Precolumn synthesis of trimethylsilyl derivatives of aqueous

chlorpheniramine, about 0.1% of the secondary amine and about 0.6% of the primary amine. Under actual conditions, the metabolites are present in quant...
2 downloads 0 Views 308KB Size
I51

3.0

3.5

4.0

4.5

5.0

5.5

PH

Interference of two chlorpheniramine metabolites, the mono(V)and didemethylated (0)compounds, under different pH conditions Figure 1.

chlorpheniramine, about 0.1% of the secondary amine and about 0.6% of the primary amine. Under actual conditions, the metabolites are present in quantities of about the same order of magnitude as chlorpheniramine, so this type of efficiency is more than adequate. Human control plasma containing 10.0 ng/mL (calculated as the free base) of chlorpheniramine maleate was analyzed using the procedure given above. The mean f SD obtained from ten replicates was 10.0 f 0.52. The sensitivity of the procedure is adequate for the determination of as little as 0.5-1.0 ng/mL of drug.

The need for clean glassware and pure reagents cannot be overemphasized. Other potential problem areas are the evaporation of ether just before the oxidation step and loss of sample in nonpolar solvents. If the permanganate is depleted during the oxidation, it is almost certain that some residual ether was present and more stringent measures must be instituted for its removal. The other potential problem is adsorption of the ketone during storage of the sample in nonpolar solvents. This can be prevented by adding a small amount of acetone to the sample. Some typical results are given in Table I. Human volunteers received a single dose of chlorpheniramine maleate and blood samples were taken at intervals through 24 h. There were four subjects in each group. Based on these results, the terminal half-life in serum was 18 h. ACKNOWLEDGMENT The authors thank Ms. P. A. Huwel for her valuable technical assistance. The mono- and didemethylated metabolites were a generous gift of the Smith Kline and French Laboratories, Philadelphia, Pa. LITERATURE C I T E D (I) R. B. Bruce, J. E. Pins, and F. M. Pinchbeck, Anal. Chem., 40, 1246

(1988). (2) T. D. Doyle and J. Levine, J . Assoc. Off. Anal. Chem., 51, 191 (1968)

RECEIVED for review February 25,1977. Accepted March 23, 1977.

Precolumn Synthesis of Trimethylsilyl Derivatives of Aqueous Phosphate For Gas Chromatographic Analysis Rlchard H. Getty, Julia Stone,' and Richard H. Hanson" Chemistry Department, University of Arkansas at Little Rock, Little Rock, Arkansas 72204

The synthesis of volatile derivatives from nonvolatile samples frequently is performed prior to analysis by gas chromatography. When this process can be accomplished on the column or in a precolumn, the speed of analysis is faster and smaller samples can be analyzed. Several authors have studied the on-column silylation process. Esposito (1) showed that by using an on-column synthesis technique, it was possible to silylate compounds dissolved in reactive solvents such as water or alcohol. The concentration of the samples was between 1 and 10%. No data were presented to indicate if the reactions were quantitative. Morrow (2) constructed a precolumn assembly where solvent vapor was purged to the atmosphere prior to silylation. All volatile reaction products were swept onto the column and analyzed. Conversion to the orthophosphate derivative was less than 50% with a 48 pg sample. Matthews (3) was able to lower the detection limit, on orthophosphate to 1 pg, but was unable to reproduce peak areas. The peaks were poorly shaped with serious tailing. Wiese and Hanson (4) quantitatively silylated orthophosphate in a precolumn in the 10-100 pg range. Above the upper limit, conversion was not complete. This report describes a technique which was used to quantitatively analyze aqueous orthophosphate in the 0.25-5 pg range using a precolumn reaction system. The problems of peak shape, tailing, and reproducibility were minimized. 'Present address, Chemistry Department, University of Indiana, Bloomington, Ind. 47401 1086

ANALYTICAL CHEMISTRY, VOL. 49, NO. 7, JUNE 1977

Some organic acids were also successfully silylated and analyzed. The technique of reaction gas chromatography was applied to the problem of silylating aqueous inorganic anions to simplify the reported procedures (5,6). In both reports, the anion was converted to the ammonium form by an ion exchanger prior to synthesis. The first report required 10 mg of salt dissolved in a nonaqueous solvent. Silylation was accomplished overnight in a reaction vial and 125 pg of the salt injected for analysis. In the second paper, silylation was accomplished in a much shorter time. Two hundred mL of 0.1 part per million orthophosphate were concentrated, silylated, and analyzed, As little as 0.02 pg of phosphate were detected. In the technique developed and described below, much less volume of sample was required. Quantitative conversion of 0.25 pg of orthophosphate in 5 pL of sample was possible without treatment with an ion exchanger or concentrating into a nonaqueous solvent. EXPERIMENTAL Apparatus. A Hewlett-Packard 5750 gas chromatograph was

equipped with dual flame ionization detectors. The analytical column was 6-f00t, '/.,-inch glass packed with 5% OV-225on 60/80 mesh Chromosorb W, DMCS treated, and acid washed. The flame detector was operated at 250 "C, the injection port at 150 "C, and the column at 135 "C. Nitrogen carrier gas flow was 60 mL/min at 50 psi. The precolumn was the commercially available Pyrolysis Sampling System from Hamilton Company. This accessory, though designed as a pyrolysis attachment, was easily adapted

t--i-

- 4 - 3

4

Sketch of reaction precolumn. (1) Oven, (2) Furnace, (3) Unheated zone, (4) Sample introduction port, (5) Gas flow in, (6) Heated transfer line, (7) Injection needle, (8) Large glass tubing, and (9) Sample tube Flgure 1.

for the reactive part of the system. The controlled temperature zone nearest the opening of the process tube, the furnace, area 2 in Figure 1, was not packed. The oven zone, area 1 in Figure 1,closer to the heated line and downstream of the first zone, was packed with the same material as the analytical column. A heated transfer line, marked 6 in Figure 1,connected this assembly from the oven zone to the injection port of the gas chromatograph. Reagents. Phosphate standards were prepared from ammonium monohydrogen phosphate or from sodium phosphate. N,O-bis(trimethylsily1)trifluoroacetamide with 1% trimethylchlorosilane was obtained from Regis Chemical. Procedure. A glass tube, 3 mm 0.d. and 80 mm long, was used to introduce the sample, in the ammonium or acid form, into the precolumn reaction system. The tube was prepared by rinsing with 1:3 nitric acid-distilled water, and then silylated. A plug of silylated glass wool was put in both ends, leaving the center portion empty. Five WL of the sample were transferred to the tube with a Hamilton syringe. The tube was then placed in a vacuum oven at 98 "C and the pressure reduced 24 psi. Water from the sample vaporized in about 3 min. Eighty WL of BSTFA-1% TMCS were added to the tube, now at room temperature, so that the deposited sample was covered with reagent. The tube was put into the tube of the Pyrolysis Sampling System through the port, 4 in Figure 1,and moved into the furnace zone, 2, at 80 "C. Reaction at this temperature occurred for 15 min. The temperature was next increased to 140 OC for 5 min. All of this was done with the carrier gas flow off. Carrier gas was then swept through the system, and the phosphate derivative condensed on the packing in the oven area, 1,at 45 "C, while excess BSTFA-1% TMCS and some reaction by-products passed through the oven and heated transfer line, 6, and were vented to the atmosphere. The venting step required 10 min. During this time, the tubing upstream of the furnace, area 3 in Figure 1,was heated for 3 min with a hot air gun to remove any condensed silyl compounds. Carrier gas flow was stopped and the oven temperature increased to 140 "C. The heated transfer line, at 150 OC, was injected into the injection port of the chromatograph and the carrier gas flow again resumed. The derivative was thus introduced into the analytical column for separation from residual silylating reagent and reaction by-products. In 3 min, the derivative hit the detector. If the sample was not in the ammonium or hydrogen form, a 45 M excess of ammonium chloride or hydrochloric acid was added to the sample prior to the start of the analysis.

RESULTS AND DISCUSSION Quantitative conversion to the volatile derivative was accomplished when the sample was mixed with a large excess of silylating reagent and reacted near the latter's vaporization temperature. Experiments where injections of sample and silylating reagent were put directly on the analytical column above the vaporization temperature of the silylating reagent gave irreproducible and nonquantitative conversion to the derivative. This confirms Esposito's (1)data that the reaction is more quantitative in the condensed phase. His results showed that the reaction gave higher yields if the liquid phase load on the column was higher, thus dissolving more silylating reagent. He interpreted this to mean the reaction occurred in the liquid phase of the coating. In the technique presented in this paper, no liquid phase coating was required to dissolve both sample and reagent. The reagent, kept near its va-

-

0

2

I

I

TIMIN.)

2

4

Figure 2. Typical chromatogram. (1) Acetone, (2)2-Octano1, (3) Residual BSTFA, and (4) (TMS)3P04

porization temperature, dissolved the sample and allowed the reaction to occur. Results that a t first appeared to indicate a problem of reproducibility were quickly explained by recognizing the detector sensitivity changed with each burn of silylating reagent, that sample was being lost during the vaporization step and also during the separation of excess reagent from the derivative by condensing the derivative and venting the reagent. The silicon oxide coating buildup on the electrodes was minimized by the separation of reagent and derivative in the Pyrolysis Sampling System. Most silicon-containing compounds were vented to the atmosphere. Prior to introducing the derivative into the analytical column, 5 pL of Freon 113 were injected into the gas chromatograph, followed by 1 p L of 1 % 2-octanol in acetone. The response to the 2-octanol was taken as a reference and the ratio of this response to that of the derivative calculated. A typical chromatogram is shown in Figure 2. The second problem was eliminated by using tubes of sufficient volume to accommodate any bumping of sample that might occur. The third source of error required very close temperature control of the oven area in the Pyrolysis System. In order to get the 0.25 pg to 5 pg of sample introduced into the system, 5 p L of 50 ppm to 5 WLof 1000 ppm phosphate were injected into the small tube. The lower limit was a function of detector signal-to-noise and the upper limit was governed by the volume of the tube needed to accommodate the silylating reagent. By simply increasing the tube volume to accommodate enough solution to have 0.25 pg of sample or by some prior concentration step, less concentrated samples could be handled. For example, 50 p L of 5 ppm phosphate would provide the 0.25 pg of phosphate. A larger tube could hold more silylating reagent and thus more than 5 wg of sample could be analyzed. Phosphate in any form other than hydrogen or ammonium gave no conversion to the volatile derivative. A large molar excess, 45 times, of ammonium chloride or hydrochloric acid relative to phosphate, was added to the samples prior to analysis and synthesis was accomplished. A 5-pg sample of sodium phosphate was treated with hydrochloric acid and ANALYTICAL CHEMISTRY, VOL. 49, NO. 7, JUNE 1977

1087

another with ammonium chloride. The first response was 3.92 relative units and the second 4.20 units, the average being 4.06 f 3.5%. Memory checks with 50 to 1pg of phosphate gave no response. Reproducibility studies with 50 pg of the same phosphate sample gave results that were within 1 % relative error. Tests where sodium phosphate WBS treated with excess ammonium chloride gave results that were consistently slightly lower than the corresponding concentration of phosphate made from diammonium hydrogen phosphate. Analysis of samples, whether in the sodium, ammonium, or hydrogen forms, could be accomplished with accuracy of 1 7 % . Studies are currently being conducted to determine if aqueous mixtures of oxyanions can be synthesized and separated. Preliminary results with some organic acids showed this technique can also be used for organic compounds

1088

ANALYTICAL CHEMISTRY, VOL. 49, NO. 7, JUNE 1977

provided they are not volatile below 100 “C. LITERATURE CITED (1) G. Esposito, Anal. Chem.. 40, 1902 (1988). (2) R. Morrow, “A Study of the Atomic Emission and Absorption of Organic-Bound Silicon in Oxygen-And...”, Xerox University Microfilms, Ann Arbor, Mich,, 1970. (3) D. Matthews, “The Gas Chromatographic Determinations of Trace Anions In Aqueous Media”, Xerox Universlty Microfilms, Ann Arbor, Mich., 1972. (4) P. Wiese and R. Hanson, Anal. Chem., 44, 2393 (1972). (5) W. C. Butts and W. T. Rainey, Jr., Anal. Chem., 43, 538 (1971). (6) D. R. Matthews, W. D. Shults, and M. R. Guerin, Anal. Chem., 43, 1582 (1971).

RECEIVED for review January 31, 1977. Accepted March 18, 1977. This research was supported by the Office of Water Resources Research (Grant B-049-ARK) and the Faculty Research Fund.