David Greenfield and David Scantlebury
Corrosion and Protection Centre, UMIST, PO Box 88 Manchester M60 1QD, UK
This review considers the past fifty one years of study in the field of corrosion prevention by organic coatings. It concentrates mainly on the mechanisms associated with the paint binder. Areas covered include the importance of the coating ionic resistance and the concept of D and I areas. The effect of the environment on the ionic resistance is examined. Blistering and delamination is considered in the light of movement of ions through coatings and along interfaces. Specific attention is paid to the metal paint interface with emphasis on adhesion, underfilm contamination and surface tolerant coatings.
Keywords: anti-corrosion paints, ionic transport, blistering, cathodic disbonding, underfilm contamination, surface tolerant coatings
In this review, we are concerned with the effects of the polymeric binder and the anti-corrosion properties of paints. Aspects of corrosion inhibition by anti-corrosion pigments have not been emphasized. Also within the field of underfilm corrosion and blistering, we have not considered the the highly important and topical area of filiform corrosion.
Any discussion which considers the prevention of aqueous metallic corrosion, usually takes as its starting point the electrochemical model of the corrosion of mild steel in a neutral electrolyte with the four important processes involved linked in series namely; the anodic reaction, the cathodic reaction and the conductive pathways for ions and electrons. Inhibition of the anodic or cathodic reactions or an interruption of the ionic flow will cause the rate of corrosion to be significantly reduced. Thus, all corrosion prevention measures are aimed at removing or suppressing one of the three elements. One of the most convenient, and certainly the oldest method of protecting a substrate from the detrimental effects of the environment, is to coat it with a polymeric barrier paint to isolate it from its surroundings., This barrier might be due purely to the properties of the polymer, or to the inclusion of inert pigments that act to increase the length of the diffusion path through the coating.
The inclusion of inert pigments into the film can increase the effectiveness of the coating providing the formulation is not too heavily loaded. Too high a concentration of pigment is liable to reduce the effectiveness of the coating. The Critical Pigment Volume Concentration has been shown to be a transition point where a number of significant features of the coating change., The loading of a coating with pigment particles is made more complex by the effect of the distribution of pigment particle size. A wide range of pigment particle sizes favours a high Pigment Volume Concentration. David � is PVC the same as CPVC � define it!
The nature of this barrier has been the subject of much work over many years by workers in the field of protective organic coatings. The original assumption was that organic coatings act as a barrier to water and oxygen from the environment. Scientific investigations of the subject over the last half century have determined that the limiting factor in the protective mechanisms of barrier coatings is frequently their resistance to the flow of ionic current.
Figure 1 : Schematic of resistance behaviour of coatings in immersed metals after ref 
The importance of the electrolytic resistance of organic coatings has long been appreciated. Bacon, Smith and Rugg, after measuring the resistances of over 300 coating systems, determined a direct correlation between these resistances and the ability of the coating to protect the underlying steel from corrosion. Three general classifications, based on a sustained resistance value, were established during this investigation: good, fair and poor - see Figure 1. The classification that a coating was allocated was dependant upon its long term resistance value upon immersion in solution �Most coatings, if continuous, have resistances in the neighbourhood of Log R = 9 during the first 5 to 10 minutes of immersion, followed by a decrease in resistance which may vary considerably in steepness and duration�, the units of R were Ohms for one sq. cm. Their subsequent behaviour determined which category they should be placed into. All coatings were found to exhibit an initial decrease in resistance, which varied in terms of rate and duration. For a good coating, this initial decrease was followed by an abrupt recovery to around the original value. The subsequent resistance of the coating remained either essentially constant or fluctuated within the high resistance region. A coating designated as fair either levelled off or displayed a slight increase in resistance. However, a subsequent sharp drop in resistance was a pre-cursor to coating failure, which occurred within six months of the exposure period. The resistance of a poor coating continued to decrease resulting in failure within sixty days. A coating that maintained a resistance of 108 Ohm cm2 provided good corrosion protection while one whose resistance fell below 106 Ohm cm2 did not.
Although this work identified the relationship between the ionic resistance of the coatings and their ability to protect, the permeability of other corrosive agents other than ionic species was not considered. Therefore, the results were presented as providing a measure of the likely protection that could be expected from a particular system rather than proposing a mechanism for the protective action of organic coatings.
A further important result from the work of Bacon, Smith and Rugg identified the effect of the environment on the protective merit of coated submerged steel. One of the variables taken into consideration was the level of dissolved salts in the test solution. It was found that the resistance of the coating fell with increasing salt concentration and from this result it was concluded that the dissolution of environmental water into the coating was more important than the uptake of salt from the solution by the coating. That the level of dissolved salts of the exposure environment had an effect on the properties of coatings had been reported previously by Kittelberger and Elm . This work showed a relationship between the level of dissolved salts in the solution, expressed as osmotic pressure, and the water uptake of the films under investigation, expressed as percentage weight gain. It was shown that the nature of the solute did not affect the results; rather it was the osmotic pressure that determined the level of imbibition is this a word? of water into the coating. When the coatings were exposed to a salt solution, after an initial high rate of water absorption, the curve levelled off and eventually an equilibrium was attained. Where the films were exposed to distilled water, however, no equilibrium was reached and the film continued to gain in weight. It was concluded that the degree of water absorption was a function of the activity of the water in solution, due to the differences in osmotic pressure between the exposure solution and that within the film, see Figure 2.
Figure 2: The relationship between osmotic pressure and water uptake after ref 
In order that a paint coating should offer protection from corrosion, it should interact with one or other of the processes involved in the overall series of corrosion reactions. This was considered by Mayne, who determined that the permeability of organic coatings to both water and oxygen is so high that the rate at which they arrive at the interface of the coating and the steel is greater than that required for corrosion to proceed and so could not be rate determining. Therefore, Mayne reasoned that the method by which an unpigmented coating protects steel from corrosion was by the introduction of a path of high electrolytic resistance in between the metal and the environment; the term resistance inhibition was coined for this mechanism.
Further work by Mayne  showed that, upon immersion in water or an aqueous solution of electrolyte, most organic coatings acquire a negative charge. According to Mayne, the acquisition of this charge has the effect of creating a selectively permeable membrane, which is preferentially permeable to cations; that is, a film that has gained a negative charge due to immersion may be regarded as a large poly-anion. An alternative view is expressed by Corti et al , who claim that a charged film would slow down the passage of counter ions (those having the opposite charge) whilst having no effect on the movement of co-ions through the film.
These contradictory views stem from the proposed mode of transport of ions across the film. According to Corti, the rate of permeation across the film is affected by the presence of small imperfections or pores, which extend through the film and have cross sections distinctly larger than the free areas normally present between the atomic groups in the membrane matrix. Mayne�s model, on the other hand, infers that the passage of ions is through the bulk matrix of the film. If the ionic conduction takes place through pores, it is reasonable to claim that a negative charge on the polymer would impede the passage of cations due to electrostatic attraction between the ions and the pore walls. However, if the movement of ions through the coating is considered to be through the bulk of the film, the proposal that a negative charge on the film would act as a repulsive force towards anions seems plausible.
Transport of cations through polymer films has been observed elsewhere with the passage of iron ions through a range of coatings on steel immersed in NaCl. These coatings were continuous and free from defects and the ionic transport was considered to have taken place through the bulk polymer. Of the coatings tested, a �practical anticorrosive coal tar epoxy� was found to retard the process. The inference of this work is that the conduction of ions through the film takes place through the bulk polymer. The work reviewed in the following section proposed mechanisms to account for this.
Ionic conduction through the bulk of the polymer matrix was studied by Maitland and Mayne. This work, which mainly considered the properties of an unattached, unpigmented pentaerythritol alkyd varnish, found that the resistance of the film was affected by the state of the exposure solution and identified two distinct processes, which took place in succession of each other: The Fast Change and The Slow Change. The Fast Change was observed to take place within a few minutes of immersion where the films attained a steady resistance value, Ro. This was followed by The Slow Change, which took place over a number of weeks or months.
The Fast Change was shown to be a reversible process and was independent of both the nature of the solute in the exposure solution and the state of the film in respect of the slow change. The controlling factor in the process was seen to be the osmotic pressure or water activity of the solution; this process is shown in Figure 3
Figure 3: The Fast Change; Comparison of the effect of 3.5N KCl with isotonic sucrose solution after ref. 15
The Slow Change was considered to be controlled by the concentration of the electrolyte in solution. This was confirmed by experiment, whereby the reaction was accelerated by an increase in temperature.
Maitland associated this phenomenon with an ion exchange process where cations from the electrolyte were exchanged with hydrogen ions from carboxyl groups within the polymer. The ion exchange behaviour was found to be affected by the pH of the system or the concentration of potassium ions in solution. An increase in either the pH or the potassium concentration resulted in a greater rate of ion exchange, causing a fall in the resistance of the coating.
Work on the effect of the Slow Change on the reversibility of the Fast Change was carried out by Cherry and Mayne,, who found that the coatings under investigation altered their mode of conduction at a given value of pH, the value of which varied according to the particular coating. Working on the assumption that the slow change was an ion exchange process, it was postulated that the resistance of the film should be a function of the ratio of the activities of the metal cation from the electrolyte (in this case potassium) and that of the hydrogen ions. The term rK was defined, where:
Films were conditioned in solutions of various rK values and then the temperature coefficient of resistance for each film was measured. It was found that there was a small range of rK values, specific to each polymer, where the mode of conduction ceased to run counter to that of the solution and commenced to follow it. The point where this change in the conductive behaviour of the film took place was named the reversal point and was seen to indicate a change in the mechanism of ion transport from one of activated diffusion to that of aqueous conduction. This conduction was considered to take place through virtual pores.
The model whereby the electrolytic conduction takes place through pores is supported by Miskovic et al who, as a result of their AC impedance examination of electrodeposited epoxy, proposed a model whereby electrolytic conduction through a coating is dependant only on conduction through macro-pores. These pores were examined microscopically and were determined to have radii between 3 and 7.2�m.
The question of the possibility of conduction through pores was addressed by Kinsella. This work identified different modes of ionic conduction through polymer films. Two distinct forms of conduction were identified. The first was I, or inverse type conduction, where the resistance of the film coincided with that of the solution; this type of conduction was in line with that identified by Bacon, Smith and Rugg .
The second mode of through-film transport that was identified was named D, or direct type conduction where the film�s resistance ran counter to that of the solution�s. Initially, the films that exhibited D type conduction were thought to be due to coating deficiencies such as dust wicks passing across the film. However, investigation of the temperature coefficients of resistance for these films showed them to be higher than that of aqueous conduction. In addition, the films were shown to be preferentially cation permeable and thus considered to be continuous.
Figure 4: Schematic plot of I and D type conduction, after ref 
A series of papers were produced as a result of this work, which addressed various different factors affecting ionic conduction in polymer films. The first, examined the influence of the electrolyte on the resistance of films. It was found that the films investigated followed the general trend exhibited in Figure 4. However, when an I Type film was exposed to a solution of very low electrolyte concentration, and therefore a high water activity, the film displayed D Type characteristics over a small range of concentrations, at the lower end of the scale. Once the electrolyte concentration reached a critical level, the film began to exhibit more typical I Type characteristics. This led to the oft quoted conclusion that the difference between D and I Type conduction was �more one of degree rather than of kind�.
Scantlebury examined the distribution of D and I Type areas across a given film. It was found that the films investigated were all very inhomogeneous and had a distribution of both D and I Type areas over the film. Two conclusions, relating to the mechanisms responsible for the different modes of conduction were drawn from this investigation. Firstly, that D Type conduction could not be attributed to the presence of pores unless they were of molecular dimensions.
The second conclusion that was drawn from this study, was that the resistance of both D and I Type areas bore a relationship to their hardness values, as determined via microhardness tests. D Type areas were found overall to have a lower hardness value than that obtained for I Type. The hardness of the film was equated to its crosslinking density. This result was reinforced by swelling experiments, which showed that D Type areas swelled to a greater degree than I Type. Therefore, it was concluded that D Type conduction was a function of lower crosslink density in the film.
This conclusion reinforced Kinsella�s view  that the initial drying conditions had a marked effect on the final properties of the film. Other work on crosslink densities of polymer films has shown that the crosslink density appears only to affect the ionic conductivity of the film. The study was concerned with the effect of crosslink density upon water permeation through films and concluded that the permeation of water was not noticeably affected by alteration of the crosslink density. The final two papers in the series by Mayne and co-workers considered the influence of temperature and finally the effect on the barrier properties of a coating due to the inclusion of inert pigments.
Evidence that D Type areas are detrimental to the protective properties of organic coatings was produced by Mills,  who showed a direct relationship between the presence of D Type areas in the film and the occurrence of underfilm corrosion. With regard to the term �virtual pores� put forward by Cherry17 to explain the mode of conduction in an apparently continuous film, Mills suggests the term �conductive polymer phase� as more appropriate. This term is intended to describe more accurately the apparent properties of D Type films.
Mills further observed that the proportion of D Type areas was a function of the thickness of the film. It was found that the incidence of D Type areas decreased significantly once the film�s thickness was increased to around 75μm, this finding was used to conclude that D Type areas were spherical and approximately 75μm diameter.
Amongst the most common forms of failure found in organic coatings are those of blistering and delamination. Much debate has ensued as a result of the research into these failure modes and a number of governing mechanisms have been put forward. The mechanisms involved in the two modes of failure are similar, but it is unclear whether they are the same phenomena.
Within each of these failure modes, there are sub-classifications, which may or may not be important for a particular system. The components of a system are the substrate, the coating and the exposure environment. Factors affecting the performance of a system include surface preparation, coating application, cure regime and film integrity.
The blistering mechanism of paint coatings upon exposure to aqueous environments was addressed by Gay, who found that all the systems examined displayed four common features:
In the majority of cases, before any blistering occurred, some corrosion was observed at weak points in the coating and no substantial amount of blistering was found where corrosion was not present. The suggested sequence of events leading to the formation of blisters is as follows.
The principle of endosmotic transfer of water through the coating was supported by Mayne  who concluded that the contribution of osmotic transfer of water was much smaller; 6% in the case of the coatings studied.
Figure 5: Stages in the development of an osmotic blister.
Subsequent work on blistering has produced three individual classifications
This form of failure is one consequence of a contaminated substrate., , . Soluble salts at the interface can form a concentrated salt solution that, due to decreased water activity, acts to draw water through the coating, which behaves as a semi-permeable membrane, from the exposure environment. The osmotic pressures exerted can be expected to range between 2500 and 3000 kPa compared with a mechanical resistance to deformation of a coating ranging from 6 to 40 kPa In addition, the pressure within the blister becomes greater than the atmosphere on the free side. So, provided the film is not ruptured in the process, the conditions exist for a continually expanding blister.
Figure 5 schematically represents the stages in the formation of an osmotic blister. In Figure 5A a layer of interfacial water in contact with soluble salts is shown, in B the salts at the interface have formed a concentrated solution and finally C represents the stage when the osmotic pressure, produced by the concentration gradient across the coating, has drawn water through from the outside environment and formed the blister. The foregoing is attractive due to its simplicity; however, it has been shown29 that electroosmosis plays a predominant role in this form of blistering with osmosis only contributing to about 6%.
Gowers and Scantlebury were able to measure currents flowing between different areas of the substrate under artificial blisters. They found that contaminating the surface of the substrate with NaCl promoted the corrosion process, the current flow and the blister formation.
One suggestion that has been made to reduce the tendency for a coating to suffer osmotic blistering hinges on the formulation of the film. In a solvent borne coating, it is normal to find that a combination of different solvents is employed. According to Storfer and Yuhas, hydrophilic solvents can facilitate and accelerate several of the generally recognized blister-formation mechanisms, particularly osmotic blistering. An inverse relationship was determined between the solubility of water in the last solvent in the formulation to evaporate (the tail solvent) and the blister resistance of the coating. Hydrophilic solvents are more prone to cause blistering.
This mode of failure was addressed by Koehler, who considered liquid filled blisters to be anodic in nature. Experiments were conducted an on epoxy phenolic resin exposed to 0.01N calcium chloride. Calcium was chosen as the cation because it does not produce an alkaline environment. No deliberate faults or holidays were introduced into the coating. The samples were anodically polarised and blisters appeared after seven days. The substrate in the centre of the blister was dished and corroded while the periphery was bright, indicating cathodic detachment. The explanation of these results was that during the test, chloride ions passed through the coating at thin points and created an acid environment under the coating within the blister. The cathodic reaction was postulated as being the reduction of the ferric oxide on the steel to the soluble ferrous state. This type of failure requires a low pH environment to proceed. Such low values may be found in many canned foods and drinks.
It is to this form of blistering that the majority of the literature is directed. Cathodic blistering is the result of an alkaline environment under the coating caused by the cathodic reaction, associated with corrosion that occurs at a damaged site of the film., ,  Early work by Kittelberger  identified a relationship between bare areas of steel and the occurrence of blistering on immersed panels; it was shown that an exposed area of of the coated specimen was enough to exert a significant influence on the blistering of the coating. Figure 6 shows the proposed failure mechanisms, which may occur when a film containing a fault is exposed to a corrosive medium.
The fault may take the form of mechanical damage to the coating or may be an inherent fault of the coating i.e. pores or holidays. The primary pre-requisite for this form of failure is that the substrate should support a cathodic reaction. In the case of a neutral or alkaline environment the cathodic reaction would be the reduction of oxygen. Tests with cathodically polarised steel coated with polybutadiene, exposed to a NaCl solution, developed blisters containing a solution of high pH after 7 days, which coalesced in 18 days; no intentional damage was made to the coatings in this experiment.
Figure 6: Possible consequences of a damaged coating.
As no defects were introduced into the coating material, the question of the route taken by the reactants is raised. Mayne12 showed that the transmission of water and oxygen through the film was in excess of that needed for corrosion and attributed the protective mechanism to the film�s high ionic resistance. If we accept this hypothesis, some pathway must exist through the film to allow the sodium ions to the interface in order to produce the alkaline environment. These pathways could be due to pores. However, an alternative theory is proposed by Leidheiser, who in a cautionary technical note suggests that above a given concentration, alkali cations may have a deleterious effect on the coating, which leads to morphological changes hence introducing conductive pathways to the interface.
Akin to cathodic blistering, cathodic delamination is also the result of alkalinity at the interface. Again, this alkalinity is the result of cathodic activity under the coating. It is associated with faults, either inherent or induced, in the coating. Cathodic polarisation may be a consequence of either corrosion at the point of damage or the application of cathodic protection. Resulting from experiments carried out by Smith and Dickie on primer failure, it has been shown that under impressed cathodic conditions, corrosion inhibitive pigments play no part in the reduction of disbonding. Under such conditions, the performance of the system reflects the resistance of the primer resin system to alkali displacement. During salt spray exposure, it was found that the corrosion inhibitive pigments did exhibit a degree of control of the rate of failure by inhibiting the anodic reaction.
A number of explanations have been put forward whereby the alkaline environment under the film affects the integrity of the metal - polymer interface, or perhaps more properly the interface between the oxide and the polymer. Koehler, showed that this form of failure, which he called �halo detachment�, only occurs when there are alkali metal cations available in the environment to act as counter ions to the cathodically generated OH-.
One explanation is that dissolution of the air-formed oxide layer is responsible for the loss of adhesion. Various techniques have been adopted to confirm this view. Ritter carried out ellipsometric studies on a range of coatings in tandem with an investigation of uncoated steel exposed to saturated NaOH. The coatings studied were collodion, polystyrene and two proprietary acrylics, all of which were cured at room temperature. Additionally, a heat-cured acrylic was examined. The argument put forward was that delamination is a direct result of oxide dissolution and this is supported by the presence of iron ions in the interfacial region of the polymer in the heat cured specimens. The oxide layer present on the metal under these circumstances, is thicker than the normal air-formed layer, as a result of the curing regime.
Further work carried out using this technique concluded that the high pH generated under the coating was conducive to oxide growth and surface roughening of the substrate. This conclusion was the result of tests carried out on uncoated substrates in a high pH environment. Taken in conjunction with the findings cited in the previous paragraph, the inference of this could be that the thicker oxide layer, developed due to the high pH generated under the film, could cause the oxide film to become mechanically weaker and result in a cohesive failure.
Castle and Watts, in an XPS investigation of cathodic delamination, concluded that dissolution of the oxide layer was not a significant factor in the process. It was found that the oxide within the disbonded region was reduced only in localised patches and the disbonded area extended well beyond this.
In a similar vein, Wiggle et al considered different pre-treatments and the effect upon them of cathodically produced hydroxide. Both zinc and iron phosphate pre-treatments along with bare metal was included in the study. Initial tests were carried out to determine the porosity of the conversion coatings, which yielded results of 0.2% for the zinc and 23% for the iron phosphate; these porosity values were linked to the poor resistance of the iron phosphate to anodic undermining. Experiments were carried out under both anodic and cathodic polarisation as well as at open circuit. Cathodic polarisation trials resulted in dissolution of the zinc phosphate and hence loss of adhesion of the coating. Knaster and Parks  also identified dissolution of zinc phosphate under paint coatings on pre-treated steel but considered this to be a secondary reaction. The contention in this case is that delamination of coatings from phosphated panels, due to a defect, is caused by oxygen depolarisation along the phosphate-paint interface and hydrogen evolution at the delamination front that results in a force, which disrupts the adhesion of the polymer to the phosphate layer.
Dissolution of phosphate conversion coatings on steel was also examined by Sommer and Leidheiser. The variables they considered were the effect of the pH and also the alkali cation involved. The pH effects were investigated by exposing test panels to 0.01, 0.1 and 1.0-M solutions of NaOH for ten minute intervals up to one hour. The importance of the cation involved was determined with 0.1-M solutions of hydroxides of Na, Cs, K and Li. It was found that increasing pH values from 11.5 to 13.5 resulted in a higher rate of dissolution of the phosphate ions from the conversion coating.
With regard to the cation type, the investigators were surprised to find that Na was the most effective of the cations in the dissolution of the coating (1.6, 2.6, & 4.1 times K, Cs & Li respectively). Additionally, the cation type not only affected the rate but also the mode of dissolution. With Na, Cs and K, the phosphate ions were dissolved more rapidly than the zinc ions, whereas in the case of Li, the zinc was the one to be dissolved at a higher rate.
Dissolution of the conversion layer is also suggested by Dickie, after an investigation of the failure of an epoxy coating. This mechanism is proposed to account for the slower delamination and longer delay time observed with alkali resistant coatings.
An alternative view to the dissolution of the substrate theory is that the alkaline solution under the film attacks the coating itself. Koehler�s investigation  of oleoresinous and polybutadiene coatings on steel concluded that delamination was due to saponification of the polymers. In his discussion, he states that disbonding may occur at pH 11.7, which is well below that required to dissolve the oxide layer. That higher pH�s may form under the coating at cathodic sites is not disputed, however it is suggested that this may take place subsequent to the loss of coating adhesion. Furthermore, it is proposed that regardless of any other phenomena a film of water is required under the coating for disbonding to proceed.
Attack on the polymer is also a mechanism supported by Hammond et al  who examined three different epoxy systems on steel with a coating range of 10 - 30 �m. In their examination of the interfacial composition of delaminated surfaces, using XPS, predominant traces of polymer were in evidence on the metal surface relative to iron oxide; this was in conjunction with little or no traces of iron in the interfacial polymer. Their conclusion from this was that the failure mechanism was due to resin degradation and that the locus of failure was within the polymer, resulting in a failure that was cohesive in nature.
Another XPS investigation was undertaken by Castle and Watts , who looked at fusion-bonded epoxy coated mild steel. The results of this study suggested that the failure was essentially adhesive in nature. One of the variables in the experiments carried out was surface texture, ranging from grit blasted (Ra 3.80 �m) to a mirror finish polish with 1�m diamond paste (Ra 0.15 �m). It was found that as the Ra values increased, there was a slight tendency toward cohesive failure of the oxide layer. However, although some iron was transferred to the polymer, the amount was a trivial component of the locus of failure and was consistent with cohesive failure of the oxide covering at protruding asperities.
The nature of the delamination mechanism is not the only area of debate within the overall subject of cathodic delamination. Whilst there is agreement as to the species involved in the process, it is still open to conjecture how those elements arrive at the delamination front. Considering a coated steel substrate, immersed in an electrolyte of neutral or near neutral pH, the half-cell reaction responsible for the delamination process is generally agreed to be oxygen reduction. This reaction generates OH- at the cathodic site and is responsible for the alkaline environment at the delamination front. The elements required for the process to proceed are water, oxygen and free electrons. The electrons may be generated by either an anodic reaction or through the application of cathodic protection. Additionally, in order to maintain electroneutrality, some form of counter ion is required; the high pH values recorded at the reaction zone indicate that these counter-ions are alkali cations rather than protons.
If we assume that cathodic delamination is a result of a damaged coating, there are two possible routes that the reactants for the cathodic reaction may take. The two alternatives are either through the coating or along the polymer-metal interface. An extensive review of the delamination process was carried out by Leidheiser et al . The results of the experiments carried out indicated that water was transported to the reaction zone through the coating. It was suggested that a certain fraction of this could be in the form of a cation such as H3O+, as the cathodic nature of the front may favour the transmission of water associated with an ion possessing a positive charge. The supply of oxygen to the cathodic site was found to be largely through the coating, with a small contribution from interfacial transport, in the case of the epoxy coating studied. Other work on the subject of the supply of reactants to the delamination front with pigmented coatings, suggests that supply of oxygen along the interface may be the rate-controlling factor.
The major unresolved factor in the overall picture is the route taken by the charge-neutralising cations. Leidheiser�s study identified a linear relationship between the diffusion coefficient in aqueous solutions of the different cations, used in his experiments, and the delamination rate. This suggested that the cations were supplied along the interface. However, as delamination rates showed a strong dependence on film thickness, through film migration of cations could not be discounted.
An important feature identified in Leidheiser�s work was an incubation time needed before delamination commenced. This was dubbed the �delay time� and the factor controlling this parameter was considered to be cation ingress into the coating, due to an observed correlation between ionic mobility and the delay time.
Interfacial migration of cations is supported by the work of Castle and Watts55. A number of factorsdrew them to this conclusion. Firstly, EPMA examination of a cross section of the polymer showed Na ingress into the coating from both the interface and the bulk solution; however there was no Na detected in the bulk of the polymer, suggesting that Na+ did not fully penetrate the film. Also, the disbondment velocity (Dk) was related to the surface profile of the substrate and the parameter referred to as tortuosity (τ) as follows:
A tortuous interface is one whereby the route that must be taken by a species in order to diffuse from one point to another, is made more difficult by roughening the substrate, thereby making the effective path longer. It follows from this that a rougher surface should result in a lower lateral degree of delamination, if the controlling mechanism is the interfacial migration of a species.
Castle and Watts� results for a steel/epoxy system produced a constant disbondment velocity (Dk*) of 0.4 � 0.08 mm day-1. This again was felt to point towards interfacial migration as being important to the process.
The effect of the substrate upon the delamination process was discussed by Skar and Steinsmo  in an investigation of the importance of through film ionic migration. Their experiment, which aimed to clarify the confusion as to the route taken by the cationic species, showed a strong linear relationship between the dry film thickness and the disbonding rate. The explanation offered for this result considered three alternative routes by which the counter-ions may reach the delamination front:
These alternative routes are illustrated in Figure 7. The first option was discounted due to the strong dependence on film thickness. Of the remaining two alternatives, it was felt that the second was the most likely.
Figure 7: Suggested ionic migration routes from ref. 
The logic behind this conclusion stems firstly, from work carried out by Mayne and Mills,25 whose investigation of detached films showed a notable difference between the ionic resistances of attached, compared to detached, films. Secondly, it was proposed that the conduction mechanism of the detached film exhibited �D� type behaviour, as defined by Kinsella, due to the high pH under the coating. Whilst the second proposition may well be valid, the first is dubious. The conclusions drawn by Mayne and Mills, were indeed that detached films showed lower resistances, as quoted, but not on ordinary mild steel substrates - where the detached films showed similar resistances to those attached to the steel. It was when the coatings were applied to inert substrates (platinum and steel passivated with a zinc chromate pigment) that the quoted difference in resistance was found.
Working with both pigmented and unpigmented chlorinated rubber coatings, Sharman  identified a number of features worthy of note. Referring to the delay time defined by Leidheiser, he concluded for the system under investigation that it was essentially independent of cation mobility. Furthermore, it was found that the iron oxide pigmented systems showed longer delay times than the clear varnish coated substrates; this was unexpected as the former had higher ionic permeabilities. This situation was considered by pre-soaking the pigmented coatings in water prior to testing; this resulted in a reduction in the delay times from 5-7 hours to 1-2 hours. This result pointed to ingress of either water or oxygen as the factor responsible for the duration of the delay time.
With regard to the species responsible for the delamination process and the route taken, different mechanisms were seen to be responsible for the two systems. For the unpigmented coating, interfacial cation migration was considered to be the determining factor as the delamination rate constant increased in a linear fashion with cation mobility. With the pigmented coating, which displayed a greater through-film ionic transport rate, coupled with lower oxygen permeability, interfacial migration of oxygen was suggested as a likely rate determining mechanism; this was suggested to be virtually independent of cation mobility. A further result, which reinforced the hypothesis of interfacial migration of a species as the rate-determining factor, was that the rate of disbonding eventually became zero. The explanation for this was that the interfacial pathways became blocked with corrosion products.
Armstrong and Johnson, working with unpigmented chlorinated rubber coatings subjected to cathodic protection, concluded that the corrosion current passed through faults in the form of cracks or crevices in the film. It is proposed that degradation of the film due to OH- was essentially mechanical in nature. This argument is supported by a number of factors. Free standing films exhibited conductive behaviour only after long exposure to concentrated OH-; this was lost after the films were dried in a dessicator. It was felt this result supported the hypothesis of conduction through cracks or crevices rather than through the bulk film. FTIR (Fourier Transform Infra Red) spectra of membranes exposed to OH- displayed no detectable chemical change from the untreated film.
The underside of the films (that in contact with the metal), which was subjected to the alkaline conditions, was found to contain holes that were not visible on the upper face of the coating. This is thought to be due to pockets of high solvent or plasticiser concentration; it is suggested that the OH- opens up such defects by a solvent cracking process, which reinforces the assertion that the breakdown is mechanical rather than chemical.
In an ideal world, the coatings that we applied to protect against corrosion would be a perfectly homogeneous mix and be free of defects of any kind. Also, the surface to which the coatings were applied would be in an ideal condition, free of any form of contamination and with an appropriate finish.
This utopian situation does not occur in the real world and the substrates are commonly far from pristine. There are occasions where adequate preparation of the surface to be coated is either technically unfeasible or prohibitively expensive. In such circumstances, there is a need for a coating that will perform well on a contaminated substrate. Mayne�s12 model of underfilm corrosion, considered earlier, was applicable to a clean surface. In the presence of contaminants on the surface, resistance inhibition does not come into play, as all the required elements for corrosion to proceed are present at the interface.
Surface contamination is conventionally considered to be due to such factors as salt deposits or corrosion. Indeed, this is the source of a great deal of concern and has been the basis of much work,    . Additionally, it has been demonstrated that, even on substrates cleaned to laboratory standards, minor inclusions such as sulphides can have a detrimental effect on the performance of an organic coating.
An established method for dealing with the problem of painting over steel that is rusted, contaminated with salts or, as is usually the case, a combination of the two, has been to coat with an oil-based coating pigmented with red lead (Pb3O4). Environmental concerns have limited the use of lead compounds and forced formulators to look to alternative means of addressing the problem. In the pursuit of a viable alternative, the red lead formulation is still used as a benchmark by which to compare the various contenders , , . The exact mechanisms by which red lead protect a compromised substrate are unclear, but its efficiency is undisputed.
Rust, in itself, is not seen as a problem; rather it is the contaminants that the rust contains. Mayne  identified that sulphates were embedded into adherent rust close to the metal surface; this rust layer was not easily removed by wire brushing. These sulphate �nests� are felt, to a large degree, to be responsible for the formation of blisters at anodic sites . Such blisters lead to failure, due to local bulging and eventual cracking of the coating.
Mayne�s experiments were carried out in Cambridge, over the period 1945 to 1953. The contamination of his specimens was determined to be due to the burning of fossil fuels, this was justified by a seasonal variation in the degree of contamination giving higher levels in the winter months. One could possibly argue that with the advent of smokeless zones in Britain that, in non-industrial urban areas, this problem should be greatly alleviated. Having said this, industrial areas still provide an adequate level of contamination to require attention. The problem of sulphates in industrial areas becomes one of chlorides in marine environments.
In a search for environmentally acceptable coatings that may be applied over compromised metal surfaces, attention has been focused on the field of �surface tolerant coatings�. Frondistou-Yiannas  surveyed the range of these coatings and their reported modes of action. The mechanisms considered were: conversion of rust to magnetite, conversion of iron oxides to other compounds and inertisation of soluble salts. In addition to these, Thomas  also considered barrier coatings, the ability of the coating to wet and penetrate the rust and anti-corrosive pigments. The field trials of the former investigator, which were carried out in both marine and industrial areas, concluded that none of the systems tested was expected to provide long-term protection in the aggressive environments used in the tests.
So called rust converters are mainly formulations based upon tannic acid. An extensive review of the mechanisms involved in the action of tannin based coatings, details the reaction mechanisms by which it is claimed that tannins help to protect steel. The action of tannin-based treatments is explained by the inhibition of the formation of magnetite, which if allowed to develop, would increase the area available for oxygen reduction. An alternative name of �rust deactivators� is proposed for these formulations, which may lead one to think that they have some effect upon the contaminants that are invariably present. However, this statement is followed by the fact that tannins are unable to neutralise surface contaminants such as sulphates.
In her review of rust conversion preparations, Thomas  states that the premise behind their use consists of the stabilisation of the rust, so that the redox reaction between Fe(II) and Fe(III) is no longer possible. In this case, in principle, corrosion would be prevented. It is pointed out, however, that tannic acid reacts with Fe(III) within regions underlying thick layers of surface rust; therefore, it would be necessary to wire brush the surface prior to treatment.
A further assessment of the action of tannic acid has been conducted by Morcillo et al. Panels that had been exposed outdoors for up to two years were thoroughly wire brushed and immersed in tannic acid of various concentrations. Fresh panels were also immersed, to assess the effect on clean steel. A film of ferric tannate was found to form on the surface of the rusty panels, this film was found to be extremely porous. The results from these tests showed that tannic acid promotes rusting on clean steel and does not significantly reduce that of rusted steel. Furthermore, it was suggested that due to the nature of the ferric tannate film formed on the surface, that the tannic acid might attack the iron at the base of the pores.
Perhaps the most poignant point comes from Frondistou-Yiannas� paper where, all the rust converter systems under examination did not make it to the field trials, as they failed the initial laboratory screening due to blistering and intercoat disbonding.
An alternative method of dealing with contaminated surfaces proposed by Sykes is the use of water-borne paint. The contention here is that if the requirements of the coating fall into any of the following criteria:
Then an aqueous medium, in the form of a water-borne coating probably offers the best chance of success; providing a method to either dissolve or bind the soluble ions while the paint is drying.
Experimental work was carried out on a 120�m thick water-borne vinyl acrylic latex system applied over pre-rusted steel. The substrates were coated in the as-received condition, wire brushed or grit blasted prior to application of the coating. The formulation from which the coating was taken normally has its pH adjusted to 4.5 - 5.0. During the investigation, the pH was taken down to 2.7. This lower pH was found to aid in the removal of light deposits of rust resulting in a decrease in the degree of blistering of the coating.
Another feature highlighted in this work was that dry grit blasting could exacerbate the problem of coating rusty steel by exposing contaminants buried in the rust layer. This feature has been noted elsewhere. Research sponsored by Nuclear Electric determined that dry blasting not only exposes buried salts but also spreads them. The research was primarily aimed at the determination of an acceptable level for residual salt contamination. Substrates were pre-contaminated with two salt compositions as shown in Table 1, prior to being coated with a three-coat alkyd paint system.
Table 1: Composition of salt mixtures, after ref. 
Exposure tests were carried out on panels contaminated with progressively higher levels of salt contamination. The exposure trial ran for 6 � years, after which pull-off adhesion tests were carried out. With zero contaminant, the mode of failure was cohesive within the primer coat. As the salt levels increased, progressively less primer remained adhered to the steel surface. At higher contamination levels, there was a change from mixed to total adhesive failure of the primer. This change was abrupt and corresponded to the values shown in Table 2
Table 2: Critical Contamination Levels Applicable to Different Environments.
Critical Contamination Level
2 �g cm-2
6 - 20 �g cm-2
The conclusion drawn by this investigation was that the adhesion, and therefore the performance, of the system was severely affected by salt contamination of the substrate, where the levels were in excess of 2 �g cm-2 when the salts were predominantly chlorides.
An epoxy coating used as a tank lining was also tested. In this case, both the marine and the urban salt contamination resulted in coating failure at levels above 20�g cm-2. The marine salt, however, brought about the failure in 6 days compared with 56 days for the urban salt composition.
The more aggressive nature of the chloride ion is attributed to higher osmotic pressures exerted. Additionally, chlorides tend to be more soluble than other salts and of a lower formula weight. Hence, a given salt burden of a chloride represents a higher molar concentration, which produces a higher osmotic pressure.
As often, in the field of organic coatings, the literature provides what appears at first sight, to be contradictory data. The critical chloride levels determined by Morcillo  for example was determined to be 50�g cm-2. It is important that one realises that each different coating/metal system is likely to have various parameters, including the chloride levels it can tolerate, that are unique to itself.
Another effect of contaminated surfaces found by Morcillo was the degradation of a chlorinated rubber primer applied over contaminated rusted steel. The chloride levels found under the films in this investigation were too high to be accounted for by the initial contamination or the exposure environment. An XPS analysis was carried out which indicated that the C-Cl bonds in the polymer underwent a chemical transformation that yielded Cl- and accounted for the increased chloride concentration in the rust at the interface. So the, initially present or subsequently formed, interfacial rust appeared to catalyse the decomposition of the polymer.
Morcillo discussed a further aspect of the importance of proper surface preparation, in a paper that considered the significance of the size and the type of abrasive used during pre-cleaning operations. One important point highlighted was that there is a critical surface profile, the value of which is determined by the environment along with the type and thickness of the coating. As the coating increases in thickness, the effect of the surface profile diminishes. The critical surface profile was found to be a function of the aggressiveness of the environment � a more aggressive environment resulted in a lower critical profile.
In conclusion, the surface to be coated is as important a part of the overall system as any other, so the state and condition of the substrate prior to painting significantly affects the demands that are placed upon the coating. Traditionally tried and tested methods for coating compromised steel are no longer environmentally acceptable, therefore new, safe, ecologically sound alternatives are demanded. Should the emerging water-borne paint technology provide a solution to the contamination problem, it would also have the added advantage of addressing the impending volatile organic content legislation.
Within the debate regarding the important features of the protection offered by organic coatings, the argument over the importance of adhesion is amongst the most contentious. It is often claimed (or at least inferred) that the adhesion is of paramount importance, to the exclusion of all other features. According to Bullett and Prosser �the ability to adhere to the substrate throughout the desired life of the coating is one of the basic requirements of a surface coating, second only to the initial need to wet the substrate�, Funke describes adhesion as the most important and decisive property of a coating.
In order to enter into the debate as to the importance of adhesion in the protective mechanisms of paint systems, it is necessary to consider the process of adhesion itself. In particular, the principle of adhesive failure needs to be addressed. For two surfaces to adhere to each other, be they of the same or of different materials, there is a need for intimate contact on an atomic scale between them. In the case of a liquid coating, the degree of intimate contact is governed by the ability of the coating to wet the substrate. This wetting ability needs to be present not only in the static sense, where the contact angle between the substrate and the liquid may be used to determine how efficient the wetting is; but also in a dynamic sense, where the rate of wetting of the substrate is not countered by the rate of shear involved in the application of the coating.
According to Bikerman a true adhesion failure is improbable, a view shared by others.,  If one subscribes to this point of view, then it must be the case that all failures that involve detachment of the coating from the substrate, must include some element of either cohesive failure of the coating or the oxide layer on the surface of the steel. This view is reinforced by XPS work which has been carried out by several workers, who have found that traces of polymeric material is normally found on the �metal� surface of a metal/polymer interfacial fracture, which may have appeared to be a purely adhesive failure from a visual examination., ,  Although the explanation of a cohesive polymer or oxide failure may account for the observed failure modes, this is probably still an oversimplification.
The function of the adhesive joint is to effectively remove the interface from between the two different materials. In order to do this, a gradual change must occur in the chemical composition of the coating and the substrate in the interfacial area. Plato elucidated this concept in his explanation of how the different elements in the universe could exist as one:
It is not possible for two things to be fairly united without a third for they need a bond between them which shall join them both, that as the first is to the middle, so is the middle to the last, then since the middle becomes the first and the last, and the last and the first become the middle, of necessity, all will come to be the same. And being the same with one another, all will be a unity.
The inference of this is that in a true adhesive joint, there is no clearly defined interface and that one needs to see the interfacial area as a gradual transitory phase where the coating and the substrate affect and are effected by each other. So, in the same way that the atomic structure of the metal gradually changes from that of the bulk phase to that of the oxide on the surface � it is proposed that the properties of the surface of the uncoated metal are changed when a polymeric coating is applied to it. In the same way, Leidheiser suggests that the structure of the coating varies as one passes through the bulk of the film to the interfacial region with the metal. This region, which extends into the polymer, is known as the interphase. The size of the interphase is not known, but is thought to be in the nanometer range.
Whilst loss of adhesion may occur under a number of situations, a major form of this type of failure is loss of adhesion resulting from exposure to aqueous or humid environments., , ,  As a result of work into loss of adhesion due to water ingress, it has been concluded that adhesion tests in the dry state are of little use for applications in immersed or humid environments.,  Indeed, it has been proposed that, in isolation, adhesion tests are insufficient to determine the ability of a coating to control corrosion.
It is true that the adhesion plays an important role in the protective mechanism of coatings, but one should take care when claiming that it is the overriding consideration. When one considers the process of cathodic delamination it is clear that, once the paint has become detached from the substrate, the underlying metal is exposed to the environment and is no longer afforded any protection from the coating system. One should, however, be careful to ensure that the whole process is viewed in context. Whilst the loss of adhesion, resulting from the delamination process, effectively negates the protection afforded by the coating, it is important to consider whether the original adhesion is the deciding factor in the delamination process. Gowers and Scantlebury suggested that the beneficial role of the adhesion of a paint coating is due to the impairment of the formation of a layer of electrolyte at the coating/substrate interface, preventing ionic current flow and the spread of corrosion over the surface.
A case in point is that of the use of silanes as adhesion promoters. There has been a considerable amount of work carried out in this field and the results may, at first sight, appear contradictory. The fundamental reason why the results of various investigations lead to different conclusions is the starting point and the assumptions made by the investigators.
Some work has been carried out which has shown that the use of silane coupling agents leads to an improvement in wet adhesion. This result has been used to assert that this treatment will therefore result in enhanced protection against such phenomena as underfilm corrosion or cathodic delamination. This claim is based on the assumption that good adhesion to the substrate will necessarily lead to improvement in the protection afforded.
The work of Marsh et al has tested this assumption and has shown that although silanisation does in fact provide better wet adhesion, the treatment afforded no improvement in resistance to delamination and that underfilm corrosion was significantly worse on the silane treated specimens. It was pointed out by Marsh that the use of the cathodic delamination test as a determinant of adhesive strength is inappropriate, as the work showed that improved adhesion played no part in the rate of delamination. Furthermore, the cathodic delamination test is therefore a measure of cationic mobility rather than adhesive strength. Other work on salt spray exposure of coated panels has found that no adhesion loss had occurred although severe underfilm corrosion had taken place.
Good surface preparation is the key to good adhesion but the type, as well as the condition, of the substrate has been found to have a strong influence upon the initial dry, and the subsequent wet, adhesion of a metal/polymer system. Gosselin showed that the metal used as the substrate affected the strength of the final dry bond and Arslanov found that the adhesive strength of a coating on aluminium initially decreased upon exposure to water, but subsequently got stronger.
One method that has been suggested by Funke to increase the adhesive strength of coated systems is the application of thin layer technology.,  Here, it is suggested that the adhesion to the substrate could be enhanced as a result of co-operative bonding between the intermediate thin polymer layer and the main bulk of the coating.
The proposed thickness of these coatings is in the nanometer range.  While this is in conflict with common coating practice whereby the base primer coat is thicker than the surface profile of the metal, when viewed in conjunction with Leidheiser�s contention regarding the interfacial region, and Kumins� model of restricted chain mobility within thin layers; this may well be an avenue worthy of investigation.
In conclusion, though the adhesion of a coating to the substrate is an important factor, which must be taken into consideration when addressing the mechanisms of corrosion protection afforded by the coating, the overall picture is much more complex and the overall mechanism that must be considered is a combination of various contributory factors. Karyakina and Kuzmak have produced an extensive overview of the processes of adhesion and other contributory factors, together with a review of experimental techniques for the evaluation of coating properties.
 Mayne J.E.O. Paints For The Protection Of Steel - A Review Of Research Into Their Modes Of Action. Br. Corr. Journal. Vol. 5, May 1970, pp 160 - 111.
 Hare C.H. Barrier Coatings. J. Prot. Coat. Linings, Feb 1989, pp 59 - 69.
 Royston I. A Ha'p'orth O' Tar, SURCON '97, Birmingham, Sept 1997.
 Asbeck W.K. and Van Loo M. Critical Pigment Volume Relationships. Ind. Eng. Chem. Vol. 41, No. 7, 1949, pp 1470 -1475.
 Bierwagen G.P. Critical Pigment Volume Concentration As A Transition Point In The Properties Of Coatings. J.Coat.Tech. Vol. 64, No. 806, 1992, pp 71 - 75.
 Bell S.H. The Structure Of Paints. JOCCA, Vol. 38, Oct 1955, pp 595 - 623.
 Thomas, N.L. The Barrier Properties of Paint Coatings Prog Org Coat, Vol. 19, 1991, pp 101-121
 Dickie R.A. Smith A.G. How Paint Arrests Rust, Chemtech, January 1980, pp 31 - 35.
 Bacon C.R. Smith J.J. and Rugg F.M. Electrolytic Resistance in Evaluating Protective Merit of Coatings on Metals. Ind Eng Chem, Vol.40, No 1, 1948 pp 161-167.
 Kittelberger W.W. and Elm A.C. Water Immersion Testing of Metal Protective Paints: Role of Osmosis in Water Absorption and Blistering. Ind Eng Chem Vol. 38. No. 7. 1946, pp 695-699.
 Mayne J.E.O. The mechanism of the protective action of an unpigmented film of polystyrene JOCCA Vol. 32 No 352 1949, pp 481-487.
 Mayne J.E.O. The Mechanism of the Inhibition of the Corrosion of Iron and Steel by Means of Paint Official Digest, Feb. 1952, pp 127-136.
 Corti H. Fern�dez-Prini R. and G�mez D. Protective Organic Coatings: Membrane Properties and Performance Prog Org Coat, Vol. 10, 1982, pp 5-33.
 Skoulikidis T. and Ragoussis A. Diffusion of Iron Ions Through Protective Coatings on Steel, Corrosion, Aug 1992, pp 666-670.
 Maitland C.C. and Mayne J.E.O. Factors Affecting the Electrolytic Resistance of Polymer Films Official Digest, September 1962.
 Cherry B.W. and Mayne J.E.O. The Resistance Inhibition of Corrosion in Unpigmented Systems Official Digest Vol. 33 No. 435. 1961, pp 469-480.
 Cherry B.W. and Mayne J.E.O. Ionic Conduction Through Varnish Films 1st Int Cong on Mett Corr London 1961, pp 539-544.
 Miskovic-Stankovic V.B. Drazic D.M. and Teodorovic M.J. Electrolyte Penetration Through Epoxy Coatings Electrodeposited on Steel Corrosion Science, Vol. 37 No. 2, 1995, pp 241-252.
 Kinsella E.M. and Mayne J.E.O. Ionic Conduction in Polymer Films 3rd International Congress on Metallic Corrosion, Moscow 1966, pp 117-120.
 Kinsella E.M. and Mayne J.E.O. Ionic Conduction in Polymer Films: I. Influence of Electrolyte on Resistance. Br. Polym. J. Vol. 1. July 1969, pp 173-176.
 Mayne J.E.O. and Scantlebury J.D. Ionic Conduction in Polymer Films: II Inhomogeneous Structure of Varnish Films. Br. Polym. J. Vol. 2, September 1970, pp 240-243.
 Muizebelt W.J. & Heuvelsland W.J.M. Permeabilities Of Model Coatings: Effect Of Cross Link Density And Polarity. Polymeric materials for corrosion control. ACS symposium 322 1986, pp 110-114.
 Kinsella E.M. Mayne J.E.O. & Scantlebury J.D. Ionic Conduction In Polymer Films III Influence Of Temperature On Water Absorption. , Br Polym J, Vol. 3, 1971, pp 41 - 43.
 Mayne J.E.O. & Scantlebury J.D. Ionic Conduction In Polymer Films IV The Effect Of Pigmentation With Iron Oxide, Br Polym J, Vol. 3, 1971, pp 237 - 239.
 Mayne J.E.O. and Mills D.J The Effect of the Substrate on the Electrical Resistance of Polymer Films. JOCCA Vol. 58, 1975, pp 155-159.
 Mills D.J and Mayne J.E.O. The Inhomogeneous Nature of Polymer Films and its Effect on Resistance Inhibition. in Corrosion Control by Organic Coatings Ed H.Leidheiser Jr. 1981, pp 12-17.
 Mills D.J. PhD Thesis, Cambridge, 1973.
 Gay P.J. Blistering of paint films on metal, JOCCA, Vol. 32, No. 352, 1949, pp 488-498.
 Mayne J.E.O. The Blistering of Paint Film. Part II Blistering in the Presence of Corrosion. JOCCA Vol. 31. No 12, 1950, pp 538-547.
 Leidheiser H. Jr. Corrosion of painted metals - A review Corrosion-NACE, Vol. 38, No. 7, 1982, pp 374 - 383.
 Appleman B.R. Painting over soluble salts: A perspective J. Prot. Coat. Lin. Oct 1987, pp 68 - 82.
 Hare C.H. Blistering Of Paint Films On Metal, Pt.1: Osmotic Blistering, J.Prot.Coat.Lin. Feb 1998, pp 45-63.
 Funke W Toward a unified view of the mechanism responsible for paint defects by metallic corrosion Ind. Eng. Chem. Prod. Res. Dev. Vol. 24, No. 3, 1985, pp 343 -347.
 van der Meer-Lerk L.A. & Heertjes P.M. Mathematical model of growth of blisters in varnish films on different substrates, JOCCA. Vol. 62, 1979, pp 256 - 263.
 Gowers K.R. & Scantlebury J.D. Blistering Phenomena On Lacqured Mild Steel, Corrosion Science, Vol. 23, No. 9, 1983, pp 935-942.
 Storfer S.J. & Yuhas S.A. Jr. Mechanism of Blister Formation in Organic Coatings, Mat Perf, Jul 1989, pp 35-41.
 Koehler E.L. Underfilm corrosion currents as the cause of failure of protective organic coatings Corrosion Control by Organic Coatings, Ed. Leidheiser H. Jr. 1981, pp 87 -96.
 Schwenk W. Adhesion Loss For Organic Coatings Causes And Consequences For Corrosion Protection. Corrosion Control by Organic Coatings, 1981, pp 103-110.
 Hare C.H. Non-Osmotically-Induced Blistering Phenomena On Metal, J Prot. Coat Lin, Mar 1998, pp 17-34.
 Nguyen T. Hubbard J.B. & McFadden G.B. A mathematical model for the cathodic blistering of organic coatings on steel immersed in electrolytes J. Coat. Tech. Vol. 63, No. 794, 1991, pp 43 - 52.
 Kittleberger W.W. Water Immersion Testing Of Metal Protective Paints. Influence Of Bare Metal Areas, Ind Eng Chem, Vol. 34, No. 9, 1942, pp 943-948.
 Leidheiser H. Jr. Cathodic delamination of polybutadiene from steel - A review J. Adhes. Sci. Tech. Vol. 1, No. 1, 1987, pp 79 - 98.
 Leidheiser H. Jr. Alkali metal ions as aggressive agents to polymeric corrosion protective coatings Corrosion - NACE, Vol. 43, No. 5, 1987, pp 296 - 297.
 Smith A.G. & Dickie R.A. Adhesion failure mechanisms of primers Ind. Eng. Chem. Prod. Res. Dev. Vol. 17, No. 1, 1978, pp 42 - 44.
 Koehler, E.L. The influence of contaminants on the failure of protective organic coatings Corrosion - NACE, Vol. 33, No. 6, 1977, pp 209 - 217.
 Ritter J.J. Ellipsometric studies on the cathodic delamination of organic coatings on iron and steel J. Coat. Tech. Vol. 54, No 695, 1982, pp 51 - 57.
 Ritter J.J. & Kruger J. Studies on the subcoating environment of coated iron using qualitative ellipsometric and electrochemical techniques in Corrosion Control by Organic Coatings, NACE, Ed. H Leidheiser Jr. 1982, pp 28 - 31.
 Castle J.E. & Watts J.F. Cathodic Disbondment Of Well Characterised Steel/Coating Interfaces Corrosion Control by Organic Coatings, Vol. 1981, pp 78 - 86.
 Wiggle R.R. Smith A.G. & Petrocelli J.V. Paint adhesion failure mechanisms on steel in corrosive environments J. Paint Tech. Vol. 40, No. 519, 1968, pp 174 - 186.
 Knaster M. & Parks J. Mechanism of corrosion and delamination of painted phosphated steel during accelerated corrosion testing, J. Coat. Tech. Vol. 58, No. 738, 1986, pp 31 - 38.
 Sommer A.J. & Leidheiser H. Jr. Effect of alkali metal hydroxides on the dissolution of a zinc phosphate conversion coating on steel and pertinence to cathodic delamination Corrosion, Vol. 43, No. 11, 1987, pp 661 - 665.
 Dickie R.A. Chemical Studies Of The Organic Coating/Steel Interface After Exposure To Aggressive Environments, Critical Issues in Reducing the Corrosion of Steels, 1985, pp 379-396.
 Koehler E.L. The mechanism of cathodic disbondment of protective organic coatings - aqueous displacement at elevated pH, Corrosion, Vol. 40, No. 1, 1984, pp 5 - 8.
 Hammond J.S. Holubka J.W. deVries J.E. & Dickie R.A. The application of X-ray photo-electron spectroscopy to a study of interfacial composition in corrosion-induced paint de-adhesion Corrosion Science, Vol. 21, No. 3, 1981, pp 239 - 253.
 Watts J.F. & Castle J.E. The application of X-ray photoelectron spectroscopy to the study of polymer-to-metal adhesion J. Mat. Sci. Vol. 19, 1984, pp 2259 - 272.
 Leidheiser H. Jr. Wang W. & Igetoft L. The mechanism for cathodic delamination of organic coatings from a metal surface Prog. Org. Coat. Vol. 11, 1983, pp 19 - 41.
 Sharman, J.D.B. Sykes, J.M. & Handyside,T. Cathodic disbonding of chlorinated rubber coatings from steel, Corrosion Science, Vol. 35, No. 5-8, 1993, pp 1375-1383.
 Leidheiser H. Jr. & Wang W. Some substrate and environmental influences on the cathodic disbonding of organic coatings J. Coat. Tech. Vol. 53, No. 672, 1981, pp 77 - 84.
 Skar J.I. & Steinsmo U. Cathodic disbonding of paint films - transport of charge Corrosion Science, Vol. 35, Nos. 5 - 8, 1993, pp 1385 - 1389.
 Kinsella E.M. PhD Thesis Cambridge 1967
 Armstrong R.D. & Johnson B.W. An investigation into the cathodic delamination of unpigmented chlorinated rubber films Corrosion Science, Vol. 32, No. 3, 1991, pp 303 - 312.
 Baxter I.K. Surface tolerant protective coatings, Step Into the 90�s, Transactions of the 1st Joint Conference on Corrosion, Finishing and Materials, Broadbeach, Queensland, 1989.
 Gross, H Examination Of Salt Deposits Found Under German Painted Steel Bridge Decks, Materials Performance, Vol. 22, No. Oct 1983, pp 28-33.
 Feliu S. Bastidas J.M. Galv�n J.C. Feliu S.Jr. Simancas J. & Morcillo M. Electrochemical Determination Of Rusted Steel Surface Stability. , J. Applied Electrochemistry. Vol. 23, No. 2, 1993, pp 157 - 161.
 Feliu S. Galvan J.C. Feliu S.Jr. Bastidas J.M. Simancas J. Morcillo M. & Almeida E.M. An Electrochemical Impedance Study Of The Behaviour Of Some Pretreatments Applied To Rusted Steel Surfaces. Corrosion Science. Vol. 35, No. 5-8, 1993, pp 1351-1358.
 Morcillo M. The Detrimental Effects Of Water-Soluble Contaminants At The Steel/Paint Interface. Panoramic View Of The Authors Research On The Topic, Preceedings of 12th International Corrosion Congress. Sept 19-24 Houston, Texas. NACE INTERNATIONAL. 1993, pp 87 - 98.
 Neal D and Whitehurst T Chloride Contamination Of Line Pipe, Its Effect On FBE Coating Performance , Mat. Perf. Feb 1995, pp 47-52.
 Costa I. Faidi S.E. & Scantlebury J.D. Substrate effects on the corrosion performance of coated steels under immersed conditions Corrosion Control for Low Cost Reliability, 12th International Corrosion Congress, NACE, Houston, Texas, 1993, pp 437 - 448.
 Mickalonis J.I. & Leidheiser H. Jr. Corrosion inhibition of steel by lead pigments Corrosion NACE, Vol. 45, No. 8, 1989, pp 631 - 636.
 Bittner A. Advanced Phosphate Anticorrosive Pigments For Compliant Primers. J.Coat.Tech. Vol. 61, No. 777, 1989, pp 111 - 118.
 Scantlebury J.D. Organic Coatings Systems And Their Future In Corrosion Protection, Paper presented at Eurocorr 82, Budapest 1982
 Thomas, Noreen L The Protective Action of Coatings on Rusty Steel. J. Protective Coatings & Linings, Vol. 6, No. 12, 1989, pp 63 - 71.
 Frondistou-Yiannas S. Evaluation of rust tolerant coatings for severe environments J. Prot. Coat. Lin. August 1986, pp 26 - 35.
 Thomas N.L. Coatings for rusty steel: Where are we now? Surface Coatings International (JOCCA), Vol. 74, No. 3, 1991, pp 83 - 88, 97.
 Mayne J.E.O. The problem of painting rusty steel J. Appl. Chem. Vol. 9, Dec. 1959, pp 673 - 680.
 Mayne J.E.O. Current Views on How Paint Films Prevent Corrosion, JOCCA. Vol. 40, Mar, 1957, pp 183 - 199.
 Landolt D. & Favre M. Environment-friendly coatings for steel based on tannins: A critical review and new results Progress in the Understanding & Prevention of Corrosion, 10th European Corrosion Congress, Eds. Costa J.M. & Mercer A.D. Inst. Of Mats. 1993, pp 374 - 386.
 Morcillo M. Feliu S. Simancas J. Bastidas J.M. Galvan J.C. Feliu S. Jr. & Almedia E.M. Corrosion of rusted steel in aqueous solutions of tannic acid Corrosion, Vol. 48 No. 12, 1992, pp 1032 - 1039.
 Sykes J.M. Smith H.E.M. Moreland P.J. & Padget J.C. Protection of poorly prepared rusty steel surfaces by water borne paint Advances in Corrosion Control by Organic Coatings II, Eds. Scantlebury J.D. & Kendig M. 1994, pp 7 - 14.
 McKelvie A.N. Steel Cleaning Standards - A Case For Their Reappraisal, JOCCA, Vol. 60,1977, pp 277-237.
 Allan S.J. May R. Taylor M.F. & Walters J. The effect of salts on steels and protective coatings GEC Journal of Research, Vol. 12, No. 2 1995, pp 86 � 92
. Morcillo M. Feliu S. Galvan J.C & Bastidas J.M. Some observations on painting contaminated rusty steel J. Prot. Coat. Lin. Vol. 4, No. 9, 1987, pp 38 - 43.
 Morcillo M Simancas J. Fierro J.L.G. Feliu S. Jr. & Galvan J.C. Accelerated degradation of a chlorinated rubber paint system applied over rusted steel Prog. Org. Coat. Vol. 21, 1993. Pp 315 - 325.
 Morcillo M. Bastidas J.M. Simancas J Galv�n J.C. The Effect Of The Abrasive Work Mix On The Paint Performance Over Blasted Steel. Anti Corrosion Methods & Mats., Vol. 36, No. 5, 1989, pp 4 - 8.
 Bullett, T.R. & Prosser, J.L. Measurement Of Adhesion, Prog Org Coat, Vol. 1, 1972, pp 45 - 71.
 Funke W The Role Of Adhesion In Corrosion Protection By Organic Coatings. JOCCA, Vol. 68, No. 9, 1985, pp 229 -232.
 Bullett T.R. & Rudram A.T.S. The Coating And The Substrate, JOCCA, Vol. 44, No. 11, 1961, pp 787-815.
 Bikerman J.J. The Science of Adhesive Joints 2nd Edition Academic Press, New York and London, 1968, pp 137 � 150.
 Walker P. Organo Silanes As Adhesion Promoters For Organic Coatings. J.C.T. Vol. 52, No. 670, 1980, pp 49 - 61.
 Zorll U. New Insights Into The Process Of Adhesion Measurement And The Interactions At Polymer/Substrate Interfaces, JOCCA, Vol. 7, 1983, pp 1983.
 de Vries J.E. Holubka & Dickie R.A. X-ray Photoelectron Spectroscopy Study of Corrosion-Induced Paint Adhesion Loss on Conversion-Coated Steel, Ind. Eng. Chem. Prod. Res. Dev., Vol. 22, 1983, pp 256-261.
 Plato Timaeus
 Leidheiser H. Jr. and Deck P.D. Chemistry of the Metal-Polymer Interfacial Region Science Vol. 241, No 2, 1988, pp 1176-1181.
 Leidheiser H. Funke W. Water Disbondment and Wet Adhesion of Organic Coatings on Metals: A Review and Interpretation. , JOCCA, Vol. 70, No. 5, 1987, pp 121 - 132.
 Downey S.J. and Devereux O.F The Use Of Imperance Spectroscopy In Evaluating Moisture-Caused Failure Of Adhesives And Paints, Corrosion, Vol. 45, No. 8, 1989, pp 675-684.
 Crossen J.D. Sykes J.M. Knauss D. Briggs G.A.D. & Lomas J.P. The influence of water on the coating-metal interface, adhesion measurements & scanning acoustic microscopy. , *****CONFERENCE*************, Vol. ***, No. ***, 1994, pp 274-283.
 Nguyen T. Bentz D. & Byrd E. A Study Of Water At The Organic Coating/Substrate Interface , J.C.T., Vol. 66, No. 834, 1994, pp 39 - 50.
 Gosselin C.A. Effect Of Surface Preparation On The Durability Of Structural Adhesive Bonds, Polymeric Materials For Corrosion Control. ACS symposium 322 1986, pp 180-193.
 Funke W. How Organic Coating Systems Protect Against Corrosion. Polymeric Materials For Corrosion Control. ACS symposium 322 1986, pp 223-228.
 Troyk P.R. Watson M.J. & Poyezdala J.J. Humidity Testing Of Silicon Polymers For Corrosion Control Of Implanted Medical Electronic Protheses. Polymeric Materials For Corrosion Control. ACS symposium 322, 1986, pp 299-313.
 Gowers, K R & Scantlebury, J D An Electrochemical Investigation of the Effect of the Adhesion of a Lacquer Coating on the Underfilm Corrosion. JOCCA. Vol. 4, No. 71, 1988, pp 114-121.
 Marsh J. Scantlebury J.D. & Lyon S.B. Silinisation � A Dangerous Surface Treatment for Steel? CONFERENCE 1994 pp 243 � 253.
 Jin, X H Gowers, K R & Scantlebury, J D The Effect of Environmental Conditions on the Adhesion of Paints to Metals. JOCCA. Vol. 3, No. 71, 1988, pp 78-81.
 Arslanov V.V. & Funke W. The Effect Of Water On The Adhesion Of Organic Coatings On Aluminium, Prog Org Coat, Vol. 15, 1988, pp 355-363.
 Arslanov V.V. & Funke W. Improvement of the resistance to water of an adhesive joint between polymers and aluminium by using thin adhesion layers, Prog Org Coat, Vol. 15, No. 4, 1988, pp 365-372.
 Funke W. Improvement Of Wet Adhesion Of Organic Coatings By Thin Adhesion Layer, Surface Phenomena and Latexes in Waterborne Coatings and Printing Technologies, 1995, pp 115-122.
 Funke W. Thin-Layer Technology In Organic Coatings, Prog Org Coat, Vol. 28, No. 1996, pp 3-7.
 Kumins C.A. Physical Chemical Models For Organic Protective Coatings. J.C.T., Vol. 52, No. 664, 1980, 1981, pp 39 - 53.
 Karyakina M.I. & Kuzmak A.E. Protection By Organic Coatings: Criteria, Testing Methods And Modelling. Prog. Org. Coat. Vol. 18, 1990, pp 325 - 388.