Document LLLBXrLw4d6YpmN50904Y923

378 INDUSTRIAL AND ENGINEERING CHEMISTRY VOL. 32, NO. 3 for the unexposed ester and the line across the bottom of the chart represents the transmission after 8-hour exposure of the wet film to the radiation of the mercury arc. If the film is permitted to dry in air before the mercury arc exposure, the absorption spectrum iB little different from that of the un exposed film, with the exception of the disappearance of the conjugation band at 10.0 microns and broadening of the 8.4-mioron band. The development of high opacity when the film is exposed to the mercury arc radiation before drying iB believed to be due to polymerization of the ester to large molecular aggregates which effectively scatter the infrared radiation. Normal air drying modifies this polymerization, either because of the general rigidity developed in the molecule or through deactivation of the atomic groups which function in the polymerization reaction. Similar results are obtained when wet films of the glyceryl eleostearate are baked and compared with normal air-dried films. This polymerization effect on baking or mercury-arc exposure of the wet films is shown by glyceryl trilinolenate and glycol dieleostearate as well as by glyceryl trieleostearate, but by none of the other esters. It thus appears that a high degree of functionality in the molecule is necessary in order to obtain it. Acknowledgment The authors' thanks are due to J. G. Smull of Lehigh Uni versity for the preparation of the esters of the four drying pil fatty acids and to John Fleming who determined all the infrared absorption spectra. Literature Cited (1) Barnes, R, B., Froo. 5th Summer Conf. Speotrosoopy, Mass. Inst. Tech., p. 71; Rev. Sci. Inst., 7, 266 (1936); J. Chem. Education, 15,26(1938). (2) Gamble, D. L., and Barnett, C. E., In d . En g . Ch b m., Anal. Ed., 9, 310 (1937). (3) Gillette, R. H., and Daniels, F., J. Am. Chem. Soc., 58, 1139 (1936). (4) Herman, R. C., and Hofatadter, R., J. Chem. Phye., 7, 460 (1939). (5) Leoompte, M., in Grignard's "Treatise on Organio Chemistry", Vol. II, Paris, Masson et Cie, 1936. (6) Morrell, R. S., J. Soc. Chem. Ind., 56, 795 (1937). (7) Pfund, A. H., S. Optical Soc. Am., 23, 375 (1933). (8) Stair, R., and Coblentz, W. W., Bur. Standards J. Research, 15, 295 (1936). (9) Weniger, W., Phye, Ret., 31, 388 (1910). Pr k b e n t k d before the Division of Faint and Varniah Chemistry at the 98th Meeting of the Amerioan Chemical Sooiety, Boston, Mass. Formation and Deterioration of Paint Films SEVERALyears ago one of the authors (2) de Changes in Films of Methyl scribed the results ob tained in an investigation of Esters of Several Unsaturated Therefore when it became possible to resume this work, it was decided to investigate by infrared absorption as the changes taking place in films of trilinolenic glyceride Fatty Acids under Exposure well as by chemical and other physical methods the during exposure to room conditions. This investiga to Ultraviolet Light changes taking place during exposure in ample unsatu tion had been undertaken rated fatty ester films. The in the hope that a study of a.simple unsaturated glycer J. L. OVERHOLT AND A. C. ELM present program includes a study of the methanol, ide might yield information The New Jersey Zinc Company, Palmerton, Penna. glycol, and glycerol esters of which would prove helpful oleic, linoleic, linolenic, and in the explanation of the . eleostearic acids. The use formation and deterioration of the more complex drying of infrared absorption spectra in the study of these compounds oil systems and the effect exerted upon them by pigments, is described in this issue by Gamble and Barnett (3). especially zino oxide. One: of the more important results of the investigation was the realization that compounds less complicated than trilinolenic glyceride would have to Experimental Procedure be studied if the basic reactions involved in the drying process were to be uncovered and explained., The methods of chemical analysis commonly employed in the examination of drying oils had been used to follow the changes in the tri linolenic glyceride films. Since then it has become increasingly apparent that considerable uncertainty was attached to some of the more important experimental results and that, therefore, conclusions based on them might be seriously in error. There is always the question as to whether the com pounds or groups indicated by the chemical analysis were The oleates and eleostearates used in this investigation were prepared from vacuum-distilled methyl esters which had been obtained by esterifying recrystallized fatty acids. The linoleates and eleostearates were obtained from debrominated methyltetra- and methylhexabromides which had been purified by repeated recrystallization. The constants of the esters given below indicate that they were of satis factory purity and had not been bodied appreciably in the course of preparation: actually present in the film or whether they were formed in the course of the analysis under the influence of the reagents used. Physical methods of testing offer a possibility of answering this, question. Of the physical methods available, infrared absorption spectra Beem to offer the most promise. Ester Methyl ole&te Methyl linoleate Methyl linolenate Me.thyl eleoetearate Aeid No. 8.9 2.0 None 2.7 WlJ8 Iodine Value 81.5 173.2 258.8 183.3 Saponi fication No. 19&2 191.9 Visooeity, Poises o.os 0.05 0.02 0.2 MARCH, 1940 INDUSTRIAL AND ENGINEERING CHEMISTRY 379 Co n d it io n s o f Ex p o s u r e. In an investigation which is expected to yield information on the behavior of drying oil films under the influence of normal atmospheric surroundings, a great variety of exposure conditions might logically be chosen. In the present case it was desirable to amplify ex posure conditions as much as posable even at the risk of ob taining data of limited practical value in order to facilitate interpretation and correlation with infrared absorption measurements. It was therefore decided to limit exposure conditions as far as convenient to ultraviolet light. This exposure factor was preferred because of its importance in the formation and deterioration of oil and paint films. The cabinet constructed for use in this investigation consists essentially of a circular table, 30 inches (76.2 cm.) in diameter and rotated at 1.2 revolutions per hour, and a 6-inch (15.2-cm.) horizontal type "Uviarc" mercury arc light mounted 20 inches above the center of the table. The light draws 6-6 amperes at' 110-125 volts and is screened by a piece of 100-mesh cooper screen to reduce its intensity so that the changes in the oil nuns can be followed conveniently. The oil samples are exposed in thin films in Petri dishes placed near the periphery of the circular table. The whole is mounted in a constant-temperature room kept at 66 per cent relative humidity and 25 C. (77 F.). At more or less regular intervals samples of the exposed esters were taken and analyzed for change in weight, degree and type of unsaturation, degree and type of oxidation, ultimate analysis, molecular weight, viscosity, density, and refractive index. The method used in determining the change in weight needs no description, since it is well known and frequently practiced in the industry. An accurate knowledge of the reactions taking place during the formation and de terioration of paint films is necessary for a better understanding of the mechanism of failure and the development' of improved exterior house paints. A study of the dry ing of trilinolenic glyceride carried out several years ago, although it yielded some interesting results, had failed to furnish the desired information perhaps because of the complexity of the drying oil com pound investigated. It was then felt that a systematic study of simple drying oil esters should be undertaken to furnish the fundamental knowledge necessary for a satisfactory explanation of the drying and aging processes. Such a study is now under way. The changes taking place in the methyl esters of several fatty acids when exposed in thin films to ultraviolet are discussed in this paper. Unsaturation The methods commonly employed in measuring the degree of unsaturation, however, are of questionable value when ap plied to oxidized oils or films. Marshall's method (<?) which appears to yield satisfactory results in the presence of per oxides is affected by acidic compounds. To correct for this error, a blank determination is made parallel with the deter mination according to Marshall's method; the oil sample and all reagents are used except the iodine chloride solution. Although in this manner more plausible values are being ob tained, negative values for unsaturation continue to be ob served occasionally, especially with samplesjwhich have been exposed for a long period and do not completely dissolve in carbon tetrachloride. Since negative iodine addition values seem illogical, an effort is being made to find the Bource of this error. There is some evidence that perhaps it is to be found in the absorption and occlusion of halogen solution by the film substance which is too slowly desorbed during the first titration step to be titrated completely. The iodine addi tion values shown in Figure 3 were obtained by Marshall's method corrected for free acids. Conjugation The question of the behavior of conjugated double linkage systems is of particular interest. A study of the ohanges tak ing place in eleostearates was expected to yield data of con siderable value, since the oxidation of conjugated systems may be expected to differ in some essential points from the oxidation of systems of isolated double linkages. Maleic anhydride addition is commonly used as a measure of the degree of conjugation existing in an unsaturated compound. Athough this method may yield reliable information in the analysis of raw tung oil, the results obtained with it in the examination of oxidized oils must be interpreted with a great deal of caution. The best features of the analytical procedures proposed by Ellis and Jones (I) and by Kaufmann (6) were combined into a method which yielded satisfactory results with 0.1-gram samples. Fig u r e 1.. Ca l ibr a t io n o f t h e Ro l l in g Ba l l Vis c o s it y Ap p a r a t u s i Oxidation Peroxides were measured by a slight modification of a method proposed by Hamilton and Olcott (4), and hydroxyl groups were determined by a method of West, Hoagland, and Curtis (S), slightly modified to adapt it to the small-size samples available for this investigation. To obtain an approxi mate measure of the rate of decomposition of the various esters on prolonged exposure to ultraviolet light, aldehydes were measured according to Hamilton and Oloott (4). Acid and ester values were determined on the same sample to con serve material. After the acid number titration is com- 380 INDUSTRIAL AND ENGINEERING CHEMISTRY VOL. 32, NO. 3 previously pointed out (B), any molecular weight determina tion involving solution of film material is of questionable value because of the danger of decomposition and depoly merization. Therefore, it seemed desirable to search for some other method which might yield an indication of the relative molecular size of the esters after exposure and oxi dation. Since viscosity measurements of solutions have been used extensively for this purpose, it appeared that a similar measurement applied directly to the film material itself might prove of considerable value. Such a method was de scribed by Wolff and Zeidler (9) and is based on the speed of a steel ball rolling over an inclined glass panel coated with the oil film. The instrument was calibrated empirically at 65 per cent relative humidity and 25 C. by the use of a series of kettle-bodied linseed oils whose viscosities had been carefully determined in a vacuum plastometer. When the results ob tained with these oils were entered on logarithmic coordinate paper, plotting time against the sine of the angle of inclination of the glass panel, a series of straight lines was obtained which permit the conversion to poises of the reading obtained with an oil of unknown viscosity. The calibration chart shown in Figure 1 is based on the use of a Vs-inch (3.18-mm.) steel ball, a film spread by a 0.004-inch (0.1016-mm.) spreader, a rolling distance of 4 cm., and a temperature of 25 C. The viscosity in poises is obtained from the point of intersection of the inclined line of the test oil with the broken vertical line. If the glass panel is set at the angle indicated by the viscosity line (approximately 22.5 ) , the rolling time in seconds is numerically equal to the viscosity in poises. Refractive Power As pointed out in the introduction, optical measurements offer certain advantages of particular value in the study of drying oils. In the hope that correlation of the data collected in this phase of the investigation with the infrared absorption spectra might be facilitated, refractive index determinations were also made. Refractive index measurements, or rather the specific or mole refraction of organic compounds, have frequently been used to aid in the interpretation of chemical data and as an indication of structural changes. The refrac tive index of the esters was determined at 25 C. with an Abb6 refractometer and a Mazda daylight bulb as the light source. The density measurements necessary for the calcula tion of the specific refraction were made with the aid of a micropycnometer. This pycnometer has a capacity :of only about Vs ml. and yields very reproducible results. Reading from top to bottom: Fig u r e 2. Ch a n g e in We ig h t ; Fig u r e 3. Io d in e Va l u e s ; Fig u r e 4. Ma l e ic An h y d r id e Ad d it io n Va l u e s pleted, an excess of alcoholic potassium hydroxide solution (0.1 normal) is added, the mixture is refluxed .until saponifica tion is complete, and the unused potassium hydroxide is back-titrated in the usual manner. To obtain a final check on the composition of the oxidized esters, they were analyzed for carbon, hydrogen, and oxygen. Polymerization An attempt was made to determine the rate and degree of polymerization by cryoscopic molecular weight determina tions using benzene or fert-butyl alcohol (7) as solvent. The latter proved to.be the better solvent for the oxidized esters and yielded quite reproducible results, provided care was taken to prevent absorption'of moisture by the solvent. As Results and Discussion The results obtained with these methods are presented in the form of graphs on which the various values for the methyl esters are plotted against the logarithm of the exposure time. In general, no startling new effects are indicated, and a com plete analysis of these data will not be attempted at this time. The values of these data will be found in the fact that they reduce to a common denominator the changes taking place during a standard exposure in drying oil esters varying in type and degree of unsaturation. When these results are contrasted with similar data obtained with the more complex glycol and glycerol esters, it should be possible to obtain a clearer insight into the mechanism of film formation of the drying oils. An interpretation of such experimental data should not be attempted except in the full realization of the fact that individual sets of data may vary widely because of unavoidable or unsuspected variations in exposure conditions or analytical procedure. In this paper attention will be called only to the more important points brought out by this inves tigation. nable polyi for f the oxibeen nilar itself i deof a ; the > per s of ully i *>1> \ rnte I fcion \ ned { ned | mrt | m.) ; ler, 1 :he ion | cal : ihe ids ed in as er re al in i it jY !I MARCH, 1940 INDUSTRIAL AND ENGINEERING CHEMISTRY Reading from top to bottom: Fig u r e 5. Re f r a c t iv e In d ex Fig u r e 6. Sp e c if ic Re f r a c t io n f r o m t h e Lo r e n t z-Lo r e n z Eq u a t io n Specifio refraction = (n*""+~3) (j)' w*lere n's refraotive index and d is density. Fig u r e 7. Pe r o x id e s Fig u r e 8. Hy d r o x y l Ox y g e n Fig u r e 9. Al d e h y d e Ox y g e n The amount of weight gained and the rate of gain (Figure 2) are proportional to the degree of unsaturation, the eleostearate falling between the linoleate and linolenate. {In this and all subsequent curves the following numbering scheme is used: (1) methyl oleate, (2) methyl linoleate, (3) methyl linolenate, (4) methyl eleostearate.] On prolonged exposure the esters begin to decompose and lose weight, the rate of loss increasing with decreasing degree of unsaturation. The methyl esters occupy a similar relative position in most other properties. The iodine addition values (Figure 3) are of particular in terest since they seem to illustrate more clearly than any other property measured the three stages in which the drying apparently takes place. The induction period is followed by a rapid decrease in unsaturation during the second stage, which coincides with the maximum in the peroxide curves. During the third stage this method yields questionable results, per haps due to interference by oxidation and decomposition products but primarily due to the difficulty of getting the oxidized esters to dissolve or disperse completely in the carbon tetrachloride used as the solvent. Nevertheless, there is considerable evidence that no measurable unsaturation re mains in the ester films. It should be remembered, however, that only two of the three double linkages present in the eleostearate are measured originally. The methyl eleostea rate begins to show a rapid decrease in iodine addition value after approximately 1 hour of exposure and reaches zero addi tion value after about 35 hours. The maleic anhydride value starts to fall off at about the same time, but the rate of de crease is somewhat slower than that of the iodine addition value. As Figure 4 shows, the individual points obtained do not yield a smooth curve. The various humps are not due to experimental error, as check experiments indicate. On the other hand, there is no reason to believe that conjugated systems are formed again after they are once destroyed by oxidation and/or polymerization. On the basis of other measurements, especially infrared spectra, we are inclined to believe that these humps are due to interference by oxidation products, especially peroxides. For similar reasons it is felt that the sharp rise in the maleic anhydride value of the methyl linolenate during the first 10 hours of exposure does not indi cate the formation of conjugated double linkage systems but is merely due to a reaction between the maleic anhydride and the peroxides under the conditions of the experiment. It is significant that the humps in the eleostearate curve and the maximum in the linolenate curve correspond with the maxima in the respective peroxide curves. The refractive index curves (Figure 5) also show that the conjugated double linkage .system of the methyl eleostearate reacts quickly in the first stage of the drying process. Its high refractive index, due to the conjugation, drops off sharply, and after about 50 hours of exposure this ester, takes its place between methyl linolenate and linoleate. When i 381 382 INDUSTRIAL AND ENGINEERING CHEMISTRY VOL. 32, NO. 3 Reading from, top to bottom,: Fig u b b 10. Acre Va l u es ; Fig u b b 11. Es t e r Va l d e s ; Fig u r e 12. To t a l Ox y g e n b y Ul t ima t e An a l y s is these measurements are converted to specificirefraction (Fig ure 6), it becomes apparent that the eleostearate differs some what from the other esters which yield similar curves. The peroxide curves (Figure 7) present ijhe same general picture as the weight change curves, as far as the relative loca tion of the maxima is concerned. The highest peroxide con centration present at any time, even in the most unsaturated of the compounds (the methyl linolenate) is equivalent to only a fraction of one double linkage. This means that as the peroxide concentration approaches its maximum, the rate of decomposition or polymerization of the peroxides becomes appreciable. There is some reason to believe that only the peroxidic end groups are measured; if this is so, it should be possible to estimate the approximate size of the molecules from peroxide determinations. Decomposition, however, must be taken into account if serious errors are to be avoided. Reading from top to bottom: Fig d b e 13. Mo l e c u l a r We ig h t in Be n z e n e ; Fig d b e 14. Mo l e c u l a b We ig h t in <erf-BuTYL Al c o h o l ; Fig u r e 15. Vis c o s it y ; Fig u r e 16. De n s it y (d"*a) MARCH, 1940 INDUSTRIAL AND ENGINEERING CHEMISTRY 383 The hydroxyl concentration (Figure 8) reaches its maxi mum after the peroxides have started to decompose. The highest amount of hydroxyl present at any one time is again equivalent to only a fraction of a double linkage. A similar picture is presented by the aldehyde concentration (Figure 9) except that its maximum coincides roughly with the maximum of the peroxides. This observation must give rise to the question of whether these various methods are measuring peroxide, hydroxyl, and aldehyde oxygen, respectively, or whether they are not actually measuring the same thing. This question becomes more annoying when we attempt to interpret acid numbers (Figure 10) and ester values. Al though there is hardly any question that some acids are formed by oxidative decomposition of drying oil esters, the concentration of free acids in the ester films can hardly be more than a small fraction of the values indicated by the acid numbers. The same comments apply to the ester values (Figure 11). There appears to be little doubt that both of these determinations consume appreciable quantities of alkali in some reaction other than the neutralization of free fatty acids and the saponification of ester linkages. A more careful study of these reactions might easily furnish the solu tion to the question as to what happens to the peroxides dur ing the second stage of the drying process. When plotting the oxygen concentration in the four methyl esters as determined by ultimate analysis against the time of exposure, the curves shown in Figure 12 are obtained. The three more highly unsaturated esters show surprisingly little difference in the rate at which the oxygen concentration in creases. The methyl oleate differs somewhat from the other esters. The first maximum coincides roughly with the maxi mum in peroxide concentration. The sharp upturn after 100-hour exposure of this ester may be attributed to the for mation of oxidative decomposition products. The molecular weight curves obtained when using benzene (Figure 13) and fert-butyl alcohol (Figure 14) as solvents differ somewhat from one another in the maximum degree of poly merization indicated and the length of exposure required to reach these maxima. Both methods, however, indicate that the eleostearate passes through a maximum of molecular weight slightly before its maximum peroxide concentration is observed. The fact that the apparent molecular weight de creases on prolonged exposure only serves to add considerable support to the contention that cryoscopio methods are of doubtful value in the determination of the molecular weight of drying oil films. The viscosity curves (Figure 15), for ex ample, show a tremendous increase in the viscosity of the esters which is difficult to reconcile with the molecular weight data. It seems hardly possible that viscosity increases to one hundred times the original value could be observed at a time when molecular weight measurements show at best a dimeri zation. The magnitude of this change is also illustrated by the density measurements (Figure 16). Acknowledgment The authors wish to thank J. Smull and E. P. Glocker of Lehigh University who prepared the esters used in this in vestigation and A. J. Farber who did much of the analytical work. Literature Cited (1) Ellis, B. A., and Jones, R. A., Analyst, 61, 812 (1936). (2) Elm, A. C., In d . En g . Ch e m., 23, 881 (1931). (3) Gamble, D. L., and Bamett, C. E., Ibid., 32, 375 (1940). (4) Hamilton, L. A., and Oloott, H. S., Ibid., 29, 217 (1937). (5) Kaufmann, H. P., Ber., 70, 2554 (1937). (6) Marshall, Arthur, J. Soc. Chem. Ind., 19, 213 (1900). (7) Parks, G. S., Warren, G. E., and Green, E. S., J. Am. Chem. Soc., 57, 616 (1935). (8) West, E. S., Hoagland, C. L., and Curtis, G. H., J. Biol. Chem., 104, 627 (1934). (9) Wolff, Hans, and Zeidler, Gerhard, Paint Varnish Production Mgr., 15, 7 (July, 1936); 15, 7 (Aug., 1936). Pr e s e n t ed before the Division of Paint and Varnish Chemistry at the 98th Meeting of the American Chemical Society, Boston, Mass. PoLYBUTENB OP HlGH Mo l e c u l a r We ig h t Be in g Wo r k e d b e t w e e n t h e Ro l l s o p a Ru b b e r Mil l (See article, page 299.)