Available online at www.sciencedirect.com
CERAMICS INTERNATIONAL
Ceramics International 41 (2015) 8630–8636 www.elsevier.com/locate/ceramint
Mullite-based cellular ceramics obtained by a combination of direct foaming and reaction bonding Thomas Koneggern, Ruth Felzmann, Birgit Achleitner, Dominik Brouczek Institute of Chemical Technologies and Analytics, Vienna University of Technology, Getreidemarkt 9/164-CT, 1060 Vienna, Austria Received 5 February 2015; received in revised form 11 March 2015; accepted 13 March 2015 Available online 21 March 2015
Abstract Mullite-based cellular ceramics were prepared via the polymer precursor route using poly(silsesquioxane) in combination with particulate alumina or alumina/aluminum mixtures. The multi-functional preceramic polymer was used as pore-forming agent by employing a self-foaming process during the polymer cross-linking step, as well as a precursor for reactive silica, one of the reagents in mullite formation. The size of filler particulates was found to strongly affect foaming of the polymer/filler mixtures, with coarser particles facilitating an improved foaming performance. Thermal conversion in air at 1600 1C resulted in the formation of cellular ceramics with high mullite contents. The partial substitution of alumina with aluminum in the initial mixtures resulted in improved mechanical properties at comparable porosities, resulting in compressive strengths of 0.3 MPa at total porosities of 93%. A correlation between thermal analysis data and crystalline phase development during the thermal treatment allowed for the clarification of processes taking place during heat treatment, yielding information for a future process optimization approaches. & 2015 Elsevier Ltd and Techna Group S.r.l. All rights reserved.
Keywords: A. Precursors: organic; A. Shaping; B. Porosity; D. Mullite
1. Introduction Mullite (3Al2O3 2SiO2) is a structural ceramic material commonly used in high-temperature applications owing to its unique thermal and mechanical properties. A low coefficient of thermal expansion in combination with low thermal conductivity and high creep resistance as well as suitable mechanical properties at high temperatures have led to the use of mullitebased materials for a wide variety of both traditional and advanced ceramic applications [1,2]. As an alternative to conventional powder-based production techniques, the use of Si-based preceramic polymers in combination with Al2O3 or Al fillers has been suggested for the production of mullite-based ceramics [3–6]. A general, inherent advantage of the polymer route for the production of ceramics is the wide variety of forming techniques due to the polymer nature before thermal polymer-to-ceramic conversion [7]. Furthermore, a n
Corresponding author. Tel.: þ43 1 58801 16161; fax: þ 43 1 58801 16199. E-mail address:
[email protected] (T. Konegger).
http://dx.doi.org/10.1016/j.ceramint.2015.03.073 0272-8842/& 2015 Elsevier Ltd and Techna Group S.r.l. All rights reserved.
thermo-oxidative degradation of the polymer compound yields highly reactive amorphous silica, thus effectively reducing the temperature necessary for mullite formation. Even lower mullite formation temperatures were achieved by using nano-scaled Al2O3 fillers [8]. In the past years, the polymer precursor route has been increasingly considered for the formation of cellular ceramic structures [9–11]. Cellular ceramics enable further improvement of properties relevant to applications at high temperatures, including improvement of the thermal shock behavior or thermal insulation [12]. The generation of cellular structures from polymer precursors can either be obtained by application of a blowing agent [13], by in-situ foaming of the precursor component [14], or by the use of sacrificial fillers within the polymercontaining initial mixture [15]. Following these concepts, a combination of polysiloxanes with Al2O3 and sacrificial fillers yielding cellular mullite has first been reported by Kim et al. [16]. In this contribution, we present the generation of cellular mullite employing Si-based polymer compounds with multifunctionality, acting as foaming agent, matrix material, and
T. Konegger et al. / Ceramics International 41 (2015) 8630–8636
reagent in combination with particulate Al2O3 as well as an active Al filler within a single processing routine. Pore generation is accomplished by an in-situ foaming process, taking place during the cross-linking of the preceramic poly(silsesquioxane) polymer, a technique developed and extensively studied by Greil and co-workers [14,17–19]. By expanding on initial reports showing the formation of mullite when introducing particulate Al2O3 to this preceramic polymer [14,20], our investigations are focused on the identification of relevant parameters influencing the final product as a result of the strong correlation between foaming behavior and starting materials, aiming towards homogeneous pore structures and high mullite yields. Another closely related goal is the clarification of the thermal conversion process leading to the formation of mullite and, subsequently, the preservation of the structural and mechanical features established during the initial foaming step.
2. Materials and methods 2.1. Pore generation by direct foaming Direct foaming was accomplished by using a phenyl methyl poly(silsesquioxane) (PMPS; Silres H44, Wacker Chemie, Germany) with a nominal elemental composition of SiO1.51 C4.03H4.06 [18]. PMPS is solid at room temperature, and starts to form a viscous melt at around 60 1C. During a thermally-induced cross-linking starting at around 200 1C, condensation reactions between small amounts of hydroxy and ethoxy functionalities within the PMPS compound lead to the formation of water or ethanol, effectively acting as in-situ foaming agents within the viscous polymer melt. Further cross-linking of the obtained cellular matrix leads to the formation of a stable, porous thermoset [17]. As fillers, particulate Al2O3 with two distinct particle sizes (fine: CT 3000 SG, d50 ¼ 0.7 mm; or coarse: CT 1200 SG, d50 ¼ 1.3 mm; both from Almatis, Germany) and Al (DG38, d50 ¼ 15–20 mm; Ecka Granules, Germany) were used. For the preparation of cellular mullite from PMPS and Al2O3, relative amounts of educts were chosen in order to yield stoichiometric mullite (3Al2O3 2SiO2) after total oxidation. Preliminary investigations of the thermal conversion of PMPS in air showed a SiO2 yield of 51% after heating to 1000 1C. For the samples containing Al in addition to PMPS and Al2O3, the ratio of PMPS to particulate fillers was slightly lower in order to account for the difference in PMPS volume fraction and to allow for a comparable foaming behavior. The compositions used are listed in Table 1. The mixtures of PMPS and fillers were homogenized in a tumbler mixer for 1 h. For a typical sample, 50 g of the obtained powder was filled into an aluminum sleeve (diameter 50 mm, height 140 mm). After a thermal preconditioning to improve the foaming characteristics of two of the mixtures (4 h at 100 1C for PMPS/Al2O3-f, and 1 h at 200 1C for PMPS/Al2O3/Al), the container was put into a preheated furnace at 270 1C in air for a duration of 4 h in order to obtain the porous thermoset. Information about the impact of the Al2O3 particle size on the cross-linking and foaming behavior of PMPS was gained by thermogravimetric investigations of the PMPS/filler mixtures up
8631
Table 1 Sample compositions, in wt%.
PMPS Al2O3 (d50 ¼ 0.7 mm) Al2O3 (d50 ¼ 1.3 mm) Al (d50 ¼ 15–20 mm)
PMPS/Al2O3-f
PMPS/Al2O3-c
PMPS/Al2O3/Al
43.5 56.5 – –
43.5 – 56.5 –
47.2 – 34.5 18.3
to a temperature of 300 1C (10 K min 1, flowing Ar; TG 209 F3 Tarsus, Netzsch, Germany). 2.2. Formation and characterization of cellular ceramic foams The porous thermosets were cut into block-shaped specimens (typical dimensions: 35 35 25 mm3) using a diamond cutoff wheel. Ceramization was conducted in a box furnace in air. The heat treatment involved an initial heating step to 1200 1C with a heating rate of 1 K min 1 and a 2 h hold. Subsequently, the system was heated to 1600 1C with a heating rate of 1 K min 1 and held at this temperature for 4 h before cooling. After heat-treatment, the bulk density ρ was calculated from the weight-to-volume ratio of the block-shaped specimens. The porosity Ф of specimens was calculated following the relation Ф ¼ 1 (ρ/ρth), assuming a theoretical density ρth ¼ 3.2 g cm 3, ρth being the density of mullite [2]. The microstructure was investigated by light-optical microscopy (SteREO Discovery.V20, Zeiss, Germany) as well as scanning electron microscopy (Quanta 200, FEI, the Netherlands). The crystalline phase composition was determined by powder X-ray diffraction analysis (XRD) employing Cu Kα radiation (X'Pert Pro, Philips, the Netherlands) after crushing the samples in a planetary ball mill. For selected samples, a quantitative estimation of the relative phase contents following Rietveld's method was carried out using the Topas software package (Topas R 2.1, Bruker, USA). Al-containing specimens in cross-linked state were investigated by simultaneous thermal analysis including thermogravimetry and differential thermal analysis in flowing synthetic air up to 1600 1C (STA 449C Jupiter, Netzsch, Germany). The compressive strength of block-shaped specimens (dimensions after ceramization: 30 30 15 mm3) was determined with a universal testing machine (Model 1474, Zwick, Germany) with a cross-head speed of 0.5 mm min 1. Before testing, the top and bottom sample surfaces were planeparallelized by grinding with SiC paper. 3. Results and discussion 3.1. Influence of particulates on direct foaming process In the PMPS system used, temperature-induced polycondensation reactions lead to the cross-linking of polymer chains and the evolution of gases responsible for foaming of the viscous
8632
T. Konegger et al. / Ceramics International 41 (2015) 8630–8636
mixture, including H2O and CH3CH2OH. The interaction between gas formation temperature, polymer melt viscosity, as well as degree of cross-linking, i.e., stabilization of the foamed structure, has been shown to be of major importance for the resulting foam morphology [19]. In this work, the presence of particulates in the PMPS component was found to significantly affect the cross-linking behavior of the PMPS component and, subsequently, the foaming behavior of the composites. After melting of the PMPS component at temperatures of 60–70 1C, the presence of Al2O3 particulates led to a shift of the onset temperature of polycondensation reactions towards lower temperatures, which was monitored by the mass change behavior of polymer/filler mixtures (Fig. 1). While the filler-free PMPS showed a typical onset of mass loss at around 200 1C, the addition of Al2O3 powders resulted in a lowering of the onset to
115–140 1C, depending on the particle size of the Al2O3 powders used. This, in turn, significantly affected the morphology of the cross-linked foamed specimens (Fig. 2). The presence of sub-mm Al2O3 particles deterred the formation of a porous thermoset with a homogeneous pore size distribution. Due to the early onset of gas formation in the foaming process, no stabilization of gas bubbles within the polymer network can be achieved, leading to a collapse of the porous structures upon further heating. A preconditioning of the mixture at 100 1C – a temperature well below initial gas formation – led to a slightly improved foaming behavior, albeit collapse of foams remained prevalent (Fig. 2a). In contrast, the foaming characteristic was found to be significantly different for PMPS/Al2O3-c and PMPS/Al2O3/Al mixtures, both containing coarser Al2O3 particles (d50 ¼ 1.3 mm). In both cases, a stable foam formation was observed, leading to cross-linked materials with significantly higher porosities and typical pore sizes ranging from 1 to 4 mm (Fig. 2b and c). In case of PMPS/Al2O3/Al, an additional preconditioning at 200 1C for 1 h was found to improve the reproducibility of foaming results. 3.2. Morphology and properties of ceramic foams
Fig. 1. Mass loss behavior upon heating of (a) pure PMPS, (b) PMPS/Al2O3-f, (c) PMPS/Al2O3-c, and (d) PMPS/Al2O3/Al. The arrows indicate the respective onset of mass loss.
After thermal conversion, the ceramized specimens were evaluated in terms of morphology, porosity, and mechanical properties. Significant differences in the microstructure were found depending on the initial composition, in analogy to observations during the foaming process as discussed in the previous section. The use of sub-mm Al2O3 resulted in porous materials with significantly lower porosity than materials prepared with the coarser filler particles. In contrast, the PMPS/Al2O3-c and PMPS/Al2O3/Al specimens exhibited comparable pore structures with pores in the size range of 1– 3 mm (Fig. 3) and total porosities of up to 94% (Table 2). A critical factor during the preparation of polymer-derived ceramic foams is the retention of the structural integrity during the polymer-to-ceramic conversion. The loss of volatiles, in this case amounting to nearly 50% of the initial weight of the precursor component, as well as significant increases in density during the conversion process result in stresses within the struts and,
Fig. 2. Morphology of the cross-section of foamed samples: (a) PMPS/Al2O3-f, (b) PMPS/Al2O3-c, and (c) PMPS/Al2O3/Al. The arrow indicates the sample foaming direction.
T. Konegger et al. / Ceramics International 41 (2015) 8630–8636
8633
Fig. 3. Cellular macro- and microstructure of materials after thermal conversion: (a) PMPS/Al2O3-f, (b) PMPS/Al2O3-c, (c, d) PMPS/Al2O3/Al.
Table 2 Structural and mechanical properties of cellular ceramic materials.
Green density (g cm 3) Final density (g cm 3) Final porosity (%) Linear shrinkage (%) Compressive strength (MPa)
PMPS/ Al2O3 f
PMPS/ Al2O3 c
PMPS/ Al2O3/Al
0.68 0.87 73 15 1.1
0.15 0.20 94 15 o0.1
0.19 0.22 93 10 0.3
subsequently, in the formation of cracks during the thermal treatment. For PMPS/Al2O3 specimens, a linear shrinkage of 15% was observed, independent of the Al2O3 particle size used. The compressive strength values were influenced by both the total porosity and the presence of macroscopic cracks in the material. In the case of PMPS/Al2O3-c, shrinkage-dependent crack formation during the thermal treatment process resulted in very low compressive strength values well below 0.1 MPa, which can be considered inadequate for most potential applications. The addition of Al as an active component reduced the linear shrinkage of PMPS/Al2O3/Al specimens to 10% without negatively affecting the evolution of an open-cell pore network. The lowered shrinkage resulted in a reduction of the crack formation tendency. Even though surface cracks along the pore walls were present after thermal treatment (Fig. 3c and d), the absence of macroscopic cracks across struts resulted in an increase in compressive strength values at total porosities comparable to PMPS/Al2O3-c. The stress/displacement curves
Fig. 4. Typical stress/displacement curves during compressive testing of ceramized samples (PMPS/Al2O3/Al).
observed during compressive testing of PMPS/Al2O3/Al demonstrated a good reproducibility of results, with a typical increase in compressive stress up to the collapse of the full cross-sectional strut network, and a progressing collapse of individual cells and sample regions afterwards (Fig. 4). 3.3. Thermal conversion to mullite The clarification of the thermal conversion process from the starting mixtures to the final ceramic material is of significant interest for understanding and facilitating the proposed process
8634
T. Konegger et al. / Ceramics International 41 (2015) 8630–8636
to form cellular mullite. After a general overview of the final crystalline composition of all three starting compositions, the focus of this section is set on PMPS/Al2O3/Al due to its promising structural and mechanical properties. An XRD investigation of the crystalline phase composition after the thermal treatment at 1600 1C showed that all investigated materials primarily consist of mullite (Fig. 5). In all cases,
Fig. 5. XRD diffraction patterns of (a) PMPS/Al2O3-f, (b) PMPS/Al2O3-c, and (c) PMPS/Al2O3/Al after thermal treatment at 1600 1C.
unreacted Al2O3 was found in diffractograms. For PMPS/Al2O3 samples the lack of significant amounts of crystalline SiO2 leads to the assumption that the residual SiO2 is most likely present in amorphous state, assuming a complete oxidation of the polysiloxane-derived reaction products. According to XRD investigations, the PMPS/Al2O3/Al material contains small amounts of elemental Si after thermal treatment (Fig. 5c). The formation of elemental Si during the heat treatment of polysiloxane/Al mixtures has been reported in the literature previously, and has been attributed to an aluminothermal reduction of either SiO2 or SiOxCy derived from the preceramic polymer [5,6]. Since the mullitization process is carried out well beyond the melting point of Si at 1414 1C, Si accumulates as small droplets on the inner pore walls of the cellular structure (Fig. 6). Droplet formation can potentially be explained by the densification of the ceramic backbone during mullitization, leading to an extrusion of the Si melt. Oxidation of the melt during the heat treatment results in the formation of a thin oxide layer on the droplet surface, which ruptures upon further heating, as observed by SEM and EDX investigations (Fig. 6 b, c and d). To obtain a more detailed clarification of the thermal conversion process in PMPS/Al2O3/Al samples, simultaneous thermal analysis (Fig. 7) was correlated with qualitative and quantitative crystalline phase composition data at different stages during the heat treatment, acquired by cut-off experiments (Fig. 8). Several distinct phases can be observed during the thermal conversion process. In the temperature range between 400 1C and 800 1C, the thermal decomposition of PMPS takes place,
Fig. 6. Presence of Si/Al droplets on pore walls of PMPS/Al2O3/Al after thermal conversion, recorded by (a) optical microscopy and (b) SEM. Al- and Si-rich areas in (b) are highlighted by EDX mapping (c,d).
T. Konegger et al. / Ceramics International 41 (2015) 8630–8636
8635
Fig. 7. Simultaneous thermal analysis (TG and DTA) of the thermal conversion process for PMPS/Al2O3/Al.
Fig. 8. X-ray diffraction patterns of PMPS/Al2O3/Al after various stages of the thermal conversion process (a) with samples 1–5 obtained through cut-off experiments at process stages shown in (b), and the resulting quantitative crystalline phase composition of the respective samples estimated by Rietveld refinement of diffraction data (c).
resulting in a continuing mass loss, and eventually leading to amorphous products such as SiO2 or SiOxCy. Up to 600 1C, no significant oxidation of Al can be observed, the crystalline phase ratio of Al2O3/Al being in the range of the initial powder composition (taking into account error margins inherent to the XRD data refinement method used). The melting of the Al compound at 660 1C can be identified by a small endothermic peak in the DTA. Starting at 800 1C, the mass loss is slowly being compensated by oxidation of Al. The exothermic peak found at 880 1C can be attributed to the rupture of the oxide scale on Al melt droplets due to the thermal expansion of the melt, resulting in rapid oxidation of the liquid phase, an observation reported for a variety of reaction-bonded aluminum-based systems [21]. The rapid mass gain at this stage can be attributed to this process. From 900 1C to 1200 1C, crystallization of the amorphous SiO2 to cristobalite occurs, alongside the aluminothermal reduction of SiO2 to Si 4Al þ 3SiO2-2Al2O3 þ 3Si
(1)
In the final stage from 1200 1C to 1600 1C, the remaining Si and Al are further oxidized (resulting in a mass gain in this temperature region), and mullite is formed: 3Al2O3 þ 2SiO2-3Al2O3 2SiO2
(2)
A small endothermic peak indicates the melting of Si at 1410 1C. From this point on, the weight gain and, correspondingly, the oxidation rate increase. During the final hold at 1600 1C, the remaining crystalline SiO2 is consumed, further increasing the mullite content of the sample. The observed stages of the phase evolution during the thermal conversion process are in general accordance with findings by Michalet et al. [5] and Anggono and Derby [6], who investigated the phase evolution in dense polydimethylsiloxane/Al and polymethylsiloxane/Al systems, respectively. However, the formation of Si droplets on the specimen surface appears to be a phenomenon inherent to cellular materials, potentially due to the high surface-to-volume ratios of the strut/pore wall regions. An adjustment of the thermal conversion program can be
8636
T. Konegger et al. / Ceramics International 41 (2015) 8630–8636
expected to mitigate this process, further increasing the mullite yield of the proposed processing approach. The observed formation of elemental silicon can generally be expected to be detrimental to the overall strength of the material; however, the reduction in shrinkage and the subsequent mitigation of crack formation during the heat treatment can be assumed as the main reason for increased mechanical performance of Al-containing samples in comparison to Al-free samples with comparable porosity investigated in this work. 4. Conclusions Cellular mullite-based ceramics can be produced via the polymer precursor route by combining PMPS with Al2O3 and Al fillers through a combined direct foaming/reaction bonding technique. The multi-functional preceramic polymer acts as the foaming agent, mechanically stabilizes the matrix during the thermal conversion process, and its decomposition product, silica, acts as a reagent in the mullite formation process. In addition to the requirement for a precise temperature control during conditioning and foaming, the characteristics of the added fillers significantly affect the foaming process. The presence of particulate fillers resulted in a shift of the mass loss onset temperature of the preceramic polymer towards lower temperatures. While filler particle sizes larger than 1 mm led to the reproducible generation of structures with porosities of up to 94%, the use of sub-mm sized Al2O3 particles deterred the foaming process, eventually leading to an increased collapse of pore structures during foaming. By partial substitution of Al2O3 with elemental Al in the initial mixture, the shrinkage during the thermal conversion process can be reduced, thus facilitating the preservation of the foamed structures without the formation of macroscopic cracks, as opposed to specimens without Al addition. This, in turn, improves the mechanical properties of the final cellular ceramics while keeping the total porosity and resulting mullite contents at comparably high values. The observed formation of elemental Si through aluminothermal reduction, an unwanted side reaction leading to decreased mullite yields, can potentially be avoided by a well-tailored heat treatment regime based on the detailed knowledge of the phase evolution during the thermal conversion process. Acknowledgments The authors gratefully acknowledge Erich Halwax for support with XRD investigations, as well as Wacker Chemie AG, Ecka Granules GmbH, and Almatis GmbH for provision of raw materials.
References [1] H. Schneider, S. Komarneni, Mullite, Wiley-VCH Verlag, Weinheim, 2005. [2] H. Schneider, J. Schreuer, B. Hildmann, Structure and properties of mullite-a review, J. Eur. Ceram. Soc. 28 (2008) 329–344. [3] D. Suttor, H.J. Kleebe, G. Ziegler, Formation of mullite from filled siloxanes, J. Am. Ceram. Soc. 80 (1997) 2541–2548. [4] T. Michalet, M. Parlier, A. Addad, R. Duclos, J. Crampon, Formation at low temperature with low shrinkage of polymer/Al/Al2O3 derived mullite, Ceram. Int. 27 (2001) 315–319. [5] T. Michalet, M. Parlier, F. Beclin, R. Duclos, J. Crampon, Elaboration of low shrinkage mullite by active filler controlled pyrolysis of siloxanes, J. Eur. Ceram. Soc. 22 (2002) 143–152. [6] J. Anggono, B. Derby, Intermediate phases in mullite synthesis via aluminum- and alumina-filled polymethylsiloxane, J. Am. Ceram. Soc. 88 (2005) 2085–2091. [7] P. Colombo, G. Mera, R. Riedel, G.D. Sorarù, Polymer-derived ceramics: 40 years of research and innovation in advanced ceramics, J. Am. Ceram. Soc. 93 (2010) 1805–1837. [8] E. Bernardo, P. Colombo, E. Pippel, J. Woltersdorf, Novel mullite synthesis based on alumina nanoparticles and a preceramic polymer, J. Am. Ceram. Soc. 89 (2006) 1577–1583. [9] P. Colombo, J. Hellmann, Ceramic foams from preceramic polymers, Mater. Res. Innov. 6 (2002) 260–272. [10] B.V.M. Kumar, Y.W. Kim, Processing of polysiloxane-derived porous ceramics: a review, Sci. Technol. Adv. Mater. 11 (2010) 044303. [11] A.R. Studart, U.T. Gonzenbach, E. Tervoort, L.J. Gauckler, Processing routes to macroporous ceramics: a review, J. Am. Ceram. Soc. 89 (2006) 1771–1789. [12] M. Scheffler, P. Colombo, in: Cellular Ceramics, Wiley-VCH Verlag, Weinheim, 2005. [13] P. Colombo, M. Modesti, Silicon oxycarbide ceramic foams from a preceramic polymer, J. Am. Ceram. Soc. 82 (1999) 573–578. [14] T. Gambaryan-Roisman, M. Scheffler, P. Buhler, P. Greil, Processing of ceramic foam by pyrolysis of filler containing phenylmethyl polysiloxane precursor, Ceram. Trans. 108 (2000) 121–130. [15] P. Colombo, E. Bernardo, Macro- and micro-cellular porous ceramics from preceramic polymers, Compos. Sci. Technol. 63 (2003) 2353–2359. [16] Y.-W. Kim, H.-D. Kim, C.B. Park, Processing of microcellular mullite, J. Am. Ceram. Soc. 88 (2005) 3311–3315. [17] J. Zeschky, M. Scheffler, P. Colombo, P. Greil, Cellular oxide ceramics from filler loaded silicone resins, Ceram. Eng. Sci. Proc. 23 (2002) 285–290. [18] J. Zeschky, F. Goetz-Neunhoeffer, J. Neubauer, S.H. Jason, Lo, B. Kummer, M. Scheffler, P. Greil, Preceramic polymer derived cellular ceramics, Compos. Sci. Technol. 63 (2003) 2361–2370. [19] J. Zeschky, T. Höfner, C. Arnold, R. Weißmann, D. Bahloul-Hourlier, M. Scheffler, P. Greil, Polysilsesquioxane derived ceramic foams with gradient porosity, Acta Mater. 53 (2005) 927–937. [20] T. Gambaryan-Roisman, M. Scheffler, T. Takahashi, P. Buhler, P. Greil, Formation and properties of poly(siloxane) derived ceramic foams EUROMAT '99, in: G. Müller (Ed.), Ceramics – Processing, Reliability, Tribology and Wear, 12, Wiley-VCH Verlag GmbH, Weinheim, 2000, pp. 247–251. [21] S. Wu, D. Holz, N. Claussen, Mechanisms and kinetics of reactionbonded aluminum oxide ceramics, J. Am. Ceram. Soc. 76 (1993) 970–980.