Effect of solid loading on properties of reaction bonded silicon carbide ceramics by gelcasting

Abstract

Reaction bonded silicon carbide (SIC) ceramics were successfully fabricated by gelcasting using a new non-aqueous, low-toxicity gel system based on the polymerization of a phenol-formaldehyde resin and furfuryl alcohol. The effect of solid loading on the rheological properties of slurries, and the properties of green bodies and sintered ceramics, have been systematically investigated. The solid loading of a suspension able to meet the requirements of the casting process is as high as 70 wt%. The drying shrinkage and green strength of the SiC/polymer network green bodies decrease with solid loading ranging from 55 to 70 wt% because of the reduction of pre-mixed solutions. On the other hand the fracture toughness and bending strength of the sintered ceramics increase with increase in solid loading. The highest values of strength and toughness were observed at 378 ± 16 MPa and 4.22 ± 0.13 MPam^1/2, respectively, for composites produced at a solid loading of 70 wt%.

---

1. Introduction

Because of its obvious benefits of near net shape forming, high dry density and strength, low levels of organic additives, and a high degree of homogeneity, gelcasting has received considerable attention and has been found to be particularly useful in the fabrication of high-quality, complex-shaped ceramic parts.^1–4 In the gelcasting process ceramic powders are dispersed by ball milling into a pre-mixed solution containing monomer, crosslinker and solvent, to form a castable slurry. After adding catalyst and initiator, the slurry is placed in a pore free mold of the required shape and the entire system is polymerized in situ to immobilize the dispersed ceramic powders in the gelled portion, and a green body with excellent mechanical properties but containing only a few percent of polymer is obtained. Dried green bodies are thus easily machined.

Up to the present time, acrylamide (AM) has been the gel monomer most widely used in gelcasting of ceramics. However, the industry has been reluctant to use AM, firstly because it is a neurotoxin and, secondly, because its polymerization is inhibited by oxygen.^5,6 The search for a non-toxic, gel forming system that still has the ability to provide mechanical properties similar to the AM-based system has been a focus for research for more than 10 years.^7–11 In our previous study,¹² a new non-aqueous gelcasting system comprising a phenol-formaldehyde resin (PF) and benzenesulfonyl chloride was developed for the casting of reaction bonded silicon carbide ceramics. Ethylene glycol was the solvent and furfuryl alcohol (FA) was used to modify polymerization during gelcasting. Additionally, the porous carbon materials formed by the pyrolysis of the resin–glycol mixture were used as the carbon source. Gel time was modified by altering the ratio of FA, phenolic resin and curing catalyst in the pre-mix, and the viscosity of the silicon carbide (SiC) suspensions was adjusted by varying the proportion of polyethylene glycol dispersant (PEG 400).

However, the green body shows considerable shrinkage during curing and carbonization and this tends to cause deformation or cracking, especially in areas of complex shape.^13,14 In addition, the sintered bodies contain a relatively high volume fraction of free silicon. In the case of reaction bonded SiC, residual silicon within the residual pores is known to have a detrimental effect on the mechanical properties and reliability,^15,16 resulting from the fact that the strength and fracture toughness of silicon are less than 100 MPa and 1 MPam^1/2, respectively.

The primary purpose of the present study was to prepare reaction bonded SiC ceramics of complex shape and low residual silicon, based on gelcasting and reaction sintering. The effect of blend ratio and solid loading on the rheological behavior of non-aqueous SiC–resin slurries were also investigated, together with an evaluation of the microstructure and mechanical properties of the green and sintered samples.

2. Experimental

2.1 Sample preparation

The starting materials and flow chart for the gelcasting process have been illustrated in our previous study.¹² Two commercial α-SiC powders, labeled A and B (purity 98.5%; Zibo Huanyu Grinding Material, China), with a particle size of 3 μm and 45 μm, respectively, were used as raw materials. The polymeric components were PF and FA, and were used as binders as well as providing an additional source of carbon. Ethylene glycol (EG) was used both as a fugitive phase and as solvent. PEG 400 was chosen as the dispersant. The raw powders, together with 1.0 wt% PEG 400, were added to a pre-mixed solution of PF, FA and EG, followed by ball milling using SiC balls. The initial milling was conducted for 20 h using a planetary ball mill (QM-3SP4, Nanjing NanDa Instrument Plant, China) at a rotational speed of 240 rpm. Benzenesulfonyl chloride was added to the suspension as a curing catalyst and ball milling was continued for a further 2 h. The slurry was degassed for 15 min in a rotary evaporator (TP–08, Beijing Orient Sun-Tec Co. Ltd., Beijing, China) and then poured into a non-porous plaster mold. After pre-curing for 2 h at 80 °C, the rigid body was cured at 150 °C for 16 h and pyrolyzed at 800 °C at a programmed heating rate of 2 °C min⁻¹ in a flow of nitrogen. Finally, the green body obtained was sintered with liquid silicon infiltration at 1700 °C for 30 min in vacuo.

2.2 Measurements

The rheological behavior of the concentrated suspensions after ball milling at 240 rpm for 20 h was measured on a strain-controlled rotation rheometer (Model NXS-11, Chengdu Instrument Plant, China) at room temperature. A couette (cup diameter: 36.8 mm; bob diameter: 35.0 mm; bob length: 37.3 mm) was used for the rheological measurements. To avoid the influence of mechanical history, all samples were preconditioned for 1 min at a shear rate of 50 s⁻¹ and allowed to stand for a further 1 min before collecting the data. The microstructure of the polished and fracture surfaces was observed using a laser scanning confocal microscope (OLS3100, Olympus, Japan) and by scanning electron microscopy (SEM). The pore structure of the bodies after pyrolysis was examined using a mercury porosimeter (PoreSizer 9500, 2020M, Micromeritics, USA). Before measurement, all the samples were degassed under vacuum for 3 h at 200 °C.

The bulk density of the sintered samples was determined using Archimedes' principle. The composition of the specimens was determined by X-ray diffraction (XRD, D/Max–RB, Rigaku, Japan) using Cu Kα radiation, at a scanning rate of 2° min⁻¹, with scanning angles ranging from 20° to 80° and a sampling width of 0.02°. The flexural strength of the green and sintered bodies was measured using a three-point bending test (Model 5569 Materials Testing System, Instron, USA) on 36 × 4 × 3 mm test bars, using a loading span of 30 mm and a crosshead speed of 0.5 mm min⁻¹ at room temperature. The fracture toughness of the sintered bodies was evaluated using single-edge notched bend beams of 22 × 4 × 2 mm, and a notch depth and width of 2 mm and 0.2 mm, respectively, using a span of 16 mm and a crosshead speed of 0.05 mm min⁻¹. A minimum of five specimens were tested under each experimental condition.

3. Results and discussion

3.1 Rheological properties of SiC suspension

3.1.1 Effect of blend ratio. Preparation of a stable, low viscosity, high solids ceramic suspension is a prerequisite for successful gelcasting. The particle size distribution influences the packing ability of the powders within a suspension, the finer particles filling the voids between the larger particles, and as a result determining the rheological properties of the suspension. The use of bidisperse suspensions offers two advantages,^17–19 firstly, that for a given solid volume fraction the viscosity of a bidisperse suspension is generally lower than that of a monodisperse suspension, and secondly, that for a suitable viscosity the use of a bidisperse suspension can give a higher solid volume fraction than that of a monodisperse suspension.

Fig. 1 illustrates the effect of the blend ratio of SiC powders on the equilibrium viscosity of the suspensions at a constant solid loading of 60 wt% and dispersant content of 1.0 wt%. Suspensions containing a high proportion of SiC B (the coarser powder) gave lower viscosity at shear rates below 10 s⁻¹. As more SiC A was gradually replaced by SiC B the viscosity decreased, reaching a minimum when the volume fraction of SiC B reached 80%. At shear rates above 10 s⁻¹, however, suspensions with a blend ratio of 70% gave the lowest viscosity. A number of experimental and theoretical findings^20–22 have shown that, for a given solid loading and ratio of large to small diameter particles, there is a critical value of blend ratio at which the viscosity is at a minimum.

Fig. 1 Steady shear viscosity of two-component SiC suspensions with different weight fractions of SiC B.

In addition, Fig. 1 also shows that the suspensions all displayed shear thinning, with the formation of a plateau at high shear rates. At low shear rates the suspension structure was close to the equilibrium structure at rest, because thermal motion dominates the viscous force.²³ At higher shear rates the viscous forces affect the suspension structure more, and shear thinning occurs. Shear thinning behavior generally indicates a more favorable two-dimensional layered arrangement of the particles, because a flow-induced, layered arrangement can decrease the resistance of particle movement between layers.²³ At very high shear rates the viscous forces are dominant, and the plateau in the viscosity in this region is a measure of the resistance to flow of a suspension comprising a completely hydrodynamically controlled structure.

3.1.2 Effect of solid loading. The suspension for gelcasting has to satisfy two requirements, good fluidity and high solid loading.²⁴ On the one hand, a suspension with good fluidity has a uniform structure and ensures that the slurry can be poured successfully into the mold; on the other hand, a higher solid content can result in both higher density and less deformation and can cause defects in both green and sintered parts. Fig. 2 shows the relationship of viscosity to shear rate for SiC–resin slurries of different solid loading. The viscosity of all the slurries decreased with increasing shear rate. At 55 wt% solid loading the viscosity was approximately 780 mPa s at 60 s⁻¹, and as the solid loading increased, the viscosity also increased correspondingly. The viscosity of slurries at 60, 65, 70 and 75 wt% solid loading was 1.0, 1.58, 3.57 and 7.35 Pa s, respectively.

Fig. 2 Properties of SiC suspensions at a range of solid loadings.

In gelcasting, the preparation of high solid slurries is a key issue, because a higher amount of solids tends to reduce shrinkage and carbonization during curing and reduces the possibility of inhomogeneity and cracking. Fortunately, the viscosity of slurries with solid loadings as high as 70 wt% may still meet the requirements of the gelcasting process that follows.

3.2 Characterization of green bodies

The relative shrinkage of green bodies versus solid loading of the slurry is shown in Fig. 3. The relative shrinkage during curing decreases with decreasing solid content, reducing from 10.2% at 55 wt% to 5.6% at 75 wt%. In addition, the shrinkage of green bodies during carbonization also decreases from around 2.4% to as low as 0.2%.

Fig. 3 The effect of solid loading on the linear shrinkage of green bodies.

After casting in a mold at a constant temperature of 70 °C SiC/resin suspensions became consolidated within 10 min, and well shaped green bodies without a surface exfoliation layer were obtained. The mechanical properties of the green bodies at different solid loading are illustrated in Fig. 4, and it will be seen that the mechanical properties are significantly better than those obtained by conventional forming procedures, such as slip casting, tape casting or some other gelcasting procedures,^25–27 and the green parts can, therefore, be readily handled and machined in the green state. The high strength results from the crosslinking gel network and homogenous packing of the particles.

Fig. 4 Flexural strength of the cured bodies as a function of solid loading.

As the solid loading was increased from 55 to 70 wt%, the flexural strength of the green bodies gradually decreased from 21.2 ± 1.9 MPa to 12.7 ± 1.5 MPa. With increased solid loading, the polymer networks retain more of the SiC particles and the crosslinking of the powders is weakened, giving a decrease in green strength. However, when the solid loading was 75 wt%, the green mechanical properties showed a sharp decline. This might have been because of the presence of pores of a few hundred micrometers in size, as shown in Fig. 5, which are not observed in green bodies at a lower solid loading. When the solid loading is too high, gases in the slurry cannot easily escape, because of the high viscosity.

Fig. 5 Fracture morphology of green bodies at a solid loading of 75 wt%.

The fracture surface of the green bodies was examined by SEM. From the microstructure shown in Fig. 6, a homogeneous structure with no obvious flaws, cracks or bubbles could be seen, suggesting that a well-dispersed slurry, satisfactory de-airing and good casting can be achieved by the new gelcasting system. SiC powders are connected by polymer networks, shown in Fig. 6, and these account for the high strength of the green body. With increasing solid loading from 55 to 70 wt%, the thickness of the polymer networks on the SiC powder obviously decreases. An analysis of the polymer networks shows many interconnected pores of nanometer size, as illustrated in the inset to Fig. 6. It is, therefore, believed that phase separation occurs during the course of polymerization of PF and FA, only two phases occurring in the premixed solutions, a polymeric resin-rich phase and an ethylene glycol-rich phase.²⁸ After curing and carbonization, the polymeric resin-rich phase becomes the carbon matrix and the ethylene glycol-rich phase is removed, leaving pores in the carbon matrix.

Fig. 6 Micrograph of the fractured surface of the green bodies at different solid loading: (a) 55 wt%, (b) 60 wt%, (c) 65 wt%, and (d) 70 wt%.

After carbonization at 800 °C for 0.5 h in a flow of nitrogen, the pore size distribution of the SiC/porous carbon samples with different solid contents was determined by mercury intrusion porosimetry. As shown in Fig. 7, the pore size showed a monomodal distribution. A single peak distribution in pore size indicates that the SiC/porous carbon bodies were homogeneous in structure. The pore size distributions of all the samples were almost equal and very narrow, most of pores lying within the range of 30–90 nm. According to Xu,²⁹ the pore structure of the porous carbon monoliths produced by polymerization induced phase separation is defined by the ingredients of the pre-mixed solutions and the thermal history of the polymer system. Furthermore, the maximum distribution of pore size in green pieces decreases with increasing solid content of the slurries.

Fig. 7 Pore size distribution of SiC/porous carbon bodies with different solid contents.

Bulk density and the apparent porosity of the SiC/porous carbon bodies at different solid loadings was obtained using mercury intrusion porosimetry, and the results are shown in Fig. 8. As the solid loading increased from 55 wt% to 70 wt%, the bulk density of the carbonized bodies increased from 1.30 g cm⁻³ to 1.63 g cm⁻³, whereas the apparent porosity decreased from 32.0% to 25.2%. Higher solid loadings could lead to higher green density and lower apparent porosity, mainly because of the packing behavior of the bimodal SiC particles and the extent of the porous polymer networks.

Fig. 8 Bulk density and apparent porosity of the green bodies as a function of solid loading.

3.3 Characterization of the sintered pieces

After machining in the green state followed by carbonization and reaction sintering, reaction bonded silicon carbide (RBSC) ceramic parts of complex shape can be produced. From the optical microscope image shown in Fig. 9, SiC/porous carbon bodies with different solid loading were all completely infiltrated by liquid silicon under vacuum and showed a dense polished surface, indicating that the excess pores were fully filled by residual Si following siliconization. Relative homogeneous distribution can be observed, and neither residual carbon (black colored regions) nor large areas of silicon were present. As a result, the fraction of residual Si was determined by the porosity of the green body. In the image, the areas in gray represent the original α-SiC and newly formed β-SiC particles, and those in white indicate silicon – in reality, the pores between the particles were filled with silicon.

Fig. 9 Morphology of the polished surface of RBSC at a solid loading of (a) 55 wt%, (b) 60 wt%, (c) 65 wt%, and (d) 70 wt%.

Phase composition was assessed using the XRD patterns. The XRD analysis was conducted on the polished surface, and Fig. 10 shows the XRD patterns of materials at different solid loadings. The sintered ceramics are mainly comprised of three phases: α-SiC, β-SiC and Si itself, whereas the carbon phase was not detected. The specimens of higher solid loading contained a lower fraction of Si phase, shown by the strong peak intensity of Si of the four specimens. Such a variation in silicon content is compatible with the bulk density change in the carbonized samples in Fig. 9. In this case the low content of Si is explained by the reduced porosity of the SiC/porous carbon bodies. The density of SiC and Si are 3.21 and 2.34 g cm⁻³, respectively. Because of the high densification of RBSC, a variation in the Si fraction in the composite implies a corresponding variation in bulk density. The bulk density as a function of solid loading is listed in Table 1, and it is seen that the density increases with solid loading within the range 55–70 wt%.

Fig. 10 Composition of the SiC ceramic.

Table 1 Properties of the sintered bodies

Sample, including solid loading	Bulk density (g cm^−3)	Silicon content (%)	Flexural strength (MPa)	Fracture toughness (MPam^1/2)
SC-55	2.94	31.0	244 ± 15	3.08 ± 0.15
SC-60	3.00	24.1	262 ± 18	3.37 ± 0.19
SC-65	3.04	19.5	300 ± 20	3.87 ± 0.19
SC-70	3.06	17.2	358 ± 16	4.22 ± 0.13

---

The fracture surface of the sintered SiC ceramics is shown in Fig. 11. Based on the SEM micrograph, the main fracture mode was intergranular fracture of SiC, coupled with brittle fracture of the residual silicon. The flexural strength and fracture toughness of the SiC ceramics prepared at different solid loadings are summarized in Table 1. The flexural strength and fracture toughness each increased dramatically with solid loading, and when the solid loading reached 70 wt% the sintered body exhibited the highest bending strength and fracture toughness, 358 ± 16 MPa and 4.22 ± 0.13 MPam^1/2, respectively.

Fig. 11 Morphology of the fractural surface of the SiC ceramic at various solid loadings: (a) 55 wt%, (b) 60 wt%, (c) 65 wt%, and (d) 70 wt%.

4. Conclusions

The use of PF combined with FA was developed in order to consolidate suspensions in the manufacture of silicon carbide ceramics. The experimental data leads to the following summary:

1. Slurries with high solid loading and low viscosity were prepared using PEG 400 as dispersant and by optimizing the particle size distribution.

2. When the solid loading was below 75 wt% the bending strength of the green body exceeded 10 MPa, because of the fact that the SiC particles were held in a three-dimensional polymer network. This bending strength would be satisfactory for machining.

3. With increasing solid loading, the bulk density, bending strength and fracture toughness increased. Sintered bodies with solid loading of 70 wt% showed the highest bending strength and fracture toughness, 358 ± 16 MPa and 4.22 ± 0.13 MPam^1/2, respectively.

4. When the solid loading was increased to 75 wt%, gases in the slurry which would normally escape were trapped because of the high viscosity and could be seen in the cured bodies, resulting in a decrease in green strength.

Acknowledgements

This study was supported by the Major State Basic Research Development Program of China (2014cb046505) and the Research Fund for the Doctoral Program of Higher Education of China (20132302120022).

References

1. A. C. Young, O. O. Omatete, M. A. Janney and P. A. Menchhofer, Gelcasting of alumina, J. Am. Ceram. Soc., 1991, 74(3), 612–618.
2. J. Yang, J. Yu and Y. Huang, Recent developments in gelcasting of ceramics, J. Eur. Ceram. Soc., 2001, 31, 2569–2591.
3. R. He, P. Hu, X. Zhang and C. Liu, Gelcasting of complex-shaped ZrB2-SiC ultra high temperature ceramic components, Mater. Sci. Eng., A, 2012, 556, 494–499.
4. C. Zhang, X. Huang, Y. Yin, F. Xia, J. Dai and Z. Zhu, Preparation of boron carbide–aluminum composites by non-aqueous gelcasting, Ceram. Int., 2009, 35, 2255–2259.
5. P. Sepulveda and J. G. P. Binner, Evalution of the in situ polymerization kinetics for the gelcasting of ceramic foams, Chem. Mater., 2001, 13, 3882–3887.
6. F. Li, H. Chen, R. Wu and B. Sun, Effect of polyethylene glycol on the surface exfoliation of SiC green bodies prepared by gelcasting, Mater. Sci. Eng., A, 2004, 368, 255–259.
7. B. Chen, D. Jiang, J. Zhang, M. Dong and Q. Lin, Gel-casting of α-TCP using epoxy resin as a gelling agent, J. Eur. Ceram. Soc., 2008, 28, 2889–2894.
8. Y. Jia, Y. Kanno and Z. P. Xie, Fabrication of alumina green body through gelcasting process using alginate, Mater. Lett., 2003, 57, 2530–2534.
9. W. Wan, J. Yang, J. Zeng, L. Yao and T. Qiu, Aqueous gelcasting of silica ceramics using DMAA, Ceram. Int., 2014, 40, 1257–1262.
10. M. Potoczek, Hydroxylapatite foams produced by gelcasting using agarose, Mater. Lett., 2008, 62, 1055–1057.
11. A. J. Millan, M. I. Nieto and R. Moreno, Aqueous gel-forming of silicon nitride using carrageenans, J. Am. Ceram. Soc., 2001, 84, 62–64.
12. Y. Zhang, Z. Yuan and Y. Zhou, Gelcasting of silicon carbide ceramics using phenolic resin and furfuryl alcohol as the gel former, Ceram. Int., 2014, 40, 7873–7878.
13. C. Zhang, T. Qiu, J. Yang and J. Guo, The effect of solid volume fraction on properties of ZTA composites by gelcasting using DMAA system, Mater. Sci. Eng., A, 2012, 539, 243–249.
14. W. Wan, J. Yang, J. Zeng, L. Yao and T. Qiu, Effect of solid loading on gelcasting of silica ceramics using DMAA, Ceram. Int., 2014, 40, 1735–1740.
15. R. S. Kumar, D. Sivakumar, K. Venkateswarlu and A. S. Gandhi, Mechanical behavior of molybdenum disilicide reinforced silicon carbide composite, Scr. Mater., 2011, 65, 838–841.
16. P. Y. Dai, Y. Z. Wang, G. L. Liu, B. Wang, Y. G. Shi and J. F. Yang, et al. Fabrication of highly dense pure SiC ceramics via the HTPVT method, Acta Mater., 2011, 59, 6257–6260.
17. A. A. Zaman and B. M. Moudgil, Rheology of bidisperse aqueous silica suspensions: a new scaling method for the bidisperse viscosity, J. Rheol., 1998, 42, 21–39.
18. J. M. F. Ferreira and H. M. M. Diz, Pressure slip casting of bimodal silicon carbide powder suspension, Ceram. Int., 1999, 25, 491–495.
19. S. Aroati, M. Cafri, H. Dilman, M. P. Dariel and N. Frage, Preparation of reaction bonded silicon carbide (RBSC) using boron carbide as an alternative source of carbon, J. Eur. Ceram. Soc., 2011, 31, 841–845.
20. B. Dames, B. R. Morrison and N. Willenbacher, An empirical model predicting the viscosity of highly concentrated, bimodal dispersions with colloidal interactions, Rheol. Acta, 2001, 40, 434–440.
21. C. Logos and Q. D. Nguyen, Effect of particle size on the flow properties of a south Australian coal–water slurry, Powder Technol., 1996, 88, 55–58.
22. Q. Huang, W. Li, M. Gu, P. Chen, H. Zhang and Y. Qin, Effect of blend ratio on rheological properties of aqueous SiC suspensions, J. Eur. Ceram. Soc., 2004, 24, 2157–2161.
23. L. Bergstrom, Shear thinning and shear thickening of concentrated ceramic suspensions, Colloids Surf., A, 1998, 133, 151–155.
24. G. Ankur, K. Ankit, N. Balashanmugam, M. V. Arun and M. Giridhar, Optimization of rheological properties of photopolymerizable alumina suspensions for ceramic microstereolithography, Ceram. Int., 2014, 40, 3655–3665.
25. Y. Li, J. Lin, J. Gao, G. Qiao and H. Wang, Fabrication of reaction-bonded SiC ceramics by slip casting of SiC-C suspension, Mater. Sci. Eng., A, 2008, 483–484, 676–678.
26. Z. H. Luo, D. L. Jiang, J. X. Zhang, Q. L. Lin, Z. M. Chen and Z. R. Huang, Preparation of reaction-bonded silicon carbide with well controlled structure by tape casting method, Ceram. Int., 2012, 38, 2125–2128.
27. X. Wang, Z. P. Xie, Y. Huang and Y. B. Cheng, Gelcasting of silicon carbide based on gelation of sodium alginate, Ceram. Int., 2002, 28, 865–871.
28. Z. Yuan, Y. Zhang, Y. Zhou and J. Han, Preparation and characterization of porous carbons obtained from mixtures of furfuryl alcohol and phenol-formaldehyde resin, Mater. Chem. Phys., 2014, 143, 707–712.
29. S. Xu, J. Li, G. Qiao, H. Wang and T. Lu, Pore structure control of mesoporous carbon monoliths derived from mixtures of phenol resin and ethylene glycol, Carbon, 2009, 47(8), 2101–2111.

---