Measuring the intensity of spectral lines from discrete counts of line spectra

In atomic emission spectrometry, photodetector arrays are widely used in spectrum analyzers. A spectrum obtained with detectors of this type is a discrete sequence of digital values of photocell output signals. One way to quantify the concentration of an element in a test sample is to measure the intensity of its analytical line by integrating a region of the spectrum over several counts in the vicinity of this line or by approximating a region of the spectrum with a line shape profile. As a rule, due to the high saturation of atomic emission spectra with spectral lines, the region for calculating the spectral line intensity is limited to several counts. In the case of spectral line drift relative to the photocells of photodetector arrays, this limitation leads to an intensity measurement error, which is the greater, the smaller the number of counts used in integration. The objectives of this study are to determine the optimal size of the computational domain and develop an optimal method for calculating the line intensity to reduce the intensity measurement error. To simulate the drift of spectral lines relative to photocells, we have simulated and recorded a set of spectra of a hollow cathode lamp (Cu, Zn) with different positions of spectral lines relative to the photocells of BLPP-2000 and BLPP-4000 photodetector arrays. In each next spectrum of the set, the spectral lines were shifted relative to those in the previous spectrum by 2 and 1 μm for BLPP-2000 and BLPP-4000, respectively. It has been shown that compared to stepwise interpolation, linear interpolation significantly reduces the effect of the drift of spectral lines on the RMSD of the measured intensities for both types of arrays. In addition, this effect can be further decreased by choosing an optimal range of integration. In the linear interpolation for the selected spectral lines, the minimum measurement error due to the spectrum drift relative to the photocells of photodetector arrays for BLPP-2000 is 0.25 and 0.23% for a range of integration of 1.6 and 3.1 counts, respectively, and for BLPP-4000, it is 0.4 and 0.28% for 1.0 and 2.7 counts, respectively.