Two methods are presented for dealing with variable background signals in radiofrequency glow discharge optical emission spectroscopy (r.f.-GDOES). Their aim is to improve elemental analysis at trace levels, in bulk analysis or in compositional depth profiling, without having to measure background signals during the analysis. Each method uses background signals measured away from the analytical emission lines of interest during calibration only. The background signal is first determined during calibration for each material type of interest. During analysis in the first method the estimated background signal is varied according to the material type being analysed. In depth profiles this means identifying the various layers present as different material types, hence the name 'layer model'. The second method is a more conventional approach, where part of the background signal is estimated as a spectral interference. Results are presented for the bulk analysis of a tool steel and for two depth profiles: TiO₂ coating on silicon and TiN-coated tool steel. The two methods give similar results in the depth profiles, both significantly better than with a constant background.
Surface and Interface Analysis Vol. 36, Issue 10, p. 1384-1391