Expansion Behavior of Cement- Bonded Alumina Magnesia Refractory Castables

Expansion Behavior of Cement- Bonded Alumina Magnesia Refractory Castables To attain better performance in steel ladle applications, the engineered microstructural design of alumina magnesia castables is presented and discussed based on spinel and formation and its effects on castable properties. M.A.L. Braulio, D.H. Milanez, E.Y. Sako, L.R.M. Bittencourt and V.C. Pandolfelli The tough environment of steel ladles requires materials with high performance that are able to deal with various requirements, such as corrosion, mechanical wear and thermal shock damage. Taking this into account, alumina magnesia castables present an outstanding set of properties, including high corrosion and slag infiltration resistance coupled with high refractoriness. 1 These properties are attained by in-situ spinel formation, which is followed by significant volume expansion. 2 At temperatures in the range of 1000 1500 C, alumina and fine magnesia present in the matrix react with each other, which provides spinel (MgAl 2 O 4 ) formation. Various parameters affect this reaction and, consequently, the final expansion. Therefore, the alumina and magnesia reactivity, 3,4 magnesia grain size 5 and fumed silica content, 6,7 among other parameters, must be analyzed to control the alumina magnesia properties of castables. An additional factor that contributes to the expansive behavior of these castables is the binder agent. Generally, calcium aluminate cements (CACs) are used in alumina magnesia castables and have an important role in the expansion phenomenon because of calcium hexaluminate ( ) formation at temperatures >1400 C. 8,9 Fuhrer et al. 10 have evaluated the microstructural evolution of in-situ spinel castables bonded with CAC and have pointed out the growth of crystals in a needlelike morphology. Furthermore, these crystals have been observed in the matrix and in tabular alumina grains, which leads to their disintegration. Considering this aspect, cement content control is a key issue when considering a suitable set of properties, because excess could result in a porosity increase and, consequently, mechanical strength deterioration and decrease in infiltration resistance. Conversely, a controlled development of could improve some castable properties (especially corrosion and thermal shock resistance) because of its low solubility in iron-containing slags and the development of needlelike crystals, which increases the toughness of the material. In this article, the effect of CAC content (6, 4 and 2 wt%) on the properties of alumina magnesia castables has been analyzed to verify which amount could lead to better properties and higher performance. Considering that alumina magnesia castables are commonly used as steel ladle well blocks, these material properties have been assessed using mechanical strength, thermal expansion, thermal shock and creep resistance, which are the main application requirements. Additionally, spinel and formation have been analyzed with the purpose of detecting their contribution to the expansion behavior of alumina magnesia castables. American Ceramic Society Bulletin, Vol. 86, No. 12 9201

Modulus of rupture (MPa) Preparation and Characterization of Castables Vibratable high-alumina refractory castables containing 6 wt% of dead-burned magnesia (95 wt% MgO, CaO/SiO 2 = 0.36, d 45 µm; Magnesita SA, Brazil), 7 wt% of reactive alumina (CL370; Almatis), 1 wt% of fumed silica (971 U; Elkem, Norway) and various CAC contents (6, 4 and 2 wt%; Secar71, Kerneos, France) were designed using the q = 0.26 Alfred packing model. Various tabular alumina grades (d 6 mm; Almatis, Germany) were used as the refractory aggregates. The dispersion was assured by adding 0.2 wt% of a poly(carboxylate ether)-based dispersant (Bayer, Germany). The water content was 3.9 wt%, which provided suitable shaping. To complement the results, two compositions were prepared: an alumina magnesia castable without CAC and an alumina CAC castable without magnesia. In both cases, CAC or magnesia was replaced by a fine tabular alumina (<45 µm; Almatis). The same dispersant applied for the alumina magnesia castables was used, but there was a slight increase in the water content (4.1 wt%), likely because of changes in the packing and surface area of the castable particles (Table 1). Castable mixing was conducted in a paddle mixer for 5 min, and water was added in two steps, which probably led to better mixing effectiveness, according to Pileggi et al. 11,12 The characterization of the castables comprised the thermal expansion evaluation, mechanical strength, apparent porosity and permanent linear expansion (PLE) after firing at three temperatures. Creep and thermal shock resistance also were determined. Quantitative X-ray diffractometry (XRD) analyses and scanning electron microscopy (SEM) also were performed to support the obtained results. Fig. 1 Mechanical behavior, and permanent linear expansion (PLE), of alumina magnesia castables that contain various CAC content (6, 4 and 2 wt%). 14 Table 1 Castables Compositions Raw material AMC AC AM Tabular alumina (d 6 mm; Almatis) (wt%) 80 84 80 80 Reactive alumina (Almatis) (wt%) 7 7 7 Dead-burned magnesia (Magnesita) (wt%) 6 0 6 Tabular alumina (d 45 µm; Almatis) (wt%) 0 6 6 Microsilica (Elkem) (wt%) 1 1 1 Calcium aluminate cement (Kerneos) (wt%) 2 6 6 0 AMC is alumina magnesia containing various CAC content, AC is alumina CAC and AM is alumina magnesia. The expansion analysis and the creep tests were conducted using refractoriness under load equipment (Model RUL 421E, Netzch, Germany). Cylindrical samples were prepared according to the 51053 DIN standard, cured at 50 C and dried at 110 C for 1 d, followed by prefiring at 600 C for 5 h before testing. For thermal expansion evaluation, samples were heated to 1500 C at a heating rate of 3 C/min and kept at this temperature for 5 h. The compression load applied was 0.02 MPa. Concerning the creep measurements, samples were prefired at 1550 C for 24 h and then analyzed at 1450 C for 24 h under constant compression load of 0.2 MPa. For the mechanical tests (PLE and thermal shock analysis), 25 25 150 mm bars were prepared. After they were cured (at 50 C for 1 d), dried (at 110 C for 1 d) and calcined (at 600 C for 5 h), the samples were fired under a heating rate of 1 C/min at 1150, 1300 and 1500 C. The modulus of rupture was obtained using the three-point bending test (Model 810, MTS Systems) according to ASTM C133-94 ( Standard Test Methods for Cold Crushing Strength and Modulus of Rupture of Refractories, ASTM Book of Standards, ASTM International, West Conshohocken, Pa.). The PLE was measured using the difference (in percent) between the final and initial bar lengths. Additionally, apparent porosity of representative samples of these castables was measured using the Archimedes technique. Thermal shock tests were conducted using multiple cycles for the samples fired at 1500 C. The furnace PLE (%) 9202 American Ceramic Society Bulletin, Vol. 86, No. 12

temperature was kept constant at 1025 C, and the dwell time at this temperature was 15 min. After this period, the samples were thermally shocked in air, which led to a temperature gradient of ~1000 C. Subsequently, the elastic modulus was evaluated using the resonance method ( Standard Test Method for Dynamic Young s Modulus, Shear Modulus and Poisson s Ratio for Advanced Ceramics by Sonic Resonance, ASTM C-1198). Later, the samples were returned to the furnace and the same procedure was repeated for 10 cycles. The microstructure parameters were characterized using XRD quantitative analysis and software based on the Rietveld method 13 and SEM with back-scattering electron imaging (BSI). The target of these evaluations was to quantify the final spinel and content as well as to identify their distribution in all the castables. Effect of CAC Content on Overall Expansion of Alumina Magnesia Castables In previous work, 14 a strong effect of CAC was detected on the expansive behavior and mechanical strength of alumina magnesia refractory castables, which pointed out that spinel formation is not the only expansion source of the system. The mechanical strength and the PLE of alumina magnesia castables containing 2, 4 and 6 wt% of CAC were determined (Fig. 1). The higher-cement-content castable resulted in the largest expansion and the lowest mechanical strength. Two mechanisms were related to these results: spinel formation and CA6 formation. To understand these mechanisms, three compositions were evaluated: an alumina magnesia castable without cement (6 wt% of magnesia), an alumina cement castable without magnesia (6 wt% CAC) and a standard composition, which was an alumina magnesia cement-bonded castable (6 wt% magnesia and 6 wt% CAC). Expansion profiles of these castables were prepared (Fig. 2(a)). The alumina cement castable (with no magnesia) presented a small expansion (~0.8% after heat treatment). The alumina magnesia castable (with no CAC) showed an intermediate expansion (~2.0% after testing). On the other hand, when cement and magnesia were used together, the result was a higher overall expansion (~3.0%). The derivatives of the expansion curves, which led to the expansion rate, also were prepared (Fig. 2(b)). The first peak observed was attributed to in-situ spinel formation, because it also was present in the alumina magnesia castable without CAC. In Time (h) contrast, the second peak was related to formation, because it also was detected in the alumina CAC sample without magnesia. Therefore, for the alumina magnesia castable containing CAC, the higher overall expansion was a consequence of spinel and formation effects. These spinel and expansion rate peaks, which were observed, could explain the expansive behavior of the alumina magnesia castables with various CAC contents. 14 It was possible to detect that the final expansion of alumina magnesia castables was affected by the CAC content (Fig. 3(a)) and that the higher the CAC content, the higher the expansion, because of greater CA6 formation (Fig. 3(b)). L/L 0 (%) Expansion rate (%/min) Time (h) Fig. 2 (a) Expansion behavior and (b) expansion rate of alumina CAC castable with no magnesia (AC), alumina magnesia castable with no CAC (AM) and alumina magnesia castable bonded with CAC (AMC). American Ceramic Society Bulletin, Vol. 86, No. 12 9203

L/L 0 (%) Expansion rate (%/min) Time (h) Time (h) Fig. 3 (a) Expansion behavior and (b) expansion rate of alumina magnesia castables that contain various CAC amounts (6, 4 and 2 wt%). 14 To confirm these results more efficiently, XRD quantitative analysis was performed on the samples fired at 1500 C. The same amount of spinel in the three compositions (~21 wt%) was detected, because the magnesia content was the same for the three samples (6 wt%). The sample with 6 wt% CAC presented 14 wt% of this phase, whereas the sample with 4 wt% CAC presented 8 wt% of this phase. For the sample with 2 wt% of CAC, the technique was not able to detect because of its low amount. The microstructural analyses of these three castables fired at 1500 C were studied (Fig. 4). The sample with the higher cement content (6 wt%) presented welldeveloped crystals throughout the entire matrix, especially around the tabular alumina grains. Additionally, it was possible to assert that microcracks were generated in the tabular alumina grains. On the other hand, the sample with 2 wt% CAC densified and contained a small amount of. The sample with 4 wt% CAC presented an intermediate behavior: crystals were not so well developed as in the 6 wt% CAC castable, but there was much more than in the 2 wt% CAC sample. The large expansion of the castable that contained 6 wt% CAC led to a high level of apparent porosity after firing at 1500 C (Fig. 5). Because of the sintering of the 2 wt% CAC sample, there was a decrease in the apparent porosity from 1300 to 1500 C, which meant that the spinel expansion during its formation was overcome by sintering. In the range of 1150 1300 C, the porosity behavior of the three castables was similar, which meant that the CAC content had no influence when the spinel was being formed. Conversely, at 1500 C the effect of formation was clearly highlighted, because the higher-cement-content castable resulted in higher apparent porosity. In addition, the behavior of apparent porosity followed the same trend of PLE for the three castables. Therefore, a higher CAC content in alumina magnesia castables led to a higher apparent porosity, development of microcracks and lower mechanical strength because of larger expansion associated with the formation. However, a minimum value of CAC had to be incorporated if the target was to take advantage of the benefits associated with the presence of this phase. Thermomechanical Behavior of Alumina Magnesia Castables with Various CAC Contents A good thermal shock resistance can be attained by avoiding cracking initiation and by preventing crack propagation. The severe thermal shock conditions in steel ladle applications make it practically impossible to inhibit thermal shock damage initiation. Thus, the main solution to cope with thermal shock is promoting toughening mechanisms by microstructure design, such as crack deflection and crack branching. 15 The presence of can enhance refractory castable thermal shock resistance. 16 The needlelike crystals can act as in-situ whiskers, which increase the required energy for crack propagation. Therefore, it is important to have a good linkage between the crystals and the other castable components (especially tabular alumina and spinel) to provide a bridging toughening mechanism. 9204 American Ceramic Society Bulletin, Vol. 86, No. 12

Elastic modulus loss (in percent) as a function of thermal shock cycles for the alumina magnesia castables with various CAC content fired at 1500 C has been plotted (Fig. 6). The castable that contains 4 wt% CAC presents the best thermal shock behavior, according to its lower elastic modulus loss. The 2 wt% CAC castable shows the higher thermal shock damage for these sets of compositions. As analyzed before, it presents a densified microstructure, a high mechanical strength and a small amount of CA6. These features indicate that a high elastic energy is stored and that a low crack propagation resistance is attained. The microstructure of the materials with 6 and 4 wt% CAC indicate a great amount of in the castables. It should be expected that the sample that contains the greater amount (6 wt% CAC) presents the higher thermal shock resistance. Nevertheless, this result has not been observed after the thermal shock tests. It is believed that the better performance obtained by the sample with 4 wt% CAC might be a result of crack coalescence in the sample with 6 wt% CAC because of a high formation, which highlights the importance of CAC control in the castables. The creep resistance of refractory castables depends especially on three microstructural features: apparent porosity; amount, composition and distribution of glassy phases; and size, morphology and distribution of the crystalline phases. The creep behavior of the alumina magnesia castable with various CAC contents has been determined (Fig. 7), and it follows the same trend of the thermal shock results. Once again, the castable that contains 4 wt% CAC presents the best performance, which reinforces the need for CAC content control. Even with the lowest apparent porosity, the 2 wt% CAC castables present the worst creep resistance. The best results obtained for the samples with 4 and 6 wt% CAC might be related to development and its toughening mechanism ability, which results in lower deformation. Furthermore, the sharp initial decrease observed in the 2 wt% CAC sample might characterize a higher glassy phase formation in this composition. This aspect will be addressed in coming publications. CAC Content Optimization The selection of the suitable CAC content for alumina magnesia castables involves various parameters and a systemic analysis. The first point to be considered is the overall expansion of these castables, because formation imparts an extra contribution to the alumina magnesia castables expansion, in addition to the spinel formation. This aspect is of utmost importance to design of the steel ladle well blocks, and this microstructural engineered expansion control ability (by CAC variation) is an important tool to adapt the castable according to the application requirement. The second point regards the thermomechanical properties, in which the formation clearly indicates a better performance of alumina magnesia castables that contain 4 and 6 wt% CAC. The third point is that an intermediate CAC content can lead to a better balance for alumina magnesia castables, because an extremely low or an extremely high content can result in the degradation of properties, which decreases castable performance and its working life. Acknowledgments The authors are thankful to the Federation for International Refractory and Education (FIRE), Magnesita SA (Brazil) and the Brazilian Research Founding FAPESP for supporting this work. Additionally, the authors would like to gratefully acknowledge A. Genty and J. Poirier (Polytech Orléans) for the SEM analyses and G.B. Cintra (Federal University of São Carlos) for the thermal shock evaluation. About the Authors M.A.L. Braulio, D.H. Milanez, E.Y. Sako and V.C. Pandolfelli are faculty members in the Materials Engineering Dept. and FIRE Associate Laboratory, Federal University of São Carlos, São Carlos, SP, Brazil. L.R.M. Bittencourt is a research staff member of the Research and Development Center, Magnesita SA, Contagem, MG, Brazil. American Ceramic Society Bulletin, Vol. 86, No. 12 9205

References 1 S. Zhang and W.E. Lee, Spinel-Containing Refractories ; pp. 215 58 in Refractories Handbook. EUA, 2004. 2 W.E. Lee, W. Vieira, S. Zhang, A. Ghanbari, H. Sarpoolaky and C. Parr, Castable Refractory Concretes, Int. Mater. Rev., 46 [3] 145 67 (2001). 3 R. Sarkar, S. Chatterjee, B. Mukherjee, H.S. Tripathi, M.K. Haldar, S.K. Das and A. Ghosh, Effect of Alumina Reactivity on the Densification of Reaction Sintered Nonstoichiometric Spinels, Ceram. Int., 29, 195 98 (2003). 4 H.S. Tripathi, B. Mukherjee, S. Das, M.K. Haldar, S.K. Das and A. Ghosh, Synthesis and Densification of Magnesium Aluminate Spinel: Effect of MgO Reactivity, Ceram. Int., 29, 915 18 (2003). 5 J. Soudier, Understanding and Optimization of MgO Hydration Resistance and Spinel Formation Mechanisms for Increasing Performances of DVM Used in Crucible Induction Furnaces Melting Steel ; pp. 679 83 in UNITECR 05 Proceedings (Orlando, Fla., 2005). 6 H. Sarpoolaky, K.G. Ahari and W.E. Lee, Influence of in Situ Formation on Microstructural Evolution and Properties of Castables Refractories, Ceram. Int., 28, 487 93 (2002). 7 S.K. Chen, M.Y. Cheng, S.J. Lin and Y.C. Ko, Thermal Characteristics of Al 2 O 3 MgO and Al 2 O 3 Spinel Castables for Steel Ladles, Ceram. Int., 28, 811 17 (2002). 8 K. Ide, T. Suzuki, K. Asano, T. Nishi, T. Isobe and H. Ichikawa, Expansion Behavior of Alumina Magnesia Castables, J. Tech. Assn. Refract., 25 [3] 202 208 (2005). 9 Y.C. Ko and J.T. Lay, Thermal Expansion Characteristics of Alumina Magnesia and Alumina Spinel Castables in the Temperature Range 800 1650 C, J. Am. Ceram. Soc., 83 [11] 2872 74 (2000). 10 M. Fuhrer, A. Hey and W.E. Lee, Microstructural Evolution in Self-Forming Spinel/Calcium Aluminate Castable Refractories, J. Eur. Ceram. Soc., 18, 813 20 (1998). 11 R.G. Pileggi, A.R. Studart, V.C. Pandolfelli and J. Gallo, How Mixing Affects the Rheology of Refractory Castables Part 1, Am. Ceram. Soc. Bull., 80 [6] 27 31 (2001). 12 R.G. Pileggi, A.R. Studart, V.C. Pandolfelli and J. Gallo, How Mixing Affects the Rheology of Refractory Castables Part 2, Am. Ceram. Soc. Bull., 80 [7] 38 42 (2001). 13 H.M. Rietveld, A Profile Refinement Method for Nuclear and Magnetic Structures, J. Appl. Crystallogr., 2, 65 71 (1969). 14 M.A.L. Braulio, D.H. Milanez, E.Y. Sako, L.R.M. Bittencourt and V.C. Pandolfelli, The Effect of Calcium Aluminate Cement on the in Situ Spinel Expansion ; pp. 540 43 in UNITECR 07 Proceedings (Dresden, Germany, 2007). 15 A. Ratle, V.C. Pandolfelli, C. Allaire and M. Rigaud, Correlations between Thermal Shock and Mechanical Impact Resistance of Refractories, Br. Ceram. Trans., 96 [6] 225 30 (1997). 16 G. Büchel, A. Buhr, D. Gierisch and R.P. Racher, Bonite A New Raw Material Alternative for Refractory Innovations ; pp. 462 67 in UNITECR 05 Proceedings (Orlando, Fla., 2005). 9206 American Ceramic Society Bulletin, Vol. 86, No. 12

Fig. 4 Microstructural features of alumina magnesia castables that contain 6, 4 and 2 wt% CAC after firing for 5 h at 1500 C.

Apparent porosity (%) PLE (%) Fig. 5 Apparent porosity and permanent linear expansion (PLE) of alumina magnesia castables that contain various CAC contents (6, 4, and 2 wt%). Elastic modulus (%) Thermal shock cycles Fig. 6 Elastic modulus loss for alumina magnesia castables that contain 6, 4 and 2 wt% CAC fired at 1500 C after thermal shock cycles.

L/L 0 Time (min) Fig. 7 Creep behavior of alumina magnesia castables with various CAC content (6, 4 and 2 wt%).