KALMAN FILTERING OF DATA FROM OVERLAPPING LINES IN INDUCTIVELY COUPLED PLASMA-ATOMIC EMISSION-SPECTROMETRY

In inductively coupled plasma-atomic emission spectrometry (ICP-AES) spectral interferences give rise to severe problems. Detection limits worsen considerably, and line selection depends on matrix composition. It is investigated whether the detection limits can be reduced towards values obtained from isolated lines by use of a multicomponent analysis method, such as the Kalman filter. Our Kalman filter approach uses as many data as are readily available prior to analysis. The flatness of the innovations sequence is shown to be a criterion for correction of the optical instability of the monochromator. Spectral scans have intentionally been taken from a medium resolution monochromator in a fast scanning mode. The spectral scans do not require background correction in advance. The filter can handle structured as well as unstructured background and allows for noise averaging. Detection limits are obtained which are even lower than those based on pure component solutions. This is illustrated by the measurement of Nd in the presence of Pr, P in the presence of Cu, Cd in the presence of Fe and In in the presence of Mo and W. These measurements give examples of overlapping lines with identical shape, with large or small peak separations and/or with large intensity differences. The results suggest that line selection is almost independent of matrix composition. The approach is fast, deals with multicomponent systems and with one-component systems, and yields accurate concentration values in all cases. Data reduction in ICP-AES can be fully based on Kalman filtering. © 1990.