The multlelement atomic fluorescence-emission spectrometric system (continuum source of excitation) has been evaluated for indlvldual noise contributions as a function of flame type, modulation approach, and atom type. The flames studied included Ar-shielded airlacetylene, Ar-shielded N20/ acetylene, Ar-shlelded N,O/propane, and an air/acetylene flame with a liquid fuel component (Isooctane and jet engine oil). The modulatlon methods included AM (amplitude modulation) and WM (wavelengths modulation) as well as CW (contlnuous wave excltation with DC detection). The elements and wavelengths studied Include Cd (228.2 nm), Mg (285.2 nm), Cu (324.7 nm), Ca (422.7 nm), and Na (589.0 nm); Bi (306.8 nm) was studled in the air/acetylene flame only. Useful concluslons were that the N,O/propane flame was less useful than the two acetylene-based flames, that AM is superior for transitlons wlth wavelengths 5350 nm and WM for ones 5350 nm, that analyte emission/fluorescence flicker noise becomes slgnificant for analyte concentrations above -1OOX the limltlng detectable concentratlon, and that the
air/acetylene flame should always be used instead of the N,O/acetylene flame unless atomizatlon is insufflcient to obtaln reasonable signal levels.A signal-to-noise ratio approach is used to evaluate and optimize the multielement atomic fluorescence spectrometer (MEAFS) previously described (1). A review and tutorial discussion of noise and signal-to-noise ratios in analytical spectrometry have already been given (2-4), and so no attempt will be made here to review the basic concepts. In this manuscript, the total measured noise is partitioned between individual noise components; the influence of flame type,