CENGIZLER. H.. illitl ERIC, R. 1-1. Thermodynamic activity of m:lIlg:mesc oxide in ferromanganese slags. alld the. di~triblltioll of man­ganese beLween the melal and slag phases. INFACON 6. Procccllillgs of the 6Th IlITel'lWl;o/lu{ Fer,.mll/oy.\· C(mgrl'.u. ClI/le TOII'II. Voluille I.Johannesburg, SAIMM. 1992. pp. 167-174.

Thermodynamic Activity of Manganese Oxide inFerromanganese Slags, and the Distribution ofManganese between the Metal and Slag phases

H. CENGIZLER AND R. H. ERICUlliversity of the Witwatersrand, Johannesburg, 50l/lh It/rica

A study was Illade of the MnO-CaO-MgO-Si02-AI20) slags that are typical of theproduction of ferromanganese in submerged-arc furnaces. The AbO] content of theslags was kept constant at 5 per cent by mass. The study was directed at thethermodynamic activities of MnO in ferromanganese slags at 1500 °C. The thermo­dynamic activities of Mn in Pt-Mn alloys were redetermined at 1300, 1400. and1500 °C for calibration purposes. The equilibrium between carbon-saturatedMn-Si-Fe-C alloys and MnO-CaO-MgO-SiOz-AI:.!O) slags was studied under aCO atmosphere at 1500 0c.

In the experiments, it was found that (/MnO values increase with increasing MnO,and lend to increase with an increasing CaO-to-MgO ratio. The (/MnO values alsoincrease with an increasing basicity ratio. In the slag-metal equjlibrium experiments.the carbon and silicon contents of the metal phase were inversely related. At con­stant silicon contents, as the Mn-to-Fe ratio increases, the carbon solubility in themetal phase increases. The silicon content of the metal increases with increasingSi02 content of the slag, and increases as MgD replaces CaD in the slag. Anincrease in the slag basicity decreases the silicon conlent of the metal phase.

Introduction

Manganese is one of the most widely used alloying materi­als in modern steel production. It is also used in non-ferrousmetallurgy. As a strategic and critical material, the opti­mization of the recovery of manganese in smelting andrefining processes is essential. The efficiency of smeltingdepends. in part, on the distribution of the valuable man­ganese between the alloy, slag, and gas phases. which, inturn, depends on t.he thermodynamic properties of the alloyand slag phases and on the phase equilibria among them.Knowledge of these factors, when coupled with informationon the viscosity and electrical conductivity, is extremelyuseful in maximizing the efficiency of the process, and inpredicting the changes that may occur owing to the use ofdifferent raw materials and to changes in furnace operation.

Literature on the thermodynamic activity of MnO in fer­romanganese slags is very limited. although a substantialamount of data has been accumulated on the binary Si02­

MnO system I, and on MnO activities in blast-furnace slagsand in CaO-SiO,-MnO and MgO-SiO,-MnO ternary sys­tems Z- 13• However. it is fell that the data provided are notaccurate enough owing to discrepancies, at the time whenthese experiments were conducted. in the standard freeenergies of formation of the MnO and x MnO.y SiOz com­pounds. In complex systems similar to ferromanganeseslags, only one study, done by Warren l4 , has been carriedout on slags with higher basicities than those used in thepresent study.

According to the work described by Warren 14, changes inthe concentration of MnO in the slag, within the composi­tional limits of the study, did not affect its own activitycoetIicient. An increase in the AI203 content for a givenbasicity ratio (mass % (CaO + MgO):mass % SiO,)decreased the activity coefficient of MnO, especially in themore basic slags. The main aim of the present study was todetermine MnO activities in MnO-CnO-MgO-Si02­AbO) slags with basicities ranging from about 0,4 to 1,2and MnO contents ranging from 5 to 30 per cent.

Activity--eomposition relations in Pt-Mn binary systemsare well established ls . 17 . This metallic system was re-inves­tigated in the present study for calibration purposes at1300, 1400, and 1500 0c. The slags were equilibrated withplatinum strips during the determination of MilO activities.The equilibrium between carbon-saturated Mn-Fe-Si-Calloys and MnO~aO-Mgo-SiO,-A1,03 slags was alsoinvestigated at 1500 °C under a CO atmosphere in thepresent work in order to eSlablish slag-metal equilibriapertinent to the smelting of manganese ores.

Experimental ProcedureThe initial compositions of the metal and slag phases areshown in Tables I and II respectively. The slags that wereequilibrated with those metal charges were chosen from 55slags of different compositions. The gas-equilibration tech­nique was used for the detennination of activity--<.:ompositionrelations both in Pt-Mll alloys and in ferromanganese slags.

In the slag experiments, the equilibrium runs were carried

THERMODYNAMIC ACTIVITY OF MANGANESE OXIDE IN FERROMANGANESE SLAGS 167

TARLE ITHE INITIAL COMPOSITIONS Or-HIE METAL PHASES (IN PERCENTAGES)

Phase Mn Fe Si C

Fenomangancse 82,19 9,34 1,5 7

79,77 11,73 1,5 7

74,56 16,94 1,5 7

Silicomanganese 66,64 19,3 12,34 1,7

62,55 19,3 16,46 1,7

57,04 19,3 21,94 1,7

TABLE IITHE INITIAL COMPOSITIONS OFTHE SLAGS (IN PERCENTAGES)

SIng MoO C"O MgO Si02

I 5,0 20.0 11.9 58.12 5,0 25.0 6.9 58,13 5.0 2.'i,O 15.0 50.04 5,0 JO.O 10,0 50,05 5,0 35,0 5,0 50.06 5,0 32,1 15,0 42,97 5,0 35,11 12,1 42,98 5,0 35,11 17,5 37,59 10,0 20.0 10.2 54.8

10 10,0 25,0 5,2 54.8II 10,0 25.0 Il.R 47,212 IO,D )O,n 7.8 47.213 10,0 35,0 2,8 47,214 10,0 311,0 14.5 40,515 10,0 35,0 9,5 40,516 10,0 32.8 16.8 35,417 10,0 35,0 14.6 35.418 15,0 20.0 8.4 51.619 15,0 25.0 3.4 51.620 15,0 20,0 15.6 44,421 15,0 25,0 10,6 44.422 15,0 30,0 5,6 44.423 15,0 26.9 15,0 38,124 15,0 31.9 10,0 38,125 15,0 35.0 6,9 38,126 15.0 311.0 16.7 33.327 15,0 35,0 11.7 33328 211,0 20,0 6,6 48,429 20,0 20,0 13,3 41,730 20.0 25.0 8,3 41.7

31 20.0 35,0 3,3 41,732 20,0 24.lJ 15,0 35,733 20,0 30,0 93 35.734 20,0 35.0 4.3 35.735 20,0 28,7 15.0 3L336 20,0 33,7 10,0 31,337 20,0 35,0 8,7 31,338 25,0 20.0 4.8 45,2

39 25,0 20.0 ILl 38.940 25,0 25,0 6.1 38.941 25,0 21.7 15.0 33,3

42 25,0 26,0 10,0 33,343 25,0 30.0 6,7 33,344 25,0 25.8 15.0 29,245 25,0 30.11 10,8 29,246 25,0 35,0 5.8 29.247 30.0 20.0 3.1 41.948 30,0 2U,0 8,9 36.149 30,0 25.0 3.9 36,1

50 30.0 lO,n 14.1 30,951 30,0 25.0 9.1 30.952 30,0 29.1 5.0 30,953 30,0 22.9 15.0 27.154 30.0 27.9 10.0 27.155 30,0 32.lJ 5.0 27.1

168

out with different slag compositions under two different P02

values: 4,76 x 10-7 and 5,80 x 10-8 atm. Tn total, 55 samplesof slags of different compositions, containing MnO, CaO,MgO, SiOl, and AI 20 3 were prepared from analyticaJ­reagent grade oxides. All the oxides, except MnO. werefirst calcined in an electrical re.sistance-heated muffle fur­nace for abollt 24 hours at 1200 °C. After cooling, theoxides were mixed in the desired proportions 10 give slagsof the desired compositions. The mixtures were pelletized,and the pellels were recalcined I'or 12 hours at about 1200 'Cin order to promote some lIseful sintering. and therebyhomogenization, of the samples. On cooling, the pelletswere crushed and ground in an agate mortar and mixed withthe required quantities of reagent-grade MnO before everyindividual run, The rest of the pellets were stored in sealedbottles for fe-use. All the mixing procedures were carriedout under acetone in order to secure the best possible mix­ing of the oxide components. The samples were heated forseveral hours at 110°C to drive off the acetone. cooled, andcharged into the manganese-saturated platinum crucibles.

The platinum specimens for equilibration were prepared ascoupons with dimensions of 4 mm by 4 mm, cut from foil0,025 111m thick. Prior to any experimental run with a plat­inum specimen, the platinum crucibles were equilibrated withpure MnO at the chosen gas mixture at 1500 dc. The platinumcrucibles were made by the welding of platinum sheets0,125 mm thick. The slag samples for the slag experimentsand the pure MnO for the Pt-Mn calibration tests werecharged, in I g quantities, into MnO-saturated platinum cru­cibles with the following dimensions: 7 mm in diameter, 25 mmin depth, and a wall thickness of 0, 125 mIn. Seven platinumstrips were immersed in the slag (or pure MnO) sample atdifferent levels. In the case of the slag experiments, theplatinulll crucibles charged with slags of different composi­tions were placed in an AI-size AhO) crucible SUIToundedby AbO) bubbles, so that it was possible to run four or fivedifferent experiments at a time. The AbO] crucible wasplaced on the pedestal and pushed up slowly into the even­temperature region of the furnace over a period of about 40minutes. The reaction tube of the furnace was tlushed forabout 15 minutes with spectrographically pure argon beforethe mixed gases were added. In the Pt-Mn calibration tests,the samples were left in the furnace for 48 hours at 1300 °Cand for 24 hours at 1400 and 1500°C, However, all theslag samples were equilibrated in the furnace at 1500 °Cfor 24 hours. Previous studies mention an equilibration timeof about 8 hours for slag experiments l5 , Profiles obtainedby X-ray energy-dispersive analysis using a scanning elec­tron microscope (SEM- EDAX) of manganese on platinumstrips verified that equilibrium had been attained in that theprofiles were very smooth. At the end of the equilibration,the samples were rapidly withdrawn into the water-cooledend of the furnace and quenched in water.

In the slag-metal distribution experiments, the metalcharge was prepared from electrolytic manganese and ironpowders, high-purity silicon. and spectrographic-gradegraph.ite powder. From these. a number of master alloyswere prepared by melting the elements under an argonatmosphere in graphite crucibles. After the master alloyshad been crushed and ground, they were stored in closedbonIes jn a vacuum desiccator. The compositlons of theactual alloys for the experimental runs were adjusted by theaddition of the necessary pure clements in powder form tothese master alloys. Graphite crucibles, 14 mm in internal

TNFACON 6

0,----------------,

0,5

•

•

0,4

•• •

•

0,3

••

0,2

••

0,1

X...FIGURE I. The change of the activity coefficient of manganese with

composition in Pt-Mn alloys at 1500 °c

chiometric MnO in the non-stoichiometric oxide that wouldbe in equilibrium with P'02' Values of P'02 and {I'MnO havebeen mellsured by Richardson at al. 15 and by Davies andRichardson 18 respectively.

Presenl work •

Gaskelpl ---

Extensive experimental runs were carried out in order todetermine MnO activities in slags at 1500 ClC, The AI20Jcontent of the slags was kept at about 5 per CCIll by mass.The basicity ratios of the slags varied between 0,41 and 1,18.

A quadratic multivariable regression model was devel­oped to predict the activities of MilO in ferromanganeseslags as a function of slag compositions based on the gath­ered data. The seven independent variables chosen initiallywere the mass percentages of MnO, CaO, MgO, Si02, andAbOJ in the slag, as well as the basicity ralio, B, and theCaO-to-MgO ratio. The regressions were carried out by theuse of routines from the SAS (Statistical AnalysisSoftware) system. Initially, the SAS procedure RSREG.which tils a complete quadratic response surface to thedata, wns used, nnd the correlmion coefficient was found tobe 0.982. For a seven-variable model, including qundraticand cross-product terms, the analysis resulted in a modelhaving 36 parameters. which were regarded as too many.The SAS procedure RSQUARE was then used, and the cor­relation coefficient was found to be 0,979. This procedureselects the optimum combination of independent variablesin the model. With lhis approach. it was found that essen­tially similar fits could be obtained from a model with far

Results and DiscussionActivity-eomposition relations in the Pt-Mn system werere-determined for clllibration purposes at 1300, 1400, and1500 °C and at Po, values between 5,40 x 10.6 and 4.54 x10- 13 atm. In Figure I. the activity coefficient of manganeseis plaited against its mole fraction. The results of the pres­ent work are in good agreement with those obtained in aprevious study l7.

-2

-4

..{j

~~

.!3-8

-10

-12

-14

°

diameter and 44 mm in height, were used for the equilibri­um runs. The charge consisted of about 6 g of metal and 4g of slag of selected compositions. The nowrate of CO waskept at 600 cm3/nUn. After equilibration at 1500 °C for 24hours. the crucibles with their contents were quenched inwater. The slag and metal phases were then removed fromthe crucible. The graphite Lhat had stuck onto bOlh phaseswas removed with emery paper or a wire wheel.Afterwards. both phascs were kept in sealed boules untiltheir contents were analysed. A portion of the samples waskept for EDAX work.

The analytical work was done at Mintek. The manganesecontents of the equilibrated PI-Mn coupons were deter­mined by the standard periodate (photometric) method. thefinal basic constituents of the slag were analyseu by X-rayfusion, and the iron content of the slag was detennincd by theo-phcanthroline colorimetric method. The carbon content ofthe metal was analysed by means of a Leco analyser. andthe other constituents of the metal by the inductivelycoupled plasma (lCP) method.

In all the aClivity and slag-metal distribution experiments,the initial Ab03 content of the slags was kept at around5 per cent by mass.

A gas-tight reaction fumace heated electrically withmolybdenum wire as the resistance was used. All the gasesfirst passed over drying columns of silica gel and anhydrousmagnesium perchlorate. Argon gas was also passed throughcopper turnings kept at about 500 ClC so that as much oxygenas possible WtlS removed. The ancillary equipment thus in­volved a gas-purifying train and a gas mixer of conventionaldesign. Type B Ulennocollples were lIseu both for the measure­ment of sample temperatures and for control of the furnacetemperature using Eurotherm controllers. The low oxygenpartial pressures were obtained by the mixing of CO andC02 gases through capi lIary tubes connected to tlowmeters.

The method adopted for the measurement of MnO activi­ties in ferromanganese slags was similar to that used byAbraham el al. 2: the platinum strips were equilibrated withslag in a platinum crucible. and a panial pressure of oxygenwas imposed by gas mixtures of CO and C02. If initiallypure MnO is used instead of a slag containing MnO, theequilibrium constant for the reaction will be given by

MIlO(l) = MIl(I) + 1/20,(g) (pure MilO) [1.1

K' = a' M';v'(P'O,)/a'M"O· [2.1If a slag containing MnO is used. the equilibrium constantwill be

MIlO(l) = MIl(I) + 1/20,(g) (slag) [3J

K = aM;;)(j'O,)/aM"o· [4J

Since aClivities are used, and nOI conccnlrdlion terms. theequilibrium constants given by equations [2.1 and 14] will bethe true constants and will be equal. Equaling and re­arranging give

"M"O =Y(jJo/p'O,)"'M,,O("M/a'Mn) [51

and. when Uro.'In",,{I'Mn' equation [51 becomes

{lMnO =Y(jJO/p'02)a'MnO.

Therefore, Po" is the partial pressure of oxygen prevailingat the time of the experiment, P'O"l is the pressure of oxygenthat would be in equilibrium with pure MnO and an alloythat contains the same amount of manganese as was formedin the slag experiment, and (I'MnO is the activity of' the stoi-

THERMODYNAMIC ACTIVITY OF MANGANESE OXIDE IN FERROMANGANESE SLAGS 169

fewer parameters. The analysis resulted in a model having16 parameters, and this was referred to as the 'reducedmodel'. Thus, the equation used to predict the activities ofMnO in ferromanganese slags using the reduced modelwould be written asaMoO =0,7265 - 0,0249 (o/oSiOz) - 2,3115(B}' + 0,0619 (%MnOx B)-O,OO03 (%MnOJ2 + 0,1451 (%CaO xB) -0,0018 (%CaOx %MnO) - 0,0023 (%CaO)' + 0,1439 (%MgO x B) - 0,00019(%MgO x %MIlO) - 0,0045 (%MgO x %CaO) - 0,0022(%MgOJ' - 0,0506 (%SiOz x B) + 0,0005 (%SiOz x %MIlO) +0,0015 (%Si02 x %CaO) + 0,0015 (%SiOz x %MgO).

In Figures 2 and 3, the individual data groups w1th simi­lar basicity ratios and different CaO-to-MgO ratlos aresuperimposed. The curves represent the predicted valuescalculated from the empirical model, and the symbols rep­resent the experimental values. From a large number ofmeasurements of the optical basicity of glasses l9 usingmainly Pb2+ as probe ion, the optical basicity of an oxide, A,was found to be related to the Pauling electronegativity, x,of the cation involved by the expression

C.O/MgO-3,415

B.R.·O,157!O,Ol

A -0,68

• CIIO/MgO-3._415:!.O,oe

A. - O,74/(x - 0,26).

Hence, by use of the relationship

A = XAA.A+XBA.B + ...,

[7]

[8]

o 4 8 12 16 20 24 28 32 36

MnO, mass, %FIGURE 3. Activities of MnO in Si02-AbO)-MgO-CaO-MnO slags for

various CaO-to-MgO ratios

FIGURE 2. Activities of MnO in SiOz---Ah03-MgO-CaQ-.-MnO slags forvarious CaO-lo-MgO ratios

it was possible to calculate the bulk value of A (the opticalbasicity) for a slag of any composition involving theseoxides. In equation [8], X is the equivalent cation fractionbased on the fraction of negative charge neutralized by thecharge on the cation concerned. For any component, X wascalculated in terms of mole fractions by the equation

(mole fraction of component· number ofoxygen atoms in oxide molecule)

x=-------------· [9JL. mole fraction of component· number of

oxygen atoms in oxide molecule

Normally, an increase ir: the mass concentration of MnOincreases the MnO activities, and an increase in the CaO-to­MgO mass ratio increases the QMnO values. The increase inthe activity of MnO with increasing concentration can beexplained in terms of the principle of the ionic structure inliquid slags. At anyone concentration of MnO, a certainnumber of Mn2+ ions will be associated with the silicate andaluminate ions, and a certain number only with free oxygenions. An increase in the MnO content of the slag will causea simultaneous increase in the number of free oxygen ionsand in the number of associated manganese ions. In otherwords, there will be excess Mn2+ and 0 2- ions in the moltenslag. Hence, the activity of MnO increases with increasingconcentration of MnO in the slag.

The result obtained during the runs showed that CaO hasa slight effect in increasing the aMnO values. This is inagreement with the work of Warren l4 . In the present work,the AI20 3 content was kept at around 5 per cent by mass inall the experimental runs. When the free energy of the for­mation of the orthosilicates Ca2Si04' Mg2Si04, andCaMgSi04 are taken into consideration, it can easily beseen that Ca2Si04 has the most negative free energy of for­mation2o-23. Furthermore, the free energy of formation ofCaMgSi04 is more negative20-23 than that of Mg2Si04 . Inother words, the interaction of Ca2+ ions with Si02 isstronger than those of mixed Ca2+ and Mg2+ ions and Mg2+,Thus, one can conclude that an increase in the amount ofCaO can lead to an increase in the formation of Ca2Si04orthosiJicate, leading to higher QMnO values in the slags.Therefore, because of the interaction with Sial, Mn2+ ionsbecome freer, i.e. less associated with the Si02 network.and therefore the QMnQ values increase.

In Figure 4, the individual data groups w1th similar CaO­to-MgO ratios and different basicity values are superim­posed. The curves represent the predicted values calculatedthrough the empirical model, and the symbols represent theexperimental values. An increase 1n the basicity ratio clear­ly increases the MnO activity. This can be explained in a

CllO/MllO-'..90

C.O/MllO-0,23

CaO/MgO-t•• r

16 20 24 28 32 36

MnO, mass %

1264

o CIIO/MOO-1.47~O.06

B.R,-0.Q1:!.O,03

1\ -0.68

:14,,------------------,

'~fnv 04 "I 0 •~ CaO/MgO-2,QO~O,0_4 .,'/ i

016

1

1

,-;;/ II

0'-:/. / I

.' / I

008 r .roo);/" I..~~ I

o[_--,---~_,----_'-.L',_,:,Y_.M~;--_"/-----'_---'--_-'-------.J._.Jo

170 INFACON 6

very similar way in tenns of modern slag theory. At lowbasicities. almost all lhe metal cations are associated wilhthe large SiOl anionic groups, and only a few 0 2• ions exist.The activity of MnO in such melts is thus low (high slagviscosily). In other words, at low basicities, the Si02 net­work is not disrupted. As the concentration of the basicoxides increases, the SiOl network is broken up into small­er anionic groups (lower slag viscosity) and the proportionof free oxygen anions begins to increase. Thus. an equalnumber of divalent cations. which were associated withSiOl anions. associate with free oxygen anions. At thispoint, divalent cations like Ba2+, Ca:!+. and Mg2+, becauseof their higher interaction with SiCh ions. are preferentiallyassociated wilh silicate ions. and thus Mnl+ ions becomefreer from association with silicate ions. In other words. thefree energy of formation of MnSi03 is less negative thanthat of MgSi03. CUSi03, etc. Thus. Mn2+ cannot overcomethe interaction between tJlese metal cations and silicateions, and lherefore an increase in basicity increases thellMnO values in the slag.

and this behaviour can be explained in terms of the stabilityof carbides. As silicon lowers the solubility of carbon. theconditions become less favourable for carbide fonniltion.Thermodynamically, this means that silicon decreases theactivity of carbon.

5,-----------------,

Mo:Fe = 3.74

Mn:Fe = 7,42

c ~,_---'~_-'---_----'-__L__-'---_----L_---'__.l__ _'

MoO. mass. %FIGURE 4. Activities of MoO in SiOz-AhO:r-MgO-CaQ....MnO !'Iags for

various basicity ratios

FIGURE 6. The effect of the silicn content of the sing phase 011 thesilicon content of the Illet;.l phase;1I lliffcrcnt Mil-tn-Fe rmios

5,------------------,

Mn:Fe = 6,7

16~-------:;-~----"*8----~9:------:-'10

C,%

FtGURE 5. The effect of the silicon content on the carbon content of metalal different MrHo·Fc ratios

4

Figure 6 shows the effect of the Mn-to-Fe ratio of themetal phase on the relationship between the silicon contentof the metal and the SiO, content of the slag_ As the Si02content of the slag increases. the silicon content of themetal increuses, and this effect is even more noticeable atthe higher Mn-to-Fe ratios. The increase in the Si02 activityin the slag with increasing SiD:! concentration results in thetransfer of silicon from the slag to the metal.

ll' Mn:Fe::= 3,7

§. 3

lJ,

Mn:Fc = 5,52

Mn:Fc = 6,7

135 37 39 4t 43

Q.A.oO.•3

R R 0'.17

113 20 24 28 32 ~E128

'4 .41203":6

o 8 R 01.17:0.06

<; ..0'111;0 0 0,20-0,28

8,R.oO.&?=O.05

o 8.R,00,i3=0.03

---------

o

04 rI

I0.:32[

III

o I0.,,4 \'

iI

o '6 ~III

0 ')0 L,~ I

,

Extensive experimental runs were carried out on theslag-metal equilibrium at 1500 °e. The metal phase wasalways saturated with respect to graphite. The actual basici­ty ratios of the sings ranged betwecn 1,00 and 1,39 for dif­ferent ferromanganese compositions2-l. For the ferroman­ganese-metal phase. four different Mn~lO-Fe ratios wereselected: 3.7, 5,5, 6,7, and 7,4. Ln Figure 5, three differentdata groups with differenr Mn-to-Fe ratios are superim­posed. showing the relationship between dissolved carbonand silicon. It can be seen that, at constant silicon contents.as the Mn-to-Pe ratio increases, the carbon solubility in themetal phase increases. The carbon and silicon concentra­tions in the metal phase are inversely related, as expected.

THERMODYNAMIC ACTIVITY OF MANGANESE OXIDE IN FERROMANGANESE SLAGS 171

Figure 7 shows the effect of the CaO-to-AI,O, ratio on5iO, reduction at two different Mn-to-Fe ratios in the metalphase. As the CaO-to-A1,O, ratio in the slag increases (atconstant Ah03 content), the silicon content of the metaldecreases. As the Mn-to-Fe ratio increases, the effecl ofincre.1sing CaO-to-AhO, ratio on the silicon content of themetal is even greater. This indicates that the activity of Si02

in the slag is decreased by the substitution of CaO in accor­dance with the network-modification power of CaO. Thisresult is similar to the findings of Fulton and Chipman25 .

Another study" showed that, as AhO, was replaced withCaO in slags of low 5iQ, content, the equilibrium siliconcontent of the metal increased. In the case of slags with ahigh Si02 content, the opposite effect was observed. This isalso in agreement with the findings of the present study.

5

0,4,.-------------------,

"0,3'

~ •w ~

•~~

~ •~

0,2,-'

0,1' L- ----'

{l,1 0,3 0,5 0,7 ' 0,9

(MgO)%:(CaO)%

FIGURE 7. The effect of the CaO-[lrAIzO] ratio of the slag on the siliconCOnlenl of the metal at different Mn-to+Fc mtios

13~---4L------l5----'6---.J7L---...J8

(CaD), %'(AJ,O,), %

1,51,3

8

Basicity

°0

o

1,1

o

4,6

4,0

3,4

!I'

§. 2,8

2,2

1,6

1,0.0,9

FIGURE 8. The effect of the MgO-to-CaO ratio on Ihe silicon distnbUlion(al 6,7% AhO); 63, I% Si02)

In the experiments, it was found that QMnO values increasewith increasing concentration of MoO, and tend to increasewith an increasing CaO-to-MgO ratio. The aMnO valuesincrease with an increasing basicity ratio. In the slag-metalequilibrium experiments, the carbon and silicon contents ofthe metal phase were inversely related. At constant siliconcontents, as the Mn-to-Fe rntio increases. the carbon solu­bility in the metal phase increases. The silicon content ofthe metal increases with increasing Si02 content of the slag,and increases as MgO replaces CaO in the slag. An increasein the slag basicity is associated with a decrease in the sili­con content of the metal phase.

As far as the behaviour of slag is concerned, the mostimportant factors are the MnO content and the basicity ratio,

FIGURE 9. The effect of basicity on the silicon content of the meLal at1500 °c under an atmosphere of carbon monoxide

Conclusions

Mn:Fe = 7,4

o---<>--_-.- Mn:Fc = 3,7

2

0

4 0

0

00

0

"~ 3

In Figure 8, the effect of the MgO-to-CaO ratio of theslag on the silicon distribution between the metal and theslag phases is analysed at constant Al,O, and 5iO, levels.As MgO replaces CaO in the slag, the distribution ratio ofsilicon decreases, and the silicon content of the metalincreases relative to that of the slag. This can be explainedwith reference to the MnO activities. It was observed thatCaO had a slight effect in increasing the QMnO values owingto the different free energies of fonnation of the complexorthosilicates. As MgO replaces CaO. it causes a decreasein MilO activity, and there is less tendency for Ca2Si04species to fonn in the slag. This may cause an increase inthe activity of 5iO, in the slag, and therefore the siliconcontent' of the metal increases.

In Figure 9, it can be seen that the silicon content of themetal decreases as the basicity ratio of the slag increases.This can also be explained in a very similar way in terms ofthe breakup of ti,e 5iO, network on the addition of basicoxides. Because of the strong interaction between divalentcations Ca2+, Mg2+, and silicate ions, the activity of Si02decreases and the silicon content of the metal decreaseswith increasing basicity. Figure 10 shows the dramaticdecrease of the Mn content of the slag that occurs when thebasicity of the slag is increased.

172 INFACON6

o 0

10,-------------------,

Basicity

FIGURE 10. The effect of basicity on the manganese content of the slags at[500 °C under an atmosphere of carbon monoxide

AcknowledgmentsThis paper is published by permission of Mintek and theFerro AHoy Producers' Association of South Africa. Theauthors are deeply grateful to both these organizations fortheir flnancial support.

References1. Glasser, F.P. (1958). The system MnO-SiO,. Am. I.Sci.,

266, pp. 398-412.

2. Abraham, K.P., Davies, M.W., and Richardson, FD.(1960). Activities of manganese oxide in silicate melts..IISI, 196, pp. 82-89.

3. Mehta, S.R., and Richardson, F.D. (1965). Activities ofmanganese oxide and mixing relationships in silicateand aluminate melts. IISI, 203, pp. 524-528.

4. Korber, F., and Oelsen, W. (1932). The relation betweeniron containing manganese and slags consisting almostentirely of manganous and ferrous oxides. Milt. KaiserWilhelm/llst., 14, pp. 181-204.

5. Chipman, J., Gero, J.B., and Winkler, B. (1950). Themanganese equilibrium under simple oxide slags. Trans.MME, 188, pp. 341-345.

6. Bell, H.B. (1963). Equilibrium between FeO-MnO­MgO-SiO, slags and molten iron at 1500 0c, JlSf, 201,pp. 116-121.

7. Fisher, W.A., and Fleischer, H.J. (1961). The manganesedistribution between iron melts and FeO slags in MnOcrucibles and temperature range 1520 °C to 1770 °C.Arch. Eisehuttenw., 32, pp. 1-10.

8. Turkdogan, E.T., and Pearson, J. (1953). Activities ofconstituents of iron and steel making slags. Part LI­Manganous oxide. IISI, 175, pp. 393-398.

9. Bishop, H.L., Grant, N.l., and Chipman, J. (1958). Theactivity coefficient of MnO and FeO in open hearthslags. TrailS. AlME, 212, pp. 890-893.

10. Filer, E.W., and Darken, L.W. (1952). Equilibriumbetween blast furnace metal and slag as determined byremelting. Trans. AlME, 194, pp. 253-257.

II. Turkdogan, E.T. (1956). Silicon and manganese re­actions in ferromanganese blast furnace processes. JISI,182, pp. 74-79.

12. Rankin, W.J., and See, J.B. (1978). The slag-metal equi­librium and the activities of slag and metal componentsin the production of high-carbon ferromanganese. NIMreport, No. 1959.

13. Rao, B.KD.P. (1979). Activities of manganese oxide inmanganese oxide containing melts. Ph.D. thesis.

14. Warren, G.F. (1972). Liquidus temperatures and activi­ties of manganese (IJ) oxide in slags associated with theproduction of high carbon ferromanganese alloys. NIMreport, No. 1463.

15. Abraham, K.P., Davies, M.W., Barton, J.L., andRichardson, F.D. (1960). Activities of manganese insolid platinum. Acta. Met., 8, pp. 888-890.

16. Smith, P.N., and Davies. M.W. (1970). The activity ofmanganese in Pt-Mn alloys. Trans. Ins!. Min. Metal!.,79, pp. CV60-C61.

17.Rao, B.K.D.P., and Gaskell, D.R. (1981). The thenn­odynamic activity of Mn in Pt-Mn alloys in the temper­ature range 1300 to 1600 0c, Met. Trans., 12A, pp.207-211.

18. Davies, M.W., and Richardson, F.D. (1959). The non­stoichiometry of manganous oxide. Trans. Farad. Soc.,55, pp. 604-610.

00

0

0 0

00 0 o.

0

1,3

aOo0

000

00

8

as has been clearly demonstrated in this work. It was shownin recent work24 that an increase in MnO concentrationdecreases the viscosity and increases the electrical conduc­tivity. Thus, a lower viscosity implies a greater diffusivemobility of all the components in the slag, in which case thekinetics of the chemical reactions should improve. In orderto improve the recovery of manganese, its activity in theslag must be max.imized without increasing its concentra­tion. The activity of MnO can be increased by employingslags of higher basicities. However, this also results in high­er liquidus temperatures of slag24 , requiring higher powerinput in the furnace. Therefore, an optimum must bereached. Both the manganese content of the slag and the sil­icon content of the metal phase decrease rapidly asthe basicity of the slag increases. It appears that a basicityratio of around 1,0 to 1,2 results in reasonably optimumconditions where the silicon content of the metal is low andthe manganese content of the slag is around 9 per cent(about 12 per cent MnO). In the previous work24

, it wassuggested that a slag composition having an optimumaround 15 per cent would be the range for furnace opera­tion. In terms of optimum electrical conductivities, viscosi­ties, and liquidus temperatures, this is in excellent agree­ment with the findings of the present work, where QMnO

could be maximized without an increase in its concentrationin the slag by operating the slags at a basicity ratio of about1,0 to 1,2 and an MnO content of approximately 12 percent. When this is done, better recoveries of manganese willbe achieved with reasonably high MoO activities in theslag, low viscosity, low liquidus temperature, and reason­able slag resistivity.

4

THERMODYNAMIC ACTIVITY OF MANGANESE OXIDE IN FERROMANGANESE SLAGS 173

J. (1954). Slag-metalactivity of silica in

TrailS. AiM£., 200,

19. Sommerville, J.D., and Sosinsky, DJ. (1984). Theapplication of the optical basicity concept to metallurgi­cal slags. Proc. 211d 1m. Symp. on Metallurgical Slagsalld Fluxes, pp. 1015-1026.

20. Pankratz, L.B., Stuve, I.M., and Gokcen, N.A. (1978).Thermodynamic data for mineral technology. Bulletill677, U.S. Bureau of Mines.

21. Robie, R.A., Hemingway, B.S., and Fisher, J.R. (1978).Thermodynamic properties of minerals and related sub­stances at 298,15 K and I bar (10' pascals) pressure andat higher temperatures. Bulletill 1452, U.S. GeologicalSurvey.

22. Barin, R.A., and Knacke, O. (1973). Thenllochemicalproperties of inorganic substances. Springer-Verlag,Berlin.

174

23. Kubaschewski, 0., Evans, E.U., and Alcock, C.B.(1979). Metallurgical thermochemistry, 5th ed.,Pergamon Press.

24. Eric, R.H., Hejja, A.A., and Stange, W. (1991).Liquidus temperature and electrical conductivitiesof synthetic ferromanganese slags. Millerals Eng., 4,pp.1315-1332.

25. Fulton, J.C., and Chipman,graphite reactions and thelime-alumina-silica slags.pp.1136-1146.

26. Rein, R.H., and Chipman, J. (1963). The distributionof silicon between Fe-Si-C alloys and SiO,-CaO- MgO-AI,03 slags. Trans. AiME, 277,pp. 1193-1203.

[NFACON 6