Web http://www.jcse.org/

Volume 2 Paper 37


Effect of Barrier Pigments on Cathodic Disbonding Part 2: Mechanism of the Effect of Aluminium Pigments

O.�. Knudsen, and U. Steinsmo

SINTEF Materials Technology, 7465 Trondheim, Norway
E-Mail Adress:

Abstract

In Part 1 it was shown that aluminium flake pigments decreased cathodic disbonding of epoxy coatings significantly. In this article the mechanism of the effect of the aluminium pigments is studied. Two possible mechanisms have been investigated: a diffusion barrier mechanism and a mechanism where the aluminium pigments are chemically active. Water uptake, water diffusion rate, oxygen diffusion rate and ionic resistance were measured as a function of aluminium content in the paint. Adding 10% aluminium pigments decreased the oxygen diffusion rate about six times. However, the oxygen permeability was much less affected. The water diffusion rate decreased only slightly. The ionic conductivity increased when the aluminium pigment concentration increased. It was shown that the aluminium pigments only decrease the rate of cathodic disbonding when they are applied in the first coat, directly on the steel. The effect was therefore related to their presence at the steel/coating interface. This indicates that the aluminium pigments were chemically active in decreasing cathodic disbonding, i.e. that they affected the environment on the steel/coating interface. It was also shown that the aluminium pigments in the disbonded film corroded. The alkaline environment under the disbonded coating therefore reached the pigment particles in the coating.

Keywords: Organic coatings, cathodic disbonding, aluminium pigments

Introduction

In Part 1 we reported about the effect of aluminium and glass barrier pigments on cathodic disbonding of epoxy coatings. We then found that the barrier pigments had a large effect on cathodic disbonding. Epoxy and epoxy mastic binders were pigmented with between 0% and 20% (wt.) aluminium pigments. Coatings with 10% or more aluminium pigments applied on blast cleaned steel had no disbonding after 18 months exposure. After two years of testing the coating with 10% aluminium had disbonded about 3 mm. In the same test period the coatings without aluminium pigments disbonded 65 mm on average. When the coatings were applied on ground steel the disbonding of the coating without aluminium was two decades faster than the coating with 10% aluminium. The purpose with this paper is to investigate the mechanism for the effect of the aluminium pigments on cathodic disbonding.

A short review of cathodic disbonding and the use of barrier pigments in protective paints were given in the previous paper. Cathodic disbonding is caused by the formation of a highly alkaline aqueous film under the coating [1]. The alkalinity is produced by the cathodic reaction under the coating. Under the conditions we have used oxygen reduction is probably the dominating cathodic reaction under the coating . In addition to oxygen and water the oxygen reaction requires cations to maintain the charge balance at the cathode:

 
O2 + 2 H2O + 2 e- = 4 OH-

(1)

The aluminium pigments are shaped as flakes and tend to orient themselves parallel to the substrate. They are therefore expected to improve the barrier properties of the coating [2]. The aluminium pigments may have decreased cathodic disbonding by decreasing the transport through the coating of either of the reactants in Equation 5-1. A barrier mechanism implies that the aluminium pigments reduce the transport through the coating of at least one reactant, and that this reactant affects the rate of the disbonding. We have therefore studied the transport of oxygen, water and ions through an epoxy film as function of the concentration of aluminium pigments. The amide cured epoxy binder used in the previous paper was used again, with the same concentrations of aluminium: 0%, 5%, 10% and 20% by weight in the wet paint. The paints are called Epoxy 0, Epoxy 5, Epoxy 10 and Epoxy 20.

Besides any barrier effects, the aluminium pigments may decrease cathodic disbonding by affecting the chemical environment at the steel/coating interface. Leidheiser et al. have suggested that aluminium pigments may decrease cathodic disbonding by a buffer reaction [1]:

 
Al2O3 + 2 OH- = 2 AlO2- + H2O

(2)

The reaction will consume hydroxide and may therefore decrease the pH at the steel/coating interface. We have tried to verify or disprove this theory. However, we have not been able to measure the pH under the coating, which would be the most direct test of this theory. If the aluminium pigments decrease cathodic disbonding by being chemically active, it must be important to apply the aluminium pigmented coat directly on the steel and not in an intermediate or topcoat. If the effect of the aluminium pigments is connected to their presence at the interface, this will be a strong indication that the pigments are chemically active. However, it will not tell whether a buffer effect is the mechanism for the reduction in cathodic disbonding.

If the aluminium oxide on the surface of the pigments is unstable, then the aluminium pigments will corrode:

 
4 Al + 3 O2 + 4 OH- = 4 AlO2- + 2 H2O

(3)

This reaction also consumes hydroxide and will contribute to a buffer mechanism.

Experimental Method

Composition of the coatings and preparation of the test specimens were described in detail in Part 1.

Assessment of Water Diffusion Coefficient

Water uptake was measured gravimetrically. Prior to exposure the samples were kept in a desiccator and weighed regularly until constant weight was obtained. The samples were then clamped to cells filled with substitute seawater [3]. After 30 days they were removed, wiped carefully and weighed. The samples were then put in a desiccator and weighed regularly until constant weight. The temperature was held at 23�C during both adsorption and desorption. The water uptake was calculated by:

 

(4)

where W0 is the weight of the water saturated sample, W is the weight of the sample after desorption, A is the exposed area, d is the film thickness and r is the density of water. The diffusion coefficients were calculated from the water desorption rate. For short desorption time the diffusion coefficients are given by [4]:

 

(5)

where Mt is the mass desorbed water at time t (Wt - W ), M is the total amount absorbed water (W0 - W ) and D is the diffusion coefficient. Three parallels cut from the same panel were tested for each system. Mt/M was plotted against the square root of time and D was calculated from the slope of the curve. The curves were linear for short times, which means that the desorption followed Fick's law and that Equation 5-5 is valid. The water diffusion rate in a paint film may depend on the concentration of water in the polymer. The coefficients reported her were calculated from the initial part of the desorption process, i.e. when the water concentration in the coating was high.

Assessment of Oxygen Permeability and Diffusion Coefficient

Oxygen diffusion coefficients were found by the delay-time method. The samples were exposed in substitute seawater at 23�C and polarised to -700 mV SCE. Figure 1 shows the apparatus for exposing the samples in electrolytes with controlled oxygen concentration. The exposed area of the samples was 19.6 cm2. The oxygen was removed from the electrolyte by nitrogen until a current of about 0 pA was reached. This took about 30 days. The electrolyte was then suddenly saturated with air and the current was measured every 15 minutes. The diffusion coefficient was calculated as a function of the break-through time [5]:

 

(6)

The break-through time tb is found by extrapolating the linear part of the initial raising current to zero. Three parallels, cut from the same sample, were tested for each system.

Figure 1. Apparatus for exposing samples in electrolytes with controlled oxygen concentration.

Oxygen permeability was measured with a Mocon Ox-Tran 10/50 according to ASTM D3985 [6]. The temperature was kept at 23�C, and the relative humidity was about 75%. Free films were prepared by applying the paint on Teflon sheets. The films were applied in two coats to about 200 �m dry film thickness. The first coat was allowed to cure for 24 hours before the second coat was applied. After 24 hours, the paint films were gently peeled off the Teflon surface. The adhesion was very weak, and the films were still rather elastic, so the strain on the paint film was low. We therefore expect that the films were insignificantly affected by the preparation.

Ionic Conductivity and Open Circuit Potential

Polarisation curves were obtained with the same samples and apparatus as used in the oxygen diffusion experiments, when the electrolyte was saturated with air. The potential sweep rate was 20 mV per minute. The open circuit potential was found by extrapolating the polarisation curve to zero current.

Cathodic Disbonding

Steel samples, 80 x 80 x 5 mm, were cut from mild steel. The samples were ground with emery paper to roughness Ry = 5 �m. The paints were applied with an applicator in two coats. The first coat was allowed to cure for about 24 hours at room temperature before the second coat was applied. Coating thickness was measured with a magnetic gauge. The coatings were given a circular damage in the middle of the samples, 6 mm in diameter. Cathodic disbonding was tested in substitute seawater [3] at room temperature. The samples were polarised to -1050 mV SCE by a potentiostat. The disbonded diameter was measured in four directions, and the disbonded area was calculated from the average diameter.

TEM Analysis of the Aluminium Pigments

Epoxy 10 was applied on ground steel plates in two coats with a total dry film thickness of 200 �m. The samples were exposed in substitute seawater at 25�C, and polarised to -1050 mV SCE by a potentiostat. The panels were exposed until the disbonding front had moved about 10 mm from the initial holiday in the coating. TEM samples were prepared by ultramicrotomy [7]. Cross sections of the disbonded film were embedded in epoxy. Ultrathin slices of the cross sections were cut with a diamond knife and investigated in TEM. Element composition of the aluminium pigments was investigated by Energy Dispersive X-ray Spectroscopy (EDS).

Experimental Results And Discussion

Effect of Aluminium Pigments on Transport of Water, Oxygen and Ions

We have measured the transport parameters for oxygen, water and ions through the amide cured epoxy as a function of aluminium pigment concentration. Table 1 shows the transport parameters for water and oxygen in the four coatings.

Table 1. Transport of reactants through epoxy coatings with various aluminium concentrations. Water uptake, water diffusion and oxygen diffusion was measured on applied films of 300 �m DFT. Oxygen permeability was measured on 200 �m free films. Temperature: 23�C.

 
Water
Oxygen
 
Uptake
D
D
P
 
[Vol.%]

[10-10 cm2/s]

[10-9 cm2/s]

[10-9cm2/s]

Epoxy 0

3.3 � 0.07
5.5 � 0.4
32 � 2.5
1.1

Epoxy 5

3.3 � 0.04
4.3 � 0.5
10 � 0.6
-

Epoxy 10

3.2 � 0.03
4.2 � 0.7
5 � 0.7
0.7

Epoxy 20

3.8 � 0.22
2.4 � 0.7
5 � 0.8
0.6

All results are averages of three parallels, except oxygen permeability where only two parallels were tested. The deviation between the parallels for the oxygen permeability was less than the uncertainty in the instrument, which was 4%.

The water uptake was between 3.2 vol% and 3.8 vol% for the four coatings (i.e. the water concentration was between 1.8 and 2.1 mol/l). Epoxy 0, Epoxy 5 and Epoxy 10 had the same water uptake. The difference between Epoxy 20 and the others was statistically significant on the 2.5% significance level (i.e. 97.5% possibility for the conclusion to be true). The water uptake was therefore relatively little affected by the aluminium pigments. The water diffusion coefficients were calculated from the desorption rates, as described in the experimental section. Figure 2 shows the water desorption plotted according to Equation 5.5. The approximately straight lines obtained at short times in the diagram show that the desorption followed Fick's law and that Equation 5-5 can be used. The water diffusion coefficients depended on the water concentration in the film. The water diffusion coefficients in Table 1 were calculated from the short time desorption (the first 8 hours after exposure), when the water concentration in the film was high. This is most relevant for a submerged paint film, which is saturated with water. The difference between Epoxy 0 and Epoxy 5 is statistically significant on the 5% level. The difference between Epoxy 10 and Epoxy 20 is also statistically significant on the 5% level.

Figure 2. Water desorption rates for amide cured epoxy with various concentrations of aluminium at 23�C. Three parallels of each system. (Click the image to view an enlarged version)

The oxygen diffusion coefficient decreased when the aluminium content increased, as predicted by the barrier mechanism theory. Table 1 shows that the oxygen diffusion coefficient was about six times lower for Epoxy 10 than for Epoxy 0. The decrease was statistically significant on the 1% level. Increasing the aluminium concentration above 10% did not seem to have any effect on the oxygen diffusion coefficient. Figure 3 shows the current transient on three samples of Epoxy 5 after suddenly saturating the electrolyte with air. The break-through time in Equation 5-6 was found by extrapolating the linear part of the initial raising current to its crossing with the current level at the start. The break-through time is indicated for one of the samples in the figure.

The rate of oxygen transport depends on both the diffusion coefficient and the oxygen solubility in the film. The permeability coefficient includes the effect of the solubility and is therefore probably a better measure for the effect of aluminium on the oxygen transport. The aluminium pigments affected the permeability coefficient much less than the diffusion coefficient. Epoxy 0 and Epoxy 10 differed by a factor less than two. The aluminium pigments therefore seems to have increased the solubility of oxygen in the film significantly. Thomas has earlier reported that the oxygen permeability of an amide cured epoxy resin was about 130 cm3 100 �m (m2 day atm)-1 [8]. Converted to this unit, Epoxy 0 had a permeability of 98 cm3 100 �m (m2 day atm)-1. The permeability coefficients we have found therefore seem reasonable.

Figure 3. Oxygen diffusion delay time for three parallels of Epoxy 5. Current transient after changing the gas in the electrolyte from nitrogen to air. Exposed area 19.6 cm2. Conditions: substitute seawater, -700 mV SCE, 23�C.(Click the image to view an enlarged version)

Figure 4 shows the polarisation curves of the painted samples. The relationship between the current and the polarisation potential was linear. The effect of the external polarisation on the potential of the steel under the coating must therefore have been very small. Hence, the cathodic current was entirely limited by the resistance in the paint film [9]. The calculated resistivities are given in Table 2. The resistivity of the coatings decreased with increasing aluminium content. This probably means that the pigments introduce new pathways for transport of ions through the coating. There are several possible explanations for this. The curing of the binder around the pigments may be incomplete if the pigments inhibit diffusion of the curing agent. The pigments may also be surrounded by water filled voids with high conductivity. The decrease in the resistivity is large. The resistivity of Epoxy 0 and Epoxy 20 differed by more than two decades.

Figure 4. Polarisation curves obtained for painted steel samples. Substitute seawater, 21�C, DFT 300 �m, exposed area 19.6 cm2, average of 3 parallels. (Click the image to view an enlarged version)

The open circuit potential was also affected by the aluminium pigment concentration in the coatings. Table 2 shows that increasing the aluminium concentration from 0% to 20% decreased the open circuit potential from -126 mV to -340 mV SCE. It is difficult to say exactly what factor that made the open circuit potential decrease.

Table 2. Open circuit potential and resistivity calculated from the polarisation curves plotted in Figure 4. Averages of three parallels.

 

Open circuit potential

Resistivity

 

[ mV SCE]

[1010 W cm]

Epoxy 0

-126 � 10

1100 � 160

Epoxy 5

-125 � 26

170 � 29

Epoxy 10

-187 � 38

81 � 3.5

Epoxy 20

-340 � 26

7 � 0.6

Transport Properties Compared to Cathodic Disbonding

A barrier mechanism for the effect of aluminium pigments on cathodic disbonding implies that the pigments decrease the transport of at least one of the reactants, and that this reactant affects the rate of disbonding. The rate of cathodic disbonding should then correlate with the transport rate of this species. Figure 5 shows the transport properties and disbonding rate for the films as function of aluminium pigment concentration. The disbonding rates are taken from the disbonding experiments on ground steel substrate in Part 1. The responses are normalised to the results for Epoxy 0.

Figure 5. Cathodic disbonding and transport properties as function of aluminium pigment concentration. The responses are normalised to the result for Epoxy 0. (Click the image to view an enlarged version)

Figure 5 shows that the decrease in water diffusivity, oxygen diffusivity and oxygen permeability was relatively small compared to the effect of aluminium on cathodic disbonding. The transport of charge through the film increased with increasing aluminium content, i.e. opposite to the effect of the pigments on cathodic disbonding. Neither of the transport parameters correlates well with the disbonding rate. This indicates that the mechanism of the effect of the aluminium pigments on cathodic disbonding is something else than the barrier effect.

Active Aluminium Pigments

The aluminium pigments are covered with oxide. According to the supplier the pigments contain 1-3% aluminium oxide. Leidheiser et al. have suggested that the oxide on the pigments may act as a buffer that decreases the pH beneath the film [1]. Hydroxide generated by the cathodic reaction at the steel/coating interface is generally accepted as being responsible for cathodic disbonding, as discussed in the introduction in Part 1. If hydroxide is removed from the steel/coating interface by buffering reactions with the coating, disbonding may also decrease. According to the Pourbaix-diagram for aluminium [10], solid aluminium oxide is not stable above pH 9-10 and will react with hydroxide (Equation 5-2). A simple experiment was performed to test the buffer theory. Epoxy 0 and Epoxy 10 were applied in two-coat films in the four possible combinations of the two paints. The application and coating thickness is illustrated in Figure 6. If there is a buffer effect from the aluminium pigments, it will be important to have the aluminium pigments in the first coat. Then only system 3 and 4 will show decreased cathodic disbonding. If the effect of the aluminium pigments on cathodic disbonding is a general barrier effect, then system 4 will perform best, system 2 and 3 will perform equally and system 1 will perform worst. The disbonding rates are given in Figure 7.

Figure 6. Application of Epoxy 0 and Epoxy 10 in two-coat films. (Click the image to view an enlarged version)

Figure 7. Cathodic disbonding for two coat films with and without aluminium pigments applied on top of each other. Test conditions: Substitute seawater, -1050 mV SCE, 25�C, 4 parallels.

Figure 7 shows that the disbonding rate was low and almost the same for system 3 and 4. System 1 and 2 also had nearly the same disbonding rate, but much higher than system 3 and 4. This experiment clearly demonstrates the importance of applying the aluminium pigmented coat directly on the steel substrate. This also favours the buffer theory over the barrier theory. System 2 and 3 should have the same barrier properties, but the disbonding of system 3 is much slower than for system 2. In addition system 2, which should have been a better barrier, has only slightly less disbonding than system 1. The same is the case with system 3 compared to system 4. These results demonstrate that the positive effect of the aluminium pigments is linked to their presence at the steel/coating interface.

Corrosion of the Aluminium Pigments

Aluminium is not passive in aqueous solutions with pH above 9-10, and will corrode according to Reaction 5-3. Ritter et al.[11] and McLeod et al.[12] have found that the pH under the disbonded film can be as high as 14. We can then ask whether the aluminium pigments will corrode or be protected by the binder. If the pigments corrode they may gradually disappear and leave large voids in the film. The corrosion of the pigments may also have a positive effect on e.g. cathodic disbonding. Both oxygen and hydroxide are consumed in the corrosion reaction, which may decrease the disbonding rate. The buffer theory for the reduction in cathodic disbonding presented above presupposes that the sodium hydroxide under the coating has access to the pigments. If we can verify that the pigments corrode, it will also support the buffer theory.

Figure 8 shows a micrograph of a cross section of Epoxy 10 after cathodic disbonding. The coating was 360 �m thick and the sample had been exposed for three months in substitute seawater at 25�C, polarised to -1050 mV SCE with a potentiostat. The disbonding propagated along the cross section, and the disbonded distance was about 10 mm. The section of the film shown in Figure 8 is about 3 mm, and the disbonding front is to the right. The dark area in the cross section is filled with metallic aluminium flakes, while in the bright areas the aluminium pigments have corroded and are converted to aluminium oxide. In the left end of the section, the film has probably been disbonded for several weeks. Here the aluminium pigments have corroded into about 1/4 of the film thickness. Towards the disbonding front at right end of the picture, the corrosion of the pigments is less severe. Near the front, the bright area is only a thin line. Here the film has disbonded recently and only the pigments close to the interface have had time to corrode. Cross sections of Epoxy 10 outside the disbonded region (adhering coating), taken after two years exposure in the long term test in Part 1, showed no sign of pigment corrosion. The corrosion of pigments deep inside the film therefore seems to take place mainly after the coating has disbonded. Under an intact and adhering film, the cathodic reaction is probably too slow to give significant corrosion of the aluminium pigments.

Figure 8. Micrograph of a cross section of Epoxy 10 after disbonding.

We also prepared thin cross sections of the disbonded film by ultramicrotomy and analysed the pigments by TEM as described in the experimental section. Figure 9 shows micrographs of pigments taken at two positions in the film indicated below the pictures. The pigments shown in Figure 9A was found near the steel surface, less than 1 mm from the disbonding front. The picture shows several pigments located in a cluster. The pigments consist of a crystalline structure, which due to the different grain orientations leads to the varying contrast. Analysis of the electron diffraction pattern together with the Energy Dispersive X-ray Spectroscopy (EDS) seen in Figure 10A shows that the pigments consist of aluminium only. The Small aluminium oxide contribution seen in Figure 10A is due to oxidation of the TEM sample after preparation.

Figure 9B shows a cross section of two pigments near the steel surface, but about 5 mm from the disbonding front. Pigment 1, found on the left in the picture, has only corroded slightly and aluminium oxide is only found on the surface. Pigment 2, in the centre of the picture, is completely corroded. Electron diffraction pattern analysis together with EDS analysis shown in Figure 10B confirmed that this structure was amorphous and consisted primarily of aluminium oxide. The excess of oxygen is due to crystal water in the oxide. This crystal water started evaporating during the electron beam irradiation, forming the cellular structure seen in pigment 2.

This shows that the electrolyte under the coating had access to the pigments in the film. The corrosion of the pigments will consume hydroxide formed in the cathodic reaction at the steel/coating interface. The corrosion of the pigments is therefore another strong indication for that the aluminium pigments reduce cathodic disbonding by a buffer mechanism.

Figure 9. TEM micrographs of aluminium pigments. A: Aluminium pigments near the steel surface, less than 1 mm from the disbonding front. B: Aluminium pigments near the steel surface, about 5 mm from the disbonding front

Figure 10. Chemical composition of aluminium pigment particles after cathodic disbonding. A: Pigment from Figure 9A. B: Pigment 2 in Figure 9B.

Conclusion

In a previous article we have shown that aluminium pigments decrease the rate of cathodic disbonding for two different epoxy binders. In this article we have investigated two possibilities for mechanism behind the effect: A barrier mechanism and a mechanism where the aluminium pigments are chemically active.

The aluminium pigments decreased the oxygen diffusion rate through the epoxy coating. However, the oxygen permeability coefficient was much less affected. The transport of water was only slightly affected by the aluminium pigments. The ionic conductivity of the epoxy increased when the aluminium pigment concentration increased.

The aluminium pigments had insignificant effect on cathodic disbonding unless the aluminium containing film was applied directly on the steel surface. The effect was therefore connected to their presence near the steel/coating interface. This indicates that the aluminium pigments had an additional effect on cathodic disbonding besides any barrier effects. Others have suggested that buffering reactions with the oxide on the surface of the pigments may decrease cathodic disbonding. We have also shown that the aluminium pigments in the disbonded film corroded after cathodic disbonding. The alkaline environment under the disbonded coating therefore had access to the pigment particles in the coating.

Acknowledgements

We gratefully acknowledge the following companies and institutions for supporting the project: Kv�rner Oil & Gas, Aker Engineering, Statoil, Hydro, Saga, Conoco, Scana Offshore, Carboline, Jotun, International, Marintek, The Norwegian Petroleum Directorate, Norwegian Research Council. We also want to express our thanks to the following persons for their contribution: Rolf Berntsen at Jotun for preparing the paints, Nina Evje at Matforsk for performing the oxygen permeability measurements and Jan Halvor Nordlien at SINTEF for performing the TEM analysis.

References

  1. H. Leidheiser, W. Wang and L. Igetoft, "The Mechanism for the Cathodic Delamination of Organic Coatings From a Metal Surface", Prog. Org. Coat., Vol: 11, No: 1, p. 19-40 (1983).
  2. C. H. Hare, "Protective Coatings - Fundamentals of Chemistry and Composition", Technology Publishing Company, Pittsburgh (1994).
  3. ASTM D1141-90, Standard Specification for Substitute Ocean Water (1990).
  4. J. Crank and G. S. Park, "Diffusion in Polymers", Academic Press, London (1968).
  5. M. A. V. Devanathan and Z. Stachurski, "The adsorption and diffusion of electrolytic hydrogen in palladium", Proc. Roy. Soc. A, Vol: No: 270, p. 90-102 (1962).
  6. ASTM D3985-95, Oxygen Gas Transmission Rate Through Plastic Film and Sheeting Using a Coulometric Sensor (1995).
  7. T. F. Malis and D. Steele, "Ultramicrotomy for Materials Science", In: R. Andersen (ed), Materials Research Symposium Proceedings, Materials Research Society, Pittsburgh, (1990).
  8. N. L. Thomas, Journal of Protective Coatings and Linings, Vol: 6, No: 12, p. 63 (1989).
  9. U. Steinsmo and E. Bardal, "Factors Limiting the Cathodic Current on Painted Steel", Journal of Electrochemical Society, Vol: 136, No: 12, p. 3588-3594 (1989).
  10. M. Pourbaix, "Atlas of Electrochemical Equilibria in Aqueous Solutions", Pergamon Press, London (1966).
  11. J. J. Ritter and J. Kruger, "Studies of the Subcoating Environment of Coated Iron Using Qualitative Ellipsiometric and Electrochemical Techniques", In: H. Leidheiser (ed), Corrosion Control by Organic Coatings, Nace, Houston, TX, (1981).
  12. K. McLeod and J. M. Sykes, In: K. N. Strafford, P. K. Datta and C. G. Googan (ed), Coatings and Surface Treatment for Corrosion and Wear Resistance, Ellis Horwood Ltd., Chichester, UK, 295-314 (1984).