Y-H. Koh et al. Journal of the European Ceramic Society 24(2004)699-703 wt% alumina(HP-DBM, Reynolds, Bauxite, AK, 3. Results and discussion USA)as sintering additives. It was then ball-milled in ethanol with silicon nitride balls as media for 24 h. After The typical microstructure of the Si3 N4/BN fibrous milling, the mixture was dried in a rotary vacuum eva monolith is shown in Fig. 1. Flat hexagonal Si3 Na cells orator and subsequently passed through a 60-mesh were surrounded by continuous BN cell boundaries screen. The powder was mixed again with a polymer According to the XRD analyses, the Si3 N4 was com binder(methylcellulose), plasticizer(glycerol), and sol- pletely transformed into the B-phase and no reaction vent(distilled water)at room-temperature for extrusion. occurred between the Si3N4 and the BN. However, EDS The cell boundary material was prepared by mixing analyses showed that a glass phase(Si-Y-Al-N-O)was the bn powder (Grade MBN, Boride Ceramics present at the cell boundaries as well as in the cells Composites Ltd, UK) with 20 wt Al2O3 as a sintering implying that the glass phase migrated into the cell aid. A polymer binder(PVB), dispersant Menhaden boundaries during the hot-pressing fish oil), and solvent(trichloroethyleneethanol) were The thickness of the cell boundary layer was con- also added, and mixed by ball milling for 6 h. The trolled by adjusting the bn concentration in the dip amount of BN-containing material increased up to 25 coating slurry, and was measured by image analyses of wt% for different cell boundary thicknesses. The as- SEM micrographs(see Fig. 1). The boundary layer received bn powder showed platelets with a dimension thickness was found to be linearly dependent on the bn of 1-10 um and a thickness of 0.1-0.5 um. The total concentration as shown in Fig. 2. Therefore, fibrous oxygen content, including B2O3, was less than 5 wt % monoliths with desired cell boundary thicknesses were The cell was made by extruding the Si3 N4-polymer fabricated simply by changing the bn concentration in compound into 300 um diameter fibers and was the slurry. The spec coated with bn by passing through the BN-containing thickness represents the monolithic Si3N slurry, and subsequently arrayed by an automatic The cell boundary thickness was found to have a winding machine. The layered green billets were cut into strong influence on the fracture behavior of the fibrous the desired dimension, and then dried in the oven at materials. Representative stress versus deflection curves 80C for 12 h to improve the shape and strength of cell of the specimens with different cell boundary thick by hardening. The layered green billets were inserted nesses are shown in Fig. 3. Monolithic Si3 N4 exhibited into a mold of 40x40 mm and pressed at 0.5 MPa typical catastrophic failure [Fig. 3(A)] as is frequently Binder burnout of billets occurred in 900oC with heat- observed in other ceramics. However, the Si3 N4 fibrous ing rate of 2-3C/min and maintained for 3 h in flowing monolith with 18-um thick bn cell boundary was not nitrogen. Billets were hot-pressed at 1800C for I h completely fractured even after reaching critical stress, under an applied load of 30 MPa in a flowing nitrogen as shown in Fig. 3(B). The gradual fractures became atmosphere. more profound with increased cell boundary thickness Specimens for mechanical testing were machined into When the cell boundary thickness reached 37 um, the a bar shape with dimensions of 3x4x25 mm and material was fractured in a non-brittle manner [Fig. 3 ground with a 800-grit diamond wheel. The tensile side ( C)], which implied active crack interactions at the cell specimens was p with diamond slurries boundaries down to I um, and then chamfered to minimize the machining flaws. The flexural strength was measured using a four-point flexural configuration with a cross- head speed of 0.05 mm/min, and inner-and outer-spans of 10 and 20 mm, respectively. From the load versus crosshead deflection response, the apparent work-of- fracture was calculated by estimating the area under the load-deflection curve and dividing by twice the cross- sectional area of the sample. Apparent fracture toughness was measured using the single-edge-notched beam (SENB)method. A straight notch with depth and width of 1. 2 and 0.3 mm, respectively, was made at the center of the tensile surface using a thin diamond blade. The notched specimen was fractured using the four-point flexural configuration mentioned earlier. The density of the specimens was measured using the Archimedes 400m method. The specimens were characterized by scanning electron microscopy(SEM), energy dispersive spectro- Fig. 1. A representative SEM micrograph of cross-section of a fibrous metry(EDS), and X-ray diffraction(XRD) Si3N4/BN monolithwt.% alumina (HP-DBM, Reynolds, Bauxite, AK, USA) as sintering additives. It was then ball-milled in ethanol with silicon nitride balls as media for 24 h. After milling, the mixture was dried in a rotary vacuum eva￾porator and subsequently passed through a 60-mesh screen. The powder was mixed again with a polymer binder (methylcellulose), plasticizer (glycerol), and sol￾vent (distilled water) at room-temperature for extrusion. The cell boundary material was prepared by mixing the BN powder (Grade MBN, Boride Ceramics & Composites Ltd, UK) with 20 wt.% Al2O3 as a sintering aid. A polymer binder (PVB), dispersant (Menhaden fish oil), and solvent (trichloroethylene/ethanol) were also added, and mixed by ball milling for 6 h. The amount of BN-containing material increased up to 25 wt.% for different cell boundary thicknesses. The as￾received BN powder showed platelets with a dimension of 1–10 mm and a thickness of 0.1–0.5 mm. The total oxygen content, including B2O3, was less than 5 wt.%. The cell was made by extruding the Si3N4-polymer compound into 300 mm diameter fibers and was coated with BN by passing through the BN-containing slurry, and subsequently arrayed by an automatic winding machine. The layered green billets were cut into the desired dimension, and then dried in the oven at 80 C for 12 h to improve the shape and strength of cell by hardening. The layered green billets were inserted into a mold of 4040 mm and pressed at 0.5 MPa. Binder burnout of billets occurred in 900 C with heat￾ing rate of 2–3 C/min and maintained for 3 h in flowing nitrogen. Billets were hot-pressed at 1800 C for 1 h under an applied load of 30 MPa in a flowing nitrogen atmosphere. Specimens for mechanical testing were machined into a bar shape with dimensions of 3425 mm and ground with a 800-grit diamond wheel. The tensile side of the specimens was polished with diamond slurries down to 1 mm, and then chamfered to minimize the machining flaws. The flexural strength was measured using a four-point flexural configuration with a cross￾head speed of 0.05 mm/min, and inner-and outer-spans of 10 and 20 mm, respectively. From the load versus crosshead deflection response, the apparent work-of￾fracture was calculated by estimating the area under the load-deflection curve and dividing by twice the cross￾sectional area of the sample. Apparent fracture toughness was measured using the single-edge-notched beam (SENB) method. A straight notch with depth and width of 1.2 and 0.3 mm, respectively, was made at the center of the tensile surface using a thin diamond blade. The notched specimen was fractured using the four-point flexural configuration mentioned earlier. The density of the specimens was measured using the Archimedes method. The specimens were characterized by scanning electron microscopy (SEM), energy dispersive spectro￾metry (EDS), and X-ray diffraction (XRD). 3. Results and discussion The typical microstructure of the Si3N4/BN fibrous monolith is shown in Fig. 1. Flat hexagonal Si3N4 cells were surrounded by continuous BN cell boundaries. According to the XRD analyses, the Si3N4 was com￾pletely transformed into the b-phase and no reaction occurred between the Si3N4 and the BN. However, EDS analyses showed that a glass phase (Si–Y–Al–N–O) was present at the cell boundaries as well as in the cells, implying that the glass phase migrated into the cell boundaries during the hot-pressing. The thickness of the cell boundary layer was con￾trolled by adjusting the BN concentration in the dip￾coating slurry, and was measured by image analyses of SEM micrographs (see Fig. 1). The boundary layer thickness was found to be linearly dependent on the BN concentration as shown in Fig. 2. Therefore, fibrous monoliths with desired cell boundary thicknesses were fabricated simply by changing the BN concentration in the slurry. The specimen with 0 mm-cell boundary thickness represents the monolithic Si3N4. The cell boundary thickness was found to have a strong influence on the fracture behavior of the fibrous materials. Representative stress versus deflection curves of the specimens with different cell boundary thick￾nesses are shown in Fig. 3. Monolithic Si3N4 exhibited typical catastrophic failure [Fig. 3(A)] as is frequently observed in other ceramics. However, the Si3N4 fibrous monolith with 18-mm thick BN cell boundary was not completely fractured even after reaching critical stress, as shown in Fig. 3(B). The gradual fractures became more profound with increased cell boundary thickness. When the cell boundary thickness reached 37 mm, the material was fractured in a non-brittle manner [Fig. 3 (C)], which implied active crack interactions at the cell boundaries. Fig. 1. A representative SEM micrograph of cross-section of a fibrous Si3N4/BN monolith. 700 Y.-H. Koh et al. / Journal of the European Ceramic Society 24 (2004) 699–703