Micro-tensile bond testing of resin cements to dentin ... - EAPGOIAS

dental materials Dental Materials 18 (2002) 609±621

www.elsevier.com/locate/dental

Micro-tensile bond testing of resin cements to dentin and an indirect resin composite Yiu-Fai Mak a, Shirley C.N. Lai a, Gary S.P. Cheung a, Alex W.K. Chan a, Franklin R. Tay a,*, David H. Pashley b a

Conservative Dentistry, Faculty of Dentistry, University of Hong Kong, Prince Philip Dental Hospital, 34 Hospital Road, Hong Kong SAR, People's Republic of China b Department of Oral Biology and Maxillofacial Pathology, School of Dentistry, Medical College of Georgia, Augusta, GA, USA Received 21 May 2001; revised 10 November 2001; accepted 11 December 2001

Abstract Objectives: Micro-tensile bond strength (mTBS) evaluation and fractographic analysis were used to compare four resin cement systems (AC: All-Bond 2/Choice; RX: Single Bond/RelyX ARC; SB: Super-Bond C&B; and PF: Panavia F) in indirect composite/dentin adhesive joints. Methods: Flat dentin surfaces were created on extracted human third molars. The resin cements were used according to the manufacturers' instructions for bonding silanized composite overlays to deep coronal dentin. 0.9 £ 0.9 composite±dentin beams prepared from the luted specimens were stressed to failure in tension. Dentin sides of all fractured specimens were examined by scanning electron microscopy (SEM) to examine the failure modes. In group PF, morphologic features that could not be resolved at the SEM level were further validated by transmission electron microscopy (TEM) examination of the SEM specimens. Results: Statistical analyses revealed signi®cant difference (p , 0.05) among mTBS and failure modes in the resin cement groups. The two groups (AC and RX) with highest mTBS failed predominantly along the composite overlay/cement interface. Cohesive failure in resin cement was primarily observed in group SB that exhibited intermediate mTBS values. In group PF with the lowest mTBS, failure occurred mostly along the dentin surface. Globular resin agglomerates seen by SEM on PF-treated dentin were distinguished from silica ®llers by TEM. Signi®cance: The bond between the processed composite and the luting resin cement was the weak link in indirect composite restorations cemented with AC or RX. Super-Bond C&B exhibited intermediate tensile strength and Panavia F is less reliable when used in conjunction with a self-etching primer for bonding indirect restorations to dentin. q 2002 Academy of Dental Materials. Published by Elsevier Science Ltd. All rights reserved. Keywords: Resin cement; Micro-tensile; Fractography; Scanning electron microscopy; Coronal dentin

1. Introduction Indirect adhesive procedures constitute a substantial portion of contemporary oral rehabilitation treatment. Metal and metal-free inlays, veneers, crowns, orthodontic brackets, resin-bonded ®xed prostheses and even posts are now routinely bonded to tooth substrates via the use of adhesive resin cements [1±6]. They have become popular clinically because of their ability to bond to both the tooth structure and the restoration, reduced solubility, and more forgiving nature on the accuracy of ®t when compared with the use of non-adhesive luting cements [7]. * Corresponding author. Tel.: 1852-28590251; fax: 1852-25593803. E-mail address: [email protected] (F.R. Tay).

Being un®lled or less highly-®lled, the magnitude of polymerization shrinkage of resin cements can be expected to be higher than direct restorative composites with comparable resin compositions [8]. Unlike direct composite restoratives, resin cements are usually applied in thin layers. The amount of curing stress generated in resin cements varies with the ®lm thickness as well as the degree of compliance (i.e. ability to relieve shrinkage strain) of the bonding substrates [9,10]. In the presence of low compliance (i.e. restricted shrinkage strain), such as bonding of rigid, full coverage crowns or cast dowel cores and posts, the negative effects of polymerization shrinkage in bonded resin cements can be extreme. This is due to the very limited free surfaces around these adhesive joints that are available for ¯ow of the setting cement to compensate for shrinkage

0109-5641/02/$ - see front matter q 2002 Academy of Dental Materials. Published by Elsevier Science Ltd. All rights reserved. PII: S0109-564 1(02)00005-2

610

Y.-F. Mak et al. / Dental Materials 18 (2002) 609±621

stress development during the gel phase of polymerization [11]. Under these circumstances, the bonds formed between low compliant adhesive joints may be spontaneously disrupted by developing shrinkage stresses [12,13]. Such a scenario may be further aggravated by reductions in ®lm thickness [9] and reductions in internal air porosities [14] as well as an accelerated rate of polymerization [15] that alters the viscoelastic properties of the setting resin cement [16]. Bonding of composite inlays and onlays to dentin represent adhesive joints of moderate compliance in which shrinkage stress may be partially or totally compensated by elastic deformation of the bonded tooth [17] and/or composite substrates [18]. In the absence of high C-factors, bonding of a composite overlay to ¯at dentin surfaces further represents a situation of maximal compliance, in which there is unrestricted shrinkage strain (i.e. free curing contraction) of the resin cement. It is likely that these factors are responsible for the favorable results recently reported on the use of contemporary resin cements in indirect resin composite/dentin adhesive joints [19,20]. Under these compliant conditions, tensile strengths and failure modes of the test assemblies are dependent upon the ®lm thickness of resin cements [21]. As adhesion to tooth substrates continues to improve, the issue of interfacial compliance, that was previously only considered valid for strong metalto-metal bonds [21], should be taken into account in crosssectional or longitudinal studies of resin cements to dentin. Adhesion of resin cements to processed composites has traditionally been dif®cult to achieve [22]. Sandblasting of the intaglio surfaces followed by silanization has been recommended as a predictable means for enhancing the retention between resin cements and indirect composite restorations [23]. It is not known whether the strengths of these bonds are comparable with those achieved when the same restorative composite is used for direct restorations. However, comparisons between the bond strengths of indirect and direct composite restorations are dif®cult to interpret, due to the difference in interfacial stress conditions that exist in prepared cavities with ®xed boundary conditions [24], and the use of different materials for indirect and direct restorations. This study examined the micro-tensile bond strengths (mTBS) of four resin cement systems in highly compliant indirect composite/dentin adhesive joints with relatively unrestricted shrinkage strain. All specimens that were stressed to failure were examined by scanning electron microscopy (SEM) for assessment of the percentage distribution of failure modes. Morphologic features that were unresolved at the SEM level were further clari®ed using transmission electron microscopy (TEM). Where applicable, mTBSs obtained for indirect restorations were compared with those derived from direct restorations that were bonded using the same adhesives and restorative composites, under the same condition of relatively unrestricted shrinkage. The null hypotheses tested were: (1) there

is no difference in the mTBS and failure mode distribution among the resin cements when the same ®lm thickness is used for indirect restorations, and (2) there is no difference in the mTBSs of indirect versus direct restorations when the same adhesive and restorative composite are employed. 2. Materials and methods 2.1. Tooth preparation Bonding was performed on the occlusal surfaces of deep coronal dentin of 18 human third molars that were stored in 0.5% chloramine T and used within one month following extraction. The occlusal enamel and roots of the teeth were removed using a slow-speed saw (Isomet, Buehler Ltd, Lake Bluff, IL, USA) under water lubrication to form 5±6 mm thick, parallel-sided crown segments. A 180-grit silicon carbide (SiC) paper was used under running water to create a smear layer of clinically relevant thickness on the surface of the coronal dentin [25]. The bonding surfaces were examined under a stereoscopical microscope (Nikon SMZ10, Tokyo, Japan) to ensure that they were free of retained enamel. The crown segments were randomly divided into six groups each containing three specimens. 2.2. Composite overlay preparation An experimental heat- and light-activated hybrid resin composite (Bisco, Inc., Schamburg, IL, USA) was formulated for the experiment. The rationale for using a composite with dual-activation modes was to enable subsequent comparisons to be made between indirect and direct restorations, in two of the four resin cement systems that utilize universal dentin adhesives. Layers of composite, 5 mm in thickness, were dispensed into 4 £ 2 ¯at Te¯on molds (Electron Microscopy Sciences, Fort Washington, PA, USA). The molds containing the uncured composite were placed inside an experimental composite inlay processing chamber (Nitro-Therma-Lite; Bisco, Inc.) and light-activated under a pressurized nitrogen atmosphere maintained at 551.6 kPa (i.e. 80 psi) for one complete cycle at 125 8C for 10 min. This produced void-free, oxygen inhibition layer-free, polymerized composite blocks that were optimized in their degree of polymerization. After processing, the composite blocks were reduced with an Isomet saw under water lubrication to produce smaller blocks that approximate the dimensions of the teeth to be bonded. Each reduced block was then sectioned with the Isomet saw to produce 3 mm thick, parallel-sided composite overlays. The bonding surface of each overlay was ground with 180-grit SiC paper and further sandblasted with 50 mm alumina for 10 s. This created a roughened surface so that adhesive failures along the inlay±resin cement interface could be readily identi®ed under SEM, after the bonded specimens were stressed to failure.


611

Table 1 Adhesives and resin cements used in this study (4-META: 4-methacryloxyethyl trimellitate anhydride; 5-NMSA: N-methacryloxyl-5-aminosalicyclic acid; BHT: butylated hydroxytoluene (2,6-di-tert-butyl-para-cresol); Bis-GMA: bisphenol A diglycidyl ether dimethacrylate; BPDM: biphenyl dimethacrylate; HEMA: 2-hydroxyethyl methacrylate; MDP: 10-methacryloyloxydecyl dihydrogen phosphate; NTG-GMA: N-tolylglycine-glycidyl methacrylate; TEGDMA: triethylene glycol dimethacrylate; UDMA: urethane dimethacrylate) Adhesive/resin cement

Components

Composition

Lot number

Total etch adhesive All-Bond 2/Choice (Bisco, Inc., Schaumburg, IL, USA)

Uni-Etch

32% Phosphoric acid gel, xanthum gum thickener

0100001442

Primer A Primer B Pre-Bond resin D/E bonding resin

NTG-GMA, acetone, ethanol, water BPDM, photoinitiator, acetone Bis-GMA, TEGDMA, benzoyl peroxide, BHT Bis-GMA, UDMA, HEMA

0000011967 0000011968 0100002118 0000007876

Adhesive paste

0000011679

Dual-cure catalyst paste

Strontium glass, amorphous silica, Bis-GMA, UDMA, photoinitiator Amorphous silica, Bis-GMA, TEGDMA, benzoyl peroxide

0000007490

Conditioner

35% Phosphoric acid, silica thickener

OUF

Self-priming adhesive

Bis-GMA, HEMA, bisphenol A glycerolate dimethacrylate, diurethane dimethacrylate (HEMA-TMDI) a, polyalkenoic acid copolymer, ethanol, water

OER

Two-paste system in Clickere dispenser

Bis-GMA, TEDGMA, dimethacrylate polymer, zirconia ®ller, silica

BHBH

Dual-cure, ®lled resin cement

Total etch adhesive Single Bond/RelyX ARC (3M Dental Products, St. Paul, MN, USA)

Dual-cure, ®lled resin cement

Self-cure, un®lled resin cement with separate enamel and dentin etching protocols Red (enamel) activator 65% Phosphoric acid, polyvinyl alcohol Super-Bond C&B (Sun Medical Co. Ltd, Shiga, Japan) also marketed by Parkell, Inc. (Farmingdale, NY, USA) as C&B Metabond Green (dentin) activator 10% Citric acid, 3% ferric chloride, polyvinyl alcohol, water Monomer Methyl methacrylate, 4-META Polymer (L-type Clear) b Polymethyl methacrylate Catalyst S Partially oxidized tri-N-butyl borane, acetone Self-etching primer

Dual-cure, ®lled resin cement Panavia F (Kuraray Co. Ltd, Osaka, Japan)

a

VV1 EE3 VM2 VX12

ED primer A ED primer B

HEMA, MDP, 5-NMSA, water, accelerator 5-NMSA, accelerator, water, sodium benzene sulphinate

00107B 00114B

Universal paste

Hydrophobic aromatic dimethacrylate, hydrophobic aliphatic dimethacrylate, hydrophilic dimethacrylate, sodium aromatic sul®nate (TPBSS), N,N-diethanol-ptoluidine, surface-treated (functionalized) sodium ¯uoride, silanated barium glass MDP, hydrophobic aromatic dimethacrylate, hydrophobic aliphatic dimethacrylate Hydrophilic dimethacrylate, silanated silica, photoinitiator, dibenzoyl peroxide

00116A

Catalyst paste

b

EE1

00044B

HEMA-TMDI: 4,4,6,16-tetramethyl-10,15-dioxo-, 2-[(2-methyl-1-oxo-2-propenyl)oxy]ethyl ester, a type of UDMA. L-type represents new surface-modi®ed polymer powder with extended working time.

2.3. Bonding of indirect composite overlays Four different adhesive resin cement systems (Table 1) were examined in this study: Choice (Bisco, Inc.), RelyX ARC (3M ESPE, St. Paul, MN, USA), Super-Bond C&B (Sun Medical Co. Ltd, Shiga, Japan; also marketed in North America as C&B Metabond by Parkell, Inc., Farmingdale, NY, USA), and Panavia F (Kuraray Co. Ltd, Osaka, Japan). All-Bond 2 (Bisco, Inc.), a multiple-step dentin adhesive was used in conjunction with Choice. Similarly, Single

Bond (3M ESPE), a two-step self-priming (`single-bottle') adhesive, was applied on the dentin substrate before bonding with RelyX ARC according to the manufacturer's instructions. Prior to bonding, a composite overlay was randomly assigned to each dentin disk and their total thickness was ®rst recorded with a digital micrometer (Digimatic Micrometer, Mitutoyo, Tokyo, Japan). Three crown segments were used for each of the four experimental groups: AllBond 2/Choice (group AC), Single Bond/RelyX ARC

612


Table 2 Techniques of application of the adhesive resin cement systems for indirect composite restorations Resin cement systems

Dentin conditioning

Priming

Luting procedures

Choice

Uni-etch for 15 s, rinse and kept slightly moist Etching gel for 15 s, rinse and kept slightly moist Green activator for 10 s, rinse and dry

Mixed All-Bond 2 primer A and B, airdry light-cured for 20 s; Pre-Bond resin Single Bond, air-dry, light-cured for 20 s Wet etched dentin with monomer 1 Catalyst S

Mixed equal volume of Choice adhesive paste and dual-cure catalyst paste Mixed equal volume of base and catalyst paste from Clickere dispenser Bulk-mix technique: Pre-chilled dispensing dish mixture of monomer 1 Catalyst S and L-type polymer at standard polymer/monomer ratio Mix Panavia F paste A and B, protect with Oxyguard II to ensure anaerobic polymerization

RelyX ARC Super-Bond C&B Panavia F

Mix ED primer A and B, apply for 60 s, no rinsing, brie¯y air-dried

(group RX), Super-Bond C&B (group SB) and Panavia F (group PF). The procedures in which the dentin bonding surfaces were prepared are summarized in Table 2. Before bonding of the composite overlays, the intaglio surface of each overlay was cleaned with a silica-free, 32% phosphoric acid gel (Uni-Etch, Bisco, Inc.), dried with oil- and dust-free air (Dust-Off Plus, Falcon Safety Products, Inc., Branchburg, NJ, USA), and silane-treated using RelyX ceramic primer (3M ESPE). The adhesive cements were all used in their chemical-activation modes, mixed and applied immediately according to the manufacturers' instructions (Table 2). For Choice, a thin layer of All-Bond 2 Pre-Bond resin was applied to the primed dentin to avoid premature setting of the resin cement caused by uncured resin monomers with amine functional group in the oxygen inhibition layer of the cured primer. Upon placement of the adhesive resin cement, each bonded assembly was held centrally between the two measuring arms of the vertically-positioned digital micrometer. The micrometer arms were slowly adjusted to register a reading that was 100 (10) mm (mean ^ SD) thicker than that initially recorded for the respective dentin disk and composite overlay. For Panavia F, Oxyguard II was liberally applied around the resin cement to ensure complete anaerobic polymerization. The bond assembly was left in this position until complete setting of the resin cement. No attempt was made to polymerize the resin cement in the light-activation mode even when such an option was available. 2.4. Direct composite buildups The universal adhesives employed in two of the four resin cement systems (i.e. All-Bond 2 and Single Bond) are indicated also for direct composite restorations. Incremental composite buildups were performed using these two adhesives and the same experimental resin composite in the light-activation mode so that comparisons could be made with the indirect restorations. Three crown segments were used for each adhesive. For All-Bond 2, the Pre-Bond resin was mixed with D/E bonding resin to render the latter autocurable. After adhesive placement, ®ve 1 mm thick increments of the experimental composite were light-activated

separately for 40 s each using a halogen light-activation unit (Demetron Optilux 401, Kerr Co., Orange, CA, USA) with an output of 600 mW/cm 2. 2.5. m TBS evaluation After storage in distilled water at 37 8C for 24 h, each tooth was sectioned occluso-gingivally into serial slabs using an Isomet saw under water lubrication. The two widest slabs from each tooth were further sectioned into 0.9 £ 0.9 composite±dentin beams, according to the technique for the `non-trimming' version of the micro-tensile test reported by Shono et al. [26]. Three teeth from each group yielded 22±27 beams for bond strength evaluation with no premature failures for any of the specimens during sectioning. Specimens were stressed to failure under tension in a Bencor Multi-T device (Danville Engineering, San Ramon, CA, USA), using a universal testing machine (Model 4440; Instron Inc., Canton, MA, USA) at a crosshead speed of 1 mm/min. The dentin side of all fractured beams from each group were air-dried, sputter-coated with gold/palladium and examined with a scanning electron microscope (Cambridge Stereoscan 440, Cambridge, UK) operating at 10±20 kV. They were not dehydrated with organic solvents [27] to avoid the possibility of extracting partially polymerized oligomers or swelling of uncross-linked polymers (such as polymethyl methacrylate (PMMA)) from the fractured interfaces. The exact bonding area of each specimen was derived from image analysis of digitized micrographs using an image analysis software (NIH Image 1.60, Scion Corp., Frederick, MD, USA). 2.6. Statistical analysis The bond strength data obtained for the four resin cement groups AC, RX, SB and PF (i.e. indirect restorations) were statistically analyzed with Kruskal±Wallis one-way ANOVA on ranks, using SigmaStat Version 2.03 (SPSS, Chicago, IL, USA). Statistical signi®cance was set in advance at the 0.05 probability level. Multiple comparisons were done with Dunn's test at a 0:05: In addition, the mTBS of indirect restorations bonded with each of the two universal adhesives (i.e. All-Bond 2 or Single Bond) was


613

Table 3 Micro-tensile bond strengths (mTBS) of the four adhesive resin cement systems following luting of a resin composite overlay to sound dentin (*values are means (SD). Numbers in square brackets are the number of specimens tested. Groups along this column that are identi®ed by the same lower case letter superscripts are not signi®cantly different (p . 0.05). **Values are means (SD). Numbers in square brackets are the number of specimens tested. Direct restorations were compared with indirect restorations along the same row. Groups identi®ed by different symbol superscripts are signi®cantly different (p , 0.001)) Group (designation)

Indirect restoration a (MPa)*

Direct restoration b (MPa)**

All-Bond 2/Choice (AC) Single Bond/RelyX ARC (RX) Super-Bond C&B (SB) Panavia F (PF)

38.2 (8.4) [25] a,² 34.5 (7.6) [22] a,³ 24.7 (3.8) [25] b 16.1 (3.9) [27] c

48.9 (11.4) [22] ²² 51.0 (11.3) [25] ³³ NA NA

a Indirect restorations were performed using an experimental heat- and light-activated resin composite overlay that was pre-cured under heat, light and pressurized nitrogen gas. Resin cements were employed for luting the composite overlays to sound dentin. b For the two systems utilizing universal adhesives, mTBSs of direct restorations using the same composite in the light-activation mode were included for comparison. No resin cement was employed in these two groups.

compared with the mTBS of the corresponding direct restorations, using the Mann±Whitney rank sum test at the same probability level. 2.7. SEM fractographic analysis Fractographic analysis was performed only for the four resin cement groups. Failure modes were categorized as: (a) adhesive failure along the overlay±cement interface; (b) cohesive failure within the resin cement; (c) cohesive failure along the cement±adhesive interface; and (d) adhesive failure along the dentin surface. The fractional area of each

failure mode in a fractured beam was determined from the SEM micrographs using the image analysis software. The area occupied by each failure mode in one group was ®rst determined and expressed as a percentage of the total bonding surface area of that particular group. For statistical analysis, the fractional areas of the four failure modes in a fractured beam were converted to percentage surface areas for that beam. Differences in distribution of the percentage surface area for each failure mode among the four groups were statistically analyzed using Kruskal±Wallis ANOVA on ranks and Dunn's multiple comparison tests at the 0.05 probability level. 2.8. TEM preparation

Fig. 1. Failure mode distribution for the four resin cement systems when they were used for bonding processed composite overlays. The percentage surface area of a particular failure mode in each group represented the ratio of the fractional surface areas exhibited by that failure mode to the total surface area in all the fractured specimens. There was no cohesive failure within the processed composite overlays. Superscript 1: denotes cohesive fracture within the bulk of the resin cement. Superscript 2: in All-Bond 2/ Choice and Single Bond/RelyX ARC, this interface was easily identi®ed. In the case of Super-Bond C&B and Panavia F, this refers to the retention of a thin layer of fractured resin material along the dentin surface (i.e. without exposed dentin).

This procedure was only employed for those resin cement groups in which morphological features could not be resolved at the SEM level. As all fractured beams were utilized for SEM examination, the same beams that have been sputter-coated with gold/palladium were reused for TEM examination. Before retrieval, the fractured surface of each beam to be examined was carefully protected with a drop of electron-lucent, solventfree, light-curable resin (D/E bonding resin, Bisco). The dehydrated beams were ®rst immersed in 100% ethanol for 24 h. They were then placed in propylene oxide for 48 h to remove the gold/palladium coating from those regions that were unprotected by bonding resin, and to act as a transition medium for the in®ltrating epoxy resin. TEM embedding protocol and preparation of 90 nm thick undemineralized sections followed the protocol described by Tay et al. [28]. The sections were collected on single slot, carbon- and formvarcoated copper grids (Electron Microscopy Sciences, Fort Washington, PA, USA). They were examined either without staining, or double-stained with 2% uranyl acetate and Reynold's lead citrate, using a transmission electron microscope (Philips EM208S, Eindhoven, The Netherlands) operating at 100 kV.

614


Fig. 2. SEM micrographs of the dentin side of a representative fractured beam in group AC (All-Bond 2/Choice). (a) Low magni®cation view showing cohesive failure within the resin cement (C) and the adhesive (A). Fracture along the cement±adhesive interface (asterisk) is shown in higher magni®cation in (b). Bar 30 mm. (b) High magni®cation view of the area labeled with the asterisk in (a). Remnant fractured cement (C) were present on the surface of the fractured adhesive (A). Bar 3 mm.

3. Results Signi®cant difference (p , 0.05) was observed among the mTBS of indirect restorations that were bonded with the four resin cements (Table 3). The highest mTBS were obtained using All Bond 2/Choice and Single Bond/RelyX ARC (no signi®cant difference). The lowest mTBS was produced by Panavia F, that was signi®cantly (p , 0.05) lower than the other three resin cements. Super-Bond C&B resin cement gave an intermediate bond strength that was also signi®cantly different from all others. When indirect and direct restorations in groups AC and RX were compared, the mTBS of the indirect restorations in each

group was found to be lower than the corresponding direct restorations, and the differences were both signi®cantly different (p , 0.001; Table 3). Fig. 1 shows the distribution of failure modes in the four resin cement groups. All 99 beams from these indirect restorations debonded within the adhesive joint and there were no cohesive failure within the composite overlay or dentin. In groups AC and RX, a large percentage (46.1 and 72.4%) of the failures were adhesive failures along the overlay±cement interface. There were no mixed or adhesive failures involving exposed dentin in these two groups. However, a substantial percentage (46.1%) of the failures in group AC occurred cohesively between the un®lled


615

Fig. 3. SEM micrographs of the dentin side of representative fractured beam in group RX (Single Bond/RelyX ARC). (a) Low magni®cation view. Absence of polishing groups that were consistently observed in adhesive failures along the inlay±cement interfaces indicated that this was a cohesive failure within the resin cement (C). Air-voids that were incorporated during mixing of the cement could be identi®ed. Fracture of the adhesive (A) at the lower left corner exposed the underlying dentin surface (D) and revealed the full thickness of the adhesive layer (between arrows). Bar 20 mm. (b) High magni®cation view of the fracture cement±adhesive interface. Fractured resin cement (C) was present on top of the exposed un®lled adhesive (A). Angular ®ller particles (pointers) that were about 1 mm in size could be seen within the fractured resin cement. Bar 1 mm.

adhesive and ®lled resin cement. In group SB, the predominant failure mode (73.9%) was cohesive failure within the un®lled cement, the majority of which occurred within the bulk of the cement rather than adjacent to the bonded dentin. Nevertheless, 19.5% of the fractured surface areas involved exposed dentin. By contrast, no adhesive failure was observed along the overlay±cement interface in group PF. Failures were predominantly adhesive failures along the dentin surface (65.9%) and cohesive failures within the cement that were adjacent to the exposed dentin (31.7%). SEM micrographs of the dentin sides of representative

fractured beams from the four resin cement groups are shown in Figs. 2±5. Only characteristic features that were observed in at least ®ve specimens were included. Morphologic features in group PF that could not be resolved at the SEM level were further clari®ed by TEM examination (Fig. 6). In adhesive failures involving dentin (mainly the PF group), failure occurred on top of the 0.5 mm thick hybrid layers. Furthermore, globular agglomerates identi®ed on the surface of PFbonded specimens by SEM and unstained TEM were found not to be resin cement ®llers, as they could pick up heavy metal TEM stains, unlike silica ®llers.

616


Fig. 4. SEM micrographs of dentin side of representative fractured beam in group SB (Super-Bond C&B). (a) Low magni®cation view of a specimen that exhibited cohesive failure within the un®lled PMMA resin cement. The idiosyncratic manner in which stress dissipation occurred in this group resulted in a unique fracture pattern. This was characterized by the presence of a central circular crater (asterisk) that was surrounded by a slightly elevated ridge. The peripheral rim of the fractured beam consisted of circumferential hackles (arrow). Bar 20 mm. (b) High magni®cation view of another specimen showing the cabbage leaf-like, partially detached lamellae at the base of the central depression. Bar 10 mm.

4. Discussion For the two indirect restoration groups that exhibited high mean mTBS (Table 3), a substantial percentage of the failure modes was adhesive failure (Fig. 1) along the composite overlay/resin cement interface (46% for group AC and 72% for group RX). These results are different from those reported by Alster et al. [21]. In that study, thin composite layers were stressed to failure within compliant, silanized metal-to-metal adhesive joints. Failures were exclusively cohesive within the composite layers when their ®lm thickness were between 50 and 400 mm. In the present study,

bonding substrates with different adhesive strengths to the resin cements were located on either side of the adhesive joints. Our results thus con®rmed previous reports that the bond between processed composite inlays and the luting composite was the weak link in indirect composite restorations [29,30]. This is further supported by the higher mTBS obtained when the universal adhesives associated with these two resin cements were employed for direct restorations using the same experimental composite (Table 3). The lower percentage of adhesive failures between the composite overlay and the resin cement in group AC was counterbalanced by a combined increase in cohesive


617

Fig. 5. SEM micrographs of dentin side of a representative fractured beam in group PF (Panavia F). (a) Low magni®cation view showing a mixed failure within the resin cement (C) and along the dentin surface (D). Bar 20 mm. (b) A high magni®cation view of the region labeled `D' in (a). Remnant of the fractured, ®lled resin cement could be seen along the dentin surface. The part of the dentin demineralized by the self-etching primer was also partially detached, and contained exposed denuded collagen ®brils (pointer). At the SEM level, it could not be resolved whether the regions labeled (X) was the surface of the hybrid layer or the subsurface undemineralized dentin. Dentinal tubules were obliterated with smear plugs (arrow), some of which appeared to have been in®ltrated by resin. Bar 1 mm. (c) High magni®cation view of the region labeled `C' in (a). The fractured surface consisted entirely of a porous agglomerate of spherical bodies that were separated by larger voids (arrow). Bar 1 mm.

failures within the cement and adhesive, the predominance of which occurred along the cement/adhesive junction (Fig. 1). This was probably caused by the inclusion of a high concentration of butylated hydroxytoluene, a polymerization inhibitor, in All-Bond 2 Pre-Bond resin. This inhibitor is included in order to temper the accelerated rate of polymerization of the resin cement that is caused by the presence of a tertiary amine-based resin monomer in primer A of this adhesive. The presence of additional inhibitor might have reduced the degree of conversion along the

cement/adhesive junction when this adhesive was employed for indirect restorations. The observation that no adhesive failure occurred along the resin±cement/dentin interface when indirect restorations were bonded using All-Bond 2 or Single Bond as dentin adhesives has to be interpreted with caution before extrapolating to bonding in vital deep dentin. In the absence of dentin perfusion, these two total-etch adhesives created dentin bonds that were stronger than the cohesive strengths of the respective resin cement. However, different results

618


Fig. 5. (continued)

could have occurred under in vivo conditions in the presence of pulpal ¯uid transudation [31] that adversely affect the bond strengths of these adhesives [32]. Prophylactic sealing of indirect preparations with adhesives before impression taking and temporization, followed by their reapplication during the application of resin cements may alleviate potential problems associated with increased dentin permeability after removal of the smear layer in deep vital dentin [33,34]. Contrary to groups AC and RX, cohesive failure within resin cement was the predominant failure mode (74%) in group SB. A unique fracture pattern could also be distinguished irrespective of the variation in failure modes. This consisted of a central crater that was surrounded by a peripheral rim of circumferential hackles [35] within the cohesively fractured resin cement (Fig. 4). From the radial arrangement of the hackles, it appeared that the crack propagated from the center of the beam toward the periphery (as opposed to corner fracture), with the central crater being the crack initiation site (mirror region) [35]. Super-Bond C&B is different from the other resin cements examined in that it is an un®lled cement that consists mostly of linear polymer chains of PMMA. Such a low modulus of elasticity resin cement is likely to undergo plastic deformation under loading [36]. Studies on the fracture behavior of PMMA bone cements showed that debonding in these cements consisted of an initially linear phase with high stiffness until the yield strength was reached (pre-yield phase), followed by an exponential softening phase (post-yield phase) until frank fracture occurred [37,38]. Based on those studies, it is speculated that as the beam in our study was stressed under tension, an initial crack site was produced in the center of the beam. The crack may have eventually propagated along a plane with the largest number of inherent

¯aws, such as the air porosities incorporated within the cement during hand-mixing (not shown) [39]. This may have resulted in the characteristic fracture markings (hackles) along the peripheral rim of the fractured beam. As stress was relieved after crack initiation, subsurface plastic deformation of the linear polymer may have occurred. This may have caused the detachment of the polymer in the form of `cabbage leaf-like' lamellae (Fig. 4). The mTBS recorded for Super-Bond C&B was not as high as groups AC and RX. This is probably due to the absence of ®llers and and/or cross-linking of the polymer. However, the fractographic results suggested that the bonded interface in Super-Bond C&B may not fail abruptly during loading as shown by the cabbage leaf-like lamellae. Such an interface may absorb a substantial amount of energy beyond the yield point, hence demonstrating a creep-like behavior when the specimen was stressed in tension. This may account for the recent report that the fracture toughness of this PMMA cement bonded to two metal alloy plates was 10 times as high as that recorded for a ®lled resin cement [40]. The high magnitude of polymerization shrinkage associated with an un®lled PMMA cement indicates a potential for greater interfacial stresses. However, this undesirable factor may be offset by a reduction of the modulus of elasticity of the set cement [8]. Moreover, with the recent introduction of the L-type surface modi®ed polymer, the curing time of SuperBond C&B is further extended to 8.5±9.5 min. The curing time may further be increased to 15.5±18 min by altering the polymer±monomer ratio. The long setting time of this cement is advantageous in permitting ¯ow of the material during setting to relieve polymerization shrinkage stress. This may account for the excellent track record of Super-Bond C&B in clinically demanding situations in which low and moderate substrate compliance are encountered [41±43].


Fig. 6. TEM micrographs taken from epoxy resin-embedded, undemineralized sections of SEM specimens in group PF. (a) A stained section taken from an area that corresponded with the region labeled D in Fig. 5a. The layer of electron-dense gold/palladium (arrow) that covered the fractured surface was covered by the protective bonding adhesive resin (E) during TEM preparation. The fractured resin cement (R) contained ®ller particles (F). A 0.5 mm thick hybrid layer (H) was retained above the undemineralized dentin (U). Globular structures within the resin cement layer that were 200±300 nm in diameter (pointers). Bar 200 nm. (b) A stained section taken from an area that corresponded with the region labeled C in Fig. 5a. The agglomerates were separated by voids (V) and were made up globular resin bodies (pointer) that were neither ®llers nor bacteria. Bar 500 nm.

Group PF exhibited the lowest mean mTBS among the resin cements investigated. This resin cement system utilizes a selfetching primer (ED primer) for simultaneous etching and priming of dentin. It is also dual-curable and contains polysiloxane-coated sodium ¯uoride ®llers for additional ¯uoride release. Our previous studies on self-etching primers produced by the same company (e.g. Clear®l SE Bond, Kuraray Co. Ltd) showed that they produced much higher bond strengths to dentin [44] than Panavia F. The manufacturer suggested that the bond strength of Panavia F to dentin is low due to its much higher ®ller content compared with the bonding agent in Clear®l SE Bond, and the incorporation of hydrophobic mono-

619

mers that facilitates coupling to metal substrates (Dr Junichi Yamauchi, personal communication). As the other two ®lled resin cement systems investigated also contain hydrophobic monomers, this does not fully explain why the mTBS obtained for Panavia F was signi®cantly lower. When the fractured interfaces of Panavia F were examined using SEM, a layer of globular agglomerates (Fig. 4) was consistently seen along with the fractured resin cement remnants. These 200±300 nm globular bodies were initially assumed to be colloidal silica particles that were incompletely dispersed within the cement during the manufacturing process. Subsequent examination with the use of TEM revealed that they were not silica particles, and were only detectable after TEM staining (Fig. 5). In our previous report on the adverse reaction between chemically-activated resin composites and acidic singe-bottle adhesives, similar globular structures were also observed along the fractured composite/adhesive interfaces, albeit to a lesser extent [45]. We speculate that they were responsible for the low bond strength observed in Panavia F. Self-etching primers are more acidic than single-bottle, self-priming resins by virtue of their increased concentration of acidic resin monomers [46]. In Panavia F, the resin cement is directly applied over the surface of the ED primer-treated dentin and allowed to polymerize via a self-curing mechanism. This may allow suf®cient time for acidic monomers from the ED primer to diffuse into the setting resin cement for an adverse chemical interaction to occur. Although both ED primer and Panavia F resin cement contain aromatic sulphinate salts as ternary catalysts, a high concentration of uncured acidic monomers is probably present within the primed dentin. Aromatic sulphinate salts are effective oxygen scavengers (reducing agents) that reduce the thickness of the oxygen inhibition layer and the concentration of uncured acidic monomers therein. However, in the absence of acid scavengers, an adverse reaction may still occur between the tertiary amines of the resin composite and the phosphoric functional groups of incompletely polymerized or even optimally converted acidic monomers. In conclusion, the resin cements in this study were tested under the condition of maximal substrate compliance with relatively unrestricted shrinkage. This was re¯ected by the absence of premature failure when specimen beams were prepared for mTBS evaluation. This leads to a rejection of the ®rst null hypothesis. Moreover, under the same condition of relatively unrestricted shrinkage, indirect and direct restorations fabricated with the same adhesives and experimental restorative composite exhibited signi®cant differences in their mTBS and this could be due to the prolonged light curing of the bonding resin during the curing of composite build up in the direct restorations. We therefore have to reject the second null hypothesis. Acknowledgements This study was based on the work performed by Y.F. Mak

620


for partial ful®llment of the Advanced Diploma in Endodontics, the University of Hong Kong. We thank Amy Wong of the Electron Microscopy Unit, the University of Hong Kong for technical assistance. The Super-Bond C&B used in this study was generously supplied by Sun Medical Co. Ltd. The experimental heat- and light-activated inlay composites were generously supplied by Bisco Inc. This study was supported, in part, by the Faculty of Dentistry, the University of Hong Kong and by grant DE 06427 from the NIDCR, USA. The authors are grateful to Michelle Barnes for secretarial support.

References [1] Manhart J, Scheibenbogen-FuÈchsbrunner A, Chen HY, Hickel R. A 2-year clinical study of composite and ceramic inlays. Clin Oral Invest 2000;4:192±8. [2] Peumans M, Van Meerbeek B, Lambrechts P, Vanherle G. Porcelain veneers: a review of the literature. J Dent 2000;28:163±77. [3] Mitchell CA, Abbariki M, Orr JF. The in¯uence of luting cement on the probabilities of survival and modes of failure of cast full-coverage crowns. Dent Mater 2000;16:198±206. [4] Knox J, Kralj B, Hubsch PF, Middleton J, Jones ML. An evaluation of the in¯uence of orthodontic adhesive on the stresses generated in a bonded bracket ®nite element model. Am J Ortho Dent Orthop 2001; 119:43±53. [5] el-Mowafy O, Rubo MH. Retention of a posterior resin-bonded ®xed partial denture with a modi®ed design: an in vitro study. Int J Prosthod 2000;13:425±31. [6] Mannocci F, Innocenti M, Ferrari M, Watson TF. Confocal and scanning electron microscopic study of teeth restored with ®ber posts, metal posts, and composite resins. J Endod 1999;25:789±94. [7] Platt JA. Resin cements: into the 21st century. Comp Cont Educ Dent 1999;20:1173±8 see also p. 1180±2. [8] Labella R, Lambrechts P, Van Meerbeek B, Vanherle G. Polymerization shrinkage and elasticity of ¯owable composites and ®lled adhesives. Dent Mater 1999;15:128±37. [9] Alster D, Feilzer AJ, de Gee AJ, Davidson CL. Polymerization contraction stress in thin resin composite layers as a function of layer thickness. Dent Mater 1997;13:146±50. [10] Alster D, Venhoven BA, Feilzer AJ, Davidson CL. In¯uence of compliance of the substrate materials on polymerization contraction stress in thin resin composite layers. Biomaterials 1997;18:337±41. [11] Feilzer AJ, De Gee AJ, Davidson CL. Increased wall-to-wall curing contraction in thin bonded resin layers. J Dent Res 1989;68:48±50. [12] Davidson CL, Van Zeghbroeck L, Feilzer AJ. Destructive stresses in adhesive luting cements. J Dent Res 1991;70:880±2. [13] Verzijden CW, Feilzer AJ, Creugers NH, Davidson CL. The in¯uence of polymerization shrinkage of resin cements on bonding to metal. J Dent Res 1992;71:410±3. [14] Alster D, Feilzer AJ, De Gee AJ, Mol A, Davidson CL. The dependence of shrinkage stress reduction on porosity concentration in thin resin layers. J Dent Res 1992;71:1619±22. [15] Kinomoto Y, Torii M, Takeshige F, Ebisu S. Comparison of polymerization contraction stresses between self- and light-curing composites. J Dent 1999;27:383±9. [16] Dauvillier BS, Feilzer AJ, De Gee AJ, Davidson CL. Visco-elastic parameters of dental restorative materials during setting. J Dent Res 2000;79:818±23. [17] Meredith N, Setchell DJ. In vitro measurement of cuspal strain and displacement in composite restored teeth. J Dent 1997;25:331±7. [18] Sakaguchi RL, Ferracane JL. Stress transfer from polymerization

[19] [20] [21] [22]

[23] [24] [25] [26] [27] [28] [29] [30] [31] [32] [33] [34] [35] [36] [37] [38] [39]

[40] [41]

shrinkage of a chemical-cured composite bonded to a pre-cast composite substrate. Dent Mater 1998;14:106±11. Kitasako Y, Burrow MF, Katahira N, Nikaido T, Tagami J. Shear bond strengths of three resin cements to dentine over 3 years in vitro. J Dent 2001;29:139±44. Guzman-Ruiz S, Armstrong SR, Cobb DS, Vargas MA. Association between microtensile bond strength and leakage in the indirect resin composite/dentin adhesively bonded joint. J Dent 2001;29:145±53. Alster D, Feilzer AJ, De Gee AJ, Davidson CL. Tensile strength of thin resin composite layers as a function of layer thickness. J Dent Res 1995;74:1745±8. Roulet J-F, LoÈsche GM. Tooth-coloured inlays and insertsÐlong term clinical results. Proceedings of Conference on: Clinically appropriate alternatives to amalgam: biophysical factors in restorative decision making: 1996, October 30±November 2, Munich. Dash W, Watts DC, editors. Trans Acad Dent Mater 1996:200±15. Nilsson E, Alaeddin S, Karlsson S, Milleding P, Wennerberg A. Factors affecting the shear bond strength of bonded composite inlays. Int J Prosthod 2000;13:52±8. Hubsch PF, Middleton J, Knox J. A ®nite element analysis of the stress at the restoration±tooth interface, comparing inlays and bulk ®llings. Biomaterials 2000;21:1015±9. Koibuchi H, Yasuda N, Nakabayashi N. Bonding to dentin with a self-etching primer: the effect of smear layers. Dent Mater 2001;17: 122±6. Shono Y, Ogawa T, Terashita M, Carvalho RM, Pashley EL, Pashley DH. Regional measurement of resin±dentin bonding as an array. J Dent Res 1999;78:699±705. PerdigaÄo J, Lambrechts P, Van Meerbeek B, Vanherle G, Lopes AL. Field emission SEM comparison of four post®xation drying techniques for human dentin. J Biomed Mater Res 1995;29:1111±20. Tay FR, Moulding KM, Pashley DH. Distribution of nano®llers from a simpli®ed-step adhesive in acid-conditioned dentin. J Adhes Dent 1999;2:103±17. Krejci I, FuÈllemann J, Lutz F. Clinical and long-term scanning electron microscopic studies of composite inlays. Schweizer Monatsschrift fuÈr Zahnmedizin 1994;104:1351±6 in German. Shortall AC, Baylis RL, Wilson HJ. Composite inlay/luting resin bond strength-surface treatment effects. J Dent 1996;24:129±35. Itthagarun A, Tay FR. Self-contamination of deep dentin by dentin ¯uid. Am J Dent 2000;13:195±200. Escribano N, Del-Nero O, de la Macorra JC. Sealing and dentin bond strength of adhesive systems in selected areas of perfused teeth. Dent Mater 2001;17:149±55. Pashley EL, Comer RW, Simpson MD, Horner JA, Pashely DH, Caughman WF. Dentin permeability: sealing the dentin in crown preparations. Oper Dent 1992;17:13±20. Lam CW, Wilson PR. Crown cementation and pulpal health. Int Endod J 1999;32:249±56. FreÂchette VD. Failure analysis of brittle materials, Advances in ceramics, vol. 28. Westerville, OH: The American Ceramic Society, 1990 p. 7±42. White SN, Yu Z. Physical properties of ®xed prosthodontic, resin composite luting agents. Int J Prosthod 1993;6:384±9. Mann KA, Werner FW, Ayers DC. Modeling the tensile behavior of the cement±bone interface using nonlinear fracture mechanics. J Biomech Engng 1997;119:175±8. Mann KA, Allen MJ, Ayers DC. Pre-yield and post-yield shear behavior of the cement±bone interface. J Orthop Res 1998;16:370±8. James SP, Jasty M, Davies J, Piehler H, Harris WH. A fractographic investigation of PMMA bone cement focusing on the relationship between porosity reduction and increased fatigue life. J Biomed Mater Res 1992;26:651±62. Ai H, Nagai M. Effect of the adhesive layer thickness on the fracture toughness of dental adhesive resins. Dent Mater J 2000;19:153±63 in Japanese. Degrange M, Charrier JL, Attal JP, Asmussen E. Bonding of luting

Y.-F. Mak et al. / Dental Materials 18 (2002) 609±621 materials for resin-bonded bridges: clinical relevance of in vitro tests. J Dent 1994;22(Suppl 1):S28±32. [42] Burke FJ, Wilson NH, Watts DC. Fracture resistance of teeth restored with indirect composite resins: the effect of alternative luting procedures. Quintess Int 1994;25:269±75. [43] Mendoza DB, Eakle WS. Retention of posts cemented with various dentinal bonding cements. J Prosthet Dent 1994;72:591±4. [44] Tay FR, Carvalho R, Sano H, Pashley DH. Effect of smear layers on

621

the bonding of a self-etching primer to dentin. J Adhes Dent 2000;2:99±116. [45] Sanares AME, Itthagarun A, King NM, Tay FR, Pashley DH. Adverse surface interactions between one-bottle light-cured adhesives and chemical-cured composites. Dent Mater 2001;17:542±56. [46] Watanabe I, Nakabayashi N, Pashley DH. Bonding to ground dentin by a phenyl-P self-etching primer. J Dent Res 1994;73:1212±20.