Isolation of Sulfur Compounds from a California Straight-RunGasoline The present investigation was undertaken to examine a product obtained by treating high-sulfur naphtha with anhydrous aluminum chloride at ordinary temperature. Aluminum chloride was found to remove sulfur compoundspractically quantitatively in the form of a heavy fluid sludge, which was readily separated from the naphtha. Hydrolysis of the sludge liberated an oil which analyzed 20 weight per cent sulfur. Seventy per cent of the oil boiled in the original naphtha range. Close-cut fractions of this material yielded several thiophanes identical with those previously reported in the literature. Improved methods of segregating individual sulfur compounds are reported.
L
ITTLE progress has been made in the art of removing sulfur compounds from petroleum oils, with the result that otherwise valuable high-sulfur crudes are discriminated against by the refiner. The only sulfur compounds removed commercially a t present are the mercaptans, but these constitute only a small portion of the total sulfur compounds which are present in a sulfurous oil. The residual sulfur bodies are known to be polymethylene sulfides, commonly called “thiophanes”. Of these compounds our knowledge depends on the results of & few academic studies which, though successful, have failed to produce techniques which can be expanded to commercial processes. The removal of sulfur compounds from high-sulfur crudes can be justified by several considerations. High-sulfur crudes are discriminated against because they are extremely corrosive to refinery equipment and are difficult to r e h e to low-sulfur oils. From the petroleum refiner’s viewpoint the elimination of sulfur in any form is an end in itself, and various catalytic processes such as hydroforming have been developed toward this end. Hydroforming eventually converts the sulfur compounds to hydrogen sulfide, sulfur dioxide, and sulfur. It is possible that the organic sulfur compounds originally present, if they could be segregated in sufficiently pure form, might eventually find novel applications and prove t o be more valuable than decomposition products such as those derived from hydroforming operations. At present straight-run blending naphthas are important, particularly since demands for aviation gasoline have developed. Here a technology for sulfur removal would be desirable, since gasolines high in sulfur respond poorly to addition of tetraethyllead, and traditional acid treatment represents an extremely costly method of sulfur reduction.
0.L.POLLY,A.C. BYRNS,AND W. E. BRADLEY Union Oil Company of California, Wilmington, Calif.
Of less interest to petroleum technologists, but of potential interest to others, is an enormous supply of organic sulfur compounds with unique chemical and physical properties and with possible utility as organic intermediates and special solvents. The availability of these compounds not only opens new horizons among the sulfur compounds themselves but, because they contain a reference point, may provide a key to the riddle of the structure of the higher naphthenes, of which these compounds may be considered to be the analogs. Many California crudes are noteworthy for their sulfur content. For example, crudes from Santa Maria Valley contain up to 5 per cent sulfur and yield gasolines carrying 0.7 weight per cent sulfur. No work appears to have been published on the nature of sulfur compounds in virgin naphthas from these stocks, although it has been tacitly assumed that the compounds would be the same as those found by Mabery and his associates in Canadian oils (3). This assumption has been found correct. An excellent summary of the work on the isolation of sulfur compounds from petroleum was written by Reid (4). The classical work on the isolation of petroleum thiophanes was done by Mabery and Quayle (5) who found that the compounds could be isolated as difficultly soluble mercuric chloride complexes. By a system of fractionation a homologous series of ring sulfides was obtained. We have found that the sulfur compounds in Santa Maria Valley straight-run gasolines form complexes with aluminum chloride a t ordinary temperatures to produce a dark-colored fluid sludge in which substantially all of the sulfur is concentrated. The separated sludge may be hydrolyzed with water to recover sulfur compounds together with varying proportions of hydrocarbon material. The present investigation was undertaken to study the oils obtained by this aluminum chloride treatment.
Santa Maria Valley Gasoline The gasoline selected for extraction was a typical straightrun naphtha obtained from Santa Maria Valley California crude. Its Engler distillation properties were: Gravity Initial b. p.
50.2’ A. P. I.
111’ F. 166‘ F. 217’ F. 252O F.
70% over 90% over 95% over Max. over
Recovery
284’ F. 320’ F. 346’ F. 364O F. 97%
The total sulfur content of the gasoline was 0.39 per cent by weight. A preliminary examination indicated that only negligible amounts of mercaptans and thiophenes were present and therefore that the sulfur compounds present were presumably alkyl sulfides, thiophanes, and similar saturated materials. In order to determine the sulfur distribution, a small portion of the above gasoline was carefully distilled in
7ss
756
Vol. 34, No. 6
INDUSTRIAL AND ENGINEERING CHEMISTRY
a twenty-theoretical-plate laboratory column packed with Nichrome helices. The fractions were analyzed for sulfur by the modified Ethyl Gasoline Corporation method Figure 1, curve A , shows the sulfur distribution in weight per cent throughout the gasoline, except for the region between 21 and 42 volume per cent overhead where no analyses were made.
Extraction with Aluminum Chloride A total of approximately five barrels (250 gallons, 1651 pounds) of the above gasoline were treated with slightly more than 2 moles (55 pounds) of technical anhydrous aluminum chloride per gram atom of sulfur present. The operation was
VOLUME PER CENT
carried out by treating one-barrel portions in a cone-bottom stainless steel agitator fitted with a motor stirrer. The aluminum chloride was added portionwise over a period of an hour, and stirring was continued several hours thereafter. The course of desulfurization, counting from the time the first aluminum chloride was added, was as follows: Sulfur in Elapsed Time, R a 5 n a t e Oil, Elapsed Time, Hours wt. % a Hours 4.25 0.0 0.39 5.25 1.0 0.097 6.25 2.0 0.092 3.25 0.053 4 Determined by modified Ethyl Corporation method. b Two other batches were 0.005% eulfur.
Sulfur in Raffinate Oil,
wt. yo'
0.039 0.032 0.016)
After settling overnight, the fluid sludge layer was drawn off and washed twice with pentane. Some hydrocarbon oil, essentially of naphthenic character, remains in the sludge. The high viscosity of this sludge makes complete removal of occluded or dissolved hydrocarbons difficult. It is possible that some of the aromatic hydrocarbons may form similar aluminum chloride complexesin the presence of small amounts of hydrogen chloride. Hydrolysis of the sludge was effected by pouring it onto chopped ice. The aqueous phase was then drawn off and the oily phase washed with water. The aluminum oxychloridehydroxide floc is best separated from the oil by addition of Super-Cel. The floc may then be washed with pentane. The total pentane-free extract weighed 26.2 pounds, equivalent to about 1.6 per cent weight yield.
Fractionation of Extract Oil The extract was a fluid, dark-green oil with the following properties: Gravity Initial b. p.
;:q E:; 5 0 4 over
22.6' A . P . I .
70
286' F.
Max. over Sulfur
156' F.
314' F. 330' F.
oTer
90% over
352' F. Cracked 682' F. 17.5% by wt.
Evidently more than 70 volume per cent of the oil distilled in the original gasoline range. The remainder appeared to be high-boiling, unstable material. The green oil was steamdistilled into four overhead portions, leaving a heavy viscous residue. The distillates, in turn, were subjected to a program of precise fractionation in twenty-plate glass stills, vaccum-jacketed and packed with Fenske helices. Following the first pass the individual cuts were shaken with 50" BB. caustic solution. In this way acidic materials were removed without the formation of troublesome emulsions. Related fractions were recombined and redistilled until the properties of the cuts were essentially constant (five to six passes). Figure 1, curve B , presents a distillation obtained by adjusting all boiling points to atmospheric pressure. A b o v e t h e 140" C. f r a c t i o n t h e t e m p e r a t u r e s ,given a r e f o r c u t s which had only one or two passes through the stills. Fractions boiling below 105' C. contained very little sulfur. (Tetraniethylene sulfide boils a t 119" C.) Chief attention was directed to material boiling between 109" a n d 139" C . T h e p r o p e r t i e s of frac0 tions obtained in this region are listed in Table I.
Preparation of Mercuric Chloride Complexes Further resolution of the fractions into their sulfur-bearing constituents was effected through mercuric chloride complexes according to the classical methods of Mabery and his associates ( 3 ) . However, complexes formed from water or alcohol solutions tended to be tinted pink. These products are lightsensitive and become black on exposure. The phenomenon is probably due to mercuration and may be avoided completely by working in ether or acetone solutions as described later. Apparently Faragher, Morrell, and Comay (2) were the first to appreciate that alkyl sulftdes form. both mono- and dmercuric chloride complexes. The troublesome liquid complexes mentioned by early investigators may be avoided by using 2 moles of mercuric chloride, whereupon stable solid products are obtained which are nearly always the di-complexes. These complexes may be recrystallized repeatedly and, after they have been purified, can be left in open beakers for a month without change in melting point. The use of dicomplexes has greatly expedited the labor of fractionating by recrystallization, since the mono-complexes frequently undergo a gradual change to the di-form with a consequent loss of thiophane in the mother liquor. The determination of melting points of the complexes is also beset with some difficulties. Thierry (6) in his study of complexes from a Persian petroleum observes: "Sintering frequently occurs during melting point determinations and this with the tendency t o dissociate makes trustworthy observations difficult
INDUSTRIAL AND ENGINEERING CHEMISTRY
June, 1942 L
OF DISTILLATE FRACTIONS TABLEI. PROPERTIES
B. P. (Atm. Cut No. RL-1 2
Pretsure), C. 109 121
3 45 6 7
125.5 128.9 130 131 133.5 134 134.5 135.5 136.3 136.5 137 137 138 139 139 139 139 140
3
9 10-1 10-2 11-1 11-2 12-a 12-1 12-2 12-3 12-4 12-b
Sulfur,
ny
% by Wt.
1.4581 1.4262 1.4425 1.4489 1.4466 1.4438 1.4428 1.4450 1.4595 1.4600 1.4620 1.4640 1.4664 1.4685 1.4718 1.4732 1.4734 1.4734 1.4722 1.4734
4.1 9.3 6.6 1 15 3 .. 8 5 9.3 5.4 3 .. 56 7 8.2 9.1 7.6 10.6 12.7 17.5 20.4 20.0 20.0 17.0 24.4
Volume,
M1. 160
111
31 45 28 20 40 132 51 52 51 51 51 53 46 60 53 50 55 55
751
stirring. In a short time the complex begins to crystallize. Often a fair separation of the less soluble complex occurs under these conditions. The more soluble material is then fractionally crystallized from the warmed mother liquor by judicious addition of pentane. A standard fractionation scheme directed by the results of melting point determinations is followed until discrete constant products appear. Meltin points were determined in slightly oversize capillary tubes. $he sintering temperature is taken when the walls first appear wetted. At the melting temperature the bottom of the tube rapidly Ells with liquid. No special precautions were taken in heating the bath. The temperatures are uncorrected on standardized thermometers. Oils were recovered from their complexes by solution of the complex in concentrated hydrochloric acid under a layer of pentane in a separatory funnel. One extraction sufficed. The pentane layer waa water-washed, filtered, and evaporated. An excess of pentane checks the partition of undecomposed complex and mercuric chloride into the oil layer.
TABLE11. DISTRIBUTION OF MERCURIC CELORIDE COMPLEXES= AMONG DISTILLATE FRACTIONS' ACCOWING TO MELTING POINT (IN O C.) RL-1 2 3 4 7 8 110) 110 9 i107) 108 10-1 10-2 11 12-1 P platee, N
-
(126) 127 A
(125) 128 P
1
(148) 150 (146) 148 (131) 131 P
B
1
(127) 129 P
-
82;; i f 5 (159) 160
(148) 149 146) 148 N 11431 143 {I,,) 147 N F 143 143 146) 146
needles.
TABLE111. MAJORCOMPLEXES ISOLATED
to obtain.''
We used the difference between sintering temperature and liquefaction temperature as a criterion of purity, since repeated recrystallization shows that the two temperatures tend to coincide. Further, the presence of another complex in the sample may or may not lower the melting point, but can usually be detected by cloudiness in the melt. Table I1 attempts t o classify all fractions which satisfied reasonably well the criteria of constancy of melting point following recrystallization and sharpness of melting point. The numbers in parentheses are the sintering temperatures; the others are liquefaction temperatures. The major complexes are evident, since they were obtained from several fractions. Their distribution is interesting in view of the number of distillations each fraction has undergone. The lesser complexes were obtained in amounts too small for characterization. Their behavior on recrystallization, in several cases, suggested they were mono-complexes. Corresponding complexes from the various RL fractions were grouped and subjected to final recrystallization. The more important complexes so obtained are presented in Table I11 together with their properties and the results of analysis.
Regeneration of Sulfides from Complexes Several methods for regenerating organic sulfides from their mercuric chloride complexes are available. Mabery and Quayle used hydrogen sulfide, Thierry used concentrated hydrochloric acid, Faragher, Morrell, and Comay used sodium thiosulfate, and we have found that aqueous ammonia is satisfactory. A given complex decomposed by any of these methods yields oils having identical ProDerties. We Drefer the hydrckhloric acid metcod as outgned later. The iegencrated oils from the more important complexes are described in Table IV.
Procedure The complexes are best prepared by dissolving the required
amount of mercuric chloride in a considerable excess of acetone. The crude thiophane is then poured into the cool solution with
Boilhg Range, C. 109
Fraotion No. RL-1 4
1
8 10-1
1
12-1 10-2 11-1,2 12-1
(126) 127
N
Analysis" RS.HgCli
121-9
B
(158) 159
P
RS.2HgClr
134
C
(126) 126
P
RS.HgClr
134.5-136.3
D
(137) 138
N
RS.2HgClr
136.3-139
E
(168) 169
N
RS.2HgCIr
136.5-139
F
(147) 148
N
RS.2HgClr
-
The values in parentheses are the sintering temperatures. needles, P Analysis by metCo$?sfte$aragher, Morrell and Comay (I).
b N 8
1
Complex y .C.0' P. Form)
No. A
TABLEIV. PROPERTIES OF OILB FROM MERCURIC CHLORIDE COMPLEXES nF
Oil B. P., C.
1.4810 1.4817 1.4830 1.4770
IS6 5 138.5 138.5 139.5 139.5
I
Complex A
B
C D E P
M. P., 126) 158) 126) 137) 168) 147)
I
' C. 1271,
159 126 138 169 148
.... 1.4909
Analysis0 CbHk
....
dH1;S C6H118
Analysis by Ethyl Corporation lamp-sulfur determination ( 8 ) . b Insuffiioient sample available.
Boiling points were determined bv Siwoloboff's caDillarv-cuD method a f t e r which the sample W& used for refrakive hdek and suliur analyses. The temperatures are uncorrected. A molecular weight determination on oil F , Table IV by Rast'a method gave good results (found 116, theoretical 116.2)when the camphor blends were weighed into tubes, sealed, and repeatedly melted and cooled until the transition temueratures were constant. Calculated molecular weights from- determinations at several concentrations were then extrapolated to zero oil concentration.
758
I N D U S T R I A L A N D E N G I N E E R I N G CHEMISTRY
Identification of Complexes Our data do not indicate the presence of complexes such as RS.3HgC12 or (RS)z (HgC12), which have previously been reported. We are in substantial agreement with the work of Teutsch (6) who isolated thiophanes from Panuco (Mexican) crude. Teutsch reported the complexes listed in Table V, together with their identification from data in the literature. With no desire to belittle the excellent work of Teutsch, we are inclined to disagree with several identifications of Table V since the presence of simple pentamethylene sulfides does not seem t o have been established. We concur in identifying our complex A as that of tetramethylene sulfide. It is of interest to note that we found tetramethylene sulfide, as did Teutsch, in the 105-110” C. gasoline fraction, although the pure compound boils a t 119” C. according to the literature. Complex R, which appears to be Teutsch’s compound melting a t 152-154” C., is believed to be that of 2-methyl tetramethylene sulfide since i t agrees closest with the data for Trochimovski’s compound (7) which melted with decomposition at about 150” C., and was made from an oil of 132.5” C. boiling point and 1.4886 refractive index at 18.5’ C. Table IV shows that the oil obtained from this complex boiled a t 130.5’ C.
Vol. 34, No. 6
that these compounds are the di-complexes of 3-methyl tetramethylene sulfide and not pentamethylene sulfide. Complexes E and F do not fit any data found in the literature. The low refractive indices of oils recovered from these complexes suggest that the rings are tetramethylene sulfides bearing alkyl groups. These oils, as well as the fractions from which they came, have rather pleasant camphoraceous odors. This agreed with Thierry’s observation on the Persian-oil thiophanes. Apparently the “extremely penetrating putrid odors’’ reported by investigators (1) who have prepared thiophanes synthetically have been due to side products, probably unsaturated mercaptans.
Discussion
If the boiling points of the oils recovered from complexesare located on boiling-point curve B of Figure 1, an interesting correlation with curve A appears. Each thiophane is associated with a high-sulfur region in the original gasoline. It seems likely, therefore, that the aluminum chloride extraction has not altered the nature of these compounds and that they also existed as thiophanes in the crude. Further, these oils are completely soluble in cold concentrated sulfuric acid and can be recovered unchanged on dilution with water, a fact which indicates their stability. In passing, it is interesting to note that Thierry (6) recovered thiophanes from acid sludge. However, in the present investigation i t was found that because of a number of factors, particularly distribution ratio, TABLEV. MERCURICCHLORIDECOMPLEXES FROM PANUCO aluminum chloride is far more effective than !sulfuric acid for CRUDE the segregation of thiophanes. It was beyond the scope of -FractionBd P y.C.P., Complex the project to determine the structures of these thiophanes. C:’ Composition Identification The high refractive indices suggest that no alkyl sulfides are 105-110 126 C4HsS. HgCh Tetramethylene sulfide present. 120-125 125-7 CaHloS. HgClz rr-Methyl tetramethylene sulfide The curves of Figure 1 suggest that even more interesting 120-130 137-9 CaHloS . HgClr Pentamethylene sulfide results may be anticipated in the fractions which boil above 152-4 C1HioS.2HgClz 140-143 163-5 CeHizS. HgClz i 140” C. At this point, however, one passes the frontier. There are no literature references suficient to identify any compounds which may be encountered. Cursory experiments TABLEVI. COMPARISON OF PETROLEUM THIOPHANES WITH SYNTHETIC THIOPHANES suggest that the higher sulfides may be true thioterpanes resembling the cineoles. The authors are tempted to call Troohimovski (7) attention to the unique position of the thiophanes in the Author’s Penta@-Methyl Complex methylene tetramethylelucidation of naphthene chemistry. Since they are “vulD Teutsch (6) sulfide enesul6de nerable” a t a reference point, they are readily susceptible to B. P. of parent fraction, isolation, purification, and structure studies. There is no c. 134-6 132-6 137-9 lit: 5 ail3 M. P. of complex e C. (137) 138 reason to believe the outlined methods cannot be extended to 2 1 1 No. HgCln rnoledles Crystal form Needles .1. . Plates .. the higher boiling fractions we have obtained, except that the number of constituents, even in a close-cut fraction may Oil by Synthesis Oil from Complex be very large. A gas-oil cut was treated with mercuric chloB. P., C. 138.5 ... 141.5-42 137-8.5 (747 mm.) (740 mm.) ride and a crystalline product was obtained, but no further 1.4817 ... 1.5046 1.4886 nv work was done on it. A great aid to further research in the (18.5’ C.) (18’ C.) field would be t o prepare pure mercuric chloride complexes of synthetic thiophanes and determine their properties and the properties of the oils recovered from them. Complex C, which is evidently Teutsch’s compound meltLiterature Cited ing a t 125-127” C., is believed to be 3-methyl tetramethylene sulfide monomercuric chloride. This conclusion is based on (1) Bost, R. W., and Conn,M. W., Oil Gus J., 32, 17 (June 8,1933). (2) Faragher, W. F., Morrell, J. C., and Comay, S., J. Am. Chem. the boiling point and refractive index of the oil recovered Sac., 51, 2774-81 (1929). (C, Table IV), in spite of the fact that Trochimovski gives the ( 3 ) Mabery, C . F., and Quayle, W. O., Am. Chem. J., 35, 404-32 melting point of the “beta” complex as 83” C. (TableVI). (1906). Complex D has the same melting point as Teutsch’s com(4) Reid, E. E., “Science of Petroleum”, Vol. 11, pp. 1033-41, London, Oxford Univ. Press, 1938. pound melting a t 137-139” C. but is a dimercuric chloride, (5) Teutsch, I., Petroleum Z.,30, 1-6 (May 16, 1934). while his is reported to be the mono-derivative. Apparently (6) Thierry, E. H., J. Chem. SOC.,127,2756-9 (1925). Teutsch’s identification hinges upon this analysis, which fits (7) Trochimovski, E.G., J . Russ. Phys. Chem. SOC.,48,901,928,944 Trochimovski’s description of pentamethylene sulfide mono(1916);Brit. Chem. Abstracts, 1917,i, 164-7. (8) Wilson, V. W., Buell, A. E., with Schulze, W. A., Oil Gus J.,37, mercuric chloride. Our difference of opinion is based on the 76-8 (March 23,1939). properties of the oil recovered from complex D,which we have assumed to be identical with Teutsch’s complex. Table VI PREBBNTED before the Division of Petroleum Chemistry at the 103rd is presented to summarize the data which seem to indicate Meeting of the AMBRIOAN CHEMICAL SOCIETY, Memphis, Tenn. O