Four sets of samples using the compositions shown in Table 1 were produced. The first set of the ultralow cement type used water as the mixing liquid and was named ULCC_water. The second set of the ultra-low cement type used the Al polymeric resin as mixing liquid and was named ULCC Al 2 O 3 . The first set of the no cement type used water as the mixing liquid and was named NCC_water. The second set of no cement type used the Al polymeric resin as the mixing liquid and was named NCC Al 2 O 3 .

¹Alphabond 300. ²STP = Sodium Tripolyphosphate. ³APA = Amonium Polyacrilate. ⁴The quantity of 8 wt% of liquid polymeric resin was equivalent to the addition of 1 wt% of alumina nanoparticles after pyrolysis.

The compositions were devised to approximate an Andreasen’s particle size packing with a distribution modulus q = 0.26, aiming an almost self flowing rheological condition.

The basics of the Pechini’s methodology [6] to produce the polymeric resin, consists in the dissolution in deionized water at 70˚C, of a salt of the cation of interest, in this case, Al(NO₃)₃. In the sequence, citric acid is added forming the metallic citrate. In order to promote the esterification reaction, ethylene glycol is added and the temperature is raised to 90˚C, keeping the system for two hours under this condition. The resin viscosity can be controlled by evaporating the residual water simply expositing it at 110˚C for specific time periods. In this study, a resin viscosity of 8cps was used in the molding of the castables containing it.

The specimens used in the testing were molded under light vibration in the form of rectangular bars with dimensions of 25 × 25 × 150 mm.

After demolding, they were dried at 110˚C for 24 hours. After drying, specimens to be used in physical and mechanical testing were sintered at 1450˚C, using the firing program of Table 2, with a dwell time of 2 hours at the final temperature. On the other hand, specimens intended for dynamic elastic modulus and thermal expansion tests were only dried.

1) The evaluated physical properties after firing at 1450˚C/2h of specimens were:

· Apparent Porosity

· Apparent Density

· Pore size distribution

· Nanoparticles dimensions and specific surface

2) The Mechanical Strength of specimens fired at 1450˚C/2h was evaluated by flexural testing in 3 point bending.

3) The Elastic Modulus versus Temperature testing of dried specimens (110˚C/24h), was evaluated by means of the impulse excitation technique, using a Resonant Frequency Damping Analyzer (RFDA-HT1600-IMCE-Belgium) in line to ASTM-1548-2 (2012). The test was conducted from room temperature to 1500˚C, under a 3˚C/minute heating rate.

4) The thermal shock testing was conducted by measuring both the residual dynamic elastic modulus, as well as the residual 3 point bending strength, after a single thermal shock test, under the sudden gradients of 300˚C, 600˚C and 1000˚C. The thermal shock was conducted by specimens cooling in running water at 24˚C.

5) The thermal expansion behavior was determined from room temperature to 1250˚C.

In Table 3, the characteristic properties of the nano Alumina powder after pyrolisis and burning of the resin at 1000˚C/1h, are presented.

X-ray diffraction is a convenient method for determining the mean size of nanocrystallites in nanocrystalline bulk materials. The first scientist, Paul Scherrer, published his results in a paper that included what became known as

*Calculated from the Specific Surface area. **Calculated by the Scherrer Equation.

the Scherrer equation in 1981 [1] . This can be attributed to the fact that “crystallite size” is not synonymous with “particle size”, while X-Ray diffraction is sensitive to the crystallite size inside the particles. From the well-known Scherrer formula the average crystallite size, L, is:

L = K λ / β cos θ

where λ is the X-ray wavelength in nanometer (nm), β is the peak width of the diffraction peak profile at half maximum height resulting from small crystallite size in radians and K is a constant related to crystallite shape, normally taken as 0.9 [8] .

The results confirm the nano size of the alumina powder that was generated after pyrolysis and burning of the carbon content of the liquid resin.

In Figure 1 it can be seen that the resin indeed renders pure Al₂O₃ product after pyrolysis and oxidation starting at 900˚C.

In Figure 2 and Figure 3 the pore size distribution and the respective cumulative porosity of the investigated castables are presented.

It can be noticed that the ULCC Al 2 O 3 castable molded with the liquid resin presented eminently pores with equivalent diameters in the range from 0.1 µm to 1.0 µm.

On the other hand, the ULCC_Water castable molded with water presented a bigger pore size distribution range from 0.5 to 3.0 µm.

For the NCC castables, the one molded with water presented population of the pore of smaller sizes from 0.08 µm to 2.0 µm, while the NC Al 2 O 3 molded with the Al₂O₃ resin presented pore population sizes in the range from 0.1 µm to 1.0 µm.

Considering the observed range of dimensions, in all the cases, the pore sizes presented by the investigated castables, indicate a high concentration of virtually non-permeable pores to liquid metals or slag (<5 µm).

In Table 4 the results of Apparent Density and Apparent Porosity, after drying at 110˚C/24h and after firing at 1450˚C/2h, are were presented.

It can be noticed that the Apparent Densities after drying and firing are substantially bigger for the specimens containing the resin. As a consequence, the Apparent Porosity was smaller for the specimens molded with it. This fact was unexpected, considering that the samples produced using the resin as the liquid mixing vehicle, used a larger amount of molding liquid (8 wt% against 5.8 wt%

of water used in the the ULCC_water and NCC_water). In addition, after firing, there was an additional mass loss because of the volatiles coming from the pyrolisis and burning of the carbon of the resin plus the eventual water excess. The results indicated a better sintered microstructure upon addition of the Al₂O₃ resin.

From Figure 4 it can be noticed that the overall dimensional variation independent of the type of castable, including the thermal expansion coefficient (inclination of the curve), are smaller for the samples molded with the Al₂O₃ containing resin. The difference is particularly noticeable in the range from 400˚C to 900˚C. This indicates a lesser volume variation for eventual castable refractory linings molded with castables containing the Al₂O₃ resin.

In Figures 5(a)-(c), the X-ray diffraction patterns of ULCC Al 2 O 3 and NCC Al 2 O 3 are presented. They explain why the castables molded with the Al₂O₃ resin displayed smaller thermal expansion coefficients. The reason resides in the fact that the aluminum oxide nanoparticles coming from the resin, reacted more readily in the aggregates environment, leading to the formation of bigger amounts of mullite (3Al₂O₃·2SiO₂) and ebonite (CaO·6Al₂O₃) in the final phase assembly of these castables, than the castables molded only with water. This contributed to smaller thermal expansion coefficients for castables molded with the Al resin.

In Table 5 and Figure 6, the results of flexural strength in three-point bend, of the castables are presented. It is clear that after firing at 1450˚C/2h, the castables molded with the liquid resin presented a substantially bigger average flexural strengths than the castables molded with water. This fact indicates clearly that the nano alumina particles generated by the liquid resin were able to contribute to a much stronger castables matrices, under a microstructure condition of pores of comparable sizes and volume fraction.

The elastic modulus of ceramics upon heating to high temperatures for certain

*Average over 5 specimens – standard deviations in the range ±0.3 MPa.

periods of time is sensitive to microstructure alterations caused by reactions between components, phase transitions, liquid phase development, pyroplasticity, densification (sintering) and/or defects opening.

In this work, in order to investigate the differences in microstructural evolution with the temperature of the castables, the dynamic elastic modulus of only dried specimens, was evaluated continuously from room temperature to 1450˚C, under a heating rate of 3˚C/minute. The testing was made using the impulse excitation technique IET using the equipment RFDA 1600 HT from ICEM Belgium [7] [9] [10] .

The results shown in Figure 7 and Figure 8 indicate that similar to what happened with the flexural strength, the room temperature elastic modulus of the resin molded castables were substantially higher than those molded with water. Upon heating from room temperature to approximately 1300˚C, the elastic modulus of the castables molded with liquid resin continued to present substantially higher values of elastic modulus and display a volumetric behavior more stable than the castables molded with water.

This difference is made clearly on the respective heating up and cooling down plots as well as on the hysteresis curves, which displayed a much smaller area for the castables molded with resin. It can be seen that after cooling down, their elastic modulus at room temperature, get back very close to their initial startup value. This fact indicates that the microstructures of the castables containing resin, were less altered by a heating up and cooling down cycle, and that they

somewhat healed themselves from eventual defect opening upon heating and cooling, ending as strong as at the beginning of the test. Conversely, the behavior of the castables molded with water was marked by a strong hysteresis loop indicating some relevant microstructure alteration. Their elastic modulus at room temperature after the heating and cooling cycle, were still smaller than those of dried startup value, implying that just one heating up and cooling down cycle caused a reasonable level of permanent damage on their final microstructure [11] .

For both castables, the ups and downs in the plots at temperatures close to 1400˚C can be a credit to increasingly damping effects, caused by pyroplastic deformation originated from liquid phase development.

An examination of the cooling down plots in Figure 7 and Figure 8, also indicates a slight increase of elastic modulus for both types of castables, from 1450˚C down to approximately 700˚C. We may credit this behavior to sintering, microstructure healing caused by the contraction of the specimens bulk phase and also to a decrease in the pyroplastic character with the hardening of the eventual liquid phase developed within the matrices of the investigated castables. Finally, from 600˚C and down, the elastic modulus of both types of castables experiences a decrease (which is much bigger for the water-molded samples). This fact may indicate a possible opening of microstructure defects, caused by contraction cracking and hardening below the glass transition temperature of the vitreous phase.

By comparing the behavior of both castables from Figure 7 and Figure 8, it may then be concluded that after a firing cycle, the castables molded with the liquid resin are stronger and reached a more stable final microstructure than the castables molded with water. This is indicative of the strong sintering effect that just 1 wt% of alumina nanoparticles had in the castables matrices.

Specimens of both castables (ULCC and NCC), with and without Al₂O₃ resin, previously fired at 1450˚C/2h, were submitted to thermal shock by a single quenching in running tap water at 24˚C, under the sudden gradients of 300˚C, 600˚C and 1000˚C.

In order to evaluate the thermal shock damage, the residual elastic modulus of the shocked specimens was continuously measured from room temperature to 1450˚C. In addition, the residual flexural strength after thermal shock were also determined [12] [13] . The resulting elastic modulus and residual strength versus temperature behavior after thermal shock, are shown in Figures 9-13.

It is evident that the first thermal shock at 300˚C causes the most significant damage, displaying a fall of about 60% of the elastic modulus and flexural strength, relative to the specimens without thermal shock damage. Under the point of view of thermal shock thermo elastic theory, this temperature can be related to the critical ΔT of pure alumina materials. A second point to be observed is that the specimens with the presence of resin have a higher resistance

to thermal shock damage, due to their higher final values of elastic modulus. Finally, the specimens of the ULCC type when compared to the specimens of type NCC show a better behavior against thermal shock as can be seen on its leveling off the residual flexural strength, after ΔT = 300˚C.

In order to check the effectiveness of alumina nanoparticles formed in situ coming from the resin addition, comparatively to alumina nanoparticles added in powder form, the flexural strength and dynamic elastic modulus behavior of NCC castable in both situations were evaluated. It has to be pointed out that the addition of 8 wt% of liquid resin in the castable batch composition, results after resin pyrolisis and oxidation, on only 1 wt% of Al₂O₃ nano particles in situ. In this manner, a volume of liquid resin equivalent to 8 wt% of the castable’s total mass, was separately pyrolised and burnt to 1000˚C/1hour. The resulting powder was mixed in substitution of 1 wt% of the calcined reactive alumina, in the batch composition being nominated NCC Al 2 O 3 * . In this case, water was used as the mixing liquid in the amount of 5.8 wt%. The results of the flexural strength of fired specimens are shown in Figure 14, while the results of dynamic elastic modulus of only dried specimens are shown in Figure 15. It can be seen that the development in situ of the alumina nanoparticles coming from the resin, was

much more effective on increasing both the flexural strength and the dynamic elastic modulus, than by adding the nanoparticles in powder form. The flexural strength after the addition of the nanoparticles powder in the NCC Al 2 O 3 * , was almost the same as that of the NCC_water composition (without resin), showing no properties improvement at all. This fact leads to the conclusion that the development in situ of the alumina nanoparticles, is indeed a powerful novel route, much more effective on improving castables thermal and mechanical properties, than adding the same quantity of nanoparticles in powder form.