Instrumental noise distribution in electronic absorption spectrometry

Abstract

The instrumental noise distribution in electronic absorption spectrometry was studied by recording 27 instrumental replicates at five concentration levels of a series of dilutions of pyrimidine. It was found that the dependence of noise on concentration was weak, especially at the base of peaks; heteroscedasticity where noise is greatest at the tops of peaks was also not observed. The noise was greatest at the sides of peaks where derivatives were highest, suggesting that a major cause was uncertainty in wavelength calibration. It is proposed that first derivatives of the spectrum can be employed as a weighting function in multivariate calibration.