Dentin contamination during repair procedures: A threat to universal adhesives?

Abstract Objective This study examined the influence of surface contamination during repair procedures with hydrofluoric acid, silane, or ammonium polyfluoride on the bond strength of universal adhesives to dentin using different etching modes before and after thermocycling. Materials and Methods Dentin surfaces of human molars were contaminated in different ways (silane, hydrofluoric acid, ammonium polyfluoride, and no pretreatment as control) followed by application of a universal adhesive (etch & rinse or self‐etch mode). After a composite build‐up was placed onto each tooth, sticks for the microtensile bond strength (MTBS) test were sectioned. Half of the sticks were tested after water storage for 24 h, the other half after thermocycling (15,000 cycles, 5/55°C). The MTBS data were analyzed statistically by using the Kolmogorov–Smirnov test, one‐way analysis of variance, and Tukey HSD test (p < 0.05). The fracture patterns of all specimens were evaluated and analyzed using a χ 2 test. Results Dentin contamination with a multifunctional silane does not influence microtensile bond strength irrespective of aging. Contamination with hydrofluoric acid or an ammonium polyfluoride primer leads to a significantly lower bond strength after aging, irrespective of the etch mode. Conclusion Dentin contamination with hydrofluoric acid or ammonium polyfluorides during repair procedures should be avoided, as they appear to decrease the bond strength of universal adhesives.

restorative materials (Demirel & Baltacioğlu, 2019;. Currently, the proportion of repairs of existing restorations relative to all treatments performed in daily practice is rising . The advantages of repairs include the facts that they are minimally invasive, require less time in certain cases and allow clinicians to avoid the risk of damaging the pulp (Brunton et al., 2017). However, many repairs involve not only the restorative material but also to a variable extent the tooth structure. As the complexity of these clinical situations increases, it is often difficult to decide which approach to use, especially as there exist a variety of different recommendations for repair procedures in the literature, which are not consistent (Altinci et al., 2017). For instance, the surface pretreatment of ceramic for adhesive cementation and repair depends on its composition (Awad et al., 2017). For a successful cementation of a glassceramic, for example, lithium disilicate, its surface has to be etched with hydrofluoric acid and coated with a silane (3-methacryloxypropyltrimethoxysilane) (Tian et al., 2014).
Applying only a universal adhesive after hydrofluoric acid etching, even if it contains a silane, cannot be recommended due to the instability of the silane component resulting in lower bond strengths, as compared to the conventional method, that is, hydrofluoric acid etching and silanization (Guimaraes et al., 2018;Maier et al., 2019;Yoshihara et al., 2016). When repairing restorations in clinical practice, adjacent enamel and dentin surfaces may be "contaminated" with agents used for restoration pretreatment. In the case of glass ceramics, this may apply to both hydrofluoric acid, which can be used intraorally for ceramic repairs in the form of buffered 9% hydrofluoric acid with strict safety precautions, and silanes (Hickel et al., 2013;. Dentin contamination with a silane before the use of a universal adhesive does not seem to influence initial bond strength to dentin, neither in the self-etch nor in the etch & rinse mode (Chen et al., 2017;Kanzow et al., 2020). However, enamel or dentin contamination with hydrofluoric acid should be avoided because it negatively affects the bond strength to the tooth structure (Loomans et al., 2010;Saracoglu et al., 2011).
The use of ammonium polyfluoride-based primers is a new option for pretreating glass-ceramic surfaces. Prior etching with hydrofluoric acid and subsequent application of silane is not necessary when using this pretreatment method, as the surfaces are etched by ammonium polyfluoride and silanized by trimethoxysilylpropyl methacrylate. These self-etch "glass-ceramic primers" provide bond strengths comparable to those achieved by conventional pretreatment with hydrofluoric acid and silanes, and therefore seem to be an interesting option for repairs of glass ceramic (Al-Harthi et al., 2018;El-Damanhoury & Gaintantzopoulou, 2018;Maier et al., 2019).
The null hypotheses of this investigation were set forth as follows: (1) Surface contamination with hydrofluoric acid, silane or ammonium polyfluoride does not influence the bond strength of universal adhesives to dentin.
(2) The bond strength of universal adhesives to dentin does not differ between the self-etch and etch & rinse mode, irrespective of the kind of contamination.
(3) The bond strength is not influenced by aging.

| MATERIALS AND METHODS
For this study, 40 caries-free human molars were used after obtaining the patients' informed consent. The teeth were stored in 0.5% chloramine-T solution at 8°C for no longer than 6 months before testing. The use of human teeth for bond strength testing has been approved by the Ethics Committee of Hannover Medical School (no. 2092-2013). The teeth were randomly divided into eight main groups with each five teeth and kept moist throughout the testing period ( Figure 1). The main groups were further subdivided into groups that were tested after 24 h water storage versus groups that were aged via thermocycling. All teeth were cleaned from debris and embedded in gypsum parallel to the tooth axis. A low-speed saw (IsoMet Low-Speed Saw; Buehler, Esslingen, Germany) was used under constant water cooling to separate the coronal part of the crown from the tooth at a right angle and expose the dentin. The cut was created 1 mm below the deepest part of the fissure. Therefore, a u-shaped device was placed in the deepest part of the fissure in order to determine the height of the cut. The cutting process is illustrated in Figure 2. The dentin surface was then checked for any remaining enamel areas with dental loupes at ×4 magnification. When the surface still included enamel, slightly more tooth structure was removed until only dentin was exposed. The dentin surface was then roughened with moist abrasive paper (600-grit, SiC Grinding Paper; Buehler, Esslingen, Germany) and thoroughly rinsed with water to create a clinically relevant smear layer.
Afterwards, the dentin surfaces were pretreated depending on the groups the teeth had been assigned to. In half of the test groups, the universal adhesive was used in the self-etch mode, and in the other half in the etch & rinse mode. The universal adhesive was F I G U R E 1 Test and control groups with their respective surface pretreatment (adhesive application [all groups] and contamination [test groups only]). Number of teeth per main group (n), resulting number of sticks per test group (n) applied as described in Table 1. In the self-etch mode, the adhesive was applied to the dentin, carefully rubbed onto the surface for 20 s, and gently air-dried for approx. 5 s; in the etch & rinse mode, the dentin was etched with 35% phosphoric acid for 15 s, rinsed with water for 15 s and gently dried, before adhesive application. The surfaces were "contaminated" with silane, ammonium polyfluoride or buffered hydrofluoric acid in the test groups ( Figure 1). In the selfetch mode, the dentin was contaminated directly before adhesive application, and in the etch & rinse mode, between phosphoric acid etching and adhesive application. In the control groups (ER and SE), the dentin did not undergo any contamination. The universal adhesive was light-cured with an LED unit (Bluephase; Ivoclar Vivadent, Schaan, Liechtenstein) for 10 s (output > 1000 mW/cm 2 ). Before each polymerization cycle, the light output of the unit was checked with a radiometer (Bluephase Meter; Ivoclar Vivadent, Schaan, Liechtenstein). Then, a composite build-up (Z100 MP Restorative, shade A3, 3M Oral Care; 3M Deutschland GmbH, Seefeld, Germany) was placed onto each tooth in three layers of 2 mm. Each layer was light-cured F I G U R E 2 Illustration of the cutting procedure. Initially, a u-shaped device was placed in the deepest part of the fissure in order to determine where the cut had to be placed. The upper part of the crown was cut was 1 mm below the deepest part of the fissure. After pretreatment of the dentin surface according to the respective treatment protocol and placing of the composite build-up, five cuts in x-and y-direction were made resulting in 16 sticks per tooth

| Fracture analysis
The fracture analysis of all specimens tested by μTBS before and after thermocycling showed overall 64.0% adhesive, 5.7% cohesive, and 30.3% mixed fractures. After thermocycling, more adhesive fractures (52.2% before TC vs. 75.7% after TC) and fewer cohesive (7.6% before TC and 3.9% after TC) and mixed (40.2% before TC and 20.4% after TC) fractures were present (Figure 4). The differences in fracture patterns before and after aging were significant (p < 0.001).

| SEM
The results of SEM are shown in Figures 5 and 6   Regarding the methodology, the microtensile bond strength test was used to determine the group-specific effects of different dentin pretreatments. Some groups in our study have a high standard deviation (cf. Table 2 and Figure 3). A high standard deviation might show a wider distribution of the measured values. This in turn could be associated with a higher uncertainty of the measured mean value.
Nevertheless, a high number of specimens in each group (n = 40) was chosen in order to increase the validity of the method.
During adhesive procedures, dentin poses a greater challenge to adhesive bonding than enamel because of its composition (on average: 50% by vol. inorganic material, 30% by vol. collagen, and 20% by vol. water), and any negative influence resulting from dentin surface contamination may considerably affect bond strength (Perdigão, 2010 (Yoshida et al., 2004). For optimal bonding of 10-MDP adhesive systems to dentin, manufacturers recommend carefully "rubbing" the adhesive into the surface to maximize monomer infiltration (Carrilho et al., 2019). Based on this recommendation (Moritake et al., 2019;Saito et al., 2020) the adhesive used in this study was also actively rubbed into the dentin surface with a brush for 20 s, gently air-dried for 5 s, and light-cured. This universal adhesive contains a silane, so it can theoretically bond to glass-based ceramics (Matinlinna et al., 2018). However, it should be noted that silanes added to adhesive systems only have a positive influence on bond strength immediately after their addition, as they will be inactivated by a dehydration and condensation reaction after a very short time Yoshihara et al., 2016). This is why a separate surface pretreatment with a silane is recommended when repairing glass ceramics, as any degradation processes at the interface can be prevented (Elsayed et al., 2017;Silva et al., 2020;Yoshihara et al., 2016). Concerning the intraoral repair of direct and indirect restorations, there have not been any consistent protocols to date, and the literature holds numerous and various approaches . During intraoral repair procedures, the tooth structure may be "contaminated" with a wide variety of repair agents. In a study conducted by Chen et al. (2017), no negative effect on initial shear bond strength to dentin was found, neither in the self-etch nor in the etch & rinse mode, when using a universal adhesive after contamination with a silane (Chen et al., 2017 universal adhesive in combination with a silane might be advantageous when glass-ceramics need to be repaired and dentin is involved, as any contamination of the dentin surface with a silane will not negatively affect bond strengths (Table 2 and Figure 3). However, as the repair of glass-ceramics not only requires silanization, but also the creation of surface roughness to increase the surface energy, silanization cannot be considered as an isolated factor, but only as a cofactor together with other pretreatment methods. Dentin contamination with ammonium polyfluoride (MEP) or hydrofluoric acid (PE) did not influence initial bond strength in our study. After thermocycling, however, the μTBS values after contamination with MEP or PE were significantly lower than those of the control groups. This was found irrespective of the adhesive application mode used (selfetch vs. etch & rinse, cf. Table 2 and Figure 3). Therefore, the first null hypothesis has to be partially rejected because contamination with ammonium polyfluoride or hydrofluoric acid significantly influences bond strength. Since this effect occurred after thermocycling for these two agents, the third null hypothesis also has to be partially rejected. Similar results for dentin contamination with hydrofluoric acid after phosphoric acid etching were found by Saracoglu et al. (2011). The bond strength to contaminated dentin was significantly lower compared to the control group, but only 1000 thermocycles were performed, much fewer than in our study, and a classic twostep etch & rinse adhesive was used instead of a universal adhesive.
In contrast, another investigation showed that contamination with hydrofluoric acid does not significantly influence bond strength when a universal adhesive is used in the etch & rinse mode. In the self-etch mode, however, bond strengths to contaminated dentin were significantly lower, which corresponds with the results of our study (Kanzow et al., 2020).
The results of the fracture analyses support the findings of the 61.5% vs. 90.5%). Therefore, a high amount of adhesive failure might be an indicator for the impact of the factors "contamination" as well as the "aging." Regarding the SEM analyses, the smear layer was not removed from the dentin surfaces by 60-s "contamination" with hydrofluoric acid in our study (cf. Figures 5a vs. b and 6a vs. b), which is in consistence with the results published by Szep et al. (2000), who found an amorphous layer of the undefined precipitate. After etching dentin with phosphoric acid, in contrast, the smear layer was completely removed (cf. Figures 5e vs. f and 6e vs. f). The significantly lower bond strengths measured after thermocycling in comparison to the control groups cannot be attributed to the dentin structure visible with SEM. Nevertheless, EDX analyses showed that the application of hydrofluoric acid to nonetched dentin leads to an accumulation of fluoride ions in the surface (Szep et al., 2000). This accumulation might cause the fluoride ions to react with the residual calcium ions of the dentin surface to form CaF 2 (Szep et al., 2000). As 10-MDP also bonds to residual calcium ions, the reaction mechanism initiated by hydrofluoric acid application might inhibit this chemical bonding process. Loomans et al. (2010) also showed that bond strength to dentin significantly decreases after contamination with hydrofluoric acid and found certain changes in contaminated dentin via When dentin had been etched first with phosphoric acid and then with hydrofluoric acid, the hybrid layer was thinner, and there were mineral precipitates below the hybrid layer and areas with distinct nanoleakage (Loomans et al., 2010). These changes occurring at the microscopic or molecular levels might explain the lower bond strengths.
F I G U R E 4 Fracture analysis of the control and test groups before and after thermocycling, a distinction between adhesive, cohesive and mixed fractured was made F I G U R E 5 (a-h) Scanning electron microscope images of dentin surfaces at ×1000 magnification; (a-d): non-etched, smear-layer-covered dentin without contamination and after contamination with hydrofluoric acid (PE), silane (MP) or ammonium polyfluoride (MEP); (e-h) analogous as described for (a-d), but dentin etched with 37% phosphoric acid, smear layer completely removed, dentinal tubules exposed. The use of ammonium polyfluoride primers for intraoral repairs of glass-based ceramics could be an alternative to hydrofluoric acid.
However, this kind of contamination also led to significantly lower bond strengths after thermocycling, as compared to the controls, irrespective of the adhesive application mode (self-etch vs. etch & rinse, cf. Table 2 and Figure 3). In a shear test conducted by Kanzow et al. (2020) specimens that were contaminated with an ammonium polyfluoride primer and artificially aged did not differ from the control group when the etch & rinse mode was used, but for the self-etch mode, a significant decrease in bond strength, comparable to our results, was found. Scanning electron microscopy showed that, without prior phosphoric acid etching, the smear layer was superficially removed by the use of the one-component primer and the dentinal tubule openings were visible, but apparently sealed with precipitates (cf. Figure 6a vs. d). Similar results for this test group were described by Kanzow et al. (2020). In our study, the bond strength of the universal adhesive after contamination with the ammonium polyfluoride primer was significantly lower after artificial aging than that of the controls in both the self-etch and the etch & rinse modes, so an interaction between the fluorides contained in the one-component primer and the dentin surface at the molecular level, similar to the process described for hydrofluoric acid, may explain this decrease. However, as no other study in the literature describes such an interaction between ammonium polyfluoride primers and dentin, additional research will be necessary to clarify this phenomenon.
The second null hypothesis, stating that the bond strength of universal adhesives to dentin does not differ between the self-etch and etch & rinse modes, irrespective of the kind of contamination, can be confirmed. When the dentin surfaces were contaminated with the same agent, there was no difference between the self-etch and etch & rinse modes, neither initially nor after aging.

| CONCLUSION
When repairing restorations intraorally, it is mandatory to differentiate between repairs limited to the restoration material and repairs involving the tooth structure. Dentin contamination with a multifunctional silane before the use of a universal adhesive does not influence bond strength irrespective of aging. Contamination with hydrofluoric acid or an ammonium polyfluoride primer leads to a significantly lower bond strength after aging. As the surface of glassbased ceramic restorations has to be pretreated for establishing an adhesive bond, dentin contamination with substances involved in the repair procedure should be avoided.

| CLINICAL SIGNIFICANCE
Repair procedures help to maintain the long-term stability of dental restorations. During repair procedures, dentin contamination with hydrofluoric acid or ammonium polyfluoride primers should be avoided, as they impair the bond strength of universal adhesives after artificial aging.

ACKNOWLEDGMENTS
The authors do not have any financial interest in the companies whose materials are included in this article.