Stir Bar Sorptive Extraction for the Determination of ppq-Level Traces

After 15 min of extraction, the stir bar was desorbed in a thermal desorption unit .... structure and constant temperature along the whole length of t...
17 downloads 0 Views 69KB Size
Download PDF
Anal. Chem. 2001, 73, 1509-1514

Stir Bar Sorptive Extraction for the Determination of ppq-Level Traces of Organotin Compounds in Environmental Samples with Thermal Desorption-Capillary Gas Chromatography-ICP Mass Spectrometry Jordy Vercauteren,*,†,§ Christophe Pe´re`s,§ Christophe Devos,§ Pat Sandra,§ Frank Vanhaecke,† and Luc Moens†

Laboratory of Analytical Chemistry, Ghent University, Proeftuinstraat 86, B-9000 Ghent, Belgium, and Laboratory of Organic Chemistry, Ghent University, Krijgslaan 281, B-9000 Ghent, Belgium

The extraction and preconcentration capabilities of a new extraction technique, stir bar sorptive extraction, were combined with the separation power of capillary gas chromatography (CGC) and the low limits of detection (LODs) of inductively coupled plasma mass spectrometry (ICPMS) for the determination of the organotin compounds tributyltin (TBuT) and triphenyltin (TPhT) in aqueous standard solutions, harbor water, and mussels (after digestion with tetramethylammonium hydroxide). Throughout, tripropyltin for TBuT and tricyclohexyltin for TPhT were used as internal standards to correct for variations in the derivatization and extraction efficiency. Calibration was accomplished by means of single standard addition. Derivatization to transform the trisubstituted compounds into sufficiently volatile compounds was carried out with sodium tetraethylborate. The compounds were extracted from their aqueous matrix using a stir bar of 1-cm length, coated with 55 µL of poly(dimethylsiloxane) (PDMS). After 15 min of extraction, the stir bar was desorbed in a thermal desorption unit at 290 °C for 15 min, during which the compounds were cold-trapped on a precolumn at -40 °C. Flash heating was used to rapidly transfer the compounds to the GC where they were separated on a capillary column with a PDMS coating. After separation, the compounds were transported to the ICP by means of a homemade heated (270 °C) transfer line. Monitoring of the 120Sn+ signal by ICPMS during the run of the GC provided extremely low LODs for TPhT in water: 0.1 pg L-1 (procedure) and 10 fg L-1 (instrumental) and a repeatability of 12% RSD (n ) 10). In harbor water, concentrations of 200 pg L-1 for TBuT and 22 pg L-1 for TPhT were found. In fresh mussels, a concentration of 7.2 ng g-1 (dry weight) TPhT was found. The accuracy of the method was checked by the determination of TPhT in CRM477 (mussel tissue) and comparison of the result to that of an analysis of the same material with a classical liquid/liquid extraction with isooctane. The interest in speciation has grown significantly over the past few years due to the awareness that many organometallic 10.1021/ac000714s CCC: $20.00 Published on Web 03/06/2001

© 2001 American Chemical Society

compounds are considerably more toxic than the corresponding free metals. As far as organotin (OT) compounds are concerned, it is well known that the trisubstituted species, and especially tributyltin (TBuT) and triphenyltin (TPhT), are the most toxic. TBuT has been widely used for many years as an antifouling compound, added to paints intended for boats, and TPhT is still being frequently used as a fungicide in agriculture, mainly against potato blight (Phytopthora infestans). Although the use of TBuT has now been drastically restricted, it is still allowed on larger boats and desorption of OTs from contaminated sediments could be an important source of future OT pollution.1 Therefore, it is very important to have a method that allows fast detection in the subnanogram per liter range. In earlier studies, OTs were often extracted using tropolone and n-hexane and determined with GCflame photometric detection (FPD) after a Grignard derivatization.2-8 In more recent work, the Grignard derivatization is replaced by in situ ethylation with sodium tetraethylborate9-12 and FPD is sometimes replaced by atomic emission detection (AED).10,12 Quite recently, solid-phase microextraction (SPME) was introduced as an elegant and practicable extraction technique for volatile OTs13 and semivolatile OTs,14 in both cases combined with capillary gas * Correcponding author. E-mail: [email protected]. † Laboratory of Analytical Chemistry. ‡ Research Assistant of the Fund for Scientific ResearchsFlanders. § Laboratory of Organic Chemistry. (1) Weidenhaupt, A.; Arnold, C.; Mu ¨ ller, S. R.; Haderlein, S. B.; Schwarzenbach, R. P. Environ. Sci. Technol. 1997, 31, 2603-2609. (2) Caricchia, A. M.; Chiavarini, S.; Cremisini, C.; Morabito, R.; Scerbo, R. Anal. Chim. Acta 1994, 286, 329-334. (3) Fent, K.; Hunn, J. Environ. Sci. Technol. 1991, 25, 956-963. (4) Harino, H.; Fukushima, M.; Tanaka, M. Anal. Chim. Acta 1992, 264, 9196. (5) Gomez-Ariza, J. L.; Morales, E.; Ruiz-Benitez, M. Analyst 1992, 117, 641644. (6) Nagase, M.; Hasebe, K. Anal. Sci. 1993, 9, 517-522. (7) Mu ¨ ller, M. D. Anal. Chem. 1987, 59, 617-623. (8) Van de Broek, H. H.; Hermes, G. B. M.; Goewie, C. E. Analyst 1988, 113, 1237-1239. (9) Følsvik, N.; Brevik, E. M. J. High Resolut. Chromatogr. 1999, 22, 177-180. (10) Ceulemans, M.; Witte, C.; Lobinski, R.; Adams, F. C. Appl. Organomet. Chem. 1994, 8, 451-461. (11) Morcillo, Y.; Porte, C. Trends Anal. Chem. 1998, 17, 109-116. (12) Lobinska, J. S.; Ceulemans, M.; Lobinski, R.; Adams, F. C. Anal. Chim. Acta 1993, 278, 99-113.

Analytical Chemistry, Vol. 73, No. 7, April 1, 2001 1509

chromatography-inductively coupled plasma mass spectrometry (CGC-ICPMS). Although classical liquid/liquid extraction is still being used very often, it suffers from low sensitivity, because only a few microliters of the total extract are being used. Over the last two decades, several techniques applying poly(dimethylsiloxane) (PDMS) have allowed transfer of the total amount of the compounds extracted to the analytical instrument (often GC or LC). One of the first approaches using PDMS for extraction and preconcentration was open tubular trapping (OTT).15,16 Briefly, with this approach, a sample (water, air) is pumped through a capillary (similar to a capillary gas chromatography column) coated with a thin layer of PDMS (5-100 µm).16-18 Sampling is normally performed until any of the analytes of interest is no longer fully retained by the PDMS (breakthrough sampling). Desorption is performed either by heating or extraction with an organic solvent. Despite the advantages of this technique, it never gained widespread acceptance due to the limited sampling capacity. About 10 years ago, Arthur and Pawliszyn introduced a new technique applying PDMS for sample enrichment: solidphase microextraction.19 In this approach, a PDMS layer is coated onto the outside surface of a needle of a syringelike device. The needle can be inserted directly into the aqueous sample or into the headspace above it. Although not a real necessity, sampling is often performed until equilibrium is reached. An important advantage of SPME is the possibility of desorption directly into the analytical instrument. Due to its simplicity, SPME has gained a lot of interest and its only drawback is the small volume of the PDMS coating used (e0.5 µL), which implies a large (sample/ PDMS) phase ratio, hence allowing only compounds with a very high octanol-water partitioning coefficient (Ko/w >20 000 for a 10-mL sample) to be extracted with a high efficiency. About three years ago, a technique based on extraction/preconcentration of analyte compounds on a short bed packed with PDMS particles was introduced by Baltussen et al.20,21 The packed bed contains ∼300 µL of PDMS, resulting in a higher sensitivity compared to OTT and SPME. Sampling is performed until breakthrough or, for even higher sensitivity, until all analytes are in equilibrium with the sorbent. This technique can be applied for gaseous and liquid samples, although for the latter drying is required, which induces a loss of volatile compounds, thereby limiting its application range in aqueous samples to nonvolatile compounds. Very recently, a technique combining the sensitivity of packed PDMS beds with the application range (in terms of volatility) of SPME was introduced.22 Stir bar sorptive extraction (SBSE) applies stir (13) Moens, L.; De Smaele, T.; Dams, R.; van den Broek, P.; Sandra, P. Anal. Chem. 1997, 69, 1604-1611. (14) Vercauteren, J.; De Meester, A.; De Smaele, T.; Vanhaecke, F.; Moens, L.; Dams, R.; Sandra, P. J. Anal. At. Spectrom. 2000, 15, 651-656. (15) Burger, B. V.; Munro, Z. M. J. Chromatogr. 1986, 370, 449-464. (16) Bicchi, C.; D’Amato, A.; David, F.; Sandra, P. J. High Resolut. Chromatogr. 1989, 12, 316-321. (17) Roeraarde, J.; Blomberg, A. J. High Resolut. Chromatogr. 1989, 12, 139. (18) Burger, B. V.; LeRoux, M.; Burger, W. J. G. J. High Resolut. Chromatogr. 1990, 13, 777-779. (19) Arthur, C. L.; Pawliszyn, J. Anal. Chem. 1990, 62, 2145-2148. (20) Baltussen, E.; Janssen, H.-G.; Sandra, P.; Cramers, C. A. J. High Resolut. Chromatogr. 1997, 20, 385-393. (21) Baltussen, E.; Janssen, H.-G.; Sandra, P.; Cramers, C. A. J. High Resolut. Chromatogr. 1997, 20, 395-399. (22) Baltussen, E.; Sandra, P.; David, F.; Cramers, C. J. Microcolumn Sep. 1999, 11, 737-747.

1510 Analytical Chemistry, Vol. 73, No. 7, April 1, 2001

Figure 1. Schematic representation of a stir bar applied for SBSE.

bars, varying in length from 1 to 4 cm, coated with a relatively thick layer of PDMS (0.3-1 mm), resulting in PDMS volumes varying from 55 to 220 µL. The stir bar is added to an aqueous sample for stirring and extraction, and after a certain stirring time, the bar is removed from the aqueous sample and thermally desorbed into a gas chromatograph. Due to the much larger volume of the PDMS-phase extraction efficiency is far better than for SPME. SBSE has been applied for the determination of volatile and semivolatile compounds in beverages.23 The aim of this work was to combine the sensitivity of SBSE with the unsurpassed detection power of CGC-ICPMS for organometallic compounds and to apply this combination for the determination of trace amounts of organotin compounds in environmental samples. Although TPhT was the main compound of interest, TBuT was also determined in certain samples. Since confusion is unlikely to occur, triphenyltin hydroxide and the derivatized compound triphenylethyltin will both be abbreviated as “TPhT” throughout this work. The same applies for tricyclohexyltin (“TCT”), “TBuT”, and tripropyltin (“TPrT”). EXPERIMENTAL SECTION Instrumentation. Stir Bar Sorptive Extraction. The preparation of the stir bars has been described in detail elsewhere.22 Stir bars of ∼1 cm in length containing ∼55 µL of PDMS were used. PDMS-coated stir bars are also commercially available from Gerstel GmbH (Mu¨llheim a/d Ruhr, Germany) under the trade name Twister. Stir bars are conditioned at 300 °C for 4 h. For all extractions, 40-mL closed-cap vials with Teflon-coated silicone rubber septa, both from Alltech Associates (Lokeren, Belgium), were used. A schematic representation of a PDMS-coated stir bar is presented in Figure 1. Thermal desorption (TD). A modified thermodesorption unit from Chrompack (Middelburg, The Netherlands) was used. After extraction, the stir bar is removed from the sample by means of a big magnetic stir bar, transferred into a glass thermal desorption tube, and desorbed for 15 min at 290 °C. A He flow of ∼100 mL min-1 is passed through the glass tube during desorption. Compounds are trapped on a deactivated fused silica precolumn (15 cm, 0.53-mm i.d.) obtained from Supelco (Bellefonte, PA) that is cooled to -40 °C using liquid nitrogen. Subsequently, flash heating (15 °C s-1) up to 290 °C is applied to the precolumn for a rapid transfer of the compounds into the GC. To make sure (23) Sandra, P.; Baltussen, E.; David, F.; Tredoux, A.; Hoffmann, A., submitted for publication.

Figure 2. Schematic representation of the desorption unit. Table 1. Instrumental Parameters for GC-ICPMS gas chromatograph column injection system oven temperature program carrier gas/inlet pressure transfer line/temperature ICPMS Rf power carrier gas auxiliary gas plasma gas dwell time

HP 5890 series II (Hewlett-Packard) FSOT, DB-1 (100% PDMS), 30 m, i.d. 0.25 mm, df 0.25 µm splitless 50 (1 min) to 270 °C (2 min) at 45 °C min-1 He (99.996%)/20 psi homemade/270 °C HP 4500 ICPMS (Hewlett-Packard) 1250 W ∼1.7 L min-1 ∼1 L min-1 ∼16 L min-1 120Sn: 30 ms

that all compounds are removed from the precolumn, it is kept at 290 °C for 4 min. The nature of this injection process causes a small amount of peak broadening compared to a normal syringe injection; this, however, has no influence on the analysis. A schematic representation of the desorption unit is given in Figure 2. After desorption, the bar is immediately ready for a new extraction. Gas Chromatography-Inductively Coupled Plasma Mass Spectrometry. Parameter settings for GC and ICPMS are summarized in Table 1. Oven temperature program and inlet pressure were set for a compromise between column efficiency and speed of analysis. For the coupling of GC with ICPMS, a homemade transfer line was used. The basic design of this transfer line has been described in detail elsewhere.24 In brief, it consists of a deactivated fused-silica capillary running through a stainless steel capillary that is directly heated by an ac power supply. The Ar nebulizer gas for the ICP is heated in a similar stainless steel capillary before it is introduced into the steel transfer capillary. It then flows around the fused-silica capillary, such that cold spots (24) De Smaele, T.; Verrept, P.; Moens, L.; Dams, R. Spectrochim. Acta Part B 1995, 50, 1409-1416.

are avoided. The transfer line used in this work was a new version of the model described earlier. The main improvement was that the capillary, which provides the heated Ar nebulizer gas, was incorporated in the length of the transfer line, thereby providing a more rigid structure and constant temperature along the whole length of the instrument. Reagents, Standards, and Samples. Triphenyltin hydroxide (95% purity) and tricyclohexyltin hydroxide (TCT, 95% purity) were purchased from Supelco Inc. (Bellefonte, PA), tributyltin chloride (97% purity) was from Fluka Chemie AG (Buchs, Switzerland), and tripropyltin chloride (98% purity) was from Merck (Hohenbrunn, Germany). Stock solutions of 10 mg L-1 were prepared in ethanol (p.a., Panreac, Barcelona, Spain). Further dilutions were also carried out with ethanol. All stock solutions were stored in the dark at 4 °C. Isooctane was obtained from Vel (Leuven, Belgium). Sodium tetraethylborate was purchased from Strem Chemicals (Newburyport, MA). A 1% (m/v) solution in Milli-Q water (Millipore, Milford, MA) was prepared daily. Tetramethylammonium hydroxide (TMAH; 10% in water), used to digest the mussel samples, was purchased from Fluka Chemie. Buffer solutions with a pH value ranging from 8 to 9 were prepared by adding appropriate amounts of KOH to 10 g L-1 NH4Cl solution; for buffer solutions with a pH value ranging from 4 to 5, an appropriate amount of HOAc was added to 16 g L-1 NaOAc. Harbor water samples were taken in a small recreational harbor in Merelbeke near Ghent, Belgium, and analyzed within 24 h. As polyethylene flasks are known to release trace amounts of OTs, 100-mL glass flasks were used. Mussels originating from Zeeland, The Netherlands, were purchased in a Belgian supermarket and analysis was performed within 24 h. Sample Preparation and Derivatization. Aqueous test samples were freshly prepared each measuring day by adding an appropriate amount of OT stock and 500 µL of a 1% NaBEt4 solution to a mixture of 30 mL of Milli-Q water and 10 mL of buffer solution in a closed-cap 40-mL vial. For harbor water samples, 10 mL of buffer solution was added to 30 mL of sample; subsequently, an appropriate amount of internal standards (TCT for TPhT and TPrT for TBuT) and 500 µL of a 1% NaBEt4 solution were added. To remove the shell, fresh mussels were immersed in liquid nitrogen and crushed, whereby shell and mussel could easily be separated. Samples were digested afterward, on the basis of a procedure developed by Nagase and Hasebe.6 About 1 g of mussel sample was put into a 40-mL vial and 5 mL of TMAH solution was added together with appropriate amounts of internal standards and spikes for standard addition. This mixture was then heated (60 °C) for 1 h. After cooling to room temperature, 30 mL of buffer was added. In previous work,12,14 an optimum pH of 8-9 was suggested for the derivatization of TPhT and TCT. For TBuT and TPrT, the optimum pH value for derivatization is 4-5.25,26 When the latter pH value was used, the sensitivity for TPhT was only slightly lower; therefore, a pH value of 4-5 was used when TPhT and TBuT were determined simultaneously. To neutralize the excess of TMAH, an appropriate amount of HCl (12 mol L-1) was slowly added to the mixture until the pH was restored. Buffer was added before the acid to avoid very low pH values and high temperatures at local scale in the solution. In the next step, 500 (25) De Smaele, T. Ph.D. Thesis, University of Ghent, Belgium, 1998. (26) Ceulemans, M.; Lobinski, R.; Dirkx, W. M. R.; Adams, F. C. Fresenius J. Anal. Chem. 1993, 65, 256-262.

Analytical Chemistry, Vol. 73, No. 7, April 1, 2001

1511

Figure 3. Extraction time profile of SBSE-TD-CGC-ICPMS for 100 pg L-1 TPhT and TCT in 40 mL of water.

µL of NaBEt4 solution was added and the solution was manually shaken. After ∼5 min, extraction was started. When the liquid/liquid extraction was used as comparison, 1 mL of isooctane was added and the mixture was shaken vigorously. Ten minutes later ∼300 µL of the organic layer was transferred to a vial and ∼15 mg of silica gel was added for sample cleanup.25 A 2-µL aliquot was injected into the GC’s split/splitless liner. RESULTS AND DISCUSSION Extraction Time Profile. To achieve the best agitation, the maximum stirring rate was used (1400 rotations min-1). As can be observed in Figure 3, equilibrium could not be reached within 1 h for TPhT and TCT and probably several hours are required. This relatively long period of time can be explained by the volume of the PDMS phase. As SBSE applies much larger volumes of PDMS than SPME, more material needs to be transferred to the PDMS phase, and thus, more time is required to reach equilibrium. Mass transfer in the PDMS itself might also influence extraction kinetics. Bearing in mind the difference in polarity, the similar extraction behavior of TPhT and TCT is somewhat surprising, but this was also observed in previous work, in which headspace SPME using a 100-µm PDMS fiber was applied. This similar behavior is also an indication that TCT is a good internal standard for TPhT as far as the extraction is concerned. As with SPME, working at equilibrium is not necessary as long as extraction conditions are kept constant. Taking the high sensitivity of CGC-ICPMS into account, an extraction time of 15 min was selected. As desorption also requires 15 min, a fast sample throughput can be guaranteed. The considerable variation in the signal profile is due to the use of different stir bars for this experiment and the fact that only one sample per time value was analyzed. Analytical Figures of Merit. Linear Dynamic Range. Standard samples containing a fixed concentration of 0.25 ng L-1 internal standard (TCT) and a TPhT concentration varying from 0.025 to 100 ng L-1 were analyzed. A regression coefficient of 0.9988 was obtained, indicating a linear behavior over the whole range of concentrations used. Also, no deviations from the theoretical straight line were observed at the highest concentrations; neither did it occur when working without IS correction, which could mask deviations that occur in the same way for both compounds. For concentrations above 1 ng L-1 in clean samples, 1512

Analytical Chemistry, Vol. 73, No. 7, April 1, 2001

a reduced detector voltage was applied to the ICPMS. This was done to avoid automatic shutdown of the detector, an action that occurs to prevent rapid deterioration or damage to the detector surface when very high signals are registered. As an internal standard was used, this change in detector voltage did not influence the signal ratio. Another way to avoid high amounts of ions entering the mass spectrometer is by applying split instead of splitless injection. Most likely, the linear range exceeds concentrations of 100 ng L-1, but these concentrations are beyond the area of application and were therefore not measured. In previous work,14 a linear range of up to at least 6000 ng L-1 was found for the determination of TPhT with headspace SPME-CGCICPMS. As the PDMS volume of SPME (∼0.5 µL) is far lower than that of SBSE, it can be expected that the linear area for SBSE also extends into the microgram per liter area. Limits of Detection (LODs). The limit of detection for TPhT in water samples was calculated in two ways. For both approaches, the sensitivity (counts per unit of concentration) was obtained by integrating the peak area of a standard. The standard deviation on the background was obtained in two different ways. First, for the “procedure LOD”, 10 blanks were analyzed, and for each of these, an area with the same width and position as used for the standard was integrated. Using the 3s criterion, a procedure LOD of 100 fg L-1 was obtained. In a second approach, the “instrumental LOD” was calculated using the standard deviation on 10 “blank areas” within the chromatogram of the standard. The width of these areas was the same as that of the standard peak; the position of course was not. In this way, possible contamination at this very low level does not have any influence because the blank area is taken in a region where only background noise is present. An instrumental LOD of 10 fg L-1 was calculated based on the 3s criterion. This type of LOD should be regarded as an indication of the instrumental possibilities of this technique. As for many highly sensitive techniques, the procedure LOD is determined mainly by the sample preparation (purity of reagents and glassware) and/or memory effects and not by the instrumental detection power. These values for the LODs are 2-3 orders of magnitude lower than the LODs found for headspace SPMECGC-ICPMS.14 To the best of our knowledge, these LODs are the lowest ever reported for OT determination in aqueous samples. Recently, Tao et al.27 developed a technique with a similar instrumental LOD but a more complex and labor-intensive liquid/ liquid extraction was applied in combination with large volume injection, programmed temperature vaporization (PTV), and shieldtorch conditions for the ICP. An attempt to apply shieldtorch conditions with our type of transfer line failed as the plasma could not be ignited. Repeatability. Ten consecutive extractions with the same bar of standard solutions containing 100 pg L-1 TPhT and TCT (as internal standard) resulted in an RSD of 12%, a satisfying result considering that several processes (derivatization, extraction, separation, and detection) and low analyte concentrations are involved. It has to be noted that the use of different stir bars increased the variation, but this is largely compensated for by using an internal standard. Most likely, this difference among stir bars will be much smaller for the commercially available stir bars. (27) Tao, H.; Rajendran, R. B.; Quetel, C. R.; Nakazato, T.; Tominaga, M.; Miyazaki, A. Anal. Chem. 1999, 71, 420-4215.

Figure 5. Chromatogram for a fresh mussel sample, TCT was added as internal standard (1 ng).

Figure 4. (a) Chromatogram for a harbor water sample. TBuT was quantified and TPrT was added as internal standard (50 pg). (b) Chromatogram for a harbor water sample; TPhT was quantified and TCT was added as internal standard (50 pg).

Accuracy. As no certified reference material exists for TPhT, CRM477 (mussel tissue) was chosen because the indicative value suggested a relatively high amount of TPhT (1.58 µg g-1, s ) 0.43 µg g-1, as cation). As reported by Morabito et al.,28 TPhT was not found to be stable at any storing temperature and degradation of TPhT in this CRM was expected. Therefore, the value that was found with SBSE analysis (1.17 µg g-1, s ) 0.23, as cation) was compared to the value found by an analysis carried out a few days later using classical liquid/liquid extraction with isooctane (0.97 µg g-1, s ) 0.38, as cation), hereby showing good agreement. Results were calculated using standard addition. Taking the results and the uncertainties into account, it is hard to say something about any degradation that might have taken place in the CRM. Analysis of Real-Life Samples. Harbor Water. Chromatograms obtained for a harbor water sample are presented in Figure 4a (TBuT) and b (TPhT). Using standard addition and internal standardization, 200 pg L-1 (s ) 25 pg L-1) TBuT and 22 pg L-1 (s ) 2 pg L-1) TPhT were determined. The sample was analyzed in 3-fold. The concentrations that were found are as expected, considering the size of the harbor and the fact that only small (