Microtensile bond strength tests:scanning electron microscopyevaluation of sample integritybefore testing

Marco Ferrari1, Cecilia Goracci1,Fernanda Sadek2, Paulo EduardoCapel Cardoso2

1Department of Dental Materials, University ofSienna, Sienna, Italy, 2Department of DentalMaterials, University of S¼o Paolo, S¼o Paolo,Brazil

Shear and tensile bond strength tests have long repre-sented the most common laboratory trials for evaluatingthe adhesion of bonding systems to enamel and dentin.These tests are relatively easy to carry out, are widelyapplied in dental research, and they provide the bulk ofthe currently published data on bonding systems. How-ever, it has been shown that both tensile and shear bondstrength tests can be greatly affected by the variability inspecimen geometry and experimental loading conditions(1, 2).Microtensile tests have been developed (3) to overcome

some of these limitations, and are now regarded asthe most predictable bond strength tests that can beperformed (4).The microtensile technique offers several advantages

over the other procedures (4–7). One of the mainobjectives of the method, as it was first introduced, wasto avoid the occurrence of cohesive fractures of dentin onloading. Cohesive failures in dentin, while the resin–dentin bond remains intact, have been frequentlyreported since the introduction of newer adhesive sys-tems that create bond strengths on dentin of 20–25 MPa(4). The occurrence of failure of the substrate itselfprevents the measurement of interfacial bond strengthand, hence, the evaluation of improvements in bonding

procedures or formulations. Using the microtensile testresults in a more uniform distribution of loading stressesacross a smaller bonded interface, thereby reducing thefrequency of cohesive fractures in dentin compared withconventional shear bond strength tests (4).The structural variability of the substrate in small

bonding sites is expected to be limited, thus allowing fora more accurate analysis of the bonding mechanism.Indeed, the microtensile technique has found a specificapplication in highlighting the differences in bondingcharacteristics between small regions of dental tissues (4),such as normal vs. adjacent carious dentin (8), enamel vs.dentin, coronal dentin vs. root dentin (4, 9–11), andocclusal vs. middle vs. cervical enamel (12).In addition, since a number of microtensile specimens

can be obtained from a single tooth, collecting suitablenumbers of teeth that meet the statistical criteriabecomes easier (4).However, a �trade-off � for the simplified sample col-

lection exists in the fact that microtensile testing is avery �technique-sensitive� procedure. Proper specimenpreparation requires special testing equipment and askillful investigator. The method involves cutting abonded tooth into a number of slabs, which are thenfurther sectioned into sticks of 0.5–1.5 mm thickness.

Ferrari M, Goracci C, Sadek F, Cardoso PEC. Microtensile bond strength tests:scanning electron microscopy evaluation of sample integrity before testing. Eur J OralSci 2002; 110: 385–391. � Eur J Oral Sci, 2002

The failure of a certain number of microtensile specimens during their preparation andbefore loading is a common and undesirable occurrence. This study was aimed atobserving, under a scanning electron microscope, enamel and dentin microtensilespecimens, in order to find structural faults that might be responsible for their pre-mature failure. In a sample of 80 sticks, none of the specimens was found to be free ofdefects. These may consist of microcracks in enamel, most often at the periphery of thestick, or in dentin at the level of hybrid layer. Gaps were often seen at the interfacesbetween the substrates. Voids were sometimes visible within the resin compositethickness. Enamel specimens tended to exhibit more defects than dentin specimens. Itis fair to suspect that, because of the brittleness of the tissue, enamel microtensilespecimens are intrinsically more prone to failure, thus yielding bond strengths whichare not significantly higher than those measured on dentin specimens. This leads one toquestion the reliability of the microtensile method for testing adhesion on enamel. Itseems sensible to develop a method for a quantitative assessment of specimensintegrity before loading as a possible predictor for their performance under load.

Marco Ferrari, Research Center for DentalHealth, Piazza Attias 19, I)57120 Livorno, Italy.

Telefax: +39–0586–898305E-mail: md3972@mclink.it

Key words: dental materials; scanning electronmicroscopy; bond strength; dentin bondingagents

Accepted for publication July 2002

Eur J Oral Sci 2002; 110: 385–391Printed in UK. All rights reserved

Copyright � Eur J Oral Sci 2002

European Journal ofOral SciencesISSN 0909-8836

Each stick is made up of the two substrates (i.e. resincomposite and enamel or dentin), which are bonded atthe interface to be tested. The stick can be left in a beamshape in the �non-trimming technique� (5, 13–16) or slabscan be trimmed with burs at the bonding site (5, 11, 16,17) to create an hourglass profile that reduces thebonding surface even more, further concentrating theloading stress.All the cutting procedures, particularly the bur-trim-

ming of the hourglass shape likely transmit vibrations tothe specimens. A common occurrence when preparingmicrotensile specimens, especially if the bond strengthsare relatively low (5–7 MPa) (4), is a premature failure ofthe specimen that makes it useless (5, 14, 16).The number of prematurely failed, discarded speci-

mens in each test is probably related to the �aggressive-ness� of the preparation procedure. In this regard, thenon-trimming technique should prove as the least trau-matizing and most efficient (4, 5).Scanning electron microscopy (SEM) observations can

be of great help in revealing interfacial voids, gaps orcracks in microtensile specimens that can be responsiblefor their premature failure. However, SEM investiga-tions of microtensile specimens in order to assess the typeof failure have only been conducted after tensile testing(18, 19).No previous SEM study has evaluated the integrity of

microtensile samples after preparation, but before load-ing, in order to detect those structural defects in thesubstrates or the adhesive interface that can significantlyaffect the bond strength test results.The purpose of this study was to determine if SEM

observations of sticks prepared from resin-bondedenamel and dentin could identify interfacial defects priorto microtensile testing.

Material and methodsTwenty-eight sound human molars, recently extracted forperiodontal reasons, were selected for the study. Any resid-ual soft tissue was removed from the roots with a scaler. Theteeth were then rinsed with water, and stored in a salinesolution at 4�C for no longer than 3 months.On all of the specimens, the roots were cut off at the

middle third with a diamond disc. Each tooth was thenrandomly assigned to one of two groups: group A includedsamples for resin–enamel bond-strength testing. On thesespecimens, some of the most superficial enamel was cut offfrom the buccal or lingual aspect of the tooth with a cooleddiamond disk on a Labcut 1010 machine (Extec Corp.,Enfield, CT, USA) (Fig. 1a). Care was taken not to exposeany portion of the underlying dentin but to create a flatsurface of enamel, which was then polished with wet sand-paper (Fig. 1a).Group B samples were used to evaluate resin–dentin bond

strengths. On these teeth, all of the occlusal enamel andsome superficial dentin were removed with a cooled dia-mond disk to obtain a flat surface in mid-coronal dentin(Fig. 1b).All polishing of enamel or dentin bonding surfaces was

done with wet sandpaper to create a standard smear layer.Abrasive (SiC) papers of 220, 320, and 400 grit were used in

sequence, each one for 10 s, followed by a final polishingwith a 600 grit sandpaper for 60 s. Finally, the bondingsurface was rinsed with water and lightly dried with an airstream.Within each group, two subgroups were then randomly

formed in which two different bonding systems were tested.They were a self-etching primer system (Clearfil SE; Kura-ray, Morita, Japan – subgroup 1), and a �total-etch system�(Excite, Vivadent, Schaan, Liechtenstein – subgroup 2).These materials were used according to the manufacturer’sinstructions (Table 1).After applying the adhesive system, a proprietary com-

posite resin block of approximately 5 · 5 · 5 mm was builton the bonding surface, following the incremental tech-nique. Each layer of composite was individually curedfor 40 s, with an Optilux 401 light [Demetron, (Kerr Co,Danbury, CT, USA) 600 mW cm)2 intensity] (Fig. 1c).

Fig. 1. (a) Enamel specimens preparation involved the removalof a portion of superficial tissue without exposing the under-lying dentin. (b) Tooth prepared for dentin test: the occlusalthird was removed with a diamond disc, creating a flat surface.(c) Resin build-up over the enamel and the dentin surface. (d)Cutting of the tooth along the X and Y axis and the resultingstick.

386 Ferrari et al.

The bonded specimens were placed in a saline solution at27�C for 24 h. Using a diamond blade, each bonded toothwas sectioned vertically into a number of slabs 0.8 mmthick. By rotating the sample 90� and again sectioning itlengthwise, multiple beam-shaped sticks, each with a cross-sectional surface area of approximately 0.8 · 0.8 mm(¼ 0.64 mm2), were obtained. The lower half of the stickswas made up of the dental substrate, while the upper halfwas composed of the resin build-up (Fig. 1d).Two to three sticks from each subgroup were randomly

selected for microscopic analysis. A total of 80 sticks, 40with enamel and 40 with dentin as bonding substrates, wereprocessed for SEM observation.The preparation involved a gentle surface decalcification

of the sticks with 36% phosphoric acid for 10 s, and a briefdeproteinization of the surface of the interface between resinand dentin with 2% sodium hypochlorite solution for 60 s.After rinsing with water, the specimens were dehydrated inascending acetone concentrations (30, 50, 70, 90 and 100%),and critical-point dried (CPD 030; Balzers, Liechtenstein).Finally, each stick was mounted on aluminum stubs, sput-ter-coated with gold by means of the Edwards Coater S150Bdevice (Edwards Ltd., London, UK), and observed under aPhilips 515 scanning electron microscope (Philips Co.,Amsterdam, the Netherlands). Only two of the four sides ofthe sticks could be imaged, as one side was on the stub andone side was directed away from the scanning beam.

Microphotographs were taken at different standardizedmagnifications (·120, ·710, and ·1010). The low mag-nification provided an overview of each stick (Fig. 2a),whereas the photomicrographs taken at higher magnifica-tions revealed the quality of the bonding interface, as well asvarious structural defects of the specimens (Fig. 2b). Thesewere classified as microfractures within the composite resin(CR), within the adhesive resin (AR), the hybrid layer (HL),dentin (D) or enamel (E). Gaps were sometimes seen run-ning along the interfaces, such as the interface betweencomposite and adhesive (CR–AR), between the adhesiveand the top of hybrid layer (AR–THL), between the bottomof the hybrid layer and dentin (BHL–D) or the bottom ofthe hybrid layer and enamel (BHL–E) (Table 2). The mor-phological characteristics of the hybrid layer and the resintags penetrating the dentin substrate were also analysed(Fig. 2c). Every defect of each sample was counted, evalu-ating the location of the defect and including it in the cor-responding group.

Results

No premature failures of the bonds occurred duringthe preparation of the sticks. However, none of themicrotensile specimens observed under the scanning

Fig. 2. (a) Overview of a stick from the Clearfil SE group under the scanning electron microscope at a low magnification (bar,0.1 mm). (b) Higher magnification of the previous stick to detect the presence of gaps along the adhesive interface or fractures withinthe dental substrate ( bar, 0.1 mm). (c) View of the adhesive interface, to reveal the morphological characteristics of the hybrid layerand the resin tags (arrows) penetrating the dental substrate (bar, 10 lm).

Table 1

Recommended steps for the handling of the tested bonding systems

Clearfil SE Excite

Apply SE Primer and wait 20 s Apply H3PO4 for 15 sGently air-dry Rinse and dry gently, leaving the surface dampApply SE Bond Apply the adhesive for 10 sLightly blow with air Lightly blow with airLight cure for 10 s Light cure for 20 s

Defects of microtensile specimens 387

electron microscope appeared free of structural defects.These faults consisted of microcracks in the dental sub-strate, either in enamel (Fig. 3) or in dentin at the levelof hybrid layer (Fig. 4). Another type of defect that wasoften seen was the presence of gaps, either between

enamel and resin or between the hybrid layer and resin;voids within the resin composite thickness were alsosometimes visible (Fig. 5). On a few dentin specimens, asmall portion of residual enamel remained becauseof incomplete occlusal preparation (Fig. 6). A verycommon observation was the presence of microcracks in

Table 2

Frequency and location of structural defects as seen under the scanning electron microscope

Substrate Adhesive CR AR or HL D or E CR-AR AR-THL BHL-D/ENo linearinterface

Defect ofsubstrate

Dentin Clearfil SE 3 4 4 25 32 1Excite 5 5 2 22 34

Enamel Clearfil SE 4 14 4 30 20 1Excite 3 10 4 15 23 3

CR, defects (voids) within composite resin; AR, defects within adhesive resin; HL, defects within hybrid layer; D, defects withindentin; E, defects within enamel; CR-AR, defects between composite resin and adhesive resin; AR-THL, defects (microcracks)between adhesive resin and top of hybrid layer; BHL-D/E, defects between bottom of hybrid layer and dentin or enamel substrate; nolinear interface means that the adhesive interface was not positioned linearly; Defect of substrate means that the dental substrateshowed defects per se.

Fig. 3. Microphotograph of a stick taken to reveal the presenceof microcracks in the enamel substrate. An enamel crack run-ning at 90� from the bonded interface is shown (arrows, a). Aninterfacial gap is also visible (arrows, b) (bar, 10 lm).

Fig. 4. Microphotograph of a dentin stick exhibiting cracks atthe top of the hybrid layer (arrows) (D, dentin; HL, hybridlayer; CR, composite resin) (bar 0.1 mm).

Fig. 5. A stick from the Clearfil SE group exhibiting a void(arrow) in the composite layer (CR, composite resin; D, dentin)(bar, 0.1 mm).

Fig. 6. Residual enamel (E and arrows) at the periphery of astick prepared to test the dentin substrate (D) (CR, compositeresin; bar, 0.1 mm).

388 Ferrari et al.

the enamel substrate at the corners of the sticks.Regardless of the material used for bonding, the findingof microfractures was more frequent in enamel than indentin samples. Hybrid layer and resin tag formationwas noted in all samples (Figs 2 and 7).Table 2 summarizes the observed defects and their

location for each subgroup.

Discussion

The development of the microtensile method can beregarded as a significant contribution to the science ofadhesion testing (7). One of the limits of the conventionaltensile bond strength technique, revealed through FiniteElement Analysis analysis by van Noort et al. (1) andde Hoffe et al. (20), is the highly non-uniform stressdistribution across the bonded surface on which specimengeometry, material stiffness and loading configurationhave an effect. In the microtensile technique, the surfacetested is so small that the variability introduced by thesefactors is greatly reduced. Further, the regional differencesin the structure of dental tissues can be controlled, and thestress distribution across the bonding surface is thought tobe much more uniform. This allows for a more realisticand reliable appraisal of resin–tooth bond strengths.In the present study, an SEM analysis was performed

in order to detect any structural defects exhibited atresin–enamel or dentin interfaces in microtensile speci-mens before loading, as a result of the bonding or pre-paration procedures. It was revealed that none of thespecimens was flawless, and that the majority of theflaws were located on the enamel side or in resin–toothbonds.In a microtensile bond strength test performed using

the same protocol as the present research (Cardoso et al.2001, AADR, Chicago), the resin bond strengths meas-ured on enamel were not significantly higher than thoseachieved on dentin. As years of research and clinicalexperience have clearly demonstrated, bonding toenamel with both a self-etching primer and a �total-etch�system is far more reliable than adhesion on dentin; it is

therefore fair to accept that enamel samples might failunder relatively lower loading levels owing to theintrinsic brittleness of this tissue in the reduced surfaceareas used in microtensile specimens (21). This calls intoquestion whether microtensile testing is an appropriatetrial for enamel, which is fragile, anisotropic, and has awater content lower than dentin.The suspicion of an intrinsic weakness of the enamel in

microtensile specimens was confirmed in the SEM ana-lysis of the sticks. As mentioned earlier, microscopicanalysis of the sticks before loading revealed a morefrequent occurrence of microcracks in enamel than indentin. Microcracks were also most often located at theperiphery of the sticks, suggesting that these defects wereinadvertently introduced by the specimen preparationprocedures, particularly the vibrations of the cuttingdevices, disks and burs. Had the enamel cracks devel-oped before resin bonding, they would have been filledwith resin. The fact that the cracks were empty indicatesthat they developed after bonding was done.In this research, the �non-trimming� method of speci-

mens preparation was followed, which was expected tobe less traumatic than the methods where an hourglassprofile is created with burs at the bonding interface (5).In the present study, no premature failure of the resin–enamel or resin–dentin bond occurred. Nevertheless, allof the bonded interfaces in the sticks exhibited structuraldefects at various locations.It might be argued that the use of ascending concen-

trations of acetone could have been responsible for theextraction of poorly polymerized materials, thereby cre-ating voids that did not exist before such treatment.Similar technique artifacts might have been introducedwith the use of ethanol. However, if these materials were�acetone-extractable�, they were probably not contribu-ting much to the bond. Conversely, it is also true thatsome linear polymers which are soluble in acetone, suchas Polymethylmethacrylate and poly Hydroxiethyl-metharylate, can provide good bond strength. In addi-tion, microscopic images such as the example given inFigs 2 and 7 clearly demonstrate that when the resin–dentin bond is of good quality, it is also able to with-stand the challenge of exposure to acetone solutionswithout developing defects.It should then be pointed out that, in order to be

observed under the SEM, the sticks underwent a vacuumdesiccation and that the stress imposed to the specimensby this procedure may be responsible for some of thedefects detected. If epoxy resin replicas of the specimenshad been made for microscopic evaluation, bubbles andother artifacts might have been introduced. This tech-nique also tends to lower the resolution of microscopicimages to a certain extent. Detailed, high-resolutionimaging up to ·5000 can be performed with the epoxyresin replica technique using a proper impression mater-ial, adequate degassing and a high-quality microscope;however, handling the size of microtensile samples can bevery difficult.Mannocci et al. (22), in a recent analysis of dentin

microtensile specimens using a confocal microscopeat normal atmosphere pressure, frequently observed

Fig. 7. Microtensile specimen from the Excite dentin subgroupwith a clear adhesive layer (arrows). Integrity was not affectedby the preparation procedures (bar, 0.1 mm).

Defects of microtensile specimens 389

fractures or cracks of the dental substrate. In order tolimit the occurrence of this phenomenon, the authorssuggested preparing specimens no thinner than 1.5 mm.The thickness of the specimen seems to be critical in

determining its ability to survive the preparation proce-dures for microtensile testing. Bouillaguet et al. (5)reported a high incidence of premature failures (26%)during hourglass trimming of microtensile dentin speci-mens 0.5 mm thick. They suggested that, especially whenusing the trimming technique, the slabs be made thickerprior to trimming. Phrukkanon et al. (23) recommen-ded never reducing the cross-sectional area at thebonding interface to less than 1.1 mm2, since, in a pilotstudy, specimen failures greatly increased below this size.The authors believed that a 1.5 mm2 cross-sectionalsurface is the most appropriate, at least for the trimmingtechnique, as they reported a minimal percentage ofpremature failures when handling specimens of thissize (23). In the present study, the specimens were pre-pared in a beam shape, with a cross-sectional area of0.8 mm · 0.8 mm, using the non-trimming technique.Thus, although none of the specimens failed duringpreparation, many resin–enamel beams appear to havedeveloped cracks. As enamel has clearly proved to bestiffer and consequently more brittle than dentin, it maybe that, for proper microtensile bond strength testing,resin-bonded enamel does require a different specimensize from that of resin-bonded dentin. However, severallimitations of this study must be considered. Cracksobserved between the base of the hybrid layer and theunderlying mineralized dentin might correspond to the�submicron hiati� that can result from the dehydrationartifacts produced during specimen preparation. Becausethe specimens beams used in this study were small(0.8 · 0.8 mm), they easily are susceptible to dehydra-tion. The same may also be true for the microcracks thatwere noted along the surface of the hybrid layer.There is still some controversy over whether the

specimens that fail during preparation should be inclu-ded as �zero values� in the computation of the samplemean bond strength, as proposed by Shono et al. (14).Bouillaguet et al. (5), however, preferred to discard theprematurely failed specimens to avoid biasing the sam-ple. These authors reported the percentage of specimensin each experimental group that debonded prematurely,and related it to the mean tensile bond strength measuredfor that group. Through this analysis it was revealed thatthe specimens that failed prematurely most likely had abond strength of 13 MPa or lower (5). These specimenshad been prepared with the trimming technique. Sincethe non-trimming method has proved able to measurebond strengths as low as 5 MPa (24), the findings ofBouillaguet’s study provides some indirect support tothe idea that the non-trimming technique is a less trau-matic procedure for microtensile testing.For a more thorough and meaningful appraisal of the

amount of structural defects exhibited by each experi-mental specimen prior to testing, it would be desirable todevelop a method that could non-destructively detectinterfacial defects across the entire bonded surface ofthe stick. A technique able to provide good resolution

images of the specimens without exposing them to ex-treme conditions of pressure, temperature, and humidity,would be ideal. For that purpose, a field-emission envi-ronment microscope (FESEM) might be ideal.It would also appear desirable to develop a method for

a quantitative assessment of the structural integrity ofresin-bonded interfaces in microtensile specimens beforeloading. This method might permit the recording alldefects seen through a microscopic section of the wholesurface of each specimen, to finally arrive at a void ordefect score, which is a quantitative indicator of itsintrinsic strength.Going a step further, if statistical analysis revealed the

existence of a correlation between the defect score ofthe stick before testing and its measured bond strength,the score could be taken as a predictor of the specimen’sperformance under load. Through this quantitative ana-lysis of samples’ integrity before testing, the degree ofaggressiveness of the different procedures for preparingmicrotensile specimens could be better appreciated.Furthermore, by selecting from the experimental

sample those specimens that are expected to have aboutthe same intrinsic strength, one would be more confidentthat what is being tested is the actual bond strength ofthe adhesive interface. The proposal of a scientificmethod for assessing the structural integrity of micro-tensile specimens before loading will be the aim of afuture study.In conclusion, the microtensile technique is a versatile

and reliable method to test the quality of adhesion ofdental materials to different substrates. However, withinthe limitations of these study, microtensile testing shouldbe regarded as a very �technique-sensitive� method thatshould be handled with care. In order to make the testmore accurate, a standard procedure for specimen pre-paration, which places the least possible stress on thebonds, should be defined. Furthermore, a scientificmethod for assessing the structural integrity of the sticksbefore loading should be developed, in order to detectthose specimens which, as a result of an intrinsic weak-ness, might yield bond strength values that would biasthe outcome of the trial.

References1. van Noort R, Cardew GE, Howard IC, Noroozi S. The

effect of local interfacial geometry on the measurement of thetensile bond strength to dentin. J Dent Res 1991; 70: 889–893.

2. van Noort R, Noroozi S, Howard IC, Cardew G. A critiqueof bond strength measurements. J Dent 1989; 17: 61–67.

3. Sano H, Shono T, Sonoda H, Pashley DH. Relationshipbetween surface area for adhesion and tensile bond strength-evaluation of a microtensile test. Dent Mater 1994; 10: 236–240.

4. Pashley DH, Ciucchi B, Sano H, Yoshiyama M, Carvalho

RM. Adhesion testing of dentin bonding agents. A review. DentMater 1995; 11: 117–125.

5. Bouillaguet S, Ciucchi B, Jacoby T, Wataha JC, Pashley

D. Bonding characteristics to dentin walls of Class II cavities,in vitro. Dent Mater 2001; 17: 316–321.

6. Inoue S, Vargas MA, Abe Y, Lambrechts P, Vanherle G,Sano H, van Meerbeek B. Microtensile bond strength of ele-ven contemporary adhesives to dentin. J Adhes Dent 2001; 3:237–245.

390 Ferrari et al.

7. Schreiner RF, Chappell RP, Glaros AG, Eick JD. Micro-tensile testing of dentin adhesives. Dent Mater 1998; 14: 194–201.

8. Nakajima M, Sano H, Burrow MF, Tagami J, Yoshiyama

M, Ebisu S, Ciucchi B, Russel CM, Pashley DH. Tensilebond strength and SEM evaluation of caries-affected dentinusing adhesives. J Dent Res 1995; 74: 1679–1688.

9. Yoshiyama M, Carvalho RM, Sano H, Horner JA, Brewer

PD, Pashley DH. Regional bond strength of resin to humanroot dentin. J Dent Res 1996; 24: 435–442.

10. Yoshiyama M, Sano H, Ebisu S, Tagami J, Ciucchi B,Carvalho RM, Johnson MH, Pashley DH. Regional bondstrength of bonding agents to cervical sclerotic root dentin.J Dent Res 1996; 75: 1404–1413.

11. Zhang Y, Agee K, Nor J, Sachar B, Russel C, Pashley DH.Effect of acid etching on the tensile properties of demineralizeddentin matrix. Dent Mater 1998; 14: 222–228.

12. Shono Y, Terashita M, Pashley EL, Brewer PD, Pashley

DH. Effects of surface area on resin–enamel tensile bondstrength. Dent Mater 1997; 13: 290–296.

13. Pashley DH, Tay FR. Aggressiveness of contemporary self-etching adhesives. Part II. Etching effects on unground enamel.Dent Mater 2001; 17: 430–444.

14. Shono Y, Ogawa T, Terashita M, Carvalho RM, Pashley

DH. Regional measurement of resin–dentin bonding as anarray. J Dent Res 1999; 78: 669–705.

15. Tay F, King NM, Suh BI, Pashley DH. Effect of delayedactivation of light-cured resin composites on bonding of all-in-one adhesives. J Adhes Dent 2001; 3: 207–225.

16. Tay FR, Smales RJ, Ngo H, Wei SHY, Pashley DH. Effect ofdifferent conditioning protocols on adhesion of a GIC to den-tin. J Adhes Dent 2001; 3: 153–166.

17. Armstrong SR, Keller JC, Boyer DB. The influence of waterstorage and C-factor on the dentin–resin composite microten-sile bond strength and debond pathway utilizing a filled andunfilled adhesive resin. Dent Mater 2001; 17: 268–276.

18. Armstrong SR, Boyer DB, Keller JC. Microtensile bondstrength testing and failure analysis of two dentin adhesives.Dent Mater 1998; 14: 44–50.

19. Armstrong SR, Keller JC, Boyer DB. Mode of failure in thedentin–adhesive resin–resin composite bonded joint as deter-mined by strength-based (TBS) and fracture-based (CNSB)mechanical testing. J Dent 2001; 17: 201–210.

20. de Hoff PH, Anusavice KJ, Wang Z. Three-dimensional finiteelement analysis of the shear bond test. Dent Mater 1995; 11:126–131.

21. Carvalho RM, Santiago SL, Fernandes CAO, Suh B,Pashley DH. Effects of prism orientation on tensile strength ofenamel. J Adhes Dent 2000; 2: 251–257.

22. Mannocci F, Scheriff M, Ferrari M, Watson TW. Micro-tensile bond strength and confocal microscopy of dental adhe-sives bonded to root canal dentin. Am J Dent 2001; 14: 100–104.

23. Phrukkanon S, Burrow MF, Tyas MJ. The influence ofcross-sectional shape and surface area on the microtensile bondtest. Dent Mater 1998; 14: 212–221.

24. Pashley DH, Carvalho RM, Sano H, Nakajima M,Yoshiyama M, Shono Y, Fernandes CA, Tay F. The micro-tensile bond test: a review. J Adhes Dent 1999; 1: 299–309.

Defects of microtensile specimens 391