Fresenius J Anal Chem (2001) 371 : 431–436 DOI 10.1007/s002160101080
S P E C I A L I S S U E PA P E R
H. Below · H. Kahlert
Determination of iodide in urine by ion-pair chromatography with electrochemical detection
Received: 4 April 2001 / Revised: 8 August 2001 / Accepted: 13 August 2001 / Published online: 14 September 2001 © Springer-Verlag 2001
Abstract A variety of parameters affecting the determination of iodide in biological materials by ion-pair chromatography and electrochemical detection were examined in detail. It became apparent that the pH value, the ionpair concentration, the proportion of organic solvent and of organic bases as a component of the buffer solution, as well as the salt concentration in the eluent system could effectively influence the retention characteristics of iodide in the chromatographic system, resulting in the separation of potential interfering substances. The presence of other anions in the sample matrix has to be taken into consideration, particularly thiocyanate because of its long retention time. Investigations of the electrochemical detection mechanism revealed that the reaction hitherto assumed to be responsible for detector signal generation (formation of AgI) is incorrect. In addition, a much more sensitive detection of iodide than that cited in the literature to date is possible if the detector potential is optimally selected and any anticipated interfering substances are removed by chromatography. Use of a gold electrode rather than a silver electrode also considerably enhances the reliability of the procedure.
Introduction Numerous biochemical processes are influenced by the trace element iodine. Since the natural resources of iodine in Germany are generally considered as inadequate [1], iodide analysis is important in ascertaining alimentary io-
H. Below (✉) Ernst Moritz Arndt University Greifswald, Institute of Hygiene and Environmental Medicine, Hainstrasse 26, 17493 Greifswald, Germany e-mail:
[email protected] H. Kahlert Ernst Moritz Arndt University Greifswald, Institute of Chemistry and Biochemistry, Soldmannstrasse 16, 17487 Greifswald, Germany
dine deficiency. Moreover, the use of antiseptics containing iodine is widespread. Yet little information is currently available about the resorption of iodine on application of iodophors and its toxicological relevance. Iodide usually occurs in concentrations of less than 1 mg L–1 in biological materials such as urine or serum, making sensitive detection essential. Thus the number of analytical methods available is very limited. The classical wet chemical method of quantifying the iodide concentration is based on the Sandell-Kolthoff reaction, the Wawschinek modification of which [2] has been used for several years in our working group [3]. This reaction is very sensitive to interference and dependent on a variety of factors [4, 5, 6]. Determination of iodide in serum samples, for example, is not possible with this procedure. Radiochemical processes [5] are distinguished by a low limit of detection and a high degree of accuracy. The same applies to ICP–MS (inductively-coupled plasma–mass spectrometry), a standard method for the determination of iodine within the framework of food analysis in terms of LMBG, Article 35 (German Federal Food Legislation) [7]. Use of these procedures is restricted by the availability of the analytical technology. Further, HPLC processes are used. Particularly ion-pair chromatographic procedures [6, 8, 9, 10] are used alongside with ion-exchange methods [11, 12, 13]. In the course of ion-pair chromatography, iodide is separated from the bulk of the sample using various ion-pairing reagents. An electrochemical detector (ECD) offers the low detection limit required. Silver electrodes are used for this purpose since it is assumed that the reaction giving rise to the potential is the formation of AgI at the electrode. Preliminary readings, taken after combination of various ion-pair chromatographic procedures described in the literature [9, 10, 14, 15] showed, however, that measurements in biological material are only possible for short initial period. After that time, oxidation at the silver electrode gave rise to serious problems. It could also be shown, using a UV detector, that the chromatographic separation of iodide from other matrix components was incomplete. For this reason, the chromatographic conditions and the parameters of elec-
432
trochemical detection were investigated, with the objective of developing a reliable routine method.
Experimental Apparatus Liquid chromatography was performed with ICI DP 800 HPLCcontrol and evaluation software, a LC 1110 pump, a uniflows-degasser DG1310, a variable UV-detector LCD 500 (all Gamma Analysentechnik GmbH), and a Coulochem II ECD with silver, gold, platinum, and carbon electrodes (Bischoff Chromatography). An Aspec XLi (Abimed) was used as sampler and for solid phase extraction (SPE). Cyclic voltammetry (CV) was performed using an AUTOLAB system with a PSTAT 10 (Eco-Chemie, Utrecht, Netherlands) in conjunction with a three-electrode system and a personal computer (IBM compatible). The reference electrode was an Ag/AgCl electrode (3 M KCl) (Metrohm, Herisau, Switzerland) with a potential of 0.207 V vs. SHE at 25°C. A glassy carbon rod served as an auxiliary electrode. Prior to the measurements, the solutions were purged for 300 s with nitrogen to prevent interference by oxygen. HPLC columns and SPE unit HPLC columns used were LiChroCart 125×4.0 mm, 5 µm (Merck), Spherisorb 100×4.6 mm, 5 µm (Bischoff), Ultrasep ES Pharm RP 18, 150×3 mm, 7 µm (Sep Serv), Luna C 18(2), 150×4.6 mm, 5 µm with SecurityGuard (Phenomenex). SPE unit used was 3 mL (500 mg) C18 cartridge (Machery and Nagel). Reagents Demineralized water was produced by a Seradest SR-1400 plant and purified with the RS 40 EZ facility (both from Water Preparation and Regeneration Station Co.). KI (Suprapur) was used to prepare the calibration standard and KI (analytical grade) for the control standard (both from Merck). A mixture of methanol or acetonitrile (both LiChroSolv from Merck) with a buffer was used as the HPLC eluent system. The buffer comprised tetra-n-butylammonium hydroxide (Merck), triethylamine (Merck), or n-octylamine (Sigma), Na2HPO4.2H2O (Merck) and KH2PO4 (Merck). Adjustment of the pH value was carried out with dilute (approx. 4%) phosphoric acid (Riedel de Haën). NaCl (Merck) was also used. Methanol (LiChroSolv from Merck) was used for SPE unit preconditioning. Samples and sample pretreatment Urine samples (spontaneous samples) were stored at –20°C prior to their examination and only thawed out immediately before analysis. After thorough mixing, the samples were centrifuged and subjected to SPE. The SPE columns were preconditioned with 3 mL methanol and subsequently rinsed with 3 mL of ultrapure water. After this, 0.5 mL of urine sample or standard was applied and the eluate discarded. The eluate derived from a newly applied urine sample or from a standard (25 µL) was injected directly into the HPLC column.
Results and discussion Chromatographic separation Since a very low concentration of iodide accompanied by a large number of interfering substances at much higher
Fig. 1 Influence of the pH value on the chromatographic behavior of iodide in ion pair chromatographic systems. LiChroCart (125 mm×4.0 mm; 5 µm), buffer without organic solvent (5 mmol TBAOH, 10 mM n-octylamine, 20 mM KH2PO4, pH adjusted with H3PO4)
concentrations has to be determined in biological materials, the greatest possible degree of chromatographic separation of the analytes from the interfering substances is an absolute necessity. The pH value, the proportion of organic solvent, the salt content, and the concentration of the ion-pairing reagent in the eluent system were investigated as criteria exerting a primary influence on the retention characteristics of iodide. The retention time of iodide increases with decreasing pH value, as expected. If the pH value is decreased from 7 to 3.5, the capacity factor (k’=ts/tm) doubles, as does the retention of iodide in the chromatographic system (Fig. 1). Even small amounts of organic solvents in the eluent system induce a marked reduction in the retention of iodide. For example, the addition of 10% of methanol at a pH value of 3.5 leads to a decrease in the capacity factor from 32.8 to 11.8 (Table 1). It may generally be assumed that an increase in the concentration of ion-pairing reagent will lead to an increase in the retention time [16]. However, this tendency is less pronounced in the present chromatographic system. A doubling of the ion-pair concentration and the n-octylamine concentration (as buffer additive) at pH 3.5 only instigates a minimal increase in the capacity factor. This applies equally well to an eluent system with an additional 10% of methanol as to a system containing no methanol (Table 1). At a pH value of 4.6, on the other hand, an increase in the ion-pair concentration from 5 mmol to 10 mmol, accompanied by a simultaneous doubling of the n-octylamine content, leads to a reduction of the retention time. Apparently, the retention time prolongation effect of the ion-pairing reagent is counteracted by the n-octylamine additive (Table 1). Biological materials usually contain higher levels of NaCl. Moreover, a certain ionic strength has to be assured for the purpose of electrochemical detection. For these reasons the effect of the salt concentration on the retention characteristics of iodide was investigated. Increasing the salt concentration in the eluent system has a conspicuous shortening effect on the retention time.
433 Table 1 Influence of the ion pair concentration (as tetrabutylammonium hydroxide; TBAOH), n-octylamine and the proportion of organic solvent at the retention of iodide. LiChroCart (125 mm×4.0 mm; 5 µm), 20 mM KH2PO4, pH adjusted with H3PO4
TBAOH (mM)
5 10
n-Octylamine (mM)
10 20
TBAOH (mM)
5 10
pH 3.5
pH 4.6
Eluent with 10% methanol k’
Eluent without methanol k’
Eluent without methanol k’
11.8 12.8
32.8 33.0
26.2 16.5
in concentrations from 0.1 to 10 mg/L. Together with a UV signal, thiocyanate also generates a clearly negative ECD signal, implying that the risk of chromatogram overruns exists in the course of serial analyses as a result of its long retention time (Fig. 3). Most analytical methods described in the literature refer to analysis times of under 10 min. In our opinion, such times are too short to definitely eliminate thiocyanate interference.
Electrochemical detection (ECD)
Fig. 2 Influence of the salt concentration on the chromatographic behavior of iodide in ion pair chromatographic systems. LiChroCart (125 mm×4.0 mm; 5 µm) buffer without organic solvent (5 mM TBAOH, 10 mM n-octylamine, pH 4.0 adjusted with H3PO4) Table 2 Retention time of some anions (LiChroCart 125 mm×4.0 mm; 5 µm; buffer: 5 mM TBAPO4, 10 mM n-octylamine, 20 mM KH2PO4, pH 4.0 adjusted with H3PO4) Compound
tret (min)
ECD
UV 225nm
IO3 Uric acid Nitrate Oxalate Thiosulfate Iodide Thiocyanate
3.6 4.1 7.1 11.2 14.1 15.5 40.3
weak negative peak negative peak no yes yes yes negative peak
yes yes yes yes yes yes yes
The capacity factor in the range between 0 and 100 mmol/L NaCl is diminished from 21.5 to 6.1. In a solution of 20 mmol L–1 KH2PO4, the capacity factor of 25.3 is of the same order of magnitude as that of a solution containing 10 mmol L–1 NaCl (Fig. 2). As shown, the retention time of iodide in ion-pair chromatographic systems can be influenced by means of the pH value, the ion-pair concentration, the proportion of organic bases as buffer components, and the proportion of organic solvent, resulting in an optimized separation of the substrate components from the analytes. Additionally, the retention characteristics of different anions expected in the biological matrix were checked. ECD active species were also observed as well as anions which can only be recorded by means of UV detection (Table 2). Interference is to be expected in particular from thiocyanate, which invariably occurs in biological materials
As a rule, a silver electrode is used for the electrochemical detection of iodide. The basic assumption is that the formation of Ag(I) is the signal generating reaction. The main problem in the use of silver electrodes is the high oxidation sensitivity of silver, leading to rapid electrode surface coating, loss of sensitivity, and ultimate destruction of the electrode. In our experience, silver electrodes are unsuitable for use in routine analyses. The shape of the iodide signal peak is conspicuous in electrochemical detection processes (Fig. 4). The first half of the iodide peak has a Gaussian distribution, followed by a long tail-end descending below the baseline. Attempts to influence this tailing by changing the chromatographic conditions (pH value, eluent composition, flow, etc.) or the electrode material (gold, platinum, carbon) were unsuccessful. Tailing was most strongly pronounced when peak symmetry disturbances were already identifiable in the UV region. For this reason, the use of endcapped columns (Ultrasep ES Pharm RP 18 and Luna C 18(2)) proved to be superior to the use of conventional columns. The deviations in peak shape derived during EC detection compared with the symmetrical peaks obtained by UV detection (Fig. 4) can only be explained in terms of electrode reactions. Surface contaminants or secondary reactions involving oxidation/reduction products are conceivable. The use of normal voltammetric methods gives rise to a further problem with regard to the optimal working electrode potential. On the one hand it is clear that the analyte has to be electrochemically active at the chosen potential. The procedures most commonly described in the literature for determining the iodide concentration in biological materials define working potentials between –35 and +100 mV [13, 14, 15]. On the other hand, it is unclear from the cited articles which reference system and which electrode configuration is supposed to be used. In other words, it remains unclear which potential is valid for the
434 Fig. 3 Separation of some anions. LiChroCart (125 mm×4.0 mm; 5 µm) buffer/methanol = 90/10 (buffer: 20 mM TBAOH, 10 mM n-octylamine, 20 mM KH2PO4, 0.1 mol NaCl/L, pH 6.5 adjusted with H3PO4)
Fig. 4 Comparison of iodide-peak by DC (–35 mV) or UV (225 nm). LiChroCart (125 mm×4.0 mm; 5 µm) buffer/methanol = 90/10 (buffer: 20 mM TBAOH, 10 mM n-octylamine, 20 mM KH2PO4, 0.1 M NaCl/L, pH 6.5 adjusted with H3PO4, 1.5 mL/min)
working electrode. In contrast to the potential range given above, Lookabaugh et al. [9] pointed out that iodide is electrochemically active only at about +0.8 mV with reference to Ag/AgCl. This is fully consistent with our cyclic voltammetry results. No oxidation or reduction reaction is detectable in a cyclic voltammogram with a potential range between –600 and +600 mV vs. Ag/AgCl, and only the background current is measurable, as can be seen in Fig. 5. Extension of the potential range up to +1000 mV in the positive direction induces an oxidation of iodide with a reduction signal displayed at around +100 mV (Fig. 6). This confirms the fact that iodide is electrochemically active only at around +600 mV and that the product of this oxidation is reduced at a potential of around +100 mV.
The authors cited above [10, 14, 15] investigated the potential range between –150 and +100 mV, assuming that they had found maximum sensitivity in this range. We expanded the potential range under investigation from –600 to +700 mV. The results of our studies with hydrodynamic voltammetry (HPLC in conjunction with the ECD) undertaken in a potential range from –400 to +700 mV are shown in Fig. 7. Scarcely any changes in sensitivity are to be found in the potential range from 400 to –300 mV. A marked increase in sensitivity is recorded at potentials below –400 mV, increasing exponentially down to –600 mV. The sensitivity in the positive potential range increases steadily up to +600 mV, after which negative peaks are recorded at around 700 mV, the detector signal having re-
435
Fig. 5 Cyclic voltammogram in the range from –600 to +600 mV. 40 µg iodide/L; buffer/acetonitrile = 90/10 (buffer: 5 mM TBAOH, 10 mM triethylamine, 20 mM KH2PO4, pH 6.0 adjusted with H3PO4)
Fig. 7 Sensitivity of the iodide determination using hydrodynamic voltammetry. Luna RP 18(C2) (150 mm×4.6 mm; 5 µm) buffer/ acetonitrile = 90/10 (buffer: 20 mM TBAOH, 10 mM triethylamine, 20 mM KH2PO4, pH 6.0 adjusted with H3PO4), Analytical cell 5040 with Au-electrode (Bischoff)
Fig. 8 Time dependence of the calibration function. Luna RP 18 (C2) (150 mm×4.6 mm; 5 µm) buffer/acetonitrile = 90/10 (buffer: 5 mM TBAOH, 10 mM triethylamine, 20 mM KH2PO4, pH 6.5 adjusted with H3PO4) 1.4 mL min–1; Au-electrode; 0 mV) Fig. 6 Cyclic voltammogram in the range from –600 to +1000 mV. 40 µg iodide/L; buffer/acetonitrile = 90/10 (buffer: 5 mM TBAOH, 10 mM triethylamine, 20 mM KH2PO4, pH 6.0 adjusted with H3PO4)
versed its current direction. Oxidation reactions are now in progress. On the one hand, these findings make it clear that measurements with a detector potential at around 0 mV are relatively insensitive to iodide, but nonetheless selective. On the other hand, these results show a certain antithesis to the results of cyclic voltammetry. The main problem is that the potential of the ECD working electrode is less well defined than it was for cyclic voltammetric measurements. Measurement of the potential of the palladium reference system in the ECD, such that the said potentials are defined by our system has not yet been possible. It is still remains difficult to explain why iodide generates a signal in a potential range where it is not electrochemically active without prior oxidation. Further investigations will be necessary in this respect. A further problem of electrochemical detection is the time consistency of the signal generated. A decline in sen-
sitivity is invariably to be expected from the deposition of contaminants on the surface of the detector or oxidation of the electrode potential. A continuous decline in sensitivity is discerned even where oxidation-stable gold electrodes are used (Fig. 8). It is clear that continuous coating of the electrode surface occurs. Attempts to inhibit electrode contamination using pulse polarographic measurements were ineffective. A stable electrode condition could not be achieved when setting up the positive and negative cleaning potentials necessary for the above purpose. The reason is most probably the uncontrolled course of oxidation and reduction reactions. Shifts of the calibration function can only be compensated by insertion of a calibration standard after every 5 samples during the measuring process. The calibration function is traced back by iteration and every measurement evaluated on the basis of the recalculated calibration function. The limit of detection is 1 µg iodide/L (routine conditions, Table 3) and the limit of quantification 2 µg iodide/L. Under tuned conditions (freshly polished elec-
436 Table 3 Recovery from urine (standard method for routine analysis). Luna RP 18 (C2) (150 mm×4.6 mm 5 µm), buffer/acetonitrile = 90/10 (buffer: 5 mM TBAOH, 10 mM triethylamine, 59 mM
Measurements (n) Mean value (µg L–1) Standard deviation (µg L–1) Recovery (%) Standard deviation (%)
KH2PO4, 8 mM Na2HPO4×2H2O, pH 6.5 adjusted with H3PO4), 1.5 mL/min, injection volume 25 µL, Au electrode 0 mV, SPE as described under samples and sample pretreatment
Urine
Urine + 100 µg I–/L
Urine + 200 µg I–/L
Urine + 400 µg I–/L
10 80 3
10 179 6 99.2 3.3
10 275 9 98.2 3.2
10 463 29 96.5 6.2
3.7
trodes), a detection limit below 0.1 µg iodide/L is possible for a short time. The calibration curve is linear from 2 µg iodide/L to 600 µg iodide/L in a working range between 50 and 400 µg/L. As repetitive measurements with our standard routine method show (Table 3), ion-pair chromatography is entirely suitable for routine determinations of iodide in urine, provided the various influencing factors are taken into account.
Conclusion The chromatographic separation of iodide and matrix components can be optimized for the determination of iodide in biological materials by varying the pH value, the ion-pair concentration, the proportion of organic solvent, the amouts of organic bases as buffer components, and the salt content of the eluent system. Gold electrodes used for the electrochemical detection have proven superior to silver electrodes. If a highly sensitive iodide detection is required, and if the matrix components have been completely removed, detector potentials between –400 and –700 mV or above 300 mV should be chosen. A two-electrode arrangement is also favorable where an electrode with oxidative potential is positioned upstream and an electrode with reductive potential downstream. The investigations conducted provide a basis for iodide determination using ion-pair chromatography with elec-
trochemical detection in studies of iodide uptake associated with alimentary iodine deficiency, in the measurement of iodide resorption after clinical application of iodophors, and in the determination of iodide traces in sea water.
References 1. Hampel R, Kühlenberg T, Zöllner H, Klinke D, Klein K, Pichmann E-G, Kramer A (1996) Münch med Wschr 6:78–82 2. Wawschinek O, Eber O, Petek W, Wakonig P, Gürakar A (1985) Berichtet der OGKC 8:13–15 3. Zöllner H, Kramer A, Hampel R (1995) GIT Labor-Medizin 18:330–332 4. Lauber K (1975) Anal Chem 47:769–771 5. Haldimann M, Zimmerli B, Als C, Gerber H (1998) Clin Chem 44:817–824 6. Rendl J, Börner W (1995) Nuklearmediziner 18:286–290 7. Fechner P, Habernegg R (1999) Lebensmittelchemie 53:40–41 8. Small H, Stevens TS, Baumann WC (1975) Anal Chem 47: 1801–1809 9. DeKleijn JP (1982) Analyst 107:223–225 10. Buchberger W (1988) J Chromatogr 439:129–135 11. Hurst WJ, Snyder KP, Martin RA (1983) J Liq Chromatogr 6:2067–2077 12. Lookabaugh M, Krull IS, LaCourse WR (1987) J Chromatogr 387:301–312 13. Odink J, Bogaards JJP, Sandman H, Egger RJ, Arkesteyn GA, DeJong P (1988) J Chromatogr 431:309–316 14. Hurst JH, Stefovic JW, White WJ (1984) J Liq Chromatogr 7:2021–2030 15. Rendl J, Seybold S, Börner W (1994) Clin Chem 40:908–913 16. Weiß J (1991) (eds) Ionenchromatographie VCH
