Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

The origin of inclusions in the electroslag remelting process Bell, Mark Moshe 1971

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Notice for Google Chrome users:
If you are having trouble viewing or searching the PDF with Google Chrome, please download it here instead.

Item Metadata

Download

Media
831-UBC_1974_A7 B44.pdf [ 8.02MB ]
Metadata
JSON: 831-1.0078782.json
JSON-LD: 831-1.0078782-ld.json
RDF/XML (Pretty): 831-1.0078782-rdf.xml
RDF/JSON: 831-1.0078782-rdf.json
Turtle: 831-1.0078782-turtle.txt
N-Triples: 831-1.0078782-rdf-ntriples.txt
Original Record: 831-1.0078782-source.json
Full Text
831-1.0078782-fulltext.txt
Citation
831-1.0078782.ris

Full Text

THE ORIGIN OF INCLUSIONS IN THE ELECTROSLAG REMELTING PROCESS by MARK MOSHE BELL B.A.Sc. Un i v e r s i t y of B r i t i s h Columbia, 1969 A THESIS SUBMITTED IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF APPLIED SCIENCE i n the Department of METALLURGY We accept t h i s thesis as conforming to the required standard THE UNIVERSITY OF BRITISH COLUMBIA July, 1971 In p r e s e n t i n g t h i s t h e s i s in p a r t i a l f u l f i l m e n t o f the r e q u i r e m e n t s f o r an advanced degree at the U n i v e r s i t y of B r i t i s h C o l u m b i a , I a g r e e that: the L i b r a r y s h a l l make i t f r e e l y a v a i l a b l e f o r r e f e r e n c e and s t u d y . I f u r t h e r agree t h a t p e r m i s s i o n f o r e x t e n s i v e c o p y i n g o f t h i s t h e s i s f o r s c h o l a r l y purposes may be g r a n t e d by the Head o f my Department o r by h i s r e p r e s e n t a t i v e s . I t i s u n d e r s t o o d t h a t c o p y i n g o r p u b l i c a t i o n o f t h i s t h e s i s f o r f i n a n c i a l g a i n s h a l l not be a l l o w e d w i t h o u t my w r i t t e n p e r m i s s i o n . The U n i v e r s i t y o f B r i t i s h Columbia Vancouver 8, Canada Depa r tment i i . ABSTRACT The aim of this work was to determine whether (ASTM E-45) D-type inclusions are a necessary component of ESR ingots. Therefore, experi-ments were undertaken to ascertain the origin of the inclusions. Pure iron (FVE) and electrolytic nickel were melted through CaF2+CaO+Al20^ slags, and the changes in the metal compositions (with respect to calcium, aluminum and oxygen)were measured. The composition changes could not be accounted for by simple chemical slag/metal reac-tions. Hence, electrochemical slag/metal reactions were inferred. These reactions were detected, and the accompanying re c t i f i c a t i o n of the 60 Hz A.C. melting current was measured. It was thus concluded that both chemical and electrochemical slag metal reactions contribute to the f i n a l composition of the metal. The cooling and s o l i d i f i c a t i o n paths of the various metal compo-sitions were then studied. With the use of the classical nucleation theory i t was found that the metal compositions after reacting with the slag, provide sufficient supersaturation for inclusion nucleation at the latter stages of s o l i d i f i c a t i o n . As in the latter stages of s o l i d i f i c a t i o n l i t t l e time and material is available for inclusion growth i t was con-cluded that small inclusions are to be expected in the ESR s o l i d i f i c a t i o n mode. i i i . Since the chemical and electrochemical reactions are an inherent part of the electroslag process, we are able to conclude that small globular oxide inclusions are a necessary result of the normal method of electroslag melting. iv. TABLE OF CONTENTS Page TITLE PAGE i ABSTRACT i i TABLE OF CONTENTS l v LIST OF FIGURES v i i i LIST OF TABLES x i LIST OF SYMBOLS x i i i CHAPTER I. INTRODUCTION 1 CHAPTER II. THEORETICAL CONSIDERATIONS 3 2.1 Oxygen in Steel 3 2.2 Deoxidation Thermochemistry.... 7 2.3 Thermochemistry of Slag/Metal Reactions 12 2.4 Thermodynamics of Slag Systems Used 14 2.4.1 Binary CaF 2 + A l ^ 14 2.4.2 Ternary CaF 2 + A l ^ + CaO 17 2.5 Chemical Interaction Between Liquid Iron and CaF 2 + CaO + A l ^ Slag 24 2.5.1 Justification for Considering Iron/Al 20.j Interaction Only 24 2.5.2 Calculated value of the Equilibrium Constant 26 2.5.3 Experimental Value 27b 2.5.4 Deviation from Henrys' Law 28 V. Page 2.5.5 Prediction of Extent of the A l ^ / L i q u i d Iron Interaction at Equilibrxum. 34 2.5.5.1 At Unit A l , ^ Activity 34 2.5.5.2 At Variable A l ^ Activity 38 2.5.6 Possibility of Secondary Reactions 40 2.5.6.1 Reactions Causing Excess Aluminum........ 40 2.5.6.2 Reactions Causing Excess Oxygen 43 2.6 Precipitation from Fe-O-Al alloys 45 2.6.1 Alumina Precipitation 47 2.6.2 Hercynite Precipitation 48 2.6.3 Nucleation Considerations • • • • 50 2.7 Chemical Interaction Between Liquid Nickel and CaF 2 + CaO + A120 slag 51 2.7.1 Thermodynamic Consideration of Liquid Nickel/A^O^ Interaction 53 2.7.2 Prediction as to the Extent of Liquid Nickel/Al 20 3 Interaction 56 2.7.3 Thermodynamics of CaO/liquid Nickel Interaction 57 2.7.4 Thermodynamics of CaO + A l ^ ^ / L i q u i d Nickel Interaction CHAPTER III. EXPERIMENTAL APPARATUS, PROCEDURE, AND RESULTS 65 3.1 Melting Apparatus-Electroslag Furnace 65 3.2 Melting Experiments 65 3.2.1 CaF 2 + CaO + Si0 2 Slag .Experiments 67 3.2.2 CaF 2 + CaO + A l ^ 70 v i . Page 3.2.3 Aluminum Deoxidation Experiments 73 3.2.4 Temperature Measuring Experiments 76 3.2.5 Rectification Study Experiments 77 3.3 Ingot Analysis 83 3.3.1 Analysis Procedure 83 3.3.2 Total Aluminum Analysis 87 3.3.3 Total Oxygen Analysis 89 3.3.4 Inclusion Analysis 89: 3.3.4.1 Optical Analysis 90 3.3.4.2 Electron Micro-probe Analysis 90 3.3.4.3 Inclusion Extraction 92 3.3.4.4 X-ray Diffraction Analysis 95 3.3.4.5 Atomic Absorption Analysis for Inclusions' Fe Content 95 CHAPTER IV. DISCUSSION 98 4.1 Slag/Metal Reactions T 98 4.1.1 Analysis of Results in Terms of Chemical Thermodynamics 98 4.1.1.1 Iron/Slag 98 4.1.1.1.1 Introduction 98 4.1.1.1.2 Variable A1 20 3 Activity 100 4.1.1.1.3 Deoxidation Experiments 102 4.1.1.2 Nickel/Slag Interaction 104 4.1.1.2.1 Nickel/Al 20 3 + CaO 104 4.1.2 Possibility of Higher Slag/Metal Temperatures and the Operation of Electrochemical Reactions 106 v i i . Page 4.1.3 Possibility of Faradaic Deposition 106 4.1.4 Analysis of Rectification Results I l l 4.1.4.1 Rectification in A l ^ + CaF 2 Slag I l l 4.1.4.2 Rectification in CaF 2 + CaO + ^-2°3 116 4.1.4.3 Effect of Electrode/Ingot Size Ratio 118 4.2 Precipitation 119 4.2.1 Precipitation from Fe-O-Al Compositions 119 4.2.1.1 Analysis i n Terms of Stoichiometry 120 4.2.1.2 Analysis in Terms of Equilibrium Thermodynamics 121 4.2.1.3 Analysis in Terms of Supersaturation Needed for Nucleation 123 4.2.1.3.1 Supersaturation by Addition and Cooling 124 4.2.1.3.2 Supersaturation During Solidi f i c a t i o n 126 4.2.1.3.2.1 Introduction 126 4.2.1.3.2.2 Comparison with Analytical Results. 128 4.2.2 Precipitation from Ni-0-Ca-Al Compositions 129 CHAPTER V. CONCLUSION 132 BIBLIOGRAPHY. 134 APPENDIX 1 137 APPENDIX II ' 139 APPENDIX III 140 APPENDIX IV 142 APPENDIX V 146 v i i i . LIST OF FIGURES Page1 Figure 1. Temperature-Composition diagram for the system Fe-0 at 1 atm. 4 Figure 2. C-0 equilibrium i n steel furnaces 6 Figure 3. Alumina and S i l i c a equilibrium lines in l i q u i d iron 11 Figure 4. ^a-^ 2 + "^ "2^ 3 k i n a r v phase diagram 16 Figure 5. CaO + A^O^ binary phase diagram 18 Figure 6. CaF 2 +CaO binary phase diagram 20 Figure 7. CaO and A1 20 3 a c t i v i t i e s at 1600°C 23 Figure 8. A1„0„ activity at low concentrations in CaO 25 Figure 9. Aluminum and oxygen Henrian act i v i t i e s in l i q u i d iron at unit alumina activity (1600°C) 29 Figure 10. Wt.% Aluminum and oxygen in liq u i d iron at unit A1 20 3 activity 32 Figure 11. Wt.% Aluminum and oxygen in liq u i d iron in equilibrium with alumina at various temperatures 35 Figure 12. Wt.% Aluminum and oxygen in liq u i d iron i n equilibration with alumina of various ac t i v i t i e s 39 Figure 13. FeO + A1 20 3 binary phase diagram 46 Figure 14. A1 20 3 and hercynite equilibrium lines in liquid iron 49 Figure 15. Effect of interfacial- tension on nucleation of oxides in Fe-O-Al system, 1538°C. 52 ix. L i s t of Figures (cont'd) Page Figure 16. Wt.% Aluminum and oxygen in l i q u i d nickel in equilibrium with alumina 55 Figure 17. Wt.% Calcium and oxygen and aluminum and oxygen in equilibrium with CaO or A1 20 3 60 Figure 18. Extent of reaction 2.7.XI 63 Figure 19. Argon Slag Cap 66 Figure 20. CaO + S i 0 2 binary phase diagram 68 Figure 21. Ca 2SiO^ + CaF2.pseudobinary phase diagram 69 Figure 22. Aluminum addition to iron electrodes 74 Figure 23. Temperature measuring probe 78 Figure 24. Rectification measuring cir c u i t 80 Figure 25. Effect of arcing to the mold on melting current 84 Figure 26. Copper deposited on the slag due to arcing 85 Figure 27. Sectioning of an ingot s l i c e for analysis 86 Figure 28. Extraction apparatus 93 Figure 29. Polarization curve for the system Fe/CaF 2+Al 20 3 108 Figure 30. Current and voltage wave form in A.C 109 Figure 31. Occurrence of both chemical and electro-chemical slag/metal interactions 113 Figure 32. Composite nucleation-segregation diagram during s o l i d i f i c a t i o n in Fe-0-Al system 127 X . L i s t of Figures (cont'd) Page Figure 33. Inclusion Alignment 130 Figure A.I.I. Oxygen deposited (by reaction 1) dependence on the amount of liqu i d iron interacting with the slag 1 3 8 b Figure A.IV.l. Inclusions in the iron matrix 142 Figure A.IV.2. Aluminum and oxygen in the inclusions 143 Figure A.IV.3. X-ray powder pattern of the extracted inclusions 144 x i . LIST OF TABLES Table 1. Table 2. Table 3. Table 4. Table 5. Table 6. Table 7. Table 8. Table 9. Table 10. Table 11. Table 12. Table 13. Table 14. Table 15. Page Values of log K' 27 Predicted Aluminum and Oxygen Deposited by A1 20 3 72 Analytical Results of Total Aluminum and Oxygen Contents (ingots 1-6) . 72 Predicted Calcium and Oxygen Deposited by CaO Dissociation (ingots 5-7) ........ 73 Predicted I n i t i a l Aluminum and Oxygen Contents (ingots 8-11) 75 Predicted Final Aluminum and Oxygen Contents (ingots 8-11) 75 Analytical Results of Total Aluminum and Oxygen Contents (ingots 8-11) 76 Temperature Results 76 D.C. Current Measured in Various Slags 81 D.C. Current Measured Using Various Ingot Diameters 81 Optical Observation of Inclusions 91 Electron-Microprobe Analysis of Inclu-sions 94 X-ray Diffraction Analysis of Inclusions 96 Atomic Absorption Analysis for the Determination of the Inclusions' Iron Content 96 Prediction of Aluminum and Oxygen Con-tents by Electrochemical Reactions, and by a Combination of Both Chemical and Electrochemical Slag/metal Reactions (ingots 1, 2 and 5) 115 x i i . L i s t of Tables (cont'd) Page Table 16. Prediction of Aluminum and Oxygen Con-tents by Electrochemical Reactions, and by a Combination of Both Chemical and Electrochemical Slag/Metal Reactions (ingots 3,4 and 6) . 117 x i i i . LIST OF SYMBOLS a = Raoultian activity A = Constant depending on the complexity of the molecular species forming and on the charac-ter of the means by which atoms and molecules are transported to and from the site where the 25 new interface is to form ( A = 1 0 for hercy-3 0 nite and 1 0 for FeO). A * = A adsorbed A ° = A i n neutral state A ^ = A in gas phase A Q J = A in l i q u i d phase A , N = A in solid phase (s) [ A ] . , = A dissolved in M M ( A ) = A dissolved in the slag [ A ] , ^ a, . w hypothetical 1 wt.% Henrian activity of A in M 1 wt.% m M = 3 V J ( A ) = concentration of A in supersaturated solution s • s • ( A ) = concentration of A in equilibrated solution eq C^^ = concentration of an element in the residual l i q u i d C Q = concentration of an element in the l i q u i d i n i t i a l l y F = Faraday's constant f. = Henrian activity coefficient of A A f = fraction s o l i d i f i e d s xiv. Henrian activity of A Boltzmann's constant equilibrium constant i n terms of Raoultian a c t i v i t i e s in the metal and the slag and p a r t i a l pressure in the gas. Henrian a c t i v i t i e s in the metal, Raoultian a c t i v i t i e s in the slag, and p a r t i a l pressure in the gas. wt.% in the metal, Raoultian a c t i v i t i e s in the slag, and partial pressure in the gas deoxidation constant in terms of wt.% in the metal segregration coefficient mole fraction of A part i a l pressure of A activity product of the supersaturated liquid temperature in degrees Kelvin molar volume frequency alternate current amperes atmosphere coulombs direct current electroslag remelting Ferrovac-E melting point molecular weight parts per million XV. sec = second VAR = vacuum arc remelting wt.% = weight percent AF° = standard free energy change for reaction A in cal per mole of reactants cr i t ^hom = free energy required for the homogeneous nucleation 3 of X nuclei per sec per cm Y° = Raoultian activity coefficient of A a A _ R = int e r f a c i a l tension of phases A and B x v i . ACKNOWLEDGEMENT I wish to thank my research director, Dr. A. Mitchell, for his enthusiastic guidance and encouragement. I also extend my gratitude to my fellow graduate students, staff and faculty members of the Metallurgy Department for their advice, to the workshop staff for their service, and to a l l others who assisted with this work. I gratefully acknowledge financial assistance from the International Nickel Company of Canada. x v i i . Alumina i n c l u s i o n s on the f r a c t u r e d s urface of an ESR s t e e l . (x3000) CHAPTER I INTRODUCTION The deleterious role that inclusions play in the mechanical behaviour of metals i s well acknowledged. It is found that in many instances inclusions reduce the impact and fatigue properties of metals. The degree to which inclusions w i l l reduce the mechanical properties of the parent metal depends inter a l i a upon both the size and the composi-tion of the inclusions. A good deal of progress has been made in reducing the inclusion size distribution in the metal. Better understanding of inclusion floatation and the development of processes such as VAR and ESR has made that progress possible. It i s an acknowledged practical fact, however, that inclusions cannot be entirely removed from the metal, thus leaving behind a clean metal matrix. ESR ingots are known to contain characteristically a typical type of inclusions, classified D-type, according to the ASTM E45 standards. They are usually evenly distributed globular oxides and do not exceed 12 um in diameter. L i t t l e is known about the effect of inclusion composition on the mechanical properties of the metal.' It is generally accepted, however, that small spherical oxides (D-type) are potentially harmful in many 2. areas particularly i n ro l l i n g contact fatigue. The f i r s t aim of this work was to define whether Inclusions are indeed a necessary component in ESR ingots arising in slag/metal contact during processing. This was done by determining the origin of inclusions in the systems: I. Fe melted through CaF2+Al203+CaO slags II. Ni melted through CaF2+Al203+CaO slags The second aim of this work was to study the effect of the system used on the composition of the inclusions precipitated. In this context, the precipitation path followed by Fe-O-Al alloys produced by melting through CaF^+Al^O^+CaO slags was studied. 3. CHAPTER II THEORETICAL CONSIDERATIONS 2.1 Oxygen i n S t e e l At s t e e l making temperatures the s o l u b i l i t y of oxygen i n pure i r o n i s very high."'" This i s shown i n Figure 1 where the s o l u b i l i t y l i m i t as a function of temperature i s i n d i c a t e d as l i n e B-B'. At 1600°c for example, the s a t u r a t i o n s o l u b i l i t y of oxygen before l i q u i d i r o n oxide forms i s approximately 0.23 wt.%. In actual s t e e l making p r a c t i c e , however, the oxygen content of the i r o n i s c o n t r o l l e d by and r e l a t e d to the carbon-oxygen e q u i l i b r i u m as determined from re a c t i o n 2.1.1. CO, v = [ C ] n , „. . „ + [0], t „. . ^ 2.1.1 (g) L J 1 wt.% i n Fe 1 J l wt.% i n Fe 2 Chipman and E l l i o t reviewed the work done on the e q u i l i b r i u m and showed that: l o g K = 1168/T - 2.07 c. • JL • X where the e q u i l i b r i u m constant IC, ^ ^ i s given by: 2.1.1 P CO OXYGEN (Wt. %) Figure 1. Temperature-composition diagram for the system Fe-0 at 1 atm.^ 5. Using K« . T at P - 1 atm, and the appropriate interaction coefficients, Chipman and E l l i o t were able to plot the equilibrium values of carbon and oxygen dissolved in the iron (Figure 2). The actual value of oxygen and carbon dissolved, or rather the product [wt.%0 ] • [wt.%0 ] , observed in steel making practice is not the same as the equilibrium value. In Figure 2 are shown curves for carbon versus oxygen derived from several studies of the open hearth and ele c t r i c furnace processes. In each case the product is substantially greater than that corresponding to equilibrium with carbon monoxide of 1 atm. at 1600°C. The difference is associated with the mechanism of the overall process by which oxygen diffuses through the metal to the point at which the bubble formation occurs. It is postulated that this point i s the furnace hearth. In order to avoid excessive precipitation of oxide particles, or the formation of CO bubbles, in cooling and s o l i d i f i c a t i o n , the oxy-gen content in steels is often required to be less than 10 ppm. If equilibrium were achieved i t would require 2 wt.% of carbon at this latter content of oxygen. But since equilibrium is not obtained and also since lower carbon contents are often desired the oxygen con-tent is further lowered by either: 1. Lowering the pressure of carbon monoxide, or 2. Deoxidizing the steel with an element which forms a very stable oxide. 6. Figure 2. C-0 equilibrium in steel furnaces. A. Equilibrium, 1600°C, equation 2.1.1 B. Ele c t r i c furnace C. Open hearth 7. If i t was desired to lower the content of both carbon and oxygen from point D (Figure 2) to point E (Figure 2) reaction 2.1.1 could be allowed to proceed at a lower P . The resulting carbon and oxygen CO content w i l l thus follow path D-E. One potential drawback of this procedure in ingot degassing practice is the formation of blowholes which are acceptable only in rimming steels. The second drawback is that refractory/metal interactions may be found at reduced pressures. In general manufacture of steels, therefore, a deoxidizer which forms a very stable oxide and has an appreciable so l u b i l i t y in iron is used. With the use of a deoxidizer such as manganese, s i l i c o n , or aluminum the reaction product i s not a gas and thus the formation of blowholes is avoided. An example of a reaction path followed using a deoxidizer other than carbon is shown as l i n e D-F on Figure 2. 2.2 Deoxidation Thermochemistry Low contents of oxygen in iron can be achieved, as previously mentioned, by deoxidation with an element which forms an oxide of high thermochemical s t a b i l i t y . Examples of these would be carbon, manganese si l i c o n or aluminum. The deoxidant is allowed to react with the dis-solved oxygen at some high temperature. It then forms a precipitated oxide leaving very small amounts of dissolved oxygen at equilibrium. Typically, a deoxidation equilibrium would be described by the following chemical reaction: 8. M°(s),(l), or (g) " ^ F e + ^°he 2 ^ If the free energy change for the reaction is known i t is possible to calculate the reaction's equilibrium constant K, from the relation: where A F2.2.1 - - R T l n K 2 . 2 . i I f the deoxidant and the oxygen follow Henrys' Law at dilute solutions then reaction 2.2.1 can be written as: M ( s ) (1), or (g) = [ M ] 1 wt.% in Fe + [ 0 ] 1 wt.% i n Fe 2.2.II with a different equilibrium constant [ h j - [ h j K' _ "MJ L"0-A strong deoxidizer w i l l have a very small value of K', which corresponds to small values of M and 0 dissolved i n the iron i n equilibrium with MO. Some deoxidants such as aluminum and s i l i c o n have Henrian activity coefficients such that when multiplied by the activity coefficients of oxygen they are found to give a constant product. Thus for s i l i c o n and aluminum the following empirical relationships hold: 9. f g i ' f = constant = A f., • f = constant = B Al 0 K' can now be expressed as: f M[wt.%M].f n[wt.%0] K' = = constant [wt.%M] •[wt.%0] Thus, a new equilibrium constant can be written for reaction 2.2.II. K" = ^' — ^' f • f n constant M 0 [wt.%M] • [wt.%0] ( aM0 ) 3-9 Experimental studies made to determine K", or the two-phase equilibrium between the dissolved elements in the metal and MO were us-ually carried out in the presence of MO of unit activity. Thus the equilibrium constant can be expressed as: K"1 = [wt.%M]-[wt.%0] which i s commonly called the deoxidation constant. Thermodynamic data obtained for the deoxidizing reactions is 2-9 available from many sources. It is clear, however, that the published data is not self-consistent. Work done by various researchers produced different results for A F ° and hence for the deoxidation constants. In the case of s i l i c o n deoxidation the thermodynamic data is fa i r l y consistent. Results produced by several workers"'"^ '"''"'" suggest 10. that for the reaction S;l02(s) " [ S i ] Fe + 2 [ (V 2' 2- i n K"2 2 I I T= -31,000/T + 12.0 In the case of aluminum deoxidation the thermodynamic data i s inconsistent. The aluminum deoxidation w i l l be f u l l y described later, however, for the sake of comparison Gokcen and Chipman"* obtained for reaction: A l 2 0 3 ( s ) = 2 [A1]F...+ 3 [ 0 ] p e 2.2.IV the following deoxidation constant: K1" = -64,000/T + 20.48 2.2.IV It is possible to plot log [wt.%M]_ against log [wt.% 0] at any temperature resulting i n a series of straight lines.Any point on one of these lines corresponds to possible concentrations of M and 0 for a particular temperature in equilibrium with MO of unit activity. Figure 3 shows the equilibrium lines for aluminum and s i l i c o n at various tem-peratures. It is clear that aluminum is the stronger deoxidizer since we observe that for any particular temperature much lower concentrations of aluminum and oxygen then s i l i c o n and oxygen are dissolved in the iron. For example, K1" at 1600°C i s : K"', = 2 .8 x 10 ^ for s i l i c o n deoxidation -14 K"-,' = 2.5 x 10 for aluminum deoxidation 11. Si (Wt %) Al (Wt.%) Figure 3. Alumina and s i l i c a equilibrium lines in l i q u i d iron. 12. Thus, at 1600°C, 10 ppm of s i l i c o n dissolved in iron w i l l be in equilibrium with 1750 ppm of oxygen at unit activity of Si02« On the other hand, 10 ppm of aluminum dissolved in iron w i l l be in equilibrium with 28 ppm of oxygen dissolved at unit activity of alumina. Deoxidation thus removes the dissolved oxygen from solution in the iron melt to form an oxide of the deoxidant used. It is to be noted that deoxidation is a misnomer in the sense that the oxygen con-tent of the bulk li q u i d i s not reduced unless the oxide formed is me-chanically removed. The MO produced (as oxide inclusions) has a lower density than the iron matrix, and hence floats to leave behind a cleaner iron melt, thus making deoxidation meaningful. It is important to note, however, that the bouyancy force developed between the inclusions and the matrix i s a function of the inclusions' diameter as well as on their relative density. The smaller inclusions take longer to float away from the matrix, and to remove a l l the inclusions entirely including the ones in o the size range of 300 A i t requires an impractically long time. Thus, the iron melt is not l e f t entirely free of inclusions. 2.3 Thermochemistry of Slag/metal Reactions Slag/metal reactions can be considered as simple dissociations: (M0) = [M] F e + [ 0 ] p e 2.3.1 13. However, i f the reaction sites can be separated by a potential gradient, the system must be considered as electrochemical: (MO) = ( M 2 +) + (0 2~) 2.3.II (M2*) + 2e = [. M] F e 2.3. I l l (0 2 _) = [ 0 ] + 2e 2.3.IV Fe The sum of equations 2.3.II - 2.3.IV is equal to equation 2.3.1 but reactions 2.3.Ill and 2.3.IV are considered to take place at separated sites. Most steelmaking slag/metal systems contain potential gradients but are clas s i c a l l y considered as reactions of type 2.3.1. Some workers"'""'"' have attempted to describe these reactions in electrochemical terms. 13 Littlewood succeeded in expressing the desulfurization of steel in this manner. He also contended that i f oxidation of impurities from the molten iron is to proceed, electrons must be removed continuously from the slag/metal interface, e.g. [Mn] - 2e = (Mn 2 +) 2.3.V Fe while at the slag/gas interface the following reaction occurs: l/20„ + 2e = (0 2~) 2.3.VI (g) Littlewood neglects to explain, however, how the electrons mi-grate between the two sites, the slag/metal interface and the slag/gas interface, as the slag is ionic. This could be accomplished by a simple redox reaction,or electrons could migrate i f the walls of the container 14. were conducting, e.g. made from a conductive material such as graphite. An electroslag furnace contains at least two separated s i g n i f i -cant slag/metal reaction sites, one being the electrode tip/slag inter-face and the other being the metal pool/slag interface. When the fur-nace is operating on either D.C. or A.C. these sites are separated by a potential gradient. In the case of 60 Hz melting the potential gradient reverses 120 times per second, and the net Faradaic change in the system should be close to zero, i f the operating electrochemical reactions are fast and symmetrical. If indeed the electrochemical reactions are fast and symmetrical, and zero net Faradaic deposition results, the reactions may be the con-sidered as simple chemical dissociations as described by reaction 2.3.1. This type of reaction w i l l be considered to take place under the melting conditions used in this work, and w i l l thus be. treated later. 2.4 Thermodynamics of Slag Systems Used 2.4.1 Binary CaF 2 + A l ^ The two solid phases in this system are alumina (melting point at 2050°C) and CaF 2 (Melting point at 1423°C), which co-exist with the single phase liquid at the eutectic point. There is no detectable solid 14 15 solubility in this system. Considerable disagreement ' exists con-cerning the position of the eutectic in the phase diagram. Recent work 15. done by Mitchell and B u r e l ^ suggests that the eutectic is at 10 wt.% alumina and 1290°C. They contend that different results were obtained due to impure CaF^ used (containing CaO). The phase diagrams offered 14 15 by the various workers ' are shown in Figure 4. When CaF^ and alumina exist as a single phase l i q u i d , however, there is no reason why the liquid should not be considered as having 2+ 3+ 2- 1-four components as Ca , Al ,0 and F rather than two components, as CaFry and alumina. Hence, the l i q u i d systems CaO + AlF^ and CaF 2 + A1 20 3 are formally equivalent. However, in considering how this system reacts, i t has a thermochemical potential for CaO, AlF^, CaF 2 and Al^O^ such that a„ _ and a. 1 T, are very small compared to a„ _ and a,, _ . In CaO "^^3 CaF 2 2 3 other words, the reaction: 3CaF2 + A1 20 3 = 3 CaO +2A1F3 2.4.1 has a very small equilibrium constant. Therefore the system can be treated in other reactions (e.g. Al 20 3/iron) as CaF 2 +A1 20 3 < There is no published thermochemical data on the system. But, as the system is a simple eutectic with no solid s o l u b i l i t y , in the liquid phase i t is postulated to be nearly ideal with a slight positive deviation. 16 25 One of the slag systems commonly used in electroslag melting ' is 25 wt.% A1 20 3 i n CaF 2. Using this l i q u i d slag,temperatures of about 1650-1700°C are obtained in the electroslag furnace. At these tempera-tures the slag is a li q u i d close to saturation in alumina according to the 14 phase diagram proposed by Kuo-Chu-Kun. Thus we may assume that Figure 4. CaF 2 + A1 20 3 binary 17. the alumina activity i s approximately unity. As seen from figure 4 the. slag is l i q u i d at temperatures of 1650°-1700°C in the composition range of 0-30 wt.% A l ^ . As the li q u i d phase exhibits approximately ideal beahviour (disregarding Al-F-0 interaction to form (Al-F-0) n complexes), and as there is no solid m solub i l i t y i t is possible to determine the alumina activity anywhere within that composition range by: a A l 0 ^ a t x w t , % A 12°3^ NA1 0 ^ a t X w t , % M 2 ° 3 ^ 2.4.II N M 0 (at 30 wt.% A1 20 3) N M 2 ° 3 0.25 2.4.2 Ternary CaE^.+Al203+Ca0 The equilibrium phase diagram of the ternary system CaF2+Al203+Ca0 17-19 is not well established. There are various proposed phase diagrams for the system, but they show extensive disagreement. This is to be expected as of the three binaries CaF 2+Al 20 3, CaF2+CaO, and Al^+CaO, only CaO+Al^ is well acknowledged. The binary CaO+Al^O^ is shown in Figure 5. Five different solid compounds can form at various compositions. The alumina and CaO a c t i v i -ties in the various compounds are shown in Figure 5. The CaF 2+Al 20 3 binary was described previously in section 2.4.1. 18. u o 2 41 a. E 4> H 2400 CaO AitQ,(w»8) A li°3 20 Figure 5. CaO + A 1 2 0 3 binary. 1 9 . 91 -23 Several workers^ have studied the CaO+CaF2 system (Figure 6). 22 Budnikov et a l . have shown the system to be a simple eutectic (1360°C, 82.2 wt.%CaF2) with conflicting results on the CaO side of the 21 eutectic. Baak presented a phase disgram with an immiscibility gap in the l i q u i d state on the CaF2~rich side of the diagram. The position of the eutectic found in his work was at the same temperature reported by 22 Budnikov et a l . , but at 86.4 wt.% CaF2» Recent work done by Korpachev 23 et a l . showed the system to have an immiscibility gap, but over a much smaller composition range than is shown by Baak. In section 2.4.1 i t was shown that for the binary CaF2+Al203 i t is possible to consider an alumina phase being diluted by inert CaF^ • Adding CaO w i l l only alter this situation by CaCH-A^O^ interaction, as seen in the binary CaCH-A^O^ system. As long as the system is a single phase liq u i d i t is ,thus, j u s t i f i a b l e to treat the CaF2+CaO+Al20.j system as a CaO+A^O^ system diluted by inert CaF2- This system is analogous to the 24 CaO+Si02+CaF2 system where Mitchell showed that i t can be treated as . a calcium s i l i c a t e + CaF2 pseudobinary. An appropriate slag for electroslag melting should have }inter 25 a l i a , the following properties; 1. The liquidus temperature of the slag should be less than the melting point of the metal. 2. The melting point of the slag's primary phase should be greater than that of the metal. 20. 2600, Budnikov et al. . 22 Mukerji 21 Baak a 22 2200 Temp-^C I Liquid region 1800 2 Liq.region / / / / / 2 5 7 0 ^ l\ i i i i I \ I I I I I I I I / I 1400 / / / / CaO + Liquid t CaF?+ Liq- , C Q F | 2 ^ Ca(0 Cafc 20 60 40 CaO (wt.%) 80 CaO Figure 6. CaF 2 + CaO binary. 21. As the melting point of CaF2 (1418°C) is below that of both iron (1538°C) and nickel (1450°C) a calcium aluminate compound of higher melting point than the metal, alumina, or CaO (M.P.= 2614°C) would be required as a primary phase. The compounds thus available would be CaO'Al 20 3 (1608°C), Ca0-2A1203 (1770°C) or Ca0-6A1203 (1860°C) in addition to A1 20 3 and CaO. 19 Research on the calcium aluminate + CaF 2 pseudobinaries has demonstrated that they are not simple eutectic binaries. Some exhibit compound formation and liq u i d immiscibility gaps. However, in the ab-sence of published data, we have assumed that the liquids used were single phase, and precipitated the appropriate Ca0+Al 20 3 primary com-pound on freezing. CaO and A1 20 3 interact in a base/acid relationship in CaF 2, e.g. (Al 26 3) + (0 2 _) = 2 (A10~) which may be followed, for example, by: (A10~) + 4 (F~) = (A10F^~) + (0 2") This l a t t e r reaction would modify the acid/base relationship i f 3-the ion AlOF^ had significant s t a b i l i t y . However, as we have assumed pseudobinary behaviour for the system CaF2+CaO+Al203 we must also assume 3-that the exchange reaction giving rise to AlOF^ has an equilibrium lying extensively to the l e f t at a l l reasonable values of oxide ion ac-t i v i t y . 22. We have available two methods of controlling the alumina activity in the liq u i d system CaF^+CaO+Al^O^. The f i r s t of these is to use dilution in the binary CaF2+Al203, and the second is to alter the ratio of CaO/AJ^O^. Both of these methods, of course, carry the boundary restriction that the chosen composition must have physical properties compatible with i t s use as an ESR slag. In one experiment (ingot 3) i t was desired-to lower the alumina activity in the slag by an order of magnitude from unity to 0.1. The following slag was thus used: 70 wt.% CaF2 23,5 wt.% A1 20 3 6.5 wt.% CaO The activity of the alumina in the slag was determined in the following way: The mole fraction of A l ^ in CaO^A^O.^ is 0.666. Thus from 2 a plot of log alumina activity versus mole fraction of alumina (Figure log activity of alumina was determined to be -0.13. The activity was therefore 0.74. As CaO"2Al203 was diluted by CaF 2 the activity in the slag was f i n a l l y determined to be 0.74(0.115) = 0.086, where 0.115 is the mole fraction of CaO^A^O.^ in the slag. In another experiment (ingot 4) i t was desired to lower the alumina activity s t i l l further. The following slag was thus used: Figure 7. CaO and A1„0, act i v i t i e s at 1600°C. 24. 60 wt.% CaF2 1.335 wt.% A1 20 3 38.665 wt.% CaO The activity of alumina was determined as follows: The activity of A l ^ in 3 C a O - A l ^ is antilog (-0.8) = 0.16 (Figure 7). Assuming alumina behaves ideally in 3 CaO-A^O^, i t s activity in a slag of composition: 1.335 gm Al^O^ + 38.665 gm CaO or: 3.34 wt.% A1 20 3 + 96.66 wt.% CaO is 0.013 (Figure 8). Allowing for the dilution in CaF2 the fin a l activity of alumina in the ternary is (0.013) x (0.477) = 0.006. Thus by decreasing the alumina content of the slag and increasing that of CaO the alumina activity was lowered by another order of magnitude. 2.5 Chemical Interaction Between Liquid Iron and CaF2+Ca0+Al203 Slag 2.5.1 Justification for Considering Iron/Al 20 3 Interaction Only In section 2.4.2 i t was shown why i t is possible to consider that a CaF2+CaO+Al203 melt reacts as CaF2 and calcium aluminate. The feasible chemical reactions between the above slag and liquid iron are: (CaF 2) = [ C a ] F e + 2 [ F ] F e 2.5.1 (CaO) = I C a ] F e + [ 0 ] F e 2.5.II (A1 20 3) = 2 [ A l J p e + 2 [ 0 ] p e 2.5.Ill 25. Figure 8. A1 90„ activity at low concentrations in CaO. 26. 26 Calcium, however, i s insoluble in l i q u i d iron , thus, reactions 2.5.1, and 2.5.II cannot take place. As reaction 2.5.Ill i s thus the only possible one i t is valid to consider only the interaction between liqu i d iron and alumina. 2.5.2 Calculated Value of the Equilibrium Constant 2 27 Various workers ' have shown that i t is possible to calculate the free energy of reaction 2.5.IV. A l o 0 , » . = 2 [ Al] . „ . _ + 3 [ 0] n _ „ . _ 2.5. IV 2 3(s) 1 wt.% in Fe 1 wt.% in Fe knowing the free energy of formation of Al^O^^y 2 Al,.. + 3/20o = A1_0,, 2.5.V (1) 2 2 3(s) AF° v= -401,500 + 76.91 T and the free energy of solution of aluminum (2.5.VI) and oxygen (2.5.VII) in l i q u i d iron. Al „ . = [ Al] . _ „ . _ 2.5. VI (1) 1 wt.% m Fe AF° V].= -10, 300 - 7.71 I l/20„, . = [0] ... 2.5.VII 2(g) 1 wt./o m Fe AF° = -28,000 - 0.69 T 2.5.VII Combining reactions 2.5.V, 2.5.VI and 2.5.VII, AF° for reaction 2.5.IV can be calculated as follows: 27.a AF° = - AF° + 2 AF° +3 AF° 2.5.IV 2.5.V 2.5.VI 2.5.VII = - [- 401.50C4-76.91 T ]+ 2 [-10,300-7. 71 T ] + 3 [-28,000-0.69 T ] - 296,900-93.71 T Since AF° = -RT ln K' 2.5.IV log K' = - AF°/2.303 RT = -64.75/T + 20.20 2.1.VIII The equilibrium constant K' which is expressed as: [ h A 1 ] 2 ' [ h J 3 K , _ Al-1 !_Q can be evaluated using equation 2.5.VIII at any temperature. K' can be expanded to: [f • wt.% A l ] 2 [ f . w t . % 0 ] 3 K» = — —2 < aAl 20 3> [ wt.% Al] 2[wt.% 0 ] 3 f 2A. f 3 _ J Al Q If aluminum and oxygen follow Henreys' Law then f and f can A-L _ both be considered to be unity and the equilibrium constant can be written as: K.. = [wt.%Al] 2[wt.%0] 3  C a A l 0 > A i2°3 27.b 2.5.3 Experimental Value 3-9 Many workers have tried to determine the actual, or experi-mental equilibrium constant, K'. During the experiments they a l l maintained the alumina activity at unity by using an alumina container. For unit activity of alumina the equilibrium constant is expressed as: K' = [wt.% Al] 2- [wt.% 0] 3- f^- f Q 3 5 7 9 29-31 Recent work ' ' ' has produced K' values quite close to the calculated value as seen from Table 1 where log K1 values obtained by various researchers are given. Table 1 Values of Log K' Worker log K' Gokcen and Chipman^ -64000/T + 20.48 31 Rohde et a l . -64000/T + 20.57 29 Sawamura et a l . -64800/T + 20.63 30 McLean and Ward -64090/T + 20.41 9 McLean and Bell -64090/T + 20.41 D'entremont et al.'' -64000/T + 20.48 g Hilty and Crafts -58600/T + 22.75 2 8 Wentrap and Hieber -71200/T + 27.90 Calculated value -64750/T + 20.20 28. 8 28 Earlier work done by Hilty and Crafts , and Wentrup and Hieber produced much higher values of K'. Their work has later been questioned by other researchers.^'^ A graphical comparison between the various values of K' is shown in Figure 9 where the Henrian activities of aluminum and oxygen using various K's at 1600°C are plotted. 2.5.4 Deviation from Henrys' Law Gokcen and Chipman^ were f i r s t to detect.the deviation from Henrys' Law at dilute concentrations of oxygen and aluminum in li q u i d iron. They observed a negative deviation from Henrys' Law which is indicative of aluminum-oxygen interaction in the li q u i d iron solution. 32 Following Wagner's interpretation of interaction coefficients, Gokcen and Chipman obtained the following values for the interaction coefficients at 1600°C. Al e Q - -12 0 on 6 Al - " 2 0 where and while e log f« Al ^ 0 2.5.IX 0 [wt.%Al] 0 l Q g f A l 2.5.X 6A1 [vt.%0] 0 = 27 Al 6A1 16 e0 2.5.XI It is useful to recall that f = f Al f 0  r A l r A l X r A l Figure 9. Aluminum and oxygen Henrian a c t i v i t i e s in li q u i d iron at unit alumina activity (1600°C). 30. and f = f ° x f A l r0 r0 x r 0 5 Al Gokcen and Chipman assumed e and eO to be negligible, as 9 was later confirmed by McLean and Bell. Relations 2.5.IX and 2.5.X can thus be rearranged to read: f = f° r A l r A l = antilog [ e . [ wt.% 0] ] and f = f^ 1 0 -o = antilog [ e^ 1 • twt.%Al]] Other workers have since found other values for the interaction coefficient 0 Al 6A1 e 0 D'entremont et a l , ' -1.7 -1 9 McLean and Bell -7.8 -4.6 Buzek 3 6 -1.65 -.96 31 Rohde et a l . -2.0 -1.17 The interaction coefficients were found to be inversely propor-tional to the absolute temperature by a l l the above mentioned workers. This is demonstrated by the values obtained for the interaction coefficients at 1800°C. 3 1 . 0 Al Al 0 D'entremont et a l . ^ -1.56 -.52 Gokcen and Chipman"* -9 -5.5 Buzek 3 3 -1.46 -.86 From the above data i t is clear that the Henrian activity coefficients f and f ^ are not unity. The degree of deviation however, i s , as shown, a matter of controversy. The equilibrium constant K', as shown previously can be written as follows for alumina of unit activity. K' = [wt.%Al] 2. [wt.%0] 3 f 2^- f 3 thus, log K' = 2 log [wt.%Al]+ 3 log [wt.%0] + 2-log f +3 log f A-L U = 2 log [wt.%Al]+ 3 log [wt.%0] + 2e° [wt.%0]+ 3e A 1[wt.%Al ] A-L (J in view of relation 2.5.XI. 2. 5.XII log K' = 2 log [wt.%Al] + 3 log [wt.%0 ]+ 3 e^ [1.12 [wt.% 0 ] 0 1 + [wt.% All] 3-9 28-31 Using equation 2.5.XII various workers ' have plotted (Figure 10) the actual wt.% of dissolved aluminum and oxygen rather than the Henrian a c t i v i t i e s . Most research was carried out in the temperature range 1600°C-1850°C. Only D'entremont et a l . studied the interaction at 1900°C. Figure 10. wt.% aluminum and oxygen i n liq u i d iron at unit A l o 0 ~ activity. 33. From Figure 10 i t can be seen that at 1600°C in the range of -4 -2 10 wt.% aluminum to 2 x 10 wt;% aluminum the interaction between aluminum and oxygen has l i t t l e effect on the amount of oxygen and aluminum dissolved. This means that In this region the equilibrium lines are almost straight, and are positioned very closely to the Henrian activity lines. Review work done by Hopp^ tends to dismiss the studies of Gokcen and Chipman^ as overestimating the interaction coefficients. Avoiding 9 the work of McLean and Bell , Hopp contends that l i t t l e or no interaction occurs up to 0.3 wt.% aluminum. Later work done by Chipman' at higher temperatures demonstrates that he indeed overestimated the interaction. It can be seen from Figure 10 that the interaction at 1800°C is less than at 1600°C. Studies made by McLean and B e l l ^ and D'entremont et all have shown that l i t t l e deviation from the Henrian activity lines appear up to 10 ^ wt.% aluminum. 2 3 For the composition range of no interaction the product f ^ " fg is constant and thus a new equilibrium constant can be written for alumina of unit activity: K ~ 2 3 Al 0 = [wt.% A l ] 2 - [wt.% 0 ] 3 The values of K1" are similar to or slightly higher than those 2 3 of K' as shown by Figure 10. Thus f ^ f n m a v D e considered to be a constant with a value close to unity. 34. 2.5.5 Predictions of the Extent of Al^O^/liquid Iron Interaction at Equilibrium 2.5.5.1 At Unit Alumina Activity As seen in Figure 10, no one set of equilibrium lines for the various temperatures can be used to determine the extent of the inter-action between alumina and l i q u i d iron. Thus Figure 10 was re-plotted giving only the limits of oxygen and aluminum contents dissolved in liquid iron at equilibrium with alumina of unit activity (Figure 11). A l l recent data obtained for the system l i e s within the limits plotted. Equilibrium lines have also been estimated and plotted for the follow-ing temperatures: 1700°C, 2000°C and 2200°C. According to reaction 2.5.Ill aluminum and oxygen dissolving in, or coming out of solution in the metal are locked in the alumina stoichiometry ratio of 3 oxygen atoms per 2 aluminum atoms. Thus, for any i n i t i a l oxygen and aluminum content of the iron i t is possible to determine the path to equilibrium at any temperature. 34 35 Previous workers ' have found that the highest temperature achieved by the slag in the electroslag furnace used is in the range of 1650-1700°C. If, now, pure iron (FVE), with i n i t i a l aluminum and oxy-gen contents of 19 and 30 ppm respectively (point A on Figure 11), were to equilibrate with a slag of unit alumina activity at 1700°C the content of aluminum and oxygen dissolved in the metal would rise toward position A' on Figure 11. Where the shape of line A-A' is the stoichiometric ratio of wt.%0/wt.%Al = 48/54. If equilibrium were Figure 11. wt.% aluminum and oxygen in liquid iron in equilibrium with alumina at various temperatures. 36. achieved the composition at point A' would be reached and the f i n a l content of the metal would thus be 45 ppm oxygen and 36 ppm aluminum. If, on the other hand, the slag/metal interface temperature were 2000°C,and i f equilibrium were obtained the f i n a l oxygen and aluminum content would be 280 and 300 ppm correspondingly. In steel making practice often the iron melt is deoxidized by additions of aluminum. If aluminum were added to FVE the interaction between the iron and the slag would be greatly affected. If, then, the i n i t i a l content of the metal were changed to 30 ppm oxygen and 3730 ppm aluminum (point B on Figure 11)(as done in making ingot 10) upon interaction with a slag of unit alumina activity at 1700° oxygen and aluminum w i l l be removed from the metal in the stoichiometric ratio and dissolve in the slag as alumina. If equilibrium were achieved the f i n a l content of the iron would be 12 ppm oxygen and 3710 ppm aluminum (point B' on Figure 11). In another experiment (ingot 11) greater amounts of aluminum were added such that the i n i t i a l content of the metal was 15900 ppm aluminum and 30 ppm oxygen (point C on Figure 11). It can be seen from Figure 11 that this composition is within the composition range obtained at 1700°C upon equilibration of alumina of unit activity and iron. Thus no interaction is expected to take place between this iron melt and the slag. Aluminum deoxidation was also carried out on ARMCO iron which contains 700 ppm oxygen. In one experiment (ingot 8) 416 ppm aluminum 37. were added (point D on Figure 11). Upon interaction with alumina slag at 1700°C the f i n a l content of oxygen and aluminum is expected to de-crease to 331 and 2 ppm (point D' on Figure 11) correspondingly. In another experiment (ingot 9) 3730 ppm aluminum were added (point E on Figure 11). If equilibrium were achieved with a slag of unit alumina activity at 1700°C the f i n a l oxygen and aluminum content would decrease to 2880 from aluminum and 11 ppm oxygen (point E'). If the i n i t i a l content of aluminum and oxygen in the iron were less than that predicted by equilibrium at a certain temperature,then i f equilibrium were not achieved the resulting aluminum and oxygen con-tent would necessarily be less than equilibrium (undershoot). If, on the other hand, the i n i t i a l aluminum and oxygen contents were greater than equilibrium,and i f equilibrium were not reached the f i n a l aluminum and oxygen contents would exceed the equilibrium values (overshoot). A l l researchers who obtained data for this system tried to ob-tain equilibrium by using an i n i t i a l aluminum and oxygen content greater than equilibrium. As i t is extremely d i f f i c u l t to obtain equilibrium i t i s l i k e l y that their experimental equilibrium values exceed the true equilibrium. In this work, this may be the case with the aluminum de-oxidation experiments. However, in the rest of the melting experiments the opposite is true as the i n i t i a l aluminum and oxygen contents are less than the expected equilibrium content. 38. 2.5.5.2 Variable Alumina Activity In section 2.5.2 i t is shown that [wt.%Al] 2 [wt.%0] 3 f 2 • f 3 K' = , ^ ° ( a A l 0 } AI 2U 3 Thus for any particular value of alumina activity i t is possible to write: K' ( a A 1 . ) = [wt.%Al] 2 [wt.%0] 3 f 2 • f 3 2.5.XIII A1 20 3 Al U It is clear from expression 2.5.XIII that as the activity of 2 3 alumina in the slag decreases so does the product [wt.%Al] •[wt.%0] This i s demonstrated by Figure 12 in which wt.% aluminum versus wt.% oxygen equilibrium lines are given for various alumina act i v i t i e s used in this work. Using Figure 12, and the fact that oxygen and aluminum can be absorbed or precipitated only in the alumina stoichiometry, i t is again possible to predict the fi n a l aluminum and oxygen contents of the ingots, knowing the i n i t i a l contents in the electrodes,and the slag/metal inter-face temperature. The aluminum and oxygen contents of pure iron (FVE) are 19 and 30 ppm correspondingly (point A, Figure 12). The interaction would then follow the stoichiometric path E-D. If equilibrium were reached at a 39. Figure 12. wt.% aluminum and.oxygen in liq u i d iron in equilibrium with alumina of various a c t i v i t i e s . 40. slag/metal interface temperature of 1700°C,the f i n a l aluminum and oxygen content for the various alumina ac t i v i t i e s are given in Table 12. 2.5.6 Possibility of Secondary Reactions Up to this point only reaction 2.5.Ill was considered. Accord-ing to this reaction aluminum and oxygen are deposited in the alumina stoichiometric ratio of 3 oxygen atoms and 2 aluminum atoms. However, other reactions may be of consequence and instrumental in destroying the stoichiometric restriction in the dissolution of aluminum and oxygen in iron in the presence of alumina. 2.5.6.1 Reactions Causing Excess Aluminum One possible reaction that may cause excess aluminum i s : M2°3(s) + 3 F e ( l ) = 3 ( F e 0 ) + 2 t M ] 1 wt.% in Fe 2 ' 5 ' X I V  A F2.5.XIV = 2 1 0 ' 2 0 0 ~ 5 6 , 8 7 T a S A F2.5.XIV = " R T l n K2.5.XIV i t is possible to calculate that log K _ -46,000 + 12.4 S 2.5.XIV T thus, at 1600°C,K2 = 7.1x10 and at 1800°C,K2 = 1.58xl0" 1 0 41. As ( aFeO ) 3 t - h A l } 2 K2.2.XIV - ( a M 2 0 T V T -and since a A 1 „ and a„ can both be considered as unity K' ,. „ T„ can Al^O^ Fe J 2.5.XIV be written as 3 2 K2.5.XIV= ( aFeO } ' t h A l J At 1600°C, for example, i f the &j Q in the slag were 0.1, the resulting h A 1 would be 2.66x10 ~*. If f w e r e unity then 2.66x10 ^ wt.% Al Al or 0.266 ppm aluminum would dissolve in the liquid iron. That is a -4 negligible amount. For °^ 0.01 in the slag 8.45x10 wt.% or 8.45 ppm Al would dissolve. This amount i s also negligible as i t is within the experimental error. However i f the ap eQ in the slag were 0.001, 2.66x10 wt.% or 266 ppm aluminum would dissolve and would thus eliminate the stoichiometric argument used in predicting the f i n a l oxygen and aluminum content in the iron. In the case of a slag/metal temperature of 1800°C the situation is aggravated as shown: -1 -4 For a F eQ = 10 , 4 x 10 wt.% or 4 ppm aluminum would result -2 -2 For a F eQ =10 , 1.26 x 10 wt.%, or 126 ppm aluminum would result -3 -1 For ap eg =10 , 4 x 10 wt.%, or 4000 ppm aluminum would result It is expected, however, that the concentration of FeO in the slag would reach 1 wt.% at the later stages of the melts and i t is an-ticipated that the Raultian activity coefficient of FeO would be 10 and thus the a _ in the slag would be approximately 0.1. At that latter r eu 42. activity no interference from reaction 2.5.XIV would be expected. Reactions other than reaction 2.5.XIV may also be involved in removing the stoichiometric restriction in the dissolution of aluminum and oxygen in liquid iron in the presence of alumina by introducing excess aluminum. Liquid iron may reduce alumina to form A^O^ ^ and A10, . as follows: (g) 2Fe.... + A l o 0 o / . = 2 FeO.... + A1„0. . 2.5.XV (1) 2 3(s) (1) 2 (g) AF° x v = 240,700 -66.33 T and Fe-., + A1„0„, . = FeO-.v + 2 A10, . 2.5.XVI (1) 2 3(s) (1) (g) ^ . 5 X V I - 351,600 - 91. 7 T ron to A^O^ ^ ami A10^^ may, in turn, react with the liq u i d i deposit aluminum and oxygen according to their stoichiometry. Thus Al^O^^ w i l l dissociate to give dissolved aluminum to oxygen mole ratio of 2/1. A10 dissociation w i l l result in Al/0 ratio of 1/1. This compares with the ratio of 2/3 of aluminum to oxygen resulting from Al^O^ disso-ciation. In treating reactions 2.5.XV and 2.5.XVI in the same manner as reaction 2.5.XIV i t can be seen, however, that reactions 2.5.XV and 2.5.XVI w i l l proceed to a lesser extent than reaction 2.5.XIV. As reac-tion 2.5.XIV was considered to be negligible the other two were, therefore, also dismissed. In addition, since Al^O and A10 are in the gas phase at A3. electroslag furnace temperatures, and as gases are continuously flushed out of the system very l i t t l e interaction between A ^ O ^ and liquid iron is expected. 2.5.6.2 Reactions Causing Excess Oxygen Alumina in the slag may decompose to i t s gaseous oxides A^O and A10 according to the following reactions: (A1 20 3) = 2 [wt .%0] p e + A l 2 0 ( g ) 2.5.XVII & F2.5.XVII = 2 9 8 > 5 ° ° " 9 1 ' 3 7 T and (A1„0J = [wt.ZO]„ + 2A10 . . 2.5.XVIII 2 3 Fe (g) AF° X V I I I = 380,580 - 104.22 T The extent of these reactions is very small. For example, in order to equilibrate with 1 ppm oxygen in the liquid metal at 1600°C, -7 -9 partial pressures less than 1.5x10 atm Al^O, and 1.7x10 atm A10 would be required at unit alumina activity. Alternative possible reactions that may increase the oxygen con tent in the iron in excess of the alumina stoichiometry may be: (CaO) = Ca, . + [01 2.5.XIX (g) Fe A F2.5.XIX = 1 5 9 > 9 0 0 " 4 6 - 4 T 44. and (CaO) = (Ca)_ „ + [01 2.5.XX CaF^ Fe AF° ^ has not been calculated as the free energy of solution of calcium in the slag is unknown. Thus, as an approximation, the same AF° as for reaction 2.5. XIX was considered for reaction 2.5.XX. From AF° ^  xiX' K2 5 XIX W a S c a l c u ± a t e < ^ t o be 2.5x10 9 at 1600°C. Thus, ( P C a ) , [ h 0 3 F e -9 K2.2.XIX= \J = 2.5 x 10 9 at 1600°C and K2.2.XX- ( a j ) " 2 ' 5 " 1 0 » 1 6 0 0 C assuming that calcium behaves ideally in the slag. To determine the extent of the reaction we w i l l consider the "worst case" i.e. when aCaO = 1 ' T h u s > a t 1 6 0 0 ° > (PCa>- t h0^Fe = 2 ' 5 x l ° " 9 ^ • [ h Q ] F e =2'5x10"9 Calcium and oxygen, according to reactions 2.5.XIX and 2.5.XX, are bound to the CaO stoichiometric ratio such that (wt.Ca, . , )/ (g,or in slag) [wt.0]pe = 40/16. This results in a restriction on the equilibrium which can be shown (Appendix I) to give rise to a negligible quantity of oxygen in the ingot. 45. 2.6 Precipitation From Fe-O-Al Alloys In ESR the metal reacts with the slag as i t melts, as i t drops through the slag,and until i t leaves the slag/ingot reaction zone. The metal and the slag move towards equilibrium according to reaction 2.5.III. Recall reaction 2.5.Ill: (A1 20 3) = 2 [ A l ] p e + 3 [ 0 ] p e This reaction introduces aluminum and oxygen to the iron as des-cribed previously (section 2.5.2). As the metal leaves the reaction zone, i t experiences the axial and radial temperature gradients existing in the ingot pool. It w i l l experience a temperature decrease causing the ex-solution of dissolved oxygen and aluminum. In the presence of a slag in which the activity of alumina was less than unity, we would expect equilibrium to proceed according to Figure 12. However, since the system is now removed from the slag/metal reaction zone, i t w i l l come to equilibrium with precipitate particles, rather than with the alumina activity in the slag. The composition of the possible precipitates can be seen in the FeO + Al^O^ binary (Figure 13). Three types of precipitates can form: liquid FeO+A^O^ melts (range X-Y on Figure 13), solid hercynite, (FeO-A^O^) and solid alumina. Both hercynite and alumina have f i n i t e solid s o l u b i l i t i e s for each other. It can also be seen from Figure 13 A l 2 0 3 (MoleS) 2 0 0 0 Tempft) 1 6 0 0 1 1 i « I L H+L / / J / A+L /A x . _ ^ r ^ T l \-d (z* / H + L \ A+H \ l / F + L ' H + F i 1 i i — 3 6 0 0 3 2 0 0 2 8 0 0 4 2 4 0 0 FeO 2 0 4 0 6 0 8 0 AUO3 A l 2 0 3 (wtfc) A = A l 2 0 3 H = F e O A l ^ F = FeO L = Liquid Figure 13. FeO + A1,0- binary. 47. that upon cooling,if no nucleation problems exist,the precipitates composition w i l l approach either a two-phase solid mixture containing alumina and hercynite (range Y-Z on Figure 13),or FeO and hercynite (range X-Y on Figure 13). 2.6.1 Alumina Precipitation If alumina were to precipitate i t would be according to reaction: M2°3(s) = 2 [ A 1 ] F e + 3 [ 0 ] F e 2 ' 6 ' 1 Thus i f the aluminum and oxygen contents prior to the precipitation were known i t would be possible to predict the f i n a l content of oxygen and aluminum in solution,and also the amount of alumina precipitated. If indeed pure iron (FVE) equilibrated with slag of unit alumina activity at 1700°C,and then cooled, the reaction path would then move down line A'-K on Figure 11. At the s o l i d i f i c a t i o n tmperature, 1538°C, the iron would thus contain 21 ppm oxygen and 9 ppm aluminum in solution (point K in Figure 11). This means that 45-21 = 24 ppm oxygen combined with 36-9 = 27 ppm aluminum to form 51 ppm of alumina precipitates, assuming equilibrium prevails. If equilibrium were not achieved on precipitation higher amounts of aluminum and oxygen would remain in solution with iron. Further changes in composition below the s o l i d i f i c a t i o n tempera-ture are small as at that temperature the metal is already l e f t with 48. minute amounts of aluminum and oxygen in solution. In addition s o l i d -state diffusion processes are involved and these require a long period of time to achieve equilibrium. Thus the reaction may be considered quenched with the metal's s o l i d i f i c a t i o n . 2.6.2 Hercynite Precipitation Up to this point only alumina has been considered as the possible reaction product. However, hercynite, is another feasible reaction 36-39 39 product. The thermodynamic data for the formation of hercynite is as follows: FeO • A l o 0 _ , v = 2 [wt^Al]^ + 4.[wt.%0]_ + F e , 2 . 6 . I I 2 3(s) Fe Fe (1) AF° , T T = 334,140 - 109.63 T £ . 0 . 1 1 Using this data along with the interaction coefficients obtained 9 by McLean the hercynite equilibrium lines were plotted (Figure 14). Alumina precipitation lines were plotted,in addition,for comparison. It can be seen from Figure 14 that thermodynamically alumina is much more stable than hercynite at any given temperature. Only at very low aluminum concentrations does hercynite formation become more favour-able. At 1538°C, for example, the alumina is more stable at concentra--4 tions greater than 10 wt.% aluminum. At lower concentrations of alumi-num hercynite is more stable, as the hercynite line f a l l s below that of alumina. 49. Figure 14. A l ^ and hercynite equilibrium lines in liquid iron. 50. Hercynite may also precipitate after an i n i t i a l alumina pre-39 cipitation when the metal i s sufficiently depleted in aluminum. The equilibrium in this case is between alumina, hercynite, the dissolved oxygen,and the liquid iron,following the reaction: FeO • A1„0 = Fe,-,. + [01 + Al o0,. , 2.6.Ill 2 3(s) (1) 1 wt.% in Fe 2 3(s) AF° , T T T = 34,950 - 12.99 T Z , o . I l l At 1600°C, for example, hercynite would precipitate when the oxygen content in the iron melt is greater than 0.058 wt.%. For this reaction to occur i t is clear that the oxygen dissolved must be in excess the alumina stoichiometry. 40 Chipman has suggested that both alumina and hercynite may occur due to incomplete mixing of the deoxidizer, i.e. aluminum, in the melt. This relates to the deoxidation experiments done with high i n i t i a l 41 aluminum contents. It is believed, though, that the metal pool in the electroslag furnace is well mixed. 2.6.3 Nucleation Considerations 36 Turpin and E l l i o t in their work on "nucleation of inclusions in iron melts" have found that hercynite was formed from iron melts with concentrations of aluminum and oxygen which thermodynamically should have precipitated alumina. They suggested that supersaturation is necessary cr i t for homogeneous nucleation. The supersaturation term AF, should be hom 51. i added to AF° to represent the actual equilibrium. The supersaturation terra is dependent on the interfacial tension between the matrix and the new phase. Turpin and E l l i o t were then able to show the effect of inter-f a c i a l tension on the nucleation of oxides in the Fe-O-Al system demon-strated by Figure 15. It can be seen from this figure that i f AF, horn of alumina (and necessarily also the in t e r f a c i a l tension between alumina and liquid iron) were much greater than that of hercynite, the precipi-tation of hercynite may become more favourable. From Figure 15 i t can also be seen that i f the in t e r f a c i a l ten-sions of the alumina/iron and hercynite/iron couples, for example, were both very high FeO would precipitate. 2.7 Chemical Interaction Between Liquid Nickel and CaF2+CaO+Al203 Slag In section 2.3 i t is shown that the slag CaF2+CaO+Al203 may be considered to react as a CaF 2 and calcium aluminate. As no fluorine sol u b i l i t y in nickel has been reported CaF2 may thus be considered as inactive in the presence of li q u i d nickel. As calcium, aluminum and oxygen a l l have a f i n i t e s o l u b i l i t y in liquid nickel the interaction of calcium aluminate with liquid nickel w i l l thus be considered. In i -t i a l l y , however, the interaction of alumina and CaO (independently) with li q u i d nickel w i l l be discussed. Figure 15. Effect of i n t e r f a c i a l tension on nucleation of oxides in FP-(1-A1 Q V R t - p m . lSlflOf:. (n in P T - O Q I r-m1\36 53. 2.7.1 Thermodynamic Considerations of Liquid Nickel Alumina Interaction Very l i t t l e data is available concerning the thermochemistry of reactions between nickel and alumina. In order to find the AF° for the dissociation reaction: Al.,0,, , = 2 [Al] . . _ . M . + 3 [ 0 ] . . „ . ... 2.7.1 2 3(s) lwt.% in Ni lwt.% in Ni the following three reactions have to be considered: M2°3(s) = 2 M ( l ) + 3 / 2 °2(g) 2 ' 7 - 1 1 2A1 ( 1 ) = 2 [Al] l w t > % l n N . 2 . 7 . I l l 3/20o, . = 3 [0] . „ . ... 2.7.IV 2(g) lwt.% m Ni AF for reaction 2.7.II was obtained from Chipman and E l l i o t where the following value was given: AF° = 401,500 - 7 6 . 9 1 T To determineAF0 for reaction 2 . 7 . I l l the following relation was used: 1/2AF° T T = RT ln [y° • M ' W , N i ] 2.7.V A X 100.M.W. . Al = RT ln + RT l n 58.7/100-27 = 4.576(1873)logO.00025+4576Tlog0.0218 = -31,000-7.6T o 42 The value of y A 1 (0.00025) was obtained from Samarin. 54. To determine AF° f o r reaction 2.7.IV the following r e l a t i o n was 27 obtained from Kubaschewski and Evans: 1/3 AF° 7 I V = 36504-4.575 T l o g N 2.7.VI where N i s one weight percent of oxygen i n n i c k e l expressed i n mole ' f r a c t i o n . Thus, 1/3 AF° I V= 3650 + 4.574 T l o g 0.0214 = 3650 - 7.65 T Reactions 2.7.II to 2.7.IV can be added to form re a c t i o n 2.7.1. In the same manner AF° 7 ^ to AF° 7 ^ can be added to give the free energy of reaction 2.7.1. Hence, the free energy of reaction 2.7.1 was found to be: AF° 7 = 350,450 - 115.06 T The eq u i l i b r i u m constant f o r reaction 2.7.1 can be expressed as: [ v 2 [ v 3 2 ' 7 ' 1 ( a A l 2 0 3 > as A F ° > 7 > I = R T l n K ' ^ ^ log = -76,700/1+25.2 Thus, i t i s possible to p l o t the Henrian a c t i v i t i e s of aluminum and oxygen i n n i c k e l as shown i n Figure 16. I f the Henrian a c t i v i t y c o e f f i c i e n t s of aluminum and oxygen can be considered to be unity at Figure 16. Wt.% aluminum and oxygen in liquid nickel in equilibrium with alumina. 56. small concentrations of those elements in nickel the Henrian act i v i t i e s on Figure 16 can be considered as actual wt.%. As data on the activity coefficients and the interaction coefficients is not available,this w i l l be considered to be the case. Hence, at unit alumina activity, K;; 7 > i = [wt.%Ai] 2 N. [wt.%o] 3 N. 42 A similar graph obtained by Samarin showed a higher value of K"' for a particular temperature. At 1600°C K"' calculated in this work has the value of 2 x 10 ^ whereas that of Samarin has the value of l O " 1 3 2.7.2 Predictions as to the Extent of Liquid Nickel/Al^O^ Interaction Upon interaction with alumina, i f no other sources of aluminum and oxygen are available, oxygen and aluminum deposited in the nickel w i l l necessarily be related by the alumina stoichiometry as shown in equation 2.7.1. In terms of wt.%, aluminum and oxygen w i l l deposit in the nickel in the ratio 54/48. Thus, i f the i n i t i a l content of aluminum and oxygen in the nickel were known i t would be possible to predict the amount of aluminum and oxygen deposited,for any slag/metal interface temperature fif,indeed tequilibrium prevailed. The i n i t i a l content of aluminum and oxygen in the nickel elec-trodes was 33 ppm aluminum and 11.ppm oxygen (point A on Figure 16). 57. If, now, the nickel were to react with a slag of unit alumina activity at 1700°C the reaction path would follow lin e A-D, whose slope i s the alumina stoichiometric ratio of wt.%0 to wt.%Al. If equilibrium were obtained the composition of point D would be reached and the f i n a l aluminum and oxygen contents would increase to 35 and 13 ppm correspondingly. The slags used in the nickel melting experiments had alumina ac t i v i t i e s of 0.48 (ingot 5),and 0.006 (ingot 6). In the f i r s t case the f i n a l contents of aluminum and oxygen are expected to drop sli g h t l y to 32 ppm aluminum and 10 ppm Oxygen (point C on Figure 16). In the case of alumina activity of 0.006 more aluminum and oxygen are expected to be lost to the slag as alumina, and i f equilibrium were obtained the aluminum and oxygen contents would reduce to 24 ppm aluminum and 3 ppm oxygen (point B on Figure 16). In principle the Al/0 stoichiometry might deviate from 2/3 by reaction analogous to 2.5.XIV, 2.5.XV, and 2.5.XVI. However, since NiO has a smaller thermochemical s t a b i l i t y than FeO the significance of such reactions would be even smaller in the present case than was found for liquid iron systems. 2.7.3 Thermodynamics of Ca0/liqi.-d nickel Interaction In the interaction of CaO with l i q u i d nickel the following reaction is of interest: 58. CaO, . = [Ca] + [ 0 ] , . 2.7.VII (s) 1 wt.% m Ni 1 wt./4 in Ni L i t t l e thermodynamic data is available concerning this reaction. AF° of the reaction can be calculated from the AF° of the following reactions: CaO. . = l / 2 0 o / s + Ca . . 2.7.VIII (s) 2(g) (g) 1/20- = [0], _ . ... 2.7. IX 2 1 wt.% in Ni C3 / v = Cd/.\ 2 • 7 • X (g) (i) CH / 1 \ = [ Ca J . « • • 2 • 7 • XI (1) 1 wt.% in Ni 2 Chipman and E l l i o t give the following A3? for reaction 2.7.VIII. A F2.7.VIII= 137,900- 45.7 T AF° for reaction 2.7.IX was obtained as shown in section 2.7.1. AF for reaction 2.7.IX i s given by Chipman and E l l i o t . AF° _ = -35,050 + 19.92 T For reaction 2.7.X}AF0 was evaluated using the following relation: M.W.... AF°=-RT ln (v° ) Ca 100 M.W. _ ; Ca o M.W = -RT In Y " - RT ln u , w ' N i Ca 100 M.W.„ Ca 58.7 = -4.576 (1973) log (0.005) - 4.575 T log 40x100 = -20,800 - 8.4 T The Raultian activity coefficient, y° is not available in the — U a 43 literature, thus i t was evaluated by the Hauffe and Wagner method. 59. By addition of the free energies of reactions 2.7.VII, 2.7.IV, 2.7.IX and 2.7.X, AF° = 135,700-41.83 T , = - 29,600 + 9.15 l 0 g -2.7.VII T The Henrian activity coefficients for reaction 2.7.VII are not known. If they are considered to be unity, the following e q u i l i -brium constant can be written at unit CaO activity: .^Vvil = I w t.ZCa] N £ [wt.%0] N. The equilibrium values of wt.% calcium and wt.% oxygen in the presence of various CaO activities are shown in Figure 17, along with the values of wt.% aluminum and wt.% oxygen in equilibrium with alumina of unit activity. Upon interaction between liquid nickel and CaO, i f no other source of calcium and oxygen were available, oxygen and calcium would be absorbed by the liquid nickel in the stoichiometric ratio as dictated by reaction 2.7.VII. Thus, the ratio of absrobed calcium to that of oxygen would be 40/16. Considering the pure nickel used in this work, the i n i t i a l calcium and oxygen contents were 2 ppm and 11 ppm correspondingly (point A on Figure 17). If the liq u i d nickel were to react with a slag with CaO activity of 0.383 at 1700°C (as was the case with ingot 7), calcium and oxygen would be absrobed by the liquid according to path A-A' (on Figure 60. Ca(Wt%) Figure 17. Wt.% calcium and oxygen and aluminum and oxygen in equilibrium with CaO or A1„0_. 6 1 . 17). If equilibrium were obtained composition A' would be achieved, thus 5 ppm calcium and 13 ppm oxygen would be observed in the liq u i d nickel. Figure 17 also indicates that the interaction between liq u i d nickel and CaO, of a certain activity, at a temperature T, would be of lesser extent than the interaction between liq u i d nickel and alumina of a similar activity as the CaO lines l i e below those of alumina. 2.7.4 Thermodyanmics of CaO+A^O^/liquid nickel Interaction The interaction between CaO+Al^O^ and liquid nickel can be re-presented by the following reaction: 3 [Ca]. „ . ...+A1„0„ , . » 3 CaO, . + 2 [Al], . „ . ... 2.7.XI 1 wt.% in Ni 2 3 (s) (s) .1 wt.% m Ni The free energy of reaction 2.7.XI can be evaluated knowing the free energy of reactions 2.7.1 and 2.7.VII in the following manner: AF° = AF° - 3 AF° 2.7.XI 2.7.1 2. 7.VII = (350,450-115.06 T) - 3(135,700-41.83 T) = -56,650 + 10.43 T ^ A F2.7.XI = " R T l n K 2 . 7 . X I . where, . . 3 2 ( a C a 0 ) ! f h A l ^ '2. 7. XI , . r , T 3 ( a A l 2 0 ) - [ h C a ] 62. log K„ 7 can be calculated to be: log K2' = 12,380/T - 2.28 2 3 It is now possible to plot log (a n>V (a )versus [h ] /[h ] L a u ^X2^3 at various temperatures (Figure 18) In one experiment (ingot 6) both A^O^ and CaO were used in the slag. The alumina activity was 0.006 and the CaO activity was 0.56. T h U S > ( a C a Q ) 3 (0.56) 3 1^^)- = -0T006-= 29.3 3 Corresponding to that value of ( a ^ a g ) / ( a ^ i Q ) ^ r o m Figure 18 i t can be seen that: [h ] 2 A 1 •• = 700 at 1600 C (point A on Figure 18) [h„ ] L a = 340 at 1700 C (point B on Figure 18) = 170 at 1800°C (point C on Figure 18) 40 at 2000°C (point D on Figure 18) This means that [ h A 1] > [h ], and i f aluminum and calcium follow Henrys' Law in liquid nickel, then wt.% of dissolved aluminum would be much greater than wt.% of dissolved calcium. 2 3 3 To bring the ratio of [ h ] /[h ] to 10, a ratio of (a ) / (a ) of about 1000 would be necessary at 1700°C (Figure 18). As 0.56 AI 2 U 3 is about the highest CaO activity commonly obtained in the electroslag 63. Figure 18. Extent of reaction 2.7.XI. 6A. furnace slags, an alumina activity of 0.00018 would thus be required 3 to achieve ( a 0 _) / ( a A 1 _)= 1000. Thus we can conclude that as long as CaO Al^O^ ° alumina is present in the slag alumina w i l l react with l i q u i d nickel to. a much greater extent than CaO. This, however, is not obvious from Figure 17, which shows the interaction of CaO and A^O^ with li q u i d nickel separately. The value of Figure 18 is thus seen in representing the extent to which CaO and A1„0„ react with liquid nickel when both are present in the slag. 65. CHAPTER I I I EXPERIMENTAL APPARATUS, PROCEDURE AND RESULTS 3.1 Melting Apparatus - Electroslag Furnace The electroslag furnace used was described previously by M. 44 Etienne. In this work only A.C. power was used. To start the melting operation, however, D.C. current was used, and when the system generated sufficient heat the current was switched to A.C.. Since the starting procedure of our electroslag furnace requires that the power supply to-lerate transient short circuits,D.C. power was used to establish a molten slag pool. At this point the mode was switched to line frequency A.C.. Argon gas cover was provided around the electrode and over the slag to maintain a very low partial pressure of oxygen over the slag effectively -10 44 (^10 atm. 02)' This was done to ensure that no atmospheric oxygen entered the system. The design of the argon cap is shown in Figure 19. A l l ingots, but ingot 14}were made in a two inch mold. Ingot 14 was made in a three inch mold. 3.2 Melting Experiments I n i t i a l l y i t was the intention of this work to study the extent i 1 Ui? mt777f/m g, c ap 66. Water cooled copper electrode Colorllth Gas inlet Rubber bellows Consumable electrode Gas outlet Gas seal Copper mold Water-jacket Slag Metal liquid pool Solidified metal. Figure 19. Argon slag cap. 67. of slag/metal reactions by melting pure iron (FVE) and pure nickel (electrolytic) through both CaF2+CaO+Si02 and CaF2+CaO+Al203 slags. 3.2.1 CaF2+CaO+Si02 Slag Experiments The CaF2+CaO+Si02 slags were prepared by fusing CaCC>3 with Si0 2 at temperatures above 1000°C to allow CaCO^ to decompose to CaO and C02 gas. The calcium s i l i c a t e fused product was then mixed with CaF,,. Electroslag melting requires a slag of lower liquidus temperature than the melting temperature of the metal,and a primary phase of higher 25 melting point than the metal. To ensure a high melting point primary phase CaC0o and SiO„ were fused in such a ratio as to produce Ca„SiO. J I . 2 4 45 (composition A in Figure 20). The fused product was then diluted with CaF 2 to lower the slag's liquidus temperature to below the melting point of the metal (composition range A-B in Figure 21). The drawback of slags with Ca 2SiO^ as a primary phase is that upon cooling a-Ca2SiO^ transforms to Y-Ca2SiO^, and the transformation is accompanied by a 10% volume increase. The slag was thus broken up into powder which rendered i t useless for electroslag melting operation. As an alternate CaSiO^ (melting point 544°C) was chosen as the primary phase. Composition B,on Figure 20, was thus chosen,and was in turn diluted with CaF 2 > The resulting slag composition was: 30 wt.% CaF2 37 wt.% CaO 33 wt.% S i 0 2 68. SiojMoleg) JO 2 0 3 0 4 0 5 0 6 0 7 0 8 0 9 0 1 0 0 2 2 0 0 2 0 0 0 1 1 8 0 0 I 1600r V 2 o w 4) a E 14001 12001 aCa2Si04 4" Co3Si207 /9Co2Si04 + Ca3Si207 ' 1000 CoO + 0Ca 2 SiO« 800 u ->Ca25iO< + Co3Si207 6001 CoO + 7Ca2Si04 675 C03S.2O7 + aCoSiOj aCoSiOj + Tridymite CojSijOy + ^CaSiOj H3800 3400 3000 V {2600 2 o a a E u I 2200 ^CaSiOj + Tridymite 1800 1400 0CoSiO3+ Quartz 1 0 20 30 A 40 5 0 B 60 1 0 0 0 7 0 80 90 C a O S i O t fwts) 100 SiO. Figure 20. Ca0+Si02 binary Figure 21. Ca ?SiO +CaF_ pseudobinary. 70. Pure iron was melted through this slag, however, the melting 1 was unstable as arcing occurred between the metal and the mold. The arcing took place because the slag skin was not sufficiently thick. Due to the arcing and the limited composition range available^ the CaF2+ Ca0+Si02 slag system was abandoned. 3.2.2 CaF2+Ca0+Al203 Slag Experiments The f i r s t set of melting experiments was done to determine the effect of varying the alumina activity in the slag on the amount of dissolved oxygen and aluminum i n pure iron (FVE). Iron bars, 1 1/4 inch in diameter and 20 inches in length were melted through the following slags with the following alumina a c t i v i t i e s : 1. 75 wt.% CaF2+25 wt.% A1 20 3 ( a ^ Q = 1) 2. 93 wt.% CaF2+7 wt.% A^O.^ ( a M = 0.22) 3. 70 wt.% CaF2+23.5wt.% A120 ( a ^ = 0.086) + 6.5 wt.% CaO 4. 60 wt.% CaF2+1.335 wt.% A l ^ ( a ^ Q = 0.006) +38.665 wt.% CaO It was then desired to determine the effect of varying the alumina and CaO ac t i v i t i e s in the slag on the amount of dissolved oxygen, aluminum and calcium in nickel. Nickel bars, 1 3/8 inches in diameter and 18 inches i n length, were then melted through the following slags: 71. 5. 85 wt.% CaF2+15 wt.% A l ^ ( a ^ Q = 0.48) 6. 60 wt.% CaF2+1.335 wt.%.Al2<)3 ( a ^ • = 0.006) +38.665 wt.% CaO ( a „ _= 0.56) CaU 7. 75 wt.% CaF2+25 wt.% CaO (a C a 0= 0.383) The f i n a l aluminum and oxygen content of ingots 1 to 6 can be predicted as described in sections 2.5.5 and 2.7.2. For a slag/metal interface temperature of 1700°C, i f equilibrium is obtained between the alumina and the metal, the predicted contents of the various ingots are given in Table 2. The analytical results for the total aluminum and oxygen of the ingots are given in Table 3, and plotted on Figures 12 and 16. The metho by which aluminum and oxygen were determined are later described in sections 3.3.2 and 3.3.3. In section 2.7.4 i t was shown that as long as A1 20 3 was present in the slag, effectively no interaction between the liq u i d nickel and CaO should take place. Therefore, ingots 5 and 6 should not contain any calcium. Ingot 7 melted through CaF2+Ca0 slag was predicted to contain 5 ppm calcium and 13 ppm oxygen,after equilibrating with the slag at 17G0°C. Calcium analysis were carried out on the nickel ingots by the Falconbridge Nickel Co. Ltd.,and are given in Table 4. Furnace running conditions for a l l melting experiments are given in Appendix II. 72. Table 2 Predicted Aluminum and Oxygen Amounts Deposited by Alumina Dissociation i n i t i a l f i n a l Ingot No. point on Al(ppm) O(ppm) Al(ppm) O(ppm) f i g . 12 f i g . 16 1 19 30 36 45 D 2 19 30 25 35 C 3 19 30 20 31 B 4 19 30 10 22 E 5 33 11 32 10 C 6 33 11 24 3 B Table 3 Analytical Results and Oxygen of Total Contents Aluminum analytical point on Ingot No. Al(ppm) 0(ppm) f i g . 12 f i g . 16 1 150 225 D' 2 220 240 C 3 170 235 B' 4 70 45 E' 5 771 130 c 6 160 31 B' 73. Table 4 Predicted Calcium and Contents Deposited by CaO Oxygen Dissociation predicted analytical Ingot No. Ca(ppm) O(ppm) Ca(ppm) 0(ppm) 5 0 0 0 _ 6 0 0 65 -7 5 13 100 4 3.2.3 Aluminum Deoxidation Experiments The second set of melting experiments was done to determine the effect of aluminum addition to the metal on the amount of oxygen and aluminum dissolved in the iron. Iron bars 1 1/4 inches in diameter and 18 inches in length, and of two i n i t i a l oxygen contents were used. Iron with 30 ppm oxygen (FVE), and with 700 ppm oxygen (ARMCO) were used. Aluminum in wire form was attached to the electrodes as shown in Figure 22. Various amounts of aluminum were added while the slag composition was held constant at 25 wt.% A^O^ in CaF2» The following additions were made to the iron electrodes which were, in turn, electroslag remelted (Table 5). 74. Figure 22. Aluminum addition to iron electrodes (ingots 9 and 10). 75. Predicted I n i t i a l Aluminum and Oxygen Contents Ingot No. Electrode I n i t i a l 0 Content Al (added (gm) Intended Al content (ppm) point on Fig. 11 8 ARMCO 700 1.15 416 D 9 ARMCO 700 10.3 3730 E 10 FVE 30 10.3 3730 B 11 FVE 30 44.0 15900 C Upon interaction with the slag at 1700°C the following alumin and oxygen contents are predicted from Figure 11. Table 6 Predicted Final Content of Aluminum and Oxygen Ingot No. Final 0 content (ppm) Final Al (ppm) content point on Fig. 11 8 331 2 D' 9 11 2880 E' 10 12 3710 B' 11 30 15900 C The experimental fi n a l aluminum and oxygen content are given in Table 7 and are plotted on Figure 11. 76. Table 7 Analytical Results of Total Aluminum and Oxygen Contents Ingot Analytical Point on No. Al(ppm) 0(ppm) Fig. 11 8 285 300 D" 9 1850 195 E" 10 2656 122 B" 11 8885 37 C" 3.2.4 Temperature Measuring Experiments The f i n a l aluminum and oxygen contents in the ingots seemed to suggest that the apparent slag/metal interface temperature was in the range of 1950-2100°C. Previous workers, 3^' 3 5 however, found the tem-perature to be in the range of 1650-1700°C. The temperature was measured during the melting of nickel through 25 wt.% alumina (ingot 5), and during the melting of ARMCO iron through a 25 wt.% alumina slag (ingot 9). Results obtained are given in Table 8. Table 8 Temperature Results Ingot Max. temperature No. 5 1650°C 9 1640°C The slag/metal interface temperature was measured by attaching a W/3% Re - W/'25% Re thermocouple probe to the electrode which was to be melted. The thermocouple wires were inserted inside alumina thermo-couple tubing which was protected from dissolution in the slag by a BN cover. The temperature measurement apparatus i s shown in Figure 23. The temperature was continuously recorded using an SRG Sargent Recorder, and the maximum reading registered by the recorder is in the region between the electrode tip and the slag/ingot interface. This temperature is given in Table 8. In Figure 23 i t can be seen that the temperature was measured off centre whilst the temperature i s expected to be the highest immediately underneath the electrode. Thus, the results obtained could be less than the maximum slag/metal temperatures. The slag, however, i s postulated to undergo strong convection which combined with the fact that i t i s transparent to thermal radiation should prevent any large temperature gradients developing within the slag. It is anticipated, therefore, that the temperature measured w i l l be within 50°C of the actual maximum slag/ metal temperature obtained. Hence, the slag/metal interface temperature w i l l be considered to be 1700°C, which agrees well with the heat balance 35 proposed by S. Joshi. 3.2.5. Rectification Study Experiments As the slag/metal temperature was indeed found to be in the range of 1650°-1700°C the higher than equilibrium content of aluminum and oxy-gen could only be explained by the presence of electrochemical reactions Figure 23. Temperature measuring probe. 79. rather than by purely chemical slag/metal reactions. As electrochemical reactions require a net direct current flow in the system, A.C. r e c t i f i -cation was suspected,and the D.C. component was measured by the apparatus as shown i n Figure 24. The device is essentially a shunt measurement of current which discriminates against A.C. potentials by means of an RC circuit,where RC = 1 sec. The instantaneous values of re c t i f i c a t i o n were observed on a Keithly micro-voltmeter (Model 153) . Simultaneously, the current and voltage wave forms were observed on a Tectroniks oscilloscope (Type 564B) (Figure 25). The voltmeter reading was in turn integrated with respect to time by a Lectrocount (Model II) which was previously standardized by passing a known amount of D.C. current through the c i r c u i t . Using the standardization factor the Lectrocount integration was converted to Amp.sec which were } i n turn ,divided by the time period over which the integration was made to give the average direct current in the c i r c u i t , in Amps. The D.C. current was measured whilst melting pure iron (ARMCO) through the following slags: 1. 25 wt.% Al 20 3+75 wt.% CaF 2 (slag used to melt ingots 1, 8 to 11) 2. 1.335 wt.% Al203+38.665 wt.% CaO+60 wt.% CaF2) (slag used to melt ingots 4 and 6) The D.C. potential was developed such that the electrode became negative with respect to the ingot. The D.C. results are given in Table *1 00 c n o> n rt H-Hi H> O o 3 to cn c l-t H-3 era n H-H n c B C IOPF CRO INTEGRATOR A -Ve -vW— lookn 81. Table 9 D.C. Current Measured in Various Slags Amount of Al and 0 de-posited at 100% eff. Ingot Slag (wt.%) D.C. Melt Rate Al 0 No. (CaF 2/CaO/Al 20 3) (Amps) (gm/sec) (ppm) (ppm) 12 (75/0/25) 1.8 3.14 53.5 47.5 13 (60/38.665/1.335)3.2 2.82 106 94 To determine whether the r e c t i f i c a t i o n occurred within the electro-slag furnace or externally in the e l e c t r i c a l c i r c u i t the electroslag fur-nace was replaced with a graphite bar. A.C. power was then applied but no rectification was observed, nor was there any observable change in the pure sine wave. The rectification was ,thus ,attributed to the electroslag furnace. The rectification in the system was then believed to be inherent in the large differences between the electrode and ingot diameters. As a l l ingots were produced in a 2 inch mold, another melting experiment (ingot 14) was carried out by melting ARMCO iron through 25 wt.% A1 20 3+ 75 wt.% CaF2 slag,using a 3 inch mold. The results are given in Table 10. Table 10 D.C. Current Measured Using Various Ingot Diameters Ingot Ingot Diameter Melt Rate D.C. No. (cm) (gm/sec) (Amps) 12 5.4 3.14 1.8 14 7.4 3.28 1.52 82. It is worth noting that before the Lectrocount was used, only the D.C. voltage was read (from the Keithly voltmeter). Under various melting conditions the voltmeter reading continuously fluctuated between 0.3 and 2 mV. To obtain the range of D.C. current going through, the resistance of the system was measured. The D.C. resistance was found to be 0.6 mQ while the A.C. resistance was found to be 1.4 mft . As the D.C. component sees only the D.C. resistance, by the relation V = IR,the D.C. current was calculated to range between 0.5 to 3.3 Amps in a l l experi-ments. The fact that the D.C. resistance was less than the A.C. resis-tance suggested that there is an additional reactance in the system which would cause a phase difference in A.C. between the potential,and the current. This was,in fact,observed on the oscilloscope, and the phase angle was estimated to be 10°. It can be shown (Appendix III) that a 10° phase sh i f t corresponds to an impedence which when added to the D.C. re-sistance is equivalent to 1.4 m^  reactance in the electroslag furnace's c i r c u i t . An interesting phenomenon occurred when arcing took place between the electrode,through the slag skin, and to the mold. The D.C. potential reading on the voltmeter shifted such that the electrode became positive with respect to the ingot. The D.C. potential also approached 30 mV which i s more than ten times the potential achieved in the opposite direction in the absence of arcing. When arcing ceased i t took 5 to 10 seconds for the potential to revert to i t s original value. 83. Since this arcing puts the cir c u i t : electrode -* mold ->- ingot,in parallel 46 with the slag resistance, the melting current increases slightly dur-ing the arc period, as is demonstrated by Figure 25. The arc exists around the copper mold more significantly. This effect is shown by Figure 26, which corresponds to the arc of Figure 25. 3.3 Ingot Analysis 3.3.1 Analysis Procedure After the ingots were made in the electroslag furnace a standard analysis procedure was carried out. The ingots were f i r s t microscopically examined. The surfaces of the ingots were observed and related to the melting conditions, in re-spect of the pool volume. The ingots were then cut,and a s l i c e 1/4 inch thick was removed from just below the fi n a l molten pool. The slice,2 inches in diameter and 1/4 inch thick,was,in turn,cut as shown in Figure 27. Sections A to H (on Figure 27) were then used for the ingot analysis as follows: Section A total aluminum analysis Sections B to F total oxygen analysis Section G optical microscopy and microprobe analysis Section H inclusion extraction analysis. 84. Figure 25. Effect of arcing to the mold on melting current. 86. Figure 27. Sectioning of an ingot s l i c e for analysis. 87. The following stages were involved in the ingot analysis: 1. total aluminum analysis 2. total oxygen analysis 3. inclusion analysis The inclusion analysis was composed of the following steps: a. optical microscopy of the inclusions in s i t u b. electron microprobe analysis of the inclusions in si t u c. inclusion extraction d. X-ray diffraction analysis of the extracted inclusions e. atomic absorption analysis of the extracted inclusions. As the ingots were made for different purposes not a l l ingots were analyzed to the same extent. Ingots 1-4 and 8-11 were analyzed completely as the precipitation study was carried essentially on iron ingots. Ingots 5 to 7 were analyzed for aluminum and oxygen contents, and the inclusions were only studied optically 5and with the electron micro-probe. An example of the entire inclusion analysis procedure is given in Appendix IV (for ingot 1). 3.3.2 Total Aluminum Analysis The total aluminum analysis was carried out by the B.C. Research 47 Council using the neutron activation technique with the Sherlock system. It is a non-destructive analysis of section A (Figure 27). It uses an antimony 124 source which generates Y_rays as in reaction 3.3.1. 88. 124 123 Sb 4 ->• S b i / J + Yi 3.3.1 The y r a y s s t r i k e a b e r y l l i u m block to release neutrons by the following r e a c t i o n : 9 Q Be + Y i Be + n 3.3.II. Water i s used to moderate the neutrons to thermal energies. The thermal neutrons then bombard the specimen (A), and some are cap-tured by the n u c l e i of the elements present i n the sample. In the case of aluminum the following reaction takes place; 7 7 7 R A l .+ n A l + Y 2 3.3.III. -14 where i s energy released w i t h i n a short period of time (10 sec) a f t e r the neutron capture. The n u c l e i are thereby changed to an excited, unstable s t a t e , and then decay back (2.5 minutes h a l f l i f e ) to a s t a b l e s t a t e emitting Y ^ -rays i n the process,as follows: A l 2 8 -»• A l 2 7 + Y 3 3.3.IV. The Y 3-rays, provide a unique y-ray spectrum that can be used to i d e n t i f y the r a d i o a c t i v e isotopes of the element. The aluminum spec-trum has an i d e n t i f i a b l e peak at 1780 keV ( i . e . the energy o f Y 3 f ° r aluminum i s 1780 keV). In t h i s analysis aluminum s u l f a t e standard was used f o r c a l i -b r a t i o n , and the expected accuracy i s approximately - 10% of content. 89. 3.3.3 Total Oxygen Analysis The total oxygen content of the ingots was determined by the standard inert gas fusion technique, using a Leco 537 induction furnace Model 507-800, and a Leco rapid oxygen analyzer Model 509-600. The metal samples (B to F on Figure 27) were prepared by the standard method. Due to the high temperatures (> 2500°C) achieved by the induc-tion furnace,and since the metal is fused in graphite, the metal becomes carbon s turated. Upon equilibration with carbon a l l but negligible amounts of oxygen (< 1 ppm) are driven off. Also,at the high tempera-tures obtained carbon reduces a l l possible oxides in the metal. Thus i t i s valid to expect that the values obtained represent the total oxy-gen content of the metal. The major error in the oxygen analysis arises in the nature of the dispersed second phase oxide. When this fact is combined with the small sample weight (^0.5 to 2 gm) i t is seen that the analysis error arises primarily in sampling. The overall error i s estimated to be ±10% of content. 3.3.4 Inclusion Analysis The analytical results from the various methods are combined to give a comprehensive analysis of the inclusions in the following manner: The electron microprobe results indicated what elements were present in the inclusion with the exception of iron. The X-ray results 90. gave the phase composition of the i n c l u s i o n s . The atomic absorption determination of the i r o n content of the extracted i n c l u s i o n s allowed an approximate evaluation of the wt.% of the various phases present i n the in c l u s i o n s . I t also complemented the ele c t r o n microprobe analysis which, as previously mentioned, was not useful i n determining the presence of i r o n i n the i n c l u s i o n s . F i n a l l y , the o p t i c a l analysis of the i n c l u s i o n s determined whether various phases existed a l l i n the same i n c l u s i o n , whether they exi s t e d each i n a separate i n c l u s i o n , or as a combination of the two. The o p t i c a l analysis also allowed the probable compositions of the i n c l u s i o n s to be determined along with t h e i r maximum s i z e . 3.3.A.l O p t i c a l Analysis The o p t i c a l analysis specimens were mounted i n Kold Mount, ground on Emery paper, and diamond polished. Care was taken during the grinding and p o l i s h i n g not to•remove i n c l u s i o n s from the surface. The polished samples were then observed on the Reichert O p t i c a l Microscope, and on the Bauch and Lomb o p t i c a l microscope. The o p t i c a l analysis observations are given i n Table 11. 3.3.4.2 El e c t r o n Micro-probe Analysis The o p t i c a l specimens were also used f o r probe a n a l y s i s . The specimens were u l t r a s o n i c a l l y cleaned, and were then coated with conduc-t i v e s i l v e r paint. The probe used was JXA-3A E l e c t r o n Probe X-ray Microanalyzer. The in c l u s i o n s i n the specimens were analyzed f o r 91. Table 11 Optical Observations of Inclusions Ingot No. Observations s i z e (urn ) shape colour probable composi-tion occurrence frequency 1 < 5 round grey FeCA^O^ a l l 2 < 5 round grey FeCA^O^ a l l 3 < 8 angular grey A1~0„ some < 4 round grey FeO.Al^O^ most 4 < 3 round grey FeO.Al„0 a l l or FeO 5 < 10 a l l shapes dark grey ^2^3 a ^ 6 < 3 angular dark grey ^2^3 some < 5 a l l shapes light grey Ca/Ni most 7 <15 a l l shapes light grey Ca/Ni a l l 8 <3 round grey FeO.A^O^ most < 4 angular grey 3 ^eW 9 < 3 a l l shapes grey FeO.A^O^ a l l 10 < 3 angular grey A1~0~ most < 3 round grey FeD-JU^O^ few 11 < 3 angular dark grey A1„0_ almost a l l < 1 round grey FeO.A^O^ very few 92. aluminum and calcium using 25 kV and for Fluorine and oxygen using 10 kV excitation voltage. The probe analysis was useful for qualitative analysis only, as the inclusion size rarely exceeded 10 um . For i n -clusions of size less than 1 vm. even qualitative analysis was im-possible. The electron microprobe results are given in Table 12. 3.3.4.3. Inclusion Extraction The inclusions were extracted from the iron matrix by the bro-48 mine-methanol technique suggested by Bohnstedt. The bromine, while dissolving a l l the elemental species, leaves the inclusions intact, while anhydrous methanol is added to control the dissolution reaction. Anhydrons rather than hydrated methanol is used to avoid dissolving FeO inclusions. The apparatus used in this work is shown in Figure 28. Once the dissolution step was completed the procedure in this work was much simpler than the procedure suggested by Bohnstedt, as only oxide inclusions are expected to be present in the samples used. This allows us to eliminate steps meant to remove carbide inclusions. Thus, the inclusions were separated from the bromine solution by centrifuging, and were then dried. Due to the small size of the inclusions (< 10 um ) f a n d the amount of handling required,only a qualitative analysis was done on the inclu-sions (X-ray). Quantitative analysis, mainly the determination of the 93. S i l i c a gel Methanol S i l i c a ger Ascarite Bubbler Condenser Bromine & iron sample Methanol Heater and s t i r r e r Figure 28. Extraction app aratus 94. Table 12 Electron Micro-probe Analysis of Inclusions Ingot Al 0 Ca No. Comments 1 present present absent absent Al and 0 present in a l l i n -clusions 2 present present absent absent Al and 0 present in a l l 3 present present absent absent Al and 0 present in a l l very strong Al count 4 present present absent absent Al and 0 present in a l l Al present in most 5 present present absent absent Al and 0 present in a l l 6 absent absent present absent Ca and Ni in inclusions. Inclusions are Ca/Ni second phase 7 absent absent present absent Ca and Ni in inclusions. Inclusions are Ca/Ni second phase 8 present present absent absent Al and 0 in a l l 9 present present absent absent Al and 0 in a l l . Stronger Al than 7 10 11 present present absent absent Al and 0 in a l l . Stronger Al than 8 present present absent absent Al and 0 present in a l l Very strong Al counts total inclusion content in the iron, was impossible, as an appreciable amount of inclusions, particularly in the lower size range (< 1 um) w a s believed to be lost during the separation. 3.3.4.4 X-ray Diffraction Analysis The extracted inclusions were analyzed for phase composition by the Debye-Scherrer technique using the Philips X-ray machine. A cobalt tube along with an iron f i l t e r was used. The machine setting was 36 kV and 18 mA for a period of three hours. A 114.83 mm diameter camera, was used. The camera is designed such that 2 mm measured on the film corre-sponds to 1°16. The distance along the film between the zero point and the reference end is 180 mm. Results of the analysis are given in Table 13. 3.3.4.5 Atomic Absorption Analysis for Inclusions' Fe Content A weighed amount of extracted inclusions was fused with potassium pyrosulfate which 9in turn ?dissolved the inclusion. The fused product was then dissolved in 3NHC1 solution. The resulting solution was analyzed for iron content in the SP 90 atomic spectrophotometer unit. The results were calibrated using a standard solution containing similar potassium pyrosulfate and HC1 concentrations, and are given in Table 14. Due to the small amount of extracted inclusions available, very low concentrations of inclusions were present in the solutions to be Table 13 X-ray Diffraction Analysis of Indus ions Ingot No. Results 1 Fe0.Al 20 3+ weak A l ^ 2 FeO.Al 20 3+ weak A l ^ 3 A 1 2 ° 3 + V" W e a k F e ° 4 FeO + weaker F e O . A l ^ 8 FeOAl 20 3 + weaker A l ^ 9 A1 20 3 + weak F e O . A l ^ 10 A1 20 3 + weak FeO.Al 20 3 11 A1 20 3 + v. weak FeO.Al 20 3 Note: v = very Table 14 Atomic Absorption Analysis for the Determination of the Inclusions' Fe Content Ingot Fe content probable composition No. (wt.%) in inclusions wt.% A120 wt.% FeOAl20 wt.% FeO 1 22 30 70 2 22 30 70 3 13 60 40 4 67 ~ 14 86 8 19 40 60 9 2 95 5 10 3 90 10 11 1 98 2 analyzed. Thus the iron concentration was also very low,and we were work ing at the lower end of the operating range of the unit, where the accur-acy i s only ±30%. The results are useful, though, in terms of relative amounts of Fe in the various inclusions. It was originally intended to measure the total aluminum content of the ingots in this manner. However, the large concentrations of iron relative to the aluminum present in the solutions rendered the analysis impractical. 98. CHAPTER IV DISCUSSION 4.1 Slag/Metal Reactions 4.1.1 Analysis of Results in Terms of Chemical Thermodynamics 4.1.1.1 Iron/Slag 4.1.1.1.1 Introduction Electron microprobe analysis has shown that neither calcium nor fluorine were present in the inclusions (Table 12). This reaf-firms the reported insolubility of calcium and fluorine in l i q u i d iron. Thus,it is indeed j u s t i f i a b l e to consider alumina as the only reactive component in the CaF^+CaO+Al^O^ slag system}upon interaction with l i q u i d iron. This also means that no liquid slag was trapped in the metal's liquid pool. 49 The alumina primary phase was reported to contain some dissolved CaO as CaO^Al^O^ since CaO is always present in f i n i t e amounts (500 ppm) in the CaF^ used. As no calcium was detected in the inclusions i t is also possible to conclude that no slag primary phase was entrapped in the liquid metal pool. Therefore, a l l the inclusions in the iron ingots indeed arise by precipitation. 99. The extent of the interaction is determined from the total aluminum and oxygen found in the ingots as compared to the i n i t i a l contents in the electrodes. Before analyzing the results i t is f i r s t necessary to determine how meaningful, or how valid are the total aluminum and oxygen results. One factor that may question the validity of the results is that D.C. power was used to start the melting operations. FVE ingots melted entirely through 25 wt.% KL^O^I^ w t>% CaF2 slag, using D.C. (electrode negative), as was used to i n i t i a t e our melting experiments, are known^^ to contain 140 ppm aluminum and 450 ppm oxygen. This compares with A.C. contents of 150 ppm aluminum and 225 ppm oxygen (ingot 1). 3 These differences are negligible as the molten pool volume ( 15 cm = 120 gm) containing 450 ppm oxygen and 140 ppm aluminum would contribute only about 15 ppm oxygen and 5 ppm aluminum in the ingot i f i t were distribu-ted evenly. Actually only the region very near to the position where D.C. was switched to A.C. is expected to be greatly affected, and as the samples were taken well away from that region no detectable effect is expected. The results also depend on a l l the precipitated second phase inclusions remaining in the matrix, as opposed to floating out and dissolving in the slag. If the metal pool is quiescent we can apply Stoke's Law to determine the inclusions' terminal rising velocity. If the Stokes rising velocity of the inclusions did not exceed the ingot's so l i d i f i c a t i o n rate the inclusions would indeed remain in the metal. 100. -2 At the s o l i d i f i c a t i o n rates obtained i n our furnace, (a, 10 cm/sec) accord 1'ig to Stoke's Law no i n c l u s i o n s smaller than 20 um i n diameter should escape. The l a r g e s t i n c l u s i o n s found i n ingots 1 to A, and 8 to 11,were well below t h i s l i m i t ( % 10 um i n diameter). The inclusions, however, were found to coalesce (figure A.IV.2). This would increase t h e i r r i s i n g v e l o c i t y " ^ to perhaps beyond the s o l i d i f i c a t i o n rate. Thus, i t i s conceivable that coalesced i n c l u s i o n s may f l o a t away. I f , on the other hand, the metal pool i s turbulent with v e l o c i t i e s i much greater than the i n c l u s i o n s ' terminal v e l o c i t i e s , S t o k e s Law would _3 not apply. A terminal v e l o c i t y of 3 x 10 cm/sec i s attained by an i n -clus i o n 10 pm i n diameter,following Stoke's Law. Hence, turbulant v e l o c i t i e s w ell below t h i s value would be required for a simple bouyancy model to apply. I t i s , t h u s , p o s s i b l e that the t o t a l aluminum and oxygen analyzed was l e s s than the contents obtained by i r o n / s l a g i n t e r a c t i o n . 4.1.1.1.2 Variable Alumina A c t i v i t i e s Assuming a slag/metal i n t e r f a c e temperature of 1700°C, the f i n a l aluminum and oxygen contents of pure i r o n , remelted through slags of various alumina a c t i v i t i e s are given i n Table 2,and p l o t t e d on Figure 12, as points B to E. The a n a l y t i c a l r e s u l t s of the aluminum and oxygen content i n the ingots, given i n Table 3,and shown i n Figure 12,as points B' to E', d i f f e r e d g r eatly from the predicted values. The a n a l y t i c a l r e s u l t s are well i n excess of the predicted values. I f the slag/metal 101. interface is indeed 1700°C,the results are quite surprising as the i n i t i a l aluminum and oxygen content is below equilibrium, and,thus, the maximum possible content of aluminum and oxygen should be that of equilibrium. In these experiments the expected equilibrium contents were in fact exceeded. The results, would indicate equilibration between the metal and the slag at much higher temperatures Q.900-2000°C). Even though the only source of additional aluminum and oxygen accessible to the metal was the alumina in the slag, the analytical results did not l i e on the alumina stoichiometry path A-F. The results were at positions far enough from the stoichiometry path such that experimental error could not account for the difference. Ingots 1-3 showed excess oxygen,while aluminum was in excess of the stoichiometric ratio in ingot 4. These variations cannot be explained in terms of different slag/metal interface temperatures. A mechanism of aluminum and oxygen deposition,other than chemical equilibration,is, hence, re-sponsible for the deviation from path A-F. The validity of the predicted chemical interaction between the metal and the slag i s also questioned by the relative contents of alumi-num and oxygen in ingots 1 to 4. Even though the values of dissolved aluminum and oxygen in the metal should decrease with decreasing alumina activity in the slag, the analytical results of ingots 2 and 3 exceeded those of ingot 1. Recall the alumina activity in the slag in making the various ingots: 102. Ingot Alumina A c t i v i t y 1 1 2 0.22 3 0.086 4 0.006 The value of aluminum and oxygen i n ingot 4 was well below those of ingots 1 to 3, as was predicted. From these r e s u l t s we can conclude that not only do we have a deviation from the numerical equi-l i b r i u m values of aluminum and oxygen, but that the system does not follow a q u a l i t a t i v e d e s c r i p t i o n of the equilibrium i n respect of alumina a c t i v i t y v a r i a t i o n s . 4.1.1.1.3 Deoxidation Experiments The aluminum and oxygen contents of ingots 8 to 11,given i n Table 7,and i n Figure 11 }are w e l l i n excess of the predicted values given i n Table 6. The r e s u l t s i n d i c a t e e q u i l i b r a t i o n with the s l a g at 2000 -2100 C rather than at 1700 C. Thus whereas the metal was expected to lose aluminum and oxygen as alumina to the slag>at times the oppo-s i t e took place. The r e s u l t s i n these experiments i n d i c a t e an e q u i l i -brium temperature even higher than the experiments c a r r i e d out with no aluminum deoxidation. The differ e n c e may mean that aluminum deoxidation i s not e f f i c i e n t and that excess aluminum i s required to lower the oxygen content to any extent even.when considering the slag/metal rcac-103. tion to occur at 1900 C,as is the case in the experiments done without aluminum deoxidation. The inefficiency of aluminum deoxidation i s well acknowledged, and often 100 times the theoretical amounts of aluminum are added in steelmaking practice in order to achieve the desired de-oxidation level. It is d i f f i c u l t in these experiments to deter ine how closely the alumina sto ichiometry was followed in the slag/metal interaction since the aluminum content of the metal melting at the electrode tip is unknown. Table 5 gives the theoretical aluminum contents, had a l l the aluminum reached the iron melt. This,however, is unlikely as aluminum may have been oxidized by the oxygen in the atmosphere above the slag,following the reaction: 2 A l ( 1 ) + 3/2 0 2 ( g ) = A l 2 0 3 ( s ) The extent of this reaction is negligible as to oxidize only 100 ppm aluminum an atmosphere equivalent to a i r would be necessary. It is also possible that some aluminum deoxidized the slag as follows: 2 A l ( i ) + 3(Fe0) = F e ( 1 ) + A l ^ The most probable reaction causing a loss of aluminum is the evaporation reaction. Since the condensation site would be the cold mold wall the aluminum does not reflux, and i s , hence, removed from the system. 104. Assuming that a l l the aluminum did reach the iron melt,Figure 11 demonstrates that while ingot 8 has excess aluminum, ingots 9 to 11 show excess oxygen. This indicates the likelihood that some aluminum was lost prior to dissolution in the iron melt. The basic relation between aluminum and oxygen in solution in liqu i d iron is well demonstrated. Figure 11 shows that as the aluminum content increases the oxygen content does in fact decrease. 4.1.1.2 Ni/Slag Interaction 4.1.1.2.1 Ni/(A1 20 3 + CaO) The i n i t i a l aluminum and oxygen contents of the nickel electrodes (5 and 6) are given in Table 2 along with the predicted values upon the metal's interaction with slags of various alumina activities,at 1700°C. The analytical results of the aluminum and oxygen content in the ingots are given in Table 3. The i n i t i a l contents (point A) and predicted con-tents (points B and C) and the analytical results (points B' and C') of oxygen and aluminum }dissolved in liquid nickel, are a l l plotted on Figure 16. In this system, paralleling the iron/A^O^ system, the analytical results are well in excess of the predicted values. Here, again, the reactants are seeing an apparent slag/metal temperature of about 2000°C. In this system, as well as in the iron/Al^O^ system, aluminum and oxygen are not absorbed in the alumina stoichiometric ratio as indi-cated by path A-E on Figure 16. The f i n a l aluminum content is well 105. in excess of the stoichometric ratio. Thus another mechanism other than the chemical dissolution of alumina is involved in the dissolution of aluminum and oxygen in the metal. 3 In section 2.7.4 i t was shown that even when (a,., „ ) / (a„ n) = 10 the interaction between alumina and nickel w i l l be predominant. In fact,at 1700°C the ratio(h^) 2/h-Q ) 3 is about 10. Ingot 6, however, was found to have 65 ppm calcium,which according to Figure 18 again indicates an apparent slag/metal interface temperature well above 2000°C. If the temperature is indeed 1700°C then, again, a mechanism other than chemical dissociation of aluminum is operslive. Ingot 7, made through a CaF2+CaO slag, was predicted to contain 5 ppm calcium and 13 ppm oxygen (point A' on Figure 17),upon equilibra-tion with CaO. The analytical contents were 100 ppm calcium and 4 ppm oxygen (point A" on Figure 17). Figure 17 demonstrates that the slag/metal interface sees an apparent temperature above 1850°C. It also shows that CaO stoichiometry had l i t t l e bearing on the li q u i d nickel/CaO interaction, as 98 ppm calcium were absorbed by the nickel whereas 9 ppm oxygen were lost. The results of the CaO/nickel interaction, thus, are in accord with the results obtained from the A^O^/nickel interaction. For a slag/ metal interface temperature of 1700°C,chemical dissociation of the slag components is not solely responsible for the f i n a l content of aluminum, oxygen and calcium in the metal. 106. 4.1.2 Possibility of Higher slag/metal Temperatures and the Operation of Electrochemical Reactions The analytical results of the total aluminum, calcium and oxygen contents of ingots 1 to 11 have clearly demonstrated that the slag/metal interaction is no simple chemical dissociation of alumina, or CaO. How-ever, the results would be much more meaningful in terms of chemical interaction i f the slag/metal interface temperature were ,indeed,in the range of 1900 to 2000°C. The results given in Table 8 show that the slag/metal temperature is approximately 1650-1700°C. Therefore, the only other reasonable explanation for the results is the operation of electrochemical reactions i n our system,rather than chemical interactions. If electrochemical reactions do occur then the aluminum, calcium and oxygen contents in excess of equilibrium at 1700°C would be explained. Electrochemical reactions would also explain the lack of alumina, or CaO, stoichiometry between the aluminum and oxygen(or calcium and oxygen absorbed,or lost by the metal,upon interaction with alumina, or CaO, in the slag. 4.1.3 P o s s i b i l i t y of Faradaic Deposition In the previous section i t was shown that the only reasonable interpretation of the analytical results is the operation of electro-chemical rather than chemical slag/metal reactions. For current to pass through an interface between a metal and an ionic slag a Faradaic reaction is required. Hence, an electroslag fur-nace contains at least two electroactive slag/metal interfaces whether 107. 52 i t is being operated with A.C. or D.C. power. It has been proposed that the anodic reaction on an iron/slag interface i s : Fe ( i^ F e 2 + + 2e 4.1.1. and the cathodic reactions can be any of: Fe 2 ++ 2e + Fe, v 4.1. II. (1) Ca 2 ++ 2e •*• Ca* 4.1. III. Ca* ->- Ca°, metal or slag 4.1. IV. Ca* -> Ca°(g) 4.I.V. A l 3 4 + 3e •+ A l * 4.1.VI. A l * • -»• [ Al] ' 4.1.VII. A l * -»• A l ( 1 ) 4.1. VIII. 2A1* + A l 3 + -»• 3(A1+) 4.1. IX A l 3 4 " + 2e -»• (Al +) 4.I.X. Schematically we may represent the polarization behaviour of this interface by the curve shown in Figure 29, where the anodic plateau re-presents the establishment of a region of slag saturated with FeO. It is evident that i f reactions 4.1.1.and4.1.II. are fast and reversible, alternating current may flow through the slag/metal interface without net Faradaic decomposition of the slag. However, i f the reaction is asymmetrical, or i f reactions 4.1.III. to 4.I.X. are produced during cathodic polarization we w i l l have a net Faradaic reaction,and a result-ing D.C. component. This asymmetry is observed in the 60 Hz wave form as shown in Figure 30. CO ^ h v U J \ T5 CM «E u \ a £ .-P c g o m CM q o <£<5 t o CM _2 h o a E <^  . JO C CM \ \ Figure 29. Polarization curve for the system Fe/CaF 2+Al 20 109, Pbtential (OV/divi'sbf-fl Current (400Amp/ " divisionj // / T P 1 P Time (KT3sec/division) fbtential Current Figure 30. Current and Voltage wave form in A.C. 110. D.C. potential measurements showed the electrode in the electro-slag furance to be negative with respect to the ingot. Thus, the elec-trode would behave as the cathode with the following possible net elec-trochemical reactions taking place: ( A l 3 + ) + 3 e - [ A l ] F e > N . A.l.XI (Ca 2 +) + 2e [ Cal . A.l.XII. At the anodic ingots, oxygen would diffuse into the metal from 52 the FeO saturated layer following the mechanism proposed by Beynon. As electroslag melting is a consummable electrode operation, the elec-trode drops containing the aluminum and calcium eventually join the ingot, and so both the cathodic and the anodic deposits reach the ingot. It is possible to determine the magnitude of the D.C. current needed to deposit any amount of aluminum, calcium and oxygen in the metal. In order to deposit 100 ppm aluminum, for example, the amount of D.C. current required, at an electrode melt rate of 3 gm/sec,would be calculated as follows: gm of aluminum melted/sec = (10 ^gm Al/gmFe) (3 gm Fe/sec) = 3 x 10 ^  gm Al/sec no. of equivalents/sec = (3 x 10 ^ gm Al/sec)/g gm Al/equivalent = 3.33 x 10 ^  equivalents/sec Amps needed = (3.33 x 10 ^  equivalent/sec) 96500 coul/ equivalent = 3.21 Amps 111. At the same melt rate 1.44 Amps would be needed to deposit 100 ppm calcium, and 3.61 Amps would be necessary to deposit 100 ppm of oxygen. At a melt rate of 1 gm/sec one third of the D.C. Amps would be required to deposit 100 ppm of the element. The calculations presume 100% efficiency of the electrochemical reaction. It is l i k e l y , however, that the efficiency would be less than that due to the following reasons: 1. Some electronic conduction may take place such that less charge transfer takes place at the interfaces. 2. The elements in their neutral state may not dissolve in the iron. The f i r s t reason is highly unlikely as the slag^at 1700°C)is known to be ionic. The second reason is much more li k e l y as some of the charge transfer may be carried out by calcium which is insoluble in iron, and thus dissolves in the slag or vaporizes. The efficiency of the anodic reaction may be less than 100% i f FeO dissipates in the slag. Therefore, greater current than calculated would be needed in practice to deposit 100 ppm of the elements mentioned. 4.1.4 Analysis of Rectification Results 4.1.4.1 Rectification in Al203+CaF2 slag The maximum amounts of aluminum and oxygen that could be elec-trochemically deposited by the D.C. current found in making ingots 12 112. and 13 were calculated by assuming 100% efficiency of the deposition reactions with respect to aluminum and oxygen. The results are given in Table 3. In the case of ingot 12 where CaF2+Al203 slag was used, a maxi-mum of 53.5 ppm aluminum and 47.5 ppm oxygen could be deposited. These contents, combined with the i n i t i a l aluminum and oxygen contents in the electrodes, are well below the oxygen and aluminum contents found in ingots melted through a similar slag (ingots 1 and 2). Hence i t is unlikely that electrochemical reactions are solely responsible for the aluminum and oxygen deposition. A possible explana-tion for the observed behaviour is that the chemical and electrochemical slag/metal reactions occur conjointly, as shown in Figure 31a. The iron electrode interacts with the slag electrochemically,.following the reac-tion: ( A l 3 4 ) + 3e -»- [All 4.1.XIII. Fe The metal drop then f a l l s through the slag,and since at this point no potential gradient exists between the drop and the slag i t reacts with i t chemically,as follows: M2°3(s) " . 2 [ M ] Fe +Al + 3^ ^  Fe+Al 4 - 1 ' X I V ' The drop then reaches the ingot surface where the diffusion of oxygen from the anodic FeO layer takes place: 113 . di o (fl 00 u CO fi o •H 60 4-1 CO CJ rH cO co cu rt o 4J 0) o o <U CO • CU n cu Xi CU 4-1 n a c 60 -H •H Figure 31. Occurrence of both chemical and electrochemical slag/metal intei-actions 114. ( 0 )Fe0 * [ ° W + 0 A ' 1 - X V -The composition changes resulting from various reactions f o l -lowed by ingot 2, assuming the electrochemical reactions (paths A-B and C-D) are 100% efficient with respect to aluminum and oxj.^en, and that the chemical interaction achieves equilibrium (path B-C),are shown on Figure 31b. It can be shown (Appendix V) that a metal drop with a residence time of 0.1 sec in the slag could readily come to equilibrium, assuming the bulk mass transfer rate across the slag/metal interface to be rate determining. Even i f 100% efficiency i s assumed in the electrochemical de-position, and even i f the metal is assumed to obtain.equilibrium with the slag, the combined aluminum and oxygen contents predicted for ingots 1, 2 and 5 ( a l l melted in CaF2+Al203 slag) are s t i l l below those found. This is shown in Table 15. The situation with nickel melting (ingot 5) is more complex than is the case with iron melting. Even though lower aluminum amounts are predicted in the case of nickel resulting from chemical equilibra-tion, the f i n a l content of aluminum is much higher than in iron. It is important to re c a l l that we are measuring 3 Amps D.C. in the presence of 1000 Amps A.C,using a shunt technique where the measured output is less than 1 mV D.C. This system could only be c a l i -brated using pure D.C. current,and hence a substantial error in the 115. Table 15 Prediction of Aluminum and Oxygen Contents by Electrochemical Reactions, and by a Combination of Both Chemical and Electrochemical Slag/Metal Reactions Ingot Predicted values Analytical No. Electrochemical Electrochemical and values reactions chemical reactions Al (ppm) O(ppm) Al (ppm) 0 (ppm) Al(ppm) O(ppm) 1 72 77 73 78 150 225 2 72 77 62 68 220 240 5 86 58 83 55 771 130 D.C. measurements is not unlikely. The results may also be explained i f the electrochemical equiva-lent weight of aluminum were ,in fact,higher than 9. This means that some alumina ions might have been present in the slag. This, however, was proven unlikely in section 2.5-6, and also we can propose no equiva-lent mechanism for oxygen. Even though electrochemical deposition may not be entirely re-sponsible for the aluminum and oxygen dissolved in the metal, the pre-sence of their reactions explains the lack of alumina stoichiometry in the dissolved aluminum and oxygen contents. The efficiency of the electrochemical deposition at the cathode.(electrode) and anode (ingot) need not be the same. In fact,the excess oxygen found in ingots 1 and 116. 2 suggests that the anodic deposition is of higher efficiency than the cathodic deposition. The oxygen excess may also be attributed to transient electrode-mold arcing. The electrpactive interfaces upon arcing are the electrode/ slag,and the mold/slag interfaces, with the electrode being electro-positive with respect to the mold. This would cause oxygen deposition on the electrode while aluminum and calcium may form at the mold/slag interface,and dissolve in the slag. 5.1.4.2 Rectification in CaFo+Ca0+Alo0„ Slag / 2 J The re c t i f i c a t i o n found while melting iron through CaF2+CaO+Al2P3 slag (ingot 13) was observed to be higher than in the absence of CaO. Almost twice the D.C. current was measured when iron was melted through the CaO containing slag. Ingots 3,4 (iron), and 6 (nickel) were melted through slags containing CaO. The higher D.C. current is s t i l l insufficient to account for the aluminum and oxygen amounts found in ingot 3 (Table 16), but i t may serve to explain why ingot 3 which was melted in a slag with a _ = A J 2 U 3 0.086 had a higher aluminum and oxygen content than ingot 1 melted through a slag of unit alumina activity. The D.C. current was sufficient to deposit the aluminum and oxy-gen amounts found in ingot. 4 (Table 16). In fact> qurrent efficiencies in the range of 50% to 70% would be adequate to deposit the amounts found. 117. Table 16 Prediction of Aluminum and Oxygen Contents by Electrochemical Reactions, and by a Combination of Both.Chemical arid Electrochemical Slag/Metal Reactions Ingot Predicted Value Analytical No. electrochemical electrochemical and Values reactions chemical reactions Al (ppm) 0 (ppm) Al (ppm) 0 (ppm) Al (ppm) 0 (ppm) 3 125 124 105 105 170 235 4 125 124 38 33 70 45 6 139 105 128 95 160 31 The case of nickel melting through a CaO containing slag i s more complex. The D.C. current was sufficient to deposit the oxygen content of ingot 6, but insufficient to deposit the aluminum content. Even though the current was insufficient to deposit aluminum alone, 65 ppm of calcium were also found in the ingot. Calcium could only deposit electrochemically, since i t was shown previously (section 2.5.7) that according to chemical slag/metal equilibrium a negligible amount of calcium would deposit in nickel in the presence of alumina in the slag. This is a strong indica-tion that electrochemical reactions play an important role in the slag/ metal interaction, and that the measured D.C. currents were,in fact,lower than the actual D.C. currents passing through the system. The reason why a low oxygen content (31 ppm) was found in ingot 6 may be explained by the fact the calcium was also found in the metal. 118. As CaO is more stable than alumina, and as the equilibrium lines l i e below those of alumina, lower oxygen contents are expected. This also explains the low oxygen content (4 ppm) found in ingot 7 which was melted through CaF2+CaO slag in the absence of alumina. Experiments in which calcium was deliberately added i n large quantities (equivalent to the aluminum deoxidation experiments with pure iron) would not be feasi-ble in this case as the oxygen content at 100 ppm calcium is already at the limit of our experimental analysis method. Therefore, i t is l i k e l y that chemical as well as electrochemical slag metal interactions take place in our system. 4.1.4.3 Effect of Electrode/Ingot Size Ratio The re c t i f i c a t i o n in the system was believed to be inherent in the large differences in diameters between the electrode and the ingot. Thus, when a FVE electrode (1 1/4 inch) was melted into a 3 inch mold (ingot 14) the rectification was expected to exceed that obtained when a 2 inch mold was used (ingot 1 to 13). The measured D.C. currentshowever, was close to that of ingot 12. This can be possibly explained in terms 52 of the diagram (Figure 29) produced by Beynon. The diagram gives a natural log of the current density (.inversus the anodic and cathodic overpotentials for steady state D.C. polarization. The current densities of the electrode (point A on Figure 29) and ingots 12 (point B on Figure 29)»and 14 (point C on Figure 29),are shown in both sides of the overpotial 119. curve,and i t can be seen that the current densities are sufficiently high to cause the asymmetry. If the polarization time is also current density dependent we would observe rectification of the A.C. The above results would indicate that the electrode current density is the deter-mining factor in producing asymmetry. To reduce the rectification the electrode and ingot diameters should be increased to a point such that the current density would tie in the symmetrical section of the over-potential curve (range E-F on Figure 29). A typical industrial electroslag furnace would have an electrode 18 inches in diameter,and a mold 24 inches in diameter. The current 2 densities in both (^5-10 Amp/cm ) are such that asymmetry in the over-potential curve is expected. 4.2 Precipitation 4.2.1 Precipitation From Fe-0^A1 Compositions The phase compositions of the second phase precipitates were determined by the various analyses }and given in Tables 11 to 14. The analytical results are complementary and self consistent. The analytical results w i l l now be compared with the second phase precipitation as pr.-.:diced by: 1. Stoichiometry 2. Equilibrium thermodynamics 3. Supersaturation a. by cooling b. by s o l i d i f i c a t i o n 120. The predictions w i l l be carried out considering the total aluminum and oxygen analytical values given in Tables 3 and 7, as the i n i t i a l aluminum and oxygen contents in the liquid iron prior to the onset of precipitation. 4.2.1.1 Analysis in Terms of Stoichiometry The alumina and hercynite stoichiometry lines are given in Figure 19. If the ingot composition were to be on the alumina stoichio-metry line alumina would be expected to precipitate. Similarly,hercynite should precipitate i f the ingot's composition were to l i e on the hercynite stoichiometry line . If the composition were to l i e between the two lines both phases should come out of solution. In the case where there was an excess oxygen with respect to the hercynite stoichiometry,hercynite and FeO should precipitate. If aluminum were in excess of the alumina stoichiometry,alumina alone should precipitate,with the excess aluminum le f t in solution in the liq u i d iron. The analytically determined ingot aluminum and oxygen contents are plotted on Figure 14. From the diagram i t can be seen that in ingots 1 and 2 hercynite is expected to precipitate along with some FeO. Hercynite was,in fact, found to be the predominant phase in the inclusions, however, minute amounts of alumina were also detected,rather than FeO. Ingot 3 was found to contain alumina with some FeO inclusions,rather than predomi-nantly hercynite,along with some FeO,as suggested by Figure 14. This 121. could have easily occurred due to inadequate mixing in the pool. Ingot 4 contained FeO and hercynite,whereas alumina is predicted. However, i t would only requite an analytical error of 30 ppm aluminum to s h i f t the composition beyond the hercynite stoichiometry line,and,thus,coincide with the analytical results. Ingot 8 is predicted,and found,to have both alumina and hercynite present in the inclusions. For ingots 9 to 11 alumina is predicted to be found. Alumina was,in fact,determined to be the major constituent of the inclusions, however, traces of hercynite were also detected. Again, the presence of hercynite may be due to uneven mixing of aluminum and oxygen in the liquid metal. Generally, the predictions are useful in determining the predo-minant constituent in the inclusions. The minor constituents do not seem to follow the predictions and might,consequently,be associated with uneven mixing of aluminum and oxygen in the l i q u i d metal. It is important to note, however, that this analysis takes no account of nucleation phenomena. Neither is i t cpncerned with the rela-tive thermodynamic s t a b i l i t y of the various phases. Therefore, this approach is a great oversimplification of the problem. 4.2.1.2 Analysis in Terms of Equilibrium Thermodynamics It was shown previously (section 2.6.2) that in the case where a l l particles have the same nucleation energy,alumina rather than hercynite,is the thermodynamically favourable phase to precipitate out of the Fe-O-Al system in the range of compositions with which we are 122. concerned. This is also shown in Figure 14 as the equilibrium line for alumina formation are below those for hercynite formation ,at any particular temperature. Alumina was indeed found to be the major constituent in inclusions from ingots 3, 9, 10 and 11, and in minor concentrations in inclusions from ingots 1, 2 and 8. Hercynite, however, was analyzed as the pre-dominant second phase in ingots.1, 2 and 8,and detected in minor concen-trations in ingots 4> 9, 10 and 11. FeO was the major second phase in ingot 4. The results indicate that the precipitation agrees closely with the thermodynamic predictions only at high concentrations of aluminum, well in excess of the alumina stoichiometry. But even under these con-ditions some hercynite did precipitate. 39 • McLean and Ward in their study on the thermodynamics of hercynite formation showed that hercynite may precipitate i f during alumina preci-pitation the concentration of aluminum dropped sufficiently such that hercynite became thermodynamically more stable. This would explain the presence of minor amounts of hercynite i n ingots 1 to 4 and 8, but would not hold for ingots 9 to 11 where the alumina precipitation path could never reach the region where hercynite precipitation is thermodynamically favoured. McLean and Ward also studied the following reaction: F e 0' A 12°3(s) " F e ( l ) + [ 0 ] 1 wt.% in Fe + A 12°3(s) 2 - 6 ' 3 ' 123. and have shown that hercynite would precipitate in the presence of alumina at oxygen content of 0.058 wt.%. This reaction could also explain the presence of both phases in the inclusions. The reaction, however, was obtained in the absence of dissolved aluminum in the iron liquid. As this i s not the case in our system,the reaction is not applicable. Thermodynamic restrictions do not seem to play a major role in the second phase precipitation in the Fe^O-Al system. It is therefore, believed that other factors, mainly the supersaturation needed for nucleation of the various phases, are also involved in determining the most favourable phase to precipitate. . .4.2.1.3 Analysis in Terms of Supersaturation Needed for Nucleation Previous discussion would lead us to believe that the second phase precipitation involves an important nucleation step. This is often the case when the melt does not contain particles or surfaces on which i t is relatively easy for the equilibrium precipitation product to form. Hence, the precipitation reaction cannot proceed until the system becomes sufficiently supersaturated to cause an oxide phase to form homogeneously. Supersaturation can be obtained in our system by: 1. cooling a melt which equilibrated with alumina at a given temperature 2. addition of aluminum, oxygen or both to the melt 3. segregation of oxygen and aluminum in the residual liquid during freezing. 124. In our system supersaturation may have been obtained by a l l three mechanisms. The system was found to be supersaturated even with respect to the highest temperature obtained (mechanism 2). The system was cooled from 1700°C to 1538° (mechanism 1) and allowed to freeze (mechanism 3). 4.2.1.3.1 Supersaturation by Additions and Cooling The degree of supersaturation of a melt is often given by the supersaturation ratio. Q /K'''. , In the Fe-O-Al system : 2 3 Q (f...• wt.%Al) s.s . ( f n • wt.% 0) s.s. s AI v K'" ( f A 1 - wt.%Al)2eq • ( f 0 • wt.% 0)3eq 36 53 Turpin and E l l i o t using Volmer and Weber's classical theory of nucleation suggested that the supersaturation sufficient to overcome the barrier,imposed by the int e r f a c i a l tension of the metal/oxide sur-face ,can be expressed in terms of: = - 4 - 5 7 6 T l D g ( ' ~ 2 ' 7 V ( ( ° F e - o x i d P 3 / k T l o * A > 1 / 2 (Q /K'11-) represents the c r i t i c a l supersaturation required to create a certain number of nuclei, of a given phase,per unit time, per unit volume of liq u i d . The results are very insensitive to whether one 3 chooses a rate of 1 or 10000 nuclei per cm ,per sec. For convenience Turpin and E l l i o t used1 1 nucleus per cm-*,per sec. Turpin and E l l i o t t applied t;he classical treatment of nucleation to the Fe-0-Al system, and were able to determine the effect of 125. supersaturation on the nucleation of the various oxides. Upon addition c r i t of AF, to the equilibrium free energy for reactions 2.6.1 and 2.6.II.. horn they were able to demonstrate (Figure 15) the effect of interfacial ten-sion on the oxide precipitation. They obtained the following values for the in t e r f a c i a l tensions: 2 °Al 20 3-Fe - 2 4 0 0 e r 8 s / c m aherc-Fe = 1600-1800 ergs/cm2 2 a ' _ = 250 ergs/cm < ; Feo-Fe & For these values of interfacial tensions i t is d i f f i c u l t to determine (Figure 15) which of the phases,alumina or hercynite is more favourable for nucleation,as both lines are very close together. The supersaturations required for the precipitation of either phase at the above mentioned interfacial tensions were never attained by any of our melts. Therefore, according to this theory, no inclusions should have precipitated in the l i q u i d before freezing. This, as men-tioned previously, was not the case. It is possible, however, that 2 39 a , • „ is nearer 1000 ergs/cm . If this i s true then i t is possi-herc-Fe ble to explain sin terms of this theory^why hercynite precipitated in ingots 1 and 2. The rest of the results, however, cannot be interpreted in terms of this theory, The treatment, therefore, is only useful in suggesting a qualitative condition under which hercynite may become a more favourable phase to precipitate out of the Fe-0-Al system. 126. 4.2.1.3.2 Supersaturation During Solidification 4.2.1.3.2.1 Introduction 37 Forward and E l l i o t t considered the possibility of second phase nucleation due to supersaturation during s o l i d i f i c a t i o n . They postulated that the development of conditions for homogeneous nuclea-tion during s o l i d i f i c a t i o n may be relatively easy for two reasons: 1. The dendrites of the solid iron growing through the melt should tend to subdivide the remaining liquid into nu-merous small volumes,and thereby isolate many of these volumes from particles that may be i n i t i a l l y present i n the liquid,and which might act as substrates for hetero-geneous nucleation. 2. The normal segregation of aluminum and oxygen into the residual liquid potentially could provide a very high supersaturation. Forward and E l l i o t t assumed uniform li q u i d composition and no aluminum and oxygen diffusion in the solid iron,and were,thus ,able to write: J U L = d - f ) <*•-!> u s o Using K° = 0.92 for aluminum"'4 and 0.05 for oxygen"'"' they were able to show (Figure 32) the concentrations of aluminum and oxygen to be reached in the residual liquid,for various fractions of metal solid fied. The plot assumes the starting composition to be the equilibrium deoxidation line of the liquidus temperature; the arrows showing the 37 direction of the change in composition. Forward and E l l i o t then 39 superimposedAFj^^ obtained from Turpin and E l l i o t t analysis,and were 127. Figure 32. Composite nucleation-segregation diagram during s o l i d i f i c a t i o n in Fe-O-Al system. 128. thus able to estimate the point during s o l i d i f i c a t i o n at which an oxide phase might be expected to appear.by homogeneous precipitation. The curves also show which phase should predominate in the competition for nucleation. 4.2.1.3.2.2 Comparison With Analytical Results The i n i t i a l aluminum and oxygen concentrations in our melts, assuming no nucleation occurred above the liquidus temperature,were well above those indicated by Figure 32. Upon s o l i d i f i c a t i o n , therefore, the aluminum and oxygen concentration moved up further than the figure suggests. It is,thus,very l i k e l y that the i n i t i a l supersaturation for nucleation of alumina and/or hercynite was surpassed in the early stages of s o l i d i f i c a t i o n . 2 If, for example, a,. n ._ = 2000 ergs/cm ,and a , _ = 1500 Al„0„-Fe herc-Fe 2 ergs/cm sthen most of the analytical results can be explained. Upon so l i d i f i c a t i o n the melts of ingots 1-4 and 8 would be expected to pre-cipitate hercynite i n i t i a l l y , followed ultimately by alumina precipita-tion. This was the case with ingots 1, 2 and 8. The melts of ingots 9 to 11 would precipitate alumina initially,and in the latter stages of s o l i d i f i c a t i o n hercynite would also become favourable to nucleate. This agrees exactly with the analytical results. The presence of FeO as the major second phase in ingot 4 cannot be explained by this theory. It is interesting to note that oxygen segregates very strongly during s o l i d i f i c a t i o n whereas aluminum does not. Accordingly, for homo-geneous nucleation the c r i t i c a l supersaturation arises chiefly because of increase in oxygen in the liquid. As a consequence, nucleation of 129. phases of r e l a t i v e l y high oxygen content i s favored,whereas much of the aluminum i s locked up i n the s o l i d i f i e d structure. There i s a f u r t h e r reason to believe that the i n c l u s i o n s pre-c i p i t a t e d homogeneously during s o l i d i f i c a t i o n . Work done on a s t a i n -l e s s s t e e l (Type SAE 304) ESR ingot showed that i n c l u s i o n s appeared i n the i n t e r d e n d r i t i c p o s i t i o n s only. In the present study i n c l u s i o n alignment (Figure 33) suggests that t h i s i s also the case f o r i r o n ingots. This suggests that the i n c l u s i o n s were formed i n the l a t t e r stages of the s o l i d i f i c a t i o n process. I t i s also p o s s i b l e that the i n c l u s i o n s were pushed into the i n t e r d e n d r i t i c positions by the advancing dendrites, 37 however, t h i s was shown to be very u n l i k e l y . Supersaturation, obtained by cooling, aluminum a d d i t i o n and s o l i d i f i c a t i o n , w a s shown to be s u f f i c i e n t to cause homogeneous nucleation upon s o l i d i f i c a t i o n . Most of the a n a l y t i c a l r e s u l t s can also be ex-plained i n terms of the c l a s s i c theory of nucleation which states that the b a r r i e r i s the work necessary to the formation of the new i n t e r f a c e between the l i q u i d metal and the newly formed oxide. It also gives a reason why ESR i n c l u s i o n s are t y p i c a l l y so small, as upon s o l i d i f i c a t i o n l i t t l e time or material i s a v a i l a b l e f o r growth. 4.2.2 P r e c i p i t a t i o n from Ni-0-Ca-Al compositions L i t t l e i s known concerning the p r e c i p i t a t i o n from Ni-0-Ca-Al a l l o y s . Disregarding the nucleation phenomena i t i s expected, though, 130. Figure 33. Inclusion Alignment (x800) 131. from Figure 17 that CaO would precipitate preferentially to alumina since the CaO equilibrium lines l i e below those of alumina. The inclusions in ingot 5 were a l l alumina as no calcium was present in the ingot. In ingot 6 which contained 65 ppm calcium, 160 ppm aluminum,and 31 ppm oxygen surprisingly^no CaO was detected. The inclusions were analyzed to be alumina and Ca/Ni second phase. Hence, i t is possible that a nucleation probl .1 does exist, and that CaO pre-cipitation requires greater supersaturation than alumina precipitation. Ingot 7 which contained 100 ppm calcium and k ppm oxygen, also had no CaO inclusions; only Ca/Ni second phase was detected. This, however, may be entirely due to the. fact that very l i t t l e oxygen was present in the ingot. 132. CHAPTER V CONCLUSION It was proposed in this work to determine whether D-type i n -clusions are indeed a necessary component in the ESR ingots. In relation Co this the following conclusions can be reached. Qualitatively,the metal compositions follow the equilibrium path. This is clearly demonstrated by aluminum deoxidation experiments where aluminum additions caused the oxygen contents of the iron to be decreased. In considering the results quantitatively,the large excess of aluminum, calcium and oxygen contents of the ingots, well beyond those prediced by chemical equilibrium, lead us to believe that the elements are not deposited by a simple chemical slag/metal interaction. It is most probable that both chemical and electrochemical slag/metal reactions are responsible for the fin a l metal composition. Hence, as long as the metal i s in contact with the slag i t i s impossible to obtain an inclusion-free ingot since slag elements w i l l deposit in the metal, by combinations of chemical and electrochemical mechanisms. 133. As the work was carried out on pure iron containing aluminum and oxygen,and on pure nickel containing aluminum, calcium and oxygen, i t is l i k e l y that the results also relate to a l l steels and nickel-base materials. In studying the effect of the liquid metal composition on the resulting inclusions i t was shown that the fi n a l inclusion composition cannot be explained by the classical nucleation theory applied to the liquid pool. According to the theory insufficient supersaturation was obtained in the liquid pool to allow of nucleation. In the latter stages of s o l i d i f i c a t i o n , however, the supersaturation required by the theory is achieved. Thus the probable inclusion nucleation and growt] site is within the freezing interface. This,in turn,explains why the inclu-sions found were small (D-type), as l i t t l e time or material is available for their growth. Hence,the inclusion", in ESR ingots are expected to be of the small globular oxide D-type, and are a necessary result of electroactive slag/metal interface reactions inherent in the process. 134. BIBLIOGRAPHY . 1. D i s t i n P.A., Whiteway C.R. and Masson C.R., Con. Met. Quarterly, 10, No. 1, 13 (1971) . 2. Chipman J . and E l l i o t t J.F., E l e c t r i c Furnace Steelmaking Vol. I I , Interscience Publishers, N.Y. (1963). 3. Korber F., Stahl and Eisen, 54, 535 (1934). 4. Swisher J.H., Trans A.I.M.E., 239, 123 (I, 1967). 5. Gokcen N.A. and Chipman J . , Journal of Metals, 173 ( I I , 1953). 6. Hopp H.V., A.f.d. Eisenhuttenwesen, _40, 205 (IV, 1969). 7. D'Entremont J . C , Guernsey D.L. and Chipman J . , Trans A.I.M.E., 227, 14 ( I I , 1963). 8. H i l t y D.C. and Crafts W., Trans A.I.M.E., 188, 414 ( I I , 1950). 9. McLean H. and B e l l H.B., J.I.S.I., 203, 123 ( I I , 1965). 10. H i l t y D.C. and Crafts W., Trans A.I.M.E., 188, 425 (1950). 11. Gokcen N.A. and Chipman J . , Trans A.I.M.E., 194, 171 (1952). 12. Chipman J. and Chang L.C., Trans A.I.M.E., 185, 191 (1949). 13. Littlewood R., Trans A.I.M.E. 233, 772 (IV, 1965). 14. Kuo Chu-kun and Yen Tung-Shung, Acta Chim. S i n i c a , 30, 381 (1964). 15. Burel B., M.A; Sc. Thesis, Dept. of Metallurgy, U.B.C (VII, 1969). 16. ^Mitchell A. and Joshi S., P r i v a t e communication, Dept. of Metallurgy, U.B.C. (1971). 17. Hoyle G., Dewsnop P., S a l t D.J. and Bares E.M., BISRA Open Report, 33 (1966). 18. Zhmoidin G.I., I z v e s t i c a Met. Navk, SSSR, No. G, 9 (1969). 135. 19. Gutt W., Chattergee A.K. and Zhmoidin G.I., Journal Met. Sc., _5, 360 (1970). 20. E l l i o t t J.F., Gleiser M. and Ramakrishna V., Thermochemistry for Steelmaking, Addison-Wesley (1963). 21. Baak T., Acta chem. Scand., 8, No. 9, 1727 (1954). 22. Budnikov P.P. and Tresvyatskii S.G., Dokoladii Acad. Nauk., SSSR, 89, No. 3, 479 (1953). 23. Korpachev V.G., Sryvalin I.T. and Buryler B.P., Tremodin Svoistva Rasplanov, U.S.S.R, 87 (1969). 24. M i t c h e l l A., Trans Faraday S o c , 63, 1408 (1967). 25. M i t c h e l l A. and B e l l M.M., some observations on the surface quality of electroslag ingots, Accepted by J.I.S.I., s e r i a l No. 3788. 26. Sponsellor D.L. and F l i n n R.A., Trans A.I.M.E. 230, 776 (VI,1964) 27. Kubaschewski 0. .and Evans E., Me t a l l u r g i c a l Thermochemistry, Perganon Press, N.Y. (1958). 28. Wentrap H. and Hieber G., A.f.d. Eisenhuttenwesen, _13, 15 (1939). 29. Sawamura H. and Sanok., Subcommittee for physical chemistry of steelmaking, 19th committee, 3rd d i v i s i o n , Japan Society for the promotion of science. Special Report No. 3 (XII, 1963). 30. McLean A. and Ward R.G., Journal of Metals, 17_, 526 (1965). 31. Rohde L.W., Choudhury A. and Wohlsten M., A.f.d. Eisenhuttenwesen, 42_, 1966 ( I I I , 1971). 32. Wagner C., Thermodynamics of A l l o y s , Addison-Wesley, 51 (1952). 33. Buzek Z. and Hutla A., S c i e n t i f i c Report, Met. tech. Inst. (Ostrava), 11, No. 3, a r t i c l e 180 (1965). 34. Paton B.E., Electroslag Welding, Am. Welding S o c , N.Y., 22 (1962). 35. Joshi S., private communication, Dept. of Metallurgy, U.B.C. (1971). 36. Turpin M.L. and E l l i o t t J.F., J.I.S.I., 204, 217 (1966). 136. 37. Forward G. and E l l i o t t J.F., Journal of Metals, 54, (V, 1967). 38. Woehibier F.W., Rengstorff G.W., Journal of Metals, 50, (V, 1967). 39. McLean A. and Ward R.G., J.I.S.I., 8 ( I , 1966). 40. Chipman J . , Trans A.I.M.E., 224, 1288 (1962). 41. Cambell J . , Journal of Metals, 23 (VII, 1970). 42. Samarin A.M., Jernkont. Ann. 151, 181 (1967). 43. Hauffe K. and Wagner C., Z. Electrochem., j46, 160 (1940). 44. Etienne M., Ph.D. Thesis, Dept. of Metallurgy, U.B.C. (X, 1970). 45. Gutt W. and Osborne G.J., Trans B r i t . Ceramic S o c , 65_, 521 (1966). 46. Cameron J . , Etienne M. and M i t c h e l l A., Met. Trans. 1, 1839 (VII, 1970). 47. Godfrey K.V. and Downs W.E., Neutron a c t i v a t i o n and on stream analysis i n the mineral i n d u s t r i e s using Sherlock I I I . B.C. Research (X, 1970). 48. Bohnstedt U., Z. Anal. Chem., 199, 109 (1964). 49. M i t c h e l l A. and Etienne M. , Trans A.I.M.E. 7A1_, 1462 ( V i i , 1968). 50. Beynon G., P r i v a t e communication, Dept. of Metallurgy, U.B.C. (1971). 51. Kozakevitch P. and Lucas L.P., Revue de Met., 589 (IX, 1968). 52. Beynon G. and M i t c h e l l A., Electrode p o l a r i z a t i o n i n the D.C. e l e c t r o s l a g melting of pure i r o n . , Dept. of Metallurgy, U.B.C. (1971). 53. Volmer M. and Weber A., Z. Phys. Chem., 119, 277 (1926). 54. Bogdandy L., A.f.d. Eisenhuttenweseu, J34_, 235 (1963). 55. Hepworth M.T., Smith R.P. and Turkdogan E.T., Trans A.I.M.E., 236, 1278 (1966). 137. APPENDIX I Extent of Reaction 2.5.XX Two reactions are actually possible which may cause excess oxygen in iron through CaO dissociation: (CaO) = (Ca) + [0] ± ^ ± n p e AF° = 124,850 - 26.5 T or, (CaO) = C a ( g ) ; + [0] l ^ ± n p e AF° = 159,900 - 46.4 T reaction 1 w i l l be considered f i r s t . since AF° =-RT ln at 1600°C K x = 2.2 x 10~9 hence (a C a) [wt.%0] 2.2 x 10"9 = ; r ( a C a 0 ) assuming a ^ = 1 and a ^ = N ^ 2.2 x 10~9 = (N„ ) [wt.% 0] Ca The amount of calcium dissolved in the: slag is related to the amount of oxygen dissolved in the iron by the CaO stoichiometry of 40/16. Thus,it is possible to write: 2.2 x i o " 9 = • ( M ; w - ° f f r 5 1 ^ ) ) • [ - f - ( i o o ) ] 40 16 wt. of slag W X = wt. (gm) of dissolved oxygen in W gm iron MW of slag = 85 gm/mole wt. of slag = 400 gm 138.a It is now possible to calculate and plot (Figure A.I.I) the amount of dissolved oxygen versus the amount of iron (W) considered' to react with the slag. W is believed to be in the range of 50 to 200 gm. From Figure A.I.I i t can be seen that a maximum of 5 ppm oxygen would thus result. This is a negligible amount of oxygen. The calculation, however, depends upon the calcium being soluble in the slag and having a low vapor pressure. This is in fact true as the -4 resulting mole fraction of calcium in the slag is 4.6x10 , giving -3 vapor pressure of only 1.8x10 atm. Thus the calcium w i l l remain in the slag and reduce the oxygen deposition in iron by mass action. From this argument i t is clear that the extent of reaction 2 is even less than reaction 1,and,thus,also may be considered to contribute a negli-gible amount of oxygen to the metal. 138b Figure A.I.I. Oxygen deposited (by reaction 1) dependence on the amount of liquid iron interacting with the slag. 139. APPENDIX II E l e c t r o s l a g Furnace Running Conditions Ingot Electrode Slag (wt.%) Current Voltage Melt Rate No. (CaF„/CaO/Al 0 ) Amps(rms) (Volts(rms) gm/sec 1 FVE (75/0/25) 650 24.2 2.21 2 FVE (93/0/7) 650 21.5 2.40 3 FVE (70/6.5/23.5) 600 23.1 2.92 4 FVE (60/38.3/1.7) 865 23.8 3.26 5 Ni c k e l (85/0/15) 690 24.6 3.42 6 Ni c k e l (60/38.3/1.7) 610 22.2 2.68 7 Ni c k e l (75/25/0) 700 24.2 3.20 8 ARMCO (75/0/25) 825 25.0 2.90 9 ARMCO (75/0/25) 560 28.0 3.32 10 FVE (75/0/25) 800 25.1 2.88 11 FVE (75/0/25) 650 24.9 2.92 12 ARMCO (75/0/25) 675 25.5 3.14 13 ARMCO (60/38.3/1.7) 810 23.5 2.82 14 ARMCO (75/0/125) 800 27.7 3.28 140. APPENDIX III Relation Between Phase Shift'and Reactance i n the C i r c u i t A D.C. resistance (r) of 0.64 mft was measured over a se c t i o n of the o v e r a l l ESR u n i t c i r c u i t , a s shown i n Figure 24. Over the same sec t i o n A.C. reactance (Z) of 1.4 mfi was measured. This i n d i c a t e d the presence of impendence (L) i n the c i r c u i t . I t w i l l now be shown that the 10° phase s h i f t observed between the current and the voltage i s i n d i c a t i v e of an impedence which when added to the resistance w i l l r e s u l t i n the measured reactance, as: tan Q = wL/R = 2 TT 60 L/R and 0 = 10° thus, _L_ . -7 1 — 4.7 x 10 x K where L = o v e r a l l impedence R = o v e r a l l resistance = 20V 500A = 4 x 10" 2 Q. hence, (from 1) L = (4.7xl0~ 7) (4xl0~ 2) H = 1.9xl0" 5 H The impedence (1) i n the part of the c i r c u i t measured i s about L/5 . 141. hence, 1 = 3.8xl0" 6 H It i s now possible to determine the resistance in the part of the c i r c u i t measured. Z = ( r 2 + x 2 ) 1 / 2 - ( r 2 + w 2 l 2 ) 1 / 2 = (3.6xl0~ 7 + 1 . 9 x l 0 _ 6 ) 1 / 2 = 1.5 mfi This compares closely to the 1.4 mfi measured. 142. APPENDIX IV Inclusion Analysis Done on Ingot 1 The inclusions were analyzed by the following methods: 1. Optical observation of inclusions in s i t u (Figure A.IV.l). 2. Electron microprobe analysis of inclusions in s i t u (Figure A.IV.2). 3. X-ray diffraction analysis on extracted inclusions (Figure A.IV.3). 4. Atomic absorption analysis on extracted inclusions for the determination of their iron content. 1 Optical Observation Figure A.IV.l. Inclusions in the (xlOOO) iron matrix 143. Figure A.IV.2. Aluminum and oxygen in the inclusions (xl300) 144. 3 X-ray Analysis Intensity 9 ( ° ) d(A°) phase > \ svery weak \ \ \ \ \ \ \ \ \ x w e a k \ \ \ \ ^strong > x \ \ \ \ \ xvery w< ^\ \ v eak V \ \ \ \ vstrong \ \ \ »strong \ ^strong 18.05 2.19 hercynite 18.45 2.83 alumina 19.15 2.73 alumina 21.35 2.46 hercynite 22.70 2.32 alumina 23.35 2.26 alumina 26.30 2.02 he cynite 26.70 1.99 alumina 34.20 1.57 hercynite 35.35 1.55 alumina 38.30 1.44 hercynite 39.50 1.41 alumina 40.00 1.39 alumina 145. 4 Atomic absorption Analysis A solution containing 69.8 mgm of dissolved inclusions per l i t r e of solution 5was measured to contain 15.2 mgm Fe per l i t r e of solution. Thus, i f a l l the iron was present as FeOA^O^ the amount of FeO-A^O^ inclusions would be: / i c o -n / i \ (174gm/mole hercynite) l n „ , . . ... (15.2 mgm Fe/1) c, z = 1 = 47.2 mgm hercynite/1 56 gm/mole Fe The remaining would be alumina inclusions: 69.8 mgm inclusions/1 - 47.2 mgm hercynite/1 = = 22.6 mgm alumina/1. The inclusion composition can thus be calculated to contain 6.75 wt.% hercynite and 32.5 wt.% alumina. I 146. APPENDIX V Kinetics of Drop/Slag Equilibration It is possible to show that i f the residence time of the metal drop in the slag is 0.1 sec,the flux of aluminum, oxygen and calcium through the drop/slag interface could easily allow chemical equilibra-tion between the two. 44 Etienne showed that the flux of the elements across the inter-•~ A 2 face is approximately 10 moles/sec.cm . Thus, i f aluminum is con-sidered: —A 2 Molar Flux in 0.1 sec = (10 moles/sec.cm ) (0.1 sec) = 10 moles/cm2 -5 2 Flux of Al in gms = (10 moles/cm ) (27 gm/mole) in 0.1 sec _^ 2 = 2.7x10 gm/cm 2 Surface area of metal drop = 4 TT (0.5 cm) = 3.14 cm2 ~*A 2 2 Flux of Al through surface = (2.7x10 gm/cm ) (3.14 cm ) in 0.1 sec , = 8.5x10" gm 9 3 Wt. of metal drop = 4/3 TT (0.5 cm) (7.1 gm/cm ) = 3.7 gm Final Al content in the drop after 0.1 sec in the slag 8.5xl0~ 4 gm (100%) 3. 7 gm = 0.023 wt.% = 230 ppm Al 147. The same c a l c u l a t i o n for calcium and oxygen gives the f o i l , v, ing amounts of each: Calcium = 340 ppm Oxygen = 154 ppm Hence, s u f f i c i e n t amounts of aluminum, oxygen and calcium ca deposit to allow chemical e q u i l i b r a t i o n of the metal drop with the sl a g . This assumes that the reaction rates are f a s t at the tempera-tures concerned ( -1700°C). 

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            data-media="{[{embed.selectedMedia}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
https://iiif.library.ubc.ca/presentation/dsp.831.1-0078782/manifest

Comment

Related Items