Anal. Chem. 1999, 71, 4208-4215
Tin Speciation in the Femtogram Range in Open Ocean Seawater by Gas Chromatography/ Inductively Coupled Plasma Mass Spectrometry Using a Shield Torch at Normal Plasma Conditions Hiroaki Tao,* Ramaswamy Babu Rajendran, Christophe R. Quetel,† Tetsuya Nakazato, Mamoru Tominaga, and Akira Miyazaki
National Institute for Resources and Environment, 16-3, Onogawa, Tsukuba, Ibaraki 305-8569, Japan
A sensitive method for the determination of ultratrace organotin species in seawater is described. The merits and demerits of derivatization methods using Grignard reagent or sodium tetraethylborate (NaBEt4) were evaluated in terms of derivatization efficiency, applicability to the programmed temperature vaporization (PTV) method, and procedural blanks. The sensitivity of the gas chromatography/inductively coupled plasma mass spectrometry (GC/ICPMS) was improved by more than 100-fold by operating the shield torch at normal plasma conditions, compared with that obtained without using it. The absolute detection limit as tin reached subfemtogram (fg) levels. Furthermore, the detection limit in terms of relative concentration was improved 100-fold by using the PTV method, which enabled the injection of a large sample volume of as much as 100 µL without loss of analyte. When the organotin species in seawater were extracted into hexane with a preconcentration factor of 1000 after ethylation with NaBEt4 and a 100 µL aliquot of the extract was injected into the GC, the instrumental detection limit in relative concentration reached 0.01 pg/L in original seawater. Sources of contamination of organotin species during the sample preparation were examined, and a purification method of NaBEt4 was developed. Finally, the method was successfully applied to open ocean seawater samples containing organotin species at the level of 1-100 pg/L. The use of organotin compounds as antifouling paints has led to toxic effects for nontarget aquatic species. They cause deleterious effects, such as shell anormalies in oysters and imposex in gastropods, even at concentrations below a few ng/L. 1 In many countries, the use of tributyltin and triphenyltin compounds as antifouling paints for small boats is now restricted by law. However, they remain in use for large vessels, and about 69% of all large ships are reported to use them.2 In addition, large amounts of other organotin compounds, such as dibutyltin and * Corresponding author. Fax: 81-298-58-8308. E-mail:
[email protected]. † Present Address: Institute for Reference Materials and Measurements, Retieseweg, B-2440 Geel, Belgium. (1) Fent, K. Crit. Rev. Toxicol. 1996, 26, 1-117. (2) Ambrose, P. Mar. Pollut. Bull. 1994, 28, 134.
4208 Analytical Chemistry, Vol. 71, No. 19, October 1, 1999
dioctyltin, are used as stabilizers for plastics such as poly(vinyl chloride) (PVC). The toxicity of these species depends on the chemical form, which therefore makes the speciation of organotin compounds necessary. The most popular current analytical method involves the use of gas chromatography, in conjunction with a flame photometric detector (GC-FPD).3,4 However, the method sometimes suffers from serious interference from concomitant materials in the sample. Therefore, so-called hyphenated methods, based on the on-line coupling of chromatographic separation systems, such as GC, liquid chromatography (LC), and supercritical fluid chromatography (SFC), to element-specific detection systems, such as atomic absorption spectrometry (AAS), microwave induced plasma atomic emission spectrometry (MIPAES) and inductively coupled plasma mass spectrometry (ICPMS), are in current use. A variety of combinations, such as LC/ ICPMS,5-8 SFC/ICPMS,9,10 GC/AAS,11 GC/MIPAES,12-18 and GC/ ICPMS,19-23 have been recently applied to the speciation of tin. (3) Mu ¨ ller, M. D. Anal. Chem. 1987, 59, 617-623. (4) Michel, P.; Averty, B. Appl. Organomet. Chem. 1991, 5, 393-397. (5) McLaren, J. W.; Siu, K. W. M.; Lam, J. W.; Willie, S. N.; Maxwell, P. S.; Palepu, A.; Koether, M.; Berman, S. S. Fresenius’ J. Anal. Chem. 1990, 337, 721-728. (6) Yang, H. J.; Jiang, S. J.; Yang, Y. J.; Hwang, C. J. Anal. Chim. Acta 1995, 312, 141-148. (7) Inoue, Y.; Kawabata, K.; Suzuki, Y. J. Anal. At. Spectrom. 1995, 10, 363366. (8) Koori, M.; Sato, K.; Inoue, Y.; Ide, K.; Ohokouchi, H. Bunseki Kagaku 1995, 44, 537-542. (9) Shen, W. L.; Vela, N. P.; Sheppard, B. S.; Caruso, J. A. Anal. Chem. 1991, 63, 1491-1496. (10) Vela, N. P.; Caruso, J. A. J. Anal. At. Spectrom. 1992, 7, 971-977. (11) Kuballa, J.; Wilken, R.-D.; Jantzen, E.; Kwan, K. K.; Chau, Y. K. Analyst (Cambridge, U.K.) 1995, 120, 667-673. (12) Lobinski, R.; Dirkx, W. M. R.; Ceulemans, M.; Adams, F. C. Anal. Chem. 1992, 64, 159-165. (13) Sta¨b, J. A.; Cofino, W. P.; Van Hattum, B.; Brinkman, U. A. T. Fresenius’ J. Anal. Chem. 1993, 347, 247-255. (14) Ceulemans, M.; Lobinski, R.; Dirkx, W. M. R.; Adams, F. C. Fresenius’ J. Anal. Chem. 1993, 347, 256-262. (15) Ceulemans, M.; Adams, F. Anal. Chim. Acta 1995, 317, 161-170. (16) Ceulemans, M.; Slaets, S.; Adams, F. Talanta 1998, 46, 395-405. (17) Chau, Y. K.; Yang, F.; Maguire, R. J. Anal. Chim. Acta 1996, 320, 165169. (18) Szpunar, J.; Schmitt, V. O.; Lobinski, R.; Monod, J.-L. J. Anal. At. Spectrom. 1996, 11, 193-199. (19) Kim, A.; Hill, S.; Ebdon, L.; Rowland, S. J. High Resolut. Chromatogr. 1992, 15, 665-668. (20) Prange, A.; Jantzen, E. J. Anal. At. Spectrom. 1995, 10, 105-109. 10.1021/ac990087a CCC: $18.00
© 1999 American Chemical Society Published on Web 08/20/1999
Among these, the GC/ICPMS and the GC/MIPAES give the lowest detection limits at the subpicogram level, as absolute amounts, or at sub-ppt levels as relative concentrations. These detection limits are sufficient for the monitoring of organotin compounds in polluted areas, such as harbors, bays, and inland seas, but, to clarify the occurrence and behavior of organotin compounds in the open ocean, more sensitive methods are required. Consequently, investigations have been reported on the contamination of coastal seawater,24-26 but few on that in the open ocean. The purpose of the present study is to develop a speciation method which enables the determination of ultratrace levels of organotin compounds in open-ocean seawater. As one of the techniques to improve the detection limit, a large-volume injection method using programmed temperature vaporization (PTV) was investigated. Compared with LC, the injection volume into a capillary GC is generally limited to 1 µL, and this poses serious restrictions on the detection limit in terms of relative concentration. To overcome this problem, the PTV method using Tenax powder as a packing material has been recently employed for the speciation of tin.14-16 The choice of packing material is critical for obtaining sharp chromatographic peaks and a quantitative recovery of the analytes. This study reports on the use of silanized quartz wool, an inert material. To select the most suitable derivatization method for PTV, three derivatizations, namely, ethylation using sodium tetraethylborate (NaBEt4), propylation using propylmagnesium bromide (PrMgBr), and pentylation using pentylmagnesium bromide (PeMgBr), were examined, and the efficiencies of these reagents were compared. During the experiment, the authors found that the detection limit as absolute amounts was also improved by more than 2 orders of magnitude by using a shield torch at normal plasma conditions. The shield torch has been generally used in conjunction with cool-plasma conditions to eliminate polyatomic-ion interference. This finding leads to the possibility for ultratrace speciation of organometallic compounds at the femtogram level. By using both techniques, the lowest detectable concentration was improved by more than 3 orders of magnitude, compared with prior reported values. Following an investigation of the sources of contamination in the procedural blank, this method was successfully applied to the determination of organotin compounds in open ocean seawater. EXPERIMENTAL SECTION Reagents. Monobutyltin (MBT) chloride, monophenyltin (MPhT) chloride, diphenyltin (DPhT) chloride, and tripentyltin (TPeT) chloride were purchased from Aldrich (Milwaukee, WI). Dibutyltin (DBT) chloride and tributyltin (TBT) chloride were purchased from Wako Pure Chemicals (Osaka, Japan). Triphenyltin (TPhT) chloride was purchased from Kanto Chemicals (21) De Smaele, T.; Moens, L.; Dams, R.; Sandra, P. Fresenius’ J. Anal. Chem. 1996, 355, 778-782. (22) De Smaele, T.; Moens, L.; Dams, R.; Sandra, P.; Van der Eycken, J.; Vandyck, J. J. Chromatogr., A 1998, 793, 99-106. (23) Moens, L.; De Smaele, T.; Dams, R.; Van Den Broeck, P.; Sandra, P. Anal. Chem. 1997, 69, 1604-1611. (24) Alzieu, C.; Sanjuan, J.; Deltreil, J. P.; Borel, M. Mar. Pollut. Bull. 1986, 17, 494-498. (25) Dowson, P. H.; Bubb, J. M.; Lester, J. N. Mar. Pollut. Bull. 1993, 26, 487494. (26) Kubilay, N.; Yemenicioglu, S.; Tugrul, S.; Salihoglu, I. Mar. Pollut. Bull. 1996, 32, 238-240.
(Tokyo, Japan). Tripropyltin (TPrT) chloride and tetrabutyltin (TeBT) were purchased from Merck (Darmstadt, Germany). Stock standard solutions of individual organotin compounds (1000 mg/L as Sn) were prepared by dissolving appropriate amounts of the respective compounds in methanol (pesticide-analysis-grade, Wako); these were then stored at 4 °C until required. Working standard solutions were prepared daily by dilution from the stock solutions. TPrT chloride and TPeT chloride were used as an internal standard for pentylation by PeMgBr (2 M diethyl ether solution, Aldrich) and propylation by PrMgBr (2 M THF solution, Tokyo Kasei, Tokyo, Japan), respectively. TPrT chloride was also used as an internal standard for ethylation reactions with NaBEt4 (Strem Chemicals, Newburyport, MA). The 5% NaBEt4 solution was filtered with a 0.2-µm hydrophilic poly(tetrafluoroethylene) (PTFE) membrane filter (Sumplep CLR25-LG, polyethylene housing, Millipore, Bedford, MA) and was then centrifuged at 2000 rpm for 3 min to remove particulate organotin impurities. The solution was then extracted three times with a one-tenth volume of hexane in order to remove dissolved impurities. Hydrochloric acid, acetic acid, and ammonium hydroxide were of ultrapure grade (Merck). Acetate buffer (1 M, pH 5) was prepared by mixing acetic acid and ammonium hydroxide. After adding a small aliquot of the purified NaBEt4 solution, the buffer was purified by two successive hexane extractions. Sodium chloride (analytical-reagentgrade, Wako) and anhydrous sodium sulfate (pesticide-analysisgrade, Wako) were heated at 500 °C for 24 h to decompose organotin impurities. Tropolone (95%, Wako) was used without purification. Hexane, dichloromethane, and ethyl acetate were of pesticide-analysis-grade (Wako). The Xe gas (981 ppm diluted in Ar gas), used for optimization of the ICPMS parameters, was purchased from Takachiho Kagaku (Tokyo, Japan). Organotin concentrations given in this paper are expressed as tin. Instrumentation. An HP6890 GC equipped with a PTV inlet system (Hewlett-Packard, Wilmington, DE) was coupled to an HP4500 ICPMS (Hewlett-Packard-Yokogawa, Tokyo, Japan) by a laboratory-made transfer line, as described in a previous paper.27 Injections were made by an automatic injector through the septumless head in a quartz liner which was packed with silanized quartz wool. A maximum injection volume was 25 µL per injection, and as many as 10 multiple injections were possible. After venting the solvent, analytes were transferred into a capillary column (DB-1, 0.32-mm i.d. × 30-m-long × 0.25-µm film thickness, J&W Scientific, Folsom, CA) by rapidly increasing the inlet temperature. Optimum operating conditions of the PTVGC/ICPMS instrument are given in Table 1. A Turbo Vap II evaporator (Zymark, Hopkinton, MA) was used for preconcentration of the extract. Sample Preparation. Propylation or Pentylation by a Grignard Reagent. Twenty milliliters of 6 M HCl, 20 g of NaCl, and 100 µL of ca. 5 µg/L TPeT chloride (for propylation) or TPrT chloride (for pentylation) as an internal standard were added to approximately 1 L of seawater, accurately weighted in advance, in a Pyrex glass vessel. Five milliliters of 0.25% tropolone hexane and 5 mL of dicholomethane were added, and the solution was stirred vigorously for 30 min by a magnetic stirrer. After phase separation, the extraction was repeated with the same organic solvents. The combined organic layer was dried by passing (27) Tao, H.; Murakami, T.; Tominaga, M.; Miyazaki, A. J. Anal. At. Spectrom. 1998, 13, 1085-1093.
Analytical Chemistry, Vol. 71, No. 19, October 1, 1999
4209
Table 1. Optimum Operating Conditions for GC/ICPMS ICPMS parameter forward power (kW) Ar plasma gas flow rate (L/min) Ar auxiliary gas flow rate (L/min) Ar make-up gas flow rate (L/min) sampling depth (mm) measured amu dwell time (ms) transfer line parameter transfer line column heater 1 (torch side) temp (°C) heater 2 (GC side) temp (°C) GC parameter injection mode inlet temp (°C) (hold time (min)) inlet temp ramp rate (°C/min) He vent gas flow rate (mL/min) vent end time (min) He purge gas flow rate (mL/min) purge start time (min) column oven temp (°C) (hold time (min)) oven temp ramp rate (°C/min) sample volume (µL) He carrier gas flow rate (mL/min) a
1.2 16.4 0.93 1.0 5.8 120 100 inactivated capillary column (1.5 m × 0.32 mm) 240 280 PTV solvent vent mode -10(1.5a, 4.1a)-450(4.0)-250(0) 720(-10-450), -60(450-250) 100 1.4a, 4.0b 50 2.7a, 5.3b DB1, 30 m × 0.32 mm × 0.25 µm 50 (2.8a, 5.4b) - 180 (0) - 220 (0) - 300 (2.7) 90 (50 - 180), 20 (180 - 220), 80 (220 - 300) 25 - 100 2
For a 25 µL sample volume. b For a 100 µL sample volume.
through an anhydrous sodium sulfate column (Sep-Pak DRY, Waters, Milford, MA). After washing the column with 2 mL of hexane, the combined organic layer was preconcentrated to 1 mL under a gentle stream of Ar at 25 °C using the Turbo Vap II evaporator. One milliliter of Grignard reagent was added to the concentrate, which was then reacted for 30 min at room temperature. Ten milliliters of 0.1 M H2SO4 was gradually added, with cooling, to decompose the unreacted Grignard reagent, and the organic layer was separated. Five milliliters of hexane was added to the aqueous layer to recover the residual organotin derivatives. The combined organic layer was passed through an anhydrous Na2SO4 column, which was washed with 2 mL of hexane, and the combined effluent was then preconcentrated to 1 mL under a gentle stream of Ar at 25 °C using the Turbo Vap II. Ethylation Using NaBEt4. Five milliliters of 1 M CH3COOH/ CH3COONH4 buffer was added, together with 100 µL of ∼5 µg/L of TPrT chloride as an internal standard, to approximately 1 L of seawater, accurately weighted in advance, in a Pyrex glass vessel to adjust the pH to 5. If the sample had been acidified in advance, 6 M ammonium hydroxide would be used for neutralization. Two milliliters of 5% NaBEt4 and 30 mL of hexane were added, and the solution was stirred vigorously for 20 min on a magnetic stirrer. After phase separation, the hexane layer was placed in a centrifuge tube. Two grams of anhydrous Na2SO4 was added to the extract, followed by shaking for 1 min to remove water. After this, the centrifuged hexane layer was transferred into a preconcentration tube. Two milliliters of hexane was added to the centrifuge tube to recover the residual hexane layer. The combined hexane layer was preconcentrated to 1 mL under a gentle stream of Ar at 25 °C using the Turbo Vap II evaporator. RESULTS AND DISCUSSION Comparison of Derivatization Yields. The yields for the three derivatization methods were estimated by comparing the 4210 Analytical Chemistry, Vol. 71, No. 19, October 1, 1999
Figure 1. Comparison of the derivatization yields for several organotin species by Grignard reagents and NaBEt4. Neither the propylation of TPrT nor the pentylation of TPeT were examined.
signal peak area for each derivative. Filtered seawater samples containing approximately 200 ng of each organotin species were processed as described in the Experimental Section. The results are shown in Figure 1. It is assumed that the peak area depends, not only on the derivatization yield but also the extraction efficiency of the derivatives from water and the sensitivity of the GC/ICPMS. To calculate the derivatization yields, it was assumed in the present experiment that all organotin species gave the same extraction efficiencies after being fully derivatized. It was also assumed that all the organotin derivatives gave the same sensitivity with respect to tin in the GC/ICPMS. In cases in which the analyte neither decomposed during GC separation nor coeluted with the solvent, this assumption was found to be valid for mercury species, as was reported in a previous paper.27 Based on these assumptions, the derivatization yield could be obtained by normalizing the individual peak area with that of TeBT, which was originally fully derivatized. It can be seen from Figure 1 that the ethylation yields for all the organotin species except for DPhT chloride were approximately 100%. The reason for the low yield for DPhT chloride is not clear, but a lower recovery for this compound from a sediment sample has also been reported for ethylation with ethylmagnesium bromide,17 although the recovery
Figure 3. Effect of solvent vent time on the recovery of (a) ethyl derivatives and (b) propyl derivatives at a vent temperature of -10 °C.
Figure 2. Effect of solvent vent time on the recovery of (a) ethyl derivatives, (b) propyl derivatives, and (c) pentyl derivatives at a vent temperature of 20 °C.
in that case included not only the derivatization yield but also the extraction efficiency from the sediment. Alternatively, some problems might exist with regard to the stability of DPhT chloride in the standard since the other derivatization methods also gave lower yields in this study. This issue is now under investigation. A comparison of the individual derivatizations shows that the pentylation of bulkier molecules such as phenyltin species gave considerably lower yields. This is probably due to steric hindrance between the phenyl and pentyl groups. Lower yields were also observed for monosubstituted species such as MBT and MPhT, which might be explained by the fact that these species must react three times with the bulky pentyl group to give the final products. Similar phenomena were reported in the comparison of methylation and pentylation by Grignard reagents.13 The high value encountered for pentylation of TBT chloride was due to an impurity in the PeMgBr reagent which is discussed below. In terms of derivatization yield, ethylation by NaBEt4 is preferred to propylation or pentylation by Grignard reagents. Optimization of the Vent Time and Vent Temperature for the PTV Method. To use the PTV method, the boiling point of the analyte must be considerably higher than that of the solvent. Since the boiling point of the derivatives will increase in the order of ethyl < propyl < pentyl derivatives, pentylation appears to be the most suitable procedure. The effect of vent time on the recovery of each derivative was examined at a vent temperature of 20 °C and a vent gas flow of 100 mL/min. The injection volume was set to 25 µL. Figure 2 shows a comparison of the recovery obtained for the three types of derivatives. The recovery was calculated by normalizing the signal peak area with that of the
most nonvolatile species, TPhT derivative. It is reasonable to assume that the recovery of the TPhT derivative is 100% because of its high boiling point. The venting efficiency of the solvent (hexane, bp 68.7 °C) was monitored by measuring 40Ar13C+ polyatomic ion counts instead of measuring the 12C+ or 13C+ signal which was too strong to measure. Most of the hexane was vented just before 0.3 min. From an experiment on recovery in the preconcentration step using the Turbo Vap II evaporator, it was found that the analytes were rarely lost in the presence of solvent but began to vaporize rapidly after the solvent vent was completed. Even the most volatile species, i.e., tetraethyltin (TeEtT, mp -112 °C, bp 181 °C), showed a recovery of 93.7%. However, almost no signal for TeEtT was observed even at 0.3 min with the PTV method. This is because vaporization of the solvent took place from the upper part of the quartz wool packing and therefore most of the solvent had been already vaporized before the solvent in the bottom of the liner was vented. The recoveries for ethyl derivatives of MBT, TPrT, and DBT were also low. Although the recoveries for propyl derivatives were superior to those for ethyl derivatives, they were still inferior to those for pentyl derivatives, which showed almost 100% recoveries at 20 °C. These facts support the conclusion that pentylation is preferred over the other derivatizations for the PTV method, especially for the case of volatile organotin species. However, since pentylation was not preferred because of the low derivatization yield and the high blank signal, the possibility of obtaining a better recovery for the ethyl or propyl derivatives was pursued by decreasing the vent temperature. Recovery of the volatile tin species increased with decreasing vent temperature. A longer time was required to vent the solvent at lower temperature; for example, 0.6 min was required at 10 °C, 0.8 min at 0 °C, and 1.3 min at -10 °C, but the period between the end of solvent vent and the beginning of analyte vaporization was also extended. This longer period allowed not only the complete removal of the solvent but also a multiple injection of the sample solution. Figure 3 shows the recoveries of ethyl and propyl derivatives at a vent temperature of -10 °C. It can be seen that all the ethyl derivatives, including TeEtT, were retained for Analytical Chemistry, Vol. 71, No. 19, October 1, 1999
4211
Figure 5. Comparison of contours of 124Xe signal intensity as a function of the sampling depth and the plasma power observed (a) with the shield torch and (b) without the shield torch. Makeup gas flow rate is (a) 1.0 L/min and (b) 1.1 L/min, respectively. The numbers on the curves are in 103 counts/s. Figure 4. Effect of transfer time on signal peak areas of (a) ethyl derivatives and (b) propyl derivatives using the PTV method.
at least 4 min. Since the boiling points of propyl derivatives are higher than those of ethyl derivatives, higher recoveries for tetrapropyltin (TePrT) and tripropylbutyltin (TPrBT) would be expected, compared with those of TeEtT and triethylbutyltin (TEtBT), respectively. However, the opposite result was observed. This might be related to the difference in coexisting materials in the final hexane extract with Grignard derivatization as opposed to NaBEt4 derivatization. NaBEt4 or its reaction product such as BEt322 might dissolve in the final extract and play a role in retarding the vaporization of the ethyl derivatives. A similar phenomenon was observed with regard to the effect of transfer time mentioned later. Consequently, a vent temperature of -10 °C was used in all subsequent experiments. Optimization of the Transfer Time for the PTV Method. After venting the solvent, the analytes were transferred from the PTV inlet to the capillary column by rapidly increasing the inlet temperature. The ramp rate was set at the maximum value of 720 °C/min. Figure 4 shows the effect of the transfer time on the signal peak areas of the ethyl or propyl derivatives. Transfer time is defined as the period between the ramp start time and the purge start time in this study. When the transfer time was too short, nonvolatile species, such as TPhT and DPhT derivatives, were not transferred quantitatively. Despite the lower boiling points of the ethyl derivatives relative to those of the propyl derivative, the transfer efficiency for the former were rather poorer than those of the latter at a transfer time of 0.4 min. This appears to be related to the differences in coexisting materials in the final hexane extract, as mentioned above. However, since the transfer efficiencies for all the tin species were found to be nearly 100% after 1.0 min for both types of derivatives, the transfer time was set at 1.2 min. Increase of Sensitivity in Relative Concentration by Multiple Injection. Since the maximum volume per injection was limited to 25 µL with the present PTV system, a multiple injection was used in order to achieve a larger volume injection. For the 4212 Analytical Chemistry, Vol. 71, No. 19, October 1, 1999
propyl derivatives, up to a triple injection was possible within the vent time of 2.8 min without loss of analyte. The interval between the injections requires optimization. If the interval is too short, the accumulated sample exceeds the holding capacity of the liner and causes a deteriorated chromatographic separation of the analytes. Conversely, if the interval is too long, the last injected solvent cannot be vaporized sufficiently within 2.8 min, which also causes a deteriorated separation. From an experiment on the effect of the interval on solvent vent and analyte recovery, it was found that an interval of between 35 and 50 s was appropriate, and 40 s was selected for the propyl derivatives. A good linearity between injection volume and signal peak area was observed for all the propyl derivatives up to 75 µL with the regression coefficients ranging from 0.9986 to 0.9999. It can be seen from Figure 3 that the recovery of TeEtT is approximately 90% with a single injection of 25 µL at a vent time of 2.8 min. However, the recovery increased when triple injection at the same vent time was used, because of the longer presence of the solvent. For the case of the ethyl derivatives, as many as four injections were possible within a vent time of 4 min and a 40-s interval. A good linearity was also observed for all the ethyl derivatives up to 100 µL with the regression coefficient ranging from 0.9974 to 0.9997. Increase of Sensitivity in Absolute Amounts by Using the Shield Torch. The shield torch has been typically used with coolplasma conditions to decrease polyatomic ion interference such as from ArO+ and Ar2+. However, in the present study, the shield torch is used with normal (hot) plasma conditions. The sensitivity was drastically improved with this condition. Figure 5 shows a comparison of contours of Xe signal intensity obtained with and without the shield torch as a function of sampling depth and plasma power. Since a good correlation between Xe signal and Sn signal was reported,21 this continuous Xe signal could be used for the optimization of the ICPMS parameters instead of a transient Sn signal. Both contours were totally different. The Xe signal intensity obtained without using the shield torch increased with decreasing power, reaching a maximum at a power of 0.7 kW and at a depth of 12 mm. However, the power had to be set at 1.2 kW for a practical analysis because of poor stability and robustness
Table 2. Comparison of Typical Blanks of Propylation with Propylmagnesium Bromide and Ethylation with Sodium Tetraethylborate (ng/L)
c
Figure 6. Chromatograms of the propyl derivatives of a mixture of tin standards (a) without the shield torch and (b) with the shield torch. A, TPrT (5.8 pg); B, MBT (5.8 pg); C, DBT (5.3 pg); D, TBT (4.9 pg); E, TeBT (4.7 pg); F, MPhT (4.6 pg); G, TPeT (6.4 pg); H, DPhT (3.8 pg); and I, TPhT (4.1 pg).
at lower power. On the contrary, the Xe signal intensity increased with increasing power and with decreasing depth when using the shield torch, and it was much higher than that obtained without the shield torch. Chromatograms of the propylated organotin species, obtained with and without the shield torch at the respective optimum condition, are shown in Figure 6. The sensitivity for tin with the shield torch was approximately 100 times higher than that without the shield torch, while the background signals were nearly the same. The reason for this improvement of the sensitivity is not clear at this time, but it might be because a secondary discharge at the interface region is diminished by using the shield torch, which would decrease the dispersion of ion energy and consequently increase ion transmission to the mass spectrometer. Procedural Blanks. Reduction of the procedural blank was essential for the ultratrace determination presented here. Since pentylation with PeMgBr gave a higher blank, especially for TBT, the use of pentylation was abandoned. Contamination of the commercially available PeMgBr reagent by TBT species has been reported.13 Typical blank levels of propylation with PrMgBr and ethylation with NaBEt4 are given in Table 2. Compared with the latter method, the former method gave considerably higher blank signals. Since a number of steps are necessary for Grignard derivatization, the possibility of contamination increases. For example, the NaCl reagent, used for the extraction of organotins as chlorides prior to the derivatization, was found to be contaminated by MBT at 6.6 pg/g and DBT at 0.5 pg/g. Although this contamination can be removed by heating at 500 °C for 24 h, the main source of contamination was the Grignard reagent, and it was not possible to remove this because of the high reactivity to water. Therefore, it was concluded that propylation with PrMgBr
species
PrMgBr
NaBEt4a
NaBEt4b
NaBEt4c
Sn MBT DBT MPhT TBT TeBT TPeT DPhT TPhT
23 1.3 0.36 0.023 0.11 nd 0.75 0.0076 0.0041
0.71 0.057 0.19 0.0023 0.0049 nd 0.080 0.011 0.046
0.22 0.058 0.12 0.0004 0.0014 nd 0.049 0.0006 0.014
0.15 0.015 0.056 ndd 0.0009 nd 0.015 0.0004 nd
a Without purification. b Purified by filtration and centrifugation. Purified by filtration, centrifugation, and extraction. d nd, not detected.
was unsuitable for the determination of organotins at ultratrace levels unless a high purity reagent can be obtained. A purification method for a NaBEt4 solution by solvent extraction has been reported.4 The method was modified in the present study by incorporating filtration and centrifugation steps prior to the solvent extraction in order to obtain more pure NaBEt4. The solution, prepared by dissolving NaBEt4 in water, was turbid, and it was not transparent, even after filtration through a 0.45-µm hydrophilic polyvinyldene difluoride membrane filter (Millex-HV, PVC housing, Millipore). However, the solution became transparent after centrifugation, and the impurity was decreased significantly. These facts suggest that large portions of the impurity exist as particulates of less than 0.45 µm in diameter. More recently, it was found that the solution became transparent by filtration with a 0.2-µm hydrophilic PTFE membrane filter, and the impurity level was significantly reduced by the filtration. After centrifugation, the solution was further cleaned by three successive hexane extractions. The pH of the NaBEt4 solution was far from the optimum value (pH 5) for ethylation, which suggests that the organotin impurity was not ethylated and was not removed by hexane extraction at such alkaline conditions. It has been reported that di- and trisubstituted species can be extracted over a wide pH-range but that monosubstituted species, such as MBT and MPhT, are not extracted efficiently under alkaline conditions.14 Thus, the extraction efficiency was checked from the ratio of organotin concentration in the first and second extracts by assuming the same efficiency in both extractions. The extraction efficiencies were in excess of 90% for all the tin species except for inorganic tin for which the extraction was 57%. The high efficiency observed here may be explained by assuming that the organotin species in the NaBEt4 reagent may have been already ethylated during the long storage time and therefore was rapidly extracted, even under alkaline conditions. On the other hand, the freshly added monosubstituted species could not be extracted in such a short extraction time because of the slow reaction rate under alkaline conditions as has been reported in ref 14. It can be seen from Table 2 that inorganic tin, MBT, and DBT still gave higher values, even when purified NaBEt4 was used. The source of this contamination has not yet been identified, but is now under investigation. Detection Limits and Repeatability. The instrumental detection limit for each species is defined as the amount which would Analytical Chemistry, Vol. 71, No. 19, October 1, 1999
4213
Table 3. Detection Limits and Repeatability RSDa
detection limit (fg) instrumental
method
without internal standard (%)
with internal standard (%)
2.6 1.4 0.9 1.6 0.9 0.7 1.3 1.8 1.5 1.4
510 50 4.5 170 4.7 3.8 3.5 35 3.3 3.0
9.1 9.0 9.0 7.9 9.3 8.3 7.1 6.2 6.6 6.0
4.2 1.1
Sn MBT TPrT DBT MPhT TBT TeBT TPeT DPhT TPhT a
1.5 2.1 1.1 2.6 4.3 4.3 5.4
Relative standard deviation.
Table 4. Comparison of Absolute Detection Limits Reported by Various Hyphenated Techniques for the Determination of Butyltin Species technique
detection limita (pg)
ref
GC/FPD GC/AAS GC/MS GC/MIPAES GC/MIPAES GC/MIPAES GC/MIPAES GC/ICPMS GC/ICPMS GC/ICPMS GC/ICPMS GC/ICPMS GC/ICPMS LC/ICPMS LC/ICPMS
0.2 25 0.5-1 0.4 1 0.05b 0.15 0.3-0.8 0.052-0.17 0.015-0.021b 0.05 0.0007-0.0016b 0.0038-0.17 8-9 20-40
4 11 13 18 17 12 14 23 22 21 20 this method this method 7 5
a
Method detection limit. b Instrumental detection limit.
give three times the standard deviation of the integral values of the baseline noise around the respective retention time when injecting 100 µL of hexane. The method detection limit is defined as the amount which would give three times the standard deviation of the peak areas for six replicates of the blank. These values are listed in Table 3. The higher method detection limits encountered for inorganic tin, MBT, and DBT were due to high contributions from the blank. Table 4 compares the absolute detection limits reported for butyl-tin species. It can be seen that the detection limit obtained in this work is the lowest obtained thus far and is approximately 2 orders of magnitude lower than the values obtained with a similar GC/ICPMS, due, primarily, to the operation of the shield torch at normal plasma conditions. Repeatability was evaluated from five replicates of 1 L of seawater spiked with 100 µL of a mixed standard containing ∼5 µg/L of each species. The relative standard deviations of the peak area with and without the internal standardization by TPrT chloride are given in Table 3. It can be seen that the repeatability was improved by the internal standardization. Analysis of Seawater Samples. Seawater samples were collected during the cruise of the R. V. Yoko-maru in October 1998 from the East China Sea. Surface seawater was taken with a polyethylene bucket. Approximately 1 L of seawater was transferred to a Pyrex glass vessel immediately after sampling 4214 Analytical Chemistry, Vol. 71, No. 19, October 1, 1999
Figure 7. Chromatograms of (a) a seawater sample (collected at station 6), (b) a blank, and (c) hexane. I, impurity arising from internal standards; and U, unidentified species.
without filtration on board. This vessel was also used for derivatization and extraction to prevent possible contamination or losses arising from sample transfer because of adsorption on the glass wall. The samples were stored in the dark at ambient temperature and were analyzed 20-22 days after sampling. Since the cruise of the R. V. Yoko-maru was carried out before the evaluation of each derivatization performance, the authors tentatively planned to apply Grignard derivatization, and added HCl and NaCl (without heating) to the Pyrex glass vessel in advance for the extraction of the organotin species as chlorides. However, since the later experiments on the evaluation of each derivatization performance showed that ethylation with NaBEt4 was preferable, the sample treatment was changed from Grignard derivatization to NaBEt4 ethylation. Therefore, the addition of NaCl was actually unnecessary, and the added volume of HCl should have been decreased, although acidification by HCl or CH3COOH below pH 2 is considered to be necessary prior to ethylation to prevent microbial growth and to release organotin species eventually included in the carbonate matrix of the suspended particulate matter. Chromatograms of the seawater sample and the blank containing TPrT and TPeT chlorides as internal standards are shown in Figure 7, along with the baseline obtained with hexane injection. The sample injection volume was set at 25 µL. A shoulder at 5.46 min is due to an impurity in the TPrT chloride, and small peaks at 6.37, 6.92, and 7.52 min are attributed to impurities in the TPeT chloride. Unidentified peaks at 6.51 and 7.82 min might be ascribed to mono and dioctyltin species, respectively, considering
Table 5. Concentration of Organotin Species in Surface Seawater (ng/L) station no.
latitude (.N)
longitude (.E)
inorg Sn
MBT
DBT
TBT
MPhT
DPhT
TPhT
1 2 3 4 5 6 7 8 9b 10 11 12 13 14 15 16 17 18 19
31. 50.1′ 31. 07.9′ 30. 57.0′ 30. 30.2′ 29. 40.0′ 29. 18.5′ 31. 50.0′ 31. 20.1′ 31. 20.1′ 30. 40.7′ 30. 22.0′ 31. 50.0′ 29. 06.0′ 28. 52.5′ 28. 45.7′ 28. 41.7′ 28. 31.1′ 29. 26.3′ 30. 20.1′
127. 19.8′ 126. 54.9′ 127. 28.0′ 126. 39.3′ 126. 46.4′ 125. 49.6′ 126. 38.2′ 126. 17.9′ 126. 17.9′ 125. 59.5′ 126. 04.9′ 128. 00.1′ 126. 14.5′ 126. 39.7′ 126. 54.2′ 127. 04.8′ 127. 23.2′ 127. 11.3′ 127. 11.0′
0.58 0.37 0.36 0.21 0.17 0.26 0.25 0.20 0.68 0.15 0.23 1.38 0.12 0.084 0.32 0.15 0.11 0.42 0.058
0.28 0.046 0.26 0.078 0.23 0.50 0.53 0.48 0.19 0.31 0.75 1.12 0.10 0.081 0.25 0.12 0.15 0.19 0.038
1.15 0.69 0.93 0.73 0.78 0.94 0.80 1.23 2.40 0.77 1.01 0.85 0.52 0.53 0.60 0.58 0.52 0.55 0.12
0.085 0.074 0.081 0.071 0.070 0.41 0.077 0.065 0.041 0.063 0.082 0.42 0.058 0.055 0.10 0.028 0.025 0.052 0.077
0.014 nda nd 0.0011 nd 0.022 0.0066 0.0042 0.0017 0.0051 0.038 0.0073 0.0007 nd 0.0019 0.0011 nd 0.0010 nd
0.0080 0.0037 0.0027 0.0023 0.0030 0.0023 0.0013 0.0039 nd 0.0036 0.014 0.0035 0.0048 nd nd nd nd nd 0.0022
0.0045 nd 0.0032 0.0033 0.0042 0.0029 0.0047 0.0051 0.011 0.0041 0.029 0.0039 0.0037 nd nd nd nd nd 0.0061
a
Not detected. b Sampled at the depth of 10 m.
nucleophilic attack by HCl.3 Taking a closer look at the data, it is interesting to note that the concentrations of butyltin and phenyltin species increase with decreasing distance from the coasts of China and Japan. This study is being carried out in a study of pollution mechanisms of organotin compounds in the East China Sea. Detailed results of this study will be published elsewhere.
Figure 8. Occurrence of TBT in surface seawater in the East China Sea in October 1998. The numbers on the map indicate the station numbers listed in Table 5.
the retention order reported in ref 11. Throughout the analysis of seawater samples, the NaBEt4 solution, which was purified by a single extraction with hexane, was used. This solution and the unheated NaCl were the main sources of the large blank signal for MBT and DBT. Since the concentrations of these species in the sample were considerably higher than the blank values, they did not become a serious problem. However, It should be noted that the triple extraction of NaBEt4 solution should be performed when analyzing the samples of lower organotin species content. Analytical results of seawater samples are given in Table 5 and in Figure 8 to demonstrate the importance of ultratrace speciation in the clarification of the occurrence and behavior of organotin species in the open ocean area. The concentrations were determined against the internal standard of TPrT chloride. TPrT chloride was preferred to TPeT chloride at this low concentration level because unidentified peaks were sometimes observed at the same retention time as that for TPeT chloride. The concentration figures shown for phenyltin species should be considered as indicative values, since it was reported that acidification with HCl to pH 2 caused the degradation of TPhT species through
CONCLUSIONS The present study offers a very sensitive method which makes it possible to perform organotin speciation at the ultratrace level in open ocean seawater for the first time. The improvement of the sensitivity of GC/ICPMS was achieved by using programmed temperature vaporization which allows large-volume injections up to 100 µL and by operating the shield torch at normal-plasma conditions which enhances the signal intensity by 2 orders of magnitude. This enhancement is not considered to be specific to tin but, rather, appears to be more general to other elements. A similar improvement was also observed for mercury. This breakthrough is expected to permit elemental speciation at femtogram levels for a variety of elements. ABBREVIATIONS MBT, monobutyltin; DBT, dibutyltin; TBT, tributyltin; TeBT, tetrabutyltin; MPhT, monophenyltin; DPhT, diphenyltin; TPhT, triphenyltin; TPrT, tripropyltin; TPeT, tripentyltin; PrMgBr, propylmagnesium bromide; PeMgBr, pentylmagnesium bromide; NaBEt4, sodium tetraethylborate. ACKNOWLEDGMENT The authors would like to express their gratitude to Dr. M. Kunugi at National Institute for Environmental Studies and Drs. Y. Kiyomoto and K. Okamura at Seikai National Fisheries Research Institute for collecting seawater samples from the East China Sea. This work was financially supported by the Japanese Environment Agency “Global Environment Research” program. Received for review January 28, 1999. Accepted June 25, 1999. AC990087A Analytical Chemistry, Vol. 71, No. 19, October 1, 1999
4215
