- Email: [email protected]
TEM study of the early stages of Ti2AlC oxidation at 900 °C
Available online at www.sciencedirect.com
Scripta Materialia 67 (2012) 830–833 www.elsevier.com/locate/scriptamat
TEM study of the early stages of Ti2AlC oxidation at 900 °C Bai Cui,⇑ Daniel Doni Jayaseelan and William Edward Lee Centre for Advanced Structural Ceramics (CASC) and Department of Materials, Imperial College London, South Kensington Campus, London SW7 2AZ, UK Received 21 June 2012; accepted 30 July 2012 Available online 4 August 2012
The cross-sectional microstructure on Ti2AlC oxidized for 1 h at 900 °C was investigated using focused ion beam/transmission electron microscopy. An Al-depleted intermediate layer formed between the oxide scale and the substrate that consisted of nonstoichiometric Ti2AlxC nanocrystallites. The oxide scale consisted of a TiO2-rich outer layer and an Al2O3-rich inner layer. Ti2AlC has been theoretically predicted to preserve its crystal structure down to a sub-stoichiometry of Ti2Al0.5C, and such a Ti2AlxC phase was observed in this study. Ó 2012 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved. Keywords: MAX phases; Oxidation; Microstructure; Ti2AlC
Ti2AlC is a promising material for various hightemperature applications, such as electrical heating elements [1], hot corrosion-resistant coatings [2], high-temperature electrodes [3] and cladding materials in leadcooled fast nuclear reactors [4]. Ti2AlC is a typical MAX phase [3,5] in the family of ternary compounds, with general formula Mn+1AXn, where M is a transition metal, A is a group IIIA or IVA element, X is C or N, and n is from 1 to 3. A major difficulty with non-oxide ceramics such as MAX phases is their tendency to oxidize in air. Understanding the oxidation behaviour of Ti2AlC is critical for its application at elevated temperature in oxidizing atmospheres, and is also essential for the control of crack healing ability via surface oxidation [6,7]. The oxidation behaviour of Ti2AlC has been investigated under isothermal conditions at high temperatures (1000– 2000 °C) [8–10] and at intermediate temperatures (500– 1000 °C) [11]. Parabolic oxidation kinetics have been observed by various researchers, and the growth of oxide scale on Ti2AlC is believed to be caused by inward diffusion of O2 and outward diffusion of Al3+ and Ti4+ [12,13]. The high-temperature oxidation resistance of Ti2AlC involves selective oxidation of Al, leading to formation of continuous and protective Al2O3-rich scales [8–10,14]. After isothermal oxidation from 1100 to 1300 °C, a dense and continuous inner a-Al2O3 layer
⇑ Corresponding
author. [email protected]
Tel.:
+1
217
333
4120; e-mail:
and a discontinuous outer TiO2 layer are formed [8– 10]. At 1400 °C, a mixed outer layer of TiO2 and Al2TiO5 and a cracked a-Al2O3 inner layer are formed [8,14]. The cracked oxide scale provides channels for rapid ingress of oxygen to the body, leading to heavy oxidation above 1400 °C [8]. There has been disagreement over the detail of the intermediate-temperature oxidation behaviour of Ti2AlC, particularly that at 900 °C [8,15,16]. In our previous publication [8], X-ray diffraction (XRD), secondary electron imaging (SEI) and energy-dispersive spectroscopy (EDS) in a scanning electron microscope revealed that rutile TiO2 was mainly present after 1 h of oxidation at 900 °C in air. Based on cross-sectional focused ion beam (FIB)/scanning electron microscopy (SEM) observation and thermodynamic arguments, Yang et al. [15] argue that a continuous Al2O3 layer (250 nm thick) forms in the oxide scale on Ti2AlC after 1 h of oxidation at 900 °C along with some large TiO2 grains dispersed in the Al2O3 scale. Wang and Zhou [11] investigated the isothermal oxidation behaviour of Ti2AlC at 500–900 °C by XRD, Raman spectroscopy and SEM, identifying transitional polymorphs of Al2O3 (c-, d-, h- and a-) and TiO2 (anatase and rutile) in the oxide scale. After 20 h of oxidation at 900 °C in air, the oxide scale formed is rutile TiO2 and a-Al2O3 [11]. Reasons for these different observations likely include different as-made ceramic microstructures (such as grain size, porosity and impurity contents), and different oxidation temperatures and times used. These results
1359-6462/$ - see front matter Ó 2012 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.scriptamat.2012.07.045
B. Cui et al. / Scripta Materialia 67 (2012) 830–833
do, however, demonstrate that the Ti2AlC oxidation behaviour at 900 °C is not fully understood. In all these studies, the oxide scale formed at 900 °C is thin (<1 lm) and nanosized grains are formed on the surface [8,11,15]. To fully understand the early stages of Ti2AlC oxidation at 900 °C, cross-sectional transmission electron microscopy (TEM), EDS and electron diffraction are necessary since only TEM investigations can provide all the structural, phase and crystallographic information of the materials down to atomistic levels [17]. The current work presents a detailed FIB/TEM examination of cross-sectional microstructures on Ti2AlC oxidized for 1 h at 900 °C. Ti2AlC samples were prepared by uniaxial hot-pressing Ti, Al and graphite starting materials at 1400 °C for 1 h at 30 MPa in flowing Ar [18], and were supplied by Prof. Yanchun Zhou (Shenyang National Laboratory for Materials Science, Chinese Academy of Sciences). The as-received samples are predominantly single-phase Ti2AlC ceramics as determined by XRD analysis, with densities of 4.10 g cm 3 (99.8% theoretical) measured by the Archimedes method. The size of the lamellar grains in the Ti2AlC ceramics is 50–100 lm long, 3–10 lm thick and 10–20 lm wide. Ti2AlC samples for oxidation experiments were cut by diamond saw to a size of 4 4 4 mm3. The surfaces were ground to 1200 grit and polished to 1 lm, followed by degreasing in acetone. During oxidation experiments, samples were placed on a Pt crucible and heated to temperature at a rate of 20 °C min 1 in an open hearth furnace. Isothermal oxidation runs were performed for 1 h at 900 °C in air before samples were removed and air cooled. Sections, 100 nm thick, were then prepared for TEM using the in situ lift-out technique on a dual-beam FIB instrument (FEI Helios 600 NanoLab, FEI Company, Acht, Eindhoven, The Netherlands). During FIB section preparation, a Pt protective layer was deposited to protect the feature of interest from Ga+ milling, then a crosssection was in situ lifted-out and attached to a support grid that fits into a TEM specimen holder. TEM was carried out with an electron microscope operated at 200 kV (JEM 2000FX, JEOL Ltd., Tokyo, Japan) and a high-resolution electron microscope operated at 200 kV (JEM 2010, JEOL Ltd., Tokyo, Japan). Microstructures were examined in the TEM using bright-field (BF) and selected area electron diffraction (SAED) crystallographic analysis (operated using a double-tilt holder). The aperture size used for the selected area diffraction patterns is 200–300 nm. The diffraction patterns were solved using the SingleCrystal software (CrystalMaker Software Ltd., Yarnton, Oxfordshire, UK). EDS chemical analysis in the JEM 2000 was performed using an ultrathin window detector (Oxford Instruments INCA, Oxford, UK). An electron-beam probe of <10 nm diameter was obtained in this instrument for EDS chemical analysis. Figure 1a shows typical plan view surface morphology of the oxide scale formed after 1 h of oxidation at 900 °C. The EDS analysis indicates that the small (<1 lm) grains were composed of titanium and aluminium oxides. The rectangular frame in Figure 1a shows the region where a Pt protective layer was deposited and then the cross-sectional lamina was in situ lifted-
831
Figure 1. (a) Plan view of the surface morphology (SEI taken from FIB) of Ti2AlC oxidized for 1 h at 900 °C. The rectangular frame indicates the region where a Pt protective layer was deposited and then the cross-sectional lamina was lifted out. (b) Low-magnification BFTEM image showing a cross-section of Ti2AlC oxidized for 1 h at 900 °C. A: Ti2AlC substrate; B: intermediate layer; C: oxide scale. The arrowed region is a thin bent layer caused by Ga+ milling during FIB.
out for TEM observations. Figure 1b is a low-magnification BF-TEM image of the cross-section of Ti2AlC oxidized for 1 h at 900 °C. The arrowed region in Figure 1b is a thin bent layer caused by Ga+ milling during FIB. As revealed by EDS and SAED, the cross-section consisted of three parts (from the left to right): Ti2AlC substrate (A), intermediate layer (B) and oxide scale (C). An enlarged view of the A/B interface is shown in Figure 2a. A SAED pattern from A (Fig. 2b) indexed as Ti2AlC [0 0 0 1]. A ring pattern from B (Fig. 2c) also indexed as polycrystalline Ti2AlC structure (space group P63/mmc, ICDD 29-0095). EDS from A (not shown) suggests that the Ti/Al ratio is close to 2. EDS from B (not shown), however, indicates that it is close to 4. The combined SAED and EDS results indicate the intermediate layer B is likely to be nonstoichiometric Ti2AlxC (x < 1). High-resolution TEM (HRTEM; Fig. 2d) revealed nanocrystallites (<20 nm) distributed in the intermediate layer B. A typical fringe periodicity is 1.30 nm (arrowed in Fig. 2d), coinciding with the the-
Figure 2. (a) An enlarged view of the A/B interface in Figure 1b. A: Ti2AlC substrate; B: intermediate layer. (b) SAED pattern from A, indexed as Ti2AlC [0 0 0 1]. (c) SAED pattern from B, indexed as polycrystalline Ti2AlxC. (d) HRTEM image of intermediate layer B, revealing Ti2AlxC nanocrystallites (<20 nm).
832
B. Cui et al. / Scripta Materialia 67 (2012) 830–833
Figure 3. An enlarged view of the B/C interface in Figure 1b. B: intermediate Ti2AlxC layer; C: oxide scale; C1: TiO2-rich layer; C2: Al2O3-rich layer. (b) SAED pattern from C1. (c) SAED pattern from C2.
oretical lattice spacing of Ti2AlC (0 0 0 1) planes (d = 1.36 nm, ICDD 29-0095). A theoretical investigation of the phase stability of Ti2AlC containing Al vacancies predicted that Ti2AlC may contain Al vacancies down to a sub-stoichiometry of Ti2Al0.5C [19]. When x < 0.5, Ti2AlxC tends to decompose to twinned Ti2C and Al [19]. In the layered structure of Ti2AlC, Al monolayers intercalate between twinned Ti2C layers, with the Al layer as the mirror plane [3,5]. The thermodynamically stable Ti–C–Ti slabs with strong covalent bonds are separated by softer Ti– Al–Ti slabs with weaker bonds [3,5]. Due to the weaker Ti–Al bonds, Al atoms can diffuse out, leaving Al vacancies. Theoretical investigation has demonstrated that the energy barriers involved in the vacancy migration in Ti2AlC are 0.83, 2.38 and 3.00 eV for VAl-, VTi- and VC-assisted diffusion, respectively [20]. These results suggest that vacancy-mediated diffusion in Ti2AlC is most energetically favourable for VAl. However, this research is one of the first experimental observations of Ti2AlxC. Song et al. [21] studied the ablation behaviour of Ti2AlC ceramics under an oxyacetylene flame at a temperature up to 3000 °C and observed a porous subsurface layer containing Ti2Al1 xC (x < 0.5) and TiCxOy (x + y 6 1). Zhang et al. [22] investigated the phase stability of a similar MAX phase, Ti3AlC2, in a Cu matrix, and revealed that Ti3AlC2 retains its structure under partial loss of Al at 950 °C (although they did not quantify the extent of the loss). The depletion of Al results in highly defective Ti3AlC2 which finally decomposes and transforms into cubic TiCx. The Al-depleted intermediate layer between the oxide scale and the substrate (i.e. layer B in Figs. 1b and 2a) has also been observed in other studies of Ti2AlC oxidation. Rao et al. [10] investigated the initial oxide scales of Ti2AlC after 3 min of oxidation at 1200 °C and observed a Ti-rich (Al-depleted) intermediate layer (100 nm) that consists mostly of TiC derived from the transformation of the Ti2AlC matrix due to the depletion of Al atoms. Lin et al. [23] found a thin (50– 100 nm) Al-depleted layer at the Ti2AlC/oxide scale interface after Ti2AlC oxidation at 1200 °C for 10 h. An enlarged view of the B/C interface is shown in Figure 3a. The intermediate layer B was polycrystalline
Ti2AlxC, as discussed above. The oxide scale C consisted of an outer layer, C1 (200–300 nm thick), and a mottled inner layer, C2 (400 nm thick). EDS from layers C1 (not shown) and C2 (not shown) indicates that C1 was a TiO2-rich layer (Ti:Al 2:1) while C2 was an Al2O3rich layer (Al:Ti 2:1). Because the probe size for EDS chemical analysis in this TEM is several nanometres, it appears that both C1 and C2 were a mixture of TiO2 and Al2O3. SAED from the oxide layers C1 (Fig. 3b) and C2 (Fig. 3c) reveals similar diffuse ring patterns. It is difficult to distinguish between broad ring patterns from amorphous material and those from small nanocrystals [24], but further comparison between these two may be informative since the inner ring from region C1 appears to be slightly sharper than in C2 (identical conditions were used when forming both SAED patterns). The mottling in the BF image of region C2 (Fig. 3a) is probably strain contrast arising from strain between the Al2O3-rich oxide nanocrystals in the C2 layer. This contrast is not present in the outer C1 TiO2-rich layer, perhaps because this region is completely amorphous, as suggested by the more diffuse SAED pattern (Fig. 3b). There are two possible reasons for these observations: (i) 1 h at 900 °C may be insufficient time for complete crystallization and growth of rutile TiO2 and a-Al2O3. During Ti2AlC oxidation, many rutile TiO2 (5–10 lm) and a-Al2O3 (<1 lm) crystals form after 1 h at or above 1000 °C [8,9]. (ii) Ga+ ion beam damage in FIB and e beam damage during the TEM may have amorphized the nanocrystals, especially if they are embedded in a glassy matrix [24]; such electron beam damage is more likely in the thinner outer regions of the TEM samples. The fact that the TiO2-rich (Ti:Al 2:1) outer layer C1 is over the Al2O3-rich (Al:Ti 2:1) inner layer C2 may be explained by the higher growth rate of TiO2 than of Al2O3 [6,10] and the outward diffusion of Ti through Al2O3 grain boundaries [23]. The higher growth rate of TiO2 is due to a greater lattice defect concentration in TiO2 [25]. Al2O3 is a fairly stable oxide, and has a growth rate that is more than five orders of magnitude lower than that of TiO2 during the oxidation of Ti–Al
B. Cui et al. / Scripta Materialia 67 (2012) 830–833
alloys [26,27]. In addition, the outward diffusion of Ti through Al2O3 grain boundaries [23] may result in the outgrowth of TiO2 on top of the Al2O3. Therefore, TiO2 outgrows the inner Al2O3 to form a TiO2-rich outer layer over an Al2O3-rich inner layer. Based on these microstructural observations, the cross-sectional microstructure of Ti2AlC oxidized 1 h at 900 °C is composed of Ti2AlC substrate, Ti2AlxC intermediate layer and oxide scale. The intermediate layer formed between the oxide scale and the substrate is an Al-depleted layer consisting of nonstoichiometric Ti2AlxC (x < 1) nanocrystallites (<20 nm), transforming from Ti2AlC due to depletion of Al. The oxide scale consists of a TiO2-rich outer layer and an Al2O3-rich inner layer, formed by the faster outgrowth of TiO2 than of Al2O3. Ti2AlC was theoretically predicted to preserve its crystal structure down to a sub-stoichiometry of Ti2Al0.5C, and such a Ti2AlxC phase has been experimentally observed in this study. Valuable comments and discussions on this manuscript from Prof. Jeff De Hosson (University of Groningen, The Netherlands) are greatly appreciated. B.C. gratefully acknowledges the Lee Family Scholarship for financial support, and thanks Prof. Yanchun Zhou (Shenyang National Laboratory for Materials Science, China) for providing Ti2AlC samples. [1] M. Sundberg, G. Malmqvist, A. Magnusson, T. ElRaghy, Ceram. Int. 30 (2004) 1899. [2] Z.J. Lin, Y.C. Zhou, M.S. Li, J.Y. Wang, J. Eur. Ceram. Soc. 26 (2006) 3871. [3] M.W. Barsoum, Prog. Solid State Chem. 28 (2000) 201. [4] L.A. Barnes, N.L.D. Rago, L. Leibowitz, J. Nuclear Mater. 373 (2008) 424. [5] B. Cui, R. Sa, D.D. Jayaseelan, F. Inam, M.J. Reece, W.E. Lee, Acta Mater. 60 (2012) 1079. [6] H.J. Yang, Y.T. Pei, J.C. Rao, J.T.M. De Hosson, S.B. Li, G.M. Song, Scripta Mater. 65 (2011) 135.
833
[7] G.M. Song, Y.T. Pei, W.G. Sloof, S.B. Li, J.Th.M. De Hosson, S. van der Zwaag, Scripta Mater. 58 (2008) 13. [8] B. Cui, D.D. Jayaseelan, W.E. Lee, Acta Mater. 59 (2011) 4116. [9] X.H. Wang, Y.C. Zhou, Oxid. Met. 59 (2003) 303. [10] J.C. Rao, Y.T. Pei, H.J. Yang, J.T.M. De Hosson, G.M. Song, S.B. Li, Acta Mater. 59 (2011) 5215. [11] X.H. Wang, Y.C. Zhou, J. Mater. Res. 17 (2002) 2974. [12] M.W. Barsoum, N. Tzenov, A. Procopio, T. El-Raghy, M. Ali, J. Electrochem. Soc. 148 (2001) C551. [13] M.W. Barsoum, J. Electrochem. Soc. 148 (2001) C544. [14] J.W. Byeon, J. Liu, M. Hopkins, W. Fischer, N. Garimella, K.B. Park, M.P. Brady, M. Radovic, T. El-Raghy, Y.H. Sohn, Oxid. Met. 68 (2007) 97. [15] H.J. Yang, Y.T. Pei, G.M. Song, J.Th.M. De Hosson, Scripta Mater. 65 (2011) 930. [16] B. Cui, D.D. Jayaseelan, W.E. Lee, Scripta Mater. 65 (2011) 933. [17] W.E. Lee, W.M. Rainforth, Ceramic Microstructures: Property Control by Processing, Chapman & Hall, London, 1994. [18] X.H. Wang, Y.C. Zhou, Int. J. Mater. Res. 93 (2002) 66. [19] J.Y. Wang, Y.C. Zhou, T. Liao, J. Zhang, Z.J. Lin, Scripta Mater. 58 (2008) 227. [20] T. Liao, J.Y. Wang, Y.C. Zhou, Scripta Mater. 59 (2008) 854. [21] G.M. Song, S.B. Li, C.X. Zhao, W.G. Sloof, S. van der Zwaag, Y.T. Pei, J.Th.M. De Hosson, J. Eur. Ceram. Soc 31 (2011) 855. [22] J. Zhang, J.Y. Wang, Y.C. Zhou, Acta Mater. 55 (2007) 4381. [23] Z.J. Lin, M.J. Zhuo, Y.C. Zhou, M.S. Li, J.Y. Wang, J. Am. Ceram. Soc. 89 (2006) 2964. [24] D.B. Williams, C.B. Carter, Transmission Electron Microscopy, Springer Science & Business Media, New York, 2009. [25] D.B. Lee, T.D. Nguyen, Mater. Corros. 59 (2008) 624. [26] M. Yamaguchi, H. Inui, in: R. Darolia, J.J. Lewandowski, C.T. Liu, P.L. Martin, D.B. Miracle, M.V. Nathal (Eds.), Structural Intermetallics, TMS, Warrendale, PA, 1993, pp. 127–142. [27] M.R. Yang, S.K. Wu, Bulletin of the College of Engineering, National Taiwan University, vol. 89, 2003, p. 3.
Scripta Materialia 67 (2012) 830–833 www.elsevier.com/locate/scriptamat
TEM study of the early stages of Ti2AlC oxidation at 900 °C Bai Cui,⇑ Daniel Doni Jayaseelan and William Edward Lee Centre for Advanced Structural Ceramics (CASC) and Department of Materials, Imperial College London, South Kensington Campus, London SW7 2AZ, UK Received 21 June 2012; accepted 30 July 2012 Available online 4 August 2012
The cross-sectional microstructure on Ti2AlC oxidized for 1 h at 900 °C was investigated using focused ion beam/transmission electron microscopy. An Al-depleted intermediate layer formed between the oxide scale and the substrate that consisted of nonstoichiometric Ti2AlxC nanocrystallites. The oxide scale consisted of a TiO2-rich outer layer and an Al2O3-rich inner layer. Ti2AlC has been theoretically predicted to preserve its crystal structure down to a sub-stoichiometry of Ti2Al0.5C, and such a Ti2AlxC phase was observed in this study. Ó 2012 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved. Keywords: MAX phases; Oxidation; Microstructure; Ti2AlC
Ti2AlC is a promising material for various hightemperature applications, such as electrical heating elements [1], hot corrosion-resistant coatings [2], high-temperature electrodes [3] and cladding materials in leadcooled fast nuclear reactors [4]. Ti2AlC is a typical MAX phase [3,5] in the family of ternary compounds, with general formula Mn+1AXn, where M is a transition metal, A is a group IIIA or IVA element, X is C or N, and n is from 1 to 3. A major difficulty with non-oxide ceramics such as MAX phases is their tendency to oxidize in air. Understanding the oxidation behaviour of Ti2AlC is critical for its application at elevated temperature in oxidizing atmospheres, and is also essential for the control of crack healing ability via surface oxidation [6,7]. The oxidation behaviour of Ti2AlC has been investigated under isothermal conditions at high temperatures (1000– 2000 °C) [8–10] and at intermediate temperatures (500– 1000 °C) [11]. Parabolic oxidation kinetics have been observed by various researchers, and the growth of oxide scale on Ti2AlC is believed to be caused by inward diffusion of O2 and outward diffusion of Al3+ and Ti4+ [12,13]. The high-temperature oxidation resistance of Ti2AlC involves selective oxidation of Al, leading to formation of continuous and protective Al2O3-rich scales [8–10,14]. After isothermal oxidation from 1100 to 1300 °C, a dense and continuous inner a-Al2O3 layer
⇑ Corresponding
author. [email protected]
Tel.:
+1
217
333
4120; e-mail:
and a discontinuous outer TiO2 layer are formed [8– 10]. At 1400 °C, a mixed outer layer of TiO2 and Al2TiO5 and a cracked a-Al2O3 inner layer are formed [8,14]. The cracked oxide scale provides channels for rapid ingress of oxygen to the body, leading to heavy oxidation above 1400 °C [8]. There has been disagreement over the detail of the intermediate-temperature oxidation behaviour of Ti2AlC, particularly that at 900 °C [8,15,16]. In our previous publication [8], X-ray diffraction (XRD), secondary electron imaging (SEI) and energy-dispersive spectroscopy (EDS) in a scanning electron microscope revealed that rutile TiO2 was mainly present after 1 h of oxidation at 900 °C in air. Based on cross-sectional focused ion beam (FIB)/scanning electron microscopy (SEM) observation and thermodynamic arguments, Yang et al. [15] argue that a continuous Al2O3 layer (250 nm thick) forms in the oxide scale on Ti2AlC after 1 h of oxidation at 900 °C along with some large TiO2 grains dispersed in the Al2O3 scale. Wang and Zhou [11] investigated the isothermal oxidation behaviour of Ti2AlC at 500–900 °C by XRD, Raman spectroscopy and SEM, identifying transitional polymorphs of Al2O3 (c-, d-, h- and a-) and TiO2 (anatase and rutile) in the oxide scale. After 20 h of oxidation at 900 °C in air, the oxide scale formed is rutile TiO2 and a-Al2O3 [11]. Reasons for these different observations likely include different as-made ceramic microstructures (such as grain size, porosity and impurity contents), and different oxidation temperatures and times used. These results
1359-6462/$ - see front matter Ó 2012 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.scriptamat.2012.07.045
B. Cui et al. / Scripta Materialia 67 (2012) 830–833
do, however, demonstrate that the Ti2AlC oxidation behaviour at 900 °C is not fully understood. In all these studies, the oxide scale formed at 900 °C is thin (<1 lm) and nanosized grains are formed on the surface [8,11,15]. To fully understand the early stages of Ti2AlC oxidation at 900 °C, cross-sectional transmission electron microscopy (TEM), EDS and electron diffraction are necessary since only TEM investigations can provide all the structural, phase and crystallographic information of the materials down to atomistic levels [17]. The current work presents a detailed FIB/TEM examination of cross-sectional microstructures on Ti2AlC oxidized for 1 h at 900 °C. Ti2AlC samples were prepared by uniaxial hot-pressing Ti, Al and graphite starting materials at 1400 °C for 1 h at 30 MPa in flowing Ar [18], and were supplied by Prof. Yanchun Zhou (Shenyang National Laboratory for Materials Science, Chinese Academy of Sciences). The as-received samples are predominantly single-phase Ti2AlC ceramics as determined by XRD analysis, with densities of 4.10 g cm 3 (99.8% theoretical) measured by the Archimedes method. The size of the lamellar grains in the Ti2AlC ceramics is 50–100 lm long, 3–10 lm thick and 10–20 lm wide. Ti2AlC samples for oxidation experiments were cut by diamond saw to a size of 4 4 4 mm3. The surfaces were ground to 1200 grit and polished to 1 lm, followed by degreasing in acetone. During oxidation experiments, samples were placed on a Pt crucible and heated to temperature at a rate of 20 °C min 1 in an open hearth furnace. Isothermal oxidation runs were performed for 1 h at 900 °C in air before samples were removed and air cooled. Sections, 100 nm thick, were then prepared for TEM using the in situ lift-out technique on a dual-beam FIB instrument (FEI Helios 600 NanoLab, FEI Company, Acht, Eindhoven, The Netherlands). During FIB section preparation, a Pt protective layer was deposited to protect the feature of interest from Ga+ milling, then a crosssection was in situ lifted-out and attached to a support grid that fits into a TEM specimen holder. TEM was carried out with an electron microscope operated at 200 kV (JEM 2000FX, JEOL Ltd., Tokyo, Japan) and a high-resolution electron microscope operated at 200 kV (JEM 2010, JEOL Ltd., Tokyo, Japan). Microstructures were examined in the TEM using bright-field (BF) and selected area electron diffraction (SAED) crystallographic analysis (operated using a double-tilt holder). The aperture size used for the selected area diffraction patterns is 200–300 nm. The diffraction patterns were solved using the SingleCrystal software (CrystalMaker Software Ltd., Yarnton, Oxfordshire, UK). EDS chemical analysis in the JEM 2000 was performed using an ultrathin window detector (Oxford Instruments INCA, Oxford, UK). An electron-beam probe of <10 nm diameter was obtained in this instrument for EDS chemical analysis. Figure 1a shows typical plan view surface morphology of the oxide scale formed after 1 h of oxidation at 900 °C. The EDS analysis indicates that the small (<1 lm) grains were composed of titanium and aluminium oxides. The rectangular frame in Figure 1a shows the region where a Pt protective layer was deposited and then the cross-sectional lamina was in situ lifted-
831
Figure 1. (a) Plan view of the surface morphology (SEI taken from FIB) of Ti2AlC oxidized for 1 h at 900 °C. The rectangular frame indicates the region where a Pt protective layer was deposited and then the cross-sectional lamina was lifted out. (b) Low-magnification BFTEM image showing a cross-section of Ti2AlC oxidized for 1 h at 900 °C. A: Ti2AlC substrate; B: intermediate layer; C: oxide scale. The arrowed region is a thin bent layer caused by Ga+ milling during FIB.
out for TEM observations. Figure 1b is a low-magnification BF-TEM image of the cross-section of Ti2AlC oxidized for 1 h at 900 °C. The arrowed region in Figure 1b is a thin bent layer caused by Ga+ milling during FIB. As revealed by EDS and SAED, the cross-section consisted of three parts (from the left to right): Ti2AlC substrate (A), intermediate layer (B) and oxide scale (C). An enlarged view of the A/B interface is shown in Figure 2a. A SAED pattern from A (Fig. 2b) indexed as Ti2AlC [0 0 0 1]. A ring pattern from B (Fig. 2c) also indexed as polycrystalline Ti2AlC structure (space group P63/mmc, ICDD 29-0095). EDS from A (not shown) suggests that the Ti/Al ratio is close to 2. EDS from B (not shown), however, indicates that it is close to 4. The combined SAED and EDS results indicate the intermediate layer B is likely to be nonstoichiometric Ti2AlxC (x < 1). High-resolution TEM (HRTEM; Fig. 2d) revealed nanocrystallites (<20 nm) distributed in the intermediate layer B. A typical fringe periodicity is 1.30 nm (arrowed in Fig. 2d), coinciding with the the-
Figure 2. (a) An enlarged view of the A/B interface in Figure 1b. A: Ti2AlC substrate; B: intermediate layer. (b) SAED pattern from A, indexed as Ti2AlC [0 0 0 1]. (c) SAED pattern from B, indexed as polycrystalline Ti2AlxC. (d) HRTEM image of intermediate layer B, revealing Ti2AlxC nanocrystallites (<20 nm).
832
B. Cui et al. / Scripta Materialia 67 (2012) 830–833
Figure 3. An enlarged view of the B/C interface in Figure 1b. B: intermediate Ti2AlxC layer; C: oxide scale; C1: TiO2-rich layer; C2: Al2O3-rich layer. (b) SAED pattern from C1. (c) SAED pattern from C2.
oretical lattice spacing of Ti2AlC (0 0 0 1) planes (d = 1.36 nm, ICDD 29-0095). A theoretical investigation of the phase stability of Ti2AlC containing Al vacancies predicted that Ti2AlC may contain Al vacancies down to a sub-stoichiometry of Ti2Al0.5C [19]. When x < 0.5, Ti2AlxC tends to decompose to twinned Ti2C and Al [19]. In the layered structure of Ti2AlC, Al monolayers intercalate between twinned Ti2C layers, with the Al layer as the mirror plane [3,5]. The thermodynamically stable Ti–C–Ti slabs with strong covalent bonds are separated by softer Ti– Al–Ti slabs with weaker bonds [3,5]. Due to the weaker Ti–Al bonds, Al atoms can diffuse out, leaving Al vacancies. Theoretical investigation has demonstrated that the energy barriers involved in the vacancy migration in Ti2AlC are 0.83, 2.38 and 3.00 eV for VAl-, VTi- and VC-assisted diffusion, respectively [20]. These results suggest that vacancy-mediated diffusion in Ti2AlC is most energetically favourable for VAl. However, this research is one of the first experimental observations of Ti2AlxC. Song et al. [21] studied the ablation behaviour of Ti2AlC ceramics under an oxyacetylene flame at a temperature up to 3000 °C and observed a porous subsurface layer containing Ti2Al1 xC (x < 0.5) and TiCxOy (x + y 6 1). Zhang et al. [22] investigated the phase stability of a similar MAX phase, Ti3AlC2, in a Cu matrix, and revealed that Ti3AlC2 retains its structure under partial loss of Al at 950 °C (although they did not quantify the extent of the loss). The depletion of Al results in highly defective Ti3AlC2 which finally decomposes and transforms into cubic TiCx. The Al-depleted intermediate layer between the oxide scale and the substrate (i.e. layer B in Figs. 1b and 2a) has also been observed in other studies of Ti2AlC oxidation. Rao et al. [10] investigated the initial oxide scales of Ti2AlC after 3 min of oxidation at 1200 °C and observed a Ti-rich (Al-depleted) intermediate layer (100 nm) that consists mostly of TiC derived from the transformation of the Ti2AlC matrix due to the depletion of Al atoms. Lin et al. [23] found a thin (50– 100 nm) Al-depleted layer at the Ti2AlC/oxide scale interface after Ti2AlC oxidation at 1200 °C for 10 h. An enlarged view of the B/C interface is shown in Figure 3a. The intermediate layer B was polycrystalline
Ti2AlxC, as discussed above. The oxide scale C consisted of an outer layer, C1 (200–300 nm thick), and a mottled inner layer, C2 (400 nm thick). EDS from layers C1 (not shown) and C2 (not shown) indicates that C1 was a TiO2-rich layer (Ti:Al 2:1) while C2 was an Al2O3rich layer (Al:Ti 2:1). Because the probe size for EDS chemical analysis in this TEM is several nanometres, it appears that both C1 and C2 were a mixture of TiO2 and Al2O3. SAED from the oxide layers C1 (Fig. 3b) and C2 (Fig. 3c) reveals similar diffuse ring patterns. It is difficult to distinguish between broad ring patterns from amorphous material and those from small nanocrystals [24], but further comparison between these two may be informative since the inner ring from region C1 appears to be slightly sharper than in C2 (identical conditions were used when forming both SAED patterns). The mottling in the BF image of region C2 (Fig. 3a) is probably strain contrast arising from strain between the Al2O3-rich oxide nanocrystals in the C2 layer. This contrast is not present in the outer C1 TiO2-rich layer, perhaps because this region is completely amorphous, as suggested by the more diffuse SAED pattern (Fig. 3b). There are two possible reasons for these observations: (i) 1 h at 900 °C may be insufficient time for complete crystallization and growth of rutile TiO2 and a-Al2O3. During Ti2AlC oxidation, many rutile TiO2 (5–10 lm) and a-Al2O3 (<1 lm) crystals form after 1 h at or above 1000 °C [8,9]. (ii) Ga+ ion beam damage in FIB and e beam damage during the TEM may have amorphized the nanocrystals, especially if they are embedded in a glassy matrix [24]; such electron beam damage is more likely in the thinner outer regions of the TEM samples. The fact that the TiO2-rich (Ti:Al 2:1) outer layer C1 is over the Al2O3-rich (Al:Ti 2:1) inner layer C2 may be explained by the higher growth rate of TiO2 than of Al2O3 [6,10] and the outward diffusion of Ti through Al2O3 grain boundaries [23]. The higher growth rate of TiO2 is due to a greater lattice defect concentration in TiO2 [25]. Al2O3 is a fairly stable oxide, and has a growth rate that is more than five orders of magnitude lower than that of TiO2 during the oxidation of Ti–Al
B. Cui et al. / Scripta Materialia 67 (2012) 830–833
alloys [26,27]. In addition, the outward diffusion of Ti through Al2O3 grain boundaries [23] may result in the outgrowth of TiO2 on top of the Al2O3. Therefore, TiO2 outgrows the inner Al2O3 to form a TiO2-rich outer layer over an Al2O3-rich inner layer. Based on these microstructural observations, the cross-sectional microstructure of Ti2AlC oxidized 1 h at 900 °C is composed of Ti2AlC substrate, Ti2AlxC intermediate layer and oxide scale. The intermediate layer formed between the oxide scale and the substrate is an Al-depleted layer consisting of nonstoichiometric Ti2AlxC (x < 1) nanocrystallites (<20 nm), transforming from Ti2AlC due to depletion of Al. The oxide scale consists of a TiO2-rich outer layer and an Al2O3-rich inner layer, formed by the faster outgrowth of TiO2 than of Al2O3. Ti2AlC was theoretically predicted to preserve its crystal structure down to a sub-stoichiometry of Ti2Al0.5C, and such a Ti2AlxC phase has been experimentally observed in this study. Valuable comments and discussions on this manuscript from Prof. Jeff De Hosson (University of Groningen, The Netherlands) are greatly appreciated. B.C. gratefully acknowledges the Lee Family Scholarship for financial support, and thanks Prof. Yanchun Zhou (Shenyang National Laboratory for Materials Science, China) for providing Ti2AlC samples. [1] M. Sundberg, G. Malmqvist, A. Magnusson, T. ElRaghy, Ceram. Int. 30 (2004) 1899. [2] Z.J. Lin, Y.C. Zhou, M.S. Li, J.Y. Wang, J. Eur. Ceram. Soc. 26 (2006) 3871. [3] M.W. Barsoum, Prog. Solid State Chem. 28 (2000) 201. [4] L.A. Barnes, N.L.D. Rago, L. Leibowitz, J. Nuclear Mater. 373 (2008) 424. [5] B. Cui, R. Sa, D.D. Jayaseelan, F. Inam, M.J. Reece, W.E. Lee, Acta Mater. 60 (2012) 1079. [6] H.J. Yang, Y.T. Pei, J.C. Rao, J.T.M. De Hosson, S.B. Li, G.M. Song, Scripta Mater. 65 (2011) 135.
833
[7] G.M. Song, Y.T. Pei, W.G. Sloof, S.B. Li, J.Th.M. De Hosson, S. van der Zwaag, Scripta Mater. 58 (2008) 13. [8] B. Cui, D.D. Jayaseelan, W.E. Lee, Acta Mater. 59 (2011) 4116. [9] X.H. Wang, Y.C. Zhou, Oxid. Met. 59 (2003) 303. [10] J.C. Rao, Y.T. Pei, H.J. Yang, J.T.M. De Hosson, G.M. Song, S.B. Li, Acta Mater. 59 (2011) 5215. [11] X.H. Wang, Y.C. Zhou, J. Mater. Res. 17 (2002) 2974. [12] M.W. Barsoum, N. Tzenov, A. Procopio, T. El-Raghy, M. Ali, J. Electrochem. Soc. 148 (2001) C551. [13] M.W. Barsoum, J. Electrochem. Soc. 148 (2001) C544. [14] J.W. Byeon, J. Liu, M. Hopkins, W. Fischer, N. Garimella, K.B. Park, M.P. Brady, M. Radovic, T. El-Raghy, Y.H. Sohn, Oxid. Met. 68 (2007) 97. [15] H.J. Yang, Y.T. Pei, G.M. Song, J.Th.M. De Hosson, Scripta Mater. 65 (2011) 930. [16] B. Cui, D.D. Jayaseelan, W.E. Lee, Scripta Mater. 65 (2011) 933. [17] W.E. Lee, W.M. Rainforth, Ceramic Microstructures: Property Control by Processing, Chapman & Hall, London, 1994. [18] X.H. Wang, Y.C. Zhou, Int. J. Mater. Res. 93 (2002) 66. [19] J.Y. Wang, Y.C. Zhou, T. Liao, J. Zhang, Z.J. Lin, Scripta Mater. 58 (2008) 227. [20] T. Liao, J.Y. Wang, Y.C. Zhou, Scripta Mater. 59 (2008) 854. [21] G.M. Song, S.B. Li, C.X. Zhao, W.G. Sloof, S. van der Zwaag, Y.T. Pei, J.Th.M. De Hosson, J. Eur. Ceram. Soc 31 (2011) 855. [22] J. Zhang, J.Y. Wang, Y.C. Zhou, Acta Mater. 55 (2007) 4381. [23] Z.J. Lin, M.J. Zhuo, Y.C. Zhou, M.S. Li, J.Y. Wang, J. Am. Ceram. Soc. 89 (2006) 2964. [24] D.B. Williams, C.B. Carter, Transmission Electron Microscopy, Springer Science & Business Media, New York, 2009. [25] D.B. Lee, T.D. Nguyen, Mater. Corros. 59 (2008) 624. [26] M. Yamaguchi, H. Inui, in: R. Darolia, J.J. Lewandowski, C.T. Liu, P.L. Martin, D.B. Miracle, M.V. Nathal (Eds.), Structural Intermetallics, TMS, Warrendale, PA, 1993, pp. 127–142. [27] M.R. Yang, S.K. Wu, Bulletin of the College of Engineering, National Taiwan University, vol. 89, 2003, p. 3.