162 FRIEND NOTES ON THE EFFECT OF XVK-Notes on the E’ect of Heat and Oxidation on Linseed Oil. By JOHN ALBERT NEWTON FRIEND. ALTHOUGH a considerable amount of work has been done on certain properties of linseed oil and the manner in which they are affected by heat and oxidation a few of the constants have not received the attention they deserve. Two years ago therefore the author decided to investigate the subject and although the work has had to be abandoned temporarily it was thought desirable t o place on record the results obtained. Two samples of oil were used namely best Baltic and best Cal-cutta raw linseed oil. They were obtained from a reputable firm, and having been purified by recognised commercial methods, further purification was deemed unnecessary HEAT AND OXIDATION ON LINSEED OIL.163 The Effect o f Hent on Linseed OiT. The molecular weight of linseed oil has occasionally been deter-mined by different investigators the figure obtained being of the order of 800. No systematic study of the subject however appears to have been made. Fahrion (Zeitsch. angew. Chem. 1892 5 171) noted that after being heated linseed oil lost much of its power of absorbing iodine and suggested that polymerisation had taken place. Fokin found that by heating the oil to 250-300° in an inert atmosphere a substance having a molecular weight approach-ing 2000 is obtained. Simiiar results have been achieved by Morrell ( J . SOC. Chem. Itid. 1915 34 105). E X P E R I M E N T A L . Samples of the same batch of pure Baltic oil were obtained, through the kindness of Dr.Morrell which had been "thickened" by heat but without appreciable oxidation as follows : Sample 1.-Raw oil. ,, ,, ,, 2.-Oil heated for thirty hours at about 200O. 3.-Oil heated for thirty-six hours a t about 300O. 4.-Oil heated for forty-two hours at about 300O. relative viscosit,ies and molecular weights were determined. The densities coefficients of expansion with rise of temperature, TABLE I. CoefIicient of Density Density expansion Relative Oil. at 0". at 50". 0-50". viscosities. 1 0.94444 0.90956 0.00076 100 2 0.94289 0.9091 2 0.00071 73 3 0.95804 0.9238 1 0.00074 459 4 0.9 7 2 6 7 0,93901 0*00072 3770 The coefficient of expansion steadily falls whilst the density and viscosity rise with the temperature and length of treatment.Sample 2 is interesting in that its density and viscosity are slightly lower than those for the raw oil. Possibly this is due to the pre-cipitation or flocculation " of certain constituents of the raw oil. The molecular weights were determined by the freezing-point met,hod in benzene solution.* * The author desires to thank Dr. W. E. S. Turner of Shcffield University, in whose laboratories the determinations were made for the necessary supply of benzene. Its melting point was 5.934" 164 FRIEND NOTES ON THE EFFECT OF Grams of oil in 100 grams Oil. of benzene. 1 17.53 10.22 9.89 2.69 Calculated a t infinite dilution . . . . . . 2 28.45 16.76 7.66 Calculated a t infinite dilution . . . . . . TABLE 11. Molecular weight, 624 673 704 733 744 594 667 738 760 Grams of oil in 100 grams Oil.of benzene. 3 31.74 21.07 18.84 10.33 Calculated at infinite dilution . . . . . . 4 22.42 14.58 7 . 6 2 Calculated a t in-finit,e dilution.. . Molecular weight. A75 778 794 897 lo00 995 1118 1270 1420 I n every case the apparent molecular weight increases with dilution. The results show that polymerisation takes place a t temperatures as low as 200° and increases in extent with the temperature and length of exposure t o the same. The Oxidation of Linseed Oil. When linseed oil is exposed t o the air it increases in weight, slowly a t first but rapidly later due t o the absorption of oxygen. A maximum point is ultimately reached after which a gradual loss in weight occurs.Many attempts have been made t o deter-mine the value for this maximum in the case of oils spread in thin layers on glass or other non-absorbent plates and the figures obtained have been used as bases of comparison of the relative I' drying" powers of the oils. I n 1913 Ingle (?7. Soc. Chem. Inc?., 1913 32 639) concluded that the maximum increase in weight is exactly double that' calculated from the iodine value thus suggest-ing the formation of peroxides during oxygen absorption. This afforded strong support to the belief that the iodine number is an approximate measure of the drying power of an oil as suggested by Hazura (Zcitsch. angew. CIicin. 1888 1 312). These experi-ments however ignore the fact that linseed oil during practically the whole period of its oxidation also loses appreciable quantities of water vapour carbon dioxide and organic vapours so that the total weight of oxygen absorbed exceeds the observed increase in weight of the oil by the weight of the vapours evolved.The latter amount is not negligible; in one experiment the total weight of oxygen absorbed was 2.5 times the observed increase in weight' of the oil a t the maximum point (Friend Proc. Paint Varnish Soc., May 14th 1914). The maximum point therefore does not represent the stage a HEAT AND OXIDA4TION ON LINSEED OIL. 165 which oxygen ceases t o be absorbed but a point of equilibrium a t which the oxygen absorbed exactly counterbalances the loss in weight due to escaping gases an6 vapours.Any factor that assists or retards the evolution of these vapours will proportionately lower or raise the maximum point. This serves to explain very largely the varied nature of the results obtained by different investigators, which results have ranged from 13 per cent. (Redman Weith and Brock J . Itid. Eiig. Cheni. 1913 5 630) t o 25.6 per cent. (Ingle, Zoc. c i t . ) the usual amount being approximately 18 per cent. in the case of raw oil spread in thin layers on glass plates to the extent of about 0.1 to 0.2 gram of oil per 100 sq. cm. (Lippert, Zeitsch. nrulew. Chetn. 1898 11 412; Weger ibid. 502). Whilst therefore the formation of peroxides may take place a11 accurate proportionality between the increase in weight and the iodine value must not be expected Ingle's results being largely a matter of accident.The foregoing methods of determining the drying powers of oils by noting the alteration in weight is thus merely of comparative value and only serviceable when carried out under precisely similar conditions for the diff erent oils concerned. It is essential t o emphasise these points in connexioii with the s u cc e ed ing results . Although the change in weight has been made the subject of repeated research no records appear to exist of any attempt to study the corresponcling changes in volume. That some change takes place is evident from the fad t h a t thick layers of oil in-variably crinkle on setting The results in table 111 were obtained as follows: Determinations 1-4 were made by passing air through a weighed quantity of pure Calcutta oil contained in a flask noting the alteration in weight and subsequently finding the density a t cliff erent temperatures, The fourth determination is slightly less accurate than the-previous three as the oil had thickened and frothed a small quantity rising in the form of bubbles up the flask and setting in the neck.The error was certainly small but appeared impossible to avoid. The device employed by Procter and Holmes ( J . SOC. C'hem. Znd. 1905 24 1287) could not be adopted as the oil was relatively near its setting point and any drops rapidly solidified when separated from the bulk of the liquid. I n a somewhat similar experiment Mulder (" Die Chemie der austrocknenden Oele," Berlin 1867 p. 114) found t h a t his oil became solid after increasing in weight by 10.5 per cent.This however can only be an approximate figure owing to frothing so no attempt was made to coiifirm it the remaining data in table I11 being- obtaine 166 NOTES ON THE EFFECT OF HEAT AND OXIDATION ETC. iii another way. The oil was spread on glass plates in thin layers containing about 0.1 to 0.2 gram per 100 sq. cm. This thickness gives trustworthy results (Lippert loc. cit. ; Weger Zoc. c i t . ) and has the advantage that i f through unequal distribution the oil is even considerably thinner a t any place no appreciable error is entailed (Wise and Duncan J . Znd. Bng. Chem. 1915 7 202). The alterations in weight were noted and the densities deter-mined by removing portions of the film with a knife under air-free water to which sodium chloride solution was subsequently added until the liquid possessed the same density as the solid.The fifth determination was difficult to obtain as the tacky film adhered to the knife. Eventually however the film was scraped off with crystals of zinc sulphate and as these dissolved in the water the oil was left without attachment and sank or floated according to circumstances. It is unfortunate that tha fourth and fifth results should be slightly less accurate than the others since it is impossible to decide from them whether any contraction or expansion takes place at the time of setting. TABLE 111. Experi-ment. 1 2 3 4 5 6 7 8 9 Percen-tage increase Condition in Density of oil.weight. a t 0". Raw oil ... ... - 0.94208 Liquid ... ... 2.08 0.95906 Do. ... ... 5.83 0.98736 Thick frothy liquid ... 9.66 1.01161 Tacky ... ... ... 14.14 -Just set ... ... 17.34 -Solid linoxyn ... 17.90 -Do. at point of maxi-mum weight ... 18-57 -Do. three months old 10.3 -Coeffic-cient of expan -Density sion a t 15". 0-15". 0,93179 0.00074 0.94850 0.00074 0,97696 0.00071 1.00123 0*00069 1.0424 -1.0582 -1.0656 -1.0902 -1.1054 -Perccn-tage increaso in volume a t 15". -0.28 0.87 2.06 2.0 3.3 3.1 1.4 -7 Under the particular conditions of the experiments therefore, the raw oil on setting a t 1 5 O expanded by 3.3 per cent. and then slowly contracted. The maximum increase in weight was 18'57 per cent.but the increase in volume reached its maximum befqre this namely at the setting point. The density of the oil Bteadily increased whilst the coefficient of expansion fell. Sabin ( J . Incl. Eng. Chem. 1911 3 84) mentions that a film of raw oil exposed to the air for eight months yielded linoxyn of density 1.098 the total gain in weight of the oil being about 2 per cent. No statement is made of the density of the original oil but assuming it to have the value of 0.932 the shrinkage must hav RELATION BETWEEN CHEMICAL CONSTITdTION ETC. 167 been 13.4 per cent. This lends support t o determination 9 in table 111. It is important to observe that the expansion is a function of the increase in weight. Any factor assisting the decomposition of the oil lowers the maximum point and reduces the expansion. Sabin (loc. cit.) and Gardner ( J . Ind. Eng. CJbenz. 1914 6 91) found that linseed oil when mixed with even chemically inert powders such as barytes and silica exhibits a smaller increase in weight 011 setting the powder apparently catalytically assisting the decomposition of the peroxide compounds. It would thus appear difficult to calculate the amount of expansion t o be ex-pected in any particular case. Tauber (Chern. Zeit. 1909 33 85 94) suggests that the crack-ing of paint surfaces is due to tension caused by electrical action of the pigments suspended in the linoxyn. In view of the fore-going results it is evident that the contraction suffered by the linoxyn on prolonged exposure to air is quite sufficient explanation for the cracking of old paint,. THE VICTORIA INSTITUTE, WORCESTER. [Received February 15171 1917