Comparison of the Gold Reduction and Stripping Processes at Platinum, Rhodium, Iridium, Gold and Glassy Carbon Micro- and Macrodisk Electrodes Alan M. Bonda, Steven Kratsisa, Shelly Mitchellb and Jan Mocakc a Department of Chemistry, Monash University, Clayton, Victoria 3168, Australia b Department of Chemistry, Latrobe University, Bundoora, Victoria 3083, Australia c Department of Analytical Chemistry, Slovak Technical University, SK-81237 Bratislava, Slovak Republic The gold AuIII + 3e2 ? Au0 reduction and Au0 ? AuIII + 3e2 oxidation stripping processes in dilute aqua regia electrolyte (0.1 m HCl + 0.32 m HNO3) were examined at platinum, rhodium, iridium, gold and glassy carbon disk electrodes. After ascertaining that the preferred material was platinum, the effect of electrode size was evaluated by using nine different platinum disk electrodes having diameters ranging from 2 to 2000 mm.The optimum analytical response was obtained with a 50 mm diameter platinum disk electrode.With this electrode diameter, a sharp symmetrical gold stripping peak was obtained and the deposition process occurred predominantly under conditions of radial diffusion so that stirring of the solution was not required. In contrast, larger sized platinum electrodes produced a broader, asymmetric stripping response for the gold oxidation peak, whereas electrodes of smaller diameter provided poorer signal-to-noise ratios. The limit of detection and limit of quantification were calculated to be 4.4 3 1027 m (86 ppb) and 13.1 3 1027 m (258 ppb), respectively, at the 50 mm diameter platinum disk electrode under conditions of linear sweep stripping voltammetry at a scan rate of 200 mV s21 and a 140 s deposition time.The optimum electrode gave a very well defined gold oxidation signal with negligible background current when applied to the determination of gold in a gold ore sample. Keywords: Gold determination; voltammetry; optimum electrode size and electrode material There are numerous permutations and combinations of electrode materials and size available for use in analytical applications of voltammetry.Although the nature of the electrode material is often carefully considered in the development of a voltammetric method of analysis, there are relatively few studies in which the performance has been studied as a function of electrode size. However, the importance of electrode size is now emerging more frequently as a significant variable with the commercial availability of microdisk electrodes having diameters of around 10 mm.In contrast, common commercially available macrodisk electrodes typically have diameters in the millimetre range. Numerous analytical methods are available for the determination of the precious metals and many of them have recently been reviewed by Qu.1 A wide range of carbon, metal and chemically modified electrodes have been proposed for the voltammetric determination of gold.1,2 However, the specific dependence of the analytical response on the nature and size of the electrode material has yet to be determined.In this paper, the reduction of gold(iii): AuIII + 3e2 ? Au0 (1) and the stripping of gold: Au0 ? AuIII + 3e2 (2) in dilute aqua regia electrolyte (0.1 m HCl + 0.32 m HNO3) has been studied at platinum, rhodium, iridium, gold and glassy carbon disk electrodes to determine the optimum electrode surface in the analytical sense.Having ascertained that platinum is the preferred electrode material, the response at a range of electrode diameters varying from 2 to 2000 mm was examined to determine the optimum electrode size. The limit of detection (LOD) and limit of quantification (LOQ) were calculated for a typical set of conditions and the optimum electrode was applied to the determination of gold in a gold ore sample. Experimental Chemicals The dilute aqua regia electrolyte (0.1 m HCl + 0.32 m HNO3) was prepared by dilution of analytical-reagent grade nitric and hydrochloric acids.Dilution of the acids was effected with high-purity water (17 MWcm; Nanopure, Barnstead, IA, USA). Tests for interferences from iron and copper were undertaken using analytical-reagent grade FeSO4·7H2O and CuSO4·5H2O, respectively. Gold standard solutions were prepared by dilution of a 1000 ppm Gold Spectrosol solution (BDH, Poole, Dorset, UK) with the dilute aqua regia electrolyte.All glassware was cleaned in a detergent solution, followed by boiling nitric acid and then rinsing with water prior to drying at 200 °C. Preparation of the Gold Ore Sample The method of sample preparation used prior to the voltammetric analysis of gold ore samples, donated by Unichema International, was as follows: 1. The ore sample was dried at 105 °C and crushed to less than 100 mm particle size. 2. A 25 g amount of the sample was accurately weighed into a 250 ml beaker and 50 ml of 5 m HCl were added.The resultant mixture was left to stand at room temperature for 30 min to destroy sulfides. 3. The solution was heated on a hot-plate at approximately 120 °C for 2 h, with the beaker being covered with a watchglass. 4. After the solution had cooled, 10 ml of concentrated HCl and 10 ml of concentrated HNO3 were added. The solution was then heated for 4 h, allowing the liquid volume to decrease but not to go to dryness. 5. The solution was allowed to cool and 25 ml of 8 m HCl were added.After stirring, the solution was filtered into a separating funnel with the residues being rinsed with 2 m HCl. 6. The filtrate was extracted three times with diethyl ether (3 3 20 ml). Analyst, October 1997, Vol. 122 (1147–1152) 11477. The ether extract was washed with 0.1% HCl to remove iron. 8. The ether extracts were evaporated over 10% HCl using a heater. When the ether had volatilised, the gold remained in the acid solution.The sample was then added to dilute aqua regia electrolyte (3 : 1 ratio) and was ready for voltammetric analysis. The use of diethyl ether solvent extraction to remove interferences has been reported,3,4 as have other forms of solvent extraction.5–9 Equipment All voltammetric measurements were undertaken at 20 °C with a BAS 100A Electrochemical Analyzer (Bioanalytical Systems, West Lafayette, IN, USA) equipped with a BAS PA-1 preamplifier and Faraday cage.Where necessary, oxygen was removed by degassing solutions with nitrogen. The standard three-electrode arrangement was used with a platinum wire counter electrode, an Ag/AgCl (3 m KCl) reference electrode and the following working electrodes: 2000, 500, 250, 100, 70, 50, 25, 10 and 2 mm diameter platinum disk electrodes, 500 mm diameter glassy carbon, rhodium and iridium disk electrodes and a 2000 mm diameter gold disk electrode. The working electrodes were made via methods related to those described in the literature10 using metal wires from Goodfellow Metals (Cambridge, UK).Prior to each voltammetric experiment, the working electrode surface was polished on a LECO polishing pad with 0.6 mm alumina slurry, then rinsed with distilled water and carefully dried. Method for the Evaluation of Detection Limits According to the IUPAC definition, the limit of detection is the lowest concentration of an analyte that an analytical process can reliably detect.Originating from this definition, the signal counterpart of the limit of detection (LOD) is located 3sb (three times the standard deviation) or 6sb above the gross blank signal or, equivalently, 3sb or 6sb above zero when using the net signal. The use of 6sb instead of 3sb is substantiated by the tendency to diminish type II error11 causing an inadequate judgment of a true signal as the blank, although this is not important in our further discussion of the problem. If the calibration plot is a perfect straight line passing through the origin (net signals are assumed), then the limit of detection is simply determined by projecting the intercept of 3sb with the calibration plot on to the concentration axis.By analogy, the limit of quantification (LOQ) refers to the lowest concentration, or another quantity, which can be quantitatively measured with reasonable reliability by a given procedure. Its signal counterpart was defined as 10sb above the gross blank signal, or 10sb above zero when the net signals are used.In ref. 11, a 9sb (nine times the population standard deviation) approach is advocated instead. Also in this case, the use of a calibration plot is necessary for the calculation of the LOQ. When using the described standard method of obtaining the LOD, three main factors can considerably influence the reported result. The first occurs commonly because the number of measurements made on the blank and/or the sample itself is not sufficiently large to fulfil the conditions required to ensure that the normal distribution is valid for the measured signal.This problem may be overcome by replacing the value 3sb by the corresponding product t(n, a)sb where t(n,a) is the critical value of the t-distribution, n is the number of degrees of freedom and a is the confidence level.11 The second problem is related to inconsistency of the position of the zero net signal and the intercept of the calibration plot, denoted q0.The accuracy of the evaluated LOD depends on how positive or negative q0 is. When q0 is negative, the projection through the calibration plot leads to an over-estimation of the LOD, whereas a positive value leads to an under-estimation or even in extreme cases to a negative LOD value. If the intercept q0 is ignored, the LOD signal is projected through a line parallel to the net signal calibration plot, having the same slope q1 but a zero intercept (which is equivalent to calculation of the LOD as 3sb/q1) via this procedure, the negative LOD values are prevented and the over- and under-estimation proceed again but in the opposite way.The third problem is associated with the uncertainty of the position of the calibration line itself. A line of best fit could be shifted by chance or even by a statistical manipulation in such a way that a desired (usually lower) LOD is obtained owing to the improper values of regression parameters.Flaws in the ways in which the IUPAC definition of the LOD is commonly interpreted can be minimised by using a method known as the upper limit approach (ULA),11 which also is used in this paper. Using the ULA, the upper one-sided confidence limits of individual signal observations are calculated, which depend on the errors in regression and also on the type of the calibration model. If a one-parameter straight-line calibration model is statistically correct (the straight line passing through the origin), then the upper confidence limit for any chosen concentration c0 is distant from the regression line by the value of t(n,a)s, where n is now n 2 1 and s is defined by the equation s = (1 + c0 2/Sci 2)1/2 sy (3) with sy denoting the residual standard deviation.For the calculation of the limit of detection by the ULA, it is necessary to consider the case where c0 = 0 in eqn. (3). Under these conditions, the product in eqn. (3) refers to the upper limit signal at zero concentration.Further, when c0 = 0 and thus s = sy, the limit of detection can be simply calculated as LOD = t(n,a) sy/q1 (4) where n = n 2 1 and q1 is the slope of the calibration plot. Differences in the standard approach and the ULA method are illustrated in Fig. 1. Results and Discussion Selection of the Optimum Electrode Material The initial investigations concerned the selection of the optimum working electrode material for the voltammetric determination of gold(iii) in 0.1 m HCl + 0.32 m HNO3 electrolyte, which for convenience is refered to as dilute aqua regia.For this study, 2000 mm diameter platinum and 500 mm diameter platinum, glassy carbon, rhodium and iridium electrodes and a 2000 mm diameter gold electrode were chosen. Platinum macrodisk electrodes Cyclic voltammograms at 2000 and 500 mm diameter platinum macrodisk electrodes in dilute aqua regia showed a well defined response with the AuIII + 3e2?Au0 reduction and (Au0 ? AuIII + 3e2) oxidation peak potentials of 0.550 and 0.900 V versus Ag/AgCl, respectively, giving a peak-to-peak separation of 0.350 V with the 500 mm diameter electrode under the conditions in Fig. 2(a). An important feature with the use of this electrode material is the sharp gold oxidation peak which is required for the sensitive stripping method. Two closely spaced stripping peaks have been observed in other work,12 but under the conditions of the present study only a single stripping peak was found.At the platinum macrodisk electrodes, both the reduction and oxidation peaks when using cyclic voltammetry varied linearly with gold concentration over the range 2 3 1148 Analyst, October 1997, Vol. 1221025–2 3 1023 m and the limit of detection was found to be 2 3 1026 m and the limit of quantification 2 3 1025 m for the reduction process at a scan rate of 375 mV s21. Further, the reduction peak current was found to be dependent on the square root of the scan rate over the scan rate range 20–1000 mV s21 with a slope of 0.51 being obtained from a log(scan rate) versus log (peak height) plot.These data are consistent with a (linear) diffusion-controlled reduction process. The stripping process gave a slope of 0.76 when plotting log(scan rate) versus log (peak height) when the switching potential was 400 mV versus Ag/AgCl. Iridium electrode The first scan at the iridium working electrode exhibited similar characteristics to that found at the platinum electrode, with a sharp gold oxidation peak, reduction and oxidation peaks of 0.530 and 0.870 V versus Ag/AgCl, respectively, and a peak-topeak separation of 0.340 V under the conditions in Fig. 2(b). However, on repetitive cycling, the nature of the response changes as dissolution of the electrode occurs. With repetitive cycling, the gold reduction peak increases in current and shifts to more negative potentials (0.05 V shift by the fourth scan).Conversely, the gold stripping peak decreases in height with each cycle, but does not shift in potential [Fig. 2(b)]. The changes in the voltammetry with repetitive cycling are attributed to dissolution of the iridium metal electrode which is accompanied by the development of an IrIII–IrIV redox couple. 13,14. The dissolution is detected voltammetrically by the development of an oxidation current at 0.700 V versus Ag/ AgCl. Rhodium electrode Potential cycling of a gold(iii) solution using a rhodium electrode in dilute aqua regia also was accompanied by rhodium metal electrode dissolution [Fig. 2(c)]. Problems associated with dissolution of a particular electrode material are detected via an enhanced background current when monitoring the voltammetry of dilute gold solutions. For example, stripping voltammograms obtained from 3 31026 m gold(iii) solutions at platinum, rhodium and iridium electrodes with 2 3 1026 m standard additions using a 140 s deposition time and a scan rate of 200 mV s21 are compared in Fig. 3. The background with the platinum electrode where no dissolution occurs is very small in comparison with that observed with the rhodium and iridium electrodes. Glassy carbon electrode Under the conditions in Fig. 2(d), the cyclic voltammogram of gold(iii) with a glassy carbon working electrode showed reduction and oxidation peak potentials of 0.400 and 0.950 V versus Ag/AgCl, respectively, and a large peak-to-peak separation of 0.550 V for the gold(iii) reduction and gold metal reoxidation.However, apart from the considerable irreversibility of the process, the response is well defined [Fig. 2(d)]. At other forms of carbon electrode15 two gold stripping peaks have been reported, although only a single response was observed under the conditions of this study. Gold electrode A gold electrode can be used for the determination of gold via use of the gold reduction process. As shown in Fig. 2(e), the Fig. 1 Illustration of differences in determining the limit of detection by (a) the upper limit approach for the case when the calibration model is represented by a straight line passing through the origin (ULA1) and (b) the standard approach (SA) based on the IUPAC definition, calculating LOD1 by means of the calibration line and LOD2 by using an auxiliary line parallel to the calibration line (the same slope but a zero intercept). Fig. 2 Cyclic voltammograms of a 3.5 3 1024 m gold(iii) solution obtained in dilute aqua regia electrolyte at a scan rate 20 mV s21 using (a) platinum (500 mm), (b) iridium (500 mm), (c) rhodium (500 mm), (d) glassy carbon (500 mm) and (e) gold (2000 mm) working electrodes.Analyst, October 1997, Vol. 122 1149peak potential for reduction of gold when using a gold electrode was 0.550 V versus Ag/AgCl, which is comparable to values obtained at platinum and iridium electrodes. Obviously, stripping voltammetry of gold at a gold electrode is not possible.Further Considerations Concerning the Choice of Working Electrode Material The above data suggest that either platinum and glassy carbon may be the preferred electrode materials for the determination of gold. However, at low gold(iii) concentrations, problems with the background current arise when using the glassy carbon electrode. Various oxygen-containing functional groups have been identified by various workers which can be present on the carbon surface (e.g., phenol, carbonyl and quinone groups).16,17 Scanning to positive potentials oxidises these functional groups, forming a multilayer oxide which increases the background current.An additional problem with the glassy carbon electrode arises from the fact that the reduction peak potential for gold(iii) is almost 200 mV more negative than found at a platinum electrode. Therefore, when using a stripping technique for the gold determination, a considerably more negative deposition potential has to be used, which can cause problems with impurities such as iron, copper and lead, which are reduced at these more negative potentials.Hence, these elements, which are often present at relatively high levels in gold ore samples, may cause interference in the determination of gold at a glassy carbon electrode, but not at platinum. The problems with the presence of iron and the use of a glassy carbon electrode can be seen by comparing voltammograms obtained at platinum and glassy carbon electrodes for a 10 mg l21 gold(iii) solution spiked with 15 mg l21 [Fig. 4(a) and (b)] and 150 mg l21 iron(iii) [Fig. 4(c) and (d)]. The platinum electrode also is favourable for minimising interferences from copper because a deposition potential of 250 mV versus Ag/ AgCl can be used, which is more positive than the potential for the reduction of copper. Fig. 4(e) shows a stripping voltammogram of a 10 mg l21 gold(iii) solution spiked with 15 mg l21 copper(ii).Further details on intereferences likely to be encountered in the voltammetric determination of gold are available.18–20 Glassy carbon electrodes have been widely used for the determination of gold;5,6 in contrast and perhaps suprisingly, according to our findings, platinum indicating electrodes have been only rarely used.12,18 Determination of the Optimum Platinum Electrode Size for the Determination of Gold Evaluation of the optimum size platinum electrode was undertaken by comparing voltammograms obtained with a 3.5 3 1024 m gold(iii) solution using nine different platinum disk electrode sizes ranging from 2000 to 2 mm in diameter at a scan rate of 20 mVs21 (Fig. 5). At large electrode sizes (diameters of 2000, 500 and 250 mm) with this scan rate the reduction process exhibits a peak-shaped response whereas at smaller diameters (100–2 mm) a sigmoidal-shaped response is observed as the diffusion of gold(iii) to the electrode surface changes from linear at the larger diameter macrodisk electrodes to radial at the smaller diameter microdisk electrodes.21 The gold oxidation stripping peak also changes in shape from relatively broad and slightly asymmetric at larger diameters to sharp and symmetrical with a well defined baseline for diameters @50 mm.Further, with the smaller electrode sizes, stirring of the solution is not required during the deposition step22 so that microelectrodes may offer advantages over conventionally sized electrodes when using stripping methods.The change in voltammetry as a function of electrode diameter at a constant scan rate can be represented graphically by plotting Fig. 3 Stripping voltammograms obtained in dilute aqua regia electrolyte at 500 mm diameter (a) platinum, (b) rhodium and (c) iridium electrodes using a 140 s deposition time and a scan rate of 200 mV s21 for 3 3 1026 m gold(iii) solutions with 2 3 1026 m gold(iii) standard additions.Fig. 4 Stripping voltammograms obtained in dilute aqua regia electrolyte using a 140 s deposition and a scan rate of 200 mV s21 for a 10 mg l21 gold(iii) solution after addition of 10 mg l21 iron (iii) at (a) platinum (500 mm) and (b) glassy carbon (500 mm) electrodes and after addition of 150 mg l21 iron(iii) containing the same (c) platinum and (d) glassy carbon electrodes. (e) Stripping voltammogram obtained at a platinum (500 mm) electrode for 10 mg l21 gold(iii) solution containing 15 mg l21 copper(ii). 1150 Analyst, October 1997, Vol. 122the ratio of stripping peak current to the reduction peak or limiting current versus the logarithm of the area of the electrode. From this form of data analysis it can be observed that initially there is an increase in the ratio of oxidation to reduction currents as the size of the electrode is decreased until a diameter of about 50 mm is reached (Fig. 6). At very small sizes, the reproducibility for the gold process decreased, presumably owing to variable electrode areas resulting from greater difficulty in reproducibly polishing the smaller sized electrodes.Therefore, a 50 mm diameter electrode was selected as optimum on the basis of possessing a sharp, symmetrical gold oxidation peak, efficient electrolysis (oxidation/reduction value) and a favourable signalto- noise ratio. In practice, the 50 mm diameter electrode also represents close to the largest size electrode available for achieving near steady-state behaviour in practical analytical voltammetry.Hence this size of electrode is also about the largest for which stirring of the solution is not required to enhance the gold deposition process. Determination of the Limits of Detection and Quantification Using a 50 mm Diameter Platinum Microdisk Electrode The limits of detection and quantification under a typical set of conditions were determined using the 50 mm diameter platinum disk electrode and linear-sweep stripping voltammetry.Using a 140 s deposition time and a scan rate of 200 mV s21, a small but reproducible signal could be obtained with a 4 3 1027 m gold(iii) solution. A calibration plot of peak current versus concentration was constructed using 4 3 1027 m as the lowest analyte concentration and 4 3 1027 m standard additions (Fig. 7). Using the upper limit approach, the LOD was determined to be 4.4 3 1027 m (86 ppb) and the LOQ 13.1 3 1027 m (258 ppb).Obviously, the LOD and LOQ are strongly influenced by scan rate, deposition time and deposition potential. However, the conditions used above certainly provide adequate sensitivity for the determination of gold in gold ore samples, which was the problem of practical interest in this study. Determination of Gold in a Gold Ore Sample The 50 mm platinum microdisk electrode was used to determine gold in ore samples provided by Unichema International (Port Fig. 5 Cyclic voltammograms obtained in dilute aqua regia electrolyte at a scan rate of 20 mV s21 for 3.5 31024 m gold(iii) solutions using platinum electrodes with sizes varying from (a) 2000 to 70 mm and (b) 50 to 2 mm in diameter.Fig. 6 Variation in the ratio of anodic to cathodic peak or limiting currents versus log (electrode area) for the gold oxidation and gold(iii) reduction processes. Experimental conditions as in Fig. 6. Fig. 7 Stripping voltammogram obtained in dilute aqua regia electrolyte for a 140 s deposition time and a scan rate of 200 mV s21 at a 50 mm platinum disk electrode for a 4 3 1027 m gold(iii) solution with 4 x 1027 m standard additions of gold(iii).Analyst, October 1997, Vol. 122 1151Melbourne, Victoria, Australia). The stated concentration in the sample considered in Fig. 8 was 32 mg l21. A well defined stripping response was observed with a 140 s deposition time and a scan rate of 200 mV s21 . Standard additions of 13 mg l21 gold(iii) were made to the solution (Fig. 8) and a plot of peak current versus concentration of added gold(iii) was constructed. The gold concentration was calculated from the intercept and found to be 28 mg l21, which compares favourably with the stated value of 32 mg l21. Equally, well defined gold stripping voltammograms were obtained for other gold ore samples provided by Unichema International. Differential-pulse Stripping Voltammetry Differential-pulse stripping voltammetry is widely used in trace analysis.However, for the determination of gold at a platinum electrode, stripping studies at the 1026 m concentration level produced broader and more complex signals of lower reproducibility than obtained under linear-sweep conditions. The lack of improvement associated with the use of the differential-pulse method is probably associated with the extreme irreversibility of the gold redox chemistry. Consequently, the linear-sweep method is preferred for the determination of gold at platinum disk electrodes.Conclusions The preferred electrode material for the voltammetric determination of gold in 0.1 m HCl + 0.32 m HNO3 electrolyte was found to be platinum. Rhodium, iridium, glassy carbon and gold electrode materials exhibited less well defined responses or also exhibited surface interferences or dissolution problems in the acid electrolyte. A 50 mm diameter electrode was selected as an ideal size for a platinum disk electrode on the basis of generating the maximum (stripping) peak to reduction current ratio, possessing a sharp, symmetrical gold oxidation peak and having advantages associated with microelectrode (radial diffusion) characteristics.An excellent response was obtained when the optimum electrode was used to determine gold in gold ore samples. The authors thank T. Hughes and G. Scollary for numerous helpful discussions and Unichema International for financial assistance and the provision of the gold ore samples.References 1 Qu, Y. B., Analyst, 1996, 121, 139. 2 Turyan, I., and Mandler, D., Anal. Chem., 1993, 65, 2089. 3 Lintern, M., Mann, A., and Longman, D., Anal. Chim. Acta, 1988, 209, 193. 4 Kaplin, A. A., Pichugina, V. M., and Filichkina, O. G., Zavod. Lab., 1988, 54, 4. 5 Hall, G. E. M., and Vaive, J. E., Chem. Geol., 1992, 102, 41. 6 Jakubec, K., and Sir, Z., Anal. Chim. Acta, 1985, 172, 359. 7 Brainina, Kh. Z., Gornostaeva, T. D., and Pronin, V.A., Anal. Chem. (USSR), 1979, 34, 831; Zh. Anal. Khim., 1979, 34, 1081. 8 Gornostaeva, T. D., and Pronin, V. A., Anal. Chem. (USSR), 1971, 26, 1549; Zh. Anal. Khim., 1971, 26, 1736. 9 Larkins, P. L., Anal. Chim. Acta, 1985, 173, 77. 10 Koppenol, M., Cooper, J. B., and Bond, A. M., Am. Lab., 1994, 26, July, 25. 11 Mocak, J., Bond A. M., Mitchell, S., and Scollary, G., Pure Appl. Chem., 1997, 69, 297. 12 Bruk, B. S., Pozina, M. I., and Rozenfeld, E. I., Anal. Chem. (USSR), 1979, 34, 842, Zh.Anal. Khim., 1979, 34, 1095. 13 Bard, A. J., Encyclopedia of Electrochemistry of the Elements, Marcel Dekker, New York, 1976, vol. 6, p. 232. 14 Llopis, J., Catal. Rev., 1968, 2, 161. 15 Vasileva, L. N., and Koroleva, T. A., Anal. Chem. (USSR), 1973, 28, 1875; Zh. Anal. Khim., 1973, 28, 2107. 16 McCreery, R. N., in Electroanalytical Chemistry. A Series of Advances, ed. Bard, A. J., Marcel Dekker, New York, 1991, vol. 17, p. 259. 17 R. E. Panzer and P. R. Elving, Electrochim.Acta, 1975, 20, 635. 18 Huiliang, H., Jagner, D., and Renman, L., Anal. Chim. Acta, 1988, 208, 301. 19 Alexander, R., Kinsella, B., and Middleton, A., J. Electroanal. Chem., 1978, 93, 19. 20 Gao, Z., Li, P., Dong, S., and Zhao, Z., Anal. Chim. Acta, 1990, 232, 367. 21 Bond, A. M., Analyst, 1994, 119 (11) 1R. 22 Brainina, Kh. Z., and Bond, A. M., Anal. Chem., 1995, 67, 2586. Paper 7/02632C Received April 17, 1997 Accepted June 18, 1997 Fig. 8 Determination of gold(iii) in a gold ore sample by using linearsweep stripping voltammetry in dilute aqua regia electrolyte at a 50 mm platinum disk electrode with 4 3 13 mg l21 standard additions of gold(iii) with a deposition time of 140 s and a scan rate of 200 mV s21. 1152 Analyst, October 1997, Vol. 122 Comparison of the Gold Reduction and Stripping Processes at Platinum, Rhodium, Iridium, Gold and Glassy Carbon Micro- and Macrodisk Electrodes Alan M. Bonda, Steven Kratsisa, Shelly Mitchellb and Jan Mocakc a Department of Chemistry, Monash University, Clayton, Victoria 3168, Australia b Department of Chemistry, Latrobe University, Bundoora, Victoria 3083, Australia c Department of Analytical Chemistry, Slovak Technical University, SK-81237 Bratislava, Slovak Republic The gold AuIII + 3e2 ? Au0 reduction and Au0 ? AuIII + 3e2 oxidation stripping processes in dilute aqua regia electrolyte (0.1 m HCl + 0.32 m HNO3) were examined at platinum, rhodium, iridium, gold and glassy carbon disk electrodes.After ascertaining that the preferred material was platinum, the effect of electrode size was evaluated by using nine different platinum disk electrodes having diameters ranging from 2 to 2000 mm. The optimum analytical response was obtained with a 50 mm diameter platinum disk electrode. With this electrode diameter, a sharp symmetrical gold stripping peak was obtained and the deposition process occurred predominantly under conditions of radial diffusion so that stirring of the solution was not required.In contrast, larger sized platinum electrodes produced a broader, asymmetric stripping response for the gold oxidation peak, whereas electrodes of smaller diameter provided poorer signal-to-noise ratios. The limit of detection and limit of quantification were calculated to be 4.4 3 1027 m (86 ppb) and 13.1 3 1027 m (258 ppb), respectively, at the 50 mm diameter platinum disk electrode under conditions of linear sweep stripping voltammetry at a scan rate of 200 mV s21 and a 140 s deposition time.The optimum electrode gave a very well defined gold oxidation signal with negligible background current when applied to the determination of gold in a gold ore sample. Keywords: Gold determination; voltammetry; optimum electrode size and electrode material There are numerous permutations and combinations of electrode materials and size available for use in analytical applications of voltammetry.Although the nature of the electrode material is often carefully considered in the development of a voltammetric method of analysis, there are relatively few studies in which the performance has been studied as a function of electrode size. However, the importance of electrode size is now emerging more frequently as a significant variable with the commercial availability of microdisk electrodes having diameters of around 10 mm. In contrast, common commercially available macrodisk electrodes typically have diameters in the millimetre range.Numerous analytical methods are available for the determination of the precious metals and many of them have recently been reviewed by Qu.1 A wide range of carbon, metal and chemically modified electrodes have been proposed for the voltammetric determination of gold.1,2 However, the specific dependence of the analytical response on the nature and size of the electrode material has yet to be determined.In this paper, the reduction of gold(iii): AuIII + 3e2 ? Au0 (1) and the stripping of gold: Au0 ? AuIII + 3e2 (2) in dilute aqua regia electrolyte (0.1 m HCl + 0.32 m HNO3) has been studied at platinum, rhodium, iridium, gold and glassy carbon disk electrodes to determine the optimum electrode surface in the analytical sense. Having ascertained that platinum is the preferred electrode material, the response at a range of electrode diameters varying from 2 to 2000 mm was examined to determine the optimum electrode size.The limit of detection (LOD) and limit of quantification (LOQ) were calculated for a typical set of conditions and the optimum electrode was applied to the determination of gold in a gold ore sample. Experimental Chemicals The dilute aqua regia electrolyte (0.1 m HCl + 0.32 m HNO3) was prepared by dilution of analytical-reagent grade nitric and hydrochloric acids. Dilution of the acids was effected with high-purity water (17 MWcm; Nanopure, Barnstead, IA, USA).Tests for interferences from iron and copper were undertaken using analytical-reagent grade FeSO4·7H2O and CuSO4·5H2O, respectively. Gold standard solutions were prepared by dilution of a 1000 ppm Gold Spectrosol solution (BDH, Poole, Dorset, UK) with the dilute aqua regia electrolyte. All glassware was cleaned in a detergent solution, followed by boiling nitric acid and then rinsing with water prior to drying at 200 °C. Preparation of the Gold Ore Sample The method of sample preparation used prior to the voltammetric analysis of gold ore samples, donated by Unichema International, was as follows: 1.The ore sample was dried at 105 °C and crushed to less than 100 mm particle size. 2. A 25 g amount of the sample was accurately weighed into a 250 ml beaker and 50 ml of 5 m HCl were added. The resultant mixture was left to stand at room temperature for 30 min to destroy sulfides. 3. The solution was heated on a hot-plate at approximately 120 °C for 2 h, with the beaker being covered with a watchglass. 4. After the solution had cooled, 10 ml of concentrated HCl and 10 ml of concentrated HNO3 were added. The solution was then heated for 4 h, allowing the liquid volume to decrease but not to go to dryness. 5. The solution was allowed to cool and 25 ml of 8 m HCl were added. After stirring, the solution was filtered into a separating funnel with the residues being rinsed with 2 m HCl. 6. The filtrate was extracted three times with diethyl ether (3 3 20 ml).Analyst, October 1997, Vol. 122 (1147–1152) 11477. The ether extract was washed with 0.1% HCl to remove iron. 8. The ether extracts were evaporated over 10% HCl using a heater. When the ether had volatilised, the gold remained in the acid solution. The sample was then added to dilute aqua regia electrolyte (3 : 1 ratio) and was ready for voltammetric analysis. The use of diethyl ether solvent extraction to remove interferences has been reported,3,4 as have other forms of solvent extraction.5–9 Equipment All voltammetric measurements were undertaken at 20 °C with a BAS 100A Electrochemical Analyzer (Bioanalytical Systems, West Lafayette, IN, USA) equipped with a BAS PA-1 preamplifier and Faraday cage.Where necessary, oxygen was removed by degassing solutions with nitrogen. The standard three-electrode arrangement was used with a platinum wire counter electrode, an Ag/AgCl (3 m KCl) reference electrode and the following working electrodes: 2000, 500, 250, 100, 70, 50, 25, 10 and 2 mm diameter platinum disk electrodes, 500 mm diameter glassy carbon, rhodium and iridium disk electrodes and a 2000 mm diameter gold disk electrode. The working electrodes were made via methods related to those described in the literature10 using metal wires from Goodfellow Metals (Cambridge, UK). Prior to each voltammetric experiment, the working electrode surface was polished on a LECO polishing pad with 0.6 mm alumina slurry, then rinsed with distilled water and carefully dried.Method for the Evaluation of Detection Limits According to the IUPAC definition, the limit of detection is the lowest concentration of an analyte that an analytical process can reliably detect. Originating from this definition, the signal counterpart of the limit of detection (LOD) is located 3sb (three times the standard deviation) or 6sb above the gross blank signal or, equivalently, 3sb or 6sb above zero when using the net signal. The use of 6sb instead of 3sb is substantiated by the tendency to diminish type II error11 causing an inadequate judgment of a true signal as the blank, although this is not important in our further discussion of the problem.If the calibration plot is a perfect straight line passing through the origin (net signals are assumed), then the limit of detection is simply determined by projecting the intercept of 3sb with the calibration plot on to the concentration axis. By analogy, the limit of quantification (LOQ) refers to the lowest concentration, or another quantity, which can be quantitatively measured with reasonable reliability by a given procedure.Its signal counterpart was defined as 10sb above the gross blank signal, or 10sb above zero when the net signals are used. In ref. 11, a 9sb (nine times the population standard deviation) approach is advocated instead. Also in this case, the use of a calibration plot is necessary for the calculation of the LOQ.When using the described standard method of obtaining the LOD, three main factors can considerably influence the reported result. The first occurs commonly because the number of measurements made on the blank and/or the sample itself is not sufficiently large to fulfil the conditions required to ensure that the normal distribution is valid for the measured signal. This problem may be overcome by replacing the value 3sb by the corresponding product t(n, a)sb where t(n,a) is the critical value of the t-distribution, n is the number of degrees of freedom and a is the confidence level.11 The second problem is related to inconsistency of the position of the zero net signal and the intercept of the calibration plot, denoted q0.The accuracy of the evaluated LOD depends on how positive or negative q0 is. When q0 is negative, the projection through the calibration plot leads to an over-estimation of the LOD, whereas a positive value leads to an under-estimation or even in extreme cases to a negative LOD value.If the intercept q0 is ignored, the LOD signal is projected through a line parallel to the net signal calibration plot, having the same slope q1 but a zero intercept (which is equivalent to calculation of the LOD as 3sb/q1) via this procedure, the negative LOD values are prevented and the over- and under-estimation proceed again but in the opposite way.The third problem is associated with the uncertainty of the position of the calibration line itself. A line of best fit could be shifted by chance or even by a statistical manipulation in such a way that a desired (usually lower) LOD is obtained owing to the improper values of regression parameters. Flaws in the ways in which the IUPAC definition of the LOD is commonly interpreted can be minimised by using a method known as the upper limit approach (ULA),11 which also is used in this paper.Using the ULA, the upper one-sided confidence limits of individual signal observations are calculated, which depend on the errors in regression and also on the type of the calibration model. If a one-parameter straight-line calibration model is statistically correct (the straight line passing through the origin), then the upper confidence limit for any chosen concentration c0 is distant from the regression line by the value of t(n,a)s, where n is now n 2 1 and s is defined by the equation s = (1 + c0 2/Sci 2)1/2 sy (3) with sy denoting the residual standard deviation.For the calculation of the limit of detection by the ULA, it is necessary to consider the case where c0 = 0 in eqn. (3). Under these conditions, the product in eqn. (3) refers to the upper limit signal at zero concentration. Further, when c0 = 0 and thus s = sy, the limit of detection can be simply calculated as LOD = t(n,a) sy/q1 (4) where n = n 2 1 and q1 is the slope of the calibration plot.Differences in the standard approach and the ULA method are illustrated in Fig. 1. Results and Discussion Selection of the Optimum Electrode Material The initial investigations concerned the selection of the optimum working electrode material for the voltammetric determination of gold(iii) in 0.1 m HCl + 0.32 m HNO3 electrolyte, which for convenience is refered to as dilute aqua regia. For this study, 2000 mm diameter platinum and 500 mm diameter platinum, glassy carbon, rhodium and iridium electrodes and a 2000 mm diameter gold electrode were chosen.Platinum macrodisk electrodes Cyclic voltammograms at 2000 and 500 mm diameter platinum macrodisk electrodes in dilute aqua regia showed a well defined response with the AuIII + 3e2?Au0 reduction and (Au0 ? AuIII + 3e2) oxidation peak potentials of 0.550 and 0.900 V versus Ag/AgCl, respectively, giving a peak-to-peak separation of 0.350 V with the 500 mm diameter electrode under the conditions in Fig. 2(a). An important feature with the use of this electrode material is the sharp gold oxidation peak which is required for the sensitive stripping method. Two closely spaced stripping peaks have been observed in other work,12 but under the conditions of the present study only a single stripping peak was found. At the platinum macrodisk electrodes, both the reduction and oxidation peaks when using cyclic voltammetry varied linearly with gold concentration over the range 2 3 1148 Analyst, October 1997, Vol. 1221025–2 3 1023 m and the limit of detection was found to be 2 3 1026 m and the limit of quantification 2 3 1025 m for the reduction process at a scan rate of 375 mV s21. Further, the reduction peak current was found to be dependent on the square root of the scan rate over the scan rate range 20–1000 mV s21 with a slope of 0.51 being obtained from a log(scan rate) versus log (peak height) plot. These data are consistent with a (linear) diffusion-controlled reduction process. The stripping process gave a slope of 0.76 when plotting log(scan rate) versus log (peak height) when the switching potential was 400 mV versus Ag/AgCl. Iridium electrode The first scan at the iridium working electrode exhibited similar characteristics to that found at the platinum electrode, with a sharp gold oxidation peak, reduction and oxidation peaks of 0.530 and 0.870 V versus Ag/AgCl, respectively, and a peak-topeak separation of 0.340 V under the conditions in Fig. 2(b). However, on repetitive cycling, the nature of the response changes as dissolution of the electrode occurs. With repetitive cycling, the gold reduction peak increases in current and shifts to more negative potentials (0.05 V shift by the fourth scan). Conversely, the gold stripping peak decreases in height with each cycle, but does not shift in potential [Fig. 2(b)]. The changes in the voltammetry with repetitive cycling are attributed to dissolution of the iridium metal electrode which is accompanied by the development of an IrIII–IrIV redox couple. 13,14. The dissolution is detected voltammetrically by the development of an oxidation current at 0.700 V versus Ag/ AgCl. Rhodium electrode Potential cycling of a gold(iii) solution using a rhodium electrode in dilute aqua regia also was accompanied by rhodium metal electrode dissolution [Fig. 2(c)]. Problems associated with dissolution of a particular electrode material are detected via an enhanced background current when monitoring the voltammetry of dilute gold solutions.For example, stripping voltammograms obtained from 3 31026 m gold(iii) solutions at platinum, rhodium and iridium electrodes with 2 3 1026 m standard additions using a 140 s deposition time and a scan rate of 200 mV s21 are compared in Fig. 3. The background with the platinum electrode where no dissolution occurs is very small in comparison with that observed with the rhodium and iridium electrodes.Glassy carbon electrode Under the conditions in Fig. 2(d), the cyclic voltammogram of gold(iii) with a glassy carbon working electrode showed reduction and oxidation peak potentials of 0.400 and 0.950 V versus Ag/AgCl, respectively, and a large peak-to-peak separation of 0.550 V for the gold(iii) reduction and gold metal reoxidation. However, apart from the considerable irreversibility of the process, the response is well defined [Fig. 2(d)].At other forms of carbon electrode15 two gold stripping peaks have been reported, although only a single response was observed under the conditions of this study. Gold electrode A gold electrode can be used for the determination of gold via use of the gold reduction process. As shown in Fig. 2(e), the Fig. 1 Illustration of differences in determining the limit of detection by (a) the upper limit approach for the case when the calibration model is represented by a straight line passing through the origin (ULA1) and (b) the standard approach (SA) based on the IUPAC definition, calculating LOD1 by means of the calibration line and LOD2 by using an auxiliary line parallel to the calibration line (the same slope but a zero intercept).Fig. 2 Cyclic voltammograms of a 3.5 3 1024 m gold(iii) solution obtained in dilute aqua regia electrolyte at a scan rate 20 mV s21 using (a) platinum (500 mm), (b) iridium (500 mm), (c) rhodium (500 mm), (d) glassy carbon (500 mm) and (e) gold (2000 mm) working electrodes.Analyst, October 1997, Vol. 122 1149peak potential for reduction of gold when using a gold electrode was 0.550 V versus Ag/AgCl, which is comparable to values obtained at platinum and iridium electrodes. Obviously, stripping voltammetry of gold at a gold electrode is not possible. Further Considerations Concerning the Choice of Working Electrode Material The above data suggest that either platinum and glassy carbon may be the preferred electrode materials for the determination of gold.However, at low gold(iii) concentrations, problems with the background current arise when using the glassy carbon electrode. Various oxygen-containing functional groups have been identified by various workers which can be present on the carbon surface (e.g., phenol, carbonyl and quinone groups).16,17 Scanning to positive potentials oxidises these functional groups, forming a multilayer oxide which increases the background current.An additional problem with the glassy carbon electrode arises from the fact that the reduction peak potential for gold(iii) is almost 200 mV more negative than found at a platinum electrode. Therefore, when using a stripping technique for the gold determination, a considerably more negative deposition potential has to be used, which can cause problems with impurities such as iron, copper and lead, which are reduced at these more negative potentials.Hence, these elements, which are often present at relatively high levels in gold ore samples, may cause interference in the determination of gold at a glassy carbon electrode, but not at platinum. The problems with the presence of iron and the use of a glassy carbon electrode can be seen by comparing voltammograms obtained at platinum and glassy carbon electrodes for a 10 mg l21 gold(iii) solution spiked with 15 mg l21 [Fig. 4(a) and (b)] and 150 mg l21 iron(iii) [Fig. 4(c) and (d)].The platinum electrode also is favourable for minimising interferences from copper because a deposition potential of 250 mV versus Ag/ AgCl can be used, which is more positive than the potential for the reduction of copper. Fig. 4(e) shows a stripping voltammogram of a 10 mg l21 gold(iii) solution spiked with 15 mg l21 copper(ii). Further details on intereferences likely to be encountered in the voltammetric determination of gold are available.18–20 Glassy carbon electrodes have been widely used for the determination of gold;5,6 in contrast and perhaps suprisingly, according to our findings, platinum indicating electrodes have been only rarely used.12,18 Determination of the Optimum Platinum Electrode Size for the Determination of Gold Evaluation of the optimum size platinum electrode was undertaken by comparing voltammograms obtained with a 3.5 3 1024 m gold(iii) solution using nine different platinum disk electrode sizes ranging from 2000 to 2 mm in diameter at a scan rate of 20 mVs21 (Fig. 5). At large electrode sizes (diameters of 2000, 500 and 250 mm) with this scan rate the reduction process exhibits a peak-shaped response whereas at smaller diameters (100–2 mm) a sigmoidal-shaped response is observed as the diffusion of gold(iii) to the electrode surface changes from linear at the larger diameter macrodisk electrodes to radial at the smaller diameter microdisk electrodes.21 The gold oxidation stripping peak also changes in shape from relatively broad and slightly asymmetric at larger diameters to sharp and symmetrical with a well defined baseline for diameters @50 mm.Further, with the smaller electrode sizes, stirring of the solution is not required during the deposition step22 so that microelectrodes may offer advantages over conventionally sized electrodes when using stripping methods. The change in voltammetry as a function of electrode diameter at a constant scan rate can be represented graphically by plotting Fig. 3 Stripping voltammograms obtained in dilute aqua regia electrolyte at 500 mm diameter (a) platinum, (b) rhodium and (c) iridium electrodes using a 140 s deposition time and a scan rate of 200 mV s21 for 3 3 1026 m gold(iii) solutions with 2 3 1026 m gold(iii) standard additions. Fig. 4 Stripping voltammograms obtained in dilute aqua regia electrolyte using a 140 s deposition and a scan rate of 200 mV s21 for a 10 mg l21 gold(iii) solution after addition of 10 mg l21 iron (iii) at (a) platinum (500 mm) and (b) glassy carbon (500 mm) electrodes and after addition of 150 mg l21 iron(iii) containing the same (c) platinum and (d) glassy carbon electrodes. (e) Stripping voltammogram obtained at a platinum (500 mm) electrode for 10 mg l21 gold(iii) solution containing 15 mg l21 copper(ii). 1150 Analyst, October 1997, Vol. 122the ratio of stripping peak current to the reduction peak or limiting current versus the logarithm of the area of the electrode. From this form of data analysis it can be observed that initially there is an increase in the ratio of oxidation to reduction currents as the size of the electrode is decreased until a diameter of about 50 mm is reached (Fig. 6). At very small sizes, the reproducibility for the gold process decreased, presumably owing to variable electrode areas resulting from greater difficulty in reproducibly polishing the smaller sized electrodes. Therefore, a 50 mm diameter electrode was selected as optimum on the basis of possessing a sharp, symmetrical gold oxidation peak, efficient electrolysis (oxidation/reduction value) and a favourable signalto- noise ratio.In practice, the 50 mm diameter electrode also represents close to the largest size electrode available for achieving near steady-state behaviour in practical analytical voltammetry. Hence this size of electrode is also about the largest for which stirring of the solution is not required to enhance the gold deposition process.Determination of the Limits of Detection and Quantification Using a 50 mm Diameter Platinum Microdisk Electrode The limits of detection and quantification under a typical set of conditions were determined using the 50 mm diameter platinum disk electrode and linear-sweep stripping voltammetry. Using a 140 s deposition time and a scan rate of 200 mV s21, a small but reproducible signal could be obtained with a 4 3 1027 m gold(iii) solution.A calibration plot of peak current versus concentration was constructed using 4 3 1027 m as the lowest analyte concentration and 4 3 1027 m standard additions (Fig. 7). Using the upper limit approach, the LOD was determined to be 4.4 3 1027 m (86 ppb) and the LOQ 13.1 3 1027 m (258 ppb). Obviously, the LOD and LOQ are strongly influenced by scan rate, deposition time and deposition potential. However, the conditions used above certainly provide adequate sensitivity for the determination of gold in gold ore samples, which was the problem of practical interest in this study.Determination of Gold in a Gold Ore Sample The 50 mm platinum microdisk electrode was used to determine gold in ore samples provided by Unichema International (Port Fig. 5 Cyclic voltammograms obtained in dilute aqua regia electrolyte at a scan rate of 20 mV s21 for 3.5 31024 m gold(iii) solutions using platinum electrodes with sizes varying from (a) 2000 to 70 mm and (b) 50 to 2 mm in diameter.Fig. 6 Variation in the ratio of anodic to cathodic peak or limiting currents versus log (electrode area) for the gold oxidation and gold(iii) reduction processes. Experimental conditions as in Fig. 6. Fig. 7 Stripping voltammogram obtained in dilute aqua regia electrolyte for a 140 s deposition time and a scan rate of 200 mV s21 at a 50 mm platinum disk electrode for a 4 3 1027 m gold(iii) solution with 4 x 1027 m standard additions of gold(iii). Analyst, October 1997, Vol. 122 1151Melbourne, Victoria, Australia). The stated concentration in the sample considered in Fig. 8 was 32 mg l21. A well defined stripping response was observed with a 140 s deposition time and a scan rate of 200 mV s21 . Standard additions of 13 mg l21 gold(iii) were made to the solution (Fig. 8) and a plot of peak current versus concentration of added gold(iii) was constructed. The gold concentration was calculated from the intercept and found to be 28 mg l21, which compares favourably with the stated value of 32 mg l21. Equally, well defined gold stripping voltammograms were obtained for other gold ore samples provided by Unichema International.Differential-pulse Stripping Voltammetry Differential-pulse stripping voltammetry is widely used in trace analysis. However, for the determination of gold at a platinum electrode, stripping studies at the 1026 m concentration level produced broader and more complex signals of lower reproducibility than obtained under linear-sweep conditions. The lack of improvement associated with the use of the differential-pulse method is probably associated with the extreme irreversibility of the gold redox chemistry.Consequently, the linear-sweep method is preferred for the determination of gold at platinum disk electrodes. Conclusions The preferred electrode material for the voltammetric determination of gold in 0.1 m HCl + 0.32 m HNO3 electrolyte was found to be platinum.Rhodium, iridium, glassy carbon and gold electrode materials exhibited less well defined responses or also exhibited surface interferences or dissolution problems in the acid electrolyte. A 50 mm diameter electrode was selected as an ideal size for a platinum disk electrode on the basis of generating the maximum (stripping) peak to reduction current ratio, possessing a sharp, symmetrical gold oxidation peak and having advantages associated with microelectrode (radial diffusion) characteristics. An excellent response was obtained when the optimum electrode was used to determine gold in gold ore samples. The authors thank T. Hughes and G. Scollary for numerous helpful discussions and Unichema International for financial assistance and the provision of the gold ore samples. References 1 Qu, Y. B., Analyst, 1996, 121, 139. 2 Turyan, I., and Mandler, D., Anal. Chem., 1993, 65, 2089. 3 Lintern, M., Mann, A., and Longman, D., Anal. Chim. Acta, 1988, 209, 193. 4 Kaplin, A. A., Pichugina, V. M., and Filichkina, O. G., Zavod. Lab., 1988, 54, 4. 5 Hall, G. E. M., and Vaive, J. E., Chem. Geol., 1992, 102, 41. 6 Jakubec, K., and Sir, Z., Anal. Chim. Acta, 1985, 172, 359. 7 Brainina, Kh. Z., Gornostaeva, T. D., and Pronin, V. A., Anal. Chem. (USSR), 1979, 34, 831; Zh. Anal. Khim., 1979, 34, 1081. 8 Gornostaeva, T. D., and Pronin, V. A., Anal. Chem. (USSR), 1971, 26, 1549; Zh. Anal. Khim., 1971, 26, 1736. 9 Larkins, P. L., Anal. Chim. Acta, 1985, 173, 77. 10 Koppenol, M., Cooper, J. B., and Bond, A. M., Am. Lab., 1994, 26, July, 25. 11 Mocak, J., Bond A. M., Mitchell, S., and Scollary, G., Pure Appl. Chem., 1997, 69, 297. 12 Bruk, B. S., Pozina, M. I., and Rozenfeld, E. I., Anal. Chem. (USSR), 1979, 34, 842, Zh. Anal. Khim., 1979, 34, 1095. 13 Bard, A. J., Encyclopedia of Electrochemistry of the Elements, Marcel Dekker, New York, 1976, vol. 6, p. 232. 14 Llopis, J., Catal. Rev., 1968, 2, 161. 15 Vasileva, L. N., and Koroleva, T. A., Anal. Chem. (USSR), 1973, 28, 1875; Zh. Anal. Khim., 1973, 28, 2107. 16 McCreery, R. N., in Electroanalytical Chemistry. A Series of Advances, ed. Bard, A. J., Marcel Dekker, New York, 1991, vol. 17, p. 259. 17 R. E. Panzer and P. R. Elving, Electrochim. Acta, 1975, 20, 635. 18 Huiliang, H., Jagner, D., and Renman, L., Anal. Chim. Acta, 1988, 208, 301. 19 Alexander, R., Kinsella, B., and Middleton, A., J. Electroanal. Chem., 1978, 93, 19. 20 Gao, Z., Li, P., Dong, S., and Zhao, Z., Anal. Chim. Acta, 1990, 232, 367. 21 Bond, A. M., Analyst, 1994, 119 (11) 1R. 22 Brainina, Kh. Z., and Bond, A. M., Anal. Chem., 1995, 67, 2586. Paper 7/02632C Received April 17, 1997 Accepted June 18, 1997 Fig. 8 Determination of gold(iii) in a gold ore sample by using linearsweep stripping voltammetry in dilute aqua regia electrolyte at a 50 mm platinum disk electrode with 4 3 13 mg l21 standard additions of gold(iii) with a deposition time of 140 s and a scan rate of 200 mV s21. 1152 Analyst, October 1997, Vol. 122