The results of extract assays are shown in Fig. 3 and Fig. 4. The data have been analysed by one-way ANOVA, and the minimum significant difference at p<0.05 has been calculated and displayed on the histograms. There were no significant differences between any extracts of sample groups and the negative control group except the mullite group. A significant decrease of OD value happened in the positive control group, indicating that the tests were valid and L929 mouse fibroblasts cells were susceptible to the degrees of cytotoxicity.
The OD values (490nm) of each set of samples are shown in Fig. 3. Low OD values of all experimental groups were found after the cells had been cultured by extracts for 1 day, which was mainly because the cells did not fully grow and divide, and the cell concentration was low. Three days later, the OD values increased significantly except for the pure mullite group with a concentration of 3cm2/mL and the positive control group. After being incubated for five days, the OD values of each experimental group were almost twice that of the cells cultured for 1 day, and each of them was not lower than that of the negative control group except the pure mullite group and positive control group.
Combined with the comparison between RGR value (>100%) and determination standard of cytotoxicity level, ZrO2 ceramic and mullite/3Y-TZP had no cytotoxicity, and morphologies of the cells of these groups also confirmed this result, as shown in Fig. 4(a-b) . Although RGR value of mullite group ＜100%, the result of morphologies of the cells cultured by extract of mullite for 5 days indicated that mullite had no cytotoxicity or slight cytotoxicity compared with that of the positive and negative control groups, as shown in Fig. 4(c-e).
No adverse reaction was observed in petri dishes by inverted microscopy in the direct contact experiments, and morphologies of the cells are shown in Fig. 5. As depicted in Fig. 5(a-c), no abnormalities or dead cells were found around the samples, and no cell-free transparent regions were observed. Density and morphologies of the cells of these sample groups were similar to those of the negative control group (Fig. 5(d)), while there were significant death cells in the positive control group (Fig. 5(e)). The results showed that those three kinds of samples were no-cytotoxicity, which was a further proof of the previous experimental results.
3.2 Microstructure characterization
Fig. 6 shows the microstructure of mullite/3Y-TZP, and there are two obviously different phases in the sample. The preliminary experimental results proved that the black phase was mullite generated by the reaction of Al2O3 and SiO2 during the sintering process, while the gray area was ZrO2 [24, 25]. In the ternary eutectic system of Y2O3-SiO2-Al2O3 formed by Y2O3 with SiO2 and Al2O3, local liquid phase appeared in the sample at high temperature, then the contact reaction and nucleation occurred between SiO2 and Al2O3. After that, the core of mullite crystal grew into the columnar crystal through mutual diffusion, as shown in Fig. 6(a) [37-41]. The size of the columnar mullite was about 10μm, as shown in Fig. 6(b), and there were ZrO2 particles uniformly distributed in the interior of mullite, which can improve the strength of columnar mullite by the pinning effect. Meanwhile, columnar mullite will also significantly enhance and toughen the composite ceramic for this reason. In addition, previous studies have shown that Y2O3 can enter the crystal lattice of ZrO2 to stabilize the tetragonal/cubic phase .
The surface morphologies of the tooth enamel before the tribological tests are shown in Fig. 6(c-d). It retained complete character with no protrusions or microcracks on the surface, providing a reliable premise for subsequent friction experiments. EDS analysis indicated that the main components of the tooth enamel surface were calcium and phosphorus, and the atomic ratio of the two elements was about 1.6:1, confirming that tooth enamel was indeed made of hydroxyapatite (Ca10(PO4)6(OH)2), as shown in Fig. 6(d), which provided a theoretical basis for subsequent analysis of wear debris.
3.3 Wear behavior
3.3.1 Coefficient of friction (μ) and mass loss
Ra and 3D topographies of the polished tooth enamel and mullite/3Y-TZP before the tribological tests are shown in Fig. 7. Their Ra values were 33.6nm and 148.22nm respectively, indicating that the surfaces of these two kinds of materials using for the experiments were smooth, which can significantly reduce the frictional resistance (Fx) and μ . As is known to us, there should be an appropriate μ when the dental ceramics, especially ceramic crowns, sliding against natural teeth, so as not to cause excessive wear on one side or affect the chewing of food. The results were obtained by repeating the friction experiments for three times, and one of them was shown in Fig. 8. It can be seen that after a brief run-in period, the μ finally stabilized at 0.464. Combined with the results of Wang et al. (as shown in Fig.9), this value was between the μ of glass ceramics and Au-Pd alloy when rubbing against tooth enamel, and was particularly similar to the result of polished zirconia and porcelain . Meanwhile, we also understood the influence of the surface treatment on the results of friction experiments of dental ceramics from their research, and this is why the frictional couple were polished before the tests. The reason for getting this result was that even though the surfaces of this couple were very smooth before the tribological tests, they would be destroyed after the initial contact under the action of applied load, resulting in unstable μ and Fx until new wear surfaces were generated. In the process of stable friction, the existence of artificial saliva played an important role in lubrication and cooling, but also could wash away the debris generated during the friction process and clean the wear surface. Therefore, the μ and Fx were decreased and extremely stable, and they were the factors that determined the mass loss of the teeth and mullite/3Y-TZP [34, 42]. This frictional behavior indicated that there was a good match between the tooth enamel and mullite/3Y-TZP.
The mass loss of the teeth and mullite/3Y-TZP was very small, even though the former value was slightly higher than that of the later, both of them lost a few milligrams, only were 0.5±0.1mg and 0.3±0.1mg, indicating that the wear resistance of the friction couple was well in this environment. Stable friction process and low mass loss proved that mullite/3Y-TZP had application potential in the field of oral cavity.
Moreover, Lee et al . found that ZrO2 would undergo phase transition under the action of load during the friction process, and the low mass loss of mullite/3Y-TZP may partly attributed to transformation toughening induced by flash-temperature. In this study, the diffraction peaks of different ZrO2 phases of mullite/3Y-TZP did not show significant changes before and after the tribological tests, as shown in the XRD patterns of Fig. 10. The diffraction peaks intensity of m-ZrO2, t-ZrO2 and c-ZrO2 were almost unchanged, which may be caused by stress dispersion and cooling effects of artificial saliva. In addition, the internal structure of mullite/3Y-TZP did not change before and after the tribological tests, as shown in the Raman spectrum of Fig. 11 (black and red curves). The intensity of the diffraction peak and Raman shift of m-ZrO2 and t-ZrO2 did not change significantly . The -OH peak (3625cm-1) was observed in the Raman spectrum of mullite/3Y-TZP after the tribological tests, and Lang et al. [45, 46] mentioned that Y2O3 could react with water to form α-Y(OH)3 due to the action of water and pressure during the friction process. However, compared with the polished sample whose -OH peak most likely generated during the pretreatment process before the tribological tests, no new -OH peak formed during the friction process. This hypothesis was demonstrated by immersing mullite/3Y-TZP in artificial saliva for 5 days and then performing a Raman test, and the spectrum was shown in the blue curve. No significant changes happened in the intensity and Raman shift of the -OH peak, which was similar to those of the previous two results. These results indicated that mullite/3Y-TZP had good stability during friction and artificial saliva environments.
3.3.2 Wear appearances
Different observation methods were selected based on the different characteristics of the wear surface. 3D morphologies of the wear surface of the tooth enamel were measured by LSM700, and they showed a flat wear surface with almost no grooves or undulations, which can be obtained from Fig. 12(a). Surface roughness of the wear surface was measured to be only 4.166μm, even though it was not as smooth as that of the original surface of tooth enamel, and the curve at the bottom of Fig. 12(a) also illustrated no significant fluctuation. This phenomenon may be caused by the formation of smooth film on the wear surface due to the deformation of the debris falling from tooth enamel under action of artificial saliva and load during the friction process, which indicated that no significant abrasive wear had occurred on the surface of enamel, providing a basis for the following wear mechanism. Mullite/3Y-TZP, as a counterpart, did not produce severe wear on its wear surface compared to enamel, as shown in Fig. 12(b). The 3D morphologies measured by VHX-500 demonstrated that the worn area of mullite/3Y-TZP was very shallow, and the vertical height difference between the centre of the pit and the unworn surface was only 8.37μm, as shown by the curve below Fig. 12(b). This was mainly because both mullite/3Y-TZP and enamel have high hardness, it was difficult to destroy the surface structure of mullite/3Y-TZP and cause serious wear during the friction process with the lubricating of artificial saliva,.
Fig. 13(a) represents overall appearance of the wear surface of enamel at low magnification, and it showed a flat surface without significant scratches, which was consistent with its 3D morphologies. Fig. 13(b) shows the morphology obtained by local magnification of Fig. 13(a). A small amount of abrasive debris adhered to the flat surface, the analysis of EDS showed that they were calcium-phosphorus compounds (hydroxyapatite) derived from the surface of enamel, and contain a small amount of elements of mullite/3Y-TZP and artificial saliva, as shown in the upper right corner of Fig. 13(b). Fig. 13(c) shows a portion of the edge of the worn region, and it was apparent that delamination of the layered debris had occurred in this region, and the size was about 25μm, while cracks presented around it, which was closely related to stress-induced fatigue fracture.
Fig. 13(d) represents overall appearance of the wear surface of mullite/3Y-TZP at a low magnification, and only a very small amount of scratches presented on the flat surface. A small amount of abrasive debris adhered to the surface of the counterpart, and EDS analysis showed that their compositions were the same as that of the debris on enamel surface, as shown in Fig. 13(e). Unlike the wear surface of enamel, the surface of mullite/3Y-TZP did not exhibit large-scale peeling, even though some scratches and microcracks appeared, as shown in Fig. 13(f), which was closely related to the mechanical properties of mullite/3Y-TZP. In addition, the pinning effect of mullite and good combination between the ZrO2 particles reduced the likelihood of particle flaking.
Similar study of Wang et al . (as shown in Fig. 14) can be compared with the study in the text. Fig. 14(a-c) represents the wear surface morphology of enamel after sliding against polished zirconia ceramic, and the enamel surface generated micro cracks and a layer of wear debris, which was almost the same as the results of this study, indicating that the tooth enamel had appeared fatigue wear; What was different from the results of this experiment was that the wear surface of zirconia ceramic had obvious particle shedding phenomenon (as shown in Fig. 14(d)), that is, abrasive wear had occurred on the surface of zirconia ceramic, which was closely related to the brittleness of zirconia ceramic. Because mullite/3Y-TZP used in this experiment had better fracture toughness, the nailing effect of mullite and alumina particles reduced the possibility of zirconia particles falling off . Fewer particles such as hard zirconia would reduce the damage to the friction couple during the friction process. Therefore, mullite/3Y-TZP is more suitable as dental materials than pure zirconia ceramic.
In order to further analyze element distribution of the wear surface of mullite/3Y-TZP, EPMA analysis was performed, as shown in Fig. 15, where Al and Si elements were not indicated. As shown in Fig. 15(a-d), internal components Zr and Y of mullite/3Y-TZP were detected, and it clearly proved that Y element distributed uniformly in ZrO2, which played a role in stabilizing t-ZrO2. In addition to the components of ceramic matrix, there were also Ca and P elements on the wear surface, as shown in Fig. 15(e-f), indicating that there was wear debris from enamel presented on the wear surface, which can be found in the grooves in combination with Fig. 15(a). The representative elements of artificial saliva, such as Na, Cl, etc. also existed on the wear surface, and their distribution was consistent with that of Ca and P elements besides a small amount of them evenly distribute in other areas, which was mainly because of the adsorption of wear debris to artificial saliva. It was apparent that the filling of grooves and cavities with wear debris and artificial saliva can significantly lubricate the wear surface and reduce friction resistance.
3.3.3 Wear mechanism
The shape of the wear debris provides a reliable clue to the wear mechanism of the specimen. Fig. 16 shows morphologies of the wear debris obtained after tooth enamel sliding against mullite/3Y-TZP in artificial saliva, from which abrasive grains and layered wear debris with different sizes can be seen. Abrasive grains showed small sizes with obvious aggregation and mutual adhesion, while the layered debris had two kinds of morphologies: (Ⅰ) large layered debris with a rough surface, as shown in Fig. 16(a-b), and (Ⅱ) layered debris with a smooth surface, as shown in Fig. 16(c-d). On the surface of the first kind of lamellar debris, there were aggregated fine particles and obvious microcracks, which was caused by the aggregation of abrasive debris under the combined action of artificial saliva and load during the friction process, or directly from the lamellar shedding on the surface of the enamel. The second kind of lamellar debris had clear outline with distinctly straight boundaries and microcracks, indicating that they were mainly caused by brittle fracture. The elements of these wear debris were shown in Fig. 17(a-d), it can be seen that the mass ratio of three elements of calcium, phosphorus and oxygen in the wear debris was high, reaching 26%, 18% and 24% respectively, which was consistent with the composition of enamel, and they were distributed almost every debris combined with Fig. 17(a). In addition, the surface of the wear debris was evenly distributed with a small amount of debris from the surface of the mullite/3Y-TZP and elements in the artificial saliva, as shown in Fig. 17(e-i), which indicated that only slight wear occurred in mullite/3Y-TZP, thus it was difficult to find large pieces of ZrO2 particles in the wear debris. The filling and lubrication of the wear surface by artificial saliva reduced the frictional resistance and mass loss, and result in the adhesion of elements, such as Na and K on the surface of the enamel and counterpart.
The morphologies of the wear debris combined with the character of the wear surface indicated that enamel mainly experienced fatigue wear. Because enamel on the surface of tooth is hard and brittle, repeated friction and load will cause stress concentration in the contact part, resulting in fatigue fracture . In addition, the shape of the edge portion of the wear surface indicated that adhesive wear happened locally (Fig.13(a)), which was due to the abrasive was wetted and pressed to form film on the wear surface, and then the film was peeled off due to repeated friction for a long time. Thereby, an adhesive wear zone was formed.
However, whether it was from the mass loss and wear surface morphologies of mullite/3Y-TZP or the elemental analysis of the wear debris, mild wear of mullite/3Y-TZP could be obtained. The uniform distribution of Zr, Al and Si elements in the wear debris indicated that mullite/3Y-TZP did have a slight particle flaking phenomenon and only occur mild abrasive wear, which was also consistent with a few scratches on its wear surface. The lubrication and cooling effects of the wear debris and artificial saliva which filled in the pits of the wear surfacec of mullite/3Y-TZP maintained the entire friction process at an extremely low mass loss during the repeated tribological tests. Because the hardness of the enamel was lower than that of mullite/3Y-TZP and artificial saliva had lubrication and cooling effects, it was difficult to cause extensive fatigue wear on the composite ceramic. In addition, almost no phase transformation of ZrO2 was obtained by the previous XRD analysis during the friction process, and Raman analysis found no significant reaction between Y2O3 with water, indicating that the internal structure of mullite/3Y-TZP was retained, which played an important role in retaining high mechanical properties. This phenomenon showed that mullite/3Y-TZP had good stability and wear resistance in the human oral environment.