Thermodynamic Assessment of the Alumina-Sodium Oxide-Silica using the Modified Quasichemical Model
Guillaume LAMBOTTE (PhD Student) and Patrice CHARTRAND (Professor)
Centre for Research in Computational Thermochemistry (CRCT)
École Polytechnique de Montréal
Info: [email protected]
Abstract
Binary Systems
Alumina
Feeder
For the production of aluminum electrolytic cells are used (Fig. 1), in which the
electrolysis of alumina (Al2O3) dissolved in a cryolitic (Na3AlF6) based bath is
done at around 4.2 V and 300 kA.
A critical review of the available data for the binary, ternary and reciprocal subsystems has been performed, and the data are being used in
the on-going optimization process. The presented figures represent the work in progress.
Hood
Crust
Breaker
According to Øye & al. if the chemical erosion of the carbon cathode block
The ordering compositions in the liquid phase are set to the following composition in the different binary subsystems:
Frozen
Ledge
Carbon
Ramming Paste
Anode
• Al2O3-Na2O: we have decided to extend the stability of the associate NaAl4+ from the ternary into the binary system in order to optimize it
while maintaining cation distribution in the reciprocal system.
Side Lining
(Al4C3 formation and dissolution) was the only failure of the cell, then it could
operate around 11 years. However it is never the case in the industry because
failure due to corrosion of the refractory layers is the most common reason for
cell shut down.
Cryolitic Bath
Aluminum
•Al2O3-SiO2: at the mullite (3 Al2O3,2 SiO2) composition XAl2O3=0.6 .
Carbon Cathode
In the cell the refractory lining is mainly composed of aluminosilicate materials.
•Na2O-SiO2: at the orthosilicate composition, Na4SiO4 or XNa2O=0.8 .
Current Collecting Bar
Thermodynamic properties of solids are mainly based on JANAF and Barin thermodynamic tables and the previous optimization of the
Dense Refractory Lining (Al2O3-SiO2)
Insulation Refractory Lining
system by P. Wu (PhD Thesis, CRCT, 1993)
• Al2O3-Na2O
1300
1100
(source: http://www.ceramics-research.com)
corrosion of aluminosilicate refractory materials by the cryolitic bath is the Al,
Na, Si // F, O reciprocal system (Fig. 3). Interactions with carbon, sodium and
aluminum are also to be taken into account.
The NaF-AlF3 join represents the cryolitic bath, and the aluminosilicate
Na2Al12O19
refractory lining has its composition along the Al2O3-SiO2 join.
Al2Si2O4F2
(fluor-topaz)
(β’’-alumina)
1400
The special features of this system are:
SiF4
(malladrite)
NaF
(cryolite)
AlF3
Na3AlF6 Na5Al3F14 (chiolite)
0.2
0.4
F
O
Na
,
Si
O
F
Na
SiO2
Na
…; reciprocal system AlF3-NaF-Al2O3-Na2O
Al
F
O
Na
0.8
0.6
0.3
0.8
0.7
0.6
0.8
1.0
Reciprocal System
performed (P. Chartrand, 2002). The integration of this previous work to
our optimization of the oxide subsystem and the further optimization of
the reciprocal system should be done without major changes.
Polythermal Projection 600°C - 1100°C
1725°C and sup.
(Na2O)3
0.6
0.5
0.4
0.3
0.2
0.1
Al2O3
0.2
2053°C
0.9
Weight fraction
0.4
Mole fraction Na2O
0.3
0.4
0.5
0.6
0.7
0.8
Al2O3
…
0.3
0.4
0.5
0.6
0.7
0.8
0.2
O
O
0.2
0.1
Si
0.0
0.1
…; binary system, Na2O-SiO2
1.0
0.2
Na
0.8
0.3
F
F
0.6
Optimization of the NaF-AlF3-Al2O3 subsystem has already been
2000°C
NaAl9O14
Na
0.4
1890°C Al6Si2O13
1410°C
Na2Al12O19
, NaF
-8
-10
1450 to 1475°C
0.2
…
Zaitsev 1999
900°C
1000°C
1100°C
1150°C
1200°C
1300°C
1400°C
Rego 1985
1300°C
1400°C
Yamaguchi 1983
1100°C
1300°C
1500°C
1200 to 1225°C
0.5
1088°C
1015°C
Na6Si2O7 1122°C
1025°C
Na4SiO4 1083°C
0.4
Na2SiO3
1585°C
Al
0.2
950 to 975°C
1868°C
AlNaO2
O
O
0.0
1595°C
789°C
Na6Si8O19 807°C
Na2Si2O5 874°C
837°C
Calculated (1100, 1300 , 1500°C)
-6
Silica Saturation
Tilley 1933 & Schairer 1956
700 to 725°C
1723°C
Na2O
Al
1.0
-4
Mole fraction Na2O
1132°C
• Gibbs energy :
from pure component, Al2O3
1.0
-12
Polythermal projection 700°C - 2050°C
0.7
Na
Al
-90
Ternary System
0.1
O
… reciprocal
0.8
0.4
O
0.6
0.5
Na
Si
0.4
Mole fraction Al2O3
0.9
F
,
0.2
0.8
F
0.8
0.7
Si
Al
0.0
0.6
O
… binary
0.6
0.5
O
1.0
0.4
Al
Si
Mullite saturation
Fan 91 (glass) 1450K
Fan 91 (gel) 1450K
Mole fraction SiO2
0.3
F
,
0.8
-70
-130
0.2
F
Haller 1975
Hammel 1966
Andreev 1964
Moriya 1964
-110
0.1
Al
0.6
-50
600
Strong interactions are observed between the components, both in the solid and liquid states, resulting in strong short range order in the liquid solution:
unary
0.4
-30
800
0.0
second nearest neighbour cationic ordering is modeled using the Modified Quasichemical Model in the Quadruplet Approximation (MQMQA). However in
order to model the complete composition of the system we need to take into account anion interactions (mixing of O2- and F- on the second sublattice).
MQMQA : 2 sublattice model
Na
Si , anion sublattice F
O .
• Components : cation sublattice Al
• Quadruplets :
-1
1200
0.9
Fig. 3: Al, Na, Si // F, O Reciprocal system
0.2
-10
1000
• Strong short range order in the liquid phase
9 In the binary subsystems Na2O-SiO2 and Na2O-Al2O3.
9 Charge compensation effect in the silica-rich region of the liquid slag
(replacement of a tetrahedral Si4+ by a tetrahedra formed by an
(Na-Al)4+ pair in the silicon network).
• Limited solid solubilities.
• Metastable miscibility gaps in the molten oxides (of technological
importance for ceramic industry).
• Stable miscibility gaps in the oxifluoride melts.
Na2SiF6
0.2
0.5
(β-alumina)
0.4
0.0
0.0
Enthalpy of mixing [kJ.mol ]
NaAl9O14
1950 K
-2
JANAF
Kracek 1930
Dan's 1930
Kracek 1939
Morey, Bowen (taken from Kracek 1939)
Willgallis 1964
Williamson 1965
Yoder & Schairer 1970
Meshalkin 2003
1600
Na6Si8O19
Al2O3
NaAlO2
1.0
1850 K
0.6
Mole fraction Al2O3
Na2Si2O5
(sodium aluminate)
0.8
Na2SiO3
Na2O
Al6Si2O13 (mullite)
(liq. ref. state)
1100
Takamori 1973
Galakhov 1976
Averyanov 1989
Takei 2000 (MD)
1800
Na6Si2O7
NaAlSiO4 (nepheline)
0.6
2021 K
Bondar 2005
• Na2O-SiO2
The chemical system corresponding to the simplified problematic of the
NaAlSi3O8 (albite)
0.4
Na4SiO4
SiO2
1500
Mole fraction Al2O3
T [°C]
Na6Si8O19
Na2Si2O5
Na2SiO3
Na6Si2O7
(orthosilicate) Na4SiO4
(disilicate)
(metasilicate)
0.2
Bjorkvall 2001
2150 K
900
0.0
Thermodynamic Modeling
2400 K
1300
900
2200 K
2300 K
1700
1500
Shornikov 2000
Calculated (2200, 2300, 2400 K)
0.8
0.4
the equilibrium calculation tool.
Fig. 2: Post-mortem analysis of
an aluminum electrolysis cell
1.0
Aksay 1973
Klug 1987
Shornikov 2000
0.3
A CALPHAD type modeling forms the backbone of such a prediction tool. Thermodynamic software will provide
Aramaki 1962
Horibe 1967
Staronka 1968
Davis 1972
(liq. ref. state)
Insulation
Layer
1700
Dense Refractory
Lining
Greig & Bowen 1924
Toropov 1951
Shairer 1955
Troemel 1957
Gelsdorf 1958
Aramaki 1959
Welch 1960
1900
T [°C]
a great advantage to the industry, resulting in a better design of cell refractory lining in order to resist corrosion,
have longer lifespan and more stable thermal and mechanical properties.
1900
NaAlO2
Predicting the corrosion reactions and the corrosion products at operating temperatures (around 960ºC) will be
Corroded
Refractory
Lining
Current
Collecting
Bar
2100
JANAF
Schairer 1956
Rolin 1963
Rolin 1965
Weber 1970
ß''-Alumina (Na 2Al12O19)
Cathode
T [°C]
based on an analysis of the phases observed at room temperature after the cell shut down.
• Al2O3-SiO2
2100
Activity SiO2
Side
Lining
log10(Activity Na2O)
Ramming
Paste
Post-mortem analysis (Fig. 2), is so far the best solution to understand the mechanism of corrosion, but is
Mullite (Al 6Si2O13)
Fig. 1: Schematic view of an aluminum electrolysis cell
ß-Alumina (NaAl 9O14)
The corrosion of this lining is due to the penetration of bath and aluminum
through the carbon cathode block. Different corrosion mechanisms are involved
which will lead to the progression of the corrosive media further in the refractory
layer.
• Quadruplets mix randomly with overlapping of pairs and ions (the configurational entropy term take this into account).
• Equilibrium is obtained by Gibbs energy minimization constrained by an elemental mass balance.
0.2
(NaF)6
(AlF3)2
Foster, P.A.Jr. (1975), Journal of the American Ceramic Society 58(7-8).
The first step in the thermodynamic modeling of this reciprocal system is to obtain model parameters for the Al2O3-Na2O-SiO2 system. In the case of the
oxide subsystem we only have one anion mixing on the second sublattice. Second nearest neighbour coordination numbers have to be defined which
allow to reproduce the composition of the maximum short-range ordering in the different subsystems. To model the charge compensation effect we have
introduced a NaAl4+ associate, which will mix on the cation sublattice.
This project was supported by
E.F. Osborn and A. Muan (1964), extracted from Phase Diagram for Ceramists, fig. 1-501, The American Ceramic Society.