Analytica Chimica Acta 556 (2006) 145–150

A highly stable and sensitive chemically modified screen-printed electrode for sulfide analysis Dong-Mung Tsai, Annamalai Senthil Kumar, Jyh-Myng Zen ∗ Department of Chemistry, National Chung Hsing University, 250 Kuo-Kuang Road, Taichung 40217, Taiwan Received 15 March 2005; received in revised form 15 May 2005; accepted 17 May 2005 Available online 5 July 2005

Abstract We report here a highly stable and sensitive chemically modified screen-printed carbon electrode (CMSPE) for sulfide analysis. The CMSPE was prepared by first ion-exchanging ferricyanide into a Tosflex anion-exchange polymer and then sealing with a tetraethyl orthosilicate sol–gel layer. The sol–gel overlayer coating was crucial to stabilize the electron mediator (i.e., Fe(CN)6 3− ) from leaching. The strong interaction between the oxy–hydroxy functional group of sol–gel and the hydrophilic sites of Tosflex makes the composite highly rigid to trap the ferricyanide mediator. An obvious electrocatalytic sulfide oxidation current signal at ∼0.20 V versus Ag/AgCl in pH 7 phosphate buffer solution was observed at the CMSPE. A linear calibration plot over a wide range of 0.1 ␮M to 1 mM with a slope of 5.6 nA/␮M was obtained by flow injection analysis. The detection limit (S/N = 3) was 8.9 nM (i.e., 25.6 ppt). Practical utility of the system was applied to the determination of sulfide trapped from cigarette smoke and sulfide content in hot spring water. © 2005 Elsevier B.V. All rights reserved. Keywords: Ferricyanide; Tosflex; Sol–gel; Screen-printed electrode; Sulfide

1. Introduction Sulfide toxicity is well known and the corresponding risk is associated with exposure in several occupational settings [1–6]. Continuous and high concentration exposure of sulfide can cause various physiological and biochemical problems. Therefore easy and fast detecting analytical tools are pressing need to effectively control the risk of sulfide-toxicities. Electrochemical detection assays using chemically modified electrodes (CMEs) were found to be advantageous for a wide range of practical assays in field analysis [7]. Electroanalysis can permit miniature assemblies/chips with disposable type of low-cost screen-printed electrodes (SPEs) [8]. Reliable measurements of dissolved sulfide in the environment require analysis immediately on sample collection because of the rapid oxidation rate of sulfide by dissolved oxygen. Vari∗ Corresponding author. Tel.: +886 4 22840411/506; fax: +886 4 22854007. E-mail address: [email protected] (J.-M. Zen).

0003-2670/$ – see front matter © 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.aca.2005.05.038

ous CMEs based on different modifiers, pre-oxidized nickel, palladium and copper electrodes were reported for the sulfide assays [9–18]. However, most of these systems are limited to practical extendibility and the common problems of stability and sensitivity are yet to be improved. We report here a highly stable and sensitive chemically modified screen-printed carbon electrode (CMSPE) for sulfide analysis. The CMSPE was prepared by first ion-exchanging ferricyanide into a Tosflex anion-exchange polymer (SPE/Ts-FeCN) and then sealing with a tetraethyl orthosilicate sol–gel layer (SPE/Ts-FeCN/SG). In general, sol–gel-based systems with efficient electrocatalysis, fouling protection, exclusion of interferants, are of great importance for the development of chemical sensors. Very recently, our group applied such a system for the mediated oxidation of ascorbic acid in neutral pH with high stability for repetitive measurements [19,20]. The sol–gel overlayer was essential to stabilize the Fe(CN)6 3− -Tosflex system without any fouloff and deterioration behavior in flow injection analysis. The chemical interaction between the hydrophilic sites of Tos-

146

D.-M. Tsai et al. / Analytica Chimica Acta 556 (2006) 145–150

flex and oxy- and/or hydroxy-surface functional groups of silicate was proposed to result in such a high stability. In this study, the proposed sensor was applied for the sensitive detection of sulfide down to the nano-molar range in physiological pH. Note that ferricyanide was reported as a good redox mediator useful for the oxidative sensing of sulfide and other mediators are mostly unstable in neutral pH [21–25]. Under the optimized experimental conditions, the CMSPE (i.e., SPE/TS-FeCN/SG) was successfully used for the determination of sulfide trapped from cigarette smoke and sulfide content in hot spring water.

2. Experimental 2.1. Chemicals and reagents Tosflex membrane (IE-SA 48) was purchased form Tosoh Soda (Japan). Tetraethyl orthosilicate (TEOS, 98%), ferricyanide and Na2 S were purchased from Aldrich. Disposable SPEs were bought from Zensor R&D (Taichung, Taiwan). All the other compounds were analytical reagent grade chemicals and used without further purification. Aqueous solutions were prepared with doubly distilled deionized water. Unless otherwise stated, a 0.1 M, pH 7 phosphate buffer solution (PBS) was used in all electrochemical measurements. 2.2. Apparatus Voltammetric measurements were carried out with a CHI 832A electrochemical workstation (Austin, TX, USA). The three-electrode system consists of one of the following working electrode: SPE, SPE/Ts-FeCN and SPE/Ts-FeCN/SG, an Ag/AgCl reference electrode (Model RE-5, BAS) and a platinum auxiliary electrode. The FIA system consisted of a solvent delivery system with microprocessor pump drive (BAS PM-80), a Rheodyne Model 7125 sample injection loop (20 ␮l) with an interconnecting Teflon tube and a BAS wall-jet electrochemical detector. The assembling of working electrodes in FIA is similar to our previous procedure [20]. In FIA, the SPE/Ts-FeCN/SG was equilibrated in PBS at 0.3 V versus Ag/AgCl under a hydrodynamic flow rate (Hf ) of 500 ␮l/min until the current became constant. It normally took less than 2 min. The FIA oxidation peak current signals were taken as the quantitative parameter.

solution was collected. The concentration of the dissolved polymer was determined gravimetrically from an evaporated portion of the solution. The optimal coating solution was brought to a final concentration of 1.3 wt.% by dilution with methanol. As to the preparation of SPE/Ts, 5 ␮l Tosflex (1 wt.%) was spin-coated onto SPE at a spin rate of 3600 rpm for 3 min. The Fe(CN)6 3− mediator can then be doped into the SPE/Ts to prepare the SPE/Ts-FeCN by first dipping in 1 mM Fe(CN)6 3− for 10 min followed by gentle washing with double distilled water. The electrodes were then continuously scanning by cyclic voltammetry until the voltammograms became stable in pH 7 PBS solution. Normally it took 3–5 min. Finally, 7 ␮l of sol–gel composition consisting of 1.5 ml TEOS + 0.33 ml H2 O + 25 ␮l 1 M HCl was spin-coated on the SPE/Ts-FeCN at 3600 rpm for ∼1 min to prepare the SPE/Ts-FeCN/SG for sulfide analysis. Prior to the experiments, the freshly prepared sol–gel solutions were sonicated for 5 min and then left overnight at room temperature. Such a mixture can be used up to 4 days from the preparation. 2.4. Real samples A famous brand of manufactured cigarette was purchased from a local market. The cigarette smoke solution was collected using in 0.1 M NaOH solution (more facile medium to trap gaseous sulfides) containing gas-washing bottle as illustrated in Scheme 1. The setup consists of a 100 ml Quickfit B¨uchner flask with a 19/26 ground neck and a bubbler (0.8 cm diameter) ending with a micron filter reaching the bottom of the flask. The upper part of the bubbler was attached to a poly(vinyl chloride) tube and a 60 ml aliquot of deionized water was added in the flask. Sulfur-containing hot spring water was collected from Yangmingshan National Park (Taipei, Taiwan) and thoroughly filtered (Millipore 0.2 ␮M white Nylon diameter 47 mm) prior to routine analysis. Two hot spring water samples obtained from different sources possess different color of yellow (pH ∼4–5) and green (pH ∼2), respectively.

2.3. Electrode preparation The preparation of Tosflex and sol–gel solutions generally followed our previous reports [19,20]. In brief, 2.5 g of Tosflex membrane was finely cut and heated to boiling in 10 ml of water/methanol/2-propanol aqueous/alcoholic solution for ∼20–50 h. After cooling, the undissolved membrane was separated by centrifugation and a clear, yellowish

Scheme 1. Schematic view of the setup used for collecting cigarette smoke.

D.-M. Tsai et al. / Analytica Chimica Acta 556 (2006) 145–150

147

Fig. 1. Cyclic voltammetric response of the CMSPE (i.e., SPE/TsFeCN/SG) in the presence/absence of 1 mM sulfide at a scan rate of 100 mV/s in pH 7 PBS. The dashed line represents the CV response of SPE in the presence of 1 mM sulfide. Insert figure is the plot of anodic peak current (ipa ) against sulfide at the CMSPE.

3. Results and discussion 3.1. Electrocatalytic mediation of sulfide As shown in Fig. 1, a defined peak corresponding to the encapsulated Fe(CN)6 3−/4− redox couple at a potential of ∼0.15 V was observed at the SPE/Ts-FeCN/SG by cyclic voltammetry in pH 7 PBS. Since sulfide ion can exist in various forms of H2 S, HS− and S2− at different pH solutions (pKa1 = 6.96 and pKa2 = 11), it is the HS− species that actively involve in the mechanism in neutral pH [9]. In the presence of sulfide, the anodic peak was found to increase considerably accompanied with a decrease in the cathodic peak current. This indicates that the sulfide oxidation reaction is mediated through the Fe(CN)6 3−/4− redox couple on the SPE/Ts-FeCN/SG surface. In comparison, a bare SPE was also subjected for the sulfide oxidation under similar experimental conditions. Instead of the appearance of a peak like behavior through mediation, direct oxidation of sulfide was observed at a potential of ∼400 mV more positive than that at the SPE/Ts-FeCN/SG. The lowering in overpotential for the sulfide oxidation at the SPE/Ts-FeCN/SG is advantageous since it can not only provide a better sensitivity but also avoid the interference of other easily oxidizable compounds. For example, Cl− , Br− and SO3 2− get oxidized with/without the mediator on the carbon surface at a high positive potential of ∼1.0 V versus Ag/AgCl in neutral pH. By choosing a mediator like Fe(CN)6 3− can allow to lower the working potential and hence avoid the oxidation of these coexisting compounds. Salimi et al. reported a sol–gel derived carbon ceramic electrode fabricated by [Ru(bpy)(tpy)Cl]PF6 (bpy = bipyridyl and tpy = terpyridyl) mediator for the sulfide detection at 0.8 V versus Ag/AgCl in pH 7 PBS [26]. Chen reported a water-soluble cobalt tetrakis (N-methyl-2pyridyl)porphyrin sulfide sensor at 0.6 V versus Ag/AgCl in

Scheme 2. (A) Conceptional representation of the CMSPE. (B) Proposed mechanism for the sulfide oxidation reaction. Sulfide exists mainly as the HS− form in pH 7 (pKa1 = 6.96).

pH 2 solution [27]. Unfortunately, both systems show a profound catalytic activity to SO3 2− and are thus not selective for the sulfide detection. The SPE/Ts-FeCN/SG shows a selective oxidation signal without any SO3 2− interference. Further CV studies show that the anodic peak current (ipa ) signals for the sulfide oxidation is linear up to 100 ␮M sulfide; after that it tends to reach a constant. This behavior is similar to our earlier studies on ascorbic acid oxidation using a similar system [19,20]. The sulfide oxidation is related to the surface saturation kinetics with a Michaleis–Menton type of reaction pathway [28]. Scheme 2B depicts a possible reaction mechanism for the sulfide oxidation at the SPE/Ts-FeCN/SG. In the first step, the active Fe(CN)6 3− directly oxidize sulfide. The reduced form of the mediator can then be re-oxidized to the active state at an applied potential of 0.2 V. This is the typical example for EC type of coupled chemical reaction [29,30]. The surface coverage of Fe(CN)6 3− (Γ FeCN ) with/without sol–gel overlayer were measured as 8.3 × 10−10 and 3.8 × 10−10 moles/cm2 , respectively. The observation denotes multiple Fe(CN)6 3− monolayers on the surface with a considerably thick-film CME layer characteristics [20]. The Γ FeCN difference is due to the monolayer blocking effect of the sol–gel overlayer coating. Since the sol–gel overlayer was porous in nature, the reactant/product can diffuse freely on the outer surface of the SPE/Ts-FeCN/SG as illustrated in Scheme 2B. Plots of the catalytic current are linearly increased with the square root of the scan rate, v1/2 (data not shown) suggesting the reaction is diffusion-controlled with the diffusion of sulfide as the rate-determining step (rds) for the overall oxidation reaction [29,30]. Meanwhile, a plot of anodic peak current against v1/2 also shows a good linearity starting from zero in the absence of sulfide. This indicates

148

D.-M. Tsai et al. / Analytica Chimica Acta 556 (2006) 145–150

that the charge transport pathway of the doped Fe(CN)6 3− is due to electron hoping mechanism similar to the behavior in ionomer and redox polymer [31]. The fact that there is no marked alteration in the peak current or peak potential upon continuous scans further indicates that the electrode is free from any surface poisoning effect. Note that Pd, Cu, Ag and Au-based systems were reported to be highly unstable due to surface foul-off behavior with sulfide [17–18,32,33]. The free from foul behavior of the SPE/Ts-FeCN/SG is a clear advantage for sulfide detection assay. 3.2. Flow injection analysis of sulfide Experiments regarding FIA of sulfide at the SPE/TsFeCN/SG were further studied. Compared to the results of SPE and SPE/Ts-FeCN, the advantage of the SPE/TsFeCN/SG in FIA is clearly demonstrated in Fig. 2. As can be seen, the SPE shows a very feeble signal; the SPE/TsFeCN, on the other hand, starts with a huge peak but continuously decreases with time. The signal even decays to the level of bare SPE after ∼5000 s of operation time. Presumably the active Fe(CN)6 3− site was completely leaching out from the working surface under the hydrodynamic working condition. The SPE/Ts-FeCN/SG, however, is highly stable without any decrease or alternation in the FIA signals. The sol–gel overlayer coating was thus essential to stabilize the Fe(CN)6 3− mediator from leaching. It is believed that the oxy–hydroxy functional group of sol–gel strongly interacts with the hydrophilic sites of Tosflex (Scheme 2A) and makes the composite highly rigid to trap the Fe(CN)6 3−

Fig. 2. FIA of 50 ␮M sulfide using SPE, SPE/Ts-FeCN and SPE/TsFeCN/SG in pH 7 PBS. Eapp = 0.3 V. Hf = 0.2 ml/min.

Fig. 3. Effects of (A) applied potential (Eapp ) and (B) hydrodynamic flow rate (Hf ) on the detection of 50 ␮M sulfide by FIA at the SPE/Ts-FeCN/SG using pH 7 PBS as carrier solution.

mediator [20]. Since Fe(CN)6 3− can be blocked from crossing anionic interfaces, it is more likely that the oxy–hydroxy functional groups of the sol–gel can electrostatically repel the Fe(CN)6 3− and hence prevents it from leaching out.

Fig. 4. Typical calibration plot in FIA at the SPE/Ts-FeCN/SG using pH 7 PBS as carrier solution. Insert figure is the results of continuous injection (n = 6) of 0.1 ␮M sulfide in FIA. Other conditions were the same as in Fig. 2.

D.-M. Tsai et al. / Analytica Chimica Acta 556 (2006) 145–150

149

Table 1 Interference effect on the detection of 1 ␮M sulfide at the SPE/Ts-FeCN/SG in pH 7 PBS carrier solution by FIA Interferant 2−

SO3 SO4 2− CO3 2− C2 O4 2− F− , Br− , Cl− NH4 + Na+ K+ Mg2+ Ca2+ a

[Interferant]:[sulfide]

Actual ipa (nA)a

10:1 100:1 100:1 100:1 100:1 1:1 100:1 100:1 100:1 100:1

38.19 38.43 40.99 39.15 43.01 40.18 38.43 43.01 38.31 41.26

Signal alteration (%) −5.94 −5.34 +0.97 −3.57 +5.94 −1.03 −5.34 +5.94 −5.64 +1.62

ipa = 40.60 nA without interfering analyte.

The FIA conditions were further optimized at the SPE/TsFeCN/SG using a pH 7 PBS carrier solution. As can be seen in Fig. 3, the FIA signals regularly increase as the applied potential (Eapp ) increases positively up to 0.3 V and start to decrease as Eapp > 0.3 V. The behavior is in accordance with the electrochemical characteristics of the Fe(CN)6 3−/4− redox couple. By taking the optimized Eapp of 0.3 V, ∼10 times higher in the current signal was observed at the SPE/Ts-FeCN/SG than that of SPE. As to the effect of hydrodynamic flow rate (Hf ) in FIA, a peak-like trend to the current response with a maximum at 0.2 ml/min was observed (Fig. 3B). Presumably the optimal Hf was favorable for the diffusion controlled mechanism and was thus used in subsequent experiments. Under the optimal conditions, a linear calibration plot in the window of 0.1 ␮M to 1 mM with a current sensitivity and regression coefficient of 5.60 nA/␮M and 0.9989, respectively, was obtained. Five continuous injections of 0.1 ␮M sulfide showed a R.S.D. value of 3.12% indicating a detection limit (S/N = 3) of 8.9 nM for 20 ␮l injection (i.e., 25.6 ppt) at the SPE/Ts-FeCN/SG. The DL value is at least three

orders lower than the recent reported 0.8 ␮M at pre-oxidized nickel electrode, 0.3 ␮M at Pd-coated glassy carbon electrode and 5 ␮M at hexadecylpyridium bis(chloranilato)-antimonyV modified SPE [15–17]. The low DL and R.S.D. values are clear advantage of applying the present systems to practical analysis (Fig. 4). 3.3. Analytical performance Interference effect was checked for variety anionic and cationic species in the presence of 1 ␮M sodium sulfide. As shown in Table 1, acceptable tolerance of ∼5% variation can be obtained according to the relative responses for a 100fold higher concentration of SO4 2− , CO3 2− , C2 O4 2− , F− , Br− , Cl− , Na+ , K+ , Mg2+ , Ca2+ , a 10-fold higher concentration of SO3 2− and an equal concentration of NH4 + compared to the response shown with pure solution of 1 ␮M sulfide taken as 100%. The major interferences in cigarettes smoke as reported earlier are alkanes and gas (such as ethane, propane, CO2 , CO) [34]. The alkanes are mostly electro-inactive; as

Fig. 5. FIA of sulfide in real samples of (A) cigarette smoke and (B) green sulfur hot-spring water at the SPE/Ts-FeCN/SG. The dilution factor is 5 and 10 for (A) and (B), respectively. Other conditions were the same as in Fig. 2. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of the article.)

150

D.-M. Tsai et al. / Analytica Chimica Acta 556 (2006) 145–150

for gas, the influence might be only in the change in pH. Since we sucked cigarette smoke into 0.1 M NaOH solution, the interference does therefore not exist. Interferences from other anions (Br− , HCO3 − , NO2 − , CH3 COO− , BrO3 − ) were studied. Reponses with 10× excess of the mentioned anions in a concentration of 1 ␮M sulfide was found to be almost negligible for the proposed system. The common sulfide corrosion problem with metallic electrodes also did not occur in this approach [35]. Overall, the SPE/Ts-FeCN/SG showed a remarkable selectivity towards sulfide. Finally the procedure is extended to real sample analysis by detecting the sulfide trapped from cigarette smoke and sulfide content in hot spring water using standard addition method. The determination of toxic components in cigarette smoke is important because cigarette smoking causes human respiratory diseases not only to the smokers but also to nonsmokers exposed to the cigarette smoke. Since low amounts of sulfide are enough to cause physiological distress, highly sensitive and rapid methods for its detection in cigarette smoke is important. As shown in Fig. 5, the detected value is 18.35 ␮M and the result is close to the measured value of 19.16 ± 1.56 ␮M by both the standard methylene blue reaction (i.e., Caro’s method) and cinder-tetracyano nickel based electrochemical methods from our previous report for the same brand of cigarette [36]. Also shown in Fig. 5, the detected values are 12.35 and 32.82 ␮M, respectively, for the two hot spring waters. Good recoveries in the window of 98.6–100.6% were achieved for all three real samples. These results clearly indicate the practical utility of the present approach.

4. Conclusion The SPE/Ts-FeCN/SG was found to be quite efficient for the sulfide oxidation in neutral pH without any foul-off behavior. The electrocatalytic behavior was explained in terms of the mediation through Fe(CN)6 3−/4− redox couple with the HS− species. Such performance characteristics along with high stability compare favorably to those of existing methods for detecting sulfide [6]. Extended practical assay for the cigarette smoke and hot spring water samples gave promising results and further offer application into more real samples.

Acknowledgment The authors gratefully acknowledge financial support from the National Science Council of Taiwan.

References [1] R.P. Pohanish, S.A. Greene, Hazardous Materials Handbook, Van Nostrandeinhold, New York, 1996. [2] P.A. Patnaik, Comprehensive Guide to the Hazardous Properties of Chemical Substances, 2nd ed., Wiley, New York, 1999. [3] R.G. Hendrickson, A. Chang, R.J. Hamilton, Am. J. Ind. Med. 45 (2004) 346. [4] S. Gao, K.K. Tanji, S.C. Scardaci, Agron. J. 96 (2004) 70. [5] Y. Maki, J. Gen. Appl. Microb. 33 (1987) 391. [6] N.S. Lawrence, J. Davis, R.G. Compton, Talanta 52 (2000) 771. [7] J.-M. Zen, A.S. Kumar, D.-M. Tsai, Electroanalysis 15 (2003) 1073. [8] J. Wang, Acc. Chem. Res. 35 (2002) 811. [9] J.P. Hart, A.K. Abass, Anal. Chim. Acta 342 (1997) 199. [10] J.-M. Zen, P.-Y. Chen, A.S. Kumar, Electroanalysis 14 (2002) 513. [11] M. Thompson, N.S. Lawrence, J. Davis, L. Jiang, T.G.J. Jones, R.G. Compton, Sens. Actuators B 87 (2002) 33. [12] M. Pandurangappa, N.S. Lawrence, L. Jiang, T.G. Jones, R.G. Compton, Analyst 128 (2003) 473. [13] N.S. Lawrence, L. Jiang, T.G.J. Jones, R.G. Compton, Anal. Chem. 75 (2003) 2054. [14] Y.-H. Tse, P. Janda, H. Lam, A.B.P. Lever, Anal. Chem. 67 (1995) 981. [15] M.I. Prodromidis, P.G. Veltsistas, M.I. Karayannis, Anal. Chem. 72 (2000) 3995. [16] D. Giovanelli, N.S. Lawrence, L. Jiang, T.G.J. Jones, R.G. Compton, Analyst 128 (2003) 173. [17] I.G. Casella, M.R. Guascito, E. Desimoni, Anal. Chim. Acta 409 (2000) 27. [18] I.G. Casella, M. Contursi, E. Desimoni, Analyst 127 (2002) 647. [19] J.-M. Zen, D.-M. Tsai, A.S. Kumar, V. Dharuman, Electrochem. Commun. 2 (2000) 782. [20] J.-M. Zen, D.-M. Tsai, A.S. Kumar, Electroanalysis 15 (2003) 1171. [21] P. Jeroschewski, K. Haase, A. Trommer, P. Gr¨undlet, Electroanalysis 6 (1994) 769. [22] P. Jeroschewski, S. Braun, Fresenius, J. Anal. Chem. 354 (1996) 169. [23] P. Jeroschewski, C. Steuchart, M. K¨uhl, Anal. Chem. 68 (1996) 4531. [24] N.S. Lawrence, M. Thompson, C. Prado, L. Jiang, T.M.G. Jones, R.G. Compton, Electroanalysis 12 (2002) 499. [25] M.R. Milani, N.R. Stradiotto, A.A. Cardoso, Electroanalysis 15 (2003) 827. [26] A. Salimi, S. Pourbeyram, M.K. Amini, Analyst 127 (2002) 1649. [27] S.-M. Chen, J. Electroanal. Chem. 432 (1997) 101. [28] A.S. Kumar, J.-M. Zen, Electroanalysis 14 (2002) 671. [29] A.J. Bard, L.R. Faulkner, Electrochemical Methods, John Wiley, New York, 2001. [30] F. Scholz (Ed.), Electroanalytical Methods, Springer, Berlin, 2002. [31] M. Majda, in: R.W. Murray (Ed.), Molecular Design of Electrode surface, Wiley, New York, 1992, p. 159. [32] R.L. Sanchez, I. Gonzalez, G.T. Lapidus, J. Appl. Electrochem. 32 (2002) 1157. [33] M. Alanyalioglu, H. Cakal, A.E. Oeztuerk, U. Demir, J. Phys. Chem. B 105 (2001) 10588. [34] P.A. Patnaik, Comprehensive Guide to the Hazardous Properties of Chemical Substances, 2nd ed., Wiley, New York, 1999. [35] F. Hissner, J. Mattusch, K. Heinig, J. Chromatogr. A 848 (1999) 503. [36] J.-M. Zen, R.-L. Chang, P.-Y. Chen, R. Ohara, K.-C. Pan, Electroanalysis 17 (2005) 739.