Materials Science and Engineering C 33 (2013) 21–27

Contents lists available at SciVerse ScienceDirect

Materials Science and Engineering C

j ourna l homepage: www.e lsev ie r .com/ locate /msec

Depletion of water molecules during ethanol wet-bonding with etch and rinsedental adhesives

Geneviève Grégoire a,⁎, Patrick Sharrock b, Mathieu Delannée a, Marie-Bernadette Delisle c

a Department of Biomaterials, Faculty of Odontology, University Toulouse III, 31062, Toulouse, Franceb Medical and Spatial Imaging Laboratory, University Toulouse III, Ave. Pompidou, 81104, Castres, Francec Faculty of Medicine, University Toulouse III, 31062, Toulouse, France

⁎ Corresponding author at: Faculté de Chirurgie Dent31062 Toulouse Cedex 9, France. Tel.: +33 5 62 17 29 2

E-mail address: gregoire@cict.fr (G. Grégoire).

0928-4931/$ – see front matter © 2012 Elsevier B.V. Alldoi:10.1016/j.msec.2012.07.040

a b s t r a c t

a r t i c l e i n f o

Article history:Received 2 May 2012Received in revised form 6 July 2012Accepted 26 July 2012Available online 8 August 2012

Keywords:Human dentinAdhesionInterfaceHybrid layerEthanol wet bonding

The treatment of demineralized dentin with ethanol has been proposed as a way to improve hydrophobicmonomer penetration into otherwise water saturated collagen fibrils. The ethanol rinse is expected topreserve the fibrils from collapsing while optimizing resin constituent infiltration for better long termadhesion. The physico-chemical investigations of demineralized dentin confirmed objectively these workinghypotheses. Namely, Differential Scanning Calorimetry (DSC) measurements of the melting point of watermolecules pointed to the presence of free and bound water states. Unfreezable water was the main type ofwater remaining following a rinsing step with absolute ethanol. Two different liquid water phases were alsoobserved by Magic Angle Spinning (MAS) solid state Nuclear magnetic Resonance (NMR) spectroscopy.Infrared spectra of ethanol treated specimens illustrated differences with the fully hydrated specimensconcerning the polar carbonyl vibrations. Optical microscopy observations as well as scanning electronmicroscopy showed an improved dentin-adhesive interface with ethanol wet bonding. The results indicatethat water can be confined to strongly bound structural molecules when excess water is removed withethanol prior to adhesive application. This should preserve collagen from hydrolysis upon aging of the hybridlayer.

© 2012 Elsevier B.V. All rights reserved.

1. Introduction

The phenomenon of nano-leakage is thought to cause the longterm degradation of the base of the hybrid layer leading to clinicalloss of restorations. Water hydrolysis of collagen results fromincomplete impregnation of the collagen fibrils with resin or fromwater absorption due to the hydrophilicity of the monomers andpolymerized polymers used in bonding. This arises because themonomers cannot infiltrate thoroughly the open spaces left in collagencompletely demineralized to several micrometers depth and soakedwith water following the rinsing step. Partial demineralization,accompanied with rubbing of the bonding agent was experimentedto test milder self-etch products that could bind directly to theremaining hydroxyapatite in the treated dentin, and form a tighter sealby preserving collagen protected with hydroxyapatite minerals and bywetting more efficiently the interface by use of ethanol as a volatilesolvent promoting reactive monomer infiltration. This was shown tofavor nanolayering on dentin and enamel [1].

aire, 3 chemin des Maraîchers,9; fax: +33 5 61 25 47 19.

rights reserved.

The beneficial role of ethanol was investigated in a study of ahydrophobic adhesive applied to acid-etched and ethanol drieddentin [2]. However, residual water found after ethanol chemicaldehydration compromised polymerization of the adhesive resin. Thiswas attributed to rapid evaporation of ethanol solvent leaving bulkremnant water in the interfibrillar spaces and dentinal tubules. Thusethanol wet bonding may help hydrophobic monomer penetration,but ethanol saturated demineralized dentin was not obtained evenafter 30 s application of absolute ethanol.

In an examination of the technique sensitivity of ethanol wetbonding, the collapse of collagen fibrils was observed by Atomic ForceMicroscopy(AFM) when long ethanol exposure times were used.Progressive water replacement by using ethanol gradients wasproposed based on the rationale that collagen shrinks with free andloosely bound water removal, but does not collapse, leaving essentialwater inside the fibrils to preserve the triple helix structure. Thelarger interfibrillar spaces are then better infiltrated by hydrophobicmonomers [3].

A thicker resin infiltrated layer was observed in a study of anexperimental adhesive based on hydrophobic BisGlycidylmethacrylate(BisGMA) present in 100% ethanol solvent [4]. The diffusion of BisGMAmonomer differed between water and ethanol wet bonding. In theethanol wet bonded specimens, longer resin tags were observed by a

22 G. Grégoire et al. / Materials Science and Engineering C 33 (2013) 21–27

differential staining technique indicating high infiltration into tubules.Ethanol wet bonding increased the penetration of BisGMA in theintertubular regions and better infiltration of the monomers into thecollagen matrix was confirmed by the micro-Raman results.

Ethanol wet bonding has been proposed to coax hydrophobicresins into demineralized dentin. To avoid phase separation of resinsand to favor penetration of monomers up to the demineralized front,the removal of water by ethanol is thought to provide a method thatincreases aging stability, prevents water uptake and subsequentcollagen fibril degradation by water and enzymes. Indeed the long-term stability of the resin–dentin bond is the weak spot in tooth-colored restorations, and much work has been devoted to amelioratethe resin-dentin bond stability. Therapeutic agents have been tested,MMP inhibitors tried, but ethanol washing seems to be the simplestway to achieve better results in resin-dentin bond strength andresistance to nano-leakage and carious recurrence.

The aim of this work was to compare the hybrid layers made byusing the standard etch and rinse procedure and by using the modifiedprocedure including an absolute ethanol rinse prior to application ofthe resin bonding solution. Our objective was to provide physicochem-ical data to combine this with microscopic evidence to reveal anydifference caused by the water vs. ethanol wet bonding procedures andto illustrate and explain the fate of water when the ethanol rinse isapplied to demineralized dentin.

2. Materials and methods

2.1. Thermal methods

Thermogravimetry was performed with a SDT Q600 instrumentfrom TA Instruments (Guyancourt, France) and differential scanningcalorimetry with a DSC 92 instrument from Setaram Instrumentation(Caluire, France). The ground samples (approximately 10 to 15 mg)were placed in platinum sample holders and heated under an air flowof 100 mL min−1 at a rate of 5°min−1. Aluminium pans were used forDSC, with empty pans placed in the reference compartment. De-viations in the baseline were due to slightly inequivalent weightsbetween sample and reference. The sample and reference compart-ments were cooled to −50 °C by use of a liquid nitrogen containingDewar flask while dry nitrogen gas was flowed inside the heatingchamber at a constant rate. The heating rate was 5°min−1 from −50to +110 °C.

2.2. Spectroscopy

The samples studied were derived from dentin slabs where theenamel layer had been removed with a diamond burr. The dentin wasground under liquid nitrogen cooling to prevent collagen burning,and sieved to recover dentin powder with particle sizes lower than250 micrometer in diameter. They were further ground in an agatemortar for infrared spectroscopy [5]. The proton NMR was registeredon a Brukerspectrospin 400 MHz solid-state MAS spectrometer(BrukerBiospin, Wissembourg, France) at room temperature (21 °C).Infrared spectra were recorded on a Genesis II spectrometer fromMattson (Thermo Electron, Courtaboeuf, France) using a diamondanvil ATR from Graseby (Specac, Slough, UK).For the infrared andthermogravimetric investigations, the samples were ground samplesheld in N° 4 glass frit crucibles, etched, rinsed and treated inside the

Table 1Product, manufacturer, component, batch number, chemical composition of the dentin adh

Adhesivetested

Manufacturer Classification B

ExciTE® F Ivoclar Vivadent, Schaan,Liechtenstein

single-component adhesive in conjunctionwith the total-etch technique.

N

crucibles and then sucked solvent free by applying a vacuum in thesupporting Erlenmeyer flask.

2.3. Microscopy

30 human healthy molars extracted because of developmentalpathologies were used for the study. After extraction, they were keptin chloramine T at 4 °C until use. The teeth were cut parallel to theocclusal face using a low-speed saw (Isomet 2000, Buehler, Dardilly,France) equipped with a rotating diamond-impregnated copper disk(11-4244-15 HC, Buehler) under water spray. Deep dentin waschosen for the tests to favour the diffusion of the dentin bondingsystem in demineralized dentin, thus providing the best conditionsfor the resin–dentin interaction. The dentin disks were randomlyassigned to two groups: group I was etched with phosphoric acidfollowed by application of 2 μL of absolute ethanol during 10 s thatwas renewed once after gentle drying with absorbing paper, and thenwas treated with Excite F (see composition in Table 1). Group IIreceived a similar application of Excite F preceded by a phosphoricacid etch but without the ethanol treatments. The adhesive wasapplied to 30 dentin disks according to the manufacturer's step bystep instructions (1 — Apply phosphoric acid gel. Rinse with waterspray and blot excess water with a foam pellet. Do not over dry thedentin! 2 — Saturate dentin with a generous amount of ExciTE Fand agitate the adhesive on the prepared surface for at least10 seconds. 3 — Disperse ExciTE F to a thin layer with a weak streamof air. 4 — Light-cure ExciTE F for 10 s at a light intensity of more than500 mW/cm2). The prepared specimens were kept for 48 h in closedvials saturated with humidity by inserting a water impregnated cottonswab containing 1% chloramine in the inside cover of the samplecontainer.

2.3.1. Specimens preparation for optical microscopyTen disks of each subgroup were used for optical microscopy. They

were fixed in 10% formaldehyde (Rhone-Poulenc Ltd., Manchester,England) for 2 days, demineralized with 5% trichloroacetic acid(Merck, Darmstadt, Germany) for 5 days, and then rinsed in waterfor 2 h. The specimens were dehydrated in ascending grades ofalcohol then cleared in toluene for ½ h before being impregnatedwith paraffin (Merck) for 2 days. They were then embedded inparaffin, and 5 μm sections were cut with a microtome (Reichert–Jung NuBlock,Germany). The sections were successively stained withGoldner's trichrome, a classic bone stain, and with crystal violet andpicric acid [6,7]. The stained sections were observed and photo-graphed using an optical microscope (020-507-101 Wild Leitz GmbH,Heerbrugg, Switzerland).

2.3.2. Specimens preparation for scanning electron microscopyTen dentin disks from each subgroup were used for scanning

electron microscopy. They were fractured into two parts, decalcified inphosphoric acid for 20 s and deproteinised for 60 s in 2% sodiumhypochlorite. The specimens were fixed in 2.5% glutaraldehyde, thenrinsed and dehydrated in alcohol. They were dried by immersion inpure hexamethyldisilazane (HMDS) for 20 min. TheHMDSwas allowedto evaporate for 15 min in air before specimens were sputter-coated ingold palladium. The specimenswere examined and photographed withan SEM (S450, Hitachi, Tokyo, Japan) using an acceleration voltage of15 kV.

esive tested. aHEMA: 2-hydroxyethyl methacrylate.

atch # Composition

75726 Phosphonic acid acrylate, HEMAa, dimethacrylate, highly dispersed siliconedioxide, initiators, stabilizers and potassium fluoride in an alcohol solution.

23G. Grégoire et al. / Materials Science and Engineering C 33 (2013) 21–27

The aim of the optical microscopy and scanning electron micros-copy observations was to detect any differences resulting from theethanol wet bonding technique compared to the classical protocol. Thestructure analysed was the architecture of the outer face of the etchedand treated dentine disks.

3. Results

3.1. Thermal methods

The DSC traces are presented in Fig. 1. Untreated dentin (curve 1-a)did not show any pronounced DSC curves within the temperature zoneinvestigated. However, when dentin powder was demineralized andrinsed with water and air-dried, two strong endothermal events tookplace during heating from low temperatures, one centered at 0 °Crelated to water melting, and another near 65 °C correlated withcollagen denaturation or melting (curve 1-b). When the demineralizeddentin was vacuum dried at 20 °C, the free water signal at 0 °C was nolonger observed, but a weak signal arose near −20 °C (curve 1-c)which can be attributed to strongly bound water molecules, such asthose present in hydrated collagen fibrils responsible for 3D structuralintegrity of the macromolecule. When the water rinse was followed bya washing step with water containing 30v% ethanol (curve 1-d), thewater thawing peak decreased in intensity andwhen 100% ethanol wasused, the 0 °C peak nearly disappeared and was replaced by a broaderpeak located between −30 and −10 °C (curve 1-e). The collagendenaturation shifted to higher temperatures as previously reported.

Fig. 1. DSC diagrams of: a=untreated dentin; b=demineralized dentin rinsed withwater and air dried; c=demineralized dentin dried at 100 mbar 1 h.; d=demineralizeddentin+ethanol 30%; e=demineralized dentin+ethanol 100%; f=d+adhesive;g=e+adhesive; h=b+adhesive. The stars indicate normal freezable water melting,and the os show the collagen denaturation temperatures.

When some free water was still present (30% ethanol use) andadhesive added, the DSC trace showed the broad peak centered at−20 °C and the collagen denaturation near +65 °C (curve 1-f). Butwhen most of the free water had been removed by absolute ethanoland adhesive added (curve 1-g) then no signals were observed in theDSC trace. On the other hand, a strong water thawing peak is observedat 0 °C following the standard manufacturer's method for adhesiveapplication without ethanol rinse (curve 1-h).

Thermo gravimetric analyses revealed the water losses attemperatures between room temperature and 150 °C under air flow.Table 2 lists the results obtained for the samples treated differentways. Demineralized dentin contained the largest amount of waterwith 33.8%, and rinsing with 30% ethanol removed only little water,with 28% water remaining. The 100% ethanol rinse step removedmore water with only 15.3% water left. Application of adhesiveremoved approximately the same amount of water as ethanol did,with 10.5% water left in the sample following demineralization andresin infiltration. Because the remaining water is expressed as % of theinitial mass and the initial mass of dentin was increased by the massof resin, the % of water is smaller than that observed with ethanolalone, but is of similar magnitude. When resin was infiltratedfollowing absolute ethanol rinse, the final water contents were only5.3%, near the percentage observed for untreated dentin. Thus bothethanol and adhesive partially removed water from demineralizeddentin.

3.2. Spectroscopy

Proton NMR results are illustrated in Fig. 2. Two types of protonswere distinguished: one centered at 0.29 ppm, and the other locatedat 4.67 ppm. The upfield proton is recognized as pertaining to “free”OH groups, such as the hydroxyl group in hydroxyapatites, while thedownfield proton is attributed to “bound” water molecules, the de-shielding being due to electron density sharing with neighboringmolecules or surfaces to which the protons are attached. The relativeintensities of the NMR peaks allow calculating the proportions of freeand bound protons (F/B ratio). The ethanol treated sample had an F/Bratio of 0.28, and the same sample with adhesive yielded a ratio of0.22. If ethanol was not used, the F/B ratio found was 0.267. Globally,this indicates that the ethanol treated sample followed by adhesiveapplication had less free protons than bound ones, compared to thedentin not rinsed with ethanol and treated with adhesive only.

The 31P MASD NMR spectrum of the demineralized dentin rinsedwith ethanol showed a strong peak at 3.50 ppm, with no otherfeatures more than spinning side bands. On the other hand, thedentin samples treated with Excite showed a strong peak at 3.50 ppmand a weak but easily detected small peak located at 24.01 ppm. Thestrong peak at 3.5 ppm was attributed to the phosphate groups indental hydroxyapatite whereas the small peak at 24.01 ppm wasrelated to the phosphorous atom present in the phosphonate group ofthe adhesive. It did not change with or without the ethanol rinse [8,9].

The infrared spectra of dentin showed the water and aminestretching frequencies near 3500 cm−1, the amide I vibration at1650 cm−1, and the mineral groups at 1456 and 1413 cm−1 forcarbonates and the major peak at 1013 cm−1 for phosphate . These

Table 2Thermogravimetric data resulting from different dentin treatments.

Sample wt.% water loss

Untreated dentin powder 6.4Demineralized dentin 33.8Demineralized dentin+30% EtOH 28Demineralized dentin+100% EtOH 15.3Demineralized dentin+adhesive 10.5Demineralized dentin+100%EtOH+adhesive 5.3

Fig. 2. ProtonNMRspectraof demineralizeddentinwashedwith ethanol (1), demineralizeddentin with ethanol and adhesive (2) and demineralized dentin and adhesive (3). Thebroad peaks correspond to hydrogen bonded water molecules.

24 G. Grégoire et al. / Materials Science and Engineering C 33 (2013) 21–27

features are illustrated in Fig. 3 (see spectrum 3-a). Demineralizationcaused the disappearance of the phosphate band and of the carbonatebands, but revealed the amide I band and broad OH and NH absorptionbands (spectrum 3-b). Dying under vacuum (100mbar) decreased

Fig. 3. Infrared spectra of normal dentin (a), demineralized dentin rinsed with waterand air-dried (b); demineralized dentin dried under vacuum at 20 °C (c), anddemineralized dentin rinsed with ethanol (d). In the spectra b, c and d, the phosphatepeak is no longer present.

slightly the water bands but did not modify the amide peak (spectrum3-c). Rinsing with ethanol further decreased the intensity of the waterbands and of the amide peak (spectrum3-d). The peak at 1045 cm−1 inspectrum3-Dwas attributed to ethanol residues. Focusing on the amideI region, shown in Fig. 4, one can observe that this amide vibration issmall in normal dentin, and much stronger in demineralized dentinwater washed (Fig. 4, line c). Rinsing with ethanol reduced the peak at1642 cm−1. When the adhesive was applied to the water washeddemineralized dentin (spectrum 4-e), a carboxyl peak appeared at1702 cm−1. However, when the demineralized dentinwas treatedwith

Fig. 4. Infrared spectra in the collagen amide I region of: normal dentin(a); demineralized dentin washed with ethanol (b); demineralized dentin water washed(c); b with Excite (d); c with Excite (e). The arrow points to the carbonyl vibration ofExcite with water.

Fig. 6. (A) and (B). (magnification 320×). Representative light micrographs of thedentin treated with phosphoric acid, followed by application of Excite F. The hybridlayer is perturbed, not constant, with voids. Tag formation is less pronounced, most arenot homogeneous.[6a: mineralized dentin is yellow, the hybrid layer (HL) and tags (T)

25G. Grégoire et al. / Materials Science and Engineering C 33 (2013) 21–27

ethanol prior to adhesive application, the carbonyl only appeared as ashoulder on the major 1636 cm−1 peak (4-d).

3.3. Microscopy

3.3.1. Interface structure observed by optical microscopyRepresentative optical micrographs using Goldner's trichrome and

the staining technique with crystal violet and picric acid are shown inFigs. 5 to 6b. With the first stain, mineralized dentin is blue and theadhesive is beige, resin impregnated demineralized dentin is pink.Demineralized dentin that is not infiltrated with resin remainsunstained. With the second stain, mineralized dentin is yellow, thehybrid layer and the tags are bright pink, and the adhesive is beige.Non infiltrated demineralized collagen stays colorless.

The interface observed in group I showed a regular hybrid layerwith an average thickness of 4 μm, and conical tags (Fig. 5). Theinterface observed in group II showed occasionally an irregular hybridlayer and less numerous tags, or even none at all (Fig. 6a and b). WithGoldner's trichrome demineralized zones free of infiltrated resinwere often seen, sometimes accompanied with water droplets.

3.3.2. Interface structure observed by scanning electron microscopyNumerous and thick tags with V shaped bases were identified in

group I specimens (Fig. 7a). The hybrid layer was continuous andshowed a regular thickness of nearly 3 μm. At the composite/adhesiveinterface of group II, occasional water droplets were seen (Fig. 7b).

4. Discussion

DSC was used previously to examine the denaturation of solvent-treated demineralized fibrils [10].

DSC can be utilized in the lower temperature ranges also, toinvestigate the properties of frozen water molecules and differentiatefree water from loosely or strongly bound water molecules. Ofparticular importance is the observation of supercooled water whichcan be qualified as unfreezable water, a phenomenon associated withstrong binding of the water molecules with the matrix or adsorbentwalls [11,12].

Our results show that untreated dentin has mostly unfreezablewater, with no detectable DSC signal near zero °C or lower tempera-tures. Demineralized dentin (Fig. 1b) shows a strong endothermalsignal corresponding to free water at 0 °C, and a collagen denaturation

Fig. 5. (magnification 320×). Representative light micrograph of the dentin treatedwith phosphoric acid, followed by application of 2 μL of ethanol, and Excite F. Thehybridized complex is thick with numerous tags (T). V-shaped plugs are visible at thetubuli mouths. [Mineralized dentin is blue, resin impregnated demineralized dentin ispink].

are pink, non infiltrated dentin remains unstained. 6b: mineralized dentin is blue, resinimpregnated demineralized dentin is pink. Non infiltrated demineralized collagenstays colorless].

signal near 65 °C, as reported previously [10]. Furthermore, inagreement with other observations, the infiltration with adhesive(Fig. 1h) left a strong frees water signal and a collagen denaturationtemperature above 90 °C. On the other hand, rinsing with 30% ethanolonly slightly decreases the freewater signal and induces little change incollagen denaturation temperature. Vacuum drying the demineralizeddentin removed the free water but left trace amounts of loosely boundwater with a weak signal near −20 °C (Fig. 1c). When the deminer-alized dentin was treated with 30% ethanol followed by adhesiveapplication, sufficient water remained (as determined by the −20 °Csignal intensity in Fig. 1f) to cause denaturation of collagen at 65 °C.Pure ethanol reduced the water contents to traces of free water andbound water molecules (Fig. 1e) This in turn, following adhesiveinfiltration, yielded a DSC trace with no observable water or collagensignals (Fig. 1g), similar to untreated dentin despite the presence of 5%water by weight. This means the water present in untreated dentin andin 100% ethanol rinsed demineralized dentin treated with adhesiveboth represent unfreezable water, tightly bound in the sound dentinand tightly bound in the hybrid layer.

Fig. 7. a. (original magnification 2500×). Representative SEM micrograph showing theadhesive-dentin interface obtained with phosphoric acid, followed by application of2 μL of ethanol, and Excite F. A conical swelling can be observed. Note the uniform andthick hybridized complex and the numerous tags (T) with conical swelling at theirjunction with the hybrid layer (HL). b. (original magnification 2500×). RepresentativeSEM micrograph showing the adhesive-dentin interface with phosphoric acid andExcite F. Note the numerous visible porosities (droplets D) at the interface. The tags (T)appear non-homogeneous and weakened with voids.

26 G. Grégoire et al. / Materials Science and Engineering C 33 (2013) 21–27

Kopp et al. [13] found that for collagen, the most cross-linkedcollagen was able to bind less water but binding of this water wasmore thermostable. Also, the collagen denaturation temperaturesincreased when unfreezable water decreased. Finch and Ledward [14]showed that cross-linked fibers were stabilized by hydrophobicinteractions which rupture exothermally. Haly and Snaith [15]measured specific heat of water in rat tail tendon and resolvedwater into four states. 1) Unfreezable water below 0.3 g water/gprotein. 2) Freezable water from 0.3 to 0.6 g water/g protein forwhich ice enthalpy and temperature of fusion were different frombulk water one. 3) Freezable water with the same ice enthalpy as bulkwater but with a temperature of fusion inferior to 0 °C. 4) Bulk waterabove 1 g water/g protein.

Because ethanol strongly binds to water molecules, all thefreezable water can be removed by dissolution in ethanol. Only theunfreezable water is preserved (see curve g in Fig. 1) following theethanol rinses and adhesive used in our experiments. This explainsthe presence of residual water (observed by infrared and thermo-gravimetry) in the samples treated with the ethanol rinse andaccounts for the thick and homogeneous hybrid layer observed byoptical micrography. Residual water is needed to prevent fibrilcollapse and allow extensive resin encapsulation. The elimination ofinter-fibrillar water accounts for all the results observed and indicatesthat the ethanol rinse can prepare the demineralized layer to betteraccept more hydrophobic monomers to seal the hybrid layer fromfurther water uptake.

Our proton NMR results are analogous to previously publishedfindings which show the presence of free and bound water in variousproportions [16]. Recent work on the mineral-matrix interface incalcium phosphate biomaterials shows how the NMR technique isversatile to unravel the complex intermolecular interactions inbiological materials [17,18]. Proton NMR studies of collagen fibrilshave led to recognize several types or water molecules in collagen.These are classified as strongly held water bridges, cleft water chains,surface water and bulk water. When the tissue contains sufficientwater, then protons are in fast exchange between the variouscompartments, so that the simplified model with two types of water(bound and free) is sufficient to describe the water properties. Clearlythe water impacts collagen fibril structure and stability, and fibrilmolecular mobility has been observed by 13C NMR indicating motionof the polypeptide chains related to water contents. All recent studieson solid state NMR of dentin have shown strong interactions betweenthe collagen organic moieties and the hydroxyapatite mineral phase[19]. Demineralization replaces these interactions with water-collagen bonds, and the aim of adhesive use should be to restorestrong collagen bonds with encapsulating and infiltrated resin.

Infrared spectroscopy has been used to describe variations inmineral densities in mantle dentin and circumpulpaldentin at variousstages of tooth maturation [20]. Di Renzo [21] also used infraredspectroscopy to evaluate chemical modifications of human dentinsurfaces upon demineralization. The extent of demineralization couldbe followed as a function of time and the nature of the acid used. Thedentin-resin interface was investigated by Miyazaki [22] with laserRaman spectroscopy, to determine the resin gradient in the deminer-alized zone, and explore the gap between the demineralization frontand the resin penetration. Laser Ramanmicroscopy was further used tostudy the variation in chemical composition across the hybrid layerobtained with a single-step adhesive [23]. Comparing infrared andRaman spectroscopies in dentin etch chemistry, Lemor [24] concludedthat infrared was limited in resolution to 15 μm due to diffraction,because the wavelengths used to characterize the interfaces were inthe 1000 cm−1 region, corresponding to 10 μm. Thus absorption datacollected at one point contains information from a much larger region,blurring the gradients. In contrast, the Raman spectra obtained at thewavelength used of 647 nm correspond to a diffraction limit near 1 μm,and allow more precise determination of concentration profiles across

27G. Grégoire et al. / Materials Science and Engineering C 33 (2013) 21–27

an interface.These authors found a broad interface associated with thephosphoric acid etched dentin with an overall width of 14.8 μm,consisting in 9 μm completely demineralized zone, and a 6 μm widepartially demineralized zone. The adhesive resin reached 9 μm insidethe demineralized zone and had no direct overlap with the zone overwhich the apatite gradient exists. In fact, there exists an 8 μm widezone where collagen is themajor constituent (apatite and resin are lessthan 50%) and there is a region of 0.5 μmwith nomeasurable apatite orresin concentration. This gap in the resin-sound dentin interface leavesexposed collagen prone to hydrolysis.

Optical microscopy and SEM observations reveal differences inadhesive/dentin interfaces depending on the type of bonding methodused. Our results agree with other morphological observations [4].With water wet bonding, staining techniques as well as SEMinvestigations occasionally reveal droplets previously detected incomparative evaluations of various adhesive systems. These dropletswere seen in the presence of excess water. A more homogeneous andclearly improved interface exists with ethanol wet bonding. Our resultsconfirm the hypothesis put forward by Osorio et al. [3].

To improve the understanding of infiltration of resin monomers inthe demineralized collagen, Vaydianathan [25,26] proposed toanalyze collagen–ligand interactions through computer visualizationof the collagen–monomer complexes. Results confirmed stericcomplementarity as the primary driving force for collagen–ligandinteractions. Electrostatic complementarity was the second source ofmolecular interaction acting on bonding to collagen, with hydrogenbonding reinforcing the adhesive forces due to micromechanicalinterlocking (the steric effects). The authors claimed that computermodeling of molecular conformations confirmed the previouslyreported Raman spectroscopic evidence supporting the formation ofhydrogen bonds between the collagen polymer and the adhesivemonomers. Despite the evidence that monomers have severalfavorable conformations for binding to collagen, the interaction maybe limited to the surface features of collagen, presumably because ofthe tight internal structure of the three stranded superhelix. Indeed,Spencer [27] found no evidence for any covalent bonding betweenCollagen and monomers in a spectroscopic study. Hydrophilicmonomer based resin adhesives may be better adapted to infiltrateand bind to hydrogen-bonded demineralized collagen than hydro-phobic monomers. However, if the water-soluble monomers absorbtoo much water, the adhesive dilution may cause poor polymeriza-tion and leave some collagen fibrils exposed. This is the reason whyrinsing with Ethanol may contribute to better monomer infiltrationby removing excess water and favoring collagen interactions withadhesive components.

J.W. Ager [28] focused on the organic phase of dentin and foundthat the specific amide I vibration was sensitive to the environment ofthe collagen molecule in dentin. The amide I feature changed notablywith age, and increased in dentin that had been demineralized anddehydrated. The amide I vibration is essentially an in-plane carbonylstretch. The observed frequencies (1665 to 1675 cm−1 in solution)depend on the secondary structure of the protein. A strong amide Ifeature at 1660 cm−1 is observable in gel-like collagen. However,compared to collagen in solution, inhomogeneous broadening occursin dentin due to the heterogeneous environment. A significantincrease in the amide I peak was observed for demineralized anddehydrated dentin, but not if the samples were kept hydrated.Demineralization occurs first in the extra-fibrillar compartment thenin the intrafibrillar more protected part of collagen. Partly dehydrateddemineralized collagen shrinks, and then collapses when completelydehydrated. Consistent with the Raman data, loss of chemicallybound water destabilizes the quaternary structure of collagenmolecules. The amide I vibration intensity is insensitive to mineral

contents as long as the structure remains hydrated, but perturbationsdue to exposure to non-physiological solvents increase the amide Iintensity. Intrafibrillar movement and increased interaction betweencollagen fibrils upon dehydration cause an increase in the amide Ipeak height. In dentin dehydrated with alcohol, the amide I peakincreases associated with increased stiffness. The higher meltingpoints observed by DSC analysis of alcohol treated demineralizeddentin can also be attributed to increased stiffness of the fibrils in thepresence of less water.

5. Conclusion

When water molecules are present in the freezable compartmentthey are prone to hydrogen bond other polar mobile substances,including ethanol solvent. However, unfreezable water molecules willremain bonded to the collagen fibrils and resist release by solventwash-out or evaporation under low pressure. They are bonded aswater molecules with high thermal stability. To preserve collagenfrom hydrolysis upon aging of the hybrid layer, ethanol wet bondingcan be used to limit the water presence at the interface to stronglybound structural water that cannot jeopardize the longevity of thehybrid layer.

Acknowledgements

The authors wish to thank Marie-Paule LACOMBLET from Depart-ment of Biomaterials, Faculty of Odontology, University Toulouse III, forher technical help in experimentations.

References

[1] K. Yoshihara, et al., Acta Biomater. 7 (8) (2011) 3187–3195.[2] F.T. Sadek, A. Mazzoni, L. Breschi, F.R. Tay, R.R. Braga, J. Dent. 38 (4) (2010)

276–283.[3] E. Osorio, M. Toledano, F.S. Aguilera, F.R. Tay, R. Osorio, J. Dent. Res. 89 (11)

(2010) 1264–12699.[4] T.P. Shin, X. Yao, R. Huenergardt, M.P. Walker, Y. Wang, Dent. Mater. 25 (8)

(2009) 1050–1057.[5] S. Elfersi, G. Grégoire, P. Sharrock, Dent. Mater. 18 (7) (2002) 529–534.[6] G. Grégoire, P. Guignes, K. Nasr, J. Dent. 37 (2009) 691–699.[7] G. Grégoire, F. Dabsie, M. Delannée, B. Akon, P. Sharrock, J. Dent. 38 (2010)

526–533.[8] M.A. Bayle, G. Grégoire, P. Sharrock, J. Dent. 35 (2007) 302–308.[9] M.A. Bayle, K. Nasr, G. Grégoire, P. Sharrock, Dent. Mater. 24 (2008) 386–391.

[10] S.R. Armstrong, J.L. Jessop, E. Winn, F.R. Tay, D.H. Pashley, J. Endod. 32 (2006)638–641.

[11] P.G. Debenedetti, H.E. Stanley, Phys. Today 56 (2003) 40.[12] J. Wolfe, G. Bryant, K.L. Koster, Cryo. Lett. 23 (2002) 157–166.[13] J. Kopp, M. Bonnet, J.P. Renou, Matrix 9 (1989) 443–450.[14] A. Finch, D.A. Ledward, Biochim. Biophys. Acta 278 (1972) 433–439.[15] A.R. Haly, J.W. Snaith, Biopolymers 10 (1971) 1681–1699.[16] C. Jäger, T. Welzel, W. Meyer-Zaika, M. Epple, Magn. Reson. Chem. 44 (2006)

573–580.[17] J.V. Bradley, et al., Chem. Mater. 22 (2010) 6109–6116.[18] F. Pourpoint, et al., Appl. Magn. Reson. 32 (2007) 435–457.[19] A. Vyalikh, P. Simon, T. Kollmann, R. Kniep, U. Scheler, J. Phys. Chem. C 115 (2011)

1513–1519.[20] K. Verdelis, L. Lukashova, J.T. Wright, R. Mendelsohn, M.G.E. Peterson, S. Doty, A.L.

Boskey, Bone 40 (2007) 1399–1407.[21] M. Di Renzo, T.H. Ellis, E. Sacher, I. Stangel, Biomaterials 22 (2001) 787–792.[22] M. Miyazaki, H. Onose, B.K. Moore, Dent. Mater. 18 (2002) 576–580.[23] M. Sato, M. Miyazaki, J. Dent. 33 (6) (2005) 475–484.[24] R. Lemor, M.B. Kruger, D.M. Wieliczka, P. Spencer, T. May, J. Raman Spectrosc. 31

(2000) 171–176.[25] J. Vaidyanathan, T.K. Vaidyanathan, P. Yadav, C.E. Linaras, Biomaterials 22 (2001)

2911–2920.[26] J. Vaidyanathan, T.K. Vaidyanathan, J.E. Kerrigan, Acta Biomater. 3 (2007)

705–714.[27] P. Spencer, T.J. Byerley, J.D. Eick, J.D. Witt, Dent. Mater. 8 (1992) 10–15.[28] J.W. Ager III, et al., J. Bone Miner. Res. 21 (2006) 1879–1887.