THE DIRECT INTRODUCTION OF POWDERS FOR PLASMA EMISSION WITH FLUIDIZED-BED AND SLURRY NEBULIZATION METHODS - PARTICLE FLOW-RATE AND ANALYTE HETEROGENEITY

The magnitude of measurement uncertainty caused by particle sampling statistics in the direct ICP-AES analysis of powders was investigated using silica-immobilized 8-hydroxyquinoline with analyte elements spiked homogeneously and at varying degrees of heterogeneity. At panicle flow rates above 10 000 particles (25 μm diameter) per measurement, the statistical uncertainty in the number of particles (Poisson statistics) was less than uncertainty introduced from other sources. Uncertainty originating from sample heterogeneity (analyte elements distributed on a fraction of the sample particles) was over 10% when analyte was distributed on fewer than 1 in 100 particles (flow rates about 25 000 particles per measurement). The measured uncertainty in the emission of Cu agreed well with theory, other elements (Cd, Co) showed considerably higher than expected relative standard deviation. Internal standard compensation for signal noise originating from variation in the particle flow became ineffective as the sample heterogeneity increased. Measurement of emission signals of homogeneous powder samples showed an apparent level of "effective heterogeneity" which set the lower limit to which internal standardization effectively compensated for noise. Increasing the numbers of sample particles by grinding the sample to the smallest possible size is shown to be the most effective practical approach to decrease contribution of particle sampling statistics to the measurement uncertainty. Instrumental improvements which permit simultaneous measurement of analyte lines and off-peak background offer promise for reduction of measurement uncertainty. © 1990.