2.1 Starting Materials

A photopolymerizable Si₃N₄ slurry suitable for DLP-based ceramic 3D printing was prepared and optimized in this study. The primary ceramic powders used in the slurry formulation included Si₃N₄ (SN-E10, UBE Corp., Japan), Al₂O₃ (AKP-53, Sumitomo Chemical, Japan), and Y₂O₃ (Yttrium oxide, Sigma-Aldrich, USA). The average particle sizes of these powders were confirmed to be 1.67, 0.192, and 1.099 μm for Si₃N₄, Al₂O₃, and Y₂O₃, respectively.

The ceramic composition was initially set to a weight ratio of Si₃N₄:Y₂O₃:Al₂O₃ = 90:5:5 based on previous studies and literature reports [¹⁰]. This composition was selected to facilitate the α-phase to β-phase transformation of Si₃N₄ and promote densification during sintering [¹¹]. Sintering additive ratios were slightly adjusted following mechanical property evaluations; however, the final optimized composition remained consistent at Si₃N₄ 90 wt%, Y₂O₃ 5 wt%, and Al₂O₃ 5 wt%.

A photocurable resin system was formulated using two commercially available acrylate-based monomers with distinct viscosities: monomer A, Isobornyl Methacrylate (240 cP), which is a monofunctional monomer primarily used for viscosity control and to achieve a more moderate curing speed, and monomer B, Trimethylolpropane Trimethacrylate (TMPTMA) (76 cP), a highly reactive trifunctional monomer chosen for its ability to form a high crosslinking density network, thus ensuring rapid curing and enhanced mechanical properties of the cured layers.. To enhance the interlayer adhesion and uniformity of the cured resin while precisely regulating the slurry viscosity, two types of urethane acrylate-based oligomers were blended into the formulation. High-viscosity oligomer A (4,560 cP) is a hexafunctional urethane acrylate oligomer, incorporated to achieve low shrinkage upon curing and a high crosslinking density, thereby contributing to improved durability and the dimensional precision of the final parts. However, its high functionality can lead to increased brittleness. Conversely, low-viscosity oligomer B (384 cP) is a difunctional urethane acrylate oligomer, primarily added to enhance interlayer adhesion and mitigate the brittleness associated with oligomer A, while also assisting in viscosity control. A photoinitiator (Irgacure 819, BASF, Germany) was used, as well as Dispersant (BYK-110, BYK, Germany) to ensure the stable dispersion of the ceramic powders within the slurry.

2.2. Slurry Formulation and Preparation

Nine photocurable resin compositions were designed and categorized into three groups to optimize slurry formulation: M-A, M-B, and M-AB. These formulations were prepared by varying the photoinitiator content (2, 3, and 4 wt%) and weight ratio of monomers A and B. The curing characteristics and layer-by-layer printing stability were evaluated for each formulation. A constant 50 wt% loading of Si₃N₄ powder was used in each composition to measure the curing depth, and the most suitable monomer composition for layer-by-layer DLP processing was selected based on these results.

The slurry was prepared using a two-step mixing process: The monomer–oligomer mixture, ceramic powder, and dispersant were premixed using a planetary mixer (Thinky Mixer ARE-310, Thinky Corp., Japan) at 500 rpm for 10 min. Subsequently, the resulting premix was transferred to an attrition mill (Netzsch, Germany), where it was further dispersed at 500 rpm for 30 min using zirconia beads as the milling medium. This dispersion step was intended to eliminate agglomerates of the ceramic particles and ensure a homogeneous slurry. Following mixing, the slurry was subjected to vacuum degassing for 10 min to remove entrapped air bubbles.

The rheological properties of the slurry were measured at 25 °C using a rotational viscometer (Brookfield DV2-Pro) immediately after preparation and again after 24 h of ambient storage. The measurements were conducted at sufficiently low shear rates considering the inherently high viscosity of the slurries. The variation in viscosity over time was used to assess the stability of each formulation. Further, the effects of oligomer content on slurry viscosity and cured body (green body) strength were systematically investigated. In this series of experiments, monomer A was used exclusively (M-A series), and the content of high-viscosity oligomer A was increased incrementally from 0 to 10 wt%. The slurry viscosity, printability, and three-point bending strength of the cured bodies were measured for each composition.

These results confirmed that the addition of 8 wt% oligomer A produced the most favorable balance between mechanical strength and printability. Subsequently, partial replacement with the low-viscosity oligomer B was explored to mitigate the increase in viscosity associated with oligomer A. The total oligomer content was fixed at 8 wt%, and the ratio of oligomers A to B was varied from 8:0 to 3:5. The viscosity and mechanical strength of each formulation were compared, and the optimal oligomer combination was identified.

2.3. Printing and Evaluation Process

A custom-built DLP 3D printer was utilized to evaluate the curing behavior and layer-by-layer printing characteristics of the photocurable slurry. The system employed a 405 nm wavelength ultraviolet light emitting diode (UV-LED) projector as the light source, with an XY resolution of 40 × 40 μm and a build platform size of 50 × 50 mm. The printing process was conducted in a bottom-up configuration, and the layer thickness was set to 50 μm.

Prior to printing, the curing depth of each slurry composition was measured, and this confirmed controllable depths in the range of 25–60 μm. It was determined that an energy dose of ~1000 mJ/cm² was required per layer under the 50 μm condition. Accordingly, each layer was exposed to the same energy dose during printing. After each UV exposure, the build platform was raised by 50 μm, followed by the application of a new slurry layer and subsequent pattern exposure in a repeated cycle. The photocurable resin system undergoes a free radical chain-growth polymerization mechanism upon UV irradiation during the printing process. This fundamental chemical reaction proceeds in three main steps:

Initiation Step: A radical (R·) generated from the photoinitiator by UV exposure reacts with the double bond of an acrylate or methacrylate, initiating the polymerization.

Propagation Step: The newly formed radical continuously reacts with the double bonds of other monomers, leading to extension of the polymer chain.

Termination Step: The polymerization reaction concludes when two radicals combine, forming a crosslinked polymer network.

During the printing process, slurry spreadability, interlayer adhesion, and deformation caused by curing shrinkage were monitored in real time to evaluate the printing stability of each formulation. After printing was completed, the dimensional accuracy and morphological stability of the cured specimens were assessed by comparing the printed shapes against the CAD model, with particular attention to warping and cracking.

The printed green parts were subjected to a post-curing process at 50 °C for 1 h, followed by a multistep debinding and sintering treatment. The debinding profile was determined using thermogravimetric analysis (TGA), which identified the decomposition range of the organic components. Further, the temperature was increased from 25–400 °C at a ramp rate of 1 °C/h to ensure the complete removal of organics.

Sintering was performed in a nitrogen (N₂) atmosphere at temperatures ranging from 1,700–1,900 °C, with a holding time of 2 h. Atmospheric pressure sintering and gas pressure sintering (GPS) were applied for comparison. Although atmospheric sintering was conducted under ambient N2 pressure, the GPS was conducted under 1 MPa N2 at both 1,800 and 1,900 °C. The mechanical properties of the sintered bodies were evaluated using three-point bending strength, Vickers hardness, and fracture toughness. The flexural strength test was conducted at room temperature in accordance with KS L ISO 14704, and the Vickers hardness was determined from the five measurements of polished surfaces in accordance with KS L ISO 14705. Fracture toughness was estimated based on the fracture characteristics of the specimens.

3.1. Effect of Monomer Composition and Photoinitiator Content

In the DLP 3D printing process, the viscosity of the ceramic slurry is a key parameter that determines both stacking stability and the dimensional precision of printed structures. A slurry with an excessively low viscosity can result in sedimentation of the ceramic particles, which leads to layer-wise inhomogeneity and printing defects [¹²]. In contrast, an excessively high viscosity can reduce slurry flowability, which results in poor coating and process delays [¹³]. this study investigated, the effects of monomer composition and photoinitiator content on slurry viscosity and time-dependent stability.

The effect of photoinitiator content was evaluated using monomer-A-based formulations (M-A series, 60 wt% ceramic). Fig 1(a) shows that increasing the photoinitiator content from 2−4 wt% led to a gradual decrease in the initial slurry viscosity. This trend is attributed to the partial action of the photoinitiator as a solvent-like component, which reduces the overall viscosity of the system [¹⁴]. For example, the viscosity decreased from ~12,500 cP at 2 wt% to ~10,800 cP at 4 wt%. However, the rate of viscosity reduction diminished above 4 wt%, and the excessive addition of photoinitiator resulted in solid dilution and undesirable curing shrinkage. Therefore, the 3 wt% was selected as an optimal concentration to balance curing efficiency and viscosity control.

Further, the time-dependent stabilities of different monomer formulations were compared. For the M-A-3 and M-B-3 formulations that comprised monomers A and B, respectively, M-A-3 showed the lowest initial viscosity but exhibited more than a 20% viscosity increase after 24 h, which implies reduced temporal stability. In contrast, M-B-3 had a slightly higher initial viscosity; however, it maintained good stability with less than a 5% change over the same period. The mixed monomer formulation M-AB-3 (A:B = 1:1 by weight) exhibited an intermediate initial viscosity of ~11,400 cP and the highest temporal stability, with less than a 3% increase after 24 h, as shown in Fig 1(b). These results are consistent with previous studies, which report that monomer A tends to undergo phase separation or aggregation over time, thereby increasing viscosity while the highly reactive monomer B promotes early partial network formation, stabilizing the slurry [¹⁵]. The observed differences in viscosity and stability with the monomer formulations (M-A-3, M-B-3, M-AB-3) are not solely due to the inherent viscosity differences of monomer A (240 cP) and monomer B (76 cP), which contribute to a dilution effect. More critically, the distinct chemical functionalities of these monomers govern the photopolymerization kinetics and the resultant polymer network structure. Monomer A, being a monofunctional Isobornyl Methacrylate, contributes to viscosity control and a moderate curing speed, but its limited crosslinking capability results in a less rigid network, potentially leading to the observed higher viscosity increase over 24 h and post-printing warpage. In contrast, monomer B, Trimethylolpropane Trimethacrylate (TMPTMA), is a highly reactive trifunctional monomer that facilitates the rapid formation of a highly crosslinked polymeric network. This dense network contributes to faster curing and improved initial stability, but can induce higher polymerization shrinkage, causing delamination. Therefore, the optimal mixed monomer formulation (M-AB-3) leverages both the viscosity-reducing effect of monomer B and the balanced crosslinking capability offered by the combination of monofunctional monomer A and trifunctional monomer B. This synergistic blend effectively harmonizes curing shrinkage, viscosity, and printing quality, leading to superior dimensional stability and print reliability.

In addition, stacking experiments were conducted using a DLP 3D printer for all nine resin formulations listed in Table 1. For the M-A series composed solely of monomer A, no significant deformation occurred during the stacking process. However, after printing, warping was observed at the edges of specimens as depicted in Fig 2(a). For the M-B series (monomer B only), a high curing shrinkage stress led to interlayer delamination during the stacking process, also clearly visible in Fig 2(b). In contrast, the mixed monomer formulations (M-AB series) exhibited superior flowability, interlayer adhesion, and curing stability throughout the printing process, resulting in the well-defined structures shown in Fig 2(c), thereby enabling the formation of structures without notable geometric distortion or delamination. The M-AB-3 formulation achieved the highest dimensional stability after curing, improving it by ~15% and 10% points compared to that of M-A-3 and M-B-3, respectively [¹⁶]. Further, this can be attributed to the balanced shrinkage behavior achieved by the mixed monomer composition, which effectively harmonizes the curing shrinkage, viscosity, and printing quality.

Curing depth is also considered an important performance metric. Monomer A yielded shallow curing depths with a lower shrinkage stress, which resulted in reduced warping but insufficient layer adhesion. In contrast, monomer B cured more deeply and rapidly; however, it induced excessive shrinkage stress, thereby leading to delamination. The M-AB-3 formulation demonstrated an optimal balance between curing depth and shrinkage stress, which ensures high dimensional accuracy and stable layer-by-layer stacking behavior. Consequently, the M-AB-3 formulation, which is composed of a 1:1 weight ratio of monomers A and B with a 3 wt% photoinitiator, was selected as the reference resin formulation for this study. Subsequently, this formulation was used as the basis for the further investigation of oligomer addition and optimization of ceramic solid loading.

3.2. Effect of Oligomer Addition on Viscosity and Strength

Incorporating oligomers into photopolymerizable resin systems can enhance the mechanical strength of cured structures and mitigate polymerization shrinkage. However, an increase in oligomer content leads to an increase in slurry viscosity, which can adversely affect printability during the layer-by-layer deposition process. In this study, the effects of high-molecular-weight oligomer A were investigated using a series of formulations. A base slurry composed of a 60 wt% Si₃N₄ ceramic and monomer A was prepared, and oligomer A was added at concentrations ranging from 0−10 wt%. Each formulation was evaluated for printability using a DLP printer and the flexural strength of a cured green body.

The slurry without oligomer A (oli-A-0) exhibited a very low viscosity of ~5,420 cP, which resulted in the poor adhesion of the cured layer to the build platform and led to print failure because of outflow. Similarly, the 2 wt% formulation (oli-A-2) failed to print consistently, which was attributed to insufficient viscosity, preventing stable layer formation. In contrast, formulations with oligomer A concentrations of 4 wt% or higher were printable, and the flexural strength of cured specimens increased with an increasing oligomer content [¹⁷]. The flexural strength was measured to be 38, 96, and 122 MPa for oli-A-4, oli-A-6, and oli-A-8, respectively. However, a decline in strength was observed at 10 wt% (oli-A-10), dropping to 83 MPa. This reduction can be attributed to a decrease in the internal bonding strength within the cured structure caused by excessive oligomer content or reduced light penetration and non-uniform curing resulting from increased viscosity [¹⁷]. The viscosity showed a near-linear increase with oligomer content, which reached ~9,100 cP at 8 wt% oligomer A (Table 2).

Additional experiments were conducted by partially substituting the low-viscosity oligomer B to address the issue of increased viscosity associated with oligomer A. The total oligomer content was fixed at 8 wt%, and six formulations (oli-AB-0 to oli-AB-10) with varying A:B ratios were prepared. The viscosity and flexural strength of the cured structures were evaluated for each formulation. The results demonstrated a consistent decrease in viscosity with increasing B content, while the flexural strength increased to a maximum and then declined. The A:B = 4:4 formulation (oli-AB-8) exhibited the most favorable performance with a reduced viscosity of ~7,000 cP and a maximum flexural strength of 154 MPa [¹⁸]. This result indicates an optimal balance between polymer density and viscosity control, thereby suggesting that oligomer B not only improves the flow characteristics of the slurry but also contributes to the formation of a sufficiently crosslinked network. This optimized performance is a direct consequence of both the differential viscosities of the oligomers and the contributions of their distinct chemical functionalities to the polymer network. High-viscosity oligomer A (4,560 cP) is a hexafunctional urethane acrylate oligomer. Its incorporation significantly increases the crosslinking density of the cured body, leading to higher green body strength (e.g., 122 MPa for oli-A-8) and improved durability, albeit with a tendency towards increased brittleness due to its high functionality. Conversely, low-viscosity oligomer B (384 cP) is a difunctional urethane acrylate oligomer. It contributes to reducing the overall slurry viscosity for improved processability, but more importantly, it provides a more flexible chain structure within the crosslinked network. This flexibility effectively enhances interlayer adhesion and mitigates the brittleness associated with oligomer A's high crosslinking density, thereby balancing the overall mechanical performance with optimized flowability In contrast, the formulation with 5 wt% oligomer B (oli-AB-10) exhibited the lowest viscosity (6,480 cP) but reduced flexural strength (133 MPa) [¹⁹]. These findings suggest that an appropriate combination of high- and low-viscosity oligomers is crucial for achieving a balance between mechanical performance and printability in photocurable resin systems (Table 3).

Based on the optimized formulation (monomer A/B = 1:1, photoinitiator 3 wt%, oligomer A/B = 4 wt%/4 wt%, and dispersant 4 wt%), a systematic evaluation of the ceramic solid loading from 60-68 wt% was conducted to determine the printability threshold. Viscosity measurements showed that slurries containing up to 66 wt% ceramic maintained a moderate increase in viscosity, achieving a value of ~15,000 cP, which ensured stable layer deposition. However, at 68 wt%, the viscosity exceeded 20,000 cP, thereby resulting in difficulties in layer re-coating and material spreading [²⁰]. These results are consistent with previous reports indicating that slurry viscosities beyond 20,000 cP compromise layer uniformity in DLP processes. Accordingly, 66 wt.% was established as the practical upper limit for the ceramic solid content in the slurry.

The final slurry formulation with 66 wt% ceramic content exhibited excellent layer-wise buildability and mechanical strength, and allowed the successful fabrication of complexshaped test pieces and spinal cage prototypes. This demonstrates the validity of the proposed formulation strategy for high-solid-content Si₃N₄-based photopolymerizable slurries for DLP-based additive manufacturing.

3.3. Optimization of Ceramic Solid Content

As summarized in Table 4, five formulations (S_62–S_70) were prepared by fixing the optimized resin composition (monomer A/B = 1:1, photoinitiator 3 wt%, oligomer A/B = 4 wt%/4 wt%, and dispersant in the range of 3.0–3.8 wt%) while incrementally increasing the ceramic powder content from 62 to 70 wt% in 2 wt% intervals. The viscosity of each slurry was measured using a rotational viscometer. The target upper limit for the slurry viscosity was set to 20,000 cP for ensuring defect-free and dimensionally accurate printing in the DLP 3D printing process.

According to the viscosity measurement results shown in Fig 3, the viscosity increased moderately from ~11,800 to 15,000 cP with an increase in solid content from 62 to 66 wt%. However, a further increase from 66 to 68 wt% caused a sharp increase of more than 33%, which exceeded 20,000 cP. For the formulation with 70 wt% solid content (S_70), the viscosity surpassed 30,000 cP, and during the actual printing, frequent defects such as blade coating failure and incomplete layer filling were observed, thereby making it difficult to achieve stable layer-by-layer printing [²¹].

Ceramic solid loading affects photopolymerization depth in addition to viscosity. The curing depths of the three formulations(S_64, S_66, and S_68) were measured to evaluate curing behavior. The experiment was conducted by irradiating a fixed range of energies (400–1400 mJ/cm²) using a DLP processor. The results are presented in Fig 4. The curing depth increases linearly with an increase in exposure energy. However, for the high-solid-loading formulation S_68, the curing depth was significantly lower than that of S_64 and S_66 under the same exposure conditions [²²]. This behavior can be attributed to enhanced light scattering by the ceramic particles in the slurry, which reduces the effective UV penetration depth, thereby limiting the achievable curing thickness. The rate of increase in the curing depth sharply declinedwhen the exposure energy exceeded 50 mJ/cm², thereby approaching the saturation region. This finding indicates that maintaining the exposure energy below 50 mJ/cm² is more favorable to achieving stable curing. These results experimentally demonstrate that solid loading influences not only the rheological properties but also the photopolymerization mechanism, and it supports the conclusion that S_66 offers the most balanced performance in terms of printability and curing behavior during the subsequent layer-by-layer fabrication and sintering processes.

3.4. Thermal Stability Evaluation of the Slurry via TGA Analysis

Analyzing the thermal stability and decomposition behavior of the organic components within the slurry is necessary as the photopolymerizable ceramic slurry undergoes debinding and sintering after printing to fabricate the final structure. In this study, TGA was conducted to investigate the decomposition characteristics of individual components (monomers A, B, A, and B). The results are presented in Fig 5. Further, TGA was performed on the green body fabricated using the optimized slurry to evaluate the thermal decomposition behaviors of the binders. As discussed in the experimental procedure, the TGA curve showed a total weight loss of ~16 wt% up to 400 °C, after which little to no further change occurred. This implies that most of the organic components, including monomers, oligomers, and dispersants, were thermally decomposed below 400 °C [²³]. The initial steep weight loss near 120 °C is attributed to the evaporation of low-molecular-weight monomers, while the gradual loss between 150 and 400 °C corresponds to the thermal decomposition of oligomers.

Based on these findings, the debinding process was designed to heat the green body up to 400 °C. After thermal treatment at this temperature, the residual weight of the sample was ~83.5 wt%, which corresponds to the combined weight of the ceramic powders (66 wt%) and sintering additives (Y₂O₃ + Al₂O₃, 9.9 wt%), thereby confirming the complete removal of organics. The debound specimen retained its shape without any visible shrinkage or cracks, which suggests that the internal stress caused by porosity during decomposition was minimized [²⁴].

Beyond 400 °C, negligible weight changes were observed for all resin components, which indicate that the major organic species were completely decomposed at this temperature. Accordingly, the optimal temperature for the debinding process for removing organic constituents in the slurry was established to be ~400 °C. The formation of residual pores in the final sintered body was expected to be minimized by applying a debinding profile based on the temperatures that completely remove the monomers and oligomers.

3.5. Evaluation of the Mechanical Properties of the Sintered Bodies

X-ray diffraction (XRD) analysis of the sintered specimens confirmed the successful phase transformation from α-Si₃N₄ to the desired β-Si₃N₄ phase after sintering, as presented in Fig 10. This verifies the Si₃N₄ ceramic had the proper intended formation.

Under sintering conditions at temperatures above 1800 °C all of the compositions (S_64, S_66, and S_68) achieved a relative density exceeding 99% (Fig 6(a)). It was observed that with an increase in solids loading densification began at lower sintering temperatures. Furthermore, the S_64 composition, with its lower solid content and density, exhibited clustered pores and defects, as clearly seen in its microstructure (Fig 6(b)). In contrast, the S_66 composition (Fig 6(c)) showed a significantly denser microstructure with reduced porosity. The S_68 composition, having the highest solid loading, achieved the highest relative density, and its microstructure (Fig 6(d)) consistently showed minimal bulk porosity, indicative of highly effective densification. The overall trend of decreasing bulk porosity and increasing densification with higher solid content is clearly depicted across the microstructures presented in Fig 6(b-d). An increase in density with higher sintering temperatures is one of the commonly observed characteristics in ceramics [²⁵]. However, despite excellent bulk density, issues such as layer delamination were observed during printing, which subsequently affected its three-point flexural strength.

The mechanical strengths of the sintered bodies were evaluated using a three-point bending test on specimens with dimensions of 4 × 3 × 35 mm. The specimens were fabricated in both horizontal (laid on the X-Y plane) and vertical (standing on the X-Y plane) build orientations (Fig 7), and ten specimens were tested for each condition to determine the average and standard deviation. The results confirmed that the flexural strength increased with higher sintering temperatures across all compositions, with S_66 exhibiting the highest strength at 1800 °C (Fig 8) [²⁶]. However, the high solid loading composition S_68 showed a significant reduction in strength of ~30–40% in the vertically printed specimens, which was attributed to interlayer bonding defects and layer delamination, which were directly observed during the printing process and post-printing inspection. Representative SEM images (Fig 9) clearly illustrate these interlayer bonding defects and delamination phenomena in the vertically printed S_68 specimens, particularly at high solid loadings where slurry flowability and uniform layer spreading are hindered, resulting in poor print quality.

Vickers hardness measurements were conducted under a load of 4.9 N (500 gf) using a Vickers hardness tester. All of the compositions exhibited hardness values exceeding 1500 HV when sintered above 1750 °C, with S_66 demonstrating consistent hardness stability across the entire temperature range. The fracture toughness was calculated based on crack lengths emanating from Vickers indentations using the indentation fracture (IF) method. The results confirmed a gradual increase in fracture toughness with sintering temperature, followed by a sudden decrease at 1850 °C. This decrease was attributed to excessive grain growth during oversintering, which increased the brittleness of the ceramic body (Fig 9). The comprehensive optimization process, from initial resin formulation to ceramic solid loading, and the resultant key performance indicators of the developed Si₃N₄ slurry are summarized in Table 5.