UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Sensing and modelling for oxygen lead softening Kapusta, Joël Patrick Thierry 1994

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

Item Metadata


831-ubc_1995-059855.pdf [ 3.53MB ]
JSON: 831-1.0078460.json
JSON-LD: 831-1.0078460-ld.json
RDF/XML (Pretty): 831-1.0078460-rdf.xml
RDF/JSON: 831-1.0078460-rdf.json
Turtle: 831-1.0078460-turtle.txt
N-Triples: 831-1.0078460-rdf-ntriples.txt
Original Record: 831-1.0078460-source.json
Full Text

Full Text

SENSING AND MODELLINGFOR OXYGEN LEAD SOFTENINGbyJOEL PATRICK THIERRY KAPUSTAMaItrise de Science, Université de LYON I, France, 1985Diplôme d’Etudes Approfondies, Université de NANCY I, France, 1986A THESIS SUBMITTED IN PARTIAL FULFILLMENT OFTHE REQUIREMENTS FOR THE DEGREE OFDOCTOR OF PHILOSOPHYinTHE FACULTY OF GRADUATE STUDIES(Department of Metals & Materials Engineering)We accept this thesis as conformingto the required standard/THE UNIVERSITY OF BRITISH COLUMBIAAugust 1995© Joel Patrick Thierry Kapusta, 1995In presenting this thesis in partial fulfilment of the requirements for an advanceddegree at the University of British Columbia, I agree that the Library shall make itfreely available for reference and study. I further agree that permission for extensivecopying of this thesis for scholarly purposes may be granted by the head of mydepartment or by his or her representatives. It is understood that copying orpublication of this thesis for financial gain shall not be allowed without my writtenpermission.(Signature)Department of___________________The University of British ColumbiaVancouver, CanadaDate S1, ct95DE-6 (2/88)ABSTRACTLead ores generally contain significant amounts of arsenic and antimony. The process oflead softening, corresponding to the removal of these hardening impurities, is therefore of necessityin any pyrometallurgical lead refining circuit. Even in a smelter where electrorefining is used, apartial softening might be required. The treatment of silver bearing concentrates, while providingadditional revenues, also introduces additional amounts of arsenic and antimony into the circuit.The efficiency of the Betts electrorefining process depends on the stability of the corroding leadanodes. Anode slime stability is directly related to the antimony and arsenic content of the anodes.Thus, control of bullion quality prior to anode casting is key to optimization of the electrorefiningoperations, improvement of lead production, and addition of revenues. The (partial) softening stageis the ideal location in a smelter for such a control. When composition spikes are regular occurrences,the batch softening circuit tends to become somewhat complicated to reduce their effects. In thisthesis work, the option of continuous single pass softening has been explored. In particular, thetwo critical components to this revision, i.e. a method of continuously monitoring bullion quality,and a control strategy, were investigated.An oxygen probe for continuous measurements in molten lead has been designed in thelaboratory prior to being tested in an industrial environment. A commercially available yttriastabilized zirconia serves as solid electrolyte. The reference system is composed of a Cu-Cu20mixture. Both lead wire and conducting lead (probe housing) are made of 316 stainless steel.Sealing is achieved by means of a high temperature magnesia cement. An additional plug of copperpowder, isolated from the reference system by a layer of alumina powder, serves as oxygen getter— 11 —to eliminate oxygen ingress from the atmosphere. The extremity of the lead wire that is insertedinto the probe is coated with cement to avoid any short-circuit with the Cu plug. These featureswere decisive in the success of the probe which provides a continuous measurement for severalconsecutive days. Once it was established that the probe was giving satisfactory measurements inthe laboratory, i.e. quick response, proper response to temperature changes and oxygen potentialchanges, a testing campaign was carried out in the plant. The campaign was successful and acorrelation between measured emf and As+Sb bullion content was established.Since any control decision would be based on the probe readings, it is crucial to regularlyensure the proper functioning of the probe. A method based on the thermal arrest technique hasbeen tested and calibrated to provide for a quick assessment of the probe reliability.A thermodynamic model of the current semi-batch process was developed. An analysis ofthe process showed that a thermodynamic model can be used to represent process operation, andan equilibrium analysis gave a reasonable fit to operating data. The model developed for semi-batchsofteningwas modified to continuous softening in order to simulate continuous single pass softening.The preliminary calculations showed that the concept of continuous single pass softening will meetthe target set for lead softening, assuming the process operates close to thermodynamic equilibrium.A slag high in As+Sb can be produced at bullion compositions in the target range for electrorefining.Based on the data presented in this thesis, a simple feedback control algorithm could be developedto regulate °2 injections to a continuous softener on the basis of the measured level of As+Sb inoutput bullion.— 111 —TABLE OF CONTENTSPageABSTRACTTABLE OF CONTENTS ivLIST OF TABLES viiiLIST OF FIGURES xLIST OF NOMENCLATURE xivACKNOWLEDGEMENTS xviiCHAPTER 1INTRODUCTION: LEAD REFINING 11.1 Lead Softening Practices 31.2 Cominco’s Lead Softening Practice 13CHAPTER 2LITERATURE REVIEW: OXYGEN PROBES AND PROCESS CONTROL 232.1 Oxygen Probes 242.1.1 Oxygen Probe Principle 252.1.2 Oxygen Probe Applications in Process Control 302.1.2.1 Steelmaking Industry 302.1.2.2 Non-Ferrous Industry 322.1.3 Oxygen Probe Applications in Molten Lead 39- iv -2.2 Process Modelling 442.2.1 Thermodynamic Models 452.2.2 Kinetics Models 482.2.3 Conclusions 49CHAPTER 3SCOPE AND OBJECTIVES 51CHAPTER 4OXYGEN PROBE DESIGN 554.1 Selection of Probe Components 564.1.1 Solid Oxide Electrolyte 564.1.2 Reference System 604.1.3 Lead Wire and Conducting Lead 674.1.4 Conclusions 684.2 Probe Development in Laboratory 704.2.1 Experimental Apparatus 704.2.2 Cell Assembly 754.2.3 Measurements in Molten Lead 84CHAPTER 5INDUSTRIAL TRIALS 945.1 Softening Process Investigations 955.1.1 Preliminary Sampling Campaign 955.1.2 Softening Chemistry and Oxygen Efficiency 103-v5.1.3 Conclusions 1065.2 Oxygen Probe Plant Tests 1105.2.1 Oxygen Potential Measurements 1105.2.2 Results and Discussion 1125.2.3 Conclusions 1185.3 Thermal Arrest Technique 120CHAPTER 6SOFTENING PROCESS MODELLING AND CONTROL STRATEGY 1276.1 Model Formulation 1286.1.1 Assumptions 1286.1.2 Computation of Equilibrium Composition 1306.1.3 Solution Modelling 1346.2 Model Validation and Calibration 1386.2.1 Solution Modelling Validation 1386.2.2 Model Calibration 1396.2.3 Analysis of the Batch Process 1446.2.4 Batch Process Operation 1516.3 Continuous Single Pass Softening Process 1536.3.1 Assumptions 1536.3.2 Process Analysis 1556.3.3 Control Strategy 164- vi -CHAPTER 7SUMMARY AND RECOMMENDATIONS 1667.1 Summary 1666.3 Recommendations 168REFERENCES 169APPENDIX 1 EXPERIMENTAL EQUIPMENT 176APPENDIX 2 PLANT DATA 181- vii -LIST OF TABLESTable 1.1 Typical assays of lead bullion and slag in the softening circuit (early1992) 15Table 2.1 Characteristics of emf cells used for oxygen measurements in moltenlead 43Table 4.1 Equilibrium oxygen pressure of various metal-metal oxide systems attemperatures between 500 and 700°C 65Table 4.2 Properties ofY203fully stabilized zirconia from Coors Ceramic Co 68Table 4.3 Assays for Cu and Cu20powders from Cerac Inc 69Table 4.4 Computed values ofO2(Pb/Pbo)’o2(cuIcu)’ and Eth 81Table 4.5 Comparison of theoretical and measured emf for various probesimmersed in oxygen saturated molten lead 88Table 4.6 Measured emf in laboratory tests and corresponding lead bullion assays. 93Table 5.1 Assays of softener bullion for vertical gradient assessment (measurement#1) 100Table 5.2 Assays of softener bullion for vertical gradient assessment (measurement#2) 101Table 5.3 Assays of softener bullion for vertical gradient assessment (measurements#3 and #4) 101Table 5.4 Assays of softener slag from opposite sides of the vessel forvertical gradient assessment (measurements #5 and #6) 102Table 5.5 Process parameters and assays of softener slag and bullion for softenerchemistry assessment (test #1, June 1994) 108Table 5.6 Process parameters and assays of softener slag and bullion for softenerchemistry assessment (test #2, October 1994) 109Table 5.7 Measured emf in softener vessel and corresponding lead bullion assays 117Table 6.1 Equilibrium reactions representing the oxygen softening process 129— viii —Table 6.2 Molar balance equations of the chemical system representing the oxygensoftening process 129Table 6.3 Expressions for the temperature dependence of the standard Gibbs freeenergies of all species 135Table 6.4 Oxygen saturation in liquid lead at different temperatures. Comparisonbetween the model computations and the relationship from Alcock andBelford (1964, ref. 42) 139Table 6.5 Batch softening model parameters (June plant data and correspondingmodel calculations) 142Table 6.6 Batch softening model parameters (October plant data and corresponding model calculations) 143Table 6.7 Batch softening model parameters 144Table 6.8 Equilibrium constants of As and Sb oxidation (Reactions (III) and (IV)) 150Table 6.9 Continuous softening model parameters 154Table 6.10 Continuous softening model parameters 155- ix -Figure 1.1Figure 1.2Figure 1.3Figure 1.4Figure 1.5Figure 1.6Figure 1.7Figure 1.8Figure 1.9Figure 1.10Figure 1.11Figure 1.12Figure 2.1Figure 2.2Figure 2.3Figure 2.4LIST OF FIGURES245891012161718192226293133Lead production flowchart (Adapted from Davey, 1980, ref. 1)Pyrometallurgical lead refining flowchart (Adapted from Davey,1980, ref. 1)Harris machine (From Leroy et at., 1970, ref. 7)Batch refining kettle (Adapted fromBowers and Johnston, 1984, ref. 9).Simplified view of a reverberatory softening furnace (Adapted fromGilges et at., 1981, ref. 10)Softening rate as a function ofbullion composition (Adapted fromGreen,1950, ref. 13)Oxygen-assisted kettle (Adapted from Blanderer, 1984, ref. 14)Cominco’s lead refining process flowchartScale drawing of Cominco’s original softener (Adapted from de Grootetal., 1989, ref. 16)Cominco’ s original lead softening circuitCominco’ s current lead softening circuitSchematic drawing of Cominco’s current softenerSchematic representation of pure zirconia and calcia stabilized zirconia.Schematic diagram of emf generation in an oxygen concentration cellusing calcia stabilized zirconia as solid oxide electrolyteSchematic drawing of a typical oxygen probe for application in moltenmetalProgress of silicon-deoxidation recorded by the electromotive forcemeasurements of an oxygen probe at 1600°C (From Turkdogan andFruehan, 1972, ref. 26)Figure 2.5 Sublance system for automatic dipping of oxygen sensors into asteelmaking converter (FromGoto, 1988, ref. 37, after Ariga and Ogawa,1977, ref. 35) 34Figure 2.6 Computing system to control the content of soluble aluminum in Al-Sikilled steel and the magnitude of the deoxidation of Si-semi-killed steel(From Goto, 1988, ref. 37, after Hiromoto et al., 1977, ref. 36) 35Figure 2.7 Improvements in deviation from target value of the content of solublealuminum in Al-killed steels before and after the use of oxygen sensors(From Goto, 1988, ref. 37, after Hiromoto et al., 1977, ref. 36) 36Figure 2.8 Feed-back control system for continuous casting of tough pitch copperbillet using an oxygen probe at the exit of the launder (From Dompasand Lockyer, 1972, ref. 39) 37Figure 2.9 Schematic illustration showing how the oxygen content in flowingmolten copper is continuously measured in the launder (From Goto,1988, ref. 37) 38Figure 2.10 Correlation between oxygen partial pressure in the slag and lead contentof the slag (From Fontainas et al., 1985, ref. 52) 41Figure 3.1 Continuous single pass softening circuit 52Figure 4.1 Schematic drawing showing the effect of temperature and oxygenpressure on the conductivity of a solid oxide electrolyte (Adapted fromPatterson, 1971, ref. 74) 58Figure 4.2 Schematic representation of (a) partial electrical conductivities, (b) ionictransference number in Y203 - doped Th02, and (c) effect of dopantconcentration on partial conductivities in Th02 - Y203 solid solutions(Adapted from Choudhary et al., 1980, ref. 75) 59Figure 4.3 Conductivity of various solid oxide electrolytes as a function of tempperature (Adapted from Choudhary et al., 1980, ref. 75) 61Figure 4.4 Electrolytic domain of various zirconia solid oxide electrolytes as afunction of temperature and oxygen partial pressure (Adapted from DeSchutter et al., 1991, ref. 57) 62Figure 4.5 Characteristics of commercial zirconia solid electrolytes (Adapted fromNippon Kagaku Togyo Co. Ltd. marketing pamphlet, 1991, ref. 83). 63Figure 4.6 Overvoltage versus current for various metal-metal oxide systems at900CC (Adapted from Worrel and Iskoe, 1973, ref. 84) 66- xi -Figure 4.7 Schematic representation of the experimental set-up for laboratory tests. 71Figure 4.8 Temperature profile along the vertical axis of the chamber with locationand length of the hot zone (± 1°C) for a temperature set point of 6 10°Cusing the outer thermocouple (S-type) 72Figure 4.9 Schematic representation of the chamber with measuring devices inposition for laboratory tests 74Figure 4.10 Schematic representation of the possible cell assembly geometries foroxygen probes 76Figure 4.11 Schematic representation of the first design of cell assembly used in theearly experiments 77Figure 4.12 Scale drawing of the steel sleeve used as housing for the probe tip(35 mm long zirconia thimble shown in position) 78Figure 4.13 Emf and temperature readings recorded during a test with a probefabricated according to the design adopted in the early tests 83Figure 4.14 Schematic representation of the improved design of cell assembly. .. 85Figure 4.15 Schematic representation of the probe and the probe holding device. 86Figure 4.16 Typical probe response upon immersion into an oxygen saturated leadmelt at 697°C ± 1°C 89Figure 4.17 Probe response to a temperature change. Initial temperature 683°C,intermediate temperature 730°C, and final temperature back to 683 °C 91Figure 4.18 Probe response to additions of antimony at temperatures around 625°C 92Figure 5.1 Close-up of softener circuit 96Figure 5.2 Dimensions of the softener unit as of June 1994 99Figure 5.3 Typical recording of emf and temperature measured in the softenervessel 113Figure 5.4 Plot of plant and laboratory emf measurements versus total impuritycontent of lead bullion 115Figure 5.5 Plot of plant and laboratory emf measurements versus As+Sb content oflead bullion 116- xii -Figure 5.6Figure 5.7Figure 5.8Figure 5.9Figure 5.10Figure 5.11Figure 6.1Figure 6.2Figure 6.3Figure 6.4Figure 6.5Figure 6.6Figure 6.7Figure 6.8Figure 6.9Figure 6.10Figure 6.11Figure 6.12119121122123124126132133145147148157158159160162163Schematic representation of a probe tip with a steel cup for sludgeprotectionThermal arrest apparatusCooling curve of pure leadCooling curve of “North Pot” bullion containing 2 wt% total impurity.Calibration curve for thermal arrest techniquePlot of As+Sb content in output bullion versus total impurity content.Simplified flow diagram of the whole algorithmSimplified flow diagram of the minimization routineAs+Sb removal during a softening cycle in batch and continuous slagremoval modesOxygen partition during a softening cycle in batch modeEvolution of the slag composition during a softening cycle in batch andcontinuous slag removal modesEffect of temperature on the slag composition during a softening cyclein batch modeEffect of impurity level and 02 injection rate on output bullion quality(615°C, 15.5 tph input bullion)Relationship between slag composition and impurity level in outputbullion (615°C)Variation of P with impurity level in output bullion (6 15°C)Sensitivity of impurity removal rate to 02 injection rate (615°C, 15.5 tphinput bullion)Effect of bullion flow rate on impurity level in output bullion (6 15°C).Effect of temperature on impurity level in output bullion (40 Nm3fh 02,15.5 tph bullion)149— xiii —LIST OF NOMENCLATUREa. Activity of species i in phase jB. Absolute mobility of species ic1 Concentration of species ie Elementary chargeE Electromotive forceE Measured electromotive forcemeasE Open circuit electromotive forceE th Theoretical electromotive forceF Faraday constantStandard Gibbs free energy of species i in phase jI,]ex Excess Gibbs free energyG* Reduced Gibbs free energyFlux of particle iK Equilibrium constantn Number of moles of species i in phase jOxygen partial pressureR Gas constantT TemperatureTbUllIOfl Bullion temperatureTsiag Slag temperaturet. Transference number of species i- xiv -t Transference number of electronselectront. Transference number of ionsionxl, Mole fraction of species i in phase jz Valency of species iGreek SymbolsActivity coefficient of species i in phase jActivity coefficient of species i at infinite dilution0 Self interaction coefficient of oxygenElectrochemical potential of species iChemical potential of species ia Electrical conductivity of species ia0 Electron free conductiona. Ionic conductioniona Positive hole conductionElectrostatic potentialSubscripts(s) solid phase(1) liquid phase(g) gas phaseOther SymbolsNm3 Normal cubic meter- xv -mm Minutetph Tonne per hourat% Atom percentemf Electromotive force(Bullion) Dissolved element X in lead bullion- xvi -ACKNOWLEDGEMENTSI would like to express my sincerest thanks to Dr. Greg Richards for his confidence in meand inspiration to me. I would like to extend my utmost gratitude to Dr. Ray Meadowcroft whograciously offered his help part way through this project and generously dealt with most of myworrying and complaining. Special thanks to my wife, Radka, for her patience, support, and helpall along this research work. I am also very grateful to my daughter, Claire, who convinced me tocomplete my thesis, and successfully motivated me to do so.I would also like to acknowledge the assistance of Cominco personnel, and in particular,Gerry Toop, Tom De Groot, Mark Plamondon, Wayne Teague and Dan Brew without whom theplant trials would not have been possible.The financial assistance of Cominco Ltd., the B.C. Science Council, and the Natural Sciencesand Engineering Research Council is gratefully acknowledged.Thanks are also due to those members of the Department ofMetals andMaterials Engineeringwho helped in this endeavor, and in particular to Peter Musil for his invaluable contributions to themanufacturing and maintenance of my experimental apparatus. Special thanks to my office mates,Cornelius Muojekwu and Sunil Kumar, and to Bernardo Hernandez Morales for their friendship,support, and helpful discussions.I’ aimerais aussi remercier très sincèrement mes parents pour leur aide, leur soutien, et leursencouragements répétés durant toutes mes longues années d’étude.- xvii -CHAPTER 1INTRODUCTION: LEAD REFININGIn the course of the production of lead fromore to refined product, each stage can be consideredas a refining operation since impurities are separated from the valuable product in each of thesesteps, as shown in Figure 1.1. However, in this figure, Davey’1restricts the term “refining’ to theoperations carried out downstream of the smelting stage. Lead refining usually involves a numberof different steps depending on the type and amount of impurity elements present in the moltenmetal. Copper, arsenic, antimony, sulphur, tin, bismuth, silver and gold are the main impuritiespresent in crude lead. Most of the copper is removed from the lead by cooling the molten bullionto near its freezing point and precipitating the copper and the copper compounds in a dross or matte.Since there is no practical alternative to the pyrometallurgical process for copper removal, leadbullion is always thermally drossed as a first refining step2. Following copper drossing, twoalternative routes are possible: either electrolytic or pyrometallurgical refining. Electrolytic refiningis the only process that can adequately produce refined lead with a low level of bismuth (less than10 ppm). Due to its high capital cost, this route is less attractive and is adopted only where high-puritymetal (low bismuth level) is required by the market21 or where there is cheap power. In most ofthe lead refineries around the world, operations are entirely pyrometallurgical, with a few exceptionssuch as Cominco Ltd. (British Columbia)31,Monteponi & Montevecchio Co. (Sardinia)41,Cerrode Pasco Co. (Peru)’5, or Shenyang Smelter (China)61 where electrolytic techniques are used.Pyrometallurgical refining, which involves reacting the impurity elements with sulphur, oxygen,zinc, calcium, etc., in order to produce various slags or drosses, is reasonably selective in impurity—1—1 INTRODUCTION: LEAD REFIMNGOre in situMining Country rockOreGangue, andMilling other metalconcentratesLead concentrateRoasting Sulfur, CadmiumSinterResidues Smelting SlagCrude leadRefining By-product metalsMarket lead99.99+ % PureFigure 1.1 - Lead production flowchart (Adapted from Davey, 1980, ref. 1).-2-1.1 Lead Softening Practicesremoval and the refined metal can meet the majority of the market requirements21. A typicalpyrometallurgical route for lead refining is shown in Figure 1.2. In each successive step, the impuritylevel of one or more elements is decreased to trace level. Of particular interest for this project, thesoftening step corresponds to the removal of the hardening impurity elements, namely antimony,arsenic and tin, by selective oxidation due to differences in the free energies of formation of theoxides of these elements and lead. Current lead softening practices are reviewed and discussed inthe following section. Details of Cominco’ s softening process are then presented to provide theindustrial background of the project.1.1 Lead Softening PracticesWhen lead bullion from a smelter is refined via an electrolytic route, antimony and arsenicare commonly added to the bullion prior to anode casting in order to achieve the desired hardnesslevel required in the refinery. Removal of antimony, arsenic and tin, or softening, is usually performed in pyrometallurgical refineries. The preferential oxidation of the hardening elements canbe accomplished at either low or high temperature.In the low temperature alternative, softening is carried out in a kettle at about 450°C by addinga strong oxidizing agent, such as sodium nitrate, and absorbing the resulting sodium antimonate,arsenate and stannate into a caustic soda melt. This softening variant is commonly called the Harrisprocess”7’81.AttheMetallurgie Hoboken refinery71,lead softening is performed in Harris machinesas shown in Figure 1.3. A reaction cylinder rests above a holding pot with its lower part immersedin liquid lead. The softening cycle starts by pouring molten caustic soda into the reaction cylinder.Then lead is pumped from the pot and injected into the cylinder where it migrates through the-3-1.1 Lead Softening PracticesCrude leadCopper Drossing______________SofteningDesilverisingDezincingDebismuthisingFinal RefiningMarket lead99.99+ % PbDross [Cu]Slag [Sb, As, Sn]Crust [Zn, Ag]Metallic zincDross [Bi]— Caustic drossFigure 1.2 - Pyrometallurgical lead refining flowchart (Adapted from Davey, 1980, ref. 1).-4-1.1 Lead Softening PracticesFigure 1.3 - Harris machine (From Leroy et al., 1970, ref. 7).4CAUSTIC SLAG—--—GRANULATINGSPOUT-5-1.1 Lead Softening Practicescaustic slag before returning to the pot through the valve located at the bottom of the cylinder. Eachtime lead flows through the slag, dissolved tin, arsenic or antimony together with some lead reactwith air or sodium nitrate and collect in the caustic slag layer. The reaction involving antimony,for example, can be described as follows:2 j (Bullion) + 2 NaNO3 + 4 NaOH <- 2Na3SbO4 + N2 + 2H20The Harris process oxidizes very little lead and can be operated selectively for arsenic and tin withrespect to antimony21.This selective softening can be achieved by means of several Harris machinesin series, the first machines removing arsenic and tin with traces of antimony, provided the sodiumnitrate is added in the appropriate proportion, the last machines removing the remaining antimony.The caustic soda, which can be regenerated by evaporation, not only participates in the oxidationreactions but also acts as a suspension medium for the reactants, facilitating their separation fromthe lead bullion111.The high temperature alternative involves oxidation by injecting air or oxygen into the moltenlead at 600°C to 750°C to form a liquid slag containing most of the hardening impurities19131.Thebasic reactions involved are the following:1/2 °2 (g) (Bullion)Pb (I) + (Bullion) E4 PbO (Slag)](Bullion) + 3/2 Q(Bullion) <-4 Sb015 (Slag)A(Bullion) + 3/2 (Bullion) <—* As015 (Slag)-6-1.1 Lead Softening PracticesThis softening variant can be performed in batch or continuous mode. Batch softening is usuallycarried out in kettles of up to 350-tonne capacity, with most medium sized refineries (30,000tons/year) having kettles of about 50 to 1 00-tonne capacity with a diameter of 2 to 3 metres (seeFigure 1.4). Reverberatory furnaces (see Figure 1.5) are usually used for continuous softening inrefineries with large bullion throughput (over 100,000 tons/year).The BHAS refinery at Port Pine, Australia, is well known for its high performances as a resultof continuous efforts devoted to research’ Their softening process is the most efficient processin the world and serves as a reference and a source of inspiration for many refineries, includingCominco’ s. For that reason, it seems legitimate to provide more details of this particular operation.Softening at Port Pine is carried out continuously in a reverberatory furnace to reduce the antimonycontent of the bullion from about 0.8 % to about 0.03 %, and the arsenic from about 0.2% to lessthan 0.001%. The oxidation of the hardening impurities is enhanced by the strong agitation of thebath accompanying the air injection. The temperature of crude lead bullion is about 400°C to 450°Cwhile the temperature of the softener bath is in the range 700°C to 760°C. The oxidation reactionscontribute largely to the heat requirements of the process. The rate of input bullion into the furnaceand the amount of air injected are adjusted in such a manner that the impurity level in the bathremains in the composition range corresponding to the shaded area in Figure 1.6. This is to takeadvantage of the effect of varying antimony concentration on the rate of oxidation. As Green’31points out, only a continuous process can gain the greatest advantage of the oxidation rate phenomenon by oxidizing the antimony as fast as it is supplied. The slag produced has roughly thefollowing composition: 75% Pb, 12% Sb and 2% As. Removing the last traces of antimony isusually performed in a final refining step using caustic soda.-7-1.1 Lead Softening PracticesFigure 1.4 - Batch refining kettle (Adapted from Bowers and Johnston, 1984, ref. 9).VentilationExhaustFuel&air-8-1.1 Lead Softening PracticesFigure 1.5 - Simplified view of a reverberatory1981, ref. 10).softening furnace (Adapted from Gilges et al.,Air lances/GasFeedI Iim 5-ton pot-9-1.1 Lead Softening Practicesci)I0Vx0Per cent antimonyFigure 1.6 - Softening rate as a function of bullion composition (Adapted from Green, 1950,ref. 13).200150100500I I1tv_%11: I0 .02 .04 .06 .08 0.1 0.2- 10 -1.1 Lead Softening PracticesA third alternative combines the use of caustic soda and high temperature. It is usually carriedout in a stirred kettle at a temperature of about 630°C by addition of caustic soda into molten leadto form a dross which can be either dry or sticky. Air injection assists the process and promoteshigh softening rates. Blanderer1141,who patented a softening method using caustic soda in a kettle,claims to be able to achieve efficiencies close to those obtained in Port Pine by injecting additionalpure oxygen and adjusting the stirrer speed. A schematic drawing of Blanderer’ s oxygen-assistedkettle is given in Figure 1.7. With good control, i.e. limited arsenic vaporization, appropriateproportion of caustic soda, Quigley and Happ’51 report that selective oxidation of arsenic withrespect to antimony can be achieved. The arsenic removal reaction can be described as follows:(Bullion) + 5 NaOH + 5/2 02 —* Na3AsO4.2 NaOH + 3/2H20The caustic soda requirements can be calculated from this reaction. Quigley and Happ151 alsoobserve that the selective extraction of arsenic occurs at the beginning of the process when thecaustic slag is liquid. Once the slag is saturated withNa3AsO4.2NaOH, its viscosity increases andthe selectivity decreases.The choice between batch and continuous softening is site dependent since both haveadvantages and drawbacks. Batch softening requires that large holding pots or kettles are held athigh temperature for long periods of time, which is costly in terms of energy requirements. It has,however, the advantage of being able to treat bullion of °abnormal” composition, such as highimpurity feed bullion, by simply adjusting the operating conditions for the duration of the “abnormality”. The number of kettles required depends on the throughput of the refinery. Large- 11 -1.1 Lead Softening PracticesRing tubeRing tubeStirrerTop ViewFigure 1.7 - Oxygen-assisted kettle (Adapted from Blanderer, 1984, ref. 14).Side ViewKettleStirrerAir- 12 -1.2 Cominco’s Lead Softening Practicerefineries may employ up to 5 kettles. Continuous softening on the other hand can be performedfor large throughputs using smaller vessels at very high extraction rates. Since the process iscontrolled by the feed rate, the bullion and slag composition, the temperature and the degree ofagitation, any ‘abnormal” bullion composition cannot be easily handled. When it comes to hygiene,it is difficult to assess whether the batch or continuous alternative is better since hygiene dependson temperature (higher temperature favors large amounts of flue gases and dusts), vessel size (largervessels produce larger amounts of by-products), and on the type of by-product (manual drossskimming or automatic slag tapping).1.2 Cominco’s Lead Softening PracticeAt the Trail smelter of Cominco, lead is produced in blast furnaces from sinter made from amixed feed of concentrates and zinc plant residues. In the subsequent electrolytic refining by theBetts process, certain impurities contained in the anodes, such as copper, antimony, arsenic, bismuth,silver and gold, remain on the corroding anode surface as an adhering layer of porous slime, whilezinc, cadmium, nickel and cobalt dissolve into the electrolyte. High current efficiencies and theproduction of high-purity refined lead depend on the stability of the slime31. In order to ensure thatthe slime remains on the anode surface, specific levels of antimony and arsenic in the cast anodesare required. The optimum content of combined antimony and arsenic is in the range 1.5 to 2.0%.Historically, antimony and arsenic were supplemented downstream of the blast furnaces to maintainan appropriate composition of the anodes. Mainly due to increasing levels of antimony and arsenicin the Sullivan concentrate in the mid 1 980s and the purchase of larger quantities of silver bearingconcentrate inherently rich in antimony and arsenic, the content of antimony and arsenic has- 13 -1.2 Cominco’s Lead Softening Practiceexceeded the limits required for the Betts process. Removing the excess antimony and arsenic inthe bullion became necessary. After reviewing the various options currently available for leadsoftening, a choice was made at Cominco to undertake partial softening with pure oxygen.The original partial softening vessel, commissioned in 1986 and installed downstream of theContinuous Drossing Furnace (C.D.F.), as shown in Figure 1.8, was designed to operate semicontinuously’61.It consists of a 20-tonne capacity vessel built from a blast furnace settler with ametal depth of about 0.8 meters (see Figure 1.9). It is serviced with four to six oxygen lances eachfittedwith an oxygen flow meter and carrying up to 20 normal cubic meters per hour of 98% oxygen.The impurities, mainly antimony and arsenic, together with some lead are oxidized and form a slagwhich is kept fluid at about 700°C by means of a natural gas burner. A schematic representationof the original softening circuit is shown in Figure 1.10. Copper is removed from the crude bullionin the C.D.F. Two 200-tonne capacity holding pots, called “North” and “South” pots due to theirgeographical positions in the plant, allow mixing of crude bullion from the C.D.F. with partiallysoftened lead from the softener. In the summer of 1991, the circuit was modified in an attempt toincrease the softening capacity and increase control on the anode composition. The new lay-out isgiven in Figure 1.11. The desired target of 1.5 to 2.0% combined antimony and arsenic in theanodes is achieved by intermittent pumping between the two pots. During the process, the bulliontemperature in the softener increases due to the exothermic character of the oxidation reactions. Apump is interlocked with a thermocouple in order to maintain the temperature at the bottom of thevessel between 6 15°C and 625°C. At the upper limit of the temperature cycle, the pump is turnedon and lead bullion at 450°C is drawn from the “North” pot, while partially softened bullion andslag are discharged, the bullion returning to the same pot. When the temperature drops to the lowerlimit, the pump is turned off, the bullion and slag discharge ceases, and softening proceeds. The- 14 -1.2 Cominco’s Lead Softening Practicetime duration between the two pumping actions corresponds to a softening cycle and lasts about20 minutes. The slag is produced at a rate of about 15.8 metric tons per day, corresponding to about220 kg per cycle. Typical assays of the various bullions and the slag are shown in Table 1.1.Table 1.1 - Typical assays of lead bullion and slag in the softening circuit (early 1992). Leadconstitutes the balance together with minor amounts of tin and silver.Location Sb As Bi Cu(wt%) (wt%) (wt%) (wt%)C.D.F. Bullion 3.00 0.46 0.10 0.19SouthPotBullion 1.37 0.18 0.10 0.14North Pot Bullion 1.19 0.14 0.10 0.14SoftenerBullion 0.76 0.07 0.11 0.14Softener Slag 26.8 5.4 0.01 0.10- 15 -1.2 Cominco’s Lead Softening PracticeCrude leadCopper DrossingSofteningElectrorefiningDross [Cu]Slag [Sb, As, Sn]Slime [Sb, As, Cu, Bi, Ag, Au]Final RefiningMarket lead99.99+ % PureFigure 1.8 - Cominco’s lead refining process flowchart.- 16 -1.2 Cominco’s Lead Softening PracticeOxygen lancesNatural gasburner1 meterFeed bullionProduct bullionFigure 1.9 - Scale drawing of Cominco’s original softener (Adapted from de Groot et al., 1989,ref. 16).VentilationSlagU- 17 -1.2 Cominco’s Lead Softening PracticeOxygen SlagCrude leadDustCu-matteFigure 1.10 - Cominco’s original lead softening circuit.ContinuousCopper DrossingAnode casting- 18 -1.2 Cominco’s Lead Softening PracticeOxygenCrude leadSlagDustCu-matteFigure 1.11 - Cominco’s current lead softening circuit.ContinuousCopper DrossingAnode casting19 -1.2 Cominco’s Lead Softening PracticeTo accommodate the treatment of increasing quantities of silver bearing concentrate, a largersoftener unit was commissioned and installed in 1992. The new unit consists of a circular 100-tonnecapacity vessel serviced with up to eight oxygen lances (see Figure 1.12). A steel pipe, with adiameter of about 25 cm, is submerged into the molten lead and acts as the outlet for the softenedbullion. A thermocouple, immersed one metre inside the pipe, serves as temperature controller tomaintain the bullion temperature between 615°C and 630°C. Thus, the softener operates in a similarcyclic manner as the original unit. Two natural gas burners keep the slag fluid. As for the softeningcircuit, it remains unchanged with the exception that continuous pumping, rather than intermittentpumping, between the two holding pots was initiated to improve control over the anode composition.Since the plant-scale softener was commissioned in 1989, Cominco lead furnaces staff havehad to continuously modify and adapt the process to better respond to the changing operatingconditions in the drossing plant: long term increase of bullion hardness due to the treatment ofincreasing amounts of silver concentrate, as well as daily fluctuations of incoming bullion composition due to changing operating conditions in the smelter. Holding a composition target is crucialin ensuring proper production in the refinery. Whatever the improvements achieved in theconfiguration of the process circuit or in the design of softening vessels, added revenues from silverrecovery with good control of the production requires an ability to treat harder bullion withappropriate control of the output bullion composition from the softening stage. On-line compositioncontrol presents the best potential to achieve this goal.Such a control requires an understanding of process fundamentals, and in particular processchemistry. Although much work as been devoted to the softener, no quantitative analysis of theprocess has been done. A number of fundamental questions are unanswered. For example, it is- 20 -1.2 Cominco’s Lead Softening Practicenot known whether composition and temperature gradients are present in the vessel. The oxygenefficiency has only been assessed qualitatively on the basis of the number of bubbles breaking thesurface of the bath. The influence of bullion composition, and in particular high versus low As+Sblevel and the ratio As/Sb, has been assessed based on an improper analysis of oxidation rate suchas represented on Figure 1.6. Lower levels ofAs+Sb are believed to favor higher process efficiency.No characterization of the temperature effect on slag quality during normal operation, as opposedto during start-up, has been attempted. Moreover, only limited data are available in the literatureon the chemical system Pb-Sb-As-O.The following Chapter presents, on one hand, a review on sensors used in pyrometallurgyfor process control purposes, and on the other hand, a review on process modelling.- 21 -1.2 Cominco’s Lead Softening PracticeFigure 1.12 - Schematic drawing of Cominco’ s current softener.Fume hoodInput bullion EEEE IOutput bullionThermocoupleOxygen lancev Slag outletBricksSteel pipe- 22 -CHAPTER 2LITERATURE REVIEW: OXYGEN PROBES AND PROCESS CONTROLA knowledge of the behavior of solutes such as oxygen, nitrogen or sulphur in metallicsolutions is important for numerous chemical and metallurgical processes. In particular, thebehavior of oxygen in molten metals governs industrial refining practices such as steel deoxidation,copper deoxidation or lead softening. In these refining operations, temperature and oxygen activityare the most important thermodynamic parameters. Consequently, the control of a refining processrequires the monitoring of these two parameters. Temperature measurement has long been acommon practice in ferrous and non-ferrous pyrometallurgy with the use of thermocouples. Onthe other hand, on-line oxygen determination without the need for sampling hot metals was notpossible until the development of galvanic cells using solid oxide electrolytes. Since the pioneeringwork of Kiukkola and Wagner’71,tremendous progress has been achieved in improving the characteristics of solid oxide electrolytes and designing oxygen probes for rapid and accurate measurements of oxygen in molten metals.A literature survey has been carried out covering the multiple aspects of oxygen probes:principle, design, and applications, and is presented in the first section. Even though the literaturewas extensively analyzed to elucidate the various design concepts, it is believed that it is best suitedto present these concepts in Chapter 4, which specifically deals with the design, construction andtesting of an oxygen probe for application in lead softening. The scope of the review in this Chapteris thus limited to the principle and the applications of oxygen probes.- 23 -2.1 Oxygen ProbesThe chemistry and kinetics of numerous pyrometallurgical processes were poorly understoodat the time the first furnaces were built. Improvements in process efficiency were usually made bygradually implementing modifications in a process based on plant observations and experience. Inthis manner, new practices evolved, frequently leading to new furnace designs. The lead softeningprocess in use at BI-IAS in Port Pine, Australia, is a case in point. Over several years, a number offurnace designs were constructed and tested in the production circuit at very high costs until asatisfactory furnace was obtained’1’31.With the development of computers, attempts at quantitatively analyzing pyrometallurgical processes in terms of chemistry and kinetics were made.Starting with simple assumptions, mathematical models of various processes were developed andgradually expanded as a better understanding ofprocess fundamentals was gained. Since the adventof personal computers in the 1 980s, mathematical modelling has become an integral part of theanalysis of metallurgical processes. Richards et al.’8 described modelling as a very powerful toolwhen combined with industrial measurements in the elucidation ofprocess kinetics, troubleshootingand optimization.A review of the mathematical modelling of pyrometallurgical processes is presented in thesecond section of this Chapter. The objective is not to provide an exhaustive review ofmathematicalmodels available but rather to discuss, illustrated with some examples, the approach and purposesof process modelling.2.1 Oxygen ProbesThis section is composed of three main parts. First, the operating principle of an oxygenprobe is briefly introduced. The most prominent applications of oxygen probes in process control- 24 -2.1 Oxygen Probesof fenous and non-ferrous metallurgy are then presented to demonstrate that oxygen probes havebecome an indispensable tool in pyrometallurgy. In the last section, the emphasis is set ontoreviewing the state-of-the-art of the oxygen probe applications in molten lead environments.2.1.1 Oxygen Probe PrincipleAn oxygen probe is an electrochemical cell with a solid oxide electrolyte that operates as anoxygen concentration cell. Consider the generalized electrochemical cell written in concise notationas followsPt, 02 (P 02) MO I 02 (P o) Pt (A)where MO is a solid oxide electrolyte in which oxygen ions, 02, are mobile. Calcia stabilizedzirconia, Zr02 - CaO, is a typical example of solid oxide electrolyte. The addition of calcia intopure zirconia generates oxygen ion vacancies within the lattice which compensate for the differenceof valency between zirconium and calcium ions (see Figure 2.1). These vacancies facilitate thetransport of oxygen ions through the electrolyte. Further details on the conduction properties ofsolid oxide electrolytes are given in Chapter 4.When there exists a gradient of oxygen partial pressure between P2and ‘ 02’ cell (A)generates an electromotive force. The electrochemical phenomenon involved can be explainedfollowing the derivation of Jacob and Mathews’9. Due to the difference of oxygen chemicalpotential between the two sides of the cell, there exists a flux of charged particles in the solid- 25 -2.1 Oxygen ProbesFigure 2.1 - Schematic representation of pure zirconia and calcia stabilized zirconia.Pure ZirconiaZr02Calcia Stabilized ZirconiaZr02- l5mol%CaO0000000000000000000000000000000000000 0OD0000000• 00000D0000• 0 00000000000 0000000DOO0 0 • 0High temperature cubic structure Stabilized cubic structure0 Oxygen ions (valency: - 2)0 Zirconium ions (valency: + 4)• Calcium ions (valency: + 2)Oxygen vacancies- 26 -2.1 Oxygen Probeselectrolyte along the gradient. The particle flux for diffusion of a mobile species i in an activitygradient and electrical field is given byj1 = -c1B grad (2.1)where c, B. , and r are the concentration, absolute mobility and electrochemical potential of themobile species i, respectively. If no current is drawn from the solid electrolyte, under steady stateconditions, maintaining the charge neutrality at every point requires thatzj = 0 (2.2)where z is the valency of the mobile species i. Combining the Equations (2.1) and (2.2), it follows- c1zB gradr1 = 0 (2.3)The absolute mobility of species i can be expressed asB1 = oI(z12ce) (2.4)where and e are the electrical conductivity and elementary charge respectively, and the electrochemical potential can be written ast. + z.FØ (2.5)where p and 0 are the chemical potential of species i, and the electrostatic potential, respectively.Substituting Equation (2.4) and (2.5) into (2.3) and rearranging yields- 27 -2.1 Oxygen Probesgrad = - , (t1/z) grad ji (2.6)where t, = a I a is the transference number of species i. For an oxygen concentrationcell using a solid oxide electrolyte, the mobile species are mainly oxygen ions and electrons. It isseen, from Equation (2.6), that the tendency of the mobile species to diffuse along the chemicalpotential gradient is compensated by an electric field that opposes ionic motion. As a consequence,in response to an oxygen potential gradient, an oxygen concentration cell generates an open-circuitemf,E0. This emf is an integral quantity measured between the left and right-hand side of thecell, and proportional to the difference of oxygen potential as expressed by the following equationE0 = pr = -! f , (t1Iz) d j.t, (2.7)where rand 1 stand for right and left. This electrochemical phenomenon is schematically illustratedin Figure 2.2. In electrochemical probes, predominantly ionic conducting solid electrolytes areused, i.e. only oxygen ion is mobile and its transference number is equal to unity, then the emf ofcell (A) is given by the well-known Nernst equation as followsRT o2E0 = in —— (2.8)The above oxygen pressures P and P can be exerted by pure oxygen gas, by a gas mixture,by a metal/metal oxide mixture or by oxygen dissolved in a molten metal. A measure of theopen-circuit emf,E0 generated by an oxygen concentration cell in which the oxygen potentialofone electrode is known (reference electrode), will provide a determination of the unknown oxygen- 28 -2.1 Oxygen ProbesEmfVacancies/?QD.0000000GDooou0000’ GD P,,cooD co co200O000000 2GDcoOQOQQDQ GDGDoxygen ions- 0 + 2 e — 0 + 2 e2 2diffusion 2 2Figure 2.2 - Schematic diagram of emf generation in an oxygen concentration cell using calciastabilized zirconia as solid oxide electrolyte.- 29 -2.1 Oxygen Probespotential of the other electrode. In this manner, the cell operates as an oxygen probe which can beused to measure the oxygen content of a molten metal. Oxygen probes based on this principle havebeen used in various ferrous and non-ferrous pyrometallurgical processes where oxygen control isrequired. These processes include steelmaking, copper smelting, copper refining, and nickelsmelting. A schematic drawing of a typical oxygen probe is given in Figure 2.3. The next sectionpresents some of the most significant applications of oxygen probes in process control.2.1.2 Oxygen Probe Applications in Process ControlOxygen probes have increasinglybeen used in pyrometallurgy to either investigate the processmechanisms or even to control certain operations. This section presents some of the main applications of these probes in process control. The first part covers the steelmaking industry while thesecond part deals with the non-feffous industry. Steelmaking IndustryInvestigations of the potential application of oxygen probes in steelmaking operations havebeen reported as early as the mid-sixties by Fitterer and co-workers20221.The first work on in-situapplication of oxygen probes was published by U.S. Steel researchers in the early seventies2326,followed by numerous Japanese and German papers in the seventies and eighties27361. Resultingfrom tremendous efforts devoted to designing oxygen probes for process control purposes, disposable devices for single readings were produced and commercia1ized24’5.Such devices havecontributed significantly to the improvements made in the steelmaking industry over the last twentyyears. These probes have made possible better monitoring and control ofmetal processing practices- 30 -2.1 Oxygen ProbesConducting wireCementMetal housingCementMolten metalSolid electrolyteReference systemEmfFigure 2.3 - Schematic drawing of a typical oxygen probe for application in molten metal.- 312.1 Oxygen Probessuch as decarburization, dephosphorization or deoxidation, as well as slag formation and composition. The progress of silicon-deoxidation recorded by oxygen probes is shown in Figure 2.4. Themeasurement of the dissolved oxygen can be carried out by manually dipping the probe into moltensteel. Performing this measurement just after refining in a converter provides a measure of thedissolved oxygen remaining in the liquid steel. This information is extremely valuable in decidingwhether or not to proceed with further processing. When the use of oxygen probes became standardin Japan, more sophisticated systems such as the sublance system shown in Figure 2.5 wereincorporated into steelmaking practices. Figure 2.6 shows a computing system to control the contentof soluble aluminum in Al-Si-killed steel. The improvements in the deviation from target valuesfor soluble aluminum in Al-killed steels before and after the use of oxygen probes are shown inFigure Non-Ferrous IndustryThe most successful oxygen probe application to date in non-ferrous metallurgy is in copperrefining38411.Dompas and Lockyer [39] report on how copper deoxidation prior to continuous castingat Metallurgie Hoboken (Belgium) is monitored by oxygen probe measurements. Oxygen potentialmeasurements in flowing copper in the launder is used to control deoxidizer additions andcombustion air to the launder heater (see Figures 2.8 and 2.9).The copper industry also provides examples of the application of oxygen probes for investigating the mechanism involved in a process. Kemori et al. [42] measured the oxygen potential ofthe matte in a copper flash smelting furnace using disposable oxygen probes. The oxygen potential- 32 -2.1 Oxygen Probes300O.054%Si 800I I %Si ADDED 600( ) % Si RESIDUAL IN“S SOLUTION 400— ‘. 10.135%Si 300(0.025)”. 2002 o0.305%Si 2* ISV 5—,(0.I5i... 100100 — 0282% Si(045)50(0.73)I I I I I0 20 40 60 80 100 120 140 160TUE,MINUTESFigure 2.4 - Progress of silicon-deoxidation recorded by the electromotive force measurements ofan oxygen probe at l6OOC (From Turkdogan and Fruehan, 1972, ref. 26).- 33 -2.1 Oxygen ProbesSublanceOxygenSensor—Main LanceFigure 2.5 - Sublance system for automatic dipping of oxygen sensors into a steelmaking converter(From Goto, 1988, ref. 37, after Ariga and Ogawa, 1977, ref. 35).CraneWinch forSublanceSublQnce GuideSteelmcikingConverter- 34 -2.1 Oxygen ProbesFigure 2.6 - Computing system to control the content of soluble aluminum in Al-Si killed steeland the magnitude of the deoxidation of Si-semi-killed steel (From Goto, 1988, ref.37, after Hiromoto et at., 1977, ref. 36).Display board(Al or Si tobe added)Consoleriter®Operatingdata- 35 -2.1 Oxygen Probes>0C4,CSoIAIJ (x103I.)Figure 2.7 - Improvements in deviation from target value of the content of soluble aluminum inAl-killed steels before and after the use of oxygen sensors (From Goto, 1988, ref. 37,after Hiromoto et al., 1977, ref. 36).10>sUBefore use of oxygen sensor n=72X:&517,4910539363330272421 1815 129 6 30 3 6 9 121518 21Z 2730333639• After use ol oxygen sensor X =1.2 n41j —39•36-33•3027•24-21•1S-15-12•9 6-3 0 3 6 9 12151821242730333639- 36 -2.1 Oxygen ProbesFigure 2.8 - Feed-back control system for continuous casting of tough pitch copper billet using anoxygen probe at the exit of the launder (From Dompas and Lockyer, 1972, ref. 39).OxygenSensoroIJs- 37 -2.1 Oxygen ProbesFigure 2.9 - Schematic illustration showing how the oxygen content in flowing molten copper iscontinuously measured in the launder (From Goto, 1988, ref. 37).e terminoloxygensensorair supply recorderflowingcopper- 38 -2.1 Oxygen Probesof the slag in the settler below the shaft, at the tap holes and in the electric furnace downstream ofthe flash furnace were also measured. Even though incomplete, their study suggests that thepyrometallurgicai oxidation of the copper concentrate is completed in the shaft or in the settler.On-line oxygen monitoring for liquid sodium is another application field for oxygen probes.Liquid sodium is used as coolant in nuclear reactors. Dissolved oxygen in liquid sodium has to bemonitored to minimize corrosion of the stainless steel piping as well as for early detection of oxygeningress into sodium circuit to prevent a fire hazard431. Due to the low temperature of application,it is customary for a probe to take up to a week before recording a stable emf. Nevertheless, anysubsequent change in dissolved oxygen is quickly detected. Such probes can then become part ofan on-line warning system.2.1.3 Oxygen Probe Applications in Liquid LeadAfter demonstrating how the use of oxygen probes dramatically improved the productionquality of some metallurgical processes, it seems important to investigate the past uses and applications of oxygen probes in such a corrosive environment as molten lead. In the course of thisresearch, a literature survey on electrochemical measurements in lead has been carried out. In anattempt to gather as much information as possible, thermodynamic studies involving electromotiveforce measurements in lead as well as research oriented towards designing oxygen probes for leadprocesses were considered.Electrochemical techniques were successfully used to determined a number of thermodynamic properties in lead systems. Alcock and Be1ford1determined the free energy of formation- 39 -2.1 Oxygen Probesof lead oxide, the partial free energy of solution and the solubility of oxygen in lead. Charette andF1engas451measured the free energy of formation of various oxide including PbO. Szwarc et al. [46]investigated the solubility and diffusivity of oxygen in liquid lead. Bandyopadhyay and Ray471evaluated the kinetics of dissolution of oxygen in lead. Jacob and Jeffes48, and Otsuka andKozuka49’501determined the activity relations of oxygen in lead and lead alloys. Taskinen511carriedout a comprehensive study on oxygen-metal interactions in dilute molten lead alloys. Temperaturedependences of the oxygen self interaction parameter as well as first order interaction parametersof Ag, Au, Bi, Cu, In, Ni, Sb, Sn, and Te were determined.Contrary to experience in the steel and copper industry, very little attempt has been made touse oxygen probes in lead pyrometallurgy. Fontainas et aL52 report the application of disposableoxygen probes to control the reduction level of lead smelter slags. They claim to reproduciblyachieve an accuracy of about ± 10% in the determination of lead content of the slag. The correlationbetween logP02 and wt% Pb in the slag from their industrial measurements in the blast furnaceis shown in Figure 2.10. Due to frequent probe failures and occasional out of range readings, theauthors stipulate that two probes are routinely necessary for one measurement. Sometimes a thirdprobe is required when the two readings differ by more than 10 mV (20-25% relative error on Pbcontent). When three probes are used, the cost of one measurement is said to be comparable withthe cost of a routine X-ray fluorescence slag analysis.Continuous oxygen monitoring in liquid lead or liquid lead alloys for process control in thelead smelting or refining industry has not been published in the literature to date. The only reporteduse of a long-life oxygen probe in liquid lead comes from the nuclear industry, and more preciselyfrom the Center for Nuclear Energy5358 in Mol, Belgium. Liquid metal, either pure lithium or the- 40 -2.1 Oxygen Probes11,311.211,1• ..110 ••.• .••• \. •109-.2’.\•e•— 4 I I I 1 11‘\\\%%\%\%\0,7 0,8 0,9 1,0 1,1 1,2 1,3 1,4 1.3 2 2.5 3/. Pb (stag)Figure 2.10 - Correlation between oxygen partial pressure in the slag and lead content of the slag(From Fontainas et al., 1985, ref. 52).- 41 -2.1 Oxygen Probeseutectic Pb-l7Li, has been considered in a number of fusion reactor designs. Liquid metal can beused as coolant for the plasma facing components, or as a protection for the structural materials ofa reactor against highly energetic particles and high heat fluxes581. Using Pb- l7Li as a coolantraises similar problems to the use of liquid sodium: the non-metallic impurities of the coolant, andin particular dissolved oxygen, must be controlled for safety reasons. An oxygen probe wasdeveloped to continuously monitor the oxygen content of the liquid Pb-i 7Li alloy in the temperaturerange 350°C and 500°C. More than 100 prototype probes were tested amongst which 55% werereliable and could be calibrated. Dekeyser and De Schutter58report that the majority of the failuresare due to cracks in the solid electrolyte. Cracks can be caused by thermal shock especially at theinitial immersion into the liquid alloy. When subject to temperature changes, the probe respondssluggishly, especially upon heating. This behaviour is explained by Dekeyser and De Schutter581by assuming slow kinetics of the system Pb-Li and oxygen, i.e. slow restoration ofoxygen saturationafter a temperature increase. On the other hand, the probe responds quickly to a decrease oftemperature (no effect of slow kinetics of the system), but a stable emf is not achieved until after20 to 50 hours. This behaviour is acceptable when monitoring Pb-Li over long periods of time, i.e.months or years, to detect oxygen pick-up, but would not be acceptable for on-line monitoring ofa lead bullion whose composition varies on a daily or even hourly basis.Kozuka591 reports the use of an oxygen probe in the forehearth of a lead blast furnace inSumiko ISP smelter, but cannot confirm that the oxygen concentrations measured correspond tothe oxygen potentials in the furnace.The characteristics of the various oxygen electrochemical cells used in the above works aresummarized in Table 2.1. The results of this survey are very encouraging for two major reasons.- 42 -2.1 Oxygen ProbesTable 2.1 - Characteristics of emf cells used for oxygen measurements in molten lead.Reference Conducting TemperatureSolid Electrolyte Electrode Lead Range Applications Ref.System (Wire) (°C)Zr02-14mol% MgOor Ni-NiO Jr 440-800 Thermodynamics 44Zr02-15 mol% Th02Zr02-1 mol% CaO Ni-NiO Pt 500-1100 Thermodynamics 45Zr02-CaO Air Chromel 740-1080 Thermodynamics 46(Pt)Zr02-CaO Ni-NiO Steel 750 Thermodynamics 47Zr02-7mol% CaO Cr-cermetor Ni-NiO 500-1200 Thermodynamics 48Zr02-4.5 mol%Y203 (Pt)Zr02-10.4 mol% CaO Air Jr 800-1050 Thermodynamics 49-50(Pt)Zr02-7mol% CaO Air Cr-cermet 760-1000 Thermodynamics 51(Pt)Ni-NiO FeZr02-CaO or or 1150-1350 Probe (lead slags) 52Mo-MoO2 PtZr02-Y03 In-1n203 Steel 400-600 Probe 53-58(liquid Pb- l7Li)Zr02-MgO Fe-FeO Steel 1000-1100 Probe (crude lead) 59Zr02-MgO Cr-Cr203 Mo 1000-1100 Probe (lead slags) 59- 43 -2.2 Process ModellingFirst of all, application of electrochemistry in molten lead has been performed in the past withsuccess, especially in the determination of thermodynamic properties. Solid electrolytes withphysical characteristics suitable for molten lead are thus available. Secondly, oxygen potentialmeasurements at temperatures as low as 400°C have been successful even though slow kinetics inthe solid electrolyte are expected. The maj or challenge is then to design a portable probe forcontinuous measurement that would give satisfactory results when used in the specific conditionsof lead softening, i.e. the presence of an oxidic slag, temperature and composition fluctuations onan hourly basis, and a demanding industrial environment.2.2 Process ModellingSince the advent of computers process modelling has become an integral part of the metallurgical field. Numerous process models have been developed covering the fields ofpyrometallurgy,hydrometallurgy andmetal forming. These models cover awide spectrum of simulation capabilities,from smelting and converting to casting and rolling, with the ability to predict the evolution ofprocess characteristics ranging from chemical composition to microstructure. In particular, thermochemical process models are primarily concerned with the simulation of the chemical composition of the various phases involved in a process. Starting from simple models in the 1950s, theybecame rather comprehensive in the 1980s. Such models have traditionally been developed in twostages. First, the process is treated as a “black box” operating at thermodynamic equilibrium anda “thermodynamic model” is formulated based on idealized assumptions. Then, the model isexpanded to include kinetic factors that have been identified as important. Mass transfer conceptsand reaction rates are implemented at this stage. The resulting “kinetics model” then becomes apowerful tool in the elucidation of the rate-limiting steps of the process.- 44 -2.2 Process ModellingInvestigating process fundamentals through modelling avoids the need to carry out extensiveand expensive plant tests that would otherwise be required to analyze and optimize a process. Inthis section, thermodynamic models are discussed first, followed by kinetics models, and finallysome conclusions are drawn.2.2.1 Thermodynamic ModelsThe fundamental assumption is that the process under consideration runs at thermodynamicequilibrium and consequently, that all reactions involved in that particular process proceed toequilibrium. The main purpose of the model is to determine the equilibrium composition of thechemical system representing the process.A traditional approach to determining equilibrium compositions of a multiphase systemconsists in examining the various independent equilibrium reactions at a given temperature, andderiving a set of non-linear simultaneous equations involving the equilibrium constants, the totalpressure, and the elemental material balances. Thermodynamic models of slag fuming6°21,nickelconverting63,andcopper converting64,are all based on this technique. The derived set ofequationsis solved by means of a nonlinear equation-solving routine. For example, Kyllo and Richards63use a Quasi-Newton routine whereas Goto[MJ employs a Brinckley-Newton-Raphson routine. Thisapproach can provide good results when reasonable assumptions are made to simplify the systemof equations. However, this exercise requires some prior knowledge of the system as well asmetallurgical intuition. The technique also presents some difficulties with regard to initial guesses,generality, and convergence.- 45 -2.2 Process ModellingIn the late 1 950s, White et al. [65] developed a new technique, the Gibbs free energy minimization method, that eliminated the above difficulties. Rather than considering the independentequilibrium reactions involved in the process, this technique considers all possible speciesrepresenting the process and computes the equilibrium composition by minimizing the Gibbs freeenergy of the system subject to two constraints: 1) the elemental balances, and 2) the equilibriumcomposition values determined must be positive. Oliver et al. [66] used the two alternative ways tosolve the same problem in order to demonstrate the differences between the two methods. Numerouspapers on this new methodwere later published, notably by Eriksson who developed the well-knownSOLGASMIX routine67’8.Although the underlying principle for all of the programs described inthese papers is the same, the mathematical techniques differ. The SOLGASMIX routine, forinstance, uses Lagrange’s method of undeterminedmultipliers to set up a system of linear equationsthat is then solved with a Gaussian elimination technique. The thermodynamic model of theOutokumpu flash smelting furnace of Vartiainen et al. [69] is based on this technique and uses amodified version of the SOLGASMIX routine.Since the slag fuming process provides a very good illustration of the role and purpose ofthermodynamic modelling, it was selected for a more detailed review. In 1954, Bell et al. [60], whowere the first to attempt a quantitative analysis of the process chemistry, assumed that slag fumingwas described by the following reactions:C+l/202 -* COC+O2 -> CO2ZnO (slag) + CO Zn (g) + CO2- 46 -2.2 Process ModellingH2 + 1/202 <-* H20ZnO (slag) + H2 -* Zn (g) + H20They developed a model based on mass balances on carbon, hydrogen, oxygen and nitrogen, andon the two equilibriaZnO (slag) + CO <-> Zn (g) + CO2H20+C0 <-> H+CO2The unknown activity of zinc oxide was computed from plant data by the addition of a zinc balance.A system of seven equations was derived and solved for the seven unknowns, F P2PH2’‘H2O ‘ PN2 Zn’ andWith this simple model, Bell et al. [601 simulated slag fuming operations and examined theeffect of changes in a number of operating parameters. Their model, which accounted fairly wellfor the furnace heat balance, was successful in predicting the improvements in fuming rates byoxygen enrichment and preheating of the blast. For a simple model, this was quite an achievementeven though themodel was in variance in predicting the effect of other operating parameters. Furtherrefinements by Ke1logg6 1, and Grant and Barnett621, brought the model to such a degree ofsophistication, predicting the effect of most variables including the behavior of iron, sulphur, lead,and temperature with time, that it reinforced the belief that slag fuming was operating at thermodynamic equilibrium.Thermodynamic models usually give realistic results. This is often due to the fact that in- 47 -2.2 Process Modellingpyrometallurgical processes, reactions are usually relatively rapid and, for the major species in areactor, advance a considerable way towards the final equilibrium composition in a short time. Forthat same reason, a thermodynamic model constitutes a useful tool by providing a first step towardsthe understanding of the working of a process. Paradoxically, a thermodynamic model can alsoprovide evidence against its primary assumption of thermodynamic equilibrium. The nickel converter equilibrium model of Kyllo and Richards63 corresponds to such an example. From a thermodynamic study, the author show that the nickel convertingprocess is actually governed by kineticsrather than by thermodynamic equilibrium. They conclude that a kinetics model would be moreappropriate to investigate the process. The following section presents the background for kineticsmodels.2.2.2 Kinetics ModelsContrary to thermodynamic models which assume all reactions to be instantaneous with nospecific reaction site, kinetics models take into account both reactions rates and local heat and masstransfer. Instead of being treated as a black box, the process is investigated on a microscopic scaleusing a mechanistic approach. Reactions are considered to take place at specific locations, such asgas bubble-liquid interfaces or solid particle-liquid interfaces. Although reactions at the variousinterfaces are usually considered to be instantaneous, this approach still permits the assessment ofthe rate-limiting steps. These controlling steps often correspond to mass transport to and from thereaction interfaces.Again, the slag fuming process provides a good illustration of kinetics modelling. The earlythermodynamic models of slag fuming, as described in the previous section, all assume that the- 48 -2.2 Process Modellinginjected coal and air react and come to thermal and chemical equilibrium with the slag. Based onan investigation of several industrial zinc slag fuming furnaces, Richards et al. [70] suggest that zincfuming takes place on entrained coal particles in the slag, and that the process is kinetically controlled. They argue that the fuming furnace consists of two reaction zones created by the divisionof coal between the slag and the tuyere gas stream. The fraction of coal entrained in the slag reducesZnO andFe304,while the fraction of coal remaining in the gas stream combusts and provides theheat for the endothermic reduction reactions and heat losses. A kinetic model was developed byRichards and Brimacombe71’21,and used to analyze the fuming process. By fitting the model tofuming cycles from five different companies, the fractions of entrained, combusted, and unreactedcoal as well as the oxygen utilization were characterized. With the fitted parameters values, themodel is able to predict fuming rates with reasonable accuracy. In the light of the kinetic descriptionof the slag fuming, Richards and Brimacombè71’21conclude that process improvements should beachieved by increasing the amount of coal entrained in the slag, which in turns increases the amountof reductant available for zinc oxide. Cockcroft et al. [731 report on tests carried out with a highpressure coal injection unit installed on a slag fuming furnace to increase the entrained coal fraction.The results of their tests show that significant improvements in fuming rates (1.5 to 2 times) areobtained. This example demonstrates that kinetics modelling combined with industrial measurements constitutes a very powerful tool to improve the understanding ofexisting processes. Improvedunderstanding usually leads to process optimization.2.2.3 ConclusionsBHAS compensated for a limited understanding of process fundamentals of oxygen leadsoftening by extensive research, including the design and testing of several furnaces, in order to- 49 -2.2 Process Modellingachieve satisfactory performances from their lead softening process. Although their process isregarded as an example around the world, their task was facilitated by the regularity of concentratecomposition.In the case of a partial softening process, which would also be required to handle bullioncomposition spikes, a better understanding of process fundamentals are crucial in the success ofany control strategy. In view of the previous sections, process modelling is undoubtedly a powerfultool to quantify process chemistry and kinetics. To date, no mathematical model of the oxygenlead softening process has been published. The development of a thermodynamic model is anappropriate first step towards the understanding of the basic reactions involved in the oxygen leadsoftening process.- 50 -3 SCOPE AND OBJECTIVESCHAPTER 3SCOPE AND OBJECTIVESLead ores generally contain significant amounts of arsenic and antimony. The process oflead softening, corresponding to the removal of these hardening impurities, is therefore ofnecessityin any pyrometallurgical lead refining circuit. Even in a smelter where electrorefining is used, apartial softening might be required. The treatment of silver bearing concentrates, while providingadditional revenues, also introduces additional amounts of arsenic and antimony into the circuit.The efficiency of the Betts electrorefining process depends on the stability of the corroding leadanodes. Anode slime stability is directly related to the antimony and arsenic content of the anodes.Thus, control of bullion quality prior to anode casting is key to optimization of the electrorefiningoperations, improvement of lead production, and addition of revenues. The (partial) softening stageis the ideal location in a smelter for such a control. When composition spikes are regular occurrences,the batch softening circuit tends to become somewhat complicated to reduce their effects.The objective of this research work was to explore the option of continuous, single passsoftening for the production ofbullion of consistent quality. In this case, the softening circuit wouldappear as in Figure 3.1. Two components are critical to this revision:• a method of continuously monitoring bullion quality for control purposes, and• a control strategy based on an understanding of process fundamentals.- 51 -3 SCOPE AND OBJECTIVESContinuousCopper DrossingFigure 3.1 - Continuous single pass softening circuit.Crude leadDustCu-matteSlag Softener TAnode casting- 52 -3 SCOPE AND OBJECTIVESThe scope of this thesis work was primarily to develop an on-line sensor for bullion qualitycontrol. Continuous monitoring of bullion quality involves a continuous measurement in-situ ofthe arsenic and antimony content of the bullion. Solid state sensors, or electrochemical probes, arethe most common devices for on-line continuous measurement of liquid metal composition. Anantimony (or arsenic) sensor would be most suitable to directly measure the antimony (or arsenic)content of the bullion. No antimony-ion (or arsenic-ion) conducting solid electrolyte is commercially available. The development of such an electrolyte, from fabrication (including setting up theappropriate experimental equipment), to characterization (electrical conductivity, ionic domain,..), and finally testing in molten lead (stability, corrosion resistance, ...), would by itself representa PhD research project. Since this project was primarily concerned with oxygen lead softening,including designing a sensor forbullion quality control, rather than developing one single componentof a sensor in the form of a new solid electrolyte, an alternate route had to be adopted. The amountof dissolved antimony and arsenic in the bullion is directly related to the amount ofdissolved oxygendue to thermodynamic constraints. Thus, measuring the content of dissolved oxygen by means ofan oxygen probe would provide an indirect measurement of the antimony and arsenic content,assuming a relationship between dissolved antimony and arsenic and dissolved oxygen is established. Both laboratory and industrial tests were required in the oxygen probe development program,the former to perform all necessary design experimentation on the individual components as wellas on the whole assembly, and the latter to test the suitability ofthe probe in an industrial environmentas well as to calibrate the probe.The second objective was to develop a preliminary control strategy based on a continuousmeasurement of bullion quality as measured by the combined arsenic and antimony level. Such anassessment required the understanding ofprocess fundamentals, and in particular process chemistry.- 53 -3 SCOPE AND OBJECTIVESAs mentioned in Chapter 1, a number of fundamental issues, such as oxygen efficiency, influenceof bullion composition, effect of temperature, have not been addressed in the literature. Processmodelling, combined with industrial measurements, constitutes a very powerful tool to analyze aprocess. Since no mathematical model of oxygen lead softening has been published to date, athermodynamic model of the process has been developed as a first step towards the quantificationof softener chemistry. This model was designed to allow the simulation ofbothbatch and continuousoperating modes. Once calibrated with plant measurements for the batch mode, the model couldbe used to analyze the continuous single pass softening option for a variety of conditions in orderto demonstrate its ability tomeet targets for bullion quality. The acquisition and analysis of industrialdata has thus been a major thrust in achieving this objective. Visits to a smelter provided theopportunity to gather data on an industrial softener.- 54 -CHAPTER 4OXYGEN PROBE DESIGNDesigning an oxygen probe for continuous measurements in a lead softening process requiredthe completion of a number of steps. First, the appropriate materials were selected in order to satisfythe requirements of operating in molten lead within a range of temperatures and oxygen partialpressures characteristic of lead softening. Once the materials had been selected, a cell assembly,i.e. the manner in which the various components are assembled together, was adopted taking intoaccount that a number of prerequisite parameters had to be satisfied. The probe needed to becorrosion resistant, robust and portable with a life-span of at least a day. The next step consistedin carrying out laboratory tests in pure molten lead to confirm the adequacy of each component andof the assembly as a whole. The reversibility, thermal resistance and response of the probe wereinitially the main concerns. The probe was then calibrated to ensure proper oxygen measurements.Finally, the response of the probe to temperature and oxygen partial pressure changes was investigated. The oxygen partial pressure changes were generated by successive additions of antimonyinto pure lead. At this point, a final laboratory probe design had been achieved. A number ofprobes could then be constructed for testing in an industrial environment. It was expected that somemodifications to the design would probably have to be implemented later to improve the probecharacteristics and make it a tool for process control. This chapter is divided in two main parts. Inthe first section, the selection of materials is discussed. In the second section, the probe design isdescribed and the laboratory tests are presented.- 55 -4.1 Selection of Probe Components4.1 Selection of Probe ComponentsSince it is intended to integrate an oxygen probe in a process control system to monitor thelead softening operations at Cominco Ltd plant in Trail, this probe should provide an accurate andcontinuous measurement of the oxygen content of a lead bullion in the temperature range of 450°Cto 750°C, for oxygen pressures in the range of 10b0 atm and 1022 atm, and for periods of time of atleast several days. One of the critical aspects in the design of such a device consists in selectingthe propermaterials according to the specific set of operating conditions. In this section, the selectionof the probe components such as solid electrolyte, reference system, and conducting lead material,is discussed.4.1.1 Solid Oxide ElectrolyteThe transport properties of solid electrolytes are primarily due to the presence and mobilityof lattice defects. In particular, electrical conductivity is determinedby the nature and concentrationof various ionic and electronic point defects such as vacancies, interstitial or misplaced atoms,impurities, and free electrons or electron holes. All these defects coexist in a solid electrolyte.However, for given temperature and pressure conditions as well as given composition and crystalstructure, a particular type of defect predominates due to energy considerations yielding a specifictype of conduction. A solid electrolyte will then be either an ionic or an electronic conductordepending on the predominance of either ionic or electronic defects (mixed conduction can also beobserved). A knowledge of the defect structure is therefore essential. In the case of solid oxideelectrolytes, the generation of lattice defects and the variation of their concentrations are temperature and oxygen partial pressure (or oxygen potential, 1102) dependent. Consequently, the type- 56 -4.1 Selection of Probe Componentsof conduction in a solid oxide electrolyte will also be temperature and oxygen pressure dependentas illustrated in Figure 4.1. A solid oxide electrolyte exhibitsfree electron conduction (as, n-type)for low oxygen pressures, positive hole conduction (a,p-type) for high oxygen pressures, andionic conduction (a10) for intermediate oxygen pressures as shown in Figure 4.2 (a). Ionic conductivity in solid oxide electrolytes, which occurs only within a limited oxygen pressure range asshown in Figure 4.2 (b), is due to the presence of oxygen ion vacancies. The addition of aliovalentimpurities as dopants can increase the number of vacancies to such an extent that these vacanciesgenerate a highly defected structure through which oxygen ions can migrate inside the electrolyte.The most typically used dopants are calcia, CaO, magnesia, MgO, and yttria,Y203. Figure 4.2 (c)shows the effect ofY203 addition on Th02 conductivity.The success of the probe depended largely on selecting amaterial that was an ionic conductorover the range of oxygen partial pressures encountered in the softening process, and that had anadequate ionic conductivity in the temperature range of application. The properties and characteristics of solid oxide electrolytes have been extensively reported in the literature7482. Fourprincipal types of solid oxide electrolyte are available: ö-Bi203,CeO2, Th02, and Zr02 basedmaterials, CaO, MgO,Y203being the most commonly used dopants. The relevant characteristicsof these electrolytes can be summarized as follows. The 6-Bi203based materials cannot be usedfor oxygen determination in molten metals due to problems caused by their instability at roomtemperature82.The CeO2 based materials have excellent characteristics at low temperatures, andin particular have a very good conductivity for temperatures below 500°C, but their oxygen pressurerange of application is very limited in the temperature range of 500 to 700°C. It extends only downto about 1013 atm at 600°C [771• The Th02based materials have properties suitable for use as solidoxide electrolytes, but their conductivity is the lowest of the group in the temperature range of- 57 -4.1 Selection of Probe Components0-JypFigure 4.1 - Schematic drawing showing the effect of temperature and oxygen pressure on theconductivity of a solid oxide electrolyte (Adapted from Patterson, 1971, ref. 74).an- 58 -4.1 Selection of Probe Components00)0-J1.00.8g 0.6- 0.40.2000)0-JFigure 4.2 - Schematic representation of (a) partial electrical conductivities, (b) ionic transferencenumber inY203-doped Th02, and (c) effect of dopant concentration on partial conductivities inTh0-Ysolid solutions (Adapted from Choudhary et at., 1980, ref.75).n-type mixed ionic mixed p-typea(a)(b)(c)Log P02- 59 -4.1 Selection of Probe Components450°C and 750°C (see Figure 4.3). The Zr02 based materials have been widely used due to theircommercial availability and their wide range of applicability: good electrical conductivity over alarge temperature range (see Figure 4.3) and wide electrolytic domain (see Figure 4.4). For thesereasons, zirconia based electrolytes were the most appropriate materials for the probe.From the comparative review of past applications of zirconia electrolytes in molten lead,presented in Chapter 2 and summarized in Table 2.1, and in light of the above restrictions, twomaterials appeared to have the highest potential for application in lead softening:• Zr02-15 mol% MgO for its corrosion resistance properties and good conductivity attemperatures below 800°C, and• Zr02-6 or 8 mol%Y203 for its large electrolytic domain and long life span.Figure 4.5 provides a comparison between the electrical conductivity of commercial zirconia solidelectrolytes doped with yttria and magnesia. Yttria stabilized zirconia was chosen since it has ahigher conductivity in the temperature range 450°C to 750°C than magnesia stabilized zirconia.4.1.2 Reference SystemWhen selecting the reference system, two choices were made: first deciding whether to usea gaseous or a solid reference, and secondly deciding which chemical system was best suited. Theserious drawback of a gaseous reference system, i.e. diffusion of molecular oxygen through thefine pores and microcracks of the electrolyte, only occurs at a negligible rate at temperatures between450°C to 700°C. However, a gaseous reference system has the disadvantage ofbeing less convenient- 60 -4.1 Selection of Probe ComponentsTemperature, °C1100 1000 900 800 700 60010I I I I I8120326,110,%10_I‘7io %-c05-.,C.) io0C) 1•04,>E2I I I0.7 0.8 0.9 1.0 1.13 -11)TxlO ,(°K)Figure 4.3 - Conductivity of various solid oxide electrolytes as a function of temperature (Adaptedfrom Choudhary et al., 1980, ref. 75).- 61 -4.1 Selection of Probe ComponentsC”000)0-j2000 1400 1000Temperature, °C800 600 400I I J I I IySZ403020100-10-20-30-40-50-60-70-80 I I4 6 8 10 12 14 161/Tx104,(°K)Figure 4.4 - Electrolytic domain of various zirconia solid oxide electrolytes as a function oftemperature and oxygen partial pressure. YSZ stands for Yttria Stabilized Zirconia,CSZ for Calcia Stabilized Zirconia, and YDT for Yttria Doped Thoria (Adapted fromDe Schutter et al., 1991, ref. 57).I I- 62 -4.1 Selection of Probe Components00i!6 7 8 9 10 11 12 13lITx (°K)1Figure 4.5 - Characteristics of commercial zirconia solid electrolytes. The letters correspond tothe type of dopant, withM for magnesia, C for calcia and Y for yttria, and the numberscorrespond to the mole percent of dopant. (Adapted from Nippon Kagaku Togyo Co.Ltd. marketing pamphlet, 1991, ref. 83).Temperature, (°C)1300 1100 900 700 600 500ZR-8Y-6Y-1-2-3 —IC-5I__________I I I I I- 63 -4.1 Selection of Probe Componentsthan a solid reference system: loss of portability, problem of gas flow rate regulation, and large emfreadings resulting in loss of accuracy231. On the other hand, a solid reference system seems to bemore adapted to the objectives of this project, even though Kozuka591suggests that a solid referencemight not retain its true oxygen potential for long periods of time. A solid reference system isusually a mixture of fine powders ofmetal and its oxide. Ni-NiO, Cr-Cr23Cu-Cu20,and Fe-FeOare some of the commonly used mixtures.Since solid electrolytes are not fully ionic, t10, 0.99, an experimental difficulty may occurwhen attempting to maintain a desired oxygen potential at the electrode-electrolyte interface. Theelectronic conductivity contribution of the solid electrolyte, even though very small, te1ecofl 0.01,generates a short-circuiting current through the electrolyte, and subsequently a shift in the oxygenpotential at the electrode-electrolyte interface from that established by the metal-metal oxideequilibrium to that fixed by the steady state oxygen transfer. This electrode polarization phenomenon is characterized by the generation of an overvoltage. If the short-circuiting current isvery small, a stable emf can be obtained which includes an overvoltage contribution. Measurementsof overvoltage versus current for symmetrical electrochemical cells of the typeMetal, Metal Oxide I Zr02 + CaO I Metal Oxide, Metalperformed by Worrel and Isko&84 are reported in Figure 4.6. In order to minimize polarization,the reference system should have an equilibrium oxygen pressure close to that prevailing in themolten lead at the temperature of interest. Values of P02 for various metal-metal oxide mixturesat temperatures between 500 and 700°C are given in Table 4.1. The equilibrium oxygen pressureswere calculated using data from Barin and Knackè851,- 64 -4.1 Selection of Probe ComponentsFrom this review, it appeared that the system Cu-Cu20would give the lowest level ofpolarization as it exhibits an oxygen equilibrium pressure closest to that prevailing in a lead bullionin equilibrium with an oxidic slag. For these reasons, a Cu-Cu20mixture was considered the bestsuited reference for application in lead softening.Table 4.1 - Equilibrium oxygen pressure of various metal-metal oxide systems at temperaturesbetween 500 and 700°C.Equilibrium Oxygen PressureMetal-Metal Oxide System (atm.)500°C 600°C 700°CPb(S) + 1/2 °2(g) > PbO(s,yelIow) 1.1 x 10.19 2.7 x 10.16 1.3 x 1013Ni (s) + 1/2 °2 (g) NiO 1.1 x 1023 5.3 x 10° 4.3 xFe(S) + 1/2 °2(g) FeO(5) 1.3 x i0° 1.9 x 1026 3.7 x 10232 Cr(S) + 3/2 °2(g) Cr203(5) 2.8 x 1042 1.7 x 1036 6.7 x 10322 CU(S) + 1/2 °2(g) Cu20(5) 7.0 x 1016 2.8 x i0’ 3.1 x 10h1- 65 -4.1 Selection of Probe Components>a)Cu0a)>0Figure 4.6 - Overvoltage versus current for various metal-metal oxide systems at 900°C (Adaptedfrom Worrel and Iskoe, 1973, ref. 84).I INi-MO100101.00.10.01Fe-FeOCu-Cu200.1 1.0 10 100i, Current (pA)1000- 66 -4.1 Selection of Probe Components4.1.3 Lead Wire and Conducting LeadA lead wire is a metallic wire that connects the reference electrode to a measuring devicesuch as an electrometer. Platinum is the most commonly used material for lead wires in laboratoryoxygen sensors. A conducting lead corresponds to a rod or tube of conducting material that isdipped into a molten metal in order to provide electrical contact with the melt. The conductinglead actually corresponds to the counter electrode of the probe. Since the conducting lead is incontinuous contact with the melt, one of its most important characteristics is that it should becorrosion resistant.Various metals such as platinum, iridium, molybdenum, as well as stainless steel and Crcermet have been successfully employed as conducting lead materials (see Table 2.1). Even thoughplatinum has excellent characteristics in terms of chemical stability, electronic conduction andavailability, it might not be the most adequate material for temperatures in the order of 500 to 700°Cdue to its slow response to oxygen exchange reactions at the electrode-electrolyte interface.Moreover, its cost could be prohibitive for an industrial device. Cr-cermet has been reported todissolve in molten lead for temperature higher than 850°C, whereas iridium does not show anysolubility in molten lead481. Stainless steel 316 is very appealing due to its commercial availabilityin various shapes and sizes (wires, rods and tubes) at reasonable cost. Provided it possesses goodcorrosion resistance characteristics in molten lead and antimonial lead slag, stainless steel 316 isbelieved to be the best choice for an oxygen probe in lead softening. To avoid the use of dissimilarleads, stainless steel 316 was also chosen as lead wire material.- 67 -4.1 Selection of Probe Components4.1.4 ConclusionsIn view of the above operating requirements inherent to lead softening, the followingmaterialshave been selected for the main components of the probe.• Solid electrolyte: commercially available yttria stabilized zirconia (Zr02-6 or 8 mol%Y203)from Coors Ceramic Co. and Nippon Kagaku Togyo Co. (see Table 4.2 for properties).• Reference system: Cu and Cu20powders from Cerac Inc. (see Table 4.3 for assays).• Lead wire and conducting lead: 316 stainless steel wire and tubing.Table 4.2 - Properties ofY203 fully stabilized zirconia from Coors Ceramic Co.Properties Units ValueBulk density g/cc 5.60Electrical conductivity (Log a at 650°C) - 2.5Coefficient of thermal expansion (25-1000°C) 106/°C 10.5Thermal shock resistance (maximum tolerable temperature difference) °C 150Maximum use temperature °C 2400- 68 -4.1 Selection of Probe ComponentsTable 4.3 - Assays for Cu and Cu20powders from Cerac Inc.Impurities Cu powder Cu20powder% (-200, +325 mesh) (-200 mesh)Typical purity 99.5 % 99.9 %Ag - 0.008Al 0.001 0.001Ca - 0.001Fe 0.005 0.004Mg - 0.005Mn - 0.001Ni - 0.003Pb 0.005 -Si 0.001-0.01 0.005Sn 0.005 0.05Zn 0.005 -- 69 -4.2 Probe Development in Laboratory4.2 Probe Development in LaboratoryOnce the materials had been selected, the various components of the probe were assembledtogether according to a chosen probe geometry. The cell assembly design was then tested forthermal and corrosion resistance, calibrated to ensure proper oxygen measurements, and eventuallymodified to improve its characteristics. This section is divided in three parts. First, the experimentalapparatus built for this project is described. Then the development of the cell assembly design isdiscussed. Finally, the laboratory tests of the probe in molten lead are presented and a probe designfor industrial tests is discussed.4.2.1 Experimental ApparatusThe experimental set-up is composed of three main components: a furnace, a temperaturecontroller, and a data acquisition system as shown in Figure 4.7. The furnace, specially designedand constructed for this project, is a vertical unit with a height of 1 metre and a square base of 0.5metre. The heating element is made of chromel ribbon with variable pitch windings to providerigorous temperature control of± 1°C within ahot zone of about 10cm. A temperature profile alongthe vertical axis of the furnace chamber is given in Figure 4.8. This rigorous control is required inorder to avoid temperature gradients within the probe, which would otherwise influence the probereadings. A sheet metal shield provides protection around the hot zone to reduce the electricalpick-up as well as electromagnetic effects created by the chromel windings. The chamber is madeof a mullite tube with an inside diameter of about 7 cm which allows the use of crucibles with anouter diameter of up to 6.7 cm. This is particularly important since large crucibles allow simultaneous testing of two probes without restricting access to the crucible for other instrumentation- 70 -4.2 Probe Development in LaboratoryMillivoltmeterEXP-16multiplexerFigure 4.7 - Schematic representation of the experimental set-up for laboratory tests.1125.4 LII0-probe leadsTIC leadsFurnaceControl T/C/Personal computerwith DAS-8 boardTemperature controller- 71 -C-)a)a-)C-)0-a)a)a-)C)(Ii-U;2n254.2 Probe Development in Laboratory201510__50—5—10620 650Temperature, °CFigure 4.8 - Temperature profile along the vertical axis of the chamber with location and lengthof the hot zone (± 1°C) for a temperature set point of 6 10°C using the outer thermocouple (S-type).630 640- 72 -4.2 Probe Development in Laboratorysuch as reagent addition tube, sampling tube, and thermocouple. It also provides some bufferingeffect for composition control with the use of large volumes of molten lead. The chamber can beclosed off with top and bottom gas-tight caps. In these circumstances, all instruments are insertedinto the chamber through appropriate gas-tight Swagelok tube fittings. Gases can be introducedin the chamber to control the atmosphere when required. For this purpose, a gas purification train,for moisture, oxygen and hydrogen removal using silica gel, phosphorus pentoxide and BASFcatalyst R3-11, is kept on stand-by. The chamber is schematically represented in Figure 4.9.Temperature control is achieved by an Omega CN9000 Series microprocessor controllerwhich features PID (Proportional-Integral-Derivative), PD (Proportional-Derivative), Proportional,and On/Off control. Two separate thermocouple locations can be used for temperature control:mid-way along the vertical axis of the furnace at the outer periphery of the heating element, orinside the chamber close to the top of the crucible. In both alternatives, a S-type thermocouple(Pt/Pt-Rh) is used. The experimental temperature is measured by means of a K-type thermocoupledipped into the molten lead sample and protected by an alumina sheath closed at one end.The data acquisition system hardware is composed of a Keithley EXP-16 multiplexer and aKeithley DAS-8 board. Labtech Notebook is used as data acquisition software. Prior to themultiplexer, the probe leads are connected to a Corning 130 pH and millivolt meter. Such aninstrument has an input impedance greater than 1012 ohms, which is required to obtain a reliablereading from an oxygen probe that has a very high resistance.Further technical details of the furnace, the millivoltmeter, and the data acquisition systemare given in Appendix 1.- 73 -4.2 Probe Development in LaboratoryFigure 4.9 - Schematic representation of the chamber with measuring devices in position forlaboratory tests.TIC in alumina sheathI II I II I IChromel windings ii ;j77’ /I/ // // /I/ /// // /Mullite tubes/ /7-,’I/// // // /Metallic shield —i 7 // // /I/ /// /I/ // /Metallic basket -//////// JAlumina crucible i —I I I I IIOxygen probe7- 7/ // // // // // // // // // // /J/// // // // // /// // // // // // /H / // // /___/ /_/ // // // /II////////7///Alumina support plate Molten lead- 74 -4.2 Probe Development in Laboratory4.2.2 Cell AssemblyTwo cell geometries are possible as illustrated by Figure 4.10. The geometry using a diskcemented at the tip of a silica tube is typical of single reading or disposable probes used in steel-making operations. One of the main advantage of solid electrolyte disks is their low cost comparedwith thimbles. Problems of leakage at the zirconia-silica joint is however a serious drawback ifcontinuous measurements for long periods of time is desired. For this project, the geometry usinga solid electrolyte thimble was the preferred option. Figure 4.11 shows a schematic drawing of atypical probe design for continuous measurements which was adopted to start the experimentalwork. The dimensions of the zirconia thimbles are: 6 mm OD, 4 mm ID, and 35 mm long.The behavior of this assembly upon heating was the first concern to be addressed. Thecomponent materials have very dissimilar coefficients of thermal expansion: 10.2, 12.7 and 17.4106 1°C for the solid electrolyte, the cement and the stainless steel, respectively. Stresses withinthe assembly and generated during heating or cooling cycles can lead to cracking of the zirconiatube. Two parameters can be adjusted to reduce the risk of cracking due to thermal propertydifferences between the components: the thickness of the stainless steel housing, and the amountof cement used for the metal-ceramic joining. A commercially available, high temperature, magnesia based cement, Ceramabond 571 from Aremco Product Ltd., was chosen for the metalceramic bonding. Experiments demonstrated that the thinner the steel housing and the larger theamount of cement, the better the resistance of the probe to thermal stresses. The success of theseexperiments, no cracking even after several heating and cooling cycles with the same probe, resultedin the design of a customized steel sleeve to serve as probe tip housing. A scale drawing of thissteel sleeve is given in Figure 4.12.- 75 -4.2 Probe Development in LaboratoryFigure 4.10 - Schematic representation ofthe possible cell assembly geometries for oxygen probes.Alumina, silicaor steel tubingLead wireCementReference electrodeSingle readingSolid electrolyteContinuous reading- 76 -4.2 Probe Development in Laboratory316 stainless steel wire316 stainless steel housingMagnesia cementMetal-ceramic joiningZr02-Y03 electrolyteFigure 4.11 - Schematic representation of theexperiments.first design of cell assembly used in the earlyCu-Cu20mixture- 77 -4.2 Probe Development in LaboratoryFigure 4.12 - Scale drawing of the steel sleeve used as housing for the probe tip (35 mm longzirconia thimble shown in position).9.5 mmThread6.2 mm15.0mm 10.0mm- 78 -4.2 Probe Development in LaboratoryThe next major concern consisted in ensuring that the probe readings were within reasonableaccuracy. For this purpose, the probe was immersed in oxygen saturated lead bullion obtained bymelting high purity lead in an oxidizing atmosphere and leaving the melt until a layer of yellowlead oxide was formed on the surface. Using oxygen saturated lead provides a solution that exertsan oxygen partial pressure which is temperature dependent only. Monitoring the deviation betweenthe theoretical and measured oxygen partial pressure of the melt permitted the validation of a designor the identification of any problem that needed to be addressed. Using data fromJacob and Jeffes48,the temperature dependance of the equilibrium oxygen pressure for the formation of lead oxideaccording to the reactionPb + 1/2 °2(g) E4 PbO (s, yellow)is given by_52468— 23.79- T(4.1)where T is the temperature in Kelvin. Similarly, the temperature dependance of the equilibriumoxygen pressure for the formation of copper (I) oxide according to the reaction2 Cu (s) + 1/2 °2(g) Cu20(s)is given by40341.2ln O(Cu/CO) = 17.225- T(4.2)- 79 -4.2 Probe Development in LaboratoryThe oxygen partial pressures 0 and P02 (CCO) were computed at different temperaturesusing Equations (4.1) and (4.2). The probe theoretical emf, Eth, was then calculated for the sametemperatures using Equation (2.8) inwhich the appropriate oxygen partial pressures were substitutedas followsRT O2(Cu/CuO)Eth = — in (4.3)4F P02 (Pb/PbO)The computed oxygen partial pressures as well as the theoretical emf are given in Table 4.4. Aregression analysis yielded the following expression of the temperature dependence of EhEth (mV) = 222.6 - 0.1414 T (°C) (4.4)The standard Gibbs free energies of formation of PbO and Cu20from Jacob and Jeffes148,AlcockandBelford44,and Charette and Flengas451,all obtained by an emf technique, yield theoretical emfvalues in agreement within ± 2 mV , whereas data from Barin and Knacke851,and Kubaschewskiand A1cock861 yield theoretical emf values 9 mV and 13 mV lower, respectively. The choice to usedata from Jacob and Jeffes481 was motivated by the fact that among the investigators they were theonly ones to use an emf technique and a temperature range which includes the temperaturescharacteristic of lead softening. In addition, their data is supported by the results of two of the otherstudies.- 80 -4.2 Probe Development in LaboratoryTable 4.4 - Computed values of P02 (Pb/PbO)’ P02 (Cu/CO)’ and Eh.Temperature P0,(Pb/PbO) 1O (Cn/Cu20) E th(°C) (atm.) (atm.) (mY)500 7.263 x 1020 6.632 x 1016 151.9550 4.479 x 10’s 1.577 x 144.8600 1.723 x 10.16 2.6 10 x 1013 137.7615 4.753 x 1016 5.695 x l0’ 135.6650 4.462 x 3.187 x 1012 130.7700 8.273x10’4 3.008x10’ 123.6750 1.153x10’2 2.281x10’° 116.5800 1.257x10’ 1.432x10°9 109.5- 81 -4.2 Probe Development in LaboratoryA number of probes were built according to the design presented in Figure 4.11, and theyrecorded an emf out of range while immersed in the molten lead. Figure 4.13 shows typical emfand temperature recordings of such a probe for a test performed at a temperature of 675°C for whichthe theoretical emf is 127.1 mV. The probe emf reached about 170 mV within one minute afterimmersion and gradually increased for about one and a half hours until it stabilized around 270mV. Modifications of parameters such as amount of Cu-Cu20mixture, sintering of the Cu-Cu20mixture within the thimble, amount of cement, duration and temperature of cement curing, wereunsuccessful in attempting to achieve proper emf readings. Even though the stabilization of themeasured emfargues against the possibility ofcracks in the zirconia tube, carefully cut cross-sectionsof the probe tips were examined under a microscope for confirmation. No cracks were apparent.Further observations of the same cross-sections revealed that the reference electrode had lost itsintegrity during the tests. A layer of Copper (II) oxide (CuO) had formed at the reference electrode/solid electrolyte interface indicating that oxygen had reached and contaminated the referencesystem. Since within the temperature range of the tests, P02 (CCO) > P02 (Pb/PbO)’ the hypothesisof oxygen migration from the melt through the zirconia solid electrolyte to oxidize the Cu-Cu20mixture is thermodynamically disputed. It was concluded that the oxidation of the Cu-Cu20mixturewas caused by oxygen ingress from the top of the probe. The oxygen could either come fromleftover water contained in the cement or from the oxidizing chamber atmosphere through theporous cement. In either case, this oxygen ingress had to be stopped to maintain reference electrodeintegrity. The solution was to integrate an oxygen getter between the reference electrode and thecement seal. A plug of copper powder appeared to be well adapted for the purpose. The term“p1ug is used because of the property of the copper powder to partially sinter after some exposureto the experimental temperature, becoming more compact and acting as a stopper. Alumina powderwas used to separate the copper plug from both the Cu-Cu20mixture and the cement. The lead- 82 -4.2 Probe Development in Laboratory700690- Temperature - 450680 400L) 670 - - 3500660 -- 300I E620- 100610 -50600— I 00 10 20 30 40 50 60 70 80 90 100Time, minutesFigure 4.13 - Emf and temperature readings recorded during a test with a probe fabricatedaccording to the design adopted in the early tests.- 83 -4.2 Probe Development in Laboratorywire needed to be electrically insulated to avoid any short-circuit by contact with the copper plug.A cement coating was applied on the lead wires for that purpose. To eliminate the risk of oxygencontamination by the water contained in the cement, the cement coating was allowed to completelycure in a drying oven according to the curing treatment recommended by the manufacturer. Thisimproved probe design is schematically represented in Figure 4.14. In order to provide room forthe additional alumina and copper powders, longer zirconia thimbles were used. Their dimensionsare as follows: 6 mm OD, 4 mm ID, and 70 mm long.4.2.3 Measurements in Molten LeadA probe tip, as represented in Figure 4.14, was assembled to a holding device in order to beinserted into the vertical furnace for measurement in molten lead. A probe holder, serving asconducting lead as well, was designed using 316-stainless steel tubing so that a probe tip could bescrewed on at one end. A quartz tube was used to insulate the lead wire from the conducting leadin order to prevent electrical contact which would otherwise short-circuit the probe. Pre-purifiedargon could be introduced through the quartz tube, flush the probe holder and exit at the top of theassembly. Flushing the probe holder provided extra protection for the reference system againstoxygen ingress from the atmosphere. Figure 4.15 gives a schematic representation of the probeand its holding device.A number of experiments were carried out to investigate the characteristics of the probe suchas response time and accuracy, cell reversibility, life-span, and reproducibility. All the requiredtests initially involved molten lead saturated with oxygen obtained as previously described bymelting lead in an oxidizing atmosphere and leaving the melt until a layer of yellow lead oxide was- 84 -4.2 Probe Development in LaboratoryMagnesia cement coatingMagnesia cementAlumina powderCopper powder316 stainless steel sleeveMagnesia cementAlumina powderCu-Cu20mixtureZr02-Y03 electrolyteFigure 4.14 - Schematic representation of the improved design of cell assembly.316 stainless steel wire85 -4.2 Probe Development in LaboratoryArgon inletQuartz tube31 6-stainless steel tubeFigure 4.15 - Schematic representation of the probe and the probe holding device.Stainless steel fittingArgon outletProbe tip- 86 -4.2 Probe Development in Laboratoryformed on the surface. Ten probes were built and used for laboratory testing, and only two weredysfunctional (out of range emf). Figure 4.16 shows a typical probe response upon immersion inoxygen saturated molten lead. It can be seen that the probe responded immediately and a stableemf of 125 mV ± 1 mV was obtained in less than fifteen minutes at a temperature of 697°C ± 1 °C.The theoretical emf value calculated from Equation (4.4) is 124.0 mV. All the successful probesgave emf readings in very good agreement with the theoretical emf value when immersed in oxygensaturated molten lead (see Table 4.5). Each probe was tested in molten lead obtained by meltingpieces of solid lead from the same lot of ingots. Any discrepancy between measured and theoreticalemf could be caused by three main factors: 1) uncertainty in the data from which the theoreticalemf was calculated, 2) the polarization of the Cu-Cu20reference electrode which would decreasethe measured emf, and 3) the presence of impurities in the Cu-Cu20reference, such as CuO, whichwould increase the equilibrium oxygen pressure of the reference system, and consequently increasethe measured emf. The choice of the standard Gibbs free energies of formation of PbO and Cu20from Jacob and Jeffes481 has been discussed previously. The Cu-Cu20reference electrode wasspecifically chosen to provide, if any, the minimum polarization overvoltage. The Cu and Cu20powders, although not of the highest purity (see Table 4.3), seem to provide a reasonably goodreference. Since differences ranging from 0.2 to 3.0 mV between measured and theoretical emfwere observed, which is within an acceptable error, it was concluded that in the laboratory the probeprovides an accurate and reproducible measurement of oxygen dissolved in molten lead.During some of the above tests in oxygen saturated molten lead, the probe was positioned atvarious vertical levels within the melt to assess the effect of any potential temperature gradient.The measured emf remained constant which indicated an absence of oxygen potential gradientwithin the molten lead, and consequently an absence of temperature and composition gradients.- 87 -4.2 Probe Development in LaboratoryTable 4.5 - Comparison of theoretical and measured emf for various probes immersed in oxygensaturated molten lead.Probe # Temperature E th E meas E meas - E th(°C) (mV) (mV) (mV)1 697.5 124.0 124.4 0.42 698.7 123.8 127.0 0.23 678.6 126.6 127.0 0.44 679.9 126.5 128.0 1.55 682.0 126.2 127.6 1.26 680.4 126.4 128.0 1.67 697.0 124.0 127.0 3.08 663.0 128.8 130.0 1.2- 88 -4.2 Probe Development in Laboratory710 I I I • 300700 250IIITemperatureo 690- 2000) t:rjz660- 1500)0) 670- Emf - 100660 -- 50650 I I I I I I 00 10 20 30 40 50 60 70 80Time, minutesFigure 4.16 - Typical probe response upon immersion into an oxygen saturated lead melt at 697°C± 1°C.- 89 -4.2 Probe Development in LaboratoryOnce it was established that the probe gives a satisfactory response, it was necessary to ensurethat the probe was reproducible. For this purpose, the probe was subject to a temperature change.Increasing the temperature from 683 °C to 730 °C in two steps, as shown in Figure 4.17, generateda decrease of emf from 128 mV to 122 mV. This emf decrease with a temperature increase isconsistent with Equation (4.4). Once the temperature was set back to 683 °C, the emf increasedback to 127.5 mV proving the reproducibility of the probe.The life-span of a probe depends on the amount of copper powder used in the probe construction. Below a critical amount, the probe would last only a couple of hours after which the emfwould start deviating. With copper powder in excess of the critical amount, the probes lasted forlonger periods of time, up to several days. The amount of copper powder is measured as the lengthof the zirconia thimble that is filled with powder, and the critical amount corresponds to a lengthof about two centimetres.The next step consisted in testing the probe response to changes of oxygen partial pressure.For this purpose, solid antimony was added to the melt, where it reacted with the dissolved oxygento form Sb015,and a new oxygen partial pressure was established. Figure 4.18 shows the changesin the probe emf in response to successive additions of solid antimony. The probe response wasvery rapid which was very encouraging since the objective is to use the probe to monitor the oxygenlevels in a vessel running in a 30-minute cycle mode. It should be mentioned that the temperaturefluctuations observed during the experiment were not large enough and not in the right directionto be responsible for the emf changes. The introduction in the chamber near the controllingthermocouple of a “cold” reagent addition tube for antimony addition created the drop in temperatureto which the controller responded by a slight overshoot. As a result, an overall temperature increase- 90 -4.2 Probe Development in LaboratoryFigure 4.17 - Probe response to a temperature change. Initial temperature 683°C, intermediatetemperature 730°C, and final temperature back to 683°C.C-)00)zct5ci)ci)F-7507407307207107006906806706606506406301801701601501401301201100 1 2 3 4 5 6 7 8 9Time, hours- 91 -4.2 Probe Development in Laboratory650 250640 - 240210Q) 610 - SbZ 200w600-Sb590 -Emf 180Q)E- 580 -170160570 -560 - 150550 I 1400 1 2 3 4Time, hoursFigure 4.18 - Probe response to additions of antimony at temperatures around 625°C.- 92 -4.2 Probe Development in Laboratoryof about 10°C was measured between the start and end of the experiment. A number of tests withantimony additions were carried out and samples of molten lead taken. The measured emf andcorresponding lead assays are summarized in Table 4.6.Following the success of the probe in the laboratory, ten probes were built according to thedesign given in Figure 4.14 in order to carry out a plant testing campaign. The results of thiscampaign are discussed in the following Chapter.Table 4.6 - Measured emf in laboratory tests and corresponding lead bullion assays.Measured Emf Temperature As + Sb Total Impurity As + Sb(mV) (°C) (wt%) (%) Total Impurity147 615 0.22 0.22 1.00143 620 0.24 0.24 1.00170 621 0.35 0.35 1.00151 620 0.26 0.26 1.00182 615 0.61 0.64 0.95168 615 0.37 0.61 0.61- 93 -CHAPTER 5INDUSTRIAL TRIALSIn order to elucidate the chemistry of the softener unit, as well as to complete the researchprogram on the oxygen probe, time was spent in the plant. For this purpose, visits to the smelterin Trail were planned with three major objectives in mind: investigation of softening processparameters, collection of specific data for process modelling, and testing of the oxygen probe in anindustrial environment. Three visits of up to several weeks took place in March, June, and October1994. Daily plant sampling at various locations in the circuit is routinely done and the subsequentX-ray assays constitute an important source of information to comprehend the overall chemistryof the softening circuit. Monthly tabulations of these plant assays were collected and constitutedthe original source ofdata for a mathematical model ofthe process. The details of the model togetherwith the data analysis are presented in Chapter 6. A number of short visits, of one or two days,were also paid to the smelter throughout the duration of the project. These short visits maintainedcontact with Cominco personnel, provided information about the modifications made to the softening process, and provided an opportunity to present the progress made in the project. In thisChapter, the first section covers the investigation of the softening process, including the datacollection, while the second section reports on the oxygen probe testing program in the softenerunit. In the last section, a thermal arrest technique that provides an instantaneous measurement ofthe impurity level of a lead bullion is described.- 94 -5.1 Softening Process Investigations5.1 Softening Process InvestigationsInvestigating the chemistry of a softening cycle consisted in assessing the composition of thebullion in the softener vessel at the beginning and end of the cycle, the composition and amount ofslag generated during the cycle, and the amount of oxygen injected during the cycle. A number ofpreliminary samples were collected in advance to establish the best approach to the monitoring ofa softening cycle. These tests, then, also provided the information required to find the mostappropriate timing and method for taking the various samples.5.1.1 Preliminary Sampling CampaignAs described in Chapter 1, during a pumping action, unsoftened lead bullion at 450°C isdrawn from the north pot into the softener, while partially softened lead bullion at about 630°Coverflows out of the softener back into the north pot. Simultaneously, slag is discharged onto avibrating cooling launder and collects in a 5-ton pot. A close-up of the softening circuit is givenin Figure 5.1 on which the various material flows in and out of the softener unit are indicated.Samples of input and output bullions could easily be obtained by scooping some bullion andquenching it in water. Slag samples were simply collected on the vibrating cooling launder. Sinceno difficulty arose in sampling the above material streams, their compositional evolution during apumping action was thoroughly examined. For this purpose, input and output bullion samples, aswell as slag samples, were taken every minute over the entire length of a pumping action whichlasts from 5 to 10 minutes under normal operating conditions. In an effort to obtain meaningfuldata, the sampling was repeated several times and carried out only on days when the softener wasrunning under normal operating conditions. The following observations were made.- 95 -C/i0 CD 0 C) CD (I)Figure5.1-Close-upofsoftenercircuit.BullionSamplerOxygenToSouthPotInputSampleNorthpotSoftenerSlagpot5.1 Softening Process Investigations• Characterization of input bullionThe As+Sb content of the input bullion from the north pot remains constant during the pumpingaction. This suggests that the north pot bullion is fairly homogeneous, or that any vertical composition gradient in the north pot does not affect the softener chemistry within the time frame of acycle. One input bullion sample should suffice to provide the composition of the unsoftened bullion.• Characterization of output bullionTheAs+Sb content ofthe outputbullion gradually increases by only about 0.1 wt% during apumpingaction, typically from 0.7 wt% up to 0.8 wt%. This corresponds to the effect of mixing unsoftenedbullion from the north pot into the softener vessel. From this observation, it was concluded thatvery strict timing should be followed when taking the bullion samples at the beginning and end ofa pumping action. Any delay would result in meaningless data.• Characterization of overflowing slagThe slag composition also remains constant during the period in which it is overflowing. A sampletaken at any time during the slag overflow will provide the slag composition corresponding to thepreceding cycle, assuming that the slag produced during a cycle has the same composition as theoverflowing slag. This hypothesis can be considered valid only when the softener has reached asteady state regime, that is to say, the cycles have been regular for at least 3 to 4 hours.The characteristics of the bullion and slag within the softener were also examined. In particular, the possibility that composition and temperature gradients develop within the vessel, bothhorizontally and vertically, has been explored. The dimensions of the softener vessel as of June- 97 -5.1 Softening Process Investigations1994 are given in Figure 5.2 to provide for a scale of the required sampling equipment. The vesselhas an inner diameter of 2.70 m for a depth of 1.80 m. The bath reaches a level of about 1.50 m,including a slag layer of 15 to 25 cm thick.Composition gradient in molten leadAssessing the presence of a compositional gradient along the depth of the vessel required that fiveor six samples be taken simultaneous along the vertical axis of the furnace. This apparently simpleconcept proved to be a rather challenging sampling exercise. However, a sampling device wasdesigned and fabricated with materials from the maintenance shop. It consisted of a 4 metre longsteel pipe made of tubing material normally used to fabricate the oxygen lances. A series of ironsample tubes, about 10 cm long, were clamped on the pipe at various levels. The opening of eachsampling tube was closed offwith several layers ofmasking tape. Tests were performed to determinethe number of layers required to provide a 30 seconds delay before lead bullion would collect inthe tubes. Such a delay was indispensable to permit the positioning of the device inside the vessel.Measurements were laboriously performed, the weight and length of the device involved combinedwith its high temperature making it difficult to handle, especially in times of high fume emissions.Two measurements were carried out on separate days. The assays of the first measurement did notreveal the existence of any composition gradient within the depth of the lead bath (see Table 5.1).A relative antimony depletion 75 cm below the slag surface, or 25 cm above the oxygen lances tip,and a copper and arsenic enrichment at about the same level were the only noticeable facts. Thissurprising discovery was confirmed by the second measurement. Even though the assays were notas consistent as in the case of the first measurement, no pronounced gradient was, however, detected(see Table 5.2). Two additional series of samples collected in the top 50 cm of the lead bullionagain confirmed the absence of a gradient (see Table 5.3).- 98 -5.1 Softening Process InvestigationsFigure 5.2 - Dimensions of the softener unit as of June 1994.Oxygen lancePot coverLead wellL1 = 2.lOm L2= 1.50m L3= O.30mL5L4= 2.80m L5= 1.80m L6= 2.70m- 99 -5.1 Softening Process InvestigationsComposition gradient in molten slagSlag samples were taken simultaneously or within 10 second intervals near the slag surface andclose to the slag-bullion interface at various locations in the vessel. Extreme care was put intotaking the samples, and it is assumed that no local mixing took place during the sampling. Theassays did not reveal any composition difference, indicating that the slag was homogeneous incomposition over the entire vessel. Assays of slag samples taken at opposite sides of the vessel aregiven in Table 5.4.Table 5.1 - Assays of softener bullion for vertical gradient assessment (measurement #1).Distance below slag surface Sb As Bi Cu(cm) (wt%) (wt%) (wt%) (wt%)15 0.79 0.09 0.23 0.0425 0.80 0.10 0.23 0.0550 0.81 0.23 0.20 0.2775 0.63 0.10 0.21 0.13100 0.79 0.08 0.23 0.05125 0.80 0.08 0.23 0.05Softener output bullion 0.73 0.08 0.23 0.06Slag sample 7.2 22.4 - --100-5.1 Softening Process InvestigationsTable 5.2 - Assays of softener bullion for vertical gradient assessment (measurement #2).Distance below slag surface Sb As Bi Cu(cm) (wt%) (wt%) (wt%) (wt%)25 0.90 0.27 0.44 0.0750 0.90 0.10 0.49 0.0675 0.85 0.20 0.45 0.44100 0.91 0.17 0.47 0.36125 0.99 0.25 0.48 0.52135 0.92 0.10 0.47 0.06Softener output bullion 0.92 0.11 0.48 0.08Slag sample 7.8 22.6 - -Table 5.3 - Assays of softener bullion for vertical gradient assessment (measurements #3 and #4).Distance below slag surface Sb As Bi Cu(cm) (wt%) (wt%) (wt%) (wt%)25 0.73 0.09 0.21 0.1035 0.73 0.09 0.20 0.0945 0.76 0.09 0.18 0.0925 0.72 0.10 0.20 0.0935 0.70 0.08 0.19 0.0845 0.70 0.09 0.20 0.08- 101 -5.1 Softening Process InvestigationsTable 5.4 - Assays of softener slag from opposite sides of the vessel for vertical gradientassessment (measurements #5 and #6).Distance below slag surface Sb As Pb(cm) (wt%) (wt%) (wt%)0 19.9 9.3 5910 20.1 9.3 5915 20.1 9.2 590 20.0 9.3 5915 19.8 9.0 58• Temperature gradient in the bath (slag and bullion)Similar to the composition gradient measurement, assessing the presence of a temperature gradientwithin the depth of the vessel required the fabrication of a special device to measure the temperatureat six different levels simultaneously. A “temperature probe” was made by Cominco’s workshopand consisted of six K-type thermocouples inserted in and making contact with a steel pipe closedat one end. Numerous problems were encountered when attempting to use the device in the softenerbath. After several trials, a recording was performed over more than 2 hours with only one thermocouple functioning improperly. The remaining five thermocouples recorded temperatures within5 to 10°C of each other, indicating that no pronounced thermal gradient was established in the bath.When other measurements were carried out on following days, none of the thermocouples seemedto function properly any longer. It has been suggested that the thermocouples were no longer incontact with the pipe giving a meaningless measurement.- 102 -5.1 Softening Process InvestigationsMonitoring of slag-bullion interface temperatureA thermocouple similar to the type used to control the softener, was immersed in the bath to measurethe slag-bullion interface temperature during a cycle. The measurements showed that the interfacetemperature was very close to the controlling temperature, which is measured deeper in the vessel,confirming that no temperature gradient is established during normal operation, at least in the tophalf of the bath.No precise data on the amount of oxygen injected in the lead bath was available at the timeof the short visits, and only an estimate of the “average normal injection rate” was obtained.Determining the amount ofoxygen injected in the lead bath constituted one of the key measurementsto be performed during the extended trials. Two alternatives were possible: the use of rotametersthat normally equip each lance, or a vortex meter installed on the main oxygen line. The firstalternative was hindered by the fact that some lances had defective rotameters and long delays wereexpected before they could be replaced. The second alternative benefited from the availability ofvortex meters on site. The advantage of a vortex meter is that it provides a total oxygen injectionrate instantaneously. Its dependence on the gas pressure in the main delivery pipe is, however, amajor drawback. Any deviation of the gas pressure from the calibration pressure results in anerroneous oxygen measurement.5.1.2 Softening Chemistry and Oxygen EfficiencyWhether the injected oxygen equilibrates with the bath or kinetic processes control thesoftening rate is a critical question to answer in understanding softener chemistry. Important- 103 -5.1 Softening Process Investigationsinformation on this issue can be obtained by assessing the efficiency with which oxygen is utilizedin the process. First, a sampling procedure for investigating the softening process chemistry wasdevised from the information provided by all preliminary tests. The procedure, which correspondsto a succession of steps carried out during each pumping action for the entire length of the test, isgiven below in chronological order.• first sample of output bullion taken 1 minute after the pump is turned on (provides the finalcomposition of the previous cycle)• sample of input bullion• sample of slag• second sample of output bullion taken 1 minute after the pump is turned off (provides theinitial composition of the following cycle)The total oxygen injection rate and the slag-bullion interface temperature were continuouslymeasured for the entire duration of the test by a vortex meter and a thermocouple respectively.A first test was carried out in June 1994 using this sampling procedure for a duration of sixcycles at a time when the softener had been operating at three cycles per hour for more than fourconsecutive hours. In such circumstances, the slag collected on the launder could be consideredrepresentative of the slag generated during a cycle. Unfortunately fume generation was unusuallyhigh during this period. The oxygen injection rate was measured by a vortex meter at an averageof 73.0Nm3/hour. The slag-bullion interface temperature was monitored and evolved in a cyclicmode between 602°C and 626°C. The slag pot was weighed by means of a crane scale prior to andafter the sampling test. The slag produced during the test amounted to a total of 1451 kg, corre- 104 -5.1 Softening Process Investigationssponding to an average of about 242 kg per cycle. Expressed in terms of oxides, the average slagcomposition was the following: 29.1 wt% Sb015, 9.1 wt% As015, 61.2 wt% PbO and 0.5 wt%Sn02,with Sb01,As01,PbO and 5n02totaling 99.9 % of the slag weight. The oxygen efficiency,calculated as the amount of oxygen utilized in slag formation divided by the total amount of oxygeninjected, was found to be about 80%. Fuming, unusually high, should be accounted for in order toprovide a proper calculation of the oxygen consumption. Unfortunately, the amount and composition could not be assessed. The process parameters and average assays of slag and bullion samplesare given in Table 5.5. The complete assays for all six cycles are given in Appendix 2.Using a crane scale and weighing the large slag pot was not, in retrospect, the most appropriatemethod to accurately determine the amount of slag produced during the test. Any large inaccuracyin the slag weight measurement could lead to a substantial uncertainty in the oxygen efficiencycalculation. Moreover, it is difficult to determine the effect of fuming on the calculation of oxygenefficiency. A second test, three days later, was unfortunately aborted after sampling three cyclesdue to operating trouble in the circuit.A new test was not performed until October 1994 when certain improvements were made inthe sampling procedure. All lances were, by then, equipped with brand new rotameters, providinga better measurement of the oxygen injection rate than the vortex meter alone. Due to the variationsfrom day to day in the oxygen pressure in the main delivery line the vortex meter was reliable onlyon days when the delivery pressure was identical to the calibration pressure. The slag was collectedin barrels during the test rather than into the usual 5-ton pot. In this manner, the barrels could behandled without the need for a plant operator to maneuver the crane, and also be weighed with acalibrated plate scale. This method is believed to provide better accuracy in measuring the amount- 105 -5.1 Softening Process Investigationsof slag produced. A total of three tests were attempted, only the third one was carried to completionwithout any problems. A steady state regime had prevailed for more than 5 hours. The test waslimited to three cycles rather than six, to minimize the chances of interruption, and lasted 54 minutes.The oxygen injection rate was set at 74.8Nm3/hour. The temperature recorded by the controllingthermocouple varied between 592°C and 614°C, which is about 10 degrees lower than in the testperformed in June. A total of 784 kg of slag, corresponding to an average of 261 kg per cycle, wascollected with the following average composition: 22.5 wt% Sb015, 11.2 wt% As015, 64.6 wt%PbO, and 0.9 wt% Sn02,the sum ofSb01,As01,PbO and Sn02totaling 99.2% of the slag weight.The oxygen efficiency was calculated to be about 92%. Fuming, which did not appear to be anyhigher than “normal”, was not accounted for in the calculation. The process parameters and averageassays of slag and bullion samples are summarized in Table 5.6. Complete assays for all threecycles are given in Appendix ConclusionsThe main purpose of the preliminary sampling campaign was to characterize the varioussoftener streams and develop a procedure to investigate the process. The major outcome can besummarized as follows:• No composition gradient (both vertical and horizontal) has been observed in the slag.• No composition gradient has been observed within the depth of the bullion.• No vertical temperature gradient has been observed in the bullion (up to the slag-bullioninterface) under normal operating conditions.Another important result from this preliminary campaign was to define so called “normal operatingconditions” as follows: 3 to 4 cycles per hour, regular temperature cycle within the control range,- 106 -5.1 Softening Process Investigationsbullion composition within the desired target range, minimum lance consumption, and minimumfuming. When the process is running under the normal operating conditions as described above,very few bubbles break the surface, suggesting a high oxygen efficiency is achieved. The majorresult from the softener investigations is the confirmation that the oxygen efficiency is high, in theorder of 90%. It can then be assumed that under these operating conditions, kinetic constraintshave only a limited influence on the process. To summarize the major findings of the investigations,the softening process chemistry is characterized by the following parameters:• initial bullion composition: 0.75 to 0.85 wt% As+Sb• final bullion composition: 0.65 to 0.75 wt% As+Sb• slag composition: 35 to 40 wt%As015+5b• slag production: about 250 kg/cycle• slag layer thickness: 20 to 25 cm• oxygen injection rate: 68 to 76 Nm3/h• oxygen efficiency: about 90% (fume formation not included)• temperature cycle: within the range 600 to 630°CTo maintain the steady state regime as described above, good maintenance is essential. However,goodmaintenance does not provide control over input bullion composition, and dramatic alterationsin the process regime are mainly due to composition changes. Such changes need to be detectedin order to adjust the process parameters accordingly. Continuous monitoring of the bullion impuritycontent is a potential solution that would give a warning of any deviation from11normal operatingconditions’t. The following section reports on industrial tests with oxygen probes to determinewhether composition control could be achieved by monitoring the oxygen potential of softenerbullion.- 107 -5.1 Softening Process InvestigationsTable 5.5 - Process parameters and assays of softener slag and bullion for softener chemistryassessment (test #1, June 1994).Process Parameters ValuesNumber of cycles sampled 6Total sampling time 120 mmOxygen injection rate 73 Nm/hTemperature cycle 602°C - 626°CTotal slag weight 1451 kgAverage slag amount per cycle 242 kgAverage initial bullion composition 0.76 wt% Sb0.07 wt% As0.31 wt% Bi0.11 wt% Cu0.01 wt% SnAverage final bullion composition 0.69 wt% Sb0.06 wt% As0.31 wt% Bi0.11 wt% Cu0.01 wt% SnAverage slag composition 29.1 wt% Sb0159.1 wt% As01561.2 wt% PbO0.5 wt% Sn02Oxygen utilization 80% (calculated)- 108 -5.1 Softening Process InvestigationsTable 5.6 - Process parameters and assays of softener slag and bullion for softener chemistryassessment (test #2, October 1994).Process Parameters ValuesNumber of cycles sampled 3Total sampling time 54 mmOxygen injection rate 74.8 Nm3/hTemperature cycle 592°C - 614°CTotal slag weight 784 kgAverage slag amount per cycle 261 kgAverage initial bullion composition 0.67 wt% Sb0.09 wt% As0.25 wt% Bi0.07 wt% Cu0.01 wt% SnAverage final bullion composition 0.61 wt% Sb0.08 wt% As0.26 wt% Bi0.08 wt% Cu0.01 wt% SnAverage slag composition 22.5 wt% SbO1511.2 wt% AsO1564.6 wt% PbO0.9 wt% Sn02Oxygen utilization 92% (calculated)- 109 -5.2 Oxygen Probe Plant Tests5.2 Oxygen Probe Plant TestsA number of oxygen probes were tested in the plant during the October 1994 visit to thesmelter in Trail. At that time, a number of operating problems occurred and perturbed the normaloperation of the softener in such a way that the combined arsenic and antimony level in the vesselvaried from as low as 0.4 wt% to as high as 1.9 wt%. These peculiar operating conditions hadmixed repercussions on the oxygen probe testing program. On one hand, this situation was beneficialsince it permitted investigating the probe response to a wide range of lead bullion compositions.On the other hand, at times of low antimony and arsenic concentrations, oxidation by contact withthe atmosphere was very fast, and large amounts of oxide sludge were generated on the surface ofthe bullion inside the lead well (bullion outlet as shown in Figure 5.2). Obtaining proper measurements was difficult because of sludge adhesion onto the probe within an hour after immersion.This section covers the various aspects of the oxygen probe testing program, describing thedifficulties encountered and the solutions used to overcome them. The results are discussed and acalibration curve of the measured emf as a function of the combined As+Sb content is given. Toconclude, some recommendations for improving the probe design to better suit the industrialenvironment are discussed.5.2.1 Oxygen Potential MeasurementsThe oxygen potential measurements were done in the lead well at the lead bullion outlet.Simultaneous temperature measurements were performed with a K-type thermocouple inserted intoa closed end alumina tube. The oxygen probe was connected to a Corning 125 millivolt and pHmeter similar to the instrument used in the laboratory. Both emf and temperature were recordedwith a chart recorder with a digital panel for quick reading.- 110 -5.2 Oxygen Probe Plant TestsBreakage of the zirconia tube upon immersion and molten lead infiltration through the probethread were the two major problems encountered in the early stages of the testing campaign. Thetemperature difference between the atmosphere just above the bath surface and the lead bullionwas measured to be about 300°C, a differential twice as large as the maximum tolerable by zirconia.Slowly lowering the probe close to the bath surface and leaving it there for a few minutes to allowthermal equilibration would therefore not be sufficient to avoid breakage upon immersion due tothermal shock. To remedy this problem, a piece of steel tube about 10 cm long was attached to thetip of the probe holding tubing. The role of this “extension tube” was to make contact with the leadbullion before the probe itself so that the temperature of the probe tip would slowly increase byheat conduction. This worked on all but two occasions. Ineffective conduction due to poor metalcontact is the most likely reason for the failures rather than the possible presence of defects in thezirconia tubes. Once a probe had survived immersion, the possibility of a zero millivolt readingwas the next problem. Such a reading is typically due to molten lead infiltration through the probethread and short-circuiting of the lead wire and conducting lead. The only way to avoid this problemwas to make sure that the probe tip was not immersed too deep into the lead bullion thus avoidinghigh pressures in the liquid lead.Another factor that determined the length of time that a probe would last in the bullion wasthe problem of a “sludge” coating. As mentioned earlier, at times of low antimony and arseniccontent, an oxidic sludge formed on the surface of the lead bullion and gathered around anyinstrument present in the lead well. This sludge, accumulating with time and slowly creeping downto the probe tip, ended up covering the probe completely. As a result, a constant emf reading wasobtained which corresponded to the oxygen potential of the sludge layer directly in contact withthe zirconia. This phenomenon could take place within an hour after immersion thus considerably- 111 -5.2 Oxygen Probe Plant Testslimiting the duration of a meaningful test. Even though the probe was still functioning and couldbe removed, cleaned, and reused, it did not fulfill its role of a continuous measuring device. It isworth mentioning that such operating conditions are not normal but merely the results of operatingtroubles that might easily be avoided if a probe was used to control the system.5.2.2 Results and DiscussionThe emfmeasurements successfully obtainedwith five differentprobes on various days duringthe trials and for which a bullion assay is available are given in Table 5.7. The corresponding leadbullion temperatures and assays are also given in the same Table. The range of measured emf,between 178 mV and 312 mV, corresponded to a total impurity content ranging from 0.97 wt% to2.41 wt%. As seen in Table 5.7, the measurements were not carried out at constant temperature,but actually within the range 600°C to 623°C. These relatively small temperature differences donot have a large influence on the measured emf. In the laboratory experiments with oxygen saturatedlead bullion, a temperature change of 10°C corresponded to an emf change of 1.4 mV. A typicalrecording of emf and temperature is given in Figure 5.3. The probe response was qualitativelyconsistent with theNernst equation: an increase oftemperature and/or a decrease in impurity contentgenerated an emf decrease. However, a quantitative analysis of the probe response would requirethe measurement of both temperature and bullion composition changes as well as an understandingof the fluid flow and mixing in the vessel. The velocity of the output bullion stream, which variedfrom very high at the start of a softening cycle to almost nil at the end, appeared to have a temporaryeffect on the emf readings as marked by the discontinuity of the emf trace on Figure 5.3. Themeasured emf given in Table 5.7 correspond to the stable emf obtained at low bullion velocitytowards the end of a cycle and read on Figure 5.3 just prior to the discontinuities.- 112 -ETemperature_______ZZ___-____________________3_______::________600——-____—____—-—-..—-----———0-—-—--—_______-—--E”EEEmf______________—-___.—-.____—___-200mI,---____-_______3_______:1201059075604530150Time,minutesFigure5.3-Typicalrecordingofemfandtemperaturemeasuredinthesoftenervessel.5.2 Oxygen Probe Plant TestsThe emf data from both plant and laboratory tests (see Tables 5.7 and 4.6 respectively) wereplotted versus the total impurity content. As can be seen in Figure 5.4, these data show noticeablescatter. On the other hand, when the measured emf was plotted versus the combined As+Sb, abetter correlation was obtained as shown in Figure 5.5. This demonstrates that the impurities otherthan arsenic and antimony, i.e. bismuth, tin, silver, and copper, do not participate in the oxygenequilibria. Thus, the emf reading is almost solely dependent on the As+Sb content of the bullion.The correlation between measured emf and As+Sb content was determined by a least-squareregression analysis as follows(mV)meas = 133.5 + 95.65 (wt%)AS+Sb (5.1)where (mV) meas is the measured emf in mV, and (wt%) As-i-Sb is the combined As+Sb content in wt%obtained from the bullion samples assays. The intercept of the above equation gives an emf valuefor the “zero impurity content” of 133.5 mV. This value compares well with the theoretical valueof 135.9 mV calculated with Equation (4.4) at 613 °C, the average measured temperature from Table5.7. The good correlation between measured emf and As+Sb content given by Equation (5.1) canbe rearranged as follows(wt%)AS+Sb = - 1.396 + 10.45 x i0 (mV)meas ± 0.05 (wt%) (5.2)This relationship provides the necessary calibration curve that will permit the use of the probe inthe plant.- 114 -Figure 5.4 - Plot of plant and laboratory emf measurements versus total impurity content of leadbullion. The solid line corresponds to a least-square analysis.5.2 Oxygen Probe Plant Tests350300250200150I1000 1 2Total Impurity Content, wt%3- 115 -3503002502001500As+Sb Content, wt%5.2 Oxygen Probe Plant TestsFigure 5.5 - Plot of plant and laboratory emf measurements versus As+Sb content of lead bullion.The solid line corresponds to a least-square analysis.I1001 2 3- 116 -5.2 Oxygen Probe Plant TestsTable 5.7 - Measured emf in softener vessel and corresponding lead bullion assays.Measured emf Temperature As + Sb Total impurity As + Sb(mV) (°C) (wt%) (wt%) Total impurity196 623 0.52 1.08 0.48186 620 0.49 1.02 0.48187 617 0.54 1.10 0.49190 620 0.59 1.17 0.50191 617 0.59 1.14 0.52185 600 0.50 1.09 0.46178 614 0.46 1.05 0.44179 614 0.49 1.07 0.46180 614 0.50 1.10 0.45185 612 0.54 1.16 0.46192 612 0.54 1.14 0.47192 600 0.53 0.97 0.55296 618 1.79 2.35 0.76312 608 1.87 2.41 0.78258 610 1.23 1.70 0.72- 117 -5.2 Oxygen Probe Plant Tests5.2.3 ConclusionsThe oxygen probe, which was designed and tested in laboratory, was successfully used in anindustrial environment. One of the key achievements of the program was determining a correlationbetween measured emf and combined As+Sb content of the softener bullion. Out of ten probesbuilt for the industrial tests, five provided satisfactory measurements for various length of timedepending on the conditions prevailing in the vessel. Two probes continuously monitored theoxygen potential of the bullion for more than a day, thus proving the potential of the design. Inmost cases, the reason for interrupting a test was the fact that the tip of the probe was coated withan oxidic sludge. However, a probe with such a coating could possibly be removed from the melt,cooled down, cleaned, and reused immediately or later. The fact that a number of probes failedupon immersion or shortly after suggest that some improvements in the “laboratory design” of theprobe should be implemented in order to upgrade the probe to an “industrial design”. In particular,failure due to thermal shock is the most critical issue. Completely eliminating the problem is notrealistic since zirconia tubes can have defects that become points of weakness upon heating.However, the failures can be reduced by ensuring that the probe is preheated to a temperature closeto that prevailing in the bath. This could be achieved by integrating the “extension tube” conceptused in the plant into the probe design. An alternative would be to slowly preheat the probe to thedesired temperature by means of a small furnace prior to immersion. Slag and sludge protectionis another critical issue to be dealt with. The probe tip could be enclosed in a protective steel cupthat would stop the sludge from creeping down along the probe holding tube and prevent it fromcoating the zirconia (see Figure 5.6). As for the lead infiltration problem, the attachment betweenthe probe tip and the probe holder could be modified in such a way that the thread would not bebelow the liquid surface, therefore eliminating the risk of infiltrations.- 118 -5.2 Oxygen Probe Plant TestsSludge accumulationProtective steel capFigure 5.6 - Schematic representation of a probe tip with a steel cup for sludge protection.Oxygen probe- 119 -5.3 Thermal Arrest Technique5.3 Thermal Arrest TechniqueOne of the major concerns about the use of an oxygen probe to continuously monitor thebullion composition is the possibility of erroneous emf readings. In a situation where the probehas failed, unless its reading is suddenly strongly off-range, the malfunction would not be detecteduntil routine assay results were available, which can take several hours. A quicker assessment ofthe state of the probe would be extremely valuable. The thermal arrest technique provides such aninstantaneous check. The Quick-Cup system from Heraeus Electro-Nite was chosen, tested andcalibrated for that purpose. A schematic description of the apparatus is given in Figure 5.7. Theprinciple of the technique is based on the fact that the liquidus temperature of a metallic alloy iscomposition dependent, the higher the alloying element(s) content, the lower the liquidus temperature. Thus, the plotting of the temperature versus the time of cooling yields a time-temperaturecooling curve. Quenched bullion samples of various composition from the plant were remelted inthe laboratory and their cooling curves recorded. The change of slope in the curve indicates theliquidus temperature of the sample. For pure lead, a well defined plateau is obtained, as shown inFigure 5.8. For samples containing more than 1 wt% impurity, the change of slope in the curve isusually abrupt without a well defined plateau, as shown in Figure 5.9. A plot of liquidus temperatureversus total impurity content is given in Figure 5.10. A regression analysis yielded the followingrelationship(°C) Liquidus = 324.4 — 8.2 (wt%)ToIjmpu, (5.3)where (°C) Liquidus is the liquidus temperature in Celsius, and (wt%) Total impurity is the total impuritycontent of the bullion. The obtained value of 324.4°C instead of 327.2°C for the melting point ofpure lead shows that the technique still requires some fine tuning.- 120 -Figure 5.7 - Thermal arrest apparatus.5.3 Thermal Arrest TechniqueEXP-16multiplexerTIC leadsMolten leadQuick-CupQuick-Cup stand/_ ___\Personal computerwith DAS-8 board- 121 -5.3 Thermal Arrest Technique350340330320C)o 310ci-)- 300ci)290280270260250 I I I0 1 2 3 4 5Time, mm.Figure 5.8 - Cooling curve of pure lead.- 122 -5.3 Thermal Arrest Technique350 I I340330320C-)o 310ci)z- 300ci)290280270260250 I0 1 2 3 4Time, mm.Figure 5.9 - Cooling curve of “North Pot” bullion containing 2 wt% total impurity.- 123 -3305.3 Thermal Arrest TechniqueTotal Impurity Content, wtFigure 5.10 - Calibration curve for thermal arrest technique.C)0ci)ci-)Sci)H(12320310300290\\c\\\I I I I I I I I I I0 1 2 3 4- 1245.3 Thermal Arrest TechniqueAs opposed to the insensitivity of dissolved oxygen to impurities other than As and Sb, eachimpurity element has a non negligible effect on the liquidus temperature drop of the lead bullion.Temperature drops of 6.8°C, 14.0°C, 9.5°C and 3.7°C for each one wt% of Sb, As, Ag, and Bi,respectively, have been estimated from phase diagrams. For that reason, it is not possible to directlyassess the As+Sb content of the softener bullion with the thermal arrest technique. However, if acorrelation between As+Sb content and total impurity content was established, an indirectassessment would be possible. A total of 130 softener bullion samples were examined and a plotis given in Figure 5.11. A least-square regression analysis yielded the following equation(wt%) As+Sb = - 0.473 + 0.964. (wt%) Total impurity (5.4)where (wt%) As+Sb and (wt%) total impurity are the softener bullion As+Sb content in wt%, and the totalimpurity content in wt%, respectively. By combining Equations (5.2) and (5.3) and rearranging,the following correlation between liquidus temperature and As+Sb level in the softener bullion isobtained(wt%) As+Sb = — 0.117 (°C) Liquidus ± 0.15 (wt%) (5.5)The characteristics of the softener bullion, that is to say both the absolute and relative amountsof the various impurities, have evolved since the start of this research project. For example, therelative amount of bismuth has regularly increased since 1992. In order to consider the long termchanges in the feed bullion composition, the relationship corresponding to Equation (5.4) shouldbe updated on a regular basis by analyzing the current daily assays of the softener bullion.- 125 -5.3 Thermal Arrest TechniqueTotal Impurity Content, wt3.0Figure 5.11 - Plot of As+Sb content of output bullion versus total impurity content.2.01.5Q)1.00U-DU)+C120.50.00.5 1.0 1.5 2.0 2.5- 126 -CHAPTER 6PROCESS MODELLINGANDCONTROL STRATEGYAssessing a control strategy for an improved softening process required the understandingof the process fundamentals, and, in particular, the process chemistry. The collection of industrialdata on Cominco’s current softener, as reported in Chapter 5, provided the necessary basis to carryout an analysis of the process. A thermodynamic model was developed to serve as a processinvestigation tool. This model was validated by performing an analysis of operating plant data.Once it was ascertained that the thermodynamic model gave a reasonably good representation ofthe batch process operation, and that an equilibrium analysis gave a good fit of operating data, themodel was modified to simulate continuous softening. In this manner, the option of a continuous,single pass process could be examined, and a preliminary control strategy assessed. First, all aspectsof the model formulation, and in particular the assumptions and the solution modelling, are covered.Then, the model validation and calibration is presented. Finally, the continuous single pass softeningalternative is discussed, and a number of conclusions are drawn.- 127 -6.1 Model Formulation6.1 Model FormulationTo determine the equilibriumcomposition of the multiphase system representing the softeningprocess, it was decided to develop a simpler version of the free energy minimization algorithm thanthe previously mentioned SOLGASMJX routine. In this manner, changing and adapting the modelto perform various process simulations, i.e. batch, semi-batch or continuous modes for example,would be easily and quickly done. In this Section, the model assumptions are first presented, thena description of the mathematical algorithm is given, and finally, the solution modelling of thechemical system representing the lead softening process is discussed.6.1.1 AssumptionsAs a first step in the analysis of the softening process it was assumed that the process operatesat thermodynamic equilibrium. Although this is not strictly correct because the oxygen efficiencywas measured in the plant to be about 90%, it still provides a basis on which to understand the basicreactions taking place in the process. A thermodynamic process model essentially assumes that allreactions are instantaneous and that the degree of reaction is controlled by thermodynamics alone.Daily plant assays of Cominco’ s softener bullion consider up to seven metallic elements, i.e. lead,antimony, arsenic, bismuth, tin, copper, and silver. In order to simplify the computations, thechemical system chosen to represent the process has been limited to the four elements involved inthe main reactions, i.e Pb, Sb, As, and 0. Thus, the molar balance considers four species in thebullion, Pb(l) and three dissolved elements, i (Bullion)’ (Bullion)’ and Q (Bullion)’ and three speciesin the slag, PbO, Sb015 andAs015. The equilibrium reactions and molar balance equations whichdescribe the above system are given in Tables 6.1 and 6.2.- 128 -6.1 Model FormulationTable 6.1 - Equilibrium reactions representing the oxygen softening process.Equilibrium reactions 1/2 02 (g) > (Bullion)Pb + 1/2 02 (g) ‘E—> PbO (Slag)Sb(1) + 3/4 °2(g) <> SbO15 (Slag)As (I) + 3/4 °2(g) As01,5(Slag)Table 6.2 - Molar balance equations of the chemical system representing the oxygensoftening process (‘n” stands for the number of moles).Elemental balances n Pb, Tot = ‘‘ Pb + n PbOSb, Tot = (Bullion)+ n Sb01GAs, Tot M(Bullion)+ nASOl5O,T0t = + nPbO + 1.5nb015 + 1.51AsO5(Bullion)Phase balances n Bullion = n Pb + 0 + (Bullion) + As (Bull iou)(Bullion)n Slag PbO + ‘ SbO + ‘ AsO- 129 -6.1 Model FormulationOxygen gas does not appear in the balance because the equilibrium partial pressure in thesystem is well below one atmosphere. As a result, thermodynamics predicts all oxygen is consumedbelow the bath surface. The model operates by reacting a given amount of oxygen with the bullionand minimizing the free energy of the system by shifting the oxygen between PbO, Sb015As015,and Q (Bullion) The temperature is considered to be constant and equal to the slag-bullion interfacetemperature. It is also assumed that no fuming and no interactions with the gas phase take place.6.1.2 Computation of Equilibrium CompositionThe Gibbs free energy function of a multiphase system is defined as follows,Gystern = n1 (G + RTlna,) (6.1)phases specieswhere n , G , and are the number of moles, the standard Gibbs free energy, and theactivity of species i in phase j, respectively. Expressed in an expanded form, Equation (6.1) givesGsystem = n (G1 + RT1nX1,3+ RT lny) (6.2)phases specieswhere X1, and y are themole fraction and activity coefficient of species i in phasej, respectively.Following the assumption that only two phases are considered, the lead bullion and the liquid slag,the Gibbs free energy function for the system becomesGsystem 1’i, bullion ( Gjbullion + RT ln X, bullion + RT ln ‘y1, bullion )+ 1, slag ( G slag + RT ln X,, slag + RT ln y, slag ) (6.3)- 130 -6.1 Model FormulationThe reduced Gibbs free energy of the system, G,stem, defined as follows,G tosystem G i, bullionGsystem= RT = i, bullion 1\ RT + in X1, bullion + in 7i, bullion+ i, slag [ Gstag + in X, slag + ln 7 slag] (6.4)is minimized subject to two constraints: 1) the elemental balances, and 2) the equilibrium composition values determined must be positive.In the search for a minimum value of the Gibbs free energy function of a system, subject tothe mass balance relations, various mathematical techniques have been used. The SOLGASMIXprogram, for instance, uses Lagrange’s method of undetermined multipliers to set up a system oflinear equations that is then solved with a Gaussian elimination technique. The EQUIL program,written in FORTRAN IV and based on a technique presented by Rao871,uses a Newton-Raphsonroutine to achieve final convergence. The algorithm developed for this project was inspired byRao’s technique. Simplified flow diagrams of the whole algorithm and the minimization routineare given in Figures 6.1 and 6.2. As in the EQUIL program, an estimate of the minimum thatsatisfies the mass balance relations is first determined, but the final value of the minimum is thenobtained in a manner that only requires the computation of the first derivative of the reduced freeenergy function. The Newton-Raphson routine of the EQUIL program requires the computationof both first and second order derivatives. The minimum of the G stem function is mathematicallydefined by- 131 -Figure 6.1 - Simplified flow diagram of the whole algorithm.6.1 Model Formulation- 132 -Figure 6.2 - Simplified flow diagram of the minimization routine.6.1 Model FormulationDo Ito MaxI(EachIndeendentVanabIeFmd Upper and Lower Limits(constraint 2)1 toMaCompute First Approximation of MinimumLas Middle = Upper Limit - Lower LimitCompute Number of Moles of All Species- 133 -6.1 Model Formulation*system= o (6.5)‘k )k=1,2 Nwhere N is the total number of species in the system (bullion and slag). Although the derivativeof the G stem function must be equal to zero at the minimum, in practice a zero value is very seldomattained. An accuracy limit of 10 was therefore chosen for the computations.The algorithm has been designed so that additional elements, species or phases can easily beimplemented provided the corresponding thermodynamic data required in the computations areavailable. However, the molar balance is limited to a maximum of 30 species altogether.6.1.3 Solution ModellingIt is apparent from Equations (6.2) to (6.4) that the Gibbs free energy minimization algorithmrequires the knowledge of two thermodynamic quantities for each species, the standard Gibbs freeenergy, G, and the activity coefficient, ‘y,. The standard states were chosen as follows: pureliquid for lead, antimony and arsenic, pure gas for oxygen, pure solid for lead oxide and antimonyoxide, and pure liquid for arsenic oxide. Expressions for the temperature dependence of the standardGibbs free energies of all species were derived using data from Barin and Knacke851,Pankratz881,andBarin891. These expressions are given in Table 6.3.- 134 -6.1 Model FormulationTable 6.3 - Expressions for the temperature dependence of the standard Gibbs free energies ofall species.Standard Gibbs Free Energy(J I mole)= -4251.0 + 125.405T -29.455T1nT= 8054.1 + 147.516T - 32.38 T1nTG = 11862.9 + 144.823T - 29.156T1nT - 0.002T2Gg = - 9679.0 - 2.204 T - 29.956 T in T - 0.002 T2 + 83680 T= -233580.9 + 241.37T - 44.819T1nTGbo = -364331.2 + 212.573T - 39.957TlnTG0 = - 344661.7 + 447.448 T - 76.358 T in T- 135 -6.1 Model FormulationThe activity coefficients of the bullion species have been derived from published data. Leadis assumed to obey Raoult’s law, consequently YPb = 1. The details relevant to the solutionmodelling of the dissolved elements are given below.OxygenOxygen is considered to form an infinitely dilute solution in liquid lead. For an infinitely dilutesolution, the basic equation for the activity coefficient of a solute isln’y1 = iny + (6.6)where ‘? and are the activity coefficient ofsolute i at infinite dilution, and interaction coefficientof solute j on solute i, respectively. Taskinen511measured the oxygen activities in lead and dilutelead alloys in the temperature range from 762°C to 1000°C, and derived expressions for the temperature dependence of the activity coefficient as well as self-interaction coefficient of oxygen atinfinite dilution in lead as follows14040my0 = 6.133— T(6.7)andag = 52.0 — 70900 (6.8)It is assumed that the above expressions for in yg and a g are valid in the temperature rangecharacteristic of lead softening, i.e. between 600°C and 630°C, and up to saturation. This hypothesiswill be discussed in the next section. Due to the scarcity of available and reliable interaction- 136 -6.1 Model Formulationcoefficients for antimony and arsenic, only the two above coefficients are used for the solutionmodelling of oxygen.ArsenicItagaki et al. [90] investigated the liquid Pb-As system in the temperature range between 464°C and582°C. They showed that a slight positive deviation from Raoult’ s law is observed for dilutesolutions of arsenic in liquid lead when the arsenic atom fraction is less than 0.15 at a temperatureof 464°C. However, this deviation becomes so small at 582°C that arsenic is considered to forman ideal solution in liquid lead, as suggested by Davey’.AntimonyAntimony is considered to obey Henry’s law over the composition range characteristic of leadsoftening. Extrapolating the data from Hultgren et al. [911, and Seltz and DeWitt921, the followingexpression for the temperature dependence of the activity coefficient of antimony dissolved in leadwas derivedlny = -0.0027- 219.1(6.9)This expression provides data in agreement with results from Zunkel and Larson931.The activity relations in the slag constitute the greatest uncertainty due to the scarcity of datain the literature. Values of the activity coefficients of the slag species have been obtained by fittingof the model to industrial data. These values were then compared to the limited data available tocheck their validity. The procedure is detailed in the next section.- 137 -6.2 Model Validation and Calibration6.2 Model Validation and CalibrationPrior to using the model to investigate the concept of continuous single pass softening, anumber of preliminary steps were performed in order to demonstrate that a thermodynamic modelcan be used to represent process operation. In particular, the validity of the solution modelling wastested, and the batch process currently used at Cominco was analyzed. These two aspects arepresented in this section.6.2.1 Solution Modelling ValidationAs a first step towards validating the model, the system Pb(l)- °Pb - PbO(s, yellow) wasconsidered, and the oxygen saturation in molten lead was computed with the model in the temperature range from 527°C to 827°C. The results are presented in Table 6.4. The oxygen saturationdata computed with the relationshipln (at% 0)- 12065.55+ 10.477 (6.10)fromAlcock andBelford44,are also given in Table 6.4 for comparison. A better agreement betweenthe two sets of values is obtained in the temperature range 577°C to 677°C. Even though theexpressions for in °yg and were determined by Taskinen511 in the temperature range between762°C and 1000°C, they appeared to be suitable for the temperature conditions of lead softening,and thus their use in the model is justified.- 138 -6.2 Model Validation and CalibrationTable 6.4 - Oxygen saturation in liquid lead at different temperatures. Comparison between themodel computations and the relationship from Alcock and Belford (1964, ref. 44).Oxygen Saturation(at%)Temperature From Equation (6.10) From Model(°C)527 1.003 x 10” 0.957 x577 0.243 x 0.239 x i0615 0.447 x i0 0.448 x i0627 0.536x10 0.540x103677 O.108x102 0.113x102727 0.205 x 102 0.220 x 102777 0.363x102 0.408x102827 0.613x102 0.723x106.2.2 Model CalibrationIndustrial data collected at the Cominco smelter were used to determine the activity coefficients of the slag species by fitting them to the model. This was achieved by running the simulationfor a given amount of oxygen consumed and changing the values of the coefficients until the finalamount and composition of the slag generated in the simulation match the industrial data. The plantdata collected in June and October 1994, and presented earlier in Tables 5.5 and 5.6, were used as- 139 -6.2 Model Validation and Calibrationprocess parameters in the simulations. The “June test” data were gathered by monitoring 6 softeningcycles for a total duration of 2 hours, whereas the “October test” data were obtained by monitoring3 softening cycles for a total duration of 54 minutes. The slag composition was normalized for theternary PbO-Sb015-As0 system. The input (plant data) and output (calculations) processparameters are summarized in Tables 6.5 and 6.6. The activity coefficient of PbO was obtainedusing data from Sugimoto et al.941 and preliminary results from Chaskar95. Sugimoto et a!.measured the activity of PbO at 800°C for the whole range of composition in the binary systemPbO-Sb015 using an emf technique. Chaskar carried out a study of the ternary system PbOSb015-As using a Knudsen cell technique. The activity coefficients of Sb015 and As015 werethen fitted using the above procedure. Their values are presented in Table 6.7.With the above procedure, from a set of activity coefficients including the two fitted values,the mole fractions of all species are calculated by minimizing the free energy of the system. Theactivity of each species can subsequently be computed from the knowledge of its mole fraction andactivity coefficient. Once the activities are known, the validity of the fitted activity coefficientscan be verified in the following manner. The equilibrium constants of the equilibria3/2 PbO (s) + Sb f4 3/2 Pb (1) + SbO15 (s) (I)and3/2 PbO(,) + As(1) <—> 312Pb() + As015() (II)can be calculated through the relationships- 140 -6.2 Model Validation and Calibrationa Sb01 a3/2K1 =3/2 (6.11)a Sb a PbOandaMOK11= 3 (6.12)aMa rcOusing the activities obtained in the simulations. At this point, it should be emphasized again thatthese activities were not obtained with Equations (6.11) and (6.12) but through the computation ofthe mole fractions of all species as described above. The simulation results using the “June test”data yield values of 303.8 and 82.9 whereas the results using the “October test” data yield valuesof 303.5 and 82.9 for K1 and K11 , respectively.The equilibrium constants K1 and K11 can also be computed using standard Gibbs freeenergy data. This alternative provides values of 303.0 and 80.1 for K1 and K11 ,respectively. Thisvery good agreement between the two sets of values demonstrate that the activity coefficientsobtained with the fitting procedure provide self consistent results in the model calculations.The final bullion composition obtained in the simulations also compares very well to theoperation data, as presented in Tables 6.5 and 6.6, demonstrating that the thermodynamic modelcan give a reasonably good fit of operating data.- 141 -6.2 Model Validation and CalibrationTable 6.5 - Batch softening model parameters (June plant data and corresponding model calculations).Parameters Plant Data Model CalculationsInitial bullion compositionSb 0.76 wt%As 0.07 wt%Final bullion compositionSb 0.69 wt% 0.66 wt%As 0.06 wt% 0.04 wt%Normalized Slag compositionPbO 61.5 wt% 61.6 wt%Sb015 29.3 wt% 29.3 wt%As015 9.2wt% 9.lwt%Slag production 242 kg/cycle 239 kg/cycleTemperature 615°COxygen injection 73.0m3/hourOxygen efficiency 80%Vessel capacity 60 tonnesSoftening cycle length 20 mm.- 142 -6.2 Model Validation and CalibrationTable 6.6 - Batch softening model parameters (October plant data and corresponding modelcalculations).Parameters Plant Data Model CalculationsInitial bullion compositionSb 0.67 wt%As 0.09 wt%Final bullion compositionSb 0.61 wt% 0.59wt%As 0.08 wt% 0.06 wt%Normalized Slag compositionPbO 65.7wt% 65.7wt%SbO15 22.9 wt% 22.9 wt%As015 11.4wt% l1.4wt%Slag production 261 kg/cycle 259 kg/cycleTemperature 615°COxygen injection 74.8m3/hourOxygen efficiency 92%Vessel capacity 60 tonnesSoftening cycle length 18 mm.- 143 -6.2 Model Validation and CalibrationTable 6.7 - Batch softening model parameters.Activity Coefficients SourceJune Data October DataBullionPb 1 1Sb 0.78 0.78 91-92As 1 1 1,900 dependent on dependent on 51composition compositionSlagPbO 0.29 0.31 94-95Sb015 0.405 0.56 fittedAs015 0.030 0.04 1 fitted6.2.3 Analysis of the Batch ProcessOnce the activity relations in the slag were obtained, the model was used to analyze themechanism of softening in a fully batch mode and with continuous slag removal. The processparameters from the October test (see Table 6.6) and their corresponding activity coefficients (seeTable 6.7) were used in this analysis. In the initial stage of the process, injected oxygen is largelyabsorbed into the liquid lead with little slag formation. This effect is really only significant whenstarting with an oxygen free lead bullion, as done in the simulations. In the actual process, thisinitial stage is much shorter since the bullion is close to oxygen saturation. However, once thebullion is saturated, slag formation begins and relatively high rates of antimony and arsenic removalare achieved (see Figure 6.3). With time, as the concentration of antimony and arsenic in the bullion-144-6.2 Model Validation and Calibration85 I I-C\20- 80 -II v Continuous70 / Slag Removal•Batch65 I I0 5 10 15 20Softening Time, minutesFigure 6.3 - As+Sb removal during a softening cycle in batch and continuous slag removal modes(computed from model simulations).- 145 -6.2 Model Validation and Calibrationdrops, the softening efficiency also falls and more lead is oxidized to PbO (see Figure 6.4). In anormal 18 minute cycle, starting with oxygen free lead, this would mean that roughly 57% of theoxygen consumed reacts to remove antimony and arsenic.Continuous removal of the slag during softening, however, leads to slightly higher efficiencybecause the high antimony and arsenic slag formed in the earlier stages of the cycle is removed anddoes not back react with lead as time goes on (see Figures 6.3 and 6.5). Under these conditions,about 60% of the oxygen consumed reacts to remove antimony and arsenic.The improvement provided by continuously removing the slag is very small, there is only a3% increase in oxygen utilized to remove As and Sb. This is due to the limited variation range ofbullion composition during a cycle. The difference between (As+Sb) start and (As+Sb) end , where(As+Sb) corresponds to the wt% of As+Sb in the bullion, is less than 0.1 wt%. This means thatthe slag composition in equilibrium with the bullion in the early stages of the cycle is only slightlydifferent than the final slag composition. Thus, even though the activity coefficients of the slagspecies are fitted to match the final slag composition, they still provide a reasonable estimate overthe whole cycle. This leads to the conclusion that the benefits of continuously removing the slagwould only justify the additional operating requirements if large variations in bullion compositionwere occurring during a cycle.Temperature plays a role in determining softener efficiency as shown in Figure 6.6. Theactivity coefficients of the slag species were calculated at temperatures other than 6 15°C assumingthat RT ln y is constant. Increasing temperature results in a decrease in process efficiency. This- 146 -6.2 Model Validation and Calibration1. i:c: I I90v SbO5 + As0180 vPbO•o— (Bullion)70___________________ci)60-o /.- V50-Cd40 - —_v_v_v_v_v_v_v_v V1)V30 -CSoftening Time, minutesFigure 6.4 - Oxygen partition during a softening cycle in batch mode (computedfrom model simulations).- 147 -6.2 Model Validation and Calibration3 5 I I34 -v Continuous33 Slag Removal -• Batch32-31-29 - -+28 --27 --26 --25 I I I0 5 10 15 20Softening Time, minutesFigure 6.5 - Evolution of the slag composition during a softening cycle in batch and continuousslag removal modes (computed from model simulations).- 148 -6.2 Model Validation and Calibration35 I34• 595°Cv 615°C• v 635°C3231 VN:7vv.-D+ V2726 -25 I .1 I0 5 10 15 20Softening Time, minutesFigure 6.6 - Effect of temperature on the slag composition during a softening cycle in batch mode(computed from model simulations).- 149 -6.2 Model Validation and Calibrationarises due to changes in equilibrium constants of Reactions (Ill) and (IV) with temperature, seeTable 6.8.(Bul1ion) + °2(g) Sb015 (s) (Ill)and(Bullion) + °2(g) E—> AsO,5 (IV)Table 6.8 - Equilibrium constants of As and Sb oxidation (Reactions (III) and (IV)).Temperature K(°C)600 1.56 x iO’4 3.93 x iO’3625 3.85 x rn’3 1.11 x iO’3650 1.03 x i’ 3.35 x 1012675 2.94 x 1012 1.08 x 1012700 8.99x 10” 3.70x10- 150 -6.2 Model Validation and Calibration6.2.4 Batch Process OperationThe analysis of the process with the model permitted the identification of a number ofimportant parameters. However, it cannot be emphasized enough that any conclusions drawn fromthe analysis are subject to the assumption of thermodynamic equilibrium. With this in mind, thefollowing remarks concerning process operation can be made.TemperatureAlthough higher temperatures may assist with process kinetics, from a thermodynamic perspectivelower temperatures favor As+Sb deportment into the slag. Operating at the lowest possible temperature is thus preferred. Observations during the plant visits showed that temperature is also akey issue regarding lance life. High temperatures in the vessel increase the corrosiveness of theslag and reduce the life span of the lances. In 1987, H. Salomon-De-Friedberg961examined dailyassays of various softener streams in an attempt to assess relationships that might explain elementaldeportment and lance life. The prominent conclusion of his analysis was that the arsenic level inthe input bullion determined softener performance and lance life. More specifically, arsenic levelsin input bullion higher than 0.6 wt% at 620°C, and higher than 0.5 wt% at 640°C drastically reducedlance life. During the visits in March, June, and October 1994, the arsenic content in the bullionnever exceeded 0.5 wt%, but lance consumption increased on several occasions, including oneinstance where it reached 150 lances per shift. It was observed that increased lance consumptionwas always associated with high temperatures in the vessel. Two types of lance failure can occur,caused either by metal ignition with the oxygen stream, or by PbO corrosion at the slag level. Thefirst type could be explained by the ‘arsenic level effect”. As suggested by Salomon-De-Friedberg,the higher the temperature, the lower the tolerable level. The second type is directly correlated to- 151 -6.2 Model Validation and Calibrationthe corrosiveness of the slag which is strongly dependent on temperature. Temperature control inthe softener thus appears to be critical, temperature playing an important role in softener performanceand influencing considerably lance life. It should be pointed out, however, that temperatures veryclose to the slag melting point would render the softener very vulnerable to minor impurity elementssuch as tin, which can considerably lower the slag freezing point even when present in minuteamounts.Cycle lengthThe process analysis showed that short cycles, in the order of 15 minutes, are preferred to takeadvantage of the higher removal rate ofAs and Sb in the early stages. When the softener runs underthe “normal operating conditions” as defined during the plant trials and presented in Chapter 5, theduration of a cycle is about 15 to 20 minutes.As-i-Sb levelA high level of combined As+Sb favors high slag quality minimizing Pb oxidation. Depending onthe softening capacity, i.e. the ability to remove sufficient amounts of As+Sb with short cycleswhile meeting the target, As+Sb should be increased rather than lowered to boost the efficiency.Fast kinetics at low levels are effective only within the range 0.02 to 0.06 wt%, which is too lowin a partial softening process to be practical.As/Sb ratioWith respect to lance protection, low As content is preferable.- 152 -6.3 Continuous Single Pass Softening Process6.3 Continuous Single Pass Softening ProcessIn order to simulate the process, the equilibrium model developed for the batch process wasmodified to continuous softening through a number of important assumptions. The model was thenused to analyze the continuous, single pass softening route for a variety of conditions. A controlstrategy was devised as a result of the analysis. The first section deals with the assumptions of themodel, the second covers the process analysis, while the last presents the control strategy.6.3.1 AssumptionsWhen modifying the equilibrium model developed for the batch process, the followingimportant assumptions were made:• the bullion is fully backmixed,• the slag is fully backmixed, and• isothermal conditions prevail.Work on the existing softener showed that the slag phase is relatively well mixed. Bullion mixingin the continuous process would depend on flow rate and vessel geometry. With mechanical stirringit could be kept close to backmixed conditions if desired. Without deliberate mixing however, thereis a likelihood that there would be compositional gradients that would affect output bullion composition. At this stage of the analysis the heat balance was ignored in order to clarify the processchemistry. Furthermore, a heat balance would depend on a host of factors such as vessel size,refractory thickness, and use of natural gas for heating the freeboard.- 153 -6.3 Continuous Single Pass Softening ProcessThe critical model parameters used in the analysis are given in Table 6.9. Given the uncertaintyat this point as to the dependence of the slag activity coefficients on temperature and compositionthese values were held constant at levels that give reasonable predictions of the batch reactorperformance at higher As+Sb content. To simulate operating conditions, the model calculation wascarried out for short time steps (1 mm) starting at an arbitrary bullion and slag composition. Duringeach time step the appropriate weight of input bullion is added and then the amount of 02 injectedis equilibrated with the bullion and slag present. The excess bullion and slag are then removed andthe process repeated until steady state is reached. Thus, although the results presented in this thesisdeal only with steady state, the model is capable of simulating transient behaviour.Table 6.9 - Continuous softening model parameters.Activity coefficientBullionPb 1Sb 0.78A0 dependent on compositionSlagPbO 0.245b015 0.42As015 0.05- 154 -6.3 Continuous Single Pass Softening Process6.3.2 Process AnalysisContinuous single pass softening was analyzed for a variety of conditions as summarized inTable 6.10. In the base case, 15.5 tph of bullion with 2.9 wt% As+Sb is treated with 40 Nm3/hofoxygen at 615°C. The predicted output composition is 1.73 wt% As+Sb, in the middle of the targetrange of 1.5 to 2.0 wt% As+Sb.Table 6.10 - Continuous softening model parameters.Base RangeBullionoutput target As+Sb (wt%) 1.73 1.5 - 2.0input As+Sb (wt%) 2.9 2.3 - 3.5input flow (tph) 15.5 10.3 - 25.8Oxygeninjection rate (Nm3Ih) 40 32 - 56Temperature (°C) 615 595 - 635Figure 6.7 shows the effect of impurity level in input bullion on output bullion composition.As expected, output impurity level increases linearly with impurity level. The slope of the lines isapproximately 0.82 which indicates that a 1% rise in impurity level results in roughly a 0.82% risein output impurity level. This is due to the increasing thermodynamic efficiency of impurity removalas impurity level in softener bullion rises. As suggested by Reactions (Ill) and (IV),- 155 -6.3 Continuous Single Pass Softening Process(Bullion) + 3/4 °2(g) SbOl.5(S) (III)(Bullion) + (g) As015 (1) (IV)increasing the concentrations (activities) of [Sb] Pb and [As] Pb would shift the equilibrium to theright and put more Sb and As into the slag phase. This effect is illustrated in Figure 6.8.It is interesting to note that there is not a direct proportionality between As+Sb in slag andbullion as might be suggested by the equilibrium constants of Reactions (Ill) and (IV),[Sb015] = K111 [Sb] P4 (6.13)and[As015] = K [As] P’ (6.14)This arises because as As+Sb increases in bullion the oxygen dissolved in bullion, and hencedecreases. Figure 6.9 shows the variation of P with impurity level.The other feature to note from Figure 6.7 is the effect of 02 injection rate on impurity level.Increasing 02 injection results in an almost proportionate rise in impurity removal. However, thereis a reduction in efficiency as illustrated in Figure 6.10 where absolute impurity removal in gramsof As+Sb is plotted against 02 injection rate. Efficiency falls, in this case, for the same reasons itrises with output impurity level. As °2 injection rate increases, more impurity is removed from thebullion and output bullion impurity level falls. At lower (As+Sb) Pb(l) the slag is not as rich inAs+Sb and an increased fraction of 02 reacts to form PbO.- 156 -6.3 Continuous Single Pass Softening ProcessI I I I I I I I I I I I I I I I I I I I I I I I2.5--1.0- Oxygen Injection Rate -(12 3+ v32Nm/hU)30.5- O4ONm/h -A 46 Nm3/hD 56 Nm3/h0.0 I I I I I I I I I I I I I I I I I I I I I I1.5 2.0 2.5 3.0 3.5 4.0 4.5As+Sb in Input Bullion, wt%Figure 6.7 - Effect of impurity level and 02 injection rate on output bullion quality (6 15°C, 15.5tph input bullion).- 157 -6.3 Continuous Single Pass Softening ProcessAs+Sb in Output Bullion, wt%Figure 6.8 - Relationship between slag composition and impurity level in output bullion (6 15°C).555045403530tndU)-oU)+CU‘VWV,“4w“VEEE EZV25 —0.0 0.4 0.8 1.2 1.6 2.0 2.82.4- 158 -6.3 Continuous Single Pass Softening ProcessAs+Sb in Output Bullion, wtFigure 6.9 - Variation of P with impurity level in output bullion (6 15°C).ct50.1C’D01. I I IY.v,.\.I I I I I I I I I1.0 1.5 2.0 2.5- 159 -6.3 Continuous Single Pass Softening Process3Oxygen Injection Rate, Nm /hFigure 6.10 - Sensitivity of impurity removal rate to 02 injection rate (6 15°C, 15.5 tph inputbullion).C\2a-0-DU)+Q)ct50ci)1101051009590I I I I I I I I I I I I I I Iv_._.--v-.I I I I ] I I I I I I I I I I30 35 40 45 50 55 60- 160 -6.3 Continuous Single Pass Softening ProcessA comparison between Figures 6.10 and 6.3 shows that continuous single pass softeningpresents efficiency gains with respect to batch softening: removal rates in the order of 100 g(As+Sb)/mole 02 and 80 g (As+Sb)/mole 02 for the continuous and batch routes, respectively.The effect of bullion flow rate is shown in Figure 6.11. As the flow is increased at a givenimpurity level, the impurity level in the output bullion rises. The effect is not linear, however. Thisresults from the relatively constant rate of As+Sb removal. The output (As+Sb) Pb is given by,% (As+Sb)00 = % (As+Sb)1 — x 100% (6.15)where “a” is the As+Sb removal rate in tph, and “f” is the bullion flow rate in tph. Therefore doublingthe bullion flow basically cuts the change in % (As+Sb) out in half. Again, however, there is asecondary effect which arises as output % (As+Sb) out increases. As discussed above, as % (As+Sb)in output bullion increases the process becomes more efficient due to an increase in % (As+Sb) inslag, see Figure 6.8.Temperature plays a role in determining softener efficiency as shown in Figure 6.12.Increasing temperature results in a decrease in process efficiency. As previously mentioned, thisarises due to changes in equilibrium constants of Reactions (III) and (IV) with temperature, seeTable 6.8. It is clear that monitoring of softener temperature will be an important part of the controlsystem.- 161 -0-p-)0Cl)+UI6.3 Continuous Single Pass Softening ProcessHCD3. in Input BuHion, wt%Figure 6.11 - Effect of bullion flow rate on impurity level in output bullion (615°C).2.0 2,5 3.0 3.5 4.0 4.5- 162 -6.3 Continuous Single Pass Softening ProcessctjcnCfD+024.5Figure 6.12 - Effect of temperature on impurity level in output bullion (40 Nm3/h 02, 15.5 tphbullion).50454035301.5 2.0 2.5 3.0 3.5 4.0As+Sb in Input Bullion, wt%- 163 -6.4 Control Strategy6.4 Control StrategyThe above calculations show that the concept of continuous single pass softening will meetthe target set for lead softening, assuming the process operates close to thermodynamic equilibrium.A simple feedback system would control 02 injection into the softener on the basis of the measuredlevel of As+Sb in the output bullion. If the % (As+Sb) in the bullion increased, the 02 injectionrate would be increased to keep the As+Sb within a target range.Based on data presented in this Chapter, a control algorithm could be developed to regulate02 flow to a continuous softener. The essential elements of a control system would be:• a probe to measure As+Sb in the bullion output stream,• a temperature probe in the softener, and• a controller to control (1) 02 injection rate and (2) burners/coolers in the softener.The critical measurement of combined arsenic and antimony level in the lead bullion wouldbe provided by the oxygen probe, whose design was described in Chapter 4 and industrial calibrationwas presented in Chapter 5. The simple feedback system described above, although providing thenecessary control to keep the As+Sb within the target range, would not address the aspect of processoptimization. The thermodynamic model could then be incorporated in the controller, providedfurther development was carried out to adjust for the idealized assumptions built into the model.Once part of the controller, the model would have the powerful role of providing the necessary datato run the process by predicting the optimum process parameters in response to the continuously- 164 -6.4 Control Strategychanging bullion characteristics. Contrary to a simple feedback control system, which would onlyadjust the 02 injection rate to achieve bullion control, the model would be able to predict the effectof adjusting a number ofprocess parameters to a change in bullion quality. For example, in responseto an increase in emf, i.e. an increase of impurity level, the model would simulate the effect of anincrease in 02 injection rate, a decrease of input bullion flow, and a decrease of temperature. Basedon the computations, the best alternative to meet thebullion target, with the highest oxygen efficiencyand the best slag quality, would be suggested to the operator. In this manner, the operator wouldbe able to take the most appropriate action to ensure process efficiency while meeting the composition target.- 165 -CHAPTER 7SUMMARY AND RECOMMENDATIONS7,1 SummaryAn oxygen probe for continuous measurements in molten lead has been designed in thelaboratory prior to being tested in an industrial environment. A commercially available yttriastabilized zirconia serves as solid electrolyte. The reference system is composed of a Cu-Cu20mixture. Both lead wire and conducting lead (probe housing) are made of 316 stainless steel.Sealing is achieved by means of a high temperature magnesia cement. An additional plug ofcopperpowder, isolated from the reference system by a layer of alumina powder, serves as oxygen getterto eliminate oxygen ingress from the atmosphere. The extremity of the lead wire that is insertedinto the probe is coated with cement to avoid any short-circuit with the Cu plug. These featureswere decisive in the success of the probe which provides a continuous measurement for severalconsecutive days. Once it was established that the probe was giving satisfactory measurements inthe laboratory, i.e. quick response, proper response to temperature changes and oxygen potentialchanges, a testing campaign was carried out in the plant. The campaign was successful and acorrelation between measured emf and As+Sb bullion content was established as follows(wt%)AS+sb = - 1.396 + 10.45 x iO (mV)ms ±0.05 (wt%)where (wt%) As+Sb is the combined As+Sb bullion content in wt%, and (mV)Mur is the measuredemf in mV.- 166 -7.1 SummarySince any control decision would be based on the probe readings, it is crucial to regularlyensure the proper functioning of the probe. A method based on the thermal arrest technique hasbeen tested and calibrated to provide for a rapid assessment of the probe reliability. The followingcorrelation between liquidus temperature and As+Sb bullion content was obtained(wt%)M÷sb = 37.57 — 0.117 (°C)LUjd ±0.15 (wt%)where (wt%) As+Sb is the combined As+Sb bullion content in wt%, and (°C) Liquidus is the measuredemf in mV. In order to take into account any long term changes in bullion characteristics, i.e.amount of As+Sb relative to the total impurity content, this relationship should to be re-assessedon a regular basis. Such updating would only involve a simple analysis of daily plant assays.A thermodynamic model of the current semi-batch process was developed. An analysis ofthe process showed that a thermodynamic model can be used to represent process operation, andan equilibrium analysis gave a reasonable fit to operating data. The model developed for semi-batchsofteningwasmodified to continuous softening in order to simulate continuous singlepass softening.The preliminary calculations showed that the concept of continuous single pass softening will meetthe target set for lead softening, assuming the process will operate close to thermodynamic equilibrium. A slag high in As+Sb can be produced at bullion compositions in the target range forelectrorefining. Based on the data presented in this thesis, a simple feedback control algorithmcould be developed to regulate 02 injections to a continuous softener on the basis of the measuredlevel of As+Sb in output bullion. This critical measurement would be provided by the oxygenprobe designed for this project.- 167 -7.2 Recommendations7e2 RecommendationsThis thesis has served as an attempt to explore the option of continuous single pass softeningwhich would considerably simplify the softening circuit and lead to a bullion of consistent quality.Such a revision relies on the ability to continuously monitor the bullion quality, and on theunderstanding of process fundamentals.Although successful in the plant, the oxygen probe would require a number of modificationsin order to improve its reliability in an industrial environment. These modifications would include(1) improving the thermal shock resistance by a proper preheating prior to immersion,(2) improving the slag and sludge protection, and(3) improving the resistance to liquid lead infiltrations.The assessment of a control strategy, which would be based on a continuous measurementof the bullion quality, requires a proper understanding of the process. A model was developed tohelp understand the basic reactions assuming thermodynamic equilibrium. Further developmentof the model should be made to(1) adjust for deviations from idealized assumptions,(2) include certain kinetic effects, and(3) include a heat balance.Some laboratory tests should also be performed to develop a better model of the slag thermochemistry and to verify that the thermodynamic model applies to softening at high As+Sb levels.- 168 -REFERENCES(1) T.R.A. Davey, “The Physical Chemistry of Lead Refining”, Lead-Zinc-Tin’80, J.M. Cigan,T.S. Mackey and T.J. O’Keefe, Eds., The Met. Soc. ofAIME, New York, 1980, PP. 477-507.(2) J.F. Castle and J.H. Richards, “Lead Refining: Current Technology and a New ContinuousProcess”, Advances in Extractive Metallurgy, M.J. Jones, Ed., Inst. Mm. Met., London, 1977,pp. 217-234.(3) G.R. Larouche, “Antimonial and Arsenial Lead Production at Cominco Trail Operations”,Primary and Secondary Lead Processing, M.L. Jaeck, Ed., The Met. Soc. of CIM, 1990, pp.103- 109.(4) E.R. Freni, “Electrolytic Lead Refining in Sardinia”, J. Metals, November 1965, pp.1206-1214.(5) C.A. Aranda and P.J. Taylor, “Electrolytic Lead Refining as Practiced by the Cerro de PascoCorporation at La Oroya, Peru”, Extractive Metallurgy of Lead and Zinc, C.H. Catterill andJ.M. Cigan, Eds., Vol. II, The Met. Soc. of AIME, New York, 1970, pp. 891-915.(6) D. Shen, “Lead Electrorefining at Shenyang Smelter, China”, Mineral Processing andExtractive Metallurgy, M.J. Jones and P. Gill, Eds., Inst. Mm. Met., London, 1984, pp.599-605.(7) J.L. Leroy, P.J. Lenoir and L.E. Escoyez, “Lead Smelter Operation at N.V. MetallurgieHoboken S.A.”, Extractive Metallurgy of Lead and Zinc, C.H. Catterill and J.M. Cigan, Eds.,Vol. II, The Met. Soc. of AIIVIE, New York, 1970, pp. 824-852.(8) K. Emicke, G. Holzapfel and E. Kniprath, “Lead Refinery and Auxiliary By-ProductRecoveries at Norddeutsche Affinerie (N.A.)”, Extractive Metallurgy of Lead and Zinc, C.H.Catterill and J.M. Cigan, Eds., Vol.11, The Met. Soc. ofAIME, New York, 1970, pp. 867-890.(9) J.E. Bowers and R.D. Johnston, “Continuous Refining of Secondary Lead”, Mineral Processing and Extractive Metallurgy, M.J. Jones and P. Gill, Eds., Inst. Mm. Met., London,1984, pp. 63-7 1.(10) R.N. Gilges, E.K. Hudson and F.E. Pierce, “Lead Refining Practice at Bunker Hill”, TMSPaper A81-35, 1981.(11) G.K. Williams, “Development and Application of the Continuous Lead Refining Process”,McCarron, Bird and Co., Ed., Melbourne, Australia. -(12) G.K. Williams, “Continuous Lead Refining at Port Pine, South Australia”, Trans. Am. Inst.Mi Engrs., Vol. 121, 1936, pp. 226-263.- 169 -(13) F.A. Green, “The Refining of Lead and Associated Metals at Port Pine, South Australia”,The Refining of Non-Ferrous Metals, The Inst. Mm. Met., London, 1950, PP. 297-325.(14) J. Blanderer, “The Refining of Lead by Oxygen”, J. Metals, December 1984, pp. 53-54.(15) G.C. Quigley and J.V. Happ, “Kinetics of Arsenic Removal From Lead Bullion”, Researchand Development in Extractive Metallurgy, Adelaide Branch, AusIMM, Adelaide, SouthAustralia, May 1987, pp. 187-195.(16) E.T. de Groot, M.T. Martin and G.W. Toop, “Partial Softening ofLeadBullion with Oxygen”,Non-Ferrous Smelting Symposium, AusIMM-1E Aust Port Pine Combined Group, Port Pine,South Australia, September 1989, pp. 61-63.(17) K. Kiukkola and C. Wagner, “Measurements onGalvanic Cells Involving Solid Electrolytes”,J. Electrochem. Soc., Vol. 104, 1957, pp. 379-387.(18) G.G. Richards, D. Dreisinger, E. Peters and J.K. Brimacombe, “Mathematical Modellingof Zinc Processes”, Computer Software in Chemical and Extractive Metallurgy, W.T.Thompson, F. Ajersch and G. Eriksson, Eds., The Met. Soc. of CIM, Montreal, Canada,1988, pp. 223-252.(19) K.T. Jacob and T. Mathews, “Solid State Electrochemical Sensors in Process Control”,Indian J. Technology, Vol. 28, June-August 1990, pp. 413-427.(20) G.R. Fitterer, “Progress in the Development of a Device for Direct Determination ofOxygenin Liquid Steel”, J. Metals, August 1966, pp. 96 1-66.(21) G.R. Fitterer, “Further Development of the Electrolytic Method for the Rapid Determinationof Oxygen in Liquid Steels”, J. Metals, September 1967, pp. 92-96.(22) G.R. Fitterer, C.D. Cassley and V. Vierbicky, “The Rapid Determination of Oxygen inCommercial Steel With the Solid Electrolyte Probe”, J. Metals, June 1968, pp. 74-78.(23) R.J. Fruehan, L.J. Martonik and E.T. Turkdogan, “Development of a Galvanic Cell for theDetermination of Oxygen in Liquid Steel”, Met. Trans., Vol. 245, 1969, pp. 1501-1509.(24) C.K. Russell, R.J. Fruehan and R.S. Rittiger, “Probing for More Than Temperature”, I.Metals, November 1971, pp. 44-47.(25) D.A. Dukelow, J.M. Steltzer and G.F. Koons, “Oxygen Sensor: ANew Base forDeoxidationPractices?”, J. Metals, December 1971, pp. 22-25.(26) E.T. Turkdogan and R.J. Fruehan, “Review of Oxygen Sensors for Use in Steelmaking andof Deoxidation Equilibria”, Can. Met. Ouarter., Vol. 11, 1972, pp. 37 1-382.(27) M. Kawakami, K.S. Goto and M. Matsuoka, “A Solid Electrolyte Oxygen Sensor forSteelmaking Slags of the Basic Oxygen Converter”, Met. Trans., Vol. 1 1B, 1980, pp. 463-69.- 170 -(28) D. Janke, “Fundamental Aspects of the Electrochemical Oxygen Measurement in MoltenIron and Iron Alloys”, Applications of Solid Electrolytes, T. Takahashi and A. Kozawa, Eds.,JEC Press, Cleveland, 1980, pp. 154-163.(29) K.I. Suzuki and K. Sanbongi, “Application of Oxygen Probes to Steelmaking Process”,Applications of Solid Electrolytes, T. Takahashi and A. Kozawa, Eds., JEC Press, Cleveland,1980, pp. 164-170.(30) T. Kawawa, R. Imai and H. Tokunaga, “Problems in Application of Oxygen Probes inSteelmaking Process”, Applications of Solid Electrolytes, T. Takahashi and A. Kozawa,Eds., JEC Press, Cleveland, 1980, pp. 171-176.(31) M. Matsuoka and I. Ikedo, “Recent Applications ofOur Oxygen Probe for Sub-Lance Systemin Japanese Steelmaking Industry”, Applications of Solid Electrolytes, T. Takahashi and A.Kozawa, Eds., JEC Press, Cleveland, 1980, pp. 177-180.(32) W. Pluschkell, “Application of Oxygen Concentration Cells to Steel Making Problems”,Applications of Solid Electrolytes, T. Takahashi and A. Kozawa, Eds., JEC Press, Cleveland,1980, pp. 181-192.(33) K.S. Goto, K. Nagata and S. Yamaguchi, “A Review of New Uses for Oxygen Sensors forIron and Steelmaking Slags”, lion and Steelmaking, November 1983, pp. 43-52.(34) K.S. Goto, “Use of Oxygen Sensors for Steelmaking in Japan”, Scan. J. Met., Vol. 12, 1983,pp. 43-44.(35) S. Ariga and H. Ogawa, Report no 17 of 2nd National Meeting of Oxygen Probes forSteelmaking, Tokyo, July 1977.(36) T. Hiromoto, T. Saeki, Y. Igaki, T. Nisugi and K. Ishikura, “Development and Applicationof Oxygen Sensors to Steelmaking”, J. lion Steel Inst. Japan (Tetsu-to-Hagane), Vol. 63December 1977, pp. 2326-2334.(37) K.S. Goto, “Solid State Electrochemistry and its Applications to Sensors and ElectronicDevices”, Materials Science Monographs, No 45, Elsevier, 1988.(38) J. Dompas and J. Van Melle, “Operation of EMF Oxygen Probes in a Copper Refinery”, LInst. Metals, Vol. 98, 1970, pp. 304-309.(39) J.M. Dompas and P.C. Lockyer, “Oxygen Control in Liquid Copper by the Oxycell”, M.c.tTrans., Vol. 3, 1972, pp. 2597-2604.(40) W.T. Thompson and P. Tarassoff, “Determination ofOxygen in Copper with an Emf Probe”,Can. Met. Ouarter., Vol. 10, No 4, 1971, pp. 3 15-321.(41) J.M. Floyd and D.S. Conochie, “Trials ofOxy-Probes in Liquid Converter Slags”, C.S.I.R.O.Division of Chemical Engineering, 1976.- 171 -(42) N. Kemori, Y. Shibata and M. Tomono, “Application of Oxygen Probes to a Copper FlashSmelting Furnace”, Met. Rev, of MMIJ, Vol. 4, No 1, 1987, pp. 40-52.(43) O.M. Sreedharan, C. Mallinka and J.B. Gnanamoorthy, “Metallurgical Application of SolidElectrolyte Oxygen Sensors: A Review”, Bull. Electrochem., Vol.2, No 3, 1986, pp. 297-302.(44) C.B. Alcock and T.N. Belford, “Thermodynamics and Solubility ofOxygen in LiquidMetalsfrom E.M.F. Measurements Involving Solid Electrolytes, Part I - Lead”, Trans. Faraday Soc.,Vol. 60, 1964, pp. 822-835.(45) G.G. Charette and S.N. Flengas, “Thermodynamic Properties of the Oxides of Fe, Ni, Pb,Cu, and Mn, by EMF Measurements”, J. Electrochem. Soc., Vol. 115, 1968, pp. 796-804.(46) R. Szwarc, K.E. Oberg and R.A. Rapp, “The Diffusivity and solubility of Oxygen in LiquidLead from Electrochemical Measurements”, High Temperature Science, Vol. 4, 1972, pp.347-356.(47) G.K. Bandyopadhyay and H.S. Ray, “Kinetics of Oxygen Dissolution in Molten Lead”,MLTrans., Vol. 2, 1971, pp. 3055-306 1.(48) K.T. Jacob and J.H.E. Jeffes, “Thermodynamics of Oxygen in Liquid Copper, Lead andCopper Lead Alloys”, Trans. Inst. Mm. Met., Vol. 80, 1971, pp. C32-C41.(49) 5. Otsuka and Z. Kozuka, “Activities of Oxygen in Liquid Lead and Antimony fromElectrochemical Measurements”, Met. Trans., Vol. lOB, 1979, pp. 565-574.(50) S. Otsuka and Z. Kozuka, “Further Study on the Activities of Oxygen in Liquid Lead andAntimony by a Modified Coulometric Titration Method”, Met. Trans., Vol. 12B, 1981, pp.616-620.(51) A. Taskinen, “Oxygen Metal (Ag, Au, Bi, Cu, In, Ni, Sb, Sn, Te) Interaction in Dilute MoltenLead Alloys”, Acta Polytech. Scand., No 146, 1881.(52) L. Fontainas, D. Verhuist and P. Bruwier, “Oxygen Potential Measurements in Lead SmelterSlags by Means of Disposable Solid Electrolyte Probes”, Can. Met. Ouart., Vol. 24, 1985,pp. 47-52.(53) Ja. Dekeyser, F. Casteels, H. Tas, M. Soenen and M. Alberty, “On-Line Monitoring andControl in Dynamic Lithium and Static Lithium-Lead”, Proceedings 13th Conference onFusion Technology, Italy, 1984, pp. 1037-1044.(54) Ja. Dekeyser, F. Casteels, H. Tas, N. Rumbaut and M. Soenen, “On-Line Monitoring andControl of Non-Metallic Impurities in Lithium and Sodium”, Liquid Metal Engineering andTechnology, Brit. Nuclear Energy Soc., London, 1985, pp. 163-170.(55) F. De Schutter, Ja. Dekeyser, H. Tas and S. De Burbure, “Compatibility and Performanceof Electrochemical Oxygen and LithiumMonitors in Liquid Pb- l7Li”, J. Nuclear Materials,Vol. 155-157, 1988, pp. 744-748.- 172 -(56) F. De Schutter, Ja. Dekeyser, J. Luyten and H. Tas, “Monitoring System for ChemicalCharacterization of Liquid Breeder Alloys”, Sensors and Actuators, Vol. B 1, 1990, pp.446-450.(57) F. De Schutter, Ja. Dekeyser, R. Deknock, R. Jacobs and H. Tas, “Electrochemical Sensorsfor Oxygen and Lithium Detection in Pb-l7Li”, Fusion Engineering and Design, Vol. 14,1991, pp. 241-248.(58) J.C. Dekeyser and F. De Schutter, “Development of Electrochemical Sensors for Pb-l7Li”,Final Report 89-92, EnergyDivision of theVlaamse Instelling voorTechnologisch Onderzoek(VITO), March 1993.(59) Z. Kozuka, “Oxygen Potentials in Molten Metals, Mattes and Gases”, Australia JapanExtractive Metallurgy Symposium, AusIMM, Sydney, Australia, 1980, pp. 335-341.(60) R.C. Bell, G.H. Turner and E. Peters, “Fuming of Zinc from Lead Blast Furnace Slag”, I.Metals, March 1955, pp. 472-477.(61) H.H. Kellogg, “A ComputerModel of the Slag-Fuming Process for Recovery ofZinc Oxide”,Trans. Met. of AIME, Vol. 239, 1967, pp. 1439-1449.(62) R.J. Grant and L.J. Barnett, “Development and Application of the Computer Model of theSlag Fuming Process at Port Pine”, South Australia Conference 1975, AusIMM, Melbourne,1975, pp. 247-265.(63) A.K. Kyllo and G.G. Richards, “A Mathematical Model of the Nickel Converter: Part I.Model Development and Verification”, Met. Trans., Vol. 22B, 1991, pp. 153-161.(64) S. Goto, “Copper Metallurgy - Practice and Theory”, 1MM, London, 1974, pp. 23-34.(65) W.B. White, S .M. Johnson and G.B. Dantzig, “Chemical Equilibrium in ComplexMixtures”,J. Chem. Phys., Vol. 28, May 1958, pp. 751-755.(66) R.C. Oliver, S.E. Stephanou and R.W. Baier, “Calculating Free-Energy Minimization”,Chem. Eng., February 1962, pp. 121-128.(67) G. Eriksson, “Thermodynamic Studies of High Temperature Equilibria, III. SOLGAS, aComputer Program for Calculating the Composition and Heat Condition of an EquilibriumMixture”, Acta Chem. Scand., Vol. 25, 1971, pp. 265 1-2658.(68) G. Eriksson and E. Rosen, “Thermodynamic Studies of High Temperature Equilibria, VIII.General Equations for the Calculation of Equilibria in Multiphase Systems”, Chem. Scripta,Vol. 4, 1973, pp. 193-194.(69) A.Y. Vartiainen, P.A. Taskinen and A.T. Jokilaakso, “Thermochemical Description ofAntimony and Arsenic in the Suspension Stage of the Outokumpu Flash Smelting Furnace”,H.H. Kellogg International Symposium, QuantitativeDescription of Metals ExtractionProcesses, N.J. Themelis and P.F. Duby, Eds., TMS, 1991, pp. 45-67.- 173 -(70) G.G. Richards, J.K. Brimacombe and G.W. Toop, “Kinetics of the Zinc Slag Fuming Process:Part I. Industrial Measurements”, Met. Trans. B, Vol. 16B, 1985, pp. 5 13-527.(71) G.G. Richards and J.K. Brimacombe, “Kinetics of the Zinc Slag Fuming Process: Part II.Mathematical Model”, Met. Trans. B, Vol. 16B, 1985, pp. 529-540.(72) G.G. Richards and J.K. Brimacombe, “Kinetics of the Zinc Slag Fuming Process: Part III.Model Prediction and Analysis of Process Kinetics”, Met. Trans. B, Vol. 16B, 1985, pp.541-549.(73) S.L. Cockcroft, G.G. Richards and J.K. Brimacombe, “High-Pressure Coal Injection in ZincSlag Fuming”, Met. Trans. B, Vol. 20B, 1989, pp. 227-235.(74) J.W. Patterson, “Conduction Domains for Solid Electrolytes”, J. Electrochem. Soc., Vol.118, l9’7l,pp. 1033-1039.(75) C.B. Choudhary, H.S. Maiti and E.C. Subbarao, “Defect Structure and Transport Properties”,in Solid Electrolytes and Their Applications, E.C. Subbarao, Ed., Plenum Press, 1980, pp.1-80.(76) T.H. Etsell and S.N. Flengas, “The Electrical Properties of Solid Oxide Electrolytes”,Chemical Reviews, Vol. 70, 1970, pp. 339-376.(77) H.L. Tuller and A.S. Novick, “Doped Ceria as a Solid Oxide Electrolyte”, J. Electrochem., Vol. 122, 1975, pp. 255-259.(78) H.S. Maiti and E.C. Subbarao, “Electrical Conduction in CaO-Doped Thoria Electrolytes”,J. Electrochem. Soc., Vol. 123, 1976, pp. 1713-1718.(79) P.M. Mc Geehin and A. Hooper, “Review of Fast Ion Conduction Materials”, J. Mater. Sc.,Vol. 12, 1977, pp. 1-27.(80) A. Hooper, “Fast Ionic Conductors”, Contemp. Phys., Vol. 19, 1978, pp. 147-168.(81) E.C. Subbarao and H.S. Maiti, “Solid Electrolytes with Oxygen Ion Conduction”, SolidState Tonics, Vol. 11, 1984, pp. 3 17-338.(82) A.J. Burggraaf, B.A. Boukamp, I.C. Vinke and K.J. de Vries, “Recent Developments inOxygen-Ion Conducting Solid Electrolyte and Electrode Materials”, Adv. Sol. State Chem.,Vol. 1, 1989, pp. 259-293.(83) Nippon Kagaku Togyo Co. Ltd., Marketing pamphlet.(84) W.L. Worrell and J.L. Iskoe, “Kinetics of Oxygen Transfer at the Interface Between aZirconia-BasedElectrolyte and aMetal-Metal Oxide Electrode”,Fast Ion Transport in Solids,W. Van Gool, Ed., TUPAC, North-Holland, Amsterdam, 1973, pp. 5 13-521.- 174 -(85) I. Barin and 0. Knacke, “Thermochemical Properties of Inorganic Substances”,Springer-Verlag, Berlin, 1973.(86) 0. Kubaschewski and C.B. Alcock, “Metallurgical Thermochemistry”, 5th Ed., PergamonPress, 1979.(87) Y.K. Rao, “Stoichiometry and Thermodynamics of Metallurgical Processes”, CambridgeUniversity Press, 1985.(88) L. Pankratz, “Thermodynamic Properties of Elements and Oxides”, Bulletin No 672, USBureau of Mines, 1982.(89) I. Barin, “Thermochemical Data of Pure Substances”, VCH, Weinheim, Germany, 1989.(90) K. Itagaki, T. Shimizu and M. Hino, “Thermodynamic Studies of the Liquid Lead-ArsenicSystem”, Report No 767, Research Institute of Mineral Dressing and Metallurgy, TohokuUniversity, Japan, 1978.(91) R. Hultgren, P.D. Desai, D.T. Hawkins, M. Gleiser and K.K. Kelley, “Selected Values ofthe Thermodynamic Properties of Binary Alloys”, America Society for Metals, 1973.(92) H. Seltz and B.J. De Witt, “A Thermodynamic Study of the Lead-Antimony System”, j.Aim Chem. Soc., Vol. 61, 1939, pp. 2594-2597.(93) A.D. Zunkel and A.H. Larson, “Slag-Metal Equilibria in the Pb-PbO-Sb203 System”, Trans.Met. Soc. AIME, Vol. 239, 1967, pp. 473-477.(94) E. Sugimoto, S. Kuwata and S. Kozuka, “Activity Measurements in the Pb-Sn and Pb-SbAlloys by E.M.F. Method with Y203 Doped Stabilized Zirconia at Low Temperatures”,Nippon Kogvo Kaishi, Vol. 98, May 1982, pp. 39-45.(95) V. Chaskar, “Thermodynamics of Softener Slags”, unpublished work, University of BritishColumbia, Vancouver, Canada, 1995.(96) H. Salomon-De-Friedberg, “Softener Metallurgy”, Cominco Metals Development, InternalMemorandum No 840-10, June 1987.- 175 -APPENDIX 1EXPERIMENTAL EQUIPMENT- 176-Custom-made Furnace SpecificationsDimensions 1.0 m x 0.5 m x 0.5 mHeating element Chromel A-i ribbonNumber of windings 104 with variable pitch as follows (from top to bottom):7 in. with 4 windings/in.4 in. with 31/2 windings/in.8 in. with 21/2 windings/in.4 in. with 31/2 windings/in.7 in. with 4 windings/in.Insulating materialsInner core Solid alumina particles (with traces of titanium oxide)Walls Diatomite insulating bricks (Diatherm 23)Fiber fraxMeasured resistance 12.4 2Maximum temperature 1100°CHotzone(± 1°C) 10cmChamber diameter 7 cmElectromagnetic shield Stainless steel foilOther specifications • Water cooled top and bottom gas-tight caps to allow controlof the atmosphere• Power supply unit• Temperature controller (Omega CN9000 Series)• S-type control thermocouple- 177-Corning 130 pH and Millivolt MeterSpecificationsRange 0 to ± 1800 mVResolution 0.1 mVRelative accuracy ± 0.2 mVRepeatability ± 0.1 mVRecorder output AdjustableModes STANDBY, pH, REL mV, mVInput impedance > 1012 ohms- 178-DAS-8 BoardFeatures• 8 analog input channels• 12-bit resolution• 7 digital 110 bits (4 out, 3 in)• 4000 samples per second using Call Driver• On-board sample pacer clockSpecificationsNumber of channels 8, single-endedResolution 12 bits (2.4 mV/bit)Accuracy 0.01% of reading ± bitFull scale ±5 voltA/D type Successive approximationLinearity ± 1 bitCoding Offset binaryOvervoltage Continuous single channel to ±35 VInput impedance i07 ohmsInput current 100 nA max at 25°C- 179-Exp-16 Channel MultiplexerFeatures• Expands any analog input to 16 differential inputs• Cold-junction compensation for thermocouples• Open thermocouple detection• Shunt terminals for current measurement• Daisy-chain up to 8 boards• Instrumentation amplifier with switch-selectable gains of 0.5, 1,2, 10, 50, 100, 200, and 1000• Input filteringSpecificationsInput bias current 2 nA typ, 6 nA maxTemperature coefficient 5 ppm typ, 15 ppm maxOvervoltage protection ±30 V continuousCommon mode voltage ±10 V maxAnalog output voltage ±5 V maxAnalog output current 20 mA maxCold-junction compensation +24.4 mVI°C (.1°C/bit)0.0 V at 0.0°CThermocouple types accepted J, K, T, E, S, R, B- 180-APPENDIX 2PLANT DATA- 181 -Sbwt% Aswt% Biwt% Cuwt%Input bullion - cycle 1 1.20 0.18 0.35 0.08Softener bullion - Start of cycle 1 0.77 0.07 0.31 0.11Softener bullion - End of cycle 1 0.72 0.07 0.33 0.11Slag sample - End of cycle 1 24.1 6.9 - -Table A2. 1 - Assays of input and output bullion, and slag from cycle 1 (June test).Sbwt% Aswt% Biwt% Cuwt%Input bullion - cycle 2 1.19 0.18 0.35 0.09Softener bullion - Start of cycle 2 0.78 0.08 0.32 0.11Softener bullion - End of cycle 2 0.71 0.06 0.31 0.11Slag sample - End of cycle 2 24.3 6.9 - -Table A2.2 - Assays of input and output bullion, and slag from cycle 2 (June test).Sbwt% Aswt% Biwt% Cuwt%Inputbullion—cycle3 1.16 0.17 0.34 0.10Softener bullion - Start of cycle 3 0.77 0.07 0.31 0.11Softener bullion - End of cycle 3 0.69 0.06 0.32 0.11Slag sample - End of cycle 3 24.1 6.9 - -Table A2.3 - Assays of input and output bullion, and slag from cycle 3 (June test).- 182-Sbwt% Aswt% Biwt% Cuwt%Inputbullion-cycle4 1.13 0.16 0.32 0.11Softener bullion - Start of cycle 4 0.78 0.08 0.32 0.11Softener bullion - End of cycle 4 0.67 0.05 0.29 0.11Slag sample - End of cycle 4 24.4 6.9 - -Table A2.4 - Assays of input and output bullion, and slag from cycle 4 (June test).Sbwt% Aswt% Biwt% Cuwt%Inputbullion-cycle5 1.13 0.16 0.32 0.10Softener bullion - Start of cycle 5 0.75 0.07 0.30 0.11Softener bullion - End of cycle 5 0.68 0.06 0.31 0.11Slag sample - End of cycle 5 24.5 6.9 - -Table A2.5 - Assays of input and output bullion, and slag from cycle 5 (June test).Sbwt% Aswt% Biwt% Cuwt%Inputbullion-cycle6 1.13 0.16 0.33 0.10Softener bullion - Start of cycle 6 0.72 0.06 0.31 0.10Softener bullion - End of cycle 6 0.68 0.06 0.31 0.11Slag sample - End of cycle 6 24.5 6.9 - -Table A2.6 - Assays of input and output bullion, and slag from cycle 6 (June test).- 183-Sbwt% Aswt% Biwt% Cuwt%Input bullion - cycle 1 0.93 0.21 0.26 0.07Softener bullion - Start of cycle 1 0.68 0.10 0.25 0.07Softener bullion - End of cycle 1 0.62 0.08 0.26 0.08Slag sample - End of cycle 1 18.7 8.6 - -Table A2.7 - Assays of input and output bullion, and slag from cycle 1 (October test).Sbwt% Aswt% Biwt% Cuwt%Input bullion - cycle 2 0.91 0.20 0.26 0.08Softener bullion - Start of cycle 2 0.67 0.09 0.25 0.08Softener bullion - End of cycle 2 0.61 0.08 0.26 0.08Slag sample - End of cycle 2 18.8 8.5 - -Table A2.8 - Assays of input and output bullion, and slag from cycle 2 (October test).Sbwt% Aswt% Biwt% Cuwt%Input bullion - cycle 3 0.89 0.19 0.26 0.07Softener bullion - Start of cycle 3 0.66 0.09 0.26 0.07Softener bullion - End of cycle 3 0.59 0.08 0.25 0.08Slag sample - End of cycle 3 18.8 8.6 - -Table A2.9 - Assays of input and output bullion, and slag from cycle 3 (October test).- 184-


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



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"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items