Atom diffusion in furnaces — models and measurements

Experimental as well as the theoretical approach to estimate diffusion coefficients for several analyte elements with different behavior in graphite furnaces, lead, gold, indium and chromium, were investigated. ‘Close’ graphite furnaces of two designs differing in the size of end apertures and the diameter of the injection port were used. The furnaces were fast heated at rates of approximately 10 000 K s−1. The peak absorbance of all studied analytes was independent of geometry, suggesting that the separation of atomization and removal was attained. Residence times of the analytes in the two different furnaces were determined from absorbance tail shapes. In the case of gold, the influence of temperature in the range between 1800 and 2200 K on the residence time in both furnaces was also found. The residence times measured in the two different furnaces under otherwise identical conditions, made possible to select the accurate model of diffusional removal from several possible models. The knowledge of the accurate model allowed the estimate of experimental diffusion coefficients. They were thus compared with those semiempirically calculated from kinetic theory of gases, extended to allow for the intermolecular forces. The accuracy of these calculations is limited since the input data (critical temperatures, boiling temperature or melting temperature, molal volumes at the critical, boiling and melting points, metallic crystallographic radii and dissociation constants of metal dimers) are not known with adequate accuracy. The comparison of ‘theoretical’ and ‘experimental’ values of diffusion coefficients makes possible to assess value of using individual sources of input data for the semiempirical calculations.