- Email: [email protected]
mullite composites formation by gas pressure infiltration
Materials Characterization 114 (2016) 9–17
Contents lists available at ScienceDirect
Materials Characterization journal homepage: www.elsevier.com/locate/matchar
Transmission electron microscopy observations on phase transformations during aluminium/mullite composites formation by gas pressure infiltration M. Pawlyta a,⁎, B. Tomiczek a, L.A. Dobrzański a, M. Kujawa a, B. Bierska-Piech b a b
Silesian University of Technology, Institute of Engineering Materials and Biomaterials, Konarskiego 18A, 44-100 Gliwice, Poland Silesian Centre for Education and Interdisciplinary Research, 75 Pułku Piechoty 1A, 41-500 Chorzów, Poland
a r t i c l e
i n f o
Article history: Received 10 September 2015 Received in revised form 10 February 2016 Accepted 11 February 2016 Available online 13 February 2016 Keywords: Mullite Composite Interface Electron microscopy X-ray diffraction
a b s t r a c t The porous ceramic preforms were manufactured using the powder metallurgy technique. First, the start-up material (halloysite with the addition of carbon fibres as the pore-forming agent) was slowly heated to 800 °C and then sintered at 1300 °C. Degradation of the carbon fibres enabled the open canals to form. At the end of the sintering process, the porous ceramic material consisting mainly of two phases (mullite and cristobalite) was formed, without any residual carbon content. During infiltration, the liquid metal filled the empty spaces (pores) effectively and formed the three-dimensional network of metal in the ceramic. The cristobalite was almost entirely decomposed. In the areas of its previous occurrence, there are new pores, only in the ceramic grains. The mullite, which was formed from halloysite during annealing, crystallized in the Pbam orthorhombic space group, with the (3Al2O3·2SiO2) stoichiometric composition. The mullite structure does not change during the infiltration. The composite components are tightly connected. A transition zone between the ceramics and the metal, having the thickness of about 200 nm, was formed. The nanocrystalline zone, identified as γ-Al2O3, was formed by diffusing the product of the cristobalite decomposition into the aluminium alloy matrix. There is an additional, new phase, identified as (Mg,Si)Al2O4 in the outer parts of the transition zone. © 2016 Elsevier Inc. All rights reserved.
1. Introduction Mullite is one of the most important ceramic materials. Because of the formation conditions (high temperature, low pressure), it is rarely formed in natural conditions. Nevertheless, its importance in traditional and advanced ceramics (including electronic, optical, and hightemperature structural applications) is huge and is still increasing [1, 2]. Mullite consists of many phases, which differ in aluminium, oxygen and silicon content [3]. The general composition of mullite can be collectively described as Al4 + 2xSi2 − 2xO10 − x (0 b x ≤ 1) [4]. The first of this group is sillimanite Al2SiO5, for which x = 0. The structure of all mullite phases can be described as modified sillimanite where aluminium atoms are exchanged for silicon atoms while overall charge neutrality is maintained [5]. The charge neutrality condition causes part of the oxygen atoms to be removed, and vacancies appear in these places. The parameter x gives the number of oxygen vacancies per unit cell. Theoretically, all mullite compositions with x between zero and one are possible, but so far only phases with x between 0.2 and 0.9 have been observed [6]. Most frequently described are [7,8]: a. “Sinter-mullite” with stoichiometric composition (3Al2O3·2SiO2) obtained as a result of heat treatment, typically via solid-state
⁎ Corresponding author. E-mail address: [email protected] (M. Pawlyta).
http://dx.doi.org/10.1016/j.matchar.2016.02.003 1044-5803/© 2016 Elsevier Inc. All rights reserved.
reactions. In this case x = 0.25 corresponding to about 72 wt.% of Al2O3, b. “Chemical-mullite”, also obtained as a result of heat treatment but with no stoichiometric composition. Composition of these mullites strongly depends on the starting materials and the temperature treatment, c. “Fused-mullite” (2Al2O3·SiO2) produced by crystallizing aluminosilicate melts. In this case x = 0.40 corresponding to about 78 wt.% Al2O3.
The structure and chemical composition change depending on the process formation temperature and velocity. The mechanical properties and electronic structures of mullite can be calculated based on its structural models [9]. Therefore, the knowledge and understanding of mullite are so much important. Nevertheless, the structure and chemical composition change depending on the process formation temperature and velocity. All mullites are crystalline in the orthorhombic space group Pbam [10,11] and do not transform to a tetragonal form [12]. If mullites are obtained from homogenous, powdered precursors in low temperatures (below 1200 °C), mullite with high Al2O3 (70 mol% Al2O3) forms first and without regard to the initial material composition. Mullite crystals formed this way have an a lattice close to the b lattice (“mullite with pseudotetragonal metric”) [13,14]. The degree of diversification between a
10
M. Pawlyta et al. / Materials Characterization 114 (2016) 9–17
and b can be rated by analysis of the shape of the X-ray diffraction (XRD) reflex that appears at the angle of about 2θ ≈ 26° (for CuKα radiation). For a = b, it should be a single, symmetrical reflex. The larger the difference between a and b, the more asymmetric it becomes, and finally it splits into two separate reflections: (120) and (210), corresponding to two interplanar distances (d120 = 0.3428 nm, d210 = 0.3390 nm) [15]. Incorporation of SiO2 into the mullite structure causes differentiation in the lattice sizes and a corresponding structure change. Mullite can be used in one of the following forms: monolith mullite ceramics, mullite coatings and mullite matrix in composites. In particular, investigations in the field of mullite composites have been intensively conducted in the last two decades. The main aim of this research focused on improving the toughness. Despite previous intense efforts, no significant breakthrough has been achieved. The unresolved problem is the inhomogeneous dispersion of mullite grains in the metal matrix. A new, promising approach is the application of mullite in the form of continuous fibres or three-dimensional skeletons (preforms), which are then filled with molten metal, usually aluminium or its alloys [1,16]. A successful example of that approach is the obtaining of a threedimensional network of metal in the ceramic [17]. In this case, porous ceramic preforms were manufactured using a powder metallurgy technique. The start-up material, halloysite powder (delivered by NaturalNano) with the addition of 30 wt % carbon fibres (SGL Carbon Group) as a pore-forming agent, was subjected to dry mixing in a Fritsch Pulverisette 6 ball mill for 10 min. The prepared powder mixture was compacted in a steel die with an inside diameter of 30 mm on a LabEcon600 platen press at 100 MPa pressure and then sintered at 1300 °C (heating rate ~2.5 K/min). Thermal degradation of the carbon fibres facilitated the formation of open porosity. Sintering of nanotubular halloysite up to 1300 °C induces the formation of a mullite phase [18]. At the end of the sintering process, porous crystalline mullite was formed without any residual carbon content. In order to fabricate the composite materials reinforced by the previously prepared porous preforms, the gas pressure infiltration technique was used. The infiltration process was performed in a specially designed autoclave [19]. The AlSi12 alloy was heated to 800 °C in silicon carbide crucible. The molten aluminium alloy was infiltrated into the pores of the preform under the influence of nitrogen pressure (3 MPa for 30 s). As a result, the infiltrated composite characterized by a very homogeneous distribution of the mullite phase and the absence of unfilled pores was obtained. A detailed description of the process mentioned above was presented in [17]. The critical significance of the interfacial properties of the composite components on the essential properties of the composite causes their recognition and understanding to be of the paramount importance [20]. The aim of the present paper is to investigate the phasetransformation process during the infiltration of aluminium into porous mullite preforms.
2.2. Characterization The ceramic preform and composite samples were characterized using the XRD (PANalytical X'Pert Diffractometer PW 3040/60 equipped with Cu-Kα radiation (λ = 0.15406 nm) in Bragg–Brentano geometry). The preform sample for the transmission electron microscopy (TEM) study was prepared by crushing in an agate mortar and dispersing ethanol suspensions of the obtained powder onto lacy carbon-coated copper grids. The composite sample was prepared by the focused ionbeam milling and in-situ lift-out. TEM investigations were undertaken with a field-emission transmission electron microscope (FEI Titan 80–300 TEM/STEM) with a super twin lens operated at 300 kV and equipped with an annular dark-field detector. The chemical composition was determined in the same apparatus using energy dispersive spectroscopy (EDS). Information about the structure of the investigated materials came from the database maintained under the care of the Mineralogical Society of America and the Mineralogical Association of Canada [21]. The Eje-Z program, available on the University of Cádiz (Spain) server, http://www2.uca.es/dept/cmat_ qinor/catalisis/tem-uca-server.htm, was used for phase identification by nano diffraction, and fast Fourier transform (FFT) obtained for the high-resolution images, [22]. 3. Results and discussion 3.1. XRD analysis Comparison of the X-ray diffractograms of the ceramic preform and composite sample obtained by aluminium infiltration is given in Fig. 1a.
2. Experimental procedure 2.1. Samples preparation During the investigations, the structures of ceramic (mullite) preform before and after infiltration were compared. The ceramic preform was sintered at a temperature of 1300 °C. Its white colour confirms thermal decomposition of the carbon fibres. The preform hardness was 22 HRF. The bulk density of the preform was 1.32 g/cm3. The volume fraction of the pores was equal to 55.9%; the volume fraction of mullite 44.1% (accordance with Archimedes' principle). The composite material was obtained as a result of infiltration when liquid AlSi12 aluminium alloy filled the pores in the ceramic preform. After infiltration, the colour of the preform changed to metallic, and its hardness increased five times up to a value of about 111 HRF.
Fig. 1. (a) X-ray patterns of the ceramic preform (PREFORM) and composite sample after infiltration (COMPOSITE). (b) Magnification of part of the image from Figure (a). Description: C — cristobalite high, M — mullite, Al — aluminium, Si — silicon, Q – quartz.
M. Pawlyta et al. / Materials Characterization 114 (2016) 9–17
11
Table 1 Summary of the crystallographic data of the identified phases. Phase name
Chemical composition
Space group
Reference
Number at the American Mineralogist Crystal Structures Database [21]
Mullite Cristobalite high Quartz Aluminium γ-Al2O3 Silicon Spinel
3Al2O3·2SiO2 SiO2 SiO2 Al Al2O3 Si MgAl2O4
Pbam Fd3m P3221 Fm3m Fm3m Fd3m Fd3m
[1] [34] [35] [36] [37] [36] [30]
1618 17,665 789 11,137 10,553 11,243 1398
In the ceramic preform sample, two phases were identified: cristobalite high and mullite. After infiltration, additional reflections characteristic of aluminium and silicon appeared. The crystallographic data of the identified phases are given in Table 1. The position of the mullite and cristobalite reflections in both diffractograms do not change. In a separate figure (Fig. 1b) magnification of part of both diffractograms in the 2θ angle range between the 20° and 28° are presented. In that range, the cristobalite reflex (111) and mullite reflex, described as (210) in Fig. 1a, are visible. Two observations can be made. First, after infiltration, the intensity of cristobalite reflection significantly decreases. Second, in both diffractograms one can see that the mullite reflex described as (210) is asymmetrical, indicating that it is a superposition of two separate reflections. Precise analysis indicates that their
positions correspond to two interplanar distances (d120 = 3.429 Å, d210 = 3.389 Å). This confirms a difference between the lattice parameters a and b, and consequently the orthorhombic structure of mullite, before and after the infiltration [15]. Taking into account the relationship between the lattice parameters of mullite and the Al content [12], the Al2O3 content was estimated at about 60% mol (3Al2O3·2SiO2, stoichiometric composition). 3.2. TEM study of ceramic preform Images from the transmission electron microscopy of ceramic preform samples are presented in Fig. 2a. Observations carried out by means of scanning transmission electron microscopy (STEM)
Fig. 2. (a) HAADF-STEM image of ceramic preform (scale bar, 100 nm). Exemplary mullite grain is marked A, cristobalite grain B. (b) Elemental maps — Al (orange), O (red) and Si (yellow). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
12
M. Pawlyta et al. / Materials Characterization 114 (2016) 9–17
Fig. 3. (a, b) STEM-BF images of mullite grains obtained for different magnifications (the ceramic preform sample). (c) The result of chemical analysis (atomic content: aluminium — 28% silicon — 13%, oxygen — 69%). (d) The corresponding FFT of the image shown in Figure b.
mode with the high-angle annular dark field (HAADF) detector demonstrated the presence of two components in the studied material, identified as mullite and cristobalite. The components are diverse in chemical composition, as confirmed by the elemental mapping of Al, Si and O (Fig. 2b). The mullite grains contain all the listed elements; by contrast, in cristobalite, only Si and O are present (for example, the cristobalite grain marked B in Fig. 2a is not visible on the aluminium map).
In higher magnification images, one can see that mullite grains are inhomogeneous with respect to contrast (Fig. 3a). Cristobalite grains do not have that feature (Fig. 4a). On that basis, we can state that mullite is characterized by a finely divided structure, consisting of crystallites homogeneous in size and shape. The needle elongated crystallites, about 100 nm long, dominate (Fig. 3a). In the case of crystallites aligned parallel to the microscope's optic axis, the square cross section (about 30–50 nm in size) is visible (Fig. 3b). Chemical analysis confirms the presence of
Fig. 4. (a) STEM-HAADF image of cristobalite grains (the ceramic preform sample, scale bar, 50 nm). B indicates cristobalite grain. (b) The result of chemical analysis (atomic content elements: aluminium — 3%, silicon — 18%, oxygen — 79%).
M. Pawlyta et al. / Materials Characterization 114 (2016) 9–17
13
Fig. 5. (a) HAADF-STEM image of the transition zone between the ceramic and metal (the composite sample after infiltration, scale bar, 1 μm). (b) Maps of the distribution of aluminium (yellow), oxygen (red) and silicon (green). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
three elements (Fig. 3c), with atomic content equal to aluminium — 28%, silicon — 13%, oxygen — 69%, corresponding to the 3Al2O3·2SiO2 phase. From the FFT obtained for the mullite high-resolution image (Fig. 3d) interplanar distances were determined at 7.6 Å, 5.6 Å and 7.3 Å, as well as angles between the crystallographic direction of 44° and 88°. Based on the chemical and diffraction analysis, the observed phase was identified as 3Al2O3·2SiO2 mullite [11,23]. Fig. 4a shows an STEM-HAADF image of cristobalite grains (marked B), confirmed by the morphology (uniform contrast), and the result of the analysis of the chemical composition (Fig. 4b). 3.3. TEM studies of the composite Fig. 5a depicts the structure of the alloy matrix composites reinforced by the mullite porous preform. Mapping of the chemical composition of the aluminium, oxygen, and silicon was carried out in selected parts (Fig. 5b). Based on the results obtained, one can indicate both components of the composite. On the left side, there is a ceramic skeleton. The skeleton is made of mullite, which can be identified by its characteristic morphology, described earlier in the article (finely divided structure, consisting of homogeneous needle crystallites, similar in size and shape). Cristobalite grains are not visible. They should
be recognizable based on a uniform contrast. Noticeable is the presence of pores that are irregular in shape, of a size of up to about 1 μm. Their morphology indicates that there is no carbon fibre residue, but empty space after the cristobalite grains, which has got decomposed in the infiltration process. Confirmation of this hypothesis is the occasional presence of small areas with high silicon content; an example is the distribution of silicon in the place marked as 1 on the map (Fig. 5b). Moreover, in Fig. 5a four characteristic points of the analysed area are indicated: 1 – a fragment of the ceramic skeleton with increased content of Si (magnification, and the result of chemical analysis in Fig. 6a, b); 2 — a fragment of the ceramic skeleton with typical chemical composition of mullite (Fig. 6c, d); 3 — the transition zone between the ceramic (ceramic skeleton) and the metal (Fig. 6e, f); and 4 — the aluminium matrix. The grain in the ceramic skeleton with increased Si content was recognized as the remaining part of the cristobalite grain, not completely decomposed during infiltration. 3.4. Mullite structure After the process of infiltration, mullite is the only component of the ceramic skeleton. Fig. 7a shows the image of a single crystallite mullite in [001] orientation together with a magnified view of the selected
14
M. Pawlyta et al. / Materials Characterization 114 (2016) 9–17
Fig. 6. (a, c, e) HAADF-STEM images of the magnified parts indicated by numbers 1–3 in Fig. 5a and corresponding to the results of the chemical composition. (b, d, f) (the composite sample after infiltration).
fragment (Fig. 7b). The cross section of the crystallite has a shape close to a square with sides of 30–40 nm. As it is a high-resolution image, it is possible to determine the FFT, which is equivalent to the selected area electron diffraction (SAED), and then the phase identification. FFT can be obtained from an area with a diameter equal to a few nanometres. In the case of SAED, the smallest diameter is about 170 nm. The resulting transform (Fig. 7c) was compared with the simulated electron diffraction of mullite in the direction [001] (Fig. 7d), giving good agreement. Fig. 7a and b were obtained by using the HAADF detector. Its advantage is the efficient use of electrons scattered over larger angles [24]. Thanks to this, interpretation is ambiguous (the diffraction contrast is eliminated nearly in full), and it is not necessary to undertake advanced simulations (unavoidable in the past [23,25]) to interpret the result correctly. Fig. 7e shows the model of the four mullite elementary cells in the orientation
[001]. Aluminium and silicon atoms, having a higher value of Z, strongly scatter electrons, which in the HAADF image should be seen as brighter points. One can see that in the magnification of the mullite crystal (Fig. 7f) where, for convenience, groups of atoms which strongly scatter the electrons are marked. For a similar process, when the liquid metal (aluminium) fills the pores in the ceramic (mullite) preforms it was observed that the mullite phase was transformed to α-alumina and an aluminaaluminium composite was formed [26]. The probable cause that this phenomenon did not occur in our case was the short infiltration time (30 s, in the described process it was 1 h). As the gas pressure during infiltration was relatively low (up to 3 MPa), the amorphization of the ceramic skeleton did not occur, as it requires much higher pressure values [27].
M. Pawlyta et al. / Materials Characterization 114 (2016) 9–17
15
Fig. 7. (a) The high-resolution STEM-HAADF image of mullite grain oriented in [001]. (b) Enlarged fragment of (a). (c) The corresponding FFT pattern. (d) Simulation of the electron diffraction of mullite in the direction [001] (ߛ — forbidden reflections, space group absence. (e) Model of four mullite elementary cells: black — Al atoms, blue — Al/Si atoms, yellow — O atoms. (f) Magnification of mullite crystal from (a) with characteristic elements depicted from the mullite model (Fig. 7e). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
3.5. Transition zone Between the ceramic skeleton and the metal matrix a distinct transition zone is visible with a width of about 200 nm (Fig. 8a and already in Fig. 6e). The chemical composition (Al, O) of the transition zone is different from the two adjacent components of the composite (Fig. 6f). The transition zone is polycrystalline, as confirmed by the SAED diffraction obtained with the lowest available selective aperture. Fig. 6b shows the Fourier transform calculated from highresolution transmission electron microscopy (HRTEM) image of the
region of the transition zone. The size of the crystallites (specified as an area where the structure and the crystallographic orientation, assessed by Fourier transform, is unchanged) ranges from several to tens of nanometres. Based on the results of the chemical and diffraction analysis phase, a transition zone can be identified as γAl2O3 (characteristic of a high heating rate and subsequent crystallization [28]). The change in the Al, O and Si contents in the interface range indicates that it was formed by the atomic diffusion, which played an essential role in wetting the mullite particles with aluminium [29].
16
M. Pawlyta et al. / Materials Characterization 114 (2016) 9–17
Fig. 8. (a) HAADF-STEM image of the transition zone between the ceramic and the metal (scale bar, 200 nm). (b) The corresponding FFT pattern calculated for HRTEM image of the transition zone (the composite sample after infiltration)
The presence of the relatively large crystallites is detected in the outer part of the transition zone (as viewed from the side of the mullite). This phenomenon is characteristic of the whole sample and is present in all the areas, including areas not shown in this paper. The crystallite size exceeds 100 nm. Moreover, they are wider than the mullite needle crystals. An example of one of the largest crystals is shown in Fig. 9a. The chemical composition of the large crystallites differs from the chemical composition of the transition zone by the
additional presence of magnesium and silicon (Fig. 9b). The atomic content of aluminium is approximately 31%, silicon — 7%, oxygen — 59%, magnesium — 3%. The crystal structure is confirmed by the highresolution imaging. From the computed Fourier transforms of the high-magnification image (Fig. 9c) the lattice spacing (4.1 Å; 4.8 Å; 3.0 Å) and the angles between the crystallographic directions (54° and 34°) were determined. The described phase was identified as a spinel of the chemical formula (Mg,Si)Al2O4, and then compared with the
Fig. 9. (a) HAADF-BF image of crystallite (Mg,Si)Al2O4, which is typical of the outer part of the transition zone between the ceramic and the metal. (b) The result of chemical analysis (atomic content: aluminium — 31%, silicon — 7% oxygen — 59%, magnesium — 3%). (c) The corresponding FFT pattern calculated for HRTEM image of the crystal shown in Figure c. (d) Simulation of electron diffraction for MgAl2O4 phase in the direction [110] (the composite sample after infiltration).
M. Pawlyta et al. / Materials Characterization 114 (2016) 9–17
results of computer simulations of the MgAl2O4 phase [30,31] (Fig. 9d), giving good agreement. It is worth highlighting that MgAl2O4 spinel formation was confirmed in the mullite ceramics formed by sintering at relatively low temperatures (1500–1550 °C) from the recycled fly ash and bauxite with MgO addition as the raw materials [32]. The nanostructured MgAl2O4 spinels can be synthesized even at the significantly lower temperatures by a direct conversion process from γ-Al2O3 [33]. 4. Conclusions Based on the XRD and high-resolution electron microscopy investigations the phase process transformations during the infiltration of aluminium into porous mullite preforms can be characterized. During infiltration, the liquid metal effectively fills empty spaces (pores) and forms a three-dimensional network of metal in the ceramic. Cristobalite is decomposed almost in full. In the areas of its occurrence, there are new pores only in the ceramic grains. Mullite, which was formed from halloysite during annealing, crystallizes in the orthorhombic space group Pbam, with stoichiometric composition (3Al2O3·2SiO2 or Al2·25Si.75O4.871). The unit cell parameters are: a = 7.543 Å; b = 7.692 Å; c = 2.884 Å; α = β = γ = 90°. During infiltration, the mullite structure does not change. The composite components are tightly connected. Between the ceramic and the metal, a transition zone, having a thickness of about 200 nm, is formed. That zone is nanocrystalline, identified as γ-Al2O3. In the outer parts of the transition zone, there is a new phase, formed by the reaction of diffusing the product from cristobalite decomposition to the aluminium alloy matrix. The new phase was identified as (Mg,Si)Al2O4. Acknowledgement The project was financed by the National Science Centre grant according to the DEC-2011/03/B/ST8/06,076 decision number. References [1] H. Schneider, J. Schreuer, B. Hildmann, Structure and properties of mullite—a review, J. Eur. Ceram. Soc. 28 (2) (2008) 329–344, http://dx.doi.org/10.1016/j.jeurceramsoc. 2007.03.017. [2] I.A. Aksay, D.M. Dabbs, M. Sarikaya, Mullite for structural, electronic, and optical applications, J. Am. Ceram. Soc. 74 (10) (1991) 2343–2358, http://dx.doi.org/10.1111/j. 1151-2916.1991.tb06768.x. [3] R. Sadanaga, M. Tokonami, Y. Takeuchi, The structure of mullite, 2Al2O3. SiO2, and relationship with the structures of sillimanite and andalusite, Acta Crystallogr. 15 (1) (1962) 65–68, http://dx.doi.org/10.1107/S0365110X62000134. [4] W.E. Cameron, Composition and cell dimensions of mullite, Am. Ceram. Soc. Bull. 56 (11) (1977) 1003–1007. [5] H. Schneider, S. Komarneni (Eds.), Mullite, John Wiley & Sons, 2006. [6] R.X. Fischer, H. Schneider, D. Voll, Formation of aluminum-rich 9:1 mullite and its transformation to low alumina mullite upon heating, J. Eur. Ceram. Soc. 16 (2) (1996) 109–113, http://dx.doi.org/10.1016/0955-2219(95)00139-5. [7] R.X. Fischer, H. Schneider, The mullite-type family of crystal structures, in: H. Schneider, S. Komarneni (Eds.), Mullite, Wiley-VCH, Weinheim 2005, pp. 1–46. [8] M. Schmucker, H. Schneider, Mullite-type gels and glasses, in: H. Schneider, S. Komarneni (Eds.), Mullite, Wiley-VCH, Weinheim 2005, pp. 93–128. [9] S. Aryal, P. Rulis, W.Y. Ching, Mechanical properties and electronic structure of mullite phases using first-principles modeling, J. Am. Ceram. Soc. 95 (7) (2012) 2075–2088, http://dx.doi.org/10.1111/j.1551-2916.2012.05172.x. [10] J. Ylä-Jääski, H.U. Nissen, Investigation of superstructures in mullite by highresolution electron microscopy and electron diffraction, Phys. Chem. Miner. 10 (2) (1983) 47–54, http://dx.doi.org/10.1007/BF00309584. [11] D. Balzar, H. Ledbetter, Crystal structure and compressibility of 3: 2 mullite, Am. Mineral. 78 (11−12) (1993) 1192–1196.
17
[12] R.X. Fischer, H. Schneider, M. Schmucker, Crystal structure of Al-rich mullite, Am. Mineral. 79 (1994) 983–990. [13] J. Ossaka, Tetragonal mullite-like phase from co-precipitated gels, Nature 191 (1961) 1000–1001. [14] H. Schneider, T. Rymon-Lipinski, Occurrence of pseudotetragonal mullite, J. Am. Ceram. Soc. 71 (1988) C162–C164. [15] B.R. Johnson, W.M. Kriven, J. Schneider, Crystal structure development during devitrification of quenched mullite, J. Eur. Ceram. Soc. 21 (14) (2001) 2541–2562, http:// dx.doi.org/10.1016/S0955-2219(01)00268-0. [16] E.C. Hammel, O.R. Ighodaro, O.I. Okoli, Processing and properties of advanced porous ceramics: an application based review, Ceram. Int. 40 (10) (2014) 15351–15370, http://dx.doi.org/10.1016/j.ceramint.2014.06.095. [17] B. Tomiczek, M. Kujawa, G. Matula, M. Kremzer, T. Tański, L.A. Dobrzański, Aluminium AlSi12 alloy matrix composites reinforced by mullite porous preforms, Mater. Werkst. 46 (4–5) (2015) 368–376, http://dx.doi.org/10.1002/mawe. 201500411. [18] J. Ouyang, Z. Zhou, Y. Zhang, H. Yang, High morphological stability and structural transition of halloysite (Hunan, China) in heat treatment, Appl. Clay Sci. 101 (2014) 16–22, http://dx.doi.org/10.1016/j.clay.2014.08.010. [19] A. Boczkowska, P. Chabera, A.J. Dolata, M. Dyzia, A. Oziębło, Porous ceramic-metal composites obtained by infiltration methods, Metalurgija 52 (3) (2013) 345–348. [20] J.S. Moya, S. Lopez-Esteban, C. Pecharromán, The challenge of ceramic/metal microcomposites and nanocomposites, Prog. Mater. Sci. 52 (7) (2007) 1017–1090, http://dx.doi.org/10.1016/j.pmatsci.2006.09.003. [21] R.T. Downs, M. Hall-Wallace, The American Mineralogist Crystal Structure Database, Am. Mineral. 88 (2003) 247–250. [22] S. Bernal, F.J. Botana, J.J. Calvino, C. Lopez-Cartes, J.A. Perez-Omil, J.M. RodrıguezIzquierdo, The interpretation of HREM images of supported metal catalysts using image simulation: profile view images, Ultramicroscopy 72 (3) (1998) 135–164, http://dx.doi.org/10.1016/S0304-3991(98)00009-6. [23] T. Epicier, M.A. O'Keefe, G. Thomas, Atomic imaging of 3:2 mullite, Acta Crystallogr. A: Found. Crystallogr. 46 (12) (1990) 948–962, http://dx.doi.org/10.1107/ S0108767390001118. [24] R. Brydson (Ed.) Aberration-corrected Analytical Electron Microscopy, vol. 3, John Wiley & Sons, 2011. [25] T. Epicier, Benefits of high-resolution electron microscopy for the structural characterization of mullites, J. Am. Ceram. Soc. 74 (10) (1991) 2359–2366, http:// dx.doi.org/10.1111/j.1151-2916.1991.tb06769.x. [26] R. Soundararajan, G. Kuhn, R. Atisivan, S. Bose, A. Bandyopadhyay, Processing of mullite–aluminum composites, J. Am. Ceram. Soc. 84 (3) (2001) 509–513, http:// dx.doi.org/10.1111/j.1151-2916.2001.tb00691.x. [27] P.E. Kalita, H. Schneider, K. Lipinska, S. Sinogeikin, O.A. Hemmers, A. Cornelius, Highpressure behavior of mullite: an X-ray diffraction investigation, J. Am. Ceram. Soc. 96 (5) (2013) 1635–1642, http://dx.doi.org/10.1111/jace.12191. [28] G. Di Girolamo, C. Blasi, L. Pilloni, M. Schioppa, Microstructural and thermal properties of plasma sprayed mullite coatings, Ceram. Int. 36 (4) (2010) 1389–1395, http://dx.doi.org/10.1016/j.ceramint.2010.02.007. [29] H. Zhang, N. Maljkovic, B.S. Mitchell, Structure and interfacial properties of nanocrystalline aluminum/mullite composites, Mater. Sci. Eng. A 326 (2) (2002) 317–323, http://dx.doi.org/10.1016/S0921-5093(01)01500-3. [30] R.C. Peterson, G.A. Lager, R.L. Hitterman, A time-of-flight neutron powder diffraction study of Mg-Al2O4 at temperatures up to 1273 K, Am. Mineral. 76 (9–10) (1991) 1455–1458. [31] H. Sieber, D. Hesse, S. Senz, J. Heydenreich, X. Pan, TEM investigations of spinelforming Solid state reactions: mechanism, film orientation, and interface structure during MgAl2O4 formation on MgO (001) and Al2O3 (1 1.2) single crystal substrates, Z. Anorg. Allg. Chem. 622 (10) (1996) 1658–1666, http://dx.doi.org/10. 1002/zaac.19966221007. [32] Y. Dong, S. Hampshire, J.E. Zhou, Z. Ji, J. Wang, G. Meng, Sintering and characterization of fly ash based mullite with MgO addition, J. Eur. Ceram. Soc. 31 (5) (2011) 687–695, http://dx.doi.org/10.1016/j.jeurceramsoc.2010.12.012. [33] J.F. Al-Sharab, F. Cosandey, A. Singhal, G. Skandan, J. Bentley, TEM characterization of nanostructured MgAl2O4 synthesized by a direct conversion process from γ-Al2O3, J. Am. Ceram. Soc. 89 (7) (2006) 2279–2285, http://dx.doi.org/10.1111/j.1551-2916. 2006.01020.x. [34] R.W. Wyckoff, IX. Die Kristallstruktur von β-Cristobalit SiO2 (bei hohen Temperaturen stabile Form), Z. Krist.-Cryst. Mater. 62 (1) (1925) 189–200. [35] L. Levien, C.T. Prewitt, D.J. Weidner, Structure and elastic properties of quartz at pressure, Am. Mineral. 65 (9–10) (1980) 920–930. [36] R.W. Wyckoff, Cubic closest packed, ccp. structure, Cryst. Struct. 1 (1963) 7–83. [37] E.J.W. Verwey, The structure of the electrolytical oxide layer on aluminium, Z. Krist.Cryst. Mater. 91 (1) (1935) 317–320.
Contents lists available at ScienceDirect
Materials Characterization journal homepage: www.elsevier.com/locate/matchar
Transmission electron microscopy observations on phase transformations during aluminium/mullite composites formation by gas pressure infiltration M. Pawlyta a,⁎, B. Tomiczek a, L.A. Dobrzański a, M. Kujawa a, B. Bierska-Piech b a b
Silesian University of Technology, Institute of Engineering Materials and Biomaterials, Konarskiego 18A, 44-100 Gliwice, Poland Silesian Centre for Education and Interdisciplinary Research, 75 Pułku Piechoty 1A, 41-500 Chorzów, Poland
a r t i c l e
i n f o
Article history: Received 10 September 2015 Received in revised form 10 February 2016 Accepted 11 February 2016 Available online 13 February 2016 Keywords: Mullite Composite Interface Electron microscopy X-ray diffraction
a b s t r a c t The porous ceramic preforms were manufactured using the powder metallurgy technique. First, the start-up material (halloysite with the addition of carbon fibres as the pore-forming agent) was slowly heated to 800 °C and then sintered at 1300 °C. Degradation of the carbon fibres enabled the open canals to form. At the end of the sintering process, the porous ceramic material consisting mainly of two phases (mullite and cristobalite) was formed, without any residual carbon content. During infiltration, the liquid metal filled the empty spaces (pores) effectively and formed the three-dimensional network of metal in the ceramic. The cristobalite was almost entirely decomposed. In the areas of its previous occurrence, there are new pores, only in the ceramic grains. The mullite, which was formed from halloysite during annealing, crystallized in the Pbam orthorhombic space group, with the (3Al2O3·2SiO2) stoichiometric composition. The mullite structure does not change during the infiltration. The composite components are tightly connected. A transition zone between the ceramics and the metal, having the thickness of about 200 nm, was formed. The nanocrystalline zone, identified as γ-Al2O3, was formed by diffusing the product of the cristobalite decomposition into the aluminium alloy matrix. There is an additional, new phase, identified as (Mg,Si)Al2O4 in the outer parts of the transition zone. © 2016 Elsevier Inc. All rights reserved.
1. Introduction Mullite is one of the most important ceramic materials. Because of the formation conditions (high temperature, low pressure), it is rarely formed in natural conditions. Nevertheless, its importance in traditional and advanced ceramics (including electronic, optical, and hightemperature structural applications) is huge and is still increasing [1, 2]. Mullite consists of many phases, which differ in aluminium, oxygen and silicon content [3]. The general composition of mullite can be collectively described as Al4 + 2xSi2 − 2xO10 − x (0 b x ≤ 1) [4]. The first of this group is sillimanite Al2SiO5, for which x = 0. The structure of all mullite phases can be described as modified sillimanite where aluminium atoms are exchanged for silicon atoms while overall charge neutrality is maintained [5]. The charge neutrality condition causes part of the oxygen atoms to be removed, and vacancies appear in these places. The parameter x gives the number of oxygen vacancies per unit cell. Theoretically, all mullite compositions with x between zero and one are possible, but so far only phases with x between 0.2 and 0.9 have been observed [6]. Most frequently described are [7,8]: a. “Sinter-mullite” with stoichiometric composition (3Al2O3·2SiO2) obtained as a result of heat treatment, typically via solid-state
⁎ Corresponding author. E-mail address: [email protected] (M. Pawlyta).
http://dx.doi.org/10.1016/j.matchar.2016.02.003 1044-5803/© 2016 Elsevier Inc. All rights reserved.
reactions. In this case x = 0.25 corresponding to about 72 wt.% of Al2O3, b. “Chemical-mullite”, also obtained as a result of heat treatment but with no stoichiometric composition. Composition of these mullites strongly depends on the starting materials and the temperature treatment, c. “Fused-mullite” (2Al2O3·SiO2) produced by crystallizing aluminosilicate melts. In this case x = 0.40 corresponding to about 78 wt.% Al2O3.
The structure and chemical composition change depending on the process formation temperature and velocity. The mechanical properties and electronic structures of mullite can be calculated based on its structural models [9]. Therefore, the knowledge and understanding of mullite are so much important. Nevertheless, the structure and chemical composition change depending on the process formation temperature and velocity. All mullites are crystalline in the orthorhombic space group Pbam [10,11] and do not transform to a tetragonal form [12]. If mullites are obtained from homogenous, powdered precursors in low temperatures (below 1200 °C), mullite with high Al2O3 (70 mol% Al2O3) forms first and without regard to the initial material composition. Mullite crystals formed this way have an a lattice close to the b lattice (“mullite with pseudotetragonal metric”) [13,14]. The degree of diversification between a
10
M. Pawlyta et al. / Materials Characterization 114 (2016) 9–17
and b can be rated by analysis of the shape of the X-ray diffraction (XRD) reflex that appears at the angle of about 2θ ≈ 26° (for CuKα radiation). For a = b, it should be a single, symmetrical reflex. The larger the difference between a and b, the more asymmetric it becomes, and finally it splits into two separate reflections: (120) and (210), corresponding to two interplanar distances (d120 = 0.3428 nm, d210 = 0.3390 nm) [15]. Incorporation of SiO2 into the mullite structure causes differentiation in the lattice sizes and a corresponding structure change. Mullite can be used in one of the following forms: monolith mullite ceramics, mullite coatings and mullite matrix in composites. In particular, investigations in the field of mullite composites have been intensively conducted in the last two decades. The main aim of this research focused on improving the toughness. Despite previous intense efforts, no significant breakthrough has been achieved. The unresolved problem is the inhomogeneous dispersion of mullite grains in the metal matrix. A new, promising approach is the application of mullite in the form of continuous fibres or three-dimensional skeletons (preforms), which are then filled with molten metal, usually aluminium or its alloys [1,16]. A successful example of that approach is the obtaining of a threedimensional network of metal in the ceramic [17]. In this case, porous ceramic preforms were manufactured using a powder metallurgy technique. The start-up material, halloysite powder (delivered by NaturalNano) with the addition of 30 wt % carbon fibres (SGL Carbon Group) as a pore-forming agent, was subjected to dry mixing in a Fritsch Pulverisette 6 ball mill for 10 min. The prepared powder mixture was compacted in a steel die with an inside diameter of 30 mm on a LabEcon600 platen press at 100 MPa pressure and then sintered at 1300 °C (heating rate ~2.5 K/min). Thermal degradation of the carbon fibres facilitated the formation of open porosity. Sintering of nanotubular halloysite up to 1300 °C induces the formation of a mullite phase [18]. At the end of the sintering process, porous crystalline mullite was formed without any residual carbon content. In order to fabricate the composite materials reinforced by the previously prepared porous preforms, the gas pressure infiltration technique was used. The infiltration process was performed in a specially designed autoclave [19]. The AlSi12 alloy was heated to 800 °C in silicon carbide crucible. The molten aluminium alloy was infiltrated into the pores of the preform under the influence of nitrogen pressure (3 MPa for 30 s). As a result, the infiltrated composite characterized by a very homogeneous distribution of the mullite phase and the absence of unfilled pores was obtained. A detailed description of the process mentioned above was presented in [17]. The critical significance of the interfacial properties of the composite components on the essential properties of the composite causes their recognition and understanding to be of the paramount importance [20]. The aim of the present paper is to investigate the phasetransformation process during the infiltration of aluminium into porous mullite preforms.
2.2. Characterization The ceramic preform and composite samples were characterized using the XRD (PANalytical X'Pert Diffractometer PW 3040/60 equipped with Cu-Kα radiation (λ = 0.15406 nm) in Bragg–Brentano geometry). The preform sample for the transmission electron microscopy (TEM) study was prepared by crushing in an agate mortar and dispersing ethanol suspensions of the obtained powder onto lacy carbon-coated copper grids. The composite sample was prepared by the focused ionbeam milling and in-situ lift-out. TEM investigations were undertaken with a field-emission transmission electron microscope (FEI Titan 80–300 TEM/STEM) with a super twin lens operated at 300 kV and equipped with an annular dark-field detector. The chemical composition was determined in the same apparatus using energy dispersive spectroscopy (EDS). Information about the structure of the investigated materials came from the database maintained under the care of the Mineralogical Society of America and the Mineralogical Association of Canada [21]. The Eje-Z program, available on the University of Cádiz (Spain) server, http://www2.uca.es/dept/cmat_ qinor/catalisis/tem-uca-server.htm, was used for phase identification by nano diffraction, and fast Fourier transform (FFT) obtained for the high-resolution images, [22]. 3. Results and discussion 3.1. XRD analysis Comparison of the X-ray diffractograms of the ceramic preform and composite sample obtained by aluminium infiltration is given in Fig. 1a.
2. Experimental procedure 2.1. Samples preparation During the investigations, the structures of ceramic (mullite) preform before and after infiltration were compared. The ceramic preform was sintered at a temperature of 1300 °C. Its white colour confirms thermal decomposition of the carbon fibres. The preform hardness was 22 HRF. The bulk density of the preform was 1.32 g/cm3. The volume fraction of the pores was equal to 55.9%; the volume fraction of mullite 44.1% (accordance with Archimedes' principle). The composite material was obtained as a result of infiltration when liquid AlSi12 aluminium alloy filled the pores in the ceramic preform. After infiltration, the colour of the preform changed to metallic, and its hardness increased five times up to a value of about 111 HRF.
Fig. 1. (a) X-ray patterns of the ceramic preform (PREFORM) and composite sample after infiltration (COMPOSITE). (b) Magnification of part of the image from Figure (a). Description: C — cristobalite high, M — mullite, Al — aluminium, Si — silicon, Q – quartz.
M. Pawlyta et al. / Materials Characterization 114 (2016) 9–17
11
Table 1 Summary of the crystallographic data of the identified phases. Phase name
Chemical composition
Space group
Reference
Number at the American Mineralogist Crystal Structures Database [21]
Mullite Cristobalite high Quartz Aluminium γ-Al2O3 Silicon Spinel
3Al2O3·2SiO2 SiO2 SiO2 Al Al2O3 Si MgAl2O4
Pbam Fd3m P3221 Fm3m Fm3m Fd3m Fd3m
[1] [34] [35] [36] [37] [36] [30]
1618 17,665 789 11,137 10,553 11,243 1398
In the ceramic preform sample, two phases were identified: cristobalite high and mullite. After infiltration, additional reflections characteristic of aluminium and silicon appeared. The crystallographic data of the identified phases are given in Table 1. The position of the mullite and cristobalite reflections in both diffractograms do not change. In a separate figure (Fig. 1b) magnification of part of both diffractograms in the 2θ angle range between the 20° and 28° are presented. In that range, the cristobalite reflex (111) and mullite reflex, described as (210) in Fig. 1a, are visible. Two observations can be made. First, after infiltration, the intensity of cristobalite reflection significantly decreases. Second, in both diffractograms one can see that the mullite reflex described as (210) is asymmetrical, indicating that it is a superposition of two separate reflections. Precise analysis indicates that their
positions correspond to two interplanar distances (d120 = 3.429 Å, d210 = 3.389 Å). This confirms a difference between the lattice parameters a and b, and consequently the orthorhombic structure of mullite, before and after the infiltration [15]. Taking into account the relationship between the lattice parameters of mullite and the Al content [12], the Al2O3 content was estimated at about 60% mol (3Al2O3·2SiO2, stoichiometric composition). 3.2. TEM study of ceramic preform Images from the transmission electron microscopy of ceramic preform samples are presented in Fig. 2a. Observations carried out by means of scanning transmission electron microscopy (STEM)
Fig. 2. (a) HAADF-STEM image of ceramic preform (scale bar, 100 nm). Exemplary mullite grain is marked A, cristobalite grain B. (b) Elemental maps — Al (orange), O (red) and Si (yellow). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
12
M. Pawlyta et al. / Materials Characterization 114 (2016) 9–17
Fig. 3. (a, b) STEM-BF images of mullite grains obtained for different magnifications (the ceramic preform sample). (c) The result of chemical analysis (atomic content: aluminium — 28% silicon — 13%, oxygen — 69%). (d) The corresponding FFT of the image shown in Figure b.
mode with the high-angle annular dark field (HAADF) detector demonstrated the presence of two components in the studied material, identified as mullite and cristobalite. The components are diverse in chemical composition, as confirmed by the elemental mapping of Al, Si and O (Fig. 2b). The mullite grains contain all the listed elements; by contrast, in cristobalite, only Si and O are present (for example, the cristobalite grain marked B in Fig. 2a is not visible on the aluminium map).
In higher magnification images, one can see that mullite grains are inhomogeneous with respect to contrast (Fig. 3a). Cristobalite grains do not have that feature (Fig. 4a). On that basis, we can state that mullite is characterized by a finely divided structure, consisting of crystallites homogeneous in size and shape. The needle elongated crystallites, about 100 nm long, dominate (Fig. 3a). In the case of crystallites aligned parallel to the microscope's optic axis, the square cross section (about 30–50 nm in size) is visible (Fig. 3b). Chemical analysis confirms the presence of
Fig. 4. (a) STEM-HAADF image of cristobalite grains (the ceramic preform sample, scale bar, 50 nm). B indicates cristobalite grain. (b) The result of chemical analysis (atomic content elements: aluminium — 3%, silicon — 18%, oxygen — 79%).
M. Pawlyta et al. / Materials Characterization 114 (2016) 9–17
13
Fig. 5. (a) HAADF-STEM image of the transition zone between the ceramic and metal (the composite sample after infiltration, scale bar, 1 μm). (b) Maps of the distribution of aluminium (yellow), oxygen (red) and silicon (green). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
three elements (Fig. 3c), with atomic content equal to aluminium — 28%, silicon — 13%, oxygen — 69%, corresponding to the 3Al2O3·2SiO2 phase. From the FFT obtained for the mullite high-resolution image (Fig. 3d) interplanar distances were determined at 7.6 Å, 5.6 Å and 7.3 Å, as well as angles between the crystallographic direction of 44° and 88°. Based on the chemical and diffraction analysis, the observed phase was identified as 3Al2O3·2SiO2 mullite [11,23]. Fig. 4a shows an STEM-HAADF image of cristobalite grains (marked B), confirmed by the morphology (uniform contrast), and the result of the analysis of the chemical composition (Fig. 4b). 3.3. TEM studies of the composite Fig. 5a depicts the structure of the alloy matrix composites reinforced by the mullite porous preform. Mapping of the chemical composition of the aluminium, oxygen, and silicon was carried out in selected parts (Fig. 5b). Based on the results obtained, one can indicate both components of the composite. On the left side, there is a ceramic skeleton. The skeleton is made of mullite, which can be identified by its characteristic morphology, described earlier in the article (finely divided structure, consisting of homogeneous needle crystallites, similar in size and shape). Cristobalite grains are not visible. They should
be recognizable based on a uniform contrast. Noticeable is the presence of pores that are irregular in shape, of a size of up to about 1 μm. Their morphology indicates that there is no carbon fibre residue, but empty space after the cristobalite grains, which has got decomposed in the infiltration process. Confirmation of this hypothesis is the occasional presence of small areas with high silicon content; an example is the distribution of silicon in the place marked as 1 on the map (Fig. 5b). Moreover, in Fig. 5a four characteristic points of the analysed area are indicated: 1 – a fragment of the ceramic skeleton with increased content of Si (magnification, and the result of chemical analysis in Fig. 6a, b); 2 — a fragment of the ceramic skeleton with typical chemical composition of mullite (Fig. 6c, d); 3 — the transition zone between the ceramic (ceramic skeleton) and the metal (Fig. 6e, f); and 4 — the aluminium matrix. The grain in the ceramic skeleton with increased Si content was recognized as the remaining part of the cristobalite grain, not completely decomposed during infiltration. 3.4. Mullite structure After the process of infiltration, mullite is the only component of the ceramic skeleton. Fig. 7a shows the image of a single crystallite mullite in [001] orientation together with a magnified view of the selected
14
M. Pawlyta et al. / Materials Characterization 114 (2016) 9–17
Fig. 6. (a, c, e) HAADF-STEM images of the magnified parts indicated by numbers 1–3 in Fig. 5a and corresponding to the results of the chemical composition. (b, d, f) (the composite sample after infiltration).
fragment (Fig. 7b). The cross section of the crystallite has a shape close to a square with sides of 30–40 nm. As it is a high-resolution image, it is possible to determine the FFT, which is equivalent to the selected area electron diffraction (SAED), and then the phase identification. FFT can be obtained from an area with a diameter equal to a few nanometres. In the case of SAED, the smallest diameter is about 170 nm. The resulting transform (Fig. 7c) was compared with the simulated electron diffraction of mullite in the direction [001] (Fig. 7d), giving good agreement. Fig. 7a and b were obtained by using the HAADF detector. Its advantage is the efficient use of electrons scattered over larger angles [24]. Thanks to this, interpretation is ambiguous (the diffraction contrast is eliminated nearly in full), and it is not necessary to undertake advanced simulations (unavoidable in the past [23,25]) to interpret the result correctly. Fig. 7e shows the model of the four mullite elementary cells in the orientation
[001]. Aluminium and silicon atoms, having a higher value of Z, strongly scatter electrons, which in the HAADF image should be seen as brighter points. One can see that in the magnification of the mullite crystal (Fig. 7f) where, for convenience, groups of atoms which strongly scatter the electrons are marked. For a similar process, when the liquid metal (aluminium) fills the pores in the ceramic (mullite) preforms it was observed that the mullite phase was transformed to α-alumina and an aluminaaluminium composite was formed [26]. The probable cause that this phenomenon did not occur in our case was the short infiltration time (30 s, in the described process it was 1 h). As the gas pressure during infiltration was relatively low (up to 3 MPa), the amorphization of the ceramic skeleton did not occur, as it requires much higher pressure values [27].
M. Pawlyta et al. / Materials Characterization 114 (2016) 9–17
15
Fig. 7. (a) The high-resolution STEM-HAADF image of mullite grain oriented in [001]. (b) Enlarged fragment of (a). (c) The corresponding FFT pattern. (d) Simulation of the electron diffraction of mullite in the direction [001] (ߛ — forbidden reflections, space group absence. (e) Model of four mullite elementary cells: black — Al atoms, blue — Al/Si atoms, yellow — O atoms. (f) Magnification of mullite crystal from (a) with characteristic elements depicted from the mullite model (Fig. 7e). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
3.5. Transition zone Between the ceramic skeleton and the metal matrix a distinct transition zone is visible with a width of about 200 nm (Fig. 8a and already in Fig. 6e). The chemical composition (Al, O) of the transition zone is different from the two adjacent components of the composite (Fig. 6f). The transition zone is polycrystalline, as confirmed by the SAED diffraction obtained with the lowest available selective aperture. Fig. 6b shows the Fourier transform calculated from highresolution transmission electron microscopy (HRTEM) image of the
region of the transition zone. The size of the crystallites (specified as an area where the structure and the crystallographic orientation, assessed by Fourier transform, is unchanged) ranges from several to tens of nanometres. Based on the results of the chemical and diffraction analysis phase, a transition zone can be identified as γAl2O3 (characteristic of a high heating rate and subsequent crystallization [28]). The change in the Al, O and Si contents in the interface range indicates that it was formed by the atomic diffusion, which played an essential role in wetting the mullite particles with aluminium [29].
16
M. Pawlyta et al. / Materials Characterization 114 (2016) 9–17
Fig. 8. (a) HAADF-STEM image of the transition zone between the ceramic and the metal (scale bar, 200 nm). (b) The corresponding FFT pattern calculated for HRTEM image of the transition zone (the composite sample after infiltration)
The presence of the relatively large crystallites is detected in the outer part of the transition zone (as viewed from the side of the mullite). This phenomenon is characteristic of the whole sample and is present in all the areas, including areas not shown in this paper. The crystallite size exceeds 100 nm. Moreover, they are wider than the mullite needle crystals. An example of one of the largest crystals is shown in Fig. 9a. The chemical composition of the large crystallites differs from the chemical composition of the transition zone by the
additional presence of magnesium and silicon (Fig. 9b). The atomic content of aluminium is approximately 31%, silicon — 7%, oxygen — 59%, magnesium — 3%. The crystal structure is confirmed by the highresolution imaging. From the computed Fourier transforms of the high-magnification image (Fig. 9c) the lattice spacing (4.1 Å; 4.8 Å; 3.0 Å) and the angles between the crystallographic directions (54° and 34°) were determined. The described phase was identified as a spinel of the chemical formula (Mg,Si)Al2O4, and then compared with the
Fig. 9. (a) HAADF-BF image of crystallite (Mg,Si)Al2O4, which is typical of the outer part of the transition zone between the ceramic and the metal. (b) The result of chemical analysis (atomic content: aluminium — 31%, silicon — 7% oxygen — 59%, magnesium — 3%). (c) The corresponding FFT pattern calculated for HRTEM image of the crystal shown in Figure c. (d) Simulation of electron diffraction for MgAl2O4 phase in the direction [110] (the composite sample after infiltration).
M. Pawlyta et al. / Materials Characterization 114 (2016) 9–17
results of computer simulations of the MgAl2O4 phase [30,31] (Fig. 9d), giving good agreement. It is worth highlighting that MgAl2O4 spinel formation was confirmed in the mullite ceramics formed by sintering at relatively low temperatures (1500–1550 °C) from the recycled fly ash and bauxite with MgO addition as the raw materials [32]. The nanostructured MgAl2O4 spinels can be synthesized even at the significantly lower temperatures by a direct conversion process from γ-Al2O3 [33]. 4. Conclusions Based on the XRD and high-resolution electron microscopy investigations the phase process transformations during the infiltration of aluminium into porous mullite preforms can be characterized. During infiltration, the liquid metal effectively fills empty spaces (pores) and forms a three-dimensional network of metal in the ceramic. Cristobalite is decomposed almost in full. In the areas of its occurrence, there are new pores only in the ceramic grains. Mullite, which was formed from halloysite during annealing, crystallizes in the orthorhombic space group Pbam, with stoichiometric composition (3Al2O3·2SiO2 or Al2·25Si.75O4.871). The unit cell parameters are: a = 7.543 Å; b = 7.692 Å; c = 2.884 Å; α = β = γ = 90°. During infiltration, the mullite structure does not change. The composite components are tightly connected. Between the ceramic and the metal, a transition zone, having a thickness of about 200 nm, is formed. That zone is nanocrystalline, identified as γ-Al2O3. In the outer parts of the transition zone, there is a new phase, formed by the reaction of diffusing the product from cristobalite decomposition to the aluminium alloy matrix. The new phase was identified as (Mg,Si)Al2O4. Acknowledgement The project was financed by the National Science Centre grant according to the DEC-2011/03/B/ST8/06,076 decision number. References [1] H. Schneider, J. Schreuer, B. Hildmann, Structure and properties of mullite—a review, J. Eur. Ceram. Soc. 28 (2) (2008) 329–344, http://dx.doi.org/10.1016/j.jeurceramsoc. 2007.03.017. [2] I.A. Aksay, D.M. Dabbs, M. Sarikaya, Mullite for structural, electronic, and optical applications, J. Am. Ceram. Soc. 74 (10) (1991) 2343–2358, http://dx.doi.org/10.1111/j. 1151-2916.1991.tb06768.x. [3] R. Sadanaga, M. Tokonami, Y. Takeuchi, The structure of mullite, 2Al2O3. SiO2, and relationship with the structures of sillimanite and andalusite, Acta Crystallogr. 15 (1) (1962) 65–68, http://dx.doi.org/10.1107/S0365110X62000134. [4] W.E. Cameron, Composition and cell dimensions of mullite, Am. Ceram. Soc. Bull. 56 (11) (1977) 1003–1007. [5] H. Schneider, S. Komarneni (Eds.), Mullite, John Wiley & Sons, 2006. [6] R.X. Fischer, H. Schneider, D. Voll, Formation of aluminum-rich 9:1 mullite and its transformation to low alumina mullite upon heating, J. Eur. Ceram. Soc. 16 (2) (1996) 109–113, http://dx.doi.org/10.1016/0955-2219(95)00139-5. [7] R.X. Fischer, H. Schneider, The mullite-type family of crystal structures, in: H. Schneider, S. Komarneni (Eds.), Mullite, Wiley-VCH, Weinheim 2005, pp. 1–46. [8] M. Schmucker, H. Schneider, Mullite-type gels and glasses, in: H. Schneider, S. Komarneni (Eds.), Mullite, Wiley-VCH, Weinheim 2005, pp. 93–128. [9] S. Aryal, P. Rulis, W.Y. Ching, Mechanical properties and electronic structure of mullite phases using first-principles modeling, J. Am. Ceram. Soc. 95 (7) (2012) 2075–2088, http://dx.doi.org/10.1111/j.1551-2916.2012.05172.x. [10] J. Ylä-Jääski, H.U. Nissen, Investigation of superstructures in mullite by highresolution electron microscopy and electron diffraction, Phys. Chem. Miner. 10 (2) (1983) 47–54, http://dx.doi.org/10.1007/BF00309584. [11] D. Balzar, H. Ledbetter, Crystal structure and compressibility of 3: 2 mullite, Am. Mineral. 78 (11−12) (1993) 1192–1196.
17
[12] R.X. Fischer, H. Schneider, M. Schmucker, Crystal structure of Al-rich mullite, Am. Mineral. 79 (1994) 983–990. [13] J. Ossaka, Tetragonal mullite-like phase from co-precipitated gels, Nature 191 (1961) 1000–1001. [14] H. Schneider, T. Rymon-Lipinski, Occurrence of pseudotetragonal mullite, J. Am. Ceram. Soc. 71 (1988) C162–C164. [15] B.R. Johnson, W.M. Kriven, J. Schneider, Crystal structure development during devitrification of quenched mullite, J. Eur. Ceram. Soc. 21 (14) (2001) 2541–2562, http:// dx.doi.org/10.1016/S0955-2219(01)00268-0. [16] E.C. Hammel, O.R. Ighodaro, O.I. Okoli, Processing and properties of advanced porous ceramics: an application based review, Ceram. Int. 40 (10) (2014) 15351–15370, http://dx.doi.org/10.1016/j.ceramint.2014.06.095. [17] B. Tomiczek, M. Kujawa, G. Matula, M. Kremzer, T. Tański, L.A. Dobrzański, Aluminium AlSi12 alloy matrix composites reinforced by mullite porous preforms, Mater. Werkst. 46 (4–5) (2015) 368–376, http://dx.doi.org/10.1002/mawe. 201500411. [18] J. Ouyang, Z. Zhou, Y. Zhang, H. Yang, High morphological stability and structural transition of halloysite (Hunan, China) in heat treatment, Appl. Clay Sci. 101 (2014) 16–22, http://dx.doi.org/10.1016/j.clay.2014.08.010. [19] A. Boczkowska, P. Chabera, A.J. Dolata, M. Dyzia, A. Oziębło, Porous ceramic-metal composites obtained by infiltration methods, Metalurgija 52 (3) (2013) 345–348. [20] J.S. Moya, S. Lopez-Esteban, C. Pecharromán, The challenge of ceramic/metal microcomposites and nanocomposites, Prog. Mater. Sci. 52 (7) (2007) 1017–1090, http://dx.doi.org/10.1016/j.pmatsci.2006.09.003. [21] R.T. Downs, M. Hall-Wallace, The American Mineralogist Crystal Structure Database, Am. Mineral. 88 (2003) 247–250. [22] S. Bernal, F.J. Botana, J.J. Calvino, C. Lopez-Cartes, J.A. Perez-Omil, J.M. RodrıguezIzquierdo, The interpretation of HREM images of supported metal catalysts using image simulation: profile view images, Ultramicroscopy 72 (3) (1998) 135–164, http://dx.doi.org/10.1016/S0304-3991(98)00009-6. [23] T. Epicier, M.A. O'Keefe, G. Thomas, Atomic imaging of 3:2 mullite, Acta Crystallogr. A: Found. Crystallogr. 46 (12) (1990) 948–962, http://dx.doi.org/10.1107/ S0108767390001118. [24] R. Brydson (Ed.) Aberration-corrected Analytical Electron Microscopy, vol. 3, John Wiley & Sons, 2011. [25] T. Epicier, Benefits of high-resolution electron microscopy for the structural characterization of mullites, J. Am. Ceram. Soc. 74 (10) (1991) 2359–2366, http:// dx.doi.org/10.1111/j.1151-2916.1991.tb06769.x. [26] R. Soundararajan, G. Kuhn, R. Atisivan, S. Bose, A. Bandyopadhyay, Processing of mullite–aluminum composites, J. Am. Ceram. Soc. 84 (3) (2001) 509–513, http:// dx.doi.org/10.1111/j.1151-2916.2001.tb00691.x. [27] P.E. Kalita, H. Schneider, K. Lipinska, S. Sinogeikin, O.A. Hemmers, A. Cornelius, Highpressure behavior of mullite: an X-ray diffraction investigation, J. Am. Ceram. Soc. 96 (5) (2013) 1635–1642, http://dx.doi.org/10.1111/jace.12191. [28] G. Di Girolamo, C. Blasi, L. Pilloni, M. Schioppa, Microstructural and thermal properties of plasma sprayed mullite coatings, Ceram. Int. 36 (4) (2010) 1389–1395, http://dx.doi.org/10.1016/j.ceramint.2010.02.007. [29] H. Zhang, N. Maljkovic, B.S. Mitchell, Structure and interfacial properties of nanocrystalline aluminum/mullite composites, Mater. Sci. Eng. A 326 (2) (2002) 317–323, http://dx.doi.org/10.1016/S0921-5093(01)01500-3. [30] R.C. Peterson, G.A. Lager, R.L. Hitterman, A time-of-flight neutron powder diffraction study of Mg-Al2O4 at temperatures up to 1273 K, Am. Mineral. 76 (9–10) (1991) 1455–1458. [31] H. Sieber, D. Hesse, S. Senz, J. Heydenreich, X. Pan, TEM investigations of spinelforming Solid state reactions: mechanism, film orientation, and interface structure during MgAl2O4 formation on MgO (001) and Al2O3 (1 1.2) single crystal substrates, Z. Anorg. Allg. Chem. 622 (10) (1996) 1658–1666, http://dx.doi.org/10. 1002/zaac.19966221007. [32] Y. Dong, S. Hampshire, J.E. Zhou, Z. Ji, J. Wang, G. Meng, Sintering and characterization of fly ash based mullite with MgO addition, J. Eur. Ceram. Soc. 31 (5) (2011) 687–695, http://dx.doi.org/10.1016/j.jeurceramsoc.2010.12.012. [33] J.F. Al-Sharab, F. Cosandey, A. Singhal, G. Skandan, J. Bentley, TEM characterization of nanostructured MgAl2O4 synthesized by a direct conversion process from γ-Al2O3, J. Am. Ceram. Soc. 89 (7) (2006) 2279–2285, http://dx.doi.org/10.1111/j.1551-2916. 2006.01020.x. [34] R.W. Wyckoff, IX. Die Kristallstruktur von β-Cristobalit SiO2 (bei hohen Temperaturen stabile Form), Z. Krist.-Cryst. Mater. 62 (1) (1925) 189–200. [35] L. Levien, C.T. Prewitt, D.J. Weidner, Structure and elastic properties of quartz at pressure, Am. Mineral. 65 (9–10) (1980) 920–930. [36] R.W. Wyckoff, Cubic closest packed, ccp. structure, Cryst. Struct. 1 (1963) 7–83. [37] E.J.W. Verwey, The structure of the electrolytical oxide layer on aluminium, Z. Krist.Cryst. Mater. 91 (1) (1935) 317–320.