PROCEDURE
Bromine, National Formulary, and reagent grade acetic acid and pyridine were used. Sample phenols (“purified” and reagent) were used as supplied. An amount of sample to require about 20 ml. of 0.15M bromine solution was dissolved in 37 ml. of glacial acetic acid plus 3 ml. of pyridine. This solution was titrated to a permanent voltage drop. +4 blank titration was made on the glacial acetic acid and pyridine minus sample. Blanks were 0.1 to 0.2 ml. Presumably, further purification of the pyridine would reduce this small blank; however, this was not attempted. Standard bromine solutions were prepared both by standardizing approsimate solutions with thiosulfate and by :tccurately weighing chilled bromine and diluting to volume. The solutions, once prepared, gave consistent results over
a period of a month indicating that, with reasonable handling, there is not significant loss of bromine during this time. The titration volume was not critical, and smaller volumes could be accommodated with the apparatus described. The usual time for titration was about 20 minutes.
respond to those found by lngberman (3). Troublesome precipitation of bromination products, noted in some cases by Ingberman, was prevented, since no aqueous solutions were used.
RESULTS AND DISCUSSION
( 2 ) Englert, S. , M. E., McElvain, S. M., IIbid., b 51,863 (1929). (3) Ingberman, A. K., ANAL. CHEM.30,
Results obtained for a series of phenols are shown in Table I. Three of the phenols titrated are of special interest. Theoretically, 2,4dinitronaphthol does not brominate in any position. Only a n amount of titrant corresponding to the blank was necessary to bring about the end point voltage change. p tert-Butylphenol was titrated satisfactorily, but 2,6-ditert-butylphenol overbrominated. These results cor-
LITERATURE CITED
(1) Djerassi, C., Scholz, C. R., Chem. SOC.70,417 (1948).
J. Am.
\ - ,
1003 (1958). (4) Kolthoff, I. M., Belcher, R., “Volumetric Analysis,” Vol. 111, pp. 534 ff., Interscience, New Ynrh York, 1~157 1957. (5) Komeschaar. M. F.. 2. anal. Chem. . ,15,259 (1876).’ (6) Reilley, C. N., Cooke, W. D., Furman, N. H., ANAL.CHEM.23,1223 ( 1951). RECEIVEDfor review August 7, 1961. Accepted December 4, 1961. Division of Analytical Chemistry, 140th Meeting, ACS, Chicago, Ill., September 1961.
Structural Group Analysis of High Sulfur Content Mineral Oils CLARENCE KARR, JR.,l R. T. WENDLANDI2 and W. E. HANSON Mellon Institute, Pittsburgh, Pa.
b A method for the calculation of the average ring composition for a high sulfur content mineral oil is based on the observation that replacement of one or two methylene groups of a hydrocarbon molecule by a sulfur atom will raise the refractive index and density, on the average, b y amounts which are proportional to the weight per cent of sulfur so introduced. The proportionality factor appears not to vary greatly with hydrocarbon type and, for mixtures of the sort encountered in petroleum, it can b e taken as essentially constant. Adjusting the observed refractive index and density values of an oil in accordance with its sulfur content thus gives the values which the mixture would have if all sulfur compounds present were converted to their hydrocarbon analogs. The average ring analysis can then b e calculated by any of the well known structural group methods. The technique is illustrated by application to mixtures of hydrocarbons and sulfur compounds and to selected sulfur concentrates from high sulfur content crude oils. Nomographic charts are provided for routine work.
0
the past 20 years, numerous methods have appeared in the literature for determining the structural group composition of mineral oils. Following the original “direct method,” which was based on the molecular weight and ultimate analysis of the oil fraction before and after exhaustive hydrogenation, the procedure was gradually simplified to the point where physical property correlations, derived from measurements on petroleum oils of diverse types, substituted for the difficult hydrogenation. The method most commonly employed is that proposed by van Xes and van Westen (3). The aim of every structural group analysis scheme is to estimate the proportion of the carbon present as aromatic ring structures, as naphthenic ring structures, and as paraffinic side chains. With certain assumptions as to the type of rings present and the extent t o which they are condensed, it is possible to express the carbon distribution data as the mean number of aromatic and cycloparaffinic rings per molecule. Hydrocarbon type analysis has found its greatest use in the heavier oils-Le., for oils of average molecular VER
weight of 200 and above, and in particular for wax-free (alkane-free) oils. This latter simplification permits the allocation of the “paraffinic carbon” strictly to alkyl side chains on the ring structures. I n developing analytical methods, one must be cognizant of the fact that sulfur and, to a considerably lesser extent, nitrogen and oxygen may be present. The latter two elements have been dismissed because of their relatively slight contribution on a mole percentage basis in most mineral oil fractions, Sulfur, on the other hand, may be an important component of oils but, since the nature of the sulfur compounds present is usually unknown, the correction has often been ignored. van Nes and van Westen have advocated an empirical correction, but the sulfur contrnt of thc oils r.mployed in deriving the factor to be applied did not exceed 2.3y0. Caniisa and Fratta (1) have shovn that oils of higher sul-
Present address, U. S. Bureau Mines, Morgantomn, W. Va. 2 Present address, Winona State College, Winona, Minn. VOL. 34, NO. 2, FEBRUARY 1962
249
fur content (up to 7y0)can be handled by the extrapolation of data obtained on mixtures of the high sulfur content oil with some similar sulfur-free petro-
90
100
I
I
Molecular Weight (M) 150 175 200 250 I I I I
125
110 I
leum fraction. In no case, however, were these authors dealing with oils of high ring content such as are considered here.
I
400 600 1000 111l11
1
1
0.9-
1
-
Correction Factors for Density and Refractive Index of High Sulfur Content Oils. Figure 1 shows a plot,
1
1
II Figure 1
.
1
2
3 4
5
1
1
IO
1
1
1
9
1
8
1
7
8 9
Sulfur Compound Classes Alkane thiols Alkane sulfides Cycloalkane thiols (1 ring) Cycloalkane sulfides (1 ring) Cyclic sulfides (1 ring) Aromatic thiols (1 ring) Aromatic sulfides (1 ring) Cycloalkane thiols (2 fused rings) Cycloalkane sulfides ( 2 nonfused rings; 2 fused rings) Aromatic thiols ( 2 fused rings; 1 aromatic)
Cyclic sulfides ( 2 fused rings; 1 aromatic) Aromatic thiols (2 fused rings) Aromatic sulfides ( 2 fused rings; 2 nonfused rings) Alkane disulfides Thiaalkylalkanesulfides Alkane dithiols 1,3-Dithiacycloalkanes (1 ring) 1,3-Dithiacycloalkanes ( 2 nonfused rings; 1 aromatic)
10 Thiophenes (1 ring) 11 Thiophene thiols (1 ring) Thienylalkane sulfides (1 ring) 12 Benzothiophenes PhenvlthioDhenes
250
(
1
6 IOOO/M
'
1
5
'
1
4
I
1
3
'
1
2
'
1
I
'
Interrelations of Sulfur and Hydrocarbon Compound Classes
Aromatic sulfides ( 2 fused rings; 1 aromatic)
6
I
7
Density-molecular weight relationship: hydrocarbons-sulfur compounds
Table 1.
Plot
In some work in this laboratory, a number of wax-free, sulfur compound concentrates were prepared by distillation and chromatography from a series of crude oils of high sulfur content. Because the sulfur contents of these concentrates were in the 3 t o 9% range, i t was evident t h a t none of the methods for structural group analysis would apply directly. The need for information on the character of these oils led to the development of the refractive index and density correction terms described below.
ANALYTICAL CHEMISTRY
ilnalogous Hydrocarbon Classes Alkanes
4.5
1.7
Cycloalkanes
5.8
2.1
Benzenes
6.4
2.7
Bicycloparaffins ( 2 fused rings) Dieycloalkanes (2 nonfused rings) Benzocycloalkanes ( 2 fused rings; 1 aromatic) Cycloalkylbenzenes ( 2 fused rings; 1 aromatic)
5.5
1.9
7.3
2.8
6.9
2.9
5.4
1.9
Cycloalkanes 6.1 Benzocycloalkanes 7.1 ( 2 fused rings; 1 aromatic) Cycloalkylbenzenes ( 2 nonfused rings; 1 aromatic) Alkylbenzenes 4.1 Alkylbenzenes 10.9
2.3 2.8
Naphthalenes Biphenyls and diphenylalkanes Alkanes
Piaphthalenes BiDhenvls and Diphknylalkanes Average
kd
x 108
6.4
kn
x 108
0.6 2.2 2.6
6 . 4 & 1 2 . 2 =k 0 . 5
typical of a series, relating t h e density for various classes of sulfur compounds to t h e reciprocal of t h e molecular weight. The compounds were grouped according t o structure and, when values for more than one isomer were available, an arithmetic average was taken as the representative value for the particular molecular weight. On each plot, a similar curve was constructed for the analogous hydrocarbon series and, again, an isomer average was taken to secure the representative property value for a given molecular weight, The term "analogous hydrocarbon" is defined as the hydrocarbon one would get were one t o replace each sulfur atom in aliphatic combination by a methylenic (-CH,-) or a n ethylenic (-CH2-CH2-) group and each sulfur atom in aromatic combination with an acetylenic (-CH=CH-) group. This rather arbitrary convention permits one to relate the various classes of petroleum sulfur compounds with the major classes of petroleum hydrocarbons. Thus, thiacyclopentanes or -hexanes become Ca- or Cs-naphthenes; thiophenes become benzene derivatives; benzothiophenes become naphthalenics; etc. The sulfur compound classes incorporated in each plot, and the corresponding hydrocarbons are listed in Table I. Figure 2 illustrates a similar plot, typical of a series, relating refractive index with reciprocal molecular weight. No attempt is made to list the sources of the data used. I n plotting the values, it was evident that many were in error and, for this reason, it was necessary to exercise judgment in the selection. In constructing the curves, greater weight was given to data taken on samples of evident high purity. In any case, the curves must pass through the limiting values for the paraffins (dzoOc. = 0.8510; C. = 1.4750) and, in few instances, what apparently would have been the best line drawn through the literature values had to be shifted somen.hat to have it meet this condition. Three important generalizations may be derived from the plots. A satisfactory linear relationship exists between
the properties of density and refractive index for sulfur compounds and the reciprocal of the molecular weight. While this relationship is knon-n to hold for homologous hydrocarbon series and should hold similarly for individual sulfur compound series, a reasonably good approximation to linearity is maintained for the isomer averages. Sulfur compounds of the same basic ring and chain structures fall on the same line. For any given molecular weight, the decrease m density in going from sulfur compounds to hydrocarbons of comparable structure is approximately the same. When two sulfur atoms are present, the effect is approximately doubled. h similar relationship is obtained for the refractive index plots.
-r
t
-Cd
=
An =
-,1 and M
(1)
-cn-M1
where Cd and C, are the proportionality constants for density and refractive index, respectively, a t 20" C., for sulfur compound mixtures containing a single atom of sulfur per molecule. While i t is doubtless true that the vast majority of all sulfur-containing niolecules in petroleum contain but a single atom of sulfur, the average probably rises appreciably as one goes into the asphaltic range. Operating to lower the average in a petroleum fraction is the content of hydrocarbons, especially those of aromatic type which tend to be selectively concentrated by the usual methods for preparing sulfur compound concentrates. To account for this variation in the number of sulfur atoms per molecule, Equations 1 and 2 must be multiplied by a factor, Sfif equal to the average number of sulfur atoms per molecule, where S is the sulfur content in weight per cent. Equations 1 and 2 then become
3m,
A n = -Cn-
s 32 X lo2
=
-k,,S
(4)
Thus, if each sulfur atom in a sulfurrich petroleum fraction were replaced by a methylenic, ethylenic, or acetylenic group, depending on whether the sulfur is in aliphatic or aromatic combination, the resulting decrease in density and refractive index is given simply by the product of the sulfur content in per cent and the appropriate constant. With the adjusted values for the physical properties and the molecular weight (essentially unchanged from
103 110 I
I25 I
Idoleculcr Weigkt (M) 150 175 200 250 I i
400 600 IO00 i
Q)
17111111
1.50
-Alkanes -Alkane Thiols 0 -Alkane Sulfides 0 Q
I n consequence of the linearity of the plots, the decreasc in density or rvfractive index in going from sulfur compounds to analogous hydrocarbons is proportional to the reciprocal of the molecular iveight. Thus: Ad
90
(Figures Indicate Numbers of Isomers)
7 6 5 4 3 2 I O 1000 / M Refractive index-molecular weight relationship: hydrocarbons-sulfur II
Figure 2. compounds
IO
9
8
the original), the average structural group analysis can be carried out by the van Nes-van Westen or by any other suitable method. The values for Cd and C, may be deduced immediately from the plots in Figures 1 and 2. Since the relationships are all linear and the lines for hydrocarbons and analogous sulfur compounds must intersect at a value of zero for l/Jf) i t is necessary only to take the property difference a t some convenient value of l/X and divide by this value of l/X. Thus, a t the l / X value of 0.010, the density increment in going from alkanes to alkane thiols or sulfides is 0.834 to 0.691 or 0.143. Dividing by 0.010 gives 14.3 as the value for C d . Dividing further by 32 X IO2 yields 4.5 for the value of kd
.
Table I lists the k values. for the refractive index and density, for the 12 different structural types of sulfur compounds considered. I n view of the wide variety of compound classes represented, the spread in values is surprisingly small. For gross mixtures such as petroleum fractions, it appears justifiable to employ the mean values given in the table. A consideration of the uncertainties involved in any procedure of this sort and the inherent errors in any of the structural group analysis methods suggest t h a t the approximately 10 to 20YGmean deviation in the k values is not significant. An alternate approach, of course, would be simply to calculate the quantities
$ and
An
-
s
for the several sulfur compound classes
directly using data for individual sulfur compounds and their corresponding hydrocarbon analogs. Unfortunately, reliable data for sulfur compounds are as j e t too limited, and the method using reciprocal molecular weight plots appeared likely to yield a better average for the property increment per atom (or per cent) of sulfur. Measurements at Higher Temperatures. Because of t h e highly viscous nature of many sulfur concentrates, i t is often necessary t o employ higher temperatures for physical property measurements t h a n iiould otherwise be the case. For the handling of viscous oils in ordinary hydrocarbon type analysis, 70" C. has been generally accepted as a n alternate standard t o the conventional 20" C. Unfortunately, there are practically no reliable physical property (lata available for sulfur compounds a t elevated temperatures and, in the absence of such information, it is inipossible to say what effects a temperature rise would have on the physical property correction factors. If it is assumed, however, that the temperature coefficients for any particular sulfur compound type are not markedly different from those of the analogous hydrocarbons, which seems likely, the correction terms should be applicable at any temperature consistent with t h e requirements of the van Nes-van Westen or other structural group analysis method cniployed. Application of Correction Factors to Structural Group Analysis Equations. I n t h e v a n Nes-van Westen method of structural group analysis, t h e per cent carbon in t h e form of aromatic rings, yo C A , a n d t h e per VOL. 34, NO. 2, FEBRUARY 1962
251
cent carbon in cyclic (naphthenic and 0 CR, for aromatic) structures, 7 measurements at 20' C., are given in terms of two parameters, v a n d w, respectively, which are, in turn, linear functions of the density and refractive index (3). Introducing the above derived sulfur correction terms into the equations for aromatic carbon, the following are obtained:
% C A = 430 v
the density and refractive index correction terms for sulfur:
lo-%'),
+
for a positive; (9)
+
RA = 0.44 0.080M ( V 0.9 X 10--35), for u negative; and (10) RT = 1.33
+ 3660/M +
+ 0.146M (W - 4.0
X
10-3s),for w positive; (11)
0.385( 0 , positive) (5)
% C A = 670 v
+ 0.055-V( a + 0.9 X
RA = 0.44
RT
+ 3660/M +
=
1.33
+ 0.180M (W - 4.0
X
lo-%), for w negative (12)
0.59s(a, negative) (6) The corrections for % ' CA and RA are obviously small, and the differences between values calculated by Equations 5, 6, 9, and 10, and those by the original van Nes-van Westen equations, for which no sulfur correction was recommended, will show up only at relatively high sulfur contents. The differences between the sulfur corrections in the present relationships for yo Ca (w,positive) and RT and the empirical corrections advocated by van Nes and van Westen are also slight, especially if the sulfur content is low; only for % ' C E (w,negative) is the refractive index-density correction for sulfur appreciably different. For oils showing relatively large values for yo C A and
Similarly, for cyclic carbon:
% CR = 820 w
+ 10,000/M3.25(w,positive) (7)
% C R = 1440 w
+ 10,60O/M 5.75(w,negative) (8)
The van Nes-van Westen method (3) also provides equations, valid for measurements at 20' C., for the calculation of the average number of rings of aromatic type per molecule (RA) and the average number of rings of both aromatic and saturated types per molecule (RT). These relationships assume only six-membered, kata-condensed ring systems to be present. Again, applying
Table
Mixture
II.
n y o Cs d20" C. 1.5528 0.9856 180" 1754 1.5582 0,9918 11 1.5623 0.9973 111 1.4996 0.9017 IV V 1.5401 0.9559 1.5173 0.9357 +I 245= 1.5111 0.9258 VI1 0.9207 255" VI11 1 ,5082 242O 0.9654 1.5391 IX a Calculated from component values. Ebullioscopic method. E Combustion analysis. Calculated, van Nesvan Westen method (3). Calculated, Hazelwood method ( 2 ) . Table 111.
% Sa 9.88 7.05 5.44 nil nil 8.32 5.55 4.16 5.55
81.37" 84.31a 85.91" 87.37c 89.32c 80.27c 82.7gC 83.59~ 84.85.
RT RA l.04a 1.12d 1.57* l.73c I 1.30" I1 1.67" 1.57* 1.97" 1.10" 1.15d 2.12~ 1.13" l.17d 111 1,8ga 2.036 1.215 1.4gb 1.46c VI 0.71* 0.83d 0.72d VI1 1.55" . 1.79* 1.64~ 0.65. 1.73c 0.63" 0.6W VI11 1.78" 1.94' 2.13~ 1.06" 1.12d IX 1.8P 2.09' a van Nes-van Westen, with empirical correction (3). b van Nes-van Westen, with TLD - d correction (Eq. 11). Haeelwood, with WJ - d correction (Eq. 14). d van Nes-van Westen, with TLD - d correction (Eq. 9). e Haeelwood, with n D - d correction (Eq. 1 3). f van Nes-van Westen with TZD - d correction (Eq. 7 ) . 0 Haeelwood, with %D - d correction (via Eq. 14). h Van Nes-van Westen, with TLD - d correction (Eq. 5 ) . i Haeelwood, with TLD - d correction (via Eq. 13). ANALYTICAL CHEMISTRY
More recently, Haeelwood (2) has introduced a new structural group correlation which has the advantage of being applicable to highly aromatic oils. Limiting conditions are that RA should exceed 0.9 and that RT should be greater than 1.2; sulfur is assumed t o be absent or present only in negligible amounts. Because much of the sulfur in high sulfur content oils is combined in ring structures and because, in the usual methods of physical separation, aromatic hydrocarbons are concentrated with sulfur compounds, it appeared that this method, appropriately corrected for sulfur, might provide more reliable results than those obtained by the corrected van Nesvan JT7esten procedure. The Hazelwood equations were developed on the basis of measurements made at 20" C. Introducing the density and refractive index corrections for sulfur, the Hazelwood equations become : RA = 0.040M [2.75( n - 1.4750)( d - 0.8510) 0.3 X lO-%S]f 0.58 (13)
+
Rr = 0.08021.f [(d - 0.8510)0.50(n - 1.4750)- 5.3 X . lO-SS] 1.10 (14)
+
RT 1.73" 1.88" 1.97" 2.31d 2 . 65e 1.70" 1.82. 1.975 2.070
RA 1.45. 1.500 .. ~
1.53" 0.61d 1.37O 1.03" 0.91a 0.84" 1.31'
% CR 69.3" 77.94 . . .. 81.26 54.2d 63.Oe 52.9" 53.3" 53.5" 61.4"
% CA 59.5" 63.9" 65.0" 17.5d 38.5e 34.8" 29.1" 26.2O 44.0"
Summary of Ring Analyses for Test Mixtures
Test Mixture
252
v and w are positive.
Properties and Ring Analyses of Test Mixtures
Molecular Weight
I
yo CR, it is usual that the parameters
1.17e 1.21~ 1.25E 0.89' 0.84c 0.8le 1.20*
65.5" 75.7" 82.4" 49.6" 52.6" 53.6" 60.1"
% CR
-
- % CA
63.1' 74.01 81.11 47.6' 51.31 52.61 58.8'
68.00 77.10 53.08 47.79 48.70
46.7a 50.la 52.6. 25.2h 21.7" 20.2' 35.ln
48.80
60.09
50.3* 52.7h 54 6 h 28.2h
23.7h 21.7h 37.1h
50.9' 53.4' 55.4' 33.8' 30.5' 28.7' 38 8'
Table IV.
Test Mixture I I1 I11
Per Cent Deviation Ring Analyses, Calculated as indicated, vs. Analyses Deduced from Composition
(Positive sign denotes calculated value is high; negative sign denotes calculated value is low.) KT
-24.9. -11.2" -4.1" -28.8" -13.2'
-9.3b -0.5b
OC
-27 .4a -26.74 -26.10 -31.0" -28.6"
RA -22.8' -23.3' -23.5' -19.4' -20.gd -19.0d -14.5d
+4.@ +3. l b +i.6c -12.3' VI -14.1C -1.7b VI1 -10.6C -1.P VI11 -9.6" -12.2C -25.0" IS -9.20 1.O b -19.1' +2.9" a van Xes-van F~esten,with empirical correction (3). b van Xes-van Westen, with ~ Z D- d correction (Eq. 11). Hazelwood, with n~ - d correction (Eq. 14). van Nes-van Westen, with I ~ D- d correction (Eq. 9). a Haeelwood, with TLD - d correction (Eq. 13). van Xes-van JTesten, with n~ - d correction (Eq. 7). g Hazelwood, with TLD - d correction (via Eq. 14). h van Xes-van Westen, with n~ - d correction (Eq. 5). Hazelwood, with n~ - d correction (via Eq. 13).
+
Table V.
- 19.3e -5.5"
-19.4' -18.3" -13.6'
-7.7E
-3.6a -10.7*
-2.8"
f1.54
-5.7" -1.3" +o. 2" -2.1"
% CR -8.9' -5, Of -0.1f
-10.0f -3.7f -1.7f -4.21
- 1.98
-21.5"
+2.28
-19.1" -27.6' -25.4" -22.9" -20.2'
- 1.0s -21.6"
-lo.@ -8.6' -8.9
-2.29
-15.5* -17.5h -16.0h -19.0h -18.5h -17.2h -15.7'
-14.4' -16.4' -14.8' -2.9' +4.8' t9.5'. -11.8'
Typical Structural Group Analyses of Petroleum Sulfur Concentrates
Sample Code: A, Rest Texas (Permian) B, Kuwait (Burgan) C, Venezuela (Light Mnra)
Moleculara yo Sample nZ,O O C . d2ODc. Weight Sulfur RT RA 70CR A-1 1.5365 0.9570 315 4.08 2.32' 2.45' 1.34' 1.391 49.40 52.0' C-1 1.5393 0,9616 390 4.62 2.53" 2.78' 1.60' 1.64f 42.70 46.7' A-2 1.5718 1,026 434 4.01 4 . 6 5 ~ 4.74' 2 . 1 4 ~ 2.19' 75.60 66.6h C-2 1.5856 1.000 185 5.43 1.46b~C 2.06' 1.776 1.77' 58.Ob80 74.8h B-1 1.6102 1.058 265 9.07 2.15b.c 2.79' 2.526 2.351 54.gbJ 67.5b B-2 1.8349 1,092 246 7.87 2.49b.c 3.15' 2.708 2.55f 67.3'80 79.gh A-3 1.6391 1,086 251 6.70 2.30b~c 3.46' 2.920 2.60f 61.5*#l 84.8h C3 1.6469 1.098 223 6.80 2.28b.c 3.43' 2 . 7 4 ~ 2.63f 68.gb*# 94.2' C-4 1.8i65 1.145 264 6.63 3.04bmC 4.44' 3.546 3.331 74.1'80 99.2h A4 1.6784 1.1615 289 7.14 3.70b,~ 5.03d 3.728 3.42f 80.9b3' a Ebullioscopic method. Note that aromatic ring carbon exceeds total ring carbon; also that aromatic rings exceed total rings. van Nes-van Westen, with nD - d correction (Ea. . * 11). Hazelwood, arith n~ - d correction (Eq. 14). van Xes-van V-esten, with n D - d correction (Eq. 9). f Hazelwood, QTith TLD - d correction (Eq. 13). van Nes-van Westen, with n~ - d correction (Eq. 7). Hazelwood, with nD - d correction (via Eq. 14). van Yes-van Westen, with n D - d correction (Eq. 5). 2 Hazelwood, with TLD - d correction (via Eq. 13).
% (3.4 33.9' 32.9' 34.2' 77.0' 73.9' 86.6' 93.1' 98.0'
loo+'
100f'
33.3' 30.5' 34.3' 66.3' 58.4i 66.7j 66.3' 75.0' 77.0" 71.8'
6
Equation3 for yo CA and yo C x remain the same, except that the values of RA and Rr are n o w those calculated by Equations 13 and 14. The value for C , the total number of carbon atoms per molecule, may still be gotten by direct analysis, provided the determined value for carbon is increased by the substitution of two carbons for each sulfur atom present. Evaluation of t h e Corrected van Nes-van Westen and Hazelwood Equations. I n the original work on pure hydrocarbon oils, v a n Nes and v a n Westen a n d more recently Hazelwood estimated analytical errors by applying their equations t o a series of oils, t h e ring compositions for which had been determined independently by elemental analysis before and after exhaustive hydrogenation. KO similar approach appears feasible for the present problem. Khile it seems likely that refractive index and
density corrections of the type developed here will allow the basic equations to be applied to sulfur-rich oils, it remains t o be demonstrated that the results obtained are even approximately correct. EXPERIMENTAL
As a compromise, the revised equations were tried on a series of test mixtures, made u p from pure hydrocarbon and sulfur compounds and from sulfur-free hydrocarbon oils, and the results were compared with the ring analyses as calculated from the components. To this end, two mixtures were preparedone containing seven pure aromatic and naphthenic hydrocarbons and one containing 10 divalent sulfur compounds. The former consisted of one mole each of Tetralin, 2-methylnaphthalene, phenanthrene, diphenylmethane, diphenylethane, and triphenylmethane admixed with two moles of Decalin (L1\57ay.,
165.7; % C, 91.48; dZo0'., 0.9980; naZo c*, 1.5737; C T , 12.6); the latter, of equimolar amounts of tert-butylthiophene, ditert-butylthiophene, ditertoctylthiophene, benzothiophene, diphenylene sulfide, ethyl tolyl sulfide, diphenyl sulfide, dibenzyl sulfide, dipentyl sulfide, and diheptyl sulfide (MWav., 192.1; % S, 16.68; % C , 74.42; C T , 13.9). These two mixtures were combined in various proportions to produce three secondary mixtures with sulfur contents ranging from 5.4 to 9.9% (I, 11, 111). Each was examined for refractive index and density; the molecular weights and sulfur contents were computed from the composition. Further, a sulfur-free mineral oil (IVj and an aromatic hydrocarbon concentrate (V) derived from i t were examined in the same way, and their structural characteristics were determined. Three synthetic sulfur concentrates (VI, VII, VIII) were then prepared by the addition to the oil of varied amounts of the sulfur compound mixture; a single blend (IX) was made up from this VOL. 34, NO. 2, FEBRUARY 1962
253
sulfur compound mixture and the aromatic concentrate. The physical properties of these composites, as well as their calculated ring analyses, are listed in Table 11. The values of RT, R.4, % C E , and % C.$ were now calculated for the three sulfur compound concentrates and the four petroleum blends using the several alternate equations. Results are shown in Table 111. Per cent deviations of these results from those calculated as indicated in Table I1 are listed in Table IV. Finally, for comparison, ring analyses are given in Table V, computed by the sulfur-corrected van Xes-van Westen and Hazelwood equations, for a series of petroleum sulfur concentrates which had been prepared by elution chromatography over alumina and distillation from three high sulfur content crude oils. DISCUSSION
9 survey of the results leads to sevcral conclusions regarding the validity o f the various equations for structural group analysis. The van Ses-van Westen method (with empirical sulfur correction) re-
sults in large negative errors for RT which become worse as the sulfur content increases. Application of the refractive index and density corrections to the van Nesvan Westen or to the Hazelwood equations produces notable improvements in the RT value. Deviations are relatively smaller and are both positive and negative. When an oil is rich in both sulfur and aromatics as, for example, in the case of the heavier of the petroleum sulfur concentrates (Table V), the value for RT, calculated by the van Nes-van Westen equations, frequently falls below that for RA, even when the refractive index and density corrections are applied. This results, of course, in a negative naphthene content. Use of the corrected Hazelwood equations, however, leads to more consistent figures, a t least to the extent that RT exceeds RA by what appears to be a reasonable amount. With regard to Ra, the refractive index and density corrections improve somewhat the van Nes-van Westen equations (compare the empirical sulfur correction); still better results are obtained with the corrected Hazelwood equations, although the deviations are appreciable and all negative,
For the calculation of C1, the van Xes-van Testen equations with empirical sulfur corrections are very poor; notable improvement results, however, when the refractive indes and density corrections are applied. The corrected Hazelwood equations are the best, both with respect to the sign and the magnitude of the deviations. For the calculation of the yo , C R , the van Xes-van Westen equations appear superior; the evidence points to the possibility that the sulfur correction term, advocated originally by these authors, may be better than the slightly higher value resulting from the refractive indes and density corrections. The corrected Hazelwood equations also make a relatively poorer showing. However, in all three cases, the deviations in yo C R are far less than those for RT, R,, and % C A and, hence, the choice of method for % C R is probably not critical. An additional argument for the corrected Hazelwood equations is the observation that, for natural petroleum sulfur concentrates (Table V) , high aromaticity and sulfur content lead to yo C E values by the corrected van Xes-van Westen equations which are less than the corresponding values for % CA, implying negative contents of naphthenes.
2CL *O
16500
164CO 1.6300
1.6230
1.6lW
-
1
11
10
r
10
-
20
-
9
E
E
8
7
30-
6 -
R,*O.OWM [d-O.Sn-0.1135-5.3
I
lO-'S]
t 1.10
C~*O,O7OMtO.3R~tO.IR~
5
40
-
so
-
4
3
2
MEAN TOTAL N U M B E R OF R i N G S PER
MOLECULE
Ref. R.N. Hazelwood, Anal. Chem...E,1073 ( 1 9 5 4 ) l
60
-
(Corrected tor Sulfur 1 A p p l i c a b l e to Petroleum oils with: I. R e t Ind. > 1.57; Den. > 0.99 or 2. Ret. Ind.< 1.57; Den.< 0.99 but S 3 %ond M W. 400
>
I4600
70 -
1
I
14500
Figure
254
ANALYTICAL CHEMISTRY
3. Nomograph for RT
>
,6500
,6500
L
,63011
I$??)
-
t -
16'20
I6300
-
15930
I.20C
1.5800
1.15c
15700 %s
l.lOC
100
I5600
1.0%
1.5590
00
15400
1.5l0C
-
-
I49CL
-
1.4'X
MEAN NUMBER OF AROMATIC RINGS PER MOLECULE
0.90C
,L?C'
,483:
R A
09%
1.5300 Z
1.5200
I.OOc
a,1073 (13541
Ref. R.N.Hazelwood, &d m, 08X
(Corrected f o r S u l f g r l 099:
A p p l i c c b l e t o P e f r o l e i l m oils with I. Ref. Ind.) 157, Den.> 0 9 9 o r 2. Ref. Ind.< 157; Cen 0 99 b u t
3 2 snd M W . > 400
~-
:-
k
Figure 4. RECOMMENDATIONS
The foregoing results indicate clearly that some discrimination must be used in the choice of equations for the four structural charactr>ristics: RT, RA, 70 C E , Yo C q . There is no sharp line that separate. aromatic-rich, high sulfur content oils from the less aromatic, lower sulfur content oils nhich are amenable t o treatment by the van Kcsvan Westen procedure. Furthrr, the sulfur compound mixtures and the resultant blends x i t h hydrocarhono and with petroleum oils, by which the severnl equations are being judged, are fen in number and hardly typical of true d f u r concentratcs from petrolrum. I'he percentage deviation figures (Table 1V) must be regarded, therefore, as a qualitative rather than a quantitatim guide. The results of analyses on true petroleum sulfur concentrates are valuable in this connertion only to the e ~ ; tent that they show TT hrn a n equation is badly in error (inconsistent rmuit R T ) . Experience with the several methods, c o u p l d n i t h the present study, leads to the following recommendations :
OILSWITH SULFUR CONTEST LESS
Nomograph for
RA
TH~X 3y0. Refractive index 0.99; use Hazelwood equations with refractive index-density corrections for sulfur. OILS WITH SULFUR CONTEXT GREATERT a 4 37,. ~ Refractive index 0.99; use Hazelwood equations with refractiy:e index-density corrections of sulfur.
ment, can be made easily in the v a n Nes-van Westen structural analysis, it is necessary to have nomographs only for RT and RA as given by the Hazelwood equations, with corrections for sulfur content; the calculations for C T , % C R ,and % C A are so simple as to require no such graphical aid. As follows from the recommendations in the foregoing section, the nornographs presented as Figures 3 and 4 will serve for oils of the following characteristics: refractions index > 1.57, dencity > 0.99; rrfractive index < 1.57, drnsity < 0.99, but sulfur > 356 and molecular weight > approximately 400. LITERATURE CITED
NOMOGRAPHIC CHARTS
(1) Camisa, A. di, Fratta, C. A,, Xetistu dei Conibustibili 12, 315 (1953). ( 2 ) Hazelwood, R. N., .ANAL. C m x . 26,
I n practice, it is inconvenient to use equations of the sort presented here. As a n aid in the analysis, van Xes and van Westen reduced their method to nomographic form, with a resulting substantial saving in calculation time, Since sulfur corrections, either empirical or by refractive index-density adjust-
van Ses, K., van Westen. H. il., (3"A4spects of the Constitution of Mineral Oils," pp. 335-49, Elsevier, New York, 1951. RECEIVED for review May 5 , 1955. Resubmitted -4uguct 10, 1961. Accepted December 8, 1961. Kork was sponsored by the Gulf Research dl- Development Company as part of the research program of the Multiple Fellowship on Petroleum.
1073 (1954).
VOL. 34, NO. 2, FEBRUARY 1962
255
