Manufacturing and Characterization of Dental Crowns Made of 5-mol% Yttria Stabilized Zirconia by Digital Light Processing

We herein report manufacturing of dental crowns made of 5-mol% yttria partially stabilized zirconia (5Y-PSZ) with desired mechanical properties, optical translucency and dimensional accuracy using digital light processing (DLP). To this end, all processing parameters were carefully controlled and optimized. First, 5Y-PSZ particles with a bimodal distribution were prepared via calcination of as-received granules and subsequent ball-milling and then used to formulate 5Y-PSZ suspensions with a high solid loading of 50 vol% required for high densification after sintering. Dispersant content was also optimized. To provide high dimensional accuracy, initial dimensions of dental crowns for 3D printing were precisely determined by considering increase and decrease in dimensions during photopolymerization and sintering, respectively. Photopolymerization time was also optimized for a given layer thickness of 50 μm to ensure good bonding between layers. A multi-step debinding schedule with a slow heating rate was employed to avoid formation of any defects. After sintering at 1500 °C for 2 h, 5Y-PSZ could be almost fully densified without noticeable defects within layers and at interfaces between layers. They had high relative densities (99.03 ± 0.39%) with a high cubic phase content (59.1%). These characteristics allowed for achievement of reasonably high mechanical properties (flexural strength = 625.4 ± 75.5 MPa and Weibull modulus = 7.9) and % transmittance (31.4 ± 0.7%). In addition, 5Y-PSZ dental crowns showed excellent dimensional accuracy (root mean square (RMS) for marginal discrepancy = 44.4 ± 10.8 μm and RMS for internal gap = 22.8 ± 1.6 μm) evaluated by the 3D scanning technique.


Introduction
Dental computer-aided design and computer-aided manufacturing (CAD/CAM) technology is still prevalent in the manufacturing of dental crowns made of yttria (Y 2 O 3 )-doped zirconia (ZrO 2 ) ceramics [1,2]. However, this technique has several intrinsic limitations, including high consumption of cutting tools and great loss of discarded material during the subtractive machining process. Thus, the use of 3D printing of ceramics, which can construct 3D structures via direct addition of materials in a layer-by-layer fashion, has recently received a great deal of attention in modern dentistry to substitute for traditional dental CAD/CAM technology [3,4].
From the viewpoint of resolution and accuracy of 3D printing techniques, digital light processing (DLP) and stereolithography (SLA) have been most extensively investigated for the manufacturing of zirconia dental crowns [3][4][5][6][7][8][9][10][11][12][13][14][15]. These techniques can selectively photocure thin layers of zirconia suspensions, where zirconia particles are uniformly dispersed in photocurable monomers, using high-resolution UV engines according to pre-designed 2D images layer-by-layer [3,4]. 3D-printed zirconia crowns can then be heat-treated for debinding to remove organic phases and for sintering to densify zirconia particles. This approach can precisely replicate intricate 3D geometries of dental crowns designed for Calcined granules were then vigorously crushed into smaller particles by ball-milling at a rotation speed of 90 rpm for 24 h using zirconia balls as media. Morphologies and size distributions of as-received granules, calcined granules and crushed particles were characterized using a field emission scanning electron microscope (FE-SEM; JSM-6701F, JEOL Techniques, Tokyo, Japan) and a laser particle size analyzer (CILAS 1090; Orleans, France), respectively. Calcined granules were then vigorously crushed into smaller particles by ball-milling at a rotation speed of 90 rpm for 24 h using zirconia balls as media. Morphologies and size distributions of as-received granules, calcined granules and crushed particles were characterized using a field emission scanning electron microscope (FE-SEM; JSM-6701F, JEOL Techniques, Tokyo, Japan) and a laser particle size analyzer (CILAS 1090; Orleans, France), respectively. Table 1 summarizes the constituents of photocurable 5Y-PSZ suspensions with a solid loading of 50 vol% used for manufacturing dental crowns by our DLP process. Unless otherwise notified, all reagents were used as purchased. To obtain reasonably low viscosities at high solid loadings, 1,6-hexanediol diacrylate (HDDA; Sartomer, Exton, PA, USA) and decalin (Sigma Aldrich, St. Louis, MO, USA) were employed as the photopolymerizable monomer and diluent, respectively [26]. An alkylol-ammonium salt of a copolymer with acidic groups (DISPERBYK-180; BYK-Chemie GmbH, Wesel, Germany) was used as a dispersant. Diphenyl(2,4,6-trimethylbenzoyl) phosphine oxide (TPO; Sigma Aldrich, St. Louis, MO, USA) was used as a photo-initiator.

Preparation of 5Y-PSZ Suspensions
For preparing 5Y-PSZ suspensions, a solution of 65 wt% HDDA-35 wt% decalin was used as a photocurable liquid medium. Predetermined amounts of HDDA, decalin, dispersant (2 wt% with respect to the 5Y-PSZ particle content) and zirconia balls as a media for vigorous mixing were added into plastic bottles. The mixtures were then blended by a planetary centrifugal mixer (Hantech Co. Ltd., Ansan-si, Gyeonggi-do, Republic of Korea) for 10 min at a rotation speed of 1000 rpm. Subsequently, submicron-sized 5Y-PSZ particles were added into liquid medium and then vigorously mixed by the planetary centrifugal mixer for 30 min. This mixing process was repeated three times using predetermined amounts (1/2, 1/4 and 1/4 of the total weight in sequence) of particles, in order to achieve uniform dispersion of particles in liquid mediums. Three different types of 5Y-PSZ suspensions with a range of solid loadings (44 vol%, 48 vol% and 50 vol%) were prepared to examine their rheological behaviors. For photocuring by DLP, 2 wt% of photo-initiator with respect to the HDDA content was mixed with 5Y-PSZ suspensions for 10 min.

Custom-Built DLP Process
5Y-PSZ specimens and crowns were manufactured using our custom-built DLP machine employing a specially designed feeding and recoating systems (Veltz3D, Incheon, Republic of Korea), as shown in Figure 2A, whose basic operation principle was proposed in our previous report [6]. To offer high accuracy for dental crown applications, a digital micromirror device (DMD) light engine with 1920 × 1080 pixels was employed, which could illuminate a UV power of~14.95 mW/cm 2 at a peak wavelength of~405 nm. Schematics showing the DLP process used for manufacturing 5Y-PSZ dental crowns are illustrated in Figure 2B. Our DLP machine could extrude highly concentrated 5Y-PSZ suspensions in a controlled manner using a stepper motor and then spread them uniformly using a recoater. In addition, after selective photopolymerization of the first layer, the second fresh layer could be directly deposited onto the first layer. This approach could eliminate a potential risk of interfacial delamination between layers that can occur during detachment of photopolymerized layers from a vat in the conventional DLP process. To ensure complete photocuring of 50-micron-thick layers of 5Y-PSZ suspensions and good bonding between layers, fresh layers were photocured for various time periods (1 s-10 s) and thicknesses of photocured layers were measured by a micrometer. In addition, to obtain high dimensional accuracy for dental crown applications, increases in dimensions via UV scattering by ceramic particles during the photocuring process were evaluated quantitatively [38,39,40]. To this end, disk-shaped regions with different diameters (100 μm-1000 μm) in 50-μm-thick-layers of 5Y-PSZ suspensions were photopolymerized for 3 sec, after which photocured disks were washed with ethanol and then their diameters were calculated by FE-SEM. The degree of discrepancy between designed and printed diameters was computed, representing broadening.

Manufacturing of 5Y-PSZ Specimens and Dental Crowns
5Y-PSZ specimens and dental crowns were manufactured by our DLP process. To this end, 50-micron-thick layers of 5Y-PSZ suspensions with a solid loading of 50 vol% were photocured for 3 s. Two types of specimens (rectangular bars and disks) were manufactured to estimate mechanical properties and optical translucency of 5Y-PSZ dental crowns. To achieve high accuracy, initial dimensions of dental crowns (c.f., Figure  2B) were precisely determined by considering opposing changes in dimensions during the entire process-increase due to UV scattering by 5Y-PSZ particles during photopolymerization process and decrease due to densification of 5Y-PSZ particles via sintering at high temperatures. Based on our preliminary studies, we employed an enlargement of ~25.2% and 24.5% in the x-y and z-directions, respectively, in order to design an initial CAD model for the DLP process.
As-printed 5Y-PSZ specimens and dental crowns were removed from build platforms and then rinsed with ethanol several times to remove uncured suspensions, after which green specimens and dental crowns underwent heat-treatment for debinding and sintering. To design a schedule for debinding, the thermal decomposition behavior of a green specimen was characterized by thermogravimetric analysis (TGA; STA 449 F3 Jupiter; Netzsch GmbH, Selb, Germany). For this evaluation, a specimen was heated up to 600 °C at a heating rate of 10 °C /min in air and its weight loss was recorded as a function of temperature. A representative TGA curve is shown in Figure 3. Thermal decomposition To ensure complete photocuring of 50-micron-thick layers of 5Y-PSZ suspensions and good bonding between layers, fresh layers were photocured for various time periods (1 s-10 s) and thicknesses of photocured layers were measured by a micrometer. In addition, to obtain high dimensional accuracy for dental crown applications, increases in dimensions via UV scattering by ceramic particles during the photocuring process were evaluated quantitatively [38][39][40]. To this end, disk-shaped regions with different diameters (100 µm-1000 µm) in 50-µm-thick-layers of 5Y-PSZ suspensions were photopolymerized for 3 sec, after which photocured disks were washed with ethanol and then their diameters were calculated by FE-SEM. The degree of discrepancy between designed and printed diameters was computed, representing broadening.

Manufacturing of 5Y-PSZ Specimens and Dental Crowns
5Y-PSZ specimens and dental crowns were manufactured by our DLP process. To this end, 50-micron-thick layers of 5Y-PSZ suspensions with a solid loading of 50 vol% were photocured for 3 s. Two types of specimens (rectangular bars and disks) were manufactured to estimate mechanical properties and optical translucency of 5Y-PSZ dental crowns. To achieve high accuracy, initial dimensions of dental crowns (c.f., Figure 2B) were precisely determined by considering opposing changes in dimensions during the entire process-increase due to UV scattering by 5Y-PSZ particles during photopolymerization process and decrease due to densification of 5Y-PSZ particles via sintering at high temperatures. Based on our preliminary studies, we employed an enlargement of~25.2% and 24.5% in the x-y and z-directions, respectively, in order to design an initial CAD model for the DLP process.
As-printed 5Y-PSZ specimens and dental crowns were removed from build platforms and then rinsed with ethanol several times to remove uncured suspensions, after which green specimens and dental crowns underwent heat-treatment for debinding and sintering. To design a schedule for debinding, the thermal decomposition behavior of a green specimen was characterized by thermogravimetric analysis (TGA; STA 449 F3 Jupiter; Netzsch GmbH, Selb, Germany). For this evaluation, a specimen was heated up to 600 • C at a heating rate of 10 • C/min in air and its weight loss was recorded as a function of temperature. A representative TGA curve is shown in Figure 3. Thermal decomposition started at a temperature of 150 • C due to removal of diluent, dispersant, and photo-initiator. Above 300 • C, remarkable weight losses were observed due to intensive removal of photopolymerized HDDA. On the basis of the TGA analysis, a multi-step debinding schedule was carefully designed, where slow heating rates below 470 °C were employed to remove photopolymerized HDDA and organic phases without formation of defects such as cracks within layers and at interfaces between layers ( Table 2). After debinding, 5Y-PSZ specimens and dental crowns were densified via sintering at 1500 °C for 2 h.

Characterization of as-Printed and Sintered 5Y-PSZ Specimens
3D geometries of as-printed and sintered 5Y-PSZ specimens and dental crowns were examined by optical microcopy. Microstructures of as-printed and sintered 5Y-PSZ specimens and dental crowns were characterized by FE-SEM. External and internal structures of dental crowns were examined by micro-computed tomography (μ-CT; SkyScan1273, Bruker Corp., Billerica, MA, USA). A voxel size of 40 μm was used for image analyses.
Shrinkages after sintering at 1500 °C for 2 h were computed by comparing dimensions of sintered specimens to those of as-manufactured specimens. Densities of sintered specimens were measured using Archimedes' principle. Their relative densities were computed based on measured densities and theoretical density of 5Y-PSZ (6.034 g/cm 3 ) [41]. However, it should be noted that theoretical density of 5Y-PSZ should be affected by the contents of tetragonal and cubic phases. Grain size was measured by the linear intercept method. In addition, an image analysis program (MIPAR, MIPAR Image Analysis, Columbus, OH, USA) was used to roughly examine distribution of grain sizes.
Crystalline phases of sintered 5Y-PSZ specimens were characterized using an X-ray diffractometer (DMAX-2500, Rigaku Corporation, Tokyo, Japan) using Cu Kα radiation. On the basis of the TGA analysis, a multi-step debinding schedule was carefully designed, where slow heating rates below 470 • C were employed to remove photopolymerized HDDA and organic phases without formation of defects such as cracks within layers and at interfaces between layers ( Table 2). After debinding, 5Y-PSZ specimens and dental crowns were densified via sintering at 1500 • C for 2 h.

Characterization of as-Printed and Sintered 5Y-PSZ Specimens
3D geometries of as-printed and sintered 5Y-PSZ specimens and dental crowns were examined by optical microcopy. Microstructures of as-printed and sintered 5Y-PSZ specimens and dental crowns were characterized by FE-SEM. External and internal structures of dental crowns were examined by micro-computed tomography (µ-CT; SkyScan1273, Bruker Corp., Billerica, MA, USA). A voxel size of 40 µm was used for image analyses.
Shrinkages after sintering at 1500 • C for 2 h were computed by comparing dimensions of sintered specimens to those of as-manufactured specimens. Densities of sintered specimens were measured using Archimedes' principle. Their relative densities were computed based on measured densities and theoretical density of 5Y-PSZ (6.034 g/cm 3 ) [41]. However, it should be noted that theoretical density of 5Y-PSZ should be affected by the contents of tetragonal and cubic phases. Grain size was measured by the linear intercept method. In addition, an image analysis program (MIPAR, MIPAR Image Analysis, Columbus, OH, USA) was used to roughly examine distribution of grain sizes.
In addition, an energy dispersive X-ray spectroscope (EDS; Horiba, Ltd., Kyoto, Japan) attached to a FE-SEM (SU-70, Hitachi Ltd., Hitachi, Japan) was used to examine chemical compositions of grains with tetragonal and cubic phases. To this end, EDS spot analyses with a spot size of 0.4 µm were conducted in two different types of grains (large and small grains) and 14 points were tested for each grain.

Mechanical Properties Evaluation
To estimate mechanical functions of 5Y-PSZ dental crowns, fracture strength and Weibull modulus of sintered specimens were measured using three-point flexural strength tests according to ISO 6872 (Dentistry-Ceramic materials) [42][43][44]. Rectangular specimens were grounded to obtain dimensions of 3 mm × 4 mm × 25 mm and then polished to ensure smooth surfaces. A constant crosshead speed of 1 mm/min was applied to the center of each specimen using a screw-driven load frame (Instron 4467, Instron, MA, USA). Load versus displacement curves were recorded until specimens were fractured. Flexural strengths were then computed by considering loads at fracture and dimensions (thicknesses and widths) of specimens. Twelve specimens were tested to obtain mean and standard deviation. In addition, a Weibull plot of failure probability (P f ) versus measured flexural strengths (σ) was constructed to determine Weibull modulus (m) representing mechanical reliability of 5Y-PSZ specimens and characteristic strength (σ o ) [45].

Optical Translucency Evaluation
To estimate optical translucency of 5Y-PSZ dental crowns, a computer-controlled spectrophotometer (CM-3600A, Konica Minolta, Inc., Osaka, Japan) was used to directly measure optical transmittances of sintered disks. For these evaluations, the sintered disks were slightly ground to obtain a thickness of 1 mm and then polished to ensure flat, smooth surfaces. Spectral transmittance data over the spectral range from 360 nm to 740 nm at 10 nm intervals were collected. % transmittance was computed according to the standard (CIE standard illuminant D65) [46]. Three samples were tested to obtain mean and standard deviation.

Dimensional Accuracy Evaluation
To examine potentials of 3D-printed 5Y-PSZ dental crowns for clinical use, their dimensional accuracies in terms of marginal and internal fits were closely examined by a 3D scanning technique [13,47,48]. For these evaluations, 5Y-PSZ dental crowns were scanned by a dental scanner (Medit T500, Medit Corp., Seoul, Republic of Korea) having a precision of 7 µm. Scanned data were recorded using a scan software (colLab 2017 v2.0.0.4, Medit Corp., Seoul, Republic of Korea). Color maps were then generated by superimposing scanned data of 5Y-PSZ dental crowns over original CAD data and then analyzed using a 3D analysis software program (Geomagic ControlX 2020.1 ver.; 3D Systems, Inc., Rock Hill, SC, USA). Blue and red colors indicate negative and positive deviations compared to the reference (original CAD data), respectively, while green color indicate deviations within ±50 µm. Internal discrepancies between original CAD design and manufactured 5Y-PSZ dental crowns were quantified in terms of root mean square (RMS) obtained from the entire inner area of dental crowns [49,50]. Five specimens were tested to obtain mean and standard deviation.

Statistical Analysis
Statistical analysis was carried out using one-way analysis of variance (ANOVA) with MATLAB (The MathWorks, Inc., Natick, MA, USA) and Tukey-Kramer Post Hoc Test (OrginLab Corp., Northampton, MA, USA). p values less than 0.05 were considered statistically significant.

Morphologies and Size Distributions of 5Y-PSZ Granules and Particles
To manufacture dental crowns, commercially available 5Y-PSZ granule, which has been widely used to produce zirconia blanks for dental CAD/CAM, was employed owing to its high translucency [36]. However, as-received granules showed very large sizes with several tens of microns ( Figure 4A). In addition, they could not be crushed into smaller particles because of a small amount (~4 wt%) of polymeric binder. Thus, asreceived granules were heat-treated to 1000 • C for 1 h to completely remove polymeric binder, while their spherical morphologies and sizes changed negligibly ( Figure 4B). After ball-milling of calcined granules, fine particles were obtained ( Figure 4C). In addition, 5Y-PSZ nanoparticles were partially bonded together due to calcination at a relatively high temperature of 1000 • C ( Figure 4D). On the other hand, when granules calcined at a lower temperature of 500 • C were ball-milled, they were crushed into very fine particles ( Figure 4E) since nanoparticles were weakly bonded together. This finding suggests that a high calcination temperature of 1000 • C could effectively reduce the specific surface area of particles, thus being beneficial to achieve high solid loadings with reasonably low viscosities [26]. However, it should be noted that particle size distribution can also strongly affect viscosities of 5Y-PSZ suspensions [25,[51][52][53].  Size distributions of as-received 5Y-PSZ granules and crushed particles obtained via ball-milling of granules calcined at 1000 °C were characterized by a laser particle size analyzer. Results are shown in Figure 5A. As-received granules showed relatively large sizes ranging from 5 μm to 100 μm. The mean size was as large as ~50 μm. On the other hand, crushed particles obtained after calcination at 1000 °C and subsequent ball-milling showed much smaller sizes. In addition, they showed a bimodal distribution with two peaks at ~0.3 μm and ~15 μm. A cumulative particle size distribution of crushed particles is displayed in Figure 5B. The fractions of particles with size ranges of 0.1-1 μm, 1 μm-3 μm, 3 μm-10 μm and 10 μm-30 μm were 34.04%, 19.08%, 16.71%, and 29.42%, respectively. This finding suggests that calcined granules could be efficiently crushed into smaller particles by ball-milling. In addition, a wide particle size distribution could be beneficial for preparing 5Y-PSZ suspensions with reasonable viscosities at high solid Size distributions of as-received 5Y-PSZ granules and crushed particles obtained via ball-milling of granules calcined at 1000 • C were characterized by a laser particle size analyzer. Results are shown in Figure 5A. As-received granules showed relatively large sizes ranging from 5 µm to 100 µm. The mean size was as large as~50 µm. On the other hand, crushed particles obtained after calcination at 1000 • C and subsequent ball-milling showed much smaller sizes. In addition, they showed a bimodal distribution with two peaks at~0.3 µm and~15 µm. A cumulative particle size distribution of crushed particles is displayed in Figure 5B. The fractions of particles with size ranges of 0.1-1 µm, 1 µm-3 µm, 3 µm-10 µm and 10 µm-30 µm were 34.04%, 19.08%, 16.71%, and 29.42%, respectively. This finding suggests that calcined granules could be efficiently crushed into smaller particles by ball-milling. In addition, a wide particle size distribution could be beneficial for preparing 5Y-PSZ suspensions with reasonable viscosities at high solid loadings since voids formed between large particles could be filled by smaller particles [25,52]. On the other hand, particles obtained via ball-milling of granules calcined at a relatively low temperature of 500 • C showed much smaller sizes (<1 µm) with a unimodal distribution ( Figure 5C).
ball-milling of granules calcined at 1000 °C were characterized by a laser particle size analyzer. Results are shown in Figure 5A. As-received granules showed relatively large sizes ranging from 5 μm to 100 μm. The mean size was as large as ~50 μm. On the other hand, crushed particles obtained after calcination at 1000 °C and subsequent ball-milling showed much smaller sizes. In addition, they showed a bimodal distribution with two peaks at ~0.3 μm and ~15 μm. A cumulative particle size distribution of crushed particles is displayed in Figure 5B. The fractions of particles with size ranges of 0.1-1 μm, 1 μm-3 μm, 3 μm-10 μm and 10 μm-30 μm were 34.04%, 19.08%, 16.71%, and 29.42%, respectively. This finding suggests that calcined granules could be efficiently crushed into smaller particles by ball-milling. In addition, a wide particle size distribution could be beneficial for preparing 5Y-PSZ suspensions with reasonable viscosities at high solid loadings since voids formed between large particles could be filled by smaller particles [25,52]. On the other hand, particles obtained via ball-milling of granules calcined at a relatively low temperature of 500 °C showed much smaller sizes (<1 μm) with a unimodal distribution ( Figure 5C).

Effect of Particle Sizes on Rheological Behaviors of 5Y-PSZ Suspensions
To examine the effect of particle size distribution on rheological behaviors of 5Y-PSZ suspensions, four different types of suspensions were prepared using as-received granules, granules calcined at 1000 • C and crushed particles obtained via ball-milling of granules calcined at 1000 • C and 500 • C, and 2 wt% of dispersant was employed for uniform dispersion of particles in photocurable vehicles. Apparent viscosities of 5Y-PSZ suspensions were measured as a function of shear rate using a cone/plate viscometer. Results are displayed in Figure 6. All suspensions showed shear thinning behaviors, where viscosity decreased with an increase in shear rate. However, when as-received granules were used, suspensions showed very high viscosities due to large size of granules. On the other hand, granules calcined at 1000 • C could be partially crushed during vigorous mixing for preparation of suspensions, thus resulting in lower viscosities than as-revived granules. In addition, when crushed particles obtained via ball-milling of granules calcined at 1000 • C were used, significantly lower viscosities were observed. This reduction in viscosity was mainly attributed to a broad particle size distribution (c.f., Figure 5B), and thus small particles could fill voids formed between large particles [26,52]. However, when a low calcination temperature of 500 • C was used, nanoparticles within granules were weakly bonded together, and thus very fine particles could be obtained by ball-milling (c.f., Figure 5C). Thus, 5Y-PSZ suspensions prepared using these fine particles showed higher viscosities than those prepared using particles obtained via ball-milling of granules calcined at a higher temperature of 1000 • C. This finding suggests that it is crucial to prepare 5Y-PSZ particles with proper size ranges by optimizing the calcination temperature (c.f., 1000 • C in this study) and ball-milling process, which allows for achievement of high solid loadings with reasonably low viscosities.
ball-milling (c.f., Figure 5C). Thus, 5Y-PSZ suspensions prepared using these fine part showed higher viscosities than those prepared using particles obtained via ball-millin granules calcined at a higher temperature of 1000 °C . This finding suggests that it is cr to prepare 5Y-PSZ particles with proper size ranges by optimizing the calcina temperature (c.f., 1000 °C in this study) and ball-milling process, which allow achievement of high solid loadings with reasonably low viscosities.  Figure 6. Apparent viscosities as a function of shear rate observed for 5Y-PSZ suspensions prepared using as-received, granule calcined at 1000 • C, and crushed particles obtained via ball-milling of granules calcined at 1000 • C and 500 • C (solid loading = 44 vol%).

Optimization of Dispersant Content
To obtain high solid loadings with low viscosities, a solution of low-viscosity HDDA as a photopolymerizable monomer and decalin as a diluent was employed as a liquid medium. In addition, to induce strong repulsion between 5Y-PSZ particles, a kind of an alkylol-ammonium salt of a copolymer with acidic groups was employed as a dispersant [6]. To optimize dispersant content in relation to particle content, 5Y-PSZ suspensions were prepared using a range of dispersant contents (1 wt%, 2 wt%, 3 wt% and 4 wt%). For these tests, a solid loading of 44 vol% was used. Figure 7A shows apparent viscosities as a function of shear rate observed for different dispersant contents. All 5Y-PSZ suspensions showed a decrease in viscosity with an increase in shear rate. Such shear thinning behaviors are highly beneficial for the DLP process [51]. When the dispersant content increased from 1 wt% to 2 wt%, remarkable reductions in viscosity were observed over all shear rates ( Figure 7B). However, with further increases in dispersant content (3 wt% and 4 wt%), viscosity increased. This finding suggests that 2 wt% of the dispersant used in this study could sufficiently adsorb on entire surfaces of 5Y-PSZ particles and induce sufficient repulsion between particles, thus offering significantly reduced viscosities. However, when excessive dispersant was used, viscosity increased due to an increase in thickness of the adsorbed layer [26].

Optimization of Maximum Solid Loading
For manufacturing zirconia dental crowns with desired mechanical properties and optical translucency, it is crucial to achieve high densification with negligible defects after sintering at high temperatures. To this end, 5Y-PSZ suspensions should have high solid loadings since highly concentrated ceramic particles can be densified more favorably, while reasonably low viscosities should be maintained to allow for the formation of thin, uniform layers by a recoater for sequential DLP processes. 5Y-PSZ suspensions were prepared using a range of solid loading (44 vol%, 48 vol% and 50 vol%) and their rheological behaviors were characterized to determine the maximum solid loading obtainable using our compositions (c.f., Table 1). The same dispersant content of 2 wt% was used for all suspensions. Apparent viscosities as a function of shear rate observed for 5Y-PSZ suspensions prepared using a range of solid loadings are shown in Figure 8A. Shear thinning behaviors with a decrease in viscosity with an increase in shear rate were observed for all 5Y-PSZ suspensions. On the other hand, higher solid loading resulted in higher viscosities over all shear rates ( Figure 8B), which is a common phenomenon of ceramic suspensions for colloidal processes [26,27,53]. However, 5Y-PSZ suspensions with the highest solid loading of 50 vol% showed reasonably low viscosities (1.37 ± 0.19 Pa·s) at a high shear rate of 200 s −1 , suggesting that they could be spread uniformly by a recoater for sequential DLP processes. In addition, reasonably high viscosities (4.66 ± 0.73 Pa·s) were observed at a low shear rate of 10 s −1 and thus fresh layers of 5Y-PSZ suspensions deposited on previously photopolymerized layers could be maintained fairly during photopolymerization. Effect of dispersant content on apparent viscosities of 5Y-PSZ suspensions with a solid loading of 50 vol% was also examined, as shown in Figure 8C. At a low shear rate of 10 s −1 , 3 wt% showed lower viscosities than 2 wt%. However, at high shear rates (100 s −1 and 200 s −1 ), 2 wt% and 3 wt% showed no significant differences in viscosity. It should be noted that HDDA content decreases with an increase in dispersant content, thus resulting in lower green strength after photopolymerization. Thus, it is reasonable to suppose that 2 wt% of dispersant used in this study is sufficient to uniformly disperse 5Y-PSZ particles and offer reasonably high green strengths. However, when a higher solid loading of 52 vol% was used, 5Y-PSZ suspensions showed very high viscosities (data now shown here). In addition, air was often trapped within layers during DLP. Thus, a solid loading of 50 vol% was employed for manufacturing 5Y-PSZ specimens and dental crowns, in order to achieve almost full densification without imperfections after sintering at 1500 • C for 2 h.

EER REVIEW 12 of 23
To obtain high solid loadings with low viscosities, a solution of low-viscosity HDDA as a photopolymerizable monomer and decalin as a diluent was employed as a liquid medium. In addition, to induce strong repulsion between 5Y-PSZ particles, a kind of an alkylol-ammonium salt of a copolymer with acidic groups was employed as a dispersant [6]. To optimize dispersant content in relation to particle content, 5Y-PSZ suspensions were prepared using a range of dispersant contents (1 wt%, 2 wt%, 3 wt% and 4 wt%). For these tests, a solid loading of 44 vol% was used. Figure 7A shows apparent viscosities as a function of shear rate observed for different dispersant contents. All 5Y-PSZ suspensions showed a decrease in viscosity with an increase in shear rate. Such shear thinning behaviors are highly beneficial for the DLP process [51]. When the dispersant content increased from 1 wt% to 2 wt%, remarkable reductions in viscosity were observed over all shear rates ( Figure 7B). However, with further increases in dispersant content (3 wt% and 4 wt%), viscosity increased. This finding suggests that 2 wt% of the dispersant used in this study could sufficiently adsorb on entire surfaces of 5Y-PSZ particles and induce sufficient repulsion between particles, thus offering significantly reduced viscosities. However, when excessive dispersant was used, viscosity increased due to an increase in thickness of the adsorbed layer [26].

Optimization of Maximum Solid Loading
For manufacturing zirconia dental crowns with desired mechanical properties and optical translucency, it is crucial to achieve high densification with negligible defects after sintering at high temperatures. To this end, 5Y-PSZ suspensions should have high solid loadings since highly concentrated ceramic particles can be densified more favorably, while reasonably low viscosities should be maintained to allow for the formation of thin, uniform layers by a recoater for sequential DLP processes. 5Y-PSZ suspensions were prepared using a range of solid loading (44 vol%, 48 vol% and 50 vol%) and their However, when a higher solid loading of 52 vol% was used, 5Y-PSZ suspensions showed very high viscosities (data now shown here). In addition, air was often trapped within layers during DLP. Thus, a solid loading of 50 vol% was employed for manufacturing 5Y-PSZ specimens and dental crowns, in order to achieve almost full densification without imperfections after sintering at 1500 °C for 2 h.

Optimization of Photocuring Process for DLP
To completely consolidate layers of 5Y-PSZ suspensions with high dimensional accuracy, UV illumination time for photocuring should be carefully determined. Thus, layers of a 5Y-PSZ suspension with a solid loading of 50 vol% were photocured for a range

Optimization of Photocuring Process for DLP
To completely consolidate layers of 5Y-PSZ suspensions with high dimensional accuracy, UV illumination time for photocuring should be carefully determined. Thus, layers of a 5Y-PSZ suspension with a solid loading of 50 vol% were photocured for a range of illumination time periods (1 s to 10 s). Photocured layers were then washed with ethanol and their thicknesses were measured using a micrometer. Figure 9A shows measured thicknesses of photocured layers as a function of UV illumination time. The photocured thickness increased almost linearly from 37.8 ± 1.2 µm to 93.2 ± 1.2 µm with an increase in UV illumination time from 1 s to 4 s. A UV illumination time of 2 s showed photocured thickness (56.7 ± 4.7 µm) slightly thicker than layer thickness (50 µm) of 5Y-PSZ suspensions, which would have a risk of imperfections between photocured layers particularly for manufacturing of complicated structures (e.g., dental crowns). Thus, to ensure strong bonding between photocured layers, a UV illumination time of 3 sec, which could result in a photocured thickness of 81.4 ± 1.0 µm, was employed for the DLP process. designed diameters were smaller than 400 μm, relatively large discrepancies were observed presumably due to resolution of a DLP engine. However, percent discrepancy decreased considerably with an increase in diameter and then changed slightly. Thus, based on these measurements, initial dimensions of dental crowns could be determined reasonably.  Figure 9(B) represent photocured diameters and discrepancies, respectively.

Microstructures of as-Printed 5Y-PSZ
Microstructures of as-printed 5Y-PSZ specimens were examined by FE-SEM, as shown in Figure 10A-C. Owing to the use of 50-μm-thick layers of a 5Y-PSZ suspension with UV illumination time of 3 s, specimens could be manufactured without notable defects ( Figure 10A). In addition, photocured layers could be strongly bonded together For manufacturing 5Y-PSZ dental crowns using the DLP technique, special care should be paid to determine initial dimensions of dental crowns since there should be enlargement and shrinkage during photopolymerization and sintering, respectively. More specifically, when predetermined regions of thin layers of 5Y-PSZ suspensions are exposed to UV illumination, their vicinities can also be photopolymerized due to UV scattering by ceramic particles, resulting in notable increase in dimensions. This broadening effect can be greatly influenced by many factors, including refractive index of ceramic particles and solid loading [38,39]. Thus, the degree of discrepancy between designed and photocured dimensions should be carefully characterized and controlled for individual ceramic suspensions to achieve high accuracies for dental crown applications. To this end, disk-shaped regions with different diameters (100-1000 µm) were photocured for 3 sec to completely photopolymerize 50-µm-thick layers of 5Y-PSZ suspensions. Figure 9B shows measured diameters of disks in accordance with their initial values and percent (%) discrepancies between measured and initial values. When designed diameters were smaller than 400 µm, relatively large discrepancies were observed presumably due to resolution of a DLP engine. However, percent discrepancy decreased considerably with an increase in diameter and then changed slightly. Thus, based on these measurements, initial dimensions of dental crowns could be determined reasonably.

Microstructures of as-Printed 5Y-PSZ
Microstructures of as-printed 5Y-PSZ specimens were examined by FE-SEM, as shown in Figure 10A-C. Owing to the use of 50-µm-thick layers of a 5Y-PSZ suspension with UV illumination time of 3 s, specimens could be manufactured without notable defects ( Figure 10A). In addition, photocured layers could be strongly bonded together without interfacial delamination ( Figure 10B). The fractured surface showed that 5Y-PSZ particles could be distributed in photocured HDDA without notable aggregates or voids ( Figure 10C). without interfacial delamination ( Figure 10B). The fractured surface showed that 5Y-PSZ particles could be distributed in photocured HDDA without notable aggregates or voids ( Figure 10C).

Shrinkages and Relative Density of Sintered 5Y-PSZ
Following a tightly controlled debinding process with slow heating rates below 470 o C to remove photopolymerized HDDA and organic phases, 5Y-PSZ specimen and dental crowns were densified by sintering at 1500 °C for 2 h. Table 3 summarizes linear sintering shrinkages and relative densities of 5Y-PSZ disks. Linear shrinkages in diameter and thickness were 19.12 ± 0.18% and 19.65 ± 0.27%, respectively. It should be noted that these values should be carefully considered to design initial dimensions of dental crowns for the DLP process, which could offer high dimensional accuracy after sintering. Sintered 5Y-PSZ disks showed high densities (5.976 ± 0.023 g/cm 3 ) measured by Archimedes' principle, which corresponded to 99.03 ± 0.39% of theoretical density of 5Y-PSZ (6.034 g/cm 3 ) [41]. However, it should be noted that theoretical density of 5Y-PSZ should be affected not by the contents of tetragonal and cubic phases, but also Y2O3 contents in both phases.

Microstructures and Grain Sizes of Sintered 5Y-PSZ
Densification behavior and microstructures of sintered 5Y-PSZ specimens were characterized by FE-SEM, as shown in Figure 11A-C. Fractured surfaces of 5Y-PSZ specimens revealed high densification without notable defects, such as large voids and cracks ( Figure 11A). In addition, no interfaces between layers were visible, suggesting strong interfacial bonding without imperfections, which is one of the most striking advantages of our DLP process using a specially designed recoating system. Only a few residual pores were observed ( Figure 11B). To examine morphologies and of grains, the surfaces of specimens were polished and then heat-treated at 1350 °C for 20 min to reveal grain boundaries via thermal etching. Figure 11C show a representative FE-SEM image. Two different types of grains (large and small grains) were observed due to co-existence of tetragonal and cubic phases.

Shrinkages and Relative Density of Sintered 5Y-PSZ
Following a tightly controlled debinding process with slow heating rates below 470 • C to remove photopolymerized HDDA and organic phases, 5Y-PSZ specimen and dental crowns were densified by sintering at 1500 • C for 2 h. Table 3 summarizes linear sintering shrinkages and relative densities of 5Y-PSZ disks. Linear shrinkages in diameter and thickness were 19.12 ± 0.18% and 19.65 ± 0.27%, respectively. It should be noted that these values should be carefully considered to design initial dimensions of dental crowns for the DLP process, which could offer high dimensional accuracy after sintering. Sintered 5Y-PSZ disks showed high densities (5.976 ± 0.023 g/cm 3 ) measured by Archimedes' principle, which corresponded to 99.03 ± 0.39% of theoretical density of 5Y-PSZ (6.034 g/cm 3 ) [41]. However, it should be noted that theoretical density of 5Y-PSZ should be affected not by the contents of tetragonal and cubic phases, but also Y 2 O 3 contents in both phases.

Microstructures and Grain Sizes of Sintered 5Y-PSZ
Densification behavior and microstructures of sintered 5Y-PSZ specimens were characterized by FE-SEM, as shown in Figure 11A-C. Fractured surfaces of 5Y-PSZ specimens revealed high densification without notable defects, such as large voids and cracks ( Figure 11A). In addition, no interfaces between layers were visible, suggesting strong interfacial bonding without imperfections, which is one of the most striking advantages of our DLP process using a specially designed recoating system. Only a few residual pores were observed ( Figure 11B). To examine morphologies and of grains, the surfaces of specimens were polished and then heat-treated at 1350 • C for 20 min to reveal grain boundaries via thermal etching. Figure 11C show a representative FE-SEM image. Two different types of grains (large and small grains) were observed due to co-existence of tetragonal and cubic phases. Grain sizes were measured via the linear intercept method [54]. Values measure considering all grains were 1.45 ± 0.18 μm. To roughly examine a distribution of g sizes, an image analysis program was used. Figure 12A show a representative col Grain sizes were measured via the linear intercept method [54]. Values measured by considering all grains were 1.45 ± 0.18 µm. To roughly examine a distribution of grain sizes, an image analysis program was used. Figure 12A show a representative colored image reconstructed from a FE-SEM image, where different colors represent different sizes. Counts as a function of equivalent circle diameter-diameter of a circle closely matching the perimeter of a grain [55]-are shown in Figure 12B. Roughly calculated sizes of large (pink color) and small grains (green color) representing were 1.58 µm and 0.73 µm, respectively. It should be noted that large and small grains correspond to cubic and tetragonal phases, respectively. Grain sizes were measured via the linear intercept method [54]. Values measu considering all grains were 1.45 ± 0.18 μm. To roughly examine a distribution o sizes, an image analysis program was used. Figure 12A show a representative image reconstructed from a FE-SEM image, where different colors represent d sizes. Counts as a function of equivalent circle diameter-diameter of a circle matching the perimeter of a grain [55]-are shown in Figure 12B. Roughly calculate of large (pink color) and small grains (green color) representing were 1.58 μm a μm, respectively. It should be noted that large and small grains correspond to cu tetragonal phases, respectively. It should be noted that densification behavior and crystalline phases (e.g., tetr and cubic phases) of 5Y-PSZ are strongly affected by sintering conditions s temperature and dwelling time, thus resulting in different mechanical and properties. In addition, sintering conditions should be optimized individua different particles since they have different particle sizes and specific surface are example, it was reported that the highest hardness and fracture toughness were o after sintering at 1450 °C for 2 h [56]. Meanwhile, other reports demonstrated that a sintering temperature of 1550 °C was used to achieve high mechanical properties [ high optical translucency [58]. Thus, a further study is required to examine e sintering conditions (temperature and dwell time) on densification, grain grow crystalline phases of 5Y-PSZ manufactured using our DLP technique, in order to op mechanical properties and optical transmittance simultaneously for dental applications.

Crystalline Phases of Sintered 5Y-PSZ
Crystalline phases of sintered 5Y-PSZ specimens were characterized by X expected, two different types of peaks corresponding to tetragonal (JCPDS card 079-1767) and cubic (JCPDS card no. 01-081-1550) phases were observed, as sh It should be noted that densification behavior and crystalline phases (e.g., tetragonal and cubic phases) of 5Y-PSZ are strongly affected by sintering conditions such as temperature and dwelling time, thus resulting in different mechanical and optical properties.
In addition, sintering conditions should be optimized individually for different particles since they have different particle sizes and specific surface areas. For example, it was reported that the highest hardness and fracture toughness were obtained after sintering at 1450 • C for 2 h [56]. Meanwhile, other reports demonstrated that a higher sintering temperature of 1550 • C was used to achieve high mechanical properties [57] and high optical translucency [58]. Thus, a further study is required to examine effect of sintering conditions (temperature and dwell time) on densification, grain growth, and crystalline phases of 5Y-PSZ manufactured using our DLP technique, in order to optimize mechanical properties and optical transmittance simultaneously for dental crown applications.

Crystalline Phases of Sintered 5Y-PSZ
Crystalline phases of sintered 5Y-PSZ specimens were characterized by XRD. As expected, two different types of peaks corresponding to tetragonal (JCPDS card no. 01-079-1767) and cubic (JCPDS card no. 01-081-1550) phases were observed, as shown in Figure 13. Contents of tetragonal and cubic phases computed by Rietveld refinement using X'Pert Highscore Plus software were~40.9% and~59.1%, respectively. Measured values were comparable to those of commercial 5Y-PSZ [37]. This finding suggests that 5Y-PSZ dental crowns manufactured using our DLP process could offer reasonable translucency when they are highly densified without notable defects. It should be noted that cubic phase content can be further controlled by adjusting sintering temperature if necessary [43,59].
values were comparable to those of commercial 5Y-PSZ [37]. This finding 5Y-PSZ dental crowns manufactured using our DLP process could offe translucency when they are highly densified without notable defects. It sho that cubic phase content can be further controlled by adjusting sintering te necessary [43,59].

Chemical Compositions of Tetragonal and Cubic Grains
Chemical compositions of two different types of grains with tetragon phases were characterized by EDS spot analyses. Table 4 summarizes the elements (Zr, Y, and O) taken of large and small grains. Large grains showed contents (7.0 ± 0.3%) than small grains (3.8 ± 0.3%). This finding suggests th small grains have cubic and tetragonal phases, respectively.

Mechanical Properties of Sintered 5Y-PSZ
For clinical applications, 5Y-PSZ dental crowns manufactured using our should have reasonable flexural strengths to withstand applied loads functions [40]. Thus, flexural strengths of 5Y-PSZ specimens were measur point flexural strength tests. Twelve specimens were tested to calculate standard deviation. All specimens showed reasonably high flexural strength MPa), as shown in Figure 14A. These values were comparable to those ob dental CAD/CAM [37,59]. In addition, all specimens showed flexural stre than 500 MPa. This finding suggests that our DLP technique using highly 5Y-PSZ suspensions with a solid loading of 50 vol% can manufacture not on prostheses (dental crowns) but also three-unit prostheses (dental bridges) [4 Figure 13. Representative XRD pattern of a 5Y-PSZ specimen sintered at 1500 • C for 2 h. "t" and "c" indicate tetragonal and cubic phases, respectively.

Chemical Compositions of Tetragonal and Cubic Grains
Chemical compositions of two different types of grains with tetragonal and cubic phases were characterized by EDS spot analyses. Table 4 summarizes the contents of elements (Zr, Y, and O) taken of large and small grains. Large grains showed higher Y 2 O 3 contents (7.0 ± 0.3%) than small grains (3.8 ± 0.3%). This finding suggests that large and small grains have cubic and tetragonal phases, respectively.

Mechanical Properties of Sintered 5Y-PSZ
For clinical applications, 5Y-PSZ dental crowns manufactured using our DLP process should have reasonable flexural strengths to withstand applied loads during their functions [40]. Thus, flexural strengths of 5Y-PSZ specimens were measured by three-point flexural strength tests. Twelve specimens were tested to calculate mean with standard deviation. All specimens showed reasonably high flexural strengths (625.4 ± 75.5 MPa), as shown in Figure 14A. These values were comparable to those obtained using dental CAD/CAM [37,59]. In addition, all specimens showed flexural strengths higher than 500 MPa. This finding suggests that our DLP technique using highly concentrated 5Y-PSZ suspensions with a solid loading of 50 vol% can manufacture not only single-unit prostheses (dental crowns) but also three-unit prostheses (dental bridges) [42]. computed with the following equation When a bending stresses of 300 MPa was applied, Pf value was as low as 0.0019. finding suggests that single crowns made of 5Y-PSZ using our DLP techniques would break during their functions under normal conditions.

Optical Translucency of 5Y-PSZ Disks
For clinical uses, dental crowns should have reasonable optical transluc [35,59,60,61]. Thus, light transmittance capabilities of 5Y-PSZ specimens manufact using our DLP technique was roughly evaluated. For these evaluations, 5Y-PSZ disks a thickness of 1 mm were placed onto papers with colored marks, as shown in Figure  The 5Y-PSZ disk clearly revealed colored marks, suggesting its reasonably high transmittance capability attributed to high densification with negligible defects Optical transmittances of 5Y-PSZ disks were quantified by spectrophotometry. Result displayed in Figure 15B. Three disks were tested to obtain mean and standard devia Spectral transmittance data were collected at 10 nm intervals from 360 nm to 740 Percentage transmittance values computed to according to the standard (CIE stan illuminant D65) [46] were 31.4 ± 0.7%, which would be proper for some clinical However, these values were lower than those obtained using dental CAD/CAM du lower densification. Thus, enhancing densification behavior of zirconia, for exampl increasing solid loading in zirconia suspensions and/or using new sintering proc (e.g., microwave sintering [62] and two-step sintering [63]), still remains as one of the challenging tasks in DLP of zirconia for dental applications [6]. Mechanical reliabilities of 5Y-PSZ specimens were evaluated by computing Weibull modulus (m) and characteristic strength (σ o ) from a Weibull plot [45]. Figure 14B shows a Weibull plot of measured flexural strengths. m is a measure of the distribution of flexural strengths. Thus, for dental crown applications, 5Y-PSZ should have reasonably high m to provide high mechanical reliability with small deviations of flexural strengths and longterm mechanical function without failure. Computed m value of 5Y-PSZ was 7.9, which was comparable to those obtained by SLA process [22] and dental CAD/CAM [37]. On the other hand, σ o can be defined as a level of strength at which the P f becomes 63.2% and represents a statistical indicator of overall strength of the material. Thus, higher σ 0 values are required for dental crown applications. Computed σ o of 5Y-PSZ was 661.8 MPa, which was comparable to those obtained by dental CAD/CAM [37]. From these m and σ o values, failure probability (P f ) of 5Y-PSZ dental prostheses under flexural stress (σ) could be computed with the following equation When a bending stresses of 300 MPa was applied, P f value was as low as 0.0019. This finding suggests that single crowns made of 5Y-PSZ using our DLP techniques would not break during their functions under normal conditions.

Optical Translucency of 5Y-PSZ Disks
For clinical uses, dental crowns should have reasonable optical translucency [35,[59][60][61]. Thus, light transmittance capabilities of 5Y-PSZ specimens manufactured using our DLP technique was roughly evaluated. For these evaluations, 5Y-PSZ disks with a thickness of 1 mm were placed onto papers with colored marks, as shown in Figure 15A. The 5Y-PSZ disk clearly revealed colored marks, suggesting its reasonably high light transmittance capability attributed to high densification with negligible defects [6]. Optical transmittances of 5Y-PSZ disks were quantified by spectrophotometry. Results are displayed in Figure 15B. Three disks were tested to obtain mean and standard deviation. Spectral transmittance data were collected at 10 nm intervals from 360 nm to 740 nm. Percentage transmittance values computed to according to the standard (CIE standard illuminant D65) [46] were 31.4 ± 0.7%, which would be proper for some clinical uses. However, these values were lower than those obtained using dental CAD/CAM due to lower densification. Thus, enhancing densification behavior of zirconia, for example, by increasing solid loading in zirconia suspensions and/or using new sintering processes (e.g., microwave sintering [62] and two-step sintering [63]), still remains as one of the most challenging tasks in DLP of zirconia for dental applications [6].

Dimensional Accuracy of 5Y-PSZ Dental Crowns
To demonstrate utility of our DLP technique for dental applications, single crowns were manufactured using 5Y-PSZ suspensions. Their dimensional accuracies were then evaluated in terms of marginal and internal discrepancies [13,47]. A CAD model of a single crown was employed as a proof of concept (c.f., Figure 2B). To achieve high dimensional accuracy, initial dimensions of dental crowns were carefully determined by considering both enlargement by photopolymerization and shrinkage by sintering. In other words, initial dimensions were oversized than target dental crowns by a factor of ~25.2% and ~24.5% in the x-y and z directions, respectively. Figure 16A shows representative OM images of as-printed and sintered 5Y-PSZ dental crowns. Remarkable decrease in dimensions was observed due to high densification of 5Y-PSZ particles during sintering at 1500 °C for 2 h. However, μ-CT analyses revealed that no noticeable defects, such as voids, cracks or severe distortion, were observed for dental crowns even with sophisticated geometries ( Figure 16B). In addition, dental crowns were translucent under light illumination ( Figure 16C). It is reasonable to suppose that these dental crowns can have optical transmittances close to those of 5Y-PSZ disks (31.4 ± 0.7%), suggesting their potential for clinical uses. For offering long-term clinical success, dental crowns should have high dimensional accuracy (e.g., marginal gap ≤ 120 μm and internal gap ≤ 300 μm) [64]. Thus, 3D scanning technique was employed to perform quantitative analyses with high reliability and reproducibility [48,50]. For these evaluations, color maps were generated by

Dimensional Accuracy of 5Y-PSZ Dental Crowns
To demonstrate utility of our DLP technique for dental applications, single crowns were manufactured using 5Y-PSZ suspensions. Their dimensional accuracies were then evaluated in terms of marginal and internal discrepancies [13,47]. A CAD model of a single crown was employed as a proof of concept (c.f., Figure 2B). To achieve high dimensional accuracy, initial dimensions of dental crowns were carefully determined by considering both enlargement by photopolymerization and shrinkage by sintering. In other words, initial dimensions were oversized than target dental crowns by a factor of~25.2% and~24.5% in the x-y and z directions, respectively. Figure 16A shows representative OM images of as-printed and sintered 5Y-PSZ dental crowns. Remarkable decrease in dimensions was observed due to high densification of 5Y-PSZ particles during sintering at 1500 • C for 2 h. However, µ-CT analyses revealed that no noticeable defects, such as voids, cracks or severe distortion, were observed for dental crowns even with sophisticated geometries ( Figure 16B). In addition, dental crowns were translucent under light illumination ( Figure 16C). It is reasonable to suppose that these dental crowns can have optical transmittances close to those of 5Y-PSZ disks (31.4 ± 0.7%), suggesting their potential for clinical uses.

Dimensional Accuracy of 5Y-PSZ Dental Crowns
To demonstrate utility of our DLP technique for dental applications, single crowns were manufactured using 5Y-PSZ suspensions. Their dimensional accuracies were then evaluated in terms of marginal and internal discrepancies [13,47]. A CAD model of a single crown was employed as a proof of concept (c.f., Figure 2B). To achieve high dimensional accuracy, initial dimensions of dental crowns were carefully determined by considering both enlargement by photopolymerization and shrinkage by sintering. In other words, initial dimensions were oversized than target dental crowns by a factor of ~25.2% and ~24.5% in the x-y and z directions, respectively. Figure 16A shows representative OM images of as-printed and sintered 5Y-PSZ dental crowns. Remarkable decrease in dimensions was observed due to high densification of 5Y-PSZ particles during sintering at 1500 °C for 2 h. However, μ-CT analyses revealed that no noticeable defects, such as voids, cracks or severe distortion, were observed for dental crowns even with sophisticated geometries ( Figure 16B). In addition, dental crowns were translucent under light illumination ( Figure 16C). It is reasonable to suppose that these dental crowns can have optical transmittances close to those of 5Y-PSZ disks (31.4 ± 0.7%), suggesting their potential for clinical uses. For offering long-term clinical success, dental crowns should have high dimensional accuracy (e.g., marginal gap ≤ 120 μm and internal gap ≤ 300 μm) [64]. Thus, 3D scanning technique was employed to perform quantitative analyses with high reliability and reproducibility [48,50]. For these evaluations, color maps were generated by For offering long-term clinical success, dental crowns should have high dimensional accuracy (e.g., marginal gap ≤ 120 µm and internal gap ≤ 300 µm) [64]. Thus, 3D scanning technique was employed to perform quantitative analyses with high reliability and reproducibility [48,50]. For these evaluations, color maps were generated by superimpos-ing scanned data of 5Y-PSZ dental crowns over an original CAD data for a target crown. Representative color maps are shown in Figure 17A. Most regions of internal surfaces showed green color, indicating good internal fit with a small discrepancy within ±50 µm. On the other hand, orange and blue colors were also observed at the marginal position. Mean RMS values for the internal gap and marginal discrepancy were 22.8 ± 1.6 µm and 44.4 ± 10.8 µm, respectively ( Figure 17B). These values are comparable to those obtained for zirconia dental crowns manufactured using DLP [11,12]. These findings suggest that our DLP technique can offer reasonably high dimensional accuracy with assistance of tightly controlled processing parameters, including initial dimensions of dental crowns, sequential photopolymerization process, and heat-treatment (debinding and sintering). superimposing scanned data of 5Y-PSZ dental crowns over an original CAD data for a target crown. Representative color maps are shown in Figure 17A. Most regions of internal surfaces showed green color, indicating good internal fit with a small discrepancy within ±50 μm. On the other hand, orange and blue colors were also observed at the marginal position. Mean RMS values for the internal gap and marginal discrepancy were 22.8 ± 1.6 μm and 44.4 ± 10.8 μm, respectively ( Figure 17B). These values are comparable to those obtained for zirconia dental crowns manufactured using DLP [11,12]. These findings suggest that our DLP technique can offer reasonably high dimensional accuracy with assistance of tightly controlled processing parameters, including initial dimensions of dental crowns, sequential photopolymerization process, and heat-treatment (debinding and sintering).

Conclusions
5Y-PSZ dental crowns with desired mechanical properties, optical translucency and dimensional accuracy for clinical uses could be manufactured by carefully controlling processing parameters for our DLP process. The use of a high solid loading of 50 vol% in 5Y-PSZ suspensions allowed sintered 5Y-PSZ to have high relative densities (99.03 ± 0.39%), thus offering high flexural strength (625.4 ± 75.5 MPa) and % transmittance (31.4 ± 0.7). In addition, high dimensional accuracy (RMS for marginal discrepancy = 44.4 ± 10.8 μm and RMS for internal gap = 22.8 ± 1.6 μm) was achieved by carefully designing initial dimensions of dental crowns and photocuring time for 3D printing.

Conclusions
5Y-PSZ dental crowns with desired mechanical properties, optical translucency and dimensional accuracy for clinical uses could be manufactured by carefully controlling processing parameters for our DLP process. The use of a high solid loading of 50 vol% in 5Y-PSZ suspensions allowed sintered 5Y-PSZ to have high relative densities (99.03 ± 0.39%), thus offering high flexural strength (625.4 ± 75.5 MPa) and % transmittance (31.4 ± 0.7). In addition, high dimensional accuracy (RMS for marginal discrepancy = 44.4 ± 10.8 µm and RMS for internal gap = 22.8 ± 1.6 µm) was achieved by carefully designing initial dimensions of dental crowns and photocuring time for 3D printing.