aromatic peak prediction was also possible but with increased error. Overall scatter in the calibration data n-as about 0.9 vol. yo aromatics, calculated as yo of naphtha boiling about 150-390" F. Standard deviation of replicate chromatographic hydrocarbon-type analyses was 1 0 . 2 vol. Yo naphthenes and rt0.1 vol. % aromjtics. Routine analyses over a period of several months averaged within 1.3 vol. % naphthenes and 0.7 vol. yo aromatics of mass spectrometer analyses (Table 11). Applications. An impressive amount of information was derived from a single chromatographic analysis of crude oil (Table 111). The chromatogram yields boilingrangr distribution dataneeded for refinery crude tower control and composition analysis adequate for stream value prediction. Hydrocarbon-type distribution analyses are directly applicable to units processing low boiling stocks and are also useful for prediction of quality of higher boiling streams. The entire crude oil analysis, including computations, can be completed in less than 2 hours by technician-level personnel. The chromatographic technique can be used routinely to guide crude oil purchases, for quality control of crude
oil receipts a t the refinery gate, to check crude oil blending facilities and for feed forward control on refinery units. The method has also been advantageously used for steam analysis (Figure 7 ) where speed and low cost are prerequisites for quality control. ACKNOWLEDGMENT
The authors gratefully acknowledge the contributions of Harrison M.Stine for initiating this work, Jean K. Evans for technical assistance, Richard Thutt' for the quantitative mass spectrometer analyses, and C. B. McKinney of the Bureau of Mines for supplying samples of Bradford, Conroe, and Hastings crude oils. LITERATURE CITED
( 1 ) Barras, R. C., Boyle, J. F., 27th
Midyear Meeting of the American Petroleum Institute's Division of Refining, San Francisco, Calif., May 16, 1962. ( 2 ) Bassette, R., Whitnah, C. H., J. Dairy Science, 44, S o . 6, 1164 (June 1961). (3) Brown, R. A,, ANAL. CHEM.23, 430 (1951). (4) Desty, D. H., Goldup, A,, Swanton, W. T., ISA Proceedings, 1961 Internatjonal Gas Chromatography Symposium of Instrument Society, of America, p. 83, Michigan State University, June 1961.
(5) Dorsey, J. A., Hunt, R. H., O'Neal, M. J., ANAL.CHEM.35,511 (1963). (6) Ebert, A. A., Jr., Ibid., 33, 1865 i1961). ( 7 ) Eggertsen, F. T., Groennings, S., Holst, J. J., Ibid., 32, 904 (1960). (8) Gohlke, R. S.,Ibid., 31, 535 (1959). (9) Holliman, R. C., Smith, H. M., McKinney, C. M., SDonsler. C. R.. Cnited States Department of the In: terior, Bureau of Mines Technical Paper 722 (1950). 10) Levy, E. J., Doyle, R. R., Brown, R. A,, Melpolder, F. R., ANAL.CHEM. 3 3 . 6- 9- -8 i\ -l -R- -6, .l ) li, Levy, E. J., Miller, E. D., Beggs, W. S.,Ibid., 35, 946 (1963). 12) Lindeman, L. P., Annis, J. L., Ibid., 32. 1742 11960). 13) 'Martin, R. 'L., Winters, J. C., Ibid., 31, 1954 (1959). 14) Messner, A. E., Rosie, D. M., Argabright, P. A., Ibid., 31, 230 (1959). (15) Polgar, A. G., Holst, J. J., Groennings, S.,Ibid., 34, 1226 (1962). (16) Rvsselberee. J. van. Znd. Chim. Belge24, 1023 (1959). ' (17) Webb, T. P., 7th Detroit Anachem I
Conference, Wayne State Cniversity, Detroit, Mich., 1959. (18) Wiley, W. C., Science 124, 817 (1956). (19) Wiley, W. C., McLaren, I. H., Rev. Sci. Inst. 26, 1150 (1956).
RECEIVEDfor review August 21, 1963. Resubmitted April 27, 1964. Accepted April 27, 1964. Sixth World Petroleum Congress, Frankfurt, Germany, June 1926, 1963.
Chromatographic Separation of Polycyclic Aromatic Hydrocarbons on Columns Containing s- I rinitrobenzene 'C..
0 .
I
RUSSELL TYE and ZEB BELL' Kettering Laboratory, Department o f Preventive Medicine and industrial Health, College o f Medicine, University of Cincinnati, Cincinnati, Ohio
b The presence of Lewis acids has a marked effect upon the distribution of Lewis bases between a polar and a nonpolar liquid. This principle has been investigated with respect to polynitro aromatic compounds in each of three glycols, as stationary phases, with isooctane as the mobile phase, for use in liquid-liquid partition chromatography. Mixtures of polycyclic aromatic hydrocarbons which are refractory to other separation methods may be separated by using 0.258M s-trinitrobenzene in a polyethylene glycol (Carbowax-400), supported on 100-200 mesh Columpak. The chromatographic retention volumes agree with volumes calculated from distribution coefficients. This aids in the identification of compounds and may permit making useful advance estimates of the adequacy of a given column in separating two compounds if their distribution coefficients can be measured. 1612
ANALYTICAL CHEMISTRY
T
HE SEPARATION AKD IDENTIFICATIOS of polycyclic aromatic hydro-
carbons (PAH) in such complex mixtures as coal tars or residua from the refining of petroleum have received considerable attention, especially because these mixtures may be carcinogenic, and they may be encountered environmentally. Many combinations of compounds are very difficult to separate by adsorption chromatography, the method most commonly used, and other methods, including paper chromatoggraphy, have been applied (6). The formation of complexes of P I H with polynitro aromatic compounds has long been used as a means of purification and identification ( I ) . McCaulay and Lien have utilized the differences in basicity of the alkyl benzenes for their selective extraction from n-heptane (4). I n a series of papers, Hammick and coinvestigators have reported extensive experimentation on the acidbase interaction between polynitro aro-
matic compounds and various others including P.IH ( 2 ) . This paper describes a liquid-liquid chromatographic method based on the format,ion of complexes of PA\H and s-trinitrobenzene (TNB) xithin the chromatographic column. The complexes, being more polar t,han the hydrocarbons, are selectively partitioned into the polar stationary phase with a corresponding increase in the chromatographic retention volume. Compounds, which in the absence of T S B have identical distribution coefficients, may have such differences in the stability of their complexes with TKB that, marked differences in their distribution coefficients and great ease of separat,ion result. Pyrene and fluoranthene are two such compounds, and the effects on their distribution coefficients of the addition of varying 1 Present address, Kaiser Steel Corp., Fontana, Calif.
Fluoranthene, best grade, Uatheson, Coleman and Hell. 2-~Iethylfl~ioi,anthexie, 1.2-dimethylfluoi,anthcnc, 1,2,3-trimethylfluorant h e w , purificd, 1 h . A . c'. Laca.>agne. 5-Nethylbenz [ala11thracenc, 1-ethylp y i w ~4,5 ~ - ciiiiietli!,li)hrnanthrene, benzo[c!phcnanthi,c,iie, 2 - methylbmzo [ c ]~iheiinnthrcne, 2,9 - dime thy llw n z o [ c ]phenanthrene 5 6diniethylchi,!..,cne, l)unifird! Dr. 11. S.S e w n a n . Apparatus and Instruments. Convent ion a 1 110r o si li c a t e g la .- 5 c 11ro mat ographic tulles (the column used in obtaining the data on Table I1 was approximately 1.O-ciii. i.tl. a n d 4 5 cm. in length, and the column used t o obtain the d a t a of Figure 1 was allproximately O h ' X 30 e m . ) ; 12-nil. vials with polyethylene >toppers ; and ultraviolet si,ectroi,hotonieter l k k man DK-2. M e t h o d U s e d in Comparison of Stationarv Solvents. Solution> of pyrene (6 93 x 1 0 - 4 ~ ) and of duoranthene (1.44 X 10-'!M) in isoocatane separately were pai,titioned by shaking 4 ml. of cach in 12-1111. vials \\-ith 4 ml. of each of a series of six conc,c~nti,ations of ' l Y I 3 , ranging fi,oni 0 to 0.258~44,in each of thrw glycaols (IIEC;, TEG> and C-400). The rcsulting solutions in isooctane were tranefcved to other vials by means of a nidicinc d i ~ p p o rand shaken with 4 nil. of 12\. aquooiii SaOH t o rcinove TSH. The ultmviolctt absorbances of t h c aromatic h>di,oc.aybons in the isooctane \vert n i w w i ~ dinitially and following the waahixig with SaOH to deterniinc thc ~iot~tioniiai~itioned into the glycol. The effect of alkylation upon the‘ distribution cocficic~nt~of ai.oiiintic4 hydrocarbons w . 4 determinctl by the ianie procedure. 1he Iiiwtical. c h r o ~ i i a t o ~ i ~ a ~ asihic~ pect of the di>ti,ibution cocffic+nts cd and htahetwecn the ~ ~ r o p o ~moliilr tionary sol\.cntb is thr Icngtli of column requircd to q ) a i ' a t c c,oniliounds of interest. Although cxac't rdcwlatioIi> of ~
V o l u m e of Effluent, rnl
Figure
1.
Chromatogram of pyrene and fluoranthene
0.3-ml sample, on TNB C-400 column, 0.8 X 30 cm.
amounts of Th'B were measured in three solvents with different polarities, diethylene glycol, t,riethylene glycol, and Carboivax-400 (DEG, TEG, and C-400). 'lXH was selected over s-trinitrophenol (picric acid), on the basis of findings that coniplexes of 4-5 ring aromatics with picric acid were so unstable that the retention volumes were too small to provide separation of similar compounds.. Cornillexes of PA%H with 2.4,i-t,rinitrofluorenone were st'able enough but were so insoluble that revcxrsible precipitation occurred within the column. This caused the occlusion of other caompounds, with consequent impairment of separation. This happrned with '1'TB also, in DEG, but not in C-400. Several different compounds were chroniatographcd on columns that inc>luded '1'Xl$>with good separations k i n g obtained. . i l k ~ . l PAH may tw markedly more basic than their parcnt nuclei ( 4 ) and might bc rspcctcd to clistrihute more r c d i l y into the polar phase, with larger retention volumes resulting. This point was investigated.
Triethylene glycol, Carbide and Carbon Chemicals Co. Polyethylene glycol (Carbowax-400), Carbide and Carbon Chemicals Co. s-Trinitrobenzenr, Eastman, Distillation Products Industries. Isooctane, pure grade, Phillips l'etisoleum Co. l'hcnanthrene, chrywne, and benz [a]anthracene, reagent grade, Ilistillation Products Industries. Pyrrne, 11 H-benzo [blfluorene, best grade, Edcan Laboratories.
Table I.
Columpak, Fisher Scientific Co., 30-60 mesh ground to mesh 100-200, sieving wet. Diethylene glycol, Matheson, Coleman and Bell.
Carbowas400
r .
Comparison of Merits of Three Glycols as Solvents of Stationary Phase in Partition Chromatography
EXPERIMENTAL
Materials.
~
0
s
0 016 0,032
S i$
O.O(i4
0.12'3 0.258
4s
0 I(i 10 1 11 5 I :1 . :3
s
t X h !):;
!I , i l 10 !I 12 .i 1 .i0
Ir)llp
vrl'!'
IO x .>.,
-
,)
I 5 1 2 0 !I
VOL. 36, NO. 8, JULY 1964
1613
Table II.
Predictability of Retention Volumes of Several PAH on TNB Column
R value
K Compound
Calc. 1.95 2.05
Acenapht hene 4,5-l>irrietli~lphenant hrene Anthracene 1-Ethylpyrene 11 H-henzo [b]flnorene Phenanthrene
6.74
8.07 8.28 9 00 11.28 15.0 13 5 1 59 21 0 22 8 , 5 l .1
5,6-I>inieth?.lchr~.sene
Pyrene Fluorant heiie .5-~lethyli)enz[a]anthracene
Benz [alanthracene C hrysene Iliheiiz [a,h ] anthracene a
From col. 2 27 2 00 7 07 6 hl 8 94
8 4
9 15 13 12 19 21 55
74 3 5 1 6 8 4
Cak. 1 10 1 0 0 0 0 0 0 0
0 0 0 0
06 07 69 68 GO 63 56
43
1 1 0 0 0 0 0 0
47 42 34 32 16
0 0 0 0 0
47 50 36 33 16
09 70 64 GO 60 52
Reten tion Val. Peak Cak. Observed
From col.
32 32 51 55 56 58 67 82 7*5 84 103 110 221
43
33a
32 510 50 58. 56n 61 82Q 7t i a t inn of tlissolvotl gascs. 'ThL ~ i a p c i~b (#onc ~ ) i ~ ivith x ~ l t h c x trc,hriicluc. \vhic*h wcms I hc nio-t I)romibing for thc ~iui'iiosc, narnt>I y x:i> I ' O I ~ ItBogi,ai h y . 'Thib :innl>.si.:tliffrr- f m m R not,mal gas c~ht~riiiiato~isl,hic. ~li:iixtion bec~auscof LTHOUGH GRl..4T
(811
the presence of a very large excess of water. Water is irreversibly adsorbed on porous column materials, and moreover very small quantities are liable to upset the working of a katharometer detector (other types of detector are not suitable for the quantitative determination of permanent gases). I t is necessary either (a) to inject samples of gas previously extracted from the liquid, in which case the excellent capabilities of gas chromatography are limited by the numerous disadvantages of the extraction process (S), or (b) to take special precautions to avoid t,he carry over of the liquid into the detector. Method (a) has been used by 13ovijn etal. ( I ) and by Ramsey (6), but neither technique is suitable for the rapid analysis of very small saniples. Previous workers in category (b) ( 2 , 5 ) were concerned with hydrocarbons. These dissolve appreciably more gas than does water and are more readily separated from the gas. Extension of the method to permanent gazes dissolved in water calls for higher equilimcnt sciisitivities and more efftvtive means of gas liquid separation. EXPERIMENTAL
Apparatus. .I flow diagram for the apparatus finally used is shown in Figure 1. T h e detector employed was a very high sensitivity katharometer designed by t,he Irniversity Department of Physical Chemistry. It was of t,he direct flow type with 0,001-inch
diameter tungsten wire filaments fiwd along the axis of 2-mm. diameter cylindrical holes of 8.6-cm. length drilled in a stainless steel block. Inlet and outlet ports for gases were drilled a t right angles to the wires. Off-balance signals were measured with a conventional Wheatstone Bridge linked to a Control Instruments 1-mv. iiotentiometric recorder and a n Electro-Methods dial type integrator. The Aensitivity, S, of the katharometer was 47,500 mv.ml./mg. when determining hydrogen using argon as carrier gas a t a floiv rate of 62.5 ml./minute. The precolumn was an electrically heated 10-inch long> ';,-inch diameter tube filled with granular 6- to 12-mesh calcium sulfate; the selection of this material was governed by factors discussed later. Liquid samples were injected into the top of the liarking 1vit.h a 1 4 . Record type syringe through a small rubber serum cap. For analysis of gas mixtures, a separating column of 44- to 60-mesh 5:i molecular sieve jreact,ivated hy heating was placed hetween thc to 340" precolumn and the measuring ,