Measurements were made by peak area using a Disc Integrator on a Minneapolis Honeywell Class 15 recorder. The sample solutions were used as standards and the fractions analyzed in terms of percentage recovery of each solute in each fraction. Because of the large number of cuts, replicate runs were not always made and analyses are considered to be accurate to only 3 to 5OJo0. Solute recoveries, from the sum of the contents of each cut, ranged from about 80 to 110%. Lower recoveries were generally associated with more strongly adsorbed solutes which eluted over a larger number of cuts. More accurate determinations would be required to determine whether actual losses are involved or whether the lower indicated recoveries are accumulated errors for the more dilute solutions. Distribution Constants. Equal volumes of the stationary phase and a
hexane solution of the sulfide were shaken in vials sealed by serum caps. Samples of the hexane layer were withdrawn by a syringe at various intervals from 1 to 24 hours and analyzed by GLC. The distribution constant, K , was calculated assuming that the decrease in sulfide concentration in the hexane layer represented the sulfide concentration in the stationary phase layer. In all cases, the value of K did not change appreciably after 1hour of shaking but the tabulated values are based on analyses made after 24 hours. Most measurements were made with an initial sulfide concentration of 6.66 pl./ml. in the hexane. ACKNOWLEDGMENT
The author expresses his gratitude to C. H. Calvert for assistance in the laboratory.
LITERATURE CITED
(1) Birch, S. A,, McAllan, D. T., J . Inst. Petroleum Tech. 37, 443 (1951). (2) Challenger, F., “Aspects of the Organic Chemistry of Sulfur,” pp. 1-20, 73-90, Butterworth, London, 1959. (3) Challenger, F., Haslam, J., Bramhall, R . L., Walkden, J., J . Inst. Petroleum Tech. 12, 106 (1926). (4) Emmott, R., Zbid., 39, 695 (1953). (5) Hopkins, R . L., Coleman, H. J., Thompson, C. J., Rall, H. T., U . 8. Bur. Mines, Rept. Invest. 6458 (1964). (6) Keulemans, A. I. M., “Gas Chromatography,” pp. 96-129, Reinhold, New York, 1957. (7) McAllan, D. T., Cullum, T. V., Dean, R. A., Fidler, F. A., J . Am. Chem. SOC.73, 3627 (1951). (8) Purnell, H., “Gas Chromatography,” p. 93, Wiley, New York, 1962. RECEIVEDfor review June 13, 1966. Accepted July 25, 1966.
Identification of Thiaindans in Crude Oil by Gas-Liquid Chromatog rap hy, DesuIfurization, and Spectral Techniques C. J. THOMPSON,
H. J.
COLEMAN, R. 1. HOPKINS, and H. T. RALL
Barflesville Petroleum Research Center, Bureau of Mines,
b Knowledge of the sulfur components in petroleum is of both theoretical interest and practical value to the petroleum industry. The apparent absence of thiaindons in petroleum has been of interest to sulfur and petroleum chemists for many years. Recently the authors have identified 1 -thiaindan and 17 alkylthiaindans, in Wasson, Texas, crude oil by a combination of gas-liquid chromatography, desulfurization, and spectral techniques. This represents the first known identification of this class of sulfur compound in petroleum. These identifications and the techniques employed are described.
C
of the sulfur compounds in petroleum has special significance for petroleum scientists. Information, both qualitative and quantitative, concerning these compounds is valuable, not only as an addition to fundamental knowledge but also as an aid in processing applications. The presence in petroleum of thiols ( I , S), chain sulfides (9), cyclic sulfides (1, 6, IO), thiophenes ( I , II), and benzothiophenes ( 2 ) is well documented. While the presence or absence of thiaindans in petroleum has been of int,erest for some time, their identification has HARACTERIZATION
1562
ANALYTICAL CHEMISTRY
U. S.
Department of the Interior, Bartlesville, Okla.
not been reported gespite their expected presence because of a close similarity in structure to the relatively abundant benzothiophenes. This paper describes the procedures used for the isolation and identification of thiaindans in a Wasson, Texas, crude oil distillate boiling at 200’ to 250’ C. It discusses the p r e p arat,ion of the thiaindan concentrate, the positive identification of l-thiaindan,
a ,
%methyl-1-thiaindan,
DC ,
2,2-dimethyl-1-
thiaindan,
and the tenta-
and
m z
tive identification of alkylthiaindans.
15 additional
EXPERIMENTAL
Preparation of Thiaindan Concentrate. The 200” to 250” C. distillate in which the thiaindans were identified was obtained by a series of isothermal, molecular, and fractional distillations as outlined in Figure 1. The all-glass, steam-heated, isothermal stripping unit previously described (8) was used to strip the lower boiling distillate from the crude oil. The higher boiling distillate was removed using a Consolidated Vacuum Corp. centrifugal molecular still. The residence time in the heated zone of both the isothermal
still and the centrifugal molecular still was but a few seconds. This short contact time and the low temperatures employed tended to protkct the sulfur compounds from thermal breakdown or rearrangement. The distillate prepared from the crude oil then was fractionated into selected boiling ranges a t 15 mm. Hg through a concentric tube column having an annulus of 0.04 inch and a lengtKof 48 inches. The 200’ to 250” C. boiling ranee distillate produced in the above tre& ment was percolated through silica gel to provide two fractions-one a hydrocarb )n discard essentially free of sulfur and the other a sulfur compound concentrate. This concentrate was treated successively with sodium hydroxide and sodium aminoethoxide (6) to extract thiols and phenols. The thiol-free concentrate was filtered through a 15foot by 2-inch diameter bed of H-41 alumina using gradient elution chromatography. This treatment produced 27 fractions as indicated at the bottom of Figure 1. The sulfur-containing fractions-Le., 19 through 26-were treated, as shown in Figure 2, with anhydrous H I in butane solution at -78” C. to extract the sulfides (4)after those substances which crystallize upon chilling were first removed. The residues from the HI treatment and the precipitates from the butane treatment were combined and treated twice with trinitrobenzene to remove benzothiophenes and
FRACTIONS 19 TO 26, INC. 57,6 gms
(100 .O%)
I
C H I L L I? Q H I O
-Dirt iIlot ion
I
DISTILLATE
(19.5%)
(80.0%)
HI EXTR'ACTION
Irothhnol D i'st iIlot ion
RESIDUE
EXTRACT
HI EXTRfCTION
DISTILLATE
I
(72.O%)
RES I D U.E
I
Vacuum Fractionation
* FIE]
I
HI EXTR'ACTION
,I I
I
D IS TI LLATE (17.0%)
19.4 grns
17.5 grns
T R E A T WITH T R I N I T R . 0 8 E N 2 E NE
I
Silica Gel Adrorpt ion
I
FILTRATE. 13.8 grns
17.2 grns
HI E X T ~ A C T I O N HYDROCARBON DISCARD
I 1
CONCENTRATE su~uR '
I
Thiol Extraction
I
A THIOLS & PHENOLS
I I
9 : 3 grns
4.2 gms
I.
THIOL FREE CONCENTRATE
I
I
AluAino Adrorot ion
TREAT' WITH TRINITROBENZENE
, A{,"!J?T
II
ANDODUC;T
II
i
27 FRACTIONS
-
Figure 2. fractions
Treatment of Wasson
I
THlAlNDAN CONCENTRATE
200'
I
to 250' C. adsorption
Figure 1. Outline of treatment of Wasson, Texas, crude oir to produce adsorption fractions
naphthalenes as adducts. This treatment produced a final residue containing approximately 60% thiaindans of a molecular weight 136 to 200, 23% alkylbenzenes, 9% naphthalenes, and small amounts of other sulfur compounds, as determined by mass spectrometry. Preliminary desulfurization studies supported the presence of thiaindans and prompted a detailed study which resulted in the identifications reported herein. Positive Identification of 1-Thiaindan, 2-Methyl-1-thiaindan, and 2,2Dimethyl-1-thiaindan. The positive
identification of thiaindans in petroleum, either individually or as a class, was a prime objective of this investigation. Toward t h a t end, a preliminary evaluation of the thiaindan concentrate was achieved using gasliquid chromatography (GLC). The chromatogram shown in Figure 3 was produced in the analysis of a 7-pl. charge of the concentrate in a Perkin-Elmer vapor fractometer, Model 154D, using B
l/rinch by @-foot column packed with 30- to 42-mesh GC-22 Super Support impregnated with silicone rubber SE-30. A comparison of the retention times of the peaks in the chromatogram with the indicated retention times of the two reference thiaindans shown suggests the
possible, if not probable, presence of 1thiaindan in the material producing the shaded peak a t 40 minutes and of 2methyl-1-thiaindan in the material producing the shaded peak a t 42 minutes. 1 n . m effort to identify these thiaindans, the material producing these shaded peaks was trapped for further investigation. GLC analysis of the
z
-
0 I-
u W
J U W 0
W z
n
a
0
W V
a I 0
92
84
- d l , l 76
68
60
52
44
36
28
TIME, minutes
Figure 3. crude oil
Gas-liquid chromatogram of a thiaindan concentrate from Warson
VOL 38, NO. 1 1 , OCTOBER 1966
1563
4 00
LA 0
0
K
60
56
52
48
44
36
40
TIME, minuias
32
-
Figure 4. Gas-liquid chromatogram of material trapped during minute interval in Figure 3
products of Raney nickel desulfurization (7') of the trapped material revealed the presence of at least three components in the material producing the large shaded peak in Figure 3. By the alternate use of nonpolar and polar substrates, it was possible to separate individually four components represented by the two shaded peaks of Figure 3. This was accomplished by repeated trapping of the effluent material (39 t o 44.5 minutes) from the nonpolar column as shown in Figure 3. The material so obtained was then charged to a polar column (1/4-inch by 25-foot1 packed with 45- to 50-mesh Chromosorb G coated with about 6% Reoplex 400) to effect the separation shown in the chromatogram of Figure 4. The materials producing the four significant peaks then were collected and analyzed by infrared and mass spectrometry. The peak (material producing the peak) with a retention time of 53 minutes was identified positively as 1thiaindan by infrared and mass spectrometry and by GLC retention time. The peaks a t 46 and 35 minutes were identified in like manner as 2-methyl-1thiaindan and 2,2-dimethyl-l-thiaindanJ respectively. The excellent agreement of the infrared spectra of the materials isolated from the crude oil with the spectra of the reference thiaindans, as shown in Figure 5, is unequivocal proof of the presence of these thiaindans in petroleum. Desulfurization data support the identifications just discussed-e.g., ethyl benzene was produced from 1-thiaindan, n-propylbenzene from 2-methyl-1thiaindan, and isobutylbenzene from the 2,2-dimethyl-l-thiaindan. The composition of the peak at 27 minutes in Figure 4 was not established in this study. The reference samples of thiaindans used in this investigation were purified by GLC and, before use as reference compounds, their structure was confirmed by desulfurization, mass, and infrared analyses. The investigation has established the presence of thiaindans in general arid specifically 1-thiaindan, 2-methyl-1-thiaindan, and 2,2-dimethyl-l-thiaindanin 1564
ANALYTICAL CHEMISTRY
39- to 44-
petroleum. Related desulfurization studies have resulted in additional identifications discussed below. Tentative Identification of Thiaindans. The absence of reference thiaindans or spectra thereof, other than t h a t of the 1-thiaindan, 2-methyl-1thiaindan, and 2,2-dimethyl-l-thiain-
n ,Material
-
dan, has precluded additional positive identifications. However, the combination of gas-liquid chromatography and Raney nickel desulfurization was effective in providing the data required to permit the tentative identification of additional thiaindans. These identifications were accomplished by the Raney nickel desulfurization of a 10-pl. sample of the thiaindan concentrate. The individual alkylbenzenes produced in this desulfurization were isolated by GLC trapping techniques, and those present in sufficient quantity were identified by infrared and mass analyses. The minor components were identified by GLC retention time. Figure 6 is a chromatogram of these desulfurization products obtained with a '/*-inch by 40-fOOt column packed with 30- to 42mesh GC-22 Super Support coated with about 20% chlorinated polyphenylether (five rings). The salient data of this desulfurize tion study are summarized in Table I. Briefly, the material producing each major peak was trapped and identifiede.g., the peak at 22.7 minutes, designated in Figure 6 as peak No. 2, was
isolated from crude oil (in
WAVELENETH, microns
cs2)
.'-
Figure 5. Comparison of infrared spectra of materials isolated from Wasson crude oil and spectra of reference thiaindans
4
Figure 6. crude oil
Chromatogram of the products of Raney nickel desulfurization of thiaindan concentrate from Wasson
trapped and identified by both infrared and mass analyses as isopropylbenzene. I n this instance, with the identity of the aromatic now established the thiaindan precursor is limited to only 3methyl-1-thiaindan, as listed in columns 4 and 8 of Table I. Similar treatment of each peak in Figure 6 provided a total of 15 tentative
identifications in addition to the three positive identifications described earlier. Several of the identified aromatics could have been produced from two different thiaindans. These are listed in the table on an and/or basis and reported as a single tentative identification. Some of the identified aromatics in peaks 12, 13, and 15 of Figure 6 could
originate from not only the 1-thiaindans shown in Table I but also from 2-thiaindans. Because of instability toward oxygen and because of greater chemical reactivity the 2-thiaindans are not considered likely components in this sample and therefore are not listed in Table I. Small amounts of the aromatics observed could have been derived from
Table 1.
Date Pertaining to Raney Nickel Desulfurization Products From Thiaindan Concentrate as Shown in Chromatogram of Figure 6 Identity of material Identity of material
producting peak As determined by GLC by infrared
As indicated
Peak No.
Precursor of identified aromatic Peak No.
producting peak As determined by GLC by infrared
As indicated
Precursor of identified aromatic
9’
10
c ) y c
J/
I I
C
12
I3
14
I 5
8
Sam le of insufficient size to permit analysis Unifentified Slash indicated “and/or”.
VOL 38, NO. 11, OCTOBER 1966
1565
the desulfurization of trace quantities of benzothiophenes and aryl alkyl sulfides present in the thiaindan concentrate. However, any such contribution would be too small to invalidate the identifications reported. 1Thiatetralin, retention time of 61 minutes in Figure 3, was looked for but not found in this concentrate. The tentative identifications based upon this desulfurization study are reasonably certain but lack final confirmation by spectral and GLC comparison with authentic reference samples of these compounds. These identifications are therefore listed as tentative. CONCLUSIONS
With the techniques described in this paper, the following 18 thiaindans were identified (three positively, 15 tentatively) in a Wasson, Texas, 200’ to 250’ C. distillate: 1-thiaindan, 2methyl-1-thiaindan, 3-methyl-1-thiaindan, bmethyl-1-thiaindan and/or 7 - methyl - 1 - thiaindan, 3,3 - dimethyl 1-thiaindan , 2,3-dimet hy 1- 1-thiaindan and/or 3-ethyl-l-thiaindan, 2,2dimethyl-1-thiaindan, 3,5-dimethyl-1thiaindan and/or 3,7-dimethyl-l-thiaindan, 3,6-dimethyl-l-thiaindan, 2,5-dimethyl-1-thiaindan and/or 2,7-dimethyl-1-thiaindan, 2-ethyl-1-thiaindan, 2,6-dimethyl- 1-thiaindan, 2methyl-3-ethyl-1-thiaindan, 2,4-di-
man, M. L., Richardson, D. M., J . methyl-1-thiaindan, 2,2,5-trimethyl-IChem. Ena. Data 6.464-8 (1961). thiaindan and/or 2,2,7-trimethyl-lColema;, H. J.; Thompson, C. J., thiaindan, 2,3,4trimethyl-l-thiaindan (3) Hopkins, R. L., Rall, H. T., Ibid., 10, and/or I-ethy1-4-methyl-l-thiaindan, 80-4 (1965). (4)Hopkins,’ R. L., Coleman, H. J., 2,2,4-trimethyl-l-thiaindan, and 2Thompson, C. J., Rall, H. T., U . S. methyl-2-ethyl-1-thiaindan. This repBur. Mines Rept. Inv. 6458, 20 pp., resents the first known identification 1964. ~ . . ~ of this class of sulfur compounds in (5)Hopkins, R. L.,Smith, H. M., ANAL. CHEM. 26,206-7 (1954). petroleum. (6)Mabery, C. F., Quayle, W. O., J. SOC. \ - - - - I .
ACKNOWLEDGMENT
The authors thank R. F. Kendall, J. E. Dooley, and B. H. Eccleston of the Bartlesville Petroleum Research Center for the spectral data presented and so necessary in the identifications achieved. N. G. Foster, now of Texas Woman’s University, Denton, Texas, supplied the preliminary mass spectral type analysis on the concentrates. The sample of 2-methyl-1-thiaindan used in this investigation was generously applied by Cal Y. Meyers of Southern Illinois University, Carbondale, 111. The sample of 2,2-dimethyl-l-thiaindanwas donated by Harold Kwart of the University of Delaware. LITERATURE CITED
(1) Birch, S. F., J . Inst. Petrol. 39, 185205 (1953). (2) Coleman, H. J., Thompson, C. J., Hopkins, R. L., Foster, N. G., Whis-
Chem. Ind. 19, 505-6 (1900); Am. Chem. J . 35,404-32 (1906). (7) Mozingo, Ralph, Wolf, Donald E., Harris, Stanton A.. Folkers. Karl, J . A m ; Chem. SOC.65,’1013-16 (1943). ‘ (8)Rall, H. T., Hopkins, R. L., Thomp son, C. J., Coleman, H. J., Proc. Am. Petrol. Inst., Sec. V I I I 42, 46-50 (1962). (9) Thompson, C. J., Coleman, H. J., Hopkins, R. L., Rall, H. T., J. Chem. Eng. Data 9,473-9 (1965). (10)Ibid.. 10. 279-82 11965). (11)Thompson, C. J.; Coleman, H. J., Ward, C. C., Rall, H. T., Ibid., 4, 3478 (1959).
Received for review April 18, 1966. Accepted June 14, 1966. Reference to specific brands is made for identification only and does not imply endorsement by the Bureau of Mines. Investigation performed m part of American Petroleum Institute Research Project 48 on “Production, Isolation, and Purification of Sulfur Compounds and Measurement of Their Properties” which the Bureau of Mines conducts at Bartlesville, Okla., and Laramie, Wyo. Presented at the 152nd Meeting, ACS, New York, September 1966.
Adsorption of Traces of Silver on Container Surfaces FOYMAE KELSO WEST and PHILIP W. WEST Coafes Chemical Laboratories FRANK A. IDDINGS Nuclear Science Center Louisiana State University, Baton Rouge, La.
b This study was designed to measure losses of silver due to its adsorption on container surfaces, and to evaluate the use of ligands for prevention of silver losses during prolonged storage of potable water samples. It was found that the greatest loss of silver to container surfaces occurred between 10 and 30 days. The adsorption characteristics of borosilicate ’glass-, flint-, polyethylene-, and silicone-coated containers have been established. It was found that sodium thiosulfate prevented losses both at 0.05 mg. per liter and 1.0 mg. per liter silver concentrations.
T
1962 U.S. Public Health Service Drinking Water Standards of a maximum permissible concentration of 0.05 mg. per HE
1566
SPECIFICATION
0
IN
THE
ANALYTICAL CHEMISTRY
70803
liter of silver emphasized the need for sensitive, specific, reliable, and economical methods for the routine determination of silver in potable waters, Inasmuch as many researchers have noted the high and often erratic adsorption of silver on laboratory glassware, it appeared mandatory that this phenomenon be investigated in detail prior to the development of analytical methods for determining trace concentrations in water. The work reported here was designed to study the adsorption of silver on various common storage container materials and to evaluate the use of ligands for complexing the silver as a means of preventing losses. In routine laboratory operation, samples must frequently be stored from several hours to several weeks, depending upon the
distance of sampling point and upon the fluctuating work load of the laboratory. Chambers and Proctor (2) reported that the adsorption of silver varied with the strength and composition of the solution and the kind of material used in the container, and was also affected by the previous treatment of the container. They reported adsorption of 15-45y0 on borosilicate glass flasks from a neutral solution of 0.1 mg. per liter silver. Dagnall and West (3) noted losses of silver at the 0.01 mg. per liter level to be significant despite use of silicone-coated glassware. EXPERIMENTAL
Apparatus and Materials. An R C L Model 10D decade scaler was used in the preset time mode for all counting. Electronic integrity of the instrument