tassium, and cesium was tested. Onlv iron, COPPCr, and manganese nitrates were soluble to any appreciable extent, and these all yielded brightly colored solutions. ACKNOWLEDGMENT
The authors acknowledge the help of Julian Alvarez in preparing the drawings, and H. R. Bayne for checking the method.
LITERATURE CITED
(1) Flynn, K. F,, Glendenin, L. E., Phys. Rev. 116, 744 (1959). ( 3 ) Hawk, p. B., Oser, €3. L., Summerson,
R. H., “Practical Physiological Chemistry,” p. 644, Blakiston, New York, 1954. (3) Kolthoff, I. M., Sandell, E. B., “Textbook of Quantitative Inorganic Analysis,” p. 337, Macmillan, New York, 1952. (4)Lutwak, L., ANAL. CHEM. 31, 340 (1959). ( 5 ) .Packard, L. E., in “Liquid Scintillation Counting” C. G. Bell, F. N. Hayes,
eds., p. 63, Pergamon Press, London, 1958. (6) Packard Tri-Carb Liquid Scintillation Spectrometer Operation Manual, Packard Instrument Co., LaGrange, Ill. ( 7 ) Peng, C. T., . ~ K A L . CHEM.32, 1292 (1960). (8) Ronzio, A . R., Intern. J. A p p l . Radiation and Isotopes 4, 196 (1959). (9) Von Erdtmann, G., Hermann, G., 2. Elektrochem. 64, 1092 (1960). RECEIVED for review Kovember 24, 1961. Accepted March 9, 1962. Work waa partially supported by U. S. Public Health Service Grant D-877.
Effect of Preheater Contamination on Gas Chromatographic Analysis of Strongly Adsorbed Substances EDGAR
D. SMITH
and AUBREY B. GOSNELL
Graduate Institute of Technology, University of Arkansas, little Rock, Ark.
b
The presence of carbonaceous deposits in the preheater section of a gas chromatographic apparatus may lead to serious complications in the gas chromatographic analysis of polar organic compounds. The main effects of such a deposit are low detector responses and false peaks. Work with fatty acids and amines has shown that these effects may sometimes be of such a magnitude as to invalidate either qualitative or quantitative analyses. These effects can be eliminated by rigorous cleaning of the preheater section, but they gradually return as the carbon deposit again builds up through thermal decomposition of the samples injected.
the very first reported application of gas chromatography involved the analysis of organic acids ( 3 ) and bases (2, 4, there have been LTHOUGH
10-4
3 4
L 4
0 INJECTION
NUMBER
Figure 1. Typical plot of recorder response vs. injection number
0=
propionic acid injection numbers 1-8 and
21-31 0 = n-valeric acid injection numbers 9-20
646
ANALYTICAL CHEMISTRY
relatively few similar applications since these early papers. The major reasons for the lack of progress in these fields are discussed in recent articles by Hunter, Xg, and Pence (1) and by Smith and Radford ( 5 ) , and methods for overcoming these experimental difficulties are suggested. The present work is concerned with still another obstacle which must be surmounted before highly polar organic compounds can be analyzed satisfactorily by gas chromatography. It has been discovered that the preheater section of the chromatograph must be maintained free of carbon deposit when strongly adsorbed materials are to be analyzed. If this is not done, low and erratic detector responses, as \Tell as false responses, may be obtained. The false responses are apparently due to partial displacement of one or more components from previously injected samples which had been retained on the carbon deposit in the preheater. Numerous false responses may be obtained from a badly contaminated preheater, and accordingly this phenomenon has been called the repeater effect. The effects described are principally illustrated in this work using propionic and valeric acids, but similar effects have been observed with other polar organic compounds. These effects are particularly noticeable with acids and amines, but they have also been observed to a minor extent with alcohols and ketones. EXPERIMENTAL
-4 Perkin-Elmer Model 154-C Vapor Fractometer equipped with a thermistor detector was used. One meter stainApparatus and Reagents.
less steel columns were packed with t h e partition phases and substrates listed below, t h e partition phases being applied by the conventional slurry a n d evaporation techniques using acetone as the solvent. -411 columns were conditioned overnight a t 180’ C. in a stream of helium gas, and the columns were reweighed after conditioning to determine the final per cent of partition phase.
A. 12.4% LAC-296 on firebrick (regular 60/80 mesh, Wilkens Instrument and Research Inc.. Cat. Eo. XA-165). B. 15.5% LAC-296 on unactivated Celite (Perkin-Elmer Corp., Cat. No. 154-0048). C. 19.5% L4C-296 on Chromosorb-W (30/60 mesh, Johns-Manville Corp., Cat. S o . -1-472). D. 10% LAC-296 S% Hap04 on acid washed Chromosorb-R (Johns-
+
“I
h,e
I
I
S S C O L U M N ISSCOLUMN S S C O L U M N
Ill
p
2
1
’
I
Ill
INJECTION NUMBER
Figure 2. Recorder response v5. injection number interchanging two identical columns of packing D in stainless. steel
0 = propionic acid 0 = n-valeric acid
INJECTION NUMBER
Figure 3. Recorder response vs. injection number for identical packings in stainless steel and in copper
hfanville Corp., Cat. Xi). A-472, which had been leached with 6N HCl until the acid remained colorless, then thoroughly rinsed with distilled water). iv1 chemicals used in this work were reagent grade. The LAC-296 stationary phase is a polydiethylene glycol adipate obtained from Wilkens Instrument and Research, Inc. (Cat. S o . XA-155-0044). Procedure. Individual 10-111. Hamilton syringes were used for each of t h e acids studied to be sure t h a t the results were not confused by contamination of the syringes. The existence of the repeater effect was established by injecting a series of 1-p1. samples of propionic acid until a constant recorder response was obtained. Generally, about 10 samples were nrcessarp to achieve this end. At this point, repetitive samples of valeric acid were injected, and the recorder responses for both this acid and the displaced propionic acid were measured. This was continued until the valeric acid responses reached a constant value, a t which point the propionic acid response had been reduced to practically zero. Injections of propionic acid were then commenced and continued until constant responses were obtained for this acid again, the responses due to displaced valeric acid also being measured until they had been reduced to a negligible value. The data obtained were summarized by plotting injection numbers us. the peak area (peak height times half-band width) for each acid. Similar tests were carried out under modified conditions in order to clarify the cause of the repeater effect as described in the next section. DISCUSSIONS AND RESULTS
Figure 1 shows the results of a typical experiment in which the existence of the repeater effect was established. Similar results were obtained for all four of the columns tested showing that the column packing did not influence these results t o any significant degree. This same point was confirmed by the results in Figure 2. I n this series, two fresh stainless steel columns of packing D were employed. The first of these columns was used to demonstrate the repeater effect. When this was com-
pleted, propionic acid was injected until a constant recorder response was again obtained for this acid. At this point, the second fresh column (which had been kept a t oven temperature during the preceding tests) was quickly exchanged with the first, and injections of valeric acid were started. Repeater responses for propionic acid were obtained despite the fact that this second column had never been exposed.to this acid. When the repeater responses had been reduced to a very low value, the first column (“saturated” with propionic acid) was again placed into service, and the valeric acid injections were continued. The repeater responses for propionic acid continued to fall off in a regular way showing that these responses were independent of the previous exposure of this column packing to propionic acid.
‘01
INJECTION
NUMBER
Figure 4. Recorder response vs. injection number after mechanical cleaning of preheater
0=
propionic acid
0 = n-valeric acid
Figure 3 shows similar data for two identical packings-one in stainless steel and one in copper. After presaturation of the apparatus and stainless steel column with propionic acid, the copper column was substituted, and repeated injections of propionic acid showed no further change in the recorder response for this acid. Switching back to the stainless steel column gave the same result. Finally, the propionic acid was displaced by repeated injections of dichloroacetic acid which appeared to be no more effective than valeric acid in this respect. The peaks for dichloroacetic were not observed until the series was completed because of its very long retention time a t this temperature. The results described above shon-ed that the repeater responses were independent of the column and packing. The symmetry of the peaks and the reproducible retention times strongly indicated that they were due to acid retained a t the front of the column which was partially released each time a new sample was injected. An obvious explanation of these results was that
0
5
10 15 INJECTION NUMBER
20
Figure 5. Recorder response vs. injection number after oxygen purging of preheater
0=
propionic acid
0 = n-valeric acid
the acids were being adsorbed on a film of carbon laid down in the preheater zone by thermal decomposition of the many samples injected into this hot zone. This was confirmed by ~-isual inspection and by the results shown in Figures 4 and 5 . After mechanical cleaning of the preheater, most of the carbon deposit was removed and the repeater effect was nearly eliminated. Purging of the preheater zone with an oxygen stream overnight a t about 272” C. further reduced this effect to the point that it was barely detectable Finally, the effect was roughly reproduced by placing a small plug of glass wool contaminated m ith carbon in the front of a typical gas chromatographic column. In this case, horrever, the separation was impaired and low trailing peaks mere obtained for all of the acid. due to the relatirely strong adsorption
A.
INJECTION No. I CONTAMINATED PREHEATER
B.
INJECTION No. 5 CONTAMINATED PREHEATER
n INJECTION No. I CLEAN PREHEATER
D.
TIME
+
Figure 6. Typical chromatograms illustrating the effect of a contaminated preheater on the response of n-butyl amine VOL 34, NO. 6, MAY 1962
647
of the acids a t the column operating temperature. Thc repeater effect described in this work has been observed with a variety of polar organic compounds and with two gas chromntographic instruments. For example, Figure 6 shows the results obtained for n-butyl amine on a Model 20 Barber-Colman gas chromatograph. The first sample injected gave a relatively low response for the major component, and no response was noted for a minor impurity emerging just after this main peak. Subsequent injections
gave much better responses and clearly showed the presence of this impurity. injection of this same sample into a freshly cleaned preheated section gave a still higher response and better impurity resolution with the first sample injected. It s e e m certain that any material which is adsorbed appreciably by carbon (at the flash heater operating temperature) will show similar effects. Obviously, these effects are of fundamental importance to both users and manufacturers of gas chromatographic equipment.
LITERATURE CITED
(1) Hunter, I. R., Ng, Hawkins, Pence, J. W.,ANAL.CHEX 32, 1767 (1960). (2) James, A. T., Biochem. J . 5 2 , 242 i19.52 'i \ - - - -
(3) James, Ai.T., Martin, -4.J. P., Zbid.,
50, 679 (1952). (4)James, A. T., Martin, -4. J . P., Smith, G. H., Zbid., 5 2 , 238 (1952). (5) Smith. E. D.. Radford. R. D.. ANAL. CHEM.33, 1160 (1961).
RECEIVED for reviem December 1, 1961. Accepted March 12, 1962. Southwest Regional ACS Meeting, S e K Orleans La., Dee. 1961.
Separation of Organo- and Organobromoarsenic Compounds by Gas-Liquid Chromatography B. J. GUDZINOWICZ and H. F. MARTIN Monsanto Research Corp., Everett, Mass.
b This investigation demonstrates the feasibility of applying gas-liquid chromatography to the separation and quantitative analysis of organo- and organobromoarsines. Qualitative chromatographic data are presented for eight arsines varying in molecular weight from 156 (trivinylarsine) to 306 (triphenylarsine). Nearly linear relationships between the logarithm of the net retention time and either the boiling point or the molecular weight of each component of a homologous series are reported. Trivinylarsine and dimethylbromoarsine are used to demonstrate the applicability of gasliquid chromatography to the quantitative analysis of these compounds.
I
sensitive materials (10). Furthermore, various tetra-alkyltins and alkyltin halides were synthesized, and their gas chromatographic behaviors were studied (13). Gas chromatographic techniques were extensively employed by Brinckman and Stone ( 2 ) in the preparation of pure vinylhaloborane samples for spectrometric examination and vapor pressure measurements. In the past, these organo- and organobromoarsines have been char-
N RECENT YEARS, greater emphasis
has been put on gas chromatographic methods for the separation and purification of organo-metallic compounds. Abel, Sickless, and Pollard (1) reported the chromatographic separation of tetramethyl derivatives of the Group 11% metals (silicon, germanium, tin, and lead) and obtained linear relationships for the logarithm of the retention time us. either the molecular weight of the compound or its boiling point within a homologous series. Kaesz, Phillips, and Stone (9) resolved tin perfluoroalkyls with diisodecylphthalate on crushed firebrick, whereas Carbowax 4000 (Union Carbide polyethylene glycol) was the preferred liquid phase for the lead derivatives. i n the synthesis and cleavage of perfluorovinyltin compounds, instrumental modifications were cited for the separation and collection of these air-
648
ANALYTICAL CHEMISTRY
acterized by physical properties and elemental analyses (11, 1 2 ) . The purpose of this paper is to show that these arsenic derivatives can be analyzed by gas-liquid chromatography. QUALITATIVE STUDIES
9 modified Barber-Colnian Model 10 argon ionization detector chromatograph (5, 6) m s employed for the qualitative studies of organo- and organobromoarsenic mixtures. Using isothermal column operating conditions, the chromatograms for the separation of the arsenic derivatives investigated (Table 1) Fere obtained with a 6 foot by inch 0.d. stainless steel column packed Kith 5% by weight dimethyl silicone polymer (General Electric SE-30 silicone gum rubber) as liquid stationary phase on 80 to 100 mesh Chromosorb W.
Table 1.
Organo- and Organobromoarsines
JIolecBoiling ular Point, Compound Weight C. Dimethylbromc185 128-130/720 arsine mm . Methylethylbroino- 199 152-155 arsine Methylbutylbromo- 227 172-178/720 arsine mm . Tribut ylarsine 246 114/10 mm. Trivinylarsine 156 130 Triphenylarsine 306 Methyldibromo250 179-181/720 arsine mm . T'inyldibromoarsine 262 74-76/14 mm . O
1 0
1
3
I
1
6 9 12 15 RETENTION TIME, MINUTES
I 18
I 21
Figure 1. Gas chromatogram of ( 1 ) benzene, (2) trivinylarsine, (3) divinylbromoarsine, and (4) vinyldibromoarsine Operating conditions: column temperature, 80' C.; flash heater, 167' C.; detectortemperature, 235' C.; cell voltage, 1250 v.; argon flaw rate, 38 cc./minute