ANALYST, JUNE 1991, VOL. 116 573 Direct Reductive Amperometric Determination of Nitrate at a Copper Electrode Formed ln Situ In a Capillary-fill Sensor Device Arnold G. Fogg, S. Paul Scullion and Tony E. Edmonds Chemistry Department, Loughborough University of Technology, Loughborough, Leicestershire LE 1 I 3TU, UK Brian J. Birch Unilever Research, Colworth House, Sharnbrook, Bedfordshire MK44 ILQ, UK A method has been developed for determining nitrate amperometrically by direct reduction at a freshly deposited copper electrode surface in a capillary-fill device (CFD). Copper(l1) is added to the nitrate sample which is then taken up into the device. The potential of the screen-printed carbon electrode is held at -0.75 V versus the screen-printed silver reference electrode. At this potential, copper is plated onto the carbon electrode forming a freshly prepared copper electrode.At the same time dissolved oxygen is reduced. The potential is then scanned to more negative potentials and the signal at -0.90 V, due to the reduction of the nitrate, is measured. The method for determining nitrate given here is preliminary to the production of CFDs in which chemical reagents, copper sulphate and potassium hydrogen sulphate (used to produce the acidity), are screen-printed or otherwise coated onto the upper plate within the device. Keywords: Capillary-fill device; disposable sensor; nitrate determination; amperometric detection; in situ copper electrode Several of the methods that have been developed for use on-line in flow injection with amperometric detection appeared to be ideally suited for use in the capillary-fill device (CFD) that has been developed and patented by Unilever Research.These methods were listed in the introduction to the previous paper in the present series,l together with a description of the Unilever device. In this previous paper1 a preliminary study of the adaptation of a flow injection method for the amperometric determination of phosphate, based on its on-line reaction with an acidic molybdate reagent, followed by the electrochemical reduction at a glassy carbon electrode of the 12-molybdophosphate formed, for use in the CFD, was described. Pre-formed 12-molybdophosphate was shown to respond reproducibly and rectilinearly within the device. Recently, flow injection methods for the determination of nitrate, based on an on-line nitration reaction2 and on its on-line reduction to nitrosyl chloride ,3 have been developed.However, these methods require the injection of concentrated sulphuric acid and cannot be adapted easily for use in CFDs. The direct reduction of nitrate is not generally possible at the more commonly used electrodes, such as glassy carbon, platinum, gold and mercury electrodes. However, several voltammetric methods have been reported for the determina- tion of nitrate based on the use of solid electrode materials that are capable of catalysing the reduction of nitrate.4-13 Davenport and Johnsonh.7 used a rotating cadmium electrode which gave a linear response over a narrow range of nitrate concentration and a limit of detection of 1 x 10-4 mol dm-3 for nitrate.Bodini and Sawyer4 observed that the reduction of nitrate was catalysed by the simultaneous deposition of copper and cadmium at a pyrolytic graphite electrode and obtained a limit of detection of approximately 1 x 10-6 mol dm-3 nitrate. Johnson and Shenvood8 used a rotating cadmium electrode coated with a layer of metallic copper to determine nitrate down to 1 X 10-4 mol dm-3 and adapted the technique for use with high-performance liquid chromatography.' Xing and Scherson1(),11 described a rotating ring-disc electrode method for the determination of nitrate in acidic media. Their method i s based on the measurement of the ring currents associated with the oxidation of nitrite ions that are generated by the reduction of nitrate ions on a gold disc electrode covered by a layer of underpotentially deposited cadmium. An advantage of this technique is that the background currents for the oxidation process are significantly lower than those for the reduction of nitrate, and limits of detection were of the order of 20-30 ppb of nitrate.The use of copper cathodes for the direct reduction of nitrate has also been investigated.5.12 Pletcher and Poor- abedil2 found that the reduction was particularly sensitive to halide ions which had the effect of shifting the nitrate reduction wave to more negative potentials. Albery et aZ.5 used a packed bed wall-jet electrode to measure nitrate in a flowing stream and were able to re-generate the electrode after each measurement; reduction is only reproducible on a fresh copper surface.Almhofer and Frenzel13 described recently a flow injection method for the determination of nitrate using a tin electrode as the amperometric detector. The measurement range was 5 X 10-5-1 X 10-2 mol dm-3 of nitrate but the response was non-linear and chloride, nitrite and sulphate interfered when they were present at concentrations greater than the nitrate. All the methods mentioned above require solutions to be de-oxygenated before the measurement is made and the electrode surface had to be re-formed between measurements owing to poisoning of the electrode surface. Methods using cadmium are undesirable in a hydroponic environment owing to possible contamination problems. As indicated above, the methods of determining nitrate in a flow injection system273 involved the injection of concentrated sulphuric acid and are unsuited for easy adaptation for use in the CFD.For that reason the flow injection method of Albery et aZ.5 has been adapted here for that purpose. The work of both Pletcher and Poorabedi12 and Albery et aZ.5 has indicated that nitrate can be reduced directly at a copper electrode but only if the electrode surface is freshly prepared. In the method developed here copper ion is added to the nitrate sample and copper is deposited freshly on the screen-printed carbon electrode in the CFD before the device is used to determine nitrate reductive1 y . Experimental The CFDs have been described in detail previously.' They were filled by dipping the ends of the devices that were remote from the electrode connections into the appropriate solution such that the solution was taken up by capillary action.The574 ANALYST, JUNE 1991, VOL. 116 CFDs were placed flat on the bench before a potential sweep or step was applied. Details of the voltammetric experiments applied in the present study have been described in full previously.' Reagent Solutions Copper(r1) sulphate pentahydrate. A 1 x 10-1 mol dm-3 solution was prepared by dissolving 12.48 g of CuS04.5H20 in 500 ml of distilled water. Potassium chloride (KCl). A 1 x 10-2 rnol dm-3 solution was prepared by dissolving 0.373 g of KCl in 200 ml of distilled water and diluting to 500 ml. Potassium nitrate (KN03). A 1 x 10-2 rnol dm-3 solution was prepared by dissolving 0.506 g of KN03 in 200 ml of distilled water and diluting to 500 ml. Sulphuric acid (2 rnol dm-3 H2S04).A 2 rnol dm-3 solution was prepared by carefully adding 54.3 ml of the concentrated acid to 300 ml of water, cooling and diluting to 500 ml. Solutions for voltammetry were prepared by adding appropriate amounts of the stock solutions (described above) to a 50 ml calibrated flask and diluting to the mark. When testing the effect of various ions on the response, a stock solution of either 5000 or lo00 mg 1-1 of the ionic component was prepared and the appropriate amount was added to the calibrated flask before dilution. Results and Discussion The reduction of nitrate at a freshly polished copper rod electrode gave a fairly well defined wave with a peak potential of -0.49 V in 0.5 mol dm-3 sulphuric acid (see Fig.1). The effect of chloride ions on the nitrate reduction at this electrode is also shown in Fig. 1; the peak potential is shifted to increasingly negative potentials and eventually the peak merges with the background current as the chloride concentra- tion is increased. The small peak at -0.36 V, which was not present when the solution had been previously de-oxygen- ated, was also dependent on the chloride concentration, being higher and narrower when chloride was present. Pletcher and Poorabedil2 investigated the reduction of nitrate at a copper electrode and showed that in the over-all reaction the nitrate was reduced to ammonia. The reaction was shown to be 60 50 40 f ? 30 3 u controlled by diffusion at high acidities 20 10 0 -0.7 -0.6 -0.5 -0.4 -0.3 -0.2 -0.1 PotentialN versus SCE Fig. 1 Effect of chloride on the linear sweep voltammogram for the reduction of nitrate at a newly polished copper rod electrode.Nitrate concentration, 1 X mol dm-3; sulphuric acid concentration, 0.5 mol dm-3; scan rate, 5 mV s-1. Chloride concentration: A, 0; B, 0.1 x 10-3; C, 1.0 x 10-3; and D, 2.5 x 10-3 mol dm-3 (>0.1 rnol dm-3 perchloric acid) but no reaction could be detected at a pH>3. They speculated that the effect of chloride ions on the half-wave potential of the nitrate reduction was due to a double-layer effect, with the chloride ions being adsorbed at the electrode. Their attempts to prove this hypothesis by double-layer capacitance experiments were unsuccessful due to the presence of faradaic currents over most of the potential range.It was envisaged in the present work that any sensor employing a copper electrode would be prone to the formation of an oxide film on storage,l4 and that reproducibility would be improved by generating the copper electrode in situ immediately prior to use. Copper was electrolysed from a 1 rnol dm-3 sulphuric acid solution onto glassy carbon and screen-printed silver and carbon electrode surfaces, and the reduction of nitrate at these surfaces was studied. As shown by the results in Table 1, the surface on which the copper was deposited had a marked effect on the nitrate reduction peak potential at the screen-printed electrodes, presumably due to the greater resistance of these screen-printed electrodes. The screen-printed silver electrode gave an additional peak at approximately -0.3 V on the first scan which was not observed in subsequent scans.Oxide formation on the silver surface could have occurred on exposure to the atmosphere.14 Hitchman and co-~orkers15,16 investigated the use of silver electrodes for the potentiometric determination of amino acids and found that chemical or electrochemical cleaning of the electrode improved the repeatability of the results. Whatever the cause of the additional peak it was decided here to concentrate on using the screen-printed carbon ink elec- trodes in which the carbon overlaid a silver coating; these electrodes have a lower resistance than those in which carbon is screen-printed directly onto the substrate. Table 1 Effect of electrode substrate on nitrate reduction at a copper plated electrode.Scan speed, 10 mV s-I; plating time, 2 min at -0.5 V (-0.3 V for silver electrode); solutions de-oxygenated Peak potentialN versus SCE Bare electrode surface Glassy carbon -0.490 -0.550 Screen-printed silver -0.51s Screen-printed carbon on silver Screen-printed carbon Wave poorly defined 1 .o 0.8 2 0.6 2 E 0.4 0.2 I 0 -0.1 -0.2 -0.3 -0.4 -0.5 -0.6 -0.7 PotentialN versus SCE Fig. 2 Typical linear sweep voltammograms for the reduction of nitrate at a copper plated screen-printed carbon ink on silver electrode. Sulphuric acid concentration, 0.5 mol dm-3; scan rate, 10 mV s-1. Nitrate concentration: A, 0; B, 1.0 X C, 2.5 X lo-'+; D, 5.0 x 10-4; E, 7.5 x 10-4; and F, 10.0 x mol dm-3ANALYST, JUNE 1991, VOL. 116 575 De-oxygenation of the plating solution made little dif'fer- ence to the signals obtained.Typical voltammograms obtained for nitrate reduction at a copper plated screen- printed electrode are shown in Fig. 2; the peak at -0.3 V can be removed by de-oxygenation. The use of differential-pulse voltammetry gave a better separation of the oxygen and nitrate peaks but some overlap was still present. The voltammetric response of nitrate reduction over a measured concentration range of' 1 x 10-4-1 x 10-3 rnol dm-3 in non-de-oxygenated solutions is shown in Table 2. Chloride shifted the nitrate reduction peak to more negative potentials, as with the copper rod electrode, but the shift was proportion- ately greater at an equivalent chloride concentration.The nitrate reduction peak became indistinguishable from the background reaction at a chloride concentration of 5 x 10-4 rnol dm-3 as compared with a similar effect at a chloride concentration of 5 x 10-3 mol dm-3 when using the copper rod electrode. It was found that nitrate could be measured directly in the copper plating solution, because the peak potential for copper deposition was 200 mV more positive than that for nitrate reduction (Fig. 3). Thus, it should be possible to add copper(ii) sulphate to nitrate samples, which are to be presented to the CFD, and to deposit copper from the sample solution before determining the nitrate Concentration. The effect of chloride on the response at the screen-printed Table 2 Voltammetric characteristics of nitrate reduction at a copper plated carbon ink on silver electrode (all potentials versus SCE) Nitrate concentration/ 10kJ rnol dm--' 1 2 4 6 8 10 LSV* D PVt E,IV i&A -0.55s 30 - 0.540 82 - 0.545 170 -0.565 300 -0.570 45s - 0.550 640 E,IV idClA -0.530 250 -0.530 570 -0.535 1250 -0.455 1950 -0.545 2700 -0.540 3450 * Linear sweep voltammetry. Scan speed, 10 mV s-I.t Differential-pulse voltammetry. Scan speed, 2 mV s - * ; pulse amplitude, 25 mV; and pulse interval, 1 s. 1.6 1.4 1.2 Q 1 E 0.8 L 3 0.6 0.4 0.2 0 -0.8 -0.6 -0.4 -0.2 0.0 PotentialN versus SCE Fig. 3 Linear sweep voltammograms of a solution containing copper(i1) and nitrate at a screen-printed carbon ink on silver electrode. Sulphuric acid Concentration. 0.5 rnol dm-3: copper sulphate concentration, 1 x 1W2 rnol dm-3; nitrate concentration, 1.0 X mol dm-3; scan rate, 5 mV s-1.Chloride concentration: A, 0; B , 1 x 10-3; C, 2.5 x 10-3; and D, 7.5 x 10-3 rnol dm-3 electrode, also shown in Fig. 3, is to heighten and narrow both peaks, with the copper peak potential being relatively unchanged whilst the nitrate peak potential was shifted to more negative values with increasing chloride concentration. Attention now was turned to studying the determination of nitrate in the CFD. The effect of chloride on the copper deposition process in the CFDs was to heighten and narrow the peak as shown in Fig. 4. Typical voltammograms obtained for nitrate reduction in a CFD, with and without the addition of chloride, are shown in Fig. 5. Two main peaks were ob- tained; the copper deposition peak at approximately -0.58 V and the nitrate reduction peak at approximately -0.88 V.The peak potentials obtained are typically more negative than those obtained in bulk solution, presumably due to the nature of the reference electrode [Ag-Ag+ as opposed to the saturated calomel electrode (SCE)]. It is evident that the presence of chloride is beneficial in that a complete separation of the copper and nitrate peaks can be achieved, making 200 150 $. ? : 100 3 0 50 0 -0.2 -0.4 -0.6 -0.8 -1.0 PotentialN versus internal reference electrode Fig. 4 Linear sweep voltammograms showing the effect of chloridc on the copper deposition process in a CFD. Scan ratc. 2 mV s-I; sulphuric acid concentration, 1 rnol dm--'; chloride concentration: A and C, 0; B and D, 2 x 10k4 mol dm--3.Copper(i1) concentration. A and B. 0; C and D , 5 x 10-3 rnol dm-3. No nitrate added 250 200 5 1 5 0 . w 2? 3 100 50 0 B -0.2 -0.4 -0.6 -0.8 -1.0 Potent i a IN versus i n t e r n a I reference electrode Fig. 5 Linear sweep voltammograms showing the effect of chloridc on the reduction of copper(i1) and nitrate in a CFD without a delay during the scan to pre-reduce the nitrate. Scan rate, 2 mV s-I; sulphuric acid concentration, 1 rnol dm-3; nitrate concentration, 1.0 x 10-3 mol dm-3; copper(i1) concentration. 5 X lo-' rnol dm-3. Chloride concentration: A, 0: and B. 2 X rnol dm-'576 0.18 0.16 0.14 a 2 €0.12 2 50.10 0 0.08 0.06 0.04 ANALYST, JUNE 1991, VOL. 116 - - - - - - - - measurements easier. An estimate of the peak area showed no change in the amount of charge passed during the deposition process, but the rate at which it was passed increased as chloride concentration increased.Tam and Christiansen17 studied the effect of chloride in copper plating baths on the electrochemical processes occurring and found that chloride enhanced the rate of deposition of copper on platinum electrodes. They attributed this rate enhancement to the formation of a metakhloro-copper(u) complex which they believed lowered the activation energy for the reduction of cu2+. Another possible explanation for the changes in the shape of the voltammograms upon addition of chloride is the stabilization of the reference electrode. The silver ions produced at the counterheference electrode would be precipi- tated out of solution by the chloride ions present, thereby reducing the change in the silver electrode potential.It should be noted, however, that the concentration of silver ion above the silver electrode could become fairly large during the period of copper deposition. Calculations show that this concentration of silver ion could be as great as 1.47 x 10-2 rnol dm-3. A chloride concentration of 1 X 10-4 mol dm-3 would not be expected to reduce the silver ion concentration significantly; in the thin-film device the lateral diffusion of chloride is negligible. In previous work involving the develop- ment of a CFD for the determination of phosphate,' a 1 x 10-1 mol dm-3 chloride concentration was adopted for use in the device and the potentials obtained were approximately those that would be expected with a silver-silver chloride reference electrode. The reduction potentials obtained here are 400 mV more negative than those obtained under semi-infinite diffusion conditions, which is more consistent with the standard potential for the silver-silver ion couple.By increasing the chloride concentration the nitrate reduc- tion peak is shifted to more negative potentials and the peak is increased in height and decreased in width. Again it is thought that a stabilization of the reference potential may be respons- ible for these effects. A greater tolerance to large concentra- tions of chloride was apparent with the CFD, as is shown in Fig. 6; the nitrate reduction peak is clearly distinguishable from the background reaction even at a chloride concentration of 1 x 10-2 rnol dm-3, in contrast to the situation with the screen-printed electrode in bulk solution (see Fig.3). Thus, the major difference between the measurements in the CFDs and measurements in bulk solution is the greater degree of separation between the nitrate reduction peak and the background reaction. This is thought to be due to differences in the peak potentials obtained when using thin-layer cells. A comparison of the equations for peak potentials under both semi-infinite diffusion and thin-layer conditions shows that reductions occur at more positive values in the latter instance. For reversible electrode reactions there is little difference in the peak potential for semi-infinite diffusion conditions: E,, - EO = -28.5/n millivolts at 25 "C (where n = number of electrons, E,, = peak potential for the cathodic process and Eo = standard electrode potential) and for thin-layer conditions E,, = Eo. For irreversible electrode reactions, however, the ana- logous equations for peak potential are: for semi-infinite diffusion conditions an,Fv E -Eo= - - RT [ 0.78 + In (s) + In (F)'] PC arn,F and for thin-layer conditions log[ ] 2.303 RT E -Eo=---- PC nllF an,F(-v)V where ar = transfer coefficient; n, = number of electrons involved in the rate determining step; v = scan rate in V s-1; o.20 I 0.02 1 ' I I I 1 -1.3 -1.2 -1.1 - 1 -0.9 -0.8 PotentialN versus internal reference electrode Fig.6 Linear sweep voltammogram showing the effect of chloride on the reduction of nitrate at an in situ generated copper electrode in a CFD.Scan rate, 2 mV s-1; sulphuric acid concentration, 0.5 rnol dm-3; nitrate concentration, 1 X rnol dm-3; copper(r1) concentration, 1 X rnol dm-3. Chloride concentration: A, 0; B, 2.5 x 10-3; C, 5 x 10-3; and D, 10 x 10-3 mol dm-3 Table 3 Comparison of the peak potentials obtained for irreversible electrode reactions at different values of ko and (2: Epc - EON Semi-infinite kolcm s- (2: Thin-layer cells diffusion 1 x 10-6 0.25 -0.471 -0.706 1 x 10-6 0.50 -0.271 -0.371 1 x 10-6 0.75 -0.194 -0.254 1 x 10-8 0.25 -0.944 -1.179 1 x 10-8 0.50 -0.508 -0.607 1 x 10-8 0.75 -0.352 -0.411 ko = heterogeneous rate constant in cm s-1; A = area of thin-layer cell in cm2; V = volume of thin-layer cell in cm3; and R, T, F and D have their usual significance.Using the following typical values: n = 1, v = 2 mV s-1, AIV = 50 pm-1, T = 298 K and D = 1 x 10-5 cm2 s-1, a comparison of peak potentials at different values of ko and ar are presented in Table 3. The advantage of using thin-layer cells in order to study irreversible reactions is apparent. The linear range of the reaction was investigated using the same CFD throughout. Measurements were made by holding the potential at -0.75 V until the current decayed to a steady value and then applying a negative potential scan. At a potential of -0.75 V it was found that copper was deposited effectively and the oxygen was reduced, whilst the reduction of nitrate was minimal. Typical voltammograms are shown in Fig. 7 and the voltammetric response obtained over the nitrate concentration range studied is given in Table 4.The measure- ment of peak height versus concentration gave a linear response over the nitrate range 1 x 10-5-1 x 10-3 mol dm-3 (Y = 0.996) with the results becoming erratic at higher nitrate concentrations. It should be noted that the measurement of the linear range of the devices in this way is only of value at the developmental stage. The device will only be of use as a sensor when the reagents are screen-printed and a separate device is used for each measurement. The concentration of the different reagents was varied in order to find their optimum values and the results obtained areANALYST, JUNE 1991, VOL. 116 577 shown in Table 5 . Different concentrations of chloride were added in an effort to find the conditions under which varying the chloride concentration would have the least effect.The term E, refers to the point on the voltammogram at which the current starts to increase following the nitrate reduction peak and the value, E, - E,, is a rough guide to the resolution of the nitrate reduction peak from the background reaction. The use of sulphuric acid at a concentration of 0.5 rnol dm-3 gave the best results in terms of peak height and in minimizing the effect of varying the chloride concentration. The highest copper concentration used here, 2.5 x 10-2 rnol dm-3, gave the best results, but background currents were high and the recorder had to be offset by 20-30 PA, presumably because of copper diffusing into the area of the working electrode and being deposited.It would appear that the use of a solution containing 1 x 10-2 rnol dm-3 copper, 0.5 rnol dm-3 sulphuric acid and 1 X 10-4 rnol dm-3 chloride gives the optimum results. As the concentration of chloride in real samples will vary, it would be beneficial to remove all chloride before filling the device. Dionex On-Guard (Dionex, Sunnyvale, CA, USA) Table 4 Voltammetric characteristics of nitrate rcduction at CFDs. Scan speed, 2 mV s--l Nitrate concentration/ 10-5 rnol dm-3 0 1.0 2.5 5.0 7.5 10.0 25.0 50.0 75 .o 100.0 250.0 500.0 750.0 1000.0 E f l - -0.91 -0.90 -0.90 -0.90 -0.90 -0.91 -0.90 -0.91 -0.93 -0.96 -0.98 - 1.01 - 1 .oo i&A 0 1.50 2.25 5.50 8.25 11.75 33.5 55 .O 97.0 140.0 495.0 580.0 875.0 850.0 silver columns were designed for the removal of chloride prior to the injection of samples in ion chromatography. Their use here was tested by passing the sample solution through the column, filling the device and recording the peak voltage at which nitrate reduction occurred.It was found that concentra- tions of chloride of up to 0.5 rnol dm-3 could be removed with the use of these columns. The effects of various ionic species commonly found in hydroponic fluids on the voltammetric response of nitrate reduction were tested. The results obtained are shown in ' O 1 I G I -0.7 -0.8 -0.9 - 1 PotentialN versus internal reference electrode Fig. 7 Linear sweep voltammograms obtained to produce a calibra- tion graph for the determination of nitrate at an in situ generated copper electrode in a CFD. Scan rate, 2 mV s-l; sulphuric acid concentration, 0.5 rnol dm-'; copper(I1) concentration, 1 x lo-' rnol dm-3; chloride concentration, 1 x 10-4 rnol dm-3.Nitrate con- centration: A, 0; B, 0.10 x C, 0.25 x D, 0.50 x E, 0.75 x 10-4; F, 1.0 x 10-4; G. 2.5 x 10-4; and H, 5.0 x 10-1 rnol dm-3 Table 5 Optimization of reagent concentrations [Cu'+]/ [H2SO41/ [CI-]/ 10-3 mol dm-3 mol dm-3 10-4 mol dm-3 5 2 1 5 2 5 5 2 10 5 1 1 5 1 5 5 1 10 5 0.5 1 5 0.5 5 5 0.5 10 5 0.1 1 5 0.1 5 5 0.1 10 1 0.5 1 1 0.5 5 1 0.5 10 10 0.5 1 10 0.5 5 10 0.5 10 25 0.5 1 25 0.5 5 25 0.5 10 EdV -0.915 -0.970 - 1 .020 -0.930 -0.980 - 1 .005 -0.940 -0.985 -0.995 -0.930 -0.970 - 1.020 -0.900 -0.940 -0.965 -0.930 -0.930 - 1.010 i&A E,"IV 13.5 -0.985 10.0 - 1.020 --t - 1.040 15.5 - 1.010 11.5 - 1 340 9.0 - 1.045 16.8 - 1.030 13.0 - 1 .OS5 13.5 - 1 .055 Waves poorly defined and difficult to measure 11.0 -1.020 10.0 - 1.030 4.5 - 1.060 15.3 - 1 ,050 16.8 - 1.010 16.8 - 1.030 18.8 - 1.010 19.3 - 1.010 16.0 -1.080 E , -EdV 0.07 0.05 0.02 0.08 0.06 0.04 0.09 0.07 0.06 0.09 0.06 0.04 0.15 0.07 0.065 0.08f: 0.08f: 0.07$ * E, = potential at which the current begins to increase after the nitrate reduction peak.1- Indistinguishable from background current. f: Deposition current stabilized at 20-30 FA above baseline.578 ANALYST, JUNE 1991, VOL. 116 ~~ Table 6 Effect of ionic species on peak height of nitrate reduction Amount Effect on peak Ion* added/mg 1-1 height (YO ) Ca2+ 100 100.8 500 1 0 . 3 K+ 100 500 Na+ 100 500 Fe3' 10 100 Fez+ 10 100 500 Zn2+ 10 100 Ni2+ 10 100 Mn2+ 10 100 NH4 + 10 100 ~ 0 ~ 3 - 100 500 H2B03- 10 100 Mg*+ 100 98.6 95.9 101.4 101.3 99.1 94.7 104.3 100.8 95.0 104.0 106.5 105.5 101.6 104.8 101.7 102.8 98.2 101.9 95.8 99.3 104.3 101.4 N02- 10 126.7 100 Response erratic M004'- 10 98.9 100 97.7 ~ 1 3 + 10 101.6 100 98.5 EDTA 10 (Ep=-l.OOV) 62.1 (Ep = - 1.04V) 42.5 100 * The cations were added as the sulphate and the anions as the sodium salts.[NO3-] = 5 x 10-4 mol dm-3. Table 6. The peak voltages obtained were -0.900 k 0.03 V unless stated otherwise. Although little change in peak height was observed on addition of molybdate there was a marked change in the shape of the voltammogram (see Fig. 8). It is known that molybdate catalyses the reduction of nitrate18 and heteropoly compounds of similar structure have been shown to catalyse hydrogen evolution,19 which is believed to be the background reaction, and this appears to be what is occurring here.The fact that little increase in peak height occurred as the molybdate concentration was increased appears to be fortuitous. Nitrite appear:; to be reduced at the same potential as nitrate and the addition of 10 ppm of nitrite caused an increase in peak current of >25%. Pletcher and Poorabedil2 studied the reduction of nitrite and nitrate at a copper electrode and found that both species were reduced at approximately the same potential. Although the effect of iron(ii1) on peak height is not marked, the decrease in peak height was approximately 6% for a 100 pprn iron(Ir1) solution; the iron(rn) is reduced before both the copper and nitrate. A solution of 50 pprn of iron(rI1) in a 1 rnol dm-3 solution of sulphuric acid in the CFD gave a peak on scanning at -0.27 V and a shoulder at -0.14 V.Ethylenediaminetetraacetic acid (EDTA), which is nor- mally added to hydroponics in the form of the iron(r1r) salt, has a particularly marked effect on the nitrate reduction peak, as shown in Table 6. However, at the acidity used in the CFD, 0.5 mol dm-3 sulphuric acid, and the low EDTA concentrations used in hydroponics, it is unlikely that the complexing of the copper would have any effect on the use of the device. It also i ' o . 2 v 4 0.10 0.08 a E 3 EQ.06 0 L 3 0.04 0.02 - - -0.7 V - I ! I I / i I I I \ I I I 'L/ I I / -4 / t .0.7 V t -0.7 V I \ I I I ' I \I o t PotentialN versus internal reference electrode Fig.8 Linear sweep voltammograms showing the effect of molyb- denum(v1) on the determination of nitrate at an in situ generated copper electrode in a CFD. Molybdenum concentration: A, 0; B, 10; and C, 100 mg 1-I. Other parameters and concentrations as in Fig. 7 Table 7 Effect of EDTA on nitrate reduction at a copper electrode (all potentials versus SCE). Scan speed, 10 mV s-l; results are the average of three scans [EDTA]/ Peak Peak mg I-' potentialN current/pA 0.00 -0.48 175 4.95 -0.57 186 9.80 -0.57 190 19.2 -0.58 186 Table 8 Effect of EDTA solutions on nitrate reduction at a copper rod electrode (all potentials versus SCE). Results are the average of three scans Peak Peak Electrode treatment potentiaW current/pA Untreated -0.495 183 Placed in 0.5 rnol dm-3 H2S04 for 1 min and then washed with water -0.480 180 Placed in 50 ppm EDTA-0.5 mol dm-3 H2S04 for 1 min and then washed with water -0.547 188 Placed in 50 ppm EDTA (pH 5.2) and then washed with water -0.495 180 seems unlikely that the EDTA interacts with the nitrate in any way.Addition of EDTA to nitrate solutions resulted in a shift in the nitrate peak potential, but had little effect on peak height as is shown in Table 7. That the effect of the EDTA was due to its adsorption on the electrode surface was confirmed by placing a freshly polished copper rod electrode in different solutions, washing with water and then scanning in a nitrate solution. The results, which are shown in Table 8, indicate that it is the fully protonated EDTA that is adsorbed onto the copper surface.Workers who have investigated the effect of additives on the electrodeposition of copper previously,20-22 found that compounds containing amino or carboxyl groups cause an increase in the overpotential which they attributed to adsorption on the copper electrode surface. It was noted that amino groups gave a larger overpotential effect than carboxyl groups and that compounds containing two or more of these groups produced a proportionately larger effect. Gunawar- dena et aZ.23 observed that low concentrations of EDTA retarded the nucleation of silver on vitreous carbon and attributed this to an adsorption phenomenon.ANALYST. JUNE 1991, VOL. 116 579 Attempts to remove the effect of EDTA by the application of high positive potentials were only partially successful.It has been shown that EDTA is oxidized at pre-treated glassy carbon electrodes24 and at doped lead oxide electrodes,25 although the products of the electrode reaction are not known. At a screen-printed carbon electrode no clear oxidation wave was observed but an increase in current was seen at potentials near to the cut-off potential. The application of positive potentials (1.4-1.6 V for 1 min) to the CFD containing a nitrate-EDTA mixture, before determining nitrate, caused a shift in the nitrate peak potential to more positive values, and some increase in peak current. However, the peak current could not be increased beyond 80% of the value obtained when no EDTA was present. The present study is the second in a series aimed at assessing the use of CFDs as disposable amperometric sensors.In the previous study it was shown that phosphate could be deter- mined readily in the device as 12-molybdophosphate directly at a screen-printed carbon electrode, and here we have shown that nitrate can be determined by direct reduction at a copper electrode produced in the device immediately before carrying out the determination. So far the reagents for these determi- nations have not been screen-printed or otherwise deposited within the device; they have been added to the sample solution before this was taken into the device. Work on these and other systems is continuing and it is hoped shortly to provide information on the screen-printing or coating of these various reagents . A. G. F., S. P. S. and T.E. E. thank the Agricultural and Food Research Council for financial support for this project. References Fogg, A. G., Scullion. S. P., Edmonds, T. E., and Birch, B. J., Analyst, 1990, 115, 1277. Fogg. A. G., Scullion. S. P.. and Edmonds. T. E., Analyst, 1989,114,579. Fogg. A. G.. Scullion, S. P., and Edmonds, T. E., Analyst, 1990, 115, 599. Bodini. M. E., and Sawyer, D. T., Anal. Chern., 1977,49,485. 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 Albery, W. J., Haggett, B. G. D., Jones, C. P., Pritchard, M. J., and Svanberg, L. R., J. Electroanal. Chem., 1985, 188,257. Davenport, R. J . , and Johnson, D. C., Anal. Chem., 1973,45, 1979. Davenport, R. J., and Johnson, D. C.. Anal. Chern., 1974,46, 1971. Johnson, D. C., and Shenvood, G. A., Anal. Chim. Acta, 1981, 129, 87. Johnson, D. C., and Sherwood, G. A., Anal. Chim. Acta. 1981, 129, 101. Xing, X., and Scherson. D. A.. J. Electroanal. Chem., 1985. 188, 257. Xing. X., and Scherson, D. A., Anal. Chem., 1987, 59, 962. Pletcher, D., and Poorabedi, Z., Electrochim. Acta, 1979, 24, 1253. Almhofer, N., and Frenzel, F., Fresenius 2. Anal. Chem., 1988, 330, 494. Massey, A. G., in Comprehensive Inorganic Chemistry, ed. Trotman-Dickenson, A. F., Pergamon Press, Oxford, 1973, vol. 3. Hitchman, M. L., and Nyasulu, F. W. M., J . Chem. SOC., Faraday Trans. I, 1986,82, 1223. Hitchman, M. L., Aziz, A., Chingakule, D. D. K., and Nyasulu. F. W. M., Anal. Chim. Acta, 1985, 171, 141. Tam, T. M., and Christiansen, P. J., Plast. Surf. Finish, 1988, 75. 70. Edmonds, T. E., Anal. Chim. Acta, 1980, 116. 323. Keita, B., and Nadjo, L., J. Electroanal. Chem., 1985,191,441. Schneider, H., Sukava, A. J., and Newby. N. J., J. Electro- chem. Soc., 1965, 112, 568. Schneider, H . , Sukava, A. J., McKenney , D. J., and McGregor, A. T., J . Electrochem. Soc., 1965, 112, 570. Sukava. A. J., and Chu, A. K. P., J. Electrochem. Soc., 1969, 116, 1188. Gunawardena, G., Hills, G.. Montenegro, I., and Scharikker, B., J. Electroanal. Chem., 1982. 138,225. Fogg, A. G., Fernandez-Arciniega, M. A., and Alonso, R. M., Analyst, 1985, 110, 1201. Johnson, D. C., Polta, J. A., Polta, T. Z., Neuberger, G. G., Johnson, J.. Tang, A. P. C., Yeo, 1. H., and Baur, J., J. Chem. Soc., Faraday Trans. I, 1986, 82, 1081. Paper 0lO3962 D Received August 31st, I990 Accepted November 13th, 1990