dental

materials
Dental Materials 17 (2001) 456470
www.elsevier.com/locate/dental

Bonding of contemporary glass ionomer cements to dentin

H.K. Yip a,*, F.R. Tay a, H.C. Ngo b, R.J. Smales b, D.H. Pashley c
a
Faculty of Dentistry, The University of Hong Kong, Hong Kong, SAR, China
b
Dental School, Adelaide University, Adelaide, Australia, 5005
c
Department of Oral Biology and Maxillofacial Pathology, School of Dentistry, Medical College of Georgia, Augusta, GA 30912-1129, USA
Received 1 February 2000; revised 1 December 2000; accepted 25 January 2001

Abstract
Objective: The objective of this study was to investigate the microtensile bond strength (mTBS) of contemporary glass ionomer cements
(GIC) to sound coronal dentin.
Methods: Three specimen teeth were prepared for each material tested: Fuji IX GP (GC), ChemFlex (Dentsply) and Ketac-Molar Aplicap
(ESPE). GIC buildups were made according to the manufacturers' instructions. After being stored at 378C, 100% humidity for 24 h, the teeth
were vertically sectioned into 1 1 mm beams for mTBS evaluation. Representative fractured beams were prepared for scanning (SEM) and
transmission electron microscopic (TEM) examination.
Results: Results of the mTBS test were: Fuji IX GP (12.4 ^ 8.6 MPa), ChemFlex (15.0 ^ 9.3 MPa) and Ketac-Molar Aplicap
(11.4 ^ 7.7 MPa). One-way ANOVA and a multiple comparison test showed that ChemFlex had a statistically higher mTBS (p , 0.05).
SEM fractographic analysis showed that the predominant failure modes were interfacial and mixed failures. The GIC side of the fractured
beams revealed dehydration cracks, a high level of porosity, and voids with an eggshell-like crust. TEM analysis of the demineralized dentin
sides of the fractured beams revealed the presence of an intermediate layer along the GICdentin interface. This zone was present on the
fractured dentin surface in the case of interfacial failure, and beneath GIC remnants in specimens that exhibited a mixed failure mode.
Signicance: The ndings suggest that the bonding of GIC to dentin is not weak and that the mTBS values probably represent the weak
yield strengths of GICs under tension. q 2001 Academy of Dental Materials. Published by Elsevier Science Ltd. All rights reserved.
Keywords: Microtensile bond strength; Glass ionomer cement; Ultrastructure; Surface interaction zone

1. Introduction and trials [1214]. Frankenberger et al. [12] reported a two-

Since their introduction in 1972, glass ionomer cements
(GICs) have been widely used as dental restorative materials,
luting cements and base materials [13]. In particular, they
have been successfully used for the restoration of cervical
lesions [4,5]. Their main advantages are relative ease of use,
chemical bonding to tooth substrate, long-term uoride ion
release, low coefcient of thermal expansion and acceptable
esthetic quality [6]. Among the disadvantages are sensitivity to
moisture and desiccation during the initial setting stages, and
relatively poor physical properties [7]. Moreover, despite the
reported short- and long-term uoride release and uptake of
GICs [8,9], recent analyses of clinical studies failed to eluci-
date a direct relation between this uoride releasing potential
and their caries-inhibition effects [10,11].
The clinical performance of recently developed highly-
viscous GICs has been reported in only a few clinical reports
* Corresponding author. Tel.: 1852-2859-0286; fax: 1852-2559-9013.
E-mail address: hkyip@hkusua.hku.hk (H.K. Yip).
0109-5641/01/$20.00 + 0.00 q 2001 Academy of Dental Materials. Published by Elsevier Science Ltd. All rights reserved.
PII: S 0109-564 1(01)00007-0
year success rate of 72% for Hi-Dense (Shofu, Kyoto,
Japan), with more fractures observed in Class II than
Class I lesions. This result was due to the low exural
strength of the resin-free GIC compared to resin-modied
GICs and compomers [15]. The development of high
strength GICs allowed these materials to be used in poster-
ior teeth for the atraumatic restorative treatment (ART)
approach [16]. Survival rates for single-surface restorations
were satisfactory after three years [17,18].
GICs bond chemically to tooth substrates, and bond
strengths have been studied extensively in sound and carious
dentin using conventional shear testing methods [19,20]. The
bond strengths of GIC have been reported in the range of 3
4 MPa [21,22]. Recently, the microtensile bond strength
(mTBS) technique [2326] was developed for measuring
resin-tooth bond strength in several regions within the same
tooth. The use of a dumbbell-shaped design enables the high-
est stress to be focused on the bonded interface [25]. However,
this technique was not suitable for evaluation of materials of
relatively low bond strength to dentin (ca. 57 MPa) as
 H.K. Yip et al. / Dental Materials 17 (2001) 456470 457

premature bond failure often occurred during `free-hand' bur-
trimming of the specimens [27]. A recent `non-trimming'
version of the microtensile bond test replaced the use of dumb-
bell-shaped test specimens with beams of uniform cross-
sectional area, that were sectioned from the bonded restora-
tions [28,29]. The technique was very conservative in terms of
the number of teeth required for testing bond strength to
dentin, as up to 30 beams could be produced from one sound
molar tooth [27]. Preparation of beams of reduced cross-
sectional areas also enabled the mapping of regional bond
strength of a restorative material along the entire surface of
coronal dentin [29]. The `non-trimming' version of the micro-
tensile bond test should be useful for measuring bond strength
of contemporary high viscosity GICs, as this would reduce
premature bond failure during specimen preparation and
also allow stress to be directed along the interface, resulting
in less substrate failures within the GIC.
The objectives of this study were to evaluate the regional
tensile bond strengths of three restorative GICs to coronal
dentin and to examine the fractured GICdentin interfaces
using scanning (SEM) and transmission electron micro-
scopy (TEM). The hypothesis to be tested was that failure
of GICs under tensile stress does not reect their true adhe-
sive strength to dentin.
2. Materials and methods
2.1. Tooth preparation
Recently extracted human third molars were collected
from the Discipline of Oral and Maxillofacial Surgery and
stored in phosphate buffered saline at 48C. Only teeth that
were free of caries and cracks when examined under a
stereoscopical microscope (Nikon SMZ10, Tokyo, Japan)
at 10 magnication were used. A shallow bur mark was
made along the root of each tooth to designate the location
of the mesiobuccal line angle. The specimen teeth were
prepared by rst removing the occlusal enamel using a
slow-speed saw with a diamond-impregnated disk (Isomet,
Buehler Ltd., Lake Bluff, IL, USA) under water lubrication.
The dentin surface was further examined to be free of
enamel. A 180 grit silicon carbide paper was used under
running water to create a smear layer on the dentin surface.
2.2. GIC bonding and specimen preparation
Three contemporary conventional, highly-viscous
restorative GICs were evaluated (Table 1). All products
were used according to manufacturers' instructions. Three
prepared specimen teeth were randomly assigned to each
GIC group comprising Fuji IX GP, ChemFlex and Ketac-
Molar Aplicap. Fuji IX GP and Ketac-Molar capsules were
mixed with a triturator (Silamat-mix Type S3, Vivadent,
Schaan, Liechtenstein) for 10 s each, while ChemFlex was
prepared by hand-mixing.
For Fuji IX GP, the dentin surface was conditioned with a
polyacrylic acid (Cavity Conditioner; pH 1.65) for 10 s.
For the ChemFlex group, the dentin surface was conditioned
with a polyacrylic acid (pH 2.34) for 10 s. As recom-
mended by the manufacturer, the same liquid that was used
for mixing the cement was employed as the dentin condi-
tioner. For the Ketac-Molar group, a polyacrylic acid condi-
tioner (Ketac-Conditioner; pH 2.90) was applied to the
dentin surface for 25 s. Following rinsing of the conditioner

Table 1
Glass ionomer cements investigated in this study

Group Manufacturer Batch no. Conditioner Composition

Fuji IX GP GC International, 091085 GC Condition (10% Powder: alumino-uorosilicate glass

Tokyo, Japan polyacrylic acid) for 10 s
Polyacrylic acid powder
Pigments
Liquid: water
Polyacrylic acid
Polybasic carboxylic acid
ChemFlex Dentsply/DeTre, 9810000648 GIC liquid (10% Powder: strontium alumino-
Konstanz, Germany polyacrylic acid) for 10 s uorosilicate glass
Polyacrylic acid
Tartaric acid
Iron oxide and titanium oxide
pigments
Liquid: water
Polyacrylic acid
Ketac-Molar Aplicap ESPE GmbH, F0049611 Ketac-Conditioner (25% Powder: calcium aluminium-
Seefeld, Germany polyacrylic acid) for 25 s lanthanum-uorosilicate glass
Acrylic acid-maleic acid copolymer
pigments
Liquid: water
Acrylic acid-maleic acid copolymer
Tartaric acid
458 H.K. Yip et al. / Dental Materials 17 (2001) 456470
with distilled water for 10 s, each tooth was blot-dried and left
inverted on a piece of wet, lint-free tissue until ready for the
placement of the GIC.
A matrix band was secured around each specimen tooth
and the encapsulated GICs were syringed circumferentially
from the periphery to the center until the whole dentin
surface was covered with a thin layer. Additional incre-
ments of GICs were placed and condensed to form a core
5 mm in height. For the ChemFlex group, each scoop of
powder was mixed with one drop of liquid. The GIC was
dispensed through a Centrix syringe equipped with a
RibbonTube (Centrix Corp., Shelton, CT, USA), for build-
ing up the 5 mm core. Following the setting of the GICs,
Fuji IX GP and ChemFlex specimens were coated with GC
Fuji Varnish, while the Ketac-Molar specimens were coated
with Ketac-Glaze to protect the set material from the loss or
gain of moisture. All nine restored specimens were then
stored at 378C, 100% humidity for 24 h before sectioning.
2.3. Microtensile bond testing
The specimens were individually secured with sticky wax
to a Plexiglass sectioning block. Each tooth was aligned
with the mesiobuccal line angle to the lower right of the
sectioning block. Using the `non-trimming' technique [29],
beams with a cross-sectional area of approximately
0.81 mm 2 were prepared, with the GIC comprising the
upper half of the beam and dentin comprising the lower
half. Each tooth was sectioned with the Isomet saw through
the GIC buildups and dentin at 0.9 mm increments, to
produce a series of 0.9 mm-thick slabs. Individual slabs
from the same tooth were identied and then further
sectioned occluso-gingivally into 0.9 mm 0.9 mm
beams. Each tooth yielded between 19 and 24 beams for
bond testing. The use of three teeth resulted in a total of 64
65 beams for each of the GICs evaluated. Beams with
premature bond failure were assigned a null bond strength
Table 2
Microtensile bond strength (mTBS) of the glass ionomer cements investi-
gated
Group mTBS (MPa) a mTBS (MPa) a Area (mm 2) a
Individual teeth Group Group
Fuji IX GP 14.9 (9.3) [23] 12.4 (8.6) [64] a,b 0.81 (0.04) c
13.3 (7.5) [21]
8.5 (8.0) [20]
ChemFlex 21.1 (7.5) [19] 15.0 (9.3) [65] a 0.84 (0.04) c
10.4 (9.4) [23]
14.5 (7.9) [23]
Ketac-Molar Aplicap 11.3 (6.7) [21] 11.4 (7.7) [65] b 0.83 (0.04) c
11.4 (10.6) [24]
11.4 (3.8) [20]
a strength, that was not signicantly different when compared
Values are means (SD). Numbers in square brackets are the number of
specimens tested. Groups identied with the same superscripts are not
signicantly different (p . 0.05). The high SDs are due to inclusion of
zero bond strengths for 1719% of the bonds that failed during handling
prior to testing.
value and were included in the compilation of the mean
tensile bond strength as well as the failure mode assessment.
Specimens were stressed to failure under tension using a
universal testing machine (Model 4440; Instron Inc.,
Canton, MA) at a crosshead speed of 1 mm per min, accord-
ing to the technique reported by Shono et al. [29].
Bond strength data obtained for the three groups of GICs
were analyzed with a one-way ANOVA design, using a statis-
tical package (SigmaStat Version 2.03, SPSS, Chicago, IL,
USA). Statistical signicance was set in advance at the 0.05
probability level. Multiple comparisons were done using the
StudentNewmanKeuls test at a 0.05.
The failure modes of the bonds were initially evaluated at
30 with a stereoscopic microscope. Failures were classi-
ed as interfacial failure between dentin and the GIC, cohe-
sive failure within the GIC, or mixed (combinations of
cohesive failure in the GIC and interfacial failure along
the dentin surface).
2.4. SEM and TEM preparation
The dentin and GIC sides of four fractured beams from
each group were dehydrated, coated with gold/palladium,
and examined using a scanning electron microscope
(Cambridge Stereoscan 440, Cambridge, UK) operating at
1020 kV. Four additional fractured beams from each
group that were initially classied as adhesive failures
were prepared for TEM examination. The dentin sides of
the fractured beams were rst covered with a layer of adhe-
sive resin (D/E bonding resin, Bisco Inc.) to protect any GIC
that were still remaining. They were then completely demi-
neralized and processed for TEM examination according to
the method described in Tay et al. [30]. 70 nm thick sections
were prepared and double-stained with 1% aqueous phos-
photungstic acid at pH 3.2, followed by 2% uranyl acetate
for 10 min each. They were examined with a transmission
electron microscope (Philips EM208S, Eindhoven, The
Netherlands), operating at 80 kV.
3. Results
3.1. Microtensile bond strength
Microtensile bond strengths for the three GICs tested are
shown in Table 2. The mean mTBS varied from 11.4 to
15.0 MPa. A one-way ANOVA indicated a signicant
difference among the three GICs tested. The Student
NewmanKeuls multiple comparison test showed that the
hand-mixed ChemFlex had a marginally better mean bond
with encapsulated Fuji IX GP. However, it was signicantly
better than encapsulated Ketac-Molar. Failure modes as
determined by optical microscopic examination are shown
in Table 3.
 H.K. Yip et al. / Dental Materials 17 (2001) 456470 459

Table 3 failure, the exposed dentin surface could be recognized by the

Distribution of failure modes as observed with optical microscopy presence of polishing grooves. However, no exposed dentin
Group Interfacial Mixed Cohesive Beams with tubules could be observed, and numerous agglomerates were
failure failure failure in premature still attached to the dentin surface. At a higher magnication,
(%) (%) GIC (%) failure (%) a these agglomerates were covered with granular clusters (Fig.
1(b)). The surface layer to which these agglomerates were
Fuji IX GP 56.3 23.4 20.3 18.8
ChemFlex 67.7 20.0 12.3 18.5 attached had a smooth texture that differed from that of smear
Ketac-Molar 43.1 32.3 24.6 17.0 layer-covered dentin [30].
Aplicap The corresponding TEM micrograph showed that the
a agglomerates were remnant GIC material that was retained
Beams with premature failure during specimen preparation were
included in the calculation of mean bond strength as well as failure mode on the dentin surface (Fig. 1(c)). Glass particles were
assessment. evident within the remnant GIC, being enclosed within a
ne granular hydrogel matrix. The underlying dentin could
3.2. Ultrastructure of fractured GICDentin interface be recognized by the presence of banded collagen brils.
The GIC was connected to the dentin via a surface inter-
3.2.1. Fuji IX GP mediate layer [31]. At a higher magnication, this layer
A SEM micrograph of the fractured dentin side of a beam appeared electron-lucent and could be distinguished from
with a mixed failure mode is illustrated in Fig. 1(a). Within the GIC matrix by the presence of denatured collagen
the portion of the fractured beam that exhibited interfacial from the original smear layer (Fig. 1(d)). By contrast,

Fig. 1. (a) SEM micrograph of the dentin side of a fractured beam from the Fuji IX GP group that exhibited a mixed failure mode. G: region with cohesive
failure within the GIC; D: region with interfacial failure along the dentin surface, where numerous agglomerates (arrow) could be seen. Bar 30 mm. (b) A
higher magnication of the agglomerates (arrows) from Fig. 3(a) revealed that they were particulate and covered with granular clusters. The surface to which
they were attached was smooth. Bar 300 nm. (c) TEM micrograph of an apparent interfacial failure, showing that remnant GIC material was retained on the
surface of dentin (D). Glass particles (G), each with an electron-dense, beaded, necklace-like periphery (arrow), corresponded with the SEM observation in
Fig. 3(b). However, the matrix in Fig. 3(b), being a hydrogel, had collapsed due to the loss of water in SEM processing. M: GIC matrix; Z: surface intermediate
layer; E: unlled epoxy resin that was used for laboratory specimen embedding. Bar 300 nm. (d) Remnants of the smear layer, existing as denatured
collagen microbrils (arrow) were present within the surface intermediate layer (Z). Unraveling of the collagen brils (arrowhead) was evident along the
dentin surface. D: laboratory demineralized dentin; G: remnant GIC glass particle. Bar 300 nm. (e) TEM micrograph of a Fuji IX GP specimen that
demonstrated a true interfacial failure. Failure occurred between the GIC matrix and the surface intermediate layer (Z) and no glass particles could be seen. In
addition to the presence of denatured collagen microbrils, electron-dense plate-like structures (arrows) were present within the surface interaction zone. D:
laboratory demineralized dentin; E: epoxy resin. Bar 300 nm. (f) TEM micrograph of a glass particle (G) along the fractured dentin surface, revealing a
concentric layered morphology. The glass particle core was surrounded by an electron-lucent, siliceous gel layer (arrow). Highly electron-dense beads
(pointer) could be observed along the periphery of the glass particle, giving it a necklace-like appearance. They are probably artifacts caused by deminer-
alization of the specimen for TEM processing. M: GIC matrix. Bar 100 nm.
460 H.K. Yip et al. / Dental Materials 17 (2001) 456470

Fig. 1. (continued)

collagen brils from the underlying dentin were intact and
exhibited characteristic cross-banding. A gradual transition
could be seen in the matrix as some of the collagen brils
became unraveled at the surface where they extended within
the surface intermediate layer. In some specimens, interfa-
cial failure occurred between the GIC and dentin and no
glass particles could be observed. However, an intermediate
layer was present on the dentin surface shown in Fig. 1(e).
In this TEM micrograph, this layer was about 1.5 mm thick
and, apart from the denatured collagen strands, additional
 H.K. Yip et al. / Dental Materials 17 (2001) 456470 461

Fig. 1. (continued)

electron-dense, plate-like structures were seen that may
represent re-deposition of calcium and phosphate salts
during the ion diffusion process between the GIC and dentin
[32]. Glass particles that were present adjacent to the dentin
surface exhibited a characteristic concentric layered struc-
ture (Fig. 1(f)). The bulk of the glass particle consisted of an
electron-dense core. This was surrounded by a scalloped
electron-lucent zone that probably represents the siliceous
layer formed around the glass particles [33]. In addition, an
artifactual layer consisting of localized, electron-dense
462 H.K. Yip et al. / Dental Materials 17 (2001) 456470
Fig. 1. (continued)
domains was observed long the periphery of the glass
particle.
3.2.2. ChemFlex
Fig. 2(a) is a SEM micrograph of the dentin side of a
fractured beam showing an exclusive interfacial failure
along the GICdentin interface. The fractured surface
revealed in relief, the presence of two morphologically
distinct surfaces. The lower level was that of the dentin
substrate, wherein polishing grooves and dentinal tubules
could be observed. In contrast, the surface that was located
at the upper level was devoid of these features, but exhibited
sporadic attachment of isolated glass particles. TEM exam-
ination of the fractured GICdentin interface revealed that
the two morphologically distinct surfaces represented differ-
ent extents of smear layer removal by the polyacrylic acid
conditioner. In Fig. 2(b), only a thin layer (ca. 300 nm) of
denatured collagen could be seen between the remnant GIC
and banded collagen from the underlying intact dentin. The
dentinal tubule was still occluded with smear layer debris. In
Fig. 2(c), the layer of denatured collagen was about 1.5 mm
thick and was partially covered with electron-dense gran-
ules. The concentric layer morphology of the glass particles
was similar to that observed in Fuji IX GP.
3.2.3. Ketac-Molar
The ultrastructure of the fractured Ketac-Molar-dentin
interface revealed unique features that were not observed
in the other two GIC groups. Fig. 3(a) is a SEM micrograph
of a mixed failure from the dentin side of a fractured beam.
Remnants of fractured GIC with protruding glass particles
could be discerned from the fractured surface. Along the
areas that exhibited interfacial failure, patent dentinal
tubules that were surrounded by a circular cuff of peritubu-
lar dentin were observed. A higher magnication of the
junction between sites of cohesive GIC and interfacial fail-
ure (Fig. 3(b)) showed what appeared to be irregularly-
shaped glass particles surrounded by a gel-like matrix in
areas that demonstrated cohesive failure. In areas that exhib-
ited interfacial failure, a similar gel-like matrix could be
seen around protruding collagen brils. The corresponding
GIC side of the fractured beam (Fig. 3(c)) clearly shows the
existence of a surface intermediate layer that was attached to
the GIC surface. In addition, short tags of GIC that have
been pulled out of the dentinal tubules were retained on the
underside of the fractured surface intermediate layer.
TEM examination of the fractured GICdentin interface
revealed the presence of a layer of GIC matrix on the surface
of dentin (Fig. 4(a)). Electron-dense glass particles were
devoid of a concentric layered structure. They were
surrounded by electron-lucent zones that probably repre-
sented a surface siliceous hydrogel layer. The adjacent inter-
particulate matrix was grayish and could be clearly
discerned from the underlying surface interaction layer
that extended about 0.5 mm into the underlying intact
dentin. This relationship was better visualized with a TEM
micrograph of a higher magnication (Fig. 4(b)). At the
lower left corner of the micrograph, collagen brils with
cross-banding that were separated by well-dened, elec-
tron-lucent interbrillar spaces were evident. Within the
 H.K. Yip et al. / Dental Materials 17 (2001) 456470 463

surface interaction zone, the interbrillar spaces were
obscured by a faint, granular matrix that was similar to
the interparticulate matrix of the GIC. Electron-dense
plate-like depositions were also found around the periphery
of the banded collagen brils. In Fig. 4(c), extension of the
GIC as short cement tags into the dentinal tubules conrmed
the previous SEM observation. This is the only GIC of the
three that were examined, in which cement tags were
observed within the dentinal tubules.
4. Discussion
This study used the microtensile method to evaluate the
bond strength of highly-viscous GICs to sound dentin. The
original design [23] used dumbbell-shaped specimens that
allowed the tensile stress to be more uniformly directed toward
the weakest interfacial region [25,34]. However, we were
unsuccessful in reproducing these dumbbell-shaped speci-
mens from GIC-restored dentin, as even slight vibration
caused by eccentric movement of a high-speed bur resulted
in premature bond failures. Therefore, we adopted the `non-
trimming' version of the microtensile bond test that uses
beam-like specimens with a small, but uniform cross-sectional
area throughout the entire length of the beam [29]. This tech-
nique produces less stress on the bonded interface during
specimen preparation [27], so that we are able to study materi-
als that produce relatively low bond strengths. The disadvan-
tage is that cohesive failures are more liable to occur when
testing materials such as GIC that have relatively low tensile
bond strengths [35,36].
Previous studies of the bond strength of GIC to dentin
using tensile or shear tests indicated that bond strengths
above 5 MPa [20,21,3739] were seldom achieved, and
failure modes were largely cohesive within the cement
[22]. On the other hand, the tensile bond strengths of the
GICs investigated in the present study were in the range of
1215 MPa, and more interfacial (adhesive) and mixed fail-
ures were observed. This apparent increase in bond strength
should be viewed with caution. Increases in bond strength
up to three times the previously reported values [40] do not
represent a dramatic improvement in the physical properties
of contemporary highly-viscous GICs developed for the
ART approach [22,41,42]. Tensile or shear bond strength

Fig. 2. (a) SEM of an interfacial failure along the dentin side of a fractured beam from the ChemFlex group. The area containing polishing grooves represented
the dentin surface (D). Patent dentinal tubules were present (arrowheads), being surrounded by a cuff of peritubular dentin. The area devoid of tubular orices
represented the top of the surface intermediate layer (Z) between the GIC matrix and dentin. Some glass particles (pointer) were retained on this surface. Areas
in between where the surface interaction layer and dentinal tubules could simultaneously be observed (asterisk) probably represented either a thin surface
interaction zone or failure that involved splitting of this zone. Bar 2 mm. (b) TEM micrograph showing the presence of a thin layer of denatured collagen
(arrow) along the dentin surface that was in close approximation with an adjacent glass particle. This layer of denatured collagen was probably part of the thin
surface intermediate layer (Z) between the laboratory demineralized dentin (D) and the matrix of the GIC (M). T: dentinal tubule; Arrowhead: glass particles
within the GIC matrix. Bar 300 nm. (c) TEM micrograph showing the retention of a thick surface intermediate layer (Z) on the dentin side of a fractured
beam. Smear layer remnants that were not completely removed by the polyacrylic acid conditioner were trapped within this zone. The transition from cohesive
failure within the GIC to interfacial failure along the top of this zone is indicated by the arrow. D: laboratory demineralized dentin; Arrowhead: glass particles;
E: embedding epoxy resin. Bar 300 nm.
464 H.K. Yip et al. / Dental Materials 17 (2001) 456470

Fig. 2. (continued)

values are a function of stress distribution and concentration
that is characteristic of the particular strength-based test
employed [43,44]. The higher tensile bond strengths
reported in this study are the result of the use of smaller
specimen cross-sectional areas [23,29]. Moreover, the
higher incidence of interfacial failures observed in a micro-
tensile test represents a more uniform stress distribution
along the bonded assembly [45,46], that is probably caused
by the decrease in the number and size of critical aws
concurrent with specimen size reduction [27]. It is important
to emphasize that the higher bond strengths obtained in this
study at small cross-sectional areas are not improvements on
 H.K. Yip et al. / Dental Materials 17 (2001) 456470 465

Fig. 3. (a) SEM micrograph of a mixed failure that was present along the dentin side of a fractured beam from the Ketac-Molar group. Patches of GIC (G) with
retained glass particles (arrow) were present. Along the sites of interfacial failure (D), patent tubular orices (pointers) could be observed on the surface of the
exposed dentin. Bar 2 mm. (b) A higher magnication of the exposed dentin surface in Fig. 3(a). A dentinal tubule (T) could be seen in the center of the
micrograph. The dentin surface appeared to be covered by irregularly-shaped glass particles (arrows) that were surrounded by a gel-like matrix that had
collapsed over the dentin surface. A similar gel-like matrix could be seen around protruding collagen brils (pointer). Bar 200 nm. (c) SEM micrograph of
the corresponding GIC side of the fractured beam examined in Fig. 5(a). Exposed glass particles (arrows) were evident in areas that exhibited cohesive failure
within the GIC (G). The underside of the detached surface intermediate layer (Z) could be seen in regions that exhibited interfacial failure. Numerous cement
tags that have been pulled out of the dentinal tubules were still attached to this surface (pointers). Bar 10 mm.
466 H.K. Yip et al. / Dental Materials 17 (2001) 456470
Fig. 3. (continued)
the lower bond strengths reported previously using larger
cross-sectional areas [47]. Direct comparison of the present
GIC microtensile bond strength data to values obtained
using conventional bond tests may only be made to materi-
als that are tested using the same cross-sectional areas and
stressed under the same conditions [48].
Few ultrastructural studies of GICs are available in the
literature [33]. In particular, TEM examination of the inter-
action between GIC and dentin is lacking. Previous SEM
studies described the existence of an intermediate layer [49],
an interaction zone [31], an interdiffusion zone [50] or an
interface [32] between GIC and dentin. Spectroscopic
studies showed that this interphase consists of uoridated
carbonatoapatite [49] or the reciprocal diffusion of ions
between the GIC and dentin [32]. In the present study,
correlative SEM and TEM examination conrmed the
presence of an intermediate layer that is between 0.5 and
1.5 mm thick in all the three groups of GIC. This agrees with
the results presented by Ngo et al. [31], but is contrary to
those of Ferrari and Davidson [50], who reported a 6 mm
thick interdiffusion zone between Fuji IX and dentin. Our
TEM observations further showed that the surface interac-
tion zones in Fuji IX GP and ChemFlex were similar in
appearance, consisting of plate-like materials dispersed
among denatured smear layer remnants. The observations
were different in Ketac-Molar Aplicap, where the plate-like
materials were present within the interbrillar spaces of
intact, banded collagen brils. The occurrence of nanometer
sized plate-like materials within the surface intermediate
layer could represent re-precipitation of calcium and phos-
phate salts during the ion exchange process between the
GIC and dentin [32,49]. This characteristic interaction is
probably universal to all the three groups of GICs examined.
On the other hand, inclusion of either smear layer remnants
or banded collagen brils within the surface intermediate
layer may be explained by the aggressiveness of different
conditioning protocols in removing the smear layer and
demineralizing the underlying intact dentin. This is related
to the concentration of the polyacrylic acid employed as
well as the etching time that is recommended by each manu-
facturer [51].
In the Fuji IX GP and ChemFlex groups, the dentin was
conditioned with 10% polyacrylic acid for 10 s. The
presence of smear layer remnants within the surface inter-
mediate layer (Figs. 1(e) and 2(c)) suggests that the smear
layer was not completely removed by this conditioning
protocol. Chemical bonding of polyacrylic acid or poly-
acrylic acid-co-maleic acid with the residual hydroxyapatite
from the smear layer might have resulted in the retention of
these polyelectrolytes on the dentin surface instead of being
rinsed off [52]. This could have produced the gel-like, glass-
free layer (Fig. 3(b)) that facilitated subsequent chemical
exchanges between the leached ions from the setting glass
ionomer matrix and the calcium and phosphate ions from
the partially demineralized smear layer. Such a surface
intermediate layer that incorporated smear layer remnants
was often retained on the dentin surface in specimens that
exhibited interfacial or mixed failures.
In the case of Ketac-Molar Aplicap, a more aggressive
conditioning protocol was employed that consisted of
 H.K. Yip et al. / Dental Materials 17 (2001) 456470 467

treating the smear layer-covered dentin with 25% poly-
acrylic acid for 25 s. Our SEM and TEM observations
suggested that such a protocol rendered the dentinal tubules
patent and formed cement tags (Figs. 3(c) and 4(c)). More-
over, the smear layer was completely removed and the
underlying dentin was demineralized to a depth of about
0.5 mm (Fig. 6(d)). This is contrary to the results of Hosoya
and Garcia-Godoy [53], who reported the absence of cement
tags or a hybrid layer for Ketac-Molar Aplicap. On the other
hand, Hewlett et al. [21] reported a higher degree of dentinal
tubule patency with the use of 25% polyacrylic acid as a
conditioner. Rinsing of the conditioner probably resulted in
a collagen-rich zone that contained retained polyelectro-
lytes. Subsequent ion exchange between the setting GIC
and the partially demineralized collagen brils could result
in the formation of a surface intermediate layer within the
smear layer-depleted, intact dentin. There is a possibility
that interbrillar spaces may not be completely inltrated
by the polyelectrolytes, which may account for the lower
bond strength observed when 25% polyacrylic acid was
used as the conditioner in this and other studies [21]. It is
further speculated that the situation may even be worsened
when conditioned and rinsed dentin is desiccated before the
application of the GIC, as collapse of the collagen network
during air-drying [54] will further limit the diffusion of the
polyelectrolytes. The results from the present study suggest
that complete removal of the smear layer, with more aggres-
sive conditioning protocols that etch into the intact dentin,
does not enhance the dentin-glass ionomer restorative bond
strength. The observation of short cement tags that pulled
out of the dentinal tubules further suggests that they have a
limited contribution to the retention of GICs.
Controversy exists in the application of GICs to dentin
covered with thick smear layers. Peutzfeldt and Asmussen
[55] reported that 10 or 25% polyacrylic acid treatment
improves the bond of GIC to dentin with greater surface
roughness and thicker smear layers, but not for smoother
dentin surfaces with thinner smear layers. On the other
hand, recent clinical studies suggest that there is no advan-
tage in pre-conditioning with polyacrylic acid over cleaning
with pumice and water in the placement of GIC in cervical
teeth lesions [4,5]. As the surface intermediate layer

Fig. 4. (a) TEM micrograph of the dentin side of a fractured beam from the Ketac-Molar group that exhibited a cohesive failure within the GIC matrix (M). The
dentin surface was devoid of smear layer remnants. The surface intermediate layer (Z) extended 0.5 mm (arrows) into the underlying intact dentin. E: epoxy
resin; D: laboratory demineralized dentin; Arrowheads: glass particles within the retained GIC. Bar 1 mm. (b) A higher magnication of the surface
intermediate layer (Z) in Fig. 4(a). Banded collagen brils within this zone were partially obscured by a material that resembled the overlying GIC matrix (M).
Unraveling of the surface collagen brils was also evident (arrow). Numerous electron-dense, plate-like depositions (arrowheads) were also observed around
the interbrillar spaces of the surface interaction zone. In contrast, the interbrillar spaces (pointer) within the underlying laboratory demineralized dentin (D)
were devoid of these plate-like structures and appeared electron-lucent. Separation between the GIC and the surface intermediate layer (asterisk) was probably
the result of laboratory specimen preparation. Bar 200 nm. (c) The dentin conditioning regime employed in the placement of Ketac-Molar resulted in
complete removal of the smear layer and removal of the smear plugs in some dentinal tubules. Extension of the GIC glass particles (arrowhead) and matrix (M)
into the dentinal tubule resulted in the formation of a cement tag (arrow). The large clear area around the cement tag was originally occupied by peritubular
dentin matrix that was removed by laboratory demineralization and replaced with epoxy resin. Z: surface intermediate layer with obscured interbrillar spaces;
D: laboratory demineralized dentin; E: epoxy resin. Bar 1 mm.
468 H.K. Yip et al. / Dental Materials 17 (2001) 456470

Fig. 4. (continued)

observed in this study, and that reported by Ngo et al. [31],
was less than 2 mm, it is not known whether a surface inter-
mediate layer could be formed to incorporate the entire
thickness of a smear layer that is more than 2 mm thick.
Failure to do so might result in areas where the GIC is
only bonded to the supercial part of the smear layer. Fail-
ure could then occur between the deeper part of the smear
layer and dentin, resulting in the different levels of interfa-
cial failure that were observed in Figs. 2(a) and 3(a).
The use of the microtensile testing method enabled more
 H.K. Yip et al. / Dental Materials 17 (2001) 456470
interfacial failures to be observed along the GICdentin
interface than was previously possible. However, since
only one-half to two-thirds of the failure modes from speci-
mens tested in different groups were interfacial, the results
were still supportive of the tested hypothesis that failure of
GICs under tensile stress does not reect their true adhesive
strength to dentin. Ultrastructural observation of an intact
surface intermediate layer on the dentin surface, even in the
event of an interfacial or a mixed failure, further suggested
that bonding of conventional GIC to dentin is not weak, and
that the mTBS values probably represented the relatively
low yield strength under tension of the investigated GICs.
The surface intermediate layer was found to include either
smear layer remnants or intact, demineralized collagen
brils, depending on the aggressiveness of the polyacid
conditioning protocol. The present study was performed
using sound, healthy dentin. However, the association of
reduced bond strength with the more aggressive condition-
ing protocol and the ultrastructural appearance of the
surface intermediate layer within demineralized dentin
would have clinical implications for bonding GIC to carious
dentin. Further research should be aimed at studying the
morphological variation of the surface intermediate layer
in carious dentin. This is of particular relevance to the atrau-
matic restorative treatment (ART) approach, in which
carious dentin may not be completely removed before the
placement of GIC, and which may contain extensive bacter-
ial and deteriorated collagen remnants. Understanding the
ultrastructural interaction between GICs and these patholo-
gical entities may further improve the success rate of the
ART approach.
Acknowledgements
We thank Y.Y. Chui and Joyce Yau of the Oral Biology
Unit, Prince Philip Dental Hospital, and W.S. Lee of the
Electron Microscopy Unit, The University of Hong Kong,
for their technical support. This study was supported by
CRC grant 10201258 from The University of Hong Kong,
and by grant DE 06427 from the National Institute of Dental
and Craniofacial Research, USA. The materials used in this
study were generously supplied by Dentsply Asia (Hong
Kong), ESPE GmbH (Seefeld, Germany) and GC Interna-
tional Corporation (Tokyo, Japan). The authors are grateful
to Shirley Johnson for secretarial support.
References
[1] Wilson AD, Kent BE. A new translucent cement for dentistry. The
glass ionomer cement. Br Dent J 1972;132:1335.
[2] McLean JW. Glass-ionomer cements. Br Dent J 1988;164:293300.
[3] Smith DC. Development of glass-ionomer cement systems. Bioma-
terials 1998;19:46778.
[4] Tyas MJ. The effect of dentine conditioning with polyacrylic acid on
the clinical performance of glass ionomer cement3 year results.
Aust Dent J 1994;39:2207.
469
[5] van Dijken JW. Four-year evaluation of the effect of 10% polyacrylic
acid or water rinsing pretreatment on retention of glass polyalkenoate
cement. Eur J Oral Sci 1996;104:6466.
[6] Naasan MA, Watson TF. Conventional glass ionomers as posterior
restorations. A status report for the American Journal of Dentistry.
Am J Dent 1998;11:3645.
[7] Mount GJ. Glass ionomers: a review of their current status. Oper Dent
1999;24:11524.
[8] Forsten L. Fluoride release and uptake by glass-ionomers and related
materials and its clinical effect. Biomaterials 1998;19:5038.
[9] Yip HK, Lam WT, Smales RJ. Fluoride release, weight loss and
erosive wear of modern aesthetic restoratives. Br Dent J
1999;187:26570.
[10] Mjor IA. Glass-ionomer cement restorations and secondary caries: a
preliminary report. Quintessence Int 1996;27:1714.
[11] Randall RC, Wilson NH. Glass-ionomer restoratives: a systematic
review of a secondary caries treatment effect. J Dent Res
1999;78:62837.
[12] Frankenberger R, Sindel J, Kramer N. Viscous glass-ionomer
cements: a new alternative to amalgam in the primary dentition?
Quintessence Int 1997;25:66776.
[13] Ho TFT, Smales RJ, Fang TSD. A 2-year clinical study of two glass
ionomer cements used in the atraumatic restorative treatment (ART)
technique. Community Dent Oral Epidemiol 1999;27:195201.
[14] Peng D, Gao W, Smales RJ, Yip HK. Clinical evaluation of two glass
ionomer cements for ART in an hospital environment. Caries Res
1999;33:3145.
[15] Hickel R, Manhart J. Glass-ionomer and compomers in paediatric
dentistry. In: Davidson CL, Mjor IA, editors. Advances in glass-
ionomer cements, Quintessence Publishing Co., Inc, 1999.
p. 20126.
[16] Frencken J, Makoni F. A treatment technique for tooth decay in
deprived communities. World Health 1994;47:1517.
[17] Frencken JE, Makoni F, Sithole WD. ART restorations and glass
ionomer sealants in Zimbabwe: Survival after 3 years. Community
Dent Oral Epidemiol 1998;26:37281.
[18] Holmgren CJ, Lo ECM, Hu DY, Wan HC. ART restorations and
sealants placed in Chinese school childrenresults after three
years. Community Dent Oral Epidemiol 2000;28:31420.
[19] McInnes-Ledoux PM, Weinberg R, Grogono A. Bonding glass-iono-
mer cements to chemomechanically-prepared dentin. Dent Mater
1989;5:18993.
[20] Burke F, Lynch E. Chemomechanical caries removal. The effect of
chemomechanical caries removal on the bond strength of glass poly-
alkenoate cement to dentine. J Dent 1994;22:28391.
[21] Hewlett ER, Caputo AA, Wrobet DC. Glass ionomer bond strength
and treatment of dentin with polyacrylic acid. J Prosthet Dent
1991;66:76772.
[22] Ewoldsen N, Covey D, Lavin M. The physical and adhesive proper-
ties of dental cements used for atraumatic restorative treatment. Spec
Care Dent 1997;17:1924.
[23] Sano H, Shano T, Sonoda H, Takatsu T, Ciucchi B, Carvalho R,
Pashley DH. Relationship between surface area for adhesion and
tensile bond strengthevaluation of a microtensile bond test. Dent
Mater 1994;10:23640.
[24] Nakajima M, San H, Burrow MF, Tagami J, Yoshiyama M, Ebisu S,
Ciucchi B, Russell CM, Pashley DH. Tensile bond strength and SEM
evaluation of caries-affected dentin using dentin adhesive. J Dent Res
1995;74:167988.
[25] Pashley DH, Sano H, Ciucchi B, Yoshiyama M, Carvalho RM. Adhe-
sion testing of dentin bonding agents: a review. Dent Mater
1995;11:11725.
[26] Yoshiyama M, Shano H, Ebisn S, Tagami J, Ciucchi B, Carvalho RM,
Johnson MH, Pashley DH. Regional strengths of bonding agents to
cervical sclerotic root dentin. J Dent Res 1996;75:140413.
[27] Pashley DH, Carvalho RM, Sano H, Nakajima M, Yoshiyama M,
470 H.K. Yip et al. / Dental Materials 17 (2001) 456470
Shono Y, Fernandes CA, Tay F. The micro-tensile bond test: a review.
J Adhes Dent 1999;1:299309.
[28] Cardoso PEC, Braga RR, Carrilho MRO. Evaluation of micro-tensile,
shear and tensile tests determining the bond strength of three adhesive
systems. Dent Mater 1998;14:3948.
[29] Shono Y, Ogawa T, Terashita M, Carvalho RM, Pashley EL. Regional
measurement of resin-dentin bonding as an array. J Dent Res
1999;78:699705.
[30] Tay FR, Carvalho R, Sano H, Pashley DH. Effect of smear layers on
the bonding of a self-etching primer to dentin. J Adhes Dent
2000;2:99116.
[31] Ngo H, Mount GJ, Peters MC. A study of glass-ionomer cement and
its interface with enamel and dentin using a low-temperature, high-
resolution scanning electron microscopic technique. Quintessence Int
1997;28:6369.
[32] Sennou HE, Lebugle AA, Gregoire GL. X-ray photoelectron spectro-
scopy study of the dentin-glass ionomer cement interface. Dent Mater
1999;15:22937.
[33] Hatton PV, Brook JM. Characterisation of the ultrastructure of glass-
ionomer (poly-alkenoate) cement. Br Dent J 1992;173:2757.
[34] Nakabayashi N, Watanabe A, Arao T. A tensile test to facilitate
identication of defects in resin-bonded dentin specimens. J Dent
1998;26:37985.
[35] Kerby RE, Knobloch L. Strength characteristics of glass-ionomer
cements. Oper Dent 1992;17:1704.
[36] Galun EA, Saleh N, Lewinstein I. Diametral tensile strength and
bonding to dentin of type I glass ionomer cements. J Prosthet Dent
1994;72:4249.
[37] Smith DC, Ruse DN, Zuccolin D. Some characteristics of glass iono-
mer cement lining materials. J Can Dent Assoc 1988;54:9038.
[38] Cattani-Lorente MA, Godin C, Meyer JM. Early strength of glass
ionomer cements. Dent Mater 1993;9:5762.
[39] Berry EA, Powers JM. Bond strength of glass ionomers to coronal and
radicular dentin. Oper Dent 1994;19:1226.
[40] van Noort R, Noroozi S, Howard IC, Cardew G. A critique of bond
strength measurements. J Dent 1989;17:6167.
[41] Powers JM, Burgess JO. Performance standards for competitive
dental restorative materials. In: Proceedings of Conference on Clini-
cally Appropriate Alternatives to Amalgam: Biophysical Factors in
Restorative Decision-making. Munich: Academy of Dental Materials.
1996; pp 6876.
[42] Guggenberger R, May R, Stefan KP. New trends in glass-ionomer
chemistry. Biomaterials 1998;19:47983.
[43] DeHoff PH, Anusavice KJ, Wang Z. Three-dimensional nite element
analysis of the shear bond test. Dent Mater 1995;11:12631.
[44] Versluis A, Tantbirojn D, Douglas WH. Why do shear bond tests pull
out dentin? J Dent Res 1997;76:1298307.
[45] Armstrong SR, Boyer DB, Keller JC. Microtensile bond strength
testing and failure analysis of two dentin adhesives. Dent Mater
1998;14:4451.
[46] Schreiner RF, Chappell RP, Glaros AG, Eick JD. Microtensile testing
of dentin adhesives. Dent Mater 1998;14:194201.
[47] Shono Y, Terashita M, Pashley EL, Brewer PD, Pashley DH. Effects
of cross-sectional area on resin-enamel tensile bond strength. Dent
Mater 1997;13:2906.
[48] Sudsangiam S, van Noort R. Do dentin bond strength tests serve a
useful purpose? J Adhesive Dent 1999;1:5767.
[49] Geiger SB, Weiner S. Fluoridated carbonatoapatite in the intermedi-
ate layer between glass ionomer and dentin. Dent Mater 1993;9:33
36.
[50] Ferrari M, Davidson CL. Interdiffuson of a traditional glass ionomer
cement into conditioned dentin. Am J Dent 1997;10:2957.
[51] Barakat MM, Powers JM, Yamaguchi R. Parameters that affect in
vitro bonding of glass-ionomer liners to dentin. J Dent Res
1988;67:11613.
[52] Yoshida Y, Van Meerbeek B, Nakayama Y, Snauwaert J, Hellemans
L, Lambrechts P, Vanherle G, Wakasa K. Evidence of chemical
bonding at biomaterial-hard tissue interfaces. J Dent Res 2001 (in
press).
[53] Hosoya Y, Garcia-Godoy F. Bonding mechanism of Ketac-Molar
Applicap and Fuji IX GP to enamel and dentin. Am J Dent
1998;11:2359.
[54] Gwinnett AJ. Chemically conditioned dentin: a comparison of
conventional and environmental scanning electron microscopy nd-
ings. Dent Mater 1994;10:1505.
[55] Peutzfeldt A, Asmussen E. Effect of polyacrylic acid treatment of
dentin on adhesion of glass ionomer cement. Acta Odontol Scand
1990;48:33741.