2.1

Evaluation of interfacial integrity

Twelve extracted maxillary and mandibular human premolars were stored in a saline solution at 4 °C and used within 3 months of extraction. Single-rooted premolars free of caries and fractures were selected ( Fig. 1 a). Protocols were approved by the Ethics Committee of Osaka University. The crowns of teeth to be restored with post cores were removed with a low speed diamond saw (Isomet III, Buehler, Lake Bluff, IL, USA) at the cemento-enamel junction. Teeth were then endodontically instrumented with a size 40 file, and the root canals were obturated with gutta-percha and a resin-based endodontic sealer (AH-Plus, Dentsply, York, PA, USA) ( Fig. 1 b). The premolars were then divided into three groups, with four roots in each ( Fig. 1 c).

Flared root canals with at least 1 mm of dentin remaining in the walls were prepared by endodontic drills to two-thirds of the root depth in eight endodontically obturated roots. The thickness of the dentin walls was standardized at 1 mm by marking reference points on the top surface of the roots and preparation was performed by measuring wall thickness with digital calipers (Digimatic Calipers, Mitsutoyo, Tokyo, Japan). The thickness of the dentin walls was confirmed with bucco-palatal and mesio-distal radiographs . Ferrules with a height of 2.0 mm were also prepared.

The 12 premolars were divided into three groups of four specimens in each. Those in the FRP group were then reconstructed with a 1.1-mm optical FRP (i-TFC fiber post, Sun Medical, Shiga, Japan). Specimens in the FRP/sleeve group were reconstructed with an FRP wrapped in a hollow tubular fiber sleeve (i-TFC fiber sleeve, Sun Medical). The remaining four obturated roots served as a control group. In these roots, a 1.3-mm cylindrical post space with ferrules was prepared and restored with an FRP alone. After conditioning with 3% EDTA for 2 min followed by irrigation with saline solution, the post space in all teeth was coated with a self-etching adhesive agent (Hybrid Coat II, Sun Medical). Fiber materials pretreated with a silane coupling agent (Porcelain Liner M, Sun Medical) were then bonded with resin cement (i-TFC post resin, Sun Medical) and light-cured for 40 s. All specimens were stored at 37 °C with 100% humidity for 24 h. To prevent additional light penetration through the thinned dentin wall, all roots were wrapped with opaque adhesive tape.

The flared post space in each root was divided into three areas—cervical, middle, and apical—according to vertical location. Two 1.1-mm slabs were obtained from each area by slicing perpendicular to the tooth axis with a low-speed diamond saw ( Fig. 1 d). Six slabs were obtained from each prepared root, for a total of eight slabs from each vertical region from each group.

To evaluate the integrity of the resin–dentin and resin–fiber interfaces, a 1.0% acid red propylene glycol dye solution (Caries Detector, Kuraray Noritake Dental, Tokyo, Japan) was applied for 5 s, then rinsed and gently dried. A digital image of each slab was obtained with a stereoscopic system (NIS Elements D 3.00, Nikon, Melville, NY, USA) at a magnification of 40×. The ratio of the gaps at each interface was calculated using image analysis software and expressed as a percentage of the entire length of the interface.

The results were presented by means and standard errors, since averages of two values from the each third vertical location were used. Three-way analysis of variance (ANOVA) and Student–Newman–Keuls tests were used to analyze the effects of restoration type, vertical location in the root, and adhesion interface on the ratio of gaps at a significance level of α = 0.05.

2.2

Fracture testing

Thirty-two single-rooted maxillary and mandibular human premolars were selected ( Fig. 1 a), and their bucco-palatal and mesio-distal dimensions and root lengths were measured with digital calipers (Digimatic Calipers). The teeth were then divided into four groups ( n = 8 teeth each) so that there were no significant differences among groups in terms of bucco-palatal or mesio-distal dimensions, according to the Bartlett test and ANOVA at a 95% level of confidence.

Flared roots were prepared and restored using the same methods as those described for FRP and FRP/sleeve groups in the interfacial integrity evaluation ( Fig. 1 b and c). In addition, the non-bonded FRP and FRP/sleeve groups were prepared as the non-bonded equivalents of the FRP and FRP/sleeve groups, respectively. That is, the restoration procedures were the same, except that the adhesive agent was replaced by a resin separator (Super-Bond SEP, Sun Medical), leaving the fiber materials non-bonded to the root dentin.

All flared roots were restored with a core build-up (i-TFC core resin, Sun Medical) with a standardized shape produced with a plastic former. After restoration with a metallic full crown (Casting Gold Type II, Morita, Osaka, Japan) luted with resin cement (Panavia F 2.0, Kuraray Noritake Dental), the root surface was coated with approximately 200 μm of polyvinylsiloxane impression material (Duplicone, Shofu, Kyoto, Japan) to simulate the periodontal ligament. Finally, the restored specimen was embedded in an epoxy-acrylic resin block (Quetol 812, Nisshin EM Corporation, Tokyo, Japan) at a depth 2 mm below the cemento-enamel junction. All specimens were stored at 37 °C and 100% humidity for 24 h before fracture testing.

A 45° oblique load was applied to the center of the functional cusp of the restored teeth at a crosshead speed of 0.5 mm/min with a universal testing machine (Autograph AG-IS 20-KN, Shimadzu, Kyoto, Japan) until fracture ( Fig. 1 e) . Crack propagation areas were stained with caries detector dye and identified with a stereomicroscope at 40× magnification. Crack propagation was classified into one of three categories: cervical (fractures extending ≤1/3 the length of the root longitudinally from the cervical portion), middle (fractures extending between 1/3 and 2/3 from cervical toward apical portion), or apical (fractures extending longitudinally to the apical third of the root).

Fracture loads were recorded and compared among groups using two-way ANOVA and the Student–Newman–Keuls post hoc test. The distributions of crack propagation were compared with Fisher’s exact test. All statistical analyses were performed at a significance level of α = 0.05.

2.3

Finite element analysis ( Fig. 1 f)

ABAQUS 6.10 software (Rising Sun Mills, Providence, RI, USA) was used to perform FEA. A representative intact premolar was scanned with micro-computed tomography (XT H 225, Nikon Metrology, MI, USA), to obtain 1295 slices. Of these, 648 were selected for modeling and converted into a digital 3D finite element model. Thereafter, a digital representation of a flared root canal was created, in which the thickness of the remaining dentin wall in the coronal 2/3 of the root was constant at 1 mm, by enlarging the root canal.

The properties of materials used for the analyses were summarized in Table 2 . All materials were considered homogeneous, linear-elastic, and isotropic, with the exception of the orthotropic FRP. A finite element mesh was generated with linear isoparametric tetrahedral elements; 620,584 tetrahedron elements and 116,132 nodes were created. The nodes in the upper and lateral portions of the bone were fixed at 6 degrees of freedom.

A set of digital models ( Fig. 2 a; Model 0 represents the original mode) was established for each of the four experimental groups used in the fracture testing. In the FRP and FRP/sleeve groups, fiber materials were programmed to be perfectly bonded to dentin, while in the non-bonded FRP and FRP/sleeve groups, the resin–dentin interfaces were defined as completely non-bonded (i.e., the cement could slide along or separate from the dentin wall). The friction coefficient was 0.3 .

Finite element analysis of the premolars with flared root canals restored with fiber materials under bonded and non-bonded conditions. (a) A series of models was established for each group to simulate the progress of interfacial failure during static loading. The white interfacial lines indicated by arrows represent the interfacial detachment modes identified. (b) Stress distribution in the original mode (Model 0) in each group. (c) The maximum principle stress developed in the components including dentin, resin cement and fiber materials of the restored teeth during progressive interfacial failure.

To simulate the initiation and propagation of interfacial failure during the fracture test loading, a series of models was developed for each group according to the extent of interfacial failure along the resin–dentin interface ( Fig. 2 a, Models 1–5). A 100 N static load was applied at the center of the lingual cusp at a 45° angle to the long axis of the tooth. The stress distributions and magnitudes were recorded for each model.