ANALYST, JUNE 1986, VOL. 111 625 Voltammetry of Hexacyanoferrates Using a Chemically Modified Carbon-paste Electrode Kurt Kalcher lnstitut fur Analytische Chemie, Karl-Franzens Universitat, Universitatsplatz 1, A-80 10 Graz, Austria A simple electroanalytical method for the quantitative determination of hexacyanoferrate(l1) and -(Ill) has been developed using a carbon-paste electrode, chemically modified with a weakly basic anion exchanger. The complexes can be pre-concentrated at the electrode surface prior to voltammetric determination. For differential-pulse voltammetric measurements with suitable methodical parameters, a linear relationship between the current and concentration was found for 0.3 to 6000 pg I-' of Fe. The method is applicable to the quantitative determination of hexacyanoferrate(l1) in wine down to 6 pg 1-1 of Fe.Factors influencing the signal response like pre-concentration time and ionic effects have also been investigated. Keywords: Chemically modified carbon-paste electrode; liquid anion exchanger; hexacyanoferrate determination; differential-pulse voltammetry; wine analysis The development of chemically modified electrodes (CME) as an electroanalytical tool has been growing rapidly during the last few years. The aim of increasing the selectivity and the sensitivity of common polarographic and voltammetric methods can be achieved by attaching functional groups to the surface of the working electrode. The feasibility of pre- concentrating electroactive species, based on chemical or physico-chemical reactions with the modifying agent, often lowers the detection limit down to the ultratrace level.Ion exchangers in particular have attracted much interest in this respect, and thus many ionic components have been analysed using electrodes chemically modified with ion exchangers for voltammetric determinations, e.g., chromate,l copper(II),2 iodide,3 gold (III)4 and iridium(IV).S Although the trapping of hexacyanoferrate(II1) in a cationic polymeric film on a platinum electrode was shown to be possible by Facci and Murray6 it has not been used for trace analytical applications. The aim of this work was to establish the optimum experimental conditions for analysing hexacyanoferrates at low concentrations. The usual methods, which include spec- trophotometry of the coloured aggregates of iron, titration of the hexacyanoferrate(I1) with permanganate or cerium(IV), determination of iodine [oxidised by hexacyanoferrate(III)], voltammetry using electrodes with working ranges at positive potentials or decomposition into iron and hydrogen cyanide ,7 are not very selective or sensitive.By adding a liquid ion exchanger to a carbon paste, voltammetric determination can be preceded by a pre- concentration period during which the ionic electroactive species accumulate on the surface of the electrode. This makes trace level analysis possible. Additionally, as this step involves no change of oxidation states, no potential needs to be applied and pre-concentration can be performed independently from the measurement. This offers another advantage; the medium exchange after the accumulation step separates off many components that interfere with the electrochemical determi- nation; these components often show little or no affinity to the ion exchanger.As it is possible to regenerate the surface to its initial state, the electrode filling can be used repeatedly, resulting in fast analyses and a low consumption of the carbon paste accompanied by greater precision. The method described above has been examined to see if it can be applied to the determination of hexacyanoferrates in wine. For the treatment of wine, K4[Fe(CN),] is often used as a fining agent and is added to the wine in order to precipitate heavy metals, including iron, and the resulting (blue) lees are filtered off. It is evident that the amount added must be evaluated on the basis of a chemical analysis to avoid too high dosages.The excess of hexacyanoferrate(I1) remaining in the wine can produce very toxic cyanocompounds and, by decomposition , potassium cyanide itself. Although many other substitutes, such as aferrine, have been introduced, hexacyanoferrate(I1) is still used owing to its low price. Because of the potential danger, it seems important to control the contents of this compound in wine even at low concentra- tions. Pure wine cannot be used directly for the accumulation phase as the high contents of organic compounds block up a large area of the surface and reduce the sensitivity drastically, so it must be diluted in a ratio of at least 1 + 9 or better still, 1 + 19. In this manner the detection limit is acceptably low with respect to the trace analysis of toxic hexacyanoferrate( 11).Of course, as hexacyanoferrate(II1) is also pre-concentrated at the electrode to almost the same degree, it cannot be distinguished from [Fe(CN)$- and is determined together with it. Experimental Apparatus For the voltammetric measurements, a PAR 264 A polaro- graph (Princeton Applied Research) was used in combination with a self-constructed electrode assembly of Plexiglass.8 The cell consisted of a titration vessel of glass (EA 880-20, 20-90 ml, from Metrohm) with a platinum wire as the counter electrode and a saturated calomel electrode (SCE) as a reference. The latter was in contact with the solution over a salt bridge with a Vycor frit, filled with 1 M KC1 solution.The curves were registered on a two-channel recorder. For determining peak heights smaller than 5 mm the data were evaluated digitally on a HP 1000 mini-computer after A - D conversion by an appropriate interface.9 The receptacle for the test solution was a 50-ml glass beaker equipped with a PTFE-coated stirring bar (30 mm, 0.d. 7 mm), which was driven by a variable speed magnetic stirrer. It was placed under a suitable holder for the electrode. Working Electrode The body of the working electrode was fabricated from a PTFE rod (0.d. 10 mm) with a 3 mm deep bore hole (diameter 7 mm) on one side for the carbon-paste filling. Contact to this was made by a platinum wire glued into a bore hole in the centre of the rod.626 ANALYST, JUNE 1986, VOL. 111 The carbon paste was prepared according to Monien et al.1" A 5-g mass of spectral carbon powder (RWB, Ringsdorff- Werke) was mixed thoroughly with 1.8 ml of liquid paraffin (Uvasol, Merck). Commercially available carbon paste (EA 267 C from Metrohm) can also be used. For each gram of this paste, 0.05 ml of liquid ion exchanger, Amberlite LA2 (Fluka), was added, and the substances were mixed thoroughly to a homogeneous consistency. This mixture was placed in the corpus of the electrode using a spatula of PTFE and smoothed off. An electrode prepared in this way can be used, if regenerated appropriately, for at least ten determina- tions without any notable change in the signal response. Reagents De-ionised and doubly distilled water that had been purified by a cartridge de-ionisation system (Nano-pure from Barn- stead) was used throughout.Hydrochloric acid and sodium hydroxide were of Supra-pure quality (Merck) and all other reagents were of analytical-reagent grade (pro analysi, Merck). Stock solutions of the hexacyanoferrates were prepared with concentrations of 1 mg ml-1 of Fe by dissolving 151.27 mg of K4[Fe(CN)6].3H20, or 117.92 mg of K,[Fe(CN),], respectively, in 20 ml of water. They were stored in the dark at 4 "C and could be used for at least one week. Solutions with lower concentrations were made freshly by dilution. For the regeneration of the electrode, 20 ml of saturated, aqueous sodium chloride solution were mixed with 0.2 ml of NaOH (30% solution). It was prepared freshly when needed. As standard wines, two Austrian types were chosen which contained no determinable amount of hexacyanoferrate( 11).The white wine was a Griiner Veltliner, a light table wine (referred to as the standard white), and the red wine (standard red) was Blaufrankisch, both originating from Burgenland. They were used directly from the bottle without any pre- treatment. Procedure Pre-concentration A 20-ml volume of the test solution containing the hexa- cyanoferrate(I1) or-(111) was acidified with hydrochloric acid to a pH of 2 (0.02 ml of 10 M HCl). When analysing wine, 1 ml of the sample was diluted with 19 ml of water and acidified with 0.02 ml of 10 M HCI. The electrode was placed in the holder exposing the carbon paste surface to the solution, which was then stirred at a rate of 300 rev.min-1. After allowing to stand for the required time, the electrode was removed, rinsed with water for about 0.5 s, placed in the voltammetric cell and connected to the polarograph. Voltammetry Water (20 ml), acidified with 0.05 ml of 10 M HCI to give a final HCl concentration of 2.5 x 10-2 M as the supporting electrolyte and de-aerated by passing through pure nitrogen (99.999%) for 3 min, was used for the voltammetric measure- ments. Quantitative analyses were performed in the differen- tial-pulse mode (DPV). The potential range was set from -0.15 to +0.4 V vs. SCE in the anodic direction regardless of the oxidation state of the iron in the cyano complex. The pulse height was 50 mV and the scan rate 10 mV s-1 with a drop time of 0.5 s. In order to settle the solution and eliminate the high background currents an equilibration phase of 15 s with an applied initial potential preceded all the voltammetric measurements. Cyclovoltammograms (CV) were recorded from -0.3 to +0.5 V vs.SCE for hexacyanoferrate(I1) and in the reverse direction for [Fe(CN),I3- The scan rate was 50 mV s-1 in both instances. The current range was selected according to the expected peak height. For accurate evaluations of the peak heights for low concentrations , the background current was synthesised and subtracted from the curve .9 Higher concen- trations yielded peaks that could be determined manually from the recorder output with sufficient precision by means of a tangent fit. Regeneration After trapping the iron complex on the surface of the electrode with a subsequent voltammetric determination, the functional groups of the ion exchanger were regenerated by a 2-min treatment with a well stirred, alkaline saturated sodium chloride solution (to remove the complex) followed by exposure for 2 min to stirred 2 M HCl (for re-formation of the ammonium groups). This procedure was also applied to the freshly prepared, virgin electrodes to yield reproducible results.Evaluation of quantitative results A quantitative determination consisted of a regeneration step, the registration of the blank of the unloaded electrode with DPV, the accumulation of the iron complex and the recording of the potential - current curve with DPV. This method was repeated twice with the appropriate addition of specified amounts of the iron complex to the test solution (standard additions method).The concentration was obtained from the peak currents by the usual mathematical methods. The peak height was calculated either by a tangent fit for higher concentrations, or synthesis and subtraction of the back- ground. The blank subtract method is also applicable. Results and Discussion In acidic aqueous solutions both hexacyanoferrate(I1) and -(III) form ion aggregates with water-immiscible, weakly basic ion exchanger molecules at the electrode - solution interface: [Fe(CN),I4- + 4R2NH,'C1- ++ (R2NHi)4[Fe(CN)6] + 4C1- (1) [Fe(CN),I3- + 3 R2NH,'C1- ++ (R2NHi)3[Fe(CN)6] + 3 C1- There is no evidence that all the negative charges are compensated by ammonium ions; protons may replace them to a certain degree.Owing to the very similar behaviour of both species, at least a one-step protonation of the hexa- cyanoferrate(I1) ion in equation (1) is probable, which would lead to an exchange reaction analogous to equation (2). However, the absolute composition of the ion pairs formed is not of great importance in the method; as these ions associate generally with ammonium compounds both complexes can be pre-concentrated at the electrode surface. As equations (1) and (2) or variations of them involve no change in the oxidation states or electron transfers, accumulation can be carried out without the application of a potential. When, after trapping the complex on the surface of the electrode, an electrical field varying with time during voltammetric measurement is imposed, oxidation of Fe(I1) or reduction of Fe(II1) can occur according to the potential and the oxidation state of the metal ion as follows: (2) (3) This electron transfer leads to a current response and can be seen clearly in the corresponding cyclovoltammograms (Figs.1 and 2). The pictures show almost reversible processes for both complex species as expressed in equation (3). The potentials of the cathodic and anodic maxima, E,, and E,,, are displayed in Table 1. For the reduction, E,, shifts much more with increasing cycle number than does E,, for the oxidation. Because of this, all differential-pulse measurements are also carried out in the anodic direction to give better reproducibil-ANALYST, JUNE 1986, VOL. 111 ity. If hexacyanoferrate(II1) is accumulated it is reduced to hexacyanoferrate(I1) during the equilibration phase and re-oxidised when recording the voltammogram. In CV, the current of the maxima decreases with growing cycle number because small amounts of the trapped-in cyanoferrate diffuse from the electrode surface leading to a signal reduction.The regeneration process makes use of this diffusion and supports it in two ways: firstly, the high concentration of chloride competes with the active sites of the exchanger and replaces cyanoferrate; and secondly, the alkaline medium reduces the protonation of the ammonium groups and thus prompts the detach reaction. Of course, the cationic state must be remade by acid treatment. As stated above , anodic differential-pulse voltammetry (ADPV) has been chosen for quantitative determinations because of its high sensitivity and accuracy.Fig. 3 shows the voltammogram of hexacyanoferrate(n) carried out in this mode. For the oxidised species the resembling curves are recorded. The peak potentials are slightly dependent on the amount of accumulated ions, being shifted in the cathodic 627 direction with increasing current. As the background shows quasi-linear functionality within the peak range, the tangent fit method can be applied fairly exactly for evaluation of peak height. Only for very small signals should more sophisticated methods be applied.9 For analysing wine, it was first checked that pre- concentration could be carried out in the alcoholic medium. The results are summarised in Fig. 4, which shows the dependence of the voltammetric peak current on the ethanol contents of an acidified aqueous solution of [Fe(CN)&-.The relationship between alcohol concentration and signal response is a linear function where increasing amounts reduce the peak height. Thus, with respect to ethanol, determinations are possible because, at low concentrations, the changes in the signal are not disastrous. Two table wines, a white and a red, with undetectable amounts of hexacyanoferrate(I1) were chosen to serve as standard media for working out the analytical conditions of the method. Practice has shown that pure, undiluted wine contains many organic compounds that produce no voltammetric signals by themselves but block the r 1 0.3 0.1 -0.1 -0.3 PotentialN vs. SCE Fig. 1. Cyclovoltammogram of hexacyanoferrate(I1) solution.[Fe(II)], 1 ng 1-1; accumulation time, 1.5 min 25 f % g o 3 0 -25 I I I I 0.5 0.3 0.1 -0.1 PotentialN vs. SCE Fig. 2. Cyclovoltammogram of hexacyanoferrate(II1) solution. [Fe(III)], 1 mg I - * ; accumulation time, 1.5 min -0.1 0.1 0.3 PotentiaW vs. SCE Fig. 3. Anodic differential pulse voltammogram of hexacyano- ferrate(I1). A, Background; B, 200 pg 1 - l of Fe(I1); and C, 400 pg 1-1 of Fe(I1). Accumulation time, 30 s 0 5.0 10.0 15.0 20.0 Ethanol content, % Fig. 4. [Fe(II)], 100 pg 1-1 as [Fe(CN),]4-; accumulation time, 1 min Relationship between peak current and ethanol content. Table 1. Maximum cathodic and anodic potentials (Ep, and Ep,) with their corresponding currents (ip, tration of Fe, 1 mg 1-1 Oxidation state Fe(I1) . .. . . . Fe(I1) . . . . . . Fe(I1) . . . . . . Fe(II1) . . . . . . Fe(II1) . . . . . . Fe(II1) . . . . . . Fe(1I) . . . . . . Fe(II1) . . . . . . Method cv cv cv cv cv cv ADPV ADPV Deposition tirnels 60 60 60 60 60 60 30 30 Cycle 1 2 3 1 2 3 - EP 1 Vvs. k E 0.030 0.033 0.035 0.029 0.030 0.031 - EP 1 Vvs. k E 0.135 0.135 0.134 0.130 0.130 0.130 0.080 0.085 34.0 32.5 31.6 28.2 27.0 26.8 - and ips). Concen- 37.0 32.8 31.5 27.0 26.8 26.6 45.0 27.6628 ANALYST, JUNE 1986, VOL. 111 surface of the electrode by adsorption and hence reduce the extent of the ion exchange drastically. Therefore, it is necessary to dilute the wine to at least 10 times (or better to 20 times) its volume prior to the accumulation step. In this way a loss of sensitivity of about 10-20% compared with aqueous solutions still permits the determination of trace levels of hexacyanoferrate.For the following investigations with respect to wine, 1-ml samples were mixed with 19 ml of water and used as described under Experimental. The effect of pre-concentration time on the peak current for an aqueous medium is displayed in Figs. 5 and 6. Similar graphs have been obtained for both types of cyano complexes. With standard white wine the dependence of the voltammetric current on the accumulation time is shown in Fig. 7. An almost identical graph has been registered for the red wine. Both are very similar to pure aqueous solutions with respect to their qualitative shape. A linear run for short periods of time soon deviates into a hyperbolic function approaching a limiting value.This behaviour can be explained by the mass action law, which governs the exchange process according to equations (1) and (2). It is self-evident that accumulation can take place to some extent only where equilibrium is reached and desorption equals sorption. Hence, the maximum amount of hexacyanoferrate on the electrode surface is controlled by the number of functional exchanger groups , the concentration of ions in the bulk solution, and the concentration distribution I I I I 1 0 1 2 3 4 5 Ti me/m i n Fig. 5. Dependence of the peak current on the accumulation time for solutions of hexacyanoferrate(I1) [Fe(II)]: A, 250 pg 1-l; and B, 5 mg 1-1 120 - 40 1 - 0 1 2 3 4 5 Time/m in Fig. 6. Relationship between the peak current and accumulation time for solutions of hexacyanoferrate(II1).[Fe(III)]: A, 250 pg I-’; and B, 5 mg 1 - 1 coefficient Dc, which expresses the affinity of an ionic species to an ion exchanger , and is defined as The subscript ex. designates the concentration of the complex as the ion aggregate with the exchanger and sol the concen- tration in the solution. The state of motion during pre- concentration also influences the absolute shape of the graphs but does not alter either the qualitative result or the equilibrium, i.e., the limiting value of the current. There may be two reasons why this maximum current is reduced in wine compared with aqueous solutions containing no alcohol: firstly, the concentration distribution coefficient [equation (4)] is decreased (as can also be seen in Fig.4); and secondly, the adsorption of organic molecules inhibits the ion exchange process notably. A transition of the ion exchanger from the surface into the alcoholic medium could not be confirmed as the reproducibility of the electrode is maintained by the regeneration step. The relationship between concentration and peak current for pure aqueous solufions at selected accumulation times, t,, is shown in Figs. 8 and 9. In spite of the different oxidation states of the metals the two types of complexes behave similarly, leading to the conclusion that their affinities to the 1.0 2.0 3.0 4.0 5.0 0 Ti me/m in Fig. 7. Dependence of the peak current on the accumulation time. [Fe(II)] [as hexacyanoferrate(II)], 250 pg I-’ after dilution of 1 ml standard white wine to 20 ml I /u f . 40 2 3 0 20 0 0.2 0.4 0.6 0.8 1 .o Fig.8. Dependence of the peak current on the concentration of hexacyanoferrate(I1) solutions. Accumulation times and concen- trationranges: A, 5 min, 0-10 pg 1-1 of Fe, current axis extended 10 times; B, 3 min, 0-100 pg 1 - 1 of Fe; C, 30 s, 0-1 mg 1 - 1 of Fe; and D, 5 s, 0-10 mg 1 - 1 of Fe Concentration rangeANALYST, JUNE 1986, VOL. 111 629 Concentration range Fig. 9. Dependence of the peak current on the concentration of hexacyanoferrate(II1) solution. Accumulation times and concen- tration ranges as in Fig. 8 I I 0 0.2 0.4 0.6 0.8 1 .o Concentration range Fig. 10. Dependence of the peak current on the concentration of hexacyanoferrate(I1) in standard white wine. Concentration ranges and accumulation times: A, 0-100 pg 1-1 of Fe, 3 min, current axis extended 5 times; B, 0-1 mg 1-1 of Fe, 2 min; C, 0-10 mg 1-I of Fe, 30 s; D , 0-100 mg 1-l of Fe, 5 s 1 1 0 0.2 0.4 0.6 0.8 1 .o Concentration range Fig.11. Dependence of the peak current on the concentration of hexacyanoferrate(I1) in standard red wine. Concentration ranges and accumulation times: A , 0-100 pg I-' of Fe, 3 min, current axis extended 10 times; B, C , and D as in Fig. 10 ion exchanger must be of the same magnitude although that for hexacyanoferrate( 11) is slightly higher. Direct propor- tionality exists over a wide range of concentration if an appropriate pre-concentration period is chosen. Hence , linear relationships can be observed for up to 6 mg 1-1 of iron for [Fe(CN)6I4- and up to 8 mg 1-1 of iron for [Fe(CN)6I3-.Therefore, the best way of determining hexacyanoferrate concentrations is the internal standard additions method. Solutions containing concentrations higher than 4 mg 1-1 of Fe must be diluted prior to analysis. In practice, the detection limit lies at 0.3 pg 1-1 of Fe, which corresponds to about 1.1 pg 1-1 of hexacyanoferrate and a molar concentration of 5.4 x M. For a series of five determinations of 50 pg 1-1 of Fe(I1) as [Fe(CN)&- (t, = 2 min) the coefficient of variation was calculated as 1.5% with two standard additions. (It is advisable to add at least two standards in order to check the linearity between the concentration and the signal response.) As shown in Figs. 10 and 11, a linear relationship exists between the concentration and voltammetric peak heights over a wide range for both white and red wine, with suitable accumulation times. Thus, reduction in the signal height by components like alcohol does not imply a breakdown in the linearity.It can be seen that both types of wine, white and red, behave similarily . With short deposition times linearity can be observed for up to 60 mg 1-1 of Fe as [Fe(CN)6]4- in the wine. In practice the detection limit lies at about 6-7 pg 1-1 of Fe in wine, corresponding to about 25 pg 1-1 of [Fe(CN)6]4-. Of course, hexacyanoferrate(II1) , [Fe(CN),I3- , which could be present in wines by oxidation processes, is co-accumulated on the electrode surface to almost the same extent as the hexacyanoferrate(I1) complex leading to a combined determi- nation of both species.From a toxicological point of view this fact is not important, as a distinction between the two types is not particularly relevant. The recovery rate was determined with spiked wines when the test solution was measured once with only one standard addition (Table 2). At low concen- trations the deviation is usually larger than at higher ones. The estimated maxima of deviations, resulting from practical experience, are also listed in Table 2. The error of the result can be decreased by performing one measurement twice, in addition to the standard additions. However, it is usually not necessary to do so because the accuracy, as indicated in Table 2, is often adequate. The fact that the recovered concen- trations of the white standard lie above (and those of the red wine below) the real contents is an artefact of this series and cannot be reproduced in any way.At this point it should be stressed that all the steps of the procedure should be carried out with the greatest care, and always in the same manner, in order to achieve the utmost homogeneity of results. Many errors arise from the non- equivalent treatment of electrodes (regenerating, rinsing, stirring, etc.) resulting in non-representative amounts of complex being accumulated by the ion exchanger. It should also be pointed out that the measurements of added standards have to be performed with the same electrode filling to avoid discrepancies on the electrode surface. For pure aqueous solutions, the influence of other ionic components on the current was investigated (Table 3).Similar behaviour was found for hexacyanoferrate(I1). Only per- chlorate and salicylate cause severe disturbances. It should be emphasised that changes in peak heights due to other ions that are larger than the standard deviations do not imply that the determination is impossible. In most instances the linear relationship between signal and concentration is maintained and only the sensitivity is altered. However, this must be checked in every single instance. Finally, the hexacyanoferrate(I1) content of ten Austrian white and red wines and two champagnes was investigated. It was shown that wines with higher sugar contents behaved analytically in the same way as the standard types used. Six of the wines had no determinable amounts of hexacyanoferrate630 ANALYST, JUNE 1986, VOL.111 ~~ Table 2. Recovery rates and maxima of deviations of hexacyanoferrate(I1) in spiked wines; evaluation with one standard addition Standard wine White . . . . White . . . . White . . . . Red . . . . Red . . . . Red . . . . Accumulation time . . 3 min . . 1 min . . 10 s . . 3 min . . 1 min . . 10 s Spiked concentration of Fe/ yg 1 - 40 400 4000 40 400 4000 Concentration of Fe foundlyg ml-1 44.7 415 4080 38.9 381 3940 Estimated maxima of deviation, '/O f 15 f 10 2 5 t 15 + 10 +5 Table 3. Interference of anionic components to the determination of hexacyanoferrate(I1) in aqueous solution. Concentrations: Fe, 100 yg 1-1; and anionic component, 10 mg I-'. Accumulation time, 60 s Change in Component Added as Concentration/M peak current, YO Br- .. * . I- . . . . * . NO2- . . . . NO,- . . . * HW4- . . . . so32- . . . . so42- . . . . CN- . . . . Acetate . . . . Oxalate . . . . Tartrate . . . . Citrate . . . . Ascorbate . . Salicylate . . Tannin . . . . c104- . . . . KBr 1.3 x 10-4 KI 7.9 x 10-5 NaN02 2.2 x 10-4 KN03 1.6 x 10-4 KH2P04 1.0 x 10-4 Na2S03 1.2 x 10-4 K 8 0 4 1.0 x 10-4 NaCIO, 1.0 x 10-4 KCN 3.8 x 10-4 G I 3 4 0 2 1.7 x 10-4 C2H204 1.1 x 10-4 C4H606 6.8 X 10-5 C ~ H S N ~ ~ O ~ 5.3 X lop5 C6H806 5.0 x 10-5 C7H603 7.2 x 10-5 C76H52046 -5.9 X lop6 +3.1 -15.0 + 10.4 -1.2 -2.6 +4.7 -7.9 -43.7 -0.2 -4.5 +7.0 + 10.2 + 10.0 -3.5 -46.5 -7.2 and four contained between 7 and 8 pg 1-1 of Fe as hexa- cyanoferrate(I1). As the wines are at least one year old it is not possible to deduce whether the original content of hexa- cyanoferrate(I1) was low or whether the complexes had vanished owing to decomposition processes. The champagnes did not show any detectable hexacyanoferrate(I1). Conclusion The basic conditions for a method to determine trace levels of hexacyanoferrate( 11) and -( 111) with a chemically modified carbon paste electrode have been described. As a linear relationship is achieved between concentration and voltam- metric signal over a wide range, the standard additions method can be applied to quantitative evaluations. The determination of hexacyanoferrate in both white and red wines is possible. The ease and simplicity of preparing the electrodes is also advantageous. 1. 2. 3. 4. 5 . 6. 7. 8. 9. 10. References Cox, J. A., Anal. Chim. Acta, 1983, 154, 71. Wang, J., Greene, B., and Morgan, C., Anal. Chim. Acta, 1984, 158, 15. Kalcher, K., Fresenius 2. Anal. Chem., 1985, 321, 666. Kalcher, K., Anal. Chim. Acta, 1985, 177, 175. Kalcher, K., Fresenius 2. Anal. Chem., in the press. Facci, J., and Murray, R. W., J. Phys. Chem., 1981,85,2870. Burger, N., Talanta, 1985, 32, 49. Kalcher, K., Fresenius 2. Anal. Chem., 1986, 323, 238. Kalcher, K., and Jorde, C., Compput. Chem., in the press. Monien, H., Specker, H., and Zinke, K., Fresenius 2. Anal. Chem., 1967,225, 342. Paper A5f398 Received November 4th, 1985 Accepted December 30th, 1985