AF-OSR.

TR

gO~

Final COI4RACT C09TRCT OF

Report on

ICALCULATION

F44620-76-C-0060 CERAMIC PHASE DIAGRAMS

1 December

1975

to

30

September to

1979

Submitted

Air Force

Office

of Scientific Force Base, 1979

Research D.C.

(AFSC) ''

Boiling Air 30

20332

November by Larry

Kaufman Inc.
'

ManLabs,

21 Erie Street Cambridge, Massachusetts

02139

1301
LLJ

80 1

29

87

Approved f or publlie rfV1OO distribution unlimitdt

--

-~

77

Unclassified
SECURITY CL$

.11 R PORT DOCUMENTATION PAGE

4. ITLE(andSubttle)S.

IrIt,%TON

OF

THIS PAGE (noru- Dots Emoed)

__________t_________

BEFORE CONPWLF.TING FR

OOVT ACCESSION NO. 3. RECIPIENTS ClkTALOG NIIMAIER

TYPE Of REPORT A, PERIOO COVERED

talculation of Ceramic Phase Diagrams.

ia
to

eprDcebr17
P
r

6. AUTHOR(s)

PERFORMING ORG. REPORT NUMBER

I-

CONIRACT OR CRAtJT NMUER(1)

Larry Kaufman

F446;6-76-,A00j
,

~.ManLabs, Inc.ARA&WR

PERFORMING CRGANIZAT,ON NAME AND ADESS-

21 Erie St. Cambridge, Massachusetts 02139

I1. CONTROLLING OFFICE NAME AND ADDRESS

QJ6~O
].J

10.

PROGRAM FI.FMCNT. PROJECT. TASKC

-(;

NINUES

Air Force Office of Scientific Research//Vt/6

Boiling Air Force Base

Washington, D.C. 20332
14.~t --

130 Novw
13.;UMBE
___________y__

9A79
OF/AGE

I S. SECURITY CLASS. *.f IhI.4 t..irft)

ISs. DECLASSIVIC SCHEDULE

16. DISTRIBUTION STATEMENT (of 11.ld Report)

Unclassified_ I~N. 53"CiRACIINf

_

IF.~. -

Ahr!avod ?'or public release,

u o n: tin~ 1fi int ed.

17.

DIST RI3UTION STATEMENT

(oft he obstracI entered in Block 20. It dlifernt from Repot)

IS. SUPPLEMENTARY

NOTES

19. KEY WORDS (Cnnrue on ioerRe aide It ntco'esary and Identify by block nurribet)

Data Base Quasibinary Hydrogen storage. Phase Diagrams Quasiternary Electroptical semiconductors Ceramic Systems Superalloys HEM silicon solar cells Sialons Molten Salts metatectic reactions in hafnium alloys Thermochemistry Battery Electrodes Doped Germanium Borosilicate Glasses_. ,0. ABSTRACT (Continue on roveres aide It nec.. aery arndIdentify by block nuffber) A data base is being developed for calculation of quasibinary and quasiterna4 phase diagrams of oxide and oxynitride ceramic systems. The current base ; covers combinations of Cr 0O-, MgO, A15Og. FejO%, Fe0, FeO, CaO, SiO, SitN- and AlN. Lattice Stability, Solution and Compound Phase Parameters were derived covering the liquid, spinel, periclase, corundum alpha and beta.

DD IJAN73 1413

EDITIOW OF I NOV4SS IS OSSOLETE

(/Ai/Ec-;/

Z7

77
'I

StCURITy
'
i '4

CU:ASSII ICATIeN (OF T4IS OWE .F. f7w-n Dea

Enterod)

5 -

-r

'*:Y

sit_

ICATON OF THIS 90a#0000 PAGE(vm D-at

quartz, crystobalite, tridymite and beta SiSN4 structures in all of the

quasibinary systems composed of the data base compounds.

Evaluation of these

parameters permit calculation of the thermochemical properties of these

,Si4N d

systems in good agreement with limited thermochemical data. In addition, isothermal quasiternary systems of selected oxide and oxynitride systems composed of data base compounds have also been calculated over a wide range of temperature. The free energy of formafion of beta sialons in the -Al0'O-AlN system and the X phase in the SiO '-AI O -Si4 Nt- system have syste 2 2 3 34
been calculated. Recent examples of the use of data bases and computer

techniques in solw~ng problems associated with: in-situ formation of

columbium, nickel and cobalt based superalloys, sigma phase formation, multicomponent electroptical semiconductors, sialon ceramics, molten salt,

sulfide and oxide slags, high temperature battery electrodes, hardenability, synthesis of hydrogen storage compounds, grain refinement in aluminum alloys, metatectic reactions in hafnium alloys and the growth of HEM silicon solar cell materials. The initial results obtained by applying these techniques to CVD synthesis of doped germanium-borosilicate glasses for C3 1 applications are reported.

V ,'%

'i?~;:
-

-i'

- .'i/I

,U
,A

',,',
,'-

1ECURIrYf CLAWMCATWN OF TNt, PAGI(wheft Data

941-04

DISTRIBUTION LIST AFOSR/NE AFML/LLM

Attn: Maj. W.C. Simmons Bldg #410 Bolling AFB Washington, D.C. 20332 (16 copies) AFML/LLS Attn: Branch Chief W-P AFB, Ohio 45433 U.S. Army Research Office Metallurgy & Materials Sciences P.O. Box 12211 Research Triangle Park, NC 27709 Office of Naval Research Attn: Dr. A.M. Diness 800 N. Quincy Street ARlington, VA 22203 Materials Research Lab (AMMRC) Attn: Dr. R.N. Katz Watertown, MA 02172 Prof. A.H. Heuer Prof. T.E. Mitchell Dept. of Materials Science Case Western Reserve University Cleveland, Ohio 44106 Prof. D.L. Johnson Department of Materials Science Northwestern University Evanston, Illinois 60201 Dr. D.D. Cubicciotti Stanford Research Institute Materials Research Center Menlo Park, California 94025

Attn: Branch Chief W-P AFB, Ohio 45433

AFML/LLN Attn: Branch Chief W-P AFB, Ohio 45433 AFML/MBM Attn: W.L. Baun W-P AFB, Ohio 45433 N.S.F. (Materials Res.) Attn: R.J. Reynik 1800 G Street, N.W. Washington, DC 20550 U.S. Naval Research Laboratory Attn: Dr. R.W. Rice, Chem. Div. Washington, DC 20375 Professor T.Y. Tien University of Michigan Materials Engineering Ann Arbor, Michigan 48105 Dr. F.F. Lange Dr. A.G. Evans Rockwell International Science Center Thousand Oaks, California 91360 Dr. Fred S. Pettit Pratt & Whitney Aircraft East Hartford, Conn. 06108
Dr. Harold Posen
RADC-ESMC

Building 1126 Rm. 102

OTCBedford,

Hanscom AFB Mass. 01731 (5 copies)

RESEARCH (AFSC) AIR FORCE OFFICE OF SCiENTIF!C DDC TO TRANSIMITTAL OF NOTICE

and is This technical rpcrt has been reviewed (7b). 190-12 AFR IAN release approved fox' public Distribution is unlimited.
V,

Siii

A. D. BLOSE Technical Information Officer

_______________

ABSTRACT A data base is being developed for calculation of quasibinary and quasiternary phase diagrams of oxide and oxynitride ceramic systems. The current base covers combinations of A 2 03 , MgO, A1 2 03 , Fe2 0 3 , Fe3 0 4 , FeO, CaO, SiO 2 , Si3 N4 and AIN. Lattice Stability, Solution

and Compound Phase Parameters were derived covering the liquid, spinel, periclase, corundum alpha and beta quartz, crystobalite, tridymite and beta Si 3 N 4 structures in all of the quasibinary systems composed of the data base compounds. Evaluation of these

parameters permit calculation of the thermochemical properties of these systems in good agreement with limited thermochemical data. In addition, isothermal quasiternary systems of selected

oxide and oxynitride systems composed of data base compounds

have also been calculated over a wide range of temperature.

The free energy of formation of beta.sialons in the Si3 N4 -AI2 03 -AIN system and the X phase in the SiO 2 -AI2 03 -Si3 N 4 system have been calculated. Recent examples of the use of data bases and computer

techniques in solving problems associated with: in-situ formation of columbium, nickel and cobalt based superalloys, sigma phase formation, multicomponent electroptical semiconductors, sialon ceramics, molten salt, sulfide and oxide slags, high temperature battery electrodes, hardenability, synthesis of hydrogen storage compounds, grain refinement in aluminum alloys, metatectic reactions in hafnium alloys and the growth of HEM silicon solar cell materials. The initial results obtained by applying these techniques to CVD synthesis of doped germanium-borosilicate applications are reported. iv glasses for C31

'

~*

71

TABLE OF CONTENTS

I.

Introduction and Summary ................................ I a. Program Personnel............................... 4 b. Interaction with Other Technical Activities......4 Calculation of Quasibinary and Quasiternary Systems-l (CALPHAD,2 35 (1978).............. 5

II.

III.

Calculations of Quasibinary and Quasiternary Systems-II (CALPHAD, 3 27 (1979) ................... 25 IV. Calculations of Quasibinary and Quasiternary
Oxynitride-III (CALPHAD, 3 273 (1979) ................ 45

V. Utilization of Data Bases and Computer Techniques for Solving Metallurgical Problems (Proceedings of a Joint National Physical Laboratory Chemical Society of London Symposium of Industrial Use of Thermochemical Data-September 1979) ................. 62
IV. Thermochemistry of Doped Germanium Borosilicate Glasses for Communications Fibers ...................... 83

,: "

p~cTAB jU, ~tif Iai Unmnoun onde

,A

.st

speCi

"ik

Section I. Introduction and Summary The utility of computer based methods for developing and utilizing coupled phase diagram and thermochemical data on metallic systems has been well documented. Sections II and V of the current report contain suitable references and examples. Not withstanding the development and applications of metallic data bases the growth of similar facilities for ceramic and oxide systems has proceeded at a much slower pace. This was certainly the case in 1975 when the present program was initiated (see page 6). Accordingly, the current work was initiated in order begin the development of a data base for oxide and oxynitride ceramic systems which could be applied to computation of phase equilibria and stability over a wide range of temperature, pressure and composition. One of the objectives of this study was to develop the capability of dealing with SIALON systems which are based on combinations of Si N4, AlN, Al2 03 and SiO 2* Reference to Sections III and VI shows that this objective has been achieved through our own efforts and those which were carried on by other workers i1 the "CALPHAD group" (i.e. references 2,3 page 42 and references 3 and 4 page 61). Section 4 provides a number of examples of computed ternary sections through systems composed of oxynitride ceramics. The dataib"'se-which has been developea for the purpose of performing the calculations displayed in Section 4 can be employed to -exp Jore stable and metastable equilibrium in multi component oxynitride systems. The three technical papers contained in Sections Il-IVtrace the development of the present data base which currently covers Cr203 , A1203 , MgO, FeO, Fe 2 03 , Fe3 04 , SiO 2 , CaO, .Si N4 and AlN. Section II discusses the initial segment dealing with combinations of Cr203, MgO, A1203 , Fe203 , FeO and Fe3O 04 The twelve quasibinary systems and three quasiternary

1--4

e1 "I

A"

systems were calculated and compared with limited experimental

data. The development of lattice stability, solution and compound phase parameters for the liquid, spinel, corundum and periclase phases can be employed to calculate phase equilibria activity, vapor pressure and free energy of formation of solution phases and compounds in quasibinary and quasiternary systems composed of the above listed compounds. The ternary isothermal sections which were calculated in this work include the FeO 4 -Cr 2O-Al 20 3 system between 1600 and 2400K, the MgO-Fe3 O4 -Al 2 03 system between 0 1600K and 2850 K and the MgO-Cr 2 03 -Al 2 03 between 1300*K and 2800K. In the last case the computed results were found to compare favorably with experimental results. The second phase of the development covered in Section III deals with inclusion of CaO and SiO Each of the quasibinary systems between their compounds and the forgoing group (i.e. Cr2 03, Al 203, MgO, FeO, Fe203 and Fe3 04 ) was analysed permitting calculation of phase diagram and thermochemical free energy of formation of quasibinary compounds. This description covered the liquid, spinel, periclase, corundum, crystobalite, tridymite and the alpha and beta quartz structures which occur in these quasibinary systems. In addition, ternary isothermal sections were computed between 1600 and 2800K for the MgO-Fe 3 04 -Cr 2 03 system. The extension to oxynitride ceramic systems described in Section IV was carried out by developing lattice stability, solution and compound phase parameters for each of the quasibinary systems between Si3 N4 and AIN and MgO, Al203 and SiO 2. The beta Si3N 4 and hexagonal AIN structures were included along with the liquid and previously listed solid phase structures. The melting point and decomposition temperatures of Si3N4 and AIN at one atmosphere pressure were calculated in the course of tis analysis. Isothermal sections in the SiO 2 -AI 2 03 -Si 3 N4 and Si3 N4 -Al 03 -AIN systems were computed which are in good agreement with experimental findings. Moreover, the free energy of formation of the ternary compound phase X which forms in the

1-2-

T'-

<V

..

- -

42

SiO 2 -AI203 -Si 203 -A1N 3 N4 and the beta sialon phase of the Si3 N4 -Al system wure calculated. Comparison of the latter results with independent findings obtained by the Stuttgart group yields good results which can be further improved by iteration. Section V provides a listing of recent cased in which thermochemical data have been employed successfully in dealing with practical problems. Such documentation is difficult to obtain because the secrecy and confidentiality restrictions which surround economically viable processes necessarily limit publication of such information. Nevertheless, such examples come to light over a period of years. The discussion in Section V covers processing of in-situ eutectic composite formation in columbium, nickel and cobalt base superalloys, sigma phase formation in high temperature alloys, multicomponent semiconductors for electroptical devices, SIALON ceramics, molten salt, sulfide and oxide slags for ore reduction and refining, high temperature battery electrodes, hardenability calculations in alloy steels, synthesis of hydrc- n storage alloy compounds, grain refinement in aluminum alloys, metatectic reactions in hafnium alloys and the growth of HEM silicon solar cell materials. The final section of this report discusses application of the present computational methods to the CVD processing of doped germanium-borosilicate glasses. Section VI illustrates the utilization of Data Bank information in computing the temperature/pressure range over which the halides of potential dopents can be employed effectively in the CVD processing of preforms. As the first step in an analysis of these glasses, Section VI contains an analysis of the SiO 2 -GeO 2, SiO 2 -B203 and B2 03 -GeO 2 quasibinary systems as well as the SiO 2 -GeO 2 -B 203 quasiternary system over the temperature range from 1000K to 1600K. Future activities will include consideration of the addition of Sb205, P205 and CeO 2 to this system.

-3-

i.77 T'7>
*:. ____ ____ ___ ____

7
-AM

Section Ia.

Program Personnel

Technical activities on this program have been performed by L. Kaufman, H. Nesor, E.P. Warekois, D. Pendse, J. Smith, G. Zilberstein, C. Thielen, K. Taylor, A.P. Miodownik, and J. Nell during the course of this study. Section Ib. Interaction with Other Technical Activities

During the course of this study technical presentations covering this work were given at the ASM/AIME Materials Congresses

in Chicago, Illinois (October 1978) and Milwaukee, Wisconsin (September 1979), CALPHAD VI-Harwell, Oxford England April 1977, CALPHAD VII-Stuttgart, West Germany (April 1978), TechnionHaifa Israel (May 1978), CALPHAD VIII-Stockholm Sweden (May 1979) and National Physical Laboratory/Chemical Society of London Symposium on Industrial Use of Thermochemical Data, University of Surrey Guildford Surrey England (September 1979). Specific interactions with scientific personnel at the Max Plank InstituteStuttgart, West Germany, University of Surrey, Guildford Surrey, England and at the Israel Institute of Technology-Technion City Haifa Israel were most helpful.

-4-

7, "0",
77,>

7F

Section 11.

Calculation of QuasiBinary and QuasiTernary Systems 1, CALPHAD (1978) 2 pp 35-53

7-P7

CALPHAD Vol. 2, No. 1, pp. 35-53. Pergamon Press, 1978. Printed in Great Britain.

CALCULATION OF QUASIBINARY AND QUASITERNARY OXIDE SYSTEMS - I* Larry Kaufman and Harvey Nesor ManLabs, Inc., 21 Erie Street Cambridge, Massachusetts 02139, USA ABSTRACT. A data base is being developed for calculation of quasi-binary and quasi-ternary phase diagrams of ceramic systems. The initial segment of this base covers combinations of Cr2 3, MgO, A 2 03 , Fe2 0 3 Lattice Stability, Solution and Compound Phase Fe3 04and "FeO". Parameters are presented covering the liquid, spinel, periclase and corundum phases in systems composed of these oxides. These parameters permit calculation of the thermochemical properties and phase diagrams of combinations of these oxides in good agreement w-.th limited experimental data. Twelve calculated quasi-binary phase diagrams and three calculated quasi-ternary phase diagrams are presented illustrating the applicability of the data base.

1. Introduction The utility of computer based methods for coupling phase diagrams and thermochemical data for metallic systems has been demonstrated in many papers published in this journal. A considerable effort is being applied toward Numerous examdevelopment of a suitable data base for metallic systems (1). ples have been published showing how such data on metallic systems can be synthesized to successfully predict the thermochemical properties and phase relations in ternary metallic systems (1-12). Such predictions can be applied practically to explore conditions for in-situ eutectic growth, metallic glass formation, spinpdal decomposition, metastable equilibrium, and diffusionless transformations. In a recent workshop on "Applications of Phase Diagrams in Metallurgy and Ceramics" held in January, 1977 at the National Bureau of Standards in Gaithersburg, Maryland (13), many informational gaps in ceramic phase diagram data were noted and the need to develop computational techniques to bridge these gaps became evident. Such efforts have been proceeding individually for some time (14,15). However, in view of the success achieved in developing a tractable data base for metallic systems, it seemed to be an auspicious time for launching the development of a similar data base for ceramic systems. This is particularly true, since the need for such information is apparent (i.e., metastable reactions are quite common in ceramic systems,pressure is an important variable, etc.) and experimental data is limited and difficult to obtain. Accordingly, the present work should be considered as a beginning, much like the initial attempt at estimating the lattice stability of metals (16,17). The only difference is the experience gained with development of a data base for metals over the past ten years and the availability of suitable computational programs.

*This work has been sponsored by the Air Force Office of Scientific Research, Bolling AFB, Washington, D.C., under Contract F44620-76-C-0060.

6-

L. Kaufman and H. Nesor

0(L

cIxy.'j M

GIT L

LL

LE ELL

(A)

(B)

Figure 1. Illustration of Quasi-Binary Approximation of the Ext:ess Free Incr~v of Mixing of Solution Phases, EGL in Terms of the Ternary Approximiation to the Free Energy G (x,y,T) [Mass Basis FEqual to One Gram Atom or One Mole of Atoms).

2. Description Of The Thermochemical System Employed To Describe Solution And :cmpound Phases The method utilized for describing solution and compound phases is essentially the same as that employed earlier (1,3), except that some symboiic usage has evolved which facilitates data handling as indicated below. The free energy, GL, of a liquid (solution) phase, L, in the binary system I-J is given by Equation (1) where T,is in Kelvins, x is the atomic fraction of J and n mole of atoma (i.e., a gram atom). the mass basis i!;
GL,(l-x)* 0 L + *GL + RT(xtnx + (l-x)Xn(l-x)) + x(l-x) (LIJ(l-x) + xIJi]()

In this expression, 06L and eGt. are the free energies of a gram atom of pure liquid I and J, R -8.314 J/g.at.*K, and LIJ and LJI are functions of temperature (but not of compostion). Similar equations can be written

-7,-

_4_ _ _ _ _ _ _ _ _ _ _ _ __ _ __ _ _ _

<!'f-s'

..~ .4 .

7,77"

QUASIBINARY AND QUASITERMARY OXIDE SYSTEMS

TABLE

Lattice Stability, Solution Phase and Compound Parameters Employed for Calculation of Several Metal - Oxygen Systems (All units in Joules per grdm atom (mole of atoms), T in Kelvins) Lattice Stability Parameters FEFEVL = 371200 - 118.4T, ALALVL = 301250 - 107.9T 3410 - 37.9T CRCRVL = 367360 - 120.5T, OOOOVL = 4.2T 230,MGMGVL = 133890 - 98.3T, OOOOLB = Solution Phase Parameters 0 to SO a/o Oxygen LFEOO LOOFE LCROO LOOCR LMGOO LOOMG LALOO LOOAL 0 -1000810 0 -17656S0 -90790 -1940120 -1287830 -1287830 + 125.ST + + + + 234.3T 411 3T 626.3T 147.7T 147.7T (Base Phase is BCC) Structure Periclase (P) Spinel (S) Corundum (C) Corundum (C) Periclase (P) Corundum (C) 50 to 100 a/o Oxygen 423420 -1424230 -882820 -882820 -90790 -1940120 -1287830 -1287830 + + + + + 4S6.1T S81.6T 117.2T 117.2T 411.3T 626.3T 147.7T 147.7T

Compound Phase Param6ters Compound Fe0 50.5 Fe0 42900.571 Fe0 400.6 Cr0 4006 Mg0
5 00. 5

Compound Parameter 28030 + 12.5ST 40170 + 11.63T 51040 + 11.71T 112130 - 8.37T 182004 - 16.74T 96650 - 2.51T

AX 0 400.6

etc., for the B=bcc, for each solution phase with GJ replaced Uy G*, G E=hcp phases etc., and with LI? replaced by BIJ and EIJ etc. In the present symbolism, one uses the symbol FEFELB to designate the free energy of pure liquid iron minus the free energy of pure bcc form of iron. Thus FEFELB= -FEFEEL. Similarly, the free energy of pure liquid chromium minus the free energy of pure bcc chromium is designated by CRCRLB. If I=iron and J=nickel in Equation (1), then Table 10 of Reference 1 showz that LFBNI= -8368 + 2.72T (J/g.at.) and LNIFE= -32217 + 9.205T (J/g.at.). Figure 1 and Appendix I shows how the free energy of a solution phase can be described along a specific quasi-binary join in a given ternary system. This approximation is used here to reduce the ternary description to a quasi-binary description in order to employ the information stored in the data bank to deal with high order systems. The free energy. is defined by Equation (2), (see Equation 14 compound G0, of (1)) as phase i(1-x*)ix 60 aof page G. * (G'-x,) *G 1 + xG + x,(1-x,) [(l-x,)LIJ 4 x,LJI-CtT]1

(2)

where 0 refers to the base phase selected for the compound and CIT] (which is a temperature dependent function) is the compound parameters. Tables 1-4 define the iron-oxygen, chromium-oxygen, aluminum-oxygen, magnesium-oxygen, ironmagnesium, chromium-magnesium, iron-aluminum, iron-chromium, aluminum-chromiin,

-8-

L. Kaufman and H. Nesor.

TABLE 2 Comparison of Calculated and Observed (18,20) Free Energy of Formation of Binary Oxide Compounds (All units in Joules per mole, T in Kelvins) (P-Periclase, S-Spinel, C=Corundum) Compound Free Energy of Formation Calculated Observed FeO (P) -267860 + 67.1ST -271120 + 66.94T Fe 3O0 -1166600 + 382.21T -1086S80 + 292.88T 4 (S) Fe 0, (C) -8943.30 + 322.59T -803328 + 2Sl.04T O.SFeO-0.l2SFe + 0.125 Fe 3O0 4 ; AG a -11900 0.133Fe3 0i-. 03340.,*0.2Fe 03 AG =-23700 Cr 03 (C) T-1000 1500 2000 At 203 (C) T-1000 1500 2000 MgO (P) Tz 500 1000 1500 2000 2500 3000 Calculated -1204990 + 276.77T -874460 -747260 -618810 Calculated -1359800 -1202900 -1046000 Calculated -552960 -499360 -426810 -322210 -217610 -113010
+ +

14.18T; To 13.71T; To
+

839K 1726K

Observed -1129680 -928220 -789830 -651450 Observed -1359800 -1196620 -1035540 Observed -547520 -492460 -423840 -319240 -215060 -110460

2S1.04T

TABLE 3 Solution Phase Parameters for Several Binary Metal Systems (All units in Joules per gram atom (mole of atoms), T in Kelvins) LFENG - 64850 LMGFE - 64850
LCRMG - 510SO LMGCR - SlOSO

LFECR

= 87000 - 10.46T LCRFE = 87000 - 10.46T LALCR = -46440 LCRAL -- 46440 +

LFEAL LAV!E

-62760 -96230

4.60T 33.05T

LALMG = -10040 LMGAL - -10040

+ +

3.35T 3.3ST

---

[7

QUASIBINARY AND QUASITERNARY OXIDE SYSTEMS

TABLE 4 Ternary Liquid Solution Phase Parameters (For Several MetalzMetal-Oxygen Systems) (All units in Joules per gram atom (mole of atoms, T in Kelvins)
System TRNL. System TRNL

At-O-Cr
Fe-O-Cr Cr-O-Mg

0
1439300 912.1T 2217520 - 125S.OT TABLE

Fe-O-Al
AL-O-Mg

1757280 - 1125.5T 1882800 - 1121.3T

Lattice Stability Parameters for Oxides (All units in Joules per gram atom (mole of atoms), T in Kelvins) Symbol C S P Abbreviations Example HR 10 At2 03 CF 56 Fe3 04 (Co3 4 ) Periclase CF 10 MgO (NaCt) WO - Fe0.500. 5, FO = Fe0.400 .6 HO = Fe0.42900.571 Corundum Spinel AO = At0. 4 0 0 .6, CO = Cr0.400 .6 WOWOLC - 7280 - 7.28T WOWOLS - 9017 - 6.44T WOWOLP a 14016 - 8.54T FOFOLC - 18117 - 8.37T FOFOLS a 15355 - 7.29T FOFOLP - 9414 - 6.28T HOHOLC - 9707 - 9.71T HOHOLS - 15816 - 8.45T HOHOLP - 12217 - 7.20T MO g0 .500. 5 Name Type

AOAOLC - 23640 - 10.21T AuAOLS - 19205 - 9.08T AOAOLP - 4477 - 12.84T COCOLC - 34350 - 13.47T COCOLS - 22594 - 12.SST COCOLP = 34058 - 15.48T MOMOLC - 22594 - 12.55T MOMOLS -u 19560 - 10.46T MOMOLP - 47698 - 15.40T

and aluminum-magnesium systems in order to apply the description shown

in

Figure 1 and Appendix I. In these Tables, L refers to the liquid phase while B and V refer to the bcc and vapor phase respectively. The symbol MG is used for magnesium, AL for aluminum and 00 for 1/202. Thus, the basis in all cases is one mole of atoms or one gram atom. The numerical values chosen for the lattice stability, solution and compound phase parameters were derived from relevant thermochemical and phase diagram data for oxygen-containing systems
(18-24) supplemented by previous assessments of lattice stability (1,17) and solution phase parameters (i.e., iron-chromium (1,2), iron-aluminum and aluminum-chromium (2), aluminum-magnesium (9)). The selection of the ternary liquid interaction parameters for the aluminum-oxygen-chromium, iron-oxygenchromium, chromium-oxygen-magnesium, iron-oxygen-aluminum and aluminum-oxygenmagnesium systems shown in Table 4 was made on the basis of the data contained in Reference (20). The description of Tables 1-4 permits calculation of the thermochemical properties of the above noted oxide systems and calculation of their phase diagrams under standard (one atom phase pressure) conditions. In addition, it permits calculation of the excess free energy of the liquid phase in the quasi-binary join of a ternary metal-oxygen-metal system as shown in Figure 1 and Appendix I. This facility will be employed subsequently. The calculated iron-oxygen and chromium-oxygen phase diagrams resulting from this

-10-

L. Kaufman and H. Nesor

1/a At203L 10
TIZ Iiqi .d Fu * Vapor' SO, 2130 2S t i

"2!
L 0.7 en Pase

P.2
172

Liquids19165 232
F r
3

5
CrO A : 3 C
0

igur

l C lcu 2 .

a1 1oedI80y 0.4S Fe.29

0 0 5 1 3 8

FCC L
1200 OCC - Fe0.O.S

_7
00.6* Feo0.4 240 .

0400

39 600
1 C 0 Fe10042900.S71 10 0

C 700

Corundum

e 2

. .6 r0. 4006

0.4006

40O6

Figure 2. Calculated Iron-O

Phase D xygen

iagrams.

Figure 4.

Calculated Cr 0 .400.6

Ago40.6

Phase Diagram.

A120S tS F. .C
54 06

g S

2o

t 78

46 93 eoo

Cr 4ro.00, VapO0r

CrF &CC - Liquid 1600 inhIe $00 C Cr .400. 6e

W 1200 Cro.4Ou. 6 600 ui (71

W2

rA C atVapor /, o

1670

OLE!9

.C into Fe an31Sa torundum p o e 17 6 ) 0 4ap00 Miscibility

Feo0.400. 6

C1"

Cro0.400. 6

Ato0.4O0.6,

Pilure3.

Calculated Chromium-Oxygen Phase Diagram.

Figure S.

Calculated Ao.400.6-Fe0.4010.6 Phase Diagram. 0tb y 6 Decomposes Fe0. Figure 2 shows (Note: nd that I2 bove 1726K). into Feo.20,7

L, pielSperclse Pan counum

description are shown in Figures 2 and 3.

th oxde A 2, , ucre of.?,

These diagrams are in reasonably

good agreement with the observed diagrams if one allows for the idealizations inherent in treating iron oxide (wustite) as a line co mp ou nd at 50/50 stochiomerry. e. The Lattice Stab(lity Solution and Compound Phase Parameters lor the Spinel Periclase and Corundum Structures
Table S shows the values derived for the lattice stability of the liquid, L, spinel, S, periclase, P, and corundum, C, structures of the oxides AZ203 ,

Fe203 and "FeO"l. These lattice stability values were 4-15. phase values diagrams shown infor Figures employed compute atom It are given a gram noted that the the quasi-binary lattice stability should be to
Cr203, MgO, Fe304,

(one mole of atoms) of each of the oxides! Moreover, examination of Table S shows that a gram atom of each of the oxides (i.e., Cro.4Oo.6] has been assign-

The g-atom free energy of the ed a specific abbreviation (i.e.) Co z Cro.4Oo.6. liquid minus the corundum form of Cro.4Oo-6 is designated by the symbol COCOLC
and equals 34350 - 13.47T Joules as shown in Table 5.

In direct analogy to the,

discussion of solution phase parameters in Section 2 above, the solution phase

3

17'
'-1-

v-

Z,

QUASIBINARY AND QUASITERNARY OXIDE SYSTEMS - I

TABLE 6 Quasibinary Solution Parameters for Oxide Systems (All units in Joules per gram atom (mole of atoms), T in Kelvins)
LAOCO SAOCO PAOCO CAOCO LAOHO SAOHO PAOHO CAOHO = = = -7422 -7422 -7422 -7422 + + + + 13.05T 13.05T 13.05T 13.'0ST LAOFO SAOFO PAOFO CAOFO LAOMO SAONIO PAOMO CAOMO
= -14297 + 5.65T = -14297 + 5.65T 6623 + 5.65T =

6623 + 5.65T

= -20350 + ll.2ST = -20350 + 11.25T = 570 + 11.25T 570 + II.25T = 25.73T 25.73T 25.73T

= -33530 + 22.93T = -33530 + 22.93T = 29230 - 4.69T = -12610, + 22.93T 13.05T 13.05T 13.OST

LAOWO = -39330 + SAOWO = -39330 + PAOWO = -18410 + CAOWO = -18410 + LCOFO -- -26080 +

25.73T
19.16T

SCOFO = -26080 + 19.16T

PCOFO = -26080 + 19.16T

LCOAO = -7422 + SCOAO = -7422 + PCOAO = -7422 + CCOAO = -7422 + LCOHO = -37780 + SCOHO = -37780 + PCOHO =

13.05T
26.19T

26.19T

-4310 + 15.73T

CCOFO =
LCONIO SCOMO PCOMO CCOMO LFOAO SFOAO PFOAO CFOAO LFOMO SFOMO PFOMO CFOMO LHOCO SHOCO PHOCO CHOCO
=

7390 +
780 780 21700 21700 + + + +

8.70T
7.45 7.45 3.26 7.45

CCOHO =
LCOWO SCOWO PCOWO CCOMO LFOCO SFOCO PFOCO CFOCO LHOAO SHOAO PHOAO CHOAO LHOMO SHOO PHOMO CHOMO

-4310 + 15.73T
44.22T 44.22T 44.22T 33.76T 18.49T 18.49T 18.49T 18.49T

=
= =

= -70760 + = -70760 + = -49840 + = -37290 +

=

= = =
=

58150 - 30.03T 58150 - 30.03T 79070 - 30.03T 79070 - 30.03T

+ + + + 26.82T 26.82T 14.27T 26.82T 15.44T 15.44T 15.44T 1S.44T

= = =
=

41000 41000 41000 41000

= -51350 = -S1350 = -9510 = -30430

=

=
=

59390 - 35.77T 59390 - 35.77T

803"0 - 35.77T 80310 - 35.77T + 19.75T + 19.75T + 1.38T + 19.75T

=
= =

37810 37810 37810 37810

= -30040 = -34220 6360 = = -9120

LMOAO -

SMOAO =

PMOAO = CMOAO -

59910 - 27.11T 59910 - 27.11T 122670 - 54.73T

80830 27.11T

LNIOCO =

87520 - 39.96T

SMOCO =

PMOCO = 108440 - 44.14T CMOCO = 87520 - 39.96T

87520 - 39.96T

LMOFO = -22340 + 13.56T SMOFO = -22340 + 13.56T 18500 + 1.OOT PMOFO = CMOFO = -2420 + 13.56T
LMOWO a SMOIVO = PMOWO = COWO a 16890 16890 8520 8520 4.06T 4.06T 4.06T 4.06T

LMOHO - -10370 + l0.54T

SMOHO = -14556 + 10.54T PMOHO = 26030 + 2.18T CNIOHO - 10550 + 10.54T LWOAO SWOAO PWOAO = CWOAO = 63180 63180 84100 84100 31.71T 31.71T 31.71T 31.71T

-12-

"7

7,~

L. Kaufman and H. Nesor

TABLE 6 LWOCO a 31440 SWOCO a 31440 PWOCO - 50840 CWOCO - 31440

-

(continued)
+ + + +

10.007 10.007 10.007 10.OOT

LWOMO a 16800 SWOMO a 16800 PWOMO - 8430 CWOMO - 8430

3.93T 3.93T 3.93T 3.93T

TABLE 7 Quasibinary Compound Parameters for Oxide Systems (All units in Joules per gram atom (mole of atoms) T in Kelvins) Compound (AZL. 2 86 00 .42 8 ).M
1 4 3 %O 1 43 ) 0 1 43 )

Structure (SP) (SP) (SP) (SP) (SP) (SP)

*Base (S) (S) (S) (S) (S) (S)

Compound Parameters 74050 40580 134720

+
-

(AX0.28 6 00 . 42 8).(Fe. 14 6 (Cr

0

7.11T 7.11T

.M

0 4

(Cr0.2 86 (Fe0 . 2 86 (Fe0 . 306

14 3O 00 1 4 3 ) 0 4 28 )'(Mg 0 . 00 1 4 3 ) 14 3O 0 4 28 )'(g 0 . 0 4 08

38910 + 16.74T 18410

+

13.72T

"M9

0. 1 43O 0 0 14 3 )

ThBLE 8 Comparison of Calculated and Observed (20) Free Energies of Formation of Complex Spinel Oxides (P-Periclase, C-Corundum, SP=Spinel) Free Energy of Formation, Joules, T in Kelvins Reaction FeO(P) MgO(P) MgO(P) P.0(P)
+ + + +

-40070 -39740 -40770 Cr2 0(C)+*MgO-Cr 2 0 3 (SP) -41120 -36961 At 2 03 (C)-MgO-A' 2O3 (SP) -35560 -46100 Cr.0 3 (C)eFeO.Cr2O0 3 (SP) e 03 e. C)Mg~eO 3 C)-45190 -20940 20g3(C) -19330 A9 2 0 3 (C).FeO.AI 2 03 (SP)

+ +
-

6.74T 7.12T 13.03T 14.007 3.22T 2.05T 8.32T 8.38T 2.20T 2.057

(CALC) COBS) (CALC)

(OBS)

(CALC) COBS) (CALC) (OBS) (CALC) COBS)

parameters for one gram atom of the liquid on the A12 01-MgO Join of the At 0-Mg ytem are designated by LAOMO and LMOAO. These parameters have been

comutdsfrom Tables 1-4 and the procedure outlined in Appendi Iadrels displayed in Table 6. Reference to Table 6 shows that LAOMO w-33530 + 22.93T and LMOAO - 59910 - 27.11T respectively. phssTables 6-8 summarize the solution phabe parameters for the remaining phss(i.e., C, corundum, S, spinel, F, pe ri clase). Thus the solution paraotn meters for one gram atom of the corundum solution in the Ata.40 0 6-Cre.0. 422 + of the At-0-Cr system, can be read as CAOCO -- 7422 + M3OST and CCOAO z 13.05 Joules respectively. Tables 7 and 8 display the compound parameters in Figures 4-15. employed in calculating the quasi-binary phase diagrams showh generate the free Table 8 shows that these same parameters can be employed to energy of formation of complex spinel oxides in agreement with observation (20).

=1

___________________

__

;7

QUASIBINART AND QUASITERNARY OXIDE SYSTEMS

V/0 A 2C 3__ 6 is 21 19

38

48 -C

FeO

12

20 L;-

28

w/o At 20 3 36 46

S7

70

84

100

[=L-Liquid lgO0 S SpinetSolidSolut ion 1100

~1800

li0

14

P"
lD0 1140 Misibiit S2 60070 Fe 7;il" At0. 0 0 0 FeO 0 (

ISM
() Spinet (SP)

~1200Peils
TPAS

0.714

A1 .4006

.Fe (Ato.29 0 14200.141) 60 0.428)'SP Figure 6. Calculated Fe0 42900 S7' AtO 400 6 Phase Diagram. Figure 8. Calculated Fe0 . 5 00 .SAt 0 . 4 00 6 PhaseDIagram. (Note: Figure2 howsthat Fe intoFe and Fe 0 O05 Decomposes 0 42900571 below 39). 0 3 10 I so 21 " 32 41 51 I" 61

/o At T1 0 20 " 300 s00 0

NCO 00 I

=4

71

90 I

10

I Liquid(L)

so0

0 231S 2380L.PCorundum

2100 (C) perilaso TASABLtI

20
1000

501400

700

*( 04140

Figure 9.

Calculated Cr0 .400. 6 -Feo.400.6 Phase Diagram (Note: Figure 2 shows that Fe 0 .400.6 Decomposes.
into F00 .42900.071 and2i. .bove 1729K.

Figlure 7. Calctilated At 0 .400. 6 -Hgo 0

o.

Phase Diagram,.

Thus the definitions of Equations Cl) and C2) coupled with the quasi-binary approximation shown in Appendix I and parameters shown in Tables 1-8 generate the one atmosphere, thermochemical properties and phase diagrams for the quasi-binary joins displayed in Figures 4-15. 4. Discussion Of The Results Of The Quasi-Binary Approximatiorr

Table 5 provides the "liquid-solid" free energy differences (and the lattice stability parameters) for the six oxides of current interest. Thus, AOAOLC is equal to the free energy of liquid At0 ,0 8 (i.e.,I/SAZ,0,) minus the free energy oi the corundum form of Ato.0 0 : Information on the heat 6. of fusion for the st:,ble forms of each oxide was taken from review publications and recent experimental studies (18,20,25). As indicated earlier, AOAOLS and AOAOLP represent the corresponding free energy differences between the liquid and spinel and the liquid and periclase forms. This description permits calculation of the free energy differences between the C, S and P forms since

* - -.

' I

_- -

..

. .. . ..

1 4-..
*I4

. ..

...

L. Kaufman and H. Nesor

"/a Cr 20 3 -0 1 94 19 29 39 48 SO 69 80 89 100 FeO

o In at Ip

C r 20 3 . . . .

w All
77

2100

20

700

C700

WAALI Spinet(SP')

F 0.429 O.S 71 Figure 10. Calculated Fe 0 . 5 7 1 'Cr. 4 2 0O 0 . Phase Diagram. "/o u"so Se0 7 14 22 10 Afl "1
61
4 0

Cr 0 .400.6
6

(Fe 0. 1 4 690. 0e.SO0*S

1 4 3 )CCr,

-8 6 0,,

28 6

)'Sp 074CO10 2 (Nute: Figure PhaseDiagram. intoFe and Fc 0 42900.S7l

Fe 0O 0 . Figure 12. Calculated 40 5 .Cro. thatFe show~s 0 so halW

s Decompoes

72

96 10(

Fe0

w/.) M1 27 S 39

91

S59

3000 2SO 27 2000 100090 boPericlase

L-

0$ (I')

3000 L.C 2000 9S

39

SPSI'

aho00that Ce

4 0

DoP

oe

isoF

420

l.13T0J/g4at 4435 AOAOLS of AL0 ,6 00 .6,0AOAOSC,is.equal.to AOAOLC he eigamNtlpguey for006"0.. drw acuante 13 A 02 asla concusio ) reslt obere
-. -

dieffrene

ofrg 272ogaf

spinel the ewe

e.Oo formso h ndry corundum aommnstefe

derived from the difference FOPOLC - FOPOLS given in Table 5 (26,28). The lattice stability parameters shown in Table 5 were derived by using experimenperforming in"t'e missing components by tal values as a base and "filling 4-15 as in the case of metals (16,17). an "ideal solution" iteration on Figure-, fixed by are 6 Table in shown liquid the for The solution phase parameters metal-oxygen, the for results the I and Appendix in outlined procedure the metal-metal and metal-oxygen-metal systems shown in Tables 1-4. Reference to Table 6 discloses that the corresponding values for the S, P and C solid solutions are usually equal to, or more positive than the L parameters. The S, spinel solution parameters are usually equal to the liquid parameters while the periclase, P, and corundum, C, parameters are usually more positive than

v-15-

rgr

QUASIBINARY AND QUASITERNARY OXIDE SYSTEMS

L and S (and equal). These small differences were obtained by iterative calculation of Figures 4-15 starting with equal values of L, S, P and C. Finally, the quasi-binary compound parameters shown in Table 7 were fixed to yield conformity with the phase diagram and the observed thermochemical

W/o Hgo Fe34 6

1 2 29 3

S9 71

8S

100

3000

T-ta shown in Table 8.

5. Calculation of the Activity of Quasibinary and Ternary Components of Solution Phases

2000

' 1000 o, . ,
1

s,.

Fe0 .4 2 900. 57

8O.s0
00. 5

Defining the activity of

quasi-binary components across a given join is quite straightforward (2) based on the defini-

Figure 14. Calculated Fe0 .42900 .S7 1 -g 0 , Phase Diagram.

/oHo
FeO 6 12

tions of Equation (1).

Thus,

31,

46

s7

70

84

too

the activity of CO and AO in the C phase of the CO-AO system (Fig. 4) where x is the fraction of AO and the reference states are pure corundum CO and pure corundum AO is given by Equation (3) and (4) as

, L
3000

2000

C Rfnaco

RT~n(1-x)

(3)

100

14

2 +x [CCOAO+(2x-1) (CAOCO-CCOAO)]

ME TABLEI

and C = RTnx
Fe0.50.5
M0.$00.S

RTinaAO
+ (l-x)
2

(4)
[CCOAO+2x(CAOCO-CCOAO)]

O.5 -HgO.sO0 . Phase Figure IS.Calculated Fe0 .s

Diagram. (Note: Figure 2 shows that Feo.$O 0 5 Decomposes into Fe and

Similar expressions can be Feo., 29 Oo.$T1 Below 8390. generated for each of the phases. Derivation of equations for the activity of the ternary components (i.e., Cr, 0 and At) can only be carried out for the liquid phase where the free energy has been explicitly stated in terms of ternary components (and not in terms of end members on a given binary join). To illustrate this proceL to compute RT~naA, dure, consider the liquid At-O-Mg case where it is desired PL /,L and In the case of the first two components, aL A At 0AM L L pAL and pL are the vapor pressures of Al and Mg over the AL m and pL are the vapor pressures of the pure metals at the temperature in question. A similar result holds for the oxygen with a / L 2 Lp cmpictin slgh uetoL 2 use of (1/2)02 sa component! Thus ao= (p cthe 0 / en at oneatthhe 37.9T O.5RTtnlbp + 2ROOOOIOV and since L1 is L where0 L02 L the pressure in atmospheres.Th weep0 R O 0 680+7.T RTn02 and RT.naL RTfnaL L 'kgL Pi Pg where aut = gMgL liquid oxide and opL equations for component activities and vapor pressures can now be written explicitly as Equations (S)-(7) for the AL-O-Mg case where x is the fraction of 0 and y is the fraction of Mg. Note that along the A0. 4 00 .6 -Mgo. 5 0 0.5 join the

Z.

-16-

L.Kaufman and H. Nesor

fraction Of 0 varies from x 6 Equations LCS)-(7) apply to *0.6 when y w 0 to x 0. all values of x and y (17). -0.2y. Nevertheless, ALjnj~ RT2IiCI-x..y) * x2 MAoo +,y2 LALMG + xy(LALOO + LAL,G _ LOHMG + CZx -y ( LOO AL-LA LDO ) . -i ( MG A ..I LHG (l-2 ( -11(x)LLA4;
_X-2 (Xi+Y)

(LMG0-LaOMIG
2

xY[l -2 (x~y) TRNLCA-oMg)

2 LLO + Y L'ALHG Y(1-)[LALOO + LALMG
-11

(S)

a = RT~nx + (laX)
+ X2 1 (1-Y)

( 2-3L 2y
xCl-Y)y

LfY3HG]

(W0AL-LALOO

1 -

LL L G

2
(X+Y)2 L2 Stna

MGLAM)

CLMGO-L00M)
2

+ (Y(l-.Y)-xy(3-2Cx+y))] TRNL CAt-0.-Mg) ALOO + LALMG LOOMG]

RTn

1 AL00

(I _y) 2LALXG-x(z..Y) fX2j +

2

~y~

I (LGAL-LALMG)

(7)

(~y) 2{2x+yxy.y ]

(G0o..W0MG)

x~l-x)xy(3..zx+Y))] TRNL (A.-O-Mg)

the activity and vapor pressure Of each elemental component in the liquid Phase at any com-2M Position- Conversely they can be used to calculate the composition at I given OXygen pressure! 6. Calculation of Diagrams The extension to quasi-ternary systems ca be eadlyaccomplished by letting each Of comXponents i.e., CO,the AO, F0 etc., Play the role of the elements in Appendix I and References 16-21 show the resulting isothermal sections comnWued for the Pe,0 4PCr 2 O,-AXA0A, NgO-Fe30o 4 Mt2 0 I and MgOCr 2 o,Ax208 at Selected tbm-A peratures. In line with previous practice,

' 2%S1*O49.7 be emnloyed to calculateF0049.S1.0.90,7

These equations Can

2400t

10

Liquid

LConwmdumc

A& 0 . 400. 6 O

000k SAL

4"0 ()S

t Cr

Figure2

4 0

Cslcuisee4 liothermai sections in the F#0

A0406Sse.049051
04 6 Syt.

cO0 r0406

y-

QUASIBINARY AND QUASITERNARY OXIDE SYSTEMS

the free energy of ternary solution phases are synthesized from their binary edge components by using Kohler's Equation (Appendix I) with a zero ternary
interaction parameter. The free energy of ternary compound phases is defined

along specific joins (see Reference 2, p 2117, Eqs. (1,8). As an example, we consider the spinel compound phase in the system AO-MO-CO (see Figure 21).
This compound runs between the A20.MgO and Cr2 Os.MgO compositions designated 8 )'2o(Mg. 1 3 Oo. 1 4j. If x is the fraction MO and y is the fraction CO then the free energy of the spinel (SP) compound AO iMOx1 - CO(1x)MO is given by Equation (8) as

(Table 7) as (Ato.aasOo. 0e).(Mgo.1 2 3Oo.is) and (Cro. 2 06 Oo. 4

GSP

S * xGS + YGco S + RT[yiny(l-x-y)ln(l-x-y)-(l-x)Ln(l-x)] (l-x-y)GO +I-

X 6A

(-;Y -)AG
l-x B

+ AiG

(8)

where

LAOMO xLMOAO "(l-x,) CA] AGE - 0, AG. (1-x.)x, (l-x)x, [(l-x,) LCOMO + x, LMOCO - CB](9 AG and11
x, + yp ; p = (x, - x )/(I-x,), x' = 0.286, x, 0.286, p 0 (10) CA = 74050 and CB = 38910 + 16.74T 7. Discussion of Results The present description of the iron-oxygen system summarized in Tables 1 and 2 which generate Figure 1, lead
e. 429 00371 Fe 0. 429 0 . 571

(From Table 8)

to the calculated expressions

for the free energy of formation of "FeO", Fe3O4 and Fe 2 0 3 shown in Table 2. These calculated free energies of formation can be employed to examine the decomposition of "FeO" into iron and Fe3O 4 on cooling or the decomposition of Fe2 O3 into Fe3O4 and oxygen gas on heating. Table 2 shows these calculations which are the basis upon which the reaction isotherms shown in Figure 2 were constructed. Any self consistent sv.tem of thermochemical data for this system would be required to generate the appropriate values of To for these decomposition reactions (i.e., 839K and 1726K respectively). Nevertheless in many instances,, this test is
not made!

ISS09

60

s
C C

I=(
. 00 . 4 6

o0.

0.4.6

AA
Calculated Isothermal Sections in the Fe

0 0 cro.O.0

Figure17.

0 .429 037C

Figure 2 shows that "FeO" is not stable below 839K. Consequently the quasi-binaiy Joins which

-,o. 400 6 "

0.40 0. 6

Syste."

___7

L. Kaufman and H. Nesor

include the Feo.500.5 (i.e., WO) component (e.g., Figs 8, 12 15) Y.03o.s %.s%.S
P 28SOK
L.P 2500K

are

metastable with

respect to the decomposition of WO below 839K. In a similar

fashion, Figure 2 shows that Fe 0 .,O0.6

decomposes above 1726K. Thus each of

the joins containing
L.Friclase (P)

the component (FO) is

metastable above 1726K. As a consequence,

Figures 5, 9 and 13 illustrate this metastability. Thus the lattice stability,

Liquid M.) At.4o. 6

0.06

L SP(AIo.2 600 . Fe o.30O60.408) 42 h.

Fe .0 42 9O0.$7 C.Corundu
PPericlae

"'0.
2350
.

.4.) 413 01

solution and compound phase parameters for the elemental components, and their

22SO

2P. sP.P
SP.P.L

binary combinations,
shown in Tables 1-3, were employed to characterize the
thermochemical pro-....

SPS
SP.L
L 0.400.6
Figure 18.

CL

perties and the phase

diagrams of the iron-

0.42%.S7
Calculated Isothermal Sections in the Hg0.500.s-Feo.42900.571 -

oxygen, aluminum-oxygen, At0. 4 00. 6 System. magnesium-oxygen, chromium-oxygen, ironmagnesium, chromium- magnesium, iron-aluminum, iron-chromium, aluminum-chromium and aluminum-magnesium systems. These binary systems form the edges of the At-O-Cr, Fe-O-Cr, Cr-O-Mg, Fe-O-AZ and At-O-Mg ternary systems in which the quasi-binary "joins" of interest lie. Therefore, the development of the present base is built on defining the three edge binary systems (i.e., generating a small number of lattice stability, solution and compound parameters capable of being used to compute the binary phase diagrams and thermochemical properties) and then using the quasi-binary approximation shown in Appendix I to develop a description of the liquid free energy on a specific quasi-binary "join" in the ternary! The results of this procedure is shown in Table 6. These results are then combined with the lattice stability values for the quasi-binary components shown in Table 5 and the remaining solution phase parameters shown in Table 6 to compute the "quasi-binary" phase diagrams shown in Figures 4-15. Stable compound phases (mainly spinels (SP) are added by defining a base-phase (S) and a compound parameter as shown in Table 7. The resulting free energy of the compound must be consistent with the observed free energy of formation of the complex compound from it's "quasi-binary" components and the quasi-binary phase diagram. The means employed to define the free energy of the quasi-binary compound is identical to that employed for binary compounds (1-5, 17) and can be seen explicitly by considering the case of the spinel AO0 .7 1 4MO 0 .2 8 6 , which can be described as (1/7) (AZ20 3 .MgO) or (Ato.2se0o.4e)'(MgO.143O0.14 ). Reference to Tables 6 and 7; Figure 7 and Equations 8 and 9 illustrate how the present definitions fix the free energy of this compound. For example, the discussion of Equations 8-10 deals with the
AOc.7I4MOo.,ss - CO0O.
71

MO0. 2

86

join in which the fraction of CO is equal to y.

When y a 0, Equation (8) provides a description of the free energy of the AO , 4 MO0.426 (SP) spinel which is equal to: o . 1
P GS

0.71

AO

+ 0.286G

+ (0.714)(0.286)[0.714LAOMO + 0.286LMOAO - C]

(11)

AO"2

S. -

'w_

.. 49-

rd

H = '

-.-

QUASIBINARY AND QUASITEENARY OXIDE SYSTEMS

Reference
-

to Tables 6 and
55

7 show that LAOMO = -33530 + 22.93T, LMOAO = 59910

27.IIT and C = 74050

C'.rr

0.' s

Joules respectively! *

These
2000

\
/

c.cor.n
Iss2st

numerical parameters provide a means of defining compound (1/7) (Ai the free energy of 2 0,-,MgO) the (SP). The free energy of

p..
P.S&

,
s

SsPr

formation of this compound from the stable forms of sp.c-s

alumina and magnesia

C
.oo.o
S.(.
0 0 .4

(e.g., (3/5) is A 2 O,(C) and C (.1/2),IgO(P) readilys,,o.8O.2.

obtained by adding the At0.40.6 .

.os /

0600.408)

quantities 0.714AOAOSC and 0.286 MOMOSP to the ex-

pression for GSp in Equation (11). The result, is then multiplied by

solid
16ooc
Solution

-5280 - 0.460T Joule/g.at.

P, s

seven and displayed in

Table 8 as the calculated
free energy change in

;
s

SP
L29..

Joules for the formation

of ?.gO'Az 2 O,(SP)
A
2

sg.s
Fes
,

from
i.e. ,
A

"7

03(C)

and

MgO(P)

0.6

.429o.S7

-36961 - 3.22T Joules. It is also compared with

Figure 19.

Calculated Isothermal Sections in the H -At0.4o0. 6 System.

the observed value (20)

of -35560 - 2.05T Joules.

In summary, Tables 5-8 and Figures 4-15 illustrate how the descriptions of the elemental edge binary systems are employed to synthesize the quasi-binary joins of interest within ternary systems which are composed of the elemental edge binaries. The stability of the computed "quasi-binary Joins" must be considered within the frame work of the approximations noted in the earlier discussions of "FeO' (WO) and Fe2 O In particular, each of the computed "joins" shown in 3 (FO). Figures 4-15 can be undercut by more stable interactior.s along intersecting joins
in the same ternary.
CO, AO, MO and 110.

Such interactions are not likely across the joins between

However joins in which FO and WO are partners may be sus-

ceptible to such intersections by competing pairs of compounds or solution phases within the elemental ternary system which lie on intersecting joins! Such interactions must be considered in computing thc most stable configuration.

The extension of the computations into quitsi-ternary systems is illustrated in Figures 16-21 for selected isothermal s ctions of the HO-CO-AO, MO-HO-AO and MO-CO-AO systems over a range of tempi catures. These computations are carried out in a straightforward manner emplo)ing the methods and definitions detailed earlier. Development of such sections as a means for guiding experimental establishment of isothermal sections in such systems is probably the most useful application of the present framew:rk. Indeed very little experimental data exist with which to compare the predicted results. Only the isothermal section of Greshkovich and Subican (30) in the MO-CO-AO system at 1973K which is shown in Figure 21 could be founafor comparison with the computed results. The current framework can also be employed to predict interesting compositions for eutectic solidification or precipitation reactions on aging after high temperature solution treatment. The periclase solid solut;oi in the MO-HO-AO and MO-CO-AL systems displayed in Figures 18-21 e:xhibits an extensive range of stability at high temperature which is sharply restricted

"~

I4

-.

-20-

Jr

-Il

777-7

77

'1

L. Kaufman and i.Nesor

at lower temperatures. In both systems the complex spinel phase limits the range of stability of the periclase structure. Under such conditions, solution treatment atL high temperatures by low temperature aging
(i.e.
,

fto
28008

~
C.Corundum 262SK

t.0 0.S
P

below 1300K) could

L.Liquld

be employed to strengthen the perislase matrix. 8. Summary

At 0 . 400.4

Paiis L-SP

59.(Cr 0 21te00.420.4

29600.426)

CrS.4006(

A data base has been presented for43

describing the therno-

P'"015013
2S008SPP-

r.0

chemical properties

and the phase diagrams for a group of six oxides. Treatment of these oxides as components of quasibinary joins in true ternary systems has

SP
At0400.6 Figure 20. At.
4 0 0 .6

1.

been employed to

deveop aquai-biary

Cr,.Q.

which can J.iroximation then be used to compute

higher order systems for purposes of phase diagram computation and prediction of .hermochemical properties.

Calculated Isothermal Sections in the Hgu 5 SO 0. 5 'Cr 04

System..406

06

"g~o.Soo. 5
P
2375K Vp Liquid (L) PP 2280K P*Priclase C.Corundsu

"103 0.s
p

t SPC

sP

SP

SP.

C C

AtA 0400.6 ,4'0

~ 4(LP

SP.(Cr0 28 600 4 28 ,0.2$60 0. 428) p(r,10, 640 A Nv

Cr 0 00

bserved) P SPS SP P2 SP SP. C Ato 4064 At~ 0400.6 SYStem. Observed Section at 1973Kafter Greshkovich and RI.

Figure 21. Calculoted andObserved Isothermal Sections inthe Hs0. . 500 .Cr 0. 40 0. 6 Stubican (LO)

4006 0

4,4,

___ ___ _ ___ ___ ___ __ ___ ___ ___ ___ ___ ___ __ ___ ___ ___ ___ ___ __

QUASIBINARY AND QITASITERNARY OXIDE SYSTEMS

V)

References 1. 2. 3. 4. L. Kaufman CALPHAD 1, 7 (1977). L. Kaufman and H. Nesor, Metallurigcal Transactions 5, 1617, 1623 (1975, 6A 2115, 2123 (1976). L. Kaufman and H. Nesor, Canadian Metallurgical Quarterly 14, 221 (1975). L. Kaufman and H. Nesor, Proceedings of the Second Conference on In-Situ Composites p 11, M. R. Jackson, J. L. Walter, F. D. Lemkey and R. W. Hertzberg Eds., Xerox Individualized Publishers, Lexington, Massachusetts (1976).

L. Kaufman and H. Nesor, Proceedings of International Symposium on Computer Simulation for Materials Applications p 63. Nuclear Metallurgy 20 (Part 1) R. J. Arsenault, J. R. Beeler and J, A. Simmons Eds., AIE, New York, N.Y., (1976). 6. M. Hillert and M. Waldenstrom, CALPHAD 1 97 (1977). 7. R. Lundberg, M. Waldenstrom and B. Uhrenius, CALPHAD 1 1S9 (1977). S. 8. C. Chatfield and M. Hillert, CALPHAD 1 201 (1977).

9. M. L. Saboungi and C. C. Hisu, CALPHAD 1 237 (1977).

10. A. D. Pelton and C. W. Bale, CALPHAD 1 253 (1977). 11. T. G. Chart, J. F. Counsell, G. P. Jones, W. Slough and P. J. Spencer, International Metallurgical Reviews 20 57 (197S). 12. T. Margaria, C. Alibert, I. Ansara and J. Driole, High Temperatures-High Pressures 8 451 (1976). 13. Proceedings of a Workshop on "Applications of Phase Diagrams in Metallurgy and Ceramics" National Bureau of Standards, Gaithersburg, Maryland, C. G. Carter and L. H. Bennett, Eds., (1977). 14. A. D. Pelton and H. Schmalried, Met. Trans. 4 1395 (1973). 15. G. Kaestle and K. Koch, "Zur Berechenung Von Phasendiagramen oxidischer Mehrstoff system mit Hilf Von mathematisch thermodynamischen Modellen" (to be published) CALPHAD 2 (1978). 16. L. Kaufman, Phase Stability in Metals and Alloys, P. S. Rudman, J. Stringer and R. Jaffee Eds, p 125 McGraw Hill, New York (1967). 17. L. Kaufman and H. Beinstein, Computer Calculation of Phase Diagrams, Academic Press, New York (1970). 18, M. Gleiser and J. F. Elliott, Thermochemistry for Steelmaking, AddisonWesley, Reading, Massachusetts. 19. R. Hultgren, P. D. Desai, D. T. Hawkins, M. Gleiser and K. K. Kelly, Selected Values of the Thermodynamic Properties of Metals (and of Binary Alloys) (2 Volumes) ASM, Metals Park, Ohio (1973). 20. 0. Ku aschewski, High Temperatures-High Pressures 4 1 (1972). 71. M. Hansen and K. Anderko, Constitution of Binary Alloys, McGraw Hill, New York (1958). 22. R. P. Elliott and F. Shunk, First Supplement (Ibid) (1965). 23. F. Shunk, Second Supplement (Ibid) (1969). 24. E. M. Levin, C. R. Robbins and H. F. McMurdie, Phase Diagrams for Ceram~ists, American Ceramics Sociey Columbus, Ohio (1964), First Supplement (Ibid) (1969), Second Supplement Ibid (1975). 25. E. N. Fomichev, V. P. Bondorenko and V. V. Kandyba, High TemperaturesHigh Pressures 5 1 (1973).

-22-

~r.;4$I~(~4.M_

L. Kaufman and H. Nesor 26. 27, 28. 29. J. P. Sharples and A. Navrotsky, 3. Solid State Chem. 12 126 (1975). T. Yokokawa and 0. J. Kieppa, J. Phys. Chem. 68 3246 (1964). A. Ferrier, Comptes Rend. Academy Sci., Paris C246 919 (1967). W. B. Pearson, Lattice Spacing and Structure of M.'etals 2 3 (1967), Pergamon Press, Oxford. 30. C. Greshkovich and V. S. Stubican J. American Ceramics Soc., S1 43 (1968). APPENDIX I

Quasibinary Approximation to the Free Energy of Solution Phases (Mass Basis, One Gram Atom or One Mole of Atoms) (See Figure 1) L L RTlx-tnl-yxnxyy] G.(l-x-y)OGL+xOGy* Ex y,T](A1

(xLJK+yLKJ]+xy(l-x-y)TRNL (A-2) where LIJ, LJI, LIK, LKI, LJK, LKJ and TRNL are all temperature dependent functions. If one considers the vertical plane which cuts GL and contains the compositions x-xi, y-o and xwx*, y-l-r,, with a quasi binary definition of the excess free energy as

+xy(x+9')

where E-(x-xl*) (x,-x*) 1, y-E(I-x*) and (1-x-Y)=(l-x1) (l-o) The explicit relationship between L, and)L, and the terms LIJ, LJI, LIK, LKI, LJK, LKJ and TRNL shown in Equatioo (A-2)c n be obtainedby equating the ternary and quasi binary descriptions shown in Figure 1 at &*0, 1/3, 2/3 and 1. The result is given by the following equations; (A-4) ,, Q(l-xj), S.Cl-x.) L 1 2M-N, L 2 2N-M, P-(x*-xb 1x+0333P)Q[QO.667LIJ+(x +P0.333LJ1)] M1 'u0.667 (l-O.333S)lx (x*+O.667P)Q[QO.333LIJ+(xi+PO.667LJI)] N~033lO67) M2 .. 222(Q-0.333Pf-1 QS[QO.667LIK.S0.333LKI] N2 -0.222(Q-0.667)- IQS[QO.333L1K+SO.667LK1] (A-5) (A-6) (A-7) (A-V8) (A-9) A-0 (A-11)

M3 0.333C4+0.333Q)S(x+.333P)[C4+.333P)LJK+SO.333LKJ]
N3 n.667Cx1*+O.667Q)_ S(x*+O.66 P) [(xi+O.667P)LJK+SO.667LKJ] M4 O.222TRNL QS(xj+O.333P) 1
7

M -RT[O.667Q9.nO.667Q+(x +0.333P)in(x +O.333P)+0.33StnO.333S1

N uRTC0.333QinO.333Q+(x*+O.667P)in(xl*+O.667P)+0.667SZnO.667S] 1 1 M 667Qx* (QLIJ+x*UJI)

(A-13) (-4 C-4

'1

-23-

__

____

____

_____

____

____

QUASIBINARY AND QUASITERNARY OXIDE SYSTEMS-I

N =0.333QxlCQLIJ.x*LJI) M =O. 333Sx*CSLKJ+x*LJK)CA17 NsO . 667Sx* CSLKJ+x*LJK)

m 8 -0.66 7 RTCx*tflx*+QkflQ)+0.333RTcx~tlx*+StnS-1.910)

C-6

C-8 (A-19) A )
(A-21)

+m +m +* - .910) +0.667 RT(xAknx*+SknS) M8 . 333R41 x+m CM 4 2 M 3*. M+ - 6 M7 ma) N-4.SCN +N +N +N +N -N6 N7 -N8 )

(A-22)

-24-

Section 111.

Calculation of QuasiBinary and QuasiTernary Systems (1979) 3 pp 27-44

-ICALPHAD

-257
5(I -

W"

CALPHAD Vol.3, No.1, pp.27-44. Pergamn Press Ltd. 1979. Printed in Great Britain.

CALCULATION OF QUASIBINARY AND QUASITERNARY OXIDE SYSTEMS - II* Larry Kaufman ManLabs, Inc., 21 Erie Street Cambridge, Massachusetts 02139, USA ABSTRACT. A data base is being developed for calculation of quasi-binary and quasi-ternary phase diagrams of ceramic systems. The initial segment of this base covers combinations of Cr2 0 3 , MgO, A1 20 3 , Fe203 Fe 304 and "FeO". Lattice Stability, Solution and Compound Phase Parameters were derived covering the liquid, spinel, periclase and corundum phases in systems composed of these oxides. These parameters permit calculation of the thermochemical properties and phase diagrams of combinations of these oxides in good agreement with limited experimental data. This base has been extended to include SiO 2 and CaO by carrying out an analysis of the thirteen binary systems composed of the above mentioned systems with silica and calcium oxide. The current base of eight oxides can thus be employed to compute fifty six ternary systems. Isothermal sections in the MgO-Fe 304-Cr 201 are presented between 2800K and 1600K illustrating the usefulness of the data base. 1. Introduction

The utility of computer based methods for coupling phase diagrams and thermochemical data for metallic systems has been demonstrated in many papers published in this journal. A considerable effort is being applied toward development of a suitable data base for metallic systems. Numerous examples have been published showing how such data on metallic systems can be synthesized to successfully predict the thermochemical properties and phase relations in ternary metallic systems. Such predictions can be applied practically to explore conditions for in-situ eutectic growth, metallic glass formation, spinodal decomposition, metastable equilibrium, and diffusionless transformations. In addition, recent development of SIALON composites, combining silicon and aluminum nitrides with oxides of silicon, aluminum, magnesium and other metals, has provided additional motivation for developing methods for predicting multicomponent phase diagrams of ceramic systems. Accordingly present plans include further expansion of the current data base to cover Si3 N. and AN. In addition to the present work (1) other efforts described in CALPHAD are currently in progress in order to develop predictive methods which can be used to compute multicomponent phase diagrams for ceramic systems (2,3).

*This work has been sponsored by the Air Force Office of Scientific Research, Bolling AFB, Washington, D.C., under Contract F44620-76-C-0060.

-26MA 77 77

L. Kaufman

2. Description Of The Thermochemical System Employed To Describe Solution And Compound Phases The method utilized for describing solution and compound phases-is essentially the same as that employed earlier (1) except that some symbolic usage has evolved which facilitates data handling as indicated below. The free energy, GL, of a liquid (solution) phase, L, in the binary system I-J is given by',Equation (1) where T is in Kelvins, x is the atomic fraction of J and the mass basis is a mole of atoms (i.e., a gram-atom). GL - (1-x)*G + x*G + RT(xnx + (l-x)Xn(l-x))+ x(l-x)[LIJ(l-x) + xLJI] (1)

In this expression, *GL and *G are the free energies of a gram atom of pure liquid I and J, R = 8.314 J/g. ai .*K, and LIJ and LJI are functions fr cof temperature (but not of composition). The composition, x, is the atom 0 tion of J. Si ilar equations can be written for each solution phase with Gt replaced by *Gj, eGC, etc., for the P=Periclase, C=Corundum phases, etc., and with LIJ replaced by PIJ and CIJ etc. In the present symbolism, one uses the symbol DODOLP to designate the free energy of pure liquid CaO minus the free energy of pure Periclase form of CaO. Reference-to Tables 1 and 2 show that DODOLP=39748-13.723T Joules, and that the liquid phase of the CaO-MgO system (i.e. DO-MO with i-DO, j=MO and x=atom fraction of MO) is described by LDOMO-20920 and LMODO=-20920. The thermochemical data for stable phases listed The free energy, GY, in Table 1 was taken from the ManLabs-NPL Data Bank (4). (2) as, of a compound phase i(lx*)Jx* is defined by Equation G- . (l-x,) 'G + x, -G + x,(l-x*)[(l-x,)LIJ + x, LJI-C[T]] (2)

where e refers to the base phase selected for the compound and C[T] (which is a temperature dependent function) is the compound parameter . Table 3 lists the compound parameters and base phases assigned to compounds in the systems of interest. Thus the free energy of one gram atom of the compound CaO*Cr 2 O3 . which is defined as (I/7)(CaO.Cr2 O3 ) or DOw. 7 14 is defined as 8 6 CO 0 G 0.286 OG D + 0.714 OGSo + (0.286)(0.714)[(0.2R6)LDOCO+a.714LCODO-C] (3)

Reference to Table 2"and 3 shows that LDOCO=LCODO=O and the compound parameter C in this case is equal to 101671. Thus the free energy of formation of DOo.2 9sCOo.7 1 4 from the spinel forms of DO and CO is equal to -20741 Joules. In order to compute the free energy of formation of DOO.2 8 6 CO0. 7 14 from the stable P(Periclase) form of DO (i.e. CaO) and the stable C(Corundum)form of CO (i.e. Cr2 O3 ) it is necessary to include the lattice stability terms DODOSP and COCOSC. The former is contained in Table 1 (DODOSP=DODOLP-DODOLS-28137-4.93T) while the latter was presented earlier (1) (i.e. COCOSC=117S6-0.92T). With this additional information the free energy of formation of D00. 2 0 6 CO 0. 7 14 from the P form of DO and the C form of CO can be computed as is shown in Table 4. Here the free energy of formation of (I/7)(CaO.Cr2 O3 ) from (i/7)CaO (i.e. 0.286DO) and (l/7)(Cr 2 O3 ) '(i.e. 0.714 CO) with DO in the Periclase form and CO in the Corundum form. Reference to Tables 4 and 5 discloses that there are differences between the calculated %alues and the assessed "Data Bank Values" which are largely based on direct thermochemical measurements. However, it should be pointed out that there are several instances of "internal inconsistencies" in the "Data Bank Values". Thus for example the "assessed values" in Table 4 suggest that CaO.Fe 2 O3 decomposes into Fe2 O3 and 2CaO.Fe2 O3 and that 3CaO.A 203 decomposes into CaO and CaO.A9 2O 3 at 300K. Moreover the "assessed value:;" in Table 5 suggest decomposition of mullite (3AZ 2O 3 .2SiO 2 ) into SiO 2 and AZ20 3 at low temperatures. Since none of the above reactions are actually observed it must be assumed that these

-27.;
77* 7

.............
_

QUASIBINARY AND QUASITERNARY OXIDE SYSTEMS

11I

TABLE 1 SUMMARY OF LATTICE STABILITY PARAMETERS (All units in Joules per gram atom (mole of atoms), T in Kelvins) P - Periclase, T -Tridynite, V AO - (1/S)Ar2O0 3 F0 - (1/S)Fe 03 C =Corundum, S = Spinel, X =Crystobalite,

quartz), R =trigonal (B quartz) H = hexagomal (ai

CO

(l/S)r 2O3 ,

MO

=(l/2)MgO,

WO

=(l/2)FeO

HO = (1/7)Fe 04

SOSOLX*= (1/3)SiO2 (Liquid) - (1/3)Si02 (Crystobalite) SOSOLT -(1/3)SiO (Liquid) - (/)i 2 (Tridymite) SOSOTX = (1/3)Si02 (Tridymite) - (1/3)Si0 2 (Crystobalite) SOSOLX = SOSOLT +SOSOTX SOSOLX = 3347 SOSOLT =3933
SOSOTH - 858 SOSOHR
=
-

1.674T

2.008T
0.795T 0.293T
=SOSOLS =-2.092T

251

SOSOLC

= SOSOLP

DODOLP = 39748 DODOLC = 14644 DODOLS -11611 DODOLX =

DODOLT
=

-13.723T

10l.878T
-8.786T

1 .674T
-2.008T

AOAOLX - COCOLX = MOMOLX = WOWOLX = HOHOLX = FOFOLX = -1.674T AOAOLT - COCOLT = MOMOLT = WOWOLT = HOHOLT =FOFOLT x-2.008T
*These

differences specify the free energy of one phase (i.e. liquid) minus the free energy of the second phase (i.e. Crystobalite) of a given compound.

-28* -

t 4V~'

L. Kaufman

TABLE 2 Quasibinary Solution Parameters for Oxide Systems (All units in Joules per gram atom (mole of atoms), T in Kelvins) DO-AO (l/2CaO-l/SA
2 03 )

SOz.P

(1/3Si0 2 -1/2CaO)
4

SO-FO (l/3Si0 2 -1l/SFe2O0 3) LSOFO = 54392 CSOFO = 138072 XSOFO = 138072 TSOFO = 138072 LFOSO = 163176-83.68T CFOSO = 246856-83.68T XFOSO = 246856-83.68T TPOSO = 246856-83.68T SO-HO (1/3Si0 2 -1/7Fe 3 O4 OjH~O 7 LSOHO =80542-10.46T XSOHO = 164222-10.46T TSOHO = 164222-10.46T LHOSO = 213384-125.52T XHOSO = 297064-125.52T THOSO = 297064-125.52T *0.7< XHO1.0
SH

LDOAO = -22595 CDOAO = 102926 PDOAO = 102926 LAODO = -22595 CAODO = 102926 PAODO = 102926 DO-CO (l/2CaO-l/5Cr2O0 3) =DCO = 80 PDOCO = 83680 LCODO = 0 CCODO = 83680 PCODO = 83680 DO-MO (1/2CaO-1/2NIgO) =~MO z 23012 PDONIO = 41840 LMODO = 29288 PMOPO = 68199 DFO(1/2CaO-1/2Fe2 0O 3 23 TUMUFO = -20920 CDOFO = 104600 PDOFO = 104600 LFODO = -20920 CFODO = 104600 PFODO = 104600 DO-HO (1/2CaO-1/7FeO0 ) LDOHO = 0 0 SHOO 1752LSOMO PHODO 117152 DO-WO C1/2CaO-1/2FeO)

O<XDO <O.. 0 LSODO -0 XSODO = 83680 TSODO = 83680 LDOSO = -393631+104.6T XDOSO .-_3099S1+104.6T TDOSO = -3099S1+104.6T 0.4 0 xDlO LSODO -1570 -18 PSODO = -195476+41.8T LDOSO = -100416+41.8T PDOSO = -16736+41.8T SO-AO C1/3Si0-/A1O T-=O = 3849 31203) LSOAO = 122173 XSOAO = 122173 TSOAO = 122173 LAOSO = 284S1 XAOSO = 112131
XOO=123

SDOHO POOHO1 LHODO

= 117152 = 117152

TAOSO =112131 SO-CO (1/3SiC -1/5Cr2 O0 2 ) L90-O = 68r9-9 CSOCO = 151879 XSOCO = 151879 LCOSO - 68199' CCOSO = 151879 XCOSO - 151879 SO-MO (1/3SiO 2 -1l/2MgO) M .0 = -24267+24 271'XSOMO - 59413424.27T TSOMO =59413+24.27T

LDOWO = -7113 PDOWO . 30962 LWODO = -20501 PWODO = 17573040x

SSOHO = 83680 10O LHOSO = 247902-10O SH = 331582-130.OT SWO(1/3Si0 2 -1/2FeO) 5 0 qsxwo~0 . LSOWO -29288 XSOWO = 112968 TSOWO = 112968 LWOSO = 324678-225.94T XWOSO = 408358-225.94T TWOSO = 408358-225.94T O.O<xwo 'L0 LSOWO =162339-112.97T

LMOSO = -229283+68.62T XMOSO = -145603+68.62T TMOSO - -145603+68.62T 1 <. LSOMO w-106274+42.01T PSOMO = 19246+42.01T LMOSO -- 106274+42.01T

PSOWO = 246019-112.97T LWOSO =191627-112.97T PWOSO =275307-112.97T

PMOSO a 19246+42.01Tf

-211:
7,'

(qASITERNARY OXIDE SYSTEM4S QUASIBINARY AMD

-II

TABLE 3 SUMARY OF CMPOUND PARANETERS (AlAl units in Joules per gr.am atom (mole of atoms),

Tin Kelvins)

Compound l/9(2CaO-Fe 0 3 l/7(CaO-Fe 2O0 3) l/32(Ca0-6A1 2O0 3)

1/12(CaO*2Al 2O0 3)

Stoi~hiometrw DO
44

Base P S C
C

Compound Parameter (Joules/g .atom) 9623 - 3.43T 46861 - 15.48T 16736 + 13.18T
15481 + 10.46T

FO0

~.
37

DO 0 . 2 8 6 FO 0 .71 4

DO 0 063 AO %

DO 0 . 1 6 7 A G9. 3

1/7(CaO*Al 2O0 3)
1/1(3CaO-A 2 O03 )

DO 0 . 2 8 6 AO 0 A4
[O100546 AO 0 -454

S
P

38493
16318 + 13.51T

1/7CCaO*Cr 2O0 3) lI5(CaO~Sio) 1/7(2Ca&-SiO 2 ) 1/21C2SiO '3AI 2 0 3 l/S(MgO*SiO) 1/7(2NMgO.5i0 2 ) 1/7(Si0 2 *2Fe0)

DO 0. 2 8 6 DO0 0 DO0 . 5 71 0028 MO0 4 MO0 . 5 71 s0042

CO 0 -.. 4 s so 04.29 AO 0 4 so so0 Z W0 0

S P P C P P P

101671 -62383 + 49.29T -56902 + 43.93T 41840 +4.73T -66735 +47.28T -64434 + 41.13T 217986 - 120.5T

MO - (l/2)MgO, SO -(l/3)Si0 S - Spinel

2

DO WO

(l/2)CaO,
=(l/2)FeO,

FQ

(l/5)Fe 2 0 3 9

AO

=(l/5)A12

03

C - Corundum,

P -Periclase

inconsistencies represent cumulative errors which can occur in assessment of thermochemical data when the compe~ition between adjoining phases (i.e. the phase diagram) is not considered. 1En the present process such competition is considered as a natural part of the -evaluation method. 3. Discussion of Computed Phase Diagrams Binary phase diagrams bertween CaO and Si0 2 with the oxides At.:'C 3 FeO, Fe2O3 and Fe3 01. awe provided in the compilation series Although some of the phase diagrams prepared by Levin, Robbins and McMtoidie(5). are sketzchy and conflicting experimnental data are presented tbL results have been used in concert with the experimental thermochbmical dzata(4) and prior lattice stability values(l) to select the values for thp solution and compound parameters which are shown in Tables 2 and 3. Figur':6 1-13 are the binary phase diagrams which can be generated from the results displayed in Tables 1-3. The calculated AL 2 0 3 -CaO binary diagram in Figure 1 is in relatively good experimental diagrams shown in Figures 231,232,2295,2296 and agreement with the S. Moreover, Tabile 4 shows that the computed free energie s 4308 of reference of formation of compounds in this lzinary system are in relatively good agreement with the "Data Bank Values" derived from thermochemical data. The solution phases in this system are apprarximated by regular solution descriptionst (i.e. LDOAO=LAODO etc.). Figure 2 siows the computed Cr 2 O 3 -CaO phase diagram
Cr2 01 N1g01

-30-

Ti>

-.-

---

.yj7.7'.

-.

s0

!'7.

* popSt.> ,

L. Kaufman

TABLE 4 SUMMARY OF CALCULATED AND ASSESSED (4) FREE ENERGIES OF COMPOUND FORMATION (JOULES, 'K) DATA BANK AG VALUES (4) (2/9)CaO + (l/9)Fe2 0 l/9(2CaO.Fe 2 0 3 ) 3 AG - -2714-0.313T (calculated) (1/7)C40 + (I/7)Fe 2 0 3 - 1/7(CaO.Fe 2 0 3 ) AG - -3808 + 0.977T (calculated) (l/32)CaO + (6/32)Al2 03 (1/32)(CaO.6A12 0 3 ) -5117 at 300K -7627 at 1400K -3112 at 3.00K -4284 at 1400K .. .. .. ..

AG - -739-0.957T (calculated) (1/12)CaO + (2/12)A1203 (1/12)(CaO2AI203 )

-1916 at 300K -5381 at 1600K -3096 at 300K -6720 at 1600K '1632 at 300K -S786 at 1600K ...... ..

AG = -1101-1.928T (calculated) (l/7)CaO + (I/7)A12 03 . (I/7)(CaO.A1 203 ) AG - -1248-2.219T (calculated) (3/11)CaO + (1/II)A1 2 0 3(I/II)(3CaO.AI20 3 )

AG = -947-2.15ST (calculated)

(l/7)CaO + (l/7)Cr 2 0 3 - (1/7)(CaOCr203 )

AG = -4300-2.067T (calculated)

which compares well with Figure 37 of reference 5. Table 2 discloses that this syftem is computed with regular solution paraneters and an ideal liquid phase (i.e. LDOCO=LCODO=O). Figure 3 displays the computed CaO-MgO system which is obtained with assymetric excess free energies (LDOMO LMODO) but ideal mixing entrooies (dL/dT=0). The latter agrees with Figure 229 of reference S. Figures 4-6 contain the calculated CaO-Fe 2O 3 , CaO-Fe 3 O0 and CaO-FeO phase diagrams which are in good agreement with those shown in Figures 43-48 of reference(S). Reference to Table 2 shows that the CaO-Fe2 Os and CaO-Fe 3O cases are approximated by regular models (i.e. the latter as an ideal liquid) while the CaO-FeO case is assymetric (LDOWO9LWODO) with ideal entropies of mixing for solution phases. It should be noted that the computed CaO-Fe 2O 3 diagram shown in Figure 4 is metastable since Fe2O3 decomposes into Fe3 0. and 02 above 1760K under atmospheric conditions. In contrast to the results encountered in analysis of the CaO-based systems, reference to Table 2 shows that analysis of the Si0 2 -based binary systems proved to be more difficult. In this group, only the 3iOj,-Cr 2O 3 binary calculated in Figure 9 could be approximated by a regular solution model. The resulting phase diagram compares well with the experimental diagram shown in Figure 332 of reference S. Figure P shows the computed SiO 2 -A 2 0 diagram which is in good agreement with the exi.3rimental results which are given in Figures 313,4374 and 4375 of-Reference S. This system is described by assymetric solutions (LSOAOtLAOSO) and ideal solution entropies (dL/dT etc.=0). Moreover, examination of Table 5 shows that the computed free energy of mullite (3A1 2O3.2SiO2) from the stable forms of A12 03 and SiO 2 is in relatively good agreement with the values listed in the Data Bank(4). The. latter are determined by assessment of thermochemical data. The computed Si0 2 -Fe2O 3 diagram shown in Figure 11 is metastable above 1760K due to decomposition of FeZ03 into Fe$O and O at one atmosphere. This diagram is in acc'rdance with

Figure 80 of reference S and is computed with assymetric so. tion models

A:

-31-

QUASIBINARY AND QUASITERNARY OXIDE SYSTEMS

-II

TABLE 5 SUMMARY OF CALCULATED AND ASSESSED (4) FREE ENERGIES OF COMPOUND FORMATION (JOULES,0 K) DATA BANK AG VALUES (l/S)CaO + (l/5)SiO 2- (l/5)(CaO*5Si0 2 ) AG - -19791-2.402T (calculated at 300K) AG - -20808-l.S47T (calculated at 1800K) (2/7)CaO + (l/7)SiO2-* (l/7)(2CaO. Si0 2 ) AG -- 1848S-O.954T (calculated at 300K) AG - -19212-0.343T (calculated at 1800K) (2/21)Si0 2 + (3/21)A 203 - (l/21)(2SiO 2 3A1 2 03 ) AG a -706-l.2S4T (calculated at 300K) AG - -1191-0.846T (calculated at 1900K) (1/S)MgO + (1/5)SiO 2 - (l/S)(Mg0-Si0 2 ) AG -- 6464-1.864T (calculated at 300K) AG - -7481-1.009T (calculated at 1800K)
(2/7)MgO + (1/7)Si02 - (1/7) (2MgO* Si 2 )'

-17970 at 300K -18146 at 1800K -19430 at 300K -19539 at 1800K +397 at 300K -2013 at 1900K -7273 at 300K -5799 at 1800K -8301 at 300K -7761 at 1800K -6845 at 300K -4067 at 1400K

AG - -8088-0.216T (calculated at 300K) AG a-8815 + 0.395T (calculated at 1800K) (l/7)Si0 2 + (2/7)FeO. (1/7)(Sio2 *2FeO) AG a -7373 + 1.413T (calculated at 300K) AG - -7849 +. l.880T (calculated at 1400K)

(i.e. LSOFO#LFOSO) with finite excess entropies of solution(i.e. dL/dT#O) as shown in Table 2. Analysis of the remaining binary systems i.e. Si0 2 -CaO, Si0 2 -M9O, SiO 2 -Fe 3 O% and Si0 2 -FeO could only be carried out by segmenting the description of the liquid phase as is illustrated in Table.2. Thus, in the last of the above named cases, the excess free energy of the liquid phase (i.e. the last term in Equation 1) is set equal to x(l-x)[(1-x)29288.x(324678-225.94T)] where ii iet equal to x(l-x)[(1-x)(162339-112.97T)+x(191627)-1T2797T)I. Note that both expressions yield the identical result, 44246-28.24T at xWO=xS0=0.S. The calculated Si0 2 -FeO phase diagram shown in Figure '13 is in relatively good agreement wi-h the observed diagram in Figure 80 of reference 5 except that liquid monotectic is reported(S) at 1970K and 45 w/o FeO. Examination of Table S shows that the computed fre~e energy of formation of Fayalite (2Fe05Si0 2 ) from the stable forms of SiOz and FeO is in relatively good agreement with the "asesedvaues(4 oerthe temperature' range of its stability. InL the Si02-Fe3O4 case the thermochemical description of the liquid phase shown in Table 2 is divided into two segments corresponding to the composition range O<xHO<0. 7 O (i.e. H0=(I/7)1*e Tie c'5mputed 3 O4) and a second covering the range 0.70<xjj0 <l.0. binary phase diagram shown in Figure 12 compares favorably witfh that shown in Figure 87 of reference 5 except that the observed liquid monotictic is reported

Ox<0.5 and x=xwoatom fraction of WO.

In the range 0.5<x<1.0 the last term

(5)to be at 1930K and 90 w/o Fe 3O%.

L. Kaufman

w/o CaO
At 2 03 13 TOK 26 37 48 58 67 76 85 93 CaO h(liquid) L C (Corundum)

P(Periclase)
1
9 21765

1600 4.A0

I
0 7 1 4 DO0. 2 8

A0. 4 S4 DO0. 5 4 6

AO0. 9 37 DO0. 0 6 , 800 A0. 4 S4 DO0. 5 4 (

4 4. 0.S4 6 D AO 0.063 0.167 0.286 DO A0.400.6 A0.45 4D0.5 4 6 =(1/l) (3CaO.A 203) Ca0.500. 5
Figure 1. Calculated Al0.400 .6 -Ca0 . 500.5 Phase Diagram w/o CaO Cr2 03 TOK 9 18 27 37 47 57 67 78 89 CaO

L(liquid)

2400

-P(Periclasi)
223 0

C (Corundum) 1600 + "CO0. 7 1 4 DO0. 2 8

CO CO

D 7 1 4 D0.286
+

P 800

CO Cr 0 40.6

0.286-(l/7) (Cr2 03 -CaO)=CO0. 7 1 4 DO0 286 DO Ca 0 sO0

.!.

-Figure 2. Calculated Cr0 . 40 0.6 -Ca0.s0 .5 Phase Diagram

-33-

'77

.:R"

QIJASIBINARY AND QUASITERNARY OXIDE SYSTEMS

-II

W/o mgO CaO

7 15 23 32 42 52 63 74 87 MgO

(Periclaseclse) MicbliyGp

tCa

00
5 ,Mg 0 .SO

Figure 3. Calculated Ca0 . 00

. Phase Diagram

CaO

11

22

33

43

S3

63

73

82

91

Fe 2 0 3

+ P(ericaseL+C (Corundum) 1600 s

+
S00
DO 0 . 445FO0

METAS'ABLE!

10['o
Fe0 .4 0 0 . 6

Ca .S O.S(1/9)(ZCaOFe 2 O3 )uDO 0 .44 FO 0 55 5

Figure 4. Calculated Ca GS 0 .5 F 0.400.6 Phase Diagram (Note: Fe 0 .4 00 . and 1/20 2 6 Decomposes into Fe 0 .4290 .71 above 1760K)

IN
________________________4L____:_ -. N

L. Kaufman

w/o Fe3O4 CaO 12 23 34 44

34

-a73 83 91 Fe 3 04

63

T*

DOt tCa

0 0a

Fe

9 0

HO 4

Figure 5. Calculated Ca 0 .50 .- Fe 0 .4290 0.571 Phase Diagram w/o FeO 46 56 36

-

CaO T*K

13

24

66 L CLiquid)

75

83

92

FeO

240C L
+

P(Periclase)

1::

Periclase Miscibility Gap)

DO

WO

Ca 0 . 00. Figure 6. Calculated Ca 0 .s0o

5 -Fe0

Fe0 .5 00 . 5 . 00.S Phase Diagram

p'

-;4

-3-

41

~7

nr

Q9JASIBINARY AN~D QUASITERNARY OXIDE SYSTEMS

-II

w/o CaO SiO 2 13 26 38 48 S8 68

77

85

93

CaO

T*K

L(Liquid) X-Crystobalite
24If 0
2406@

P(Periclase)

1600.TCTridymite) +SO .6 D 0.4 108

110

so 0 . 4 2 9 DO 0 . 57 1

so 0 .6 DO0

so(- '--.6-0.4 50.4 2 9 D 0.571

S1 Si 0

(15t(~iOSi2=SO0. 6 DO0.4

s0. 4 2 9 D0.571 (l7)(ZCaO. Si 2 03 )

Ca

00 5

Figure 7. Calculated Si0 .3330 0.667- Ca 0 .50 0.5 Phase Diagram

w/o A
S
2

3---8
51 61 70 80 90 A
20 3

10

20

30

TK L+XCCrystlobalite)

TOI
41 L(Liquid) 5 - &log Mullit

T(Tridymite).Miullite

C (Corundum)J

1200arz+Mlt
R(aQuartz)+Mullite 600

lai

so MULLITE-(l/21)(2SiO2 2AL

O3 >50 0 2

86

AOA A 0 4 0. 6

i 0 ,.333 0.667 Figure 8. Calculated Si 0 30

6

- A?

400

Phase Diagram

-36-

777

L. Kaufman

w/o SiO 2 Cr2 03 7 14 22 31 40 50 61 73 86 Si02

TOK

L(Liquid) 300 L+L 2 (Liquid Miscibility Gap)

2000

C(Corundum) + L
+

/0o0
1750

Crystobalite,

C + T(Tridymite)

100 C

H CcQuartz)

C + R ($Quartz)
p
p

I..

Co

Cr.4 00.6
6 -Si0

So Si0 .33300.667 33300.667 Phase Diagram

Figure 9. Calculated Cr0 400

w/o MgO
Si0 2 TK 10 20 30 40 50 60 70 80 90 MgO L(Liquid)

240Gap
XeS0

LI+L 2 X-Crystobalite 240 iquid Miscibi *yP(Periclase) i4

L
.42160

1$45
I

0 6 M00 . 4

1600 "T(Tridyimite) + SO ioso 0 .6M00 . 4 800 R(oQuartz) S0.6M0.4 +

SO0. 6 MO 0 4 SO0. 0 4 29

So 0

4 2 9M0 0 5 7 1

00.571

so (1/5)(SiO .MgO)-SOo

6 M0o 4

#
Mgo
6 6 7 -Mg 0

MO
2

I
Sio.33300.667

SO0 .42 9 MOO. 5 7 1 "1/7)(SiO

5

2MgO)

O0 . 5

Figure 10. Calculated Si0 33300

.5005 Phase Diagram

-37,?''V'? .-

QUASIBINARY AND QUASITERHARY OXIDE SYSTEMS

-Il

w/o Fe2O3 S O 15 29
LK 1

-71 79 86 94 Fe 2O0 3

41

52

62

L(Liquid)
12

X(Crystobalite)+L
1800 T.L

10
172 5

META_ STABLE!

T(Tridymite)+P 1200 XHczQuartz) 't P 600 R(BQuartz)

1og0

,es 0

~+P(Periclase)

Si0

Fe

Figure 11. (Note: Fe S2

Calculated Si 0 .33300. 667 -Fe0 .40 0.6 Phase Diagram Decomposes into Fe 7 and 1/2 02 above 1760K) 0 O0 0 4200 w/o Fe 0 79 87 94 e3O4 16 29 41 52 62 71
_

TK

L+L 2
+

T+L L
+
1 7 50

L(Liquid)

1800 -XCCrystobalite)

S
0

1200

T(Tridymite)

S(Spinel)

.H~cQuartz) + S

600

RC8Quartz)

S(Spinel)

so S7Fe i 0 . 3 3 0 .6 6 Figure 12. Calculated S Diara S 0

667 0. 3 33 O0 . 0

HO 0 Fe F 0. 5 71 4 2 9O 0 . 4290 0 571 Phase

-38-

itW

L.

Kaufmnn

w/o Sio 17 31 44 55

FcO 64 73 81 88 94

FeO

T0 K 1800

L1+ L L(Liquid) 2 Liquid Miscibility G P X=Crystobalite X+ T+L

'q Gp

L
1 50 0

1200

T(ridymite)+ SO .429WO0.571

10800

soO. 5 29w 0.
6001 R(BQuartz)

s 0. 571 429w0.*
+prcls

so
Si 0 .333 0 0667

S0.429W0 O. 7iClI17)CSiO Z*2FeO) s

6 6 7 -PC 0

WO

FeOS 00.5 . 0O. Phase Diagram

0 Fi,-ure 13. Calculated Sij 0 .330

In the remaining SiO 2 -CaO and SiO 2 -MgO cases, Table 2 shows tha~t 11c 6 0. liquid thermochemistry is described in two composition ranges, from '.0>~.'(; and from 0.6>xSQ>0. The resulting computed binary diagrams shown in Fi-gures 7 and 10 ccpare favorably with those shown in Figures 237, 2331, 2303 and 266, 2331 of reference S. In these cases the greatest discrepancies are encountered again in the position of the liquid monotectic where the experimental locations

are at 1970K and 27 w/o CaO in the SiO 2 -CaO case and at 1960K and 31 w/o MgO in
the SiO2-MgO case. Comparison of the computed and "assessed" free energies of

compound formation for- SiO 2 -CaO and SiO 2-MgO compounds listed in Table 5 shows while the computed low temperature result% for the (l/5)Q(fgO'SiO 2 )
that in both cases excellent agreement is observed for the (l/7)(2MgO.SiO2) However, (i.e. M00. 0. 571S0 0. 429) cases. 5 71SO 0.4 2 9) and (l/7)(2CaO*5i0 2)(i.e. D0 (i.e. MOO.4SO 0.. ) and (l/5)(CaO-SiO 2 ) (i~.D0 0. 4SO. 6 ) agree with the "assessed" values, the computed high temperature values are in poor agreement with the "assessed" high temperature values. Thus, it appears that while the present model canl be readily applied to obtain a good description of tile CaO based systems difficulity is encountered in its application to some of the SiO 2 based sys tems.

4. Calculation of Quasi-Ternary Phase DiagramsI

The objective of developing the present data base for characterizing quasi-binary ceramic systems is to provede a means for computing stable and meta stable equilibria in quasi-ternary and higher order ceramic systems. Several examples of such systems were presented earlier(l). In these cases the quasi-ternary systems were computed by synthesis of the edige l'rlary characteristic for solutions on the basis of the Koehler equation. Th-S Lor the (l/2)NMgO-(l/7)Fe 3 O4-(l!5)Cr2O3 (i.e. MIO-HO-CO) case the free ecrr_- of the liquid is defined as

-39-7
-,.-'--,.;',~,

14
~
-

-*

~~~~

-~4,k

N ,-

QUASIBINARY AND QUASITERNARY OXIDE SYSTEMS - 11

GL.I-x-y) GLO + xGL

+

+ YG o + RT((l-x-y)1n(l-x-y)+xknx+yny)

x(l-x-y)(I-y) 1 [(l-x-y)LMOHOxLHOMO] +y(1-x-y)(I-x) 1 [(I-x-y)LMOCO+yLCOMO] +xy(x+ey) 1[xLHOCO + yLCOHO] +TRNL(l-x-y)xy (4)

where x is the atom fraction of HO, y is the atom fraction CO and TRNL is a ternary interaction parameter. Similar equations can be written for the Periclase(P), Corundum(C), Spinel(S) and other solution phases by replacing the L's and TRNL by P's and TRNP etc. If TRNL=O then all 6f the terms in Equation(4) have been explicitly defined(l). Thus in this approximation the free energy of ternary solution phases are synthesized from their binary edge components by using Kohler's Equation with a zero ternary interaction parameter. The free energy of ternary compound phases is defined along specific joins As an example, we consider the spinel compound phase in the system NZ HO-CO (see Figure 14). This compound runs between the Fe30.MgO and Cr2O 3 "MgO, compositions designated (Table 7(1)) as (Feo.,osO0.4 0o)'(MgO.14 3Oo0. 4 3 ) and Cr. 2 s 6 O0 4 2 0)'(Mg0 1 4 Oo. 3 1 43 ): If x is the fraction MO and y is the fraction HO then the free energy of the spinel (SP) compound CO. I.MO I-HO MO given by Equation(s) as
SP G * (l-x-y)G o + xG o + yG O + RT(ytny+(l-x-y)Ln(l--y)-(l-x)Ln(l-x)]

(]AGA + ( " )AGB where

AGE-

+ AGE

O,AGA - (1-x*)X[(1-xb) LCdMO+x!LMOCO AGB - (1-x,)x,[(l-x,) LHOMO + x, LMOHO

-

CA] C(6)

and
x-x! + yp ;p

(x

x)/(l-x,),xl - 0.286, xs

0.286, p - 0

CB

0 and CA

38910+16.74T

(From Table 7 of reference 1)

(7)

Since all of the parameters required to calculate the free energy of the liquid, spinel, periclase and corundum solution phases in the MO-HO-CO system as well as .the free energy of the spinel compound phase, the entire ternary phase diagram can be computed. The results are shown as a series of isothermal sections between 2800K and 1600K in Figures 14 and 15. These results can be employed to locate melting points and liquid decomposition products thus providing a guide for structural studies. Moreover these sections can be employed to locate interesting solid state decomposition reactions which can be employed to produce strengthening reactions attainable via heat treatment. One such example in the present computations is the decomposition of the periclase field indicated on cooling from 1850K to 1600K in Figure 15. Thus an "alloy" composed of 0.7MO-0.25HO-0.USCO (i.e. 88.4 mol percent MgO-9.1 mol percent Fe3 O,-2.5 mol percent Cr2 O3) would be expected to be strengthened by solution treatment at 1850K and subsequent aging at 1600K.

-4o0.. -

-7-, 7 7L

L. Kaufman

0 0 4..

.,4,.4

oo
(nn

es

in
0

W+

oc
0

00 U

00

C4

.,4

0 0CD .J~0

0
'00~i

+.01 006 -

4' Inn

z.
-1S.

CUC

0'

, FRI"

Vi

QUASIBINARY AND QUASITERHARY OXIDE SYSTEMS

-II

In

-4 V 0

C4

0 (n

.m 0

0 4

Inn

kn cc 0 0o

+)

+ 41

cc :3 d
r-

.1
>u Enn

C; 0 Inr
cn

c +. r I

IC
1~14

L. Kaufman

References 1. 2. 3. 4. L. Kaufman and H. Nesor, CALPHAD 2 35(1978) G. Kaestle and K. Koch, CALPHAD 2-198(1978) L.J. Gaukler, H.L. Lukas7T-i .-Henig and G. Petzow, (CALPHAD) 2 349(1978) ManLabs-NPL Data Bank, ManLabs, Inc. Cambridge, Massachusetts USK, National Physical Laboratory, Teddington-Middlesex, England S. E.M. Levin, C.R. Robbins and H.F. McMurdie, Phase Diagrams for Ceramists, American Ceramics Society Columbus, Ohio (1964), First Supplement (Ibid) (1969), Second Supplement Ibid (1975).

-3

I.

'i

7-RT477 n'

77

Section IV. Calculation of QuasiBinary and QuasiTernary OxyNitride Systems, CALPHAD (1979) 3 pp 273-89

CALCULATION OF QUASIBINARY AND QUASITERNARY OXYNITRIDE SYSTEMS - III* Larry Kaufman ManLabs, Inc., 21 Erie Street Cambridge, Massachusetts 02139 USA (This paper was presented at CALPHAD VIII, Stockholm, SwedenMay 1979) ABSTRACT. A data base is being developed for calculation of quasi-binary and quasi-ternary phase diagrams of ceramic systems. Initially the base was restricted to the following oxides: Cr203 , MgG, Fe20., Fe 304 , FeO, A12 03 , CaO and SiO 2. Lattice Stability, Solution and Compound Phase Parameters were derived covering the liquid, spinel, corundum, periclase, crystobalite, tridymite and quartz phases which appear in the binary systems composed of these oxides. These parameters were selected in accordance with the observed thermochemical properties and phase diagrams of these binary systems. The usefulness of this analytical description was illustrated by computing isothermal sections for ternary systems composed of these oxides. The above mentioned data base has now been expanded to cover combinations of the oxides with silicon nitride, Si.3N 4, and aluminum nitride, AIN thus permitting computation of oxynitride SIALON ceramic systems which are of current practical interest. The current analysis provides isothermal sections in the Si3 N4 -A1 203 203 -AIN and Si0 2 -A1 of of formation free energy for the as computed values as well Si N systmes 3 4 0 3the beta sialons in the Si3N4-A 203-A1N system and the X phase in the Si0 2-A2 calculations with independent are compared former results system. The Si N 3 4 reported by Doerner et.al. at CALPHAD VIII and published in the current issue of the CALPHAD Journal. 1. Introduction The utility of computer based methods for coupling phase diagrams and thermochemical data for metallic systems has been well documented in many papers published in this journal. A considerable effort is being applied toward developing an extensive data base for metallic systems. Recently similar efforts have begun with the aim of developing similar data bases for ceramic systems (1-4). Recent development of SIALON composites, combining silicon and aluminum nitrides with oxides of silicon, aluminum, magnesium and other metals has provided additional notivation for developing methods for predicting multicomponent phase diagrams of ceramic systems. The present paper describes work conducted at ManLabs during the past year and compares the results obtained independently at the Max Plank Institute in Stuttgart (4). 2. Description of the Thermochemical System Employed to Describe Solution and Compound Phases The iethod utilized for describi.ag solution and compound phases is the same as that employed earlier (1,2) employing some symbolic usage which facilitates data handling as indicated below. The free energy, GL, of a liquid (solution) phase, L, in the binary system I-J is given by Equation (1) where T is in Kelvins, x is the atomic fraction of J and the mass basis is a mole of atoms (i.e., a gram-atom).
L G . (I-xJ GL + xoGL + RT(xtenx + (1-x)tn(1-x))+ x(l-x)[LIJ(1-x) + xLJI]

(1)

This work has been sponsored by the Air Force Scientific Research, Bolling APB, Washington, D.C., under Contract F44620-76-C-0060.

-45-

?
._~~~ " " a I -, I . , . .. "IJI p

In this expression, G
1

and

0G

are the free energies of a gram atom of pure

*J

liquid I and J, R = 8.314 J/g.at. *K, and LIJ and LJI are functions of temperature (but not of composition). The composition, x, is the atom fractionPof oC J. Similar equations L can be written for each solution phase with .GLi replaced by Gi, Gi , etc., for the P=Periclase, C=Corundur phases, etc., and with LIJ replaced by PIJ and CIJ etc. In the present symbolism, onr uses the symbol SNSNLB to d3signate the free energy of pure liquid Si3N 4 minus the free energy of the beta form of Si3N4 . Reference to Tables 1 and 2 show SNSNLB=33949-12.SlT Joules/g.at., and that the liquid phase of the SN-AO, (l/7)Si 3N4 -(l/5)A1 203 system (i.e. SN-AO with i=SN, j=AO and x=atom fraction of AO) is described by LSNAO=29288 and LAOSN=29288. G= The free energy, G , of a compound phase i (lx,)Jx* is defined by Eq.(2) as (l-x.) G i
+

x
.

G? + x*(l-x*)[(i-x*)LIJ + x. LJI-C[T]]

(2)

where 8 refers to the base phase selected for the compound and C[T] (which is a temperature dependent function) is the compound parameter. Table 2 lists the compound parameters and base phases assigned to compounds in the systmes of interest. Thus the free energy of one gram atom of the compound AlN'Al 203 which is defined as (/7)(AlN'Al 203 ) or AN 0. 28 6A 0.714 is defined as .8 B G7= 0.286 GAN + 0.714 GAO + (0.286)(0.714)[(0.286)LANAO 0.714 LAOAN-C] (3) Reference to Table 2 shows that LANAO=LAOAN=29288 and the compound parameter C in this case is equal to 146440. Thus the free energy of formation of AN0. 2 86AO0. 7 14 from the beta In order to compute the free energy

Si3N4 forms of AN and AO is equal to -23922 Joules.

of formation of Al 0.28 6A 0.714 from the stable N(hexagonal) form of AN (i.e. AIN) and the stable C(Corundum) form of AO (i.e. A1203 ) it is necessary to include the lattice stability terms ANANBN and AOAOBC. The former is contained in Table 1 (ANANBN=ANANLN-ANANLB"425930.80T) while the latter AOAOBC=AOAOLC-AOAOLB=23640+2.30T. With this additional information the free energy of formation of AN0 .2 8 6AO0.714 from the N form of AN and the C form of AO can be computed as is shown in Table 3. Here the free energy of formation of (1/7) (AN'Al 203 ) from (1/7)A1N (i.e. 0.286AN) and (l/7)(Al 20 3)(i.e. 0.714 AD) with AN in the hexagonal, N, form and AO in the Corundum form as illustrated by Equation 4. AG =0.286 GN N) + 0.7t
= 0.286 (ANANBN) + 0.714 (AOAO3C) - 23922

(S) (6)

= 5138 + 1.418T Joules/g.at.

Tables 1 and 2 summarize the lattice stabilities, solution phase and compound phase parameters for the quasi-binary oxynitride systems of interest while Table 3 displays the free energy of formation of the compound phases which form in the subject binary systems. The quasi-binary systems computed on the basis of Tables 1 and 2 in combination with the lattice stability values presented earlier are shown in Figures 1-7 which are all computed at one atmosphere pressure! Thus these diagrams represent metastable equilibria at temperatures above the decomposition points of AN and SN. This matter will be taken up in Section 3. Reference to Table 3 shows that several metastable phases are described in the AN-AO system. These phases designated,AN0. 7 06AO0. 2 9 4 (G) and AN0.286AO0.714(B) display positive free energies of formation of 1.121T Joules/g.at. and 5138+1.418T Joules/g.at. respectively. Indeed the latter was discussed in Equations (3-6) above. In equation 2 it was shown that the free energy of formation of this AN0 .28 6AO0.714(B) phase from AN(B) and AO(B) was negative i.e. -23922 Joules/g.at. However Equations (3-6) which compute the free energy of formation if this phase from AN(N) and AO(C) yield the positive value in Table 6. Thus this phase does not appear on the AN-AO diagram shown in Figure 3. Reference to Figure 3 shows that neither the above mentioned AN0 7 06 AO0 2 94 (G) nor the AN0 28 6A00 714(8) phase

S-46-

..-

-~

TABLE 1 S4UMARY OF LATTICE STABILITY PARAMETERS (All units in Joules per gram atom (mole of atoms), T in Kelvins) AO (I/S)A1 203 , SO = (I/3)SiO 2, MO = (1/2)MgO, AN = (I/2)AlN, SN = (l/7)Si 3 N4

P a Periclase, C = Corundum, S = Spinel, X = Crystobalite, T = Tridymite H = a quartz, R = B quartz, B = beta Si3 N4, N = hexagonal AIN, L = Liquid SNSNLB- = (1/7) Si3N4 (Liquid) - (1/7) Si3N4 (beta) ANANLN = (1/2) AIN (Liquid) - (1/2) AIN (hex) etc.

SNSNLB = 33949 - 12.SIT SNSNLN = SNSNLP = SNSNLC = SNSNLX = - 13.31T - 12.SST - 10.21T
-

ANANLN = 42593 - 13.31T

ANANLB = - 12.SlT - 10.21T - 10.21T
-

ANANLP =
ANANLC = ANANLX = ANANLS = (reference I) AOAOLB = (reference 2)

1.67T 8.37T

1.67T

SNSNLS = MOMOLB = MOMOLN =

SOSOLC = SOSOLB = SOSOLN =
*

- 8.37T

OMWOLP = 47698 - 15.40T - 12.SIT - 13.31T 2.09T -2.09T -2.09T

- 12.SIT

- 13.31T AOAOLN = AOAOLC = 23640 - 10.21T (reference 1)

SOSOLX = 3347 -1.674T (reference 2) These differences specify the free energy of one phase (i.e. liquid) minus the free energy of the second phase (i.e. beta) for a given compound.

is stable. Indeed these phases are the unstable counterphases for the stable SO.412ANO.588 (G) and the SN0 .9 80ANO.020(B) phases which appear in the SN-AN system shown in Figure 2. This definition of the metastable AN0 .70 6AO0.294 (G) and A"O.286AO0.714(B) phases are required for calculation of the SN-AO-AN ternary system described in Section S. Tables 1 and 2 permit computation of the melting points of these metastable compounds. In particular subtraction of equation 2 from equation 1 provides the description of the free energy of fusion of compound phase V when x=x..

Thus4
AGL GL-G7-(l-x.)[ L OGO]+x,[*G LOGjI+RT(x. nx,+(l-x,)tn(-x,))+x,(l-x,)C (7)

The.free energy change on fusion for the AN0 . 2 86 A0. 714 (B) phase with efB and C=146440 is given by GL-G

LB
-

0.286(ANANLB)+0.714(AOAOLB)-4.977T+29904 29904-17.487T Joules/g.at

(8) (9)

yielding a melting point of 1710"K which lies below the stable 2200"K eutectic in Figure 3. A similar calculation can be performed for the AN0 7 06A0o. 29 4 (G) phase with e=N and C262760 from Table 2.

-47-

--

-K5

TABL1" 2 S1P4AY OF SOLUTION AND COMPOUND PARAMETERS (All units in Joules per gram atom (mole of atoms), T in Kelvins) Solution Parameters
LANSO = LSOAN = LSNAN = LANSN = LANAO = LAOAN = LANJO = LMOAN = 29288 LSNSO = LSOSN = LSNAO = LAOSN = t1NOSN = LSNMO = 29288

NANSO = NSOAN = BSNAN = BANSN = NANAO = NAO.AN = NANMO = NWOAN =

XANSO = XSOAN = TANSO r TSOAN = HNSO = HSOAN = RANSO = PSOAN NSNAIN = NANSN = 125520 CAAtO = CAOAN = 125520

125520

PAN.O = PWOAN = 125520 BSNSO = BSOSN = XSOSN = XSNSO = TSNSO = TSOSN = WNSO BSNAO = BAOSN = CSNAO = CAOSIN = 125520

HSOAN = RANSO = RSOAN = 125520

BSNMO = B OSN = PSNMO = PMOSN = 125520 Compound Parameters Compound (1/50) (7Si 3N4 -(1/2)AIN) S AIN) (1/17) A203) (I/17) (6AIN -l (1/ 7)(AIN" Al203 ) (1/17)(AIN 3A1 2 03 ) (l/12)(SAIN-MgO) (1/SO) (7Si3 N4 . (l/3)SiO 2) (/10)(SNO.700SO00 (llO(i 2) 3 ,i (1/SO) (7Si3N4 (1/S)A1203 ) (1/SO)((1/2)MgO.7Si3 N 4) Here GL-G G = O.706(ANANLN)+0.294(AOAOLN)-5.036T+13027 z 43098-18.346T Joules/g.at. (10) (11) Name B G G B SP Stoichiometry SN0.980ANO.020 SN AN Stability Stable Stable Metastable Metastable Stable Stable Stable Stable Stable Stable Base B N N B S N a B B 9 Compound Parameter Jouies/g.at. 83680 + 4.184T 96232 62760 146440 12008112.678T

(/1)(Si 3N4 SAN

0.412

0.588

ANo 706AO0.294 AN0 . 28 6A0.714 AN0 .II 8 AO0 88 AN0 833400.167 SN0.980500.020' 0N.700 0O.300

Q
B

104600 - 8.368T 71128 + 4.184T 115060-- 2S.104T 65270 + 4.184T 112968 + 4.184T

B B

SN Mo

9 8 0 AO 0 2 0 0 2 0 SN0 9 8 0

yielding a melting point of 23490K which is unstable since it lies within the two phase L+N field shown in Figure 3. Finally mention should be made of the Sn0.9 8 0ANo. 0 2 0 (B), Sn 0.80SO0.020(B) and 3 and in Sn 0 . 98 0 A00. 0 20 (B) and MO 0 . 0 20 5no.980(B) cimpound phases shown in Tables 2 Figures 2,5,6 and 7. These "compound" phases are defined for mathematical convenience only so that the extension of the Si-N 4 or SN(B) phase into a ternary system can be made along the lines used previously for metallic systems(S). Examination of the free energy of formation of these compounds show very s.nall negative values which could be representative of a dilute solution phase.

-48-

,7

L!

TABLE 3 SUMMARY OF CALCULATED FREE ENERGIES OF COMPOUND FORMATION (JOULES, OK) REACTION (7/5O)Si3N4 (B) + ji/IO0)AlN(N)-(I/SO)(7Si 3N4 -(1/2)AlN)(B) (1/17)Si 3N4 (B) + (S/17)AIN(N) (Si 3(/17) 3 4 .SAIN)(G) AG -213 - 0.098T -2234 + 0.326T + 1.121T 5138 + 1.418T -SIS - 0.2S9T -2510 + 0.81ST -720 - 0.102T -16496 + 4.971T -234 - 0.036T -686 - O.140T

(6/17)AIN(N) + (1/17)A120 3 (C)-(I/17)(6AlN.A1 203) (G) (I/7)AIN(N) + (1/7)A1 203 (C) +(1/7)(A1N.A1
2 03 )(B)

(1/17)AlN(N) + (3/17)A1203(C) -(1/17)(A1N.3A1 203 )(SP) (S/12)A1N(N) + (1/12)MgO(P) -(1/12)(SAIN.MgO)(Q)

(7/S0)Si3N4 (B) + (1/ISO)SiO 2 (R)-(l/SO)(7Si3N4 (1/3)SiO2 )(B) (1/1O)Si 3N4 (B) + (l/1O)SiO 2 (R).(I/lO)(Si3N4*SiO 2) (R) (7/SO)Si3Ni *(I/SO)(7Si3N4 "(1/S)A1 203 )(B) 4 B) + (1/S)AI20 3 (C) (1/100)MgO(P) + C7/50)Si 3N4 (B).(I/SO)((I/2)MgO.7Si 3 N4 )(B)

3. Calculation of the Melting and Decomposition Temperatures of AIN and h!3 4 Examination of Figures 1-7 shows that the decomposition temperatures for AIN and Si3N 4 are computed to be 2850K and 2129K respectively at one atmosphere.lhese temperatures are below the computed melting SNSNLB in Table I at 3200K and 2714K decomposition temperatures of AN and (1,6-12) by defining Equations 1 and points of AN and SN which are defined by ANANLN and respectively. Calculation of the melting and SN was carried out along the lines developed previously 2 for the liquid and A. 50 No.50 phase in the Al-N Table 4 shows the

system and the liquid and Si0o 4 29N0.571 phase in the Si-N system.

results in terms of Lattice Stability Parameters (4.1), Solution Phase Parameters (4.2), Compound Parameters (4.3) and Free Energy changes on Fusion 4.4. The latter provides the basis for the values of ANANLN and SNSNLB in Table 1. The equations displayed earlier in connection with the melting point of AN0 .28 6AOo. 7 14 (B) and ANo. 70 6AOo. 2 94 (G) and derived by subtraction of equations 1 and 2 (i.e. Eqs. 7-11) are used in defining the resuIts in (4.4). The Table 4 description can also be employed to compute the free enezgy of formation of AN and SN from nitrogen and liquid or solid forms of aluminum and silicon. The results shown in 4.S are in good agreement with the assessed Data Bank values (8). These results yield the predicted one atmosphere decomposition points. Finally, the sojubility of nitrogen in liquid silicon and liquid aluminum can be computed (4.6) and are found to be in good agreement with observed values (9,10), 4. Calculation of the "i02-AI 2 0 3 -Si 3 N4 Ternary System The isothermal section at 2200K in the SiO2-A 2 0 3 -Si 3 N44 syste was computed and

is shown in Figure 8. The ternary liquid(L), corrundum(C) and B(beta) solution phases were described by means of Kohler's equation with zero values for the ternary interaction parameters. The mullite(M) and silicon oxynitride(R) compoundswere defined as shown in Figure 8 with the counterphase paravters CB=O and CAB=O along the lines previously set forth in Equations 4-6 of reference (2). In addition a ternary X phase was includedr corresponding to SO0. 20 AO0.6SSN 0 . The compound parameter for this phase 1 5 (3,4).

-49-

~a~1 ~
-,

-.

r.

'

~
ry,

jVj~
.~i
4

-?-J

-L.@

y: .,

-~;

no,1

W/o S'02, AIN 10 TOK M"etastable! 3000 20 30 40 so 60 70 80 90 NOTE: A1N decomposes /] above 28S0K at I atm. SiO 2

..... L.+X

.... it

2000 NiT (Tridymite) 1000 N+H(a Quartz) 850 N+R(a Quartz)

f, I

"17S01080

,.,

,.

I ,,

I . .

Al0

5N0 5=AN

SO=Si

500

667

Phase Diagram at one -Si 0 N Figure "1. Calculated Al 0S 0.5- 0.33350.667 atmosphere w/o AIN Si3N4 T*K MetastableLiquid 3000
___above

10

20

30

40

50

60

70

80

90

AIN

2500 20 :--Si3N4 decomposes above 2129K at I atm. into N 2 and Si 1000 B+G 2430

I Metastable! AIN decomposes 2850K at 1 atm. into Al and N2

SN=Si 0

4 2 9 N0 5 7 1

(1/17)[Si 3N 4 SA1N]=G=SN0. 4 12AN0.588

4 2 9N 71

A10.5N0.5 =A N Phase Diagram

Figure ..2. Calculated Si

0.333

at one atmosphere

"

-I-

AiM 10 TK Metastable 3000 L+JN 2000. 20 30

w/o Al 20 3A1 60 s0 40

2 03

70

80

90

AiN decomposes into Al and N2 atl1atm. above 28S0K (iud L(iud 2207 2200 2190

woo 1000

NSP

SP
+

ANq=Al10 5 N0 . 5 *Figure

SP=(l/17) (3A1 093 *A1N)

0.1 AO 0.8=SP

Al 0'40 0.=AO

Phase Diagram at one 0 -Al N ".3. Calculated Al 0.4 0.6 0.5 0.5 atmsphre w/o MgO-AiM T*K 10 Metastable 20 30 40 s0 60 70 80 90 MgO

'AlN decomposesinto

N+

Q.P (Periclase)

AN=A

5 0 5N0.

Q=A 0. 83?00. 167' (l/12)(SAlN*MgO)

Mg 0 . PO 5=1

S -M~g0* 5 Phase Diagram at one 500 . Figure 4. Calculated Al 0 . 0 *5 5N atmosphere

-4,
K

ORO
4f

"4go

10

20 2434

30

W/o S N 60 40 so

70

80

90

Si 3N4

24000 P(Periclase) 1600

+

B Metastab le Si 3N4 decomposes above 2129 at 1 atm into N2and S

800

om~O.M 0 500.5 Figure

SN=Si 0 . 429 N05 71 S. Calculated Mg O0 O S -Si 0 . 429 N 5.71 Phase Diagram at one atmosphere

(i.e. C=47,700) shown in Figure 8 is defined on the basis of Table VIII of reference 12 which defines the free energy of a ternary compound by means of Equation (12) as
=X

0.20 -GC+0.6S 'GA+O.lS

[(.0(.S(.S 0.2OLSOAO+0.65LAOSO-C]

.(0.20)(0.15)(0.3S)-

[0.2OLSOSN.0.lSLSNSO-C].(0.65)(0.lS) (0.80)- [0.6SLAOSN+0.lSLSNAO-C] (12)

Since LA0S0=284Sl and LSOAO=38492 from reference (2) -946 Joules/g.at. C0 Gx 0.20 OGC.O 6 A 0.1 5OC SN-96 S;O.6 or the free energy change for the reaction 0.067Si 13Al 203 4 (B>-*[0.067Si0 2 .0' 2 (x).0.13A 2 03 (C)+0.0214Si3 N
=

(13)

'0.0214Si N 4 (X)

(14) (5

AG 0.20(SOSOCX)+0.lS(SNSNCB)-9460

-3700-0.261T Joules/g.at.

S. Calculation of the Si N -Al 0 -AlN Ternary System

The ternary isothermal section for the SN-AO-AN system shown in Figure 6 was calculated along lines similar to those described in Section 4. The solution phases covered included the liquid, L, corundum, C, hexagonal, N, and Beta, B, phases described by Kohler's equation with zero valued ternary interaction parameters. In addition, the spinel AN 0 118A0 0 882 (SP), SN 0 412 AN 0. 588 "AN 0. 706 A0 0 294 (G) and SN 0 980 A0 0 .020AN0 .286A0. 714 (B)compound phases are included. The spinel compound phase is des%:ribed with a SN 0. 118 A0.882 (SP counterphase with O=S, C8 and CAB equal to zero. The base phases and the compound parameters for the B and G compound phases are shown in Figure 9. The prior discussion in Section 2 described the metastable counterphases AN0 . 706 A0.294 (G) (B). These metastable phases are described with base phases of N and Bt anid AN 0 2 AO respectively and with compound parameters which are 62760 and 146440 Joules/g.at. respectively as shown in Table 2. Panels (a)through (c)in Figure 9 show the computed equilibrium between the N/L, B/L and GIL pairs at 2200K. The B/L equilibrium shown in

3 4

Pane

(b

wascopuedwih__vluof________nrdroetedheBbeai
-,4

*Si
-T*K

3N4

10
-1 -

20
I

30
I

40
I

W/o SSi0 70 60 s0
I I III

80

90

So2

2428 2100 -- Metastable Si 3N 4decomposes above 2129K intjS 140 N2and Si B+R
700
*

L+XX=Crystoba l if

L 7x 14 R+T(Tridymite) R+ct(Quartz)
850

1080

R+8 (Quartz)

L~L0.429

0.571

0.700.30=

'2 2~'

.. 33

.6

*Figure,

Si3N4

6. Calculated Si 0.429 N0.571 -Si 0.333 0 0.667 Phase Diagram at one atmosphere w/o Al* 0 0 2 .5 Al 0 SiN 90 23 70 80 50 60 20 30 40 10 TOK Liquid

1400

Metastable Si 3N4decomposes above 2129K at 1 atm. into N 2 and Si B+C (Corrundum)

SN=Si 0 . 429N0 . 571 Calculated SiO.2 N0.7 Figure .7. 0.2 0.7 atmosphere

AO=Al 0 4 00 . 6 -Al 0006 Phase Diagram at one 0406

VO
_ U_

TABLE 4 SUMM4ARY OF ALUMINUM-NITROGEN AND SILICON-NITROGEN SYSTEM ANALYSES FOR ESTIMATION OF THE MELTING TEMPERATURES AND HEATS OF FUSION OF AiN and Si 3N 4(Joules, g.at. @K) V=Vapor, L=Liquid, D=Diamond Cubic, A=fcc, B~bcc 1. Lattice Stability Parameters ALALLB= 628- 6.69T ALALLA=10711-ll.SlT ALALBA=10083-4.82T SISIBD=44350-l9.54T SISILD=S0626-30.06T SISILB= 6276-10.46T NNNNLB= '879-1757T (reference 6) ItNNNNBV=-3682+S3.64T to NNNNLV=-2803+36.07T (reference 5) o t 2. Solution Phase Parameters LALNN=-S0208, LNNAL=-912112+2S6.48T, LSINN=14644, Lx-674879+23S.l4T 3. Compound Phase Parameters (Al 0. 0
5

; C.zl67360+l8.4lT C=12S520+31.38T

base=bcc base=bcc

(0 . 4 29N O.71)

4. free Energy Change on Fusion Al 0

5N 0

AG

O.S(628-6.69T)+0.S(879-17.57T)-5.77T.,25 (167360+18.41T)
s42593-13.31T =

ANANLN(Table 1)

; Melting Point=3200*K

Si 0

42

N0 AG = 0.429(6276-10.46T) + 0.571(879-17,57T)
-

S.68T+0.24S(12S520+31.38T)

a33949-12.SlT = SNSNLB(Table 1) ; Mel..ing Point=2713*K 5. Free Energy of Formation 0.5 Al(L) + 0. 25N 2 V) 1.N.

AG - 0.50Cr-628+6.69T) + 0.5(-3682+53.64T) + 0.25 (0O5(-50208)+O.50(-912112+256.48T)-167360-18.41T) =-164285 + S7.625T; Decomposition at 1 atmosphere at 2850*K

0.5 Al(fcc) + 0.25N (V) - Al 0 5 N0

AG
=O.SO(10083-4.82T)

+ 0.5(-3682+53.64T)

+ .25(0.5O(-S0208)+O.SO C-912112+256.48T)-167360-18.41T)
=-158929

+ S1.87T

-54

TABLE 4 (concluded) 0.429 SiCL) + 0.2855N 2 (V) + Si 0 .4 29 N0 .571 AG = 0.429(-6276+10.46T) + 0.571(-3682+53.64T) + 0.245(0.429(14644) + 0.571(-674879+235.14T)-12SS20-31.38T) a -128420+60.32T; Decomposition at 1 atmosphere at 2129*K 0.429 Si(D) + 0.2855N 2(V) Si0. 4 29 N0 .571

AG = 0.429(44350-19.54T) + 0.571(-3682+53.64T) + 0.245(0.429(14644) + 0.S71(-674879+235.14T)-12S520-31.38T)

= -106702+47.44T

6. Calculation of Nitrogen Solubility in Liquid Aluminum and Silicon Al-N (Solubility of nitrogen in liquid aluminum) RTlnx = ALALBL + NNNNBL - LALNN + 0.50(0.5(-50208)+0.S0(-912112+256.48T)-167360-18.41T) z -275559 + 79.18T ; x=0.0001=0.Olatomic percent at 1773K Si-N (Solubility of nitrogen in liquid silicon) RTlnx = (0.429)(0.571)- 1 SISIBL+NNNNBL-LSINN + 0.429[0.429(14644)+0.571(-674879+235.14T)-125520-31.38T) z -236701+69.55T ; x=0.00017=0.017 atomic percent at 1673K. compound phase out into the ternary system (3,4). Panel (d) shows a composite of the B/L, GIL, N/L and S/L equilibrium. This composite constitutes the ternary section at 2200K and should provide a picture of the phase constitution at lower temperatures. Calculation of ternary sections at lower temperatures, (wher. the liquid disappears) can be readily carried out on the basis of the parameter values developed here. It should be noted that the isothermal sections shown in Figures 8 and 9 are metastable with respect to the one atmosphere decomposition of Si3N 4 at 2129*K (see Section 3). 6. Calculation of the Free Energy of Formation of Beta SIALONS and Comparison with the Computed Results of the Stuttgart Group(4) The foregoing results can be compared directly with those presented earlier (3,4) by the Stuttgart group with respect to phase equilibria and the free energy of formation of the beta SIALONS. Figures b and 9 should be compared with Figure 7 of reference 3 and Figure 16 of reference 4. With respect to the former comparison Figure 8 covering the SN-SO-AO ternary section should be compared with the upper left portion of the SiO 2 Al 20 3 -AlN-Si 3 N 4 square in Figure 7 of reference 3 which corresponds to the Si N -SiO-Al 203 right isosceles triangle. Here the X1 phase shown in Figure 7 of reference 3 corresponds to the current X phase in Figure 8. Figure 9 should be compared with the lower right isoceles triangle Si 3N4 -A1 2 03-AIN shown in Figure 7 of reference 3. The present diagram includes the "beta Si 3N 4" phase and the G phase. The latter corresponds to the X7 phase in in reference 3 are not included in Figure reference 3. The phases labeled X2 ,X 4 ,X S and X

9. The next step in the comparison is computation of the free energy of formation of beta sialons." This calculation is detailed in Table 5 starting with the expression for GB given by Equation (16) in Table 5. Reference to Equation (16) and the succeeding relations through Equation (21) shows that the free energy of the beta sialon phase which projects into the Si 3N4 -A1 20 3 -AIN ternary from the Si 3N 4 corner toward the AIN.Al2 03 or Al 30 3N composition can be defined explicitly in terms of the compound parameter for the

-55-

17,

v.,gf

itg

4)
C14

If~

00

C44

ei

+4

L)

+0

or4.
H

0q C/)

0404.'

co

4JC

C
i r

) 04

.:

4 0 0N 0 0 H .0

dO
.9 :3

C4

on) V-

V)

-5d6-

M 0
9.44

%0)

~-

ti))

go4J

00

4J

4J

4J

U))

k.

0o10

J 0

Q.C)

~4J
e.c 00 Ln C4
C; It4

V-#

P-4 E

'4

U.4 4

'V-

.oC4

0
IN

)
00

''

,..)ZI

zro

(D~I~

%:4

Z-

4')

C'

TABLE S CALCULATION OF THE FREE ENERGY OF FORMATION OF BETA SILICON NITRIDE - ALUMINUM OXYNITRIDE SOLID SOLUTIONS (Joules, *K) G
=z

OGSN
+

GAO + y
+

%A + Cl-(y/(lx)))6GA + (y/(l-x) AGB

-

RT(ylny

zlnz

C1-x)ll(l-x))

AGE

(from reference 5)

(6

where x~XAo and Y=A

and z=zSN=l-x-y as in Figure 9 .074

Since the solid solution runs from Sn AD to AN A then n 0.286A 0.714 0.980 0.020 020and1 x!0.2 an p=(x+-x.)/Cl-x0 )=2.426. Thus

~O74
(17)
(18)

x=x*+yp
z=l-x,

0.02 + y~l~p)
=

.46y
-

(from reference 5) 3.426y

0.98

with AGA = (0.02) (0.98) [0.O2LSNAO+0.98LAOSN-C]=0.0196[29288-65270-4.184T]

a-705
-

0.082T

(19)

and

&GB = (0.286) (0.714) (O.286LANAO+0.7l4LAOAfl-CI=0.2042[29288-1464401

a -23922

(20)

since the compound parameters, C, in these equations refer to the values listed in Table 2
frS0 . 9 80A0. 020 adA Thus B G ,(O.98-3.426y)*Gs
+
0. 28 6A0 0

714 respectivl B B
+ y (0.98-2. 4 26y)-l (-23922)

+ (0.02+2.426y)GAO + yG

(1-y(0 .98-2. 42 6y) .14)(-705-0.082T)

+ RT(ylny+(0.98-3.426y) ln(0.98. ..426y)-(0.98-2.426y)ln(0.1)8-2 .426>'))

+ CABy(O.98-3.426y)(0.98-2.426y) 1 The free energy change for the reaction (O.14-0.490y)Si 3 N4 (B)e(0.004.0.485y)A 203 (C)+0.Sy AIN(N) [(Si 3N4) (0.14-0.409>') (Al 203)(0.004+0.'e85y) (AIN)0 .5>'](B) is AG
=

(21)

(22)

G8

(0.98-3.426y)*GNGC G AD
=

(0.02+2.426')*GA

(23) (24)

or since *G B AO

AOAOBC=AOOCAAL=34+.0 OOCAAL=34+.0

from Table 1 and Reference I and

0GB 0G

ANANBN=42S93-0.80T from Table 1

(25)

AN77777

TABLE S (concluded) Then Equation (23) yields the following result when CAB=-83862 J/g.at. AG = (0.02+2.426y)(23640+2.30T) + y(42S93-0.80T)
+
4 (1-y(O.98-2.426y) I)(-70S-0.082T) + y(O. 9 8-2 . 26y)- (-23922)

+ RT(ylny+(O.98-3.426y)ln(O.98-3.426y)-(0.98-2.426y)ln(O.98-2.426y))

-83862y (0.98-3 A26y)(0. 98-2.426y) -l

(26)

for Oy5.O.286. Thus Eq. (26) permits explicit calculation of the free energy of formation of the beta silicon nitride-aluminum oxynitride solid solution as a function of concentration along the join betweenSN(I/7)Si3N4 ) aid ANo.286AO0.714((1/7 AIN'A20 3)or (1/7)A
3 N30.

If Y is the iole fraction of Al3N30 along the Si3N4 -A1 3N30 join then Y and

y (the atom fraction of AN in the SN-AO-AN ternary shown in Figure 9)are simply related by Y = (0.98/0.286)y+0.02 = 3.426y+0.02 (27)

metastable AN0. 2 86 AO0.714 phase, C=146440, given in Table 2 and the value for CAB=-83862 Joules/g.at. which was used to compute Figure 9. Equations 21 through 26 show how the free energy of formation of the beta sialon phase can be computed as a function of composition when the compounds- beta Si3N4 , the corundum form of A1203 and the hexagonal form of AIN are considered to be reactants. Thus equations (26) and (27) can be employed to compute the free energy of formation of beta sialons according to equation 22 as a function of composition. Figure 10 shows the result of such a calculation which was kindly carried out by Dr. H.L. Lukas (3,4) as a function of the mole percent Al303 N. The mole percent is J00 times the mole fraction, Y, given in Equation 27. In Figure 10, the curve labeled L. Kaufman is that given by Equation 26 while the curve labeled P. Doerner et al is the identical free energy change as computed by the Stuttgart group from a completely different starting point using an entirely different model for the liquid and beta sialon phase. The differences betveen the two curves are less than 4KJ/g.at. (i.e. 1 kcal/g.at.) which is quite remarkable. If the values of CAB-83862 and C=146400 Joules/g.at. employed in Table S to develop Equation 26 are reconsidered,the difference between the two curves can be reduced to less than 1600 J/g.at. (i.e. 400 cal/g.at.). Thus the current value of C=146440 yields a positive free energy of formation for the metastable compound AN0. 2 8 6A0 0 .714 as noted in Section 2, Equations 4-6 and Tables 2 and 3. Moreover the melting point of this compound, based on C=146,400 Joules/g.at. is 1710K according to Eq. 9. If the value of C were increased to C=167,360 Joules/g.at. Equations 4-6 and 9 would lead a free energy of formation (from hexagonal AIN and A120 3 (C)) of 866+1.418T Joules/g.at. as compared with the current value of S138+1.418T listed in Table 3. Thus the increase in C would make the free energy of Ano. 28 6 AO0. 71 4 more negative than it is currently, but still positive. Finally alteration of C from 146,400 to 167,360 would increase the melting point of AN0 AO0 from 1710*K to 1956*K, which is still below the 2200K eutectic in Figure 3. Thus if C=167,360 Joules/g.at. for AN0 . 2 8 6AO0.714 the upper curve in Figure 10 would be sheared downward until it coincided with the lower curve at 100 mole percent Al 303 N. Conversely one could inquire as to the accuracy of the lower curve at 100 mole percent Al303 N since the latter phase is unstable! The final step in bringing the upper curve close to the lower curve is decreasing CAB from -83862 Joules/g.at. to -92048 Joules/g.at. The effect of such a change on the computed phase diagram (along with the aforementioned change in C from 146400 to 167360) is shown in Figure 11. The net result is an increase in the field of stability of beta Si3 N4.

-59-

-7 7
gr _

*
-

6. Conclusions

lW

*o.sw

ow

IF

The current extension of the quasibinary synthesis of oxynitride ceramic systems has provided a means for synthesizing the thermochemical properties and the phase diagrams of complex multicomponent systems. The utilization of metastable phase descriptions as a means for assessing phase stability has been shown to be of value in deriving information on- the stability of beta sialons. The current description has also been applied to compute the free energy of formation of the X phase in the SiO 2-Si N4-Al 0 system and could be applied to calculation of other oxynitride solid solution phases system). It A' (i.e. the G phase in the Si N -A1 2 0 -AlN 3 4 2 3 hoped that the future development of this interesting and inportant CALPHAD area will provide future opportunities to perform comparison of independent comparisons along the lines of Figure 10 with increasing levels of agreement!

L.

AFNA

P. DOEIRNER El AL.

M011-Z AL303M

Figure 10. Comparison of Current Computed Free Energies of Formation of Beta Si N Solid Solutions at 2000K with those due to Doerner et al (4). SN=(1/7)Si N4

B.L

G+L

L
(1 I7)CS

FAN=(l/2)AlN
Figure 11. Recalculated Composite Equilibrium in the

SPO(15AlO

N4

fI/S)A1O..

(112)AIN

System at 2200K with CAB=-92048 Joules/g.at. and C=167360 Joules/g.at.

-60-jf

414

References 1. L. Kaufman and H. Nesor, CALPHAD 2 35 (1978) 2. L. Kaufman, CALPHAD 3 27 (1979) 3. L.J. Gaukler, H.L. Lukas, E.'h. Henig and G. Petzow, CALPHAD 2 349 (1979) 4. P. Doerner, L.J. Gaukler, H. Krieg, H.L. Lukas, G. Petzow and-J. Weiss CIkLPHAD 3 239 (1979) S. L. Kaufman and H. Neso-, Met. Trans. 6A 2115 (1975) 6. L. Kaufman, CALPHAD 3 45 (1979) 7. L. Kaufman and H. Nesor, CALPHAD 2 325 (1979) 8. ManLabs-NPL Data Bank 9. M Hansen and K. Anderko, Constitution of Binary Alloys, McGraw Hill, N.Y. (1958) 10. R.P. Elliot and F. Shunk, First and Second Supplements (Ibid)(1965) and (1968) 11. L. Kaufman and H. Nesor, Titanium Science and Technology, R.I. Jaffee and I1.Burte, Eds. Plenum Press, New York 2 773 (1973) 12. L. Kaufman and H. Nesor, Met. Trans. S 1623 (1975) 13. L. Kaufman, Summary of ihe Proceedings of the Eighth CALPHAD MEETING, CALPHAD 3 160 (1979)

,,

3 r

Section V. Utilization of Data Bases and Computer Techniques for Solving Metallurgical Problems (to be published in the) Proceedings of a Joint National Physical Laboratory/Chemical Society CONFERENCE ON INDUSTRIAL USE OF THERMOCHEMICAL DATA (September 1979)

!.2

-627 V

Joint
National Physical Laboratory Chemical Society CONFERENCE ON INDUSTRIAL U$~i OF THERMOCHIIECAL DATA held at the

University of Surrey-Guildford, Surrey England 11-IS September 1979 "Utilization of Data Bases and Con titer Techniques for Solving Metallurgical Problems" by Larry Kaufman ManLabs, Inc. 21 Erie Street Cambridge, Massachusetts U.S.A. 01239 telephone 617-491-2900 20 illustrations

-63___ ________ ____ __ _ ____ ____

H 1711"

UTILIZATION OF DATA BASES AND COMPUTER TECMIQUES FOR SOLVING METALLURGICAL PROBLEMS" by Larry Kaufman ManLabs, Inc. Cambridge, Massachusetts 02139 U.S.A. ABSTRACT Documentation of specific instances in which thermochemical data have been employed successfully to economic advantage is difficult because of the secrecy and confidentiality restrictions which of necessity limit publication of such information. Nevertheless, such examples come to light over a period of years and have been properly documented. Moreover, development or computational facility can be assessed as providing a vehicle for evaluating, guiding and synthesizing potential economic projects. The CALPHXD group is an international association of practicing thermochemists devoted to fostering technological advances in thermochemical knowledge and its application to real problems. The fourth meeting of the CALPHAD group held in August 1975 at the National Bureau of Standards inGaithersburg, Maryland provided a list of examples demonstrating the importanc2 of phase diagram and thermochemical data in developing industrial processes. The proceedings of this meeting were published in the CALPAD Journal. &ucceeding meetings of the CALPIL1D group, as well as subsequent journal papers have provided many noresituations The in the fields of: in situ columbium, current paper will review these cases nickel and cobalt base superalloys, sigma phase formation in high temperature alloys, multicomponent semiconductors for electro-optical devices, SIALON ceramics,molten salt, sulfide and oxide slags for ore reduction and refining, high temperature battery electrodes, hardenab'ility calculations in alloy steels, synthesis of hydrogen storage alloy compounds, grain refinement in aluminum alloys, metatectic reactions in hafnium alloys and the growth of HEM silicon solar cell material. Introduction
5 which have The symposia and publications of the CALPHAD group 1-

taken place during the past five years have provided many examples of the application of thermochemical data bases to the solution of industrial problems The CALPHAD Journal, published regularly by Pergamon Press'of Oxford, England provides a vehicle for timely exposition . of informatio, and instances and such applications. Nevertheless, it is useful to compile additonal illustrations *from time to time in order to demonstrate the wide range of applications and to review examples presented in other publications. paper will cover recent cases Accordingly, the present

related to the topics listed in the abstract

-,

which demonstrate how data bank thermochemical data can be applied to the solution of real problems.

~I

'This work has been sponsored by the Air Force Office of Scientific Research Bolling AFB, Washingt-n, D.C. under Contract F 44620-76-C-0060 and the Metallurgy Program, Metallurgy and Materials Section, Division of Materials Research, National Science Foundation, Washington, D.C. under Grant

-64-

7,

1. Calculation of ',uperalloyPhase

Lagrams.

Today there is strong interest

in the development of directionally solidified eutectic alloys for the blades and vanes of the gas turbine engine which are being subjected to increasingly higher temperatures and strLsses. These alloys derive their superior high temperature strength from the directional solidification of antalloy of eutectic composition in a high temperature gradient producing aligned high strength microdimensional fibers or lamellae in a relatively ductile matrix. The potential for increases in average operating metal temperatures of over 100*C have already been domonstrated for the y/y'+6 and carbide reinforced superalloy in situ composites. Limitations in terms of oxidation resistance, These may be ternary or A substantial ductility, and/or castability in these first generation alloys indicate the need for discovering other eutectic Plloy systems.

higher order systems, so there is a requirement for systematically and efficiently identifying promising multi-element eutectics. number of publications are devoted to displaying methods for computing ternary

isothermal sections of superalloy systems 6-1 0. Detailed computations of the Cb-Al-Cr system10 published in 1973 and shown in Figure 1 predicted a eutectic with a minimum melting point at 1317*C and a composition of 4lCr-25Cb-34A1 (wt.%) within the region of the Cr-Cb-CbA13 -Cr2Al 3 composition space. This

investigation suggested that directional solidification of an alloy of this composition should result in an in situ composite having a CbAl3 reinforcing phase in a matrix of 64Cr-iOCb-26A1 (wt.), providing a combination of good high temperature strength and oxidation resistance. investigation of this system by M.K. Thomas
11

Subsequent experimental

shown in Figure 2 verified the

eutectic and matrix compositions and disclosed a meiting point at 13S50C within 33VC of the predictions 10 . Extensive data bases covering binary 13 metallic systems are being developed in the U.s. 1 2 , France 9 ' Germany13 England 4 , Sweden 1 , and Japan which can be employed to compute ternary superalloy systems. Figures 3 and 4 present recent ternary calculations for the Co-Cr-Zr system which are part of a systematic and integrated progran of work buing carried out at the National Physical Laboratory, aimed at identifying and characterizing eutectic behaviour in ternary and quaternary Co + Cr or Ni + Cr based alloys. Calculations of the analogue systemcs Co-Cr-Nb The phase diagram calcuand Co-Cr-Ta, of the corresponding Ni-Cr-X systems, and of some quaternary modifications to these systems have also been made.

1415S

Sen

16

17,18

lations have been made using the well established CALPHAD approach using a newly developed NPL system "ALLOYDATA" which stores critically assessed therm.dynamic and structural data for elements, binary and ternary alloys, retrieves relevant data for a chosen multicomponent system and calculates a wide range of types of two and three-phase equilibria in ternary and quaternary systems. The unusual feature of the sections shown in Figures 3 and 4 is the

-6S-

F'R

0"* 4. *

an NOo

vs

~~

l0a
Q0.

C3

ft

ca. 040

.4alt

r-66

7!

stability of the (Co,Cr),Zr, Laves phase clear across the

02

ternary section. The computed isothermal sections

Sfor

1300-1900K indicate the composition and temperature -ranges over which a number of
2',

eutectic troughs exist, and alo summarize the composition ranges over which the embrittling sigma phase, Co Cr 6 , is calculated to ") exist15 . Of parti-

"04 Figure 3.Calculated 1700K Co-Zr-Cr Section is

cular relevance to the aim of the calculations is the eutectic valley between fcc Co-based solutions (a) and

the (Co,Cr)2 Zr Laves phase which may be of potential use for turbine blade applications. The isothermal sections indicate the solidification path along this valley, and demonstrate how the detailed calculation of phase diagrams can assist in the understanding of fairly complex solidification processes. The calculated compositions of this eutectic valley have been verified by subsequent experimental studies to be within 0.5-2 mol %, although the calculated eutectic temperatures are somewhat higher than the,experimentally detendned values. Bearing in mind the lack of experimental thermodynamic
c.
A

data for the Co-Zr system, and

the uncertain nature of the

liquidus and solidus curves for

the Co-Cr system, the agreement

is very satisfactoryis 2. Sigma Phase. Due to the wide range of applicability of "4
.P

-nickel and cobalt base superalloys and of high chromium

9*,65
8

stainless steels substantial

effort has been directed toward

'.

computing ternary diagramsti.

establishing the stability range of the sigma ce---0 8 ,9 ,13 p1 7. 1 phase ' Figure 5 shows a recent result by Chart, [

Figure 4.Calculated 1600K Co-Zr-Cr SectionlS

-67-

.'.iw.

Y'VI'

'j7'*.

'7
Si

Cr

Ni

Fe
4

figure 5. Calculated Stability Range of the Sigma Phase in the Cr-Fe-Ni-Si System at 800K, for Alloys Containing up to 0.08 Mol Fraction Si. Phase Relationships in the Ternary Cr-Fe-Ni Systems are Indicated' Putland and Dinsdale 14 for quarternary section in the Fe-Cr-Ni-Si system at 800K. This system is of particular interest in the development of stainless

steel tubing for nuclear reactors which is resistant to void swelling under . .. .. 19,20 irradiation1 . A new procedure for evaluating the sigma phase precipitation tendency of superalloys has been developed by Machlin and Shao2 1 . It is called SIGMA-SAFE. Phase diagrams were used to determine the degree of supersaturaIt was found that this supersaturation tion with respect to precipitation of the sigma phase as a function of temperature and alloy compositions.

number, A, distinguishes between 24 superalloys according to their known tendency to precipitate or not to precipitate the sigma phase, except for one superalloy. This capability of predicting the sigma phase precipitation

tendency of an arbitrary' new superalloy composition is not shared by PHACOMP. It was shown that A correlates quantitatively to wt pct sigma precipitated in a specific alloy. SIGMA-SAFE also yields the multicomponent phase diagram applicable to each superalloy21 . The method depends upon fixing the geometry of the two and three phase fields bounding the y,y' and o fields in Multicomponent nickel base superalloys as illustrated in Figure 6. Machlin and Shao carried this out with the help of computed isothermal ternary sections han hnrp A nnqtrntPd Pre11plnt noreement as illustrated Table 1.

-68-

,- ,,."-,
________________TM

-/

Tals. 1. Comprhor of SIGMA-SAFE Predictiins

toBe Alloys Cbwervtd &gnu Prone
-IN

roith Obtanstiofl 2

Sigma Free INCONL700 MAR MI200 MAR M246 TRW 1900 NICROTUNG TRW NASA IVY TRW NASA VI A WASPALOY UNITEM.P AF 21D UN2TELIPAF21DA 80 PEIMONIC ISIONIC 90 32900 M22 W2

, *PDRL

INCO 713C l 0.3

163

4'-~-

1) ____m

I700
= 1

?/
I/~
,-~,.' '

_/TRW *0320 *

'/~ /
--

NX I 7218 1800 USoo U1 Nirnonic I IS RENE41

(0243) (Dordlcrtine) (OhJal) (1.333) (0.465) (0393) (06s2) (Boidertuti) (0.93) (2.562)

%4r

C18)

uiril. eru al ioniit umbc r.dic fined toequal thre Value Inprcenlhe isIie pe so o) 1 ., lcetr rml rco(gc l pt,), -(~ t (c Fig.6-P0crspective view of high N1 region of Ni-Cr-Nlo-Ai Mo) coordntes Inthre phascirrows dotevectors p rcto the (tAlan(p quaternary. Cross-ha~tched region Meines y + '' +c 3O qualcnycorrsondntothleCrciateritortie constant Cr scion of t lu rc.glon. Below and above tho ctoss-h.ttchcd region thc rcqcal.y spcctive 3 phase region* cach contain y + 1, 4 '.21

.'The assessment of their SIGMA-SAFE proposals shows that for the 10 superalloys known to be prone to precipitation of the sigma phase at 87S*C, SIGMA-SAFE correctly predicts all to be supersaturated with respect to precipitation of the sigma phase. Out of a total of 14 superalloys reported to have been held at 875% for long time and in which no sigma phase was found 13 were predicted by SIGMA-SAFE to be in equilibrium phase fields that do not contain sigma while The supersaturation numbe~ir one was predicted to lie in a y+ca field (M252). predicted by SIGMAk-SAFE correlates to wt pct sigma quantitatively, as well as does the NVnumber predicted by PHACOMP. The critical composition for precipitation of sigma in IN 100 is correctly predicted by SIGMA-SAFE. The boundaries between the y,y+a,l+y' and y+y'+i phase fields in the multicornpnent phase diagrams corresponding to Ni base superalloy compositions are predicted by SIGM1A-SAFE as a function of temperature. The y' solvus temperature predicted by SIGMA-SAFE agrees with those observed for 4 superalloys within a 1006C maximum deviation. The predictions of PHACOMP and SIGMA-SAFE for the effects of specific alloying elements in promoting sigma differ quantitatively, and for some elements, qualitatively. This difference cam be important for certain superal loys. 3. High Temperature Battery Electrodes. Solid Li-Al alloys appear to be one -of the leading candidates for the negative -ectrode of high temperature batteries. Nevertheless a loss of capacity, accompanied by changes in morphology of the Li-Al components has been encountered after cyclic operation. Modification of cell design and alteration of the electrode composition is being considered in an effort to ovqrcome the above noted shortcomings and Addition of Mg has been Li diffusion in the a-Al solution22 improve considered as a viable solution, however such a route requires knowledge of the Al-Li-Mg phase diagram. To fill the gap in experimental data, computations of the component binary systems and the ternary sections between 648 and 773K

-69-

T'673 K

Occ.

FCC Co
M9G:FCC

T= 673K
6CC#FCC*

Cc
FCCA FCC -Mg 2 Ca

LFC8

L.8CC-

FCCL.C

MgzCo.L 1.gC~aco

mg 2Ca

Li
HCP

8CC L

FCC.L \FCC#1H49CO4.L

Mg

L.8CC 0.

G-CoL8CC

\-Uq 2CGcBC

T -723K

&cCo
3CC ?P^ga.BCC L BcC

T-773 K

Occ
L*8CC

M2

_--

M92CO-L AUo MOCO-BCC.HKP

L. OC*@L Mg 2 Ce IA9 2Ca.BCC 2\MgCa. LHCP . 'Mg2Co.IICP

U M9&Ce.L.SCC'.u"

1.1

//
-

'

QCO
MgCP
.

IM92 CO-OCC

L.8CC-' *'MgCa*L.CC

Figure 7. Calculated Isothermal Sections in the Ca-Mg-Li System 23 22 were performed . These computations define the field of temperaturescomposition stability for alloys in this system. Recently these computations have been extended to cover the binary components and ternary sections in the 23 Ca-Mg-Li system , Figure 7 shows the results obtained by Hsu and Sabotingi between 673 and 773K in order to define the composition range of stability of the solid and liquid phases. This system is of interest in the development.. of high-temperature batteries in which iron sulfide is the positive electrode, calciumT-magnesium alloys are the negative electrode and mixtures of LiCl-KCl*CaCI2 salts serve as electrolytes. Thus,reaction of the lithiumu from the electrode with the Ca-Mg alloy can lead to formation of cernary Ca-Mg-AlA 4. Hydrogen Storage Compounds. The intermetallic compound FeTi is most likely

to be the first to undergo commercia~l development for energy storage purposes.

It can be hydrided at room temperature under a few atmospheres pressure; the hydrogen can subsequently be released at a few tens of atmospheres by heating4 to SO-100*C. This range of pressure and temperature make iron-titaniume h.;4ride a natural energay storage medium for peak-shaving system.s, solar energy

-70-

Mixs

CC)ost '~site

(d) Experinental

\\ \ %/ ,
oh).

v
A t.=(Fe,11n)o.sTio.
s

-\TX
FeTt FrTe " "

fe

Figure 8. Recalculated (Left Panel) Composite (c) and Experimental Equilibrium(d) in the Manganesa-Titaniufn-Iron System at 1273K 2 4 .

CC) ce)

N ,s,.

+0D

Experimental

C\T+ d)

Al

Fe

re

Ti

Figure 9. Recalculated (Left Panel),

Composite (c)and Experimental(d)

Equilibrium in the Aluminum-Titanium-Iron System at 1273K 24 . conversion and storage, and fuel for transportation systems. The hydriding characteristics of the FeTi intermetallic can be manipulated by addition of a third element. Replacement of iron by manganese improves the stability of the compound hydride, making it more suitable for peak-shaving applications. It is important to have single phase FeTi. Excess iron yields ru2ri Lave's phase, which does not hydride. Excess titanium leads to solid solution phase, whose hydride i~r so stable that the hydrogen is not recoverable. The presence of either reduces the hydriding capacity and A iron-titanium-manganese V / /i =:'Cx_ efficiency, -el of the alloy. Knowledge of the ternary .. c l :*2.~..T.Icr "eA phase diagram is necessary for producing single phase material. In addition, the alloy cost is high, directly melted from sponge titaniium and scrap iron, the cost is estimated to be from $2.SO-3.OO per lb. The compound could be produced by direct reduction from ilmenite (an ore which cont~'ins Fe and Ti

,V/i

__

41J%"

70700C .0.
680 ~LL

I,

0.60
+Ai

Lt9
60

66 6
cc4 hi

c040
1

/
0

A
+*'

60 Q-20

Al+

L.AIG

L.-.60

5aL-

600Sao808 0 1 2 3 4 5 6 7 0 1 02 03 04 T i.oa--% 05 06

% S . a .-

Partial vertical section at 0.05%TI for the calculated N-Ti-S system IL-liquid. 0-TiA,).

Partial vertical section at 0.50%S! for the calculated Al-Ti-Si system (L-liquid, 8-TiAI,).

25 Figure 10. Calculated Vertical Sections in the Al-Ti-Si System

in about equal proportions) at a lesser cost.

Presently only aluminothermic

reduction is thought to offer cost reductions below $2.00 per..lb. However, this process will lead to aluminum in the alloy. Again single phase material is desirable, and a knowledge of the iron-titanium-aluminum phase diagram is essential. To provide such information, synthesis of the manganese-titaniumiron and aluminum-titanium-ircn systems were carried out by using the analytical descriptions developed for the Fe-Ti, Fe-Mn, Ti-Mn,Al-Ti and PeAl binary systems and combining these to generate ternary sections

24 . To provide

additional experimental information a study of selected portions of the Mn-Ti-Fe and Al-Ti-Fe system were carried out at I000C(1273K). These sections are

shown in Figures 8 and 9 with a recalculation of the Mn-Ti-Fe and Al-Ti-Fe systems to provide closer agreement with the experimental results S. Grain Refinement of Aluminum. Grain refinement of Al by Ti increases

markedly when the Ti concentration exceeds 0.08 a/o resulting from the onset of a peritectic reaction. Evaluation of the role of Si additives has been carried out by Youdelis 2 S by constructing the vertical sections shown in Figure 10 from computed Al-Ti-Si isothermal sections generated between 700 and 1000K from binary descriptions. These results show that silicon additions tend to move the liquidus line defining the two phase (L+O) field to lower Ti concentrations thus lowering the critical Ti concentration (for the onset of the peritectic) to lower levels. This calculated result agrees with the

observation that enhanced grain refinement is observed at silicon leve!s as low as 0.05 a/o.

-72-

____ ___ ___

___ ____

____ ___

____ ___

____

__

___

____

___

____

___

.3

~~~~.~~

-.

__

__

T.,L

Tijt

I.

aL

...

TL;-.. Tiat
I

.- cT

T
T.) * c

T
aT

TL To L .I

Tot-

T*

T
Ta) compound~

aT

"tigure11. Different types of three-phase equilibrium between u,S and L phases 2 6 to . the position of the intersection point of the in relationship (xoT ) curves 6. M~etatectic Reactions in Hafnium. Alloys. of alloys. The netatectic reaction in which

Oo+L,on cooling has a deleterious effect on the high temperature properties This transformation can cause excessive segregation during solidification, welding or heat treatment.. Moreover, when there is limited

between trnfrmto and tepaeso foration Diffeen typfesrfheepasnqulbru melin results or peritectic phases, a betw;een low temperature solubility pito ction h of eutectic the interse n elmentondhito the sinrynrbtion thesol alloy for high of Ce at parametecurve lower temperatures , which reduces the capaebility Sii
temperature service. ! Pieraggi and Dabosi
6

have used a regular solution model

to predict the occurence of this reaction in hafnium base alloys as illustrated for metatectic tendency eatonin3hc related they have 11. In particular in Figure 6.MeatctcRecton n afiu Alys Te the eatctc

delterous ffet onthehightemeratre ropetie ~-.uL oncooing as of aloy. Tis tansormtioncancaue exessve egreatin drin

7. ilardenability of Alloy Steels.

Kirkaldy, Thomson and Baganis 2 7 have

coupled thermochemical and kinetic data for alloy steels to develop an efficient method for computing 12 and 13. CCT and TTT diagrams illustrated in Figures This has been carried out by using free energy changes and activity expansion of the activity coefficient. The latter

data for iron and its binary and ternary alloyswhich are used to evaluate the general linear series (agner)

have been employed in turn for the accurate thermodynamic determination of the liquidus line and the y field in steels with additions of Mn, Si, Ni, Cr, Mo, Cu, V, Nb, 1 and Co. A computer program based on several derived analytic Indeed, the predictions.

formulae is accurate for total additions up to 5 wt.%. compendia.

compare favorably with observations on over 200 steels from international The predicted Ae3 temperatures and diffusion data have been incorporated into a Zener-type expression for the transformation kinetics of ferrite and a semi-empirical formula has been optimized to adequately predict the ferrite start line of the TMT diagrams of low alloy steels. The TIT Compari-

diagrams are converted in turn to CCT diagrams via the Avrami rule.

sons of the predictions with the CCT curves for a representative set of 27 consercial steels show good agreement 8. Growth of HEM Silicon Solar Cell Material. The heat exchanger method (HEM)

developed to grow large sapphire crystals (25 cm diam. x 25 cm. high) has been applied to the growth of silicon solar cell materials. In this method the

seed is placed at the bottom of the crucible and the temperature in the melt increases upwards. This suppresses convection that causes temperature and concentration fluctuations at the solid-liquid interface. Early experiments indicated that SIC particles were found in crystals solidified by the HEM. However, even with the presence of SiC particles,

large grains have been grown with limited interface breakdown during solidification. This observation differs from Czochralski (CZ) growth where interface breakdown due to SiC is followed by twin/polycrystalline growth. The basic elements of the REM and the CZ growth furnaces and the processes are quite similar (heaters, crucibles, insulation,etc.). the furnace. A silica crucible loaded with the charge and set in a graphite retainer is placed in The chamber is evacuated, and after melting the charge crystal In the HEM process, the chamber is typically evacuated For CZ growth an argon blanket is used and the In early experiments with

growth is achieved.

during growth to 0.1 Torr.

chamber pressure can vary from 10 Torr to 1 atm.

HEM growth it was quite surprising that high carbon concentrations were found in the silicon ingots. A thErmochemical analysis was conducted to assess the source of the problem2 8 . This analysis, illustratnd in Figure 14,was subsequently confirmed experimentally. Analysis of the equilibrium between C, ZO, CO2 , Si, SiO and SiO, over the temperature and pressuire

I2

range of interest shows

~-74-

_1Z_1 -Il

Soo 700

eC

PREOeCTED

AE3
SA"T

1763-c)

EXP

LXP I%
PREO. O.1".

Soo 500 '

.I

400

J01MNY D'PTH
I *
I S

1
S

2 80 (SEC)

8 100

82

16tf* 1000

0.5

5 TIME

Figure 12. Comparison of predicted and experimental CCT start 27 or steel type 5140 (0.1-0 and 1%) curves
1I

goo PREDICTED AE3 (761 'C

700

600

S00

400

E.

100

IO00

TIME (SEC)

'Figure 13. Comparison of predicted and experimental TTT start

(0.1%) curvcs
-

for steel type 514027

-75-

+,+++.+:-+,

- I+--.

.,
,.,

++ -,
-'

:.

.
.

" --'-- --++,,;- ,,.

7-M+-1 0. F*+.+. . .

+++.-, "'~~

+-+

++

+ +

'+...Q" *"++ ,.. - +" .. .. +:'

.. .

/,*

_+++

..

..

..

.. . .

. ..
.

'++

++*+ +
.. ,

. II I

..
. .

V- .. ... j:P+.. + ..

-I<

q ,". +g!.iiS'

Jl I :7-1

+.,.,,+. '!a

++ ++ . . ... :-- ... +' _+jZ +'++.'+'+ +++. , ,.r.

. .++ ...

.......................

1735

168 D

S~mp.

173A65 t

-P.

C"S10s=SiO+CO , 10.

3C+SiO
STABLE 9- -2 SiO 0

api :

9,

SIC+CO STABLE.

STABLE

Si+SiO2 STABLE

Ix
,1

'

1aSiO+CO STABLE
C+S1O

ML0o1

STALE WORKING RANGE 3C+SiO=SiC+2C0

# .4 $ I I a . I

WORKING RANGE
.. L..._.L. 4 1 I 0 6 -_ t 7 "t I a I

1. Corrutated presure-terperture relotian for reoct;-fl . 2. Equ;;lr;', v,;or pressure of gaseous SO as a funct;on 9181hMte 8a.a silica ?o rom silicon carbide, silicon montoi;d on, of temperature for the leactiona Si (condensed) + Sa , (condensed) garo, onoxid,. - 25;0 (gas).

p p

p "':p|p

'

i p

100

100

10.

CO+3S0 3 S;O 0" CO+ _ STABLE

~~SiC+ 2 SiO2
STABLE

. -

3 SABESAL Si+2 CO STABLE

"

2 SiC+SO 2 STABLE

a-

ia-OOI

-0.5

WORKING RANGE

Z NRANGE

WORKING

a 4

a a

a a (leoriq

tI 7

I P S

*.

j 4

t 5 a (lo'rigK

I 7

3. Eqalabrium total pressure of CO end S; as a funct;s, of 4. [quilib4um pressure of CO as a fau-cton cf temperature ltmproture for the reaction CO + 3SsO -* SiC + 2S;O'j. for lte rcacion 2C0 + 3Si -* 2SiC + SXO 2.

Figure 14.

Calculated Pressure-Temperature Fields of Stability for SiC Stability During HEM Growth of Silicon Solar Cell Materials in the Presence of Graphite 2 8 .

-76-

10,.-

that CO is present in the

a -

54.11, iso) M-Ur A. ep Aew~* --

.4
11761of

working range. The reaction ctrbon monoxide with sili-

goo Iso
AD
'.tion 0

con to yield silicon carbide and silicon monoxide is the source of the silicon carbide. 0 One scnurce of CO is the reacof Sio 2 crucible with the graphite retainer. Re09 ID

01

02

03 co ad

04

05 Idel

06

07

0s

duction of the CO, effected

switching to a Mo retainer

~by Lwd
o"

colt simple itig.lar liquid and iealo sold soltio~n (a.UO a

reduced the carbon concentrain the silicon2

~ak

AS-,od.4tio

Figure 15. Calculated Pseudobinp y system (G lb -X(lb 30tors.

9. Multicornponent SenticonducSimple solution models.

generally based on regular solutions are applied successfully to compute the phase diagrams for alloy semiconductors, I11-V compounds and a variety of multi-component systems which
AM 1 20 30 40 50 so

are important for clectro-optical

0%
A~o -"

.06,1
% L700

aa~d light emitting diode

applications 1 . The Yocation of the liquidus phase boundaries is critical to successful growth of such compositions from th, melt29 Such calculation als provide insight into the distribution of impur~ities likely to
be developed in a crystal grown

0
150 V10L

L
-13C

126

i~r I1* t-~~.e

tiltill______ 11 F l

from the melt. Recently, Linnebacgohas considered the

effect of using a partially .associated solution (PAS) model in defining the solutions in the pseudobinary system GaSb-AlSb.
900

10----111 1,,,

itP as
ING6 8I.

$o
?00

The results of these computations are shown in Figure 15. lu SufdMotnSlt xd

It.SlieMle.ilOie

20
F

30
W
%S

40

50

and Ceramic Systems. During the~ past ten years, substantial progress has been made in . .vcloping data bases, models

Figure 16. The calculated Fe-S diagram with isobar values of S W)tip to 1 atm 1 2

-77-

9'E

KCI

and computational methods for calculating the phase diagrams and th'ermochemical properties of metallic systems. The development of similar data bases and models for dealing with nonmetallic systems bas

proceeded at a .lower pace. Neverde less, measurable advances have been made and it is likely that the next few years will see even more rapid development of this non-metallic field than the "metallic" field. This is due to the fact

CoC12

SrC.

Figure 17. Calculated Liquidus Contours in the KCSrC12-CaCI2 System 35 .

that the "non-netallic" group covers a very broad spectrum of materials where development of accessible data bases and models for estimating properties can be quickly translated into new solutions for important industrial problems. Indeed the development of data bases for the "non-metallic" systems has begun and prior CALPRAD symposia 1 -5 as well as current publications31 "40 provide a measure of this development. Figure 16 shows a recent example of a computed
32 Fe-S phase diagram developed by Sharma and Chang31 . Chang and Hu have been developing a description of MetAl-Metal-Oxygen and Metal-Metal-Sulphur systems covering associated LiCI solution, defect thermodynamic

and'Wagnervmodels for dealing with these "mixed" metal-nonuetal systems. Similar studies have been carried out by . 33 Schwerdtfeger and Pitsch
Pelton, Bale and ".

"ers -37have been ..oping data bases cover-

'i

ing molten salt and oxide systems. Figures 17 and 18 show some of their recent results systems. for fused salt The thernodynamic properties of all phases in the fifteen binary systems NOCI KCI Figure 18. Calculated Liquidus Contours in the LiCl-KCI-NaCI System 3s.

-78-

-'ILI

can be expressed as differences of single-ion site preference energies. The site preference
"u-

energies and redox energies, used to give the best agreement with the high temperature phase diagra s are in reasonable agreement with values predicted by earlier analyses of thermo1

,d...

(0

O 0
' Z

'"r*.

.t.

.0.

dynamic and spectral data as

well as with calculations from crystal field theury. In the case of the Fe 304 spinel 3 04 -Co phase, it was necessary to propose a small additional positive regular solution term
A.40..

to obtain quantitative agreement between the measured and calculated phase diagrams at
tLqU P ",s .P *28u

"'"P"
P

all compositions. In the other systems, the simple model with no additional terms was found - 36 to be sufficient .40.40.4

':

SP L.
X I\
St M 286Pl.429 A ],

SP

Additional efforts3-4
have been directed toward development of data bases
SP

P'/ S . -w-

cap4ble of computing ternary

and higher order phase diagrams

from the thermochemical properties of.oxide, nitride and mixed systems which have direct Figure 20. application in slag-metal equilibria, high temperature refractories and (SiO 2 -A 203 -. SiN Cr2O,-A120, system. .ne
4 -AlN)

,o.,Oo. Calculated Mg.

5O 0 . 5 -Cr0.400 .6

AI0.400. 6 System

38

synthesis of SIALON ceramics.

Figure

20 shows an example of calculated and observed isothermal sections in the MgOof the chief advantages which this effort will provide is the ability to generate metastable equilibria data which can provide infor-

38

mation on supercooled liquids and the nature of the phase diagram when one or more of the conkpeting phases is suppressed. Such information is of great practical value when dealing with oxide ceramics and glass forming systems.

-.79-

.. .++

..

.'.

..

'.

-"r

+t'

. -

.. ... . ..

,-

,t

$77-7777

7-D*

#F0-COO SOLID SO.OSOM

PC. IDS

L I

"ON
*0.60 0

LOGIO(P02) * CO0 ANMt *

SOM40A * 303 -- > 0100(i 3S I350.000 13.,000 2" 000000500555400S .22s.000 12n0000 $1 00< 3223.000 ItS,.000 %t3,000
*O 0.30 0.20 0.30 0.45 0 .20 i 0. " 0.70 70 0.90 0s 0.90 DO 3.00 a 00

$ SS S $SSSS$SSMSSS 5555$ ....... . ..S

V00,VV0,o
S

gU WW

UZVi 44

10.000
t00.000 o0.000 aSSS 925.000

$IS$55 05S 5 s.... *5S$S$SSSSSSSS S 5 5ssss0050 0SSS$$ISS;;$$SS 'SS$ 5 5S5 SSSS 5S0502 .1SSSSSM SS .11SSSSSS SSSSSSS $3S05$ NSSS SS0.4 6 SsS0S00.... $. S SSSS$S$50 -S $ $SS s5 S S S 6 .;!5 SSSSSS S5055 0 0sS 5S 0

s S S $ SS

* 1 lltlul w u
000 0

SS

SSSSS0S

SSSSSSiSS0SS0 S

0SSS

UVWW_
. S
0SSSS

9.0*0 00.000 900 *0 *25.0 cooS 00 000 7a20.000

S ...... ... 500055 Os$00o000;00000ss-.s-ssss S0SSS5S03 SS S SSSSSSS$55 00SSSSSSSSSSSSS SSSS$ 0SSSS0S0SSSSSS A$ 0SSSSS $SSSS SSSSSSSSSS

S00000,00000 OO $050055 OOISOO00.OOOs

SSSSSSSS S SSSS SSSSSSSSSS S S %ss0s303' 5S005 001 sOsl0303501 0500055 000003000$00st'

000sss 05500000sssssSS0SS5 136 3 Figure 19. Calculated Phase Diagram for the Fe-Co-O System in Air

Points

are Experimental. formed among the six salts LiCi, NaC1, KC1, MgCl 2, CaCl2 , and SrCl 2 were obtained and expressed in mathematical form by a critical analysis of the measured binary phase diagrams and of published binary thermodynamic data. Interpolation precedures were then used to estimate the thermodynamic properties of the twenty ternary systems formed among the six salts, and all twenty ternary phase diagrams were calculated using a computer program designed for the purpose. Good agreement with the measured phase diagrams for the eight 35 ternary systems for which measured diagrams are available was obtained Synthesis of thermochemical properties and phase diagram data has also been performed 36 ,37 for the Fe-Cr-O, Fe-Ni-O, Cr-Ni-O, Mn 0-Co0, 34C34.' Fe04 -11n P 4 and Fe 304-Co 3 04 systems. Figure 19 shows an example of a section in the latter system computed at an oxygen pressure corresponding to air at one atmosphere. In the former study 3 5 high temperature phase equilibrium data The thermodynamics of the spinel phases was determined was compiled and assessed to evaluate the thermodynamics and structural models of the spinel phases. in terms of the distribution of the cations between the tetrah)dral and octahedral sites. The phase diagram analysis approach to examining this cation

distribution has the advantage that a true high temperature equilibrium is being examined. The high temperature phase diagram data was shown to be consistent

with the assumption that the equilibrium constants fox the exchange and redox reactions are independent of composition and that the free energies of thef three site exchange reactions, but not of the redox reaction, are independent of temperature. Furthermore, the free energies of the site exchange reactions

-80-

. ,.. ++::.,.

11 R
, , . .. .. . ..
*,-2, o..
-+

....
. -

+- -+ + +:+ ,

Conclusion The foregoing set of examples show how far the utilization of data bases and computer methods for solving industrial problems has progressed in just the past five years. Not too long ago it was a difficult task to consider computing a ternary system. Moreover, there were very few, systems for which sufficient data existed which would permit a realistic computation to be carried out. By contrast, the current situation is such that computation techniqu s for routine calculation of ternary systems Cat reasonable cost) are available and are being applied to a very wide spectrum of real problems. Higher order systems are also being computed where necessary. Moreover, data bases covering a wide range of materials systems are developing. The community of effort which has achieved this progress nust be continued and attention should be paid to educating the technical and scientific communizy at large to the efficacy of this development.

-81-,

References 1. 2. 3. 4. S. 6~ 7. 8. 9. 10. .11. 12. 13. 14. 15. 16. 17. 18. L. Kaufman, "Proccedings of the Fourth CALPIL:W Meeting", A-ugust 1975
CALPILND 1977,1,7.

19. 20. 21. 22. 23. 24. 2S. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43.

W. Pitsch and G. Inden "Project IMeeting, CALPILND V 1 MPI -Dusseldorf, Wfest Germany, June 1976. L. Kaufnan, G. McHugh, A.P. Miodo%,nik and N.H. Rand, "Sunary of the Proceedings of the Sixth CALPa*XD Meeting", Harwell-Oxford England, April 1977, CALP!. 1977, 1, 293. G. Petzow, I 'h. Hoenig and H. Lukas, "Project IMeeting CALPILAD VII"I, MPIStuttgart, hest Cernany, April 1978, Stummary-P.J. Spencer and T. Chart, CALPHAD, 1978, 2, 197. B. Lihrenius and N1. Hillert, CALPHL\D VIII, Royal Institute of TechnolkcgyStockholm, Sweden, May 1979, Stux-ary-1,. Kaufman, CALPHAD, 1979, 3, 159. L. Kaufman and If. Nesor, Zeit. f. MetallkUnde , 1973, 64, 249. L. Kaufman and II. Nesor, Annual Revievo;of Materials Science, 1973, 3, 1. L. Kaufman and H. Nesor, Met. Tr., 1974, 5, 1617, 1623, 1975, 6A, 2115, 2123. 1. Ansara, International Metals Review, 1979, No. 1, p.2 0 , (Review No. 238). L. Kaufman I~fW Nesor, Conference on In-Situ Composites-lI, M.R. Jackson et al Editors, Xerox Publishing, Lexington, Mass., 1976, p.11 . M.K. Thomas, Ibid p.37. L. Kaufman, CALPHAD 1978, 2, 55, 81, 117, 295, 325; CALPIIAD, 1979, 3, 45. 20 C. Allibert, C. Bernard, I.D. Nussler and P.J. Spencer, reference 5, P. . T. Chart, F. Putland and A. Dinsdale, reference 5, P.1 8 . T. Chart and F. Putland, CALPHAD, 1979, 3, 9. B. Sundman, reference 5, p.102. T. Nishazawa, N. Hasebe and N. Ko, reference S, p.1 13 . M. Hasebe and T. Nishazawa, "Application of Phase Diagrams in Metallurgy and Ceramics", NBS Special Publication 496-Symnposium of January 1977, 1978, 2 , 911. L. Kaufman, J.S. Uatkin, J.H. Gittus and A.P. Niodownik, CALPHAD, 1977, .3, 281. 3.5. Ilatkin, reference 5, p.161. E.S. Machlin and J. Shao, Net. Trans., 1978, 9A, 561. M.L. Sahoungi and C.C. Hsu,_ECA%1Tj4D,_1977, 1, 237. C.C. Hsu 3nd M.L. Saboungi, Zeit.f. M.eta11kunde, 1978, 69, 581. D.* Dew-Hughes and L. Kauifman_, CALPILAD175- 199, 180IS. WV.V. Youdelis, Metal Science, 1978, (Auvust),_p.363. -Dabosi,J. 'atrials Science, 1979, 14, 416.1 B. Pieragg~i and F. 3.5. Kirlkaldy, B.A. Tfiompson, and F.A. Baganis, "iardenability Concepts with Applications in Steel", A.I.M.E., 1978. F. Schmid, C.P. Khattuk, T.G. Diggs and L. Kaufman, J. Flectrochem. Soc., 1979, 126, 935. M. Illegems, reference 4, p.60. R. Linneback, reference 4, p.61 and reference 5, P.2 7 . R.C. Sharma and Y.A. Chang, reference 4, p.17 8. Y.A. Chang and D.C. Hu,reference 4, P.17. K. Schwerdtfoger and If.Pitsch, reference 2, pIT. 2-1. m.L. Saboungi, Perisei and 1-. Blander, reference 4, p.1641. P.L. Lin, A.D. Pelton and C.11. Bale, 3. Ceramics Soc., 1979, (In Press). A.D. Pelton, Hf. Schmalzried and 3. Sticher, Ber. Bunsqcag es. Phys. Chem. 1979, 83, 241. A.D. Pelton, If.Schmalzried and 3. Sticher, .1. Phys. Chem. Solids, 1979, (In Press). L. Kaufman and H. Nesor, CALPILID, 1978, 2, 35. 6. Kaestle and K. Koch, reference 4, p6 L. Kaufman, CALPILAD, 1979, 3, 27. L.J. Gaukler, II.L. Lukas, ETTh. Ifen-g and G. Petzow, CALPIUND, 1978, 2, 349. L. Kaufman, reference 5, p.11 . P. Darner, L.J. Gaukler, If. Krieg, H.L. Lukas, G;.Petzow and J. Weiss, reference S, p.12.

t.

-J

VI. Thermochemistry of Doped Germanium Borosilicate Glasses for Communications Fibers Recent efforts at RADC/ET-Hanscom Field, Massachusetts have led to the development of the only fiber drawing tower in the DoD capable of producing multikilometer lengths of jacketed fiber by a process which involves drawing the 150 micron diameter fibers from a glass preform which has been passed through an annular hot zone. The preforms from which fibers are drawn are prepared by the CVD process such that the layers (and the distribution of elements providing a refractive index variation) can be deposited in a controlled manner. This insures light guiding of the communications signal. One current objective is improvement of the induced transmission loss due to exposure to ionizing irradiation. At present the evidence in hand indicates that doping the preforms with multivalent ions and/or ions providing deep energy band levels in the silicate glass may provide the best route to radiation resistance improvement. The ratio of oxidation/reduction levels of the various dopant ions such as Ce+ /Ce+ 4 or Sb+ 3 /Sb+ s as a function of oxidation/ reducing atmosphere has proven to be critical in some of the glasses. The preforms are fabricated in a thin-walled fused silica reaction tube which will serve as a cladding material for the fiber which is supported on both ends in a glassworking lathe. While rotating, it is heated with an oxygen-hydrogen burner which traverses along its length. Electronic grade

chemicals (SiCI 4, GeCl 4 , POC 3 ) are kept in separate containers

at controlled temperatures and ultra-pure oxygen is bubbled through these liquids. The chemical vapors entering the reaction tube are controlled using a mass flow control system

I~

incorporating of oxygen and a pressurized system, since

a feedback loop to control the desired quantity chloride vapors. Ultra-pure BCI 3 vapors from tank are also monitored and controlled by the B2 03 can be used in chemical vapor deposition as an

3i

-83-

__ _ __ _ __

intermediate cladding or core component. The mixture of oxygen and reactants is passed through the fused quartz tube continuously, the heat for the reaction being provided by the traversing burner. The temperature of the tube (1600-1650C) is continuously monitored and controlled using an optical pyrometer. In the reaction zone, both the heterogeneous reaction at the tube
wall and a homogeneous gas phase reaction take place. The homogeneous reaction produces fine solid patticles of oxide glass product, some of which are driven to the tube wall. At the high temperature employed, they are fused to a bubble-free glassy film as the burner passes along the tube. After the desired number of passes has been made, tne burner temperature is increased in excess of 18001C, and the tube is collapsed to form a waveguide preform. Since the CVD process is largely empirical, ManLabs, Inc. has initiated application of the computer based thermochemical and data bank analysis of the CVD problem in support of the

RADC effort. The most expeditious route to provide this support is to utilize the methods described in Sections II-V %;ith specific consideration of the phase diagrams composed of B203 -SiO 2 -GeO and additions of P2 05 , Sb205 and CeO 2 .
The current effort will attempt to generate state diagrams for the glass CVD process which can be applied to

a variety of materials of current interest in optical wave guides.

These state diagrams should specify the temperature-pressureconcentration ranges where specific products can be formed starting with known reactants. These systems should be representative of those of current interest in the doped silicate glass preform technology. Thus, the best state diagram information

may consist of free energy-temperature-oxygen (or chlorine)

pressure information which is capable of being combined to deal

with mixtures or solutions of components in the gaseous or condensed state. The particular glass systems of interest will cover

It

-84-

the GeO 2 -SiO 2 - B20 3 system with the following dopants: P20 5, Sb2 05 Sb203, Ce2 03 and CeO 2. The current study will examine the CVD process as it relates to fabrication of doped germaniumborosilicate glass preforms used for optical communication fibers with special emphasis on the current RADC/ET mixing chamber design employed to investigate introduction of a variety of dopants. In this configuration the dopants are introduced as chloride compounds which are decomposed and oxidized to be finally deposited as oxides at 1550 to 1650 0 C. The decomposition and reformulation steps will be treated separately and then combined so as to simulate the oxide configuration of the final preform. Special attention will be given to the oxidation state of the dopants in the germanium-borosilicate glass. In order to illustrate how the current technique can be applied to generate such info .tion it is instructive to consider the following examples: a. Utilization of the ManLabs-NPL Data Bank to Define Vapor Pressure of Dopants Thermochemical data on over 2000 inorganic compounds are held in the data bank. Many of the halides of the dopants ef interest are included. The chlorides of antimony,SbCl 5 and SbCl 3 in the gaseous and condensed state,are included. Figure 1 and Table 1 illustrate how the bank can be employed to compute the vapor pressure of SbCl 5 over solid SbCl 5 as a function of temperature. Table 1 shows the reaction -Cl5SB + Cl5SB(G), which is the computer language for the vaporization of solid SbCl 5 displayed in Figure 1. Entry of this notation in the ManLabs-NPL system causes the computer to output the specific heat, entropy, enthalpy and free energy change for the reaction as a function of temperature as shown in Table 1. In addition LOG1OKP is tabulated. The latter can be employed to compute the vapor pressure of SbCl 5 since K p
=

P(SbCls)(atm) = P(SbCls)(torr)/760

(1)

-85-

1000

100 w0
)T-K U)
tt

10

300

350

400 T(*K)

450

TEMPERATURE Figure 1. Calculated Vapor Pressure-Temperature Relations for SbCl5

TABLE I DATA BANK OUTPUT FOR VAPORIZATION OF SbCl 5 7"9 0126

THERrIOY'rAr IC FUrICT I OISl:

-CLV.+CL5 : B.: G
T K DEL'TrCP CRL/K P1OL DELTAS C:AL/K r'lOL 21. DELTAH", CAL/NEIL 11900 11;877 11531 11300 110.30 10771 DEL TiG CAL/rIOL 3850 3.00 :3139 2500 1E1., I1 12o 94 LO31 Off P -2.. -,. 76,113 -2.111 -1. 56! -1. 0196 -0.6149 0.l -3

29 E '. 15
300.00 35.00 :350. 00 375. Oc 4 000 4,25. 00

-12. 190
-11.528 -11. 000 -10.572 -1 0. 21 -9.-.

-12. 245

2.. 924

000

25. 976 25.142 24.3.:8. 2,3.727 117

101

-86-

Nn:I

The vapor pressure of SBCI S in torr calculated from Eq.(l) and the data in Table 1 is displayed in Figure 1. This shows that at 350*K(i.e. 77*C) the vapor pressure of SbCI S is 20.8 torr while at 300cK (i.e. 27*C) it is only 1.6 torr which is too low for prac:Lcal use. In fact, the 20.8 torr pressure is used by maintaining the SbCl 5 feed lines at 77'C. b. Calculation of the Relative Proportion of SbCl and SbCl 5 as a Function of Temperature and Pressure The foregoing example illustrates how the aauz bank can perform a simple problem. A second example of a more complex problem can be illustrated by considering the decomposition of SbCl 5 into SbCl 3 and Cl 2 which can be written as SbCl + Cl2 SbCl 5 (2)

31

Table 2 shows the Data Bank results for the forward The fifth column tabulates AG0 for the forward reactionthus at equilibrium reaction. AG=0=AG' + RTln PSbCl 5 /(Pc2) (Psbcl3 Since PC12 = PSbCI 3 = Pt x (where Pt is the total pressure and x is the fraction of SbCl 3 ) and PSbCl 5 Equation(3) can be recast as -AG*=RTln (l-2x)/x2 Pt thus permitting solution of x (and l-2x) as a function of temperature and pressure. Figure 2 displays the results for Pt=1 atm and Pt=0.1 atm (76 torr). The results show that at I atmosphere and 300K the mixture consists largely of SbCl 5 with very little SbCl 3 As the temperature is increased the SbC1 fraction increases substantially. Lowering the pressure to 76 torr has a very marked effect on the ratio. Thus control of the pressure could be an easy way to vary the ratio of
=

(3)

Pt(1-2x).

-87-

iM

i ... .

....

~ :~ :~: ~5 .

. ' .

n!....."...'

'

Fraction SbC1 (Pt=l atm)

Fraction SbCl 5 (Pt=76 torr)

L)

94

0.10/ 0.10-Fraction SbCl 3

k
/
Fraction SbC1 3 CPt=1 atm)

-Pt=76
LU

torr)

~/
z
0

0.01

CI 2 (G)+SbCI 0.001 300 1 I

(G) I 400

SbC1 5 (G) I 500

TEMPERATURE, T0 K Figure 2. Calculated Fraction of SbCI 3 and SbCl 5 as a Function of Temperature and Total Pressure in Chlorination of SbCI 3

TABLE 2 DATA BANK OUTPUT FOR CHLORINATION OF SbCl 3

-CL2 (G.> - .E:CL3 |5) +.SBC:L5 0_.,

T K DELTACP CRL/K MiL DELTA. CAL/K MnL DELTAH CAL/MOL DELT tG CAL/MOL LJi31OKP

298.15 300.00

325. 00 350.00 5 .OO.8.0 :375.O0

0.170 0. 199
0.550 0.830 1.057
.

-35.290 -35.289
-35.259 -3',5.._207 -35.142

-18450 -18450
-16440 -18423 -18:399

-7928 -7863
-6981 -6100 -5221

5.812 5.728

4. 6954 -. 1-.9 :3.:09 3.04:3

400.00
425; 00

1.244
1.4 00

-35. 068
-34.987

-18370
-18337

-4343
-3468

2. 373
1.78:3

500.00

450.010 475.00
525.00 550.00 575.00 600.00

1.532 1.645

1. 742
1.827 1.901 1.967 2. 02P

-34.903 -34.818

-34. 731

-18300 -18261

-18218

-2594 -1722

-853

1.260 0.?92
-. 0106 -0.349 -0.662 -0.948

0.37:3

-34.644 -34.557 -34.471 -34.386

-18174 -18127 -18079 -18621

14 879 1742 2603

-88-

...... .

..

Sb+ 3 /Sb+ S ions entering the reaction chamber. Thus the data bank offers a means for investigating the stability of Sb+3/Sb +5 interchanges in the reaction 6SbCI 5 (G) + SSb 2 0 3 (S) -)-0SbCI 3 (G) + 3Sb 2 O5 (S) as a function of temperature and pressure. (5)

Similar investigation

can be undertaken for chloride and oxides of phosphorous, and cerium in order to evaluate their tendency to yield a predominate valence state in the glass. Although the CVD process is controlled by kinetic factors as well as by thermodynamic factors and the former may prevent attainment of equilibrium, the equilibrium calculations can still serve as a useful guide in outlining the field in which certain species predominate (i.e. Figure 2). Since the dopants finally are deposited in the glass phase it is proposed that the techniques developed in the present program for computing multicomponent oxide phase diagrams be applied in dealing with the condensed state reactions which take place during the preform synthesis at high temperatures. As the first step in this analysis the GeO 2 -SiO 2 , B2 03 -SiO 2 and B2 0 3 -GeO 2 binary systems have been analysed along the lines described in Sections II-IV. The results are shown in Table 3 and Figures 3-5. The binary solution phases in these systems are described by equations similar to Equ'tion 1 on page 27. These results can be employed to compute ternary sections of the B2 03 -GeO 2 -SiO 2 system between 1600K and 1000K which are shown in Ffgures 6 and 7. The latter are computed by employing the des.ription contained in Table 3 and ternary solution phase free energies similar to those depicted by Equation 4 on page 40 with a zero valued ternary interaction parameter for each phase. The current program will next consider effects on this system due to additions of P 2 Os, CeO 2 and Sb2 0 5 .

-89-I

-00 .

T I

TABLE 3 SUMMARY OF LATTICE STABILITY AND SOLUTION PHASE PARAMETERS FOR GERMANIUM BOROSILICATE SYSTEMS (All Units in Joules per gram atom (mole of atoms),T in Kelvins) BO = (1/5)B
20 3

SO = (1/3)SiO 2

GO = (1/3)GeO 2

Q=B 2 0 3 structure, R=low temperature GeO 2 structure, X=Crystobalite, T=Tridymite, H=a quartz, B=O quartz structure, U=low temperature GeO 2 GOGOLH* = (i/3)GeO 2 (Liquid)-(1/3)GeO 2 (a quartz) SOSOLQ= -6.15T SOSOHU=-4791-5.69T GOGOLH= 3891-2.80T GOGOHU= 7443-5.69T GOGORU= 7109-5.40T GOGOLX= -1.67T GOGOLT= -2.01T GOGOLQ= -6.1ST ; ; ; ; ; ; BOBOLQ=4448-6.1ST BOBOLX= -1.67T BOBOLT= -2.01T BOBOLH= -2.80T BOBOLR= -3.10T BOBOLU= -8.49T

LBOSO=64852-41.84T, LSOBO=-4184+20.92T XBOSO=XSOBO=TBOSO=TSOBO= HBOSO=HSOBO=83680 RBOSO=RSOBO=QBOSO=QSOBO=33472 LGOSO=LSOGO=HGOSO=HSOGO=RGOSO=RSOGO=4184+12.552T TGOSO=TSOGC=XGOSO=XSOGO=83680; UGOSO=USOGO=41840 LBOGO=LGOBO=12552, QBOGO=QGOBO=33472, HGOBO=HBOGO=83680 RBOGO=RGOBO=83680

* These differences specify the free energy of one phase (i.e. liquid) minus the free energy of the second phase

(i.e. for a given compound).

"-90-

T*

T",'.

w/o GeO 2 6 13 20 28

i2 37

46

57

70

84

SiO2 L+X L+T

1500 1307 1000 5000

G0=13)e
Figure 3. Calculated GeO 2
-Sio

(1/3)SciO -so
2

Phase Diagram

W/o Sio 2

~2 3 14
T0 K 1500

26

44

49

59

68

77

85

93

5i

L+X L

L+T
10000 5000

Q.R

85

L.4'

233 -91-

j~i

w/o GeO 2 B 203

TOK

22

38

52

63

75

79

85

91

96

L
1500

L+fl
10

13J7
000

Q+U

BO= (1/5) B 0 Figure 5. Calculated B 03 Ge0 2 Phase Diagram

(1/3)Ge02=GO

-92-

M0

c~4J

04 00

44

r4

to)

0-

oo
LIir

44

00

100

040

0-4

r_

'-44

.,4

+
+

+
+.

0
C4O

tol

0 M4 00

>1

~~-94-

J9ij