pentane are shown in Table 11. For comparison, data taken from the work of Sternberg et d.(8) were converted to the same basis and relative responses of their Beckman flame ionization detector are included in the table. Because different systems were used to run the samples, no precise measurements of detector sensitivities to the chlorinated hydrocarbons were made. However, the same column and column conditions were used and general conclusions can be drawn. The triode argon detector was enough more sensitive than the flame detectors to be more useful in examining trace constituents. The standard flame ionization detector was much more sensitive than the experimental high temperature flame ionization detector. The choice of detector would depend upon the nature of the sample t o be analyzed. Both types of detector would require calibration for each component. Based on weight per cent calibration, the triode requires smaller correction factors than the flame (Table 11); as the flame detector is a carboncounting device, its correction factors would be smaller on a per cent carbon basis. For the analysis of chlorinated compounds in trace quantities in hy-
drocarbons, the ehoice would be the triode, and for hydrocarbons in chlorinated compounds, the flame. The use of two detectors which give different responses, such as the triode and flame, can yield qualitative information about the identity of a compound. If, under otherwise identical conditions, two different detectors are employed, the chromatograms will often show differences in the relative heights of the corresponding peaks. If the relationship between the response of the two detectors for various classes of substances is previously established with mixtures of pure compounds, these differences in peak heights may be utilized for purposes of identificationi.e., any paraffin, naphthene, or aromatic compound as a constituent of a mixture would give a much greater relative response on the flame detector than would a chlorinated hydrocarbon. From comparisons of the relative responses of two or more detectors, unknown substances can often be classified according to type. Combination of response data with retention data would often suggest the exact identity of unknowns. Future work for qualitative and quantitative analysis of mixtures of
chlorinated hydrocarbons foresecs the use of electron capture techniques as described by Lovelock (5, 6) and Gregory and Lovelock (9). These techniques should be particularly useful in trace constituent determinations of chlorinated compounds because of the very high sensitivity of such detectors to these compounds. LITERATURE CITED (1) Desty, D. H., Goldup, A., Whyman, B. H. F., J . Znst. Petrol., 45, 287 (1959). (2) Gregory, N. L., Lovelock, J. E., Proc. 1961 International Gas Chromatog-
raphy Symposium, East Lansing, hlich., June 1961, pp. 151-8. (3) Hollis, 0. L., ANAL.CHEM.33, 352
(1961). (4) Lovelock, J. E., ANAL. CHEM. 33, 162 (1961). (5) Lovelock, J. E., “Gas Chromatography,” 1960, R. P. W. Scott, ed., p. 16, Butterworths, London, 1960. (6) Lovelock, J. E., Nature 189, 729 (1961). (7) MacWilliam, I. G., Dewar, R. A.,
“Gas Chromatography,” D. H. Desty, ed., p. 142, Butterworths, London,
1958. (8) Sternberg, J. C., Gallaway, W. S., Jones, D. T. L., Proc. 1961 International
Gas Chromatography Symposium, East Lansing, Mich., June 1961, pp. 159-84. RECEIVED for review December 15, 1961. Accepted June 11, 1962.
Determination of Alkanes and Cycloalkanes through C, and Alkenes through C, by Capillary Gas Chromatography A. G. POLGAR, J. J. HOLST, and SIGURD GROENNINGS Shell Development Co ., Emeryville, Calif.
b Gas chromatographic methods have been developed for the determination of the C1 through Ca alkanes, cyclopentanes, and cyclohexanes and of the C? through C, alkenes. Analysis is accomplished by making runs at two temperatures in a 300-foot silicone oil-coated glass capillary column with hydrogen flame ionization detector. Because of the marked difference in pattern of emergence at the two temperatures, peaks of the crowded areas of one chromatogram are resolved to a considerable extent in the other. Thus four fifths of the possible saturated and unsaturated compounds can be quantitatively accounted for as individuals. Applicability of the methods is demonstrated with the Cq to 130’ C. fractions of saturates from a platformer feed and product, and with two CT-range olefinic materials.
1226
ANALYTICAL CHEMISTRY
T
utility of capillary gas chromatography has been amply demonstrated in the numerous articles of the past few years dealing with the theory of capillary columns as well as with the various aspects of the technique, including the design of sensitive ionization detectors. As coated capillaries possess enormous fractionating efficiencies, they are eminently suited for the examination of complex mixtures containing many close-boiling components. An obvious application is, therefore, their use in the analysis of natural petroleum fractions and hydrocarbon products of similar complexity. The resolution of the lower range of saturates up to and including the C, compounds has been demonstrated on a vsriety of column coatings. The extension of the capillary column technique into the middle range of gasoline saturates was recently reported by Desty, HE
Goldup, and Swanton (3). Using a 900-foot glass column coated with squalane, 122 peaks were resolved from the C3 to C g portion of the American Petroleum Institute’s Ponca City crude in a 20-hour run. More than half of these peaks (a few of them containing two or three compounds) were identified, mainly in the C3 to Cs range. Full component analysis of the Cp alkanes, including some low-boiling Cl0 alkanes, is possible through their retention times determined on squalane by Simmons, Richardson, and Dvoretzky (IS). The novel methylene insertion reaction was employed to synthesize 19 unavailable C9 isomers. I n contrast, component analysis of saturates by packed columns has not been attempted beyond the C, compounds (6, 11). The potentialities of coated capillaries for the examination of olefinic mixtures have not been utilized in any systematic
.
Sample N,
VAPCRIZER
i
Discard .tre*rn
Figure 1. Gas chromatograph with sample splitter, glass capillary column, and flame ionization detector
manner. P d e d column techniques have been estended only to the end of the CC range; however, the resolution 011 a simple substrate is incomplete (IO), and full component analysis requires the use of three columns (9). The analytical schemes proposed in this paper cover the range of Cl to Cs alkanes, cyclopentanes, and cyclohexanes with several of the overlapping Cs alkanes and cyclopentanes, and the C2 to C, alkenes. Only one column is employed, a 300-foot glass capillary coated with silicone fluid 96. When operated a t 25' C., this column yields several multiple peaks. However, rerunning the sample a t another temperature (0' or 50' C.) produces a sufficiently different emergence sequence (IS) to permit the quantitative determination RS individuals of all but a few of the possible isomers. Peak identifications lvere based on blendsof pure compounds.
COLUMN.The column consisted of a glass capillary tubing coiled in the form of a doughnut, having a coating of General Electric Co.'s silicone fluid 96, viscosity grade 1000 (SF-96), deposited from a 10% n-hexane solution. [The technique of manufacturing springlike coils has been described by Desty, Harsnape, and Whyman (4).] Its operating parameters are given in Table I. DETECTOR. The hydrogen Aame ionization detector was of conventional design. The outer shell (not shown in Figure 1) was a 5-cm. brass cylinder enclosing the flame jet (+) and the ion collector (-), a 12 x 12 mm. cylindrical platinum screen positioned 12 mm. above the jet. The polarizing potential across the electrodes was 300 volts. The detector response was coupled to a Cary Model 31 vibrating reed electrometer and a Brown potentiometric recorder with 1-mv. span and 1-second full scale pen speed.
Calibration Compounds. SATUThe range of compounds intended for coverage in this paper was the C1to C, alkanes, cyclopentanes, and cyclohexanes. All but one (1,cis2,cis-4trimethylcyclopentane) of the 73 possible compounds of this group mere available, .mainly from the Samples and Data Office of the American Petroleum Institute (API), in 99% or better purity. Several of the lower boiling Ca alkanes and cyclopentanes were expected to emerge in the range of the Cg compounds. Of these, nine alkanes were obtainable as standard API samples, and small quantities of four cyclopentanes were received from API Research Project (APIRP) 58A. One of these samples was designated as a mixture of 1,1,cis-2,trans-4- and 1,1,cis-2,cis-4-tetrainethylcyclopentanes. The boiling point, calculated by APIRP 44 for both isomers, is 130' C. Although the gas chromatographic check of this material showed two peaks, one of them was only 4% of the total area and emerged far too early for a 130' C. boiling point compound. We concluded, therefore, RATES.
EXPERIMENTAL
Apparatus. A schematic diagram of the apparatus is shown in Figure 1. The sample splitter, column, and hydrogen mixing chamber of the detector were enclosed in a n insulated box (indicated by the dashed lines) which served as a constant temperature oven when elevated temperatures were required. For operation a t 0' C., the column coil was immersed in an ice-water slurry. Dead volumes in the column assembly are virtually eliminated by inserting the column ends directly into the sample splitter and the hydrogen mixing chamber through silicone rubber plugs which do not contact that part of the sample going through the column. The streamLsplit ratio was controlled by a variable length of nylon tubing attNhed to the discard stream line. This ratio was approximately 1000 in all our experiments. VAPORIZER. A X 2 inch brass block with a straight 3/8-1nch bore, located on the top of the oven, served as a sample vaporizer. A cartridge heater was used to keep it a t 225" C.
that this smsllcr pc;ik is an impurity, and the larger peak contained both spatial configurations. Uneupectetlly, l,cis-3,cis-5-triniethylcyclohexane (-4PI No. 1070) with a boiiing point of 138.4' C. emerged within the Cs range, just ahead of the terminal compound, npropylcyclopentane (b.p. 130.9" C.). Cycloheptane was also included, but other cyclic structures, assumed to be rare in gasoline, mere not considered. The list of these calibration compounds is given in Table 111. OLEFINS. All but three of the 65 possible alkenes of the C2 to C7 range n.ere available commercially (mainly as API standards) in 99% or better purity. The unavailable compounds were three Cf alkenes-namely, cis-2-methyl-3-hesene (b.p. 86.0' C.), cis-5-methyl-2-hesene (b.p. 89.5' C,), and 2-ethyl-lpentene (b.p. 94.0' C.). The last of these was fortuitously discovered in one of the process heptene mixtures esamined. It could not be isolated, but it appeared permissible to add this sample to the 62-compound master blend. A full list of these 63 compounds is given in Table 1V. Emergence times xere also obtained for all of the possible CS and Ce cyclenes: Cyclopentene 1-Methylcyclopentene 3-Methylcyclopentene 4-Methylcyclopentene Cyclohexene Several of the C7cyclopentenes should emerge in the region of the C, alkenes according to their boiling points; unfortunately, however, none of them is available. Master Blend Chromatograms. SATURATES. The chroniatogram Of our master blend of S8 compounds, obtained a t room temperature, is reproduced in Figure 2. Blends of increasing complexity were used to obtain peak identifications for the congested areas. The 67 peaks in this chromatogram comprise 51 singlets, 12 doublets, three triplets, and one quad-
Table
1.
Coated Glass Capillary Column Dimensions 0,012 inch X 300
feet
Volume Coating
6 . 5 ml. SF-96 (from
Liquid phase Av. liquid thickness Gas (Nt)/liquid volumes Inlet pressure Linear flow rate Exit flow rate Sample size Plates calcd. from injection (?&) H.E.T.P. Resolution (n-C,/ n-G)
13 .O p.s.i.g. 8 . 6 cm./sec. 0 . 6 cc./min. 0 2 to 2 p g . 360 ,000
10% in n-C6) 0.015 ml. 0 . 2 micron 425
0 . 2 5 mm. 68
VOL. 34, NO. 10, SEPTEMBER 1962
1227
COLUMN 24
2 y 7 2 3 W 7
TEMP
- 26.C
3tP 7
7
7
lt2DMCP
7
ECP
Yln"ll,
20
trom ,"lac! 0"
25
30
COLUMN T E M P SO - 52'C
35
40
50
45
6
ll3DYCP
CH
3iP
55
60
65
ro
7 7
.
ICWI,
Figure 2.
Chromatograms of master blend of C1 to C8 alkanes, cyclopentanes, and cyclohexanes (plus some Cs's) Column. Glass capillary, 0.01 2 inch X 300 feet, coated with silicone fluid 96 Detection. Hydrogen flame ionization
ruplet. A significant rearrangement in the sequence of emergence takes place when the column temperature is raised to 50' C. The chromatogram of the master blend, obtained a t this temperature, is also shown in Figure 2. Although the total number of peaks decreased to 62, several of the compounds under the multiple component peaks of the room temperature chromatogram are now either singlets or appear associated with peaks of compounds that
1228
ANALYTICAL CHEMISTRY
were singlets in the room temperature chromatogram. The analytical usefulness of this auxiliary 50" C. chromatogram rests on the varying temperature dependence of retention volumes. The plot of the adjusted retention volume of a compound against the reciprocal of the absolute temperature is, in general, a straight line. These lines have characteristic slopes which are related to the apparent heats of solution of the
vapors in the column substrate (11). Close lines with sufficiently different slopes may cross, which means that the sequence of emergence for the two compounds represented by these linea is now reversed. The combined interpretation of the two chromatograms permits the quantitative determination of all but 15 of the 88 components of the master blend. These 15 unresolved compounds appear as six doublets and one triplet:
Table II. Multiple Peaks in the Czto C7Alkenes Master Blend Chromatogram at 24', , 'O and 50" C. Nultiple Peak0 At 24' C. At 0' C. At 50" C. 21MPr'/lB' No separation No separation KO separation 3311MlB'/c2P' Partial separation Well resolved; 33DMlB- emerges first S o separation 4hf 1P-/3M 1P' Partial separation N o separation S o separation tdHs'/c3Hx' Partial separation No separation N o separation 44DAV1P-/3Mt2PNo separation Measurable separation (3Mt2P' is now a No separation singlet, but 44DiMlP- coemerges with 2M2P') 33 1)M 1PP/23DM2B' Partial separation Well resolved; 33DiM1P' emerges first Xo separation 3411M 1P'/44DMc2P' Partial separation Measurable separation; 34DRflP' N o separation emerges first 5M 1Hs=/23DlI 1P' S o separation No separation Well reBolved (but 23DRIlP- now coemerges with 2Mt3Hx,= and 5MlHs' is only partly separated from 3ElP') KO separation 4Mc2Hx- is measurably separated ahead No separation of 3M2E1B'/4Mt2Hxm doublet S o separation No separation No separation No separation No separation No separation Partial separation No separation Well resolved; t2Hp' emerges first No separation Well resolved; 23DM2P' emerges first No separation I
1,cis-2,trans-4-Trimethylcyclopentane +l,l,czs 2,2,4,4-tetramethylpentane - 3,trans - 4 - Tetramethylcyclo-
+
pentane cyclopentane
I,cis-2,trans-3-trimethyl-
+
1,cis-4-Dimethylcyclohexane 1,trans-3-dimethylcyclohexane l,l,cis - 2,cis - 4 - Tetramethylcyclopentane 1,1,cis-2,trans-4-tetramethylcyclopentane 1-Ethyl-cis-2-methylcyclopentane 2,2,3,4-tetramethylpentane 1,cis-3,cis-5-Trimethylcyclohexane n-propylcy clopentane 1,cis - 2 - Dimethylcyclopentane methylcyclohexane 2,2,3,3-tetramethylbutane
+
+ + +
+
One of the Cs compounds, l,n's-2,cis-4-trimethylcyclopentane, was not obtainable for calibration. According to its calculated boiling point (117' C.) this is expected to emerge in the 3-ethyl3-methylpentane/l ,cis-3-dimethylcyclohexane region. Indeed, during the analysis of our platforming feed sample (see below), a discrepancy was found in area measurements between the room temperature and 50' C.curves that could be resolved only by assuming an unlisted compound coemerging with 2-methylheptane a t rooni temperature, and appearing with the 3-ethylhexane 1,cis2,trans-3-trimethylcyclopentane doublet a t the elevated temperature. We felt that the evidence was fairly conclusive for identification of this species as 1,cis2,cis-4-trimethylcyclopentane. Although we tested all available C9 saturates, the 15 compounds included in the master blend do not represent the complete account of the overlap with the Cs range. Listed below are some additional Cg alkanes and cyclopentanes, several of which should be expected to emerge ahead of the terminal C8, a-propylcyclopentane, either at room temperature or a t 50' C.:
+
B.P., O
2,4-Dimethylheptane 4-Ethyl-Smethylhexane 1,trans-2,cis-3,trans-4-Tetra-
methylcyclopentane 1,l ,cis-2,trans-3-Tetramethylcyclopentane 1,trans-2,trans-3, cis-4-Tet ramethylcyclopentane 1,cis-2,trans-3, cis-4-Te tramethylcyclopentane 1,1,2,2-Tetrarnethylcyclopentane
c.
132.9 133.8 127.4 130. 131. 131. 133.
None of these compounds is available at present, although the third, fourth, and seventh are on the priority list for synthesis by APIRP 58A. In analyzing a platformer feed and product saturates (see below), a few peaks were observed towards the end of the C, compounds that do not appear in the master chromatogram. For instance, there is a well defined peak ahead of n-octane in the room temperature chromatogram of the feed. This peak is absent in the product chromatogram; therefore, it may be assumed to represent a cyclic compound. We tentatively identified it as l1trans-2,cis3,trans-4-tetramethylcyclopentane (b.p. 127.4' C,), In addition, two small peaks have been observed emerging a t room temperature after l,l,cis-3,cis-4tetramethylcyclopentane, and between 2,2dimethylheptane and 2,2,3-trimethylhexane. The first of these is believed to be a cyclane, the second may be an alkane (2,4dimethylheptane?). There were two additional unidentified peaks in this region of the 50' C. chromatogram. OLEFINS. The chromatogram of our master blend of 63 alkenes, obtained a t room temperature, is reproduced in Figure 3. The positions of the corre-
sponding normal alkanes, as well as those of the five cyclenes investigated, are also indicated in this master chromatogram. As with the saturates, blends of increasing complexity were used to obtain peak identifications for the congested areas. The 49 peaks in this chromatogram comprise 36 singlets, 12 doublets, and one triplet. Six of the doublets show partial separations which are satisfactory for qualitative identification, although not sufficiently pronounced for quantitative evaluation. With the C1 to Cs saturates the significant rearrangement in the sequence of emergence a t 50' C. was primarily due to the presence of two different hydrocarbon types, alkanes and cyclanes. Rerunning the alkenes master blend a t 50" C. proved to be less helpful. The sequence of emergence did not change and only two of the doublets of the room temperature chromatogram split to a calculable degree. Lowering the column temperature to 0' C., on the other hand, was considerably more useful. Two of the doublets that showed no separation in the room temperature chromatogram and three of the doublets Kith partial separation were resolved to a measurable extent; one component of the triplet also became measurable. The multiple peaks of the room temperature chromatogram are listed in Table 11; the extent of their resolution a t 0' and 50" C. is given in the last two columns of this table. The 0" and 50" C. runs take 120 and 31 minutes, respectively, to the end of the master blend (cis-2heptene), as compared with 55 minutes required by the 24" C. chromatogram. Retention Volumes. SATURATES. Expressed on a relative basis, retentions determined for the 88 compounds of our master blend a t 25' and 51" C. are listed in Table 111. VOL. 34, NO. 10, SEPTEMBER 1962
1229
W I COLUMN TEMP.
2M2P' 6
-
2 3 25'C.
Minulel f r o m inleclion
- ,
I
18
17
20
19
22
21
23
24
25
26
29
28
21
31
30
1
I
32
33
,
I
35
34
0
j
34DMc2P 34DMc2P-
7
7
inp%7
2YlHx.
7 7
I3Mp'
CAP*
f
7
3Y13Hr.
7
223DM2P. 3 D y
7
i
5M12nt'
rl
335 :
II
7
7
34DM
4MIUx.
12P.
,
36
,
37
L
,
,
,
39
38
40
v ,\ 41
Figure 3.
,
42
J
,
,
44
43
,
45
,
46
,
48
47
Chromatogram of master blend of
L
,
,
J
,
50
49
,
1
52
51
53 53
54 54
I
55 55
C, to C, alkenes
Cdumn. Glass capillary, 0.01 2 inch X 300 feet, coated with silicone Auld 96 Detection. Hydrogen flame ionization
As customary, gas holdup measured to the front of the methane peak was
mers likewise tend to follow straight lines. Isomers of simpler structures are located between these boundaries. 411 lines, except that of the geminally substituted cyclopentanes, have approximately the same slope. OLEFINS. The relative retentions for the 63 alkenes of the master blend are listed in Table IV. Data for the five cyclenes are also included. The slight type selectivity observed with the alkanes and cyclanes was also noted here. The usual plot of the relative retentions G'S. boiling points for all
used to adjust the observed retention volumes, since the flame ionization detector does not respond to air. Both the 25" and 51' C. relative retention volumes were reproducible to zk0.01. SF-96 is generally considered to be a fair "boiling point substrate." We observed a slight type selectivity, illustrated in Figure 4. The alkanes, as well as alkyl cyclopentanes and alkyl cyclohexanes, lie on straight lines; their complexly substituted iso-
compounds shows a fairly narrow band (Figure 5 ) . However, a retardation of the normal and monomethyl cis-2and cis-3-alkenes relative to the mono(except 2-methyl) and disubstituted 1-alkenes is discernible. The line of the trum-2-, trans-3-, and 1-alkenes is intermediate. Representatives of other configurations are scattered between the two boundary alkene lines. The five cyclenes investigated are distinctly outBide the_band of alkene points. Resolution of Racemates of 3,4Dimethylhexane. One of the -Wl
. C
1
I
n-ALOL CYCLOHEXANES
70
Figure 4.
I 10
I
90
I
I i 20
I
I30
Type selectivity of SF-96 for saturates at
25' C. 1230
,
too !IO BOILING POIXT. 'C.
ANALYTICAL CHEMISTRY
-_
0,2 20,
Figure 5. 25' C.
I .G
50
I LO BOlLlRG POIKT.
I 70
I 80
C
IO0
c.
Type selectivity of SF-96 for alkenes at
standard hydrocarbons, 3,4-dimethylhexane (API No. 240; claimed purity, 99.8%), gave a peak with a shoulder that amounted to 40y0 of the total area in the room temperature chromatogram. Identical pictures were obtained with samples from two other ampuls of this material, the last of them recently purchased and newly opened. At 2" C. column temperature, the separation became more prominent, but at 50" C. the shoulder moved completely under the main peak. As 3,4dimethylhexane contains two structurally identical asymmetric carbon atoms, i t is proposed that the two peaks from the SF-96 column represent a partial separation of its racemic (dl) and internally compensated meso forms. A similar observation involving the resolution of the two racemic forms of 2,3,4-trimethylhexane has been reported recently on squalane-coated stainless steel capillary by Simmons, Richardson, and Dvoretzky (IS). This latter pair also separates on SF-96, but 3,4-dimethylhexane may not be resolved on squalane between 0" and 50" C. Quantitative Analysis. 911 our area measurements were made b y planimeter (triplicates), and an averwas established. age precision of =t5yO This figure is similar t o those arrived at by J a d k in a systematic study of area measurements (8). Deviation of the determined area percentage values from the per cent by weight composition of the blends was, on the average, + 10%. Interestingly enough, there appeared to be a definite emergence time dependence; the relative sensitivities diminished with increasing emergence time, and this effect was greater with alkanes than with cyclopentanes and cyclohexanes. I n light of recent information (6), i t is obvious that a modified stream splitter design would improve our quantitative accuracy. APPLICATIONS
The application of the method is demonstrated by the analysis of two saturated and two olefinic samples. Platformer Feed and Product. A specific purpose of this work was to gain a better understanding of the composition of reformed products. I n catalytic reforming, cyclohexanes are directly dehydrogenated to aromatics, while cyclopentanes are first isomerized t o cyclohexanes. Excess of one or the other cyclane type in the reformed product can indicate deficiencies in the metal and acid functions of an experimental catalyst; also the type of substitution on the cyclanes influences the ease of conversion. Hence, i t is desirable not only to distinguish between five- and six-membered rings, but also to segregate each according to number and kind of substituents. Our method can furnish such a detailed
Table 111.
Adjusted Retention Volumes of Saturates Relative to n-Hexane at 25" and 51 O C.
Column. SF-96 on glass capillary Relative Retention Components in Order of Emergence At 51" C. B.P., O C. At 25' C. a t Room Temperature 0.01 0.005 - 161.49 Methane 0.02 0.01 - 88.63 Ethane 0.07 - 42.07 0.03 Pro ane 0.17 0.07 - 11.73 2-ethylpr propane 0.21 0.11 - 0.50 n-Butane 0.23 0.13 9.50 2,2-Dimethylpropane 0.36 0.26 27.85 2-Methylbutane 0.44 0.35 36.07 n-Pentane 0.59 0.51 49.74 2,2-Dimethylbutane 0.75 0.68 49.26 Cy clopentane 0.75 0.69 57.99 2,3-Dimethylbutane 0.77 60.27 0.72 2-Methylpentane 0.87 0.84 63.28 3-Methylpentane 1.00 1.00 68. 74 n-Hexane 1.23 1.28 79.20 2,2-Dimethylpentane 1.27 1.30 71.81 Methylcy clopentane 1.29 1.37 80.50 2,4-Dimethylpentane 1.37 1.43 80.87 2,2,3-Trimethylbutane 1.61 1.75 86.07 3,3-Dimethylpentane 1.68 1.78 80.74 Cyclohexane 1.77 90.05 2.02 2-Methylhexane 1.81 2.02 89.78 2,3-Dimethylpentane 1.87 2.06 87.85 1,l-Dimethylcyclopentane 1.91 2.19 91.85 3-Methylhexane 2.04 2.29 90.77 1,cis-3-Dimethylcyclopentane 2.09 91.72 2.35 1,trans-3-Dimethylcyclopentane 2.16 2.43 91.87 l,trans-2-Dimethylcyclopentane 2.09 93.48 2.43 3-Ethylpentane 2.16 99.2k 2.48 2,2,4-Trimethylpentane 2.35 2.82 98.43 n-Heptane 2.79 99.53 3.29 l,eis-2-Dimethylcyclopentane 2.81 3.29 100.93 Methylcyclohexane 2.81 106.47 3.29 2,2,3,3-Tetramethylbutane 2.88 3.42 104,89 1,1,3-Trimethylcyclopentane -2.84 3.49 106.84 2,2-Dimethylhexane 3.09 103.47 3.71 Ethylcyclopentane 3.14 3.81 109.84 2,2,3-Trimethylpentane 3.09 3.87 109.10 2,ELDimethylhexane 3.14 109.43 3.93 2,4-Dimethylhexane 3.31 4.06 109.29 1,trans-2,cis-4-Trimethylcyclopentane 3.34 4.14 111.97 3,3-Dimethylhexane 3.52 110.2 4.36 l,trans-2,cis-3-Trimethylcyclopentane 3.60 4.49 113.47 2,3,4Trirnethylpentae 3.73 4.61 114.76 2,3,3-Trimet hylpentane 3.98 4.93 113.73 1,1,2-Trimethylcyclopentane 3.93 115.61 5.09 2,3-Dimethylhexane 3.98 115.65 5.09 3-Ethyl-2-methylpentane 4.07 5.09 117.96 1,1,3,3-Tetramethylcydopentane 4.26 118.26 5.40 3-Ethyl-3-methylpentane 4.14 5.46 117.65 2-Methylheptane 4.19 5.53 117.71 4-Methylheptane 4.26 5.53 117.72 3,4-Dimethylhexane 4.34 5.53 116.73 l,cis-2,trans-PTrimethylcyclopentane 4.36 5.53 122.28 2,2,4,4-Tetramethylpentane 4.50 5.67 120.09 l,czs-3-Dimethylcyclohexane 4.45 5.74 121 . 5 1,1,~is-3,trans-PTetrarnethylcyclopentane 4.50 5.74 117.5 1,cis-2,trans-3-Trimethylcyclopentane 4.55 5.78 119.35 I,trans-4-Dirnethylcyclohexane 4.41 5.85 118.92 3-Methylheptane 4.45 5.88 118.53 3-Ethylhexane 4.76 6.02 119.54 1,l-Dimethylcyclohexane 4.90 6.37 120.8 1-Ethyl-trans-3-methylcyclopentane 5.04 6.37 118.79 Cy cloheptane 4.84 6.49 124,023 2,2,5-Trimethylhexane 4.99 6.49 121.4 1-Ethyl-cis-3-methylcyclopentane 5.04 121.2 6.56 1-Ethyl-trans-2-methylcyclopentane 5.13 6.62 121.52 1-Ethyl-1-methylcyclopentane 5.13 6.76 126.54 2,2,4-Trimethylhexane 5.25 6.76 123.42 1,trans-2-Dimethylcyclohexane 5.54 7.12 123.0 1,cis-2,cis-3-Trimethylcyclopentane 5.54 124.32 7.26 1,cis4Dimethylcyclohexane 5.59 124.45 7.26 l,trans-3-Dimethylcyclohexanc (Continued)
-
VOL 34, NO. 10, SEPTEMBER 1962
1231
Table 111.
(Confinued)
Column. SF-96 on glass capillary Components in Order of Emergence a t Room Temperature n-Octane Isopropylcyclopentane I, 1,cis-2-trans-4-Tetraniethylcyclopentane
1,l,cis-2,cis-4-Tetraniethylcyclopentane
2,4,4-Trimethylhexane 1,1,cis-3,cis-4-Tetramethylcyclopentane 1-Ethyl-cis-2-methylcyclopentane 2,2,3,4-Tetramethylpentane
2,3,5-Trimethylhexane 3-Ethyl-2,2-dimethylpentane I,cis-2-Dimethylcyclohexane
2,2-Dimethylhe tane 2,2,3-Trimethylffexane Ethylcyclohexane 1,cis3,cis-5-Trimethylcyclohexane n-Propylcyclopen tane
Table IV.
O
C.
125.66 126.42 129.4 129.4 130.65 133.0 128.05 130.02 131.34 133.83 129.73 132.69 133.60 131,78 138.41 130.95
Relative Retention At 51" C.
A t 25' C.
7.66 7.75 7.86 7.86 7.86
8.49 8.49
5.54 5.87 5.91 5.91 5.87 6.03 6.41 6.41
8.75 8.91 8.91 9.16 9.28 9.44 9.52 9.57
6.33 6.60 6.72 6.51 6.72 7.02 7.02 7.02
8.11
Adjusted Retention Volumes of Olefins Relative to n-Hexane at 24' C.
Column. SF-96 Componentg in Order of B.P., Relative Emergence C. Retention Ethylene -103.71 0.003 Pro ylene - 47.70 0.034 2-dethylpropene - 6 90 1-Butene - 6:26} 0'11 trans-2-Butene - 0.88 0.14 cis-2-Bu tene 0.18 3.72 3-Methyl-1-butene 0.22 20.06 I-Pentene 0.31 29.97 2-Methyl-1-butene 0.34 31.16 trans-2-Pentene 0.38 36.35 3,3-Dimethyl-l-butene 41.25 0.40 cis-2Pentene 0.41 36.94 2-Methyl-2-butene 0.44 38.57 Cyclopentene 0.60 44.24 4-Methyl-1-pentene 53.87 0.63 3-Methyl-1-pentene 54.18 0.64 2,d-Dimethyl-l-b~55.62 0.70 tene cis-4-Methyl-2-pentene 0.71 56.39 trans-4-Methyl-2entene 0.74 58.61 2-kethyl-1-pentene 62.11 0.88 1-Hexene 63.49 0.89 2-Ethyl-1-butene 64.68 0.99 n-Hexane 1.00 68.74 trans-3-Hexene 67.09 1.03 cis-3-Hexene 66.45 1.03 trans-2-Hexene 67.88 1.05 3-Methylcyclopentene 1.08 2-Met hyl-2-pentene 4,4-Dimethyl-1pentene trans-3-Methyl-21.11 pentene 67.70 72.52\ 4-Methylcyclopentene 65.67 1.12 cis-2-Hexene 68.89 1.16 cis-3-Me thyl-2-pen70.46 1.23 tene trans-4,4-Dimethyl2-pentene 76.74 1.31 3,3-Dimethyl-1pentene 77.48 1.36 2,3-Dimethyl-2butene 73.21 1.37 2,3,3-Trimethyl-lbutene 77.89 1.40
:;:;; 5
-
1232
B.P.,
ANALYTICAL CHEMISTRY
on glass capillary Components in Order of Emergence 3,CDimethyl-lpentene &-4,4Dimethvl-2pentene 2,4-Dimethyl-lpentene l-hlethylcyclopentene 3-Methyl-1-hexene 3-Ethyl-1- entene 2,4-Dimet&l-2entene 5-heth yl-1-hexene 2,3-Dimethyl-1pentene trans-2-Methyl-3hexene 4-Met hyl- 1-hexene cis-4-Methyl-2-hexene 2-Ethyl-3-methyl-lbutene trans-4-Methyl-2hexene trans-5-Methyl-2hexene Cyclohexene &-3,4-Dimethy1-2pentene %Methyl-1-hexene trans-3,4-Dimethyl2-pentene I-Heptene 2-Ethyl-1-pentene trans-3-Methyl-3hexene trans-3-Heptene cis-3-Heptene trans-3-Methyl-2hexene 2-Methyl-2-hexene n's-3-Methyl-3hexene trans-2-Heptene 3-Ethyl-2-pentene cis-3-Methyl-2hexene 2,3-Dimethyl-2pentene cis-2-Hep tene
B.P., Relative OC. Retention 1.52 1.53 1.60 1.63 1.68 1.73 1.77 1.80 1.87 86.31 86.371
1.92 1.97 2.02 2.08 2.25 2.42 2.46 2.50 2.65 2.76 2.84 2.84 2.87 2.90 2.95 2.97 3.07 3.19
account of the comporirnts distilling up to 133' C. This is shown by the analysis of thc C4 to 130" C. saturates portions of a platformer feed and composite reformed product. The range analyzed comprises nearly one half of the feed and four fifths of the product saturates. The samples were prepaied by isolating the aggregates of saturates from 0.5 liter of feed and product by displacempnt chromatography on silica gel, and distilling these aggregates to 130' C. The combined interpretation of the chromatograms obtained at room temperature and 50' C. yield the summary type compositions shown in Table
v.
Total cyclanes contents by mass spectrometric analysis and by the refractivity intercept method (7) are included for comparison. Gulf Heptenes. This material was obtained from Gulf Oil Corp. and is assumed t o be representative of the feedstock currently being used in the oxo process for the production of iso-octyl alcohols. Because the fluorescent indicator adsorption (FIA) analysis (a) showed a 4.4% saturates content (no aromatics), the sample was first fractionated by displacement development on silica gel. Results of the gas chromatographic analysis are given in Table VI. Ziegler C7 Olefins. This material represents a narrow distillation fraction of the product from propylene and 1-butene by the Ziegler process. It was analyzed without prior fractionation, as F I A analysis showed it t o be 100% olefinic. I n addition t o a few trace components, only two major peaks were found which, with the aid of the original 62-component master blend, were identified 2-methyl-lhexene (65%) and 1-heptene (33%). Production of such a high concentration of 1-heptene by this process is improbable. I n its place, the formation of 2-ethyl-l-pentene is expected. As 2-ethyl-1-pentene was one of the three C7alkenes missing from the original master blend, i t was immediately suspected that the 33% peak represented this compound, or perhaps its mixture with 1-heptene, rather than 1-heptene alone. They boil only 0.4' C. apart; therefore, their coemergence should not be unexpected. Mass spectrometry contributed some corroborating evidence. Although the mass spectrum of 2-ethyl-1-pentene was not available, it was conclusively established that, if one of the components of the Ziegler material is 2-methylI-hexene, the residual spectrum could not be due t o 1-heptene. Decisive proof was obtained by hydrogenating the sample and reanalyzing the product in the capillary column. The 2-methyl1-hexene content checked well with the 2-methylhexane concentration found in
Table V. Type Composition of the Cr to 130' C. Saturates ex Platformer Feed and Product
(Volume per cent) Method.of Cyclo- CycloAnalysis Alkanes pentanes hexanes Feed Gas chromatographya 51.7 21.2 27.1 Mass spectrometry 53.3 46 7 Refractivity intercept 52.4 47.6 Product Gas chromatographya 94.7 4.7 0.5 Mass spectrometry 95.7 4.3 Refractivity intercept 94.3 5.9 For unidentified CO compounds (4.3 and 3.0% by weight in feed and product, respectively) parsffin-naphthene ratio found for rest o sample was assumed. 0
Table VI.
Composition of Gulf Heptenes
Components
Peak Area, %
Olefins 4,4-Dimethyl-l- entene
0.3
trans-4,4~imetR~1-2-Den- -
tene 3,3-Dimethyl-l-pentene 2,3,3-Trimethyl-l-butene 3,CDimethyl-1-pen tene n's-4,4-Dimethyl-2-pentene 2,4-Dimethyl-l-pentene 3-Methyl-1-hexene 3-Ethyl-1-pentene 2,4Dimethyl-2-pentene 5-Methyl-1-hexene 2,3-Dimethyl-l-pentene trans-2-Methyl-3-hexene cis-4Methyl-2-hexene trans-4Methyl-2-hexene 2-Ethyl-3-methyl-1-butene trans-5-Methyl-2-hexene
2.6 0.8 0.4 0.4
0.1 7.5 Trace 0.2
1
20.7 7.2
1
0.4 3.3 1.7 4.9
1.1
cis-3,4-Dimethyl-2-pentene
2-Methyl-I-hexene 1 .o trans-3,4-Dimethyl-2-pentene 15.0 2-Ethyl-1-pentene 0.8 l-HeDtene trans:3-Methyl-3-hexene 1.2 trans-3-Heptene 0.4 cis-3-Heptene 1.9 trans-3-Methyl-2-hexene 2-Methyl-2-hexene 4.3 cis-3-Methyl-3-hexene 2.2 3-Ethyl-2-pentene 0.7 trans-2-Hep tene 0.2 n's-3-Methyl-2-hexene 2.7 2,3-Dimethyl-2-pentene 13.6 cis-2-Heptene Trace Total olefins 95.6 Saturates n-Hexane Trace ( ? ) 2,2-Dimethylpentane Trace 2,4Dimethylpentane 0.8 2,2,3-Trimethylbutane 0.2 2-Methylhexane 0.5 2,3-Dimethylpentane 2.5 3-Methylhexane 0.4 3-Ethylpentane Trace 2,2,4-Trimethylpentane Trace Total saturates 4.4
1
1
the hydrogenated sample. Since the 2heptene and 3-heptene peaks accounted for only 3 of the 5% n-heptane found in the hydrogenated sample, the unhydrogenerated sample had to contain about 2% 1-heptene. In fair agreement, the 3-methylhexane peak was 3% smaller than the corresponding peak of the Ziegler olefins. The coemergence of 2ethyl-1-pentene with 1-heptene was further confirmed in an addition experiment with 1-heptene. Cracking during hydrogenation was negligible, as only one small strange peak (0.2y0 2-methylpentane) was found in the product. DISCUSSION AND CONCLUSIONS
In developing a gas chromatographic scheme for the analysis of a complex system such as the homologous series discussed above, the initial problem is to arrive a t a substrate that provides the maximum resolution of all possible components. The next problem concerns the complete understanding of the resolution achieved, SF-96 appeared to be the best capillary column coating to fulfill the first requirement. This conclusion was reached after comparing it with squalane and n-hexadecane, the other two commonly employed substrates for hydrocarbon analysis. This stand may need to be reappraised, however, as a result of a recent report by Ahlberg and Walker (1). I n a 200-foot stainless steel capillary coated with nhexadecene, these investigators resolved a t 26" C. column temperature, 42 of the 44 components of a 30' to 114' C. boiling range gasoline. I n addition to all but one (2,2,3,3-tetramethylbutane) of the possible saturates of this range, the sample also contained benzene and toluene, but no olefins. Only one overlap occurred (3-methylhexane with l,cis-3-dimethylcyclopentane)which is a notably better performance than that of the SF-96 coating at a single temperature. For full comparison, however, the resolution obtainable for the 114' t o 133" C. range saturates, as well as information on the useful life of this column, will be needed. The combined interpretation of the chromatograms permits the quantitative analysis of 73 of the 88 compounds of the saturates master blend, and 51 of the 63 compounds of the alkenes master blend, as individuals, with the balance as doublets and one triplet. Although several of the doublets show a tendency for more complete resolution, which may be obtained on a longer column, further progress toward a complete component analysis of both ranges will be more readily achieved with the aid of an auxiliary column whose substrate either provides a different sequence of emergence or at least alters the spacing of peaks in the troublesome areas of the SF-96 chromatogram. n-Hexadecene
appears to be a promising candidate for this task. In actual cases the outlook for full component analysis is much brighter because all possible compounds of the range will not be present in every sample. Origin and history of prior treatment of the sample will be of great help in this respect. I n future applications some steps in the experimental work described for the platformer feed and product may be simplified or omitted without impairing the outcome of the analysis. First, it will suffice to separate only a few milliliters of saturates. Second, the fractionation by distillation of the saturates is omitted and the calculation of the C4 to n-propylcyclopentane range is based on an internal standard, Benzene is suitable, since it emerges as a singlet in both chromatograms. The column is rapidly purged by raising the oven temperature to 100' C. after the emergence of n-propylcyclopentane. I n the development of our methods, the interpretation of the fundamental chromatograms was based on blends of calibration compounds. This approach has been possible, since all but one of the C1to C8saturates and all but two of the C2 to C7 alkenes were available in reliable purity. This picture becomes less complete if the boiling point (and emergence time) overlap of the Cg saturates range is considered; a t least six additional unavailable Cgalkanes and cyclopentanes, above those incorporated in the master blend, are expected to emerge ahead of the terminal compound of the C8 range, n-propylcyclopentane. The chromatogram of the alkenes is more definitive. There is no boiling point overlap with the Cg compounds; however, the emergence of a few of the lower boiling C, isomers ahead of the terminal C7alkene, cis-2-heptene, cannot be excluded. Because of the lack of calibration compounds, the expansion of this scheme of analysis to include Cs and higher saturates or C8 and higher alkenes could not be contemplated. ACKNOWLEDGMENT
The authors are indebted t o F. M. Nelsen for development of the equipment, and to Kenneth W. Greenlee, director of APIRP 58A, for samples of four recently synthesized tetramethylcyclopentanes. LITERATURE CITED
( 1 ) Ahlberg, D. L., Walker, J. Q.,Hydrocarbon Process. Petrol. Rejiner 40, 338
(November 1961).
(2) Am. SOC.Testing Materials, "ASTM
Standards on Petroleum Products and Lubricants," D 1319-60T. (3) Desty, D. H., Goldup, A., Swanton, W. T.,ISA Proceedings, "1961 Inter-
VOL. 34, NO. 10, SEPTEMBER 1962
1233
national Gas Chromatography Symposiuni of Instrument Society of America,” p. 83, Michigan State
University, June 1961. Harsnape J. N., Whyman, B. H. F., ANAL. HEM. 32, 302 (1960). (5) Durrett, L. R., Simmons, M. C., Dvoretzky, I., Division of Petroleum Chemistry, 139th Meeting, ACS, St. Louis, Mo., March 1961. (4) Desty, D. H.,
(6) Eggertsen, F. T., Groennings, S., ANAL.CREM.30,20 (1958). (7) Groennings, S., ASTM Bull. No. 227, 64 (January 1958) [TP 30). ( 8 ) Jan&, J., J . Chromatog. 3,308 (1960). (9) Knight, H. S., ANAL.CHEM.30, 9 (1958). (10) Martin, R. L., Ibid., 32, 336 (19601. (11) Martin, R. L., Winters, J. C., Zbid., 31, 1954 (1959). (12) Porter, P. E., Deal, C. H., Stross,
F. H., J. Am. Chem. SOC.78, 2999 (1956). (13) Simmons, M. C., Richardson, D. B., Dvorypky, I., in “Gaa Chromatography 1960, R. P. W. Scott, ed., p. 211, Butterwortha, London, 1960. RECEIVEDfor review March 19, 1962. Accepted June 20, 1962. Division of +a1 tical Chemistry, Sym osium Honoring {. Zechmeister, 141st heeting, ACS, Washington, D. C., March 1963.
Gas Chromatographic Determination of Ethyl Alcohol in Blood for Medicolegal Purposes Separation of Other Volatiles from Blood or Aqueous Solution KENNETH
D. PARKER, CHARLES R.
FONTAN, JOHN L. YEE, and PAUL L. KIRK
School o f Criminology, University of California, Berkeley, Calif.
b A gas chromatographic method is described which, applicable to the medicolegal determination of the ethyl alcohol content of blood and aqueous solutions, offers advantages of rapid analysis, improved accuracy, simplicity, and specificity. Retention data for 56 volatiles indicate the resolution of the castor wax column used and illustrate its general utility for presumptive identification of volatile materials. Seven minutes were required to prepare and quantitate ethyl alcohol in a sample. Use of ethyl acetate as an internal standard obviated the necessity of precise measurement of blood samples. Standard blood and water samples with 0.023 to 0.1 80yoalcohol were analyzed with a precision of 2% and an accuracy of about 4%.
C
n E M I c h L and enzymic methods for the determination of ethyl alcohol in blood have been reviewed and evaluated by Lundquist ( 4 ) . Gas chromatographic methods have been described (1-3, 6),and their routine use has been attcmpted with varying degrees of success. The major problem has been the difficulty of dealing adequately with the high proportion of water in blood. In the use of both the thermalconductivity and the argon detectors, water lengthened the time required for a given analysis unless special chromatographic techniques (8) or special sample preparation ( 1 ) was employed to eliminate water. Chundela and Janak ( 2 ) employed 1-butanol and 2-butanone as internal standards, which avoided the necessity of accurate measurement of injected samples. Many liquid phases, operating parameters, injection methods, and chro-
1234
ANALYTICAL CHEMISTRY
matographic and sample preparation techniques were tested before selecting the described method. The hydrogen flame detector, castor wax column, and direct injection of prepared samples have advantages of rapid analysis, improved accuracy, simplicity, and specificity for routine determinations of ethyl alcohol in urine and blood samples. EXPERIMENTAL
Apparatus and Reagents. The HyFi gas chromatograph (Wilkens Instrument and Research Co., Walnut Creek, Calif.) with hydrogen flame ionization detector, Aerograph Model 600, and the Leeds & Northrup Speedomax H, 0- t o 1-mv. recorder, Model S, equipped with the Disc chart integrator Model 207, were employed. The Hamilton microsyringe of 1 4 capacity, Model 7001N, was used to inject blood and aqueous solutions. The chromatographic column was a stainless steel tube inch in o.d., 0.093 inch in i.d., 10 feet in length. It was pack’ed with 60- to 80-mesh Chromosorb W, acid-washed, coated with castor wax 40% by weight. The column was preconditioned a t 190’ C. for about 8 hours. The operating conditions were: injector temperature 170’ C., oven temperature 120’ f 1’ C., flow rate of carrier gas (nitrogen) 13.6 ml. per minute, flow rate of hydrogen 22 ml. per minute, attenuation X8, volume injected 1 pl. The blood, ethyl alcohol, ethyl acetate, and water used to prepare the stock standards were free of impurities when examined by the gas chromatograph a t high sensitivity. Aqueous ethyl alcohol standards were prepared by diluting 5 ml. of C.P. ethyl alcohol, 95 to 98%, to loo0 ml. with water. The ethyl alcohol
concentration of this standard was 0.360 f 0.006% (w./v.) determined by direct oxidation by acid dichromate. Volumetric dilutions of portions of this standard were made, providing the other aqueous ethyl alcohol standards used : 0.180, 0.135, 0,090, 0.068, 0.045, and 0.023% ethyl alcohol. Blood ethyl alcohol standards were made by diluting 5.00 ml. of the appropriate standard to 10.0 f 0.02 ml. with blood. The synthetic blood ethyl alcohol samples contained 0.180, 0.090, 0.068, 0.045, and 0.023’% ethyl alcohol. Ethyl acetate standards were prepared by diluting 15.0 ml. of C.P. ethyl acetate to 2000 ml., making the 0.676% ethyl acetate standard from which, by volumetric dilution, the other aqueous standards-0.338, 0.169, 0.127, 0.085, and 0.043% ethyl acetate (w./v.)-were prepared. Only the 0.16970 ethyl acetate standard was used in the method, however. The other ethyl acetate standards were used only to investigate the linearity of chromatographic response. Procedure. Blood and water, standard and unknown samples, were prepared serially by an identical procedure. The same ethyl acetate standard, serum bottles of the same size, and the same two pipets were used. Pipets, bottles, and liquids were a t room temperature. Using the volumetric pipet, exactly 1.00 ml. of 0.16970 ethyl acetate standard was serially delivered into each clean, dry, and labeled serum bottle and stoppered until the sample containing the ethyl alcohol was delivered. The sample, 1.00 ml. of blood or water, was then placed in the bottle, and the pipet waa washed with the mixed contents of the bottle. The stoppered and mixed prepared samples, kept a t room temperature, were used within 4 hours, and discarded. Although ethyl acetate hydrolyzes a t room temperature and a t the p H of blood,