(21) Martin, R. L., ANAL. CHEX. 33, 347 (1961). (22) Moore, S., Stein, W. H., Ann. Rev. Biochem.21, 521 (1952). (23) Pharmacopeia of the United States of America, 16th rev., p. 554, Mack Printing Co., Easton, Pa., 1960. (24) Rogers, L. B., Spitaer, J. C., ANAL. CHEM.33, 1959 (1961). (25) Schola, R. G., Brandt, W. W., 1961 International Gas Chromatography Symposium, Michigan State University, East Lansing, Mch., 1961.
(15) Hoel, P. G., “Introduction to Mathematical Statistics,” Wiley, New York, 1947. (16) Jost, W., “Diffusion,” p. 25, Academic Press, New York, 1960. (17) Keller, R. A., Stewart, G. H., J. Chromatog.9, 1 (1962). (18) Keulemans, A. I. M., “Gas Chromatography,” 2nd ed., p. 192, Reinhold, New York, 1959. (19) Martin, A. J. P., Ann. Rev. Biochem. 19, 517 (1950). (20) Martin, A. J. P., Endeavour 6, 21 (1947).
ROY-4. KELLER Department of Chemietry University of Arizona Tucson, Aria. GEORGE H. STEWART Department of Chemistry Gonaaga University Spokane, Wash. RECEIVED for review Bugust 10, 1962. Accepted September 27, 1962. Work supported in part by the National Institutes of Health, RG 7046 Bio (Cl).
Inverse Temperature Programming in Gas Chromatography SIR: Gas chromatography has been widely used for the analysis of hydrocarbon mixtures, and a variety of column packing materials and substrates have been applied to this purpose. Typical of the column substrates used in analyses of such compounds are long-chain saturated paraffins such as n-hexadecane ( 3 ) . I n this laboratory primary interest centers on analysis of hydrocarbons in the CZ through C6 range, and extensive use has been made of n-hexadecane columns. However, n-hexadecane has serious drawbacks as a partitioning agent, in that compounds beyond the C i s are retained for an excessive length of time, with the result that peaks for the C i s and C i s are broad and unsymmetrical. Moreover, the column cannot be heated, since hexadecane bleeds severely from the column above room temperature. Eicosane has been previously reported to be effective for hydrocarbon determinations @), and since this compound can be heated to some extent without destroying the column, it was felt that such a substrate might be useful for analyses of Cz to C6 hydrocarbons. Therefore an eicosane column consisting of 25y0 by weight eicosane on 30- to 60-
t s
mesh firebrick was prepared and evaluated, and some rather novel results have been obtained.
A mixture containing saturated and unsaturated Ca and Cs hydrocarbons, butane, isobutane, and n-hexane was used as a test mixture. The analyses were run on a Burrell K-2 gas chromatograph using a 2.5-meter column, with a helium carrier a t a flow rate of 45 cc. per minute. The thermal conductivity detector was operated a t a current of 180 ma. and maintained a t a temperature of 150” C. Resolution of the components in the mixture was extremely poor when the column was operated a t room temperature. Figure 1 shows a typical run, in which the column was heated after elution of the C4components in an attempt to reduce retention time for the nhexane. It is evident that this type of operation does not give a useful analysis. I n view of the resolution obtained for the lower hydrocarbons with n-hexadecane, the performance of the eicosane column was surprising. The only apparent difference between the two materials which might conceivably account for the large divergence in their per-
formance as substrates is their physical state a t room temperature, eicosane being a solid while n-hexadecane is a liquid. With this thought in mind, we heated the eicosane column to a point just above the melting point of the substrate (- 40’ C.). The sample mixture previously mentioned was injected and the components through the C1’s were nicely resolved, the longest retention time being of the order of 10 minutes. The hexane, however, remained on the column for a period comparable to that observed in the initial programmed run, and the peak shape was the same. I n a repeat run, the sample was again injected while the column temperature was about 40’ C., but after elution of the C4’s, the column was allowed to cool to room temperature. This resulted in elution of the hexane (plus some Csand c6 impurities) in less than 30 minutes, as shown in Figure 2. This procedure, which we have chosen to term ‘‘inverse programming,” clearly produces the most rapid analysis and the most satisfactory peak shapes. I n attempting to understand the mechanism by which the column functions, the possible significance of various terms included in the van Deemter H E T P equation has been considered for
* Li
I
::
0 ‘
0
Figure 1.
1838
Conventional program
ANALYTICAL CHEMISTRY
Figure 2.
Inverse program
isothermal operation of the column at 25" and 40' C. Kieselbach has given a modified form of this equation and a detailed discussion of the terms involved (1). If it is assumed that the partitioning agent is present on the packing material of the column as a solid at room temperature and as a liquid a t the elevated temperature used in these studies, a preliminary consideration would suggest that the diffusion coefficient of the gaseous sample into the liquid phase might be a controlling factor in determining the HETP. Proceeding with this idea, the WET? of the eicosane column \vas found to be 0.60 a t room temperature and 0.42 a t 40' C. for nbutane. Within the accuracy of our data this quantity may be given by HETP
=
.4 f B Dgo -f
:
UO
p--
(1
-
(pL?po)
where the terms conform to the nomen-
clature of Kieselbach (1). The terms A and B Dg,/% were determined experimentally under the two isothermal conditions and the various parameters evaluated for our conditions of flow rate, pressure, etc. These data then led t o the relationship
ions about the factors involved in column operation. The inverse programming method, however, appears t o be a useful analytical technique, and its extension t o other types of columns where feasible operating temperatures straddle the melting point of the substrate may improve column efficiency. LITERATURE CITED
I n view of published values for diffusion coefficients for analogous systems, this ratio appears too low t o attribute the change in column resolution primarily to the difference in the rate of diffusion of the gaseous sample into the . effecpartitioning phase. Since d ~ the tive film thickness, probably is not very different for these two conditions, the constant C, the coefficient of resistance to mass transfer in the partitioning agent, is apparently a n important factor in this phenomenon. The extent of the data obtained is insufficient to allow any further conclus-
(1) Kieselbach, R., ANAL. CHEM.33, 23
(1961).
(2) Meyer, R. A., in "Gas Chromatography,"~.93, V. J. Coates, H. J. Noebels,
I. S. Fagerson, eds., Academic Press, New York, 1958. (3) Taramasso, M., Ric. Sci. 2 6 , 887 (1956). T. 0. TIERNAN J. H. FUTRELL
Chemistry Research Laboratory Aeronautical Research Laboratories Office of Aerospace Research Wright-Patterson Air Force Base, Ohio
RECEIVEDfor review August 31, 1962. Accepted October 11, 1962.
Analysis of Ferrocenes by Mass Spectrometry SIR: Aromatic and olefinic compounds have been analyzed by low voltage mass spectrometry for several years (1-3). Attempts have been made recently a t this laboratory to apply the same technique to ferrocene compounds. At the outset, it was desired to determine whether the iron t o cyclopentadienyl ring bonding in ferrocene would be cleaved under the conditions used in low voltage analysis, thus yielding fragment ions rather than molecule ions, which are normally produced at low electron energies. Spectra of several substituted ferrocenes were obtained using a Consolidated Electrodynamics Corp. Model 21-103 mass spectrometer at a n inlet temperature of 350" C. and a nominal ionizing voltage of 8 e.v. After examining these spectra, it became apparent that in all cases very intense molecule ion peaks n-ere produced with little or no fragmentation. This fact would allow the determination of molecular weights of substituted ferrocenes for identification and qualitative determination of purity. Quantitative estimations could be made by a comparison of peak intenqities. However, accurate quantitative results could not be obtained because low voltage sensitivities had not been studied. This latter study is made difficult by the lack of a large number of very pure compounds. It is possible that the low voltage sensitivities of substituted ferrocenes depend upon the nature and posi-
tion of the substituent groups, as has been found with substituted benzenes (1).
Table I lists the compounds for which low voltage spectra were obtained. Also included are the mass number of the molecule ion containing the most abundant isotope and the intensity of this peak. Since unequal sample sizes were used in obtaining the various spectra,
Table
I.
Intensities of Ferrocene Molecule Ions
Intensity of molecule mle of
chart divi-
186 370 254
811.0 222.0 307.0 708.0 646.0 594.0 399.0
ion
Ferrocene Biferrocenyl 1,l'-Dichloroferrocene 1,l'-Diethylferrocene
242
1,l'-Di-n-butylferrocene 298
1,l'-Dibutyrylferrocene Trimethylsilylferrocene 1,l'-Bis( trimethylsily1)ferrocene Triphenylsilylferrocene Dimethylethoxysilylferrocene 1,l'-Bis( dimethylethoxysily1)ferrocene Vinyldimethylsilylferro.~
cene 1,l '-Bis(vinyldimethy1sily1)ferrocene
ion,
molecule
326 258
sions
330 444
2244.0 439.0
288
3740.0
390
3210.0
270
1226.0
354
436.0
the intensities observed do not indicate a measure of the sensitivities of the compounds. These data are included only to show that very large intensities can be obtained in this class of compounds. Compounds containing acid substituents [COOH, SOaH, B(OH)2] decomposed in the mass spectrometer and spectra could not be obtained. The use of mass spectrometry in analysis of ferrocenes was demonstrated by a n examination of the product from an ethylation of ferrocene. The product was found t o contain predominantly mono- and diethylferrocene and a small amount of triethylferrocene. Further, it appears that ferrocenes would be easily identified in the presence of other types of compounds which do not contain any iron, if use is made of the existence of the four iron isotopes having atomic weights of 54, 56, 57, and 58. LITERATURE CITED
(1) Crable, G. F., Kearns, G. L., Norris, M. S., ANAL.CHEM.32, 13 (1960).
(2) Field, F. H., Hastings, S. H., Ibid., 28, 1248 (1956). (3) Lumpkin, H. E., Ibid., 30, 321 (1958). DONALD J. CLANCY~ ILGVARS J. SPILNERS Gulf Research & Development Co. Pittsburgh 30, Pa. 1 Present address, Research Division, W. R. Grace & Co., Clarksville, Md. RECEIVED for review September 10, 1962. Accepted October 18, 1962. VOL 34, NO. 13, DECEMBER 1962
0
1839
