Resistance to bond degradation between dual-cure resin cements and pre-treated sintered CAD-CAM dental ceramics

Objective: To evaluate the bond stability of resin cements when luted to glass-reinforced alumina and zirconia CAD/CAM dental ceramics. Study design: Eighteen glass-infiltrated alumina and eighteen densely sintered zirconia blocks were randomly conditioned as follows: Group 1: No treatment; Group 2: Sandblasting (125 µm Al2O3-particles); and Group 3: Silica-coating (50 µm silica-modified Al2O3-particles). Composite samples were randomly bonded to the pre-treated ceramic surfaces using different resin cements: Subgroup 1: Clearfil Esthetic Cement (CEC); Subgroup 2: RelyX Unicem (RXU); and Subgroup 3: Calibra (CAL). After 24 h, bonded specimens were cut into 1 ± 0.1 mm2 sticks. One-half of the beams were tested for microtensile bond strength (MTBS). The remaining one-half was immersed in 10 % NaOCl aqueous solution (NaOClaq) for 5 h before testing. The fracture pattern and morphology of the debonded surfaces were assessed with a field emission gun scanning electron microscope (FEG-SEM). A multiple ANOVA was conducted to analyze the contributions of ceramic composition, surface treatment, resin cement type, and chemical challenging to MTBS. The Tukey test was run for multiple comparisons (p < 0.05). Results: After 24 h, CEC luted to pre-treated zirconia achieved the highest MTBS. Using RXU, alumina and zirconia registered comparable MTBS. CAL failed prematurely, except when luted to sandblasted zirconia. After NaOClaq storage, CEC significantly lowered MTBS when luted to zirconia or alumina. RXU decreased MTBS only when bonded to silica-coated alumina. CAL recorded 100 % of pre-testing failures. Micromorphological alterations were evident after NaOClaq immersion. Conclusions: Resin-ceramic interfacial longevity depended on cement selection rather than on surface pre-treatments. The MDP-containing and the self-adhesive resin cements were both suitable for luting CAD/CAM ceramics. Despite both cements being prone to degradation, RXU luted to zirconia or untreated or sandblasted alumina showed the most stable interfaces. CAL experimented spontaneous debonding in all tested groups. Key words:CAD/CAM ceramic, alumina, zirconia, resin cement, surface pre-treatment, sandblasting, silica-coating, chemical aging, bond degradation, microtensile bond strength.


Introduction
Densely sintered zirconia and glass-infiltrated alumina have expanded clinical indications for all-ceramic tooth-and implant-supported fixed dental prostheses (FDPs) (1,2). Even when CAD/CAM-based ceramic restorations can be fixed with conventional cements such as zinc-phosphate or resin-modified glass-ionomer, adhesive cementation has been recommended for improving the clinical retention and marginal fit (1). Nevertheless, searching for a suitable luting strategy to achieve durable bonds between resin cements and high-strength ceramic cores is still a matter of concern (3,4). Alternative porcelain surface treatments such as sandblasting or silica-coating have been suggested, as neither hydrofluoric acid etching nor silanization result in a satisfactory bond to alumina or zirconia due to the absence of a silicon oxide phase in their composition (4,5). However, the influence of conditioning methods on resin bond strength and durability has not been accurately quantified for CAD/CAM dental ceramics (6,7). New dual-cure resin cements, such as the phosphate monomer-containing Clearfil Esthetic Cement and the self-adhesive RelyX Unicem have widely been indicated for luting sintered ceramic cores (4,6,8). However, little information is available in the literature about the longevity of these bonds (8,9); and many factors, e.g., ceramic wettability, porcelain surface roughness or the bonding agents' composition and performance may affect the quality of the resin cement/ceramic adhesion (10). Furthermore, acidic compounds in dentinal fluids, salivary enzymes, and proteolytic residues produced by oral bacteria may hamper the stability of adhesive interfaces (10,11), and have recently been considered as potential sources of chemical bond degradation (10,12). Besides, bonded ceramic restorations promote higher crevicular fluid accumulation (with bacterial products and host-derived factors) than do dental tissues (13,14). Based on the presence of sodium azide in artificial saliva, immersion in a sodium hypochlorite aqueous solution (NaOCl aq ) has been proposed as a suitable and less time-consuming aging technique (15,16) that may reproduce in just a few hours the long-term hydrolytic effect of the mentioned bond biodegradators present in saliva (12,(17)(18)(19). Thus, NaOCl solutions have been des-cribed as potent deproteinizing and biological oxidants with the capability of accelerating natural bond deterioration (18). Such storage medium has been considered to have similar efficacy and lower variability than water aging or thermocycling (16). Because the aim of this study was to test the resistance to chemical degradation of different dual-cure resin cements luted to pre-treated alumina and zirconia ceramic surfaces, immersion of bonded specimens in NaOCl aq was performed. The null hypothesis tested was that neither ceramic composition, nor conditioning method, nor resin cement type influence the bond stability of resin cement/ceramic interfaces.

Material and Methods
-Experimental design Eighteen cubic-shaped (edge = 19.5 mm) sintered and glass-infiltrated (15 vol % quartz glass) blocks of alumina (batch no. 7803, Al Cubes for Cerec, Vita Zahnfabrik; Bad Säckingen, Germany) and eighteen cylindershaped (Ø19.5 mm × 5.25 mm high) sintered zirconia blanks (batch no. 18004627, Cercon Zirconia, Dentsply; Konstanz, Germany) were selected for the study. CAD-CAM aluminum and zirconium oxide ceramic blocks were randomly divided into three groups (n = 6 for each ceramic type) according to the different pre-treatments carried out: Group 1: No surface treatment; Group 2: Sandblasting using 125 μm aluminum-oxide (Al 2 O 3 ) powder (Supradental; Madrid, Spain) applied perpendicularly to the ceramic surface for 10 s at a working distance of 5 mm under a pressure of 75 ± 10 psi; and Group 3: Tribochemical silica-coating with 50 μm Al 2 O 3 -particles modified by silica oxide (Supradental; Madrid, Spain). Thirty-six composite specimens (height = 4 mm) were made by layering 2 mm-thick increments of a microhybrid composite (batch no. J27435, Tetric Evo Ceram, Ivoclar-Vivadent; Schäan, Liechtenstein) using a square silicon mold for the alumina cubes and a rounded one for the zirconia ceramic cylinders. Each composite film was condensed with a clean plastic filling instrument to avoid contamination, and light-cured for 40 s at 600 mmW/ cm 2 (BluePhase, Ivoclar-Vivadent; Schäan, Liechtenstein). The last increment was compressed using a glass microscope slide in order to obtain a flat surface. After removing the specimens from the e671 mold, the portions that were previously in contact with the silicone pattern were irradiated for an extra 40 s. -Luting procedure Composite samples were randomly bonded to the treated ceramic surfaces using different resin cements: Subgroup 1 (CEC) used an MDP-containing resin cement: Clearfil Esthetic Cement (batch no. 00002A/0001AB, Kuraray Medical; Okayama, Japan); Subgroup 2 (RXU) used a self-adhesive resin agent: RelyX Unicem (batch no. 245776, 3M ESPE; Seefeld, Germany); and Subgroup 3 (CAL) used a conventional Bis-GMA resin cement: Calibra (batch no. 060112 (base) / 051151 (catalyst), DeTrey Dentsply; Konstanz, Germany). All materials were handled following the manufacturer's instructions at room temperature (RT) of 23.0°C ± 1.0°C. The chemical composition and application mode of the cements tested are detailed in table 1.
The ceramic-to-composite luting procedures were carried out by means of a customized metallic tool that produced a constant seating pressure of 1 kg (1,249 MPa). The compressive force was applied for 5 min, leaving the resin cement to set in the self-curing modality. Finally, the specimens were photo-activated for 40 s on each side of the blocks (BluePhase: 600 mmW/ cm 2 ) to ensure optimal polymerization. Bonded specimens were stored in distilled water for 24 h at 37 °C.

-Microtensile bond strength test
After 24 h of water storage, all samples were vertically sectioned under water cooling into 1 ± 0.1 mm 2 sticks using a slow-speed diamond saw (Accutom 50, Struers GmbH; Copenhagen, Denmark) according to the "nontrimming" method of the microtensile test. One-half of the specimens from each experimental subgroup were tested for microtensile bond strength (MTBS). The Apply the primer on the ceramic surface and air-dry.

ChEMICAl COMpOsITIOn AnD ApplICATIOn MODE OF ThE REsIn CEMEnTs TEsTED In ThE sTuDy
Mix the cement pastes.
Apply the mixture on the ceramic surface.
Mix the cement.
Apply the mixture on the ceramic surface, self-cure (5 min) and photocure (40 s). Apply the silane on the ceramic surface.

Calibra
Gently air-dry after 5 s.
Light-cure for 20 s.
Apply the mixed cement on the ceramic surface. Selfcure (5 min).

Results
Ceramic composition, surface treatment, resin cement type, and NaOCl aq immersion significantly affected bond strengths to CAD-CAM ceramics (p < 0.001). Interactions were also significant except the interaction between ceramic composition and challenging procedure (p = 0.558        Representative FEG-SEM images are presented in figures 1 to 3. Sandblasted substrates evidenced edgeshaped micro-roughness ( Fig. 1) whereas silica-coated beams showed a slight undulation in their surface texture with homogeneously distributed micro-irre-gularities (Fig. 2). At 24 h, remaining cement layers mainly luted to pre-treated ceramics were observed (Figs. 3,4). After NaOCl aq immersion, cement residuals of CEC (Figs. 5,6) and RXU (Fig. 7) persisted above the ceramic substrates.

Discussion
Resin cement/ceramic bonded interfaces are susceptible to degradation. Challenging these bonding sites at the laboratory is required to test the bond stability of different luting systems proposed for oxide-based ceramics (1). A microtensile test was performed, since it allows for a homogeneous distribution of stress across the adhesive interface and a sensitive evaluation of bond strength (20). The findings of the present study require rejection of the null hypothesis because ceramic composition, conditioning treatment and resin cement type influenced the bond stability at the resin cement/ ceramic interfaces. At 24 h evaluation, the bond strength of CEC to pretreated zirconia reached the highest MTBS values ( Ta-ble 2). Prior to CEC application, the ceramic primer was spread over the porcelain surface to improve adhesion and to protect against moisture. 3-MPS (3-methacryloxypropyl trimethoxysilane) and 10-MDP (10methacryloxy-decyl-dihydrogen phosphate) monomers present in the CEC ceramic primer composition (Table  1) optimize the surface wettability (21) and create crosslinks: (a) with the 10-MDPs dispersed in the CEC resin matrix and (b) with the hydroxyl radicals (OH-) of the zirconia surfaces (2,22). Consistently, the main failure type recorded in the CEC groups before challenging was mixed (Table 3).
In the first minute of the mixture of RXU, the pH is purported to be low, less than 2 pH units, due to the presence of multifunctional phosphoric methacrylates ( Table 1). Even though this acidity seemed to promote adhesion to a glass-porcelain (IPS Empress, Ivoclar Vivadent) in a previous study (23), it remains to be ascertained whether this chemical interaction also occurs when bonded to CAD/CAM ceramics (8). In fact, RXU mostly failed adhesively at 24 h (Table 3). CAL confirmed pre-testing failures with complete detachment from the ceramic substrates except when luting sandblasted zirconia surfaces (Tables 2 and 3) due to their rough texture (4,8). An AFM analysis developed in a former investigation revealed a significant increase in the average roughness of zirconia surfaces after sandblasting (4). The lack of chemical interaction between CAL and the tested ceramics may be due to the absence of adhesive functional monomers in the composition of the resin cement and the silane coupling agent (24). Actually, the high content of solvents in the silane formulation may interfere with the resin polymerization, jeopardizing adhesion. Concerning the ceramic type, glass-reinforced alumina has been reported to be rather resistant to air-abrasion and/or chemical surface reactions (25). This may explain why CEC and RXU attained comparable MTBS results despite the alumina surface pre-treatment (Table 2). Several in vitro methods have been proposed to replicate the clinical conditions that may cause the adhesive interfaces to fail (15). Short-term immersion in 10 % NaOCl aq has previously been used to reproduce the long-term action of salivary enzymes (10,18,19,26). In the current research this solution hampered the bond strength of CEC and CAL when combined with either type of ceramic and conditioning method. After NaOCl aq storage, the structural deterioration of CEC was observable such that the resin portion seemed to be dissolved displaying protruding filler particles (Figs. 5,6). The use of resin cements containing adhesive phosphate monomers has been deemed the best luting strategy for both alumina (27) and zirconia (28) mainly after air-abrading the ceramic surfaces. Nevertheless, microstructural changes in the CEC resin ce-e676 ment were noticeable in the comparison of FEG-SEM images taken before and after chemical aging (Figs. 4,6 respectively). Sandblasting is also supposed to somewhat compensate for the lack of adhesive monomers in the CAL Bis-GMA matrix (4,8), but it has not contributed to bond stability (Table 2). Moreover, the 3-MPS hydrophilic monomers of the silanes used prior to CEC and CAL resin cements may have expedited interfacial sorption and hydrolytic effects thanks to the establishment of hydrogen bonds throughout the hydrophilic groups (29). This could justify the bond strength decrease of both cements after challenging.
In RXU debonded specimens, cement dissolution areas (with rounded margins) were detected by FEG-SEM after chemical aging (Fig. 7). The adhesive interface of RXU has been compared to that of some conventional luting agents such as silicate or zinc-phosphate cements (30). The inorganic fillers (glass silicate) may react with the acidic phosphoric ester forming a silicate gel in which glass particles are entrapped. Therefore, RXU luted to zirconia and untreated or sandblasted alumina demonstrated greater bond stability than that of the CEC and CAL subgroups. Possible differences in effective silica deposition and nanomorphological alterations produced by silica-coating on alumina and zirconia substrates deserve further investigation. Despite not being the focus of the study, it might contribute to explain why RXU decreased MTBS only when luted to silica-coated alumina. Overall, given that silanized interfaces become unstable in contact with moisture (1), RXU was the only resin cement capable of maintaining bond strength after aging, as it requires neither primer application nor silanization (Table 1). Saliva contamination has been shown to affect the resin bond to CAD/CAM ceramics and its durability (10). Accordingly, filtrations of the NaOClaq solution at the adhesive interfaces have lead to degradation of the resin matrix of both CEC (Figs. 5,6) and RXU (Fig. 7) polymeric materials (19,26), thus increasing the percentages of adhesive failures in most groups (Table 3). Nonetheless, the results of this experiment provide only an indication of the possible performance of resin cements to zirconium and aluminum oxide-based ceramics and should be reasonably extrapolated to the clinical environment. Further in vitro long-term storage studies as well as controlled clinical trials are needed to redefine these findings (1,8).
Within the limitations of this study, the following conclusions could be drawn: (a) resin-ceramic interfacial longevity depended on cement selection rather than on surface pre-treatments; (b) the MDP-containing (CEC) and the self-adhesive (RXU) resin cements provided adequate bond strength levels to alumina and zirconia CAD/CAM ceramics after challenging; (c) despite both cements being prone to degradation due to resin matrix dissolution, RXU luted to zirconia or untreated or sandblasted alumina showed the most stable interfaces; and (d) the conventional Bis-GMA resin cement (CAL) experimented spontaneous debonding in all tested groups regardless of the ceramic surface treatment.