Fracture toughness dependence on grain size in molybdenum silicide, titanium silicide and aluminum nitride

The fracture toughness of MoSi2, Ti5Si3 and AlN as a function of grain size was measured using the controlled-flaw method in conjunction with the miniaturized disk-bend test (MDBT). The materials investigated had grain sizes of 3.5, 7.0, 11.2 and 16 mum for MoSi 2, 2, 4, 7 and 10--20 mum for Ti5Si 3 and 2.5 and 4.5 mum for AlN. The specimens used in the experiments were 3 mm in diameter and varied in thickness from 280 to 593 mum. These were indented using a Vickers pyramid indentor to indentation loads varying from 10 to 80 N. Indentation cracking was experienced at all indentation loads and R-curve behavior was exhibited. The fracture toughness, Kinfinity, was calculated using a straightforward graphical procedure involving an empirical R-curve equation. Kinfinity of MoSi2 was determined to be relatively grain-size independent, with a value of ∼4 MPa·m1/2. Kinfinity, of Ti5Si3 showed a strong dependence on grain size, with maximum of 3.56 +/- 0.41 MPa·m1/2 at a grain size of ∼4 mum. For AlN Kinfinity decreased from 2.85 +/- 0.40 to 2.32 +/- 0.21 MPa·m1/2 as the grain size increased.;The grain growth behavior of polycrystalline MoSi2, Ti 5Si3 and AlN was studied during static annealing at 1400°C in an argon atmosphere. MoSi2 exhibited abnormal grain growth, Ti5Si3 showed normal grain growth behavior and annealing of AlN produced no visible increase in the grain size. The presence of Ni in contact with MoSi2 during annealing significantly increased the rate of grain growth. MoSi2 with grain sizes of 11.2 and 16 mum contained traces of Ni and exhibited a decrease in hardness compared to the Ni-free samples with grain sizes of 3.5 and 7.0 mum.;Stresses arising in non-cubic materials during processing, due to the presence of thermal expansion anisotropy, were calculated for AlN, Al 2O3, MoSi2, SiC, Ti5Si3 and ZnS. These stresses, sigmamax, provide information on the maximum grain sizes, ds, that can be tolerated by these materials before they fracture spontaneously without application of an external stress. Values of sigmamax, ds and other material parameters were estimated using a graphical procedure, which allowed normalization of the fracture toughness. Normalized experimental values of fracture toughness for the above-mentioned materials were in good agreement with each other.