In order to better characterise a permanent modifier based on iridium deposited on zirconium or tungsten treated platforms of transversely heated graphite atomizer, and to gain additional information about its chemical behavior directed to an eventual further optimization, a series of experiments were carried out, both by surface techniques such as scanning electron microscopy (SEM), X-ray photoelectron spectroscopy (XPS or ESCA) and X-ray fluorescence (XRF) and by electrothermal atomic absorption spectrometry on the iridium release from unmodified and various other modified pyrolytic graphite platforms. Special attention was paid to the influence of the amount of iridium, zirconium carbide coating of the platform surface and the presence of citric acid on the iridium vaporization during pyrolysis and atomization. The processes of iridium losses during pyrolysis and atomization and peak maximum alignment depend on the amount of the iridium deposited on the pyrolytic graphite coated platforms in the presence of nitric acid. A fractional order of release which suggests an atom vaporization from the surface or edges of the iridium islands was estimated. In the presence of citric acid, mass independence and zero order of the atom release were found. The zirconium treatment of the platform results in change of the spatial distribution of iridium and hence its vaporization. Vaporization temperatures as high as 2100 degrees C, and first order of the process of atom generation were obtained, While it was possible to study the iridium atomization from uncoated and zirconium coated surfaces, evidencing a different order for the release process, the same was not possible for the tungsten coated platforms due to an 'overstabilization' that brought the iridium atomization temperature out of the working range of the instrument used, The different chemical behavior of tungsten and zirconium was also confirmed by XPS investigations, With tungsten, evidence of both W-C and W-O bonding was found, while zirconium on the contrary shows only Zr-O bonding and no evidence of carbide bonding. The SEM revealed a highly dispersed distribution of spot-like features whose smallest average diameter was of the order of 0.1 mu m. The XRF asserted the confinement of iridium in these features and a strict association with zirconium in the case of zirconium treated surfaces. It is worth mentioning that such structure was preserved also after 400 thermal cycles simulating an atomization step at 1900 degrees C despite a quite evident deterioration of the graphite surface, thus confirming the excellent durability of this modifier. (C) 1997 Elsevier Science B.V.