Wavelength calibration and spectral sensitivity correction of luminescence measurements for dosimetry applications: method comparison tested on the IR-RF of K-feldspar

TL;DR

This study compares two methods for wavelength calibration and spectral sensitivity correction in luminescence measurements, demonstrating their impact on spectral interpretation and dating accuracy in K-feldspar IR-RF analysis.

Contribution

It introduces a comparison of SRF approximation methods using calibration lamps versus manufacturer curves, highlighting their effects on spectral analysis accuracy.

Findings

Both correction methods produce similar results broadly.

Differences in correction methods can cause 3-13 nm shifts in peak wavelength.

Calibration lamps are recommended for SRF determination.

Abstract

Spectroscopic investigations provide important insights into the composition of luminescence emissions relevant to trapped-charge dating of sediments. Accurate wavelength calibration and a correction for the wavelength-dependent detection efficiency of the spectrometer system are crucial to ensure the correct spectrum interpretation and allow for its comparison with those obtained from other systems. However, to achieve an accurate detection efficiency correction, it is necessary to obtain the device-specific spectral response function (SRF). Here, we compare two SRF approximation methods by using either a calibration lamp of known irradiance or calculating the product of efficiency curves provided by the manufacturers of all known optical elements. We discuss the results using radiofluorescence (RF) measurements of two K-feldspar samples as an example. Feldspar infra-red (IR) RF spectra are known to be composed of several overlapping emissions, whose variation with sample mineralogy is still poorly understood and requires more extensive investigations. We find that both methods of sensitivity correction yield broadly similar results. However, the observed differences can alter a spectrum's interpretation. For example, we observe that after peak deconvolution the maximum signal wavelength of the IR-RF peak used for dating applications differs by ~3-13 nm between the two methods, depending on sample and diffraction grating. We recommend using calibration lamps to determine a device's SRF but highlight the need to consider issues such as higher-order signals in the choice of filters to establish the SRF's reliable wavelength range. Additionally, we find that a simple and inexpensive fluorescent white light yields an acceptable wavelength calibration comparable to that obtained from a specialized light source.