- Email: [email protected]
Amino-functionalization of colloidal alumina particles for enhancement of the infiltration behavior in a silica-based ceramic core
Ceramics International xx (xxxx) xxxx–xxxx
Contents lists available at ScienceDirect
Ceramics International journal homepage: www.elsevier.com/locate/ceramint
Amino-functionalization of colloidal alumina particles for enhancement of the infiltration behavior in a silica-based ceramic core Gye Seok Ana, Soo Wan Choia, Tae Gyun Kima, Jae Rok Shina, Yong-In Kima, ⁎ Sung-Churl Choia,⁎⁎, Young-Gil Jungb, a b
Division of Materials Science and Engineering, Hanyang University, 222 Wangsimni-ro, Seongdong-gu, Seoul 04763, Republic of Korea School of Materials Science and Engineeing, Changwon National University, 20 Changwondaehak-ro, Changwon, Gyeongnam 51140, Republic of Korea
A R T I C L E I N F O
A BS T RAC T
Keywords: Amine-functionalization Precursor Colloidal alumina Ceramic core Infiltration
The surface characteristics of nano-sized alumina particles were modified to include amine functional groups to enhance the infiltration behavior in a silica-based ceramic core, inhibiting crystallization of the cristobalite phase in order to improve mechanical properties. In this study, polymer-based polyethylenimine (PEI) or silane-based (3-Aminopropyl)triethoxysilane (APTES) were used as precursors in amine group addition. The specimens were characterized by the C-H stretching vibration and N-H bending vibration or the symmetric stretching vibration of siloxane group (Si-O-Si) in the FT-IR spectra, depending on the nature of the molecular structure of the precursor molecule. The dispersion properties were also improved through electrostatic repulsion or steric hindrance. These modifications resulted in a PEI-grafted colloidal alumina particle that induced a more rapid infiltration process with greater efficiency than did the APTES-treated particle. The rapid infiltration of the ceramic core suppressed the formation of the cristobalite phase, improved the flexural strength from 3.2 to 10.2 MPa, and reduced the linear shrinkage rate from 1.91–1.05%.
1. Introduction Ceramic cores produced by injection molding have been extensively used in hollow gas turbine blades and vanes as part of the internal cooling path [1–5]. The ceramic cores are typically based on fused silica (SiO2) and zircon (ZrSiO4) materials, which are able to endure thermal stress induced by the high temperature investment casting process. These substances are selected due to their thermal shock resistance and chemical inertness against molten metal [6,7]. In order to endure these extreme conditions, the ceramic core must possess appropriate mechanical properties. High-temperature sintering is essential in the production of ceramic cores, but the fused silica is also converted to the phase at temperatures greater than 1300 °C. The volume reduction that results from the β to α phase transformation of cristobalite is the main cause of the shrinkage of the ceramic cores. Additionally, the microcrack produced in this phase transformation reduces the flexural strength. At temperatures near 350 °C, when the β phase of cristobalite is present, the flexural strength increased; it is decreased at room temperature. The changes in these mechanical properties are caused by the microcrack formed in the phase transformation, as described above [8–11]. A way to prevent the crystallization
⁎
of the fused silica is needed in order to suppress the degradation of the silica-based ceramic core. In our previous work, we introduced a method to prevent the crystallization of fused silica via infiltration with a colloidal alumina particle. The crystallization of cristobalite was limited due to the mullite phase formed by the reaction between the surface of fused silica and the absorbed alumina particle. The obtained infiltrated fused silica with 2.54% alumina content had a relatively high physical property. Even with this improvement to physical properties of the ceramic core, the efficiency is to still insufficient to apply to the production process due to relatively long infiltration time (150 min) [12]. In this study, the surface of alumina particles was modified in order to enhance the performance of the infiltration process. The silica surface is negatively charged due to the large amount of hydroxyl groups (-OH); therefore, it is advantageous to introduce a positive charge to improve the adsorption of the alumina particles through electrostatic attraction. The positively charged functional group most often utilized is the amine group. Polymer-based polyethylenimine or silane-based (3-Aminopropyl)triethoxysilane were used as precursors in the modification of the alumina particle surface. Infiltration effi-
Corresponding author. Co-corresponding author. E-mail addresses: [email protected] (S.-C. Choi), [email protected] (Y.-G. Jung).
⁎⁎
http://dx.doi.org/10.1016/j.ceramint.2016.09.127 Received 8 September 2016; Received in revised form 13 September 2016; Accepted 19 September 2016 Available online xxxx 0272-8842/ © 2016 Elsevier Ltd and Techna Group S.r.l. All rights reserved.
Please cite this article as: An, G.S, Ceramics International (2016), http://dx.doi.org/10.1016/j.ceramint.2016.09.127
Ceramics International xx (xxxx) xxxx–xxxx
G.S. An et al.
Fig. 1. High resolution SEM images of (a) as-prepared alumina, (b) PEI grafted alumina, and (c) APTES treated alumina.
distributions were measured using dynamic light scattering (DLS) (Zetasizer Nano ZSP, Malvern, United States) in order to estimate the dispersion behavior for different functionalizing methods. The weight loss of infiltrated specimens was measured using a thermogravity/differential thermal analyzer (TG/DTA, SDT 2960 Simultaneous, TA instruments, United States) and particle images were obtained by field emission scanning electron microscopy (FESEM, S-4800, Hitachi, Japan) and EDX mapping (Link Pentafet, Oxford Instruments, United Kingdom). The crystal structure of sintered specimens was determined using an X-ray diffractometer (XRD, D/Max-220, Rigaku, Japan) under Cu Kα radiation (λ=1.54178 Å ). The flexural strength and shrinkage rate were estimated based on 3-point bending test with a universal testing machine with a span size of 80 mm and crosshead speed of 1.0 mm/min (UTM, H10SK, Hounsfield, United Kingdom).
ciency, the mechanical properties, and the crystal structure were improved through this modified preparation of colloidal alumina particles. 2. Experimental process Commercially available nano-sized aluminum oxide (Al2O3, APS 40–50 nm, 99.5%, Alfa-Aesar, United States) was surface-modified through hyper-branched polyethylenimine (PEI, Mw 60,000, SigmaAldrich, United States), and (3-Aminopropyl)triethoxysilane (APTES, 99%, Sigma-Aldrich, United States). Alumina powder (5g) was suspended in 500 ml of deionized water and stirred at 400 RPM at room temperature for 1 h. 3 wt% of PEI in deionized water was added to the well-dispersed alumina solution and mixed at 80 °C for 3 h. The resulting colloidal particles were centrifuged at 10,000 RPM for 1 h. The refined particles were rinsed with deionized water several times and collected. To modify the surface by silane grafting, alumina powder (5 g) in 500 ml of deionized water was mixed with 100 ml of 1N ammonia solution at 250 RPM at 80 °C for 3 h. The sample was then cooled by the addition of 200 ml of deionized water and mixed with 150 ml of ethanol at 250 RPM at 60 °C for 12 h. The silane precursor solution, composed of 2 ml of APTES, 1 ml of ethanol, and 1 N of ammonia solution, was added 30 min into the procedure. After heating was complete, this product was washed as previously described for PEI. The silica-based matrix of the ceramic core was prepared by ceramic injection molding. 63 wt% of fused silica (amorphous SiO2, 325 mesh, IMERYS, United States) was used as the base material and was modified by addition of 21 wt% of zircon (ZrSiO4, 1 µm, Cenotec, Korea). A thermoplastic binder (15 wt%) composed of paraffin wax (Nippon-seiro, Japan), microcrystalline wax (Nippon-Seiro, Japan), stearic acid (C19H36O2, Samchun Pure Chemical, Korea) and oleic acid (C19H34O2, 97%, Samchun Pure Chemical, Korea) was used as feedstock for injection molding. The feedstocks were injection molded using a C-frame ceramic injection molding machine (CTM-CI-CF-35100HT, Cleveland Tools and Machines, United States) with 6 mm x 8 mm x 90 mm green bodies in accordance with ASTM C 1161-13. After calcination of the thermoplastic binders at 0.2 K/min, the specimens were pre-sintered at 1100 °C for 2 h with a heating rate of 5 K/ min using a box furnace (UAF-15–27-LHE, Lenton, United Kingdom). Pre-sintered specimens were infiltrated by 20% supensions of the prepared surface-modified colloidal alumina in deionized water under vacuum. The length of reaction was varied to achieve different levels of the alumina content. Resulting specimens were dried at 110 °C for 12 h to evaporate residual solvent and sintered at 1300 °C for 2 h. The surface functional groups of the modified alumina particles were determined by Fourier transform infrared spectroscopy (FT-IR, IRAffinity-1S, Shimadzu, Japan). The zeta-potential value at various pH was measured using 0.1 M HNO3 or 0.1 M NaOH in deionized water by electrokinetic sonic amplitude measurements (ESA, Zeta Finder, Matec Applied Sciences, United States). The particle size
3. Results and discussion The morphology of as-prepared and each surface modified alumina particles were examined by SEM in Fig. 1. The as-prepared alumina particles have slightly keen edged quasi-spherical shape with about 40 nm of average diameter approximately. However, the morphology of PEI grafted alumina particles exhibit rugged and distorted surface compared with the raw alumina particles due to grafting of polymer, which has randomly arranged polymeric structure. On the other hand, the morphology of APTES treated alumina particles are similar with the as-prepared one. It is possibly considered that silane layer formed around the surface of alumina particles by condensation of silanol group is homogeneous and very thin which cannot identify with naked eye, thus the morphology of both as-prepared and APTES treated alumina shows no significant difference.. The identity of the surface functional groups of the as-prepared alumina particles and PEI or APTES surface-modified particles through PEI or APTES was verified from the FT-IR spectra (Fig. 2). Absorption peaks for the Al-O stretching vibration were observed at 575 and 787 cm−1 and a broad absorption band of the surface hydroxyl group (Al-OH) in the range of 3700–3000 cm−1 due to the alumina base material [13,14]. Additionally, the N-H bending vibration peak at 1630 cm−1 from the functionalized amine group and the absorption bands between 3000 and 2700 cm−1 due to the C-H stretching vibration of alkyl chains of PEI and APTES were observed [15,16]. For the APETS-grafted specimen, the symmetric stretching vibration peaks of siloxane group (Si-O-Si) were observed at 1080 and 1120 cm−1[16]. The broad absorption band between 3700– 3000 cm−1 in both the surface modified alumina than as-prepared alumina particle is thought to be due to overlap of the –OH and N-H stretching vibrations [17].. In Fig. 3, the zeta potential values at each pH were used to evaluate the level of surface charge. The fused silica has a negative zeta potential value (between −3 to −45 mV) between pH 2 and 11. In this same range, the zeta potential of as-prepared alumina particle was 5 to 2
Ceramics International xx (xxxx) xxxx–xxxx
G.S. An et al.
Fig. 4. Particle size and distribution of the as-prepared alumina, PEI grafted alumina, and APTES treated alumina. Fig. 2. FT-IR spectra of the as-prepared alumina, PEI grafted alumina, and APTES treated alumina.
APTES-modified specimens ( Fig. 4). The advanced surface modification led to an improvement in the dispersion behavior of colloidal alumina particles. This enhancement of dispersion performance, more pronounced in the PEI-grafted particles, was evaluated on the basis of electrostatic repulsion and steric hindrance as well [18]. The large positive charge density of the free surface amine groups bonded to alkyl chain repels the alumina particles because of the high positive zeta potential value. In contrast, the free amine group of APTES exists in low numbers at the surface terminal site of siloxane bond. Due to these inherent structural differences, the two particle types have different surface charges, impacting the dispersion. The molar structure of each precursor affects dispersion behavior in additional ways. For the PEI polymer, the long and branched alkyl chains induce significant steric interference between the PEI-grafted alumina particles. The alkyl chains in the APTES-grafted alumina lead do not lead to significant steric effects because the alkyl chain of the APTES grafted alumina is the much smaller propyl group, bonded to the terminal amine group [19]. Consequently, the dispersion stability of the PEI-grafted alumina is outstanding because of the combination of the strong electrostatic repulsion and limited steric hindrance. Fig. 5 shows the alumina content and apparent porosity of the fused-silica matrix of prepared specimens with different infiltration times. Although the infiltration efficiency varies, the alumina content was increased with an increase in infiltrating time. The PEI-grafted colloidal alumina particle had the most rapid saturation of alumina content and decreasing apparent porosity, which were achieved at 45 min. The APTES-treated colloidal alumina particle achieved saturation in 75 min. However, the raw colloidal alumina particles did not reach the saturation point until 130 min of infiltrating time. Both surface modified specimens had 2.54% alumina content at the saturation point, with a continual subsequent weak increase past this point. At the alumina content achieved in the PEI-grafted specimen after 45 min of infiltration time, the APTES-treated specimen was only 1.41% (60% efficiency); the raw colloidal alumina achieved 72% efficiency.. Fig. 6 illustrates the differences between the raw prepared fused silica and PEI-grafted colloidal alumina infiltrated specimen for 45 min through high resolution SEM and EDX measurements. Through a comparison of those two microstructure images, it is confirmed that the fine particles filled the pores with fused silica substrate in the infiltration process. This is clearly represented in the EDX mapping of Si and Al. The EDX data indicates that the alumina coagulated homogeneously in silica-based porous specimen via infiltration, excluding the fractured large fused silica area (indicated in black).. The infiltration efficiency of the PEI-grafted alumina was outstanding compared with the APTES-treated specimen. The dispersion
Fig. 3. Zeta-potential variation of prepared alumina specimens and fused silica at different pH values.
−14 mV, with an isoelectric point (IEP) at pH 5.6. The difference in zeta potential between these two raw materials is the greatest at pH 3.92. However, this difference is still notably small due to the electrostatic attraction effect.. The alternations of zeta-potential values were expanded to the large number to positive charged direction, and the IEP was shifted to a higher pH number for both amine-functionalized particles. The zetapotential values at neutral pH were found to be 34 mV for the PEIgrafted specimen (IEP of pH 10.4) and 21 mV for APTES-treated specimen (IEP of pH 9.3). A pH of 6.8 (neutral) was found to have the maximum polarity difference between fused silica (−38 mV) and surface-modified colloidal alumina particles. The difference in the zetapotential value is thought to be due to the greater number of free surface amine groups in the PEI-grafted alumina is more than the APTES-treated alumina, leading to larger positive charge densities. Increase in zeta-potential value as a driving force of the dispersion mechanism was also observed represented in particle size distribution of prepared specimens. The average size of the raw alumina particles was 178 nm, larger than the 40–50 nm primary alumina particles. The inherent high surface energy of the small particles and nearby neutral surface charge induces aggregation the alumina particles. A broad monomodal distribution was observed. The PEI and APTES surface modified specimens were of smaller average size and with a narrower monomodal particle distribution. The average size of the PEI- and APTES-grafted alumina particles were 60 and 93 nm, respectively. The PEI-grafted alumina particles had a narrower distribution than the
3
Ceramics International xx (xxxx) xxxx–xxxx
G.S. An et al.
Fig. 5. Infiltrated alumina content and apparent porosity as a function of infiltration time. The filled and open symbols indicate the alumina content and apparent porosity, respectively.
Fig. 7. The X-ray diffraction patterns of the sintered specimens as a function of infiltration time with PEI-grafted colloidal alumina particles.
stability was enhanced via electrostatic repulsion and steric hindrance. The large positive surface charge of the PEI-grafted alumina improved the adsorption capability because of the strong opposition between the negatively charged fused silica particles and positively charged alumina particles. The easy movement of alumina particles onto the surface or pores of the fused silica enhances the infiltration process. The adsorption capability of the APTES-grafted alumina is comparatively weak because of the enhanced dispersity and because the Van der Waals force between the fused silica matrix and siloxane groups bonded to the surface of alumina is weaker than the effect of strong dispersion stabilization and positive surface charge of the PEI-grafted alumina. Therefore, the saturation point of the alumina content was reduced to 45 min in the PEI-grafted alumina. After 75 min, the APTES-treated
alumina achieved the same level of porosity reduction and alumina saturation. Fig. 7 depicts the X-ray diffraction patterns of the sintered specimens at 1300 °C according to the infiltrating time of the PEI-grafted colloidal alumina particle. Phase transformation is proposed due to the coexistence of zircon (JCPDS 6-266) and cristobalite (JCPDS 39-1425). The mullite phase (JCPDS 15-0776) peaks were revealed with infiltration times exceeding 45 min, and cristobalite peaks were not observed. A slight excess of infiltrated alumina (JCPDS 46-1212) was observed specimens with 60 min of infiltration.. The effect of alumina infiltration on mechanical properties are
Fig. 6. Microstructures and EDX mapping results of fused silica based ceramic cores with and without the infiltration of PEI grafted colloidal alumina: (a) non-infiltration, (b) 45 mininfiltration, (c) Al mapping after 45 min-infiltration, and (d) Si mapping after 45 min-infiltration.
4
Ceramics International xx (xxxx) xxxx–xxxx
G.S. An et al.
hibited the crystallization of fused silica via the formation of mullite phase. Acknowledgments This work was supported by the National Research Foundation of Korea (NRF) grant funded by the Korea government (2014R1A2A1A11050220), and the Power Generation and Electricity Delivery of the Korea Institute of Energy Technology Evaluation and Planning (KETEP) grant funded by the Ministry of Trade, Industry and Energy, Korea (20142020103400). References [1] I. Huseby, M. Borom, C. Greskovich, High temperature characterization of silicabase cores for superalloys, Am. Ceram. Soc. Bull. 58 (1979) 448–452. [2] A.A. Wereszczak, K. Breder, M.K. Ferber, T.P. Kirkland, E.A. Payzant, C.J. Rawn, E. Krug, C.L. Larocco, R.A. Pietras, M. Karakus, Dimensional changes and creep of silica core ceramics used in investment casting of superalloys, J. Mater. Sci. 37 (2002) 4235–4245. [3] C.-H. Chao, H.-Y. Lu, Optimal composition of zircon-fused silica ceramic cores for casting superalloys, J. Am. Ceram. Soc. 85 (2002) 773–779. [4] S. Pattnaik, D.B. Karunakar, P.K. Jha, Developments in investment casting processa review, J. Mater. Process. Technol. 212 (2012) 2332–2348. [5] M. Gromada, A. ´Swieca, M. Kostecki, A. Olszyna, R. Cygan, Ceramic cores for turbine blades via injection moulding, J. Mater. Process. Technol. 220 (2015) 107–112. [6] W.D. kingery, H.K. Bowen, D.R. Uhlmann, Introduction to Ceramics, John Wiley & Sons, New York, 1976. [7] L.L. Hench, S.H. Wang, The sol-gel glass transformation of silica, Phase Transit. 24–26 (1990) 785–834. [8] C.-H. Chao, H.-Y. Lu, Stress-induced β→α-cristobalite phase transformation in (Na2O+Al2O3)-codoped silica, Mater. Sci. Eng. A 328 (2002) 267–276. [9] A. Kazemi, M.A. Faghihi-Sani, H.R. Alizadeh, Investigation on cristobalite crystallization in silica-based ceramic cores for investment casting, J. Eur. Ceram. Soc. 33 (2013) 3397–3402. [10] A. Kazemi, M.A. Faghihi-Sani, M.J. Nayyeri, M. Mohammadi, M. Hajfathalian, Effect of zircon content on chemical and mechanical behavior of silica-based ceramic cores, Ceram. Int. 40 (2014) 1093–1098. [11] R.C. Breneman, J.W. Halloran, Effect of cristobalite on the strength of sintered fused silica above and below the cristobalite transformation, J. Am. Ceram. Soc. 98 (2015) 1611–1617. [12] Y.-H. Kim, J.G. Yeo, S.-C. Choi, Shrinkage and flexural strength improvement of silica-based composites for ceramic cores by colloidal alumina infiltration, Ceram. Int., 42, pp. 8878–8883. [13] J. Gangwar, B.K. Gupta, S.K. Tripathi, A.K. Srivastava, Phase dependent thermal and spectroscopic responses of Al2O3 nanostructures with different morphogenesis, Nanoscale 7 (2015) 13313–13344. [14] L. Wu, Z. Yin, Sulfonic acid functionalized nano γ-Al2O3 catalyzed per-O-acetylated of carbohydrates, Carbohydr. Res. 365 (2013) 14–19. [15] L. Feng, H. Zhang, P. Mao, Y. Wang, Y. Ge, Superhydrophobic alumina surface based on stearic acid modification, Appl. Surf. Sci. 257 (2011) 3959–3963. [16] A. Huang, N. Wang, J. Caro, Seeding-free synthesis of dense zeolite FAU membranes on 3-aminopropyltriethoxysilane-functionalized alumina supports, J. Membr. Sci. 389 (2012) 272–279. [17] A.S.M. Chong, X.S. Zhao, Functionalization of SBA-15 with APTES and characterization of functionalized materials, J. Phys. Chem. B 107 (2003) 12650–12657. [18] T. Xia, Y. Guan, M. Yang, W. Xiong, N. Wang, S. Zhao, C. Guo, Synthesis of polyethylenimine modified Fe3O4 nanoparticles with immobilized Cu2+ for highly efficient proteins adsorption, Colloids Surf. A 443 (2014) 552–559. [19] R.A. Bini, R.F.C. Marques, F.J. Santos, J.A. Chaker, M. Jafelicci Jr., Synthesis and functionalization of magnetite nanoparticles with different amino-functional alkoxysilanes, J. Magn. Magn. Mater. 324 (2012) 534–539.
Fig. 8. Flexural strength (σ3pt) and linear shrinkage (ΔL/L0) as a function of infiltration time with PEI-grafted colloidal alumina particles.
summarized in Fig. 8, showing the flexural strength and linear shrinkage against infiltrating time. As reference, the silica-based green body specimen had 3.2 MPa of flexural strength and 1.91% of shrinkage rate, respectively. These mechanical properties were improved to 10 MPa and 1.04% after up to 45 min of infiltration. Because of the drastic reduction of the rate of increase of alumina content after additional infiltrating time, the mechanical properties were also not further improved.. 4. Conclusions In this study, amine-functionalized alumina for enhanced infiltration was created by surface modification with uniquely structured PEI and APTES precursors in order to improve the mechanical properties of the fused silica-based ceramic core. From the observed C-H stretching and N-H bending or symmetric stretching of siloxane group (Si-O-Si) vibration peaks on the FT-IR spectra, the functionalization with the amine group was confirmed. As a result, the enlarged zeta potential value was 34 and 21 mV for the PEI- and APTES-grafted alumina at neutral pH, respectively. The average particle sizes were 60 and 93 nm with a narrower particle distribution for the PEI-grafted alumina. The dispersion stability and positive surface charge of the PEI-grafted alumina are greater than that of the APTES-grafted alumina. This accelerates the infiltration behavior, reducing the saturation time to 45 min for the PEI-grafted alumina, whereas 75 min was required for APTES-grafted alumina to achieve the same porosity reduction and alumina saturation. These results were explained by the fundamental structural differences of the PEI and APTES, which are more efficient for improving the dispersity at the PEI grafted alumina through a number of amine groups and longbranched alkyl chain induced by the polymeric structure of PEI. Increased infiltration of the pre-sintered ceramic core using the PEI grafted alumina increased the flexural strength to 10 MPa and in-
5
Contents lists available at ScienceDirect
Ceramics International journal homepage: www.elsevier.com/locate/ceramint
Amino-functionalization of colloidal alumina particles for enhancement of the infiltration behavior in a silica-based ceramic core Gye Seok Ana, Soo Wan Choia, Tae Gyun Kima, Jae Rok Shina, Yong-In Kima, ⁎ Sung-Churl Choia,⁎⁎, Young-Gil Jungb, a b
Division of Materials Science and Engineering, Hanyang University, 222 Wangsimni-ro, Seongdong-gu, Seoul 04763, Republic of Korea School of Materials Science and Engineeing, Changwon National University, 20 Changwondaehak-ro, Changwon, Gyeongnam 51140, Republic of Korea
A R T I C L E I N F O
A BS T RAC T
Keywords: Amine-functionalization Precursor Colloidal alumina Ceramic core Infiltration
The surface characteristics of nano-sized alumina particles were modified to include amine functional groups to enhance the infiltration behavior in a silica-based ceramic core, inhibiting crystallization of the cristobalite phase in order to improve mechanical properties. In this study, polymer-based polyethylenimine (PEI) or silane-based (3-Aminopropyl)triethoxysilane (APTES) were used as precursors in amine group addition. The specimens were characterized by the C-H stretching vibration and N-H bending vibration or the symmetric stretching vibration of siloxane group (Si-O-Si) in the FT-IR spectra, depending on the nature of the molecular structure of the precursor molecule. The dispersion properties were also improved through electrostatic repulsion or steric hindrance. These modifications resulted in a PEI-grafted colloidal alumina particle that induced a more rapid infiltration process with greater efficiency than did the APTES-treated particle. The rapid infiltration of the ceramic core suppressed the formation of the cristobalite phase, improved the flexural strength from 3.2 to 10.2 MPa, and reduced the linear shrinkage rate from 1.91–1.05%.
1. Introduction Ceramic cores produced by injection molding have been extensively used in hollow gas turbine blades and vanes as part of the internal cooling path [1–5]. The ceramic cores are typically based on fused silica (SiO2) and zircon (ZrSiO4) materials, which are able to endure thermal stress induced by the high temperature investment casting process. These substances are selected due to their thermal shock resistance and chemical inertness against molten metal [6,7]. In order to endure these extreme conditions, the ceramic core must possess appropriate mechanical properties. High-temperature sintering is essential in the production of ceramic cores, but the fused silica is also converted to the phase at temperatures greater than 1300 °C. The volume reduction that results from the β to α phase transformation of cristobalite is the main cause of the shrinkage of the ceramic cores. Additionally, the microcrack produced in this phase transformation reduces the flexural strength. At temperatures near 350 °C, when the β phase of cristobalite is present, the flexural strength increased; it is decreased at room temperature. The changes in these mechanical properties are caused by the microcrack formed in the phase transformation, as described above [8–11]. A way to prevent the crystallization
⁎
of the fused silica is needed in order to suppress the degradation of the silica-based ceramic core. In our previous work, we introduced a method to prevent the crystallization of fused silica via infiltration with a colloidal alumina particle. The crystallization of cristobalite was limited due to the mullite phase formed by the reaction between the surface of fused silica and the absorbed alumina particle. The obtained infiltrated fused silica with 2.54% alumina content had a relatively high physical property. Even with this improvement to physical properties of the ceramic core, the efficiency is to still insufficient to apply to the production process due to relatively long infiltration time (150 min) [12]. In this study, the surface of alumina particles was modified in order to enhance the performance of the infiltration process. The silica surface is negatively charged due to the large amount of hydroxyl groups (-OH); therefore, it is advantageous to introduce a positive charge to improve the adsorption of the alumina particles through electrostatic attraction. The positively charged functional group most often utilized is the amine group. Polymer-based polyethylenimine or silane-based (3-Aminopropyl)triethoxysilane were used as precursors in the modification of the alumina particle surface. Infiltration effi-
Corresponding author. Co-corresponding author. E-mail addresses: [email protected] (S.-C. Choi), [email protected] (Y.-G. Jung).
⁎⁎
http://dx.doi.org/10.1016/j.ceramint.2016.09.127 Received 8 September 2016; Received in revised form 13 September 2016; Accepted 19 September 2016 Available online xxxx 0272-8842/ © 2016 Elsevier Ltd and Techna Group S.r.l. All rights reserved.
Please cite this article as: An, G.S, Ceramics International (2016), http://dx.doi.org/10.1016/j.ceramint.2016.09.127
Ceramics International xx (xxxx) xxxx–xxxx
G.S. An et al.
Fig. 1. High resolution SEM images of (a) as-prepared alumina, (b) PEI grafted alumina, and (c) APTES treated alumina.
distributions were measured using dynamic light scattering (DLS) (Zetasizer Nano ZSP, Malvern, United States) in order to estimate the dispersion behavior for different functionalizing methods. The weight loss of infiltrated specimens was measured using a thermogravity/differential thermal analyzer (TG/DTA, SDT 2960 Simultaneous, TA instruments, United States) and particle images were obtained by field emission scanning electron microscopy (FESEM, S-4800, Hitachi, Japan) and EDX mapping (Link Pentafet, Oxford Instruments, United Kingdom). The crystal structure of sintered specimens was determined using an X-ray diffractometer (XRD, D/Max-220, Rigaku, Japan) under Cu Kα radiation (λ=1.54178 Å ). The flexural strength and shrinkage rate were estimated based on 3-point bending test with a universal testing machine with a span size of 80 mm and crosshead speed of 1.0 mm/min (UTM, H10SK, Hounsfield, United Kingdom).
ciency, the mechanical properties, and the crystal structure were improved through this modified preparation of colloidal alumina particles. 2. Experimental process Commercially available nano-sized aluminum oxide (Al2O3, APS 40–50 nm, 99.5%, Alfa-Aesar, United States) was surface-modified through hyper-branched polyethylenimine (PEI, Mw 60,000, SigmaAldrich, United States), and (3-Aminopropyl)triethoxysilane (APTES, 99%, Sigma-Aldrich, United States). Alumina powder (5g) was suspended in 500 ml of deionized water and stirred at 400 RPM at room temperature for 1 h. 3 wt% of PEI in deionized water was added to the well-dispersed alumina solution and mixed at 80 °C for 3 h. The resulting colloidal particles were centrifuged at 10,000 RPM for 1 h. The refined particles were rinsed with deionized water several times and collected. To modify the surface by silane grafting, alumina powder (5 g) in 500 ml of deionized water was mixed with 100 ml of 1N ammonia solution at 250 RPM at 80 °C for 3 h. The sample was then cooled by the addition of 200 ml of deionized water and mixed with 150 ml of ethanol at 250 RPM at 60 °C for 12 h. The silane precursor solution, composed of 2 ml of APTES, 1 ml of ethanol, and 1 N of ammonia solution, was added 30 min into the procedure. After heating was complete, this product was washed as previously described for PEI. The silica-based matrix of the ceramic core was prepared by ceramic injection molding. 63 wt% of fused silica (amorphous SiO2, 325 mesh, IMERYS, United States) was used as the base material and was modified by addition of 21 wt% of zircon (ZrSiO4, 1 µm, Cenotec, Korea). A thermoplastic binder (15 wt%) composed of paraffin wax (Nippon-seiro, Japan), microcrystalline wax (Nippon-Seiro, Japan), stearic acid (C19H36O2, Samchun Pure Chemical, Korea) and oleic acid (C19H34O2, 97%, Samchun Pure Chemical, Korea) was used as feedstock for injection molding. The feedstocks were injection molded using a C-frame ceramic injection molding machine (CTM-CI-CF-35100HT, Cleveland Tools and Machines, United States) with 6 mm x 8 mm x 90 mm green bodies in accordance with ASTM C 1161-13. After calcination of the thermoplastic binders at 0.2 K/min, the specimens were pre-sintered at 1100 °C for 2 h with a heating rate of 5 K/ min using a box furnace (UAF-15–27-LHE, Lenton, United Kingdom). Pre-sintered specimens were infiltrated by 20% supensions of the prepared surface-modified colloidal alumina in deionized water under vacuum. The length of reaction was varied to achieve different levels of the alumina content. Resulting specimens were dried at 110 °C for 12 h to evaporate residual solvent and sintered at 1300 °C for 2 h. The surface functional groups of the modified alumina particles were determined by Fourier transform infrared spectroscopy (FT-IR, IRAffinity-1S, Shimadzu, Japan). The zeta-potential value at various pH was measured using 0.1 M HNO3 or 0.1 M NaOH in deionized water by electrokinetic sonic amplitude measurements (ESA, Zeta Finder, Matec Applied Sciences, United States). The particle size
3. Results and discussion The morphology of as-prepared and each surface modified alumina particles were examined by SEM in Fig. 1. The as-prepared alumina particles have slightly keen edged quasi-spherical shape with about 40 nm of average diameter approximately. However, the morphology of PEI grafted alumina particles exhibit rugged and distorted surface compared with the raw alumina particles due to grafting of polymer, which has randomly arranged polymeric structure. On the other hand, the morphology of APTES treated alumina particles are similar with the as-prepared one. It is possibly considered that silane layer formed around the surface of alumina particles by condensation of silanol group is homogeneous and very thin which cannot identify with naked eye, thus the morphology of both as-prepared and APTES treated alumina shows no significant difference.. The identity of the surface functional groups of the as-prepared alumina particles and PEI or APTES surface-modified particles through PEI or APTES was verified from the FT-IR spectra (Fig. 2). Absorption peaks for the Al-O stretching vibration were observed at 575 and 787 cm−1 and a broad absorption band of the surface hydroxyl group (Al-OH) in the range of 3700–3000 cm−1 due to the alumina base material [13,14]. Additionally, the N-H bending vibration peak at 1630 cm−1 from the functionalized amine group and the absorption bands between 3000 and 2700 cm−1 due to the C-H stretching vibration of alkyl chains of PEI and APTES were observed [15,16]. For the APETS-grafted specimen, the symmetric stretching vibration peaks of siloxane group (Si-O-Si) were observed at 1080 and 1120 cm−1[16]. The broad absorption band between 3700– 3000 cm−1 in both the surface modified alumina than as-prepared alumina particle is thought to be due to overlap of the –OH and N-H stretching vibrations [17].. In Fig. 3, the zeta potential values at each pH were used to evaluate the level of surface charge. The fused silica has a negative zeta potential value (between −3 to −45 mV) between pH 2 and 11. In this same range, the zeta potential of as-prepared alumina particle was 5 to 2
Ceramics International xx (xxxx) xxxx–xxxx
G.S. An et al.
Fig. 4. Particle size and distribution of the as-prepared alumina, PEI grafted alumina, and APTES treated alumina. Fig. 2. FT-IR spectra of the as-prepared alumina, PEI grafted alumina, and APTES treated alumina.
APTES-modified specimens ( Fig. 4). The advanced surface modification led to an improvement in the dispersion behavior of colloidal alumina particles. This enhancement of dispersion performance, more pronounced in the PEI-grafted particles, was evaluated on the basis of electrostatic repulsion and steric hindrance as well [18]. The large positive charge density of the free surface amine groups bonded to alkyl chain repels the alumina particles because of the high positive zeta potential value. In contrast, the free amine group of APTES exists in low numbers at the surface terminal site of siloxane bond. Due to these inherent structural differences, the two particle types have different surface charges, impacting the dispersion. The molar structure of each precursor affects dispersion behavior in additional ways. For the PEI polymer, the long and branched alkyl chains induce significant steric interference between the PEI-grafted alumina particles. The alkyl chains in the APTES-grafted alumina lead do not lead to significant steric effects because the alkyl chain of the APTES grafted alumina is the much smaller propyl group, bonded to the terminal amine group [19]. Consequently, the dispersion stability of the PEI-grafted alumina is outstanding because of the combination of the strong electrostatic repulsion and limited steric hindrance. Fig. 5 shows the alumina content and apparent porosity of the fused-silica matrix of prepared specimens with different infiltration times. Although the infiltration efficiency varies, the alumina content was increased with an increase in infiltrating time. The PEI-grafted colloidal alumina particle had the most rapid saturation of alumina content and decreasing apparent porosity, which were achieved at 45 min. The APTES-treated colloidal alumina particle achieved saturation in 75 min. However, the raw colloidal alumina particles did not reach the saturation point until 130 min of infiltrating time. Both surface modified specimens had 2.54% alumina content at the saturation point, with a continual subsequent weak increase past this point. At the alumina content achieved in the PEI-grafted specimen after 45 min of infiltration time, the APTES-treated specimen was only 1.41% (60% efficiency); the raw colloidal alumina achieved 72% efficiency.. Fig. 6 illustrates the differences between the raw prepared fused silica and PEI-grafted colloidal alumina infiltrated specimen for 45 min through high resolution SEM and EDX measurements. Through a comparison of those two microstructure images, it is confirmed that the fine particles filled the pores with fused silica substrate in the infiltration process. This is clearly represented in the EDX mapping of Si and Al. The EDX data indicates that the alumina coagulated homogeneously in silica-based porous specimen via infiltration, excluding the fractured large fused silica area (indicated in black).. The infiltration efficiency of the PEI-grafted alumina was outstanding compared with the APTES-treated specimen. The dispersion
Fig. 3. Zeta-potential variation of prepared alumina specimens and fused silica at different pH values.
−14 mV, with an isoelectric point (IEP) at pH 5.6. The difference in zeta potential between these two raw materials is the greatest at pH 3.92. However, this difference is still notably small due to the electrostatic attraction effect.. The alternations of zeta-potential values were expanded to the large number to positive charged direction, and the IEP was shifted to a higher pH number for both amine-functionalized particles. The zetapotential values at neutral pH were found to be 34 mV for the PEIgrafted specimen (IEP of pH 10.4) and 21 mV for APTES-treated specimen (IEP of pH 9.3). A pH of 6.8 (neutral) was found to have the maximum polarity difference between fused silica (−38 mV) and surface-modified colloidal alumina particles. The difference in the zetapotential value is thought to be due to the greater number of free surface amine groups in the PEI-grafted alumina is more than the APTES-treated alumina, leading to larger positive charge densities. Increase in zeta-potential value as a driving force of the dispersion mechanism was also observed represented in particle size distribution of prepared specimens. The average size of the raw alumina particles was 178 nm, larger than the 40–50 nm primary alumina particles. The inherent high surface energy of the small particles and nearby neutral surface charge induces aggregation the alumina particles. A broad monomodal distribution was observed. The PEI and APTES surface modified specimens were of smaller average size and with a narrower monomodal particle distribution. The average size of the PEI- and APTES-grafted alumina particles were 60 and 93 nm, respectively. The PEI-grafted alumina particles had a narrower distribution than the
3
Ceramics International xx (xxxx) xxxx–xxxx
G.S. An et al.
Fig. 5. Infiltrated alumina content and apparent porosity as a function of infiltration time. The filled and open symbols indicate the alumina content and apparent porosity, respectively.
Fig. 7. The X-ray diffraction patterns of the sintered specimens as a function of infiltration time with PEI-grafted colloidal alumina particles.
stability was enhanced via electrostatic repulsion and steric hindrance. The large positive surface charge of the PEI-grafted alumina improved the adsorption capability because of the strong opposition between the negatively charged fused silica particles and positively charged alumina particles. The easy movement of alumina particles onto the surface or pores of the fused silica enhances the infiltration process. The adsorption capability of the APTES-grafted alumina is comparatively weak because of the enhanced dispersity and because the Van der Waals force between the fused silica matrix and siloxane groups bonded to the surface of alumina is weaker than the effect of strong dispersion stabilization and positive surface charge of the PEI-grafted alumina. Therefore, the saturation point of the alumina content was reduced to 45 min in the PEI-grafted alumina. After 75 min, the APTES-treated
alumina achieved the same level of porosity reduction and alumina saturation. Fig. 7 depicts the X-ray diffraction patterns of the sintered specimens at 1300 °C according to the infiltrating time of the PEI-grafted colloidal alumina particle. Phase transformation is proposed due to the coexistence of zircon (JCPDS 6-266) and cristobalite (JCPDS 39-1425). The mullite phase (JCPDS 15-0776) peaks were revealed with infiltration times exceeding 45 min, and cristobalite peaks were not observed. A slight excess of infiltrated alumina (JCPDS 46-1212) was observed specimens with 60 min of infiltration.. The effect of alumina infiltration on mechanical properties are
Fig. 6. Microstructures and EDX mapping results of fused silica based ceramic cores with and without the infiltration of PEI grafted colloidal alumina: (a) non-infiltration, (b) 45 mininfiltration, (c) Al mapping after 45 min-infiltration, and (d) Si mapping after 45 min-infiltration.
4
Ceramics International xx (xxxx) xxxx–xxxx
G.S. An et al.
hibited the crystallization of fused silica via the formation of mullite phase. Acknowledgments This work was supported by the National Research Foundation of Korea (NRF) grant funded by the Korea government (2014R1A2A1A11050220), and the Power Generation and Electricity Delivery of the Korea Institute of Energy Technology Evaluation and Planning (KETEP) grant funded by the Ministry of Trade, Industry and Energy, Korea (20142020103400). References [1] I. Huseby, M. Borom, C. Greskovich, High temperature characterization of silicabase cores for superalloys, Am. Ceram. Soc. Bull. 58 (1979) 448–452. [2] A.A. Wereszczak, K. Breder, M.K. Ferber, T.P. Kirkland, E.A. Payzant, C.J. Rawn, E. Krug, C.L. Larocco, R.A. Pietras, M. Karakus, Dimensional changes and creep of silica core ceramics used in investment casting of superalloys, J. Mater. Sci. 37 (2002) 4235–4245. [3] C.-H. Chao, H.-Y. Lu, Optimal composition of zircon-fused silica ceramic cores for casting superalloys, J. Am. Ceram. Soc. 85 (2002) 773–779. [4] S. Pattnaik, D.B. Karunakar, P.K. Jha, Developments in investment casting processa review, J. Mater. Process. Technol. 212 (2012) 2332–2348. [5] M. Gromada, A. ´Swieca, M. Kostecki, A. Olszyna, R. Cygan, Ceramic cores for turbine blades via injection moulding, J. Mater. Process. Technol. 220 (2015) 107–112. [6] W.D. kingery, H.K. Bowen, D.R. Uhlmann, Introduction to Ceramics, John Wiley & Sons, New York, 1976. [7] L.L. Hench, S.H. Wang, The sol-gel glass transformation of silica, Phase Transit. 24–26 (1990) 785–834. [8] C.-H. Chao, H.-Y. Lu, Stress-induced β→α-cristobalite phase transformation in (Na2O+Al2O3)-codoped silica, Mater. Sci. Eng. A 328 (2002) 267–276. [9] A. Kazemi, M.A. Faghihi-Sani, H.R. Alizadeh, Investigation on cristobalite crystallization in silica-based ceramic cores for investment casting, J. Eur. Ceram. Soc. 33 (2013) 3397–3402. [10] A. Kazemi, M.A. Faghihi-Sani, M.J. Nayyeri, M. Mohammadi, M. Hajfathalian, Effect of zircon content on chemical and mechanical behavior of silica-based ceramic cores, Ceram. Int. 40 (2014) 1093–1098. [11] R.C. Breneman, J.W. Halloran, Effect of cristobalite on the strength of sintered fused silica above and below the cristobalite transformation, J. Am. Ceram. Soc. 98 (2015) 1611–1617. [12] Y.-H. Kim, J.G. Yeo, S.-C. Choi, Shrinkage and flexural strength improvement of silica-based composites for ceramic cores by colloidal alumina infiltration, Ceram. Int., 42, pp. 8878–8883. [13] J. Gangwar, B.K. Gupta, S.K. Tripathi, A.K. Srivastava, Phase dependent thermal and spectroscopic responses of Al2O3 nanostructures with different morphogenesis, Nanoscale 7 (2015) 13313–13344. [14] L. Wu, Z. Yin, Sulfonic acid functionalized nano γ-Al2O3 catalyzed per-O-acetylated of carbohydrates, Carbohydr. Res. 365 (2013) 14–19. [15] L. Feng, H. Zhang, P. Mao, Y. Wang, Y. Ge, Superhydrophobic alumina surface based on stearic acid modification, Appl. Surf. Sci. 257 (2011) 3959–3963. [16] A. Huang, N. Wang, J. Caro, Seeding-free synthesis of dense zeolite FAU membranes on 3-aminopropyltriethoxysilane-functionalized alumina supports, J. Membr. Sci. 389 (2012) 272–279. [17] A.S.M. Chong, X.S. Zhao, Functionalization of SBA-15 with APTES and characterization of functionalized materials, J. Phys. Chem. B 107 (2003) 12650–12657. [18] T. Xia, Y. Guan, M. Yang, W. Xiong, N. Wang, S. Zhao, C. Guo, Synthesis of polyethylenimine modified Fe3O4 nanoparticles with immobilized Cu2+ for highly efficient proteins adsorption, Colloids Surf. A 443 (2014) 552–559. [19] R.A. Bini, R.F.C. Marques, F.J. Santos, J.A. Chaker, M. Jafelicci Jr., Synthesis and functionalization of magnetite nanoparticles with different amino-functional alkoxysilanes, J. Magn. Magn. Mater. 324 (2012) 534–539.
Fig. 8. Flexural strength (σ3pt) and linear shrinkage (ΔL/L0) as a function of infiltration time with PEI-grafted colloidal alumina particles.
summarized in Fig. 8, showing the flexural strength and linear shrinkage against infiltrating time. As reference, the silica-based green body specimen had 3.2 MPa of flexural strength and 1.91% of shrinkage rate, respectively. These mechanical properties were improved to 10 MPa and 1.04% after up to 45 min of infiltration. Because of the drastic reduction of the rate of increase of alumina content after additional infiltrating time, the mechanical properties were also not further improved.. 4. Conclusions In this study, amine-functionalized alumina for enhanced infiltration was created by surface modification with uniquely structured PEI and APTES precursors in order to improve the mechanical properties of the fused silica-based ceramic core. From the observed C-H stretching and N-H bending or symmetric stretching of siloxane group (Si-O-Si) vibration peaks on the FT-IR spectra, the functionalization with the amine group was confirmed. As a result, the enlarged zeta potential value was 34 and 21 mV for the PEI- and APTES-grafted alumina at neutral pH, respectively. The average particle sizes were 60 and 93 nm with a narrower particle distribution for the PEI-grafted alumina. The dispersion stability and positive surface charge of the PEI-grafted alumina are greater than that of the APTES-grafted alumina. This accelerates the infiltration behavior, reducing the saturation time to 45 min for the PEI-grafted alumina, whereas 75 min was required for APTES-grafted alumina to achieve the same porosity reduction and alumina saturation. These results were explained by the fundamental structural differences of the PEI and APTES, which are more efficient for improving the dispersity at the PEI grafted alumina through a number of amine groups and longbranched alkyl chain induced by the polymeric structure of PEI. Increased infiltration of the pre-sintered ceramic core using the PEI grafted alumina increased the flexural strength to 10 MPa and in-
5