Author’s Accepted Manuscript

Lightweight design of bauxite-SiC composite

refractories as the lining of rotary cement kiln using
alternative fuels

Ren Bo, Li Yawei, Sang Shaobai, Jin Shengli

www.elsevier.com/locate/ceri

PII: S0272-8842(17)30960-4
DOI: http://dx.doi.org/10.1016/j.ceramint.2017.05.148
Reference: CERI15308
To appear in: Ceramics International
Received date: 7 February 2017
Revised date: 21 May 2017
Accepted date: 21 May 2017
Cite this article as: Ren Bo, Li Yawei, Sang Shaobai and Jin Shengli,
Lightweight design of bauxite-SiC composite refractories as the lining of rotary
cement kiln using alternative fuels, Ceramics International,
http://dx.doi.org/10.1016/j.ceramint.2017.05.148
This is a PDF file of an unedited manuscript that has been accepted for
publication. As a service to our customers we are providing this early version of
the manuscript. The manuscript will undergo copyediting, typesetting, and
review of the resulting galley proof before it is published in its final citable form.
Please note that during the production process errors may be discovered which
could affect the content, and all legal disclaimers that apply to the journal pertain.
Lightweight design of bauxite-SiC composite refractories as the

lining of rotary cement kiln using alternative fuels

REN Boa, LI Yaweia,*, SANG Shaobaia, JIN Shenglib
a
The State Key Laboratory of Refractories and Metallurgy, Wuhan University of Science and Technology, Wuhan

430081, China
b
Chair of Ceramics, Montanuniversitaet Leoben, Peter-Tunner-Straße 5, 8700, Leoben, Austria

Abstract: To meet the demand of energy-saving and adapt to the change from coal to the alternative fuel in the

rotary cement kiln, bauxite-SiC refractories were fabricated by the incorporation of silica sol coated lightweight

mullite aggregates in order to achieve low thermal conductivity and superior alkali vapor attack resistance

simultaneously. Furthermore, the mechanism of resistance to alkali vapor attach was investigated by means of

X-ray diffraction (XRD) and scanning electron microscopy (SEM). Results showed that the thermal conductivity

of bauxite-SiC specimens decreased gradually with increasing amounts of silica sol coated lightweight mullite

aggregates while changes in the alkali vapor attach were not detectable. The shell-covered structure with a silica

sol coating on the surface of lightweight aggregates hindered alkali vapor diffusion into the aggregates at high

temperature. Bauxite-SiC refractories possessed lower thermal conductivity, superior alkali attack resistance and

higher mechanical properties compared with the specimens which contained pristine lightweight aggregates.

Keywords: Bauxite-SiC composite refractories; Silica sol coating; Lightweight mullite aggregate; Thermal

conductivity; Alkali vapor attack resistance

1 Introduction

Nowadays tremendous attention on heat loss reduction of industrial kilns and furnaces is paid
towards decreasing the consumption of fossil fuel in the energy intensive industries [1-4]. In the
last decade, several research works [5-8] have brought up a concept of lightweight lining of
furnaces which possess appropriate mechanical properties, corrosion resistance and excellent
thermal insulation performance compared with the traditional dense linings. In this case,
lightweight microporous aggregates were partly incorporated into the linings, substituting for the
dense aggregates. Guangping Liu et al. [6-8] prepared the lightweight alumina-magnesia castables
as the lining of steel ladle using lightweight CA6 aggregates and microporous corundum
aggregates respectively. Enhanced heat insulation performance, good thermal shock resistance and
molten slag resistance were obtained [9-11].
In the rotary kiln, the lightweight magnesia-spinel bricks were also newly developed to
reduce the heat loss on the surface of the transition zone [12-13]. Decades ago, the hazardous

*
Corresponding author. Tel.: +86 2768862188; Fax: +86 2768 862188.
E-mail address: liyawei@wust.edu.cn.

1 / 29
wastes were used as an alternative fuel in cement manufacturing process. These wastes contained
aggressive elements such as alkalis (K, Na), S and Cl which were volatilized during combustion.
They became more concentrated and continued to circulate in the system, which made the
eventual alkali attack on the exposed lining more severe as time progressed. Among all of these
elements, alkalis are more aggressive and can quickly penetrate into the open pores of refractories
[14-19]. In this regard, lightweight aggregates are susceptible to alkali vapor due to their porous
structures. Besides magnesia-spinel bricks, bauxite-SiC refractories are also commonly used in
China as the lining at the transition zone of rotary kilns because of their excellent wear resistance
and good thermal shock resistance. The lining in this area is completely in contact with flowing
cement clinkers and aggressive vapor [20-26], experiencing severe corrosion. Consequently, the
alkali vapor attack of the lining in the transition zone has to be taken into consideration as the
lightweight refractories are designed to reduce the heat loss of kilns.
In this research, the commercially available lightweight mullite aggregates and silica-sol
coated ones were introduced directly into the bauxite-SiC refractories. The silica-sol coating acted
as a barrier between the aggregates and matrix to hinder the penetration of alkali vapor into
aggregates. Based on our previous investigations, silicon powder was also introduced to prevent
the active oxidation of SiC aggregates [27] and to improve the mechanical properties as well as to
optimize the pore structure in the matrix [28]. The effects of varying quantities from 50% to 100%
of lightweight aggregates on the thermal conductivity and alkali attack resistance of the
refractories were investigated to evaluate the feasibility of using lightweight materials for
bauxite-SiC refractories in the rotary kiln. Meanwhile the interfacial bond between lightweight
aggregates and the matrix was studied before and after alkali exposure to elucidate the alkali
attack mechanism.
2 Experimental
2.1 Preparation of silica sol coated lightweight aggregates
The silica sol-coated lightweight mullite aggregates (SC-LMA) were prepared by the
following procedure. The commercial lightweight mullite aggregates (LMA) (4-3 mm and 3-1 mm,
69 wt% Al2O3, 26 wt% SiO2, 1 wt% Fe2O3, 2 wt% TiO2, Yanshi Guangming High-tech Refractory
Products Co., Ltd., China) were immersed in the silica sol (the solid content is 31 wt%, PH=9.0,
Wuhan Tianxiang Chemical Co., Ltd., China) and stirred for 30 min using a plastic mixing blade
controlled by an electronic stirrer (JJ-1, Changzhou Guohua Electric Appliance Co., Ltd., China).
The rotating speed was 80-100 rpm. The mass ratio of LMA to the silica sol was 1:1. Afterwards,
the as-received products were filtrated by a vacuum filtration setup (SHB-III, Zhengzhou Great
Wall Scientific Industry and Trade Co., Ltd., China) to remove excess silica sol. The as-prepared
aggregates were dried at 60 °C for 24 h. The dried aggregates had a 12.1 wt% mass increase. The

2 / 29
fabrication flow chart is shown in Fig. 1. Additionally, the 1-0 LMA was determined inappropriate
to substitute the dense aggregates due to the following technical issues: (1) extremely tiny
granularity; and (2) difficulty in separation after coating and drying.
The physical appearance of LMA (4-3 mm) and their microstructures before and after the
coating process are shown in Fig.2. Compared with the pristine LMA (Fig.2a), the milky coating
can be observed by the naked eyes on the surface of SC-LMA(Fig.2b), indicating that the LMA
was successfully coated with silica sol. The microstructure observation showed that the rugged
surface with some open pores was found on the LMA surface (Fig.2c). However, after the coating
process, the surface became much smooth (Fig.2d-e) and a layer of amorphous silica was detected.
This suggested that a part of the open pores were filled with silica sol.
2.2 Preparation of bauxite-SiC specimens
Homogenized bauxite aggregates, high grade bauxite, SiC aggregates, Guangxi clay, micron
sized α-alumina, silicon powder as well as LMA or SC-LMA were used as raw materials while
Al(H2PO4)3 solution (liquid, the specific gravity is 1.35) was used as a binder. The detailed
information of the raw materials is illustrated in Table 1. Due to the fact that the bulk densities of
LMA and SC-LMA are much lower than the dense bauxite aggregates, the latter were substituted
gradually by the former (LMA and SC-LMA, respectively) by the same volume fraction according
to their apparent densities (as shown in Table 2) in order to closely maintain the same particle
gradation as the reference specimen. Therefore, each specimen possessed the same volume
fraction of aggregates and the same mass fraction of the matrix part as long as the specimens were
compacted into the same dimension. The only difference lay in the type of aggregates. Thus, the
roles of various aggregates in bauxite-SiC composite refractories can be evaluated directly.
As can be seen in Table 3, different bauxite-SiC compositions were formulated by varying
the type of aggregates. The reference specimens are the dense bauxite-SiC composite refractories
widely used as the lining of cement kiln. The lightweight specimens containing LMA and
SC-LMA were designated as “U” and “C”, respectively. In these lightweight compositions, 5-3
mm and 3-1 mm dense aggregates were substituted by the commercial lightweight ones with sizes
of 4-3 mm and 3-1 mm, respectively. 50% and 100% of the total amount of these dense aggregates
were substituted. For instance, 50% of the volume fraction of bauxite aggregates (5-1 mm) was
substituted by the SC-LMA, which was labeled as C50. The 1-0 mm LMA was not introduced into
the system due to the technical issues. It is worth noting that all of the compositions but the
reference one are not 100 wt% because they were not normalized.
All the raw materials were mixed for 20 min in a mixer (JJ-5, Wuxi Jiangong Test Facility
Co.Ltd., China) with the rotating speed of 80-100 rpm. After kneading, cylindrical specimens
(Ф50×50 mm3 for the mechanical properties test, Ф50×30 mm3 for alkali exposure and Ф180×20

3 / 29
mm3 for thermal conductivity test) were compacted under a pressure of 20 MPa by a hydraulic
press, then cured at 110 °C for 24 h in a muffle furnace. Subsequently, the as-obtained specimens
were treated with a heating rate 5 oC/min to 1500 oC, and then soaked at that temperature for 3 h
in air using an electric furnace.
2.3 Testing and characterization methods
The bulk density (BD) and apparent porosity (AP) of bauxite-SiC specimens after firing were
measured according to Archimedes’ Principle. Cold crushing strength was determined on a
universal testing machine (LM-02, Longsheng Test Facility Co. Ltd., China) at the crosshead
speed of 0.5 mm/min based on GB/T 5072.2-2004. The above properties measured for each
composition were average values from at least three test pieces.
The alkali vapor attack test was an optimized case based on the reference [29] and GB/T
14983-2008. The specimens were embedded in premixed carbon black (30 nm, 99.8 wt% fixed
carbon, Pingdingshan Aobote Xiangsu Zhuji Co., Ltd., China) and potassium carbonate
(Analytically pure, 99.8wt% K2CO3, Tianjin Bodi Chemical Engineering Co., Ltd., China)
powders in the graphite crucibles (Ф80×220 mm3). The mass ratio of carbon black to potassium
carbonate was 1:1. The mass ratio of the specimens to the premixed powder was 1:1, with
sufficient mixed powder added to cover all of the specimens. All of the reactant materials and
specimens were sealed inside the graphite crucibles to prevent a leak of K vapor. The graphite
crucibles were then put inside a cubic alumina crucible (240×130×120 mm3). To avoid oxidation
of the graphite, they were embedded in coke. Afterwards, the alumina crucible was fired at
1000 °C for a soaking time of 10 h in an electric furnace. The schematic diagram of the test is
shown in Fig. 3.
To evaluate the phase composition, approximately 10 g of each specimen was ground to a
particle size below 74 μm. The phases present in the fired specimen and reacted ones were
determined by X-ray powder diffractometry (XRD, X’ pert Pro, Philips, using Ni-filtered, Cu Ka
radiation at a scanning speed of 2 deg/min at the temperature of 20 °C). The patterns were
recorded from 10° to 90° 2θ in steps of 0.008°. Diffractograms of specimens were compared to the
International Centre for Diffraction Data-ICDD PDF2 database. The phase analysis was conducted
using the software of Philips X’ pert Pro High Score Plus (version 3.0.4).
Thermal conductivity of the fired specimens was measured on disk specimens (Ф 180×20
mm3) at 300 °C, 500 °C, 800 °C and 1000 °C, respectively, using the standard water flow plate
method (PBD-30, Sinosteel Luoyang Institute of Refractories Research Co. Ltd., China) [30-32].
All the specimens for SEM observation were cut, and then were polished using B4C polishing
materials. The cross section of the SC-LMA was observed after the fired specimens were etched
by hydrofluoric acid for 10 min to remove the amorphous phase. After alkali exposure, samples

4 / 29
were firstly embedded in the epoxy resin under vacuum, then cut into smaller specimens.
Subsequently, the as-obtained specimens were polished carefully with the aid of absolute ethyl
alcohol, coated with gold, and then the edge part was observed by scanning electron microscope
(SEM, Quanta 400, FEI Company, USA). The mass percentage and distribution of K in the matrix
of corroded specimens were detected by the quantitative energy dispersive spectroscopy (EDS,
INCA IE 350 Penta FET X-3, Oxford, UK) SEM microanalysis.

3 Results and Discussions

3.1 Physical properties and thermal conductivity
The physical properties (bulk density, apparent porosity and cold crushing strength) and
thermal conductivity of the fired specimens are shown in Table 4. In contrast with reference
specimen, the substitution of 50% and 100% LMA caused a decrease of bulk density (BD) and
cold crushing strength (CCS), while the change of apparent porosity resulted in the opposite
tendency. Meanwhile, the thermal conductivity (TC) of specimens U50 and U100 at various
temperatures was lower than the reference specimen. For example, the TC of specimens U50 and
U100 at 1000 °C was 1.220 and 1.087 W·m-1·k-1, respectively, and 1.349 W·m-1·k-1 for the
reference specimen.
The use of SC-LMA in samples also impacted the physical properties and thermal
conductivity in the same way as LMA. Nevertheless, the CCS of the specimens C50 and C100
was slightly higher than the U50 or U100 when the same volume fractions of dense aggregates
(50% and 100%) were substituted. It implied that the silica sol coating could enhance the
interfacial bond between the lightweight aggregates and the matrix. Additionally, the TC of the
specimens C50 and C100 was lower than the specimens U50 and U100, respectively, which could
be caused by the SiO2-rich interfacial phase located in refractories.
3.2 Alkali exposure testing
Physical appearances of the specimens before and after the alkali exposure are presented in
Fig.4. As can be seen, all of the specimens were quite stable according to their macroscopic
morphology. No cracks or damage can be detected except for the variation of color on their
surface.
To directly evaluate the roles of both kinds of lightweight aggregates in bauxite-SiC
refractories, the phase compositions and microstructure were investigated. The phase
compositions of specimens at their edge after firing and alkali exposure are shown in Fig. 5. As
for the fired specimens, XRD analysis did not display any remarkable differences in the
crystallographic phases; quartz was present in almost all the specimens along with mullite,
corundum and β-SiC phases (Fig.5a). Quartz resulted from the oxidation of SiC aggregates and
silicon powder [27]. With regard to the specimens U50 and U100, the peak intensities of quartz

5 / 29
were nearly the same as those of the reference specimen. In contrast, the amount of quartz in
specimens C50 and C100 increased, indicating that the silica sol had transformed into quartz
during heating-up. After the alkali test, kalsilite (KAlSiO4) was formed in all specimens, while the
amount of quartz phase reduced (Fig. 5b). It is worth noting that the amount of KAlSiO4 in
reference specimen was higher than all other specimens, implying more K was present.
The microstructures of polished cross sections of specimens U100 and C100 (LMA and
SC-LMA containing materials) after firing are shown in Fig. 6. Apparently, SC-LMA was much
denser than LMA (Fig. 6a, 6b), showing clearly that the silica sol penetrated into lightweight
aggregates, forming the coating on surface of aggregates. Compared with U100, the CCS of C100
increased from 35.4 to 39.8 MPa. The enhanced strength was attributed to greater sintering
between the SC-LMA and matrix (Fig. 6b). Further investigation of its SEM micrographs revealed
a shell-core structure and in-situ formation of many tiny mullite crystals at the coating layer after
removing the amorphous phase by hydrofluoric acid (Fig. 6c,6d). Additionally, quartz was found
to decrease in the local areas (Fig. 6d).
However, the microstructural evolution of specimens at their edge after the alkali exposure
was different with respect to KAlSiO4, depending on the starting composition of the sample, as
shown in Fig. 7. With regard to the reference specimen, no evidence of chemical attack can be
found inside the dense bauxite aggregate (Fig. 7a), while K was concentrated in the matrix,
resulting in abundant KAlSiO4 formation (Fig. 7b). The KAlSiO4 phase was verified by XRD. In
the case of U50 specimen, the alkali attack occurring to lightweight mullite aggregate (LMA)
resulted in an ambiguous edge between this aggregate and the matrix. A large crack also emerged
in the matrix and passed through the aggregate (Fig. 7c). As can be seen from the EDS mapping of
Fig. 7c, a high level of the K vapor penetrated into the lightweight aggregates, which was
distributed rather homogeneously in the matrix and porous aggregate (Fig. 7d). At higher
magnification, some laminar KAlSiO4 can be easily seen inside the pores of LMA, along with
micro-cracks (Fig. 7e).
Compared with specimens U50 and U100, specimens C50 and C100 had different
microstructures. The K was concentrated in the matrix of C50 and C100 specimens similar to
reference ones where SC-LMA remained intact (Fig. 7f). It is clearly observed that the K mainly
surrounded around SC-LMA in corroded areas (Fig. 7g).
To further characterize the penetration of K vapor, the weight percentage and distribution of
K in the matrix were also evaluated by quantitative EDS microanalysis using the SEM. The
relative content of K among Al, Si, O, K in the matrix as a function of the infiltration distance
after alkali exposure testing is illustrated in Fig. 8. Among all the specimens, the reference
exhibited the highest level of K content, experiencing the most severe penetration. More than 10

6 / 29
wt% of K was left in the matrix when the penetration depth was approximate 7 mm. The porosities
for U50 and U100 are higher than the reference specimen, which would lead to the severer
penetration. In fact, a lower level of K was exhibited in the matrix of U series specimens (Fig. 8a).
This can be attributed to the fact that a part of the K vapor infiltrated into the LMA, resulting in a
lower concentration of K. Although the introduction of SC-LMA improved the porosity of
refractories, the content of K in C series specimens was lower than the U series specimens when
the penetration depth was the same, indicating better alkali attack resistance. It can be deduced
that the silica coating can react with K vapor and prevent it from infiltrating into the pores of
lightweight aggregates during the alkali exposure (Fig. 8b).
3.3 Discussions
The above results show the significant influence of aggregate type on the corrosion behavior
of K vapor during the alkali exposure. For a better uderstanding, three kinds of penetration
processes are proposed to explain the chemical and microstructural changes in these aggregates,
which is graphically presented in Fig. 9.
Fig. 9a is a schematic representation of the penetration mode which can be used to explain
the penetration process of K vapor into the reference specimen. Such aggregates were quite stable
due to their dense structure, leading to a high concentration of K in the matrix. A large amount of
KAlSiO4 was formed in the matrix, which was in good agreement with the SEM and EDS resluts.
In the case of LMA (Fig. 9b), the K vapor infilitrated into the pores of aggregates to form
KAlSiO4 due to their porous structure. As a result, substantial volume expansion occurred,
followed by crack formation[17,34]. Accordingly, the concentration of K in the matrix was lower
than the reference specimen because K vapor was partly consumed inside the LMA. In contrast,
the corroson behaviour is similar to the reference case once a layer of silica was formed on the
LMA, a layer which limited penetration inside the aggregates (Fig. 9c). Instead, a lower
concentration of K was obseverd in both aggregates and the matrix, which suggested that part of
the alkali was consumed at the interfacial region.
To reveal the interaction between silica coating and K vapor, the phase evolution of the
specimens during alkali exposure was simulated by using the Equilib module in Factsage (version
6.4). Two kinds of chemical reaction processes are demonstrated in the thermodynamic
calculation.
In the first case, only the LMA was considered for thermodynamic calculations. The
calculation selected 10 g LMA as the starting compositions, and Alpha was defined as the mass
ratio of the K vapor to aggregates because the K vapor was introduced directly in the corrosion
test. When Alpha was 0.2, the calculation was carried out with 2g K vapor and 10 g LMA. Results
of thermodynamic calculations are shown in Fig. 10. Thermodynamically, the main components,

7 / 29
such as mullite, Al2O3 and SiO2, can be highly active to react with K vapor, leading to the
generation of several kinds of K-aluminosilicates (KAlSiO4,KAlSi2O6, etc). However, such
products were not found in the laboratory experimental due to the fact that they were intermediate
products based on the previous research [28-29,33-34]. The experiment results showed that K
vapor can easily penetrate into LMA due to their porous structure, causing the degradation of
refractories. These results are in accordance with the XRD and SEM analysis. In this experiment,
only the KAlSiO4 was found in the exposed specimens because other K-aluminosiliactes are likely
to transform into KAlSiO4 based on our previous investigations [28,33]. The mass ratio of K to
refractory was in the range of 0.08~0.26 when KAlSiO4 was formed.
The second case demonstrated the chemical reactions which occurred on the SC-LMA
surface. In order to make this calculation comparable to the first case, 10g SC-LMA was also
chosen to carry out the reaction. However, the thermodynamic reactions changed due to the
presence of the coating. From the results of SEM observation (Fig.6c-d), the coating can be
considered as a three-layered structure: a thin layer of mullite, a transitional layer and a silica layer,
as illustrated in Fig. 11a. The K vapor would react with this region first instead of penetrating into
the aggregates as long as the composite coating layer was formed. Therefore, a representative of
the corrosion process that took place at the interfacial region of SC-LMA was studied.
Considering that only one variate can be used in the Equilib module, the mass of K was regarded
as a constant, which was chosen as 1g because under the same condition, enough KAlSiO4 can be
formed by reacting with 10g LMA in the first case. Therefore, a 1.1g coating layer (the mass of
coating on 10g SC-LMA) on SC-LMA was chosen to react with 1g K vapor in this case. Beta was
defined as the mass ratio of SiO2 to the coating. When Beta was 0.7, the calculation was carried
out with 1g K vapor, 0.33g mullite, and 0.77g SiO2. The result is shown in Fig. 11b. The surface
of coating, which was basically composed of mullite, was more favourable to react with K vapor.
However, if the reaction occurred at the transitional layer or the silica layer, the K vapor would
transformed into a K-rich liquid phase instead of turning into K-containing compounds, thus
avoiding crack. Furthermore, all of the K vpaor can be transformed into liquid at the silica layer.
By comparsion, the SiO2 component in the LMA was consumed and tranformed into a
K-aluminosilicate phase due to its limited amount. The SC-LMA provided better K vapor attack
resistance due to the silica-rich coating on its surface.

4 Conclusions
Lightweight bauxite-SiC composite refractories exhibiting low thermal conductivity were
successfully fabricated by the introduction of lightweight mullite aggregates. Use of a silica sol
coating technique of the lightweight aggregates could potentially be applied to the lightweight
refractory linings. In all the compositions investigated, the appropriate mechanical properties,

8 / 29
thermal conductivity and alkali attack resistance of refractories can be balanced by the use of 50
vol.% lightweight aggregates. It was verified that silica sol-coated aggregates in the form of a
shell-covered structure hindered the alkali penetration into lightweight aggregates, and the
silica-rich rims acted as a protection layer for lightweight aggregates to absorb K vapor during
exposure. Because the silica coatings on the porous mullite aggregates decreased thermal
conductivity and improved alkali resistance in laboratory tests, additions of the lightweight
aggregate to a refractory material used in rotary kilns hold the potential for displacing a portion of
the denser aggregates used in rotary kiln refractory linings and for improving the thermal
efficiency of the process.

9 / 29
References
[1] ZX Zhang, Y Chi, LI Shui qing, JH Yan, KF Cen, Heat transfer model and the numerical simulation of rotary

kiln, Chem. Eng. (China) 31(2003) 27–31.

[2] N.N. Zhao, Discussion on calculation of loss of heat through wall of rotary kiln, Cem. Eng. (China) 4(2012)

23–26.

[3] F. Liu, Z.C. Song, L. Yang, L. Jian, Study of the temperature distribution in cement rotary kilns, Energy for

Metall Ind. 28(2009) 21–25.

[4] G. Xiao, X. Ding，H. Chen, S. Cheng, Research status of the mathematical models of the heat transfer process

of rotary kilns and their development, Chem. Machinery 34(2007) 47–50.

[5] W. Yan, J. Chen, N. Li, Slag resistance of lightweight corundum-mullite refractories, Rare Metal Mat. Eng.

44(2015) 442–446.

[6] G. Liu, X. Jin, W. Qiu, G. Ruan, Y. Li, The impact of bonite aggregate on the properties of lightweight

cement–bonded bonite–alumina–spinel refractory castables, Ceram. Int. 42(2016) 4941–4951.

[7] H. Chen, L. Zhao, X. He, W. Fang, Z. Lei, H. Chen, The fabrication of porous corundum spheres with

core-shell structure for corundum–spinel castables, Mater. Des. 85(2015) 574–581.

[8] L. Fu, H. Gu, A. Huang, M. Zhang, X. Hong, L. Jin, Possible improvements of alumina–magnesia castable by

lightweight microporous aggregates, Ceram. Int. 41(2015) 1263–1270.

[9] C. H. Peng, N. Li, B. Q. Han, Study on slag corrosion resistance of microporous magnesia-spinel carbon

Refractories, Bull. Chin. Cream. Soc. 28(2009) 307–311.

[10] L. Fu, H. Gu, A. Huang, M. Zhang, Z. Li, Slag resistance mechanism of lightweight microporous corundum

aggregate, J. Am. Ceram. Soc. 98(2015) 1658–1663.

[11] W. Yan, J. Chen, N. Li, Slag Resistances of lightweight corundum-mullite refractories, Rare Metal Mater. Eng.

44(2015) 442–446.

[12] W. Yan, J. Chen, N. Li, W. Qiu, Y. Wei, B. Han, Preparation and characterization of porous MgO–Al2O3

refractory aggregates using an in–situ decomposition pore–forming technique, Ceram. Int. 41(2015) 515–520.

[13] W. Yan, X. L. Lin, J.F. Chen, N. Li, Y. Wei, B. Han, Effect of TiO2 addition on microstructure and strength of

porous spinel (MgAl2O4) ceramics prepared from magnesite and Al(OH)3, J. Alloys Compd. 61(2015)

287–291.

[14] J. Szczerba, Chemical corrosion of basic refractories by cement kiln materials, Ceram. Int. 36(2010)

1877–1885.

[15] R. Ceylantekin, C. Aksel, Improvements on corrosion behaviours of MgO–spinel composite refractories by

addition of ZrSiO4, J. Eur. Ceram. Soc. 32(2012) 727–736.

[16] G. Liu, N. Li, W. Yan, et al. Composition and microstructure of a periclase–composite spinel brick used in the

burning zone of a cement rotary kiln, Ceram. Int. 40(2014) 8149–8155.

[17] D. Madej, J. Szczerba, Fundamental investigations on the high temperature corrosion of ZrSiO 4-containing

10 / 29
 andalusite refractories in cement kiln preheater, J. Eur. Ceram. Soc. 36(2016) 875–883.

[18] W. Juntao, C. SongLin, Y. Lin, et al, Corrosion of disposing urban domestic wastes to refractories for cement

rotary kilns, Naihuo Cailiao 49(2015) 165–167.

[19] C. Shujinag, L. Yaowen, L. Guohua, et al. Corrosion of alternative fuels to periclase-compound spinel bricks

for cement rotary kilns, Naihuo Cailiao 49(2015) 161–164.

[20] B. Sahin, C. Aksel. Developments on the mechanical properties of MgO–MgAl2O4 composite refractories by

ZrSiO4–3 mol% Y2O3 addition, J. Eur. Ceram. Soc. 2012, 32(1): 49–57.

[21 E.A. Rodríguez, A.K. Limones, J.E. Contreras, et al. Effect of hercynite spinel content on the properties of

magnesia–calcium zirconate dense refractory composite. J. Eur. Ceram. Soc. 2015, 35(9): 2631–2639.

[22] Á. Obregón, J. Luis, Rodríguez–Galicia, et al, MgO–CaZrO3–based refractories for cement kilns, J. Eur.

Ceram. Soc. 31(2011) 61–74.

[23] E.A. Rodríguez, G. A. Castillo, T. K. Das, et al, MgAl2O4 spinel as an effective ceramic bonding in a

MgO–CaZrO3 refractory, J. Eur. Ceram. Soc. 33(2013) 2767–2774.

[24] W. Meng, C. L. Man, T. Z. Ge, et al, Effect of zircon addition on the physical properties and coatability

adherence of MgO–2CaO • SiO2–3CaO • SiO2 refractory materials, Ceram. Int. 42(2016) 9032–9037.

[25] Z. Guo, P. Stefan, R. Michel, Reaction characteristics of magnesia-spinel refractories with cement clinker, Int.

J. Appl. Cem. Technol. 2(2005) 327–335.

[26] W. Yan, X. L. Li, N. Li, Effect of spinel content of lightweight aggregates on the reaction characteristics of

periclase–spinel refractories with cement clinker, In: Proceedings of the 9th UNITECRS’15, Austria,2015.

[27] B. Ren, S. Sang, Y. Li, Y. Xu, Effects of oxidation of SiC aggregates on the microstructure and properties of

bauxite–SiC composite refractories, Ceram. Int. 41(2015) 2892–2899.

[28] B. Ren, S. Sang, Y. Li, S. Jin, Correlation of pore structure and alkali vapor attack resistance of bauxite–SiC

composite refractories, Ceram. Int. 41(2015) 14674–14683.

[29] L. M. Scudeller, E. Longo, J. A. Valela, Potassium vapor attack in refractories of the alumina-silica system, J.

Am. Ceram. Soc. 73 (5) (1990) 1413-1416.

[30] X. Deng, J. Wang, J. Liu, H. Zhang, L. Han, S. Zhang, Low cost foam-gel casting preparation and

characterization of porous magnesium aluminate spinel(MgAl2O4) ceramics, Ceram. Int. 42 (2016)

18215–18222.

[31] L. Fu, H. Gu, A. Huang, M. Zhang, X. Hong, L. Jin, Possible improvements of alumina–magnesia castable by

lightweight microporous aggregates, Ceram. Int. 41 (2015) 1263–1270.

[32] L. Han, F. Li, X. Deng, J. Wang, H. Zhang, S. Zhang, Foam-gelcasting preparation, microstructure and

thermal insulation performance of porous diatomite ceramics with hierarchical pore structures, J. Eur. Ceram.

Soc.http://dx.doi.org/10.1016/j.jeurceramsoc.2017.02.032.

[33] Y. Xu, S. Sang, Y. Li, B. Ren, L. Zhao, Y. Li, S. Li, Pore Structure, permeability, and alkali attack resistance of

Al2O3-C Refractories, Refract. Metall. Mater. Trans. A 45 (2014) 2885–2893.

[34] J. Stjernberg, M. A. Olivas-Ogaza, M.-L. Anttia, J.C. Iona, B. Lindblom, Laboratory scale study of the

11 / 29
 degradation of mullite/corundum refractories by reaction with alkali-doped deposit materials, Ceram. Int. 39

(2013) 791–800.

12 / 29
List of table captions:
Table 1 The detailed information of the raw materials

Table 2 The apparent density, bulk density and apparent porosity of the aggregates
Table 3 The batch compositions of the specimens
Table 4 The bulk density, apparent porosity, cold crushing strength and thermal conductivity of the specimens

13 / 29
 Table 1 The detailed information of the raw materials
Material Index Particle size Chemical composition Place of origin
5-3 mm;
Homogenized bauxite 73 wt% Al2O3, 21 wt% SiO2, 1 wt% Yangquan Jinyutongda
3-1 mm;
aggregates Fe2O3, 4 wt% TiO2 Refractory Co., Ltd., China
1-0 mm;
Shandong Jinmeng Silicon
SiC 1-0 mm; 97 wt% SiC
Carbide Co., Ltd., China
86 wt% Al2O3, 8 wt% SiO2, 2 wt% Xiaoyi Hezhongxing
High grade bauxite -0.088 mm;
Fe2O3, 3 wt% TiO2 Refractory Co., Ltd., China
30 wt% Al2O3, 46 wt% SiO2, 2 wt% Nanning Guini Mineral Co.,
Guangxi clay -0.075 mm;
TiO2 Ltd., China
Kaifeng Special Refractory
Micron sized α-alumina -2 μm 99 wt% Al2O3
Co., Ltd., China
Anyang Yuhong Metallurgy
Silicon powder -0.045 mm 98 wt% Si & Refractory Co., Ltd.,
China

14 / 29
 Table 2 The apparent density, bulk density and apparent porosity of the aggregates
Index Apparent density Bulk density Apparent porosity
(g/cm3) (g/cm3) (%)
Homogenized bauxite aggregate 2.94 2.62 13.5
Porous mullite aggregate 1.60 1.44 49.6
Silica sol coated porous mullite aggregate 1.60 1.49 42.4

15 / 29
 Table 3 The batch compositions of the specimens
Compositions(wt%)
Raw materials
Spec. Ref U50 U100 C50 C100
Homogenized bauxite aggregate 5-1mm 50 25 - 25 -
1-0mm 8 8 8 8 8
LMA 4-1mm - 13.8 27.5 -
SC-LMA 4-1mm - - - 13.8 27.5
SiC aggregate 1-0mm 10
High grade bauxite powder ≤0.088μm 17
Silicon powder ≤0.045μm 3
Guangxi clay ≤0.045μm 8
α-alumina 2μm 4
Al(H2PO4)3 liquid +3

16 / 29
 Table 4 The bulk density, apparent porosity, cold crushing strength and thermal conductivity of the specimens
No. Bulk density Apparent porosity Cold crushing Thermal conductivity(W·m-1·k-1)
(g/cm3) (%) strength(MPa) 300 °C 500 °C 800 °C 1000 °C
Ref 2.42±0.00 19.2±0.1 48.4±0.5 0.747 0.980 1.094 1.349
U50 2.25±0.01 25.2±0.3 43.1±0.9 0.592 0.786 1.039 1.220
U100 2.02±0.02 32.5±0.4 35.4±1.3 0.554 0.690 0.914 1.087
C50 2.25±0.01 24.3±0.3 50.2±0.8 0.607 0.783 1.037 1.180
C100 2.00±0.02 31.1±0.3 39.8±1.1 0.543 0.632 0.884 1.038

List of figure captions:

Fig. 1 Fabrication flow chart of silica sol coated lightweight aggregates

Fig.2 Physical appearance and microstructure of the LMA before and after the coating process: (a) pristine LMA;

(b) SC-LMA; (c) pristine LMA; (d) SC-LMA; (e) SC-LMA

Fig.3 Schematic diagram of the alkali exposure: (a) graphite crucible; (b) graphite crucible was put in the alumina

crucible embedded by coke

Fig.4 Physical appearances of the specimens: (a) before the alkali exposure; (b) after the alkali exposure

Fig. 5 Phase compositions of (a) the fired specimens and (b) the specimens after alkali exposure

Fig. 6 Microstructures of porous aggregates after firing at 1500°C: (a) the LMA in specimen U100; (b) the

SC-LMA in specimen C100; (c) the edge of SC-LMA, a coating layer was found; (d) a higher magnification of the

coating layer of SC-LMA

Fig. 7 Microstructures of the specimens after alkali exposure: (a) a dense aggregate in the reference specimen; (b)

the matrix in the reference specimen; (c) the LMA in specimen U50; (d) the EDS mapping of figure (c); (e) a

higher magnification of the LMA in (c); (f) the SC-LMA in specimen C50; (g) the EDS mapping of figure (f)

Fig. 8 Relative content of K element among Al, Si, O, K in the matrix: (a) reference specimen, U50 and U100; (b)

reference specimen, C50 and C100

Fig. 9 Corrosion behaviours of K vapor to the aggregate: (a) dense aggregate; (b) LMA; (c) SC-LMA

Fig. 10 Thermodynamic analysis of the LMA in K vapor environment at 1000°C

Fig. 11 Thermodynamic analysis of the SC-LMA in K vapor environment at 1000°C: (a) The schematic diagram of

coating layer; (b) The result of thermodynamic calculation

17 / 29
 Fig. 1 Fabrication flow chart of silica sol coated lightweight aggregates

18 / 29
Fig.2 Physical appearance and microstructure of the LMA before and after the coating process: (a) pristine LMA;

(b) SC-LMA; (c) pristine LMA; (d) SC-LMA; (e) SC-LMA

19 / 29
Fig.3 Schematic diagram of the alkali exposure: (a) graphite crucible; (b) graphite crucible was put in the alumina

crucible embedded by coke

20 / 29
 Fig.4 Physical appearances of the specimens: (a) before the alkali exposure; (b) after the alkali exposure

21 / 29
 Fig. 5 Phase compositions of (a) the fired specimens and (b) the specimens after alkali exposure

22 / 29
 Fig. 6 Microstructures of porous aggregates after firing at 1500°C: (a) the LMA in specimen U100; (b) the

SC-LMA in specimen C100; (c) the edge of SC-LMA, a coating layer was found; (d) a higher magnification of the

coating layer of SC-LMA

23 / 29
24 / 29
 Fig. 7 7 Microstructures of the specimens after alkali exposure: (a) a dense aggregate in the reference specimen;

(b) the matrix in the reference specimen; (c) the LMA in specimen U50; (d) the EDS mapping of figure (c); (e) a

higher magnification of the LMA in (c); (f) the SC-LMA in specimen C50; (g) the EDS mapping of figure (f)

25 / 29
Fig. 8 Relative content of K element among Al, Si, O, K in the matrix: (a) reference specimen, U50 and U100; (b)

reference specimen, C50 and C100

26 / 29
 Fig. 9 Corrosion behaviours of K vapor to the aggregate: (a) dense aggregate; (b) LMA; (c) SC-LMA

27 / 29
 Fig. 10 Thermodynamic analysis of the LMA in K vapor environment at 1000°C

28 / 29
Fig.11 Thermodynamic analysis of the SC-LMA in K vapor environment at 1000°C: (a) the schematic diagram of

the coating layer; (b) The result of thermodynamic analysis

29 / 29