Abstract
CAD/CAM milling systems provide a rapid and individual method for the manufacturing of zirconia dental restorations. However, the disadvantages of these systems include limited accuracy, possible introduction of microscopic cracks, and a waste of material due to the principle of the ‘subtractive process’. The hypothesis of this study was that these issues can be overcome by a novel generative manufacturing technique, direct inkjet printing. A tailored zirconia-based ceramic suspension with 27 vol% solid content was synthesized. The suspension was printed on a conventional, but modified, drop-on-demand inkjet printer. A cleaning unit and a drying device allowed for the build-up of dense components of the size of a posterior crown. A characteristic strength of 763 MPa and a mean fracture toughness of 6.7 MPam0.5 were determined on 3D-printed and subsequently sintered specimens. The novel technique has great potential to produce, cost-efficiently, all-ceramic dental restorations at high accuracy and with a minimum of materials consumption.
INTRODUCTION
The introduction of CAD/CAM milling systems in the dental field enabled zirconia ceramics to be used as a standard material for dental prosthetic restorations (Luthardt et al., 1999; McLaren and Terry, 2002). In the meantime, more than 20 milling systems have been introduced into the market. Among CAD/CAM milling systems, two types can be differentiated. For ‘hard machining’, a restoration is milled out of a sintered monoblock, whereas a white monoblock is milled for ‘soft machining’, with subsequent sintering. The disadvantage of both systems is the considerable amount of waste of raw material, because the unused portions of the monoblocks must be discarded after milling, and recycling of the excess ceramic material is not feasible. Advantages of restorations produced by ‘hard machining’ are accurate shape and precise dimensions (Bindl and Mörmann, 2007). However, the tooling of sintered high-strength ceramics is costly and time-consuming. The tools are exposed to heavy abrasion and therefore withstand only short running cycles. Moreover, there is a considerable risk of microscopic cracks that can be introduced into the ceramic surface due to the tooling process of the brittle material (Wang et al., 2008). Surface damage does not occur during ‘soft machining’, because the shaping is performed prior to sintering. Furthermore, the milling of white monoblocks results in shorter machining times and longer service life cycles of the tools. However, the accuracy of contour and shape of ‘soft-machined’ restorations is more critical compared with that of the ‘hard-machined’ components because of the shrinkage during subsequent sintering, which must be considered and controlled. Additionally, quality assurance of the white monoblocks is difficult relative to storage and shipping, and with respect to the sintering process, which is performed not in an (controlled) industrial, but in a dental laboratory environment.
So-called generative manufacturing techniques exhibit the potential to overcome the described deficiencies. With these techniques, a three-dimensional component can be built up layer by layer. While the generative manufacturing of metallic- and polymer-based materials is state-of-the-art and commercially available, generative production with ceramic materials, worldwide, is still in development (Tay et al., 2003). For ceramic materials, 5 generative manufacturing techniques are of special interest: (i) stereo-lithography (Doreau et al., 2000); (ii) 3D-P, i.e., printing of a polymeric or inorganic binder into a ceramic powder bed (Uhland et al., 1999); (iii) selective laser sintering (Bourell et al., 1992); (iv) selective laser melting (Hollander et al., 2003); and (v) direct inkjet printing (Zhao et al., 2002). Only porous structures, however, can be created by the first 4 technologies (i–iv) mentioned above. In contrast, direct inkjet printing of a ceramic suspension provides the possibility of generating dense green bodies at a high resolution and complex shape (Ebert et al., 2008; Özkol et al., 2009). Besides, only thin walls of some 100 μm of thickness and a height of 1 mm at most, or small pillar-shaped arrays of less than 100 μm of diameter and a few 100 μm of height have been generated up to now (Zhao et al., 2002; Noguera et al., 2005; Lewis et al., 2006).
The objective of the present study was to develop a tailored direct inkjet printing process that can be used to build up dental prosthetic restorations made of high-strength zirconia ceramics. A tailored additive system was developed that allows for the printing of a suspension with a high solid content of zirconia powder, with the use of direct inkjet printing technology with conventional drop-on-demand inkjet print heads (Uibel et al., 2006). A well-balanced drying device was developed for the creation of three-dimensional structures of the size of dental prosthetic restorations at high accuracy with respect to its dimensions. Additionally, a tailored cleaning unit, based on a modified ultrasonic bath, was integrated into the printer, which allows for long printing periods without nozzle clogging. With our study, we tested the hypothesis that it is possible to overcome the major issues of CAD/CAM milling sytems—limited accuracy, possible introduction of microscopic cracks, and a waste of raw material—by the direct inkjet printing technique.
MATERIALS & METHODS
Synthesis of the Ceramic Suspension
The ceramic suspension consisted of approximately 27 vol% of zirconia powder, 55% distilled water, a boehmite sol, and dispersants (Uibel et al., 2006). The boehmite sol was used to prevent agglomeration of the ceramic particles and to increase the green body strength. We synthesized the boehmite sol by adjusting distilled water to a pH value of 2.0. Boehmite (Disperal P2, Sasol, Hamburg, Germany) was added at a temperature of 80°C. A 3 mol% quantity of yttria partially stabilized zirconia powder (TZ-3YS-E, Tosoh, Tokyo, Japan) was added to the sol. The mean ceramic particle size was 90 nm, and the specific surface area was 7 m2/g. The bulk density of the powder was 6.05 g/cm3. The pH value of the ceramic ink was recorded by potentiometry (In-Lab 417, Mettler Toledo, Gießen, Germany). The viscosity of the slurry was determined with the use of a rotational rheometer (Viscolab LC 10, Physica, Stuttgart, Germany).
Set-up of the Printing Station and Printing Procedure
The ceramic suspension was injected into an empty standard HP cartridge (HP 51645a, 42 mL, Hewlett Packard, Palo Alto, CA, USA) by means of a syringe. The printing device, based on a modified drop-on-demand deskjet printer (HP DeskJet 930c, Hewlett Packard), consisted of the following units. The original printing carriage held the cartridge, driven by the original servo- motor. The cleaning system was comprised of an ultrasonic bath (Carrera 2309, 50 W/50 Hz, Lutter & Partner, München, Germany) and stripping rollers. The printhead was automatically soaked and cleaned in the ultrasonic bath after each printing cycle, when the cartridge returned to the starting position in the x-direction. A mixture of water and ethanol was used as cleaning fluid. To prevent jams or error messages such as ‘printer out of paper’ after the substitution of paper feed by a z-axis, we developed a paper-simulating unit. As in commercial two-dimensional printers, printing occurred line by line in the x-direction, and the maximum line width (y-direction) was determined as the maximum width for the printed component. A z-drive (Servo motor, Isel Elektronik, Eichenzell, Germany) was implemented to allow for the printing of specific cross-sections, layer on layer, to build up the three-dimensional components. We used special software (CEC TestPoint, Capital Equipment Corp., Billerica, MA, USA) to control the vertical motion. A step size, i.e., a resolution in the z-direction of Δz = 5 μm, was achieved. The drying unit consisted of 3 components: 2 narrow-spot spotlights (1000 W, PAR 64 can, Showtec, Köln, Germany), magnifying glasses to focus the light, and a fan (Minebea Co. Ltd., NMB 2408NL-04W-B40, Ayutthaya, Thailand) to decrease the humidity in the printed area. The temperature in the printing zone was approximately 90°C. Graphite plates (Ringsdorff Werke, Bad Godesberg, Germany) of 4 mm thickness were used as substrates. The three-dimensional components were printed page-by-page from Microsoft Word files that contained the black-colored cross-sections. Each cross-section represented a slice with a thickness of 5 μm of the respective section in the z-direction (height) of the component.
Heat Treatment and Characterization of the Printed Components
The printed 3D components were first dried in a chamber dryer (T 5022, Heraeus Kulzer, Hanau, Germany) at 80°C for 12 hrs. The organic additives were then removed in a ceramic furnace at 550°C, and the parts were subsequently fired at 1450°C for 2.5 hrs. The density of the as-fired specimens was determined according to the principle of Archimedes. SEM micrographs (Leo 440i, Carl Zeiss, Jena, Germany) were taken from the cross- sections of cut specimens to analyze the microstructure. Printed and subsequently sintered specimens (1.5 x 3.0 x 30.0 mm3, n = 21) were ground on a precision surface grinding machine (PS R300, G&N, Erlangen, Germany) to determine the Weibull parameters, i.e., characteristic strength σ0 and Weibull modulus m (Munz and Fett, 1999). The grid size of the final diamond charged grinding wheel was 46 μm. Additionally, printed and sintered specimens (3.0 x 6.0 x 30.0 mm3, n = 4) were ground to determine the fracture toughness K Ic . These K Ic -specimens were notched by means of a diamond-charged cut-off wheel (thickness: approx. 200 μm). The notches (depth: 20% of specimen thickness) were sharpened by the razor-blade method (Kübler, 1997) (SEVNB, i.e., single-edge V-notched beam). Note that SEVNB specimens were used only for fracture toughness measurements, and specimens for strength measurements had inherent flaws without any pre-cracking. The specimens were mechanically tested in a universal testing machine (model 1186H0425, Instron, Darmstadt, Germany) in four-point bending mode. The inner and outer roller spans were 12 and 24 mm, respectively. The stressing rate was set at 100 MPa−1 to avoid subcritical crack growth during testing.
RESULTS
The pH value of the suspension was set at 8.5. The relative density of the fired specimens was 96.9%. An isotropic shrinkage of 20 vol% was determined. SEM micrographs of the printed and sintered specimens (cross-section) revealed a rather homogeneous microstructure, with some submicronsized pores. The characteristic strength of the ground bars was σ0 = 763 MPa, with a 90% confidence interval of [678;859]. The Weibull modulus was m = 3.5 [2.4;4.4] (Fig. 1). The fracture toughness of the SEVNB specimens was KIc = 6.7 ± 1.6 MPam0.5. The SEM analysis of the fracture surfaces of the flexural strength specimens revealed homogeneous cross-sections. Only single larger defects were detected on a few specimens (Fig. 2). These process-related defects were a result of single clogged nozzles that were either dried up or blocked by agglomerates during the printing process (Fig. 3). It was possible for crack-free components to be built up on the centimeter scale after optimization of the drying and cleaning process. Based on a CAD file, three-dimensional components of the size of a crown, with its characteristic occlusal surface topography, were built up (Fig. 4).
DISCUSSION
In contrast to results of studies published previously (Zhao et al., 2002; Noguera et al., 2005; Lewis et al., 2006), not only thin structures of some 100-μm thickness, but also components of high shape accuracy can be produced by direct inkjet printing. It was demonstrated that it is possible, by this technology, to build up dense three-dimensional components of the size and shape of a dental crown out of high-strength zirconia ceramics. Although the microstructures of the printed and fired samples were not completely free of process-related defects, the obtained density was at 96.9% of the theoretical density—high enough to provide mechanical properties (σ 0 = 763 MPa, KIc = 6.7 ± 1.6 MPam0.5) that can be compared with those of conventionally produced 3Y-TZP via cold isostatic pressing (NN, 2006).
The strong scattering of the strength (m = 3.5) is attributed to the clogging of single nozzles during printing, which was demonstrated on some single specimens and was responsible for the decreased strength of those samples. However, strengths of up to 1200 MPa were achieved on those specimens that contained no large defects due to clogged nozzles during printing. The strengths of those specimens without process-related large defects were determined by the inherent microscopic flaw distribution of the zirconia material itself.
It is remarkable that the developed ceramic suspension, with 27 vol% of solid content, was printable through original HP inkjet nozzles (diameter: approx. 28 μm), although the system had been developed originally for inks with a solid content of less than 5 vol%. The successful printing arose from the nano-scaled ceramic powder and the tailored additive system, which has been described in more detail elsewhere (Özkol et al., 2009). Moreover, the build-up of crack-free three-dimensional parts of the size presented was obtained due to the well-designed drying and cleaning system.
Regarding adjusted drying, pre-heating of the substrate ensured the avoidance of temperature gradients leading to internal stresses and bending of layers during drying. Concerning the cleaning system, an ethanol content of < 10 wt% in aqueous solution was determined as sufficient in terms of nozzle cleaning, without intense evaporation and the formation of bubbles on the printed surface during drying that would lead to later delamination.
In the next step of development, an advanced 3D printer will be used. The most important new feature will be the control of each single nozzle of the print heads by special software, whereby, in case of sudden nozzle-clogging during printing, the cartridge can be immediately moved to the cleaning unit, where the clogged nozzle will be re-opened, and the printing process can proceed. Moreover, this advanced printer will consist of more than one cartridge, to print support material in parallel. This will allow for the manufacture of not only a three-dimensional occlusal surface of a restoration, but also complete crowns and bridges with hollow spaces. It should be noted that shrinkage due to drying or sintering can be a critical issue in individually made dental ceramic prostheses. Isotropic shrinkage was determined on rectangular dense specimens. This may differ when cap structures with various wall thicknesses are to be printed. Therefore, the optimization of the drying process and a tailored multiple-stage sintering process, as well as an advanced design and scale of the three-dimensional data, are additional steps for further development.
Weibull plot with strength distribution of fired zirconia specimens (N = 21) that were built up by the direct inkjet printing technique. Fracture surface of a flexural strength test specimen. Single nozzle blocked by agglomerated particles. SEM micrograph showing the 3D-printed occlusal surface of a dental crown.
Footnotes
Notes
Acknowledgements
This work was supported by Heraeus Kulzer, Hanau, Germany.