work suggests that a small turbulence contribution may be contributing to the high velocity fall-off of plate height under normal chromatographic conditions. I t is hoped that this problem will be considered further by others on both experimental and theoretical grounds. APPENDIX
Individual run data for the columns u.sed in this study are presented in Table V I . The specific properties of each of t h e columns used can be found in Table I1 t. LITERATURE CITED
(1) Bohemen, J., Purnell, J. H., J . Chem. ( L o n d o n ) 1961, 360. (2)T;Zarman, P. C., "Flow of Gases irough Porous Media", Academic Prc?ss, Kew York, 1956. ( 3 ) L)al Nogare, S., Chiu, J., ANAL. CH1OM. 34,890 (1962).
( 4 ) Dal Nogare, S., Juvet, R. S., Jr.,
"Gas Liquid Chromatography-Theory and Practice", p. 38, Interscience, New York, 1962. ( 5 ) DeFord, D. D., Loyd, R. J., Ayers, B. O., ANAL.CHEM.35,426 (1963). (6) Dorr, Frank, "Gas Measurement and Flow Efficiency Using Gas Tracer Technique," Summer Seminar for Gas Technology, Institute of Gas Technology, Chicago, Illinois, Au ust 1963. ( 7 ) Giddings, J. C., ANAL. 34,
HEM.
1186 (1962). ( 8 ) Zbid., 35, 1338 (1963). ( 9 ) Zbid., p. 353. (10) Zbid., p. 439. (11) Giddings, J. C., Robison, R. A,, Ibid., 34, 885 (1962). (12) Giddings, J. C., Seager, S. L., J . Chem. Phys. 33, 1579 (1960). (13) Giddings, J. C., Seager, S. L. Stucki, L. R., Stewart, G. H., ANAL.CHEM.32, 867 (1960). (14) Golay, M.,J. E., in "Gas Chromatography 1958, D. H. Desty, ed., pp. 36-55, Butterworths, London, 1958. (15) Halasz, I., Heine, E., Nature 194, 971 (1962).
(16) Heine, E., Doctoral Dissertation, Univ. of Frankfort (1963). R., 'ANAL. CHEM. 35,
McLaren, L., Zbid., 35 7 McLaren, L., I b i d . , 36,
1477 119641. '(20) Norem, S. D., Ibid., 34, 40 (1962). (21) Seibel, A,, "A High Temperature
Hydrogen Flame Ionization Detector," 15th Pittsburgh Conference on Analytical Chemistry and Applied Spectroscopy, March 2, 1964. (22) Sternberg, J. C., Poulson, R. E., ANAL.CHEM.36, 58 (1964). (23) Sternberg, J. C., Stillo, H. S., Schwendeman, R. H., Ibid., 32, 85 (1960). (24) van Deemter, J. J., Zuiderweg, F. J., Klinkenberg, A., Chem. Eng. Sci. 5, 271-289 (1956).
RECEIVEDfor review January 27, 1964. Accepted April 13, 1964. 2nd International Symposium on Advances in Gas Chromatography, University of Houston, Houston, Texas, March 23-26, 1964.
Wide-Range Programmed Temperature Gas Chromatography in the Separation of Very Corn lpIex Mixtures CHARLES M.ERRITT, Jr., J. T. WALSH, D. A. FORSS, P I 0 ANGELINI, and
S. M. SWIFT
Pioneering kesearch Division, U . S. Army Natick laboratories, Natick, Mass.
b The rangfe of programmed cryogenic tem p e ra tur e gas chromatography has been widened to include to over temperatures from -196" -,C200" C. Various and variable rates of temper'ature rise may b e em,Dloyed and automatic programming can b e provided. The necessity for empiloying very low starting temperatures for mixtures containing very volatiire components is demonstrated, and a i l example of a separation of a mixture%of coinpounds having a boiling range IFrom - 161 " to f200" C. is given. The technique has been applied tcb the total analysis of the carbon a'ioxide, center cut, and water fractions of the volatile compounds from irra$diated beef. The efficacy of the selwrations enhances the use c>f a rapid scanning mass spectrometer ft3r identifi cation of the components in the eluatbs.
I
ana lysis of the volatile corn,?ounds is, d a t e d from natural products, ari extre mely complex variety of mixtb res is encountered by the invest igator \v ho must separate and ' h : THE
1,502
ANAL' 7TICAL CHEMISTRY
identify the components of such mixtures. I n this laboratory, for example, quantitative analyses are required of such diverse mixtures as the compounds isolated from fresh ground coffee, deteriorating fish, irradiated meat, oxidized fat, microbially decomposed fuels, insect secretions, and human sweat. Such mixtures have several features in common. They contain a great many components encompassing many different types of compounds. Some of the components are present in appreciable amount; others, only as traces. The relative volatility of the compounds may vary from substances which are fixed gases to compounds having boiling points above 200" C. The use of programmed temperature gas chromatography is essential for the separation of such mixtures. Programmed temperature gas chromatography commencing a t room temperature or above has been found inadequate for the effective separation of the low boiling constituents of many of the mixtures encountered. The development of a technique for separating multicomponent mixtures of highly volatile compounds by pro-
grammed temperature gas chromatography in the low temperature rangei.e., ibelow room temperature-was described ( 4 ) , its effectiveness was demonstrated, and the apparatus for programming from -80" C. to room temperature was described. The range of programmed temperature gas chromatography ha3 now been extended to include temperatures from - 196' to over $.200" C. This papel shows the appreciable improvement iri the separation of very complex mixtures when programming is carried out over a very wide range. EXPERIMENTAL
Apparatus. Most of the data presented in this paper were obtained on modified commercial apparatus. Two different setups were employed. The Eiarber-Colman Series 5000 modular system for subambient temperature programming was used without modification for programming from -65" to f200" C. Liquid nitrogen was used to cool the column chamber for chrom;atograms requiring starting temperatures below -65" C. Another
130
-100
-50
50
0
TEMPERATURE I
I
Figure 1. Column.
150 CENTIGRADE I
I
I
20 RETENTION
0
IO0 DEGREES 40
TIME
MINUTES
Chromatograms of “hydrocarbon mixture” obtained using different temperature ‘programs
5 % SE-30 on 80- to 100-mesh Chromosorb W , Barber-Colman apparatus.
Log recorder response used to provide attenuotion for more
abundant components
apparatus used the Wilkens Instrument Co. ALerographModel 50013 Hy-Fi c.olumn oven as the column chamber and an F & h1 Scientific Co. Model 240 programmer, modified for subambient programming as described by Robertcon, Issenberg, and Nerritt (6) Liquid nitrogen 01 powdered d r j ice n a s used to cool the column chamber t o the appropriate starting temperature Spontaneous narming of the column (,hamher (4) waq used in the liquid nitrogen to dry ice temperature range and the modified F & AI 240 programmer wab used to control the tempeiature of the column chamber above -80” C -4 conventional .\erograph Hy-Fi gas chromatograph mas ubed t o pro\ide a flame detector, ~ i t the h column oven serving to pieheat the eluate from the column before entering the detector
Procedures. Details concerning the columns employed, operating parameters, and nature of the samples studied are given mainly with t’he accompanying figures. Some procedures, however, were generally used throughout. ALllcolumn3 were stainless steel, 10 feet x ‘/g inch. All samples were swept onto the column from gas-sampling devices described in detail by Bazinet and Walsh ( 1 ) and by Forss, Bazinet, and Swift ( 2 ) . Lab gas used for very low temperature studies was obtained from the bench tap, but to ensure uniform composition, a single large volume was collect,ed, the methane content was reduced by vacuum distillation a t - 196’ C., and small aliquots were t’aken. The gas was described by the local supplier as “natural gas, fortified with cracked
petroleum.” The various samples of volatile compounds from irradiated beef were obtained according to procedures described by Merritt et al. ( 3 ) . The “hydrocarbon mixture” was prepared by combining a commercial white gasoline with a petroleum ether (b.1). 30’ to 60” C.), in which a mixt,ure of light gases, including met,hane, ethane, and propane, was dissolved. This mixture was prepared to provide a widely boiling mixture. RESULTS A N D DISCUSSION
Although the improved separation obtained with subambient programming was demonstrated for the analysis of complex mixtures of volatile compounds isolated from coffee arid irVOL. 3 6 , NO. 8, JULY 1 9 6 4
1503
radiated beef ( 4 ) , the necessity for wide-range programming of the temperature is not readily apparent. Comparison of the three chromatograms in Figure 1 demonstrates the obvious advantage of starting the program below -80" C. and completing it a t a temperature above the elution temperature of the component having' the highest boiling point. The sample is the "hydrocarbon mixture" whose most volatile component is methane and whose least volatile component has a boiling point above 200" C. The chromatogram in the upper portion of t'he figure was obtained using a conventional program of 5" per minute with a starting temperature of 25" C. The entire inixt'ure was very rapidly eluted
from the column and very little separation was obtained. The chromatogram in the central portion was obtained by programming from -130" to 25" C. a t a program rate of 5' per minute, and then holding a t 25" C. The lower boiling, more volatile constituents of the mixture are well separated, but the higher boiling members tend to be eluted with poor efficiency and very marked loss of resolution a t the isothermal end of the chromatogram. I n the lowest chromatogram the column was heated from -130" to +150° C. a t 5" per minute. More than 55 components can easily be discerned in this mixture. Although on the whole the separation of the mixture on this SE30 column is good, the T !ry low boilitig
members of the mixture--e.g., the first two componentr-are resolved poorly. The third peak is very broad. More effective separatioi may be obtained on polar columns such as diethylene glycol succinate or 1,2,3-tris(2-cyanoethoxy)propane. Figure 2 compares the separations on three types of liquid phase. The samples are equal aliquots of lab gas. The top chro natogram shows an excellent separation obtained on tris(cyanoethoxy)propane. There is a loss of column efficiency on SE-30 at low temperature, as seen in the center chromatogram, but its overall resolving power is better. There is loss of both efficiency and resolution on the cyanoethyl silicone-diethylene glycol suc-
TR S
SE-30
w tn
z
0
a v, w U
11:
w n U 0 0
w
a
ECNSS-S
MINUTES Figure 2. TRIS.
SE-30.
Behavior of lab gas aliquots on three columns
570 1,2,3,-tris(2-~yanoethaxy)prapanean 60- to 80-mesh firebrick
5% methyl silicone polymer on 8 0 - to 100-mesh Chromosorb W
ECNSS-S.
370 cyanoethyl silicone-ethylene glycol succinate copolymer an 8 0 - to 1 00-mesh Gas Chrom P
Temperature programmed from -170' Aerograph,
1504
F&
ANALYTICAL CHEMISTRY
M apparatus
C. a t -3'
C./minute
w v)
z 0
a tn w a
a
w n
a
0 0
w
a
-170
-I 35
- 80
-25
0 50 C E NTIGRADE
TEMPERATURE
I
0
30
I
I
60
90
TIME Figure 3.
100
MINUTES
Programmed temperature chromatograms of lab gas starting at different temperatures Column.
5% TRlS on 60- to 80-mesh firebrick
Temperature program.
m S o C./minute
Aerogroph, F & M apparatus
cinate copolymer, as shown in the bottom chromatogram. Several liquid phases have been evaluated for use a t very low temperatures and, in general, most polar phases seem to work well. Among those which have been found satisfactory are tris(cyanoethoxy)propane, oxydipropionitrile, diethylene glycol succinate, Carbowax 20M, and Carbowax 4000. Among the nonpolar phases which have been found satisfactory are squalane, Apiezon 11, silicone SF-96, and, to a limited extent, SE-30. .In important quePtion is whether it
is necessary or desirable to go to extreme low temperatures; in particular, whether or not it is worth while programming below dry ice temperature. The separations obtained on a tris(cyanoethoxy)propane column of three identical aliquots of lab gas are shown in Figure 3. I n each case the temperature program rate was 5" per minute, but the starting temperature a t the top was O", in the center -80'. and a t the bottom -170" C. The very low starting temperature is seen to be necessary for complete resolution of the mixture.
Low temperature gas chromatography is very important, in the analysis of natural pr0duct.s where often the major portion of the volatile material isolated is water and carbon dioxide and only trace amounts of the flavor and odor compounds are present. Thus, in isolating the off-odor product,s in irradiated beef, it is necessary to carry out' a preliminary separation to remove the gross amount': of water and carbon dioxide ( 3 ) . This is usually done by low temperature-high vacuum distillation, followed by gas chromatographic separation of the components in the two VOL. 36, NO. 8, JULY 1964
1505
I
I
W v)
z
0
a v)
W
a
U W
n
a
0 0 W
a
I
I
-170
-I 35
I
0
I
I
-25 TEMPERATURE
-6 5
I
I
0
60
I
I20 C E N T I GRADE
i
I
I
30
60
90
TIME
MINUTES
Figure 4. Programmed cryogenic temperature gas chromatograms of carbon dioxide and center cut fractions from irradiated beef volatiles Column. 5% TRlS on 60- to 80-mesh firebrick Aerograph, F & M opporotus
fractions. The choice of temperature range for these vacuum separations is vital. The situation is illustrated by chromatograms shown in Figure 4. The upper chromatogram shows the separation obtained when a COz fraction-Le., the distillate obtained below - 140' C.-is programmed from - 170" C. I n this case, although the sample is mainly carbon dioxide, the separation of the trace components may readily be observed. Since this fraction is separated by vacuum distillation a t a temperature of - 140' C. and a pressure of torr, only the most volatile compounds of the initial sample are present here. On a low temperature program such as this they are all separated and eluted by the time the column temperature has reached - 80' C. The lower chromatogram shows the separation of the center cut fraction1506
ANALYTICAL CHEMISTRY
that is, the material distilled between -140' and -80" C. a t torr. This fraction was programmed from -170' C. and no components, or very few components, were eluted below -80' C. The components of the center cut are separated and eluted above -80' C. Thus, while a very low temperature program is required to separate the compounds in the COz fraction, a program starting a t -80' C. would separate the compounds in the center cut. The amount of overlap between these two chromatograms indicates the effectiveness of the preliminary separation by vacuum distillation. A new technique employing evaporation of a solvent extract onto a cold column has recently been developed for isolating and separating the volatile flavor compounds present in the very
dilute aqueous solutions obtained as water fractions (3). Although described elsewhere in detail (Z), it should be noted here that since the solvent extract method now permits analysis of the water fraction and since the low boiling compounds in the C02 and center cut fractions can readily be separated from water by vacuum distillation a t -80' c. and analyzed by low temperature gas chromatography, complete identification of flavor compounds in complex systems is now possible. When the system is not complicated by the presence of large excesses of water or carbon dioxide, widerange temperature programming may be employed directly-for example, in the analysis of the rancid odor compounds present in oxidized butter oil. Cryogenic separations are reproducible. This is shown by the two
w v)
z
0
a v)
w
a
[L
w
n
a
0
u w
a
I
1
-170
-I 35
I
I
-6 5 -25 TE M PER ATU RE
I
60
TIME Figure 5. aliquots
I
I20 C E NT I GRADE
60
30
0
I
0
90 MINUTES
Duplicate programmed cryogenic temperature gas chromatograms of irradiated beef center cut fraction Column. 5% TRlS on 60-to 80-mesh flrebrick Aerograph, F & M apparatus
chromatograms of aliquots of irradiated beef center cut volatiles programmed from - 170' C. on tris(cyanoeth0xy)propane (Figure 5). The temperature rise rate for the upper chromatogram is slightly faster than the lower. The ultimate goal of the analysis of natural products is usually to identify the compounds isolated. Gas chromatography separates the compounds. 111this laboratory, identification is then accomplished by means of a rapid scanning mass spectrometer. The mass spectra of the components of the mixture are obtained as they are eluted from the gas chromatograph. The quality of the gas chromatographic separation, therefore, is reflected in the quality of the mass spectra obtained. The use of subambient and wide-range temperature programming has been of tremendous value in providing the type of separation required to get definitive mass spectra of well separated component s.
Figure 6 shows some examples of mass spectra of compounds isolated from a center cut fraction of the volatiles in irradiated ground beef. .4t -60' C. and 2 minutes after the start of the chromatogram, the compound being eluted into the ion source of the mass spectrometer is n-butane. Half a minute later and at -58' C. the butane has disappeared and the spectrum seen is that of n-butene-1. This spectrum shows no residual interferences from butane and demonstrates a clearcut separation. Another example is seen at 18.5 minutes, when the temperature has risen to -25' C. and the mass spectrum shows that the compound being eluted is n-heptane. A minute later, another peak is eluted which the mass spectrum indicates is a mixture of two compounds not separated on the chromatographic column. In this case, however, heptene-1 and dimethyl sulfide are both identified in thc mixture from
their contributions to the mass spectrum. n-Heptane and n-octane a r e separated by more than 10 minutes. Moreover, it is interesting to note the relatively low temperature a t which these hydrocarbons are eluted, nheptane a t -25' C. and n-octane at -8' C. Elution of compound.; a t lower temperatures reduces the possibility of decomposition on the column. I n addition to its usefulness as a means of separating mixtures with a wide range of boiling points, wide-range temperature programming also permits a more general approach to the separation of a variety of types of mixtures. Thus, the same column may be used to separate a mixture of light hydrocarbons or fixed gases and for a mixture of high boiling fatty acid esters. Only the selection of a11 appropriate temperature program is required. This great versatility of the method not only greatly enhances the usefulness of gas chroVOL. 36, NO. 8, JULY 1964
e
1507
n-bu t a n e
41 43
TIME
TEMP.
2.0min.
-60' C.
2.5
-58
18.5
-2 5
58
butene -I A.. 41 43
56
n- hep t one
41
29
27
41
1
-
5557
47
55
71
62
70
84
83
100
TO
-22
29.0
-8
98100
n - o c t one
MASS
19.5
CHARGE
RATIO
Figure 6. Time-of-flight mass spectra of some components eluted from programmed cryogenic temperature gas chromatogram of an irradiated beef center cut fraction Column. 5% P,Poxydipropionitrile on 80- to 1 00-mesh firebrick to + 3 0 ° C. at 3'/minute Temperature program. -65'
matography itself, but also permits greater utilization of ancillary methods such as rapid scanning mass spectrometry. ACKNOWLEDGMENT
The authors are indebted to hl. L. Bazinet and Phillip Issenberg for their contributions to the experimental work. 1508
ANALYTICAL CHEMISTRY
LITERATURE CITED
(1) Baxinet, ?*I. L., Walsh, J. T., Rev.
sei. Instr. 31, 346 (1960).
( 2 ) Forss, D. A , , Bazinet, M. L., Swift, S.M., J . Gas. Chromatog. 2 , 134 (1964). (3) Merritt, C., Jr., Bresnick, S. R.,
Bazinet, M. L., Walsh, J. T., Angelini, P., J . Agr. Food Chem. 7, 784 (1959). (4) hlerritt, C., Jr., Walsh, J. T., AKAL. CHEM.35, 110 (1963).
(5) Robertson, 11 H , Issenberg, P , Merritt, C Jr "Facts and Methods for Scientdc Risearch, Vol 5 , T o 1, F & 11 Scientific. Co , Avondale, Pa 1964 RECEI~ED for rewen April 3, 1961 Accepted Aprd 30, 1961 2nd International S\mposium on -4dvances in Gas Chroniatographj , UniversltT of Houston, Houston, Texas, March 23 to 26, 1964
