78

[32] K.N. Allahar, M.E. Orazem, K. Ogle, Corros. Sci. 49 (2007) 3638. [33] M. Huang, C. Allely, K. Ogle, M.E. Orazem, J. Electrochem. Soc. 155 (2008) C279. [34] M. Mulder, Basic Principles of Membrane Technology, Kluwer Academic Publishers, Dordrecht, 1996. [35] C. Carneiro, F. Oliveira, J. Nogueira, A. Mendes, J. Coat. Technol. 3 (2006) 159. [36] J. Crank, The Mathematics of Diffusion, Oxford University Press, Oxford, 1967. [37] A.L. Glass, J. Smith, J. Paint Technol. 39 (1967) 490. [38] A.L. Glass, J. Smith, J. Paint Technol. 38 (1966) 203. [39] J. Hu, J. Zhang, J. Zhang, C. Cao, J. Mater. Sci. 39 (2004) 4475. [40] J.M. Hu, J.Q. Zhang, C.N. Cao, Prog. Org. Coat. 46 (2003) 273. [41] D.R. Lide, CRC Handbook of Chemistry and Physics, Taylor and Francis, Boca Raton, 2007.

[42] R. Mills, Rev. Pure Appl. Chem. 11 (1961) 78. [43] R. Posner, K.S. Wapner, M.G. Grundmeier, Electrochim. Acta 54 (2009) 891. [44] R. Posner, T. Tizt, K. Wapner, M. Stratmann, G. Grundmeier, Electrochim. Acta 54 (2009) 900. [45] W.J. van Ooij, T. Child, Chemtech 28 (1998) 26. [46] T.F. Child, W.J. Van Ooij, Trans. Inst. Met. Finish. 77 (1999) 64. [47] H.P. Schrieber, R.Y. Qin, A. Sengupta, J. Adhes. 68 (1998) 31. [48] M.N. Sathyanarayna, M. Yaseen, Prog. Org. Coat. 26 (1995) 275. [49] A. Miszczyk, H. Szalinska, Prog. Org. Coat. 25 (1995) 357. [50] M. Rebhan, M. Rohwerder, M. Stratmann, Mater. Corros. 54 (2003) 19. [51] E.M. Pettrie, Handbook of Adhesives and Sealants, McGraw-Hill, 2000. [52] G. Grundmeier, M. Stratmann, Annu. Rev. Mater. Sci. 35 (2005) 571.