Detection of Reversible Redox Species by Substitutional Stripping

of reversible redox species by a new type of stripping analysis named substitutional stripping voltammetry (SSV). This analysis requires an interdigit...
0 downloads 0 Views 770KB Size
Anal. Chem. 1994,66, 1224-1230

Detection of Reversible Redox Species by Substitutional Stripping Voltammetry Tsutomu Horluchl,' Osamu Niwa, and Hisao Tabei NTT Basic Research Laboratories, Nippon Telegraph and Telephone Corporation, Tokai, Ibaraki 3 19- I I, Japan

We have achieved the highly sensitiveelectrochemicaldetection of reversible redox species by a new type of stripping analysis named substitutional stripping voltammetry (SSV). This analysisrequires an interdigitatedarray (IDA) microelectrode in a sample solution combined with a macroelectrode in a second cell. The second cell contains an electroactive species which deposits reversibly on the electrode. The charge generated by the redox cycling reaction of the sample at the IDA is transmitted to the macroelectrode and accumulated by the deposition of metal ion species on the macroelectrode. A large signal can be obtained immediately by stripping the deposited species. A linear relationship was achieved between the concentration and the stripping peak current when ruthenium hexaammine was used as the sample and silver ions were used as the reversibly deposited species. A very low detection limit of 10 pmol/dm3 was also obtained because the underpotential deposition of silver ions accompanies the ruthenium hexaammine redox cycling. This is because the stripping peak potential is different from that of ordinary deposition which occurs in the initial stage of the potential scan of the macroelectrode. The charge accumulation by SSV is superior to the electrical integration of the redox current of ruthenium hexaammine at the IDA. A highly sensitive and simple method is required in order to detect trace amounts of materials which may affect their biological, environmental, or industrial surroundings. The electrochemical method has been widely used to analyze electroactive species in solution, because it is simple and inexpensive. Voltammetry is commonly used to detect redox speciesand is capable of both qualitative and quantitative analyses.' However, the detection limit of cyclicvoltammetry using a conventional sized electrode is in the micromole per cubic decameter range. Various pulse methods have been p r o p o ~ e dwhich ~ , ~ have improved the detection limit by 1 or 2 orders. However, these detection limits are not low enough for application to biological, industrial, or environmental samples. On the other hand, stripping voltammetry has the lowest detection limit ( 10-9-10-10 mol/dm3) of the commonly used electroanalytical techniques."-lo This excellent sensitivity is ( 1 ) Smyth, W. F. Voltammetric Determination of Molecules of Biological Significance; John Wiley & Sons: Chichester, UK, 1982. (2) Parry, E. P.; Osteryoung, R. A. Anal. Chem. 1965, 37, 1634. (3) Flato, J . B. Anal. Chem. 1972, 44, 75A. (4) Bard, A. J.; Faulkner, L. R. Electrochemical Methods; John Wiley & Sons:

New York, 1980; pp 4 1 3 4 2 0 . (5) Kissinger, P. T.; Heineman, W . R. Laboratory Techniques in Electroanalytical Chemistry; Marcel Dekker, Inc: New York, 1984; pp 499-521. (6) Florence, T. M. Anal. Chim. Acta 1982, 141, 73.

1224

Analytical Chemistry, Vol. 66,No. 8,April 15, 1994

attributable to the preconcentration step. However, stripping analysis requires an analyte which deposits on the electrode. Therefore, only a limited number of analytes such as some heavy metal ions or halides can be measured using the stripping technique. Various efforts have been made to widen the range of detectable species using modified electrodes or ligands to form adsorptive c0mplexes.11-~~ However, when these techniques are applied, suitable ligands or materials must be found with which to modify the electrode and an electrode modification method must be developed for each sample in order to deposit the analyte efficiently on the electrode surface. Recently, we developed a highly sensitive detection method named substitutional stripping voltammetry (SSV)I6J7 to detect trace amounts of reversible redox species which cannot be deposited on the electrode without electrode surface modification or ligand techniques. A closely spaced twin microelectrode was applied in this stripping analysis method. SSV is expected to be applied to the detection of a wide range of reversible redox species such as hydroquinone, catecholamine, and metal complexes, with a remarkably low detection limit in the picomole per cubic decameter range. SSV requires a twin cell system. In a twin cell system (inset in Figure l ) , the two cells are connected by a salt bridge and a twin microelectrode is placed in the left cell compartment, which is filled with a sample solution. A macroelectrode is placed in the right cell compartment, which is filled with another solution containing a speciesthat is reversiblydeposited on the electrode when the potential is changed. The macroelectrode is connected to theone working electrode (collector) of the twin microelectrode by a lead. When electrolysis is carried out by potentiostating the other electrode (generator) of the twin microelectrode, the reverse reaction is induced at the collector and deposition occurs simultaneously on the macroelectrode because of the selfinduced redox cycling effect.18 The amount of deposited speciescan be estimated by conventional stripping voltammetry (7) Anderson, L.: Jagner, D.: Josefson, M. Anal. Chem. 1982, 54, 1371. Anal. Chem. 1983, 55, 416. (8) Sadana, R. S.Anal. Chem. 1983, 55, 304. (9) Hu, A.; Dessy, R. E.; Graneil, A. Anal. Chem. 1983, 55, 320. (10) Van den Berg, C. M. G. Anal. Leu. 1984, 17. 2141. ( I 1 ) Van den Berg, C. M. G.; Jacinto, G. S. Anal. Chim. Acta 1988, 211, 129. ( 1 2 ) Wang, J.; Varughese, K. Anal. Chim. Acta 1987, 199, 185. (13) Wang, J.; Farias,P. A. M.; Mahmoud, J. S.An4l. Chim. Acta 1985,171,215. (14) Van den Berg, C. M . G. Analyst 1989, 114, 1527. (15) Moreira, J . C.: Zhao, R.; Fogg, A. G . Analyst 1990, l l 5 , 1561. (16) Horiuchi, T.; Niwa, 0.;Morita, M.; Tabei, H. Anal. Chem. 1992.64, 3206. (17) Horiuchi, T.;Niwa, 0.;Morita, M.; Tabei, H. Denki Kagaku 1992,60,1130. (18) Horiuchi,T.; Niwa, 0.;Morita, M.;Tabei. H. J . Electrochem. SOC.1991,138, 3549.

0003-2700/94/0366-1224$04.50/0

@ 1994 American Chemical Society

4

0

b

preelectrolysis 1

2

+strip pin;

3

time /min. Figure 1. Preelectrolyslsand stripping current in SSV. Inset is a schematic diagram of the measurement system. The plcoammeter was inserted when the preelectrolysisand stripping current were monitored directly. The sample solution was 1 pmol/dm3 ruthenium hexaammine In the left cell compartment. The silver ion concentration in the right cell was 10 pmoi/dm3.

and is related to both the preelectrolysis time and the concentration of the redox species in the left cell. The concentration of the sample redox species in the left cell can be estimated from the large stripping peak, if preelectrolysis is carried out for a given period. A long preelectrolysis time induces a large amount of deposition because the deposition continues as long as the redox cycling continues in the left cell. As a result, a lowconcentration sample can be detected quantitatively from greatly amplified detection signals. In this paper, we confirm the current enhancement mechanism in SSV and demonstrate quantitative analysis in the picomole per cubic decameter region. Ruthenium hexaammine solution was used as a typical reversibel redox species. Dopamine was also used as a typical example of an analytically important compound.

EXPERIMENTAL SECTION Electrodes. The precise experimental conditions have been described in detail in previous papers.16J7 The twin microelectrode used in this experiment was an interdigitated array (IDA) electrode. The IDA electrode consisted of two series of finger electrodes. The finger widths and gaps were all 2 pm, and each finger was 2 mm long. The IDA consisted of 750 pairs and was fabricated on a thermally oxidized silicon wafer by photolithographic, lift off, and dry etching techniques as described el~ewhere.’~-~l A glassy carbon electrode (GCE) 1 mm in diameter (specially ordered from BAS, Tokyo, Japan) and a silver electrode 1.6 mm in diameter (BAS, West (19) Sanderson, D. G.; Anderson, L. B. Anal. Chem. 1985, 57, 2388. (20) Aoki, K.; Morita, M.; Niwa, 0.; Tabci, H. J . ElectroanaL Chem. 1988,256, 269. (21) Niwa, 0.;Morita, M.; Tabei, H. J . Elecfroonol. Chem. 1989, 267, 291.

Lafayette, IN) were used as the deposition electrodes for detection of ruthenium hexaammine and dopamine, respectively. The reference and auxiliary electrodes were a Ag/ AgCl electrode in 3 mol/dm3 sodium chloride (BAS) and a platinum wire, respectively. Chemicals and Apparatus. Hexaammineruthenium(II1) chloride ( [ R u ( N H J ) ~ ] Cruthenium ~~, hexaammine) (Johnson Matthey/Alfa Products, Ward Hill, MA), dopamine (Tokyo Kasei, Tokyo, Japan), 0.05 mol/dm3 standard buffer solution (pH 4.0; Nakarai Chemicals Ltd., Kyoto, Japan), 0.1 mol/ dm3 phosphate buffer solution (pH 7.0; Nakarai Chemicals Ltd.), potassium nitrate (Wako Chemicals, Osaka, Japan), potassium iodide (Wako Chemicals), and silver nitrate (Kanto Chemicals, Kyoto, Japan) were used as purchased. Electrochemical measurements were performed using a twin potentiostat HECS 990 (Huso, Kanagawa, Japan), a potential sweep unit 175 (Princeton Applied Research, Princeton, NJ), and anX-Yrecorder 3025 (Yokogawa Denki, Tokyo, Japan). A magnetic stirrer (PC-351; Iwaki Glass, Tokyo, Japan) was used to stir the electrolyte during the preelectrolysis period. A picoammeter 486 (Keithley, Cleveland, OH) was inserted between the IDA collector and the G C electrode when the preelectrolysis and stripping current were monitored. Cells. As shown in Figure 1, a twin cell was used in the SSV experiment. Two 20-mL Teflon cells were connected by a salt bridge. The bridge was composed of a glass tube filled with saturated potassium nitrate solution. Each end of the tube was connected to a Vycor disk, GO070 (Princeton Applied Research). For the detection of ruthenium hexaammine, a 0.05 mol/dm3 standard buffer solution containing ruthenium hexaammine was used in the left cell compartment and the Analytical Chemistry, Vol. 66, No. 8, April 15, 1994

1225

glassy carbon electrode was immersed in a 0.1 mol/dm3 potassium nitrate solution containing silver nitrate in the right cell compartment. For the detection of dopamine, 0.1 mol/ dm3phosphate buffer solution containing dopamine was used in the left cell compartment and the silver electrode was immersed in a 0.1 mol/dm3 potassium nitrate solution containing potassium iodide. The Ag/AgCl reference electrode, the platinum wire auxiliary electrode, and the IDA electrode were placed in the left cell compartment. Highly pure argon gas was passed through each electrolyte to remove oxygen before the experiments. The reference and auxiliary electrodes were directly connected to the potentiostat. The terminal of the potentiostat for the working electrode, both generator and collector electrodes of the IDA, and the stripping electrode (glassy carbon or silver) were connected to a switch box which was used to switch between the preelectrolysis and stripping states. In the preelectrolysis state, the IDA generator was connected to the potentiostat and the IDA collector was connected to the stripping electrode. In the stripping state, the stripping electrode was directly connected to the potent iost at. Procedures. Before the SSV experiment, it is necessary to pretreat the stripping electrode to improve the reproducibility and accuracy of the stripping peaks. After the stripping electrode had been polished using an alumina polishing kit (BAS), cyclicvoltammetry was observed in the right cell under the same conditions as those of the SSV stripping stage as described below. The potential sweep was continued until the voltammogram stabilized. As with the conventional stripping method, SSV also requires preelectrolysis and stripping steps. For ruthenium hexaammine detection, the IDA generator was potentiostated at -0.4 V and the IDA collector was connected to the GC electrode during the preelectrolysis step. The electrolyte solution was stirred with a magnetic stirrer during preelectrolysis in the right cell compartment. After a given period, stirring was stopped and the solution was left for 10 s. After the rest period, the switch was immediately changed to the stripping state, and the potential of the GC electrode was swept from -0.4 to +0.5 V at a scan rate of 20 mV/s. For dopamine detection, the preelectrolysis potential was +0.55 V and the silver electrode was swept from +0.25 to -0.4 V at a scan rate of 20 mV/s. Other procedures were the same as those for the ruthenium hexaammine.

RESULTS AND DISCUSSION Mechanism. Figure 1 shows both the preelectrolysis and stripping currents measured with an ammeter which was inserted between the IDA and the stripping electrode in the SSV experiment. The ruthenium hexaammine and silver ion concentrations were 1 and 10 pmol/dm3, respectively. In the preelectrolysis stage, we observed a steady-state current caused by the self-induced redox cycling.*8 The steadystate preelectrolysis current of the 1 pmol/dm3 ruthenium hexaammine agreed well with the theoretical limiting current of a twin potentiostated IDA.*O In our previous paper, we reported that the potential difference between one working IDA and the macroelectrodes caused by inserting a 10-MQ resistor was 0.27 V.I8 This is induced by the concentration 1226

Analytical Chemistry, Vol. 66,No. 8,April 15, 1994

Preelectrolysis time /min. 0

1

2

3

4

no RuHx

5

6

1

C

-60 -80

LL-

3

- 100 - 120 L

lpM RuHx

Flgure 2. Preeiectroiysls current observed with and without a solution of 1 wmol/dm3 ruthenium hexaammlne. The silver Ion concentration was 10 wmolldm3.

gradient of the redox species during preelectrolysis and is sufficiently large for silver ions to be deposited on the stripping electrode. A large stripping peak current flows in the opposite direction in the stripping stage. The total charge flowing in the preelectrolysis period was 12 p C whereas the peak area of the stripping peak was 6.6 p C . The conversion efficiency of 55% is not very high despite the fact that the silver ion concentration was higher than the sample concentration.17 This is because not all of the large amount of silver ions deposited on the GC electrode could be dissolved in a single potential scan. The stripping peak was also observed in the second potential scan. However, this was observed only when the analyte concentration was high. We have developed SSV in order to detect trace amounts of a species in the picomole to nanomole per cubicdecameter range. No peakcan beobserved in thesecond scan in the lower concentration region. Figure 2 shows a comparison of the preelectrolysis current with and without 1 hmol/dm3 ruthenium hexaammine. The silver ion concentration was 10 pmolldm3. When there was no ruthenium hexaammine, no preelectrolysis current was observed. This indicates that the self-induced redox cycling current of the sample species caused the silver ion deposition. There was a delay in attaining a steady-state preelectrolysis current. There are two possible reasons for this delay. One is that the potential of the collector/GC pair was not applied by an external device. The IDA operated in the self-redox cycling mode took a much longer time to reach a steady state than one operated in the dual potentiostated mode.18 The other reason is that the silver ion deposition rate seems to be low and this becomes a rate-determining step at the initial stage of preelectrolysis. The overpotential of the silver deposition may be high at the initial stage of preelectrolysis and decrease during preelectrolysis, which may also delay the response time. It is important in SSV that the preelectrolysis current is in a steady state, even if the sample volume is extremely small (microliter order). Conventional stripping analysis is usually performed with a fairly large volume sample (milliliter order) by stirring the solution. However, it will bedifficult tomeasure very small volume samples (microliter order) by conventional stripping analysis, because the analyte concentration decreases during a fairly long preelectrolysis in accordance with the amount of deposition on the electrode due to the consumption of analyte by the deposition reaction. Therefore, the stripping peak is not proportional to the preelectrolysis time, and it is difficult to detect species quantitatively when the sample solution is very small.

With SSV, the analyte concentration changes very little even if preelectrolysis is carried out over a long period for a small volume sample. This is in contrast to the conventional stripping method. A compact diffusion layer is formed between the two working electrodes of the IDA and maintains a steady state during preelectrolysis.** It is also possible to use SSV even when the change in the concentration of the deposition species is very slight because the concentration or the volume of the deposition species solution can be set at higher or larger values than those of the sample species solution. This indicates that SSV can be applied to detect samples with a very small volume (less than 0.1 mL) or a very low Concentration (pmol/dm3 region), which need a long preconcentration period. The SSV method is more sensitive than the conventional stripping method, because the current induced during the preconcentration stage is already enhanced by the redox cycling at the IDA. As a result, it is easy to apply SSV to quantitative analysis because the height of the stripping peak is proportional to the preelectrolysis time.17 Detection Limit. Glassy carbon and silver ions were used as a stripping electrode and deposition species, respectively, in the ruthenium hexaammine detection. Glassy carbon electrodes have a wide potential window both in the anodic and cathodic potential regions, and silver ions can be reversibly deposited on the glassy carbon. Other electrodes such as a hanging mercury electrode can be used for silver deposition, but the narrow potential window in the anodic region limits the scan range during the stripping stage. Since a sufficient deposition rate is needed to consume the charge generated by the self-induced redox cycling of ruthenium hexaammine, silver and mercury ions have a suitable potential for a deposition species because their deposition potential is sufficient high compared with the ruthenium hexaammine oxidation potential. Silver ions were chosen for the deposition species because mercury ions tend to form complexes with anions in the electrolyte. In a previous paper,17 we reported the detection of 1 P 10-8mol/dm3 ruthenium hexaammine using a 3-mm-diameter glassy carbon electrode immersed in an electrolyte containing 10-6 mol/dm3 silver ions. In order to improve the detection limit, the experimental conditions should be optimized. There are certain parameters which have to be considered such as stripping electrode size, deposition species concentration, and potential scan rate in the stripping stage. When the sample concentration is low, the S/N ratio between the stripping signal and charging current will decrease because the amount of silver ions deposited in a given preelectrolysistime is very small, although the charging current and faradaic current from the impurities do not change. It is difficult to obtain a flat baseline because of the existence of undesired charging currents or currents from impurity reactions at a large stripping electrode. Since the charging current and the faradaic current of impurities can be reduced by decreasing the electrode size, we confirmed that a GCE with a diameter of 1 mm can be used to obtain a sufficient silver deposition rate during preelectrolysis in place of the 3-mm GCE used in our previous paper. (22) Niwa, 0.; Morita, M.; Tabci, H.Sew. Acruarors B 1993, 13-14, 5 5 8 .

Silver ion concentration is also an important factor in the improvement of the SSV detection limit. If the silver ion concentration is equivalent to or lower than that of the sample concentration, the conversion efficiency is much lower than 100% because the silver ion deposition becomes the ratedetermining step and cannot consume all the cliarge induced by the ruthenium hexaammine redox cycling daring preelectrolysis. In our previous paper,17 we studied the conversion efficiency of 1 pmol/dm3 of a ferrocene derivative (aqferrocene) when using 1 and 10 pmol/dm3 iodide ions as the deposition species of a 1.6-mm-diameter silver disk electrode. The conversion efficiency was 100%when 10pmol/dm3 iodide was used. However, only about 3% efficiency was obtained when the iodide concentration was 1 pmol/dm3. This result suggests that the deposition rate of iodide is slower than the aq-ferrocene reaction and that the silver concentration must be higher than that of aq-ferrocene in order to consume the charge generated in the preelectrolysis. A similar result was obtained for the detection of ruthenium hexaammine by SSV. We evaluated the conversion efficiency by holding the silver ion concentration at 1 pmol/dm3 or 100 nmol/dm3 and changing the ruthenium hexaammine concentration from 10 pmol/dm3 to 1pmol/dm3. The collection efficiency was 100% with ruthenium hexaammine concentrations of 1-1 00 nmol/ dm3 and 10-100 pmol/dm3 (also perhaps at 1 nmol/dm3) for respective silver ion concentrations of 1 pmol/dm3 and 100 nmol/dm3. These results clearly indicate that a silver ion solution with a concentration of 100 nmol/dm3 is sufficient for the detection of ruthenium hexaammine at picomole per cubic decameter concentrations. The scan rate of the stripping electrode plays an important role in improving reproducibility and accuracy in SSV. If the scan rate is high, a large stripping peak is observed. However, the baseline inclines and it is difficult to dissolve all the silver on the stripping electrode with a single potential scan. It is not possible to employ an extremely slow scan rate because the amount of undesired silver deposition which occurs in the early potential of the stripping stage increases. The scan rate of 20 mV/s satisfied the above limitations. We started the stripping potential at the same value as the preelectrolysis potential of the generator electrode in the IDA. Higher starting potential is available, but a transitional current caused by switching from the preelectrolysis to the stripping stage may affect the silver stripping peak current. Under our conditions, the transitional current was sufficiently decayed before the electrode potential reached the silver stripping potential. Figure 3a shows the result of SSV applied to 10 pmol/dm3 ruthenium hexaammine with preelectrolysis times ranging from 0 to 20 m. It was confirmed that all the silver ions deposited on the glassy carbon electrode dissolved in the first potential scan because there was no stripping peak in the second scan. The voltammograms shown in Figure 3b were obtained in the same way as those shown in (a) without ruthenium hexaammine. In Figure 3a, the silver stripping peaks observed at 0.3 V increased when the preelectrolysis time was increased. In contrast, there were no peaks at the same position in Figure 3b. These results indicate that the electrolysis of 10 pmol/ dm3 ruthenium hexaammine resulted in silver ion deposition. Analytical Chemistry, Vol. 66, No. 8, Aprii 15, 1994

1227

t

10 min.

/I0 $ V

InA

20 min 10 min

5 min. 0 min.

lOmin

0 05V 1 Omln

5 min

0 min

OV 5min

-0 05V 0 5 min

4 1 V 0 5”

-0 4V 0 5m1n

1

1

1

/ / I l l

1

1

1

1

1

1

1 Orin

I

The peak height after 20 min of preelectrolysis was about 0.5 nA. The theoretical limiting current of 10 pmol/dm3 ruthenium hexaammine in the twin potentiostated mode was 0.8 pA using the steady-state equation of the IDA.*O Therefore, the signal detected with SSV was about 630 times larger than that detected with cyclic voltammetry in the twin potentiostated mode. The area of this peak was calculated at about 1.1 nC by converting the horizontal axis to unit of time. On the other hand, the totalchargeduringpreelectrolysis was 0.96 nC, which was calculated by assuming that thesteadystate current continued to flow for 20 min. From these results, the Coulombic efficiency of the substitution from ruthenium hexaammine oxidation to silver deposition was calculated to be 115%. This high Coulombic efficiency of the substitution was due to the concentration of silver ions being much higher than that of the ruthenium hexaammine. The 15% excess error may be caused by the effect of dissolved oxygen for the following reasons. Before preelectrolysis, the oxygen was removed completely by bubbling argon gas through the sample solution. However, we could not deoxygenate the sample solution with argon during preelectrolysis because the ruthenium hexaammine redox cycling is very sensitive to convection. Therefore, a very small amount of oxygen may dissolve during 20-min-long preelectrolysis. This is because the efficiency is 100% for the charge obtained with the peak of the shorter (10 min) preelectrolysis. The efficiency was also 100% when calculated using the result for a sample solution of more than 100 pmol/dm3. Calibration curves of SSV in the picomole per liter region (10-1 00 pmol/dm3) of ruthenium hexaammine concentration were obtained. The relationship between the stripping peak height (Zp in nA) and the sample concentration (c in pmol/ dm3)could be calculated from the limiting current of the IDA and the preelectrolysis time ( t in min). I p = atc Here a was calculated a t 0.00225 from the limiting current of 1 pmol/dm3 ruthenium hexaammine a t the IDA and the conversion factor from the stripping peak area to the peak height when the scan rate was 20 mV/s. Preelectrolysis 1228

Analytical Chemistry, Vol. 66, No. 8, April 15, 1994

1

I

8

I

l

l

1

I

Figure 4. Conventional stripping voltammograms of silver using a glassy carbon electrode and different preelectrolysispotentials. Silver ion concentration was 100 nmol/dm3.

experiments of 10- and 2-min duration were performed to obtain the calibration curves of ruthenium hexaammine by SSV. The 10-min experiment showed a linear relationship between the sample concentration and the peak height. The proportional factor at was 0.0207 obtained using the least squares method. The square of the multiple correlation coefficient, which is given by the ratio of the model sum of the squares to the total sum of the squares, was 0.994. The proprotional factor obtained with the least squares method agreed well with the calculated value of 0.0225, when t was 10. On the other hand, the proportional factor for the 2-min experiment was 0.001 63 and the square of the multiple correlation coefficient was 0.977. The proportional factor for the 2-min experiment was below the calculated value. This is because the initial stage of the conversion efficiency in the preelectrolysis was not high, as shown in Figure 2. Silver Ion Deposition Mechanism. As shown in Figure 3, there were other peaks a t 0.15 V in both (a) and (b). These peaks did not depend on the preelectrolysis time, and all had the same height. The reason for the appearanceof these timeindependent peaks in SSV is discussed below. Figure 4 shows the result of conventional anodic stripping voltammograms (ASV) of the 100 wmol/dm3 silver ions for different preelectrolysis times and potentials using the same glassy carbon electrode used in SSV. The preelectrolysis time and potential are shown on the right of each voltammogram. The bottom voltammogram is the GC baseline. When the preelectrolysis potential was sufficiently lower than the silver deposition potential (-0.4 or -0.1 V), large peaks with same height were observed for the same preelectrolysis time of 0.5 min. When the preelectrolysis potential was -0.05 V, the peak height decreased because this value is not sufficiently low. When the preelectrolysis potential was increased to 0 V for a preelectrolysis time of 0.5 min, the

voltammogram was almost same as the baseline. A preelectrolysis time of 5 min was needed to obtain a stripping peak with a similar height. In this voltammogram, two stripping peaks were observed as with the SSV result. When the preelectrolysis potential was further increased to 0.05 V, it took a long preelectrolysis time of 10 min before the second peak was observed. In this voltammogram, two clearly separated peaks were observed at 0.15 and 0.3 V, respectively. The peakat 0.3 V wasnot observed at a preelectrolysispotential of 0.1 V even for a preelectrolysis time of 10 min. These results indicate that the stripping peak at 0.15 V is the bulkstripping peakcaused by thedeposition at a sufficiently low potential and that at 0.3 V is the monolayer stripping peakcaused by underpotential deposition (UPD). It is known that metal ions are deposited on an electrode of a different metal at more positive potential and this under potential shift AUbetween bulk and monolayer stripping peaks is proportional to A@,which is the difference between the work functions of substrate and absorbate materials with a proportional factor of 0.5.23 Results for the work function of carbon vary greatly; however, a value of 4.6 eV is recommended.24 The AU is calculated as 0.15 V by use of the silver work function of 4.3 eV,24which agreed well with our experimental result. The voltammogram of SSV applied to 10 pmol/dm3 ruthenium hexaammine had two peaks as with theconventional stripping method. These peaks were also caused by normal deposition and UPD. The bulk stripping peaks from silver deposition at 0.15 V, which is observed in conventional stripping voltammetry and in SSV both with and without ruthenium hexaammine, occurred early during potential sweeping in the stripping stage when the stripping electrode potential was sufficiently low to deposit silver. On the other hand, the monolayer stripping peaks of silver UPD at 0.3 V, which were observed in SSV only with ruthenium hexaammine, were caused by underpotential deposition during preelectrolysis. The peak area of 1.1 nC, observed at 0.3 V by using SSV applied to 10 pmol/dm3 ruthenium hexaammine with 20-min preelectrolysis, corresponded to silver atoms covering 0.0006 of the monolayer, which was roughly estimated from the number of silver atoms and their radius. As discussed above, the monolayer stripping peak caused by the self-induced redox cycling current of the low concentration sample is separated from the bulk stripping peak. This indicates that the SSV stripping peak does not interfere with the bulk stripping peak. This peak separation effect is useful for detecting a picomolar sample where its peak height is very small. Comparison with Digital Integration. SSV operates by substituting another electrolyte current for the sample detection current and current integration in the form of deposition. The high sensitivity of SSV is based on detection current enhancement by this integration method. However, a digital integration method which uses an ammeter may be used to enhance the detection current. Figure 5 compares SSV with or without the insertion of an picoammeter. The sample solution contains no ruthenium

(d) 5min. (c)Smin. (ammeter inserted)

(b) 2min. (ammeterinserted)

(a) 5 min.

u 0

0.5 Potential /V

Flgure 5. SSV observed wlth and without picoammeter insertion.

(23) Kolb, D. M.; Przasnyski, M.; Gerischer, H. J . Elecrroonul. Chem. 1974, 54,

hexaammine, and the silver ion concentration was 10 pmol/ dm3. Voltammograms b and c were obtained by inserting the picoammeter between one working IDA and a glassy carbon electrodewith preelectrolysis times of 2 and 5 min, respectively. Voltammograms a and d were obtained without the picoammeter and with 5-min preelectrolysis. The voltammograms were measured in alphabetical order. When the picoammeter was inserted, stripping peaks were observed even though there was no ruthenium hexaammine, as shown in (b) and (c). The height of these peaks depended on the preelectrolysis time. From these peak areas, it was estimated that a current of approximately 80 pA flowed during the preelectrolysis period. As shown in Figure 2, the preelectrolysis current without ruthenium hexaammine contained a 10-nA noise band. It was necessary to increase the picoammeter current range a 200 nA full scale in order to measure this noise current. In general, exact current measurement is difficult when the dynamic range is wide because the accuracy becomes poor. The 80-nA current estimated above was within the accuracy of the 200-nA range. This error current becomes an off-set current which flows during the preelectrolysis period and which results in the undesired stripping peak. However, for electrochemical integration in SSV there is no current source except for the IDA working electrode, which can operate as an ideal integrator. The SSV method has the considerable advantage of being able to analyze a lowconcentration sample whose detection current is very low. Application of SSV to Dopamine Analysis. Various electrochemical methods have been developed to detect catecholamine such as voltammetry and the LCEC method. We achieved a low detection limit of 10 nmol/dm3 when we applied IDA microelectrodes to the generation collection voltammetry of dopamine.25 We applied cathodic SSV to detect lower concentrations of dopamine using an IDA/silver electrode combination. Iodide was chosen as the deposition species. The experimental conditions described for the detection of ruthenium hexaammine were also optimized for dopamine detection. A cathodic stripping peak of iodide was obtained which increased in proportion to the amount of dopamine preelectrolysis.

25. (24) Morcos, I. J . Electroonal. Chem. 1975, 66, 250.

( 2 5 ) Niwa, 0.; Morita, M.; Tabei, H.EIectrounolysis 1991, 3,

163.

Analytical Chemistw, Vol. 00, No. 8, April 15, 1994

1229

Dopamine could be detected at nanomolar levels by cathodic SSV. The dopaminedetection limit is better than that obtained with generation collection voltammetry. However, it is much higher than that of ruthenium hexaammine. One reason is that the detection limit of iodide cathodic stripping is 1 or 2 orders poorer than that of anodic stripping. Another possible reason for this higher detection limit is that the dopamine was quasi-reversible on the gold electrode. We are now developing a method for the fahrication of a carbon-based IDA microelectrode and studying the electrochemical properties of dopamine on the electrode.26 Since dopamine shows good electrochemical reversibility a t a carbon IDA electrode, the dopamine detection limit will be lowered by using a carbon IDA electrode.

CONCLUSION SSV was applied to a sample solution of ruthenium hexaammine as a typical reversible redox species using silver ions as a reversibly deposited species. The current enhancement mechanism of SSV caused by analyte substitution and current integration was confirmed directly by an ammeter insertion experiment. A linear relationship was obtained ( 2 6 ) N i w a . 0 : Tabei, H. Anal. Chem. 1994, 66, 2 8 5 .

1230

Analytical Chemistry, Vol. 66, No. 8, April 15, 1994

between the ruthenium hexaammine concentration and the stripping peak current. A very low detection limit of 10 pmol/ dm3 was also obtained because the underpotential deposition of silver ions accompanies ruthenium hexaammine redox cycling. This is because the stripping peak potential is different from that of ordinary deposition which occurs a t the initial stage of the potential scan of a macroelectrode. The charge accumulation by SSV is superior to the electrical integration of the redox current of ruthenium hexaammine a t the IDA. SSV was demonstrated as a simple and quantitative analytical technique for low-concentration samples. SSV can also be applied for the detection of dopamine using the cathodic stripping system.

ACKNOWLEDGMENT We thank M . Morita, NTT Basic Research Laboratories, for helpful discussions. Received for review October 1, 1993. 1994.@ e

Accepted January 25,

A b s t r a c t published in Advance A C S A b s f r o c t s . M a r c h 1 , 1994.