Interfacial Reactions, Thermodynamics and Kinetics in Doped
Aluminosilicate Ceramics/Liquid Al-Alloy Contacting Systems
M. Oliveira1, S. Agathopoulos1, J. Lino2, J.M.F. Ferreira1
1
Ceramics & Glass Engineering Dept., University of Aveiro, CICECO, 3810-193 Aveiro, Portugal
2
Dept. of Mechanical Engineering and Industrial Management, Faculty of Engineering, University
of Porto, 4200-465 Porto, Portugal
Keywords: Interfacial reactions, alumino-silicates, MgO, CaO, BaO, Al-alloys.
Abstract. Doping of alumino-silicate crucibles with MgO, CaO, or BaO aimed for improving their
performance towards Al-alloy melts in aluminium foundry industry. Depending on the particular
oxide, doping considerably affects microstructure and crystalline state of resulting ceramics, as well
as interfacial reactions with Al-alloys at elevated temperatures. Analysis of reaction features, in
terms of reaction zone structure, mechanism and kinetics, for each particular doping oxide, indicates
that reaction intensity and kinetics follow the order MgO>BaO>CaO.
Introduction
The aluminium industry often uses alumina-made crucibles. Crucibles of pure alumina (Al2O3)
anticipate long lifetime and preservation of purity of the melt, but their production is rather
expensive due to high sintering temperatures. Addition of silica (SiO2) considerably reduces
sintering temperature and lowers the thermal expansion coefficient (CTE). Aluminosilicate
crucibles are currently used in the aluminum industry [1, 2].
This work aims at providing an insight of the performance of aluminosilicate ceramics doped with
alkaline earth oxides, such as MgO, CaO and BaO, towards molten Al-7wt%Si foundry alloy. In
particular, the article focuses at the factors which control the interfacial ceramic/liquid-metal
reactions, in terms of reaction products, thermodynamics and reaction kinetics.
Materials and Experimental Procedure
Table 1 shows the chemical composition of the metals used in this study (named as “130.0” and
“356.0”). Table 2 summarizes the characteristics of the investigated ceramics.
Ceramic crucibles were obtained by slip casting using silica, as natural diatomite (Soc. Portuguesa
de Diatomite, Portugal) and α-Al2O3 powder (Grade A16SG, Alcoa Chemicals, USA). MgO, CaO
and BaO additives were introduced as 4MgCO3.Mg(OH)2.5H2O (Merck, Germany), CaCO3 (M1,
Mineraria Salcilese, Italy) and BaCO3 (Grade 99+%, Aldrich, Austria), respectively. Sintering was
carried out in air at 1400ºC for 2 h [3, 4]. Porosity and pore size distribution were measured by
mercury porosimetry (PoreSizer 9320, Micromeritics, USA). Crystalline phases were identified by
X-ray diffraction (XRD, Rigaku Geigerflex D/Mac, C Series, Cu Kα radiation, Japan).
Al-metal blocks were melted inside the crucibles under dynamic vacuum (pO2<2x10-3 Pa) at
several temperatures between 750º and 1050oC and soaking times between 30 minutes and 24 hours
(heating rate 10º/min). Apart slow cooling, quenching experiments with ceramic-crucible/Al-alloy
couples sealed in borosilicate glass capsules under vacuum (1 Pa), were also carried out for up to
1250oC for 4 hours. Polished cross-sections of ceramic/metal interfaces were examined by scanning
electron microscopy (SEM, Hitachi S-4100, Japan) under secondary and back-scattering electrons
mode and energy dispersion spectroscopy (EDS) under point analysis mode.
Table 1 Content of elements (except Al) (wt%) in the aluminium, “130.0”, and the Al-7wt%Si
alloy, “356.0” (given by the manufacturer, according to Aluminium Association, “AA”).
Metal
Si
Fe
Cu
Zn
Mg
Mn
Ti
Cr
V
“130.0”
0.1
0.05
← < 0.025 →
−
−
−
“356.0”
6.5-7.5
0.6
0.25
0.35
0.25-0.45 0.35
0.25
−
−
Table 2 Characteristics of the investigated ceramics. The volume and molar Al2O3/SiO2 ratios
were 1/1 and 51.55/48.45, respectively.
I.D.
Al2O3⋅SiO2
5%MgO
5%CaO
5%BaO
10%BaO
15%BaO
Additive (%)
vol (mol)
0 (0)
5 (11.5)
5 (8.1)
5 (4.9)
10 (9.9)
15 (14.9)
Element ratio (at%)
O / Al / Si / Me
62.47/25.53/12.00/0.00
61.65/24.02/11.30/3.03
61.88/24.50/11.52/2.10
62.10/24.93/11.73/1.25
61.77/24.25/11.40/2.57
61.42/23.54/11.07/3.97
Porosity
%
Pore size (µm)
50
0.45, 75
43
1.5
1
N/A
6
75
18
2
25
1
Phases
Al2O3,SiO2,MU
Al2O3,MU,Cordierite,*
Al2O3,MU,Anorthite
Al2O3,SiO2,MU,Glass
Al2O3,Celsian
Al2O3,Celsian
Me: Metal of dope oxide, MU: Mullite, N/A: not applicable.
*: In MgO ceramics, traces of Mg7Al18Si3O40 were identified.
Wetting and Reaction Zone (RZ)
Poor wettability (contact angle >90o) was obtained for temperatures up to 900º-950oC, due to the
formation of superficial Al-oxide layer on the surface of the liquid aluminum [5-7], whose presence
depended on temperature rather than holding time (until 24 hours). In industry, stirring and agitation
are employed to disrupt this oxide layer [8], whereas, on a laboratory scale, prolonged heating at
elevated temperatures under high vacuum results in Al2O-gas, whose formation weakens the oxide
film [6-10]. However, Al should have been diffused itself through the superficial Al-oxide barrier
towards the ceramic, as proven by the blackening of the crucibles near the ceramic/metal interface.
At temperatures >950oC, the Al-oxide layer was evidently disrupted because the melt wetted the
crucible walls (contact angle <90o). Strong reaction occurred, identified by a black reaction zone
(RZ) formed always in the ceramic but never in the metal. RZ was strongly adhered to the metal
phase that it was often spontaneously separated from the mother ceramic by cracks after cooling.
Evidently, cracks and separation were possible due to the mismatch of CTEs of the contacting
zones, indicating that RZ largely behaved alike the metal phase rather than the ceramic.
Reactions with Pure Al, “130.0”
The reaction between Al2O3-SiO2 and Al resulted in a wide RZ (probably due to the large porosity
of the ceramics, Table 2), which comprised two zones (Fig. 1a). A wide zone, adjacent to the intact
ceramic, was infiltrated by Al (Al/O>2/3) and some Si. Between this wide zone and the metal, there
was a tiny zone (some tens of microns), depleted of Si. The reduced Si was detected in the metal
(maximum Si ~10wt% close to the RZ).
Doping of ceramics with BaO considerably changed the microstructure of the RZ (Fig. 1b),
which was much thinner, probably due to the lower pore volume fraction of BaO-doped ceramics.
RZ comprised an Al2O3 skeleton infiltrated by Al, while Si and Ba were detected neither in RZ nor
dissolved in bulk metal, but exclusively in some discrete particles, either adhered to the ceramic
(i.e. the RZ, inset of Fig. 1b) or dispersed in the metal phase.
Fig.1: Cross section of ceramic/metal interfaces: (a) Al2O3-SiO2/“130.0”, 1050ºC/30 minutes. (b)
5%BaO/“130.0”, 1050ºC/4 h. (M: metal phase, RZ: reaction zone, C: intact ceramic).
Reactions with Al-7%Si Alloy, “356.0”
Figure 2 shows three characteristic microstructures of interfaces between MgO-, CaO- and BaOdoped ceramics and “356.0” Al-alloy. Evidently, the doping-oxides greatly affected the
microstructure of the RZ. Apparently, the thickness of the RZ follows the general trend
MgO >> BaO > CaO
(1)
Increasing temperature generally widens the RZ, without however altering the sequence (1), at
least within the investigated range of temperature and time. In the following, one shall shed some
light on expression (1) from different view points.
Fig.2: Characteristic microstructures of reaction zones formed between “356.0” alloy and the investigated
ceramics (900-1000oC/4 h): (a) 5%MgO. (b) 5%CaO (the β-Al2CaSi2 precipitates are visible only in
back-scattering mode, inset). (c) 5%BaO. (M: metal phase, RZ: reaction zone, C: intact ceramic).
Porosity, Pore Size, and Crystalline State. At first sight, order (1) may be plausibly attributed
to the different pore volume fraction of ceramics (Fig. 3) except the case of 10%BaO (Fig. 3b).
Apparently, the glassy phase in 5%BaO (Table 2), presumably associated to BaO and SiO2, has
likely two opposing effects in the RZ formation. Being located at grain boundaries, it resulted in a
dense ceramic structure but it makes ceramic vulnerable to chemical attacks comparing to the
corresponding crystalline phases [11, 12]. Therefore, in the case of 5%BaO, a presumably
intergranular reaction resulted in a thick RZ, a non-planar RZ onset front towards the intact ceramic
and enhanced porosity of RZ compared with the porosity of intact ceramic.
20
200
10
800
>1000ºC
20
2000
10
RZ thickness ( µm)
400
3000
750-800ºC
900-950ºC
1050ºC
porosity
Porosity (%)
40
30
30
4000
RZ thickness ( µm)
600
50
900ºC
1050ºC
porosity
Porosity (%)
RZ thickness ( µm)
800
600
1000ºC
400
900ºC
800oC
200
1000
<800ºC
0
0
0
5% CaO
5% BaO
5% MgO
0
0
5% BaO
10% BaO
15% BaO
(a)
(b)
Fig.3: Correlation of RZ thickness for several temperatures (4
hours) with the initial porosity of ceramics (Table 2).
0
4
8
12
16
20
24
Time (h)
Fig.4: Isotherms for the RZ thickness
in 5%BaO/ 356.0 system (24 h).
The absence of RZ for 15%BaO at low temperatures (Fig. 3b) should be due to the poor wetting
regime, whereby the liquid alloy, covered by the superficial Al-oxide, could not infiltrate inside the
narrow pores (1 µm, Table 2).
Consequently, porosity favours infiltration and widens the RZ as long as good wetting occurs.
The RZs developed in dense ceramics should result from reactions apparently taken place by solid
diffusion. In the latter case crystalline state is of crucial importance. Hence, considerable alterations
in the microstructure of the initial ceramic (i.e. in the RZ) took place for 5%MgO, moderate for
5%BaO and almost imperceptible for 5%CaO (Fig. 2). Evidently, the aforementioned analysis is
consistent with expression (1).
Reaction Products and Mechanism. The reaction products, determined by XRD, SEM and EDS
analyses, were as follows. In the case of doping with MgO, RZs consisted of Al2O3 skeleton
infiltrated by liquid Al, Mg and Si (Fig. 2a). The bulk metal was homogeneous with an average
composition 68Al-30Si-2Mg (at%) at the highest tested temperature. The RZs of CaO-doped
ceramics comprised Al2O3 skeleton infiltrated by Al, whereas neither Ca nor Si was detected in this
zone. In the metal phase, precipitous shaped particles, assigned to the equilibrium phase β-Al2CaSi2
[13], were adhered at the interface but never forming a continuous layer (Fig. 2b). No Ca was
detected in bulk metal. Similarly, the RZs in the BaO-doped ceramics (Fig. 2c) comprised Al2O3
infiltrated by liquid Al, but depleted of Ba and Si. The reduced Ba and Si were seemingly
concentrated exclusively in precipitates, found either dispersed in the bulk metal and assigned to the
equilibrium phase α-Al2BaSi2 [13], or adhered to the interface whose composition was 30Al-12Ba58Si (at%), which must be considered as metastable one since it is not anticipated by the Al-Ba-Si
phase diagram [13]. The formation of α-Al2BaSi2 precipitates was also confirmed in quenching
experiments from 1250oC, although the ternary Al-Ba-Si phase diagram anticipates precipitation of
α-Al2BaSi2 between 1040º-647oC [13]. Apparently, if the formation of α-Al2BaSi2 does not result
from insufficiently rapid cooling during quenching, then a miscibility gap at the liquid phase might
be conjectured, at least for the investigated system under vacuum.
Accordingly, the reaction mechanism can be represented by the following chemical equation:
Oxide + Al → Me + Al2O3
(2)
Table 3 compares the driving force, in terms of chemical potentials (i.e. ∆G) at the interface, for
each particular case [14]. Obviously, within the investigated temperature range, any reaction with
the alkaline earth oxides is thermodynamically non favorable. Therefore, the observed chemical
reactions should occur because of SiO2 incorporated in the alkaline earth oxide associated phases
(Table 2), resulting in negative ∆G (Table 3). If the reaction of the glassy phase in 5%BaO (Table
2) is considered as a blend of celsian and BaO, then the ∆G of the reaction with 5%BaO would be
slightly higher than with celsian and the order of ∆G would also agree with expression (1).
Table 3. Gibbs free energy (in kJ/mol Al) calculated for several optional cases (see Table 2) of
eq.(2) at two different temperatures [14]. (*: Chemical equation yields also Al2O3).
Oxide
Me
CaO
MgO
BaO
SiO2
Mg7Al18Si3O40 *
CaAl2Si2O8 *
Mg2Al4Si5O18 *
BaAl2Si2O8 *
Al6Si2O13 *
Ca
Mg
Ba
Si
Mg, Si
Ca, Si
Mg, Si
Ba, Si
Si
560oC
118.84
64.04
4.59
−133.71
−27.23
−83.20
−100.76
−106.05
−133.71
∆G (kJ/mol Al)
1125oC
115.72
50.66
5.25
−118.58
−27.45
−71.72
−90.38
−93.82
−118.58
Kinetics. Processes, at which the same diffusion mechanism is the rate-limiting step of kinetics,
feature parabolic growth of the RZ over time [15], while in porous ceramics, the RZ growth is
linearly dependent on time [10]. The time evolution of the RZ thickness for the 5%BaO/“356.0”
system is depicted in Fig. 4. Apparently, porosity resulted in linear growth for short times (≤ 4 h),
while the exponential trend at prolonged experiments would be due to the effect of the precipitates
in the apparent reaction kinetics, acting as a diffusion barrier between ceramic and melt.
In the case of both CaO and BaO ceramics, after reduction by liquid Al, the reduced Ca and Ba,
respectively, seem to drift the reduced Si from the RZ towards the metal, forming the precipitates.
Evidently, the formation of the equilibrium phases, specifically β-Al2CaSi2 and α-Al2BaSi2 [13],
lead both CaO and BaO systems to equilibrium regime. Meanwhile, in the case of BaO ceramics,
the metastable precipitates adhered at the interface can be considered as precursors of the
equilibrium ones. This was not the case for MgO-ceramics because the Al-Mg-Si phase diagram
anticipates no ternary equilibrium phases [8]. Thus Mg and Si were dissolved in Al in the RZ and in
the metal, and no phase separation occurred in the metal phase.
Consequently, considering that the attached precipitates hinder interfacial reactions, the
aforementioned analysis predicts strong protection for CaO ceramics, moderate for BaO and
completely vulnerable interface for MgO. Hence, kinetics also supports expression (1). Moreover,
the exponential trend of the isotherms for 5%BaO ceramics in Fig. 4 might deal with the coverage
fraction of the interface by the metastable precipitates and its dependence on temperature and time.
Acknowledgements
This work was supported by the Agency of Innovation S.A., Portugal, under Project P051-P31B05/97-DISASVAL, and the Portuguese Foundation for Science and Technology under Grants
PRAXIS-XXI BD/18605/98 and BPD/1619/00.
References
[1] A.M. Wynn: Br. Ceram. Trans. Vol. 91 (1992), p. 153.
[2] C.R.V. Cruz: Refractários para Fundição, (Refractories for Foundry). Culture Commission of
the Chamber of Engineers, Lisbon, Portugal, 1984.
[3] M. Oliveira, S. Agathopoulos and J.M.F. Ferreira: Acta Materialia Vol. 50 (2002), p. 1441.
[4] M. Oliveira, S. Agathopoulos and J.M.F. Ferreira: J. Mater. Res. Vol. 17 (2002), p. 641.
[5] M.I. Pech-Canul, R.N. Katz, M.M. Makhlouf and S. Pickard: J. Mat. Sci. Vol. 35 (2000), p.
2167.
[6] K. Landry and N. Eustathopoulos: Acta Mater. Vol. 44 (1996), p. 3923.
[7] C. Allaire and P. Desclaux: J. Am. Ceram. Soc. Vol. 74 (1991), p. 2781.
[8] J.R. Davis, Aluminum and Aluminum Alloys, ASM Specialty Book. Materials Park, OH,
USA, 1996.
[9] M. Allahevrdi, S. Afshar and C. Allaire: JOM Vol. 52 (2000), p. 43.
[10] E. Saiz and A.P. Tomsia: J. Am. Ceram. Soc. Vol. 81 (1998), p. 2381.
[11] W.E. Lee and R.E. Moore: J. Am. Ceram. Soc. Vol. 81 (1998) 385.
[12] S. Afshar and C. Allaire: JOM Vol. 50 (1998), p. 30.
[13] G. Petzow and G. Effenberg, “Ternary Alloys: A comprehensive Compendium of Evaluated
Constitutional Data and Phase Diagrams”; pp. 310 and 682. VCH, Weinheim, Germany, 1988.
[14] JANAF Thermochemical Tables, J. Phys. Chem. Ref. Data Vol. 14 (1985), p. 156, 348, 711,
1469, 1673.
[15] F.J.J. van Loo: Prog. Solid St. Chem. Vol. 20 (1990), p. 47.