Influence of the Compacts Homogeneity on the Incidence of Cracks during Thermal Debinding in Ceramic Injection Molding

Transcription

1 264 J. Mater. Sci. Technol., Vol.25 No.2, 2009 Influence of the Compacts Homogeneity on the Incidence of Cracks during Thermal Debinding in Ceramic Injection Molding Xianfeng Yang, Zhipeng Xie and Yong Huang State Key Laboratory of New Ceramics and Fine Processing, Department of Materials Science and Engineering, Tsinghua University, Beijing , China [Manuscript received March 26, 2008, in revised form July 22, 2008] During thermal debinding in ceramic injection molding, the inhomogeneity of green body is a key origin of cracks. In this study, the impact of low molecular weight binders on the homogeneity of the green body was investigated. Incidence of cracks during thermal debinding indicated that the volume ratio of wax to stearic acid should be out of high viscosity and incompletely wetting region. In these two formulation regions, typical inhomogeneous microstructures were observed. By mercury intrusion method, it was shown that pore size distribution of the debinded compacts was determined by thermal degradation of low molecular weight binders. A particle-rich region model was established to predict the nucleation of cracks caused by solid loading fluctuation. The criterion of cracks nucleation was that local capillary force from solid loading fluctuation was larger than the suction force from the surroundings. KEY WORDS: Ceramic injection molding; Cracks; Debinding; Homogeneity 1. Introduction Ceramic injection molding is a near net shaping technology for producing complex shaped ceramic parts. Different with dry-pressing or cold isostatic pressing process, higher content of multi-component binders are applied in ceramic injection molding to ensure the flowability of the ceramic powder-binder mixtures. So the pores in the green bodies are filled with organic binder, called closed pore structure [1,2]. In initial stage of thermal debinding, the connected porosity has not been formed. The gaseous products from low molecular binder evaporation or monomer from polymer decomposition can not escape out of the green body quickly [3,4]. Therefore the imbalance in the competition between degradation of the binder and the outward diffusion of the gaseous products may result in cracking or bloating [5]. There are lots of literature focused on exploring the mass transport phenomena and optimizing the debinding schedule to reduce these defects. In early studies, shrinking undegraded core model and distributed porosity model were applied to investigate the influence of heating rate, product size or atmosphere on cracking [6 9]. However these two extreme models can not reflect the real situation. Lam et al. [10] and Ying et al. [11] developed an integrated model valid for tow dimensional analysis to simulate multi-phase fluid movement, heat transfer, polymer pyrolysis, particle packing, strain, stress and their interactions. Although these approaches have established the basic relationship between the formation of defects and debinding parameters, they share the same assumptions that the spatial distribution of ceramic particles, organic binder and pores is uniform, which does not reflect the physical reality of the debinding process. It is generally accepted that the aggregates Corresponding author. Prof., Ph.D.; Tel.: ; address: xzp@mail.tsinghua.edu.cn (Z.P. Xie). of the ceramic particles, voids, inadequate melt of the polymer, segregation between binder and particles always exist [12]. Even if the green body can be considered as uniform, capillary driven viscous flow of liquid binder and inter-region deference of debinding rate will change the initial distribution during debinding process [13,14]. In essence, cracking is the results of imbalance between inner stress and bonding strength among ceramic particles, based on the local distribution of melt binder, solid particle and gaseous products [15]. So it is necessary to better understand the relationship between composition distribution in the green body and cracking. The degree of the uniformity of the composition distribution is called compact homogeneity in this study. However we do not know well about the influencing factor of the homogeneity of compact and few studies have attempted to explore the mechanism of cracking experimentally or theoretically. It is mainly because there is no standard evaluation criterion for the homogeneity of the particle filled feedstock, especially high solid loading system. Traditional thermal weight loss method turns to be useless due to the limited precision of the balance measurement. On the other hand, it is almost impossible to observe the structure evolution during thermal debinding by electronic or optical microscope when the compacts are composed of over 50 vol. pct of ceramic particles of several hundred nanometers. In the present work, the effect of binder composition on the compact homogeneity, incidence of cracking during thermal debinding and structure evolution was investigated. The nucleation of cracks caused by solid loading fluctuation was predicted. 2. Experimental Commercial zirconia powder with average particle size (d 50 ) of 0.16 µm and BET specific area of 8.3 m 2 /g (YSZ-F-DM-3.0, Farmeiya Advanced Mate-

2 J. Mater. Sci. Technol., Vol.25 No.2, the low molecular weight components (wax and SA) from boiling. Ten samples were taken out from the debinding furnace each time at 250 C, 350 C and the final debinding temperature 450 C for inspecting the cracks. The green bodies and debinded samples were observed by scanning electron microscopy (SEM, LEO 1530, Germany). Thermalgravimetric analysis (TGA) was performed (Dupont 2100, USA) under nitrogen atmosphere with a heating rate of 10 C/min. The pore size distribution of the debinded bodies was evaluated by mercury intrusion method (Autopore 9500, USA). Fig. 1 SEM micrograph of the zirconia powder Incidence of cracks / % Cannot be injected High viscosity Safely debinding Incompletely wetting V(wax) / V(SA) Fig. 2 Influence of v(wax)/v(sa) on the incidence of cracks rials Co., Ltd., China) was used. The SEM (scanning electron microscopy) micrograph of the zirconia powder is shown in Fig. 1. Zirconia powder at 89 wt pct or 58 vol. pct was rolling mixed with polypropylene (PP, K8303, Beijing Yanshan Petrochemical Co., Ltd., China), paraffin wax (Shenyang Paraffin-wax Chemical Co., Ltd., China) and stearic acid (SA, Shantou Xilong Chemical Factory, China) at 170 C for 30 min. The volume fraction of PP was 0.2. The total volume fraction of wax and SA was 0.8. In experiments, the volume fraction ratio of wax to SA (v(wax)/v(sa)) was set at 0.3/0.5, 0.4/0.4, 0.5/0.3, 0.6/0.2, 0.7/0.1 and 0.8/0, respectively. Rectangular bars of 4.5 mm 6.0 mm 42.0 mm (eight samples each mold) were injection molded on the injection molding machine (JPH30, Hengli Plastic Machinery CO., Ltd, China). The injection temperature was C from inlet to nozzle. The samples were immersed in the powder bed in the cubic and debinded in the Muffle furnace. Based on the previous experiments, the heating rate for thermal debinding, 4.5 C/h, was low enough to prevent 3. Results and Discussion 3.1 Incidence of cracks and homogeneity of the green body It was found that cracks were always formed from room temperature to 250 C and the incidence of cracks with varying v(wax)/v(sa) is shown in Fig. 2. For the given powder and solid loading, when v(wax)/v(sa) is under 0.2/0.6, the viscosity of the feedstock is so high that the mold cavity can not be filled successfully. When v(wax)/v(sa) increases from 0.2/0.6 to 0.8/0, the organic binder system can be classified into high viscosity, safely debinding and incompletely wetting region, respectively (Fig. 2). In high viscosity region, the content of the wax is relatively low and the flowability of the feedstock is poor. The typical features of the microstructure of the feedstock in this region are the rod like high molecular binder and voids on the fracture surface shown in Fig. 3(a). In this region, cracking can not be avoided in thermal debinding; especially at left-hand, cracks are formed in all the samples. When the volume ratio of wax to SA reaches 0.5/0.3, the samples can be debinded without cracks. However, when the volume ratio of wax to SA is over 0.6/0.2, cracking during debinding occurs again. In this region, the content of SA is not enough to wet the ceramic particles. Ceramic particles or aggregates without organic binder are found on the fracture surface (Fig. 3(b)). Compared with the homogeneous green body for safely debinding (Fig. 3(c)), it can be inferred that cracking during thermal debinding is related directly with the inhomogeneity of the green body. During thermal debinding, the organic binder is substituted by the pores gradually, so development of pore size distribution of the compact is considered as the reflection of structure evolution in binder removal. Nonuniform distribution of binder and ceramic particles will evolve into the nonuniform porosity in the Fig. 3 Microstructure of the fracture surface of samples from: (a) high viscosity region, (b) incompletely wetting region, (c) safely debinding region

3 266 J. Mater. Sci. Technol., Vol.25 No.2, 2009 Intrusion volume / (ml/g) C 350 C 450 C 450 C (2 h) Pore diameter / nm Fig. 4 Pore size distribution at each stage during thermal dedinding (v(wax)/v(sa)=0.6/0.2) Fraction of binder remaining / % SA Wax PP Temp. / o C Fig. 5 Weight loss of the binder components compact. It has been widely accepted that nonuniform shrinkage from porosity difference is the main underlying mechanism controlling the formation of cracks corresponding to a wider pore size distribution [16]. In this study, development of pore size distribution was measured to set the relationship between green body structure and incidence of cracks. 3.2 Pore size distribution of the debinded compacts Figure 4 presents the pore size distribution of the compact at each debinding step. It indicates that pore structure of the debinded compacts turns to be mature from C because pore size distribution changes little from 350 to 450 C. While the thermal gravimetric curves of the binder component in Fig. 5 shows that starting degradation temperature of SA is about 150 C, wax: 200 C and PP: 400 C. If the influence of interaction between binder components and ceramic particle surface on thermal degradation is neglected, it can be inferred that the degradation of polymer has little influence on the final pore structure of the compact. Figure 6 presents the pore size distribution curve with the variation of v(wax)/v(sa) when the content of polymer in the binder was fixed at 0.2. It shows that v(wax)/v(sa) has little effect on the pore size distribution of the final debinded compacts. The most frequent pore size is in the narrow region of about nm. In this study, the expected broader pore size distribution caused by inhomogeneous structure was not shown by mercury intuition method [17]. This may Intrusion volume / (ml/g) / / / / Pore diameter / nm Fig. 6 Pore size distribution at different v(wax)/v(sa) be from that small number of voids or cracks can not be characterized by the variation of mercury intrusion volume. Despite this, it is confirmed that the binders of low molecular weight determine the structure evolution in thermal debinding. So optimizing the constitution of low molecular-weight binders and debinding schedule in low temperature range is crucial to getting uniform microstructure to reduce the formation of cracks [18]. In quantitative study and actual processing control, it is more interesting to decide the acceptable inhomogeneous level in green body, which will not lead to cracks in debinding. However, the physical and chemical processes during thermal debinding are complex. Several mass transport mechanisms of viscous flow, vapor diffusion and vapor convection coexist. The actual liquid-vapor interface is infinitely coarse during the formation of connected porosity, which was similar to the process of drying [19]. Additionally, the positions of the ceramic particles and distance among them are not stable under the action of capillary force. So it is necessary to do some essential simplifications to model this physical process. In this study, viscous flow of binder is considered as the controlling mass transport mechanism to predict the nucleation of cracks caused by the inhomogeneity of the compacts. 3.3 Particle-rich region model The ideally uniform ceramic particle-organic binder mixture can not be prepared easily because the specific surface area of ceramic powder is always very large and the interaction between particles is strong. Under high shear force during kneading, extreme state of particle or binder aggregates can be avoided. The actual state is always particle-rich region existing in the uniform surroundings, schematically shown in Fig. 7(a). It is assumed that there is a transition zone between them. In thermal debinding, there is a capillary suction pressure from liquid-vapor meniscuses at surface or neck area between two particles. The viscous flow of melt binder driven by this capillary suction force in the porous body is the main homogenization mechanism. However in the transition region shown in Fig. 7(a), there is a local opposite resistant pressure from solid loading fluctuation. For spherical particle packing, higher solid loading results in higher packing factor and smaller capillary diame-

4 J. Mater. Sci. Technol., Vol.25 No.2, Fig. 7 Particle-rich region model for cracking during thermal debinding: (a) particle-rich region in the homogeneious surrounding, (b) nucleation of cracks Fig. 8 (a) Solid loading fluctuation in the form of aggregate in the green body, (b) cracks around the aggregate in the debinded compact ter than the surroundings. According to the Young- Laplace equation, liquid in the porous compact tends to be transported from capillaries of larger diameter to the smaller. When the local resistant capillary force is larger than the homogenizing suction force, the binder in the transition region will be transported into the particle-rich region gradually. At the end, the binder at the interface disappears and the two regions are separated. Without the binder, the particle in and outside the particle-rich region can not move or rotate consistently. This state is considered as the nucleation of cracks as shown in Fig. 7(b). In this study, the solid loading fluctuation in the form of aggregates in green body and cracks around the dense aggregates in debinded compact were observed, as shown in Fig. 8. So the upper limitation of solid loading fluctuation is that local capillary force near the transition region is smaller than the suction force. This model is called particle-rich region model and deduced as follows. The capillary suction pressure P c from surrounding can be expressed as [20] : ( ψ ) 1 2 P c = σj(s l ) (1) K J(S l ) = 0.364{1 exp[ 40(1 S l )]} (1 S l ) (2) S l 0.09 where σ is the surface tension. The porosity ψ is the function of the solid loading φ and liquid saturation S l, given as: ψ = (1 φ)(1 S l ) (3) K is the permeability of porous media of liquid binder, which can be written as K = (1 φ)3 d 2 36kφ 4 (4) where k is Kozeny constant related with pore shape and tortuosity factor. In this calculation, k=5. So P c can be written as: ( 1 φ ) 1 2 P c = σf(s l ) = 6 kφ 2 K (1 φ)d σf(s l) (5) f(s l ) = (1 S l ) 1 2 J(Sl ) (6) f(s l ) is just the function of S l, so it is defined as binder saturation factor. The local capillary force from solid loading fluctuation in the transition region, P S, can be deduced from the Young-Laplace equation as [21] : P s = 2σ r, r = 1 φ 3φd (7) P s = 6σ ξ, ξ = φ r φ s (8) d 1 φ r 1 φ s where P s is capillary force, r is capillary radius. The equivalent capillary radius r can be calculated from solid loading φ and particle diameter d when the influence of particle-rich on the surroundings is neglected. φ r and φ s are solid loading of the particle-rich region and surroundings, respectively. φ s is assumed to be equal to φ. ξ is defined as inhomogeneous factor.

5 268 J. Mater. Sci. Technol., Vol.25 No.2, 2009 Critical inhomogeneous factor vol. fraction 54 vol. fraction 56 vol. fraction 58 vol. fraction 60 vol. fraction Liquid saturation fraction Fig. 9 Critical value of inhomogeneous factor From P s P c, the upper limit of inhomogeneous factor is written as: ξ φ2 kf(sl ) (9) 1 φ The utility of Eq. (9) is that it establishes the quantitative relationship between cracks nucleation and solid loading, compact structure and residual binder fraction. These three parameters are corresponding to three items of the right hand of the equation. On the assumption of liquid binder flow in this model, a given porosity in the compacts is desired. So the equation is only valid for pendular state or funicular state, marked by the dotted line and arrow in Fig. 9 [22]. The result shows that with increasing the solid loading, the predicted critical value of inhomogeneous factor becomes larger. So it is reasonable that higher solid loading is desired to reduce the appearance of cracks. Figure 9 also indicates that higher liquid saturation fraction leads to little critical value of inhomogeneous factor. It means that at the early stage of debinding, lower solid loading fluctuation in the compacts will result in the nucleation of cracks. This prediction is inline with that observable cracks, which are always formed in the lower temperature range. However, the drawback of this model is that the effect of the size of the inhomogeneous region on cracking was not involved. The nucleation of cracks does not mean the detected cracks in the final debinded or sintered parts. The small particle-rich region results in short original cracks and small nonuniform shrinkage, which can be eliminated in the final sintering stage. The relationship between the size of nonuniform region and cracking in thermal debinding will be discussed in our future publications. 4. Summary The relationship between compacts homogeneity and cracking in thermal debinding was investigated from experimental and theoretical perspectives. Binder composition plays a key role in the homogeneity of the green body. Inhomogeneous structure in the green body evolves to cracks in thermal debinding. Thermal degradation of low molecular weight binders determines the development of pore size distribution of the debinded compacts. In the particle-rich model, the inhomogeneous factor is defined and the critical value of cracks nucleation was predicted. The results show that higher solid loading was beneficial to reducing the cracking defections and much care should be taken in the low temperature debinding stage. Acknowledgements The work was financially supported by the National Nature Science Foundation of China (NSFC) under grant No and the National High Technology Research and Development Program of China ( 863 Program ) under grant No. 2007AA03Z522. REFERENCES [1 ] J.A. Lewis: Annu. Rev. Mater. Sci., 1997, 27, 147. [2 ] V.M. Kryachek: Powder Metall. Met. Ceram., 2004, 43(7-8), 336. [3 ] J.H. Song, J.R.G. Evans and M.J. Edirisinghe: J. Mater. Res., 2000, 15(2), 449. [4 ] R.M. German: Int. J. Powder Metall., 1987, 23(4), 237. [5 ] S.A. Matar, M.J. Edirisinghe and J.R.G. Evans: J. Am. Ceram. Soc., 1996, 79(39), 749. [6 ] Z. Shi, Z.X. Guo and J.H. Song: Acta Mater., 2002, 50(8), [7 ] W.J. Tseng and C.K. Hsu: Ceram. Int., 1999, 25(5), 461. [8 ] J.R.G. Evans: J. Eur. Ceram. Soc., 1997, 17(2-3), 161. [9 ] S.J. Lombardo: J. Am. Ceram. Soc., 2003, 86(12), [10] Y.C. Lam, S.J. Ying and S.C.M. Yu: Metall. Mater. Trans. A, 2000, 31(10), [11] S.J. Ying, Y.C. Lam and J.C. Chai: Comput. Mater. Sci., 2004, 30(3-4), 496. [12] R.Y. Wu and W.C.J. Wei: J. Am. Ceram., Soc., 2005, 88(7), [13] M. Trunec and J. Cihlár: J. Am. Ceram. Soc., 2001, 84(3), 675. [14] M.J. Edirisinghe: J. Mater. Sci. Lett., 1991, 10(22), [15] Z.C. Feng, B. He and S.J. Lombardo: J. Appl. Mech., 2002, 69(7), 497. [16] H.M. Shaw and M.J. Edirisinghe: J. Eur. Ceram. Soc., 1995, 15(2), 109. [17] S.W. Kim, H.W. Lee and H. Song: Ceram. Int., 1996, 22(1), 7. [18] D.M. Liu and W.J. Tseng: J. Mater. Sci. Lett., 1997, 16(6), 482. [19] G.W. Scherer: J. Am. Ceram. Soc., 1990, 73(1), 3. [20] M.J. Cima, J.A. Lewis and A.D. Devoe: J. Am. Ceram. Soc., 1989, 72(2), [21] S.J. Ying, Y.C. Lam and J.C. Chai: Model. Simul. Mater. Sci. Eng., 2004, 12(2), 311. [22] S.W. Kim, H.W. Lee and H. Song: Ceram. Int., 1999, 25(5), 671.