You are on page 1of 8

19_02-446B-G.

qxd 8/28/03 3:45 PM Page 631

Phase Equilibria and Thermodynamic Properties of the


Cu2O-CaO-Na2O System in Equilibrium with Copper
PASCAL COURSOL, ARTHUR D. PELTON, and MANUEL ZAMALLOA

The liquidus surfaces of the Cu2O-CaO, Cu2O-Na2O, and Cu2O-CaO-Na2O phase diagrams in equilibrium
with metallic Cu were measured by thermal analysis at compositions varying from approximately 0 to
35 wt pct Na2O and 0 to 15 wt pct CaO. Solubilities in the solid binary terminal solutions were also
measured by wavelength dispersive X-ray spectrometer analysis. Copper oxide activities in binary
liquid slags were determined from the measured oxygen content of the metallic copper equilibrated
with the slags. The ternary system is a simple eutectic system. No ternary compounds were observed.
The Cu2O-CaO binary eutectic was measured at 1140 °C  10 °C at 10  1 wt pct CaO and the
Cu2O-Na2O binary eutectic was measured at 803 °C  15 °C at 28  2 wt pct Na2O. The liquid slag
was thermodynamically modeled with the modified quasi-chemical model, while the solid Cu2O-rich
solution was treated as Henrian ideal. All data from the present work and from the literature (phase
diagrams and activities) for the binary systems were evaluated simultaneously by least-squares
optimization in order to obtain the best model parameters. With only these binary parameters, the
calculated ternary liquidus surface is in very good agreement with the measurements. Finally, using
the model, the liquidus projection of the Cu2O-CaO-Na2O system in equilibrium with Cu was calculated
as well as the oxygen content of the equilibrated Cu as a function of slag composition.

I. INTRODUCTION used. In the thermal analysis experiments, the reference cru-


cible was not used. The entire assembly was enclosed in a
BASIC fluxes are commonly used to reduce the concen- covered SiC crucible and purged with Ar at 6 l/min. The
trations of certain minor elements in blister copper. Despite assembly was vibrated to reduce undercooling by attaching
the proven effectiveness of Na2CO3-CaO fluxes in removing the thermocouple supports to a motor oscillating at 30 Hz.
As and Sb,[1–4] various thermochemical aspects of the slag In general, undercooling was less than 5 °C. Samples were
produced industrially are not fully understood. As described held at 1260 °C for 1 hour before cooling at 4 °C/min.
by Riveros et al.[5], Na2CO3 and Cu2O are not fully miscible. Heating curves were also obtained when appropriate, for
Depending upon the overall composition and the oxygen con- example, to determine eutectic temperatures more accurately.
tent of the copper, the slag formed during fluxing may con- The deviation of the measured liquidus temperatures from
sist of two liquid phases, one rich in Na2CO3 and one rich in the true values is estimated as 10 °C when Cu2O is the
Cu2O. An understanding of the phase equilibria and thermo- primary solid phase. When CaO is the primary solid phase,
dynamics of this system is essential for process simulation the error with this technique is about 20 °C because the
and control. Kuxmann and Kurre[6] studied the solubility of CaO liquidus is very steep. These errors are the combined
CaO in liquid Cu2O in the presence of metallic copper. Oishi effects of undercooling, contamination, thermocouple accu-
et al.[7] have studied the partial pressure of O2 in equilibrium racy, and data treatment.
with liquid copper and CaO-Cu2O liquid solutions at 1300
°C. Thermodynamic modeling of the Cu-CuO-Cu2O-CaO sub-
system was done by Ranniko[8] using the association model. B. Measurement of Cu2O Activity by Slag/Metal
No other literature data were found. In the present article, Equilibration
we report our measurements and thermodynamic modeling Activities of Cu2O in the slag were measured by
of the Cu2O-CaO-Na2O system in equilibrium with copper. equilibration with liquid copper followed by quenching and
analysis of the Cu phase for dissolved oxygen. Since the
II. EXPERIMENTAL activity of oxygen in molten Cu as a function of concentra-
tion is known from the literature,[12,13] the activity of Cu2O
A. Thermal Analysis and Differential Thermal Analysis can then be determined.
The DTA apparatus is shown in Figure 1. Recrystallized Samples of 40 g slag and 5 g Cu were held in cov-
Al2O3 crucibles of 99.8 pct purity (Vesuvius McDanel) were ered Al2O3 crucibles (of the same type as in Figure 1) in a
flowing Ar atmosphere for 3 hours at 1200 °C and were then
quenched in water. The water did not come into contact with
PASCAL COURSOL, Process Scientist, is with the Alcan Research and the slag.
Development Centre, P.O. Box 1250, Jonquiere, PQ G7S 4K8, Canada. The quenched copper was analyzed for oxygen by a com-
ARTHUR D. PELTON, Professor, is with the Centre de Recherche en Calcul bustion technique. Samples of 1.0 g were weighed, heated
Thermochimique, Ecole Polytechnique de Montreal, Montreal, PQ, Canada in a graphite crucible under Ar at 1175 °C for 1 hour, and
H3C 3A7. Contact e-mail: apelton@polymtl.ca MANUEL ZAMALLOA,
Senior Scientist, is with the Noranda Technology Center, Pte. Claire, PQ
cooled under Ar in the furnace. After combustion, the
H9R 1G5, Canada. graphite crucible was not wetted by the copper, hence all
Manuscript submitted September 10, 2002. the metallic copper could be completely separated from the

METALLURGICAL AND MATERIALS TRANSACTIONS B VOLUME 34B, OCTOBER 2003—631


19_02-446B-G.qxd 8/28/03 3:45 PM Page 632

which was obtained by combining the enthalpy of fusion


of 64,200 (J/mol) reported by Mah et al.[10] with the melting
point of 1229 °C reported by Neumann et al.[11] under the
assumption that Cp  0 (such that Sfusion
o
 64,200/(1229
 273.15)  42.739 (J/mol K)).
Oxygen activities in liquid Cu were taken from the critical
evaluation of Sigworth and Elliott.[12] These are very close
to the values given in the compilation of Oishi and Ono.[13]
The following equation for PO2 and aCu as a function of
the oxygen mole fraction XO in liquid Cu was obtained by
using the parameters given by Sigworth and Elliott[12] in the
unified interaction parameter formalism.[14]

10,266
O2 (bar)  Xo exp c a
P1/2  5.44b
T
[4]
X2o
 a7.80  b aXo  b d
24,000
T 2

1
aCu  (1  Xo) exp c a7.80  b X2o d
24,000
[5]
2 T
From the measured values of XO, one can calculate PO2 and
aCu from Eqs. [4] and [5]. Substitution into Eq. [2] then gives
aCu2O relative to pure (metastable) liquid Cu2O as standard state.
Fig. 1—Differential thermal analysis apparatus.
Solubility of CaO in liquid Cu2O
A very similar technique was used to measure the solubil-
crucible. The dissolved oxygen content was then calculated
ity of CaO in molten Cu2O (i.e., the liquidus). The technique
from the weight loss. Because the dissolved oxygen content
and results are described in Section IV–A–1
was high (0.1 to 2.5 wt pct), this method permitted precise
measurements to be made. For each equilibration run, the
average of three analyses was taken. Some samples were C. Solubilities in Solid Solutions
also analyzed with a LECO* oxygen analyzer. Results The mutual solubilities of CaO, Cu2O, and Na2O in the
terminal solid solutions are, at most, a few percent. Slag samples
* LECO is a trademark of LECO Corporation, St. Joseph, MI.
of 40 g were held in covered Al2O3 crucibles under flow-
ing Ar for 1 hour at a temperature where the slag was liquid.
obtained by the two methods were in good agreement. The
They were then cooled to the measurement temperature and
error of the measured oxygen contents is estimated to be 5
held for 3 hours in order to attain equilibrium and to permit
pct of the measured value.
the growth of solid particles large enough for analysis. The
For the reaction,
samples were then quenched in water (avoiding contact between
the slag and the water) and dry polished. The primary particles
2Cu (liquid)  1/2O2(gas)  Cu2O (liquid) [1] were subsequently analyzed with a JEOL* 8900 electron micro-
the equilibrium constant at 1200 °C is JEOL is a trademark of Japan Electron Optics Ltd., Tokyo.

aCu2O scope equipped with a wavelength dispersive X-ray spectro-


K1200 °C  1  119.9 [2] meter. For analysis of Na, Ca, and Cu, the following standards
Po/22 aCu2
were used, respectively, Na2Al2Si6O16, CaMgSi2O6, and Cu2O.
where aCu2O is the activity of Cu2O in the liquid slag at The errors in the measured CaO, Cu2O, and Na2O contents
1200 °C relative to the pure liquid (metastable) Cu2O as are estimated as 10 pct of the measured values as long as
standard state, aCu is the activity of liquid copper, and PO2 Na2O was not the main component of the solid solution (i.e.,
is the equilibrium oxygen potential. The value of K1200  119.9 not in the Na2O-rich terminal solid solution). When Na2O was
in Eq. [2] was calculated from the Gibbs energies of Cu the solvent, results were erratic, probably because of hydration
(liquid), O2 (gas), and Cu2O (solid) from the FactSage or oxidation, and percentage errors are estimated as 50 pct.
5.1 database,[9] and from the following Gibbs energy of
fusion of Cu2O:
D. Starting Materials
Cu2O (solid)  Cu2O (liquid) Metallic Cu of 99.99 pct purity from Noranda Inc. (Pte.
Claire, Quebec) and CaO of 99.95 pct purity from Alfa Aesar
G0fusion  G0Cu2O (liquid)  G0Cu2O (Solid) (Ward Hill, MA) were used. The Cu2O was produced by Alfa
[3]
 64,200  42.739 T (J/mol) Aesar by oxidation of high-purity Cu.

632—VOLUME 34B, OCTOBER 2003 METALLURGICAL AND MATERIALS TRANSACTIONS B


19_02-446B-G.qxd 8/28/03 3:45 PM Page 633

The Na2O was purchased from Alfa Aesar who supplied Cu. A similar large decrease in the solubility of Cu in Cu2O
the following analysis: 87 wt pct Na2O, 12 wt pct Na2O2, and is expected with the addition of Na2O. Indeed, when we cooled
1 wt pct Na2CO3. At the high temperatures of the experiments, Cu2O that was equilibrated with liquid Cu, we observed
it was assumed that all Na2O2 is decomposed to Na2O and significant precipitation of Cu from the oxide. However, if
O2 and all Na2CO3 is decomposed to Na2O and CO2. 3 wt pct CaO or Na2O was added to the Cu2O, the amount
of Cu precipitate observed was negligible.
E. Use of Al2O3 Crucibles The order of magnitude of the Cu2 concentration in
Cu2O in equilibrium with metallic Cu can be calculated by
Preliminary tests with MgO crucibles showed a large considering the equilibrium
penetration of the slags into the crucible walls. With recrys-
tallized Al2O3 crucibles (99.8 pct, Vesuvius McDanel (Beaver Cu (liquid)  CuO (liquid)  Cu2O (liquid) [6]
Falls, PA)), no such penetration was observed. With Cu2O-CaO aCu2O
slags, a dense passivating layer of calcium aluminate rapidly K  [7]
forms on the crucible surface (Figure 2), which was shown by aCuOaCu
energy dispersive X-ray spectroscopy analysis to be of the Since no data for the Gibbs energy of fusion of CuO are
approximate composition Ca3Al2O6. From measurements of available, we have estimated K as being approximately equal
the thickness of this layer (20 m), and assuming that the
to K for the corresponding reaction among Cu (liquid), CuO
layer is Ca3Al2O6 (which is the calcium aluminate that has the
(solid), and Cu2O (solid). At 1200 °C, K  15.39.[9] Setting
highest density and highest CaO content), we calculate, in the aCu  1 and aCu2O  1 in Eq. [7] gives aCu2O  0.065. If
worst-case scenario, that CaO losses in the slag are less than Cu and Cu ions are assumed to form an approximately
0.25 wt pct, which can be neglected. With Cu2O-Na2O slags, ideal (Temkin) mixture, then the ratio of Cu to Cu ions
a passivating layer was also observed, although a successful in the melt is of the order of Cu/Cu  0.065.
analysis of this layer was not obtained. Analysis of the melt
by inductively coupled plasma mass spectroscopy (ICPMS)
after the test showed that the Na2O losses were negligible IV. RESULTS
(within the experimental error of the ICPMS analysis).
The slags were also analyzed for Al2O3 by ICPMS. A. Cu2O-CaO System
Alumina contents of less than 1 wt pct were observed in the 1. Phase diagram
slags after the tests. This indicates that contamination from The phase diagram of the Cu2O-CaO system equilibrated
the alumina crucibles was negligible. with metallic copper is shown in Figure 4. The Cu2O
liquidus points shown on Figure 4 were measured by
thermal analysis (cooling curves) as were the eutectic
III. SOLUBILITY OF Cu0 and Cu2 IN
THE MOLTEN OXIDE temperatures (heating curves). The melting point of Cu2O
in equilibrium with metallic Cu was measured at 1219 °C,
From the Cu-Cu2O phase diagram[11] in Figure 3, it can be which may be compared with the value of 1223 °C [11]
seen that the solubility of Cu in pure molten Cu2O is high. shown in Figure 3.
However, with small additions of CaO or Na2O, this solubility For measuring the steep CaO-liquidus, thermal analysis
decreases markedly. Oishi et al.[7] equilibrated Cu2O-CaO is too imprecise. Hence, the following technique was used.
melts with liquid copper followed by quenching and analysis. Eleven samples containing different concentrations of CaO,
With the addition of 3 wt pct CaO to Cu2O, the Cu solubility either in the single-phase liquid or two-phase (liquid  CaO)
at 1300 °C was reduced from 14 wt pct (Figure 4) to region, were equilibrated with Cu at 1235 °C for 2 hours and
2 wt pct, due to the high electropositivity of Ca relative to quenched, as described in Section II–B. The oxygen content

Fig. 2—Inner surface of alumina crucible showing calcium aluminate pas-


sivation layer (layer became detached during cooling). Fig. 3—Phase diagram[11] of the Cu-Cu2O system.

METALLURGICAL AND MATERIALS TRANSACTIONS B VOLUME 34B, OCTOBER 2003—633


19_02-446B-G.qxd 8/28/03 3:45 PM Page 634

of the Cu phase was subsequently analyzed and plotted vs analyzed for CaO. The solubility was found to be less than
the CaO content of the slag, as shown in Figure 5. 0.1 wt pct CaO in solid Cu2O at 780 °C.
In the two-phase region, the oxygen potential is inde- The phase boundaries drawn on Figure 4 were calculated
pendent of composition. Hence, from Figure 5, it can be seen from the model, as discussed in Section V. The experimental
that the liquidus composition at 1235 °C occurs at points of the present study are systematically approximately
(11  1) wt pct CaO. This point is plotted in Figure 4. 5 °C below the calculated boundaries. We believe that this
Measurements of the solubility of CaO were also probably reflects a real systematic error due to undercool-
performed by Kuxmann and Kurre[6] who used CaO ing and to contamination by dissolved alumina.
crucibles. After quenching, the slag and metal phase were
both analyzed. As can be seen from Figure 4, there is good
agreement between the results of Kuxmann and Kurre and 2. Activity of Cu2O in the slag
the present study. The activity of Cu2O in molten Cu2O-CaO slags equili-
Mutual solid solubilities of Cu2O and CaO were measured brated with Cu was measured at 1235 °C, as described in
as described in Section II. C. An equimolar mixture of Cu2O Section II–B. Results are shown in Figure 6. A slight neg-
and CaO was equilibrated at 1235 °C and quenched, and the ative deviation from ideal solution behavior of the activity
solid primary lime particles were analyzed for Cu2O. The of CuO1/2 (aCuO1/2  aCu2O1/2) is seen when plotted vs mole
copper oxide (either CuO or Cu2O) solubility in solid CaO fraction in the CuO1/2-CaO system, thereby indicating a
(solidus composition) at 1235 °C was measured as 1.5 wt. pct. small negative deviation from ideal (Temkin) random mixing
To measure the solubility of CaO in solid Cu2O, a ternary of Cu and Ca ions.
mixture (80 wt pct Cu2O, 5 wt pct CaO, 15 wt pct Na2O) These same results are also plotted as equilibrium oxygen
was heated to 1260 °C, cooled slowly to 780 °C (20 °C below pressure (calculated from Eq. [4]) vs composition in Figure 7,
the ternary eutectic temperature), held at 780 °C for 3 hours where the results of Oishi et al.[7] at 1300 °C are also plotted.
and quenched. The Cu2O-rich phase of the eutectic was These authors measured PO2 using CaO-ZrO2 solid electrolytes
with Ni/NiO reference electrodes.

Fig.4—Cu2O-CaO-(Cu) phase diagram in equilibrium with Cu (solubili- Fig. 6—Activity of CuO1/2 (liquid standard state) in CuO1/2-CaO slags equi-
ties in terminal solid solutions  0.1 wt pct). librated with Cu at 1235 °C.

Fig. 5—Measurement of solubility of CaO in liquid Cu2O in equilibrium Fig. 7—Equilibrium oxygen partial pressure of Cu2O-CaO slags equili-
with Cu at 1235 °C. brated with copper.

634—VOLUME 34B, OCTOBER 2003 METALLURGICAL AND MATERIALS TRANSACTIONS B


19_02-446B-G.qxd 8/28/03 3:45 PM Page 635

B. Cu2O-Na2O System C. Cu2O-CaO-Na2O System


1. Phase diagram The DTA measurements were performed at constant
Results of differential thermal analyses are presented in Cu2O compositions of 70, 80, and 90 wt pct (which is the com-
Figure 8. After the experiments, several slag samples were position range of most practical interest) in the Cu2O-CaO-Na2O
analyzed. Within error limits, the compositions were the system in equilibrium with Cu. Results are presented in
same as the original weighed-in compositions, indicating Figures 11 through 13. Due to its steepness, measurements of
that sodium losses by volatilization were negligible. the CaO liquidus are less precise.
A ternary mixture (80 wt pct Cu2O, 5 wt pct CaO, and
15 wt pct Na2O) was heated to 1260 °C, cooled slowly to
780 °C, held at 780 °C for 3 hours, and quenched. The Na2O V. THERMODYNAMIC MODELING
content of the solid Cu2O was found to be 5  1 wt pct, as The phase diagrams and the thermodynamic properties
shown in Figure 8. are related by the equation
As discussed in Section II–C, it was very difficult to
measure the Cu2O content of the solid Na2O. In general, RT ln ai (liquidus)  RT ln ai (solidus)  g0fusion(i) [8]
however, the results suggest a solubility of 1 wt pct or less.
where ai (liquidus) and ai (solidus) are the activities of
2. Activity of Cu2O in the slag component i at equilibrium on the liquidus and solidus (relative
The activity of Cu 2 O in molten Cu 2 O-Na 2 O slags to the liquid and solid standard states, respectively) and
equilibrated with Cu was measured at 1200 °C, as gfusion(i)
0
is the Gibbs energy of fusion of component i at
described in Section II–B. Results are shown in Figure 9. temperature T. The Gibbs energy of fusion of Cu2O was given
A slight negative deviation of aCuO1/2 from ideal mixing in Eq. [3]. Those of Na2O and CaO were calculated by taking
behavior is seen. The same results are plotted as equi-
librium oxygen pressure (calculated from Eq. [4]) vs com-
position in Figure 10.

Fig. 10—Equilibrium oxygen partial pressure of Cu2O-Na2O slags equili-


brated with copper at 1200 °C.

Fig. 8—Cu2O-Na2O-(Cu) phase diagram in equilibrium with Cu.

Fig. 9—Activity of CuO1/2 (liquid standard state) in CuO1/2-NaO1/2 slags Fig. 11—Isopleth of the Cu2O-CaO-Na2O-(Cu) system at 70 wt pct Cu2O
equilibrated with Cu at 1200 °C. (equilibrated with copper).

METALLURGICAL AND MATERIALS TRANSACTIONS B VOLUME 34B, OCTOBER 2003—635


19_02-446B-G.qxd 8/28/03 3:45 PM Page 636

Fig. 12—Isopleth of the Cu2O-CaO-Na2O-(Cu) system at 80 wt pct Cu2O Fig. 13—Isopleth of the Cu2O-CaO-Na2O-(Cu) system at 90 wt pct Cu2O
(equilibrated with copper). (equilibrated with copper).

the enthalpies and temperatures of fusion from the FactSage By simultaneous least-squares optimization of all the data
5.1 database[9] and assuming Cp(fusion)  0. This gives (phase diagram and activities) for the Cu2O-CaO and Cu2O-
Na2O binary systems, the following expressions were obtain
Na2O (solid)  Na2O (liquid) [9] for the model parameters:
g0fusion  47,698  33.95T J/mol
gCuCa  (44,296  19T)  34,984 (YCaO  YCuO1/2)
CaO (solid)  CaO (liquid) [10]  26,236(YCaO  YCuO1/2)2 (J/mol)
[16]
g0fusion  79,496  27.94T J/mol
As shown in Sections IV–A–2 and B–2, the liquid phase gCuNa  18,229  22,906 (YNaO1/2  YCuO1/2)
is close to an ideal (Temkin) solution of Cu,Ca, and Na  21,745(YNaO1/2  YCuO1/2)2 [17]
ions on cation sites. That is, the Gibbs energy of mixing is  8083 (YNaO1/2  YCuO1/2)3 (J/mol)
close to
The Cu2O-rich terminal solid solution was modeled as a
gideal  RT (XCuO1/2ln XCuO1/2 [11] Henrian ideal solution of NaO1/2 and CuO1/2 with the
 XCaO ln XCaO  XNaO1/2 ln XNaO1/2) following activity coefficient, independent of composition
where the Xi are the mole fractions of the components CuO1/2, at a given temperature:
CaO, and NaO1/2. As shown in Section III, the concentra-
tion of Cu is small and the concentration of Cu0 is small 21,000(J/mol)
ln gNaO1/2  [18]
as long as the content of (CaO  Na2O) is above 3 wt pct. RT
Hence, in the model, it is assumed that all copper in the slag
All lines shown in Figures 4 through 13 were back-calculated
is present as Cu.
from the model equations. The model gives a melting point
The liquid was modeled with the modified quasi-chemical
of Cu2O of 1229 °C rather than 1223 °C because of the neglect
model described elsewhere,[15] which takes into account the
of Cu0 dissolution in the slag.
small degree of short range ordering that must occur. This
In the Cu2O-Na2O system, the model indicates that the
model requires “cation-cation” coordination numbers, which
enthalpy of liquid-liquid mixing is slightly negative in Cu2O-
were chosen as ZCa  2ZNa  2ZCu  2.7548 in order to be
rich solutions and nearly zero or slightly positive in Na2O-
consistent with earlier optimizations of other slag systems,
rich solutions. This gives rise to the prediction of a very flat
thereby allowing the eventual incorporation of the present
Na2O-liquidus in Na2O-rich solutions, as shown by the
model for Cu2O-Na2O-CaO slags into our larger database
dashed line in Figure 8. The model predicts similar behav-
for multicomponent slags.[9]
ior in the Cu2O-CaO system, with a negative enthalpy of
“Equivalent fractions” Yi are also defined as
mixing in Cu2O-rich solutions, becoming nearly zero or
YCaO  2XCaO /(2XCaO  XCuO1/2  XNaO1/2) [12] positive in CaO-rich solutions. Hence, also in this system,
a flat CaO-liquidus is predicted.
YCuO1/2  XCuO1/2 /(2XCaO  XCuO1/2  XNaO1/2) [13] Since the deviations from ideal solution behavior in the
YNaO1/2  XNaO1/2 /(2XCaO  XCuO1/2  XNaO1/2) [14] liquid are small, the degree of short-range ordering is small.
Hence, as discussed by Pelton et al.,[15] the thermodynamic
The parameters of the modified quasi-chemical model properties of the binary liquids can be calculated approxi-
are the Gibbs energies gAB of the second-nearest-neighbor mately by taking the sum of gideal from Eq. [11] and an
pair-exchange reactions: excess Gibbs energy term equal to
[A  A]  [B  B]  2[A  B] gAB [15] ZCu XCuO1/2  ZCa XCaO
gE  a b YCuO1/2 YCaO gCuCa [19]
where A, B  Cu, Na, or Ca. 2

636—VOLUME 34B, OCTOBER 2003 METALLURGICAL AND MATERIALS TRANSACTIONS B


19_02-446B-G.qxd 8/28/03 3:45 PM Page 637

in the CuO1/2-CaO system, and tures into the molten copper, the eventual formation of excess
Cu2O must be taken into account. For example, Riveros et
ZCu XCuO1/2  ZNa XNaO1/2 al.[5] have shown that at 1250 °C, the solubility of copper
gE  a b YCuO1/2 YNaO1/2 gCuNa [20] in Na2CO3 is very low when the oxygen potential PO2 is
2
below 106 bar, but above PO2  105 bar, the formation of
in the CuO1/2-NaO1/2 system. a Cu2O-rich slag takes place. As shown in Figure 8, Na2O
lowers the melting point of Cu2O, leading to the eventual
The thermodynamic properties of the ternary liquid formation of a eutectic at 800 °C and approximately 28 wt
Cu2O-Na2O-CaO solution were predicted from the binary pct Na2O. Since Na2CO3 is stable as a carbonate melt, con-
model parameters of Eqs. [16] and [17] using the “sym- ditions leading to its decomposition must be taken into
metric approximation,” as discussed by Pelton and Char- account (e.g., (1) level of impurities such as As and Sb, (2)
trand. [16] No additional ternary parameters were used. presence of other oxides such as SiO2 and FeOx, or (3)
(Ideality was assumed in the CaO-NaO1/2 binary liquid control of PCO2 by adjusting the furnace draft or burner func-
solution.) The calculated (predicted) phase boundaries tions). In general, depending on the process requirements,
shown in Figures 11 through 13 agree well with the the engineer may envisage alternative conditions and use
measurements, except for the CaO-liquidus, where, as dis- these systems to limit or promote the presence of immis-
cussed in Section IV–A–1, the experimental points are cible layers. In fact, the concept of two immiscible layers
probably too low. Some of the thermal arrests at the 795 °C of slag for copper fire refining was recently used by Zamal-
invariant exhibited double differential thermal analy- loa et al.[17] in a process where a slag layer of Cu2O combined
sis peaks, as shown in Figures 12 and 13. This remains with CaO or Na2O coexists with a molten (Na2SO4-CaSO4)
unexplained. layer containing AsOx and SbOx. The present model for
The model may be used to calculate various thermo- Cu2O-CaO-Na2O equilibrated with molten copper may be
dynamic properties and phase equilibria. For example, the considered as the base system for the slag layer (copper
entire calculated liquidus surface of the Cu2O-Na2O-CaO oxide–rich phase) in both soda-lime-fluxing and sulfate-
system in equilibrium with Cu is shown in Figure 14, and refining[17] processes.
the calculated oxygen content of liquid Cu in equilibrium
with the slag as a function of slag composition is shown on
Figure 15 (where Eqs. [4] and [5] were used for the copper VII. CONCLUSIONS
and oxygen activities in molten Cu).
Based upon the present measurements, the Cu2O-CaO-
Na2O system in equilibrium with Cu has been modeled
VI. DISCUSSION thermodynamically at compositions from approximately
0 to 35 wt pct Na2O and 0 to 15 wt pct CaO. The para-
Industrially, the effectiveness of impurity removal from meters of the model are stored in the database of the
blister copper by oxidation under a layer of soda slag FactSage[9] system and can be used along with the other
(Na2CO3-CaO) is highly dependent on oxygen potential. FactSage databases and Gibbs energy minimization software
Although this may be accomplished by injecting air/O2 mix- to calculate phase equilibria and all thermodynamic

Fig.14—Calculated liquidus of the Cu2O-Na2O-CaO-(Cu) system (equili- Fig. 15—Calculated content of oxygen in liquid copper equilibrated with
brated with copper). Cu2O-CaO-Na2O slags at 1200 °C.

METALLURGICAL AND MATERIALS TRANSACTIONS B VOLUME 34B, OCTOBER 2003—637


19_02-446B-G.qxd 8/28/03 3:45 PM Page 638

properties of this system. For example, the liquidus 5. G. Riveros, Y.J. Park, Y. Takeda, and A. Yazawa: J. Min. Metall. Inst.
projection, as well as the oxygen content of the equilibrium Jpn, 1986, vol. 102, pp. 415-22.
6. J. Kuxmann and K. Kurre: Erzmetallurgy, 1968, vol. 21, pp. 199-209.
liquid copper phase as a function of slag composition, has 7. T. Oishi, Y. Kondo, and K. Ono: Trans. Jpn. Inst. Met., 1986,
been calculated. vol. 27. (12), pp. 976-80.
8. H. Ranniko: Acta Polytech. Scan., 1995, Chem. Technol. Ser., vol. 229,
pp. 1-78.
ACKNOWLEDGMENTS 9. C.W. Bale, A.D. Pelton, and W.T. Thompson: F*A*C*T (Facility for
the Analysis of Chemical Thermodynamics), Ecole Polytechnique, Mon-
Constructive discussions with Dr. P. Chartrand during the treal (http://www.crct.polymtl.ca), 2002.
entire course of this project are much appreciated. Financial 10. A.D. Mah, L.B. Pankratz, W.W. Weller, and E.G. King: Report of
Investigation 7026, U.S. Bureau of Mines, United States Government
support from the Noranda Technology Center and from the Printing Office, Washington, DC, 1967.
Natural Sciences and Engineering Research Council of 11. J.P. Neumann, T. Zhong, and Y. Chan: Phase Diagram of Binary Cop-
Canada is gratefully acknowledged. per Alloys, ASM INTERNATIONAL, Material Park, OH, 1994,
pp. 288-93.
12. G.K. Sigworth and J.F. Elliott: Can. Metall. Q., 1974, vol. 13 (3),
REFERENCES pp. 455-62.
13. T. Oishi and K. Ono: Trans. Jpn. Inst. Met., 1986, vol. 25 (4),
1. C.T. Eddy: Trans. AIME, 1931, vol. 96, p. 104. pp. 291-300.
2. W.J. Hillendbrand, B.K. Poule, and H.C. Kenny: Trans. AIME, 1933, 14. A.D. Pelton: Metall. Mater. Trans. B, 1997, vol. 28B, pp. 869-77.
vol. 100, p. 483. 15. A.D. Pelton, S.A. Degterov, G. Eriksson, C. Robelin, and
3. Q. Jiao, E. Carissimi, and D. Poggi: Proc. Copper 91-Cobre 91 Int’l. Sym- Y. Dessureault: Metall. Mater. Trans. B, 2000, vol. 31B, pp. 651-60.
posium, Ottawa, 1991, vol. 4, pp. 341-357, Pergamon Press, Elmsford, NY. 16. A.D. Pelton and P. Chartrand: Metall. Mater. Trans. A, 2001, vol. 32A,
4. U.C. Rojas and J.C. Gomez: Proc. Howard Worner Symp. on pp. 1355-60.
Injection in Pyrometallurgy, TMS-AIME, Warrendale, PA, 1996, 17. M. Zamalloa, Y. Tremblay, P. Coursol, and E. Carissimi: U.S Patent
pp. 275-87. No. 6, 395, 059 B1, 2001.

638—VOLUME 34B, OCTOBER 2003 METALLURGICAL AND MATERIALS TRANSACTIONS B

You might also like