Specimens from a previous study had a strength distribution that did not fit a two- or three-parameter Weibull model. Fractographic examination revealed that failure originated from multiple flaw types. Objectives: To determine a statistical model for accurately predicting the failure probability of the specimens as a function of stress and stressing rate. Methods: Specimens were prepared from an alumina-zirconia-glass ceramic (In-Ceram® Zirconia; VITA Zahnfabrik). Rectangular beam specimens having dimensions of 25x4x1.2mm were fabricated and randomly assigned to five groups of 30 specimens each. After glass infiltration, excess glass was removed using SiC abrasive (Mizzy Heatless® Wheels). Specimens were subjected to multiple firings to mimic the thermal history encountered by cores of layered all-ceramic prostheses. Specimens were fractured in four-point flexure in deionized water using four different stressing rates (0.01, 0.1, 1.0, and 10 MPa/s). The fifth group was tested in an inert environment (oil) at 1000MPa/s. Results: Fractographic examination revealed that failure sometimes originated from specimen edges (8.7%). After excluding those specimens, remaining specimens failed from either volume flaws associated with incomplete infiltration (4.7%) or surface flaws (95.3%). The data fit well to a partially concurrent flaw model (Johnson, 1979) having volume flaws in some specimens and surface flaws in all specimens. The model parameters corresponding to best fit were 4.8% probability of a volume flaw with Weibull modulus (m_V=30) and characteristic strength (s_0V=527MPa) independent of stressing rate. Surface flaw parameters and correlation were dependent on stressing rate:

	Failure Origin
Stressing Rate(MPa/s)	Surface(%)	Volume(%)	mS	s0S	R2
0.01	92.3	7.7	4.4	348	0.975
0.1	89.7	10.3	3.6	381	0.963
1.0	100.0	0.0	6.9	467	0.974
10	100.0	0.0	18	479	0.977
inert	95.8	4.2	11	677	0.989

Conclusion: A partially concurrent flaw model can accurately predict the failure probability of a framework ceramic with glass infiltration defects. NIH-NIDCR Grant DE013358.