Document kmMe97rLG5RwOVMpX6kpE7Lb

304 INDUSTRIAL AND ENGINEERING CHEMISTRY VoL 41, No. 2 (29) Kass, presented before the Division pf Paint, Varnish, and Plastics Chemistry at; the Memphis Section Meeting of the Ame r ic a n Ch e mic al So c ie t y , Memphis, Tenn., 1942. (30) Kass, presented as a part of the Symposium on Drying*OiIs before the Division of Paint, Varnish, and Plastics Chem^try . it the Minnesota Section Meeting of the Ame r ic an Ch e mic a l So c ie t y , Minneapolis, Minn., 1947. (31) Kass and Burr, j. Am. Chem. Soc., 61,1062 (1939). (32) Ibid., p.3292. . - (33) "Kass and Shell, presented before the Division of Organic Chemistry at the Detroit Section Meeting of the Ame r ic a n Ch e mic a l So c ir t y v Detroit,vMich., 1943. (34) Kirschenbauer, XL S. Patent 2,389,260 (1945). (35) Kuhn and Meyer, Z. Physiol. Chem., 185,204 (1929). (36) Lewis, unpublished wort, Northern Regional Research Labora tory. (37) Lewis, Moser, and Cowan, unpublished work. Northern Re gional Research Laboratory. (38) Mattiello, "Protective and Decorative Coatings," Vol. IV, Chap. 12, New York, John Wiley & Sons, 1944. (39) Mitchell and Kraybill, J. Am. Chem. Soc., 64, 988 (1942). (40) , Mitchell, Kraybill, and Zcheile, In d . En g . Ch e m., An a l . Ed .. . 15,1(1943). ' (41) Moore, Bidchem. 31, (1937). (42) 'Morrell arid Davis, J. Chem. See. (Loudon), 1936, p. 1481. (43) Morrell and Davis, Trans. Faraday Soc., 32, 209 (1936). (44) Morton, Heilbron, and Thompson, Biochem. 25,20 (1931). (45) Myers, Kass, and Burr, OH A Soap, 18,107 (1941). (46) Nicola, Herb, and Riemenschreider, unpublished work. Eastern Regional Laboratory. (47) Northern Regional Research Laboratory, unpublished work. (48) Novak, V. S. Patent 2,178,604 (1939). (49) O'Hare and Withrow, In d . En g . Ch e m., 39,101 (1947). (50) Priest and Von Mikusch, Ibid., 32,1314 (1940). (51) Radiove, Teeter, Bond, Cowan and Kass, Ibid., 38,997 (1946). (52) Rhdlove, Teeter, and Cowan, Official Digest, Federation Paint ' yamish Production Clubs, 265/74 (1947). (53) .Ralston and Turinsky, U. 8. Patents 2,411,111-3 (1946). (54) Rose, Freeman, and McKinney, In d . En g . Ch e m., 34, 612 (1942). ` (55) Schicht and Grun, German Patent 287,660 (1914). (56) Spitzer, Ruthruff, and Walton, Am. Paint J., 26, No. 12, 68 (1941)* (57) Steger and Van Loon, Fettchem. Umschau, 43,17 (1936). (58) Strain, J. Am. Chem. Soc., 63,3448 (1941). (59) . Sunderland, J. OH & Colour Chemists' Assoc., 28,137 (1945). (60) Cowan, Teeter, Bachman, and Bell, In d * En g . Ch e m., in press. (61) Teeter,. Radlove, and Cowan, unpublished work, Northern Regional Research Laboratory. (62) Touchin, Paint Manuf. 16 (No. 7), 237 (1946). (63) Turk, and Boone, tJ.S. Patent 2,405,380 (1946). (64) Turk, Dawson, and Soloway, Am. Paint J., 28, No. 9 16 (1943). (65) Turk and Feldman, Paint, Oil, Chem. Rev. 106, No. 13, 10 (1943). (66) Varrentrapp, Ann., 35, 196 (1840). (67) Vlodrop, van, Chem. Weekblad, 38,150 (1941). (68) Von Mikusch and Frazier, In d . En g . Ch e m., An a l . Ed ., 15. 109(1943). (69) Waterman and van Vlodrop, Alien Property Custodian, 359,978 (Filed 1#40); Waterman and co-workers, Verfkronick, 13, 130-6,180-2 (1940). Re c e iv e d February 26, 1948. Recent work on isomers of conjugated octadecadienoic acid indicates that revised constants are necessary in apectrophotometric determinations of polyunsaturated acids (Paper 19, Am. Oil Chemists* Soc., New York, Nov. 15--17, 1948; also Papers 16, 18, 33, 55, and 56 on isomerisation of drying oils). Mechanism of the Oxidation of Drying Oils P. O. Powers Battelle Memorial Institute, Columbus, Ohio The steps encountered in the oxidation of drying oils explanation of all the phenomena observed in the oxidation of by air are peroxide formation, decomposition of peroxides, drying oils has not been achieved. polymerization, and degradative oxidation. An induction This failure often has been attributed to the complexity of the period occurs prior to uptake of oxygen and this has been drying oils themselves, and it is only in recent years that an ac shown to be due to the presence of inhibitors. .Recent curate analysis of the composition of linseed oil was possible. work has shown the presence of hydroperoxides, particu However, our knowledge of the composition of the common dry larly when oxidation is conducted under mild conditions, ing oils is now on a quantitative basis, and the accuracy of the but there is also considerable evidence for the presence of methods of analysis has been greatly improved. A great deal of cyclic peroxides. Peroxides apparently decompose by de progress has been made in the separation of the pure fatty acids. hydration and by reduction. It seems possible that per Methyl esters of these acids serve as useful methods of interpret oxide decomposition rand polymerization are intimately ing the behavior of the drying oils on oxidation. The study of the associated. Polymerization inevitably occurs subsequent behavior of mixtures of esters does not always parallel the behav to oxidation but it has not been established clearly whether ior of the pure ester since the oxidation of oleate esters is greatly the polymer chains are formed through either ether or increased (18) in the presence of a small amount of a linoleic acid peroxide linkages or through carbon-to-carbon bonds. ester. Rate of oxidation is increased, however, with the- unsatu Considerable evidence for the latter structure has been ration (55), and when mixtures of trienoic and dienoic esters (15) accumulated. are present, the trienoic esters are oxidized first. Also, when di enoic and oleic esters are present, the dienoic .esters are oxidized TJRING recent years a careful study has been made of the first. Boxidation of animal and vegetable oils; still, the statement In one respect the methyl esters are not complete models for of Miilas (24) made in 1932, " .. . the mechanism of the oxidationthe drying oils. They do not dry and thus remain susceptible to of unsaturated oils is still relatively obscure owing to the como xidation to a . greater degree than do the glycerides. Thus, plexity of the oxidation products and the uncertainty of their greater amounts of oxygen may be absorbed than will be absorbed structure," remains an accurate estimate of our present knowl by the corresponding glyceride (88). Such information may be ap edge. Progress has been apparent in the intervening 15 years and plicable to the study of deterioration but is a further stage of oxi much valuable work has been done by university, government, dation than usually is encountered in the drying process. and industrial laboratories both here and abroad. In spite of all The complexity of the products of oxidation of the drying oils this activity, it must be admitted that an entirely satisfactory is well known and is the principal reason for the difficulties en- N41567 February 1949 INDUSTRIAL ANVD ENGINEERING CHEMISTRY 30S countered in studying the mechanism of oxidation. The reactions occurring are often consecutive and apparently in some instances are competitive. It has been found (26) that conditions can be controlled to favor oxidation rather than polymerization. In many oxidation reactions there seems to be a preliminary phase where oxidation alone occurs; subsequently polymerization takes place with little oxidation. Whether this behavior is more ap parent than real remains to be established. One factor that may explain in part the lack of a satisfactory mechanism of the oxida tion process is the divergent aims of groups of investigators. Much of the investigation of the oxidation process has been con ducted with the purpose of preventing oxidation completely. Other workers are interested in completing the oxidation as quickly as possible and are primarily interested in accomplishing polymerization of the oils. Methods of Research t ime--- It is becoming increasingly apparent that no one method of at- Figure 1 tack will reveal the whole mechanism of oxidation of the drying oils. Chemical methods of analysis in many cases are limited severely by the instability of the oxidized products. Also, the in solubility of the gels resulting from oxidation severely restricts such methods of attack. Physical methods offer promise of sup 0 plementing chemical analysis. Spectroscopic information is prov ing especially helpful in this direction, particularly the ultraviolet absorption spectra. The effect of the carbonyl group is recog nized, but a complete quantitative measure of its presence in con jugated structure is not established yet. Infrared spectra of oxi dized oils have been measured but further work with model sub stances will be required before the results will give more than spec ulative information on the structure of the oxidized oils. The temperature of oxidation (37), exposure to light, presence of driers, access to oxygen, film thickness, and other factors influence the course of oxidation. Volatile products which are evolved in the course of the oxidation, particularly water, may yield useful clues on the course of the oxidation. Information on the nature and amount of these products and of the stage of oxidation when they are evolved is still incomplete. The dielectric constant of blown oils (26) has been measured at various stages of oxidation. The 1 Grignard reagent has been employedrecently(39) and gives promise in establishing the function of the oxygen in the oxidized oils. hibitors. It has been found (18) that synthetic unsaturated esters do not show an induction period. Also, the induction period can*be extended by the addition of inhibitors; the in hibition period is prolonged with each successive increment of antioxidant. The induction period is shorter as the temperature is increased and is often negligible at 100 C. in the presence of the amounts of inhibitors naturally present in drying oils. Materials which can be oxidized to quinones are among the most effective antioxidants although many other phenols and aromatic amines show some antioxidant effect (//, 27). Of the dihydric phenols, catechol and hydroquinone have pronounced antioxidant power whereas resorcin is less effective. The ethers or esters of hydroquinone are much less active inhibitors. Aminophenols and substituted aminophenols are often active anti oxidants, as are phenylenediamines and other aromatic di amines. Of the more highly substituted phenols, pyrogallol and gallic acid and its esters have been found to possess pro nounced antioxidant power. a:-Naphthol is more effective than 0-naphthol (27).. Many of the natural oils exhibit a definite induction period and all of the nonlipid components of the oil have been suspected of being responsible for the effect. It has been established that the Oxidation Process inhibition is due to the presence of the tocopherols. On removal of these sterols, the oils dry more rapidly, and by addition of The stages in the oxidation of drying oils by air may be broken tocopherol concentrates obtained by molecular distillation, the down into inhibition, peroxide formation, peroxide decomposition, stability of oils to oxidation is increased. and polymerization. In Figure 1, the changes encountered are The effectiveness of antioxidants often is found to be increased plotted to illustrate the behavior at each stage of the oxidation. by the presence of acids. Phosphoric acid has been found to be The scale is purely arbitrary and may vary greatly according to especially effective although citric and ascorbic acid also exhibit conditions or materials employed. In the inhibition step, the synergistic properties. It has been found that the amount of end of the phase is quite apparent, and at early stages of the oxi phosphoric acid which is soluble in an oil (0.0002%) is not ef dation, peroxides are the only products formed. Beyond this fective (8) in promoting the antioxidant action. Oxalic acid point, separation of phases of the various steps is-largely artificial. is active as an antioxidant in addition to any synergistic effect Peroxidation does not cease necessarily when rearrangement be (23). gins. Polymerization may well begin with the decomposition of In vinyl polymerization it is found that a small amount of peroxides. These steps, however, do represent a period in the antioxidant can prevent the polymerization ofa large amount of oxidation when a given process appears to predominate. vinyl compound because the chains are long. In the case of the The induction period ends when oxygen absorption begins, and oxidation of drying oils, the length of the chains has been esti the peroxidation step may be considered to continue to the maxi mated as rather short (32). If oxidation is autocatalytic, how mum peroxide value. Peroxides subsequently largely disappear al ever, a small amount of antioxidant should be effective. though oxygen may continue to add to the oil. The polymeriza tion phase is marked by the rapid increase in viscosity. Oxida FORMATION OF PEROXIDES tion continues long after the polymerization phase is apparently complete. There is general agreement that oxygen first adds to drying oils with the formation of peroxides; their structure and position INHIBITION in the oil molecule has not been entirely clarified. The method of Wheeler for determination of peroxides usually is employed. However, such effects may be caused by physical conditions Determination of peroxides by oxidation of the ferrous Io q . has which limit the rate of oxygen uptake. The induction period been employed often to determine peroxides' from the hydro now is considered generally to be caused by the presence of in carbons, but this apparently gives unexpectedly high values when 306 INDUSTRIAL AND ENGINEERING CHEMISTRY v . VoL 41, No, 2 applied to fatty acid peroxides (4).. There is indication that peroxides of varying stability occur, since, on heating, the per oxide value drops to a constant value which decreases but slowly on further heating (29). Types of Peroxides. Until recently, the Cyclic peroxide formed by the addition of a molecule of oxygen to the double bond has been generally accepted. Thisrview has been reinforced by the almost generally observed drop of the iodine value corresponding to the disappearance of one double bond with the addition of one molecule of oxygen. --CH=CH-- II 0--0 The rupture of the carbon chain at the double bond by decom position of the peroxides also is indicative of addition of oxygen at the double bond. This structure, however, has been considered (12) sterically so unstable as to eliminate any but momentary occurrence. A modification of this structure may be considered also-- the ethylene oxide type structure proposed by Paquot (12) \z O. This structure should exhibit the same properties as O the four-membered ring peroxide. In addition, it suggests the possibility of rupture of the 0=0 bond to yield an ethylene oxide derivative. The high content of esters (9t 28) formed in oxidized 60s and particularly oxidized unsaturated acids suggests such a structure, as does the isolation (28) of a dihydroxystearic acid by hydrolysis. The hydroperoxide theory proposed by Farmer (12) has many points in its favor. In the case of the drying oils, Farmer (12) believes the initial addition of oxygen is at the double bond. Addition to the o-methylenic carbon atom is regarded as im probable because of the high energy of activation- required to break the C--H bond. The oxygen is believed to add to the double bond to form a biradical: v 00* ' --CH*CH=CH 4- O, -->- --CHiCHCH-- This combines with another unsaturated group to yield two radicals: * 00* --CH*CH=CH----\r --CH*CH--CH---------- * * OOH --CHCH=CH-- d--rCH2GHCH-- These radicals continue the chain reaction and lead to the formation of a-methylenic hydroperoxides. Thus, a small amount of addition at the double bond will lead to the formation of a large amount of hydroperoxide if the chains are long. One piece of evidence that does not fit smoothly into the hydroperoxide structure is the drop in iodine value on oxidation; this would not be expected: if*the peroxide added at the a-methylene carbonratom. This decrease in iodine value may in part he explained by the formation of conjugated structures which are known to add less than the theoretical amount of iodine. The other questionable point is the failure to find (41) active hydrogen in air-dried films. The expected amount of active hydrogen, however, has been found (4) in blown ethyl linoleate. Investiga tion of this field is active and further work undoubtedly will be done to clarify these points. Since the behavior of the oleate esters, the unconjugated poly ene esters, and the conjugated unsaturated esters each exhibit their own peculiar behavior pattern on oxidation, they will be considered separately. Oleate Esters. Relatively pure hydroperoxides have been obtained only from methyl oleate. A methyl oleate hydroper oxide apparently of over 90% content has been isolated; it con sists of two oleate hydroperoxides differing in the position of the hydroperoxide group. The hydroperoxide has been concentrated by molecular distillation from, methyl oleate slightly oxidized at moderate temperatures. The further separation from un oxidized methyl oleate is accomplished by chromatographic separation (14) or by low temperature crystallization from ace tone (85). This hydroperoxide apparently is quite stable and failed to decompose a-tocopherol after several weeks' exposure. It has an iodine value corresponding to one double bond. On oxidation with potassium permanganate, azelaic and suberic acid were formed as were caprylic and pelargonic acids (85). These acids were formed apparently in approximately equivalent amounts, indicating that the hydroperoxide groups are at the 8 and 11 positions, the methylenic groups alpha to the double bond. It has been found that when methyl oleate is oxidized at 120 C., no indication was found of the presence of the 8 hydroperoxide ester (2) although the same investigators found evidence of its presence in sunlight at 20 C. Since methyl oleate hydroperoxide exhibits an iodine value and a conjugated structure is not present, the iodine value of methyl oleate should give an index to the extent of hydroperoxide forma tion. It is found that while the iodine value does not decrease to the extent expected if the peroxide group added to the double bond there is a considerable drop in iodine value on peroxidation which is apparently 70% of the expected drop (2). . Unconjugated Polyene Acid Esters. Considerable study has been made recently of the peroxidation of methyl linoleate. However, hydroperoxides have not been isolated. It has been found that the ultraviolet absorption at 2325 A. increased with the absorption of oxygen. At moderate temperatures all of the absorbed oxygen is present in the form of peroxides and 70% of these peroxides are conjugated (22). This is in agreement with the mechanism proposed for the formation of the hydroper oxides (6} 18). The 11 carbon atom becomes the center of a radical formed in the oxidation process. This may rearrange to radicals having their center at the 9 or 13 carbon. The hydro peroxide then can add at either the 9, 11, or 13 carbon atom;, the probability is assumed to be equal for each position. Since the 9 and 13 hydroperoxides are conjugated a value of two thirds of the peroxides containing a conjugated structure is in agreement with these assumptions. However, by hydrogenation of methyl linoleate hydroperoxide, 9 and 13 hydroxystearic acid were identified (3) but no proof of the presence of the 11 hydroxy compound could be obtained. * t-C--CH=CHCH=CH-- * --CH=CHCHCH=CH 13 11 9 --CH=CHCH=CHCH-- * --CH=CHCHCH=CH It has been shown that the diene conjugation (17) increases parallel to the rise of the peroxide value, but a sharp drop in diene conjugation occurs just before the decrease in peroxide content which oiccurs as oxidation proceeds. This is.in contradic tion to the observation (17) that the diene content is not changed when peroxides are decomposed by heating in vacuum. The methyl esters of soybean acids on oxidation at five tem peratures from 15 to 100 C. show (29) that during the early stages of oxidation the iodine value drop is less than the expected value for the disappearance of one double bond at low tempera tures. As the temperature is raised, the iodine value approaches the expected value and coincides at-100. C. The iodine number drop exceeds the peroxide value as the oxidation is carried, fur ther. This condition occurs at progressively lower peroxide values as the temperature is raised. In this, connection, it' has 41, No. 2 dj it con- tion of the acentrated i y oxidized un- aphic acestable and exposure. Dond. On id suberic icids {85). equivalent are at the uble bond, it 120 C., roperoxide 2nce of its value and of methyl ide formadecrease to the double iroxidation study has linoleate. i has been eased with ires all of s and 70% agreement hydroperenter of a to { Jfdto- te^&tom; on. Since two thirds agreement of methyl acid were 1 hydroxy )H--CH-- * =CHCH-- )H--CH ) increases rp drop in d peroxide contradicot changed t five tem5 the early ie expected v teinperaipproaches ne number I ^ fur* /bxide hasiX/setfj'it February 1949 INDUSTRIAL AND ENGINEERING CHEMISTRY . 307 been found {22) that at low stages of oxidation, the conjugation is'measured by the peroxide value at several oxidation tempera tures ranging from 40Q to 100 C. This evidence points to a cyclic peroxide possessing a conjugated structure. The ultra violet spectrum of methyl oleate hydroperoxide {85) shows no evidence of diene structure. 1 The behavior of methyl linolenate parallels that of methyl linoleate. Diene conjugation increases with peroxide. content. Triene conjugation also is found but . the amounts are rather smaller than the diene content, usually less than 20% of the diene content. The kinetics (4) of the oxidation of ethyl linoleate has been studied recently. The oxidation is autocatalytic since the rate of oxygen pickup is proportional to the peroxide content. When benzoyl peroxide is added to ethyl linoleate, the rate of oxygen uptake is proportional to the square root of the peroxide content. Since the radicals liberated by benzoyl peroxide are well recog nized, 'examination of products from benzoyl peroxide promoted oxidation might afford information on the mechanism of the oxidation. The heats of oxidation of ethyl linoleate and lino lenate have been measured {5). The value of each ester was substantially the same (), 52 kg-cal. per mole of peroxide. A value of about 66 kg-eal. was estimated as the bond energy of the O--O bond in the hydroperoxide, an estimate considerably greater than earlier values. Conjugated Esters. In the case of the conjugated esters, the evidence seems clear that the addition is at the double bond. There is no evidence of active hydrogen during the early stages of the oxidation {89). Although there is a notable increase in conjugation on oxidation,of-Unconjugatedsystems, there is every evidence of disappearance of conjugation on oxidizing con jugated systems. It has been suggested that the addition of oxygen in conjugated systems is 1-4. However, there is evidence that conjugation is still present in methyl oleostearate after the addition of 1 mole of oxygen; this is suggestive of 1-2 addition. The formation of polymeric peroxides has been sug gested frequently as occurring in conjugated systems. Evidence to this effect is shown by the lower content of polymers obtained in dilute solution oxidation. However, the same effect would be expected whatever the mechanism of the polymerization reaction. The reaction for the formation of cyclic peroxides is often written as occurring in one step. The presence of two molecules of oxygen and two fatty acid segments in an active state at the same time is too remote for serious consideration. 2--CH=CH-- + 02 V--CH--CH- iA --CH--G'H- If the reaction is considered to occur in steps, it is surprising that two links between segments are required. One peroxide link between chains would be sufficient to form a cross link and it might be expected that the second oxygen molecule would add at another double bond in the molecule or to an unoxidized molecule. DECOMPOSITION OF PEROXIDES In the oxidation of drying oils the peroxide value accounts for practically all the oxygen added during early stages of oxidation. At low temperatures this maybe the case until 0.5 mole of oxygen has been added {29). At higher temperatures and under more severe conditions other oxygenated products besides peroxides are formed earlier in the oxidation. It is apparent that the oxygen first adds as peroxide and other oxygen compounds are their decomposition products. In addition to temperature, light driers, acids, and bases accelerate the decomposition of peroxides. In the case of peroxidized petroleum oils (41),three modes of decomposition are recognized; dehydration with the liberation of a molecule of water; reduction with the liberation of a mole cule of active oxygen; or by further oxidation with the splitting of at least one C--C' bond. Aldehydes, ketones, and acids are among the principal products formed. The same type of re actions are observed in the case of the drying oils and polymeriza tion reactions induced by oxidation also occur. ! Type reactions by which peroxides decompose may be indicated as follows: 0-0 --CH2(!)H(1)H --> --CH==CHCO-- + H20 Dehydration 0-0 O --CHCH-- -->: --CHCH-- -f* O 0--0 --CHCH--- --> --CHOHCO-- 0-0 v --culm-- --> 2-CHO Reduction Rearrangement Rupture These reactions are expressed in terms of the cyclic peroxide structure, which has been used generally to explain the products formed. Each of these reactions will be considered separately. Dehydration. Water is evolved in the drying of oil films; as much as 6% has been measured {16). It is well known that high humidity retards the drying of oil films. It apparently does not greatly affect the pickup of oxygen, hence dehydration may be involved in the polymerization step. The increase in conjugation found in oxidized oils has been ascribed {20, 25) to the formation of unsaturation by the dehydration step. It has been found that methyl linoleate evolves 1 molecule of water on exposure to air in thin films and methyl linolenate evolves between 1 and 2 moles per mole of ester {89). The glycerides evolve the same amount of water on drying {40). This water is evolved after considerable oxygen has been added to the esters. Increase in triene conjugation when oxidized oils are treated with alkali is often quite large {19, 20, 22); this suggests the enolization of carbonyl groups adjacent to a double bond. Such a structure would be anticipated from the dehydration of a per oxide: 0--0 II KOH --CH==CHCH2CHCH --^ --CH=CH--CH===CHCO-------- > ;K O --H^CHCH=-CH--=-OH-- Reduction of Peroxides. The evolution of active forms of oxygen has been postulated and evidence {84) has been presented for the evolution of hydrogen from drying oils. It is to be ex pected that active oxygen if evolved would be absorbed readily by the drying oils. The increase in ester number encountered in oxidized oils and the isolation of dihydroxystearic acid {9) are suggestive of this type of decomposition of peroxides. Rearrangement of Peroxides. This reaction has been sug gested to explain the disappearance of peroxides. However, there is little quantitative evidence in its favor. If peroxides in drying oils decomposed largely in this manner, a much higher content of hydroxyl groups would be found. Oxidized oils often contain less than 2% hydroxyl {28). Chain Rupture. The final reaction suggested for the decom position of peroxide results in breaking of the carbon chain. This undoubtedly occurs in the oxidation of drying oils. However, 308 12 INDUSTRIAL AND ENGINEERING CHEMISTRY Vol. 41, No. 2 14 16 OXYGEN, % Figure 2 IS is in agreement with the behavior of the. esters of the two adds on heat bodying. However, in neither case is there quantita tive evidence of the extent of dimer or trimer formation on oxidation. An analysis (1) of the polymer structure on polymer izing unsaturated glycerides has been developed. This analysis was developed to interpret the behavior of heat bodied oils but is equally applicable to oxygen induced polymerization. Oxygen Linked Polymers. While the air dried oil films give clear evidence of a cross linked polymeric system, our knowledge of the bonds between the chains is still fragmentary and has been arrived at largely by inference and as yet no rigorous determina 20 tion of structure has been made. The view held for years and recently supported by Treibs (89, 40), is that a dioxane link joins the fatty acid segments. This ring has been assumed to be formed by the rearrangement or decomposition of a perdioxane ring: --CH--CH-- /A vA--CH--CH--. --CH--CH-- + o. ~^OH---- H O --C--CH-- o7 V -^c h --<5o !OH- in the drying of films it does not occur to any great extent. It may be an important factor in the eventual destruction of the films. Dicarbonyl groups also have been found present in drying oils (31). Their content in oxidized oil is not large, however. They could be formed by the dehydration of a peroxide adjacent to a carbonyl group. Change in hydroxyl content and iodine value has been measured when peroxides are decomposed (29) by heating 190 hours at 55 C. The hydroxyl value did not change appreciably even with a decrease of 2000 in the peroxide value. The iodine value also showed no appreciable decrease. This indicates some other mechanism than rearrangement, and is not in agreement with the dehydration reaction unless the unsaturation formed is saturated by some attendant reaction, such as polymerization. There are several serious objections to this mechanism. There has never been any explanation offered of why it does not operate in the case of oleic acid esters which show a negligible tendency to polymerize under usual conditions. The mechanism of such a reaction would be exceedingly complex to the point of being improbable. A peroxide formed at the double bond would have to combine with ah unsaturated group on another fatty acid seg ment, the second mole of oxygen would then add and complete the ring. The reaction often is written as comprising a single step. The probability of two molecules of oxygen and two un saturated groups arriving at the required arrangement in space at the same time is too remote for serious consideration. Other objections to this theory include the fact that the presence of the dioxane ring has never been established in oxidized oils (10). Also the only evidences for the presence of the neces sary amount of ether oxygen are indirect and inconclurive. Carbon-Carbon Linked Polymers. A more reasonable bond between segments is by a carbon-carbon linkage. This might result from an aldoi condensation between a carbonyl group in one chain and a methylene group in a second chain. This type of bond has been suggested (20) as the cause of chromophores developed in oxidized oils. Changes in physical and chemical POLYMERIZATION OF DRYING OILS The polymerization of oils induced by oxidation is now recog nized as the process responsible for the drying of oil films. The concepts of high polymer chemistry have thrown considerable light on this process. The background of this interpretation was outlined by Bradley (7) in 1936. It was not recognized then that three fatty acid segments could combine in a single chain. In the case of heat bodied oils, the presence of such trimers is now well substantiated. In the case of the air dried materials, evidence for trimer formation exists in the observation of Elm (28) that glycol linoleate forms a gel while glycol linolenate dries to a hard film. This ihay be interpreted as an indication that linoleic acid shows a slight tendency to form trimers while the tendency is much greater in the case of linolenic acid. This Figure 4 41, No. 2v two acids quantitanation on i polymer- films give knowledge 1 has been leterminaby Treibs segments, igement or 4- Os [-- m. There tot operate mdency to of such a t of being /ould have i )|ete ig ^ Jngle id two un it in space . that the in oxidized the necesdve. table bond This might roup in one lis type of omophores i chemical 30 February 1949 ' INDUSTRIAL AND ENGINEERING CHEMISTRY 30ST properties accompanying the polymerization step indicate that aldol formation is not the principal step responsible for the poly merization of drying oils. It has been proposed (50) that two of the unsaturated groups in adjacent fatty acid segments combine by vinyl polymerization. The pronounced changes in density, decrease in iodine value, and decrease in refractivity accompanying the drying of films with little change in oxygen content are indicative ofsuch a mechanism. It was shown that linoleic esters show these changes to a more pronounced degree than do linolenic esters. This might be expected since two double bonds are sufficient to give air-drying properties. With linolenic acid the third double bond affords a . site for further oxidation which partially masks the effect of the polymerization reaction. To determine if blown oils show the same behavior as do films, linseed oil was blown at 80 C. and samples taken at intervals. Density and refractivity were determined, also the carbon and Ihydrogen content. The curves (Figures 2 and 3) show that the iblown oils do indeed show the same behavior as the films on drying. It should be noted that the blown oils were carried only to the gel stage while films become hard. Therefore, the trend exhibited at the end of the blowing might well be expected to continue as drying is completed. In an attempt to throw additional light on the mechanism of polymerization, the heat of formation of blown oils and oil films at various stages of oxidation was calculated from data in the literature (21, 86). This work was supplemented by determina-tion of the heat of combustion of samples of linseed oil blown under conditions stated above. The results were in general comparable for all the data obtained (Figure 4). The heat of formation increased up to the point where about two molecules of oxygen were added by the glyceride. Beyond this point, the heat of formation decreases. No definite conclusions have been drawn. It was felt that since volatile products, including carbon dioxide, were lost from the system, the analysis of the process was not accurate until the effect of these materials was included in an analysis of the whole system. However, all values were reduced to an oxidized tri glyceride and should be roughly comparable. The polymeriza tion reaction would be expected to give a greater heat of forma tion and such a result was expected. A possible explanation is the formation of unsaturated groups in the oxidation process. It has been suggested above that the liberation of water from the oxi dized oils may lead to unsaturation. Such groups would give a lower heat of formation. It has been shown (50) that the loss of unsaturation, increase in density, and drop in molecular refractivity accompanying the $ -drying of unsaturated ester films might equally well be interl preted as reflecting bonding between the chains by a Diels re|| action. --CO--CH=CH--CH==CHC0-- + --OH=-CH--CH=-CH-- --^ --COCH--CHCH=CHC0-- --CHCH=CH--CH- The weight of evidence accumulated on the behavior of oxidized -oils now seems to favor this view above the possibility of vinyl polymerization. Speetroscopic evidence points to the presence of carbonyl groups adjacent to carbon-carbon double bonds, and possibly such carbon-carbon double bonds with carbonyl groups alpha to both carbon atoms and also a considerable content of con jugated dienes. Such systems might readily be expected to enter into a Diels reaction. The changes in density and specific refraction with little change in oxygen content are in agreement with this type of reaction and the decrease in iodine value is more nearly in agreement with this type of reaction than is vinyl polymerization. The results of the study of the decomposition of peroxides cited above suggest that the. dehydration and polymerization steps may be closely associated and occur in immediate sequence. Such a mechanism would explain the failure to observe a drop in iodine value on decomposition of the peroxides (#0), the double bond formed in each chain becoming saturated in the formation of the sax-membered ring. The evolution of water, associated with the drying of oils, is suggestive of such a reaction. Literature Cited (1) Adams, H. E., and Powers, P. O., J. Applied Phys., 17, 325, (1946). (2) Atherton, D., and Hilditeh, T. P., J. Chem. 80c., 1944, p. 105. (3) Bergstrom, S., Nature, 156,717-18 (1945). (4) Bolland, J. L., Proc. Royal Soc. (.London) 186A, 218 (1946). (5) Bolland, J. L., and Gee, G., Trans. Faraday Soc., 42, 244 (1946). (6) Bolland, J. L., and Koch, H. P., J. Chem. Soc., 1945, p. 445.' (7) Bradley, T. F., In d . En g . Ch e m., 29,440, 579 (1937). (8) Calldiis, V. P.# J. Am. Chem. Soc., 69.384 (1947). (9) Deatherage, F. E., and Mattill, H. A., In d . En g . Ch e m., 31,1425 (1939). (10) Elm, A. C., Ibid., 23, 881 (1931). (11) Evans, E. A., J. Inst. Petroleum, 32, 392 (1946). (12) Farmer, E. H., Trans Faraday Soc., 42, 228, (1946). (13) Farmer, E. H., Koch, H. P., and button, D. A., J. Chem. Soc., 1943, p. 541. (14) Farmer, E. H., and Sutton, D. A., J. Chem. Soc., 1943, p. 119. (15) Filer, L. J., Jr., Mattil, K. F,, and Longenecker, T. E., Oil A Soap, 22, 196-201 (1945). (16) Gardner, W. H.f and Waddell, R. B., In d . En g . Ch e m., 33, 629 (1941). (17) Gunstone, F. D,, and Hilditeh, T. P.# J. Chem. Soc., 1945,836. (18) Hilditeh, T. P., and Sleightholme, J. J., J. Soc. Chem. Ind., 51, 39T (1932). (19) Holman, R. T., and Burr, G. O., J. Am. Chem. Soc., 68, 562 (1946). (20) Holman, R. T., Lundberg, W. O., and Burr, G. O., Ibid., 67, 1386 (1945). (21) Long, J. S., Zimmerman, E. K., and Nevins, S. C., In d . En g . Ch e m., 20, 806 (1928). (22) Lundberg, W. O., and Chipault, J. R.f J. Am. Chem. Soc., 64, 833 11947). (23) Mattill, H. A., Oil & Soap, 22, 1 (1945). (24) Milas, N. A., Chem. Rev., 10, 308 (1932). (25) Mitchell, J. H., Jr., and Kraybill, H. R.f J. Am. Chem. Soc., 64, 988 (1942). (26) O'Hare, G. A., and Withrow, W. J., In d . En g . Ch e m.* 39, 101 (1947). (27) Olcott, H. S., J. Am. Chem. Soc., 56, 2492 (1934). (28) Overholt, J. L., and Elm, A. C., In d . En g . Ch e m., 32, 378 (1940); 32, 1348 (1940); 33, 658 (1941). (29) Paschke, R. F., and Wheeler, D. H.f OH Sc Soap, 21, 52 (1944). (30) Powers, P. O., Overholt, J. L., and Elm, A. C.t In d . En g . Ch e m., 33, 1257 (1941). (31) Prill, E. A., OH Sc Soap, 19,107 (1942). (32) Roger, W.f Jr.t and Taylor, H. S., J. Phys. Chem., 3, 1334 (1926). (33) Stirton, A. J., Turer, J., and Riemenschreider, R. W., OH Sc Soap, 22, 81 (1945). (34) Stutz, G. F. A., Nelson, H. A., and Schmutz, F. C., In d . En g . Ch e m,, 17, 1138 (1925). (35) Swift, C. E., Dollear, F. G., and O'Connor, R. T., Oil Sc Soap 23 355 (1946). (36) Taylor, R. S.f and Smufl, J G., In d . En g . Ch e m., 28, 193 (1936). (37) Treibs, W., Ber., 75, 203 (1942). (38) Ibid., p. 632. (39) Ibid., p. 953. (40) Ibid., 76, 670 (1942). (41) Zuidema, H. H., Chem. Rev., 38, 197 (1946). Re c e iv e d February 26,1948.