Amperometric detection of amines using cobalt electrodes after separation by ion-moderated partition chromatography

Talanta E L S E V I E R Talanta44(1997) 239 248

Amperometric detection of amines using cobalt electrodes after separation by ion-moderated partition chromatography

A. H i d a y a t ~, D . B . H i b b e r t ~'*, P . W . A l e x a n d e r b

"Department of Analytical Chemistry, University of New South Wales, Sydney, NSW 2052 Australia bDepartment o[' Physieal Science, Universi O, of Tasmania, P.O. Box 1214, Launeeston, Tasmania 7001, Australia

Received 21 February 1996; revised 25 June 1996: accepted 26 June 1996

Abstract

A simple and low-cost amperometric sensor for amines has been developed using a cobalt wire electrode working in alkaline solution. The sensor may be used as a detector for high-performance liquid chromatography (HPLC) that avoids the need for derivatization or post-column reaction. Experimental conditions for flow injection analysis (FIA) and HPLC separation, including the applied potential, pH and concentrations of organic modifier and carrier solution, were optimized. A cobalt wire electrode, in the constant potential amperometric mode, gives an excellent response toward amines in ion-exclusion chromatography in unbuffered solution. The sensitivities of the detection and separation of amines on the column are affected by flow rate, the concentration of the mobile phase and the concentration of organic solvent in the mobile phase, whereas the applied potential only affects the sensitivity of the detector. A cobalt electrode is more sensitive than a copper electrode, and comparable in sensitivity to a UV detector for most amines tested. The detection sensitivity is comparable to that obtained with GC methods, but the procedures are far simpler. The detection limits of the order of nanomoles obtained under the chromatographic conditions used offer an alternative for the determination of amines in a variety of matrices, such as in environmental, biomedical and pharmaceutical samples.

Kevwords: Amines; Amperometric detection; Cobalt electrodes; Ion-moderated partition chromatography

1. Introduction

The determination of amines by gas chro- matography, when effective, is a sensitive tech- nique [1] , but derivatization is required for

* Corresponding author.Fax: (61)2-9385-6141.

non-volatile amines and the polar nature of amines may cause severe tailing of peaks [2]. In the last decade, several electrochemical ap- proaches have been explored to detect amines after separation by high-performance liquid chro- matography (HPLC). HPLC methods for the sep- aration of biogenic amines tend to use reversed-phase sorbents with pre-or post-column derivatization and UV detection. However, this

0039-9140/97/$17.00 Copyright © 1997 Elsevier Science B.V. All rights reserved PII S0039-9140(96)02039-5

240 A. Hidayat et al. / Talanta 44 (1997) 239-248

procedure adds complexity and may lead to problems associated with side-reactions of the derivatizing compound. HPLC on ion-moder- ated partition columns with sodium hydroxide as eluent has been reported to separate up to 10 volatile and non-volatile amines [3]. Electro- chemical detection has employed polarography, square-wave voltammetry, pulsed amperometry and biosensors. The last method used avocado, mushroom and potato tissue on a Clarke oxy- gen electrode [4]. However, the most favourable electrochemical detectors for these compounds are limited to carbon-based electrodes [5-7]. Drawbacks of electrochemical detectors are a high overpotential for the oxidation of amines, which leads to poor selectivity, and the use of derivatization procedures or post-column reac- tions, which also complicates the method [5-7].

Metal or oxidized metal electrodes are pre- ferred to carbon-based electrodes in many re- spects. They are simpler in construction, robust and have lower oxidation potentials for many redox compounds, and therefore have greater sensitivity and selectivity. As a consequence, there has been extensive exploration of metal/ metal oxide electrodes for the constant-potential amperometric detection of organic compunds, including Pt, Au, Cu, Ni, Ag, Pd, Rh, Ir, Fe and W [8-15].

Cobalt, as the phthalocyanine, has also been used as a material for the construction of chemically modified carbon-paste electrodes for many compounds, including amines [16-18]. However, there has been no report on the use of cobalt wire for the constant-potential amper- ometric detection of these compounds. As metallic wire electrodes respond to complexing agents only in alkaline solution (pH ~>6) [12,19,20,21], our attempts were focused on the use of an alkaline solution as carrier. We re- port here the performance of cobalt wire as a simple and low-cost amperometric detector for the determination of amines after HPLC sepa- ration using an ion-exchange column (Aminex HPX-72-O) with sodium hydroxide as eluent. Its action is compared with that of a copper electrode.

2. Experimental

2. I. Apparatus

The cyclic voltammetric study was performed with a BAS-100B potentiostat equipped with Ag/ AgCI (3 M NaC1) reference and platinum wire auxiliary electrodes. An EG&G Princeton Applied Research Polarograph, type 174, was used for constant-potential amperometric detection with a single-line flow injection analysis (FIA) manifold and HPLC systems. The output was recorded and processed by a Macintosh II VX microcomputer.

The flow cell used in both FIA and HPLC detection was made from a Perspex block (5 × 3 × 1 cm) with a 1.2 mm channel allowing the analyte first to contact the working electrode (1 mm diameter wire) along a 1 cm length and then flow away past auxiliary and reference electrodes [22]. The Ag/AgCI reference electrode, prepared by electrolysis at a silver wire electrode in 1.0 M KC1, was coated with 1.0 M KCI in 4% agar gel and was located out of the flowing stream. Each electrode was polished with emery paper and rinsed with deionized water before use. After pre- treatment (see below), the potential was kept fixed at the optimum level. Only after a stable baseline current had been obtained in the supporting elec- trolyte were analytes added and their peak cur- rents determined.

The separation of amines was performed using a Model 510 HPLC pump, U6K injection valve, and Model 484 tunable absorbance detector, all from Waters (Milford, MA, USA) with a resin- based, modified partition column (Aminex HPX- 72-0,300 mm×7.8 mm i.d. x 11 mm o.d.) purchased from Bio-Rad Laboratories (Sydney, Australia).

2.2. Chemicals and solutions

All chemicals were of analytical grade and were used without further purification. Methylamine (me), n-propylamine (pro), n-butylamine (but), trimethylamine (tma), cysteamine hydrochloride (cys) and benzylamine (ben) were purchased from Sigma Chemical (St. Louis, MO, USA). Stock solutions of the amines 10 mM were prepared in

A. Hidayat et al. /Talanta 44 (1997) 239-248 241

Milli-Q water and diluted to the required concen- trations before use. The solutions were filtered through a Millipore 0.45 pm membrane filter and degassed in an ultrasonic bath prior to use. All water was distilled and passed through a Milli-Q water purification system.

2.3. Procedure

Cyclic voltammetric studies (10 cycles) were performed with a potential scan rate of 50 mV s -1. In the flow injection system, the carrier solu- tion was continuously pumped through the detec- tor cell at a constant flow rate until a stable baseline current was obtained. All experiments were performed at room temperature with sodium

08

A

At cobalt vcn'e electrode

04 oo 4 4 -08 -~2

v

B

At copper ~xre electrode f ,~ / ~

1 I I 08 04 00 -04 -08

v

Fig. 1, Cyclic voltammograms at (A) cobalt and (B) copper wire electrodes in 0.1 M NaOH (dotted line) and after addition of 1 mM methylamine (solid line). Sweep rate, 50 mV s -~.

120 '

100'

80"

60'

40'

20'

0

0 Cobalt

200 400 600

mV

800

Fig. 2. FIA peak height of 25 nmol (cobalt electrode) and 20 nmol (copper electrode) of methylamine as function of applied potential. Carrier solution, 0.1 M NaOH: flow rate, 0.75 ml min ~.

hydroxide solution as a carrier. The hydrody- namic voltammograms were obtained in the flow- through cell, point-by-point, allowing for stabilization of the current.

In HPLC, the effects of eluent concentration, electrode pretreatment, addition of organic sol- vent and flow rate on the electrode response were investigated by applying each parameter at vari- ous levels. The electrochemical detector perfor- mance of a cobalt electrode was compared with that of a copper electrode and UV detection. The wavelength used for UV detection recommended by Bio-Rad (210 nm) [23] was chosen with a

Table 1 Effect of pretreatment on the slope of the FIA of methylamine standards at cobalt and copper wire electrodes.

Treatment Slope (nA nmol t)

Cobalt Copper

None b 4.50* 4.07* Cycled 10 times 4.48* 4.06* +2 V for 2 min 4.27** 3.89** +2 V for 10 min 1.41"** 3.71"**

a , ** and *** are statistically significant (95%) groupings by analysis of variance (ANOVA) and least significant difference. b Apply potential at optimum level directly after rinsing with water.

242 A. Hidayat et al. / Talanta 44 (1997) 239-248

Table 2 Effect of mobile phase concentration on the resolution of the most overlapping peaks (methylamine and propylamine)

NaOH (mM) Resolution a

25 0.68 50 0.69 75 0.73

100 1.35

Flow rate, 0.75 ml min-~; sensor, cobalt at +0.40 V vs. Ag/AgC1.

Average of three replicates.

sensitivity setting of 0.1 a.u.f.s. (absorbance units full-scale). All experiments were performed at room temperature with 0.075 M sodium hydrox- ide solution as the mobile phase. The injection volume was 20/11.

3. Results and discussion

3.1. Principles of electrode response

The effect of amines on the electrochemistry at a cobalt surface is twofold. At more positive potentials at which Co TM is formed ( > 0.5 V vs. Ag/AgC1, 0.075 M NaOH), a so-called catalytic oxidation may occur, in which the amine is oxi- dized by Co m, which is regenerated electrochemi- cally [8-11,14-17]. Schematically the reactions are

Co "~ + a m i n e ~ C o II + amine oxidation products(I)

Co ~I ~ Co HI + e (2)

The species present at the surface include oxides and hydroxy compounds such as CoO, Co(OH)2 and CoO(OH) [24,25]. Nickel has been shown recently to participate in these types of reaction [9]. The cyclic voltammograms in Fig. 1 demon- strate this effect for both cobalt and copper in the presence of methylamine, above +0.5 V.

In constrast to the above mechanism, amines may absorb at potentials below that at which they are oxidized, blocking the surface and lead- ing to a reduction in the oxidation current of the metal. As an example of this behaviour, Luo et

al. [11] reported that the addition of glucose and lactic acid reduced the oxidation current of cop- per wire electrodes. Hui and Huber [26] also reported that the oxidation current of a nickel electrode was reduced by 100 nA after addition of 1.0 mM glycine at an optimum potential of 0.55 V. Similarly to the published cyclic voltam- mograms of glucose, lactic acid and glycine at copper electrodes [11,26], our results show that the oxidation currents of cobalt and copper wire in 0.1 M NaOH (dotted lines, Fig. 1), were di- minished at low potentials ( < +0.5 V) with the addition of 1 mM of methylamine (solid lines, Fig. 1). The broad oxidation peak of cobalt at +0.4 V was significantly reduced when 1.0 mM of methylamine was added. A similar trend was observed at a copper electrode with significant depression at the maximum in the oxidation cur- rent (+0.05 V).

There is a difference between the dynamic na- ture of cyclic voltammetry and constant-potential amperometry. However, a similar trend was ob- tained at a constant potential, with a decrease in the current as the amines were introduced into the system at a cobalt wire held at +0.4 V, and an increase in current at a copper wire held at +0.7 V. The behaviour is consistent, and we note that for cobalt at +0.4 V the baseline was maintained constant over many hours of opera- tion.

3.2. Selection of working potential

Flow-injection studies were carried out to de- termine whether cobalt and copper electrodes could be used with ion-exclusion chromatogra- phy. Methylamine standards were injected into a single-line FIA system using NaOH as the carrier solution and the FIA conditions were optimized for high sensitivity with acceptable precision. First the current was recorded as a function of the applied potential to determine the working potential giving the highest response. Amounts of 25 nmol (cobalt electrode) and 20 nmol (cop- per electrode) of methylamine were injected into the flow system at applied potentials between 0.0 and + 0.7 V.

A. Hidayat et al./ Talanta 44 (1997) 239-248 243

The results (Fig. 2) show consistency between FIA and cyclic voltammetry, but the changes in current were greater in the flow system. The great- est reduction in current due to injection of methy- lamine at a cobalt wire occured at +0.40 V. At a copper electrode the FIA peaks steadily increased with increase in applied potential. They were poorly reproducible and the baseline noise was high, giving poor precision when used to calibrate for amines. Subsequent work was therefore con- centrated on the use of a cobalt electrode for practical analysis of amines.

3.3. Electrode pretreatment and position

Stitz and Buchberger [9] reported that the sensi- tivity of a nickel electrode for the detection of carbohydrates and related compounds is strongly influenced by the pretreatment of the electrode surface. Procedures affecting the amount of oxide on the electrode are likely to affect the sensitivity. In this experiment, after polishing with emery paper and rinsing with dionized water, the elec- trodes were pretreated with four different proce- dures as follows: (1) the potential was kept fixed at the optimum working level: (2) the electrodes were cycled initially in the blank solution over the desired potential range ( - 0 . 4 to 1.1 V), for 10 cycles; (3) a potential of + 2 V was applied for 2 min; (4) a potential of +2 V was applied for 10 min.

Table 1 shows that the highest sensitivity was achieved by applying the optimum potential di- rectly after polishing with emery paper and rins- ing with deionized water. The same sensitivity was shown when the electrode was cycled 10 times.

Table 3 Effect of organic modifier (methanol) on the peak height and the resolution of the most overlapping peaks (methylamine and propylamine)

Methanol Peak height (nA) concentration (%) Methylamine Propylamine

Resolution

0 15 3.4 0.73 0.25 7.4 2.0 1.07 0.5 4.6 1.3 1.08

A

8.5 nA 6 nmol

4 ITUn.

I

6 nmol ~ 9 nA

Fig. 3. Typical FIA peaks of the methylamine standards de- tected at (A) cobalt and (B) copper electrodes. Carrier solu- tion, 0.075 M NaOH; flow rate, 0.75 ml min ~ ~; potential, 0.4 V (cobalt) and 0.5 V (copper).

The sensitivity was reduced significantly when the electrode was pretreated by oxidation at 2 V. The longer the oxidation at a cobalt electrode, the lower was the sensitivity. Completing a thick layer of oxide by this treatment would leave little to be affected by an absorbed amine.

3.4. Concentration of NaOH in the mobile phase

The effect of the concentration of the mobile phase on the detection response was examined to obtain a compromise between maximum sensitiv- ity and solute retention on the chromatographic column. Since only two amines were overlapped (methylamine and propylamine), we focused on the separation of these two compounds.

Table 2 shows the effect of the concentration of NaOH in the range 0.025-0.100 M on the resolu- tion of selected amines. The retention times of

244 A. Hidayat et al. / Talanta 44 (1997) 239 248

compounds increased slightly as the concentration of sodium hydroxide increased, which improved the resolution. The best overall resolution of six compounds was achieved by using 0.075 M sodium hydroxide as the mobile phase. The reso- lution of methylamine and propylamine increased from 0.68 to 1.35 on increasing the NaOH con- centration 0.025 to 0.10 M (Table 2). Similar observations have been reported where the con- centration of ion-interaction reagent in the mobile phase increased, leading to a corresponding in- crease in solute retention [3,13].

An optimum concentration of alkaline medium is essential to develop the oxide film on the sur- face of the electrodes to respond sensitively to amines. Detailed studies by UV-visible reflec- tance spectroscopy and FTIR spectrometry at a nickel electrode surface showed that the main requirement for sensitive detection consists in the formation of a nickel(III) oxide-hydroxide layer in alkaline medium [9]. The same conditions may also be expected to apply to cobalt and copper electrodes. At low alkaline concentration, the oxi- dation process may not fully develop, and the oxide film is not sufficient to respond to the addition of amines. At high alkaline concentra-

A

4.5hA I

6hA

4 n'lJll.

Fig. 4. FIA peaks showing the precision for 12 mmol methy- lamine detected at (A) cobalt and (B) copper electrodes. Carrier solution, 0.075 M NaOH, flow rate, 0.75 ml min-~; potential, 0.4 V (cobalt) and 0.5 V (copper).

tion, however, an unfavourable response occurs because of increasing participation of the equi- librium involving hydroxide ions. We found that the lowest detection limit was achieved at the optimum NaOH concentration of 0.075 M. Stulik et al. [13] also reported that copper electrodes respond to complexing agents only in a solution of pH > 6.0, and that the sensitivity increases with increasing pH and decreasing ionic strength.

3.5. Flow rate

Decreasing the flow rate improves the resolu- tion of the overlapping peaks, and the best com- promise between resolution and retention time was achieved at a flow rate of 0.75 ml min ~. At higher flow rates, propylamine moves in and masks the methylamine peak. These results sup- port the previous report [22] that the optimum flow rate was achieved at 0.6 ml min -1. Mass transfer in Aminex columns is slow, requiring a commensurately low flow rate to avoid band broadening.

3.6. Concentration of organic solvent

As the sensor is used in conjuction with the HPLC system, the effect of added organic solvent was studied. The addition of solutions of methanol between 0.1% and 2% to the carrier solution decreased the response of the cobalt wire electrodes to amines. The result is in agreement with those described in previous studies [13]. Ac- cording to Stulik et al. [13], the presence of or- ganic solvent suppresses the solubility of the passivating film and the interaction of the test substances with the passivating film. This in turn decreases the relative permittivity in the carrier, which generally causes a decrease in the rate constants for complex reactions. In contrast to the above, we found a positive response at a copper wire electrode when amine was injected into a carrier solution containing methanol. This result is also consistent with several reports on the am- perometric detection of organic compounds by a copper wire electrode in an alkaline medium [9,11,13], in which direct oxidation yielding posi- tive currents occurs.

A. Hidayat et al.

Table 4 Comparison of stability of cobalt and copper electrodes

Relative response (%)

Injection number or time Cobalt Copper

10 100.0 100.0 20 99.5 99.0 50 102.4 74.4 After 3 h 99.2 55.0 After 6 h 100.8 42.5 After 9 h 99.4 RSD ('V,,) (n = 10)" 1.3 1.9

Electrode potential (vs. Ag/AgC1), Co +0.94 V and Cu +0.5 V: sample, 12 nmol methylamine. ~ (For the first 10 injections).

Although the FIA experiments at cobalt sug- gested that the addition of methanol signifi- cantly reduced the peak height, we invesitgated whether the addition of a low methanol con- centration could improve the resolution of the amine peaks. An organic modifier penetrates and swells the organic backbone of the resin, so that decreased osmotic pressure decreases the intra-particle water volume. This to some extent may affect the resolution of the peaks.

The results in Table 3 show that the im- provement in resolution of ovelapping peaks was not significant when methanol was added to the mobile phase. On the other hand, the peak heights were significantly reduced and the baseline was drifting. Therefore, the addition of organic modifier is not recommended with this method.

The effect of atmospheric oxygen in the solu- tions was studied, since the dissolution of cop- per in complex-forming media often depends on the presence of oxidants. However, no change occurred when oxygen was removed by passage of nitrogen for 30 min. Therefore, we conclude, in agreement with Stulik et al. [13], that the presence of atmospheric oxygen does not affect the detection.

Fig. 3 shows typical peaks obtained in the FIA system when responding to methylamine using both metallic cobalt and a copper wire electrode as detectors. Various structural types of amines

Talanta 44 (1997) 239 248 245

were also investigated. Aliphatic primary, sec- ondary and tertiary and aromatic amines can be

/ /

. . . . I I

1

,si i i I

'i 5

J~

f , I /

C

5na r i

If V ' 3

t / \t3 'iL! , ',J

B 2 4

I0 min

Fig. 5. Chromatograms of a mixture of aliphatic amines sepa- rated by ion-moderated partition chromatography and de- tected by (A) cobalt electrode, (B) copper electrode and (C) UV detector: Peaks 1= methylamine: 2 - propylamine: 3 = butylamine; 4 = cysteamine hydrochloride; 5 = trimethy- lamine; 6 = benzylamine.

246 A. Hidayat et al. / Talanta 44 (1997) 239-248

Table 5 HPLC calibration data for selected amines with amperometric detection at a cobalt electrode at +0.4 V (vs. Ag/AgCI)

Analyte Linear concentration Slope (nA nmol-~) r 2" LOD (nmol) b range (nmol)

Methylamine 2.6 130.0 1.65 0.989 0.46 n-Propylamine 3.4-101.0 0.77 0.984 0.98 n-Butylamine 5.4-135.0 0.65 0.987 1.15 Trimethylamine 4.2- 91.0 0.79 0.989 1.92 Benzylamine 9.2 - 166.2 0.12 0.999 6.30

Mobile phase 0.1 M NaOH; flow rate 0.75 ml min -~ Coefficient of determination (the fraction of variance accounted for by the linear model).

b LOD (limit of detection) = concentration that yields a current of three times the standard deviation of the background noise.

detected by this method. The sensitivity de- creased with increasing chain length and with increasing number of substituents on the nitro- gen atom.

3. 7. Prec&ion and s tabi l i ty

The precision of the electrode response was examined by injecting a sequence of 10 repli- cate samples of 12 nmol methylamine under the opt imum conditions, using both cobalt and copper electrodes (Fig. 4). The mean values of the peak heights recorded were 23.0 and 14.3 nA with relative standard deviations (RSDs) of 1.3 and 1.9% for the cobalt and copper elec- trodes, respectively.

The stability of the electrode was tested by injection of 10, 25 and 50 replicate samples af- ter 3,6 and 9 h without resurfacing the elec- trodes. The results in Table 4 indicate that good stability is obtained particularly with the cobalt electrode. The response eventually begins to decrease after 3 - 4 days of use but the for- mer sensitivity was easily restored by cycling 10 times between - 0 . 4 0 and +2.6 V. The long- term stability of the reference electrode must also be monitered. Redox processes at each of the electrodes may lead to poisoning and loss of performance.

The great stability offered by these systems may be due partly to the principle of detection, i.e. oxidation at a reversible layer consisting of high-valent metal oxide and metal hydroxide. The response thus results from the interaction

of the analyte with higher valent metallic ions contained in the outer layer. This layer can be converted very rapidly to and from a layer of lower valence metal oxide, electrochemically or chemically thus renewing itself.

The detector cell design, which eliminates possible reactions with the Ag/AgCl reference electrode, may also contribute to the good sta- bility of these sensors.

m 8.

40

= . + - . = .

30

20

10

0 1 0 2 0 3 0 4 0

UV result/nmol

Fig. 6. Amperometric sensor with cobalt wire compared with UV spectrophotometric detection for amines. UV wavelength, 275 nm, carrier solution, 0.1 M NaOH; flow rate, 0.75 ml min 1.

A. Hidayat et al. / Talanta 44 (1997) 239-248 247

3.8. Analytical performance of the electrodes in the HPLC system

Typical chromatograms of six amines under the optimum chromatographic conditions, using an amperometric detector with cobalt and copper electrodes and a UV detector for comparison, are shown in Fig. 5. The amines separated in order of elution were methylamine, propylamine, buty- lamine, cysteamine hydrochloride, trimethylamine and benzylamine, with retention times of 10.1, 12.2, 19.8, 24.3, 39.0 and 67.9 min, respectively. All peaks were resolved sufficiently to allow the HPLC system to be of practical use for determining the six amines.

A good separation of the six amines was achieved in this system using only a 0.075 M NaOH carrier, without pre- or post-column deriva- tization. The retention times of isopropylamine and isobutylamine are identical with those of n-propylamine and n-butylamine, which suggests that the isomeric amines cannot be separated with this method.

The order of elution suggests that the less polar compounds (benzylamine) are retained more strongly than more polar compunds (methy- lamine), which shows that a reversed-phase parti- tion is involved in this separation.

The detectors used, amperometric with cobalt and copper wire electrodes and a UV detector, gave similar chromatograms, with reasonable base- line separations of the amines tested. The RSDs of the retention times are < 2.5%, which is good for an HPLC method.

3. 9, LineariO,, sensitivity and limit of detection

The calibration data for selected amines with amperometric detection at cobalt are shown in Table 5. The calibration graphs exhibit good lin- earity (correlation coefficients ranging from 0.984 for n-propylamine to 0.999 for benzylamine). The linear range of the voltammetric detector was between 30 (for benzylamine) and 300 (methy- lamine) times the limit of detection.

There was also a difference between the sensitiv- ities of the various compounds shown by the amperometric sensors: the simpler the amines, the

higher was the sensitivity. Their detection limits were consistent with their sensitivity which follows their degree of structural complexity. Methylamine could be detected at levels as low as 0.5 nmol whereas benzylamine could not be detected until about 6.3 nmol was present. The cobalt electrode exhibited higher sensitivities than the copper elec- trode for all amines tested. Compared with the electrode resonse in FIA, the amperometric re- sponse toward separated amines with a metallic electrode is lower. This discrepancy is probably due to the higher dispersion in the HPLC system and the difference in dynamic conditions between the two experiments.

Satisfactory precision (the RSD of the peak height was < 2%) was obtained with five repli- cates, even at the lowest measured concentrations. Compared with a chemically modified electrode (CME) and mercury/gold amalgam electrodes, a cobalt electrode is more stable. A newly resurfaced gold amalgam electrode may operate efficiently for only a few minutes or hours. Using a CME, after some time the response is reduced. Some of the decrease in the electrode response is due to the gradual leaching of the modified chemicals from the CME surface. In fact, some decrease in re- sponse has been observed previously even for conventional carbon paste electrodes in electro- chemical detection in liquid chromatography upon long-term exposure to a binary mobile phase con- taining a small fraction of organic components.

The design of the Ag/AgCI reference electrode may also contribute to the stability of this sensor. In this cell design, the silver wire which has been electrolysed with KCI solution was covered with solid agar gel containing 1.0 M KC1. The function of the solid agar gel containing 1.0 M KC1 is similar to that of the Ag/AgCI reference electrode with a double junction.

The retention times are comparable to those achieved by Stamler and Loscalzo [27] using capil- lary zone electrophoresis for separation. As deriva- tization is not necessary, our method is simpler and cheaper. To our knowledge, this represents the first report on the use of cobalt and copper electrodes for constant-potential amperometric detection in conjunction with ion-moderated partition HPLC for the determination of amines.

248 A. Hidayat et al. / Talanta 44 (1997) 239-248

3. I0. Comparison with UV spectrophotometric detection

The performance in amine determination of a cobalt electrode was compared with that of a copper electrode and UV detection using the wavelength suggested by the instrument manu- facturer [23]. Fig. 6 demonstrates the close agreement between amperometric detection with a cobalt electrode and UV detection.

Fig. 5 shows that compared with a UV de- tector, the cobalt electrode is more sensitive for the detection of methylamine and butylamine but less sensitive for propylamine, trimethy- lamine and benzylamine. Benzylamine absorbed UV radiation more strongly than other amines which are not aromatic. The UV detector placed prior to the amperometric detector also caused more dispersion. The constraints of working at UV wavelengths is that severe inter- ference by organic compounds other than amines may occur.

References

[1] X.H. Yang, C. Lee and M.I. Scranton, Anal. Chem., 65 (1993) 572.

[2] H.B. Hucker and S.C. Stauffer, J. Chromatogr., 138 (1977) 437.

[3] R.B.H. Wills, J. Silalahi and M. Wootton., J. Liq. Chro- matogr., 10 (1987) 3183.

[4] F. Mazzei, F. Botre, M. Lanzi, G. Lorenti, F. Porcelli and C. Botre, Sens. Acutuators, 7 (1992) 427.

[5] K. Shimada, T. Oe, M. Tanaka, and T. Nambara, J. Chromatogr., 487 (1989) 247.

[6] J.B. Barek, F. Pacakova, K. Stulik and J. Zima, Talanta, 32 (1985) 279.

[7] R.M. Smith and A.A. Ghani, Electroanalysis, 2 (1990) 167.

[8] T. Ueda, R. Mitchell and F. Kitamura., J Chromatogr., 592 (1992) 229.

[9] A. Stitz and W. Buchberger., Electroanalysis, 6 (1994) 251.

[10] A. Stitz and W. Buchberger, Fresenius' J. Anal. Chem., 339 (1991) 55.

[11] P. Luo, F. Zhang and R.P. Baldwin, Anal Chim. Acta., 244 (1991) 169.

[12] P. Luo, F. Zhang and R.P. Baldwin, Anal. Chem., 63 (1991) 1702.

[13] K. Stulik, V. Pacakova, K. Le and B. Hennissen, Talanta, 35 (1988) 455.

[14] R.E. Reim and R.M. Van Effen, Anal. Chem., 58 (1986) 3203.

[15] P.W. Alexander, A. Hidayat and D.B. Hibbert, Electro- analysis, 7 (1995) 290.

[16] E.G. Cookeas and C.E. Efstathiou, Analyst, 119 (1994) 1607.

[17] K. Korfhage, M.K. Ravichandran and R.P. Baldwin, Anal. Chem., 56 (1984) 1514.

[18] X. Qi, and R.P. Balwin, Electroanalysis 6 (1994) 353. [19] W.Th. Kok, H.B. Hanekamp, P. Bos and R.W. Erei,

Anal. Chim. Acta, 142 (1982) 31. [20] K. Stulik, V. Pacanova, M. Weingart and M. Podolak, J.

Chromatogr., 367 (1984) 311. [21] T.A. Gill and J.W. Thompson., J. Food Sci., 49 (1984)

603. [22] P.W. Alexander, P.R. Haddad and M. Trojanowicz, Anal

Chim. Acta, 171 (1985) 151. [23] Guide to Aminex HPIC Columns for Food and Beverage

Analysis, Bio-Rad Richmond, CA, 1993. [24] R.D. Cowling and A.C. Riddiford, Electrochim. Acta, 5

(1969) 627. [25] N.A. Hampson, Electrochemistry, 2 (1972) 138. [26] B.S. Hui and C.O. Huber, Anal. Chim. Acta, 134 (1982)

211. [27] J.S. Stamler and J. Loscalzo, Anal. Chem., 64 (1992) 779.