of Sulfoxides from Petroleum ation Exchange Resin Chromatography kuno, D. R. Latham, and W. E. Haines Laramie Petroleum Research Center, Bureau of Mines, U.S . Department of the Interior, Laramie, Wyo. Petroleum chemists have generally considered that ases in petroleum are nitrogen compounds. Evidence is presented to show that a portion of the weak bases are sulfoxides. Present evidence indicates that the sulfoxides do not occur naturally in crude oils but are readily formed in some oils by mild air oxidation. The sulfoxides were concentrated by cation exchange resin chromatography from petroleum fractions boiling from 250' to 500' C. The amount of sulfoxides, titrated potentiometrically as weak bases in acetic anhydride with perchloric acid, was quantitatively related to the intensity of the infrared S=O absorption band at 1040 cm-l and to the increase in sulfide content upon reduction with lithium aluminum hydride. Analytical data, including mass and nuclear magnetic resonance spectrometry, suggest that the sulfoxides have saturated cyclic structures.
CONSIDERABLE STUDY has been directed toward the weak bases that are present as nitrogen compounds in petroleum. These constituents, traditionally classified with the nonbasic nitrogen compounds, are less basic than the so-called basic nitrogen compounds which, as defined by Richter et ai. (I), have pK, values 2 2. The more commonly known examples of weakly basic compounds in petroleum are pyrroles and indoles. Extensive studies of methods for the analysis of these compounds have been made by Snyder and uell (2, 3'). Other weakly basic nitrogen compounds that occur in petroleum are amides (4, 5 ) including quinolones (6, 7). Recent studies in this laboratory have shown that certain sulfur compounds also behave as weak bases in petroleum. Previously, the authors reported a method for the type analysis of nitrogen bases in crude oils using a combination of nonaqueous potentiometric titration and lithium aluminum hydride reduction (5). The method differentiated compounds that titrate as weak bases in acetic anhydride into three types, BI, Bz,and Ba. Nitrogen compounds that would be classified as types B1and B3 had been reported to be present in petroleum. Nitrogen compounds that would be classified as type Bz were not known. Type B2 bases, as defined by the type-analysis method, are those compounds that titrate as weak bases in acetic anhydride and are reduced with lithium aluminum hydride to compounds that are not titratable. Although the type Bzbases were believed to be nitrogen compounds at the time of the development of the method, more recent studies have shown these bases to be sulfoxides (1) F. P. Richter, P. D. Caesar, S. L. Meisel, and R. D. Offenhauer, Ind. Eng. Chem., 44,2602 (1952). (2) L. R. Snyder and B. E. Buell, ANAL.CHEM., 36,767 (1964). (3) L. R. Snyder and E. E. Buell, Anal. Chim. Acta, 33,285 (1965). (4) N. N. Bezinger, M. A. Abdurakhmanov, and G. D. Galpern, Petrol. Chem. (U.S.S.R.),1,13 (1962). (5) I. Okuno, D. R. Latham, and W. E. Haines, ANAL.CHEM., 37, 54 (1965). ( 6 ) N. N. Bezinger, M. A. Abdurakhmanov, and G. D. Galpern, Petrol. Chem. (U.S.S.R.), 1, 485 (1962). 67) E. C. Copelin, ANAL. CHEM., 36,2274 (1964).
1830
e
ANALYTICAL CHEMISTRY
that are readily formed in some crude oils by air oxidation under mild conditions; sulfoxides have been shown to titrate as weak bases in acetic anhydride (8). This paper describes the separation of sulfoxides from petroleum fractions by cation exchange resin chromatography. Analytical data are presented to show that they are saturated cyclic sulfoxides. Analytical methods used include gas chromatography, potentiometric titration, and infrared, ultraviolet, mass, and nuclear magnetic resonance spectrometry. CONCENTRATION OF WEAK BASES BY ION EXCHANGE CHROMATOGRAPHY
Six high-sulfur and high-nitrogen crude oils containing from 0.9 to 3.8 weight per cent sulfur and from 0.1 to 0.65 weight per cent nitrogen were distilled at reduced pressure. Potentiometric titration showed that the 250" to 500" C (at 760 mm Hg) fraction contained the largest percentage of weak bases. This fraction was subsequently used in the chromatographic procedure to concentrate the weak bases. Duolite C-10 cation exchange resin, 40-100 mesh (Diamond Alkali eo., Redwood City, Calif.), was used to concentrate the bases. The resin was washed three times in boiling methanol, then Soxhlet extracted for 24 hours each with methanol, benzene, and n-pentane. The chromatographic column consisted of a 10-mm-0.d. glass column packed loosely (without tapping) with 5 grams of air-dried resin. A 50-gram sample of the 250" to 500' C fraction of the crude oil was passed through the chromatographic column with 2 psi of nitrogen pressure. Fifty milliliters each of npentane, benzene, methanol, and 10 isopropylamine in methanol were successively used as eluents. Each chromatographic fraction was titrated potentiometrically in acetic anhydride after removing the eluent solvent. No basic nitrogen or weak bases were found in the fraction eluted with pentane, indicating that all of the bases were removed from the petroleum sample and retained on the ion exchange column, The benzene-eluted and the methanol-eluted fractions contained only weak bases. Neither fraction contained basic nitrogen which required the stronger aminemethanol solvent for elution. The methanol-eluted fraction was selected for further study because it contained the largest concentration of weak bases. The methanol-eluted fraction from each crude oil was analyzed by the nitrogen type-analysis method (5) to determine the type Bz content. Because of limited quantities of material, it was necessary to modify the method by using 0.01N instead of 0.1Nperchloric acid as the titrant to facilitate titrating smaller samples. Fifteen milliliters of a 1 : 2 mixture of benzene and acetic anhydride was used as the titration solvent, Table I compares the concentration of the type Bz bases, calculated as weight per cent sulfur, in the 250" to 500" C fractions with the concentration in the methanoleluted ion exchange fractions. As shown in the table, the (8) D. C. Wimmer, Ibid., 30, 2060 (1958).
FREQUENCY, Crn-1
1600
r1 '
1200
1400 I
I
I
1100 l
l
1
1000
950
I
I
---
n11er1 r,d"rlien A l l i f rlduclior
Table I. Comparison of Type Bz Bases in 250" to 500" C Fractions and in Ion Exchange Chromatographic Fractions Type Bz,wt Z S MeOH250" to Crude oil 500" C fraction eluted fraction 6.13 I 0.11 6.19 I1 0.05
0.05 0.01
111
IV V
VI
3.61 1.67
0.00
0.05
0.00
0.00
Table 11. Comparison of Type BZ Content and Increase in Sulfide Content in Methanol-Eluted Ion Exchange Fractions Figure 1. Infrared spectra of a methanol-eluted ion exchange fraction, before and after reduction with lithium aluminum hydride
type Bz bases in the 250" to 500" C fractions of crude oils I through IV have been substantially concentrated in the ion exchange chromatographic fractions. The methanol-eluted fractions of crude oils V and VI contained little or no BZ bases. REDUCTION OF TYPE Bz BASES IN THE METHANOL-ELUTED FRACTIONS TO SULFIDES
Evidence of sulfoxides is presented by the increase in sulfide content when the fractions were reduced with lithium aluminum hydride. The spectrophotometric procedure of Drushel ( 9 ) which measures the iodine-sulfide complex was used for the determination of sulfides. Table I1 compares the type Bzsulfur with the change in sulfide sulfur. The table shows the fractions that contain larger concentrations of type Bz sulfur have greater net increases in sulfide content. The higher net sulfide change relative to the percentage of type Bz sulfur in each fraction possibly results in part from the average absorptivity used in calculating sulfide sulfur. The absorptivity of 400 liters gm-l cm-l, as suggested by Drushel (9) may be too low for the types of sulfides being determined in this molecular weight range and consequently yield results that are excessively high. The discrepancy may also result from the presence of compounds that do not titrate as weak bases but which are reduced to sulfides with lithium aluminum hydride. Although sulfones are one possibility, ion exchange tests with model compounds have indicated that they would not be in the methanol-eluted fractions. Aryl sulfoxides are not believed to be present in the methanol-eluted fractions because ion exchange tests have shown that diphenyl and dibenzyl sulfoxides are not strongly adsorbed but are readily eluted from the column with benzene. The large net increase in sulfide sulfur observed in the reduced fractions is therefore presumably due to nonaryl sulfides. Data reported by Drushel (9) show that the iodine complex absorptivities of aryl sulfides range from 1.5 for diphenylthiamethane to 72 liters gm-l cm-l for 1,3-diphenylm2-thiapropane as compared to approximately 400 for the saturated aliphatic and heterocyclic sulfides. Hence, it can be reasonably assumed that aryl compounds do not constitute a significant portion of the sulfides being determined. (9) H. V. Drushel and J. F. Miller, ANAL.CHEM., 27,495 (1955).
x
Type B, Crude oil as wt % S I 6.13 I1 6.19 I11
3.61
IV V
1.67 0.05
VI
0.00
Before reduction 1.50 0.74 0.97 1.08
1.13 0.53
Sulfide, wt S After Net reduction increase 10.11 11.61 10.71 9.97 7.16 8.13 5.47 4.39 1.85 2.98 1.75 1.22
Table 111. Infrared Absorption Frequencies and Molar Absorptivities of Sulfoxides in Carbon Tetrachloride Sulfoxides S=O frequencies, cm-1
Q
Methanol-eluted ion exchange fractioins Thiacyclopentane-1-oxide 2-Methylt hiacyclopentane-1-oxide
1040 1035(303), 1095(93) 1035(250), ¶060(153) 1035(302), 1095(94) 3-Methylthiacyclopentane-1-oxide 1-Cyclopentyl-1-thiapropane-1-oxide 1055(292) 1055(295)ref. (10) 1-Cyclohexyl-1-thiaethane-1-oxide 1055(320) ref. (10) 1-Phenyl-1-thiaethane-1-oxide 1055(326), 1095(281) Diphenyl-1-thiamethane-1-oxide 1070(360) Thiapropane-2-oxide 1070(247), 1040(223) Thiacyclohexane-1-oxide a Values in parentheses are molar absorptivities in liters mole-' cm- l.
INFRARED ANALYSIS OF THE METHANOL-ELUTED ION EXCHANGE FRACTIONS
Infrared analysis of the methanol-eluted fractions as films, Figure 1, showed an absorption band at 1040 cm-' in the samples containing large amounts of type Bs bases; this band is characteristic of the S=O stretching vibration of sulfoxides. As shown in the figure, the absorbance decreased substantially when the samples were reduced with lithium aluminum hydride, a decrease which would be expected if sulfoxides were being converted to sulfides. Barnard et al. (10) have reported that the intensity of the S=O band of sulfoxides in carbon tetrachloride solution is proportional to the concentration according to Beer's law. Table 111 shows the frequencies and molar absorptivities of various sulfoxides in carbon tetrachloride solution. The absorptivities of the S=O bands in the 1035-1055 cm-' region are shown to be approximately 300 liters mole-l cm-'. Infrared analyses of the ion exchange fractions in carbon tetrachloride showed that the intensity of the band at 1040 cm-1 was semiquantitatively related to the concentration of type Bz bases. Table IV compares the type Bz sulfur, as determined by the type-analysis method, with (10) D. Barnard, J. M. Fabian, and H. P. Koch, J. Chem. Soc., 1949, p. 2442. VOL. 39, NO. 14, DECEMBER 1967
@
1881
5 00 3,000 0,tf
FREQUENCY, ern-1 2,000 1,600 1,400 1,200 1,000 900
BOO
0.45
700 650
0.40
~
0.35
z
-1
Figure 2. Infrared spectrum of material from gas chromatographic separation of a methanol-eluted ion exchange fraction
sulfoxide sulfur, as determined by infrared. Infrared calculations for sulfoxide sulfur were made by measuring the absorbance of the sample in a 0.1-mm pathlength sodium chloride cell, and using an extinction coefficient of 300. A general correspondence of the values from the two determinations is evident. An exact quantitative correspondence would not be expected because of different interferences in the two methods. Exact classification of the sulfoxides in the ion exchange fractions by infrared was not possible because suitable reference compounds were not available. None of the reference compounds shown in Table 111 exhibits si sulfoxide band which is in complete agreement with the ion exchange fractions, Certain types of sulfoxides can be eliminated on the basis of infrared data. The five-membered saturated heterocyclic sulfoxides display S=O frequencies in the range of the ion exchange fractions. However, in contrast to the ion exchange samples, these compounds show doublets with less intense absorptions at slightly higher frequencies. The six-membered thiacyclohexane-1-oxide also shows a doublet with the stronger absorption at 1070 cm-l than at 1040 cm-'. The dialkyl, cycloalkyl-alkyl, and aryl-alkyl sulfoxides display single absorption bands, but at considerably higher frequencies than the ion exchange fractions. later suggest that two types of sulfoxides not included in the table, the dicycloalkyl sulfoxides and the saturated heterocyclic sulfoxides containing two condensed rings, are most representative of the sulfoxides in the ion exchange fractions. GAS C ~ R O ~ A T O 6 R A ~SEPARATION ~IC OF A METHANOLELUTED FRACTION
Gas chromatogram of the methanol-eluted ion exchange fractions containing large concentrations of type Bz bases showed a prominent peak which was absent in the chromatograms of the fractions that were reduced with lithium aluminum hydride. A sample of the material under the peak was trapped for further examination. Potentiometric titration of
3
s: 0.30
!i
? 0.25
a
Y 5 0.20 vi
w
VI
a
m 0.15
3 0.10
0.05
0
10
20 30 40 50 60 OXlDATlON TIME, DAYS
70
80
90
Figure 3. Air oxidation of crude oil PI1
the isolated sample showed that it Contained 9.1 weight per cent sulfur as weak bases. High resolution mass spectral analysis of the trapped material showed that the main component had an mje 228. Exact mass measurements indicated a compound with the empirical formula C13H24S0. Saturated heterocyclic sulfoxides with two condensed rings of dicycloalkyl sulfoxides would be included in this molecular weight. Infrared and nuclear magnetic resonance data support these types of structures. The infrared spectrum of a film of the gas chromatographic sample is presented in Figure 2. The characteristic S=O absorption is shown to occur at 1040 cm-l. Evidence of nonaromaticity is indicated by the absence of the aromatic CM stretching frequency in the 3100-3000 cm-1 region as well as the GH bending vibrations in the 900650 cm-l region. No significant absorption bands, indicative of aromatic or olefinic C=C stretching, occur at 16801480 cm-'. The strong methylene CH stretching bands at 2924 and 2855 cm-1 suggest aliphatic character. In support of the infrared data, nuclear magnetic resonance analysis showed that the sample contained aliphatic or naphthenic hydrogens and no aromatic hydrogens.
FORMATION OF WEAK BASES IN CRUDE OILS BY AIR OXIDATION Table IV. Comparison of Type BZ Content and Sulfoxide Content as ~ e t e r ~ by i ~ Infrared ~ d in ~ e ~ h a ~ o Ion ~ - E ~ ~ ~ e ~ The origin or nature of the type B2 bases was Qf interest Exchange Fractions
Crude oil
Type B2wt % S
Infrared (1040 cm-l), wb sulfoxide S
I XI JII IV
6.13 6.19 3.61 1.67
5.92 5.16 3.50
VI
0.05 0.00
v
~~
1832
o
ANALYTICAL CHEMISTRY
1.36
0.64 0.15
because it was not known if the weak bases occurred naturally in the crude oils or were formed later during storage; the crude oils that were studied had been in storage and had not been completely protected from exposure to air. A fresh sample of crude oil 111 was taken at the wellhead and subsequently titrated potentiometrically. No weak bases were present. Weak bases appeared and increased rapidly when a stream of air was bubbled through the oil. The change is presented graphically in Figure 3. The weak bases increased
to 0.07 weight per cent sulfur after only 3 days of air-oxidation at room temperature. After 58 days of exposure, the sulfoxide sulfur increased to 0.20x. Additional exposure of the oil to air at 85°C for 10 days resulted in a sharp increase of the bases to 0.42 weight per cent sulfur before leveling off. The results show that these weakly basic materials were not present in the virgin crude but were easily formed under mild oxidizing conditions. Weak bases do not form in all crude oils, as evidenced by a similar air-oxidation test with crude oil VI. Prolonged exposure of this oil to air resulted in no increase in weakly basic material, indicating the absence of compounds which form or promote the formation of sulfoxides.
RECEIVED for review March 27, 1967. Accepted May 18, 1967. Division of Chemistry Petroleum, 150th National Meeting, ACS, Miami, Fla., April, 1967. This investigation was performed as a part of the work of American Petroleum Institute Research Project 60 on the Characterization of the Heavy Ends of Petroleum, carried out by the Bureau of Mines at Laramie, Wyo. Work was done under cooperative agreements between the Bureau of Mines, U S . Department of the Interior; the American Petroleum Institute; and the University of Wyoming. Reference to specific commercial materials or models of equipment in this report is made to facilitate understanding and does not imply endorsement by the Bureau of Mines.
p-Type Analysis by tisn Mass pectromet E. J. Gallegos, J. W. Green, L. P. Lindeman, R. L. LeTourneau, and R . M. Teeter Chewon Research Co., Richmond, Calif. We describe the first reported multicomponent grouptype analysis utilizing high resolution mass spectrometry for analyzing high boiling petroleum stocks without requiring fractionation via silica gel chromatography or its equivalent. The application of mass spectrometer group-type analysis to high boiling fractions has been restricted because it is necessary to employ costly and lengthy separations to reduce the number of spectral interferences caused by the great variety of components. It is well known that high resolution mass spectrometry eliminates many of these interferences. We have found that the data from the MS-9 high resolution spectrometer are sufficiently reproducible to take advantage of the small mass differences and thus eliminate the physical separation. This group-type procedure determines 19 components; seven saturated hydrocarbons with zero to six rings, nine aromatic hydrocarbons with one to four rings, and three aromatic sulfur compound types.
BECAUSE OF THE ABILITY of the early commercial mass spectrometers to give information on the carbon-hydrogen ratio and molecular sizes of hydrocarbon mixtures, mass spectrometry became an important analytical tool for assaying petroleum fractions. R . A. Brown (1) in 1951 was the first to use this spectral information for analyzing for compound types rather than for individual compounds. His procedure applied to gasoline received immediate and nearly universal acceptance in the petroleum industry. Increasing demand for processing of high boiling petroleum stocks caused interest in extending group-type analysis to these materials. O’Neal and Wier ( 2 ) in 1951 were the first to employ the necessary instrumentation and techniques for recording mass spectra of the heavier materials. Most heavy petroleum fractions contain such a large variety of components that a single spectrum cannot be interpreted in terms of compound types as can be done with gasolines. By 1956, however, several mass Spectrometer group-type methods (1) R. A. Brown, ANAL.CHEM., 23, 430 (1951). ( 2 ) M. J. O’Neal, Jr., and T. P. Wier, Jr., Ibid.,p. 830.
were published for determining separately the compositions of saturate and aromatic fractions. The disadvantage of this scheme is the time-consuming silica gel separation step which doubles the number of samples that must be analyzed, increases the opportunities for error, and limits the overall accuracy by the precision of the split. In this work, we made use of high resolution mass spectrometry, which has only recently become commercially available, to eliminate most of the interferences between saturated and aromatic hydrocarbons and, further, to eliminate the interferences of many hydrocarbons with a number of sulfur compounds. This is the first published report of a quantitative group-type procedure which takes advantage of high resolution mass spectrometry for the analysis of high boiling petroleum fractions without a preliminary silica gel separation. This analysis is useful as a rapid and inexpensive procedure for supplying the petroleum chemist with a quantitative estimate of the major components in the saturate and aromatic portions of crude oils and refinery stocks. No attempt has been made to include all molecular types present in petroleum products. This deficiency becomes more pronounced with the higher boiling fractions. A number of additional components could be added to the present scheme with additional effort. However, the law of diminishing returns eventually forces one to abandon many of the minor components with a sacrifice of some completeness and accuracy. Alternatively, one is forced to use a more elaborate scheme involving separation techniques with consequent increase in cost of analysis. As presently constituted, the method is considered useful for distillate fractions in the 500 ’950” F range. EXPERIMENTAL
Instrumentation. The high resolution instrument used is an MS-9 made by Associated Electrical Industries, Ltd. A 2-pl sample is introduced into an all-glass heated inlet (3). The instrument is adjusted to give 5000 resolution at mass 250. This ensures optimum resolution for the portion of ~
~~
(3) R. M. Teeter and W. R. Doty, Rev. Sei. Innstr., 37,792 (1966). VQL. 39, NO. 14, DECEMBER 1967
* 1833
