Objectives: To measure subcritical crack growth parameters in vitro and to correlate those with clinical observations from the 12 years recall of a prospective clinical study.
Methods: Bending bars were manufactured and the inert fracture strength was determined in four-point bending and silicon oil. Weibull statistics were applied and the parameters m and sigma(0) were calculated. Dynamic fatigue experiments were performed in water at four decreasing loading rates from 1.3 to 0.0013MPa/s. The parameters of subcritical crack growth n and A were calculated. Strength-fracture probability-life time (SPT) predictions were derived for 1, 4, 8 and 12 years, based on a static crack growth mechanism. The 12 years clinical recall of a prospective clinical study on the same material was performed. Bulk, chipping and marginal fractures or detoriations were focused here. Failure rates were calculated according to Kaplan-Meier survival analysis and merged into the SPT diagram.
Results: Inert fracture strength of the glassceramic was measured to sigma(0)=134MPa and the Weibull modulus to m=8.1. The subcritical crack growth parameter n was calculated to n=19.2 and the extrapolated crack velocity to A=0.0014m/s. Based on a clinical relevant failure probability of P(F)=5%, material strength was predicted to decrease from initial sigma(0.05)=93MPa down to sigma(0.05)=33MPa after 12 years (-64%). The clinical survival rate dropped from 100% (1 year) to 93% (4 years), 92% (8 years) and 86% after 12 years. The incidence of inlay defects like chipping and marginal fractures increased from 1% at baseline, 7% after 4 years, 26% after 8 years to 57% after 12 years.
Significance: Clinical data match the slow crack growth measurements in terms of dramatically increased clinical bulk fractures and detoriations from marginal and chipping fractures after 12 years. Clinical survival rate seems to converge towards the in vitro lifetime predictions with increasing time. A failure level of P(F)=5% is clinically exceeded after 4 years of clinical service, which corresponds to an experimental prediction of fracture releasing static loading of 35.5MPa. Since average chewing pressure over time is less deleterious compared to static loading, this regression analysis provides a conservative threshold value.