Corrosion Science xxx (xxxx) xxx–xxx

Contents lists available at ScienceDirect

Corrosion Science
journal homepage: www.elsevier.com/locate/corsci

Degradation mechanisms of alumina-chromia refractories for secondary

copper smelter linings
⁎
Liugang Chena,b, , Annelies Malflieta, Jef Vleugelsa, Bart Blanpaina, Muxing Guoa
a
KU Leuven, Department of Materials Engineering, Arenberg Kasteelpark 44, Leuven, 3001, Belgium
b
Zhengzhou University, Henan Key Laboratory of High Temperature Functional Ceramics, Zhengzhou, 450001, China

A R T I C LE I N FO A B S T R A C T

Keywords: To assess the viability of an alumina-chromia refractory alternative for secondary copper smelters, the de-
Ceramic gradation behavior of alumina-chromia refractories in a secondary copper smelting slag and a Cu-CuxO mixture
High temperature corrosion was studied via lab scale refractory finger testing. Microstructural characterization of the worn refractory
EPMA samples resulted in a comprehensive understanding of the corrosion and penetration resistance and a compar-
SEM
ison with the degradation of magnesia-chromite refractories. Alumina grains in the refractory are preferentially
Copper
corroded by the slag. The alumina-chromia refractory exhibits superior infiltration resistance to the slag and the
Cu-CuxO mixture compared to magnesia-chromite refractories.

1. Introduction
Long-lasting refractory linings for furnaces in copper-making in-
dustries, where high cost magnesia-chromite bricks are the most ex-
tensively used material, are desirable based on economic and en-
vironmental considerations. An extended lifespan of the refractory
lining can reduce production losses during downtime of the furnace [1].
Because of the possible leaching of Cr(VI), used Cr2O3-containing re-
fractories are treated as potentially hazardous waste [1,2]. A prolonged
lining lifetime could alleviate the generation of spent magnesia-chro-
mite bricks. Therefore, continuous and consistent endeavors are being
made to improve the lifetime of refractory linings of various copper-
making furnaces via lining concept optimization [1], slag engineering
[3,4] ceramic/refractory engineering [5] and/or process engineering
[1]. Among these approaches, the most clear-cut way to enhance the
performance of furnace linings is through the use of new refractory
types that better withstand the combination of mechanical, thermal and
chemical stresses imposed on the lining [1]. This has led to the con-
sideration of replacing the magnesia-chromite brick type by alternative
refractory types [6–12].
Alumina-chromia, which was also written as Al2O3-Cr2O3 [6–8],
refractories are characterized by a high corrosion resistance to acidic
slags and excellent mechanical properties [6,7]. Therefore, this specific
refractory type, which is commonly used to withstand rigorous en-
vironments like waste incinerators [6], glass tanks [7] and coal gasifiers
[7,9], has been considered as an alternative material for the lining of
copper-making furnaces. Among the mechanical, thermal and chemical
loads, the chemical degradation of the refractory lining caused by liquid
slags is typically the most severe and can lead to a loss of structural
integrity of the brick [1,4], making the lining more susceptible to
thermal shock, mechanical stresses and erosion. Therefore, under-
standing the occurring degradation mechanisms in the refractory in
contact with slag is indispensable in assessing the viability of the alu-
mina-chromia refractory for copper-making furnaces.
In some studies the corrosion and infiltration resistance of alumina-
chromia and magnesia-chromite refractories in different copper pro-
duction processes have been compared [8,10–12]. The alumina-
chromia brick exhibits a superior performance against slag attack,
thereby having a longer service life. Laboratory scale experiments
comparing the wear behavior of alumina-chromia and magnesia-chro-
mite bricks against fayalite slag have been carried out by Zou et al. [11]
and Gregurek et al. [12] via rotary kiln tests. The alumina-chromia
refractories were less worn than the magnesia-chromite refractories
under the same testing conditions. Microstructural characterization and
thermodynamic analyses indicate that the newly-formed FeCr2O4 and
FeAl2O4 spinel phases from the slag/refractory interactions generate a
protective layer and thus hinder the further corrosion of the refractory
[11,13]. Similarly, Rigby [10] figured out that a reaction layer was
generated from the interactions between the Cr2O3-rich matrix of alu-
mina-chromia refractories and infiltrating CuxO, thereby preventing the
further ingress of slags and the spalling of the refractory. As a result, for
instance, the skimming lip of an anode vessel mouth lined with 30 wt%
Cr2O3 alumina-chromia refractories has shown a longer lifetime (over
nine months) than magnesia-chromite refractories (three months) [10].

⁎
Corresponding author at: KU Leuven, Department of Materials Engineering, Arenberg Kasteelpark 44, Leuven, 3001, Belgium.
E-mail address: liugang.chen@kuleuven.be (L. Chen).

https://doi.org/10.1016/j.corsci.2018.03.039
Received 6 December 2017; Received in revised form 19 March 2018; Accepted 21 March 2018
0010-938X/ © 2018 Elsevier Ltd. All rights reserved.

Please cite this article as: Chen, L., Corrosion Science (2018), https://doi.org/10.1016/j.corsci.2018.03.039
L. Chen et al. Corrosion Science xxx (xxxx) xxx–xxx

The success of such initiatives consequently make alumina-chromia Table 2

refractories to be considered as replacement materials with a prolonged Composition and properties of the investigated alumina-chrome refractory.
lifetime in lining secondary copper smelters. Overall composition (in wt%) Bulk Apparent Cold
However, the degradation mechanisms and outstanding perfor- density porosity crushing
mance of alumina-chromia refractories in fayalitic slags may not be Al2O3 Cr2O3 MgO SiO2 Fe2O3 (g·cm−3) (%) strength
completely applicable for secondary copper smelting processes. (MPa)

Compared to the primary copper production, a broad stream of sec-
ondary raw materials containing different contaminations are fed in
secondary copper smelters [1,14]. This difference in feed will lead to a
difference in slag composition. ZnO, for instance, can be present in the
secondary copper smelting slag and changes both the physical and
chemical properties of the slag, thus affecting the corrosion resistance
of refractory linings. The presence of ZnO in the slag was found to in-
fluence the degradation behavior of magnesia-chromite refractories
[15–17]. The degradation behavior of alumina-chromia and magnesia-
chromite refractories in contact with calcium-ferrite type slags (57 wt%
Fe2O3 and 19 wt% CaO) has also been tested by Gregurek et al. [12].
Unlike in fayalitic slags, alumina–chromia bricks were more severely
corroded than magnesia-chromite refractories in calcium-ferrite slags.
Therefore, as a precondition for assessing the feasibility of the alumina-
chromia refractory alternative for secondary copper smelting, it is vital
to identify and understand the occurring degradation mechanisms of
the refractory in slag systems that are relevant for secondary copper
smelting.
In this work, the degradation mechanisms of alumina–chromia re-
fractories in a Cu-CuxO mixture and a secondary copper smelting slag
were investigated. The influence of slag agitation, which was mimicked
via refractory finger rotation, and temperature on the corrosion beha-
vior was also studied. The degradation mechanisms of alumina-chromia
refractories are discussed and compared to those of the magnesia-
chromite refractories reported in previous work [15–18] to evaluate the
viability of using alumina-chromia refractory linings in secondary
copper smelters.
2. Experimental
82.76 11.00 0.74 0.77 0.09 3.52 12.4 194
Ar gas (passing the gas through silica gel and Mg turnings furnace
operating at 500 °C to remove traces of moisture and oxygen) was
blown into the furnace at a flow rate of around 0.4 l/min to simulate the
reducing atmosphere used in the industrial process. A cylindrical re-
fractory finger sample was moved down and maintained at a distance of
10 mm above the Mo crucible for 10 min to preheat the sample. Then
the refractory finger sample was immersed into the molten slag for 4 h
with a rotation speed of 12 rpm or 36 rpm. To investigate the infiltra-
tion behavior of the Cu-CuxO mixture, a static refractory finger test was
used. 400 g Cu-CuxO mixtures (1 wt% Al2O3, 4 wt% Cu2O and 95 wt%
Cu) [17] were added in an alumina crucible thereby making sure that
there was enough liquid to immerse the samples. The filled alumina
crucible was put in the vertical tube furnace and heated to 1180 °C to
melt the mixtures. After melting the mixtures for 30 min, the finger
sample was inserted in the furnace and kept above the alumina crucible
at a distance of 10 mm to preheat the sample for 10 min. Subsequently,
the refractory finger was immersed into the molten mixtures for 30 min.
Ar gas, which generally contains around 2 ppm O2, was blown into the
furnace tube with a flow rate of 0.6 l/min during the experiment. After
immersion for the required time, the refractory finger sample was re-
moved from the molten slag or Cu-CuxO mixture and quenched in N2
steam. The refractory samples were cut with a diamond saw at 1 cm
below the bath line for microstructural analysis. The slag was cooled
down to room temperature in the furnace, removed from the Mo cru-
cible and ground to a particle size smaller than 80 μm for compositional
analyses.

2.1. Materials
The composition of the secondary copper smelting slag, which is of
the fayalitic type, as supplied by a copper producer is listed in Table 1.
Minor elements, such as K, S, P, Cu, Mn, Pb, and Ni, are accounted as
“others” in Table 1. The cylindrical refractory samples (dia-
meter = about 15 mm; length = around 50 mm) were cut from a
commercially available alumina-chromia brick. The chemical compo-
sition and physical properties of the investigated alumina-chromia
brick are provided in Table 2.
2.2. Experimental set-up and procedure
The experimental set-up and the procedure for refractory finger
tests are similar to the one used in our previous work in which the
degradation of magnesia-chromite refractory has been tested [15,17].
The parameters for the refractory finger tests are listed in Table 3.
Approximately 170 g of the as-delivered secondary copper smelting slag
was placed in a Mo crucible (inside diameter = 40 mm;
height = 80 mm), heated in a vertical tube furnace (GERO HTRV
100–250/18, with MoSi2 heating elements) to 1200 °C or 1180 °C, and
kept at that temperature for 50 min to ensure the molten state. Purified
2.3. Sample analysis techniques
The sliced refractory sections were embedded in a low viscosity
resin (Epofix) by vacuum impregnation, ground with silicon carbide
papers and polished with diamond paste. The polished specimens were
carbon coated for compositional and microstructural analysis.
Compositional analyses were performed by full quantitative electron
probe X-ray microanalysis coupled with wavelength dispersive spec-
troscopy (EPMA-WDS, JEOL JXA-8530F). It was operated using an ac-
celeration voltage of 15 kV and a probe current of 15 nA. The oxygen
content was not measured directly. Instead the oxidation state of the
element was selected a priori. Periclase (MgO), hematite (Fe2O3),
willemite (Zn2SiO4), fluorspar (CaF2), chromium oxide (Cr2O3) and
obsidian ((Na,K,Al,Fe) silicate glass) were used as standards.
Microstructural analyses were obtained using scanning electron mi-
croscopy (SEM, FEI, XL-30). The ground slag before and after the cor-
rosion tests was analysed with a wavelength-dispersive X-ray fluores-
cence spectrometer (WDXRF, Zetium PANalytical). Although both Fe2+
and Fe3+ are possibly present in the slag and chromite spinel phases, all
Fe is calculated in the form of “FeO” for the purpose of presentation.

Table 1
The chemical composition of the secondary copper smelting slag used in the experiments (by XRF).
A2O3 ZnO FeO SiO2 Na2O MgO CaO Cr2O3 PbO CuO Others

wt% 3.1 9.4 55.6 23.7 2.3 0.6 0.7 0.8 1.0 0.6 1.9
mol% 2.2 8.2 54.8 28.0 2.6 1.1 0.9 0.4 0.9 0.5 1.7

2
L. Chen et al.
Table 3
Parameters of the refractory finger tests.
Exp. no. Rotation speed (rpm) Temperature (°C)
R12 12 1200
R36 36 1180
S0 0 1180
a
The oxygen partial pressure is measured in the outlet of the Ar gas (Rapidox 2100, Cambridge Sensotec Ltd.).
2.4. Methodology of thermodynamic calculation
The solubility limit of Al2O3 was calculated according to the same
approach reported for MgO [16,18]. Thermodynamic calculations were
performed with the FactSage software [19] (version 6.4). Equilibrium
calculations were performed with the equilibrium module EQUILIB,
which is based on the minimization of the Gibbs free energy, and the FT
oxide, FS stel and Fact PS databases. The following possible solution
phases are chosen in the calculations: (1) FT oxide-slag (molten oxide
phase); (2) FT oxide-monoxide (oxide solid solution); (3) FT oxide-
spinel (spinel solid solution); (4) FT oxide-olivine; (5) FT oxide-zincite;
(6) FT oxide-willemite, (7) FS stel-FCC (metallic iron) and (8) FACT PS
(gas phase). The dissolution of Al2O3 was modeled at 1200 °C by re-
lative addition of pure Al2O3 (x, in gram) into the slags ((100-x), in
gram). ZnO, FeO, SiO2, Al2O3 and CaO are the slag components for the
calculation. FeO/SiO2, Al2O3/SiO2 and CaO/SiO2 mass ratios in the slag
were kept at 1.4, 0.18 and 0.03, respectively. ZnO content in the slag
varied from 0 to 20 wt%. The maximum Al2O3 dissolution amount into
the ZFS slag is determined when a solid phase precipitates from the
liquid slag.
3. Result and discussion
3.1. As-delivered refractories
Fig. 1 shows backscattered electron (BSE) images of the as-delivered
alumina-chromia refractory. The chemical composition of the phases in
the refractory are presented in Table 4. The investigated alumina-
chromia refractory is a direct-bonded mixture of alumina (Al2O3, ‘Am’
in Fig. 1), alumina-chromia solid solution (Al2O3-Cr2O3, ‘AC’ in Fig. 1),
zirconia (ZrO2, ‘Z’ in Fig. 1), and CaO·6Al2O3 (CA6, ‘CA6’ in Fig. 1) and
Al2O3·P2O5 (‘AP’ in Fig. 1b) impurity phases. The alumina grains appear
dark gray in BSE mode, with a maximum grain size of 3 mm (Fig. 1a)
and a chromia-rich rim, directly binding with the refractory matrix
(Fig. 1b). Between the alumina grains and the refractory matrix, mi-
crocracks can be seen (Fig. 1a). Inside the alumina grains, micro-pores
and cracks, zirconia grains and CaO·6Al2O3 phases can be observed. The
Fig. 1. Backscattered electron (BSE) images of the as-delivered alumina-chromia refractory samples, (b) enlarged view of the selected area in (a); Am = alumina,
AC = alumina-chromia solid solution, Z = zirconia, CA6 = CaO·6Al2O3, AP = Al2O3·P2O5; P = pore.
Corrosion Science xxx (xxxx) xxx–xxx
Time (hours) pO2 (atm) Others
a
4 < 10−19 Sample size: D = 15 mm × L = 50 mm
Slag amount: ∼170 g
0.5 2 ppm O Mixture amount: 400 g
refractory matrix is rich in Cr2O3 and composed of alumina-chromia
solid solution grains, smaller alumina grains, zirconia and Al2O3·P2O5.
The observed Al2O3·P2O5 originates from the phosphate binder for the
brick preparation, and zirconia is commonly employed to improve the
thermal shock resistance of this refractory type [6,8]. The CaO·6Al2O3
and Al2O3·P2O5 impurity phases are dispersed in respectively the alu-
mina grains and refractory matrix. Interconnected pores and cracks
form a network in the as-delivered refractory brick.
3.2. Visual inspection and global slag analyses
Fig. 2 shows the refractory fingers before and after corrosion testing.
Before the rotating test, gray-appearing alumina grains with a max-
imum size of 5 mm and dark gray-appearing refractory matrix are ob-
served on the surface of the refractory fingers (Fig. 2a and c). After
testing for 4 h with a rotating speed of 12 rpm (Fig. 2b) or 36 rpm
(Fig. 2d), large cavities appear at the locations of the alumina grains
although the surface is covered with a layer of frozen slag. These ob-
servations suggest that the alumina grains in the alumina-chromia re-
fractory are preferentially attacked by the secondary copper smelting
slag whereas the Cr2O3-rich refractory matrix is little corroded, similar
as direct-bonded magnesia-chromite refractories [15–17].
The preferential corrosion of alumina grains is also implied by the
combination of increased Al2O3 level with little increase in the Cr2O3
level in the tested slag (Table 5 and Fig. 3). The high solubility limit of
Al2O3 in the slag provides the thermodynamic driving force for the
alumina dissolution. The results also indicate that the decrease of op-
erating temperature could hinder the corrosion of alumina grains. As
shown in Table 5 and Fig. 3, the Al2O3 content increase in the slag in
test R36, where a stronger slag agitation and a lower testing tempera-
ture were employed, is 5.0 wt%, whereas that in the R12 test is 5.6 wt%.
To compare the corrosion resistance of alumina-chromia and mag-
nesia-chromite refractories, the increase of Al2O3 and MgO in the tested
slag and the predicted solubility limits of Al2O3 and MgO under the
same experiment/calculation conditions are shown in Fig. 4. The so-
lubility limit of Al2O3 was calculated according to the same approach
reported for MgO [16,18]. The MgO content increases with 0.8 wt%
3
L. Chen et al. Corrosion Science xxx (xxxx) xxx–xxx

Table 4
Chemical compositions of the phases in the as-delivered alumina-chrome refractory, as determined by EPMA-WDS.
Phases Chemical composition (wt%)

Al2O3 Cr2O3 ZrO2 CaO MgO SrO Ce2O3 P2O5

Alumina 93.6 ± 2.9 5.4 ± 2.7 – – – – – –

Alumina-chromia solid solution 47.8 ± 24.3 49.7 ± 23.2 – – – – – –
Zirconia – – 93.8 ± 1.0 – – – – –
CaO·6Al2O3 88.2 ± 0.6 – 5.2 ± 0.5 0.9 ± 0.2 1.6 ± 0.1 1.8 ± 0.7 –
Al2O3·P2O5 39.5 ± 1.3 – – – – – – 55.6 ± 2.0

Fig. 2. Appearance of the refractory finger samples in R12 (a, b) and R36 (c, d)
tests, (a) and (c) before testing, and (b) and (d) after testing, showing the
corrosion of alumina grains.
after testing at 1200 °C with a rotation speed of 36 rpm (Fig. 4a) [16]. In
comparison, the Al2O3 content increases with 2.5 wt % and 1.9 wt% in
the tests R12 (12 rpm at 1200 °C) and R36 (36 rpm at 1180 °C), re-
spectively, for the same corrosion duration, although the slag before
testing has a higher Al2O3 level. The extent of dissolution of alumina
from alumina-chromia refractories is therefore more pronounced than
that of periclase from magnesia-chromite refractories in a slag with
∼9 wt% ZnO. This would imply an inferior slag corrosion resistance of
alumina-chromia bricks compared to magnesia-chromite refractories.
The different extents of dissolution of alumina and periclase should
be ascribed to the different solubility limits of Al2O3 and MgO. In the
slag containing 10 wt% ZnO the solubility limit of Al2O3 (4.4 wt%) is
higher than that of MgO (2.1 wt%) (Fig. 4b), thereby providing a
greater thermodynamic driving force for the Al2O3 dissolution. How-
ever, in the secondary copper smelting process, the ZnO level in the slag
can vary depending on the feed materials and/or operation conditions
[16], leading to different solubility limits of MgO and Al2O3 in the slags.
The solubility limit of Al2O3 continuously decreases with increasing
ZnO content (Fig. 4b), suggesting that the increase of ZnO concentra-
tion can suppress the Al2O3 dissolution. Once the ZnO content in the
slag is raised to 20 wt%, the solubility limits of Al2O3 and MgO are
comparable and ∼2.0 wt%. As a result, the extent of dissolution of
alumina from alumina-chromia refractories is expected to be of the
same level as that of periclase in magnesia-chromite bricks. This will
Fig. 3. Comparison of the Al2O3 content in the secondary copper smelting slag
before and after refractory finger tests.
lead to a comparable corrosion resistance of alumina-chromia and
magnesia-chromite refractories in the secondary copper smelting slag
having 20 wt% ZnO. However, this merit needs to be further in-
vestigated.
In summary, alumina grains in the alumina-chromia refractory are
preferentially corroded by the secondary copper smelting slag like
periclase in magnesia-chromite refractories, and the extent of dissolu-
tion of alumina grains is more pronounced than that of periclase in
secondary copper smelting slag with 9 wt% ZnO, but would be com-
parable in a slag with 20 wt% ZnO.
3.3. Worn refractory microstructure analyses
3.3.1. Microstructure observations
The worn microstructures from the tested refractory fingers are il-
lustrated in Figs. 5–7. The chemical composition of the phases in the
worn refractory sample from the test R12 are presented in Table 6. The
refractory surfaces are covered with frozen slag layers of 200 μm and
500 μm for respectively test R12 and R36 test (Fig. 5). The thicker slag
layer in test R36 is indicative for a higher slag viscosity, which is ex-
pected from the lower testing temperature in R36. The concave shape of
the interface of the alumina grains with the slag indicates that the
alumina grains were corroded by the slag. As a result, the Cr2O3-rich
refractory matrix protrudes into the slag (Fig. 5b). These observations
further confirm that the alumina grains in the refractory are preferably

Table 5
Chemical compositions of the slags after rotating refractory finger tests as measured with XRF.
Exp. no. Chemical composition

Al2O3 ZnO FeO SiO2 Na2O MgO CaO Cr2O3 PbO CuO Others

R12 wt% 5.6 7.6 50.6 28.2 2.5 0.7 1.0 0.1 0.8 0.4 2.5
mol% 3.8 6.5 49.3 32.9 2.8 1.2 1.3 0.0 0.6 0.3 1.3
R36 wt% 5.0 8.6 51.3 26.6 2.5 0.6 0.9 0.7 0.9 0.4 2.5
mol% 3.5 7.5 50.3 31.3 2.8 1.1 1.1 0.3 0.6 0.4 1.1

4
L. Chen et al.
Fig. 4. Comparison of (a) the increase in Al2O3 and MgO [16] concentration in secondary copper smelting slags and (b) the predicted solubility limits of Al2O3 and
MgO synthetic fayalite slags with a fixed FeO/SiO2 ratio of about 1.4.
corroded by the secondary copper smelting slag, whereas the Cr2O3-rich
matrix is attacked less. At the refractory/slag interface (Figs. 5 and 6), a
continuous, dense gray-appearing layer (‘H’ in Figs. 5 and 6) with a
thickness of 20 μm covers the fresh surface of the alumina grains.
Chemical analyses by EPMA-WDS on this gray layer indicate that this
would be a hercynite-based spinel consisting of 30.2 mol% FeO,
20.5 mol% ZnO, 2.6 mol% MgO, 44.9 mol% Al2O3 and 1.2 mol% Cr2O3
(Table 6).
The infiltration behavior of secondary copper smelting slag in the
alumina-chromia refractory is shown in Fig. 7. A densified refractory
matrix is observed over a distance of 4.5 mm from the refractory surface
towards the center after 4 h testing. An elemental mapping of the
marked area in the densified zone indicated in Fig. 7 is shown in Fig. 8.
Since Si, Fe and Ca are not detectable in the matrix of the as-delivered
refractory (Table 4), the region containing Fe, Si and Ca is the in-
filtrated zone. Fig. 8 illustrates that the infiltrating slag is mainly found
at the grain boundaries, and in the cracks and/or open pores in the
refractory. Spinel phases, identified by the region composed of Al, Cr
and Fe in Fig. 8, are observed next to the infiltrating slag in the den-
sified refractory zone.
Fig. 9 presents a detailed microstructure of the CaO·6Al2O3 phase in
the bulk refractory sample from the R12 test, showing the corrosion
behavior of CaO·6Al2O3. Next to the CaO·6Al2O3, a Na, K, Ca, Al, Si and
S containing phase is observed (point 001, Table 7), indicating the
diffusion of Na2O, K2O and SiO2 into the CaO·6Al2O3 phase. With the
continuous diffusion of Na2O, K2O and SiO2 from the slag, the melting
point of CaO·6Al2O3 phase will decrease, thereby leading to the
Fig. 5. Back scattering electron (BSE) images of the worn samples from R12 (a) and R36 (b) tests, showing the refractory surface zone: Am = alumina; AC = alumina-
chromia solid solution; Cra = crack; Z = zirconia; H = (Zn,Fe,Mg)(Al,Cr)2O4 hercynite-based spinel.
5
Corrosion Science xxx (xxxx) xxx–xxx
transformation of CaO·6Al2O3 phase into a liquid phase. As a result of
the CaO·6Al2O3 corrosion by the infiltrating slag, new paths are gen-
erated for the further slag infiltration.
A number of large cracks with a maximum width of 15 μm are ob-
served between the larger alumina grains and the refractory matrix in
the densified refractory zone (Figs. 5 and 7). These cracks are believed
to be generated during quenching or cutting of the refractory samples,
because these cracks are not infiltrated. Between the newly-formed
spinel and alumina grains/alumina-chromia solid solution, no obvious
cracks can be found (Figs. 5–7).
3.3.2. Spinel phase formation
As discussed in Section 3.3.1, only spinel phases were generated
from the slag/refractory interactions. The spinel phases are observed at
the refractory surface and in the densified refractory zone next to the
infiltrating slag (Figs. 5–7).
The composition analyses of the spinel phases at the refractory
surface (Table 6) indicate that this is a (Zn,Fe,Mg)(Al,Cr)2O4 hercynite-
based spinel. It is believed that such hercynite-based spinel precipitates
at the refractory/slag interface once the slag is locally saturated with
Al2O3 according to the following reaction:
FeOslag + ZnOslag + MgOslag + Al2O3,slag + Cr2O3,slag/refracory →
(Zn,Fe,Mg)(Al,Cr)2O4,refractory (1)
This newly-formed continuous and dense hercynite-based spinel
layer can act as a protective layer between the slag and the alumina
L. Chen et al. Corrosion Science xxx (xxxx) xxx–xxx

Fig. 6. BSE images of the worn samples from R12 (a and b) and R36 (c) tests, showing the slag/refractory interactions at the refractory surface: Am = alumina;
AC = alumina-chromia solid solution; P = pore; Z = zirconia; IS = infiltrating liquid slag; Fay = (Fex,Zn1−x)2SiO4 fayalite; M = (Fe,Zn)(Fe, Al)2O4 magnetite-based
spinel; H = (Zn,Fe,Mg)(Al,Cr)2O4 hercynite-based spinel; Sp = (Fe,Zn,Mg)(Cr,Al)2O4 chromia-rich spinel.

grains. The infiltration of the slag into the alumina grains through the
grain boundaries can thus be physically inhibited. Like olivine on
periclase grains [15,16], the newly-formed spinel layer on alumina
grains would also protect alumina from the direct attack of the slag and
slow the alumina dissolution into the slag.
To identify the formation mechanism of the spinel next to the in-
filtrating slag in the densified refractory zone, the spinel grains at a
distance of 2 mm from the refractory surface were selected for mi-
croprobe analysis. As shown in Table 6, this spinel contains a higher
level of Cr2O3 (22.5 mol%) compared to the hercynite-based spinel
(Cr2O3, 1.2 mol%) at the refractory surface. Because of the low solu-
bility of Cr2O3 in the secondary copper smelting slag [15], the high
Cr2O3 content indicates that these spinel phases are most likely formed
by removing FeO, ZnO and MgO from the infiltrating slag. The reaction
forming this Cr2O3-rich spinel phase can be written as:
FeOslag + ZnOslag + MgOslag + Al2O3,slag,refractory + Cr2O3,refractory →
(Fe,Mg,Zn)(Cr,Al)2O4,refractory (2)
Studies on the formation mechanisms of spinel layers in the FeO-
Cr2O3[20], MgO-Cr2O3 [20] and MgO-Al2O3 [21] etc. systems show
that in all systems the spinel phases form by the counter-diffusion of
cations, i.e. the outward diffusion of Al3+ and Cr3+ from the chromite
to the slag and the inward diffusion of Fe2+, Zn2+ and Mg2+ from the
slag to the chromite, through the newly formed solid spinel layer.
Therefore, this Cr2O3-rich spinel phase grew on both the Al2O3-Cr2O3/
spinel and spinel/slag interfaces, due to the supply of fresh Al2O3 and
FeO from the bath through the liquid channels. The growth of this

Fig. 7. Overview of the worn refractory sample from the R12 test, showing the liquid slag infiltration in the refractory: Am = alumina; AC = alumina-chromia solid
solution; CA6 = CaO·6Al2O3; P = pore; Cra = crack; Z = zirconia; IS = infiltrating liquid slag; H = (Zn,Fe,Mg)(Al,Cr)2O4 hercynite-based based spinel;
Sp = (Fe,Zn,Mg)(Cr,Al)2O4 chromia-rich spinel.

6
L. Chen et al. Corrosion Science xxx (xxxx) xxx–xxx

Table 6
Chemical compositions of the phases in the alumina-chrome refractory from Test R12, as determined by EPMA-WDS (in mol%).
Phase Position Chemical composition

FeO ZnO Al2O3 Cr2O3 MgO SiO2 CaO Na2O P2O5 Cu2O

Spinel Surface 30.2 ± 0.6 20.5 ± 0.3 44.9 ± 0.9 1.2 ± 0.6 2.6 ± 0.2 – – – – –
2 mm from the surface 40.0 ± 1.0 5.3 ± 0.3 28.3 ± 6.3 22.5 ± 5.6 2.8 ± 0.4 – – – – –
Slag Surface 41.6 ± 2.0 4.7 ± 0.4 5.2 ± 0.2 – 2.0 ± 0.3 44.2 ± 1.8 1.4 ± 0.1 1.1 ± 0.1 – –
2 mm from the surface 16.5 ± 0.6 0.2 ± 0.1 16.3 ± 0.4 – 0.5 ± 0.1 56.5 ± 2.5 3.4 ± 0.5 2.6 ± 0.1 2.5 ± 1.0 –

newly-formed Cr2O3-rich spinel fills the open pores and cracks in the
refractory, thereby densifying the refractory.
The microstructural observations could be summarized as follows:
(1) only spinel phases are generated from the slag/refractory interfacial
reactions; (2) slag infiltrated into the refractory along the open-pore
network and the grain boundaries. Cr2O3-rich spinel was generated next
to the infiltrating slag. The newly-formed spinel continuously grew with
the supply of fresh Al2O3 and FeO from the infiltrating slag and thus
filled the open pores and cracks in the refractory, leading to the den-
sification of the refractory; (3) the corrosion of CaO·6Al2O3 phases oc-
curs in the slag infiltrated refractory part, which provides paths for the
further ingress of the slag.
3.4. Slag and Cu-CuxO infiltration behavior
The chemical composition of the infiltrating slag at a distance of
2 mm from the refractory surface is shown in Table 6. Compared with
the frozen slag at the refractory surface, the infiltrating slag has a lower
FeO, ZnO and MgO content, but higher levels of SiO2, Na2O, CaO, Al2O3
and P2O5. The depletion of FeO, ZnO and MgO is mainly due to the
consumption of these oxides in the refractory/slag interactions forming
spinel phases, whereas the slag constituents which react little with the
refractory components, such SiO2 and Na2O, are enriched. The in-
creased Al2O3 and CaO content in the infiltrating slag is caused by the
combined effect of the depletion of slag constituents and the dissolution
of refractory components, for example CaO·6Al2O3. The presence of
P2O5 in the infiltrating slag is mainly caused by the dissolution of
Al2O3·P2O5 phase from the refractory. Because of the depletion in net-
work breaking oxides (FeO, ZnO and MgO) and the enrichment of
network forming oxides (SiO2 and P2O5), the viscosity of the infiltrating
slag increases. Additionally, as discussed in Section 3.3.2, the newly-
formed Cr2O3-rich spinel could fill the pores and cracks in the refractory
matrix and thus block the channels for further slag infiltration into the
refractory.
Fig. 10 shows an overview microstructure of the alumina-chromia
refractory finger sample after testing for 30 min in molten Cu-CuxO.
Limited Cu-CuxO penetrated into the alumina-chromia refractory. In-
filtrated Cu-CuxO mixtures are mainly observed at the grain boundaries
and/or at the interconnected open-pore network up to a distance of
400 μm from the refractory surface. Deeper in the refractory, the Cu-
CuxO mixture is dispersed as micrometer sized particles. These ob-
servations indicate that the alumina-chromia refractory has a superior
infiltration resistance to molten Cu-CuxO.
Compared with magnesia-chromite refractories [15–17], alumina-
chromia refractories have a better infiltration resistance against both
the secondary copper smelting slag and the Cu-CuxO mixture. It is ob-
served in previous work [15–17] that both the slag and Cu-CuxO mix-
ture infiltrated up to the center (7.5 mm) of magnesia-chromite re-
fractory finger samples after testing for 4 h and 30 min, respectively. In
contract, the infiltration depths for the slag and Cu-CuxO mixture in the
alumina-chromia refractory fingers are 4.5 mm and 400 μm after testing
for 4 h and 30 min, respectively. The superior infiltration resistance to
the slag and Cu-CuxO mixture of alumina-chromia refractory can thus
hinder the corrosion in the infiltrated refractory part and severe

Fig. 8. (a) EPMA-WDS elemental mapping and (b) a combination of the elemental mappings of the selected zone ‘x’ in Fig. 7, showing the presence of infiltrating slag
in the refractory matrix and formation of (Fe,Zn,Mg)(Cr,Al)2O4 chromia-rich spinel: P = pore. The green area in (b) is because of the depth variation of the pore.

7
L. Chen et al. Corrosion Science xxx (xxxx) xxx–xxx

Fig. 9. (a) BSE image of CaO·6Al2O3 phase in an alumina grain, and (b) energy dispersive spectroscopy (EDS) spot measurements of selected point in the figure,
showing the corrosion of CA6 by the infiltrating slag: Am = alumina; CA6 = CaO·6Al2O3; IS = infiltrating slag.

Table 7 Additionally, the alumina-chromia refractory has a superior infiltration

The chemical composition of the selected point in Fig. 9a, as determined EDS
(atom%).
O Ce Na Mg Al Si S K Ca Ti
52.61 0.03 0.15 1.53 15.67 17.47 1.23 1.72 9.55 0.04
spalling [10]. The lower apparent porosity of 13.6% for the alumina-
chromia refractory compared with 15–16.5% for the magnesia-chro-
mite refractories [15,17], is considered to be the cause for this. Besides,
the differences in alumina-alumina and periclase-periclase interfaces,
and alumina-chromia and periclase-chromite interfaces, can affect the
wettability, and thereby the infiltration, of liquid slag and the Cu-CuxO
mixture. For instance, the Cu–CuxO preferentially wets the Fe-enriched
surface layer of chromite grains in fused-grain magnesia-chromite re-
fractories instead of the chromite grains in direct-bonded refractories
[17]. In the alumina-chromia refractory, the Cr2O3-rich matrix does not
contain Fe2O3 (Table 4), and thus would be less wetted by the Cu–CuxO
mixture. However, supporting literature data on the wetting behavior
and contact angles have not been found. Additionally, the blockage
effect of the newly-formed spinel phase would be another reason to
prevent the further ingress of slag and Cu-CuxO mixture.
3.5. Industrial relevance
Due to the higher solubility limit of Al2O3, the extent of dissolution
of alumina grains is pronounced compared to that of periclase in the
slag having 9 wt% ZnO. When the ZnO level reaches to 20 wt%, the
solubility limits of Al2O3 and MgO are comparable. We expect that this
will lead to a similar extent of dissolution of alumina and periclase, and
thus a comparable corrosion resistance of alumina-chromia and mag-
nesia-chromite refractories in the slag containing 20 wt % ZnO.
resistance against both the slag and Cu-CuxO mixture compared to
magnesia-chromite refractories. These results can lead to a conclusion
that the alumina-chromia refractory alterative may have a longer life-
time compared to magnesia-chromite refractories for secondary copper
smelters operating with ZnO-rich fayalitic slags. However, since the
wear of refractory linings is caused by a combination of chemical,
thermal and mechanical loads, industrial trials are deemed necessary to
validate our results.
4. Conclusion
Refractory finger corrosion tests were performed to investigate the
degradation mechanism of alumina-chromia refractories in the sec-
ondary copper smelting slag with a ZnO content of ∼9 wt% as well as
Cu-CuxO mixture. From the macroscopical and microstructural ob-
servations, the following conclusions are drawn:
(1) It is preferentially that the alumina in the alumina-chromia re-
fractory is corroded by the slag. The Cr2O3-rich phase in the re-
fractory is little attacked. The high solubility limit of Al2O3 in the
slag is the thermodynamic driving force for the faster dissolution of
the alumina grains.
(2) The slag infiltrated into the alumina-chromia refractory through the
open-pore network and the grain boundaries, and by the corrosion
of the CaO·6Al2O3 impurity phases. The infiltration of Cu-CuxO
mixture into the refractory via the open-pore network and the grain
boundaries was limited.
(3) (Fe,Mg,Zn)(Al,Cr)2O4 spinel phases are generated from the slag/
refractory interactions at the refractory surface and in the in-
filtrated refractory. The spinel was formed next to the infiltrating
slag and filled the open pores and cracks in the refractory, leading
to a densified refractory zone.

Fig. 10. Light optical microscopy images of the infiltration of Cu-CuxO mixture in the alumina-chromia refractory for 30 min at 1180 °C; Am = alumina, Cu = Cu-
CuxO mixture.

8
L. Chen et al.
Acknowledgements
The authors are grateful to the support from the Hercules
Foundation (project ZW09-09) in the use of the FEG-EPMA system and
the close co-operation with engineers from Metallo-Chimique. L. Chen
would like to thank the Research Fund of KU Leuven for his post-doc-
toral fellowship (PDM/16/140).
References
[1] A. Malfliet, S. Lotfian, L. Scheunis, V. Petkov, L. Pandelaers, P.T. Jones, B. Blanpain,
Degradation mechanisms and use of refractory linings in copper production pro-
cesses: a critical review, J. Eur. Ceram. Soc. 34 (2014) 849–876.
[2] V. Petkov, P.T. Jones, E. Boydens, B. Blanpain, P. Wollants, Chemical corrosion
mechanisms of magnesia-chromite and chrome-free refractory bricks by copper
metal and anode slag, J. Eur. Ceram. Soc. 27 (2007) 2433–2444.
[3] K.H.T. Matsui, T. Ikemoto, K. Sawano, A consideration on corrosion of magnesia
refractory brick by silicate slag, J. Tech. Assoc. Refract. 22 (2002) 302–309.
[4] L. Scheunis, A.F. Mehrjardi, M. Campforts, P.T. Jones, B. Blanpain, E. Jak, The
effect of phase formation during use on the chemical corrosion of magnesia –
chromite refractories in contact with a non-ferrous PbO–SiO2 based slag, J. Eur.
Ceram. Soc. 34 (2014) 1599–1610.
[5] Y.Y. Deng, H.Z. Wang, H.Z. Zhao, Influence of chrome-bearing sols vacuum im-
pregnation on the properties of magnesia-chrome refractory, Ceram. Int. 34 (2008)
573–580.
[6] D. Chen, A. Huang, H. Gu, M. Zhang, Z. Shao, Corrosion of Al2O3–Cr2O3 refractory
lining for high-temperature solid waste incinerator, Ceram. Int. 41 (2015)
14748–14753.
[7] M. Nath, A. Ghosh, H.S. Tripathi, Hot corrosion behavior of Al2O3-Cr2O3refractory
by molten glass at 1200 °C under static condition, Corros. Sci. 102 (2016) 153–160.
[8] S. Breyner, K. Santowski, T. Prietl, A. Spanring, A. Franzkowiak, Thermal shock
resistant alumina-chromia products for the copper industry, Proceeding Copper,
Santiago, Chile, 2013, pp. 451–460.
Corrosion Science xxx (xxxx) xxx–xxx
[9] J.P. Bennett, K. Kwong, Refractory liner materials used in slagging gasifiers,
Refract. Appl. New 9 (2004) 20–25.
[10] A.J. Rigby, The use of alumina-chrome refractories in the high wear areas of anode
refining vessels, Proceeding Copper, Santiago, Chile, 2013, pp. 497–505.
[11] M. Zou, M. Jiang, Y. Qian, X. Shi, Simulation study on corrosion resistance of Al2O3-
Cr2O3 and MgO-Cr2O3 bricks to ISA furnace slag, Refract. Naihuo Cailiao 41 (2007)
180–182 (in Chinese).
[12] D. Gregurek, A. Ressler, V. Reiter, A. Franzkowiak, A. Spanring, T. Prietl, Refractory
wear mechanisms in the nonferrous metal industry: testing and modeling results,
JOM 65 (2013) 1622–1630.
[13] Z. Chen, On refractory material and formation of protection layer for Ausmelt
copper smelting furnace, China Nonferrous Metall. 2 (2005) 27–30.
[14] J. Wood, S. Creedy, R. Matusewicz, M. Reuter, Secondary copper processing using
Outotec Ausmelt TSL technology, Proceeding of Met Plant, (2010), pp. 460–467.
[15] L. Chen, M. Guo, H. Shi, L. Scheunis, P.T. Jones, B. Blanpain, A. Malfliet, The in-
fluence of ZnO in fayalite slag on the degradation of magnesia-chromite refractories
during secondary Cu smelting, J. Eur. Ceram. Soc. 35 (2015) 2641–2650.
[16] L. Chen, M. Guo, H. Shi, S. Huang, P.T. Jones, B. Blanpain, A. Malfliet, Effect of ZnO
level in secondary copper smelting slags on slag/magnesia-chromite refractory in-
teractions, J. Eur. Ceram. Soc. 36 (2016) 1821–1828.
[17] L. Chen, S. Li, P.T. Jones, M. Guo, B. Blanpain, A. Malfliet, Identification of mag-
nesia-chromite refractory degradation mechanisms of secondary copper smelter
linings, J. Eur. Ceram. Soc. 36 (2016) 2119–2132.
[18] L. Chen, M. Guo, H. Shi, P.T. Jones, B. Blanpain, A. Malfliet, Influence of FeO/SiO2
and CaO/SiO2ratios in iron-saturated ZnO-rich fayalite slags on the corrosion of
MgO, J. Am. Ceram. Soc. 99 (2016) 3754–3760.
[19] C.W. Bale, E. Bélisle, P. Chartrand, S.A. Decterov, G. Eriksson, A.E. Gheribi, K. Hack,
I.H. Jung, Y.B. Kang, J. Melançon, A.D. Pelton, S. Petersen, C. Robelin, J. Sangster,
P. Spencer, M.A. Van Ende, FactSage thermochemical software and databases,
2010–2016, Calphad 54 (2016) 35–53.
[20] K. Nagata, R. Nishiwaki, Y. Nakamura, T. Maruyama, Kinetic mechanisms of the
formations of MgCr2O4 and FeCr2O4 spinels from their metal oxides, Solid State
Ionics 49 (1991) 161–166.
[21] I. Ganesh, A review on magnesium aluminate (MgAl2O4) spinel: synthesis,proces-
sing and applications, Int. J. Mater. Res. 58 (2016) 63–112.