With the theoretically predicted relative standard deviation (RSD) of
measurements as a criterion, the ultra-violet detection wavelengths ar
e optimized in a capillary electrophoresis system equipped with a diod
e-array, when analyte signals are baseline-separated in the time-direc
tion and the measurements are the integration results in the time and
wavelength directions. Some cold drugs and additives are taken as exam
ples. The auto-correlation of an instrumental baseline at a diode (tim
e-correlation) and the cross-correlation of measurements between diffe
rent diodes (wavelength-correlation) are both taken into account for a
better prediction of RSD. A simple method for estimating the waveleng
th-correlation is presented with a theoretical proof. This paper also
shows that the measurement precision is governed to a large extent by
the zero level setting of the time integration which accumulates the i
nstrumental output relative to the zero level. The optimum (minimum RS
D) is selected from among the wavelengths (190 - 300 nm) and possible
zero level settings. The validity of the RSD prediction is statistical
ly verified.