THE UNIVERSITY OF MICHIGAN INDUSTRY PROGRAM OF THE COLLEGE OF ENGINEERING THE EFFECT OF TEMPERATURE AND OXIDATION ON THE REACTIONS BETWEEN IRON AND ALUMINA-SILICA REFRACTORIES John Edward Brokloff A dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy in the University of Michigan 1964 March, 1964 IP-662

ACKNOWLEDGMENT I wish to express my appreciation to all those individuals who aided in the completion of this investigation. I especially acknowledge the following: Professor Lo H. Van Vlack, Chairman of the Doctoral Committee, for his guidance throughout the course of the investigation. His criticisms and suggestions were most helpful, and his encouragement aided greatly in the completion of the work. Professors R. A. Flinn, W. Co Bigelow, L. 0. Case and Associate Professor R. Do Pehlke for their assistance, interest and cooperation throughout the entire work. The American Foundrymen's Society, International Nickel Company, General Motors Corporation and General Electric Company for the financial support provided me as a graduate student of the University. Professor P. K. Trojan and Professor Do Lo Sponsellor for their helpful suggestions in the construction of the experimental equipmento Miss Ellen L. Philip for typing and assistance in preparation of the final drafto My fellow graduate students for their valuable assistance in running the experimental apparatus. ii

TABLE OF CONTENTS Page ACKNOWLEDGMENTS.............................................. o ii LIST OF TABLES................................................. v LIST OF FIGURES............................................... vi LIST OF APPENDICES.................................... ix ABSTRACT..................................................... x CHAPTER Io INTRODUCTION.........................................o 1 IIo REVIEW OF LITERATUREo..................................o Literature Pertaining to Phase Equilibriao........ 3 Literature Pertaining to Refractory Reactionso...... 12 III. EXPERIMENTAL PROCEDURE................................. 14 Phase Rule Application........................ 16 Experimental Materialso............................ 19 Temperature Control............................... 21 Atmosphere Control............................... 23 Reaction Technique............................. 25 Phase Identification............................... 26 IVo EXPERIMENTAL RESULTSo...........................o 28 Equilibration Studies in the Iron-Oxygen-HydrogenAluminum System................................... 28 Equilibration Studies in the Iron-Oxygen-HydrogenSilicon System..................................... 43 Equilibration Studies in the Iron-Oxygen-HydrogenAluminum-Silicon System............................ 47 V. DISCUSSION OF RESULTSo........................ 60 Iron-Alumina Equilibria............................ 60 Iron-Silica Equilibria............................ 64 Iron-(Alumina+Silica) Equilibria................. o 65 Correlation of Equilibration Results with Large Scale Refractory-Metal Reactions...............o 70 iii

TABLE OF CONTENTS CONT'D Page VI. CONCLUS IONS............................................ 74 APPENDICES I. CALCULATION ON THE UNIVARIANT EQUILIBRIUM IRON-ALUMINAI'IRCYlrTrITE-GAS.................................... 88 II. THE LABORATORY FURNACE.................................. 89 III. APPLICATION OF THE PHASE RULE........................... 90 IV. SAMPLE CALCULATION OF WATER VAPOR:HYDROGEN RATIO FROM OPERATING DATA.......................................... 93 V. NOMENCLATURE............................. 95 REFERENCES................................................... 96

LIST OF TABLES Table Page I Summary of Pillay, Floridis, and Thurner Data.............................................. 10 II Chemical Analysis of Kaolinite.................... 20 III Summary of Equilibration Experiments.............. 76 IV Heat Record for Refractory- Metal Reactions....... 80 V

LIST OF FIGURES Figure Page 1 Flow Diagram of the Experimental Apparatus....... 15 2 Partial Equilibrium Diagram for the Iron-OxygenHydrogen-Aluminum System After Pillay, et al(24).,, 17 5 Photograph showing: (A) Furnace Tube Seal with Experimental Sample Boat, (B) Calibration Thermocouple, (C) NBS Thermocouple, and (D) Gas Sample Tube...................o.............. 22 4 Schematic Drawing of the Furnace Tube Seal.......... 22 5 Microstructure of Sample from Experiment A-45..... 29 6 Microstructure of Sample from Experiment A-27...... 50 7 Microstructure of Sample from Experiment A-28,..... 50 8 Microstructure of Sample from Experiment A-27.... 352 9 Microstructure of Sample from Experiment A-29,..... 52 10 Microstructure of Sample from Experiment A-29..... 55 11 Microstructure of Sample from Experiment A-21...... 3355 12 Microstructure of Sample from Experiment B-40o,.... 54 15 Microstructure of Sample from Experiment B-40......o 34 14 Microstructure of Sample from Experiment C-41..... 54 15 Microstructure of Sample from Experiment D-47..... 56 16 Microstructure of Sample from Experiment D-48.......oo 56 17 Microstructure of Sample from Experiment D-5000... 57 18 Microstructure of Sample from Experiment D-50.......o 57 19 Microstructure of Sample from Experiment D-450.... 59 20 Microstructure of Sample from Experiment D-42.....oo 39 vi

LIST OF FIGURES CONT'D Figure Page 21 Microstructure of Sample from Experiment E-54....... 40 22 Microstructure of Sample from Experiment E-55....... 40 23 Microstructure of Sample from Experiment E-53....... 42 24 Microstructure of Sample from Experiment E-56*....... 42 25 Microstructure of Sample from Experiment A-24....... 44 26 Microstructure of Saw le from Experiment A-25....... 44 27 Microstructure of Sample from Experiment A-22....... 46 28 Microstructure of Sample from Experiment A-27...... 46 29 Microstructure of Sample from Experiment B-4 0....... 46 30 Microstructure of Sample from Experiment C-41....... 48 31 Microstructure of Sample from Experiment D-50....... 48 32 Microstructure of Sample from Experiment D-47....... 48 33 Microstructure of Sample from Experiment D-52....... 49 34 Microstructure of Sample from Experiment A-29....... 51 35 Microstructure of Sample from Experiment B-40....... 51 36 Microstructure of Sample from Experiment D-50....... 52 37 Microstructure of Sample from Experiment E-56...... 52 38 Microstructure of Sample from Experiment D-51....... 54 39 Microstructure of Sample from Experiment D-52...... 54 40 Microstructure of Sample from Experiment D-52....... 55 41 Microstructure of "as received" Mullite............. 55 42 Microstructure of Sample from Experiment D-47....... 57 vii

LIST OF FIGURES CONT'D Figure Page 43 Microstructure of Sample from Experiment A-29...... 57 44 Microstructure of Sample from Experiment B-40...... 58 45 Microstructure of Sample from Experiment D-50o...... 58 46 Equilibrium Diagram for the Iron-Oxygen-HydrogenAluminum System............................... 81 47 Plot of the Logarithm of the Chemical Equilibrium Constant Against Reciprocal Absolute Temperature for the Iron-Oxygen-Hydrogen-Aluminum System........ 82 48 Equilibrium Diagram for the Iron-Oxygen-HydrogenSilicon System After Colligan.(15)................. 83 49 Equilibrium Diagram for the Iron Oxide-A1203-SiO2 System After Schairer and Yagi. (35)................ 84 50 Microstructure of Unattacked Silica Refractory..... 85 51 Microstructure of Silica Refractory from Heat Number F-43........................................ 85 52 Microstructure of Alumina Refractory from Heat Number F-420....................................... 86 553 Microstructure of Fireclay Refractory from Heat Number F-42......................................o 86 54 Microstructure of Fireclay Refractory from Heat Number F-43 0.................................... 86 55 Scale Drawing of the Laboratory Furnace........... 87 viii

LIST OF APPENDICES Appendix Page I CALCULATION ON THE UNIVARIANT EQUILIBRIUM IRON-ALUMINA-HERCYNITE-GAS AT 800~C AND 900 ~C.... 88 II THE LABORATORY FURNACE.......................... 89 III APPLICATION OF THE PHASE RULE.................... 90 IV SAMPLE CALCULATION OF WATER VAPOR: HYDROGEN RATIO FROM OPERATING DATA............................... 93 V NOMENCLATURE...................................... 95 ix

ABSTRACT This investigation is concerned with determining the effect of temperature and atmosphere on the reactions of iron with alumina and/or silica. The specific systems chosen for study included: a) the iron - oxygen - hydrogen - aluminum system b) the iron - oxygen - hydrogen - silicon system c) the iron - oxygen - hydrogen - aluminum - silicon system. The phase rule was used to establish a program of experiments and to interpret the final equilibrated phase assemblageso Microstructures of samples from each system revealed the basic equilibrium situation in terms of the presence or absence of reaction products. The experimental procedure involved equilibration of charges in a resistance tube furnace under a controlled atmosphere of hydrogenwater vapor-argono Charges for systems (a) and (b) were alumina with iron or hercynite, and silica with iron, respectively. Charges for system (c) included iron-mullite (3Al2Q3o2SiO2) and iron-kaolinite (2A12Si205(OH)4). Temperatures were varied from 1450~C to 1700~C, while atmosphere conditions were maintained at levels of pH /pH2 < 0~50. Phase identification was accomplished by examination of polished sections with a reflecting light microscope. X-rays were used to verify the existence of reaction products. The results from equilibration of iron and alumina locate the four phase univariant boundaries in the equilibrium diagram for x

the iron-oxygen-hydrogen-aluminum system. Solid hercynite (FeOA1203) is the product that results from the reaction of iron and alumina at temperatures less than 1700~C. The equation for the free energy of the reaction Fe(5) + A1203 + H20(g) FeOAl203 + H2 is: AF~ = -2100 - lo7T For the reaction Fe(^) + A120l + H20(g) \FeO0AkO0 + H2 (Q) 2 3 b (g) 2 3 2(g) the equation is: AF~ = -5850 + 0O37T Results from iron-silica charges agree with previously published data, thereby confirming the existing equilibrium diagram for the iron-oxygen-hydrogen-silicon system and substantiating the overall experimental techniques of this study. The equilibration of iron-mullite produces corundum and liquid, for the range of temperatures 1450 C to 1700~C and approximate atmosphere conditions of PH 0/PH = 0.22. A liquid phase forms from iron-mullite combinations at lower temperatures and under more reducing conditions than for iron-silica equilibria. The equilibration of ironkaolinite produces mullite and liquid, for the range of temperatures 1450~C to 1535~C and approximate atmosphere conditions of pH20/PH = 0.220 The liquid phase forms more readily compared to results from iron-silica or iron-mullite reactions. xi

Results from equilibration of iron-mullite and iron-kaolinite are interpreted in terms of the primary phase fields of the iron oxide-A1203-SiO2 equilibrium diagram. Finally, the equilibrium results are correlated with large scale refractory-metal reactions involving alumina, silica, and fireclay refractories with molten steel. xii

CHAPTER I INTRODUCTION This investigation is concerned with determining the effect of temperature and atmosphere on the reactions of iron with alumina and/or silicao A study of this nature finds considerable practical application for two different reasons: (a) furnace refractories such as alumina, silica, and fireclay are subject to attack from iron oxide and thereby increase operating costs and cause production difficulties and (b) the reaction products that form can be a source of macroinclusions in steel castingso In view of this, a study concerned with the fundamental equilibrium relationships involving iron, alumina and/or silica can be of significant valueo The specific systems selected for study in this investigation are: (a) the iron-oxygen-hydrogen-aluminum system, (b) the iron-oxygen-hydrogen-silicon system0 (c) the iron-oxygen-hydrogen-aluminum-silicon system0 The system iron-oxygen-hydrogen-aluminum is especially important since present equilibrium relationships are not well defined0 This is particularly true for very high temperatures of approximately 1700~C and at temperatures near the melting point of irono Selection of the system iron-oxygen-hydrogen-silicon affords a good opportunity to check the overall experimental technique. Finally, investigation of the iron-oxygen-hydrogen-aluminum-silicon system is of interest -1

-2because of its direct application to the problem of fluxing action of oxidized iron on fireclay refractories. In this study the phase rule was used to establish a program of experiments and to interpret the final equilibrated phase assemblages., Microstructures of samples from each system revealed the basic equilibrium situation in terms of the presence or absence of reaction productso The equilibrium diagrams for systems (a) and (b) are presented, and results from large scale refractory - met reactions are explained in terms of equilibrium relationships developed for system (c).

CHAPTER II REVIEW OF THE LITERATURE A review of the literature is presented in order to provide a background for the present investigation. For purposes of clarity, the literature survey is divided into the following two categories: (a) Pertinent phase equilibria (b) Practical refractory reactions. Literature Pertaining to Phase Equilibria Fe-O In phase equilibrium studies involving a transition type element like iron, the gas phase plays an important role in the determination of the final iron oxidation state, Muan discussed this fact and reviewed pertinent iron-oxygen data with regard to (a) an experimental technique involving indirect gas systems and (b) a convenient method of presenting equilibrium data through pressure-temperature diagrams. Both (a) and (b) are used in this investigation. Darken and Gurry(2) determined the equilibrium for iron-solid iron oxide in the temperature range of 1038 C to 1365 C by using mixtures of CO-CO20 Analysis of the oxidized product formed at 910 C showed an iron content of 76085 weight percent, corresponding to a formula of Fe 09510o Darken and Gurry(3) also determined the gas compositions in equilibrium with liquid iron oxide and liquid iron at 1600 Co The equilibrium pCO /PCO ratio was 0O187, corresponding to an equivalent PH20/PH2 value of 0O772~ -3

Philips and Muan(4) combined their results with that of Darken and Gurry(3) to show that: (a) hematite (Fe203) is the stable condensed phase in air until 1590~C -a.iL (b)'!,^ne-i-e (e. e' Feo-) exists fio0l 13' O~C to 15i:1~C with liquid the stable phase above 1591~Co Richards and White(5) reported a temperature of transition for Fe203-Fe304 in air of 1385 + 5 C which is in good agreement with that of Greig, et al.() Al-O-Si (7)_ Bowen and Greig's evaluation of the Ao103-SiO2 system enabled them to construct a reasonably complete binary diagram, The existence of sillimanite (A1203Si02) as an equilibrium phase was doubted. Mullite (3A120532SiO2), however, was cited as a true compound with optical and crystallographic properties very similar to those of natural sillimanitee They concluded that mullite melts incongruently at 1810~C. Natural sillimanite was thought to be stable to 1545~C. Budnikov and Mchedlov(8) asserted that mullite was a defect lattice with considerable amounts of impurities. For firing in extremely reducing atmospheres, Barta and Barta(9) reported the formation of corundum from various starting ratios of A1203/SiO2 due to silica vaporization. Evidence for the congruent melting of mullite was also presented. The conclusions of Aramaki and Roy corroborated those of previous authors in that mullite melts congruently at 1850 C,

-5and that equilibrium toward sillimanite was non-existent. The A1203 Si02 binary equilibrium diagram was revised, however, to the extent that mullite solid solution exists from 60 to 63 mole percent alumina, and that a eutectic of mullite (ss) and alumina exists at 1840~C and 67 mole percent alumina! Fe-O-Si Development of phase relationships between iron oxide and silica has progressed since the early work on the system FeO-SiO2 by Bowen and Schairer. (11) A binary diagram was constructed showing the congruent melting of fayalite (2FeO0SiO2) at 1220~C and two (12) liquid formation at high silica levels. Darken and Gurry ) developed phase relations in the FeO-Fe203-SiO2 system~ Muan(13) determined boundary curves for the liquidus surface in the FeO-Fe203-SiO2 system using controlled atmosphere techniquesO The following three ternary invariant points were established at 1140~C, 1150~C, and 1455~C, respectively: (a) fayalite, magnetite, tridymite, liquid, and gas (b) fayalite, magnetite, wustite, liquid, and gas and (c) tridymite, magnetite, hematite, liquidaand gas,. The low temperature for liquid formation (11400 C) was attributed to the influence of the highly reducing condition of po = 10' atmo Darken's(14) work emphasized the role of gas composition in determining the melting temperatures of oxides in the presence of excess silica. The experimental technique involved the passage of CO/C02 gas mixtures over a silica rod coated with hematite in a gradient furnace. With the results obtained and other thermodynamic

-6data, an equilibrium diagram of temperature versus (pCO /Pco) was constructed. Figure 48 which will be discussed later illustrates the equilibrium diagram. The ratio (PCo /PCo) can be converted to p0 by using the equilibrium constant for the reaction: CO +! O2 ~ Co2 (1) The following univariant equilibria were determined: (a) fayalite, solid iron, silica, and gas (b) fayalite, magnetite, silica, and gas and (c) fayalite, melt, silica, and gas. In addition, the position of the univariant equilibrium of melt, solid iron, silica, and gas was estimated. This involved the assumption that the activity of iron oxide in the melt phase is reduced by 1/3 due to the presence of excess silica. At the melting point of iron, 1533~C, the reaction CO +- 2 0 Co2 (2) gives: log -o.)68 ( ) With a 1/3 reduction in the gas mixture ratio, Equation (3) becomes: Pc_2 log (-pC = -o.68 - log 3 = -1.16 (4) The univariant equilibria of melt, solid iron, silica, and gas was determined by this point and the quintuple point of fayalite, melt, solid iron, silica, and gas. Further experimental investigation of the iron-oxygen-carbonsilicon system was carried out by Colligan. Using a molybdenum

-7resistance furnace with gaseous mixtures of CO and C02, Colligan equilibrated vacuum melted iron with pure silica at temperatures of 1250~C, 1400~C, and 1550~C. The results accurately determined the position of the univariant equilibria of melt, solid iron, silica, and gas. A good estimation of the quintuple point of melt, solid iron, liquid iron, silica, and gas was also made. From Colligan's data, Darken's original estimate on the depression of activity of iron oxide in a liquid silicate melt was checked. Actually, the estimate of 1/3 was low. The equilibrium value of log (Pc2 /PCO) at 15330C was -1.42. This indicated that the activity of iron oxide in the melt with excess silica was depressed to 1/5 or 1/6 of that in the absence of silica. From the work it was concluded that an atmosphere must be more reducing than originally estimated to prevent silicate melt formation. Fe-O-Al Hansen and Brownmiller's(6) early work on the Fe203-A1203 binary showed the extent of solid solubility of each compound in the other. Richards and White(5) indicated. however, that the equilibration time of one hour used. by Hansen and. Brownmiller may have been too short, McIntosh, et al.(17) developed the FeO-A1203 binary equilibrium diagram. The existence of the phase 3FeO0A1203 up to 1225~C was reported. Also, the existence of a eutectic of iron oxide (ss) and hercynite (FeO-A120) at 1550~C and subsequent congruent melting of hercynite at 14400C were cited.

-8Recent work by Fisher and Hoffman(18) using both microscopic and x-ray analysis techniques led to a revision of the FeO-A1203 binary, At a temperature of 13350~C the existence of the eutectic of iron oxide (ss) and hercynite was again established. At 1750~C and 65 percent alumina another eutectic of hercynite and alumina (ss) was reported. The hercynite composition varied from 59 weight percent alumina at 1500~C to 64.2 weight percent alumina at 1750~C. Congruent melting of hercynite was established at 1780~C. The work of Muan and Gee(19) on the Fe203-A1203 system showed the importance of an atmosphere in determining the composition of condensed phases. Data obtained at po = 0.21 atmosphere and p = 1 atmosphere was presented on pseudo-binary equilibrium diagrams. The compositions of the condensed phases must be read from an accompanying FeO-Fe203-A1203 ternary diagram. The role of the increased oxygen pressure from 0.21 to 1 atmosphere is to (a) increase the temperature of decomposition of Fe203.A1203(ss), (b) reduce the existing area of spinel (ss), and (c) move the composition of the condensed phases further away from the FeO0Fe203-FeOoAl203 joino Gokcen and Chipman (20) studied the reaction: A12053(s) 2 Al + 3 0 (5) by equilibration of water vapor-hydrogen mixtures with melts of iron in alumina crucibles at temperatures of 16950~C 1760oCcand 18660Co The gas mixtures were such that the water vapor: hydrogen ratio ranged from 0.002 to 0.0135. Subsequent examination revealed that the melts

-9had clean surfaces and were loosely attached to the alumina crucibles. There was no discoloration of the crucibles. No evidence of the reaction product hercynite (FeO-Al203) was reported. Richards and White(5) emphasized the importance of temperature and oxygen partial pressure in determining the various phase assemblages in multicomponent oxide systems containing iron. Through the efforts of Richard and White, the work by Schenk, Franz, and Willeke(21) is applicable to this study. Previously, the three investigators had reduced mixtures of Fe20O-A1203 in a CO-CO2 atmosphere under isothermal conditions. From the results, Richards and White were able to ascertain that a solid iron-alumina-hercynite-gas univariant equilibrium existed at 4.0 percent CO2 and 5.4 percent CO2 for 800~C and 900~C, respectively. Using the equilibrium constant for the reaction CO2 + H2 2 H20 + CO (6) the above data can be converted to a water vapor: hydrogen ratio and related directly to this investigation. The calculation for the data is included in Appendix I. Using platinum electrodes to measure e.m.f. potentials in (22) FeO-A1203 mixtures, Fisher and Hoffman determined the free energy of formation of hercynite (FeO'A1203) as -7500 to -15,000 calories/mole at 15000~C From a graphical integration of calorimetric curves, Fisher and Lorenz(23) determined the heat of reaction for FeO + A1203 FeOA1203 (7)

-10at 820~C, 950~C, and 960~C. The average value was reported as AH = -14,600 calories/mole + 25 percentO Pillay, et al (24) determined experimentally the gas ratio pH O/PH in equilibrium with liquid iron, alumina, and hercynite at 17000~C Also, data obtained by Floridis and Chipman(25) in a study of oxygen activity in liquid iron alloys were incorporated into the investigation. The data were obtained by visual check of the appearance of remnants of the alumina crucibles. For certain oxygen concentrations in the metal, the interior surfaces of the crucibles were black. The blackened material was identified by x-rays as hercynite. Finally, some of the data of Thurner(26) in his study of sulfuroxygen equilibrium at 1586 ~C was used. The data are summarized in the following table. TABLE I SUMMARY OF PILIAY, FLORIDIS, AND THURNER DATA PH20/PH2 Tempo (~C) REDUCING OXIDIZING Investigator (No hercynite) (hercynite) 1700~C 0.31 0.31 Pillay 1600~C o0253 0.279 Floridis 1550 C 0.274 0.288 Floridis 1586 ~C - 0305 Thurner Pillay used the above data to plot log(p20/pH2) against 1/T as a straight line with the equation: log ( ) = - T +0.07 (8)

-11From this equation the free energy of formation of hercynite for the reaction A1203 + Fe(B) + H20 < FeOA1203 + H2 (9) was determined as: AF~ = -5260 + 0.532To (10) Pehlke(27) in his work on the solubility of nitrogen in liquid iron alloys obtained data pertinent to the iron-oxygen-aluminum system, For high oxygen concentrations in the Sievert's apparatus, a reaction occurred between the liquid iron and the alumina crucible. The product, hercynite, formed according to the reaction: Fe + 0 + A1203 -Fe0A1203 (11) Analysis of eight heats gave a mean value of the activity of oxygen in equilibrium with hercynite at 1606~C of o.o635 This result was combined with data of Taylor and Chipman (28) and Floridis (25) to obtain a free energy of formation of hercynite from constituent oxides of AF166oC = -4,290 calories/gm-moleo Fe-0-Al-Si (29) Snow and McCaughey(29) studied the system FeO-A1205-SiO2 using mixtures containing up to 60 percent iron oxide and 55 percent alumina. The boundary curves on the liquidus surface of the ternary were reasonably well established in the low alumina region. Three quintuple points were established but the existence of the phase iron cordierite went undetected.

-12Schairer and Yagi(30) studied the system iron oxide-A1203-Si02 under strongly reducing conditions (p0 = 10-13 atmo). The phase iron cordierite was shown to exist in a very limited region. Muan(3) determined the boundary curves of the liquidus surface for the system iron oxide-A1203-Si02 in air. Comparison of this diagram with that for the highly reducing condition reveals the absence of the primary phases wustite, fayalite, and iron cordieriteo Literature Pertaining to Refractory Reactions Iron Oxide Attack on A1203-SiO2 Refractories Snow and McCaughey(29) provided an explanation for the formation of corundum during iron oxide attack of fireclay brick. Muan's(31) comparison of iron oxide attack on al-unina-silica refractories in air and under strongly reducing conditions (P0 = 10-1 atm.) involved the following: (a) In air the absorption of iron oxide by a brick of approximately 60 weight percent alumina and 40 weight percent silica results in liquid formation at temperatures decreasing from 1595~C to 1380~Co Increased absorption of the iron oxide does not lower the temperature for liquid formation but does increase the amount of liquid phaseo (b) For strongly reducing conditions, sufficient absorption of iron oxide by alumina-silica brick can result in liquid formation at temperatures as low as 1088~C. Kaolinite Decomposition Considerable literature exists on the problem of kaolinite decomposition. Brindley and Nakahira(32) discussed previous work and

-13introduced new and pertinent information on the problem. Initially, the authors outlined the series of reactions that occur in kaolinite decompositiono These include: (a) an endothermic reaction at 500~C resulting in the loss of structural water and the formation of metakaolin, (b) an exothermic reaction at 925~C resulting in the formation of gamma alumina and mullite with accompanying crystallization of cristobalite, and (c) an exothermic reaction at 1200~C which results in the disappearance of gamma alumina. Using an experimental technique of x-ray analysis of macrocrystalline kaoliniteg Brindley and Nakahira (3234) were able to determine that: (a) meta-kaolin was a layered structure with two dimensional regularity, (b) a cubic phase, spinel type structure formed from the meta-kaolin during the second stage of decomposition, and (c) the second stage of decomposition was completed with the formation of mullite at 1050~Co (35) Finally, Comer( showed through electron microscopy that the orientation of the newly formed mullite was related to that of the original kaolinite crystal0

CHAPTER III EXPERIMENTAL PROCEDURE An understanding of the elevated temperature reactions of iron with alumina and/or silica is best obtained by investigation of the basic equilibrium that exists between the reacting constituents. In this investigation the equilibrium was studied under a controlled atmosphere of water vapor-hydrogen-argon in a LECO high temperature resistance, tube furnace. A detailed description of the laboratory furnace is given in Appendix II. Also, a flow diagram of the complete experimental apparatus is shown in Figure 1. With a water vapor-hydrogen mixture, the dissociation pressure for oxygen is obtained from the reaction: H2(g) 2 2(g) (g) (12) Prior experimental work has determined the equilibrium data for this reactionJ(36) iF~ - 60,180 - 13593T (13) log K = - 1, + 3o45 (14) T For a water vapor: hydrogen ratio of 0.25, the oxygen dissociation pressure at 1550~C is calculated as 2.9 x 10-10 atmo In order to place emphasis on the important experimental aspects of the investigation, the procedure is divided into the folloing categories: -14

k BQ ) i ) o a)' H 3a 9 Y +), e CO 4 M - ri 0 X 0 a E s s a [gIo _ | o rh * 1) | rd 1 ~aS Qcl~ ) O H r a) $ +'rQ A d k cjC b O'rg a) L 33 --— < l + ~ - - S 4 Cd a)a)Had ajk Cd0 PI W O D o j S __1P o co 9La C^ ~~~~~- N -H o^, _ x I -- c=3-o - ~ ~=5 -- I z ^ ^-H \^ ^\~\\\\\~ \.B ~~h~~\\\\ ~ ~ o~ ~'"'-~~V CJ~~~~~~~~~~~ =' ~ =~~~~~~~~~t -— =-o —-o-~ ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~r

-16(a) phase rule application (b) experimental materials (c) temperature control (d) atmosphere control (e) reaction technique (f) phase identificationo Phase Rule Application A phase rule analysis of the four component iron-oxygenhydrogen-aluminum system was helpful in establishing a program of experimentso A similar application of the phase rule to the ironoxygen-carbon-silicon system has been made by Colligan(5) Since the investigation was carried out under a condition of constant pressure slightly greater than one atmosphere, the phase rule can be expressed as: P + V C + 1 (15) where P =the number of phases V = variance or degrees of freedom C = the number of components, For the four component iron-oxygen-hydrogen-aluminum system, the partial equilibrium diagram shown in Figure 2 has been developed (24t) by Pillay(24 There must exist a quintuple point at the melting point of irono The five phases in equilibrium at this point are alumina, solid iron, liquid iron, hercyniteand gaso From a consideration of phase rule principles, there are five possible four-phase univariant equilibria which can emanate from this qunituple pointo

-170.50 0.40 ALUMINA + HERCYNITE + GAS.i. 0.30 I 0.20 - ALUMINA + LIQUID IRON + GAS - FLORIDIS o THURNER 0.10 o- PILLAY 1500 1600 1700 TEMPERATURE, C Figure 2. Partial Equilibrium Diagram for the Iron-Oxygen-HydrogenAluminum System After Pillay, et al.(24)

-18These are: (1) (2) (3) (4) (5) A1203 A1203 A1203 A1203 FeO'A1203 Fe(s) FeO'A103 FeO'A12l3 Fe Fes) Fe(g) Fe(s) Fe() Fe() Fe() gas gas gas FeO'A1 0 gas Equilibria (4) and (5) do not exist because of the respective experimental restrictions of (a) a constant gas pressure of one atmosphere, and (b) an ever present excess phase, alumina (A1203). Application of the rigorous phase rule technique outlined by L. O. Case(37) aids in a better understanding of such a system. Included in Appendix III are analyses of iron-alumina equilibria using Case's method. In establishing a series of experiments, the four phase boundaries in the iron-oxygen-hydrogen-aluminum system were first approached from the hercynite field. That is, the atmosphere was determined which would just prevent the formation of any reaction product. Subsequently, in order to verify that localized equilibrium was being achieved, coupled charges of iron-alumina and hercynitealumina were used to approach the boundary from the reducing side, In this case) hercynite (FeO~A1203) broke down into the constituents, iron and alumina (corundum). For the iron-oxygen-hydrogen-silicon system, only iron-silica charges were usedo

-19The temperatures of equilibration used in the investigation were 1450~C, 1500CC, 1517~C, 15355C, and approximately 1700 ~C Experimental Materials The materials used in the investigation of the iron-refractory reactions included: (a) Ferrovac "E" iron, (b) reagent grade BakerAdamson alumina, (c) reagent grade Fisher silica, (d) mullite, and (e) kaolinite, The reported analysis of the Ferrovac "E" iron is 99.9 percent iron, 0.025 percent carbon, 0.003 percent manganese, 0,007 percent silicon, 0o007 percent oxygen, and 0~002 percent nitrogen. All of the iron slugs were cleaned with a dilute solution of HC1, rinsed in deionized water and acetone and then dried. Weights of the iron slugs ranged from approximately 0.015 to 0.045 gramso Reagent grade Baker-Adamson alumina contained the following maximum limits of impurities: (a) C - 0.005 percent, (b) Si02 - 0.05 percent, and (c) S04 - 0.005 percent. Grain size of the material ranged from 115 to 120 mesh. Silica was reagent grade 140 mesh material from Fisher Laboratory Chemical Companyo Mullite was obtained from the Carborundum Company, Niagara Falls, New York. Clear crystals were selected from the aggregate and crushed to a size of 120 mesh. Polished sections of the "as received" material were examined under the reflected light microscope. This revealed that the mullite was completely free of corundum but did contain a small amount of silicate glass.

-20Natural kaolinite used in this study originated from the aggregate sample obtained by the American Petroleum Institute for its Clay Mineral Standards Project No. 49 (1950). Through the courtesy of Dr. R, M. Denning, small quantities of this aggregate sample were available from the Department of Geology and Mineralogy, The University of Michigan. The overall analysis of the kaolinite as listed in the project report is: TABLE II CHEMICAL ANALYSIS OF KAOLINITE Constituent Weight Percent SiO2 45.58* A1203 37.62* 2 O 57362* Fe203 100 FeO 0.13 MgO 0.03 CaO 0352 Na20 0.42 K20 0.49 H20+ 13,42 HO" 0.63 TiO2 1.42 SO3 C *The SiO2:A1 0 ratio for the calcined product or oecomposed kaolinite is 54.8/45.2.

-21Temperature Control A constant temperature was maintained in the LECO resistance tube furnace with a Wheelco controllero The controller was actuated by a platinum-platinum 10 percent rhodium thermocouple positioned near the furnace heating elementso Protection for the thermocouple was provided by a Triangle RR alumina thermocouple protection tubeo Prior to each experimenal run, the hot zone of the furnace tube was calibrated to the desired temperature with a separate platinum-platinum 10 percent rhodium thermocouple It had been previously determined that the hot zone length is two incheso The calibrating thermocouple was anchored in a recrystallized alumina combustion boat filled with powdered. reagent grade alumina0 Figure 4 illustrates the calibrating thermocouple in the alumina boato This duplicates the exact configuration of the experimental samples in which iron slugs were depressed in alumina and/or silica powdero The exit to the tube was closed with a brass plate sealo A detailed drawing and a photograph of the furnace tube seal are shown in Figure 3 and 4, respectively The calibration was carried out with a water vapor-hydrogen-argon atmosphere in the tubeo Once the furnace was calibrated to the desired temperature, the alumina boat with the "anchored" thermocouple was removed. An experimental sample boat was immediately inserted into the end of the tubea Because of the reducing characteristic of the water vaporhydrogen-argon atmosphere, contamination of the calibrating thermocouple occurredo It was necessary to clean the thermocouple frequently

-22GAS EXIT - SIGHTPORT / FURNACE TUBE 0.025" TUNGSTEN WIRE BRASS CAP SILICONE RU BER GASKET Figure 3. Schematic Drawing of the Furnace Tube Seal. Illlllllllg^^^^^.......g. g...,..i... "' D %|l~ ^ - f: B::.::: S.::::::::: -:::::::::::::::::::::::: ^i::^~:::............::: - eB.:-E: -::::::::::..::::::.:; ^^ ii.i. Figure 4. Photograph showing: (A) Furnace Tube Seal with Experimental Sample Boat, (B) Calibration Thermocouple, (C) NBS Thermocouple and, (D) Gas Sample Tube. gB B g ggR g g ~ aB f i l:ll;0j: ~ i0 tl;2 l.;;if400i;00iti.;00|04::i0j:.......... -:: i000000;:0 o a taB~f E hf B~aiv Ea~af LB. E:E E:E::- E E:::::::::E:::::::.:.:.::.:..:..:..... g g sg ~ fg:g g g i~ i g %;:::::s;: g~i~s: ~ i:X-:::;:-:S|;f;::-;L:-;;::4::;:0;::;: i::::-;:4:000;0 070500: ti::;t;:. ~E R~~fE::EB:: -...........................:::::.:::::.:.: f R s s: l. E: E::: E E:.:::.:::.::.:::.::::::::.::.: Figure~~~~~~~~~~~~~~~ a a:8g:| gB Photgraph0l0g 0T::::;:;::::!::j::;::j show0ging::t::.:: (A) Furnac Tube: Seal with;;44:: hf ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ J a %fE~i EE:::::::::::i::::::

-23in hot hydrochloric acid, The thermocouple was then checked against a similar platinum-platinum 10 percent rhodium thermocouple calibrated by the National Bureau of Standards to 1300~Co This standardizing procedure was carried out in the LECO laboratory furnace in air at temperatures of 1450~C, 1550~C, and 1600~Co Data for the NBS calibrated thermocouple was linearly extrapolated to the higher temperatures. The uncertainty in the NBS thermocouple is rated at not more than 2 degrees at temperatures of 1450~C and greater. With this uncertainty included, the temperature variation in the hot zone was determined as + 2o50C up to 15355~C For temperatures at 1700~C, the variation was determined as + 5~C. Atmosphere Control Commercial grade Airco hydrogen and argon were used to provide the atmosphere within the furnace tubeo Purification of the hydrogen was accomplished with an Engelhard Deoxo Gas Purifier. Argon was passed over copper gauze at 600~C (11100F) and then over calcium chips at 650~C (39) (l200'F) in order to remove any extraneous oxygeno Prior to mixing of the argon with the stream of wet hydrogen, the argon was thoroughly driedo This was accomplished by passing the gas through quantities of calcium sulfate (Drierite), magnesium perchlorate (Anhydrone), and phosphorous pentoxideo The water, oxygen, and nitrogen contents of the dried argon were checked with gas samples analyzed in the mass spectrograph. Both hydrogen and argon were carefully metered to maintain a minimum 6 to 1 argon to hydrogen ratio. Flow rates were 275 to 600

-24millimeters per minute for hydrogen and up to 3600 millimeters per minute for argono This corresponded to a linear gas velocity within the furnace tube of about 15 feet per minuteo With high gas flow rates, a 6 to 1 argon to hydrogen ratio, and a resistance furnace, Gokcen(20) and Floridis(25) contend that thermal diffusion effects are minimized, The desired water vapor:hydrogen ratio was obtained by first passing the hydrogen through a presaturator and then through two saturator units immersed in a water batho The presaturator was maintained at a temperature near that of the water bath, The saturator units were Corning Glass No. 31760 gas wash bottles equipped with coarse fritted discs of 40 to 60 micron pore size, The discs provided a uniform dispersion of hydrogen gas bubbles in the distilled water within the wash bottles, A third wash bottle, containing only a small amount of water, acted as an entrainment chamber. The heat sources for the constant temperature water bath were two stainless steel sheath immersion heaters, The bath was constantly stirred and the temperature was controlled with a special platinum-mercury thermoregulator, Bath temperatures were measured with a 50 - 1000C thermometer calibrated against a standard thermometer from the National Bureau of Standards, The temperature variation for the bath was determined as 0,03~C0 The performance of the saturators was checked by passage of the water vapor-hydrogen mixture through a Nesbitt absorption bottle containing calcium sulfate, magnesium perchlorate and phosphorous

-25pentoxide. The gain in weight of the bottle was determined after passage of a given volume of hydrogen which was collected over watero The results showed that saturation was achieved and agreed to within 2 percent of the calculated PH20/P2 ratio. Appendix IV contains a sample calculation for run D-48. Gas samples obtained after passage of the water vapor-hydrogen stream through the Nesbitt bulb were analyzed in the mass spectrograph. This provided a check on the oxygen, nitrogen, and possible residual water contentso Also, samples of the gas stream obtained at the exit end of the furnace tube were analyzed for oxygen and nitrogen. Reaction Technique Small amounts of refractory material were placed in a Triangle RR recrystallized alumina boat, Each material was separated from the other by a recrystallized alumina plateo A Ferrovac "E" iron slug of approximately 0.015 to 0.045 grams was forced into each sand to achieve intimate contact between the iron and the sand, For experiments involving iron-alumina equilibria near a univariant boundary, the experimental sample boat contained a charge of previously formed hercynite and aluminao* Following calibration of the furnace to the desired temperature, the filled sample boat was placed in the cold end of the furnace tubeo During this operation, argon was passed through the tube to act as a flushing medium. a*For this report the combined sample of iron-alumina and hercynitealumina is referred to as a coupled charge or coupleo

-26After the tube was sealed the atmosphere with the desired p 20/PH2 ratio was admitted~ Saturator performance was checked by obtaining a gain in weight sample from the Nesbitt absorption bulbo The alumina boat was retained for 30 minutes in the cold end of the furnace with the water vapor-hydrogen-argon stream flowing through the tube. The boat was then correctly positioned in the hot zone of the furnace tube by means of a tungsten wire (0.025 in, diameter) previously attached to the boat. At the completion of the run, the sample boat was pulled to. the cold end of the tube and allowed to cool in the furnace atmosphere for 15 minutes. Just prior to removal of the boat, a gravimetric sample of the gas stream was obtained to verify that saturation was still being accomplished, Phase Identification After completion of the run, the alumina boat was removed from the furnace tube and each sample was examined visually. It was necessary to check that the iron-refractory charges had not been disturbed during positioning in the hot zone of the furnace, Careful inspection also provided some indication of the presence or absence of reaction products, For temperatures below the melting point of iron and under reducing conditions, the iron-alumina and iron-silica samples exhibited little adhesion of sand grains to the iron slugo For iron-alumina charges at higher temperatures under oxidizing conditions, a quantity of hercynite was produced that was clearly discernable as black grains adherring to the remaining unreacted iron

-27For microscopic examination, the samples were impregnated in a partial vacuum with Palmer Cement, a thermosetting resin. This eliminated the friability of the granular material and also preserved the important contact area between the iron slug and the individual sand grains, Examination of polished sections under the reflected light microscope provided positive identification of phases formed in the localized zone of equilibrium~ Previous studies have used optical microscopy to identify the phases encountered in this investi(4o)(4l) gation, (4)4 Powder x-ray diffraction techniques were used to accomplish the following.: (a) verify the presence of hercynite (FeOoAl203) for oxidized iron-alumina samples fired at 1450~C and 15355C, and (b) verify the presence of mullite (3A120302SiO2) for iron-kaolinite samples fired at 1450~C and 15355~C

CHAPTER IV EXPERIMENTAL RESULTS The results of this investigation are presented according to the following systems of chemical elements: (a) the iron-oxygenhydrogen-aluminum system, (b) the iron-oxygen-hydrogen-silicon system, and (c) the iron-oxygen-hydrogen-aluminum-silicon system. Equilibration Studies in the Iron-Oxygen-Hydrogen-Aluminum System The reaction product formed on equilibrating iron and alumina at oxidation levels used in this investigation (PH 0/PH2 <_ 050) is hercynite (FeOoA1203)o Figure 5 is a low magnification view of an experimental sample equilibrated at 1535~C for 5 hours at PH o/PH = 0o41l The reaction zone with numerous grains of hercynite in equilibrium with corundum is clearly visible. Because of the slowness of the reaction, a considerable portion of the iron slug still remainso The Solid Iron-Alumina-Hercynite-Gas Boundary Charges of iron-alumina at 1450~C bracket this four phase boundary between the oxidizing atmosphere of pH20/PH2 = 0.256 and the reducing atmosphere of pH20/PH2 = 0.226. Figures 6 and 7 illustrate the results at the respective water vapor:hydrogen ratioso For the above oxidizing condition hercynite is present in the localized region between the iron and aluminao Under the above reducing condition there is no evidence of any reaction product between the charged materials -28

-29Iron-alumina charge PO2' 2,4l Magnification: 1 15350C - 5 hours Figure 5. Microstructure of Sample from Experiment A-45.* The code symbol for the phases in the photomicrographs is given in Appendix V.

-30Iron-alumina charge p / = 0.236 Magnification: 500X 14500C - 5 hours Figure 6. Microstructure of Sample from Experiment A-27. 1 aXri~~~~~~~~~~~~~~~~vli Iron-alumina charge poo/PpH = 0.226 Magnification: 250X 1450C 4 hours Figure 7. Microstructure of Sample from Experiment A-28.

The results of approaching the four phase univariant boundary with hercynite-alumina charges are equally confirming Figure 8 illustrates the reaction zone of hercynite-alumina equilibrated at 1450~C under the oxidizing atmosphere of PH20/PH2 = 0.256~ The charge remains completely stable with no reduction occurring~ Figure 9 exhibits the reaction zone for hercynite-alumina at 14500C and the reducing atmosphere of PH 0/PH = 0.218. The hercynite (FeOoA1205) has begun to undergo decomposition0 Small particles of iron are clearly visible within the existing hercynite grains, Figure 10 shows the results of the iron-alumina portion of the couple equilibrated at 1450~C and PH o/PH = 0o218. For this reducing atmosphere, there is no evidence of any reaction product in the region between the charged constituentso The results of an iron-alumina charge at 1450~C under the more oxidizing condition of SH20/PH2 = 0o41 is shown in Figure 11 There is extensive formation of hercynite with delineation between the two equilibrated phases, alumina and hercynite, quite marked0 Results obtained on firing a coupled charge of iron-alumina, hercynite-alumina at 1500~C position the four phase boundary above a water vapor:hydrogen ratio of 0o217o Figure 12 shows alumina immediately adjacent to the iron slugs There is no evidence of any reaction product in the region between the two constituents0 For the hercynite-alumina charge, Figure 13 illustrates appreciable reduction of numerous hercynite grains to s.iall particles of iron and corundum. With sufficient equilibration time, the decomposition of all hercynite could be expected with only the phases iron and corundum remainingo

-32Hercynite-alumina charge 0.256 Magnification: 500X 14500C 2 5 hours Figure 8. Microstructure of Sample frcn Experiment A-27, Hercynite-alumina charge P 0/p t 0.218 Magnification: 250X 4 -7 hours Figure 9. Microstructure of Sample from Experiment A-29.

-33Iron-alumina charge 0.218 Magnification: 250X 1500C 7 hours Figure 10. Microstructure of Sample from Experiment A-29. Iron-alumina charge p20/p2 l. Magnification: 250X 14500~C - 45 hours. Figure 1l. Microstructure of Sample from Experiment A-21.

-34Iron-alumina charge p /PR = 0.217 Magnification: 250X 1500~C - 6 hours Hercynite-alumnina charge Magnification: 250X Figure 13. Microstructure of Sample from Experiment B-40. Iron-aliumina charge pEo = 0.240 Magnification: 250X 1517~C - 5 hours Figure 14. Microstructure of Sample from Experiment C-41.

-35Results of an iron-alumina charge at 15170C places the four phase univariant boundary below PH 2/PH = 0O240o Figure 14 exhibits the formation of hercynite in the localized region between the iron and aluminao Using the mean value of the results of the limiting charges at 1450~C,. 1500~C and 1517~C, the four phase univariant equilibria can be positioned as follows: (a) PH 0/PH = 0230 at 1450~C and (b) pH20/ = Oo235 at 15000C. The Liquid Iron-Alumina-Hercynite-Gas Boundary The results of iron-alumina charges at 15355C bracket the four phase boundary of liquid iron-alumina-hercynite-gas between the water vapor:hydrogen ratios of 0o256 and 0o231o Figure 15 illustrates the results for the oxidizing condition (PH20/PH2 = 0o256)a The formation of hercynite immediately adjacent to the iron slug is clearly evidento For the reducing condition (PH20/pH2 = 0o231), Figure 16 illustrates that no reaction product has developed. The four phase boundary was also approached at 15355C from the reducing side with a coupled charge of hercynite-alumina and ironalumina. Figure 17 illustrates the result of the hercynite-alumina portion at 1555~C and PH20/PH2 = 0213o Decomposition of the hercynite grains is complete to give small particles of iron and corundum. The large mass of iron appearing at the edge of the photograph is a remnant of the original iron-alumina charge used to form the hercyniteo The result of simultaneous equilibration of the iron-alumina portion of the couple is shown in Figure 18o There is no evidence of any reaction

-36Iron-alumina charge p2 2/p. = 0.256 Magnification: 250X 15550C - 5 hours Figure 15. Microstructure of Sample from Experiment D-47. Iron-aliuina charge p p 0.251 Magnification: 250X 15355C - 4 hours Figure 16. Microstructure of Sample from Experiment D1s

-57Hercynite-alumina charge P 2 = 0.213 Magnification: 250X 1535OC - 5 hours Figure 17. Microstructure of Sample from Experiment D-50.:.::.. 1 1... Iron-alumina charge Magnification: 250X Figure 18. Microstructure of Sample from Experiment D-50.

v-8productQ The striations present on the alumina grains are a characteristic occurrence in this investigation. They result from the cutting action of the 5 micron diamond compound during the polishing sequence. Subsequent polishing with a water - Linde B suspension does not totally remove the marks from the hard aluminao Under the more oxidizing conditions of PH 0/PH = 0o416 and 2 2 00514, iron-alumina charges undergo extensive reaction as shown in Figures. 19 and 20, respectively. The formation of hercynite is clearly evident with excellent delineation between the two equilibrated phases, Results of exposure of iron-alumina charges at 1708~C and water vapor:hydrogen ratios of 0O314 and 0278 position the univariant boundary below 00278 for this temperatureo Figures 21 and 22 for the two oxidizing conditions show an equilibrium of hercynite and corundumo For temperatures of approximately 17000C, the formation of hercynite in this study is established at oxidation levels somewhat lower than (24) formerly cited by Pillay. ) From the results at 1535~C and 1708~C, together with pertinent thermodynamic calculations to be discussed later, the position of the liquid iron-alumina-hercynite-gas phase boundary is established. The Quintuple Point, Solid Iron-Liquid Iron-Alumina-Hercynite-Gas and the Solid Iron-Liquid Iron-Alumina-Gas Boundary Intersection of the univariant boundaries of solid iron-aluminahercynite-gas and liquid iron-alumina-hercynite-gas, together with available data on the melting point of iron, determines the position of the quintuple pointo

-59Iron-alumina charge PH o/P~ = 0o416 Magnification: 250X 1535~C - 5 hours Figure 19. Microstructure of Sample from Experiment D-45Iron-alumina charge pJ 0 = 0.514 Magnification: 250X 1535~C - 3.5 hours Figure 20. Mierostructure of Sample from Experiment D-42.

14o. Iron-alumina charge p /P2 0.314 Magnification: 500X 1708~C - 4.5 hours Figure 21. Microstructure of Sample from Experiment E-54. Iron-alumina charge P O /PH 2 0.278 Magnification: 500X 1708 C - 3 hours Figure 22. Microstructure of Sample from Experiment E-55.

41The univariant phase boundary of solid iron-liquid iron-aluminagas must emanate from this quintuple point as outlined in a previous section involving phase rule analysiso For this investigation the partial pressure of oxygen in the vicinity of the quintuple point is about 1010 atmo The solubility of alumina in either solid or liquid iron can be considered negligibleo It is also known that the melting point of iron is lowered by the addition of other elements, in this case, oxygeno With these considerations, a straight line approximation of the four phase univariant boundary can be madeo Phase Field Boundaries at Very High Temperatures In. accordance with phase rule principles, the univariant four phase boundary of liquid iron-alumina-hercynite-gas must terminate at an upper quintuple point of liquid iron-alumina-hercynite-melt-gaso Also, from this quintuple point, there must emanate two additional four phase univariant boundaries. The results of an iron-alumina charge at 1715~C and PF2O/PH2 = O~367 is shown in Figure 235 An oxide melt and corundum crystals are the phases in equilibrium. Also, from the phase diagram (18) of FeO-A1203 established by Fisher and Hoffman, it is known that the equilibrium of alumina-hercynite-liquid exists in air (p0 = 0.21 ~2 atmo) at 1750~Co With this information it is possible to approximate the four phase univariant boundary of alumina-hercynite-melt-gas in a vertical direction from the upper quintuple point0 Figure 24 illustrates the result of exposure of an iron-alumina charge to a reducing atmosphere of PH n/PwH = Ool99 at 1717~Co There

-42Iron-alumina charge P32 o = ~367 Magnification: 250X 1715C 3 hours Figure 23. Microstructure of Sample from Experiment E-53....... Iron-alumina charge p20p2 = 0.199 Magnification: 250X 171 C - hours n 24 icaon 2 17170C 5 hours

-453is no evidence of reaction. Some small particles of iron are shown situated within the pores of the alumina graino With this information the four phase univariant boundary of liquid iron-alumina-melt-gas can be approximated in a horizontal direction from the upper quintuple pointo Equilibration Studies in the Iron-OxygenHydrogen-Silicon System Simultaneous equilibration of iron-silica charges with each alumina-iron and/or hercynite sample served two significant purposes: (1) The results obtained could be checked against the equilibrium diagram already established by Darken ) and Colligan5) and shown in Figure 48, For a given experimental run, any discrepancy in ironsilica equilibria would serve to cast doubt on the validity of the results obtained for the other systems. (2) Iron-silica charges equilibrated at 14500C permit closer positioning of the solid ironsilica-melt-gas univariant boundary at this temperature. The results of this system are presented in terms of (1) and (2), with the latter being considered firsto The Solid Iron-Silica-Melt-Gas Boundary The results of iron-silica charges at 1450~C are in good agreement with previous experimental work at 1250 ~C 1400~C, and 1500~Co(5) The univariant phase field boundary of solid iron-silicamelt-gas is bracketed between the oxidizing atmosphere of pH20/PH = 0.313 and the reducing atmosphere of PH2O/pH2 0.278. Figures 25 and 26 illustrate the results for the respective water vapor:hydrogen

-44Iron-silica charge PI H2 0.313 Magnification: 250X 1450~C - 6 hours Figure 25. Microstructure of Sample from Experiment A-24. Iron-silica charge pH20/P2 = 0.278 Magnification: 250X 1450~C - 4 hours Figure 26. Microstructure of Sample from Experiment A-25.

-45, ratioso Under the oxidizing condition there is definite formation of an iron silicate melto In addition} a small particle of iron has formed because of reduction of the melt on cooling through the iron stable fieldo For the reducing atmosphere there is no evidence of any reaction product in the region of the iron-silica interface. Figure 27 shows the results of an iron-silica charge equilibrated at 1450~C under the oxidizing condition of PH20/pH = 0o3477 An extensive amount of iron silicate liquid is in evidenceo Also, from the morphology of the grains, tridymite is the most abundant form of silica presento For the more reducing condition of PHo0/pH = 0236 at 1450~C, Figure 28 illustrates the interface of iron and silicao There is no evidence of any reaction producto However, the corresponding iron-alumina charge equilibrated at this temperature and atmosphere results in hercynite formation as shown in the previous section (Figure 6)o The Iron Silicate Phase Field Included in this section are the results of iron-silica equilibrations simultaneously obtained with critical limiting iron-alumina equilibrations already discussedo The results conform with the iron-oxygen-hydrogen-silicon equilibrium diagram to which reference has already been made (Figure 48)~ At 1500~C and 15170~C the equilibration of iron-silica charges for PH 0/PH = 0O217 and 0O240, respectively, results

-46Iron-silica charge P I2/PoI = 0.547 Magnification: 250X 1450~C - 5 hours Figure 27. Microstructure of Sample from Experiment A-22. Iron-silica charge 02 Magnification: 250X 14500C - 5 hours Figure 28. Microstructure of Sample from Experiment A-27. Iron-silica charge P02 0217 Magnification: 250X 1500~C - 6 hours Figure 29. Microstructure of Sample from Experiment B-40.

-47in appreciable iron silicate liquid formation as shown in Figures 29 and 300 For both experimental runs, the results from corresponding iron-alumina charges helped locate the solid iron-alumina-hercynitegas boundary. For a temperature of 15355C the results of iron-silica mixtures at PH20/PH2 = 0.213 and 0~256 are shown in Figures 31 and 32~ An extensive amount of iron silicate liquid is in evidence in the reaction zoneo Spherical particles of iron formed on cooling through the iron stable region are also presento For both these experimental runs, the results from corresponding iron-alumina charges aided in positioning the liquid iron-alumina-hercynite-gas boundary~ The Iron Stable Phase Field The result of an iron-silica charge under reducing conditions of PH20/PH2 = 0.08 at 1535~C is shown in Figure 33~ There is no evidence of any interface reaction0 Iron and silica are the only phases present, in agreement with the iron-oxygen-hydrogen-silicon equilibrium diagram0 The significance of the results from the corresponding iron-mullite charge for this experimental run will be discussed latero Equilibration Studies in the Iron-Oxygen-HydrogenAluminum-Silicon System Results from the equilibration of charges of iron and alumina + silica can be interpreted in terms of the five component system of ironoxygen-hydrogen-aluminum-silicon. In general the phase rule stipulates that the addition of another component to a given system simply increases the degree of freedom or variance by oneo

-48Iron-silica charge Pr oP/ p 0.240 Magnification: 250X 15170C - 5 hours Figure 30. Microstructure of Sample from Experiment C-41. Iron-silica charge P)20/PH2= 0.213 Magnification: 250X 1535C - 5 hours Figure 31. Microstructure of Sample from Experiment D-50. Iron-Silica charge O2/PH = 0. Magnification: 25OX 1535~C - 5 hours Figure 32. Microstructure of Sample from Experiment D-47.

-49Iron-silica charge o/PH = 0.08 Magnification: 250X 15T55O - 3 hours Figure 33. Microstructure of Sample from Experiment D-52.

-50Results for each of the related four component systems have already been presented~ In this composite five component system, experimental procedures have determined the equilibrium phase assemblage for a given set of conditions of temperature and atmosphere, The variance or degree of freedom associated with each phase assemblage can be calculated using the phase ruleo The results obtained for charges of iron and alumina + silica are discussed in terms of the two excess phases used: (a) mullite (3Al203o2Si02) and (b) kaolinite (2A12Si205(OH)4)o Iron-Mullite Charges The firing of iron and an excess of mullite formed an equilibrium zone somewhat similar to that found in the iron-alumina serieso A considerable number of holes and pits developed which necessitated careful impregnation with Palmer Cement in order to obtain satisfactory polished sectionso Examination of the microstructure with the reflecting light microscope revealed the phases in the localized zone of equilibriumo Iron-mullite charges were equilibrated at an oxidation level of approximately p20/PH = 0,22 for the temperatures 14500~C 1500~C, 1535~C, and 1700~Co In each case three specific condensed phases and a gas constitute the equilibrium situationo The three condensed phases are: (a) corundum, (b) liquid, and (c) mulliteo Figures 34, 5, 6, and 37 illustrate the results at each temperature and water vapor:hydrogen ratioa

-51Iron-mullite charge PH20/PT2 0.218 Magnification: 250X 1450~C - 7 hours Figure 34. Microstructure of Sample from Experiment A-29. Iron-mullite charge p 0/P 0.217 Magnification: 250X 1500~C - 6 hours Figure 35. Microstructure of Sample from Experiment B-40.

-52Iron-mullite charge Pf20/PH2 = 0.213 Magnification: 250X 1535~C - 5 hours Figure 36. Microstructure of Sample from Experiment D-50. Iron-mullite charge oPH = 0.199 2 2 Magnification: 250X 1535~C - 3 hours Figure 37. Microstructure of Sample from Experiment E-56.

-53Also, at 1535~C the phase assemblage remains the same for oxidation levels as low as PH O/PH = 0.125. Figure 38 exhibits the equilibrium zone with corundum, liquid, and mulliteo Just as in the systems involving individual iron-alumina and iron-silica charges, the reducing conditions necessary for high temperature iron-mullite stability are also determined. An extremely low water vapor:hydrogen ratio is necessary to prevent the formation of reaction products from iron-mullite charges at 1535~Co Figure 39 illustrates the results of the iron-mullite combination at this temperature and pH20/PH2 = o008. There are no reaction products visible in the photomicrographo A small amount of liquid is entrained within the mullite grains. This liquid is, however, not the result of any reaction, rather it is liquid that is contained within the original charged mulliteo Figure 40 shows the results of mullite fired simultaneously with the previous sample under the conditions of 1535~C and PH O/pH2 = 00o8, but in the absence of iron. Small amounts of liquid are visible, and its similarity to the liquid in Figure 39 can be notedo Finally, Figure 41 depicts unfired "as received" mullite at 150 magnificationo A small amount of liquid is clearly visible within the mullite grains. From the revised Si02-A1203 phase diagram of Aramaki and Roy(10), the overall composition can be estimated in the vicinity of 60 mole percent aluminao

-54Iron-mullite charge PH O/P2 = 0.125 Magnification: 250X 1535~C - 3 hours Figure 38. Microstructure of Sample from Experiment D-51. Iron-mullite charge PH20/PH2 = 0.08 Magnification: 250X 1535C - 5 hours Figure 59. Microstructure of Sample from Experiment D-52.

-55Iron-mullite charge /p o.o8 Magnification: 250X 1555 C 3 hours Figure 40. Microstructure of Sample from Experiment D-52. Fi__g__re__ 41pirsrueo a eevd" E Mt. FiguS O~~~~~:iiire-i 4 i.i. M-i - cr-o- trt —Er of ="as receive -d." Mu:iliteg-:.-i.-: Ni-;. Figure 41. Microstructure of Aas received" Mullite.

-56Iron-Kaolinite Charges The results of iron-kaolinite charges (Si02/A1203ratio = 54o8/45o2) at oxidation levels and temperatures identical to those used for iron-mullite can again be described in terms of three condensed phaseso These include (a) mullite, (b) liquid, and (c) "decomposed kaoliniteo" Positive identification of each phase was accomplished by examination of the reaction zone with the reflecting light microscopeo In addition, x-ray diffraction studies of finely crushed material also from the reaction zone verified the presence of mulliteo Unlike the results from charges of iron-alumina, iron-silica, or iron-mullite, the results from combination of iron-kaolinite indicate an almost complete disappearance of the irono The accompanying formation of a liquid phase is extensiveo That portion of the kaolinite not in the vicinity of the iron remains unattacked but does undergo decomposition, accompanied by appreciable shrinkageo The overall result after firing is a sample which is quite different from that initially positioned in the furnaceo Figure 42 is a low magnification view of the iron-kaolinite sample after firing at 1535~C and PH2O/PH = 0256~ The region originally occupied by the iron is clearly evident. Bakelite now fills the voido The region immediately adjacent is liquid with mullite crystals becoming more predominant in the outlying areaso Figures 43, 44 and 45 depict results under high magnification for iron-kaolinite charges at 1450~C, 1500~C, and 1535~Co The oxidation

-57Iron-kaolinite charge PH 0/PH = 0. 256 Magnification: 150X 1555 C - 5 hours Figure 42. Microstructure of Sample from Experiment D-47. Iron-kaolinite charge PH o/P = 0.218 Magnfication: 250X 1450 - 7 hours Figure 43. Microstructure of Sample from Experiment A-29.

-58Iron-kaolinite charge PH20/PH2 = 0.217 Magnification: 250X 15000C - 6 hours Figure 44. Microstructure of Sample from Experiment B-40. Magnification: 250X 1535~C - 5 hours Figure 4~[5. Microstructure of Sample From Experiment D-50. Magnif i cat ion: 25OX 15350C - 5 hours —Fiur 4- icotrctreofSape ro Epei

-59level for each sample in this series is approximately the same (pH O/PH2 = 0.22) For each of the iron-kaolinite samples the two condensed phases of primary interest are (a) mullite and (b) liquid. The third condensed phase jdecomposed kaolinite" is shown only in Figure 44. Details concerning the decomposition reactions involving kaolinite have already been reviewed in Chapter IIo

CHAPTER V DISCUSSION OF RESULTS The results of this investigation are discussed in terms of the systems already presented in the previous section. namely the equilibria of iron-alumina, iron-silica, and iron-(alumina+silica). In addition, a final section is included which describes results obtained from a study on non-metallic macroinclusions in steel castingso These data are discussed in relation to the results established by the equilibrium studies of this investigation. Iron-Alumina Equilibria Experimental results obtained for this system permit construction of the equilibrium diagram shown in Figure 460 Pertinent data already discussed in Chapter II is also included for reference purposes, The positions of the following equilibria are established; (a) solid iron, hercynite, alumina, and gas (b) liquid iron, hercynite, alumina, and gas (c) solid iron, liquid iron, alumina, and gas (d) the quintuple point-solid iron, liquid iron, hercynite, alumina, and gaso In addition, for very high temperatures the locations of the equilibria listed below are estimated: (e) liquid iron, melt, alumina, and gas (f) hercynite, melt, alumina, and gas -60

-61-, (g) the quintuple point-liquid iron, hercynite, melt, alumina, and gaso This equilibrium diagram defines clearly the effect of both temperature and atmosphere on the reactions involving iron-oxygen-hydrogen-aluminumo Interpretation of the experimental measurements in a quantative manner requires the use of a plot of the logarithm of the chemical equilibrium constant against the reciprocal absolute temperatureo The limiting, gas ratios defining the univariant equilibria for the ironalumina system are plotted in this fashion in Figure 47- Straight lines are drawn through the mean values of the gas ratios as a best estimate of equilibrium conditionso The equation of the line for the solid iron region is calculated as: n) 1076 + o084 (16) The line is terminated at the melting point of iron which is taken as 1533~C for highly reducing conditions. Also, from the straight line representation, the free energy of formation of hercynite can be calculated for the reaction: Fe(8) + A1203+ H20(g) 2 FeOoAl203 + 2 (17) AF~ = -2095 - lo7T (18) Equation (18) can be combined with the results of Darken and Gurry(6) on the reaction: Fe0() + H2,(g) Fe(s) + H2O(g) (19)

-62This permits calculation of the free energy of formation of hercynite from constituent oxides according to the reaction: FeO(X) + A1203 - Fe0OA1203 (20) AFO = -5350 cal/mole (21) 1500~C (22) In comparison, Fisher and Hoffman(2) have estimated the free energy of formation of hercynite from e.omfo measurements on iron oxide-alumina mixtures at 1500 0C The form of the Faraday equation used in the calculation is: AF~ = -Z1(nl-n2)(96,490)(0.239)E (22) For the hercynite forming reaction FeO + A1203 v FeOAl203 (23) Z1 is set equal to 6, E is determined from experiment as 0.27 volts and (nl-n2) is estimated from 0~2 to 0.40 The latter quantity being an estimate of the transport mobility of the positive ions, aluminum and irono Calculation gives values of -7500 to -15,000 calories/gm-mole for the free energy of formation of hercynite at 15000Co Considering the estimate made by Fisher and Hoffman on the ionic mobility factor, the e.m.fo results corroborate the findings of this investigation on iron-alumina interaction in a satisfactory manner. From Equation (18) the heat of reaction is -2100 calories/ gm-mole, indicating a slightly exothermic reactiono This is of the

-63same order of magnitude as other exothermic reactions involving the oxidation of solid phases by gaseous oxygen, namely the formation of 2FeOoSiO2, 2FeOoTiO2, and 2FeOoCr203(5) By including a term for the heat of fusion of iron, Equation (16) becomes applicable for higher temperatures. In Figure 47 the equation of the straight line in the liquid iron region is: p 2940 In -L.\ = - 0.19 (24) H2 T From this equation the free energy of formation of hercynite can be calculated for the reaction: Fe() + A12 3 + H20(g) Fe 23 2(g) (25) iF0 = -5850 + 0037T (26) Pillay reports an equation F~ = -5260 + 03.2T (27) Agreement between the two sets of data can be considered satisfactoryo An explanation on the positioning of the boundary is, nevertheless, necessary~ For 1700~C Equation (24) is used to calculate the equilibrium gas ratio for the univariant line, liquid iron-hercynite-alumina-gas. The value of PH20/PH2 is 0.271. Results from a charge of ironalumina at 1708~C and PH 2/PH = 0.279 show the formation of the reaction product, hercyniteo This is in good agreement with the calculation. Consequently, the univariant phase field boundary is positioned lower than that originally determined by Pillayo

From the data obtained at temperatures in the vicinity of 1700~C, the two univariant equilibria, (e) and (f), and the quintuple point (g) are approximately located..An important phase relationship at these higher temperatures can be pointed outo If Figure 46 was expanded, a portion of the "air line" representing the locus of the PH20/PH2 ratios for p0o = 0.21 at each temperature for the reaction HH +0 HO(28) (g) (g) 2 (g) (28) could be included. The equilibrium of alumina-hercynite-melt(liquid)gas lies on this line at 1750~C as reported by Fisher and Hoffman (18) The univariant boundary must then extend from this point and terminate at quintuple point (g). Iron-Silica Equilibria As stated previously, an evaluation of results from charges of iron-silica have provided: (a) a check on the overall experimental technique and (b) additional data to position the univariant solid iron-silica-melt-gas phase boundary at 1450~C, a temperature at which no previous work has been reported. Figure 48 is the pertinent equilibrium diagram resulting from the work of Darken(14) and Colligan (15) Originally the ordinate was PC /PCO~ For this investigation, however, it is more convenient to denote the ordinate as PH2O/PH2o Conversion of the diagram from the CO - C02 gas system to the H2 - H20 system is accomplished by use of the reaction:

-65Co2 + H CO + H20 (29) log ( ~= _ 13 + 1.360 + log ( (30) 1H2 T kPco 0 The results of all the iron-silica equilibrations from this investigation are plotted on the diagram. Both iron stable and melt stable samples are appropriately marked as sucho Comparison of this diagram with the equilibrium diagram for the iron-oxygen-hydrogenaluminum system shows that iron-silica equilibrations bracket the important univariant boundaries, thereby supporting the results obtained for the corresponding charges of iron-alumina and/or hercynite-aluminao Finally, experimental results at 1450~C accurately position the univariant boundary of solid iron-silica-melt-gaso The location of the boundary at this temperature agrees very well with previous work at 12500~C 1400~C, and 15000CO (5) Iron-(Alumina+Silica) Equilibria It has already been indicated that the addition of another component to a given system simply increases the degree of freedom or variance by oneo For the four component systems studied in this investigation, three condensed phases and gas constitute univariant equilibriumo This is represented by a line on the pressure-temperature diagram. For the five component system of iron-oxygen-hydrogenaluminum-silicon, three condensed phases and gas constitute bivariant

-66equilibrium. With a given atmosphere the specific phase assemblage exists over a range of temperatures. For a given temperature the same phase assemblage exists over a range of atmosphereso The phase combinations developed in this system are reviewed in terms of the following three categories: (a) excess mullite, (b) excess kaolinite,and (c) application to the iron oxide-A1203-SiO2 equilibrium diagram0 Excess Mullite The results obtained from iron-mullite charges are first interpreted by noting the characteristics of the equilibrium diagrams (Figures 46 and 48) for the two four component systemso The diagram for a strong excess of silica is characterized by a large expanse of liquid silicate phaseo The solid iron-silicamelt-gas boundary decreases rapidly from 1250~C to 1533~Co There, a highly reducing atmosphere of PH20/PH2 = 0o146 is necessary to prevent melt formationo On the other hand, the diagram for an excess of alumina shows no liquid reaction product at any atmosphere condition until temperatures in excess of 1700~C are attained0 Also, at lower temperatures (\ 1535~C) for iron-alumina samples the formation of the solid reaction product, hercynite, can be prevented with atmospheres not nearly as reducing as those necessary to prevent melt formation in the iron-oxygen-hydrogen-silicon system0 Results of iron-mullite equilibrations have been presented for a range of temperatures at reducing conditions of approximately

-67PH o/PH2 = 022~ A liquid reaction product was produced at all temperatures from 1717~C down to 1450~Co The formation of a liquid reaction product occurs even with temperatures low enough and atmospheres sufficiently reducing to inhibit the formation of any reaction product in both the iron-alumina and iron-silica systemso The assemblage of condensed phases (a) corundum, (b) liquid, (c) mullite, and the gas phase constitutes bivariant equilibrium with V = 2~ The phase rule can be applied as follows: V = C + 1 P (31) V = 5+1-4 V 2 This phase combination exists over a range of temperatures (1450~C to 1717~C) for a given atmosphere condition (PHO/PH2 - 0o22)o Also, as shownfrom the results, it exists over a range of atmospheres (PH2 /PH2 = 00125 to PHH20/H2 = 00213) for a given temperature (1535~C)o Finally, in order to suppress the formation of reaction products from charges of iron-mullite, an extremely reducing atmosphere is necessary. At 1535 ~C the required condition is PH20/PH = 0o08 This corresponds to a partial pressure of oxygen of 2~2 x 1013 atmospheres Excess Kaolinite The results obtained from iron-kaolinite charges can also be described in terms of a phase assemblage of three condensed phases and a gas phaseo Kaolinite is represented by the formula

-682Al2Si205(OH)4j, with the calcined material having a SiO2/A1203 ratio of 54~8/45.2. Mullite, on the other hand is represented by 3Al2032SiO2 with a stoichiometric composition of 28o1 percent silica and 719 percent alumina. With the lower alumina and higher silica contents, kaolinite is subject to greater attack by iron oxide. Also, unlike mullite, kaolinite undergoes decomposition on heating. Experimental results have been presented for temperatures 14500C, 15000C, and 15350C at the approximate atmosphere condition of PH o/P. = 0.22. Also, for temperatures 1450~C and 1535~C, the PH 0/pH level was varied. For all cases the phase assemblage consisted of the condensed phases (a) mullite, (b) liquid, and (c) "decomposed kaolinite" and the gas phase. The formation of a liquid phase was extensive, confirming the susceptability of kaolinite to iron oxide attack0 With the phase combination remaining the same over an area of temperature and atmosphere levels, the variance of the equilibrium situation is 2~ This is in agreement with the phase rule as applied to this five component system: V =C + 1 -P (32) V = 5+1-4 V = 2 Correlation with the Iron Oxide-A1203-Si02 Diagram Results obtained from the iron-mullite and iron-kaolinite reactions can be further interpreted and explained with the help of the iron oxide-A1203-SiO2 ternary diagram0 Figure 49 is a liquidus plot showing phase relationships for the system at a reducing

-69atmosphere of about p = 1013 atmospheres(30 The liquidus plot 2 shows the isotherms at which the liquid is saturated with solid phaseso For this investigation all samples were cooled by rapidly pulling them to the cold end of the furnace tube. The ternary diagram cannot, therefore, be used to predict a possible sequence of crystallization for any of the sampleso This would necessitate slow cooling under equilibrium conditionso An interpretation of a qualitative nature is, nevertheless, possible. Comparison of results for iron-mullite and iron-kaolinite reactions shows that corundum was a reaction product for the ironmullite series, and mullite was a reaction product for the ironkaolinite serieso The overall composition for the iron-mullite and iron-kaolinite charges must lie on lines (A) and (B), respectively, as shown in Figure 49o This line connects the apex, iron oxide, with the appropriate composition on the A1203-SiO2 side of the triangleo Also, the experimental condition of a strong excess of mullite or kaolinite positions the overall composition near the A1203-Si02 side of the diagram. Rectangles locate these approximate compositionso The primary fields of crystallization for the two composition rectangles are, however,, differento The overall composition for iron-mullite reactions is situated in the primary corundum field, while for the iron-kaolinite series the overall composition is located in the primary mullite field. For the temperatures used in this investigation the equilibrium condensed phase assemblage for the

excess constituents (a) mullite and (b) kaolinite is: (a) The equilibrium of liquid with corundum and the excess phase, mulliteo (b) The equilibrium of liquid with mullite and the excess phase "decomposed kaolinite " Finally, tie lines can be projected through the composition rectangles on lines (A) and (B) from the alumina apex and the mullite composition point, respectivelyo This locates the approximate liquid compositions attained for the temperatures used in this investigationo Circles in Figure 49 locate these approximate compositionso Correlation of Equilibration Results with Large Scale Refractory-Metal Reactions The equilibration samples from this investigation are particularly significant when compared to the results from large scale refractory-metal reactions. This section of the report includes results from an evaluation of refractory-metal interaction (42) as a source of macroinclusions in steel castings () The original work described only the nature and extent of the reactionso No analysis of the important reaction zone in terms of microstructures was included, Consequently, it was felt that a correlation of these microstructures with the equilibration results from this study would prove both interesting and informativeo In order to provide the reader with some background on the previous work, some of the important aspects involving experimental procedure are presented. This is then followed by a discussion of

-71the microstructures obtained from interaction of refractory and molten steelo The reactions are considered in terms of the three types of refractories evaluated: (a) silica refractories, (b) alumina refractories, and (c) fireclay refractories. Experimental Procedures The technique employed in the laboratory experiments involved immersion of refractory samples in a steel bath in air at temperatures up to 29700F (1630~C)o The bath consisted of AISI No, 1018 steel bar stock melted down in a sixty pound induction furnace. The final desired composition of the bath was obtained by making furnace additions of ferrosilicon, ferromanganese, and high carbon pig irono The steel composition was determined from analysis of pin samples taken at the beginning and at the end of each heato After the necessary alloy additions were made, refractory brick samples (0~5 x 0.5 inches cross-section) were immersed in the bath for five or ten minutes. Longer immersion times were avoided as a precaution against accumulation of excessive amount of reaction products. The test atmosphere was airo For some heats nitrogen was used to cover the melt, thereby reducing the oxidation of the molten batho Evaluation of the refractory-metal reaction zone was accomplished by metallographic examination of polished sectionso Results on Silica Refractory-Metal Reactions The attack of silica refractories in molten steel has been described by Van Vlack et al(41)It is concluded that there is no described by Van Vlack, et al, It is concluded that there is no

l72direct reaction between the metal and the silicao Rather, it is felt that the iron oxide formed from oxidation of the metal dissolves the silica to form an iron silicate liquido Figure 50 is a cross section of a portion of unattacked silica brick used in the immersion testso Only a limited quantity of liquid is visible in the photomicrograph. Figure 51 shows the reaction zone for identical silica brick after immersion in molten steel for ten minutes in air at 29700F (1630~C)o Comparison of these two figures shows that the reaction at the refractory-metal interface results in extensive silicate liquid formationo The silicate liquid has been absorbed into the refractory and completely surrounds the silica grainso This condensed phase combination is in accord with the equilibrium diagram of Figure 48 where melt and silica coexist in air at temperatures in excess of approximately 1450~Co Results on Alumina Refractory-Metal Reactions Prior work has shown that there is very little alteration of high alumina brick (60 to 70 percent alumina) by molten steelo Figure 52 illustrates the affected zone of high alumina refractory after immersion in a steel bath for ten minutes at 2830~F (15500C)0 The formation of the solid reaction product, hercynite, is evidento This is in accord with the equilibrium diagram for the system which indicates tha xtensive melt formation occurs only at more elevated temperatureso

-75Results on Fireclay Refractory-Metal Reactions The behavior of fireclay refractories of approximately 65 percent silica and 35 percent alumina when in contact with molten steel has been outlined by Van Vlack, et al, () The reaction products that form during the alteration include a highly fluid silicate liquid and mulliteo Figure 53 illustrates the reaction zone of fireclay refractory after immersion in molten steel for ten minutes in air at 2830 T (1555~C)o The microstructure is identical to that developed for the iron-kaolinite equilibrium studies under reduced atmosphereo The approximate composition of the altered reaction products can be indicated on the iron oxide-A1203-SiO2 equilibrium diagram. A microstructure of particular interest is shown in Figure 54~ From the equilibrium studies it should be remembered that corundum was a reaction product for the iron-mullite series only, not the iron-kaolinite serieso Likewise, it has been concluded that the alteration products for fireclay refractories contain corundum only if aluminum is dissolved in the steelo(4) Figure 54 illustrates this to be trueo Corundum is evident in the reaction zone of fireclay refractory immersed in molten steel containing 0o2% aluminum.

CHAPTER VI CONCLUSIONS The following conclusions may be drawn from this work: 1o The results from equilibration of iron and alumina accurately position the important four phase univariant boundaries in the equilibrium diagram for the iron-oxygen-hydrogen-aluminum system. 20 Solid hercynite (FeOoA1203) is the product from the reaction of iron and alumina at temperatures less than 1700~Co The equation for the free energy of the reaction Fe(5) + A1203 + H20(g) - FeO~Al120 + H2(g) (33) is: YF~ = -2100 - 17T (34) The equation for the reaction Fe(i) + A12~3 + H2~(g) v FeOoA1203 + H2( (35) is: AF~ = -5850 + 0037T (36) An oxide liquid is the reaction product at temperatures in excess of 1700~Co 36 The agreement of iron-silica results with previously published data confirms the existing iron-oxygen-hydrogen-silicon equilibrium diagram and substantiates the overall experimental techniques of this research0 -74

-754. For the range of temperatures investigated (1450~C to 1717 C) and approximate atmosphere conditions of PH 2/pH2 = 0.22, the equilibration of iron-mullite produces corundum and liquid as reaction products. A liquid phase forms from iron and mullite at lower temperatures and under more reducing conditions than for iron-silica equilibria. 5. For the range of temperatures investigated (1450~C to 1555~C) and approximate atmosphere conditions of PH20/PH2 = 0.22, the equilibration of iron-kaolinite produces mullite and liquid as reaction products. There was more extensive liquid phase formation in the iron-kaolinite reactions than in the iron-silica or iron-mullite reactionso 6. The results from equilibration of iron-mullite and iron-kaolinite can be interpreted with the help of the primary phase fields of the iron oxide-A1203-SiO2 equilibrium diagram. Corundum forms as a reaction product from iron and mullite, and mullite forms as a reaction product from iron and kaolinite. 7. The equilibrium results of this investigation can be correlated with engineering type refractory-metal reactions involving alumina, silica, and fireclay refractories with molten steelO

H 0 0 U> *1HI d d Cd C3 C Cd od Cd d Cd (aM U2 U U U U U U U Q d { U U P 41-HlH H rH H H- H H H H H-H HH *H,Q O 10 1I 0 110 1 IO*H I O I I*H O Prd H a) *-H a) *H a) a *H a) a) *H a) *H a) *H- ) *H ) a- *H o.H -~ H -PH - -P-H - - H -PH - r-P +-P H I - IH U)^4 IC)) ^Ico ^CQ ItQ a1 O a o aco aco-p co I-Pco 0) -' - ri -' - H'1'-'-.1'a-r-.H H H.- I II -*HI 0 -P 0o-P 0o0 -P 0o0-P - o O r-IH OO' 0Hr-! 0? H rA ~ | ^zH? H?0O kl0 _ir- 0 H 0 r- t U) 0 a a) - ) )O co Ci a) a H o' o E-i E-1 0 0 0 ue cd H i *H > > 2 (\ rd HH - 0 P 0 q 0 +VD P4 p Q.-H I C E 0 cJ -P Lh ~r (D OJ HM H ~ o C)o o 0rd H H u]q 0 HpQ o 0 0o0 0 C' H H H H pq H E g H V H ~ OH <H OCd p \ J - ON H L &3 ~0O ) -zi 1- M 1 K CX\ (' rz O E -U) 0 o o o0 a 0 0 0 0 0 0 0 r* Po\ 0 i-iO K ~ o LC ir\ \D ir\ Lt.uL CO E-I: -: P1 0 0 0 0 0 0 0 0 o0 LroLf L\ L\ Lr\ Lf Lr\ Lr Cd d Cd Cd CUJ C\j g a )( ci cOJ.CH -iH'l -r -*H! CiH C IH-H o o 3 r -I I c o3'H c4 r -I Is r-!'H UC *.HO *HO -H I0 *H I 0'HO *H 0 *H'H I' - H I *H 0 (D r-A *H ~ E *H) E^.,- 4.H *-* r-I q (D 0 -E -- o -1 *H r- H r- H 1H H r —H H.Hc H H-H r-H H H ci O O OC O Oz O O z I O O<CO O S O < O CO Q O CIO O co II I o I^A:0b1 I III I -I I >I oI 00 o 0 0? 0 0 o 0o0 000 0 0 0 00 HH HH HIHH H H HH HH HIH HJ H MH H Ha) OJ OJ OJ O H 0J OJ OJ O! Cc^ oI I I I I I I I V] Wi < < ~ < < < o l

-77oj (I (D I ) ~T a — P -tP U r-l ~l r. —! d g rd rd rdrd rd rd 0r * C l * *Hl * H *r-l'-I' pe'H * S H a a S S - a a.H a aaH a)H aH aH aH - PL Cd 9i Di c31 <v Cd <i Cd <1 e3 o a~ d'H I'-HI d d I'-PI d I-H 1 H -P H o o H *H *H *H a H H *ci H HH.H *H H H.H *, HH' H'H *H TH U12 *4U -)IF ~4 4H (~)~ (1 U- (D.H a) U-H Oa) r- IH H I I I IHH 1. 11 I IH H 4 I 4o I I - I - I) (1 H 4 1H H d a~ H 0- 0 0~) 0~'w 0- P Cq O 0 H O D H 0 OH EH H 3 H 4 H 4 H H *H'HO e.H -1H.H 2 1)H. aH a t a) a) a) ta H O O 0 0 H) H ~ -I H CH.r-I (1' HA 0:'0 a). (D - fto l0 0coI 0 I0-P Zr' H i kffi k HO O O) H''....rH O HH CM CQ H d H H 0 O O H * H 4 I P 0 P~i, r —- ^1 K r O E 0H n OC Oh O O O | H 0 H ( ciic' H ON. 0o\ -=- 0 CMJ vo c3 CM C\ H H ON 0 < 0 Ca U2 C Mj CM C J L o o-\ 0 o o, 0 o0 0 0 P4 0 0 0 0 0 0 0 0 rQ P1L(D 0 HC\j X, I uO{ Y I Lr\- 4- ( m- 4f\ Lf\ O n n P \ Hq r H H r H A H H H H H H H I H I I a) | g)rra CMl CM MoN Ir-A r-i 0 -P r-A 0 d -P r - -4-', ci a)o c )o c o')- iH U ci iO C _) H H0 *H H'H I * *H *H I L 0 *H H0 00 0 * H O H HO 0 U -PH HO O H SH H H z dH'^ H H H'rH c i H'H'H M c H*H H'H H'H H*H H'H ci <co <(1RW <;1U);4'mU) ~,U <'Ao <co ~: n v A o A Ari I U 17 R HHHH HMHH HMHHH HH HH HH HH HH o -. - L X O N0X 0X CL C ^I I I II I I! v)~ H; HH ~ m ~ I — H I — H H H H

-78csd *rl~H *r-H Hr- *rrC P d C n D d dd bi r d 4r o la I p H *HH * HH * *H *Hr-: H-I H r-I -I, -I E F4 4 H D?1 *H H H o cr-'-t' r-I i cI - *Hr Q I ~ 0 C0 1 *\ *rl * rl a * *H *H rl - *.r -l *Hl Hr'l *H O rH-i *H O-r *Hr pArd.H (a) aH (a.H~(-) ~1 ~H- 1 ci IH ri H. ~l -'Hk 4'H Hr 0oa) ~ ~H w -PHI 1 1 P H 3DH Ii H I 3)H H C) C ***H H *H * Hcarl H r* C-c i i *H H H *H Hc ra *H H H *H 2 0 a) I *H *H I *H *H I II I I *H H I I * I l Strd C) )o ) 0 cJ -P H i^'a14 F0' 4-)1 - H0 4 (D U zzH sr H H O rd H OOriH 00 rO 0H 000 *r-HO a)a) O a)a(HI iHv % <??i(a ) I-a)HH?f 8 V' > hI p a, ~ > H mq I H H M H H ~ H H ~< H H H rd Hl a) 0 \ I c rH CJ H oCMH *Hr C~J C-J I Il ci ci i m *H o *CO 0-p a a) a) a) CMI E- EP E- P b~o o P4 *H Hrq E~'H > Lt o\ r H Z C - d ^ I 0 0 0( 0 0 - H H H O O E lao1 0 0 0 H H H V 0 MH 0 H( d p r- CO H )L L n C V 4- 0 L^ L C J C O CO N = ~ O V): CU CM CM CU CM H O pq O E- v) e e * 0 o o Eq o C) 0OO 0 0 0 0 0 rn) P Lf' O-l L LF\ -t > 0 LC\ PLA \\ L(\ \ LC\ LF\ Lt\ 0) Q LT Lr\ Lr\ Lr i\ rI\ E-3 o *H *HH *H g < d g g d c -H ai a)*H c a)*H cir- Ci iH )'H ci O (D a ) *H I * - H *r- * - * - *H IC) l H I * * * - C) *H *HO *rl v) S O *(T *H i H H S*Hf-*i-H ( a)H *Hi-H *H H V O r O OOHHO O -r- O O O OO O O OO O O rH - -H i H H H H H H H HHHH HH H H HH H H H H H 0 0 00- 0 CO 0 O 0 H CM H.D, -l -i P CQ I' I I I I I 0. O P L P P!. _ Lx.

79rI * A *H H 0 d r H *H rI I > T\J*H r-.r-I c'DH d H'H' H 8)4 H'.*r ~ C I H, P l l aH *H 0H r. D IH r- H o dH H P 4 4 I I O (D ( KC Ea * ~ o na) 0 a I rl *O aSU.r? o o d'HO Oi a, a,, O~ O 0 4)( CPI *H1 IrOH "',0d' - cO H E-I H >a *rJ o o o 0 b L C OO O O H H H 0 PH' H m C +q P O CO e i OH C O O O> ON 0 o 0 0 0 0 U EH 0 O (D l H V O O0 00 0 H H H H H H H a)D H

-80TABLE IV HEAT RECORD FOR REFRACTORY - METAL REACTIONS Steel Analysis Aim Temperature Immersion Time (%) (%) Heat No. ~C OF (Minutes) C Si Mn Al Al F-42 1550 2830 10 0.38 0.36 0.37 - F-43 1630 2970 10 0.36 0.38 0.42 0.2

-81I I I I I I 0.4- H2'AIR 0 ALUMINA 0 + HERCYNITE LIQUID + + o GAS o ALUMINA 0 1~ + 0.3- GAS 0~ 0 oI Q-1 I ---- \E LIQUID IRON +. 0.2 - cN SOLID IRON ALUMINA as + + ALUMINA GAS GAS 0 OXIDIZING 0.1 - El REDUCING + CALCULATED OS0~~~ I PILLAY, FLORIDIS 1450 1500 1550 1600 1650 1700 1750 TEMPERATURE (~C) Figure 46. Equilibrium Diagram for the Iron-OxygenHydrogen-Aluminum System.

-821.5 0 Q OW I.4 - Cl 1.3 II 0 DATA + BY CALCULATION 52 5.4 5.6 5.8 ( xl4) I / TEMPERATURE (~K)Figure 47. Logarithm of the Chemical Equilibrium Constant versus Reciprocal Absolute Temperature for the Iron-OxygenHydrogen-Aluminum System.

-850.4 0 O SILICA + LIQUID + GAS \ 0 0.3 0 o o I\p O ester 0.2 SILICA + SOLID IRON + GAS SILICA + LIQUID IRON + GAS 0.1 + 0 IRON SILICATE LIQUID STABLE I+ IRON STABLE 1400 1500 1600 TEMPERATURE (t) Figure 48. The Equilibrium Diagram for the Ison-Oxygen-TydrogenSilicon System After Colligan. 15)

-b4I 0 "'""-i / I i~,~Z u^ ^^-,' 2F.OSiQao$K1^ 3F0 ^r "0 y^ RUNDU 3A1A2SiQ,'F0 >0 20'0 40o FeO.AI20,. 8.') - ALO2i WT.% Figure 49. Equilibrium Diagram for the Iron Oxide-A1203-SiO2 System After Schairer and Yagi. (30)

-85Figure 50. Microstructure of Unattacked Silica Refractory. 150X Figure 51. Microstructure of Silica Refractory from Heat Number F-43. 150X

-86Figure 52. Microstructure of Alumina Refractory from Heat Number F-42. 150X Figure 52. Microstructure of Fireclay Refractory from Heat Number F-42. 150X Figure 54. Microstructure of Fireclay Refractory from Heat Number F-43. 150X

-87CRUSILITE RODS ALUMINA BRICK,\~ ~,~L. _/~ ~ INSULATION \FURNACE TUBE GAS EXIT / \ \\\ GAS ENTRANCE HOT ZONE TOP VIEW Figure 55. Scale Drawing of the Laboratory Equipment Corporation Furnace 8"

APPENDIX I CALCULATION ON THE UNIVARIANT EQUILIBRIUM IRON-ALUMINA-BERCYNITE-GAS From the data of Schenk, et al. (21) Richards and White(5) determined that the iron-alumina-hercynite-gas univariant equilibrium existed at 4.0 and 5.4 percent CO2 for 800~C and 900~C, respectively. With the remaining percentage of gas being CO, the PCO /PCO can be calculated as 0.0417 and 0.0571 for each temperature. Gas reactions involving carbon, oxygen, hydrogen and water vapor are related by the "water gas reaction." CO2 + H2 - CO + H20 (37) Equilibrium data for this reaction is available in the literature,(36) log K = -1395 + 1360 (38) T Also, from thermodynamic considerations log K = log P2 - log (PCO2 (39) H2 CO For 800~C and 900~C Equations (38) and (39) are combined with the known C02:CO ratios to give log (PH20/PH2) values of -1.320 and -1o073, respectively. Taking the antilogarithm of these values results in water vapor:hydrogen ratios of 0.048 and 0.085 for the univariant boundary at 800~C and 900~C. -88

APPENDIX II THE LABORATORY FURNACE The furnace used in the equilibration studies is Laboratory Equipment Corporation Model 2600, A detailed drawing of the furnace is shown in Figure 55, The heat source is four Crusilite Type X silicon carbide heating elements. Dimensions of the elements include: (a) an effective heating length of 10 inches (b) a diameter of 3/4 inches and (c) an overall length of 25 inches. Nominal resistance is about 108 ohms, Although the maximum operating temperature recommended for the Crusilite elements is 15750~C equilibration runs at temperatures in excess of 1700~C were achieved. Careful installation of the rods together with a slow rate of increase of temperature help significantly in obtaining these temperatures. Element life at 17000C was, however, limited to about thirty hours. Power is supplied to the elements from a 5 KW auto transformer with single phase connection, A divided voltage control provides voltage independently to the upper and lower sets of elements in incremental steps of two volts, With such control, equal amperage is easily maintained on both sets of elements, thereby increasing element life significantly, The available secondary voltage ranges from 24 to 96 volts, The furnace is equipped with Triangle RR alumina tubes or Triangle H5 mullite tubes. Nominal dimensions of the tubes are 1 1/4 inches outer diameter, 1 inch inner diameter and 30 inches overall length. -89

APPENDIX III APPLICATION OF THE PHASE RULE An aid to understanding the systems studied in this investigation is the application of phase rule principles outlined by Caseo(37) The technique is a rigorous one and uses the following terms: N = The number of chemical individuals or substances represented by a distinct chemical formula. G = The number of independent distributions of the same individual between a pair of phaseso E = The additional relations between concentration variables, ieo, mass action equilibria0 U = The number of variables, i0eo, one less concentration variable than the number of individuals in each phase and the additional variable, temperature, For the iron-oxygen-hydrogen-aluminum equilibrium diagram, the following three important points are analyzed: (a) a bivariant area under oxidizing conditions, (b) a bivariant area under reducing conditions, and (c) a univariant phase field boundary, (a) A Bivariant Area Under Oxidizing Conditionso System: H2(g), H20(g); A1203(s); FeOAl1203(s) -90

-91P = 3 gas, solid alumina, solid hercynite N = 4 H2, H20, A1203, FeO G = 1 A1203(s) ~WA1203(hercynite) E = O U = 3 2-1=1 (gas); 2-1=1 (hercynite) plus 1 (temperature) C = N- E C = 4 c=4 Vp = Up - G - E Vp = C + 1 - P = 2 V 2 P P For this particular analysis, the phase rule variables are temperature, hercynite composition and gas compositiono In the experimental studies the system was fixed by controlling both temperature and gas compositiono (b) A Bivariant Area Under Reducing Conditions. System: H2(g), H20(g); Al203(s),H in Fe(s) P = 3 gas, solid alumina, solid iron N = 4 H2, H20, A1203, Fe G = 1 H2 2H (g) E = 0 U = 3 2-1=1 (gas); 2-1=1 (solid iron) plus 1 (temperature) C = N - E C = 4

-92V = U - G - E V = C+ 1-P V = 2 V = 2 In this particular situation the phase rule variables are temperature, solid iron composition and gas composition. For the experimental investigation the system was fixed by controlling the temperature and gas compositiono (c) A Univariant Phase Boundaryo System: H2(g) 2(g); A1203s) FeOoAl2o0; H in Fe 2 3'^ (s)'111'^(s) P = 4 gas, solid alumina, solid hercynite, solid iron N = 5 XH2 H20, A1203, FeO, Fe G = 2 A1203 1203(hercynite) H2/ _ 2H (g) E = 1 Fe(8) + A1203(s) HO(g) (FeOAlO() + H2g) U = 4 2-1=1 (gas); 2-1=1 (hercynite) 2-1=1 (iron); plus 1 (temperature) C = - E C = 4 V = U - G E V = C + 1 - P P P P V = 1 V = 1 P P For the phase boundary analysis the phase rule variables are temperature, iron composition, hercynite composition and gas compositiono With the variance equal to one, the system is fixed by controling either the temperature or gas composition0

APPENDIX IV CALCULATION OF WATER VAPOR:HYDROGEN RATIO FROM OPERATING DATA The following is a sample calculation for determination of pH2 /PH2 from operating data using two different techniques involving total pressure and weight gain results, respectively~ Operating Data Experimental number D-48 Average saturator temperature 58o33~C (corrected) Atmospheric pressure 73553 mm Hg Atmospheric pressure corrected to 0~C 732.3 mm Hg Total pressure 73706 mm Hg Temperature of water in volumetric flask 28.0~C Room temperature 2820~C Gas sample analysis (H2) 99o90% Gain in weight of Nesbitt absorption bulb 0.1578 gmso By determining the total pressure with only hydrogen and water vapor in the system, the following equation is applicable: PH20 PH20 PH2 PT -PH20 From steam tables the vapor pressure of water in the saturator at 58.33~C is 138.22 mm Hg. -93

PH20 138.22 PH2 737.6 - 138.22 PH20 PH = 00231 PH2 Using the gain in weight technique, the following equation is applicable: PH2O Wto 22 414 - 1 x = - PJH 18 V(ST) After the hydrogen passed through the Nesbitt absorption bottle, it was collected over water in a 1000 mlo volumetric flask, PDry 273 VolH = VolH x 760 X T 2(STP) 2 760 T Dry Patm(O~C) - V0P H20 (28~C) PDry = 7323. - 28.4 Py = 70359 mm Hg Vol = (999) (I7 501(.2 ) H2(STP) 760 301o2 VolH = 838.69 ml. 2(STP) PH20 0O.1578 22,414 PH2 18 838.69 PH0 H20 0.234 The percentage difference between the two values is: 0.234 - 0.231 Percentage difference = 100 0.2-4 Percentage difference = 153%

APPENDIX V NOMENCLATURE Throughout the text of this report the following terms were used: Alumina A1203 Mullite 3A1203 2Si02 Kaolinite 2A12Si205( OH)4 Hercynite FeOoA1203 Quartz SiO2 stable form to 870~C Tridymite SiO2 stable form from 870~C to 1470~C Iron silicate liquid x (FeO) y (SiO2) The following code is used to identify the phases shown in the photomicrographs: Symbol Phase B Bakelite C Corundum (alumina) H Hercynite I Iron L Liquid M Mullite P Palmer Cement aQ Quartz T Tridymite DK Decomposed kaolinite CRI Cristobalite -95

REFERENCES lo Muan, A., "Phase Equilibria at High Temperatures in Oxide Systems Involving Changes in Oxidation States," Am. Jouro Science, 256, No. 3 (1958) 1710 2o Darken, Lo So, and Ro Wo Gurry, "The System Iron-Oxygen. Io The Wustite Field and Related Equilibria," Journal Amo Chemo Soco 67, No. 8 (1945) 13980 30 Darken, Lo So, and Ro Wo Gurry, "The System Iron-Oxygeno II. Equilibrium and Thermodynamics of Liquid Oxide and Other Phases," Journal Amo Chemo Soco 68, No. 5 (1946) 7980 4, Philips, B., and Ao Muan, "Stability Relations of Iron Oxides: Phase Equilibria in the System Fe304-Fe203 at Oxygen Pressures up to 45 Atmospheres," Journal Phys. Chemo 1961. 5. Richards, Ro G., and J. White, "Phase Relations of Iron-Oxide Containing Spinels: I, Relations in the System Fe-Al-O," Trans. Brito Ceramo Soco, 53 (1954) 259. 60 Greig, Jo Wo, Eo Posnjak, Ho E. Merwin and R. B. Sosman, Am. Jouro Science, 30 (1935) 239~ 7. Bowen, N. L., and J. W. Greig, "The A1203-Si02 System," Journal Amo Ceramo Soco, 7, No, 4 (1924) 2380 80 Budnikov, P. P., and 0. P. Mchedlov-Petrosyam, "Theory of the Transformation of Kaolinite on Heating," Journal Amo Ceramo Soco Ceramo Absto, 42, No, 5 (1959) 145fo 9. Barta, Ro, and Cho Barta, "Study of the System Al 0 -SiO " Journal Amo Ceramo Soco, Ceramo Abst., 42, No. 2 t159) 67co 10o Aramaki, Shigeo and Ro Roy, "Revised Phase Diagram for the System Al 0 -SiO," Journal Am. Ceramo Soco, 45, No- 5 (1962) 229. 11. Bowen, No A., and J. F. Schairer, "The System FeO-SiO2 Amo Jouro Scio 24, (1932) 177o 12o Darken, Lo S., and Ro Wo Gurry, "The System FeO-Fe203-Si02" Journal Am. Ceramo Soco, 72, (1950) 3910. 135 Muan, Ao, "Phase Equilibria in the System Fe0-Fe203-Si02" Trans. AIME, 203 (1955) 9710 -96

-9714o Darken, Lo' So, "Melting Points of Iron Oxides on Silica, Phase Equilibrium in the System Fe-Si-0 as a Function of Gas Composition and Temperature," Journal Am. Chem. Soco, 70, (1948) 2046. 15o Colligan, Go A., "Iron Silica Sand Interface Reactions," PhoD. Thesis, Horace Ho Rackham Schhol of Graduate Study, The University of Michigan, Ann Arbor, Michigan, 1959. 16. Hansen, Wo C., and Lo To Brownmiller, Am. Jouro Science, 15, (1928) 2250 17o McIntosh, A. B., JO R. Rait and R. Hay, Jour. Royo Techno Collo Glasgow, 4 (1937) 720 18. Fisher, W. Ao, and A. Hoffman, Archivo Eisenhuttenw, 27, No 5, (1956) 343o 19. Muan, A., and Co L. Gee, "Phase Equilibrium Studies in the System Iron Oxide-A1203 in Air and at 1 Atmosphere Oxygen Pressure," Journal Am. Ceram, SocO, 39, No. 6 (1956) 2070 20. Gokcen, No A., and J, Chipman, "Aluminum-Oxygen Equilibrium in Liquid Iron," Trans. AIME, 197 (1953) 1735 21o Schenk, R., Ho Franz and Ho Willeke, Z. anorg. Chem., 184, No. 1, 1929. 22, Fisher, W. Ao, and A. Hoffman, Archiv. Eisenhuttenw,, 26, Noo 1 (1955) 430 235 Fisher, Wo A., and Go Lorenz, "Entwicklung eines Kalorimeters fur Temperaturen bis 1100~C und Bestimmung der Reaktion-enthalpi des Spinells Fe0oA1203 zwischen 8000C und 1000~C," Archiv. Eisenhuttenw, 27, Noo 6 (1956) 375, 24. Pillay, To Co Mo, Jo D'Entremont and Jo Chipman, "Stability of Hercynite at High Temperatures," Journal Am. Ceramo Soco,.43, No. 11 (1960) 5830 25. Floridis, To P., and JO Chipman, "Activity of Oxygen in Liquid Iron Alloys," Transo AIME, 212 (1958) 549. 26. Thurner, Co Ro "'Activity of Oxygen in Liquid Iron-Sulfur Alloys," MoSo Thesis, Massachusetts Institute of Technology, 1958, 27. Pehlke, Ro Do, "A Thermodynamic and Kinetic Study of the Solubility of Nitrogen in Liquid Iron Alloys," PhoDo Thesis, Massachusetts Institute of Technology, 1960,

-9828. Taylor, C, R,, and J, Chipman, "Equilibria of Liquid Iron and Simple Basic and Acid Slags in a Rotating Induction Furnace," AIME Tech. Pubo No. 1499 (1942) 180 29. Snow, Ro Bo, and W. Jo McCaughey, "Equilibrium Studies in the System FeO-A1203-Si02," Journal Am, Ceram. Soc., 25, No. 6 (1942) 151o 30. Schairer, JO F,, and K. Yagi, "The System FeO-A1203-Si02" Am, Jouro Scio, Bowen Volume, Part 2 (1952) 4710 31, Muan, A., "Phase Equilibria at Liquidus Temperatures in the System Iron Oxide-A120-SiO2 in Air Atmosphere," Journal Am. Ceramo SocO, 40, No. 4 (1957)121. 32. Brindley, Go W., and Mo Nakahira, "The Kaolinite-Mullite Reaction Series: I, A Survey of Outstanding Problems," Journal Am. Ceram. Soc., 42, No, 7 (1959) 311. 330 Brindley, G. Wo, and Mo Nakahira, "The Kaolinite-Mullite Reaction Series: II, Metakaolin, Journal Amo Ceramo Soc. 42, No, 7 (1959) 314o 34. Brindley, Go W., and M. Nakahira, "The Kaolinite-Mullite Reaction Series: III, The High Temperature Phases," Journal Am. Ceramo Soc., 42, No, 7 (1959) 319. 355 Comer, Jo JO "Electron Microscope Studies of Mullite Development in Fired Kaolinites," Journal Am. Ceramo Soco, 43, No, 7 (1960) 378. 36. Basic Open Hearth Steelmaking, AIME, Edwards Brothers, Inc.o Ann Arbor, Michigan, 1951. 37. Case, L. 0., Elements of the Phase Rule, The Edwards Letter Shop, Ann Arbor, Michigan, 1939. 38. Kerr, Po Fo, P. K. Hamilton, and Ro Jo Pill, "Analytical Data on Reference Clay Materials," American Petroleum Institute, Project No, 49, 1950. 39. Mallett, M, W., "Purification of Argon," Industrial and Engineering Chemistry, 42, No. 10 (1950) 2095. 40. Flinn, R. Ao, and L, Ho Van Vlack, "Deoxidation Defects in Steel Castings," Trans. AFS, 67 (1959) 295.

-9941o Van Vlack, L. Ho, R. A, Flinn and G, A, Colligan, "Nonmetallic Macroinclusion Causes in Steel Castings Deoxidized with Aluminum, Part I," Trans. AFS, 68 (1960) 132. 42o Van Vlack, L. Ho, J. E, Brokloff, and Ro A. Flinn, "Refractories as Sources of Macroinclusions, An Evaluation," Trans, AFS, 69, (1961) 198.

UNIVERSITY OF MICHIGAN 3 9015 02523 0734