Partial Ceramic Crowns: Influence of Ceramic Thickness, Preparation Design and Luting Material on Fracture Resistance and Marginal Integrity In Vitro

Functional and esthetic restorations of extensively damaged teeth require materials that should be bio-compatible, mechanically stable for oral use and tooth-colored. In general, ceramic inlays and partial ceramic crowns (PCC), that is, restorations with one or more cusps being restored1–2 are clinically accepted alternatives to cast gold restorations and amalgam fillings.3–4 However, failures occur, and the predominant reasons for failures are fractures,5 one of the major failure mechanisms being damage accumulation.6 

Fracture toughness of dental ceramics has been mainly tested by basic laboratory, bending and indentation methods, for example, four-point bending test, biaxial flexure test, Knoop or Vickers indentation, where tensile stress is applied on the indented surface of the specimen until fracture.7–8 These methods use the acute single loading failure as the test methods; however, a lifetime of dental restorations is limited by the accumulation of contact damage during oral function, and the strength of dental ceramics are significantly lower after multi-cycle loading than after single-cycle loading.6,9 

Fatigue is described as a change of material characteristics over time under cyclic conditions.6,10 For dental ceramics, microscopic surface flaws and defects, which may develop as a result of thermal, chemical or mechanical processes, act as localized stress concentrators.11 Subcritical crack growth in ceramics is attributed to “corrosion-assisted” stress at the crack tip or at any pre-existing defect in the ceramic.12 Cracks have been shown to be sites of fracture initiation and, consequently, failure.13–15 

These problems have resulted in attempts to improve the mechanical properties of all-ceramic restorations, without compromising the aesthetics of the restoration or causing further damage to opposing teeth and other restorative materials.16 One approach is to use industrially prefabricated feldspathic ceramic, which is milled using computer-aided design and computer-aided manufacturing (CAD/CAM) device technology,17–18 that is, the Cerec system. This ceramic possesses a better structural homogeneity and fracture strength compared to laboratory-processed dental ceramic materials.9,19 Still, in stress bearing areas, the literature generally recommends a minimum ceramic thickness of 1–2 mm.20–21 However, this recommendation is mainly based on basic mechanical testing, without taking the clinical system into account.

Reasons for ceramic restoration fractures, however, may not only be related to mechanical properties of the ceramic, but also to the preparation design and corresponding outline of the restoration. Different preparation designs have been described in the literature. On one hand, preparation designs for PCC have been based on traditional concepts, utilizing a conventional retention form.22–23 On the other hand, designs that neglect retentive elements have been described, where retention of the all-ceramic restoration depends solely on the adhesive luting agent.24–25 Again, only limited information is available regarding the influence of the preparation design on the fracture resistance of PCC.

The third factor of interest in this context is the adhesive luting material. Adhesion to both tooth hard tissue and the ceramic have to be sufficient to guarantee high bond strength and guide masticatory forces from the restoration to the tooth.14 Gap formation may impair this bond and, thus, may promote fractures. Recently, a new self-etching luting material has been marketed. No information is available to date about gap formation and its influence upon fracture resistance of PCC for this and other luting composites.

Therefore, this in vitro study evaluated the influence of different ceramic thicknesses, preparation design and luting agents on ceramic fracture and marginal integrity of PCC. It was hypothesized that these three variables would affect fracture resistance and marginal integrity of PCC. Visible cracks in the ceramic have been used as early indicators of ceramic fracture.

Figure 1 summarizes the procedures followed in this study. Eighty extracted human molars, stored in 0.5% chloramine solution after extraction, were cleaned, mounted in Pattern Resin (GC Corporation, Tokyo, Japan) and stored in physiological saline solution until use. The molars were assigned randomly into two groups of 40 teeth each. Diamond burs (Cerinlay Set, Intensiv, Viganello, Lugano, Switzerland) with a taper of ∼4–6° in a high-speed handpiece with sufficient water cooling were used to perform one of the following preparations on each tooth (Figure 2):

Preparation of MOD-cavities (about 5.0 mm in width/about 4.0 mm in depth);

Preparation A: Coverage of functional cusps (lingual cusps in upper molars, buccal cusps in lower molars) about 1.5 mm/divergence angle of buccal and lingual walls ∼4–6° plus butt joint preparation about 1.0–1.5 mm/cusp coverage convergence angle ∼ 4–6°.

Preparation B: Horizontal reduction of functional cusps about 1.5 mm.

Non-functional cusps were not covered; proximal margins were placed 1 mm below the CEJ within cementum/dentin, the depth of the box being approximately 1.5 mm. Internally, rounded line angles were prepared.

The CAD/CAM method (Cerec 3 software version 1.0, Sirona, Bensheim, Germany) was used to construct and machine-mill the partial ceramic crowns. Mark II ceramic blocks (Vita, Bad Säckingen, Germany) were used to fabricate the PCC using the Cerec 3 system and corresponding software (Sirona Cerec 3 software version 1.0). After fabrication of the PCC, the ceramic thickness of the functional cusps was adjusted to (group 1) 0.5–1.0 mm and (group 2) 1.5–2.0 mm, using diamonds in a high-speed handpiece with sufficient water-cooling.

Following try-in and adjustment to the prepared cavities, the PCCs were finished with Komet finishing diamonds (Brasseler, Lemgo, Germany) and polished with Sof-Lex flexible discs (3M, St Paul, MN, USA), with decreasing roughness under sufficient water cooling. The PCC were inserted using one of the following luting material/bonding system combinations (10 specimens each per luting system, preparation design and ceramic thickness):

VL: Variolink II/Excite (Vivadent, Schaan, Liechtenstein) dual-cured resin composite luting agent.

RX: RelyX Unicem-Universal Aplicap (3M ESPE, Seefeld, Germany) self-adhesive universal resin cement.

The restorative procedures were performed in a device simulating proximal contact to adjacent teeth to match the clinical situation as closely as possible. Luting material was applied to the cavity surfaces following adhesive conditioning of tooth substances and PCC surfaces. The luting procedure of Variolink II is summarized in Table 1, in comparison with RelyX Unicem.

Excess luting material was removed prior to curing. Following insertion procedures, finishing was performed with Komet finishing diamonds (Brasseler, Lemgo, Germany), and the restorations were polished with Sof-Lex flexible discs (3M). Before thermocycling and mechanical loading (TCML), samples were stored in physiological saline solution at 37°C for 24 hours. The samples were exposed to thermocycling (5,000×5° at 55°C and 30 seconds/cycle) and mechanical loading (500,000×72.5 N at 1.6 Hz) simultaneously. Mechanical loading was performed by means of a cyclic (1.6 Hz) increase in pressure (72.5N) upon a metal stop, representing the opposing cusp. The metal stop was statically placed in the occlusal central fissure of the restoration.

Visible cracks, discernable under a light microscope (Wild Makroskop M420, Heerbrugg, Germany) at 12x magnification, were used as early indicators of ceramic fractures. Before and after TCML (Figure 3) digital images of the specimen were taken from each aspect (mesial, distal, occlusal, palatal). Digital images were also taken after dye penetration (Figure 3), in order to better visualize crack formation.

After TCML, microleakage at occlusal and palatal locations and for the ceramic- and tooth-interfaces were determined separately by means of dye penetration. Except for areas within 1 mm of restoration margins, the specimens were covered with nail varnish and placed in a 0.5% basic fuchsine solution for 16 hours at 37°C. After dye penetration with fuchsine, the specimens were cleaned, mounted onto stubs with acrylic resin and longitudinally sectioned in the mesio-distal direction into as many approximately 300 μm thick sections as possible4–8 using a rotating diamond saw (blade thickness 300 μm) (Innenlochsäge Leitz 1600, Leitz) and water cooling. The sections were approximately 300 μm thick, each section providing two sites for the evaluation of dye penetration. Digital images of the sections were recorded, and microleakage along both the tooth- and ceramic-restoration interfaces was recorded for the multiple sections using an image analyzing system (Optimas 6.1, Stemmer, Munich, Germany). The extent of dye penetration was expressed as a percentage of the entire length of the restoration wall (100% reference), as shown in Figure 4. Data for the occlusal and palatal location of each interface were pooled, because no statistical difference was observed between the two evaluation locations. Thus, for each section, four dye penetration measurements were recorded, rendering between 16 and 36 measurements/tooth (4 × 4–8 sections/tooth). The maximum value was selected for each tooth and used for further statistical analysis.

Digital images of the sections were recorded, and the actual ceramic thickness on the multiple sections was evaluated with an image analyzing system (Optimas 6.1, Stemmer, Munich, Germany). Two points for cavity design A and one point for cavity design B were measured and documented (Figure 5).

Non-parametric statistical analysis was considered appropriate for analyzing the data, because of the lack of normal distribution. Medians and 25% to 75% percentiles for each of the different criteria were determined for all interfaces separately. Statistical analysis was performed using Mann-Whitney U and Wilcoxon's rank sum tests (PC+ version 6.0 software) (SPSS, Chicago, IL, USA) for pairwise comparisons among groups. The level of significance was set at α=0.05. For evaluating the influence of preparation design, luting material, and ceramic thickness in general, the level of significance was adjusted to α*(k)=1–(1–α)1/k by application of the error rates method (k=n of paired tests performed). Differences in fracture rates were analyzed with the χ²-test (SPSS).

The actual ceramic thicknesses at the covered cusps areas measured after sectioning the teeth ranged from 0.6 mm to 0.9 mm (medians) for subgroups of group 1 and from 1.5 mm to 1.8 mm (medians) for subgroups of group 2 (Table 2). Within each of these two groups, there was no statistically significant difference between ceramic thicknesses of the subgroups with varying preparation designs and luting materials.

An example of visible cracks within a PCC is shown in Figure 6. One can see two crack lines, one which runs centrally from the functional cusp to the non-functional cusp, and the other, which runs at the mesial margin along the ceramic/luting agent interface. The results of the visible crack evaluations are shown in Table 3. All cracks of PCCs occurred after TCML. Seventeen out of 80 ceramic restorations cracked, 15 in group 1 (0.5–1.0 mm), and 2 in group 2 (1.5–2.0 mm). The difference between the two groups, group 1 (0.5–1.0 mm) and group 2 (1.5–2.0 mm), was statistically significant, as was the difference between the groups before and after TCML.

The results of dye penetration for RelyX Unicem are summarized in Figure 7A/Table 4. Generally, dye penetration was statistically significantly higher at the tooth/luting agent interface, with medians ranging from 36.9% to 52.7% compared to the ceramic/luting agent interface, with medians ranging from 7.3% to 14.6%. In general, there was no statistically significant difference in dye penetration between the ceramic thickness of group 1 (0.5–1.0 mm) and group 2 (1.5–2.0 mm). In detail, single pairwise comparisons revealed that, only in one case (ceramic/luting agent interface, group 1 [13.5%] vs group 2 [8.1%] in preparation design A) was there a statistically significant difference. Dye penetration was also, in general, not influenced by preparation design. In detail, single pairwise comparisons showed inconsistent results, with preparation design A having more penetration than design B in two cases and, in one case, it was the other way around.

The results of dye penetration for Variolink II (VL) are summarized in Figure 7B/Table 5. Generally, dye penetration was significantly higher at the ceramic/luting agent interface, with medians ranging from 36.0% to 78.4%, compared to dye penetration at the tooth/luting agent interface, with medians ranging from 12.3% to 19.3%. In general, no influence of ceramic thickness on dye penetration was found. Single pairwise comparisons showed, with one exception, no statistically significant difference between the data of dye penetration of groups 1 and 2 at both interfaces. The dye penetration of group 1 (36.0%) was statistically significantly lower than in group 2 (78.4%) at the ceramic luting agent interface of preparation design B. Dye penetration was, in general, not influenced by preparation design. In detail, pairwise comparison showed lower dye penetration data for preparation design A than for design B in two cases (ceramic/luting agent interface, preparation design A (68.4%) vs preparation design B (36.0%) for group 1 (0.5–1.0 mm) and ceramic/luting agent interface, preparation design A (40.3%) vs preparation design B (78.4%) for group 2 (1.5–2.0 mm).

In this study, the influence of three parameters (ceramic thickness, cavity design and luting material) on two endpoints (ceramic fracture resistance and marginal integrity) was evaluated.

In this study, the appearance of visible cracks in the ceramic has been used as an early indicator for ceramic fractures. Ceramics are brittle materials that are susceptible to failure beyond a critical stress, which is dependent upon internal and surface flaw distributions.7 Fractographic analysis of clinically failed “all-ceramic” restorations has proven that fracture had, in fact, always originated with the formation of cracks.15,26 

Basic laboratory methods use acute single loading failure, specimen loading until fracture,7–27 as test methods. Some authors,7 however, doubt whether prediction of all ceramic materials' performance or long-term success from such data is possible, but it is hypothesized that materials with improved fracture toughness should have better clinical success and longevity.27 In the patient situation, however, many fatigue-related fractures are observed.28 

TCML were used to induce fatigue. It is a method that is usually applied to simulate clinical situations, such as masticatory forces under wet conditions. It has been published that TCML fatigue significantly decreases the fracture resistance of all-ceramic full crowns.29 

Marginal integrity was evaluated by dye penetration, which is a commonly applied method to test the sealing of adhesive, tooth-bonded restorations. According to Kidd,30 microleakage is defined as the passage of bacteria, fluids or molecules between a cavity wall and the restorative material applied to it. Dye penetration after TCML was assessed in order to better simulate the in vivo situations. Krejci and others31 postulated that 120,000 in vitro loadings approximated six months of clinical use. The current study should thus simulate approximately 2–2.5 years of clinical use. A definite relationship, however, between in vitro dye penetration data and results from clinical (in vivo) testing still remains to be established, with problems probably arising due to evaluation deficiencies for both the in vitro and in vivo methods. However, in this study, dye penetration was also used to evaluate risk factors for in vitro crack formation in order to compare the two luting materials.

The results of this study show a definitive influence of ceramic thickness upon visible crack formation. In the literature, a number of recommendations have been published. Hansen25,32 indicated a ceramic thickness of at least 1.0 mm (occlusal) for inlay preparations. Jackson and Ferguson33 described an inlay-onlay preparation with occlusal reduction of at least 1.5 mm.

Based on these reports in the literature and in order to be on the safe side, a thickness of 1.5–2.0 mm was chosen for group 2. The rationale for the selection of 0.5–1.0 mm for group 1 was that, clinically, the final ceramic thickness may actually fall short of the required thickness, because the occlusal adjustment of the restoration is predominantly performed following adhesive luting procedures.

However, all these recommendations have been based on basic laboratory testing, without simulating the clinical conditions. To the best of the authors' knowledge, this is the first experimental indication in a clinical simulation experiment that, for PCC, a minimum thickness of at least 1.5 mm in stress-bearing areas is necessary.

In the current study, no influence of the two preparation designs on visible crack formation was found. Broderson22 describes the disadvantages for cavity design A, which create a high amount of stress points in the ceramic restoration and may lead to fracture. This assumption cannot be backed up by the results of the current study. However, the data are in agreement with van Dijken and others,24 who found no statistical differences among the four preparation types of partial and complete, posterior ceramic restorations employed in their five-year follow-up of dentin/enamel bonded ceramic coverage.

This study indicates no statistically significant influence of the luting materials tested upon visible crack formation. Variolink II is a clinically accepted adhesive luting material, and the results are in line with the accepted idea that adhesive luting is the basis for fracture prevention of dental ceramics.9,13 This apparently holds also for the new self-adhesive luting material RelyX Unicem.

In contrast to the results for visible crack formation, marginal integrity was shown to be independent of ceramic thickness but dependent upon the luting material. However, in line with the data for crack formation, marginal integrity was independent of preparation design. Apparently, there is no direct link between crack formation and marginal integrity.

The mechanism of adhesion seems to differ between the two luting materials, because dye penetration data for the two interfaces (ceramic/luting agent and tooth/luting agent) differed significantly. Variolink seems to better adhere to tooth substrate; whereas, RelyX Unicem has a better adhesion to ceramic. The results of VL are in agreement with other research.34–36 The authors reported that the adhesion of Variolink to the tooth/luting agent interface is more effective than to the ceramic/luting agent interface. Leakage may result in hydrolysis, loss of the restoration or fracture because of failure.

The tooth substrate in this study was mainly enamel, and it has been shown in several studies37–40 that adhesion of RelyX Unicem to enamel is less effective than to dentin. This may result in hypersensivity, recurrent caries, eventual pulpal pathoses41 and loss of the restoration. Morphological SEM and TEM evaluation of the interface by De Munck and others37 revealed that RelyX Unicem only superficially interacted with enamel. It can be speculated that better adhesion can be obtained here, if enamel is selectively etched before applying RelyX Unicem, as was proposed by DeMunck and others37 and Hikita and others.38 

Within the limitations of this study, it can be concluded that PCC fabricated from industrially sintered feldspathic ceramic should have at least a thickness of 1.5–2.0 mm in stress bearing areas to prevent crack formation and, in the long run, ceramic fractures. The cavity design had no significant influence on both crack formation and marginal integrity. The self-adhesive luting composite was as effective as a conventional multi-step preparation in preventing fractures.

The authors express their thanks and appreciation to Prof Dr Loys J Nunez, Memphis, TN, USA, for his constructive criticism and advice regarding the manuscript.