- Email: [email protected]
Highly selective electrogenerated chemiluminescence (ECL) for sulfide ion determination at multi-wall carbon nanotubes-modified graphite electrode
Analytica Chimica Acta 582 (2007) 267–274
Highly selective electrogenerated chemiluminescence (ECL) for sulfide ion determination at multi-wall carbon nanotubes-modified graphite electrode Rongfu Huang, Xingwang Zheng ∗ , Yingjuan Qu School of Chemistry and Material Science, Shaanxi Normal University, Xi’an 710062, PR China Received 25 July 2006; received in revised form 17 September 2006; accepted 19 September 2006 Available online 26 September 2006
Abstract In the present work, a novel method for immobilization of carbon nanotubes (CNTs) on the surface of graphite electrode was proposed. We further found that superoxide ion was electrogenerated on this CNTs-modified electrode, which can react with sulfide ion combing with a weak but fast electrogenerated chemiluminescence (ECL) emission, and this weak ECL signal could be enhanced by the oxidative products of rhodamine B. In addition, the rate constant of this electrochemical reaction k0 was investigated and confirmed that the speed of electrogenerating superoxide ion was in accordance with the subsequent fast CL reaction. Thus, the fast CL reaction of superoxide ion with target brought in the possibility of high selectivity based on time-resolved, relative to other interferences. Based on these findings, an excellently selective and highly sensitive ECL method for sulfide ion was developed. Under the optimum conditions, the enhancing ECL signals were linear with the sulfide ion concentration in the range from 6.0 × 10−10 to 1.0 × 10−8 mol L−1 , and a 2.0 × 10−10 mol L−1 detection limits (3σ) was achieved. In addition, the proposed method was successfully used to detect sulfide ion in environmental water samples. © 2006 Elsevier B.V. All rights reserved. Keywords: Electrogenerating superoxide ion; Chemiluminescence; Multi-wall carbon nanotubes; Sulfide ion determination; Modified graphite electrode
1. Introduction The detection of sulfide ion has been paid much attention as a consequence of the toxicity of hydrogen sulfide and the corresponding risk associated with exposure in a number of occupational settings [1–3]. Thus, the determination of sulfide ion was particularly important from industrial, environmental, and biological point of view. For this purpose, many analytical schemes, such as titrimetric [4], spectrometric methods [5], electroanalytical methods with chemically modified electrode technique [6], polarographic [7], ion chromatographic [8], catalytic kinetics methods [9,10], and fluorescence [11], chemiluminescence (CL) [12–14] methods, have been developed. Among these methods reported for sulfide ion, the CL methods had been paid considerable attention since this analytical technique promised high sensitivity, wide linear range
∗
Corresponding author. Tel.: +86 29 8530 8184. E-mail address: [email protected] (X. Zheng).
0003-2670/$ – see front matter © 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.aca.2006.09.035
and requirement of simple instrumentation [15]. Although these methods presented straightforward approach to high sensitivity, some powerful or unstable oxidants used in these systems led to poor selectivity. Thus, a highly selective combined with excellently sensitive CL method for the determination of sulfide ion is strongly desirable. As reviewing the powerful CL systems, it was obviously found that most of the useful CL reactions were relatively fast, which led to better selectivity based on the CL speed-resolution power. For example, the CL intensity of hypobromite with urea or ammonia reached a maximum within 20 ms after mixing and no CL signal was detectable after 120 ms. Based on this advantage, two highly selective CL methods for urea or the ammonia were developed [16,17]. Thus, to obtain high CL reaction speed, the powerful CL reaction systems often concerned the addition of some catalysts, the use of instable reagents, and so on. Among those instable reagents used to speed up the CL reaction, such as OH− , BrO− , ClO− , Mn3+ , Co3+ , etc., the superoxide ion was paid much more attention [18,19]. The
268
R. Huang et al. / Analytica Chimica Acta 582 (2007) 267–274
main reason was as followings: firstly, superoxide ion, relative to hydrogen peroxide, presented not only the faster CL reaction speed, but also the higher oxidizing potential from the thermodynamic and kinetic points; secondly, due to the much lower absorption ability in UV–vis region, it nearly did not present the inner filter effect for CL emitter when it used as the CL reaction oxidant. In addition, no interference species was brought in by its reductive product, H2 O. Consequently, some better CL analytical performances, such as the excellent sensitivity and high selectivity, were obtained. Up to now, although some schemes such as biochemical reaction and photochemical reaction, etc. [20,21], have been developed to produce superoxide ion and explored its CL analytical applications, these two methods suffered from poor selectivity due to the seriously chemical interferences from the relating precursors which were used to produce superoxide ion. Compared with these methods, electrochemical reduction dioxygen was paid more attention [22,23]. However, the reduction of dioxygen to superoxide ion depends on the reaction medium and electrode material. For instance, a reversible, one-electron reduction occurs to yield the superoxide ion in aprotic media; Taylor and Humffray [24] suggested that dioxygen was reduced to superoxide ion on glass-carbon electrode first and Appleby and Marie [25] concluded the mechanism of dioxygen reduction on carbon black electrode in alkaline solution, in which superoxide ion was produced first and further oxidated to H2 O2 . On the other hand, carbon nanotubes (CNTs), both singlewall and multi-wall CNTs, have attracted increasing attention because of its remarkable nanostructures, since its discovery by Iijima in 1991 [26]. One promising application is based on the promotion ability of electron-transfer reactions of target molecules when used as an electrode material in electrochemical reactions. In addition, it has been also applied to dioxygen reduction [27,28]. For instance, the multi-wall carbon nanotubes (MWNTs) and cobalt prophyrin-modified glass-carbon (GC) electrode exhibited electrocatalytic activity to the reduction of oxygen; Hu et al. have investigated the electrochemical reduction of dioxygen on the single-wall carbon nanotubes-dihexadecyl phosphoate (SWNTs-DHP) film electrode and the possible mechanism of the reduction of dioxygen was proposed. But up to now, no work was developed to study the CL performances with the superoxide ion electrogenerated on MWNTs-modified electrode for analytical purpose. In this paper, it was found that the instable CL oxidant, superoxide ion, could be rapidly generated at the surface of multiwall carbon nanotubes (MWNTs)-modified graphite electrode and further reacted selectively with sulfide ion, due to speedresolution, producing a weak but fast ECL signal. At the same time, it was also found that this ECL signal could be strongly enhanced in the presence of the oxidative products of rhodamine B (OPRB) based on an energy transfer CL mechanism. Based on these observations, a highly selective and excellently sensitive ECL method for sulfide ion detection at MWNTs-modified graphite electrode was developed. In addition, the proposed ECL method was successfully applied to the determination of sulfide ion in environmental water samples.
2. Experimental 2.1. Reagents All the reagents were of analytical-reagent grade and were directly used for following experiments without purification. Cetyltrimethylammonium bromide (CTAB), NaOH, paraffin, rhodamine B, were purchased from Xi’an & Shanghai Chem. Ind. Co. MWNTs, were purchased from Shenzhen Nanotech Port Co. Ltd. The pretreatment method was showed in next part this paper. Hydrogen peroxide 30% (v/v) was diluted with doubledistilled water daily. Solutions of rhodamine B (1.0 × 10−3 mol L−1 ) were prepared by accurately weighing and diluting to appropriate concentration. Stock sulfide ion solutions were prepared by dissolving appropriate amount of Na2 S·9H2 O in water, which were standardized using iodine solution and sodium thiosulfate solutions. The instability of very dilute sulfide ion solution was well known. For this reason, the calibration solutions were prepared serially from the solution of next highest concentration and analyzed immediately. 2.2. Apparatus The applied potential for different electrolytic ways such as linear sweep, multi-potential step and cyclic voltammeter were performed with a conventional three-electrode cell linked to a CHI660b Electrochemistry Working Station (CH Instruments, Inc.). The electrolytic cell utilized a conventional three-electrode setup and was arranged as shown in Fig. 1. The cell was made of a microbeaker (high: 35 mm, i.d.: 25 mm). The working electrode was the MWNTs-modified graphite electrode (Shanghai, 12.6-mm2 surface area); a Pt flake (7 mm × 7 mm) and Ag/AgCl (saturated KCl solution) were used as the auxiliary electrode and pseudo-reference electrode, respectively. The ECL intensity was transformed into an electrical signal by an R456 photomultiplier (PMT) (Xi’an Remax Electronic Science Tech. Co. Ltd., Xi’an, China), which was operated at −800 V and the ECL cell was placed in front of the PMT.
Fig. 1. The block diagram of ECL detection system. W, working electrode; R, Ag/AgCl reference electrode; C, counter electrode (Pt); L, KNO3 salt bridge; PMT, photomultiplier; NHV, negative high voltage supply.
R. Huang et al. / Analytica Chimica Acta 582 (2007) 267–274
2.3. Preparation of MWNTs-modified graphite electrode Multi-wall carbon nanotubes (MWNTs) were sonicated in concentrated nitric acid at 25 ◦ C for about 24 h, then filtered and washed with double-distilled water several times for further purification until the filtrate became neutral and dried in an oven at 37 ◦ C. The graphite electrode was carefully polished on rough and fine sandpapers successively, and the electrode was then subjected to be sonicated to remove adsorbed particles and rinsed with double-distilled water several times. About 1.0 mg nitric acid-treated MWNTs and some amount of CTAB were dispersed together in 10 mL liquid paraffin to give a black suspension with the aid of ultrasonic agitation for about 1 h. Subsequently, the above electrode was immersed in and sonicated for a few minutes. A stable and uniform MWNTsmodified graphite electrode was prepared. 2.4. The analytical procedure When adding 1.0 × 10−7 mol L−1 hydrogen peroxide, 5.0 × 10−8 mol L−1 rhodamine B to 0.1 mol L−1 NaOH solution, the difference between UV–vis spectra of rhodamine B and the mixture solution indicated that rhodamine B was oxidized by hydrogen peroxide. 5.0 mL blank solution, the above oxidative products of rhodamine B solution (OPRB), was added into the ECL cell and a stable blank signal was recorded when the electrolytic potential was applied to the working electrode; the sample or standard sulfide ion solution containing not only the oxidative products of rhodamine B, but also an appropriate concentration of sulfide ion was added to the ECL cell, and the ECL signal was recorded. The concentration of sulfide ion was quantified via the peak height of the relative ECL emission intensity which was obtained by subtracting the blank ECL emission intensity from that of the sample or standard sulfide ion solution.
269
According to the literatures, Comptom and co-workers [34] found CNTs’ effective electrocatalytical behavior on basal plane pyrolytic graphite electrodes, immobilized by abrasive attachment and film modification (dissolved in acetone); Cai and Chen [35] directly dispersed the CNTs in a CTAB solution and fabricated a CNT-modified electrode which has promotion effects on the direct electron transfer of glucose oxidase. In our work, we found that the nitric acid-treated MWNTs (carboxylic CNTs) could form special ion complex (CTAB–CNTs) with cationic surfactant CTAB through electrostatic interaction. Meanwhile, this ion complex CTAB–CNTs can be well dispersed into paraffin by ultrasonication for a few minutes. We further found that this suspension was very stable after 3 weeks. In addition, it was well-know that liquid paraffin can well permeate into the graphite layer or the pore on the surface of the bare graphite electrode and form a very stable paraffin dipping film on the graphite electrode, due to the hydrophobic nature of graphite plane. Thus, while the bare graphite electrode was immersed into the paraffin, which contained the CTAB–CNTs complex, for a given time, the CTAB–CNTs was more stable modified to graphite electrode surface. Compared with Cai’s work [35], not only MWNTs were effectively immobilized, but also it provided a hydrophobic nature on the surface of graphite electrode. Furthermore, the electrochemical behavior of the MWNTsmodified graphite electrode was investigated. Fig. 2 showed the electrochemical response of K3 Fe(CN)6 at different work electrodes: the cyclic voltammogram of K3 Fe(CN)6 on liquid paraffin with CTAB dipping graphite electrode (curve (a)) is poor; the potential difference (Ep ) between the oxidation and the reduction peaks is 256 mV and its stability is unsatisfactory. However, the voltammetric response is apparently improved at the MWNTs-modified graphite electrode, reflected by the enlargement of the peak currents (Ip ) and the decline of the potential difference (Ep = 63 mV), with a pair of excellent symmetrical redox peaks (curve (b)). And we found the peak potential and currents remained stable after a few cycles.
3. Results and discussion 3.1. The immobilization of the MWNTs on the surface of graphite electrode Though the carbon nanotubes-modified electrodes, in comparison with other forms carbon electrode, showed better performance due to its dimensions, the electronic structure, and the topological defects present on the tube surface, a major barrier for developing such CNTs-modified electrode is the insolubility of CNTs in most solvents [29]. However, many efforts have been made to disperse CNTs into suitable solvents, such as dimethylformamide (DMF) [30], acetone [31], concentrated sulfuric acid [32] and perfluorosulfonated polymer Nafion [33], and then, this suspension was cast on a substrate electrode to form a carbon nanotube film. Even now, these CNTs casting films on the electrode surface often presented the poor mechanical stability and resulted in the poor reproducibility of the electrochemical response. In this case, the new immobilization schemes for CNTs are highly desired to broaden the application of CNTs-modified electrode.
Fig. 2. Cyclic voltammograms at different work electrode in 1 mmol L−1 K3 Fe(CN)6 solution: (a) liquid paraffin with CTAB dipping graphite electrode; (b) MWNTs-modified graphite electrode.
270
R. Huang et al. / Analytica Chimica Acta 582 (2007) 267–274
The reason for the better performance of the MWNTs-modified electrode might arise from the promotion of electron-transfer reaction on the CNTs surfaces. Meanwhile the CNTs increased the effective area of the electrode, so the peak current increased significantly. 3.2. Electrogenerating superoxide ion at MWNTs-modified graphite electrode The cyclic voltammograms of dioxygen in 0.1 mol L−1 airsaturated NaOH solution at MWNTs electrode is shown in Fig. 3a and b (scan rate: 0.1 and 0.01 V s−1 ) when the potential initially sweeps from 0.2 to −1.2 V. According to the cyclic voltammograms, it is obvious that two well-defined dioxygen reduction peaks are observed at about −0.3 and −0.9 V, being consistent with similar observations by others, in which the dioxygen was firstly reduced to hydrogen peroxide [28]. In addition, when superoxide dismutase (SOD) was added into the reaction system, the first reduction peak decreased greatly. Based these observation, it was inferred that superoxide ion can be produced in the first step as an intermediate and further reduced to H2 O2 [25]. On the other hand, the second reduction peak is still visible at low scan rate in 0.1 mol L−1 air-saturated NaOH solution, indicating that the first reductive product is quite stable for the subsequent further reduction, which may result from the hydrophobic micro-environment of the waxed graphite electrode surface (Fig. 3b). Correspondingly, the bare graphite electrode shows quite contrary effect to the reduction of dioxygen (Fig. 3c) combined with a low background current and a flat, broad reduction peak. Compared with two peaks at MWNTs-modified graphite electrode, the separation of two peaks is due to the difference of rate constant of two steps. Then, the two peaks do not separate [36].
Fig. 4. Plot of Ep vs. log v. The influence of scan rate on the first reduction-peak potential of dioxygen in air-saturated 0.1 mol L−1 NaOH solution.
In addition, the heterogeneous rate constant of the first step of dioxygen reduction was investigated at MWNTs-modified graphite electrode and it was approximately considered as the rate constant of electrogenerated superoxide ion. According to Fig. 3, the electrochemical behavior of dioxygen was irreversible. As an irreversible reaction, the peak potential of dioxygen shift negatively with the increase of scan rate (v) and have a linear relationship with the natural logarithm of scan rate [36]. The first reduction-peak potential changes according to the Eq. (1). The plot of Ep versus log v is present in Fig. 4, with a correlation coefficient of 0.9816. The electron-transfer coefficient α can be calculated from the slope of 0.0771 to be 0.62:
Ep = E0 + ×
RT (1 − α)nF
1/2 DR (1 − α)nFv 1/2 0.780 + ln 0 + ln k RF
(1)
In Eq. (1), E0 is the formal potential of the reduction of superoxide ion. We can get E0 from the slope of Ep versus v at low scan rate when v limited to 0. On the assumption that the value of D (diffusion coefficient of dioxygen in aqueous solution) is 1.51 × 10−5 cm s−1 [37], the standard heterogeneous rate constant k0 at the MWNTs-modified graphite can be obtained from the intercept of −0.3179 V, and in the present case it was 5.48 × 10−2 cm s−1 . 3.3. The ECL reaction properties of superoxide ion with sulfide ion
Fig. 3. Cyclic voltammograms of dioxygen reduction in air-saturated 0.1 mol L−1 NaOH: (a) MWNTs-modified graphite electrode, scan rate (0.1 V s−1 ); (b) MWNTs-modified graphite electrode, scan rate (0.01 V s−1 ); (c) bare graphite electrode, scan rate (0.1 V s−1 ).
In our work, it was found that there was a weak but fast ECL emission in sulfide ion solution when a certain potential was applied to this MWNTs-modified electrode. Moreover, this weak ECL signal was enhanced greatly in the presence of oxidative products of rhodamine B as a sensitizer. Thus, as reviewing useful CL systems, it was found that most of these reactions were relatively fast, which showed some unique analytical
R. Huang et al. / Analytica Chimica Acta 582 (2007) 267–274
reaction: r TEC = 0 k
271
(2)
3.4. The other experiment conditions
Fig. 5. ECL intensity–time profile. OPRB solution: 5 mL; sulfide ion: 4.0 × 10−9 mol L−1 ; the applied potential: −0.2 V (vs. Ag/AgCl).
characteristics. Then, the kinetic characteristics of this ECL reaction sensitized by OPRB were studied in detail (as shown in Fig. 5). It was found that the rate of the reaction in solution was relative fast. The ECL intensity reached a maximum within 0.1 s and no ECL was detectable after 0.6 s. In addition, we found that 1000-fold of SO3 2− , I− , NO2 − , C2 O4 2− and equal amount of ascorbic acid did not cause interference. Then, we suggested that the ECL reaction of superoxide ion with sulfide ion was much faster, which may led to higher selectivity based on the speed-resolved, due to difference of their reaction speed with superoxide ion. In general, the ECL reaction procedure concerns two steps: one is electrochemical reaction, and the other is the subsequent CL reaction of the electrogenerated species with themselves or with co-existing ones in the solution. Thus, the speed of the ECL procedure is strongly affected by the rate of either the electrochemical reaction or the subsequent CL reaction, and decided by the lower rate step. In the present work, the CL reaction between superoxide ion and sulfide ion showed high reaction rate, which brought in better analytical performance. To confirm whether the previously electrochemical reaction is in accordance with the subsequent CL reaction, we estimated the time of generating superoxide ion in the electrochemical step according to the standard heterogeneous rate constant. When the dioxygen in solution was adsorbed on the surface of the electrode, the distance between dioxygen and electrode surface is the hydration radius of dioxygen (r), which is also the electron-transfer distance as a reductive potential was applied to this electrode. Then, the time of electron transfer can be calculated in this way: (Eq. (2)) where k0 is the standard heterogeneous rate constant. Accordingly, the time of electron transfer, equal to the electrogenerating superoxide ion, is just a few microseconds. The CL intensity reaches a maximum within 0.1 s, indicating that the time of CL reaction is about 0.1 s. Then, we are concluded that the speed of the previous electrochemical reaction (TEC ) is much faster than subsequent CL
3.4.1. The selection of electrochemical parameters When multi-potential step, linear-sweep potential, cyclic voltammeter was applied, respectively, ECL signal of sulfide ion was observed. It was found that the strongest ECL intensity was obtained when the multi-potential step mode was employed. Therefore, this mode was chosen in this work because of its high sensitivity. Moreover, an investigation of the effect of the applied potential on the ECL signal (as shown in Fig. 6) displays multipotential step at bare graphite electrode and MWNTs-modified graphite electrode. With a maximal ECL signal the bare graphite electrode in the presence of sulfide is observed with a peak potential of −0.6 V (as shown in Fig. 6a). In contrast, the MWNTsmodified electrode exhibited remarkably different voltammetric behavior with a substantial positively shift in the reduction-peak potential. Thus, the more positive reductive potential may lead to higher selectivity. The effect of the applied potential on the ECL signal with the MWNTs-modified electrode (as shown in Fig. 6b) showed that the signal achieving its maximum value at −0.2 V and the enhancement decreased slowly with potential being more negative. We also found that there was a little difference between the applied potential for best ECL intensity and the first peak potential of dioxygen reduction. To be compromised, the optimal electrolysis potential for the system was −0.2 V. 3.4.2. Selection of ECL sensitizer A few fluorescing compounds were tested as energy transferreagents in the CL reaction (data unshown), but most of other energy transfer-reagents produce minor increase of CL efficiency except for the oxidative products of rhodamine B
Fig. 6. Effect of applied potential on the relative ECL intensity: (a) on bare graphite electrode; (b) on MWNTs-modified graphite electrode. OPRB solution: 5 mL; sulfide ion: 4.0 × 10−9 mol L−1 .
272
R. Huang et al. / Analytica Chimica Acta 582 (2007) 267–274
Fig. 7. Effect of oxidative time for OPRB on ECL intensity—the rhodamine B concentration: 5.0 × 10−8 mol L−1 ; H2 O2 concentration: 1.0 × 10−7 mol L−1 ; NaOH concentration: 0.1 mol L−1 ; sulfide ion concentration: 4.0 × 10−9 mol L−1 ; the applied potential: −0.2 V (vs. Ag/AgCl).
Fig. 8. ECL spectrum of the reaction. OPRB solution: 5 mL; sulfide ion: 4.0 × 10−9 mol L−1 ; the applied potential: −0.2 V (vs. Ag/AgCl). The HV of photomultiplier (PMT): −1000 V.
3.4.3. The effect of ECL reaction medium The medium of the proposed ECL reaction system not only affected the generation of superoxide ion but also was the key factor that affected the performances of the subsequent CL reaction. In order to obtain better analytical performances, some medium, such as NaOH, Na2 CO3 , NaHCO3 , CH3 COONa and borax buffer solutions were investigated. The cyclic voltammograms at MWNTs-modified graphite electrode showed two well-defined dioxygen reduction peaks in NaOH solution and it was suitable for the subsequent CL reaction (Fig. 3). In addition, the effect of sodium hydroxide concentration was also investigated. The ECL signal increased when the sodium hydroxide concentration was increased up to 0.1 mol L−1 , but decreased at a higher concentration. Therefore, 0.1 mol L−1 sodium hydroxide was used in all subsequent studies.
is accompanied by weak CL, in which the CL emission was due to the excited sulfur dioxide (SO2 * ) [38]. But in our work, there was no interference in the presence of 1000-fold sulfite, which suggested that SO2 * is not produced in this system. In addition, sulfide ion also can be oxidized to exited sulfur (S2 * ), and the emission spectrum is ranged from 275 to 425 nm [15]. At the same time, the ECL spectrum was investigated through a series of optical filters, which were set separately between the cell and the photomultiplier tube. The result was shown in Fig. 8. It can be seen that the profile has two peaks at around 460 and 560 nm. Unfortunately, no obvious ECL spectrum was obtained for the blank signal due to the weak ECL emission. However, the exploration of the emission spectrum of sulfur and absorption spectrum of OPRB indicated a great overlap (approximate 260–310 nm), which may be satisfied with the requirement of energy transfer. In our work, although H2 O2 may be produced at the electrode surface, there was still amount of hydrogen peroxide in OPRB solution, which was added in for the oxidative products of rhodamine B. In addition, the total concentration is rather low and there was obvious interference when a large amount of was H2 O2 added in. Therefore, the hydrogen peroxide produced by electrochemical reaction had no obviously effect on the CL reaction. Based on these considerations, we proposed that sulfide ion was oxidized to S2 * by superoxide ion. Then, energy is transferred from S2 * to the sensitizer (Eq. (3)), and when the exited state of sensitizer falls back to the ground state, light emission occurs (Eq. (4)). The ECL scheme is postulated as follows (Scheme 1).
3.5. Discussion on the mechanism of the ECL reaction
3.6. Analytical performance
To the best of our knowledge, the oxidation of sulfide ion by some oxidants such as potassium permanganate and so on
Under the above optimum conditions, the calibration of emission intensity versus sulfide concentration was linear in the range
(OPRB). On the other hand, the concentration of rhodamine B and hydrogen peroxide as well as the mixing time were further examined. Firstly, a compromise among the concentration of H2 O2 , rhodamine B and oxidative time had to be made. As the concentration of H2 O2 increased and rhodamine B decreased, the oxidative time decreased. However, the large excess H2 O2 will have a negative effect on the sulfide ion and superfluous sensitizer may lead to self-absorption of the emission. Therefore, 1.0 × 10−7 mol L−1 hydrogen peroxide and 5.0 × 10−8 mol L−1 rhodamine B were respectively selected as optimum for the oxidative products of the rhodamine B. At the same time, the oxidative time selected in this system was 6 h, because the ECL signal increased greatly with the mixing-time going, fell down gradually after 10 h and remained almost constant about 6 h (shown in Fig. 7).
R. Huang et al. / Analytica Chimica Acta 582 (2007) 267–274
273
pool water was diluted 100 times before direct analysis. The analytical results are shown in Table 1. The reasonable agreement was found between the recovery and 100% with t-tests showing no significant difference between the methods at the 95% confidence level. 4. Conclusion Scheme 1. The possible ECL reaction mechanism.
from 6.0 × 10−10 to 1.0 × 10−8 mol L−1 , and the regression equation was I = 26.89C + 79.91 (r = 0.9964), where the concentration (C) is measured in 10−9 mol L−1 . The detection limit (3σ) was 2 × 10−10 mol L−1 . The relative standard deviations for seven replicate measurements of 4 × 10−9 mol L−1 of sulfide is 3.7%. A satisfactory precision of the proposed method is observed.
A novel MWNTs-modified graphite electrode was successfully employed for the generation of superoxide ion. The speed of the electro-reduced superoxide ion is confirmed to be well coupled with the subsequent rapid CL reaction. Then, the ECL performances were decided by the subsequent CL reaction. Due to fast CL reaction between superoxide ion and sulfide ion, as well as the low reductive potential, we got some excellently analytical performances, especially in selectivity. In addition, the possible mechanism was investigated in detail. Acknowledgement
3.7. Interference study In order to assess the selectivity of the proposed method, interference from various cations and anions were investigated by studying their effects on the determination of 4 × 10−9 mol L−1 sulfide. The interference of metal ions were avoided due to the alkaline medium and could be eliminated by a cation exchange column or form metal complexes when added in EDTA, EGTA or citrate as complexants, if necessary. The tolerance of other foreign species was taken as the largest concentration yielding an error of less than ±5% in the analytical signal of sulfide. No interference could be found when including up to 1000-fold K+ , SO4 2− , SO3 2− , Cl− , PO4 3− , CO3 2− , C2 O4 2− , CH3 COO− , I− , NO2 − , F− , EDTA, EGTA, citrate, glucose and equal amount of ascorbic acid. 3.8. Application In order to assess the accuracy of the proposed method, the proposed method was applied to the determination of sulfide in environmental water samples: tap water and pool water. The
Table 1 Determination of sulfide in environmental water samples by the proposed method Found* (nmol L−1 )
Added (nmol L−1 )
Tap water
8.5 8.8 8.6
10 15 20
18.2 24.1 27.6
98.4 101.3 96.5
Pool water
1.1 1.0 1.2
1 2 3
2.0 3.1 4.3
95.2 103.3 102.4
Samples
Recovered (nmol L−1 )
Recovery (%)
* Average of three determinations. OPRB solution: 5 mL; Applied potential: −0.2 V (vs. Ag/AgCl)
The financial support of the National Natural Science Foundation of China (No. 20575040) is gratefully acknowledged. References [1] P. Patnaik, A Comprehensive Guide to the Hazardous Properties of Chemical Substances, 2nd ed., Wiley, New York, 1999. [2] R.P. Pohanish, S.A. Greene, Hazardous Materials Handbook, Van Nostrand Reinhold, New York, 1996. [3] W. Puacz, W. Szahun, Analyst 120 (1995) 939. [4] S. Komatsu, A.J. Shundo, Chem. Soc. Jpn. Pure Chem. Sect. 84 (1967) 1679. [5] C.F. Wood, I.L. Marr, Analyst 113 (1988) 1635. [6] J. Kurzawa, Anal. Chim. Acta 173 (1985) 343. [7] D.R. Canterfold, Anal. Chem. 47 (1975) 88. [8] K. Han, W.F. Koch, Anal. Chem. 59 (1987) 1016. [9] H. Weisz, T. Lenz, Anal. Chim. Acta 70 (1974) 359. [10] A. Safavi, M. Mirzaee, Fresen. J. Anal. Chem. 367 (2000) 645. [11] J. Rodriguez-Fernandez, J.M. Costa, R. Pereiro, A. Sanz-Medel, Anal. Chim. Acta 398 (1999) 23. [12] S.R. Spurlin, E.S. Yeung, Anal. Chem. 54 (1982) 320. [13] J.X. Du, Y.H. Li, J.R. Lv, Anal. Chim. Acta 448 (2001) 79. [14] A. Safavi, M.S. Karimi, Talanta 57 (2002) 491. [15] S.R. Spurlin, E.S. Yeung, Anal. Chem. 54 (1982) 320. [16] X.C. Hu, N. Takenaka, M. Kitano, H. Bandow, Y. Maeda, M. Hattori, Analyst 119 (1994) 1829. [17] X. Hu, N. Takenaka, S. Takasuna, Anal. Chem. 65 (1992) 3489. [18] J.S. Sun, S.G. Schulman, J.H. Perrin, Anal. Chim. Acta 338 (1997) 1. [19] S. Kim, R. DiCosimo, J.S. Filippo Jr., Anal. Chem. 51 (1979) 679. [20] L. Greenlee, I. Fridovich, P. Handler, Biochemistry 1 (1962) 779. [21] C.D. Jaeger, A.J. Bard, J. Phys. Chem. 82 (1979) 3146. [22] T. Okajima, T. Ohsaka, Luminescence 18 (2003) 49. [23] T. Okajima, K. Tokuda, T. Ohsaka, Bioelectrochem. Bioenerg. 41 (1996) 205. [24] R.J. Taylor, A.A. Humffray, J. Electroanal. Chem. 54 (1975) 63. [25] A.J. Appleby, J. Marie, Electrochim. Acta 24 (1978) 1243. [26] S. Iijima, Nature (London) 354 (1991) 56. [27] J.Y. Qu, Y. Shen, X.H. Qu, S.J. Dong, Electroanlysis 16 (2004) 1444. [28] F. Wang, S.S. Hu, J. Electroanal. Chem. 580 (2005) 68. [29] J. Chen, M.A. Hamon, H. Hu, Y. Chen, A.M. Rao, P.C. Ehlund, R.C. Haddon, Science 282 (1998) 95. [30] H. Luo, Z. Shi, N. Li, Z. Gu, Q. Zhuang, Anal. Chem. 73 (2001) 915.
274
R. Huang et al. / Analytica Chimica Acta 582 (2007) 267–274
[31] F.H. Wu, G.C. Zhao, X.W. Wei, Electrochem. Commun. 4 (2002) 690. [32] M. Musameh, J. Wang, A. Merkoci, Y. Lin, Electrochem. Commun. 4 (2002) 743. [33] J. Wang, M. Musameh, Y. Lin, J. Am. Chem. Soc. 125 (2003) 2408. [34] R.R. Moore, C.E. Banks, R.G. Comptom, Anal. Chem. 76 (2004) 2677.
[35] C.X. Cai, J. Chen, Anal. Biochem. 332 (2004) 75. [36] A.J. Bard, L.R. Faulkner, Electrochemical Methods: Fundamentals and Application, 2nd ed., Wiley, New York, 2001, p. 232. [37] D.H. Evans, J.J. Lingane, J. Electroanal. Chem. 6 (1963) 283. [38] J. Staff, W. Jaeschke, Naturforschung B 293 (1978) 33B.
Highly selective electrogenerated chemiluminescence (ECL) for sulfide ion determination at multi-wall carbon nanotubes-modified graphite electrode Rongfu Huang, Xingwang Zheng ∗ , Yingjuan Qu School of Chemistry and Material Science, Shaanxi Normal University, Xi’an 710062, PR China Received 25 July 2006; received in revised form 17 September 2006; accepted 19 September 2006 Available online 26 September 2006
Abstract In the present work, a novel method for immobilization of carbon nanotubes (CNTs) on the surface of graphite electrode was proposed. We further found that superoxide ion was electrogenerated on this CNTs-modified electrode, which can react with sulfide ion combing with a weak but fast electrogenerated chemiluminescence (ECL) emission, and this weak ECL signal could be enhanced by the oxidative products of rhodamine B. In addition, the rate constant of this electrochemical reaction k0 was investigated and confirmed that the speed of electrogenerating superoxide ion was in accordance with the subsequent fast CL reaction. Thus, the fast CL reaction of superoxide ion with target brought in the possibility of high selectivity based on time-resolved, relative to other interferences. Based on these findings, an excellently selective and highly sensitive ECL method for sulfide ion was developed. Under the optimum conditions, the enhancing ECL signals were linear with the sulfide ion concentration in the range from 6.0 × 10−10 to 1.0 × 10−8 mol L−1 , and a 2.0 × 10−10 mol L−1 detection limits (3σ) was achieved. In addition, the proposed method was successfully used to detect sulfide ion in environmental water samples. © 2006 Elsevier B.V. All rights reserved. Keywords: Electrogenerating superoxide ion; Chemiluminescence; Multi-wall carbon nanotubes; Sulfide ion determination; Modified graphite electrode
1. Introduction The detection of sulfide ion has been paid much attention as a consequence of the toxicity of hydrogen sulfide and the corresponding risk associated with exposure in a number of occupational settings [1–3]. Thus, the determination of sulfide ion was particularly important from industrial, environmental, and biological point of view. For this purpose, many analytical schemes, such as titrimetric [4], spectrometric methods [5], electroanalytical methods with chemically modified electrode technique [6], polarographic [7], ion chromatographic [8], catalytic kinetics methods [9,10], and fluorescence [11], chemiluminescence (CL) [12–14] methods, have been developed. Among these methods reported for sulfide ion, the CL methods had been paid considerable attention since this analytical technique promised high sensitivity, wide linear range
∗
Corresponding author. Tel.: +86 29 8530 8184. E-mail address: [email protected] (X. Zheng).
0003-2670/$ – see front matter © 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.aca.2006.09.035
and requirement of simple instrumentation [15]. Although these methods presented straightforward approach to high sensitivity, some powerful or unstable oxidants used in these systems led to poor selectivity. Thus, a highly selective combined with excellently sensitive CL method for the determination of sulfide ion is strongly desirable. As reviewing the powerful CL systems, it was obviously found that most of the useful CL reactions were relatively fast, which led to better selectivity based on the CL speed-resolution power. For example, the CL intensity of hypobromite with urea or ammonia reached a maximum within 20 ms after mixing and no CL signal was detectable after 120 ms. Based on this advantage, two highly selective CL methods for urea or the ammonia were developed [16,17]. Thus, to obtain high CL reaction speed, the powerful CL reaction systems often concerned the addition of some catalysts, the use of instable reagents, and so on. Among those instable reagents used to speed up the CL reaction, such as OH− , BrO− , ClO− , Mn3+ , Co3+ , etc., the superoxide ion was paid much more attention [18,19]. The
268
R. Huang et al. / Analytica Chimica Acta 582 (2007) 267–274
main reason was as followings: firstly, superoxide ion, relative to hydrogen peroxide, presented not only the faster CL reaction speed, but also the higher oxidizing potential from the thermodynamic and kinetic points; secondly, due to the much lower absorption ability in UV–vis region, it nearly did not present the inner filter effect for CL emitter when it used as the CL reaction oxidant. In addition, no interference species was brought in by its reductive product, H2 O. Consequently, some better CL analytical performances, such as the excellent sensitivity and high selectivity, were obtained. Up to now, although some schemes such as biochemical reaction and photochemical reaction, etc. [20,21], have been developed to produce superoxide ion and explored its CL analytical applications, these two methods suffered from poor selectivity due to the seriously chemical interferences from the relating precursors which were used to produce superoxide ion. Compared with these methods, electrochemical reduction dioxygen was paid more attention [22,23]. However, the reduction of dioxygen to superoxide ion depends on the reaction medium and electrode material. For instance, a reversible, one-electron reduction occurs to yield the superoxide ion in aprotic media; Taylor and Humffray [24] suggested that dioxygen was reduced to superoxide ion on glass-carbon electrode first and Appleby and Marie [25] concluded the mechanism of dioxygen reduction on carbon black electrode in alkaline solution, in which superoxide ion was produced first and further oxidated to H2 O2 . On the other hand, carbon nanotubes (CNTs), both singlewall and multi-wall CNTs, have attracted increasing attention because of its remarkable nanostructures, since its discovery by Iijima in 1991 [26]. One promising application is based on the promotion ability of electron-transfer reactions of target molecules when used as an electrode material in electrochemical reactions. In addition, it has been also applied to dioxygen reduction [27,28]. For instance, the multi-wall carbon nanotubes (MWNTs) and cobalt prophyrin-modified glass-carbon (GC) electrode exhibited electrocatalytic activity to the reduction of oxygen; Hu et al. have investigated the electrochemical reduction of dioxygen on the single-wall carbon nanotubes-dihexadecyl phosphoate (SWNTs-DHP) film electrode and the possible mechanism of the reduction of dioxygen was proposed. But up to now, no work was developed to study the CL performances with the superoxide ion electrogenerated on MWNTs-modified electrode for analytical purpose. In this paper, it was found that the instable CL oxidant, superoxide ion, could be rapidly generated at the surface of multiwall carbon nanotubes (MWNTs)-modified graphite electrode and further reacted selectively with sulfide ion, due to speedresolution, producing a weak but fast ECL signal. At the same time, it was also found that this ECL signal could be strongly enhanced in the presence of the oxidative products of rhodamine B (OPRB) based on an energy transfer CL mechanism. Based on these observations, a highly selective and excellently sensitive ECL method for sulfide ion detection at MWNTs-modified graphite electrode was developed. In addition, the proposed ECL method was successfully applied to the determination of sulfide ion in environmental water samples.
2. Experimental 2.1. Reagents All the reagents were of analytical-reagent grade and were directly used for following experiments without purification. Cetyltrimethylammonium bromide (CTAB), NaOH, paraffin, rhodamine B, were purchased from Xi’an & Shanghai Chem. Ind. Co. MWNTs, were purchased from Shenzhen Nanotech Port Co. Ltd. The pretreatment method was showed in next part this paper. Hydrogen peroxide 30% (v/v) was diluted with doubledistilled water daily. Solutions of rhodamine B (1.0 × 10−3 mol L−1 ) were prepared by accurately weighing and diluting to appropriate concentration. Stock sulfide ion solutions were prepared by dissolving appropriate amount of Na2 S·9H2 O in water, which were standardized using iodine solution and sodium thiosulfate solutions. The instability of very dilute sulfide ion solution was well known. For this reason, the calibration solutions were prepared serially from the solution of next highest concentration and analyzed immediately. 2.2. Apparatus The applied potential for different electrolytic ways such as linear sweep, multi-potential step and cyclic voltammeter were performed with a conventional three-electrode cell linked to a CHI660b Electrochemistry Working Station (CH Instruments, Inc.). The electrolytic cell utilized a conventional three-electrode setup and was arranged as shown in Fig. 1. The cell was made of a microbeaker (high: 35 mm, i.d.: 25 mm). The working electrode was the MWNTs-modified graphite electrode (Shanghai, 12.6-mm2 surface area); a Pt flake (7 mm × 7 mm) and Ag/AgCl (saturated KCl solution) were used as the auxiliary electrode and pseudo-reference electrode, respectively. The ECL intensity was transformed into an electrical signal by an R456 photomultiplier (PMT) (Xi’an Remax Electronic Science Tech. Co. Ltd., Xi’an, China), which was operated at −800 V and the ECL cell was placed in front of the PMT.
Fig. 1. The block diagram of ECL detection system. W, working electrode; R, Ag/AgCl reference electrode; C, counter electrode (Pt); L, KNO3 salt bridge; PMT, photomultiplier; NHV, negative high voltage supply.
R. Huang et al. / Analytica Chimica Acta 582 (2007) 267–274
2.3. Preparation of MWNTs-modified graphite electrode Multi-wall carbon nanotubes (MWNTs) were sonicated in concentrated nitric acid at 25 ◦ C for about 24 h, then filtered and washed with double-distilled water several times for further purification until the filtrate became neutral and dried in an oven at 37 ◦ C. The graphite electrode was carefully polished on rough and fine sandpapers successively, and the electrode was then subjected to be sonicated to remove adsorbed particles and rinsed with double-distilled water several times. About 1.0 mg nitric acid-treated MWNTs and some amount of CTAB were dispersed together in 10 mL liquid paraffin to give a black suspension with the aid of ultrasonic agitation for about 1 h. Subsequently, the above electrode was immersed in and sonicated for a few minutes. A stable and uniform MWNTsmodified graphite electrode was prepared. 2.4. The analytical procedure When adding 1.0 × 10−7 mol L−1 hydrogen peroxide, 5.0 × 10−8 mol L−1 rhodamine B to 0.1 mol L−1 NaOH solution, the difference between UV–vis spectra of rhodamine B and the mixture solution indicated that rhodamine B was oxidized by hydrogen peroxide. 5.0 mL blank solution, the above oxidative products of rhodamine B solution (OPRB), was added into the ECL cell and a stable blank signal was recorded when the electrolytic potential was applied to the working electrode; the sample or standard sulfide ion solution containing not only the oxidative products of rhodamine B, but also an appropriate concentration of sulfide ion was added to the ECL cell, and the ECL signal was recorded. The concentration of sulfide ion was quantified via the peak height of the relative ECL emission intensity which was obtained by subtracting the blank ECL emission intensity from that of the sample or standard sulfide ion solution.
269
According to the literatures, Comptom and co-workers [34] found CNTs’ effective electrocatalytical behavior on basal plane pyrolytic graphite electrodes, immobilized by abrasive attachment and film modification (dissolved in acetone); Cai and Chen [35] directly dispersed the CNTs in a CTAB solution and fabricated a CNT-modified electrode which has promotion effects on the direct electron transfer of glucose oxidase. In our work, we found that the nitric acid-treated MWNTs (carboxylic CNTs) could form special ion complex (CTAB–CNTs) with cationic surfactant CTAB through electrostatic interaction. Meanwhile, this ion complex CTAB–CNTs can be well dispersed into paraffin by ultrasonication for a few minutes. We further found that this suspension was very stable after 3 weeks. In addition, it was well-know that liquid paraffin can well permeate into the graphite layer or the pore on the surface of the bare graphite electrode and form a very stable paraffin dipping film on the graphite electrode, due to the hydrophobic nature of graphite plane. Thus, while the bare graphite electrode was immersed into the paraffin, which contained the CTAB–CNTs complex, for a given time, the CTAB–CNTs was more stable modified to graphite electrode surface. Compared with Cai’s work [35], not only MWNTs were effectively immobilized, but also it provided a hydrophobic nature on the surface of graphite electrode. Furthermore, the electrochemical behavior of the MWNTsmodified graphite electrode was investigated. Fig. 2 showed the electrochemical response of K3 Fe(CN)6 at different work electrodes: the cyclic voltammogram of K3 Fe(CN)6 on liquid paraffin with CTAB dipping graphite electrode (curve (a)) is poor; the potential difference (Ep ) between the oxidation and the reduction peaks is 256 mV and its stability is unsatisfactory. However, the voltammetric response is apparently improved at the MWNTs-modified graphite electrode, reflected by the enlargement of the peak currents (Ip ) and the decline of the potential difference (Ep = 63 mV), with a pair of excellent symmetrical redox peaks (curve (b)). And we found the peak potential and currents remained stable after a few cycles.
3. Results and discussion 3.1. The immobilization of the MWNTs on the surface of graphite electrode Though the carbon nanotubes-modified electrodes, in comparison with other forms carbon electrode, showed better performance due to its dimensions, the electronic structure, and the topological defects present on the tube surface, a major barrier for developing such CNTs-modified electrode is the insolubility of CNTs in most solvents [29]. However, many efforts have been made to disperse CNTs into suitable solvents, such as dimethylformamide (DMF) [30], acetone [31], concentrated sulfuric acid [32] and perfluorosulfonated polymer Nafion [33], and then, this suspension was cast on a substrate electrode to form a carbon nanotube film. Even now, these CNTs casting films on the electrode surface often presented the poor mechanical stability and resulted in the poor reproducibility of the electrochemical response. In this case, the new immobilization schemes for CNTs are highly desired to broaden the application of CNTs-modified electrode.
Fig. 2. Cyclic voltammograms at different work electrode in 1 mmol L−1 K3 Fe(CN)6 solution: (a) liquid paraffin with CTAB dipping graphite electrode; (b) MWNTs-modified graphite electrode.
270
R. Huang et al. / Analytica Chimica Acta 582 (2007) 267–274
The reason for the better performance of the MWNTs-modified electrode might arise from the promotion of electron-transfer reaction on the CNTs surfaces. Meanwhile the CNTs increased the effective area of the electrode, so the peak current increased significantly. 3.2. Electrogenerating superoxide ion at MWNTs-modified graphite electrode The cyclic voltammograms of dioxygen in 0.1 mol L−1 airsaturated NaOH solution at MWNTs electrode is shown in Fig. 3a and b (scan rate: 0.1 and 0.01 V s−1 ) when the potential initially sweeps from 0.2 to −1.2 V. According to the cyclic voltammograms, it is obvious that two well-defined dioxygen reduction peaks are observed at about −0.3 and −0.9 V, being consistent with similar observations by others, in which the dioxygen was firstly reduced to hydrogen peroxide [28]. In addition, when superoxide dismutase (SOD) was added into the reaction system, the first reduction peak decreased greatly. Based these observation, it was inferred that superoxide ion can be produced in the first step as an intermediate and further reduced to H2 O2 [25]. On the other hand, the second reduction peak is still visible at low scan rate in 0.1 mol L−1 air-saturated NaOH solution, indicating that the first reductive product is quite stable for the subsequent further reduction, which may result from the hydrophobic micro-environment of the waxed graphite electrode surface (Fig. 3b). Correspondingly, the bare graphite electrode shows quite contrary effect to the reduction of dioxygen (Fig. 3c) combined with a low background current and a flat, broad reduction peak. Compared with two peaks at MWNTs-modified graphite electrode, the separation of two peaks is due to the difference of rate constant of two steps. Then, the two peaks do not separate [36].
Fig. 4. Plot of Ep vs. log v. The influence of scan rate on the first reduction-peak potential of dioxygen in air-saturated 0.1 mol L−1 NaOH solution.
In addition, the heterogeneous rate constant of the first step of dioxygen reduction was investigated at MWNTs-modified graphite electrode and it was approximately considered as the rate constant of electrogenerated superoxide ion. According to Fig. 3, the electrochemical behavior of dioxygen was irreversible. As an irreversible reaction, the peak potential of dioxygen shift negatively with the increase of scan rate (v) and have a linear relationship with the natural logarithm of scan rate [36]. The first reduction-peak potential changes according to the Eq. (1). The plot of Ep versus log v is present in Fig. 4, with a correlation coefficient of 0.9816. The electron-transfer coefficient α can be calculated from the slope of 0.0771 to be 0.62:
Ep = E0 + ×
RT (1 − α)nF
1/2 DR (1 − α)nFv 1/2 0.780 + ln 0 + ln k RF
(1)
In Eq. (1), E0 is the formal potential of the reduction of superoxide ion. We can get E0 from the slope of Ep versus v at low scan rate when v limited to 0. On the assumption that the value of D (diffusion coefficient of dioxygen in aqueous solution) is 1.51 × 10−5 cm s−1 [37], the standard heterogeneous rate constant k0 at the MWNTs-modified graphite can be obtained from the intercept of −0.3179 V, and in the present case it was 5.48 × 10−2 cm s−1 . 3.3. The ECL reaction properties of superoxide ion with sulfide ion
Fig. 3. Cyclic voltammograms of dioxygen reduction in air-saturated 0.1 mol L−1 NaOH: (a) MWNTs-modified graphite electrode, scan rate (0.1 V s−1 ); (b) MWNTs-modified graphite electrode, scan rate (0.01 V s−1 ); (c) bare graphite electrode, scan rate (0.1 V s−1 ).
In our work, it was found that there was a weak but fast ECL emission in sulfide ion solution when a certain potential was applied to this MWNTs-modified electrode. Moreover, this weak ECL signal was enhanced greatly in the presence of oxidative products of rhodamine B as a sensitizer. Thus, as reviewing useful CL systems, it was found that most of these reactions were relatively fast, which showed some unique analytical
R. Huang et al. / Analytica Chimica Acta 582 (2007) 267–274
reaction: r TEC = 0 k
271
(2)
3.4. The other experiment conditions
Fig. 5. ECL intensity–time profile. OPRB solution: 5 mL; sulfide ion: 4.0 × 10−9 mol L−1 ; the applied potential: −0.2 V (vs. Ag/AgCl).
characteristics. Then, the kinetic characteristics of this ECL reaction sensitized by OPRB were studied in detail (as shown in Fig. 5). It was found that the rate of the reaction in solution was relative fast. The ECL intensity reached a maximum within 0.1 s and no ECL was detectable after 0.6 s. In addition, we found that 1000-fold of SO3 2− , I− , NO2 − , C2 O4 2− and equal amount of ascorbic acid did not cause interference. Then, we suggested that the ECL reaction of superoxide ion with sulfide ion was much faster, which may led to higher selectivity based on the speed-resolved, due to difference of their reaction speed with superoxide ion. In general, the ECL reaction procedure concerns two steps: one is electrochemical reaction, and the other is the subsequent CL reaction of the electrogenerated species with themselves or with co-existing ones in the solution. Thus, the speed of the ECL procedure is strongly affected by the rate of either the electrochemical reaction or the subsequent CL reaction, and decided by the lower rate step. In the present work, the CL reaction between superoxide ion and sulfide ion showed high reaction rate, which brought in better analytical performance. To confirm whether the previously electrochemical reaction is in accordance with the subsequent CL reaction, we estimated the time of generating superoxide ion in the electrochemical step according to the standard heterogeneous rate constant. When the dioxygen in solution was adsorbed on the surface of the electrode, the distance between dioxygen and electrode surface is the hydration radius of dioxygen (r), which is also the electron-transfer distance as a reductive potential was applied to this electrode. Then, the time of electron transfer can be calculated in this way: (Eq. (2)) where k0 is the standard heterogeneous rate constant. Accordingly, the time of electron transfer, equal to the electrogenerating superoxide ion, is just a few microseconds. The CL intensity reaches a maximum within 0.1 s, indicating that the time of CL reaction is about 0.1 s. Then, we are concluded that the speed of the previous electrochemical reaction (TEC ) is much faster than subsequent CL
3.4.1. The selection of electrochemical parameters When multi-potential step, linear-sweep potential, cyclic voltammeter was applied, respectively, ECL signal of sulfide ion was observed. It was found that the strongest ECL intensity was obtained when the multi-potential step mode was employed. Therefore, this mode was chosen in this work because of its high sensitivity. Moreover, an investigation of the effect of the applied potential on the ECL signal (as shown in Fig. 6) displays multipotential step at bare graphite electrode and MWNTs-modified graphite electrode. With a maximal ECL signal the bare graphite electrode in the presence of sulfide is observed with a peak potential of −0.6 V (as shown in Fig. 6a). In contrast, the MWNTsmodified electrode exhibited remarkably different voltammetric behavior with a substantial positively shift in the reduction-peak potential. Thus, the more positive reductive potential may lead to higher selectivity. The effect of the applied potential on the ECL signal with the MWNTs-modified electrode (as shown in Fig. 6b) showed that the signal achieving its maximum value at −0.2 V and the enhancement decreased slowly with potential being more negative. We also found that there was a little difference between the applied potential for best ECL intensity and the first peak potential of dioxygen reduction. To be compromised, the optimal electrolysis potential for the system was −0.2 V. 3.4.2. Selection of ECL sensitizer A few fluorescing compounds were tested as energy transferreagents in the CL reaction (data unshown), but most of other energy transfer-reagents produce minor increase of CL efficiency except for the oxidative products of rhodamine B
Fig. 6. Effect of applied potential on the relative ECL intensity: (a) on bare graphite electrode; (b) on MWNTs-modified graphite electrode. OPRB solution: 5 mL; sulfide ion: 4.0 × 10−9 mol L−1 .
272
R. Huang et al. / Analytica Chimica Acta 582 (2007) 267–274
Fig. 7. Effect of oxidative time for OPRB on ECL intensity—the rhodamine B concentration: 5.0 × 10−8 mol L−1 ; H2 O2 concentration: 1.0 × 10−7 mol L−1 ; NaOH concentration: 0.1 mol L−1 ; sulfide ion concentration: 4.0 × 10−9 mol L−1 ; the applied potential: −0.2 V (vs. Ag/AgCl).
Fig. 8. ECL spectrum of the reaction. OPRB solution: 5 mL; sulfide ion: 4.0 × 10−9 mol L−1 ; the applied potential: −0.2 V (vs. Ag/AgCl). The HV of photomultiplier (PMT): −1000 V.
3.4.3. The effect of ECL reaction medium The medium of the proposed ECL reaction system not only affected the generation of superoxide ion but also was the key factor that affected the performances of the subsequent CL reaction. In order to obtain better analytical performances, some medium, such as NaOH, Na2 CO3 , NaHCO3 , CH3 COONa and borax buffer solutions were investigated. The cyclic voltammograms at MWNTs-modified graphite electrode showed two well-defined dioxygen reduction peaks in NaOH solution and it was suitable for the subsequent CL reaction (Fig. 3). In addition, the effect of sodium hydroxide concentration was also investigated. The ECL signal increased when the sodium hydroxide concentration was increased up to 0.1 mol L−1 , but decreased at a higher concentration. Therefore, 0.1 mol L−1 sodium hydroxide was used in all subsequent studies.
is accompanied by weak CL, in which the CL emission was due to the excited sulfur dioxide (SO2 * ) [38]. But in our work, there was no interference in the presence of 1000-fold sulfite, which suggested that SO2 * is not produced in this system. In addition, sulfide ion also can be oxidized to exited sulfur (S2 * ), and the emission spectrum is ranged from 275 to 425 nm [15]. At the same time, the ECL spectrum was investigated through a series of optical filters, which were set separately between the cell and the photomultiplier tube. The result was shown in Fig. 8. It can be seen that the profile has two peaks at around 460 and 560 nm. Unfortunately, no obvious ECL spectrum was obtained for the blank signal due to the weak ECL emission. However, the exploration of the emission spectrum of sulfur and absorption spectrum of OPRB indicated a great overlap (approximate 260–310 nm), which may be satisfied with the requirement of energy transfer. In our work, although H2 O2 may be produced at the electrode surface, there was still amount of hydrogen peroxide in OPRB solution, which was added in for the oxidative products of rhodamine B. In addition, the total concentration is rather low and there was obvious interference when a large amount of was H2 O2 added in. Therefore, the hydrogen peroxide produced by electrochemical reaction had no obviously effect on the CL reaction. Based on these considerations, we proposed that sulfide ion was oxidized to S2 * by superoxide ion. Then, energy is transferred from S2 * to the sensitizer (Eq. (3)), and when the exited state of sensitizer falls back to the ground state, light emission occurs (Eq. (4)). The ECL scheme is postulated as follows (Scheme 1).
3.5. Discussion on the mechanism of the ECL reaction
3.6. Analytical performance
To the best of our knowledge, the oxidation of sulfide ion by some oxidants such as potassium permanganate and so on
Under the above optimum conditions, the calibration of emission intensity versus sulfide concentration was linear in the range
(OPRB). On the other hand, the concentration of rhodamine B and hydrogen peroxide as well as the mixing time were further examined. Firstly, a compromise among the concentration of H2 O2 , rhodamine B and oxidative time had to be made. As the concentration of H2 O2 increased and rhodamine B decreased, the oxidative time decreased. However, the large excess H2 O2 will have a negative effect on the sulfide ion and superfluous sensitizer may lead to self-absorption of the emission. Therefore, 1.0 × 10−7 mol L−1 hydrogen peroxide and 5.0 × 10−8 mol L−1 rhodamine B were respectively selected as optimum for the oxidative products of the rhodamine B. At the same time, the oxidative time selected in this system was 6 h, because the ECL signal increased greatly with the mixing-time going, fell down gradually after 10 h and remained almost constant about 6 h (shown in Fig. 7).
R. Huang et al. / Analytica Chimica Acta 582 (2007) 267–274
273
pool water was diluted 100 times before direct analysis. The analytical results are shown in Table 1. The reasonable agreement was found between the recovery and 100% with t-tests showing no significant difference between the methods at the 95% confidence level. 4. Conclusion Scheme 1. The possible ECL reaction mechanism.
from 6.0 × 10−10 to 1.0 × 10−8 mol L−1 , and the regression equation was I = 26.89C + 79.91 (r = 0.9964), where the concentration (C) is measured in 10−9 mol L−1 . The detection limit (3σ) was 2 × 10−10 mol L−1 . The relative standard deviations for seven replicate measurements of 4 × 10−9 mol L−1 of sulfide is 3.7%. A satisfactory precision of the proposed method is observed.
A novel MWNTs-modified graphite electrode was successfully employed for the generation of superoxide ion. The speed of the electro-reduced superoxide ion is confirmed to be well coupled with the subsequent rapid CL reaction. Then, the ECL performances were decided by the subsequent CL reaction. Due to fast CL reaction between superoxide ion and sulfide ion, as well as the low reductive potential, we got some excellently analytical performances, especially in selectivity. In addition, the possible mechanism was investigated in detail. Acknowledgement
3.7. Interference study In order to assess the selectivity of the proposed method, interference from various cations and anions were investigated by studying their effects on the determination of 4 × 10−9 mol L−1 sulfide. The interference of metal ions were avoided due to the alkaline medium and could be eliminated by a cation exchange column or form metal complexes when added in EDTA, EGTA or citrate as complexants, if necessary. The tolerance of other foreign species was taken as the largest concentration yielding an error of less than ±5% in the analytical signal of sulfide. No interference could be found when including up to 1000-fold K+ , SO4 2− , SO3 2− , Cl− , PO4 3− , CO3 2− , C2 O4 2− , CH3 COO− , I− , NO2 − , F− , EDTA, EGTA, citrate, glucose and equal amount of ascorbic acid. 3.8. Application In order to assess the accuracy of the proposed method, the proposed method was applied to the determination of sulfide in environmental water samples: tap water and pool water. The
Table 1 Determination of sulfide in environmental water samples by the proposed method Found* (nmol L−1 )
Added (nmol L−1 )
Tap water
8.5 8.8 8.6
10 15 20
18.2 24.1 27.6
98.4 101.3 96.5
Pool water
1.1 1.0 1.2
1 2 3
2.0 3.1 4.3
95.2 103.3 102.4
Samples
Recovered (nmol L−1 )
Recovery (%)
* Average of three determinations. OPRB solution: 5 mL; Applied potential: −0.2 V (vs. Ag/AgCl)
The financial support of the National Natural Science Foundation of China (No. 20575040) is gratefully acknowledged. References [1] P. Patnaik, A Comprehensive Guide to the Hazardous Properties of Chemical Substances, 2nd ed., Wiley, New York, 1999. [2] R.P. Pohanish, S.A. Greene, Hazardous Materials Handbook, Van Nostrand Reinhold, New York, 1996. [3] W. Puacz, W. Szahun, Analyst 120 (1995) 939. [4] S. Komatsu, A.J. Shundo, Chem. Soc. Jpn. Pure Chem. Sect. 84 (1967) 1679. [5] C.F. Wood, I.L. Marr, Analyst 113 (1988) 1635. [6] J. Kurzawa, Anal. Chim. Acta 173 (1985) 343. [7] D.R. Canterfold, Anal. Chem. 47 (1975) 88. [8] K. Han, W.F. Koch, Anal. Chem. 59 (1987) 1016. [9] H. Weisz, T. Lenz, Anal. Chim. Acta 70 (1974) 359. [10] A. Safavi, M. Mirzaee, Fresen. J. Anal. Chem. 367 (2000) 645. [11] J. Rodriguez-Fernandez, J.M. Costa, R. Pereiro, A. Sanz-Medel, Anal. Chim. Acta 398 (1999) 23. [12] S.R. Spurlin, E.S. Yeung, Anal. Chem. 54 (1982) 320. [13] J.X. Du, Y.H. Li, J.R. Lv, Anal. Chim. Acta 448 (2001) 79. [14] A. Safavi, M.S. Karimi, Talanta 57 (2002) 491. [15] S.R. Spurlin, E.S. Yeung, Anal. Chem. 54 (1982) 320. [16] X.C. Hu, N. Takenaka, M. Kitano, H. Bandow, Y. Maeda, M. Hattori, Analyst 119 (1994) 1829. [17] X. Hu, N. Takenaka, S. Takasuna, Anal. Chem. 65 (1992) 3489. [18] J.S. Sun, S.G. Schulman, J.H. Perrin, Anal. Chim. Acta 338 (1997) 1. [19] S. Kim, R. DiCosimo, J.S. Filippo Jr., Anal. Chem. 51 (1979) 679. [20] L. Greenlee, I. Fridovich, P. Handler, Biochemistry 1 (1962) 779. [21] C.D. Jaeger, A.J. Bard, J. Phys. Chem. 82 (1979) 3146. [22] T. Okajima, T. Ohsaka, Luminescence 18 (2003) 49. [23] T. Okajima, K. Tokuda, T. Ohsaka, Bioelectrochem. Bioenerg. 41 (1996) 205. [24] R.J. Taylor, A.A. Humffray, J. Electroanal. Chem. 54 (1975) 63. [25] A.J. Appleby, J. Marie, Electrochim. Acta 24 (1978) 1243. [26] S. Iijima, Nature (London) 354 (1991) 56. [27] J.Y. Qu, Y. Shen, X.H. Qu, S.J. Dong, Electroanlysis 16 (2004) 1444. [28] F. Wang, S.S. Hu, J. Electroanal. Chem. 580 (2005) 68. [29] J. Chen, M.A. Hamon, H. Hu, Y. Chen, A.M. Rao, P.C. Ehlund, R.C. Haddon, Science 282 (1998) 95. [30] H. Luo, Z. Shi, N. Li, Z. Gu, Q. Zhuang, Anal. Chem. 73 (2001) 915.
274
R. Huang et al. / Analytica Chimica Acta 582 (2007) 267–274
[31] F.H. Wu, G.C. Zhao, X.W. Wei, Electrochem. Commun. 4 (2002) 690. [32] M. Musameh, J. Wang, A. Merkoci, Y. Lin, Electrochem. Commun. 4 (2002) 743. [33] J. Wang, M. Musameh, Y. Lin, J. Am. Chem. Soc. 125 (2003) 2408. [34] R.R. Moore, C.E. Banks, R.G. Comptom, Anal. Chem. 76 (2004) 2677.
[35] C.X. Cai, J. Chen, Anal. Biochem. 332 (2004) 75. [36] A.J. Bard, L.R. Faulkner, Electrochemical Methods: Fundamentals and Application, 2nd ed., Wiley, New York, 2001, p. 232. [37] D.H. Evans, J.J. Lingane, J. Electroanal. Chem. 6 (1963) 283. [38] J. Staff, W. Jaeschke, Naturforschung B 293 (1978) 33B.