CHOICE OF A SUITABLE MATERIAL FOR CONSTRUCTION OF A BATTELLE TYPE IMPACTOR TO MINIMIZE SYSTEMATIC-ERRORS IN SAMPLING OF AIRBORNE DUST

Citation
H. Bayer et al., CHOICE OF A SUITABLE MATERIAL FOR CONSTRUCTION OF A BATTELLE TYPE IMPACTOR TO MINIMIZE SYSTEMATIC-ERRORS IN SAMPLING OF AIRBORNE DUST, Mikrochimica acta, 119(1-2), 1995, pp. 167-176
Citations number
14
Categorie Soggetti
Chemistry Analytical
Journal title
ISSN journal
00263672
Volume
119
Issue
1-2
Year of publication
1995
Pages
167 - 176
Database
ISI
SICI code
0026-3672(1995)119:1-2<167:COASMF>2.0.ZU;2-Q
Abstract
Systematic errors can occur when airborne dust is sampled with an impa ctor for subsequent element analysis. As could be shown, blank values, collection losses and memory effects are caused by the impactor and t he material it is made of. Four different materials were chosen to bui ld up a two-stage Battelle type impactor: Makrolon(R), titanium, alumi nium and stainless steel. The latter was applied without and with a co ating of TIN. Air dust was collected on Vaseline(R) coated impaction p lates and directly analyzed by total reflection X-ray fluorescence (TX RF). Up to 15 elements were determined simultaneously after internal s tandardization. Blank values arising from the impactor walls in contac t with the gas flow were measured by sampling air from a clean bench. Relative collection losses were determined by sampling ambient air wit h two impactors in parallel and by comparing the two sets of results. Memory effects were measured by first collecting airborne particulates and then, after having replaced the impaction plates, pulling air thr ough the impactor in a clean bench. The three effects led to relative systematic errors of a few % up to several 10% depending on element an d impactor material. Stainless steel causing the largest errors has to be considered the least suitable material, even if coated with TIN. A luminium led to moderate systematic deviations. Only titanium and Makr olon(R) were proven to be widely suitable impactor materials, the latt er one, however, only when short sampling periods are applied.