Equilibrium Sampling through Membranes of Freely Dissolved Copper

Zhao Dong , Christopher G. Lewis , Robert M. Burgess , James P. Shine ... Antonio López-López , Jan Ake Jönsson , Manuel García-Vargas , Carlos Mo...
0 downloads 0 Views 99KB Size
Anal. Chem. 2005, 77, 7605-7611

Equilibrium Sampling through Membranes of Freely Dissolved Copper Concentrations with Selective Hollow Fiber Membranes and the Spectrophotometric Detection of a Metal Stripping Agent Roberto Romero,† Jing-fu Liu,† Philipp Mayer,‡ and Jan Åke Jo 1 nsson*,†

Department of Analytical Chemistry, Lund University, P.O. Box 124, SE-221 00 Lund, Sweden, and Department of Environmental Chemistry and Microbiology, National Environmental Research Institute, P.O. Box 358, 4000 Roskilde, Denmark

A sensitive spectrophotometric method for the determination of freely dissolved copper concentrations in aqueous samples after preconcentration with hollow fiber membrane extraction has been developed. The method is based on the equilibrium sampling through a selective membrane into an acceptor solution containing 4-(pyridyl2-azo)resorcinol (PAR), which serves as stripping agent and metal indicator. Negligible extraction of interferences and equilibrium enrichment of copper allowed for selective spectrophotometric determination of the Cu-PAR complex. Some important extraction parameters such as acceptor composition, shaking, equilibrium time, and sample volume were studied. The optimized methodology showed good linearity in the range of 5-100 µg/L, an enrichment factor of 93, good repeatability and reproducibility (RSDs < 6%, n ) 6), and a detection limit of 4 µg/ L. The cationic metals Ni2+, Co2+, Cd2+, Fe3+, Pb2+, Zn2+, and Mn2+ were shown not to interfere with the measurement of Cu2+. Measurements on samples containing mixtures of various ligands and cations were in good agreement with theoretically calculated concentrations, and the method was also applied to environmental samples. The developed technique requires less labor and less sophisticated equipment than conventional methods typically based on atomic absorption spectrometry or ICP. Total concentrations of metals are poor indicators of their toxicity in natural aquatic systems since metals are distributed in various chemical forms such as free hydrated ions, dissolved inorganic, organic complexes, and metals associated with colloidal particles. These different forms can vary greatly in terms of bioavailability,1 and it is thus important to quantify freely dissolved concentrations of metals in aqueous samples for approximate * Corresponding author. E-mail: [email protected]. Fax: +46 46 222 45 44. Tel: +46 46 222 81 69. † Lund University. ‡ National Environmental Research Institute. (1) Slaveykova, V. I.; Parthasarathy, N.; Buffle, J.; Wilkinson, K. J. Sci. Total Environ. 2004, 328, 55-68. 10.1021/ac050763a CCC: $30.25 Published on Web 10/26/2005

© 2005 American Chemical Society

characterization of their bioavailable fraction.2 Several methods have been developed and applied for this purpose, including ionselective electrodes3, anodic or cathodic stripping voltammetry,4,5 and also ion-exchange resins.6 Besides the well-known advantages of these instrumental techniques (e.g., precision and accuracy), they also have disadvantages related to, for instance, chemical interferences, complexity, difficulty to apply in situ, or disturbance of solution equilibria.7 Other emerging speciation techniques include diffusive gradients in thin films,8,9 Donnan membrane,2,7 and the in-situ sampling device “gellyfish”.10 They are all based on the selective enrichment of the free or labile metal concentration and the subsequent instrumental metal analysis. These methods generally apply sampling times of days, and they require a subsequent instrumental analysis by atomic absorption spectroscopy (AAS) or ICPMS. There remains a need for methods that combine shorter sampling times with a simple and widely available detection method that preferably can be operated in the field. In the present study, we apply “equilibrium sampling through membranes (ESTM)” to generate an enriched and purified extract that allows the spectrophotometric detection of copper. The approach is based on (1) a hollow fiber supported liquid membrane that is selective for copper, (2) the metal stripping agent 4-(pyridyl-2-azo)resorcinol (PAR) that simultaneously is used as metal indicator, and finally (3) the simple spectrophotometric detection of the Cu-PAR complex. Supported liquid membrane (SLM) extraction11 is an emerging trace metal separation and preconcentration tool that has been (2) Temminghoff, E. J. M.; Plette, A. C. C.; Van Eck, R.; Van Riemsdijk, W. H. Anal. Chim. Acta 2000, 417, 149-157. (3) Yoshimoto, S.; Mukai, H.; Kitano, T.; Sohrin, Y. Anal. Chim. Acta 2003, 494, 207-213. (4) Sauve´, S.; Norvell, W. A.; McBride, M.; Hendershot, W. Environ. Sci. Technol. 2000, 34, 291-296. (5) Kogut, M. B.; Voelker, B. M. Environ. Sci. Technol. 2001, 35, 1149-1156. (6) Liu, Y.; Ingle, J. D. Anal. Chem. 1989, 61, 525-529. (7) Nolan, A. L.; Mclaughlin, M. J.; Mason, S. D. Environ. Sci. Technol. 2003, 37, 90-98. (8) Zhang, H.; Davison, W. Anal. Chem. 1995, 67, 3391-3400. (9) Zhang, H.; Davison, W. Anal. Chem. 2000, 72, 4447-4457. (10) Senn, D. B.; Griscom, S. B.; Lewis C. G.; Galvin, J. P.; Chang, M. W.; Shine, J. P. Environ. Sci. Technol. 2004, 38, 3381-3386. (11) Jo ¨nsson, J. Å.; Mathiasson, L. Trends Anal. Chem. 1999, 18, 318-325.

Analytical Chemistry, Vol. 77, No. 23, December 1, 2005 7605

successfully applied for the study of trace metal speciation.12,13 In a carrier-facilitated SLM, a thin impregnated microporous membrane separates the source phase (donor) and the acceptor phase solution and the membrane is impregnated with a waterimmiscible solvent containing a suitable lipophilic ligand as carrier. Different physical formats of porous membranes are available for the application in SLM, including flat sheet, spiral wound, and hollow fiber.14 The application of hollow fibers in SLM is termed by some authors HFSLM15 and by others “liquid-phase microextraction using hollow fiber membranes”.16 It has a number of advantages over flat liquid membranes such as stability, which may be ascribed to the absence of solubilization, emulsion formation, and osmotic pressure difference.17 Besides, HFSLM is a highly robust, reliable, and low-cost alternative to SLM based on flat membranes. Further, the low cost and easy construction of a hollow fiber device allows for its disposable usage. Nevertheless, only very few studies with HFSLM have so far been performed on trace metal speciation.12,18 In the present study, we combined hollow fiber extraction with a copper-selective liquid membrane that has been developed previously19 using 1,10-dibenzyl-1,10-diaza-18-crown-6 as carrier, oleic acid in the membrane, and dihexyl ether as organic solvent. The metal ion transport across the membrane occurs by (1) complexation of the metal ion, M, with the carrier to form CML at the sample/membrane interface, (2) partitioning into the organic phase, (3) diffusion of the CML complex to the membrane/ acceptor solution interface, (4) release of M at the membrane/ acceptor interface, and (5) complexation of the metal with the stripping agent at the acceptor solution. The fatty acid (oleic acid) acts as countercation (L) during the complexation of the heavy metal in the membrane.20 Diffusive sampling techniques such as SLM and solid-phase microextraction can be operated and optimized in many different ways. They can be operated in either the kinetic uptake regime or under equilibrium conditions.21 Further, they can be tuned to extract as much analyte as possible from a sample, or alternatively, they can sense freely dissolved concentrations without depleting them and without disturbing any speciation equilibria.21-23 ESTM refers to the combination of “equilibrium sampling” and “negligible depletion”,24 and it can also be characterized as equilibrium sampling within an infinite bath. The hollow fiber format is very (12) Parthasarathy, N.; Pelletier, M.; Buffle, J. Anal. Chim. Acta 1997, 350, 183195. (13) Parthasarathy, N.; Pelletier, M.; Buffle, J. J. Chromatogr., A 2004, 1025, 33-40. (14) Jo ¨nsson, J. Å.; Mathiasson, L. J. Sep. Sci. 2001, 24, 495-507. (15) Fonta´s, C.; Palet, C.; Salvado´, V.; Hidalgo, M. J. Membr. Sci. 2000, 178, 131-139. (16) Rasmussen, K. E.; Pedersen-Bjergaard, S. Trends Anal. Chem. 2004, 23, 1-10. (17) He, T.; Versteeg, L. A. M.; Mulder, M. H. V.; Wessling, M. J. Membr. Sci. 2004, 234, 1-10. (18) Parthasarathy, N.; Pelletier, M.; Tercier-Waeber, M. L.; Buffle, J. Electroanalysis 2001, 13, 1305-1314. (19) Romero, R.; Jo ¨nsson, J. Å. Anal. Bional. Chem. 2005, 381, 1452-1459. (20) Guyon, F.; Parthasarathy N.; Buffle, J. Anal. Chem. 2000, 72, 1328-1333. (21) Pawliszyn, J. Solid-phase microextraction: theory and practice; Wiley: New York, 1997. (22) Vaes, W. H. J.; Urrestarazu-Ramos, E.; Verhaar, H. J. M., Seinen, W.; Hermens, J. L. M. Anal. Chem. 1996, 68, 4463-4467. (23) Mayer, P.; Vaes, W. H. J.; Wijnker, F.; Legierse, K. C. H. M.; Kraaij, R.; Tolls, J.; Hermens, J. L. M. Environ. Sci. Technol. 2000, 34, 5177-5183. (24) Liu, J.; Jo ¨nsson, J. Å.; Mayer, P. Anal. Chem. 2005, 77 , 4800-4809

7606

Analytical Chemistry, Vol. 77, No. 23, December 1, 2005

suited for ESTM because it provides a high surface-to-volume ratio that is crucial for the fast attainment of a thermodynamic equilibrium.25 Additionally, the small size of the hollow fibers minimizes the impact on the sample. Up to now, the sampling of trace metals using SLM has traditionally be combined with the analysis using AAS12,26 or voltammetric methods,18 although recently, a fluorometric method has been published.27 Spectrophotometric measurements would normally provide insufficient sensitivity and selectivity for this purpose, unless preceded by the highly selective enrichment and purification of SLM. The application of spectrophotometry for the determination of copper and other transition metals in waters is advantageous because it is easily operated, widely available, and inexpensive when compared to the previous techniques. It also has advantages for the construction of field instruments, where a sensitive detection can be obtained with simple means. The enrichment of copper within the present ESTM method is driven by the complex binding of Cu2+ to PAR. PAR is highly suited for this purpose because it forms a water-soluble copper complex with a high molar absorptivity (104 L mol-1 cm-1) and hence exhibits high sensitivity for photometric detection.28 PAR forms complexes with many other cationic trace metals, and the selectivity of a spectrophotometric method based on PAR would thus be insufficient without the selectivity of the membrane. The purpose of this work is to develop an improved and simple method based on equilibrium sampling through membranes for the spectrophotometric determination of the freely dissolved fraction of copper in natural waters. This is to be seen as a model system for a particular application; devices like the one presented here could be developed for other metal ions, thereby extending the possibilities for determination of freely dissolved concentrations with simple equipment. EXPERIMENTAL SECTION Chemicals and Materials. Stock standard solution of 1000 mg/L copper in dilute nitric acid was prepared from copper(II) nitrate trihydrate (Merck, Darmstadt, Germany) extra pure (g99.5%); this solution was used for further dilutions. Nitric acid (65% Suprapur), phthalic acid (GR for analysis), Na2EDTA (GR for analysis), catechol (Reag. Ph Eur), salicylic acid (Reag. Ph Eur) (all from Merck), and humic acid (Fluka, Buchs, Switzerland) were used in the experiments. The organic liquid membrane was optimized previously (see ref 19) and consisted of 6.61 mM 1,10dibenzyl-1,10-diaza-18-crown-6 (DbzDA18C6), g97% (Fluka) and 8.61 mM oleic acid (Fluka) dissolved in dihexyl ether, g97% (Fluka). The optimized acceptor solution consisted of a solution of 0.15 mM PAR, (Merck), previously dissolved in water, and 0.1 M ammonium acetate (Merck) at pH 8.7. For investigation of Cu(II) extraction, the synthetic solutions used as source solutions during optimization consisted of 50 µg/L Cu(II) in pure water at pH 7 (pH was adjusted with Merck (25) Mayer, P.; Tolls, J.; Hermens, J. L. M.; Mackay, D. Environ. Sci. Technol. 2003, 37, 184A-191A. (26) Mendiguchı´a, C.; Moreno, C.; Garcı´a-Vargas, M. Anal. Chim. Acta 2002, 460, 35-40. (27) Ueberfeld, J.; Parthasarathy, N.; Zbinden, H.; Gisin, N.; Buffle, J. Anal. Chem. 2002, 74, 664-670. (28) Bin-Abas, M. R.; Takruni, I. A.; Abdullah, Z.; Tahir, N. M. Talanta 2002, 58, 883-890.

Suprapur KOH, 0.1 M), and the free fraction of copper was estimated using Visual Minteq V 2.30 (see below). For the optimization of acceptor pH, different buffer solutions were prepared, for a final concentration in the acceptor of 0.1 M. Thus, pH 4 and 5 was prepared by acetic acid and sodium acetate (Merck), pH 6 with citric acid and NaOH (Merck), pH 7 with sodium dihydrogen phosphate and NaOH (Merck), pH 8-9 with ammonium acetate, adjusting the pH with NH3 (Merck), and pH 10 with sodium carbonate and sodium bicarbonate (Merck). Other chemicals were of analytical reagent grade. Ultrapure water was obtained from a Milli-Q Gradient water system (Millipore, Bedford, MA). All glassware was cleaned with 2 M nitric acid for several days and finally rinsed with high-purity reagent water immediately before use. 50/280 Accurel PP polypropylene hollow fiber tubing (50-µm wall thickness, 280-µm inner diameter, 0.1-µm pore size) was obtained from Membrana GmbH (Wuppertal, Germany) at a cost of less than 0.04 Euro/m. BD Micro-Fine Syringe (with a needle of 0.30-mm outer diameter and 8-mm length, 0.5 mL, prepared for U-100 insulin injection), obtained from BD Consumer Healthcare, was used to fill the acceptor into the lumen of the hollow fiber for extraction and to flush out the acceptor into a small glass vial (50 µL, Alltech) after extraction. Equipment and Software. Spectrophotometric detection was carried out using an Ocean Optics PC1000 spectrometer (Dunedin, FL), with a LS-1 tungsten light source. Optical fibers (50-µm diameter) were connected to the light source and to the detector. FIAlab for Windows 5.8 software from FIAlab Instruments (Bellevue, WA) was used for data acquisition. A quartz cell microcuvette with 5-mm path length and 2.5 µL of chamber volume from Hellma (Mu¨lheim, Germany) was used for spectrophotometric measurements. Determination of total copper concentration was performed by electrothermal atomic absorption GFAAS, using a GBC 932 system with System 3000 automated graphite furnace (GBC Scientific Equipment, Dandenong, Australia). A pH 211 microprocessor pH meter (Hanna Instruments) was used to adjust the sample and acceptor buffer pH and to measure the sample pH. Aqueous solutions were shaken with a HT Infors orbital shaker from Infors AG (Bottmingen, Switzerland). Statgraphics Plus for Windows 4.1 (Statistical Graphics, Rockville, MD) was used for data manipulation. The concentrations of free and complexed copper were determined by Visual Minteq V. 2.30 (http://www.lwr.kth.se/ english/OurSoftware/Vminteq/index.htm), Cheaqs V L20.1 (A Program for Calculating Chemical Equilibria in Aquatic Systems, RIVM, Bilthoven, The Netherlands) and WinHumicV for Win95/ 98/NT (http://www.lwr.kth.se/english/OurSoftWare/WinHumicV/ index.htm). Membrane Preparation and Extraction Procedure. The hollow fiber was cut manually and carefully into a 15-cm length, and the lumen of the hollow fiber was filled with acceptor solution using the needle of the BD Micro-Fine Syringe. Then, the fiber was immersed in the organic phase for ∼10 s in order to form the organic liquid membrane. The fiber was washed with water in order to remove the excess of the organic phase, and fresh acceptor solution was injected into the hollow fiber, making sure

that the lumen was completely filled with acceptor without air bubbles. To prevent leakages, the fiber was sealed using a strip of aluminum foil and inserted into a piece of small glass tubing. After this, the hollow fiber device was ready for sampling. A more detailed description of the applied hollow fiber supported liquid membrane device has been reported by Liu et al.24 The whole extraction device was completely immersed into the sample solution of 100 mL contained in a 250-mL volume flask. The sample flasks were continuously shaken for 75 min at 125 rpm on an orbital shaker. Finally, the acceptor solution was collected by flushing it into a 50-µL vial with a BD Micro-Fine Syringe. After that, the collected acceptor was injected in the quartz cell microcuvette and the absorbance was measured at 504 nm. The preparation of the extraction device and the subsequent sampling consist of rather simple steps and they do not require laboratory facilities. This makes the proposed method applicable for passive field sampling. Sample Collection. Three leachate waters from Kristianstad (Skåne, Sweden) were collected, and the hollow fiber supported liquid membrane device was applied without any further pretreatment of the water. The leachate waters were directly from landfill, from the influent to a treatment plant, and from the outflow from the same treatment plant. RESULTS AND DISCUSSION The enrichment factor, Ee was selected as response variable for the optimization of the extraction process, and it was defined as

Ee ) CA/CS

(1)

where CA is the measured copper concentration in the acceptor and CS is the freely dissolved copper concentration in the sample.11,24 Influence of Acceptor-Phase Composition on the Extraction of Copper. Copper diffuses through the membrane into the acceptor solution where it can form a water-soluble chelate with PAR, and this is the basis for the continuous enrichment of copper until a thermodynamic equilibrium is reached. These processes depend on the properties of the membrane and of the acceptor solution. The subsequent experiments were carried out to optimize the composition of the acceptor solution, since the liquid membrane has been optimized in an earlier study,19 by using the flat sheet configuration. The initial conditions in the acceptor solution, based on the work of Bobrowska-Grzesik and Grossman,29 were as follows: 0.19 mM PAR, previously dissolved in water and 0.1 M ammonium acetate at pH 8.2. First, the concentration of PAR was varied between 0.05 and 1 mM. The enrichment factor, Ee, increased with increasing PAR concentration (0.05-0.1 mM) and then remained constant at a value of 62 for PAR concentration up to 0.2 mM (Figure 1). Above 0.2 mM, Ee decreased to values lower than 1 (no enrichment), and this coincided with the coloring of the membrane likely caused by the precipitation of PAR on the inner membrane surface. (29) Bobrowska-Grzesik, E.; Grossman, A. M. Fresenius J. Anal. Chem. 1996, 354, 498-502.

Analytical Chemistry, Vol. 77, No. 23, December 1, 2005

7607

Figure 1. Influence of PAR concentration on copper extraction. Extraction conditions: source solution 50 µg/L Cu2+ at pH 7. Acceptor phase: Different concentrations of PAR previously dissolved in water and 0.1 M ammonium acetate at pH 8.2. Organic phase: 6.6 × 10-3 M DbzDA18C6 and 8.6 × 10-3 M oleic acid in dihexyl ether. Sample volume, 100 mL; extraction time, 60 min; shaking rate, 135 rpm.

Figure 2. Effect of the pH acceptor solution on the enrichment factor. Extraction conditions: source solution 50 µg/L Cu2+ at pH 7. Acceptor phase: 0.15 mM PAR previously dissolved in water at different pHs, using 0.1 M of buffer concentration. Other conditions are the same as in Figure 1.

To study whether the Cu-PAR complex was adsorbed in the membrane when high concentrations of PAR were used, two extractions were performed at PAR concentrations of 0.15 and 0.75 mM. After collecting the acceptor, the hollow fibers were immersed in a tube containing 10 mL of 1 M HNO3 for 2 h to dissolve the complex from the hollow fiber, and copper concentrations were subsequently measured by GFAAS. The amount of copper measured in the acceptor solution was 14.0 ng at the lower and 0.86 ng at the higher PAR concentration, whereas the amount of copper extracted from the membrane was 1.34 ng at the lower and 23.6 ng at the higher PAR concentration, indicating that when high concentrations of PAR were used, the complex was sorbed to the membrane instead of staying in the acceptor solution. Based on these studies, a PAR concentration of 0.15 mM was selected as the optimal concentration for further experiments. The effect of the pH in the acceptor phase was studied between 4 and 10, and for each pH, five replicates were made. Figure 2 indicates that there was no significant extraction of copper at pHs lower than 6. Then Ee increased until a maximum between pH 8 and 9, which is the optimum range for the formation of the complex Cu-PAR, and after that it decreased. At pH 10, some drops of organic phase were observed in the acceptor solution, indicating that the organic phase was not stable at higher pH values. Besides, it should be noted that relative standard deviation (RSD) values are low, ranging from 3.1 to 9.5, and in the pH range 7-9, RSD values are lower than 5%. Finally, it should be mentioned that maximum wavelength and spectra did not change at different pH values. A pH of 8.7 was selected as optimum for the next experiments. 7608

Analytical Chemistry, Vol. 77, No. 23, December 1, 2005

Figure 3. Effect of agitation on the enrichment factor. Extraction conditions: source solution 50 µg/L Cu2+ at pH 7. Acceptor phase: 0.15 mM PAR previously dissolved in water and 0.1 M ammonium acetate at pH 8.7; extraction time, 30 min. Other conditions are the same as in Figure 1.

Likewise, the concentration of the buffer was also studied between 0.02 and 0.50 M. The enrichment factor increased slightly from 83.1 to 92.1 when buffer concentration changed from 0.02 to 0.05 M, and it did not increase further at higher buffer concentrations. A buffer concentration of 0.1 M was selected for further experiments. Finally, the stability of the PAR-Cu complex was determined by measuring the absorbance at 504 nm as a function of storage time. The results obtained demonstrated that the absorbance remained stable during at least 10 h, when the complex was stored at room temperature. This fact gives the analyst the chance of carrying out the extraction of the freely dissolved copper concentration in field and measuring the sample in the laboratory within 10 h after the sampling step. Study of the Shaking Speed and Equilibrium Sampling Time. One important parameter in HFSLM is the shaking speed during extraction. Shaking decreases the thickness of the unstirred boundary layer adjacent to the hollow fiber membrane, and this is known to be an efficient way to accelerate the extraction. Figure 3 illustrates the effect of the shaking speed on Ee after 30 min of shaking. This parameter increased with shaking speed until 150 rpm, reaching its highest values between 100 and 150 rpm. When shaking speed higher than 150 rpm was used, Ee decreased, probably due to the formation of air bubbles in the hollow fiber and the loss of the organic solvent, which can form microemulsions in the sample and in the acceptor phases.30 Therefore, 125 rpm was selected as the optimum shaking speed for further experiments. Analyte molecules need time for their mass transfer through two interfaces and the liquid membrane before reaching a thermodynamic equilibrium between sample and acceptor. Extraction kinetics were investigated under static conditions as well as at a shaking speed of 125 rpm by plotting Ee values against sampling time (Figure 4). The experimental values were also fitted to a first-order one-compartment uptake model (modified from ref 25):

Ee(t) ) D(1 - e-kt)

(2)

where k is the rate constant and D is the distribution coefficient (30) Wen, X.; Tu C.; Lee, H. K. Anal. Chem. 2004, 76, 228-232.

Figure 4. Extraction time profile with shaking speed 125 rpm (circles), and static (triangles). Extraction conditions: source solution 50 µg/L Cu2+ at pH 7. Acceptor phase: 0.15 mM PAR previously dissolved in water and 0.1 M ammonium acetate at pH 8.7. Other extraction conditions are the same as in Figure 1. Table 1. Results Obtained after the Application of a First-Order One-Compartment Uptake Model to the Extraction of Copper

shaking static a

D

sa

k (min-1)

t90% (min)

t95% (min)

t99% (min)

94.2 97.5

2.1 3.2

0.042 0.010

56 226

73 293

112 451

Standard deviation of D.

between the source solution and acceptor that is equivalent to Ee at equilibrium. Although this model is not a precise reflection of the extraction process, it is suited to fit the data and to estimate equilibrium sampling times (e.g., t95% ) ln(0.05)/(-k)). Results are presented in Table 1. Shaking enhanced the mass transfer by a factor of 4 leading to a reduction of the estimated equilibration time (t95%) from 293 to 73 min. Although the equilibration time could be seen as long, especially in the static mode, a large number of different samples can be extracted simultaneously, increasing the capacity of the method. An important observation is that there was no difference between the estimated distribution coefficients (D) under the two different conditions, which is in full accordance with the principles of ESTM.24 Sample Depletion. Equilibrium sampling devices should be applied in a manner that minimizes the depletion of the analyte concentration in the sample, which in turn minimizes the impact on equilibria between freely dissolved and bound analytes.25 Depletion can be considered to be negligible when the extracted amount is kept below 5% of the amount that was dissolved in the sample.22 The depletion of the present ESTM method was shown to be minimal. Experimental Ee values were 91, 94, and 91 at sample volumes of 100, 250, and 500 mL, and the extracted amount was only 0.86 (100 mL), 0.33 (250 mL), and 0.16% (500 mL) of the amount added to the sample. This indicates that, in all the cases, depletion was negligible, and the smallest sample volume of 100 mL was used in further studies. However, the described ESTM method can also be applied at higher sample volumes and it could also be used for passive field sampling, where the sample volume can be considered infinite. Performance Characteristics. Once the experimental conditions of the HFSLM method have been established, a number of performance parameters such as linearity, detection limit, precision, and selectivity were evaluated. The linearity of the method was tested within the range of 5-100 µg/L (solutions were

prepared using the procedure described in Chemicals and Materials section) and analyzing each level in triplicate. Using leastsquares regression, the absorbance at 504 nm (A504) was proportional to the freely dissolved copper concentration (Cu2+) over the studied range (A504 ) 0.035 + 0.038Cu2+) with a determination coefficient (R2) of 99.55%. The performance characteristics provided an analytical sensitivity, defined as the least variation of concentration that the analytical method is able to discern,31 of 1.9 µg/L, a linearity, expressed as relative standard deviation of slope, of 99.92, and a detection limit of 4.0 µg/L. This detection limit is only slightly higher than the detection limit of the recently reported HFSLM method with fluorescence detection.27 The overall precision of the method was evaluated by performing repeatability and reproducibility experiments by carrying out six extractions during one day at a level of 35 µg/L of the freely dissolved copper concentration (repeatability) and two replicates at three different days (reproducibility). The RSDs were calculated to be 3.95 and 5.64% for repeatability and reproducibility, respectively. Furthermore, an external calibration was used to evaluate the concentration of copper in the acceptor solution when different copper concentrations were tested in the sample (range from 5 to 100 µg/L). The calculated acceptor concentrations were plotted versus sample concentrations, and least-squares regression was applied. In this case, the slope should give an estimation of Ee, over the range of studied concentrations in the sample. The obtained slope had a value of 93.3 ( 1.8, in accordance with the values obtained previously as well as with D. This means that Ee is constant within the concentration range studied Finally, the selectivity of the HFSLM method was evaluated by studying a number of metal ions as potential interferences for the determination of copper. Although the selectivity of the applied membrane had already been demonstrated in an earlier study,19 where GFAAS was used as detection technique, we thought that this study should be done again because the detection system has been changed, and now it could be less selective than GFAAS. For this purpose, different concentrations of Ni2+, Co2+, Cd2+, Fe3+, Pb2+, Zn2+, and Mn2+ (all of them can form stable complexes with PAR) were added to solutions without copper and solutions containing 50 µg/L copper. The concentrations of the selected cations ranged from 10 to 1000 µg/L, and after the extraction, the absorbance of the acceptors were measured. The data indicate that the different cations were not extracted (enrichment factor for all the cations equal to zero) and they did not affect the enrichment of copper (the slight variations of Ee can be attributed to experimental error). This clearly demonstrates the high selectivity of the developed ESTM technique due to the high selectivity of the transport through the membrane. Validation and Application of the Developed Method. Freely dissolved copper concentrations in different synthetic samples were measured using the calibration graph obtained previously, and these concentrations were then compared to calculated concentrations. The samples contained phthalic acid, EDTA, humic acids, catechol, and salicylate (see Table 2). The program Cheaqs V L20.1 was used for the ligands phthalic acid, EDTA, catechol, and salicylate, and the program WinHumic V (31) Cuadros-Rodrı´guez, L.; Garcı´a-Campan ˜a, A. M.; Bosque-Sendra, J. M. Anal. Lett. 1996, 29, 1231-1239.

Analytical Chemistry, Vol. 77, No. 23, December 1, 2005

7609

Table 2. Composition of Different Sample Solutions Used To Study the Effect of Different Ligands on Copper Extraction source sample

ligand

Conligand

pH

total Cu (µg/L)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19

humic acids humic acids humic acids humic acids phthalic acid phthalic acid phthalic acid EDTA EDTA EDTA EDTA catechol catechol catechol catechol salicylate salicylate salicylate salicylate

2.5 mg/L 1 mg/L 0.5 mg/L 0.1 mg/L 10-5 M 10-4 M 10-3 M 10-8 M 10-7 M 10-6 M 5 × 10-6 M 10-7 M 5 × 10-7 M 10-6 M 10-5 M 10-6 M 10-5 M 5 × 10-5 M 10-4 M

6.2 7.7 7.1 7.5 7.6 7.0 6.2 7.6 7.5 7.2 6.9 7.7 7.6 7.2 6.9 6.3 6.4 7.4 6.6

300 10 100 50 25 50 100 40 75 10 400 30 200 60 90 50 80 150 20

Table 3. Analysis of Freely Dissolved Copper Concentration in Synthetic Samples

pH total Cu (µg/L) Ni(II) (µg/L) Co(II) (µg/L) Fe(III) (µg/L) Pb(II) (µg/L) Zn(II) (µg/L) Cd(II) (µg/L) Mn(II) (µg/L) humic acids (mg/L)) phthalic acids (M) EDTA (M) catechol (M) salicylate (M) free Cu2+theoretical (µg/L) free Cu2+experimental (µg/L) s (n ) 5) p-value

S. water 1

S. water 2

S. water 3

7.4 200

7.5 100 50

7.9 50 100 40

10 40 75 10 25 1

100 100 300 50 10-4 10-7

61.9 58.8 3.8 0.14

32.8 31.4 2.5 0.28

10-7 10-6 12.4 12.2 2.0 0.87

Table 4. Analysis of Freely Dissolved Copper Concentrations in Leachate Waters

leachate influent effluent

pH

free Cu2+ (µg/L)a

total copper (µg/L)b

ratio (%)c

8.03 7.90 7.60

4.9 (3.2)4d 16.3 (4.0)4 8.7 (1.9)4

798 (4.5)5 259 (5.1)6 64 (15.5)6

0.6 6.3 13.7

a Measured using the proposed spectrophotometric method. b Measured by GFAAS. c Ratio between the freely dissolved and total copper concentration. d Relative standard deviation in percent. Subscript indicates the number of replicates.

Figure 5. Free Cu2+ concentrations measured by HFSLM extraction technique against concentrations estimated using Cheaqs and WinHumic Software. (s): line with slope 1 and intercept 0. Extraction conditions: acceptor phase, 0.15 mM PAR previously dissolved in water and 0.1 M ammonium acetate at pH 8.7; extraction time, 75 min; shaking rate, 125 rpm. Other extraction conditions are the same as in Figure 1.

was used for humic acids. The measured concentrations were plotted against the calculated concentrations (Figure 5), and a linear regression gave a slope of 0.984, an intercept of 0.453, and a correlation coefficient of 0.991. A joint confidence interval32 was used to evaluate that the method is not affected by systematic error (intercept equal to 0 and slope equal to 1). The results, Fcal ) 0.16 and Ftab (ν)2,17; R ) 0.05) ) 3.59, indicate that the HFSLM method can be used for the determination of the freely dissolved copper concentrations in samples containing the ligands tested. Furthermore, the reliability of the proposed method was evaluated by applying it to three synthetic samples containing copper together with cations and ligands frequently accompanying copper. Measured concentrations were in good agreement with theoretically calculated concentrations with deviations of only 2-5% (Table 3). These deviations can solely be explained by the (32) Hartmann, C.; Smeyers-Verbeke, J.; Penninckx, W.; Massart, D. L. Anal. Chim. Acta 1997, 338, 19-40.

7610 Analytical Chemistry, Vol. 77, No. 23, December 1, 2005

measurement error, since a t-test confirmed that there was no statistically significant difference between measured and theoretically calculated concentrations (p > 0.05). These results confirm that the proposed method is suitable for the determination of freely dissolved copper concentrations. Finally, the method was employed to measure the speciation of copper in three environmental samples. Freely dissolved copper concentrations were measured in quadruple using the proposed spectrophotometric method, and total copper concentrations were measured by GFAAS after acidifying the sample to pH 1 and stirring for 1 h. Table 4 shows the obtained results. The “leachate” sample was taken from a point where leachate directly comes out from a landfill. The leachate is led to a storage pond and then to a treatment system. The “influent” sample was taken after the storage pond and before a filter treatment system, and the “effluent” sample was collected after the treatment system. It can be observed that copper was most strongly complexed in the leachate, where less than 1% of the total amount is freely dissolved. It should also be noted that the amount of copper decreased during the treatment, whereas the freely dissolved fraction increased, partly because the pH in the effluent was lower than in the influent. CONCLUSION A selective enrichment method for the spectrophotometric determination of the freely dissolved copper concentration in environmental waters has been developed. The method is based

on the principles of equilibrium sampling through membranes, and it offers important advantages such as simplicity, reproducibility, and elimination of interferences. The methodology is rather inexpensive compared to other techniques, since it is based on hollow fibers, common reagents, and spectrophotometric detection. The present study has shown that the technique can be applied to discrete environmental samples, and we expect it to be further applicable as passive field samplers. Due to the simplicity and the low cost of the sampling device, the hollow fiber can be discarded after each extraction to avoid carryover effects. The technique has been successfully applied to “dirty” samples such as leachate waters, indicating it to be compatible with different types of environmental and biological samples and matrixes. In this respect, it might be applicable not only as sample enrichment technique but also as a sample cleanup procedure.

It is interesting to note that the sensitivity of the method in terms of detection limit can be improved using cuvettes or flow cells with longer path lengths. Besides, it should be commented that ESTM of copper could be combined with other detection techniques such as fluorescence, ICP or GFAAS. ACKNOWLEDGMENT The authors gratefully acknowledge “MISTRA”, Foundation for Strategic Environmental Research (Sweden), for its financial support. R.R. is also grateful to the Secretarı´a de Estado de Eduacio´n y Universidades (Spain) & European Social Fund for a postdoctoral fellowship. We thank Andreas Christensson for his help on spectrophotometric measurements. Received for review May 4, 2005. Accepted September 22, 2005. AC050763A

Analytical Chemistry, Vol. 77, No. 23, December 1, 2005

7611