INDUSTRIAL AND ENGINEERING CHEMISTRY
868
Vol. 23, No. 8
Paint Thinners I-Effect of Different Thinners on the Durability of House Paints in Outdoor Exposure Tests’ F. L. Browne FOREST PRODUCTS LABORATORY, MADISON, WIs.
KE r e a s o n u s u a l l y
0
Results of experiments o n the effect of p a i n t t h i n n e r s
of drier necessary in paint is not
on the durability of coatings of house p a i n t s a r e regiven by painters for affected by the choice of thinner, the verity of this theory ported. T h e k i n d of t h i n n e r proved to be a m i n o r preferring turpentine is doubtful; the effect of turpenfactor in durability. Nevertheless, coatings were to other thinners for house tine on the dispersion of paint definitely m o r e durable w h e n t h e paint h a d been paint is the belief that turpenpigments may well prove more t h i n n e d w i t h a deliberately oxidized t u r p e n t i n e that tine makes c o a t i n g s more important. T u r p e n t i n e apparently causes the white pigleft a considerable residue o n evaporation than when durable, especially on those ments to flocculate more than t h e p a i n t was t h i n n e d w i t h ordinary t u r p e n t i n e or softwoods in which the bands petroleum or coal-tar distillates w i t h petroleum or coal-tar distillates. T h e results of summerwood are wide (23, do (ZO),and the choice of thinpoint to t h e possibility that t u r p e n t i n e m a y find its 21). C h e m i s t s are not so ner may therefore affect the structure of the aggregates of most valuable use in paint a f t e r it has been converted well agreed on the s u b j e c t pigments in the paint coating. into a non-volatile product whose incorporation in (19 p. 44), although most of It is not yet known what sort them think that turpentine p a i n t coatings m a k e s them m o r e durable. of pigment structure in coat* ” has an advantage in the thinings makes for durabilitv. (4) The residue theory assumes that the non-volatile residue ning of priming-coat paint for resinoussoftwoods such as southern yellow pine. Painters and cheqists alike base their left behind when turpentine evaporates from thin films adds t o the durability of paint coatings (19 p. 32, 11, 26). Objection opinions’about the rehtive value i f thinners chiefly on theo- to the theory has been made (19 p. 32) because the amount of retical considerations. residue left by freshly distilled turpentine is very small, so that only old, oxidized turpentine should prove materially better .The following theories have been advanced: (1) The solvent theory maintains that turpentine obtains deeper penetration of paint in resinous wood because it is a better solvent than petroleum distillates for the resin supposed t o interfere with penetration (23, 21, 19 p. 44). If the theory is sound, coal-tar distillates should be still better paint thinners than turpentine, and for that reason benzene or solvent naphtha is sometimes recommended (10). The theory, however, rests on certain misconceptions. The cavities in the summerwood of softwoods, over which paint fails soonest, are not “filled with resin.” On the contrary, paint liquids penetrate the summerwood of both resinous and non-resinous softwoods more readily than the springwood, on which paint lasts longer ( 3 , 13). Hence poor adhesion of paint t o summemood is not due t o insufficient penetration of paint liquids ( 4 ) . Further, the solvent action of turpentine on fresh paint has been said to promote adhesion between coats (16), but under normal conditions of exposure paint never fails by separation between coats. The solvent theory, therefore, may be dismissed as inherently illogical. (2) The catalyst theory attributes to turpentine some of the properties of metallic driers. Turpentine does oxidize readily and is able t o transfer some of its combined oxygen to any linseed oil with which it may be subsequently mixed (19 p. 32, 11, 22). Yet freshly distilled turpentine has no observable drying action on linseed oil. In its ordinary use as a paint thinner turpentine does not replace any of the metallic drier. It is therefore difficult to see why the alleged drying action of the turpentine should affect the durability of the paint. (3) The dispersion theory holds that turpentine-thinned paints utllize their metallic driers to better advantage and dry more promptly and uniformly because metallic soaps remain more finely dispersed in linseed oil in the presence of turpentine than in the presence of mineral spirits (25). Since the amount 1 Received March 30, 1931. Presented before the Division of Paint and Varnish Chemistry at the 81st Meeting of the American Chemical Society, Indianapolis, Ind., March 30 to April 3, 1931. Note-The experiments described in this paper were first proposed in a conversation between the writer and F. P. Veitch of the Bureau of Chemistry and Soils. in June, 1926, I. F. Odell, Pine Institute of America Fellow at the Mellon Institute of Industrial Research, was planning similar tests at about the same time and it was decided to join forces in a single undertaking. H. K. Salzberg has since taken Odell’s place for the Pine Institute. Besides the outdoor exposure tests at five testing stations reported here, a .corresponding series of exposures was made in an accelerated testing device a t the Mellon Institute, together with additional exposure tests and laboratory experiments on other characteristics of paint thinners. The experiments at Mellon Institute, which still continue, will be reported subsequently by Salzberg. This paper is therefore the first of a series on the general subject of paint thinners.
than mineral spirits. There is evidence, however, that paint driers catalyze the oxidation of turpentine and consequently that turpentine leaves more residue in paint coatings than is commonly supposed (26). To verify the theory fully it must be shown that the residue is a desirable ingredient in paint coatings.
Few exposure tests of the effect of paint thinners on the durability of coatings have been reported. Undoubtedly many have been made but remain unpublished either because paint manufacturers conducted them for their own information or because of the common reluctance to publish results that are considered negative. Exposure tests at North Dakota Agricultural College (14, 17, 18) made on white pine, a wood with narrow bands of summerwood, failed to reveal any differences in durability that could be attributed to the kind of thinner. Tests by the American Paint and Varnish Manufacturers’ Association (12) indicated equal durability on white pine for paints thinned with gum spirits or with any of five wood turpentines. A former study a t the Forest Products Laboratory on southern cypress, eastern hemlock, and southern yellow pine ( 5 ) likewise showed no differences between thinners of widely different nature. None of these tests permitted close comparisons, however, because they were not made on matched boards. Differences in density and ring width between boards may have much more effect on the behavior of coatings over them than the kind of thinner in the paint. Another series of exposures made on matched boards (6) indicated that paint thinned with turpentine is slightly more durable on southern cypress than when thinned with benzene. Outline of Tests
The tests reported here were made on longleaf pine (Pinus palustris), which is a variety of southern yellow pine that often has wide bands of summerwood, high density, and high resin content. The boards used averaged somewhat higher in density than the general value for the species (16); according to current theories any differences in durability of coatings caused by the nature of the paint thinner should be most marked on wood of this kind. The boards for the test panels were matched; that is, long boards were each
INDUSTRIAL A N D ENGINEERING CHEMISTRY
August, 1931
cut into ten short pieces and these pieces were than built into sets of ten panels, each of which was coated with a paint thinned with one of the ten thinners tested. Tests were made with two paints, a straight white-lead paste paint and a lead-zinc inert prepared paint. The ten thinners included four turpentines, four mineral spirits, one painters' naphtha, and one coal-tar naphtha. All painting was three-coat work. Exposures were made out-of-doors a t Madison, Wis., Pittsburgh, Pa.. Washington, D. C., Gainesvillci, Fla., and Fresno, Calif. The panels were exposed in a position inclined a t 45 degrees to the vertical, facing south. The exposures began in the fall of 1927 and were completed a t all stations except Pittsburgh by June, 1930. Inspections, using methods that have been published elsewhere ( B ) , were made by the writer a t least once a year. Details of Tests
The four turpentines were steam-distilled and destructively distilled wood turpentine, ordinary gum-spirits turpentine, and gum-spirits turpentine purposely oxidized in the laboratory. To prepare the oxidized turpentine a part of the shipment of gum-spirits turpentine was warmed by placing the 5-gallon container in a bath of hot water and bubbling air into the turpentine through a glass tube reaching to the bottom of the container. THINNERS-A~~ the thinners were examined a t the Bureau of Chemistry and Soils, U. S. Department of Agriculture. The results appear in Table I. The gum-spirits and steamdistilled wood turpentines conform t o specification 7 b of the Federal Specifications Board (7) and specification D13-26 of the American Society for Testing Materials (1). The destructively distilled wood turpentine conforms to specification D236-27 of the A. S. T. M. (2). The oxidized turpentine falls outside the limits of the specifications for gumspirits turpentine. T a b l e I-Results of E x a m i n a t i o n of T h i n n e r s at B u r e a u of C h e m i s t r y and Soils. U. S. D e o a r t m e n t of Agriculture TURPEXTINES
I
TEST
i ~
Initial boiling point at 760 mm . . . . . . . . . . . . . . . . . . . Distilled below 170' C . at 760mm.. . . . . . . . . . Distilled below 180' C. at 760 m m . . . . . . . . . . . . . .
Steamdistilled wood
z;;zd
Ordinary gun1 spirits
Oxidized
wood
l54'C.
150OC.
154'13.
156'C.
937,
74%
98%
78%
....
967"
....
....
gum
spirits
MINERAL SPIRITS FROM CRUDE OIL FROM:
-
I
PennCalifornia sylvania
Mid-
, ~ ~ ~ ~ ,
~~
Copper strip test for 1 NegaNegaKegaPosisulfur.. . . . . . . . . . . I tive tive tive tive Initial boiling point at 760 m m . . . . . . . . . . . 138' C . 145' C . 15s' C. 149' C. 5% over a t . . . . . . . . . . 164O C . 159' C . 175' C . 164' C.! 97% over a t . . . . . . . 209O C . 204' C . 236' C . 218' C!, C O A L - T A R PRODUCT
varnishmakers' and painters' naphtha
1
IV&-
(160' solvent nanhtha)
Initial boiling point at 760 m m . . . . . . . . . . . . . . . . . . . . . . . . . . 126' C . 570 over a t . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1340 c . 9770 over a t . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 170' C .
The amounts of residue left when weighed samples of the four turpentines were evaporated from tared Petri dishes under arbitrarily chosen conditions were determined a t the Forest Products Laboratory as follows: TURPENTINE
% Oxidized gum spirits Steam-distilled wood Destructively distilled wood Ordinary gum spirits
5.9 2.1 1.5 1.3
869
When a sample of the gum-spirits turpentine was redistiiled, the fresh distillate left 0.1 per cent residue on evaporation. The figures are relative and are approximate only, because the amount of residue varies with the conditions under which the evaporation takes place. On evaporation from paint coatings in the presence of linseed oil, paint driers, and pigments, the residues probably are greater (26). The four mineral spirits were products derived respectively from Pennsylvania, Midcontinent, California, and Mexican crude oil. With the exception of the one from Mexican crude oil, they met specification 16 of the Federal Specifications Board (8). The mineral spirits from Mexican crude oil contained enough sulfur to give a positive copperstrip test but it did not cause noticeable discoloration of the paints, both of which contained white lead. The varnish-makers' and painters' naphtha was a product from Midcontinent crude oil. It had a much lower range in boiling point and a higher rate of evaporation than the mineral spirits. The coal-tar product was 160" C. solvent naphtha. PaINTs-The two paints represented common practice with paste paint and with prepared paint, respectively. The paste paint was commercial basic carbonate white lead containing 92 per cent pigment and 8 per cent raw linseed oil by weight. It was reduced for application as follows: THIRD COAT AT STATIONS EXCEPT
FIRST
Paste white lead, pounds Raw linseed oil, gallons Thinner, gallons Paint drier, gallons
COAT 100 1 5 2 0 0 125
SECOND GAINESCOAT VILLE 100 100 1 5 3 5 1 5 0 125 0 125 0 125
THIRD COATAT GAINESVILLB
100 2 0 1 0 0 25
The priming-coat reduction departs markedly from the customary recommendation of 4 gallons of linseed oil and 2 gallons of thinner. The increased proportion of thinner is thoroughly conventional, however, since the wood to be painted was longleaf pine, but the high content of pigment is unusual. I n the writer's opinion it had much to do with the conspicuously better senrice given in these tests by the paste paint. The second- and third-coat reductions conform to customary practice. The special third-coat reduction a t Gainesville was designed to reduce discoloration of white-lead paint by lichen and sooty mold. I n that respect it was a t best only partly successful. The prepared paint was ground a t the Forest Products Laboratory. The pigment paste was reduced for thirdcoat application as is done in manufacturing prepared paint except that the thinner was withheld so that the chosen thinner could be incorporated later. To make third-coat paint the required amount of the desired kind of thinner was added. To make second-coat paint there was added 1 pint of thinner per gallon of third-coat paint. To make priming-coat paint 3 pints of thinner were added per gallon of third-coat paint. A prepared paint does not permit reduction to the heavily pigmented priming and second-coat paints possible with paste paint. I n that respect the prepared paint was a t a distinct disadvantage in these experiments as compared with the paste paint. The composition of the prepared paint, after addition of the amount of thinner necessary for the third coat, was: % b y VI. Pigment (66% by w t . ) : Basic carbonate white l e a d . . . . . . . . . . . . . . . . . 45 Zinc oxide, lead-free.. ..................... 40 Silica . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . S Barytes., ................................ 7 Vehicle (34% by. wt.): Raw linseed o i l . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9 0 Liquid paint drier.. ....................... 5 Thinner.. ................................ 5
Figure I-Effect of Thinners on Durnbility of Paste P a i n t at Galneaville. Fla. Paint for panel G-872 was chinned with gum-spirits turpentine, p = n e l G 3 7 3 with oridiaed turpentine and panel 0-374with mineral spirits. piece of board is distinctly flat grain in panel a 7 2 but tends to become more nearly edge grain m pan& 0 3 7 4 .
This composition lies within the limits of specification 106 of the Federal Sriecifications Board. in effect a t the time these tests were started (9), and conforms closely to the great majority of paints that the Government was receiving under that specification. I n its characteristic behavior it differs from that of straight white-lead paint 8s much as any paint could that fd1 within the limits of tlie specification. The specification has since been changed in a way that excludes paint of this composition. The constants of the raw linseed oil used for grindiug the prepared p i n t and for making tlie reductions with both paints were not determined. Its acid number may have been higher than that desirable in paint containing zinc oxide. If so, the prepared paint was put a t a further disadvautage as compared with the paste paint. TEMPmms-The longleaf-pine lumber was in the fonn of hoards 1 by 6 inches hy 16 feet, surfaced on four aides, of a select grade, arid air-dry. Most of tlie boards were flat-grained. Many came froni srnall trees or from the ceuters of logs so that the annual rings curved sharply, and thus some boards presented partly Aat-grain and partly edge-grain surfaces:. A few had pith through a portion of their length, making the surfaces near one end mostly edge-grain while near tlie other end they were predominantly flat-grain (Figure 1). The hoards were sorted into groups of three, each of which provided t,lie lumber for one set of ten panels corresponding to the ten thinners. For each exposure stat,ion except 13th burgh thrro were two sets of pariels, oiie set lor paste paint and the other for prepared paint. The three Iwards for a set of panels were each cut into ten pieces 18 inches long. The thirty cuttings were assembled into ten panels, each of which had one piece from each board. The pieces were distributed among the ten panels so that successive cuttings from the boards were coated with paints thinned as follows: tiesiruciively distilled wood fiirixnlirir Steam-distilled w o d turpentine
Gum-rpirits turpentine Oxidized am-spirits turpentine &ziueral rpintr. Midcontinent Oil Ksphlhr from Midcontinent crurle oil Mineral spirits, l’ennsylvimia oil Solvent naphtha frnm cod tar Minerd spirits. Califoinia Oil Minerd spirits, Mexican di
Alternating the turpentines with the other thinners, instead of grouping them together a t one end of the boards, would lave improved this layout greatly. The three pieces composing test panel were placed side
Left-haod
by side and held firmly together by cleats nailed ou tlie back. The concealed edees of the nieces remained uncoated. Racks. ends, and edges o f the cokpleted panels were protected by two coats of aluminum paint made with long oil spar varnish. About 11 per cent of the pieces were predominautly edgegrain. In assembling the pieces no effort was made to turn them all bark side out; consequently about 48 per cent of them, distributed at random, were pit.h side out. The inferiority of the pith side of flat-grained boards war not appreciated when these tests were started. With some woods t.his shortcoming in the layout would have proved fatal to the object of the test, but boards of soutliern yellow pine do not develop loose grain easily on the pith side unless subjected to mechanical shock. Only three pieces, all from one board, developed loose grain, so that the va1idit.y of the conclusions drawn is not seriously impaired by this wcrsight. The panels for Pittsburgh were designed differently. Four i s f o o t hoards were eacli cut into ten pieces 12 inclics long. The ten pieces from each board made a set of panrls. T m sets, oiie for each paint, were exposed on the roof of the Mellon Institilk. The other two sets xwre tested in 511 aect~lerated hating device, the results of which arc not included in this report. These panels were painted on backs, ends, and edges with two coats of aluminum paint. Painting and Exposing the Panels
All painting w t s iionc at tlie Forest I’mclurta Lalioratory 1st arid Septeiniicr, 192i, and all esposnres were oven~ber14, 1‘32i. In the preparhtion, all knots were first. sealrid witli 4.5-pound ciit orange siiellac vmiisli to whicli liad been added 4 per cent by w i g h t of castor oil. Tliia sealer prevented excellently the early flaking of coatings over knots, although even with it.s help suoli coatings finally cracked characteristically. Pitch streaks were not coated with the knot sealer. Painting was donc indoors. One week was alloB,ed between coats for drying. The panels were not exposed outdoors to sunlight between coatings. The weight of paint applied to each panel was rccorded. The finished panels were exposed outdoors 011 test fences, inclined at an angle of 45 degrees from the vertical and faring south. At Pittsburgh the panels were held in racks on the roof of the Mellon Institute. Inspections
The writer recorded eight inspections a t Madison, three each a t Washington, Gainesville, arid Pittsburgh, and two at Fresno. Independent inspections at Pittsburgh and Gainesville were made by H. K. Salzbers. Exposures were con-
l.VDLTSTtllALd.TD ENGINEEEIKC? CHEMISTRY
August, 1931
871
FiCure 1 Effect of Thinnera en Durability of Prepared Paint a t Gsinesville, Fla. panel G.J@ was thioiied with gum-spirits turpentine, panel G 3 S 3 with oxidized turpentine, panel G-884 with mineral spirits ~
Paiiit fur
tiiiiii:il until tlic ~,anr!Is were rated “bad” i n integrity of coating.ytliat is, iintil much of the summerwood was left bare (Figures 1 niid 2). At some stations much of the springwood coated with prepared paint was also bare a t the cnd
,4 t.nd ... t.hn ”
and spreading rate of white-lead paint were the same whether it was thiimed with turpentine or with mineral spirits. With the lead-and-zinc paint, however, the turpentines made mixtiires I(?ssfluid than those of the mineral spirits.
““l.l
The mritcr followed his dynamic method of evaluating paint stwice (6). In addition, since the tests were on miitched boards, he made static comparisons of the panels within each set. Because of the 45-degree aiigle of exposure, the dense pine lumber, the severity of the climate toward paint at some of the stations, and the high content of zinc oxide in the preparcd paint, deterioration of the coatings, especially those of the r w a r e d paint, took place more rapidly than is desirable a t stations that could be visited only once a
A v . P ~ N T P a o a n ~ ~EXX-OS e No. OF PAN&I.S hrruao OP Av, Lhx 00 100 $0. .fl.
PAINTS
w h i t * lend:
~ , ~ ~ Lead and in^: ~
~
~
~
20 ~ 20 r i ~
20
20
t ~
s
7.37
-0.07 E O ox
7.41 5 02 4~ 76
~
-0 06 - 0~. 0 4
&
Observations of the effectof tile tiiinricrs tile consistency led to the following coIic~usions($4):
of the
V
