A study of emission spectra is presented that investigates the capabil
ities and the limitations of a multicomponent evaluation method, Kalma
n filtering, in ICP-OES. Using computer simulation the possible source
s of error in concentration estimation were investigated. With synthet
ic data we tried to simulate real spectra containing all effects that
can influence the appearance of the peaks within a real spectral windo
w. After this, model solutions with known composition were prepared, a
nalyzed, and evaluated by means of the Kalman filter. Finally, bauxite
matrices were digested and analyzed. The results of the evaluation pe
rformed by the conventional method in ICP emission spectrometry (autom
atic background correction (ABC)) and the Kalman filter were compared
to the certificate for the bauxite standard samples. It will be shown
that in interference-free cases the two evaluation procedures give com
parable results; for more complex spectra, Kalman filtering is superio
r. (C) 1997 Academic Press.