Corrosion Science 43 2001) 1095±1109 www.elsevier.com/locate/corsci Oxidation behaviour of Ti 3 SiC 2 -based ceramic at 900±1300 C in air Zhimei Sun a, Yanchun Zhou a, *, Meishuan Li b a Department of Ceramic and Composite, Institute of Metal Research, Chinese Academy of Sciences, 72 Wenhua Road, Shenyang 110015, People's Republic of China b Institute of Corrosion and Protection of Metals, Chinese Academy of Sciences, 62 Wencui Road, Shenyang 110015, People's Republic of China Received 10 January 2000; accepted 17 July 2000 Abstract The isothermal oxidation behaviour of Ti 3 SiC 2 -based ceramic containing 7 wt.% TiC at 900±1300 C in air has been investigated. The growth of the oxide scales on Ti 3 SiC 2 from 900 C to 1100 C obeyed a parabolic law, whereas at 1200 C and 1300 C, it was a two-step parabolic oxidation process. The scale was composed of an outer layer of coarse-grained TiO 2 rutile), an inner layer of a mixture of ne-grained TiO 2 and SiO 2 tridymite). Furthermore, the oxide scale at 1100 C contained a discontinuous SiO 2 ``barrier'' sandwiched in the outer TiO 2 layer. The scales formed on Ti 3 SiC 2 were dense, adhesive and have good adhesion with the substrate during the cyclic oxidation. Ó 2001 Elsevier Science Ltd. All rights reserved. Keywords: Ti 3 SiC 2 -based ceramic; Oxidation behaviour; Air; Parabolic; SiO 2 barrier 1. Introduction Titanium silicon carbide Ti 3 SiC 2 ), which was identi ed about 30 years ago [1], has recently attracted the attention of physicists as well as material scientists because it o ers a unique combination of the merits of both metals and ceramics. The high * Corresponding author. Tel.: +86-24-2384-3531, ext.: 55180; fax: +86-24-2389-1320. E-mail address: yczhou@imr.ac.cn Y. Zhou). 0010-938X/01/$ - see front matter Ó 2001 Elsevier Science Ltd. All rights reserved. PII: S0010-938X 00)00142-6
1096 Z. Sun et al. / Corrosion Science 43 2001) 1095±1109 melting point and low density [2], high strength at elevated temperatures [3], excellent thermal shock resistance, good thermal conductivity and machinability with conventional tools [2], to name a few, suggest that Ti 3 SiC 2 and Ti 3 SiC 2 -based ceramic materials be good candidates for high temperature applications. As a structural ceramic for high temperature applications, it should embody the properties of oxidation resistance, chemical stability, low volatility, resistance to creep deformation, su cient toughness at ambient temperature, and thermal shock resistance [4]. Therefore, considering Ti 3 SiC 2 as a high temperature structural material, the ability to resist oxidation is the primary criteria. Although the oxidation behaviour of Ti 3 SiC 2 or Ti 3 SiC 2 -based material has been investigated by a number of researchers, there are scatters in the reported data [5±9]. Racault et al. [5] studied the oxidation of Ti 3 SiC 2 powders under owing oxygen. They showed that the oxidation rate for Ti 3 SiC 2 was slower than that for TiC, and the Ti 3 SiC 2 powder almost totally oxidised to TiO 2 rutile) and SiO 2 cristobalite) at the temperatures between 1050 C and 1250 C. Tong et al. [6] investigated the oxidation of monolithic Ti 3 SiC 2 and Ti 3 SiC 2 =SiC composite at 1000 C in owing air for 10 h. They demonstrated that the oxidation resistance of Ti 3 SiC 2 =SiC composite was better than that of monolithic Ti 3 SiC 2. Barsoum and El-Raghy [7] reported parabolic oxidation behaviour of Ti 3 SiC 2 bulk samples consisted of 2 vol% TiC during the oxidation at 900±1400 C in air for 500 min. The calculated activation energy is 320 and 370 kj mol 1. The oxide scale formed at above 1000 C consisted of two layers: an outer layer of pure TiO 2 rutile) and the inner layer of a mixture of SiO 2 and TiO 2. Feng et al. [8] investigated the oxidation of polycrystalline Ti 3 SiC 2 bulk material containing 2 mol% TiC) at temperatures between 800 C and 1100 C for 100 min. They reported a parabolic oxidation in the temperature of 800±950 C and non-parabolic oxidation from 950 C to 1100 C with corresponding calculated activation energies of 137.7 and 312.5 kj mol 1, respectively. Radhakrishnan et al. [9] investigated the oxidation behaviour for polycrystalline Ti 3 SiC 2 containing 2 vol% TiSi 2 ) at 1000 C in air for 50 h. They demonstrated that the oxidation obeyed a paralinear law and Ti 3 SiC 2 was not a good oxidation-resistance material at 1000 C. The discrepancy between the reported oxidation behaviour mentioned above is probably attributed to the di erent impurities and manufacturing techniques, and thus di erent microstructures. In addition, the oxidation conditions varied for different investigators. In the present work, we investigated the isothermal oxidation of Ti 3 SiC 2 ceramic prepared by the in situ hot pressing/solid±liquid reaction process in the temperature range of 900±1300 C up to 20 h. In addition, the cyclic oxidation at 1100 C was also performed to investigate the adherence of the scale with the substrate. The signi cance of this work is as follows. Firstly, at 1100 C, a discontinuous SiO 2 barrier layer sandwiched in the coarse-grained TiO 2 was observed, which inhibited the further oxidation of Ti 3 SiC 2. Secondly, a two-step parabolic oxidation process was revealed at 1200 C and 1300 C. Finally, the di usion-controlling oxidation process is not a simple mode of the inward di usion of oxygen and the simultaneous outward diffusion of titanium and carbon, but a complex one. It might be the inward di usion of oxygen and simultaneous outward di usion of Ti and carbonaceous species CO), as
Z. Sun et al. / Corrosion Science 43 2001) 1095±1109 1097 well as SiO. The results are bene cial to understanding the high-temperature oxidation behaviour of this technically important material. 2. Experimental method 2.1. Specimen preparation The material used in this work was Ti 3 SiC 2, TSC ZS510, which was fabricated by the in situ hot pressing/solid±liquid reaction process [10]. Brie y, the material was made according to the following procedure. Ti, Si, and graphite powders were mixed and milled in a polypropylene jar for 10 h. After ball milling, the mixture was cold pressed in a graphite die with the diameter of 50 mm. The in situ hot pressing/solid± liquid reaction was conducted under a owing argon atmosphere in a furnace using graphite as heating element. The furnace temperature was rapidly reached to 1550 C at the rate of 40 C min 1, then the sample was held at that temperature for 1 h under a pressure of 40 MPa and then cooled down with the pressure removed. The Ti 3 SiC 2 content in the hot pressed material is 93 wt.% calculated by the Rietveld method [11] in CERIUS 2 computational program for material research MSI, USA) and the major impurity is TiC. The Rietveld method involves tting the entire observed X-ray di raction XRD) pattern, step by step, to a pattern calculated using models for the crystal structure and di raction peak pro les, which therefore provides estimates of phase proportions that are less a ected by sample aberrations, such as preferred orientation. All the peaks in the observed di raction pattern contribute to the estimate of phase content, therefore, the accuracy of measurement is greatly improved and sensitivity is signi cantly increased over the traditional methods of quantitative XRD analysis. For the oxidation experiments, rectangular bars with the dimensions of 10 3 4 mm 3 were cut by the electrical-discharge method. The surfaces were ground down to 1000 SiC paper and polished using diamond paste. All the samples were chamfered at the edges to reduce thermal stress. 2.2. Specimen examination The continuous-isothermal-mass-change measurements were performed from 900 C to 1300 C in air for 20 h. The sample was suspended in a thermobalance mtb 10-8, Setakam, France) with a Pt wire when the temperature reached to the required temperature. The cyclic oxidation experiments were performed in a furnace controlled automatically. The samples were hold at 1100 C in a furnace for one hour and then cooled down to room temperature in air for 10 min, which was de ned as one cycle. After the oxidation tests, the samples were characterised by XRD Rigaku D/ max-ra di ractometer, Japan) to determine the phase composition of the oxide scale. The surface morphology was investigated by an S-360 scanning electron
1098 Z. Sun et al. / Corrosion Science 43 2001) 1095±1109 microscope SEM) Cambridge Instruments, Ltd., UK) equipped with an energy dispersive spectroscopy EDS) system. Subsequently, the oxidised samples were cross-sectioned without polishing for further SEM analysis. In SEM analysis, a thin lm of Au was coated on the oxide scale in order to get better observation. 3. Experimental results and discussion 3.1. Oxidation kinetics The results of the isothermal oxidation at 900±1300 C in air for 20 h are summarised in Fig. 1 a). The corresponding square of weight gain per unit area as a function of time is shown in Fig. 1 b). It is seen that the speci c weight gain is relatively small from 900 C to 1100 C and the curve is in agreement with a parabolic rate law. There is almost no di erence for the weight gain between 900 C and 1000 C in Fig. 1 b). The parabolic rate constant, K p, increases from 2:4 10 9 kg 2 m 4 s 1 at 900 C to4 10 7 kg 2 m 4 s 1 at 1100 C, which is in agreement with the previous observations [7,8]. The oxidation kinetics at 1200 C and 1300 C, however, followed a complex law. Carefully analysing the square of speci c weight gain versus time curves for the samples oxidised at 1200 C and 1300 C as was shown in Fig. 1 c)), we obtained a two-step parabolic oxidation process. In the rst oxidation stage, i.e., from 2 to 9 h, the parabolic rate constants at 1200 C and 1300 C are 2:2 10 6 and 9:6 10 6 kg 2 m 4 s 1, respectively. Whereas in the second stage of oxidation, i.e. from 9 to 20 h, the parabolic rate constants are 4 10 6 and 7:7 10 6 kg 2 m 4 s 1, respectively. This two-step oxidation mechanism, or two oxidation stages at 1200 C and 1300 C have not been reported in previous works, which is not surprising considering the relatively short oxidation time in the early works. It is also interesting to note that the parabolic constant changes in di erent ways for the samples oxidised at 1200 C and 1300 C. The parabolic rate constant became lower at the second stage of oxidation when oxidised at 1300 C, which suggests that protect scale or layer that inhibits further oxidation might be formed. Contrary to the phenomenon at 1300 C, the parabolic constant became higher at the second stage of oxidation when exposured at 1200 C. The temperature dependence of the parabolic rate constants for Ti 3 SiC 2 is shown in Fig. 2. The curve of ln K p versus 1=T is near linearity, from which we obtained the activation energy of 350 kj mol 1 for the oxidation process from 900 C to 1300 C. The result is in agreement with the result of Barsoum and El-Raghy [7], 320±370 kj mol 1 and that of Feng et al. [8], 312.5 kj mol 1 in the temperature of 950±1100 C. To study spallation resistance of the oxide scale, the cyclic oxidation was conducted at the condition of 1 h furnace heating and 10 min air cooling. The accumulative time at 1100 C is 88 h. Fig. 3 shows the weight gain as a function of oxidation time for the cyclic oxidation of Ti 3 SiC 2 at 1100 C. No mass loss was detected and the mass gain would simply increase if the experiment were not be interrupted. It is therefore concluded that the oxide scale formed on Ti 3 SiC 2 is adherent and coherent, resistance to thermal cycle.
Z. Sun et al. / Corrosion Science 43 2001) 1095±1109 1099 Fig. 1. a) Weight gain per unit area versus time for Ti 3 SiC 2 oxidised at di erent temperatures. The dependence of the square of the speci c weight gain with time for Ti 3 SiC 2 oxidised b) at di erent temperatures, c) at 1200 C and 1300 C, respectively.
1100 Z. Sun et al. / Corrosion Science 43 2001) 1095±1109 Fig. 2. The temperature dependence of the parabolic rate constant for the oxidation of Ti 3 SiC 2. Fig. 3. Weight gain per unit area versus time for Ti 3 SiC 2 cycled from 1100 C to room temperature 88 times. The total time for the cyclic oxidation is 102.7 h. 3.2. Microstructure observations In Section 3.1, we demonstrated that Ti 3 SiC 2 was a material with good oxidation resistance below 1100 C. To understand the oxidation process, the surface morphology and phase composition of the oxide scales were analysed by SEM and XRD analysis, respectively. Fig. 4 shows the XRD pattern of the surface of the sample oxidised at 900 C for 20 h, where peaks from re ections of TiO 2 rutile), SiO 2 tridymite), and Ti 3 SiC 2 can be seen. The presence of Ti 3 SiC 2 was due to the very
Z. Sun et al. / Corrosion Science 43 2001) 1095±1109 1101 Fig. 4. X-ray di raction pattern from the surface of the sample oxidised at 900 C for 20 h. thin layer of the oxide scale little speci c weight gain was observed). The presence of graphite in Fig. 4 was due to contamination. The oxide products were then determined to be TiO 2 with minor amount of SiO 2. When the temperature increases to 1000 C, no peaks from Ti 3 SiC 2 could be detected not shown here), indicating the thickness of the oxide scale increased with temperature. The main phase of the oxide product was TiO 2 with trace of SiO 2. Above 1100 C, only TiO 2 was observed. Therefore, at the present oxidation conditions, the oxide products on the Ti 3 SiC 2 surface developed from TiO 2 and SiO 2 at low temperature 900 C) to rather pure TiO 2 at high temperatures above 1100 C). To con rm the above results, the surface morphology and the cross-section of the oxide layer was investigated by SEM. According to SEM investigation, the morphology of the surface scale can be divided into two groups: 1) Well-shaped crystals were formed on Ti 3 SiC 2 at 900 C and 1000 C, Fig. 5 a) shows the surface morphology of the sample oxidised at 900 C. The crystal morphology of the sample oxidised at 1000 C is similar to that of 900 C, it is therefore not shown here for briefness. The crystal size increased with increasing oxidation temperature and scale thickness. 2) The Ti 3 SiC 2 samples oxidised from 1100 C to 1300 C revealed rather di erent morphology, which were shown in Fig. 5 b)± d). The scale consisted of two parts, i.e. large crystallites with well-shaped facets and small grains embedded in them. Furthermore, some surfaces of TiO 2 of the samples oxidised at 1100 C and 1200 C were covered with bubbles, a typical morphology at 1100 C was shown in Fig. 5 e). Large crystallite as marked B) contained Ti and O, while both the bubbles as marked A) and the small grains in Fig. 5 b)± d) all contained Ti, Si and O. The corresponding EDS X-ray spectra were shown in Fig. 6 a)± c) respectively. The presence of C was again due to contamination. Comparing the data from XRD and EDS X-ray
1102 Z. Sun et al. / Corrosion Science 43 2001) 1095±1109 Fig. 5. Surface morphology of oxide scale after Ti 3 SiC 2 samples exposure in air at a) 1000 C, b) 1100 C, c) 1200 C, d) 1300 C, and e) at 1100 C showing bubbles enriched in Si on the SiO 2 surface. microanalysis, a discrepancy existed between the two results, i.e. it is free of SiO 2 in the XRD patterns for samples oxidised at 1100 C and above, however, the EDS X-ray microanalysis did con rm the presence of Si. It might be attributed to the fact that the SiO 2 content is very low so that XRD did not detect it.
Z. Sun et al. / Corrosion Science 43 2001) 1095±1109 1103 Fig. 5 continued) Typical back-scattered electron image of the cross-section scale for the samples oxidised from 1000 C to 1300 C in air for 20 h are shown in Fig. 7 a)± d). With the X-ray microanalysis, the coarse-grained bright layers on the left of the micrographs were identi ed as TiO 2 and the dark grey layers in the middle of the micrographs were recognised as mixtures of TiO 2 and SiO 2. The bright parts on the right were Ti 3 SiC 2 matrix. Therefore, the scale generally consists of an outer part of coarsegrained TiO 2 and an inner part of a ne-grained mixture of TiO 2 and SiO 2 as shown in Fig. 7 a), c) and d). Whereas in Fig. 7 b) for the oxide scale produced at 1100 C, a discontinuous sandwich layer of SiO 2 was observed in the outer layer of coarsegrained TiO 2. This sandwiched-in layer formed an intermediate barrier for the diffusion of the oxidation species. This phenomenon is similar to the oxidation of TiAl at 900 C [12], in which an Al 2 O 3 barrier formed in the border region of the outer and the inner layer and good oxidation resistance was observed. This SiO 2 barrier was also helpful to the oxidation resistance of Ti 3 SiC 2. The discontinuous character of SiO 2 barrier might be attributed to the TiC content in the matrix. Thus the oxide layer of the Ti 3 SiC 2 samples oxidised at 1100 C can be described as an outer part of coarse-grained TiO 2 sandwiched by rather pure discontinuous coarse-grained silica
1104 Z. Sun et al. / Corrosion Science 43 2001) 1095±1109 Fig. 6. EDS X-ray spectra from a) large crystallite marked as B in Fig. 5 e), b) bubbles marked as A in Fig. 5 e), c) small grains in Fig. 5 c). barrier, and an inner layer of a mixture of ne-grained TiO 2 and SiO 2. This distribution of oxides was further con rmed by X-ray dot maps shown in Fig. 8.
Z. Sun et al. / Corrosion Science 43 2001) 1095±1109 1105 Fig. 6 continued) 3.3. Transport process in the scale As discussed above, the oxidation of Ti 3 SiC 2 was a di usion-controlled process. The marker experiment in our previous work [13] showed that the oxidation of Ti 3 SiC 2 was mainly controlled by the outward di usion of titanium and inward di usion of oxygen. In the previous works by Barsoum and EI-Raghy [7] and Feng et al. [8], this process was controlled by the inward di usion of oxygen and outward di usion of titanium and carbon, and the Si sublattice was essentially immobile. We here proposed a di erent mode to explain the oxidation of Ti 3 SiC 2 and the formation of SiO 2 barrier and SiO 2 bubbles on the TiO 2 surface. At initial oxidation stage, the in situ oxidation of Ti 3 SiC 2 occurred, TiO 2 and SiO 2 formed, and CO gas left away from the surface. The total reaction is Ti 3 SiC 2 5O 2 g!3tio 2 SiO 2 2CO g Further oxidation continued by the outward di usion of titanium and inward di usion of oxygen. After certain time, a continuous TiO 2 lm formed on the surface, and SiO 2 was wrapped under TiO 2. After transient oxidation, a two layer scale is formed: the inner layer consists of a mixture of TiO 2 and SiO 2, and the outer layer is rather pure TiO 2. Due to the SiO 2 precipitates, the oxygen pressure in the inner layer is much lower than that in the outer layer. Low oxygen pressure supported the formation of SiO rather than SiO 2, and the formed SiO 2 turned into SiO [14], therefore, the oxidation takes place by the following total reaction: 1
1106 Z. Sun et al. / Corrosion Science 43 2001) 1095±1109 Fig. 7. Back scattered image of cross-section of the oxide scales after exposure in air at a) 1000 C, b) 1100 C, c) 1200 C, and d) 1300 C for 20 h. Ti 3 SiC 2 SiO 2 4O 2 g!3tio 2 2SiO g 2CO g 2 The oxidation was then controlled by the outward di usion of titanium, SiO and CO gas, and inward di usion of oxygen at high temperatures, such as at 1100 C and
Z. Sun et al. / Corrosion Science 43 2001) 1095±1109 1107 Fig. 7 continued) Fig. 8. Typical X-ray dot maps of the oxide cross-section of Ti 3 SiC 2 at 1100 C for 20 h. 1200 C. During the outward di usion process, SiO gas will turn into solid SiO 2 where the oxygen pressure is high enough to support the following reaction, such as in the outer coarse-grained TiO 2 layer. SiO g 1 2 O 2 g!sio 2 3 SiO 2 was therefore precipitated in the outer coarse-grained TiO 2 layer, forming a SiO 2 barrier. If the SiO gas di used outward the TiO 2 surface along certain di usion channel, SiO turned into SiO 2 according to Eq. 3) forming bubbles rich in Si on the
1108 Z. Sun et al. / Corrosion Science 43 2001) 1095±1109 TiO 2 surface, as shown in Fig. 5 e). At low temperatures, such as 900 C and 1000 C, the oxidation rate is very low, and the oxide scale is thin, therefore, the concentration gradient of oxygen in the scale was not steep, therefore the oxidation process followed Eq. 1) and no SiO 2 bubbles or barrier was observed. At very high temperatures, such as 1300 C, the oxidation rate was high, and pores and cracks might form, thus the di usion process along these channels was too quick to form SiO 2 barrier or bubbles. In addition, the small grains might form as follows. When outward di usion of CO or SiO came out of some TiO 2 planes, they destroyed the original well-shaped grains and left holes behind them. These holes developed with time forming the morphology of small grains on the large grains. Further work is required to study the mechanism of the formation of the small grains. 4. Conclusions The oxidation kinetics of Ti 3 SiC 2 polycrystalline samples containing 7 wt.% TiC from 900 C to 1300 C in air are parabolic with an activation energy of 350 kj mol 1. Furthermore, the oxidation at 1200 C and 1300 C can be considered as a two-step parabolic oxidation process. The oxide scale is dense and adherent, resistant to cyclic oxidation. The scale was generally composed of an outer layer of coarse-grained TiO 2 and an inner layer of ne-grained mixture of TiO 2 and SiO 2. The oxide scale formed at 1100 C contained a discontinuous SiO 2 barrier sandwiched in the coarsegrained TiO 2. Bubbles enriched in Si were also observed forming on some surface of TiO 2 at 1100 C and 1200 C. The parabolic oxidation, combining with marker experiment, supported a di usion-controlled process that was the inward di usion of oxygen and outward di usion of titanium and carbonaceous species CO). In addition, this process at high temperatures, especially at 1100 C, might also be controlled by the outward di usion of SiO to the oxide surface. Acknowledgements This work was supported by National Outstanding Young Scientist Foundation under grant no. 59925208, the National Sciences Foundation of China under grant no. 59772021 and the 863 program. References [1] W. Jeitschko, H. Nowotny, F. Benesovsky, Monatash Chem. 98 1967) 329. [2] M.W. Barsoum, T. El-Raghy, J. Am. Ceram. Soc. 79 1996) 1953. [3] Z. Sun, Y. Zhou, J. Zhou, Phil. Mag. Lett. 80 2000) 289. [4] R. Raj, J. Am. Ceram. Soc. 76 1993) 2147. [5] C. Racault, F. Langlais, R. Naslain, J. Mater. Sci. 29 1994) 3384. [6] X. Tong, T. Okano, T. Iseki, T. Yano, J. Mater. Sci. 30 1995) 3087. [7] M.W. Barsoum, T. El-Raghy, J. Electrochem. Soc. 144 1997) 2508.
Z. Sun et al. / Corrosion Science 43 2001) 1095±1109 1109 [8] A. Feng, T. Orling, Z.A. Munir, J. Mater. Res. 14 1999) 925. [9] R. Radhakrishnan, J.J. Williams, M. Akinc, J. Alloy. Comp. 286 1999) 85. [10] Y. Zhou, Z. Sun, S. Chen, Y. Zhang, Mat. Res. Innovat. 2 1998) 142. [11] R.A. Young, The Rietveld Method, Oxford University Press, Oxford, 1993. [12] S. Becker, A. Rahmel, M. Schorr, M. Schutze, Oxidat. Metals 38 1992) 425. [13] Z. Sun, Y. Zhou, unpublished work. [14] E.A. Gulbransen, K.F. Andrew, F.A. Brassar, J. Electrochem. Soc. 113 1966) 834.