2014 Volume 54 Issue 8 Pages 1823-1829
A study has been carried out to clarify the sintering behavior of silica filler sand. Sintering experiments were firstly conducted and the experimental data were compared with the thermodynamic calculations by FactSage and MELTS software.
It was found that sintering of the silica sand proceeded through the evolution of liquid phases, which were originated from alkali feldspars. The liquid proportion was found to increase with an increase in temperature. It wan considered that the viscous liquid phases connecting silica particles contributed to the strength of the sintered sand. The thermodynamic calculations revealed that MELTS software could sufficiently simulate the phases in equilibrium with each other. The comparison of the experimental data with the thermodynamic simulations demonstrated that the whole system of the silica sand did not attain equilibrium. It was hence inferred that filler sands should be designed to prevent the state fully sintered by controlling the particle size.
Filler sands are usually used for the sliding nozzle of a ladle to keep the sliding gate refractory apart from molten steel. Filler sands, put into the space upon the sliding gate, are sintered with taking heat from molten steel. The sands should be sintered to an appropriate extent to keep stable operations without any problems. The probable problems are typically classified into the following two phenomena. A surplus extent of sintering results in a failure to open the gate to pour molten steel, while an insufficient extent of sintering results in leakage of molten steel. Therefore, it is necessary to understand the sintering behavior of filler sands. However, very few studies have been so far carried out regarding filler sands.
Matsui et al.1) has studied the effect of concentrations of alkali oxides on the sintering behavior of silica filler sands to prevent surplus sinter that leads to blockage on pouring. They presumed that the sinter proceeded with the generation of liquid phases in between silica particles. Volume fractions of liquid phases were calculated using thermodynamic software with varying alkali oxide contents as a function of temperature. As a result, it was revealed that the liquid fraction increased with increasing alkali oxide contents as well as with increasing temperature. Finally, an optimum condition was found to prevent surplus sinter and the direction to successful operations was suggested. Another literature2) has pointed out that filler sands may affect cleanliness of molten steel. It is considered that improper ejection of filler sands may lead to the formation of large inclusions.
Consequently, it is important to understand the sintering behavior of filler sands to stabilize practices with ladles and to keep steel cleanliness. With this background, a study has been undertaken to clarify the sintering behavior. Sintering experiments were firstly conducted using silica filler sands. After that, microscopic observation was carried out to understand how liquid phases generate. Furthermore, the experimental data were compared with thermodynamic calculations to confirm if the prediction is probable by calculations.
Sintering experiments have been firstly conducted to understand the effect of temperature on the sintering state. In addition, the effect of particle size has been determined to better understand the sintering behavior from the view points of equilibrium and kinetics. Thermodynamic calculations were further carried out to see if the prediction is probable.
2.1. Effect of TemperatureA box type furnace schematically illustrated in Fig. 1 was used for the sintering experiments. After the furnace was heated up to 1000°C, a platinum crucible containing 15 g of sand was placed in the furnace. An R-type thermocouple protected by the alumina tubing was contacted with the crucible to allow the temperature measurements as accurately as possible. Subsequently, the specimen was heated to the aimed temperatures with a heating rate of 40 K/minute, after which the specimen was held for 2 hours to be sintered under an air atmosphere. The aimed temperatures were chosen at 1000, 1250, 1400, 1500 and 1600°C to understand the evolution of sintering with an increase in temperature. After the sintering experiments, the specimens were taken out of the crucible with crushing the sintered sand by a hammer because the specimens were stuck on the crucible wall especially at the higher temperatures of 1500 and 1600°C. Therefore, the specimens after taken out were broken to pieces of granular lumps.
Experimental apparatus.
The chemical composition of the silica filler sand employed in this study is shown in Table 1. The silica sand consists mainly of silica with a small amount of Al2O3, Na2O and K2O. Density of the sand was measured as 2.57 g/cm3 by Archimedian method. The appearance of the sand is shown in Fig. 2 and the sand material has granular morphology with relatively uniform size distribution. The distribution of the particle size was measured using sieves as shown in Table 2. It can be seen that the majority of the particles have the size of 0.5 to 1 mm.
| SiO2 | Al2O3 | K2O | Na2O |
|---|---|---|---|
| 93.0 | 4.6 | 2.0 | 0.4 |
Appearance of the sand specimen.
| Size (mm) | +1.00 | +0.50 | –0.50 |
| mass% | 22.2 | 77.7 | 0.1 |
Minerals: Silica (SiO2), Alkali feldspar (Na,K)(Si,Al)4O8
As mentioned above, the size of the sand particles are relatively large so that the sand has been crushed to have a smaller size distribution to determine the effect of particle size on sintering behavior. A vibrating mill was used to crush the sand to have the sand particles smaller than 0.106 mm. The crushed sand specimen was applied for the sintering experiments with the procedure which was basically the same as stated above. The specimens were heated at 1250°C for 12 hours and at 1600°C for 4 hours.
2.3. ObservationAfter every experiment, macroscopic observation by eyesight was made to evaluate how sintering proceeded depending on the condition. It was so difficult to express the sintering state quantitatively. For reference, therefore, the sintering index indicated in Table 3 with the schematic illustrations was used to express the sintering state in this study. Index 1 corresponds to the sand without being sintered. Index 3 corresponds to the state that the sand can be easily broken with a hammer implying the medium sintered state. Index 5 corresponds to the state being stiffly sintered.
;%0A%09%09%09newWindow.document.open();%0A%09%09%09newWindow.document.write('<img src=%22./Graphics/54_1823_tbl_3.jpg%22>');%0A%09%09%09newWindow.document.close();%0A%09%09)
|
After the eyesight observation, the specimens were mounted with resin material penetrated into the cavities. They were thereafter polished with SiC abrasive papers and finished with diamond paste of 1 μm particles. Au was coated on the specimen surface to allow microscopic observations by an SEM (Scanning Electron Microscope; HITACHI S-3500N). Image analysis of SEM image pictures was performed to quantitatively indicate the sintering state associated with the sintering index. Porosities in an image were specified to obtain area percent with the help of image analysis software of WinROOF version 7.0. Five images were taken for analysis. Thereby, porosity proportions were obtained and the average values were used for further considerations. Qualitative analysis was also carried out by an EDS (Energy Dispersive X-ray Spectrometry; HORIBA EMAX-7000) to observe the element distribution in order to understand the evolution of liquid phases. Quantitative analysis was also carried out as for every phase after checking the composed phases.
Before the observations of the sintered specimens, the original sand particles of Fig. 2 were quantitatively analyzed to identify the mineral phases as well. Five points were randomly measured for each phase and the average values were taken for analysis.
2.4. Thermodynamic CalculationRecently, it is of general recognition that thermodynamic software is a powerful tool to simulate the phase evolutions. In this study, thermodynamic software of FactSage version 6.33) was applied for the calculation. Further, thermodynamic software of MELTS version 10.4.14) typically used by geologists to simulate magma constituents was also applied for the calculation. In the both calculations, the conditions were taken as SiO2–Al2O3–K2O–Na2O quaternary system under an air atmosphere.
Table 4 shows the analytical result of the particles contained in the original silica sand of Fig. 2. The sand consists of pure silica particles and two types of the particles composed of SiO2, Al2O3, Na2O and K2O. It was obvious from the observation that majority of the particles were silica with a minor amount of the latter minerals. It should be noted that the distributions of the SiO2, Al2O3, Na2O and K2O were confirmed to be completely homogeneous by the mapping analysis. Referring to the book of rock-forming minerals,5) the particles composed of SiO2, Al2O3, Na2O and K2O phases are identified as K-rich and Na-rich alkali feldspars as denoted in Table 4. Comparing the two types of feldspars, K-rich alkali feldspars was even more dominant consistently with the chemical composition of the sand as shown in Table 1.
| SiO2 | Al2O3 | K2O | Na2O | Remarks | |
|---|---|---|---|---|---|
| 1 | 100 | – | – | – | Silica |
| 2 | 67.3 | 18.3 | 12.9 | 1.7 | K-rich Alkali Feldspar |
| 3 | 70.9 | 19.4 | – | 9.7 | Na-rich Alkali Feldspar |
Average values of five points are listed.
Figure 3 shows the appearance of the silica filler sand specimens along with the element distributions typically observed at each temperature. As indicated below the macroscopic pictures, the sintering index increases with increasing temperature. This result implies that sintered sand becomes stronger with an increase in temperature.
Appearance of the filler sand specimens and element distributions with a schematic illustration at 1600°C.
Subsequently, the porosity proportions were measured for the specimens except at 1000°C. Figures 4(a) and 4(b) show a typical example of the SEM image and the measured porosity proportions plotted against temperature, respectively. As can be seen, the porosity proportion decreases with increasing temperature, which may imply that melting proceeds as temperature rises. Further, the experiments with the crushed sand brought lower values compared at the same temperatures. As a result, the porosity proportion could quantitatively express the sintering state. This melting behavior and the lower proportions with the crushed sand will be described later in detail.
(a) SEM image of the specimen after sintering experiment of 1400°C and (b) measured porosity proportion.
In the mappings of Fig. 3 that show the element distributions, the red, blue and green colors indicate the distributions of Al, Si and K, respectively. One can see that the particles are as they are before the experiment at 1000°C and that the sand specimen contains silica particles together with alkali feldspars that contain Al2O3, Na2O and K2O. This result is well consistent with the observation of the original silica sand, as described above. At 1250°C, it can be seen that the red regions are more spread in the silica matrix than at 1000°C. The red regions appear to penetrate into between the silica particles at 1400 and 1500°C. Finally at 1600°C, the red regions are more dominated than at the lower temperatures. Furthermore, the gradation at 1600°C in the colors is dimmer than at 1500°C. Therefore, it is so difficult to differentiate the regions that an illustration was given for the phase distribution at 1600°C. The red regions of the pictures higher than 1250°C are determined as liquid phase as described below.
The red regions observed in the specimens after the sintering experiments were analyzed as listed in Table 5 and shown in Fig. 5. To make sure if the red regions are molten, we have referred to the phase diagrams of KAlSi2O6–SiO2 and KAlSi3O8–NaAlSi3O8 pseudo-binary systems.5) The diagrams tell us that the melting point of alkali feldspars is 1150°C. Thus the red regions seen above 1250°C can be determined as liquid phases, which are originated from the contained alkali feldspars. This result shows that the silica sand sinters through the evolution of liquid phases. Besides, dilution of Al2O3, Na2O and K2O in the liquid phase is obvious with increasing temperature implying that silica dissolves into the liquid phases at the elevated temperatures. At 1600°C, silica occupies the liquid phase as much as 88.1 mass%.
| Sand | temperature | SiO2 | Al2O3 | K2O | Na2O | Liquid proportion |
|---|---|---|---|---|---|---|
| Original (0.5 mm ~) | 1000 | Liquid phase was not generated. | 0 | |||
| 1250 | 68.3 | 17.9 | 12.1 | 1.6 | 26 | |
| 1400 | 76.0 | 15.8 | 7.2 | 1.0 | 29 | |
| 1500 | 78.5 | 13.3 | 6.2 | 2.0 | 35 | |
| 1600 | 88.1 | 7.3 | 3.9 | 0.7 | 63 | |
| Crushed (~ 0.106 mm) | 1250 | 83.6 | 10.3 | 5.0 | 1.1 | 46 |
| 1600 | 91.5 | 5.1 | 2.7 | 0.7 | 90 | |
Average values of five points are listed.
Effect of temperature on liquid proportion and composition.
Meanwhile, every silica phase observed in Fig. 3 was analyzed as pure silica which was the same as in the original sand. This implies that the silica phases do not contain Al2O3, Na2O and K2O. Condensation of these minor oxide species in the liquid phase can indicate the liquid proportion assuming that the density of the solid is the same as that of the liquid. In this study, employing Al2O3 as the oxide, one can obtain the proportion of liquid phases by the following equation:
| (1) |
To better understand the evolution behavior of the liquid phase, the compositions were plotted on the phase diagram of SiO2–Al2O3–K2O system6) in Fig. 6, counting Na2O equivalently as K2O. As mentioned above, no phase change takes place until 1000°C. Thereafter alkali feldspars start to melt at 1250°C. Silica particles are considered to be dissolved into the liquid phases with increasing temperature. Then, SiO2 is getting more dominant in the liquid phase at 1600°C and closer to the bulk composition. The viscosity of the liquid phase at 1600°C is estimated to be as high as 108 (poise).7)
Liquid compositions plotted on SiO2–Al2O3–K2O ternary phase diagram.
Basically, filler sand plays a role of plugging to prevent leakage of molten steel. It is considered from this experiment that this role can be attributed to the viscous property of the liquid phases, with which the sand particles are able to be connected tightly with each other. In other words, sintered sand can behave as if it is a refractory lid over a sliding gate at the interface between molten steel and sand. With getting closer to the gate, the temperature becomes lower with keeping disconnected states without being sintered. It seems that this combination can allow molten steel to be safely poured after opening a sliding gate.
3.3. Thermodynamic CalculationsAccording to the diagram of Fig. 6, the sand should be all molten because the liquidus temperature is below 1600°C at the bulk composition. This reason may be attributed to the fact that the whole system does not reach equilibrium in the sintering experiments. Therefore, thermodynamic considerations were made to better understand the sintering behavior. Figures 7 and 8 show the results by FactSage and MELTS, respectively. Basically, the both results show that alkali feldspars start to melt at around 800°C by reacting with silica. With an increase in temperature, silica phases decrease while liquid phases increase, and finally the whole system becomes molten at around 1550°C. However, the behaviors in the alkali feldspars are different from each other at the temperatures lower than 1100°C. In particular, decomposition of feldspars to form corundum can be realized in Fig. 7, which has not been observed in the sintering experiments. This fact is attributed to the definition of the solid solution of alkali feldspars. It is known that K-rich and Na-rich alkali feldspars form a solid solution.5) FactSage version 6.3 does not consider the definition of this solid solution. This result tells us that MELTS software should be more accurate for the calculation of the present sand system.
Variation of phase proportion as time calculated by FactSage.
Variation of phase proportion as time calculated by MELTS.
In the sintering process, the evolution of the liquid phases is the most important to be accounted for. The liquid proportions obtained by the both calculations are therefore plotted in Fig. 9 along with the experimentally obtained values. As realized, the experimental values are closer to the liquid proportion calculated by FactSage. Besides, the tendency is quite similar to each other in the gradual increase from 1250 to 1500°C and the rapid increase above 1500°C. However, the fact closer to the line by FactSage is inconsistent with the above explanations about the corundum formation. We therefore assumed that the deviation between the experimental results and the equilibrium state was unexpectedly very large at every temperature. Therewith, further experiments were performed to understand the factors of kinetics.
Liquid proportion plotted against temperature.
The effect of particle size on the evolution of liquid phase has been examined to ascertain which software can correctly predict the sintering behavior with respect to equilibrium. That is why the holding times have been decided to be prolonged. Figure 10 shows the experimental results using the crushed sand, indicating that the crushed sands have more sintered than the original sands. The appearances of the sintered specimens by the crushed sands look like rocks. Particularly, the sintering index increased dramatically from 2 to 5 with the crushed sand at 1250°C. This behavior of more sintered with the crushed sand is consistent with the above-mentioned results of Fig. 4.
Effects of particle size and holding time on sintering index.
The microstructures of the specimens are shown in Fig. 11. Clearly, the liquid regions are more spread in the sintered specimens by the crushed sands. Quantitative analysis shows that silica contents in the liquid phases are higher than by the original sands as provided in Table 5. The liquid proportions obtained by Eq. (1) are also higher than by the original sands. The liquid proportions were plotted in Fig. 12 along with the lines calculated by the software. It is evident that the liquid proportions experimentally obtained are in good agreement with the line by MELTS. Especially, the value at 1250°C is right on the simulation line. The experimental data of 1600°C position 10 mass% lower than the line. This could be caused by the viscous liquid phases which obstruct the dissolution of the silica particles. This result of agreement indeed proves that MELTS software is able to accurately simulate the present sand system.
Effects of particle size and holding time on microstructures of the specimens.
Calculated liquid proportion curves plotted against temperature along with experimental data.
The sintering behavior can be predicted by thermodynamic simulation, as described above. It should be notified, however, that the deviation exists between the simulations based on equilibrium and the actual sintered situations. The particle size of filler sands could be empirically determined to prevent the state fully sintered to avoid complete blockage. It is expected with this study that designing the size distribution can be more optimized by accounting for the sintering behavior.
A study has been undertaken to clarify the sintering behavior of silica filler sand. Sintering experiments were firstly conducted and the experimental data were compared with the thermodynamic calculations by FactSage and MELTS software. The following conclusions were drawn:
(1) The sintering of the silica sand proceeded through the evolution of liquid phases, which were originated from the contained alkali feldspars.
(2) The liquid proportion increased with an increase in temperature as a result of the dissolution of silica phases into the liquid phases.
(3) It was considered that the viscous liquid phases connecting silica particles contributed for the strength after sintering.
(4) According to the thermodynamic calculations, MELTS software could sufficiently simulate the phases in equilibrium with each other.
(5) The comparison of the experimental data with the thermodynamic simulations demonstrated that the whole system of the silica sand did not attain equilibrium.
(6) It was inferred that the filler sand should be designed to prevent the state fully sintered by controlling the particle size with accounting for the sintering behavior.
