Performance optimization of a membrane assisted passive sampler for monitoring of ionizable organic compounds in water

Abstract

A thin-walled silicone rubber hollow fibre membrane has been developed as a passive sampler. The inside of the tube is filled with an aqueous solution at an appropriate pH. The tube is sealed at both ends and then immersed in a water sample. In order for the ionizable permeating compounds to be trapped in the aqueous receiving phase, the pH is adjusted such that the compounds are ionized and trapped. The major advantages are its simplicity, low cost and high selectivity, since only ionizable organic compounds are trapped. Additionally, the sampler uses no organic solvent. By adjusting the pH of the acceptor phase, it is possible to control the extraction process and whether the sampler is used in the kinetic or equilibrium regime. Since it is very selective, no further clean-up of the extract is required. The membrane assisted passive sampler has been tested for extraction of chlorophenols under laboratory conditions. The extraction process was found to be linear over a 72 h sampling period. Selectivity of the passive sampler in river water was demonstrated and the extraction process was independent of sample concentration, even at lower concentration levels of analytes. However, the sample matrix in some river water samples led to incomplete trapping, thereby reducing the amount trapped in the acceptor phase. Detection limits (three times signal to noise ratio) were dependant on sample matrix and type of detection system and ranged from 0.05 µg L^–1 to about 1 µg L^–1 with a UV photodiode detector in water samples from one river and 1.0 µg L^–1 to 20 µg L^–1 in another but with an ordinary UV detector. The enrichment factors in river water were 28 for 2-chlorophenol and 44 for 2,4-dichlorophenol over a 72 h sampling period. 4-chlorophenol was poorly extracted and its enrichment factor was 3.

---

Introduction

Monitoring of the presence of various organic chemicals, such as phenols, in water bodies is important as it gives vital information on the toxic risks associated with them. The presence and levels of phenols in aquatic environments is of concern because of their wide spread release as by-products in the production of plastics and dyes, pulp and paper industries. Chlorinated phenols can also form during wastewater treatment, since chlorine is added as a disinfectant. Because of the toxicity and persistent nature of these compounds, they need to be determined, even at single µg L^–1 levels.

The common approach is to collect part of the environmental media, which is later analyzed for potential pollutants in the laboratory. This approach, among other advantages, provides manageable control over accuracy and precision of the results. However, information obtained from grab environmental samples is only about concentration levels at the time of sampling and may fail to account for episodic contamination. This can be addressed by collecting many representative samples over a time period but the cost of analysis is increased.

An alternative and more cost effective approach is to obtain time-weighted average (TWA) concentrations of pollutants using passive samplers. Sodergren¹ developed a first sampler for the monitoring of compounds in water bodies consisting of a dialysis membrane filled with an organic solvent. The major disadvantage of the system was successive loss of the organic solvent from the device through diffusion during environmental exposure and lack of selectivity. Huckins et al.,² described the second field sampler, a semi-permeable membrane device (SPMD) for passive and integrative in situ monitoring of water and air borne contaminants. The SPMD sampler consisted of layflat polyethylene tubing containing a thin film of triolein, a high molecular weight neutral lipid. This has been commercialized.³ The other advantage, apart from being very rugged, is that high preconcentration factors of the analytes are obtained and are limited by the partition coefficients into the lipid and exposure time.⁴ The disadvantage of the SPMD is the laborious recovery of the analytes, usually involving large volumes of organic solvents.

Other passive samplers have recently been reported, especially those based on solid phase material as trapping media. These include the Chemcatcher that use commercially available solid phase extraction C₁₈ Empore disks as receiving phase.^5,6 The partitioning of the analytes onto the sorbent and their subsequent desorption is similar to the solid phase extraction technique. The type of Empore disk can also be chosen according to the properties of the analytes of interest. A membrane enclosed sorptive coating (MESCO) passive sampler that uses a stir bar coated with poly(dimethylsiloxane) (PDMS) enclosed in a dialysis membrane bag as receiving phase has also been described.⁷ The stirrer bar used as a receiving phase is similar to the one used in the stir bar sorptive extraction (SBSE) technique.⁸ It combines the advantages of the passive sampling approach with solventless preconcentration of organic solutes from aqueous matrices and subsequent desorption of the sequested analytes on line with capillary gas chromatography. It avoids clean up of extracts, required for other samplers, and the whole extract is injected. A new MESCO that uses a silicone collector, instead of a stirrer bar coated with PDMS, has been reported by Paschke et al.⁹ The silicone rods are later thermally desorbed into a gas chromatographic system, as in the MESCO. A polar organic chemical integrative sampler (POCIS) has been described by Alvarez et al.¹⁰ It consists of a solid receiving phase material enclosed in microporous polyethersulfone diffusion-limiting membrane. Jonsson’s research group¹¹ has also reported to have developed an equilibrium sampling through membranes (ESTM) technique for measuring the free fraction of ionizable organic compounds in water that could be applied as a passive sampler. In this case, a porous polypropylene hollow fibre membrane is impregnated with a non-polar solvent. The inner side of the hollow fibre serves as the acceptor phase and is filled with appropriate aqueous solution. The same research group has recently reported an equilibrium sampling of freely dissolved organic chemicals into a thin film of 1-octanol supported on a porous hollow fibre membrane, which could also be applied as a passive sampler.¹² In both of the above cases, solvent loss may be a problem when the equilibrium extraction techniques are used as passive samplers. However, this could depend on environmental conditions and sampling time. As interest in passive samplers is growing, more passive samplers for monitoring organic compounds in water bodies have been reported. These are discussed in several review papers.^13–16

In this work, a simple and very selective membrane assisted passive sampler (MAPS), that does not use organic solvents, based on a silicone hollow fibre membrane for extraction of ionisable organic compounds in water bodies, is described. The potential of the sampler, especially its selectivity and simplicity, have been demonstrated in the laboratory using chlorophenols as model compounds.

Theory of the extraction process

The theory of extraction for the MAPS is similar to the one developed for the supported liquid membrane (SLM) extraction technique.^17–19 In brief, in order for the chlorophenols to dissolve into the silicone membrane from the sample, they have to be non-ionized at the sample pH. The compounds then diffuse through the membrane into the acceptor phase. Once in the acceptor phase, they are ionized and trapped. The concentration of non-ionized phenols in the acceptor phase is thus kept at zero. This maintains a concentration gradient between the two phases; the donor and acceptor phases. This way the concentration of the compounds in the acceptor solution can be increased much higher than in the original sample, without experiencing a plateau or maximum, and is limited by the sample volume and/or the extraction time. This also gives selective enrichment, since only compounds that are ionized at the pH of the acceptor phase are enriched.²⁰ Compounds that are ionized at the pH of the sample solution do not dissolve into the membrane, since they are too polar. Large molecules have slow diffusion in the membrane and are less excluded. From the developed theory,^17–19 an acidic compound (like chlorophenols) needs an acceptor pH that is 3.3 units above its pK_a value for it to be completely trapped and the sample pH must be 2 units below its pK_a for the compounds to dissolve into the membrane. Therefore, it is possible to determine the best trapping conditions for the acceptor phase in the sampler from the pK_a of the ionisable organic compound.

Two important parameters that are often measured in the liquid membrane extraction techniques are the extraction efficiency (E), and enrichment factor, E_n. The extraction efficiency is defined as the fraction of analyte in the extracted sample that is found in the acceptor phase and is given by the equation below.^17–19 It is also a measure of mass transfer between the donor and acceptor phase and is constant under specified extraction conditions.

E = C_AV_A/C_DV_D (1)

Where C_A is the concentration in the collected acceptor fraction and C_D is the concentration in the extracted sample. V_A is the collected acceptor volume while V_D is the volume of the sample that has been extracted. The enrichment factor is a ratio of concentration found in the acceptor phase to that in the original sample. This determines the detection limit of the method. It is given by the equation below.^17–19

E_n = C_A/C_D (2)

Both the extraction efficiency and the enrichment factor are constant under specified extraction conditions, such as extraction time, sampler dimensions, turbulence and sample volume.

Experimental

Chemicals

The following chlorophenol standards were used: 4-chlorophenol (>99%), 2-chlorophenol (>98%) and 2,4-dichlorophenol (>99%) were from Merck–Schuchardt (Darmstadt, Germany). Other chemicals used were trisodium phosphate (99%), proanalysis sulfuric acid (99%) and HPLC grade methanol and acetonitrile, all from Merck–Schuchardt (Darmstadt, Germany).

Solutions

Stock solutions at 1000 mg L^–1 were prepared in methanol and were stored in the refrigerator at 4 °C. Fresh stock solutions were prepared after every three months. Solutions for hollow fibre extraction were prepared by diluting the stock with deionised water, and the methanol concentration did not exceed 0.1%.

Chromatographic conditions

The three chlorophenols were best separated with a mobile phase composition of 60% water and 40% acetonitrile, at a flow rate of 1.0 mL min^–1, with the UV detector set at 280 nm. A C₁₈ column with dimensions 25 cm × 4 mm × 5 µm was used from Supelco (Bellefonte, PA, USA). Two chromatographic systems were used. At first, a Waters HPLC system (Milford, MA, USA) with a 515 pump, Waters 996 Photodiode Array detector and Millenium³² chromatographic software was used. The mobile phase was continuously degassed by a 200 series Perkin Elmer degasser (PA, USA) and an injection volume of 100 µL was used. In some experiments, an SRI (LA, California, USA) 210 HPLC system with a UV detector (VUV-24) and peak simple chromatographic system was used. The mobile phase was degassed offline and filtered before use. The injection volume was 20 µL.

Hollow fibres

Three hollow fibre membranes used for the optimization process were from Technical Products Inc., (Georgia, USA). Their dimensions are summarized in Table 1. The hollow fibres were bought as long tubes that were cut to appropriate lengths when used.

Table 1 Dimensions of the three hollow fibres used in studying how the thickness influences the extraction process

Type	Inner diameter/cm	Outer diameter/cm	Thickness/cm	Length/cm	Volume/cm^3	Surface area/cm^2
Smallest	0.0635	0.1194	0.0284	58	0.187	21.74
Medium	0.1575	0.2413	0.0420	23.5	0.458	17.81
Biggest	0.4775	0.9525	0.2375	7.8	1.396	23.33

---

Procedures

River water samples. River water samples for optimization of the extraction procedure were used without any manipulation, except pH measurements. River water samples were taken 10 cm deep from a perennial river about 5 km east of the University of Venda, Limpopo Province and from a stream near Super sport park, Centurion, Gauteng Province, South Africa, in glass containers. The containers were thoroughly washed with soap, soaked in acid and then rinsed with deionised water before use. Water samples were then spiked with known concentrations of the chlorophenols, except for the blank. Samples were extracted immediately.

Preparation of the hollow fibre membranes and extraction procedure. The basic acceptor solution was obtained by dissolving phosphate buffer in deionised water at concentrations of 0.5 M. The hollow fibre silicone rubber membrane, previously soaked in deionised water, was filled with acceptor buffer using a peristaltic pump (Minipuls 3; Gilson Medical Electronics, Villiers-Le-Bel, France). The flow system consisted of acid resistant tubing (Acid-Flexible; Elkay Products, Shrewsbury, MA, USA). The hollow fibre ends were then tightened together and made in a form of a loop about 3 cm in diameter. The outside was rinsed with deionized water thoroughly to remove any buffer spills. It was then immersed in a 750 mL sample in a conical flask and left hanging for an appropriate time. There after, it was taken out, the outside washed or flushed with deionised water and its contents transferred into a 4 mL vial. The buffer solution of the acceptor solution was adjusted by adding 100 µL of 2 M sulfuric acid for a total volume of 1246 µL. The extracts were either analyzed immediately or stored in the refrigerator at 4 °C. Fig. 1 shows a typical set-up of the extraction procedures. Hollow fibres filled with only the un-spiked water samples were also extracted to check for any chlorophenols already in the water or for any possible contamination.

Fig. 1 Schematic set-up for the passive sampling.

Optimization of extraction parameters

Thickness and length of the hollow fibres. The samplers used in this study were selected on the basis of surface area to volume ratio, where we used three fibres of different diameter. These were trimmed to different lengths to obtain approximately similar surface areas (Table 1). In addition, for the medium diameter fibre, we varied the length in order to study the influence of surface area on the mass transfer (Table 2). 0.5 M phosphate buffer was used as acceptor buffer and river water spiked with 0.5 mg L^–1 of the chlorophenols was used as sample with 24 h extraction time. Each experiment was repeated twice.

Table 2 The volume and surface area of the three different lengths of a medium hollow fibre (0.1575 cm i.d × 0.2413 cm o.d)

Length/cm	Volume/cm^3	Surface area/cm^2
29	0.565	21.97
48	0.935	36.37
56	1.090	42.43

---

Sample volume. In order to study the influence of sample volume on the extraction process of the passive sampler, various volumes (250, 500, 750 and 2000 mL) of deionized water spiked with 0.05 mg L^–1 of the chlorophenols were extracted over a 24 h period. A medium hollow fibre with a length of 48 cm was used. Each experiment was repeated twice.

Extraction time. To assess the influence of exposure time on the uptake of the chemicals in the sampler, fibres of a given size (0.1575 cm i.d × 48 cm length) were exposed for 12, 24, 48 and 72 h. A sample volume of 750 mL were used and each extraction was repeated twice. Hollow fibres were filled with 0.5 M phosphate buffer and extraction was performed with deionized water, spiked with the 0.05 mg L^–1 mixture of the chlorophenols as sample.

Influence of turbulence. Turbulence was studied by comparison of the enrichment factors under static conditions and under stirred conditions in river water. 750 mL was used as the sample volume and 0.01 mg L^–1 mixtures of the chlorophenols were extracted for 72 h. The experiment was repeated at least three times. The revolutions per minute were set at 220 and were measured by a photo/contact tachometer.

Sample matrix and limits of detection. A key assumption that is at the core of the usefulness of passive sampling techniques is that the uptake rate is concentration independent. For these experiments, 4, 10 and 20 µg L^–1 mixtures of the chlorophenols were spiked in both river water and deionized water. The spiked samples were extracted for 72 h as before. A 48 cm long medium hollow fibre was used for these experiments. Each experiment was repeated at least three times, except those of deionized water where duplicates were performed.

The detection limit was obtained from chromatograms obtained after extraction of the 10 µg L^–1 mixtures of the chlorophenols spiked in river water and extracted for 72 h. The detection limit was calculated as the concentration that would give a peak height three times the baseline noise.

Preparation of calibration curves. External calibration standards were prepared from the stock solutions in deionized water. The concentration of the phenols in the extracted samples was thus quantified by injecting along the calibration standard solutions. Peak area was used for quantification. Standard solutions in µg L^–1 ranges were prepared daily while those in mg ^–1 range were stored in the refrigerator for about 3 days.

Results and discussion

Optimizations of the sampler

Varying the thickness and length of the hollow fibres. Fig. 2a and 2b shows the normalized concentrations obtained in optimizations of hollow fibre thickness and length, respectively. The results were normalized with respect to the smallest volumes (Tables 1 and 2). The results (Fig. 2a), though not very conclusive, suggest that hollow fibre thickness of 0.420 mm (420 µm) gave better extractions. The reason why the thinner tube did not give the highest amount extracted, since analyte diffusion would not take long, is unclear. Perhaps the fibre may not have the same uniform composition, so that other materials used like silica could influence the extraction process. Jonsson et al.,¹¹ in the development of ESTM, is reported to have optimized the thickness of hollow fibres impregnated with non-polar organic liquid. In this case, a slightly thicker tubing (50 µm) was chosen instead of a much smaller one, despite the latter having fast mass transfer. This was based on the convenience to handle the bigger tubing. In this study, the medium hollow fibre was chosen as the best thickness, as it gave much improved mass transfer and was also easy to handle.

Fig. 2 (a) Plots of the hollow fiber thickness and (b) length of the hollow fiber against the amount extracted for 24 h from deionized water spiked with 0.5 mg L^–1 mixtures of chlorophenols; (c) sample volume and (d) extraction time against enrichment factor after extraction of 0.05 mg L^–1 mixtures of chlorophenols spiked in deionized water.

The influence of the hollow fibre surface area on the mass transfer (Fig. 2b) was also not very conclusive. A length of 56 cm did not give much better extraction compared to the 48 cm longer fibre despite having a higher surface area. The reason for this could be positioning in the sample container that could influence the extraction process. A length of 48 cm was chosen for convenience in handling, especially putting in the sample container. A fibre that is too long would require folding several times before putting it in the sample container.

Varying the sample volume and extraction time. In applications of passive samplers in real environments, the sample volume should be infinitely large so that it does not influence the extraction process. Fig. 2c shows the influence of sample volume on the enrichment factor. It shows that the enrichment factor increased with sample volume. However, from 750 mL of sample volume, the enrichment factor reached a plateau. A sample volume of 750 mL was therefore taken as the optimum. The optimum sample volume is influenced by the exposure time and the uptake kinetics of the chemicals from donor into the acceptor phase of the sampler. This observation has also been noted by Jonsson et al.,¹¹ in the optimization of the ESTM with chlorophenol as a model compound.

If the trapping conditions are correctly set in the acceptor phase,^17,18 the extraction can go on until all the analytes in the sample are extracted. This means that a linear relationship is expected between the extraction time and the amount accumulated in the acceptor phase. Fig. 2d shows that such a relationship was obtained. Linear relationships have been observed in other passive samplers working in the kinetic regime, such as the Chemcatcher,^5,6 MESCO^7,21 and PDMS.⁹

Influence of turbulence. Turbulence was studied by comparison of the enrichment factors under static conditions and under stirred conditions in river water. The results of the experiments are shown in Fig. 3. The passive sampler is influenced by turbulence, as shown in the results. The influence was more with 2,4-dichlorophenol. For 2-chlorophenol, turbulence had little influence because of the problem of incomplete trapping in the acceptor phase, which was limiting the mass transfer. It means that equilibrium was almost reached between the concentrations of non-ionized compounds in the acceptor and donor phases. It is therefore possible to use the sampler either in the kinetic or equilibrium regime by simply adjusting the pH of the receiving phase. Before the proposed sampler is deployed for field applications, preliminary laboratory experiments need to be performed to study the influence of matrix effects on the trapping capability. If the sampler needs to be applied in the kinetic regime and matrix effects are reducing the trapping of the target analytes, one could reduce the exposure time and/or use a much stronger buffer solution. Vrana et al.⁶ also investigated the influence of hydrodynamics in a Chemcatcher passive sampler and obtained similar observations. At low turbulence, the diffusion through the aqueous layer limits the mass transfer for more hydrophobic compounds, resulting in higher enrichment factors at high turbulence.

Fig. 3 Comparison of enrichment factors under static and stirred conditions in river water spiked with 10 µg L^–1 mixtures of chlorophenols and with extraction time of 72 h.

Memory effect. Memory effects were not studied in detail but no indication was observed during extractions that could pose a problem for the compounds investigated. This is mostly as a result of slow mass transfer between the membrane and acceptor phase.^22–25 This is generally more pronounced with compounds with high octanol–water partition coefficients.²⁵Memory effects in this case can prevent the hollow fibre from being re-used, as slowly diffusing analytes interfere with the next extraction, requiring thorough washing in between extractions. In this case, hollow fibres could easily be re-used but as a precaution, were thoroughly washed and soaked in deionized water for at least twenty-four hours.

Linearity, sample matrix and limits of detection. Experiments were performed to find out whether the extraction process was independent of the sample concentration and matrix. This is very important for quantification purposes, since the sampler has to be calibrated in the laboratory with known concentrations and used to determine unknown concentrations in real samples. A linear relationship was obtained between extracted concentration and the amount accumulated in the acceptor phase in river water (Fig. 4). The percentage relative deviation ranged from 1.3 to 34% for 2-chlorophenol and from 4 to 20% for 2,4-dichlorophenol. In other active sample preparation applications of membrane extractions,^20,26 linear relationships were also obtained between acceptor phase and sample concentrations. Such linear relationships have also been observed in other passive samplers,^4,6,7,9 as part of their performance optimization.

Fig. 4 Plot of the different extracted concentrations against accumulated amount in river water after 72 h extraction.

Fig. 5 shows a comparison of the enrichment factor in river water and deionized water extracted over a 72 h period. From the Figure, it shows that enrichment factors were much lower in river water compared to deionized water. In river water, 4-chlorophenol was not even detected, due to poor extraction. The most probable cause of this is due to co-extraction of other acidic compounds in river water that are also in high concentrations. This lowers the acceptor pH, thus leading to incomplete trapping of the target compounds. This agrees well with the results, since 2-chlorophenol, with a highest pK_a, would be affected most by pH decrease in the acceptor phase, with least affected being 2,4-dichlorophenol, with a pK_a of 7.6. In this case, a much stronger/concentrated buffer is recommended. Jonsson et al.,¹¹ studied the decrease of acceptor pH due to matrix found in various water samples, such as deionized water, river water and leachate water. pH decreased most in leachate and river waters and very little in deionized water.

Fig. 5 Comparison of the enrichment factors in river and deionized waters. ▲ and ○ represent enrichment factors in deionized water. ◆ and ■ represent enrichment factors in river water. Extraction time was for 72 h with optimized parameters.

Table 3 shows the comparison of detection limits by direct injection and after hollow fibre passive sampling in deionized and river water. The detection limit was calculated as the concentration that would give a peak height three times the baseline noise. It clearly shows that the developed sampler can be used for passive sampling of chlorophenols in water bodies at trace levels.

Table 3 Comparison of detection limits by direct injection and after hollow fiber passive extraction in deionized and river waters

Sample type	Detection limits/µg L^–1
	2-Chlorophenol	4-Chlorophenol	2,4-Dichlorophenol
a River water-2 was from a different area and was extracted for 24 h instead of 72 h. Injection volume was also 100 µL instead of 20 µL and with a photodiode array detector.
Direct injection	60	65	60
Deionised water	0.5	1.5	0.7
River water-1	1.5	20	1.0
River water-2^a	0.05	1.0	0.85

---

Fig. 6a shows the chromatogram passive extraction of a 10 µg L^–1 mixture of the chlorophenols in spiked deionized water (Fig. 6a) and river water (Fig. 6b). The clean chromatograms demonstrate the selectivity of the passive sampler.

Fig. 6 Chromatograms obtained after 72 h passive extraction of 10 µg L^–1 mixtures of chlorophenols spiked in deionized water (a) and in river water (b).

Conclusions

The potential for passive sampling of acidic compounds in MAPS has been demonstrated. The sampler is inexpensive, selective and uses no organic solvents. By changing the acceptor solution from basic to acidic conditions, the MAPS can be used to extract basic compounds. However, before the passive sampler is used for field application, the influence of environmental factors, such as temperature, turbulence and biofouling, will have to be investigated. These factors are currently being considered as part of further method development, along with enclosing the silicone hollow fibre into an enclosure, such as copper mesh.

Acknowledgements

Authors would like to thank Prof. Jönsson (Lund University-Sweden) for donating the peristaltic pump used in the experimental and Prof. Rohwer (University of Pretoria-South Africa) for donating initial silicone hollow fibre tubes. The authors also thank Water Research Commission of South Africa for funding.

References

1. A. Södergren, Environ. Sci. Technol., 1987, 21, 855.
2. J. N. Huckins, M. W. Tubergen and G. K. Manuweera, Chemosphere, 1990, 20, 533.
3. Environmental Science Technologies Inc. (USA), http://est-lab.com.
4. P. Sabaliunas and A. Sodergren, Environ. Pollut., 1997, 96, 195.
5. J. K. Kingston, R. Greenwood, G. A. Mills, G. M. Morrison and L. B. Person, J. Environ. Monit., 2000, 2, 487.
6. B. Vrana, G. A. Mills, E. Dominiak and R. Greenwood, Environ. Pollut., 2006, 142, 333.
7. B. Vrana, P. Popp, A. Paschke and G. Schuurmann, Anal. Chem., 2001, 73, 5191.
8. E. Baltussen, P. Sandra, F. David and C. Cramels, J. Microcolumn Sep., 1999, 11, 737.
9. A. Paschke, K. Schwab, J. Brummer, G. Schuurmann, H. Paschke and P. Popp, J. Chromatogr., A, 2006, 1124, 187–195.
10. D. A. Alvarez, J. D. Petty, J. N. Huckins, T. L. Jones-Lepp, J. P. Goddard and S. E. Manahan, Environ. Toxicol. Chem., 2004, 23, 1640.
11. J. F. Liu, J. Å. Jonsson and P. Mayer, Anal. Chem., 2005, 77, 4800.
12. J. F. Liu, X. L. Hu, J. F. Peng, J. Å. Jonsson, P. Meyer and G. B. Jiang, Anal. Chem., 2006, 78, 8526.
13. B. Vrana, G. A. Mills, I. J. Allan, E. Dominiak, K. Svensson, J. Knutsson, G. Morrison and R. Greenwood, TrAC, Trends Anal. Chem., 2005, 24, 845.
14. Y. Lu, Z. Wang and J. Huckins, Aquat. Toxicol., 2002, 60, 139.
15. J. D. Petty, C. E. Orazio, J. N. Huckins, R. W. Gale, J. A. Lebo, J. C. Meadows, K. R. Echools and W. L. Cranor, J. Chromatogr., A, 2000, 879, 83.
16. F. Stuer-Lauridsen, Environ. Pollut., 2005, 136, 503.
17. L. Chimuka, N. Megersa, J. Norberg, L. Mathiasson and J. Å. Jonsson, Anal. Chem., 1998, 70, 3906.
18. J. Å. Jonsson and L. Mathiasson, TrAC, Trends Anal. Chem., 1999, 18, 325.
19. J. Å. Jönsson, P. Lövkvist, G. Audunsson and G. Nilvé, Anal. Chim. Acta, 1993, 277, 9.
20. N. Megersa, T. Solomon and J. Å. Jönsson, J. Chromatogr., A, 1999, 830, 203.
21. B. Vrana, A. Paschke and P. Popp, Environ. Pollut., 2006, 144, 296.
22. M. Papantoni, N. K. Djane, K. Ndungu, J. Å. Jönsson and L. Mathiasson, Analyst, 1995, 120, 1471.
23. M. Knutsson, L. Mathiasson and J. Å. Jönsson, Chromatographia, 1996, 42, 165.
24. T. Miliotis, M. Knutsson, J. Å. Jönsson and L. Mathiasson, Intern. J. Environ. Anal. Chem., 1996, 64, 35.
25. L. Chimuka, L. Mathaiasson and J. Å. Jönsson, Anal. Chim. Acta, 2000, 416, 77.
26. N. Megersa and J. Å. Jönsson, Analyst, 1998, 123, 225.

---