QUANTITATIVE EQUILIBRIUM CALCULATIONS ON SYSTEMS WITH RELEVANCE TO COPPER SMELTING AND CONVERTING by BO BJÖRKMAN AKADEMISK AVHANDLING som med tillstå n d av rektorsämbetet vid Umeå Universitet för erhållande av filosofie doktorsexamen framlägges t i l l o ffe n tlig granskning vid Kemiska Institutionen, sal B, LuO, onsdagen den 30 maj 1984, kl 10.00 Umeå 1984 Fakultetsopponent: Dr. Philip J. Spencer, Aachen
75
Embed
QUANTITATIVE EQUILIBRIUM CALCULATIONS ON SYSTEMS …
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
QUANTITATIVE EQUILIBRIUM CALCULATIONS ON SYSTEMS WITH RELEVANCE TO
COPPER SMELTING AND CONVERTING
by
BO BJÖRKMAN
AKADEMISK AVHANDLING
som med t i l ls tå n d av rektorsämbetet v id Umeå U n ive rs ite t fö r erhållande av f i lo s o f ie doktorsexamen framlägges t i l l o f fe n t l ig granskning vid Kemiska In s titu tio n e n , sal B, LuO, onsdagen den 30 maj 1984, kl 10.00
Umeå 1984
Fakultetsopponent: Dr. P h ilip J. Spencer, Aachen
T i t le :
Author:
Address:
Abstract:
Key words:
Q uantita tive equ ilib rium ca lcu la tions on systems w ith relevance to copper smelting and converting
Bo Björkman
Department o f Inorganic Chemistry, U n ivers ity o f Umeå, S-901 87 Umeå, Sweden
The present thesis gives a summary o f resu lts obtained through theo re tica l and experimental studies o f systems w ith relevance to copper smelting and converting.
Many chemical elements are involved in the copper production processes and a deta iled experimental study would be very time- consuming and expensive. A complicating fa c t is also the co rro s i- v i ty of the liq u id phases towards container m ateria l. A powerful a lte rn a tive is equ ilib rium ca lcu la tions , in which models fo r the liq u id phases as well as re lia b le basic thermodynamic data are needed.
In the present thes is , a generalized s tructu re based model fo r liq u id s ilic a te s was developed and used in assessments o f the systems PbO-SiC^, Fe-O-SiOo, CUO0 . 5 -SÌO2 and Cu-Fe-O-SiOo. In the model, the non-ideal s i l ic a te melt is treated as an ideal solutio n but containing a few complexes. The PbO-Si02 melt could be described by introducing the complexes Pb3 Si?0 y, Pb4 Si^0 -jQ and Pbi3 S i12O37 in addition to the components PbO and Pb2Si0 4 . The species considered in the Fe-O-SiC^ melt were FeO, FeO ]^,Fe?Si0 4 , Fe3 Sio0 7 , Fe3SigO| 5 and in the CUO0 . 5 -SÌO2 melt CuOo. 5 ana CU4 SÌO4 . Trie calculated phase diagrams, the a c t iv it ie s o f meta l oxides and the oxygen p a rtia l pressures were a l l in good agreement w ith the published data.
Two o f the papers in th is thesis concern the determination o f Gibbs free energies fo r Cu2S (s ,l) and Ca2 Fe2Û5 (s) through emf measurements u t i l iz in g a so lid e le c tro ly te . A c t iv it ie s and term inal s o lu b il i t ie s in the so lid so lu tion [Fe^,Ca]0 were also determined.
The resu lts obtained from the quan tita tive equ ilib rium ca lcu la tions fo r conventional copper smelting and converting were used to o u tlin e the overa ll reactions taking place and the outcome o f changes in process parameters. Comparison w ith observed values, however, showed tha t the copper and magnetite contents in slag were calculated too low. These discrepancies could be completely explained by using a non-equilibrium approach in which the conve rte r was assumed to consist of several segments w ith concentratio n gradients between the segments.
Assessment o f s i l ic a te systems, model fo r l iq u id s i l ic a te s , lead s i l ic a te s , iron s i l ic a te s , cuprous s il ic a te s , q uan tita tive equilib rium ca lcu la tion s , emf measurements, Gibbs free energy, cuprous sulphide, dicalcium fe r r i t e , w üs tite -ca lc ia so lid so lu tion , copper smelting and converting.
ISBN 91-7174-167-4 67 pages + 5 appendices
To Tuija
Tina & Matti
QUANTITATIVE EQUILIBRIUM CALCULATIONS ON SYSTEMS WITH RELEVANCE TO
COPPER SMELTING AND CONVERTING
BO BJÖRKMAN
Department o f Inorganic Chemistry, U n ivers ity o f Umeå,
S-901 87 Umeå, Sweden
This thesis is a summary o f the fo llow ing papers, in the te x t re fe r
red to by th e ir Roman numerals I-V .
I. Q uantita tive Equilibrium Calculations on Conventional Copper
Smelting and Converting.
Björkman, B. and Eriksson, G., Can. M e ta ll. Q. 1982, 21, 329-
37.
I I . A Generalized Approach to the Flood-Knapp Structure Based Mo
del fo r Binary Liquid S ilic a te s : Application and Update fo r
the PbO-SiO^ System.
Björkman, B ., Eriksson, G. and Rosén, E., Met a l l . Trans. B.
In p r in t.
I I I . An Assessment o f the System Fe-O-SiO^ Using a S tructure Based
Model fo r the Liquid S ilic a te Phase.
Björkman, B ., To be published.
i i
IV. Determination o f the S ta b il i ty o f Cu S in the Temperature
Range 1000 K-1450 K by Solid State EMF Measurements.
Björkman, B. and Fredriksson, M., Scand. J. M e ta ll. 1982, 11,
281-86.
V. A Solid State EMF Study o f the System Ca0 -Ca2 Fe2 0 5 -Fet 0 -Fe.
Björkman, B ., Scand. J. M e ta ll. Submitted fo r pub lica tion .
i i i
TABLE OF CONTENTS
INTRODUCTION 1Aim of the present study 2
Outline o f the work 3
MODEL FOR LIQUID SILICATES 5S ilic a te structures 5
Solid s ilic a te s 5Liquid s ilic a te s 7
Present models 10Proposed model fo r liq u id s ilic a te s 18
Application o f the model 21Pb0-Si02 21Fe-0-Si02 25
Cu0 0 .5 -Si02 27Cu-Fe-0-Si02 31
Discussion o f the present s truc tu re based model 35
EXPERIMENTAL STUDIES 36Method 36
Comments to resu lts 39
Af G° fo r Cu2S (s ,l) 39The system CaO-Ca^e^g-Fe^O-Fe 40
APPLICATION TO COPPER SMELTING AND CONVERTING 41Non-equilibrium conditions 47
Calculation procedure 49Results 51
FUTURE IMPROVEMENTS OF THE THERMODYNAMIC DESCRIPTION 55
ACKNOWLEDGEMENTS 57
REFERENCES 59
1
Introduction
The chemical reactions taking place in technical and geological pro
cesses, e.g. the formation o f minerals and ores in the ea rth 's crust
and the extraction o f valuable metals from these ores, have been the
subject o f extensive studies during the la s t decades. Nevertheless’,
due to th e ir complexity, many o f these processes are not ye t fu l ly
understood. This is especia lly the case fo r the influence o f minor
elements. As an example, a survey o f the main chemical reactions ta
king place in conventional copper production through smelting and
converting can be obtained by studying the system Cu-Fe-S-O-SiC^.
However, to a rr ive a t a more complete descrip tion , the elements Ca,
Mg, A l, N i, Pb, As, Sb and Bi should also be incorporated. Such
multicomponent systems are usually d i f f i c u l t to handle experimental
ly due to th e ir complexity and because a t high temperatures the cor
ro s iv i ty o f the liq u id phases towards container material may cause
severe problems. Furthermore, a systematical study o f the influence
o f a l l components on the outcome o f a process would be very time-
consuming and expensive, especia lly i f the study is performed on a
f u l l scale process. Therefore most o f the experimental data presen
ted h ith e rto have been determined in binary systems and very ra re ly
in systems higher than ternary.
An a lte rn a tive way to study multicomponent systems is to use quanti
ta t iv e equ ilib rium ca lcu la tions , since in most cases re lia b le th e r
modynamic descriptions fo r the constituen t binary and ternary sys
tems are the only needed. Equilibrium ca lcu la tions can therefore
serve as an e ffe c tiv e tool to simulate multicomponent processes and
2
to p red ic t the outcome o f a change in process parameters.
Today, e f f ic ie n t computer programs e x is t fo r the ca lcu la tion o f equi
lib rium compositions in multicomponent systems, e.g. the free energy
m inim ization program SOLGASMIX (1 ). This computer program has success
fu l ly been applied to a number of systems, e.g. roasting o f chalcopy-
r i t e , CuFe$2 ( 2 ) , phase e q u ilib r ia in multicomponent a llo y systems
(3 ), mineral assemblages in p y ro lite and lh e rz o lite (4) and phase
e q u ilib r ia in a gas o f so lar composition (5 ,6 ). A p rerequ is ite fo r
these ca lcu la tions is tha t the Gibbs free energies are known fo r a l l
species which might appear in nonneglig ible amounts and tha t models
are given which describe the composition dependence o f the excess
Gibbs free energies o f the so lu tions.
Aim o f the present study
The aim o f th is study has been to ca lcu la te q uan tita tive equ ilib rium
compositions in systems w ith relevance to the conventional copper
smelting and converting and to study the extent to which equ ilib rium
ca lcu la tions may be used to simulate these processes. To accomplish
th is , a considerable part o f th is work has been directed towards ob
ta in ing a generalized s tructu re based model fo r s i l ic a te systems.
This model was then used in an assessment o f the most important bina
ry and ternary copper-iron slag systems.
Although a large body o f experimental data exis ts today, thermodyna
mic data fo r many important phases in systems w ith relevance to cop
per pyrometallurgy are e ith e r lacking or o f low accuracy. Part o f the
3
work has therefore been concerned w ith determination o f improved
thermodynamic data fo r C i^ S ts J ), Ca2 Fe2 0 ^(s) and the so lid so lu tion
[Fet ,Ca]0.
Putì ine o f t he work
This study commenced w ith a quan tita tive ca lcu la tion o f the e q u ilib
rium compositions in conventional copper smelting and converting ( I ) .
The composition dependence o f the a c t iv i ty co e ffic ie n ts fo r liq u id
species was described by using as simple mathematical expressions as
possible.
There have been many recent reports o f measurements by spectroscopy
and w ith ana ly tica l and high temperature X-ray d if fra c t io n techniques
on both glassy and liq u id s il ic a te s . These measurements have shown
tha t some complexation occurs between the mixed oxides, a fa c t tha t
ju s t i f ie s the use o f a s tructu re based model to describe the non
id e a lity o f s i l ic a te melts. Information on the composition o f the
complexes present in s i l ic a te melts may thereby also be gained.
A generalized s tructu re based model fo r liq u id s ilic a te s was presen
ted in paper ( I I ) and used in an assessment o f the systems PbO-SiO^
( I I ) and Fe-O-SiC^ ( I I I ) . The system PbO-SiC^ was chosen to te s t the
model because very accurate a c t iv it ie s o f PbO(l) are ava ilab le (7)
and because Raman spectroscopic measurements on PbO-Si0^ glasses (8 )
have shown tha t these glasses and presumably also the melt consist o f
a few s il ic a te complexes. The s tructu re based model has also been ap
p lied to the system CuOQ g-SiO^ and to s i l ic a saturated melts in the
4
system Cu-Fe-O-SiO^. The resu lts obtained fo r these la t te r systems
w i l l be discussed in th is summary paper. The system Cu-Fe-O-SiO^ con
ta ins the main elements fo r a descrip tion o f the slag phase in conven
tio n a l copper pyrometallurgy.
The sulphide phase (matte phase) in conventional copper pyrometal
lurgy consists mainly o f the elements Cu, Fe and S. Improved values
o f the Gibbs free energy o f formation fo r Cu2S (s , l) , obtained through
emf measurements u t i l iz in g ca lc ia s ta b ilize d z ircon ia as so lid e lec
t ro ly te , are given in paper (IV ).
The copper concentrates used in copper pyrometallurgy contain, in
many cases, rather large amounts o f CaO and/or MgO, e ith e r derived
from the copper ore or purposely added to bene fit the process. In the
M itsubishi process (9) CaO is added instead o f SiO^ to give an oxide
phase w ith the desired properties. The systems Ca-Fe-0 and Ca-Fe-0-
SÌO2 are therefore o f as great in te re s t fo r copper m e ta llu rg is ts as
fo r iron m e ta llu rg is ts .
In an assessment o f the systems Ca-Fe- 0 and C a-Fe-O -S^* a knowledge
o f the subsolidus phase re la tions is most important. However, fo r
some parts o f these systems the published data are con trad ic to ry .
Subsolidus phase re la tions in the system Ca0 -Ca2Fe2 0 ^-Fet 0 -Fe, a c t i
v it ie s o f Fe O in the so lid solu tions and the s ta b i l i t y o f Ca2 Fe2 0 ^(s)
were therefore determined from emf measurements using ca lc ia s t a b i l i
zed z ircon ia as so lid e le c tro ly te . The resu lts obtained are given in
paper (V).
Model fo r l iq u id s ilic a te s
Models fo r liq u id s ilic a te s have occurred qu ite frequently in the l i
te ra tu re . These are e ithe r based on parameterized expressions fo r the
a c t iv i ty co e ffic ie n ts o f the mixed components or on d if fe re n t assump
tions o f the melt s truc tu re . To compare these models and to develop a
generalized model fo r liq u id s ilic a te s based on sound s tru c tu ra l as
sumptions, a b r ie f review o f the lite ra tu re on the structures o f so
l id and liq u id s ilic a te s is necessary.
S i l ica te structures
Sol id s i 1ica te s . In most cases s ilic o n is in fo u rfo ld coordination
w ith oxygen, resu lting in s l ig h t ly d is to rted [SiO^] tetrahedra, but
in a very few cases the oxygen coordination has been found to be s ix
fo ld in the shape o f a s l ig h t ly d is to rted octahedron, e.g. in thauma-
s ite Ca^ISi(OH)^][COg][SO^]•1 2 ^0 , in the high pressure m odifica tion
of S i0 2 * s tis h o v ite , and in the high pressure phase
K [(S i0 7 5 ’ AIq 25^4^8-* ^ most s ilic a te s contain iso la ted or
corner shared [SiO^] groups, the only exception being fib rous S i0^
which is b u i l t up from edge shared [S iO ^ tetrahedra. With very few
exceptions the s il ic a te anions w ith iso la ted and corner shared [S i0^]
groups can be c la s s ifie d according to Table 1.
S ilic a is a strong ly ac id ic oxide and in the presence o f a more basic
oxide w i l l react according to the reaction scheme
Table 1. Broad c la s s if ic a tio n o f s ilic a te s w ith fo u rfo ld coordinated
s il ic o n atoms. From Liebau (10).
S i ng 1 e Double T rip le Quadruple
Quintup le Hexuple
Tetrahedra + + + - - -
Chains + + + + + -
Layers + + - - - -
Frameworks +
Rings + + - - - -
7
Consider the three-dimensional array o f [SiO^] tetrahedra in the 3-
c r is to b a lite s tructu re il lu s tra te d in Fig. la . 3 -c r is to b a li te is the
stable m odifica tion o f SiO^ from 1743 K and up to the melting po in t.
I f a more basic oxide MO is added, some o f the Si-O-Si bonds w i l l be
broken according to reaction ( I ) , resu lting in a formation o f new so
l id phases containing s i l ic a te anions w ith the s truc tu re o f e ith e r
in f in i te layers or chains or small d iscre te rings or chains (see Fig.
lb - f and Table 1).
The ra t io o f SiO^ to basic oxide is the most important fa c to r deter
mining the degree o f condensation o f the [SiO^l tetrahedra, but not
the only important one. The size and charge o f the cations w i l l de
termine whether a ce rta in anion is stable or not and also the perio
d ic ity o f e.g. the chain anions ( 1 0 ).
L iquid s i l ic a te s . The s truc tu ra l properties o f l iq u id s ilic a te s can
be assumed to be s im ila r to so lid s il ic a te s . The anions present in
s i l ic a te melts would thus be d iscre te o rth o s ilic a te anions, chain
anions, rings , layer anions and three-dimensional framework anions.
I t is not reasonable, however, to assume tha t the chains, layers and
three-dimensional framework anions in the melts are very la rge, but
tha t they are merely fragments o f the in f in i te anions present in so
l id s il ic a te s . These assumptions on the s tructu re o f the s i l ic a te
melts are also in agreement w ith most o f the ava ilab le experimental
resu lts concerning s i l ic a te melt s tructu res.
High temperature X-ray d if fra c t io n measurements on several s i l ic a te
melt systems have been performed by Waseda et a l. (13 through 17).
8
Fig. 1. The id e a l iz e d structures o f a) c r i s toh a l i t e , b) layers o f
SiOq te trahedra, c) id e a l iz e d s ing le pyroxene chain seen in perspective
d) d i f f e r e n t configurations o f s ing le and double chains3 e) t r i p l e rings
in b e n i to i te , BaTiSi^Og, f ) d iscre te o r th o s i l ic a te anions in Mg^SiO^;
shaded c irc le s represent Mg. Throughout th is f igu re small black c irc le s
represent Si atoms and open c irc le s represent 0 atoms. F ig . la s b> d - f
fi*om V e ils ( ! ! ) and F ia . 1c from Deer e t a l . (12).
Bild borttagen – se tryckt versionImage removed – see printed version
9
The systems investigated comprise L i20-S i025 Na20-S i02 and I<2 0 -S i0 2
(13,14), Ca0-Si02 and Mg0-Si02 (15), Fe0-Si02 (16) and F eO -F e^-
Si02 (17). These measurements have confirmed th a t the fundamental
local ordering u n it in liq u id s ilic a te s is the [SiO^] tetrahedron.
The S i-S i coordination number was shown to gradually increase by the
add ition o f Si02, ind ica ting a polymerization o f [SiO^] tetrahedra.
At the m etas ilica te composition, MSiO^, the S i-S i coordination num
ber is about three in several o f the systems.
The glasses formed on rapid cooling o f most s i l ic a te melts co n s titu
te a lin k between the melt and the so lid phases and i t is reasonable
to expect s im ila r structures in the glass and the melt. This has a l
so been shown by Waseda and Toguri (15) through high temperature X-
ray d if f ra c t io n measurements and by Kashio et a l. (18) and Sweet and
White (19) through Raman and IR spectroscopy. The q u a n tita tive d is t
r ib u tio n can, however, d i f fe r in the glassy and the liq u id s ta te , as
was emphasized by Waseda (20). Comparison o f Raman spectra fo r both
glassy and so lid s ilic a te s (21 through 24) shows tha t the same s truc
tu ra l un its are to be expected in both so lid and glassy s il ic a te s .
A large number o f experimental measurements w ith Raman and IR spec
troscopy on glasses in s i l ic a te systems have been reported: Na2 0 -
The double tube arrangement was used in the measurements on the sys
tem Ca0 -Ca2 Fe2 0 3“ Fet 0 -Fe and an extensive descrip tion o f th is tech
nique is given in paper (V).
Id e a lly a so lid e le c tro ly te working as an oxygen sensor should exhi
b i t good oxygen anion conductiv ity and no e lec tron ic conduc tiv ity .
However, using so lid e lec tro ly tes based on z irco n ia , the e lec tron ic
conduc tiv ity , although small w ith in an oxygen p a rtia l pressure range -20of 10 atm through 1 atm a t 12 00 K, can d is tu rb the measurements
by causing an oxygen permeation through the e le c tro ly te . As the oxy
gen permeation rate is proportional to the d iffe rence in oxygen par-
38
t i a1 pressure between the two ce ll compartments i t can be minimized
by choosing a proper reference system. In the study o f the system
Ca0 -Ca2 Fe2 0 ^“ Fet 0 -Fe(V) a d iffe rence in oxygen p a rtia l pressure be
tween reference and sample mixture o f less than 1.7 In p ^ ) un its
could be achieved by using Fe-Fe^O as the reference m ixture.
Another app lica tion o f the emf technique is to measure the oxygen
p a rtia l pressure of e q u ilib r ia invo lv ing also sulphur, e.g.
The technique fo r these measurements has a t our department been app
lie d to e q u ilib r ia invo lv ing su lfides and su lfa tes in a number o f
systems, e.g. Cu-S-0 (70), Pb-S-0 (71), Ca-S-0 and Mn-S-0 (72) and
Zn-S-0 (73). Paper (IV) deals w ith the resu lts obtained in measure
ments o f the equ ilib rium
In these measurements the sample m ixture, Cu and C^S, is e q u ilib ra
ted w ith a continuous flow o f SO in argon, w ith a known p a rtia l
pressure o f S0 2 » while measuring the equ ilib rium oxygen p a rtia l
pressures. Application o f the law o f mass action to equ ilib rium
(V II) gives
Cu2Û(s) + S02 (g) + 0 2 (g) <=> CuO’ CuSO^s) (V)
CaS(s) + 2 0 2 (g) ♦* CaSO^(s) (VI)
2Cu(s,l) + S02 (g) »C u 2S (s ,l) + 02 (g) (V II)
In K(V II) = In p(02) - In p(S02) (11)
A ll sulphur-containing gas species, except SÛ2 (g ), are formed in
39
neg lig ib le amounts w ith in the temperature and oxygen p a rtia l pressu
re range studied, - In p ^ ) ~ 18-27, and the equ ilib rium p^C^) in
the c e ll can be considered the same as in the flow ing gas stream. An
extensive descrip tion o f the c e ll arrangement and c e ll operation is
given in paper (IV ).
Comments to resu lts
AfG° fo r C U g S (s ,l) . Reliable thermodynamic data fo r the sulphide
CU2S cons titu te a base fo r a l l equlibrium ca lcu la tions on conventio
nal pyrom etallurgical copper production. This includes simulations
o f both the conventional copper smelting and converting as discussed
in paper ( I ) and the roasting process as described in the study by
Eriksson (2). The A^G° values fo r C^S given in the lite ra tu re d i f
fe r by as much as 12 kJ. Most o f these data have been determined
using gas e q u ilib ra tio n , which generally gives data o f much lower
accuracy than the emf technique u t i l iz in g s ta b ilize d z ircon ia as so
l id e le c tro ly te . The values o f A^G°(Cu2S,s) determined in paper (IV)
are considered ce rta in to w ith in + 650 J.
Values o f A^G°(Cu2S ,1 ) were calculated using the AfG° values fo r so
l id CU2S together w ith data from the lite ra tu re on the enthalpy o f
fusion fo r Cu and Cu2S and a c t iv it ie s o f Cu(l) and Cu2S (l) in the
liq u id sulphide in equ ilib rium w ith liq u id m e ta llic copper. These
AfG0 values were considered ce rta in to w ith in 2500 J. The good
agreement between calculated and measured values at 1 i qui dus tempe
ratures is taken as an evidence fo r the r e l ia b i l i t y o f the Af G° va
lues fo r CU2S( 1 ).
40
The system CaQ-CagFegO^-Fe^Q-Fe. To ca lcu la te the phase re la tions at
m elting temperatures in the ca lc ia rich slag systems involved in
both iron and steel m etallurgy as well as in non-ferrous m e ta llu rg i
cal processes, a knowledge o f the subsolidus phase re la tions in the
system Ca-Fe-0 is most important. The data presented in the l i t e r a
ture fo r th is system is , however, contrad ictory in some aspects. The
dicalcium fe r r i t e , Ca2 Fe2Û^, ex ists in equ ilib rium w ith e ith e r o f
the s o lid solutions [Fet ,Ca]0 or [Ca,Fet ]0 , and iron up to a melting
temperature o f 1423 K according to re fs . (74 through 76), while ac
cording to re fs . (77 through 80) the phase re la tions s h if t to co
ex is ting so lid so lu tions [Fet ,Ca]0 and [Ca,Fet ]0 in equ ilib rium w ith
iron a t a temperature between 1308 and 1343 K. The figures fo r the
terminal so lid s o lu b i l i ty o f ca lc ia in w üstite vary from 12 to 37.5
mole per cent (74,75,77,80,81) a t 1300 K. To resolve some of these
con trad ic tions, a study o f the system Ca0 -Ca2 Fe2 0 5 -Fet 0 -Fe was ac
complished (V).
The experimental emf values obtained were used to determine a c t iv i
t ie s o f Fe O in the so lid solu tions [Fet ,Ca]0 and [Ca,Fet ]0, the
compositions o f the terminal so lid solu tions and the s ta b i l i t y o f
the dicalcium f e r r i t e , Ca2 Fe2 0 5 . The fo llow ing four-parameter
Rediich -K is te r equation fo r the in teg ra l excess Gibbs energy o f the
so lid so lu tions was found to be consistent w ith both measured a c t i
v it ie s and phase re la tions
G ^ / J ’ m o l ’ 1 = X | X 2 [ A + ( x -j- X 2 ) B + ( x ^ - X 2 ) 2 C + ( x 1 - X 2 ) 3 D] ( 1 2 )
where A = 2809 + 16.6205(T/K), B = 11268, C = 5038 and D = -7575. x ]
41
and x2 correspond to X(CaO) and X(Fet 0) in the solid solutions, re
spectively. Using the data set obtained in paper (V) the phase dia
gram fo r the pseudobinary section Fe^O-CaO in equilibrium with iron
at subsolidus temperatures was calculated.
3+ 2+In the model i t is assumed that the quotient Fe /Fe in Fe O in
equilibrium with iron is unchanged by the formation of a solid solu
tion with CaO. This may be a too sim plified description. However, no
determination of this quotient could be made because of the d i f f i
culty to prepare the solid solutions in advance. In the s im ilar solid3+ 2+solution [Fe^,Mg]0 the quotient Fe /Fe has been shown to decrease
on the addition of MgO to Fe O and reaches a value of p rac tica lly
zero at a mole fraction of Fe O equal to 0.3 (c f. Fig. 10). A decrease
of the quotient Fe^+/Fe^+ may influence the oxygen partia l pressure.
This cannot be accounted fo r without using a more elaborate method
than the one used in this study. However, th is effect of Fe3+ cannot
be quantified unless extensive measurements are performed at several
Fe^+ concentrations in the w lis tite -calc ia , [Fe-j_x ,Ca]0 solid solu
tion. Such measurements could be part of an extension of the study
in paper (V), with the aim of obtaining a consistent thermodynamic
description for the m etallurgically important systems Ca-Fe-0 and
Ca-Fe-0-Si02.
Application to copper smelting and converting
Conventional copper production processes can be divided into three
steps, namely smelting, slag blowing and copper blowing. The smelt
ing step, where a copper sulphide concentrate is smelted together
42
0.10
♦ 0.08
0.06o/
0 0.2 Q4 0.6 0.8 1.0X(FetO)
3+Fig. 10. The quotients n(Fe ) / [n (F e ^ ^ )+ n (M g ) ] p lo t te d against the mole
f r a c t io n o f w u s tite , X(Fe^O)^ in the s o l id so lu tion [Fe^M g]0 in e q u i l ib
rium w ith i r o n .
Simons (8 2 )y 1573 K
----------- Katßura and Kimora (83 )3 1433 K
O Ender (8 4 ) , 1573 K
□ Giddtngo and Gordon (85)
43
with a f lux ing agent, usually s i l i c a , is carried out in a reverbera-
to ry , e le c t r ic , f lash or b las t furnace. The matte phase formed is
t ransferred to a converter w ith in which the slag blowing and copper
blowing steps are performed. In the slag blowing step, fresh s i l i c a
is added and a i r or oxygen-enriched a i r is injected u n t i l the amount
of iron sulphide in the matte phase is about 1 wt %. The slag formed
is poured o f f and is normally e ithe r rec ircu la ted to the smelting
furnace or treated in a special furnace in order to recover part o f
the copper present in the slag. In jec t ion of a i r or oxygen-enriched
a i r is then continued, and the copper sulphide begins to form l iq u id
b l is te r copper in what is terming the copper blowing step.
The temperatures normally prevail ing during a l l three steps are
1473-1523 K.
A normal copper making process (c f. Fig. 11) is defined in paper ( I )
as taking place at 1523 K. The concentrate consists o f pure chalco-
p y r i te , CuFeS25 which contains 34.6 wt % Cu. The amount o f s i l i c a
added is 0.3 mol/mol CuFeS2 and a known amount o f a i r (p(0 2) = 0 . 2 1
atm) is in jected. The oxygen e f f ic iency is assumed to be 100 %, The
composition o f the matte phase at which the smelting step is in te r
rupted is an adjustable process parameter. The converting (slag b lo
wing and copper blowing) is regarded as one step with no intermedi
ate slag removal.
The equilibr ium composition o f the gas, matte and slag phases was
calculated as a function o f the amount o f oxygen in jected ( I ) . The
resu lts , which are given in Fig. 12, were used to deduce the overall
44 Ojlair) SiOj CuFeS2
ii ii ii
T*1523 K
Gas phase- N-, SO-, S-
Slag phase; CuO05<
FeO, FeOj 333* Si02
Matte phase- CuS05< FeS
T* 1523 K
O^lair) Si0 2 J jT"
11 -Il TGas phase: N2, S02 ,S2
Slag phase CuO05' FeO, FeOj 3 3 3 ^ ^ 2
Matte phase: CuSQ5' FeS
Copper phase
Smelting
step
Converting: Slag blowing and copper blowing
F ig . 11. P rin c ip a l sketch o f the ”normal" copper making process. For .
the gas phase3 only the predominant species are given. The copper phase
does not appear u n t i l the end o f converting.
Gas phase
„ - 2
Matte phase cuSqs10
FeS
Slag phase and solid phases
F » ( W . . ~
Fig . 12. Log(p^/atm) fo r gaseous species3 amount 3 n^/mol3 o f l iq u id and
s o lid species and wt % C ulatte VS a amoun ° f ° 2 ^ ^ supplied/m ol CuFeS2) fo r the nnormal11 copper making process. Gaseous species fo r which the
p a r t ia l pressure never exceeds 10 atm are om itted.
45
reactions taking place in pyrometallurgical copper production ( I ) .
Calculations were also performed to simulate the slag removal be
tween the smelting step and the converting and to study the outcome
of a change in such process parameters as temperature, s i l i c a con
tent in slag and pa r t ia l pressure o f oxygen in in jected gas. From
the results o f these la t te r calculations the fo llow ing conclusions
could be drawn:
1 ) the i n i t i a l smelting step should, i f possible, be continued to
matte grades o f at least 65 wt % Cu,
2 ) the copper losses to slag are minimized i f the s i l i c a content
in slag and the temperature are as low as possible,
3) low s i l i c a content in slag, low temperature and oxygen enrich
ment o f injected gas w i l l increase the Fe^+ content in slag.
An increased Fe^+ content w i l l , however, increase the r is k fo r mag
n e t i te formation, and thereby a d ra s t ic a l ly increased v iscos ity and
physical entrainment o f matte part ic les in the slag phase w i l l re
s u l t . A compromise s i tua t ion may ex is t in practice.
Fig. 13 shows calculated and observed copper losses to slag p lotted
vs wt % Cu (wt % CuSq 5 ) in the matte phase. The discrepancies can
be explained by e ithe r of the fo llow ing reasons:
1 ) su 1 phi di c copper losses to slag,
46
8
0 *6o>î%
2
__100
Fig . j g . C alcu lated and p ra c t ic a lly observed copper losses to slag vs
matte g rad e . This study3 from the ca lcu la tio n s on the nnormal" copper
making process; to ta l copper losses (s u lf id ic and o x id ic ) according
to Sehnalek and Im ris (8 6 ) ; to ta l copper losses according to
Nagamori (8 7 ); operating furnaces (86 and 88 through 9 1 ); b la s t
furnace; A > e ve rbe ra t o vy f urn ace ; V e le c t r ic furnace; Q Outukumpu
furnace; fla s h furnace; ^ Horan da furnace; Q Pierce-Smi th con
v e rte r.
47
2) equil ibrium conditions do not prevail in operating furnaces,
3) thermodynamic description is too s im p li f ied .
In Fig. 13 is also given a selection of experimentally obtained va
lues (86,87) of the to ta l chemical copper losses to slag, which can
be of two kinds, oxid ic or su lph id ic . Considering the d ifference be
tween these curves and the calculated curve, which corresponds to
the oxid ic s o lu b i l i t y , i t is evident that the su lphid ic s o lu b i l i t y
is small and cannot explain the extent o f the p ra c t ica l ly observed
copper losses, especially at matte grades greater than 60 wt % Cu.
A physical entrainment o f matte pa rt ic les into the slag in operating
furnaces is more plausible and th is may account fo r part o f the ob
served discrepancy.
In paper ( I ) the l iq u id phases were treated as non-ideal mixtures
consisting of mononuclear compounds, e.g. CuOq ^,Fe0,Fe0-j 3 3 3 and
SÌO2 in the slag phase. The non-idea lity was described by using the
simplest mathematical expressions possible. The sulphur and oxide
s o lu b i l i t y in the slag and matte phase respective ly, as well as the
influence of a l l minor elements was neglected. This thermodynamic
description is perhaps too s im p li f ied and may account fo r part of
the discrepancy between calculated and observed copper losses.
Non-equilibrium conditions
Another explanation o f the discrepancy in Fig. 13 may be non-equ il ib
rium conditions in operating furnaces.
48
Table 4 . Typical process analysis fo r the slag blowing step o f con