Document Jrw0Op2eyK364vDj5dXV1rY4B

Chemistry in Cpnada for September, 1953 blistering of exterifcjjhouse paints on wood siding has boon a troublesome problem In recent years. In order to study its cause and prevention op a laboratory scale, the authors have designed a small apparatus in which a miniature wall section can be subjected to various conditions of heat and moisture. Several aspects of the blistering problem have been investigated with the aid of this apparatus including wood species and thickness of wood; effect of back-priming and vapor bar riers; use of special primers and moisture-permeable paints; value of air vents; behaviour of varnish coatings. The mechanism of blistering is discussed. Experimental Study Of Blistering Of Paints On Wood J C. Y. HOPKINS and B. C. SMART, National Research Council, Ottawa. % 1 PEHING of exterior house ated chamber so as to maintain the three coats of ordinary linseed oil 'paint, followed by peeling, has desired external temperatures. Other house paint and the opposite face ecome a serious problem in recent work on moisture migration is re (or back) was left unpainted. In years. It is not a new problem, for ferred to in the bibliography at the order to simulate the adverse con it was studied at the U.S. Forest Pro end of this paper. ditions that affect ordinary house ducts Laboratory as early as 1927 The present authors wished to siding, the experiments were begun and by many other Investigators up reduce experiments of this type to by trying various means of apply to the present time. The subject was a laboratory scale, and in so doing, ing moisture to the back of these reviewed by Browne in 1933 |l>. The to make it easier to control the ex painted panels. Absorbent cotton most recent knowledge is summar ternal temperature. It was expected was placed against the back of sev ized in the report of a conference also that results could be secured eral panels and moistened with on condensation control held in more quickly on the small scale. water frequently so that it was al 1952<*i. These objectives were realized to a ways saturated. The paint film was The present paper deals only with satisfactory degree. apparently unaffected and the test blistering caused by moisture. There may be other causes but they are Preliminary Experiments was discontinued after 18 days with out any evidence of blistering. believed to be of relatively minor Test panels of white pine % inch A panel was fastened to the open :U importance. The mechanism of moisture blistering is not yet fully thick were painted on one face with side of a container and the con- understood. The forces involved in the passage of liquid water fnd water vapor through wood are ap parently quite complex These forces are large enough to over come the powerful adhesion of paint to wood. However, there is not suf ficient information available to pro vide a clear explanation of the phy sical process by which mechanical energy accumulates at the wood- paint interface and ruptures the bond. Early studies of the phenomenon were based on cases of blistering on existing wood buildings <">. It was f ioon found that moisture blistering could be produced at will on small test huts by maintaining certain conditions of moisture concentration and temperature v >. The huts were usually placed outdoors, hence the exterior temperature was not controllable. More recently, a large test house was built at the Univer sity of Minnesota which was completely enclosed by a refriger FIGURE 1--Both panels were in contact with liquid water at the back. Panel B (water not heated) shows no effect; Panel C {water heated slightly) is badly blistered. 35 (183)-- Chemistry in Canada for September, 1953 face of the test panel, a hollow metal cooling block is mounted over a gasket in front of the panel "*o as to leave a %-inch air space. Water is circulated through the cooling block at the desired temperature from a second thermostated bath (Figure 3). Other modifications of the appara tus 'are described in the experi mental part. Procedure Test panels of white pine 2%" x 5%." x %" were used in most of the experiments. Their moisture con tent was in the range 8-10%. Paints were applied by brushing, allowing 24 hours' drying between coats and 7 days' ageing before exposure. Film thickness was in conformity with ordinary painting practice. The painted panels were attached to the blistering apparatus as de FIGURE 2--Blistering apparatus, showing (left to right): water vapor box, gasket, painted test panel, gasket, frameplate, hollow cooling block. scribed above with the back of the panel (usually unpainted) towards the Interior of the box. Tempera talner was filled .with water so that and connected by a tapered section tures at the back and the face of the unpainted back of the panel to a metal tube, closed at the bottom the test panel were recorded. The was in continuous contact with end (Figure 2). There is an open time of initial blistering and in water under a slight head. There ing at the top of the box lor a crease in number and size of blisters was no sign of blistering during a thermometer. The lower portion of was noted. .. period of 13 days (Figure 1). the tube is offset so that it can be Room temperature was faiifag These simple experiments show conveniently placed in a heating constant at 25*C. and the temper^ that blistering does not take place bath. The outside surfaces of the ture in the box was regulated as d;? readily because of moisture alone. box are coated with asbestos-cement sired up to 60C. Evidently a temperature differential insulation before use. In some experiments it was de is necessary as well. Methods of The tube is filled with water up to sired to maintain temperatures cor maintaining a difference in temper the tapered section and immersed in responding to winter conditions, ature between the back and the a thermostated bath. The test panel i Accordingly, the whole apparatus face of the test panel were examined. is mounted against the open side of was placed in a constant-tempera Test panels were placed in a car the box with a rubber gasket and a ture refrigerated room (5'C.) and bon arc water-spray weathering ma metal frame-plate. It is not neces the temperature in the box was held chine with the back towards the sary that the joint be air-tight. at 25"C. arc. Blistering occurred in two or three days. Payne cups <"> were used in an other procedure and the test panels were made from single ply birch veneer 0.01 inch thick, painted on one side. A little water was placed in the cup, the painted wood speci men was clamped on with the paint ed side out and the cup was placed on a heat source at about 60C. Blis tering took place quickly, usually in The bath temperature Is regulated so that the temperature in the box remains at the desired level. Evapor ation of the warm water In the tube raises the moisture vapor content of the air in the box. The conditions are therefore similar to those exist ing in the wall space of a building when excessive moisture is present. When it is desired to control the temperature at the outside (painted) Experimental Results Rapidity and extent of blistering was found to depend on several fac tors including the temperature dif ferential between the back and the face of the panel. In many experi ments blistering commenced within 24 hours and was profuse after 48 hours'. The newly-formed blisters always contained water. Systems one or two hours. Closer control of the temperature was desired. This was achieved by TABLE I using glass Jars instead of the Payne Rate of Blistering of Paint on Panels of Various Thicknesses cup. The test panel was fastened over the mouth of the jar and the jar was immersed to the neck in a water bath at controlled tempera Thickness of panel Temp. Diff. 25* - 5 = 20*C. 8hrs. Extent at. Blistering 24 hrs. 48 hrs. 72 hrs. ture. The bath was maintained at a few degrees above room tempera ture. Blistering was produced in 8 hours at a temperature differential 14" V none none none none slight ' trace none none marked slight trace none marked marked marked trace as low as 3C., using the thin veneer panels. It was more rapid at greater temperature differences. TABLE H Finally an all-metal apparatus was designed, with which temperatures on both sides of the test panel could be controlled. Apparatus' The apparatus consists essentially Effect of Back-priming Coating on Back of Panel Extent of Blistering 2 days 7 days Uncoated Aluminum-varnish, one coat Alumihuin-varnish, two coats White lead primer, one coat marked none none slight very marked none none marked ` of a metal box, open on one side. White lead primer, two coats none * none --36(184)-- GLD3 Chemistry in Canada for September, 7953 that did not blister In 14 days were iound to be practically immune to blistering even on much longer ex posure. The following ratings were' em ployed to describe extent of blister ing: none, trace, slight, marked, very marked. A. Reproducibility.--Test panels pre pared and exposed In triplicate showed good agreement in rapidity and extent of blistering. B. Nature of test panel.--White pine panels of varying thickness were painted with three coats of linseed oil house paint (white lead 70, zinc oxide 20, asbestine 30) and attached to the blistering apparatus. Blister ing was rapid on the thin panels and slower on thick panels but other wise no difference in results was ap parent (Table I). Panels of white pine, basswood, birch sapwood and birch heartwood, % * thick, were painted identically and exposed at a temperature dif ference of 40-25 <,=15cG. All blis tered within 48 hours to approxi mately the same extent. C. Type of paint coating.--Oil paints, enamels and varnishes were applied in three coats to white pine panels M* thick and exposed at a temperature difference of 20C. The linseed oil house paints blister ed most readily, usually In 24-48 hours.- If allowed to age for one month or more, blistering was re tarded. Enamels were less susceptible to blistering. Some commercial archi tectural enamels showed no blister ing in 14 days. However, when the 7-day ageing period was omitted and the panels were exposed after 24 hours' drying, two enamels out of three developed marked blistering. Three commercial exterior var nishes were exposed. Two showed no blistering in 14 days while the third developed one blister with no fur ther damage. The temperature dif ference was 20C. 0. influence of primer*.--Two coats of linseed oil house paint (white lead 70, zinc oxide 20, asbestine 10) were applied over three different primers on %" birch panels. After 7 days' ageing, the panels were ex posed at 20 C. temperature difference, with the following results (Figure 4): Self-primed--blistered in 2 days. White lead primer (zinc-free)-- blistered in 5 days. Aluminum-varnish primer -- no blistering in 7 days, slight blis tering in 10 dsjys. Several commercial primers for ex terior use on wood were tested in combination with the same leadzinc topcoat. There was some varia tion in resistance to blistering but all permitted some blistering of the system. E. Effect of tsack-primtng. -- The effect of painting the back of the wood was determined, using birch panels. Three coats of lead-zinc oil paint were applied to the face of the panels. The backs were coated as shown in Table n. After ageing, the panels were exposed In the blis tering apparatus for 7 days at 20C. temperature difference. In a further experiment, a panel was' painted as before and backprimed with aluminum-varnish but leaving a one-inch wide uncoated strip across the back. Upon expo sure, blistering took place rapidly opposite the uncoated strip but not elsewhere on the panel (Figure 5). F. Special paints. -- The effect of formula changes on 1 resistance to blistering was examined. (a) Paints containing diatoraaceous earth. Three paints were made up hav ing white lead and zinc oxide as the prime pigments with a diatomaoeous earth as extender. A comparison paint was made with asbestine ex tender. The pigment proportions by weight are given In Table III. TABLE III Pigment Mixtures White lead ' 60 55 55 60 Zinc oxide 30 20 15 30 Asbestine -- -- -- 10 Diatomaoeous earth 10 25 30 -- The ratio of pigment to linseed oil was 68:32 by weight in each paint. White pine panels, ^4-inch thick, were given three coats of each paint on successive days and allowed to age for 7 days. Upon exposure in the blistering apparatus at a temperature differ ence of 20C,, all four paints showed slight blistering in 24 hrs. and mark ed blistering In 3 days. There were no appreciable differences among the paints. (b) Highly pigmented paints. The effect of increasing the ratio of pigment to oil in linseed oil paints Was examined. A pigment mixture of white lead 60, zinc oxide 30, as bestine 10 (by weight) was ground in linseed oil in varying proportions. When tested as in (a) above, the results shown in Table IV were ob tained. Pigment: oil ratio by wt. 68:32 73:27 78:22 83:17 TABLE IV Blistering Test; Lead-zinc Paints 1 day . Extent of Blistering 2 days 4 days trace trace none none slight slight none none marked marked slight none 7 days marked marked slight trace FIGURE 3--Mistering assembly (test panel removed) shewing coding block, water vapor box and heating bath, and large eeld water bath with circulating pump. ----3/ (!35)-- GLD38186 Chemistry in Canada for September, 1953 FIGURE 4--Influence of primers. A, aluminum-varnish, t, white lead. C, selfprimed. lead-iinc-inert top coat on all. TABLE V Composition of Clear and Pigmented Coatings (Parts by Weight) Glyptal G2466, solids Glyptal G2466, solvent Bakelite VF1249 (a) Raw linseed oil Mineral spirits Driers White lead Zinc oxide Titanox RC Asbestine Paint No. 1 2 3 4 49 49 21 21 100 100 12 12 15 15 33 50 50 35 35 100 100 15 15 56 70 100 12 15 3 Paint No. 1 Pigmented 2 Pigmented 3 Pigmented 4 Pigmented 5 Clear 6 Clear TABLE VI Extent of Blistering, 3-day Test Clear and Pigmented Coatings Drying Time of Coating 4 days 7 days 15 days 47 days very marked marked marked marked none (7-day test) none none " none " marked none slight none A lead titanium pigment mixture was formulated as follows (parts by weight): titanium-calcium pigment (type RCHT) 50, white lead 35, as bestine 15. This pigment was ground with oil and thinner: pig ment 67, raw linseed oil 10, bodied linseed oil 6, mineral spirits 17, drier. The paint was tested as in (a) above except that panels were made up with one and two coats respectively. There was no blister ing on any of the panels in a 12day test. The pigment: oil ratio in this paint is 81:19 by weight. (c) Clear and pigmented coatings. A group of finishes was prepared to compare the blister resistance of two different vehicles (a) unpigmented, and (b) pigmented. The first was a mixture of alkyd resin with raw linseed oil. The second was a phenolic varnish. The com position of the clear and pigmented coatings is shown in Table V. (a) Composition of this vehicle is given as phenolic resin 16 parts, linseed oil 32 parts, mineral spirits and driers 52 parts. The paints were applied in 3 coats to the usual white pine panels. Duplicate sets of panels were allow ed to dry for various periods of time. They were subjected to the blistering test at a 20 temperature difference. The results are shown in Table VI. Although the dear alkyd-linseed oil coating is blister-resistant, it ap 38 (186)-- pears to lose this property when pigmented. The enamels (3 and 4) had good resistance to blistering, in spite of the heavy pigmentation. G. Effect of a vapor barrier.--Some experiments were carried out with a miniature hollow wall section as the test piece in order to study the effect of vapor barriers (Figure 6). Both the inner and outer panels were of white pine. When the inner panel was un painted and the outer panel painted with ordinary linseed oil paint, blis tering occurred Quickly under the usual test conditions. Upon opening up the test piece, liquid water was found In the wall space. The -inside surface of the inner panel was then painted with two coats of linseed oil paint as a water vapor barrier. Under the same test conditions, there was no blistering in 7 days and no visible moisture accumulation in the wall space. A thin sheet of waxed paper .was not effective as a vapor barrier. It proved to be quite permeable to water vapor and early blistering of the paint ensued. A heavy waxed membrane prevented blistering over a period of 7 days. It was found that a small break in the paper resulted in total loss of effec tiveness. H. Effect of ventilating the wall space. -- Miniature wall sections were employed for tests in which the wall space was vented by holes drilled through the outer panel. The inner panel was unpainted in'these tests. A single hole of 14" diameter in the outer panel had no apparent ef fect on the incidence of blistering. Under the usual test conditions (20 temperature difference) blisters ap peared In 72 hours on the test panel and on an unvented control. Simi larly, two holes, one near the top and one near the bottom of the outer panel, did not prevent blister ing (Figure 7). j; Effects of condensible vapors other than water. -- Several organic liquids were tried in place of water in the blistering apparatus. The usual test procedure was followed, using a white pine panel coated with linseed oil paint and a 20"C. temperature difference, for a 7-day period. Methanol as the test liquid pro duced slight blistering in 5 days. A control test with water gave very marked blistering In the same time. Eith&nol, hexane, ether, ethylene chlorohydrin, turpentine and miner al spirits did not produce any blister ing In 7 days under the same con ditions. K. Other experiments. -- A trans parent rubber membrane was stretched over the face of a wood panel so as to act as a coating in place of a paint film. When attached GLD38187 to the blistering apparatus, liquid water accumulated progressively be tween the membrane and the panel, having passed through the wood. Painted test panels were weighed before and at intervals during the blistering test to determine the amount of water absorbed by the wood. An air-dry painted panel (moisture content about 7%) weigh ing 18.5 gm. increased in weight to 37.3 gm. in 7 days. Thus its final moisture was over 100% of the ori ginal bone-dry panel weight. By weighing the test panel at in tervals it was observed that blister ing of linseed oil paint always oc curred when the moisture content of the wood exceeded 30%. How ever, there was blistering at lower moisture percentages in some cases and it was not possible to establish a threshold value. Discussion of Results Sufficient testing has been done to show that the simple blistering apparatus produces rapid blistering of ordinary oil paints on wood. The results are reproducible although the variability of the wood itself has some effect. Blistering did not occur in these tests unless there was a higher tem perature at the back of the test panel than at the painted face, even when the wood was saturated with water. Blistering occurred more rapidly on thin panels than on thick ones and the rate increased with the tem perature difference between the back and front of the panel. A few trials were made with different spe cies of wood but no relation was observed between species and rate of blistering. Linseed oil house paints had much greater tendency to blister than did enamels or varnishes. The resis tance of enamels to blistering has been observed by Vannoy <">. Cer tain enamels and varnishes did not blister at all under the conditions of our test but it is not yet certain that the conditions are as severe as those encounered in practice or that a period of weathering might not in crease the susceptibility of the coat ing to blistering. The same reserva tion applies to the favorable results obtained with aluminum primer (varnish pigmented with alumin um). This primer was more resistant than the conventional linseed oil white-pigmented primers. Painting the back of the test panel (back-priming) was effective in re tarding blistering. Two coats of aluminum-varnish on the back of the panel prevented blistering for the duration of the tests. Oil paints containing up to 30% of diatomaceous earth in the pig ment mixture did not resist blister-' ing. Increasing the pigment :oil ratio of a lead-zinc formula reduced its blistering tendency. A matt coat ing containing white lead and titan Chemistry in Canada for September, 1953 FIGURE 5--Panel 3, back-primed with aluminum-varnish. Panel 2 treated similar ly but a 1-inch strip at tne middle net back-primed. ium dioxide with a pigment:oil ratio of 81:19 by weight did not blis ter in these tests. The value of a vapor barrier on the warm side was readily shown using a miniature wall section. Two coats of linseed oil paint on the inner face of the wall prevented blistering of the exterior paint during the test. Ventilating the wall space by means of holes drilled through the outer panel did not prevent blister ing under the test conditions. (Fig ure 7). Apparently much of the water vapor must have condensed on the back of the exterior panel before It could escape through the vents. Mechanism of Blistering The experimental results do not as yet give a full understanding of the mechanism of blistering. 'They do indicate the following points: (a) No matter how wet the back of the siding, blistering does not occur un less there is a temperature differ ential. This suggests that the mech anism is not simple osmosis but rather a movement of vapor as well as liquid, (b) Blistering is not ob served unless there is liquid water on the back of the test panel. At least two theories have been advanced to explain blistering. Ac cording to one theory, the liquid ---39 (187)-- water that has condensed on the back of the siding travels through the wood capillaries by osmosis. Upon reaching the outer face it ex erts strong pressure against the paint and raises blisters. According to the second theory, which we favor, the partial pressure of water vapor is higher on the in terior of the wall than on the ex terior because of the temperature difference; as a result, water vapor travels through the wood toward the outside. Upon reaching the out side surface of the wood, it con denses because the temperature is low and the paint coating is rela tively impermeable. Thus there if liquid water trapped between thi wood and the paint. As it increase in volume it raises the paint 1: blisters. This theory is supporte by the apparent dependence on temperature difference and by th observation that newly-formed bli ters always contain liquid water. There are several features th; are not clear, however. If the mee anism depends merely on water v pour movement, why does blisterii occur only when the wood is ve wet throughout? Presumably this because the moisture vapor tra: mission rate increases with the m sture content of the wood. Experiments with liquids otl GL D38188 Chemistry in Conodo for September, ? 953 FIGURE 6 -- Schematic cross-section of blistering apparatus with miniature wall section on test. A, opening for thermometer. B, water vapor box. C, water level. 0, inner wood panel (warm side), E, outer wood panel (cold side). F, wall space. than water did not produce blister ing, except to a small extent with , methanol. Accordingly, there may be some effect related to the absorp tion and swelling of wood by water that influences the result. Conclusions The tendency of a paint coating on wood to blister under any given circumstances can be determined readily by means of the laboratory apparatus described in this paper. It has been demonstated that there is little or no tendency " to blistering unless a temperature dif ference exists between the inner face and the outer face of the paint ed wood. The experiments did not reveal any simple way to prevent blister ing of paint on buildings. The ten dency to blistering was reduced by back-priming, by use of aluminumvarnish primer under exterior paint, by heavily pigmented paints, and by use of enamel or varnish instead of oil paint. However, these pro cedures are attended by certain dis advantages and it is not suggested that they will prevent blistering entirely. It is judged that every possible effort in building construction and maintenance should be directed to wards preventing the entrance and accumulation of moisture in the sidewalls. Acknowledgement The authors are indebted to W. G. Humphreys for assistance in the ex perimental work. References (1) Brown, F. L. Some causes of Blister ing end Peeling of Paint on Bouse Biding. Report RO. 0.8. Forest Products tab. (1833). (8) Condensation Control in Buildings. Conference Report Mo. . TJB. Build ing Research Advisory Board. (1932). (3) H&sl&m, 3. B. and Werthan, S. Studies in the Painting ol 'Wood. Xnd. Xng. Chem. *3, 327 (1831). (4) Joy, P. A. Basic Concepts of Water Vapor Migration. Technical Paper Mo. 88. Pennsylvania State College. (1831). (5) Pfalssner, P. 'M. Plow of Gases and Water Vapor Through Wood. Can. 3. Research 28A, 388 (1930). (8) Bartwig. O. R. Circular 333. Katl. Paint Varnish and Lacquer Assoc. (1928). --40(18*)-- (7) Browne, F. L. Testing Bouse Paints. 3. Chem. education 10. 528, (1833). (8) Biff. J. W. and Davis, R. B. Mechan ism of Moisture Peeling. Ind. Eng. Chem. 31, MOT, 144& (1838). (8) Gardner, B. A. and Van Beudceroth, Circulars 380 and 597, Matl. Paint Var nish and lacquer Assoc. (1838-1840). (10) Rowley. P. B. and Lajole. M. H. Some Causes of Paint Peeling. Engineering Bulletin Mo. 30. University of Mlnne- iiota (1848). (11) Vannoy, W. G. Blister-resistant House Paints. Amer. Paint. J. M, No. 32. p. 82, (1852). (12) Condensation Control. Bousing and Home Finance Agency, Washington, D.C. (1830). (13) Hutcheon, N. 3. Bouses for Cold Weather, National Research Council, Ottawa. (1951). (14) Miller, W. T. and Morse. F. B. Mois ture Problems In Homes. Engineering Bulletin No. 70. Purdue Unlv. (1950). (13) Dunlap, M. X. Condensation Control tn Frame Buildings. U3. Forest Products Lab. (1830). (It) Teesdale, L. V. Remedial Measures for Building Condensation Difficulties. Report K1710. V.B. Forest Products Lab. (1847). (17) Payne, B. F. and Gardner, W. H. Permeability of Varnish Films. Ind. Sng. Chem. 28 893 (1937). (18) Wray. R. I. and Van Vorst, A. R. Permeability of Paint Films to Mois ture. Ind. Eng. Chem. 25 342. (1933). (19) Chisholm, T. H. Bow to Select Paints for Maintenance Work. Manuf. add ,, Ind. Xng. May, 1844, p. SO. MOTE--This paper was presented to the Protective Ooatlngs Division at the 3Sth Annual Conference of The Chemical In stitute of Canada. Montreal, June 2-4, 1832. Manuscript received June, 1933.* , GLD38189