UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

The influence of interfacial turbulence on the rate of oxidation and deoxidation of molten copper and… Barton, Robert Glen 1976

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

Item Metadata

Download

Media
831-UBC_1976_A6_7 B37.pdf [ 10.75MB ]
Metadata
JSON: 831-1.0079076.json
JSON-LD: 831-1.0079076-ld.json
RDF/XML (Pretty): 831-1.0079076-rdf.xml
RDF/JSON: 831-1.0079076-rdf.json
Turtle: 831-1.0079076-turtle.txt
N-Triples: 831-1.0079076-rdf-ntriples.txt
Original Record: 831-1.0079076-source.json
Full Text
831-1.0079076-fulltext.txt
Citation
831-1.0079076.ris

Full Text

THE INFLUENCE OF INTERFACIAL TURBULENCE ON THE RATE OF OXIDATION AND DEOXIDATION OF MOLTEN COPPER AND SILVER USING LOW-MOMENTUM VERTICAL GAS JETS by ROBERT GLEN BARTON B.Sc.(Honours) Chemistry, U.B.C. 1970 A THESIS SUBMITTED IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE in THE FACULTY OF GRADUATE STUDIES in the Department of Metallurgy We accept this thesis as conforming to the required standard THE UNIVERSITY OF BRITISH COLUMBIA May, 1976 In presenting th i s thes is in pa r t i a l fu l f i lment of the requirements for an advanced degree at the Un ivers i ty of B r i t i s h Columbia, I agree that the L ibrary sha l l make it f ree l y ava i l ab le for reference and study. I fur ther agree that permission for extensive copying of th is thesis for scho lar ly purposes may be granted by the Head of my Department or by his representat ives. It is understood that copying or pub l i ca t ion of th is thes is for f i nanc ia l gain sha l l not be allowed without my wr i t ten permiss ion. Depa rtment The Univers i ty of B r i t i s h Columbia 2075 Wesbrook Place Vancouver, Canada V6T 1W5 7 ABSTRACT The rate of oxidation of 99.99% and 99.999% pure copper samples at 1220°C by low-momentum jets of pure oxygen has been studied at gas flow rates 3 -1 of from 500 to 2000 cm min . Oxidation rates at a given gas flow rate were found to be constant and were governed by starvation mass transfer kinetics. Factors studied for the reaction include: effect of lance height, effect of small additions of si l i c o n and sulphur to the melt prior to oxidation, and effect of oxide patch area. Interfacial tension-generated flow, radially outward from the point of jet impingement, was observed during oxidation and surface velocity studies showed that such flow had a mean value of 26.1 ± 5.5 cm sec ^ for a l l of the experiments and was independent of oxygen gas flow rate and copper bath oxygen concentration. Surface-blockage studies indicated that the bulk of the oxygen transfer to the copper occurred over the area described by the oxide patch. Liquid-phase oxygen mass transfer coefficients were calculated using the oxidation rates and oxide patch areas, and a mean value was found to be 0.104 ± 0.012 cm sec \ independent of oxygen flow rate, bath oxygen content, and dissolved sulphur and si l i c o n contents. The rate of oxidation of 99.995% pure silver at 1100°C was studied 3 using low-momentum jets of pure oxygen at flow rates of 1500 and 2000 cm min \ and was found not to be governed by starvation mass transfer kinetics. - i i -The oxidation rate was not dependent on oxygen gas flow rate and was found to be a factor of about 50 times less than those observed for copper. No spontaneous interfacial tension-generated flow was observed during oxidation of the molten silver and a possible explanation was postulated. Liquid-phase oxygen mass transfer coefficients were found to have a mean value of -3 -1 2.88 ± .41 (10 ) cm sec , independent of gas flow rate and bath oxygen content. The effect of interfacial turbulence on liquid-phase oxygen mass transfer coefficients in molten copper was to enhance the value by about 40 times over that observed in molten silver, where interfacial turbulence does not occur on oxidation. Copper deoxidation at 1220°C using low-momentum jets of pure hydrogen 3 -1 at flow rates of 1500 and 3000 cm min was studied, and was found not to depend on hydrogen flow rate, lance height, and starting oxygen concentration. The rate-controlling step was found to be the gas-phase mass transfer of hydrogen to the liquid surface for the f i r s t 3000 sec. of deoxidation. After this, liquid-phase oxygen mass transport control predominated. Dissolved si l i c o n was found to retard the deoxidation rate, while dissolved sulphur was found to enhance the deoxidation rate through continued SO2 elimination. Interfacial tension-generated flow was observed during de-oxidation and approximate surface velocities of 10 to 15 cm sec ^  towards the point of jet impingement were observed. A mechanism for this flow was postulated. The gas-phase mass transfer coefficient was found to be 1.28 ± 0.25 cm sec ^  for copper-oxygen alloys, and was 0.89 cm sec in the presence of dissolved si l i c o n and 2.68 cm sec ^  in the presence of dissolved - i i i -sulphur. An approximate value for the liquid-phase oxygen mass transfer coefficient in the liquid-phase control region was found to be 4.9 (10 ) cm sec , and was found to be influenced by the presence of bubbling during this phase of deoxidation. The rate of deoxidation of molten silver at 1100°C by low-momentum 3 -1 hydrogen jets was studied at hydrogen flow rates of 1500 and 2000 cm min The rate was found not to depend upon hydrogen flow rate, but was found to decrease with decreasing starting bath oxygen concentration. Interfacial tension-generated flow was observed, during silver deoxidation, with approximate surface velocities of 10 to 15 cm sec towards the point of jet impingement. The rate-controlling step was found to be liquid-phase mass transfer of oxygen, and liquid-phase oxygen mass transfer coefficients were found to decrease with decreasing i n i t i a l oxygen content. These values were enhanced by the presence of bubbling during deoxidation. Interfacial turbulence during the dissolution of solid C^S, C^O, Se, and Te in molten copper was shown to occur. Values calculated for the spreading coefficient S, indicated that the spreading of these materials on molten copper was predictable. - iv -TABLE OF CONTENTS PAGE ABSTRACT i TABLE OF CONTENTS .. . . iv LIST OF FIGURES .. v i i i LIST OF TABLES x i i LIST OF SYMBOLS x i i i ACKNOWLEDGEMENTS xv 1. INTRODUCTION 1 1.1. General 1 1.2. Spontaneous Interfacial Motion 2 1.2.1. Solute Effects on Interfacial Tension 2 1.2.2. Mechanism for Spontaneous Motion 7 1.2.3. Mass Transfer Coefficient .. .. 12 1.3. Review of Previous Work .. .. 13 1.3.1. Observations of Spontaneous Interfacial Motion . . . . 13 1.3.2. Gas Jets Impinging on Liquids 18 1.3.3. Kinetics of Molten Copper Oxidation and Deoxidation .. 18 1.3.4. Kinetics of Molten Silver.Oxidation and Deoxidation .. 25 1.4. Scope of the Present Study .. 26 2. EXPERIMENTAL 28 : 2.1. Apparatus 28 2.2. Materials . 31 - v -PAGE 2.2.1. Copper 31 2.2.2. Silver 31 2.2.3. Selenium and Tellurium .. 31 2.2.4. Reagents 31 2.3. Preparation 33 2.3.1. Copper 33 2.3.2. Silver 33 2.4. Procedure 34 2.4.1. Copper Experiments 34 2.4.2. Silver Experiments 35 2.4.3. Experiments with Sulphur 35 2.4.4. Experiments with Se, Te, Cu^S and C^O solids . . . . 35 2.5. Analysis 36 2.5.1. Sample Preparation 36 2.5.2. Oxygen and Sulphur 36 2.5.3. Oxide Patch Size and Surface Velocity 36 3. RESULTS 38 3.1. Copper Oxidation 38 3.1.1. Observations .. 38 3.1.2. Oxidation Rates 42 3.1.2.1. Bulk-Phase Mixing 46 3.1.2.2. Silicon Experiments 46 3.1.2.3. Sulphur Experiments 50 - v i -PAGE 3.1.3. Surface Velocity Studies 50 3.1.4. Oxide Patch Areas 55 3.1.5. Surface Blockage Studies 55 3.2. Silver Oxidation 60 3.2.1. Observations 60 3.2.2. Oxidation Rates 65 3.3. Copper Deoxidation 65 3.3.1. Observations 65 3.3.2. Deoxidation Rates 67 3.4. Silver Deoxidation .. 72 3.4.1. Observations 72 3.4.2. Deoxidation Rates 75 3.5. Solid Dissolution Experiments 75 3.5.1. Observations 75 4. DISCUSSION 79 4.1. Copper Oxidation 79 4.1.1. Oxidation Rates 79 4.1.1.1. Bulk-Phase Mixing 80. 4.1.1.2. Silicon Experiments 8Q 4.1.1.3. Sulphur Experiments 81 4.1.2. Surface Velocity Studies .. .. 83 4.1.3. Oxide Patch Areas 87 4.1.4. Surface Blockage Studies 88 4.1.5. Mass Transfer Coefficients 88 - v i i -PAGE 4.2. S i l v e r Oxidation 93 4.2.1. Oxidation Rates .. .. 93 4.2.2. Surface Motions 94 4.2.3. Mass Transfer C o e f f i c i e n t s 95 4.2.4. Influence of I n t e r f a c i a l Turbulence on Liquid-Phase Mass Transfer C o e f f i c i e n t s 99 4.3. Copper Deoxidation 102 4.3.1. Deoxidation Rates 102 4.3.1.1. E f f e c t of Dissolved S i l i c o n and Sulphur 103 4.3.2. Surface Motions 104 4.3.3. Mass Transfer C o e f f i c i e n t s 105 4.4. S i l v e r Deoxidation I l l 4.4.1. Deoxidation Rates I l l 4.4.2. Surface Motions 113 4.4.3. Mass Transfer C o e f f i c i e n t s 113 4.5. S o l i d D i s s o l u t i o n Experiments 115 4.5.1. Spreading 115 4.5.2. Significance 118 5. CONCLUSIONS 119 6. SUGGESTIONS FOR FUTURE WORK 123 BIBLIOGRAPHY 125 APPENDIX I 129 - v i i i -LIST OF FIGURES FIGURE PAGE NO. 1.1. Surface tensions for solutions of C, P, N, 0, S, and Se in liquid Iron at 1550°C (KOZAKEVITCH9) 3 1.2. Surface tensions for Aluminum containing various solute me tals at 50° to 80° ab ove the liquidus (KOROL'KOVlO) 4 1.3. Effect of dissolved oxygen on the surface tension of molten copper (MONMA and SUTOH) • • 5 1.4. Effect of dissolved sulphur on the surface tension of molten copper. (MONMA and SUTOH) 6 1.5. Interfacial motion generated on a Micro scale due to eddy penetration • ^ 1.6. Interfacial motion generated on a Macro scale by the presence of a partially Immersed solid 10 1.7. Effect of the presence of a non-transferring surface-active solute, in an interface, on the spreading due to a transferring surface active solute H 1.8. Schlieren photographs of interfacial turbulence induced by a stream of 492 m M F e 3 + solution from a syringe onto a 0.1% In Amalgam. The bulk aqueous phase is 50 m M Fe 3 +. Width of f i e l d i s 11 mm (Brimacombe and Richardson^) 15 1.9. Interfacial turbulence on liquid t i n and liquid iron during blowing with oxygen gas at a flow rate of 120 cm3 min~l ^ 1.10. Results for a 0.5-0.54 gm droplet of copper levitated in a gas stream of C0:C02 ratio 1:25 at a flow rate of 4160 cm3 min - 1 (TOOP and RICHARDSON35) 2 1 - ix -PAGE 1.11. The uptake of oxygen from mixtures of O2 and N 2 at 1400°C by both desiliconized, 0, and untreated, A, copper drops. 1 R .. PQ 2 = 1.11 (lCT 4) a tin. (GLEN and RICHARDSON ) . . . . 23 1.12. Gross-sectional view of experimental apparatus . . . . 29 3.1. Interfacial turbulence during absorption of oxygen by molten copper .. ,. • 39 3.2. Growth of oxide patch during an oxidation experiment .. .. 40 3.3. Temperature-time plot for Run 41 showing temperature rise during oxidation. A l l other runs exhibited similar behaviour .. 41 3.4. Concentration of oxygen in copper as a function of time for four different oxygen blowing rates.. .. 43 3.5. Oxidation-time relationship for pure copper experiments at 1500 cm^-min oxygen blowing r a t e . . . . 44 3.6. Oxidation-time relationship for Run 28 at 1500 cm min -^ oxygen blowing rate. Oxygen shut off at 256 sec. to check on bulk stirring properties .. 47 3.7. Comparison of the oxidation-time relationship for Run 25 (no Si added) and Run 41 (20 ppm Si added) , both at 1500 cur* min--*- oxygen flow rate.. .. 48 3.8. Oxidation-time relationship for copper containing dissolved sulphur at 1500 crrr* min--L oxygen flow rate 49 3.9. Surface velocity as a function of radial distance from the lance for Runs 22 and 23. Vertical lines show edge of oxide patch 51 3.10. Surface velocity as a function of radial distance from thei.lance for Runs 24 and 26. Vertical lines show edge of oxide patch. 52 3.11. Surface velocity as a function of radial distance from the lance for Runs 27 and 29. Vertical lines show edge of oxide patch 53 3.12. Surface velocity as a function of radial distance from the lance for Runs 30 and 41. Vertical lines show edge of oxide patch .. . . 54 Oxide patch areas as a function of copper bath oxygen concentration for four oxygen flow rates .. Oxide patch area as a function of copper bath oxygen concentration for four i n i t i a l sulphur concentrations a l l at 1500 cm3 min~l oxygen flow rate Top view of apparatus for reduced surface area experiments, showing the configuration of the alumina disc Surface of molten silver during oxidation Temperature-time plot for silver Run 4 showing experimental temperature. A l l other runs gave similar results Oxidation-time relationship for silver experiments for two oxygen flow rates Interfacial turbulence during the deoxidation of molten copper with hydrogen Deoxidation-time relationship for copper using Ar gas at 1500 cm3 min~l flow and hydrogen gas at 1000 cm3 min - 1 flow Deoxidation-time relationship for copper at two hydrogen flow rates .. Deoxidation-time relationship for copper Run 50, showing regions of gas-phase and liquid phase control Interfacial turbulence during the deoxidation of molten silver with hydrogen Deoxidation-time relationship for molten silver experiments at two hydrogen flow rates Interfacial turbulence during the dissolution of solid Cu2S .. .. Interfacial turbulence during the dissolution of solid surface-active agents Flow pattern in liquid caused by an impinging gas jet. Note stagnant areas, labelled S - x i -PAGE 4.2. Maximum surface velocity as a function of copper bath oxygen concentration . . 86 4.3. Liquid-phase oxygen mass transfer coefficient as a function of copper bath oxygen concentration 89 4.4. Liquid-phase oxygen mass transfer coefficient as a function of copper bath oxygen concentration 90 4.5. Liquid-phase oxygen mass transfer coefficient as a function of copper bath oxygen concentration showing the effect of dissolved s i l i c o n and sulphur .. 91 4.6. Mass transfer from a pure oxygen jet to molten silver 96 4.7. Schematic representation of stirring pattern and surface flow conditions,during deoxidation of molten copper and silver with gas 106 C FINAL 4.8. Plot of -In •  •°_..T.rrT,.rAT vs. time for copper C INITIAL v r o deoxidation showing change from gas-phase to liquid phase mass transport control of deoxidation at about 3000 sec 0.37 wt % 0) 110 4.9. Liquid-phase mass transfer coefficients for oxygen during the deoxidation of molten silver with H 0 gas 114 - x i i LIST OF TABLES TABLE CONTENTS PAGE I SPECTROSCOPIC ANALYSIS OF COPPER AND SILVER SAMPLES 32 II EXPERIMENTAL RESULTS FOR OXYGEN FLUXES DURING OXIDATION OF COPPER 45 III EXPERIMENTAL RESULTS FOR OXYGEN UPTAKE RATES OVER REDUCED SURFACE AREA 59 IV ABSORPTION AND MASS TRANSFER DATA FOR LIQUID Ag OXIDATION 64 V EXPERIMENTAL RESULTS FOR OXYGEN FLUXES DURING DEOXIDATION OF COPPER 71 VI EXPERIMENTAL RESULTS FOR OXIDATION OF SULPHUR-CONTAINING MOLTEN COPPER . . 82 VII COMPARISON OF EXPERIMENTAL k 0 VALUES FOR MOLTEN SILVER WITH THOSE FROM THE LITERATURE 98 VIII COMPARISON OF CALCULATED AND EXPERIMENTAL OXYGEN MASS TRANSFER COEFFICIENTS FOR COPPER AND SILVER.. 101 IX MASS TRANSFER COEFFICIENTS FOR THE DEOXIDATION OF MOLTEN COPPER AND SILVER 108 X CALCULATED VALUES FOR THE SPREADING COEFFICIENT, S, FOR Cu90, Cu9S, Se AND Te 117 - x i i i -LIST OF SYMBOLS Symbols Description 2 A interfacial area for mass transfer, cm B constant. C , C^, solute concentration; at the interface and in the bulk phase respectively, wt % or moles cm - J where indicated. s s s ^o' ' ^ 2 ' ^ H2° concentration of oxygen, hydrogen, and water in the interface and bulk phase, respectively, wt % or moles cm-^. Bulk phase specified. Ds^ diffusion coefficient for species s in metal m, m cm^  sec~l. K equilibrium constant. k g, k Q, kH2> ^H20 mass transfer coefficient for species s, oxygen, hydrogen, and water, respectively, cm sec~l. k Qy overall mass transfer coefficient, cm sec n , n , Ajj2» nH20 mass flux of species s, oxygen, hydrogen, and water respectively, moles sec~+. [0] concentration of oxygen in the liquid phase, wt % or moles cm-3 where indicated. PQ^kj P H ^ ' ^ H^O partial pressure of oxygen, hydrogen, and water, respectively, at the interface or in the bulk gas phase, atm. R gas constant, 81.94 cm^  atm. mole ^ °K ^ , r bath radius, cm. s, s , s t n solute species; transferring across the interface, and non-transferring, respectively. - XIV-S spreading coefficient, dynes cm ^ t surface renewal time, sec. e t time, sec. T temperature, °C or °K where indicated. U interfacial velocity of spreading, cm sec ^. Y Q activity coefficient of oxygen. 3 V volume of metal bath, cm . r surface excess of solute species s at the interface, molecules cm-2, surface tensioi oxide, respectively, dynes cm" 0 , 0 , 0 , 0  ion for a metal, copper, silver and metal m Cu Ag mo ., . » rr > - XV -ACKNOWLEDGEMENTS The author wishes to express h i s gratitude to Dr. J.K. Brimacombe for h i s help with the t h e o r e t i c a l aspects of t h i s study, and h i s aid with the photography during the experiments. The assistance of members of the departmental workshop i n the design and construction of the experimental apparatus i s also greatly appreciated. Thanks are also extended to those f a c u l t y and s t a f f members and graduate students who have helped with the experimental and t h e o r e t i c a l aspects of t h i s i n v e s t i g a t i o n . F i n a n c i a l support from the National Research Council i n the form of a research assis t a n t s h i p i s g r a t e f u l l y acknowledged. 1. INTRODUCTION 1.1. GENERAL Spontaneous interfacial turbulence has long been of interest to the surface chemist and chemical engineer. Pioneering work by Thompson''" in 1855, 2 in alcohol-water systems, and later by Marangoni in 1871, established that local imbalances in interfacial tension could result in interfacial motions. Studies since this time have shown that such spontaneous interfacial motions 3 4 occur for a wide variety of aqueous and organic phases ' , and that these motions profoundly affect mass transfer in f l u i d phase-separation processes^' often increasing rates by an order of magnitude. In the f i e l d of process metallurgy, the observance of any spontaneous interfacial motions has been hampered by experimental d i f f i c u l t i e s related to the opacity of the fluids involved and the high temperature encountered. Attempts to use predictions based on low temperature aqueous-organic systems for liquid metals,have been unsuccessful owing to the vast differences in density, interfacial tension, and adsorption effects. Despite these problems i t is evident that interfacial turbulence could be very important in liquid metal systems because most metallurgical reactions occur at interfaces and, at high temperature, tend to be controlled by transport processes. Thus, the extra degree of mixing provided by spontaneous turbulence and any resulting 7 8 increase in interfacial area could significantly accelerate reaction rates ' - 2 -The present investigation was undertaken to determine the effects of interfacial turbulence on the rates of liquid—phase mass transfer during the oxidation of molten copper and silver using low-momentum (< 24 dynes) vertical pure gas jets. 1.2. SPONTANEOUS INTERFACIAL MOTION 1.2.1. SOLUTE EFFECTS ON INTERFACIAL TENSION The works of Thompson and Marangoni established that imbalances in interfacial tension across an interface can lead to surface flow directed towards regions of highest interfacial tension. Local changes in the con-centration of a surface-active solute residing at the interface, or changes in temperature can CcXU-6 2. such imbalances in interfacial tension. Unless a large thermal gradient exists, however, i t is generally considered that temperature effects are negligible. Depending upon the exact nature of solute adsorption, a surface-active solute can either raise or lower the liquid metal interfacial tension. The relationship between interfacial tension a, and solute concentration C g, for both cases can be expressed as follows:. = " r s f ( 1 ) s' T,P S s which is derived from the Gibb's adsorption equation. Values for the excess solute concentration at the interface, r g , are positive i f i ^ r J < 0 and negative i f ( — J > 0. Most surface-active solutes i n liquid metals are s positively adsorbed; some of the experimentally observed effects on a m e t a ^ are given in Figs. 1.1 to 1.4 for Fe 9, Al"*"^, Cu - 0^ and Cu - S^. Values - 3 -FIGURE II Surface Tensions for Solutions of C, P, N, 0, S, and Se in licmid iron at 1550°C (Kozakeuitch^) . - 4 -SOLUTE W-% FIGURE 1-2 Surface tensions for Aluminum containing various solute metals at 50° to 80° above the liquidus (Korol'Kov 1 0). OXYGEN W-% FIGURE 1-3 Effect of dissolved oxygen on the surface tension of molten copper (Monma and Sutol-1) . Effect of dissolved sulphur on the surface tension of molten copper (Monma and Sutoll). - 7 -of in the presence of as much as 0.88 weight - % oxygen, reported by 12 11 Eremenko et a l , agree well with those of Monma and Suto , where a lowering 13 of by 50% in the presence of oxygen was observed. Bernard and Lupis report that dissolved oxygen lowers the surface tension of liquid silver by 20%. Generally speaking, the effect of solute metals is less than for solute non-metals. Those solutes exhibiting the highest degree of surface activity in liquid metals are oxygen, sulfur, selenium and tellurium. 1.2.2. MECHANISM FOR SPONTANEOUS MOTION The cause of spontaneous interfacial motion has been established as being interfacial tension gradients, along the interface, due to non-uniform solute concentrations at the interface. Such differences in solute con-centration are usually the result of non-uniform mass transport of solute across the interface on a micro- or macro-scale. In a liquid system having a well-stirred bulk phase, micro-scale concentration fluctuations can occur as random solute-rich eddies penetrate to the interface, which is shown schematically in Fig. 1.5. Because the bulk solute concentration, C^ is larger than the interfacial solute concentration, C^, the eddy, on penetrating to the interface, w i l l establish a concentration, and hence, interfacial tension gradient at the interface. Surface flow w i l l be directed away from the region of the eddy, and the magnitude of such a flow w i l l depend upon the rate of arrival of eddies at the interface,and the degree to which the interfacial tension is lowered by the eddies. - 8 -INTERFACE C b ^ s Cg ^ C 5 CTj > CTb 4 N. I TRANSFER of s SOLUTE RICH EDDY • PENETRATES TO THE -INTERFACE AND CAUSES SPREADING Y=0 t FIGURE 1-5 Interfacial motion generated on a micro-scale due to eddy penetration. - 9 -On a macro-scale, concentration imbalances can occur as a result of the presence of a source or sink of a surface active solute, such as a partially immersed solid. For purposes of i l l u s t r a t i o n , consider the case of a partially immersed solid, shown in Fig. 1.6. If the solid is soluble in the liquid, then concentration gradients w i l l be established within the liquid phase as solute collects near the interface. Thus, a longitudinal surface tension gradient w i l l be set up, and, for a positively adsorbed solute, flow w i l l be directed along the interface away from the solid. This mechanism has been used to explain the preferential attack of refractory 14 bricks at the surface line of steel- and glass-making furnaces . Brimacombe and Weinberg^ have shown that a low-momentum (< 6 dynes) vertical oxygen jet impinging on the surface of a liquid metal is yet another example of a macro-sized local source of solute which can i n i t i a t e spontaneous interfacial motion. A surface-active species can also serve to block spontaneous surface 16 motion . Up to now the transferring solute species has been considered as being the only surface active element present in the system. However, i f another solute is present that is surface active but non-transferring, the situation may change considerably. According to Gibb's theory^, a solute which depresses the interfacial tension must adsorb at the interface. As the amount of solute adsorbed in this GZbb-i ZayeA increases, the surface can become immobile so that i t is effectively solid in that eddies rich in the transferring solute, s t, w i l l be unable to spread as they penetrate to the interface, as in Fig. 1.7. This blocking action w i l l l i k e l y retard the transfer of s t > and may serve to explain the effect of s i l i c o n on the rate of oxygen absorption by levitated copper drops observed by Glen and Richardson"^. - 10 -SOLID SOLID DISSOLVING IN LIQUID FIGURE 1-6 Interfacial motion generated on a macro scale by the presence of a partially immersed solid. - 11 -AIR LIQUID (a) OPPOSING SURFACE PRESSURE, P, TO SPREADING (b) CLEAN SURFACE EDDY CONTAINING TRANSFERRING SOLUTE , s t c ro> c rs t 1 1 1 1 1 1 NON TRANSFERRING SOLUTE, s n, ADSORBED AT THE INTERFACE . EDDY CONTAINING BOTH s f AND s n AS CONCENTRATION OF s n BUILDS UP AT THE INTERFACE, SPREADING IS IMPEEDED. Os f *03 n FIGURE 1-7 Effect of the presence of a non-transferring surface-active solute in an interface. - 12 -1.2.3. MASS TRANSFER COEFFICIENT To measure the Influence of interfacial turbulence on the rates of mass transfer in the liquid phase, a knowledge of the flux of the trans-ferring specie(s) between adjacent f l u i d phases is necessary. In principle the molar flux can be related to the solute concentration profile by Ficks f i r s t law, but in the case of liquids this is very d i f f i c u l t , owing to con-vection effects. Thus, the normal method of evaluating the molar flux is empirical in nature. The molar flux can be related to the concentration of a specie(s) by the following equation: n = k- ACC1 - C b) (2) s s s s , where and C^ are the concentrations of the transferring specie,s,at the s s interface and in the bulk phase, respectively, and k is the liquid phase s mass transfer coefficient. This equation defines the mass transfer co-efficient. Values for the mass transfer coefficient depend upon the geometry of the system, the position on the interface, the f l u i d velocity, and the bulk fl u i d properties. Because k g is normally independent of solute con-centration, i t , rather than the molar flux, is usually derived from experimental data. The mass transfer coefficient need not always be evaluated from experimental data, but also may be estimated from one of several mathematical 19 20 21 22 models that have been developed in recent years ' ' ' These models are based on an idealized concept of the behaviour of stirred fluids and are derived from the diffusion equations. Perhaps the most r e a l i s t i c of - 13 -22 these i s the Surface Renewal Model proposed by Higbie . In this model, for a system having a well-stirred bulk phase, a solute-rich eddy penetrates to the interface and remains in contact for a time, te, during which solute transfer to or from the eddy occurs by unsteady-state diffusion. At the end of this time period, the eddy i s swept away from the interface and re-placed by a new one. By assuming that'te'is so short that the concentration gradient in the eddy does not reach i t s rear boundary (i.e. the eddy i s semi-infinite in 23 thickness), the equations for this model can be easily stated . The solution of the unsteady state diffusion equation then leads to an ex-pression for the mean value of the molar flux over the time interval te, as 1 1°te /WT i b n , n = — n • dt = .2/ S/m . (C 1 - C°) (3) s(avg) te / s W —— s s ° Jo irte from which the mass transfer coefficient i s found to be k s Values for'te'are often d i f f i c u l t to obtain, but for particular s t i r r i n g conditions such as involving rising bubbles, f a l l i n g drops or impinging vertical gas jets, values for te can be more easily estimated. 1.3. REVIEW OF PREVIOUS WORK 1.3.1. OBSERVATIONS OF SPONTANEOUS INTERFACIAL MOTION In recent years, both Schlieren photography and high-speed motion picture photography have been used to study interfacial turbulence in liquid - 14 -metal systems, both at room temperature and at high temperature. Brimacombe 24 and Richardson observed interfacial turbulence, using Schlieren photography, in their experiments with aqueous phases and mercury amalgams. In one such study,a mercury-indium amalgam was placed in a container and covered with a dilute solution of fe r r i c chloride. Concentrated fe r r i c chloride solutions were then introduced to the amalgam by means of a syringe, and the ensuing turbulence was photographed, as shown in Fig. 1.8(a) and (b). In Fig. 1.8(a) the amalgam (dark)-ferric chloride (light) interface i s shown prior to introduction of the concentrated solution. There i s no detectable inter-f a c i a l motion. Figure 1.8(b) shows the same interface immediately after introduction of the concentrated solution. This Schlieren photography shows that the interface has become very highly agitated. The transfer of indium is from the amalgam to the fe r r i c chloride solution. Surface velocities of up to 30 cm-sec ^ were observed as was a three-fold increase in the 25 aqueous-phase mass transfer coefficient. For this system, Brimacombe has shown that the observed turbulence is accompanied by a fluctuation in the electrical potential across the interface and so the effect may be electro-capillary in nature. More recently, Brimacombe and Weinberg"'"^  observed interfacial turbulence in high-temperature liquid metal systems where oxygen gas was jetted vertically onto the liquid metal surface. A bright oxide patch was observed to form directly beneath the jet and to spread rapidly over the liquid surface. Analysis of high-speed motion pictures of iron, t i n and copper melts gave the following velocities: - }'t -FIGURE l"8 Schlieren photographs of interfacial turbulence induced by a stream of 492 m MFe 3 + solution from a syringe onto a 0.1% In amalgam. The bulk aqueous nhase is 50 m MFe3+. Width of fiel d 11 mm. (a) ferric ion solution f a l l s from svringe onto interface (b) interface twitches, spreads rapidly from point of ferric ion impingement. (Brimacombe and Richardson 2^). - 16 -(i) iron (1600°C): U p e * 25 cm/ sec ( i i ) t i n (1100°C): U n = 80-150 cm/sec Sn ( i i i ) copper (1100°C): U = 50-100 cm/sec The oxide patches for iron and t i n are shown in Fig. 1.9 Momentum transfer from the jet (< 6 dynes) was eliminated as a possible source of interfacial motion by observing the effect of an argon jet, of 6 dynes momentum, on alumina particles on the liquid surfaces. Observed motions were less than 5 cm-sec They proposed the following mechanism for the process: (a) the oxygen concentration at the point of jet contact i s at the saturation value y (b) at the walls,the oxygen concentration is less} (c) this concentration imbalance leads to changes in a , resulting in surface spreading towards regions of high Q Q U ' For a metal surface to spread in the presence of a metal oxide, the value of the i n i t i a l surface tension of the metal, a , must be larger than the sum m of the metal oxide surface tension, a , and the interfacial tension between mo the metal and metal oxide, a , . This difference i s termed the Spreading m/mo Coefficient, S, and is defined as S = a - (a + a , ) (5) m mo m/mo which must be positive for spreading to occur. Thus, the systems observed by Brimacombe and Weinberg"^ a l l have positive values of S because the oxide patches are observed to spread over the melt surface during an experiment. (b) Interfacial turbulence on liquid t i n and iron during blowing with pure oxygen gas at a flow FIGURE 1*9 r a t e °f 120 cm3 min~l. (a) bright oxide patch on surface of liquid t i n vibrates and spreads rapidly over the surface. Velocity of the oxide pieces was estimated to be 80 to 150 cm sec~l. (Brimacombe and Weinbergl5). (b) spreading of oxide patch on liquid iron. The patch was observed to pulsate and then spread over the surface at a rate of 25 to 50 cm sec~l. (Brimacombe2^). - 18 -As oxidation continues, the value of a w i l l decrease, and so S w i l l decrease m during an experiment. The interfacial motions observed by these workers 27 have not been mentioned in other similar oxidation studies 1.3.2. GAS JETS IMPINGING ON LIQUIDS Because use is made of low-momentum gas jets in the present study, the literature of gas jets impinging verti c a l l y onto liquid surfaces i s reviewed briefly here. A l l previous work has dealt with high-momentum gas jets due to their use in metallurgical processes such as the BOF and TBRC. The action of such jets i s to vigorously s t i r the metal bath which, in turn, speeds the refining process. Several studies on the behaviour of these jets include the 19 20 29 basic stirring patterns ' , fl u i d dynamics , circulation and penetration 26 30 31 32 33 depths ' ' ' and mathematical model descriptions for fl u i d flow patterns and liquid-and gas-phase mass transfer c o e f f i c i e n t s ^ ' ^ . 34 The most relevant of these studies i s that of Davenport et al on the mass transfer rates across gas-liquid interfaces using jets impinging normal to the liquid surface. Liquid phase mixing was caused by the jet force, and for high jet momenta (e.g. 56,000 dynes) the bulk liquid appeared to be well stirred. Water model studies, using hollow plastic beads to show both flow patterns and flu i d velocities, indicated that the surface motion due to jet force was in a direction away from the point of.jet impingement and was found to vary with jet momentum and lance height above the bath surface. Observed -1 surface velocities ranged from 12 to 25 cm sec , and i t is of interest to note that jet velocities used were a l l significantly greater than 3000 - 19 -cm sec ^ - some even reaching as high as Mach 2. Mass transfer rates for gas jets soluble in the liquid phase were measured experimentally, and from these rates liquid-phase mass transfer coefficients were derived. These values were then compared to values calculated from an equation based on the Higbie Surface Renewal model. In this case, the eddies were assumed to contact the surface of the liquid at the point of impingement of the gas jet, and then proceed to traverse the surface as described before in section 1.2.3. Then, on solving the unsteady state diffusion equations, a relationship for the mean mass transfer coefficient, k , is found to be s k = B ( U^Ji_Ds )^ ( 6 ) ' S TC where Ur is the surface velocity, r i s the bath radius, and Ds is the diffusion coefficient for the solute species in the liquid phase. Values for the constant, B, in Eq. (6) are as follows: (i) B = .2/ — i f Ur • r is constant, 4 and ( i i ) B ~'\J~y^ i f Ur is constant. Ur The residence time for an eddy in the interface i s given by — = te (see sec. 1.2.3.). Based on experiments for CC^ jets impinging on water, they used Eq. (6) to calculate values for k . For a surface velocity of 25 cm sec \ they s calculated the following values: - 3 - 1 (i) i f Ur • r is constant, k g = 3.1 (10 ) cm sec ; -3 -1 ( i i ) i f Ur is constant, k =3.4 (10 ) cm sec s - 20 -The experimental value was k g = 1.8 (10 ">) cm sec X, and they f e l t the agreement between theoretical and experimental values was reasonable. In this instance, there is no clear preference for either value of B. In a similar study for jets impinging on water, Bradshaw and 21 Chatterjee obtained good agreement between theoretical and experimental results for k using Ur • r as a constant value. They note that k values s s increase with increasing jet momentum or decreasing jet height. 1.3.3. KINETICS OF MOLTEN COPPER OXIDATION AND DEOXIDATION During the oxidation and deoxidation of molten copper, mass transfer rates for oxygen can be controlled by transport in the gas and/or liquid phases as well as by the rate of chemical reactions at the surface. Thus, to correctly interpret mass transfer rate data, i t is necessary to establish which of these processes controls the rate of transport. Most laboratory-scale kinetic studies of copper oxidation have involved CO/CO2 mixtures flowing over levitated pure copper droplets, where the partial pressure of oxygen in the gas phase was determined by the CO/CO2 ratio used. In their study, Toop and Richardson"^ found that, at 1400°C, the observed rates of approach to equilibrium indicated that the kinetics of oxygen transfer were chemically controlled for the reaction co 2 :Z co + to] C u (7) In Fig. 1.10, the observed rates of oxygen uptake show that a constant value is reached quickly and that, at a fixed P02» temperature controls the maximum amount absorbed. 21 -30 254 20 o | 5 J 10 05 J (a) 0 o o 1581 ^ ~A" A~~ 1490 °C 1396 C 1299 °C 10 20 30 40 50 60 70 80 90 100 110 120 REACTION TIME (min) 3 5 0> 1489 °C o 1° O— 1395 °C w A A A ft A 1296 °C (b) 0 10 20 30 40 50 60 70 80 90 100 110 120 REACTION TIME (min) FIGURE HO Results for a 0.5-0.54 gm. droplet of copper leyitated in a gas stream of CQ:C02 ratio of 1:25 at 4160 ml.min - 1. Ca) oxidation; (b) deoxidation (Toop and Richardson ), - 22 -As these rate-controlling reactions occur at the interface, they f e l t that surface films could adversely affect the oxidation rates, but the exact effect was not known. In a later study on levitated copper droplets of 99.999% purity, Glen 18 and Richardson observed the same type of oxidation ratio as in Fig. 1.10 for the temperature range 1400° to 1600°C. Their results are presented in Fig. 1.11. Electron Microprobe analysis of a cooled droplet indicated that a thin surface film of s i l i c a was present. (NOTE: Spec, pure copper has a si l i c o n content < 2 ppm). To determine the effect of these films, levitated droplets were cleaned with HF solutions, dried, and relevitated for oxidation. The upper curve in Fig. 1.11 gives the results for cleaned droplets. From this study, i t is evident that a very thin film of s i l i c a on the surface of molten copper droplets can reduce the oxidation rate by from 50 to 80%. Analysis of the results indicated that, for the temperature range and gas pressures used, oxygen mass transport was gas-phase controlled. However, at temperature less than 1400°C, the equations for gas-phase control no longer appeared to apply. The exact reason for this discrepancy was not known,and they indicate that chemical kinetics were unlikely to be responsible. Nevertheless, the role of s i l i c a films in inhibiting the rate of copper oxidation has been established. 37 In a more industrial approach, Gerlach et a l studied the rates of oxidation of a molten copper bath at 1300°C using high-momentum vertical gas jets with gas mixtures of 0.72 to 21% oxygen. The Surface Renewal model was applied to calculate values for the mass flux, and, on comparison with their - 23 -TIME (min) FIGURE I'll The uptake of oxygen from mixtures of C>2 + N 2 at 1400°C, by both desiliconized 0, and untreated, A, copper drops. -^g PQ 9 = 1.11 x 10"^ atm. (Glen and Richardson ). - 24 -experimental data, they found that mass transport control resided in the gas phase. 38 Frohme et a l , in similar experiments with oxidizing gas mixtures and molten copper at 1200°C, found that the mass transport rate for oxygen was gas-phase controlled. Furthermore, their values of gas phase mass transfer coefficient agree well with the results of Gerlach. They also studied the oxidation rates of copper containing 0.8 wt % sulphur, and found that the liquid-phase mass transfer coefficient for sulphur removal was 0.09 cm sec X . Based on these studies i t appears that the oxidation rate of molten copper by dilute mixtures of oxygen gas is controlled by oxygen mass trans-port in the gas phase, and i t is unlikely that chemical reactions influence this control. For jets of pure oxygen, mass transport control of oxygen in the gas phase i s , of course, impossible, so control should reside in the liquid phase. The deoxidation of molten copper has been studied by Themelis and 39 Schmidt using a submerged vertical, jet of pure CO at flow rates from 5 to 100 l i t r e s • min X . Over the bath oxygen concentration range of 1.0 to 0.1 wt %, they found that the rate of reduction was unaffected by the changing oxygen content. For oxygen concentrations less than 0.1 %, the rate of reduction decreased on decreasing oxygen content. Thus, they concluded that the rate of deoxidation of molten copper (0.1 % < % 0 < 1 •%) by a submerged vertical jet of CO is controlled by mass transfer of CO in the gas phase. For oxygen contents of < 0.1 %, a transition to liquid-film control was observed. - 25 -In a later study on mass transfer between a submerged horizontal gas 40 jet and a liquid, Brimacombe et a l analysed the kinetic data of Themelis 39 and Schmidt in terms of mass transport theory, and showed that the de-oxidation of molten copper by a submerged jet of deoxidizing gas was trans-36 port controlled. Nanda and Geiger also studied the deoxidation of molten copper in this manner. On the basis of these studies, i t appears certain that the deoxidation of molten copper, by submerged jets of CO or H^, is mass-transport controlled, 1.3.4. KINETICS OF MOLTEN SILVER OXIDATION AND DEOXIDATION Oxidation rates for molten silver have been investigated using high-41 momentum vertical gas jets. Chatterjee et a l have found that the oxygen mass transfer rates increased on increasing jet momentum and decreasing interfacial area, and decreased with increasing lance height. The mass transfer relationships were developed in a manner similar to those for 34 Davenport et al , and the experimental values for the mean liquid-phase mass transfer coefficient range from 0.001 to 0.015 cm sec \ depending upon the above conditions, for a bath temperature of 1000°C. Though surface velocities were not measured, they f e l t that since mass transfer was related to jet momentum, a correlation between mass transfer and surface velocity ought to exist. 42 In a later study, Chatterjee and Bradshaw measured the gas-phase resistance to mass transfer for high-momentum gas jets having Pn^ = 0.1 to 0.2 atm. Values for k^ were found to vary from 1.4 to 4.8 cm sec X, - 26 -depending on flow rate. On comparison with their prior data for pure oxygen gas jets, the values for k g were found to be 20% less for air and 30% less for other gas mixtures, indicating that the presence of a gas phase resistance to mass transfer can lower the liquid-phase mass transfer rate by as much as 30%. 43 Sano and Mori studied the rates of absorption and desorption of oxygen molten silver, at 1000°C using low flow rate gas mixtures 1000 cc min . During the degassing of oxygen, surface motion was observed, and this motion appeared to decrease with decreasing oxygen concentration. The liquid-phase mass transfer coefficient obtained for pure oxygen and argon-oxygen mixtures range from 0.013 to 0.019 cm sec X . It is of interest to note that these 41 values are similar to those for Chatterjee et a l , who used high-momentum gas jets in their studies. It i s not clear whether or not Sano and Mori observed any turbulence during oxidation, but the above comparison seems to indicate that some form of large scale st i r r i n g did exist, either spon-taneous or mechanical in nature. Based on the st i r r i n g effect of jets of pure oxygen on molten silver, oxygen mass transport i s liquid-phase controlled. This changes over to gas-phase control when gas mixtures are used during oxidation. During deoxidation, oxygen mass transport is li k e l y controlled in the gas phase. 1.4. SCOPE OF THE PRESENT STUDY Although much work has been done on the kinetics of oxidation and deoxidation of molten metals, l i t t l e i s known about the relationship between spontaneous Interfacial turbulence and mass transfer rates. Also, few - 2 7 -p i c t o r i a l observations of the liquid metal surface during both oxidation and deoxidation have been made. The effects of dissolved surface active impurities on the oxidation and deoxidation rates of molten copper have only recently been studied, along with their dissolution behaviour from the solid state. Therefore^the aims of the present study were to: (a) attempt to correlate mass transfer and interfacial turbulence data for both copper and silver, through the use of high speed cine photography; (b) to study the effect of dissolved sulphur on both the oxidation and deoxidation rates for molten copper; (c) to observe photographically the dissolution behaviour of solid Cu S, Cu o 0, Se and Te. - 28 -2. EXPERIMENTAL 2.1. APPARATUS A l l experiments were performed in the apparatus shown in Fig. 1.12. The main body was fabricated from transite, with exterior dimensions of 32.5 x 32.5 x 52.5 cm. The assembled furnace was lined, to a height of 27.5 cm with porcelain wool insulation. A 17.5 cm dia. copper induction c o i l was centred in the apparatus, provision having been made for cooling water inlet and outlet. Graphite susceptors were fabricated from a 10 cm dia. graphite rod. Each finished susceptor was 5. cm thick and had a centred 5 cm hole for crucible placement. To support the graphite, a 6.25 cm thick refractory brick was shaped to f i t inside the induction c o i l . When the graphite susceptor was centred and placed so that i t extended slightly above the top level of the induction c o i l , fine powdered alumina was added to this level. A f i n a l piece of porcelain wool was placed over the alumina, with a hole l e f t for installation of the crucibles. Molten copper 3 and silver samples were held in recrystallized alumina crucibles, of 100 cm 3 and 50 cm capacity respectively, supplied by McDaniel Corp. To allow for gas jet positioning, thermocouple installation, and sample withdrawal, a 12.5 cm dia. hole was bored in the top place of the fur-nace. A circular plate, 17.5 cm dia. and 1 cm thick, was fabricated from - 29 -R - P t - 10% Rh THERMOCOUPLES SAMPLE PORT O - R I N G ADJUSTIBLE CLAMPS-GAS LANCE AUXILLARY QUARTZ /VIEW PORT HIGH-SPEED CINE CAMERA LENS ALUMINA THERMOCOUPLE_ PROTECTION TUBE ARGON SHIELD-GAS LINE WATER COOLING RING mjm FITTINGS PLATE -MOLTEN METAL RECRYSTALLIZED -ALUMINA CRUCIBLE JNDUCTION HEATING COILS REFRACTORY SUPPORT /ALUMINA / / ^ R A N S I T E > A A A A * QUARTZA * x * *• < xGRAPHITE * . t, °3I6.SS.*« FIGURE 1-12 Cross-sectional view of experimental apparatus. - 30 -316 stainless steel, and installed over the opening. A 1.6 cm dia. hole was bored in the centre of the steel plate for the lance guide. The lance guide was fabricated from 316 stainless steel rod, and had the f i n a l dimensions of 4.7 cm long, 1.4 cm dia., and 0.7 cm bore. The lance was held firmly in place by means of a Wilson seal. In a l l , five 1.6 cm dia. threaded holes, spaced 55° apart on a ci r c l e of 3.2 cm radius, were placed in the steel plate. For a l l experiments, the lance used was a 0.2 cm i.d. by 7.6 cm long mullite tube which was fitted into a 0.6 cm o.d. by 0.3 cm i.d. alumina lance 30 cm long. Fused quartz sample tubes, 0.3 cm o.d., 0.2 cm i.d. and 90 cm long, were used to withdraw from 1 to 2 gm metal samples from the crucibles. As cuprous oxide attacks quartz, that part of the sample tube in contact with the molten copper was coated with Alcoa XA-17 reactive alumina, dried, and fired at 1200°C for two hours, to provide an inert protective coating for the tubes. Temperatures were measured by two Pt - Pt - 10% Rh thermocouples held in alumina protective sheaths. The thermocouples were connected to two calibrated Honeywell chart recorders through a cold junction. Surface observations were made through a 10 x 10 x 0.5 cm quartz window, centred 37 cm from the bottom of the apparatus. During filming, additional observations were made through a 2.5 cm dia. quartz sight glass held in a removable stainless steel tube inclined 10° from the vertical in the top plate. Two copper tubes were attached to this assembly to provide for argon gas shielding. A Hycam high-speed cine camera was used to photograph the liquid metal surfaces in both black and white and colour. - 31 -Gas flow rates were metered by calibrated Gilmont flow meters. A l l gases were dried with s i l i c a gel. 2.2. MATERIALS 2.2.1. COPPER The copper used was supplied in two forms, cathode sheets and rod. The cathode sheets, supplied by NORANDA, were of 9 9 . 9 9 % purity, and the rod stock, supplied by ASARCO, was of 9 9 . 9 9 9 % purity. Spectroscopic analyses (TABLE I) give the impurity levels for each. 2.2.2. SILVER Fine silver was supplied by ENGELHARD INDUSTRIES, and was of 9 9 . 9 9 5 % purity. The spectroscopic analysis (TABLE I) gives the impurity level. 2.2.3. SELENIUM AND TELLURIUM Both the selenium and tellerium samples used were supplied by the FAIRMONT CO., and were of 9 9 . 9 9 % and 9 9 . 8 % purity respectively. 2.2.4. REAGENTS The copper sulfide (Cu2S) and copper oxide (Cu20) powders used were of reagent grade. A l l acid solutions were made from d i s t i l l e d water and reagent grade acids. The argon, oxygen, and hydrogen gases, supplied by CANADIAN LIQUID AIR, were a l l of commercial purity. Gases were used direct from cylinders but were dried by passing through s i l i c a gel. A l l graphite used was reagent grade. - 32 -TABLE I SPECTROSCOPIC ANALYSIS OF COPPER AND SILVER SAMPLES COPPER SILVER ELEMENT NORANDA ASARCO ENGELHARD Al ND ND < 10 ppm Cu MATRIX MATRIX < 5 ppm Fe 10 ppm < 0.7 ppm ND Pb ND < 1 ppm ND Mg 1 ppm ND < 10 ppm Ni ND < 1 ppm ND Se ND < 1 ppm ND Si 10 ppm < 1 ppm < 10 ppm Ag 10 ppm < 0.3 ppm MATRIX S ND < 1 ppm ND Sn ND < 1 ppm ND Ti ND ND < 10 ppm NOT DETECTED : Sb, As, Ba, Be, Bi, B, C, Ca, Cr, Co, Ga, Au, Mn, Mo, Nd, K, Na, Sr, Ta, Th, W, U, V, Zn. - 33 -2.3. PREPARATION 2.3.1. COPPER Cathode sheets were cut up and induction melted, in 6000 gm lots, in a graphite crucible. During melting, argon gas was passed over the melt surface to retard oxide formation. The copper was then cast in small cylindrical graphite crucibles, of 750 gm capacity. The average sol i d i f i e d b i l l e t measured 3.5 x 8.2 cm. A similar procedure was used for the high-purity rod stock. The sol i d i f i e d b i l l e t s were then cleaned to remove surface oxides, particularly s i l i c a , as follows: (a) immersed in 50% HNO^  for 15 min.; (b) washed in water for 5 min.; (c) washed in 5% HF for 2 hrs.; (d) washed in water for 30 min.; (e) dried with ethyl alcohol and stored in a dessicator. The samples were weighed prior to use in an experiment. 2.3.2. SILVER Fine silver, of 2000 gm weight, was melted in a box furnace and cast into cylindrical graphite crucibles of 400 gm capacity. The solidified b i l l e t measured 2.5 x 7.5 cm on average. The cleaning procedure was the same as for copper, and samples were weighed prior to use in an experiment. - 34 -2.4. PROCEDURE 2.4.1. COPPER EXPERIMENTS For each experiment a clean, weighed copper b i l l e t was placed in the alumina crucible and positioned in the susceptor. The top plate was then installed and secured. Each thermocouple was positioned, and connected to the strip-chart recorders. The lance was then lowered to within 5 cm 3 -1 of the crucible, and argon flushing at 1500 cm min was begun, 30 minutes prior to heating the copper. The induction power was then switched on and set at 5KW during the melting process. When the copper had melted, after ^ 90 min., the power was reduced to 2KW to achieve the experimental bath temperature. An i n i t i a l oxygen sample was withdrawn from the melt and logged in the record book. The lance was then lowered to ^ 0.1 cm from the liquid surface. Gas flow was changed over to oxygen at a desired flow rate and, as soon as the oxide patch appeared on the bath surface, a timer was started. Samples were withdrawn at convenient time intervals, over the course of an experiment and quenched in water. For surface velocity studies, fine alumina chips 0.15 cm long) were added to the melt, through a quartz tube, at the point of jet impingement. High speed motion pictures (300 frames • sec "*") were made of the spreading of these particles for later analysis. The average time for an oxidation experiment was 350 sec. At the end of an oxidation experiment, the oxygen was switched off, and argon at a rate of 2 l i t r e s • min ^ was passed over the melt in preparation for deoxidation with hydrogen. After purging for 10 min., an i n i t i a l sample was taken. Then, with the timer zeroed, hydrogen at a desired flow rate was passed over the melt. Filming and sampling techniques were as before, except that no alumina chips were added. Degassing experiments lasted, on average, 1000 sec. 2.4.2. SILVER EXPERIMENTS Clean, weighed, silver b i l l e t s , of ^ 400 gm mass, were placed in 3 50 cm capacity alumina crucibles, and positioned in the susceptor. Oxi-dation and deoxidation experiments were performed in the same fashion as for copper. Motion pictures (32 frames • sec ^) were made during a run at timed intervals, as before. There were no surface1 velocity studies performed. 2.4.3. EXPERIMENTS WITH SULPHUR Powdered cuprous sulphide,of 99.9% purity, was added in weighed proportions to molten copper to give starting compositions of 0.01%, 0.05%, 6.1% and 0.5% by weight sulphur. Oxidation, deoxidation and surface velocity studies were then conducted as for pure copper. Samples were analysed for both oxygen and sulphur. 2.4.4. EXPERIMENTS WITH Se, Te, Cu?Sand Cu?0 SOLIDS Weighed samples of Se, Te, C^S and Cu^O were held in baskets fabr i -cated from pure copper wire, and these baskets were suspended over molten pure copper. When partially immersed, their dissolution behaviour was - 36 -recorded by high speed photography. No samples were taken. 2.5. ANALYSIS 2.5.1. SAMPLE PREPARATION Quenched samples of both copper and silver were f i r s t f i l e d to remove any adherent alumina or s i l i c a pieces from the sample tubes. Then, the samples were cleaned with 50% HNO^  for 15 minutes, followed by a water and ethyl alcohol wash. Each 1 to 2 gm sample was trimmed to be approximately 0.06 gm for oxygen analysis and ^ 1 gm for sulphur analysis, because of limitations imposed by the upper detectable limits of the LECO analyses. 2.5.2. OXYGEN AND SULPHUR Weighed samples were analysed using LECO oxygen and sulphur analysers, having accuracies of ± 10 ppm and ^ ± 2% respectively. The small oxygen samples used were found to be reproducible within the larger i n i t i a l samples taken. 2.5.3. OXIDE PATCH SIZE AND SURFACE VELOCITY The high-speed motion pictures obtained during experiments were analysed using a Model 900-B MOTION ANALYSER from Photographic Analysis Ltd. This consisted of a 16 mm variable speed movie projector and magnifying screen, with an X-Y calibrating recorder having d i g i t a l display. - 37 -For surface velocity studies, the X-Y recorder was f i r s t calibrated for the crucible diameter at the liquid surface. Then, the X-Y coordinates for a given pellets' trajectory were recorded, from which surface velocities could be determined. Oxide patch surface areas were based on a s t a t i s t i c a l analysis of patch diameters, measured by the recorder, for a given sequence of frames in a film. The oxide patch, as observed from above, appeared uniformly circular. - 38 -3. RESULTS 3.1. COPPER OXIDATION  3.1.1. OBSERVATIONS When oxygen was blown onto the molten copper surface, a circular oxide patch was observed to form instantaneously on the liquid at the point of jet impingement, and this patch was seen to increase in area during an experiment. Figure 3.1 shows the liquid copper surface both before and after oxygen is introduced; the growth of the oxide patch with time is shown in Fig. 3.2. Surface motions were identified both through ripple formation on the clear portion of the surface (away from the patch) and tongueJ) of oxide spreading radially outward from the oxide patch, as can be seen in Fig. 3.2. The thickness of the oxide patch could not be measured, but i t should not be more than 1 to 2 mm, as this was the approximate distance of the lance tip above the liquid surface. During preliminary tests, i t was noted that the bulk temperature of the molten copper rose approximately 25°C during oxidation. Taking this into account, the temperature for each experiment was controlled so that the average temperature during oxidation was 1220°C ± 10°C, a reference 4 8 temperature for which the interfacial oxygen concentration i s known . In Fig. 3.3 a typical time-temperature curve is presented, showing the rise in temperature during oxidation. - 39 -FIGURE 3'l Interfacial turbulence during oxygen absorption by molten copper. (a) copper surface prior to oxygen introduction; (b) copper surface during oxidation. Note distortion of crucible rim reflection. Gas flow rate is 1500 cm3 min~l. - AO -FIGURE 3*2 Growth of oxide patch on molten copper during an experiment. (a) t * 40 s e c , oxide patch area = 2.02 cm2; (b) t = 183 s e c , oxide patch area - 3.21 cm^ . Note tongues of oxide spreading from edf*e of patch. Ga9 flow rate is 1500 cm^  min"*. - 41 -I400H 1200 J & 1000' Q : U J a Li 8 0 0 6 0 0 A 4 0 0 H 2 0 0 1 FIGURE 3-3 oxygen on 1233° -o-1213° Tovg= 122 0 ° - 0 - 4 ~20 30 40 50 60 To 8 0 90 100 110 120 L30~ EXPERIMENT TIME (min) Temperature-time plot for Run 41, showing temperature rise during oxidation. A l l other runs exhibited similar behaviour. - 42 -Early in the molten stage prior to each experiment, the surface of the molten copper was seen to twitch, in a random fashion, for approximately ten minutes. On attaining thermal equilibrium, the surface twitching ceased, leaving the molten copper surface clean and smooth. The cause of this twitching is not known, but i t could l i k e l y be due to the evaporation of a surface film or the desorption of a surface-active species from the bulk of the liquid. 3.1.2. OXIDATION RATES Oxidation experiments were performed on molten copper samples using 3 -1 pure oxygen jets with flow rates of 500, 1000, 1500, and 2000 cm min . Other variables considered were changes in lance height and presence of dissolved si l i c o n and sulphur. The experiments were not continued to the 49 oxygen saturation point of 2.1 wt % because as saturation was approached, the oxide patch shape became markedly irregular, making area measurements d i f f i c u l t . Also, i t was found that the lance tended to become plugged with solid copper oxide as the saturation value was approached. Typical experi-mental rates of oxygen absorption are shown in Figs. 3.4 and 3.5. In Fig. 3.5 the reproducibility of oxidation experiments at an oxygen flow rate of 3 -1 " 1500 cm min is shown. Similar behaviour was observed at other oxygen flow rates. From the slopes of the lines in Figs. 3.4 and 3.5, the oxidation rates for a l l copper experiments were calculated (moles oxygen per s e c ) , and are presented in Table II, together with the maximum possible oxygen absorption rates based on the flow rate of oxygen through the lance. - 44 -TABLE II EXPERIMENTAL RESULTS for OXYGEN FLUXES DURING OXIDATION OF COPPER Run Number Oxygen Flow Rate (cc/min) Approximate Lance Height (mm) Flux of Oxygen to surface x 10 4 moles 0 Flux of Oxygen Absorbed by Cu F 2 x 10 4 moles 0 Difference (F X - F 2) x 10 4 moles 0 sec Impurity Added (%) sec sec 22 " 500 1 7.44 5.36 2.08 -23 1000 1 14.89 12.82 2.07 -24 1500 1 22.32 19.7 2.62 -25 1500 1 22.32 20.04 2.28 -26 2000 1 29.76 25.2 4.56 -27 •'*' 2000 1 29.76 24.8 4.96 -29 1500 10 22.32 18.9 3.42 -30 1500 20 22.32 17.98 4.34 -37 1500 1 22.32 17.68 4.64 -41 1500 1 22.32 19.38 2.94 0.002% Si 43 1500 1 22.32 15.53 *6.79 +5.51 0.1% S 44 1500 1 22.32 11.37 *10.95 +1.33 0.5% S 45 1500 1 22.32 20.99 +1.33 +1.26 0.01% S 51 1500 * 1 22.32 20.46 0.05% S * not considering 0 2 lost as SO + considering 0 9 lost as S0„ - 46 -3.1.2.1. BULK-PHASE MIXING Mixing of the bulk copper phase was provided by both the inductive f i e l d and the spontaneous turbulence generated during oxidation. To determine i f the bulk copper phase was well-stirred, and hence i f the bath oxygen concentration was homogeneous, the oxygen supply was turned off after 256 sec. during an experiment, and sampling was continued to determine both the change in oxygen concentration and the time taken to reach constant composition. The results are shown in Fig. 3.6. 3.1.2.2. SILICON EXPERIMENTS The role of interfacial s i l i c o n in reducing the oxidation rate of 18 molten copper was chearly shown by Glen and Richardson . To minimize any possible effects due to sil i c o n contamination, the copper b i l l e t s used in these experiments were treated with a 10% HF solution prior to use, as described in sec. 2.3.1. The sil i c o n content of the as-cast material was 44 found to be 0.001% . To prevent contamination from the s i l i c a sampling tubes, that portion of the tube in contact with the molten copper was coated with alumina, as described in sec. 2.1. This film was found to be adherent and protective. To determine the effect of dissolved si l i c o n on the oxidation of copper in the present experiments, 0.002% si l i c o n was added to 99.999% pure copper (Table I) in Run 41. The results are given in Fig. 3.7 and Table II. - 48 -2 0 I 5 1 z o cr z U J o z o o z U J >-X o i-o H 0 - 5 H O o o % Si A 0 0 0 2 % Si — 1 — 2 5 0 3 0 0 1 | 150 2 0 0 T I M E ( s e c ) FIGURE 3'7 Comparison of the oxidation-time relationship for Run 25 (no Si added) and Run 41 (20 ppm Si added); both at 1500 cc min~l oxygen flow rate. 350 - 49 -50 100 150 2 0 0 250 3 0 0 350 T IME (sec) FIGURE 3*8 Oxidation-time relationship for copper containing dissolved sulphur at 1500 cc min~l oxygen flow rate. - 50 -3.1.2.3. SULPHUR EXPERIMENTS Four experiments were performed with varying i n i t i a l sulphur contents of 0.5%, 0.1%, 0.05%, and 0.01% to determine the effect of dissolved sulphur on the oxidation rate for molten copper. The chosen oxygen flow rate was 3 -1 1500 cm min , and both oxygen and sulphur analyses were performed. In Fig. 3.8 the results of these experiments are presented, together with an average oxidation rate for sulphur-free copper. 3.1.3. SURFACE VELOCITY STUDIES During oxidation of the copper samples, alumina particles were dropped onto the bath surface near the point of jet impingement to permit rough measurements of surface velocity. These particles were observed to spread radially outward at a very rapid rate. Analysis of motion picture films showed that the particles moved with a reasonably constant velocity across the bath surface u n t i l they approached the crucible rim, where the velocity rapidly decreased. At no time did the trajectory of the particles vary from a straight-line path radially across the surface. Those particles collecting near the rim of the crucible were seen to move only in small, tight circles at very slow rates. Results of these surface velocity studies are presented in Figs. 3.9 through 3.12, showing surface velocity as a function of radial distance from the lance. In a l l cases the lance position is at zero cm and the crucible rim i s at roughly 2.8 cm. - 51 -50< 4 0 . A 0 1 % oxygen O 0-2 " 5 0 0 cnv'min 0 a gas flow 3 0 i 2 0 4 o a> </> E o io i o O 5 0 U J > U J o 2 4 0 cc z> CO 3 0 H 2 0 H 10^. oxygen cm/min 0 2 FIGURE 3-9 » i i i i 1 i i i 1 i i i 0 0 2 0-4 0-6 0-8 10 1-2 14 16 18 2 0 2-2 2 4 RADIAL DISTANCE FROM LANCE (cm) Surface v e l o c i t y as a function of Radial distance from lance. Runs 22 and 23. V e r t i c a l l i n e s show edge of oxide patch. - 52 -5 0 4 0 H 30H 20 A u tt £ 10 E u >-§ 5 0 ' _i LU > LU § 4 0 or CO 3 0 4 2 0 4 10 T 1 * r O 0-22% oxygen A o-5 • |24 " 1500 err?/min 0 , gas flow O 0 -46 % oxygen A 0-75 " • I 0 4 " 2 0 0 0 cm /min 0 , gas flow - r T T T FIGURE 310 1 0 2 0 RADIAL DISTANCE FROM L A N C E (cm) Surface velocity as a function of Radial distance from lance. Runs 24 and 26. Vertical lines show edge of oxide patch. - 53 -FIGURE 311 Surface v e l o c i t y as a function of Radial distance from lance. Runs 27 and 29. V e r t i c a l l i n e s show edge of oxide patch. - 54 -5 0 O A 1 1 r 0- 9 4 % oxygen 1- 2 4 0 1500 c m / m i n 0, gas flow 3 0 J 20 o a> e 10 o u g so LU > UJ o g 4 0 or co O 0-3 % oxygen A 0-82 " 1500 cm/min 0 2 gas flow 3 0 A 20 J I0H T T 1-0 T T T T RADIAL DISTANCE FROM LANCE (cm) 2 0 FIGURE 3*12 Surface velocity as a function of Radial distance from lance. Runs 30 and 41. Vertical lines show edge of oxide patch. - 55 -3.1.4. OXIDE PATCH AREAS Oxide patch areas were measured from the same sequence of motion picture film used to determine surface velocities. For bath oxygen contents up to approximately 1.5 wt %, the oxide patch was uniformly circular in appearance, as seen in Fig. 3.1. Sections of the patch were observed to become detached and to dissolve as they spread across the surface. At bath oxygen contents of more than 1.5 wt %, tongues of oxide were seen to spread out from the patch across the bath surface, making patch area measurements uncertain. A l l areas were measured at fifteen time intervals during a given film sequence, and these results were s t a t i s t i c a l l y analysed. The measured areas had a standard deviation of less than 10% within each group. Results for these area measurements are presented graphically in Figs. 3.13 and 3.14. 3.1.5. SURFACE BLOCKAGE STUDIES Experiments were performed to determine the area through which the majority of oxygen mass-transfer occurs, so that the surface area term in the mass transfer equation could be correctly evaluated. A centred 2.75 cm dia. hole was bored in a 5.0 cm dia. alumina disc, and this disc was carefully placed on the molten copper surface as shown in Fig. 3.15. Oxygen was then jetted onto the open surface in the centre, and the rate of oxygen absorption 3 -1 from jets of 1000 and 1500 cm min oxygen flow was measured as before. The results are given in Table III. - 56 -7 i u 5H < ui ac < I 4 ui 9 x o 3 ' 2H \4 O 500 cm/min oxygen • 1000 A 1500 " • 2 0 0 0 " FIGURE 313 — i 1 1 1 1 1 — 0-2 0 - 4 0 - 6 0 - 8 1 0 1-2 BATH OXYGEN CONCENTRATION (wt-%) Oxide patch area as a function of copper bath oxygen concentration for four oxygen flow rates. - 57 -< Ul cr < 5H 4A u i o X 0 3 2H H -r 0 - 2 T 1 1 r 0 - 4 0 - 6 0 - 8 1 0 1-2 BATH O X Y G E N CONCENTRATION ( w t % ) FIGURE 314 Oxide patch area as a function of copper bath oxygen con-centration for four i n i t i a l sulphur concentrations at 1500 cm 3 min 1 oxygen flow rate. - 58 -FIGURE 315 Top view of apparatus for reduced surface area experiments, showing configuration of the alumina disc. - 59 -TABLE III EXPERIMENTAL RESULTS FOR OXYGEN UPTAKE RATES OVER REDUCED SURFACE AREA Run Number Oxygen Flow Rate (cm3/min) Surface Area (cm2) Oxygen Uptake Rate (moles 0/sec) K10+-4 % Difference 23 1000 25 12.82 34 1000 5.94 10.74 16.2 24 1500 25 19.7 25 1500 25 20.04 36 1500 5.94 17.87 10.0 - 60 -3.2. SILVER OXIDATION 3.2.1. OBSERVATIONS Unlike the oxidation of molten copper, vi r t u a l l y no surface motions were observed to occur during the oxidation of the molten silver samples. The only detectable surface motions were the slight twitchings and circular stirrings of less than 5 cm sec ^ velocity, as was observed for copper, just after melting had occurred. On attaining thermal equilibrium the bulk of these motions disappeared, leaving the surface clean and mirror smooth. Figure 3.16 shows the liquid silver surface both before and after oxidation had begun. A l l experiments were performed at a controlled temperature of 1100°C ± 10°, and a typical heating curve is shown in Fig. 3.17. As molten silver does not have a stable oxide at this temperature, an oxide patch similar to that for molten copper, was not expected to form during the oxidation stage. Discrete surface patches, however, were observed to form on some of the molten silver samples. These patches were li k e l y composed of oxidized impurities that are insoluble in the molten silver, but, as the level of 4 4 contamination of the silver was small , the patches should not have con-tained much material. In any event, these patches were easily cleared to the edge of the liquid surface using a probe, and, because the general surface motions were slight, patches tended to remain at the edge of the liquid surface. - 61 -(0) (b) FIGURE 3*16 Surface of molten silver during oxidation. _^ (a) Argas only; (b) oxygen gas at 1500 cm3 min flow. Note absence of interfacial motions as seen in crucible rim reflection. - 62 -10 2 0 T 1 1 1 1 1 r 4 0 6 0 8 0 100 EXPERIMENT TIME (min) FIGURE 317 Time-temperature plot for Silver Run 4 showing experimental temperature. A l l other silver runs give similar results. - 63 -- 6 4 -TABLE IV ABSORPTION AND MASS TRANSFER DATA FOR LIQUID Ag OXIDATION Run Number c b 0 (moles) ri Delivered moles 0/sec x 10 4 n Absorbed moles 0/sec x 10 4 x 10 cm/sec 4 0.0075 22.32 0.388 0.0167 22.32 0.388 4.02 0.0162 22.32 0.388 2.27 0.0227 22.32 0.388 3.53 0.0206 22.32 0.388 2.55 5 0 29.76 0.408 — 0.006 29.76 0.408 7.43 0.0098 29.76 0.408 3.78 0.0117 29.76 0.408 3.43 0.0118 29.76 0.408 2.75 0.0139 29.76 0.408 2.77 7 0.001 22.32 0.423 _ 0.003 22.32 0.423 2.55 0.007 22.32 0.423 2.64 0.009 22.32 0.423 2.7 - 22.32 0.423 -0.015 22.32 0.423 3.36 - 65 -3.2.2. OXIDATION RATES Oxidation experiments were performed on molten silver baths using 3 -1 pure oxygen jets having flow rates of 1500 and 2000 cm min . Experiments were continued for approximately 350 sec. in each case. Owing to d i f f i -culties encountered during the analysis of the samples, only four of the eight experiments attempted proved to be reasonably successful. Oxygen uptake rates for the silver experiments are shown in Fig. 3.18. Oxidation rates were determined from the slopes of the lines in Fig. 3.18 and are presented in Table IV, together with the total flux of oxygen delivered to the bath surface. 3.3. COPPER DEOXIDATION 3.3.1. OBSERVATIONS In preparation for deoxidation experiments, each sample was f i r s t 3 -1 oxidized using an oxygen jet of 1500 cm min flow for a period of approxi-mately 350 seconds. Then, the apparatus was purged with argon gas at 3 - 1 1500 cm min for approximately 800 sec. to remove excess oxygen from the furnace atmosphere before introducing the hydrogen gas. During the inert gas purge, only minor stirring motions were observed which were l i k e l y due to the sti r r i n g effect of the inductive f i e l d . When hydrogen was jetted onto the liquid copper surface, instantaneous surface flow directed to the point of jet impingement was observed. Pieces of debris were seen to collect beneath the lance and to spin around rapidly. For the most part, these surface motions appeared to be highly disorganized, - 6 6 -Interfacial turbulence during the deoxidation of molten copper with hydrogen. (a) copper surface before deoxidation started; (b) copper surface after deoxidation started. Note distortion of crucible rim reflection. - 67 -as opposed t o t h e r a d i a l f l o w ou twa rd seen i n c o p p e r o x i d a t i o n . These same random m o t i o n s were o b s e r v e d by S m a l l and coworke r s " ' " ' i n t h e i r s t u d y o f h yd rogen a b s o r p t i o n by c oppe r b a t h s c o n t a i n i n g d i s s o l v e d o x y g e n . Becau se o f t h e randomness o f t h e s e s u r f a c e m o t i o n s , s u r f a c e v e l o c i t y measurements were n o t a t t e m p t e d . However , t h e v e l o c i t i e s d i d appea r t o be i n t h e r ange o f 10 t o 15 cm sec ^ . The p i e c e s o f d e b r i s c o l l e c t i n g b e n e a t h t h e l a n c e were o f i r r e g u l a r shape and d i d n o t appea r t o change i n s i z e d u r i n g an e x p e r i m e n t . These p i e c e s a r e l i k e l y composed o f n o n - r e d u c i b l e o x i d e s such as s i l i c a and a l u m i n a . When t h e s e p i e c e s were pushed f r o m b e n e a t h t h e l a n c e by a p r o b e , t h e y were o b s e r v e d t o r e c o l l e c t q u i c k l y a t t h e p o i n t o f j e t imp i n gemen t . As t h i s patch t e n d e d t o o b s c u r e most o f t h e s u r f a c e m o t i o n s n e a r t h e l a n c e , t h e o n l y e v i d e n c e o f s u r f a c e s t i r r i n g was seen as &[f)AJiti> o f l i g h t e r c o l o u r e d l i q u i d i n t h e c l e a r p o r t i o n o f t h e l i q u i d s u r f a c e . The l i q u i d c oppe r s u r f a c e b o t h b e f o r e and a f t e r d e o x i d a t i o n had begun i s shown i n F i g . 3 . 1 9 ; d i s t o r t i o n o f t he c r u c i b l e r i m r e f l e c t i o n i s e v i d e n c e o f s u r f a c e m o t i o n . F o r a l l d e o x i d a t i o n e x p e r i m e n t s , t h e b a t h t e m p e r a t u r e was m a i n t a i n e d a t 1220°C ± 1 0 ° , as i n s e c . 3 . 1 . 1 . , and no t e m p e r a t u r e change d u r i n g d e o x i d a t i o n was o b s e r v e d . 3 . 3 . 2 . DEOXIDATION RATES U s i n g j e t s o f p u r e h y d r o g e n a t f l o w r a t e s o f 1000, 1500 , and 3000 3 - 1 cm m i n , d e o x i d a t i o n e x p e r i m e n t s were p e r f o r m e d on m o l t e n coppe r s amp le s c o n t a i n i n g v a r y i n g amounts o f d i s s o l v e d oxygen . The a v e r a g e d u r a t i o n o f an e x p e r i m e n t was 1200 s e c , w i t h one l a s t i n g f o r 5400 s e c . t o e s s e n t i a l l y z e r o 2.0 1-6 ^ C r O 0 — O O O Q O Q -1-2 H - 0-8 35 0 - 4 4 uj 2 0 o z o o i 1-6 > X o 1-24 0 8 i 0-4 J - 68 -T 1 1 i r 1500 c m / m i n Ar 1000 cm/min H , 2 0 0 4 0 0 6 0 0 8 0 0 1000 1200 TIME (sec) FIGURE 3*20 Deoxidation-time relationship for copper Runs 31 and 32. ) - 69 -• I 1 1 i | 2 0 0 4 0 0 6 0 0 8 0 0 100 0 1200 TIME (sec) FIGURE 3-21 Deoxidation-time relationship for copper experiments at two hydrogen flow rates. - 70 -2 0 -o 164 S I-2H z o UJ o z 0 -8 ' o o z 1500 crrf/min H 0 0 - 4 1 liquid- phase control gas-phase control 1000 2 0 0 0 3 0 0 0 TIME 4 0 0 0 (sec) 6000 FIGURE 3-22 Deoxidation-time relationship for copper Run 50, showing regions of gas-phase and liquid-phase control. - 71 -TABLE V EXPERIMENTAL RESULTS FOR OXYGEN FLUXES DURING DEOXIDATION OF COPPER Run Number H 2 Flow Rate (cm3/min) Approximate Lance Height (cm) Flux of H 2 to Surface (moles/sec) XIO*-4 Flux of 0 from (moles/sec) X10+-4 Run Duration (sec) 31 0 3 0 0 1200 Ar Only 32 1000 1 7.44 1.80 1200 33 1500 1 11.16 2.20 1000 37 3000 1 22.32 2.09 1200 38 1500 1 11.16 1.99 2010 39 1500 1 11.16 2.14 800 40 1500 1 11.16 2.68 1000 41 1500 1 11.16 1.42 1000 * 43 1500 1 11.16 4.3 1000 * 45 1500 1 11.16 1.63 800 50 1500 1 11.16 1.7 3000 * OXYGEN ALSO LEAVING AS S0„ BUT IN AN UNKNOWN AMOUNT BECAUSE SULPHUR ANALYSIS NOT POSSIBLE. - 72 -oxygen content. In each case the lance height was maintained at approxi-mately 10 mm above the liquid surface. The observed deoxidation rates are presented graphically in Figs. 3.20 to 3.22 as a function of experimental time, and from the slopes of these lines, the deoxidation rates (moles 0 per sec.) as presented in Table V were calculated. 3.4. SILVER DEOXIDATION 3.4.1. OBSERVATIONS Deoxidation of molten silver samples was performed in a manner similar to that described for molten copper in sec. 3.3.1. During the inert gas purge only slight s t i r r i n g motions were observed and were presumably due to the stirring effect of the inductive f i e l d . As described in sec. 3.2.1., some surface patches were observed during oxidation, but these were quickly removed 40 sec.) during deoxidation. When hydrogen gas was jetted onto the molten silver bath, surface flow directed from the edge of the crucible towards the point of jet impingement was observed to occur. This flow was punctuated by rapid, randomly directed twitchings of the whole surface, making velocity measurements d i f f i c u l t , much the same as was observed for copper. Rough estimates indicate that average velocities were in the range of 10 to 15 cm sec 1 . After about 90 s e c , a film of what appeared to be very fine bubbles formed on the surface and disappeared from the centre radially outwards. On removal of this film, the surface appeared clean and bright and motions ceased. It i s not known of what this film was composed. Throughout the deoxidation some gas bubbling in the liquid was observed to - 73 -FIGURE 3" 23 Interfacial turbulence during the deoxidation of molten silver with hydrogen. (a) molten silver surface before deoxidation started; (b) molten silver surface after deoxidation had begun. Note distortions in the crucible rim reflection and the lance tip reflection. - 74 -i 1 1 1 • r TIME (sec) FIGURE 3-24 Deoxidation-time relationship for liquid silver experiments at two hydrogen flow rates. - 75 -occur, and that by single, random bubbles, with only slight surface disruptions. In Fig. 3.23, the molten silver surface both before and after deoxidation has begun is shown. 3.4.2. DEOXIDATION RATES Using jets of pure hydrogen at flow rates of 1500 and 3000 cm sec 1 the deoxidation rates for molten silver were measured. The lance height was maintained at 10 mm throughout each experiment, and the bath temperature was controlled at 1100°C, with no observable fluctuations during deoxidation. Experiments were continued for up to 500 sec. and none went to complete deoxidation. Owing to d i f f i c u l t i e s encountered in analysis, only four of the eight experiments attempted provided satisfactory results. The observed deoxidation rates are presented graphically in Fig. 3.24, and the approxi-mate rates (moles 0 per sec.) were calculated from the slopes of the lines in Fig. 3.24 and are presented in Table IX. 3.5. SOLID DISSOLUTION EXPERIMENTS  3.5.1. OBSERVATIONS Weighed samples of commercially pure Cu^O, Cu^S, selenium, and tellurium were suspended in pure copper bcu>\izti>, and lowered to the molten copper sur-face by means of a copper wire. The ensuing events were observed and photo-graphed through the viewing port using a high-speed camera at 200 frames per second. A piece of Cu9S suspended above the bath surface can be seen in - 76 FIGURE 3*25 Interfacial turbulence during the dissolution of solid Cu2S (a) before contacting surface; (b) * 2 sec. after contacting surface. Note distortions in crucible rim reflection. (a) - 78 -Fig. 3.25(a). This photo also shows the relative quiescent nature of the molten copper just prior to introducing the Cu^S sample, as witnessed by the reflection of the crucible rim. When contact with the copper was initiated, the surface of the bath was observed to become highly agitated as shown i n Fig. 3.25(b). The reflection of the crucible rim was distorted by the presence of ripples and waves formed as a result of surface motion, directed radially away from the solid C^S, induced by the dissolution of the C^S solid. White patches visible on the bath surface are areas of molten Cu^S (mp. 1130°C), which are spreading away from the solid and dissolving in the copper. Complete dissolution of a 4 gm piece of Cu^S required only about 8 seconds. No surface velocities were measured due to experimental d i f f i c u l t i e s . Based upon the previous values obtained during copper oxidation (sec. 3.1.3.), however, the velocities encountered during Cu^S dissolution appear to be in the range of 20 to 40 cm sec X. Similar experiments were conducted for CU2O, selenium and tellurium solid additions, since these materials are 11 known to lower the surface tension of molten copper , as discussed in sec. 1.2.1. The p i c t o r i a l results of these experiments are shown in Fig. 3. A l l solids melted and spread rapidly across the surface, but, in the case of selenium and tellurium,the surface was quickly covered with a film of the molten metal, obscuring further surface motions. Fume formation was also a problem, and so these latter experiments were less satisfactory. - 79 -4. DISCUSSION 4.1. COPPER OXIDATION 4.1.1. OXIDATION RATES As seen in Figs. 3.4 and 3.5, the oxidation rates for molten copper are constant at a l l the flow rates studied. The reproducibility of oxidation 3 -1 rates at the oxygen flow rate of 1500 cm min is shown in Fig. 3.5, and these results indicate good agreement among the various experiments. Changes in the flow rate of oxygen delivered to the bath surface are seen to be matched by identical changes in the oxygen absorption rate, as is evident in Table II. Also, the rates of oxygen delivery to the bath surface and that absorbed by the copper are seen to be similar. Thus, the oxygen absorption rate for molten copper, using a low-momentum oxygen jet, varies directly with the input oxygen flow rate for the experimental conditions described. This is an effect indicative of starvation mass transfer^ In Runs 29 and 30, the lance height above the surface was increased by a factor of 10 and 20 respectively to determine the effect of increased lance height on oxygen absorption rates. From a comparison of the above rates to that of Run 25 in Table II, the observed differences are only 5 and 10 respectively, clearly not a major effect. These small differences in absorption rate on increasing lance height are most l i k e l y due to oxygen losses to the freeboard. Thus, the effect of changes in lance height, over the range studied, is small. - 80 -4.1.1.1. BULK-PHASE MIXING From Fig. 3.6, i t can be seen that only 25 sec. are needed, from the time of oxygen shut off, to reach constant bath composition. The slight increase in oxygen concentration observed, amounting to only 7.3% of the value prior to shut-off, is probably due to continued oxygen absorption from the freeboard as well as to departures from perfect mixing in the bulk liquid phase. Since only 0.06% oxygen is absorbed after oxygen shut-off, i t i s clear there are no gross errors in assuming the bulk copper phase to be well-stirred. 4.1.1.2. SILICON EXPERIMENTS 18 The experiments of Glen and Richardson clearly demonstrated the severe retardation in oxygen absorption rates due to blockage of adsorption-reaction sites by interfacial films of s i l i c a . This is an effect that would apply especially to.their experiments, where small droplets of molten copper were levitated in a slowly flowing gas stream of low oxygen partial pressure Being surface active, a few p.p.m. of dissolved s i l i c o n would naturally concentrate in mono-layer proportions over the whole surface, and present a layer of s i l i c o n to the oxidizing gas. Thus a coherent network of s i l i c a is easily formed which prevents further oxidation, as shown in Fig. 1.11. Because of the dynamic nature of the present experiments, where oxidation occurs at a point source resulting in rapid surface spreading away from the point of jet impingement and in a high degree of surface agitation, i t seems unlikely that such interfacial s i l i c a films could exist and retard the - 81 -oxidation rate. The results given in Fig. 3.7 and Table II show that added si l i c o n does not in fact retard the oxidation rate of molten copper in the present experiments. Further to this, a recent paper by Forster and 46 Richardson mentions that interfacial s i l i c a films can be removed from the surface of molten copper by molten C^O, which forms a soluble product with the s i l i c a . Since molten C^O is present in our experiments, i t is unlikely that any coherent s i l i c a films could exist at the interface and adversely affect the copper oxidation rates. The above results indicate that small amounts of dissolved s i l i c o n have l i t t l e effect on the oxidation of molten copper when large-scale surface spreading occurs. Thus, the procedures taken to ensure a low si l i c o n content in the melt may have been unnecessary. 4.1.1.3. SULPHUR EXPERIMENT S Dissolved sulphur is another surface-active solute which has much the same effect on the interfacial tension of molten copper as does oxygen*^, These experiments were undertaken to determine i f the oxidation rate of molten copper could be altered by the presence of dissolved sulphur in a manner similar to the effect of dissolved oxygen on nitrogen absorption 48 rates in molten iron . From Fig. 3.8 and Table VI i t can be seen that as the i n i t i a l sulphur content increases, the corresponding copper oxidation rate decreases. It i s worthy of note that the oxidation rates show the same constant behaviour as those for pure copper. Moreover, the rates of sulphur loss due to oxidation are also linear. By comparison with the oxidation results for sulphur-free copper in Table II, one can see that a large difference between delivered oxygen flux and absorbed oxygen flux exists for TABLE VI EXPERIMENTAL TESULTS FOR OXIDATION OF SULPHUR-CONTAINING COPPER Run Number % S Added as Cu2S Delivered oxygen x 10^ moles 0 Absorbed oxygen x 10+4 moles 0 Difference [Deliv.-Abs.] x 10+4 moles 0 Oxygen Lost as SO2 x 10+4 moles 0 Corrected Absorbed oxygen rate x 10+4 moles 0 sec sec sec sec sec 43 0.1 22.32 15.58 6.79 1.27 16.85 *44 0.5 22.32 11.37 10.95 9.62 21.99 45 0.01 22.32 20.99 1.33 0.0746 21.065 51 0.05 22.32 20.46 1.86 -* SO^  bubbling was observed during this experiment and may have affected the absorption rates indicated in this table. - 83 -i n i t i a l sulphur contents of 0.5% and 0.1%, with only minor differences occurring for the lower sulphur contents. However, when the amount of oxygen lost as S0 2 gas is taken into account, as in Table VI, the dis-crepancies between delivered and absorbed oxygen fluxes are minimized for a l l but the 0.1% sulphur case. In this instance, the cause of the dis-crepancy is most l i k e l y due to errors in the sulphur analysis. Sulphur dioxide gas bubbling was observed at only the largest i n i t i a l sulphur content of 0.5%. Overall, the effect of small amounts of dissolved sulphur on the oxidation rate of molten copper is negligible. Discrepancies between the delivered and absorbed oxygen fluxes, when dissolved sulphur is present, are due to oxygen loss as S0 2 gas from the interfacial regions of the copper bath. 4.1.2. SURFACE VELOCITY STUDIES On examination of Figs. 3.9 through 3.12 i t can be seen that velocity profiles for the particles can be divided into three regions: (a) Oxide patch region - particle velocities are low in this region lik e l y because the particles are accelerating and may be hampered by oxide viscosity effects. Near the edge of the oxide patch, the particles accelerate to the observed velocity maxima. The velocity of the liquid copper beneath the oxide is unknown, but is l i k e l y the same as for the clear surface region. (b) Clear surface region - particles move at constant velocity through this region up to a radial position of 1.4 to 1.8 cm, where deceleration - 84 -FIGURE 41 Flow pattern in liquid caused by an impinging gas iet. Note stagnant areas, labelled S. (Davenport et al^4). This same type of flow pattern should exist for the interfacial turbulence caused by low momentum oxygen jet impinging on liquid Cu, Sn, and Fe. - 85 -begins. This region experiences some rippling, which may affect the observed velocities. (c) Crucible rim region - here particles experience rapid deceleration to near zero velocity. Minor random sti r r i n g motions of less than 5 cm sec-''" were observed. Apparently this region is stagnant, and i t extends outwards from the crucible rim for approximately 0.8 cm. The existence of this stagnant region is l i k e l y a manifestation of the stirring pattern in the liquid, which is probably the same as that shown schematically in Fig. 4.1 (where stagnant areas are labelled S). Higher values of a in the stagnant region may also contribute to the deceleration of the particles. Recorded velocity maxima range from 20 to about 35 cm sec ^ for a l l the experiments. For the low momentum jets employed, observed surface velocities do not appear to vary with either oxygen flow rate or lance height. Though particle motions were too erratic for any degree of precision to be obtained, i t is f e l t that the above velocity range is representative for a l l experiments. Velocity maxima were plotted against bath oxygen concentration and are shown in Fig. 4.2. From this i t appears that the surface velocity decreases slightly on increasing bath oxygen content, as is also seen in Figs. 3.9 to 3.12. . This decrease in velocity i s l i k e l y due to decreases in the size of the clear surface region for unimpeded particle acceleration, caused by increases in oxide patch area on increasing bath oxygen content (see sec. 4.1.3.). Also, decreases in the interfacial tension driving force for surface spreading (sec. 1.2.2.) on increasing bath oxygen content could account for the observed decrease in velocity maximum on increasing bath - 86 -o w o 330H UJ > UI o CO 2 ^ 10 5 - 1 r " 1 1 " T ~ " 1 a • • • * • • O O • • • -• *o • • • A • • • • O 5 0 0 cm/min 0, • A 1000 " • 1500 " • 2 0 0 0 " -I • » 1 1 \ 0 - 2 0 - 4 0-6 0-8 I 0 1-2 BATH OXYGEN CONCENTRATION (wt-%) FIGURE 4*2 Maximum surface velocity as a function of copper bath oxygen concentration. - 87 -oxygen content. The mean velocity maximum was found to be 26.1 cm/sec, with a standard deviation of 21.1%. The straight line in Fig. 4.2. is obtained from a linear regression analysis of the data. Brimacombe and Weinberg"1"' observed spreading velocities for oxide particles on molten copper ranging from 50 to 100 cm sec X . These values are 50% larger than velocities observed in the present experiments, but as the earlier measurements were more crude in nature, having been made without the aid of the equipment described in sec. 2.5.3, this discrepancy may not be surprising. 4.1.3. OXIDE PATCH AREAS Oxide patch areas were measured so as to characterize the area involved in oxygen transport to the bulk copper phase. As seen in Figs. 3.13 and 3.14, values of oxide patch area generally increase on increasing oxygen flow rate. With the exception of experiments at an oxygen flow rate of 3 -1 2000 cm sec , where problems in area measurement arose, oxide patch areas also appear to increase with increasing bath oxygen content in a linear fashion. Figure 3.14 shows the effect of dissolved sulphur on the oxide patch areas. The observed areas are a l l less than those for sulphur-free copper at the same oxygen flow rate. At an i n i t i a l sulphur content of 0.5%, the bath surface was disrupted by S0^ gas bubbling, and so oxide patch areas here cannot be correlated to the sulphur-free cases. In these experiments with sulphur, a reduction in oxide patch area due to the presence of dissolved sulphur is not a surprising development owing to S0_ elimination - 88 -from the bath. Under these conditions, the surface regions of the copper are unable to saturate in oxygen, especially at the edge of the oxide patch, allowing for a lesser degree of surface coverage. 4.1.4. SURFACE BLOCKAGE STUDIES From the results presented in Table III, i t is clear that only slight differences in oxygen absorption rate occur when the bath surface area is reduced by about 76%. As a result, oxygen transport to the bulk copper phase can be considered to occur predominantly in the area described by the oxide patch. This area w i l l be used to calculate liquid-phase mass : transport coefficients for oxygen. 4.1.5. MASS TRANSFER COEFFICIENTS Using oxygen uptake rates and oxide patch areas previously determined, mass transfer coefficients for oxygen in the molten copper were calculated from Eq. 2, as n = k ACC1 - C b) (2) o o o o The rate of oxygen absorption by copper was shown in sec. 1.3.3. to be con-trolled by liquid-phase oxygen mass transfer. The interfacial oxygen concentration, C^, was taken from the phase 49 o diagram of Johnson at 1220 C, and was found to be 2.1 wt %. Values of k Q from Eq. 2 are presented graphically in Figs. 4.3 to 4.5 as a function of bath oxygen content; a l l exhibit l i t t l e change in k on increasing bath - 89 -0 2 i 0 1 4 u Q> in \ E Q 0-2H 0 1 4 A O 5 0 0 cm/min 0 a A 1000 " O O - a 2000 crriVmin 0, O O -O-O o o —r~ 0-2 1 1 r — 0-8 10 1-2 BATH OXYGEN CONCENTRATION (wt-%) 0-4 0-6 FIGURE 4*3 Liquid-phase oxygen mass transfer coefficient as a function of copper hath oxygen concentration. Horizontal line i s mean k Q for pure copper experiments using oxygen gas at 1500 cm3/min. - 90 -0 2 H o CO E u T 0-6 0-8 1 0 1-2 BATH OXYGEN CONCENTRATION ( w t % ) FIGURE 4*4 Liquid-phase mass transfer coefficient (k D) as a function of bath oxygen concentration. The mean value is k Q = 0.104 cm/sec, and the straight line i s from a linear regression analysis of the data. i - 91 0-2 H 1500 cm/min 0 2 0 0 0 2 % Si 0 - H [-XX o o o (/> E o 0 2 i 1500 cnf/min 0^ O 0-5 % S • 0 1 A 0 0 5 " • 0 0 1 " 0 1 I Q( • A A A -Q-i 0-2 —I 0-4 BATH 0-6 0-8 1 0 OXYGEN CONCENTRATION (wt %) - T — 1-2 FIGURE 4-5 Liquid-phase oxygen mass transfer coefficient as a function of copper bath oxygen concentration showing the effect of dissolved s i l i c o n and sulphur. Horizontal solid line is mean k Q for pure copper experiments using oxygen gas at 1500 cm^/min; horizontal dashed line i s from a linear regression analysis of the data for sulphur-bearing copper. - 92 -oxygen content. Since the surface velocity decreases by only about 18% during an experiment (sec. 4.1.2.), i t is reasonable to assume that the degree of sti r r i n g of the copper remains constant during an experiment, and so the constancy of k Q values is to be expected. In Fig. 4.4, a l l the values for experiments at an oxygen flow 3 -1 rate of 1500 cm min using pure copper are presented and show good reproducibility. The straight line i s from a linear regression analysis of the data, and the mean k value was found to be 0.104 cm sec 1 with a o standard deviation of 11.6%. For comparison purposes, this k Q value was included in Fig. 4.3 and 4.5 as a horizontal lin e . With the exception of 3 -1 the experiment at 1000 cm min oxygen flow in Fig. 4.3, k Q values in pure liquid copper do not appear to vary with incident oxygen flow rate. As the surface velocity measurements also show no variation with oxygen flow rate, this fact is not surprising. The scatter shown for the experiments at 3 -1 2000 cm min oxygen flow i s a result of d i f f i c u l t i e s encountered in measuring oxide patch areas at this high flow rate. In Fig. 4.5, the k Q value for pure copper was found to compare favourably with those obtained from experiments involving dissolved sulphur and s i l i c o n , as k (S.) = 0.099 cm sec" 1 o x and k (S) = 0.112 cm sec 1 o Since the observed deviation is only about 7% from the k Q for pure copper, i t is apparent that, in a system displaying large-scale surface spreading, low levels of surface-active solute contamination have no discernable effect on oxygen mass transfer in liquid copper. - 93 -The relevant literature on copper oxidation, reviewed in sec. 1.3., indicated that the published works differ vastly from the present study in terms of jet flow rate, gas composition, and liquid metal bath size. On the basis of this, i t is f e l t that a meaningful comparison of the results with those from the literature i s not possible at this time. 4.2. SILVER OXIDATION 4.2.1. OXIDATION RATES A small number of silver experiments were performed for comparison to the copper experiments, and these proved to be less satisfactory owing to the d i f f i c u l t i e s encountered in the analysis for oxygen. From Fig. 3.19, the oxidation of silver i s seen to be approximately linear with time for each of the experiments. The oxidation rates (moles 0 per sec), calculated from the slopes of these lines, are presented in Table IV. Here i t is seen that the oxidation rates for molten silver are a l l virt u a l l y the same despite the use of two different oxygen flow rates. On comparison with the oxygen flux to the bath surface, one can see that the oxidation rates are about a factor of 50 times less than the incident oxygen flux. This indicates that a surplus of oxygen is present and that starvation mass transfer does not apply to molten silver oxidation. The rate of oxygen absorption i s controlled by transport in the liquid phase, as was the situation for copper. As mentioned in sec. 3.2.1., some surface patches did appear on the molten silver during oxidation, but as these patches were easily swept away by a probe,it i s f e l t that they did not hinder silver oxidation to any - 94 -measurable degree. Because the silver samples were pre-cleaned in the same manner as the copper samples (sec. 2.3.1.) and, since the i n i t i a l s i l i c o n content was low (Table I), i t i s f e l t that interfacial s i l i c a films are unlikely to have occurred on the molten silver used here. 4.2.2. SURFACE MOTIONS As mentioned previously, the oxidation of molten silver is not accom-panied by any interfacial tension-induced surface flow, apart from the slight inductive f i e l d motions of less than 5 cm sec X. This i s , at f i r s t , sur-52 13 prising since Buttner et al and Bernard and Lupis have shown that adsorbed oxygen does lower the interfacial tension of molten silver by as much as -1 13 300 dynes cm (33%) . Thus, i t would appear that a driving force for surface spreading i s present even though not of the magnitude observed for molten copper (Aa - 50%) l i. The explanation for the lack of surface spread-ing during the oxidation of molten silver may l i e in the nature of oxygen g absorption on molten silver. Richardson has said that at elevated temperatures (> 900°C), adsorption kinetics are very rapid, and that the equilibrium solute excess at the interface, r , occurs virt u a l l y instan-13 taneously. For the silver-oxygen system, Bernard and Lupis have shown that the c r i t i c a l bulk silver-phase oxygen concentration for saturation of the interface with chemisorbed oxygen is only 0.004 wt %, and so can truly be considered an instantaneous process in the present case. Since silver oxide is not known to exist at these temperatures or pressures, i t i s reasonable to assume that the chemisorbed oxygen forms a monolayer covering the silver surface, and that the surface tension of the molten silver i s - 95 -lowered uniformly across the surface. Without a local surface tension imbalance occurring, as was the case in the region of the oxide patch on molten copper, surface tension-induced flow w i l l not occur. Even though pure silver eddies can (and l i k e l y do) penetrate to the interface, any resulting surface flow w i l l be instantaneously damped out because there i s a surplus of oxygen available for adsorption at any time. For molten copper, the oxygen also chemisorbs on the entire surface, but at the same time forms liquid copper oxide (Chi^O) beneath the lance in a small confined patch. This patch does not cover the entire liquid surface because of constraints imposed by bath oxygen content, interfacial tension differences between the Cu^O^ and the Cu^ (which are also controlled by the bath oxygen content), and oxygen jet flow rate, as discussed before. Thus, a point AOUAce. of low surface tension exists, and surface flow radially outwards from this point occurs as the spreads and dissolves. The fact that the silver oxidation rates are roughly a factor of 50 times less than those observed for copper is almost certainly a direct result of the lack of interfacial tension-driven flow during molten silver oxidation. 4.2.3. MASS TRANSFER COEFFICIENTS Because starvation mass transfer does not obtain during the oxidation of molten silver, the oxygen mass transfer coefficient in silver was calculated from Eq. (8), as b (8) - 96 -TIME (sec) FIGURE 4-6 Mass transfer from a pure oxygen jet to liquid silver. The straight line is from a linear regression analysis of the data. - 97 -The value for the interfacial oxygen content, C 1, was taken from the o 62 o work of Mizikar et a l at 1100 C, and was found to be 0.28 wt %, which 52 63 65 agrees with the results from other sources ' ~ . Unlike the situation for molten copper, the area for oxygen mass transfer in the silver experi-ments was considered to be the whole surface. The results of these calculations are presented in Table IV, and in Fig. 4.6 as a plot of In X vs time. As can be seen, the points in Fig. 4.6 for the two flow rates show good agreement, and give an average value for the oxygen mass transfer -3 -1 coefficient of 2.88 (10 ) cm sec . Because of the scatter of the data in Fig. 3.19 and thus also in Fig. 4.6, i t was f e l t necessary to place error limits on the calculated k Q values as shown by the dashed line in Fig. 4.6. -3 -1 The error in k is then ± 0.41 (10 ) cm sec . It is f e l t that the o d i f f i c u l t i e s encountered in the analysis of the samples were responsible in part for the scatter observed in Figs. 3.19 and 4.6. Since k Q for molten silver does not appear to vary greatly with gas flow rate in the present experiments, i t is certain that control of oxygen transport resides in the liquid phase. A comparison of the experimental oxygen k^ values in silver with those available from the literature i s given in Table VII. As was the case for copper, a direct comparison with the present experiments is not possible owing to the use of very high-momentum oxygen jets by the majority of other 41 researchers. Chatterjee et a l used oxygen jets of from 3000 to 56,000 dynes force to achieve k Q values in the range of 0.001 to 0.015 cm sec X. It i s interesting to note, however, that the mean k Q obtained here f a l l s in the lower range of the above study. . At the lowest jet force, momentum - 98 -TABLE VII COMPARISON OF EXPERIMENTAL k Q VALUES FOR MOLTEN SILVER WITH THOSE FROM THE LITERATURE Temperature (°C) PO 2 (atm) Jet force (dynes) Observed mean k Q (cm/sec) Reference 1100 1 18 and 24 .00288 Present Work 1000 1 - .04 to .09 50 1000 1 20,000 .01 to .02 34 1000 1 3,000 to 56,000 .001 to .015 41 1000 1 v 12 .014 to .019 43 - 99 -transfer to the liquid silver i s poor, and the effects of any bulk phase 41 stirring - in Chatterjee's , the bulk phase was stirred by paddle stirrers at 50 rpm. - become important in evaluating k Q. Since there is v i r t u a l l y no momentum transfer from the jet in the present experiment, the fact that 41 the observed k Q value agrees with the lowest obtained by Chatterjee i s an indication that the bulk phase stirring is similar for both experimental 43 conditions. Sano and Mori , on the other hand, achieved the same k values o as their contemporaries"^>4l>->0 using an oxygen jet of onty 12 dyn&> ^Ofidt. Clearly this is an indication that their silver melt experienced a high degree of inductive st i r r i n g , resulting in the large k Q values they obtained. 4.2.4. INFLUENCE OF INTERFACIAL TURBULENCE ON LIQUID-PHASE MASS  TRANSFER COEFFICIENTS A comparison of k Q values for copper and silver i s more meaningful than oxidation rates because this eliminates any differences in their res-22 pective driving forces. If the Surface Renewal Model of Higbie i s assumed to hold for both copper and silver, then i t would be expected that the ratio of values of k could be expressed as follows: o o 1.37 (10~4) 1.03 (10~4) 1.2 k Cu o - 100 -However, the results of the present experiments indicate that the following result obtains: k A g « 0.03 k C u o o This is a result of the lack of spontaneous interfacial turbulence during the oxidation of molten silver. Thus, i t can be said that interfacial tur-bulence is capable of enhancing k Q in the liquid phase by a factor of 40. To f u l l y appreciate the effect of interfacial turbulence on k Q values in molten copper and silver, from a theoretical standpoint, the Higbie 22 Surface Renewal Model , modified for a radial co-ordinate system, can be used to determine the effect of surface velocity on k for each metal. o Such values can be readily calculated from Eq. (6), Chapter 1, and compared to the values obtained experimentally in this work. Calculated values of k Q are summarized for different surface velocities in Table VIII where one can see that the value in copper for a stagnant system (U =0.1 cm sec X) is about a factor of 17 times less than that observed in a f u l l y turbulent system, while the observed k Q in silver appears to agree with that calculated for a stagnant system. The overall effect of surface tension-driven flow, then, is to enhance the mass transport of oxygen in the molten copper phase through vigorous stirring of the bulk metal phase. In Table VIII, i t can also be seen that the calculated and experimental k Q values for copper agree reasonably with each other, though the decrease in calculated k Q with increasing patch radius was not experimentally observed. The reason for this i s not at a l l clear at this time. For silver, the experimental k agrees with that calculated for a stagnant system, and this - 101 -TABLE VIII COMPARISON of CALCULATED and EXPERIMENTAL OXYGEN MASS TRANSFER COEFFICIENTS FOR COPPER AND SILVER T = 1220°C (Cu); 1100°C (Ag) , 2 Do/ C u = 1.029 (10-4) cm2 s e c - 1 ; D o/Ag = 1.37 (10 ) cm sec const (Eq.6) = 1.303 Liquid Patch Surface Renewal ko k 0 exptl. Metal Radius Velocity Time (cm/sec) (cm/sec) (cm) (cm/sec) (sec) Cu 0.5 0.1 5 0.0059 10 0.05 0.059 20 0.025 0.084 26 0.019 0.095 0.104 30 0.0167 0.102 Cu 1.0 0.1 10 0.0042 10 0.1 0.042 20 0.05 0.059 26 0.038 0.067 0.104 30 0.034 0.072 Cu 1.5 0.1 15 0.0034 <* 10 0.15 0.034 20 0.075 0.048 26 0.057 0.055 0.104 30 0.05 0.059 Ag 1.3 0.1 13 0.0042 1.0 1.3 0.0133 5.0 .26 0.030 10.0 .13 0.042 0.00288 - 102 -appears consistant with the observed lack of radially-directed surface flow during oxidation. Since precise values for surface motions in molten silver are not presently available, however, the observed agreement may only, be fortuitous. In other molten metals, which have high-temperature stable oxides, i t is possible that starvation mass transfer kinetics w i l l obtain for oxi-dation with low-momentum gas jets. Because of the complexity of high-tempera-ture reactions and of the number of variables involved, i t i s not possible at this time to develop equations for predicting k Q values using only the data from the present experiments. Further experiments on a wider range of liquid metals would be necessary before this is possible. 4.3. COPPER DEOXIDATION 4.3.1. DEOXIDATION RATES The observed deoxidation/time plots in Figs. 3.20 and 3.21 appear linear in a l l cases for the time periods shown. From the results of pure argon degassing in Fig. 3.20, i t is clear that oxygen w i l l not desorb into an inert, oxygen-free atmosphere, despite the existence of a large concen-tration driving force. Thus, the observed deoxidation rates are due entirely to the effects of the hydrogen gas. Changes in the flow rate of hydrogen to the copper surface do not appear to affect the deoxidation rates to any significant degree, as seen in Table V. For the low hydrogen flow rates used, the deoxidation rate is roughly constant, and, because a surplus of hydrogen gas is available for - 103 -reaction, starvation mass transfer kinetics do not obtain in these experiments. Variations in lance height were tried, but no change in the deoxidation rates was observed. Also, the variable starting oxygen concentration did not appear to affect the deoxidation rates. From the extended duration experiment in Fig. 3.22, i t can be seen that the constancy in deoxidation rate lasts for about 3000 sec (to ^  0.55 wt. % oxygen), beyond which i t decreases. This change indicates that a change is occurring in the deoxidation kinetics at this stage. A similar change was 39 observed by Themelis and Schmidt for the deoxidation of copper using sub-merged jets of CO gas, and this indicated a changeover from gas-phase mass transport control (linear) to liquid-phase control (curved), as was dis-cussed in sec. 1.3.3. Thus, the constant rate of deoxidation observed in Figs. 3.20 to 3.22 corresponds to gas-phase mass transport control (i.e. the transport of ^ ( g ) to the surface). A l l experiments were not continued for the extended duration because, as the oxygen concentration becomes small (about 0.12 wt. % ) , hydrogen begins to dissolve in the copper at an appreciable rate, as discussed by Small et a l ^ ^ . Under these conditions, water vapour w i l l form within the bulk of the copper rather than at the interface, leading to bubble formation and subsequent enhancement of k Q values. 4.3.1.1. EFFECT OF DISSOLVED SILICON AND SULPHUR As discussed in sec. 4.1.1.2., si l i c o n dissolved in copper does not exert any discernable effect on the rate of copper oxidation because i t forms - 104 -a soluble compound with C^ O"*". However, when the C^O content is reduced during deoxidation, the s i l i c a films could again reach the interface and perhaps have some influence on deoxidation rates through blockage of the interface to hydrogen gas. In Run 41, to which 20 ppm. of si l i c o n had been added prior to oxidation, a large surface patch was observed to form during deoxidation. Analysis of this patch showed i t to be composed of alumina and s i l i c a particles, and the analysis of the surfaces of the other de-oxidation experiments showed no traces of either of these materials. As seen in Table V, the deoxidation rate of Run 41 is about 30% less than the average value, indicating that the observed traces of alumina and s i l i c a retarded the deoxidation rate. In sec. 4.1.1.3., i t was found that dissolved sulphur lowered the rate of copper oxidation by an amount roughly equal to the rate of SO2 elimination. During deoxidation, i f any sulphur is present, the deoxidation rates could continue to be enhanced by the removal of oxygen as S02. Because 0.004% sulphur remained in the bath of Run 43 at the start of deoxidation, the large rate observed in Table V is l i k e l y aided by S0^ off-gassing. Sulphur contents were not measured, however, because of sample size restrictions. 4.3.2. SURFACE MOTIONS The type of surface motions described in sec. 3.3.1. were d i f f i c u l t to analyse because of their randomness, and the reported values of surface velocity of 10 to 15 cm sec 1 are only approximations. A detailed analysis of the surface flow is beyond the scope of the present study, however i t i s - 105 -useful to attempt an explanation of surface flow during the deoxidation of copper. Observations of surface flow during deoxidation have been reported ,. -, . 8,55 in the literature A schematic representation of the liquid copper interface during hydrogen deoxidation is seen in Fig. 4.7. From the data in Table V, i t i s clear that the bulk of the hydrogen flow to the interface remains unreacted, and so leaves the area as shown in the diagram. The major reaction site is at the point of jet impingement. When the hydrogen gas reaches the surface, i t reacts immediately with the dissolved oxygen to form 1^0 vapour, and analysis of the equilibrium constant (Appendix I) has shown this reaction to proceed completely to 1^0 vapour. Because of this deoxidation, the copper immediately beneath the lance now has a larger interfacial tension than the remainder of the interface, and so inwardly-directed surface flow w i l l result. In the i n i t i a l stages of deoxidation, this surface flow was found to stop when the hydrogen was brie f l y turned off. The surface flow w i l l decrease on increasing time because the surface tension driving force also decreases, and in the liquid-phase control region, no surface flow was observed. 4.3.3. MASS TRANSFER COEFFICIENTS For the interfacial reaction H 2(g) + [0] Cu (10) - 106 -FIGURE 4-7 Schematic representation of sti r r i n g pattern and surface flow conditions during deoxidation of molten copper and silver with H gas. - 107 -i t can be seen from Fig. 4.7, that the majority of the deoxidation should occur at the point of jet impingement, and that since hydrogen is relatively 55-58 insoluble in molten copper at these oxygen concentrations , no de-oxidation occurs within the bulk, but is confined to the surface. In sec. 4.3.1. i t was shown that the rate of copper deoxidation, during the f i r s t 3000 sec, is controlled in the gas phase, and so hydrogen mass transfer coefficients, kp 2, can be calculated from the rate data in Table V using Eq. ( 1 1 ) , K R 2 b i n H 2 = ~ . A ( P H 2 " PH2"> < N> where PH2> ^H^ r e f e r t o t n e bulk-phase and interfacial hydrogen gas pressures respectively, and R is the gas constant. Because the equilibrium constant is large (see Appendix I), i t is reasonable to assume that P T J 2 is zero. The value for the area A was taken as the total surface area. From Table IX, the mean value for k j ^ a t a hydrogen flow rate of 1500 3 -1 -1 cm min was found to be 1.28 cm sec with a standard deviation of 19.5%, and this value agrees with the kjj 2 found for a hydrogen flow rate of 3000 3 - 1 3 - 1 cm min . For the low flow rate of 1000 cm min , the kjj 2 value was found to be about 30% less than those observed at the higher flow rates, indicating that k^ 2 becomes dependent upon hydrogen flow rate at values less than 1500 3 , -1 cm min In Run 41, the presence of dissolved s i l i c o n reduced the k ^ value to 0.89 cm sec 1 through blockage of the interfacial regions to hydrogen. It is interesting to note that this value i s similar to the kg 2 obtained for the low hydrogen flow rate experiment. The presence of residual sulphur in Run 43 TABLE IX MASS TRANSFER COEFFICIENTS FOR THE DEOXIDATION OF MOLTEN COPPER AND SILVER Run Number H 2 Flow (cm^ min - 1) nH 2 moles ,„4 x 10 sec TO SURFACE n H 2 moles .,_4 x 10 sec REACTED A (cm2) kH 2 cm sec Copper 32 1000 7.6 1.80 25 0.88 33 1500 11.2 2.20 25 1.08 37 3000 22.3 2.09 19.6 1.30 38 1500 11.2 1.99 19.6 1.24 39 1500 11.2 2.14 19.6 1.34 40 1500 11.2 2.68 19.6 1.67 41* 1500 11.2 1.42 19.6 0.89 43+ 1500 11.2 4.30 19.6 2.68 45+ 1500 11.2 1.63 19.6 1.02 50 1500 11.2 1.70 19.6 1.06 * contained dissolved Si + contained dissolved S Run Number H 2 Flow (cm^ min - 1) nH 2 moles 0 -„4 x 10 sec TO SURFACE n H 2 moles 0 1 r t4 x 10 sec REACTED A (cm2) ko cm sec Co t=0 moles .3 3 x 10 cm Silver 4 1500 11.2 1.92 11.6 .146 0.5 5 1500 11.2 .89 11.6 .063 0.3 6 1500 11.2 .17 10.2 .010 0.25 7 3000 22.3 .13 10.2 .025 0.2 - 109 -enhanced the k j ^ value to 2.68 cm sec X through added oxygen loss as SO2 gas. Thus, i t appears that the presence of surface-active solutes other than oxygen can slightly influence the apparent rates of hydrogen mass transfer in the gas phase during copper deoxidation. As mentioned previously, there i s a changeover from gas-phase to liquid-phase control of deoxidation occurring at about 3000 sec. for the copper deoxidation experiments. Using this, i t i s possible to obtain an approximate value for the liquid-phase oxygen mass transport coefficient k Q, but owing to the increased solubility of hydrogen in the bulk of the copper, deoxidation can now occur within the bulk copper phase, so that the k Q value can only be approximate. A plot of In (^/QQ^JJITIAL^ V S * time in Fig. 4.8 shows the definite change in slope at 3000 sec, and the slope of the dashed line in the liquid-phase control region gave the k Q -3 -1 value of k = 4.9 (10 ) cm sec o which i s very much smaller than the value of k Q determined for oxidation. Removal of the remaining oxygen i s a slow process, which was also observed 38 55 by Frohme et al and Small and co-workers As a check on the deoxidation kinetics, we can calculate the bulk oxygen concentration at which changeover from gas-phase to liquid-phase mass transport control occurs using the following relationship: o" " k„ ' RT where values for kir 0 and k are those determined in the present experiments. n z o - 110 -1 1 1 1 1 I 1 I I I I I 1 1000 2 0 0 0 3 0 0 0 4 0 0 0 5 0 0 0 6 0 0 0 TIME (sec) FINAL FIGURE 4*8 Plot of - In _o_ vs. time for copper deoxidation C INITIAL o showing change from gas-phase to liquid phase mass transport control of deoxidation of about 3000 sec (^  0.37 wt % 0^  in the bath). - I l l -1.28 1 4.9 (10~3) (81.94)(1493) 2.14 (10 ) moles 0 • cm' -3 From Fig. 3.22, we can see that the observed changeover occurs at about -3 -3 0.55 wt % 0 (3 x 10 moles 0 • cm ), and this agreed with the calculated value, confirming that gas-phase mass transport control obtains for the i n i t i a l 3000 sec. of copper deoxidation. 4.4. SILVER DEOXIDATION 4.4.1. DEOXIDATION RATES As observed for silver oxidation, the results for deoxidation proved 3 -1 to have considerable scatter among the three experiments at 1500 cm min hydrogen flow (Fig. 3.24). Deoxidation experiments were carried out immedi-ately following each oxidation experiment, but i t was noticed that during the argon gas flushing, prior to hydrogen jetting, each sample would lose about 50% of i t s oxygen content. Thus, i t would appear that oxygen dissolved in silver i s unstable and requires an oxidizing atmosphere in order to remain in solution, unlike the situation observed for copper in sec. 4.3.1. 43 This trend was also observed by Sano and Mori in their study of oxygen desorption from molten silver to flowing argon gas. In Fig. 3.24 and Table IX i t can be seen that for the same gas flow rate, three different deoxidation rates were obtained. Also, though i t is d i f f i c u l t to assess accurately for the two hydrogen flow rates employed, there appears to be l i t t l e dependence of the deoxidation rate on hydrogen - 112 -flow rate. It is seen, however, that the deoxidation rate dzcAZCU>2A on decreasing bath oxygen content with the exception of Run 6, where analytical errors are believed responsible for the different behaviour observed. Since gas bubbling was observed throughout silver deoxidation and because the deoxidation rate depends on bath oxygen concentration, i t i s clear that the rate-controlling step cannot be hydrogen transport to the bath surface, but is liquid-phase transport of oxygen. As for copper deoxidation, this can also be shown in the following manner. At the change from gas-phase to liquid-phase control, the respective mass transport rates are related as follows: b k H2 b k o C o = S T • PH2 ^ Since no value of at the changeover point in liquid silver i s available, i t has to be estimated from the data. In sec. 4.2.4., i t was shown that, in the absence of interfacial turbulence during oxidation, k^ U k^ g. o o This equality should also roughly hold for deoxidation since the degree of turbulent behaviour is similar in both metals. Also, because mixing in the gas phase should be the same for both metals under the same jet conditions, vCu „ Ag kH 2 - kH 2" .Cu .Ag kH 2 k H l Thus —zr~ - —r~ > and, substituting into Eq. (9), the value for k C u k A g o o b —3 —3 C at the change of control is found to be 2.3 (10 ) moles 0 • cm in molten o silver. From Table IX, i t can be seen that the starting oxygen concentrations are a l l less than the above value, and so the rate-limiting step is oxygen transport in the liquid phase. Because bubbling was observed, i t i s l i k e l y - 113 -that some deoxidation is occurring within the liquid, and so subsequent k Q S calculations can only be considered approximate due to the enhancement in mixing caused by water-vapour bubbling. 4.4.2. SURFACE MOTIONS As observed for copper in sec. 4.3.2., the type of surface motions found for silver were d i f f i c u l t to analyse because of their randomness, and the reported values of 10 to 15 cm sec 1 are only approximations. Furthermore, because bubbling was observed during a l l silver deoxidation runs, the resulting surface motions become even more complicated. A schematic repre-sentation of the silver surface is shown in Fig. 4.7, and the flow description for copper in sec. 4.3.2. w i l l also apply to silver deoxidation, with the added complication of bubbling. The surface flow w i l l decrease with increasing time because the surface tension driving force also decreases with increasing time. 4.4.3. MASS TRANSFER COEFFICIENTS As shown in the previous section, oxygen transfer in the bath i s the rate-limiting step in the deoxidation of molten silver with hydrogen gas. Therefore, oxygen mass transfer coefficients, k^, can be calculated from Eq. (8), - 114 -- I 1 1 1 1 1 — 100 2 0 0 300 4 0 0 5 0 0 6 0 0 TIME (sec) FIGURE 4-9 Liquid-phase mass transfer coefficients for oxygen during the deoxidation of molten silver with hydrogen gas. - 115 -The value for the Interfacial oxygen content, C^, is zero in the region of liquid-phase mass transfer control, and the area A was taken as being the whole interface. The results of these calculations are presented in Table IX and in Fig. 4.9, as a plot of In x vs time. As can be seen in Table IX the values of k Q decrease with decreasing oxygen concentration. This same dependence of both deoxidation rate and k Q values on the i n i t i a l 43 silver bath oxygen content was also observed by Sano and Mori in their study of oxygen duoAptlon rates from molten silver to flowing argon gas. On comparison with the k Q value determined for copper deoxidation, the values obtained for silver are about an order of magnitude larger. It is f e l t that this difference i s due both to the bubbling observed during silver deoxidation and to the fact that silver deoxidation was accompanied by inter-f a c i a l turbulence in the liquid-phase control region, and copper deoxidation was not. Because of the scatter observed in the present data, however, i t is d i f f i c u l t to ascertain the value of this comparison. 4.5. SOLID DISSOLUTION EXPERIMENTS  4.5.1. SPREADING The experimental observations of the spreading and turbulence generated during the dissolution of Cu^S, C^O, Se and Te are very similar to those made in sec. 3.1.1. during copper oxidation. Taking Cu^S as an example, the driving force for spreading of one liquid over another is the difference between the surface tension of pure copper, ar , and the sum of the surface tension of molten Cu0S, an c, and the Interfacial tension along the i- t>U2>5 - 116 -Cu9S-Cu interface, ar „,r . This i s often expressed as the i n i t i a l spread-u ing coefficient, S; as described by Eq. ( 5 ) , sec. 1.3.1: S = Cu2S + a Cu2S/Cu ) (5) Similar expressions can be obtained for the other three materials. Naturally, spreading w i l l only occur i f S i s positive. Some of the observed spreading and interfacial turbulence may also be due to a local reduction of the surface tension in the v i c i n i t y of the partially immersed solids (sec. 1.2.2.) where a higher solute concentration is l i k e l y , relative to the copper nearer the crucible walls. The surface would then also move spon-taneously toward the region of higher CQ U« As can be appreciated, the liquid-phase surface tensions for the four solids used in this test are rather d i f f i c u l t to obtain. Some tentative values have been located in the literature so that values of S can be roughly calculated from Eq. ( 5 ) , and these are given in Table X. The values of S in Table X are a l l very much larger than zero, as might be expected, because of the observed spreading behaviour of these four materials. Thus, this calculation could prove valuable for both industrial and laboratory experiments involving partially immersed solids in liquid metals. Baes and Kellogg^ 1 have reported observations of spontaneous surface motion during the dissolution of Cu^S in partially molten copper in conjunction with their sessile drop study of the effect of sulphur on the surface tension of copper. They observed that small crystals of Cu^S were thrown off the surface of the copper by surface eruptions and that during melting, samples - 117 -TABLE X CALCULATED VALUES FOR THE SPREADING COEFFICIENT, S, FOR Cu2S, Cu20, Se, AND Te Material (m) Temperature (°C) Or-(dynes/cm) o"m/Cu (dynes/cm) m^ (dynes/cm) Cu 1200 1250 5 8 - -Cu2S 1200 59 400 59 520 330 Cu20 1200 59 500 * 7001 1 ^ 50 Se 220 1200 105.5 5 8 11 850 294.5 Te 1*68 174.4 5 8 725 1 1 346.3 - 118 -of Cu^S + copper would often spontaneously f a l l off the sessile drop plaque. However they attributed these phenomena to melting effects rather than to spreading of the surface due to sulphur. Based on the present work, the latter effect is most l i k e l y to have been the driving force for the movements they observed as well. A. 5.2. SIGNIFICANCE The finding that a macroscopic source of sulphur, selenium and tellurium, as well as of oxygen, can generate spontaneous surface motion is important since i t shows that surface driven flow at high temperatures is not unique to the oxygen/metal system. This has implications that are worth considering when studies are undertaken of the kinetics of hetero-geneous reactions in either the laboratory or in the metallurgical plant. In both cases due consideration should be given to the possibility of interfacial turbulence being present, since idealized hydrodynamic con-ditions can be fundamentally altered and reaction rates sharply accelerated. This consideration is especially important when the reaction involves a known surface-active agent. Certainly in the laboratory where experimental conditions are subject to careful control, the presence of interfacial turbulence should be checked whenever possible. For example, the mechanism outlined here and in sec. 1.2.2. has been used to explain the preferential attack of refractory bricks at the surface line of steel and glass-making furnaces. Many other metallurgical reactions yet to be examined in this way, such as re-oxidation of liquid metals during casting, are undoubtedly accompanied by interfacial turbulence. - 119 -5. CONCLUSIONS The rate of oxidation of molten copper at 1220 C by low-momentum pure oxygen jets is governed by starvation mass transfer kinetics, and i s not affected by changes in lance height or small additions of sil i c o n . The effect of dissolved sulphur i s to reduce the rate by an amount equal to the rate of sulphur elimination as S0 2 gas. Surface velocity studies show that the interfacial tension-generated flow during copper oxidation has a mean value of 26.1 cm sec 1 (± 21%) and does not depend on oxygen gas flow rate or copper bath oxygen concentration. Oxide patch areas are found to increase with both oxygen flow rate and increasing bath oxygen concentration. Surface blockage studies show that most of the oxygen transfer to the copper occurs in the area described by the oxide patch. Oxygen mass transfer coefficients in the copper liquid phase are found to be 0.104 cm sec 1 (± 11.6%), and are independent of oxygen flow rate, bath oxygen content, and dissolved si l i c o n and sulphur contents. Literature values are not available for comparison. - 120 -The rate of molten silver oxidation at 1100°C is not governed by starvation mass transfer kinetics, but is controlled by transport in the liquid phase. The oxidation rate is not dependent on oxygen gas flow rate and is seen to be a factor of 50 times less than that observed for copper. No spontaneous interfacial tension-generated flow i s observed during molten silver oxidation. The reason postulated for this observation is that oxygen absorption by silver occurs uniformly across the inter-face. Because silver has no stable oxide at the temperature studied, a point AOuAce. of low from which spreading can occur is not able to exist at the point of jet impingement. Oxygen mass transfer coefficients in the silver liquid phase are found -3 -1 to be 2.88 (10 ) cm sec , and are independent of oxygen flow rate and bath oxygen content. A direct comparison with the literature values is not possible due to experimental differences. However, i t is shown that for non-turbulent systems, the degree of external bulk-phase mixing.is important in evaluating k Q values. The effect of interfacial turbulence on liquid-phase oxygen mass transfer coefficients in molten copper i s to enhance the value by about 40 times over that observed in molten silver where interfacial turbulence does not occur on oxidation. ! o The rate of deoxidation of molten copper at 1220 C by low-momentum pure hydrogen jets does not depend upon hydrogen flow rate, lance height and - 121 -starting oxygen concentration, and is constant for the f i r s t 3000 seconds. The rate-controlling step i s the gas-phase mass transfer of hydrogen to the liquid surface. Dissolved si l i c o n retards the de-oxidation rate by blocking the interface to hydrogen gas, and dissolved sulphur enhances the rate through SO2 elimination. Surface velocities of 10 to 15 cm sec 1 towards the point of jet impingement are observed, but because of the randomness of these motions, a detailed analysis is not possible. The proposed mechanism for this flow i s inward spreading to a region of higher a directly beneath the hydrogen lance than at a point on the interface some dis-tance from the lance. As the flow depends upon the bath oxygen con-centration, i t w i l l decrease with time during refining. The hydrogen mass transfer coefficient to molten copper at 1220°C for 3 -1 a hydrogen flow rate of 1500 and 3000 cm min i s found to be -1 3 -1 1.28 cm sec (± 19.5%). At a hydrogen flow rate of 1000 cm min , this value is reduced by about 30%, indicating that a dependence on 3 -1 hydrogen flow rate occurs at values below 1500 cm min . The k ^ values are reduced to 0.89 cm sec 1 in the presence of dissolved sili c o n , and enhanced to 2.68 cm sec ^  in the presence of dissolved sulphur. An approximate k Q value for the liquid-phase mass transfer control -3 -1 region i s found to be 4.9 (10 ) cm sec , and is influenced by the presence of bubbling during this phase of deoxidation. This k Q value is similar to that for a stagnant system. - 122 -The rate of deoxidation of molten silver at llOO^C by low-momentum pure hydrogen jets does not depend upon the hydrogen flow rate, but does depend upon the starting bath oxygen concentration, decreasing as the concentration decreases, and li k e l y depends on the bubbling observed during deoxidation. The rate-controlling step is liquid-phase oxygen mass transfer. Surface velocities of the same magnitude as seen for copper deoxidation are observed for molten silver deoxidation, and the proposed mechanism for this flow is identical to that for copper, with the added com-plication of gas bubbling. Oxygen mass transfer coefficient in the liquid phase for hydrogen flow 3 -1 rates of 1500 and 3000 cm min are found to decrease with decreasing i n i t i a l oxygen concentration. These values are enhanced by both the presence of bubbling and the interfacial tension-generated surface flow during the deoxidation of the silver, but i t is impossible at this time to determine the degree of this enhancement. Interfacial turbulence during the dissolution of solid C^S, Cu^O, Se and Te in molten copper is shown to occur. Values calculated for the spreading coefficient, S, indicate that spreading of these materials on molten copper, and the resulting turbulence, is a predictable effect. - 123 -6. SUGGESTIONS FOR FUTURE WORK 1. The oxidation of molten copper using jets of air at various flow rates could be studied to determine i f interfacial turbulence w i l l occur under these conditions. 2. Other metal systems having high-temperature stable oxides could be studied to determine i f interfacial turbulence occurs during oxidation with low-momentum oxygen jets, and what effect this has on the oxidation rates. A partial l i s t of such metals could include iron and nickel. It would also be of interest to study the oxidation of these metals using low-momentum jets of a i r . 3. It would be of interest to study the deoxidation of both molten copper and silver using low-momentum jets of CO. Because CO i s insoluble in molten copper and silver, bubb-LLng during deoxidation would not be a problem, and i t would be interesting to compare the deoxidation rates with those obtained for hydrogen. 4. Surface blockage studies for molten silver could be performed to determine the effect of interfacial area on oxidation and deoxidation. This could also be studied for copper deoxidation. 5. Oxidation and deoxidation studies for molten copper, silver, iron and nickel could be performed over a range of temperatures to determine the effect of temperature on the oxidation/deoxidation rates and on the - 124 -magnitude of the interfacial turbulence in each case. The influence of surface-active agents, such as s i l i c o n , sulphur, selenium and tellurium, on the oxidation rates of silver, iron and nickel, could also be studied. - 125 -BIBLIOGRAPHY 1. Thompson, J.: Phil. Mag., 10 (4), 1855, p.330-3. 2. Marangoni, C.G.M.: Am. Phys. (Paggendorff), 143, 1871, p.337. 3. Sawistowski, H., and James, B.R.: Akh. Deut. Akad. Wiss Berlin Kl. Chem., Geol. Biol., 1966 (6), p.757. 4. E l l i s , S.R.M., and Biddulph, M.: Ch. Eng. Soi., 21_ (1966), p.1107. 5. Davies, J.T.: "Mass Transfer and Interfacial Phenomena", in Advances in Chem. Eng., Vol.4, ed. T.B. Drew, J.W. Hoopes, and T. Vermulen, Acad. Press N.Y., 1963. 6. Berg, J.C.: "Interfacial Phenomena in Fluid Phase Separation Processes", in Recent Developments in Separation Science, ed. N. L i , Chem. Rubber Co., Cleveland, 1972. 7. Sharma, B.D., and Asundi, A.K.: Trans Indian Inst. Metals, 26_ (3), 1973, p.33. 8. Richardson, F.D.: Trans. I.S.I.J., 14 (1974), p . l . 9. Kozakevitch, P.: "Surface Phenomena of Metals", Society of Chemical Industry Monograph,28 (1968), p.223. 10. Korol'Kov, A.M.: "Casting Properties of Metals and Alloys',' Consultants Bureau, New York, 1960, p.37. 11. Monma, K., and Suto, H.: Trans. J.I.M., 2 (1961), p.148-53. 12. Eremenko, V.N., Naidich, Yu. V., and Nosonovich, A.A.: Russ. J. Phys. Chem., 34^  (5), 1960, p.484. 13. Bernard, G., and Lupis, C.H.P.: Met. Trans., 2 (1971), p.2991. 14. Stanek, V., and Szekely, J.: Chem. Eng. Sci., 25 (1970), p.699-715. 15. Brimacombe, J.K., and Weinberg, F.: Met. Trans., 3_ (1972), p.2298-9. - 126 -16. Komasawa, I., Saito, T., and Otake, T.: Int. Chem. Eng., 12 (2), 1972, p.345. 17. Brian, P.L.T.: A.I.Ch.E.J., 17_ (4), 1971, p.765. 18. Glen, C.G., and Richardson, F.D.: "Heterogeneous Kinetics at Elevated Temperatures", ed. G.R. Belton and W.L. Worrel, Plenum Press, 1970, p.369. 19. van Langen, J.M.: J.I.S.I., Nov. 1960, p.262. 20. Holden, C., and Hogg, A.: J.I.S.I., Nov. 1960, p.318. 21. Bradshaw, A.V., and Chatterjee, A.: Ch. Eng. Sci., 26 (1971), p.767. 22. Higbie, R. : Trans. Am. Inst. Ch. Eng., 31_ (1935), p.365. 23. Szekely, J., and Themelis, N.J.: "Rate Phenomena in Process Metallurgy", Wiley, 1971, p.428. 24. Brimacombe, J.K., and Richardson, F.D.: Trans. Inst. Min. Met., London, section C80, 140 (1971). 25. Brimacombe, J.K.: J. Metals, 23. (1971), 41A. 26. Okhotskii, V.B., Chernyatevich, A.G., and Posouirin, K.S.: Steel in the U.S.S.R., June 1972, p.443. 27. Emi, T., Boarstein, W.M., and Pehlke, R.D.: Met. Trans., 5_ (1974), p.1959. 28. Brimacombe, J.K.: Unpublished research. 29. Turkdogan, E.T.: Ch. Eng. Sci., 21 (1966), p.1133. 30. Flinn, R.A., Pehlke, R.D., Glass, D.R., and Hays, P.O.: Trans. A.I.M.E., 239 (1967), p.1776. 31. Molloy, N.A.: J.I.S.I. Oct. 1970, p.943. 32. Chatterjee, A., and Bradshaw, A.V.: J.I.S.I. Mar 1972, p.179. 33. Szekely, J., and Asai, S.: Met. Trans., 5_ (1974), p.463. 34. Davenport, W.G., Wakelin, D.H., and Bradshaw, A.V.: Proc. Symp. on Heat and Mass Trans, in Process Met., Inst. Min. Met., (1966), p.207. 35. Toop, G.W., and Richardson, F.D.: Adv. in Extr. Met., Inst. Min. Met., (1968), p.181. - 127 -36. Nanda, C.R., and Geiger, G.H.: Met. Trans., 2 (1971), p.1101. 37. Gerlach, J., Schneider, N., and Wuth, W.: Metall wissen schaft und Technik, 25 (11), 1971, p.1245. 38. Frohme, 0., Rothman, G., and Wuth, W.: Metall. 27 Jahrgang (1973), no.11, p.1112. 39. Themelis, N.J., and Schmidt, P.R.: Trans AIME, 239 (1967), p.1313. 40. Brimacombe, J.K., Stratigakos, E.S., and Tarassoff, P.: Met. Trans. 5 (1974), p.763. 41. Chatterjee, A., Wakelin, D.H., and Bradshaw, A.V.: Met. Trans., _3 (1972), p.3167. 42. Chatterjee, A., and Bradshaw, A.V.: Met. Trans., 4_ (1973), p.1359. 43. Sano, M., and Mori, K.: J.I.S.I. Japan, 60 (10), 1974, p.1432. 44. Analysis. Can Test Ltd. F i l e no's 8681 A, 9012 A, 107 B, and 1285 B. 45. Analysis. Asarco Corp., Denver Colorado. 46. Forster, A., and Richardson, F.D.: Trans. Inst. Min. Met., 84, (1975) p.C116. 47. Chaklader, A.CD.: private communication. 48. Pehlke, R.D., and E l l i o t t , J.F.: Trans AIME, vol.227 (1963), p.844-55. 49. Johnson, R.E.: in Metals Handbook, Vol.8, p.295, A.S.M. Cleveland, Ohio. 50. Distin, P.A., Hallett, G.D., and Richardson, F.D.: J. Iron Steel Inst. London, 206_j_ 821 (1968). 51. Davenport, W: Ph.D. thesis, Royal School of Mines, Imperial College, London, 1965. 52. Buttner, F.H., Funk, E.R., and Udin, H.: J. Phys. Chem., 56, May 1952, p.657. 53. Chaston, J.C., in Metals Handbook, vol.8, p.254, A.S.M., Cleveland, Ohio. 54. Fischer, W.A., and Ackermann, W.: Arch. Eisenhuttenw, 37_ (1966), p.43. 55. Small, W.M., Radzilowski, R.H., and Pehlke, R.D.: Met. Trans., 4^  (1973) p. 2045. - 128 -56. Weinstein, M., and E l l i o t t , J.F.: Trans. AIME, 227 (1963), p.285. 57. Sacris, E.M., and Parlee, N.A.D.: Met. Trans., 1 (1970), p.3377. 58. Wright, J.H., and Hocking, M.G.: Met. Trans., 3 (1972), p.1749. 59. Handbook of Chemistry and Physics, 53rd Edition, Chemical Rubber Company, Cleveland, Ohio, 1972-73. 60. "The Physical Chemistry of Process Metallurgy: The Richardson Conference," eds. J.H.E. Jeffs and R.J. Tait, I.M.M., 1974. 61. Baes, C.F., and Kellogg, H.H.: Trans. AIME, vol.197, 1953, p.643-8. 62. Mizikar, E.A., Grace, R.G., and Parlee, N.A.D.: ASM Trans., 1963, vol. 56, pp.101-06. 63. Sieverts, A., and Hagenacker, J.: Z. Phys. Chem., 68, 1909-10, pp.115-28. 64. Donan, F.G., and Shaw, T.W.A.: J. Soc. Chem. Ind. London, 29, 1910, pp.987-9. 65. Parlee, N.A.D., and Sacris, E.M.: Trans. Am. Inst. Min. Engrs., 233, 1965, pp.1918-9. 66. Kubaschewski, 0., Evans, E., and Alcock, C.B.: "Metallurgical Thermo-chemistry", Pergamon Press, 4th Edition, 1967. 67. Seow, H.P., and Biswas, A.K.: Proc. Australas. Inst. Min. Metall., 245, 1973. 68. Gaskell, D.R.: "Introduction to Metallurgical Thermodynamics", McGraw-Hill, 1973. 69. Baker, E.H., and Talukdar, M.I.: Trans. Inst. Min. Met., 77_, 1968, C128. 70. Luraschi, A., and E l l i o t t , J.F.: Met. Trans., 6E, 1975, p.63-75. -129 -APPENDIX I CALCULATION OF THE EQUILIBRIUM CONSTANT K^un 1. KH 2O FOR THE COPPER-OXYGEN-HYDROGEN SYSTEM Hydrogen gas reacts with oxygen dissolved in copper, at the gas-liquid interface, to form H 20(g), as Cu K H 2 ° H 2(g) + [%0] C u - H20(g) (1) and the equilibrium constant Kj^n is given as T,Cu PH20 K Q = _ PR 2 • Y 0 • [%0]Cu where P^O a n ^ ^ H2 a r e i n t e r f a c i a l concentrations of water and hydrogen gases. As these interfacial pressures are hard to evaluate, i t i s desirable to determine KH2Q. from the free energy for reaction (1) by using the following expression: A G ( 1 ) = " R T l n KH20 Assuming that equation (1) is valid to the saturation l i m i t , the free energy AG°^ was derived from the following equations: - 1 3 0 -C u2°(s) 2 C u ( l ) + ° 2 ( g ) ( 2 ) .+ H 2(g) + h 0 2(g) -»• H 20(g) ( 3 ) C U 2 ° ( S ) + H 2 ( G ) * 2 C u ( l ) + H 2 0 ( g ) ( 4 ) A G ( 2 ) = 4 6 » 7 0 0 + 3 , 9 2 T 1 ° g T ~ 3 4 . 1 T (Ref. 6 6 ) + A G ° 3 ) = - 5 7 , 2 5 0 + 4 . 4 8 TlogT - 2 . 2 1 T (Ref. 6 6 ) A G ° 4 ) = - 1 0 , 5 5 0 + 8 . 4 0 TlogT - 3 6 . 3 1 T C U 2°(s) + H 2 ( G ) 2 C u ( l ) + H 2 ° ( G ) ( 4 ) - C u 20 ( s ) + [%0] C u + 2 C u ( 1 ) (5) H 2(g) + [%0] C u H20(g) (6) A G ° 4 ) = -10,550 + 8.40 TlogT - 36.31 T - A G ° 5 ) = 22,003 + 3.92 TlogT - 27.82 T (Ref. 67,70) AG*?,. = -32,553 + 4.48 TlogT - 8.49 T At the experimental temperature of 1220°C (1493°K), the free energy value i s Thus, K^o i s ^H20 oxygen concentration AG° . = -23,998 cal mole 1 °K 1 \b) KS!J0 = 3354.7, and based on a 1 wt % standard-state KROO = 5 . 8 1 ( 1 0 5 ) cm3 mole 1 - 131 -2. K^| 0 FOR THE SILVER-OXYGEN-HYDROGEN SYSTEM Ag As was the case for copper, Kj^o w a s derived from the free energy AG°^ for the following equations: A g2°(s) -* 2 A g ( l ) + h ° 2 ( g ) + H 2(g) + !j 02(g) H 20 ( g ) A g2°(s) + H 2 ( g ) 2 A g ( l ) + H 2 ° ( g ) o ( 7 ) (8) (9) AG^y) = 3,250 - 6.8 T (Ref. 68) AG° . = -57,250 + 4.48 TlogT - 2.21 T (Ref. 66) AG° 9 ) = -54,000 + 4.48 TlogT - 9.01 T A g2°(s) + H 2 ( g ) 2 A g ( l ) + H 2 0 ( g ) (9) - A g 2 ° ( s ) ' [ % 0 ] A g + 2 A g ( l ) ( 1 0 ) H 2(g) + [%0] -> H20(g) (11) AG ° 9 ) = -54,000 +4.48 TlogT - 9.01 T - A G ° 1 0 ) = -681 (Ref. 69,70) A G(11) = ~ 5 3> 3 1 9 + 4 , 4 8 T 1 ° g T " 9.01 T At the experimental temperature of 1100°C (1373°K), the free energy value is A G ( H ) = _ 4 6 > 3 9 0 c a l raole"1 ° K - 1 . Thus, K ^ G 0 is KH|O = 2.58 (10 8), and based on a 0.1 wt % standard state oxygen concentration, K ^ G 0 = 4.5 (10 1 1) cm3 mole X . 

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}]}"
                            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:
http://iiif.library.ubc.ca/presentation/dsp.831.1-0079076/manifest

Comment

Related Items