Anal. Chem. 2004, 76, 4315-4319
Selective Mercury Determination after Membrane Complexation and Total Reflection X-ray Fluorescence Analysis Pavlos E. Koulouridakis and Nikolaos G. Kallithrakas-Kontos*
Technical University of Crete, Analytical and Environmental Chemistry Laboratory, University Campus, GR-73100 Chania, Greece
A new method for low mercury concentration analysis in drinking waters is presented. Membranes containing a few micrograms of various complexing reagents were produced on the surface of quartz glasses (reflectors). The reflectors were immersed in water solutions containing various concentrations of inorganic mercury salts at low concentrations (1-40 ng/mL). The membranes were left to equilibrate in 5-500 mL of solution for many hours; when the equilibration stage was finished they were cleaned with ultrapure water and left to dry. Analyses were performed by total reflection X-ray fluorescence (TXRF). The effects of various experimental parameters (complexing agent, equilibrium time, sample volume, etc.) as well as the selectivity of the membranes were studied. The complexing reagent dithizone with a PVC-based membrane gave the best results. The limit of quantitation was 0.8 ng/mL. Mercury is a toxic element that produces many health problems. Mercury arises from natural or anthropogenic sources. The concern for a clean environment has prompted the creation of sensitive control methods for mercury detection. Minimal dangerous levels have been established at the limit of 1 ng/mL, according to the World Health Organization (WHO).1 The detection of mercury in environmental samples can today be achieved by several methods with different detection limits.2-7 The most important problems associated with mercury analysis are related to its volatile nature, the need for low method detection limits, and the need for preconcentration steps. In the present publication a method for detection of mercury in drinking water with the use * Corresponding author. Phone: +30 2821 037666. Fax: +30 2821 037841. E-mail:
[email protected]. (1) Merian, E. Metals and their Compounds in the Environment; VCH: Weinheim, Germany, 1991; p. 701. (2) Manganiello, L.; Rı´os, A.; Varca´rcel, M. Anal. Chem. 2002, 74, 921-925. (3) Shaw, M. J.; Jones, P.; Haddad, P. R. Analyst 2003, 128, 1209-1212. (4) SanVicente de la Riva, B.; Costa-Ferna´ndez, J. M.; Jun Jin, W.; Periero, R.; Sanz-Medel, A. Anal. Chim. Acta 2002, 455, 179-186. (5) De Wuilloud, J. C. A.; Wuilloud, R. G.; Silva, M. F.; Olsina, R. A.; Martinez, L. D. Spectrochim. Acta, Part B 2002, 57, 365-374. (6) Madden, J. E.; Cardwell, T. J.; Cattral, R. W.; Deady, L. W. Anal. Chim. Acta 1996, 319, 129-134. (7) Zachariadis, G. A.; Anthemidis, A. N.; Karpouzi, M.; Stratis, J. A. Proceedings of the 3rd International Conference. Instrumental Methods of Analysis Modern Trends and Applications, Thessaloniki, Greece, September 2327, 2003; Stratis, J., Ed.; Ziti: Thessaloniki, Greece; p 7. 10.1021/ac049780a CCC: $27.50 Published on Web 06/29/2004
© 2004 American Chemical Society
of total reflection X-ray fluorescence (TXRF) is proposed. TXRF is an economical (low cost per sample) and reliable method for the analyses of trace elements, and it is used in many laboratories all over the world.8,9 The significant advantages are the following: (a) It can detect exceptionally small mass quantities (ng or pg). (b) It is multielemental; this means that it can analyze simultaneously (in one spectrum) almost all the elements with an atomic number higher than that of magnesium. (c) It has the ability to be used for the direct analysis of solid deposits (like membranes). (d) The analysis time is small (typically 100400 s). In the usual TXRF analysis a small quantity of solution (a few microliters) is deposited on the center of a quartz reflector. The liquid is desiccated and analyzed; under these conditions the detection limits are of the order of 50 ng/mL, much higher than the WHO-established limit. Such a procedure is unsuitable for the determination of mercury. In the present work new membranes were prepared to analyze low mercury concentrations in natural water solutions. The membranes were placed on the center of quartz reflectors, and they contained powerful complexing reagents as well as other auxiliary compounds. They were used to collect trace elements from various water solutions of volumes from 5 to 500 mL. The use of complexing reagents was chosen because of their ability to create stable complexes with the analyzed trace element. The evaluation of this ability was determined from the stability constants. These constants were found in two of the more reliable databases, the International Union of Pure and Applied Chemistry (IUPAC) SC Database10 and the corresponding database of the National Institute of Standards and Technology (NIST).11 Commercially available ligands with very high stability constants were selected. Two types of membranes were produced. In the first, complexing reagents were dissolved in a Nafion solution, whereas membranes from the second type were produced by mixing complexing reagents with the following components: poly(vinyl (8) Van Grieken, R.; Marcowicz, A. A. Handbook of X-ray Spectrometry, 2nd ed.; Marcel Dekker: New York, 2001. (9) Klockenkamper, R. Total Reflection X-ray Fluorescence; Wiley-Interscience: New York, 1996. (10) IUPAC Stability Constants Database, SC-Database, and Mini-SCDatabase; Academic Software: Sourby Old Farm, Timble, Otley, Yorks, LS21 2PW, UK. (11) Martell, A. E.; Smith, R. M. NIST Critically Selected Stability Constants of Metal Complexes, NIST Standard Reference Database 46, version 6; U.S. Government Printing Office: Gaithersburg, MD, 2001.
Analytical Chemistry, Vol. 76, No. 15, August 1, 2004 4315
chloride) high molecular weight, dibutyl phthalate plasticizer, tetrahydrofuran, and DTBNBA. The membrane composition was selected to be similar to that of ion selective electrode membranes.12 The first type of membrane is called a “Nafion” membrane, whereas the second is called a “PVC” membrane. In a previous paper13 we used a Nafion-type membrane for the TXRF analysis of Pb. Sulfur and chlorine are included in the Nafion and PVC molecules, respectively; these elements were the only membrane matrix elements that can be detected by TXRF, but their K X-ray energies are far enough from the mercury X-ray lines so that no interference effects were produced. The cleaning procedure after the use of the reflectors was simple, as the membranes were easily removed from the quartz surface. The quantities of the deposited membranes were very small as well as their cost, and no regeneration or other time-consuming procedures were needed. EXPERIMENTAL SECTION PVC-Based Membranes. Two types of solutions were prepared. The first was prepared by dissolution of 3 mg of ligand in 2 mL of Milli-Q water. The ligands used were the following: citric acid (Mallinckrodt granular no. 0627), dithizone (Fluka Chemica no. 43820), HEDTA (Fluka Biochemica no. 54215, N-(2-hydroxyethyl) ethylenediamine-,N′,N′-triacetic acid), thiourea (Riedel-de Hae¨n no. 33717), 18ane (Chemica no. 52045, 1,4,7,10,13,16hexaazacyclo-octadecane). All ligands were in solid form. After their dissolution all the solutions were colorless except that with the dithizone ligand (black colored). The second solution included the following components: PVC 20 mg, dibutyl phthalate plasticizer 25 mg, tetrahydrofuran (THF) 5 mL, and 5,5dithiobis (DTBNBA) 5 mg. The solutions were heated in an ultrasonic bath to accelerate the total dissolution. Membrane immobilization on quartz reflectors was performed as follows: 6 µL of the first type of solution and 10 µL of the second solution were transferred to the center of the quartz glass, giving a 10-mm diameter spot; the membranes were left to dry at room temperature. The use of an IR lamp or an oven in the desiccation process has no effect on the membrane adhesion or the mercury detection capability. Nafion-Based Membranes. These were prepared by dissolution of 3 mg of ligand in 2 mL (1.748 g) of Nafion solution 5% (Aldrich Chemicals catalog no. 27,470-4). The ligands were the same as those in the PVC-based membranes. In the center of each quartz reflector was placed 6 µL of solution. The drying processes were followed as with the previous membrane. No mercury traces were determined in either the PVC- or the Nafion-based membranes. Mercury solutions were prepared at a concentration of 1000 mg/L (Reagecon stock solutions in molar HNO3). For solutions above 20 mL in volume, plastic vessels of various capacities were used. The reflectors with the membrane were put in those vessels and remained there for various equilibration times. For solutions that were 6 mL in water volume, reflectors were placed in small plastic cups (Chemplex XRF sample cups catalog no. 1340), and the bottoms of these were covered with Mylar foil. At the end of each process, the reflectors came out of the solutions and were (12) Rouhollahi, A.; Ganjali, M. R.; Shamsipur, M. Talanta 1998, 46, 13411346. (13) Koulouridakis, P. E.; Domazos, E. A.; Galani-Nikolakaki, S. M.; KallithrakasKontos, N. G. Microchim. Acta, in press.
4316
Analytical Chemistry, Vol. 76, No. 15, August 1, 2004
left to dry. The yield of each reflector had very good repetition in each measurement. Analysis was performed by TXRF. A fine focus molybdenum X-ray tube (Seifert FK 60-04 AG) with a Seifert 150-Debyeflex 3000 high-voltage generator was used for the primary X-ray beam production, at a voltage of 55 kV and current of 20 mA. The X-ray total reflection was performed by an XRF module produced at the Wien Atom Institute. The produced X-rays were detected by an Oxford Si (Li) X-ray semiconductor detector with an 80 mm2 surface area and a resolution of 155 eV at 5.9 keV. The produced signal was amplified by the detector preamplifier and a Tennelec Tc-244 spectroscopy amplifier and collected with a personal computer multichannel analyzer card (PCA-II Nucleus). A total of 2048 channels were used for every spectrum, and the analysis time was 300 s. Peak integration and background subtraction were performed by the computer program AXIL14 distributed by the International Atomic Energy Agency (IAEA). RESULTS AND DISCUSSION Ligand Selection. Reflectors with membranes of various ligands were put in 500 mL of water solution for 1 day. This solution contained 10 ng/mL of mercury as well as 150 mg/L of calcium and 30 mg/L of potassium, in order to reproduce a high salt content water. Calcium and potassium were preferred instead of magnesium and sodium because they are detectable by TXRF. Figure 1 represents the mercury, calcium, and potassium X-ray yields as a function of the membrane type and the complexing reagents. There are two categories of membranes, those containing PVC + ligand and those with Nafion + ligand. The mercury yield of the pure PVC and pure Nafion membranes gave the lowest yields, compared to the same polymer membranes containing ligands. PVC membranes were superior to the Nafion ones for mercury yield. Dithizone and thiourea gave the best Hg yield results for both membrane types; among them the highest yield was achieved with the dithizone-PVC membrane. The calcium yields were higher in Nafion membranes compared with PVC. This accumulation is undesirable as high calcium and potassium content may block the membrane active centers and limit the mercury collection. The accumulation is undesirable also because it can lead to rather strong fluorescence signals and the collection capacity (count rate) of the detector can be saturated. On the other hand, PVC membranes gave much lower calcium yields; among them the best results were achieved by PVC-dithizone and PVC-citric acid combinations. As PVCdithizone also gave the best results for mercury collection, it was selected as the most suitable membrane, and all the other experiments are referred to this one. The effect of volume on mercury yield was examined in 10 ng/mL water samples. PVC-dithizone membranes were placed in solutions with volumes of 3, 6, 25, 50, 100, and 500 mL. The quantitative Hg yield determination after 24 h of equilibration is given in Figure 2. Increasing the solution volume increases the total mercury mass in the solution, and as a result the mercury yield grows. The linearity between the mercury yield and the solution volume was very good in the 0-50 mL volume range; for higher volumes (100-500 mL) the linearity ceased. (14) Vekemans, B.; Janssens, K.; Vincze, F.; Adams, F.; Van Espen, P. Spectrochim. Acta, Part B 1995, 50, 149-169.
Figure 1. Mercury, calcium, and potassium X-ray yields as a function of membrane composition. Calcium yields have been divided by 10 for better presentation. The results for mercury yields are presented with black columns, calcium yields with white columns, and potassium yields with columns containing horizontal lines.
Figure 2. Effect of solution volume on mercury yield (Hg concentration 10 ng/mL, PVC-dithizone membrane, equilibration time 24 h). Linearity exists up to 50 mL of solution volume.
Figure 3 represents the mercury yield of the membrane as a function of equilibration time in water with 10 ng/mL of added mercury. Sample volumes of 6 mL and 500 mL of water were used. As higher yields have been achieved with longer equilibration times, the saturation is attributed to the difficulty of Hg molecules to come in contact with the membrane in higher volumes. Thus, as seen in Figure 3, for 6 mL the maximum yield reached its final stage the first day, for 500 mL a small increase prevailed after the first day that continued both the second day and at smaller percentage the third day. After 24 h of equilibration, the mercury yield reached 95% of its final yield in both volumes; after 2 days of equilibration time the yield was practically stabilized total recovery. As the rate of mass reduction of mercury is proportional to the mercury mass in the solution, the absorbed mass of mercury in the membrane follows an exponential increase of the form y ) a(1 - e-bx).13 According to Figure 3, higher yields were achieved in 500 mL of mercury solution than in 6 mL; so the yield plateau in 6 mL of solution is due to mercury exhaustion in this solution.
The mercury total recovery from 6- and 50-mL volume water solutions was cross evaluated by the immersion of a second membrane in the solution after the 2-day equilibration stage of the first membrane; the detected mercury in this second membrane was at null levels. The effect of different types of water on the mercury yield was examined in Figure 4. Four different solutions were prepared with a 50-mL volume for each one. The first one was Milli-Q water; the second was Milli-Q water with 10 ng/mL of mercury; the third solution was Milli-Q water with 10 ng/mL of mercury, 50 mg/L of calcium, and 5 mg/L of potassium. Finally, the fourth solution was local bottled water (calcium content 30.5 mg/L) in which 10 ng/mL of mercury was added. Into each solution was placed one reflector with the dithizone-PVC membrane. The yields of the samples were measured after 24 h of equilibration time, and they are showed in Figure 4. Potassium and calcium ions did not influence mercury yield in the third solution. The same result exists for the bottled water solution. Analytical Chemistry, Vol. 76, No. 15, August 1, 2004
4317
Figure 3. Mercury X-ray yields as a function of equilibration time (Hg concentration 10 ng/mL, PVC-dithizone membrane, solution volume 500 and 6 mL). The 500-mL solution yields have been divided by 10 for better presentation in the same figure.
Figure 4. Influence of potassium and calcium ions on mercury yields (Hg concentration 10 ng/mL, PVC-dithizone membrane, equilibration time 24 h, solution volume 50 mL) for four different solutions.
Figure 5. Mercury yields as a function of the zinc concentration in the solution.
The effect of zinc was studied as its concentrations may take relatively high values in natural waters.1 An amount of 50 mg/L of calcium, 5 mg/L of potassium, and various zinc concentrations (0.5, 1, 5, and 50 mg/L) was dissolved in Milli-Q water. The WHO limit for the zinc concentration in drinking water is 5 mg/L.1 The results are given in Figure 5 where the mercury LR and zinc KR X-ray yields are given as a function of zinc concentration. As is obvious from this figure, mercury yields are not affected even at high zinc concentrations and the membrane gave the expected 4318
Analytical Chemistry, Vol. 76, No. 15, August 1, 2004
mercury yield. The effect of pH on mercury yield was studied with HNO3 acidification in the pH range of 4-7, and it was found to be stable. The adhesion of the membrane was very good at all pH values. The linearity of the X-ray mercury yields was examined for 1, 5, 10, 20, and 40 ng/mL concentrations and an equilibration time of 1 day. The results gave very good linearity (R2 ) 0.99). The minimum detection limit was estimated as 3 times the background standard deviation (square root of the background15).
For 50 mL of sample volume and 300 s of irradiation time it was found to be equal to 0.2 ng/mL. The limit of quantitation was equal to 0.8 ng/mL, which is lower than 1 ng/mL, the WHO limit for drinking water.
• The minimum detection limit is 0.2 ng/mL, and the limit of quantitation is 0.8 ng/mL for a 300 s irradiation time.
CONCLUSIONS The detection of mercury in drinking water can be achieved at ng/mL concentrations with the use of membrane complexation and TXRF detection. • The dithizone-PVC combination gave the best results among the membranes examined. The adhesion of the membranes was very good. • Good linearity exists in the 1-40 ng/mL mercury concentration range. • Complete uptake of mercury is achieved in 1-2 days. • There are no interference problems from ions usually present in natural waters. The solution pH does not influence the results over a wide pH range.
Thanks are expressed to the European Union and to the Greek Ministry of Education for the financial support of this research (Iraklitos Fellowships for research of Technical University of CretesEnvironment (2.6) EPEAEK II MIS 88744).
(15) Jenkins, R. X-ray Fluorescence Spectrometry; Winefordner, J. D., Kolthoff, I. M., Eds.; Chemical Analysis A Series of Monographs on Analytical Chemistry and its Applications, Vol. 99; Wiley-Interscience: New York, 1988.
ACKNOWLEDGMENT
SUPPORTING INFORMATION AVAILABLE Typical mercury spectrum from a PVC-dithizone membrane that was placed in a 10 ng/mL mercury solution and X-ray mercury yields as a function of concentration. This material is available free of charge via the Internet at http://pubs.acs.org.
Received for review February 9, 2004. Accepted May 27, 2004. AC049780A
Analytical Chemistry, Vol. 76, No. 15, August 1, 2004
4319
