Pore evolution and its effect on slag resistanceof Al2O3–SiC–C castables

Yucheng YIN, Yonghe LIANG,³ Shan GE, Zhiqiang LIU, Jianhua NIE and JiuChang LU

The Hubei Province Key Laboratory of Refractories and Ceramics, Ministry-Province jointly-ConstructedCultivation Base for State key Laboratory, Wuhan University of Science and Technology, Wuhan, China

Al2O3­SiC­C castables with different particle size distributions (PSD) corresponding to different Andreassen distributionmodulus (q) values, were fabricated, and their apparent porosity, permeability, pore size distribution and mean pore size,examined, after heat-treatment at different temperatures. The apparent porosity of Al2O3­SiC­C castable samples dried at 110°Cfor 24 h or fired at 1450°C for 3 h, decreases with increasing the q value from 0.21 to 0.23, but starts to increase on furtherincreasing the q value to >0.23, i.e., it reaches the minimal value when q = 0.23.The change of permeabiliy with q is similar,although it reaches the minimal value when q = 0.22. The mean pore sizes of Al2O3­SiC­C castable samples dried at 110°C for24 h increase with increasing q from 0.21 to 0.25, whereas for samples fired at 1450°C for 3 h, they have no obvious regularity.Pore size distributions of Al2O3­SiC­C castable samples after treated at different temperatures, are evidently scattered whenq = 0.21, 0.22 and 0.25, but tend to become continuous when q = 0.23 and 0.24, in particular when q = 0.23. The Al2O3­SiC­Ccastable with q = 0.23 shows the best slag resistance because of its relatively narrow pore size distribution. In addition, our resultsreveal that PSD has little, whereas temperature has more obvious, effect on apparent porosity, permeability, mean pores size andpore size distribution of Al2O3­SiC­C castables.©2013 The Ceramic Society of Japan. All rights reserved.

Key-words : Particle size distribution, Al2O3–SiC–C castable, Apparent porosity, Permeability, Pore size distribution, Slag resistance

[Received May 15, 2013; Accepted July 20, 2013]

1. Introduction

Many refractory materials are readily wetted by a moltenslag, resulting in deep slag penetration into their structure. Asa result, severe corrosion occurs at high temperatures due tothe subsequent reactions between the slag and their compo-nents. Although the corrosion reactions cannot be prevented, itis possible to improve the overall slag resistance by decreasingthe slag penetration extent.It is well known that a slag penetrates into a refractory material

via pores.1) Pores are usually characterized by pore parameterssuch as apparent porosity, permeability, pore size distributionand/or mean pore size. These parameters have close relationshipswith the refractory’s slag resistance. For refractory castables,pores are mainly resultant from two sources. One is related to theloss of water (either free water or bonded water in the cementhydration products) and/or organic components,2) and the otherto the particle packing density which depends on the PSD indi-cated by the distribution modulus (q) value in the Andreassen’smodel.3) The PSD of a refractory castable has an importanteffect on its packing density and flowability. As suggested by Y.Kutmen Kalpakli, by optimizing PSD, voids between the coarseparticles could be efficiently filled by the fine particles,4) resultingin less pores in the refractory. In addition, according to BjørnMyhre et al.,5) the PSD of a refractory castable, in particular,that of the superfine part, has a great effect on its flowability.Unfortunately, few investigations on the relationship betweenPSD and pore parameters have been carried out so far.Al2O3­SiC­C castables for blast furnace main trough are

subjected to severe conditions including thermal shock, oxida-

tion, molten iron and slag corrosion.6)­9) Many measures havebeen attempted to prolong their service lives, such as addition ofan antioxidant and component/composition improvement.10)­17)

Nevertheless, the optimization on their structures has not beenpaid much attention.Understanding of the relationship between PSD, pore parame-

ters and corrosion resistance of a refractory would be very usefulfor structural optimization of the refractory, and thus for thereduction in the slag penetration. Considering this, pore evolutionand its effect on the slag resistance of Al2O3­SiC­C castablesprepared with different PSD have been investigated in this work.

2. Experimental procedures

The Andreassen’s equation was used to design the particle sizedistribution of Al2O3­SiC­C castables.

CPFT ¼ 100 � d

D

� �q

ð1Þ

Where CPFT, d, D and q indicate the cumulative percentagefiner than, particle size, the largest particle size and the distri-bution modulus, respectively. For a refractory castable to achieveself-flow, the q value should be between 0.21­0.26.18) In thisstudy, q was allowed to vary from 0.21 to 0.25 with the intervalof 0.01, and the largest particle size of grains was 8mm. ThePSD calculated according to Eq. (1) are shown in Table 1.Following our previous experiences, the contents of calcium

aluminates cement(CAC), silicon fume, micro silica and pitchball, were fixed at 3, 4, 3 and 3wt%, respectively. Also, con-sidering that the commonly used amount of silicon carbide was13­16, 13wt% of silicon carbide was added, which included8wt% of 1­0mm silicon carbide grains and 5wt% of ¯0.074mm silicon carbide power. The amounts of other components³ Corresponding author: Y. Liang; E-mail: yucheng.yin@gmail.com

Journal of the Ceramic Society of Japan 121 [10] 873-879 2013 Paper

©2013 The Ceramic Society of Japan

DOI http://dx.doi.org/10.2109/jcersj2.121.873

873

were derived according to Table 1, thus, particle size gradingscorresponding to different q values were obtained (Table 2).Castable samples of 160mm © 40mm © 40mm were pre-

pared based on Table 2. The dry raw material powders weremixed initially in a mixer for 30 s, followed by a one-stepaddition of 6.5wt% water in 10 s while the mixer was running.After 5min wet mixing, samples were cast under vibration andthen cured for 24 h at a constant temperature and humidity ina curing box, dried for 24 h at 110°C and fired at 1450°C for 3 hin a furnace. To protect carbon in the samples from oxidation onfiring, the samples were embedded in coke powders containedin a refractory saggar sealed with a refractory lid.Apparent porosity, permeability, mean pore size and pore

size distribution of samples prepared with different PSD, wereexamined, after heat-treatment at different temperatures. Apparentporosity was measured according to ISO 5017, permeability wasdetermined according to ISO 8841, and mean pore size and poresize distribution were examined by Autopore IV9510 (Micro-meritics of USA). Flowability measurements were performedusing the flow cone method described in ASTM. The testingwas performed on a vibration table set at double amplitude of0.75mm. After filling the test cone with the castable, the formerwas removed so as to allow the latter to spread under its ownweight. The castable is then subjected to 15 s vibration, and theresultant spreading is taken as the vibration flow value.The slag corrosion test was carried out using an induction

furnace. Castable samples of 160mm © 40mm © 20mm pre-pared with different PSD were pre-inserted into the lining ofthe induction furnace (21WGJL0.025-100-2.5P). Then iron (6 kg)and slag (2 kg) were fed into the furnace. The samples werefired at 1500°C for 3 h in air under the simulated condition ofblast main trough. The schematic diagrams of the experiment areshown in Fig. 1.An Axioskop40 optical microscope (Zeiss, Germany) and

an scanning electron microscope (SEM, FEI, Nova 400 Nano)equipped with energy dispersive X-ray spectroscopy (EDS,

EDAX) were used to carry out microstructural analysis on sometypical samples heat-treated at different temperatures, so as toassist clarifying the pore evolution mechanism.

3. Results and discussion

3.1 Effect of PSD on apparent porosityThe apparent porosities of Al2O3­SiC­C castables prepared

with different PSD after treated at different temperatures areshown in Fig. 2. The apparent porosities of Al2O3­SiC­Ccastables dried at 110°C or fired at 1450°C, decrease withincreasing q from 0.21 to 0.23, but start to increase upon furtherincreasing q to >0.23, which indicates that the minimum porosityis achieved when q = 0.23. It is well known that finer particlescan fill the voids between coarse grains, which reduces thefriction between coarse grains, thus improving the flowability of

Table 1. PSD calculated according to Andreessen’ model, wt%

q value 8­5mm 5­3mm 3­1mm 1­0.088mm 0.088­0mm

0.21 9.40% 9.22% 16.77% 25.83% 38.79%0.22 9.82% 9.59% 17.30% 26.21% 37.08%0.23 10.25% 9.95% 17.82% 26.54% 35.44%0.24 10.67% 10.31% 18.32% 26.83% 33.88%0.25 11.09% 10.66% 18.79% 27.08% 32.39%

Table 2. PSD of Al2O3­SiC­C castable with different q value, wt%

Raw materialsParticle size

[mm]

q value

0.21 0.22 0.23 0.24 0.25

Fused alumina

5­8 9.40 9.82 10.25 10.67 11.093­5 9.22 9.59 9.95 10.31 10.661­3 16.77 17.30 17.82 18.32 18.79

0.088­1 14.83 14.21 14.54 14.83 14.08¯0.074 10 10 10 10 10

Silicon carbide0.088­1 8 8 8 8 8¯0.074 5 5 5 5 5

¡-Al2O3 13.79 12.08 10.44 8.88 7.39Calcium Aluminates cement 3 3 3 3 3Silicon fume 4 4 4 4 4Pitch ball 3 3 3 3 3Micro silica 3 3 3 3 3Additives 0.1 0.1 0.1 0.1 0.1

Fig. 1. Schematic diagrams of the slag corrosion test in inductionfurnace.

Fig. 2. Apparent porosity of Al2O3­SiC­C castables with differentq value.

Yin et al.: Pore evolution and its effect on slag resistance of Al2O3–SiC–C castablesJCS-Japan

874

castable. A better flowability allows the particle packing to becloser to the ideal closest packing, and the castables therefore canachieve better packing density. However, as the water contentof 6.5wt% was fixed in this study, use of more finer particlesdoes not always mean a better packing, i.e, there must be anoptimal value for q. In addition, the effect of flowability needs tobe considered together. As illustrates by Table 1, with increasingq from 0.21 to 0.25, the amount of large particles increases whilethat of finer particles decreases, which must have some effect onthe flowability of Al2O3­SiC­C castable. The flowability valuesof Al2O3­SiC­C castables with different PSD were determinedand are shown in Fig. 3.As revealed by Fig. 3, the flowability of the castable reaches

the maximal value when q = 0.23. This could be explained asfollows. For a given amount of water addition (6.5wt% in thepresent case), when q is <0.23, there are too much fine parti-cles, but there is not enough water to wet them, therefore theflowability becomes poor. On the other hand, when q is >0.23,there is no enough fine particles for reducing the friction betweenthe coarse grains. As a result, castables also follow poorly. Whenq = 0.23, Al2O3­SiC­C castables have appropriate PSD andthe best flowability, therefore, the best packing density, i.e., theminimal apparent porosity, can be achieved. In additiona, it isevident from Fig. 2 that the apparent porosity of Al2O3­SiC­Ccastable treated at 1450°C for 3 h is much greater than that driedat 110°C.

3.2 Effect of PSD on permeabilityThe permeability values of Al2O3­SiC­C castable prepared

with different PSD after treated at different temperatures areshown in Fig. 4. For samples dried at 110°C or fired at 1450°Cthe permeability changes with PSD are almost the same. The per-meability reaches its minimal value when q = 0.22. It increaseswith increasing q further from 0.22 to 0.25. Meanwhile, the

permeability of samples fired at 1450°C is much higher than thatof the samples dried at 110°C. The change of permeability withq is the same as that seen in the case of apparent porosity, exceptthat permeability reaches the minimum value when q = 0.22,whereas apparent porosity reaches its minimum value whenq = 0.23. Permeability is a property used to characterize the sizeand amount of through pores in a refractory, while the apparentporosity characterizes only the amount of open pores and throughpores, so the two do not have exactly the same relationship withthe PSD of the refractory. A little bit more fine particles could bebeneficial to having smaller permeability, because fine particlescan fill the through pores, and change them to closed pores. Thisis the reason why the q for the minimal permeability is smallerthan that for the minimal apparent porosity.

3.3 Effect of PSD on pore size distribution andmean pore size

Pore size distribution quantifies the amount of each sized porein the material. It is often illustrated with a curve of cumulativeamount of pore volume versus mean radius of pores. Pore sizedistributions of Al2O3­SiC­C castable prepared with differentPSD after treated at 110°C for 24 h are shown in Fig. 5.As seen from Fig. 5, for the samples treated at 110°C for 24 h,

their pore size distribution curves look similar. There are two orthree peaks in the curves. The only main difference is that withincreasing q, the mean pore size increases as well. The mean poresize and pore size distribution were determined by the mercuryfilling method, and 4­8mm chunks cut from the mother sampleswere used. So these two parameters indicate some characteristicsof the inner pores in the material. As shown in Table 1, withincreasing q, the amount of fine particles decreases, so some bigpores cannot be filled due to insufficient fine particles, con-sequently the mean pore size increases with increasing q. Themean pore sizes of samples with different PSD treated at 1450°Cfor 3 h (Fig. 6), have no obvious regularity. However, for poresize distribution, 3 obvious peaks appear in the curves corre-sponding to q = 0.21 and 0.22, indicating the scattering in thepore size distribution, i.e., large differences between small andbig pores in the material. On the other hand, when q = 0.23 and0.24, in particular when q = 0.23, the corresponding curves showonly two obvious peaks, and the peak at around 70 nm is lowerthan that in the cases when q = 0.21 and 0.22. This means thatpores in samples with q = 0.23 and 0.24 vary in a very narrowrange, namely they exhibit a continuous distribution. But forq = 0.25, three peaks again appear.

3.4 Temperature effect on the pore evolutionFigures 3 and 4 show that apparent porosity and permeability

of Al2O3­SiC­C castables with different PSD after fired at1450°C for 3 h are much bigger than those of the samples driedat 110°C for 24 h. The pores evolution with heating has a closerelationship with dehydration of hydration products of CAC, themelting and evaporation behavior of pitch, as well as the effectfrom the silicon powder. Some residual sphere-liked pitch ballsare seen in the Al2O3­SiC­C castable structure after dried at110°C for 24 h [Fig. 7(a)], which is attributed to their highmelting point (about ³120°C). However, with increasing thetemperature to 1450°C, pitch balls transform into gaseousspecies, resulting in pores [Fig. 7(b)]. Consequently, the apparentporosity increases.The actual atmosphere during the firing process was equiva-

lent to a mixture of 35 vol% CO and 65 vol% N2. Therefore, thefollowing reactions during firing would occur.

Fig. 3. The effect of PSD on the flow ability of Al2O3­SiC­C castables.

Fig. 4. Permeability of Al2O3­SiC­C castables with different PSD.

Journal of the Ceramic Society of Japan 121 [10] 873-879 2013 JCS-Japan

875

3SiðlÞ þ 2COðgÞ ¼ 2SiCðsÞ þ SiO2ðsÞ ð2ÞSiCðsÞ þ 2COðgÞ ¼ SiO2ðsÞ þ 3CðsÞ ð3ÞSiCðsÞ þ COðgÞ ¼ SiOðgÞ þ 2CðsÞ ð4ÞSiOðgÞ þ COðgÞ ¼ SiO2ðsÞ þ CðsÞ ð5Þ

When the temperature is above 1420°C, silicon powder (2.34g/cm3) used as an antioxidant for the Al2O3­SiC­C castablewould melt into liquid phase and react with CO(g) diffused intothe inner structure through pores according to Eq. (2). SiC canalso react with CO(g) according to Eqs. (3) and (4). In terms ofthe densities of SiC(s) (3.2 g/cm3), C(s) (2.28 g/cm3) and SiO2(s)(2.38 g/cm3), the volume changes associated with Eqs. (2) and(3) are calculated as about +41% and +228% respectively. Thismeans that the above two reactions could lead to reduction inporosity and pore structure change. This could be explainedfurther based on Figs. 8 and 9.As shown in Fig. 8(a), some whisker-shaped phases formed in

the pores. SEM-EDS verified that they were SiC [see Fig. 8(b)].These newly-formed SiC whiskers fill in pores and reduce thepore size, thus changing the permeability, pore size distributionand mean pore size. However, the SiC whiskers were mainlyformed through gas phase involved reactions severely affected bythe partial pressures of involved gases, especially CO.19) Never-theless, under the test condition, the partial pressure for samples

prepared with different PSD was the same. In fact, only CO(g)was absorbed by pores in the material, it can react to form SiCwhiskers according to Eq. (2). So its amount absorbed by poreshas influence on the SiC whisker formation. It was known thatthe smaller the pore size, the bigger its specific surface area, andthe more CO(g) absorbed,20) and thus the more SiC whiskerformed. Meanwhile, the pores provide spaces for the SiC whiskerto form, so the bigger the pore size, the more SiC whiskers it canaccommodate. In terms of these, there must be a balanced pointfor pore size, which allows much more SiC whiskers to form forsamples having different mean pore sizes. This is probably thereason why samples with q = 0.23 and 0.24 showed relativelynarrow pore size distributions.It should be pointed out that reaction between SiC and CO(g)

[Eq. (3)] actually occur in two stages. Initially, the active oxida-tion Eq. (4) of original SiC particles near surface occurs, makingthe SiC particles much more porous [Figs. 9(c) and 9(d)]. Thenthe passive oxidation Eq. (5) of SiC particles occurs on the sur-face, resulting in a dense layer formed on the surface [Figs. 9(a)and 9(b)].21)

3.5 Effect of PSD on slag resistance of Al2O3–SiC–C castable

Slag resistance of Al2O3­SiC­C castables prepared with

Fig. 5. Pore size distribution and mean pore size of Al2O3­SiC­C castables with different particle size distribution, after driedat 110°C for 24 h. (a), q = 0.21, (b) q = 0.22, (c) q = 0.23, (d) q = 0.24, (e) q = 0.25.

Yin et al.: Pore evolution and its effect on slag resistance of Al2O3–SiC–C castablesJCS-Japan

876

different PSD was tested using the induction furnace method.After test, samples were cut along vertical direction in the center,and their cross-sections are shown in Fig. 10. Al2O3­SiC­Ccastable samples with q = 0.21, 0.22 and 0.25 were obviouslycorroded by the slag, especially at the interface between gas,

liquid and solid phases, while samples with q = 0.23 and 0.24,were almost not corroded. These results indicate that the optimalq value for the corrosion resistance is the same as that for thepore size distribution, which further indicates that the slag resist-ance of Al2O3­SiC­C castables with different PSD has a close

Fig. 6. Pore size distribution and mean pore size of Al2O3­SiC­C castables with different particle size distribution, after firedat 1450°C for 3 h. (a), q = 0.21, (b) q = 0.22,(c) q = 0.23, (d) q = 0.24,(e) q = 0.25.

Fig. 7. SEM morphology of Al2O3­SiC­C castables after treated at different temperatures, (a) after dried at 110°C for 24 h,(b) after fired at 1450°C for 3 h.

Journal of the Ceramic Society of Japan 121 [10] 873-879 2013 JCS-Japan

877

relationship with the pore characters, especially with the pore sizedistribution,. This could be explained according to the followingEq. (6), in which, Kp is the slag penetration depth in pore, r theradius of pore, · the surface tension of liquid slag, ª the contactangle of liquid slag with the refractory, ©p the viscosity of liquidslag in pore, and Rp a coefficient of penetration resistance of

refractory. For a specific slag and refractory, ·, ª, and ©p areconstant, so Kp is mainly affected by the mean pore size (r) andRp which has a close relationship with the pore size distribution.This well explains the above slag resistance results.

Kp ¼ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffir· cos ª

2©pR2p

sð6Þ

4. Conclusions

The apparent porosity and permeability of Al2O3­SiC­Ccastables after dried and fired at different temperatures reach theirminimal values when q = 0.23 and 0.22, respectively. Further-more, the apparent porosity and permeability of fired samplesare much greater than those of dried ones. The mean pore size ofdried samples increases with the q value, whereas for fired ones, itappears to be irregular and much bigger. The pore size distri-butions for both dried and fired samples are discrete when q =0.21, 0.22 or 0.25, but tend to be continuous when q = 0.23 and0.24. The apparent porosity, permeability, mean pore size andpore size distribution of dried Al2O3­SiC­C castables, mainly

Fig. 9. Light microscopy microstructures of 40mm © 40mm cross section view of fired Al2O3­SiC­C castable sample.(a) on the surface, (b) 4mm from the surface, (c) 12mm from the surface, (d) 18mm from the surface (almost the center).

Fig. 10. Cross section graphs of corroded Al2O3­SiC­C castable withdifferent PSD.

Fig. 8. SiC whisker formed in pore of Al2O3­SiC­C castable after fired at 1450°C for 3 h. (a) whisker formed in pore,(b) EDS pattern of enlarged whisker.

Yin et al.: Pore evolution and its effect on slag resistance of Al2O3–SiC–C castablesJCS-Japan

878

depend on the PSD, and to some extent, on the water evaporationand dehydration of the CAC hydration products. However, duringheating at high temperatures, the effect of temperature on thepore evolution becomes significant, while the characters of poresderived after drying also have some influence on the pore evolu-tion. The best slag resistance of Al2O3­SiC­C castable when q =0.23 is believed to be associated with its relatively low porosity,permeability, mean pore size, and especially the narrow pore sizedistribution.

References1) W. Chengxun, H. Jin, Z. Liang, L. Zhaoqiu, Z. Yi and Z. Hanli,

“The Damage of Refractory and Its Suppression Technology”,Ed. by Zhang Xiuzhen, Metallurgical industry press, Beijing(2009) pp. 108­112.

2) R. Salomão and V. C. Pandolfelli, Ceram. Int., 34, 173­180(2008).

3) B. Myhre, The effect of particle size distribution on flowof refractory castables. The American Ceramic Society 30thAnnual Refracotories Symposium in St. Louis, Missouri,march 25 (1994).

4) Y. Kutmen Kalpakli, Archives of Materials Science andEngineering., 34, 81­88 (2008).

5) B. Myhre and A. M. Hundere, UNITECR, 97, 43­52 (1997).6) V. G. Domiciano, M. A. Brito, T. A. Avilal, A. P. Luz and V. C.

Pandolfelli, UNITECR’09, 2009.7) Sc. Vladimir, V. primachenko, Valery V. Martynenko,

Alexander. G. Knoruzhiy, UNITECR’09, 2009.8) Adam Molin, Ondrej Kluza, Jacek Podworny, Teresa Wala,

UNITECR’09, 2009.9) D. N. Borzov, Refract. Ind. Ceram., 43, 141­144 (2002).10) T. Yoshimoto, T. Suzuki and K. Sugiyama, UNITECR, 97, 5­

14 (1997).11) L. Yousheng, Z. Tangwen and L. Nan, Refractories, 41, 188­

190 (2007).12) G. Licun, L. Bin and W. Chunfeng, Refractories, 42, 51­53

(2008).13) Z. Tao, Refractories, 36, 246 (2002).14) L. Zhaoyun, T. Junbo and W. Juncong, Refractories, 43, 260­

262 (2009).15) G. Huazhi, L. Chunyan, W. Houzhi and Z. Wenjie, Refracto-

ries, 38, 160­164 (2004).16) D. Xiaoling, S. Jialin, C. Junhong, Z. Houxing and H. Yanruo,

Refractories, 38, 82­84 (2004).17) T. Junbo, W. Juncong and L. Chaoyun, Refractories, 43, 19­22

(2009).18) B. Myhre and B. Sandberg, Proceeding of International

Seminar on Monolithic Refractory Materials, Tehran, Iran(1997) pp. 113­140.

19) Li Yawei, Chen Xilai, Li Yuanbing, Jin shengli, Ge Shan, ZhaoLei and Li Shujing, Refractories, 42, 401­404 (2008).

20) G. Liwen, X. Cangyan and C. Shuzhao, J. Hunan Univ. Sci.Technol., 23, 18­21 (2008) (Nature Science Edition).

21) A. Yamaguchi, Int. J. Appl. Ceram. Technol., 4, 490­495(2007).

Journal of the Ceramic Society of Japan 121 [10] 873-879 2013 JCS-Japan

879