dental

materials
Dental Materials 18 (2002) 359±369
www.elsevier.com/locate/dental

Micro-tensile bond strengths to sclerotic dentin using a self-etching and a

total-etching technique
S.M. Kwong a, G.S.P. Cheung a, L.H. Kei a, A. Itthagarun a, R.J. Smales b, F.R. Tay a,*, D.H. Pashley c
a
Faculty of Dentistry, The University of Hong Kong, The Prince Philip Dental Hospital, 34 Hospital Road, Hong Kong
b
Dental School, Adelaide University, 5005, Australia
c
Department of Oral Biology/Physiology, School of Dentistry, Medical College of Georgia, Augusta, GA 30912-1129, USA
Received 13 June 2000; received in revised form 2 February 2001; accepted 20 February 2001

Abstract
Objectives: To evaluate the in vitro regional micro-tensile bond strengths of a self-etching/self-priming adhesive system to sclerotic
dentin, in the absence or presence of phosphoric acid conditioning.
Methods: Naturally-occurring, non-carious cervical lesions on extracted premolars were hand-cleaned with a slurry of pumice and
chlorhexidine, then bonded without further cavity preparation. One group was bonded using Clear®l Liner Bond 2V (Kuraray Co. Ltd.,
Osaka, Japan). The other group was ®rst conditioned with K-etchant; 40% phosphoric acid gel (Kuraray) prior to the application of the self-
etching primer. Arti®cially prepared wedge-shaped cavities were also made in sound premolars and bonded with the two methods as controls.
Resin composite build-ups were made using Clear®l Protect Liner F and AP-X resin composite (Kuraray). After storage in water for 24 h, the
teeth were sectioned into 0.7 £ 0.7 mm composite-dentin beams along the occlusal and gingival walls, and at the apex of the lesions. The use
of two conditioning methods, two substrate types, and three different locations yielded 10±14 beams for each of the 12 groups. After testing
for the mTBS, representative beams that were stressed to failure were examined with SEM. Remaining fractured beams were demineralized
and processed for TEM examination. Statistical analysis was performed using a three-way ANOVA and Student-Newman±Keuls tests.
Results: Regardless of the conditioning methods, bond strengths to sound dentin were signi®cantly higher than to sclerotic dentin (P ,
0.05). With sclerotic dentin, there was no signi®cant difference for the conditioning methods used, except that K-etchant signi®cantly
improved the bond strength at the gingival aspect of the lesions. Fractographic analysis revealed that the self-etching primer could not etch
beyond the surface hypermineralized layer of sclerotic dentin. Interfacial failure occurred along the surface of the mineralized intermicrobial
matrix and/or hypermineralized layer. With the use of phosphoric acid, a hybrid layer was only seen when the surface layers were thin.
Incompletely removed sclerotic casts were evident in both groups.
Signi®cance: Removal of the surface layers of sclerotic dentin and/or conditioning with stronger acids may be bene®cial to obtain stronger
bonding to sclerotic dentin. q 2002 Academy of Dental Materials. Published by Elsevier Science Ltd. All rights reserved.
Keywords: Micro-tensile; Sclerotic dentin, Self-etch, Total-etch

1. Introduction surface of noncarious cervical lesions. Being highly acid-

Compared with enamel, dentin is a more dif®cult bonding
substrate due to its complex and variable histological
features [1]. Post-eruption physiological and pathological
changes may further result in sclerotic dentin [2] that is
less receptive to bonding protocols that are designed to be
used on sound dentin [3]. Previous studies have reported the
obliteration of dentinal tubules with both peritubular dentin
and sclerotic casts that compromise the formation of resin
tags [4±9]. Recently, Tay et al. [10] con®rmed morpho-
logically the existence of a hypermineralized layer on the
* Corresponding author. Tel.: 1852-2859-0251; fax: 1852-2559-3803.
E-mail address: kfctay@hknet.com (F.R. Tay).
0109-5641/02/$22.00 + 0.00 q 2002 Academy of Dental Materials. Published by Elsevier Science Ltd. All rights reserved.
PII: S 0109-564 1(01)00051-3
resistant, the retention of such a layer precludes optimal
acid-etching of sclerotic dentin [2,9,11,12]. The bonding
scenario may be further complicated by the presence of
bacteria overgrowth [13], and their erosion and entrapment
within the surface hypermineralized layer of these lesions
[10]. These surface obstacles, acting alone or together,
prevent optimal in®ltration of dentin adhesive into the
underlying dentin.
Contemporary self-etching primers are designed to etch
through smear layers into the underlying dentin [14]. Unlike
the total-etch technique that involves separate acid etching
and rinsing steps [15], self-etching primers do not require
rinsing and are more user-friendly. Because acidic resin
monomers are used as both the etchant and the priming
360 S.M. Kwong et al. / Dental Materials 18 (2002) 359±369

resins, the likelihood of incomplete resin in®ltration within mTBSs to sound or sclerotic dentin regardless of the bond-
the partially demineralized dentin [16] is reduced. This was ing strategy (self-etching vs total-etch) and the location
demonstrated in a previous nanoleakage study that within the cervical lesion.
compared self-etching with total-etching systems [17].
Despite the presence of a thin hybrid layer, high initial
2. Materials and methods
bond strengths were obtained when self-etching primers
were applied on sound dentin [18±20]. However, bond
Twenty premolars with deep, natural cervical wedged-
strengths were compromised when self-etching primers
shaped lesions (group N) that had been extracted for perio-
were used on sclerotic dentin [7,21]. Although unabraded
dontal reasons were collected. Another 20 sound premolars
sclerotic dentin is devoid of a smear layer, it has been shown
extracted for orthodontic reasons were used as control
that self-etching primers could not etch through the surface
groups in this study. The teeth were stored in 0.5% chlora-
hypermineralized layer in sclerotic dentin [10]. Previous
mine T solution at 48C, and used within 1 month following
studies suggested that bonding to sclerotic dentin may be
extraction.
improved by modi®cations to protocols that are normally
The natural cervical lesions were ®rst cleaned with an
employed for sound dentin. These include removal of the
ultrasonic scaler to remove all signs of calculus from the
surface hypermineralized layer [22], or the use of stronger
adjacent root dentin. The internal surface of each lesion was
acids [9]. Based on the encouraging clinical results with the
then cleaned by hand scrubbing with a bonding brush
use of the total-etch technique on noncarious cervical
together with a mixture of 2% chlorhexidine and pumice.
lesions [23], it is envisaged that bonding of self-etching
As the lesions were highly shiny before hand-cleaning, the
primers to sclerotic dentin may likewise be enhanced with
bonding surfaces were all devoid of calculus deposits. In
the adjunctive use of phosphoric acid pre-conditioning [24].
order to examine the effect of bonding to the surface hyper-
Assessment of contemporary dentin adhesives, with
mineralized layer of these lesions, we did not clean the
improved bonding ef®cacy, by using conventional shear
bonding surfaces with slow-speed rubber cups, bristle
bond strength tests have been frequently criticized for
brush or a prophy-jet device. This was due to the relative
their eccentric stress distribution that often resulted in cohe-
substantial loss of tooth substances associated with these
sive failure within the tooth substrates [25]. The micro-
cleaning techniques [36,37] when compared with the thick-
tensile bond test has been used successfully for testing
ness of the surface hypermineralized layer in cervical
specimens with small cross-sectional areas [26]. The
sclerotic dentin [38]. After cleaning, the lesions were rinsed
advantage of this technique is that regional bond strength
with distilled water and further inspected with a stereo-
differences within a speci®c part of a single tooth, such as a
scopical microscope to ensure that the bonding surfaces
cervical wedge-shaped lesion, may be evaluated [27]. This
were free from pumice and retained plaque. They were
technique also permits a more uniform stress distribution
then restored without further cavity preparation. Arti®cial
along the interface between dentin and resin during testing
wedge-shaped lesions (group A), 4 mm wide and 3 mm
[28]. The original `trimming version' of the micro-tensile
deep, were prepared in the sound premolars along the buccal
test utilized dumbbell- or hourglass-shaped specimens. This
cervical region. A high-speed handpiece equipped with a
technique has been used to evaluate regional bond strengths
tungsten carbide bur was used for preparing the arti®cial
to sclerotic dentin [7], carious dentin [29,30], super®cial
lesions, under copious water-cooling. A new bur was used
versus deep dentin [31], and different parts of root dentin
for every ®ve lesions.
[21]. Durability of dentin bonding may also be effectively
evaluated in longitudinal studies [32]. Other materials such 2.1. Restoration of prepared lesions
as glass ionomer cements and resin-modi®ed glass ionomer
cements bonded to dentin have also been examined with this Both natural and arti®cial lesions were randomly
technique [33,34]. The recently introduced `non-trimming assigned to one of the two bonding protocols (Table 1). In
version' [35] of the micro-tensile test further employs the `self-etch' protocol (S), Clear®l Liner Bond 2V (Kuraray
beams of uniform cross-sectional areas. This enables more Co. Ltd., Osaka, Japan) was applied according to the manu-
testing samples to be produced from pathological specimens facturer's instructions. The self-etching primer liquid A and
such as human natural sclerotic lesions, which are becoming B were mixed immediately before bonding. The primer
increasingly dif®cult to collect. mixture was applied to the entire lesion with a disposable
This study evaluated the micro-tensile bond strengths brush tip and left undisturbed for 30 s. Following evapora-
(mTBSs) of a self-etching/self-priming adhesive system tion of the volatile solvent, two coats of adhesive were
following self-etching or total-etching of sclerotic dentin applied to the etched lesion and cured separately for 20 s
in cervical tooth lesions. Scanning (SEM) and transmission each.
electron microscopy (TEM) were further employed for In the `total-etch' protocol (T), each lesion was ®rst
fractographic analyses of specimen beams that were conditioned with the 40% phosphoric acid gel (K-etchant;
stressed to failure under tension. The null hypothesis that Kuraray) for 15 s. The acid was rinsed with distilled water
was tested was that there is no signi®cant difference in for 20 s, and the etched surface was left slightly moist before
 S.M. Kwong et al. / Dental Materials 18 (2002) 359±369 361

Table 1
Identi®cation of specimen groups by lesion type, bonding protocol and beam location

Lesion type Bonding protocol Beam location Number of specimens Group designation
a
Arti®cial lesion (A) (sound dentin, control) Self-etch (S) Occlusal wall (O) 12 ASO
Apex of lesion (P) b 12 ASP
Gingival wall (G) 12 ASG
Total-etch (T) c Occlusal wall (O) 10 ATO
Apex of lesion (P) 11 ATP
Gingival wall (G) 10 ATG
Natural lesion (N) (sclerotic dentin) Self-etch (S) Occlusal wall (O) 14 NSO
Apex of lesion (P) 13 NSP
Gingival wall (G) 14 NSG
Total-etch (T) Occlusal wall (O) 12 NTO
Apex of lesion (P) 13 NTP
Gingival wall (G) 13 NTG
a
Clear®l Liner Bond 2V (Kuraray Co. Ltd., Osaka, Japan) only.
b
Deepest part of a wedge-shaped lesion.
c
Forty percent phosphoric acid (K-etchant, Kuraray) on enamel and dentin, rinse, followed by the application of Clear®l Liner Bond 2V to wet, deminer-
alized bonding substrates.

bonding with the Clear®l Liner Bond 2V primer mixture.
The primer mixture was gently evaporated after application
to evaporate the solvent. Two coats of adhesive were added
and light-cured, as before. We realized that this water-
containing self-etching primer might not bene®t from a
moist bonding technique when it is used on acid-etched
dentin. In view of recent reports that a hydroxyethyl metha-
crylate±water mixture could not completely re-expand
a collapsed, air-dried collagen matrix [39], and the
potential of incomplete resin in®ltration when a self-
etching primer was used on acid-etched dentin [40], we
continued to adopt a moist bonding technique in this part
of the experiment.
After the bonding procedures, each lesion was ®rst
restored with a thin layer of Clear®l Protect Liner F, a
low-viscosity resin composite (Kuraray). This was followed
by the placement of a resin composite (Clear®l AP-X;
Kuraray). After aging in water for 24 h at 378C, the existing
restoration, together with the adjacent enamel and cemen-
tum, were further etched with the self-etching primer and
treated with the adhesive resin. Additional resin composite
was then applied in several increments to form a core
over the existing restoration. Following the protocol of
Yoshiyama et al. [7], this two-step composite placement
technique ensued that adequate composite was available
for subsequent beam preparation from different parts of
the lesion surface, without creating excessive shrinkage
stress along the bonded interface (Fig. 1).
2.2. m TBS evaluation
Each tooth was sectioned in a buccolingual direction into
2±3 slabs, using a slow-speed saw with a diamond-impreg-
nated disk (Isomet; Buehler Ltd., Lake Bluff, IL, USA)
under water lubrication. Each slab was about 0.7 mm
thick, and contained a longitudinal section of the wedge-
shaped, natural or arti®cial lesion. The slabs were then
secured with sticky wax onto clear Plexiglass blocks.
Beams with a uniform cross-sectional area of approximately
0.45 mm 2 were prepared, by making parallel cuts that were
perpendicularly aligned with the lesion surfaces. By using
serial slabs, beams that included the occlusal wall (O),
gingival wall (G), as well as the apex of the lesion (P)
were produced for each arti®cial or natural lesion (Fig. 1).
The use of two lesion types (A or N), two bonding
protocols (S or T), and three lesion locations (O, G, or P)
resulted in 12 designated groups (Table 1). The `non-trim-
ming' technique [35] enabled us to prepare more than one
beam from either the gingival or occlusal interfacial
surfaces of each sectioned slab. This resulted in 10±14
beams for each group, for evaluation of the regional tensile
bond strength.
Each beam was attached to a modi®ed Bencor Multi-
T testing assembly (Danville Engineering, San Ramon,
CA, USA) using a cyanoacrylate adhesive (Zapit; DVA,
Corona, CA, USA). The beams were stressed to failure
under tension (Fig. 1), using an Instron universal tester
(Model 4440; Instron Inc., Canton, MA, USA) at a
crosshead speed of 1 mm/min. The exact dimension of
each fractured beam was then individually measured
using a digital caliper (Model CD-6BS; Mitutoyo,
Tokyo, Japan), from which the tensile bond strength was
calculated.
Failure modes of the debonded specimens were initially
evaluated at 30X with a stereoscopical microscope
(SMZ10; Nikon, Tokyo, Japan). Failures were classi®ed
as: (a) adhesive failure, if the fracture site was maintained
entirely within the adhesive; (b) mixed failure, if the
fracture site continued from the adhesive into either
the resin composite or dentin; and (c) cohesive failure, if
the fracture occurred exclusively within the resin compo-
site or dentin.
362 S.M. Kwong et al. / Dental Materials 18 (2002) 359±369
Fig. 1. Schematic representation of the protocol employed for regional
tensile bond strength evaluation and fractographic analysis.
2.3. Statistical analysis
Bond strength data obtained for the 12 designated groups
were analysed using a three-way ANOVA design to evalu-
ate the effect of the three experimental factors: (1) bonding
substrate (arti®cial control lesions vs natural lesions), (2)
conditioning protocol (self-etch versus total-etch), and (3)
beam location (occlusal, gingival and the apex of the
lesion), as well as the interaction of these three factors on
tensile bond strength of the self-etching primer. Statistical
signi®cance was set in advance at the 0.05 probability level.
Multiple comparisons were performed using Student-
Newman±Keuls tests at a ˆ 0.05.
2.4. SEM examination
For SEM fractographic analysis, the dentin and compo-
site sides of four fractured beams from each group with
mTBS that were close to the mean value were examined.
As bacteria were anticipated on the fractured surfaces, the
samples were ®xed with 2% glutaraldehyde in 0.1 M
sodium cacodylate buffer (Electron Microscopy Sciences,
Fort Washington, PA, USA) for 1 h, and further rinsed
three times in 0.1 M sodium cacodylate buffer. Both sides
of the fractured beams were then af®xed vertically to an
aluminium stub with Zapit, and allowed to air-dry com-
pletely in a desiccator. They were sputter-coated with
gold/palladium for 4 min at 5 mA, and examined using a
scanning electron microscope (Cambridge Stereoscan 440;
Cambridge, UK).
2.5. TEM examination
Four additional fractured beams from each group that
were initially classi®ed as adhesive failures were further
prepared for TEM examination. Only the dentin sides of
these beams were examined. These 0.7 £ 0.7 mm beams
were demineralized for 24 h in an aqueous solution of
0.1 M ethylene diamine tetra-acetic acid (EDTA) that was
buffered with sodium hydroxide to pH 7. The specimens
were then ®xed in Karnovsky's ®xative (2.5% glutaralde-
hyde and 2% paraformaldehyde in 0.1 M cacodylate buffer,
pH 7.3) for a minimum of 1 h and rinsed thoroughly with
0.1 M sodium cacodylate buffer. Post-®xation were per-
formed with 1% osmium tetroxide in 0.1 M sodium caco-
dylate buffer (pH 7) for 1 hr at room temperature. The ®xed
specimens were rinsed three times in cacodylate buffer with
a dwell time of 1 min each. They were dehydrated in an
ascending ethanol series (30±100%), immersed in propyl-
ene oxide as a transition ¯uid. Resin embedding was
performed using TAAB 812 epoxy resin (TAAB Labora-
tories, Aldermaston, UK) at 608C for 48 h. Thin sections
were prepared from all processed TEM specimens and
used for initial screening. Representative 1 £ 1 mm ultrathin
sections of about 70 nm thick were then prepared with
an ultramicrotome (Reichert Ultracut S; Leica, Vienna,
Austria) using a diamond knife (Diatome, Bienne, Switzer-
land), and collected on 150 mesh copper grids (TAAB
laboratories, Aldermaston, UK). Grids were double stained
with 2% uranyl acetate for 10 min, followed by Reynold's
lead citrate for an additional 7 min. After drying, the stained
sections were further coated with carbon to minimize
deterioration of the resin composite matrix that resulted
from heat absorption by the ®ller particles. The coated
sections were examined with two transmission electron
microscopes (Model EM208S and CM100; Philips,
Eindhoven, The Netherlands), operating at 80±100 kV.
 S.M. Kwong et al. / Dental Materials 18 (2002) 359±369 363

Table 2
Regional tensile bond strengths (MPa) of arti®cial control lesions and natural cervical sclerotic lesions that were bonded using either a self-etching or a total-
etching technique [Values are mean (SD). Groups that bear the same superscripts were not statistically signi®cant when analyzed using Student-Newman±
Keuls multiple comparison tests (P . 0.05)]

Lesion type Bonding protocol Beam location

Occlusal (O) Apex (P) Gingival (G)

Articial lesions (A) Self-etch (S) 66.6 (12.6) a 63.5 (20.6) a 67.4 (20.3) a
Total-etch (T) 66.0 (24.0) a 77.3 (9.8) a 62.3 (15.4) a
Natural lesions (N) Self-etch (S) 53.7 (23.2) b 47.9 (27.8) b 44.3 (19.9) b
Total-etch (T) 43.3 (15.8) b 44.3 (19.9) b 64.2 (20.0) a

3. Results The distribution of failure modes as determined by optical

microscopy is presented in Table 4. Bond failures were
3.1. m TBS bond strength largely adhesive or mixed.

Regional bond strengths of the two conditioning protocols
on sound and sclerotic dentin are presented in Table 2. Results
of the statistical analysis are shown in Table 3. Three-way
ANOVA showed that the mean bond strength to sclerotic
dentin was signi®cantly lower than that to sound dentin
regardless of the conditioning protocols and beam locations.
The mean tensile bond strength produced by the self-etching
primer alone to natural lesions (48.7 MPa) was 26% lower
than that to arti®cial lesions (65.8 MPa) when all of the bond
strengths were pooled, the result being statistically signi®cant
(P , 0.001). Similarly, the mean bond strength of pooled data
using the self-etching primer with adjunctive phosphoric acid
pre-conditioning (total-etch) to natural lesions (53.1 MPa)
was 24% lower than that recorded in arti®cial lesions
(69.8 MPa), the result also being statistically signi®cant
(P , 0.005). However, pooled data showed no signi®cant
differences between the mean bond strengths for self-etching
or total-etching, to either sound dentin (P ˆ 0.42) or sclerotic
dentin (P ˆ 0.31). Of the three factors tested (substrate,
conditioning protocol and beam location), only the type of
substrate (sound dentin vs sclerotic dentin) was found to have
a signi®cant in¯uence on bond strength (P , 0.05). Multiple
comparison tests showed that there was no signi®cant
difference in self-etching or total-etching sclerotic dentin
except for the gingival wall of the lesions, in which higher
bond strengths were obtained for total-etching (64.2 MPa).
3.2. Fractographic analysis
In arti®cial lesions that were conditioned with the self-
etching primer only, the smear layers on sound dentin were
not completely removed. The thickness of hybrid layers,
when examined using TEM, was limited to 0.5 mm (not
shown). On the other hand, phosphoric acid removed the
smear layers and smear plugs completely, exposing patent
dentinal tubules and demineralized intertubular dentin for
the formation of hybrid layers and resin tags. The corre-
sponding TEM revealed the formation of a hybrid layer of
5±6 mm in thickness (not shown). Despite the difference in
thickness of hybrid layers, there was no obvious difference
in the failure modes. Both conditioning protocols exhibited
predominantly adhesive failure between sound dentin and
resin composite.
Fig. 2(a) is a SEM micrograph of a mixed failure mode
from a natural sclerotic lesion that was treated with the self-
etching primer only. The super®cial obstacle layer, which
was not removed in the self-etching protocol, was trapped
within the interface, part of which was exposed when the
beam was stressed to failure. A higher magni®cation of the
dentin at the left side of the beam revealed colonies of
bacteria embedded in the adhesive resin [Fig. 2(b)].
The relation between the surface obstacles and bond fail-
ure was better demonstrated with the use of TEM. Fig. 3(a)

Table 3
Statistical analysis of microtensile bond strength results using three-way analysis of variance

Source of variation Degree of freedom Sum of squares Mean square F-ratio Probability level Power (a ˆ 0.05)

A Bonding substrate 1 10324.65 10324.65 27.13 0.000001 a 0.99

B Conditioning protocol 1 570.03 570.03 1.50 0.22 0.23
C Beam location 2 239.17 119.58 0.31 0.73 0.10
AB 1 2.75 2.75 0.01 0.93 0.05
AC 2 487.75 243.87 0.64 0.53 0.16
BC 2 1608.33 804.16 2.11 0.12 0.43
ABC 2 2085.22 1042.61 2.74 0.07 0.53
Residual 134 50993.87 380.55
Total (adjusted) 145 66451.32
a
Term signi®cant at alpha ˆ 0.05.
364 S.M. Kwong et al. / Dental Materials 18 (2002) 359±369

Table 4
Distribution of failure modes

Lesion type Bonding protocol Failure modes a

Adhesive Mixed Cohesive in resin composite Cohesive in dentin

Articial lesions (A) Self-etch (S) 28 (77.8%) 4 (11.1%) 4 (11.1%) 0 (0%)

Total-etch (T) 25 (80.6%) 4 (12.9%) 0 (0%) 2 (6.5%)
Natural lesions (N) Self-etch (S) 20 (48.8%) 18 (43.9%) 1 (2.4%) 2 (4.9%)
Total-etch (T) 17 (44.6%) 14 (36.8%) 4 (10.5%) 3 (7.9%)
a
For each lesion type, pooled data from the occlusal, gingival and apex of lesions bonded with the same protocol were used in the calculation of failure mode
distribution.

was taken from a beam that originated from the apex of a

Fig. 2. (a) SEM micrograph of the dentin side of a fractured beam that was
harvested from the occlusal wall of a natural sclerotic lesion. The lesion was
treated with a self-etching primer only. A mixed failure mode could be
observed, with cohesive failure within the resin composite (C), and adhe-
sive failure along the resin±dentin interface (D). A secondary crack was
also evident within the sclerotic dentin (pointer). Asterisk: site from which
a higher magni®cation was taken. Bar ˆ 30 mm. (b) A higher magni®cation
of the site marked with an asterisk in Fig. 2(a). Bacteria-like structures
(pointers) were embedded in the adhesive resin. Bar ˆ 3 mm.
highly-sclerotic natural lesion. The thickness of the hyper-
mineralized layer was about 14 mm. Two types of bacteria
could be seen, one along the surface and the other that has
eroded into the hypermineralized layer. Bond failure
occurred along the surface bacterial layer. Variability in
the clinical substrates is further illustrated in Fig. 3(b). In
this specimen, a surface complex could be recognized,
consisting of thin, isolated sections of hypermineralized
layers that were sandwiched among different layers of
bacteria. These sections were probably interconnected
three-dimensionally. The primary plane of failure was
found within the Protect Liner F (not shown). However, a
secondary crack plane was observed within the bacterial
layers. The spaces between the separated bacteria were in®l-
trated with epoxy resin. Similar to the previous sample, no
hybrid layer was observed within the intertubular dentin.
When the natural lesions were ®rst total-etched before
self-etching primer application, the etching result was
improved, but was still site-dependent and variable.
Where the hypermineralized layer and/or bacterial matrix
were thin, they were either completely or partially dissolved
and rinsed away, exposing underlying sclerotic dentin. SEM
examination showed that in areas that were devoid of
sclerotic casts, resin tags were present, and were pulled
out of the dentinal tubules upon beam failure [Fig. 4(a)].
Even in these areas, remnant bacteria could be found. In
other specimens that demonstrated adhesive failure, the
fractured dentin surfaces were covered with bacteria that
remained tenaciously attached to the sclerotic dentin even
after phosphoric acid etching and rinsing [Fig. 4(b)].
TEM examination con®rmed that hybrid layer formation in
sclerotic dentin remained highly variable despite the
adjunctive use of phosphoric acid pre-conditioning. Where a
thick hypermineralized layer was present, the hybrid layer was
either reduced in thickness or completely absent. In Fig. 5(a),
taken from the apex of a sclerotic lesion, the hybrid layer was
reduced from 5 to about 2 mm in the presence of the
hypermineralized layer. Although the failure mode was
primarily cohesive in nature within the Protect Liner F, a
secondary subsurface crack could be seen between the
 S.M. Kwong et al. / Dental Materials 18 (2002) 359±369
Fig. 3. TEM micrographs of fractured sclerotic dentin specimens that were
treated with a self-etching primer prior to composite restoration. (a) Part of
a fractured beam from the apex of a natural sclerotic lesion. Adhesive
failure occurred along the surface bacterial layer (arrow). The hyperminer-
alized layer (HM) was 14 mm thick and was partially eroded with another
type of bacteria. No hybrid layer could be seen within the underlying
sclerotic dentin. Bar ˆ 3 mm. (b) In this fractured specimen, the primary
crack plane occurred within the Protect Liner F (not shown). Separation of
the bacteria±hypermineralized layer complex also occurred (arrows), with
the resultant spaces in®ltrated by epoxy resin. A: ®lled adhesive; Arrow-
heads: hypermineralized portions of the complex. Bar ˆ 3 mm (applicable
to all ®guresÐB: bacteria; SD: sclerotic dentin; E: epoxy resin).
hypermineralized layer and the underlying sclerotic dentin.
Propagation of this secondary crack occurred along the inter-
face where there was no hybrid layer, and was resisted by the
presence of the latter. The hypermineralized layer was thin
along the gingival wall of the sclerotic lesions [Fig. 5(b)].
However, hybrid layer formation was still erratic in the
presence of a thick bacterial layer. An adhesive failure
resulted, in which part of the ®lled adhesive was retained
along the interface with a thick hybrid layer. Fracture of the
adhesive also occurred along the surface of the bacterial layer.
365
4. Discussion
In the present study, no signi®cant difference in mean
bond strengths was observed when sound dentin was either
self-etched or total-etched, in spite of the thicker hybrid
layers observed in the latter groups. This suggested that
thickness of the resin-in®ltrated zone was not related to
bond strength, which is in agreement with previous ®ndings
[20,41]. The smaller cross-sectional areas that we used for
regional tensile bond strength assessment are re¯ected by
the higher bond values obtained for both our control and
experimental groups [26,42]. This result is due to the
decreased likelihood of encountering interfacial defects
that approached the critical ¯aw size for spontaneous
crack propagation [43].
Although bonding of resin composites with dentin adhe-
sives to sclerotic cervical lesions have improved consider-
ably over the past few years [44,45], a higher clinical failure
rate was still observed when compared with the use of glass
ionomer-based materials [46] or with bonding to sound
cervical dentin [47]. For both the self-etch and the total-
etch technique employed in this in vitro study, mean bond
strengths to sclerotic dentin were about 25% lower than the
corresponding values to sound dentin, when data from
different beam locations were pooled. This is consistent
with the 25±40% drop in mTBS that was reported by
Yoshiyama et al. [7]. In our study, barriers to diffusion
along the lesion surfaces were deliberately left intact prior
to bonding. On the other hand, natural lesions were
mechanically polished with pumice and rubber cups in the
study of Yoshiyama et al. [7]. This procedure could have
removed most of the surface obstacles, including the hyper-
mineralized layer and bacteria overgrowths. It is interesting
to note that a decline in bond strength was observed for
sclerotic dentin, irrespective of how the lesion surfaces
were cleaned. This may be partly explained by the retention
of acid-resistant sclerotic casts following both cleaning
techniques, that precluded resin in®ltration within the dent-
inal tubules. Another factor to be considered in our study is
that the surface of sclerotic dentin is devoid of a smear layer
when the hypermineralized layer is left intact. Whereas
mechanical polishing may result in reducing the thickness
of the surface hypermineralized layer, a tenacious smear
layer is invariably formed along the lesion surface, consist-
ing of hypermineralized remnants that are likewise resistant
to acid dissolution. Gwinnett [48] reported that resin in®l-
tration accounted for about one third of the bond strength in
acid-etched, sound dentin that was devoid of surface diffu-
sion barriers. Thus, it is logical that retention of diffusion
barriers, either in the form of an intact hypermineralized
layer or an acid-resistant smear layer, resulted in a similar
percentage decline in bonding to sclerotic dentin.
There are two problems associated with fractographic
analysis of bonds made with self-etching primers to sclero-
tic dentin. First, hybrid layers produced using self-etching
primers are too thin to be seen by conventional SEM. The
366 S.M. Kwong et al. / Dental Materials 18 (2002) 359±369

Fig. 4. (a) A SEM micrograph of a fractured beam taken from a natural lesion that was etched with phosphoric acid before self-etching primer application.
Numerous resin tags could be seen (pointers), some of which were surrounded by a layer of bacteria (arrow). Bar ˆ 2 mm. (b) The dentin side of a fractured
beam from another phosphoric acid-etched sclerotic lesion. Failure either occurred either along the surface of, or within the bacterial layer. Bar ˆ 1 mm.

second problem is that the fractured interfaces in bonded
sclerotic dentin are far too complex to assess accurately by
SEM examination alone. TEM fractographic analysis of
bond failures provides important, complementary informa-
tion on the strengths and weaknesses of a biological joint
comprising multiple components derived from both bio-
logical and pathological substrates, and synthetic materials,
as well as the interactions among these highly variable
components [49]. TEM analysis also allows the visualiza-
tion of noncatastrophic subsurface defects beneath the
primary crack plane [50] that are usually hard to detect
with the use of SEM. By examining resin±dentin interfaces
that have not been subjected to mechanical testing as a
control, we con®rmed that these cracks were true cracks
formed during bond strength testing and not artifacts
produced by sectioning with a diamond knife [51].
When the bond strength results were examined together
with the ultrastructural features of the resin±sclerotic dentin
interface, there were three factors that may have in¯uenced
the overall decrease in bond strength found in natural
sclerotic lesions:
1. inability to etch through the surface hypermineralized
layer of the lesion, particularly when such a layer is
thicker than 0.5 mm when a mildly aggressive self-
etching primer is used alone;
2. inability to remove the sclerotic casts that obliterate the
dentinal tubules, and the consequent lack of effective
resin tag formation; and
3. bacteria entrapment within the resin layer may introduce
defects that weaken the ultimate tensile strength of the
polymerized adhesive resin.
 S.M. Kwong et al. / Dental Materials 18 (2002) 359±369
Fig. 5. TEM micrographs of sclerotic dentin specimens that were pre-condi-
tioned with phosphoric acid. (a) Part of a fractured beam taken from the
apex of a natural sclerotic lesion. Cohesive failure occurred within the
Protect Liner F (P). Additional separation of the hypermineralized layer
(HM) from the underlying sclerotic dentin occurred along the part of the
interface that was devoid of a hybrid layer. Crack propagation was resisted
(arrow) by the presence of a 5 mm thick hybrid layer (H) along the part of
the interface in which the hypermineralized layer was absent. Bar ˆ 10 mm.
(b) Part of a fractured beam from the gingival wall of a natural sclerotic
lesion. Adhesive failure occurred within the ®lled adhesive (arrow). Hybrid
layer formation (H) was inconsistent; being 5 mm thick along the interface
that was devoid of bacteria, and almost completely absent below areas that
contained a thick, surface bacterial layer (B). The hypermineralized layer
was very thin (arrowhead). Bar ˆ 5 mm (applicable to all ®guresÐSD:
sclerotic dentin; H: hybrid layer; A: ®lled adhesive; E: epoxy resin).
The complexity of failure modes depicted in the series of
micrographs presented clearly indicates that reduced bond
strength to sclerotic dentin is not related to any single factor.
Similar to other biological variations, it is possible that each
factor contributes to a variable degree in different lesions.
The summation of all these factors, however, leads to an
overall reduction in bond strength.
367
In the presence of a surface hypermineralized layer that is
greater than 0.5 mm thick, no hybrid layer could be seen
when the self-etching primer investigated was used alone
on sclerotic dentin. This is a reasonable observation, as the
self-etching primer can only form a 0.5 mm thick hybrid
layer in sound intertubular dentin. However, our results
indicated that bond strengths of self-etched sclerotic dentin
were comparable to those in total-etched sclerotic dentin.
One possible explanation is that the self-etching primer
diffused through the intermicrobial matrix and partially
demineralized the hypermineralized layer. This created
enough porosity both around and within the larger, regularly
arranged, plate-like crystallites in the hypermineralized
layer for effective resin retention [10]. This is comparable
with bonding to aprismatic enamel [52]. However, since a
layer of hybridized sclerotic dentin is not produced, the
bond of the entire assembly will be highly dependent
upon the strength of the hypermineralized layer to the intact
sclerotic dentin. This weakness is nicely illustrated in
Fig. 3(b), and may also contribute to the reduced bond
strength in bonding to sclerotic dentin. This situation is
analogous to the application of second generation dentin
adhesives to smear layer-covered sound dentin. As the
smear layer was not completely in®ltrated, bond strength
at the interface was dependent upon the strength of the
attachment of the smear layer to the underlying dentin
[53]. Unfortunately, no data are available on the interfacial
strength of the hypermineralized layer to the surface of
sclerotic dentin.
The occurrence of erratic hybrid layers of highly uneven
thickness provides additional insights on bond failures that
occurred in total-etched sclerotic dentin. In areas where a
thick hypermineralized layer was present, even phosphoric
acid could not etch through this diffusion barrier to create a
hybrid layer in sclerotic dentin [Fig. 5(a)]. The fact that the
primary crack plane occurred within the resin composite
was probably a result of some hybrid layer formation
along parts of the interface. It is interesting to observe that
separation of the hypermineralized layer from the dentin
surface occurred in the absence of an underlying hybrid
layer. On the other hand, the junction between the ®lled
adhesive and dentin was intact where the hybrid layer was
present. This further substantiates the shock-absorbing [54]
or stress-relieving [55] role played by the hybrid layer when
the specimen beam was stressed to failure under tension.
The use of Protect Liner F as an built-in shock-absorber
could also have improved the effectiveness of the adhesive
tested based on the `elastic cavity wall concept' [56,57].
Pashley et al. [58] demonstrated, using a modelling
approach, that interfacial strength is dependent upon the
strength of the adhesive resin. Although the bacteria matrix
was suf®ciently porous for diffusion of etchants and resins,
large numbers of voids were observed around the bacteria
layers that were in®ltrated with the ®lled adhesive [Fig.
3(d)]. In some specimens, splitting of the bacteria cell
walls and cellular contents were observed (not shown).
368 S.M. Kwong et al. / Dental Materials 18 (2002) 359±369
These voids may act as additional ¯aws that reduce the
strength of the adhesive resin layer, apart from the weak
cohesive strength of the bacteria. Moreover, the long-term
stability of the resin±dentin bond involving degradable
bacterial matrix remains unknown. It may be bene®cial to
remove the surface layer of sclerotic dentin, as suggested by
Gwinnett and Kanca [22], to obtain more uniform etching
and resin in®ltration into intertubular sclerotic dentin, as
well as reducing the uncertainty that arises from bonding
to the bacterial matrix.
Within the limits of our study, the results led to the rejec-
tion of the null hypothesis. Although the effect of tubular
orientation has no signi®cant in¯uence in both types of
bonding substrate, the fact that a higher mean bond strength
was observed along the gingival wall of total-etched natural
lesions merits further discussion. This ®nding suggests that
the inability of the adhesive to form resin tags in dentinal
tubules that are blocked by mineral deposits is an important
parameter that leads to the reduction in bond strength [59].
If the above hypothesis is correct, then grinding of the
surface hypermineralized layer alone may not result in an
overall increase in bond strength, as the underlying dentinal
tubules still contain dead tracts that are blocked by whit-
lockite crystallites. This hypothesis was recently con®rmed
in a clinical trial on bonding to cervical sclerotic dentin [47].
Conversely, the adjunctive use of 35±40% phosphoric acid
to ground sclerotic dentin may result in partial dissolution of
the sclerotic casts and/or complete removal of the surround-
ing peritubular dentin, allowing more optimal resin in®ltra-
tion into the dentinal tubules. Although the inclusion of
conventional mechanical undercuts (e.g. with a small
round bur) may be a viable procedure, further research is
required to examine the effect of air abrasion as a means of
providing additional micromechanical retention when bond-
ing to sclerotic dentin using total-etch systems, self-etching
primers or the timely all-in-one adhesives.
Acknowledgements
We thank W.S. Lee of the Electron Microscopy Unit,
The University of Hong Kong for SEM assistance, and
Shadow Yeung of Oral Biosciences, The University of
Hong Kong for statistical support. This study was
supported by CRC grant 1020 3278 24993 08004 323
01 from the University of Hong Kong, and by grant DE
06427 from the National Institute of Dental and Cranio-
facial Research, USA. The adhesives and composites used
were generously supplied by Kuraray Co. Ltd. The
authors are grateful to Michelle Burnside for secretarial
support. The contents of this study are based on the
dissertation submitted by S.M. Kwong to the Faculty of
Dentistry, The University of Hong Kong for partial ful®l-
ment of the degree of Master of Dental Science. S.M.
Kwong graciously acknowledges the moral support of
Dr H.K.Yip.
References
[1] Swift EJ, PerdigaÄo J, Heymann HO. Bonding to enamel and dentin: a
brief history and state of the art, 1995. Quintessence International
1995;26:95±110.
[2] Weber DF. Human dentine sclerosis: a microradiographic survey.
Archives of Oral Biology 1974;19:163±9.
[3] Levitch LC, Bader JD, Shugars DA, Heymann HO. Non-carious
cervical lesions. Journal of Dentistry 1994;22:195±207.
[4] Gwinnett AJ, Jendresen M. Micromorphological features of cervical
erosion after acid conditioning and its relation to composite resin.
Journal of Dental Research 1978;57:543±9.
[5] Yoshiyama M, Masada J, Uchida A, Ishida H. Scanning electron
microscopic characterization of sensitive vs insensitive human radi-
cular dentin. Journal of Dental Research 1989;68:1498±502.
[6] Yoshiyama M, Noiri Y, Ozaki K, Uchida A, Ishikawa Y, Ishida H.
Transmission electron microscopic characterization of hypersensitive
human radicular dentin. Journal of Dental Research 1990;69:1293±7.
[7] Yoshiyama M, Sano H, Ebisu S, Tagami J, Ciucchi B, Carvalho RM,
Johnston MH, Pashley DH. Regional strength of bonding agents to
cervical sclerotic root dentin. Journal of Dental Research 1996;
75:1404±13.
[8] Harnirattisai C, Inokoshi S, Shimada Y, Hosoda H. Adhesive inter-
face between resin and etched dentin of cervical erosion/abrasion
lesions. Operative Dentistry 1993;18:138±43.
[9] Van Meerbeek B, Braem M, Lambrechts P, Vanherle G. Morpho-
logical characterization of the interface between resin and sclerotic
dentine. Journal of Dentistry 1994;22:141±6.
[10] Tay FR, Kwong SM, Itthagarun A, King NM, Yip HK, Moulding KM,
Pashley DH. Bonding of a self-etching primer to noncarious cervical
sclerotic dentin: interfacial ultrastructure and micro-tensile bond
strength evaluation. Journal of Adhesive Dentistry 2000;1:9±28.
[11] Duke ES, Lindemuth J. Variability of clinical dentin substrates.
American Journal of Dentistry 1991;4:241±6.
[12] Mixson JM, Spencer P, Moore DL, Chappell RP, Adams S. Surface
morphology and chemical characterization of abrasion/erosion
lesions. American Journal of Dentistry 1995;8:5±9.
[13] Spranger H. Investigation into the genesis of angular lesions at the
cervical region of teeth. Quintessence International 1995;26:149±54.
[14] Watanabe I, Nikaido T, Nakabayashi N. Effect of adhesion promoting
monomers on adhesion to ground dentin. Journal of the Japanese
Society of Dental Materials and Devices 1990;9:888±93.
[15] Fusayama T, Nakamura M, Kurosaki N, Iwaku M. Non-pressure
adhesion of a new adhesive restorative resin. Journal of Dental
Research 1979;58:1364±70.
[16] Kiyomura M. Bonding strength to bovine dentin with 4-META/
MMA-TBB resin: long-term stability and in¯uence of water. Dental
Materials Journal 1987;6:860±72.
[17] Sano H, Takatsu T, Ciucchi B, Horner JA, Matthews WG, Pashley
DH. Nanoleakage: leakage within the hybrid layer. Operative Dentis-
try 1995;20:18±25.
[18] Watanabe I, Nakabayashi N, Pashley DH. Bonding to ground dentin
by a Phenyl-P self-etching primer. Journal of Dental Research
1994;73:1212±20.
[19] Yoshiyama M, Carvalho RM, Sano H, Horner JA, Brewer PD, Pash-
ley DH. Regional bond strengths of resins to human root dentine.
Journal of Dentistry 1996;24:435±42.
[20] Prati C, Chersoni S, Mongiorgi R, Pashley DH. Resin±in®ltrated
dentin layer formation of new bonding systems. Operative Dentistry
1998;23:85±194.
[21] Yoshiyama M, Matsuo T, Ebisu S, Pashley DH. Regional bond
strength of self-etching/self-priming adhesive systems. Journal of
Dentistry 1998;26:609±16.
[22] Gwinnett AJ, Kanca J. Interfacial morphology of resin composite and
shiny erosion lesions. American Journal of Dentistry 1992;5:315±7.
[23] Van Meerbeek B, Peumans M, Gladys S, Braem M, Lambrechts P,
 S.M. Kwong et al. / Dental Materials 18 (2002) 359±369
Vanherle G. Three-year clinical effectiveness of four total-etch dent-
inal adhesive systems in cervical lesions. Quintessence International
1996;27:775±84.
[24] Igarashi K, Toida T, Nakabayashi N. Effect of Phenyl P/HEMA
primer on bonding to demineralized dentin by phosphoric acid.
Japanese Journal of Dental Materials 1997;16:55±60.
[25] Tantbirojn D, Cheng YS, Versluis A, Hodges JS, Douglas WH.
Nominal shear or fracture mechanics in the assessment of composite-
dentin adhesion? Journal of Dental Research 2000;79:41±8.
[26] Sano H, Shono T, Sonoda H, Takatsu T, Ciucchi B, Carvalho RM,
Pashley DH. Relationship between surface area for adhesion and
tensile bond strengthÐevaluation of a micro-tensile bond test. Dental
Materials 1994;10:236±40.
[27] Pashley DH, Carvalho RM, Sano H, Nakajima M, Yoshiyama M,
Shono Y, Fernandes CA, Tay F. The micro-tensile bond test: a review.
Journal of Adhesive Dentistry 1999;1:299±309.
[28] Sudsangiam S, van Noort R. Do dentin bond strength tests serve a
useful purpose? Journal of Adhesive Dentistry 1999;1:57±67.
[29] Kimochi T, Yoshiyama M, Urayama A, Matsuo T. Adhesion of a new
commercial self-etching/self-priming bonding resin to human caries-
infected dentin. Dental Materials Journal 1999;18:437±43.
[30] Nakajima M, Sano H, Zheng L, Tagami J, Pashley DH. Effect of moist
vs. dry bonding to normal vs. caries-affected dentin with Scotchbond
Multi-Purpose Plus. Journal of Dental Research 1999;78:1298±303.
[31] Yoshiyama M, Carvalho R, Sano H, Horner J, Brewer PD, Pashley
DH. Interfacial morphology and strength of bonds made to super®cial
versus deep dentin. American Journal of Dentistry 1995;8:297±302.
[32] Sano H, Yoshikawa T, Pereira PNR, Morigami M, Tagami J, Pashley
DH. Long-term durability of dentin bonds made with a self-etching
primer, in vivo. Journal of Dental Research 1999;78:906±11.
[33] Tanumiharja M, Burrow MF, Tyas MJ. Microtensile bond strengths of
glass ionomer (polyalkenoate) cements to dentine using four condi-
tioners. Journal of Dentistry 2000;28:361±6.
[34] Pereira PNR, Sano H, Ogata M, Zheng L, Nakajima M, Tagami J,
Pashley DH. Effect of region and dentine perfusion on bond strengths
of resin-modi®ed glass ionomer cements. Journal of Dentistry
2000;28:347±54.
[35] Shono Y, Ogawa T, Terashita M, Carvalho RM, Pashley DH. Regional
measurement of resin±dentin bonding as an array. Journal of Dental
Research 1999;78:699±705.
[36] Galloway SE, Pashley DH. Rate of removal of root structure by the
use of the Prophy-Jet device. Journal of Periodontology 1987;58:
464±9.
[37] Thompson RE, Way DC. Enamel loss due to prophylaxis and multiple
bonding/debonding of orthodontic attachments. American Journal of
Orthodontics 1981;79:282±95.
[38] Kwong SM, Tay FR, Yip HK, Kei LH, Pashley DH. An ultrastructural
study of the application of dentine adhesives to acid-conditioned
sclerotic dentine. Journal of Dentistry 2000;7:515±28.
[39] Nakaoki Y, Nikaido T, Pereira PN, Inokoshi S, Tagami J. Dimen-
sional changes of demineralized dentin treated with HEMA primers.
Dental Materials 2000;16:441±6.
[40] Walker MP, Wang Y, Swafford J, Evans A, Spencer P. In¯uence of
additional acid etch treatment on resin cement dentin in®ltration.
Journal of Prosthodontics 2000;9:77±81.
[41] Finger WJ, Inoue M, Asmussen E. Effect of wettability of adhesive
369
resins on bonding to dentin. American Journal of Dentistry 1994;
7:35±8.
[42] Pashley DH, Sano H, Ciucchi B, Yoshiyama M, Carvalho RM. Adhe-
sion testing of dentin bonding agents: a review. Dental Materials
1995;11:117±25.
[43] Grif®th AA. The phenomena of rupture and ¯ow in solids. Philoso-
phical Transactions of the Royal Society of London (Series A)
1920;A221:168±98.
[44] Van Meerbeek B, Peumans M, Gladys S, Braem M, Lambrechts P,
Vanherle G. Three-year clinical effectiveness of four total-etch dent-
inal adhesive systems in cervical lesions. Quintessence International
1996;27:775±84.
[45] Brunton PA, Cowan AJ, Wilson MA, Wilson NH. A three-year
evaluation of restorations placed with a smear-layer-mediated dentin
bonding agent in non-carious cervical lesions. Journal of Adhesive
Dentistry 1999;1:333±41.
[46] Neo J, Chew CL. Direct tooth-colored materials for noncarious lesions:
a 3-year clinical report. Quintessence International 1996;27:183±8.
[47] van Dijken JW. Clinical evaluation of three adhesive systems in class
V non-carious lesions. Dental Materials 2000;16:285±91.
[48] Gwinnett AJ. Quantitative contribution of resin in®ltration/hybridiza-
tion to dentin bonding. American Journal of Dentistry 1993;6:7±9.
[49] FreÂchette VD. Failure analysis of brittle materials. Advances in cera-
mics, vol. 28. Westerville, OH: The American Ceramic Society Inc,
1990. p. 7±42.
[50] Tay FR, Carvalho R, Sano H, Pashley DH. Effect of smear layers on
the bonding of a self-etching primer to dentin. Journal of Adhesive
Dentistry 2000;2:99±116.
[51] Tay FR, Sano H, Carvalho R, Pashley EL, Pashley DH. An ultrastruc-
tural study of the in¯uence of acidity of self-etching primers and
smear layers thickness on bonding to intact dentin. Journal of Adhe-
sive Dentistry 2000;2:83±98.
[52] Gwinnett AJ, GarciaÂ-Godoy F. Effect of etching time and acid
concentration on resin shear bond strength to primary tooth enamel.
American Journal of Dentistry 1992;5:237±9.
[53] Eick JD, Robinson SJ, Chappell RP, Spencer P. The dentinal surface:
its in¯uence on dentinal adhesionÐPart III. Quintessence Interna-
tional 1993;24:571±82.
[54] Uno S, Finger WJ. Function of the hybrid zone as a stress-absorbing
layer in resin±dentin bonding. Quintessence International 1995;
26:733±8.
[55] Van Meerbeek B, Willems G, Celis JP, Roos JR, Braem M,
Lambrechts P, Vanherle G. Assessment by nano-indentation of the
hardness and elasticity of the resin±dentin bonding area. Journal of
Dental Research 1993;72:1434±42.
[56] Kemp-Scholte CM, Davidson CL. Complete marginal seal of Class V
resin composite restorations effected by increased ¯exibility. Journal
of Dental Research 1990;69:1240±3.
[57] Lee SY, Chiang HC, Huang HM, Shih YH, Chen HC, Dong DR, Lin
CT. Thermo-debonding mechanisms in dentin bonding systems using
®nite element analysis. Biomaterials 2001;22:113±23.
[58] Pashley DH, Sano H, Ciucchi B, Carvalho RM, Russell CM. Bond
strength versus dentin structures: a modelling approach. Archives of
Oral Biology 1995;40:1109±18.
[59] Marshall GW, Chang YJ, Saeki K, Gansky SA, Marshall SJ. Citric
acid etching of cervical sclerotic dentin lesions: an AFM study.
Journal of Biomedical Material Research 2000;49:338±44.