Digital light processing 3D printing of surface-oxidized Si3N4 coated by silane coupling agent

ABSTRACT Due to high light absorption and high refractive index of silicon nitride (Si3N4) ceramic, it is difficult to prepare Si3N4 slurry with favorable curing ability, high solid loading and low viscosity at the same time, and thus the high-quality Si3N4 parts are hard to fabricate via Digital Light Processing (DLP). In this paper, the oxidation process was used to enhance the curing behavior of Si3N4 slurry, and then silane coupling agent (KH560) was used to improve the rheological properties of the oxidized Si3N4 slurry. The effect of Si3N4 slurry with oxidation and modification on its rheological behavior, light absorption, curing ability and stability have been systematically investigated. A Si3N4 slurry with abundant photocuring ability, high solid loading, low viscosity and reliable stability was fabricated by the oxidized (1 h) and KH560 (1 wt.%) modified Si3N4 powders, and the dense Si3N4 ceramic parts were produced by this slurry via DLP 3D printing. Subsequently, the influence of oxidation and modification on microstructure, mechanical properties and thermal conductivity have been investigated. Finally, the microstructure and performances of Si3N4 ceramic were compared between different forming processes (including: DLP and cool isostatic pressing).


Introduction
Silicon nitride (Si 3 N 4 ) ceramic, one of the best comprehensive performance structural ceramics, have been widely used in aerospace, biomedical field, mechanical engineering and chemical metallurgy industries, due to its excellent hardness, high bending strength, good chemical stability, superior hightemperature resistance and biocompatibility [1][2][3][4]. Traditional manufacturing methods to obtain customer-designed and complex-shaped Si 3 N 4 parts always rely on slip casting and dry-pressing molding. With the increase of application fields of Si 3 N 4 parts, more and more complex-shaped products with high accuracy are needed, such as artificial joint, high-temperature heat exchanger, nozzle and so on [5][6][7][8][9]. Because of the high hardness and brittleness of Si 3 N 4 ceramic, the conventional fabrication process is unable to produce these sophisticated Si 3 N 4 parts with high precision. In summary, it is necessary to explore new technologies for fabricating complex-shaped ceramic products.
Over the last decade, additive manufacturing (AM) technology has attracted great attention and interest from researchers and the industry due to its characteristics of near-net shaping and dieless forming. Hence, AM technology has been expected as an ideal method to fabricate parts with sophisticated shapes [10][11][12][13][14]. With the rapid development of AM technology, more and more methods have sprung up, such as Selective Laser Sintering (SLS), Selective Laser Melting (SLM), Three-dimensional Printing (3DP), Direct Ink Writing (DIW), Digital Light Processing (DLP) and so on [15][16][17]. SLM and SLS are always used to produce metal. Owing to the high melting temperature of ceramic and the inherent high thermal gradient and residual thermal stress of the laser-based AM technologies, SLM and SLS are unsuitable for ceramic [16,18]. 3DP and DIW are the most widely applicable AM technologies, and it can be used in all material systems, including metal, ceramic and polymer. Nevertheless, the main disadvantage of 3DP and DIW is low accuracy, which limits their application prospect [19]. DLP, based on photopolymerization effect uses a light device to project a 2D pattern into the slurry pool, and then cures the photosensitive resin within a few seconds to accomplish layer-by-layer "printing" and accumulation. Subsequently, the green body via debinding and sintering process to get the finished product with greatly high printing resolution and surface finish quality. At present, DLP has been widely applied in the field of ceramic due to its high precision and surface smoothness [20][21][22][23][24].
Generally, the printing slurry contains ceramic particle, photosensitive resin and photoinitiator. During printing, ultraviolet light initiates photoinitiator to form free radical, and then the free radical induces photosensitive resin to activate photopolymerization. On the basis of the curing principle of DLP, a whitecolored or light-colored ceramic with inborn low light absorption and low refractive index always exhibits good photocuring ability [25]. Consequently, various white-colored or light-colored ceramic parts have been successfully prepared by DLP, such as ZrO 2 [20,26], Al 2 O 3 [26], AlN [27], ZTA [28], etc. On the contrary, high light absorption and high refractive index are the innate characteristics of the graycolored Si 3 N 4 ceramic. As a result, it is difficult to fabricate Si 3 N 4 parts by DLP, as well as has been reported rarely. In our previous research, we found that forming a SiO 2 layer on the surface of Si 3 N 4 ceramic via oxidation process can improve their photocuring ability by reducing the light absorption and refractive index of Si 3 N 4 powder [29]. From this, Zou et al. [11] have successfully prepared complicated Si 3 N 4 parts by this oxidation approach. Except for sufficient photocuring ability, the ceramic slurry for obtaining high-quality Si 3 N 4 parts also requires high solid loading and low viscosity [30]. The high solid loading of slurry could provide enough strength for the green body and alleviate the shrink during debinding and sintering process, which is benefit to high-quality product without defects. On the other hand, the suitable viscosity enables the slurry to form a uniform and flat coating for printing, which can promote the homogeneous of Si 3 N 4 ceramics. Nevertheless, there is a trade-off between the high solid loading and the low viscosity. Some approaches have been used to balance this issue. Liu et al. [31][32][33] found that the modified Si 3 N 4 powder with KH560 can reduce the viscosity and promote stability of slurry because of the epoxy group of KH560 forms an ether covalent bond with the hydroxyl group of photosensitive resin. However, the modification by KH560 cannot enhance the photocuring behavior of the Si 3 N 4 slurry. A desired slurry should simultaneously have abundant photocuring ability, high solid loading and low viscosity. Unfortunately, no one studied the intrinsic relationship of these problems at the same time.
In this paper, the oxidation process was used to enhance the curing behavior of Si 3 N 4 slurry, and then silane coupling agent (KH560) was used to improve the rheological properties of the oxidized Si 3 N 4 slurry. The effect of Si 3 N 4 slurry with oxidation and modification on its rheological behavior, light absorption, curing ability and stability have been systematically investigated. Furthermore, the influence of oxidation and modification on microstructure, mechanical properties and thermal conductivity of Si 3 N 4 ceramic have been investigated. At last, the effect of the different forming processes (including: DLP and cool isostatic pressing (CIP)) on microstructure and performance of Si 3 N 4 ceramic were compared.

Pretreatment of Si 3 N 4 powders
The oxidation process is as follows. α-Si 3 N 4 was placed in a high purity alumina crucible and then oxidized at 1200°C for 0.5, 1 or 2 h in the air. Then, in order to reduce particle agglomeration, the oxidized α-Si 3 N 4 was subjected to ball milling in ethanol suspension with Si 3 N 4 ball media for 1 h using a planetary ball mill (QM-QX4, Nanjing NanDa Instrument Plant, China). The rotation speed and ball-to-powder weight ratio were 200 rpm and 5:1.
The surface modification process is shown below. The Si 3 N 4 powders (pristine or oxidized) were mixed with KH560 (1, 2, 3 or 4 wt.% of ceramic powder) in ethanol suspension, and then proceeded ball milling for 6 h at 200 rpm. At last, mixed powders were made by ball mixing Si 3 N 4 doped with sintering additives (5 wt.% Y 2 O 3 and 3 wt.% MgO) for 2 h. The rotation speed was 200 rpm, and the weight ratio of ball to powders was 5:1.

Digital light processing (DLP) and post-processing
The DLP ceramic 3D printer used in this study is provided by Shenzhen RFHLASER CO., LTD. A LED light source with 405 nm wavelength was equipped in this printer (as shown in Figure 1.). The X/Y-plane resolution and X/Y-plane dimensions of the printer are 50 μm pixels and 76 × 50 mm 2 , respectively. Si 3 N 4 slurries were printed layer-to-layer with a thickness of 15 μm. The exposure intensity and exposure time were adjusted during the fabrication process. The printed green parts were first conducted by a twostep debinding process to completely remove the cured resin polymer, including a vacuum debinding and air debinding. Based on our previous research [20,26], the two-step debinding process can effectively avoid cracks due to the low decomposition rate of organics. The heating curve for debinding is shown in Figure 2(a). The green body was then sintered by pressureless sintering under a flowing nitrogen atmosphere at 1825°C for 4 h with a heating rate of 5°C/min (the heating curve as shown in Figure 2(b)). In order to compare the mechanical properties and thermal conductivity of the DLP samples with the CIP samples. The control samples were prepared by pressing in stainless steel die and then CIP at 200 MPa. The heating curve of pressureless sintering of the control samples is shown in Figure 2 (b) as well.

Materials characterization
Particle size distribution of Si 3 N 4 powder was performed by laser diffraction particle size analyzer (Mastersizer 2000, Malvern, England). The absorbance was measured using a UV-vis spectrophotometer (UVvis, UV-3600 Plus, Shimadzu, Japan). The rheology of the Si 3 N 4 suspension was tested by a rheometer (MCR301, Physica, Austria). The microstructure of the samples was characterized by a scanning electron  microscope (SEM, SU8220, Hitachi, Japan) on the polished and plasma-etched surfaces. The samples were plasma-etched with CF 4 gas for 100 s in plasma etching apparatus (L-451D-L, Anelva reactive ion etching system). The average grain size and average aspect ratio of samples were evaluated by measuring the diameter of at least 800 grains using the SEM images. Nano measure software package was used to measure and record diameter values. The morphology of Si 3 N 4 powder was characterized by a transmission electron microscope (TEM, JEM-2100 HR, JEOL, Japan). The phase identification was characterized by using the X-ray diffraction (XRD, D8 advance, Bruker, Germany). The density of sintered Si 3 N 4 samples were tested by Archimedes' method. The thermal conductivity of the samples was calculated using the formula k ¼ ραC p . The thermal conduction coefficient (α) was measured by the laser scattering method (LFA447, NETZSCH, Germany) with a dimension of 10mm � 10mm � 2 mm. The bending strength of samples was measured by a three-point bending test with a universal mechanical testing machine (Inspekt Table Blue 05, Hegewald & Peschke, Germany). The sample size for evaluating bending strength was 1.5 mm � 2.0 mm � 25 mm. The Vickers hardness and fracture toughness of samples was evaluated using a Vickers hardness tester (Tukon2500Bm Wilson, America) under a 5 kg loading condition for 15 s on a polished surface. A line grid of printed single ceramic-resin layer was used to evaluate and characterize the cure depth and excess cure width. The cure depth and cure excess width were measured using a Gauge meter (7002-10, Mitutoyo, Japan) and an optical microscope (CDM-806H2, Tuming, China), respectively. The line grid was first placed on the platform of the Gauge meter and then gaging probe was put on the grid to obtain the cure depth of the line grind. The width of the line grind was measured by an optical microscope, and then subtracted the width of line grid designed to get the excess cure width.

Characteristics of oxidized Si 3 N 4 slurries
The distribution of particle size of Si 3 N 4 powders under different oxidation times is shown in Figure 3. The particle size of Si 3 N 4 powders increased with increasing oxidation time. With increasing oxidation time from 0 h to 2 h, the particle size of Si 3 N 4 powder increased from 0.965 μm to 0.984 (0.5 h), 0.992 (1.0 h) and 1.062 (2.0 h) μm, respectively. It was because that an amorphous SiO 2 was produced by the reaction between Si 3 N 4 and Þ during the oxidation process, and the thickness of SiO 2 layer increased with the increment of holding time. Figure 4 shows the TEM micrographs of pristine and oxidation 1 h Si 3 N 4 powders. As shown in Figure 4a-b, the surface of pristine Si 3 N 4 is clear, and only a thin layer of amorphous SiO 2 can be observed, which can be attributed to the inevitable and slight oxidation during production [34]. By comparison, an obvious amorphous SiO 2 can be seen on the surface of oxidation 1 h Si 3 N 4 powders (Figure 4c-d). Consequently, these phenomena indicate that the oxidation process is an effective approach for enhancing the thickness of SiO 2 on the surface of Si 3 N 4 powders.
Considering the photocuring principle of DLP, it is necessary to investigate the effect of the oxidation process on the light absorbance of Si 3 N 4 powders, which can provide some basic information for optimizing the curing behavior of Si 3 N 4 suspensions. dark or gray color has the ability to absorb more light [35]. Owing to the oxidation process formed a SiO 2 layer on the surface of Si 3 N 4 powders and the thickness of the SiO 2 layer increased with oxidation time, the light absorbance of Si 3 N 4 decreased with oxidation time.
As it is well known, the rheology properties of the paste are another key parameter for the DLP process, and there are many factors that influence the rheology properties of ceramic suspension, such as particle size and its distribution, particle morphology, solid loading of suspension, interparticle forces, wettability between ceramic powder and resin [36]. The viscosity curves as a function of the shear rate of Si 3 N 4 slurries (solid loading: 30 vol.%) with different oxidation times are shown in Figure 5(b). It can be seen that all the suspensions exhibit clear shear thinning behavior. Furthermore, with the increment of oxidation time from 0 h to 1.0 h, the viscosity of Si 3 N 4 slurry obviously decreased. That may be because the morphology of Si 3 N 4 powders changed from irregular to near sphere by forming a SiO 2 layer on the surface of Si 3 N 4 powder (as shown in Figure 4c-d). Compared with irregular, the near sphere shape of oxidized Si 3 N 4 powders enables to reduce the friction between particles, which is advantageous to slack viscosity [37]. However, the viscosity increased significantly when oxidation time continuously raised to 2.0 h. The result  can be attributed to the aggregation of Si 3 N 4 powder by excessive oxidation. With increasing oxidation time to 2.0 h, the thickness of SiO 2 on the surface of Si 3 N 4 powder continues to grow (as shown in Figure 3). The excessive thickness of SiO 2 may encase the neighboring particles leading to serious aggregation, which will deteriorate the rheological property of suspension.
The cure depth and excess cure width as a function of the applied energy dose LnE for the Si 3 N 4 slurry (solid loading: 15 vol.%) with different oxidation times are shown in Figure 6. With the increase of the applied energy dose, the cure depth of the slurry increased. This is due to the higher applied energy, the more scattered energy initiated the polymerization of resin monomers. In addition, the cure depth increased as the oxidation time. Under a light intensity of 280 mJ/cm 2 , the cure depth of Si 3 N 4 slurry was 65 (pristine), 77 (oxidation 0.5 h), 84 (oxidation 1 h) and 89 μm (oxidation 2 h), respectively. There are three reasons to account for this phenomenon. Firstly, the cure depth is linearly related to the average particle size of powder. The essential relationship between cure depth (D c ) and the average particle size was described by Griffith and Halloran as follows [38]: where hdi is the average particle size, Δn is the refractive index difference between the ceramic and the medium, E 0 is the energy density, and Q is the scattering efficiency term. According to Eq. (1), the cure depth increases with increasing the average particle size of powder. After oxidation at high temperature, the average particle size of powder increased with oxidation time (as shown in Figure 3). Consequently, the cure depth increased with oxidation time. Secondly, it can be seen from Eq. (1), the cure depth is inversely proportional to the difference of refractive index between ceramic particle and photosensitive resin. The refractive index of Si 3 N 4 , SiO 2 and photosensitive resin is 2.1, 1.5 and 1.46, respectively [29]. The difference of refractive index between Si 3 N 4 and photosensitive resin is about 0.7. After the oxidation process, the refractive index value of Si 3 N 4 may dramatically reduce by forming a SiO 2 layer on the surface of Si 3 N 4 . Hence, the difference of refractive index between oxidized Si 3 N 4 and photosensitive resin reduced with increasing the holding time. At last, the reducing light absorbance of oxidized Si 3 N 4 can improve cure depth. The less the absorbance of the ceramic particle, the more the ultraviolet light could be used to stimulate photopolymerization. As shown in Figure 5(a), the oxidation approach can reduce the absorbance of Si 3 N 4 . As a result, the cure depth increased with oxidation time.
During the curing process, cure depth determines the bonding strength of layer-to-layer excess cure width plays a decisive role in the resolution and precision of fabricated ceramic parts [39,40]. The desired curing behavior of slurry should have high cure depth and low excess cure width at the same time [17]. The excess cure width has been investigated, as shown in Figure 6(b). In consistent with the results of cure depth, excess cure width of Si 3 N 4 slurry increased with oxidation time. This is due to the absorbance of Si 3 N 4 decreased with increasing the oxidation time. More ultraviolet light can be used to photopolymerization in the horizontal direction. In sum, because of Si 3 N 4 with oxidation 1 h not only has relatively low light absorbance and viscosity, but also achieves enough photocuring ability, it is an optimal parameter for printing.

Characteristic of surface modified oxidized-Si 3 N 4 slurry
Although Si 3 N 4 powder with oxidation 1 h presents good comprehensive behavior for the DLP process, how to improve the wettability between oxidized Si 3 N 4 powders and resins is still a big stumbling block. Surface modification by a silane coupling agent (KH560) has been proved as an effective way to improve wettability and stability of Si 3 N 4 slurry [32]. Hence, the pristine and oxidized (1 h) Si 3 N 4 powders with different content of KH560 have been researched, as shown in Table 1.
The particle size distribution of pristine and oxidized Si 3 N 4 powders with different content of KH560 is shown in Figure 7. It can be seen that the particle size of pristine and oxidized Si 3 N 4 powders increased with increasing the content of KH560, which can be attributed to the coating of KH560 increased the particle size of Si 3 N 4 powders. Figure 8 shows the TEM images of O-SN-1. It can be seen that a thin layer of film on the surface of the oxidized Si 3 N 4 powder, indicating that KH560 has modified the surface of oxidized Si 3 N 4 powder. Figure 9(a) shows the infrared absorption spectra for pristine and oxidized Si 3 N 4 modified with and without KH560. After modification, the intensity of characteristic peak ~3448 cm −1 increased significantly, which stems from the N-H absorption of the coupling agent KH560 and KH560 hydrolysis formation of -Si(OH) 3 absorption, respectively [33]. It proves that KH560 has successfully coated on the surface of Si 3 N 4 powders. Figure 9(b) presents the light absorbance of Si 3 N 4 powders under a wavelength from 300 to 800 nm. The detail of light absorbance of pristine and oxidized Si 3 N 4 powders with different content of KH560 is shown in Fig S1. It can be seen from Figure 9(b) and Fig S1, the pristine and oxidized Si 3 N 4 with and without KH560 have the similar light absorbance at 405 nm, illustrating that surface modification with KH560 has no impact on light absorbance of Si 3 N 4 powder. Additionally, due to forming a SiO 2 layer on the surface of Si 3 N 4 powder, the oxidized Si 3 N 4 powder modified with KH560 has lower light absorbance. Figure 9(c) shows the rheological properties of 30 vol.% solid loading Si 3 N 4 slurries and the detailed rheological properties of Si 3 N 4 slurries with different content of KH560 are shown in Fig S2. It can be seen from Fig S2, the viscosity of oxidized Si 3 N 4 slurries were dramatically reduced by the surface modification process. For another, the viscosity of pristine Si 3 N 4 slurries decreased linearly with the increasing KH560 content (Fig S2). This reduced effect of viscosity ascribed to KH560, acted as molecular bridge, promotes the compatibility between hydrophilic Si 3 N 4 powders and hydrophobic resins [41]. It can be seen from Figure 9 (c), after modified with 1 wt.% KH560, the viscosity of pristine Si 3 N 4 slurry dropped 72.8%, (decreased from 2350 to 650 Pa•s at a shear rate of 0.25 s −1 ). The viscosity of oxidized Si 3 N 4 slurry reduced 88.5%, decreased from 849 to 98 Pa•s. Owing to the SiO 2 containing abundant silicon hydroxyl (Si-OH) [42], the oxidized Si 3 N 4 powders is more likely to produce dehydration and condensation with the X group of KH560 [43]. Consequently, the oxidized Si 3 N 4 powders modified with KH560 can evenly distribute in the photosensitive resin and reduce more viscosity of the slurry. Figure 10 exhibits the sedimentation experiment of the Si 3 N 4 slurries (5 vol.%). As shown, SN, SN-1 and SN-2 present a slight settlement an hour later. With increasing the experiment time to 12 h or 36 h, an obvious precipitation phenomenon appeared in SN, SN-1 and SN-2. On the contrary, it can be found mild sedimentation in SN-3 and SN-4. The SN exhibits apparent sedimentation owing to the incompatibility between the hydrophilic Si 3 N 4 powders and hydrophobic photosensitive resin. In addition, due to the lack of KH560 (1 wt.% or 2 wt.%), the modification process (KH560) could not coat the overall surface of Si 3 N 4 particles. Consequently, the agglomeration of Si 3 N 4 particles was inevitable, SN-1and SN-2 presented bad stability. With increasing content of KH560, the Si 3 N 4 particle was fully covered by KH560, the particle distribution and rheological property of slurry have been further improved, and thus the stability of SN-3 and SN-4 was well. On the other hand, the stability of the oxidized Si 3 N 4 slurries with and without KH560 is significantly better. After 36 h, all oxidized Si 3 N 4 slurries still show good stability. Especially, the stability of O-SN and O-SN-1 was good after 168 h (as shown in Fig S3). This can mainly be attributed to two reasons: (1) SiO 2 coating is an effective approach for enhancing the stability of slurry. SiO 2 coating enables decrease the equipotential of particle, and then improve its distribution and prevent agglomeration [42,44]. (2) The functional group of the oxidized Si 3 N 4 powders has good chemical activity with KH560, and then improves the steric hindrance of Si 3 N 4 particles and promotes the wettability between the oxidized Si 3 N 4 powders and photosensitive resin. Figure 11 reveals the influence of KH560 modification on the cure depth and excess width of Si 3 N 4 slurry, and the detailed curing ability of pristine and oxidized Si 3 N 4 slurry with different content of KH560 are shown in Fig S4. It is clear that the surface modification with KH560 enables to reduce the cure depth of Si 3 N 4 slurry. This may because surface modification with KH560 greatly enhance the distribution of Si 3 N 4 powder in suspension. The very homogeneous distribution of Si 3 N 4 powders would occupy more gaps between particles, which lead to reducing the penetrability of ultraviolet light. Consequently, surface modification with KH560 enable to cut down the cure depth of Si 3 N 4 slurry. On the other hand, surface modification with KH560 has slight impact on excess cure width of Si 3 N 4 slurry (as shown in Figure 11(b)). The main reason for this phenomenon is that the refractive index of KH560 is similar to that of photo-sensitive resin. After surface modification by KH560, the difference of refractive index between KH560 modified Si 3 N 4 powder and photo-sensitive resin is about 0.0102 [32]. As a result, the KH560 surface modifier has no effect on the excess cure width of Si 3 N 4 slurry.

Post-processing of Si 3 N 4 ceramics
In order to obtain high cure behavior and low viscosity of Si 3 N 4 slurry and easily produce dense Si 3 N 4 ceramics, a 40 vol.% Si 3 N 4 slurry was prepared by O-SN-1 powders, and then the green Si 3 N 4 ceramic parts were fabricated by DLP. The SN-1 slurry was used as the contrast sample. The processing parameters of O-SN-1 of power density and layer thickness are 171 mJ/cm 2 and 15 μm, respectively. The processing parameters of SN-1 of power density and layer thickness are 143 mJ/ cm 2 and 15 μm, respectively. Figure 12 exhibits the macroscopic image of the green Si 3 N 4 ceramic parts by stereolithography, vacuum debinding and air debinding, respectively. It is clear that there are no surface cracks on these Si 3 N 4 ceramic parts. In order to compare the influence of 3D printing and conventional processing method, a set of Si 3 N 4 ceramic parts were fabricated by cool isostatic pressing (CIP) and pressureless sintering at the same condition, as shown in Table 2.
It can be seen from Table 2 that the density of 3D printed samples is roughly equal to the CIP samples. The density of pristine Si 3 N 4 samples is higher than the density of oxidized Si 3 N 4 samples. XRD spectrums of the sintered Si 3 N 4 samples are shown in Figure 13. The main crystalline peaks of these samples are β-Si 3 N 4 , and no α-Si 3 N 4 diffraction peaks are detected in all the sintered samples, suggesting that α-Si 3 N 4 has completely transform into β-Si 3 N 4 . In addition, there is no difference in diffraction peaks between the 3D printed samples and CIP samples. Figure 14 illustrates SEM micrographs of the polished surfaces of sintered Si 3 N 4 samples. As shown in Figure 14, all samples are basically dense after sintering, and the grain size of different samples is in the range of 1-2 μm. This is mainly due to the short sintering time under pressureless sintering, resulting in no obvious coarsening of the Si 3 N 4 grains. The microstructure of the SN-1 (DLP) (Figure 14(a)) and SN-1 (CIP) (Figure 14(b)) is compact, and a small number of pores can be found. In contrast, there are many pores on the surface of the O-SN-1 (DLP) (Figure 14(c)) and O-SN-1 (CIP) (Figure 14(d)). This may attribute to oversintering leading to rapid grain growth and restricting fully eliminating pores. The strong covalent bonds of Si 3 N 4 seriously constrain the diffusion of atoms, resulting in the densification of Si 3 N 4 ceramic relies heavily on liquid phase sintering. Compared to solid phase sintering, the liquid phase sintering has faster rate of atom diffusion. The more content of the liquid phase is, the faster the sintering driving forces is [45]. The main derivation of the liquid phase is eutectic  melting of sintering additives and SiO 2 (on the surface of Si 3 N 4 powder). Owing to the oxidation process enhanced the thickness of amorphous SiO 2 on the surface of Si 3 N 4 powders, the oxidized Si 3 N 4 powders can form more liquid phase during sintering process. Therefore, the oxidized Si 3 N 4 powder is more likely to appear oversintering under the same sintering temperature. These microstructure features agree with the results of sample density (Table 2). Furthermore, due to the oxidized Si 3 N 4 powders have faster sintering driving forces, the average particle size and aspect ratio of the oxidized samples (O-SN-1 (DLP) and O-SN-1 (CIP)) greater than that of the pristine samples (SN-1 (DLP) and SN-1 (CIP)). Lastly, the average particle size and aspect ratio of the 3D printed samples is lower than that of the CIP samples. It may attribute to the 3D printed green samples containing more point defects and having lower initial density. Hence, the sintering driving force of the 3D printed green sample is less than that of CIP samples. Figure 15 presents the mechanical properties and thermal conductivity of the sintered samples. As shown in Figure 15(    demonstrate that DLP is a favorable way to produce high-performance Si 3 N 4 ceramics. Furthermore, owing to the density of O-SN-1 (DLP) higher than that of the O-SN-1 (CIP) (as shown in  Figure 15(a) indicates that the pristine Si 3 N 4 ceramics own the higher Vickers hardness and bending strength due to higher density and smaller grain size, and the oxidized Si 3 N 4 ceramics have better fracture toughness as a result of relatively bigger grain size and aspect ratio. According to Hall-Petch relationship [46], the smaller the grain size is, the more the number of grain boundary is. The grain boundary is able to restrict the dislocation motion. Generally, the more difficult the dislocation motion is, the better the mechanical property is. Hence, grain refinement is beneficial to improve the mechanical properties, such as bending strength and Vickers hardness. On the other hand, the fracture toughness is closely related to bigger grain size and aspect ratio. The bigger grain size could enhance the distance of crack propagation, which is conducive to releasing more energy. Effective releasing energy can alleviate the fracture phenomenon and improve fracture toughness [47]. As shown in Figure 15(b), the thermal conductivity of SN-1 (DLP) and SN-1 (CIP) are 52.21 and 55.91 W/(m•K), respectively. After the oxidation process, the thermal conductivity of O-SN-1 (DLP) and O-SN-1 (CIP) are 41.38 and 38.33 W/(m•K), respectively. Evidently, the oxidation process reduced the thermal conductivity of the Si 3 N 4 ceramics from 55.91 to 38.33 W/(m•K), the drop has reached 31.4%. The main reason for this phenomenon is that the oxidation process will induce more oxygen in Si 3 N 4 ceramics, and the oxygen is able to form vacancy during the sintering process (defect formula: SiO 2 ! Si Si þ 2O c þ V Si ), which can greatly exacerbate the phonon scattering [48,49]. Consequently, the thermal conductivity of the oxidized Si 3 N 4 ceramics is bad.

Microstructure and properties of Si 3 N 4 ceramics
In addition, the thermal conductivity values of SN-1 (DLP) and SN-1 (CIP) are nearly equal, indicating that the DLP method has well potential to fabricate high thermal conductivity Si 3 N 4 ceramics.

Conclusion
In this study, the oxidation process was used to enhance the curing behavior of Si 3 N 4 slurry, and then silane coupling agent (KH560) was used to improve the rheological properties of the oxidized Si 3 N 4 slurry. The effect of Si 3 N 4 slurry with oxidation process and modification on its rheological behavior, light absorption, curing ability and stability have been systematically investigated. The oxidation process could effectively enhance the cure depth and stability of Si 3 N 4 slurry. Modification by KH560 enables to improve the rheological properties and wettability of Si 3 N 4 slurries. The cure depth of Si 3 N 4 slurries would be reduced by KH560 modification. A Si 3 N 4 slurry with abundant photocuring ability, high solid loading, low viscosity and reliable stability were fabricated by the oxidized (1 h) and KH560 (1 wt.%) modified Si 3 N 4 powders. After debinding and sintering, a dense Si 3 N 4 ceramic part was produced by this slurry via DLP 3D printing.