126
Anal. Chem. 1981, 53, 126-127
Precolumn Coulometric Cell for High-Performance Liquid Chromatography Gary W. Schieffer Analytlcai Chemistry Division, Norwich-Eaton Pharmaceuticals, Division of Morton-Norwich Products, Inc., Norwich, New York 138 15
There appear to be few reports on utilizing the resolution capabilities of high-performance liquid chromatography (HPLC) for characterizing the products of electrochemical reactions (1). A low-dead-volume glassy carbon tubular electrode cell designed for operating under moderately high pressures was inserted between the injector and column of a high-performance liquid chromatograph to generate oxidation products for chromatographic separation (2). The cell was used to separate the oxidation products of adrenaline and l-dopa from the parent compounds (2) and to test the specificity of a stability-indicating HPLC method for the analyte in the presence of an electrochemically generated degradant (3). However, stopped flow was required to achieve a maximum electrochemical yield of only 5-10%, making the cell inconvenient to use. Miner and Kissinger used a porous graphite coulometric cell connected directly to the injection valve of an HPLC apparatus for identifying a toxic intermediate of acetominophen metabolism (4). A peristaltic pump was used to pump the solution containing acetominophen into the cell and the resulting oxidation product N-acetyl-pquinonimine (produced in 100% yield) into the injector. In the present work, a similar coulometric cell, designed to yield 100% electrolysis without stopped flow, was inserted directly into the chromatographic system, similar to the precolumn tubular electrode cell. Such a cell should obviate the need for a separate pump, yield fast and reproducible transfer of electrogenerated species, and be convenient to use. In addition, in certain cases the cell might be useful for controlling selectivity by electrochemically derivatizing one or more components of an unresolved peak. EXPERIMENTAL SECTION Cell Design and Construction. A schematic of the cell, which is based on the earlier tubular electrode design (2),is shown in Figure 1. The body consisted of two 2-in. (5.1-cm) diameter Plexiglas cylinders and a 3.3-cm diameter cylinder held together with four stainless steel bolts and four machine screw (not shown). The silver-silver chloride reference electrode (SSCE) was wound around a Plexiglas post (6.4 mm long and 1.02 cm in diameter) which extended from the top cylinder into a cavity (2.74 cm in diameter, 8.3 mm deep) in the middle cylinder. Epo-Tek 349 epoxy (Epoxy Technology Inc., Billerica, MA) was used to seal 1/16-in.(1.6-mm) stainless steel HPLC tubing into the top and bottom blocks, as shown. A 0.5-1 mm thickness of epoxy around the circumference of the tube was necessary to prevent leaks at high pressures. The epoxy was also used to seal the SSCE and platinum counterelectrode leads into the upper cylinder. A channel 3.0 mm in diameter and 22.0 mm in length was drilled through the middle cylinder. To prepare the cell for use, the top and middle cylinders were bolted together compressing eight cation-exchange membrane washers (0.25 mm thick, 0.63 mm i.d., 2.72-cm o.d., Ndion XR-170, E. I. du Pont de Nemours & Co., Inc., Wilmington, DE) located in the middle cylinder. The partially assembled cell was then inverted and a Teflon screen (Spectramesh, 74-pm openings) forced through the 22 mm long channel with a steel rod until it was flush with the surface of the cation exchange membrane washers. Reticulated vitreous carbon (Fluorocarbon Co., Anaheim, CA) was powdered with a spatula to form particles about 0.1 mm in diameter. The particles were poured into the channel a little at a time with fiim tamping with a steel rod between particle additions. The filled channel was then covered with a 15 mm X 15 mm X 0.025 mm thick piece of platinum foil with a 0.6-mm punched hole for solution flow to serve as the working electrode contact. A piece of Teflon screen was placed over the punched hole and the bottom cylinder attached with flour machine screws. 0003-2700/81/0353-0128$01.00/0
The stainless steel solution inlet and outlet tubes were fitted with stainless steel ferrules and compression screws. The 0.1 M KC1 inlet and outlet consisted of Tygon tubing pressfitted and sealed with cyclohexanone. Instrumentation. All potentials reported in this paper are given with respect to the SSCE (0.1 M in KC1). Potentials were applied with a Princeton Applied Research Model 364 or 174A polarographic analyzer. A high-performance liquid chromatograph (ALC/GPC 204, Waters Associates, Milford, MA) equipped with a U6K injection valve, a Model 6000A reciprocating pump, and a Model 440 absorbance detector (Waters Associates) at 280 nm was used. A variable-wavelength absorbance detector (Model GM-770, Schoeffel Instrument Corp., Westwood, NJ) with wavelength drive (Model SFA 339, Schoeffel Instrument Corp.) was used for stopped flow scanning and for a study at 473 nm. The inlet of the cell (top cylinder in Figure 1)was connected to the injector and the outlet connected to a prepacked microparticulate reverse-phase column (p-Bondapak C18,Waters Associates). Reagents and Eluent. The l-dopa, 3-methoxytyrosine [3(4-hydroxy-3-methoxyphenyl)alanine], 3-(3-hydroxy-4-methoxyphenyl)alanine, dopamine, and hypoxanthine described previously were used (3). All other reagents were ACS reagent grade. Solutions of the catecholamines were made by dissolving the respective compounds in 0.1 M phosphoric acid and diluting to 0.01 M with distilled water. The mobile phase was prepared by adjusting a 0.01 M NaHzPOl solution to pH 2.5 with phosphoric acid (5). The solution was filtered daily through a 0.5-pm Millipore filter and degassed prior to use. RESULTS AND DISCUSSION Attachment of the cell to the HPLC system caused only a 12% reduction in the number of theoretical plates measured for l-dopa, indicating a low holdup volume (measurement of the actual holdup volume was not attempted). In addition, no leaks were observed through the reference electrode compartment or the epoxied stainless steel fittings at the flow rates and pressures employed (up to 1.7 mL/min and 1800 psi). Since the high pressure requirement necessitated placement of the counterelectrode and reference electrode downstream from the working electrode, the chromatographic peak current-potential curves were somewhat drawn-out as in Figure 2 of ref 2. This required an additional overpotential of at least 0.5 V to achieve 100% electrolysis. Catecholamines were studied initially to characterize the cell since the electrochemistry is well-known (6, 7). Except for adrenaline and isoproterenol which cyclize to form aminochromes and indoles (6),catecholamines like l-dopa and dopamine are electrochemically oxidized at low pHs to quinones (2, 6, 7). A chromatogram of dopamine for which the electrode had been potentiostated at 1.5 V is shown in Figure 2A. Dopamine is oxidized with greater than 99% efficiency to yield a single peak for the quinone, Similarly, l-dopa yields a major peak for dopaquinone and two smaller peaks (Figure 2B) that were not observed when the same system was investigated with the precolumn tubular electrode cell (2). The l-dopa analogue 3-methoxytyrosine, which is more difficult to oxidize than dopamine and l-dopa, was oxidized with 95% efficiency at 2.0 V to yield a more complicated chromatogram (Figure 2C). A similar examination of 3-(4-hydroxy-3-methoxyphenyl)alanine yielded a chromatogram nearly identical with that shown in Figure 2C, indicating a similar oxidation mechanism. A chromatogram of the oxidation products of 3-methoxytyrosine monitored at 473 nm, the absorption maximum for dopachrome (8),did not indicate the presence of an aminochrome. 0 1980 American Chemical Society
ANALYTICAL CHEMISTRY, VOL. 53, NO. 1, JANUARY 1981
127
M. ,!.
u
350 300 250
u 350 300 250
WAVELENGTH. nrn
spectra of oxidation products obtained at stopped flow. Numbers refer to peaks in Figure 2. Spectra corrected for blank (eluent). Flgure 3. W
F
Figure 1. Precolumn coulometric cell for HPLC (cell shown prior to compression by the bolts, not shown): (A) Wings; (B) cation-ex-
change membrane washers; (C) platinum sheet working electrode contact; (D) Teflon screen; (E) lead to working electrode Via platinum contact; (F) stainless steel solution outlet; (G) powdered reticulated vitreous carbon working electrode; (H) 0.1 M KCI inlet; (J) lead to reference electrode (SSCE); (K) stainless steel solution inlet; (L) 0.1 M KCI outlet; (M) lead to platinum counterelectrode.
! 1
A
6
T 0.01 AU
1
T
0.01 AU
1
J
I
C ?
- u 0
5
0
5
MINUTES
Chromatograms of dopamine and hypoxanthine: (A) precolumn cell at 0 V; (B) cell at 1.5 V; 20 pg/mL dopamine, 30 pg/mL hypoxanthine, 100 pL injected, 1.5 mL/min, 280 nm. Flgure 4.
uu0
5
0
5 0
5
IO
15
MINUTES
Chromatograms of oxidation products: (A) dopamine OXC dized at 1.5 V; (B) /dopa oxMized at 1.5 V; (C) 3-methoxytyrosine oxidized at 2.0 V. In each case, peak of unoxldized compound is shown above oxidation chromatogram: concentration 40 pg/mL, 1.5 mL/min, 100 pL injected, 280 nm. Figure 2.
Indole derivatives can also be precluded since these elute after the 3-methoxytyrosine peak (3). UV spectra of peaks 1-4 in Figure 2B,C obtained at stopped flow with a scanning UV detector are shown in Figure 3. As expected, peaks 1 and 2 yield nearly identical spectra with bands at ca. 255 and 285 nm. Peaks 3 and 4 also yield nearly identical spectra with a band at ca. 265 nm. l-Dopa and 3-methoxytyrosine both have a single band at ca. 285 nm at pH 2.5. Although no further work was done to elucidate the structures of the oxidation products, the usefulness of chromatographic and spectroscopic techniques in studying electrochemically generated oxidation products seems clear. The use of the precolumn cell also shows promise in controlling chromatographic selectivity by electrochemical de-
rivatization, especially when an electroactive compound interferes with a nonelectroactive one. Figure 4A is a chromatogram of unoxidized dopamine and hypoxanthine yielding an unresolved peak. Potentiostating the precolumn cell a t 1.5 V completely resolves the dopaminequinone from the hypoxanthine (Figure 4B). Examination of 10 solutions with the dopamine concentration constant at 12 pg/mL and the hypoxanthine concentration ranging from 2 to 20 pg/mL yielded an average value of peak height ratio (hypoxanthine/dopaminequinone) divided by concentration of 0.186 mL pg-l with a relative standard deviation of 1.1%. Quantitative determination involving the oxidation product is, of course, dependent on a lifetime of at least several minutes for the product.
LITERATURE CITED (1) Heineman, W. R.; Kissinger, P. T. Anal. chem. 1980, 52, 151-l6lR. (2) Schieffer, G. W. Anal. Chem. 1979. 51, 1573-1575. (3) Schieffer, 0. W. J . fharm. Sd. 1979. 68, 1299-1301. (4) Miner, D. J.; Klssinger, P. T. Blochem. fharmacol. 1979, 28, 3285-3290. (5) Schieffer, G. W. J . fhann. Scl. 1979, 68. 1298-1298. (6) Hawley, M. D.; Tatawawadl, S. V.; Piekarski, S.;A d a m , R. N. J. Am. Chem. Soc. 1967, 89. 447-450. (7) Stemson, A. W.; Mccteery, R.; Fekrberg, 8.; Adam, R. N. J . f k troanal. Chem. Interfecial Electrochem. 1973, 46. 313-321. ( 8 ) Graham, D. G.; Jeffs, P. W. J . Bbl. Chem. 1977, 252, 5729-5734.
R ~ E N E Dfor review September 10,1980. Accepted October 20, 1980.