Comparison of microconcentric and membrane-desolvation sample introduction systems for determination of low rare earth element concentrations in surface and subsurface waters using sector field inductively coupled plasma mass spectrometry

Abstract

Analytical performances of a microconcentric nebulizer (MCN) and a membrane-desolvation sample introduction system (Aridus) were compared for determination of low concentrations of rare earth elements (REEs) in surface and subsurface waters using a double focusing sector field inductively coupled plasma mass spectrometer. Conventional figures of merit were employed, such as sensitivities, limits of detection (LOD), REE–O + formation, matrix induced interferences, long term signal variations, and recovery from spiked sea water samples and a pristine water CRM. Sensitivity as a function of mass for the MCN was quite flat with the exception of light REE (LREE +) which was dependent on REE–O + generation. Signal responses using the Aridus were enhanced by factors of 4–10. Large sensitivity enhancements were observed for 139La +, 140Ce + and 141Pr +, due to the attenuation of their REE–O + (where the Aridus 140Ce 16O +/ 140Ce + was three orders of magnitude lower than that determined using the MCN) as a result of water analysis reduction in the plasma. Despite higher sensitivities, Aridus LODs were not significantly different from those obtained using the MCN, and values varied from 0.05 to 0.2 ng L − 1 and 0.1 to 0.2 ng L − 1 for the MCN and Aridus systems, respectively. When using the MCN, LREE + signals were closely associated with REE–O bond strengths, implying that LREE–O + production and breakdown are related to the strength of REE–O bonds in REE–O + ions suggesting that medium REE (MREE)–O + and heavy REE (HREE)–O + if formed, were thermally unstable in the plasma. In the presence of conservative Na concentrations, MCN LREE + signal suppressions varied from about 15 to 25% and about 5 to 10% for MREE + and HREE +. In contrast, regular 15% signal suppressions for all REE + were observed for the Aridus. In the case of the MCN, more than one internal standard is required for compensation for these different effects, whereas a single internal standard ( 159Tb +) compensated for REE + signal variations with the Aridus. Long term uncompensated REE + signal % RSDs using the MCN varied from about 2 to 5%, and % recoveries were about 95–110%. With internal standardization, % RSD varied irregularly from about 1 to 3% and % recoveries were not significantly improved. In contrast, with the Aridus, high % RSDs of about 8% and recoveries of about 80% were compensated with 159Tb + as the internal standard (< 0.5% and recoveries of about 100%, respectively). Accuracy was determined by analyzing spiked sea water CRM (NAAS-5) (1:100 v/v) spiked with 1 µg L − 1 REEs and CRM SLRS-4 using an external calibration procedure. With the Aridus system, REE recoveries were about 100% with 159Tb + as the internal standard. REE determinations of SLRS-4 were in good agreement with the reported values indicating that very low REE concentrations can be determined directly without matrix separation and analyte preconcentration.