Document Gm6k45606Q4eV2km875Njbdrv

t; also results in a whitened The best example is lacquer, which before about 1948 for themselves a bad name ber Of their rapid gloss loss and whit- nitrocellulose decomposition Jjy proceeds as follows: fulose nitrate, in the presence of m hydrolyzed to cellulose and 1 jusly oxidized to liberate cer- This removal of vehicle exFsurface pigment, the resulting surface having less gloss. The N02 in the gases evolved, as well as atmos pheric oxygen in the presence of light, is capable of oxidizing the cellulose to celluronic acid, the major solid deteri oration product identified. As cellulose and celluronic acid are both somewhat water-soluble, the action of rain and dew is to further expose the surface pigment and continually renew the vehicle surface for further deterioration. This material has been presented in the nature of a preliminary report since further work remains to be done. It is hoped that this information will be of use in the development of finishes which better resist photochemical de terioration. It is also hoped that further papers covering more details of this work will be published. Acknowledgment: Suggestions and contributions of members of the General Motors Re search Laboratories is hereby gratefully acknowledged. surface ith the herein anism is as |55 Photooxidative Degradation of Alkyd Films By E. B. FitzGerald ptf I h e foremost objective Jjtite present work was to develop infor* in&libn that would lead to the synthesis oLmore durable alkyd resins. As a : second objective it was expected that methods of measurement and testing would be developed that could be used to estimate the durability of new poly mers and finishes in shorter time and more reliably than any of the presently available methods. The investigation started with a qualitative examination of the gross physical effects that result from ex posure of a protective coating to out door weather or, in general, to photo oxidative conditions. It proceeded with an attempt to measure these effects quantitatively in relation to the con ditions that produce them. Finally, an effort was made to establish some of the chemical mechanisms that are involved. The results, which follow, will be pre sented in this same general order. Materials and Methods The resin used in most of the work was a typical automotive alkyd of 50 per cent oil length (6 parts linseed to 1 part tung oil) and an acid number of 18. ^hen used with no modification other than a common drier, this will be re ferred to as "clear alkyd." Pigment was used in some of the experiments and will he specified where it occurs. Films were also prepared from fattyacid esters of glycerol and pentaerythrifel* These were obtained from the d is c u s s io n o f t h is paper th ^yVlTED, either for publication or for J; attention of the author. Address all com^uincations to ASTM Headquarters, 1916 Kce St., Philadelphia 3, Pa. j q * boldface numbers in parentheses refer Paper ^ re^erences appended to this kly 1955 The physical effects of degradation in alkyd films have been quantita tively related to the conditions and chemical mechanisms that produce them. Hormel Institute as part of a coopera tive research program sponsored by the Paint and Varnish Production Clubs. Their synthesis has been reported else where (l).1 Other materials will be de scribed where they occur. Films were sprayed or doctor bladed on glass slides or steel panels using a solution of the resin in hydrocarbon solvent. Unless otherwise designated, all films were baked at 125 C for 20 min and stored for one month before use. Free films for measurement of mechanical properties were prepared by spraying and baking the clear alkyd on steel panels coated with polytetrafluoroethylene. The films were then lifted from the panels, cut to required dimen sions, and stored for one month prior to use. Four light sources were employed in the work: (1) a 125-w Hanovia quartz mercury vapor lamp; (2) a 360-w Uviare quartz mercury lamp; (3) a Westinghouse 40-w fluorescent ultraviolet lamp, and (4) a 1000-w water-cooled General Electric AH-6 lamp with Pyrex jacket. The spectral energy distributions from these sources were measured with a quartz monochrometer-thermopile-am plifier arrangement and will be speci fied where pertinent. Whenever filters were used in combination with one of these sources, their absorption spectra were measured, and the spectral dis tributions of light actually reaching the specimens were calculated. ASTM BULLETIN Results and Discussion Gloss Loss.--Deterioration of a finish which is exposed to natural weather can manifest itself in various ways. Cracking, blistering, and peel ing, for example, are important as pects of the problem, but in the present work, major attention has been given to the phenomena that re sult in disappearance of the lustrous surface which characterizes an unex posed finish. An electron microscope view of this effect, popularly known as gloss loss or, in advanced stages, as chalking, is shown in Figs. 1 and 2. From simple, qualitative observation, it has been concluded that the loss of gloss is concomitant with the appearance of pigment at the surface. This con clusion is borne out by electron- diffrac tion studies of pigmented enamels at various stages of weathering. Thus, a fresh, high gloss surface--shown in Fig. 1--shows no evidence of pigment under electron diffraction, while a chalked surface--shown in Fig. 2--gives a clear, strong pattern characteristic of the pig ment. Examination at intermediate stages of weathering has shown that the first perceptible appearance of a pig- E. B. FITZGERALD, Fabrics & Finishes De partment, E. 1. du Pont de Nemours & Co., Inc., Experimental Station. Wilmington, Del., has been studying photooxidative degradation of organic films since 1950. (TP137) 65 Exposure. merit diffraction pattern occurs before any change can be detected by the un aided eye. These patterns appear at about the same time standard reflection instruments are showing the first per ceptible loss in gloss and also about, simultaneously with the first appear ance of the protrusions seen in the electronmierograph. Effect of Light Intensity--Experi ence has shown that exposure to light is a necessary condition for the degradation processes that lead to loss of gloss and chalking. Since this clearly implies a photochemical process, it becomes of interest to ex amine the relationship between the rate of gloss loss and light intensity. A schematic diagram of an experi mental arrangement to determine this relationship is shown in Fig. 3. The specimens consisted of 4 by 3-in. steel panels coated with the standard alkyd resin containing 20 per cent rutile pig ment by volume. A set of four panels could be exposed simultaneously to radiation from a Uviarc lamp at wellseparated levels of intensity; intensity at each level was established by prior measurement using a calibrated photo cell device. Water at 25 C was circu lated over the panels, since no other way of maintaining a known and constant, temperature at the panel surfaces could be devised. Light reflection at the airwater interface was calculated from the Fresnell equation and appropriate cor-- rections were applied. Constancy of lamp performance throughout the ex periment was determined by periodic photocell measurements. The panels were taken from the water periodically and gloss was measured using the du Pont, Universal Gloss Meter (2). Results, averaged from several experiments, are shown in Fig. 4. Taking the reciprocals of the times re quired to reach various levels of glossloss, it was found that the average rates so obtained were well represented by Eq 1, where R represents gloss units per hour, E is relative incident light intensity and k is a constant. This is shown in Fig. 5. ...............(1) It will be shown later that this relation ship with light intensity is identical to Fig. 3.--Schematic Diagram of Apparatus for Exposing Films to Various Light. In tensities. 66 (TP138) ASTM BULL'ETIN that obtained for the rate of evolution of volatile decomposition products. Effect of Wavelength.--Investigation of the relationship between wave length and degradation is much more difficult than .the intensity effect. In the present work, test panels, coated with the same enamel that was used in the intensity measure ments, were exposed under various filters to the radiation of a Hanovia lamp. Exposures were carried out, under dry, open air conditions, simul taneously and at uniform distance from the lamp by use of a turntable arrange ment. Under these conditions the panels maintained a uniform tempera ture of about 40 C, Relative values of the light distribution incident upon each panel (Table I), were calculated from the measured output of the lamp and the transmission characteristics of the filters. It was found that spectral output of the lamp changed throughout its lifetime, and the only remedy found for this was to replace the bulb at inter vals during the experiment. Gloss loss of the panels, measured periodically, is shown in Fig. 6an effort to unify and correlate these re sults in the simplest possible manner, it was assumed that if the average rate of gloss loss up to 20 units of loss were to be taken from Fig. 6, then Eq 1 would still hold and the .relation including wavelength X might be represented by some function of E\'*/\ n summed over July 1955 ft .IE R t h 1. t r c i n j I ution ation veore tfiect. mels, that ,sure,rious novia out, ;imulfrom ange- the iperares of upon dated lamp ics of ectral hout found inter- sured . In. :se reinner, e rate were .vnuld uding d by Iyer 1955 all wavelengths given in Table I. Accordingly, Eq 2 in which A and B are constants, was found to be in approxi mate agreement with the data, as shown in Fig. 7. reasonable and suggests that the aver age rate of gloss loss (up to 20 units loss) in monochromatic light is propor tional to the specific extinction. This conclusion may be applicable to films other than alkyds, for it is in general agreement with the results given by Long (3), who found that the durability of a variety of films could be correlated with the area under a transmission curve between 3000 and 4000 A. Ex ceptions would be found in the case of materials containing structures of very low light transmission and very high resistance to photolysis. Surface Erosion.--Up to this point, no attempt has been made to define the means by which the chemical proc esses of photolysis result in the geo metrical effects of surface roughness and eventual appearance of pigment particles. It has been implied, how ever, that matter is being lost by the film through photo-initiated scission reactions that lead ultimately to vola tile decomposition products. It would appear simple to detect loss of material as loss in weight, but experiments have shown that, under certain circumstances the weight loss incurred up to the point of chalking may be too slight to measure accurately. For example, if it is as sumed that a measurable amount of gloss loss or chalking would be produced by loss through volatilization of the uppermost 0.05 n of the organic film material, the corresponding weight loss for a reaction confined strictly to the surface, would be 0.2 per cent in a 1-mil film. By the use of interferometry, the disappearance of such small amounts of material can be determined more con veniently and accurately than they can by weighing. The interferometric method can be applied very simply. In the present work a sharp line of demarcation be tween an exposed and unexposed film area was produced by laying a doubleedged razor blade on the film and po sitioning an ultraviolet lamp vertically above it. Upon exposure, the unpro- Equation 2 represents merely an at tempt to express empirically the re lationship between gloss loss and wave length. It is nevertheless instructive to compare this equation with one de rived for the ultraviolet transmission on a clear film of the same resin. The specific extinction, fc, of the clear alkyd is shown in Fig. 8, together with a plot of the relation: *=dlogT= where d is film thickness and C and D are constants. Although there appears to be some deviation from Eq 3 in the region of shorter wavelengths, the agreement throughout most of the range is quite July 1955 Fig. 5.--Relationship Between Rate of Gloss Loss and Incident Light Intensity. AST M BULLETIN (TP139) 67 TABLE I.--RELATIVE ENERGY DISTRIBUTIONS FROM HANOVIA LAMP WITH , VARIOUS CORNING FILTERS. carried out on the standard aJkyd mented with 25 per cent-rutile and com Wavelength^ No Filter Filter Filter Filter Filter Filter pared with gloss measurements, it ^as m#* \ , No. 791 No. 986 No. 970 No. 774 No. 733 found that the relationship between de i 205.............. 0.15 crease in thickness and gloss loss de 215.......... .. 225.......... J.. 235............ .. 24$................ 0.45 1.00 1,70 2.45 6.05 0.22 0.82 1.62 ** . pended upon the wavelength distribu * . tion of the light source used for the exposure. 255................ 205............ .. 4.70 5.75 3.57 4.77 6. is 0.75 0.05 0.29 With direct radiation from the Han- 275................ 2.40.. 2.09 0.70 0.38 ovia or Uviarc lamps, the ratio of thick 285................ 295................ 305___ :......... 2.05 5.60 7.70 1.85 5.10 7.10 0.98 < 3.41 5.54 0.72 3.25 5.77 6.22 0.84 ness decrease to gloss loss was smaller than that obtained when the light from 315................ 9.70 9.04 7.46 8.05 2.32 these lamps was filtered by a glass, such 326................ 335................ 0.30 1.50 0.28 1.39 0.23 1.18 0.27 1.37 0.12 0.79 as Corning No. 970 (Table I), although 345................ 0.15 0.14 0.12 0.14 0.10 6162 in the latter case the change took place 355................ 365................ 375......... 0.30 18.60 0.46 0.28 17.30 0.43 0.23 14.00 . 0.29 0.28 17.10 0.43 0.24 14.90 0.38 0.20 12.10 0.38 much more slowly. Decrease in thick ness can, of course, result from increase 390................ 405............ 420................ 0.20 4.30 0.20 0.18 4.00 0.18 0.07 0.20 3.96 0.18 0.18 3.90 0.18 0.18 3.90 0.18 in density, and measurements on the clear alkyd after very drastic exposure 435.;;.......... 7.05 7.55 . 6.50 7.41 6.41 indicated that shrinkage up to 3 per 495................ 545................ 0.25 5.12 0.23 4.76 0.23 4.72 0.23 4.65 0.23 4.65 cent can be accounted for in this way. 560............. 0.15 0.14 0.14 0.14 0.14 It follows that thickness decreases of 575................ 585 and up.. 4.00 1.30 3.72 1.21 3.68 1.20 3.64 1.18 3.64 1.18 the magnitude shown in Figs. 9 and 10 could be attributed entirely to change tected film decreased in thickness rela tive to the protected part and a step was formed at the boundary; the height of the step was then measured by means of a Tolansky microinterferometer (4, 5, 6). Figure 9 shows typical fringe shifts obtained with the clear alkyd upon exposure to the unfiltered Hanovia lamp and Fig. 10 gives the result of a series of such measurements plotted against time. When measurements of this type were in density unless this change were meas ured and shown to be negligible or un less the initial film thickness was so small that the decreases upon exposure exceeded the maximum attributable to density effects. Both of these procedures were applied '5 alky bine in the present work and decrease in face thickness due to net loss of material was Rad lam dial bull; seve 4 Har desp the cien vola face l08S: lam; fom witl for deej roug A dire info of of t to sing all COD; use< outf alkj the guic limi wer h one ana alb 68 (TP140) ASTM BULLETIN July 1955 Jl- ao6 (l/X*} 25 50 75 3-0 k IO c / Unexposed "cribu- -Jjg| >r the Igll Han- thickmailer |||i > from i, such hough place thick- !| crease 1 n the posure 3 per i way. 1 ip ses of j! md 10 ;hange measor un^as so il ;n- r.p 4il: posure file to pplied ase in alwas i that rented h i \ ( ,,.yj f |00 \cident 0.05 ao4 w 0.03 0.02 4 - 250 350 400 X, millimicrons Fig. 8.--Specific Extinction of Alkyd Film. alkyd result when a sufficient depth of binder has been eroded from the sur face by conversion to volatile products. Radiation from the unfiltered Hanovia lamp attacks the surface so rapidly that ' chalking becomes evident before the bulk of the film has had time to undergo severe degradation. With the filtered Hanovia lamp, attack is slower and, despite the shielding effect of pigment, the radiation penetrates the film suffi ciently to cause degradation and loss of volatile products from beneath the sur face. Thus, for equal extents of gloss loss, the film exposed to the filtered lamp will have lost more matter in the form of volatiles. The same situation with regard to loss of volatiles obtains for clear film, but the effects penetrate deeper and, although the surface is roughened, true chalking cannot occur. Analysis of the Film.---The most direct and obvious way of getting information about the chemical nature of degradation is through analysis of the film before and after exposure to photooxidative conditions. No single method is capable of yielding all the necessary information and consequently several techniques were used in this work. One of the most outstanding characteristics of baked &lkyd films--their insolubility in all but the most drastic solvents--not only guided the selection of techniques but limited the effectiveness of some that wore attempted. Infrared spectroscopy proved to be ne of the most generally informative analytical methods. Films of clear nlkyd were cast and baked on rock salt; Clear 8-Hr Exposure 2.0 Unexposed Clear 16-Hr Exposure 500 Unexposed Clear 24-Hr Exposure Unexposed Pigmented 24-Hr Exposure Fig. 9.--Interference'Fringe Shifts Produced by Exposure. 0.02 1.00 0.75 e 0.50 0.25 * 40 60 80 Exposure , hr Fig. 10.--Decrease in Film Thickness Resulting from Exposure. 1955 jly 1955 ASTM BULLETIN (TP141) 69 100 120 V 20 40 60 80 100 120 Exposure, hr 0L Fig. 12.--Changes in Functional Group Concentrations Resulting 24 6 8 10 12 14 16 from Exposure. Microns Fig. 11.--Infrared Spectra of Films Before and After Exposure. III and IY have not eliminated unsatu ration during the bake, the decrease in spectra were taken before exposure to photooxidative conditions and at vari ous periods throughout exposure. For the purposes of the present work, only a few of the absorption maxima appear ing in the unexposed film (top of Fig. 11) need be identified. The bands at 2.9, 3.4* and 5.8 /j represent hydroxyl groups, aliphatic carbon-hydrogen bonds, and carbonyl groups respectively. The doublet at 6.3 and the bands at 8.9, 9.3, 13.5, and 14.2 n are characteristic of phthalate esters. The bands at 6.9 and 7.2 are due to --CH2-- and --CH3 respec tively, while the deep band at 7.8 arises from carbon-oxygen-carbon links of the ester. There was no evidence of olefinic unsaturation in any of the spectra. The weak bands found from 10 to 10.5 m also occurred in alkyds modified with pure stearic acid and therefore cannot be attributed to unsaturation. The changes in composition that were produced by exposure depended upon the nature of the irradiating source, as might have been expected from the work on gloss loss and erosion. Certain qualitative similarities were observed, however, and a typical example pro duced by exposure to the General Elec tric AH-6 lamp (intense radiation above 3000 A) may be seen in the spectra of Fig. 11. Major structural changes bon-hydrogen links become fewer, the rate of decomposition of the new oxy genated groups exceeds the rate of for mation and their concentration de creases. Finally, when the carbonhydrogen content has diminished to about 10 per cent of its original value, the film becomes much more stable. Formation of oxygenated structures can, of course, take place at sites of original unsaturation in the oil as in reactions I to IY (7): O2 --CH2CH=CH-- --CH(OOH)CH=CH--.. .(I) --H20 --CH(OOH)CH=CH------------ --COCH=CH--.. .(II) --CH(OOH)CH==CH-----* --CHOH--CH--CH---- - \/ O --CHOH CO CH2--. . .(Ill) --CH(OOH)CH=CH---- --CHO + --CH2CHO.. .(IV) Oxidation occurring only at sites of original unsaturation cannot, however, account for the gross changes that are observed. Even if reactions such as R'OH carbon-hydrogen is so extensive that random or repetitive degradation proc esses capable eventually of attacking any or all of the oil hydrogens must be postulated. This does not imply that the degradation is unrelated to the original unsaturation for, in fact, an alkyd synthetized from pure stearic acid showed practically no change in composition, even after prolonged expo sure. Degradation of a drying oil alkyd apparently starts during the bake with the formation of hydroxyls, hydroper oxides, and polyketones at the sites of original unsaturation. Exposure to light results in photolysis of these groups followed by complex oxidation mech anisms in which similar groups are generated and subsequently photolyzed along the remainder of the oil molecule. During the period of rapid carbonhydrogen disappearance, the spectra also show a slower decrease in phthalate content of the film. Appearance of phthalic anhydride at the surface of alkyd films or, for that matter, in un opened cans, has been noted under cer tain circumstances and has been attrib uted to reactions such as V: 0 4- R'OR C--OH evident in these spectra are analyzed quantitatively in Fig. 12. Here, the immediate and rapid decrease in ali phatic carbon-hydrogen content must be attributed to replacement of drying oil hydrogen atoms by hydroxyl, hy droperoxide and carbonyl groups. As the degradation proceeds, and the car 0 0 ,A-c/O + ROH o ..(V) 70 (TP142) ASTM BULLETIN July 1955 t^the'present work, evidence in favor ^h'otooxidalive loss of phthalate was 1 in several ways: films exposed to lag degrees of degradation, as lo oted by weight losses up to 25 per were analyzed chemically for fphthalate ester content using the method f af'Shreve and Heether (8). In every the oxidized film contained the !;: weight per cent phthalate (within \-4he limits of experimental accuracy) as the original resin. Also, exposures to the direct radiation of a Hanovia lamp ^ere made on films prepared from an gjkyd analogous to the standard clear resin, but possessing carbon-14 in the phthalate carboxyl position. Complete elementary analyses were made on these films, and C02 obtained from combustion was precipitated as barium carbonate and analyzed. Results of this work (Table II), indicate a small preferential loss of phthalate. It is also apparent esulting that loss of hydrogen during exposure to the Hanovia lamp was accompanied by loss of carbon in nearly the same unsaturease in /e that n proctacking nust be ly that to the act, an / stearic i |ge in \j. expoil alkyd ke with droper- atomic ratio as existed in the original film. In the dark, however, hydrogen was lost relatively faster than carbon. \ : It is concluded that oxidation J v whether photo or thermal initiated, be< ; gins in the oil part of the molecule and j spreads to the glyceryl and phthalate :i) ester structures such that, on the averi { age, complete alkyd monomeric units ( ) are disappearing together. This sug gests that phthalate content of the oxi]; dized film may be applied to the infra red analyses as an approximate internal standard. Consequently, replotting the data of Fig. 12 relative to phthalate ie sites sure to i groups content should give an indication of the changes in functional group concentra tions corrected for decrease in film . mech- thickness (Fig. 13). lps are itolyzed olecule. carbonspectra ithalate ince of face of , in un der ceri attrib- As mentioned earlier, chemical anal ysis of the film is greatly handicapped by its insolubility; however, chromato graphic analyses of benzene-alcohol extracts of exposed film did reveal the presence of diketones and quinones. Attempts to estimate ethylenic unsatura tion by hydrogenation and other means gave erratic results. Qualitative tests, however, were most instructive. For example, films partly exposed- to the Hanovia lamp and partly protected, would, upon immersion in aqueous po tassium permanganate, acquire a layer of manganese dioxide upon the exposed area, while the unexposed area remained clean. Similarly exposed films im ... .(V) mersed briefly in aqueous iodine- Potassium iodide and then in a starch i solution turned blue on the unexposed ? afea, while the exposed area remained colorless. One possible explanation for j ! the formation of unsaturation during 1955 \ ; July 1955 TABLE II.--ELEMENTARY ANALYSIS AND RADIOASSAY OF CARBOXYL-TAGGED ALKYD FILMS. Exposure Carbon, per cent None.............. 72 hr direct Hanovia ultraviolet ...... 72 hr at 120 C in dark..................... 64.00 63.75 62.76 Hydrogen, per cent 7.28 7.32 6.70 Oxygen, per cent 28.72 28.93 30.54 Ratio, Hydrogen to Carbon 1.37 1.37 1.28 Relative Activity per Weight, Carbon 100 94.4 95.2 photolysis will be offered in the next section. Analysis of Volatile Products.-- Gravimetric analyses of volatile prod ucts were carried out using apparatus similar to that ordinarily employed for semimicro elementary analysis; details and procedure have already been published (4). Alkyd films of known area and weight were placed in a quartz tube and exposed to filtered or unfiltered radiation from a Hanovia lamp in a gentle stream of oxygen or purified nitrogen. The effluent gases were converted to carbon dioxide and wg.ter, caught in absorption tubes, and weighed daily for ten days. One of the most significant conclu sions of this work resulted from compari son of the hydrogen-carbon atomic ratio in the off-gases to that in the parent material. When films were exposed to the unfiltered radiation of the Hanovia lamp in oxygen, the hydrogen-carbon ratio in the gases was approximately 2.5 (compared to 1.35 in the film), while use of filtered radiation (Corning No. 970) gave ratios that ranged from 5 to 7, al though, of course, the gases were evolved much more slowly. The lower ratios produced by the unfiltered radiation are in agreement with the conclusions ob tained from interferometry to the effect that short wavelengths result in total removal of organic material from the film surface. High hydrogen-carbon ratios produced by filtered radiation confirm the infrared and elementary analyses quoted previously by demon strating that dehydrogenation is one of the important initial effects of photoly sis in the bulk of the film. Another significant result of the gravi metric work was realized by plotting the total weight of gases evolved in a given period of time versus the weight of resin per unit area. Such a plot is shown in Fig. 14, which is a composite of results obtained with various light sources. It is important to note, also, that when plotted in this way, data for pigmented and clear films fell on the same lines. The dotted line in Fig. 14 represents the quantity of gases that would be evolved upon total combustion of the film to car bon dioxide and water. Thus, the graph indicates that clear films thinner than about 0.3 mg per sq cm disappear completely in 100 hr of exposure to the unfiltered Hanovia, while with pig mented films, only the pigment remains. Similarly, it is concluded that a thicker film exposed to the unfiltered Hanovia for the same period of time loses approxi mately 0.3 mg per sq cm from its sur face and, in addition, loses material in Fig. 13.--Changes in Functional Group Concentrations Relative to Phthalate Content. ASTM BULLETIN (TP143) 71 Fig. 14.--Gases Evolved from Alkyd Films During 100 Hr of Exposure to Various Light Sources. microscope slides were placed inside the cells, where they could be exposed to radiation from a Westinghouse fluores cent ultraviolet lamp through the cell wall without obstructing the infrared beam used for analysis. A typical result obtained in these ex periments is shown in the spectrum (Fig. 15) of gases produced by a dry film of pentaerythritol eleostearate after 5 days' exposure in an oxygen atmos phere. The relationship between light intensity and rate of gas evolution was determined by placing the cells at vari ous distances from the light source, where the relative intensities were meas ured by a photocell; as in the chalking experiments. The optical density of the carbon dioxide absorption band was determined periodically and converted to pressure by calibration. Hate calcu lations from these data gave the same functional relationship to intensity (Eq 1) that was obtained for the rate of chalking. The effect of wavelength distribution was studied by placing filters over the cells; output of the lamp and transmis sion of the filters are given in Fig. 16. the form of volatiles from beneath its surface. Figure 14 also shows that a film ex posed to radiation from which some of the short wavelengths have been filtered loses a smaller amount of material from its surface, while the loss from beneath the surface is the same as that experi enced by a film of identical thickness exposed to the total radiation. Finally, films exposed to a lamp producing in tense radiation in the visible and infra red but probably nothing less than 4000 A evolve volatile degradation prod ucts in a manner characteristic of a simple bulk reaction. The amount of chalking produced from these three exposure conditions was in agreement with that found in the studies of the wavelength effect; films exposed to the unfiltered Hanovia chalked severely in 100 lir, while those exposed to the filtered Hanovia chalked slightly and those exposed to the heat lamp gave no detectable loss in gloss. Changes in thickness calculated from Fig. 14 agreed satisfactorily with results obtained by interferometry. A quite different and, in certain re spects, more informative means of studying volatile products was accom plished by the use of the infrared spec trograph. Exposure cells were con structed from 10-cm lengths of 5-cm (outside diameter) borosilicate glass tubing equipped with side arms and stopcocks; rock salt windows'were ce mented to the ends. Films coated on 72 (TP144) Fig. 15.--Infrared Spectrum of Gases Evolved from Dried Film of Pentaerythritol Eleo stearate Upon Exposure to Light. ASTM BULLETIN July 1955 - / - / / }ution er the asmisig. 16. 60 - 40 / // a 5 Used 1955 5 10 15 20 Exposure, days Fig, 18.--Carbon Dioxide Evolution from Dried Films of Drying Oil Esters Upon Exposure to Light. L_________________1_________ --i--------------------------- 1---------------------------- 1_________________ O 10 20 30 40 50 Exposure* doys Fig, 17.--Carbon Dioxide Evolution from Dried Film of Pentaerythritol Eleostearate Upon Exposure to Various Light Distri- butions. compressed downward (except for the saturated ester which remained con stant), but the same relationship be tween the rates was maintained. If gas evolution is, in fact, a result of Resiilts for dried films of clear penta- and it was suggested that initial attack ketone photolysis, then the reaction erythritol eleostearate are given in Fig. by light consists of photolysis of ke should proceed--or at least commence-- 17, which clearly shows the somewhat tones formed during the bake. Fur in the absence of oxygen. In order to surprising importance of the reactions ther evidence in support of this view was test this point, two identical dried films due to longer wavelengths. Qualitatively sought by infrared analysis of the vola of pentaerythritol eleostearate were ex it would appear from Figs. 16 and 17 tile degradation products arising from posed to the Westinghouse lamp in the that the rate of gas evolution from clear various model compounds during expo infrared cells. One cell was filled with films is less dependent upon wavelength sure. oxygen and the other with nitrogen; than is the rate of chalking in pigmented Air-dried films of pentaerythritol both cells were evacuated and refilled films. Intuitively, it might be supposed esters of eleostearic, linolenic, linoleic, with their respective initial atmospheres that longer wavelengths are just as oleic, and stearic acids were exposed in at approximately two-week intervals, effective as the shorter ones in producing the infrared gas cells with an oxygen while their contents--between inter volatile products but that their effect is atmosphere. Area and thickness of the vals--were monitored by infrared. The distributed through a greater depth of films, distance from the Westinghouse results (shown in Fig. 19) suggest that film. Additional work will be needed fluorescent lamp, and all other exposure two reactions are involved: (a) photol to clarify this point and, in particular, conditions were held constant. Carbon ysis of existing oxygenated groups and the modification that is to be expected dioxide, carbon monoxide, formic acid, (6) oxidation of residual unsaturation from the presence of pigment. and water were found in every case, and or of unsaturation and free radicals In view of the complexity of alkyd no other products were detected. Also, created by the photolysis. Both re resin films, it is difficult to draw detailed the relative amounts of the various actions stopped immediately (so far as inferences regarding the mechanism of gases were approximately the same for could be detected) upon removing the deterioration from experiments on the all the films, but the rates of evolution cells from the light. films themselves. The use of compounds depended upon the amount of unsatura Exposures of stearic and sorbic acids that can serve as models for the various structures present in alkyds provides a means of overcoming this difficulty. tion originally present in the ester, as shown by the data for the carbon diox ide absorption bands in Fig. 18. When were carried out as thin layers of powder in the oxygen-filled gas cells. Here, the sorbic acid decomposed rapidly to 1 example, it was shown earlier that baked films contain little unsaturation the films were baked prior to exposure, the data corresponding to Fig. 18 were give carbon dioxide, carbon monoxide, water, formic acid, and acetic acid. I July 1955 ASTM BULLETIN (TP145) 73 Normally, however, degradation is accompanied by other effects, the most important of which is change in mechanical properties. Accord ingly, as a conclusion to the present work, a brief study was made in order to relate the observable changes in cer tain of these properties to the mecha nism already postulated. Clear and pigmented free films 2.5 mils thick were prepared as previously described and cut into strips 1 cm wide. Permanent set measurements were car ried out by exposing specimens to radia tion from a Hanovia lamp while they were maintained at 10 per cent elonga tion and 100 C in a quartz-windowed oven. Clamps were applied to the strips at inked reference lines 10 cm apart, and the strips, with their clamps, were placed in the oven for a few min Fig. 19.--Evolution of Carbon Dioxide from Dried Films of Pentaerythritol Eleostearate ^ utes prior to attachment to the stretch in Oxygen and Nitrogen. ing frame. The oven could be evac uated or maintained with any desired 'Also, the residue from the sorbic acid, a viscous, yellow liquid, was found to con tain a high concentration of hydroxyl groups. The stearic acid, by contrast, proved to be very resistant to degrada tion. From these results, and all of the fore going, it was concluded that: (a) Initial attack by light occurs at ketonic groups that have been formed prior to exposure through hydroperoxidation and rear rangements at the sites of original un and in this way spreading the destruc tion to parts of the structure not directly susceptible to photolysis. (e) The ex istence of substantial concentration of ketonic carbonyl is the most probable cause of yellowing that is observed in alkyd films upon drying or aging in the dark. Mechanical Properties.--Up to this point, the major concern has been with attempting to explain the mech anism by which an alkyd film under atmosphere during exposure. Lamps were placed 10 cm from the specimens in such a way as to provide uniform illumination. At various intervals, specimens were unclamped and allowed to relax for a few minutes while still in the oven; they were then removed and allowed to relax at room temperature until constant length, as measured by a cathetometer, was obtained. Perma nent set was calculated from the equa tion given by Andrews (12): saturation. (b) Ketonic or carboxyl carbonyls conjugated with ethylenic goes loss of gloss and chalking upon exposure to photooxidative conditions. Permanent set, per cent = x--------u- X 100 unsaturation or with other carbonyls are more readily attacked, particularly by radiation above 3000 A, than are iso lated carbonyls. (c) The initial attack does not require the presence of oxygen, but proceeds by the well-known mech anism (9) shown in reactions VI and VII: RCOR' \ hv RCOCOR'f -- R 4- R'CO -- R 4-R'4-CO....(VI) hv RCH2CH2CH2COR' -- RCH=CH2 + CH3COR/... .(VII) (id) In the absence of oxygen, the avail able carbonyls are eventually consumed and the rate of degradation is drastically reduced. With oxygen present, how ever, the radicals formed in reaction VI further react to give carbon dioxide, water, and formic acid (9, 10). Fur thermore, the lower ketone formed in reaction VII is subject to repetition of both reactions, while the olefin can pro ceed, via the usual hydroperoxidation, to give more ketone. Radicals formed in reaction VI can also combine writh simi lar radicals on nearby chains to }rield new cross-links. It is probable that these radicals are capable of abstracting hydrogen from sensitive points (such as a-methylenes and tertiary hydrogens) 74 (TP146) ASTM BULLETIN July 1955 whe u x s T esti nen ical elor bet1 con eitb exp set diff twe tha trai mei but dep nea nit] diu bet cou abs Pig coi per Bill do< nei on ca2 Jul non i8 the change -Record. Vesent 'order in eermecha.ms 2.5 viouBly n wide, sre caro radia te they clongandowed to the 10 cm clamps, iw minstretche evacdesired Lamps eeimens uniform tervals, allowed s still in /ed and jerature ed by a Permaie equa- 100 00 / 1955 Exposure, hr Fig. 21.--Intermittent Stress Relaxation. At jwhere: u -- the original distance between ref erence marks, x = the distance when the specimen V was stretched on the frame, and s = the distance after exposure and relaxation. The results (Fig. 20) give a relative estimate of the number of new, perma nent bonds (cross-links) formed by chem ical reaction while the specimens were elongated. In the direct comparison between clear films and pigmented films containing 25 per cent by weight of either rutile or anatase, the clear films experienced slightly greater permanent set than did those with pigment, but no difference could be distinguished be tween the two pigments. This means that the cross-finking reactions pene trated nearly as deeply into the pig mented film as they did into the clear, hut it is impossible to infer whether this depth was very slight or comprised I nearly the whole film. Exposures in nitrogen resulted in substantially re duced permanent set, and the difference between clear and pigmented specimens could no longer be detected. In the absence of fight, no differences due to pigmentation or type of atmosphere could be measured; it is probable that permanent set occurring in the dark re sults from a simple flow mechanism and does not depend upon the formation of &ew chemical bonds. The dependence a temperature is shown by one run carried out at 2 C. According to the mechanisms pos tulated in reactions VI and VII, chain scission reactions must be occurring simultaneously with cross-linking, but relative rates of the two processes can not be obtained from permanent set data alone. The net rate of scission and cross-linking can, however, be ob tained from intermittent measurements of the modulus of a specimen main tained in a relaxed state during exposure (II, 12). Such measurements were effected by use of a simple electric strain gage apparatus constructed for the pur pose. Test strips were cut from the same films and to the same dimensions as those used in the permanent set work; exposures to the unfiltered Hanovia lamp were carried out in air at 25 C. The stress required to elongate the speci mens 10 per cent was measured periodi cally and the ratios of stress at time t to initial stress, Ft/F,, were plotted against log time. It should be noted that the lamp was turned off a few minutes before each measurement to avoid any pos sibility that the specimen was above 25 C at the moment of measurement. Results of this work (Fig. 21) indicate a slight excess of scission processes over cross-linking at an early stage in the exposure but, following this, the rate of cross-linking increases rapidly and finally becomes one of the outstanding features of photooxidative degradation in alkyd films. It was impossible, in this apparatus, to carry out measure ments at controlled temperatures above ASTM BULLETIN 30 C. However, replacement of the Hanovia lamp by a 375-w Westinghouse infrared bulb gave cross-linking effects qualitatively similar to those obtained with the ultraviolet source. Finally, measurements were made of stress re laxation at constant elongation, but it was found that molecular flow by dif fusion was too large to permit the scis sion reactions to be isolated. Summary The initial step in the degradation of drying oil alkyd enamels occurs during the air-drying or baking stages of film formation. Oxygen attacks the unsatu ration in the oil and forms hydroper oxides. These hydroperoxides serve either to initiate the a-methylenic chain reaction, which leads ultimately to the formation of useful polymers, or they decompose in side reactions to produce hydroxyl, carbonyl, and other oxy genated groups. The carbonyl groups, especially if they are conjugated with each other or with residual olefinic un saturation, absorb strongly in the near ultraviolet and significantly at longer wavelengths. Upon exposure to sunlight, these com plex ketones immediately begin to pro duce carbon monoxide, free radicals, new olefins, and lower ketones. If oxygen is present, the free radicals are oxidized to new photolyzable compounds (which may be volatile); the newly formed olefins are subject to repetition of the hydroperoxidation process and the lower ketones continue to be photolyzed until the chain in which they originally existed is consumed. In addi tion, the free radicals may attack other parts of the alkyd molecule at random or become stabilized by reaction with other radicals. While all of these re actions start in the oil molecule, their damage spreads by repetition until the glyceryl phthalate ester is attacked and ultimately converted into volatile prod ucts. Long before the film is entirely con sumed, the effects of this complex chem ical process can be observed as physical changes in the film. Very short wave lengths, as from an ultraviolet lamp, are completely absorbed in a very thin sur face layer and repetitive reactions pro ceed rapidly through complete units of the alkyd structure in this layer. Thus, if the incident light is entirely short wavelengths, the major physical effect is an erosion of the surface, which ap pears as "chalk" if the film is pigmented or as slight dulling in a clear area; in either case, the freshly exposed surface is hydrophilic, due to its high con tent of polar, oxygenated groups. When the film is exposed to poly chromatic radiation, such as sunlight, (TP147) 75 the longer wavelengths penetrate deeper than the shorter ones and initiate reac tions throughout the bulk of the film. Here, the photolytic processes are more gradual, and dehydrogenation with the formation, of carbonyl groups and cross links becomes more important than cleavage. Consequently, hydrogen in the form of water is evolved more rapidly from the bulk of the film than is carbon in the form of its oxides and volatile compounds. Asa result, shrinkage, in crease in density, embrittlement, and yellowing occur in the bulk of the film. These physical effects are diminished by the screening effect of pigment, but radiation in the longer visible and near infrared part of the sun's spectrum penetrates to the substrate interface in the films of ordinary thickness. The extent of yellowing is dependent upon the relative rates of formation and destruction of ketone and polyketone structures. Since these rates depend, in turn, upon the wavelength distribution of the incident radiation, various light sources produce characteristic equilib rium levels of yellowing. Thus, a film that has acquired a high level of yellowing through exposure to ordinary indoor illumination, or to heat, may be bleached to a lower level of yellowness7 in sunlight or to a still lower level by appropriate ultraviolet lamps. Limited access of oxygen in the bulk of the film undoubtedly has a role in governing the sequence of reactions throughout the film, particularly in the initial stages of drying, but its effects have not been specifically studied. Water vapor appears to have little in fluence on the degradation processes, but liquid water accelerates chalking by the physical effects of swelling and solution and possibly by chemical inter action as well. A cknowledgment: Appreciative acknowledgment is ex tended to the many research staff mem bers who contributed their technical skills and advice to this work. Special acknowledgment is made to J. T. Harris Jr., and C. D. Miller for assistance with the infrared work and its interpretation. Re f e r e n c e s (1) W. O. Lundberg and J. R. Chipault, Official Digest, Fed. Paint and Var nish Mfrs., January, 1944, No. 288, p. 11. (2) S.' Hochberg and M. P. Morse Official Digest, Fed. Paint and Var! niah Mfrs., July, 1950, No. 307, p. iq ~ (3) J. S. Long, Official Digest, Fed. Pai^ and Varnish Mfrs., October, 1940 No. 297, p. 648. (4) S. Tolansky, "Multiple Beam Inter** ferometry of Surfaces and Bihna London, Oxford University preL (1948). (5) E. B. FitzGerald, Industrial and Engineering Chemistry, Vol. 45 h 2545 (1953). ' (6) W. L. Grube and S. R. Rouze, Jour nal, Optical Soc. Am., Vol. 44, p 351 (1955). (7) D. Swern, J. T. Scanlon, and H. B. Knight. Journal, Am. Oil Chemists Soc., Vol. 25, p. 193 (1948). (8) O. D. Shreve and M-. R. Heether. Analytical Chemistry, Vol. 23, p 441 (1951). (9) J. E. Guillet and R. G. W. Norrish Nature, Vol. 173, p. 625 (1954), (10) A. Finkelstein and W. A. Noyes, Jr., Discussions, Faraday Soc., No. 14 p. 76 (1953). (11) M. I. Christie, Journal, Am. Chemi cal Soc., Vol. 76, p. 1979 (1954). (12) R. D. Andrews, A. V. Tobolaky, and E. E. Hanson, Journal, Applied Physics, Vol. 17, p. 352 (1946). Acid Contamination as a Source of Error in Boiling Nitric Acid Test for Corrosion-Resistant Steels By Robert J. Bendure Th e boiling nitric acid test, ASTM Recommended Practice A 262,1 has been found useful for meas uring the quality of stainless steels with respect to their ability to resist attack by nitric acid and as an indication of the effectiveness of heat treatments that may have been employed or of ad justments of composition that may have been made to accommodate other wise damaging heat treatments. Ex cellent reproducibility between .dupli cates is normally obtained, except in those instances where the penetration NOTE.--DISCUSSION OF THIS PAPER IS INVITED, either for publication or for the attention of the author. Address all com munications to ASTM Headquarters, 1916 Race St., Philadelphia 3, Pa. 1 Tentative Recommended Practice for Boiling Nitric Acid Test for CorrosionResisting Steels (A 262 - 52 TA 1952 Book of ASTM Standards, Part 1, p. 998. * R. J. Bendure, "Contamination of Nitric Acid as a Source of Error in the Huey Boil ing Nitric Acid Test," Corrosion, Vol. 10, p. 7 (1954). 76 (TP148) High and erratic penetration rates were obtained when the testing area inadvertently contained hydrofluoric acid fumes. rates are quite high. In such cases it is not unusual for duplicate specimens to show marked variation in corrosion rates. This test has been used in our research laboratories for a number of years, the testing being carried out in a laboratory in which chemical analyses are also made. The heating apparatus used consists of two multiple burner gas-fired hot plates placed side by side so that twenty-four 1000-ml wide mouth Erlenmeyer flasks can be accommodated simultaneously. Each flask is equipped with a finger-type condenser, the cooling water being taken from a mani fold-type distributor. ASTM BULLETIN For most types of material the rate of attack follows a similar pattern--the first period rate may sometimes be slightly higher than the second period rate, following which each of the suc ceeding three periods will have rates as ROBERT J. BENDURE, Chemist, Research Laboratories, Armco Steel Corp., Middletown, Ohio, since 1942, is supervisor of a group of analysts whose work has included extensive acid corrosion testing of stinle*s steels. July 1955