dental materials Dental Materials 16 (2000) 311–323 www.elsevier.com/locate/dental Effect of a sodium hypochlorite gel on dentin bonding J. Perdigão a,*, M. Lopes b, S. Geraldeli c, G.C. Lopes d, F. Garcı́a-Godoy e a Division of Operative Dentistry, University of Minnesota, Minneapolis, MN, USA Minnesota Dental Research Center for Biomaterials and Biomechanics, University of Minnesota, Minneapolis, MN, USA c Department of Dental Materials, University of Santo Amaro and University of Mogi da Cruzes, Sao Paulo, Brazil d Department of Operative Dentistry, Federal University of Santa Catarina, Florianopolis, Brazil e Department of Restorative Dentistry, The University of Texas Health Science Center at San Antonio, San Antonio, TX, USA b Received 15 September 1999; accepted 15 November 1999 Abstract It has been suggested that the hybrid layer (HL) does not play any important role in the mechanism of adhesion to dentin. To substantiate this hypothetical insignificance of the HL, sodium hypochlorite (NaOCl) has been used to remove collagen from etched dentin prior to bonding. Objectives: The present study was conducted to determine the effect of a commercial 10% NaOCl gel on the dentin shear bond strengths and HL ultra-morphology of two simplified dentin adhesives. The null hypothesis tested was that treatment of etched dentin collagen with NaOCl would not compromise dentin bonding. Methods: The labial surface of eighty bovine incisors was polished to expose middle dentin. The specimens were randomly assigned to two total-etch adhesive systems N ˆ 40† : Prime&Bond NT (Dentsply Caulk); and Single Bond (3M Dental Products Division). After rinsing off the etchant, one drop of 10% NaOCl (AD Gel, Kuraray Ltd.) was applied to the etched dentin surface and left for 0 (control), 15, 30, or 60 s. The gel was rinsed off with water and the dentin surface kept visibly moist prior to the application of the adhesive as per manufacturer’s instructions. The respective composite resin was subsequently applied and light-cured. After 24 h in water at 378C, the specimens were thermocycled for 500 cycles in baths kept at 5 and 558C and the shear bond strengths measured. The data were analyzed with two-way ANOVA. For TEM, sixteen dentin disks were taken from middle dentin of extracted human third molars, assigned to the eight treatment sequences, and observed. Results: The increase in the NaOCl application time resulted in a progressive decrease in shear bond strengths for both dentin adhesives. For Single Bond, the application of AD Gel for 60 s resulted in a reduction of bond strengths to 38% of that obtained for the control. For Prime&Bond NT, the mean bond strength obtained when AD Gel was applied for 60 s was 31% of that obtained for the control. The application of AD Gel resulted in distinct morphology for each one of the two adhesives tested. For Single Bond, the general morphology of the collagen network was maintained, regardless of the deproteinization time. The interfibrillar space within the collagen network increased with increasing deproteinization times. For Prime&Bond NT, the general appearance of the HL was maintained for deproteinization times of 15 and 30 s. When the NaOCl gel was applied for 60 s, the morphological appearance of the HL lost its fibrillar arrangement. While remnants of the collagen fibers were observed in one of the dentin disks, the other specimen showed an amorphous structure without any discernible HL morphological features. Significance: The integrity of the collagen fibrils left exposed upon acid-etching plays a major role in the mechanism of adhesion of the specific adhesive systems tested in this study. The intermingling of the adhesive monomers with the filigree of collagen fibers or HL should still be considered the paramount dentin bonding mechanism. q 2000 Academy of Dental Materials. Published by Elsevier Science Ltd. All rights reserved. Keywords: Deproteinization; Sodium hypochlorite gel; Dentin bonding 1. Introduction Enamel is a highly mineralized tissue. The use of acidic * Corresponding author. Division of Operative Dentistry, Department of Restorative Sciences, University of Minnesota School of Dentistry, 8-450 Moos Tower, 515 Delaware St. SE, Minnneapolis, MN 55455, USA. Tel.: 1 1-612-625-8486; fax: 1 1-612-625-7440. E-mail address: perdi001@tc.umn.edu (J. Perdigão). conditioners renders enamel more receptive to fluid adhesive resins due to an increase in surface free energy and formation of enamel resin tags inside the microscopic porosities created by the dissolution of hydroxyapatite crystals [1–5]. The use of phosphoric acid has, therefore, been widely accepted as a conditioning procedure to increase the adhesion of resins to enamel [6]. Dentin, on the other hand, is a dynamic substrate with a complex 0109-5641/00/$ – see front matter q 2000 Academy of Dental Materials. Published by Elsevier Science Ltd. All rights reserved. PII: S0109-564 1(00)00021-X 312 J. Perdigão et al. / Dental Materials 16 (2000) 311–323 Table 1 Adhesive System (Manufacturer) Composition of adhesive Prime&Bond NT Lot.9708000320 Etching gel: silica-thickened 35% (Dentsply Caulk, Milford, DE, H3PO4 Adhesive: PENTA, USA) UDMA 1 T-resin (cross-linking agent) 1 D-resin (small hydrophilic molecule), butylated hydroxitoluene, 4-ethyl dimethyl aminobenzoate, cetilamine hydrofluoride, acetone, silica nanofiller Single Bond Lot 19980826 (3M Etching gel: silica-thickened 35% Dental Products Division, St. H3PO4 Adhesive: BisGMA, HEMA, Paul, MN, USA) dimethacrylates, polyalkenoic acid copolymer, initiator, water, ethanol organic structure and biological activity that preclude the establishment of a reliable and durable bonding [5]. Twenty years ago Fusayama suggested the removal of the smear layer from dentin with phosphoric acid to improve adhesion [7]. In spite of relying on different agents to dissolve the smear layer, most of the current adhesive systems were designed to provide dentin adhesion through the interaction of a hydrophilic monomer dissolved in an organic solvent with a collagen-rich humid tissue [8]. It is believed that the efficacy of the current dentin adhesives depends upon the infiltration of those high-affinity hydrophilic monomers into the filigree of collagen fibers that make up the structure of acid-etched dentin. This entanglement of monomers with collagen fibers and a few residual hydroxyapatite crystals forms a hybrid tissue [9] also known as resin–dentin interdiffusion zone [10]. The depth of dentin demineralization has become an important issue in dentin bonding. Several authors have expressed some concerns about deficient or incomplete penetration of the demineralized microporous collagen network [10,11]. The deficit of resin in the area of the collagen network close to the unaffected dentin or around individual collagen fibers could result in a delicate zone inside the hybrid layer (HL) susceptible to hydrolytic degradation [11–13]. Sodium hypochlorite (NaOCl) has been used on dentin as a deproteinizing agent [14]. NaOCl is a well-known nonspecific proteolytic agent capable of removing organic material [15], as well as magnesium and carbonate ions [16]. Several researchers [17–19] have studied the role of NaOCl in dentin permeability and dentin adhesion. Depending on each testing methodology and/or specific composition of each dentin adhesive, the application of NaOCl upon etching may increase or decrease bond strengths [19–21]. It has been reported that the higher the concentration of NaOCl, the greater the dentin bond strengths until a plateau is reached at a concentration of 10%, for an application time of 60 s [22]. This in vitro study was conducted to determine the effects of a commercial 10% NaOCl gel on the dentin shear bond strengths and resin–dentin ultramorphology of two one- bottle dentin adhesives containing different solvents. The null hypothesis to be tested was that treatment of dentin collagen with NaOCl upon etching would not compromise dentin bonding. 2. Materials and methods 2.1. Bond strengths Eighty bovine teeth were obtained at a local abattoir and were refrigerated in a solution of 0.5% chloramine for up to one week until use. After removing the roots, the crowns were cleaned of debris and were mounted in phenolic rings (Buehler Ltd., Lake Bluff, IL) with Trayresin (Dentsply Trubyte, York, PA) cold-cure acrylic resin. The buccal surface of each tooth was ground with a mechanical grinder to expose middle dentin and was subsequently polished for 30 s with wet 240-, 400-, and 600-grit silicon carbide abrasive paper. Specimens were randomly assigned to two adhesive systems and four treatment sequences 2 × 4 design, Table 1), as follows: Group 1 (control, NT-0)—Dentin was etched with 34% phosphoric acid (Tooth Conditioner Gel, Dentsply Caulk, Milford, DE) for 15 s and then rinsed with water for 10 s. Excess water was removed by blotting with a tissue paper, leaving dentin visibly moist. Prime&Bond NT (Dentsply Caulk) was applied to the dentin surface in copious amounts, leaving the surface wet and undisturbed for 20 s. The adhesive was gently air-dried with oil-free compressed air from an air syringe for 5 s, keeping the air syringe 2 cm from the surface, and light-cured for 20 s. Group 2 (NT-15)—Dentin was etched with 34% phosphoric acid (Tooth Conditioner Gel) for 15 s and then rinsed with water for 10 s. Excess water was removed by gently air-drying for 2–3 s. AD Gel (Lot 00176A, Kuraray Lda, Osaka, Japan) was applied with a disposable brush to cover the dentin surface and left undisturbed for 15 s. The gel was rinsed for 60 s with water and the excess water removed by blotting with a tissue paper, leaving dentin visibly moist. Prime&Bond NT (Dentsply Caulk) was applied to the dentin surface as in Group 1. Group 3 (NT-30)—As in Group 2, but the AD Gel was left on dentin for 30 s. Group 4 (NT-60)—As in Group 2, but the AD gel was left on dentin for 60 s. Group 5 (control, SB-0)—Dentin was etched with 35% phosphoric acid (Scotchbond Etching Gel, 3M Dental Products Division, St. Paul, MN) for 15 s and rinsed for 5 s. Excess water was removed by blotting, so that the tooth surface looked visibly moist. Single Bond (3M Dental Products Division) was applied to the etched dentin surface with a disposable brush. After replenishing the brush with fresh adhesive, a second coat was applied and dried for 2 s. The adhesive was then light-cured for 10 s. Group 6 (SB-15)—Dentin was etched with 35% phosphoric J. Perdigão et al. / Dental Materials 16 (2000) 311–323 313 Table 2 Results Prime&Bond NT Single Bond a b AD GEL (s) SBS (S.D.) a (MPa) % SBS (0 s ˆ 100%) Ultramorphological characteristics of the HL HL thickness (mm) 0 (NT-0) b 15.1 (3.3) CD 100 2.5–4.0 15 (NT-15) 10.9 (4.1) DE 72 30 (NT-30) 8.0 (4.5) EF 53 60 (NT-60) 4.8 (2.3) F 31 0 (SB-0) b 25.0 (4.1) A 100 15 (SB-15) 21.0 (5.7) AB 84 30 (SB-30) 18.5 (6.8) BC 74 60 (SB-60) 9.7 (2.4) E 38 Sprinkled layer of nanoparticles above the HL. Nano-filler penetrated the HL in areas were the tubule entrances were widely exposed to form triangular tag necks. Uniform dispersion of nanofiller through the adhesive layer. Collagen banding in the HL very evident. Fibrillar aspect maintained, but spaces between individual fibers inside the NaOCl-modified HL are wider than for group NT-0. Collagen banding absent. Fibrillar aspect maintained, but collagen fibers are separated by electron lucent spaces indicating some degree of dissolution. The residual collagen structures displayed a shaggy appearance on the top the NaOClmodified HL. For EDTA-decalcified specimens, electrondense formations are identified at the transition between the modified HL and the unaffected dentin. Collagen banding absent. Remains of the collagen network still observed in the lower half of the “modified” HL. Amorphous structure replacing the HL, without the fibrillar aspect observed in the other groups. The upper half, close to the adhesive layer, displayed a more digested appearance as compared with Group NT-30. It also resulted in a wrinkled appearance, with scattered collagen residues, and an electron dense membrane separating this area from the nanofiller in the adhesive. Collagen banding absent. For EDTA-decalcified specimens, the HL is electrondense at the top and less electron-dense at the transition with unaffected dentin. An electron-dense polyalkenoic salt deposition like “floating bubbles” is evident in the adhesive layer. The polyalkenoic salt forms a continuum with the top of the HL. The morphology of the collagen fibers within the HL is different from other adhesives— fibers appear inflated regardless of their cross-section axis. Fibrillar aspect maintained, but spaces between individual fibers inside the HL are wider than for group SB-0. Collagen banding not observed. Interfibrillar spaces are wider than in group SB-15, ranging from 10-20 nm. Collagen banding absent. Fibrillar arrangement lost in most areas close to the unaffected dentin as opposed to the corresponding area in group SB-30. Fibrillar organization maintained in the remaining NaOCl-modified HL but spaces between individual fibers are now in the range of 25 to 60 nm. 2.6–4.3 2.6–4.4 2.4–5.3 2.6–3.7 2.8–3.8 2.9–4.2 4.1–5.5 Means with the same superscript letter are not significantly different at p # 0:05: Adhesive applied as per manufacturer’s instructions. acid (Scotchbond Etching Gel) for 15 s and rinsed for 5 s. Excess water was removed by gently air-drying for 2–3 s. One drop of AD Gel (Kuraray Co.) was applied and left undisturbed on dentin for 15 s. The gel was rinsed off with water for 60 s and the excess water removed by blotting with a tissue paper, leaving dentin visibly moist. Single Bond (3M Dental Products Division) was applied to the dentin surface as in Group 5. Group 7 (SB-30)—As in Group 6, but the AD Gel was left on dentin for 30 s. Group 8 (SB-60)—As in Group 6, but the AD gel was left on dentin for 60 s. A proprietary composite resin (Surefil for Prime&Bond NT; Z100 for Single Bond) was condensed into a #5 gelatin capsule (Torpac Inc., Fairfield, NJ) to fill two-thirds of the capsule and light-cured in a Triad oven (Dentsply Trubyte) for 2 min. A final increment of composite resin was inserted into the gelatin capsule just before its application on the treated surface, and the capsule was seated securely against the flattened tooth surface. Excess material was removed with a sharp explorer under × 2.5 magnifying loupes, and the composite light-cured for a total of 80 s (40 s from each opposite direction) using a Demetron 401 curing light (Demetron/Kerr, Danbury, CT). The intensity of the light 314 J. Perdigão et al. / Dental Materials 16 (2000) 311–323 Fig. 1. TEM resin–dentin interfaces formed by Prime&Bond NT (Group NT-0). (a) Decalcified/stained interface showing the accumulation of nano-filler within the tags. Note a small-hybridized lateral canal (L) and the direct contact between the adhesive filler and the superficial fibers of the HL. Bar ˆ 2 mm. (b) Non-decalcified/non-stained interface showing some diffusion of filler inside the HL (circles). Note the accumulation of a dense layer of particles on the top of the HL, which may correspond to residual silica particles left by the etchant. Bar ˆ 1 mm. (c) Decalcified/stained interface showing a detailed view of the HL, with a shag-carpet appearance on the top of the hybrid. Nano-filler is in intimate contact with these shagged collagen fibers. Bar ˆ 1 mm. (d) Another detailed view of a decalcified/stained interface showing a HL with banded collagen fibers running parallel to the interface (compare to (c) where the fibers run perpendicular to the interface). The transition between the HL and the unaffected dentin shows some electron-dense formations (ovals) which may correspond to hydroxyapatite crystals not completely dissolved by the etchant and later enveloped by the adhesive. Bar ˆ 0.5 mm. (A ˆ Adhesive; C ˆ Composite resin; H ˆ Hybrid layer; T ˆ Resin tag; D ˆ Unaffected dentin). was monitored with a Curing Radiometer (Demetron/Kerr) to be in excess of 500 mW/cm 2. After 24 h in distilled water at 378C, the specimens were thermocycled for 500 cycles between water baths held at 5 and 558C, with a dwell time in each bath of 30 s, and a transfer time of 10 s. After thermocycling, the shear bond strengths were measured with an Instron Universal Testing Machine, model 4411 (Instron Co., Canton, MA), using the MTS Testworks software (MTS Systems Co, Eden Prairie, MN) to record the data. A knife-edge-shearing rod with a crosshead speed of 0.5 cm/min was used to load the specimens until fracture. The distance from the probe to the dentin surface was monitored using a spacer of two celluloid matrices. The data were subjected to two-way ANOVA and Partial Correlation test, using the spss 8.0 for Windows (SPSS Inc., Chicago, IL) software package. 2.2. Transmission electron microscopy Sixteen extracted unerupted third molars refrigerated in a solution of 0.5% chloramine for up to three weeks after extraction were used in this study. The occlusal enamel was removed, and 16 dentin disks with a thickness of 800 100† mm were obtained from middle dentin by slowspeed sectioning with a water-cooled Isomet diamond saw (Buehler Ltd.) parallel to the occlusal surface. A smear layer was created on the top surface by wet sanding with 600-grit silicon carbide abrasive paper for 60 s. The pulpal side of each dentin disk was then isolated with Dentin/Enamel Bonding Resin (Bisco Inc., Schaumburg, IL) to prevent the organic solvents contained in the adhesives from escaping the dentinal tubules. The dentin disks were randomly assigned to the treatment groups described for the bond strength testing n ˆ 2† (Table 1). After application and photo-polymerization of the respective adhesive, a 1-mm thick layer of a flowable composite resin (Æliteflo, Bisco Inc.) was applied to the treated dentin surface and light-cured for 40 s with an Optilux 401 (Demetron/Kerr) visible light-activation unit. A flowable composite resin was used to facilitate ultra-microtomy and to avoid damage to the diamond knife. The disks were then cross-sectioned in two halves, using a water-cooled lowspeed Isomet diamond saw (Buehler Ltd.). Four small sticks with a cross-section of 1:0 × 1:0 mm2 were cut from the center of the bonded half dentin disk using a slow-speed Isomet diamond saw (Buehler Ltd.) under J. Perdigão et al. / Dental Materials 16 (2000) 311–323 315 Fig. 2. TEM resin–dentin interfaces formed by Prime&Bond NT with dentin treated with AD Gel for 15 s (Group NT-15). (a) Non-decalcified/non-stained interface showing a NaOCl-modified HL (h). Note that this modified HL is less dense than the HL depicted in Fig. 1b for NT-0. The collagen fibers are separated by electron lucent spaces indicating some degree of dissolution. Bar ˆ 2 mm. (b) Non-decalcified/non-stained interface showing another view of the NaOCl-modified HL. Note that the nano-filler accumulates at the entrance of the tags to form plugs preventing the filler from penetrating the tubules completely. Bar ˆ 2 mm. (c) Non-decalcified/non-stained interface showing another view of the NaOCl-modified HL. Note that no residual hydroxyapatite crystals are observed in the HL in any of the fields shown in micrographs a, b, or c. Bar ˆ 2 mm. (d) Decalcified/stained interface showing a detailed view of the transition between the HL and unaffected dentin. Electron-dense formations are identified at the transition line (ovals). This field was observed at the same magnification used for Fig. 1d. Note that the interfibrillar spaces are wider for NT-15 than for NT-0. Bar ˆ 0.5 mm. (A ˆ Adhesive; C ˆ Composite resin; h ˆ NaOCl-modified hybrid layer; T ˆ Resin tag; D ˆ Unaffected dentin). water. Two sticks were partially decalcified in 10% buffered EDTA for 72 h to facilitate ultra-microtomy, while the remaining two sticks were left in distilled water at 48C for 72 h to control for possible artifacts during specimen processing. After removing the sticks from the EDTA or from water, they were immersed in 2.5% glutaraldehyde/ 2% paraformaldehyde in 0.1 M sodium cacodylate buffer at pH 7.4 for 12 h at 48C. After fixation, the bonded sticks were rinsed with 10 ml of 0.1 M sodium cacodylate buffer at pH 7.4 for 2 h. The specimens were post-fixed with a solution of 2% osmium tetroxide in 0.1 M sodium cacodylate buffer for 1 h and washed in 0.1 M sodium cacodylate for 1 h. They were then rinsed with deionized water four times, with a dwell time of 5 min, and dehydrated in ascending grades of ethanol (50% for 5 min, 70% for 5 min, 95% for 5 min and 2 × 100% for 10 min each). After the final ethanol step, the specimens were immersed in propylene oxide for two periods of 10 min each. The specimens were then embedded in 50% propylene oxide/50% MedCast epoxy resin (Ted Pella Inc., Redding, CA) in a Pelco Infiltron (Ted Pella Inc.) rotator at 6 rpm. After 6 h of rotation, the specimens were transferred to 100% epoxy resin at room temperature, and placed under vacuum for 12 h to allow for resin infiltration into the specimens. Specimens were oriented in rubber molds so the resin–dentin interface corresponding to the central area of the dentin disk could be first exposed for sectioning. The molds were filled with fresh MedCast epoxy resin (Ted Pella Inc.) and left in an incubator during 12 h at 658C. The resulting resinembedded specimen blocks were trimmed and sectioned in an MT2-B ultra-microtome (Ivan Sorvall Inc., Norwalk, CT) equipped with a material-sciences type III 3.5 mm diamond knife (Micro Star Technologies Inc., Huntsville, TX). The EDTA-decalcified sticks were cut in semi-thin slices and stained with toluidine blue under an optical microscope to allow for the localization of the interface in each section. After positive identification, ultra-thin sections 85 10† nm-thick† were obtained, mounted on 150-mesh nickel grids (Ted Pella Inc.), and stained with 2% uranyl acetate for 15 min and 3% lead citrate for 10 min. For the non-decalcified sticks, they were sectioned in ultra-thin slices and mounted on 150-mesh nickel grids without any staining. After drying at room temperature, the sections 316 J. Perdigão et al. / Dental Materials 16 (2000) 311–323 Fig. 3. TEM resin–dentin interfaces formed by Prime&Bond NT with dentin treated with AD Gel for 30 s (Group NT-30). (a) Non-decalcified/non-stained interface showing a NaOCl-modified HL (h). The collagen fibers are separated by electron lucent spaces indicating some degree of dissolution. Bar ˆ 2 mm. (b) Decalcified/stained interface showing the accumulation of nano-filler at the entrance of one tubule (arrow). Bar ˆ 2 mm. (c) Decalcified/stained interface showing a detailed view of the transition between the NaOCl-modified HL and the adhesive. Note the accumulation of nano-filler at the top of the NaOClmodified HL. Compare to Fig. 1c to visualize the increase of the interfibrillar spaces with the application of NaOCl. The shaggy appearance of the residual collagen structures is well defined on the top the NaOCl-modified HL (asterisks). Bar ˆ 0.5 mm. (d) Decalcified/stained interface showing a detailed view of the transition between the NaOCl-modified HL and unaffected dentin. Electron-dense formations are identified at the transition line (ovals). This field was observed at the same magnification as Fig. 2d, which allows for the observation of the increase in interfibrillar spaces. Bar ˆ 0.5 mm. (A ˆ Adhesive; C ˆ Composite resin; h ˆ NaOCl-modified hybrid layer; T ˆ Resin tag; D ˆ Unaffected dentin). (decalcified and non-decalcified) were analyzed under a Philips CM-12 transmission electron microscope (Philips Electronic Instruments Inc., Mahwah, NJ) at an accelerating voltage of 100 kV. 3. Results The interaction between independent variables (“adhesive system” vs. “NaOCl application time”) was not significant p . 0:254†: When controlling for the independent variable adhesive system, the partial correlation coefficient (20.6858) revealed a statistically significant negative correlation between the application time of AD Gel and mean bond strengths p # 0:0001†: 3.1. Bond strengths 3.2. TEM Mean shear bonds strengths and Duncan’s rankings are shown in Table 2 and Graph. The application of the 10% NaOCl gel resulted in a decrease in bond strengths for both dentin adhesives. For Single Bond, mean bond strengths varied from 25.0 MPa for SB-0, to 9.7 MPa for SB-60 (a reduction to 38% of the SB-0 control). For Prime&Bond NT, mean bond strengths varied from 15.1 MPa for NT-0, to 4.8 MPa for NT-60 (a reduction to 31% of the NT-0 control). There was a statistically significant difference between pairs of means at p # 0:0001: For each of the four deproteinization times, Single Bond consistently resulted in statistically higher bond strengths than Prime& Bond NT. Table 2 displays the principal morphological characteristics of the resin–dentin interfaces for the different groups. The application of AD Gel resulted in distinct morphology for each one of the dentin adhesives tested, and resulted in an increase in the depth of the HL. For NT-0 (Fig. 1), the resin–dentin interface showed a very dense HL with fibers running either parallel or perpendicular to the interface, which was probably a consequence of the sectioning plan. Collagen cross banding was observed when fibers were aligned along their long axis (Fig. 1d). The nano-filler in the adhesive was filtered into the HL, but only in selected areas where the tubule openings were widely exposed. No residual mineral crystals were J. Perdigão et al. / Dental Materials 16 (2000) 311–323 317 Fig. 4. TEM resin–dentin interfaces formed by Prime&Bond NT with dentin treated with AD Gel for 60 s (Group NT-60). (a) Non-decalcified/non-stained interface showing a NaOCl-modified HL (h). Note that this modified HL is collapsed and without any morphological features. Compare with Fig. 3a to visualize the difference in penetration depth between NT-30 and NT-60. Bar ˆ 2 mm. (b) A higher magnification of the micrograph shown in Fig. 1a. Note some residual collagen filaments inside the NaOCl-modified HL. Bar ˆ 0.25 mm. (c) Non-decalcified/non-stained interface showing a NaOCl-modified HL (h) in one specific area of the specimen depicted in (a) and (b) Note the wrinkled appearance, with scattered collagen residues, and an electron dense membrane separating this area from the nano-filler in the adhesive (arrows). In this particular area of the specimen, the NaOCl gel resulted in a more fibrillar morphology than in other areas of this specimen. The top of the NaOCl-modified HL is visibly more digested than the collagen in group NT-30 (compare with Fig. 3c and d in group NT-30). Bar ˆ 0.5mm. (d) Decalcified/stained interface showing a characterless NaOCl-modified HL in another specimen. The role of the EDTA decalcification during specimen preparation in the genesis of this amorphous structure can not be ruled out. Bar ˆ 1 mm. (A ˆ Adhesive; C ˆ Composite resin; h ˆ NaOCl-modified hybrid layer; T ˆ Resin tag; D ˆ Unaffected dentin). observed within the HL in non-decalcified specimens. The general appearance of the HL was maintained for deproteinization times of 15 and 30 s, however with wider interfibrillar spaces (Fig. 2a and b). When the NaOCl gel was applied for 60 s, the morphological appearance of the HL varied drastically. While remnants of the collagen fibers were observed in one of the dentin disks, the other specimen showed an amorphous structure without any discernible morphological features. For SB-0, the HL was electron-dense at the top, and less electron-dense at the transition to the unaffected dentin for EDTA-decalcified specimens (Fig. 5). An electron-dense polyalkenoic salt deposition like “floating bubbles” was observed in the adhesive layer. The polyalkenoic salt formed a continuum with the top of the HL. The fibers within the HL appear swollen regardless of their crosssection axis (Fig. 5d). No residual mineral crystals were observed within the HL in non-decalcified specimens (Fig. 5e). For groups SB-15, SB-30, and SB-60 (Figs. 6–8) the general morphology of the collagen network was maintained, regardless of the AD Gel application time. The interfibrillar spaces increased with the application of NaOCl, to reach a maximum range of 25–60 nm for a deproteinization time of 60 s. Also, the accumulation of the polyalkenoate salt was constant for all the specimens, but more apparent for specimens subjected to EDTA-decalcification that had been stained with heavy metals (uranyl acetate and lead citrate). This polyalkenoate salt deposition was more evident with increased NaOCl application time, and lined the wall of the tubules for groups SB-30 and SB-60 (Figs.7b, c, and 8a, b), which was not frequently observed for the control group. The effect of NaOCl on collagen fibers was readily observed in the area of the transition between the HL and the unaffected dentin. When comparing Fig. 7d with Fig. 8c, shown at similar magnifications, it is observed that the lower area of the NaOCl-modified HL in group SB-60 (Fig. 8c) lost its fibrillar arrangement as opposed to the corresponding area in group SB-30 (Fig. 7d), which still shows a welldefined fibrillar structure. For group SB-60, the lower third of the dentin tubules showed an accumulation of electron dense deposits (Fig. 8a and b). This electron-density was very similar to that of unaffected dentin, but heavier than that of the polyalkenoate salt (Fig. 8a and b). 4. Discussion The interfacial integrity of the resin–dentin bond may 318 J. Perdigão et al. / Dental Materials 16 (2000) 311–323 Fig. 5. TEM resin–dentin interfaces formed by Single Bond (Group SB-0). (a) Decalcified/stained interface showing a HL with a variable electron density gradually fading from the top to the transition to unaffected dentin. Electron-dense polyalkenoic salt deposition like “floating bubbles” (arrows) in the adhesive layer. Bar ˆ 3 mm. (b) Higher magnification of the HL of a decalcified/stained specimen. Bar ˆ 2 mm. (c) Non-decalcified/non-stained interface of the same specimen shown in (b) Note that the electron-density of the HL is uniform, with well-defined collagen fibers. Also, the “floating bubbles” from the polyalkenoic salt are not electron-dense as in (a) and (b), because they were not exposed to heavy metal staining. Bar ˆ 2 mm. (d) Close-up view of the HL in a decalcified/ stained specimen. The polyalkenoic salt intermingles with the top of the HL like a continuous structure. For Single Bond, the morphology of the collagen fibers within the HL is different from other adhesives—fibers appear inflated regardless of their cross-section axis (compare with Fig. 1c and d). Interfibrillar spaces are virtually absent. Bar ˆ 0.5 mm. (e) Close-up view of the HL in a non-decalcified/non-stained specimen showing the transition between the Hl and unaffected dentin. Note that no hydroxyapatite crystals are observed within the HL. The collagen fibers within the HL continue with the fibers in the unaffected dentin (arrows). Bar ˆ 0.25 mm. (A ˆ Adhesive; C ˆ Composite resin; H ˆ Hybrid layer; T ˆ Resin tag; D ˆ Unaffected dentin; Pk ˆ Polyalkenoate salt). depend upon the contribution of both inter- and peritubular dentin [23,24]. The contribution of the peritubular part has been more controversial. The triangular-shaped hybridization at the transition between peri- and intertubular dentin has been described as resin-tag hybridization and described as fundamental for reliable bonding [25,26]. Additional retention and hermetic sealing may also be provided by the formation of microscopic resin tags in the lateral canals that radiate from the main tubules. Such nanometer-sized tags in lateral tubule branches are hybridized, a phenomenon described as lateral tubule hybridization [27]. The contribution of the HL to bond strength was reported by Gwinnett in 1994 [28]. He hypothesized that the collagen layer offers no quantitative contribution to the interfacial bond strength. For some adhesive systems, such as Prime&Bond 2.1, the removal of collagen fibers with a standard laboratory NaOCl solution has been reported to actually increase bond strengths [21]. This increase in bond strengths may be the result of a more severe interaction of this laboratory-prepared NaOCl solution with dentin, even though a recent morphological study using FE-SEM and AFM [29] did not show any residual collagen after the application of a laboratory solution of 5% NaOCl. For another acetone-based adhesive, One-Step (Bisco Inc.), the HL may not play any important role in the effectiveness of the bonding [21,30]. For a multi-bottle adhesive system, All-Bond 2 (Bisco Inc.), one study reported that the presence or absence of the HL does not affect fracture toughness of resin–dentin interfaces [20]. In the case of All-Bond 2, a different mechanical behavior of the adhesive interface would be expected. The Young Modulus of the adhesive resin is 1.8 GPa, while the Young modulus of the All-Bond 2-infiltrated HL was estimated to be 3.6 GPa [31]. Dentin has a Young modulus in the range of 11–18 GPa [18]. The presence of the collagen layer would presumably allow for the establishment of an elastic gradient at the interface. Ideally, the intermingling of resin with acid-etched dentin should be thoroughly accomplished from the top to the bottom of the intertubular area [32]. That entanglement, however, may depend on several factors, such as the aggressiveness of the acid, the post-etching moisture conditions of the dentin surface, the type of adhesive, and its diffusibility into the exposed micro-porous collagen network [8,26]. It has been demonstrated that the utilization of different phosphoric acid gels on dentin results in different thickness of the HL. The varying thickness does not, however, affect the J. Perdigão et al. / Dental Materials 16 (2000) 311–323 319 Fig. 6. TEM resin–dentin interfaces formed by Single Bond with dentin treated with AD Gel for 15 s (Group SB-15). (a) Non-decalcified/non-stained interface showing a NaOCl-modified HL (h). If compared with the specimens in Group SB-0, the morphological differences are not observed at this magnification. The accumulation of the polyalkenoic salt is visible at the top of the NaOCl-modified HL. Bar ˆ 2 mm. (b) Non-decalcified/non-stained interface of the second specimen. At this magnification, some differences are already discernible between the morphology of the NaOCl-modified HL (h) and that of the HL (H) in group SB-0 (compare with Fig. 5c). In SB-15, the collagen fibers are thinner and arranged more randomly. Bar ˆ 2 mm. (c) Decalcified/stained interface showing a NaOCl-modified HL with wider interfibrillar spaces than those observed for group SB-0. The surface of the HL is coated with the polyalkenoate salt precipitate. Bar ˆ 1 mm. (d) Higher magnification of the NaOCl-modified HL in a decalcified/stained specimen. If compared with Fig. 5d, one can easily discern the increase in the interfibrillar spaces. Bar ˆ 1 mm. (A ˆ Adhesive; C ˆ Composite resin; h ˆ NaOCl-modified hybrid layer; T ˆ Resin tag; D ˆ Unaffected dentin; Pk ˆ Polyalkenoate salt). bond strengths associated with one-bottle adhesive systems when dentin is etched for 15 s [33]. Likewise, dentin bond strengths associated with Single Bond do not depend upon the etching time [34]. In the present study, treatment of the collagen with NaOCl resulted in a decrease in bond strengths in spite of a deeper penetration of the adhesive. Consequently, while the depth of dentin demineralization may not be an important factor for dentin adhesion, the quality and/or integrity of the collagen available for resin infiltration may be of fundamental importance. As a result of presumed changes in the collagen fibrils upon application of NaOCl, dentin bond strengths significantly decreased for both adhesives. One question brought up by the recent introduction of nano-filled dentin adhesives on the market is the penetration of filler into the HL [35]. Although a controversial issue, the nano-filler has been described as being able to penetrate the HL in some areas. This is particularly noticeable when observing non-decalcified unstained interfaces like in Fig. 1b. The clinical significance of this finding remains to be discovered. Under the TEM, the application of AD Gel for 60 s did not remove the collagen layer, but altered the morphological characteristics and spatial configuration of the collagen fibrils in the HL. Overall, the deposition of the polyalkenoate in the tubules increased with the NaOCl Gel application time (Figs.7b, c, and 8a, b). This phenomenon may have been a result of the depletion of organic components by the NaOCl gel below the area etched by the acid, and consequent exposure of calcium on the surface of the dentin unaffected by the acid. This calcium may have attracted the polyalkenoic acid and formed a precipitate. For group SB-60, the lower third of the dentin tubules showed an accumulation of electron dense deposits (Fig.8a and b). This electron-density was very similar to that of unaffected dentin, but heavier than that of the polyalkenoate salt (Fig. 8a and b). It is speculated that these deposits are remnants of hydroxyapatite that resulted 320 J. Perdigão et al. / Dental Materials 16 (2000) 311–323 Fig. 7. TEM resin–dentin interfaces formed by Single Bond with dentin treated with AD Gel for 30 s (Group SB-30). (a) Non-decalcified/non-stained interface showing a NaOCl-modified HL (h) that lost part of its reticular morphology when compared to Fig. 6a. The long strand of polyalkenoate salt on the top of the NaOCl-modified HL forms a uniform coating. Bar ˆ 2 mm. (b) Decalcified/stained interface that corresponds to the non-decalcified specimen in micrograph “a”. The collagen fibers lost their characteristic disposition observed in the control group (Fig. 5b). Note how the polyalkenoate salt forms a uniform layer on the NaOCl-modified HL and within the tubules. Overall, the deposition of this polyalkenoate in the tubules increased with the NaOCl Gel application time. Bar ˆ 3 mm. (c) Higher magnification of a decalcified/stained interface. The polyalkenoate salt lines the wall of the tubule. This inner lining was not frequently observed for the control group (SB-0). Bar ˆ 1 mm. (d) Close-up view of the transition between the NaOCl-modified HL and unaffected dentin for a decalcified/stained interface. Note that the fibrillar arrangement was not disrupted at the base of the NaOCl-modified HL. The transition to the unaffected dentin displays an electron-dense formation that may be a result of denudation of hydroxyapatite crystals from collagen fibers and enveloping of these crystals with resin (ovals). Bar ˆ 0.5 mm. (A ˆ Adhesive; C ˆ Composite resin; h ˆ NaOCl-modified hybrid layer; T ˆ Resin tag; D ˆ Unaffected dentin; Pk ˆ Polyalkenoate salt). from the proteolytic action of NaOCl on unaffected dentin. In the NT-60 group, most of the residual collagen was dissolved leaving an amorphous layer, probably gelatinized collagen (Fig. 4), but the morphology was not consistent between specimens. This variance in morphological results is in agreement with the difference in bond strengths obtained with Prime&Bond NT in the present project as compared to a previous report [35]. Similarly, dentin bond strengths associated with Prime&Bond 2.1 applied on moist dentin vary in a wide range from 7 to 23 MPa [36–39]. This discrepancy can be attributed not only to the variability in individual’s interpretation of the concept of “wet-bonding technique”, but also to differences in testing assemblies. In spite of being consistently applied to moist dentin, the level of moisture may be different for each research experiment and therefore may result in a wide variation in adhesion strengths. In other words, this wet-bonding technique is difficult to standardize clinically and thus is sensitive to errors of inaccurate clinical handling [40,41]. This incon- sistency makes acetone-based adhesives very technique sensitive. The decrease in bond strengths upon the application of AD Gel on etched dentin might be a consequence of several mechanisms acting individually or simultaneously: 1. Partial dissolution of intertubular collagen—The integrity of collagen fibers upon etching has been deemed essential for obtaining reliable and durable dentin bonds because collagen denaturation has been thought to play a negative role in dentin adhesion [42]. In result, some authors have advocated the use of 10% citric acid with 3% ferric chloride as a dentin conditioner [9], as ferric chloride could prevent etched collagen from undergoing denaturation. Nevertheless, such preventive effect has never been substantiated with sound research. 2. De-stabilization of the collagen molecule—Dentin collagen, after demineralization, is in a “de-stabilized”, but not denatured state, susceptible to proteolytic degradation [43,44]. The durability of collagen J. Perdigão et al. / Dental Materials 16 (2000) 311–323 321 Fig. 8. TEM resin–dentin interfaces formed by Single Bond with dentin treated with AD Gel for 60 s (Group SB-60). (a) Non-decalcified/non-stained interface showing a NaOCl-modified HL (h). Note that the polyalkenoate salt deposition is not as apparent as in decalcified/stained specimens, but still arranged as “floating bubbles” within the adhesive layer. Bar ˆ 3 mm. (b) Close up of the area shown in the lower left corner of Fig. 8a. Note the hydroxyapatite deposition in the tubules (long arrows) and the polyalkenoate accumulation (star). The arrowheads point at residual peritubular dentin. Bar ˆ 1 mm. (c) Close-up view of the transition between the NaOCl-modified HL and unaffected dentin for a decalcified/stained interface. Compare with Fig. 7d. The fibrillar arrangement in group SB-60 is partially lost at the base of the NaOCl-modified HL. Also, the electron-dense formation at the transition to the unaffected dentin is wider and more prominent for group SB-60 than for group SB-30 (arrows), which may be a result of a more severe effect of NaOCl on the hydroxyapatite crystals of unaffected dentin. Bar ˆ 0.5 mm. (d) Close-up view of the transition between the NaOCl-modified HL and unaffected dentin for a non-decalcified/non-stained interface. Besides the presence of loose hydroxyapatite crystals at that transition (circles), note that there is an electron-lucent halo between the NaOCl modified HL and unaffected dentin (asterisks). This halo may correspond to the electron-dense formation shown in Fig. 8c. Bar ˆ 0.25 mm. (A ˆ Adhesive; C ˆ Composite resin; h ˆ NaOCl-modified hybrid layer; T ˆ Resin tag; D ˆ Unaffected dentin; Pk ˆ Polyalkenoate salt). fibers in proteolytic solutions is dependent on both the degree of cross-linking and the extent of denaturation [45]. In the present study, the lack of collagen banding in NaOCl-treated collagen may have been a result of the deproteinization effects of NaOCl on collagen fibrils de-stabilized by the etchant. Some peptide bonds in collagen fibrils become very susceptible to cleavage after denaturation takes place [45]. 3. Volumetric contraction of NaOCl-treated dentin—It has been recently reported that 10% NaOCl results in a contraction of demineralized dentin up to 12% and decreases the N/Ca ration of a dentin lesion by at least 20% [15]. This change in mineral profile and volumetric contraction of demineralized dentin may be partially responsible for the decrease in bond strengths observed in this project. 4. Changes in the crystallinity of dentin apatite upon NaOCl treatment—X-ray ion diffraction studies have suggested that recrystallization takes place after NaOCl application. The apatite crystals undergo substitution of certain ions in the crystal lattice [15]. This re-crystallization might be responsible for changes in the surface tension of the substrate, therefore it may compromise the bonding ability of the dentin surface. The null hypothesis was rejected—the application of a 10% NaOCl gel for 30 or 60 s resulted in a significant decrease in bond strengths for both adhesives. Further studies should concentrate on the potential use of NaOCl as a remineralizing agent for the areas of the dentin cushion susceptible to microleakage [46,47]. By removing organic components from dentin, NaOCl also removes mineralization inhibitors, presumably phosphoproteins, and increases the porosity of the residual dentin [15,16]. NaOCl has been shown to increase the mineral content of a carious lesion in vitro, regardless of the presence of F 2 ion [15,47]. Additionally, it would be paramount to study the “etching” effects of NaOCl on the unaffected dentin underneath the exposed collagen network. It was clear from the present project that the application of NaOCl induces morphological changes beyond the area corresponding to the deepest penetration of the phosphoric acid gel. At this point, it is still unclear what are the long-term effects of the application of NaOCl on etched dentin in 322 J. Perdigão et al. / Dental Materials 16 (2000) 311–323 vivo. Therefore, its use on a clinical setting is strongly discouraged. References [1] Gwinnett AJ, Matsui A. A study of enamel adhesives. The physical relationship between enamel and adhesive. Arch Oral Biol 1967;12:1615–20. [2] Buonocore MG, Matsui A, Gwinnett AJ. Penetration of resin into enamel surfaces with reference to bonding. Arch Oral Biol 1968;13:61–70. [3] Erickson RL. Surface interactions of dentin adhesive materials. Oper Dent 1992((Suppl. 5)):81–94. [4] Attal JP, Asmussen E, Degrange M. Effects of surface treatment on the free energy of dentin. Dent Mater 1994;10:259–64. [5] Swift EJ, Perdigão J, Heymann HO. Bonding to enamel and dentin: a brief history and state of the art. Quintessence Int 1995;26:95–110. [6] Buonocore MG. A simple method of increasing the adhesion of acrylic filling materials to enamel surfaces. J Dent Res 1995;34:849–53. [7] Fusayama T, Nakamura M, Kurosaki N, Iwaku M. Non-pressure adhesion of a new adhesive restorative resin. J Dent Res 1979;58:1364–70. [8] Van Meerbeek B, Perdigão J, Lambrechts P, Vanherle G. The clinical performance of dentin adhesives. J Dent 1998;26:1–20. [9] Nakabayashi N, Kojima K, Masuhara E. The promotion of adhesion by the infiltration of monomers into tooth substrates. J Biomed Mater Res 1982;16:265–73. [10] Van Meerbeek B, Dhem A, Goret-Nicaise M, Braem M, Lambrechts P, Vanherle G. Comparative SEM and TEM examination of the ultrastructure of the resin–dentin interdiffusion zone. J Dent Res 1993;72:495–501. [11] Pashley DH, Horner JA, Brewer PD. Interactions of conditioners on the dentin surface. Oper Dent 1992((Suppl. 5)):137–50. [12] Sano H, Shono T, Sonoda H, Takatsu T, Ciucchi B, Carvalho R, Pashley DH. Relationship between surface area for adhesion and tensile bond strength—evaluation of a micro-tensile bond test. Dent Mater 1994;10:236–40. [13] Burrow MF, Satoh M, Tagami J. Dentin bond durability after three years using a dentin bonding agent with and without priming. Dent Mater 1996;12:302–7. [14] Shellis RP. Structural organization of calcospherites in normal and rachitic human dentin. Arch Oral Biol 1983;28:85–95. [15] Inaba D, Ruben J, Takagi O, Arends J. Effect of sodium hypochlorite treatment on remineralization of human dentine in vitro. Caries Res 1996;30:218–24. [16] Sakae T, Mishima H, Kozawa Y. Changes in bovine dentin mineral with sodium hypochlorite treatment. J Dent Res 1988;67:1229–34. [17] Barbosa K, Safavi KE, Spangberg SW. Influence of sodium hypochlorite on the permeability and structure of cervical human dentin. Int Endod J 1994;27:309–12. [18] Wakabayashi Y, Kondou Y, Suzuki K, Yatani H, Yamashita A. Effect of dissolution of collagen on adhesion to dentin. Int J Prosthodont 1994;7:302–6. [19] Vargas MA, Cobb DS, Armstrong SR. Resin–dentin shear bond strength and interfacial ultrastructure with and without a hybrid layer. Oper Dent 1997;22:159–66. [20] Armstrong SR, Boyer DB, Keller JC, Park JB. Effect of hybrid layer on fracture toughness of adhesively bonded dentin–resin composite joint. Dent Mater 1998;14:91–8. [21] Inai N, Kanemura N, Tagami J, Watanabe LG, Marshall SJ, Marshall GW. Adhesion between collagen depleted dentin and dentin adhesive. Am J Dent 1998;11:123–7. [22] Tanaka J, Nakai H. Application of root canal cleaning agents having dissolving abilities of collagen to the surface treatment for enhanced bonding of resin to dentin. Dent Mater J 1993;12:196–208. [23] Chapell RP, Cobb CM, Spencer P, Eick JD. Dentin tubule anastomosis: a potential factor in dentinal adhesive bonding?. J Prosthet Dent 1994;72:183–8. [24] Gwinnett AJ, Tay FR, Pang KM, Wei SHY. Quantitative contribution of the collagen network in dentin hybridization. Am J Dent 1996;9:140–4. [25] Van Meerbeek B, Conn Jr LJ, Duke ES, Eick JD, Robinson SJ, Guerrero D. Correlative transmission electron microscopy examination of nondemineralized and demineralized resin–dentin interfaces formed by two dentin adhesive systems. J Dent Res 1996;75:879–88. [26] Eick JD, Gwinnett AJ, Pashley DH, Robinson SJ. Current concepts on adhesion to dentin. Crit Rev Oral Biol Med 1997;8:306–35. [27] Van Meerbeek B, Perdigão J, Lambrechts P, Vanherle G. Enamel and dentin adhesion. In: Schwartz RS, Summitt JB, Robbins JW, editors. Fundamentals of operative dentistry. A contemporary approach, Chicago: Quintessence, 1996. p. 141–86. [28] Gwinnett AJ. Altered tissue contribution to interfacial bond strength with acid conditioned dentin. Am J Dent 1994;7:243–6. [29] Perdigão J, Thompson JY, Toledano M, Osorio R. An ultra-morphological characterization of collagen-depleted etched dentin. Am J Dent 2000 (in press). [30] Kanca J, Sandrik J. Bonding to dentin. Clues to the mechanism of adhesion. Am J Dent 1998;11:154–9. [31] Sano H, Takatsu T, Ciucchi B, Russell CM, Pashley DH. Tensile properties of resin-infiltrated demineralized human dentin. J Dent Res 1995;74:1093–102. [32] Tay FR, Gwinnett AJ, Wei SHY. Micromorphological spectrum from overdrying to overwetting acid-conditioned dentin in water-free, acetone-based, single-bottle primer/adhesives. Dent Mater 1996;12:236–44. [33] Perdigão J, May KN, Wilder AD, Lopes M. The effect of depth of dentin demineralization on bond strengths and morphology of the hybrid layer. Oper Dent 2000 (in press). [34] Paul SJ, Welter DA, Ghazi M, Pashley D. Nanoleakage at the dentin adhesiove interface vs. m-tensile bond strength. Oper Dent 1999;24:181–8. [35] Perdigão J, Baratieri LN, Lopes M. Laboratory evaluation and clinical application of a new one-bottle adhesive. J Esthet Dent 1999;11:23– 35. [36] Barkmeier WW, Hammesfahr PD, Latta MA. Bond strength of composite to enamel and dentin using Prime & Bond 2.1. Oper Dent 1999;24:51–6. [37] Swift EJ, Bayne SC. Shear bond strength of a new one-bottle dentin adhesive. Am J Dent 1997;10:184–8. [38] Perdigão J, Ramos JC, Lambrechts P. In vitro interfacial relationship between human dentin and one-bottle dental adhesives. Dent Mater 1997;13:218–27. [39] Perdigão J, Van Meerbeek B, Lopes M, Ambrose WW. The effect of a re-wetting agent on dentin bonding. Dent Mater 1999;15:282–95. [40] Tay FR, Gwinnett AJ, Wei SHY. Micromorphological spectrum from overdrying to overwetting acid-conditioned dentin in water-free, acetone-based, single-bottle primer/adhesives. Dent Mater 1996;12:236– 44. [41] Van Meerbeek B, Yoshida Y, Lambrechts P, Vanherle G, Duke ES, Eick JD, Robinson SJ. A TEM study of two water-based adhesive systems bonded to dry and wet dentin. J Dent Res 1998;77:50–9. [42] Nakabayashi N. Bonding of restorative materials to dentine: the present status in Japan. Int Dent J 1985;35:145–54. [43] Scott PG, Leaver AG. The degradation of human collagen by trypsin. Connect Tissue Res 1974;2:299–307. [44] Okamoto Y, Heeley JD, Dogon IL, Shintani H. Effects of phosphoric J. Perdigão et al. / Dental Materials 16 (2000) 311–323 acid and tannic acid on dentine collagen. J Oral Rehabil 1991;18: 507–12. [45] Weadock KS, Miller EJ, Keuffel EL, Dunn MG. Effect of physical crosslinking methods on collagen-fiber durability in proteolytic solutions. J Biomed Mater Res 1996;32:221–6. 323 [46] Sano H, Shono T, Takatsu T, Hosoda H. Microporous dentin zone beneath resin-impregnated layer. Oper Dent 1994;19:59–64. [47] Robinson C, Hallsworth AS, Shore RC, Kirkham J. Effect of surface zone deproteinisation on the access of mineral ions into subsurface carious lesions of human enamel. Caries Res 1990;24:226–30.