Ceramics International 44 (2018) 15453–15463

Contents lists available at ScienceDirect

Ceramics International
journal homepage: www.elsevier.com/locate/ceramint

MgO fumes as a potential binder for in situ spinel containing refractory T

castables
⁎
A.P. Luza, , L.B. Consonia, C. Pagliosab, C.G. Anezirisc, V.C. Pandolfellia
a
Federal University of São Carlos, Materials Engineering Department, Rod. Washington Luiz, km 235, São Carlos, SP 13565-905, Brazil
b
RHI Magnesita, Research and Development Center, Praça Louis Ensch, 240, Contagem, MG 32210-902, Brazil
c
Institute of Ceramic, Glass and Construction Materials, TU Bergakademie Freiberg, Agricolastrasse 17, 09599 Freiberg, Germany

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

Keywords: MgO is pointed out as an alternative binder for refractory materials, mainly for systems where the presence of
MgO CaO might not be desired. Selecting the most suitable magnesia source is an important step as its purity and
Brucite reactivity should influence the hydration reaction, leading to binding effect or cracks. This work investigated the
Hydration design of vibratable high-alumina compositions bonded with MgO fumes [which is a very fine powdered oxide
Green mechanical strength
(d < 3 µm) resulting from the production process of electrofused magnesia] and/or dead-burnt magnesia
Refractory castables
(d < 212 µm). Acetic and formic acids were added to the castables during their processing steps in order to
adjust the density of active sites for Mg(OH)2 formation and control the crystal growth of this phase. The green
mechanical strength and thermomechanical performance (cold and hot mechanical strength, thermal shock,
refractoriness under load, corrosion, etc.) of designed MgO-bonded compositions were analyzed. Improved green
mechanical strength and crack-free samples were obtained when adding up to 6 wt% of MgO fumes to the
refractories and processing them with aqueous solutions with 3 wt% of formic acid. The compositions with 6 wt
% of magnesia fumes resulted in samples with flexural strength in the range of 12.0 MPa after curing at 50 °C/
24 h and similar green mechanical strength (12.9 MPa) as the ones bonded with 4.0 wt% of calcium aluminate
cement after drying at 110 °C for 24 h, which highlights the great potential of this MgO source. Despite the
enhanced green mechanical strength, alumina-based castables containing 6 wt% of MgO (fumes, dead-burnt or
their blend) showed low mechanical strength at intermediate temperatures and high linear expansion, as a
consequence of the in situ spinel phase formation above 1200 °C. Thus, better densification, improved HMOR,
thermal shock resistance and corrosion behavior were obtained for the castables prepared with less MgO fume
contents.

1. Introduction to their performance and lower cost compared to CAC products.

MgO + H2 O → Mg (OH )2 (1)
Recently some efforts have been made in the refractory field in
order to develop CaO-free binders (i.e., based on alumina-silica, mag- Selecting the most suitable magnesia source must take into account
nesia or magnesia-silica) that can replace the traditional calcium alu- the purity and reactivity of this oxide, as the smaller the crystal size, the
minate cement (CAC) in castable compositions [1–8]. Magnesia might higher the specific surface area and its hydration likelihood [6,9].
be used as a hydraulic binder due to its reaction with water (Eq. (1)), Therefore, aiming to prevent the damage associated with the MgO
resulting in brucite [Mg(OH)2] precipitation in the refractory micro- hydration in refractory castables, one may choose to (i) induce faster
structure and, consequently, increasing its overall green mechanical brucite formation before the composition setting time, when the
strength when this reaction is properly controlled. Despite the chal- molded material still has enough freedom to accommodate stresses,
lenges related to the expressive volumetric expansion associated with and/or (ii) change the morphology of this hydrated phase by adjusting
magnesium hydroxide formation (which can lead to crack generation the pH solution, temperature and using hydrating agents [5,6,10–13].
and damage to the refractory structure during curing and drying steps Ammonium chloride, magnesium acetate, magnesium nitrate, nitric
[9]), the development of MgO-based binders would result in positive acid, acetic acid, magnesium chloride, sodium acetate, and others, are
technological and economical advantages for refractory producers due some examples of compounds that can affect the MgO hydration

⁎
Corresponding author.
E-mail address: [email protected] (A.P. Luz).

https://doi.org/10.1016/j.ceramint.2018.05.201
Received 27 April 2018; Received in revised form 17 May 2018; Accepted 23 May 2018
Available online 24 May 2018
0272-8842/ © 2018 Elsevier Ltd and Techna Group S.r.l. All rights reserved.
A.P. Luz et al. Ceramics International 44 (2018) 15453–15463

reaction rate [6,10–14]. Santos et al. [10] reported that carboxylic broad temperature range (30–1600 °C).
acids can adsorb onto MgO surface in aqueous medium leading to the
formation of higher amounts of sites for brucite nucleation, hence, in- 2. Experimental
ducing faster hydration of this oxide. This process takes place by the
heterolithic decomposition of these molecules on the magnesia surface, Two magnesia sources (MgO fumes = MF and dead-burnt = M)
releasing H+ cations and carboxylate anions that can be adsorbed at the supplied by RHI Magnesita (Brazil) were evaluated in this work. MF is a
sites occupied by O2- or Mg2+, respectively [10,15]. Besides that, the very fine and reactive magnesia powder obtained during the production
density of active sites on MgO surface is affected by the size of the process of electrofused MgO, whereas the dead-burnt one resulted from
selected carboxylic acid [10,16]. Therefore, based on the studies MgCO3 calcination. More details of MF and M features are shown in
mentioned above, a limited growth of this hydrated phase can minimize Table 1.
crack and flaw generation in the refractories’ structure. Vibratable castable formulations containing coarse and fine tabular
In fact, magnesia should play two important roles when in- alumina (d < 6 mm, Almatis, Germany), calcined and reactive Al2O3
corporated into high-alumina refractories, as (i) its controlled hydration (CL370 and CT3000SG, Almatis, Germany) and magnesia (MgO fumes
can induce the binding effect among the castables’ components and and/or dead-burnt) were designed according to Alfred's particle
enhance their overall green mechanical strength and, (ii) the further packing model (q = 0.26 [19], Table 2). Firstly, castable compositions
heating of such materials should favor the Al2O3 and MgO reaction and, containing 0.5–6.0 wt% of MgO fumes were prepared in order to ana-
consequently, the spinel (MgAl2O4) phase formation (which usually lyze the binding effect derived from the hydration process of this oxide.
takes place above 1000 °C and may also give rise to high volumetric In a second moment, refractories presenting 6 wt% of MF and/or M
expansion) [17]. The in situ generation of this latter compound is de- (which should give rise to ~21 wt% of in situ MgAl2O4 formation at
sired as it can increase the castables’ slag corrosion resistance due to the high temperatures) were also evaluated as, based on previous in-
spinel ability to trap Mn2+, Fe2+ and Fe3+ ions in its structure via solid vestigations [20–23], the optimal spinel content to improve the slag
solutions. Thus, the resulting slag becomes silica-rich, leading to a more resistance and wear mechanisms of these high-alumina MgAl2O4-con-
viscous liquid, which usually inhibits its penetration into the ceramic taining castables would be in the 20–25 wt% range. Less than 20 wt% of
lining porosity [18]. Nevertheless, both transformations (brucite and spinel content usually leads to a higher corrosion rate, whereas above
spinel formation) can result in interfacial separation between ag- 25 wt% it results in greater penetration [24].
gregates and matrix or microcracks, depending on the overall expansion Additionally, dos Santos et al. [5,10] pointed out that carboxylic
and microstructural features (such as the available porosity). This is the acids can be added as agents to adjust the kinetic of MgO hydration in
reason why the effects of these transformations on the refractory's castable compositions, reducing the development of cracks even when
properties should be monitored as a function of temperature to better using very reactive magnesia sources. Therefore, formic (molar mass:
understand such reactions and provide some insights that may lead to 46 g/mol, purity = 85%, Labsynth, Diadema, Brazil) and acetic (molar
novel compositions with advanced performance. mass: 60 g/mol, purity = 99.7%, Labsynth, Diadema, Brazil) acids were
Considering the promising results presented by Souza et al. [6] and selected as nucleation site activators for further brucite crystal forma-
Santos and colleagues [5,10] when adding formic, acetic or propionic tion and growth. Aqueous solutions containing 0.3–3.0 wt% of car-
acids to high-alumina refractory castables containing caustic or dead- boxylic acids (Table 2) were prepared and added to the castables’ dry
burnt magnesia sources, this paper investigates the design of vibratable compositions during their mixing process.
high-alumina compositions bonded with MgO fumes [powdered oxide A total of 0.1 or 0.2 wt% of the dispersant Castament® FS60 (Basf,
(d < 3 µm) resulting from the production process of electrofused mag- Germany) was used during the preparation of the designed formula-
nesia] and/or dead-burnt magnesia (d < 212 µm, which is traditionally tions. The dry-compositions were homogenized for 1 min and mixed
added to MgO-containing castables due to its limited reactivity with with water for an additional 3 min in a rheometer [25]. The required
water). Acetic and formic acids were added to the castables during their liquid content and attained flow (ASTM C1445) are shown in Table 2.
processing steps in order to adjust the density of active sites for Mg After that, prismatic samples (150 mm × 25 mm × 25 mm) were cast,
(OH)2 formation and control the growth of their crystals. The most cured at 50 °C for 24 h and dried at 110 °C for another 24 h. The curing
suitable amount of these acids to be incorporated into the dry-mixes temperature of 50 °C was selected in order to compare the developed
was selected according to the different features of the used MgO pow- mechanical strength of the prepared MgO-bonded castables with
ders (Table 1). Not only the green mechanical strength (which is di- equivalent calcium alumina cement (CAC)-containing compositions.
rectly related to the binding action derived from the magnesia hydra- The green mechanical strength of the cured and dried refractories
tion), but also the thermomechanical performance (cold and hot was evaluated via 3-point bending tests (ASTM C133-97), whereas the
mechanical strength, thermal shock, refractoriness under load, etc.) of apparent porosity was determined according to ASTM C380-00 using
the designed MgO-bonded compositions were analyzed to highlight the kerosene as the immersion liquid. A total of 5 samples were analyzed
benefits and drawbacks of this refractory system when exposed to a for each selected testing condition and the presented values consist of
the average result as well as the calculated standard deviation.
Table 1 After selecting the most promising castable compositions, additional
Chemical composition and physical properties of the evaluated magnesia samples were prepared and fired in the range of 300–1550 °C for 5 h.
sources. Cold flexural strength, apparent porosity, assisted sinterability, creep,
refractoriness under load, permanent dimensional change, thermal
Oxides (wt%) MgO fumes (MF) Dead-burnt MgO (M)
shock resistance and hot modulus of rupture (HMOR) measurements
CaO 0.61 0.80 were selected to evaluate the refractories’ thermo-mechanical perfor-
MgO 97.34 98.12 mance. The latter test was carried out at 300 °C, 600 °C, 900 °C, 1200 °C
SiO2 0.64 0.34
and 1450 °C (using five samples pre-fired for 5 h at the same testing
Others 1.41 0.74
CaO:SiO2 (molar ratio) 1.01 2.52 temperature or at 1500 °C) in HBTS 422 equipment (3-point bending
Properties device, Netzsch, Germany).
Specific surface area (m2/g) 5.60 0.56 Assisted sinterability, refractoriness under load and creep mea-
Particle size (μm) surements were carried out in order to find out the linear dimensional
d10 0.25 1.77
change and thermal stability of the samples up to 1550 °C. Cylindrical
d50 0.80 15.22
d90 2.96 50.49 specimens (height and external diameter = 50 mm and central inner
diameter = 12.5 mm) were prepared according to 51053 DIN standard

15454
A.P. Luz et al. Ceramics International 44 (2018) 15453–15463

Table 2
General information of the designed refractory castables.
Raw-materials (wt%) Screening tests Additional compositions

0.5FM 1FM 2FM 4FM 6FM 6M 0.5FM + 5.5 M 1FM + 5 M

Tabular alumina (d < 6 mm) 90.0 90.0 90.0 90.0 88.0 88.0 88.0 88.0
Calcined alumina (CL370) 6.0 6.0 6.0 6.0 6.0 6.0 6.0 6.9
Reactive alumina (CT3000SG) 3.5 3.0 2.0 – – – – –
MgO fumes 0.5 1.0 2.0 4.0 6.0 – 0.5 1.0
Dead-burnt MgO – – – – – 6.0 5.5 5.0
Aqueous solution content (wt%) 4.0 4.0 4.0 4.0 4.0–4.3 4.0–4.3 4.3 4.3
Carboxylic acid amount in the aqueous solution (wt%) 0.3 0.3 0.3 0.3 0.3–3.0 0.3–3.0 2.0 2.0
Vibratable flow (%) 173 167 161 157 120–190 109–160 170 162

and calcined at 600 °C for 5 h (assisted sinterability) or pre-fired at 3. Results and discussion
1550 °C for 5 h (RUL) or 24 h (creep). During creep and RUL testing, the
castables were heated up to 1450 °C for 24 h or 1600 °C for 5 h, re- 3.1. Green mechanical strength of alumina-based castables bonded with
spectively, under a compressive load of 0.2 MPa and with a heating rate MgO fumes
of 5 °C/min. On the other hand, the calcined samples were submitted to
a heating rate of 3 °C/min up to 1550 °C and kept under a compressive The binding action of magnesia in refractory castables relies on the
load of 0.02 MPa for the assisted sinterability measurements. formation of a brucite layer on the MgO grains in the microstructure of
Additionally, permanent dimensional change (PLC, ASTM C113-14) of the cured samples, filling in the pores and, therefore, improving the
the bar samples (based on measurements of the samples’ length before overall green mechanical resistance of these products. However, MgO
and after firing at 300 °C, 600 °C, 900 °C, 1200 °C and 1500 °C for 5 h. A hydration is followed by a significant volumetric expansion due to the
total of 5 specimens were analyzed in each selected condition) were density difference between magnesia (ρMgO = 3.53 g/cm3) and brucite
performed in order to identify the influence of expected phase trans- (ρMg(OH)2 = 2.40 g/cm3), which commonly results in crack formation
formations on the castables’ final dimension. and some damage to the refractory properties [9]. Thus, it is important
Aiming to understand the elastic modulus (E) changes with tem- to master this phase transformation by using suitable sources and
perature, dried and pre-fired (1400 °C/5 h) samples of the designed amounts of magnesia, and/or adding hydrating agents to the compo-
refractories were also analyzed according to the ASTM C 1198–91 sitions.
standard using the bar resonance technique (Scanelastic equipment, In a first attempt to analyze the likelihood of using MgO fumes as
ATCP, Brazil), which is based on the sample excitation and detection of binder in high-alumina castables, compositions containing 0.5–6.0 wt%
the correspondent vibration spectrum, using piezoelectric transducers. of this magnesia source with aqueous solutions containing 0.3–3.0 wt%
Further details about this technique can be found elsewhere [26,27]. of acetic or formic acid, were processed. The selection of 0.3 wt% of the
Tests were carried out in the 30–1400 °C temperature range in air (pO2 carboxylic acids to be added to the mixtures was based on previous
= 0.21 atm) with heating and cooling rates of 2 °C/min. The same studies [6,10] when considering the incorporation of caustic or dead
method was also used to follow (at room temperature) the E decay as a burnt magnesia to refractories.
function of thermal shock cycles (ASTM C1171-91) at 1025 °C (10 cy- Fig. 1 points out the cold flexural strength and apparent porosity of
cles, 1000 °C) using four pre-fired (1400 °C/5 h) castable samples. castable samples obtained after curing at 50 °C for 24 h and drying at
Cylindrical samples (with an external diameter of 50 mm, 50 mm in 110 °C for another 24 h. Cured specimens showed a continuous increase
height, a central inner hole with 20 mm in diameter and 25 mm deep) of their flexural strength when changing the MgO content from 0.5 to
were also prepared, cured at 50 °C, dried at 110 °C and fired at 600 °C or 6.0 wt% (Fig. 1a). Besides that, the compositions processed with acetic
1500 °C for 5 h for the corrosion measurements. A total of 4 specimens acid presented a higher mechanical strength after curing than the ones
were analyzed in each evaluated condition. Before these tests, the inner containing a formic one, as highlighted in previous studies
cup of the castable specimens were filled in with 10 g of a synthetic slag [5,6,10,29–31].
(Table 3) and the set (refractory + slag powder) was placed in an Mg(OH)2 nuclei generation are not homogeneous and depends on
electrical furnace (Lindberg Blue, Lindberg Corporation, USA) and he- the physical properties of the magnesia source, as well as some pro-
ated up to 1500 °C for 2 h with a heating rate of 2 °C/min. After cooling, cessing conditions (i.e., temperature, time, pH, etc.) and additives.
the corroded samples were cut and had their cross sections polished and When adding a hydrating agent to the castables’ mixture, a larger
prepared for scanning electron microscopy analysis (FEI, Phillips, USA). number of nuclei will be quickly generated and activated at the be-
Besides that, the slag penetration area was calculated using Image J ginning of the oxide-water interaction due to the adsorption of these
1.42q software (Wayne Rasband, National Institutes of Health, USA), species onto the MgO particles’ surface (which enhances the interaction
according to the procedure described by Braulio et al. [28] in order to of other species, such as water, with this oxide) [5,6,10,16]. Ad-
quantify the corrosion damage. ditionally, the short formic acid molecules might activate more sites
The phases predicted to be formed during the initial slag-castable located close to each other and, consequently, lead to a lower growth of
interaction in the equilibrium condition at 1500 °C were calculated for the brucite crystals (due to their impingement) on the samples con-
the compositions containing 1 wt% or 6 wt% MgO and using the taining this additive [16]. This may be the reason why for the selected
FactSage™ software (databases = FactPS and FToxid and the Equilib curing time (24 h), the samples prepared with formic acid showed
calculation modulus). lower mechanical strength than the ones processed with acetic acid
(Fig. 1a). Despite the higher modulus of rupture of the acetic acid
Table 3
containing castables (Fig. 1a), the apparent porosity values of the re-
Chemical composition of the synthetic slag used in the corrosion cup tests. fractories prepared with formic acid were slightly lower than the ones
containing the acetic one (Fig. 1c).
Oxides (wt%) SiO2 TiO2 Al2O3 Fe2O3 CaO MgO K2 O SO3 F
The highest mechanical resistance for the dried samples was
Slag 28.8 0.6 10.7 1.2 42.7 7.2 0.6 1.6 6.2 reached by the compositions containing 1 wt% of MgO fumes and
greater contents of this oxide resulted in crack formation and the drop

15455
A.P. Luz et al. Ceramics International 44 (2018) 15453–15463

12 12
Cured samples (50°C/24h) Dried samples (110°C/24h)
Acetic acid Acetic acid
10 Formic acid 10 Formic acid

Flexural strength (MPa)

Flexural strength (MPa)

8 8

6 6

4 4

2 2

0 0
0.5MF 1MF 2MF 4MF 6MF 0.5MF 1MF 2MF 4MF 6MF
20 20
Cured samples (50°C/24h) Dried samples (110°C/24h)
Acetic acid Acetic acid
16 Formic acid Formic acid
15
Apparent porosity (%)

Apparent porosiy (%)

12
10

5
4

0 0
0.5MF 1MF 2MF 4MF 6MF 0.5MF 1MF 2MF

Fig. 1. Cold flexural strength (a and b) and apparent porosity (c and d) of the high-alumina castables containing 0.5–6.0 wt% of magnesia fumes (MF) after curing the
samples at 50 °C for 24 h and drying at 110 °C for another 24 h. The mixtures were processed with the addition of 0.3 wt% of acetic or formic acids and 4.0 wt% of
distilled water.

of the flexural strength of these materials (Fig. 1b). This feature might apparent porosity of the samples obtained after curing at 50 °C for 24 h
be related to the further hydration of the unreacted magnesia particles and drying at 110 °C for another 24 h.
when in contact with water vapor generated during the drying step of Due to the high reactivity of the MgO fumes, small cracks could still
the castables at 110 °C and the insufficient amount of carboxylic acid be observed on the refractories surface after drying (which led to the
used to prepare the compositions (as the actual magnesia source eval- decay of their flexural strength, Fig. 2a) even when increasing the acetic
uated in this study has a much higher specific surface area than the ones acid contents up to 3.0 wt% in the liquid phase added to the dry-mixes
analyzed in other studies [6,10]). Some cracks could be observed on the during the processing of the 6MF compositions. The optimization of the
surface of the dried samples of 4MF and 6MF refractories (leading to the kinetic and growth of the brucite grains in such castables could be
decrease of their cold mechanical strength, Fig. 1b), which points out achieved when adding 3.0 wt% of formic acid (3FAc, Fig. 2b) to the
the uncontrolled growth of the brucite phase in the resulting micro- mixture containing magnesia fumes. In this case, the cold flexural
structure during drying. Due to the presence of these cracks, apparent strength of 6MF-3FAc composition after curing was higher than the one
porosity of the dried 4MF and 6MF specimens were not measured and obtained for the reference material (CAC-bonded refractory containing
only the results of the other compositions are presented in Fig. 1d. 4 wt% of cement) considered in this study, 12.0 ± 0.5 MPa and
Comparing the cold flexural strength of the cured and dried MgO- 6.5 ± 0.9 MPa, respectively. Similar green mechanical strength could
bonded castables (Fig. 1a and b) with a similar vibratable high-alumina be detected for both materials after drying at 110 °C/24 h,
castable containing 4.0 wt% of calcium aluminate cement (Secar 71, 12.9 ± 1.4 MPa (6MF) and 13.7 ± 1.2 MPa (4CAC - reference).
Kerneos, France) as the main binder (after curing at 50 °C/24 h = Such results highlight that not only the higher reactivity of MF [its
6.5 ± 0.9 MPa and drying at 110 °C/24 h = 13.7 ± 1.2 MPa), it could specific surface area is ~5.6 m2/g whereas M (dead-burnt) is 0.56 m2/
be inferred that magnesia fumes has a great potential binding action but g], but also the length of the carboxylic acid molecules might have
additional adjustments in the carboxylic acid contents were still re- affected the magnesia hydration process. As the dissociation constant
quired to prevent the cracking of the samples during drying. (Ka) of formic acid is lower than the one of the acetic one (FAc =
1.77 × 10−4 and AAc = 1.76 × 10−5[32]), a greater number of dis-
sociated molecules of the former should be available in the aqueous
3.2. Selection of the most suitable carboxylic acid for the processing of
solutions to interact with the sites for adsorption located on the MF
castables containing 6 wt% of MgO
surface. Hence, most likely a suitable content of dissociated species
(derived from formic acid) could be obtained when adding 3.0 wt% of
Aiming to adjust the amount of the hydrating agents to prevent the
formic acid to the liquid phase during the processing of MgO fumes-
cracking of the MgO-bonded castables containing 6 wt% of this oxide
bonded castables.
(MF or M), some tests were carried out considering the addition of
On the other hand, the dead-burnt magnesia-bonded refractories did
aqueous solutions, comprised by 4.3 wt% of distilled water plus 0.3,
not show any cracking in all evaluated conditions (when varying the
1.0, 2.0 or 3.0 wt% of acetic or formic acids, during the mixing step of
carboxylic acids content and exposing the samples at 50 °C or 110 °C for
these compositions. Fig. 2 points out the mechanical strength and

15456
A.P. Luz et al. Ceramics International 44 (2018) 15453–15463

18 18
Cured at 50°C/24h: 6MF 6M Cured at 50°C/24h: 6FM 6M
Dried at 110°C/24h: 6MF 6M (a) Dried at 110°C/24h: 6FM 6M (b)
15 15
Samples with cracks Samples with cracks

Flexural strength (MPa)

Flexural strength (MPa)

12 12

9 9

6 6

3 3

0 0
0.3AAc 1AAc 2AAc 3AAc 0.3FAc 1FAc 2FAc 3FAc

Acetic acid content (wt.%) Formic acid content (wt.%)

20 20
Cured at 50°C/24h: 6MF 6M Cured at 50°C/24h: 6MF 6M
Dried at 110°C/24h: 6M (c) Dried at 110°C/24h: 6MF 6M (d)

16 16

Apparent porosity (%)

Apparent porosity (%)

12 12

8 8

4 4

0 0
0.3AAc 1AAc 2AAc 3AAc 0.3FAc 1FAc 2FAc 3FAc

Acetic acid content (wt.%) Formic acid content (wt.%)

Fig. 2. Cold flexural strength (a and b) and apparent porosity (c and d) of the Al2O3-MgO castables containing 6.0 wt% of magnesia fumes (6 MF) or dead-burnt MgO
(6 M) after curing the samples at 50 °C for 24 h and drying at 110 °C for another 24 h. The mixtures were processed with the addition of 0.3–3.0 wt% of acetic (AAc) or
formic (FAc) acids and 4.3 wt% of distilled water.

24 h). Nevertheless, the lower reactivity of this MgO resulted in speci- hydrating agent to obtain an appropriate binding effect among the
mens with reduced flexural strength (0.7–1.4 MPa or 1.9–4.3 MPa after various components of the refractory compositions.
curing or drying, respectively) due to the slower hydration rate of this
oxide. Consequently, the less intense brucite generation [10] in the
3.3. Castable compositions containing a blend of dead-burnt and MgO
resulting microstructure of the M-bonded castables led to samples with
fumes
high porosity levels (varying from 12.0% to 16.0%), as indicated in
Fig. 2c and d.
As 6MF still presented some likelihood to crack formation after
Therefore, it is important to optimize the source and amount of the
drying (especially when increasing the size of the prepared specimens),

12 20
50°C/24h 110°C/24h (a) 50°C/24h 110°C/24h (b)
Flexural strength (MPa)

15
Apparent porosity (%)

Samples
10
with
cracks
4
5

0 0
.3AA
c Ac Ac 2AAc FAc Ac Ac Ac 2AAc FAc
M-2A M-2F -5M- -5M-2 -0.3A M-2A M-2F -5M- -5M-2
-5M-0 F-5.5 F-5.5 1M F 1MF 1MF
-5M F-5.5 F-5.5 1M F 1MF
1MF 0.5M 0.5M 0.5M 0.5M

Fig. 3. (a) Cold flexural strength and (b) apparent porosity of the Al2O3-MgO castables containing a mixture of magnesia fumes (MF) and dead-burnt MgO (M) after
curing the samples at 50 °C for 24 h and drying at 110 °C for another 24 h. The mixtures were processed with the addition of 0.3 or 2.0 wt% of acetic (AAc) or formic
(FAc) acids with 4.3 wt% of distilled water.

15457
A.P. Luz et al. Ceramics International 44 (2018) 15453–15463

a third attempt to control the drawbacks of the magnesia hydration magnesia (1MF-5M-2FAc, 6FM-3FAc and 6M-2FAc) showed a perma-
reaction and obtain refractories with suitable green mechanical nent linear change around 1.2–1.8% after firing at 1500 °C (Fig. 3c). As
strength was based on preparing castables containing a mixture of a consequence of this expansion, higher porosity levels and lower me-
dead-burnt and magnesia fumes. Thus, compositions comprising chanical resistance was detected for these MgO-rich compositions after
0.5–1.0 wt% of MF plus 5.0–6.0 wt% of M (Table 2) were processed firing at high temperatures.
with aqueous solutions with 2.0 wt% of formic or acetic acids and cured As reported in the literature [36,37], many factors can affect the
at 50 °C/24 and 110 °C/24 h. overall expansion derived from spinel formation, such as the solid so-
Fig. 3 shows the cold mechanical strength and apparent porosity of lubility of alumina at high temperatures, the purity of magnesia and
the prepared samples containing both MgO sources. As pointed out in alumina particles, the number of contacts, the diffusion of MgO vapor
Fig. 3a, when only 0.3 wt% of acetic acid was used during the pre- and also the higher Mg2+ mobility in the oxygen lattice. Moreover, the
paration of the 1MF-6M castable, crack formation was still observed in Kirkendall effect [37] can take place when the inter-diffusion of one of
the dried samples, resulting in the decrease of their flexural strength. the present species (Mg2+ or Al3+) is much faster than the other,
However, the addition of 2.0 wt% of carboxylic acids in the liquid was leading to the formation of a pore in the initial location of the lower
already enough to inhibit such flaws due to the brucite generation. mobility component. Hence, despite the enhanced green mechanical
Although crack-free samples were obtained with the blend of the strength of the 6MF-3FAc refractory, the higher reactivity and reduced
evaluated magnesia, the flexural strength of such compositions (Fig. 3a) particle size of MgO fumes (see Table 1) induced a more effective spinel
was lower than the ones observed for 6MF-3FAc (Fig. 2b) and the re- formation in the resulting microstructure, which favored greater ex-
ference (4CAC) refractory, mentioned before. Among the analyzed ca- pansion and porosity increase of the samples.
stables, the best performance was identified for the 1MF-5M-2FAc This aspect (dimension stability of the materials) can also be ana-
mixture, which showed green mechanical strength of 3.7 ± 0.4 MPa lyzed via assisted sintering measurements. As indicated in Fig. 5, this in
(after curing) and 6.1 ± 0.4 MPa (after drying) and porosity levels situ method highlighted the faster reaction between Al2O3 + MgO
around 13–14%. fumes for the 6MF-3FAc sample (resulting in an expansion rate peak
Based on these data, some compositions presenting the most pro- ~1200 °C, Fig. 5b) and, consequently, an intense dL/L0 increase up to
mising performance (1MF-0.3AAc, 1MF-5M-2FAc, 6MF-3FAc and 6M- 2.9% due to the high amount of magnesia in this mixture (Fig. 5a). On
2FAc) were selected to have their cold and hot mechanical properties the other hand, instead of expansion, 1MF-0.3AAc refractory presented
evaluated after firing at 300–1500 °C for 5 h and also after curing (hot shrinkage in the 1200–1500 °C range, as the sintering and densification
elastic modulus, assisted sinterability, etc). Although the 1MF-0.3AAc of the microstructure was able to overcome the expansive behavior of
castable contained less MgO (1 wt%) and should give rise to only ~5 wt spinel generation (Fig. 5b).
% of spinel after firing at high temperatures, this refractory was crack- The other compositions (1MF-5M-2FAc and 6M-2FAc, containing
free and presented good flexural strength values (Fig. 1a and b) after 6 wt% of MgO) showed the beginning of their spinel formation around
curing and drying. Due to this performance, it was also decided to 1200 °C (Fig. 5a) and the expansion rate peak was only detected
characterize this composition and compare its results with the MgO-rich ~1450 °C (Fig. 5b) due to the lower specific surface area and coarser
systems (with 6 wt% of this oxide). size of the dead burnt MgO, when compared to the magnesia fumes.
Furthermore, according to the final dL/L0 results of the designed re-
3.4. Thermomechanical characterization of the most promising Al2O3-MgO fractories, 6MF-3FAc specimens showed a similar expansion to the ones
designed castables presented by Braulio et al. [38] when testing vibratable high-alumina
castables bonded with 6 wt% of CAC and containing 6 wt% of dead-
The previous sections presented the binding effect derived from the burnt (d < 45 µm) magnesia + 1 wt% of silica fumes (dL/L0 at 1500 °C
brucite generation, enhancing the green mechanical strength of the = 3.0%), whereas other Al2O3-MgO refractories bonded with 2 wt% of
castables. Nevertheless, when exposing the dried samples to higher CAC + 4 wt% of hydratable alumina also resulted in expansion levels of
temperatures, Mg(OH)2 decomposition should occur close to 400 °C approximately 2.0% at 1500 °C [34], as the one obtained for 1MF-5M-
[17,33], leading to the decrease of their modulus of rupture (Fig. 4a) 2FAc.
and porosity increase (Fig. 4b) at intermediate temperatures In addition to the high expansion of the castables containing 6 wt%
(~400–1000 °C). This explains why all tested castables, fired at 600 °C of MgO, pore generation should also result in the site previously oc-
and 900 °C for 5 h, showed flexural strength values around 1.0–3.1 MPa cupied by magnesia due to the distinct diffusion rate of the involved
(Fig. 4a). Such results are similar to the ones obtained by Braulio et al. species for MgAl2O4 formation [39]. Consequently, the samples’ stiff-
[34] for Al2O3-MgO (dead-burnt, d < 45 µm) refractories bonded with ness evolution and their densification level are expected to be affected
2 wt% of CAC + 4 wt% of hydratable alumina. These authors also in- by the MgO + Al2O3 reaction. Fig. 6 points out the hot elastic modulus
dicated that the same composition, but containing 6 wt% of calcium values measured for the dried (110 °C/24 h) and pre-fired (1500 °C/5 h)
aluminate cement as the binder, presented cold mechanical strength of materials during their heating-cooling cycles up to 1400 °C.
approximately 4–6 MPa between 600 °C and 800 °C due to the more In agreement with the cold flexural strength results of the dried
efficient binding action of CAC. samples (110 °C, Fig. 4a), the initial E values (showed in Fig. 6a) point
Magnesium aluminate spinel (MgAl2O4) formation took place when out the significant binding action of brucite grains in 6MF-3FAc (re-
firing the prepared Al2O3-MgO castables at 1200 °C or 1500 °C for 5 h. sulting in high stiffness at room temperature). However, the greater
This solid state transformation is carried out through the Al3+ and amount of Mg(OH)2 in the composition with 6 wt% of MgO fumes led to
Mg2+ ions counter-diffusion, at the MgO-MgAl2O4 and Al2O3-MgAl2O4 a marked drop of the measured elastic modulus between 300 and 400 °C
interfaces [17]. Moreover, according to the thermodynamic simulations due to the decomposition of this hydroxide. After this transformation, a
carried out with FactSage™ software, up to 5 wt% and 30 wt% of continuous and slightly E decrease was observed for all evaluated for-
MgAl2O4(ss) could be generated in the evaluated refractories containing mulations in the 400–1050 °C range, and only 1MF-0.3AAc refractory
1 wt% and 6 wt% of MgO, respectively, at 1500 °C and in equilibrium presented a significant densification of its microstructure, which was
conditions. This transformation is responsible for improving the mea- able to accommodate the spinel formation and result in samples with
sured flexural strength of the designed castables after firing at 1200 °C ~105 GPa after the cooling cycle (E increase of approximately 44%
and 1500 °C (Fig. 4a). Nevertheless, it was observed that the highest compared to the initial stiffness of this castable, Fig. 6a). Conversely,
mechanical strength was obtained for the 1MF-0.3AAc samples (which the other castables containing 6 wt% of magnesia showed less intense
contained only 1 wt% MgO) and, due to the expansive feature of the in sintering above 1050 °C and the pores and flaws derived from the ex-
situ MgAl2O4 formation [17,35], the compositions with 6 wt% of pansion associated with the MgAl2O4 formation induced the E drop of

15458
A.P. Luz et al. Ceramics International 44 (2018) 15453–15463

25 25
1MF-0.3AAc 1MF-5M-2FAc
6MF-3FAc 6M-2FAc (a) (b)
Cold flexural strength (MPa) 20 20

Apparent porosity (%)

15 15

10 10

5 5
1MF-0.3AAc 1MF-5M-2FAc
6MF-3FAc 6M-2FAc
0 0
300 600 900 1200 1500 300 600 900 1200 1500
Firing temperature (°C) Firing temperature (°C)
2.5
1MF-0.3AAc 1MF-5M-2FAc
2.0 6MF-3FAc 6M-2FAc (c)
Permanent linear change (%)

1.5

1.0

0.5

0.0

-0.5

-1.0
300 600 900 1200 1500
Firing temperature (°C)
Fig. 4. (a) Cold flexural strength, (b) apparent porosity and (c) permanent linear change of the Al2O3-MgO castable samples after pre-firing at 300 °C, 600 °C, 900 °C,
1200 °C and 1500 °C for 5 h.

Fig. 5. (a) dL/L0 (%) profile and (b) expansion rate (%/min) of the Al2O3-MgO castables during their first heating treatment up to 1500 °C. All samples were calcined
at 600 °C for 5 h before testing and the applied load during the assisted sintering tests was 0.02 MPa.

15459
A.P. Luz et al. Ceramics International 44 (2018) 15453–15463

180 180
1MF-0.3AAc 1MF-5M-2FAc 1MF-0.2AAc 1MF-5M-2FAc
(a) (b)
6MF-3FAc 6M-2FAc 6MF-3FAc 6M-2FAc
150 150

120 120
E (GPa)

E (GPa)
90 90

60 60

30 30

0 0
0 200 400 600 800 1000 1200 1400 0 200 400 600 800 1000 1200 1400
Temperature (°C) Temperature (°C)
Fig. 6. Elastic modulus evolution as a function of temperature of the Al2O3-MgO castables after (a) drying at 110 °C for 24 h or (b) pre-firing at 1500 °C for 5 h.

15 pre-fired samples: 6MF-3FAc = 150 GPa, 1MF-5M-2FAc = 64 GPa, 6M-

1MF-0.3AAc 1MF-5M-2FAc 2FAc = 46 GPa and 6MF-3FAc = 28 GPa.
6MF-3FAc 6M-2FAc The same trend observed for the cold flexural strength and hot
12
elastic modulus was also identified in the HMOR behavior of the Al2O3-
MgO castables between 300 and 1500 °C (Fig. 7). Hot modulus of
HMOR (MPa)

9 rupture values between 6 and 10 MPa was obtained at 300 °C and, with
the temperature increase up to 600 °C or 900 °C, a decay of the me-
chanical strength was detected due to the brucite decomposition and an
6 increase in the porosity level (Fig. 4b) in the refractories’ micro-
structure. Densification and spinel formation (X-ray diffraction results
not shown here) in the 1MF-0.3AAc formulation induced the increase of
3
HMOR already at 1200 °C, whereas 1MF-5M-2FAc and 6M-2FAc only
showed improvement of this property at 1500 °C (Fig. 7b). Additionally,
0 the major expansion of the 6MF-3FAc samples (due to spinel genera-
300 600 900 1200 1500 tion, Fig. 5) and higher porosity was responsible for inhibiting the
Testing temperature (°C) densification of this refractory, resulting in low stiffness (Fig. 6) and
cold and hot mechanical strength results (Fig. 4a and Fig. 7, respec-
Fig. 7. Hot modulus of rupture (HMOR) of the Al2O3-MgO castable samples
tively) after firing at 1500 °C for 5 h.
after pre-firing at the same testing temperature with a holding time of 5 h.
Regarding the thermal shock resistance of the designed composi-
tions, 1MF-0.3AAc presented enhanced elastic modulus values
the samples after this first heating-cooling cycle (6 M-2FAc = −8%, throughout the 10 evaluated cycles at 1025 °C (Fig. 8a), resulting in a
1MF-5M-2FAc = −37% and 6MF-3FAc = −75%, Fig. 6a). final E drop of 18.2% (Fig. 8b). On the other hand, in addition to the
Nevertheless, when submitting the specimens to a pre-firing step at lower E results, the compositions containing 6 wt% of MgO showed a
1500 °C for 5 h and measuring their E behavior in a further heating- marked decrease of their stiffness (down to 45.0–47.0% for 1MF-5M-
cooling cycle, almost no changes in the stiffness values could be de- 2FAc and 6M-2FAc or 67.5% for MF-3FAc, Fig. 8b) for the evaluated
tected for all compositions (highlighting that the main phase transfor- condition (ΔT ~1000 °C), indicating the damage of the refractory mi-
mations in such refractories took place during the first heating, Fig. 6b). crostructure due to the thermal stress.
The following elastic modulus could be obtained after analyzing the

200 0
1MF-0.3AAc 1MF-5M-2FAc 1MF-0.3AAc 1MF-5M-2FAc
(a)
6MF-3FAc 6M-2FAc -10 6MF-3FAc 6M-2FAc
160
-20

120
E drop (%)

-30
E (GPa)

(b)

80 -40

-50
40
-60

0 -70
0 2 4 6 8 10 0 2 4 6 8 10
Number of cycles Number of cycles
Fig. 8. (a) Absolute elastic modulus and (b) its percentual change for the Al2O3-MgO castables as a function of the thermal shock cycles carried out at 1025 °C (ΔT
~1000 °C). The samples were prefired at 1500 °C for 5 h before testing.

15460
A.P. Luz et al. Ceramics International 44 (2018) 15453–15463

Fig. 9. dL/L0 (%) profile of the Al2O3-MgO castables

during (a) the refractoriness under load tests up to 1600 °C
and (b) creep measurements at 1450 °C for 24 h. All sam-
ples were fired at 1550 °C for 5 h (RUL) or 24 h (creep)
before testing and the applied load was 0.2 MPa. Creep
rate values obtained at 1450 °C after 10 h and 20 h are
highlighted in (b).

Pre-fired 600°C/5h SiO2-based liquid at the boundary surface of these magnesia grains and
favor their displacement when submitted to compressive stresses.
Nevertheless, all compositions presented T0.5 > 1600 °C (maximum
recommended working temperature, Fig. 9a), indicating that the pre-
pared castables showed high refractoriness. The samples containing
6 wt% of MgO also presented higher creep rate levels (as pointed out in
Fig. 9b) at 1450 °C, highlighting that the impurities of magnesia tend to
favor the generation of some amount of liquid phase in the micro-
structure of these refractories at high temperatures even after pre-firing
the samples at 1550 °C for 24 h.
High corrosion resistance to basic slag is usually a feature attributed
to in situ spinel (MgAl2O4)-containing refractory castables. Thus, cor-
rosion cup-tests were carried out at 1500 °C for 2 h considering the
1MF-0.3AAc 6MF-3FAc interaction of a basic synthetic slag with calcined (600 °C/5 h) or pre-
1 cm fired (1500 °C/5 h) samples. After these tests, some cracks could be
mainly observed on the surface of the 6 MF-3FAc, 6 M-2FAc and 1MF-
Fig. 10. Images of the Al2O3-MgO castable samples after the corrosion cup-test 5M-2FAc corroded specimens (Fig. 10), which highlighted that the in-
at 1500 °C for 2 h. teraction of the molten slag with the refractory components resulted in
expansive transformations and the novel phases formed in these ma-
Table 4 terials were not well dimensionally accommodated. Additionally, the
Calculated infiltrated area of the castable samples after the corrosion cup-tests calcined samples (600 °C/5 h) were submitted to their first thermal
at 1500 °C/2 h. treatment up to 1500 °C during the corrosion tests (by contact with the
Compositions Infiltrated area (%)
molten slag). Consequently, such materials expanded due to the spinel
formation (as previously pointed out in the assisted sinterability mea-
Pre-fired at 600 °C/5 h Pre-fired 1500 °C/5 h surements, Fig. 5), as well as the reaction between slag-refractory.
According to the thermodynamic calculations, the first contact be-
1MF-0.3AAc 17.6 ± 0.7 25.6 ± 0.6
1MF-5M-2FAc 30.6 ± 1.2 30.4 ± 1.0
tween slag-castables should mainly lead to a partial spinel dissolution
6MF-3FAc 32.6 ± 1.6 31.2 ± 1.1 (contained in the samples’ matrix) and the formation of calcium hex-
6M-2FAc 31.3 ± 1.0 38.1 ± 1.8 aluminate (CA6) phase due to the high CaO content in the molten
synthetic slag. It was predicted that 22.1 wt% of CA6 should be gener-
ated after the interaction of alumina from the 1MF-0.3AAc matrix
Besides the lower thermal shock resistance, another important as- fraction with calcium oxide (from the slag) at 1500 °C, whereas basi-
pect is that a small liquid phase content might be present in the 6MF- cally no fine Al2O3 would be available in the compositions containing
3FAc samples at high temperature, as highlighted in RUL and creep 6 wt% of MgO (and, consequently, no calcium hexaluminate generation
results (Fig. 9). Table 1 pointed out that SiO2 and other impurities are should take place) due to the consumption of this oxide for the in situ
contained in the MgO fumes, which may induce the formation of CaO-

15461
A.P. Luz et al. Ceramics International 44 (2018) 15453–15463

spinel formation. Besides the attack of the more reactive components Center 920 of the German Scientific Foundation (8311/54753/0835/
(matrix fraction), the calculations indicated that an additional 5.2 wt% 04040299/61306) for supporting this work. The authors are also
of CA6 (resulting in a total content of 27.3 wt% in the overall compo- grateful to RHI-Magnesita and Almatis for supplying the raw materials
sition) might be formed in 1MF-0.3AAc composition, whereas 12.9 wt% used in this study.
of this oxide are expected to be generated in 1MF-5 M-2FAc, 6 MF-3FAc
and 6 M-2FAc at 1500 °C, due to the further corrosion of the coarse References
tabular alumina grains (aggregates). It is important to highlight that
such calcium hexaluminate contents obtained from the thermodynamic [1] H. Peng, B. Myhre, Microsilica-gel bonded refractory castables with improved set-
simulations represent an equilibrium condition and the influence of behaviour and mechanical properties, Refract. World. 7 (2015) 69–72.
[2] H. Peng, B. Myhre, Recent progress in microsilica-gel bonded no-cement castable
physical aspects of the samples (i.e., their porosity) in the corrosion technology, in: Proceedings of ALAFAR 2016, San Carlos de Bariloche, Argentina,
process was not considered. 2016, pp. 1–10.
As the compositions containing 6 wt% of MgO showed higher por- [3] A.P. Luz, S. Lopes, V.C. Pandolfelli, D.T. Gomes, Advanced refractory castables
containing novel alumina and silica-based binders, in: Proceedings of ALAFAR
osity level and lower mechanical resistance than 1MF-0.3AAc (Fig. 4), 2016, San Carlos de Bariloche, Argentina 2016 pp. 1–11.
the more intense slag infiltration and, consequently, more significant [4] A.P. Luz, J. Gagliarde, C.G. Aneziris, V.C. Pandolfelli, B4C mineralizing role in the
calcium hexaluminate formation resulted in the cracking of the cor- mullite generation in Al2O3-SiO2 refractory castables, Ceram. Int. 43 (2017)
12167–12178.
roded specimens (Fig. 10). The weak cohesion between coarse and fine
[5] T. Santos, A.P. Luz, C. Pagliosa, V.C. Pandolfelli, Mg(OH)2 nucleation and growth
components of the 1MF-5M-2FAc, 6 MF-3FAc and 6 M-2FAc refractories parameters applicable for the development of MgO-based refractory castables, J.
could also be observed during the cutting step of these corroded spe- Am. Ceram. Soc. 99 (2016) 461–469, http://dx.doi.org/10.1111/jace.14019.
[6] T.M. Souza, A.P. Luz, M.A.L. Braulio, C. Pagliosa, V.C. Pandolfelli, Acetic acid role
cimens, where irregular cross section surface and coarse grains de-
on magnesia hydration for cement-free refractory castables, J. Am. Ceram. Soc. 97
tachment were obtained. (2014) 1233–1241, http://dx.doi.org/10.1111/jace.12873.
When calculating the infiltrated area of the calcined (600 °C/5 h) or [7] C. Sonat, N.T. Dung, C. Unluer, Performance and microstructural development of
pre-fired (1500 °C/5 h) corroded samples, 1MF-5 M-2FAc and 6MF- MgO-SiO2 binders under different curing conditions, Constr. Build. Mater. 154
(2017) 945–955, http://dx.doi.org/10.1016/j.conbuildmat.2017.08.020.
3FAc presented similar profiles (slag infiltration ~30.4–32.6%) in both [8] N.T. Dung, C. Unluer, Influence of nucleation seeding on the performance of car-
evaluated conditions (Table 4). On the other hand, the pre-fired spe- bonated MgO formulations, Cem. Concr. Compos. 83 (2017) 1–9, http://dx.doi.org/
cimens of 1MF-0.3AAc and 6 M-2FAc compositions showed higher in- 10.1016/j.cemconcomp.2017.07.005.
[9] T.M. Souza, M.A.L. Braulio, A.P. Luz, P. Bonadia, V.C. Pandolfelli, Systemic analysis
filtration when compared to the calcined ones. In general, the best of MgO hydration effects on alumina – magnesia refractory castables, Ceram. Int. 38
performance was obtained for the 1MF-0.3AAc composition (infiltra- (2012) 3969–3976, http://dx.doi.org/10.1016/j.ceramint.2012.01.051.
tion ~17.6 or 25.6%, Table 4), which is associated with its lower [10] T. dos Santos Jr., J. Santos, A.P. Luz, C. Pagliosa, V.C. Pandolfelli, Kinetic control of
MgO hydration in refractory castables by using carboxylic acids, J. Eur. Ceram. Soc.
porosity, higher mechanical strength and less flaws and defects gener- 38 (4) (2018) 2152–2163.
ated in its microstructure due to the in situ expansive reactions that [11] K.P. Matabola, E.M. van der Merwe, C.A. Strydom, F.J.W. Labuschagne, The in-
took place at high temperature in this material. fluence of hydrating agents on the hydration of industrial magnesium oxide, J.
Chem. Technol. Biotechnol. 85 (2010) 1569–1574.
[12] D. Filippou, N. Katiforis, N. Papassiopi, K. Adam, On the kinetics of magnesia hy-
4. Conclusions dration in magnesium acetate solutions, J. Chem. Technol. Biotechnol. 74 (1999)
322–328.
[13] E.M. van der Merwe, C.A. Strydom, Hydration of medium reactive magnesium
This work pointed out the likelihood of applying MgO fumes as a
oxide using hydration agents, J. Therm. Anal. Calorim. 8 (2006) 467–471.
potential binder for in situ spinel containing refractory castables. Based [14] C. Yan, D. Xue, L. Zou, Z. Yan, Preparation of magnesium hydroxide nanoflowers, J.
on the experimental results, the adjustment of the magnesia and car- Cryst. Growth 282 (2005) 448–454.
boxylic acids source and content is a very important aspect, as im- [15] X.D. Peng, M.A. Barteau, Dehydration of carboxylic acids on the MgO(100) surface,
Catal. Lett. 7 (1991) 395–402, http://dx.doi.org/10.1007/BF00764930.
proved green mechanical strength and crack-free samples could only be [16] X.D. Peng, M.A. Barteau, Acid-base properties of model magnesium oxide surfaces,
obtained when adding up to 6 wt% of MgO fumes to the refractories and Langmuir 7 (1991) 1426–1431.
processing them with aqueous solutions with 3 wt% of formic acid. This [17] M.A.L. Braulio, M. Rigaud, A. Buhr, C. Parr, V.C. Pandolfelli, Spinel-containing
alumina-based refractory castables, Ceram. Int. 37 (2011) 1705–1724, http://dx.
very fine magnesia (fumes) presented a high specific surface area and doi.org/10.1016/j.ceramint.2011.03.049.
greater reactivity when compared to a dead-burnt MgO, which resulted [18] A.P. Luz, M.A.L. Braulio, A.G. Tomba Martinez, V.C. Pandolfelli, Slag attack eva-
in vibratable castables with similar green mechanical strength as those luation of in situ spinel-containing refractory castables via experimental tests and
thermodynamic simulations, Ceram. Int. 38 (2012) 1497–1505, http://dx.doi.org/
bonded with 4.0 wt% of calcium aluminate cement. Such results high- 10.1016/j.ceramint.2011.09.033.
lighted the great binder potential of this very fine magnesia source. [19] A.P. Luz, M.A.L. Braulio, V.C. Pandolfelli, Refractory Castable Engineering, 1st ed.,
Despite the enhanced green mechanical strength, alumina-based Goller Verlag, (Germany), Baden-Baden, 2015, pp. 340–420.
[20] A. Nagasoe, S.I. Tsurumoto, A. Kitamura, Refractory characteristics of spinels with
castables containing 6 wt% of MgO (fumes, dead-burnt or their blend) various MgO contents, Taikabutsu Overseas 11 (1991) 20–28.
showed low mechanical strength at intermediate temperatures [due to [21] J. Mori, M. Yoshimura, Y. Oguchi, T. Kawakami, Effect of slag composition on wear
Mg(OH)2 decomposition] and high linear expansion, as a consequence of alumina-spinel castable for steel ladle, Taikabutsu Overseas 12 (1992) 40–45.
[22] K. Fujii, I. Furusato, I. Takita, Composition of spinel clinker for teeming ladle
of the in situ spinel phase formation above 1200 °C. Therefore, better
casting materials, Taikabutsu Overseas 12 (1992) 4–9.
densification, improved HMOR, thermal shock resistance and corrosion [23] B. Nagai, O. Matsumoto, T. Isobe, Y. Nishiumi, Wear mechanism of castable for steel
behavior were obtained for the castables prepared with only 1 wt% of ladle by slag, Taikabutsu Overseas 12 (1992) 15–20.
MgO fumes. [24] S. Zhang, W.E. Lee, Spinel-Containing Refractories, in: C.A. (org. Schacht (Ed.),
Refract. Handb., 1st ed., Marcel Dekker, New York, 2004, pp. 215–256.
In order to overcome the sintering/densification and dimensional [25] R.G. Pileggi, V.C. Pandolfelli, A.E. Paiva, J. Gallo, Novel rheometer for refractory
stability issues of the compositions containing high MgO content, an castables, Am. Ceram. Soc. Bull. 79 (2000) 54–58.
alternative could be to add sintering additives (calcium or magnesium [26] G. Pickett, Equations for computing elastic constants from flexural and torsional
resonant frequencies of vibration of prisms and cylinders, in: Proceedings of Am.
carbonates, boron-based compounds, silica) to these refractories and Soc. Test. Mater, 1945, pp. 846–865.
favor the generation of solid or liquid compounds in these systems at [27] A.P. Luz, T. Santos Jr., J. Medeiros, V.C. Pandolfelli, Thermal shock damage eva-
high temperatures. This aspect will be investigated in more depth in a luation of refractory castables via hot elastic modulus measurements, Ceram. Int. 39
(2013) 6189–6197, http://dx.doi.org/10.1016/j.ceramint.2013.01.038.
forthcoming paper by authors. [28] M.A.L. Braulio, A.G.T. Martinez, A.P. Luz, C. Liebske, V.C. Pandolfelli, Basic slag
attack of spinel-containing refractory castables, Ceram. Int. 37 (2011) 1935–1945,
Acknowledgements http://dx.doi.org/10.1016/j.ceramint.2011.02.007.
[29] I. Markov, Influence of surface active species on kinetics of epitaxial nucleation and
growth, Mater. Chem. Phys. 49 (1997) 93–104.
The authors would like to thank FIRE (Federation for International [30] S. Sugiyama, K. Sato, S. Yamasaki, K. Kawasgiro, H. Hayashi, Ketones from car-
Refractory Research and Education) and the Collaborative Research boxylic acids over supported magnesium oxide related catalysts, Catal. Lett. 14

15462
A.P. Luz et al. Ceramics International 44 (2018) 15453–15463

(1992) 127–133. expansion behavior during synthesis of spinel, in: Proceedings UNITECR 2015,
[31] G. Mekhemer, S. Halawy, M. Mohamed, M. Zaki, Ketonization of acetic acid vapour Kyoto, Japan, 1995, pp. 1312–1319.
over polycrystaline magnesia: in situ Fourier infrared spectroscopy and kinetic [36] Z. Nakagawa, Expansion behavior of powder compacts during spinel formation,
studies, J. Catal. 230 (2005) 109–122. Mass Charg. Transp. Ceram. (1996) 283–294.
[32] D.R. Lide, Dissociation Constants Of Organic Acids And Bases, DRC Handb, CRC [37] E.Y. Sako, M.A.L. Braulio, E. Zinngrebe, S.R. Van der Laan, V.C. Pandolfelli,
Press, 2005, 〈http://www.hbcpnet.base.com〉. Fundamentals and applications on in situ spinel formation mechanism in Al2O3-
[33] T.M. Souza, A.P. Luz, M.A.M. Brito, V.C. Pandolfelli, In situ elastic modulus eva- MgO refractory castables, Ceram. Int. 38 (2012) 2243–2251.
luation of Al2O3–MgO refractory castables, Ceram. Int. 40 (2014) 1699–1707, [38] M.A.L. Braulio, D.H. Milanez, E.Y. Sako, L.R. Bittencourt, V.C. Pandolfelli,
http://dx.doi.org/10.1016/j.ceramint.2013.07.066. Expansion behavior of cement bonded alumina-magnesia refractory castables, Am.
[34] M.A.L. Braulio, L.R.M. Bittencourt, V.C. Pandolfelli, Selection of binders for in situ Ceram. Soc. Bull. 86 (2007) 9201–9206.
spinel refractory castables, J. Eur. Ceram. Soc. 29 (2009) 2727–2735, http://dx.doi. [39] T.M. Souza, A.P. Luz, V.C. Pandolfelli, Magnesium fluoride role on alumina-mag-
org/10.1016/j.jeurceramsoc.2009.03.024. nesia cement-bonded castables, Ceram. Int. 40 (2014) 14947–14956, http://dx.doi.
[35] Z. Nakagawa, N. Enomoto, I.S. Yi, K. Asano, Effect of curundum/periclase sizes on org/10.1016/j.ceramint.2014.06.092.

15463