Microstructural and Microscopic Investigations of MgO–C Bricks – Relationships between Gas Permeability and Apparent Porosity or Pore Size by Controlling Relative Density

Changes in pore morphology from measurements of gas permeability, apparent porosity and pore size distribution of MgO–C bricks with controlled relative density were investigated. When the apparent porosity of the MgO–C bricks with increased relative density was decreased, the gas permeability was also decreased. In addition, with decreasing average pore size, MgO–C bricks decreased their permeability. These results suggested for the first time that a pore morphology formed in MgO–C bricks is largely involved. The fractal dimension of the pores was increased with increasing the relative density of MgO–C bricks. This fact indicated that fragmentation of pore and diffusion of fragmented pores in MgO–C bricks were produced through the formation processes of the refractory texture.

Introduction The requirements for MgO–C bricks used as refractories for linings such as molten steel ladle and secondary refining container are mainly corrosion resistance, spalling resistance, and oxidation resistance [1]. In particular, it is generally considered that dense microstructure is a required property from the viewpoint of corrosion resistance [2]. The dense structure is achieved by controlling the packing bulk density under each condition of particle size configuration, kneaded mixture condition, and moulding pressure [3] and can be confirmed by quantitatively measuring the amount or shape of pores. The apparent porosity of the refractory can be expressed as the volume fraction of pores in the sample, but it does not provide any knowledge about the size and shape of the pores. On the other hand, the pore size distribution is useful for knowing the size of the pores. The gas permeability is the amount of gas that has passed through the microstructure of the brick sample, and depends on the size and shape of the pores. In particular, it has been reported that the gas permeability of graphite-containing refractories depends on the moulding direction [4]. In other words, the gas permeability can be said to be a measurement method that is sensitive to changes in the microstructure. Therefore, it may be possible to complementarily evaluate the pore morphology of refractory microstructure by examining the correlation between these three physical properties. Kondo et al. [5] show that there is a proportional relationship between the gas permeability and the measured apparent porosity of shaped refractories with different materials. This proportional relationship is derived from the apparent porosity of various refractories (for example, clay, silica, roseki, magnesia-chrome, dolomite) as variables, and the relationship between the same materials has not been examined. Carbon-containing refractories were in the early days of their development and gas permeability was unexamined at that time. After that, Sikano et al. reported [3] that the gas permeability after heat treatment with reducing atmosphere was more useful than the bulk density and porosity as a microstructure evaluation of MgO–C bricks, and the gas permeability was presented as one of the indices for the MgO–C brick microstructural properties. Thereafter, there have been no reports on the gas permeability of MgO–C bricks. In addition, there are no reports comparing the gas permeability to apparent porosity or pore size distribution to evaluate the microstructure of MgO–C brick. From the above, the relationship between the gas permeability, apparent porosity, and pore size distribution of MgO–C bricks due to the change in pore morphology caused by microstructural control remains to be investigated. In this paper, the microstructural densification index of MgO–C bricks is expressed as the relative density of MgO–C bricks when the bulk density at the time of packing (moulding) is divided by the theoretical density, which is the sum of the true densities of all raw materials. By evaluating the produced MgO–C bricks with several types of relative densities, changes in pore morphology during the mixture packing process can be examined. Examining the microstructural changes of MgO–C bricks when the relative density is controlled to change the pore morphology is useful for improving the moulding process technology, which is an important elemental technology. Fractal analysis was applied to study the small changes of the pores in these dense microstructures. It is necessary to quantify the process of microstructure change by analysing the fractal dimension of the pores of bricks with different relative densities. Based on the above background, we measured the gas permeability, apparent porosity, and pore size distribution of microstructures in which the relative density of MgO–C bricks was controlled, and examined their relationships. 2 Experimental method Tab. 1 shows the raw materials used in this experiment. Middle grain (1–0 mm) of electro- fused magnesia raw material contains 0,8 % of fine particles (<0,075 mm). The particle size distribution of the electro-fused magnesia was calculated with the q-value in Andreasen equation for dense packing [6] set to 0,40, and flake graphite and phenol resin (novolac type) were added as a binder and kneaded. The kneaded product was moulded into the dimensions shown in Fig. 1 a by a friction screw press machine, and the relative density was controlled by adjusting the pressure and the number of pressurizations. The relative density after moulding was calculated from Eq. (1) [7], and 13 types of samples with different relative densities were prepared as shown in Tab. 2. In eq. 1, the bulk density indicates the density of the green brick immediately after moulding, and the theoretical density indicates the total density of the raw materials used. In this sample, the theoretical densities of fused magnesia, flake graphite, and phenolic resin were 3,60 g . cm–3, 2,02 g . cm–3, and 1,17 g . cm–3, respectively. Relative density (%) = bulk density (g . cm–3) (theorotical density(g . cm–3) Å~ 100 (eq. 1) Each moulded sample was dried at 250 ÅãC for 5 h. The dried sample was embedded in coke grains and calcined at 1400 ÅãC (5 ÅãC . min–1) for 5 h. The apparent porosity after drying and firing was measured by the Archimedes method using kerosene. Tab. 2 also shows the apparent porosity of the sample and the bulk density after firing. In order to measure the gas permeability, the heat-treated specimens were processed into a shape of ∅ 50 mm x 50 mm perpendicular to the pressing direction as shown in Fig. 1 b. This was set in the gas permeability measurement device (S-1000, SERIO Inc.) and measured at 10 kPa, 20 kPa, and 30 kPa of differential pressure of nitrogen gas in accordance with JIS R 2115, and the average value of each measurement was used as the gas permeability. The pore size distribution of the heat-treated sample was measured by the mercury intrusion method (Auto Pore V9620, Shimadzu Corp./JP). A method of visualising the pores themselves with a fluorescent dye was used to observe changes in the pore morphology [9]. The sample was vacuum-impregnated with an epoxy resin in which a fluorescent dye was dissolved, the cast surface was mirror-polished, and the pore morphology was observed with a fluorescence microscope. The observed image was binarized into pores and constituent raw materials, and the fractal dimension of the pores was calculated from the binarized image using the box-counting method [10]. The fractal dimension was calculated using the fractal analysis system (National Agriculture and Food Research Organisation) [11]. 3 Results and discussion 3.1 Correlation between gas permeability and apparent porosity Fig. 2 shows the correlation between the relative density of MgO–C bricks after moulding and the apparent porosity after heat treatment. The apparent porosity decreased with increasing relative density. The relationship between the relative density and the apparent porosity can be approximated linearly, and when this was extrapolated to the theoretical density (100 %), the apparent porosity was about 8 %. This indicates the volume fraction of pores in the close-packed microstructure, which is caused by differences in the thermal expansion/ shrinkage behaviour of the raw materials and the dispersal of binder volatiles. They can be attributed as follows: 1) Residual expansion was 2,1 % after heat treatment. 2) Volatilization of the binder was 4,0– 4,3 % due to firing (From the binder amount of 2,4 mass-%, the volatile amount of binder is about 70 %. Volatile matter is converted to volume %, which is about 4,0–4,3 %. 3) Volume shrinkage was 2,0–2,5 % due to the carbonization of the phenolic resin. So, the total of these was about 8 %, which was almost the same as the apparent porosity predicted from the extrapolated value. Therefore, even if the MgO–C bricks are packed to the theoretical density, it means that about 8 % of pores will be formed after heat treatment. Fig. 3 a shows the gas permeability by changing apparent porosity. The gas permeability decreased with decreasing the apparent porosity. This means that when the volume ratio of the pores in the brick decreases, the amount of gas passing through the structure is suppressed. This relationship is the first quantitatively reported fact in MgO–C bricks. Looking at this relationship in more detail, the differential coefficient representing the variation of gas permeability was 0,25 when the apparent porosity was lower than about 12 %, but it increased to 0,88 when the apparent porosity was higher than about 12 %, an increase of about 3,5 times. It is suggested that the change in apparent porosity is not linearly related to the gas permeability of MgO–C bricks. The Reynolds number, which indicates the gas flow of nitrogen in MgO–C bricks, was 10–20 in all brick samples. Since the gas flow of nitrogen in the microstructure of MgO–C brick is a laminar flow, it was examined using Kozeny-Carman’s equation [12] expressed in eq. 2:  Therefore, it is considered that the increase or decrease in the gas permeability of MgO–C brick depends on the pressure loss of the particle packing system. On the other hand, the change in gas permeability microbefore and after an apparent porosity of 12 % in Fig. 3 a could not be explained by the theoretical equation. The pores labelled with fluorescent dye were observed with a fluorescence microscope to confirm the morphology of the pores. Fig. 4 a–c show fluorescence microscope images with relative densities of 91,6 %, 94,5 % and 96,2 %, respectively. Observations were focused on the matrix portion of the brick microstructure. Fluorescent areas with high brightness show resin-impregnated pores, while dark areas show MgO or graphite grains. Microstructural observation using such fluorescent emitting phenomenon is a method that can clearly distinguish the pores in the matrix compared to the images obtained from optical microscopy or scanning electron microscopy and has the advantage of tracking the shape change of the pores. As can be seen from the observation image, the shape of the pores was long, narrow, and angular, which was not the generally considered rounded shape. The angular shape is considered to be due to the projection of the shapes of adjacent MgO coarse grains (hereinafter, coarse grains are defined as 5–1 mm, middle grains are defined as 1–0 mm, and fine particles are defined as particles <,075 mm: see comments in Tab. 1) and graphite grains. Changes in the visualised pore region tended to decrease with increasing relative density. In the process of pore shrinkage, in addition to the gradual reduction of voids between MgO grains of all grain sizes, MgO grains of all angular grain sizes were packed in a way that they changed their orientation in the brick microstructure. It was also found that the graphite grains were bent along the MgO grains of all particle sizes. In this way, the process of pore shrinkage was clarified through pore observation using a fluorescent field of view. It was once again identified that the shrinkage of the pore was caused by the mechanical deformation of graphite grains in addition to the packing of middle and fine grains between the coarse MgO grains. Tab. 3 summarises the texture changes in above and below an apparent porosity of about 12 % (relative density of about 94 %). When the movement of particles during moulding is distinguished by the apparent porosity (relative density), it was dominant to reduce the pores around the middle and fine MgO particles around the coarse particles at above 12 % (<94 %). On the other hand, at <12 % (>94 %), the movement of middle and fine MgO grain around coarse grains began to be restricted. Instead, it is considered that the pores around the graphite grains are reduced, and the flexibility of graphite further fills the pores. 3.2 Correlation between permeability and pore size Fig. 5 shows the cumulative pore volume concerning the pore diameter of brick samples with different relative densities. When the relative density increased from 91,6 % to 94,5 %, the pore volume slightly decreased in all pore diameters, and when the density increased from 94,5 % to 96,2 %, the pore volume decreased significantly. The decrease in pore volume was particularly significant in the pore size range of 0,005–5 μm. It is considered that the pores are gradually crushed and the structure is densified due to the increase in the relative density. Fig. 6 shows the pore size distribution of brick samples with relative densities of 91,6 %, 94,5 %, and 96,2 % respectively. All samples had pores in the range of 0,05– 20 μm. This range was divided into three regions: 0,05–0,5 μm (region i), 0,5–5 μm (region ii), and 5–20 μm (region iii), respectively. Comparing the changes in the peak shape for each sample, in the region (i), a broad peak appeared in the sample with a relative density of 91,6 %, a shouldershaped peak appeared in the sample with a relative density of 94,5 %, and no peak was present in the sample with a relative density of 96,2 %. In the region (ii), the peak intensity decreased with increasing the relative density. On the other hand, in the region (iii), the peak intensity decreased with increasing the relative density from 91,6 % to 94,5 %, but did not decrease with increasing the relative density from 94,5 % to 96,2 %. These results indicate that the pores in the range of 0,05–5 μm, which are in the region (i) and (ii), tend to disappear when the structure is densely packed, while the relatively large pores in the range of 5–20 μm tend to remain when even after the structure is densely packed. 3.3 Fractal analysis Fractal analysis was applied to study the microstructural changes of pores in low porosity structures. By analysing the fractal dimension of pores in bricks with different relative densities, the process of microstructural change was quantitatively investigated. The fractal dimension of the pores was calculated using the binarized fluorescence microscope image shown in Fig. 4. Regarding the fractal analysis of pore shape [13], when the figure is divided with the similarity ratio r as the division unit, the self-similarity is established between the number of divisions N (r) and the similarity ratio r according to eq. 3. Here, D represents the fractal dimension. By transforming eq. 3, the fractal dimension D is defined by eq. 4 [14]. The box-counting method is a method for determining the fractal dimension. When the target figure is divided into an arbitrary number of square cells, and the relationship between the size of the cells and the total number of cells containing the figure is approximated by a linear line on a double-logarithmic chart, the fractal dimension is obtained from the absolute value of the slope of the linear line. Figs. 7 a–c show the relationship between the fractal dimension and (a) apparent porosity, (b) gas permeability, and (c) average porosity of MgO–C bricks, respectively: (a) The fractal dimension increased with decreasing the apparent porosity. The process of decreasing apparent porosity, i.e., the gradual discharge and shrinking of pores, could be explained by the increase in the fractal dimension of the pore morphology. The increase in fractal dimension is considered to be the complication of pore morphology (pore division, dispersion), suggesting that the shape of the pore crush is reflected in the complication. Since the decrease in porosity is related to the increase in relative density during the moulding process, the fractal dimension may be related to both. (b) In relation to the gas permeability, the decrease gradient of the fractal dimension tended to be small in the structure with high gas permeability higher than 1,5 Å~ 10–14 m2. On the other hand, the fractal dimension increased sharply in a narrow permeability range from 1,0–1,5 Å~ 10–14 m2. This result indicates that the fractal dimension increases rapidly with the densification of the microstructure below a gas permeability of 1,5 Å~ 10–14 m2. It has been shown that the prediction that pore morphology will become finer in densified microstructure with an increased relative density of bricks is more likely. It is considered that this is because the relative density increased and the microstructure was close-packed, and the pores were subdivided and diffused uniformly into the microstructure [15]. (c) The average pore diameter had a linear relationship with the fractal dimension. These results suggest that the pores to be crushed during the moulding of MgO–C bricks are selected in a stepwise manner depending on the relative density after moulding. As mentioned earlier, up to a relative density of about 94 %, the microstructure around the coarse or middle MgO grains is gradually crushed. And above the relative density of 94 %, the movement of the coarse MgO grains begins to be restricted and pores around the fine MgO grains and the graphite, mainly in the matrix, are crushed. The relationship between the gas permeability and the apparent porosity or the gas permeability and the average porosity may select the morphology (size or volume ratio) of the crushed pores. In this study, it is considered that the change in the pore shape of graphite-containing bricks during the moulding of shaped bricks can be classified into two types. 4 Summary In this study, the following findings were obtained by examining the correlation between the gas permeability, the apparent porosity, and the pore size distribution of MgO–C bricks. (1) The apparent porosity and the gas permeability decreased as the relative density increased. It might suggest that the increase and decrease of the gas permeability depend on the pressure loss of the particle packing system. The gas permeability decreased as the average pore size decreased. (2) From the relationship between the gas permeability and the apparent porosity or the gas permeability and the average porosity, the increment of gas permeability changed before and after the relative density during moulding was about 94 %. (3) Up to a relative density of about 94 %, the pores around the medium and fine grains around the coarse MgO grains are reduced, and above about 94 %, the pores around the matrix-centred fine graphite grains begin to be reduced, which is thought to change the increments of apparent porosity and gas permeability. (4) It was found that the fractal dimension of the pores of MgO–C bricks increased with decreasing the apparent porosity, and decreased with increasing the gas permeability. These results suggest that the evaluation of bricks with several relative densities can be used to evaluate the dynamic process of microstructural changes during the pressurizing filling of the mixture, which may provide a new method for evaluating microstructure.

References

[1] TARJ, ed.: Taikabutsu Techo 12th, (2015) 158

[2] TARJ, ed.: Taikabutsu Techo 12th, (2015) 160

[3] Shikano, H.; et al.: Taikabutsu 43 (1991) [2] 66–73

[4] Asakura, H.; et al.: Taikabutsu, 63 (2011) [12] 630–641

[5] Kondo, R.: Tako Zairyo, Gihodo (1973) 247

[6] Siraki, Y.: Seramikusu Techo, Gihodo (1972) 156

[7] Yogyo Kyokai Ed.: Seramikkusu Jiten, Maruzen (1986) 445

[8] Asakura, H.; et al.: Taikabutsu 61 (2009) [3] 142

[9] Maeda, K.; Hashimoto, S.; Yamaguchi, A.: Taikabutsu 48 (1996) [11] 592

[10] Vicsek, J.; Miyazima, S.: Furakutaru Seichogensho, Asakura Shoten (1990) 6

[11] Sasaki, H.; Shoji, A.; Suyama, T.: Jap. Grass. Sci. 44 (1998) 394–395 (Separate Pages)

[12] Carman, P.C.: Trans. Inst. Chem. Engin. 15 (1937) 150–166

[13] Tsuchinari, A.; et al.: J. Ceram. Soc. Jpn. 99 (1991) [7] 561–566

[14] Takayasu, H.: Furakutaru, Asakura Shoten (1989) 7–25

[15] Tsuchinari, A.; et al.: J. Ceram. Soc. Jpn. 100 (1992) [2] 216–229