Effect of in-situ carbon containing calcium aluminate cement on properties of Al2O3-SiC-C based trough castables

ABSTRACT The properties of Al2O3-SiC-C castables bonded by in-situ carbon containing calcium aluminate cement (CCAC) were investigated in this study. The results showed that after the Al2O3-SiC-C castables were dried at 110°C, their cold crushing strength (CCS) and cold modulus of rupture (CMOR) increased with the percentage of CCAC. The sample with CCAC content exceeding 2.5% by mass (named C2.5) had higher CCS and CMOR values than those of the model Al2O3-SiC-C castables (named BPS) with ball pitch as the carbon source. After being fired at 1100 and 1450°C, all the castables exhibited increased CCS and CMOR values, while the high apparent porosity of the BPS sample lowered its CCS and CMOR values. Compared with the BPS sample, the carbon materials of the castables bonded by CCAC exhibited improved dispersion in the matrix and excellent oxidation resistance, which enhanced the corrosion resistance of the refractory castables.


Introduction
Thanks to the high thermal conductivity, low thermal expansion, and molten slag/metal non-wettability of SiC and carbon materials, Al 2 O 3 -SiC-C based castables exhibit excellent corrosion resistance, thermal shock stability and good mechanical properties. Thus, they are usually used in iron troughs [1][2][3][4]. Unfortunately, the ball pitch, which is one of the main carbon sources used for Al 2 O 3 -SiC-C castables, releases harmful gases at elevated temperatures, pollutes the environment; and there is still a great challenge for the incorporation of the graphite flakes into castables owing to their poor water wettability [5]. All these drawbacks generate serious obstacles to the development of Al 2 O 3 -SiC-C based castables.
Presently, several technologies, including micropellets [6,7], coatings [8][9][10][11][12][13][14], and graphitic carbon spheres [15,16], have been employed to overcome the above issues. However, in practical applications, these technologies suffer from such drawbacks as discontinuous coating, aggregation and uneven distribution of carbon materials in the castables. More importantly, owning to the particle sizes of graphite flakes (including modified graphite), and the non-hydrophilic property of graphitic carbon spheres, it is difficult for these carbon materials to be uniformly dispersed in castable matrices, which are crucial for the corrosion resistance of castables [17,18]. Recently, the modification of refractory matrixes by in-situ carbon composites has gained popularity [19][20][21][22]. The present authors synthesized insitu carbon-containing calcium aluminate cement (CCAC) with the phase composition close to that of Secar71 via the carbon bed sintering method, and found that the water dispersion and oxidation resistance of CCAC were enhanced, and that the corrosion resistance of the MgO-Al 2 O 3 castables bonded by CCAC were remarkably improved [23][24][25].
To avoid the release of harmful gases from the pyrolysis of ball pitch, the Al 2 O 3 -SiC-C castables in the present work were prepared using CCAC as binder and carbon source replacing the ball pitch. The effects of CCAC content on the flowability, physical properties, mechanical properties, oxidation resistance, and corrosion resistance of the Al 2 O 3 -SiC-C castables were investigated. Also, for comparison, the model iron trough castables with ball pitch as carbon source and Secar71 as binder were prepared and their properties were tested.

Preparation of Al 2 O 3 -SiC-C castables
The in-situ carbon containing calcium aluminate cement (CCAC) with the phase composition close to that of Secar71 was synthesized, and the details of the process were presented in our previous works [23,25]. The in-situ carbon content of CCAC was 1.45% by mass. The specific surface areas of Secar71 (Imerys, China), and CCAC, determined according to the BET method, were approximately 4084 and 5458 cm 2 /g, respectively. The particle size distributions were determined by means of a laser particle size analyzer, and the obtained results are listed in Table 1.
Four groups of Al 2 O 3 -SiC-C castable samples were prepared using ball pitch and different CCAC contents (1.5, 2.5 and 4.0% by mass); these were named BPS, C1.5, C2.5, and C4, respectively. The water contents of all the samples were fixed at 4.8% by mass. Based on the formulations of the Al 2 O 3 -SiC-C castables (as listed in Table 2), the raw materials were weighed, dry mixed for 2 min, and then wet mixed for another 5 min following the addition of water. The castable mixtures were placed into cuboid samples (40 × 40 × 160 mm), and crucible samples (70 × 70 × 70 mm with a hole size of 40 mm×Φ 40 mm), and vibrated on a vibration table. The cast samples were cured at room temperature for 24 h and then were de-mold. They were subsequently cured at room temperature for another 24 h, and then dried at 110°C for 24 h. Finally, the dried samples were fired at 1100 and 1450°C, respectively, for 3 h in air atmosphere.

Tests and characterization methods
The flowability values of the four Al 2 O 3 -SiC-C castable mixtures were measured using the vibrating flowing table method. The bulk density (BD) and apparent porosity (AP) values of the castable samples, after they were dried and fired, were tested based on the Archimedes method. The cold modulus of rupture (CMOR) and cold crushing strength (CCS) were measured in accordance with the relevant national standards of China (GB/T3001-2007, GB/T5072-2008). The hot modulus of rupture (HMOR) was tested at 1450°C after a holding time of 0.5 h, in accordance with the GB/T3002-2004 standard. To evaluate the oxidation resistance of the Al 2 O 3 -SiC-C castables, the cross-sections of the prism-shaped samples (40 × 40 × 160 mm), which had been fired at 1450°C for 3 h, were photographed, and the decarbonation layers in the cross-sections were measured using a vernier caliper.
For the corrosion resistance test, the hole of each prepared crucible sample was filled with 30 g slag powder (its chemical composition was listed in Table 3). The crucible samples were then fired at 1450°C for 3 h. The corroded samples were cooled and then cut axially, and the corroded depth value (mm) was measured using a vernier caliper. The cross-section was further observed via a field-emission scanning electron microscopy (SEM, SU6600, Japan). The flowing values of BPS, C1.5 and C2.5 exhibited no distinct differences, but were higher than that of C4. After being set for 60 min, BPS had no pronounced change in its flowing value, while the other three Al 2 O 3 -SiC-C castables bonded by CCAC presented lower flowability owing to the higher hydration rate of CCAC resulted from its higher surface area.

Bulk density and apparent porosity
The AP and BD values of the dried and fired Al 2 O 3 -SiC-C castables are listed in Table 4. It can be seen that the BD values of the C1.5, C2.5, and C4 samples after being dried at 110°C were similar, and slightly higher than that of BPS.
The BD values of all the samples changed little after they were fired at 1100 and 1450°C. Compared with the castables after being dried at 110°C, the four Al 2 O 3 -SiC -C castables after being fired at 1100°C exhibited evidently higher AP values owing to the decomposition of the hydration products of calcium aluminate. Moreover, the AP value of BPS was higher than that of the other three samples because of the pyrolysis of ball pitch. Interestingly, the porosities of the four castable samples decreased after they were fired at 1450°C due to formation of the liquid phases, which caused the pores to close. The carbon materials from the pyrolysis of ball pitch were oxidized easily, resulting in BPS having highest AP value among all the samples.

Strengths
The CCS and CMOR values of the dried and fired Al 2 O 3 -SiC-C castables are depicted in Figure 2. After being dried at 110°C, the sample BPS had CCS and CMOR values of 60 and 11MPa, respectively. These were higher than those of C1.5 owing to the higher content of aluminate cement in BPS resulting in higher bonding strength than that in C1.5 [26]. The strength values of the sample BPS were lower than those of the C2.5 and C4 samples. It was believed that the ball pitch added to the BPS sample played negative roles in the formation of interlocked networks of the hydration products of calcium aluminates, thereby, reducing strength development. After being fired at 1100°C, the BPS, C1.5, C2.5 and C4 samples exhibited increases in their CCS and CMOR values of 13.33 and 36.36%, 35.29 and 70.00%, 11.76 and 50.00%, 7.89 and 5.00%, respectively, owing to the growth of ceramic phases forming the interlocking structures [27]. The C2.5 and C4 samples possessed higher strength values than that of the sample BPS. These can be associated with the higher AP value of the sample BPS, as an increase in cement content is beneficial to the development of bonding and aggregates. Interestingly, the CCS value of the C2.5 sample after it was fired at 1450°C, increased from 70 to 85MPa, but the strength values of the other three castables declined slightly. Although the CMOR value of the C2.5 sample declined by 36.8%, it remained higher than those of the others. Figure 3(a,b) show the microstructures of the BPS and C2.5 samples after they were fired at 1450°C for 3 h. The ceramic bonds between the matrix and aggregate in C2.5 were better than those in the BPS sample. These observations can explain why the strength values of the C2.5 sample are higher than those of the BPS sample. The lower values of CCS and CMOR observed in C4 can be associated with the formation of low melting phases owning to the increased content of CaO originating from calcium aluminate [26].
The HMOR values of the Al 2 O 3 -SiC-C castables were tested at 1450°C after a holding time of 0.5 h. The results (as shown in Figure 4) indicated that C1.5 possessed the highest HMOR value (2.15 MPa), and C4 presented the lowest value, which was attributed to the more liquid phase at high temperatures owing to the increased content of CaO originating from CCAC [28]. In addition, the HMOR value of C2.5 was 1.72 MPa, which exhibited no difference compared with that of BPS.

Oxidation resistance
The cross-sections of the Al 2 O 3 -SiC-C castables after they were fired at 1450°C for 3 h are shown in Figure 5. The oxidized regions (light gray area) were visibly observed from the image of the BPS sample, and the average thickness of the decarbonation layers was approximately 9.72 mm. However, there were no pronounced oxidation regions in the cross-sections of the castables bonded by CCAC, and the existing dark gray color of the crosssections was deepened with the percentage of CCAC. These observations implied that the carbon materials incorporated into the Al 2 O 3 -SiC-C castables through CCAC had better oxidation resistance than that of the carbon materials from the pyrolysis of ball pitch, and that the residual carbon materials in the castables increased with the addition of CCAC.

Corrosion resistance
In Figure 6, the cross-section images of the Al 2 O 3 -SiC-C castables after corrosion were depicted, indicating that the shapes of the four crucibles remained intact. Compared with those in BPS, the boundaries between the slag and crucibles in C1.5, C2.5 and C4 remained regular. The three-phase slag-air-sample junctions at the top of BPS were seriously corroded, which was not found in the cross-sections of the Al 2 O 3 -SiC-C castables bonded by CCAC. In addition to C2.5, the bottoms of the C1.5 and C4 samples exhibited distinct cracks. It can be concluded that the iron trough castables with the addition of 2.5% by mass of CCAC possessed better corrosion resistance. The corrosion depth of the C2.5 sample, as measured using a vernier caliper, was 1.2 mm, which was  lower than that of the BPS sample (3.5 mm). These results can be attributed to the pores left by the decomposition of ball pitch and oxidation of carbon materials, resulting in increased slag corrosion channel within the BPS sample. The carbon materials introduced into the castables by CCAC had improved dispersion and oxidation resistance, which resulted in lower porosity, and prevented the penetration of molten slag into the castables owing to the molten slag non-wettability of carbon materials.
The microstructures of the Al 2 O 3 -SiC-C castables after the corrosion resistance test were observed via SEM, and their backscattered electron (BSE) images are shown in Figure 7. It can be seen that in the BPS sample (Figure 7(a)), the outermost layers of the castable matrices, which were in contact with the slag, had already been corroded, leaving the corundum aggregates directly exposed to the slag. Furthermore, the slag had penetrated into the corundum aggregates, forming CA 6 and calcium aluminosilicates (CAS) in the slag-corundum aggregate interface zones. Meanwhile, the boundaries of the corundum aggregates in the C2.5 sample (Figure 7 (b)) remained intact. The reasons for these observations were that the matrix containing CCAC, which possessed excellent corrosion resistance, coated the corundum aggregates, isolating the molten slag from the oxide grains.

Conclusions
The content of CCAC had significant effects on the physical properties, strengths, and oxidation resistance, especially corrosion resistance of the Al 2 O 3 -SiC -C castables. Compared with the castables prepared with ball pitch as the carbon source, the Al 2 O 3 -SiC-C castables bonded by CCAC had lower AP values after being dried and fired. Increasing the content of CCAC improved the CCS and CMOR values of the castables, but degraded their HMOR. In addition, CCAC enhanced the dispersion and oxidation resistance of the carbon materials in the Al 2 O 3 -SiC-C castable matrices. This lowered the porosity of the castables, hindered molten slag penetration, and avoided the chemical attack of the corundum grains by molten slag. Hence, the Al 2 O 3 -SiC-C castables with 2.5% by mass of CCAC had better corrosion resistance than those of the trough castables bonded by Secar71 and using ball pitch as the carbon source. The increased content of CaO with the percentage of CCAC deteriorated the corrosion resistance of the refractory castables.