An Inter-comparison of µXRF Scanning Analytical Methods for Lake Sediments

Abstract

The acquisition of high-resolution geochemical data from wet sediment cores through µXRF scanning is an increasingly important analytical tool. A number of µXRF core scanners are in use today that measure elemental concentrations using slightly different methods and several correction methods have been proposed in order to generate more precise quantitative geochemical data. However, only few inter-device comparisons have been undertaken to date and this contribution addresses this deficiency. The main focus is based on a sediment core 1.44 m in length extracted from Loch of the Lowes, Southern Uplands of Scotland, and analysed using a Geotek XZ MSCL carrying an Olympus Delta XRF (University of Liverpool) and an ITRAX core scanner housed at the National Oceanography Centre—Southampton (NOCS). The core is strongly laminated but layers are not consistent in terms of thickness or frequency, thus it provides a good test of the comparative analytical resolution of the devices. The cores were measured at 5 and 1 mm intervals using the Geotek scanner and steps of 0.3 mm on the ITRAX instrument. Seven elements were selected for this inter-comparison: the Geotek 5 mm measurements of Ti, Zr, K and Rb concentration picked up all phases of enhanced mineral supply to the lake as characterised by light-coloured sediments. The finer detail (mm-scale laminations) was more effectively captured by the Geotek 1 mm and ITRAX scans, however Zr, Sr and Rb measured on the ITRAX were significantly more spikey or noisy in character; in particular, peaks and troughs in ITRAX count rate of similar amplitude to the light/dark layers also appeared across intervals with no visible stratigraphic variability. The Geotek less effectively discriminated thin layers within a sediment matrix of geochemically-different lithology. The decision as to which scanning resolution to choose depends on the research question being asked. If abrupt sedimentological or climatic transitions are not under investigation, then lower resolution scans, which are well within the capability of the 5 mm Geotek runs, may be sufficient to capture palaeoenvironmental change. The Itrax can also be operated to run in such a low-resolution mode but on this occasion this was not selected. Another approach investigated was to externally calibrate the Geotek data using a regression correction method (proposed by Boyle et al. 2015, this volume) was tested and consistent dry-mass equivalent concentrations were found for both techniques despite variable statistical relationships. We thus recommend that future research presents µXRF data in terms of dry-mass concentration to facilitate more effective method and field site intercomparison. A final instrument comparability test was performed by measuring a short laminated sandstone section and a homogenous, parallel faced obsidian piece on the ITRAX (NOCS), the Geotek and an Eagle III μXRF system (NOCS). As an example, the Ti profiles through the laminated sandstone output from the three instruments showed some differences: clear peaks and troughs in the ITRAX measurements effectively tracked the alternating light and dark layers while the Geotek returned a more smoothed signal. Interestingly, the Eagle III data exhibit several Ti peaks that do not fully correspond to layer thickness. For the obsidian piece, the signal-to-noise ratio was good in all cases but the relative major element composition varied between devices, possibly due to different instrument configurations’ keeping this firmly in mind when comparing geochemical data thus seems critical.