Comparison of Two-Phase Thermal Conductivity Models with Experiments on Dilute Ceramic Composites

Thermal shock resistance of cubic 8mol% yttria-stabilized zirconia (YSZ) can be increased by the addition of dilute second phases. This study addresses how these dilute second phases affect the thermal conductivity for two-phase ceramic composites of 8mol% YSZ with 10-20vol% alumina (Al2O3) or 10-20vol% mullite (3Al2O32SiO2). Thermal conductivity measurements from 310K (37 degrees C) to 475K (202 degrees C) were made using the 3 method and compared with results from 3D analytical models and a 2D computational microstructure-based model (Object-Oriented Finite Element Analysis, OOF2). The linear Rule of Mixtures was the least accurate and significantly overestimated the measured thermal conductivity at low temperatures, with errors in some cases exceeding 100%. Calculations using the Bruggeman and OOF2 models were both much better, and the deviation of less than +/- 2.5% across all compositions and temperatures is within the range of experimental and modeling uncertainty. The Maxwell Garnett equation was a close third in accuracy (+/- 8%). A sensitivity analysis for each model quantifies how small perturbations in the thermal conductivity of the dispersed second phase influence the effective thermal conductivity of the composite, and reveals that the linear Rule of Mixtures model is physically unrealistic and oversensitive to the thermal conductivity of the dispersed phase.