Objectives . To study the fracture mechanism of experimental resin composites with different filler sizes by determining their initial fracture strength (IFS), flexural fatigue limit (FFL) and by observing respective fractographic features.

Materials and methods . Four experimental resin composites were prepared containing 11 wt% of 0.04 μm silanated colloidal silica and further 68 wt% of silanated powder glass fillers of different filler sizes (d ₅₀ /d ₉₀ , respectively): group I: 0.5/0.9 μm; group II: 0.9/1.7 μm; group III: 1.2/2.6 μm; group IV: 1.9/4.5 μm. Disk-shaped specimens (12 mm in diameter; 1.2 mm thick) were produced and stored for 24 h in distilled water at 37 °C prior to testing. IFS ( n = 14) and FFL ( n = 20) were measured using the piston-on-three-balls test in an universal testing machine. FFL was determined under cyclic loading for 10 ⁵ cycles in a staircase approach. All specimens were tested in water at 37 °C. Data were analyzed using ANOVA/Tukey test ( p < 0.05). Fractography from IFS and FFL specimens was performed under a scanning electronic microscope (SEM).

Results . There were no statistically significant differences in IFS among the composites (p = 0.19), varying from 155.4 ± 18.8 MPa in group II up to 170.7 ± 23.1 MPa in group III. However, there were statistical differences regarding FFL ( p < 0.001) in which group I presented the highest (93.0 ± 18.6 ^a MPa) and group II the lowest FFL (82.5 ± 8.0 ^c MPa). There was no statistical difference between groups I and III (91.8 ± 11.1 ^ab MPa), and between III and IV (87.2 ± 3.0 ^b MPa). SEM analysis of the fracture surfaces showed evidence of brittle fracture for both loading regimes. However, after cyclic fatigue the fracture surfaces appeared smoother, while the initial fracture surfaces presented obvious fracture markings and additional material fragmentation.

Conclusions . Based on the composites under investigation, no correlation could be proved between filler sizes below 4.5 μm and IFS / FFL values. Fractographic analysis showed a different appearance comparing fracture patterns on IFS and FFL surfaces. Failure after cyclic loading could be caused by a reduced energy release during fracture. Further, matrix deformation upon visco-elastic creep might contribute, rather than crack deflection in the fatigued specimens.