Fractal analysis at varying locations of clinically failed zirconia dental implants

Previous studies have shown that the fracture toughness of ceramics can be determined from the fractal dimensions (D) of their fracture surfaces and that the surface should be leveled to obtain an accurate D measurement. This study was to determine the effects of leveling operations and distance from the failure origin on the D values.

Twelve clinically failed zirconia implants from four different manufacturers: Axis Biodental (n=7), Z-Systems (n=3), Straumann (n=1), and Swiss Dental Solutions (n=1) were obtained from one of the authors and thoroughly cleaned. Epoxy replicas were made of three locations along the crack path in the center region of each fracture surface (near origin (O), hackle (H), and near compression curl (CC)) using a light body polyvinyl siloxane impression material. Surfaces were scanned in ScanAsyst mode with a scan size of 5μm×5μm and a scan rate of 0.592Hz using the atomic force microscope. The surface scans were then leveled using 1st order flattening operation in the AFM analysis software. The height data before and after the operation were imported into a custom MathCAD script, and FRACTALS software was used to determine the D value by Minkowski Cover algorithm, which was shown previously to be the algorithm with the highest precision. A Wilcoxon signed-rank test, two-way repeated-measures ANOVA, and one-way repeated-measures ANOVA were performed as detailed below.

The data were not normally distributed (S-W p≤0.05), so a non-parametric repeated measures test (Wilcoxon signed-rank test) was selected. The median D values before and after leveling were 2.161 and 2.174, respectively. There was a significant difference before and after leveling (p<0.001). The two-way repeated-measures ANOVA showed no significant difference among the D values for different implant brands (p=0.66) and scanning locations on the fracture surface (p=0.83). After eliminating the implant brand as a factor, the data passed normality and equal variance tests (S-W p=0.88, BF p=0.15). The mean D values and standard deviations from the three locations (O, H, CC) were 2.183±0.031, 2.179±0.024, and 2.175±0.018, respectively. One-way repeated measures ANOVA showed no significant effect of scanning location (p=0.74).

The leveling operation successfully removed the tilt without decreasing surface tortuosity, as it increased the D values significantly. The fractal dimension was the same at the three locations on the fracture surfaces. This means that hackle and compression curl regions can be used to determine fracture toughness when the failure origin has been lost.