Failure mechanisms for organic coatings subjected to 0.1M sulfuric acid

If the rate of hydrogen formation is lessthan Its removal rate, there can be almost complete disbonding of ... of failure of these coatings after expo...
0 downloads 0 Views 454KB Size
Ind. Eng. Chem. Prod. Res. Dev. 1986, 25, 129-132

129

SYMPOSIA SECTION

I. Symposium on “Anticorrosion Barriers: Chemistry and Applications” Henry Leidheiser, Jr., Chairman 188th National Meeting of the American Chemical Society Philadelphia, Pennsylvania, August 1984 (Continued from September 1985 Issue)

Failure Mechanisms for Organic Coatings Subjected to 0.1 M Sulfuric Acid Malcolm L. White, Hyaclnth Vedage, Richard D. Granata, and Henry Leldhelser, Jr.’ Center for Surface and Coating Research, Department of Chemistry, Lehbh Universe, Bethbhem, Pennsylvania 180 15

The ability of a fluoropolymer and an epoxy coating to protect steel against corrosion In 0.1 M H2S04at 60 OC was evaluated by four techniques: (1) cathodic delamination, (2) tensile adhesion, (3) rate of blister formation, and (4) gas analysis of blisters. The data obtained show that acid diffuses through the coating to react with the steel surface to oxidize (corrode) it and to form hydrogen gas by the reduction of the hydrogen ion in the acid. I f the generation of hydrogen occurs faster than its removal by diffusion through the coating or into the steel, a blister will form. If the rate of hydrogen formation is less than its removal rate, there can be almost complete disbonding of the coating with no visible blisters,

Introduction Although steel is often coated with organic-based coatings for protection against acid environments, there has been relatively little work done to determine the causes of failure of these coatings after exposure to acid. Observations made on coatings that have been in service in SOzscrubbing (flue gas desulfurization) systems where low pH conditions are frequently encountered show that a common mode of failure is blistering (Berger et al., 1980). It is generally assumed that blistering is caused by migration of liquid through the coating as the result of an osmotic effect (Berger, 1978). The migration implies that the coating is acting as a semipermeable membrane, so that ionic material is excluded (Funke, 1981). However, it is well-known that ionic components such as sodium and chloride penetrate and diffuse through organic coatings (Parks and Leideiser, 1986). In the course of some studies on the use of coatings for corrosion protection of steel in dilute sulfuric acid, we have observed blisters forming in some coatings, the blisters being filled with gas, not liquid. Two types of coatings were particularly prone to blistering on exposure to acid-a fluoropolymer and a bisphenol A epoxy hardened with an amide amine. These resins are frequently used commercially as acid-resistant coatings, so a more detailed study was carried out in order to understand better the cause of the blistering. These coatings also provide a vehicle for studying the mechanism of failure of these coatings when exposed to acid solutions. The blistering was observed at the coating/metal interface, and there was corrosion of the steel. Since this 0196-4321/86/1225-0129$01.50/0

blistering was associated with a complete loss of adhesion, the evaluation techniques used were those that were related to adhesion: cathodic delamination and tensile adhesion as well as visual observations of blistering. From the data obtained by using these techniques it was possible to develop a mechanism for the blistering and also to explain the failure of these coatings to protect the underlying steel against corrosion.

Experimental Methods Materials, Application, and Exposure Techniques. Two coatings were used in this study, a fluoropolymer and an epoxy. The fluoropolymer was a low molecular weight copolymer of vinylidene fluoride and tetrafluoroethylene, which is vulcanized by reaction with an aliphatic amine using PbO or MgO as an acid acceptor, with carbon black added as a pigment. The coating was applied by spraying from a dispersion in methyl ethyl ketone and was air-dried for several hours to remove solvent. The coated substrate was either cured at room temperature for 1week or baked at 100 “C in air for 2 days. The final cured thickness in this study was 1-2 mils, far below the thickness of 40 mils recommended by the manufacturer; this thickness was used so the blistering would occur in a relatively short time. The epoxy was bisphenol A resin hardened with a poly(amideamine). I t was applied by doctor blading to a cured thickness of 8-12 mils with a Gardner film casting knife. After several hours at room temperature, the epoxy coating was baked at 60 O C for 4 h. The substrates were cold-rolled low-carbon SAE 1010 steel, 32 mils thick, which had been blasted on both sides 0 1986 American Chemical Society

130

Ind. Eng. Chem. Prod. Res. Dev., Vol. 25, No. 2, 1986

n

LEAD

YEB O L T

+ADHESIVE

+

COATlHGy STEEL

i'

SUBSTRATE

PULL

IN

TENSILE

TESTER

Figure 1. Technique for tensile adhesion testing.

with silica sand to a profile of about 1/4 mil. This profile was the maximum that could be developed without excessive warping of the panels. After the substrates were sandblasted, they were stored in a desiccator until they were coated. For some experiments, the substrates were cleaned in boiling trichloroethylene or with an aqueous alkaline cleaner (Ridoline 72, Amchem Products, Ambler, PA) at 60-65 O C . Following the latter clean, the substrates were rinsed in overflowing water and blotted dry. After the steel substrate was coated, a circular disk, 15/16 in. in diameter, was prepared with a punch and die set in a hydraulic press. This disk was put on a 125-mL-size polypropylene bottle half-filled with 0.1 M H2S04,with the coated side facing the inside of the bottle. A seal was made between the disk and the top of the bottle with a 1/16-in.-thickgasket made from Buna-N rubber (1in. i.d. in. 0.d.). The disk and gasket were held onto the and 15/16 bottle by screwing on the cap of the bottle. To facilitate making a good seal the inside portion of the cap was removed, so that the cap became a ring holding the disk only at its circumference. The coating was exposed to the acid by inverting the bottle. The inverted bottle was placed in an oven at 60 "C. Evaluation Techniques. For the cathodic delamination experiments, the coated metal substrate was cut into a 1 in. X 2 in. piece, and the back and three edges were masked with an epoxy coating, so the substrate could be partially immersed in an electrolyte with an electrical connection made to the portion of the substrate above the electrolyte. A 1/16-in.-diameter hole was made in the coating prior to immersion in order to expose the steel substrate. The sample was immersed in the electrolyte at room temperature, and a fixed cathodic potential of -1.2 V vs. SCE was applied by using a carbon counter electrode. After a given time of exposure, the substrate was removed, rinsed, and dried, and the adhesion of the coating around the hole was tested with Scotch tape. The area of coating that delaminated from the substrate was determined. The delamination process in neutral solutions is caused by diffusion of ions, water, and oxygen through the coating to the steel/coating interface where the oxygen is reduced to hydroxide ions. The increasing basicity at the interface weakens the bond between the coating and the steel, and delamination of the coating occurs (Leidheiser and Wang, 1981b) The tensile adhesion of the coatings was measured with the technique shown in Figure 1. A flat-bottom lead anchor was fastened to the coating with a cyanoacrylate adhesive (Loctite 414) and, after curing for 1h, the anchor and coating were pulled off the substrate with a Dillon

.

P I E R C I N G NEEDLE

CUTOFF BOTTLE

G A S BUBBLE

Figure 2. Technique for sampling gas in blister.

tensile tester at a pull rate of 1 in./min. The force at failure was divided by the anchor contact area to convert the results to pressure (psi). Blistering was determined by visual observation. The blisters were always gas-filled, with separation of the coating from the substrate and with extensive corrosion of the substrate. The gas in some of the blisters was collected by using the technique shown in Figure 2. The plastic bottle used for exposing the coating to the acid was cut off so that a plastic funnel could be put into the bottle to cover the sample area where the blister had formed. The narrow end of the funnel was covered with a rubber septum. The funnel was completely filled with liquid by withdrawing the air with a hypodermic needle. A piercing needle was put through the septum and the blister broken. The escaped gas collected in the narrow end of the funnel, where a sample of it was withdrawn by a sampling syringe. This gas was analyzed by injecting a 20-pL sample into a Model 4510 Finnigan GC/MS using a 30-m capillary column (DB-1701)with helium as a carrier gas and scanning from a mass/charge of 1:lOO. Results Figure 3 shows the rate of cathodic delamination of the fluoropolymer for three solutions. The rate is highest for the Na2S04solution at a near-neutral pH of 6. When H2S04is added to lower the pH to 1, the rate decreases

Ind. Eng. Chem. Prod. Res. Dev., Vol. 25, No. 2, 1986

I

/.

10

131

T

/*

0.2M Na2S04 pH=6

Figure 5. Tensile adhesion of fluoropolymer after curing at room temperature and after baking. D A Y S IN ELECTROLYTE

Figure 3. Cathodic delamination of fluoropolymer in salt and acid solutions.

Table I. Effect of pH on Blistering of Bisphenol A Epoxy HSO, concn. M measured DH time to first blister, h 0.5 0 15 0.25 0.3 22 0.1 0.7 48-77 0.02 1.4 >888 0 (HzO) 6.0 >888 Table 11. Alkali-Metal Ion Effect on Blistering of Bisphenol A Epoxy@ time to first blister, h salt added none 31-44 Li2S04 96-218 Na2S04 276-324 &So4 384-420 a O . l M HzS04

50

100

150

200

250

300

350

TIME IN O.lMH,SO,,hrs

Figure 4. Tensile adhesion of fluoropolymer with various cleaning procedures on substrates prior to coating.

significantly. When the Na2S04in the latter solution is replaced with HCl to yield a pH of less than 1, there is essentially no delamination, suggesting that acid diffusing through the coating is neutralizing the base formed by the oxygen reduction at the steel/coating interface. Figure 4 shows the effect of cleaning on the tensile adhesion of the baked fluoropolymer as a function of exposure to acid. The initial adhesion is highest for the alkaline-cleaned substrate (500+ psi), lowest for the trichloroethylene cleaning (200 psi), and intermediate (425 psi) for the samples that were not cleaned. On exposure to acid, however, the values merge and after 350 h are all about 200 psi. Thus, it appears that the value of the initial adhesion is not predictive of the values after acid exposure. These results suggest that there is a change occurring at the interface that has a "leveling" effect on the coating/ steel bond. This effect is most likely caused by acid that diffuses through the coating and reacts with the steel to destroy the initial bond between the coating and the steel. The effect of baking the fluoropolymer on the tensile adhesion is shown in Figure 5. The adhesion of a sample cured at room temperature drops rapidly in the first 100 h of acid exposure to about 50 psi, and then drops to zero after 250 h. It is important to note that significant blistering did not occur until after there was almost complete loss of adhesion. The baked sample has about the same

+ added salt (0.5 M).

initial adhesion, but a more gradual decrease in value with acid exposure. Table I shows the effect of HzS04concentration on the rate of blistering (expressed as the time to the first observed blister) for the epoxy coating. As the pH increases, the blistering time increases, with a very sharp increase in the time to first blister above pH 0.7. Below this pH the concentration of acid diffusing through the coating to the steel surface is sufficient to neutralize the hydroxide ion formed by the oxygen reduction reaction with excess amounts of acid available for reaction with the steel and formation of hydrogen gas by reduction of hydrogen ion. Above pH 0.7 the oxygen reduction reaction probably predominates and there is no hydrogen gas formation and thus no blistering in 888 h. Table I1 shows the change in blistering rate with the addition of alkali metal ions to H2S04.There is a decrease in blistering rate with increasing nominal size of the alkali-metal ion (increasing atomic weight), but because of strong hydration effects, the effective size decreases with increasing atomic weight (Moelwyn-Hughes, 1965; Leidheiser, 1981a). Thus the smaller (hydrated) ions diffuse more readily through the coating, dilute the acid, and decrease the blistering rate. Additional details of cation effects have been published (Leidheiser et al., 1984). The sample of gas collected from an epoxy-coated substrate that had been exposed to acid for about 3000 h shows peaks for H 2 ( 2 ) ,He(4), C(12), N(14), H20(18),Ne(20),N2(28), oz(32), and co2(44). There are also small peaks at 26 and 27 (acetylene ?), 12 and 13 (methane), and 41-43 and 57 (hydrocarbons ?). A sample of injected air shows no peaks for hydrogen or any of the small peaks listed above.

132

Ind. Eng. Chem. Prod. Res. Dev., Vol. 25, No. 2, 1986 H a H+

so;

1 2

H*

Fe e 2 H + - F e + * + H 2 1

BLISTERING

H2 f

>

H

2.2

Figure 6. Mechanism for coating failure in acid.

A 20-kL sample of a 1% hydrogen standard (in an Ar/He/N2 mixture) was analyzed and, from the hydrogen peak areas of the standard and the sample, the hydrogen content of the gas from blister was calculated to be about 65%. Considering the accuracy of the sampling and analysis techniques, it is reasonable to assume that a large fraction of the gas in the blister was hydrogen. The gas in a blister in the fluoropolymer coating was also sampled and analyzed in the same way. Hydrogen was detected in this sample also. Discussion The following data obtained from the four techniques show that acid is diffusing through the coatings to react with the steel surface: (1)The inhibition of cathodic delamination is a consequence of the neutralization of the hydroxyl ion (generated by oxygen reduction) by hydrogen ion diffusing to the interface (Figure 3). (2) The “leveling”effect of acid exposure on the tensile adhesion is caused by the acid that diffuses to the interface, reacts with the steel, and destroys the initial bonding between the coating and the steel (Figure 4). (3) The time required for blistering to occur is a strong inverse function of the concentration of acid used for exposure (Table I) and is affected by dilution of the acid with alkali-metal ions (Table 11). (4) Large quantities of hydrogen are found in the blisters formed after exposure to acid. These observations enable the formulation of a mechanism to explain the blistering in these coatings and the reason for their failure to protect the underlying steel against corrosion during exposure to dilute sulfuric acid. The mechanism is illustrated in Figure 6. Hydrogen ions and water diffuse through the coating, and the resulting aqueous phase provides the medium for the corrosion reaction. The iron is oxidized to the solvated ferrous state, and the hydrogen ion is reduced to hydrogen gas. The gas pressure developed by the hydrogen causes the coating/ substrate bond to break at the periphery of the aqueous phase, and new sections of the substrate are exposed to the aqueous phase. The hydrogen gas collects initially at the interface. If the hydrogen gas diffuses through the coating or into the steel faster than it is formed at the interface, the coating will remain disbonded and no blistering will occur. If, however, the hydrogen forms at a rate faster than it can diffuse away from the substrate through the coating or into the steel, the pressure at the interface will increase and a blister will form. The blisters will remain and be gas-filled as long as this condition persists. The important conclusion drawn from this mechanism is that the first stage of coating failure is the development of a liquid phase, rapidly followed by a reaction at the

interface. Blistering occurs at a later time, only if the gas pressure becomes high enough. Thus, there can be complete loss of adhesion with no blistering or visual change in the coating (cf. Figure 5). An observation of Okuda (1983) is worth citing. He measured the change in electrical resistance of epoxy coatings exposed to 2% hydrochloric acid and found that significant blistering did not occur until some time after the coating resistance had decreased to its final, stable value, and 10 times later than the calculated time for the acid to permeate the coating. One of the major unanswered questions related to the overall corrosion process is the role of the sulfate ion. Charge balance must be maintained, and the diffusion of hydrogen ions through the coating requires that an anion also must diffuse through the coating. The only anion available in the sulfuric acid solutions is sulfate, and its size suggests that it will not move through the coating at a high rate. There is the possibility that aqueous pathways of sufficient diameter exist in the coatings to allow sulfate ion diffusion. Future work using radiotracer sulfur incorporated in the sulfate should be carried out to define more clearly the role of the sulfate ion.

Summary The blistering observed when a fluoropolymer and epoxy coating are exposed to dilute sulfuric acid solutions is caused by the diffusion of acid through the coating, and the acid reacting rapidly with the steel to oxidize it and reduce the hydrogen ion to hydrogen gas. This reaction significantly reduces the adhesion of the coating by rupturing coating/steel bonds at the interface. If the generation of hydrogen occurs faster than its removal by diffusion through the coating or into the steel, a blister forms. Acknowledgment We are indebted to David Angst of AT&T Bell Laboratories, Allentown, PA, for performing the GC/MS analysis; to the Electric Power Research Institute, Palo Alto, CA, for supporting this work; and to Barry C. Syrett, in particular, for many useful comments during the course of this work. Registry No. H2S04,7664-93-9; Li2S04,10377-48-7; Na2S04, 7757-82-6; KzSO4, 7778-80-5; HC1, 7647-01-0; (vinylidene fluoride).(tetrafluoroethylene) (copolymer), 25684-76-8.

Literature Cited Berger, D. M. Chem. Eng. 1978, 8 5 , 121-22. Berger, D. M.; Trewalla, R. J.; Wummer, C. J. In Corrosion Control 8y Organic Coatings; Leidheiser, H., Jr., Ed.; National Association of Corrosion Engineers: Houston, 1980; pp 178-85. Funke, W. Prog. Org. Coat. 1981, 9 , 29-46. Leldheiser, H., Jr. Ind. Eng. Chem. Prod. Res. Dev. 1981a, 2 0 , 547-51. Leidheiser, H., Jr.; Wang, W. d . Coat. Techno/. 198lb, 53, 77-84. Leidhelser, H., Jr.; Vedage, H.; Granata, R. D.; White, M. L. J . Nectrochem. SOC. 1984, 131, 1460-61. Moelwyn-Hughes, E. A. In Physical Chemistry, 2nd ed.; Pergamon: New York, 1965; p 24 (Table 5), p 859 (Table 11). Okuda, S. I n Organic Coatings, Science And Techno/ogy;Parfitt, 0 . D., Patsls, A. V., Eds.; Marcel Dekker: New York, 1983: pp 255-71. Parks, J.; Leldhelser. H.. Jr. Ind. Eng. Chem. Prod. Res. Dev. 1988, 25, 1.

Received for reuiew November 27, 1984 Accepted January 10, 1986