Mullite–TiC–c-BN–c-ZrO₂ materials Produced by Spark-Plasma Sintering and Their Properties

Different c-BN/c-ZrO₂ ratios are shown to affect the phase composition, microstructure, relative density, open porosity, linear shrinkage, physicomechanical properties, and linear correlation of the elastic modulus and toughness of mullite–TiC–c-BN–c-ZrO₂ samples during spark-plasma sintering at pressing load 70 MPa and 1200 – 1600°C. The synthesized TiC and c-BN powders and c-ZrO₂ spark-plasma sintered at 1400°C are characterized by extensive phase crystallization. Mullite and TiC develop profusely in sintered samples with different c-BN/c-ZrO₂ ratios. Increasing the c-BN/c-ZrO₂ ratio promotes ingrowth of more c-BN than c-ZrO₂ at 1200 – 1600°C and causes a less uniformly and densely sintered crystalline microstructure with many pores to form at 1500°C. This sample has lower physicomechanical properties, a poorer linear correlation of elasticity modulus and toughness at 1200 – 1600°C, and lower crack resistance at 1500°C.

The compatibility of oxide and non-oxide powders for spark-plasma sintering deserves special attention [1, 2] because of the different diffusion coefficients in the sintered powders [1,2,3]. It is especially critical for phase transformations, e.g., in boron nitride (BN) as its content in the sintered powder mixture and the temperature increase at low compression loads [1, 4]. This leads to uneven and incomplete sintering of powder mixtures in the transverse and/or longitudinal direction with formation of nonuniformly sintered microstructure and brittleness in boundary areas of oxide and non-oxide crystalline phases. As a result, the crack resistance decreases and the physicochemical properties of the materials are poorer [1,2,3,4].

The solution to this problem is to add rare-earth metals (e.g., Y₂O₃, Sm₂O₃, Dy₂O₃) that readily form low-melting eutectics with the oxide and non-oxide powders and solid solutions in the liquid phase because of the small cationic radii of the metals in them [5, 6]. The resulting low-melting eutectics promote nucleation of crystalline phases, stimulate diffusion of matter through the eutectic melt, and increase the contact area of the sintered particles. As a result, the powder mixture is more evenly and completely sintered [5,6,7]. The solid solutions compact the structure and strengthen it on microscopic and macroscopic levels at boundary areas of oxide and non-oxide crystalline phases and grain contacts, respectively. As a result, elastic properties are generated, the crack resistance increases, and the physicomechanical properties of the materials improve [7, 8]. On the other hand, these additives oxidize the non-oxide powder, change its composition, and diminish its content. The oxide powder particles have a different effect on sintering of mixtures of oxide and non-oxide powders, grain growth of non-oxide powder, and formation of a different glass-phase composition that increases the brittleness as the additive (oxide component) content, temperature, and compression load increase [8, 9]. As a result, the physicomechanical properties of the materials decrease significantly [8, 9].

ZrO₂ [10] and ZrB₂ powders [11] were added to the mixture of oxide and non-oxide powders to replace the used oxide powder additives and to improve the physicomechanical properties of the materials. However, these components in the powder mixtures caused uneven and incomplete solid-phase sintering. The optimal amount and ratio of additives had to be found to reduce their effects [10, 11].

The goal of the work was to study the effect of different c-BN/c-ZrO₂ ratios during spark-plasma sintering with compression load 70 MPa at 1200 – 1600°C on the phase composition, microstructure, relative density, open porosity, linear shrinkage, physicomechanical properties, and linear correlation of the elasticity modulus and toughness of mullite–TiC–c-BN–c-ZrO₂ samples.

Experimental

Preparation of mixtures of Al₂O₃ and SiO₂, TiC and c-BN powders, and sintered c-ZrO₂, preparation of mixtures of oxide and non-oxide powders

Al₂O₃ and SiO₂ powders were mixed (Table 1) according to the published method [12]. TiC and c-BN powders were synthesized in a plasma-chemical unit in vacuo at 1600°C for 1 h using TiO₂ (98.0%, Aldrich, Belgium), C (97.5%, Merck, Germany), and B₂O₃ powders (97.5%, Merck, Germany) and N₂ (99.5%, Aldrich, Belgium) and the reactions TiO₂ + 2C → TiC + CO₂ and 2 B₂O₃ + 3.5 N₂ → 4 c-BN + 3 NO₂. Cubic zirconia c-ZrO₂ was spark-plasma sintered in vacuo at compression load 35 MPa for 2 min at 1400°C. Starting ZrO₂ (97.5%, Merck, Germany) and Y₂O₃ powders (99.5%, Aldrich, Belgium) were used in a 17.58/1 ratio, which corresponded to a biphasic system on the ZrO₂–Y₂O₃ equilibrium phase diagram (by Brown and Odell and Fan Fu-Kanu and Keler) [13]. The component masses (97 mol% ZrO₂/3 mol% Y₂O₃) in units of g per 100 g of mixture were 94.62/5.38.

Table 1. Mass Proportions and Component Ratios in Starting Powder Mixtures*

Samples of powder mixtures of Al₂O₃ and SiO₂, TiC and c-BN, and c-ZrO₂ were prepared by the published method [12] at compression load 70 MPa.

Method for determining properties of powders and sintered samples

The phase composition of the synthesized and sintered powders, microstructure, relative density ρ_rel, open porosity φ, linear shrinkage ∆l, elasticity modulus E, Vickers hardness HV, and surface area S of each sample (Table 1) were calculated as before [12]. The theoretical densities (g/cm³) of the powder components were 3.17, mullite; 4.93, TiC; 3.49, c-BN; and 6.27, ZrO₂.

Toughness of the sintered samples was determined by a Vickers microcompression method using an AVK-A Hardness Testing Machine (Akashi Co., Japan) and was calculated using the formula

$$ {K}_{Ic}=0.073\left(P/{c}^{3/2}\right), $$

where K_Ic is the critical strain coefficient or toughness (MPa·m^1/2); P, load applied to test sample surface (kg/cm²); c, half-length of microcracks formed around indenter impression corners (mm).

Results and Discussion

Figure 1 illustrates the phase compositions of TiC and c-BN powders synthesized by a plasma-chemical method. It shows mainly strong diffraction peaks for TiC and c-BN with a small amount of TiC_xO_y. This phase was nonstoichiometric TiC containing unreacted TiO₂ and C.

Fig. 1.

Phase composition of TiC (a) and c-BN powders (b ) synthesized by a plasma-chemical method at 1600°C: TiC_xO_y is titanium oxycarbide.

Figure 2 illustrates the phase composition of sintered c-ZrO₂, which was characterized by distinct diffraction peaks for this phase. This was explained by strain and rearrangement of the ZrO₂ structure from tetragonal to cubic under the compression load and diffusion of Y³⁺ into the c-ZrO₂ structure, according to the ZrO₂–Y₂O₃ equilibrium phase diagram (by Brown and Odell and Fan Fu-Kanu and Keler) [13].

Fig. 2.

Phase composition of c-ZrO₂ spark-plasma sintered at 1400°C.

Figure 3 shows the phase compositions of mixtures of starting components spark-plasma sintered at 1200 – 1600°C.

Fig. 3.

Phase composition of M92TiC3BN5ZrO₂ (a) and M92TiC5BN3ZrO₂ samples (b ) sintered at 1200 – 1600°C: M, mullite (3Al₂O₃·2SiO₂); h-BN, hexagonal boron nitride; B₄C, boron carbide; TiN, titanium nitride.

Samples of all compositions typically underwent extensive mullitization at 1200 – 1600°C due to structuring and stoichiometric mullite formation. In a similar manner, TiC resulted from a transition into a viscoelastic (plastic) state that promoted diffusion and structuring of TiC under these sintering conditions. Strong ingrowth of c-BN and less of c-ZrO₂ were observed with increasing c-BN/c-ZrO₂ ratio at 1200 – 1600°C. However, c-BN and c-ZrO₂ grew less vigorously than mullite and TiC. This was explained by the denser structures of these components with covalent bonds in c-BN. This limited diffusion and structuring of c-BN and c-ZrO₂ in the solid phase. Also, the ingrowth of c-ZrO2 in the sample with 5 mol% c-ZrO₂ was slightly greater than that of c-BN in the sample with 5 mol% c-BN at 1200 – 1600°C. This was explained by better diffusion into c-ZrO₂ than into c-BN. In addition, h-BN was detectable in the sample with 5 mol% c-BN but not in that with 3 mol% c-BN because of a partial phase transformation of c-BN into h-BN in the solid in a sintered powder mixture with 5 mol% c-BN. These samples had different quantitative ratios of diffraction maxima for c-BN and c-ZrO₂ at 1200 – 1600°C (Fig. 3) because of different amounts of crystalline phases. The x-ray pattern of the sample with 5 mol% c-BN showed weak diffraction peaks for B₄C and TiN that were missing for the sample with 3 mol% c-BN at 1500 – 1600°C. These side phases formed from the reaction of TiC and h-BN in the sintered powder mixture. Mullite and c-ZrO₂ did not react simultaneously with TiC and c-BN because decomposition products of mullite and oxidation products of TiC and c-BN did not form at 1200 – 1600°C (Fig. 3).

Figure 4 shows microstructures of samples spark-plasma sintered at 1500°C. The microstructure of sintered composition M92TiC3BN5ZrO₂ (Fig. 4a ) was more uniformly and densely sintered, crystalline, and fine-grained with fewer pores and weakly sintered sections than that of composition M92TiC5BN3ZrO₂ (Fig. 4b ). This was explained by the different c-BN/c-ZrO₂ ratios in the sintered compositions (Table 1). Sintering was promoted more by diffusion of the oxide component in the solid phase at high c-ZrO₂ and c-BN ratios. Viscous flow of mullite and TiC melts also helped to form microstructures with different uniformities and densities. This was explained by the different sintering behaviors of c-BN and c-ZrO₂ particles in the presence of these melts.

Fig. 4.

Microstructure of M92TiC3BN5ZrO₂ (a) and M92TiC5BN3ZrO₃ samples (b ) sintered at 1500°C.

Figures 5 – 8 show measured ρ_rel, φ, ∆l, E, K_Ic, HV, and a photograph of impressions and microstructures of samples with different c-BN/c-ZrO₄ ratios at 1200 – 1600°C and 1500°C.

Fig. 5.

Parameters ρ_rel, φ, and ∆l of samples with different c-BN/c-ZrO₂ ratios that were sintered at 1200 – 1600°C: M92TiC3BN5ZrO₂ (1 ) and M92TiC5BN3ZrO₂ (2 ).

Composition N92TiC3BN5ZrO₂ had a decreased φ and the maximum degree of sintering (91.2%) at 1600°C. This was due to the more even filling of pores because of viscous flow of mullite and TiC melts and initiation of c-ZrO₂ diffusion at 1200 – 1300°C. Diffusion of c-ZrO₂ became greater than that of c-BN in the solid phase at 1300 – 1500°C (Fig. 3a ). Sintering slowed at 1500 – 1600°C with φ > 8% at 1600°C. This was explained by less extensive and incomplete pore filling during solid-state sintering of c-BN and c-ZrO₂ particles. The sintering results correlated with the sample microstructure at 1500°C (Fig. 4a ).

The rises of ρ_rel and ∆l and fall of φ for composition M92TiC5BN3ZrO₂ at 1200 – 1600°C were irregular. Extensive sintering up to 1400°C was explained by pore filling resulting from viscous flow of mullite and TiC melts (Fig. 3b ). Slow sintering at 1400 – 1500°C was due to weaker diffusion of c-ZrO₂ (at 3 mol%) and c-BN because of the increasing partial phase transformation c-BN → h-BN in the solid phase (Fig. 3b ) with the latter formed as a layer on the sintered particles that hindered diffusion of matter between particles and pore filling. This was confirmed by studying microstructures of boundary regions of sample crystalline phases (Fig. 6b₁ – b₄). The sintering was reduced somewhat because of the formation of a certain amount of h-BN, similar crystalline phases B₄C and TiN (Fig. 3b ), and a slight porosity increase at 1500 – 1600°C. This correlated with the microstructure of the sample sintered at 1500°C (Fig. 4b ).

Fig. 6.

Microstructure of boundary regions of mullite, c-ZrO₂, TiC, c-BN, and h-BN in compositions M92TiC3BN5ZrO₂ (a, a₁) and M92TiC5BN3ZrO₂ (b, b₁ – b₄) sintered at 1500°C.

The crystalline, more uniform, and denser sintered fine-grained microstructure of the sample with 3 mol% c-BN (Fig. 4a ) helped significantly to improve the elastic properties and, as a result, to increase the resistance of the sample to external applied loads with high K_Ic and HV values at 1500°C (Fig. 7). On the other hand, the high parameters for these properties were due to insignificant mullite–TiC (Fig. 6a ), mullite–c-ZrO₂–c-BN, and c-ZrO₂–c-BN boundary regions (Fig. 6a₁). The sample had short microcracks along straight-line trajectories (Fig. 8a ). The impression showed slight damage as a chip (Fig. 8a ), indicating little accumulation of internal stresses and minimally brittle boundary regions of these crystalline phases (Fig. 6a, a₁). This corresponded to extensive solid-state sintering of c-ZrO₂ and c-BN particles at 1400 – 1600°C (Fig. 5).

Fig. 7.

Parameters E, K_Ic, and HV of samples with different c-BN/c-ZrO₂ ratios that were sintered at 1200 – 1600°C: M92TiC3BN5ZrO₂ (1 ) and M92TiC5BN3ZrO₂ (2 ).

Fig. 8.

Impressions from measuring HV on samples M92TiC3BN5ZrO₂ (a) and M92TiC5BN3ZrO₂ (b ) sintered at 1500°C.

The sample with 5 mol% c-BN had less drastic changes of elastic properties (K_Ic and HV) up to 1400°C. The increase of physicomechanical properties slowed at 1400 – 1500°C and decreased slightly at 1500 – 1600°C (Fig. 7). This was due mainly to h-BN formation at boundary sections of sample crystalline phases (Fig. 6b, b₁ – b₄) because of partial phase transformation of c-BN into h-BN at 1500 – 1600°C (Fig. 3b ). The resulting h-BN embrittled the structure with noticeable boundary regions of crystalline phases (Fig. 6b ) that were less extensive at mullite–c-BN and c-ZrO₂–c-BN boundary regions (Fig. 6b₁, b₃) and more extensive at TiC–c-BN and c-BN–h-BN boundary regions (Fig. 6b₁ – b₄). These properties decreased because the microstructure was incompletely sintered and the sample had many pores (Fig. 4b ). As a result, the crack resistance of the sample to formation of many microcracks propagating along tortuous trajectories of more weakly sintered regions (Fig. 6b₂, b₄) and chips (Fig. 8b ) was diminished. This was caused by the high brittleness of the sample at the TiC–c-BN and c-BN–h-BN boundary regions (Fig. 6b₂, b₄).

Figure 9 shows the linear correlation of E and K_Ic. A comparison of the R² parameters of the approximations with 3 and 5 mol% c-BN showed that the difference was small at 0.01 and was greater for the sample with 3 mol% BN, which agreed with the greater increase of physicomechanical properties (Fig. 7). This was due to the formation of a uniformly and densely sintered fine-grained microstructure with few pores (Fig. 4a ) and strengthening of the structure at mullite – TiC (Fig. 6a ), mullite – c-ZrO₂ – c-BN, and c-ZrO₂ – c-BN boundary regions at 1500°C (Fig. 6, a₁). As a result, the E and K_Ic values of the sample correlated well with the straight line at 1200 – 1600°C and slightly less at 1400°C. This was related to initiation of solid-state sintering because of extensive diffusion of c-ZrO₂ and less of c-BN at this temperature (Fig. 5). The high correlation of the sample properties with the straight line was indicative of uniform viscous-flow sintering at 1200 – 1300°C and the corresponding solid-state sintering at 1300 – 1600°C.

Fig. 9.

Linear correlation of E and K_Ic for samples sintered at 1200 – 1600°C.

The lower R² values for the sample with 5 mol% c-BN were explained by irregularly and incompletely sintered microstructure with many pores (Fig. 4b ) and a weaker structure with formation of thicker h-BN in the TiC – c-BN and c-BN – h-BN boundary regions at 1500°C (Fig. 6, b₂, b₄). As a result, the E and K_Ic values deviated more from the straight line at 1600°C than those at 1200, 1300, 1400, and 1500°C, which did not deviate. This indicated that the effects of these processes were greatest and caused nonuniform solid-state sintering of the composition at 1600°C. The different effects of the correlated properties on the R² values in both samples were explained by the different sintering mechanisms and, as a result, the different degrees of sintering of the corresponding compositions at 1200 – 1600°C (Fig. 5). The E and K_Ic points (values) correlated best with the straight line and, as a result, the R² values were greater for the sample with 3 mol% c-BN at 1200 – 1600°C because of the complete and uniform solid-phase sintering.

Conclusion

Different c-BN/c-ZrO₂ ratios had different effects during spark-plasma sintering of compositions at compression load 70 MPa and 1200 – 1600°C on the phase composition, microstructure, ρ_rel, φ, ∆l, physicomechanical properties, and linear correlation of E and K_Ic of mullite–TiC–c-BN–c-ZrO₂ samples. The synthesized TiC and c-BN powders and c-ZrO₂ spark-plasma sintered at 1400°C were extensively crystallized.

Sintered samples with different c-BN/c-ZrO₂ ratios showed extensive mullite and TiC ingrowth. Increasing the c-BN/c-ZrO₂ ratio promoted c-BN ingrowth more than c-ZrO₂ at 1200 – 1600°C and caused a less uniformly and densely sintered crystalline microstructure with many pores to form at 1500°C. As a result, the sample had lower ρ_rel, ∆l, and physicomechanical properties at 1200 – 1600°C, less crack resistance at 1500°C, and a poorer linear correlation of E and K_Ic at 1200 – 1600°C.