The Combination of CQ-amine and TPO Increases the Polymerization Shrinkage Stress and Does Not Improve the Depth of Cure of Bulk-fill Composites

Bulk-fill composites (BFCs) are light-cured, resin-based materials used for the direct restorations of posterior teeth that can be placed in increments up to 4 to 5 mm thick.¹  The higher depth of cure of these composites in comparison to regular resin composites can be related not only to modifications in their monomers and filler compositions but also in the photoinitiator system used. As an example, some manufacturers use some alternative photoinitiators, such as diphenyl (2,4,6-trimethylbenzoyl) phosphine oxide (TPO) and benzoyl germanium (Ivocerin) in combination with camphorquinone (CQ), claiming that this combination improves the depth of cure.^2-4 

One possible explanation is that these alternative photoinitiators are more reactive than CQ, and they also produce more free radicals capable of initiating the polymerization, thus increasing the degree of conversion.⁵  However, these alternative photoinitiators generally absorb violet light, which has a lower transmittance compared with blue light; thus, they might increase the degree of conversion only on the top of the restoration and might not increase the depth of cure of composites.⁶  While monomers (type and size), fillers (load and size), and pigments might act as competitive absorptive components for both violet and blue light transmission throughout resin-based materials, it remains unclear whether the photoinitiator system absorption as an isolated factor has a significant effect on BFC light transmission and depth of cure. One question that needs to be asked is whether the increase in degree of conversion on the top would move the refractive index of the polymer formed closer to that of the filler used.^7,8  This would reduce the mismatch between the resin and filler refractive indexes, improving the transmission of blue light to the deeper portions of the restoration, thus enhancing free-radical formation by the CQ-amine photoinitiator system at the bottom part of the restoration. This would then increase the depth of cure as claimed.

On the other hand, in vitro studies have shown that BFC restorations might fail due to internal gap formations or by enamel cracks formed in nearby restoration margins,⁹  possibly correlated to the polymerization shrinkage stress of the composite bulk.¹⁰  Moreover, clinical evidence has shown that BFCs containing CQ-amine combined with TPO had worse clinical performance than BFCs containing only CQ-amine as the photoinitiator system. Interestingly, what is striking in the study of Bayraktar and others¹¹  is that CQ-amine combined with TPO showed more incidence of discontinued anatomic form, marginal discoloration, crevices in which dentin was exposed, and secondary caries in just one year of observation. Therefore, it is possible that the combination of these photoinitiators might not only improve the depth of cure but also play a role in the polymerization shrinkage stress of BFCs with a given resin/filler configuration.

Thus, the aim of this study was to evaluate the effect of combining TPO and CQ on the depth of cure and shrinkage stress of BFCs. The tested hypotheses were that 1) the combination of CQ and TPO will increase the depth of cure of BFCs and that 2) the combination of CQ and TPO will increase the polymerization shrinkage stress of BFCs.

The monomers and filler particles used in the BFC formulation are listed in Table 1. This formulation is based on a BFC formulation patent¹²  that has monomers and filler particles with refractive indexes very similar to each other (n=1.518 for uncured resin monomer blend; n=1.549 for cured resin monomer blend; and n=1.553 for the 7.5 μm BaBSiO₂ filler). This reduces light scattering inside the composite, which better identifies the effect of light transmission throughout the BFC.^7,13,14 

First, the monomers were blended using a centrifugal mixing device (SpeedMixer, DAC 150.1 FVZ-K, Hauschild Engineering, Hamm, North Rhine-Westphalia, Germany) for 60 seconds at 3000 rpm. To this monomer blend, two different photoinitiator systems were used, as shown in Table 2. The molar ratio concentration of CQ-amine and CQ-amine/TPO were based on previous studies.¹⁵ 

Then, the fumed silica filler was mixed with the monomer blend for 30 seconds at 3000 rpm followed by the BaBSiO₂ filler for 1 minute at 3500 rpm. Finally, each resin composite was mixed for 1 minute at 3500 rpm under an 80 mmHg vacuum atmosphere.

The mean irradiance (mW/cm²) of the multi-wave light-emitting diode (LED; Bluephase G2, Ivoclar Vivadent, Schaan, Liechtenstein) was measured five times using a portable spectrometer-based instrument (CheckMARC, BlueLight Analytics, Halifax, Nova Scotia, Canada) connected to the SpectraSuite software (Spectra Suite 2.0.140, Ocean Optics, Dunedin, FL, USA). The total spectral radiant power output from the LED was divided by the cross-sectional active area (cm²) of the light tip to calculate the average irradiance (mW/cm²). These values were used to calculate the photoactivation time needed to produce a radiant exposure of 20 J/cm². Then, the multi-wave LED was attached to an x-y-z positioning device mounted on an optical bench to standardize the positioning of the light beam in contact with a diffusive surface of a frosted diffuser target (DG20-1500, Thorlabs, Inc, Newton, NJ, USA) while the resulting image was recorded using a CMOS camera (NEX-F3, Sony Corporation, Tokyo, Japan) with a 50-mm focal length lens. To assess the irradiance distribution at different light emission wavelengths, the CMOS camera beam profiler was used with the addition of bandpass filters (Thorlabs, Inc) placed in front of the camera lens. The bandpass filters were as follows: centered at 400 nm with a 4-nm full width at half maximum (FB400-40) was used to identify the LED chips with spectrum emission peaks at 410 nm and centered at 460 nm with a 10-nm full width at half maximum (FB460-10) was used to identify the LED chips generating emission peaks near 460 nm. As the bandpass filters used in this study have an optical density of approximately 0.2, no neutral density filters were necessary to attenuate the multi-wave LED emission to avoid pixel intensity saturation in the images. To produce calibrated images and data showing the irradiance patterns across the surface of the multi-wave LED, the mean power values, obtained using the CheckMARC were entered into an open source optical analysis software (Fiji, Nature Methods, National Institutes of Health, Bethesda, MD, USA). The scaled numeric data associated with each image was exported into a computer graphic software (Origin Pro, OriginLab Co, Northampton, MA, USA).

Each BFC (n=3) was placed in a cylindrical Delrin acetal homopolymer resin mold (12 mm in height, 4 mm in inner diameter) with a Mylar strip manually pressed over the applied material with a glass plate to obtain a flat surface. Afterward, the glass plate was removed and the light curing was performed on the top of the cylinder using the multi-wave LED with 20 J/cm². Then, after 24 hours of dark and dry storage at 37°C, specimens (12 × 4 × 0.5 mm) were obtained by sections from the center of the specimen, perpendicular to the top surface and parallel to the long axis of the cylinder, using an automated water-cooled, low-speed diamond saw (Isomet 1000, Buehler Ltd, Lake Bluff, IL, USA)^6,15,16  with temperature being monitored and recorded during cutting using a IR-Thermal Camera (FLIR Pro, FLIR, Wilsonville, OR, USA). The samples were not submitted to more than 3°C of increase in temperature.

The specimens were fixed onto a glass slab and placed over an automated x-y axis microscope platform. The DC was mapped along the cross section (width 6 mm; depth 4 mm) using a Fourier transform near-infrared (FT-NIR) microscope (Nicolet Continuum, Thermo Scientific, Waltham, MA, USA) coupled to an FT-NIR spectrometer (Nicolet Nexus 6700, Thermo Scientific).^6,15,17  Every 0.5 mm in depth and width, five infrared spectra were collected from the specimen. The measurements started from 300 μm below the top surface to avoid the area of oxygen inhibition. At each measurement position, the specimens' near-infrared spectra were collected in transmission mode with an aperture of 50 μm, 50 scans/spectrum, and 4 cm⁻¹ of resolution. FT-NIR readings were performed in the transmission mode rather than the reflection mode to evaluate DC in bulk rather than on the surface of the material, which reduces the influence of monomer elution from the surfaces of the section specimens. The spectra of the uncured specimens (n=3) collected with the same settings were used as a reference to measure the peak area ratio corresponding to the aromatic double bonds (4625 cm^–1) and vinyl (6165 cm^–1) stretching absorptions. The DC (in %) was calculated as follows:

where pol and non-pol correspond to the area of the methacrylate peak for the polymeric and monomeric states, respectively.

The results were exported to create color-coded maps to describe DC as a function of position in width and depth under the LED emission position using a computer graphic software (Origin Pro, OriginLab Co, Northampton, MA, USA). Each data point registered on the map was the average of three different replications using similar testing conditions. The map scales were set to indicate the maximum DC achieved and a reduction of 10% of the maximum DC in each subsequent color-coded area; a black dashed line was also drawn to represent the limit boundary of 80% of the maximum DC achieved by the BFCs.

The multi-wave LED spectral irradiance (mW/cm²/nm) was obtained with a spectrophotometer (MARC Resin Calibrator, BlueLight Analytics) that has an optical fiber coupled to a cosine corrector sensor of Ø=3.9 mm and a field of view of 180°. The equipment was pre-calibrated using a traceable light source (LS-1-CAL, Ocean Optics) according to the standards of the National Institute of Standards and Technology (NIST, Gaithersburg, MD, USA). A spectroscopy software (Spectrasuite 2.0, Ocean Optics) was used to obtain the multi-wave LED spectral irradiance over a 380-495 nm wavelength range with a time acquisition of 20 seconds. The graphs obtained were used to calculate the spectral irradiance and the radiant exposure (J/cm²) in the violet range (380-420 nm) and blue range (420-495 nm) by integrating the irradiance versus the wavelength in the spectral irradiance graphs.

LT through each BFC was recorded during the light curing using the same MARC Resin Calibrator spectrophotometer with the same parameters. Samples of each BFC (n=5) were placed in Delrin acetal homopolymer resin molds (Ø=4 mm) of different thicknesses (1 mm, 2 mm, 3 mm, 4 mm, 5 mm, and 6 mm) and positioned on the cosine corrector with Mylar strips covering the top and bottom surfaces of the samples. The spectral radiant power and the irradiance transmitted through each BFC were measured using the spectroscopy software. The spectral irradiance transmitted in the violet and blue ranges were determined for each BFC in different thicknesses. As LT is defined as the percentage of incident light that passes all the way through to the other side of the sample, LT in percent was calculated by subtracting the irradiance obtained at the bottom of each sample from the irradiance of the multi-wave LED in each wavelength range.

VS was measured using a mercury dilatometer (American Dental Association Health Foundation, Gaithersburg, MD, USA).¹⁸  Approximately 0.1 g of each BFC (n=3) was placed on a sandblasted and silanized glass slide (75×25×1 mm). As the glass slide used was 1 mm thick and could attenuate the light that reaches the BFCs during light curing, before starting the test the multi-wave LED was recalibrated to deliver the same radiant exposure (20 J/cm²) through the glass slide. After that, a glass column was clamped to the glass slide, filled with mercury, and a linear variable differential transducer probe was placed on top of the mercury. VS was monitored for 60 minutes after the photoactivation. The data recorded by the probe was used to calculate the VS using previously measured mass and density values of BFCs.

PSS was measured for 600 seconds using the Bioman.¹⁹  The BFCs were placed between a silanized glass slab and a roughened steel piston. The distance between the glass slab and the steel piston produced a 1-mm-thick composite that corresponded to a configuration factor of 5. The composites were photoactivated through the glass using the multi-wave LED with the same parameters used for the VS test. The load signal in Newton (N) was acquired and then divided by the disk area of the piston to obtain the maximum PSS values (MPa) and the maximum rate of PSS (rPSS, in MPa.s⁻¹).²⁰ 

As the flexural modulus of the composite is correlated with the magnitude of the polymerization shrinkage stress, the FM was measured using a three-point bend test. Bar-shaped specimens of 10 × 1 × 1 mm (n=10) were made in stainless steel molds. The specimen size of 1-mm thick was chosen according to previous studies to better evaluate mechanical properties.²¹  The bulk increment was light-cured through a Mylar strip using the multi-wave LED with two light irradiations of 20 J/cm², overlapping each other by half of the diameter of the curing light tip. After 24 hours of storage in distilled water at 37°C, the specimens were submitted to a three-point bending test (Instron, Canton, MA, USA) at a crosshead speed of 0.75 mm/min until failure. FM was calculated in gigapascals (GPa) according to the following equation:

where L = the maximum load at failure (N), D = distance between the supports, w = the specimen width, h = the specimen height, and d = crosshead displacement.

Data were entered into statistical analysis software (Stata/MP 13, StataCorp, College Station, TX, USA) and checked for normality using the Shapiro-Wilks test and for variance homoscedasticity using the Levene test. Statistical analyses were performed according to the different experimental designs with a level of significance of α = 0.05. A power analysis was conducted to determine the sample size for each experiment to provide a power of at least 0.8 at a significance level of 0.5 (β=0.2). DC was analyzed using a split-plot analysis of variance (ANOVA) where the independent variables were set as between-subject groups for the experimental BFC (CQ-amine or CQ-amine/TPO) and as within-subject groups for widths (0, 1, 2, 3, and 4 mm) and the depths (0, 1, 2, 3, and 4 mm). LTs in violet and blue range were analyzed independently using a two-way ANOVA where independent variables were set as BFC (CQ-amine or CQ-amine/TPO) and thicknesses (0, 1, 2, 3, 4, 5, and 6 mm). For DC and LT, the Tukey test was applied for multiple comparisons between groups (α=0.05). VS, PSS, rPSS, and FM were analyzed using a one-way ANOVA where independent variables were set as BFC (CQ-amine or CQ-amine/TPO).

The multi-wave LED provided a mean irradiance of 1151 ± 50 mW/cm² during the 17 seconds of exposure time, thus producing 20 ± 0.5 J/cm² of radiant exposure. However, 3.4 ± 0.2 J/cm² was within the violet wavelength range (380-420 nm) and 16.6 ± 0.4 J/cm² was within the blue range (420-540 nm) as represented in Figure 1A. Figure 1B,C shows the beam profile of the multi-wave LED according to the violet (380-420 nm) and blue (420-495 nm) wavelength range. The multi-wave LED had an active area of emission of 0.649 cm² and a maximum irradiance of 1386 mW/cm², but the irradiance and the wavelengths were not uniformly distributed across the light tip. In the black dashed circle area referenced to the size of the specimen, the multi-wave LED had an overall irradiance of 1025 mW/cm², where 152 mW/cm² was within the violet wavelength range (380-420 nm) and 873 mW/cm² was within the blue wavelength range (420-495 nm).

Table 3 shows the results of LT into the violet and the blue wavelength ranges for the CQ-amine and CQ-amine/TPO BFCs according to the different thicknesses. For both BFCs, there was an exponential decrease in LT as the specimen's thickness increased, regardless of the wavelength range. However, the CQ-amine BFC had a higher violet light transmittance than the CQ-amine/TPO BFC up to 1 mm in depth, but no differences were found beyond the 1-mm thickness. On the other hand, the CQ-amine/TPO BFC had a higher blue light transmittance than the CQ-amine up to 1 mm in depth, but no differences were found beyond the 1-mm thickness.

Figure 2A,B show color-coded maps representing, in every 0.5 mm in width and depth, the mean of the degree of conversion (DC, in %) for CQ-amine [A] and CQ-amine/TPO [B] BFCs according to the LED emission position, violet (380-420 nm) and blue (420-495 nm), in width and in depth. The black dashed line in Figure 2A.B represents the limit boundary where at least 80% of the maximum DC on the top of the restoration was achieved by the BFCs. Figure 2C,D shows areas in yellow where significant statistical differences were found within subjects (width and depth nested inside each BFC, CQ-amine or CQ-amine/TPO). The CQ-amine BFC had a homogeneous DC distribution across the specimen since no statistical differences were found between width and depth regions (degrees of freedom [df][125]; F=1.03, p=0.409). In contrast, the CQ-amine/TPO BFC had a nonhomogeneous DC distribution across the specimen since statistical differences were found between width and depth regions (df([25]; F=1.69, p<0.001), where the areas in yellow showed lower DC than the areas in the top part of the specimen. Figure 2E shows areas where significant statistical differences were found between subjects (for the same width and depth position, differences between CQ-amine and CQ-amine/TPO BFCs). In the areas presented in yellow, the CQ-amine/TPO BFC had higher DC than the CQ-amine BFC (df[279]; F=30.4, p<0.001)

Table 4 shows the results of VS (%), PSS (MPa), rPSS (MPa.s⁻¹), and FM (GPa) for each BFC. There was no statistical difference between CQ-amine and CQ-amine/TPO BFCs for VS (df[1]; F=1.23, p=0.310). However, the CQ-amine/TPO BFC had a higher PSS (df[1]; F=13.08, p=0.011), rPSS (df[1]; F=35.98, p=0.001), and FM (df[1]; F=6.86, p=0.020) than the CQ-amine BFC.

It is believed that LT through a resin-based composite is a result of photoinitiator absorption, surface reflection, light scattering, and refraction caused by differences in the refractive indexes of monomers and filler particles.²²  As the same monomers and filler concentrations were used for both experimental BFCs, the present study was designed to determine the effect of two photoinitiator systems in the LT properties of the experimental BFCs. The photoinitiator systems used were CQ-amine, which absorbs light in the blue wavelength range from 420 nm to 495 nm with a peak absorption at 470 nm, and CQ-amine/TPO, which absorbs light in the violet and blue wavelength range from 350 nm to 495 nm with two peak absorptions at 380 and 470 nm.¹⁵  Since several lines of evidence suggest a strong relationship between the spectral output of the light curing unit and the photoinitiator absorption,^6,22,23  the multi-wave LED used to light cure the BFCs emitted 20 J/cm² of radiant exposure and covered photoinitiator absorption of both CQ-amine and CQ-amine/TPO BFCs.

However, differences were found between the violet and blue wavelength emissions for the multi-wave LED. While 16.6 J/cm² was emitted into the blue spectrum, only 3.4 J/cm² was emitted into the violet spectrum. Nevertheless, despite the differences in the amount of violet and blue light emitted by the multi-wave LED, previous evidence presented thus far supports the idea that the spectral output of the multi-wave LED used was enough to efficiently cure BFCs containing either CQ-amine or CQ-amine associated with alternative photoinitiators.^6,10,16,26  The present study findings are in agreement with those of previous studies; however these conclusions should be interpreted with further attention since they do not rule out the influence of other factors, such as, the light beam profile received by the specimen.

Therefore, a note of caution is due here since the beam profile results are in accordance with recent studies indicating that the light distribution throughout the multi-wave LED bluephase G2 light tip is significantly influenced by the reflectors, LED chip positions, and the presence of three blue LED chips and only one violet LED chip.^24,25  Then, the most recognizable finding to emerge from the light curing unit characterization is that the light received on the top surface of the specimens used in this study was not homogeneously distributed, neither in irradiance nor in wavelength. Consequently, the key problem with the nonhomogenous light emission is that it might have important implications for the degree of conversion homogeneity throughout the entire BFC restoration.

But it is somewhat interesting that this is not true for all case scenarios, and the data reported here appear to support the assumption that CQ-amine and CQ-amine/TPO BFCs would have different behaviors regarding the degree of conversion homogeneity when light cured with a nonhomogeneous dual-peak multi-wave LED. To some degree, the CQ-amine BFC had a lower overall DC than the CQ-amine/TPO BFC, but it was homogeneously distributed across the entire specimen. In contrast, the CQ-amine/TPO BFC had a higher overall DC than the CQ-amine BFC, but there were differences between the top and bottom part of the specimen.

It is therefore likely that such connections exist between the position of the violet light beam and the presence of TPO. As shown in Figure 2E, the DC of the CQ-amine/TPO BFC was equal or higher than the DC of the CQ-amine BFC, and this discrepancy is more pronounced in areas where there is an overlap between violet and blue light. But the fact is that the presence of TPO has a significant influence on the DC. TPO is known as a high reactive photoinitiator because it has a higher molar extinction coefficient (≈520 L/mol cm) than CQ (≈28 L/mol cm). This means that TPO absorbs light more efficiently than CQ at their peak wavelength absorptions²⁷  and that TPO cleavage forms acyl (carbon centered) and phosphonyl (phosphorus centered) free radicals that have higher nucleophilicity and electron resonance than the amino radical (carbon centered) formed by CQ-amine electron donor/hydrogen abstraction reaction, Thus, initiation of polymerization is increased, leading to a higher DC.²⁷ 

Although these differences in the DC are strictly related to the top part of the restoration due to the TPO wavelength absorption within the violet wavelength range, previous studies have shown that the DC of a polymer-based composite is related to its RI.¹³  Consequently, it is expected that the polymer network with a higher DC formed by the same monomers would have a higher refractive index, closing the mismatch between polymer and filler refractive indexes and improving the LT. As shown in Table 4, the CQ-amine/TPO BFC has a higher blue LT than the CQ-amine BFC, but, contrary to expectations, this difference was found up to 1 mm in thickness, meaning that the improvement in the DC was not enough to improve the blue LT through the BFC. Therefore, the first research hypothesis tested—that the combination of CQ and TPO will increase the depth of cure of BFCs—was rejected.

Nevertheless, a higher DC on the top of the restoration might improve the wear and hydrolytic degradation resistance and perhaps the clinical performance of BFC.²⁸  Therefore, it can thus be suggested that the combination of CQ-amine and TPO would improve chemical and physical properties of BFCs. However, in a more systematic approach this study also identified how CQ-amine and TPO interact with PSS.

Despite no statistical differences being found between the CQ-amine and CQ-amine/TPO BFCs regarding the VS, the combination of CQ and TPO increased the rPSS, PSS, and FM. For CQ-based composites, it is necessary to use an amine together with the CQ to form free radicals efficiently enough to initiate polymerization⁵ ; however, when CQ and amine are combined with TPO, the reaction could increase autoacceleration (Trommsdorff-Smith effect).^29,30 

The major issue is the presence of a tertiary amine (ie, EDMAB) combined with TPO, especially because acyl and phosphonyl radicals formed by TPO cleavage can abstract a hydrogen atom from a tertiary amine, creating one more way to initiate polymerization. This reaction is so fast that it could accelerate the rate of polymerization and the vitrification of the polymer. This means that a very fast reaction does not provide enough time to relax the internal structure, and this shrinkage stress is transferred to the tooth's boundaries.³¹ 

Furthermore, the stiffness of the material affects the stress produced as a result of polymerization.^31,32  According to Hook's law, FM = Stress/Strain, and because the flexural modulus represents the ratio of the elastic stress to the elastic strain, it follows that the higher the stress for a given strain, the higher the FM can be a factor for the higher PSS of CQ-amine/TPO BFC. Therefore, the second research hypothesis tested—that the combination of CQ and TPO will increase polymerization shrinkage stress of BFCs—was accepted.

Recent evidence suggests that the effect of combining CQ-amine and TPO at a molar ratio of 1:1 can improve optical properties without reducing the depth of cure of conventional composites.¹⁵  What is surprising is that perhaps the most serious consequences of combining CQ-amine and TPO are related to PSS and, thus, the clinical performance of BFCs containing CQ-amine and TPO.

Nevertheless, the lack of clinical evidence could lead to a concern regarding the combination of these two photoinitiator systems in BFCs. Bayraktar and others¹¹  showed that a composite containing CQ-amine associated with alternative photoinitiators (TPO and Ivocerin) has a higher incidence of discontinued anatomic form, marginal discoloration, creviced in which dentin is exposed, and secondary caries than a composite containing only CQ-amine in just one year of observation.¹¹  But certainly, this should be observed with more caution because from 50 restorations placed using the composite containing CQ-amine with alternative photoinitiators in this study, only two restorations failed, which means that other factors, such as operator technique, especially during the light curing procedure, might be the cause of these failures. Still, this study seems to have overlooked one important fact that the light curing unit used to cure the CQ-amine with alternative photoinitiators composites was a single peak blue spectrum monowave LED (Elipar S10, 3M ESPE, St Paul, MN, USA). As the monowave LED used has a single peak emission at 450 nm²⁴ and the BFC used contains TPO and Ivocerin phoinitiators with peak absorption at 370 nm and 418 nm. respectively,⁶  it seems possible that these negative results are due to poor composite light curing.

In addition, Schwendicke and others³³  published a very robust meta-analysis showing the clinical performance of directly placed restorative materials. In that study, 72 randomized controlled trials published between 2005 and 2015 were selected; however, only 8 studies included BFCs. Moreover, only 3 studies were using high-viscosity BFC and none of them included BFCs with alternative photoinitiators. Given that most clinical trials analyzed were short term and showed high risk of bias, future clinical trials should focus on comparative designs and reports using materials with the same composition but different photoinitiator systems.

It is almost certain that the polymerization reaction produces stress that would not be adequately dissipated by the strain caused within the composite, even in BFCs. Therefore, it is probable that the stress is transferred to the bonded interfaces with the tooth structure creating delamination or tooth fracture whenever and wherever the localized stress exceeds the adhesion strength or the strength of the adjacent residual tooth structure. Furthermore, these stresses may increase with time, causing delayed damage to cavity margins. Although many studies have shown in vitro evidence of marginal failures using commercially available BFCs, there is abundant room for further progress determining systematically the connections between the combination of photoinitiators and the marginal failures in experimental materials. The findings in this study indicate that CQ-amine/TPO BFCs could have a higher PSS than CQ-amine BFCs. This suggests that a weak link may exist between PSS and the marginal adaptation of BFCs containing CQ-amine associated with TPO.

Nevertheless, new alternatives to improve light curing of BFCs have emerged. Oliveira and others³⁴  showed new composites containing photoinitiators systems that light cured using longer wavelength range (green or red light) than short wavelength range (violet or blue). Apparently, long wavelengths have higher light transmittance than short wavelengths, but further work is required to establish the viability of using these high-efficiency photoinitiators to improve the depth of cure of BFCs. Still, the use of high-reactive photoinitiators in a BFC might be a drawback since they could increase PSS unless the stress-relief monomer system could compensate for the increasing PSS caused by these photoinitiators. However, recent studies concerning the development of new monomer systems based on thiol-ene polymerization reactions (thio-urethanes)¹⁷  might be a promising alternative to reduce the PSS of BFCs.

Within the limitations of this study, the following conclusions can be made: Although the CQ-amine/TPO BFC has a higher DC on the top of the restoration, the combination of CQ-amine and TPO could not increase the blue light transmission to enhance the depth of cure of BFCs. Also, despite the CQ-amine/TPO BFC having some improvement in mechanical properties, such as FM, the CQ-amine/TPO BFC had a higher PSS and rPSS than the CQ-amine BFC.

This study was supported by the FAPESP (grant #2016/06019-3 and #2017/22195-9). Dayane Carvalho Ramos Salles de Oliveira is a postdoctoral researcher at FAPESP (grant #2016/05823-3 and #2017/22161-7). We would like to acknowledge Esstech and Evonik for the kind donation of the materials used in this study.

The authors of this manuscript certify that they have no proprietary, financial, or other personal interest of any nature or kind in any product, service, and/or company that is presented in this article.