1. Introduction
In dentistry, the introduction of computer-aided design/computer-aided manufacture (CAD/CAM) systems significantly increased the use of all-ceramic restorations [
1]. Zirconia is one of the primary reinforced ceramic substrates used in the CAD/CAM process [
1]. This ceramic is in popular use mainly due to its exceptional mechanical properties, as well as its biocompatibility and resistance to corrosion [
1,
2,
3], although its limited esthetics make it difficult to match the restorations with existing dentition [
3,
4].
Unlike silica-based ceramics, zirconia is a silica-free polycrystalline ceramic, which makes it resistant to traditional hydrofluoric acid etching and silane coupling agent treatments [
5]. The bond strength and durability of various bonding methods to zirconia, including several innovative ceramic surface treatments, have been investigated. However, a standardized bonding protocol for zirconia-based restorations has not yet been established [
5]. It has been shown that sandblasting pretreatment results in relatively high bond strengths to zirconia when combined with resin materials containing phosphate monomers, such as 10-methacryloyloxydecyl dihydrogen phosphate (MDP) [
6,
7]. This mechanical surface conditioning can enhance resin–zirconia adhesion by cleaning the ceramic surface, increasing the roughness and bonding area, and enhancing the wetting of resin materials [
7]. Sandblasting with silica-coated alumina particles (tribochemical silica-coating or silicatization) to improve the resin bonding with zirconia is another method used in dentistry [
8,
9]. The presence of silica on the surface is a prerequisite for the formation of durable siloxane bonding [
8]. A high blasting pressure can result in the superior attachment of silica-coated alumina particles on the zirconia surface [
9]. However, the silica particles do not deeply penetrate into the sandblasted zirconia and only loosely cover the surface [
8,
9].
Sandblasting involves impacting the substrate surface with hard particles at high velocities, thereby eroding the material and leaving a roughened surface with high wettability [
10]. In the case of zirconia ceramic, sandblasting induces undesirable tetragonal-to-monoclinic (t–m) phase transformation and microcrack formation at the near surface zone [
10]. It also creates residual stress in the zirconia surface, mainly due to local irreversible deformation at the impact sites [
10]. Chintapalli et al. [
11] showed that the extent of subsurface t–m transformation and substrate damage caused by sandblasting were dependent on sandblasting conditions (abrasive particle size and pressure). They concluded that mild sandblasting was more beneficial than the harsh one since the former induced limited damage. Okada et al. [
12] found that there was an optimal blasting pressure range to increase the zirconia flexural strength, although surface roughness increased along with blasting pressure.
Residual stress can affect the bond strength of the coating layer on a substrate [
13]. Yang and Chang [
13] reported that compressive residual stress weakened the adhesion at the interface of the hydroxyapatite (HA) coatings and the titanium. Similarly, the residual stress formed on the zirconia surface after sandblasting may affect the resin bond strength to zirconia. However, there are still relatively few studies available on the direct influence of residual stress induced by sandblasting the zirconia surface on the resin bonding to zirconia. The purpose of this study was to investigate the influence of different sandblasting conditions on the resin bonding to zirconia ceramic in terms of residual stress. The null hypotheses were that the different sandblastings would not result in (1) different residual stresses of the zirconia surfaces nor (2) different resin bond strengths.
3. Results
Figure 1 shows the SEM images of the polished and sandblasted zirconia surfaces and corresponding EDS spectra. The SEM images show the smooth surface by polishing (a) and the roughened surfaces by sandblasting (b–e). The aluminum element was not detected by EDS on the polished surface. In contrast, the sandblasted zirconia surfaces showed a small amount (1.85–2.68 at%) of remaining aluminum even after ultrasonic cleaning.
The surface roughness values (
Ra and
Rz) and water CAs (before and after roughness correction) for the polished and sandblasted zirconia specimens are summarized in
Table 1. The polished surface exhibited very low
Ra and
Rz values. As the alumina particle size and pressure increased, both the
Ra and
Rz values gradually and significantly increased (
p = 0.023). The measured CAs significantly decreased after sandblasting (
p = 0.042) and gradually decreased by increasing the abrasive particle size and pressure. The 110M-4B group showed significantly lower CA than the 50M-2B and 50M-4B groups (
p = 0.029). After roughness correction, however, there were no significant differences in CA among all the sandblasted groups (
p > 0.05).
Figure 2 and
Figure 3 show the XRD patterns of the polished and sandblasted zirconia surfaces. The spectrum of the polished specimen exhibited only the t-phase characteristic peaks (t (101), t (002), and t (110) peaks), without the appearance of the m-phase. After sandblasting, the m-phase peak (m (111)) newly appeared at an angle 2θ of 28.2°. The m-phase/(t-phase + m-phase) ratios gradually increased as the particle size and pressure increased.
The plots of strain versus sin
2ψ for the polished and sandblasted zirconia surfaces and the residual stresses calculated from the linear dependence are shown in
Figure 4 and
Figure 5, respectively. The polished zirconia surface exhibited a slight residual stress (50 MPa). The larger alumina particle size and higher sandblasting pressure resulted in higher compressive residual stress (average values ranged from 451 to 905 MPa).
Table 2 summarizes the SBS values of the resin cement to the various zirconia ceramic surfaces before and after thermocycling. Regardless of the thermocycling condition, the polished zirconia group showed significantly lower SBS values than the sandblasted groups (
p < 0.001). At 0 cycles, the 50M-4B and 110M-4B group values were significantly higher those of the 50M-2B and 110M-2B group (
p < 0.05). At both 10,000 and 30,000 cycles, however, there were no significant differences in SBS among the sandblasted zirconia groups (
p > 0.05). The polished zirconia surfaces showed only adhesive failure, irrespective of the thermocycling condition. Before thermocycling, the distribution of the failure modes for the sandblasted surfaces was 20–30% adhesive failures and 70–80% mixed failures. No cohesive failure within the resin cement was observed. After thermocycling, all sandblasted zirconia specimens exhibited exclusively adhesive failure after debonding.
4. Discussion
The purpose of the present study was to investigate the effect of four different sandblasting conditions (two blasting particle sizes and two pressures) on the resin bonding to zirconia ceramic in terms of residual stress. Based on the findings, the first null hypothesis, that the different sandblasting conditions would not result in different residual stresses of the zirconia surfaces, was rejected (
Figure 5). The resin-bonded zirconia specimens were aged by thermocycling after a week of water saturation [
15]. Thermocycling inducing the thermal stress and hydrolytic effect is widely used to evaluate the adhesion durability [
16]. As an estimate of approximately 10,000 cycles per year is provisionally suggested [
17], the 10,000 and 30,000 cycles used in this study thus represent approximately 1 and 3 years of clinical use, respectively. Before thermocycling, different SBS values were obtained by the different sandblastings (
Table 2). However, no significant differences in SBS were detected among any of the sandblasting groups after thermocycling. Thus, the second null hypothesis was partially rejected.
It is known that the sandblasting of zirconia conditions the surface to increase the roughness, as well as to clean and activate the surface [
7]. However, the alumina particles used for sandblasting may remain on the zirconia surface even after ultrasonic cleaning in isopropyl alcohol and water for 10 min each. As shown in
Figure 1, only low levels of aluminum element were detected on all the ultrasonically cleaned zirconia surfaces after sandblasting although the detected amount gradually increased with greater abrasive particle size and pressure. Therefore, the presence of only a small amount of alumina particles on the zirconia surfaces does not seem to have impaired the resin bonding. In the case of the tribochemical silica coating, in contrast, the silica particles should remain even after ultrasonic cleaning to create durable siloxane bonding [
8]. However, Lorente et al. [
18] suggested that the silica particles uniformly cover the zirconia surface after the sandblasting, but the subsequent ultrasonic cleaning nearly completely removes the loosely bound particles. Therefore, the use of alumina particles at various blasting pressures might be preferred to that of silica-coated alumina particles [
9].
The surface roughness values of the zirconia specimens were dependent on the sandblasting pressure as well as the alumina particle size (
Table 1). Likewise, the measured CA values gradually decreased by increasing the particle size and pressure. The Young CA assumes that the surface is topographically smooth and chemically homogenous [
14]. This is not true in the case of most real surfaces, especially roughened surfaces by sandblasting. In order to obtain the actual CAs, the CAs measured on the sandblasted zirconia surfaces were roughness-corrected using the
Sdr values of the surfaces [
14]. After roughness correction, the sandblasted surfaces did not exhibit significantly different CAs. This finding indicates that the more aggressive sandblasting treatments did not necessarily produce more chemically activated zirconia surfaces, the higher roughness values notwithstanding.
The XRD analysis showed that the m-phase newly formed on the sandblasted zirconia surfaces increased with an increase in particle size and blasting pressure (
Figure 2 and
Figure 3). The shape and diameter of abrasive particles are of primary importance in the sandblasting process [
9]. The m-phase/(t-phase + m-phase) ratios increased by 83% and 44%, respectively, when the particle size was fixed and the blasting pressure was increased (50M-2B to 50M-4B and 110M-2B to 110M-4B). On the other hand, when the particle size was increased and the pressure was decreased (50M-4B to 110M-2B), the increase in the ratio was only 9%. These findings indicate that the formation of the m-phase induced by sandblasting depends more on the blasting pressure than the particle size. The polishing of the zirconia surface resulted in the creation of a slight compressive residual stress (
Figure 5), without the formation of the m-phase. The compressive residual stresses gradually increased by the increase in aggressiveness of the sandblasting (larger particle size and higher pressure). Compressive residual stress in the surface layer tends to be advantageous because it tends to close cracks and slow crack propagation. Sandblasting the zirconia surface induces protective compressive residual stresses from the t–m phase transformation and, as a result, may increase the mechanical strength of the ceramic [
12]. Gentle sandblasting (small-sized abrasive and low blasting pressure) may be beneficial because it induces limited damage, confined to the transformed region where a compressive stress field exists [
11]. In some cases, zirconia ceramic may be strengthened by removing the weak grains and grinding traces using such gentle sandblasting treatment [
19]. However, aggressive sandblasting (large abrasive particle and high blasting pressure) can weaken the ceramic by creating new surface flaws and microcracks which cannot be counteracted by the compressive residual stress field [
11,
12,
19].
The combination of mechanical and chemical procedures is essential to the efficacy of the resin bonding procedure to zirconia [
1]. Bisphenol A diglycidyl methacrylate (bis-GMA)-based resin material cannot form a durable adhesion to zirconia even after roughening by sandblasting [
20]. High and durable bond strength can be achieved when an MDP-containing primer or an MDP-containing resin is applied to sandblasted zirconia [
6,
7]. It has been hypothesized that phosphate groups in the MDP molecule can potentially react with one or two zirconium atoms in zirconia crystals [
6,
21,
22]. In this study, the MDP-containing resin luting material Panavia F 2.0 was used. Without sandblasting after polishing, however, the initial bonding (average 6.2 MPa) was reduced to an average 1.9 MPa and 0 MPa after 10,000 and 30,000 thermocycles, respectively (
Table 2), with complete adhesive failure at the ceramic surface. These decreases may be attributable to the degradation of the resin luting material itself, the hydrolytic effect of water at the resin–zirconia interface, and mismatch between the thermal expansion coefficient between the resin and ceramic [
1,
23].
Sandblasting of zirconia increases the surface roughness and thus also the area for mechanical interlocking, resulting in acceptable micrometer scale roughness [
1]. At the same time, the treatment cleans the surface by removing any organic contaminants, improves the wettability, and allows resin luting agent to flow into the roughened surface more effectively [
7,
24]. Rougher zirconia surfaces may provide wider contact areas and microporosities for resin luting agents [
5]. Before thermocycling, the higher blasting pressure resulted in significantly higher SBS values within the same particle size condition (4B than 2B) (
Table 2). However, such an increase in the surface roughness and residual stress, depending on the sandblasting conditions, did not significantly affect the SBS values after thermocycling (either 10,000 or 30,000 cycles). The CA data showed that more aggressive sandblasting did not necessarily increase the surface wettability, as seen in the roughness-corrected values, notwithstanding the more increased surface roughness values (
Table 1).
Strong micromechanical interlocking can only be achieved when a resin luting material effectively penetrates and flows into the microretentions created on the sandblasted zirconia surface [
25]. However, resin luting material may not completely fill the microporosities of the sandblasted surface due to its viscosity [
26]. Thus, it is assumed that surface roughness exceeding the optimum for adequate micromechanical interlocking does not contribute to enhancement of resin adhesion to zirconia. Yang and Chang [
13] demonstrated the relationship between bond strength and residual stress when HA coatings were plasma-sprayed on a titanium substrate. As resin adhered to zirconia seems less intimate at the interface as compared to HA coating to titanium, the influence of residual stress of zirconia on resin bonding may be minimal. However, a combined use of an adhesive primer, which could potentially more effectively wet the microporosities of the sandblasted zirconia surface due to its lower viscosity, and a bis-GMA resin cement was not tested in this study. Therefore, further investigation is needed to confirm this speculation.
In this study, the effect of four different sandblasting conditions (two blasting particle sizes and two pressures) on the resin bonding to zirconia was investigated in terms of residual stress. The residual stresses created on the zirconia surfaces directly depended on the aggressiveness of sandblasting. However, such increased surface roughness and residual stress did not affect the bonding durability of the adhesive resin cement with relatively high viscosity. Although sandblasting is favorable for improved resin bonding to zirconia, it can affect the mechanical properties of the ceramic due to accelerated t–m phase transformation, which may put stress on the surfaces and cause fatigue in the structure that could lead to premature and catastrophic failure [
1]. Therefore, mild sandblasting of zirconia with small-sized abrasives and reduced blasting pressure is preferred to an aggressive procedure because increased surface roughness and residual stress do not directly affect the resin bonding durability. Future studies should include other resin cements with different viscosities to clarify the characteristics of resin–zirconia interface. It should also be noted that SBS test results tend to show high standard deviations due to uneven distribution of the stresses at the bonding interface [
27]. Therefore, the use of other bond strength testing methodologies, such as the microtensile bond strength test, might be considered when evaluating resin bonding and its durability to zirconia ceramic.