19fi9
Anal. Chem. 1981, 5 3 , 1959-1961
change at the end point is clearly detected even in the small volume of 1,2-dichloroethane. The use of the one-phase end point change method and a small volume of the organic solvent allowed the sensitive titration of cationic surfactants with a diluted titrant such as 5 :< M tetraphenylborate solution. The conventional two-phase dye transfer methods did not give a sharp end point in any range of concentration when a small volume of organic solvent was used. Larger amounts of buffer solution had no influence on the titration, but amounts of' less than 1mL caused problems in phase separation. The tiiter was constant when the pH of the aqueous phase lay within the range 4.5-8.5. The effects of water-immiscible solvents on the titration were tested: nitrobenzene, methyl isobutyl ketone, butyl acetate, icioamyl alcohol, 1,2-dichloroethane, chloroform, toluene, benzene, carbon tetrachloride, chlorobenzene, and n-hexane. Of these, 1,2-dichloroethane was the best solvent for the titration of cationic surfactant. Initial volume fluctuations of the aqueous layer (10-25 mL) and the organic layer (0.5-1.5 mL) lhad no influence on the determination of the titration end point. The results of titrations of Zephiramine at several different concentrations, shown in Table I, indicate the high sensitivity of the proposed titrimetric method. The relative error was 1-2% when 5 mlL of cationic detergent was titrated by using this method. The effect of several salts on the titration of 5 mL of 5 X lo-&M cetylpyridinium chloride indicated that the following ions did not, interfere at the 0.001 M levell: Na+, K+, Ca2+, Ba2+, Mg2+, fU3+, NO8-, I-, C1-, Br-, S o t - , and acetate. A sharp end point was obtained even in the presence of precipitates such as ferric hydroxide. Clay minerals (colloidal dispersion) such as montmorillonite, however, caused negative errors and such i3 phenomenon as Nachblauen at the end point slightly. It is reasonable to consider that cationic surfactants are adsorbed or cation exchanged by the colloidal
clay. Anionic surfactants also caused negative errors in excess of maximum permissible concentration of lo4 M. The proposed method was applied to the determination of cationic surfactant which had been used as a disinfectant iin an infirmary. The waste water and drainage were centrifuged to remove the turbidity. The turbidity would be due to the presence of fine clay or organic matter. The results obtained are shown in Table 11. The recovery tests of the Benzethonium were also carried out by adding known amounts of Benzethonium and determining the Benzethonium according to the procedure. The recoveries of Benzethonium weire 99-101%. When the sample solutions were not centrifuged, the recoveries were about 93-101%. Table I1 also shows a comparison of the proposed method with the official titrimetric method (8) using Methyl Orange as an indicator. The proposed method is not troublesome and not far behind the instrumental method such as photometric titration and spectrophotometry in the sensitivity.
LIlrERATURE CITED Cross, J. T. "Cationic Surfactants"; Jungermann, E., Ed.; Marcel Dekker: New York, 19'70; Chapter 13. Hummel, D. "Identlflcation and Analysis of Surface-Active Agents;"; Interscience: New York, 1962; Vol. 1. Mohammed, Hussain Y.; Cantwell, Frederick F. Anal. Chem. 1980, 52, 553-557. Carkhuff, E. D.; Boyd, F. J. Am. Pharm. Assoc. 1954, 43, 240-241; Chem. Absfr. 1954, 48, 72591. Johnson, C. A.; Klng, R. E. J. Pharm. Pharmacal. 1983, 15, 584588; Chem. Absfr. I S M , 59, 11190a. IModin, Rolf; Schill, Goran Tabnta 1974, 21, 90!5Jansson, Sven, 0.; 918; Chem. Abstr. 1975, 82, 10880~. Sakai, T. Bunsekl Kagaku 1978, 27, 444-447. "The Japanese Pharmacopoeia", 9th ed.;Hlrokawa Publishing: Tokyo, 1976; p '2-317.
RECEIVED for review February 17, 1981. Accepted June 18, 1981.
Capillary Radiogas Chromatography Lauren A. Ernst, Gary T. Emmons, John D. Naworal, and Iain M. Campbell" Department of Biological Sciences, University of Plffsburgh, Parran Hall, 130 DeSoto Street, Piffslxrgh, Pennsylvania 1526 1
Radiogas chromatography (RGC) allows components of mixtures to be separated and then assayed simultaneously for amount and radioisotope content. Since its introduction in the middle 1950s (I), RGC technology has been improved significantly. Modern units, based on gas flow proportional counting and having the option to use a mass spectrometer as the mass detector (see ref 21, are efficient in their radioisotope counting (80% fo:r 14C)and are sensitive in their mass detection (10-13 g or less, if the mass spectrometer is used in the selected ion monitoring mode ( 3 ) ) . Where complex mixtures are being studied, however, component resolution by the packed gas chromatographic columns that are most commonly used, can limit the technology's effectiveness. The obvious answer to this resolution problem is to resort to capillary gas chromatographic columns and the increased theoretical plate counts that such open tubular columns provide. Hamnett and IPratt ( 4 ) and Gross et al. (5) have already demonstrated the potential of such columns in RGC. The former collected the effluent from a capillary gas chromatographic column in 12-sintervals and counted the fractions in a liquid scintillation counter. By opting for off-line radio isotope assay (presumably in a total time frame that was long compared to the GC run time), Hamnett and Pratt avoided the difficulties of measuring small amounts of radioactivity 0003-2700/81/0353-1959$01.25/0
in a flowing gas stream. Each fraction could be counted for the amount of time needed to secure acceptable statistics, background levels and counting efficiencies being those expected in scintillation counting, i.e., 20-25 counts/min and 80% for 14C,respectively. Gross et al. used an on-line capillary RGC unit. Their system comprised a low dead-volume splitter which directed part of the effluent gas of a capillary GC column to a flame ionization detector and allowed the remainder to enter a custom-built cobaltous oxide combustion furnace (all organics COJ. The 14Cconlmt of the COz produced was monitored with a custom-built anthracene scintillation cell (approximate efficiency, 40%). Their system could detect 400 counts/min. In this report we describe the construction and performance of an on-line capillary RGC unit in which the counter is a commercial gas flow proportional counter.
-+
EXPERIMENTAL SECTION Incorporation of a Capillary Column into a Packard 7400 Gas Chromatograph. A radiogas chromatograph comprising a Packard Model 7400 gas chromatograph and a Packard Model 894 gas flow proportional counter was used in this study. When this unit is used with packed columns, the effluent from the gars chromatographic coluinn is split in the detector oven to allow approximately 10% of the gas flow to pass to an hydrogen flame @ 1981 American Chemical Ejoclety
I960
ANALYTICAL CHEMISTRY, VOL. 53, NO. 12, OCTOBER 1981
DETECTOR
INJECTOR
n
. B b
m -I m
0
-I 0 XI
E
cz
XI
0
m
v)
3 3
77 0
2
z v)
m
IJ
I
C-
5-b
scale: I "
Design of detector (left)and injector (right) fittings. Code: in. nickel tubing; b, graphite ferrule; c, -+ in. stainless a, steel Swagelok reducing union; c', as in c wlth hole drilled in the side to admit '/I6 in. nickel tubing; d, silver solder joint; e, in. nickel tubing; f, capillary column end; g, mass flow controller; h, septum (retaining nut not shown); i, custom-turned stainless top; j, I/, in. nickel tubing; k, custom-turned stainless bottom with hole drilled In side to admit in. nickel tubing. Arrows indicate gas flow. Flgure 1.
ionization detector. The remainder is led by heated line (ca. 250 "C) to a copper oxide furnace (700 "C) in which all organic compounds are converted to C02and H20. The water is removed in a drying tube, quench gas (propane)is introduced, and the 14C02 is counted at ambient temperature and pressure in a shielded proportional tube (volume, 20 mL). Figure 1 shows the provision that was made to incorporate a glass support coated open tubular column into the gas chromatograph (we chose to fabricate our own injector/detector fittings; commercial fittings could equally well have been used). The injector allowed for splitless introduction of a sample to the capillary column; the detector end assembly allowed make-up gas to be added coaxially to the gas stream emerging from the column. Provision was made to preheat the make-up gas in the GC column oven prior to mixing. The mixed gases were then fed to a splitter that directed gas flow to the counter (99%) and to a flame detector (1%)- Both assemblies were made predominantly of nickel. All tubing measurements in Figure 1 are outside dimensions. Operating Conditions. Helium was used as carrier and make-up gas, propane as the quench gas. For the capillary column, carrier gas flow rates were held in the range 2-3 mL/min as dictated by a van Deempter plot. The make-up gas flow rate was set to deliver 54 mL/min to the gas flow counter. The quench gas flow was 6 mL/min. Hence the total gas flow through the counter was 60 mL/min. The SP-2250 support coated open tubular column used in this study had dimensions 25.0 X 0.0005 m. It had approximately 12 600 effective theoretical plates (measured with respect to n-tetradecan-1-01at 170 "C, k' = 6.5). The packed column used in this study had dimensions 2.7 X 0.002 m and contained 3% SP-2250 on Supelcoport (8C-100 mesh). It was installed in the gas chromatograph to make use of the same injector and detector end assemblies as did the capillary column. This involved inserting reducing unions at both column ends and in. 0.d. nickel to make the employing small lengths (1cm) of connections to the fittings. A t the injector end of this column this procedure increased the dead volume somewhat and may have reduced column efficiency fractionally. This effect could not have been great since the packed column possessed approximakly 4300 effective theoretical plates (measured with respect to n-tetradecan-1-01at 200 "C, k' = 6.5). Carrier gas flow was sufficient to give a flow of 54 mL/min of helium to the counter. For the packed column, the quench gas flow was held at 6 mL/min. For both packed and capillary columns, the path length to the counter is longer than that to the flame detector. This offset can be accommodated accurately if the data are processed by computer (see ref 6); for analogue traces such as are found in Figures 2 and 3, an acceptable accommodation is to feed the counter to the front running and the flame detector to the lagging pens, respectively,
ELUTION
TIME
Flgure 2. Traces produced by capillary (A) and packed (B) columns. Upper trace is of radioactivlty, lower of mass.
f
y D
ELUTION TIMES Flgure 3. (Panel A) Section of an "ene-ol" trace showing the response to approximately 200 dpm of ndodecan-1-01, (Panels B and C) Corresponding sections of traces obtalned by running a total radlo-
labeled plant extract the packed and capillary RGC units, respectively. (Panels D and E) Radiogas chromatograms of a mixture of sterols on packed and capillary RGC units, respectively. Peaks 1 through 4 in panel D are respectively cholesterol, "desmosterol", dihydrolanosterol, and lanosterol. of a twin pen recorder (e.g., Hewlett-Packard 7128A) and physically set the pen offset to counterbalance the detector/counter offset. Sample Preparation. Most of the development work was done with a synthetic sample ("ene-01" standard) containing the nalk-1-enes of carbon number 12, 14, 15, 17, 18, and 19 and the primary n-alcohols of carbon number 12, 13, 14, 15, 16, 17, 18, and 19. In the sample the alcohols of carbon number 12,16, and 18 were also radiolabeled to the extent of approximately 640,600, and 560 dpm/pL, respectively. The biological sample was obtained by feeding U-[14C]glutamate(20 NCi) to 7-day-old seedling cuttings of Impatiens balsamina for 2 h. At the close of the isotope incorporation experiment, the tissue was homogenized in methanol and treated immediately with excess ethereal diazomethane. The ethyl acetate solubles were then concentrated to 200 pL and the stated amount used in RGC analysis. The desmosterol used in Figure 3 was part of a mixture of sterols, each obtained independently from commercial sources.
RESULTS Figure 2 compares the performance of packed and capillary columns with an approximately 1-pL sample of the "ene-01" standard. The running conditions were chosen to maximize resolution in each (capillary, 50 "C for 5 min and then to 150 "C at 30 "C/min, hold at 150 "C for 1 min and then to 270 "C a t 5 "C/min; packed, 150 OC for 1 min and then to 320 OC at 10 OC/min: full scale deflection for radioactivity was 1000 counts/min in both cases; for mass it was 32 X A for the capillary, 4 x A for the packed column). The
19’81
Anal. Chem. 1981, 53, 1961-1962
identical injector/detector/splitter/proportionalcounter combination was used to produce both traces; only the make-up gas flows were changed to ensure that the same total gas flow entered the splitter in both cases. Clearly capillary-RGC with a commercial gas flow proportional counter is feasible-in F‘igure 2.4 (that produced from the capillary system) well-defined peaks of mass and radioactivity are evident. Even more importantly, the increased resolution of which the capillary column is theoretically capable, is delivered to the RGC experiment. This is seen most palpably for the three sets of doublets (peaks 5/6,7/8,9/10). On the packed columns, base line resolution does not occur for any doublet set. With the capillary column, base line resolution is observed for all. If the concept of separation number [(retention time difference between two adjacent peaks/sum of half-bandwidths of these two peaks) - 11 can be applied legitimately to temperature programmed gas chromatography, the separation number corresponding to peaks 9/10 on the ]packed column would be zero, wlhile for the capillary column it would be 3. Figure 3 highlights some performance characteristics of capillary-RGC. Panel A shows the response corresponding to a sample of approximately 200 dpm (4ng, taking account of splitter) of n-[14C]dodecanol (peak 4 of the ‘ h e - o l ” standard, see Figure 2). The radioactivity peak is 3.8 times the peak-to-peak background signal. We predict from this datum that our capillary-RGC unit could detect 1010 dpm comfortably. Two operational advantages of capillary-RGC over packed-column RGC are shown in the remainder of Figure 3. Panels B and C show, respectively, the corresponding section of RGC traces produced by packed and capillary columns from a radiolableled plant extract. Accurate attribution of radioactivity to mass in the area marked “a”, is not possible in the trace produced by the packed column (panel B); it is possible in that, produced by the capillary column (panel C). Panels D and E provide even more cogent proof
of the value of capillary RGC. Commercial samples of desmosterol, used by us in studies of cholesterol biosynthesis, consistently gave broad peaks on packed GC columns. When spiked with [14C]desmosterol,the radioactivity peak was not commensurately broad (panel D). The capillary RGC run of the spiked sample (panel E) shows that “commercial desmosterol” is in fad a mixture of two materials. The second substance is the true desmosterol (checked by mass spectrometry).
DISCUSSION Our results establish (a) that capillary-RGC with cornmercial gas flow proportional counters is feasible and (b) that the increased resolution that it brings to the RGC experiments can be valuable operationally. Indeed the commercially based unit is 4-fold more sensitive than the custom-designed system of Gross et al. in detecting 14C(100 vs. 400 dpm). One reason for this improved performance may be that the unit designed herein permits independent optimization of gas flow through GC column and proportional counter. LITERATURE CITED Kokes, R. I.; Tobln, M.; Emmett, P. H. J . Am. Chem. SOC.1855, i.7, 5860-5862. Campbell, I. M. Angal.Chem. 1978, 51, 1012A-1021A. Doerfler, D. L.; Rosenblum, E. R.; Malloy, J. M.; Naworal, J. D.; Mcklanus, I. R.; Campbell, I. M. Biomed. Mass Spectrom. 1980, 7 , 259-264. Hamnett, A. F.; Pratt, E. G. J . Chromatogr. 1878, 158, 387-399. Gross, D.; Gutekunst, H.; Blaser, A,; Hambock, H. J . Chromatogr. 1980, 198, 389-396. Campbell, I. M.; Doerfler, D. L.; Donahey, S. A.; Kadlec, R.; McGandy, E. L.; Naworal, J. D.; Nulton, C. P.; Venza-Raczka, M.; Wlmberly, F. Anal. Chem. 1877, 49, 1726-1734.
RECEIVED for review May 20, 1981. Accepted July 13, 1981. The financial support of the National Institutes of Health (Grant No. GM 25692), the National Science Foundation (Grant No. PCM 77-03966), and the Muscular Dystrophy Association of America is gratefully acknowledged.
Automatic Liquid Nitrogen Controller for the Thermal Energy Analyzer Cold Bath James L. Owens” and Oswald E. Kinast Monsanto Company, 800 M t f h Lindbergh Boulevard, St. Louis, Missouri 63 166
Since the introduction of the thermal energy analyzer (TEA) (I), the determination of N-nitrosamines has been greatly facilitated. This instrurnent is a selective and sensitive detector and can be interfaced with either a gas or liquid chromatograph. This fealture allows for the determination of both volatile and nonvolatile N-nitrosamines in many kinds of matrices. A cold trap and bath are essential elements of the TEA’S selectivity and sensitivity. After the catalytic disruption of the N-NO bond in the TEA pyrolyzer furnace, the fragments are directed into the cold trap which is maintained at -160 OC by using an isopentane/liquid nitrogen slush. This temperature is cold enough to freeze out most organic solvents and molecular fragments but allows the NO radical to pass into the reaction chamber. The preparation of the isopentane/liquid nitrogen cold bath slush is very tedious and time-consuming. Periodic addition of liquid nitrogen is requhed to maintain the temperature cold enough for proper operation. The storage and handling of isopentane are also safety hazards. The liquid boils at approximately 28 “C and is extremely flammable.
A desirable feature for analyzing large numbers of samplw for N-nitrosamines is automatic operation, i.e., autoinjection, especially for GC-TECA operation. Because of the problenis associated with handling isopentane, autooperation is not possible. In order to circumvent these problems, we investigated other alternatives. Liquid nitrogen alone proved to be too cold for our work, decreasing sensitivity by about 40%. In addition, liquid nitrogen boils off too rapidly. We tried without success to find other solvents which would produce the desired temperature, solve the handling problems, and be amenable for autoogeration. Another alternative was the use of cascade coolirig devices such as the RdC-4-130 Multi-Cool system (FTS Systems, Inc., Stone Ridge, NY). However, these systems are bulky, noisy, and exlpensive. Thermo Electron Corp. (Waltham, MA) has introduced a disposable cartridge trap called the CTR Gas Stream Filter. These filters replace the cold trap altogether and are comvenient to use. Depending upon sample type and load, they are good for 1-2 days. For certain analytical requirements they are as good as the cold trap and slush arrangement.
0003-2700/81/0353-196’I$O1.25/00 1981 American Chemical Society
