(at http://www.jcse.org/)

Volume 2 Extended Abstract 9

Submitted 26th August 1999

The Behaviour of Epoxy Powder Coatings on Mild Steel under Alkali Conditions

A.B. Darwin and J.D. Scantlebury

Corrosion and Protection Centre, UMIST PO Box 88, Manchester M60 1QD, UK
E-Mail Address:

Keywords: Epoxy lacquers, Power coatings, Mild steel, Alkali environments, Adhesion,

Introduction

As part of a study of epoxy-coated rebar steel in concrete, work was also carried out at open circuit, using a clear epoxy powder and in various alkali environments. Variations in specific cation, pH and chloride ion concentration were examined. The mechanisms of adhesion loss is investigated.

Experimental

Cold-rolled mild steel Q-Panels, 1 mm thick, were degreased by the process of washing in detergent solution, rinsing well in de-ionised water and then spraying with ethanol.

Pigment-free epoxy resin used was supplied by Akzo Nobel Powder Coatings. The panels were heated in an oven, such that when they were removed and the epoxy powder sprayed onto them they would be in the temperature range 235-240oC, a requirement of the coating for its curing.. A Gema Volstatic manual fluidised bed powder system and PG1 powder gun was used. Sufficient powder was used to achieve a cured coating thickness of 100m m. Cooled powder coating thickness was at all times within the range 100 ± 10 µm. Panels with holidays in the powder coating were not used, the holidays being detected by using an oscilloscope to pass a sine wave across the coating. A 3.0 mm diameter holiday was introduced into the epoxy coating by means of a flat-head drill bit such that the Q-panel steel was exposed.

Test cells were then created by adhering a perspex tube of 5 cm diameter and 6 cm nominal height to the epoxy surface of the mild steel panel, such that the artificial holiday was at its centre. Adhesion was achieved by the sparing use of Silastic 732 RTV silicone sealant.

The test solutions used are shown in Table 1. All testing was made at open circuit. A minimum of three panels were used for each exposure condition. The pH of the solutions was measured daily and adjusted as necessary to maintain the correct activity. Delamination rate (apparent by a lighter halo of delaminated coating) and half-cell potential measurements were made.

Results and Discussion

Typical potential/time responses of FBE-coated panels with a 3mm holiday exposed to the different electrolytes are shown in Figure 1.

All the FBE-coated panels in solution with no chloride have potentials indicating their passivity or near-passivity. Recorded panel potential becomes more base as solution pH is increased. Their visual appearance also remained essentially unchanged. The exposed steel at the 3mm holiday remained bright and untarnished. No corrosion products were visible. Delamination of coating from substrate could be followed by measuring the expansion of a circle of lighter or occasionally darker-coloured FBE-coating from the artificial holiday (Figure 2).

The effect of the addition of 0.6M chloride to the solutions that non-chromated FBE-coated panels were exposed to is shown in Figure 3. The potentials of all the panels became more base, in the region of –400 to –550mV. The exposed substrate at the artificial holiday tended to remain bright and untarnished. Blue deposits began to appear underneath the coating adjacent to the holiday. At the edges of the artificial holidays adjacent to the blue areas under the coating, red deposits sometimes appeared. These observations strongly suggest that corrosion was occurring. The presence of a region of bright visibly uncorroded substrate adjacent to and at the delamination front indicates that delamination of the FBE coating from its substrate was not being caused by anodic undermining. The presence of a cathode region at the coating/substrate disbondment front has been widely reported previously by many authors, and is often an implicit and sometimes explicitly stated part of their postulated coating delamination mechanism. These authors include Wiggle, Smith and Petrocelli(125), Funke(121), Schwenk(168), Leidheiser(129), Watts and Castle(131), Holubka, deVries and Dickie(141), Ritter and Rodriguez(146), Nguyen, Hubbard and McFadden(163), Skar and Steinsmo(162), and Fürbeth and Stratmann(149). From potential / time data it may be seen that corrosion of the FBE-coated test specimens at open circuit initiates soon after they are placed in the test solutions containing chloride ions. The corrosion process does not take place on the exposed mild steel substrate of the holiday, but is seen to develop at its edge. A tafel plot of substrate potential / current response in pore solution can be seen to be virtually identical in comparison with pore solution with chloride (Figure 4). Only when the bond between the mild steel test specimen and its encapsulating epoxy araldite resin is poor does evidence of crevice corrosion appear. It is therefore the belief of the author that corrosion initiates in a crevice between the FBE-coating and mild steel substrate, caused by the passive current density on the steel.

The delamination data derived for each exposure condition was averaged. These values are plotted in Figure 5, where the average radial delamination of the panel is plotted against the square root of the time of its exposure to the test solution. A linear relationship is found when the delamination data is plotted using these axes.

It is also noted that on extrapolation of these plots to the x axis there is a period of time before delamination of the FBE coating from its substrate begins to take place (delay time). The calculated delay time along with the equations of the lines of the plotted data is displayed in Table 2. No data for panels exposed to pH 12.0 KOH solution is plotted because no delamination could be found after 6 months testing.

Panel Type Immersant Solution Slope,m (=Kd, Delam. Rate Constant) Constant, c Delay Time (days0.5)
NCr-FBE [No Chloride]      
" KOH: pH 13.2 0.165 -0.649 3.9
" pH 14.0 1.427 -6.886 4.9
" NaOH: pH 13.2 0.227 -1.122 4.9
" Pore Soltn:

pH 13.5

0.870 -4.502 5.2
" [With Chloride]      
" KOH/KCl 2.083 -7.064 3.4
" NaOH/NaCl 1.487 -5.427 3.6
" LiOH/LiCl 0.549 -0.781 1.4
" Pore/Cl 1.834 -7.947 4.3
Table 2: Calculated Equation of Line and Delay Time for FBE-Coated Panels at Open Circuit

Discussion

A linear relationship is derived when radial disbondment of the FBE-coating on both panels and rebars is plotted against the square root of time of immersion of the specimen. This relationship is found for corroding and passive systems. The parabolic nature of delamination shows that rate of FBE disbondment is greatest upon its initiation and subsequently decreases with time until a theoretical cessation of disbondment occurs after an infinite time period. More importantly from a mechanistic viewpoint, this relationship indicates that the path of the factor that is the rate-determining step in the speed of the delamination reaction is along the coating / substrate interface from the coating holiday. This relationship has also been found for coated systems corroding at open circuit by Leidheiser and co-workers(100,155,158), Sharman, Sykes and Handyside(159), and Nguyen and co-workers(163-167), and systems under cathodic polarisation by Kendig, Addison and Jeanjaquet(132). If transport of this delamination parameter was through the coating, then a linear relationship between radial delamination and time of immersion would be encountered.

Kendig(132) has described this parabolic relationship mathematically as:

d = d0 + Kdt1/2

where d = radial delamination from the flaw in the coating.

d0 = a constant relating to the delay time of the system.

Kd = delamination rate constant.

t½ = square root of time of immersion.

There is considerable evidence that this part of the mechanistic delamination process occurs at open circuit potentials in corroding systems by workers such as Parks and Leidheiser(100), Leidheiser, Wang and Igetoft(158), and Sharman, Sykes and Handyside(159). Referral to Table 2 informs us that delamination rates in corroding systems tend to be greater than those in passive. It is therefore likely that the rate-determining step in the delamination process of corroding and passive systems will be different.

The FBE-panels which were exposed to passivating solutions at open circuit underwent delamination. The only electrochemical reactions which took place on their substrates in these conditions would have been the extremely small passive current derived from the formation and dissolution of the substrate surface oxide film. It is therefore unlikely that the cause of delamination was an electrochemical process.

When the delamination rate constants of the FBE-panels are plotted on a logarithmic scale against the respective pHs of the passivating test solutions that they were exposed to, a linear relationship is derived (Figure 6). Concentration of hydroxyl ions in solution is therefore directly proportional to the rate of FBE coating delamination, and the rate-controlling step in the delamination process is the passage of hydroxyl ions along the coating/substrate interface. This information indicates that it is these hydroxyl ions which are directly responsible for coating delamination. Kendig, Addison and Jeanjaquet(132) contend that hydroxide ions cause coating delamination by lowering electrolyte surface tension, thereby catalysing wetting of the steel and accelerating coating displacement by electrolyte penetration. They support this claim with data showing a linear decrease in the surface tension of 0.5M NaCl on mild steel with pH in the range of 8-11 (Figure 7). Kendig's supposition may be supported by the linear delamination rate constant/pH fit generated by the flat panel data, providing the linear relationship between pH and surface tension is continued at a higher pH region. It is noted from his graph that it is based on a small number of data points, and that a logarithmic fit could just as easily be made through them, making his argument for surface tension being an important delamination parameter more tenuous. It is equally possible that the linear pH / radial delamination rate constant relationship is related to the increasingly alkaline environments affecting acid-base interactions between coating and substrate, and hence polymer adsorption and adhesion, as suggested by Fowkes and Mostafa(138).

Extrapolation of the pH / Kd graph to the point of zero delamination rate constant correlates with a bulk solution pH of approximately 13.0. This value will be associated with the minimum pH required to cause disbondment in the FBE / substrate system. At pHs below 13.0 there should therefore be no delamination. It has been shown that there was no delamination of coating from a non-chromated FBE-panel after 6 months of exposure to a pH 12.0 solution. These observations suggest that the disbondment mechanism is unlikely to be either the chemical disruption of the polymer / metal oxide bonds by the presence of polar water molecules or through creation of a hydrodynamical stress on the bonds, as suggested by Funke(116). If the disbondment mechanism were related to hydrodynamical stress on the bonds, then increasing solution osmotic pressure would be expected to result in a decrease in delamination rate. It has been shown that delamination will occur at pHs greater than approximately 13.0; the proposed oxide dissolution delamination mechanism proposed by Ritter and his co-workers(130,144-146) cannot therefore be the cause of this delamination because it will not be thermodynamically viable at a pH of 13.2(17). That the relationship between pH and Kd is linear is of interest, since this means that Kd is proportional to the logarithm of hydroxyl ion concentration in solution. The delamination mechanism is therefore not a simple one; hydroxyl ions are not involved in a {1 hydroxyl ion : 1 coating/substrate adhesive bond disrupted} relationship. It is therefore unlikely that delamination of the FBE coating ion is caused by its saponification.

Hydrated cation mobility is plotted versus delamination rate constant for corroding FBE-panels in test solutions with chloride in Figure 8. A trend is found showing that increasing hydrated cation mobility in the test solution results in an increased rate of coating delamination, an observation that is in agreement with the hydrated cation diffusion-limited delamination mechanisms at open circuit for specific coating systems postulated by Leidheiser and his co-workers(155,158,100), Sharman, Sykes and Handyside(159), and Nguyen and his co-workers(163-167).

It may noted from the derived linear relationship of Figure 8 that no delamination should take place if the cations present in solution have an absolute ionic mobility of or below approximately 1.6 x 108m2s-1V-1. An explanation for this observation may be made by consideration of the important factors responsible for maintenance of a high pH at the disbondment front. In the corroding underfilm system, the pH at the disbondment front will be controlled by the rate at which cations are transported to the cathode site and their maximum attainable pH, and the rate at which hydroxyl ions are transported away from it. From the pH / delamination rate constant data recorded for non-corroding non-chromated FBE-Panel systems it may be seen that below a given pH, delamination will not take place. It may be postulated from Figure 8 that this pH is also that which would be generated at the delamination front by a cation of absolute mobility 1.6 x 108m2s-1V-1. At cationic mobilities at or below this value the flux of hydroxyl ions away from the disbonding front would be too great for a sufficiently high pH to be generated to cause delamination.

When the delamination rate constants created with the relevant cations in the corrosion-inducing solutions are plotted on the best-line fit generated by delamination rates of FBE-coated panels in solutions of different pH (Figure 6), the corresponding pHs at the delamination front may be estimated (Figure 9). These pHs are based on the assumption that pH at the delamination front of passive panels is essentially the same as that of the bulk solution, and that delamination of an FBE coating from a corroding substrate is predominantly due to hydroxyl ions. The estimated pHs are shown in Table 3 below:

Cation Estimated pH Hydrated Cation Mobility, U (x 108m2s-1V-1)(89) pH at Saturation(89)
K+ 14.4 6.6 15.28
Na+ 14.0 4.6 15.02
Li+ 13.4 3.4 14.72

Table 3: Solution pHs at the Delamination Front for Different Cations in Corroding Non-Chromated FBE-Panels Estimated from pH-Induced Delamination Data It is noted that these pHs are less than those attainable by the cation hydroxides at saturation. The rate-controlling step in the delamination process of corroding non-chromated FBE-coated panels is therefore the passage of the hydrated cations through the holiday and along the coating/substrate interface to the cathodic site. The cations are being transported to the cathode such that charge balance may be achieved.

From the radial delamination / square root of time graphs for the FBE-panels it may be seen that there is a delay time before delamination begins to occur, as found by other researchers including Leidheiser and co-workers(158,160) and Sharman, Sykes and Handyside(159). Delay time appears to be independent of both hydrated cation mobility and solution pH (Figures 10 and 11).

Leidheiser, Wang and Igetoft(158) linked delay time to ion ingress into the coating, and Wang (173) to water diffusion through it. Examination of the FBE-panels exposed to different solutions at open circuit reveals that there is no obvious relationship between the delay times of the FBE-coated panels and cation transport through the coating.

The differing delay times of the corroding systems cannot be explained in terms of an interfacial attack process from the holiday at the coating / substrate interface, either. If this was the important mechanism in control of delay time, then delay times should again be very similar because all the samples started to corrode after approximately the same period of immersion. Delay time must therefore be related to a transport process through the FBE-coating. There is no evidence that it is related to cationic transport, so this delay time is probably a function of either water or oxygen transport. Mayne(82) has shown that oxygen transport through a coating is a relatively rapid process, and Cottis(228) has stated that a typical oxygen permeability of an epoxy resin is 5.2 x 10-12 mole.m-1s-1atm-1, so it is unlikely that this factor could account for the length of delay time recorded in these conditions (typically 3-4 days½). Oxygen transport as a factor in delay time would also not explain the differing delay times in the systems investigated. It is therefore likely that the delay time of the corroding system at open circuit is related to water transport through the coating.

Investigation of FBE-panels in contact with the test solutions where no corrosion took place reveals that delay time cannot be related to bulk solution pH or cationic transport. Examination of their delay times does however reveal that they are of the same order (between 3.1 and 5.2 days½). It is therefore possible that delay times of these near-passive systems are related to transport through the coating. With no extensive electrochemical reactions taking place, it is unlikely that cationic or oxygen transport through the coating will have a significant effect on time before initiation of coating delamination. It is possible however that water transport is the important factor.

Another interesting trend that may be derived from the delay time data is that corroding systems tend to have lower delay times than those displaying passive properties. Parks and Leidheiser(100) have shown how ionic migration through an organic coating increases in rate when the substrate is placed under polarisation at –0.80V, and they have referred to other work conducted by themselves showing a similar trend for water. A similar process may be occurring with those systems undergoing corrosion, with the polarised substrate causing an increase in water flux and hence increasing the rate of water transport through the coating, therefore shortening the delay time of the system.

Prior to the measured propagation period of the delamination process, it is possible that delamination was occurring at a rate that was sufficiently small that it was not visibly discernible on inspection of the clear FBE-coated panels, the rate being limited by the high degree of bonding between the FBE-coating and its substrate in the interfacial and interphasial region. When the water diffusing through the coating reached the coating / substrate interphasial region, it will have disrupted the interactions between the two materials, as Funke and other workers have shown(116). With this disruption, delamination may then be able to take place at an accelerated rate. This delay time ‘initiation’ mechanism would explain the traces of disbonding sometimes found prior to the extrapolated delay time by other authors such as Sharman, Sykes and Handyside(159), and also their finding that delay time was only marginally influenced by coating thickness, the more important factor being uptake of water in the region around the defect.

The possibility is therefore raised that in both active and passive systems, delay time is controlled primarily by water diffusion, which is a function of the potential gradient across the coating, and secondarily by the nature of the coating / substrate interface. In these conditions, increasing thickness of coating would be expected to result in increasing delay time, as has been found by Wang and Leidheiser(160) and Sharman, Sykes and Handyside(159).

Visual studies of passive and corroding delaminating FBE-coated systems at open circuit reveals an area of bright untarnished substrate at the disbonding front. A closer examination of these regions by SEM reveals that they are the same in morphology as the substrate prior to delamination (Figures 12 and 13). As such it is unlikely that delamination of the coating is directly caused by any anodic reactions taking place. It is therefore likely that FBE coating delamination will be caused by a variable that causes degradation of the steel substrate at the disbondment front. No FBE visible to SEM is left on the delaminated substrate surface.

The FBE-resin is already intimately mixed with its hardener in its powder form prior to spraying onto the heated workpiece (i.e., rebar or panel), and its gel time occurs in a matter at seconds(215). There is therefore no time for a separation of resin and hardener to take place, as has been suggested for solvent-based paint systems(106). The FBE coating should therefore effectively be the same in chemical and structural composition from the outer limit of the polymer / substrate interphasial region to its external surface. In this situation, if the mechanism of adhesion loss was coating degradation and cohesive failure, then it would be equally likely to occur at any depth of the FBE coating on the passive specimens at open circuit. Since failure always appeared to occur at the coating / substrate interface, this mechanism was clearly not the cause of disbondment.

Conclusions

The following conclusions may be drawn from this investigation into delamination of FBE-coatings in a simulated pore solution environment:

1) Delay time before initiation of observable delamination processes may be a function of water penetration through the coating to the interfacial or interphasial coating/substrate region.

2) Delamination of FBE coatings from steel substrates is predominantly caused by hydroxyl ions.

3) Rate of FBE delamination is controlled by transport processes from a pore in the coating and along the delaminated coating/substrate interface to the disbondment front.

4) The locus of failure of coating adhesion is in the interfacial or interphasial coating/substrate region.

5) The rate of FBE delamination in near-passive conditions is controlled by hydroxyl ion migration from the bulk external solution to the coating/substrate disbondment front.

6) The rate of FBE delamination in the condition of underfilm corrosion is controlled by hydrated cation movement to the cathode site.


Send Mail to the

Journal of Corrosion Science & Engineering Home Page
Corrosion Information Server
Centre for Electrochemical Science and Engineering, University of Virginia (JCSE Mirror Site)