Journal of Biomechanics 34 (2001) 12691277

3D-nite element analyses of cusp movements in a human upper premolar, restored with adhesive resin-based composites
P. Ausielloa,*, A. Apicellab, C.L. Davidsonc, S. Rengoa
a

School of Dentistry, University of Naples, Federico II, Policlinico Edicio 14, Via Pansini 5, 80131 Naples, Italy b Department of Material & Production Engineering, University of Naples, Federico II, Italy c Department of Dental Materials Sciences. ACTA, University of Amsterdam, The Netherlands Accepted 30 May 2001

Abstract The combination of diverse materials and complex geometry makes stress distribution analysis in teeth very complicated. Simulation in a computerized model might enable a study of the simultaneous interaction of the many variables. A 3D solid model of a human maxillary premolar was prepared and exported into a 3D-nite element model (FEM). Additionally, a generic class II MOD cavity preparation and restoration was simulated in the FEM model by a proper choice of the mesh volumes. A validation procedure of the FEM model was executed based on a comparison of theoretical calculations and experimental data. Dierent rigidities were assigned to the adhesive system and restorative materials. Two dierent stress conditions were simulated: (a) stresses arising from the polymerization shrinkage and (b) stresses resulting from shrinkage stress in combination with vertical occlusal loading. Three dierent cases were analyzed: a sound tooth, a tooth with a class II MOD cavity, adhesively restored with a high (25 GPa) and one with a low (12.5 GPa) elastic modulus composite. The cusp movements induced by polymerization stress and (over)functional occlusal loading were evaluated. While cusp displacement was higher for the more rigid composites due to the prestressing from polymerization shrinkage, cusp movements turned out to be lower for the more exible composites in case the restored tooth which was stressed by the occlusal loading. This preliminary study by 3D FEA on adhesively restored teeth with a class II MOD cavity indicated that Youngs modulus values of the restorative materials play an essential role in the success of the restoration. Premature failure due to stresses arising from polymerization shrinkage and occlusal loading can be prevented by proper selection and combination of materials. r 2001 Published by Elsevier Science Ltd.
Keywords: 3D FEA; Adhesive tooth restorations; Resin-based composites; Polymerization shrinkage; Occlusal loading; Stress distribution

1. Introduction Resin-based materials are being employed to a large extent in contemporary restorative dentistry. A prominent advantage is the possibility of bonding the restoration to tooth structure. Signicant disadvantages of these materials are the polymerization contraction and the mismatch in mechanical characteristics with tooth structures (Davidson, 2000). During hardening, polymerization shrinkage stresses the adhesively placed restoration, while the mechanical mismatch leads to interfacial stress concentrations during functioning
*Corresponding author. Tel.: +39-081-7462089; fax: +39-0817462197. E-mail address: pietausi@unina.it (P. Ausiello). 0021-9290/01/$ - see front matter r 2001 Published by Elsevier Science Ltd. PII: S 0 0 2 1 - 9 2 9 0 ( 0 1 ) 0 0 0 9 8 - 7

(mastication, temperature uctuations). Modifying both material characteristics by adding inorganic ller particles to the resin, results in shrinkage reduction and strength and stiness (E-modulus) increase. Unfortunately, increase of modulus is associated with higher polymerization shrinkage stresses. Stress control in resin-based composite restorations is a major problem in operative dentistry. Clinical studies have shown that fractures in teeth with class II MOD restorations are mainly related to insucient residual hard tooth structure, restorative procedures and material selection (Wendt et al., 1987). Occlusal loading of teeth, adhesively restored with a variety of materials (amalgam, resin-based composites and combinations of glass-ionomers with resin-based composites), did not lead to equal fracture resistance

1270

P. Ausiello et al. / Journal of Biomechanics 34 (2001) 12691277

values nor comparable fracture patterns (Ausiello et al., 1997). Resin-based composite restored teeth often show mesial-distal crown fractures with the composite material fracture located at the critical tooth-resin interface. This behavior has been attributed to the stresses arising from the polymerization shrinkage (Davidson and De Gee, 1984), which results in stress build-up in the restored tooth and cusp displacements (Pearson and Hegarty, 1987). Repeated functional loading will cause fatiguing of the restored tooth and will ultimately result in failure. Ausiello et al. (1999a, b) investigated in vitro the performance of adhesively restored large and deep cavities in maxillary premolars under cyclic loading. That fatiguing process showed that, when polymerization shrinkage pre-stressing conditions are present, the adhesive restoration-cavity wall interface initially debonds at the weakest bonded areas between the dentin and the restorative material. As reported in several previous studies (Koike et al., 1990), polymerization shrinkage stress of adhesive composite restorations depends on several factors such as restoration size, cavity shape, incremental or bulk placement technique, water sorption, composite creep and cusp movements (Feilzer et al., 1990). Moreover, it has been shown that increased exibility (reduced E-modulus) of the restoration reduces the polymerization stress (Kemp-Scholte and Davidson, 1990). Suliman et al. (1993) investigated in vitro the cusps deection in adhesively restored natural teeth under dierent polymerization compensating eects by using a microscope with a micrometer. According to that study, cusps deection dierences depend on cavity size, wet and dry storage conditions and lling material rigidity. Apparently, there are a variety of factors that are simultaneously active. Some enlarge each other, whilst other mechanisms counteract each other. Proper understanding of the physical phenomena occurring in the adhesively restored tooth during the placement and setting of the materials and throughout it is functioning is crucial in the determination of the proper placement procedures and material selection. Structural analyses of the integral process may oer an essential tool for the assessment and reduction of fracture risks, ensuring optimal performance in selection of the restorative material combinations and material application protocols. Since stress distribution investigation of teeth, in particular after restoration, is very complicated due to the complex geometry, 3DFinite Element Analysis might be a powerful tool to visualize the problems. The aim of this study was to provide FEA engineering tools for the understanding of the inuence of the shrinkage characteristics and composite rigidity on the extent of cusps displacement and localization of critical sites in a given restored tooth under (over) functional loading. Moreover, validation of the proposed procedure of the Finite Element Model (FEM) was investi-

gated by a comparison of calculated theory and mechanical experimental data.

2. Materials and methods 3D-nite element approach consists in dividing a geometric model into a nite number of elements in which the variables of interest are approximated with some mathematical functions. Biomedical applications of this method are already known in some elds of medicine (Apicella et al., 1994; Dalstra et al., 1995; Apicella et al., 1998). In our study, a 3D FEM model of a rst upper premolar was realized. The rst step was the solid model generation in which the shapes of dentin, pulp and enamel were obtained. The space for the class II MOD cavity preparation was also dened. Then the solid model was lled with elements, creating the mesh, and the material properties of dentin and enamel were assigned to the elements, which lled the corresponding regions. Thus the sound tooth FEM model was obtained. Subsequently, the FEM model of the restored tooth was realized by changing element material properties in the central zone of cavity preparation. At the end loads and geometric conditions were applied. 2.1. Tooth solid model generation Tooth solid model was generated using literature data on the tooth morphology for the denition of the dentine and enamel volumes (Wheeler, 1974) and a plaster model (Thanaka model, Japan, 1978) for the external shape denition. Crown and roots were constructed in two dierent phases and assembled afterwards. The crown was realized by digitizing an upper premolar tooth plaster model on the scale of one to ve by a Cyberware laser scanner. Over two hundred proles were generated at 0,33 mm increments by laser scanning in two dierent directions: a vertical and a horizontal one. Among all the proles, only 34 were collected, 17 vertical and 17 horizontal at 2 mm increments, and were assembled in a 3D wire-frame structure (Fig. 1) by means of a 3D CAD (Autocad 12, Autodesk, Inc., Neuchatel, Swiss, 1992). The wire-frame curves were exported in Pro-Engineer 16.0 (Parametric Technology Co., Waltham, MA, USA, 1994), where a solid model was generated tting the horizontal and vertical proles. The model was cut in the cervical area in order to obtain the nal crown (Fig. 2). The roots were modelled by means of their mesialdistal and buccal-lingual representations taken from the literature (Braden, 1976). The two representations were scanned and 8 vertical proles were generated imitating

P. Ausiello et al. / Journal of Biomechanics 34 (2001) 12691277

1271

Fig. 1. 3D Solid upper premolar model and nite element model generation.

Fig. 2. On-o experimental test with FEM simulation.

the scanned images. The roots were constructed, tting the vertical proles (Fig. 1). The pulp region was obtained in an analogous way and subtracted from the roots.

The crown and the roots with the pulp chamber were assembled in the nal model (Fig. 1). Dening parametric cutting plane, it is also possible to easily realize dierent cavities and MOD preparations.

1272

P. Ausiello et al. / Journal of Biomechanics 34 (2001) 12691277

In Fig. 5, a MOD II cavity is shown with 3.5 mm occlusal width. 2.2. FEM model generation The solid model was exported into ANSYS rel. 5.3 (Ansys Inc, Houston, USA, 1994), using the IGES format. The volumes were redened and meshed with 8-node brick and 4-node tetrahedral elements, resulting in an 11165-element and 7340-node structure. Dierent material properties were assigned to the elements according to the volume denition. Three dierent models were realized. The rst model (Mod. A) is the sound tooth; the second model (Mod. B) is a restored tooth with a more rigid composite; the third model is a restored tooth with a less rigid composite (Mod. C). Material properties are listed in Table 1. Some assumptions were made in order to simplify the calculations. Absolute bonding was considered among enamel, dentin and composite. The pulp chamber was modeled as a void because of its negligible stiness and strength. Despite their intrinsic anisotropic nature, dentin and enamel can be assumed homogeneous and isotropic (Darendeliler et al., 1998; Versluis et al., 1996) because their anisotropy belongs to a microscopic scale whereas the tooth model is macroscopic. Furthermore, all materials were considered elastic throughout the entire deformation, which is a reasonable assumption for brittle materials in non-failure conditions (Rees et al., 1995). 2.3. Model experimental validation In order to validate the FEM model, a compression test was performed on a class II MOD restored human upper premolar. The resin-based composite Prodigy (Kerr, USA), with a Youngs modulus of 12.5 GPa and polymerization shrinkage of 2.7370.31% by volume in combination with Optibond FL adhesive (Kerr, USA) was used for the restoration. The tooth was inserted until the cementum-enamel junction in a steel cylindrical ring with the apical root area in contact with the steel ring oor. Subsequently, it was lled with resin

composite material, so it was possible to record only the material deformation within the tooth. The cylinder was clamped to the test machine and the load was applied vertically by means of a 6 mm diameter steel cylinder with the axis normal to the tooth axis in order to simulate one main important force which develops during occlusion (Mc Neil, 1997). A 1 mm/min constant rate was imposed on the loading cylinder and the vertical displacement and the axial load were acquired until the restored tooth fractured (experimental curve in Fig. 3). The same test has been simulated using the FE analysis and the results are compared. The comparison test loading cylinder was modelled with the 3D elastic beams (Fig. 3). The beams have one end in common and the other one on a cusp. End rotations have not constrained. The common end was displaced in the central position of the loading cylinder section in the experimental test. The load is applied on the tooth at two points (Fig. 3), namely the beams elements (blue lines) and on the cusps (red areas). The resulting force F crossing the central position of the loading cylinder section is 400 N (Fig. 3, red arrow). The beam elastic properties were thought to be innitely rigid compared to the tooth. Resin support was not modelled and it was considered to be rigid as the loading system. Other studies (Darendeliler et al., 1998) also assumed the supporting alveolar bone rigidity. 2.4. Numerical simulations Three dierent situations were simulated: a compression test with a 400 N occlusal load, composite polymerization shrinkage and the combination of occlusal loading and shrinkage. In these simulations, the tooth model had to be constrained apically to the enamel-dentin junction. This fact permitted to eliminate the lower part of the mesh, reducing the model size to 7894 elements and 5812 nodes. Numerical results were obtained for the three models (A, B and C). Particularly, the vertical displacement of the load application point in model C can be used as the numerical datum for the model experimental validation. For the evaluation of the stresses arising from the composite polymerization shrinkage, a volumetric contraction has been applied to the composite. The shrinkage data required for this study were obtained by using a linometer (De Gee et al., 1993). The shrinkage which determines cusp deections, however, cannot be directly related to the amount measured in the linometer free shrinkage tests. This is due to the viscoelastic relaxation occurring in the material during the passage from a viscous uid into an elastic solid. It has to be noted that when a composite is adhesively placed in the tooth cavity, only a part of the shrinkage produces cusp deections and a part relaxes. In this paper an on-o test is described: composites discs were

Table 1 Youngs modulus (GPa) Dentin Enamel More rigid composite (B) Less rigid composite (C)
a b

Poissons ratio 0.23b 0.3b 0.3 0.3

18 48a 25 12.5

Wheeler (1974). Verluis (1996).

P. Ausiello et al. / Journal of Biomechanics 34 (2001) 12691277

1273

polymerized on a glass substrate. In that study, the composite shrinkage stresses should overcome the silica glass substrate strength at about 20 % of the value measured in the linometer test: nonetheless, fracture was observed in a few cases only when total shrinkage was attained. This datum was arbitrarily used as reference in this study. A 0.54% constrained volume shrinkage (2.7% free volumetric shrinkage) was assumed to occur during the inside cavity resin-based composite polymerization.

3. Results All the analyses performed are linear static. The comparison between the experimental curve and the numerically determined one is shown in Fig. 3. Fig. 4 shows the Von Mises equivalent stress which was evaluated for the three models under vertical (over)functional loading as well as total displacement. In Fig. 5 the path traces are indicated with the two yellow small circular points. Both the paths are normal to the tooth axis, mesial-distally directed, little above the pulp chamber and near the interface of models B and C between composite and natural tooth. One path is on the side of the composite restoration; the other path is on the natural tooth side. Von Mises equivalent stress was evaluated along these paths (Figs. 6 and 7). A Von Mises equivalent stress map was also extracted for models B and C under shrinkage condition (Fig. 8). Two other paths were dened on these models, following the buccal and lingual wall proles of the tooth (Fig. 9). Along these paths normal displacement to the tooth axis was determined. In Fig. 9 the cusp displacements are plotted. Lingual cusp displacement is considered negative towards the center of the tooth, while buccal cusp displacement is considered positive towards the center of the tooth.

Fig. 3. Occlusal vertical load and comparison between experimental and numerical data.

Fig. 4. Von Mises stress distribution due to vertical loading (up) and displacement distribution (down).

1274

P. Ausiello et al. / Journal of Biomechanics 34 (2001) 12691277

Fig. 5. Paths dened in tooth.

Fig. 6. Von Mises stress path in cusp (natural tooth).

Fig. 7. Von Mises stress path in cavity region (composite).

In Fig. 10 the Von Mises equivalent stress maps for models B and C under the vertical loading and shrinkage combination and model A under vertical loading only are shown.

4. Discussion Endodontically treated teeth are easily subject to fracture as a result of the modication of their natural

rigidity and to pre-stressing, generated by the restorative procedures. Conventional conservative class II MOD composite restorations may exhibit micro-cracking and interfacial failure due to internal pre-stressing from altered cusp movements as a result of resin-based composite polymerization contraction and occlusal loading. The study of the eects of restoration techniques on endodontically treated teeth usually encounters all the negative inuences associated with the use of natural tooth samples. Teeth, in fact, dier from each other for obvious reasons, patient age, sex, etc., and they are usually responsible for large standard deviations of the determined means values. Therefore, dierent laboratory test results are dicult to compare. Consequently, it is of paramount importance to rstly exclude the anatomical sample dierences and statistical variability. In the present study, the mechanical behavior of a bicuspid, subjected to polymerization contraction and occlusal loading has been investigated by means of a computer analysis and numerical results were validated by laboratory experimental data. FEM provides a powerful tool for analyzing the mechanical behavior of complex structures. For this reason in the last two decades FEM was widely used in studying biological systems. In fact, a mathematical model permits to evaluate natural systems response under various loads or geometric conditions apart from the high dispersion, which characterizes experimental data. Obviously, FEM models need an experimental validation, which is purposely prepared and denitive. If possible, suitable laboratory data for validation can also be collected from literature in order to minimize time and costs. After validation, the model can be extensively used for a wide range of studies. Only signicant changes in the FEM model require a new experimental validation. The system analysis procedures proposed in this study create a feedback mechanism that helps in closing the structures and material design loop with engineering prediction of application stresses localization and build-up coupled with a structural performance. In the last two decades many works have shown how the 2D nite elements analysis applied to dental mechanics has become a popular numerical method to investigate the critical aspects related to stress distribution in tooth (Sakaguchi et al., 1991) and in dental restoratives (Farah and Craig, 1974) or to evaluate the stress relief of shrinking resin-based materials (Versluis et al., 1996). The use of a more detailed 3D dental model to test the inuence of elastic and shrinking properties of resinbased restoring material commonly used in operative dentistry (such as light curing resin-based composite), seems to be of extreme interest to understand critical problems related to the restorative material choice and optimal application procedures denition.

P. Ausiello et al. / Journal of Biomechanics 34 (2001) 12691277

1275

Fig. 8. Von Mises stress distribution due to polymerization contraction.

Fig. 9. Displacement path in cavity region.

We focused our interest on the investigation of the mechanical behavior of a critical system such as the rst upper premolar restored by a class II MOD cavity preparation (where a consistent amount of dentin and enamel tissue is lost and the integrity of the structure is seriously altered). Methods for geometric data acquisition, geometric model creation and modication (parametric modelling) and nite element model generation though CAE ambient are presented and discussed. The FEM prediction is in good agreement with the experimental data, as shown in Fig. 3. Comparison underlines the great accuracy for the tooth model with respect to a very complex simulation as a compressive test of restored tooth. Our numerical data are also in good agreement with experimental ones obtained by Suliman et al., 1993 by means of a microscope with a micrometer stage. Considering, rstly, the occlusal loading eect alone, the sound tooth exhibits a wider high stress area (red), localized in correspondence of the occlusal enamel, than the restored teeth (Fig. 5, upper left side). This is due to the rigidity of the enamel. In particular, looking at

models B and C, in which a dierent composite rigidity was used, a wider high stress area was found in the model with the more rigid composite (Mod. B). In fact, a less rigid structure (Mod. C) releases stress in greater deformation. In this way, Fig. 4 shows that the cusp movements increase from model A to C. Even if high stress areas are reduced in Mod. C compared to Mod. B, it does not mean that its mechanical behavior is the best one. In fact, this reduction occurs in the composite restoration, which is less rigid, and its lower rigidity allows greater cusp movements. So the average stress in the entire structure is lower but the stress values in buccal and lingual cusps are higher. These preliminary considerations, however, do not take into account setting composite shrinkage. Obviously, the best behavior is relative to the sound tooth model with a very large stress area in the central enamel region and low cusp deections. This consideration is conrmed from the Von Mises stress path into the natural tooth (Fig. 6). The peaks on the left- and on the right-hand side depend on the enamel, which is more rigid than dentin. The curves have the same shape for the three models but there is higher stress for the less rigid structure. It is interesting to note the dierence between this Von Mises path and the Von Mises path in composite (Fig. 7). Along the two dierent paths the sound tooth exhibits an analogous behavior, with two peaks in enamel. These two peaks disappear for the restored models in which there is a single material: the composite. In Fig. 7, in fact, Models B and C have similar curves but Mod. B exhibits higher values. Mod. C behaves in the central zone as the dentin of the sound tooth. Applying contraction load in models B and C, the high stress areas are concentrated near the compositetooth interface, where failure can occur (Fig. 8). Even if failure does not occur during composite curing, a prestress arises in the structure. Higher modulus restoration exhibits higher stress values. In fact, a less rigid restoration can relax the applied stress by means of a

1276

P. Ausiello et al. / Journal of Biomechanics 34 (2001) 12691277

Fig. 10. Von Mises stress distribution due to vertical loading for Mod. A and to shrinkage and vertical loading for Models B and C.

greater elastic deformation. This eect is also shown in Fig. 9 where the displacements normal to the tooth axis are plotted. The less rigid composite exhibits a greater elastic deformation that is transferred to a lower deformation of the cusps. Higher stress release depends also on viscous ow of the composite during curing. But this eect was not directly taken into account in these models. Simply, the contraction load input in simulations was signicantly reduced compared to the free contraction. At the end, occlusal loading and shrinkage were combined for models B and C. A Von Mises stress map is plotted in Fig. 10 together with that of the sound tooth under occlusal loading. Models B and C exhibit a similar behavior, even if Mod. B has higher stress in the upper part of the restoration whereas Mod. C has higher stress in the cusps. However, this eect is greatly reduced compared to the single occlusal loading case. The important dierence is between these two models and the sound tooth because they show high stress values at the composite-tooth interface, in a zone where the sound tooth exhibits low stress. This study was designed to investigate the correlation between resin composite elastic properties and restored tooth stiness. Findings indicate that more rigid composites lead to lower cusp movements under occlusal loading but exhibit higher pre-loading eect. A good composite for restoration has to balance the two opposite eects. In this way a low pre-load on the cusps can be accepted in order to reach sucient restoration rigidity.

limiting cusp movements. Furthermore, it does not have to bend cusps as a consequence of its curing, otherwise the pre-loading eect can be disastrous. The problem is that rigid composites have low elastic release. In this way the solution is in balancing the two eects, accepting a low pre-loading eect in order to obtain sucient rigidity. A new point of view can be introduced considering two aspects not contemplated in the present model. On the one hand there is the viscous ow that can generate a great stress release during composite curing. But viscous ow needs time and actually the light curing composite kinetics are too fast for this. On the other hand, there is the adhesive interface, which can limit the cusp preloading eect with its elastic deformation. By changing the adhesive elastic modulus or by simply changing its thickness it is possible to obtain dierent rigidity of the interface region. In this direction the authors plan to develop the actual study, making use again of FEM simulations, which prove to be an ecient tool in complex structure analysis.

Acknowledgements Particular thanks to Eng. F. Quadrini and Eng. L. Di Palma for their collaboration in the research on the experimental model validation and numerical simulations.

References 5. Conclusions FEM simulations supply information about mechanical behavior of sound and restored teeth. The best restoration is that which allows restored tooth to react to external loads as the sound tooth does. For this, the composite has to store a great quantity of energy,
Apicella, A., Liguori, A., Masi, E., Nicolais, L., 1994. Thick laminate composite modeling in total hip replacement. In: Found, M.S., Sheeld (Ed.), Proceedings of Second International Seminar on Experimental Techniques and Design in Composite Materials. Apicella, A., Masi, E., Nicolais, L., Zarone, F., De Rosa, N., Valletta, G., 1998. A nite element model study of occlusal schemes in full arch implant restoration. Journal of Material Science: Materials in Medicine 1, 191196.

P. Ausiello et al. / Journal of Biomechanics 34 (2001) 12691277 Ausiello, P., De Gee, A.J., Rengo, S., Davidson, C.L., 1997. Fracture resistance of endodontically treated premolars adhesively restored. American Journal of Dentistry 10, 237241. Ausiello, P., Davidson, C.L., Cascone, P., De Gee, A.J., Rengo, S., 1999a. Debonding of adhesively restored deep class II MOD restorations after functional loading. American Journal of Dentistry 12, 8488. Ausiello, P., Apiciella, A., Quadrini, F., Rengo, S., Davidson, C.L., 1999b. Experimentally determined visco-elastic properties of light curing composites in a 3D FEA of shrinkage-stress release in class II MOD restorations. Proceedings of the Third International Association Dental Research Continental European Division, Nice, Abstract No. 295. Braden, M., 1976. Biophysics of the root. In: Frontiers in Oral Physiology. Basel, Karger, Vol. 2, pp. 137. Dalstra, M., Huiskes, R., van Erning, L., 1995. Development and validation of a three-dimensional nite element model of the pelvic bone. Journal of Biomechanical Engineering 117, 272278. Darendeliler, S.Y., Alacam, T., Yaman, Y., 1998. Analysis of stress distribution in a maxillary central incisor subjected to various post and core applications. Journal of Endodontics 24, 107111. Davidson, C.L., 2000. Handling of polymerization stresses in resinbased restorative materials. Dental News 7 (4), 913. Davidson, C.L., De Gee, A.J., 1984. Relaxation of polymerization contraction stresses by ow in dental composites. Journal of Dental Research 63, 146148. De Gee, A.J., Feilzer, A.J., Davidson, C.L., 1993. True polymerization shrinkage of unlled resins and composites determined with a linometer. Dental Materials 9, 1114. Farah, J.W., Craig, R.G., 1974. Finite element stress analysis of a restored asymmetric rst molar. Journal of Dental Research 52, 859866.

1277

Feilzer, A.J., De Gee, A.J., Davidson, C.L., 1990. Relaxation of contraction shear stress by hygroscopic expansion. Journal of Dental Research 69, 3639. Kemp-Scholte, C.M., Davidson, C.L., 1990. Marginal integrity related to bond strength and elasticity of the composite resin restorative system. Journal of Prosthetic Dentistry 64, 258264. Koike, T., Hasegawa, T., Menabe, A., Itoh, K., Wakumoto, S., 1990. Eect of water sorption and thermal stresses on cavity adaptation of dental composites. Dental Materials 6, 178180. McNeill, A.A., 1997. Science and Practice of Occlusion. Quintessence, Illinois. Pearson, G.J., Hegarty, S.M., 1987. Cusp movement in molar teeth using dentine adhesives and composite lling materials. Biomaterials 8, 473476. Rees, J.S., Jacobsen, P.H., 1995. Modelling the eects of enamel anisotropy with the nite element method. Journal of Oral Rehabilitation 22, 451454. Sakaguchi, R.L., Brust, E.W., Cross, M., De Long, R., Douglas, W.H., 1991. Independent movement of cusps during occlusal loading. Dental Materials 7, 186190. Suliman, A.A., Boyer, D.B., Lakes, R.S., 1993. Interferometric measurements of cusp deformation of teeth restored with composite. Journal of Dental Research 11, 15321536. Versluis, A., Douglas, W.H., Cross, M., Sakaguchi, R.L., 1996. Does an incremental lling technique reduce p polymerization shrinkage stresses? Journal of Dental Research 3, 871878. Wendt, S.L., Harris, B.M., Hunt, T.E., 1987. Resistance to cusp fracture in endodontically treated teeth. Dental Materials 3, 232235. Wheeler, R., 1974. Dental Anatomy, Physiology and Occlusion. W.B. Saunders Co., USA.