Remarks about data reliability in experimental and computational alloy thermochemistry

The determination of phase equilibria in multi-component systems, which is fundamental to the synthesis, thermal treatments and application of several materials, may be efficiently approached through a combination of experimental measurements and computational methods. To this end, however, reliable experimental data should be available, even if obtained for a reduced set of crucial composition and temperature conditions. Among these experiments a relevant and pre-eminent role is played by thermal and thermodynamic measurements. The estimation of uncertainties in thermochemical measurements, however, is especially difficult due to the presence of errors from different sources (such as sample impurities, side reactions, deviations from equilibrium, etc.) which, mainly in alloy science, are generally larger than the instrumental errors and not easy to evaluate. On the basis also of the research carried out in the Author's laboratory, various sources of errors related to a few experimental techniques in alloy thermochemistry are examined. Some comments on the reliability and consistency of results from thermochemical optimisation, which are heavily affected by the experimental data uncertainty, (especially when large temperature and/or composition extrapolations are needed) are also given. A few comments about the overall coherence of the data concerning a certain alloy system are presented and the need for enhancing and developing experimental alloy thermodynamics underlined. (C) 2003 Elsevier Ltd. All rights reserved.