Download as pdf or txt
Download as pdf or txt
You are on page 1of 11

Dental Materials (2004) 20, 554–564

http://www.intl.elsevierhealth.com/journals/dema

Residual shrinkage stress distributions


in molars after composite restoration
Antheunis Versluisa,*, Daranee Tantbirojnb, Maria R. Pintadoa,
Ralph DeLonga, William H. Douglasa

a
Department of Oral Science, Minnesota Dental Research Center for Biomaterials and Biomechanics,
University of Minnesota, 16-212 Moos Tower, 515 Delaware Street SE, Minneapolis, MN 55455, USA
b
Department of Operative Dentistry, Chulalongkorn University, Bangkok, Thailand

Received 3 December 2002; received in revised form 16 May 2003; accepted 29 May 2003

KEYWORDS Summary Objective. Experimental measurements on various restoration configur-


Restorative composite; ations have shown that restored teeth deform under the influence of polymerization
Polymerization shrinkage, but actual residual stresses could not be determined. The purpose of this
shrinkage; Strain gauge; study was to calculate and validate shrinkage stresses associated with the reported
Microhardness; Post-gel tooth deformations.
shrinkage; Composite Methods. Three different restoration configurations were applied in a finite element
restoration; Shrinkage model of a molar. The composite properties were based on experimentally determined
stress; Residual stress; composite behavior during polymerization. The occlusal deformation pattern and the
Deformation residual stress states of the tooth, restoration, and tooth-restoration interface were
calculated using a polymerization model based on the post-gel shrinkage concept.
Reported strain gauge measurements and occlusal deformation patterns were used for
validation.
Results. The shrinkage stresses depended on the configuration and size of the
restorations. The tooth’s resistance against polymerization shrinkage diminished with
loss of dental hard tissue. Larger restorations resulted in lower stress levels in the
restoration and tooth-restoration interface, but increased stresses in the tooth. The
maximum stress values found for different configurations were not decisively
different.
Significance. The validated model indicated that shrinkage stress cannot be based on
composite properties or restoration configuration alone, but has to be approached as a
distributed pattern that depends on the location and on the properties of tooth and
restoration, geometry, constraints, and restoration procedures. Tooth deformation
was indicative of stresses in the tooth rather than in the restoration or across the
tooth-restoration interface.
Q 2003 Academy of Dental Materials. Published by Elsevier Ltd. All rights reserved.

Introduction

*Corresponding author. Tel.: þ1-612-625-0950; fax: þ 1-612- It is widely accepted that polymerization shrinkage
626-1484. of current restorative composites cause residual
E-mail address: antheun@umn.edu stresses in restored teeth. These stresses are called

0109-5641/$ - see front matter Q 2003 Academy of Dental Materials. Published by Elsevier Ltd. All rights reserved.
doi:10.1016/j.dental.2003.05.007
Residual shrinkage stress distributions in molars after composite restoration 555

residual because after curing a restored tooth is process with experimental observations. Defor-
left under stress even when there is no functional mation patterns of occlusal surfaces for restored
loading. The presence of residual stresses results in molars due to shrinkage stresses have been deter-
a changed behavior of the restored tooth, which mined and observed in a series of experiments.8
may become evident in its clinical performance. While deformation could be measured experimen-
Clinical symptoms associated with residual shrink- tally, the stress distribution must still be calculated
age stresses are inadequate adaptation, microcrack using transient composite properties determined
propagation, marginal loss, post-operative sensi- from another series of experiments.9 The purpose
tivity, microleakage, and secondary caries.1 – 5 of this study was to develop and validate a
Stress is not a material property, but a local polymerization model for the calculation of
physical state that is derived from the combination residual shrinkage stress distributions associated
of material properties, geometry, boundary con- with the reported deformations, using the deter-
ditions, and history. Since residual shrinkage stress mined transient properties distributions. The vali-
is not a straightforward property that can be dated model can consequently be used to
measured directly, its quantification in restorations investigate the shrinkage stresses in restored
remains a source of controversy. Various methods teeth, and study factors that affect them.
have been used to estimate residual shrinkage
stresses, ranging from extrapolated shrinkage or
load measurements in vitro to stress analyses in Material and methods
tooth shaped anatomies using photoelastic or finite
element methods.6,7 Calculation of shrinkage stres- Deformations of occlusal surfaces have been
ses in a tooth-restoration complex is not trivial. One reported for a successive range of cavity prep-
challenge is the description of the sequence of arations in extracted human molars:8 Class I, small
changes that take place in composite during Class II OM, large Class II OM, and Class II MOD. The
polymerization. Another challenge is that stress deformation due to polymerization shrinkage was
depends on the geometry and mechanical proper- quantified by comparing the digitized occlusal
ties of surrounding tissues. surfaces before and after restoration (Fig. 1). To
Many intricacies of the biomechanical manifes- calculate corresponding residual stresses in the
tations of polymerization processes are still not tooth, a finite element simulation was carried out
well understood. As a result the development of a for a similar combination of geometry, boundary
residual shrinkage stress model (i.e. the expression conditions, material properties, and restoration
of our understanding of the event) is an interactive procedures (history).

Figure 1 Cuspal deformation patterns (mm) for a molar determined in in vitro experiments for four consecutive
restorations (Class I, small Class II OM, large Class II OM, and Class II MOD) using Cumulus software.8
556 A. Versluis et al.

Figure 2 Finite element model of molar, indicating four simulated restorations (Class I, small Class II OM, large Class II
OM, and Class II MOD), increments, and the direction of light curing (black arrows).

Four consecutive restorations were applied in a stress development. A linear elastic approach
finite element model of a molar: Class I, small Class allows a significant reduction of the stress calcu-
II OM, large Class II OM, and Class II MOD (Fig. 2). lation time, while the required properties that have
Additionally, a large Class I and small Class II MOD to be determined are reduced to the elastic
were created. The external molar shape was modulus and post-gel shrinkage. Both properties
digitized using a contact stylus profilometer,10 depend on the curing light intensity, exposure time,
while the internal enamel – dentin and pulp surfaces and storage time. In order to apply the properties in
were estimated based on general anatomy. The the finite element analysis, they have to be
molar root was fixed, simulating the embedded expressed in mathematical forms. The local light
specimens of the corresponding experimental intensity I in the restoration was approximated
design.8 using the light intensity of the curing light source,
Orthotropic elastic properties were modeled for the effect on the light intensity due to the distance
the enamel. The elastic moduli were 84 GPa between the restoration and light guide, and the
(principal direction, perpendicular to the pulp exponential relationship for light attenuation in the
surface) and 42 GPa, with a Poisson’s ratio of composite9
0.30. The dentin elastic modulus was 18 GPa, with
a Poisson’s ratio of 0.23. The composite properties I ¼ I0 exp½2ðw=10Þ3 exp½21:19d 0:72  ð1Þ
used were based on experimental data9 obtained
for Z100 Incisal shade (Lot 6BN, 3M ESPE, St Paul, where I0 is the light intensity of the curing light
MN, USA), the same as used for the companion source (mW/cm2); w is the distance from light-
study.8 Z100 is a commercially available universal guide to the restoration surface (mm); d is the
restorative composite, containing bisGMA and depth in the composite (mm). In each composite
TEGDMA resins with zirconia/silica fillers (66% element integration point (where stresses are
inorganic filler loading by volume, particle size determined by the finite element software) local
range of 3.5 – 0.01 mm). The relatively translucent properties were calculated as a function of local
composite was chosen to obtain good light pen- light intensity and time (light exposure and dark
etration in the large restorations. The maximum cure time). The reported correlation between
elastic modulus and Poisson’s ratio for a fully cured Knoop microhardness and elastic modulus11 can be
composite were 20 GPa and 0.24, respectively. employed to calculate the local elastic modulus of
In this study a linear-elastic approach was the composite as a function of light energy and
validated for the simulation of residual shrinkage time. Local light energy density Y (mJ/cm2) is
Residual shrinkage stress distributions in molars after composite restoration 557

defined as12,13 (Eqs. (6) and (7)), using the intensity output of the
curing light source rather than local intensity levels.
Y ¼ Itc ð2Þ
A bulk property means that the acquired strain gauge
where tc is the curing light exposure time (s). The data represents the response of the whole sample.
empirical relationship found for the microhardness Since post-gel shrinkage depends on the light
KMH is9 intensity and light intensity is location dependent,
the post-gel shrinkage also depends on the location.
KMH¼ð66211exp½2t=3000Þð12exp½2ðY=2500Þ1:1 Þ Therefore, the bulk value for post-gel shrinkage
ð3Þ cannot be used in the finite element analysis.
Considering that the strain gauge records the local
where t is the total time since the start of curing (s).
deformation at the gauge surface, the post-gel
The first term describes how the microhardness
shrinkage results (Eqs. (6) and (7)) can be easily
values increase during the storage time t after light
adjusted by correction of the terms that contain the
cure. The second term describes the exponential
light intensity
relationship between microhardness and light energy.
The local elastic modulus E can be obtained by I0 ¼ FI ð9Þ
E ¼Emax KMH=KMHmax ð4Þ Y0 ¼ FItc ¼ FY ð10Þ
where Emax is the maximum attainable elastic where F is the correction factor for the actual light
modulus value (Emax ¼20 GPa), and KMHmax is the intensity at the gauge surface. Using the light
maximum microhardness value used to normalize attenuation Eq. (1) the correction factor F is 6.1
KMH in Eq. (3) ðKMHmax ¼66Þ: (w ¼ 3 mm, d ¼ 1:75 mm). To validate the derived
Post-gel shrinkage values were obtained by the elastic modulus and post-gel shrinkage formulations
strain gauge method.9 Unlike microhardness, post- the strain gauge design was also simulated in a finite
gel shrinkage cannot be fully described by the element analysis. The strain gauge configuration was
applied light energy because it depends on the light modeled by a 1.75 mm high composite sample placed
intensity at which the polymerization reaction was on an 80 mm thick strain gauge (elastic modulus 6 GPa
initiated. The maximum value for the post-gel and Poisson’s ratio 0.3, manufacturer’s data) (Fig. 3).
shrinkage amax can be expressed by the empirical Light curing from above (light guide positioned 3 mm
relationship ðR . 0:999Þ above composite sample) was simulated for the four
curing light intensities used in the experimental
amax ¼ 5000ð1 2 exp½2ðI0 =500Þ0:22 Þ ð5Þ setup9 (I0 ¼ 100; 300, 500, and 700 mW/cm2). The
where I0 is the intensity of the light source with plane strain in a 1 mm2 area at the center of the strain
which the polymerization was initially initiated gauge was recorded for validation with the exper-
(mW/cm2). The polymerization process can be imental shrinkage strain results.
divided into two periods: light exposure and dark In the restoration models, light cure was applied
cure (after the light cure). Post-gel shrinkage occlusally to the composite restorations (188 off for
during ðalight Þ and after ðadark Þ light exposure can the proximal boxes) for 60 s per increment at
be expressed by ðR . 0:97Þ 650 mW/cm2, mimicking the experimental study.8
The diameter of the simulated light guide was
alight ¼ amax ð1 2 exp½2Y00:7 =ð450 þ 0:35I0 ÞÞ ð6Þ 12 mm. The Class II OM cavities were filled in two
increments, the Class II MOD in three (Fig. 2).
adark ¼ amax ð8 þ 0:06I0 Þexp½2ðY0 =1:161Þ0:186 
 ð1 2 exp½2{Y0 =ð858 þ 4:1I0 Þ}1:5 Þ ð7Þ
where I0 is the light intensity of the light source at
which the composite polymerization is initiated
(mW/cm2), and Y0 is the applied light energy ðI0 tc Þ:
Note the difference between applied and local light
intensity or energy. The applied values are the output
of the curing light source. The post-gel shrinkage
value a determined by the strain gauge becomes:
a ¼ alight þ adark ð8Þ
Unlike the microhardness data, which were deter- Figure 3 Finite element model of the strain gauge
mined as local properties (Eq. (3)), the post-gel experiment,9 showing the element distribution for the
shrinkage results were determined as bulk properties quarter mesh.
558 A. Versluis et al.

Additionally, a large Class I and a small Class II MOD generation software package (Cumulus, copyright
were modeled. The time between increments was Regents of the University of Minnesota). Surface
60 s. It is important that the successive filling changes in Cumulus are determined perpendicular to
increments are not part of the analysis before they the baseline surface taken before restoration. In
are placed. Eight-node isoparametric arbitrary order to import the finite element geometry in
hexahedral reduced integration elements were Cumulus, the before and after surfaces were
used. A custom program calculated and applied the isolated, refined and converted to object file format
material property relationships for enamel and (Alias Wavefront Object Files). From the dense
composite, performed the restorative actions of surface point-clouds, Cumulus created rendered
filling and light cure, defined and submitted the surfaces. Differences between the before and after
restoration models to a finite element solver for occlusal enamel surfaces were determined and
calculation of stresses (MSC.Marc, MSC Software visualized using a linear color scale. The quantitative
Co., Los Angeles, CA, USA), processed the results color patterns were used to compare the experimen-
and created a convenient output format. The tal and finite element results. Cusp displacements
analyses were performed on a Compaq AlphaServer (bucco-lingual direction) were determined to be
ES40 (Compaq, Houston, TX, USA). compared with values reported in the literature.
Modified von Mises equivalent stresses were used
to express the stress conditions. The advantage of
using equivalent von Mises stress is that a multi- Results
dimensional stress distribution (e.g. six Cartesian
stress components or three principal stresses) is To validate the polymerization modeling pro-
expressed into one value, which eases the interpret- cedures, deformation was calculated in two finite
ation of the overall stress condition. The used element simulations (strain gauge simulation and
equivalent stress is based on the well-known von occlusal deformation of a tooth crown), using the
Mises formulation, modified to take into account the equations for light intensity, elastic modulus, and
difference between compressive and tensile post-gel shrinkage derived from experimental data.
strength for enamel, dentin, and the composite.14 The response of the strain gauge is not solely
Since cuspal deflection is often reported for in vitro caused by post-gel shrinkage, but also by develop-
studies, the inter-cuspal distance change were ment of elastic modulus and distribution of
determined for the mesial and distal cusp pairs. properties due to the light attenuation in the
The calculated occlusal deformations were com- composite. Results (microstrain) of the experimen-
pared with the experimentally determined occlusal tal data9 were compared with the results obtained by
deformation8 (Fig. 1), using a locally developed new the finite element simulation (Table 1). The table

Table 1 Comparison of shrinkage strains measured by the strain gauge experiments9 (mean and standard deviation; n ¼ 5) and
calculated by the numerical polymerization simulation of the strain gauge setup.

Light intensity, I0 (mW/cm2) Exposure time, tc (s) Light energy, Y0 (mJ/cm2) Shrinkage strain (mstrain)

Experimental Finite element analysis

100 10 1000 451(89) 865


100 20 2000 1227(31) 1315
100 40 4000 1703(16) 1693
100 60 6000 1844(40) 1926
300 10 3000 1825(60) 1962
300 20 6000 2280(65) 2283
300 40 12,000 2560(75) 2606
300 60 18,000 2641(49) 2777
500 10 5000 2224(95) 2433
500 20 10,000 2664(69) 2665
500 40 20,000 2849(68) 2941
500 60 30,000 2907(73) 3077
700 10 7000 2602(80) 2684
700 20 14,000 2889(86) 2868
700 40 28,000 3060(44) 3122
700 60 42,000 3136(81) 3241

The results were determined 2 min after polymerization for different combinations of light intensities ðI0 Þ and light exposure times ðtc Þ:
Residual shrinkage stress distributions in molars after composite restoration 559

Figure 4 Cuspal deformation patterns (mm) for the simulated molar for four restorations (Class I, small Class II OM,
large Class II OM, and Class II MOD) using the numerical polymerization model and Cumulus software.

shows a good agreement between the experimental color scale to indicate their intensity (Fig. 5).
and finite element data. The restorations were rendered invisible to view
Applying the material relationships to a tooth the tooth-restoration interface. Note that the
model, the occlusal deformations were calculated shown interfacial stresses are extrapolated across
and visualized by color coding (Fig. 4) for compari- the tooth-restoration interface. The distributions
son with the experimental results (Fig. 1) using the show areas of high stress concentrations (yellow)
Cumulus software. Positive values are defined as and low stress areas (blue). The color patterns
higher surfaces, and negative values indicate lower illustrate that stress values were local conditions.
surfaces compared to the baseline. The defor- In general, the Class I restoration showed a larger
mation pattern and values were similar in magni- area of high stresses at the tooth-restoration
tude and distribution to the results found by the interface than the Class IIs. However, the Class II
experimental method. The finite element analysis MOD showed higher stresses in the enamel surfaces
calculated occlusal deformation lower than 10 mm than the other restorations.
for the smallest restoration (Class I), and increased The residual stresses inside the tooth, compo-
deformation with increasing restoration size, up to site, and along the tooth-composite interface were
20 mm for the Class II MOD. also plotted in distribution graphs that visualize all
The calculated inter-cuspal distance changes stresses (Fig. 6). These graphs plot the total range
(mean value of mesial and distal cusp pairs) were of internal stresses (modified von Mises equivalent
5.3, 27.1 (10.1 distal), 46.8 (16.0 distal), and stresses), ranked in ascending order, for all restor-
45.5 mm for the Class I, Class II OM (small), Class II ation variations. Note that the graphs do not
OM (large), and Class II MOD restorations, respect- contain information about local stress patterns.
ively. Note that in the OM-configurations the values The ratio between bonded and free surfaces are
for the distal cusps are given separately because also indicated in the figure. The graphs show that in
the inter-cuspal distance changes were much general the stress level in the tooth (enamel and
smaller between the distal cusps compared with dentin) is the highest for the Class II MODs, followed
the mesial cusps. The cuspal deflections for the by the Class II OMs, while the lowest overall stress
large Class I and small Class II MOD were 10.8 and level in the tooth was found for the Class I
25.5 mm, respectively. restoration (Fig. 6A). The order is reversed if
Residual stress distributions (modified von Mises general stress levels in the restoration or along
equivalent stresses) at the tooth surfaces were the tooth-restoration interface are considered
visualized for four configurations using a linear (Fig. 6B and C).
560 A. Versluis et al.

Figure 5 Residual shrinkage stress distributions (modified von Mises equivalent stresses) for the four restorations
(Class I, small Class II OM, large Class II OM, and Class II MOD) calculated by the numerical polymerization model. To show
the tooth-restoration interface, the restoration was rendered invisible.

Discussion symptoms such as microfracture, microleakage and


post-operative sensitivity. None of these symp-
The large number of publications about polymeriz- toms, however, are direct measures of shrinkage
ation shrinkage of restorative composites indicates stress. The alleged presence of shrinkage stresses is
that it is still considered a serious clinical concern. thus only known through indirect manifestations.
Shrinkage stress has been associated with clinical One of them, tooth deformation, is often not

Figure 6 Stress distribution plots representing all internal stresses (modified von Mises equivalent stresses) ranked in
ascending order in: (A) the tooth, (B) the restoration, and (C) the tooth-restoration interface. The stresses are displayed
for two restoration sizes (small and large) of three configurations (Class I, Class II OM, and Class II MOD). The ratios
between bonded and free restoration surfaces are listed.
Residual shrinkage stress distributions in molars after composite restoration 561

specified among the clinical concerns, possibly such as time. Simulation experiments do not
because it is too small to be perceived under determine single properties but rather a bulk
normal clinical conditions. It is well accepted, response that incorporates the interaction between
however, that teeth deform under loads.15 Among various basic properties, such as the deformation of
all clinical symptoms, deformation offers the a restored tooth which involves the interaction
closest relationship with shrinkage stresses. Fur- between dental tissues and restorative properties,
thermore, deformation can be quantified, unlike anatomy, fixation, and restorative sequence.
many of the other symptoms that remain substan- One of the complexities for understanding (and
tially subjective. Deformation in teeth is often thus modeling) polymerization shrinkage is knowing
quantified by measuring the flexure of cusps.16 – 20 what exactly happens to the mechanical properties
Cusp flexure indicates the deformation at particular during polymerization. The realization that not all
locations on the tooth crown. It has been shown, shrinkage contributes to shrinkage stresses is
however, that cusp flexure may not be a reliable fundamental. The concept of pre- and post-gel
indication of actual shrinkage stress levels.21 As shrinkage or time-dependent viscosity was noted by
part of this study the complete pattern of occlusal Bowen in one of the earliest publications about
tooth deformations were measured and visualized silica-reinforced polymer properties.23 Although
(Fig. 1)8 The deformation patterns registered how a this concept has thus been known in restorative
total crown deforms for different restoration dentistry since the first introduction of composites,
configurations. However, the expectation fueled its consequences are only recently gaining more
by dental literature that Class I configurations have recognition. This may be explained by the increas-
higher stresses than Class IIs seems to be reversed if ing acceptance of polymerization stress modeling
residual stresses are correlated directly with the and because of observations that polymerization
tooth deformation. kinetics affect post-gel shrinkage and bulk stresses.
Obviously, there is more to the assessment of The post-gel concept is based on a division of the
residual shrinkage stresses in restored teeth than total shrinkage into a component that does not
simply measuring deformation on a restored tooth cause stress development due to relief by viscous
(let alone extrapolation of stress conditions from flow (pre-gel), and a component that causes stress
simplified laboratory specimens). Stress is a physi- development due to elastic storage of contraction
cal condition which depends on the combination of deformation (post-gel). From a physical point of
all material properties involved, geometry, and view, polymerization is a process of simultaneous
boundary conditions. The interactions between resin densification and composite modulus change.
them obey universal mechanical laws. Unfortu- The time-dependent relationship between contrac-
nately, shrinkage stress considerations have thus tion on the one hand and the viscous (loss) and
become much more complicated because they must elastic (storage) modulus on the other hand
involve comprehensive computations. Residual determines the potential contribution of the com-
stresses can be studied using finite element posite properties for the generation of residual
analysis, which is a widely used contemporary shrinkage stresses. For the tooth model used in this
computational technique.22 Finite element analysis study, the calculation using a time-dependent
provides the framework to combine material viscoelastic model proved prohibitively time-con-
properties, with geometry and boundary conditions suming for current hardware specifications because
according to universal mechanical laws. Using of the large number of time-steps required to
computational techniques is not merely an option, manage the distributed time-dependent nonlinear
but has become inevitable to express and test our property changes. Using the post-gel concept,
current insight of dental materials behavior. This however, allowed for shrinkage stresses and tooth
contemporary modeling approach also has conse- deformations to be approached in a linear elastic
quences for the design of experimental studies. analysis. A linear elastic analysis requires consider-
Supporting experiments can be divided into two ably less time-steps, and thus a significantly
categories (as demonstrated by the design of this reduced solver time. The validity of the linear
study): experimental determination of basic prop- elastic approach was supported by the good agree-
erties9 and experimental simulation for model ment between numerical and experimental results.
validation.8 Basic property experiments are used The shrinkage stress model was experimentally
to obtain material behavior relationships required validated by the results of the strain gauge and the
in a computational model, such as elastic modulus restored molar. At first sight, the strain gauge
and post-gel shrinkage in the current study. They experiments9 may not seem an independent vali-
often have to be determined as theoretical or dation, since they also provided the data on which
empirical functions of their controlling factors, the used post-gel shrinkage relationship was based.
562 A. Versluis et al.

However, the shrinkage model also contained light polymerization action). Considering that each of
attenuation and elastic modulus equations. These these conditions affected the final stress and
were derived from different experiments, and were deformation pattern, it is unrealistic to expect an
combined with the post-gel shrinkage in the exact fit between the finite element model and the
numerical polymerization model. The very close experiments. For example, although in both cases a
results of the measured and calculated strains not mandibular molar was used, their shapes were not
only validated the numerical polymerization model, exactly identical. Note that this is not a finite
they also indicated that the experimental strain element versus ‘reality’ issue, because the five
gauge data actually represented local strain values molars investigated in the experiments were not
at the gauge surface (Table 1). This is an important exactly identical either. Other factors that could
observation because it supports the notion that, have caused variations in the results were the
when corrected for the local light intensity at the enamel and dentin distribution, the properties for
gauge surface, the strain gauge measures post-gel the dental tissues, the shape of the cavities, the
shrinkage as a local rather than a bulk value as was applied light intensity, the exact distance between
initially presumed. light guide and restoration during curing, etc.
The calculated tooth deformations (Fig. 4) also Taking into consideration the inherent variation in
compared well in patterns and magnitudes with the experiments, validation is successfully achieved
experimentally determined deformations (Fig. 1). when a general similarity in deformation patterns
Deformation can be defined in different ways. In and deformation values is demonstrated.
order to ensure identical definition and measure- The goal of polymerization shrinkage research is
ment for the numerical and experimental to understand and predict the development of
approaches, the deformation for both was quanti- residual shrinkage stresses. The calculated defor-
fied using the same software (Cumulus). Addition- mation was caused by residual stresses predicted by
ally, the calculated cuspal flexure data was in the the numerical polymerization model. Having vali-
same range as has been reported in the literature dated the deformation, the calculated stress
(16 – 45 mm).16 – 20 An important conclusion that can distributions associated with them can be con-
be drawn from the positive validation of the sidered validated too. It was found that the stress
numerical polymerization model is that post-gel levels in the tooth increased with increasing
shrinkage must be used to calculate residual restoration size, while stresses in the restoration
shrinkage stresses (and deformations) rather than and along the tooth-restoration interface decrease
total shrinkage. Using total shrinkage values, (Figs. 5 and 6). This result can be explained by the
typically about five to ten times higher than post- increasing loss of dental hard tissues for increasing
gel shrinkage, would have grossly overestimated restoration sizes. Removal of dental hard tissue
the stresses (and deformations).21 Consequently, decreases the stiffness of the tooth. Residual
shrinkage stress cannot be approximated from shrinkage stresses are caused by the resistance of
shrinkage that is determined by displaced volume the tooth structure to shrinkage deformation of the
or weight methods. Since a correlation between bonded restoration. Therefore, a decreased stiff-
post-gel and total shrinkage is improbable (post-gel ness of the supporting tooth structure decreases
shrinkage depends mainly on reaction kinetics), residual stresses in the restoration and tooth-
total shrinkage values are unsuitable for qualitative restoration interface. However, it also results in
shrinkage stress estimations (e.g. in shrinkage more tooth deformation and consequently higher
properties rankings). Note that it is not our stresses in the tooth. Deformation measurements of
intention to nullify the value of total shrinkage the tooth crown are therefore more indicative of
measurements. For example, total shrinkage plays residual stresses in the tooth structure than
a decisive role in the direction of shrinkage;24 of shrinkage stresses in the composite restoration
however, total shrinkage does not reflect shrinkage or across their interfaces. Note that the stresses at
stress expectations. the enamel surfaces and the stresses across the
Strictly speaking, post-gel shrinkage cannot be tooth-restoration interface should not be compared
extrapolated to predict shrinkage stresses either. with each other in absolute terms, since they
As discussed before, stress is not a material involve different materials. Furthermore, the inter-
property. The stresses that were calculated by the facial stress values were extrapolated across the
shrinkage model were the result of the combination tooth-restoration interface. Since the bonding layer
of all properties (tooth and restoration), the whole has not been modeled separately, the interfacial
anatomy (tooth and restoration), all boundary stresses should only be viewed as a qualitative
conditions (root and interfaces), and the complete indication of an interfacial stress distribution that a
restoration history (each incremental filling and bonding layer would encounter.
Residual shrinkage stress distributions in molars after composite restoration 563

Areas of high stresses are at a higher risk for shrinkage model that was based on experimental
failure. Although the overall stress level along the polymerization data. Although more has to be done
tooth-restoration interface was generally higher for to unravel remaining less-understood polymeriz-
the Class I configuration, the maximum stresses of ation sequences, the current study confirmed the
the other configurations approached or met the validity of a linear elastic approach based on the
maximum stress values of the Class I (Fig. 6). When post-gel shrinkage concept for the calculation of
composites were introduced, Bowen25 predicted residual stresses in a restored tooth. It was shown
that restoration type will affect the resulting stress that tooth deformation indicated the stress levels in
levels. He noted that cavities with a large free the tooth, rather than in the restoration. Shrinkage
surface in proportion to bonded cavity walls would stress could not be expressed into a single average
develop stresses of lower magnitudes. The current value based on composite properties or restoration
study confirms Bowen’s early observation as a configuration alone, but had to be approached as a
general rule-of-thumb for the stress state in the distribution that depended on the location and
composite and the tooth-restoration interface. properties of the tooth and restoration, tooth and
However, extrapolation of the ratio between restoration geometry, tooth and restoration fix-
bonded and free surfaces as a prediction for ture, and the consecutive restorative procedures.
shrinkage stresses, as has been proposed by others,
cannot hold, even for stress levels in the restoration
and across the interface. Fig. 6 shows that the ratio Acknowledgements
between bonded and free surface areas does not
consistently predict the stress levels. For example Based in part on abstract No. 498, presented at the
(Fig. 6C), the small versus large Class I restorations 80th IADR meeting in San Diego, March 6 –9, 2002.
with almost identical ratios show large differences in The authors thank Iryna B. Olson for her assistance
the interfacial stress levels. Moreover, the bonded- with the Cumulus analysis. This study was supported
free surface ratio of the large Class I is much higher by the Minnesota Dental Research Center for
than the ratio for the Class II OM, but their interfacial Biomaterials and Biomechanics, a Faculty Develop-
stress levels are similar. A last example is the large ment Grant from Chulalongkorn University, and
Class II OM, which although it has a higher bonded- NIH/NIDR Grant R01-DE12225.
free surface ratio than the small Class II MOD, seems
to have higher overall interfacial stress levels.
References
The validated residual shrinkage stress distri-
butions that were determined in this study confirm 1. Jensen ME, Chan DCN. Polymerization shrinkage and micro-
that stress is a local property, which within one leakage. In: Vanherle G, Smith DC, editors. Posterior
restored tooth may have high values in one location composite resin dental restorative materials. Utrecht:
and low values in another. Shrinkage stresses cannot Peter Szulc Publishing Company; 1985. p. 243—62.
be captured by single values, such as cusp flexure or 2. Eick JD, Welch FH. Polymerization shrinkage of posterior
composite resins and its possible influence on postoperative
restoration configuration, but must be studied as a sensitivity. Quintessence Int 1986;17:103—11.
location dependent distribution that takes the 3. Lai JH, Johnson AE. Measuring polymerization shrinkage of
whole restored tooth into consideration. In the photo-activated restorative materials by a water-filled
introduction, various clinical symptoms were listed dilatometer. Dent Mater 1993;9:139—43.
4. Unterbrink GL, Muessner R. Influence of light intensity on
that have been associated with shrinkage stresses.
two restorative systems. J Dent 1995;23:183—9.
Those symptoms also involve all aspects of the 5. Bouschlicher MR, Vargas MA, Boyer DB. Effect of composite
restored tooth, i.e. the restoration (marginal loss), type, light intensity, configuration factor and laser polym-
the interface (adaptation, microleakage, secondary erization on polymerization contraction forces. Am J Dent
caries), and the tooth structure (propagation of 1997;10:88—96.
6. Kinomoto Y, Torii M, Takeshige F, Ebisu S. Comparison of
enamel microcracks, post-operative sensitivity).
polymerization contraction stresses between self- and light-
Any discussion of residual shrinkage stresses should, curing composites. J Dent 1999;27:383—9.
therefore, involve the whole tooth, not just the 7. Ausiello P, Apicella A, Davidson CL, Rengo S. 3D-finite
composite restoration or composite properties. element analyses of cusp movements in a human upper
premolar, restored with adhesive resin-based composites.
J Biomech 2001;34:1269—77.
8. Tantbirojn D, Versluis A, Pintado MR, DeLong R, Douglas WH.
Conclusions Tooth deformation patterns in molars after composite
restoration. Dent Mater 2003;update with PII: S0109-
5641(03)00158-1.
This study used deformation patterns—measured 9. Versluis A, Tantbirojn D, Douglas WH. Distribution of
experimentally—to validate a biomechanical transient properties during polymerization of a light-initiated
564 A. Versluis et al.

restorative composite. Dent Mater 2003;update with PII: 18. Suliman AA, Boyer DB, Lakes RS. Interferometric measure-
S0109-5641(03)00159-3. ments of cusp deformation of teeth restored with compo-
10. DeLong R, Pintado MR, Douglas WH. Measurement of change site. J Dent Res 1993;72:1532—6.
in surface contour by computer graphics. Dent Mater 1985; 19. Meredith N, Setchell DJ. In vitro measurement of cuspal
1:27—30. strain and displacement in composite restored teeth. J Dent
11. Marshall DB, Noma T, Evans AG. A simple method for 1997;25:331—7.
determining elastic-modulus-to-hardness ratios using Knoop 20. Alomari QD, Reinhardt JW, Boyer DB. Effect of liners on cusp
indentation measurements. J Am Ceram Soc 1982;65: deflection and gap formation in composite restorations.
C175—6. Oper Dent 2001;26:406—11.
12. Nomoto R, Uchida K, Hirasawa T. Effect of light intensity on 21. Versluis A, Douglas WH, Cross M, Sakaguchi RL. Does an
polymerization of light-cured composite resins. Dent Mater J incremental filling technique reduce polymerization shrink-
1994;13:198—205. age stresses? J Dent Res 1996;75:871—8.
13. Suh BI. Controlling and understanding the polymerization 22. Korioth TWP, Versluis A. Modeling the mechanical
shrinkage-induced stresses in light-cured composites. Com- behavior of the jaws and their related structures using
pendium 1999;20(Suppl 25):S34—S41.
finite element (FE) analysis. Crit Rev Oral Biol Med 1997;
14. Versluis A, Tantbirojn D, Douglas WH. Why do shear bond
8:90—104.
tests pull out dentin? J Dent Res 1997;76:1298—307.
23. Bowen RL. Properties of a silica-reinforced polymer for
15. Morin D, DeLong R, Douglas WH. Cusp reinforcement by the
dental restorations. J Am Dent Assoc 1963;66:57—64.
acid-etch technique. J Dent Res 1984;63:1075—8.
24. Versluis A, Tantbirojn D, Douglas WH. Do dental compo-
16. Pearson GJ, Hegarty SM. Cusp movement of molar teeth with
sites always shrink toward the light? J Dent Res 1998;77:
composite filling materials in conventional and modified
1435—45.
MOD cavities. Br Dent J 1989;166:162—5.
17. Suliman AA, Boyer DB, Lakes RS. Cusp movement in 25. Bowen RL. Adhesive bonding of various materials to hard
premolars resulting from composite polymerization shrink- tooth tissues. VI. Forces developing in direct-filling materials
age. Dent Mater 1993;9:6—10. during hardening. J Am Dent Assoc 1967;74:439—45.

You might also like