Specimens embedment
To simulate the biological width a circumferential wax wire was adapted below the CEJ (wax wire round, Ø 2 mm, Dentaurum). The specimen embedment procedure follows a validated protocol to simulate the tooth mobility [30]. The roots were coated with a thin layer of autopolymerizing acrylic resin (Paladur, Kulzer, Hanau, Germany). After polymerization the adhesive was applied on the resin-coated roots. A-polysiloxane soft cushion material (Mollosil, DETAX, Ettlingen, Germany) was placed into the simulated socket and specimens were relocated with the aid of the positioning unit into the mould. Throughout the fabrication process of all-ceramic crown restorations the specimens were stored in humid conditions.
Post-and-core restoration (group C/Li and C/CR)
The post space cavity within the root canal was prepared with a tapered drill (RelyX Fiber Post Drill Size 2, 3M, Seefeld, Germany) to achieve a post length of 10 mm leaving at least 4 mm of the root filling in the apical portion (n = 20). Prior to post placement the post space was rinsed using the intermittent flush technique with two sequences of 2.5 ml 1% sodium hypochlorite (NaOCl) applied with passive ultrasonic irrigation (Irri Tip; VDW, Munich, Germany) for 30 s respectively and followed by 5 ml distilled water. The final irrigation sequence was 2.5 ml 99% ethanol solution for at least 60 s. Glass-fiber reinforced composite posts (RelyX Post, Size 2 RF; 3M) were adhesively luted using a self-adhesive resin composite (RelyX Unicem 2, 3M) applied using “Elongation Tips” (3M). After initial light curing for 2 s, access material was removed and final light curing was performed for 40 s (Valo broadband LED light, WL: 395–480 nm, 1.400 mW/cm; Ultradent Dental Medizinische Geräte GmbH & Co. KG; Brunnethal, Germany).
Immediately after post placement the core build-up was fabricated with an autopolymerising composite (Clearfil Core, Kuraray, Okayama, Japan). The corresponding adhesive Clearfil NewBond (Kuraray) was applied in an etch-and-rinse approach according to the manufacturer´s instructions. The core build-ups were performed by using the silicon impression moulds, which were taken from each specimen before crown removal.
Crown restoration-retentive preparation design (group C/Li and C/CR)
All teeth were prepared using Guide-pin-Diamonds (tapered chamfer, round, size 021, grit size coarse 151 µm and fine 46 µm, Egg size 023, grit size fine 46 µm, Komet Dental, Gebr. Brasseler GmbH & Co. KG, Lemgo, Germany). A chamfer design (circumferential depth of 1 mm) was prepared for the retentive crown-design to meet all-ceramic crown requirements. The composite core was standardized in height at 2 mm with a convergence angle of 12°. The finishing line was prepared in dentine, following the CEJ, 2 mm apical from the core build-up to ensure an appropriate ferrule design. Temporary crown restorations (Luxatemp, DMG, Hamburg, Germany) were prepared with the aid of Frasaco strip crowns (transparent strip crown, size 113, Frasaco GmbH, Tettnang, Germany) and cemented (TempBond NE, Kerr, Biberach, Germany).
Non-retentive crown restoration, no post-and-core and no ferrule preparation design (group NRC/Li and NRC/CR)
After endodontic treatment the orifices of the root canals were adhesively filled up to the level of crown removal using Clearfil Core with no additional core build-up. The finishing line was prepared at the level of decapitation without a separate chamfer line. The outline surface was prepared to create an angle of 90–100° between the outer tooth surface and the prepared area. The level of preparation was waved with a concave design at the tooth center (Fig. 1).
Endo-crown restoration- pulp chamber extension (group EC/Li and EC/CR)
In addition to the occlusal preparation design of the non-retentive crown restoration a 3 mm deep oval pulp chamber was prepared. The remaining circumferential tooth hard tissue was 2 mm. A circumferential bevel of 1 mm was prepared to create a smoothed transition to the pulp chamber extension (Fig. 2).
Fabrication of indirect restorations
A quadrant-model with adjacent teeth using an occludator unit whereat only the prepared tooth could be replaced was created as a scan-model to ensure a comparable crown design for each specimen. The moulds with embedded specimens were screw retained into the scan-model to position the prepared specimen in an alignment with adjacent and antagonistic teeth. The computer aided imaging was performed using an intraoral scanner (True Definition scanner, 3M). The lithium-disilicate crowns (IPS e.max CAD LT A3, LOT18887, 15529; Ivoclar Vivadent, Schaan, Liechtenstein) were fabricated as full anatomical restorations with similar ceramic wall thickness (1.3–1.5 mm at buccal/palatal aspect; 1 mm at crown margin; 2 mm at incisal edge). Crowns were manufactured strictly adhering to the milling and sintering procedure recommended by the manufacturer. After fitting the restorations on the dies a final glaze firing was conducted using IPS e.max Ceram Glaze Paste (Ivoclar Vivadent). No additional stain or characterization firing was applied.
The restorations made of CAD-CAM composite resin (Lava™ Ultimate A3, 3M, LOTN596977, 596977) were milled according to manufacturer’s instruction and polished for finalization.
Adhesive luting of final crown restorations
Prior to luting procedure the intaglio of the crown restorations were prepared for adhesive luting. The crowns made of lithium-disilicate ceramic were acid etched with 9% hydrofluoric acid (Porcelain Etch, Ultradent) for 20 s. After cleaning with waterspray for additional 20 s and air drying Monobond Plus (Ivoclar Vivadent) was applied on the surface for 1 min followed by 1 min evaporation in an air stream. The CAD-CAM composite resin crowns were sandblasted for 10 s (CoJet™, 3M, 4 bar, 10 mm distance). After cleaning with isopropanol and air drying a multimode adhesive (Scotchbond™ Universal, 3M) was applied by brushing on the surface for 20 s, followed by gently air drying for additional 20 s and light curing for 10 s (emittance 1,400 mW/cm2 at a distance of 1–2 mm; Valo, Ultradent).
Selective enamel acid etching with a 32% phosphoric acid etching gel (Scotchbond Universal Etchant, 3M) was performed at all specimens for 20 s. After water rinsing the surfaces were gently air dried and the adhesive (Scotchbond™ Universal, 3M) were applied in motion onto the preparation surfaces for 20 s. After air drying for 20 s the adhesive was light cured for 10 s (Valo, Ultradent). The dual curing luting composite (RelyX Ultimate, 3M) was applied on restoration surfaces and finaly positioned with slight finger pressure. The excess material was removed, Airblock material (Liquid Strip, Ivoclar Vivadent) was applied and light-curing was performed for 45 s from each crown aspect (Valo, emittance 1.400 mW/cm2 at a distance of 0 mm).
Replication of the luting interface
For qualitative analyses of the restoration margins identification marks were prepared below the margins in dentine at the mesial and distal aspect. Further marks were prepared at the palatinal aspect at intervals of 5 mm. Impressions were taken at the palatinal area (Honigum Pro-Light, DMG, Hamburg, Germany) and lined with epoxy resin in vacuum (Stylcast 1266, Emerson & Cuming, Westerlo, Belgium). The epoxy specimens were coated with a gold layer of 20 nm (Sputter Coaster SCD030, Detax, Ettlingen, Germany; in Argon gas, 0.05 bar, 40 mA).
A second replica of the restoration margin was produced after artificial aging of the specimens in the same way as described above.
Scanning electron microscopy (Cam Scan Maxim 2040; Cam Scan Maxim Elektron Optics, Cambridge, UK) was performed to analyse the replica at a magnification of 200x using defined criteria: 1. no change of marginal integrity, 2. change of marginal integrity, i.e. more gap or crack formation, and 3. not assessable.
Artifical aging
All specimen were subjected to thermo-mechanical fatigue loading (TCML) including 5000 thermal cycles (5°C/55°C, two minutes each cycle) and 1.2 × 106 mastication cycles along the tooth axis. A force of 50 N with 1.7 Hz was applied at the centre of the occlusal plane using a steatite ball with a diameter of 6 mm.
Linear loading
The specimens that survived TCML were subjected to static linear loading in a universal material testing machine (Zwick 1446, ZwickRoell, Ulm, Germany; load application in oral – buccal direction at the buccal cusp, loading angle 30°, cross-head speed of 1 mm/min) until failure occurred. Failure was defined as 10% loss of maximum applied force. For even stress distribution, a 0.3 mm thick tin foil was positioned between the steel piston and the specimens.
For all specimens maximum failure load value Fmax [N] were recorded. Fracture patterns were determined after a visual inspection (magnification 2.5x).
Statistical analysis
The data were analyzed for non-normal distribution (Kolmogorov-Smirnov test). A non-parametric Kruskal-Wallis test was used followed by Mann-Whitney tests to determine statistical differences in maximum load capability between the experimental groups. To study differences in frequency of failure mode a Chi-square test was performed and the data were pooled for main fracture patterns. For qualitative margin analysis images of the marked areas before and after TCML were compared according to three categories: 1. no change of marginal integrity, 2. change of marginal integrity, i.e. more gap or crack formation, and 3. not assessable. The analysis were performed using the software SPSS Statistics Version 25 (IBM Corp., ) (α = 0.05).