Decomposition of Saturated Petroleum Hydrocarbons under Oxidizing Conditions at Low Temperatures A. W. BURWELL, Alox ChemicaI Corporation, Niagara Falls, N. Y.
T
HE results obtained as lytic agent or the height of the This paper treats of the liquid-phase oxidapresented here are the column. lion, mostly by means of air, of hydrocarbons outcome of several years In general, the increase in temderived from various petroleurn sources, under of practice, primarily by means perature (pressure remaining the controlled conditions of temperature and presof air under pressure, varying same) causes more breakdown of sure, the temperature always being mainfained from 200 to 500 p o u n d s p e r molecules than increase of pressquare inch above atmospheric, sure (holding the temperature considerably below the thermal decomposition and of oxygen under pressures up constant). However, such intemperalure of the hydrocarbons. The react o 250 pounds, beginning a t creases in temperature do not tions given hazte been deduced from the products about 150. The temperatures approach anywhere near crackformed. The oxidation reactions in the case of range close to 130" C. (2S6'F.). ing temperatures. The temhydrocarbons at cracking temperature, and those Experiments were made as low as peratures employed were so far 120' and as high as 180" C. b e 1ow cracking temperatures oxidation reactions taking place where ozone or (248' and 356" F.). but seldom that stabilization of the molecule other oxygen-addition compounds haue been e x c e e d e d 160" C. (320' F.). by pressure could not be considformed w i f h unsaturated hydrocarbons and subThe apparatus used is a vertical ered as occurring by such variasequent heating, are discussed briefly. vessel of tower-like proportions. tions of pressure as were emthe height usually being five to ployed here. Again, it is unsix times the diameter. derstood that the oxidations considered here are confined t o These oxidations take place either in the presence of those reactions of oxygen with petroleum a t temperatures catalysts of various kinds (mostly manganese in the form which do not cause, by reason of the degree of temperature, of soluble fatty acid salts, usually the oleate) or in the pres- any decomposition in the molecules of the hydrocarbons ence of the partially oxidized, unsaponifiable portions of under discussion other than that already noted. the previous charge of hydrocarbons, The start for obVARIOUSCATALYTIC AGENTS taining such unsaponifiable but partially oxidized products is invariably made with a metallic catalyzer. The catalyst Of catalytic agents which appear to have a greater influence ranges froni mercury or cobalt to manganese, iron, and lead. on the oxidation and to cause the formation of undesirable All have approximately the same general effect, and it was substances such as asphaltic bodies, etc., cobalt and mercury found that any soaps of those metals which have more than seem to show this tendency more than any others examined. one valence would act as catalytic agents for the introduc- Lead soaps, made from some of the acids obtained from pretion of oxygen into the molecule of hydrocarbons. vious oxidations which had been catalyzed by means of manganese oleate, appear to have a very much milder action EFFECT OF SELECTIOK OF CLASSOF HYDROCARBOX and to necessitate more careful watching of minimum temIf the amount of straight-chain hydrocarbons was in- peratures than others. While it is probably true that creased in the distillate or other material being oxidized, scission of the carbon-carbon bond occurs under such oxidizeither by reason of crystallization (paraffin waxes) or by ing conditions, this can in no wise be attributed to temperatreatment with destructive or selective solvent agents of ture beyond that necessary to support the reaction. The various kinds to remove those hydrocarbons which were not scission is caused entirely by oxidation and not by temperaof the straight-chain order, such straight-chain hydrocarbons, ture. However, no matter what the degree of divergence without the presence of catalyst, oxidized with greater readi- from a normal general reaction may be, it is practically always ness than any of the other hydrocarbons or mixtures of other too small to be of much influence on the whole outcome when hydrocarbons. I n the case of many of the mixtures, es- we consider the main reaction. Previous to the publication of the first patent granted to pecially where apparently straight-chain hydrocarbons were in minor quantity, the amount of catalytic agent or of pre- the author, practically all of the patents then in existence viously oxidized but unsaponifiable material had to be in- assumed, apparently, that the reaction leading to the formacreased in order to make the reaction self-supporting, so far tion of acid was caused by the oxidation of the CHa grpup. as temperature, under the circumstances of the entry of air It might be observed that at higher temperatures, especially cracking temperatures, the oxidation is such as to attack or oxygen, and pressure were concerned. preferentially the end carbon atom, forming primarily aldeEFFECTO F PRESSURE AND TEMPERATURE VARI.4TIONS hydes (3). I n general, the greater the concentration of oxygen by COURSEOF REACTION UXDER CONTROLLED CONDITIONS reason of pressure, of purity, or of both, the more readily the An attempt was made to find out what the course of the material would oxidize. .41so, by elevating the temperature to a higher point, the oxidation could be induced to continue main reaction really was. It was easily reasoned from the and to be self-supporting above a given point, all other condi- analysis of end products that in practically no case was the tions again being equal. In fact, any of those factors which end or methyl group attacked; that the general reaction generally would be assumed to have an effect on the rate of involved primarily the beta- and secondarily the gammaoxidation were found to have the influence expected. In carbon, and so on toward the center of the molecule. This other words, in order to obtain more rapid or more perfect would have been observed by the first workers if they had absorption of molecular oxygen, it would be found necessary, worked with quantities of material sufficient to have disall other things being equal, to increase the amount of cata- covered the relatively small amounts of volatile acids which
INDUSTRIAL AND EXGINEERING CHEMISTRY
Februar), 1934
are always formed. The first experiments reported here involved the use of 200 pounds of hydrocarbon or hydrocarbon mixtures instead of 50 grams, and shortly after it was discovered that commercial uses could be obtained for the product if oxidation was started on not less than 1000-pound charges. Then it iinmediately became apparent that the coiirw of the rrwtinn wac: a. already indicated R E \ ( T ~ O Y1
+
203
increase the apparent molecular weight of the mixture of acids remaining insoluble in water. The main mass of acids derived from paraffin wax of 122" to 124" F. melting point, averaging around twenty-four carbon atoms which have been freed from the accompanying insoluble ketonic alcoholic bodies. nil1 have a molecular weight of 255 as pointed out above. Thi., when figured for
+ CHaCBjCHO:-I~:SI'CH,),~C~~~CHI + CH,COCH,(CHz)zuCH,CH3 + CH,CHzCOCHz(CHz)igCHzCH3 + 2H20
2CII,CR2(CH2)~1CH2CH3
0 2
=
CH3CHOHCH,(CH,),,CH,CH, 0 2
= =
Hut
111
+
HCOOHHOOCCH2(CH,) gCHzCH3 H~CCOOHHOOCCH~(CH~)~~CH~CH~CH~ the meantime, oxygen ha.. attached itself:
HOOCCHj(CHz)l5CHOHCHiCHzCHj
01
HOOCCH2(CHi)ioCHOHCH~(CHz)5CHZCHj
There seems to be no reason why more dibasic acids should not be formed, but only very small amounts (a pound or two per ton of product,) have been found, nor does the amount of formicacetic acids permit of the formation of dibasic acids. The amounts of the lower molecular weight acids up t o CIOfrom Cs are so small (small fraction of a per cent) that they cannot be considered in this connection.
REACTIOX 111. CH,CHzCH,CHz(CH,),CH?CH,
+ 20,
When it is considered that from 10 t o 15 per cent of formic and acetic acid is formed, the formic being always in greater quantity than the acetic (equivalent weight considered), it becomes evident that the reaction producing formic acid is more general than that which produces acetic. The main reaction is therefore for the oxygen first to enter into the beta-carbon atom in the chain in paraffin wax, aswming the latter t o be a straight-chain material. In the course of time, and probably with the introduction of other oxygen atoms into the chain at other points, this particular carbon atom becomes involved in a loss of hydrogen and becomes a ketone group. Then with sufficient oxygen in Yolution in the whole mass, under the influence of the general reaction, the chain splits a t the ketone group with the formation of two acids. As is already known, the ketone group almost always goes with the larger molecules and we have thus formic aCid and an acid of higher molecular weight which may be the rest of the hydrocarbon.
REACTIOXS IXCREASING OXYGENIN MOLECULE Simultaneously, or during the time that this reaction is taking place, other atoms of oxygen are added to molecules a t other points so that the mixture of acids which we finally recover will be found, from a careful checking of equivalent weights, to have an average equivalent weight of approximately 255 where the material is made from paraffin wax having an average of twenty-four carbon atoms. Such acids correspond to an average of twelve to fourteen carbon atoms 71 hen this mixture of hydrocarbon is used. The production of acids having somewhat more than the average molecular weight, if the chain had split in the center, i q undoubtedly due to the solution or volatilization of the rpadilv water-soluble or yolatile acids. This would naturally
fourteen carbon atoms, is found t o correspond closelj when we consider that we almost invariably have either a ketone group, a hydroxyl group, or both in addition. I n such cases the presence of the hydroxyl group is proved by acetylation, and t'he presence of the carbonyl group by acetylation, reduction of the mass, and reacetylation, the carbonyl group going back to the alcohol group by reduction with mercury amalgam in about a week, and the reacetylation showing the formation of the additional hydroxyl group. Care must be taken in making these acetylations t o be *ure that any lactones which might be formed by reason of the location of the additional hvdroxyl group are held open
CH$2H,CHOOHCCHz(CHz)sCHOHCH?
+ HzO
by forming the esters of the acids which would otherwise immediately form lactones upon liberation from their salt.. I n the wash waters are also found lactic acid and several of its homologs, as well as succinic acid and one or two of the other dibasic acids in small quantities. Theqe reaction6 are easy to construct from the above information. However, the main amount of recovered material runs well toward 100 per cent of the original hydrocarbon used, whereas the total product, including water formed by oxidation of the hydrocarbons, would undoubtedly go well over 150 per cent. The following acids have been separated: formic, acetic, propionic, butyric, valeric, etc., up to and including C l ~ ] in lessening quantities; none of them is seriously decomposed by boiling a t atmospheric pressure. The attempt to separate these acids more closely by fractionating their esters has not been made here, but fraction. having close to the molecular weight of the acids and boiling points reasonably close to those given in the literature have been separated. Those derived from paraffin wax have the odor and the appearance of the straight-chain acids. They differ in some other respects because they are never single bodies (pure). Even up to Clo, as disclosed by their molecular weight, they do not solidify even a t low temperatures, the small amount of other impurity doubtless preventing any such solidification; it is doubtful if they are more than 90 per cent individuals, if that. Judging that the lighter distillates which distill up to cracking temperatures from Pennsylvania petroleum are mixtures of straight-chain and branch-chain hydrocarbons, acids made from such hydrocarbons have quite a different appearance and usually a very strong and disagreeable odor; here again it i q even more difficult to separate individuals. Mixtures
206
INDUSTRIAL AND ENGINEERING CHEMISTRY
can be reproduced of approximately the same equivalent weight a t the same temperatures of distillation, but here, too, impurities of other natures creep in and are hard to eliminate. Upon each redistillation, more of the materials will break down at the higher temperatures particularly, and unsaponifiable bodies are formed. Distillation with steam will give somewhat better odor and almost invariably better color, but fractionation in the presence of steam is not so satisfactory, and it is more difficult to produce mixtures with any given equivalent weight. However, these acids are easily produced in quantity from equivalent distillates. As the loss in manufacture is rather large, the final yield of good colored, reasonably pure acids, as acids alone, seldom exceeds 60 or 75 per cent of the original hydrocarbon used. They are fairly easy of production and many uses have been found for them. These acids in particular reaching boiling points of 300"C. (572" F.) have a peculiar property. They do not become much more viscous when cooled to -60" F., and in no case do they become solid. They produce rather attractive looking, oilsoluble salts of the heavy metals such as lead, manganese, cobalt, etc., and are used to some extent for making driers. As the equivalent weight can be controlled quite readily, it is possible to obtain as much as 40 per cent of actual lead in solution in gasoline.
ESTERFORMATIO~ The formation of esters is less marked during oxidation in the case of paraffin wax than it is with most distillates. In the case of distillates the saponification to recover the acids must be with a degree of excess alkali and longer heating with agitation. Even then a good portion of the ester is not broken up. It seems to be generally accepted that most esters of secondary alcohols saponify less readily than esters of primary alcohols, and it is probably true (according to observations made here) that esters of the tertiary alcohols, which would result from the oxidation of petroleum distillate containing much branched-chain material, are very difficult of saponification. We find, for instance, after we have decomposed the oxidation mass which has been oxidized to a point where there is practically no unoxidized hydrocarbon left, that our recovery of actual acid with 25 per cent alkali and in considerable excess, does not exceed 35 per cent of the oxidized mass. If, however, we separate this soap and then attack the remaining part of apparently unsaponifiable matter with 50 per cent caustic and a t higher temperatures, we can easily recover another 30 per cent from the remaining mass. This would indicate that there qhould be present then at least 60 to 75 per cent of esters, as the molecule of alcohol is probably larger than the molecule of acid, on the average. A glance at the reactions will make this plain. This would mean that with 30 per cent acid we might even have a mixture of esters of secondary alcohols and possibly esters of tertiary alcohols. In a11 cases lactones are present, but for the purpmes of this paper lactones are considered as esters. The author i i rather of the opinion, based admittedly on too few observations, that the secondary alcohols form esters which, however, are not too difficult of saponification in open tanks with sufficient boiling and agitation, but that esters of the tertiary alcohols are much more difficult and therefore form by far the larger part of the so-called unsaponifiahle matter derived from a given distillate. There is another possible reaction which should be given consideration. It has been claimed that some of the higher ketones break up as follows under the influence of strong alkali (reaction 11). It is known that many esters of t h e higher alcohols break up upon distillation and form free
Vol. 26, No. 2
acids and h\-drocarbons ( 1 ) . We find, however, by returning all unsaponifiable matter which may contain small amounts of unattacked hydrocarbons, but mainly alcohols, ketonw, alcohol-ketones, and tertiary esters or lactones, that this material makes, with additional quantities of hydrocarbon, an easily oxidized mass. It, in turn, is treated in the same way, the oxidation in each case being carried to that point a t which initial quantities of the petroleum-insoluble, oxidation mass-insoluble materials are formed. It is these insoluble bodies out of which it is not easy to recover good quantities of acceptable acids which cause the greatest losses. There is a polymerization of acidic bodies, probably together with some esters which also have hydroxyl and ketone groups capable of forming esters, which in turn also have hydroxyl ketone groups capable of forming esters again; the resuIt finally is too large molecules to be distilled even from the oxidation mass. The actual high percentage recovery from oxidation masses is best made by saponifying with caustic soda, returning the unsaponifiable portions together with new quantities of hydrocarbon to the oxidizer, decomposing the soaps, and distilling the recovered acids. In this way a somewhat better grade of acid is produced, but the cost of caust.ic soda for the purpose, together with the necessary sulfuric acid, enhances the cost of the acid to such a point that it is cheaper to lose more of the original raw material and to recover the acids from the distilled mass of oxidized hydrocarbons. REACTIONS OCCURRING I N OTHERPROCESSES Hydrocarbons, when oxidized under cracking conditions, lead to quite different results (2). These reactions, of course, are not to be considered as being in the sanie class as those already discussed. Such reactions, and those shown by the James patents (oxidations in vapor phase and at temperatures which undoubtedly produce cracking in the molecules) produce largely aldehydes. I n this case the end or methyl group is attacked, as aldehydes could hardly be formed by any other means. (A list of James's patents is given at the end of this paper.) I n the Harries process which was used on the unsaturated bodies in distillates recovered from the low-temperature distillation of coal, ozone was employed which, it is claimed, formed ozonides or peroxides. These ozonized hydrocarbons were heated, thus forming acids. The ozone probably attacked the double bond, and two acids were formed from the molecule of unsaturated hydrocarbon. It is interesting that Harries produced by such means acids from which the ethyl esters were made. Of such esters approximately 6000 tons nere fed to the German soldiers during the war, owing to t,he shorhge of fats in German!;. LITERATURE CITED S.Patent 896,093 (1908); Lewkowitch, "Cihernical Technology and Analysis of Oils, Fats and Waxes," 6th ed., Vol. 111, p. 455, Macmillan, 1923. (2) Lenher, J. Ant. Chena. SOC.,53, 3752 (1931); 54, 1830 (1932). (3) Pope, DykRtra, and Edgar. I h i d . , 51, 1875, 2203 (1929) ; 53,3752 11931).
(1) Ellis, Carlton, U.
PATENTLITERATURE NUMBER
NAME
DATE
BRITISH
Johnson, J. Y. Imray, J. Thompson, &' P. Schmidt, A. Pardubitzer E'abrik Akt.-Ges. Mineralol-Indii~trievorm. D. Fanto & Co. Schmidt, -1.
419 12.806 13,473 109,386: 131,301-3 133,027
Oct. 11, 1686 June 19, 1884 June 16, 1903 Feb. 13, 1917 .Lug. 12, 1919 I u r r . 13. 1919
142,507
April 30, 1920
February, 1934
INDUSTRIAL AND ENGINEERING CHEMISTRY
Teichner, C. Deutshe Erdol Act.-Ges. Traun, H. O.,Forschungslaboratorium Ges. Chemische Fabriken Worms Akt.-Gee Bymes, C. P. Winternits, H., Bullinger, T., and Teichner, 0. Schicht, G., and Green, A. Pataky, W. C. H., and Nillensteijn, F. J. Penniman. W.B. D.
148,358 148,892 156,141
.July 19, 1930 July 10, 1920 Dec. 1, 1920
156.252
Jan. 4, 1921
173,750 174,642-3
Dec. 21, 1921 Jan. 30, 1922
183,186 239,178
hfarch 17, 1921 July 21, 1925
252,327 255,020 256,922
March 2. 1926
295,811
Xtay 8, 1933
DUTCH
Hulsberp and Seiler
6151, 1120
June 7, 1921
FRENCH
Pardubitzer Fabrik .4kt.-Ges. Mineral01 Industrie vorm. D. Fanto & Co. Teichner, G. Bymes, C. P.
505,126
July 2:3, 1920
521,228 52,828
July 8, 1921 July 2, 1921
QERMAN
Schaal, E. Pardubiteer Fabrik Akt.-Ges. Mineralol-Industrie vorm. D, Fanto & CO. Chemische Fabriken Worms Akt.-Ges. Zollenger. J. E.
32,705 82,057
Sept. 25, 1884 Aug. 16, 1918
91,328
Oot. 17, 1921
94,233
April 17, 1922
(THIS SYMPOEIUM WAS
95,220
July 15, 1923
346,520
Dec. 2. 19213
35,621 382,496
Oct. 3, 1923
March 25, 1922
405,636 434,923 439,354
E‘eb. 15, 1924 Sov. 4, 1924 Oct. 7, 1926 Jan. 14, 1927
88,418
.May IO, 1922
:
