Modes and mechanisms for the degradation of fusion-bonded epoxy-coated steel in a marine concrete environment.

Fusion-bonded epoxy is used extensively to protect steel reinforcing bars from corrosion. This coating has proven to be a cost-effective material for increasing the service life of reinforced concrete structures. However, field observations have reported premature corrosion of fusion-bonded epoxy-coated reinforcing bars used in marine concrete environments, leading to severe cracking and spalling of the reinforced concretes. This study was initiated to provide a better understanding of the modes and mechanisms of the corrosion of fusion-bonded epoxy-coated steel exposed in a marine concrete environment. Grit-blasted steel panels were coated with two commercial fusion-bonded powder epoxy coatings at two thicknesses. Half of the coated panels were scribed, the other half were non-scribed. The panels were immersed in a saturated calcium hydroxide aqueous solution containing 0.6 mole/liter sodium chloride maintained at either 35[degrees]C or 50[degrees]C. Degradation was characterized and followed by infrared thermography, wet adhesion, and microscopic and analytical chemical techniques. Unscribed panels exhibited only water-induced adhesion loss, most of which was recovered after drying. However, in addition to the water-induced adhesion loss, scribed panels exhibited two modes of failure: anodic blistering near the scribe mark and cathodic delamination around the anodic blisters. Anodic blistering was attributed to localized crevice corrosion under the coating followed by blistering via an osmotic pressure mechanism. Cathodic delamination was probably induced by the alkaline cathodic reaction products, and water-induced adhesion loss was due to the presence of a water layer at the coating/steel interface.

Keywords: Adhesion loss, fusion-bonded epoxy coating, concrete solution, corrosion, degradation, marine, mode, mechanism, rebars, steel.

**********

Fusion-bonded epoxy is the dominant coating used today for protecting steel reinforcing bars from corrosion. This coating has very low permeability to chloride ions, high alkaline resistance, good flexibility, and good adhesion to steel. Fabrication of fusion-bonded epoxy-coated reinforcing bars (hereafter called epoxy-coated rebar or ECR) is a fast, efficient process. Mechanically blasted steel bars of any length or diameter are continuously fed to a chamber preheated at 230[degrees]C, followed by electrostatically sprayed with a powder epoxy, which fuses into a continuous coating, followed by quenching in water at ambient condition. Within a few minutes, a 20-meter fusion epoxycoated bar is ready for shipping to the production site. Since its introduction in the mid 1970s, (1) thousands of ECR-reinforced structures have been constructed in the United States. The corrosion resistance of ECR has been found to surpass uncoated and galvanized rebars, (2-4) and the general conclusion from these studies is that fusion-bonded epoxy coating is a cost-effective material for increasing the service life of reinforced structures.

However, reports started to circulate in the late 1980s indicating that, under severe conditions, such as in the South Florida Keys, epoxy-coated rebars may prematurely corrode, leading to a reduction of structural integrity of the reinforced concrete members and eventually cracking and spalling of the reinforced concretes (Figure 1). (5,6) Since then, Sagues and coworkers (7-12) have extensively investigated this problem both in the laboratory and in the field. Based on laboratory results from as-received and artificially-damaged ECRs exposed in model solutions and extensive field studies of numerous bridges in South Florida, the corrosion of ECR in marine concrete environments has been proposed to take place by the following steps (12): disbondment near the defects incurred during handling at the construction sites, loss of coating/steel adhesion in the region further away from the defect due to the penetration of concrete pore solution, and the development of corrosion macrocells at defects when chloride ions arrive. The corroded ECR is generally described by coating delamination, accumulation of corrosion products under the coating, and pitting of the metal. The problem is viewed as a result of normal production imperfections that are aggravated by fabrication, handling, and severe yard construction environment.

[FIGURE 1 OMITTED]

This study was initiated to reexamine the effectiveness, as well as to provide a better understanding of the degradation behavior, of fusion bonded epoxy-coated steel panel and rebars exposed to a marine concrete environment. Specifically, this article presents the modes and mechanisms of the degradation of fusion-bonded epoxycoated steel panels immersed in a saturated calcium hydroxide solution (model concrete pore solution) containing 0.6 mole/liter (3.5% by mass) sodium chloride. The experimental data and knowledge gained from this study should help both polymer-coated rebar producers to design better coating/steel rebar systems to be used in severe marine concrete environments, and rebar users to develop relevant testing methods for evaluating the corrosion resistance performance of ECR.

[FIGURE 2 OMITTED]

MATERIALS AND EXPERIMENTAL PROCEDURES

Variables included in this study were: two commercial fusion-bonded powder epoxy coatings, two coating thicknesses (130 and 190 [micro]m), two coating conditions (unscribed and scribed), and two test temperatures (35[degrees] and 50[degrees]C). Two main experimental programs were conducted in parallel. The first program (corrosion test) followed the increase of the corroded area as a function of exposure time; while the second program (wet adhesion test) tracked the loss in bond strength between the coating and steel substrate as a function of exposure time in the test solution.

Materials

Four-hundred flat, hot-rolled steel panels from the same batch measuring 152 X 102 X 3.2 mm were grit-blasted to a visually white-metal finish, and had a roughness profile of 50-70 [micro]m. After grit-blasting, each panel was individually wrapped in moisture resistant paper, which was not removed until the panel was ready to be coated. Panels were coated with two commercial, onepart, amine-cured powder epoxies that have substantially different glass transition temperatures commonly used for coating reinforcing bars.

Preparation of Coated Panels

The steel panels were randomly assigned to two groups; each group was coated with a different commercial epoxy coating. The coatings were applied to the substrate on a customized line of a rebar coating plant. In this process (Figure 2), the steel panels were heated to 214[degrees]C for 11 min, electrostatically sprayed with the powder epoxy coating, and then cured at 204[degrees]C for 11 min. A preliminary experiment indicated that these conditions produced well cured films. Coated panels were conditioned at ambient condition (24[degrees]C and 45% relative humidity) for two weeks before use. The coatings' thickness, which was controlled by the number of times the electrostatic gun passed over a panel, was measured at five locations on each panel using a thickness gauge. Only panels falling in the 92-152 [micro]m (thin) and 150-270 [micro]m (thick) thickness ranges were used. One-way analysis of variance at the 0.05 confidence level indicated that the average thicknesses between the thin and thick ranges were significantly different and the thicknesses within each range were not statistically different. Coated panels were tested for holidays using a 67.5 V holiday (pores, defects, etc., extended to the metal substrate) detector; panels containing holidays were excluded from the study.

Free Films

Free standing films of both coatings were prepared to measure the coatings physical properties and maximum water uptake using the following procedure. Polished brass panels were sprayed with polytetrafluoroethylene release agent and then heated in an oven at 200[degrees]C for four hours to remove residual solvent. The release agentsprayed panels were then coated with both powder epoxy coatings on the same customized line used in coating the steel panels. The coating films were easily removed from the substrate after cure. Physical properties measured included glass transition temperature ([T.sub.g]), maximum water uptake, tensile strength, tensile modulus, and maximum elongation at break for both dry and water-saturated films. The glass transition temperatures were measured by differential scanning calorimetry at a scanning rate of 10[degrees]C/min. The dry and wet tensile strengths and moduli were measured at ambient conditions (24[degrees]C and 45% RH) in a mechanical testing machine at a crosshead speed of 2 mm/min. The wet tensile properties and maximum water uptake were measured after immersing the free films in pH 8 water for two weeks at 35[degrees] [+ or -] 0.5[degrees]C The maximum water uptake was measured gravimetrically. The films of both coatings attained a maximum water uptake after four days of immersion.

Preparation of Specimens for Exposure

Half of the panels in each group were incised with a 25.4 X 1 mm scribe mark, using a milling machine. All scribe marks were visually inspected to ensure that no cracks or delaminations were present between the coating and the substrate. For the corrosion test panels, the scribe mark was located in the center of the panel in a direction parallel to its longer side. For wet adhesion test panels, the scribe was made at one end of the panel in the direction parallel to its shorter side. A cylindrical ring of poly(methyl methacrylate) was bonded with a silicone adhesive to the top surface of each of the corrosion test panels. The silicone adhesive was allowed to cure for at least two weeks prior to filling the ring with the test solution. For the wet adhesion test experiment, a rubber sealant was applied along the perimeter of each panel to contain the test solution. Additional information on the specimen configurations for the corrosion and wet peel adhesion tests can be found in references 13 and 14, respectively.

Exposure Solution and Conditions

Coated panels were exposed to a test solution containing saturated calcium hydroxide, Ca(OH)[.sub.2], and 0.60 mole/liter (3.5%, by mass) sodium chloride, NaCl. This solution is a simplified simulated pore solution of concretes used in a marine environment. (7) Half of the panels assigned to the corrosion experiment were immersed in the test solution heated in an oven at 35[degrees]C and the other half were in the same solution heated in an oven at 50[degrees]C. The wet adhesion test was conducted at 35[degrees]C only. In both experiments, the solution was continuously aerated by bubbling filtered and desiccated air through it. The oxygen concentration of the solution was periodically measured and found to be 7.0 [+ or -] 0.2 ppm (by volume). To avoid the effect of carbonation (reaction between C[O.sub.2] of the air and (CaOH)[.sub.2]), which slowly decreased the pH of the solution, test solutions were replaced with a fresh solution each week. The solution pH was frequently monitored and was maintained between 11.8 and 12.3.

Quantification of Degradation

INFRARED THERMOGRAPHY -- Blisters were observed to develop and grow along the scribe marks. The size and total area of each blister were quantified using a computer image analysis system connected to an infrared thermographic camera, displayed in Figure 3. This technique, which is based on the difference in thermal properties between the corrosion products and the uncorroded steel, has been demonstrated as a good, nondestructive way for quantifying the corrosion and blistering of organic-coated metals. (15) The measuring system consisted of an infrared camera equipped with a standard video output and an XYZ mounting table that can be programmed to precisely move the sample to any prespecified location.

WET PEEL ADHESION TEST -- Cathodic delamination and adhesion loss due to exposure were measured using a 90[degrees] wet peel adhesion test apparatus, schematically displayed in Figure 4. The instrument consisted of a linear bearing slider fixed to a computer-controlled universal testing machine outfitted with a 2.0 kg load cell. Coated panels were removed from the oven at prespecified exposure times, and the wet surface was immediately incised with a razor blade into eight 80 X 10 mm strips. Each specimen was carefully peeled from the substrate, leaving a length of 40 mm (from the scribe mark) of the specimen unpeeled (see Figure 4). The partially peeled coated panels were then positioned in the peel adhesion test apparatus. All the adhesion tests were conducted at room temperature and at a peel rate of 20 [+ or -] 0.1 mm per minute. All reported results were the average of eight specimens. The coefficients of variation, (standard deviation/mean) X 100, between specimens were always <7%. As will be demonstrated in the Results section, this wet peel adhesion test is a good technique for studying not only the rate of water-induced adhesion loss but also the rate of cathodic delamination.

Microstructure and Chemical Analyses of Degraded Areas

After 500 hr of immersion, several panels were removed from the exposure solution and analyzed for chemical elements associated with corrosion. In performing this analysis, 1 X 2 cm cross sections of the coated panels were prepared by cutting perpendicular to the scribe length using a diamond saw, followed by several polishing steps; the final polish was with a 0.1 [micro]m paste. Ethyl alcohol was used throughout the cutting and polishing operations. The spatial distribution of Na, Fe, Cl, and Ca was determined on the polished cross sections using a scanning electron microscope (SEM) equipped with a microprobe.

[FIGURE 3 OMITTED]

The concentrations of [Na.sup.+] and [Cl.sup.-] ions and the pH of the fluid in blisters were measured after the 2230-hr immersion exposure, using mini and micro selective electrodes (minielectrode for pH and Na, and microelectrode for Cl). This was done by lancing several blisters with a hypodermic needle and removing the fluid contained within. The pH of the steel surface of several wet adhesion specimens was ascertained at the termination of exposure using a phenolphthalein indicator.

Measurement of Thickness of the Water Layer At the Coating/Substrate Interface

This experiment was performed to verify the presence of a water layer at the coating/substrate interface. The thickness of the water layer between the fusion-bonded epoxy coating and an inorganic substrate was determined by FTIR-internal reflection spectroscopy (FTIR-MIR) using the detailed procedure previously described. (16) For this measurement, a water chamber was attached to the fusion-bonded epoxy-coated 50 X 10 X 3 mm spectroscopic grade, 45[degrees] parallel Si prism, which served as the substrate. The application of powder epoxy on the Si prisms was similar to that described earlier for coating the steel panels. Before applying the coating, except for the side to be coated, all sides of the Si prisms were masked with a pressure sensitive tape. Each of the two Si prisms was then mounted on a steel substrate using a rubber adhesive. The steel panels with the mounted Si prisms were then coated with the fusion-bonded epoxies and cured in the same way as with other steel panels. Each coated Si specimen with the water chamber attached to it was placed in an internal reflection accessory holder, while the measurement of water at the interface was carried out using an FTIR spectrometer with a variable angle internal reflection accessory. After filling the chamber with distilled water at 24[degrees]C, FTIR-MIR spectra were taken automatically every 15 min without disturbing either the specimens or the spectrometer. For the specimen configuration employed in this study, the only pathway for water migration from the environment to the interface is through the film thickness. All spectra were the coadditions of 128 scans taken at a resolution of 4 [cm.sup.-1]. Unpolarized light at a 45[degrees] incident angle and purged dry air were used.

[FIGURE 4 OMITTED]

RESULTS

Physical Properties and Water Uptake Of Fusion-Bonded Epoxy Coatings

The physical properties of fusion-bonded epoxy coatings, particularly the [T.sub.g] and the modulus, are often associated with the corrosion resistance performance of ECR. Table 1 presents maximum water uptakes and physical properties of both dry and water-saturated films. Each value and its one standard deviation are averages of values from five specimens. One-way analysis of variance was performed to determine which properties within and between coatings were significantly different at the 0.05 level of confidence. The [T.sub.g] and dry and wet maximum elongation at break of Coating E1 were significantly greater than those of Coating E2. The dry tensile strength and dry modulus of elasticity of Coating E1 were significantly less than those of Coating E2, but the wet moduli of elasticity between the two coatings were not significantly different. For Coating E1, no difference was observed between the dry and wet tensile strengths, but its wet tensile modulus was significantly greater than its dry modulus. And finally, both the dry tensile strength and modulus of Coating E2 were significantly greater than those saturated with water.

Degradation Mode

UNSCRIBED COATED PANELS -- Results from visual observation, wet peel adhesion tests, and chemical and microstructural analyses clearly showed that fusion-bonded epoxy-coated steel panels containing no defects immersed for 3072 hr in an aerated saturated Ca(OH)[.sub.2] solution containing 0.6 mole/liter NaCl at 35[degrees] and 50[degrees]C suffered only water-induced adhesion loss, and this adhesion loss was mostly recovered upon drying. There was no evidence of corrosion on any of the unscribed coated panels. Our results for rebars coated with the same coatings also showed similar results; that is, no evident corrosion was observed on the defect-free ECRs after six months of exposure to the same test solution at the same temperatures. These results are consistent with previous studies of ECR in concrete exposed to different marine environments, which showed that rebars that contain no defects can provide good protection against corrosion for a long period of time. (2,3)

SCRIBED COATED PANELS -- For scribed coated panels exposed to the same solution, the steel surface at the scribe mark remained uncorroded during exposure. This is as expected because iron is passive in the pH range used in this study. (17) However, panels containing scribe marks exposed to the same test solution clearly showed, in addition to the water-induced adhesion loss similar to that observed for the unscribed samples, two failure modes: anodic blistering near and along the scribe mark and cathodic delamination around the anodic blisters, but away from the scribes. A top view and a cross-sectional view of these degradation modes, together with the water-induced adhesion loss region, and their relative locations on a scribed panel are schematically displayed in Figures 5a and 5b, respectively. For comparison, organic-coated panels containing large defects exposed to a neutral NaCl environment are described by corrosion (not anodic blistering) at the defects and cathodic delamination away from the defects. (18,19) The characteristics of each of the two degradation modes and the water-induced adhesion loss of a fusion-bonded epoxy-coated steel panel exposed to aerated saturated Ca(OH)[.sub.2] solution containing 0.6 mole/liter NaCl as described in detail below.

[FIGURE 5 OMITTED]

Anodic Blistering: As indicated earlier, the steel surface at the scribe mark remained uncorroded during exposure. However, black corrosion products were observed to form under the coating in the immediate vicinity of scribe marks within hours after exposure. Blisters started to be visible (by the naked eye) above the corroded areas after approximately 500 hr of immersion. The location and spacing of the blisters near the scribes can be seen in Figure 6a. This is a photograph of two typical, scribed coated panels for the two coatings after immersion in the test solution at 35[degrees]C for 2230 hr. Figure 6b displays the scribed panels after removing the coatings with a knife, showing a dark appearance at the blistered sites. Typical unprocessed and computer-enhanced infrared thermographic images of the blisters within the 25 X 25 mm field of view are displayed in Figures 7a and 7b, respectively, showing a good contrast between the blisters and surrounding nonblistered area of the coating. The average compressive stress of 20 blisters was 0.84 [+ or -] 0.42 MPa, which was measured by compressing the blisters in a testing machine at a crosshead speed of 0.5 mm/min. (13) Another interesting feature of the blisters formed under the test solution was that only the size of the blisters increased with exposure time, with little increase in the number of blisters.

[FIGURE 6 OMITTED]

Ion measurements of the solution inside the blisters taken after 2230 hr of immersion revealed a [Cl.sup.-] ion concentration of (2.4-3.6) mole/liter, which was four to six times greater than that of the bulk solution. The pH in the blisters was between 4 and 5, and the concentration of [Na.sup.+] ion was approximately the same as that of the bulk. Elemental distributions of Cl, Fe, Na, and Ca of a cross section enclosing a blister near a scribe mark of a 2230-hr exposure panel are displayed in Figure 8, which also includes the back-scatter SEM image. The scribe mark is located on the right-hand side of each elemental map (it is easiest to see in the SEM image or Fe map), and the blister site is located to the left of the scribe. It is clear from the Fe and Cl maps that the material inside a blister contained a substantial amount of Fe corrosion products and [Cl.sup.-] ions. Ca and a small amount of Na are seen to be confined to the scribe mark, but not within the blistered area. A close examination of both the Fe map and SEM image revealed that pitting of the steel substrate had occurred within the blistered area.

When the coating was removed, a pasty black corrosion product was revealed inside the blisters, mainly magnetite ([Fe.sub.3][O.sub.4]), similar to that observed for localized corrosion. (20) When the pasty black corrosion products were scraped away, a shallow pit was visible to the naked eye, consistent with the SEM image of Figure 8. All the previously mentioned results, i.e., low pH, large [Cl.sup.-] ion concentration, metal pitting, and type of corrosion product, clearly indicate that the products formed inside the blisters were from the anodic half cell reactions of the corrosion process and that the blisters were produced as a result of the anodic reactions (hence the term anodic blistering).

[FIGURE 7 OMITTED]

Using data generated by infrared thermography shown in Figure 7, the growth of the anodic blisters as a function of time was followed, as depicted in Figure 9. The total area covered by the anodic blisters increased with immersion time and the rate of increase was greater at 50[degrees]C than at 35[degrees]C. The growth of the anodic areas, A(t), obeyed the power law model of the form:

A(t) = a[t.sup.b], where t is immersion time and a and b are constants.

Multifactor analysis of variance study indicated that the power law coefficients were not affected by coating thickness or coating type, but were significantly affected by the exposure temperature. That is, an increase of the exposure temperature decreased the induction time and increased the total anodic area. Values of the constant 'a' varied between -1.2 and -4.0 at 35[degrees]C and between -1.2 and -6.0 at 50[degrees]C, and those of the constant 'b' ranged from 0.2-0.6 at 35[degrees]C and 0.3-0.7 at 50[degrees]C. (13)

Cathodic Delamination: After approximately 500 hr of immersion at 30[degrees]C, the bond strength in the coating area surrounding the anodic blisters and in the vicinity of the scribe mark had fallen to the point where the coating could be easily removed from the steel substrate. After removing the coating, this area had a white metallic appearance, which turned violet when treated with a phenolphthalein indicator, suggesting that the delaminated area had a pH between 10 and 11. These characteristics, which are similar to those observed around the defects of organic-coated steel panels exposed to a neutral NaCl solution, suggested that the loss of adhesion in this region was induced by the cathodic half cell reactions (hence, cathodic delamination).

[FIGURE 8 OMITTED]

This postulation is substantiated by the wet adhesion test results shown in Figure 10. This figure shows the peel strength curves of the coating specimens near the center of the panel as a function of distance from the scribe mark for different exposure times. For each peel curve, two distinct regions of adhesion loss can be observed. The first region is described by the horizontal lines just above the force required to overcome the friction of the apparatus (approximately 0.2 kN/m), and the adhesion of this region was not recovered when the coating was dried. Additionally, this region of complete adhesion loss had a metallic appearance, which turned violet after spraying with a phenolphthalein indicator, supporting our belief that the loss of adhesion was due to cathodic delamination. As seen in Figure 11, the cathodic delaminated area moved further away from the scribe with exposure time, reaching a distance of approximately 15 and 27 mm from the scribe mark after 42 and 80 days of exposure, respectively. Another feature of this region was that it delaminated uniformly, following an elliptical pattern around the scribe mark as schematically depicted in Figure 5a.

The cathodic delaminated distance from the scribe mark was plotted as a function of exposure time, and the results are displayed in Figure 11 for both Coatings E1 and E2. The cathodic delamination, which was essentially independent of coating type, increased almost linearly and rapidly with immersion time up to 50 days, but appeared to slow down at longer immersion times.

Water-Induced Adhesion Loss: Another adhesion loss region of the fusion-bonded epoxy-coated steel exposed to the test solution is described by the upper portion of the peel strength curves of Figure 10. This region lost its wet peel adhesion strength gradually with exposure time, reaching a minimum value of about 0.8 kN/m after 60 days (Figure 10, water-induced adhesion loss). This adhesion-loss region had a white metallic appearance and was unaffected by the pH indicator, indicating that the pH < 10. Further, the adhesion loss in this region was recovered after drying. The loss of adhesion in this area, which was the same for panels with and without defects, is believed to be due to the presence of a layer of water at the coating/steel interface (see the Mechanism section for a detailed explanation).

The water-induced adhesion loss and its recovery after drying at ambient condition (24[degrees]C and 45% RH) is illustrated in Figure 12. The estimated coefficients of variation between the test specimens for both adhesion loss and adhesion recovery were <7%, as indicated in the Experimental section. It should be noted that the measurement of peel adhesion loss was only possible after 12 days of immersion, because at exposure times less than 12 days, the tensile strength of the film was less than that of the wet peel strength of the coating. Figure 12 shows that the loss of the wet peel adhesion strength in the water-induced adhesion loss region almost reached the plateau after a 40-day immersion, and that most of this adhesion loss was recovered after exposure to ambient conditions.

The quantitative wet adhesion results shown in Figure 10 reveal several important differences between the cathodic delamination and water-induced adhesion loss regions. First, the demarcation between the two regions was very distinct; that is, there was no gradual adhesion change between the two regions, as clearly seen by the vertical line connecting the upper portion and lower portion of each peel strength curve. Second, even at the minimum wet strength adhesion (approximately 0.8 kN/m), the peel strength of the water-induced adhesion loss region was still much greater than that of the cathodic delamination region. Another difference between these two regions was the water-induced adhesion loss region recovered after drying, while the cathodic delaminated areas showed essentially no peel strength after drying, as indicated earlier.

Degradation Mechanism

ANODIC BLISTERING -- Microscopic and chemical analyses indicated that the formation of the anodic sites is consistent with crevice corrosion, except that in this case, localized corrosion was between the epoxy coating and the steel substrate. Corrosion apparently took place in the areas under the coating a few millimeters from and along the scribe marks. These nano-scale or, perhaps, atomic-scale crevices were large enough to permit neutral liquid entry but narrow enough to maintain a stagnant zone. (21) The nature of these crevice sites is not certain. They were probably the disbonded areas between the coating and the steel substrate near the scribe marks, which were generated during the milling. For fusion-bonded epoxy-coated steel immersed in saturated Ca(OH)[.sub.2] solution containing NaCl, we believe that [Cl.sup.-] ions migrated along the coating/steel interface from the scribe mark to these crevice sites underneath the coating where corrosion cells were established. The interfacial transport of Cl- ion postulation was substantiated by an ion diffusion experiment using atomic absorption spectroscopy and free standing films in between the test solution chamber and distilled water chamber. The results showed no evidence of Cl nor Na or Ca ions diffusion through the free film of these coatings after three months of immersion, despite the fact that these films contained numerous large pores, which were clearly seen from the SEM images of the coating cross sections (not shown). This postulation is also consistent with the fact that the solubility of ions in a typical polymer matrix is extremely low. (22)

There were apparently two main processes involved in the anodic blistering. Anodic reactions took place at the micro crevice sites under the coating following by blistering at these sites. The principal corrosion reactions of iron that took place in a saline aqueous environment at the anodic sites under the coating include (23):

[FIGURE 9 OMITTED]

Fe.[H.sub.2][O.sub.ads] + [Cl.sup.-] [right arrow] Fe[Cl.sup.-.sub.ads] + [H.sub.2]O (1)

Fe[Cl.sup.-.sub.ads] + O[H.sup.-] [right arrow] FeO[H.sup.+] + [Cl.sup.-] + 2[e.sup.-] (rate limiting step) (2)

FeO[H.sup.+] + [H.sup.+] [right arrow] [Fe.sup.++] + [H.sub.2]O (3)

In saline solution, competitive adsorption between chloride and hydroxyl ions occurs at the crevice sites. However, the former (chloride mechanism) tends to prevail over the latter (hydroxide mechanism) because of the high chloride activity and high acidity (low hydroxide activity). (24)

The net result is the production of [Fe.sup.++] ions and electrons in the solution. In the presence of [Cl.sup.-] ions, the rate of formation of the oxidized layer has been found to be several orders of magnitude higher than that in water alone. (25) The [Fe.sup.++] ions produced in reaction (3) are hydrolyzed with acidification according to any of the following reactions (20,26):

[Fe.sup.++] + [H.sub.2]O [right arrow] FeO[H.sup.+] + [H.sup.+] (4)

[Fe.sup.++] + 2[H.sub.2]O [right arrow] Fe(OH)[.sub.2] + 2[H.sup.+] (5)

3[Fe.sup.++] + 4[H.sub.2]O [right arrow] [Fe.sub.3][O.sub.4] + 8[H.sup.+] + 2[e.sup.-] (6)

We believe reaction (6) was the principal reaction that took place at the anodic sites under the coating because the main solid corrosion products observed inside the blisters were black powdery [Fe.sub.3][O.sub.4], as indicated earlier. Further, based on Pourbaix, this reaction is thermodynamically favored (large negative enthalpy of formation) in the absence of oxygen. (20)

Reaction (6) leads to the formation of [H.sup.+][Cl.sup.-], which aggressively promoted the dissolution of iron, resulting in an increase of [Cl.sup.-] ions transport to the corroded sites. The effect is a rapidly accelerating, or autocatalytic, process at the bottom of the anodic blistered area under the coatings. With the buildup of excess [Fe.sup.++] and [H.sup.+] ions within the confined local electrolytes, negatively charged ions are required to preserve the charge neutrality. These negative ions were supplied by [Cl.sup.-] ions from the bulk solution, where [Cl.sup.-] ions were abundant. The primary route of [Cl.sup.-] ion transport from the solution to the anodic sites was along the coating/steel substrate interface, as indicated earlier. The result is an increase in concentration of ferrous chloride in the blisters following the reaction (26):

[Fe.sup.++] + 2[Cl.sup.-] + 4[H.sub.2]O Fe[Cl.sub.2].4[H.sub.2]O (7)

[FIGURE 10 OMITTED]

This reaction accounts for the presence of a large quantity of soluble chloride in the blisters (six times the bulk), as observed in the liquid found in the blisters as well as in the dry corrosion products shown in Figure 8. The preferentially high concentration of chloride ions in the blisters is similar to that observed for localized corrosion of bare iron, (21) and in blisters formed at the small defects of coated steel exposed to neutral chloride electrolytes. (27)

Thus, the principal products in the anodic blisters formed near the scribe mark of a fusion-bonded epoxy-coated steel panel immersed in saturated Ca(OH)[.sub.2] solution containing NaCl consisted mostly of Fe[Cl.sub.2].4[H.sub.2]O and [Fe.sub.3][O.sub.4]. The insoluble corrosion products are ion-selective, which allowed [Cl.sup.-] ions to continuously permeate through the rust layer and reach the steel surface. The water flow that accompanied [Cl.sup.-] ion permeation (electro-osmosis effect) was probably the principal source of water required to sustain the anodic reactions underneath the rust layers for the later stages, as proposed by Sato (24) for localized corrosion of steel. He found that water transport by the concentration gradient diffusion does not provide enough water to support the corrosion and hydration reactions in the occluded areas.

[FIGURE 11 OMITTED]

The blistering process at the anodic corroded sites is believed to be due to an osmotic pressure mechanism. As hydroscopic corrosion products formed at the anodic sites, a thermodynamic water activity difference was established between these sites under the coating and the exposure solution. This set up an osmotic pressure gradient that drove water from the outside of the coating to the anodic corroded sites underneath. Osmotic pressure has been proposed as the dominant mechanism responsible for the blistering of coatings systems exposed to neutral aqueous solutions. (28,29) Electro-osmosis flow of water to the anodic regions and mechanical stress may also contribute to the blistering process.

Cathodic Delamination

The characteristics of the cathodic delamination observed in this study for fusion-bonded epoxy-coated steel panels containing a scribe mark exposed to NaCl-containing Ca(OH)[.sub.2] solution are similar to those of organic-coated steel containing large defects exposed to a neutral NaCl solution. The cathodic delamination in neutral electrolytes has been extensively studied. (30-33) Therefore, only a brief, pertinent discussion is presented in this Mechanism section.

At the cathodic sites, oxygen is reduced, and in the presence of NaCl, a highly alkaline NaOH solution is formed at the coating/steel interface following the reaction (30,31):

1/2[O.sub.2] + [H.sub.2]O + 2[e.sup.-] + [Na.sup.+] [right arrow] 2([Na.sup.+]O[H.sup.-]) (8)

The presence of [Na.sup.+] in the cathodic delaminated areas has been verified by surface sensitive techniques. (22,34) The transport of [Na.sup.+] to the cathodic delaminated fronts is believed to be along the coating/steel interface. This is supported by our ion diffusion experiment, which showed no evidence of [Na.sup.+] diffusion through the coating free films, as indicated earlier. This postulation is also consistent with the results of cation interfacial diffusion studies by Leidheiser (30) and Nguyen et al. (35) for coated steel panels exposed to neutral electrolytes. Further, based on Kelvin probe evidence, Stratmann et al. (22,36) also suggested that not only do cations transport along the coating/steel interface from the bulk to the cathodic sites but also that this transport process is the rate determining step of the cathodic delamination. The transport of [Na.sup.+] ions to the cathodic delaminated fronts was likely enhanced by the electrical potential gradient that exists between the anodes and cathodes during corrosion activity. Cathodic potentials have been shown to strongly affect the cathodic delamination rate and the rate of [Na.sup.+] ion transport along the coating/steel interface of scribed organic-coated steel exposed to neutral electrolytes. (35,36)

In neutral chloride electrolytes, the pH of the solution at the delamination front has been reported to be as high as 14. (37) However, most pH measurements are integrated over a large volume of liquid and, thus, the pH values commonly reported in the literature are closer to 10. This value is similar to those measured in this study. The alkaline products formed from the cathodic reactions (e.g., reaction (8)) has been proposed as the main factor causing the delamination of polymer coatings from a steel substrate. Three main possible mechanisms have been advocated for the cathodic delamination for a polymer-coated steel exposed to saline environments: dissolution of the oxide layer, (30,37) alkaline hydrolysis of the polymers, (32,38) and displacement of the coating by a layer of high pH water. (39) In some coating systems, more than one mechanism has been observed, either simultaneously or in stages. (33) Based on the lack of an adhesion recovery after drying in the cathodic region observed in this study, we believe that the high pH water displacement mechanism was not the main failure mechanism for this coating system. This is because if the hydrogen bonds that are formed between the amine-cured epoxy hydroxyls and steel hydroxylated oxide hydroxyls (the main molecular interactions between an amine-cured epoxy and a steel substrate) were base-catalyzed displaced by water, some level of wet adhesion strength of the cathodic delamination region should have been regained after drying, similar to that observed in the water-induced adhesion loss region. That was not the case for this coating system.

Water-Induced Adhesion Loss: The loss of adhesion, which largely recovered upon drying, in the region beyond the cathodic delamination (see Figure 12) is attributed to the presence of a water layer, many monolayers thick, at the coating/steel (oxide) interface. This conclusion is substantiated by the FTIR spectral evidence shown in Figure 13a. These are FTIR-MIR difference spectra in the 1400-1800 [cm.sup.-1] and 2800-3800 [cm.sup.-1] regions, which contain the water OH bending band at 1640 [cm.sup.-1] and stretching band near 3400 [cm.sup.-1]. These two spectral regions also include the coating CH stretchings between 2800-3000 [cm.sup.-1] and the aromatic ring stretching in the bisphenol A portion of the epoxy resin near 1510 [cm.sup.-1]. These difference spectra were obtained by subtracting the spectra collected before exposure to water from those recorded at different time intervals after adding water to the chamber, as described in the Experimental section. The thickness of the epoxy film on the Si prisms was approximately 285 [+ or -] 10 [micro]m, as measured on a film removed from the substrate at the conclusion of the experiment using a micrometer. As emphasized in the Experimental section, there was no mirror or sample adjustment during the in-situ FTIR-MIR measurement, and the only way water could enter the coating/substrate interfacial region was through the film thickness. If there is no effect of water, all difference spectra would be a straight line at zero absorbance, as shown in Figure 13a. Bands above or below the zero base line of a difference spectrum indicate an increase or a decrease, respectively, of the concentration of a functional group. Figure 13a clearly shows that the intensities of the OH stretching near 3400 [cm.sup.-1] and OH bending at 1640 [cm.sup.-1] of water molecules increased with exposure time (for a comparison with FTIR spectrum of water only see reference 16).

Figure 13a also clearly shows that the strongly absorbed stretching band of the epoxy benzene ring at 1510 [cm.sup.-1] decreased with exposure time. As discussed in details previously, (40,41) the intensity increase of the water bands and intensity decrease of the coating material bands is a result of water entering the coating/substrate interfacial region and interacting with the evanescent wave, and is not a result of the film swelling due to water uptake. As the thickness of the water layer at the coating/substrate interface increased with exposure, the amount of the coating material within the probing depth of the evanescent wave decreased, as observed for the 1510 [cm.sup.-1] band, because the coating was pushed further away from the substrate. Although the results given in Figure 13a were for distilled water, similar spectra should be obtained when saturated Ca(OH)[.sub.2] solution was used because only water, and not [Ca.sup.++] ions, was transported through the film thickness of the fusion-bonded epoxy coating. Extensive FTIR-MIR data and analyses for in-situ measurement of water at the polymer/substrate interface have been reported for other organic materials on substrates, including water-reducible epoxy, unpigmented epoxies, alkyds, and asphalts. (16,40-42)

The amount and thickness of water at the interface between a fusion-bonded epoxy coating and an Si prism substrate were determined, and the results are displayed in Figure 13b. The thickness and mass data were calculated using the FTIR-MIR in-situ data given in Figure 13a and a two-layer model derived rigorously from the internal reflection spectroscopy given previously. (16) Inputs for the model are the refractive indices of the samples (water and epoxy) and the substrate, FTIR-MIR absorbance of a water film alone on the Si substrate, the penetration depths of the evanescent wave in water and in epoxy, and equilibrium water uptake in the film. Except for the last input, which is given in Table 1, the other inputs have been given in references 16 and 41. Figure 13b shows that, after 100 hr of exposure, a layer of water approximately 50 nm thick had been accumulated at the fusion-bonded epoxy/Si substrate interface.

It should be mentioned that the in-situ study of water at the interface used a very smooth Si surface, while the grit-blasted steel panels used in the wet adhesion loss study had a much more complex surface feature with sharp peaks and valleys. However, the Si substrate surface had a layer of Si[O.sub.2] having a thickness of 2.25 nm (as measured by an ellipsometer). Because Si[O.sub.2] is a highly energetic material, under ambient conditions the Si[O.sub.2]-covered Si substrate (Si[O.sub.2]-Si) is always covered with a layer of silanol groups and adsorbed water. (41) Thus, the chemical functional groups on the surface of this substrate are similar to those of steel and aluminum, which are, under normal conditions, also covered with hydroxylated oxide. (43) Therefore, although the thickness of the water layer at the epoxy/grit-blasted steel substrate might not be the same as that for the epoxy/Si[O.sub.2]-Si system, the effect should be the same; that is, water did enter and form a layer many monolayers thick at the coating/steel substrate interface. Nguyen and coworkers (40-42) have shown that the thickness of the water layer at the coating/substrate interface varies with coating type and surface treatment of the substrate. For example, essentially no water was observed at the interface for stochiometrically mixed clear aminecured epoxy on the silane-treated Si[O.sub.2]-Si substrate, a few nanometers for a clear model epoxy, but tens of nanometers for commercial alkyds, water-reducible epoxy, and asphalts on cleaned, untreated Si[O.sub.2]-Si substrates. The difference has been attributed to the migration of hygroscopic additives, surfactants, and unreacted curing agent and resin from the commercial coatings to the interface, which enhances interfacial water uptake.

[FIGURE 12 OMITTED]

Further evidence of a thick water layer at the coating/substrate interface comes from the work of Funke and Haagen, (44) who have observed the so-called "crossover" point for systems when the coated steel panels take up more water than that of the free films. Water uptake above the crossover point indicates that a substantial amount of water has accumulated at the interface. Additional support comes from studies of water adsorption on iron and other inorganic materials. For example, Zettlemoyer et al. (45) noted that a monolayer of water exists on ferric and nickel oxides at low relative humidities, but multilayers exist at a relative humidity of 90%. Similarly, Bowden and Throssell (46) reported up to 20 molecular layers of water on aluminum, iron, and Si[O.sub.2] surfaces at ambient temperatures and relative humidities.

[FIGURE 13 OMITTED]

The sorbed water layers closest to the surface are strongly bonded and difficult to remove. However, increasing the coverage of water will, at some point, affect the strength of the steel (oxide)/polymer (epoxy) bonds. The first direct evidence of a linkage between the water layer at the polymer/substrate interface and the adhesion loss is given by Nguyen and coworkers. (47,48) Using well-cleaned, untreated and silane-treated silicone wafers and glass fibers, these authors showed a strong correlation between the thickness of the water layer at the epoxy/substrate interface and the peel adhesion strength loss of the epoxy on a flat silicon substrate and the interlaminar shear strengths of the glass fiber/epoxy composites.

Further evidence to support the adhesion loss due to water at the polymer/substrate interface was obtained from the extensive data on adhesion loss of organic coatings and adhesives resulting from exposure to water and high relative humidities. In an extensive review and analysis of water disbondment of organic-coated substrates, Leidheiser and Funke (49) concluded that the bonding strengths of coating/untreated, inorganic substrate systems decrease significantly after exposure to water and high humidities and that "water disbondment is a consequence of a discontinuous or, in some cases, a continuous water film several to many molecular layers in thickness." Extensive data on durability of adhesives joints prepared using untreated and treated metal surfaces collected in a monograph by Kinloch (50) also shows a substantial bonding strength loss with exposure to water. Further, the degree of adhesion loss depends on the type of surface treatment, and the decrease is accelerated with increasing temperature.

The loss of adhesion due to water at the polymer/substrate interface can be explained from the surface free energy and molecular interaction theories. The bonding between a polymer coating and an oxide-covered steel substrate is governed by weak secondary forces, which are generally less than 25 kJ/mole. (51) On the other hand, the magnitude of the bonds which water forms with oxide surfaces are substantially higher, typically in the 40-65 kJ/mole range. (52) Consequently, the affinity of water for an oxide is greater than that of polymer to the same substrate. Free surface energy analysis by Kinloch (51) has also shown that the reversible work of adhesion between an organic film, including epoxy, and an oxide, including [Fe.sub.2][O.sub.3], is highly negative in the presence of water, implying that the epoxy/[Fe.sub.2][O.sub.3]-steel substrate bonds are not stable in water. This means that water is likely to enter the interface and displace an epoxy from a steel surface when this system is exposed to water or to high relative humidities.

The assertion that the interactions between epoxy/steel (oxide) bonds are weak and that this weak interaction cannot resist water displacement is strongly supported by a comprehensive analysis by Bolger and Michaels (43) on the interactions between organic molecules and inorganic substrates. They showed that there are only a few organic/substrate combinations, e.g., strong acidic organic/strong basic substrate or strong basic organic/strong acidic substrate, that can resist the displacement by water. This is because the most common metal elements (including Fe) are considerably more electropositive than the carbon atoms in organic compounds. Consequently, the electron density on the oxygen atom in the [Fe.sub.2][O.sub.3] is considerably greater than that on the oxygen in water or in most organic compounds. Therefore, covalently-bonded interfaces are not commonly formed in organic film/unmodified oxide systems, and the secondary-force bonds that do occur are too weak to resist the affinity of water to the polar, high-energy substrate. Thus, if high energy surfaces (e.g., [Fe.sub.2][O.sub.3]-covered steel) are not modified, water is likely to form a layer at the interface when an organic film/inorganic solid is exposed to water or high relative humidities.

The amount of water at the coating/substrate interface is greatly increased if the interface also contains hydrophilic contaminations or an osmotic driving force existing between the interface and the outside. Unfortunately, water soluble inorganic and organic materials are almost ubiquitous contaminants at the coating/substrate interface, either present before the coating application or migrating there from the coatings or from the environments during exposure.

CONCLUSIONS

Fusion-bonded epoxy coating is used extensively to protect steel reinforcing bars from corrosion. This coating has been shown to be a cost-effective material to increase the service life of reinforced structures. However, under severe corrosion environments, fusion-bonded epoxy-coated rebars have been reportedly prematurely corroded, leading to severe cracking and spalling of the reinforced concrete structures. A study has been initiated to provide a better understanding about the degradation behavior of fusion-bonded epoxy-coated steel panels exposed to marine concrete environments. Parameters studied included: coating condition (scribed and unscribed), glass transition temperature (coating type), coating thickness, and exposure temperature. Scribed and unscribed epoxy-coated panels were immersed in a saturated calcium hydroxide solution containing 0.6 mole/liter sodium chloride at 35[degrees] and 50[degrees]C. The degradation was quantified by infrared thermography and wet adhesion techniques. The microstructure and chemical properties of the degraded areas were characterized by microscopic and chemical analytical techniques. The following conclusions are made with regards to degradation modes and mechanisms:

(1) Unscribed coated panels suffered only water-induced adhesion loss, which was mostly recovered after drying. No evidence of corrosion was observed for unscribed coated panels after more than four months of immersion in test solution at 35[degrees] or 50[degrees]C.

(2) Scribed coated panels degraded by two modes: anodic blistering near and along the scribe mark and cathodic delamination around the anodic blisters and away from the scribe mark. Water-induced adhesion loss also occurred in scribed coated panels in the region beyond the cathodic delamination.

(3) Anodic blistering occurred through two processes: localized corrosion at the anodic sites followed by blistering at these sites via an osmotic pressure mechanism. Anodic blistering was greatly accelerated with an increase in temperature of the test solution from 35[degrees] to 50[degrees]C.

(4) Cathodic delamination occurred away from the scribe and was independent of coating type or thickness. [Na.sup.+] transport from the solution to the cathodic delamination front is believed to be along the coating/steel interface. The mechanism for cathodic delamination was similar to that observed for coated panels exposed to a neutral NaCl solution; that is, it was probably caused by the alkaline cathodic reaction products formed at the delaminated front.

(5) A water film of many monolayers thick at the coating/steel interface is believed to be responsible for the water-induced adhesion loss.

The experimental data and knowledge gained from this study should greatly help polymer-coated rebar producers to design better coating/steel rebar systems to be used in severe marine concrete environments.

Table 1 -- Maximum Water Uptake and Physical Properties of Two
Fusion-Bonded Epoxy Coatings

                               Coating E1            Coatings E2
Physical Properties         Average  Stand. Dev.  Average  Stand. Dev.

[T.sub.g], [degrees]C       115          2         85          2
Tensile Strength, MPa  Dry   49.5        1.3       56.9        3.7
                       Wet   46.8        1.9       47.9        3.2
Tensile Modulus, GPa   Dry    1.37       0.03       1.71       0.17
                       Wet    1.55       0.13       1.69       0.15
Max. Elongation        Dry    5.7        0.4        5.2        0.2
at Break (%)           Wet    4.9        0.4        4.4        0.3
Max.water uptake, %           1.7        0.1        2.0        0.1
Total Surface Free
Energy, (a)
mJ/[m.sup.2]                 38          1.7       41          2.0
Surface Polarity (b)          0.18                  0.14

(a) Determined using water and methylene iodine contact angles and
harmonic-mean equations.
(b) Ratio between polar force component and total surface free energy;
both of which are obtained from the harmonic-mean equations.

ACKNOWLEDGMENTS

The authors would like to thank Dr. Y. Paul Virmani of FHWA for his support throughout this project and to FHWA for providing the research funds. The completion of this project was greatly facilitated through the help of Mr. Greg Weaver of Lane Enterprises, Inc., in preparation of fusion bonded epoxy-coated panels. Thanks are also extended to Ned Embree, Eric Byrd, and Paul Stutzman of NIST for their laboratory assistance, and Prof. Alberto Sagues for providing the picture of Figure 1.

Presented at the 81st Annual Meeting of the Federation of Societies for Coatings Technology, November 12-14, 2003, in Philadelphia, PA.

References

(1) Clifton, I.R., Beeghly, H.F., and Mathey, R.G., "Nonmetallic Coatings for Concrete Reinforcing Bars," Federal Highway Administration Report No. FHWA-RD-74-18, 1974.

(2) Satake, J., Kamakura, K., Mikami, N., and Swamy, N., in Corrosion of Reinforcement in Concrete Construction, Ellis Horwood Ltd., U.K., pp. 358-377, 1983.

(3) Yeomans, S.R., "Corrosion Testing of Black, Galvanized and Epoxy Coated Reinforcing Steel in Concrete," Corrosion/93, Paper No. 329, National Association of Corrosion Engineers, Houston, TX, 1993.

(4) Burke, D.F., "Performance of Epoxy-Coated Rebar, Galvanized Rebar, and Plain Rebar with Calcium Nitrite in a Marine Environment," Proc. International Conference on Corrosion and Corrosion Protection of Steel in Concrete, University of Sheffield, England, July, 1994.

(5) Kessler, R.J., Powers, R.G., and Langley, R.M., "Corrosion Evaluation of Substructure Cracks in Long Key Bridge," Florida Dept. of Transportation, Corrosion Report No. 86-3, Gainsville, FL, 1986.

(6) Powers, R., "Corrosion of Epoxy-Coated Rebar, Keys Segmental Bridges," Florida Dept. of Transportation, Corrosion Report No. 88-8A, Gainsville, FL, 1987.

(7) Sagues, A.A. and Powers, R., "Effect of Concrete Environment on the Corrosion Performance of Epoxy-Coated Reinforcing Steel," Corrosion/90, Paper No. 311, National Association of Corrosion Engineers, Houston, TX, 1990.

(8) Sagues, A.A., Powers, R., and Zayed, A., in Corrosion of Reinforcement in Concrete, Page, C., Treadaway, K. and Bamforths, P. (Eds.), Elsevier Appl. Sci., New York, pp. 539-549, 1990.

(9) Zayed, A. and Sagues, A., Corrosion Sci., 30, 1025 (1990).

(10) Sagues, A., Perez-Duran, H.M., and Powers, R.G., Corrosion, 47, 884 (1991).

(11) Sagues, A.A., Powers, R., and Kessler, R., "Corrosion Processes and Field Performance of Epoxy-Coated Reinforcing Steel in Marine Substructures," Corrosion/94, Paper No. 299, National Association of Corrosion Engineers, Houston, TX, 1994.

(12) Sagues, A., Powers, R., and Kessler, R., "Corrosion Performance of Epoxy-Coated Rebar in Florida Key Bridges," Corrosion/01, Paper No. 1642, National Association of Corrosion Engineers, Houston, TX, 2001.

(13) Martin, J.M., Nguyen, T., Alsheh, D., Embree, N., and Byrd, E., "Degradation of Powder Epoxy-Coated Panels Immersed in a Saturated Calcium Hydroxide Solution Containing Sodium Chloride," Federal Highway Administration, Publication No. FHWA-RD-94-174, 1995.

(14) Alsheh, D., Nguyen, T., and Martin, J.W., "Adhesion of Fusion-Bonded Epoxy Coating on Steel in Alkaline Solution," Proc. Adhesion Society Meeting, p. 209, February, 1994.

(15) McKnight, M.E. and Martin, J.W., "Detection and Quantitative Characterization of Blistering and Corrosion of Coatings on Steel Using Infrared Thermography," JOURNAL OF COATINGS TECHNOLOGY, 61, No. 775, 57 (1989).

(16) Nguyen, T., Bentz, D., and Byrd, E., "A Study of Water at the Coating/Substrate Interface," JOURNAL OF COATINGS TECHNOLOGY, 66, No. 834, 39 (1994).

(17) Uhlig, H., Corrosion and Corrosion Control, Wiley, New York, 2nd ed., p. 98, 1963.

(18) Leidheiser, H. and Kendig, M., Corrosion, 32, 69 (1976).

(19) Dickie, R.A. and Smith, A.G., Chemtech, 10, 31 (1980).

(20) Pourbaix, M., in Localized Corrosion, Staehle, R.W., Brown, B.F., Kruger, J., and Agrawal, A. (Eds.), National Association of Corrosion Engineers, Houston, TX, pp. 12-33, 1971.

(21) Fontana, M.G., Corrosion Engineering, McGraw Hill, New York, 3rd ed., pp. 51-59, 1986.

(22) Stratmann, M., Feser, R., and Leng, A., Electrochimica Acta, 39, 1207 (1994).

(23) Cohen, M., in Corrosion Chemistry, Brubaker, G.R. and Phipss, P.B.P., (Eds.), ACS Symposium Series 89, American Chemical Society, Washington, D.C., pp. 126-152, 1979.

(24) Sato, N., Corrosion Sci., 45, 354 (1989).

(25) Ashley, G.W. and Burstein, G.T., Corrosion, 47, 908 (1991).

(26) McCafferty, E., J. Electrochemical Society: Sci. Technol., 128, 40 (1981).

(27) Nguyen, T. and Lin, C., J. Adhesion, 33, 241 (1991).

(28) Kittelberger, W.W. and Elm, A.C., Industrial Engineering Chemistry, 39, 876 (1946).

(29) Funke, W., Prog. Org. Coat., 9, 29 (1981).

(30) Leidheiser, H. Jr., Wang, W., and Igetoft, L., Prog. Org. Coat., 11, 19 (1983).

(31) Funke, W., Ind. Eng. Chem. Prod. Res. Dev., 24, 343 (1985).

(32) Dickie, R.A., in Polymeric Materials for Corrosion Control, Dickie, R.A. and Floyd, F.L. (Eds.), ACS Symposium Series 322, American Chemical Society, Washington, D.C., p. 137, 1986.

(33) Watts, J.F., J. Adhesion, 31, 73 (1989).

(34) Hammond, J.S., Holubka, J.M., and Dickie, R.A., "Surface Analysis of Interfacial Chemistry in Corrosion-Induced Paint Adhesion Loss," JOURNAL OF COATINGS TECHNOLOGY, 51, No. 655, 45 (1979).

(35) Nguyen, T., Hubbard, J.B., and Pommersheim, J.M., "Unified Model for the Degradation of Organic Coatings on Steel in a Neutral Electrolyte," JOURNAL OF COATINGS TECHNOLOGY, 68, No. 855, 45 (1996).

(36) Stratmann, M., Leng, A., Furbeth, W., and Gehmecker, H., Prog. Org. Coat., 27, 45 (1996).

(37) Ritter, J.J. and Kruger, J., in Corrosion Control by Organic Coatings, Leidheiser, H. Jr. (Ed.), National Association of Corrosion Engineers, Houston, TX, p. 28, 1981.

(38) Hammond, J.S., Holubka, J.M., De Vries, J.E., and Dickie, R.A., Corrosion Sci., 21, 239 (1981).

(39) Koehler, E.L., Corrosion, 40, 5 (1984).

(40) Nguyen, T., Byrd, W.E., and Lin, C., J. Adhesion Sci. Technol., 5, 697, 1991.

(41) Nguyen, T., Byrd, W.E., and Bentz, D., J. Adhesion, 48, 169 (1995).

(42) Nguyen, T., Bentz, D., Byrd, W.E., and Lin, C., Prog. Org. Coat., 27, 181 (1996).

(43) Bolger, J.C. and Michaels, A.S., in Interface Conversion for Polymeric Coatings, Elsevier, NY, pp. 3-60, 1969.

(44) Funke, W. and Haagen, H., Ind. Eng. Chem. Prod. Res. Dev., 17, 50 (1980).

(45) Zettlemoyer, A.C., Micale, F.J., and Klier, K., in Water, A Comprehensive Treatise, Franks, F. (Ed.), Plenum Press, New York, Vol. 5, Chapter 5, 1975.

(46) Bowden, F.P. and Throssell, W.R., Nature, 167, 601 (1957).

(47) Nguyen, T., Byrd, W.E., Alsheh, D., McDonough, W., and Seiler, J., Materials Research Society Symposia, 385, 57 (1995).

(48) Nguyen, T., Byrd, E., Alshed, D., Aouadi, K., and Chin, J., Proc. International Conference on Durability of Composites Materials in Construction, Benmokrane, B. (Ed.), Advantage Inc., Montreal, Canada, p. 451, 1998.

(49) Leidheiser, H. Jr. and Funke, W., J. Oil & Colour Chemists' Assoc., 70, 121 (1987).

(50) Kinloch, A.J. (Ed.), Durability of Structural Adhesives, Applied Sci., NY, 1983.

(51) Kinloch, A.J. (Ed.) Durability of Structural Adhesives, Kinloch, A.J. (Ed.), Applied Sci., NY, pp. 1-38, 1983.

(52) Thiel, P.A. and Madey, T.E., Surface Sci. Rep, 7, 211 (1987).

Tinh Nguyen and Jonathan W. Martin--National Institute of Standards and Technology*

*100 Bureau Dr., MS8621, Gaithersburg, MD 20899.