The incorporation of pre-polymerized additives, including thiourethane oligomers, to the formulation of resin composites has shown desirable outcomes, such as the reduction in volumetric shrinkage19 and shrinkage stress4,19, and increase in fracture toughness4,5,6. However, this strategy usually leads to a non-negligible increase in viscosity, which prevents the incorporation of higher percentages of inorganic fillers6, and negatively affects handling characteristics. This ultimately limits the amount of pre-polymers that can be added before other properties start to deteriorate (6—30% TU in cements). For this reason, the functionalization of the inorganic filler surface with oligomeric species has been proposed, and, at least for thiourethanes, has shown similar reduction in polymerization stress and improvement in mechanical properties without compromising the viscosity of the composite paste7,8. In this study, TU functionalized filler particle characteristics were studied in a systematic fashion to gain insight into the mechanisms leading to the reinforcing and stress-reducing outcomes observed with thiourethane oligomers.
The same standard functionalization method was used for the thiourethane and the conventional methacrylate silane (MA-Sil). The grafting to the filler surface was accomplished in a slightly acidic aqueous alcohol solution (pH ≅ 4.5) to hydrolyze the alkoxy groups, forming reactive silanols (≡Si–OH), which further react with themselves by a condensation process to form oligomeric siloxanes (–Si–O–Si–bonds). At the same time, the growing silane networks link to the inorganic substrate through hydrogen bonds with the hydroxyl groups present on the particle surface20,21,22. The filler surface functionalization efficiency was assessed by thermogravimetric analysis (Fig. 2A). For the particles functionalized with the MA-Sil, all organics were burned off at 500 °C and no additional mass was lost at higher temperatures. For TU-Sil fillers, the maximum mass loss was observed closer to 600 °C, which was also expected based on the much greater molecular weight of the oligomer compared to the small-molecule methacrylate silane. The untreated particles (NO-Sil), used as received from the manufacturer, also showed mass loss at a similar temperature, which could be due to impurities from the manufacturing process. For all surface treatments, the mass loss increased with a decrease in the average particle size23, which was expected since at the same mass, 1 µm-sized particles have ten times more free surface area available to be functionalized than 10 µm-sized particles (Table 3). It is noteworthy that the mass loss as a function of surface area was different for each surface treatment. For the NO-Sil and MA-Sil groups, 1 µm-sized particles showed ten times more weight loss than 10 µm-sized particles: 1.61 and 0.22% for 1 µm and 10 µm-sized particles, respectively (NO-Sil) and 2.91 and 0.39% for 1 µm and 10 µm-sized particles, respectively (MA-Sil). For the TU-Sil groups, the mass loss did not linearly scale with the particle size (weight loss = 11.73% and 5.82% for 1 µm and 10 µm-sized particles, respectively). This difference can be at least partially explained by differences in molecular weight (~ 250 g/mol for the MA-Sil and 5 kDa—or roughly 5,000 g/mol—for TU-Sil)6, as well as differential thickness of the coating. The monotonic increase in weight loss with decreasing filler size is consistent with the formation of a monolayer for the methacrylate silanes. In contrast, it is possible that the thiourethane oligomer is crosslinked onto the surface, which will be further discussed later. It is also possible that the silanization process with the thiourethane oligomer is not as efficient, and that results in non-uniform functionalization, as shown in the imaging results.
SEM micrographs (Fig. 5B) were used to characterize the particle–matrix interface. While the MA-Sil particles are entirely coated with organic matrix, thiourethane particles seem to be more heterogeneously coated, with some naked regions, and others showing better interaction with the matrix (Fig. 5C). This pattern is also identified in the confocal images which show uneven distribution of the thiourethane oligomer along the filler particle surface (Fig. 7), alternating areas with a layer of 649–1885 nm in thickness depending on the filler size and other naked regions (Table 3 and Fig. 6). The coating on the larger filler particles was much more disperse, with extensive areas of the filler remaining uncovered. The bulky thiourethane oligomer is likely tethered onto the filler surface at multiple locations, creating steric constraints for additional chain grafting, ultimately limiting the overall graft density24. It is important to point out that the starting concentration of TU oligomer during the filler functionalization was the same for all filler sizes, and the mass of filler used here was the same regardless of the size. This results in much greater overall surface area and smaller inter-particle distancing in the smaller filler particles. This likely explains why the thickness of the silane layer was similar for the 1 and 3 µm particles, at around 649 and 1226 nm, respectively, but much thicker for the 10 µm particles (1885 nm). In summary, for the larger particles, the greater inter-particle spacing and smaller overall surface area, likely combined to increase the coating thickness. The surface area vs. particle size relationship also explains the mass loss data from the TGA, which shows greater mass loss with the smaller particles, again due to the greater filler surface area per volume of material afforded by smaller particles. In addition to the mechanisms that will be discussed in more detail later, it is possible that the uncoated areas act as defect sites, which are known to lead to lower stress development25.
During polymerization, the monofunctional methacrylate silane co-polymerizes with the vinyl-containing matrix, and establishes a short and rigid bond with the filler surface20, which contributes to stress generation. In contrast, the high molecular weight thiourethane silane establishes multiple covalent interactions with the polymerizing organic matrix via the pendant thiols. The bonds formed are flexible thiocarbamate bonds which can serve as sites for relaxation of the overall polymerization stress26. In the specific case of the thiuorethane oligomers studied here, some stress relaxation during polymerization is also afforded by the delayed gelation vitrification5,6,7, but more recent studies have also pointed to the possibility of stress relaxation via dynamic bond adaptation behavior in the glassy state27,28,29,30. The dynamic relaxation behavior of thiourethanes is currently being investigated and will be reported separately. Preliminary studies using time–temperature superposition experiments have already demonstrated faster relaxation times for fillers treated with TU-Sil31. Different from the strategy where the stress-relieving molecules is randomly distributed in the matrix4,5,6, in composites with TU functionalized fillers, the stress relieving molecules are localized at the filler-matrix interface, a region of stress concentration32. This may account for the significant reduction in stress transfer between the two constituent phases of the resin composites25, even at a much lower overall TU concentration than that present when TU is added directly to the resin matrix. Additionally, the thiourethane likely forms thick and highly dense polymer structures on the surface of the particle, which may also contribute to dissipate part of the generated stress33,34 and to increase plastic deformation during stress development that help accommodate changes in free volume34,35. The multifunctional nature of thiourethanes also leads to differences in crosslinking in the siloxane layer and enhanced interface adhesion25. This may compensate for the uneven coverage of the filler particle by the thiourethane silane mentioned earlier, and helps explain why mechanical properties are not compromised, in spite of the presence of naked regions on the filler particle surface. Finally, the stress development and the mechanical properties of the resin composites are also dependent on the degree of agglomeration of the filler particles36. Systems with greater filler dispersion show improved storage modulus, tensile strength, toughness and lower polymerization stress due to the enhanced matrix-filler interaction and interfacial adhesion25. When polymer chains are grafted onto the particle surface there is steric repulsion, which minimizes their tendency to agglomerate due to the van der Walls attraction and, ultimately, promotes a more uniform dispersion of the particles.
The stress development can also be correlated with the kinetics of polymerization. The fact that with NO-Sil and MA-Sil composites had similar stress behavior was unexpected since, in theory, in the absence of bonding between the particles and the organic matrix, the filler particles should effectively behave as voids, leading to stress relief37. However, in general NO-Sil composites showed markedly faster polymerization reaction, which may have led to earlier development of diffusional limitations and a rise in stiffness, minimizing the opportunity for stress relaxation. The higher RPMAX and DC at RPMAX found in NO-Sil formulations may have resulted from the increased system mobility imposed by the absence of functional silanes at the interface38. As expected, the addition of TU-Sil led to significant reduction in polymerization stress, ranging between 41 and 54% in comparison with the MA-Sil groups. The reduction in stress is explained by chain-transfer reactions of the thiols with the vinyl groups, which delays the point in conversion at which the stiffness of the networks begins to significantly increase, and past which any increase in conversion results in disproportionately higher stress. In general, the addition of TU-Sil particles into the resin composite formulations decreased the RPMAX, as well as the conversion registered at RPMAX, in agreement with previous studies4, indicating at least some effect in delayed network formation (Fig. 4). The somewhat slower polymerization reaction did not compromise the final DC; on the contrary, TU-Sil formulations showed the highest values for 1 and 3 µm-sized filler particles. For the 10 µm-sized particles, conversion was similar to the methacrylate controls. This may be due to the lower concentrations of thiourethane in this composition (Table 4), which may have been insufficient to affect the polymerization reaction kinetics. Chain-transfer reactions are also responsible for more homogeneous network formation, which decreases the development of internal stress, especially immediately after the diffusion limitation occurs5,39,40. It is also possible that the simple presence of the low Tg thiourethane on the surface of the filler particle may play a similar role in stress relief, acting as a ductile zone for plastic deformation between the filler and the organic matrix, which ultimately yields stress absorption at the interface and toughening. Interestingly, it seems that the localization of thiourethanes directly at the surface of the filler particle significantly decreases the overall concentration of thiourethane needed to produce significant reduction in polymerization stress, compared to what is needed when TU is added directly into the matrix4,5,6. Therefore, several mechanisms may be operating at the filler-matrix interface, as mentioned throughout the discussion, and summarized in Fig. 7. It is important to reiterate, however, that the stress reduction is not gained at the expense of reduced conversion.
A potential concern over the uneven coverage of the filler particles with the TU might be the susceptibility of the particles to being dislodged under load, consequently, compromising the mechanical properties. In this study, the fracture toughness results showed the opposite effect for the 1 µm TU-Sil, with significant enhancement for all the three groups containing TU-treated particles (1.60 ± 0.03 MPa•m1/2) in comparison to MA-Sil groups (1.19 ± 0.08 MPa•m1/2) (Fig. 3). This 34% increase in fracture toughness is attributed to the flexible thiocarbamate covalent bonds4, and may also be due to lower levels of internal stress accumulation41,42. For 3 and 10 µm-sized fillers, there were no statistical differences between TU-Sil and MA-Sil, which indicates that the percentage of the thiourethane incorporated into the mixtures (Table 5) was not sufficient to significantly improve the fracture toughness. In fact, the confocal images demonstrate that the TU-Sil layer was much more uneven than for the 1 µm filler, which correlates with the data for mass loss shown in Table 3, as already mentioned. The SEM micrographs of the fracture surfaces containing TU-Sil particles showed chunks of organic matrix covering some regions of the particle’s surface, which may indicate that the fracture sometimes propagated through the organic matrix, but also along the resin-filler interface in these systems. In contrast, in the MA-Sil systems the fractured surfaces were typically covered by a thin and uniform layer of resin, which indicates that the fracture propagated through resin matrix near the fillers. This adds evidence to the fact that, though not being entirely coated by the silane, the interaction of the inorganic fillers to the organic resin in thiourethane-containing systems still allows for lower stress concentration at the filler-matrix interface. In addition, the uncoated areas on the TU-Sil filler particles might provide an energy releasing path around the particles, contributing to the enhanced toughness37. Groups containing NO-Sil particles showed, as expected, the lowest fracture toughness, due to the absence of interfacial bonding between the filler particles and the organic matrix. The mass percentage of filler particles incorporated into the formulation did not impact the mechanical resistance, which is in agreement with results reported previously in the literature for composites with different levels of 1 µm-sized barium-alumina borosilicate particles43. This previous study has shown that the flexural strength decreases and the flexural modulus increases slightly as the filler particle load increases from 40 to 60 wt% and, above 60 wt%, there is a gradual increase in both flexural strength and modulus. The results were correlated with particle size distribution, particle–matrix adhesion strength, and arrangement of the filler particles into the organic matrix43. In highly loaded systems (above 60 wt%), there was a tendency for a percolated network particle structure to be formed, which are aggregates of filler associated with mechanical reinforcement43. However, it is important to highlight that there is a threshold for increase in filler particle content and increase in mechanical performance, above which the addition of higher amounts of filler leads to decreased particle–matrix adhesion strength, possibly due to the formation of agglomerates. This threshold varies according to the filler particle system. It is possible to assume that, at least for the larger filler sizes (3 and 10 µm), the load range used in this study was insufficient to result in significant differences in fracture toughness. For the 1 µm filler particles, the fracture toughness results agree with our previous work demonstrating significant increase in values when comparing methacrylate vs thiourethane silanes7.
One thing to note is that, even though the filler loading followed standardized mass ratios (50, 60, and 70 wt%), for the TU-Sil groups, a much higher percentage of the filler weight corresponded to the mass of the silane, compared to the methacrylate groups. As a consequence, the actual inorganic filler loading varied significantly among the groups (Table 5). In general, TU-Sil composites contain 10 wt% less filler than MA-Sil formulations with 1 µm-sized particles, 6.5 wt% less with 3 µm-sized particles, and 5.4 wt% less with 10 µm-sized particles. This translated into thinner film thickness for groups containing TU-Sil particles (Fig. 2C). In addition, the film thickness is also affected by the distance between particles. For smaller filler sizes, the particles become more compacted (closer to each other), which also leads to an increase in filler content (Fig. 2B). As expected, the smaller particles led to greater viscosity as more of the resin matrix is influenced by contact with the fillers, and the smaller inter-particle spacing leads to more filler-filler interactions, both of which result in thickening of the paste.
An additional potential advantage of the thiourethane coating is its hydrophobicity. As it is well known, the presence of ester bonds makes the methacrylate silane prone to hydrolytic degradation. The siloxane layer degradation, caused by the vulnerability of the oxane bonds to hydrolysis due to its significant ionic character, increases the concentration of hydroxyl ions44, which leads to an autocatalytic reaction. As result of this reaction, there is a weakening of the filler-matrix bonding, leaching of chemical compounds, generation of micro-cracks at the interface, particle debonding, and, ultimately compromised mechanical properties45. Even though it was not the main goal of the present study to investigate the hydrolytic stability of the interfaces, the use of the thiourethanes (a multifunctional, hydrophobic, crosslinked, high molecular weight oligomer) as a particle coating can be envisioned to improve the durability and the hydrolytic stability of the interfacial siloxane bond. This may ultimately improve the durability of the filler-matrix bonding.