d e n t a l m a t e r i a l s 2 5 ( 2 0 0 9 ) 15761584

available at www.sciencedirect.com

journal homepage: www.intl.elsevierhealth.com/journals/dema

Inuence of curing protocol on selected properties of light-curing polymers: Degree of conversion, volume contraction, elastic modulus, and glass transition temperature
Magali Dewaele a,b, , Erik Asmussen c , Anne Peutzfeldt c , E. Christian Munksgaard c , Ana R. Benetti d , Gauthier Finn a , Gatane Leloup e , Jacques Devaux a
Laboratory of Chemistry and Physics of High Polymers, Universit catholique de Louvain, Louvain-la-Neuve, Belgium Department of Prosthodontics, School of Dentistry and Stomatology, Universit catholique de Louvain, Brussels, Belgium c Department of Dental Materials, School of Dentistry, University of Copenhagen, Copenhagen, Denmark d Department of Operative Dentistry, Endodontics and Dental Materials, Bauru School of Dentistry, University of So Paulo, Bauru SP, Brazil e Department of Operative Dentistry, School of Dentistry and Stomatology, Universit catholique de Louvain, Brussels, Belgium
b a

a r t i c l e
Article history:

i n f o

a b s t r a c t
Objectives. The purpose of this study was to investigate the effect of light-curing protocol on degree of conversion (DC), volume contraction (C), elastic modulus (E), and glass transition temperature (Tg ) as measured on a model polymer. It was a further aim to correlate the measured values with each other. Methods. Different light-curing protocols were used in order to investigate the inuence of energy density (ED), power density (PD), and mode of cure on the properties. The modes of cure were continuous, pulse-delay, and stepped irradiation. DC was measured by Raman

Received 28 October 2008 Received in revised form 3 August 2009 Accepted 4 August 2009

Keywords: Light-curing Soft-start Degree of conversion Volume contraction Elastic modulus Glass transition temperature Dental materials Dental polymers Resin composite

micro-spectroscopy. C was determined by pycnometry and a density column. E was measured by a dynamic mechanical analyzer (DMA), and Tg was measured by differential scanning calorimetry (DSC). Data were submitted to two- and three-way ANOVA, and linear regression analyses. Results. ED, PD, and mode of cure inuenced DC, C, E, and Tg of the polymer. A signicant positive correlation was found between ED and DC (r = 0.58), ED and E (r = 0.51), and ED and Tg (r = 0.44). Taken together, ED and PD were signicantly related to DC and E. The regression coefcient was positive for ED and negative for PD. Signicant positive correlations were detected between DC and C (r = 0.54), DC and E (r = 0.61), and DC and Tg (r = 0.53). Comparisons between continuous and pulse-delay modes of cure showed signicant inuence of mode of cure: pulse-delay curing resulted in decreased DC, decreased C, and decreased Tg . Inuence of mode of cure, when comparing continuous and step modes of cure, was more ambiguous.

Corresponding author at: Cliniques universitaires St-Luc, Ecole de mdecine dentaire et de stomatologie, avenue Hippocrate 10/bte5732, 1200 Bruxelles, Belgium. Fax: +32 2 764 90 62. E-mail address: Magali.Dewaele@uclouvain.be (M. Dewaele). 0109-5641/$ see front matter 2009 Academy of Dental Materials. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.dental.2009.08.001

d e n t a l m a t e r i a l s 2 5 ( 2 0 0 9 ) 15761584

1577

A complex relationship exists between curing protocol, microstructure of the resin and the investigated properties. The overall performance of a composite is thus indirectly affected by the curing protocol adopted, and the desired reduction of C may be in fact a consequence of the decrease in DC. 2009 Academy of Dental Materials. Published by Elsevier Ltd. All rights reserved.

1.

Introduction

It is widely admitted that the inuence of photopolymerization on the nal properties of resin composites is of major importance. Literature [15] has shown that the properties of a light-curing resin composite are mainly inuenced by the amount of energy delivered during irradiation. The total amount of energy per unit area, the so-called energy density (ED) is the product of the power per unit area (power density: PD) by the duration of irradiation (curing time). For a given ED, different combinations of PD and curing time may be used to cure the composite materials. Contradictory data have been published about the effect of variation in the combination of PD and exposure duration. Contrary to the total energy concept (referring to the reciprocity between power density and exposure duration), recent studies have demonstrated that for each ED, the combination of PD and exposure duration had a signicant inuence on degree of conversion, on extent of crosslinking and on physical properties [3]. The light from dental photocuring devices can be delivered in different modes: continuous, pulse-delay or stepped irradiation. The continuous mode delivers the same PD uninterruptedly throughout the entire exposure period. The pulse-delay mode initiates cure by a short ash of light followed by a delay of one or more minutes before the nal polymerization is performed. In the step-cure mode, a low PD is used during the rst part of the polymerization period and a higher PD is used towards the end of the irradiation. The pulsedelay and the stepped mode of cure are so-called soft-start modes of cure. The soft-start modes of cure were introduced with the purpose of slowing down the polymerization reaction, which is inevitably accompanied by contraction of the material [6]. Soft-start curing modes result in signicantly reduced gap formation in cavity margins because of contraction stress relief by ow [714]. However, few studies have evaluated the effect of curing mode on the volume contraction itself. Some authors found soft-start curing to combine increased marginal integrity with identical [15] or superior physical properties when compared to continuous curing [11]. Adversely, softstart techniques have resulted in either decreased degree of conversion (DC) as compared to the one of the continuous mode, or similar DC but at the same time in polymers of lower crosslink density, and thus of decreased mechanical properties [1619]. As hypothesized in a previous study [17], it is conceivable, that the different light-curing modes will lead to polymers of different network structure, even though the DC is the same. The DC does not give a complete characterization of polymer structures because polymers of similar DC may have different extent of crosslinking [1618]. This may occur as a result of

the soft-start modes of cure (pulse or step-cure mode) probably due to the formation of relatively fewer growth centers, which may result in a more linear polymer, with decreased crosslinking. The increased concentration of crosslinks has been associated with increased physical properties and stability of polymers [16,20]. It has been shown [20,21] that the extent of crosslinking of a polymer may be assessed by measurements of the glass transition temperature (Tg ). Tg is an important parameter for polymer characterization as it marks a region of dramatic changes in the physical properties of the polymer. The Tg value represents the temperature region at which the polymer is transformed from a glassy material into a rubberlike one. Crosslinking reduces molecular mobility and thus gives rise to increased apparent Tg [22]. The curing procedure may inuence the regularity of the network and the crosslink density, and this may be reected in the Tg . To improve the overall performance of light-cured dental polymers, a detailed understanding of the effects of mode of irradiation on properties and how these properties are related to each other is necessary. The present study represents an effort in this direction. It was the aim of the study to determine, on the same experimental dental polymer, degree of conversion, volume contraction, elastic modulus, and glass transition temperature in relation with the curing protocol. The working hypothesis was that the total energy density and, for each level of ED, that the power density and the mode of cure, have an effect on the selected properties. A further aim was to attempt to correlate the data obtained for these properties with each other. In a further study [23] the data are correlated with softening and elution of monomers in ethanol as measured on identical polymers subjected to the same curing protocols.

2.

Materials and methods

The experimental resin used in this study was composed of Bis-GMA (bisphenol-A-glycidyl dimethacrylate, Heraeus Kulzer, Germany) and TEGDMA (95% triethyleneglycol dimethacrylate, Aldrich, Belgium) at a molar ratio of 1:1, along with 0.5 wt% CQ (97% camphorquinone, Aldrich) as visible light initiator and 0.5 wt% DABE (97% N,N-dimethyl-paminobenzoic acid ethylester, Aldrich) as co-initiator. Cylindrical specimens were fabricated in a brass mold. The specimens were covered on both sides with a transparent lm (Mylar) and irradiated from one side only. Specimens used for degree of conversion and contraction measurements (height: 2 mm, diameter: 5 mm) and specimens used for elastic modulus and glass transition temperature measurements (height: 1 mm, diameter: 3 mm) were of different size as dictated by the dimension requirement of the measuring device.

1578

Table 1 Investigated curing protocols. Degree of conversion (DC), volume contraction (C), elastic modulus (E), and glass transition temperature (Tg ). Means (S.D.). Irradiation (1st step) time @ power density
Continuous A B C D E F Pulse-delay G H I J K L M N O P Q R Step S T U V X Y 40 s @ 150 mW/cm2 20 s @ 300 mW/cm2 10 s @ 600 mW/cm2 80 s @ 150 mW/cm2 40 s @ 300 mW/cm2 20 s @ 600 mW/cm2 1 s @ 300 mW/cm2 1 s @ 300 mW/cm2 1 s @ 300 mW/cm2 1 s @ 300 mW/cm2 1 s @ 300 mW/cm2 1 s @ 300 mW/cm2 1 s @ 600 mW/cm2 1 s @ 600 mW/cm2 1 s @ 600 mW/cm2 1 s @ 600 mW/cm2 1 s @ 600 mW/cm2 1 s @ 600 mW/cm2 10 s @ 300 mW/cm2 10 s @ 150 mW/cm2 10 s @ 150 mW/cm2 10 s @ 300 mW/cm2 10 s @ 150 mW/cm2 10 s @ 150 mW/cm2 19 s @ 300 mW/cm2 19 s @ 300 mW/cm2 19 s @ 300 mW/cm2 39 s @ 300 mW/cm2 39 s @ 300 mW/cm2 39 s @ 300 mW/cm2 9 s @ 600 mW/cm2 9 s @ 600 mW/cm2 9 s @ 600 mW/cm2 19 s @ 600 mW/cm2 19 s @ 600 mW/cm2 19 s @ 600 mW/cm2 5 s @ 600 mW/cm2 15 s @ 300 mW/cm2 7.5 s @ 600 mW/cm2 15 s @ 600 mW/cm2 35 s @ 300 mW/cm2 17.5 s @ 600 mW/cm2

Delay (min)

Irradiation (2nd step) time @ power density

Energy density (mJ/cm2 )

6000 6000 6000 12000 12000 12000

DC (%)

C (%)

E (GPa)

Tg ( C)

69 (2.9) 68 (2.6) 65 (0.5) 71 (2.0) 72 (3.3) 69 (7.4)

7.52 (0.04) 7.49 (0.02) 7.54 (0.15) 7.63 (0.15) 7.62 (0.15) 7.69 (0.06)

2.9 (0.14) 2.7 (0.09) 2.5 (0.15) 2.7 (0.18) 2.8 (0.23) 2.7 (0.17)

45.0 (0.85) 46.8 (0.54) 45.2 (0.08) 45.3 (1.63) 47.5 (0.86) 46.6 (1.10)

d e n t a l m a t e r i a l s 2 5 ( 2 0 0 9 ) 15761584

1 2 3 1 2 3 1 2 3 1 2 3

6000 6000 6000 12000 12000 12000 6000 6000 6000 12000 12000 12000

62 (1.5) 63 (0.4) 61 (1.3) 67 (1.8) 64 (3.2) 66 (1.8) 60 (4.0) 61 (3.9) 61 (1.1) 65 (2.5) 65 (0.8) 65 (4.2)

7.35 (0.28) 7.26 (0.27) 7.30 (0.06) 7.35 (0.24) 7.31 (0.34) 7.47 (0.31) 7.40 (0.17) 7.29 (0.20) 7.57 (0.08) 7.43 (0.06) 7.42 (0.08) 7.58 (0.09)

2.5 (0.05) 2.5 (0.05) 2.3 (0.09) 2.8 (0.37) 2.8 (0.17) 2.6 (0.26) 2.5 (0.26) 2.4 (0.14) 2.4 (0.02) 2.6 (0.07) 2.8 (0.10) 2.6 (0.31)

45.4 (0.91) 44.9 (1.21) 45.6 (0.75) 46.4 (0.38) 45.1 (0.89) 46.3 (1.26) 43.8 (1.09) 45.0 (0.67) 45.0 (0.16) 47.0 (0.90) 45.7 (0.95) 45.8 (2.06)

6000 6000 6000 12000 12000 12000

61 (3.3) 64 (0.7) 69 (5.0) 71 (2.6) 72 (1.6) 69 (2.7)

7.54 (0.06) 7.61 (0.02) 7.49 (0.11) 7.63 (0.02) 7.46 (0.26) 7.73 (0.03)

2.6 (0.03) 2.6 (0.10) 2.7 (0.23) 2.9 (0.09) 2.5 (0.22) 3.0 (0.26)

46.8 (0.52) 46.4 (0.51) 46.1 (1.24) 46.0 (1.43) 47.3 (1.65) 46.0 (1.04)

d e n t a l m a t e r i a l s 2 5 ( 2 0 0 9 ) 15761584

1579

The light-curing protocols shown in Table 1 were followed in order to investigate the inuence of the energy density, the power density, and the mode of cure on the selected properties. For the pulse-delay groups, different delay periods were used. Additionally, a short initial period of light exposure (1 s) was adopted in the rst cycle of light-activation because this pulse-delay protocol demonstrated previously to reduce gaps in the margins of restorations [14]. For the step-cured groups, the use of higher or lower power densities in the beginning or in the end of the irradiation was also examined. The different power densities were obtained by changing the voltage delivered to the quartztungstenhalogen curing unit (Optilux 501, Kerr Corporation, Danbury, CT, USA), by means of a variable external resistor connected in series (Autralec Brussels, Belgium). The desired power densities were monitored with a dental curing radiometer (Optilux model 100, Demetron Research Corporation/Kerr, Danbury, CT, USA).

2.3.

Elastic modulus (E)

2.1.

Degree of conversion (DC)

Elastic modulus of the dental resins may be measured in tension, in exion, or in shear. In this work, due to the very low dimensions of the samples, shear modulus was measured by use of a dynamic mechanical analyzer (DMA/SDTA861e, Mettler Toledo, Greifensee, Switzerland). Tensile and shear elastic modulus may be designated E and G, respectively. E and G are linked by the Poisson coefcient according to E = 2(1 + )G. For most isotropic materials, is between 0.2 and 0.5. For such materials, E is equal to 2.4 to 3 times G. In the present study, the measured values of G were multiplied by 3 to give E. Six specimens were fabricated for each experimental condition and were stored dry at ambient temperature for at least 7 days before testing. Two identical specimens were placed symmetrically between shear clamps. Samples were analyzed in shear mode isothermally at 25 C with a preloading force of 6 N, a displacement amplitude limit of 0.5 m, and a frequency of 1 Hz.

The DC was measured by means of Raman micro-spectroscopy (Labram, Dilor Horiba-Jobin-Yvon, Lille, France) as previously described [6]. In brief, this method allows the evaluation of the DC by comparing the vibration bands of the residual unpolymerized methacrylate C C stretching mode at 1640 cm1 to the aromatic C C stretching mode at 1610 cm1 used as internal standard. The unpolymerized resin was used as reference (DC = 0). Three specimens were fabricated for each experimental condition and were stored dry at ambient temperature for at least 7 days. A notch was made to identify the side irradiated by the light. For each specimen, the irradiated side was facing the microscope objective. The degree of conversion was determined at three locations and the mean value calculated for each of the three specimens. From these three mean values a new mean value was calculated.

2.4.

Glass transition temperature (Tg )

2.2.

Volume contraction (C)

The same three samples of cured resin used to determine the DC were used to measure the volume contraction. The contraction ( V/Vunpolymerized ) was obtained by comparing unpolymerized (dunpolymerized ) to polymerized (dpolymerized ) specic masses and expressed as volume percent according to the formula [6]: V Vunpolymerized dpolymerized dunpolymerized dpolymerized

The glass transition temperature (Tg ) was measured by Differential Scanning Calorimetry (DSC822e/HSS7, Mettler Toledo, Greifensee, Switzerland). Four of the six samples of cured resin used to determine the modulus were used to measure the glass transition temperature. Samples were analyzed at a heating rate of 10 C/min from 20 C to 70 C under nitrogen. The Tg -value represents the temperature region at which (the amorphous phase of) a polymer is transformed from a brittle, glassy material into a tough rubberlike one. This effect is accompanied by a step-wise increase of the DSC heat ow/temperature or specic heat/temperature curve. Thus, Tg was determined from an endothermic parallel transition of the baseline [22]. An exothermic reaction occurred within the Tg region and superimposes on the heat ow shift. This exothermic reaction is due to post-polymerization occurring due to increased chain-segment mobility in the polymer at temperatures above Tg . The temperature at the intersection of the extrapolated heat ow curve, at the low temperature end, and the tangent of the ascending curve, at the inection point, was taken as the Tg (Fig. 1).

2.5.

Statistical methods

(%) = 100

The specic mass of uncured resin (liquid) was measured by pycnometry. The specic mass of the cured resin (solid) was determined by a density column (Daventest Instruments, Fareham, UK). In the density column technique, samples were immersed in a liquid having a density gradient. The liquid was prepared by mixing potassium bromide and water to reach an appropriate density range (between 1.00 and 1.41), which depends on the anticipated spread of densities in the samples to be measured. The height at which samples stabilized was observed and compared with calibrated marker oats [6].

Bartletts test was used to check the homogeneity of variances. The data were submitted to analysis of variance (ANOVA), to Newman Keuls multiple comparison test and to linear regression analysis. The level of signicance was dened as = 0.05.

3.

Results

The degree of conversion (DC), the volume contraction (C), the elastic modulus (E) and the glass transition temperature (Tg ) for each of the 24 irradiation conditions are shown in Table 1. For an energy density (ED) of 6000 mJ/cm2 , one-way ANOVA showed statistically signicant differences between the 12 mean values of DC, E and Tg , respectively. However, when

1580

d e n t a l m a t e r i a l s 2 5 ( 2 0 0 9 ) 15761584

Comparisons of the continuous and the stepped mode of cure showed a signicant increase in C with an increasing ED (p < 0.05), except when step 1 = 150 mW/cm2 and step 2 = 300 mW/cm2 .

3.3.

Elastic modulus (E)

Fig. 1 Heat ow shift of the experimental dental polymer. Endothermic parallel transition of the baseline and glass transition temperature. Superimposed exothermic reaction.

Comparisons of the continuous and the pulse-delay mode of cure showed signicant inuence of ED (p < 0.05). Increased ED resulted in increased E. Comparisons of the continuous and the stepped mode of cure showed signicant inuence of ED (p < 0.05) when step 1 = 300 mW/cm2 and step 2 = 600 mW/cm2 . An increased ED resulted in increased E. PD of the rst step being constant, an increased PD of the second step resulted in a signicantly increased E (p < 0.05).

3.4.

Glass transition temperature (Tg )

ED was 12000 mJ/cm2 , a signicant difference was detected only for DC. Regarding the volume contraction, no statistically signicant differences were found with one-way ANOVA between the 12 mean values neither at ED = 6000 mJ/cm2 nor at ED = 12000 mJ/cm2 . Linear two- and three-dimensional regression analyses were used to search for signicant relationships between energy density, power density (PD), DC, C, E and Tg . Twodimensional analysis showed signicant although weak relationships between DC and C (r = 0.54), DC and E (r = 0.61), and DC and Tg (r = 0.53), between C and E (r = 0.50), and C and Tg (r = 0.34). Furthermore, signicant relationships were found between ED and DC (r = 0.58), ED and E (r = 0.51), and ED and Tg (r = 0.44). Three-dimensional regression analysis showed a signicant positive correlation between ED and DC and between ED and E and a concomitant signicant negative correlation between PD and DC and between PD and E. Three-way ANOVA and Newman Keuls multiple comparison test were used to analyze the inuence of ED, PD, and mode of cure on DC, C, E and Tg .

Comparisons of the continuous and the pulse-delay mode of cure showed signicant inuence of ED (p < 0.05), PD (p < 0.05) and mode of cure (p < 0.05). Increased ED resulted in increased Tg . Within a given ED, an increased PD showed a decreased Tg . The pulse-delay mode resulted in decreased Tg . The length of the delay was not signicant (p > 0.05). Comparisons of the continuous and the stepped mode of cure did not detect a signicant difference between the groups (p > 0.05).

4.

Discussion

3.1.

Degree of conversion (DC)

Comparisons of the continuous and the pulse-delay mode of cure showed signicant inuence of ED (p < 0.05) and mode of cure (p < 0.05). Increased ED resulted in increased DC. The pulse-delay mode resulted in decreased DC. The length of the delay was not signicant (p > 0.05). Comparisons of the continuous and the stepped mode of cure showed an increased DC with an increasing ED, although not signicant when step 1 = 150 mW/cm2 and step 2 = 600 mW/cm2 .

3.2.

Volume contraction (C)

Comparisons of the continuous and the pulse-delay mode of cure showed signicant inuence of mode of cure (p < 0.05). The pulse-delay mode resulted in decreased C. The length of the delay was not signicant (p > 0.05).

In the present study, resin composites were polymerized according to three different curing modes: continuous, pulsedelay, and stepped. These modes included variations in ED, and power density (PD), and for the pulse-delay mode different lengths of delay. The foregoing has shown that ED, PD, and mode of cure inuenced the investigated properties: degree of conversion (DC), volume contraction (C), elastic modulus (E) and glass transition temperature (Tg ). Thus, the working hypothesis was validated. The results conrm that the properties of a light-curing polymer-based material are mainly inuenced by the amount of energy delivered to the material during irradiation. The higher ED applied to the material, the higher were DC, E and Tg . This is in agreement with the results of Halvorson et al. [2] and Peutzfeldt and Asmussen [3] who showed that DC and E increased with increasing ED. The increase in E and Tg with energy density is caused by the increase in DC and the signicant relationship between DC and E, and between DC and Tg . However, in this study, the correlation between ED and C was found to be not signicant, although DC and C were signicantly correlated. This point does not agree with other studies [4,5] and may be due to a type II statistical error. The measured values of DC, E and Tg were statistically different at 6000 mJ/cm2 although only the values of DC were statistically different at 12000 mJ/cm2 . A tentative explanation could be that DC, E and Tg increase with increasing ED

d e n t a l m a t e r i a l s 2 5 ( 2 0 0 9 ) 15761584

1581

but tend towards a maximum. E and Tg may have a tendency to level off earlier than DC. At 6000 mJ/cm2 , the material has not reached its maximum properties for DC, E and Tg , and the polymerization rate has an inuence on these properties. At 12000 mJ/cm2 , E and Tg have reached their maximum, and the polymerization rate has less inuence on these properties. As mentioned in the introduction, soft-start favors reduced gap formation [714]. However, the inuence of curing modes on the mechanical and physical properties remains conicting. For a given monomer composition, the mechanical and physical properties of the resulting polymer are to a large extent governed by the two interrelated but independent fundamental properties of network structures: degree of conversion and degree of crosslinking. In this study, in order to understand how the curing modes inuence the properties, the underlying kinetics and mechanisms of the reaction and the structural characteristics must be understood. Therefore, this paragraph discusses the most relevant aspects related to the polymerization reaction and speculates the possible outcomes for the different light-curing protocols investigated. Owing to the high similarity of the samples (nature, volume, shape) it is rst assumed that the T reached by all samples at the end of polymerization (vitrication T ) does not differ signicantly from one sample to another. The effect of heating by the light source is thus never taken into account in this work. The characterization of the crosslinked network structure is extremely difcult due to the structural heterogeneity that develops in the polymer during polymerization. A rst explanation for the heterogeneity of the network may be related to the rate of photopolymerization. The rate of polymerization is the speed at which DC is growing with time. In classical free radical processes (assuming conditions of radical stationarity), the polymerization rate scales as the square root of the rate of initiation. At low PD, a quasi-linear relationship can be assumed between PD and the rate of initiation. Thus, in classical photopolymerization, at low PD, the polymerization rate scales as the square root of PD. However, conditions of radical stationarity are almost never met in the present bulk photopolymerization, hence the square-root law is not valid anymore. Moreover, at higher PD a premature termination of radicals can occur if radicals just created react with each other, without initiating polymerization. This leads to a net loss of radicals. Kinetically, it can be demonstrated that the rate of this premature termination reaction depends on the square of the concentration of radicals. At higher PD, the relationship between PD and the rate of initiation becomes thus less than linear. Thus, at high PD, the relationship between PD and rate of polymerization becomes

highly nonlinear, with a power law smaller than the squareroot law. In simpler words, at high PD, for a given time of irradiation, increasing PD will lead to a much lower increase of DC. The case of dimethacrylate monomers with two equivalent functional (vinyl) groups is still more complicated as the functional group reactivity apparently changes with conversion in the system. This is also linked to the microstructure of the network with a large amount of pendant functional groups. To understand the polymer network, reference is made to the denitions of cyclic structures proposed by Dotson et al. [24]. When a radical involving a pendant double bond on a polymer chain propagates (i.e. a radical formed at a double bond from a unit having one double bond already reacted), a primary cycle, secondary cycle or crosslink, can be formed. When the radical reacts with a pendant double bond on its own chain, due to the vicinity in the close neighborhood, a primary cycle results. Primary cycles do not contribute to the network but trap a few pendant vinyl groups leading to microgel zones. If the radical reacts with a pendant double bond on a different chain but with which it is already (cross)linked, a secondary cycle forms. Finally, a crosslink forms when the radical reacts with a pendant double bond on a different chain [21]. Secondary cycles and crosslinks appear consecutively and are rather undistinguishable from each other. Only the crosslinks connecting different chains leading to network formation will enhance the mechanical properties of the material. However, primary and secondary cycles, as well as crosslinks, contribute to the degree of conversion. As only crosslinks inuence Tg , this explains that polymers with the same DC can exhibit different Tg s. The apparent unequal reactivity of the functional groups and the ensuing competition between primary cyclization on the one hand, and secondary cyclization and crosslinking on the other hand, are a second explanation for structural heterogeneity of the network. Indeed, in case of divinyl monomers, with two equivalent functional (vinyl) groups, the functional group reactivity apparently changes with conversion in the system. At low conversions, the few pendant functional groups neighboring an active radical chain end exhibit an apparent enhancement of their reaction rate, due to their spatial proximity with the propagating radicals, leading to primary cycles (Fig. 2). Therefore, initially, primary cyclization dominates crosslinking and secondary cyclization. This behavior accounts for the formation of a rst type of microgel regions and a rst kind of heterogeneity of the network.

Fig. 2 Schematic picture showing how primary cycles are formed and how they trap vinyl groups in microgel zones (in grey).

1582

d e n t a l m a t e r i a l s 2 5 ( 2 0 0 9 ) 15761584

As conversion increases, two phenomena may contribute to the formation of a second type of microgels: the diffusion and the concentration effects. Diffusion effect. When the increasing viscosity limits the rate of termination, a rapid gelation (the so-called Trommsdorffs effect) takes place locally. A constant amount of free radicals are still created by the illumination, but the mobility of growing radicals decreases and they cannot nd other radicals to terminate with as easily as they can nd other monomers to propagate with. In such a case, rst, the concentration of free radicals increases leading to an increase in overall reaction rate. The diffusion of radicals controls the kinetics, and this, clearly locally consumes the double-bonds belonging to mobile free (unreacted) monomers [21]. Concentration effect. At this stage, when the local viscosity increases, the concentration of free monomers decreases locally, and owing to the increase of radical concentration, pendant functional groups are more and more involved in the reaction leading to local microgel structure to form. This, in turn, decreases the diffusivity of free monomers. Thus, as conversion increases, when the reaction becomes limited by the diffusion of free monomers, the amount of secondary cycles then of crosslinks increases, a network forms and as this occurs locally, a second type of microheterogeneities occurs. All those phenomena may be understood in terms of a three-term kinetic equation: R = k1 [R ][MF ] + k2 [R ][MP ] + k3 [R ] In this equation: R is the propagation rate; [R] is the concentration of radicals; [MF ] is the concentration of vinyls included in free monomer; [MP ] is the concentration in pendant vinyls on another chain (accessible to radicals). In the very rst stage of the reaction (very low conversion), only the rst term takes a signicant value and the chains grow linearly. The third term, which takes into account the formation of primary cycles should depend on both the number of radicals and the number of pendant vinyls on the same chain, but the number of pendant vinyl groups in the volume neighboring the radical (i.e. their concentration) is almost constant, thus the rate of primary cyclization (third term) only depends on the concentration (number) of radicals. This term does not contribute to crosslinking, but to DC and, due to microgel formation, decreases the concentration of pendant vinyls accessible to other radicals [MP ]. The second term accounts for the reaction of pendant vinyls from other chains, thus for secondary cyclization as well as crosslinking. It depends on pendant vinyls accessible to other radicals [MP ]. As regards the effect of power density at a given energy density, higher PD resulted in lower DC1 and subsequently

This result (lower DC for higher PD) supports the hypothesis that there is no signicant difference of nal (vitrication) T of samples. Indeed, should it be the case, the reverse results (higher DC for higher PD) should be expected.

lower E. This result is to link with the rate of polymerization. It is conceivable that the faster initiation rates, at high power density, will produce higher concentrations of radicals that will rapidly react with each other, thus yielding premature termination of the polymerization. Premature termination of radicals can occur if radicals just created react with each other, without initiating polymerization. A slower initiation, at lower PD, will proportionally yield a higher efciency of the initiated radicals. This results in a delayed termination of polymerization leading to a lower quantity of trapped, unreacted monomers and higher DC. In the pulse-delay mode of cure, the radicals produced during the pulse will rapidly react with each other but immediately after the pulse, the propagation of the polymerization will continue (at very low concentration of radicals) until a certain conversion within the delay period. In this study, no signicant differences were observed between the delay lengths, what leads to think that this phenomenon takes less than 1 min. This is in apparent contradiction with a previous study of Asmussen and Peutzfeldt [17], which noticed the inuence of different lengths of delay in the softening of polymers in ethanol. It is likely that this contradiction results from the different experimental methods used [23]. The nal irradiation will produce a high concentration of radicals, but due to the premature radical termination and the lower [R] during the [pulse + delay] period, the growing radicals lead to microgel regions (IR cycles) formed during the delay period of decreased mobility, thus preventing the effective reaction with monomers when compared to a continuous irradiation. This leads to a decreased DC in the pulse-delay mode of cure compared to the continuous mode. The decreased contraction in the pulse-delay mode is caused by the positive correlation between DC and C. With regards the effect of PD on the crosslinking, it has been speculated that a slow start of the polymerization, at low PD, may result in relatively fewer growth centers, which may result in a more linear polymer structure, when compared to a high rate of polymerization, promoted by the a high power density with increased crosslinks [16,25]. Such speculations are rather well supported by the above equation: With slow start of the polymerization, and at low PD, [R] and [MP ] are both low. The second term is rather low (as compared with rst and third ones) thus more linear chains are formed together with primary cycles. The resulting polymer structure is preferably linear. At high rate of polymerization (high PD), [R] is high (thus the propagation is faster and [MP ] grows). This decreases the relative importance of the third term (primary cyclization). But with time, as the rate of polymerization is high (and increases), it can become higher than the diffusion rate of free monomers (diffusion control of the propagation reaction), and [MF ] will decrease. The second term takes therefore more importance, and both secondary cycles and crosslinking occur at a higher rate. Should this phenomenon occur locally, microgels form. Actually, distinction has to be made between continuous, step-cure and pulse-delay mode. For the continuous and the step-cure mode, high PD (of the rst step) will give rise to multitude of growth centers, increasing the tendency to form a branched polymer (increased degree of crosslinking). Lower PD

d e n t a l m a t e r i a l s 2 5 ( 2 0 0 9 ) 15761584

1583

will give rise to relatively few growth centers, and a relatively linear polymer is formed. Comparisons of the continuous and the pulse-delay mode of cure did not show a signicant inuence on E, despite a decreased DC with the pulse-delay mode and a positive correlation between DC and E. Actually, from Table 1, the pulse-delay mode of cure shows a decreased E compared to the continuous mode when ED = 6000 mJ/cm2 . No differences are noticeable at ED = 12000 mJ/cm2 because, as previously explained, E has the tendency to level off earlier than DC and has already reached its maximum. In this study, it was assumed that Tg was a good parameter to determine the crosslink density. The pulse-delay technique led to polymers of decreased Tg . The decreased Tg may be interpreted as the manifestation of a polymer structure having fewer crosslinks. In the pulse-delay curing mode, as already explained, after the short ash of initial light, only few growth centers remain active during the delay. Consequently, the propagation of polymerization will predominantly add one molecule of monomer after the other to the growing polymer chain. Also, the concentration of pendant monomers accessible to reaction [MP ] decreases. The result is that a relatively linear polymer is formed before the nal cure is carried out. The nal irradiation will give rise to a multitude of growth centers. Many growth centers will increase the tendency to form a branched polymer. At the end, the pulse-delay mode results in more linear chains, with proportionally decreased crosslinking. The nding that the polymers that were precured with a pulse at 600 mW/cm2 exhibited a lower Tg than the corresponding polymers precured at 300 mW/cm2 is probably linked to the propagation of the polymerization. At a pulse PD of 300 mW/cm2 relatively few growth centers are formed, with the result that a larger part of the polymerization occurs at the nal cure. At a pulse PD of 600 mW/cm2 , many more growth centers are formed, with the result that a larger part of the polymer structure will be linear before nal polymerization. Indeed, the pulse of light initiates a concentration of radicals [R] increasing very fast, then decaying (exponentially) during the delay. On the average, during the [pulse + delay] duration, a low [R] is active during the period of time that depends on the PD during the pulse. These results are in accordance with the ones of Asmussen and Peutzfeldt [17,18]. In their studies, the pulse-delay curing resulted in less crosslinked polymers compared to continuously cured polymers. The extent of crosslinking was estimated by the Wallace hardness after storage in ethanol. The softening effect also increased with the power density of the initial exposure. To conclude, the pulse-delay mode of cure has earlier been found to result in reduced gap formation [9,14]. Concerning the structure of the polymers, the pulse-delay mode resulted in decreased degree of conversion and in decreased degree of crosslinking, since it was assumed that Tg was a good parameter of the crosslink density. A signicantly lower DC and crosslink density may have possible consequences for the stability of the polymer (for example elution and softening by food components). To conrm the results obtained in this study, determination of crosslink density by measurements of softening after storage in ethanol and elution were addressed in a further study [23].

Comparisons of the continuous and the stepped mode of cure are more ambiguous. A possible explanation is that, in the step-cure mode, PD of the rst step will have effects on degree of conversion and on density of crosslinking in the opposite direction. A high PD will produce higher concentrations of radicals that will rapidly react with each other, thus yielding premature termination and resulting in lower DC. But, high PD will give rise to a multitude of growth centers, increasing the tendency to form a branched polymer (higher crosslink density). On the other side, a lower PD in the rst step will proportionally yield a higher efciency of the initiated radicals. This results in a delayed termination of polymerization and in higher DC. But lower PD will give rise to relatively few growth centers, and a relatively linear polymer is formed. This explanation is in accordance with other studies [11,17,18], which found that, depending of the PD and the duration of the rst step, step-cured composites resulted in either lower, or higher crosslinking or modulus of elasticity. Thus, the difculty lies in determining the optimal duration and PD of the two steps. In conclusion, a step-cure mode has also been found to result in reduced contraction gaps [13] although not signicant in all studies [26]. Further investigations are necessary to nd the optimal power density and duration of the steps to allow, on the one hand, enough ow to increase marginal integrity, and, on the other hand, to activate the optimal amount of radicals to form a highly crosslinked polymer, with a high degree of conversion, and consequently to obtain similar or even better physical properties. It must be remembered, though, that the optimal power density actually depends on the concentration and also on the specic chemistry of the initiator system [11].

Acknowledgements
This study was supported by a FIRST-Europe research grant of the Walloon Region (Belgium) and by the European social funds. The authors also gratefully acknowledge Thrse Glorieux, Pascal Van Velthem and Jean-Jacques Biebuyck for technical assistance.

references

[1] Yap AUJ, Seneviratne C. Inuence of light energy density on effectiveness of composite cure. Oper Dent 2001;26:4606. [2] Halvorson RH, Erickson RL, Davidson CL. Energy dependent polymerization of resin-based composite. Dent Mater 2002;18:4639. [3] Peutzfeldt A, Asmussen E. Resin composite properties and energy density of light cure. J Dent Res 2005;84:65962. [4] Asmussen E, Peutzfeldt A. Polymerization contraction of resin composite vs. energy and power density of light-cure. Eur J Oral Sci 2005;113:41721. [5] Calheiros FC, Kawano Y, Stansbury JW, Braga RR. Inuence of radiant exposure on contraction stress, degree of conversion and mechanical properties of resin composites. Dent Mater 2006;22:799803. [6] Dewaele M, Trufer-Boutry D, Devaux J, Leloup G. Volume contraction in photocured dental resins: the shrinkage-conversion relationship revisited. Dent Mater 2006;22:35965.

1584

d e n t a l m a t e r i a l s 2 5 ( 2 0 0 9 ) 15761584

[7] Silikas N, Eliades G, Watts DC. Light intensity effects on resin-composite degree of conversion and shrinkage strain. Dent Mater 2000;16:2926. [8] Uno S, Asmussen E. Marginal adaptation of a restorative resin polymerized at reduced rate. Scand J Dent Res 1991;99:4404. [9] Yoshikawa T, Burrow MF, Tagami J. A light curing method for improving marginal sealing and cavity wall adaptation of resin composite restorations. Dent Mater 2001;17:35966. [10] Yap AU, Soh MS, Han TT, Siow KS. Inuence of curing lights and modes on cross-link density of dental composites. Oper Dent 2004;29:4105. [11] Mehl A, Hickel R, Kunzelmann KH. Physical properties and gap formation of light-cured composites with and without softstart-polymerization. J Dent 1997;25:32130. [12] Watts DC, Al Hindi A. Intrinsic soft-start polymerisation shrinkage-kinetics in an acrylate-based resin-composite. Dent Mater 1999;15:3945. [13] Davidson CL, Feilzer AJ. Polymerization shrinkage and polymerization shrinkage stress in polymer-based restoratives. J Dent 1997;6:43540. [14] Saha A, Peutzfeldt A, Asmussen E. Effect of pulse-delay curing on in vitro wall-to-wall contraction of composite in dentin cavity preparations. Am J Dent 2001;14:2956. [15] Emami N, Soderholm KJM, Berglund LA. Effect of light power density variations on bulk curing of dental composites. J Dent 2003;31:18996. [16] Asmussen E, Peutzfeldt A. Inuence of selected components on crosslink density in polymer structures. Eur J Oral Sci 2001;108:2825.

[17] Asmussen E, Peutzfeldt A. Inuence of pulse-delay curing on softening of polymer structures. J Dent Res 2001;80:1570 3. [18] Asmussen E, Peutzfeldt A. Two-step curing: inuence on conversion and softening of a dental polymer. Dent Mater 2003;19:46670. [19] Soh MS, Yap AUJ. Inuence of curing modes on crosslink density in polymer structures. J Dent 2004;32:3216. [20] Tamareselvy K, Rueggeberg F. Dynamic mechanical analysis of two crosslinked copolymer systems. Dent Mater 1994;10:2907. [21] Anseth S, Bowman CN. Kinetic gelation model predictions of crosslinked polymer network microstructure. Chem Eng Sci 1994;49:220717. [22] Groenewoud W. Characterization of polymers by thermal analysis. Elsevier; 2001. pp. 245252. [23] Benetti AR, Asmussen E, Munksgaard EC, Dewaele M, Peutzfeldt A, Devaux J, Leloup G. Softening and elution of monomers in ethanol. Dent Mater 2009;25:100713. [24] Dotson NA, Macosko CW, Tirrell M. Cyclization during crosslinking free-radical polymerizations. In: Aharoni SM, editor. Synthesis, characterization, and theory of polymeric networks and gels. Plenum Press; 1992. p. 31936. [25] Hofmann N, Markert T, Hugo B, Klaiber B. Effect of high intensity vs. soft-start irradiation on light-cured resin-based composites. Part II: hardness and solubility. Am J Dent 2004;17:3842. [26] Saha A, Peutzfeldt A, Asmussen E. Soft-start polymerization and marginal gap formation in vitro. Am J Dent 2001;14:2956.