Analytica Chimica Acta 376 (1998) 325330

Simultaneous determination of traces of iron(II) and iron(III) using differential pulse anodic stripping voltammetry in a ow-through conguration on a glassy carbon electrode
J.F. van Staden*, M.C. Matoetoe
Department of Chemistry, University of Pretoria, Pretoria 0002, South Africa Received 21 October 1997; received in revised form 21 July 1998; accepted 23 July 1998

Abstract Iron(II) and iron(III) complexes of pyrophosphate can be determined simultaneously at pH 9. The peak potentials used were 0.8 and 0.5 V for Fe(III) and Fe(II), respectively. Linear calibration plots over the range 106103 mol l1 were obtained. Detailed iron-couple cyclic voltammetric studies, sample handling procedures, possible interferences and applications of the method to real samples are described. Detection limits of 108 mol l1 and a relative standard deviation of <4% (n14) were achieved at a concentration of 1106 mol l1 for both iron species. # 1998 Elsevier Science B.V. All rights reserved.
Keywords: Iron(II); Iron(III); Glassy carbon electrode; Flow injection; Flow-through

1. Introduction The determination of the form of an element in environmental compartments is important for assessment of its biological activity (bioaccumulation, bioconcentration, bioavailability and toxicity) [1,2]. For essential elements like iron, speciation methods are needed because their bioavailability and metabolism are strongly dependent on the species involved. Particularly the iron uptake is inuenced by its oxidation state and complexation by inorganic and organic ligands. Techniques used to determine iron in environmental and geological samples include spectrophotometry,
*Corresponding author. Tel.: +27-12-420-2515; fax: +27-12432-863; e-mail: [email protected] 0003-2670/98/$19.00 # 1998 Elsevier Science B.V. All rights reserved. PII S0003-2670(98)00518-2

voltammetry, and amperometry [35]. They have also been applied in the differentiation of the two oxidation states. Recent publications indicate that ow injection (FI) systems in conjunction with amperometry or spectrophotometry [6,7] and atomic absorption spectrometric detection as well as a combination of liquid chromatography (electrochemical detection) with ame atomic absorption detection [8] have made a useful contribution to iron speciation studies. Typically, one form together with total iron is determined allowing for the calculation of the iron in the second oxidate state by the difference between measured values. These methods have a limited applicability especially when the ratio of iron in the two oxidation states is almost the same. Almost all the schemes described are not simultaneous analysis but

326

J.F. van Staden, M.C. Matoetoe / Analytica Chimica Acta 376 (1998) 325330

split streams, two injections or multi-detection systems. In addition, although these methods can differentiate the two iron forms they are tedious, time consuming and mostly inaccurate [913]. Simultaneous determination of iron in each oxidation state has been done by FI and kinetic spectroscopy. The system used a silver reductor and two ow cells aligned in the same optical path to yield two peaks [14]. Amperometric detection coupled with FI and differential pulse polarography was also used in the analysis of rocks [14] and anaerobic sealants [15]. In some cases the assays were performed by simultaneous extraction into chloroform of tris(8quinolinato)iron(III) and tris(4,7-diphenyl-1,10,phenan-throline)iron(II) prior to their electrochemical reduction in propylene carbonate [16]. These methods, although they worked very well, are laborious. This paper concentrates in investigating a possible one-stream simultaneous determination of traces of Fe(II) and Fe(III) by means of differential pulse anodic stripping voltammetry in a ow-through conguration using a glassy carbon (GC) electrode. Differentiation between Fe(II) and Fe(III) was achieved by an appropriate choice of supporting electrolyte. Optimum conditions, possible interferences and the electrode reactions were studied. 2. Experimental 2.1. Apparatus The continuous ow-through trace analyser system, cell design and three electrode arrangement used were described in detail previously [17]. An exception was the use of a Mautolab system (Ecochemie, Utrecht, Netherlands) connected to a model 663 VA stand (Metrohm, Herisau, Switzerland) for voltammetric measurements replacing the 647 VA stand and 646 microprocessor (Metrohm) previously used. The Mautolab controls the experimental parameters such as initial and nal potential scan increments and provides peak location, peak height and peak area readings as well as blank corrections. The electrode used was hand polished every morning with 0.3 A alumina and electrochemically cleaned with 3 mol l1 HNO3 between samples.

2.2. Reagents and samples All chemicals used were of analytical grade. Ammonium iron(II) sulphate and iron(III) nitrate were used for preparation of the standard solutions of iron(II) and iron(III), respectively. A 200 mg l1 stock solution was prepared in deaerated 2% (v/v) HNO3 (it was stable for a week) and working standards were prepared daily by suitable dilution of this stock solution in the matrix required. High grade nitrogen was used for deaeration of samples and solvents, and during analysis. All reagents and samples were prepared with doubly distilled water (ultra pure water). 2.3. Procedure A study of the relevant literature and some optimization experiments (see below) led to the following recommended procedure. A supporting electrolyte containing 0.1 mol l1 Na4P2O7 and 1 mol l1 NH4Cl buffer at pH 9.0 were separately sucked simultaneously into the deaeration chamber (described before [18]) using the peristaltic pump. The mixture was deaerated with nitrogen and passed through the ow-through cell at 4.7 ml min1. Samples were introduced into the reagent mixture at 2 ml min1 just after deaeration in the chamber. The solutions were allowed to merge, react and ow through the cell. Subsequently the accumulation potential or deposition potential (Dp) of 1.2 V was applied for a selected time (Dt). After the deposition period the ow was stopped, a rest period of 5 min was observed and a positive going differential pulse scan (25 mV s1) was commanded. The scan was stopped at 0.1 V, and the system was immediately ready for the next run. Each set of experiments was repeated three times and the averages were calculated. 3. Results and discussion 3.1. Selection of the supporting electrolyte A supporting electrolyte was required which allowed the formation of well-dened, sufciently separated peaks with iron in the two oxidation states. Ideally there should be no interaction of a given oxidation state on each other's wave. If present, the

J.F. van Staden, M.C. Matoetoe / Analytica Chimica Acta 376 (1998) 325330

327

interaction should be small and easily correctable. Several solutions were investigated in order to choose the most efcient supporting electrolyte. They were tartrate, citrate, thiocyanate, borate, ethylenediaminetetra-acetate (EDTA), diaminocyclohexane tetra-acetate (CDTA), ethylene glycol-bis(-amino ethyl ether)N,NH tetra-acetate (EGTA), catechol, thioglycollic acid and 1,10-phenanthroline. It was found that almost all of them formed a peak with only one form of iron. EDTA and its derivatives, however, formed peaks with both iron forms. It was impossible, to use any of these reagents due to unresolved stripping peaks of both iron forms even after varying the pH of the solutions. Preliminary examination of pyrophosphate suggested that this might be a satisfactory supporting electrolyte. Its application appeared to be uncomplicated and, furthermore, detailed studies of copper and iron couple pyrophosphate systems had already been reported [19,20]. In addition Brennan and Svehla [16] had used pyrophosphate in determining the speciation of iron using a mercury electrode. Their work highlighted difculties encountered with differential pulse voltammetry, therefore it was important to study in detail the behaviour of the iron couple in basic pyrophosphate solution. It was found during these investigations that the peak potentials of iron(II) and iron(III) were affected by pH as shown in Table 1. The best resolution of the peaks for iron(II) and iron(III) was obtained with a sodium pyrophosphate buffer at pH 9.0. 3.2. Voltammograms of the iron couples in alkaline pyrophosphate solution A condition for the appearance of an anodic stripping voltammetric (ASV) signal is the accumulation of a compound (adsorbed and/or reduced) of interest on the electrode surface. With a potential shift in the anodic direction, the accumulated compound is

Fig. 1. Differential pulse anodic stripping voltammograms: (a) mixture of Fe2 (0.5 V) and Fe3 (0.8 V); (b) Fe3 alone. Concentrations of every iron solution1104 mmol l1.

stripped off the electrode and the ASV peak appears. To determine if such accumulation takes place as well as its nature, cyclic voltammograms and differential pulse (DP)ASV scans were run. The different oxidation states of iron in pyrophosphate (diphosphate) buffer at pH 9.0 formed two distinct stripping peaks, shown in Fig. 1(a). Under these conditions Fe3 forms a very stable complex with pyrophosphate, the stripping of which occurs at a more negative potential than that of Fe2, as shown by Fig. 1(b), which conrmed the work of Brennan and Svehla [15]. Fig. 2 clearly shows a difference between Fe3 and Fe2 in this medium. A Fe3 cyclic voltammetry (CV) scan (Fig. 2) at the resting potential of 1.5 V exhibits two anodic peaks; an almost symmetric large peak at 0.609 V (1) and a small wave at 0.086 V (2). The latter disappears as the resting potential becomes more positive than 1.2 V. Subsequent cathodic scans were characterised by a small wave at 0.975 V (3) not visible as a result of its small current. This cathodic scan shows that Fe3 is therefore effectively masked by the strong complex formed with the pyrophosphate. Areas and peak heights of these peaks increased as the resting potential becomes more negative. There is thus a strong indication that the deposition reaction of Fe3

Table 1 Comparison of peak voltages (mV) for Fe3 and Fe2 in different pH media using pyrophosphate Sample Peak voltage (mV) Na4P2O7 alone Fe Fe3
2

Na4P2O7 buffered at pH 4.0 597 603

Na4P2O7 buffered at pH 9.0 597 758

349 445

328

J.F. van Staden, M.C. Matoetoe / Analytica Chimica Acta 376 (1998) 325330

which depends on the resting potential, since the peaks become more negative as the resting potential becomes more cathodic. At the resting potential of 1.2 V two sets appear. It therefore seems that the pyrophosphate buffered medium at a pH 9 supplied the necessary bit of irreversibility for the determination of both Fe3 and Fe2, which conrmed the work of Parry and Anderson [19].
Fig. 2. Cyclic voltammetric scan of 1104 mmol l1 Fe3 in an alkaline pyrophosphate medium; pH9.0.

3.3. Optimisation of parameters To obtain a better response performance from the system (high sample throughput, adequate sensitivity, working range and minimum wastage of carrier solution) optimisation of the instrumental parameters was a necessity which was done by varying the ow rates, deposition time and deposition potential. Fe3 and Fe2 yield well-dened stripping peaks at 0.8 and 0.5 V, respectively, as shown in Fig. 1. The peak responses increase with increasing deposition time, ow rate and as the deposition potential becomes more cathodic. This indicates enhancement of deposition on the electrode surface. However, the peak increase is accompanied by appearance of multiple peaks with an increase in Dt and more cathodic Dp, which resulted in irreproducible stripping peaks. An accumulation potential of 1.2 V and a Dt of 3 min are recommended. A ow rate of 4.7 ml min1, which was the highest studied, was used. 3.4. Interferences Two kinds of interferences are expected in this system, namely the presence of strong complexing agents in the sample which are capable of removing both iron(II) and iron(III) from the pyrophosphate complex and the presence of metal ions in the sample which can be deposited or absorbed on the GC electrode resulting in unresolved stripping peaks from those of the iron oxidation states. There are reports that state that the Fe2 pyrophosphate complex reduces solutions of silver, gold and mercury salts [21]. Therefore the presence of these elements in the sample may affect the iron couple system. Fortunately these elements are not common in most samples and as a result they were not studied. Metals that are deposited on a GC electrode and have potentials closer to that of the two oxidation

and therefore accumulation in ASV experiments, due to the very stable complex formed by iron(III) with phosphate [15] (Fig. 1), is more adsorption controlled as opposed to reduction controlled. The Fe2 CV scan at the same resting place (Fig. 3) is characterised by two cathodic (0.909 V (a) and 0.489 V (b)) and two anodic peaks (0.760 V (c) and 0.101 V (d)). Correlation between the areas and heights of the anodic peak (c) and the cathodic peak (b) strongly suggest that we have a chemically reversible system. The two peaks maybe assigned to oxidation of Fe2 to Fe3and reduction of Fe3 to Fe2, respectively. It was further found that as the resting potential becomes more anodic, only one anodic and one cathodic wave at 0.514 and 0.438 V appear. These waves increased in size and shifted slightly to a more anodic potential of 0.304 and 0.362 V as the resting potential became more anodic (0.9 V). The system becomes more reversible as the resting potential becomes more anodic. Based on these observations we can conclude that the Fe2 accumulation in ASV is based on a simple electron transfer reaction

Fig. 3. Cyclic voltammetric scan of 1104 mmol l1 Fe2 in an alkaline pyrophosphate medium pH9.0.

J.F. van Staden, M.C. Matoetoe / Analytica Chimica Acta 376 (1998) 325330

329

states are Cu, Pb, Cd and Zn. Cu interference had been dealt with in detail in the literature. The correction methods proposed were tried successfully [20]. The effect of the other metals were studied by observing the appearance of the iron peaks in the presence and the absence of these metals. It was found that when 1103 mol l1 of each of these species (Pb, Cd, Zn) are added to 2.5104 mol l1 of Fe2 or Fe3, Zn showed no possible interference. It formed a completely resolved stripping peak from both of the iron stripping peaks and their peak heights were unchanged. On the other hand both Pb and Cd produced overlapping peaks with Fe2 and Fe3, respectively. The effects in both cases were additive. An additive interference can easily be corrected if the amount of the interferent is known. The amounts of Pb and Cd can easily be determined in the presence of iron. An unexpected interference was the presence of iron(III) in iron(II) standards, resulting in a small stripping peak, the intensity of which increased with increasing concentration. Two methods can be applied to overcome this particular problem. First, the amount of iron(III) in iron(II) can be determined and the appropriate calculation can be done. Secondly, the standard addition method can be used without any corrections, because the concentrations of the standards used are very small. During analysis of real samples no interference correction was made because the standard addition method was used. Cd and Pb interference was not present in wine samples, but tap water contained Pb. In this sample the Pb and Fe2 stripping peaks were resolved probably because their concentrations were small. Investigation of organic interference was not

Table 2 Qualitative parameters for iron(II) and iron(III) Parameters % RSD Limit of detection (LOD) (108 mol l1) Linear range (mol l1) Regression coefficient (n6) Fe3 3.7 2.4 106104 0.987 Fe2 2.3 2.7 106104 0.998

Flow rate: 4.7 ml min1, deposition potential 1.2 V at a concentration of 1106 mol l1 (n14).

done based on the assumption that their concentrations are small due to the dilution of the samples. 3.5. Analytical application Under the experimental conditions selected, analytical results in terms of the linear range of the concentration versus peak current signal plot, limit of detection (LOD) and precision expressed as per cent relative standard deviation (% RSD) are reported in Table 2. The limit of detection was determined as a signal to noise ratio of 3 from the lowest concentration. To test the accuracy and to determine the analytical application of the method, some wine and tap water samples were analysed. The samples were diluted (14) with 2% (v/v) HNO3 prior to analysis. The determinations of the different iron oxidation states were done using the standard addition method. These results were compared with those obtained by standard spectrophotometric methods based on the use of KSCN and 1,10-phenanthroline, at wavelengths of 508 and 475 nm for Fe3 and Fe2, respectively. Relevant results are summarised in Table 3. As can

Table 3 Mean resultsSD(n4) obtained for the simultaneous determination of iron(II) and iron(III) in real samples and comparison with spectrophotometric analysis Sample Proposed method (DPASV) Fe3 (mmol l1) White wine (Graca) Red wine Tap water
a

Spectrophotometry Fe3 (mmol l1) (508 nm) 0.0250.009 0.219a0.093 (0.806)b 0.4810.042 Fe2 (mmol l1) (475 nm) 0.2580.045 0.330a0.067 (0.815)b 0.1490.035

Fe2 (mmol l1) 0.2630.005 0.3380.006 0.1530.0245

0.0310.004 0.2260.007 0.4920.004

The colour of red wine interfered in the spectrophotometric determination of Fe2 and Fe3. These results were obtained by running the samples without any reagent as blank to counteract colour interference. b Results obtained without any action to counteract colour interference for the spectrophotometric methods.

330

J.F. van Staden, M.C. Matoetoe / Analytica Chimica Acta 376 (1998) 325330

be seen there is a good agreement with the standard spectrophotometric methods. The colour of red wine, however, interfered in the spectrophotometric determinations of both Fe2 and Fe3. This is clear from the results of 0.806 mmol l1 obtained for Fe3 and 0.815 mmol l1 for Fe2 compared to the values of 0.226 and 0.338 mmol l1 for Fe3 and Fe2, respectively, for the proposed system (Table 3). In both cases the spectrophotometric results were higher. More realistic values were obtained for the spectrophotometric determinations by running the samples without any reagent as blank to counteract for colour interference. This simultaneous method is therefore a promising feature for the applicability of the simultaneous speciation determination of iron(II) and iron (III) in real matrices. It is also clear from the results in Table 2 that the major advantages of the simultaneous determination of trace amounts of iron(II) and iron(III) using DPASV in a ow-through conguration on a GC electrode over spectrophotometric methods is the simultaneous speciation in a single system and the interference free results from coloured solutions which should give more accurate results. The analysis time of 57 min indicates a possibility of at least 812 runs an hour. This is very good compared to other reports of at least 10 min analysis time [16].

References
[1] T.M. Florence, Trends in Anal. Chem. 2 (1983) 162. [2] D.R. Turner, Metal Ions Biol. Systems 18 (1984) 137. [3] A.I. Vogel, Textbook of Qualitative Inorganic Analysis, 4th ed., Longman, London, 1986. [4] C.M.G. van den Berg, M. Nimmo, O. Abollino, E. Mentasti, Electroanalysis 3 (1991) 477. [5] G. Visalakshi, S.V. Narasimhan, K.S. Venkateswarlu, Anal. Chim. Acta 212 (1988) 335. [6] K. Oguma, S. Kozuka, K. Kitada, R. Kuroda, Fresenius' J. Anal. Chem. 341 (1991) 545. [7] N. Clarke, L.G. Danielsson, Anal. Chim. Acta 306 (1995) 5. [8] G. Weber, Fresenius' J. Anal. Chem. 340 (1991) 161. [9] J.N. Wilson, Analyst 109 (1984) 839. [10] R. Kuroda, T. Nara, K. Oguma, Analyst 113 (1988) 1557. [11] J. Wang, S. Mannino, Analyst 114 (1989) 643. [12] D. Berggren, Int. J. Environ. Anal. Chem. 41 (1990) 133. [13] T.P. Lunch, N.J. Kernoghan, J.N. Wilson, Analyst 109 (1984) 843. [14] J.W. Dieker, W.E. van der Linden, Anal. Chim. Acta 114 (1980) 267. [15] M.C. Brennan, G. Svehla, Anal. Proc. 26 (1989) 343. [16] L.E. Leon, D.T. Sawyer, Anal. Chem. 53 (1981) 706. [17] J.F. van Staden, M. Matoetoe, Fresenius' J. Anal. Chem. 357 (1997) 624. [18] J.F. van Staden, M. Matoetoe, S. Afr. J. Chem. 50(3) (1997) 128. [19] E.P. Parry, D.P. Anderson, Anal. Chem. 45(3) (1973) 458. [20] G.W. Higgins, P.E. Sturrock, Anal. Chem. 41 (1969) 633. [21] J.W. Mellor, Comprehensive Treatise on Inorganic and Theoretical Chemistry, vol. XIV, Longman, London, 1935.