Effect of Manufacturing Technique and Cementation Protocol on the Load-Bearing Capacity of 3D-Printed Zirconia Molar Crowns.

This study evaluated the influence of the manufacturing method and cementation protocol on the fracture load of zirconia molar crowns. A total of 32 full-contour crowns with a wall thickness of 1 mm were fabricated from 3 mol% yttria-stabilized tetragonal zirconia polycrystal (3Y-TZP). Sixteen crowns were produced by 3D printing (LithaCon 3Y 210, Lithoz) and 16 by milling (e.max ZirCAD LT, Ivoclar Vivadent). Within each manufacturing group, eight crowns were adhesively luted with a dual-cure resin cement (Bifix QM, VOCO) and eight were conventionally cemented with a glass ionomer cement (Ketac Cem, Solventum). All crown-die complexes underwent thermomechanical aging (6000 thermal cycles between 5 °C/55 °C; 1.2 × 106 cycles at 50 N) as a correlate of five years of clinical service, followed by axial loading to failure. The statistical analysis was a two-way ANOVA on log10-transformed fracture loads with manufacturing and cement as fixed effects (α = 0.05). All crowns survived thermomechanical loading without failure. Milled zirconia crowns showed significantly higher fracture loads when adhesively luted (7162 ± 1294 N) compared with conventional cementation (4021 ± 1126 N; p < 0.001). In contrast, no significant differences were observed between adhesive (4986 ± 1301 N) and conventional cementation (5581 ± 1618 N) for 3D-printed zirconia crowns. Additively manufactured crowns exhibited greater variability in fracture loads (broader interquartile ranges) than their milled counterparts. Two-way ANOVA detected a significant Manufacturing × Cement interaction (p < 0.001), indicating that the effect of cementation differed by manufacturing route. Adhesive cementation increased fracture load for milled zirconia, whereas no statistically significant cementation effect was detected for 3D-printed zirconia under the present conditions. Given the sample size and the observed variability, the study was powered to detect very large effects but not medium or small ones; therefore, this non-significant difference should not be interpreted as proof of no effect.