Relationship between Carbon-Type Composition, Viscosity-Gravity Constant, and Refractivity Intercept of Viscous Fractions of Petroleum STEWART S. KURTZ, Jr., RICHARD W. KING, WILLIAM J. STOUT', DOROTHY G. PARTIKIAN, and E. A. SKRABEK2 Sun Oil Co., Marcus Hook and Norwood, Pa. A triangular graph has been developed which relates carbon-type composition in terms of per cent aromatic carbons, per cent naphthenic carbons, and per cent paraffinic carbons to viscosity-gravity constant and refractivity intercept for viscous petroleum fractions. If both viscosity-gravity constant and refractivity intercept are known, carbon-type composition can be determined from this graph with reasonable accuracy. Viscosity, gravity, and refractive index are the only experimental data required for the use of this graph. It is useful in following the effect of solvent extraction, acid treating, hydrogenation of the viscous fractions of petroleum, and for comparing the composition of the viscous fractions from various crude oils. T WO other laboratories have been working independently on the correlation of density, refractive index, and viscosity to provide a simple method for the analysis of petroleum fractions (5, 41). The method presented here is believed to be as accurate as the other methods, and more simple to use. The information v hich has been accumulated in recent years concerning the lubricating oil fraction of petroleum indicates that the structures of the molecules in the lubricating oil fraction are fairly uniform ( I , 2, 5-11, 14, 16, 17, 20-22, 24-27, 29-33, 36-37, 39, 40, 43-46). It seems probable that only a very small portion of the theoretically possible structures actually exist in lubricating oil. Both viscosity-gravity constant (12) and refractivity intercept (16, 18, 19) were developed with the intent of obtaining constants which were indicative of the type of hydrocarbon composition and independent of molecular weight. Therefore, it seemed possible that lines of equal viscosity-gravity constant and equal refractivity intercept could be established on a three-component graph for carbon-type composition. Carbon type composition is used in the conventional sense (30, 31, 43). Specifically, per cent aromatic ring carbon means the per cent of the total carbon atoms in aromatic ring structures; per cent naphthenic ring carbon means the per cent of the total carbon atoms in naphthenic ring structures; and per cent paraffinic chain carbon means the per cent of the total carbon atoms in chain structures, free as \\ell as combined with naphthenic and aromatic rings. Van Xes and van Westen, in their book "Aspects of the Constitution of Mineral Oils" (31), have discussed both refractivity intercept (33) and viscosity-gravity constant (32) in relation to the carbon-type composition, and presented the following approximate equations: = 0.01065 ri V.G.C. = ( % Ca) + l l ( 5 6 C A ~ O - +~ )0.0103('% ~ CN) + 0.0105 ("/c C P ) (1) O.OllO(G C a ) + 0.00925(% CN)0.00743(% + CP) 2 address, University of Delaware, Newark, Del. Present address, University of Wisconsin, Madison 6, Wis. V.G.C. T7.G.C.= \-.G.C. = = 10G - 1.Oi52 log ( V , - 38) 10 - log ( J r i - 38) G - 0.21 - 0.022 log (1'2 - 3 5 . 5 ) 0.755 d - 0.1384 log ( V , - 20) 0.1526[7.14 - log (V3 - 20)] ~~ + 0,579 (3) (-1) (5) where G = = VI = V 2= V 3= d specific gravity 60/60° F. density a t 20" C./4" C. Saybolt viscosity a t 100' F. Saybolt viscosity a t 210' F. kinematic viscosity in centistokes a t 20" C. These equations all give viscosity-gravity constants equivalent to the Hill and Coats equation for 100' F., which is Equation 3. Some of the carbon-type composition data were obtained by the Martin method (16, 26) of calculation and some by the n-d-M method (30, S1, 43). For aromatic extracts and other compounds rich in aromatic rings, which are outside the range recommended for the n-d-Lf method, the Martin analytical technique was used. VISCOSITY-GRAVITY CONSTANT CORRELATION (2) These equations were presented as somewhat tentative, and were not recommended for general use because of the uncertainty in regard to the effect of aromatic carbons on the physical properties. These authors quote Leendertse (20) in regard to the reliability of the refractivity intercept for determining the 1 Present naphthene carbon content and the paraffin carbon content of aromatic-free lubricating oil fractions. They were impressed with its reliability for saturated mineral oil fractions. The intercept for paraffin carbons chosen by Leendertse (1.0502) and the value used here (1,0500) both correspond closely with the value for the limiting point of the paraffin homologous series (1.0498) (4). A lower value (1.0480) has also been investigated, because there is no reason to believe that the limiting CH, inO crement is necessarily the best increment to use in the C ~ to Cto molecular weight range. However, the higher value is better for data on lubricating oil fractions. This will be discussed in more detail below. The basic problem was to collect enough data for viscous oils and fractions of viscous oils so that the relation between carbontype composition, viscosity-gravity constant, and refractivity intercept could be established on an empirical basis. Carbontype composition can be obtained by the n-d-M procedure if data are available for density, refractive index, and molecular weight (31, 43). If molecular tveight is not available it can be estimated from published correlations of molecular weight with physical properties (28). D a t a were collected from the authors' own files (42) and from the literature (10, 25, S1) for approximately 258 lubricating oils and fractions of lubricating oils for which carbon-type composition, viscosity-gravity constant, and refractivity intercept were available. Viscosity-gravity constants were calculated with the following equations or the corresponding nomographs: I n development of Equation 2 van Nes and van Westen observed ( 3 2 ) that for saturated oils a linear relationship existed between the viscosity-gravity constant (V.G.C.) and percentage carbon in paraffinic structures ('% Cp). There also seemed to be some relation between viscosity-gravity c o n s t p t and % CP for a number of oil fractions from widely differing crude sources. Their observations suggested that it might be possible to establish lines of constant viscosity-gravity constant on a triangular coordinate system by first establishing a relationship between viscosity-gravity constant and the percentage carbon in aro- 1928 V O L U M E 2 8 , NO. 1 2 , D E C E M B E R 1 9 5 6 1929 Table I. Data Necessary for C o n s t r u c t i o n of ViscosityGravity Constant Lines" in F i g u r e s 5 and 7 Coordinate Point 1 Coordinate Point 2 V.G.C. % CA % CN % CP % CA % CN % CP 0.78 4.0 17.0 79.0 0 23.0 77.0 0.79 5.7 18.0 76.3 0 27.0 73.0 0.80 7.5 19.5 73.0 0 31.6 68.4 20 70.2 0 35.8 64.2 0.81 9.8 0.82 12.5 20 67.5 0 40.0 60.0 0.83 15.5 20 64.5 0 45.0 55.0 0.84 18.2 20 61.8 0 60.0 50.0 0.85 21.0 20 59.0 0 54.8 45.2 0.86 23.6 20 56.4 0 60.0 40.0 0.87 26.4 20 53.6 0 66.1 33.9 20 51.0 0 73.0 27.0 0.88 29.0 0.89 31.6 20 48.4 0.7 79.3 20 0.90 34.0 20 46.0 7.5 72.5 20 0.91 36.3 20 43.7 13.7 66.3 20 0.92 38.2 20 41.8 19.9 60.1 20 0.93 40.0 20 40.0 25.0 55.0 20 0.94 41.5 20 38.5 30.0 50.0 20 0.95 43.0 20 37.0 34.5 45.5 20 0.96 44.4 20 35.6 38.4 41.6 20 41.9 38.1 20 0.97 45.7 20 34.3 0.98 46.9 20 33.1 44.5 36.0 19.5 0.99 48.0 20 32.0 47.3 34.7 18.0 1.00 49.2 20 30.8 50.0 33.0 17.0 1.01 50.5 20 29.5 52.0 31.7 16.3 Table 11. Accuracy of Viscosity-Gravity C o n s t a n t Carbon-Type Correlation Source of D a t a Lit. Ref. No. of FracAv. tions V.G.C. Std. Dev. Dev. of Averagea 0 830 0 926 0.845 0,813 0.858 0.004 0 004 0.004 0,004 0.006 0.000 +O.OOQ -0.002 - 0 002 -0,004 + O 006 -0,006 -0.007 -0,006 +0.010 +0.010 -0.007 Max. Dev. Whole oils and cuts thereof Water White Oils Borneo Crude OklahomaCrude Pennsylvania Crude Webster Crude Distillation fractions from Naphthenic Crudes P a . 180 S e u t r a l Midcontinent Neutral I California Neutral Gulf Coast Keutral Rodessa Pieutral Midcontinent Seutral 2 ($6) 11 (91) (91) 5 5 (91) (SI) 4 4 (Sf) (10) 21 20 0.883 0.819 0.004 0.002 0 000 0.000 17 17 15 0.860 0 892 0.879 0.830 0.003 0.002 0.003 0.002 +O 001 0.000 0.000 - 0 002 005 -0 005 $0.006 -0 004 + O 006 (10) (10) (IO) (IO) 18 +O 19 0.849 0.004 + O 001 156 0 858 0.003 0.000 (86) (48) 12 3 0.800 0.874 0.002 0.007 0,000 +O 006 + O 004 + O 011 (42) 21 0.836 0.004 +0.003 +0.009 36 0.837 0.004 t 0 001 +0.011 (42) 3 0.802 0.008 -0 -0 (48) (48) 3 0.956 4 0.966 0.008 0.014 -0 007 - 0 014 -0,010 - 0 015 (48) 2 0.945 0.014 -0 014 - 0 015 (48) 6 0.9G9 0.005 -0 001 -0 010 1 32 0.948 0 009 - 0 006 -0 015 224 0.005 -0 001 -0 015 Coordinate points given are terminal points of the V.G.C. lines and define area of composition over which correlation has been tested. a t t 1 40 44 40 52 56 60 64 68 72 76 PER CENT Cp F i g u r e 1. Viscosity-gravity constant cs. saturated oils 7' C p for All whole oils Saturated oils Water White Oils Vebster Crude Distillation fractions froni Naphthenic Crudes All saturated oils Aromatic extracts Aromatic conc. froni solvent refined lube b Aromatic cone. from light lubes b Aromatic extracts Aromatic conc. from Webster Crudeb Aromatic conc. from Duosol and furfuralextractionsb Aromatic conc. of (10) -411 aromatic extracts Complete total .92 z z 4 L 0 ~ .90 .88 c > .86 a i- 010 010 a Deviation from average of each d a t u m averaged, taking sign into account. b Concentrated on silica gel; a t least one aromatic ring per molecule. .94 r 00; +O .84 v) 0 :-: . 8 2 > .80 Viscosity-gravity constant plots for w h o l e o i l s 1930 ANALYTICAL CHEMISTRY matic, naphthenic, or paraffinic structures on rectangular coordinates. For saturated oils, the linear relationship betvieen viscositygravity constant and % C p observed by van Xes and van Kesten (32) was confirmed, as shown in Figure 1. Because theae oils contained no carbon in aromatic structures, it was possible to transfer this line to the saturated base line of a triangular coThis gives a series of ordinate graph for % C P , % C.V, and lo CP-% CS side of the viscosity-gravity constant points on triangle. I n order to establish a second point for the viscosity-gravity constant lines, it was necessary to plot tn-o of the composition variables as a function of viscosity-gravity constant for the series of whole oils. Per cent C P and % C A gave the best approxiniations to a linear relationship when so plotted (Figures 2 and 3,. I'ISk B .90 t A plot of 70 C.V showed considerably greater scattering, and is presented in Figure 4 merely to show that a line constructed from % C A )fits the Figures 2 and 3 so that 70C.V = 100 -(70C p data reasonably well. It was possible t o fit the data for C P and % C A n-ith a straight line up to viscosity-gravity constant values of about 0.885. Above 0.885 the effect of the carbons in aromatic structures became large, and it was necessary to change the slope abruptly in order to fit the data in this region. These data mere transferred to triangular coordinates and points of equal viscosity-gravity constant connected to give a series of lines covering the viscosity-gravity constant range from 0.790 to 0.910 I n the range from 0.910 to 1.00 the lines were placed by inspection to give the best fit to the data. Adjustments to the correlation mere made until the average deviation was a minimum and the deviation of the average was close to zero. The viscosity-gravity constant lines on Figure 5 are the final adjusted lines. Table I gives the data necessary to lay out the lines for viscosity-gravity constant on a three-component graph. For the present, a t least, the lines are being restricted to that area of the graph for which supporting experimental data exist. Table I1 shows the degree of agreement obtained for 224 fractions between the viscosity-gravity constant determined experimentally and that derived from composition by using a chart corres onding to Figure 5 . The agreement between experimental and cafculated values, while not perfect, is satisfactory. It indicates that the reverse computation-that is, relating viscosity-gravity constant to composition-should be satisfactory, provided anothei function, also independent of molecular weight, could be plotted on the graph in order to obtain intersecting lines. + REFRACTIVITY INTERCEPT CORREL4TION I.80 I I I I I I I I I I I I I 0 IO 20 30 40 50 60 70 80 PER CENT C N Figure 4. Viscosity-gravity constant whole oils TS. YC C.v for It seemed reasonable to use refractivity intercept as the other plotted function, provided the effect of aromatic carbons could be evaluated on a reasonable basis. As the aromatic content of lubricating oil fractions increases, there is evidence that the proportion of condensed aromatic ring structures increases. Because the composition data for the 224 fractions that Rere available to the authors formed a fairly narrow band on the triangular diagram of carbon-type composition, it seemed reasonable to anticipate that a fairly smooth curve should be obtained if the equivalent refractivity intercept of the aromatic carbons were plotted against the observed refractivity intercept of the sample. Therefore, the equivalent refractivity interiept of the aromatic carbons was calculated using the follon-ing equation: r,A = % AROMATIC RING CARBONS % NAPHTHENE RING CARBONS Figure 5 , r,sample(100) - 1.0300(7,C\) - l . O 5 O O ( % C p ) - % PARAFFIN CHAIN CARBONS Viscosity-gravity constant in relation to carbon-type composition n Figure 6 shows the calculated refractivity intercepts for aromatic ring carbons plotted against the refractivity intercept of the sample for those samples which had 5% or more of aromatic ring carbons. Although some scattering is observed, the anticipated trend is clear. Because the available data did not include points rich in naphthene ring carbons and having low intercept values, some assumption was needed to establish tentative intercept lines in the more naphthenic portion of the composition chart. -4s there cannot be less than one aromatic ring per molecule and as it was not believed necessary to extend the chart below 1.035 refractivity intercept, the intercept value of benzene (1.0616) was plotted opposite the value 1.035 for refractivity intercept to provide a tentative terminal point for this curve. This was a rather arbitrary step which xould only be justified if the chart ultimately developed was reliable. T h e refractivity intercept lines on Figure 7 were established by interpolating betxeen 1.0300 and 1,0500 on the naphthene and paraffin side of the triangle and between 1.0300 and r,A on the naphthene and aromatic side of the triangle, and 1.0500 and riA on the paraffin and aromatic side of the triangle. r,A is determined in each case using Figure 6 or Table 111. The data necessary for establishing the intercept lines in Figure i are given in Table IV. Study of the oints in Figure 6 suggests the possibility that &ere should be a sharp downward curvature of the line for the calculated aromatic carbon intercept betneen the values 1.045 1931 V O L U M E 28, NO. 12, D E C E M B E R 1 9 5 6 ;tnd for API 42 compounds (2, 39, 40) would indicate that a paraffin intercept value of 1.0480 would be better in the range from 20 to 40 carbon atoms than a value of 1.0500. The corresponding naphthene value is 1.0320. A complete graph was constructed and evaluated. A statistical study showed little difference between this graph and Figure 7 . However, when compared by plotting data for a series of hydrogenated oils, the graph based on the paraffin intercept value of 1.0480 and a naphthene intercept value of. 1.0320 showed the completely hydrogenated naphthene fractions, falling about 2y0 below the base line representing 0% aromatic carbons. The graph based on 100% CP = 1.0500 and 1007, CN = 1.0300 showed the completely hydrogenated fractions falling on the base line (Figure 8 ) For this reason it was decided to adhere to values of 1.0500 and 1.0300 for the paraffin and naphthene apexes, respectively, In constructing this type of graph some thought needs to be given to what is meant by per cent composition. The data for per cent carbon type arp in terms of percentage of aromatic, naphthenic, and paraffinic carbons in the molecule (16, SI j. Refractivity intercept is, ingeneral, additive on a volume per cent basis for mixtures of liqui s. As shown by Leendertse (2Oj, there is good agreement between carbon-ty e composition and refractivity intercept for complex saturate8molecules. The graph in Figure 7 , as derived, represents carbon-type composition and not volume per cent composition or weight per cent compositiorl. This should be kept in mind if it is applied to blends. 'Table 111. Data Needed for Drawing Curve of Equivalent Intercept of Aromatic Carbons os. Intercept of Lube Oil a s Shown in Figure 6 Intercept Intercept of of Aromatic Oil Carbons 1 0616 1.0350 1 ,0695 1.0400 1 0775 1.0450 1 0855 1.0500 1 0926 1,0550 1.0986 1.0600 1.1042 1 ,0650 1.1094 1 ,0700 1.1140 1.0750 1.1186 1.0800 Data for Establishing Refractivity Intercept Lines" NaphtheneNaplitheneParaffinItefracParaffin Aromatic* Aromatic Base Line Side of Triangle Side of Triangle tivity % CP % CS 5% CA 7%C.V yo CA 7% C P Intercept 25 75 15 8 84 2 1 035 74.7 50 25.3 i 040 50 68.4 25 31.6 75 1.045 63.9 36.1 0 1.050 100 60 1 88.3 39 9 11.7 1.055 56 3 79.5 20 5 43 7 1 060 72.3 52 8 27 7 47 2 1 065 po 4 49 6 33 7 66 3 1 070 39 1 46 4 93 6 BO 9 1 n7.5 i &I 56.4 43 4 43.7 56 3 In constructing Figure i values for intermediate lines can he 011tained by plotting columns 2 to 7 us. column 1 and interpolating. h working graph should have lines for each 2% of carbon-type (TJYIposition, for each unit in the third decimal place of the refractirit?. intercept, and for each unit in the second decimal of viscoeit!.-gravit!. constant. Table IV. ~~ APPLICATIO\ The carbon-type composition of the lubricating oils and frwtions of lubricating oils used in developing Figure 7 , as well as :L number of other oils on n hich data were available, was determined by plotting the viscosity-gravity constant and refractivity intercept for all these points on a large graph and reading off the corresponding composition. This study showed that the aromatic. carbon contents are usually in agreement with the authors' best analyses to within 1 or 2%, although occasionally deviation3 as large as 4 or 5% may be observed. On the paraffin and naphthene compositions, the agreement is good for fractions containing 30% or less of aromatic carbons. Between 30% and 50% aromatic carbons, there is an increase of uncertainty in the values for the paraffin and naphthene carbons, because the angle at 11-hich the correlation lines intersect decreases rapidly. Thp chart should not be used for samples having viscosity-gravity constant values between 0.95 and 1.01 which fall below 20% C,vor 20% CP. This is practical because very few samples having less than 20% CNor 2O%Cp have been found in this range. Table V presents data on a number of samples for which rather complete analytical data were obtained in connection with studies and 1.050 for the intercept of the oliginal oil. This was tried, but led to an irregular spacing of the intercept lines on the triangular diagram. The lines so placed did not agree well with the limited available data for this region. Therefore, as explained above, the coordinatee 1.0616 and 1.0350 were tried as the terminal point of the curve. This led to more reasonable spacing of the intercept lines and better agreement with the data. Considerable thought has been given to the significance of pure compound data in relation to locating the intercept lines on Figure 7 . Consideration of data from the A4PI44 tables ( 3 , 35) I I In I I I I I I 1.12 m a 4: g gI 1.11 1.10 a LL ..- 0 1.09 n w tj 1.08 i8 1.07 1.030 1.040 1.050 1.060 1.070 1.080 r i OF SAMPLE Figure 6. Refractivity intercept of aromatic carbons us. refractivity intercept of sample ANALYTICAL CHEMISTRY 1932 of rubber processing oils (16, 48). I n general, the agreement between composition obtained with Figure i and the composition obtained by more complete analysis is satisfactory-that is, within 1 or 2%. -4few deviations larger than 4% are shonn. Table V also contains data obtained by the n-d-bI procedure of Tadema, van Xes, and associates (31, 43). I n the range for which the n-d-11 method is recommended-that is, for samples in which the ratio of % C A to yo Cy is 1.5 or less-the agreement is quite good. I n the higher aromatic range the data obtained by Figure 7 are in better agreement with detailed analyses b j the Martin (16, 26) method than are the data by the n-d-11 method . Table VI compares carbon-type analyses obtained x ith Figuie i and with the n-d-?\I procedure on the cuts of five tlpical oils studied by Hill and Ferris (13). The data on these oils xere used by Hill and Coats (12) in deriving the viscosity-gravity constant, The t a o methods agree xell for these typical oils. The range of carbon-type composition is about 20% for % C.4 and % C Vand about 40% for % C p . To prepare Table VI data weie read from photostatic enlargements of the curves published by Hill and Ferris. Data were corrected to 20' C. when necessary. A statistical study of all the data (Table YII) shows that, for oils having a viscosity-gravity constant of 0.900 or less, Figure i nil1 give the value of yo C A n ith a standard deviation of ahout 1.0%. For % C.V and % C p the standard deviation is roughly 1.5% for samples m-ith a viscosity-gravity constant below 0.900. Most naturally occurring crude oils have viscosity-gravity constant values below 0.900. For samples with viscosity-gravity constants between 0.900 and 1.000, the standard deviation for % C A is approximately 2.0% and for % C,y and % C p , approximately 4.5%. The samples having viscosity-gravity constant above 0.900 are, for the most part, aromatic concentrates obtained either by solvent extraction or gel separation. These samples are outside the range of materials for x hich the viscosity-gravity constant was originally developed. Table 1-111presents corresponding data arranged b j groups based on per cent carbon in aromatic rings. For samples having less than 30% carbon in aromatic rings the standard deviation is as follons: per cent aromatic carbons, 1.2%; per cent naphthenic carbons, 2 . 0 % ; per cent paraffinic carbons, 1.7%. The data used in developing this chart represent a Fide variety of samples. Consideration of the data indicates that Figure 7 is sufficiently reliable for obtaining many of the carbon-type composition data ahich are needed in practical petroleum refining. The relation betn-een refractivity intercept and composition, and beta een viscosity-gravity constant and composition is more accurately presented by Figure 7 than hy Equations 1 and 2 % AROMATIC RING CARBONS .os00 LRBONS V O L U M E 2 8 , NO. 1 2 , D E C E M B E R 1 9 5 6 1933 Table V. Comparison of Carbon-Type Analysis by Viscosity-Gravity Constant-Refractivity Intercept and Other Methods 011 No. 1 Mol. wt. V.G.C . T% n d 2 Mol. nt. V.G.C. 1L n d 3 A I O l , n.t. T.G.C. ri n d 4 1101. a t . T.G.C. ri ri d 5 698 0.798 1.0448 1.4865 0,8835 T.G.C. rt d 7 >roi.art. V.G.C. T l n d 8 3101. wi. 1.. G.C. rt 12 d 9 JIol. rvt. T.G.C. ri n d 10 IIOl. wt. V.G.C. rt n d 11 Mol. ai. V.G.C. 1% d 12 \r01.nt. V.G.C. rt 71 d 13 Mol. wt. 1'. G .C. rl n d 14 Ll01.nt. T.G.C. IL n d 1.5 16 x-z 352 0.818 1.0440 1.4748 0.8613 Fig. 7 x n-d-XI u Martin z 464 0,842 1.0419 1.4971 0.9105 Fig. 7 z n-d-XI y Martin z x-2 2-y 2-2 FIE. 7 2-2/ z-z 442 0 892 1 ,0332 1.5352 0.9600 Fig. 7 x n-d-XI y Martin z 2-Y 365 0,915 1.0501 1.5291 0,9580 Fig. 7 s n-d-11 y Martin z 427 0 927 1.0631 1 5550 0 9837 Fig. 7 x n-d-ll y Martin z 369 0.93fj1 ,066.2 Fig, 7 n-d-115 hIartin x-z x-2 x-y 2-2 z-y 2-2 x y z x-y 0.9783 2'-2 373 0.936 1 0039 1.5537 0.9797 Fig. 7 x n-d-Ma y Martin z 373 0,943 1.0654 1.5649 0,9990 Fig. 7 x n-d-MQ y LIortin z 361 0.970 1.0735 Fig. 7 x n-d-Ma y Martin z x-y x-z s-y x-z 1,5804 z-y 1.0138 5-2 Fig. 7 x n-d-Ma u Nartin z 2-y 2-2 Fig. 7 z n-d-lIe y Martin z x-y l f o l . wt. V.G.C. Fig. 7 x n-d-3Ia y n d 310 0 997 1.0734 1.5872 1.0236 x-z Martin z x-y s-2 0 70 69 70 + I 0 3 3 3 33 33 35 64 64 0 0 0 0 3 3 3 34 34 34 0 0 0 0 4 45 43 45 0 0 50 0 + I 4 64 0 0 63 63 63 0 0 51 - 2 3 3 1 1 49 47 47 + 7 49 50 50 - 1 + 2 - 1 20 19 16 1 4 38 39 42 - 1 42 42 42 0 0 25 24 25 1 0 34 35 30 i - 4 41 41 45 0 - 4 45 43 45 + ? 0 3<5 33 - 5 5 - 5 5 ++ + 25.5 22 20 3.5 5.5 ++ 36 36 33 0 3 - 4 - 1 28.5 22 26 6.5 2 5 + ++ 39 42 35 - 3 4 26 16 26 +10 0 + 37 39 33 - 2 4 35.5 42 41 - 6.5 5.5 - 29 22 33 + 7 - 4 32 - 3 0 46 - 9 1 23.5 12 19 +ll 5 4.5 + 31.5 34 35 - 2.5 - 3.5 40 55 46 - 9 0 30.5 15 30 +l.5 5 0.5 + 23.5 30 24 - 6.5 0.5 37 21 31 +16 + 6 25 -10 - 8 - 46 52 +- 44 6 2 49 59 46 -10 + 3 31 11 31 +20 0 NO. of Data 64 118 (3.4 Group 0- 9 9 10-19 9 Standard Deviation % CA 5% C Y 70 cp 1 1 1 1 2 2 1 4 1 4 1 6 - 6 - 3 3" 3.5 - 17 27 20 30 23 -10 - 3 Mol. wt. 314 Fig. 7 x 50 15 35 1 010 V.G.C. 6 31 n-d-lIa 1~ 1 0825 hlariin z 23 25 rl n 1.5997 z-y -13 + 9 A 4 1.0343 d 2-2 - 2 - 8 +lo a Beyond range f o r which n-d-11 method is recommended: also beyond range for which T.G.C. w a s derived. :; 5% 6 39 43 35 - 4 54 Table YIII. Standard Deviation of Carbon-Type ,inalj sis by Figure 7 for Per Cent 4romatic Groups ;3; + 8 43 Q 36 42 39 - 7 - 4 + +4 Table VIIS Standard Deviation of Carbon-Type Analysis by Figure 7 for V.G.C. Groups NO. V.G.C. of Standard Deviation Group Data % Ca %Cv %CP Less than 0 819 36 0 7 1 2 1 0 0 820-0 849 75 0 8 1 9 1 4 0 850-0 899 103 1 3 2 3 2 0 0 900-0 949Q 25 2 8 4 2 3 5 0 950-0 999a 16 1 8 3 5 3 4 Above 1 OOa 3 1 2 6 1 7 3 258 Maximum V.G.C. shown in Hill and Coats (12) is 0.935; above 0.935 V.G.C. is extrapolated function. 29 5 ; ; 31 - 1 Comparison of Figure 7 and n-d-3%Analysis [Data of Hill and Ferris (13) ] Characteriaation % % % V.G.C. Factor Method CA C N C P Pennsylvania 0.808 12.4 Fig. 7 6 . 5 24.5 69 n-d-11 6 . 5 24.5 69 1Iidcontinent 0.839 12.0 Fig. 7 1 2 . 5 29 58.5 n-d-;\I 12.5 2 8 . 5 59.0 Gulf Coast No, 1 0.863 11.7 Fig. 7 1 6 . 5 34 49.5 n-d-M 15 3 4 . 5 50.5 Gulf Coast No. 2 0,883 11.5 Fig. 7 17.5 42.5 40.0 n-d-M 16 42.5 41.5 Gulf Coast No. 3 0.910 11.2 Fig. 7 2 4 . 5 44.5 31 n-d-M 22 44 34 Range (maximum - minimum) 1 8 . 0 20.0 38 51 0 1 z n-d-RI y Martin z % CP 28 28 28 0 0 1 x-y Fig.7 z n-d-11 y Martin z s-Y 272 0 982 1.0724 1.5737 1.0026 - 5-2 39s 0 883 1.0491 1.5210 0.9439 1, 5 5 3 6 2 3 2 2-y Mol. nt. 278 0.983 V.G.C. 1.0700 rt n 1.5725 d 1 ,0050 Ti 17 2-y Fig. 7 x n-d-;\.I y Martin z 3101. nt. 347 1 .G.C. 0.846 1.0407 I1 1,4847 d 0.8879 1101.xi. Fig. 7 x n-d-11 y lIartin z 419 0.815 1.0441 1,4793 0.8704 rt G Composition of Oil % CA % CY Propel.iy of Oil Table VI. Table I S s h o w data for aromatic concentrates obtained by silica gel from five rubber processing oils. These data indicate that Figure 7 gives good values on the per cent aromatic carbons in such aromatic concentrates, but poor values for the yo (23, and % C P . By subtracting 0.015 from the viscosity-gravity constant value in each case, much better agreement is obtained, as shown in the table. I n general, any method of this sort must be used with caution for fractions which are too homogenous as to type and, therefore, may not fit correlations based on averages. The effect of sulfur on a correlation of this sort is important if i t is to be used with high-sulfur oils. When d a h were used for approximately 60 oils on which sulfur data were available, the following correction factors were derived for % Cy and % Cp: Sulfur correction for 5 2 C.,- = 41 C p = Sulfur correction for ~ ~ -Kt.?< 0.288 s +Wt.C7, 0.216 s (7) ( 8) The corrected values for % C.4 can be obtained from the equation 70 C a = 100 - i% C,. + % CP) (9) Table S s h o w the effectiveness of the sulfur correction in ANALYTICAL CHEMISTRY 1934 reducing the deviations between values read from Figure 7 and those obtained using more detailed methods of analysis. The tabulation only includes those samples containing more than 0.5% sulfur. I n the region above 0.8% sulfur, the agreement between the viscosity-gravity constant-refractivity intercept method and other methods may be substantially improved by applying the sulfur correction. I n addition to viscosity-gravity constant another relation between viscosity and gravity, the characterization factor, hap been widely used (46, 4 7 ) . A graph similar to Figure 7, based on characterization factor, could undoubtedly be developed. The authors have preferred to provide tables so that data recorded in terms of characterization factor can be converted to viscositygravity constant and studied with the aid of Figure i . % AROMATIC RING CARBONS A 1.0300 /e ' -1.0600 1.0400 NAPHTHENE RING CARBONS % PARAFFIN CHAIN CARBONS Figure 8. Hydrogenation and solvent extraction data in relation to carbontype composition Lines 1, 2, and 3 explained i n text Table IX. Aromatic Fractions Concentrated by Silica Gel [From rubber processing oil (48) I Aromatic Fraction of Si, relatively naphthenic Property of Oil Mol. \vt. 366 V.G.C. 0.900 ti 1,0530 1,5269 n d 0,9465 Ss, relatively Mol. u t . 435 naphthenic V.G.C. 0.886 112 1 0536 n 1 5330 d 0 9588 Sr. relatively Mol. a t . 350 paraffinic Y.G.C. 0.898 T, 1 0547 n 1 5223 d 0 9355 91. relatively Mol. wt. 418 paraffinic V.G.C. 0 879 ti 1,0530 n 1.5180 d 0.9300 As, relatively Mol. wt. 409 aromatic V.G.C. 0.957 Ti 1.0706 n 1.5767 d 1.0124 a 0.015 subtracted from V.G.C. . ~ Composition of Oil c'r C.V % CP 30 38 32 44 28 45 4-8 - 7 - 1 7-4 B Tr 42 % CA Figure 7 Correcteda Martin Figure 7 Correcteda Martin X d z 2-2 - 1 u-2 - 3 22 22 24 - 2 2: u z +-z 21-2 Figure 7 Correcteda Martin 5 Y z X-Z ll-2 Figure 7 Corrected" Martin Figure 7 Corrected" Martin 26 24 27 X u z - 1 27 25 27 0 - 2 23 21 24 2-2 - 1 21-2 - 3 x-2 42 42 42 0 u-2 0 2 21 2 50 44 28 82 +.I - 4 - 2 C 6 33.5 29 27 6 .5 A 2.0 + 30.3 27 12 + 8 5 + 5 25 22 21 + 4 + I - 39 5 46 46 6 5 0 16 5 '0 54 - 7 5 - 2 33 36 37 - 4 - 1 Table XI gives enough data so that viscositygravity constant can be obtained from characterization factor and viscosity by interpolation. An example of the application of Figure 7 is given in Figure 8. Line 1 on this graph represents data from the paper by M. R. Lipkin, C. C. Martin, and R. C. Worthing (33)shon-ing steps in the hydrogenation of a viscous oil fraction. The gradual decrease in aromatic carbon content and increase in naphthenic carbon content is clearly shown. There is also an indication that the final hydrogenation ti eatments, which were continued in an effort to eliminate the last trace of ultraviolet absorption, may have resulted in some ring opening, because there is a decrease of between 2 and 4% in the naphthene ring carbons and a corresponding increase in the paraffin chain carbons during these last hydrogenations Other data for hydrogenated samples confirm the constancy of % C p as aromatic rings are converted to naphthenic rings. Line 2 on this graph shows the change in an oil on repeated solvent extraction (34). I n this case there is a progressive decrease in V O L U M E 28, NO. 12, D E C E M B E R 1 9 5 6 1935 Table X. Correction of Carbon-Type Composition Data for Effect of Sulfur Deviation from Best Analvses V.G.C. V S . rt V.G.C. us. r , (corr. for % SI“ Lit. Sample % S % C a % C \ % CP % C a %C.v % C P Ref. NO. V.G.C. rl 16 0.997 1.0754 0.57 + 3 . 0 0 - 3 . 0 $2.4 - 2 . 0 -0.4 (42) 18 0,943 1,0607 0.73 + 3 . 6 - 1 . 3 - 2 . 3 + 2 . 7 - 3 . 8 + l , l (42) 20 1,0699 0.72 - 2 . 1 + 3 . 8 - 1 . 7 - 2 . 9 + 1 . 4 + 1 . 5 (42) 0,970 21 0.965 1.0693 0.67 - 2 . 3 + 0 . 9 t 1 . 4 - 3 . 1 - 1 . 4 1-4.5 (42) 1.0739 0.75 0 - 0 9 + 0 . 9 - 0 . 9 - 3 . 5 +4 4 23 0.982 (42) 24 0.973 1.0746 0.73 $ 2 . 5 - 3 . 6 $ 1 . 1 $ 1 . 6 - 6 . 1 + 4 . 5 (42) 25 0.980 1.0700 0 . 5 6 + 1 . 2 S i 6 - 8 . 8 +0.5 $ 5 . 7 - 6 . 2 (48) 26 1.011 1.0788 0 . 5 7 + 1 . 8 $ 5 . 3 - 7 . 1 $1.2 + 3 . 3 - 4 . 5 (42) 27 0.956 1.0642 0 . 7 7 - 3 . 2 $ 3 . 7 - 0 . 5 - 4 . 1 $1.0 + 3 . 1 (42) 11 0.936 1.0639 0.56 1 4 . 3 - 1 . 0 - 3 . 3 + 3 . 6 - 2 . 9 - 0 . 7 (42) 9 1.0631 0.62 + 3 . 0 + 2 . 5 - 5 . 5 $ 2 . 4 $ 0 . 3 - 2 . 7 0.927 (42) 14 1.0724 0.55 0 -0 5 - 0 . 5 - 0 . 7 - 1 . 3 + 2 0 0.982 (42) 1.0665 0.73 $ 3 . 8 0 -3.8 $2.9 -2.5 -0.: 10 0,936 (48) 0 70 33 0.831 1.0434 -1.5 +3 0 -1 5 -2.3 $0.6 + l . t (34) 34 1.0670 0.79 i 3 . 0 - 3 0 0 + I 5 -5.2 0.909 +3.7 (42) 35 0 - 1 . 0 +O.l -2.7 1,0560 0.77 7 1 . 0 0.929 S2.6 (49) Std. dev. ( 0 . 5 to 0 . 8 % E) 2 0 3 1 3 R 2 4 3.2 3 2 Dev. of av. +1.2 +1.1 - 2 3 f 0 . 3 -1.2 + 0 . 9 i o 9 1 4 G - 5 5 - 0 . 1 +1 8 - 1 7 19 1 0629 0.82 0 946 22 1 0709 1.08 -0 6 $ 1 5 - 2 . 1 - 0 . 6 -2 3 + 2 9 0 971 1 0816 0 . 9 2 - 1 . 0 - 3 0 t 4 0 -2 1 - 6 1 + 8 2 1011 36 1.01 0 892 1 0532 0 + 4 3 - 4 3 - 1 . 2 +O 8 + o 4 1 . 0 3 -1 0 +4 5 - 3 . 5 -2 2 +o 9 + I 3 1 0735 13 0 970 1 0543 28 0 892 2.05 + 1 . 0 +8 0 - 9 . 0 - 1 . 4 +O 9 $ 0 5 1 0498 29 0 870 1 . 6 7 $0 5 + 4 5 - 5 . 0 - 1 . 2 -1 5 +2 7 1 0468 30 0 854 1.35 + 0 . 5 +:3 0 - 3 5 - 1 . 1 -1 7 +2 31 1.12 - 1 0 + 3 5 - 2 5 - 2 3 - 0 4 +2 i 1 0452 0 844 32 1 0441 0.89 -1 0 + 3 0 - 2 0 - 2 0 -0 1 +2 1 0 837 37 1 0501 +1 5 +4 0 - 5 5 + 0 5 $1 0 -1 5 0 838 0.86 1 0505 38 0 866 1.50 + l . 5 + 5 n - 6 5 - 0 2 - 0 2 +o 4 39 0 871 1.62 + 4 0 +2 5 1 0544 +2.1 -3 1 + I 0 - n .i $0 1 $0 4 1.70 + l . 5 +8 0 - ! 40 0 892 1 0515 1.50 + 0 . 5 +9 5 - 1 0 0 - i i + 4 3 - 3 1 41 0 8.59 1 0522 42 1.51 + 1 . 0 + 8 0 - 9 0 - 0 7 +2 8 - 2 1 1 0.534 0 863 1.73 + l . 5 +9 0 - 1 0 5 - 0 5 + 3 0 - 2 5 43 1 0550 0 871 Std. dev. ( 0 . 8 to 2.0% S ) 1.4 5 -5 6.3 1.4 2 4 2 8 D e i . of av. +O.B + 4 . G -5 2 - 0 . 9 0 0 +0.9 a Correction applied to % C.Yand % CP; % CA adjusted accordingly -!.e Table XI. Viscosity-Gravity Constant Corresponding to Indicated Viscosity and Characterization Factor VlSCO’.ltJ, 1000 E Characterization Factor 12 5 12 11 5 11 10 5 10 KineSa.\niatir bolt Visrosity-Gravit> Constant 6 45 6 0 795 0 828 0 865 0 908 0 954 1 006 52 1 0 796 0 830 0 869 0 910 0 958 1 013 8 58 9 0 797 0 832 0 872 0 914 0 962 1 019 10 97 8 0 799 0 836 0 879 0 924 0 973 1 0 3 5 20 232 0 801 0 842 0 886 0 937 0 991 1 0 5 3 50 100 463 0 802 0 845 0 890 0 942 1 001 1 067 200 927 0 803 0 846 0 894 0 948 1 011 500 3317 0 804 0 848 0 898 0 953 1 023 1000 4635 0 805 0 849 0 901 0 957 1 0 3 0 2100 r _____ 5 10 20 50 100 12 7 59 3 98 5 234 467 0 796 0 803 0 806 n 809 0 811 0 837 0 884 0 938 0 995 0 844 0 892 0 947 1 005 0 849 0 898 0 954 1 015 o 854 o 904 o 960 1 021 0 856 0 906 0 963 1 0 2 3 1 082 1 073 1 080 1 086 1 087 aromatic carbon conteiit and increase in paraffinic and naphthenic carbon content up to the last stage of extraction. In the last stage of extraction, aromatic carbon content changes little and naphthenic carbon decreases as paraffinic carbon increases. The oil for which data are presented in Line 2 contained 2.05% sulfur before eutraction. The composition of this oil as read from the graph is 26% ‘2.4, 32% CY,and 42% Cp. After applying the sulfur corrections, the composition is 24% C A , 25% C.V, and 51% C P . Van S e s and van Westen give 25% C A ,24% C r , and 51% C p . Line 3 shows the composition data corrected for the effect of sulfur These data confirm both the usefulness of the plot of the raw data to show trends, and the validity of the sulfur correction. ACKNOWLEDGMENT The authors JS ish to acknoa ledge the assistance and helpful discussion of C. C. Martin, J. C. S. Wood, and J. W. Loveland. LTTER.4TURE CITED (1) American Petroleum Institute, Research Project 6, Carnegie Institute of Technology, Pittsburgh, Pa. (2) Ibid., Research Project 42, Pennsylvania State University, University Park, Pa. (3) Ihid.,Research Project 44, Carnegie Institute of Technology, Pittsburgh, Pa. (4) Beverly, J. B., Marschner, R. T., Preprints, Petroleum Division ACS, iipril 1955. ( 5 ) Boelhouwer, C., Watermarl, H. I., J . In& Petroleum 40, 116 (1954). (6) Bondi. A . , “Physical Chemistry of Lubricating Oils,” Reinhold, Sew York, 1951. (7) Charlet, E. bI.,Laniieau, K. P.. Johnson, F. B., ANAL CHEM. 26, 861 (1954). ( 8 ) Clark, R. J.. Hood, .\.. O’Keal, LI. J., Jr., Ibid., 27, 868 (1955). (9) Dudenbostel, B. F., Jr.. Priestley, William, Jr., “Chemistry of Petroleum Hydrocarbons,” Vol. 1, Chap. 12, Brooks, B. T., others, eds.. Reinhold. Nem- York, 1954. (10) Fenske, 11. R.,Carnahan, F. L., Breston, J. N., Caser. A . H., Rescorla. A . R., Ind. E ~ L Q Chem. . 34, 63846 (1942). (11) Hazelwood. I?. S’.,ANAL.CHEM.26, 1072 (1954). (12) Hill, J. B., Coats. H. B., I n d . Eng. Chem. 20, 6 4 1 4 (1928). (13) Hill, J. B., Ferris, S.W., Ibid., 17, 1250 (1925). (14) Kurtr. S. S., Jr., “Chemistry of Petroleum Hydrocarbons,” 5-01. 1, Chap. 11, Brooks, B. T., others, eds., Reinhold, New York, 1954. (15) Kurtr, S. S., Jr., Headington, C. E., IND.ENG. CHEM.,-4x.4~. ED. 9, 21 (1937). (16) Kurtr, S. S.,Jr., Martin, C. C., India Rubber World 126, 495 (1952). ANALYTICAL CHEMJSTRY 1936 Kurta, S. S., Jr., Sankin, d.,“Physical Chemistry of Hydrocarbons,” 1-01. 11, Chap. 1 , Farkas, d.,ed., Academic Press, Sew York, 1953. Kurtz, S. S., Jr., Xard, d.L., J . Franklin I71St. 222, 563 (1936). (33) Ibid., p. 370. (34) Ibid., p. 429. (35) O’Seill, J., “Applied Ibid., 224, 583, 697 (1937). Leendertse, J. J., in “Aspects of the Constitution of 11ineral Oils.” a. 368. Elsevier. Kew York. 1951. Lillard, i.G., Jones, W,C., Jr., Anderson, J. A , , Jr., I n d . Eng. Chern. 44, 2623 (1952). Lipkin, RI. R., Hoffecker, 11‘. A., Martin, C. C.,Ledley, R. E., . h A L . CHEM. 20, 130 (1948). Lipkin, AI. R., Martin, C. C., Worthing, R. C., Third World Petroleum Congress, Section VI, 1951; E. J. Brill, Leiden, Rossini, F. D., Mair, B. J., Streiff, A . J., “Hydrocarbons from Petroleum.“ Reinhold, New York, 1954. (37) Ibid., Chap. 22. (38) Rossini, others, “Tables of Physical and Thermodynamic Properties,” API 4 4 , Carnegie Press, Pittsburgh, Pa. (39) Schiessler, R. W., Rytina, C.H., Weisel, Fischl, F., JIcLaughlin. R. L., Keuhner, H. H., Proc. Am. Petroleum Inst., 26, 111, Holland, 1951. Lipkin, l f . R., Sankin, A, Martin, C.C., ANAL. CHEM.20, 598 (40) Schiessler, R . W., Whitmore, F. C., I n d . Eng. Chem. 47, 1660 (1948). AIair, B. J., Schicktane, 8. T., I n d . Eng. Chem. 28, 1446-51 (1936). Martin, C. C., Sankin, A , , A~NAL. CHEV.25, 206 (1953). llikeska. L. -4., I n d . Eng. Chem. 128, 978-84 (1936). Mills, I. IT.,Hirschler, A. E., Kurte, S.S., Ibid , 3 8 , 442 (1946). hliron, S., ANAL.CHEW27, 1947 (1955). Xes, K. van, “Chemistry of Petroleum Hydrocarbons,” Vol. 1 , Chap. 16, Brooks, B. T., others, eds., Reinhold, New York, 1954. Kea, K. van, Westen, H. A . van, “..lspects of the Constitution of Mineral Oils,“ Elsevier. K e n . York. 1951. (32) Ibid., p. 200. Mass Spectrometry,” pp. 27-46, Report Conference. published by Institute of Petroleum, London, 1954. (36) 254 (1946). (1955). Smith, Edwin E., Engineering Experiment Station, Bull. 152. Ohio State University, Columbus, Ohio, May, 1953. (42) Sun Oil Co., unpublished data. J., in “Aspects of the Constitution of Mineral Oils.” pp. 250, 317, 318, Elsevier, Ken. York, 1951. (44) Waterman, H. J., Brennstof Chemie 36, 169 (1955). (41) (45) Ibid., p . 199. (46) Katson, K. AI., “Science of Petroleum,” Vol. 2, p. 1377, Dunstan, et al., Oxford University Press, London, 1938. (47) Watson, K. lI.,Selson, E. F., I n d . Eng. Chem. 29, 880-7 (1933). (48) Weinstock, K. V., Storey, E. B., Sweely, J. S., Ibid., 45, 1036 (1953). R E C E I T - Efor D review N a y 8 , 1956. Accepted September 14, 1S5B. Composition of Lubricating Oil Use of Newer Separation and Spectroscopic Methods F. W. MELPOLDER, R. A. BROWN, T. A. WASHALL, WILLIAM DOHERTY, and C . E. HEADINGTON The Atlantic Refining Co., Philadelphia, Pa. A study was made of the composition and physical properties of fractions separated from a solvent-refined oil heart cut. The oil charge was fractionated by means of 20-stage molecular distillation, silica gel chromatography, and liquid thermal diffusion to yield saturated fractions ranging from predominantly isoparaffins to multiring cycloparaffins. Hydrocarbon-type analyses were determined by mass and ultraviolet spectrometr? and several molecular structures were postulated from mass and infrared data. One fraction which exhibited extreme ph) sical properties w-as found to contain from four to ten rings per molecule. THE present-day trend in the refining of petroleum products and the development of new and more ponerful analytical techniques have promoted a growing interest in the composition of heavy petroleum product-. Detailed information concerning the molecular structure of hydracarbon types is needed because of the many factors related to performance and stabilitj- characteristics of lubricants. A long-range approach to the problem has been extensivelj investigated during the past t v o decades by API Project 6 (14). This work has been directed toward a separation of oil according to hydrocarbon types, and measurements of numerous physical properties of the fractions. Additional data for a group of “homogeneous” fractions were recently reported for a cooperative spectroscopic study ( 9 ) . Further progress in this field has been dependent on the development of more efficient separation techniques, extension of spectroscopic methods of analysis, and the synthesis of representative hydrocarbons as standards to guide the interpretation of spectra. Despite the lack of such standards, O’Keal (IS)and Lumpkin and Johnson (8) were able t o postulate the structure of many aromatic h j drocarbons and sulfur compounds in gas oil. 3Ielpolder and coworkers were also able to make a mass spectrometric analysis of thermal diffusion fractions separated from a light lubricating oil ( I O ) . I n viex of the significant new developments which have been made in the past fen- years in both separations and spectroscopic methods, the authors have undertaken a thorough study of the composition of a heart cut lubricating oil fraction. Working on the saturated portion only, many fractions were separated from the oil, in which specific hydrocarbon types of a narrow molecular Jveight range were concentrated. Physical properties and spectrometric anal>-sesof the fractions were determined, and an effort was made to determine molecular structures from the interpretation of niass ppectrometric data. LUBRICATIYG OIL STOCK The oil charge used in this work \vas a solvent-refined lubricating oil in the SAE-20-30 grade viscosity range. The oil was obtained from a mixed crude source as a pipe still distillate, solventextracted n-ith nitrobenzene, dewaxed n-ith methyl ethyl ketone, and filtered through clay. Inspections of the oil are shown in Table I SEPARATION PROCEDURES T h e lubricating oil charge stock n-as separated by means of distillation, adsorption, and liquid thermal diffusion to yield 43 fractions containing concentrates of paraffin and cycloparaffin hydrocarbons. The procedures are summarized in the block diagram shown in Figure 1. Starting with 5.5 liters of oil, having a molecular weight range of from CIB to Cso, a series of distillations was made in a 20-stage molecular still (11) a t 1-micron pressure. A total of 14 separate equilibrium-type distillation runs and one batch distillation run was required t o process the oil. The boiling range of the resulting