Improved fluorescence peak integration in the Tekran 2537 for applications with sub-optimal sample loadings

Abstract

The Tekran 2537 is widely used for monitoring atmospheric mercury. Although the instrument was designed for sample volumes in excess of 7.5 L, some recent research applications (e.g. aircraft) have used the instrument with significantly smaller collection times and sample volumes – and therefore smaller Hg loadings per cycle – than for which the instrument was designed. We have noticed a potential for non-linear (low) response in the fluorescence peak integration scheme, and thus the reported concentrations when the Hg loading (per cycle) is less than about 10–15 pg, e.g. at around 1 pg loading, the sensitivity is 25% lower than at 10 pg. We determined that although the atomic fluorescence detector was fundamentally linear down to at least 1 pg, the default peak integration scheme appeared to be optimized for > 10–15 pg cycle −1 and so could introduce non-linearity in smaller peaks (i.e. lower mass loadings). For research applications where achieving maximum accuracy and precision of individual, high-time resolution (<5 min) points is crucial, users can mitigate this behavior by modifying the integration parameters or recording the full fluorescence peak and processing the data offline. Two offline methods of quantifying the peak also improved the precision and thus suggest an improvement in the detection limit is possible.