Akira Yazawa and Takeshi Azakami
Thermodynamics of removing
impurities during copper smelting
Abstrac~.The removal of impurities from matte and from copper
during c<>ppersmeltingis discussedthermodynamically.Two mecha-
nisms are considered: oxidation followed by slagging;and vola.tili-
c zation. In matte, most of the undesirable impurities such as le~d,
antimony, bismuth and arsenic are not removed by oxidation, but
rather by volatilization. When white metal and metallic copper are
iR coexistence, these impurities are distributed mainly into the
copper phase, from which their removal is difficult. Elimination of
the various elements from blister copper is discussed, and the
behaviour of ~ulphur and oxygen in metallic copper is analy,zed
thermodynamically.- ,~'
The behaviour of impurities during copper smelting is of
great interest, since it is desirable to retain valuable
elements such as the precious metals and to reject undesi-
rable impurities such as lead, bismuth, antimony and
arsenic. Except for some statistical data, however, there is
little information in the literature on the behaviour of
impurities or on the mechanism of their elimination. This
paper examines the removal of impurities during copper
smelting, based upon available thermodynamic data. Two
mechanisms are considered: oxidation followed by slagging;
and volatilization. Removal of the impurities from the
matte and from the copper are treated separately.
Elimination of impurities from matte into slag
The distribution of various metals between matte and slag
could be discussed on the basis of their relative affinities for
sulphur and oxygen according to the reaction
2M8 (s,l) + 02 = 2MO (s,1) + 82
From an examination of the free energy of this reaction (1 -
4), Na and Cu would both seem to be sulphidized easily. In
matte-slag systems, however, considerable amounts of iron
exist in both phases. It is therefore more reasonable to
Akira Yazawa is Professor and Takeshi Azakami is Lecturer at the
Research Institute of Mineral Dressing and Metallurgy, Tohoku
University,Sendai,Japan.
CANADIAN METALLURGICAL QUARTERLY Volume 8 Number 3
Rl§sum~. Les auteurs se servent de l'energetique pour expliquer
l'elimination d'impurites de la matte et du cuivre au cours do
smeltage.Ilsconsiderentdeux mecanismesd'elimination: l'oxydation
suivie de solubilisation des oxydes dans la scorie,et la volatilisation.
Dans Ie cas de la matte, les impurites indesireestelles que Ie plomb,
l'antimoine, Ie bismuth et l'arsenic ne disparaissent pas par
oxydation, mais par volatilisation. Lorsque la matte blanche est en
equilibre avec Ie cuivremetallique, les impuretes se trouvent surtout
dans la phase cuivre, d'ou leur extraction est difficile.L'elimination
des elements du cuivre blister est discutee et Ie comportement du
soufre et de l'oxygene dans Ie cuivre metallique est traite par
energetique.
discuss the distribution of metals on the basis of the
following equilibrium relation:
FeO (1) + M8 (s,1) = Fe8 (1) + MO (s,1) 2
aFeS • aMO
aFeO • aMS
In intermediate-grade matte, the activities of FeS and FeO,
as a first approximation, may be taken as 0.4. Then,
assuming aFeS/aFeO = 1,
where 'Yis the activity coefficient of MO or MS, NM0 is the
mole fraction of MO in the slag, and NMS is that of MS in
the matte. Hence, the distribution of the metal may be
expressed by the ratio NM o/NM s. Unfortunately, reliable
measured values of 'YMO/'YMSare not available, but data
estimated by thermodynamic methodS are shown in Table I.
The calculated values of NMO/NMS show that Cu and
Ag may enter nearly completely into the matte phase, but
nickel less so, and a higher slag loss of nickel than copper
appears inevitable. The concentration of lead in the matte
will be one order of magnitude greater than in the slag. Zinc
will be distributed about equally in both phases (5). While
sodium itself has a great affmity, for sulphur, because of the
257
 Table I. Thermodynamic data for the distribution of Table II. Vapour pressureof metal
impurities between matte and slag (1200° C). sulphides (1200°C).

Sulphide o
Element ~F~ K2 'YMO/'YMS NMO/NMS PMS
(atm)
Ag 45,400 1.7 • 10-7 1 2 • 10-7
Na 29,440 4.3 • 10-5 --10-7 --102 AS2S3 »1 2
GeS 2 • 10
eu 27,600 8.0 • 10-5 1 8 • 10-5
HgS 70
Pb 12,350 1.5 • 10-2 --0.3 --0.05
SnS 1.0
Co 10,000 3.3 • 10-2
PbS 0.6
Ni 9,220 4.3 • 10-2 --5 --0.01
Ba 5,800 0.14 ThS 0.4
CdS 0.2
Ca 2,980 0.36
Bi 1,140 0.68 Bi2S3 9 • 10-2
Zn 750 0.77 0.5 2 Sb2S3 5 • 10-2
Mn 3,030 2.8 In2S3 3 • 10-2
Sn 7,510 13 ZnS(m 9 • 10-3
Al 8,770 20 CoS 2 • 10-5
Sb -10,550 37 CU2S 2 • 10-5
Mg -17,320 3.7 • 102
MoS2 3 • 10-6
FeS 8 • 10-7
MnS 4 • 10-7
extremely small value of the activity coefficient of Na2 0 in CaS «1
slag, the concentration of Na20 is expected to be higher
than Na2S. dynamic calculations (5). Figure 1 suggests, at point C, that
In discussing the distribution of various elements the partial pressure of metallic zinc vapour will be about
between matte and slag, mutual solubilities should also be 0.02 atm at 0.2 atm of S02 gas when ZnS and ZnO are in
taken into account. The solubility of alkali and alkaline coexistence.
earth metal sulphides in slag is considerable, and even heavy
metal sulphides such as FeS and CU2S dissolve to an Distribution of metal impurities between
appreciable extent in slag. white metal and blister copper
In the above discussion, the ratio aFeS/aFeO has been In the blister making stage of copper converting, during
assumed to be unity. However, in higher-grade mattes, which white metal and blister copper coexist, the distri-
especially in the converting process, the ratio is less than bution of various elements between the two phases can be
unity. The slagging of various impurities is therefore more discussed on the basis of the reaction,
easily accomplished in the converting process than in the
matte smelting process. -6

Elimination of impurities from matte

Pzn=O.Ol
by volatilization -8
The apparent vapour pressure of various sulphides at
1200°C is shown in Table II (6). Examination of the data
shows that elimination by volatilization may be expected Pzn=O .1
-10
for some sulphides, including those of Pb, Cd, Sb, Bi, As
and Sn. For increased volatilization, higher temperatures
and a large interface area between the gas phase and the
-12
matte are desirable. The converter seems to be a fairly ideal
vessel in which to realize these conditions. Volatilization
should be carried out during the matte blow, because most zn=9.07
-14 ZnS
impurities will be distributed in the metallic phase in which
removal is more difficult. Zn (,I)
For HgS, ZnS and CdS direct reduction to metal vapour
according to the following reaction may also be important: ·-16
~10 -6 -4 -2 o
MS + 02 = M (g) + S02 4 log Ps2(atm)

The extent of zinc metal vaporization can be determined Figure 1. Isothermal reaction equilibrium diagram for the
under various conditions by stoichiometric and thermo- Zn-5-0 system at 1200°C.

258 CANADIAN METALLURGICAL QUARTERLY

 CU2S (1) + M (s,1) = MS (s,1) + 2Cu (1) 5 Table III. Data for the distribution of metal impurities
2 .• between molten copper and copper sulphide (1200°C).
0Cu OMS
Ks =
0CU2S• oM Distribution Ratio
It can be assumed that a~u/a CU2S = 1, and therefore Element Ks
Experimental
Data·
K2 = oMS/oM = ('YMS• NMS)/ ('YM• NM) 6

Bi 27,780 6.6 • lQ-5 1.8 • 10.4 0.13

Sb 26,700 1.1 • 10-4 1.4 • 10-6 0.072
Values of the ratio 'YMS/'YMcan be roughly estimated, and Sn 19,140 1.4 • 10.3 1. 7 • 10.4
the distribution ratio NMS/NM can then be derived (Table Ag 13,060 1.2 • 10-2 3.7 • 10.2
III) and compared with data obtained experimentally Pb 11,370 2.1 • 10-2 0.12 0.1 '" 0.2
(7-10). The most harmful impurities in copper metallurgy Ni 9,580 3.8 • 10-2 0.11 0.33
Co 7,190 8.6 • 10-2 0.86
are seen to be distributed mainly into the copper phase.
Cd 4,790 0.20 0.15
Fe 3,010 0.36 5.4
Behaviour of sulphur and oxygen in metallic Zn 4,950 5.4 0.60 0.67,1.0
copper Mn -20,190 9.9 • 102 7.9 • 102
Na -37,880 4.2.105
Equilibrium relationships in the Cu-8-0 system are illus-
trated in Figure 2. Using interaction parameters for activity
coefficients (11), iso-per cent lines of sulphur and oxygen • Distribution ratio = % in CU2S/% in Cu.
have been determined and are shown on the diagram. The Table IV. Thermodynamic data for the removal of
blister making and finishing stages of copper converting can impurities from molten copper (1200°C).
be conveniently discussed by referring to this figure. During
Content
these stages, converting proceeds along the line A-B-C Element *** 'Y~ (Ref.) p~
which corresponds to a partial pressure of 802 PS02 = 0.2
(%) (atm)
atm. The blister making stage is represented by point B at
which white metal is in equilibrium with molten copper Au 0.003 1.2 • 10-7 0.34 (17) 4.9 • 10-7
containing 1 per cent 8 and 0.1 per cent O. During the Hg 2.5 • 10-5 5.2 • 102
Ag 0.1 3.5 • 10-5 4.8 (12) 2.2 • 10-4
finishing stage, the composition of the metallic phase moves Pt 5.2 • 10-5 0.03 (18) 6.4 • 10-13
to point C. As shown by the diagram, CU20 cannot exist as Pd 6.2 • 10-4 0.06 (18) 8.5 • 10.7
a stable phase during converting. After converting, the Se 0.04 5.6 • 10-4 «1 66
blister copper is refined in the anode furnace by oxidation Te 0.01 7.7· 10-2 0.01 ? 3.9
and reduction, to obtain anode copper represented by point Bi 0.009 0.64 2.7 (14) 4.2 • 10.2•
D. Cu '" 99 1 4.5 • 10.6
Pb 0.2 3.8 5.7 (13) 1.9 • 10-2
Ni 0.2 25 2.8 (19) 2.8 • 10-8
Removal of impurities from copper by oxidation Cd 31 0.73 (16) 32
In copper converting, and in fire refming in anode and Sb 0.04 50 0.013 (14) 7.9 • 10-2 ••
As 0.04 50 0.0005' (15) 5 • 102 ?
cathode furnaces, the behaviour of impurities in molten
Co 0.001 1.4· 102 10 ? 3.2 • 10-8
copper is of primary importance. The removal of metallic Ge 3.2 • 102 4.2 • 10.7
impurities by oxidation may be examined by considering Sn 0.005 4.4· 102 0.11 (12) 6.5 • 10.6
the following equilibrium relation: In 8.2 • 102 0.32 (15) 8.1 • 10-4
Fe 0.01 4.5 • 103 15 (20) 7.8 • 10-8
CU20 (1) + M (1) = 2Cu (1) + MO (I,s) 7
Zn 0.007 4.7· 104 0.11 (16) 10
2
Na 1.1 • 105 8.7
0Cu • 0MO
K7 - 8 Cr 5.2 • 106 3.2 • 10.7
0CU20 • OM Mn 3.5 • 107 0.80 (21) 9.3 • 10.4
Si 0.002 5.6· 108 0.1 (22) 1• 10-6
Values of K7 at 1200° C for various metals are shown in Ti 5.8 • 109 6.3 • 10-10
Table IV. The data in this table are arranged according to Al 0.005 8.8· 1011 0.008 (18) 1.3 • 10-5
increasing values of K7, that is, according to the value of
Ba 3.3 • 1012 6.3 • 10-2
Mg 1.4 • 1013 0.067 (18) 2.1
~ F~. The metals can be divided into three groups. The first Be 5.4 • 1013 9.5 • 10-6
group of metals, from gold to tellurium, have small K Ca 4.3 • 1014 o· 13
values, suggesting that their removal by oxidation is
difficult. Fortunately, their removal is not desirable in .pOBi = PBi + PBi2· •• pOSb = PSb + PSb2 + PSb4· ··-Typical
practice because these are also the valuable metals which blister copper from Japanese smelters.

Volume 8 Number 3 259

 -2
aCu=0.3

o -3

aCu=0.6

e -4

,,
+oa
Cd 1.5%0 acu= 1
;- /
~.
bD -5
.£

-6

-7 eu
0.02

-8
-10 -9 -8 -7 -4
log PSz (atm)
Figure 2. Isothermal reaction equilibrium diagram for the Cu-S-O system at
1200°C.

are recovered from anode slime. On the other hand, because metals may be arranged in order of their increasing
of the large K values, the third group of metals, from iron tendency to oxidize:
to calcium, should be removed easily by oxidation. As -+- Sb -+- Bi -+- Pb -+-Cd -+- Sn -+- Ni -+- In -+- Co -+- Zn -+- Fe
Although small amounts of these metals are found in
However, this order is for equal mole fractions NM of the
commercial crude copper, they may not be in the metallic
metals M in molten copper, and equal values of aM o. In
state, but are most likely present as nonmetallic inclusions.
any case it is clear that As, Bi and Sb are not readily
This leaves only the removal of the second group of metal
removed by simple oxidation.
impurities, from bismuth to indium, to be discussed in
more detail. To find the limitations of impurity removal by
Elimination of impurities from copper
oxidation, we can consider molten copper saturated with
oxygen, in which case Equation 8 reduces to
by volatilization
We shall now discuss the removal of impurities from crude
aMO 'YMO a NMO copper by volatilization. The vapour pressure of impurity
K7 =-- =---- 9
aM 'Y°MaNM metals dissolved in copper is given by the following
equation:
where 'Y~ is that in an infmitely dilute solution. The 'set o 0 0
copper' condition in practice, may be close to oxygen
PM = PM a aM = PM a 'YM a NM 10

saturation. To obtain good results in removing metal
impurities, it is desirable to have small values of 'YMO and
NM 0' and large values of 'Y~, as well as K. The formation
and skimming of slag may decrease the value of NMO' The
addition of silica to remove lead, or the addition of soda
ash or lime to remove As, Sb and Sn, may decrease the
value of 1M o·
Based upon thermodynamic and available experimental
data, values of 'Y~ in various binary Cu-M alloys have been
calculated at 1200° C, and are listed in Table IV. Consi-
dering the values of both K and 'Y~, the second group of
where p~ is the vapour pressure of the pure metal. Vapour
pressures for impurity metals at 1200° C are shown in Table
IV. From the values of p~ and 'Y~ ,it is evident that Zn and
Cd should be readily removed by volatilization, while
because of their extremely small values of 'Y~, elements
such as As, Se and Te will be difficult to remove even
though they have large values of p~ • In the finishing period
of copper converting, the loss of Ag by volatilization is
sometimes mentioned. From the values of p~ and 'Y~, the
ratio of silver to copper in the gas phase may be estimated
to be about 230 times that in melt.

260 CANADIAN METALLURGICAL QUARTERLY

 Except for a few metals, elimination by volatilization is 4Cu (1) + 02 = 2CU20 (1), log K1473 = 4.566 (2)

rather difficult in copper processing at atmospheric pres-
sure. Of course, vacuum melting would be more effective in
removing impurities (23).
Conclusion
The behaviour of impurities during copper smelting has
been discussed thermodynamically. It has been shown that
elements such as As, Bi, Sb and Pb, which are harmful
Moreover, the iso-sulphur and iso-oxygen curves in Figure 2
are in approximate agreement with those derived from the
data obtained by Floe and Chipman (24) when the
concentrations of sulphur and oxygen are small.
The iso-pressure lines of S02 are shown based upon the
data for the reaction
log K1473 = 9.018 (2)

impurities in the electrolytic refining of copper and in the and the boundary line between CU2S and CU20 is
fabrication of copper and copper alloys are very difficult determined from the data for the equation
to remove by oxidation or volatilization from molten
copper. These elements 'should be removed as much as 2CU20 (1) + CU2S (I) = 6Cu (1) + S02,
(1,2)
possible while they are still in the matte, or even in the ore. log K1473 = 1.330

APPENDIX Acknowledgment
The isothermal reaction equilibrium diagram for the Cu-S-O The authors wish to express their thanks to Dr. P. Tarassoff
system (Figure 2) is based mainly on data published by of Noranda Research Centre for correction of the English in
Sano and Sakao (11) as follows: the manuscript.
References
1-02 = Q (in CU), log Ko = 3950/T- 0.584
(1) Richardson, F.D., and J.H.E. Jeffes. JISI 171: 165. 1952.
1
Ko = ao/poT, ao = /0(%0) = /bO) • /6S) • (%0) (2) Coughlin, J.P. U.S. Bur. Mines. Bull. No. 542. 1954.
(3) Elliott, J.F., and M. Gleiser. Thermochemistry for Steel-
making, Vol. 1. Reading, Mass., Addison-Wesley. 1960.
1-S2 = ~ (in Cu), log Ks = 5440/T- 0.815
(4) Kubaschewski, 0., E.L.L.Evans, and C.B. Alcock. Metallur-
1
= 18(%S) = f~S) gical Thermochemistry, 4th ed. New York, Pergamon Press.
KS = as/psI, as • f~O) • (%S)
1967.
(5) Yazawa, A., J. Min. Metall. Inst. Japan. 83: 1093. 1967.
and (6) Azakami, T., and M. Abe. Bull. Res. Inst. Mineral Dress.
Meta 11.Tohoku Univ. 20: 65. 1964.
logf6°) = - 311.3 (%0) /T (7) Asano, N., and T. Ichio. Suiyolazishi. 14: 467. 1962.
(8) Asano, N., and M. Wada. 1. Min. Metall. Inst. Japan. 79: 235.
log/g) = - 242.6 (%S) /T 1963.
(9) Asano, N.J. Min. Metall. Inst. Japan. 81: 283. 1965.
log I~S) = _ 281.6 (%8) /T (10) Azuma, K., S. Goto, and N. Tatebe. J. Min. Metall. Inst. Japan.
log f~) =- 485.2 (%O)IT 83: 607. 1967.
(11) Sano, K., and H. Sakao. J. Japan Inst. Metals. 19: 431, 435,
Combining these equations, 504,655.1955.
(12) Azakami, T., and A. Yazawa. To be published.
log Ko =- 311.3 (%0) /T - 242.6 (%8) /T + log (%0)
(13) Yazawa, A., T. Azakami, and T. Kawashima. J. Min. Me tall.
-t log P02 = 3950/T - 0.584 Inst. Japan. 82: 519. 1966.
(14) Azakami, T., and A. Yazawa. J. Min. Metall. Inst. Japan. 83:
log Ks = - 281.6 (%8) /T - 485.2 (%0) /T + log (%8) 666.1967.
(15) Azakami, T., and A. Yazawa. J. Min. Metall. Inst. Japan. 85:
-tlogPs2 = 5440/T - ~.815 97.1969.
(16) Azakami, T., and A. Yazama. J. Min. Me tall. Inst. Japan. 84:
At 1200°C
1663.1968.
-0.211 (0%) - 0.165 (%S) + log (%0) - tlogP02 = 2.098 (17) Edwards, R.K., and M.B. Brodsky. J. Am. Chem. Soc. 78:
2983. 1956.
-0.191 (s%) - 0.329 <%0) + log (%S) - tlogPS2 = 2.878 (18) Hultgren, R., et al.. Selected Values of Thermodynamic
Properties of Metals and Alloys. New York, John Wiley. 1963.
Using these two equations, iso-oxygen and iso-sulphur (19) Barbi, G.B. Annali di chimica. 56: 992. 1966.
curves are drawn as shown in Figure 2. The solubility limits (20) Yokoyama, E., and M. Kameda. Bull. Res. Inst. Mineral Dress.
Metall. Tohoku Univ. 21: 85.1965.
of sulphur and oxygen in copper at 1200° C were assumed
(21) Yeremenko, V.N., G.M. Lukashenko, and B.R. Sidorko.
to be 1 and 1.5 per cent, respectively. The boundary lines Russian Met. Mining. NO.6: 111. 1964.
between Cu and CU2S, or Cu and CU20 are in general (22) Sano, K., K. Okajima, and N. Okada. J. Japan Inst. Metals. 20:
agreement with those derived from the following equations, 323. 1956.
assuming unit activities for Cu, CU2Sand CU2O. (23) Kameda, M., and A. Yazawa. Bull. Res. Inst. Mineral Dress.
Metall. Tohoku Univ. 19: 57. 1963.
log K1473 = 6.245 (1) (24) Floe, C.F., and J. Chipman. Trans. AIME. 147:"28.1942.

Volume 8 Number 3
261