Optimization of headspace solid-phase microextraction gas chromatography-atomic emission detection analysis of monomethylmercury

Optimum conditions for headspace solid-phase microextraction (HS-SPME) in the analysis of monomethylmercury (MeHg) have been determined. Sodium tetra(n-)propylborate (NaBPr4) is used as derivatization reagent to promote volatility. A simple aluminium bar was used to cool the SPME fiber to about 2 degrees C during the equilibration phase just before extraction. HS-SPME was performed using different fibers. The 100 Rm polydimethylsiloxane (PDMS) and 65 mu m polydimethylsiloxane-divinylbenzene (PDMS-DVB) fibers showed the best results. Although the extraction efficiency for MeHg derivative of the polydimethylsiloxane-Carboxen (PDMS-CAR) fiber is similar to the other fibers, desorption of MeHg derivative from a PDMS-CAR fiber is poor. Factors affecting the HS-SPME process such as adsorption and desorption times, ionic strength (salting-out) and extraction temperature have been evaluated and optimized thoroughly. The highest extraction efficiency for the PDMS fiber was obtained by extraction at a low temperature 2 degrees C immediately after equilibration at 30 degrees C. With the PDMS-DVB and PDMS-CAR fiber improvement of extraction efficiency at lower temperatures is negligible. Repeated extraction out of the same vial revealed that about 30% of MeHg derivative is extracted from the headspace with a PDMS fiber at 2 degrees C and about 70% with a PDMS-DVB fiber. Repeated extraction with two different fiber coatings showed that the PDMS-CAR fiber also extracts about 70% but that the desorption is incomplete. Attempts to improve the desorption failed due to degradation of the MeHg derivate at high injection temperatures. The limit of detection (3 sigma) was 16 pg/L MeHg. The relative standard deviation (n = 8) for 100 pg/L of MeHg was found to be 5%. Linearity of the HS-SPME-GC-atomic emission detection method was established over at least two orders of magnitude in the range 0-2000 pg/L. Recovery of a surface water sample spiked at 2 ng/L was 85%. The suitability of the procedure was demonstrated by analysis of a surface water sample that showed a concentration of 100 pg/L MeHg. The optimized method can be used with standard commercial equipment without further adaptations. (c) 2007 Elsevier B.V. All rights reserved.