{"@context":{"@language":"en","Affiliation":"http:\/\/vivoweb.org\/ontology\/core#departmentOrSchool","AggregatedSourceRepository":"http:\/\/www.europeana.eu\/schemas\/edm\/dataProvider","Campus":"https:\/\/open.library.ubc.ca\/terms#degreeCampus","Creator":"http:\/\/purl.org\/dc\/terms\/creator","DateAvailable":"http:\/\/purl.org\/dc\/terms\/issued","DateIssued":"http:\/\/purl.org\/dc\/terms\/issued","Degree":"http:\/\/vivoweb.org\/ontology\/core#relatedDegree","DegreeGrantor":"https:\/\/open.library.ubc.ca\/terms#degreeGrantor","Description":"http:\/\/purl.org\/dc\/terms\/description","DigitalResourceOriginalRecord":"http:\/\/www.europeana.eu\/schemas\/edm\/aggregatedCHO","Extent":"http:\/\/purl.org\/dc\/terms\/extent","FileFormat":"http:\/\/purl.org\/dc\/elements\/1.1\/format","FullText":"http:\/\/www.w3.org\/2009\/08\/skos-reference\/skos.html#note","Genre":"http:\/\/www.europeana.eu\/schemas\/edm\/hasType","GraduationDate":"http:\/\/vivoweb.org\/ontology\/core#dateIssued","IsShownAt":"http:\/\/www.europeana.eu\/schemas\/edm\/isShownAt","Language":"http:\/\/purl.org\/dc\/terms\/language","Program":"https:\/\/open.library.ubc.ca\/terms#degreeDiscipline","Provider":"http:\/\/www.europeana.eu\/schemas\/edm\/provider","Publisher":"http:\/\/purl.org\/dc\/terms\/publisher","Rights":"http:\/\/purl.org\/dc\/terms\/rights","RightsURI":"https:\/\/open.library.ubc.ca\/terms#rightsURI","ScholarlyLevel":"https:\/\/open.library.ubc.ca\/terms#scholarLevel","Title":"http:\/\/purl.org\/dc\/terms\/title","Type":"http:\/\/purl.org\/dc\/terms\/type","URI":"https:\/\/open.library.ubc.ca\/terms#identifierURI","SortDate":"http:\/\/purl.org\/dc\/terms\/date"},"Affiliation":[{"@value":"Applied Science, Faculty of","@language":"en"},{"@value":"Materials Engineering, Department of","@language":"en"}],"AggregatedSourceRepository":[{"@value":"DSpace","@language":"en"}],"Campus":[{"@value":"UBCV","@language":"en"}],"Creator":[{"@value":"Lo, Wai Man","@language":"en"}],"DateAvailable":[{"@value":"2008-09-08T18:29:42Z","@language":"en"}],"DateIssued":[{"@value":"2007","@language":"en"}],"Degree":[{"@value":"Master of Applied Science - MASc","@language":"en"}],"DegreeGrantor":[{"@value":"University of British Columbia","@language":"en"}],"Description":[{"@value":"In the present study, the corrosion behaviour of several magnesia-chrome (MC) and magnesia-alumina spinel (MA) bricks against fayalite type slags was investigated and the role of the spinel phases was highlighted . The experimental results revealed that the corrosion resistance of the MC bricks was superior to the MA bricks against KIVCET slags in static and dynamic conditions . As a result of the interaction between MgO from MC bricks and the slag, a modified forsterite phase (Mg, Fe, Zn, Ca)\u2082SiO\u2084 was formed, which destroyed the precipitated complex spinel bonds at the grain boundaries of periclase and magnesia-chromia spinel . Furthermore, both MgO and MgO-MgAl\u2082O\u2084 spinel in the MA brick dissolved into the slag, which resulted in modified forsterite phases of (Mg, Fe, Zn, Ca)\u2082SiO\u2084 and (Mg, Fe, Zn)(Fe, Al)\u2082O\u2084 complex spinels, respectively. In addition, the accretion formation in the KIVCET furnace was investigated through solubility experiments of Cr\u20820\u2083 in the KIVCET slag with various amounts of lead, which revealed that the net contribution of Cr\u20820\u2083 to the spinel formation is the highest in the barren (no Pb) slag, followed by high-lead (11% Pb) and it is the lowest for the low-lead (6% Pb) slag. The amount of spinel solid solution increased consistently with increasing Cr\u20820\u2083 dissolved and the PbO existent in the slag.\nFrom examinations of several used bricks from the tuyere area of a Peirce Smith nickel converter, it was found that the corrosion is due to the interaction of the partially oxidized matte penetrating deep into the brick and the magnesia grains forming (Mg, Fe, Ni, Co) xOy spinels . Analyses of brick samples used in the KIVCET Electric Furnace roof identified deep reaching sulphation, which weakened the bonding phase between coarse magnesia grains. In the Bottom Blown Oxygen Converter, a highly aggressive lead and bismuth oxide rich slag penetrated deep into the brick, which destroyed the grain boundaries, causing the refractory to be easily eroded at the refractory-slag interface.\nOur studies concluded that the spinel phases, either as magnesium chromate, magnesium aluminate or complex spinel [(Mg, Fe)(Cr, Al, Fe)\u2082O\u2084], enhanced the corrosion resistance of a basic refractory to fayalite type slags from the non-ferrous smelting and converting furnaces.","@language":"en"}],"DigitalResourceOriginalRecord":[{"@value":"https:\/\/circle.library.ubc.ca\/rest\/handle\/2429\/1712?expand=metadata","@language":"en"}],"Extent":[{"@value":"23301902 bytes","@language":"en"}],"FileFormat":[{"@value":"application\/pdf","@language":"en"}],"FullText":[{"@value":"CORROSION OF BASIC REFRACTORIES IN NON-FERROUS CONVERTERS by Wai Man Lo B.A.Sc . (Metals and Materials Engineering), The University of British Columbia, 2004 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF APPLIED SCIENCE in THE FACULTY OF GRADUATE STUDIES (Materials Engineering) THE UNIVERSITY OF BRITISH COLUMBIA December 2007 \u00a9 Wai Man Lo, 2007 Abstract In the present study, the corrosion behaviour of several magnesia-chrome (MC) and magnesia-alumina spinel (MA) bricks against fayalite type slags was investigated and the role of the spinel phases was highlighted . The experimental results revealed that the corrosion resistance of the MC bricks was superior to the MA bricks against KIVCET slags in static and dynamic conditions . As a result of the interaction between MgO from MC bricks and the slag, a modified forsterite phase (Mg, Fe, Zn, Ca) 2SiO4 was formed, which destroyed the precipitated complex spinel bonds at the grain boundaries of periclase and magnesia-chromia spinel . Furthermore, both MgO and MgO-MgAl 2O4 spinel in the MA brick dissolved into the slag, which resulted in modified forsterite phases of (Mg, Fe, Zn, Ca)2SiO4 and (Mg, Fe, Zn)(Fe, Al)204 complex spinels, respectively. In addition, the accretion formation in the KIVCET furnace was investigated through solubility experiments of Cr203 in the KIVCET slag with various amounts of lead, which revealed that the net contribution of Cr203 to the spinel formation is the highest in the barren (no Pb) slag, followed by high-lead (11% Pb) and it is the lowest for the low-lead (6% Pb) slag . The amount of spinel solid solution increased consistently with increasing Cr 203 dissolved and the PbO existent in the slag. From examinations of several used bricks from the tuyere area of a Peirce Smith nickel converter, it was found that the corrosion is due to the interaction of the partially oxidized matte penetrating deep into the brick and the magnesia grains forming (Mg, Fe, Ni, Co) XOy spinels . Analyses of brick samples used in the KIVCET Electric Furnace roof identified deep reaching sulphation, which weakened the bonding phase between coarse magnesia ii grains. In the Bottom Blown Oxygen Converter, a highly aggressive lead and bismuth oxide rich slag penetrated deep into the brick, which destroyed the grain boundaries, causing the refractory to be easily eroded at the refractory-slag interface. Our studies concluded that the spinel phases, either as magnesium chromate, magnesium aluminate or complex spinel [(Mg, Fe)(Cr, Al, Fe) 2 O4], enhanced the corrosion resistance of a basic refractory to fayalite type slags from the non-ferrous smelting and converting furnaces . iii Table of Contents Abstract\t ii List of Tables\t vii List of Figures\t x List of Abbreviations and Symbols \t xix Acknowledgements\t xx 1 .0 Introduction\t 1 2.0 Literature Review\t 5 2 .1 Spinel Phases in Basic Refractories \t 5 2.1 .1 Crystal Structure \t 5 2 .1 .2 Magnesia-Chromia Spinel \t 7 2 .1 .3 Precipitated Complex Spinel \t 8 2 .1 .4 Secondary Spinel Considerations \t 9 2.1.5 Magnesia-Alumina Spinel \t 10 2.2 Refractory-Slag Interactions \t 12 2.2 .1 Microstructure\t 12 2.2.2 Mineralogical Considerations\t 13 2.2.3 Chemical Considerations \t 14 2.2 .3 .1 Magnesia-Chrome Refractories \t 15 2.2.3.2 Magnesia-Alumina Spinel Refractories \t 18 2.2.4 Slag Property Considerations \t 20 2.2.5 Thermo-Mechanical Considerations \t 21 iv 2.3 Industrial Non-Ferrous Converting Furnaces \t 30 2.3 .1 Nickel-Copper Converting and Refining (Peirce Smith Converter) \t 23 2.3.2 Zinc-Lead Smelting and Converting (KIVCET Furnace) \t 25 3.0 Scope and Objectives\t 26 4 .0 Experimental Procedures\t 29 4.1 Laboratory Experiments \t 29 4.1.1 Static Corrosion Method \t 29 4 .1 .2 Dynamic Corrosion Method\t 31 4.1 .3 Solubility Test in KIVCET Slag\t 33 4.1 .4 Leaching Test of Hexavalent Chromium \t 35 4 .2 Industrial Experiments \t 36 4.2 .1 Post-Mortem Failure Analysis \t 36 4.3 Samples for Laboratory Corrosion Tests and Post-Mortem Analysis \t 39 4.4 Slag Composition\t 41 5.0 Results and Discussions \t 43 5 .1 Laboratory Experiments \t 43 5.1 .1 Characterization of the KIVCET Slag \t 43 5.1.2 Corrosion Performance by the Static Corrosion Method \t 45 5.1.3 Corrosion Performance by the Dynamic Corrosion Method \t 48 5.1.4 Magnesia-Chrome Bricks with KIVCET Slag\t 54 5.1.5 Magnesia-Alumina Spinel Bricks with KIVCET Slag\t 57 5.1 .6 Solubility of Cr2 O3 in KIVCET Slag\t 60 v 5 .1 .7 Leaching Test for HczovoleotChromium \t 71 5.2 Industrial Experiments\t 72 5 .2 .1 Peirce Smith Nickel Converter \t 73 5.2 .2 Electric Furnace ofKllJCET\t 00 5.2 .3 Bottom Blown Oxygen Convertor \t 100 6.0 Summary\t 1Z9 7.0 Conclusions\t 126 8 .0 Future Work\t 128 9.0 References\t 129 Appendixes \" .\" .\" .\" .\" .\" .\" .\" .\" .\" . .\" .\" .\" .\" .\" .\" . . .\" .' .\" .' . .\" .' . .\" . . .\" . . . .'\t 133 vim List of Tables Table 1 Silicates and ferrites most likely to be found in magnesia refractories \t 14 Table 2 Silicates and ferrites most likely to be formed in magnesia refractories \t 14 Table 3 Characteristics of magnesia-chrome and magnesia-alumina spinel brick samples\t 39 Table 4 Samples to be tested by the Static, Dynamic, Leaching and Post-Mortem Failure Analysis Test Methods\t 40 Table 5 Chemical composition of the KIVCET slag \t 42 Table 6 Chemical compositions of the experimental mixes for the solubility test \t 42 Table 7 Composition of the KIVCET slag by EDS point analysis\t 43 Table 8 Corrosion and penetration depth of the samples from the static corrosion method\t 45 Table 9 Corrosion and penetration depth of the samples from the dynamic corrosion Method\t 48 Table 10 EDS analysis of MC-2 from the static corrosion method corresponding to Figure 18\t 55 Table 11 EDS analysis of MC-2 from the dynamic corrosion method corresponding to Figure 19 \t 56 Table 12 EDS analysis of MA-1 from the static corrosion method corresponding to Figure 20\t 58 Table 13 EDS analysis of MA-2 from the dynamic corrosion method corresponding to Figure 21 \t 59 vii Table 14 Preliminary quantitative XRD analysis on slag mixes LO (low-Pb) and H2 (high-Pb) fired at 1250\u00b0 for 1 hour in argon \t 62 Table 15 Measured peak intensities on samples fired at 1250\u00b0C for one hour in argon . . . 63 Table 16 Theoretical and true XRD peak position (20) of spinels \t 64 Table 17 Peak intensities for Willemite (Zn2SiO4) \u2022 \u2022 \u2022 \u2022 \u2022 \u2022\t 65 Table 18 Peak intensities for Hardystonite (Ca 2ZnSi 2 O7)\t 66 Table 19 Peak intensities for Spinel Phases Solid Solution (Fe 3 O4, FeCr2O4, ZnFe2O4 , ZnCr2O4, ZnFeCrO4)\t 67 Table 20 Trivalent and hexavalent chromium results from the leach test \t 71 Table 21 EDS analysis of slag and matte corresponding to Figure 29 \t 77 Table 22 EDS profile of elements across the interface between the matte and the MgO grain corresponding to Figure 31A \t 79 Table 23 EDS profile of elements across the interface between the matte and the Cr 2 O3 grain corresponding to Figure 31B \t 79 Table 24 EDS point analysis at the cold face of the used MC-9 brick corresponding to Figure 32 \t 82 Table 25 EDS point analysis of the unused MC-9 brick corresponding to Figure 33 \t 82 Table 26 EDS point analysis in the porous zone corresponding to Figure 44 \t 94 Table 27 EDS point analysis in the porous zone corresponding to Figure 45 \t 95 Table 28 Volume loss of the used MC-2, MC-8 and MC-10 bricks \t 104 Table 29 EDS point analysis for MC-2 corresponding to Figure 61 \t 110 Table 30 EDS point analysis for MC-8 corresponding to Figure 62 \t 111 Table 31 EDS point analysis for MC-10 corresponding to Figure 63 \t 112 viii Table 32 EDS point analysis for unused MC-2 corresponding to Figure 64 \t 113 Table 33 EDS point analysis for unused MC-8 corresponding to Figure 75 \t 114 Table 34 EDS point analysis for unused MC-10 corresponding to Figure 66 \t 115 ix List of Figures Figure 1 The Peirce Smith Converter \t 2 . . . Figure 2 The Peirce Smith Converter \t 3 Figure 3 The Peirce Smith Converter \t 4 Figure 4 The Peirce Smith Converter \t 8 Figure 5 Phase diagram of MgO-Al2O 3 system\t 11 Figure 6 Separation due to a vitrified or slagged layer \t 22 Figure 7 Illustration of the core-drilled refractory crucible for the Static Corrosion Method\t 30 Figure 8 Illustration of the Dynamic Corrosion Method : A) specimen dimensions, B) specimen formation in rotary furnace and, C) rotary furnace set up \t 32 Figure 9 Illustration of the test sample for the Solubility Method \t 34 Figure 10 Illustration of the test sample for the Leaching Method \t 35 Figure 11 Schematic of the different zones in a used brick sample \t 36 Figure 12 Specimen preparation for the Cold Crushing Strength and the Modulus of Rupture tests \t 39 Figure 13 SEM image of KIVCET slag displaying : 1) glass phase and, 2) crystalline phase, (500x)\t 44 Figure 14 XRD peaks for the KIVCET slag \t 44 x Figure 15 Cross-section of the crucibles from static corrosion method : A) MC-2, B) MC-4, C) MC-7, D) MC-8, E) MA-1 and, F) MA-4 \t 46 Figure 16 Cross-section of the samples from dynamic corrosion method : A) MC-1, B) MC-2, C) MC-3, D) MC-4, E) MC-5, F) MC-6, G) MC-7, H) MC-8, I) MA-1, J) MA-2 and, K) MA-3\t 49 Figure 17 Microstructure of the unused bricks : A) MC-8 (80x), B) MC-5 (80x), C) MC-2 (80x) and, D) MC-1 (80x)\t 53 Figure 18 SEM image of MC-2 from the static corrosion method displaying : 1) slag, 2) interface of (Mg, Fe, Zn)(Fe, Al, Cr)2O4 complex spinel and, 3) refractory, (50x). The corresponding EDS analysis is shown in Table 12 \t 55 Figure 19 SEM image of MC-8 from the dynamic corrosion method displaying: 1-4) (Mg, Fe, Zn, Ca)SiO4 spinel, 5) Cr2 O3 grain, 6) MgO grain and, 7) (Mg, Fe)(Cr, A1)2O3 spinel, (50x) . The corresponding EDS analysis is shown in Table 13\t 56 Figure 20 A) SEM image of MA-1 from the static corrosion method displaying : 1) MgAl2O3 grain, 2) MgO grain, 3) slag, 4) MgO-slag interface, 5) interface of (Mg, Fe, Zn)(Al, Fe)2O4 complex spinel and, 6) refractory-slag interface, (50x) . The corresponding EDS analysis is shown in Table 14, B) Magnified view of the MgAl2O3 grain and slag interface, (400x)\t 58 xi Figure 21 A) SEM image of MA-2 from the dynamic corrosion method displaying : 1) MgAl2 O3 grain, 2) MgO grain, 3) (Mg, Fe, Zn)(Fe, Al) 2O4 spinel, 4) (Mg, Fe, Zn, Ca)SiO 4 spinel, 5) (Mg, Fe, Zn, Ca)SiO4 spinel, 6) slag (glass phase) and, 7) slag (crystalline phase), (200x) . The corresponding EDS analysis is shown in Table 15, B) Magnified view of (Mg, Fe, Zn)(Fe, Al) 2 O4 spinel and (Mg, Fe, Zn, Ca)SiO4 spinel, (600x) \t 59 Figure 22 Crucibles after the solubility test for 1 hour in argon at : A) 1250\u00b0C, B) 1300\u00b0C and, C) 1400\u00b0C \t 61 Figure 23 XRD intensities for Willemite versus %Pb (left) and Cr 2O3 (right)\t 65 Figure 24 XRD intensities for Hardystonite versus %Pb (left) and Cr 2 O3 (right)\t 66 Figure 25 XRD intensities for spinel phases versus %Pb (left) and Cr2O3 (right)\t 67 Figure 26 Position within the Peirce Smith Nickel Converter where the bricks from the tuyere lining were sampled \t 72 Figure 27 Cross-section images of the tuyere bricks for : A)1A, B)2A and, C)3A \t 74 Figure 28 Profile of the properties for MC-9 . Apparent Porosity (AP), Water absorption (WA), Bulk Density (BD) and Modulus of Rupture (MOR) for MC-9 : A) used brick, B) unused brick and, Cold Crushing Strength for: C) used brick, D) unused brick \t 75 Figure 29 A) SEM and B) EDS mapping of the slag and the matte residing at the hot face, (200x). The corresponding EDS analysis is shown in Table 23 \t 77 xii Figure 30 SEM images (A, 150x, B, 80x) displaying cracks through MgO grains and along grain boundaries suspected to be initiated by penetrated matte \t 77 Figure 31 A) Matte and MgO interface, (800x) and, B) matte and Cr 2O3 interface, (800x) . The corresponding EDS analyses are shown in Table 24 and 25, respectively\t 79 Figure 32 SEM images at the cold face of the used MC-9 brick : A) SEM of overall surface (80x), B) Magnified view of spinel phase (1500x) and, C) Magnified view of silicate phase (1500x) \t 81 Figure 33 SEM image of the unused MC-9 brick : A) SEM of overall surface (80x), B) Magnified view of spinel phase (800x) and, C) Magnified view of silicate phase (1000x)\t 81 Figure 34 Elemental profile for the used MC-9 brick through the longitudinal cross- section\t 83 Figure 35 A) SEM and, B) EDS mapping image at 79 mm into the refractory displaying the penetration of matte high in nickel, cooper and sulphur, (60x) \t 84 Figure 36 XRD peaks for slag and matte\t 85 Figure 37 XRD profile for the used MC-9 brick through the longitudinal cross-section \t 86 Figure 38 XRD peaks for the unused MC-9 brick \t 86 Figure 39 Image of a spalled portion in the KIVCET Electric Furnace roof \t 88 Figure 40 Cross-section of the used KIVCET Electric Furnace roof brick \t 90 Figure 41 Cross-section of the used KIVCET Electric Furnace roof brick with measurements of penetration \t 90 Figure 42 Profile of the properties for MC-11 . Apparent Porosity (AP), Water absorption (WA), Bulk Density (BD) and Modulus of Rupture (MOR) for: A) used brick, B) unused brick and, Cold Crushing Strength for : C) used brick, D) unused brick\t 91 Figure 43 A-B) Large cracks in the porous zone at 70 mm from the hot face, (40x) \t 92 Figure 44 SEM image of the porous zone displaying : 1-4) sulphate phase, 5) lead oxide phase, 6) Cr203 grains and, 7) MgO grain, (150x) . The corresponding EDS analysis is shown in Table 28 \t 94 Figure 45 SEM image of the porous zone displaying : 1-5) sulphate phase, 6) lead oxide and, 7) MgO grain, (150x) . The corresponding EDS analysis is shown in Table 29\t 94 Figure 46 A-B) High concentration of lead in the \"red\" zone at 106 mm from the hot face, (40x)\t 95 Figure 47 Elemental profile for the used MC-11 brick through the longitudinal cross- section\t 96 Figure 48 SEM image at 106 mm into the brick displaying 100% lead oxide, (200x) \t 97 Figure 49 SEM image at 116 mm into the brick displaying lead oxide engulfing the refractory grains, (200x) \t 97 Figure 50 XRD profile for the used MC-11 brick through the longitudinal cross-section . 97 Figure 51 Position of the three different types of trial bricks (MC-2, MC-8 and MC-10) in the Bottom Blown Oxygen Converter \t 100 xiv Figure 52 Image of the BBOC brick : A) at the end of one regular campaign and, B) magnified image showing the brick's texture at the hot face\t 101 Figure 53 Images of the trial bricks : A) at the end of the premature campaign and, B) cracking at the hot face\t 101 Figure 54 Cross-section of MC-2 displaying : 1) macro-crack and 2) micro-crack perpendicular to the hot face, 3) micro-crack parallel to the hot face and, 4) silver\t 102 Figure 55 Cross-section of MC-8 displaying : 1) macro-crack and 2) micro-crack perpendicular to the hot face, 3) macro-crack parallel to the hot face and, 4) silver\t 103 Figure 56 Cross-section of MC-10 displaying : 1) macro-crack and 2) micro-crack perpendicular to the hot face, 3) micro-crack parallel to the hot face and, 4) silver\t 103 Figure 57 Profile of the properties : Apparent Porosity (AP), Water absorption (WA), Bulk Density (BD) and Modulus of Rupture (MOR) for MC-2 : A) used and, B) unused\t 105 Figure 58 Profile of the properties : Apparent Porosity (AP), Water absorption (WA), Bulk Density (BD) and Modulus of Rupture (MOR) for MC-8 : A) used and, B) unused\t 106 Figure 59 Profile of the properties : Apparent Porosity (AP), Water absorption (WA), Bulk Density (BD) and Modulus of Rupture (MOR) for MC-10 : A) used and, B) unused\t 106 xv Figure 60 Profile of the Cold Crushing Strength (CCS) for MC-2, MC-8 and MC-10: A) used and, B) unused \t 107 Figure 61 A) SEM image of MC-2 displaying severe penetration of 1) silver and, 2) bismuth-lead, and 3) a magnified image shown in Figure 61B, (40x), B) Magnified SEM image of MC-2 at point 3 of Figure 61A displaying 1) MgO grain, 2) Cr203 grain, 3) (Mg, Fe)(Cr, Al) 204 spinel, 4) bismuth-lead at MgO- MgO grain boundary and, 5) bismuth-lead at MgO-Cr203 grain boundary, (200x). The corresponding EDS analysis is shown in Table 31 \t 110 Figure 62 A) SEM image of MC-8 displaying severe penetration of 1) bismuth-lead, and 2) a magnified image shown in Figure 62B, (40x), B) Magnified SEM image of MC-8 at point 2 of Figure 62A displaying 1) MgO grain, 2) Cr203 grain, 3) (Mg, Fe)(Cr, Al) 204 spinel, 4) bismuth-lead at MgO-MgO grain boundary and 5) bismuth-lead at MgO-Cr 203 grain boundary, (200x) . The corresponding EDS analysis is shown in Table 32\t 111 Figure 63 A) SEM image of MC-l0 displaying severe penetration of 1) silver and, 2) bismuth-lead, and 3) a magnified image shown in Figure 63B, (40x), B) Magnified SEM image of MC-10 at point 3 of Figure 63A displaying 1) MgO grain, 2) Cr203 grain, 3) (Mg, Fe)(Cr, A1)204 spinel, 4) silver, 5) bismuth-lead at MgO-MgO grain boundary and 6) bismuth-lead at MgO- Cr203 grain boundary, (200x) . The corresponding EDS analysis is shown in Table 33\t 112 xvi Figure 64 A) SEM image of unused MC-2 displaying : 1) MgO grain, 2) Cr2 O3 grain, 3) (Mg, Fe)(Cr, Al)2O4 spinel, and 4) pores, (80x), B) SEM image displaying the precipitated (Mg, Fe)(Cr, Al) 2O4 spinel in the MgO grain, (1500x) . The corresponding EDS analysis is shown in Table 34 \t 113 Figure 65 A) SEM image of unused MC-8 displaying : 1) MgO grain, 2) Cr2O 3 grain, 3) (Mg, Fe)(Cr, Al) 2 O4 spinel, and 4) pores, (80x), B) SEM image displaying the precipitated (Mg, Fe)(Cr, A1)2O4 spinel in the MgO grain, (1500x). The corresponding EDS analysis is shown in Table 35 \t 114 Figure 66 A) SEM image of unused MC-10 displaying : 1) MgO grain, 2) Cr2 O3 grain, 3) (Mg, Fe)(Cr, Al) 2O4 spinel, and 4) pores, (80x), B) SEM image displaying the precipitated (Mg, Fe)(Cr, A1)2O4 spinel in the MgO grain, (1000x). The corresponding EDS analysis is shown in Table 36 \t 115 Figure 67 Elemental profile for the used MC-2 brick through the longitudinal cross- section \t 116 Figure 68 Elemental profile for the used MC-8 brick through the longitudinal cross- section\t 117 Figure 69 Elemental profile for the used MC-10 brick through the longitudinal cross- section\t 118 Figure 70 XRD profile for the used MC-2 brick through the longitudinal cross- section\t 119 Figure 71 XRD profile for the used MC-8 brick through the longitudinal cross- Section\t 120 xvii Figure 72 XRD profile for the used MC-10 brick through the longitudinal cross- Section\t 121 xviii PS Converter KIVCET EF BBOC MC MA ASTM SEM EDS XRD MOR CCS AP WA BD MOE NO, N1, N2, N3 LO, Li, L2, L3 HO, H1, H2, H3 List of Abbreviations and Symbols Peirce Smith Converter KIVCET Electric Furnace Bottom Blown Oxygen Converter Magnesia-chrome brick Magnesia-alumina spinel brick American Society for Testing and Materials Scanning Electron Microscopy Energy Dispersive Spectroscopy X-Ray Diffractometry Modulus of Rupture Cold Crushing Strength Apparent Porosity Water Absorption Bulk Density Modulus of Elasticity KIVCET experimental mixes N = 0% Pb (Barren slag) L = 6% Pb (Low-lead) H = 11% Pb (High-lead) 0 = 0% Cr203 1=5%Cr203 xix Acknowledgements I would like to extend my sincere gratitude to my supervisors Dr . Tom Troczynski and Dr. George Oprea for giving me the opportunity to continue my educational goal of attending graduate studies and giving me the support and guidance during my studies . I would also like to show my appreciation to Rahul Lodha for his help and discussion . I would like to give special thanks to Carmen Oprea, Mary Fletcher, Rudy Cardeno, Gary Lockhart, Ross McLeod, Carl Ng, David Torok, Glenn Smith and Randy Clark for their assistance in using the experimental facilities at UBC. I would also like to extend my appreciation to the industrial sponsors of the research consortium on refractories at UBCeram including Teck Cominco Metals Ltd ., RHI Canada Inc., Clayburn Industries Ltd ., Hatch Associates Ltd. and NSERC for providing the financial support and making my research program possible. Lastly, I would like to express my deepest gratitude to my parents, whom have always been an inspiration to me, and I would like to thank them for their patience, support and encouragement during my years in university . In addition, I would like to thank my sisters and family for their ever-lasting support . xx 1 .0 Introduction Refractories are non-metallic materials that are used for high temperature industrial applications . Their main objective is to provide a barrier between the hot \"inside\" and ambient \"outside\" temperatures . Their greatest challenge is to resist chemical attack in a corrosive environment and to withstand stresses from a mechanical and thermal origin while maintaining their physical and chemical integrity . Due to their high thermal stability, refractories have been used in many applications such as kilns, furnaces, boilers and other applications in the ferrous and non-ferrous industries. In general, refractory bricks are commercially available as pre-fired shaped bricks . Raw aggregate material of various grades and sizes is compacted by mechanical presses and fired at an elevated temperature through a tunnel kiln . The final microstructure consists of large coarse grains (up to millimeters in size) held together by fine bonds or a matrix . At the service temperature, this microstructure must tolerate high temperatures, stresses and ultimately resist corrosive attack by molten metals, slags, fluxes, fumes, vapors and hot gases. In the non-ferrous industry, there has been little progress in the research and development of refractories . Literature sources for the non-ferrous industry are scarce, in particular, for nickel, silver, zinc and lead furnaces . These furnaces include the Peirce Smith Converter for nickel, the KIVCET Electric furnace for lead and zinc, and the Bottom Blown Oxygen Converter for silver. For the latter two, literature for these furnaces is practically 1 nonexistent . It could be said that the major area for future improvement lies in the research and development of refractories for the non-ferrous industry. The Peirce Smith Converter (as shown in Figure 1) is a batch cylindrical reactor where matte, either high in nickel iron sulphide or copper iron sulphide, is converted to a nickel sulfide matte or \"blister\" copper metal, respectively . The reactions that take place in the reactor are achieved by selectively oxidizing the iron in the matte with oxygen producing SO 2 gas and Fe2 SiO4 slag as the by-products . Oxygen is supplied to the matte though what is known as the tuyere . Image Removed for Copyright Reasons Illustration of the: Peirce Smith Converter Figure 1 . The Peirce Smith Converter ril The KIVCET Electric Furnace is a continuous reactor (Figure 2) for the processing of complex ores high in zinc and is carried out in two distinct stages . Smelting first occurs in the reaction shaft, where the initial \"flash\" smelting takes place in an oxygen-fed flame, burning above the molten bath of slag and lead bullion . Lead sulphide is oxidized to lead 2 oxide producing S0 2 gas as the by-product . The second reaction occurs in the electric furnace, where the reduction stage takes place in a coke layer floating on top of the molten slag . The lead oxide is then reduced to lead, resulting in CO gas as the by-product. Separation of the lead bullion from the slag (containing zinc and other valuable metals) occurs here and is then tapped from the furnace . In this stage, the lead must be completely removed before the slag enters into the fuming furnace where the zinc is recovered. Image Removed for Copyright Reasons Illustration of the: The KIVCET Electric Furnace Figure 2. The KIVCET Electric Furnace [1l The Bottom Blow Oxygen Converter (Figure 3) is a primary cupellation process used to separate noble metals from base metals . The BBOC is a process of extracting Ag from a high Ag-Pb bullion . The batch converter turns the impure metals into oxides such as Pb to PbO (litharge) and the silver is then separated from the impurities . This reaction takes place by injecting oxygen (from a lance system located at the bottom of the furnace) directly into the metallic bath . At the end of the cupellation process, noble metals settle at the bottom, 3 separating from the oxidized impurity metals as a slag. The impurity metals include lead, zinc, antimony, bismuth, selenium and tellurium. Image Removed for Copyright Reasons Illustration of the: Bottom Blown Oxygen Converter Figure 3 . The Bottom Blown Oxygen Converter [1i In this current research, the focus has been set on establishing the failure mechanism(s) of basic bricks used in the above-mentioned furnaces . These bricks are selected on a trial and error basis due to limited literature sources and unknown reasons for premature failure . In order to extend their service life, the primary goal is to identify the main failure behaviour. As part of an on-going collaborative research program between UBC's Refractories group and Teck Cominco Metals Ltd ., RHI Refractories Ltd., Clayburn Industries Ltd . and Hatch Associates Ltd ., experiments were carried out as part of a complex study in order to evaluate the performance of several selected magnesia-chrome and magnesia-alumina spinel candidate bricks against non-ferrous slag . 4 Abstract In the non-ferrous industry, literature sources on refractories for nickel-copper and zinc-lead smelting and converting are practically nonexistent and the selection of refractories is mainly based on corrosion models or knowledge accumulated from the steel industry. In the present study, the corrosion behaviour of several magnesia-chrome (MC) and magnesia-alumina spinel (MA) bricks against fayalite type slags was investigated and the role of the spinel phases was highlighted. The micro structural changes due to the refractory- slag interactions were studied using SEM\/EDS and XRD and the results were discussed versus the initial microstructure and phase composition of the unused bricks . The experimental results revealed that the corrosion resistance of the MC bricks was superior to the MA bricks against KIVCET slags in static and dynamic conditions . As a result of the interaction between MgO from MC bricks and the slag, a modified forsterite phase (Mg, Fe, Zn, Ca) 2 SiO4 was formed, which destroyed the precipitated complex spinel bonds at the grain boundaries of periclase and magnesia-chromia spinel . Furthermore, both MgO and MgO-MgAl2O4 spinel in the MA brick dissolved into the slag, which resulted in modified forsterite phases of (Mg, Fe, Zn, Ca) 2 SiO 4 and (Mg, Fe, Zn)(Fe, Al) 204 complex spinels, respectively. In addition, the accretion formation in the KIVCET furnace was investigated through solubility experiments of Cr203 in the KIVCET slag with various amounts of lead, which revealed that the net contribution of Cr 203 to the spinel formation is the highest in the barren (no Pb) slag, followed by high-lead (11% Pb) and it is the lowest for the low-lead ii (6% Pb) slag . The amount of spinel solid solution increased consistently with increasing Cr203 dissolved and the PbO existent in the slag. From examinations of several used bricks from the tuyere area of a Peirce Smith nickel converter, it was found that the corrosion is due to the interaction of the partially oxidized matte penetrating deep into the brick and the magnesia grains forming (Mg, Fe, Ni, Co) XOy spinels . Analyses of brick samples used in the KIVCET Electric Furnace roof identified deep reaching sulphation, which weakened the bonding phase between coarse magnesia grains. In the Bottom Blown Oxygen Converter, a highly aggressive lead and bismuth oxide rich slag penetrated deep into the brick, which destroyed the grain boundaries, causing the refractory to be easily eroded at the refractory-slag interface. Our studies concluded that the spinel phases, either as magnesium chromate, magnesium aluminate or complex spinel [(Mg, Fe)(Cr, Al, Fe) 204], enhanced the corrosion resistance of a basic refractory to fayalite type slags from the non-ferrous smelting and converting furnaces. iii 2.0 Literature Review 2.1 Spinel Phases in Basic Refractories In non-ferrous furnaces, magnesia-chrome bricks are the primary refractories used due to their superior resistance in corrosion environments (such as molten liquids and gaseous environments) at high operating temperatures . Magnesite (MgCO3) and chrome ore (FeCr2 O4) are the raw materials that are used in the production of refractory bricks. However, various magnesia spinels with chromia, alumina and iron oxides (Fe2+ and Fe3+ ) are major contributing phases to the corrosion resistance against non-ferrous slags. Corrosion in this context is referred to as a combination of both chemical dissolution and erosion. 2.1 .1 Crystal Chemistry The general formula that represents the basic spinel chemistry of interest (chromite spinels) is given as AB2O4. There are also deviations from this stoichiometry but the important spinels are of the AB2O4 stoichiometry where A is a divalent iron or magnesium cation, B is a trivalent iron, aluminum or chromium cation, and 0 is an oxygen anion . However, the structure of the spinel is cubic and there are eight formula molecules per unit-cell . The unit cell crystal lattice is therefore expressed as A8B16O32\u2022 The array of oxygen ions form an almost perfect cubic close-packed lattice with cations distributed among the crystal's tetrahedral and octahedral sites . This array of anion close packing provides 64 tetrahedrally coordinated cation sites and 32 octahedrally coordinated cation sites, in which only one eighth of the 64 tetrahedral sites and half of the 32 octahedral sites are occupied by cations, 5 resulting in a total of 24 cations in the spatial arrangement of eight formula units . These 24 cations and 32 anions comprise of a face-centered cubic cell with an approximate cell edge of 8.0-8 .4 A, depending on the cations involved . Due to unfilled cation sites, different arrangements of the cations are possible . In a normal spinel structure, bivalent cations occupy the tetrahedral sites and trivalent cations occupy the octahedral sites . For inverse spinels, half the trivalent cations occupy the tetrahedral sites and both bivalent and trivalent cations occupy the octahedral sites [21 . The cation radius governs the cation interstitial sites in the crystal lattice by stoichiometric and\/or cation-deficient substitutions and by charge requirements . The cation radii allowed for the A or B sites of chromite spinels have been set at 0 .44 - 1 .0 A. In this range, a variety of different cations may substitute into chromite spinels . However, if the cation is less than 0.44 A, for example Si4+ at 0.39 A, an olivine structure results (Mg 2SiO4). On the other hand, if the cation is larger than 1 .0 A, as for Cat+ at 1 .06 A, a non-cubic but orthorhombic structure results (CaFe2O4) [21 . Spinel solid solutions of both substitutional and interstitial are possible . For the interstitial case, impurity cations such as Ca t+ may arise at interstitial sites . The ionic radius of the cation must be relatively small in comparison to the anion in order for the cation to occupy a vacant site. Once the cation occupies a vacant site in the crystal lattice, the crystal must maintain the conditions of electroneutrality . As a result, the crystal involves a lattice defect by the means of a cation-vacancy or an anion-vacancy. In addition to the impurities causing vacancies, cations residing in the crystal lattice may leave their normal position and reside at 6 the interstitial site due to possible lattice defects or various excitations . There is no change in charge because the crystal lattice maintains its neutrality. For the substitutional case, the cations and anions from the impurity host may substitute for the host lattice ion . For example, an impurity of Ca2+ and o2 - could substitute for Na+ and Cl- ions, respectively. The ionic size and charge must be very similar to achieve the substitution and maintain charge neutrality. In addition to stoichiometric substitutions, non-stoichiometric cation- deficient structures may be created with charge neutrality by the replacement of 3 divalent cations by 2 trivalent cations and a cation vacancy . For example, iron oxide can exist in the Fee+ and Fe3+ states. The formation of a trivalent ion will disrupt the lattice by introducing an excess positive charge . In order to offset the excess charge, a defect would occur by the means of a divalent vacancy with every two trivalent ions [3l 2.1 .2 Magnesia-Chromia Spinel The structure of the Mg-Cr spinel is an AB2O4 type with a close packed arrangement of 0 anions with the cations situated at the tetrahedral and octahedral sites . The \"A\" cation is the divalent magnesium ion positioned in the tetrahedral site and the \"B\" cation is the trivalent chromium ion situated in the octahedral site [3l . This spinel, also known as picrochromite (MgCr2 O4), is a conventional component in basic refractories and is an important secondary solid solution phase and bond between MgO and Cr2O3 . In the phase diagram of the MgO- Cr2O3 system (Figure 4), the MgCr2 O4 spinel is the only phase present with MgO and has a melting point at 2400\u00b0C [4] . 7 Image Removed for Copyright Reasons Illustration of the: Phase diagram of MgO-Cr20 3 system Figure 4. Phase diagram of MgO-Cr20 3 system [41 When the MgCr2O 4 spinel comes in contact with fayalitic slags, the spinel changes to a FeCr2O 4 or a (Mg,Fe)Cr204 spinel according to the following equation [51 : FeO(in liquid slag) + MgCr204(s) = FeCr204(s) + MgO(in liquid slag) 2.1 .3 Precipitated Complex Spinel In magnesia-chrome bricks, the chrome ore, which contains chromia, magnesia, iron oxide, alumina, silica, calcia and other impurities, gives way to precipitated secondary complex spinels . Precipitated complex spinels play an important role in determining the characteristics of the refractory such as the strength at room or elevated temperature, corrosion resistance and spall resistance . The spinel formed could have iron in the bivalent 8 state or iron and alumina in the trivalent state, substituting the magnesia or chrome respectively, in which the structure may be denoted as (Mg, Fe)(Cr, Fe, A1) 2 O4 . The oxides (Cr2O 3, Al2O 3, FeO, Fe2 O3) are soluble in periclase and are formed from the precipitation from the liquid phase or exsolution precipitation from the magnesia grains upon cooling [6l . Ikesue et al [6I found that the formation of the complex spinel is highly influenced by the cooling rate and is important in controlling the characteristics of the brick. By controlling the cooling rate, the growth of the spinel could be controlled and the physical properties of the brick could be changed . The quantity and the size of the complex spinel were found to increase significantly with a slow cooling rate. 2 .1.4 Secondary Spinel Considerations In order for secondary spinels to form in magnesia-chrome bricks during manufacturing, the chrome content must be significantly high due to the high solubility of chrome in the periclase and silicate phases at temperatures above 1600\u00b0C ['I . Studies have shown that the solubility of chrome ore in magnesia-chrome bricks containing 1 .75 percent SiO2 is about 23 percent at 1600\u00b0C and 29 percent at 1700\u00b0C . Clearly, the chrome content in the brick should be in excess of 23 percent . Furthermore, a high silica containing brick will require a higher amount of chrome ore in order to reach the chrome saturation limit of silica . By using higher purity materials and sintering at higher temperatures (ie . above 1800\u00b0C), it is possible to decrease the silicate content from 3-9 wt.% to 1-2 wt .% resulting in a more chemically resistant bonding structure . Direct bonded bricks generally contain less than 3 wt .% SiO2 [7i . 9 2.1.5 Magnesia-Alumina Spinel Although magnesia-chrome bricks are superior in corrosive environments, there are concerns surrounding the formation of hazardous Cr6+ species in chrome containing refractories [101 With this potential risk, interest has been developed for magnesia-alumina spinels (MgAl2 O4) as a potential alternative to chrome. The MgAl2 O4 spinel is ideal due to its exceptional mechanical, chemical, and thermal properties at ambient and elevated temperatures 181 . Structurally, the Mg-Al spinel is similar to the Mg-Cr spinel, as it is an AB 2 O4 structure with a close-packed arrangement of 0 anions with the \"A\" cations of Mg t+ located at the tetrahedral sites, and \"B\" cations of A1 3+ situated in the octahedral sites . The spinel, with a melting point at 2105\u00b0C (Figure 5), is accompanied by a large linear expansion of 2 .45% (7.35% volume increase) when formed directly from MgO and Al2O3 . This volume increase can be beneficial in the healing of thermal stress cracks or in improving resistance to erosion [81 . In several research studies for cement rotary kilns, the addition of MgAl2O4 to magnesia- based bricks has improved their thermal shock resistance, improved erosion behaviour and extended service life up to two to three times longer than conventional magnesia-chrome bricks as a result of the large difference in thermal expansion behaviour between magnesia and spinel [9] 10 Image Removed for Copyright Reasons Illustration of the: Phase diagram of MgO-Cr203 system Figure 5 . Phase diagram of MgO-Al 203 system [41 When the MgAl2 O4 spinel is in contact with fayalitic slags, the spinel changes to a (Mg, Fe)Al204 spinel according to the equation 151 : FeO(in liquid slag) + MgAl2O4(s) _ (Mgx, Fe)Al204(s) + Mg(1-x)O(in liquid slag) Out of several different spinels tested with slag, the highest amount of MgO and Al 203 dissolved into the slag occurred when the MgAl2 O4 spinel was used [51 . 11 2.2 Refractory-Slag Interactions 2.2.1 Microstructure In order to control the physical and chemical characteristics of a brick, the morphology and properties of the initial raw materials must be controlled. By selecting raw materials of different crystallite size and grade, factors such as the microstructure, porosity, strength and chemical nature of the grain, matrix and bond can be controlled. Large grains, whether single crystal or polycrystalline, must be fully reacted and well crystallized in order to obtain the full potential to resist chemical dissolution against slags [io] For a brick to obtain its full potential to resist slag penetration, a large connected network of bonded refractory grains is desired . In order to obtain this bonded network, the brick must be sintered at a high temperature to reduce the porosity and pore size . The firing temperature should be well above the brick's intended temperature of use (>1700\u00b0C). Furthermore, using graded grain sizing improves the packing density and the number of nearest neighbouring particles thereby increasing the points of bonding contact . Although the surface to volume ratio of the grains increases as their size decreases, the smaller grains are more vulnerable to erosion and chemical dissolution . As the smaller grains dissolve into the slag, and the brick at the slag interface recedes, the large grains remain embedded rigidly in the graded system, protecting the underlying grains and matrix against erosion, abrasion and chemical dissolution. Therefore, by using the graded sizing, the resistance to corrosion of the grain-matrix-pore system as a whole is improved [io] 12 2.2.2 Mineralogical Considerations The mineralogical phases in a magnesia-chrome brick after the brick has been fired are significant as they are related to chemical dissolution at the service temperature . Goto et al [111 studied the mineralogical phases present in a direct bonded magnesia-chrome brick composed of: 66.9% MgO, 22 .7% Cr2O3 , 3 .2% Al2O3 , 5 .5% Fe 2 O3, 1 .0% SiO2 and 0.7% CaO . Precipitated secondary spinels were observed within magnesia grains and along magnesia-magnesia grain boundaries and the large coarse magnesia and chrome ore grains appeared to be in direct contact at the grain boundaries . Some interconnected pores formed a continuous gap around the grains (open pores), while others were isolated (closed pores). SEM analysis identified two different silicate phases isolated between magnesia and chrome ore grains. The composition of these silicate phases was detected by EDS and was identified as containing magnesia silica and calcia or only magnesia and silica . Although these two phases could not be detected by XRD, they were in the MgO-CaO-SiO 2 system and identified as monticellite (CaO\u2022MgO\u2022SiO2) and forsterite (2MgO\u2022SiO2), respectively. Silicate phases were found throughout the microstructure but a larger percentage of monticellite was found in the matrix near primary chromite while a larger percentage of forsterite was found near magnesia grains. Cherif et al [121 determined that the most likely silicates and ferrites found in magnesia refractories are those shown in Table 1 . Monticellite (CaO\u2022MgO\u2022SiO2) and dicalcium ferrite (2CaO\u2022Fe2 O3) are the lowest melting compounds formed at 1495\u00b0C and 1435\u00b0C respectively. The CaO\/SiO 2 (C\/S) ratio of the brick is the controlling factor dictating the formation of these lowest melting compounds and Table 2 displays the silicates and ferrites 13 most likely to be formed in the presence of magnesia and the effect of varying C\/S ratios. The C\/S ratio between 0 .94-1 .47 is the critical range governing the formation of monticellite and merwinite (3CaO\u2022MgO .2SiO 2) in which the latter has a melting point of 1575\u00b0C. Table 1 . Silicates and ferrites most likely to be found in magnesia refractories [12 1 Mineral Formula Melting Temperature(\u00b0C) Periclase MgO 2800 Forsterite 2MgO\u2022SiO2 1890 Monticellite CaO\u2022MgO\u2022SiO 2 1495 Merwinite 3CaO\u2022MgO .2SiO 2 1575 Dicalcium Silicate 2CaO\u2022SiO 2 2130 Magnesium Ferrite MgO\u2022Fe2O3 1750 Dicalcium Ferrite 2CaO\u2022Fe2O3 1435 Table 2. Silicates and ferrites most likely to be formed in magnesia refractories [121 CaO\/SiO 2 Silicate Ferrite < 0.94 Forsterite, Monticellite Magnesium Ferrite = 0.94 Monticellite Magnesium Ferrite 0.94 -1 .47 Monticellite, Merwinite Magnesium Ferrite = 1 .47 Merwinite Magnesium Ferrite 1 .47 -1 .87 Merwinite, Dicalcium Silicate Magnesium Ferrite = 1 .87 Dicalcium Silicate Magnesium Ferrite > 1 .87 Dicalcium Silicate Dicalcium Ferrite, Magnesium Ferrite 2.2.3 Chemical Considerations To understand the complex chemical attack of basic refractories by molten liquids, it is essential to recognize the chemical effects of the reactions between magnesia-chrome and magnesia-spinel based bricks with various slags . 14 2.2.3.1 Magnesia-Chrome Refractories Several research studies on the corrosion behaviour between magnesia-chrome bricks and various slags were reported. Takahashi et al [13] identified a correlation between the composition of an MgO-Cr2O3 brick and the chemical resistance using the dynamic rotary test. He found that the resistance to chemical dissolution by slag increased when the ratio of Cr2O 3\/MgO in the brick increased, and the resistance to chemical dissolution decreased with increased Fe2 O3\/Cr2 O3 or Al2O3\/Cr2 O3 ratios. At a high Cr2O3\/MgO ratio, precipitated secondary chrome spinels had higher melting points and lower solubility in SiO 2 rich slags than high ratios of Fe2 O3\/Cr2O3 or Al2O 3\/Cr2O3 . In addition, Ichikawa et al [141 found that the Al2O3 in the brick lowered the dissolution resistance and Fe 2 O3 had no significant effect. This correlation is due to the appreciable smaller ionic radius of the A1 3+ (0 .053 nm) ion in comparison to the ionic radius of Fe 3+ (0.069 nm) or Cr3+ (0.075 nm) ions . With a smaller ionic radius, it would enable the A1 3+ to dissociate from the secondary spinel and dissolve in the slag much more easily than those with a larger ionic radius. Wiederhorn et al [15] examined the effects of coal slag on the microstructure of an MgO- Cr2 O3 brick. It was proposed that the A1 3+ , Fee+ and Fe3+ from the slag entered the spinel of the brick at the hot face, while the Mg t+ and Cr3+ from the refractory migrated to the slag. The A1 3+, Fe e+ and Fe3+ components would reduce the melting temperature of the spinel and enhance the dissolution of the spinel into the slag . Within the penetrated region of the refractory, the slag components (A1 3+, Fee+ and Fe3+) would replace Cr3+ and Mgt+ in the refractory, changing the composition of the slag from one close to anorthite (CaAl 2 Si2 O8) to one close to diopside (CaMgSi2O6) . The change in slag composition decreased the viscosity 15 of the intergranular glass phase in the refractory creating a more reactive glass. Simultaneously, the grains in the refractory matrix that lay between the larger coarse grains began to grow. The growth, coupled with the transport of matter from the slag to the spinel grains of the brick, caused the hot face to bloat, crack and eventually spall from the main body of the brick. Rose et al [161 studied Basic Oxygen Furnace (BOF) slags (containing CaO-FeO-Fe2O 3-SiO2) with additions of MgO, Al203 and TiO2. It was determined that the addition of TiO 2 to the slag accelerated the solubility of MgO while the slags were found to be less corrosive when the concentration of MgO in the slag increased. The addition of Al203 to the slag did not change the dissolution rate of MgO but the combination of Al 2 O3 with CaO decreased the dissolution of MgO in the slag. Nooka et al [171 studied the interaction between fayalitic type slags and the phases in the MgO-Cr203 brick. Iron oxide from the slag was able to dissolve into the chrome grains, the magnesia grains, and the spinel structure, or combine with free magnesia to form magnesioferrite (MgFe 2 O4). When the iron oxide (FeO) dissolved into MgO, magnesiowustite ((Mg,Fe)O) was formed . Although magnesiowustite and magnesioferrite are high melting compounds (>1750\u00b0C), their formation could accelerate the dissolution of MgO into the slag at the interface. Furthermore, when FeO dissolved into the magnesia- chromia spinel (MgCr 2 O4), Cr3+ was replaced by Fe3+ forming MgFe2O4, which weakened the bonds in the brick structure. Therefore, the strong primary spinel bonds between the periclase and the chrome ore grains became susceptible to dissolution . In addition, silica 16 from the slag was also able to combine with free magnesia to form forsterite . Although forsterite is a high melting phase (1890\u00b0C), the formation left the microstructure with modified chemical properties, which made it susceptible to further dissolution as described by the following reactions [171 : 3(2FeO . SiO 2) + 02 + 6MgO = 2Fe304 + 3(2MgO\u2022SiO2) Fayalite Magnetite Forsterite 2Fe3 O4 + 0.502 + 3MgO = 3(MgO\u2022Fe 2 O3 ) Magnesioferrite 2Fe3 O4 + 0.502 + 3(2MgO\u2022SiO2) = 3(MgO\u2022Fe2O3) + MgO\u2022SiO2 Magnetite Forsterite Magnesioferrite Clinoenstatite In this sequence of reactions, the high melting forsterite phase is degraded to a pyroxene phase (MgO\u2022SiO 2), which would be molten at the converter temperatures . It was also found that the chrome grains in the brick remained chemically unaltered and were believed to be detached at the hot face by abrasion. Chrome grains and secondary spinel phases in the refractory were more resistant to converter slags and were usually washed out instead of chemically corroded 1171 . Petkov et al 118 examined the effects of a conventional, high fired, direct bonded, magnesia- chrome brick in contact with a copper metal and anode slag (>50 wt .% CuO X , 30-35 wt .% FeO and 7-8 wt .% SiO2). He found that in the slag accretion layer, a complex spinel was formed from the reaction between slag (Al 203, FeO, CuO X ) and refractory (MgO, Cr203) forming (Mg2+, Cu(+) or (2+0, Fee+)(A13+, Cr3+, Fe3+ )204. The silica in the slag at the hot face 17 also interacted with the refractory forming forsterite (2MgO\u2022SiO 2 ). A considerable amount of slag components (FeO-20 wt .% and CuO,, 5 wt.%) was found to diffuse into periclase grains near the hot face and a high concentration of FeO was also found diffused at the rim of the primary chromite grain . In addition, the slag components also diffused into the spinel components, more so in the secondary spinel than the primary spinel due to the ease of diffusion along grain boundaries than bulk grain diffusion. Evrard et al [19] performed a study on chrome-magnesia refractories with slag from a Hoboken Copper converter. He determined that the bivalent cations of Fe, Zn and Ni from the slag diffused into the chrome ore forming a complex spinel of (Mg, Fe, Zn, Ni)(Fe, Al, Cr) 204. In addition, the slag components diffused into the periclase grain forming a nickel bearing periclase (Mg, Ni)O, which was followed by dissolution into slag forming a modified forsterite phase (Mg, Fe, Ni) 2SiO4 with SiO2. This mechanism had a significant contribution to the global corrosion behaviour and is typical in converter slag attack. 2.2 .3 .2 Magnesia-Spinel Refractories Several studies examined the corrosion resistance of magnesia-alumina spinel based bricks in fayalitic type slags. Schlesinger et al [201 observed that alumina based and high-spinel based refractories were highly penetrated with molten ferrous silicate slags at 1300\u00b0C . It was proposed that Fee+ may substitute for Mgt+ in both the periclase matrix and MgAl 2O4 spinel grains by diffusion, resulting in a magnesiowustite solid solution and a (Mg,Fe)Al 204 spinel . With the formation of the solid solution spinel, the refractory could fracture due to the new spinel formation and volume expansion. In an alternative reaction, Fee+ may 18 substitute for Mgt+ in both the periclase matrix and spinel grain forming again magnesiowustite while alumina dissolved in the liquid slag. Petkov et al [18] examined several chrome-free bricks with copper metal and anode slag . In the spinel brick consisting of MgO with complex spinel (Mg, Fe)(Al, Fe) 2O4, the slag components (FeO, Al2 O3 and CuOX) reacted with MgO, forming a complex spinel of (Mg, Fee+)(Al)2O4 with up to 3 wt .% of CuO, dissolved in this phase . The slag components also diffused into the spinel phase forming a new complex spinel with CuO,t likely to be (Mg2+ , Cu(+) or (2+) , Fee+)(A13+, Fe3+ )2O4. In addition, the SiO2 from the slag reacted with MgO forming a forsterite phase (MgO\u2022SiO2). In the MgO and FeAl2O4 (hercynite spinel) brick in contact with slag, the same reactions occurred as the former brick where the MgO and the FeAl 2 O4 spinel formed forsterite and a (Mg, Fee+)(Al)2 O4 spinel with CuO X dissolved in this phase. The author also stated that the tested chrome-free alternatives could not match the performance of the magnesia-chrome bricks. In a study by Goto et al [21] of an MgO-Al2O3 refractory with calcium aluminosilicate slag using the static test at 1400-1450\u00b0C, he found that the alumina existing in the slag dissolved the fine MgO matrix forming a secondary spinel of MgAl 2 O3. The secondary spinel grains were often observed with an MgO core and appeared to replace the fine MgO matrix. Further dissolution of the MgO grains into the slag and the removal of Al 2 O3 by precipitation of the secondary spinels lead to the crystallization of monticellite (CaO\u2022MgO\u2022SiO 2) and merwinite (3CaO\u2022MgO . 2SiO2). The general phase sequence forming in the refractory from the slag to the refractory was secondary spinel, monticellite and merwinite . 19 2.2.4 Slag Property Considerations Cherif et al [12] have shown that the corrosion rate of a refractory can also be correlated through the slag basicity (as a ratio of CaO \/ SiO 2) and the weight percent of the MgO content . It was identified that the minimum corrosion rate occurred when the (CaO + MgO) \/ SiO2 ratio was greater than 2 . This increase in slag basicity was a result of an increased amount of MgO content in the slag, which consequently diminished the corrosion by dissolution . In addition, the increasing level of MgO in the slag increased corrosion resistance. An increased level of MgO content in the slag would cause the slag to be more viscous, giving rise to less erosion of the refractory lining at the slag interface . Chemical wear was decreased by the increased content of MgO, which lowered the chemical affinity between the refractory and the slag . The MgO containing components in the slag altered the iron-oxygen equilibrium in the sense that the iron oxide content of the slag was reduced. Chromite grains remained ultimately un-attacked when exposed to fayalite type slags [22] . These chromite grains were usually washed away due to the dissolution of the bonding phase. A correlation between the resistance to erosion and the Cr 203 content was found . It was found that the erosion resistance of chrome spinel grains against smelter slags increased as the ratio of Cr203 \/ (Al 203 + Fe203) (CAF) increased . It was found that the CAF ratio of 1 .06 and 0.84 were both resistant against slags than a basicity of: SiO2\/FeO (SF) = 1 :3. At a SF ratio of 1 :2, only the chrome grain with the low CAF ratio was attacked . Proceeding further to an even lower basicity level to a SF ratio of 2 :1, both types of chrome gains were attacked [22] . 20 2.2.5 Thermo-Mechanical Considerations The phenomenon associated with freezing of a penetrating liquid is known as slabbing, peeling or chemical spalling . The occurrence is often characterized as a fracture of the refractory brick by one of the following [10] : 1. Temperature gradient : sufficient uneven heating or cooling resulting in large stresses causing spall failure 2. Compression in the brick structure : shear failure from the restrained expansion of the brick due to the rise in temperature 3. Variations in thermal expansion : difference in thermal expansion coefficient between the un-penetrated brick and the penetrated surface layer, due to slag infiltration, resulting in stresses causing the surface layer to spall In metallurgical processes, slabbing is often caused by the third method [101 . Molten liquids, such as slags and mattes, will penetrate into the brick at the surface causing a vitrified or slagged surface as shown in Figure 6 . The penetrated liquid in the slagged surface may cause weakening in the brick by de-bonding of the grains from the matrix . 21 \u2014Hot Face Figure 6 . Separation due to a vitrified or slagged layer [lo] Penetrated End In processes that require frequent shutdowns or intermediate delay, the brick lining may be allowed to cool causing the penetrating liquid to solidify. This solidification of the liquid is accompanied by a volume contraction . Once the process is restored to its original operating temperature, the solidified liquid may also be restored to its liquid form . This freeze-thaw cycling of the liquid is likely to be accompanied by a change in volume with each change of phase. The change in volume subjects the brick to a shear strain parallel to the hot face . The failure occurs where the brick is weakened ; specifically a fracture is induced at the penetrating front between the penetrated and the un-penetrated regions of the brick . Often a layer will break off causing the brick to recede until the refractory lining becomes too thin. Typical depths of penetration at which slabbing occurs are from 12 mm to well over 50 mm 22 2.3 Industrial Non-Ferrous Converting Furnaces 2.3.1 Nickel-Copper Converting and Refining (Peirce Smith Converter) The service life of a PS converter can range from 2 to 6 months, which corresponds to approximately 250 to 400 charges . It is dependent upon the corrosion resistance of the brick lining, in particular, the bricks of the tuyere lining . The region around and at the tuyeres has the shortest service life compared to other regions in the converter due to the tuyere being subjected to various conditions such as [23] : \u2022 Aggressive siliceous slag high in iron oxide (FeO) \u2022 Difference in thermal expansion between the penetrated and the un-penetrated regions of the brick \u2022 Attack by metal oxide phases (such as NiO, Cu 2O) \u2022 Thermal shock generated by thermal cycling \u2022 Mechanical damage caused by pneumatic punchers in order to clean the tuyeres \u2022 Changes in atmosphere from reducing to oxidizing and vice versa \u2022 Highly abrasive and high turbulent slag contact Barthel et al [241 studied the corrosion mechanisms in a tuyere brick of a PS copper converter. It was found that melt infiltration was the dominating corrosion mechanism in comparison to wear by crack formation . At the hot face region, sulphides from the matte were found to migrate into the brick where it was converted to metallic copper and copper oxide due to the increasing oxygen pressure towards the cold face of the brick . The order of reactions that 23 occurred with the increasing oxygen pressure was Cu 2S -' Cu --> Cu2O -~ CuO. Cu2O was found to highly wet between periclase grains up to 40 mm from the hot face, which resulted in a reaction product of delafossite (CuFeO 2) from the reaction with the brick . In addition, Utigard et al [251 found a correlation of the interaction between copper mattes and a magnesia-chrome brick versus temperature . The rate of infiltration rapidly increased when the temperature increased from 1100\u00b0C to 1300\u00b0C. The rate of dissolution also increased but only at a temperature increase from 1100\u00b0C to 1200\u00b0C . Above 1200\u00b0C, there was no significant change in the dissolution rate . It was also observed that the metallic copper and copper mattes were infiltrated into the pores where the matte oxidized within the pores of the brick. The oxidation process from copper matte to copper oxide would lead to a volume increase resulting in cracking and spalling of the brick within. In addition to the dissolution of the brick by molten liquids, thermal-mechanical stresses are of particular concern due to the thermal cycling between charges. It was reported by Goncalves et al [26] that at approximately 25 mm distance from the tuyere tip, the temperature could change more than 500\u00b0C within a few seconds . This drastic temperature change would lead to a freeze and melting cycle of the infiltrated matte, which could lead to the rapid deterioration of the refractory . This freeze-thaw cycle could occur up to 80 times daily and the waiting time could be up to 105 minutes between heats . The hypothesized major wear mechanism in the tuyere lining is spalling caused by liquids penetrating and solidifying in the joints or in the brick . 24 2.3.2 Zinc-Lead Smelting and Converting (KIVCET Furnace) Literature on the corrosion behaviour in lead-zinc flash smelting furnaces is limited, in particular, studies on the KIVCET furnace are practically non-existent . In a study conducted by Oprea et al [271 , the failure mechanisms in a lead and zinc Flash Smelting furnaces were discussed. The slag was identified as a composite phase composed of a crystalline zinc ferrite (ZnFe2O4) phase and silicate liquid phase composed of PbO and CaO . The slag temperature in the lead and zinc furnaces ranged from approximately 1000\u00b0C to 1250\u00b0C and the slag was a fayalitic type, which contained high contents of lead and zinc . Below the slag line, acidic liquid phases would be able to penetrate through small pores dissolving the matrix. Through bigger pores, the slag would penetrate as a whole and phase separation would occur in the refractory. As a result of separation, the liquid phase would further attack the matrix while the formation of the crystalline phase would contribute to the opening of cracks due to the large thermal expansion. The strain due to the thermal expansion would bring a stress greater than the strength of the matrix causing cracks and an unstable structure. The refractory grains would be easily eroded away due to the loose texture at the reaction interface. Above the slag line, Zn fumes were oxidized to ZnO and were deposited on the refractory surface and in pores up to 10 mm from the hot face . The ZnO reacted with MgO and Fe 2 O3 in the brick, producing either a ZnO-MgO solid solution or zinc ferrite ZnFe 2 O4 . Both reactions would be accompanied by a volume expansion resulting in cracking at the hot face. Dissolution would follow as the refractory components dissolved in the liquid phase at the hot face. It was assumed that cracking and dissolution are the main wear mechanisms above the slag line . 25 3.0 Scope and Objectives The present research is an ongoing collaborative project between UBC's Refractories Group, Teck Cominco Metals Ltd. (TCML), RHI Canada Inc ., Clayburn Industries Ltd ., and Hatch Associates Ltd . The main goal is to investigate and establish the failure behaviour of the basic bricks in the Peirce Smith Nickel Converter at Falconbridge Ltd . in Sudbury, Ontario, and the KIVCET Electric Furnace and the Bottom Blown Oxygen Converter at TCML in Trail, British Columbia, in order to extend the service life of the refractory lining. The objectives are as follows: Assessment of the mineralogical phases of basic bricks against liquid slags, mattes and gaseous phases \u2022 Emphasis on magnesia-chromia and magnesia-alumina spinel to their corrosion resistance against liquid slags against non-ferrous smelting and converting furnaces. \u2022 Investigation of the contribution of Cr2O3 to the excessive accretion formation in the KIVCET furnace. \u2022 Study of the microstructural changes due to refractory-slag interactions through laboratory experiments and industrial trials. \u2022 Evaluation of the changes in physical properties due to the penetration and corrosion of liquid slags, mattes and gaseous phases (mainly SO 2 and CO) through laboratory experiments and industrial trials . 26 II . Establish the corrosion mechanism of basic bricks used in industrial furnaces for non-ferrous smelting and converting \u2022 Emphasis on the Peirce Smith Converter, KIVCET Electric furnace and the Bottom Blown Oxygen Converter. \u2022 Correlate the laboratory and industrial trial results in order to establish the main corrosion process resulting from refractory-slag interactions. \u2022 Evaluate the contribution of the slag or matte penetration to the changes in the physical properties of the brick during the life in service. \u2022 Investigate the contribution of gas-solid reactions, mainly SO 2 and CO, to the corrosion mechanism of basic bricks in non-ferrous furnace environments. III . Evaluate the risk of formation of Cr6+ in chromium-containing basic bricks due to the interactions with liquid or gaseous phases in the furnace atmosphere. To achieve our goals, the following research tasks were conducted: 1 . Laboratory evaluation of the corrosion behaviour of magnesia-chrome and magnesia- alumina spinel bricks in contact with KIVCET slag by: \u2022 Slag-refractory dissolution using the Slag Cup Test or Static Corrosion Method \u2022 Slag-refractory corrosion using the Rotary Slag Furnace or Dynamic Corrosion Method 27 2. Solubility evaluation by XRD through: \u2022 Dissolution of chromium oxide (Cr203) powder in no-lead, low-lead and high- lead containing KIVCET slags 3 . Leaching evaluation by simulated acid rain for: \u2022 Hexavalent chromium (Cr6+) of several used magnesia-chrome bricks where potential hazardous compounds may form from the interaction between the compounds of the slag and the chrome (Cr 3+) of the refractory 4 . Microstructural studies by SEM, EDS and XRD on slag-refractory interfaces on: \u2022 Magnesia-chrome and magnesia-alumina spinel bricks from the Static and Dynamic Corrosion Method \u2022 Magnesia-chrome bricks from the Peirce Smith Nickel Converter, the KIVCET Electric Furnace Roof and the Bottom Blown Oxygen Converter 5. Evaluation of the physical properties by comparing the used bricks to their respective unused bricks by Post-Mortem Failure Analysis . The techniques include: \u2022 Visual inspection \u2022 Physical measurement \u2022 Strength Tests : Modulus of Rupture and Cold Crushing Strength \u2022 Absorption Tests : Apparent Porosity, Water Absorption and Bulk Density 28 4.0 Experimental Procedures 4.1 Laboratory Experiments The corrosion performance (erosion and dissolution) of several different bricks, in contact with KIVCET slag, was tested by three laboratory testing methods, which include : 1) the Static Corrosion Method, 2) the Dynamic Corrosion Method and, 3) the Solubility Method. In order to evaluate the potential formation of hexavalent chromium (VI) compounds from the interaction of the slag compounds and chrome ore of the refractory, a Leaching Method was performed. 4.1.1 Static Corrosion Method The slag cup test or the static corrosion method was used to characterize the chemical interaction at the slag-refractory interface and to assess the severity of the dissolution and penetration by stagnant slags . The test was performed in a core-drilled refractory crucible as illustrated in Figure 7 . The crucible was prepared by cutting a 76 mm x 76 mm x 76 mm specimen from a larger brick sample and core drilled to a depth of approximately 45 mm at a diameter of 25 mm. The crucible was filled with 55g of KIVCET slag, sealed with a refractory lid using a thermal set mortar and placed in a high temperature furnace . The temperature was ramped up at a rate of 400\u00b0C per hour to an elevated temperature of 1300\u00b0C. The sample was then allowed to soak at 1300\u00b0C for 5 hours to promote the slag-refractory interaction . Argon was used as a protective gas atmosphere to prevent the Fe e+ in slag from oxidizing in air to Fe3+. Once the experiment was completed, the crucible was allowed to cool. The crucible was then sectioned with a wet diamond saw and the corroded and 29 penetrated areas were measured and the slag-refractory interfaces were examined through SEM and EDS. The main drawback of the static corrosion method is that the slag may be rapidly saturated with the reaction products . Saturation at the slag-refractory interface and the absence of dynamic factors such as the movement of the slag will impede the chemical interaction. Since there are no contributions from dynamic factors, wear of the refractory by erosion cannot be considered in this testing method. ~- 76 mm Figure 7. Illustration of the core-drilled refractory crucible for the Static Corrosion Method 30 4.1 .2 Dynamic Corrosion Method The Rotary Slag Test or the Dynamic Corrosion Method is used to characterize the relative resistance of a refractory against slag erosion . This test method primarily focuses on the wear performance rather than chemical dissolution . The furnace movement prohibits the saturation of the slag at the interface, which allows the slag to continuously interact with the refractory. The rotary slag test is a certified ASTM standard (C874-85) [281 , which specifies that the wedge shaped specimens are to be 228 mm long with the cross-section dimensions of : 44 mm in width at the hot face, 114 mm in width at the cold face and 76 mm in height . The shape of the brick will allow 6 specimens to form a hexagonal test lining . Due to the size restriction of our rotary furnace, the dimensions of the wedge were modified in order to accommodate the limitations . In the modified procedure (*ASTM C874-85 Modified Rotary Slag Test), the cross-section was changed to : 28 mm in width at the hot face, 66 mm in width at the cold face and 49 mm in height as shown in Figure 8A. In order to obtain samples with the modified dimensions, specimens were cut from a larger brick with a wet diamond saw . The modified shape allowed eight samples to be mortared together with a thermal set mortar forming an octagonal test lining as shown in Figure 8B. The test lining was positioned midway into the shell of the rotary furnace, which was protected by a bubble alumina insulating layer as shown in Figure 8C . The assembly was first preheated with an open flame for an hour and then heated to an elevated temperature of approximately 1300\u00b0C with an oxygen-acetylene torch . With an optical pyrometer 31 Slag Feeder Insulating Layer Rotary Furnace (Alumina) Steel Shell Optical Pyrometer -------------- u Torch monitoring the temperature, the assembly required 2 .5 to 3 hours to reach the appropriate experimental temperature at a rotational speed of 3 rpm. 28 mm A) B) Refractory Test Slag Lining C) Figure 8 . Illustration of the Dynamic Corrosion Method : A) specimen dimensions, B) specimen formation in rotary furnace and, C) rotary furnace set up 32 During the experiment, slag pellets of 1000g were fed into the centre of the furnace lining every hour for 5 hours . Between each charge, the slag was tapped out by tilting the rotary furnace vertically allowing the slag to drain out . At the end of the experiment, the final slag charge was emptied and the furnace was allowed to cool . Once cooled, the refractory lining was taken out of the furnace and disassembled into its individual wedges . The wedges were sectioned in the longitudinal direction and were measured for corrosion effects (dissolution and penetration). 4.1 .3 Solubility Test in KIVCET Slag In the solubility test, pure chromium oxide (Cr 2 O3) powders were mixed in several KIVCET smelting slags in order to test chromium oxide's solubility in various slag ratios (Cr 2O3 slag) and to determine the role of lead in dissolving Cr 2O3 . Three different slags containing lead in various concentrations (0-11% Pb) were used . These slags were : 1) barren slag (N) with no Pb or Zn, 2) a low-lead slag (L) at 6% Pb (-5 .77%) and, 3) a high-lead slag (H) at 11% Pb (-10.24%). The experimental mixes consisted of slag and 0 .3\u00b5m Cr2O3 powder at various amounts : 0%, 5%, 10% and 20%. The Cr2 O3 powder and slag were uniformly mixed in a ball milled for two hours . The tests were conducted in 10 cm3 alumina crucibles and contained only 15g of slag mix to avoid overflow during firing as shown in Figure 9 . A full set of alumina crucibles, with different mixes, were fired at temperatures of 1250\u00b0C, 1300\u00b0C and 1400\u00b0C for 1 hour in pure argon . The testing program was decided after preliminary tests . Since no Cr2O3 was present after 1 hr of firing at 1400\u00b0C, we decided not 33 to perform experiments at 1350\u00b0C . In addition, 1-hour firings demonstrated that no Cr 2O3 was present; therefore longer soaking times were not required . The firing was performed in a small electric furnace with a SiC muffle, which would allow for the circulation of a protective gas (argon atmosphere at 99 .99% purity) to prevent the slag from oxidizing . After the firing was complete, the crucibles were quenched in air . By breaking the crucibles, the solidified slag was collected . The slag was then ground to a powder by hand with a pestle and mortar suitable for qualitative XRD analyses with independent measurements of the peak intensities . In order to measure the free Cr 2 O3 in all the experimental slag mixes, calibration curves for each N, L and H slags had to be prepared at 0, 5, 10 and 20% Cr 2O3 . The relative intensities were calculated as ratios between the measured Cr 2O3 peak in the experimental mix minus background and the intensity minus background for pure Cr2 O3 . As no Cr2O3 could be found in the slag, we considered that the small amount of Cr 2O3 (up to 20%) was dissolved into the slag . 10 cm3 Alumina Crucible 15 g Experimental Slag Mix Figure 9 . Illustration of the test sample for the Solubility Method 34 4.1 .4 Leaching Method for Hexavalent Chromium The Toxicity Characteristic Leaching Procedure 1311 is a method certified by the U .S. Environmental Protection Agency [33] and is used to determine the mobility of organic and inorganic compounds present in solids (hexavalent chromium) . The samples tested were taken from the penetration region of the dynamic samples where the highest interaction between the slag and the refractory occurred as shown in Figure 10 . Samples of 100g were ground to less than 1 cm in its narrowest dimension and placed in a bottle with 2000g of leaching agent . The leaching agent was composed of distilled water brought to a pH of 4 .7 with acetic acid in order to simulate acid rain . The bottle was then secured to an apparatus which was capable of agitating end over end at 30 \u00b1 2 rpm for 18 \u00b1 2 hours at 23 + 2\u00b0C. After the experiment, the leaching fluid was filtered from the precipitate and the fluid was refrigerated at 4\u00b0C until analyzed . Leach Test Sample Penetration Region Figure 10 . Illustration of the test sample for the Leaching Method 35 4.2 Industrial Experiments 4.2.1 Post-Mortem Failure Analysis The Post-Mortem Failure Analysis is a technique used to evaluate plant trial bricks, in order to better understand the corrosion behaviour and to establish a corrosion mechanism. Sampling . Sampling from the brick included all regions such as the accretion layer, and the penetrated and un-penetrated areas by molten liquids . The brick was divided into equal parts with a wet diamond saw along the longitudinal length of the brick . This sectioning method provided consecutive samples, which allowed us to generate property profiles from the hot face to the cold face . Figure 11 is an illustration of the sections, which represent the zones that were studied by post-mortem failure analysis. Accretion\t Penetration Layer Region Original Brick Region Hot Face I--- Z1 ~ Z2 I Z3\t ~ - Z4 Cold Face \t \/ Z5 -4--- Z6 --I Figure 11 . Schematic of the different zones in a used brick sample 36 Microstructure and Elemental Composition . In order to study the microstructure, Scanning Electron Microscopy (SEM) and Energy Dispersive Spectroscopy (EDS) were used to analyze the changes in composition and microstructure of the brick . The specimens were prepared by cutting a 6 .35 mm thick slice in the longitudinal direction from the hot face. The slice was then sectioned by cutting 19 .05 mm parallel to the hot face and then further segmented into 25 .4 mm sections perpendicular to the hot face, producing consecutive samples . The samples were set in epoxy resin for 24 hours and were polished down to a 6-micron paste. After polishing, the samples were carbon coated for SEM using the Hitachi S-2300 SEM with EDS detector . At one fixed magnification (200x), an EDS area line analysis was conducted from the hot face to the cold face . With this method, the depth of penetration and the elemental composition (by weight percent) could be approximated. Mineralogical Distribution . A set of consecutive samples similar to the set used for the SEM and EDS analysis was used for a phase distribution analysis . Using consecutive samples, a profile of the phase distribution from the hot face to the cold face could be determined through X-Ray Diffractometry (XRD) . The samples used for XRD were ground using a ring mill and further ground by hand with a pestle and mortar . The mineralogical composition was determined using a Rigaku Multiflex XRD system with Jade 7 .0 software for peak identification . 37 Strength and Physical Properties . Samples for the Modulus of Rupture (MOR), the Cold Crushing Strength (CCS) (Figure 12) and the Absorption properties were prepared by cutting three 25.4 mm thick sections along the longitudinal direction from the hot face with a wet diamond saw. These three slices were further cut by 25 .4 mm parallel to the hot face producing 25.4 x 25 .4 x 152 .4 mm specimens specified by ASTM standard C133-97 [30] for MOR. Due to the restriction of the brick dimensions, the specimens were modified to 25 .4 x 25 .4 x <127.0 mm rather than 152.4 mm in length (*ASTM C133-97 Modified MOR) . After the MOR test, the first half of the broken bar was further cut perpendicular to the hot face by 25.4 mm producing 25 .4 x 25.4 x 25 .4 mm cubes for CCS . Again, due to the restriction of the brick dimensions, the size of the cubes were modified to 25 .4 x 25 .4 x 25 .4 mm rather than 50.8 x 50.8 x 50.8 mm (*ASTM C133-97 Modified CCS) . The second half of the broken bar was tested for apparent porosity, water absorption and bulk density specified by ASTM standard C20-97 [311 . These properties were determined from the dry, saturated and suspended weights . From all of these properties tests, the average of the three specimens per inch distance could therefore produce a profile from the hot face to the cold face . 38 Penetration \t \t Original Brick Region Region CCS \t\/ \/\/ Hot Face MOR \/ Cold Face \/ 1-- Z1 --I Z2 --~- Z3 ---I Z4 z6 -4-- Z6 \u2014I Figure 12 . Specimen preparation for the Cold Crushing Strength and the Modulus of Rupture tests 4.3 Samples for Laboratory Corrosion Tests and Post-Mortem Analysis Bricks used for the laboratory corrosion test were selected based on the advice from industry experts as well as with the support of technical datasheets such that only the bricks with exceptional performance would be tested . The chemical, physical and mechanical properties are presented in Table 3. Shown in Table 4 are the samples to be characterized by the Static and Dynamic Corrosion Methods, the Leaching Method and the Post-Mortem Failure Analysis . These bricks are commercially available and will be referred to by the given codes as shown in the Table . 39 o o v .) cNI \u00b0 0000 If t-t a ; 0 0 o O M -+ H 4-1 1 \u20224 -q - -1 - 1 - 4 . V N -+ N n to N N M M M N M N i N 1 i Cw M M M M M M M e -4 0 O O O O O 0 0 0 0 0 0 c o O co O O c o 0 0 0 0 0 0 00 V') ~O O O ~O . - -n 0 0 O O 4 - 4 V O , -+ U0 tr .) tr) . (\"! M - , 01 \u2014i 01 c o O c o \u2022 -4 4) 4--1 N V . -+ N O COw 0 O O O O O O M a .) M N M col 1-1 - 01 D D V EOU M0 ~O O ~n ~n O O ~O ~ n M 1 4 -4 . . .Ni 0i. 01 )--; 4--1 \tN N N M M l~ U0 N N N N N N A --+ N 00 O O O O N to in O 01 ~ ~I A r--: tri I-~ V7 ~n V) n \u2022 P V Y C p i y C o 0 In 4 .1 G .14 'cr z \u2022 ~ a ,A x ~ a ~ ~ x A a a O fir . .) O IC 11 cc -4 -tt 40 NN O O O O O co O N M h 1 O O \u2014 O O co co O Ln 00 00 N N O N M 4 - 4 1 \u2022-1 ~ - -i . O O O O c o c o c o O 00 Ln 0 0 0 0 0 0 N 00 ac.14 I O IC IC a c cIrL7 y y or) co Table 4. Samples to be tested by the Static, Dynamic, Leaching and Post Mortem Failure Analysis Test Methods Test Method Type (Code) Sample Static Dynamic Leach PostMortem Magnesia- (MC-1) Narmag FG x Chrome (MC-2) BCF F15 x x x x (MC) (MC-3) BCF F17 x (MC-4) (MC-5) Durachrome Radex Fox Compact x x x x (MC-6) Radex Fox x (MC-7) (MC-8) MX18A MX55 x x x x x x x (MC-9) Rexal DB60 x x x (MC-10) Dura Mach (MC-11) B65 Magnesia- (MA-1) Magnel RS x x Alumina (MA-2) NK92 x Spinel (MA) (MA-3)(MA-4) Radex AS905 Radex A90 x x 4.4 Slag Composition In the evaluations using the Static and Dynamic Corrosion Methods, the slag used was from the KIVCET Furnace. The composition is shown in Table 5 . The KIVCET slag contains mainly iron in the forms of FeO\/Fe2O3 , which is accompanied by SiO2 and CaO and is typical in non-ferrous converting furnaces . In addition, the slag contains moderate levels of ZnO (10.1%) and PbO (4 .4%) . 41 Table 5 . Chemical composition of the KIVCET slag Slag\t Composition(%) SiO2 CaO Fe Pb Zn As Sb Bi Cu Sn S KIVCET 23.2 13 .2 26 .0 4.4 10.1 0.08 0.17 0.01 0.24 0.06 0.60 In the solubility test, the composition of the slag mixes is shown in Table 6 . The N, L and H correspond to the KIVCET slags containing no-lead, low-lead and high-lead, respectively. The 0 (0%), 1 (5%), 2 (10%), and 3 (20%) corresponds to the amount of Cr 2O3 powder in the slag mix. Table 6 . Chemical compositions of the experimental mixes for the solubility test Composition Experimental Slag Mix FeO Fe 2 O3 CaO Cr2O3 PbO SiO2 ZnO % HO 18.9 2.5 11 .9 0.05 8 .6 19.5 15 .0 Hl 18.5 1 .6 11 .3 3 .1 8 .3 17.9 14.1 112 17.6 1 .5 10.7 6.3 8 .0 16.8 13 .5 H3 16.1 1 .2 9.6 12.4 7 .2 14.5 12.2 LO 18.9 1 .2 13 .3 0.05 4.9 21 .1 16.4 Ll 18.3 1 .1 12.7 3 .2 4.8 20.0 15 .8 L2 17.2 1 .0 12.0 6.1 4.5 18 .5 14.8 L3 15 .6 0.8 10.8 12.3 4.1 15 .6 13 .2 NO 28.5 0.3 18 .1 0.06 0.05 28 .3 1 .4 Ni 27.0 0.2 17.1 3 .2 0.05 26 .4 1 .4 N2 25 .5 0.3 16 .1 6.5 0.05 24 .8 1 .3 N3 23 .3 0.1 14.7 12.7 0.07 21 .8 1 .2 42 \t5.0 Results and Discussion 5.1 Laboratory Experiments In order to investigate the corrosion behaviour of several refractory bricks (magnesia-chrome (MC) and magnesia-alumina spinel (MA)), SEM and EDS examinations of the interface between refractory and KIVCET slag and the change in microstructural features were conducted . The corrosion experiments were carried out using a dynamic corrosion method by a rotary slag furnace and a static corrosion method by a slag crucible test to simulate industrial conditions found in non-ferrous furnaces . In addition, a solubility test of Cr2O3 powder in KIVCET slag was performed in alumina crucibles and a leaching test on several used MC bricks for hexavalent chromium was conducted. 5.1.1 Characterization of the KIVCET Slag The KIVCET slag was characterized as a ceramic-glass composite containing two phases : a glass phase containing Ca and Si and a crystalline phase containing Zn and Fe as shown in Figure 13 and corresponding EDS analysis in Table 7 . From the XRD analysis (Figure 14), a crystalline phase of zinc ferrite (ZnFe 2O4) was identified. Table 7. Composition of the KIVCET slag by EDS point analysis Element, wt .% 0 Mg Al Cr Si Ca Fe Zn 1 . Glass Phase 30 .09 2 .25 0.46 0.00 19 .63 29.15 3.26 15 .18 2 . Crystalline Phase 18.36 0.28 2 .03 0.00 0.12 0.28 60.47 18.46 43 x500 0000 20kV 100aum Figure 13 . SEM image of KIVCET slag displaying : 1) glass phase and, 2) crystalline phase, (500x) \u2666 \u2666 Zinc Ferrite, ZnFe2 O4 y m C \u2666 ~~ 0 10 20 30 40 50 60 70 80 90 2-theta Figure 14. XRD peaks for the KIVCET slag 44 5.1 .2 Corrosion Performance by the Static Corrosion Method The results from the static corrosion test are presented in Table 8 and the corresponding images are shown in Figures 15A-F . As expected from the static corrosion method, no measurable dissolution was observed at the interface between the slag and the refractory due to the absence of any turbulent flow from the slag . The slag at the interface may have been rapidly saturated with the reaction products preventing further interaction . In addition, there was no measurable difference in the dissolution behaviour between the MC and the MA crucibles . With a 5-hour soak period at a temperature of 1300\u00b0C, the test did not provide enough information in order to have a measurable corrosion depth . Although dissolution was not observed, the penetration depth was observed to be minimal. Table 8 . Corrosion and penetration depth of the samples from the static corrosion method Sample Penetrated, mm MC-2 3 .90 MC-4 2 .34 MC-7 3 .82 MC-8 3 .72 MA-1 2 .04 MA-4 1 .50 45 A) B) C) Figure 15 . Cross-section of the crucibles from static corrosion method: A) MC-2, B) MC-4, C) MC-7 46 D) E) F) Figure 15 cont. D) MC-8, E) MA-1 and, F) MA-4 47 5.1 .3 Corrosion Performance by the Dynamic Corrosion Method From the dynamic corrosion method, the dissolution and penetration depth was more apparent compared to the static corrosion method . The results are presented in Table 9 and the corresponding cross-section images are shown in Figures 16A-K . Comparing the degree of the corrosion by measurement between the MC and the MA bricks, it was shown that the MC bricks were more resistant to corrosion than the MA bricks after a 5-hour soak period at 1400\u00b0C*. Comparing the MC bricks to each other, very minimal to no dissolution at the hot face was observed and therefore no comparable differences could be concluded in terms of dissolution. *Note : The testing temperature was raised from our originally intended temperature of 1300\u00b0C to 1400\u00b0C due to the slag solidifying to the brick lining at 1300\u00b0C . This was potentially due to the oxidation of Fee+ to Fe3+, resulting in the precipitation of solid spmel phases and increased viscosity. Table 9 . Corrosion and penetration depth of the samples from the dynamic corrosion method Sample Corroded, mm(Rating) Penetrated, mm (Rating) MC-1 0 .04 (1) 6.34 (10) MC-2 0.24 (8) 6.16 (9) MC-3 0.05 (3) 5 .32 (7) MC-4 0 .06 (4) 5 .29 (6) MC-5 0.08 (6) 4.43 (2) MC-6 0 .07 (5) 5.60 (8) MC-7 0 .16 (7) 5 .00 (5) MC-8 0 .04 (1) 4.14 (1) MA-1 0.68 (10) 4.60 (3) MA-2 0.54 (9) 6.16 (11) MA-3 0.70 (11) 4.62 (4) 48 Figure 16 . Cross-section of the samples from dynamic corrosion method: A) MC-1, B) MC-2, C) MC-3, D) MC-4 49 Figure 16 cont. E) MC-5, F) MC-6, G) MC-7, H) MC-8 50 Figure 16 cont . I) MA-1, J) MA-2 and, K) MA-3 Although there were no significant differences in the dissolution behaviour between the MC bricks, there was a measurable difference in the penetration depth . The resistance to impregnation at the hot face is also important to the corrosion behaviour as it changes the physical properties of the refractory, in particular, at the penetrated front . In batch processes 51 that may require intermediate delay or frequent down time, the temperature of the refractory could rapidly fluctuate causing the refractory to have large temperature gradients, which would result in thermal-mechanical stresses . These stresses could cause the refractory to spall at the penetrated front between the penetrated and the un-penetrated regions . The corrosion behaviour for the MC bricks, based on the penetration depth, is presented in the following order: MC-8 > MC-5 > MC-7 > MC-4 > MC-3 > MC-6 > MC-2 > MC-1 In order to explain the performance in terms of brick properties (MgO, Cr 203, Fe203, SiO2, and CaO content, apparent porosity, CCS and MOR), no apparent trend could be hypothesized. A further examination of the morphology and microstructure by SEM was conducted. To further explain the penetration behaviour, a microstructural analysis by SEM of the best two (MC-8 and MC-5) and the worst two (MC-2 and MC-1) performing specimens showed that the microstructure of the best two were significantly different compared to the microstructure of the worst two as shown in Figures 17A-D . Observing the grain size of specimens MC-8 and MC-5, they were much smaller compared to MC-2 and MC-1 . The small grain size of MC-8 and MC-5 promoted smaller pore dimensions, which was encouraging in terms of the resistance to slag penetration . The large pore dimensions of MC-2 and MC-1 would allow molten slags to penetrate much more easily as observed in the specimens from the dynamic test . In addition, the SEM images of the best two specimens 52 were observed to have greater contact at the grain boundaries between adjacent grains compared to the worst two . Due to the lower surface energy and greater contact between smaller grains, the distribution of the periclase and the chrome grains promoted a greater amount of direct bonds . As seen in the microstructure of MC-2 and MC-1, the larger grains seem to have minimal contact between adjacent grains at the grain boundaries. A) B) C) D) Figure 17 . Microstructure of the unused bricks : A) MC-8 (80x), B) MC-5 (80x), C) MC-2 (80x) and, D) MC-1 (80x) 53 Based on the corrosion performance of several magnesia-chromia (MC) and magnesia- alumina spinel (MA) bricks, the MC bricks were more corrosion resistant to KIVCET slag than the MA bricks in terms of erosion and dissolution . The microstructural features determined that the pore dimensions and grain size are a contributing factor to the penetration resistance against molten slags . The absence of a good network of direct bonds throughout the brick would allow the grains to be easily eroded or \"washed\" away at the hot face by slags . In addition, a good network of direct bonds and small pore dimensions would also improve and further resist molten liquids from penetrating deep into the brick. 5.1.4 Magnesia-Chrome Bricks with KIVCET Slag From the interaction of KIVCET slag and the MC bricks, an interface between the slag and the refractory was observed as shown in Figures 18 and 19 for MC-2 (static method) and MC-8 (dynamic method) respectively. An EDS analysis at the refractory-slag interface of MC-2 (point 2) identified a layer containing both compositions of the refractory and the slag, which was likely the formation of a (Mg, Fe, Zn)(Fe, Al, Cr)2O4 complex spinel . A full EDS analysis is shown in Table 10 . The formation mechanism was possibly similar case described by Wiederhorn et al [15] and Evrard et al [191 . The cations of Fe e+ and Zn2+ from the slag diffused into the periclase and the spinel structure at the hot face, while Mg t+ was partially replaced. The free Mgt+ would then enter into the slag forming a silicate phase with Fe, Zn and Si such as a modified forsterite phase (Mg, Fe, Zn) 2SiO4. This type of behaviour was typically found in converter slag attack . The formation of the spinel at the interface would act as a barrier from further refractory-slag interaction. Although this may 54 be a likely case, the spinel and the sequence of reactions are uncertain and the penetration region is too small to be examined by XRD for confirmation. Figure 18 . SEM image of MC-2 from the static corrosion method displaying : 1) slag, 2) interface of (Mg, Fe, Zn)(Fe, Al, Cr) 2 O4 complex spinel and, 3) refractory, (50x). The corresponding EDS analysis is shown in Table 10 Table 10. EDS analysis of MC-2 from the static corrosion method corresponding to Figure 18 Element, wt.% 0 Mg Al Si Ca Cr Fe Zn Pb 1 . Slag 30.13 1 .73 0.00 13 .96 11 .88 0.00 23 .48 18.82 0.00 2. Interface 24.84 3 .97 1 .92 2.71 2.48 10.54 38 .76 14.79 0.00 3. Refractory 32 .32 42.72 2.63 0.76 0.55 15.84 5 .19 0.00 0.00 In sample MC-8, it was observed that the corrosion of the brick was due to the dissolution of the MgO grains . As shown in Figure 19 points 1-4 (Table 11), the grains within the penetrated layer contained both compositions of the refractory and the slag . The MgO grain was chemically altered by the slag forming a modified forsterite phase (Mg, Fe, Zn) 2SiO 4 , which was a typically occurrence in converter slag attack . The change from an MgO phase to a modified forsterite phase would definitely destroy the grain boundaries between adjacent grains weakening the structure . With a weakened structure, the grains would be easily 55 \"washed\" or eroded away at the hot face . As seen at point 5 of Figure 19, the chrome grain remained chemically unaltered but the grain was observed to be \"pulled out\" from the refractory interface . In addition, the direct bonds between the coarse grains (close to the interface of the refractory) demonstrated the resistance to slag penetration and there was no evidence of the grain boundaries being dissolved into slag. Figure 19 . SEM image of MC-8 from the dynamic corrosion method displaying : 1- 4) (Mg, Fe, Zn, Ca)2SiO4 spinel, 5) Cr203 grain, 6) MgO grain and, 7) (Mg, Fe)(Cr, Al)204 spinel (150x) . The corresponding EDS analysis is shown in Table 11 Table 11 . EDS analysis of MC-2 from the dynamic corrosion method corresponding to Figure 19 Element, wt .% O Mg Al Si Ca Cr Fe Zn Pb 1 . (Mg, Fe, Zn, Ca) 2SiO4 spinel 2. (Mg, Fe, Zn, Ca) 2SiO 4 spinel 3 . (Mg, Fe, Zn, Ca)2SiO4 spinel 4. (Mg, Fe, Zn, Ca)2SiO4 spinel 5. Cr203 grain 38.96 39.34 37.80 42 .14 29 .50 31 .51 30 .72 25.58 30.43 12.41 0.00 0.00 0.00 0.00 6 .58 22.38 21 .73 20.50 20.94 0.10 1 .94 1 .07 3 .77 1 .91 0.00 0.00 0.00 0.28 0.49 47 .74 2.68 3 .32 4.23 2 .24 3 .68 2.54 3 .81 7.84 1 .85 0.00 0.00 0.00 0.00 0 .00 0 .00 6. MgO grain 29.52 43 .82 0 .86 0.33 0.00 7 .25 13 .23 4.99 0 .00 7. (Mg, Fe)(Cr, Al) 2 04 spinel 27.53 10.30 3.54 0.00 0 .00 26 .92 26.04 5 .66 0.00 56 Based on the results, the slag penetrated into the refractory mainly through open pores, while the direct bonds resisted the penetration of the slag . The corrosion of the magnesia-chrome brick was from the chemical dissolution of the MgO grains by slag, which formed a modified forsterite phase . The dissolution of the MgO grains destroyed the grain boundaries between adjacent grains causing the coarse grains to be washed away at the hot face through erosion. Therefore, the corrosion of the magnesia-chrome bricks was due to the dissolution of MgO. 5.1 .5 Magnesia-Alumina Spinel Bricks with KIVCET Slag Shown in Figures 20A-B of the MA-1 sample (static method), an interface containing both components of the KIVCET slag and the refractory was observed . The elemental composition detected by EDS at point 5 (Table 12) was likely the formation of a (Mg, Zn, Fe)(Fe, Al) 2O4 complex spinel . The hypothesized mechanism (similar to Petkov et al [181 ) was the cations of Fee+ and Zn2+ from the slag, which diffused into the spinel grain, replaced Mgt+ while the cation of Fe3+ from the slag replaced A13+. The formation of the spinel provided a barrier for the refractory against further interaction with the slag. From the interaction of the KIVCET slag and the MA-2 brick (dynamic method) shown in Figures 21A-B, the magnesia-alumina spinel grain at point 1 (Table 13) was observed to be surrounded by a composite slag composed of a glassy phase and a crystalline phase. Within the spinel grain (point 3), contents of the slag were observed (13 .70 wt.% Fe and 13.97 wt .% Zn), which was likely the formation of a (Mg, Fe, Zn)(Fe, Al)2O4 complex spinel . It was 57 Figure 20 . A) SEM image of MA-i from the static corrosion method displaying : 1) MgAl2 O4 grain, 2) MgO grain, 3) slag, 4) MgO-slag interface, 5) interface of (Mg, Fe, Zn)(Al, Fe)204 complex spinel and, 6) refractory-slag interface, (50x) . The corresponding EDS analysis is shown in Table 12 Figure 20 . B) Magnified view of the MgAl2O3 grain and slag interface, (400x) Table 12. EDS analysis of MA-1 from the static corrosion method corresponding to Figure 20 Element, wt .% O Mg Al Si Ca Fe Zn 1 . MgAl 2O4 grain 41 .31 17 .51 39 .93 0.25 0.12 0.39 0 .49 2. MgO grain 38 .26 60.79 0.11 0.19 0.00 0.26 0.38 3. Slag 30.61 4.33 0.00 17.76 15 .8 12.56 18 .94 4. MgO-slag inteface 23 .87 32 .00 0.00 0.00 0.00 19.08 25 .05 5. (Mg, Fe, Zn)(Al, Fe)204 spinel 28 .28 8 .99 17 .50 0.00 0.00 32 .51 12 .72 6. Refractory-slag interface 19.01 11 .18 0.00 0.00 0.00 44.29 25.52 58 Figure 21 . A) SEM image of MA-2 from the dynamic corrosion method displaying: 1) MgAl2O4 grain, 2) MgO grain, 3) (Mg, Fe, Zn)(Fe, Al)204 spinel, 4) (Mg, Fe, Zn, Ca)2SiO4 spinel, 5) (Mg, Fe, Zn, Ca) 2SiO4 spinel, 6) slag (glass phase) and, 7) slag (crystalline phase), (200x) . The corresponding EDS analysis is shown in Table 13 x 200 0000 20 kV 200 .um Figure 21 . B) Magnified view of (Mg, Fe, Zn)(Fe, Al)204 spinel and (Mg, Fe, Zn, Ca)2SiO4 spinel, (600x) Table 13. EDS analysis of MA-2 from the dynamic corrosion method corresponding to Figure 21 Element, wt.% O Mg Al Si Ca Fe Zn Mn 1 . MgAl 2O4 spinel 40 .09 17.71 40.41 0.00 0 .08 0.83 0 .77 0.10 2. MgO grain 28.67 46.13 1 .02 0.10 0.00 12 .93 10 .74 0.42 3. (Mg, Fe, Zn)(Fe, Al) 204 spinel 31 .68 11 .34 28.88 0.12 0.00 13 .70 13 .97 0.30 4. (Mg, Fe, Zn, Ca)2SiO4 spinel 34.08 26.19 0.59 21 .67 3 .62 4.27 8 .90 0.68 5. (Mg, Fe, Zn, Ca)2SiO4 spinel 34.72 32.71 0.14 21 .83 3 .08 3.41 3.79 0.30 6. Slag (glass phase) 34.37 6 .23 3 .38 22.28 20.70 10.43 2.31 0.29 7. Slag (crystalline phase) 20.87 5 .98 4.66 0.06 0 .19 55.74 11 .66 0.84 59 also observed that the MgO grains (point 4) were highly attacked by slag, which resulted in the formation of a modified forsterite phase of (Mg, Fe, Zn, Ca) 2SiO 4 . Based on the results, the magnesia and magnesia-alumina spinel grains were both highly susceptible to dissolution by slag . In addition, as demonstrated by the dynamic test, the coarse grains were highly vulnerable from being \"pulled out\" from the refractory interface by erosion. Therefore, the corrosion of the magnesia-alumina spinel brick was due to both dissolution and erosion of both magnesia and magnesia-alumina spinel grains. 5.1.6 Solubility of Cr203 in KIVCET Slag The scope of this experiment was to determine how much Cr2 O3 would remain unreacted after firing at 1250\u00b0C, 1300\u00b0C and 1400\u00b0C at 1 hour soak periods . The crucibles after firing at their respective temperatures are shown in Figure 22A-C . In order to measure the free Cr2 O3 in all the experimental slag mixes, calibration curves for each N, L and H slags were prepared at 0, 5, 10 and 20% Cr2 O3. The relative intensities were calculated using directly the measured peak intensities . The XRD analysis on all experimental mixes fired at 1250\u00b0C, 1300\u00b0C and 1400\u00b0C in argon identified two silicatic phases of willemite (Zn 2SiO4) and hardystonite (Ca2ZnSi 2 O7) and a solid solution phase of complex spinels containing Fe, Zn and Cr (appendix 1) . A minor phase was present as metallic lead (Table 14) at 2 .2 % in mix LO and 3 .3 % in mix H2, which indicated that the 0 2 partial pressure was much lower than 10-5 atm (argon of 99.999% purity). The peak intensities versus % Cr2O3 and % Pb after firing at 1250\u00b0C are presented in Table 15 . The theoretical positions for all the expected spinel phases and the real reading for their solid solution in our experimental mixes are 60 shown in Table 16. It is to be mentioned that according to the XRD charts, the main peaks of these spinels (FeCr2O4 or FeZnCrO4) are overlapping with magnetite (Fe 304) and hematite (Fe203) peaks, which are also presumed to form due to the oxidation of Fe e+ to Fe3+ in air. Correlations between the amount of the major phases willemite, hardystonite and spinel (expressed using directly the measured peak intensities) and the % Pb (0, 6 and 11%) and % Cr 203 (0, 5, 10 and 20%) are presented in Tables 17-19 and Figures 23-25, respectively. A) B) Figure 22. Crucibles after the solubility test for 1 hour in argon at A) 1250\u00b0C and B) 1300\u00b0C 61 C) Figure 22 cont. C) 1400\u00b0C Table 14 . Preliminary quantitative XRD analysis on slag mixes LO (low-Pb) and H2 (high-Pb) fired at 1250\u00b0 for 1 hour in argon cYoSample Hardystonite, % Magnetite, % Willemite, % Lead, LO 44.6 34 .2 19 2.2 H2 36.2 59 .2 1 .3 3 .3 62 Table 15 . Measured peak intensities on samples fired at 1250\u00b0C for one hour in argon Mineral Peak (from XRD chart) Exp. Mix % Pb Measured peak intensities, cps (including background) Before After testing at 1250\u00b0C for 1 hr in Ar I \/ Io, 2 0, Slag Slag Slag Slag Slag % Deg . +5% +10% +20% Cr20 3 Cr203 Cr20 3 N 0 67.5 188 .8 426.4 721 .6 796 100 35 .41 L 6 220 315 .2 884 .8 998.4 1012 Magnetite H 11 176 .5 468 .8 863 .2 968 .8 1300 (Fe304) N 0 93 132 .8 236 359.2 346 .4 31-33 30.06 L 6 135 .5 163 .2 395 .2 413.6 420 .8 H 11 101 248 .2 350.4 428.8 596 .8 N 0 87.5 74.4 60 .8 82 .4 101 .6 100 33 .41 L 6 97 .5 226.4 46 .4 44 36.8 (Fet_XCrx)203 H 11 58 69.6 44.8 43 .2 34 .4 N 0 67.5 188 .8 442.2 359 .2 346 .4 65 35.803 L 6 220 332 884.8 721 .6 796 H 11 176 .5 468.8 863 .2 968 .8 1300 N 0 66 188 .8 442.4 721 .6 796 99.9 35.974 L 6 242 332 884.8 998 .4 1012 ZnCr2 O 4 H 11 176 .5 354.4 863 .2 911 .2 1300 N 0 93 132 .8 236 359 .2 346.4 36 .3 30 .537 L 6 135 .5 332 1048 .8 398 .5 420.8 H 11 101 614.4 350.4 513 .6 596.8 N 0 67 .5 188 .8 442.4 721 .6 796 100 35.54 L 6 220 332 884.8 998 .4 1012 H 11 176 .5 468.8 863 .2 968 .8 1300 FeCr2 O 4 N 0 93 144 .8 236 359 .2 346.4 33 30.172 L 6 135 .5 163 .2 395.2 413 .6 420.8 H 11 101 248.8 350.4 428 .8 596.8 N 0 67.5 188 .8 442.4 721 .6 796 99.9 34 .179 L 6 220 315 .2 884.8 998 .4 1012 H 11 176 .5 468.8 863 .2 968.8 1300 ZnO N 0 93 91 .2 113 .6 104 .8 95 .2 56 .7 30 .811 L 6 345 1049 539.2 477.6 340.8 H 11 197 .5 614.4 409.6 513 .6 337.6 N 0 99 .5 85 .6 55 .2 39 .2 43 .2 70 31 .731 L 6 345 488 190 .4 172 .8 189 .6 H 11 166 267.2 310.4 285 .6 183 .2 FeO\u2022Cr2O3 N 0 78 .5 77 .6 56 .8 102 .4 61 .6 40 28 .799 L 6 109 625.6 333.6 268.8 206.4 H 11 53 383 .2 336.8 214.4 241 .6 N 0 67 .5 188 .8 442.4 721 .6 796 100 35 .395 L 6 220 315 884 .8 998.4 1012 H 11 176 .5 468.8 863 .2 968 .8 1300 FeZnCrO4 N 0 93 148 236 359.2 346.4 31 30 .11 L 6 135 .5 163 .2 395.2 413.6 420.8 H 11 101 248.8 350.4 428.8 596.8 63 Table 15. (Continued) Mineral Peak (from XRD chart) Exp. Mix % Pb Measured peak intensities, cps (including background) Before After testing at 1250\u00b0C for 1 hr in Ar I \/Io, 2 0, Slag Slag Slag Slag Slag % Deg . +5% +10% +20% Cr20 3 Cr203 Cr203 N 0 67 .5 188 .8 442 .4 721 .6 796 100 35 .627 L 6 220 315 884 .8 998 .4 1012 H 11 176 .5 468 .8 863 .2 968 .8 1300 Cr0 N 0 93 148 236 359.2 346 .4 31 30.298 L 6 135 .5 163 .2 395 .2 413 .6 420 .8H 11 101 248 .8 350 .4 428 .8 596 .8 N 0 50 52 .8 43 .2 71 .2 71 .2 100 26 .03 L 6 67 .5 109 .6 68 .8 56 57 .6Keatite H 11 38 75 .2 61 .6 70 .4 109 .6 (SiO2) N 0 44 .5 44 .8 40 63 .2 73 .6 41 .3 23 .845 L 6 50 .5 401 .6 227 .2 195 .2 146 .4H 11 30 207 .2 228 .8 188 170 .4 N 0 87 .5 73 .6 48 .8 43 .2 52 100 34 .33 L 6 110 .5 234 .4 58 .4 54 .4 40Willemite H 11 58 68 .8 44 .8 48 39 .2 (Zn2 SiO4) N 0 96 .5 87 .2 67 .2 41 .6 43 .2 100 31 .82 L 6 345 1049 539 .2 477 .6 340 .8 H 11 197 .2 614 .4 409 .6 513 .6 337 .6 N 0 92 87 .2 77 .6 75 .2 59 .2 100 31 .16 L 6 345 1049 539 .2 477 .6 340 .8Hardystonite H 11 197 .5 614 .4 409 .6 513 .6 337 .6 Ca2ZnSi 2O7 N 0 78 .5 86 .4 79 .2 102 .4 90 .4 60 28 .92 L 6 109 625 .6 333 .6 268 .8 206 .4H 11 53 383 .2 336 .8 214 .4 241 .6 Table 16. Theoretical and true XRD peak position (20) of spinels Theoretical True XRD Chart 20, \u00b0 FeFe 2 O4, \u00b0 FeCr2O4 , \u00b0 ZnFe 2O4, \u00b0 ZnCr2O4 , \u00b0 FeZnCrO4, \u00b0 30.06 - 30.19 30.060 30.172 30.193 30.537 30.110 35.41 - 35.63 35.410 35 .540 35.627 35 .974 35 .395 64 Table 17. Peak intensities for Willemite (Zn 2SiO 4 ) Peak Intensity (including background), cps Peak (2 0) Mix Pb, % Slag before Sla Slag + %Cr203 at 1250\u00b0C for 1 hr in Artest g 5% 10% 20% 34 .33 N 0 87 .5 73 .6 48 .8 43 .2 52 .0 (I\/1o=100%) L 6 110 .5 234 .4 58 .4 54 .4 40 .0 H 11 58 .0 68 .8 44 .8 48 .0 39 .2 31 .82 N 0 96 .5 87 .2 67 .2 41 .6 43 .2 (I\/Io=100%) L 6 345 .0 1048 .8 539 .2 477 .6 340 .8 H 11 197 .2 614 .4 409 .6 513 .6 337 .6 Intensity of Willemite phase vs % Pb @ 34 .33deg-peak 250 --~- 0 f- 200 - a -Ir 10 20 \"- 150- N 100 - 50 1---n 0 0 2 4 6 8 10 12 Pb, % Intensity of Willemite phase vs %Cr203 @ 34 .33 deg-peak 250 200 --\u2666- N --E- L -A- H a \u00b0- 150 - 50 0 5 10 15 20 25 Cr203, % Intensity of Willemite phase vs % Cr203 @ 31 .82 deg-peak -\u2666-N -FL -A-H 800 - 600 1 400 - 200 - 0 ~~~\t 1\t \u2022 0 5 10 15 20 25 Cr203, % MU Vl Cm C 1200 1000 Figure 23. XRD intensities for Willemite versus %Pb (left) and Cr20 3 (right) 65 Table 18. Peak intensities for Hardystonite (Ca2ZnSi 2 O7) Peak (2 0) Mix Pb, % 31 .16 N 0 (I \/ Io = 100%) L 6 H 11 28 .92 N 0 (I \/ Io = 59 .9%) L 6 H 11 Peak Intensity (including background), cps Slag before \t \t Slag + %Cr 203 at 1250\u00b0C for 1 hr in Ar test Slag 5% 10% 20% 92 .0 87 .2 77 .6 75 .2 59 .2 345 .0 1048 .8 539 .2 477 .6 340 .8 197 .5 614 .4 409 .6 513 .6 337 .6 78 .5 86 .4 79 .2 102 .4 90 .4 109 .0 625 .6 333 .6 268 .8 206 .4 53 .0 383 .2 336 .8 214 .4 241 .6 Intensity of Hardystonite phase vs %Pb @ 31 .16 deg-peak 1200 - 1000 - co 800 - 600 -N c el) 400 - 200 - 0 0 2 4 6 8 10 12 Pb, \u2014~\u2014 0 \u2014E\u2014 5 --10 20 Intensity of Hardystonite phase vs %Cr203 @ 28 .92 deg-peak 700 600 N --II- L \u2014A\u2014 H a 500U F, 400 c 300 - 200 - 100 i 0 0 5 10 15 20 25 Cr203, % Figure 24. XRD intensities for Hardystonite versus %Pb (left) and Cr203 (right) 66 Table 19. Peak intensities for Spinel Phases Solid Solution (Fe304, FeCr2 O4, ZnFe2 O4, ZnCr2O4, ZnFeCrO4) Peak (2 0) Mix Pb, % 35 .41-35 .63 N 0 (I \/ Io = 100%) L 6 H 11 30 .06-30 .19 N 0 (I \/ Io = 30-33%) L 6 H 11 Peak Intensity (including background), cps Slag before \t \t Slag + %Cr203 at 1250\u00b0C for 1 hr in Ar test Slag 5% 10% 20% 67 .5 188 .8 426 .4 721 .6 796 .0 220 .0 315 .2 884 .8 998 .4 1012 .0 176 .5 468 .8 863 .2 968 .8 1300 .0 93 .0 132 .8 236 .0 359 .2 346 .4 135 .5 163 .2 395 .2 413 .6 420 .8 101 .0 248 .2 350 .4 428 .8 596 .8 Intensity of spine) phases versus lead for various % Cr203 @ 35.41-35.63 deg . peak 1400 0 f 5 1200 # 10 -4E\u2014 20 n 1000 -U TN Cc 800 600 - 400 200 ~ \"-'_~ 0- 0 2 4 6 8 10 12 Pb, % Intensity of Fe304 phase versus Pb for various % Cr203 @ 30 .06-30 .19 deg-peak 0 2 4 6 8 10 12 Pb, % 700 600 500 U 400 c 300 2 200 100 0 \u2014~-0 \u2014E\u20145 -10 -~-- 20 Intensity of spine) phases vesus Cr203 for various %Pb @ 35 .41-35 .63 deg . peak 10 15 20 25 Cr203, 0 5 Intensity of Fe304 phase versus Cr203 @ for various %Pb @ 30 .06-30 .19 deg-peak 10 15 20 25 Cr203, 0 5 Figure 25. XRD intensities for spine) phases versus %Pb (left) and Cr203 (right) 67 In regards to the willemite phase: \u2022 The two measured peaks, at 20=34 .33\u00b0 and 31 .82\u00b0 showed similar curve profiles of the intensities versus % Pb and % Cr2 O3, although the peaks at 34 .33\u00b0 were much weaker than at 31 .82\u00b0. \u2022 The maximum amount of willemite was found in the pure slag LO after firing at 1250\u00b0, followed by HO and the minimum was in the slag NO . These two slags contained willemite even before firing, but the amount increased 2-3 times after firing. \u2022 The amount of willemite decreased consistently with increasing the Cr2O3 content in the slag and remained at relatively constant and low values for all N mixes (containing barren slag). Based on these results, it could be stated that the willemite phase formation and its equilibrium with the molten material depends strongly on the lead concentration and has a maximum for 6 % Pb, according to our data. In regards to the hardystonite phase: \u2022 The two major peaks at 20=31 .16\u00b0 and 28.92\u00b0 showed similar curve profiles of the intensity versus % Pb and % Cr2O3 . \u2022 The maximum amount of hardystonite was also found, similar to willemite, in the pure slag LO after firing at 1250\u00b0, followed by HO and the minimum was found in the slag NO. These two slags, LO and HO contained hardystonite even before firing and the amount increased 3-6 times after firing . 68 \u2022 The amount of hardystonite decreased consistently with increasing the Cr2 O3 content in the mix and remained at relatively constant and low values for all N mixes (containing barren slag). Although the H mixes showed higher values at 10 % Cr 2O3 for 20=31 .16\u00b0 and lower at 28 .92\u00b0, the amount of hardystonite can be considered as decreasing almost linearly with increasing the % Cr2O3 . Based on these results, it can be stated that the hardystonite phase formation and its equilibrium with the molten material depends strongly on the lead concentration and reaches a maximum around 6 % Pb, corresponding to the low-lead slag LO. In regards to the solid solution of chromite spinels: \u2022 The maximum amount of spinel solid solution was found in the mixes H for 20% Cr2O3 after firing at 1250\u00b0C in argon. \u2022 The maximum amount in the initial slag was found in the low-lead slag LO, but it was higher in the HO slag after firing, for both positions of the peaks. \u2022 Only LO and HO contained spinel before firing, while the NO showed it only after firing. The amount increased 3 times according to the intensity values for peak 35 .41-35 .63\u00b0 and only 1 .5 fold for the other peak. \u2022 The amount of spinel solid solution increased consistently after firing with increasing the Cr2 O3 , for all experimental mixes N, L and H . The sharpest increase was for the L mixes, followed by N and H, although according to the data, the maximum was reached by the H3 mix, with 20 % Cr2 O3 . 69 \u2022 The amount of spinel increased with increasing the % Pb in the slag for pure slags and it had close values for the L and H mixes with increasing the Cr2O3. Both 20 peaks reached a maximum for 20 % Cr2 O3 in the high-lead slag H (H3). Based on the experimental results, the amount of spinel solid solution increased with the increase in Cr2O3 content. In addition, the formation of spinel increased in the mixes with lead, but all the barren slag mixes, without lead, reached comparable amounts when Cr2O3 was present. This demonstrated that the Cr2 O3 contributed in the precipitation of spinel solid solution from KIVCET slags. Accretion Formation: From the solubility test, it was demonstrated that the Cr2 O3 would rather form solid spinel phases than dissolve in the slag . It is important to mention that these spinel phases already exist in equilibrium with the melt at 1250\u00b0C and the Cr2 O3 would only increase their amount . If the spent lining from the silver refinery furnaces (BBOC) was recirculated in the KNCET slag in order to recuperate the Au and Ag, the Cr 2 O3 from the refractory brick would be increased, but it would not initiate the accretion formation. The tendency would rather be considered dependent on the Pb content in the slag and the po t in the furnace . The lead content in the slag possesses a greater role in the accretion formation than the small amount of Cr2O 3 (<5%), as shown by our experimental results . A small amount of Cr2O3 would influence more on a low-lead slag (6% Pb) than on a high-lead slag (11% Pb). However, a high-lead slag without chromium would have a higher tendency of precipitating spinel at 1250-1300\u00b0C than a low-lead slag . 70 It appeared that the barren slag would be the most affected by the increase in Cr 2O3 content and the low-lead slag would be the least affected . It is important to point out that even without Cr2O3, the barren slag would have the highest tendency to form spinel or \"accretions\" in the experimental conditions . The net contribution of Cr2O3 to the spinel formation is the highest for the barren slag mixes, followed by the high-lead and it is the lowest for the low-lead mixes. 5.1.7 Leaching Test for Hexavalent Chromium In the leach test, several magnesia-chrome bricks (MC-2, MC-4, MC-7 and MC-8) in contact with KIVCET slag were leached for Cr 3+ and Cr6+ by an extraction fluid composed of distilled water, which was brought to a pH of 4 .7 (with acetic acid) simulating acid rain . The results are shown in Table 20. Table 20. Trivalent and hexavalent chromium results from the leach test Chromium, mg\/L Sample Cr3+ Cr6+ MC-2 <0.004 0.036 MC-4 <0.007 0.043 MC-7 <0.003 0.040 MC-8 <0.003 0.035 The allowed limit of Cr6+ in drinking water is 0.05 mgCr6+\/L and is considered to be unlikely to pose significant health risks [321 . In the test samples, the Cr6+ was below the allowed limit and the risk of Cr+6 formation is practically nonexistent . In addition, Cr3+ is insoluble in water and is demonstrated by the test results . 71 5.2 Industrial Experiments 5.2.1 Peirce Smith Nickel Converter In order to further study the corrosion behaviour, a Postmortem Failure Analysis was carried out on magnesia-chrome bricks (MC-9) taken from the tuyere lining of the Peirce Smith (PS) Nickel Converter . Several brick samples were taken from an area as shown in Figure 26 but only one sample, labeled 2A, was fully analyzed. wwwwwwwwwwwwwww wwwwwwwwwwwwwwww w!w wRMw~wwMIwiMwwi~~RM~ww w_M1011.t.l. A\/i~_-i~1.A M~M1wyM wUMMMIIM1 w =w-Mwwwww =T~ wiwwwwwwwwM~wwwwww~ - t i ; wwwwwwwwwwwwwwwwwwti~twwi~s~i wiRwr~i~wwi~w w_ R ~._ ~iwi~IMIZ~ I. Hall wMiiw~ .1~I.M Mw_~ 1MRM __ww r~\u2022 ''\t wwwAW WAMP 111R1wwwwww___- w ww ww wwwwwww ____w~=~w___=01M __A=Mli1M~RMwMIWMM TM~rwZMOiw ~awMMwWMMMIMI== awoIw.M.M=W wMIAIMM wiPwlam. MMMM.~iwwimw.wimmme MMAMI ~w w~ \"MwMwwi.~\"7~~i~waaw.\"\"~wwwwlwwwwwwwwlw_!l~lwwww!!!wr!lwwwwwwwwwwwwlwwwwwwwlw w! mow_= =OMMiM, MUMi~ ~~MM owww===Mwi1= M====MIMIWI- sl . ..Mlwwwwwwww=11Owwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwiwwwwwwwwwwwwwwwwwwwwwwwwwwlwwwwwwwwwwwwwwwwwwwwwwwww lw\/AlwMwww\/iwwwwl\/islwww!lwslwwwwwwMw#wMwlwlMww WMMwwwwwwwwwwwwwwwwwwwwww .wwwwwwiwwwwwiww+NllwMllwwwwwwwafwMwlwwwwwwwlmIII mmell wwwwwwmmommm.11m.m.am mm.mwwwlwwYwww\/wwim.urm m.emMFww g ww~- M a ~llw!\u2022\u2022wM \/Owflt wMwawiMM MMww'nMM RS OOMMMMMj LONGITUDINAL SECTION SAMPLES TAKEN Figure 26. Position within the Peirce Smith Nickel Converter where the bricks from the tuyere lining were sampled [33] L Cwwwwwwwwwwww wwwwwwww 0''ww~wwwwww wwrww ww ---wwww!llwwwwwwww~r=wwwwwwwwwwl~wwwwwwwwlwwwlww~iww~wwwwl~rw~tww+~ \u2022I' MOMdcMMIF n :s4-Ja`aii 'tia wasSi, . : 'a k'de-d it'iMI12r n 72 Visual Inspection . After the brick samples were cut in the longitudinal direction, visual inspection indicated severe cracking parallel to the hot face as shown in Figures 27A-C. Brick IA was severely damaged as large open cracks, measuring up to 2 to 4 .5 inches from the hot face, were identified in the densified (impregnated) layer . This sample displayed a remaining portion from a previous impregnated layer in which the missing portion was probably spalled off during operation . Brick 2A was not as severely damaged as IA but micro cracks were observed parallel to the hot face between the penetrated and un-penetrated regions . Brick 3A showed more evidence of the thermal-mechanical damage as a large open crack could be seen within 1 inch from the hot face . Based on these observations, brick IA was thought to represent the best example of thermal mechanical spalling. Physical Properties . The property profiles, which include the Modulus of Rupture (MOR), Cold Crushing Strength (CCS), Apparent Porosity (AP), Water Absorption (WA) and Bulk Density (BD), for brick 2A clearly showed a distinct interface between the penetrated region at the hot face and the original (un-penetrated) brick . The AP, WA and BD results (Figure 28A-D) showed that the impregnation was up to 90-100 mm from the hot face. The MOR results showed expected higher values in the impregnated areas towards the hot face. For the remaining original brick, the values were similar to the unused brick . The CCS results followed a similar trend to the MOR, showing 2-4 times higher values in the impregnated areas at the hot face than in the rest of the brick . The tremendous increase in strength in both the MOR and the CCS in the densified region would increase the modulus 73 Spalled Portion Figure 27 . Cross-section images of the tuyere bricks for : A) IA, B) 2A and, C) 3A 74 30 .0 Used --+\u2014 AP(%) \u2014 .\u2014WA (%) 25 .0 - --*-- BD (g\/cm3) \u2022 MOR (MPa) 20 .0 - 15 .0 - 10 .0 \u2022 \u2022 \u20225 .0 - 0 .0 a 0 25 50 75 100 125 150 175 200 225 Distance from Hot Face (mm) A) 30 .0 Unused AP (%) .\u20140\u2014 WA (%) 25 .0 - --- BD (g\/cm3) . -s-- MOR (NPa) 20 .0 - 15 .0 - 10 .0 - 5 .0 i:7=4\u201e o .o 0 50 100 150 200 250 300 3501 Distance from Hot Face (mm) B) 0 50 100 150 200 250 300 350 Distance from Hot Face (mm) C) D) Figure 28 . Profile of the properties for MC-9 . Apparent Porosity (AP), Water absorption (WA), Bulk Density (BD) and Modulus of Rupture (MOR) for : A) used brick, B) unused brick and, Cold Crushing Strength for : C) used brick, D) unused brick 75 of elasticity (MOE) resulting in an increase in stiffness . The increased stiffness in the densified region could result in cracking at the penetrated front between the impregnated and un-impregnated zones during the thermo-mechanical behaviour (freeze-thaw cycle). Although no large cracks were observed at these boundaries, this phenomenon is a reasonable assumption due to the micro cracks observed in this region . It could also be assumed that these micro cracks were formed during vessel operation or due to \"mechanical punching\" prior to brick sampling. SEM and EDS analysis . SEM images and EDS point analyses in the impregnated zone determined several details . Slag and matte were identified in the accretion layer adhered to the hot face of the brick as shown in the SEM and EDS mapping images in Figures 29A and 29B respectively. The slag was a fayalitic type (FeO\u2022SiO2) slag due to the high contents of iron, silica and oxygen detected by EDS, which was typical in non-ferrous slags . In addition, the major phase in the matte was a nickel sulphide with a minor phase as a copper sulphide. The detection of the high sulfur content in the matte by EDS confirmed that the matte remained in its sulphide phase . A full elemental analysis of the slag and matte could be found in Table 21. SEM images within the impregnated layer showed cracks through periclase grains and along periclase and chrome grain boundaries as displayed in Figures 30A and 30B . The cracks were assumed to have been developed between two pores impregnated with matte or 76 x200 0000 kV 200um A) B Figure 29 . A) SEM and B) EDS mapping of the slag and the matte residing at the hot face, (200x) . The corresponding EDS analysis is shown in Table 21 Table 21. EDS analysis of slag and matte corresponding to Figure 29 Element, wt.% 0 Mg Al Si S Cr Fe Co Ni Cu Slag 17 .84 0.41 1 .51 9 .29 1 .31 1 .49 60.36 1 .24 3 .07 3 .14 Matte (Cu) 0.63 0.30 0.08 0.08 18 .29 1 .33 8 .23 0.33 0.00 70 .74 Matte (Ni) 0.00 0.25 0.10 0 .16 21 .81 2.20 12.25 1 .78 56.68 4 .77 x150 0000 20kV 200um A) B) Figure 30 . SEM images (A, 150x, B, 80x) displaying cracks through MgO grains and along grain boundaries suspected to be initiated by penetrated matte 77 between an empty and an impregnated one . Although it was not known when these cracks were developed, it was suspected that they were formed through the interaction between the matte and the refractory. The formation of these cracks was assumed to have been caused by thermal mechanical stresses initiated by large temperature fluctuations . Although this was a likely case, cracks may also have been developed if the matte in the impregnated region was partially oxidized to NiO or Cu20 (CuO) and reacted with the refractory forming a spinel with magnesia . The formation of the spinel would be accompanied by a volume expansion resulting in localized stress initiating cracks . To further examine this hypothesis, an EDS line analysis was conducted across the periclase and matte interface as shown in the Figure 31A with the corresponding EDS analysis shown in Table 22 . At point A in the matte, the matte was in the sulphide phase. Once crossing into the periclase grain, no sulfur was detected but nickel at a concentration as high as 44 .8 wt.% was identified coexisting with magnesium and iron. From this observation, a spinel may have formed from the diffusion of Nit+ into the periclase forming a (Mg, Fe, Ni) X OX spinel . In addition, cobalt at a concentration of 4 .5 - 4.8 wt .% was detected at the periclase rim and could be a (Mg, Fe, Co) XOX spinel . Furthermore, the line analysis conducted across the matte and chrome interface revealed that no components of the matte have diffused into the chromite grain as displayed in Figure 31B with corresponding EDS analysis shown in Table 23 . 78 Matte Periclase Chrome Ore Matte x800 0012 2.0 kV 50ssm x.800 0012 20kV 50sam Figure 31 . A) Matte and MgO interface, (800x) and, B) matte and Cr203 interface, (800x The corresponding EDS analyses are shown in Table 22 and 23, respectively Table 22. EDS profile of elements across the interface between the matte and the MgO grain corresponding to Figure 31A Distance (pm) Concentration, wt .% 0 Mg Al Cr Fe S Ni Cu Co 0 1 .07 0.00 0 .00 0.00 0.00 25 .16 73 .77 0.00 0.00 10 1 .99 0.22 0 .00 0.00 0.30 18.54 72 .85 4.53 1 .56 18 2 .26 0.51 0 .00 0 .00 0 .00 24.04 73 .20 0.00 0.00 22 * 24.33 15.81 0 .11 0.63 9 .37 0.00 44 .84 0.40 4.51 30 * 25 .16 20 .80 0.07 0.94 7.54 0.00 40.39 0.30 4 .80 40 * 34.21 57.48 0.20 2 .69 4.13 0.00 0.58 0.29 0 .42 * Note : EDS point analysis in the periclase grain Table 23. EDS profile of elements across the interface between the matte and the Cr203 grain corresponding to Figure 31B Distance\t Concentration, wt.% (pm) 0 Mg Al Cr Fe S Ni Cu Co 0 1 .31 0.00 0.00 0.00 0.00 24.62 74.07 0.00 0 .00 10 1 .38 0.00 0.00 0.00 0.00 24.59 74.04 0 .00 0.00 18 1 .13 0.00 0.00 0.33 0.00 24.34 73 .75 0.44 0.00 22 * 39.49 14.48 17.60 20.85 6.36 0.08 0.66 0.25 0.23 30 * 31 .69 13 .09 14.85 33 .87 5 .56 0 .00 0.45 0 .28 0.20 40 * 38 .27 13 .78 17.28 25 .34 4.49 0.07 0.29 0 .30 0.18 * Note : EDS point analysis in the periclase grain 79 SEM images of the cold end of the used brick and the unused brick are shown in Figures 32 and 33 respectively. The following phases were identified: periclase grains, chrome grains, silicate and spinel phases. Spinel phases were both identified at the grain boundaries between grains and precipitated within the periclase grains . There were regions at the grain boundaries with no spinel precipitation and regions with fine spinel precipitation, which promoted direct bonds. The grain boundaries with little or no spinel precipitation were mostly observed between large adjacent grains with less surface contact . Silicate phases high in silica and calcia in minute amounts were found in \"pockets\" collected in-between . A full EDS analysis is shown in Tables 24 and 25 for used and unused bricks respectively. In order to determine the depth of penetration of the different elemental compositions, an EDS profile from the hot face to the cold face was created as shown in Figure 34 . The concentration profiles for magnesium, aluminum, chromium, iron, silicon and calcium, which were typically found in the refractory, showed a normal distribution though the longitudinal section . The chromium and aluminum fluctuated considerably and was mainly due to the detection of large chromite grains and precipitated spinel . There were deviations observed within the first 15 to 20 mm from the hot face . High amounts of silicon and iron were identified at the hot face, which was due to the penetration of the fayalitic slag (FeO\u2022SiO2). Magnesium at a concentration of 5-10 wt.% was detected in the slag (accretion) layer close to the interface and may suggest that the MgO grains were dissolved into the slag or the grains were \"washed\" away. 80 A. SEM of overall surface, 80x A. SEM of overall surface, 80x B. Magnified view of spinel phase, 800x .5k 0000 20kV 20mm B. Magnified view of spinel phase, 1500x C. Magnified view of silicate phase, 1500x C. Magnified view of silicate phase, 1000x Figure 32 . SEM images at the cold face of the used MC-9 brick. Figure 33 . SEM images of the unused MC-9 brick 81 Table 24. EDS point analysis at the cold face of the used MC-9 brick corresponding to Figure 32 Element, wt.% 0 Mg Al Si Ca Cr Fe Co Ni Cu Periclase 34.37 57.10 0.54 0.39 0.00 3 .16 3.46 0.15 0 .20 0 .53 Chromite 30.88 10.53 7.49 0.12 0.00 36.40 12.56 1 .17 0.00 0.85 Silicate 31 .37 17 .84 0.13 21 .01 28.09 0.23 0.34 0.23 0.18 0.57 Spinel 31 .26 25.23 10.51 0.25 0.00 20.75 10.61 0.66 0.00 0.74 Table 25. EDS point analysis of the unused MC-9 brick corresponding to Figure 33 Element, wt.% 0 Mg Al Si Ca Cr Fe Co Ni Cu Periclase 34 .04 54.38 1 .50 0.29 0.00 4.27 4.72 0.28 0 .13 0 .39 Chromite 26 .67 11 .44 6 .93 0.01 0.02 40.97 11 .99 0.76 0.00 1 .21 Silicate 34.99 17.01 5 .46 15.49 20.06 3 .51 2.76 0.43 0.00 0.29 Spinel 29.26 25 .74 9.29 0.35 2.54 17.77 13.80 0.53 0.00 0.73 Deep into the refractory, elements from the matte such as nickel, copper and sulfur were found up to 79 mm from the hot face . The EDS profile identified a higher concentration of nickel compared to copper, which indicated that the major penetrated phase was a nickel sulphide and the minor being a copper sulphide . SEM and EDS mapping images, shown in Figures 35A and 35B respectively, identified the penetrated elements residing at the grain boundaries of periclase-periclase and periclase-chrome grains and were likely deposited through pores and cracks . 82 70 .0 60.0 - Mg (wt.%)----- Fe (wt .%) \u2014~\u2014 0 (wt .%) 50.0 - 40.0 - 20 .0 10 .0 0.0 -25 0 25 50 75 100 Distance from Hot Face 125 (mm) 150 175 200 225 20 .0 \u2014 (wt.%) \u2014 Cr (wt .%) \u2014.--AI 15.0 - 10 .0 - 5 .0 0 .0 -25 0 25 50 75 100 125 150 175 200 225 Distance From Hot Face (mm) 15 .0 Ni Cu S (wt .%) Co (wt.%)\u2014\u2666-- (wt.%) -a\u2014 (wt .%) -4-- -A 1\u2014 10 .0 - 5 .0 - -25 0 25 50 75 100 125 150 175 200 225 Distance from Hot Face (mm) Figure 34 . Elemental profile for the used MC-9 brick through the longitudinal cross-section -4-- Ca (wt.%) \u2014.\u2014Si (wt.%) 6 .0 4 .0 0ril~ lf~_ av nt. 0 .0 -25 0 25 50 75 100 125 150 175 200 225 Distance from Hot Face (mm) 2 .0 - 10 .0 8 .0 83 A) B) Figure 35. A) SEM and, B) EDS mapping image at 79 mm into the refractory displaying the penetration of matte high in nickel, cooper and sulfur, (60x) X-Ray Diffraction . Through XRD, the iron and silicon detected by EDS in the accretion slag layer at the hot face was confirmed to be a fayalitic phase (FeO\u2022SiO 2), which was typical in slags found in non-ferrous converting processes (Figure 46) . Iron was also found in the form of magnetite . Magnetite (Fe3 O4) may have formed during the blow in the converting process where FeS was rapidly oxidized to hematite (Fe 2O3) rather than combining with SiO 2 to form fayalite FeO\u2022SiO 2. Magnetite was typically found at the tuyere region due to a high local concentration of oxygen, provided by the tuyere, which rapidly oxidized iron sulphide to magnetite creating an accretion layer . This accretion or build up of magnetite would eventually adhere and blocked the tuyere mouth . Therefore, frequent \"punching\" was required to clear the tuyere mouth. In addition, nickel and copper in the accretion layer were identified as heazlewoodite (Ni3S2) and chalcocite (Cu2S) respectively. 84 These two sulphides do not interact with the refractory unless it was found to be oxidized to a nickel oxide (NiO) or a copper oxide (Cu2O or CuO). Figure 36 . XRD peaks for slag and matte The XRD profile through the longitudinal cross-section showed penetration of heazlewoodite up to 95 mm into the brick as shown in Figure 37 . Although the penetration of nickel sulphide was found deep into the brick, there was not data in support of the nickel sulphide being oxidizing to nickel oxide by XRD. In addition, there was no confirmation of the chemical alteration of the magnesia grains . It could be possible that the concentration was well below the detection limit . A Heazlewoodite, Ni3 S 2 ; V Magnetite, Fe30 4 n Chalcocite, Cu2 S ; \u2022 Fayalite, FeO-SiO2 N a) 0 10 20 30 40 50 60 70 80 90 2-Theta V \u2022 v \u2022 I A A A vjt,\u201e7)i i 85 Distance From Hot Face, mm x A A 0 Periclase, MgO ; A Heazlewoodite, Ni3S 2 x Magnesiochromite, (Mg, Fe)(Cr, A1) 2 O 4 4 H1 c a) c 95 14 152 190 221 0 10 20 30 40 50 60 70 80 90 2-Theta Figure 37 . XRD profile for the used MC-9 brick through the longitudinal cross-section Figure 38 . XRD peaks for the unused MC-9 brick 0 Periclase, MgO x Magnesiochromite, (Mg, Fe)(Cr, A1) 2 O 4 U) c m 0 10 20 30 40 50 60 70 80 90 2-Theta 86 In the original portion of the used brick where penetration did not reach, XRD has identified periclase (MgO) and a spinel phase of magnesiochromite (Mg, Fe)(Cr, A1) 204. No silicate phases were found through XRD although they were confirmed though SEM and EDS . The concentration level of the silicate phases may have been too small and was likely to be below the detection limit . These phases were similarly found in the unused brick with the XRD peaks shown in Figure 38. Based on the results, several corrosion mechanisms could be hypothesized in order to explain the premature failure of the tuyere brick. The first mechanism could be due to thermal mechanical spalling caused by stresses developed from the freeze-thaw cycle . As seen in the cross-section images, cracks were observed in the densified region and large pieces were spalled from the hot face . Although this was a likely case, the period in which these events occurred was not known. The second mechanism could be due to the interaction of the oxidized matte and the refractory . The matte, high in nickel sulphide, penetrated into the refractory in which the matte partially oxidized to nickel oxide (NiO) and interacted with the magnesia grains forming a (Mg, Fe, Ni) XO X spinel . The matte could have been oxidized by the oxygen originating from the cold end of the brick . The oxidized components of the matte (Cu2O, NiO and CoO) could form spinels with magnesia and generate cracks at the grain boundaries . The cracks could propagate parallel with the hot face during thermal cycles between blows. The interface of the magnesia grain showed partially oxidized matte rims, which could explain the presence of Co, Cu and Ni as spinels precipitated inside the periclase grains . 87 5.2.2 KIVCET Electric Furnace Roof A Postmortem Failure Analysis was conducted on used bricks from a KIVCET Electric Furnace (EF) roof in order to characterize the corrosion mechanism(s) that caused the roof's premature failure. The original roof remained in operation for 10 years, while the current roof lasted only 2 years due to severe sheet spalling as shown in Figure 39 . In both roofs, the same type of brick was used . A few brick samples were recovered after the roof was decommissioned and the exact location of these bricks within the roof was unknown . In the last few years, residues of an unknown composition have been added to the charge, in which some volatile element may have contributed to the roof's failure. Figure 39. Image of a spalled portion in the KIVCET Electric Furnace roof [1] Visual Observations. After the brick was sectioned in the longitudinal direction, several details were observed as shown in Figures 40 and 41 . Penetration was measured up to 190 88 mm from the hot face, in which two different phenomena were identified. The first was a porous zone found within 30-100 mm where two large open cracks parallel to the hot face were identified at 58 mm and 67 mm . A second region identified was a \"red\" zone concentrated between 103-120 mm from the hot face from which 190 mm towards the cold face remained the original brick. Physical Properties . From the Apparent Porosity (AP), Water Absorption (WA) and Bulk Density (BD) results, the property profile through the longitudinal direction changed with the distance from the hot face as shown in Figure 42A . At the hot face, a densified zone with an AP of 5 .9% was observed, which was much lower compared to an AP of the unused brick (13.9%) shown in Figure 42B . In addition, the AP of the red zone (7-8%) was also found to be much higher than the unused brick . The high AP indicated a gain in mass, which was likely due to the precipitation of elements from the furnace atmosphere into the pores of the refractory . In the other case, a low AP was found in the porous zone (16-19%), which was likely due to a loss of refractory. Comparing the BD of the used and unused bricks, it was observed that the BD for the penetrated zones was much higher (-4 .2g\/cm3 ) and the BD for the porous zone was much lower (-2 .8g\/cm3) in comparison to the unused brick (3 .25 g\/cm3 ). The MOR and CCS profiles, shown in Figures 42A and 42C respectively, followed a similar trend to that of the absorption test . At 61mm into the sample, a low MOR (16 .75 MPa) and a low CCS (45 .93 MPa) result were determined between two densified zones . Although 89 Figure 40 . Cross-section of the used KIVCET Electric Furnace roof brick Figure 41 . Cross-section of the used KIVCET Electric Furnace roof brick with measurements of penetration 90 25 30 20 15 10 Unused -- .\u2014AP (%) -4\u2014 WA (%) --*\u2014 BD (glcm3) --\u2022\u2014 MOR (MN) 50 100 150 200 250 300 Distance from Hot Face (mm) 0 A) 120 Used \u2014.\u2014ccs (MPa) 100 - 80 - 60 - 40 - 20 - 0 0 50 100 150 200 250 300 350 Distance from Hot Face (mm) C) B) 120 Unused \u2014.\u2014ces (MPa) 100 - 80 - 60 - 40 - 20 - 0 0 50 100 150 200 250 300 Distance from Hot Face (mm) D) Figure 42. Profile of the properties for MC-11 . Apparent Porosity (AP), Water absorption (WA), Bulk Density (BD) and Modulus of Rupture (MOR) for : A) used brick, B) unused brick and, Cold Crushing Strength for : C) used brick, D) unused brick 91 these low strength results were not as low as the unused brick, it remained as an unusual phenomenon occurring in this region . The remaining original brick had values much lower compared to the un-used brick, which may signify that the brick has lost its physical integrity during operation. SEM and EDS Analysis. In Figures 43A-B, SEM images of the porous zone at 70 mm from the hot face, identified large cracks spanning the width of the brick parallel to the hot face . The coarse grains of magnesia and chrome were de-bonded from the matrix causing a \"loose\" microstructural texture . Examining the microstructure further with the support of EDS, there was a deep reaching sulphate, which mostly affected the bricks bonding phase A) B) Figure 43 . A-B) Large cracks in the porous zone at 70 mm from the hot face, (40x) 92 between the coarse magnesia grains . As shown in Figure 44 and 45 with corresponding EDS analysis (Table 26 and 27 respectively), sulphur was identified in the presence of calcium and potassium, which was likely a calcium sulphate, potassium sulphate or a calcium potassium sulphate. Typically not found in the brick, sulfur and potassium were likely from the furnace atmosphere as gas and vapors, which reacted with calcium silicates in the brick's matrix. The sulphate attack was most evident in the porous zone. Following the porous zone, there was a lead densified microstructure as shown in Figure 46A-B . A \"band\" of lead, up to 5 mm thick, was found spanning across the width of the brick, which flooded between coarse grains and throughout the matrix . The strong flooding of lead caused most of the connecting grains to de-bond, which weakened the brick's structure. From the EDS line analysis shown in Figure 47, the only foreign elements identified were lead and zinc . Both lead and zinc were found in high concentrations penetrated deep into the brick. The deep penetrated lead reached up to 221 mm from the hot face where the highest concentration was found between 106 mm and 146 mm . Within this region at 106 mm, a highly concentrated lead region, at almost 100 wt .%, was observed as shown in the corresponding Figure 48. At 116 mm, severe flooding continued through the matrix and through the grain boundaries engulfing the grains with lead (Figure 49) . Zinc did not penetrate the entire sample, as it only existed up to 65 mm from the hot face . The remainder of the brick did not contain any lead or zinc . 93 Figure 44 . SEM image of the porous zone displaying : 1-4) sulphate phase, 5) lead oxide phase, 6) Cr2O3 grains and, 7) MgO grain, (150x). The corresponding EDS analysis is shown in Table 26 Figure 45 . SEM image of the porous zone displaying : 1-5) sulphate phase, 6) lead oxide and, 7) MgO grain, (150x). The corresponding EDS analysis is shown in Table 27 Table 26. EDS point analysis in the porous zone corresponding to Figure 44 Point Element, wt .%O Mg Al Si S K Ca Cr Fe Pb 1 . Sulphate 23 .20 0.00 0.00 0.08 27 .35 0 .49 48 .88 0.00 0 .00 0.00 2. Sulphate 18 .50 0.00 0.00 0.12 27 .72 17 .32 36.35 0.00 0 .00 0.00 3. Sulphate 21 .67 0.00 0.00 0.08 20 .00 25 .51 21 .22 3 .19 8 .32 0.00 4. Sulphate 20.47 0.00 0.00 0 .04 26 .30 16 .54 36.64 0.00 0 .00 0.00 5. Lead oxide 6.78 0.00 0.00 0.00 0 .00 0 .00 0.00 10.51 0 .00 82 .71 6. Cr2 O3 grain 10.50 5 .01 2.46 0.00 0.00 0.00 0.00 46.45 35.58 0.00 7. MgO grain 39.70 60.30 0.00 0.00 0.00 0 .00 0.00 0.00 0 .00 0.00 94 Table 27 . EDS point analysis in the porous zone corresponding to Figure 45 Point Elements, wt .% 0 Mg Al Si S K Ca Cr Fe Pb 1. Sulphate 16.98 0.00 0.00 0.81 25 .15 36.15 20.91 0.00 0.00 0 .00 2. Sulphate 10.72 0.00 0.00 0.08 27.68 32.37 29 .15 0.00 0.00 0 .00 3. Sulphate 18 .56 0 .00 0.00 0.11 27 .92 14.80 38 .60 0.00 0.00 0.00 4. Sulphate 17.11 0.00 0.00 0.24 24.11 25 .88 25.60 0.00 7.07 0.00 5. Sulphate 28.20 0.00 0.00 0.04 24.75 14.99 32.03 0.00 0.00 0.00 6. Lead oxide 2.29 0.00 0 .00 0.00 0.00 0.00 0.00 9.86 0.00 87.85 7. MgO grain 13 .91 36.12 0.18 0.00 0.00 0.00 0.00 7.35 42.44 0 .00 A) B) Figure 46 . A-B) High concentration of lead in the \"red\" zone at 106 mm from the hot face, (40x) Hot Face I 95 100 80 60 40 20 350 ---- Pb2O3 t.%) ---- ZnO t .%)(w(w 0 25 50 75 100 125 150 175 200 225 250 275 Distance from Hot Face (mm) 300 325 100 80 60 40 20 -- 7 350 MgO t .%)\u2014+\u2014 (w -- AI2O3 t .%)(w 0 25 50 75 100 125 150 175 200 225 250 275 Distance from Hot Face (mm) 300 325 ' 10 8- 6- 4- SiO2 t .%)-4\u2014 (w CaO t.%)(wt \u2666 1t 0 25 50 75 100 125 150 175 200 225 250 275 300 325 350 Distance from Hot Face (mm) 40 \u2014~\u2014Cr2O3(wt.%) \u2014k\u2014Fe2O3(wt.%) 30 - 20 - i 0 0 25 50 75 100 125 150 175 200 225 250 275 300 325 350 Distance from Hot Face (mm) Figure 47. Elemental profile for the used MC-11 brick through the longitudinal cross-section 96 Figure 48 . SEM image at 106 mm into the brick displaying 100% lead oxide, (200x) Figure 49 . SEM image at 116 mm into the brick displaying lead oxide engulfing the refractory grains, (200x) X-Ray Diffraction . The XRD results (Figure 50) identified lead as lead oxide (PbO) in the forms of Misscot and Litharge, in which the intensity of the lead oxide peaks was the highest at 117 mm. In addition, lead oxide in the form of PbO2 was only found at 117 mm and zinc lead oxide (ZnPbO3) was identified in the first 60 mm . At the cold end of the brick, the main phases identified were periclase (MgO) and the complex spinel, magnesiochromite (Mg,Fe)(Cr,Al) 2 O4 similar to the unused brick. 97 ADistance from Hot 0 Periclase, MgO ; q Zinc Lead Oxide, ZnPbO 3 \u2022 Magnesiochromite, (Mg, Al)(Cr, Fe)2O4 \u2666 Misscot, PbO ; V Litharge, PbO 0 Lead Oxide, PbO 2 Face, mm A A 10 .v. . . o A 29 49 74 97 >, o ~n 117 139 160 181 202 221 236 25B 281 303 322 1 . __ A- 0 10 20 30 40 50 60 70 80 90 2-Theta Figure 50 . XRD profile for the used MC-11 brick through the longitudinal cross-section 98 Based on the results, several hypothesized mechanisms were proposed that contributed to the brick's premature failure . The first mechanism was a highly infiltrated lead zone, in particular the \"band\" of lead at 106 mm from the hot face spanning the wide of the brick. This lead infiltrated zone was created by the condensation of lead fumes from the furnace atmosphere. The severe flooding of lead through the matrix caused the bonds between adjacent grains to break, which weakened the brick's structure . The second mechanism was due to a sulphate attack, which affected the brick's bonding phases and was most evident in the porous zone . The deep reaching sulphate attacked mostly at the calcium silicate phases in the matrix, which weakened the matrix and contributed to the deterioration of the brick's structure . 99 5.2 .3 .1 Bottom Blown Oxygen Converter The performance of several magnesia-chrome bricks was tested in a Bottom Blown Oxygen Converter (BBOC) at Teck Cominco Metals Ltd ., Trail operations . The bricks were lined at the slag line where the highest corrosion was observed . Three of each different type of brick was lined side-by-side within the BBOC positioned in the third row from the bottom as shown in Figure 51 . A Postmortem Failure Analysis and full characterization were performed on all three types of brick (MC-2, MC-8 and MC-10). Figure 51 . Position of the three different types of trial bricks (MC-2, MC-8 and MC-10) in the Bottom Blown Oxygen Converter ril In a regular campaign of the BBOC, the corrosion of the lining at the slag line was so severe that the bricks were practically disintegrated as shown in Figures 52 . In order to obtain 100 bricks that were large enough to conduct a Postmortem Failure Analysis (Figure 53), the campaign of the converter, which contained the three different types of test brick, was ended prematurely. less than 75 mm out of 228 mm of brick remained Figure 52 . Image of the BBOC brick : A) at the end of one regular campaign and, B) magnified image showing the brick's texture at the hot face Figure 53 . Images of the trial bricks : A) at the end of the premature campaign and, B) cracking at the hot face 101 Visual Observation. Visual inspection on the cross-sections of all three types of brick identified cracks that were both in the parallel and perpendicular directions to the hot face. Large macro-cracks in the perpendicular direction were observed close to the hot face, while micro-cracks continued directly to the cold face as shown in Figures 54-56 . Although cracks in the perpendicular direction were identified, the severity of the cracks in the MC-8 and MC-10 bricks was greater than the MC-2 bricks . Similarly, cracks in the parallel direction in bricks MC-8 and MC-10 were more severe than MC-2 bricks, which nearly caused the brick to spall at the hot face as shown in samples MC-8 and MC-10 (Figure 55 and 56, respectively) . In addition, silver was identified penetrated at the hot faces of MC-2 and MC- 10, while no silver was observed at the hot face of MC-8. Figure 54 . Cross-section of MC-2 displaying : 1) macro-crack and 2) micro-crack perpendicular to the hot face, 3) micro-crack parallel to the hot face and, 4) silver 102 Figure 55 . Cross-section of MC-8 displaying : 1) macro-crack and 2) micro-crack perpendicular to the hot face, 3) macro-crack parallel to the hot face and, 4) silver Figure 56 . Cross-section of MC-10 displaying : 1) macro-crack and 2) micro-crack perpendicular to the hot face, 3) micro-crack parallel to the hot face and, 4) silver 103 Volume Loss . The volume loss was determined by comparing the dimensions of the remaining used brick to the original dimensions of the unused brick . The measurements were taken at 25.4 mm intervals laterally across the face of the specimen . The approximate volume of each sample was calculated and shown in Table 28. Table 28. Volume loss of the used MC-2, MC-8 and MC-10 bricks Sample No . Volume Loss, %MC-2 MC-8 MC-10 1 30.8 40.6 26.8 2 30.2 32.6 32.4 3 31 .1 33 .4 33 .3 Average 30.7 35.5 30.8 Average* 30.8 33.4 33.3 *Note : Average taken with the high and low values discarded The MC-8 bricks were found to have the greatest volume loss at 35 .5% while both MC-2 and MC-10 type bricks were identical at approximately 31% . From this observation, it could be suggested that the MC-8 brick was the least resistance to corrosion based on the average volume loss. In another scenario, eliminating the high and low values from the average would give an approximate equal volume loss in all types (\u2014 31-33%) . Therefore, it is inconclusive as to which type has the best resistance to corrosion based on volume calculations. Physical Properties . For all three types of brick, the Apparent Porosity (AP), Water Absorption (WA) and Bulk Density (BD) profiles remained constant with the increasing distance from the hot face as shown in Figures 57-59 . The AP (<1 .0%) and WA (<0 .25%) 104 were found to be very low in comparison to the unused bricks and there were no significant differences between each type of brick . The low AP and WA results were due to severe penetration through connected open pores, which resulted in a high BD of 4 .5 g\/cm 3 consistent through the longitudinal section . In the unused bricks, the physical property profiles were relatively similar to each other in terms of AP, WA and BD. The MOR results showed expected higher values due to the impregnation of molten liquids. The CCS results (Figure 60) followed a similar trend to the MOR, showing 4-6 times higher values compared to the unused brick . The tremendous increase in strength would increase the modulus of elasticity (MOE) resulting in an increase in stiffness . The increase in stiffness could initiate cracks if any thermo-mechanical behaviour were present . In addition, \u2014d\u2014AP(%) --.\u25ba\u2014WA (%) \u2014~-- BD (glcm3) --f\u2014 MOR (NPa) A) 0 25 50 75 100 125 150 175 200 225 Distance from Hot Face (mm) B) Figure 57 . Profile of the properties : Apparent Porosity (AP), Water absorption (WA), Bulk Density (BD) and Modulus of Rupture (MOR) for MC-2 : A) used and, B) unused 105 40 --+\u2014AP(%) WA (%) 35 - \u2014*\u2014 BD (g\/cm3) --~-- MOR (MPa) 30 - 25 - 20 - 15- 10- ~` \u2666 \u2022 \u2666 A 0 0 25 50 75 100 125 150 Distance from Hot Face (mm) 40 ---4--- AP (%) --s \u2014 WA (%) 35 - \u2014- BD (g\/cm3) --f-- MOR (MPa) 30 - 25 - 20 - 15- 10- 5- 0 0 25 50 75 100 125 150 175 200 225 Distance from Hot Face (mm) A) B) Figure 58 . Profile of the properties : Apparent Porosity (AP), Water absorption (WA), Bulk Density (BD) and Modulus of Rupture (MOR) for MC-8 : A) used and, B) unused A) B) Figure 59 . Profile of the properties : Apparent Porosity (AP), Water absorption (WA), Bulk Density (BD) and Modulus of Rupture (MOR) for MC-10 : A) used and, B) unused -4\u2014AP (%) \u2014.\u2014WA (%) \u2014~ BD (g\/cm3) \u2014I-- MOR (MPa) o A\t A 0 Pr-- 0 25 50 75 100 125 150 Distance from Hot Face (mm) \u2014e\u2014 AP (%) \u2014i\u2014 WA (%) \u2014, \u2014 BD (g\/crn3) --~\u2014 MOR (MPa) 0 25 50 75 100 125 150 175 200 225 Distance from Hot Face (mm) 40 35 30 25 20 15 10 5 0 40 - 35 30 25 20 15 10 106 225 --*-- MC-2 ---1~-- MC-8 --e -- MC-10 225 \u2014~\u2014MC-2 \u2014 -.MC-8 --* MC-10200 - 200 - 75 75 0 25 50 75 100 125 150 Distance from Hot Face (mm) 0 25 50 75 100 125 150 175 200 225 Distance from Hot Face (mm) 175 - 150 - L) 175 - 150 - co(L 125- Q100- U 50 25 - 50 - 25 A) B) Figure 60 . Profile of the Cold Crushing Strength (CCS) for MC-2, MC-8 and MC-10 : A) used and, B) unused propagating cracks may not be obstructed due to the highly dense brick . The pores were almost completely infiltrated resulting in a very low porosity brick, which removed any potential impediment to obstruct a propagating crack. SEM and EDS Analysis . Shown in the SEM images of the used MC-2, MC-8 and MC-10 bricks close to the hot face (Figures 61-63 respectively), very minimal to no open pores were observed due to the deep reaching lead and bismuth oxides infiltrating the open pores and the grain boundaries of the periclase-periclase and periclase-chrome ore . The EDS point analysis at the grain boundaries found no precipitated complex spinels remaining, in which 107 these spinels were usually found in magnesia-chrome bricks as seen in the unused samples in Figures 64-66 for MC-2, MC-8 and MC-10 respectively. Several regions of the strongly infiltrated microstructure were investigated with little success in identifying the original precipitated complex spinels at the grain boundaries . Although some precipitated complex spinels in the odd regions were identified, as shown in the EDS point analysis in Figure 67B point 5, they were highly infiltrated with lead and bismuth oxides . The severe infiltration of lead and bismuth demonstrated a highly aggressive liquid that dissolved or damaged the precipitated complex spinel bonds . The damage to the bonds resulted in less bonding and a weakened structure, which gave way to intergranular infiltration and caused the grains to be easily \"washed\" or eroded away at the hot face . In a hypothesized mechanism, lead and bismuth oxides may react with SiO 2 containing phases forming low melting liquids, which may further dissolve MgO. As oxides in the brick are formed at the grain boundaries, the liquid phase containing PbO and Bi2O3 would dissolve MgO and destroy the grain boundaries. Examining the microstructure of the unused bricks, the pore dimensions and grain size in the MC-2 brick were observed to be much larger than MC-8 and MC-10 and the latter two have similar microstructures . The smaller pore dimensions of MC-8 and MC-10 were considered very encouraging in regards to decreasing the slag penetration and nearly preventing the silver from penetrating. The elemental profiles by EDS through the longitudinal section found that all bricks were highly infiltrated with lead and bismuth directly to the cold face, in which the bismuth to 108 lead ratio was approximately 2 to 1 as shown in Figures 67-69 for bricks MC-2, MC-8 and MC-10 respectively . In addition, silver was also found to have penetrated into the bricks . In the MC-2 brick, silver strongly infiltrated through the entire thickness of the brick. Although MC-2 was highly infiltrated, no silver was found at the hot face of MC-8 but surprisingly, silver was identified at 118 mm away from the hot face . The deposition of silver at the cold face can be explained as follows . Sliver was found infiltrated through the mortar (or joints) between adjacent bricks in the lining, in which silver deposited where pores were large enough to accept silver, in this case at the cold face region. In the MC-10 brick, silver penetrated the hot face but there was a central region between 50-115mm where silver was not identified. This would be a similar case to the MC-8 brick where silver penetrated through the mortar and deposited into pores, large enough to accept silver. X-Ray Diffractometry. From the XRD profiles shown in Figures 70-72 for MC-2, MC-8 and MC-10 respectively, the results showed that lead oxide was in the phase PbO and bismuth oxide as Bi203 through the thickness of the brick. In addition, lead oxide was found in the form of Pb203 close to the hot face region up to 32-72 mm . In the central region and towards the cold face, bismuth was found in a phase as bismuth chromium oxide (Bi 38CrO60), which could be a hazardous compound . By balancing the valency of the cations of bismuth (Bi3+) with the valency of the anions of oxygen (O 2-), this would result in chromium being Cr6+. The chromium found in the compound Bi38CrO60 could have been from the reaction of Bi203 with the precipitated complex spinels, in which the interaction could be the main reason for not finding any original intergranular precipitated spinels usually found in magnesia-chrome bricks . 109 Figure 61. A) SEM image of MC-2 displaying severe penetration of 1) silver and, 2) bismuth-lead, and 3) a magnified image shown in Figure 61B, (40x) Figure 61 . B) Magnified SEM image of MC-2 at point 3 of Figure 61A displaying 1) MgO grain, 2) Cr2 O3 grain, 3) (Mg, Fe)(Cr, Al)2O4 spinel, 4) bismuth-lead at MgO-MgO grain boundary and, 5) bismuth-lead at MgO-Cr2 O3 grain boundary, (200x) . The corresponding EDS analysis is shown in Table 29 Table 29. EDS point analysis for MC-2 corresponding to Figure 61 Element, wt.% 0 Mg Al Cr Fe Si Ca Pb Bi Ag 1 . MgO grain 36.75 57 .40 0.00 0.28 0.52 0.19 0.00 2 .41 2 .44 0.00 2. Cr2O3 grain 28 .57 11 .53 11 .39 30 .14 11 .27 0.00 0.00 2 .67 4 .44 0.00 3. (Mg, Fe)(Cr, Al) 2 O4 spinel 30.69 11 .56 10.24 28 .02 13 .64 0 .12 0.00 2.53 3 .23 0 .00 4. MgO-MgO grain boundary 23 .52 15 .63 0.00 0.00 0.27 0.00 0.08 16 .20 44 .30 0.00 5. MgO-Cr2O 3 grain boundary 28.30 17 .79 0.21 1 .34 0.95 0.06 0.19 13 .76 37 .39 0.00 6*. Silver 8 .41 0.09 0.00 0 .00 0.00 0.00 0.00 7.19 8 .98 75 .33 * Note: EDS point analysis for silver taken at point 1 in Figure 66A 110 Figure 62 . A) SEM image of MC-8 displaying severe penetration of 1) bismuth-lead, and 2) a magnified image shown in Figure 62B, (40x) x40 0000 20kV imm Figure 62 . B) Magnified SEM image of MC-8 at point 2 of Figure 62A displaying 1) MgO grain, 2) Cr203 grain, 3) (Mg, Fe)(Cr, Al) 204 spinel, 4) bismuth-lead at MgO-MgO grain boundary and 5) bismuth-lead at MgO-Cr 203 grain boundary, (200x) . The corresponding EDS analysis is shown in Table 30 Table 30 . EDS point analysis for MC-8 corresponding to Figure 62 Element, wt .% O Mg Al Cr Fe Si Ca Pb Bi Ag 1 . MgO grain 36 .35 52.55 0.45 2 .80 2 .75 0.31 0 .09 2 .24 2 .45 0 .00 2. Cr203 grain 32 .37 11 .77 4.83 40 .45 5 .07 0.14 0 .00 2 .04 3 .33 0 .00 3. (Mg, Fe)(Cr, Al)204 spinel 4. MgO-MgO grain 27.13 10.84 4.35 36.02 14.22 0.00 0.00 2 .88 4.56 0 .00 boundary 5. MgO-Cr2O3 grain 25 .08 13 .93 1 .05 11 .57 8 .98 0.00 0 .00 21 .40 17.99 0 .00 boundary 32 .85 9.00 2.76 24.14 12 .24 0.91 6 .94 4.83 6.32 0 .00 111 Figure 63 . A) SEM image of MC-10 displaying severe penetration of 1) silver and, 2) bismuth-lead, and 3) a magnified image shown in Figure 63B, (40x) Figure 63 . B) Magnified SEM image of MC-10 at point 3 of Figure 63A displaying 1) MgO grain, 2) Cr203 grain, 3) (Mg, Fe)(Cr, A1)204 spinel, 4) silver, 5) bismuth-lead at MgO-MgO grain boundary and 6) bismuth-lead at MgO- Cr203 grain boundary, (200x) . The corresponding EDS analysis is shown in Table 31 Table 31. EDS point analysis for MC-10 corresponding to Figure 63 Element, wt .% O Mg Al Cr Fe Si Ca Pb Bi Ag 1 . MgO grain 35 .04 54.92 0.10 2 .81 1 .96 0.19 0.00 2.04 2.94 0.00 2. Cr203 grain 28 .37 10.85 5 .78 47 .02 2 .25 0 .08 0.00 2.92 2.73 0.00 3. (Mg, Fe)(Cr, Al) 204 28 .06 17.12 4.75 33 .44 11 .29 0 .10 0.00 2 .15 3 .10 0.00 spinel 4 . Silver 6.15 0.24 0.00 0.00 0 .00 0.00 0.00 6.62 10.75 76.23 5. MgO-MgO grain 12.92 0.71 0.00 0.00 0 .00 0 .00 0.00 24.42 61 .95 0.00boundary 6. MgO-Cr203 grain 10.18 0.55 0.00 0.92 0.35 0.00 0.00 24.42 63 .59 0.00boundary 112 Figure 64 . A) SEM image of unused MC-2 displaying : 1) MgO grain, 2) Cr203 grain, 3) (Mg, Fe)(Cr, A1) 204 spinel, and 4) pores, (80x) Figure 64 . B) SEM image displaying the precipitated (Mg, Fe)(Cr, Al)204 spinel in the MgO grain, (1500x). The corresponding EDS analysis is shown in Table 32 Table 32 . EDS point analysis for unused MC-2 corresponding to Figure 64 Compound, cmp.% MgO Al203 Cr203 Fe203 SiO2 CaO 1 . MgO grain 91 .55 0.00 3 .64 4.50 0.30 0.00 2. Cr203 grain 15.68 7.87 67 .77 8.58 0.00 0.10 3. (Mg, Fe)(Cr, Al)204 spinel 18 .01 10.45 48 .44 23 .03 0.00 0.07 4. MgO-Cr203 grain boundary 21 .48 8.45 34 .83 33 .99 0.59 0.66 113 Figure 65 . A) SEM image of unused MC-8 displaying: 1) MgO grain, 2) Cr203 grain, 3) (Mg, Fe)(Cr, A1)204 spinel, and 4) pores, (80x) Figure 65 . B) SEM image displaying the precipitated (Mg, Fe)(Cr, Al) 204 spinel in the MgO grain, (1500x). The corresponding EDS analysis is shown in Table 33 Table 33. EDS point analysis for unused MC-8 corresponding to Figure 65 Compound, cmp .% MgO Al203 Cr203 Fe 203 SiO2 CaO 1. MgO grain 84.27 0.26 7.22 7 .64 0.48 0.14 2. Cr203 grain 17.88 19.24 54.85 7.91 0.00 0.11 3. (Mg, Fe)(Cr, A1)204 spinel 22 .20 9 .30 49 .94 17.88 0.42 0.26 4 . MgO-Cr2 O3 grain boundary 16.23 7.34 58 .34 17.45 0.13 0.51 114 Figure 66 . A) SEM image of unused MC-10 displaying: 1) MgO grain, 2) Cr203 grain, 3) (Mg, Fe)(Cr, Al) 204 spinel, and 4) pores, (80x) x80 0011 20kV 500mm Figure 66 . B) SEM image displaying the precipitated (Mg, Fe)(Cr, A1)204 spinel in the MgO grain, (1000x). The corresponding EDS analysis is shown in Table 34 x1 .Ok 0011 20kV 50um Table 34. EDS point analysis for unused MC-10 corresponding to Figure 66 Compound, cmp.% MgO Al203 Cr203 Fe203 SiO2 CaO 1 . MgO grain 88.75 0.16 6.72 4.07 0.21 0 .09 2. Cr203 grain 17.07 7.50 67.92 7.40 0.00 0.11 3. (Mg, Fe)(Cr, A1) 204 spinel 18.54 9.94 59.89 11 .55 0.00 0 .08 4 . MgO-Cr2O3 grain boundary 27.56 11 .69 43 .79 14.69 1 .29 0.99 115 40.0 30 .0 20.0 10.0 - Mg (w t .%) (w t .%) 0 20 40 60 80 100 120 140 160 Distance from Hot Face (mm) 20 .0 --+- Cr (w t.%) --~-- Fe (w t .%) 15 .0 - 10 .0 - 5 .0 0.0 0 20 40 60 80 100 120 140 160 1 Distance from Hot Face (mm) 1 .0 - 0 .8 -4-- Si (w t .%) - Ca (w t .%) 0 .6 0 .4 ' 0.2 L L'' I\"1jihdhik 4 0 .0 AMID 0 20 40 60 80 100 120 140 160 Distance from Hot Face (mm) 40 .0 I Pb (w t .%) -,k-- Bi (w t .%) --s--Ag (w t .%) 30 .0- 20 .0 10 .0 0 .0 84ONNIPP\"OPA... \/ILA. .- AL. ALWISAII\/ft.t..- _ 0 20 40 60 80 100 120 140 160 Distance from Hot Face (mm) Figure 67. Elemental profile for the used MC-2 brick through the longitudinal cross-section 116 25 .0 --+\u2014 Cr (w t.%) \u2014u-- Fe (w t .%) 120 .0 - 15 .0 - 10 .0 - 5 .0 - 0 .0 f 0 25 50 75 100 125 150 Distance from Hot Face (mm) Figure 68 . Elemental profile for the used MC-8 brick through the longitudinal cross-section +Pb (w t.%) ----Bi (w t.%) \u2014*--Ag (wt.%) 0 25 50 75 100 125 150 Distance from Hot Face (mm) 40 .0 30 .0 20 .0 10 .0 0 .0 117 40.0 30.0 - 20 .0- 10 .0 - Al t .%t .\u00b0\/o)\u2014o--Mg (w (w CI 0 .0 0 20 40 60 80 100 120 140 160 Distance from Hot Face (mm) 20 .0 \u2014r--Cr (w t .%) \u2014o-- Fe (w t .%) 15 .0 10 .0 5 .0 0 .0 0 20 40 60 80 100 120 140 160 Distance from Hot Face (mm) 2 .0 1 .5 -41---4--Si(wt .%) 1 .0 A 0 .5 0 .0 0 20 40 60 80 100 120 140 160 , Distance from Hot Face (mm) 40 .0 --e\u2014Ag(wt .%) -4--Fb(wt.%) -- .--Bi(wt .%) ;30 .0 20.0- '~ 10 .0 - 0.0 A Ak!-A \u2022 A AA A A- 0 20 40 60 80 100 120 140 160 Distance from Hot Face (mm) Figure 69 . Elemental profile for the used MC-10 brick through the longitudinal cross-section 118 Periclase, MgO V Magnesiochromite Ferroan, (Mg,Fe)(Cr,A1)2 0 4\t0 Lead xide, Pb 20 3 \u2022 Lead Oxide, PbO 0 Bismuth Oxide, Bi2 O3 QJ Bismuth Chromium Oxide, Bi3sCrO6o \u2666 ' Distance from Hot Face, mm V \u00ae V V 22 0 44 y j 0 \u2022. \u2022.~ \u2022. QS 0 0 0 PS93 113 133 155 ~. ~._AL ~a . \u25ba. J 1 . . a- J .a\t - _ ..A~.~ -mow . _ _ 15 25 35 45 55 65 2-Theta Figure 70 . XRD profile for the used MC-2 brick through the longitudinal cross-section 119 \u2666 Periclase, MgO \u2022 Lead Oxide, PbO V Magnesiochromite Ferroan, (Mg,Fe)(Cr,AI) 2O4 0 Lead Oxide, Pb 203 0 Bismuth Oxide, Bi 203 0 Bismuth Chromium Oxide, Bi38CrOw Distance from Hot Face, mm 15 v \u00ae V V V 35 O 54 c C 72 KJ Kl XJ 1C~ 0 000 0 86 104 123 146 15 25 35 45 55 65 2-Theta Figure 71 . XRD profile for the used MC-8 brick through the longitudinal cross-section 120 Periclase, MgO \u2022 Lead Oxide, PbO V Magnesiochromite Ferroan, (Mg,Fe)(Cr,AI) 30a 0 Lead Oxide, Pb 203 Bismuth Oxide, Bi 2 03 0 Bismuth Chromium Oxide, Bi 35CrOr60 Distance From Hot Face, mm \u2022 v 15 \u2022 32 \u00ae .O v 1\u2022 V ~Jy .r._. h r M.nM.4.. .~... 51 0 0 fd 0 00 fd >' 65 VJ ~ d c 12 155 1 r 15 25 35 45 55 65 2-Theta Figure 72 . )(RD profile for the used MC-10 brick through the longitudinal cross-section 121 Based on the results, two corrosion mechanisms were hypothesized . The first mechanism, which may be considered in a lesser degree, was due to thermal mechanical spalling . As seen in the cross-section images, large macro-cracks up to one inch in thickness were observed parallel to the hot face . Although the period, at which the formation of these cracks was not known, it could be assumed that they were formed during thermal mechanical stresses. The second mechanism, which may be considered in a higher degree, was due to pure dissolution with the contribution of erosion . The highly aggressive molten liquid of lead and bismuth oxide dissolved or damaged the precipitated complex spinel bonds, which created a weakened structure that could easily be \"washed\" or eroded at the hot face by abrasion . As seen in the image of the brick from the one regular campaign, the surface at the hot face showed a rough texture exposing the coarse grains, which indicated heavy dissolution of the matrix phase exposing the coarse grains for erosion . From the XRD results, the chromium found in the compound Bi 38CrO60 could have been from the reaction of Bi2O 3 with the precipitated complex spinels and could be the main reason for the difficulty in identifying the original precipitated complex spinel . In addition, lead and bismuth oxides may react with SiO2 in the matrix of the brick forming low melting liquids, which may further dissolve MgO and destroy the grain boundaries . 122 6.0 Summary The investigation of the corrosion performance of several magnesia-chromia (MC) and magnesia-alumina spinel (MA) bricks by the static and dynamic corrosion methods (in contact with KIVCET slag) determined that the MC bricks were superior to the MA bricks. Although the static method did not provide enough information on the severity of the corrosion at the interface due to the absence of dynamic factors, SEM and EDS analyses of the refractory-slag interfaces of both the static and dynamic samples presented hypothesized corrosion mechanisms . The slag was identified as a ceramic-glass composite composed of two phases: a glassy phase containing SiO2, CaO, Fe2O3 and ZnO, and a zinc ferrite (ZnFe2O4) crystalline phase. Both of these phases penetrated and interacted with the refractory, leading to chemical dissolution . From the static corrosion method, Fe2 O3 and ZnO phases from the slag reacted with the MC and MA brick forming complex spinels of (Mg, Fe, Zn)(Fe, Al, Cr) 2O4 and (Mg, Fe, Zn)(Fe, Al) 2 O4, respectively. The reaction product layer at the interface provided a barrier for the refractory from further interaction and penetration of the slag. From the dynamic corrosion method, it was determined that SiO 2 , CaO, Fe2O 3 and ZnO from the slag reacted with the MgO grains in the MC bricks forming a modified forsterite phase of (Mg, Fe, Zn, Ca) 2SiO4. The dissolution of the MgO grains destroyed the grain boundaries between adjacent grains, which caused the coarse grains (including the unaltered Cr2 O3 grains) to be eroded away at the hot face. In addition, both the MgO grains and the MgAl 2 O4 spinel of the MA brick (from the dynamic test) were susceptible to chemical dissolution, thus forming (Mg, Fe, Zn, Ca) 2SiO4 and (Mg, Fe, 123 Zn)(Fe, A1)2 O4, respectively. The interaction with the slag caused a weakened structure, which subjected the grains to be easily eroded away. In the Solubility evaluation of Cr 2O3 in KIVCET slag with various amounts of lead, it was found that the spinel formation in the barren slag was most affected by the increase in Cr 2O3 content and the low-lead slag was least affected after firing at 1250\u00b0C in argon . It is important to mention that these spinel phases already exist in equilibrium with the slag at 1250\u00b0C and the Cr2O3 would not initiate the accretion formation but would only increase their amount . The tendency would rather be considered dependent on the Pb content in the slag and the pot in the furnace precipitating the phases already existing in the slag . There was a high tendency to form spinel or \"accretions\" even without Cr2O3 in the barren slag. The amount of spinel solid solution increased consistently after firing with increased Cr 2 O3 and Pb content, for all experiments . The net contribution of Cr2O3 to the spinel formation was the highest for the barren slag mixes, followed by high-lead mix, and it returned the lowest results for the low-lead mixes. The leaching test for hexavalent chromium from the interaction of KIVCET slag and magnesia-chrome bricks found that the highest amount of dissolved Cr 6+ was 0.043mg\/L. The amount found was below the allowed limit of 0 .05 mgCr6+\/L in drinking water. Therefore, the risk of Cr+6 formation is practically nonexistent. In the Peirce Smith Nickel Converter, matte high in nickel sulfide (NiS) penetrated deep into the refractory, in which it could have been partially oxidized to nickel oxide (NiO) by the 124 oxygen originating from the cold end of the brick . The oxidized components of the matte (Cu2O, NiO and CoO) could form spinels with magnesia and generate cracks at the grain boundaries . These cracks could propagate parallel to the hot face during thermal cycles contributing to the global failure mechanism. The interface of the magnesia grain showed partially oxidized matte rims, which could explain the presence of Co, Cu and Ni as spinels precipitated inside the periclase grains. In the KIVCET Electric Furnace roof, there was a deep reaching sulphate, which mostly affected the bricks bonding phase between coarse magnesia grains . In the furnace atmosphere, a gaseous sulphate and potassium phase reacted with the calcium silicates in the brick's matrix, which resulted in potentially forming a calcium sulphate, potassium sulphate or a calcium potassium sulphate phase . This attack caused a weakened structure and was most evident in the porous zone . In addition, a lead flooded microstructure was observed, which caused the connecting network of grains to de-bond and resulted in a weakened structure. In the Bottom Blown Oxygen Converter, a highly aggressive liquid of lead oxide (PbO and Pb203) and bismuth oxide (Bi203) dissolved or damaged the precipitated complex spinel ((Mg, Fe)(Cr, A1)203) bonds, which created a weakened structure that was easily eroded at the hot face by molten liquids . The chromium found in the compound Bi38CrO60 by XRD could have been from the reaction of Bi2O 3 with the precipitated complex spinels destroying the bonds causing this phenomenon . 125 7.0 Conclusions From the present research, the following can be concluded: I. KIVCET Electric Furnace (Teck Cominco, Trail, BC) \u2022 Static and dynamic corrosion experiments proved that magnesia-chrome bricks are better than magnesia-alumina spinet bricks against KIVCET fayalite type slags. \u2022 In the magnesia-chrome group, the brick with concentrations lower in iron, higher in chromia, and of a rebonded fused grain rather than direct bonded, perform better and have a longer life in service. \u2022 The wear mechanism is mainly due to microstructural changes due to the reactions with CO, SO 2 and metal\/metal oxide fumes (Pb\/PbO, Cd\/CdO, Zn\/ZnO) rather than due to reactions with the slag . The thermal-mechanical stresses associated with the brick impregnation with PbO and CdO, and the redox reactions between high-in-iron phases and CO, also play an important role, as the thermal spalling can become the main cause of wear in a strongly reducing atmosphere or in an SO2-rich environment. \u2022 The Cr2 O3 in magnesia-chrome bricks recycled in the KIVCET Flash furnace is not the main reason for the accretion formation, as our experiments proved that a high lead slag would have a higher tendency to form accretions than a low lead slag, usually consistent during the process under normal operating conditions . The lead content in the slag and the partial pressure of oxygen play a more important role in accretion formation than the presence of Cr 2O3 . The increase of lead in the slag, 126 above what is normally expected according to the process, would induce the tendency of magnetite and zinc ferrite precipitations causing the accretions . The low lead slag, normally existent during the process, works against precipitations (if the atmosphere remains at the right partial pressure of CO\/0 2) and will cause no accretions. \u2022 The risk of Cr+6 formation is practically nonexistent against KIVCET slag II. Bottom Blown Oxygen Converter (Teck Cominco, Trail, BC) \u2022 The mechanism of wear in the BBOC is mainly due to the mineralogical changes at the periclase-periclase and periclase-spinel grain boundaries as a result of the reactions with the litharge type slag. The secondary silicate phases would help the process and as a result, the lowest in SiO2 and CaO bricks, in particular of the rebonded fused grain, would perform the best and have the longest life in service. III. Peirce Smith Nickel Converter (Falconbridge, Sudbury, ON) \u2022 Our investigations on brick samples from the tuyere area demonstrated that thermal- mechanical spalling of the impregnated portions of the brick lining, rather than chemical corrosion, is the main mechanism of wear in the PS converter . The oxidized components of the matte, Cu 2O, NiO and CoO could form spinels with magnesia, which are able to generate cracks at the grain boundaries, and could propagate parallel with the hot face, during the thermal cycles between blows . 127 8.0 Future Work In this study of several magnesia-chrome and magnesia-alumina spinel bricks through laboratory and postmortem analyses, further investigations are recommended. 1. While it was demonstrated that the interactions with the penetrated matte and refractory can affect the microstructure and contribute to the overall corrosion of the tuyere brick, further investigation is needed in order to determine the amount of expansion due to thermal, joint infiltration and solidification of matte and slag and its contribution to the microstructural changes caused by Ni t+ , Cue+ , Coe+ and Fee+ ions diffused into the periclase. 2. Although sulphates from the KIVCET furnace atmosphere attacked the matrix of the roof brick, there was not enough evidence to prove that the porous zone was created solely by the sulphate attack . With 1-2 wt.% of silicate phases in the brick, which mainly resided in \"pockets\" between grains, this amount could not have resulted in such a high porous zone . The hypothesized cause of failure was due to the attack of a strongly reducing CO atmosphere and the redox reactions between high-in-iron phases . Further investigation of this hypothesis is needed. 3. In order to confirm that the lead oxide and bismuth oxide chemically dissolved the precipitated complex spinels, which was the hypothesized mechanism for the dissolution of the bricks from the Bottom Blown Oxygen Converter, further corrosion studies of magnesia-chrome bricks with slags high in lead oxide and bismuth oxide are required . 128 9.0 References 1. Brothers, M., \"Bottom Blown Oxygen Converter .\" Tech Cominco Metals Ltd . Trail, British Columbia (UBC Private Communications). 2. Alper, A.M., \"Oxide Spinels.\" High Temperature Oxides Part IV . New York: Academic Press, 1971. 3. Callister, W.D., \"Structures and Properties of Ceramics.\" Materials Science and Engineering: An Introduction. New York: John Wiley & Sons Inc ., 1997. 4. Levin, E.M., \"Phase Diagrams .\" Phase Diagrams for Ceramists . Columbus : American Ceramics Society, 1964. 5. Donald, J .R., J .M. Toguri and C . Doyle, \"Surface Interactions between Fayalite Slags and Synthetic Spinels and Solid Solutions .\" Metallurgical and Materials Transactions. 29B (1998) : 317-323. 6. Ikesue, A., J. Yoshitomi, H. Shikano and T . Eguchi, \"Formation of Precipitated- Complex Spinel in Magnesia-Chrome Refractories and its Characteristics .\" Interceram. 41 (1992) : 406-412. 7. Alper, A .M., \"Chromite Spinels .\" High Temperature Oxides Part I . New York: Academic Press, 1970. 8. Cunha-Duncan F .N., and R.C. Bradt, \"Synthesis of Magnesium Aluminate Spinel from Bauxites and Magnesias.\" Journal of the American Ceramic Society . 85 (2002): 2295-3003. 9. Ghosh, A., R. Sarkar, B . Mukherjee and S.K. Das, \"Effect of Spinel Content on the Properties of Magnesia-Spinel Composite Refractory.\" Journal of the European Ceramic Society . 24 (2004) : 2079-2085 . 129 10.Carniglia, S.C., and G. Barra, \"Foundation of Refractory Applications .\" Handbook of Industrial Refractories Technology . New Jersey: Noyes Publications, 1991. 11.Goto, K. and W.E . Lee, \"The Direct Bond in Magnesia Chromite and Magnesia Spinel Refractories .\" Journal of the American Ceramics Society. 78 (1995): 1753- 1760. 12. Cherif, K., V. Pandolfelli and M. Rigaud, \"Factors Affecting the Corrosion by Fayalite Slags and the Thermal Shock Performance of Magnesia-Chrome Bricks .\" Journal of the Canadian Ceramic Society. 66 (1997) : 210-216. 13.Takahashi, H., T. Kawakami and Y . Oguchi, I . Tsuchiya, \"High Corrosion Resistant Magnesite-Chrome Refractories for RH Degassing Lower Vessel .\" Taikabutsu Overseas . 11 (1991) : 37-41. 14. Ichikawa, K., and K. Minato, \"Corrosion Resistance of Magnesia-Chrome Bricks for Low Basicity Slag.\" Refractories (Tokyo) . 41 (1989) : 26-28. 15. Wiederhorn, S .M., R.F. Krause and J . Sun, \"Effect of Coal Slag on the Microstructure and Creep behaviour of a Magnesium-Chromite Refractory .\" Bulletin of American Ceramic Society. 67 (1988) : 1201-1209. 16. Rose, S., and T.D. McGee, \"Corrosion of MgO Single Crystal by BOF Slags .\" Bulletin of American Ceramic Society. 57 (1978) : 674-679. 17.Nikoo, A.S ., and F. Golestani-Fard, \"The Effect of Slag Basicity on Corrosion of Refractories in Copper Converter .\" Proceeding of Unified International Technical Conference on Refractories . 1 (2001) : 514-523. 18. Petkov, V., P.T. Jones, E. Boydens, B . Blanpain and P . Wollants, \"Chemical Corrosion Mechanism of Magnesia-Chromite and Chrome-Free Refractory Bricks by Copper Metal and Anode Slag.\" Journal of the European Ceramic Society . 7 (2007): 2433-2444 . 130 19. Evrard, L., A. Vanderlinden and C . Van Riet ., \"Mineralogy: A Tool for Processes Development in the Non-Ferrous Extractive Metallurgy .\" Proceeding of Unified International Technical Conference on Refractories . 1 (1989) : 895-905. 20. Schlesinger, M .E., M . Karakus, M. Crites, \"Chrome-Free Refractories for Copper Production Furnaces .\" Proceeding of Unified International Technical Conference on Refractories . 1 (1997) : 1703-1709. 21. Goto, K., B .B . Argent and W .E. Lee., \"Corrosion of MgO-MgAl2 O4 Spinel Refractory Bricks by Calcium Aluminosilicate Slag .\" Journal of the American Ceramic Society . 80 (1997) : 461-471. 22. Pressley, H., J. White, \"Chemical Requirements of Refractories for Non-Ferrous Extraction Processes .\" Interceram. 30 (1981) : 280-287. 23. Makipaa, M., and P. Taskinen, \"Refractory Wear in Copper Converters : Part 1. Blister Copper-Refractory Interactions .\" Scandinavian Journal of Metallurgy . 9 (1980): 273-281. 24. Barthel, H., \"Wear of Chrome Magnesite Bricks in Copper Smelting Furnaces .\" Interceram. 30 (1981) : 250-258. 25. Utigard, T ., A. Warczok, G . Plascencia, N. Behnood and M. Zamalloa, \"Interaction Between Copper Mattes and Refractories .\" Tehran International Conference on Refractories . 1 (2004) : 321-327. 26. Goncalves, G .E., and J .M . Nunes, \"Technical Considerations about basic refractories for Tuyeres Copper Converters .\" Proceeding of Unified International Technical Conference on Refractories . 2 (2001) : 657-668. 27. Oprea, G., \"Failure Mechanism of Refractory Lining for Non-Ferrous Flash Smelting Furnaces .\" Tehran International Conference on Refractories . 1 (2004) : 290-301 . 131 28. ASTM C874-85, \"Standard Practice for Rotary Slag Testing of Refractory Materials.\" Annual Book of ASTM Standards . 15.01 (1999). 29. \"Toxicity Characteristic Leaching Procedure 1311 .\" U.S. Environmental Protection Agency . July 1992. 31 Oct . 2007 . 30. ASTM C133-97, \"Cold Crushing Strength and Modulus of Rupture of Refractories .\" Annual Book of ASTM Standards . 15 .01 (1999). 31. ASTM C20-97, \"Apparent Porosity, Water Absorption, Apparent Specific Gravity and Bulk Density of Burned Refractory Bricks and Shapes by Boiling Water .\" Annual Book of ASTM Standards . 15 .01 (1999). 32. \"Chromium in Drinking Water.\" World Health Organization. 2003 . 31 Oct. 2007