The mechanism and rate of reduction of
Mamatwan manganese ore fines by carbon
by W. D. GRIMSLEY*. B.Se. (Eng.) (Rand), M.Se. (Eng.) (Rand) (Student)
J. B. SEE*. B.Se (Hons), Ph.D. (New South Wales). M.I.M.. M.1.M.M. (Member),
R. P. KINGt, B.Se. (Eng), M.Se. (Eng.) (Rand), Ph.D. (Mane.). M.I.Chem.E.,
M.S.A.I.Chem.E.
and
(Fellow)
SYNOPSIS
Samples (15 to 20 g) of pre-calcined Mamatwan manganese ore or pure Mn 304 smaller than 48 mesh were reacted
at temperatures in the range 1000 to 1300°C with samples of spectrographically pure graphite smaller than 325
mesh. Samples (I g) of ore or pure Mn02 smaller than 46' mesh were reacted at temperatures of 1000 and 1200°C
with pure graphite fines in argon, with graphite in carbor, .'1onoxide, and with carbon monoxide alone. In each case,
the loss of mass as a function of time was recorded with a ,hermobalance.
The rate and degree of reduction were found to increase with additions of carbon and with increasing temperature
up to 1300 QC, when they reached their maximum. The shape of the reduction curves, together with the results
obtained from examination of the products of reaction by microscopy and X-ray diffraction, were used as the basis
for a postulated reduction mechanism. Some details are given of an initial kinetic model that was used to verify the
proposed mechanism.
It was found that the reduction of Mamatwan ore by graphite at temperatures in the range 1000 to 1300°C
occurs in two stages:
(i) an initial stage during which the higher oxides of manganese are rapidly reduced to MnO with concurrent
formation of CaMn 20 and
(ii) a longer stage during 4'which CaMn.O 4 and MnO are reduced to carbides by carbon dissolved in the ferromanganese phase.
SAMEVATTING
Monsters (15 tot 20 g) van voorafgekalsineerde Mamatwan-mangaanerts of suiwer Mn 304 kleiner as 48 maas is by
temperatuur in die bestek 1000 tot 1300°C laat reageer met monsters van spektrografies suiwer grafiet kleiner as
325 maas. Monsters (I g) van die erts ofsuiwer Mn02 kleiner as 48 maas is by temperatuur van 1000 en 1200°C in
argon laat reageer met suiwer fyngrafiet, in koolstofmonoksied met grafiet, en net met koolstofmonoksied. Die
massaverlies is in elke geval met 'n termobalans geregistreer.
Daar is gevind dat die tempo en mate van reduksie toeneem met die byvoeging van koolstof en 'n verhoging van
die temperatuur tot 1300°C waar dit 'n maksimum bereik het. Die vorm van die reduksiekrommes, tesame met die
resultate wat deur 'n ondersoek van die reaksieprodukte deur mikroskopie en X-straaldiffraksie verkry is, is as
grondslag vir 'n gepostuleerde reduksiemeganisme gebruik. Daar word besonderhede verstrek van 'n aanvanklike
kinetiese model wat gebruik is om die voorgestelde meganisme te verifieer.
Daar is gevind dat die reduksie van Marnatwan-erts deur arafiet in die temperatUurbestek van 1000 tot 1300°C in
twee stadiums plaasvind:
(i) 'n aanvanklike stadium waartydens die hoer mangaanoksiede vinnig tot MnO gereduseer word, met die gelyktydige vorming van CaMn 20 en
(ii) 'n langer stadium waartydens 4'CaMn 20 4 tot karbiede gereduseer word deur koolstof wat in die ferromangaanfase opgelos is.
Introduction
In South Africa, higher-grade manganese ores, such
as the Hotazel and Wessels deposits, are becoming
depleted and it is therefore important
for the large
reserves of lower-grade ores like Mamatwan manganese
ore to be utilized in the production of ferromanganese
and manganese
metal. Methods currently
in use by
South African producers of ferromanganese
and manganese metal may not be optimum for the processing of
lower-grade ores, and more efficient alternative
techniques should be sought. On'J such technique is the
prereduction
of manganese ore with carbon to yield
material having a suitable carbon content. This material
may then be melted in an open-arc furnace to produce
ferromanganese
and slag by a procedure similar to the
DIMECA process for the production of medium-carbon
ferrochromium1.
The investigation
reported here was undertaken
to
determine
the nature and extent
of the reactions
-
u_-
~------
*Pyrometallurgy
Research
Group (National
Institute
for Metallurgy), Johannesburg.
tDepartment
of Metallurgy,
University
of the Witwatersrand,
Johannesburg.
occurring during the reduction of Mamatwan manganese
ore by solid carbon. It was hoped that the information
obtained regarding the characteristics,
rate, and mechanism of reduction of Mamatwan ore would be useful in
the prediction of the reduction reactions that take place
in the upper portions of ferromanganese
furnaces and in
rotary kilns for the reduction of manganese-ore
fines
in the production of manganese metal by electrolysis.
The investigation
is part of a programme
within the
Pyrometallurgy
Research Group of the National Institute for Metallurgy on the reduction of Mamatwan
manganese ore for the production of manganese alloys.
The degree and mechanism of reduction of Mamatwan
manganese ore by graphite was determined
at temperatures ranging from 1000 to 1300°C, the reduction
being carried out in a resistance furnace in argon or
carbon monoxide atmospheres.
The change in mass of
the sample during reaction was measured with a thermobalance, and the products of reaction were examined
by X-ray diffraction and optical microscopy. A model
was then developed for prediction of the rate of reduction
of the manganese ore.
JOURNAL OF THE SOUTH AFRICAN INSTITUTE OF MINING AND METALLURGY
OCTOBER
1977
51
Previous
Stability
of Manganese
Work
MnO+10/7
The manganese oxides Mn02' Mn2O3' M~04' and MnO
have been studied extensively2-15,
and it has been
established
that their relative stabilities
depend on
temperature
and partial pressur~ of oxygen. T~ermal
decomposition takes place accordmg to the reactiOns
4 Mn°2 =2Mn2O3+02
and 6 Mn2O3=4Mn3O4+02
(I)
(2)
,
,
In air, reactions (I) and (2) occur between 400 and
700°0 and between 800 and 1l00002, 5-8,12-15 respectively. M~04 is the stable form of manganese oxide in
air from 1000 °0 up to its melting point of 1567 °05,
although there is a change in structure from tetragonal
M~04 (low hausmannite)
to the cubic spinel (high
hausmannite) at about 1160008, 16.
Possible
Oxides
Reactions
in the Reduction
of Manganese
(b)
=Mn+002
52
. . . . . . . . .
MnO+O
=Mn+OO
...,
MnO+Fe30=Mn+3Fe+00
Reduction to manganese carbide:
MnO+4/3 0
=1/3 Mn3O +00
MnO+4/3 Fe30 =1/3 Mn3O +4Fe
MnO+1O/70
OCTOBER
1977
M~03+30/7
Fe+OO
(9)
(10)
. .
(15)
The standard free energy change for reaction (9) is
positive over the temperature
range 25 ° to 2000 °0, and
it is generally accepted that gaseous reduction of MnO
does not take place3, 10, 12, 14, 15, 20-25. The standard
free energy changes for the reduction of MnO by carbon
or Fe30 to metal or carbide are almost identical in the
temperature
range from 800° to 1300°0. Over this
temperature
range, M~03 is the product most likely to
result.
Effect of Heat on Mamatwan
(ll)
Manganese
Ore
The major constituents
of the ore (for which the
analysis
is given
in
Table
I)
are
braunite
(3Mn2O3 . MnSi03),
calcite
(OaO03)'
dolomite
((Oa, Mg) 003), and hausmannite (Mn304)' minor amounts
of bixbyite ((Mn, Fe)203) and hematite (Fe203) being
present12,
Several possible reactions must be considered in the
proposal of a mechanism for the reduction of manganese
oxides in the presence of OaO, Si02, and iron oxides.
Data from Elliott and Gleiser17 and McGannon18 on the
standard
free energies of formation
of the various
manganese oxides, manganese carbides (Mn3O2' Mn703)'
cementite (Fe30), carbon monoxide, and carbon dioxide
were used for calculation of the standard free energy
changes for several possible reactions as a function of
temperature.
According to Kuo and Perssonl9, (Fe, Mn)30 and
(Fe, Mn)703 are the manganese carbides most likely to be
present at temperatures
ranging from 1000 to 1300 °0.
Identification of individual carbides by X-ray diffraction
analysis of reduced samples of ore was not possible,
since insufficient diffraction data were available and
there was overlapping of the diffraction peaks of the
phases. Reduction of Mn2O3 and Mn304 can take place
according to any combination of the reactions
3 Mn203+0
= 2Mn304+ 0O
. . (3)
(4)
3 Mn2O3+00 =2~04+002
"""
3 Mn203+Fe30=2Mn304+3Fe+00
(5)
and
M~04
+0
=3MnO
+00
. . . . . . . (6)
M~04
+00
=3MnO +0O2
, (7)
~04
+Fe30=3MnO
+3Fe+00
(8)
The standard free energy changes for these reactions
as a function of temperature
reveal that, above 700°0,
the reduction of Mn2O3 and ~04
by either Fe30 or
carbon is more thermodynamically
favourable than that
by carbon monoxide,
although reduction
by carbon
monoxide remains thermodynamically
feasible.
The final stage in the reduction process may be represented by one of the following reactions
(a) Reduction to metallic manganese:
MnO+OO
Fe30=1/7
. . . . . . . .
Oxides
26-29.
Pentz12 and De Villiers29 show that the following
mineralogical
changes occur during the heating of
Mamatwan ore in air:
700 °0 Decomposition of dolomite
800 ° to 900 °0 Decomposition of calcite with the formation of OaMn03, and possibly some
Oa(OH)2
900 °0 Beginning of the formation of OaMn2O4
and the decomposition of braunite
1000°0 Beginning ofthe formation ofjacobsite
(MnFe204) while braunite decomposes
to tetragonal Mn304 and Si02.
During heating in air from room temperature
to
1000°0 Mamatwan ore loses water and carbon dioxide
from the carbonatesl2. Pentz12 recommends preheating
of the ore to at least 800 °0 to promote decomposition of
the carbonates prior to any reduction studies.
Experimental
Materials
Mamatwan manganese ore (for which the analysis is
given in Table I) was crushed and ground to material
smaller than 48 mesh Tyler (see Table II for the screen
analysis). The ore was then calcined in argon at 1000°0
for 25 minutes. Carbon for the reduction experiments
TABLE
ANALYSIS
I
OF MAMATWAN
Constituent
MnO.
MnO
H.O
CO.
Fe.O.
MgO
AI.O.
SiO.
CaD
Total Mn
Total Fe
Mn/Fe ratio
Mass
ORE
%
29,2*
25,0
1,8
15,5
6,2
2,8
0,14
4,6
13,1
37,8
4,35
8,7
(12)
+00
. . . . . . . . (13)
=1/7 M~03+00
(14)
*Calculated from the difference
between the total manganese
in the ore and that reported as
being present as MuO.
.,JOURNAL OF THE SOUTH AFRICAN INSTITUTE OF MINING AND METALLURGY
TABLE IT
SCREEN ANALYSIS OF MAMATWAN ORE
Mesh size
(Tyler)
Particle
size
mm
Mass
0,295
0,208
0,147
0,104
0,074
65
<48>
< 65> 100
< 100> 150
< 150> 200
<200
%
57,15
81,01
65,23
30,21
298,54
TABLE
PROPORTIONS
OF ORE,
Preheated
ore
Mamatwan
AND GRAPHITE
~_u_----
Carbon
added
g
Comments
Mass
g
20,0
2,73
Stoichiometric
carbon (SC)*
16,0
17,5
Mn3O.
100,0
89,3
74,1
61,8
56,1
III
OXIDE,
Ore or oxide
ore
finer than
10,7
15,2
12,3
5,7
56,1
~
Mamatwan
Cumulative
per cent
Mass
g
MnO.
0,8 to 8,0
0,33 to 3,0 SC
3,6
SC
0,14
SC
~
*Stoichiometric carbon (SC) is the amount of carbon required to
reduce the manganese and iron oxides to their metals.
was spectrographically
pure graphite having an impurities content of less than 10 p.p.m., and of size
grading 100 per cent passing 325 mesh.
The ore, oxide, and graphite were blended carefully in
the proportions indicated in Table Ill.
Design
and Calibration
of Apparatus
Large-scale experiments
Experiments were carried out on samples of approximately 20 g of ore and reducing agent. The apparatus
consisted of a vertical molybdenum-wound
resistance
furnace
with a top-pan
Mettler
electromechanical
balance, Model PE 162, mounted below the furnace. The
thermobalance
had a maximum deflection of 150 g and a
sensitivity of 10 mg.
An inert atmosphere was maintained in the furnace by
passing argon at 1 Ifmin through the thermobalance
box
and into the furnace tube. The furnace temperature
was
monitored by a Pt-6 %Rhfpt-30 %Rh thermocouple and
controlled by a Eurotherm
Thyristor controller unit,
type 070. The hot zone of the furnace was 55 mm in
length, and could be maintained at within 5 °C of the
desired temperature.
The furnace temperature
and the
mass of the crucible and its contents were recorded
continuously on a 2-pen Rikadenki chart recorder.
The samples were weighed before and after reaction on
a second Mettler balance, Model P1200, which had a
maximum deflection of 1200 g and a sensitivity of 10 mg,
as a check on the loss in mass measured with the thermobalance. In all cases, the error was less than 50 mg for a
total loss in mass of more than 3 g, i.e. less than 2 per
cent.
Small-scale experiments
Experiments
in which atmospheres
.JOURNAL
Of
l"H.E ~OUTH
AFRICAN
of argon,
1N.~TIl"UTE
carbon
OF MINING
monoxide, and air were used, were carried out with a
Stanton
Thermogravimetric
Analyser,
Model HT-M,
type no. SIO0-1-92. The thermobalance
had a total
capacity of 100 g and a sensitivity of 1 mg. This system
had the advantages
of increased sensitivity and more
rapid heating of the smaller sample. The furnace temperature was controlled by a PtfPt-13 %Rh thermocouple
placed between the furnace winding and the mullite
work tube at the centre of the hot zone of the furnace,
which was 25 mm in length. The electromechanical
balance was standardized
against a Mettler HIOW
single-pan balance having a maximum
deflection of
100 g and a sensitivity of 0,1 mg. The error in the Stanton
t,hermobalance was less than 3 mg for a total loss in mass
of 200 to 250 mg, i.e., less than 2 per cent. The furnace
temperature
and sample mass were recorded
cont,inuously on the 2-pen chart recorder incorporated in the
equipment.
Experimental
Procedure
jJ1easurement of heating rate for the sample
The average heating rates to 1000 or 12oo°C for the
1 g samples and to 1000, 1100, 1200, or 1300 °C for
the 20 g samples were measured by the rapid raising
of samples of ore into the hot zone of the furnace concerned.
A Pt-6 %RhfPt-30 %Rh thermocouple
was
lowered into the centre of the sample, and temperature
readings were taken every 30 seconds for the 1 g samples
and every 60 seconds for the 20 g samples. The measurement of sample heating rates was necessary to determine
the effect of this initial period of heating on the initial
rate of reduction.
Reduction experiments
Samples of carbon and ore or preheated ore in alumina
crucibles were rapidly introduced into the hot zone of the
furnace and reacted on the appropriate
thermobalance
in an atmosphere of argon or carbon monoxide for up to
two hours.
The change in mass was continuously recorded on a
chart recorder, and at the end of the run the reduced
samples were rapidly removed from the hot zone of the
furnace and cooled in air.
Chemical analysis
The total manganese
content,
as determined
by
potentiometric
titration, was within 1 per cent of the
contained
manganese30. The total iron content was
determined to within 2 per cent of the amount present
by an X-ray-fluorescence
technique
developed
by
Austen et al.3I.
The final carbon content of the samples of reduced ore
and oxide was determined with a Leco carbon analyser
that has a precision of approximately
0,1 per cent. This
determination
was necessary as a check on the calculations for the degree of reduction of the samples of ore
and oxide.
X-ray diffraction analysis
X-ray diffraction was used for the detection of changes
in the mineralogy
of the samples of ore and oxide
at various
stages in the
reaction
with carbon.
Cooling in air of samples of reduced ore or oxide was too
slow to prevent the reaction products from undergoing
t,ransformation of the phases. However, X-ray diffraction
AND
MI;TAI.,LURGY
OCTOBER
11177
53
.
In!:reaslng
.
eraunitCMn:<o.
Mn;O. c~::O.
qualitative'
concentration.
(curve B) .
r
:0
""-
CaMnO3 CaMnO:1 CaMn1O. MnO
CaMn~04 braunite
MnJO. CaMnzO.
MnO
CaMn1O.
35
30
/
- - - - _S.E~I£. ter:!2p~ratl!!e-- 1000(,)
--
0
I
I
25
900
,;
~
800
~
I
cfl
20
..
::J
I
c:
~
\1)
I
0
-()
,
'"0
CD
700 G)
E
600 \1)
I
15
::J
A
5 per cent(by
B 10 per cent (by
C 15 per cent (by
D 20 per cent (by
I
10
,
0::
5
mass)
graphite
mass) graphite
mass) graphite
m<Jss)graphite
0..
E
500 c
'
,
Cl)
,
0
400
0
50 60
40
30
20
10
Time,
Fig.
I-The
reaction
16 g of preheated
of
ture
Increasing
qualitative
concentration
(curve C)
with
time
as well
MnO
CaMnzO.
Mn3O.
Carbide
Braunite
Mn3O.
r
Mamatwan
is shown,
CaMnOJ
CaMn1O.
ore
as the
with
graphite
mineralogical
70
80
90
100 110 120
min
in argon
changes
at
IOOO°C.
occurring
The
during
change
the
tempera.
Carbide
MnO/CaMn:O.
Carbide
MnO/CaMnzO.
Carbide
in sample
reaction
MnO/CaMn."O.
90
Sample
,,-
80
I
I
70
temperature
---------
1300
1200
I
(,)
I
60
I
(
cfl
..
: c:
50
()
0
110'0
~\1)
I
~-
0
.-
B
1000;
-c
(
40
;
::s
,
"0
\1)
,
900
A
0..
E
30
~a::
800
\1)
I
A
,
20
B
I
10
C
D
,'
5 per cent
(by
mass)
graphite
10 per cent (by mass) graphite
15 per cent (by mass) graphite
20 per cent (by mass) graphite.
0
0
Fig. 2
54
The reaction
temperature
OCTOBER
1977
10
20
30
40
50 60
70
Time, min
80
90
100
700
Cl.
E
c
Cl)
600
500
110 120
of 16 g of preheated
Mamatwan
ore with graphite in argon at IJOO°C. The change in sample
with time is shown, as well as the mineralogical
changes occurring during the reaction
JOURNAL OF THE SOUTH AFRICAN INSTITUTE OF MINING AND METALLURGY
analyses of a sample of reduced manganese ore quenched
in water at a cooling rate of approximately
120°C per
second revealed that cooling was still too slow to prevent
the occurrence of very rapid transformations
of the
phases.
Results
Reduction
with
Solid Carbon
The effects of variations in graphite additiof1.s on the
percentage reduction of 16 g of preheated lV!amatwan
ore at 1000 and 1300 °C are shown in Figs. I and 2
respectively. (The percentage reduction is defined as the
percentage removal of the available oxygen, Le., the
oxygen present in the ore asEociated with manganese
and iron). The percentage reduction with time increased
markedly as the temperature
increased, and the extent
of reduction for a given addition of graphite increased
markedly with temperature,
as was also shown by other
runs at 1l00 and 1300°C.
Graphite addi~ions above 10 per cent (by mass) had
very little effect on the extent of reduction at 1000 and
1l00°C, but the effect of increased graphite additions
became more important
at 1200 DC, and even more
marked at 1300°C.
In Fig. 3 the rates of reduction at temperatures
of
1000, 1l00, 1200, and 1300°C are shown for a graphite
addition of 15 per cent (by mass) of 16 g of preheated
Mamatwan
ore. The amount of graphite added was
slightly less than the stoichiometric
requirement
of
16,51 per cent (by mass). Increase of the temperature
from 1000 to 1l00°C had little effect on the rate and
degree of reduction,
but when the temperature
was
increased to 1200° to 1300°C the effect was much greater.
From Figs. I to 3 it can be seen that the reduction
proceeded in two steps: a relatively short initial stage
during which the rate and degree of reduction were
affected only slightly by the carbon added and by
temperature,
and a second, longer stage in which the
effects of increases in either the carbon addition or the
temperature
were more marked. The first stage took
from 10 to 20 minutes depending on the reaction temperature, whereas the second stage continued for up to
100 minutes.
X-ray diffraction analysis of quenched samples was
used for deductioIl of the mineralogical changes ,occurring
during the reduction of preheated Ma,ma,twan ore. The
results of this study are superimposed on the curves for
reduction versus time at 1000 and 1300°C in Figs. I and
2 respectively. At 1000 DC, the reaction was limited to
the reduction of the higher oxides of manganese down
to MnO, whereas, at 1300°C, a high degree of metallization was obtained.
Determination
of the relative
proportion<; of MnO and CaMn204 in samples of reduced
ore presented some difficulties owing to overlapping of
the major X-ray diffraction peaks of these two phases.
Protlems were also experienced with identification of
the phases collectively la1::elled 'carbide' in Fig. 2. For
easier identification of the phases, 17,5 g of pure MIlaO4
was reacted with 20 per cent (by mass) graphite in
argon at 1300 DC, and the reaction products were examined by X-ray diffraction. The results of the X-ray
diffraction analyses and the graph of reduction versus
time for MIlaO4 indicated that the short initial stage of
the reduction correspond to the reduction of MfiaO4 to
MnO, whereas the longer second stage involved the
formation of manganese carbide by the reduction of
MnO with carbon and the gradual disappearance
of
MnaO4'
Reduction
Monoxide
with
Solid Carbon
and Carbon
Several experiments were carried out with Mamatwan
ore and with pure MnO2 for determination
of the role
80
'"'
70
60
c
;;!
50
...
c
~
-U
:J
'0
Cl)
B
A
40
30
0::
A 1000.C'
B '.,1100.C .
C 1200.C,
D 1300.C .
20
10
0
0
10
20
30
40
50
60 70
80
Time, min
90
100
110 120
Fig. 3- The reaction of 16 g of preheated
Mamatwan
ore with 15 per cent (by mass) graphite
in argon, The heating curves have been omitted for clarity
JOURNAL OF THE SOUTH AFRICAN INSTITUTE OF MINING AND METALLURGY
OCTOBER
1977
55
,
1100
()
~::J
1000'0
~Cl)
0E
20
Cl)
A 14 per cent (by mass) graphite In CO
B 14 per cent (by mass) graphite in argon
C In CO without graphite
~
10
900
~Cl)
0-
800
0
10
0
0
Cl)
I
I
-0
0
I
30
::J
A
I
I
I
~
0
1200
I
40
,;
B
Sample temperature
,.- - - - - - -C - - I
50
20
30
40
50
60
70
80
90
~en
700
100
Time I min
Fig. 4-The
reduction
of I g of pure
MnO.
at 1200°C.tThe
changeIin
temperature
of the sample
with timeli(also
shown
60
Sample
temperature
50 ,,- - - - -
1200
I
I
~
..
0
0
I
I
40
C
1100
C
::J
I
-
.2
... 30 I
g
,I
"0
Cl)
I
1000 ~
Cl)
.
~
900
20
:
I
I
10
I
A 16 per cent (by mass) graphite in CO
B 16 per cent (by mass) graphite in argon
C In CO without graphite
0
20
30
40
50
Time
Fig. S-The
56
OCTOBER
1977
reduction
.
I
I
10
0
..
Cl)
~
60
1
70
80
90
800
0E
Cl)
Cl)
0E
~
700
100
min
of I g of preheated
Mamatwan
ore at 1200°C
JOURNAL OF THE SOUTH AFRICAN INSTITUTE OF MINING AND METALLURGY
played
by carbon monoxide
in the reduction
of
Mamatwan ore with carbon.
The reduction
of pure MnO2 at 1200 °C for three
different sets of reducing conditions is shown in Fig. 4.
Almost complete reduction to MnO was achieved within
the first 10 minutes when MnO2 was reacted with graphite
in a carbon monoxide atmosphere,
with graphite in
argon, and with carbon monoxide alone. Further reduction of MnO occurred only with graphite in argon.
Fig. 5 illustrates the variation in percentage reduction
at 1200°C of preheated
Mamatwan
ore reacted with
carbon monoxide alone, and with carbon in atmospheres
of argon and carbon monoxide respectively. The overall
degree of reduction achieved with carbon monoxide
alone was well below complete reduction to MnO, which
suggests that CaMn2O4 is not easily reduced by carbon
monoxide. Reduction of preheated ore by graphite in an
atmosphere
of carbon monoxide
proceeded
rapidly
until the iron oxide was reduced to metal and the
manganese oxides were reduced to MnO. The reaction
then ceased abruptly, and no further reduction of MnO
to a manganese carbide occurred.
The reaction of 1 g of preheated ore with graphite in
argon gave a result similar to that for the 16 g sample
reduced with 15 per cent (by mass) graphite in argon at
1200°C (Fig. 3) although the heating rates for these two
samples differed considerably
as a result of their
difference in mass. This result indicates that the heating
rate was not a critical factor in determining the shape of
the curves for reduction.
Plate I-Mamatwan
ore heated to IOOO°C(or 30 minutes. The light areas are braunite and
Mn 30 .. the dark areas decomposed carbonates and gangue. (Magnification 250X)
Plate IJ-Mamatwan
ore reacted with 15 per cent (by mass) graphite at 1300°C (or 40
minutes. Note the random (ormation of white metallized
regions. (Magnification 250X)
.JOURNAl,.
OF THE SOUTH
AFRICAN
INSTITUTE
OF MINING AND
METALLURGY
OCTOBER
1977
57
Microscopic
Examination
Reduced MnOs
of Reduced
Ore
and
Details of the technique for mounting of the samples
are given by Grimsley32. The stages in the reduction of
preheated
Mamatwan
ore by graphite at 1300 QC in
argon are shown in Plates I to V. Plate I illustrates an ore
particle after heating in air to 1000 DCfor 30 minutes to
decompose the calcium and magnesium carbonates. The
dark areas are the decomposed carbonate and the gangue
minerals; the light areas are the bra unite and Mn304.
Intermediate
stages in the reaction between ore and
carbon are shown in Plates 11 to IV. The white areas
indicate metal; the darker areas are the unreduced MnO
or CaMn2O4' In Plate 11, which illustrates the stage of
58
OCTOBER
the reaction at which metal was first produced, random
nucleation can be seen of the metal throughout the ore.
The next stage in the production
of metal, which is
shown in Plate Ill, occurred as the result of growth of
the nuclei shown in Plate 11. However, this mechanism is
contradicted by the 'shrinking-core'
reaction illustrated
in Plate IV. Reduced particles similar to that shown in
Plate IV were not as common as those resulting from
random nucleation, and therefore it was considered that
the 'shrinking-core'
mechanism
makes an insignificant
contribution to the overall rate of reduction.
The final stage in the reduction
of preheated
Mamatwan ore by carbon is complete reduction of the
oxide particle to metal or carbide, as shown in Plate V.
Plate III-Mamatwan
minutes. The metallic
ore reacted with 15 per cent (by mass)
areas have increased in size as the reaction
2S0X)
Plate
ore reacted with 15 per cent (by mass) graphite
Note 'shrinking-core'
appearance.
(Magnification
1977
IV-Mamatwan
minutes.
graphite at 1300°C for 60
increased. (Magnification
at 1300DC for 60
2S0X)
JOURNAL OF THE SOUTH AFRICAN INSTITUTE OF MINING AND METALLURGY
Plate V-Mamatwan
ore reacted with 15 per cent (by mass) graphite at 1300QC for 120
minutes. The particle has been etched with I per cent nitric acid in ethanol to show
carbide structure. (Magnification 250X)
The particle has been etched with 1 per cent nitric acid
in ethanol to reveal a structure similar to that of carbide.
For easier interpretation
of the microscopic examination of reduced ore samples, the stages in the reduction
of pure Mn3O4 by carbon in argon were also examined32.
There was no evidence of the random nucleation observed
in samples of reduced ore, and, as the reaction progressed,
there was continued growth of the metallized areas
formed initially rather than the formation
of new
metallic
areas. The distribution
of metallic
areas
exhibited by particles of reduced ore may be due to the
finely intergrowth
nature of the ore12, 29 and to the
dispersion of the iron oxide and the manganese oxide
throughout the ore matrix.
The final product of the reduction of Mn304 by carbon
in argon at 1300 QC was a carbide similar to that illustrated in Plate V. Microscopic examination and X-ray
diffraction analysis indicated that Mn02 is reduced by
carbon in argon in the sequence Mn02-Mna04-manganese carbide.
Discussion
The Mechanism
by Carbon
of Reduction
of Mamatwan
Ore
Figs. 1 to 3 indicate that the reduction of preheated
Mamatwan
ore by carbon, at temperatures
ranging
from 1000° to 1300°C, occurs in two stages:
(i) an initial stage during which most of the higher
oxides of manganese are rapidly reduced to MnO
with concurrent formation of CaMn204 and iron
oxide is reduced to metal or carbide, and
(ii) a second, longer stage during which MnO and the
remaining CaMn2O4 are reduced to metal or carbide.
The first stage of the reaction was complete within the
first 10 to 20 minutes at all the temperatures
and for all
the carbon additions studied. During this period, the
braunite and much of the Mn3O4 were reduced to MnO
JOURNAL
OF THE SOUTH
AFRICAN
INSTITUTE
OF MINING
with concurrent formation of CaMn2O4" Fig. 1 shows
that bra unite was reduced rapidly at 1O00°C and that
Mn3O4 was reduced mainly towards the end of t,he
initial stage, both minerals being reduced to MnO. It was
more difficult for the first stage of the reaction to be
followed in detail at 1300 °C since the rate of reaction
at this temperature
was more rapid. However, it was
established that MnO was the major product of reduction
in this stage. At 1000 and at 1l00°C, reduction was
limited mainly to the first stage, since CaMn204 and
MnO were not easily reduced at these temperatures.
The second stage of the reaction was virtually absent
below 1200°C, but, at 1200 and 1300°C, the reaction
continued for up to two hours, depending on the carbon
content. During this stage, the remaining CaMn2O4 and
MnO were reduced to metal or carbide.
Pure Mna04 was reacted with carbon at 1300°C, and
the reaction products were compared with those obtained
from the reduction of ore. For the reduction of pure
Mn304' all the manganese oxide was reduced to MnO in
60 minutes at 1300°C, whereas CaMn2O4 appeared to
persist throughout the reaction of the ore with carbon.
From the work of Riboud and Muan33, it appears that
CaMn204 may be Mn304 (Mn2+Mn23+04) with some of the
Mn2+ ions replaced by Ca2+ ions. The enhanced stability
of CaMn204 cannot be readily explained
since the
structure of CaMn2O4 has not been studied in detail and,
at present, the exact mechanism of reduction of this
phase is not known34.
The Role of Carbon
Monoxide
At temperatures
above 700 QC, the reduction of Mn203
and Mn304 by solid carbon was more favourable
thermodynamically
than that by carbon monoxide.
However, these oxides are reduced to MnO by carbon
monoxide3, 10, 12,14, 15, 20-25, and Dressel and KenAND
METALLURGY
OCTOBER
1977
59
worthy3 showed that braunite is easily reduced to MnO
with carbon monoxide at 1000°0.
From Fig. 4, it can be seen that the rate of reduction
of pure Mn02 by solid carbon at 1200°0 was initially
greater than that by carbon monoxide. However, the
rate of reduction by solid carbon decreased rapidly, and
complete conversion to MnO occurred more quickly
with carbon monoxide. The fact that Mn02 was not
reduced beyond MnO by carbon monoxide is in agreement with the thermodynamic
calculations and previous
investigations.
At 1200°0, carbon monoxide rapidly reduced braunite
and Mns04 to MnO, as shown in Fig. 5. The reduction
products were MnO and OaMn2O4'
A product likely to result from the reduction of MnO
by solid carbon is ~03'
The possible reactions are
.,
(14)
MnO+IO/7 0=1/7 ~03+00
and MnO+IO/7 Fe30=1/7 ~03+30/7
Fe+OO
. (15)
At 1200°0, the equilibrium partial pressures of carbon
monoxide for these reactions are 31,4 and 18,2 kPa
(0,31 and 0,18 atm) respectively,
and, as the partial
pressure of carbon monoxide for curve A in Fig. 5 was
83,1 kPa (0,82 atm-atmospheric
pressure in Johannesburg) the reaction ceased at the formation of MnO.
Reduction by carbon monoxide is therefore limited
to the first stage of the reaction, and the reaction
sequence is
Mns04+0
=3MnO+00
Mna04+00=3MnO+002
. . . . . .
. . . . . . . . . .
(6)
(7)
002
+0=200
(16)
Reduction
is initiated
by reaction (6), and subsequently perpetuated
by reactions (7) and (16) until all
the manganese
oxide has been converted
to MnO.
This reaction sequence explains the higher initial rate of
reaction
of Mamatwan
ore with activated
char as
compared with that with graphite32 since activated char
has greater reactivity to carbon dioxide by reaction (16),
which is considered to be the rate-determining
step in
reaction mechanisms of this kind35.
The Role of Metallic Iron and Man~anese
Microscopic examination
of samples of Mna04 after
reaction with graphite indicated that the final stage of
the reaction involved reduction of MnO by carbide or by
carbon dissolved in manganese
metal. Particles
of
reduced ore showed an apparently different mechanism
in which nucleation
of the metal appeared
to be
important and to have occurred at random. This random
nucleation could have been caused by the dispersal of
iron and manganese throughout
the ore matrix in the
finely intergrown ore29, 34. Once metallic particles had
formed in a random distribution,
they increased in size
with time, since they acted as a medium through which
carbon could be transferred to the reaction interface.
Mathematical
Analysis
of the Reaction-rate
Data
For verification of the proposed mechanism, a mathematical model of the process was developed32. Qualitative
observations
of the reduction
of Mamatwan
ore by
carbon indicate that the reaction proceeds in two stages,
and that, during reduction, a clear interface develops
between the phase cont~Wng
iron and manganese
"
O~T96~R
1911
carbides (and possibly some metal) and the phase containing MnO. No such interface was observed for the
reduction of Mns°4 to MnO. In the development of the
model, it was assumed that the considerable amount of
braunite (Mn2O3) in the ore had been instantaneously
reduced to Mna04' This assumption is justified by the
large negative standard
free energy change for the
reaction of carbon with Mn2O3 over the temperature
range considered. Previous studies12 have also indicated
the amenability of this oxide to reduction.
The very rapid rate of reduction for Mn203 above
1000°0 and the absence of thermodynamic
data for
OaMn2O4 made it necessary for the model to be constructed on the basis of the reduction of Mna04 to MnO,
followed by the reduction of MnO to carbide. Also, in
the experiments carried out, the uncertainty
regarding
the actual temperature
of the solids during the very
early stages of reduction did not justify any elaboration
of the model.
Hence, the reactions considered in the model are
(3)
Initial
3Mn20a+0
=2Mns°4
+00
(7)
Reaction (1) Mns04 +00
=3MnO
+002
Reaction (2) MnO
+10/70=1/7
Mn7Oa+00
(14)
(6)
Reaction (3) Mna04 +0
=3MnO
+00
Equilibrium
conditions must be considered in the
development of the model, since, at temperatures
from
1000 to 1300 °0, the equilibrium constant for reaction (2)
is close to 1.
The model proposed is based on semi-infinite linear
geometry, and is shown in Fig. 6. It is assumed that the
system exhibits quasi-stationary
behaviour,
and that
the diffusivities of carbon monoxide and carbon dioxide
in Mna04 are equal. The detailed development
of this
model is given elsewherea2.
An example of the comparison between the predictions
of the mathematical
model and the actual reduction
data is given in Fig. 7, in which the data from Fig. 3 are
reproduced. Fig. 7 illustrates the dependence on temperature of the model for a fixed addition of carbon.
For all cases in Stage 1, the model fits the experimental
data very well except during the first few minutes of the
reaction. The init,ial effect of temperature,
as shown by
the experimental
curves, is not apparent for the theoretical curves, which are based on isothermal behaviour.
For this reason, the theoretical
curves do not pass
through the origin of the axes. The change in direction
of the curves from Stage 1 to Stage 2 is sharper on the
theoretical
curves since, in practice, not all of the
particles reach the end of Stage 1 at the same moment.
Although, as shown in Fig. 7, the theoretically derived
reduction curves fit the experimental curves fairly well,
there are deviations for which there may be several
explanations. A major cause of these differences may be
the complex mineralogy of the ore. In this respect,
several basic assumptions
were made regarding
the
reduction of OaMn204 and iron oxides, the effective
diffusivities of carbon monoxide and carbon dioxide,
and the particle-size distribution
of the ore, for which
determination
was particularly
poor, especially in the
small sizes, as can be seen from Table H.
In an attempt
to ascertain
the importance
of
mineralogical factors, the predictions of the model at
!.IOURNAL OF THI; ~QI,ITI1
AFRICAN
IN$TITUTE
OF MI~IN~
AND
fv'IETALLURGY
Stage 1
X~
CO2
X=O
MnaO. + CO
= 3MnO
+ CO2 (in the oxide)
~Mn304
Carbon
I
CO
t
MnaO. + C
= 3MnO
+ CO (at the Interface)
Stage 2
X=x
X~
Carbon
~
CO2
MnaO. + CO
= 3MnO
+ CO, (In the MnaO.)
Mn304
.
co
~,
~
.
MnO + 10/7C = 1nMn1Ca + CO (at the MnO-carbide.
Interface)
Fig. 6-Model
for the reduction
of Mamatwan
.
manganese
ore by carbon
0
60
70
60
c
"$
..
c
50
0
-
CJ
:J
'0Cl)
40
.
B
. . 'A.'
A
B
C
D'
1000.C"
1100.C'
1200.C'
1300.C .
................................................
Q: 30
'h
20
Theoretical
Experimentul
10
0
0
10
20
30
40
50
60 70
0
Timet
min
90
100
110 120
Fig. 7-A comparison between t~, ,xperimental results and predicted behaviour for the
reduction of prel'leated Marpll,fi'tp ore with 15 per cent (by mass) graphite in argon
i,~ .
1300°C were compared with the r~q.lJction q~~Q. for
pure M~O4 at this tempera~~re32. A~ pne reslliWf,j:f this
?ompanson were not ~s posItIve as h/.ll4 peen h'~p~~, the
Importance of the mIneralogy of tllrQfe h~~fiQ~ been
clearly established.
.
Variations in particle size and the nqp-uniform~~y of
particles probably also contributed to ~he devlaticm,s of
the model from the experimental
data; The modflf was
originally developed on the premise of infinite l\pear
geometry, and the particle-size distribu:tion was superimposed on the model at a later stag$: so that finite
limitations on particle size could be taken into account.
This may account for the lesser degree of curvature
~QIJRNAl
OF THE SOIJTI1 AFRlgAN
INST'T~TE
exhibited by the theoretical curves during St,~~.
For the model, it was assumed that the carffi'i\')p- the
metallized layer was uniformly distributed at'itlLtiimes.
Hence there is no apparent depletion of carbon at the
manganese carbide-MnO
interface.
However,
if in
practice the diffusion of more carbon through the metal
to the interface were relatively slow, it would have an
effect on the overall rate of reaction.
In spite of its various limitations, particularly in the
evaluation of model parameters32, it appears that the
model can be of value in the formulation of an overall
reaction mechanism
for the reduction
of Mamatwan
manganese ore by carbon.
Of MIN!NG ANt> METAllURG~
9GTOE$~R
19?1
(j1
Summary
The reduction of Mamatwan manganese ore by carbon
at temperatures
in the range 1000 to 1300°C occurs in
two stages:
(i) reduction of the higher oxides of manganese to
MnO by carbon and carbon monoxide with concurrent formation of CaMn204' and
(ii) reduction of CaMn204 and MnO to carbide by carbon
dissolved in the ferromanganese
phase.
At 1000 and 1l00°C, very little reaction occurs during
:the second stage, and the products of reaction arc
'CaMn2O4 and MnO. At 1200°C, the second stage account,s
for up to one-half of the total reduction;
at 1300°C,
if sufficient carbon is present, most of the reduction
occurs during the second stage, the product probably
being (Fe, Mn)7C3,
The rate and overall degree of reduction at 1300°C
increase as the carbon addition is increased up to 30 per
cent (by mass) or 2,4 times the stoichiometric
amount
of carbon required to reduce the iron and manganese
oxides to their metals. Further increases in the amount
of carbon added have no apparent effect on the rate and
overall degree of reduction.
Details and an evaluation are given of a mathematical
model to fit the experimental reduction data.
Acknowled~ements
This paper is published by permission of the Director
General of the National Institute for Metallurgy. Grateful acknowledgements
are due to the Ferro Alloy
Producers' Association of South Africa, which supplies a
grant towards the work on ferro-alloys of the Pyrometallurgy
Research Group of the Institute,
Mr P.
Jenner for his contribution on the 1 g samples, Dr J. P. R.
de Villiers, with whom helpful discussions were held on
the mineralogy and structural
changes of Mamatwan
manganese ore, and Dr P. R. Jochens and Professor
D. D. Howat for their suggestions during the initial
stages of the investigation.
References
1. OSSIN, D. I., and JOCHENS, P. R. The production of mediumcarbon ferrochromium
alloys by solid-state
reduction.
Johannesburg,
National Institute for Metallurgy, Report no.
1650. Aug. 1974.
2. MuAN, A. Phase equilibria in the system manganese oxideSiO. in air. Am. J. Sci., vo!. 257. Apr. 1959. pp. 297-315.
3. DRESSEL, W. M., and KENWORTHY,H. Thermal behaviour of
manganese minerals in controlled atmospheres. U.S. Bureau
of Mines, Report of Investigation no. 5761, 1961. pp. 4-15.
4. COUGHLIN, J. P. Contributions
to the data on theoretical
m!!paU1:1rgy. XII. Heats and free energies of formation of
inorganic oxides. U.S. Bureau of Mines Bull. no. 542. 1954.
5. HAHN;' W. C., Jr., and MuAN, A. Studies in the system
Mn-D. The Mn.O,-Mn,O.
and Mn,O.-MnO
equilibria.
Am. J. Sci. vo!. 258, no. 1. 1960. pp. 66-78.
6- VOLKERT, G., and FRANK, K. D. (ed.). Metallurgie der
ferrolegierungen.
2nd edition. Berlin, Springer-Verlag,
1972.
pp. 400-412. (National Institute for Metallurgy, Johannesburg, NIM TR 516).
7. ABDUL AZIM, A. A., KOLTA, G. A., and ASKAR, M. H.
Thermal behaviour of some artificial manganese dioxides.
Electrochim. Acta, vo!. 17. 1972. pp. 291-302.
8. DOLLIMORE, D., and ToNGE, K. H. Thermo-analytical
behaviour of some synthetic manganese oxides. Proceedings
3rd ICTA, Davos. Therm. Anal., vo!. 2. 1971. pp. 91-103.
9. DAVIES, M. W., and RICHARDSON, F. D. Non-stoichiometry
of manganous
oxide. Trans. Farad. Soc., vo!. 55. 1959.
pp. 604-610.
62
OCTOBER
1977
10. ISIHARA, T., and KOGOSHI, A. Fundamental
researches
on
metallurgical
treatment
of manganese
ores. I. On the equilibrium
in the reduction
of Mn,O. with carbon monoxide.
Sci. Rept. Res. Inst. Tohoku
Univ., ser. A, vo!. 5. 1953.
pp. 172-178.
11. KLINGSBERG, C., and Roy, R. Solid-solid and solid-vapour
reactions
and a new phase in the system
Mn-O. J. Am.
Ceram. Soc., vo!. 43, no. 12. 1960. pp. 620-626.
12. PENTZ, R. D. 'n Ondersoek
na die veredeling
van laegraadse
mangaanerts
met besondere
verwysing
na die reduseerbaarheid. M.Sc. Dissertation,
Potchefstroom
University,
1970.
13. DASGUPTA, D. R. Oriented
transformation
of manganite
during
heat
treatment.
Geo!.
Survey,
Calcutta,
India.
Mineral.
Mag., vo!. 35. no. 269. 1965. pp. 131-139.
14. SHANTARIN, V. D., SHESTAKOVSKII, 0. F., SUCHILNIKOV, S. I.,
PAVLOV, V. A., KUCHER, A. G., and KHITRIK, S. Kinetics
of
the reduction
of manganese
concentrates
with solid carbon.
Communication
1. Steel in the USSR, Oct. 1972. pp. 790-792.
15. HOOPER, R. T. The production
of ferromanganese.
Proceedings of the electric
flIrnaCe conference,
1967. Metall. Soc.
AIME,
vo!. 25. 1967. pp. 141-145.
16. VAN HOOK, H. J., and KEITH, M. L. The system
Fe,O.Mn,O,. Am. Mineral.,
vo!. 43, no. 1. 1958. pp. 69-83.
17. ELLIOTT, J. F., and GLEISER, M. Thermochemistry
for steelmaking.
Vol. 1. Reading,
Mass., Addison-Wesley,
1960.
18. McGANNoN, H. E. The making, shaping and treating
of steel.
9th edition.
Pittsburgh,
United States Steel, 1971.
19. Kuo, K., and PERSSON, L. E. A contribution
to the constitution of the ternary system Fe-Mn-C. Isothermal
sections at
1050°,910°
and 690°C. J. Iron Steel Inst., vo!. 178. Sep. 1954.
pp. 39-44.
20. OATES, W. A., and ToDD, D. D. Kinetics
of the reduction
of
oxides. J. Austr. Inst. Met., vo!. 7. 1962. pp. 109-114.
21. WELSH, J. Y. Reducing
to MnO the MnO. of a manganese
ore. U.S. Pat. 3375097.
App!. 18 Oct., 1965. Ace. 26 Mar.,
1968.
22. ROBIETTE, A. G. E. Improvements
in and relating
to the
manufacture
of ferromanganese
and ferro-siliconmanganese.
Brit. Pat. 936589.
App!. 25 Nov., 1960. Ace. 11 Sep., 1963.
23. MANKHAND, T. R. Kinetics
of reduction
of manganese
oxides
by hydrogen.
Banaras Met., vo!. 5. 1973. pp. 28-29.
24. SINGH, S. K., Roy, T. D., and ABRAHAM, K. P. Kinetics
of
reduction
of manganese
dioxide pellets to manganese
oxide
with hydrogen.
Trans.
Indian
Inst. Met., vo!. 27, no. 2.
1974. pp. 87-92.
25. NIKOLOV, T., and BoYANov, 0. Kinetics
of the reduction
of
manganese
oxide (Mn,O,) with carbon monoxide
and carbon
monoxide-containing
gas mixtures.
God. Vissh. Khim. Inst.
Sofia., vo!. 15, no. 4. 1970. pp. 263-276.
26. BoARDMAN, L. Manganese
in the Union of South Africa.
Seventh
Commonwealth
Mining and Metallurgical
Congress,
1961.
27. DE VILLIERS, J. The manganese
deposits
of the Union of
South Africa. Geological Survey Handbook,
no. 2. 1960.
28. DE VILLIERS, P. R. The geology
and mineralogy
of the
Kalahari
manganese
field, north of Sishen, Cape Province.
S.A. Geol. Survey Memoir, no. 59. 1971. pp. 1-84.
29. DE VILLIERS, J. P. R. A mineralogical
investigation
of
natural.
and heated
Mamatwan
manganese
ore. Johannesbttrg, '-National
Institute
for Metallurgy,
Technical
MemorandUM,
Project
07772, no. 1. May 1973. 12 pp. (Unpublished).
30. NATIONAL INSTITUTE FOR METALLURGY. Determination
of
ffiitri'ganese
in 'ores. Johannesburg,
the Institute.
Internal
Laborat",ry
Method,
no. 25/2. Jan. 1966. (Unpublished).
31. AUST,EN, C. E., RUSSELL, B. G., and STEELE, T. W. The
ana.1ysis of silicate rocks and metallurgical
slags by X-rayfluorescence
spectrometry.
Johannesburg,
National
Institute
for Metallurgy,
Report no. 1599. Feb. 1974.
32. GIti!\!:SLEY, W. D. A study of the reduction
characteristics
RI'Itj,mechanis!lls
of reduction
of manganese
ore by solid
c~rpon.
M.Sc. (Eng.) Thesis, University
of the Witwatersralld;1977.
158 'pp.
33. RIBOUD, P. V., and MuAN, A. Melting
relations
of CaOwanganese
oxide and MgO-manganese
oxide mixtures
in air.
J. Am. Ceram. Soc., vo!. 46, no. 1. 1963. pp. 33-36.
34. DE VILLIERS, J. P. R. Private
communication.
1976.
35. RAO, Y. K. The kinetics of reiuction
of hematite
by carbon.
Metall. Trans.,
vo!. 2. 1971. pp. 1439.1447.
JOURNAL OF THE SOUTH AFRICAN INSTITUTE OF MINING AND METALLURGY