J Dent Res 75(7): 1477-1483, July, 1996

Finite Element Stress Analysis of Three Filling Techniques for Class V Light-cured Composite Restorations
M.M. Winkler*, T.R. Katonal, and N.H. Paydarl
Dental Materials Division, 'Orthodontics Division and Mechanical Engineering; Biomechanics and Biomaterials Research Center, Indiana University-Purdue University at Indianapolis, School of Dentistry, 1121 West Michigan St., Indianapolis, Indiana 46202; *to whom correspondence should be addressed

Abstract. An important disadvantage of current dental resin composites is polymerization shrinkage. This shrinkage has clinical repercussions such as sensitivity, marginal discoloration, and secondary caries. The objective of this study was to compare three filling techniques in terms of the transient stresses induced at the resin composite/tooth interface during polymerization. The techniques were: bulk filling (B), three horizontal increments (HI), and three wedge increments (WI). A simple Class V cavity preparation was modeled in finite element analysis. Polymerization shrinkage was simulated by a thermal stress analogy, thereby causing 1% shrinkage due to an arbitrary coefficient of thermal expansion. Interface normal and shear stresses were calculated at nine steps during polymerization, proceeding from 0% to 100% volume of cured resin. The importance of the interface transient stresses was revealed by the finding that, in most cases, their peak values exceeded the final or residual stress. Also, the WI and B techniques consistently exhibited the highest and lowest maximum transient stresses, respectively. These results from the simple model of a Class V restoration suggest that bulk filling of light-cured resin composites should be used in restorations which are sufficiently shallow to be cured to their full depth. Key words: composites, polymerization contraction, dental stress analysis.

Introduction
Dental resin composite restoratives experience a reduction in volume of from 1 to 3% during polymerization (Bandyopadhyay, 1982; Rees and Jacobsen, 1989; de Gee et al., 1993; Lai and Johnson, 1993). This contraction results in stresses which tend to pull the resin from the tooth structure (Bowen et al., 1983). One possible result is gap formation at the restoration-tooth interface, which leads to microleakage (Asmussen, 1975). Microleakage is not completely eliminated at the gingival margins of cervical cavities even with the use of dentin bonding adhesives (Airoldi et al., 1992; Sidhu and Henderson, 1992). Leakage may lead to secondary caries, marginal stain, sensitivity, and even restoration loss. Other consequences of polymerization contraction are residual stresses (Bowen et al., 1983) high enough to cause cusp movement (Suliman et al., 1993), sensitivity, and tooth fracture (Hickman and Jacobsen, 1991; Tajima et al., 1993). Reduction of this residual stress occurs over the first few weeks after restoration placement, due to water sorption by the resin (Bowen et al., 1982; Suliman et al., 1993). Several methods of counteracting shrinkage and the resulting internal stresses, such as utilizing more adhesive bonding agents and building up restorations in several increments, have been attempted. Incremental placement techniques are needed to ensure satisfactory polymerization of light-cured restorations in deep preparations (Watts et al., 1984; Leinfelder, 1991). Increments have also been used to reduce microleakage (Lutz et al., 1986a; Lui et al., 1987; Crim, 1991), but not always successfully (Fuks et al., 1990; Mangum et al., 1994). These methods to improve resin bonding agents have been aimed at reducing the potential for microleakage and at maintaining the bond between tooth and resin composite for enhancing retention. However, success in reducing gap formation and maintaining adhesion without reducing shrinkage increases the residual stresses which may lead to tooth fracture and sensitivity (Hickman and Jacobsen, 1991; Tajima et al., 1993). Flow of the curing resin helps to reduce residual stresses (Davidson and de Gee, 1984; Uno and Asmussen, 1991), but
1477

Received May 1, 1995; Accepted March 1, 1996

1478

Winkler et al.

j Dciit Res 75(7) 1996

(Eick et al., 1993). Changes in cavity preparation design have been investigated in attempts to reduce gap formation and residual stresses (Crim, 1991; Douvitsas, 1991). Likewise, placement techniques have also been examined (Lutz et al., 1986a,b; Sidhu, 1993). In a prior study using a Finite Element Model (FEM), investigators showed that transient stresses in a bulk-filled light-cured Class V resin composite restoration can be higher than the residual stresses after curing is completed (Katona and Winkler, 1994). Based on this finding, our hypothesis is that the placement technique used with a lightcured composite restoration will affect the maximum transient stress developed during polymerization.

mm

3 mm

3 mmm

6 mm

Composite
Enamel (E)

1
s

Materials and methods

Model A linear elastic FEM of an idealized Class V resin composite restoration was developed with PATRAN (PDA Engineering, Costa Mesa, CA) on a Sun SPARCstationi computer (Sun Microsystems, Inc., Mountain View, CA). The model is a symmetric representation of a small prism centered within a larger prism. It assumes that the enamel and dentin surrounding the restoration are of uniform thickness. Analysis was performed with ABAQUS (fllibbit, Karlsson, and Sorensen, Inc., Providence, RI). The preparation (Figs. IA, IB) was 1.8 mm deep with a square (3 mm by 3 mm) outline form. The converged FE model (Fig. IC), based on node displacemenit, contains 2704 nodes and 2160 elements. To accommodate interface (enamelcomposite and dentin-composite) discontinuities, we did not average stress values across boundaries. For boundary conditions, the bottom surface of the enamel was fixed.

1.2 mm:

Composite
Dentin

1.8 mm

3.3 mm

'I

Simulation of polymerization shrinkage

3

Figure 1. (A, B) Schematics of the idealized cavity preparation for the FE model. (C) The FE model is pictured in cross-section with only half of the model present. The three-wedge increment technique is shown with the first step of the first increment shaded. DEJ is the dentino-enamel junction.
it is less effective in deeper cavity preparations and is also speculated to be less effective in light-cured restorations (Feilzer et al., 1990). To address the deficiencies associated with

polymerization shrinkage in resin composites, investigators have taken several approaches. Improvements in the resin composite systems themselves have been sought by the development of resins which do not shrink during curing

To model polymerization shrinkage in the FEM, we simulated shrinkage by a thermal stress analogy. This method has been used previously by Hickman and Jacobsen (1991) and by Katona and Winkler (1994). Contraction was achieved by the exposure of cured resin to an imaginary one-degree temperature drop. This temperature drop, combined with an arbitrarily assigned coefficient of thermal expansion, resulted in a contraction. All results were then scaled to unit (lUY/() volume contraction. The material coefficients are listed in Table 1. We modeled the dynamics of the shrinkage process bv increasing the amount of resin exposed to the temperature drop at nine different "time" steps, proceeding from the initial condition of 00, cured to 1000,, cured (step 9). We accomplished the layered cure in one discrete step by changing the elastic moduli of elements to be cured in the next step from that of uncured resin to that of fully cuied resin. Each layer is fully cured because it is assumed that the light intensity and curing time are sufficient to cure each increment fully. A flat curing front was assumed, since, in a previous study, it was found that the stresses were nearly the same whether the curing form was convex, concave, or flat (Katona and Winkler, 1994).

I Dctit Res 75(7) 1996

Table 1. Material properties

Stress Analysis of Three Fillinig Techn?iquies

1479

Material
Enamel Dentin Cured resin Uncured resin
I b

Young's Modulus (MPa)

84,000" 18,00'

Coefficient of Poisson's Thermal Expansion Ratio 0.33' 0.31' 0.38 0.45c

0.0 0.0 8 x 1()o ()

Aew' 6P

..' i. .. . 1._

T__

-A

13,50()b 1 x 1(2)2

2 3 4 5 6 7 8 9
3
mm

1.8 mm 1

-DEJ1F

Farah t anl., 1988. Charbeneau, 1988. Arbitrary.

B
Stress analysis This analysis focused on the two stress components that are responsible for adhesive failures at the interface. The normal stress component is tensile and pulls the composite away from the tooth. The shear stress component produces sliding movement of the resin composite along the interface. Perfect adhesion between resin and tooth structure was assumed, as was done previously (Katona and Winkler, 1994). Placement techniques Three resin composite placement techniques were followed in preparation restoration: bulk filling, B (Fig. 2A); three horizontal increments, HI (Fig. 2B); and three wedge increments, WI (Fig. 2C). For each technique, the curing was simulated in nine steps in numerical order. The bulk filling method was one increment of resin composite with nine steps, while the other techniques consisted of three increments, each of which was cured in three steps. The volumes of resin composite that were cured in each step were equal for the B and HI techniques but were unequal for the WI technique (Table 2).
4 5
6

7.
8 9

I' X i ; 'i : F! j.

2 3

Figure 2. The three placement techniques are: B, bulk filling (A); HI, three horizontal increments (B); and WI, three wedge increments (C). In all cases, curing is simulated according to num.1erical order, beginning with step I and ending with step 9.

Bi/lkfillihog teclzniqie. As the name implies, the bulk filling method (Fig. 2A) entailed placement of the restorative, and curing commenced with step number 1 at the cavosurface margin. Curing continued in an orderly fashion (steps 1, 2, 3, 4 ...) toward the pulpal floor.

Horizonital io creiuie'nt tecthnique. For the HIl technique (Fig. 25), the
first increment of composite was placed on the pulpal floor and then cured starting with step number I and continuing toward the pulpal floor in a total of three steps. The second increment was positioned directly above the first polymerized increment and cured in three steps (4, 5, and 6), proceeding toward the cured resin in the first increment. The third increment was positioned in a similar manner above the second increment and cured in three steps (7, 8, and 9), beginning at the cavosurface margin and proceeding toward the cured resin in the second increment.

beginning from the outermost layer and proceeding toward the line angle formed by the pulpal floor and the covered wall. The second increment was placed above the first increment so that the uncovered wall was completely covered and the first increment was half covered. Curing was simulated in three steps, proceeding from the outermost layer and continuing toward the line angle formed by the pulpal floor and the newly covered wall. The last increment was placed above the first two layers and covered the entire span between the four walls at the cavosurface margin. Curing proceeded in three steps, beginninig
at the cavosurface margin and continuing toward the pulpal floor until the full depth of the increment was cured.

Wedge incremient thclcniq/ie. Unlike the HI technique, the WI

technique (Fig. 2C) was placed in three increments of unequal volumes. The first wedge increment was placed such that the entire pulpal floor and one wall were covered, the opposing wall was not covered, and the remaining two facing walls were half covered diagonally. Curing proceeded in three steps,

Comparison of placement techniques For comparison purposes, the ratio of the maximum transient stresses to the maximum final or residual stresses was plotted as a function of percent volume cured. Separate plots were prepared for enamel, dentin wall, and dentin floor. The maximum transient stress is the maximum stress that occurs anywhere along the interfaces at the specified cured composite volume. The maximum

1480

Winkler et al.

j Dent Res 75(7) 1996

Table 2. The cured volume of resin composite as a function of "time" step

% of Volume Cured 3 Horizontal 3 Wedge Bulk Filling Increments Increments

A
--

HlI

I
1.5
co

IWI

"Time" Step

(B) 0
11 22 33 44 55 66 77 88 100

(HI) 0
11 22 33 44 55 66 77 88 100

(WI)
0 27.7
44.4 50 63.9 72.2 75 88.9 97.2 100

0
1 2 3 4 5 6 7 8 9

B
Q

2.0 1.5

a 1.0co

final stress is similarly defined, except that the percentage of cured composite is 100%. Note that the maximum residual stress is the same for all three placement techniques. In this manner, at each of the nine "time" steps, the maximum transient stress is compared with the maximum residual stress. Plots were prepared for these interface normal stress ratios (NR) and shear stress ratios (SR).

0.5

0 0.0
.

--

20

80 40 60 % volume cured

100

Results
NR (maximum normal transient stress/maximum normal residual stress) at enamel walls The plot (Fig. 3A) of NR at enamel reveals that the B and HI techniques are similar in magnitude throughout the curing process. In contrast, the WI technique generally exhibits a larger stress ratio until the final step (100% cured).

Figure 3. Plots of maximum normal transient stress/maximum normal residual stress ratio (NR) as a function of percent volume cured at enamel (A) and at dentin (B). (Note maximum normal residual stress at enamel wall = 116.0 MPa; maximum normal residual stress at dentin wall = 113.8 MPa.)

NR (maximum normal transient stress/maximum normal residual stress) at dentin walls The plots (Fig. 3B) of the NR at dentin walls for the B and the HI techniques are different during the early stages of curing. No stresses develop in the dentin for the bulk filling technique, since the composite in this region does not cure until the later stages of polymerization. Again, WI exhibited the largest ratio, but this time, only during the middle stages of polymerization. During the last 30% of curing volume, all three techniques resulted in ratios of approximately the same magnitude.
SR (maximum shear transient stress/maximum shear residual stress) at enamel walls In Fig. 4A, the plot of the maximum SR ratio at enamel reveals that the ratios for all three techniques tend to increase during the curing process. WI attains the highest value of all three techniques as a result of a spike located at approximately 50% volume cured. The bulk filling technique generally has the lowest ratio.

SR (maximum shear transient stress/maximum shear residual stress) at dentin walls Fig. 4B is a plot of SR at dentin walls. As with NR at dentin, the bulk fi]]ing technique does not generate any stress at the walls during the first half of the curing process, since the depth of the cure does not reach the dentin until this time. The ratio curve for WI is similar to the curve for enamel. It peaks at about 50% volume cured and attains the highest value of the three techniques. During the last 20% volume cured, the ratio for the bulk filling jumps above the ratio for the other two techniques before assuming the same value at 100% volume cured.

NR (maximum normal transient stress/maximum normal residual stress) at dentin floor On the floor of the cavity preparation (Fig. 5A), the NR ratios are approximately equivalent. Zero values are due to delayed curing of the B and HI techniques. Note that the final maximum normal stress at the dentin walls (113.8 MPa) is slightly higher than at the floor (109 MPa). Generally throughout polymerization, the higher normal stresses in dentin are evident at the walls, not at the floor.

j Dent Res 75(7) 1996

Stress Analysis of Three Filling Techniques

1481

A
E co
*--I co)ar:

B HI --WIl
-in-.-

A
-o
-&

Bi Hi

WIl

0 0 co

a:

B
o 0

40 co
CO)

a:

% volume cured
Figure 4. Plots of maximum shear transient stress/maximum shear residual stress ratio (SR) as a function of percent volume cured at enamel (A) and at dentin (B). (Note maximum shear residual stress at enamel wall = 50.4 MPa; maximum shear residual stress at dentin wall = 30.4 MPa.)

% volume cured
Figure 5. Plots of maximum transient stress/maximum residual stress ratios (NR or SR) as a function of percent volume cured at floor in dentin for normal stresses (A) and shear stresses (B). (Note maximum residual normal stress at dentin floor = 109 MPa; maximum residual shear stress at dentin floor = 29 MPa.)

SR (maximum shear transient stress/maximum shear residual stress) at dentin floor The SR ratio curves at the floor are shown in Fig. 5B. The ratio for the WI technique is much larger than the ratios for the other two techniques up to 75% volume cured. At 75% volume cured, the ratio for WI drops to the value for the HI technique. Again, the maximum final shear stress at the dentin is slightly higher at the walls (30.4 MPa) than at the floor (29 MPa). Similarly, the maximum shear stresses at dentin throughout curing are generally higher at the walls than at the floor.

Discussion
Generally, the highest stresses occurred prior to full curing. While loss of the bond can occur at any time, the most likely time is when the stress is higher. The calculations for the normal stresses at the dentin walls and enamel indicate that the highest stresses occur during the curing process and not when polymerization is complete (100% cured). Unexpectedly, bulk filling resulted in the lowest maximum normal transient stresses of the three filling techniques considered, at both dentin and enamel walls. As with the normal or tensile stresses at the walls, the peak of the transient shear stress at the enamel and dentin walls is highest for WI.

We simulated polymerization shrinkage by dropping the temperature of the cured resin, which then contracted due to its coefficient of thermal expansion. The final stress levels are the same for all three techniques, because the size, shape, and constraint by the preparation walls are the same when the resin is 100% cured. That is, the simulation allows for the analysis of stresses and strains due to differences in shape and constraint of the resin composite because of its adherence to adjacent tooth structure. The coefficient of thermal expansion is the same in all cases, so the stresses are dictated by the amount of resin shrinkage: the longer the linear distance of material that shrinks, and the smaller the bonding area at the walls of the preparation, the greater the stress. The resin shape and the constraint of the particular volume at any step during polymerization dictate the stresses. Generally, there is greater shrinkage force generated in the longest dimension of the particular cured volume. This helps to explain the highest values for the WI technique, because the longest dimension of the first increment spans the longest linear distance of any increment. That is, the outer layer of the first increment spans from the junction of the pulpal floor and the wall on one side to the cavosurface margin on the opposite wall. This diagonal transverse dimension is longer than any other

1482

Winkler et al.

j Dent Res 75(7) 1996

in composite materials during hardening. J Am Dent Assoc 106:475-477. Charbeneau GT, editor (1988). Principles and practice of operative dentistry. Philadelphia: Lea & Febiger, p. 270. Crim GA (1991). Microleakage of three resin placement techniques. Am J Dent 4:69-72. Davidson CL, de Gee AJ (1984). Relaxation of polymerization contraction stresses by flow in dental composites. J Dent Res 63:146-148. de Gee AJ, Feilzer AJ, Davidson CL (1993). True linear polymerization shrinkage of unfilled resins and composites determined with a linometer. Dent Mater 9:11-14. Douvitsas G (1991). Effect of cavity design on gap formation in Class II composite resin restorations. J Prosthet Dent 65:475479. Eick JD, Byerley TJ, Chappell RP, Chen GR, Bowles CQ, Chappelow CC (1993). Properties of expanding SOC/epoxy copolymers for dental use in dental composites. Dent Mater 9:123-127. Farah JW, Craig RG, Meroueh KA (1988). Finite element analysis of a mandibular model. J Oral Rehabil 15:615-624. Feilzer AJ, de Gee AJ, Davidson CL (1990). Quantitative determination of stress reduction by flow in composite restorations. Dent Mater 6:167-171. Fuks AB, Chosack A, Eidelman E (1990). Assessment of marginal leakage around Class II composite restorations in retrieved primary molars. Pediatr Dent 12:24-27. Hickman J, Jacobsen PH (1991). Finite element analysis of dental polymeric restorations. Clin Mater 7:39-43. Katona TR, Winkler MM (1994). Stress analysis of a bulk-filled Class V light-cured composite restoration. J Dent Res 73:1470-1477. Lai JH, Johnson AE (1993). Measuring polymerization shrinkage of photo-activated restorative materials by a water-filled dilatometer. Dent Mater 9:139-143. Leinfelder KF (1991). Using composite resin as a posterior restorative material. J Am Dent Assoc 122:65-70. Lui JL, Masutani S, Setcos JC, Lutz F, Swartz ML, Phillips RW (1987). Margin quality and microleakage of Class II composite resin restorations. J Am Dent Assoc 114:49-54. Lutz F, Krejci I, Oldenburg TR (1986a). Elimination of polymerization stresses at the margins of posterior composite resin restorations: a new restorative technique. Quint Int 17:777-784. Lutz F, Krejci I, Luescher B, Oldenburg TR (1986b). Improved proximal margin adaptation of Class II composite resin restorations by use of light-reflecting wedges. Quint Int 17:659-664. Mangum FL Jr, Berry EA, DeSchepper E, Rieger MR (1994). Microleakage of incremental versus compression matrix bulk filling of cervical resin composite restorations. Gen Dent 42:304-308. Rees JS, Jacobsen PH (1989). The polymerization shrinkage of composite resins. Dent Mater 5:41-44. Rees JS, Williams KR Jr, Jacobsen PH (1989). A review of the finite element method to model the restoration of posterior teeth with composite resin. In: Interfaces in medicine and mechanics. Williams KR, Lesser TH, editors. England: Elsevier Science Publishers, pp. 304-310. Sidhu SK (1993). A comparative analysis of techniques of

dimension in a preparation of this geometry. Both the B and WI techniques are easily duplicated clinically. The HI technique, which was included as a theoretical ideal for incremental placement, would be difficult and time-consuming to perform, since three uniform layers of equal depth would be required. The B technique simply requires a complete filling of the preparation. For the first two increments, the WI technique can be performed by merely wiping the typically viscous resin composite paste into the preparation and against one of the walls at the ends of the prism-shaped preparation. The third WI increment fills the remaining void. With all three filling techniques, the maximum residual stresses are the same. The actual residual stresses may be lower than these calculated maximum residual stresses, because the FE model was designed assuming no flow of uncured resin during polymerization. The differences among the techniques are the magnitudes of the transient stress. Since debonding can happen whenever the bond strength is surpassed, these high transient stresses can have important clinical implications. This study suggests that bulk-filled light-cured resins exhibit lower transient stresses and, as such, should be less likely to result in debonding. The finding that incremental techniques do not substantially reduce interfacial stresses is supported by another FEM study which analyzed residual stresses in a Class I cavity preparation (Rees et al., 1989). The present stress analysis may help to explain the results of studies which show that the incremental techniques do not eliminate microleakage (Fuks et al., 1990; Mangum et al., 1994). Based on our results with this prototype, the clinician may be advised to use a bulk filling technique where possible. Future model refinements are planned to enable both the preparation design and the placement technique to be optimized. An incremental technique is still needed for light-cured composite restorations which are too thick to cure in one increment.

Acknowledgments
The authors extend special thanks to Jian Huang for his assistance in FEM programing. This research was supported by NIDR Grant No. DE11054.

References
Airoldi RL, Krejci I, Lutz F (1992). In vitro evaluation of dentinal bonding agents in mixed Class V cavity preparations. Quint Int 23:355-362. Asmussen E (1975). Composite restorative resins. Composition versus wall-to-wall polymerization contraction. Acta Odontol Scand 33:337-344. Bandyopadhyay S (1982). A study of the volumetric setting shrinkage of some dental materials. J Biomed Mater Res
16:135-144. Bowen RL, Rapson JE, Dickson G (1982). Hardening shrinkage and hygroscopic expansion of composite resins. J Dent Res 61:654-658. Bowen RL, Nemoto K, Rapson JE (1983). Adhesive bonding of various materials to hard tooth tissues: forces developing

J Dent Res 75(7) 1996

Stress Analysis of Three Filling Techniques

1483

restoring cervical lesions. Quint Int 24:553-559. Sidhu SK, Henderson LJ (1992). Dentin adhesives and microleakage in cervical resin composites. Am I Dent 5:240-244. Suliman AA, Boyer DB, Lakes RS (1993). Cusp movement in premolars resulting from composite polymerization shrinkage. Dent Mater 9:6-10. Tajima K, Chen K, Terashita M, Kozono Y, Oshida Y, Moore BK

(1993). Occurrence of marginal enamel cracking in composite resin restored teeth (abstract). J Dent Res 72(Spec Iss) :136. Uno S, Asmussen E (1991). Marginal adaptation of a restorative resin polymerized at reduced rate. Scand J Dent Res 99:440-444. Watts DC, Amer 0, Combe EC (1984). Characteristics of visiblelight-activated composite systems. Br Dent J 156:209-215.