Parallel dual-electrode detection based on size exclusion for liquid

Department of Chemistry, New Mexico State University, Las Cruces, New Mexico 88003. A new dual-electrode detection scheme for liquid chroma- tography ...
0 downloads 0 Views 615KB Size
Anal. Chem. 1986, 58, 1019-1023

1019

Parallel Dual-Electrode Detection Based on Size Exclusion for Liquid Chromatography/Electrochemistry Lori D. Hutchins-Kumar, Joseph Wang,* and Peng Tuzhi Department of Chemistry, New Mexico State University, Las Cruces, New Mexico 88003

A new dual-electrode detectlon scheme for llquld chromatography utlllring bare and film-coated glassy carbon electrodes In a parallel-opposed Configuration Is descrlbed. Thls approach adds a new dimension of Informatlon, based on solute characterlstlc slze, to the armory of dual-electrode detectlon schemes. As a result, detectlon of elutlng compounds is Improved with respect to Selectivity and assessment of peak ldentlty purity. Selectlvlty Is Improved by obtalnlng two equlpotentlal chromatograms, one of whlch proflllng only small solutes. Dual response ratios (wlth equlpotential operation) provide unlque characterlratlon for the lndivldual components present In the sample. A lowerlng of the overpotential for certaln analytes is observed at the coated electrode. The merits of the descrlbed approach are Illustrated in the analysls of urlne and beer samples.

Liquid chromatography with electrochemical detection (LCEC) has become a widely accepted tool for the measurement of trace quantities of organic compounds (1-3). The technique couples the advantages of low detection limits, high selectivity, and low cost. The use of dual-electrode detection schemes for LCEC has gained popularity in recent years (4). Such dual-electrode LCEC can be useful for improving selectivity and/or sensitivity. A variety of dual-electrode detectors have been proposed for this purpose, of which the series and parallel configurations are most popular. The series configuration places one electrode upstream of the other; changes in the composition of the effluent associated with the redox reaction at the upstream electrode are sensed by the downstream electrode. In the parallel configuration, two identical electrodes are arranged side-by-side so that the effluent meets both electrodes simultaneously. The dual response ratio, obtained by poising the electrodes at potentials on the slope of the hydrodynamic voltammogram, provides a good estimate of the peak purity. The parallel arrangement permits also monitoring of both oxidizable and reducible species in the same injection. It is possible also to use a difference mode detection, based on recording the difference in the response of the two electrodes (5). All of the above dual-electrode detection schemes are based on the redox properties of eluting compounds. The present paper reports a new dual-electrode detection scheme for LCEC based on the size exclusion phenomenon. To implement this mode of detection, two electrodes with different transport properties-usually one bare and the second coated with a permselective polymeric film-are employed (with each electrode generating a chromatogram profiling its transport characteristics). Because the masstransport toward the surface of the coated electrode is a characteristic of the eluting compound, the ratio of the peak currents observed at the coated and bare electrodes-held at the same potential-can serve as a useful identifying parameter. (Such a ratio is commonly used to describe the permeation properties of a membrane or polymeric coating for a given solute.) In addition to the improvement in peak

identity confirmation, chromatograms are greatly simplified (without lowering the operating potential) when complex samples are analyzed. The new selectivity dimension based on molecular size makes this detection mode a powerful tool in the armory of dual-electrode detection schemes. A schematic depiction of the parallel-opposed configuration used for this purpose is shown in Figure 1;bare and cellulose acetate coated glassy carbon disk electrodes are employed. (A parallel-adjacent configuration, with both electrodes in close proximity on the same block, cannot be used conveniently because of the nature of the modification procedure.) The stability and selectivity advantages of cellulose acetate polymeric coating, using a single electrode detection, have been demonstrated recently (6, 7). As a result, high specificity toward small solutes and elimination of protein adsorption have been obtained in liquid chromatography and flow injection systems. We have shown that different permeabilities can be achieved by hydrolyzing the film in alkaline media for different time periods (7). Such control of the permeability is extremely useful for the dual-electrode operation described in the present work. Thus, the combination of the already selective film coated electrode with an advanced detection scheme addresses directly the complexity of various real samples. Analyses of urine and beer samples are used to demonstrate the (selectivity and peak identity/integrity confirmation) advantages of this size exclusion dual-electrode detection scheme.

EXPERIMENTAL SECTION Apparatus. The liquid chromatographicsystem (Bioanalytical Systems LC-303) consisted of a dual piston pump (PM-30A), a Rheodyne Model 7125 injector (20-pL loop), a Biophase CIS column (25 cm X 4.6 mm), a parallel-opposed dual electrode thin-layer cell, and two BAS-LC-4Bcontrollers. The current-time output was plotted on Houston Omniscribe strip-chart recorders. All potentials are reported vs. the Ag/AgC1(3 M NaCl) reference electrode. Reagents. Chemicals used were the following: catechol, 4methylcatechol, gentisic acid, ferulic acid, vanillic acid, vanillin, p-coumaric acid, gallic acid, norepinephrine, tyrosine, L-dopa, dopamine hydrochloride, DL-a-methyldopa,homogentisic acid, uric acid, and acetaminophen (Sigma Chemical Co.); 3,5-dimethoxy-4-hydroxycinnamicacid [sinapic acid] (Aldrich); chloroacetic acid sodium salt [practical grade] (Kodak); disodium ethylenediaminetetraacetate, ammonium acetate, methanol (HPLC grade),and ethyl acetate (Fisher);1-propanol(J.T. Baker). All chemicals were reagent grade or better and were used without further purification. All standards were prepared in the respective mobile phase. The mobile phase utilized for the biogenic amines was a 0.15 M chloroacetic acid (sodium salt) solution adjusted to pH 3.5 with phosphoric acid. For the urine sample analysis the mobile phase was adjusted t o 96% 0.15 M chloroacetic acid (pH 3.5) and 4% methanol. The mobile phase composition for the phenolic acid, benzoic acid, and cinnamic acid compounds wm 2.18% 1-propanol, 1.98% acetic acid, 8.71% methanol, and 87.13% deionized distilled water containing 0.036 M ammonium acetate. All mobile phases contained 0.20 g/L Na,EDTA. Procedures. Extractions. Ten milliliters of the beer sample (Strohs) was acidified to pH 2 with perchloric acid and saturated with sodium chloride. The solution was then extracted by shaking

0003-2700/86/0358-1019$01.50/00 1986 American Chemical Society

1020

ANALYTICAL CHEMISTRY, VOL. 58, NO. 6, MAY 1986

' I

Figure 1. Schematic depiction of the parallel-opposed dual electrode detector with bare and coated glassy carbon electrodes.

with 10 mL of ethyl acetate for 10 min. The extraction was repeated. The two ethyl acetate layers were then combined and rotor evaporated to dryness. The extraction residue was reconstituted with 5 mL of the mobile phase. Urine samples were obtained from healthy volunteers, filtered by passing through a 10-15 wm glass filter, and diluted (150) with the appropriate mobile phase solution. Surface Modification. The coating of the glassy carbon disk (lower block) and the hydrolysis of the resulting film have been described in detail previously (7). The polishing procedure was modified to minimize possible catalytic effects due to a-alumina particles. Both electrodes were polished with a 1-pm alumina slurry, rinsed with copious amounts of water, and then polished with a 0.05-wm alumina slurry. Following this, the electrodes were rinsed with dilute nitric acid and double-distilled water and sonicated for 15 min to remove residual polishing material. The electrodes were then allowed t o air-dry. Liquid Chromatography. Chromatograms were recorded while the bare and coated electrodes were held at the same potential (usually in the plateau region). Current ratios were established by injection of standards. Hydrodynamic voltammograms were generated by injecting a standard mixture at different potentials. All experiments were performed at ambient temperature, using a flow rate of 1.0 mL/min.

RESULTS AND DISCUSSION Response Characteristics. The power of the new dualelectrode detection scheme rests in its ability to obtain information on eluting compounds on the basis of their characteristic size. The current ratio (coated/bare), iL,c/iL,b, reflects the transport characteristics of solutes toward the surfaces of these electrodes (which are based primarily on molecular size). Under the mixed (film/solution) and solution-controlled transport conditions of the coated and bare electrodes, respectively, the current ratio can be expressed as

where P, and P, are the permeabilities of the solution and film, respectively, and id is the current at the bare electrode (same as iL,b). The peak current ratio is unique to each solute, as the permeability is given by D/6, the ratio of the diffusion coefficient within the film (or solution) to the thickness of the film (or the diffusion boundary layer) (8, 9). (It should be noted that these equations are for the theoretical steady-state current and not the experimental peak height that is affected by the chromatographic band broadening). Depending on the film permeability (i.e., hydrolysis time) solutes larger than a certain size are completely excluded from the surface (PmN 0). Thus, those peaks appearing at the bare electrode, and not at the coated, can be attributed at first glance to large eluting compounds. Figure 2 shows chromatograms recorded simultaneously at the bare (A) and coated (B) electrodes, operated at the same potential, for a series of biogenic amines. While the sample components have a relatively narrow molecular weight range (153-211), distinct changes in peak ratios are observed (Table

1 1

0

5

5 nA(B 5OnA(A

10 t,min

15

Flgure 2. Chromatograms obtained simultaneously with dual electrode operation: (A) bare electrode, (B) coated electrode (30 min hydrolysis); operating potential (both electrodes),+900 mV; flow rate, 1.O mL/min, mobile phase, 0.15 M chloroacetic acid solution adjusted to pH 3.5 with phosphoric acid (containing 0.2 g/L disodium EDTA); peak identities, (1) norepinephrine, (2) L-dopa, (3) epinephrine, (4) tyrosine, (5)dopamine, (6) a-methyldopa, (7)homogentisic acid, all present at the 100-ng level.

Table I. Current Ratios for Parallel Outputs for Biogenic Amines"

compound

mol wt

ip,c/ip,b

norepinephrine L-dopa epinephrine tyrosine dopamine a-methyldopa homogentisic acid

169 197 183 181 153 211

0.094 0.056 0.072

" Conditions:

168

1.300 0.077

0.042 0.100

as in Figure 2.

I). As solute characteristic size increases, permeability through the coating (and thus peak ratio) decreases. Note especially the changes for the larger compounds, L-dopa (mol wt 197) and a-methyldopa (mol wt 211)-peaks 2 and 6. Such changes can be used to confirm peak identity, as described later in the paper. (Obviously, these changes are even more pronounced for complex samples, containing a myriad of eluting compounds representing a wide range of molecular weights; see below.) The trend in peak ratios (based on solute characteristic size) is usually maintained for operation on the mass-transport limited plateau. Deviations may be observed if the electrodes are poised below the plateau potential for a given solute. For example, tyrosine (mol wt 181, peak 4) exhibits a peak ratio that is significantly larger compared to other biogenic amines of similar size (measured on the plateau potential). Nevertheless, the peak ratio remains a unique characteristic of the analyte and can be used for peak identification. Such behavior can be understood by considering the chromatographically assisted hydrodynamic voltammograms (HDV's) of biogenic amines (Figure 3). At both electrodes, the current increases as the potential is made more positive

ANALYTICAL CHEMISTRY, VOL. 58, NO. 6, MAY 1986

10

-

1021

C

b

e

0

L 0.2

0.6

1.0

E.V

fa

A

200

i,nA

100

0

0.2

0.6

1.0

E,V

Figure 3. Hydrodynamic voltammograms generated simultaneously at the bare (A) and coated (B) electrodes: (a) norepinephrine, (b) L-dopa, (c) epinephrine, (d) tyrosine, (e) dopamine, (f) a-methyldopa, and (9) homogentisic acid, all injected at the 100-ng level. Other conditions are as in Figure 2.

until a limiting current plateau is reached. The magnitudes of the limiting currents differ at both electrodes, as they reflect differences in the rate of mass-transport toward these surfaces (10- to 15-fold difference for the 30-min hydrolyzed film and the biogenic amines of Figure 3). The coated electrode exhibits some decrease in the limiting current at high potentials. Such a decrease was observed also at the bare electrode (not shown due to the different plateau regions). A surprising aspect of the cellulose acetate coated electrode is the (-200 mV) decrease in the overpotential observed for the biogenic amines. For example, while at the bare electrode the mass-transport limiting response is usually observed a t potentials more positive than +800 mV, operation a t +600 mV is sufficient for this purpose at the coated electrode. (An exception is homogentisic acid(g) for which potentials of +400 mV and +650 mV are sufficient a t the coated and bare electrodes, respectively.) The chromatograms shown in Figure 4 can be used to illustrate this “activation” effect. Chromatogram a shows the LCEC response for a mixture of 100 ng of norepinephrine, epinephrine and L-dopa as recorded a t the bare electrode poised a t +400 mV. The corresponding (simultaneously recorded) response for the coated electrode is shown in b; significantly larger peaks are observed. When the cellulose acetate coating is removed (chromatogram c), the

I

0

I

5 t,min

I

10

Figure 4. Chromatograms for injections of 100 ng of norepinephrine, L-dopa, and epinephrine obtained (a) at the bare electrode, (b) at the coated electrode, and (c) after removing the coating from electrode (b). Operating potential was +400 mV. Other conditions are as in Figure 2.

“catalytic” activity is removed and a response characteristic of the bare electrode is obtained (as indicated from comparison to chromatogram a). Analogous cyclic voltammograms recorded for L-dopa, ascorbic acid, uric acid, acetaminophen, and potassium ferrocyanide in a quiescent solution did not exhibit improvements in the reversibility at the coated electrode. Whether or not a lowering of overpotential is obtained is thus dependent on the experimental conditions (e.g., mass-transport, time scale). This is in agreement with the theoretical behavior of electrodes coated with thin films of nonelectroactive polymers (IO). When the diffusion rate is slower in the film than in the solution (as is the case under the forced-convection conditions of the thin-layer cell) the system becomes more reversible on the coated electrode than on the bare one. For phenolic acids the hydrodynamic voltammograms appeared in similar potential regions and had similar shapes-including some decrease in limiting current at high potentials-at both electrodes (not shown), i.e., no “catalytic” effect. In order to be useful analytically, the current ratiq must be independent of solute concentration and reproducible. Figure 5 shows dual-electrode chromatograms for injections of solutions containing increasing amounts (5-15 ng) of gentisic acid and catechol. These three injections are part of a series of seven successive injections of such mixture, with solute amount ranging from 5 to 100 ng. Linearity was obtained at both electrodes (with correlation coefficients of the four calibration plots ranging from 0.9997 to 1.0000). Accordingly, the current ratio was independent of solute amount: 0.290 for gentisic acid (range, 0.288-0.293; relative standard deviation, 0.8%) and 0.400 for catechol (range, 0.395-0.408; relative standard deviation, 1.6%). The reproducibility of the current ratio was estimated from a series of 1 2 successive injections of a solution containing 25 ng of gallic acid, gentisic acid, ahd catechol and 75 ng of vanillic acid. Statistical treatment of these data is given in Table 11. As indicated from the relative standard deviations (1.15-2.85%), the current ratio remains essentially the same over an unbroken 4-h period, using the same bare and coated electrodes. Changes in current ratios, ranging from 1 to 30%, may be observed in the day-to-day results due to the use of different coatings and surfaces. Such

1022

ANALYTICAL CHEMISTRY, VOL. 58, NO. 6, MAY 1986

B

B

a

if7

A

c/

4

A

0

dL

0

5

0

5

0

t.mln

hydrolysis) electrodes were operated at 4-800 mV. Flow rate was 1 mL/min. Mobile phase composition was 2.18% 1-propanol, 1.98% acetic acid, 8.71% methanol, and 87.13% deionized distilled water, 0.036 M ammonium acetate. Table 11. Reproducibility of Current Ratio”

gallic acid catechol

gentisic acid vanillic acid

30

Flgure 6. Dual-electrode chromatograms of a diluted (150) urine sample taken 4 h after normal acetaminophen dosage. Bare (A) and coated (B, 25 min hydrolysis)electrodes were operated at +900 mV. Flow rate was 1 mL/mln. Mobile phase composition was 96% 0.15 M chloroacetic acid (pH 3.5) and 4% methanol. Peak identities are as follows: (1) ascorbic acid, (2) uric acid, (3) possibly sulfonate metabolite of acetaminophen, (4) acetaminophen.

Figure 5. Dual-electrode chromatograms for increasing increments of 5 ng of gentisic acid and catechol. Bare (A) and coated (B, 40 min

compound

15

t,min

mean

range

re1 std dev, %

0.342 0.379 0.277 0.582

0.335-0.361 0.375-0.388 0.273-0.284 0.568-0.627

2.18 1.23 1.15 2.85

“Conditions: 12 successive injections of 25 ng of gallic acid, gentisic acid, and catechol and 75 ng of vanillic acid. Other conditions are as in Figure 5. changes cause no real problem as daily calibration provides accurate standard ratios for the particular electrode pair used. (Even with such changes, the ratio trend is maintained according to molecular size, as it usually reflects changes in the porosity of the film.) Similar day-to-day changes in current ratios are common to dual-electrode detection schemes based on redox properties (11, 12). Because of the equipotential operation used in the present study, changes in the reference electrode potential do not affect the current ratio as they do in dual-potential operation. Analytical Utility. The advantages of the size-exclusion dual-electrode detection scheme are best demonstrated using complex samples, containing numerous components of widely varying molecular sizes. The use of a single bare electrode, poised at high potentials, results in decreased selectivity under these conditions. By simultaneous operation of the bare and coated electrodes, such selectivity problems are eliminated for the detection of small eluting compounds. Figure 6 shows dual-electrode chromatograms of a diluted urine sample ob-

tained 4 h after normal acetaminophen dosage (two Tylenol tablets). Convenient quantitation of small eluting compounds, e.g., ascorbic acid and uric acid (peaks 1and 2), is illustrated at the coated electrode (B), as interferences due to coeluting large components are eliminated. Note that both electrodes operate at the same potential, thus the specificity toward small solutes is maintained for both easy and hard to oxidize components. (Lowering the operating potential, as is common in dual-potential operation ( 4 ) , excludes the hard to oxidize solutes.) The peak assignments advantage of the dual-electrode operation is also obvious from these profiles. For components yielding response at both electrodes, the peak current ratios are useful for assessing the peak identity/purity (via comparison to ratios for a standard mixture, Table 111). For ascorbic acid and uric acid the current ratios are 0.023 and 0.042, respectively, compared to 0.022 and 0.044 for injected standards. The slight disagreement in the values obtained for acetaminophen, 0.085 (sample) vs. 0.079 (standard), may indicate an overlapping response (possibly with its cysteine metabolite ( 4 ) ) . The agreement in the observed values of current ratios, combined with retention data, confirms the assignment of the chromatographic peak identity. As urine metabolites of acetaminophen widely range in their molecular size, the present approach can be used to differentiate between small and large metabolites. For example, peak 3 is attributed to the relatively small sulfonate metabolite of acetaminophen (mol w t 246; current ratio 0.021). Larger metabolites that are excluded by the present coating (hydrolysis time, 25 min) may be quantified by using more permeable films (i.e., longer hydrolysis times). The improvements in selectivity and peak identity information are illustrated also from the chromatograms of ethyl acetate extract of a commercial beer sample (Figure 7 ) . The coated electrode chromatogram (B) is greatly simplified (particularly in the early eluting portion) compared to that

ANALYTICAL CHEMISTRY, VOL. 58, NO. 6, MAY 1986

1023

(170) > p-coumaric acid (164). Note the substantial increase in current ratios, compared to the urine solutes, associated with the use of a more permeable coating (40 vs. 25 min hydrolysis times). Similar improvements were obtained with beer samples from other sources, as well as an ethyl acetate extract of orange juice.

CONCLUSIONS

0

10

20

30

40

t,mln

Figure 7. Dual-electrode chromatograms for an ethyl acetate extract

of a commercial beer sample. Conditions are as in Figure 5. Peak identities are as follows: (1)vanillic acid, (2) p-coumaric acid, (3)ferulic acid, (4) sinapic acid. Table 111. Comparison of Ratios for Parallel Outputs for Both Standard and Sample (Urine: Beer Extractb) Injections ip,c/ip,b

compound

sample

standard

ascorbic acid" uric acid" acetaminophenn ferulic acidb p-coumaric acidb sinapic acidb vanillic acidb

0.023 0.042 0.085 0.24 0.41 0.20 0.47

0.022 0.044 0.079 0.23

0.53 0.21 0.52

Conditions were as in Figure 6 . *Conditionswere as in Figure 7.

of the bare electrode (A). The detection of phenolic acids in the beer extract (commonly used to add flavor to the beer's aroma) is improved by comparing their current ratios to those estimated from an injection of standards (Table 111). Good agreement exists between the sample and standard ratios. An exception is p-coumaric acid (peak 2), due to its unpure peak a t the bare electrode. Slight differences in the current ratio (e.g., for vanillic acid) are attributed to matrix effects (such as electrode passivation) or other factors that affect the day-to-day precision (such as changes in the working electrode surface or mobile phase composition). Such differences are common also to dual-potential operation at two bare electrodes (11, 13) and can be largely minimized by employing an alternate standard-sample injection scheme. Note also the minimization of interferences due to partially coeluting large compounds, associated with the p-coumaric acid and vanillic acid peaks (as well as for the unknown peak at t R E 9 min). The trend in current ratios of the phenolic acids standards is according to what is expected based on their molecular weights: sinapic acid (224) > ferulic acid (194) > vanillic acid

The parallel dual-electrode approach, based on the size exclusion phenomenon, provides the trace analyst with another effective and versatile tool for measuring electroactive compounds in complex samples. The power of this approach rests in its ability to selectively detect eluted compounds on the basis of their characteristic size, This is achieved via a proper use of a chemically modified electrode that adds a new dimension to dual-electrode detection schemes. The ratio of the two resulting currents provides a real-time fingerprint of each peak in the chromatogram. As with dual-potential detection schemes a high degree of certainty in peak identity can thus be obtained in combination with retention data. The selectivity is greatly improved as the less complex coated electrode chromatogram is used to monitor small eluting compounds. Such detection capabilities reduce further the attention necessary for sample cleanup. Additional information is thus obtained at no additional expense in volume or time. Because of the nature of the modification procedure, this approach is best implemented using an amperometric thin-layer cell in the parallel-opposed configuration. Although the method was illustrated in analysis of urine and beer samples, it is applicable to a wide range of other samples. In addition to the use of coated and bare electrodes, poised a t the same potential, it may be possible to use other arrangements that would result in additional advantages. These include the use of two coated electrodes with different film permeabilities or of a bare and coated electrodes held at different potentials. The former would add the protection necessary when electrode fouling is anticipated, while the later would add a new dimension-based on redox properties-to the operation discussed in this study. Ultimately, one can expect an array of multiple electrodes possessing different transport and redox properties. Registry No. C, 7440-44-0; cellulose acetate, 9004-35-7; norepinephrine, 51-41-2;L-dopa, 59-92-7;epinephrine, 51-43-4; tyrosine, 60-18-4;dopamine, 51-61-6;a-methyldopa, 555-30-6;homogentisic acid, 451-13-8;gallic acid, 149-91-7; catechol, 120-80-9; gentisic acid, 490-79-9; ascorbic acid, 50-81-7;uric acid, 69-93-2; acetaminophen, 103-90-2;ferulic acid, 1135-24-6;p-coumaric acid, 7400-08-0;sinapic acid, 530-59-6; vanillic acid, 121-34-6.

LITERATURE CITED Kissinger, P. T. Anal. Chem. 1977, 4 9 . 447. Rucki, R. Talanta 1980, 27, 147. Stulik, K.; Pacakova, V. J. Nectroanal. Chem. 1981, 129, 1. Roston, D. A.; Shoup, R. E.; Kissinger, P T. Anal. Chem. 1982, 5 4 , 1417A. Lunte, C. E.; Kissinger, P. T.; Shoup, R . E. Anal. Chem. 1985, 57, 1541. Sittampalam, G.; Wilson, G. S . Anal. Chem. 1983, 55, 1608. Wang, J.; Hutchins, L. D. Anal. Chem. 1985, 57, 1536. Gough, D. A.; Leypoidt, J. K. J. Nectrochem. SOC. 1980, 127, 1278. Gough, D. A.; Leypoldt, J. K. Anal. Chem. 1980, 52, 1126. Delamar, M.; Pham, M. C.; Lacaze, P. C.; Dubois, J. E. Electroanal. Chem. 1980, 108, 1. Roston, D. A.; Kissinger, P. T. Anal. Chem. 1981, 53, 1695. Shoup, R . E.;Mayer, G. S. Anal. Chem. 1982, 5 4 , 1164. Lunte, C. E.; Kissinger, P. T. Anal. Chem. 1983, 55, 1458.

RECEIVED for review September 12,1985. Accepted December 16,1985. This work was supported by the National Institutes of Health (Grant No. GM30913-02) and the American Heart Association. We wish to thank BAS, Inc., for the loan of the LC-4 potentiostats used in this work.