INFACON 7, Trondheim, Norway, June 1995Eds.: Tuset, Tveit and PagePublishers: FFF, Trondheim, Norway

TIIERMODYNAMICS OF PHOSPHORUS IN CARBON-SAlURATEDMANGANESE-BASED ALLOYS

S.C.Shim and N.SanoDepartment of Metallurgy, The UniverSity of Tokyo, Tokyo, JAPAN

ABSTRACTRemoval of phosphorus from manganese-based alloys is difficult due to the preferential

oxidation of manganese. Recently we have found that the BaO-MnO flux system is efficient forthis purpose. Accordingly, the present study has been performed to provide thermodynamicdata for manganese-based alloys. A slight decrease in the activity coefficient of phosphorus inFe-Mn-Csatd. alloys was observed with increasing manganese content, reflecting a strongerinteraction between manganese and phosphorus than that between iron and phosphorus. Wealso found that activity of phosphorus in Mn-Si-Csatd. alloys significantly increases withincreasing silicon content.

INTRODUCTION

Phosphorus is known to be one of the impurities which should be removed from ferro­manganese for the production of non-magnetic Fe-Mn alloys. However, its removal by theconventional dephosphorization method by oxidation is difficult due to the preferentialoxidation of manganese. Recently we have found that the BaO-MnO system is efficient toserve this purpose as shown in FIG.I. 1

) Accordingly, a series of works have been undertakenin our group to know the thermodynamic properties of phosphorus in Fe-Mn alloys in orderthat their contribution with the information on the phosphorus distribution ratio between theflux and metals may contribute to understanding the whole picture of phosphorus behavior inthe flux treatment. Since the details of the present study will be published later, the summary ofwhat has been known up to now will be described in this presentation with emphasis ondiscussion.

I. DETERMINATION OF TIIE ACTIVITY COEFFICIENT OF PHOSPHORUS INCARBON SATIJRATED Fe-Mn ALlOYS

The phosphorus distribution ratios between carbon saturated Fe-Mn alloys and the BaO-BaF2system in a graphite crucible, Lp(Fe-Mn-Csatd.} were measured at 1573 K in an Ar-CO

mixture having the oxygen partial pressure of 3.38xlO-18 atm(Pco=0.333atm) at 1573K and

7.89xl0-17 atm(Pco=latm) at 1673K. Since the BaO-BaF2 system containing 15.0 to 20.2mass pet of BaO has a very limited solubility of MnO, as shown in FIG.2, the presence of asmall amount of MnO was assumed not to affect the activity coefficient of phosphorus in theflux.

The phosphorus distribution for carbon saturated iron without manganese, Lp (Fe-Csatd.),had been measured under the same conditions with the same flux, so that Eq.(1) holds.

611

612

Iu(Fe - Mn - Csatd.) I fp(Fe - Mn - Csatd. ) eMIl [ Mn]

og = og '""' mass pctLp{Fe - Csatd.) fp{Fe _ Csatd.) P,Csatd(1)

where epMn,Csatd. is an interaction parameter between Mn and P at carbon saturation which isnot equal to that defined for an infinitely dilute solution of molten iron, and fp is the activitycoefficient of phosphorus in metallic melts. The values for fp (Fe-Csatd.) are known as

Eq.(2), so that fp (Fe-Mn-Csatd.) can be calculated according to Eq.(1), epMn,Csatd. being

also estimated.

386log fp{Fe - Csatd ) = - - + 0.891 2) (2)

T

FIGURE 3 shows the Lp (Fe-Mn-Csatd.) as a function of manganese content. The Lp valuesdecrease with increasing manganese content in all cases. When plotted in FIG.4 as log fp (Fe­Mn-Csatd.) vs. manganese content, all plots in FIG.3 are represented by a single straight line,regardless of slag composition and temperature. The slope gives -0.0029 for epMn,Csatd.'independent of temperature, meaning that an addition of manganese to carbon saturated irondecreases the activity coefficient of phosphorus. When the solubility of carbon in Fe-Mnalloys, which was simultaneously determined and is shown in FIG.5, is extrapolated to unityfor Mn/(Fe+Mn), the manganese content of an unstable Mn-C melt at 1573 K at carbon

saturation is 92.7 mass pet, giving 0.53 for fpMn,Csatd. for the Mn-C melt, where the

standard state of phosphorus is that in carbon saturated iron. 11rls indicates that phosphorus incarbon saturated manganese is more stable than in carbon saturated iron by -8.3 kJ/mol (=8.314x10-3x1573xln 0.53). I..u3) estimated the standard Gibbs energy for Eqs.(3) and (5) asEqs.(4) and (6), respectively.

P(l) - P(X)inMn (3)

~Go ... - 203,611.4 + 41.003T (llmol) (4)

P(l) = P(X)inFe (5)

~Go ... -159,499.3+ 19.891T (llmol) (6)

Equations (4) and (6) give -139.1 and -128.1 kJ/mol at 1573 K, respectively. Equation (2)shows that phosphorus in carbon saturated iron is less stable by 19.4 kJ/mol than that in pureiron. Combining all those values, phosphorus in carbon saturated manganese is estimated lessstable by 22.1 kJ/mol than that in pure manganese, leading to give fpC (Mn-Csatd.)= 5.3,which is slightly larger than that for iron, 4.4, calculated from Eq.(2). The relationships ofthose values are schematically shown in FIG.6.

Using Eqs.(4) and (6), the values for epMn for Fe-Mn melts may be evaluated according toEqs.(7) and (8).

fMII _ Ks . MMnp - K3 MFe

10gf~1I "" e~lI[mass pet Mn]

(7)

(8)

10000 r----,-----.--.,-----.----,

613

1 0;----::20;:---47:0:------:6:'7:0-----=-8'::-0--1,.-'00

[mass%Mn]

() IJ J::,. : Baa saturation() [) 6 : MnO saturation

1000

100

10

66o

1573K, BaO-MnO-BaF,1573K, BaO-CaO-MnO

BaO-MnO• ()() 1573K

• IJ [] 1623K

.... J::,. 1673K BaF2 BaF2

~ L+BaF2

it"10 L+BaF,,+MnO

A ~ 40

60

80L

L+BaQ

Baa 20 40 60 80 MnOmass percent

Fig. 1. Phosphorus partition ratiobetween BaO-MnO flux and Fe­Mn-Csatd. alloy t 1573, 1623, and1673 K as a function of Mncontent of Fe-Mn-Csatd. alloy.

Fig. 2. Solubility of MnO in the BaO­BaF2 system at 1573 K.

2.4

o 20 40 60

8 • :20.2 mass%BaO1573K, 0 :17.9 mass%BaO

o :15.0 mass%BaO

[mass%Mn]

80 100

2

1.6

.-91.2

0.8

• :1573K

• :1673K

J I I

20 40 60

[mass%Mn]80

Fig. 3 Effect of manganese on thephosphorus partition between Fe­Mn-Csatd. alloy and BaO-BaF2

fluxes.

Fig. 4. Effect of manganese on theactivity coefficient of phosphorusin Fe-Mn-Csatd. melts.

614

where M denotes molecular weight. Equation (8) gives epMn as -0.0039 and -0.0027 at 1573

and 1673 K, respectively. They are in good agreement with the present value of -0.0029,

which is valid only at carbon saturation. Schenck et al.4) reported EpMn as 0 between .1788

and 1923 K, whereas Ban-ya et al.S) showed EpMn = -7.17 or epMn = -0.032 up to 19.3 mass

pet manganese at 1673 K. The present value is one order of magnitude smaller than the valueof Ban-ya et al.

Liu6) derived Eq.(9) for the activity coefficient of phosphorus in Mn-C alloys containing a

small amount of phosphorus relative to pure liquid phosphorus, Yp.

In Yp =-10.64 + 14.27 Xc + 30.6XC2 (9)

Substituting Xc =0.257 at carbon saturation at 1573 K into the last two terms in Eq.(9) yields295 for fpC, which is far larger than 5.3 by the present estimation, probably because Eq.(9)

might not hold at carbon saturation.

As mentioned earlier and seen in FIG.7, MnO contents of the fluxes are limited (the maximumcontent is 0.6 mass pct.), so that the effect of MnO on the activity of phosphorus in the fluxmay be ignored.

II. DETERMINATION OF TIlE ACTIVITY COEFFICIENT OF PHOSPHORUS INCARBON SATURAlED Mn-Si ALWYS

When the alloy contains a large amount of silicon, the activity coefficient of phosphorus isknown to be enhanced to have a relatively high partial pressure of phosphorus. For thisreason, the partial pressure was controlled by equilibrating carbon saturated Fe-Mn-Si alloyswith CaC2, Ca3P2 and Cat 3.33xl0-4 atm of Ppz at 1573K, according to Eqs. (10) and (11)as demonstrated in FIG.8.

3CaC2(S) + P2(g) = Ca3P2(s) + 6C(s)b.Go = -383,160 + 177f (J/mol) 7),8)

(10)

(11)

Phosphorus in the alloy is equilibrated with P2 according to Eq. (12)

1I2P2(g) =P(X) Mn-Si-Csatd.yp.xp

K12 = 1/2PP2P Si. ea

Inyp= Ep,Csatd.• XP+ Ep,Csatd.· XSI + Ep,Csatd • Xea

Rearranging Eqs.(13) and (14) yields Eq. (15).

(12)

(13)

(14)

1 peal SiIn XP - Ep,Csatd. -:XP - Ep,Csatd •Xea + '2 In PP2 - -InK12+ Ep,Csatd.· XSi (15)

If the left hand side, which is experimentally determined, is plotted against XSi, ignoring the

EpCaXCa term because of a small content of calcium in the alloy and using EpP = 16.73), a

linear relationship with a slope of EpSi and an intercept of -In K12 will be obtained.

FIGURE 9 shows the relationship between Xp and XSi at constant Ppz = 3.33x10-4 atm at

615

8---: Ahundov et. a1.

0.8

1008040 60[mass%Mn]

24mass%BaO// .

/ 17.9mass%BaO// :;,; I;) j/

// ./,;-. 15mass%BaO

/ .& 10mass%BaO

50mass%BaOI 20.2mass%BaOI

0.22• - :1573K• --. :1673K

o 0.2 0.4 0.6 0.8 1mass%Mn/(mass%Mn+mass%Fe)

I I

u~(/J(/Jc;:l

B 4

Fig. 5. Effect of manganese on thecarbon solubility of the Fe-Mn­Csatd. system

Fig.7. Relationship between manganesecontent of Fe-Mn-Csatd. melts andBaO-BaF2-MnO fluxes at 1573 K.

o--,....----......1""""-----.......- PO)

-20.6

-24.3

""'T"'i- l/2P2(g)

1Mn-Si-C(1)at XSi,max = 0.41

(59.5) (55.8)

-128.1

-80.1 ....'Iooo.t-....Jr...Mn-Csatd.O)

1-108.7 -+&..f-- Fe-C(1) --(8~\ (59.0)

(19.4) -117.0 + ~ Mn-Csatd.(1)

~_ Fe(1) . t_ (22.1)

• _139.£1\0 + Mn(1) -139.1 ....lrl...l..._ Mn(1)

(A) (B)

Fig.6 Estimated Gibbs energies of phosphorus in Fe-C and Mn-C melts relative to P(l) at1573K. (kllmol) (A): from the experimental results in section A, (B): from thosein section B.

616

1573 K and the same data (112 In Ppz (atm) = -4.0 was used in FIG.9) are plotted in FIG.10

according to Eq. (15), giving EpSi,Csatd. = 10.4 and ~G 0 12 = -59.5 kJ/mol at 1573 K.Combining this value with -20.6 kJ/mol at 1573 K for Eq.(17), the Gibbs energy ofphosphorus relative to P(l) is -80.1 kJ/mol, which means that phosphorus in carbon saturatedmanganese is less stable by 59.0 kJ/mol ( this gives 91.0 as fpC, CsatdJ than in puremanganese.

P(l) = 1/2P2(g)~Go = 62675.4 - 52.950T (J/mo!) 3)

(16)

(17)

This last value is much larger than 21.7 kJ/mol obtained in Section A. The reason fordiscrepancy is not clearly known but presumably due to employment of the value of epP,

which is valid only for a melt without carbon, and neglect of the EpCaXCa term in Eq.(14).Another conceivable reason could be that because of a relative large concentration ofphosphorus in Mn-Si-C alloys in the present case as seen in FIG.9, the activity coefficient ofphosphorus may not be described only by the term of first order of Xp and a second order termmay be needed. Accordingly, we are presently measuring the activity coefficient of phosphorusin a dilute concentration range by equilibrating the metal in a flowing argon with a fixed partialpressure of phosphorus together with molten silver as a reference.

III. DElERMINATION OF TIIE ACTIVITY OF MANGANESE AND CARBON IN AMOLTEN Mn-C MELTJ)

The activity of manganese in a Mn-C melt in an Al203 or MgO crucible was determined byequilibrating it at 1628K with a solid MnO pellet in an Ar-CO mixture according to Eq.(18).

MnO(s) + C(s) = Mn(l) + CO(g)~Go = 286,800 - 170.2(±2,500)T

aMn .Peo yMn. XMn .peoK18 z: _ "'"'------

ae yc·Xe

In peo(l-Xe) =InK18+In~Xc yMn

where In YC is described as Eq. (22).

(18)(J/mol ) 10) (19)

(20)

(21)

where YCo is the Raoultian activity coefficient of carbon at infinite dilution.The Gibbs - Duhem relationship yields Eq. (23).

(23)

At carbon saturation, Eq. (22) deduces

In YCo= -In XCsatd. - EcC[xcsatd._~(xcsatd.)2]+ ~PcCc[(xcsatd.)2

2-3" (XCsatd·P] (24)

617

Mullite tube

t--1---AJumina holder\-t+-+--- Quartz capsule

&-f--t;-+-+-- Graphite crucible+-1-+--+--- Ca3Pz

SHttilit-++-t-t-- CaCz

-+-+-+--+--- Metal

1I-11--+-t--l--'Thermocouple

0.2

""'X

• •0.1 ••SiCsatd.

•I I

0 0.1 0.2 0.3 0.4 0.5XSi

Fig. 8 Illustration of experimentarrangements for equilibration

Fig. 9 Effect of silicon on thephosphorus content in Mn-Si­Csatd. melts at 1573K.

-6

•

• ••

o

8 -1Il-<

b.O.3

I'I •

• : Experiment

-: Calculated / •

••./

/.••/

0.300.10 0.20

Mole fraction of carbon

_2 '---_'---------lL-..--l_---J._---J._---I

0.000.50.40.2 0.3XSi

0.1o

Fig. 10 Relationship between (In( I/Xp)­16.7Xp-4.0) and silicon contentin Mn-Si-Csatd. melts at 1573K.

Fig. II Relationship between partialpressure of CO and carboncontent of Mn-C alloy at 1628K.

618

where XCsatd. = 0.268 at 1628 K.Substituting Eq.(24) into Eq. (21), Eq. (21) may be written as Eq. (25).

In pco(I-XC) =In K18 -In x~atd. + ECC[XC-x~atd'+.!.(XCsatd.)2]Xc 2

3 2- -PCCC[XC2 - (XCsatd.)2 +-(XCsatd.)3] (25)

2 3

By linear multiple regression analysis of the data plots according to Eq. (25), namelyPCO(1 - Xc) th fi II . d . edIn vs. Xo e 0 owmg parameters were eterrnm :

Xc

In K18 = -0.879(KI8=OAI5), ECC = -10.9, PCCC = -46.8, PCC = 75.6

and

In YMn =5A7X2 - 46.8X2 (26)

In YC =-0.291 -10.9XC + 75.6xc2 -46.8Xc3 (27)

The standard Gibbs energy for the reaction (18) calculated from In K = -0.879 differs fromEq.(19) which gives In K=0.718(K=OA88), only by 2.2 kllmol which is within the error ofEq. (19) This probes the accuracy of the present measurement.

FIGURE 11 shows the experimental and calculated relationship between log Pca and Xc, thecurve representing the data very well.

FIGURE 12 shows the activities of carbon and manganese in the Mn-C melt at 1628K, whichare calculated using Eqs. (26) and (27), being compared with the results of otherinvestigators.10}14) At carbon saturation the activity of manganese negatively deviates fromideal behavior, being 0040 at carbon saturation.

0.3

Activityof carbon

0.1 0.2Mole fraction of carbon

~'""~~ ~Activity ,~~,'::;S::Raoult's lawof Manganese~'"

'\ , ~

'\'" ~\ '\

\~"\~, ,," ," ,\ \'. "

, ... '\\\~

V

0.0 ~_,----_,----_'----_'----_.L.----'

o

0.2

- -. Tanaka, 1673K----: Edstrom et al., 1628K- ----: Lee et aI., 1873K---: Huang, 1673K

.~ 0.4 ...: Gee et aI., 1700K

..., __ : Present work. 1628K:~

Fig. 12. The activities of manganeseand carbon of liquid Mn-Calloys.

CONCLUSIONS

The thennodynamic properties of phosphorus in carbon saturated manganese and its ferrousalloys were investigated by using a chemical equilibration technique. The results aresummarized as follows. .1. The activity coefficient of phosphorus is enhanced with adding manganese to carbonsaturated iron with the interaction parameter of epMn, Csatd. being -0.0029 between 1573 and1673K.2. The activity coefficient of phosphorus in carbon saturated manganese is enhanced by thepresence of carbon with fpC being equal to 5.

3. Silicon substantially enhances the activity coefficient of phosphorus with EpSi,Csatd.being equal to 10.4.4. The activity of manganese deviates negatively from an ideal solution with being 0.40 atcarbon saturation (XC =0.268) at 1628 K.

REFERENCES

1. Y.Watanabe, K.Kitamura, I.P.Rachev, F.Tsukihashi, and N.Sano, Metall.Trans. B,24B(1993),338

2. F.Tsukihashi, M.Nakamura, T.Orimoto and N.Sano, Tetsu-to-Hagane, 76(1990), 16643. Y.E.Lee, Metall.Trans.B, 17B(1986), 7774. H.Schenk, E.Steinmetz, and H.Gitizad, Arch. Eisenhiittenwes., 40(1969), 5975. S.Ban-ya, N.Maruyama, and Y.Kawase, Tetsu-to-Hagane, 70(1984), 656. X.Liu, Dissertation, Department of Production Technology, The Royal Institute of

Technology, Stockholm, (1993)7. D.J.Min and N.Sano, Metall.Trans.B, 19B(1988), 4338. H.Ono, F.Tsukihashi, and N.Sano, Metall.Trans.B., 23B(1992), 3139 A.Katsnelson, F.Tsukihashi and N.Sano, ISIJ. Int., 33(1993), 104510. E.T.Turkdogan, PhysicalChemistry of High Temperature Technology, Academic Press,

New York, NY, (1980), 711. R.Gee and T.Rosenqvist, Scand.J.Metall., 5(1976), 5712. A.Tanaka, J.Jpn.Inst.Met., 41(1977), 60113. Y.E.Lee and J.H.Downing, Can.Metall.Q., 19(1980), 31514. W.Huang, Scand.J.Metall.,19(1990), 2615. J.O.Edstrom and X.Liu, Dephosphorization of Ferromanganese Melts Part 1, Theoretical

Considerations, China-Sweden Symp., Stockholm, Sweden, May, (1992), 15

619

620