1419
V O L U M E 22, NO. 11, N O V E M B E R 1 9 5 0
+
equation, yn+? = na y,, the solution of which is yn. = n/3 nt/2 n3/6. To a good approximation, therefore,
+
mo
- nk'" -
+ (n/3 - n2/2 + n3/6)k'n-2 (k'
+ Ip+'
(2)
The value of k' is easily determined from Equation 2 by surcessive approximations; a first approximation is k = tt/(mrr./ m)- 1. Both ~ 1 1 ,and mIu are'determiqed a t the end.of the nitu,or, i f very multistage extraction; me is best taken as t)ittu precise values are desired, the small amounts of solute present i n the lower layers of both funnels may be calculated, using an approximate value of k', and inqluded in ma. The value of k is obtained from the relation'k = k / r .
+
The preceding discussion has explicitly treated the case where the partition coefficient is very large. If the partition coefficient is very small, exactly analogous considerations are involved, except that in this case the lower rather than the upper layer of each funnel should he analyzed MEASUREMENTS WITH o - P n ~ NYLPHE'YOL
The interchange evtraetion technique was tested by measuring the partition coefficients of o-phenylphenol for distribution in the systems, cyclohexane-water and cyclohesane-0.5 M phosphate buffer of pH 6.65. The iiiitial concentration of solute in the cyclohexane was 0.5 mg. per ml. The ratio of cyclohexane to aqueous phase was 1 to 3. The extractions were carried out in centrifuge separatory funnels equipped with capillary outlets; the funnels were centrifuged to clarify layers a t the end of each
stage. After nine stages, the concentration of o-pheriylphenol in each cyclohexane layer was determined by measuring its optical density a t 280 mp with the Cary recording spectrophotometer. Partition coefficients were then calculated by means of Equation 2. The results of these measurements are presehted in Table I. The partition coefficient in the cylohesane-buffer system is much higher than the partition ratio in the cyclohexane-water system because of t,he salting-out effect of the buffer. Both ratios are about 10 to 20y0 lower than those previously obtained by the less precise, successive extraction technique ( 5 ) . From the partition coefficient at. pH 6.65 and from the previously determined partition ratio a t pH 12.54 (S), the value of pKa was calculated (Table I). This value agrees with that obtained earlier ( 5 ) within 0.1 pK unit. ACKNOWLEDGMENT
The authors are indebted to George Goldbaeh for technical assistance. LITERATURE CITED
(1) Craig. L. C., J . Bid. Chern., 150,33 (1943). (2) Craig, L. C.. and Post, O., ANAL.CHEM., 21, 500 (1949). (3) Golumbic, C., Orchin, M., and Weller, S., J . A m . Chon. Soc.. 71, 2624 (1949).
Golumbic, C . , Woolfolk, E. O., Friedel, R. A:. and Orchin, M., Ibid.,72, 1939 (1950). (5) Woolfolk, E. O., Golumbic, C., Friedel, R. 4.. Orchin, M.,and Storch, H. H., Bur. Mines, Bitll. 487, in press. (4)
RECEIVED
&fay 12, 1950.
Determination of Aromatics and Olefins in Wide Boiling Petroleum Fractions A . E. SPAKOWSKI, ALBERT EVANS A N D R. R . HIBBARD National Advisory Committee f o r Aeronautics, Lewis Flight Propulsion Laboratory, Cleveland, Ohio A method is described for the analysis of wide boiling petroleum fractions with end points below 600" F. Chromatography is used to split the sample into four fractions: nonaromatic, intermediate, pure aromatic, and a pure aromatic "wash." Th; per cent aromatics in the intermediate fraction is determined by specific dispersion and the total aromatics in the sample is the sum of those found in the intermediate and final two fractions. Total olefins plus aromatics are determined by sulfonation and olefins by difference. Accuracies of 1% are attained with analysis times of less than 8 hours.
I
r\; ORDER to understand more fully the effect of fuel variables
in airrraft propulsion systems, it is often necessary to determine the concentration of olefin and aromatic hydrocarbons in petroleum products. The most recent turbojet fuel specifications (AN-F-58a) call for a wide boiling fuel with 5- to 7-pound Reid vapor pressure and a 600" F. end point. Aromatics up to 25% and olefins equivalent to a 30 bromine number are permitted. A review of the literature showed the following general methods which might be applied 'to fuels of the AN-F-58a type. The first of these is the A.S.T.M. procedure in which olefins and aromatics are absorbed by a mixture of sulfuric acid and phosphorus pentoxide, olefins are calculated from bromine number, and aromatics. are obtained by difference ( 2 ) . The concentrations of paraffins and cycloparaffins can also be estimated from the older A.S.T.M. method, using refractivity intercept and the density of the raffinate from the sulfonation ( 1 ) . Occasionally inaccurate results will be obtained by the use of this A.S.T.M. procedure. Its greatest weakness appears to be in the calculation of the per cent olefins from bromine number and the molecular weight where the average molecular weight of the plefins
and not of the total sample is required. The molecular weight is estimated from the 50% boiling point of the total sample in the A.S.T.M. procedure and substantial errors often result in the calculation of the per cent olefins. Errors in per cent olefins are also reflected as.equal errors of opposite sign in the determination of the per cent aromatics by A.S.T.M. procedure. Although the approved method for the analysis of AN-F-58a fuels is the A.S.T.M. procedure, these fuels may have bromine numbers above the value of 20, which is the maximum permitted in the scope of the method. The A.S.T.M. silica gel method (3)applies to fuels containing less than 1% olefins and therdore does not cover the AN-F-58a specifications. Another general method relies on chromatography alone to split samples into paraffin and cycloparaffin, olefin, and aromatic fractions. Mair (14) and Dinneen ( 7 ) have proposed this type of analytical separation. Both have shown its effectiveness on relatively simple blends and Dinneen has also given data on shale oil naphthas (7). However, it has been the authors' experience that, while good separations are obtained on multicomponent
1420
ANALYTICAL CHEMISTRY
blends of pure hydrocarbons, an olefin plateau is not obtained in the chromatographic fractionation of the complex type fuels Mhen small amounts (10% or less) of olefins are present. Thus this method is not universally applicable to fuels meeting the AN-F-58a specifications. Grosse and Wackher developed a method for the determination of aromatic hydrocarbons in fuels based on the measurement of the specific dispersion of the sample (8). The relative uncertainty of the method on turbojet-type fuels lies in the inaccuracy of the olefin determination and in not knowing the specific dispersion of the aromatics pr'esent in the fuel. Conrad has developed a rapid method for the determination of aromatic compounds in fuels of the gasoline and kerosene range, based on chromatography and ultraviolet stimulated fluorescence (6). Limited experience has shown this method to be less reliable as the complexity of the aromatics and the olefinic content increases, as is the case with AN-F-Sa fuels. Provision is not made to obtain fractions of saturates or aromatics on which further studies can be made. While the procedure of Kurtz et al. (10) is recommended for the analysis of samples having end points no higher than 437' F., it is probable that the method could be modified to include the 600' F. end point fuels. The method requires a fractional distillation :tnd the determination in each fraction of the volume per cent ahsorbed by sulfuric acid-phosphorus pentoxide, bromine number, and specific dispersion. The determination of the per cent olefins by the nitrogen tetroxide method may also be required on some fractions. Apart from the uncertainty as to whether the method could be extended to fuels of higher end point, the principal objection to this procedure is the necessity of running a fractional distillation which requires about 24 hours and the relatively large number of determinations which must be made on the many distillation fractions. A method which has been in satisfactory use a t this laboratory for over a year uses chromatography to separate a nonaromatic fraction, an intermediate fraction, and a pure aromatic fraction. To make the separation more plainly seen, Parasheen is used ( 6 ) . The per cent aromatics in the intermediate fraction is then determined by the specific dispersion method of Grosse and Rackher (8). Although high precision cannot be obtained in the analysis of wide boiling fuels by the above dispersion method, the inaccuracies are confined to only a small part (about 10%) of the whole fuel and the error in the analysis of the total fuel is only about 10% of what i t might have been had the aromatics in the total sample been determined by the same method. The per cent olefins are obtained by the difference between the total olefins and aromatics from the A.S.T.M. sulfonation (2)and the aromatics as determined by the proposed procedure. An aromatic fraction is also obtained which can be further characterized, if desired, and paraffins and cycloparaffins can be determined on the sulfonation raffinate ( 1 ) . This method for determining aromatics and olefins in jet-type fuels is described herein and supporting data on multicomponent blends are given. Some results on typical turbojet-type fuels are also included. APPARATUS A N D MATERIALS
A borosilicate glass absorption column (S), without the stopcock and ground-glass joint a t the bottom. Activated silica gel, 28 to 200 mesh. Through-200-mesh gel may also be used with similar results, but with longer analysis times (Davison Chemical Corporation, Baltimore, Md.). Graduated mixing cylinders in loo-, 25-, and 10-ml. sizes. Stoddard solvent bottles. Refractometer capable of measuring RF-RC values to 0.("302. Mercury arc ultraviolet lamp, such as Model 16200, manufactured by the Hanovia Chemical and Mfg. Co., Newark, N. J. Isopropyl alcohol, C.P. grade. If the sample is suspected to contain aromatics of high molecular weight, Carbitol used as the eluent will desorb these aromatics more completely. Acidified calcium chloride solution, 450 grams per liter plus 20 ml. of concentrated hydrochloric acid.
Parasheen (Paraflow Sales Department, Chemical Produc.t.s, Enjay Company, 26 Broadway, S e w York 4, N. Y.). PROCEDURE
The column, thoroughly cleaned and dried, was filled H ith silica gel and packed following the A.S.T.M. method ( 3 ) . A 100-ml. sample containing 2 drops of Parasheen was introduced into the reservoir and allowed to percolate into the gel. To ensure the complete transfer of the sample, the flask was rinsed with 2 or 3 ml. of eluent (isopropyl alcohol) and the rinse was added to the reservoir immediately after all the first portion of the sample had been adsorbed on the silica gel. When the rinse was completely adsorbed, the reservoir was filled with eluent and an air pressure of 3 to 10 pounds per square inch was applied to maintain the desired rate of descent (7 to 13 mm. per minute). Additional eluent was added when necessary, so as to keep the gel covered. In samples &here the aromatic content was under 1070, 90 nil. of sample were taken and 10 ml. of pure isopropyl benzene were added. When the sample was about to issue from the bottom of t h e column, a glass-stoppered 100-ml. graduated mixing cylinder was placed under the column to receive the nonaromatic fraction. These receivers were packed in ice to reduce evaporation losses of the volatile components. As the column was not equipped with a ground-glass joint to accommodate this type of receiver, a tight seal was maintained by allowing the weight of the column to rest directly on the receiver with a little glass wool serving as a gasket. This connection was not air-tight and therefore prcvided the necessary vent while keeping evaporation losses small. As the fractionation progressed, the leading edge of the aromatic portion was located in the column by shining ultraviolet light thereon in a darkened room (or in a suitable dark box). The Parasheen tracer allowed this boundary to be followed easily. Although the boundary was found to be sharp for sample having low olefin content (5% or less), i t became ICRS distinct as olefin concentration increased.
R P IN .F:
Figure 1. Specific Dispersion of Alkylbenzenes as a Function of A.S.T.M. 50% Boiling Point
\Vhen the boundary approached to within 5 cm. of the column tip, the cylinder receiving the nonaromatic fraction was replaced by a 10-ml. graduate. A 10-ml. intermediate fraction was then taken, which contained sample on either side of the break point. The receivers were again changed and the major aromatic fraction up to 2 cm. of the alcohol-aromatic boundary, which WLS always visible, was removed. A final 10-ml. fraction, including the remaining aromatics mixed with alcohol, was collected ir it Stoddard solvent bottle and the alcohol was removed by washing with acidified calcium chloride Solution until the volume of the aromatics remained constant (usually three washings). The washing was accomplished by adding calcium chloride solution to the solvent bottle so that the level of the liquid was in the graduated neck, shaking the mixture into an emulsified condition, and centrifuging until the mixture separated into two layers. The spent calcium chloride solution was removed by inverting the bottle, carefully loosening the stopper, and permitting the water solution to drain, leaving the aromatics in the bottle. Recovery of the total sample was found to be 98% or better. The losses were assumed t o be low boiling paraffins.
1421
V O L U M E 2 2 , N O . 11, N O V E M B E R 1 9 5 0
l w determined from the major aromatic fraction and wash 11.v
Table I.
Composition of Test Blends
ParafinCycloparaffin, Vol. % 75 70 65 55 70 75 80 100 90 80
Blend
4 5 6 7
a 9
10
Olefin, Vol. % 0 5 10 20 20 20 20 0 5 10
the ultraviolet spectrophotometric method of Cleaves and Carver ( 5 ) and the average number of ring3 per molecule by the method of Lipkin and Martin (12).
Arornstic, Vol. % 25
RESULTS
25
25 25 10
5 0 0 5 10
Table 11. Composition of Stocks PsraffinCycloparaffin Stock n-Heptane n-Decane Iao-octane Cyclohexane hlethylcyclohexane. Decslin
12.5
Arornstic Vol. % Stock Benzene 15.0 Toluene 15.0 Xylenes 20.0 Methylnsph25.0 thslene ( 1 , 3) Tetrslin 25.0
. -2 5 . 0
__
Vol.
%
12.5 12.5 25.0 12.5
100.0
Olefin Stock 1-Octene Diisobutylene Cyclohexene
vu1
76
33.3 33.3 33.4
___ 100.0
100.0
To calculate the volume per cent aromatics present in the sample it is only necessary to determine the aromatics in the intermediate fraction, because the subsequent cuts are pure aromatics. The method employed is based on the measurement of the specific dispersion of the cut on an ordinary Abbe refractometer. Laboratories not equipped to measure specific dispersion may fmd it possible to calculate this value from the 50% boiling point, density, and refractive index using the method of Lipkin and hfartin (11). The weight per cent of aromatics in the cut is given tiy the equation of Grosse and Wackher (8): Weight yo aromatics
=
6,,,
- 0.16 X bromine S o . -99 ~,,,
