V O L U M E 28, NO. 3, M A R C H 1 9 5 6
303
sulfolane packing, was used continuously for several months a t
35" C. and showed no change in performance. The problem of determining nonhydrocarbon gases has received little consideration in this investigation. At present it is difficult to visualize a column that will separate the uncondensable gases from one another as well as the individual hydrocarbons. A simple cold-trap separation of such samples might provide a hydrocarbon fraction for analysis by gas-liquid partition chromatography and an uncondensable fraction for adsorption chromatography (11). .\nother possible approach is separation of the uncondensable gases in a single fraction with methane b y gasliquid partition chromatography, and analysis of this fraction by mass spectrometry. The design of suitable equipment for sampling the effluent gas from the thermal conductivity cell will greatly aid this approach. Such equipment will also be valuable for collecting hydrocarbon components when further qualitative identification is required. ACKNOWLEDGMENT
The authors are grateful t o P. E. Porter for the many helpful suggestions made in the course of this work and to G. J. Pierotti
for suggesting the use of dimethylsulfolane as a selective stationary liquid for retarding olefins, LITERATURE CITED
Bradford, B. R , Harvey, D., Chalkey, D. E., J . Znst Petroleum 40, 8 0 (1954).
Craats, F . van de, Anal. Chim. Acta, in press. Dimbat, hI., Porter, P. E., Stross, F. H., -4s.4~.CHEM.28, 290 (1956).
Eggertsen, F. T., Knight, H. S.,Groennings, S., Ibid., 28, 303 (1956).
James, A . T., Bwchem. J . 52, 242 (1952). James, A. T., llartin, A. J. P., Ibid., 50, 679 (1952). Keulemans, A. I. M.,Kwantes, A . , 4th World Petroleum Congress, Rome, June 1955. Keulemans, A. I. AI., Kwantes, A, Zaal, P , Anal. Chim. Acta, i n press. Lenoir, J. JI., Univ. Ark., Eng. Expt. Sta., Bull. 18 (1953). Lingane, J. L., Jones, S. L., ANAL.CHEM.22, 1220 (1950). Patton, H. W., Lewis, J. S.,Kaye. W. I., Tbid., 27, l i 0 (1955). Porter, P. E., Deal, C. H., Stross, F. H., Pierotti, G. J.. unpublished paper. Ray, N. H.. J . A p p l . Chem. (London) 4, 21 (1954). RECEIVED for review September 18, 1955. Accepted J a n u a r y 6, 1956.
GAS CHROMATOGRAPHY Use of liquid-Modified Solid Adsorbent to Resolve 6, and C, Saturates F. T. EGGERTSEN,
H. S. KNIGHT,
and SIGURD GROENNINGS
Shell Development Co., Emeryville, Calif.
A new type of column packing for gas chromatography consists of an active solid to which a small amount of liquid solvent has been added. The amount of solvent is sufficient to prevent or greatly reduce the tailing of chromatographic peaks normally encountered with active solid supports but not to replace the solid as the principal sorbing medium. Such liquid-modified solid adsorbents therefore resemble unmodified solids in retaining paraffins relative to naphthenes. Thus, the CS naphthenes emerge ahead of the C7 paraffins, in contrast to the behavior in gas-liquid partition columns, where the reverse sequence is observed. Usefulness of the new type of column is exemplified by the analysis of a mixture containing all ten of the commonly encountered C; and CS saturates. A 50-foot column of a commercial carbon black (Pelletex) containing 1.5% squalane (hydrogenated squalene) resolved all components, including 2,3-dimethylbutane and 2-methylpentane, which could not be separated with any of a large number of partition-type columns tested. The time for analysis is about 2 hours and the average error has been about 0.5%.
Q
CASTITATIVE resolution of saturated hydrocarbons into types and individual constituents has long been a problem of practical importance. For type estimations several methods have been more or less thoroughly explored, including empirical formula, refractive index-density relations, liquid-liquid adsorption chromatography, thermal diffusion, and solvent partition. However, these methods have been only moderately successful and do not offer much promise of yielding a more detailed analysis. Infrared and mass spectrometry give essentially all individual compounds, possibly up t o Cg, with excellent results, but evtensive sample preparation by chromatography and fractional distillation is required at least for Ci and higher hydrocarbons.
A new approach to the problem of analysis of hydrocarbons has become available with the rapid development of gas chromatography. By means of this fascinating tool encouraging results have been obtained for C j and lower hydrocarbons both by gas-liquid partition ( 4 , 9, 11) and by gas-solid adsorption chromatography using solid adsorbents (5, 10, 11). I n gasliquid partition chromatography the sample components are separated by their differing solubilities in the liquid film on the support, and adsorption effects of the support are generally negligible. I n the adsorption scheme, on the other hand, no liquid is used, and the separation is based on differing adsorbabilities of the components on the solid surface. With increasing molecular weight the resolution of individual compounds is of course complicated by the rapid rise in the number of isomers. n'evertheless, reported results are encouraging. For example, Keulemans and Kwantes (8) have demonstrated that normal and isoparaffins through Cg can be distinguished by gas-liquid partition chromatography, provided the naphthene content is low. James ( 7 ) also employed the technique to determine C6 and CS constituents in commercial petroleum ethers and was able to separate many of the individual components. Thus, it was shown that gas chromatography has great potentialities for hydrocarbon analysis, and it seemed desirable to make a thorough study of the behavior of various column types as applied to hydrocarbons. I n this study a suitable column and technique were developed for the separation of Cb and CSsaturates. Gas-liquid partition chromatography columns n-ere first investigated, generally with Celite-type supports and 40% added liquid. A41though a wide variety of liquids was tested, many of which gave excellent resolution in general, none was found which would separate 2,3-dimethylbutane (boiling point 58.0" C.) and 2-methylpentane (60.3' C). Furthermore, all the partition columns retarded naphthenes relative to paraffins, so that cyclohexane, and often methylcyclopentane, emerged after the first C7 paraffins. On the other hand, all solid adsorbents were f o m d
304
ANALYTICAL CHEMISTRY
to retard paraffins relative to naphthenes, as might he expected from the liquid-solid adsorption work of Hirschler and Amon (6). These authors showed, for example, that both silica gel and carbon adsorb n-heptane more strongly than methylcyclohexane from 1 to 1 mixture, and that silica gel adsorbs 2,4-dimethylpentane preferentially t o cyclohexane. I n gas chromatography using solid adsorbents it was observed that all of the (26's emerged n-dl ahead of the C?'s, and some adsorbents resolved the ti ouhlesome pair, 2,3-dimethylbutane and 2-methylpentaiic.
1 1~
FLOW METER
11
Pelletex. This material, which was used as the adsorbent, is a pelleted furnace black (Godfrey Cabot Co., Boston, >lass.) of surface area 24 sq. meters per grain, screened to 14-48 mesh. Squalane. This C ~ , , H paraffin B~ (2,6,10,15,19,23-hexamethyltetracosane) of molecular weight 423 and boiling point 210' C. at 1 mm., is obtained by hydrogenation of squalene (an acyclic isoprenoid ex-shark liver oil) over a platinum catalyst at 200 to 1000 pounds per square inch. Its high boiling point, good thernial stability, and low viscosity make it an especially attractive liquid of the paraffin type. Preparation of Columns. The packing was made from 157 grams of Pelletex 1350 m1.1. 14-48 mesh. and 2.36 grams of iqualane in about 275 ml. of petroleum ether. Thesquarane solution m as added to the Pelletex so as to wet all particles, and the petroleum ether was evaporated on a steam bath, followed by drying a t 110" C. for 1 to 2 hours. The dry material was packed mith the aid of an electric vibrator into both ends of a 50-foot by l/c-inch (outside diameter) copper tubing, folded into U-shape The ends were then plugged with glass no01 and the tubing was wound into a coil of four concentric helices on a 13/8-inch tube as a mandrel, the finished coil being about 7 inches long and 3*/2 inches wide. Two 10-foot columns were also prepared, one with and one without sqnalane. Hydrocarbon Test Samples. Phillips Petroleum Co. ('purp'' hydrocarbons (99% minimum) were used.
5-LITER
RESULTS 4ND DISCUSSION
BUFFER TANK
Figure 1. Schematic diagram of gas chromatographic apparatus
Complete separations were difficult to obtain with solid adsorbent,s owing to overlapping of the peaks, caused by tailing. However, it was found t.hat tailing could be reduced or virtually eliminated by adding a small amount of a high-boiling or strongly adsorbed liquid to the adsorbent. With suit'ably low amounts of liquid, paraffins are retarded relative to naphthenes, as with the dry adsorbents. Therefore these new packings may be viewed as liquid-modified solid adsorbents. Of coursel as more liquid is added the packing approaches the gas-liquid partition chromatography type, until finally the effect of the surface of the active support is generally nullified. Of the modified adsorbent packing8 tested for Cb aiid CS saturates, which include small amounts of various liquids on carbons, silica gel, and Celites, the best was 1.5% squalane (a branched paraffin, C3OHa) on Pelletex (a furnace black). The results obtained with this packing in the analysis of Cj-Ce mixtures are discussed below. The performance of this and other columns will be compared in more detail in a subsequent publication.
The effectiveness of 1.5% squalane for reducing tailing on Pelletex is illustrated by Figure 2, where chromatograms for a 2 to 1 mixture of 2,4-dimethylpentane and cyclohexane on dry and modified Pelletex are presented. The greater peak symmetry obtained with the modified Pelletex column shows that the adsorption isotherm is more nearly linear than with the bare solid. Squalane affected the separation in two additional ways: The adsorptivity of the solid was reduced, as evidenced by the lower temperature required for comparable emergence times, and the paraffin emerged sooner relative to the naphthene. T o illustrate the behavior of the squalane on Pelletex column, the results of the analysis of three synthetic blends of c5-C~
2,4 -DMP w
-1
50
v, u
EXPERIMENTAL
Apparatus. The apparatus used was conventional, Kith certain modifications described by Dimbat., Porter, and Stross (3). .4 flow diagram is given in Figure 1. The sensing elements were thermal conductivity cells (Gow-Mac Instrument Co., Madison, N. J.) of the convection-diffusion type (nonflow-sensitive). The potentiometer was operated a t 5 or 10 mv. for full scale deflection. The column and the conductivity cell block were contained in separate thermostats (silicone oil) a t 40" and 100" C., respectively, and the two units were connected by short pieces of l/rinch copper tubing. Samples were injected directly into the cell block by means of a syringe connected to the carrier gas inlet line. Helium was used as carrier gas at a flow rate of about 20 ml. per minute, and when it was desired to compare emergence times, the measured values were recalculated t o a flow rate of exactly 20 ml. per minute at, 20' C. The inlet pressure was 280 mm. for the 50-foot column employed and proportionally less for the 10-foot columns tested.
20
IO
0
I
I
IO
1
I
20
30
40
50
MINUTES
Figure 2. Effect of tailing reducer in gas chromatography 10-foot X '/Anoh Pelletex 2.4-DMP CH. 2 to 1 1.5-mg. sample Flow rate, 20 ml. per minute 20 scale units = 1 m v .
+
V O L U M E 2 8 , NO. 3, M A R C H 1 9 5 6
305
sample or 3 mg. of closeneighboring components should be charged, and a temperature of 40" C. or lower is preferred. Column length is another important factor, for in order t o Component, Wt. 70 Av. achieve these separations it was Error n2,22,3n01 daiiiple 2-1IB Pentane C P DhIB DMB 2-hIP 3-AIP AICP Hexane CH /O necessary t o extend the length No. I from an initial of 10 feet to 50. Present .. .. 4.9 10.2 1.5.6 a Longer columns were not tried, 0 6 but might be expected to give 0 5 still better separations. The efPresent .. .. .. 2.5 1 2 3.6 2.4 .. 90.3 .. FoundC fect of carrier gas flow rate was 0 2 .. .. .. 2 3 1.2 3.3 3.0 .. 90.2 .. Planini. not investigated; the rate of 20 0 3 HW b ,. .. .. 2.3 1.3 3.4 3.1 .. 90.0 .. xo. 3 ml. per minute was selected, bePresent ' 2 . 4 5.7 8 9 3 4 7.1 10.8 14.7 4.1 29.0 13.9 E'oundc cause it had been found suit0 3 Planim. 2.8 5.7 9.6 3.3 6.8 100 148 4.4 28.6 14.0 able in studies with other types 0 3 HWb 2.7 5.7 9.4 3 3 6.8 10.5 16.1 4.4 285 13.6 of l/&ch columns. '& Average of triplicates. h Product of peak height and peak width a t half height. The percentage of each com.4verage of duplicates. ponent was determined by measuring the areas under the peaks and calculating the per saturates are shon-II in Table I. An adsorptograni representing cent of total for each, assuming the areas to be a measure of the the resolution of a sample containing all the commonly enweights of the hydrocarbons present. Measurements were made countered C6and c6 saturates is reproduced in Figure 3. (Xumerboth by means of a planimeter (extrapolating the beginning or end of the peaks where necessary) and by multiplying peak height ous cyclopropanes and cyclobutanes lie in this boiling range. Their occurrence is rare, however, at least in petroleum, and they by width at half height (HI\') ( 1 ) . Theresults were about equally have therefore been neglected here.) It is seen that virtually accurate, the average error being 0.3 to 0.5y0in both cases, and complete separation of all ten components was achieved and that the maximum 1% (Table I). The weight by width method is prethe last CS emerged well ahead of the first C,, the latter (when ,ferred because it is faster. present) being indicated by the dotted curve. The elapsed time As expected, the paraffins emerged in the order of their boiling for the analysis was about 2 hours. The sharpness of peaks in points, but the three naphthenes emerged with lower boiling these experiments was comparable with that obtained with the paraffins. Thus, cyclopentane (boiling point 49" C.) emerged best gas-liquid partition chromatography columns. n-ell ahead of 2,2-dimethylbutane ( 5 0 ° ) , cyclohexane appeared The peak separations became poorer with increasing sample before 2,4-dimethylpentane (both 81 "), and methylcyclopentane size and temperature. Therefore not more than about 20 mg. of (72') before n-hexane (69'). Therefore, with respect to behavior
Table I.
Gas Chromatographic Analyses of CS&
Saturates Blends
80
n
n-Hex
30 ln
t 3
w
-1
3
20
70
60
01
i
- n-Her - 3-MP - 2-MP
1
- 2,J-DMB
c6
CH
0 W
50
- 2,2,-DMB,
-cP
I 40
1
A'
27
30
57
94
45
33
60
68
105
I51
75
44
90
285
105
30
136
120
135
I50
MINUTES
Figure 3.
Gas chromatographic analysis of C6 and Ce saturates Sample 3 1.5% squalane on 50-foot X '/,-inch Pelletex 400
c.
280-mm. pressure IO-mg. sample Flow rate, 20 ml. per minute 20 scale units = 1 mv.
ANALYTICAL CHEMISTRY
306 toward naphthenes, the 1.5% squalane on Pelletex column is more nearly like a solid adsorbent than like a liquid-type column which retards naphthenes. Although the liquid-modified Pelletex column has proved useful for analysis of CS and c 6 saturates, probably the chief value of this type of column is that it separates naphthenes from paraffins in a manner opposite to gas-liquid partition chromatography columns, and therefore offers new possibilities for analysis of saturates. Thus, a “carbon number” analysis is possible, a t least through C,, according to the authors’ experience, and probably higher. When a further breakdown or type analysis is desired, the carbon number cuts can be separated into naphthenes and paraffins by means of a highly polar gas-liquid partition chromatography column, such as glycol, which greatly retards naphthenes. A few olefins and benzene were tested as to emergence time through the squalane on Pelletex column. These results are listed below, together with those for saturates of similar boiling point. Hydrocarbon 1-Pentene n-Pentane 1-Hexene n-Hexane Benzene Cyclohexane Cyclohexene
Boiling Point,
c.
30 36 64 67
Emergence Time, Min. at 20 Ml./Min. 38 36
80 81 83
91
102 114 113 132
With the liquid-modified Pelletex column unsaturation has little effect on the emergence time. All the c6 ring compounds emerge ahead of the C; band, which begins a t about 150 minutes, as shovm in Figure 3. Therefore this column is suitable for carbon number analysis of a mixture containing all three types of hydrocarbons. Interfering olefins and aromatics can be removed and a representative sample of saturates obtained for gas chromatographic analysis by liquid phase fluorescent indicator chromatography using a column similar to that described by Criddle and Le Tourneau ( 2 ) . I t consists of four sections of inside dimensions (top to bottom) 150 X 25, 350 X 10, 350 X 5 , and 700 X 2 mm., the three lower sections containing Davison’s grade 923 silica gel. Gsing a syringe to reduce evaporation losses, 2 to 3 ml. of sample are introduced below the gel surface and eluted with isopropyl alcohol under a pressure of 1 to 2 pounds per square inch. The saturates are collected in a chilled vial, protected from air, and the cut is taken when the first drop of yellow-green fluores-
cent dye forms a t the tip of the column as observed in ultraviolet light in a dark room. This technique was tested with B blend of CS-G saturates with olefins from a catalytically cracked gasoline; gas chromatographic analysis showed a maximum change of 1% in the composition of the recovered materials. For the analysis of CSand C6 saturates the gas chromatographic method described has certain advantages over spectrometric methods. Because saturates outside the C& range do not interfere, as they do in spectrometrv, extreme care in fractionation is not required. Furthermore, the presence of such higher or lower boiling constituents can be detected. However, in the absence of C7’s the infrared method is roughly comparable with the gas chromatographic method in accuracy and time per analysis. Another important advantage of the gas chromatographic method is that the apparatus is relatively simple and does not require a highly skilled operator. I t is concluded that liquid-modified adsorbents, such as the squalane-Pelletex column employed here, constitute a useful addition to the media available for gas chromatographic separations. Columns of this type differ markedly from the more familiar gas-liquid type columns in respect to the sequence of resolution of saturates. They are also “flexible,” for by varying the amount of liquid the naphthene peaks can be moved relative to the paraffin peaks almost a t will. The resolution is excellent and comparable with that of the best gas-liquid type columns tested. LITERATURE CITED
(1) Cremer, E., Muller, R., Mikrochemie 36, 553 (1951). (2) Criddle, D. W., Le Tourneau, R. L., ANAL. CHEM.23, 1620 (1951). (3) Dimbat, M., Porter, P. E., Stross, F. H., Ibid., 28, 289 (1956). (4) Fredericks, E. M.,Brooks, F. R., Ibid., 28, 297 (1956). (5) Griffiths, J., James, D., Phillips, C., Analyst 77, 897 (1952). (6) Hirschler, A. E., Amon, S., I n d . Eng. Chem. 39, 1585 (1947). (7) James, A. T., M f g . Chemist 26, 5 (1955). (8) Keulemans, A. I. M., Kwantes. A., “Analysis of T’olatile Organic
Compounds by Means of Vapor Phase Chromatography,” World Petroleum Congress, Rome, June 1955. (9) Keulemans, A. I. M., Kwantes, A., Zaal, P., A n a l . Chim. A c t a 13, 357 (1965). (10) Patton, H. W., Lewis, J. S., Kaye, W.I., ANAL.CHEY.27, 170 (1955). (11) Ray, N. H., J . A p p l . Chern. 4, 21, 82 (1954).
RECEIVED for review September IS.
1955.
Accepted January 6, 1956.
Mass Spectrometric Analysis Broad Applicability to Chemical Research FRED W. McLAFFERTY The Dow Chemical Co., Midland, Mich. The heated-inlet mass spectrometer, applied in the past mainly to petroleum compounds, is an analytical tool of broad applicability to the whole field of chemistry. Quantitative analysis approaching the high accuracy and number of components found previously with light hydrocarbons is possible. Positive identification of unknown components of complex mixtures is illustrated. Unique information on molecular structure is provided, and complete structure determination is often possible without standards. It is hoped to show that the mass spectrometer provides a valuable complement to the rapidly growing field of instruments capable of analyzing a wide diversity of samples.
T
HOUGH the primary use of the analytical mas8 spectrom-
eter has been in routine quantitative analysis of light hydrocarbons, a number of investigators have reported on the unique qualitative and quantitative applications to the general field of chemistry, especially organic chemistry. The analytical advantages of the mass spectrometer for volatile oxygenated compounds (14, 18, 21, 22, 3.5,36), thiophenes (IQ), aromatic hydrocarbons (IQ), lactones ( I S ) , acids (16), haloalkanes (1, 26, S5), amines ( 7 ) , and metallo-organic compounds (8, 9) have been described. Significance of the spectra of such compounds as ketene dimer (M), pentaborane (IO),diborane (28),dimethylphosphinoborine trimer ( 1 2 ) -phenol, thiophenol, and aniline ( 2 7 ) has also been discussed.