Corrosion of Refractories in Lead Smelting Reactors
CORROSION OF REFRACTORIES
By
LINGXUAN WEI
B.Sc, Wuhan University of Science & Technology, China 1986 M.Sc, University of Science & Technology Beijing, China 1997
A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF APPLIED SCIENCE
in
THE FACULTY OF GRADUATE STUDIES DEPARTMENT OF METALS AND MATERIALS ENGINEERING
We accept this thesis as conforming to the required standard
THE UNIVERSITY OF BRITISH .UMB1A
DECEMBER 2000
©Lingxuan Wei, ZO0O UBC Special Collections - Thesis Authorisation Form http://www.library.ubc.ca/spcoll/thesauth.html
In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission.
v 3 The University of British Columbia Vancouver, Canada
Date
lof 1 3/19/01 2:36 PM ABSTRACT
Corrosion of refractories by slag is a complex phenomenon which, depending on the particular system, involves many processes, such as chemical wear (corrosion) and physical or mechanical wear (erosion), which may act synergistically. No single model can explain all cases of corrosion nor can it explain all corrosion mechanisms of a particular refractory in different environments, but the knowledge of the microstructure combined with the chemistry of the systems are necessary to understand the corrosion mechanism of a refractory material.
There is no systematic study on the corrosion of refractories used in lead smelting reactors (KIVCET furnace and TBRC) in literature so far. In the present work, the available literature concerning corrosion of refractories in lead-smelting reactors was reviewed. By using different analytical methods such as sessile drop technique,
SEM/EDS and XRD, the interfacial phenomena at the slag-refractory interfaces were investigated, and microstructural studies on different refractory specimens were carried out.
It was observed that above 1150° C, the KIVCET slag tends to separate into two phases: a liquid and a solid. The liquid mainly consists of S1O2, CaO and PbO, while the solid primarily contained Fe2C«3 and ZnO in the form of zinc ferrite (ZnFe204) spinel. It was
proposed that ZnO from the KIVCET slag could also react with Cr203 and Fe203 from the magnesite-chrome brick and reacted slag to form spinel-type phases: zinc chromite
ii (ZnCr204) and zinc ferrite (ZnFe2C>4) respectively. The volume changes accompanying these reactions could lead to microcracks in the matrix of the brick, and eventually cause the failure of the brick.
Laboratory evaluation of the corrosion behavior of various refractory materials against industrial slags from KIVCET and TBRC furnaces was performed using both dynamic and static corrosion tests. The corrosion rating on nine different magnesite-chrome bricks was estimated, and the possible corrosion mechanism was discussed. It was found that the rebonded fused grain type of magnesite-chrome bricks have superior performance compared with direct bonded type when used in contact with KIVCET slag and the alumina-chromia bricks performed better than magnesite-chrome bricks under similar conditions.
iii Table of Content
Abstract »
List of Tables ; vii
List of Figures viii
List of Abbreviations and Symbols xi
Acknowledgements xiii
1 Introduction , 1
2 Literature Review 6
2.1 Refractory Materials Used in Lead-Smelting Reactors 6
2.1.1 Magnesite-Chrome and Chrome-Maghesite Bricks 6
2.1.2 Alumina-Chromia Bricks 7
2.2 Corrosion of Refractories by Molten Slags 7
2.2.1 Theoretical Aspects Regarding Corrosion of Refractories 10
2.2.2 Corrosion of Refractories in Contact With Various Slags.. 16
2.2.2.1 MagnesiteRefractories 16
2.2.2.2 Alumina-Chromia Refractories 18
2.2.2.3 Magnesite-Chrome Refractories 20
2.3 Wetting Behaviors of Refractories 22
2.3.1 Wettability 22
2.3.2 Wetting Behaviors of Refractories at High-Temperature 24
iv 2.4 Mechanisms of Refractory Corrosion in Nonferrous Smelting Environments 26
3. Scope and Objectives of this Work 29
4. Experimental Procedures 30
4.1 Contact Angl e Measurements 30
4.2 Corrosion Tests 31
4.2.1 Dynamic Corrosion 31
4.2.2 Static Corrosion 33
4.3 Sample Preparation 35
4.3.1 Samples for Contact Angle Measurements. 35
4.3.2 Samples for Corrosion Tests 37
4.4 Slag Compositions 39
4.5 Testing Methods 40
5 Results and Discussion 41
5.1 Characterization of KIVCET Slag 41
5.2 Interaction Between KIVCET Slag and Various Substrates 53
5.3 Postmortem Analysis of Used MC Bricks from the KIVCET Furnace 58
5.4 Results of Corrosion Tests 67
5.4.1 Magnesite-Chrome (MC) Bricks with KIVCET Slag 67
5.4.2 Alumina-Chromia (AC) Bricks with KIVCET Slag 79
5.4.3 Magnesite-Chrome Bricks with TBRC Slag 82
5.5 Summary 86
6 Conclusions 89
V 7 Future Work 91
8 References 92
vi LIST OF TABLES
Table 1. Substrates used in the contact angle measurements 35
Table 2. Characteristics of the magnesite-chrome (MC) and alumina-chromia (AC)
brick samples 38
Table 3. Chemical composition of slags 39
Table 4. Results of EDS analyses of KS slag in Figure 19 48
Table 5. Elemental compositions of KS slag in Figure 21a and 21b 51
Table 6. Calculated compositions of KS slag based on EDS analyses in Table 5 51
Table 7. Slag compositions on the used MC brick from KIVCET furnace by
EDS analyses 59
Table 8. Results of corrosion tests of the magnesite-chrome bricks with KS slag 68
Table 9. EDS analyses of MCI brick after RSF test with KS slag 79
Table 10. Results of corrosion tests of alumina-chromia (AC) bricks with KS slag 80
Table 11. Results of corrosion tests of MC and AC bricks with TBRC slag 83
Table 12. Results of EDS analyses of MC4 brick after RSF test with TBRC slag ...85 LIST OF FIGURES
Figure 1. Schematic diagram of typical refractory microstructure 2
Figure 2. KIVCET flash smelting furnace arrangement 5
Figure 3. Top blown rotary converter (TBRC) arrangement 5
Figure 4. Phase diagram of the MgO-Cr203 system 8
Figure 5. Phase diagram of the MgO-Al203-Cr203 system 8
Figure 6. Phase diagram of the MgO-Al203-Si02 system 9
Figure 7. Phase diagram of the Al203-Cr203 system 9
Figure 8. Equilibrium forces at the solid-liquid-vapor interface for an
(a) acute contact angle and (b) obtuse contact angle 23
Figure 9. Schematic diagram of a sessile drop on a surface 23
Figure 10. Schematic diagram of contact angle measurements
at high temperatures 30
Figure 11. Schematic Diagram of the Rotary Slag Furnace (RSF) 32
Figure 12. Schematic diagram of the slag cup test (SCT) 34
Figure 13. Flow chart of the refactory substrate preparation for contac angle
measurements 36
Figure 14. Wetting behavior of KS slag on pure A1203 and MC2 substrates
at different temperarues 42
Figure 15. Wetting behavior of KS slag on pure A1203 and MC2 substrates
at different soaking times 43
Figure 16. Contact angle versus temperature for KS slag 44
viii Figure 17. Contact angle versus soaking time at 1250° C for KS slag 44
Figure 18. XRD patterns of residual slag on MC2 substrate (top) and original
slag (bottom) 46
Figure 19. SEM images of residual slag on MC2 substrate 47
Figure 20. XRD paterns of the KS slag after 3h at 1250° C in zero porosity alumina
crucible 49
Figure 21a. SEM image of KS slag at the top of the alumina crucible,
3hat 1250°C 50
Figure 21b. SEM images of KS slag at the bottom of the alumina crucible,
3h at 1250° C 50
Figure 22. Schematic phase separation of the KS slag on porous refarctory
substrates 54
Figure 23 a. XRD patterns of KS slag impregnated areas of AC4 substrate
and its original 54
Figure 23b. XRD patterns of KS slag impregnated areas of MC2 substrate
and its original 55
Figure 24. Phase diagram of the AbCVSiCVPbO system 57
Figure 25. Phase diagram of a) CaO-Cr203 system b) PbO-Cr203 system 57
Figure 26. Analyzed areas of the used MC brick from KIVCET furnace 59
Figure 27a. SEM image of the used MC brick from KIVCET furnace (interface).... 60
Figure 27b. SEM image of the used MC brick from KIVCET furnace (brick) 60
Figure 28. XRD patterns of the used MC brick from KIVCET furnace 62 Figure 29. Elemental distribution versus distance from the interface in the
used MC brick from KIVCET furnace 63
Figure 30. Ternary phase diagrams of a) ZnO-Cr2C>3-Yb203 and
b) ZnO-Fe203-Yb203 66
Figure 31a. Corroded MC specimens after SCT with KS slag, 24 hours at 1350°C. 69
Figure 31b. Corroded AC specimens after SCT with KS slag, 24 hours at 1350°C .. 70
Figure 32a. Corroded MC specimens after RSF 5 hour tests with KS slag
at 1350°C 71
Figure 32b. Corroded MC specimens after RSF 5 hour tests with TBRC slag
at 1075°C 72
Figure 33. Corroded specimens after SCT with KIVCET slag, 24h at 1350° C 76
Figure 34a. SEM image on MCI brick after RSF test with KS slag 77
Figure 34b. SEM image on MCI brick after RSF test with KS slag (Field 01) 77
Figure 34c. SEM image on MCI brick after RSF test with KS slag (Field 02) 78
Figure 34d. SEM image on MCI brick after RSF test with KS slag (Field 03) 78
Figure 35. Images of AC and MC bricks after SCT test 24hr at 1350° C
with KIVCET slag 81
Figure 36a. SEM image of MC4 brick with TBRC slag (interface) 84
Figure 36b. SEM image of MC4 brick with TBRC slag (35mm from interface) 84
x LIST OF ABBREVIATIONS OF AND SYMBOLS
Abbreviations
AC Alumina Chromia brick
ASTM American Society for Testing and Materials
CCS Cold Crushing Strength
DB Direct Bonded
EDS Energy Dispersive Spectroscopy
HMOR Hot Modulus of Rupture
HTM High Temperature Microscope
KIVCET KIVCET flash smelting furnace
KS slag The slag generated from KIVCET furnace
TS slag The slag generated from TBRC furnace
MC Magnesite Chrome brick
MOR Modulus of Rupture
RFG Rebonded Fused Grain
RSF Rotary Slag Furnace
SCT Static Cup Test
SEM Scanning Electron Microscopy
TBRC Top Blown Rotary Converter
XRD X-Ray Diffraction Symbols
6 contact angle or wetting angle o/v liquid surface tension
Yiv liquid vapor interfacial energy
Ap pressure difference
ysi solid liquid interfacial energy
ysv solid vapor interfacial energy d capillary diameter
D density rj dynamic viscosity h liquid penetration depth t time
xii ACKNOWLEDGEMENTS
I would like to express my sincere gratitude to my supervisors Dr. Tom Troczynski and
Dr. George Oprea for giving me the opportunity to work with them and for invaluable guidance and support throughout the courses of my graduate study and this research work.
I would also like to thank Ms Mary Mager for her assistance with SEM/EDS and XRD analyses. The assistance of staff and students of UBCerafn group in this department:
Carmen Oprea, Dr. Ahmad Monshi, Dr. Quangzu Yang, Peter Musil, Mehrdad Keshmiri are highly appreciated. Besides the aforementioned individuals, I'm also grateful to Joan
Kitchen for her mother-like attitude towards me.
I would like to thank my wife and parents for being always there for me with their support all throughout graduation years.
Last but not least, I would like to express my gratitude to Cominco Ltd.-Trail
Operations, Clayburn Industries Ltd. and NSERC for providing the financial support and making possible my research program.
xiii 1 Introduction
Refractories are nonmetallic materials that maintain their physical and chemical identity to be used for structural purposes in the high-temperature environments encountered in the industrial processes. A large number of industries, metallurgical and non-metallurgical (iron and steel, non-ferrous, cement, glass, petrochemical, ceramic, power generation, etc ) use refractories for various purposes.
Generally, refractories are made as prefired shapes (bricks) or as unfired monolithic materials. The latter are installed by ramming, casting or gunning followed by in situ drying and firing. The resultant microstructure typically consists of large (up to several mm) aggregate grains, and fillers held together by a fine, porous bond or matrix (Figure 1). The fine texture of the bond makes it more active than the grain.
During service, refractories must not only tolerate high temperatures, but also withstand mechanical stresses, as well as resist corrosion and erosion attack by corrosive environments such as molten metals, slags, fluxes, fumes or vapors and hot gases.
In the field of non-ferrous metallurgy very little work has been done towards the
development of refractories, compared to ferrous metallurgy. The literature sources
on refractories for non-ferrous metallurgy are scarce and for lead in particular,
practically nonexistent. The development of refractories in this field is difficult,
partially due to the complexity of the systems and the lack of phase equilibria for
1 systems involving non-ferrous oxides.
Glass or fine grained Porosity matrix
Figure 1. Schematic diagram of typical refractory microstructure
In recent years we observe an increased use of basic refractories in non-ferrous metallurgy [1,2]. However, most of the refractories are still selected on a trial and error basis. The selecting factors are usually the resistance to slag attack, cost and availability.
No apparent attempt has been made to develop new refractories for specialized applications in non-ferrous metallurgical furnaces. This is true for the copper, nickel,
2 lead, zinc and other non-ferrous metallurgical industries. It could be said that the major developments and improvements to be made in the field of refractories lie in the particular non-ferrous metallurgical application.
There are many types of smelting furnaces in non-ferrous industries [3]. The direct lead smelting processes differ considerably from the conventional blast furnace process for lead concentrates. Direct smelting processes consist of two stages:
• First, oxidation of lead concentrate:
PbS + 3/2C-2 = PbO + S02
• Then, reduction of PbO:
PbO + C = Pb + CO
2PbO +C =2Pb +C02
PbO + CO = Pb + C02
The process being continuous, the two reactions take place, either in different ends of the reactor (i.e. in KIVCET flash smelting furnace) or in the same vessel at different time ( in top blown rotary converter). The schematic arrangements of KIVCET smelting furnace (KIVCET) and top blown rotary converter (TBRC) are presented respectively in Figure 2 and 3 [4,5].
The KIVCET flash smelting process (KIVCET) was developed especially for
complex ores with high zinc content [6]. The smelting and reduction reactions take
place in two distinct stages within the same rectangular reactor in which the initial
3 "flash" smelting takes place in an oxygen-fed flame, burning in a reaction shaft above the molten bath of slag and lead bullion. The reduction stage of PbO occurs in the layer of coke floating on the molten slag. Lead bullion and slag (containing zinc and other valuable metals) are tapped from the electric furnace.
The Top Blown Rotary Converter (TBRC) is used either for smelting or cupellation.
In the converter, the operation is a batch type of process composed of feed charging and melting, the oxidation and reduction reactions following one after the other in the same reactor. This means that the composition of the slag varies during the process from a PbO rich slag to a slag containing only a small percentage of PbO.
Because the amount of lead oxide in slags of direct processes is high (e.g. 20-60 wt%
PbO) compared to the slags of classic methods (only 1-2 wt% PbO) the problems with the corrosion of refractories caused by the aggressive lead oxide are totally new
[3,4].
The present research work was focused on corrosion of refractories used in non- ferrous metallurgical furnaces, especially lead and zinc smelting reactors. Corrosion experiments were carried out as part of a complex study of the performance of commercially available refractory materials in contact with various liquid phases generated in pyrometallurgical processes (using the KIVCET flash smelting furnace and the top blown rotary converter) at Cominco Trail Operation, British Columbia.
4 Figure 2. KIVCET flash smelting furnace arrangement [4]
Figure 3. Top blown rotary converter (TBRC) arrangement [5]
5 2 Literature Review
2.1 Refractory Materials Used in Lead-Smelting Reactors
There are no systematic studies in the literature on the wear of refractories in lead- smelting furnaces. However, magnesite-chrome (MgO-C^Ch), chrome-magnesite
(Cr203-MgO) and alumina-chrome (Al203-Cr203) refractories have been widely used in lead smelting reactors [1,2], and will be discussed in more details below.
2.1.1 Magnesite-Chrome and Chrome-Magnesite Bricks
Magnesite and chrome ore are used as raw materials for making these bricks. When
MgO content is more than 50 %wt, the brick is called magnesite-chrome, otherwise
chrome-magnesite brick. The phase diagram in the MgO-Cr203 system (Figure 4) [7]
shows that picrochromite MgCr204 is the only compound in the system. Although it is rarely used as refractory on its own, it constitutes a major part of the complex solid
solutions of chromites. The ternary phase diagrams of the MgO-Al203-Cr203 (Figure
5) and MgO-Al203-Si02 (Figure 6) systems [7] are relevant as far as these refractories are concerned. As seen in Figure 5, the addition of MgO to AI2O3-O2O3 mixtures results in an increase in liquidus temperature, with the exception of an area adjacent
to the spinel MgO-Cr203, where liquidus temperatures are about 100° C above those
in adjacent areas. In the MgO-Al203-Si02 system (Figure 6), there are many low temperature eutectics, in particular, in the region rich in silica, which usually
6 decreases the maximum service temperature of the refractory. As a result, the SiC>2 content in magnesite-chrome and magnesite-alumina refractories should be a minimum in order to obtain a long service life.
2.1.2 Alumina-Chromia Bricks
This brick contains mainly alumina (AI2O3) and chromia (C^Cb). The phase diagram of the system Al203-Cr203 (Figure 7) shows that the two sequioxides form a complete series of solid-solutions, suggesting that any composition in this system can be a
refractory material. But usually about 10% Cr203 is used in the matrix of the brick as
a solid solution with tabular alumina. For some special bricks, the Cr203 content can reach as high as 80%, e.g. in coal gasifier [8].
2.2 Corrosion of Refractories by Molten Slags
Molten slag attack on refractories is a complex phenomenon involving not only chemical wear (corrosion) but also physical/mechanical wear (erosion), processes which may act synergistically. No single model can explain all cases of corrosion, nor can it explain all corrosion mechanisms of a particular refractory in different environments. Knowledge of the corrosion theories in combination with the
microstructure and chemistry data of the system is necessary to identify the
mechanisms of corrosion.
7 "i 1 i 1 1 r '2800
liquid
C/ - -
• TA/? liquid i I^•215 32^0 _ \* Pi picrochromrtc—-"j j
pcriclasc 11 i picrochromitc 2 S
J L
MgO 20 30 40 50 GO^gO-Cr^ Cr203
Cr203, wt-*/.
Figure 4. Phase diagram of the MgO-Cr203 system [7]
CrjO-.
MqOC'jO
. I'M
' I ' f ,l\
MqC ai;03
Figure 5. Phase diagram of the MgO-Al203-Cr203 system [7] =—ut/ *l0 41,0, »l,Ol 311,0, 2S-0, 2M)0 241,0, 53;0, •UgO 5Ju,0, JS.O, Figure 6. Phase diagram of the MgO-Al203-Si02 system \1] 24O0f- .2265-Q sasqui-oxide • liquid sesqui-oxide leoof- 1600t 1 80 Cr_03 M_0_ 40 60 20 Cr_03 , wt-*<. Figure 7. Phase diagram of the Al203-Cr203 system [7] 2.2.1 Theoretical Aspects of Corrosion of Refractories As in any chemical reaction between a solid and a liquid, the corrosion of refractories by molten liquids involves the "contact", which enables the reaction to take place and the "mass transfer" to allow it to proceed. For both the contact and the mass transfer, the composition, ceramic texture, the nature of the bonding phase of the refractories, and the characteristics of the melt and the products, are important. Capillaries, such as open pores and microcracks in the ceramic texture, are the channels for the initial penetration of the molten phase into a refractory material. The penetration rate dh/dt of a liquid into a capillary can be expressed by Poiseuille's law [9]: dh/dt = d2Ap/(327]h) 0) where d is the capillary diameter, Ap the capillary pressure, 77 the dynamic viscosity of the liquid, h the liquid penetration depth and t is the time. The pressure Ap can be expressed versus the liquid surface tension o_i/and the contact angle 9, as follows: Ap = 4- OLV • cosOld (2) 10 After eliminating Ap from the equations (1) and (2) and integrating, it results: h2 = d- OLV COSOI4t] (3) This equation does not include the temperature as a direct variable, although the surface tension, the viscosity of the melt and the contact angle of the refractory-melt decrease with increasing temperature. For isothermal conditions, when r\, oiv and cosG remain constant (k=OLvcosO/4rj), the depth of penetration depends only on the pore diameter d of the refractory material. 2 h = kd (4) When there is a temperature gradient on the direction of penetration, the increase of the melt viscosity and surface tension will certainly slow down the penetration. The corrosion rate is a function of many variables including temperature, refractory/liquid/interface composition and liquid density, viscosity, diffusivity, and degree of agitation [10, 11]. If the reaction product is soluble or dissociates directly in the liquid slag, then the active corrosion may continue to destroy the refractory. However, if one of the products of reaction is a solid phase, which forms as a continuous adherent layer on the original refractory surface, then this layer may act to 11 reduce the overall corrosion of the underlying refractory (i.e. passive corrosion). In this case, the possible steps in determining the corrosion reaction rate include: the chemical reactions forming the layer of product of the reaction, diffusion through the layer or diffusion through the slag. Selective corrosion of refractories may also occur, in which case only certain phases in the solid are attacked. A good example of this is the decarburization of carbon containing refractories, which occurs by several mechanisms including dissolution of carbon into the molten steel [10]. Once the carbon has been removed the refractory can be wetted by the slag so that penetration and spalling of the decarburised layer can occur. Dissolution at refractory /slag interface is governed by (a) chemical reaction (or solution) at the interface, or (b) transport (or diffusion) of reacting species through liquid [11]. The rate-determining step is used to define the types of dissolution which are termed reaction- or interface-controlled for the former and transport- or diffusion- controlled for the latter. Phases of dissimilar chemical nature tend to react at high temperature, so that in general, in order to limit the extent of dissolution, the refractory and liquid in contact should be of similar nature. Si02, B2O3, P2O5 and V2O5 are acidic so that high silica refractories are used for acid liquids (some steelmaking, coal gasifiers, and glass melts). MgO and CaO are basic and so are used in contact with basic melts. 12 In direct, congruent, or homogeneous dissolution, atoms from the solid dissolve directly into the liquid melt. Direct dissolution can be reaction or interface controlled when the diffusivity of reaction products is faster than the rate of chemical reaction at the interface, or transport (diffusion) controlled if it is slower. In the former case the dissolution process may be directly controlled by a reaction that is of the first order with respect to a reactant species, whose initial rate can be expressed by [10]: J= KfA/AoJC,,, (5) where J - the dissolution rate (g/cm.s), K - the rate constant, 2 Ac- the actual area of refractory (cm ), 2 A0 - the apparent area of refractory(cm ), and C,„ - the concentration of reactant in the melt (g/cm3). Surface irregularities such as grooves and porosity, which increase the ratio AJA0 in the above equation, have a significant effect, while other microstructural features such as crystal orientation, grain boundary phases, and grain shape are neglected. In this simple treatment, stirring of the melt has no apparent effect on dissolution rate. For the direct dissolution to continue, the atoms diffuse away from the interface at a rate proportional to t]/2 (which / is time) as reactants are depleted and dissolved species build up in the absence of liquid convection or stirring [10, 12]. In a situation where the rate of removal of reaction products by diffusion is slower than the rate of chemical reaction, a solute rich boundary layer builds up, whose 13 interface with the refractory is saturated with reaction products. The dissolution process is then governed by the diffusion of reactants to, or the products away from, the interface through the boundary layer. If the boundary layer leads to formation of a solid interface, this is termed indirect, incongruent, or heterogeneous dissolution. Here, the rate of corrosion can be in terms of the Nernst equation [12]: J=D(CS-CJ/S (6) where D - the diffusion coefficient (cmV1), 3 Cs - the saturation concentration of refractory in the melt (g/cm), C„r the concentration of reactant in the melt (g cm'3), and S- the effective boundary layer thickness (cm), which is defined as: S=(Cs-C„)/(dc-dx) (7) where dc/dx is the concentration gradient over the interface. The saturation concentration of the refractory components in the melt is important, but the saturation of liquid in the solid may also be important. If the solid is unsaturated with respect to at least one component of the liquid, then solid solution reaction may occur. Stirring the melt or rotating the refractory sample enhances the rate of indirect dissolution or effectively converts it to direct dissolution by reducing the thickness of any liquid boundary layer or breaking up any solid layer. In the case of a flat slab held 14 vertically in the melt, the rate depends on the boundary layer thickness, which is limited by such variables as the degree of convective flow caused by thermal and density gradients, liquid viscosity, the mean diffusion coefficient, and the container size. It is worth noting that in practical situations, dissolution is often under mixed control and it is difficult to distinguish between the first order reaction and diffusion control. Therefore, the refractory corrosion experiments must be performed over a wide range of hydrodynamic conditions to demonstrate the dissolution mechanism convincingly. The molten slag viscosity has significant effect on both the slag penetration and refractories dissolution. A more fluid slag will be more penetrating and more likely to dissolve the solid refractory. Furthermore, if dissolving the refractory in the liquid leads to increased viscosity, then mass transport through the melt layer next to it will be slower, so that the melt layer becomes progressively saturated. On the other hand, if the viscosity of melt layer is decreased then diffusion through it becomes more rapid, and no saturated layer forms [10]. 15 2.2.2 Corrosion of Refractories in Contact With Various Slags In order to understand the chemical attack of complex multiphase refractories, it is necessary first to understand the corrosion mechanisms of the individual phases present in their microstructures before the synergistic effects of their combination in a multi- phase assemblage can be considered. 2.2.2.1 Magnesite Refractories Single crystal MgO could dissolve in a 40CaO - 20Al2O3 - 40SiO2 (wt%) melt at 1400° C [13]. Adjacent to the MgO surface is a two-phase composition that appears to lie on the 2CaO- MgO- 2Si02 - MgO and CaO- MgO- Si02 - MgO compatibility joins. The spinel on the silicate side is uniform in structure, with a layer rich in MgO and 2CaO- Al203- Si02 between the spinel and silicate. Rose and McGee [14] studied the corrosion resistance of MgO single crystal by basic oxygen furnace (BOF) slags, containing predominantly CaO- Fe203- Si02 (CFS) with additional MgO, Al203 and Ti02. They found that addition of Ti02 to the slag produced a sharp increase in MgO solubility. Slags were more corrosive when their MgO concentration was decreased, as presumably, they became unsaturated in it. Adding Al203 alone to the slag did not cause any appreciable difference in MgO 16 corrosion rate, although AI2O3 combined with the addition of CaO decreased the solubility of MgO in the liquid. The CaO-FeO-Si02 (CFS) slags containing AI2O3 attacks sintered grain MgO refractories by fluxing the C2S bond phase [15]. Adding MgO to the slag slows down the attack due to increased slag viscosity and to lower solubility of MgO in the MgO saturated slag. Dissolution of sintered magnesia into CFS liquid at 1300-1425° C proceeds through dissolution of FeO into MgO forming a layer of magnesiowustite. This phase later dissolves at a rate controlled by diffusive mass transfer, presumably of FeO and MgO, through the slag boundary layer. Further more, the C/S=2 slag is more able to dissolve MgO than the more basic C/S=4 slag, because of the higher solubility of MgO in it and its lower viscosity. Penetration into the grain boundaries of sintered MgO aggregate occurs with the C/S=4 slag. However, slags with lower C/S ratio allows for a more rapid diffusion of Fe via volume diffusion into fused MgO crystal, while grain boundary diffusion of Fe2+ in sintered polycrystalline MgO is more rapid than volume diffusion in fused MgO. Zhang and Seetharaman [16] investigated the dissolution of MgO in CaO-FeO-CaF2- Si02 slag under static conditions. They suggested that the process of MgO dissolution in a slag consist of formation of magnesiowustite solid solution and its simultaneous dissolution in the slag. Dissolution of MgO increased with CaF2 content in the slag and solid solution layer thickness increased with concentrations of CaF2 below 15wt%, but decreased substantially at higher CaF2 levels. This was most probably due 17 to decreased silicate melt viscosity arising from breaking of the linked silicate chains in the slag by fluoride. Magnesiowustite can have a range of composition from almost pure MgO (melting point, Tm =2620° C) to almost pure FeO (Tm =1700° C), so any addition of Fe2+ in solid solution with MgO lowers its refractoriness. Formation of a lower melting phase that MgO leads to accelerated corrosion if the test is conducted at a temperature above the melting point of this new phase, although the slag composition will also have some effect. It appears that the attack of MgO by FeO containing slags leads to a magnesiowustite layer, due to the rapid diffusion of Fe2+ compared with the other species. Its rate of formation is in general controlled by diffusion of Fe2+ through the boundary layer. The magnesiowustite layer is itself attacked and the balance between its dissolution at the slag side and growth at the refractory side is controlled by the temperature and slag composition. Another factor to consider here is the open crystal structure of cubic MgO, which allows it to take up the Fe2+ from the slag. At lower temperatures such behavior would lead to improved temperature resistance, but at higher temperatures the magnesiowustite would be preferentially dissolved. 2.2.2.2 Alumina-Chromia Refractories Alumina-chromia refractories have been used mostly in the lining of tanks for fiber glass production and so, most studies have been of attack by molten glass. Other studies, particularly in USA, examined their attack by coal ash slags. Thomas et al. 18 [17] examined the behavior of sintered alumina-chrome refractories in borosilicate glasses. Their corrosion resistance was markedly improved by formation of solid solution between AI2O3 and Cr203. These refractories were used extensively as floor and forehearth linings of glass tanks for the production of wool. Addition of Cr2C>3 was also added to fused AZS glass contact refractories leading to improved corrosion resistance due to the AI2O3 - C^Cb solid solution reduced the exudation of glass phase from the refractories into the tanks. Fused alumina-chromia refractories containing MgO-Cr203 spinel and alumina-chromia solid solution phase were also developed for use in tanks producing fibber glass, a particularly corrosive molten glass. Cooper et al [18] investigated the influence of tank glass redox conditions on the corrosion of such refractories at 1316° C. The redox conditions of the glass played no role in the corrosion of static refractory but markedly influenced the corrosion rate of rotated samples. A fused cast MgO-Cr203 spinel refractory exhibited excellent corrosion resistance to both acidic and basic coal ash slags at 1500° C [19]. The slag-refractory interactions were limited to the formation of a stable layer of recrystallized hercynitic spinel [(Fe,Mg)0 (Al,Fe)203] at the slag-refractory interface. 19 2.2.2.3 Magnesite-Chrome Refractories The CaO/SiC>2 ratio in the AOD slag (Argon Oxygen Decarburization) has a big influence on the corrosion of magnesite-chrome bricks [20], Chromite is more resistant than periclase to chemical attack by low C/S ratio slag. In contrast, the periclase is more resistant towards the more basic slag, as the slag penetrated between periclase crystals does not dissolve them. It is also demonstrated [21] that the texture of magnesite-chrome bricks has a direct impact on the corrosion behavior. Large crystallite clinkers or highly interlinked spinel matrix bonding confers good slag resistance. Several studies have examined the influence of brick compositions on slag attack. By using a rotary test, Takahashi et al [22] found that the slag resistance was increased with increasing Cr203/MgO ratio and decreased with increasing Fe203/Cr203 or Al203/Cr203 ratio in the brick. At high Cr203/MgO ratio, Cr203 rich secondary spinel with high melting point and lower solubility in Si02 rich slag was precipitated, but at high Fe203/Cr203 or Al203/Cr203 ratios, the spinel formed had a lower melting point and higher solubility in the slag. Ichikawa et al [23] found that AI2O3 in the brick deteriorated the slag resistance, but Fe203 had no effect. This was explained by the smaller size of the Al3+ ion, which enables it to dissociate from the secondary spinel and dissolve in the slag easier than Fe3+ or Cr3+. Hiragushi et al [24] discussed the relationship between the slag resistance and the Cr203/MgO ratio in the brick concluding that, except for highly basic slags with low Al203 content, the slag 20 resistance increased with increasing the C^Ch/MgC) ratio. The results were interpreted in terms of the different solubility of MgO in each slag composition. Direct bonded bricks, made with fused or sintered grains have been corroded at 1300° C by a fayalite slag in a rotary slag test [25]. Greater corrosion was observed with increased CaO/Fe203 ratio of the brick, which typically contained 10% Fe203 and <1% CaO, and the fused materials exhibited better resistance to attack than the sintered ones. Wiederhorn et al. [26] investigated the effect of coal slag on the microstructure of magnesite-chrome refractories. They suggested the following mechanism of the slag attack: Al3+, Fe2+ and Fe3+ from the slag enter the spinel at the hot face, while Mg2+ and Cr3+ leave the surface grains of refractory. The Al3+, Fe2+ and Fe3+ reduce the spinel grains melting temperature, a consequence of which is the enhanced solution of the refractory in the slag. Within the refractory, Al3+, Fe2+ and Fe3+ leave the penetrated slag to be replaced by Cr3+ and Mg2+, which changes the melt composition from one close to anorthite (CAS2) to one close to diopside (CMS2). Because of this change in composition the viscosity of the intergranular glass in the refractory is lowered, making this glass more reactive. At the same time, submicrometer grains that lie between the larger aggregate particles grow. This grain growth coupled with transport of matter form the slag to the spinel grains of bricks causes the hot face to bloat, crack, and eventually spall from the main body of the brick. The author suggested that the addition of a network former such as AI2O3 to the brick would locally lead to more viscous and thus less penetrating slag. Another approach to 21 improving slag resistance of magnesite-chrome brick is to add Fe-Cr powder to the mix. The expansion caused by the oxidation of the Fe-Cr powder during firing, increases the brick density and thus the corrosion resistance. 2.3 Wetting Behavior of Refractories 2.3.1 Wettability Wetting involves the interaction of a liquid with a solid, including spreading of a liquid over a surface, and penetration of a liquid into a porous medium. Study of wettability can help to characterize surfaces and to understand solid-liquid interactions. Wettability is most often described by a sessile or resting drop. The contact angle (9) is a measure of wettability (Figure 8). A low contact angle means high wettability and high contact angle means low wettability. The contact angles are always less than 180° and 0° angles are possible. By balancing the horizontal force components at the three-phase line, Young [27] developed the following expression, which relates the contact angle 9 with the interfacial energies (Figure 8): cos 0= (Ysv - ysi)/yiv (8) where ysv, Yin and 7si are the equilibrium solid-vapor, liquid-vapor, and solid-liquid interfacial energies, respectively. As seen in the above equation, the contact angles formed by a given liquid on different solids are determined by ysv - YsU a factor called the driving force for wetting. 22 Liquid 7/f r-* Solid \ < (a) (b) Figure 8. Equilibrium forces at the solid-liquid-vapor interface for an (a) acute contact angle and (b) obtuse contact angle [27]. SOLIO Figure 9. Schematic diagram of a sessile drop on a surface. Both advancing and receding angle are shown [27]. 23 There are two main techniques for measuring the contact angle. They are the sessile drop and the Wilhemy plate methods [27]. Figure 9 is a schematic diagram of a sessile drop, showing an advancing and receding angle. In the sessile drop method, a drop is placed on a horizontal surface and observed in cross section through a telescope. A goniometer in the eyepiece is used to measure the angle. The major advantages of the sessile drop technique are speed and convenience. 2.3.2 Wetting Behavior of Refractories at High Temperatures Surface and interfacial phenomena, such as wettability, reaction between the liquid and solid phases, as well as the dissolution of the solid, can greatly affect the longevity of refractory materials. Equation (3) in section 2.2.1 indicates that to minimize the infiltration of the melt, the contact angle should be greater than 90°, and the surface tension aiv should be low. Very few fundamental studies have been published regarding the wettability of refractories, and in particular for those used in non-ferrous industry, they do not practically exist [28]. The wetting of refractories by molten slags can be influenced by many factors. Comeforo et al. [29] studied the wetting of various refractories by a soda-lime silicate glass in air at 1200° C. They observed that the contact angle changed from 0° to >40° on preformed glass drops of a fixed size, during 2h isothermal soaks, but the results were affected by substrate roughness, porosity and reactivity. Cronin et al. [27] 24 investigated the temperature and composition dependence of the wetting behavior of PbO - Si02 liquid on AI2O3. The contact angle decreased as the PbO content of liquid and the temperature increased, the authors attributed this to reductions of the driving force ysv -ysi for wetting. Hiroyuki et al. [30] investigated the wetting behavior between fayalite-type slags and solid magnesia, using a single crystal of MgO as the substrate material and slags with Fe/Si02 mass ratio of 1.44 and 2.05, and found that the contact angle varied over the range of 30°-10° and was dependent on time and the Fe/Si02 mass ratio. The interactions between slags and their refractory substrates also play an important role in the wetting behaviors. Towers [31] reported on the wetting of dense alumina by calcium alumino-silicate slags at 1415°C. The convergence of the contact angle of the slag of different composition after 12 min was attributed to a decrease in the solid-liquid interfacial energy, ysi, as the slag became saturated with alumina. Hamano et al. [32] studied the wetting of dense, 99.5% MgO by MgO - A1203 - Si02 melts. They monitored the contact angle of 1mm diameter drops formed in situ at 1370° C over a 2h period. Sintered MgO with 3 mol % O2O3 yielded a contact angle of 29°, compared to 24° on sintered MgO without additives, and 30° on single- crystal MgO. The slight increase in contact angle was attributed to picrochromite (MgCr204) formation. 25 2.4 Mechanisms of Refractory Corrosion in Non-Ferrous Smelting Environments A mineralogical study was performed on reacted refractories after corrosion tests with slags from two different furnaces [33]: 1) a Hoboken Cu-converter, used for the conversion of a Pb - Cu matte into blister-Cu and 2) a cupellation furnace for the treatment of materials containing precious metals. In both cases, the behavior of two refractories (chrome-magnesite and alumina-chromia), typical for non-ferrous extractive metallurgy, was described. In basic refractories of chrome-magnesite type, the diffusion of bivalent cations of Fe, Zn, and Ni into the structure of chromite and periclase is typical for the converter slag attack. Formation of complex spinel (Mg,Fe,Zn,Ni)(Fe,Al,Cr)204 and nickel- bearing periclase (Mg,Ni)0, followed by dissolution in silica to form a modified forsterite phase (Mg,Fe,Ni)Si04 have a significant contribution to the global corrosion mechanism. The same chrome-magnesite refractory, when in contact with the slag in the cupellation furnace, shows unexpected reactions involving both periclase and chromites. Tellurium oxide, existent as a slag component and as a volatile component of the flue gases, dissolves the periclase to form MggTeOe, with a volume expansion , which explains the fissuration of the lining materials. The antimony reacts with chromites to form a complex spinel (Mg,Fe)i+xSbxOy . Sodium 26 and silica migrate along grain boundaries, dissolving some MgO and destroying the silicate binder phase. In the amphoteric alumina-chromia refractory, the converter slag causes only partial dissolution of the alumina grains, but severely attacks the mixed oxide binder phase. The lead silicate slag is enriched with alumina during infiltration. Appreciable spinellisation of alumina and chromium oxide by bivalent cations of Zn, Ni, and Mg to (Fe,Zn,Ni,Mg)(Al,Cr,Fe)204, which will protect the refractory against further attack, is observed. The cupellation slag dissolved both oxides to form Pb(Al,Cr)i20i9 and additionally, some sodalite (Na,Ca)4.8(Al6Si6024)(S04)i-2- This reaction, caused by Ca,Na,PbS04 and Si02 present in the slag and the flue gases, occurs along grain boundaries. It appears therefore that the stability of the binder phase is the critical point. The chemical interaction between matte or white metal and a refractory material was characterized by a diffusional exchange of iron and nickel, both soluble in the matte, and the magnesiowustite phase of the refractory [34]. Wetting between ternary copper-iron sulfide melt and magnesite-chrome refractory was minimal for copper- rich melts having an iron activity approximately equal to iron activity of the refractory. The penetration rate of copper-rich matte into magnesite-chrome refractory, at the near equilibrium conditions, was slow and obeys phenomenologically the general laws of capillary. Iron-rich copper matte and low 27 melting eutectic enriched with nickel sulfide impurity were able to penetrate rapidly the open pores of refractories because of the chemical reaction enhancing the wetting. E. N. Selivanov [35] investigated the periclase-chromite linings in contact with matte (sulfide - metallic melt) in shaft melting of oxidizing nickel ores. It was shown that the sulfide component of matte, preferentially iron sulfide, impregnated the refractory, whereas the metal (21.9-23.l%Ni, 1.38-1.45%Co, 73.3-76.l%Fe and 0.8- 1.0%S) remained in the gaps between bricks. The author concluded that the impregnation of refractories with sulfide melt changes the properties of the lining substantially, but no specific data were given. 28 3 Scope and Objectives of this Work The present work is part of an ongoing collaborative research project between UBC, Cominco Ltd. and Clayburn Industries Ltd., aiming to provide refractory solutions for the KIVCET and TBRC furnaces at Trail Operation, Cominco Ltd., British Columbia. The objective of this work was to determine corrosion mechanisms and evaluate the performance of magnesite-chrome and alumina-chromia refractories in KIVCET and TBRC furnaces at Cominco-Trail Operation. In order to accomplish this, the following research investigations were conducted: 1. Investigation of the interfacial phenomena at the slag-refractory interfaces: • wettability studies by the sessile drop technique at high temperatures • study of chemical reaction products by SEM/EDS and XRD 2. Microstructural studies by SEM/EDS and XRD of the slag-refractory interfaces on the followings: • magnesite-chrome brick after use in the KIVCET furnace, • specimens after laboratory corrosion tests. 3. Laboratory evaluation of the corrosion behavior of various refractory materials against industrial slags from KIVCET and TBRC furnaces by using: • rotary slag furnace for studying the dynamic corrosion, • slag cup test for studying the static corrosion. 29 4 Experimental Procedures 4.1 Contact Angle Measurements The sessile drop technique combined with high temperature microscopy (HTM) were used to determine the contact angle of a slag at high temperature on various substrates (Figure 10). Substrate Slag Furnace Alumina holder n Optical Microscope Figure 10. Schematic diagram of contact angle measurement at high temperatures In the experiments, a small cube of KIVCET slag was carefully positioned onto the center of the substrate material, and the substrate and slag were then placed onto an alumina holder, which was then positioned within the furnace hot-zone. The furnace was heated up to 1400° C at a rate of 55-60° C/min. The photographs of wetting 30 behavior were taken at intervals through a optical microscope and contact angles were measured based on the pictures taken. Two procedures of contact angle measurements were used. According to one, the slag-substrate assembly was heated up at a fast rate of 55-60° C/min and the contact angle versus temperature was measured (fast-heating mode). According to the other, the slag-substrate couple was introduced directly at 1250° C and the contact angle versus time was obtained (hot mode). 4.2 Corrosion Tests Although the refractories are usually exposed to a dynamic type of corrosion during their service in the reactors, we used both dynamic and static tests in laboratory experiments to measure the corrosion of the refractories. Rotary Slag Furnace (RSF) was used for the dynamic corrosion test and the classical Slag Cup Test (SCT) for the static. 4.2.1 Dynamic Corrosion The Rotary Slag Furnace (RSF) is the most widely used method for measuring corrosion, which is standardized by ASTM C874 "Standard Practice for Rotary Slag Testing of Refractory Materials" [36]. As shown in Figure 11, six or eight specimens constituting a test lining are positioned midway in a shell (in our experiments, eight 31 specimens were used). The slag is melted with an oxygen-acetylene flame and the furnace rotates about a horizontal axis. The slag is refreshed at periods throughout the test by tilting to remove old slag, returning to horizontal, and adding new slag. c) Schematic overall view of the RSF set-up Figure 11. Schematic diagram of rotary slag furnace (RSF) 32 The advantages of this method include the fact that many specimens can be compared in a single test, a temperature gradient can be established, and the composition and fluidity of the slag is partially controlled, although during the test, the slag composition (especially the oxidation state of iron oxide) may change. The temperatures for testing were the maximum achievable in normal industrial conditions: 1350°C for KIVCET slag and 1075°C for TBRC slag. The furnace was heated up to the testing temperature in approximately 2 to 2.5 h and for 0.5h, during which time, charged with 1.0 kg of slag pellets ( 1.0 kg/h and the reacted slag was tapped out before each feeding. The furnace rotated at a constant speed of 3 rpm. At the end of testing, immediately after shutting off the oxygen, gas and the motor, the furnace was tilted to allow the remaining slag to drain out. After the cold furnace was disassembled, each specimen was saw-cut along the 228mm length perpendicular to and at the center of the slagged face. The sectioned refractory specimens were measured for corrosion effects, and by comparing with the original cross section of the specimens, the corrosion and penetration were calculated. 4.2.2 Static Corrosion Slag Cup Test (SCT) is a classical corrosion testing method. As shown in Figure 12, a core-drilled refractory brick is filled with slag and exposed to high temperature to 33 promote slag-refractory interaction. This method is frequently used, since it is simple and specimens can be tested in a short time. However, it suffers from the drawbacks that are usually associated with any static tests, for example, there is no temperature gradient, the slag is rapidly saturated with reaction products and there is no contribution of the dynamic factors, e.g. slag movement vs refractory surface. Slag Figure 12. Schematic diagram of slag cup test (SCT) The tests were performed by filling with the slag a O20mm x 30mm crucible core- drilled in 76mm cube cut from the brick samples. The crucible-slag assembly was held for 24 hours at testing temperatures. After test, it was sectioned and the corroded areas measured. 34 4.3 Sample Preparation 4.3.1 Samples for Contact Angle Measurements KIVCET slag was ground into -100 mesh powder and pressed into 3.0x3.0x3.0 mm3 cubes using a temporary binder, then dried at 110°C before test. Six different substrates were used to determine the wetting behavior between slag and refractories, as seen in Table 1. Table 1. Substrates used in the contact angle measurements Substrate Apparent A1203 MgO Cr203 Source Porosity (%) (%) (%) (%) Pure A1203 ~0 99 - - Haldenwanger, West-Berlin Pure MgO ~0 - 99 - Ozark Technical Ceramic, INC. MC2 32.7 5.0 62.0 20.0 Refractory brick MC2 provided by Cominco Ltd. MC6 36.1 21.5 34.6 26.0 Refractory brick MC6 provided by Cominco Ltd. AC4 43.8 57.0 0.2 40.5 Refractory brick AC4 provided by Cominco Ltd. AC7 45.3 80.0 - 16.0 Refractory brick AC7 provided by Cominco Ltd. Because the textures of the refractory bricks (MC2, MC6, AC4, AC7) were too coarse to be directly used for contact angle measurements, a special preparation procedure was used in an attempt to obtain fine, uniform textures, as seen in Figure 13. 35 Refractory bricks (MC2, MC6, AC4, AC7) Crushing and ball-nulling -325 mesh powders Forming by pressing Discs (20mm in diameter, 2mm thickness) Firing at 1500°C r Specimens with fine, uniform texture Figure 13. Flow chart of refractory substrate preparation for contact angle measurements All substrates were cut into rectangular shapes of 15mmx7mmx2mm and having the contact surface holding the slag cube polished with 600-grit sand paper to obtain flat, scratch-free surfaces. 36 4.3.2 Samples for Corrosion Tests In order to explain the selection already used during the years, based on trial and error methods or on the advice of technical experts, only the refractories with acceptable performance were used in our lab experiments. Two groups of refractory bricks, 9 magnesite-chrome (MC) and 9 alumina-chrome (AC), were identified as performing acceptably in different areas of the main smelting furnaces of KIVCET and TBRC at Cominco Ltd.. Their chemical compositions and physical and mechanical characteristics are presented in Table 2. The magnesite-chrome bricks were of direct-bonded (MC5 to MC8) or rebonded fused-grain (MCI to MC4 and MC9) type. The silicate-bonded bricks were eliminated, due to their already known unsatisfactory corrosion performance. MCI to MC4 and MC9 were made of fused magnesia-chrome clinkers, others like MC5, MC7, and MC8 of seawater magnesite and MC6 of dead-burned magnesite and chromite. The alumina-chromia bricks were all of low impurity level (CaO, MgO, Na20, K20, Fe203, Ti02) except AC5, which was found to have an unusually high content of K20 and Na20. The apparent porosities varied from 13% (MCI) to 20.6% (MC6) for the MC group, and from 12.6% (AC2) to 17.3% (AC3) for the AC group. 37 "H. £ 03 on 15. 0 11. 0 10. 0 11. 0 10. 0 21. 0 31. 0 7. 6 37. 0 49. 0 6. 0 7. 0 4. 0 4. 1 MP a o Mod . o f Ruptur e < 120. 0 141. 0 134. 0 100. 0 80. 0 80. 0 69. 0 58. 0 90. 0 Strengt h 82. 0 27. 0 26. 0 28. 0 78. 0 Crush . MP a £ o >_ on o CNNOi/liriOOOO""! r--' CN r-' vo ro rr vo 18. 2 18. 0 14. 5 13. 7 14. 0 15. 8 13. 0 14. 0 03 < PH . ^ 20. 6 c £ 03 3.2 3 3.1 1 3.2 0 3.2 9 3.4 1 3.2 7 3.6 4 3.1 0 3.0 6 3.2 5 3.2 2 3.3 0 3.2 8 3.3 3 3.0 6 3.0 3 3.3 0 3.2 8 Dens . g/c m -o Bul k C 03 O O PH 16. 0 10. 0 12. 0 15. 5 17. 0 19. 1 8. 0 18. 0 7. 0 9. 8 9. 0 9. 9 40. 5 \° 31. 8 26. 0 o £ 20. 0 21. 4 29. 0 £ o O u- 00 c <0. 1 57. 4 63. 0 0. 1 Trac e Trac e 0. 2 34. 6 70. 7 62. 5 O 62. 0 62. 0 63. 0 43. 6 03 0s on 0) o i C oo co oo i/~> Tf oo ; r-; o <0. 1 00 O ^H o o © o © o ~ 0. 3 0. 1 U . 1 0. 2 od > £ o -*-» rN CN — '— CN iri ~ — CN ro o 11. 0 11. 0 13. 8 11. 8 12. 0 12. 0 10. 8 PH m 5. 5 8. 0 © o © o © © © o © o O. on o I— 0) on N O o ro rs co > 11. 3 80. 0 85. 0 57. 0 80. 0 86. 0 89. 7 89. 6 75. 7 >9 0 6. 0 03 5. 0 7. 7 9. 6 21. 5 9. 3 6. 3 6. 5 on +-» 03 _C r- o U i— O IT) Cl vo T O © O; roi O CO 03 ^H o < Chemica l Compositio n on <^ O © *-H' O O —" CN —' © O o CN CN O CO O T3 < C CN on ^HCNro^tmvor—ooON ^cNro^^vor-oooN £ VO OOOUOOUOU oououoooo O U Bric k <<<<<<<<< £ < E—1 CO 4.4 Slag Compositions Two industrial slags were selected for our laboratory corrosion tests, one from KIVCET furnace (KS), and the other from TBRC furnace (TS). According to its composition in Table 3, the KIVCET slag contains mainly Si02, CaO, and FeO/Fe203, accompanied by PbO (4.4%) and ZnO (10.1%). The composition of TBRC slag is of a very particular type, with 36% Bi, and 25.2% Pb as majors, and relatively high levels of Sb. In the wettability study, only KIVCET slag was used on different refractory substrates. Table 3. Chemical Composition of Slags Si02 CaO Fe Pb Zn As Sb Bi Cu Sn S % % % % % % % % % % % KS Slag Typical 21 12 30 3 18 - - - 0.3 - 1.4 Experim. 23.2 13.2 26 4.4 10.1 0.08 0.17 0.01 0.24 0.06 0.6 TS Slag Experim. 1.3 0.48 0.51 25.2 0.02 0.28 6.2 36 1.2 0.02 0.23 39 4.5 Testing Methods In order to investigate the interfacial phenomena and the microstructures at the slag- refractory interface, • the contact angles between KS slag and different refractory substrates ( pure AI2O3, pure MgO, MC2, MC6, AC4 and AC7) were measured by using the sessile drop technique and a high-temperature microscope, • morphologies of different specimens, including KS slag, KS slag-A^Os crucible interface, the used MC brick from KIVCET furnace and the specimens after corrosion tests, were studied using SEM/EDS, • the crystalline phases in the KS slag and at the interface were identified by XRD. Laboratory evaluation of the corrosion behavior of 18 bricks (9 MC bricks, 9 AC bricks) with KS and TS slag was performed by using: • rotary slag furnace for studying the dynamic corrosion • slag cup test for studying the static corrosion. 40 5 Results and Discussion 5.1 Characterization of KIVCET Slag When the slag-substrate couple was heated up with 55-60° C/min (fast-heating mode), the KS slag on pure AI2O3 substrate began to deform at 1150° C, formed a hemisphere at 1250°C and finally spread completely above 1400° C, as shown in Figure 14a. The contact angle (9) between KS slag and the pure AI2O3 substrate decreased with increasing temperature from 9=70° at 1150°C to 9=10° at 1400° C as shown in Figure 16. When introduced directly at 1250° C (hot mode), the KS slag on pure AI2O3 substrate immediately formed a hemisphere, then began to spread for about 30 seconds. After that, the drop didn't deform any more (Figure 15a). The contact angle measurements for pure A1203 and pure MgO (Figure 17) show that the contact angle varied with time, and after approximately 30 seconds, it remained almost constant, 9=25° -28° . 41 Room Room Temp. Temp. 1000°C 1000°C 1150°C 1100°C 1250°C 1250°C 1300°C 1300°C 1400°C I400°C a) KS slag on b) KS slag on MC2 alumina Figure 14. Wetting behavior of KS slag on pure AI2O3 and MC2 substrates at different temperatures. Heating rate: 55-60°C/min. 42 a) KS slag on b) KS slag on alumina MC2 Figure 15. Wetting behavior of KS slag on pure AI2O3 and MC2 substrates at different soaking times. The KS slag was introduced directly at 1250°C Figure 16. Contact angle versus temperature for KS slag on various substrates. Heating rate: 55-60° C/min. 140 O) 120 01 100 Q —A A a» •A -AI203 | oi 80 c -MgO < 60 11 o E -MC2 ro c 40 o _}= —0 —_| o 20 0 -I 500 1000 1500 2000 Time (s) Figure 17. Contact angle versus soaking time for KS slag on various substrates at 1250° C. The slag was introduced directly at 1250° C In the fast mode, the KS slag on MC2 substrate started to deform at 1150°C, and formed a hemisphere at 1400° C. High volume of shrinkage could be observed between 1250° C and 1400° C (Figure 14b), while the contact angle ranged from 109° to 90° (Figure 16). Similar tests on different porous substrates of MC6, AC4 and AC7 showed similar wetting behavior. In hot mode, the KS slag on MC2 substrate also showed much shrinkage with time, the contact angle decreased from 120° at the beginning to 92° after about 600 seconds (Figure 15b and Figure 17). Comparing to the wetting behavior of KS slag on AI2O3 substrate, the KS slag on MC2 substrate showed much slower equilibrium. The XRD analysis of the slag remaining on the MC2 substrate (Figure 18) identified a crystalline phase, zinc ferrite (ZnFe2C>4). This is a spinel with a cubic structure of unit cell dimension a = 8.411 A and the melting temperature at 1590° C. The standard Gibbs energies of the reactions at 1100K are as follows [37]: 1 ZnO(s) + 2/3 Fe304 +I/6O2 = ZnFe204(s), AG° = -59,090 J- mol" ZnO(s) + Fe203(s) = ZnFe204(s), AG° = - 30,160 J- mol"' . The free energies for both reactions are large negative values, suggesting that there is a high tendency for such reactions to take place at 1100K. 45 Figure 18. XRD patterns of residual slag on MC2 substrate (top) and original slag (bottom) 46 The SEM analyses also indicated the formation of a crystalline phase in the residual slag on MC2 substrate, as shown in Figure 19. All four points A, B, C and D showed mainly Fe, Zn and O (Table 4). Figure 19. SEM images of residual slag on MC2 substrate 47 Table 4 Results of EDS analyses of residual slag on MC2 substrate in Figure 19 Element Point A Point B Point C Point D 0 37 34 34 33 Mg 1.0 0.4 0.7 2.5 Al 1.0 1.2 1.4 2.3 Si 0.7 2.0 0.5 2.2 Ca 0.8 1.8 0.4 3.3 Cr 0.3 0.2 0.3 2.5 Mn 0.6 0.7 0.6 0.7 Fe 36 37 39 35 Zn 19 20 21 17 Pb 0.4 0.3 0.5 0.4 It appeared that two different phases occurred in the KS slag on MC2 substrate at the test temperature, one being the crystalline phase (ZnFe2C>4), which remained on the MC2 substrate, another being a liquid phase which penetrated into the porous substrate. This phase separation was the reason for the excessive volume variation of KS slag drop on MC2 substrate. In order to further examine this phenomenon, an alumina crucible (zero porosity) filled up with KS slag was heated up to 1250° C, soaked for 3h. The XRD analysis was run on the slag (Figure 20), and the result showed that zinc ferrite ZnFe204 and other unidentified crystals were formed in the slag at this temperature. SEM images at the interface between the slag and the crucible in Figure 21 showed that the crystalline phase clustered near the interface (Figure 21a) and in some other areas (Figure 21b), the crystalline phase scattered among the glassy phase. The results 48 of EDS analyses in different positions (A, B, D, C and E) in Figure 21a and 21b in Table 4 show that the compositions B and C in the crystalline phase area were mainly Fe and Zn and D and E in the glassy phase area were a mixture of Ca, Si, Fe, and Zn. The molar ratio ZnO/Fe2C«3 of 0.95 and 0.96 respectively for points B and C suggested that the compositions are close to zinc ferrite Zrdi^CV Figure 20. XRD patterns of the KS slag after 3h at 1250° C in zero porosity alumina crucible 49 Figure 21a. SEM image of KS slag at the top of the alumina crucible, 3h at 1250°C Table 5. Elemental compositions of KS slag in Figure 21a and 21b Element Point A Point B Point C Point D Point E 0 32 8.9 11 24 20 Na 0.0 2.8 4.9 0.0 0.0 Mg 0.0 0.0 0.0 0.0 0.0 Al 64 4.2 3.8 1.9 1.8 Si 0.2 3.3 0.0 16 17 Ca 0.4 2.5 0.3 23 24 Cr 0.0 0.0 0.0 0.0 0.0 Mn 0.0 0.6 0.1 0.3 0.3 Fe 0.2 47 50 11 13 Zn 2.6 29 30 14 16 Pb 0.5. i.o 0.0 9.7 9.9 Composition alumina ZnFe204(?) ZnFe204(?) Glassy Glassy crucible phase phase Table 6. Calculated compositions of KS slag based on EDS analyses in Table 5 Glassy Crystalline Original slag phase phase wt% wt% wt% A1203 3.0 6.3 2.5 Si02 30 3.0 26 CaO 28 1.6 14 Fe203 15 58 35 ZnO 16 31 14 PbO 9.0 0.5 8.7 51 Based on EDS analyses results in Table 4, slag compositions in terms of oxides (assuming all elements in oxidic state) were calculated and included in Table 6. According to these data in Table 6, the Si02, CaO and PbO from the original slag remained in the glassy phase while the Fe203 and ZnO were concentrated in the crystalline phase. Considering the properties of the substrates (Table 1), it could be assumed that the difference in porosity between Al203 and MC2 substrates was the reason, for this type of phase separation. The glassy phase, containing high amount of glass-modifying oxides (CaO, ZnO, PbO), penetrated the MC2 porous substrate, leaving behind the crystalline phase, initially dispersed in the glassy phase. Because the pure Al203 substrate had zero porosity and it couldn't absorb the emerging liquid from the slag at high temperature, the composite liquid - crystalline phase could then spread across the substrate. As majority of remaining KS slag on MC2 substrate was crystalline phase, the contact angle measured between the KS slag and the porous MC2 substrate actually did not represent the real wetting of the porous substrate by the KS slag. Similar phase separation was observed in the Na20 - CaO - ZnO - PbO - Fe203 - Al203 - Si02 system [38]. The radial and dendritic crystallization of spinel zinc ferrite (ZnFe204) and magnetite (Fe304) was surrounded by a glassy matrix after heat treatment at 400° C. It was pointed out that a wide range of glass separation was 52 expected in such a system because of a high content of glass-modifying ions, and liquid immiscibility is generally enhanced in the silicate glass system, containing high amount of bivalent oxides such as MgO, CaO, ZnO and PbO. 5.2 Interaction between KS slag and refractory substrates As demonstrated by our wetting experiments, the KS slag contains two phases at 1250° C: a glassy phase which is a liquid at that temperature, and a crystalline phase of mainly zinc ferrite. Because the glassy phase contains as major components Si02, CaO and PbO, it maintains its corrosiveness for basic refractories. In order to determine the possible reaction between this glassy phase of the KS slag and the refractory substrates, XRD analyses were carried out on areas impregnated by the molten slag of different substrates, as indicated in Figure 22. The XRD patterns of impregnated areas of AC4 and MC2 substrates and their unimpregnated original area are presented in Figure 23. 53 Residual crystalline Glassy phase phase impregnated area Substrate Figure 22. Schematic phase separation of the KS slag on porous refractory substrate AC4 impregnated with - -'. :; * O Corundum AI2O3 KS slag after lh at 1300° C . : ;~i ii.;; • 1 -=j-r-i-":i--i...r: •• ' ;i JO 40' AC4 original i-,:- !;,.| O Corundum AI2O3 X Chromium oxide Cr203 Z6 Figure 23 a. XRD patterns of KS slag impregnated areas of AC4 substrate and its original 54 'T___ O Magnesiochromite(Mg,Fe)(Cr,Al) 0 MC2 impregnated with KS slag 2 4 A Zinc chromite ZnCr.O* after lhatl300°C X Periclase MgO •44-4:: H-i^lrr<••••!:: i: vH _ 1-4; joe 60° O Magnesiochromite (Mg,Fe)(Cr,Al)204 MC2 original X Periclase MgO Figure 23b. XRD patterns of KS slag impregnated areas of MC2 substrate and its original From the XRD patterns (Figure 23a), it is clear that there are two phases in original AC4 specimens: corundum Al203 and free chromium oxide Cr203, but surprisingly the free Cr203 disappeared (no XRD peaks) in the slag-impregnated AC4 substrate. 55 This suggests that free Cr203 either reacted with the slag or formed a solid solution with AI2O3. Phase diagrams can provide some useful information about the refractory-slag interactions. In the Al203-Si02-PbO system (Figure 24) [39], there are many low temperature eutectics (from 694° C to 865° C) and low melting point compounds (below 1000°C), suggesting that Si02 and PbO are corrosive to the alumina refractories. In the CaO-Cr203 and PbO-Cr203 systems (Figure 25) [39], it can be also observed that low temperature eutectics are formed, and these formations would destroy the refractory matrix, and as a result could lead to the failure of the refractory. Magnesio-chromite ((Mg,Fe)(Cr,Al)204) and periclase (MgO), as the crystalline phases were found in original and impregnated MC2 substrate (Figure 23b). There is no big difference between the XRD patterns of impregnated MC2 substrate and its original. But because magnesio-chromite has a similar spinel structure with another spinel zinc chromite (ZnCr204), their XRD peaks almost overlapped, it was not clear if zinc chromite (ZnCr204) was formed as well. 56 PbO-Al203-Si02 Figure 24. Phase Diagram of the Al203-Si02-PbO system [39] a) Ca0-Cr203 2000 b) PbO-Cr203 1000 .Liquid* aA ! A Pb0 T h ,• -A l ' 1 1600 900 \- —-•^^^ Liquid* a AB, " Lf Ml' »20*" ^ .—•—f—s 800 a A,B,+ A,B A,B«xAB:.5 185' | AjB«lA,8,' Aj6*PA3ib ',J"""- ' TOT*.-" 4 f—AJB'BAJB,^ 101* 700 800 iu u 00 CaO ' " ™ Cf20, Figure 25. Phase diagram of a) CaO-Cr203 system b) PbO-Cr203 system [39] 57 5.3 Postmortem Analyses of Used MC Bricks from the KIVCET Furnace In order to further study the corrosion mechanisms, postmortem analyses by SEM/EDS and XRD, were carried out on magnesite-chrome bricks after use in KIVCET furnace at Cominco Ltd-Trail Operations. The analyses were mainly focussed on the interface between slag and brick, thus a special sample preparation procedure was applied: the sample was cut into slices 2mm in thickness, parallel to the hot face of the brick As shown in the Figure 26, SEM/EDS, and XRD analyses were performed on three different areas: the residual slag, the slag-brick interface and the brick. Each EDS analysis was done on a area of 600x500u.m. The following results in this section are averages of the results from three EDS analyses. 58 Slag Interface MC Brick Figure 26. Analyzed areas of the used MC brick from KIVCET furnace Table 7. Slag compositions on the used MC brick sample by EDS Element Elemental Corresponding Concentration Oxide (wt%) (wt%) 0 19 - Mg 2.2 3.7 Al 0.6 1.2 Si 0.5 1.1 K 0.3 0.4 Ca 0.5 0.7 Cr 0.3 0.4 Fe 1.2 1.8 Zn 60 75 Pb 15 16 The EDS analyses on the slag, presented in Table 7, suggest that the slag composition was different from that of the KIVCET slag we used in the previous experiment, which had much lower Zn and Pb and much higher Fe, Si and Ca. A powder from the slag area was prepared and phase analyzed by XRD, and it was found that the XRD pattern (Figure 26) was quite complex, and the crystalline phases could not be identified at this stage. As shown in Figure 27a, 27b, the SEM images of the used MC brick from KIVCET furnaces showed typical micrographs of the magnesite-chrome brick made of magnesia-chromite-co-clinkers, with the secondary chromite spinels precipitated within the grains (the white spots on the SEM images) [40]. The variation of elemental concentrations with the distance from the interface (Figure 29) was presented in Figure 29. Zn changes from 60 to 15%, Mg from 2.2 to 14%, Cr from 0.3 to 7.5%, Fe from 1.2 to 13% with a maximum of 18.84% at the interface, Pb from 15 to 7.3; Si from 0.5 to 3.1%, Ca from 0.7 to 2.7 %, Al from 1.2 to 3.7 %. It was found that the Fe and Cr accumulated at the interface, and this was later believed to be due to the formation of ZnFe2C>4 and ZnCr204. 61 Figure 28. XRD patterns of used MC brick from KIVCET furnace. Residual slag (top), interface (middle), brick (bottom) 62 Figure 29. Elemental distribution versus distance from the interface in the used MC brick from KIVCET furnace 63 The XRD analyses of the interface (Figure 28) showed that two spinel phases were identified: zinc chromite (ZnCr204) and zinc ferrite (ZnFe204). The phase diagrams available for the system were studied. Figure 30a shows the ternary phase diagram of ZnO - Cr203 - Yb203 [39], From the binary edge of ZnO-Cr203, it is clear that the only compound is ZnCr204- The density of the compounds (D) involved in the formation of zinc chromite and zinc ferrite are shown as follows: Compound Density (g/cm3) ZnO 5.68 Fe203 5.30 Cr203 5.25 ZnFe204 5.07 ZnCr204 5.43 From the density data above, the density changes before and after the spinel formations could be calculated. For the formation of ZnFe204, the average density before reaction is: = D average (DznO x ZnO percentage in ZnFe204) + (DFe2o3 x Fe203 percentage in 3 ZnFe204) = 5.68x33.75 %+ 5.30x66.25% = 5.43 (g/cm ) 3 While DZnFe204 = 5.07 g/cm Therefore the volume change (AV) accompanying this reaction is: AVZnFe204 = (D average - DZ„Fe204)/D Z„Fe204 = (5.43-5.07)/5.07 = +7.1 % 64 indicating the reaction is a expansion reaction. The same calculation was done on the formation of ZnCr204, and the volume change is AVz„cr2O4 = -0.55%, suggesting that the formation is a contraction reaction. Due to these volume changes accompanying the reactions at the interface of slag- refractory couple, microcracks could develop, and the brick could be damaged or even destroyed. The increase in Fe2C>3 and Cr2C>3 contents at the interface was believed to be related to the development of ZnFe2C«4 and ZnCr204 (Figure 29). 65 b) ZnO Uol X F«JOJ Figure 30. Ternary phase diagrams of a) ZnO-Cr203-Yb203 and b) ZnO-Fe203- Yb203[39} 66 5.4 Results of Corrosion Tests Using both static and dynamic corrosion tests, the corrosion experiments were carried out to simulate the industrial conditions, in order to evaluate the corrosion behavior of available refractory bricks when used in contact with molten oxidic and metallic phases generated in various pyrometallurgical processes from KIVCET and TBRC furnaces at Cominco Ltd.. 5.4.1 Magnesite-Chrome (MC) Bricks with KS Slag The results of both static and dynamic corrosion tests of magnesite-chrome bricks with KS slag are presented in Table 8, and the images of corroded samples in dynamic (RSF) and static (SCT) testing are also presented in Figure 31a and Figure 32a. The static corrosion test (Table 8, Figure 31a), even at 24 hours holding time, did not bring enough information in order to have a measurable corrosion rating for the particular couples refractory-slag used in our experiments. The penetration depth was between 5.7 mm to 15.2 mm but the corroded area was very small (between 0.12 to 1.76 mm2). 67 Table 8. Results of corrosion tests of the magnesite-chrome bricks with KS slag Brick Dynamic (RSF) Static (SCT) Bonding type and sample Penetration Corrosion Penetration Corrosion aggregates Depth Area Depth Area (mm) (xlO3) (mm) (xlO3) (mm2/mm ) (mm2/mm2) MCI 8.8 49.3 5.7 0.7 RFG*, Magnesite-chrome clinker MC2 4.4 15.7 6.5 0.3 RFG, Magnesite-chrome clinker MC3 5.9 37.9 8.5 1.7 RFG, Magnesite-chrome clinker MC4 5.3 28.4 8.2 3.0 RFG, Magnesite-chrome clinker MC5 15.7 151.9 15.2 2.5 DB**, Seawater magnesite MC6 12.9 128.6 7.2 1.0 DB, Dead-burned magnesite and chrome ore MC7 10.1 104.4 10.7 1.3 DB, Seawater magnesite and chrome ore MC8 5.5 31.6 10.2 0.3 DB, Seawater magnesite and chrome ore MC9 4.7 18.8 7.7 0.1 RFG, Magnesite-chrome clinker Note: * RFG—Rebonded Fused Grain, **DB—Direct Bonded. 68 71 Figure 32b Corroded MC specimens after RSF 5 hours with TBRC slag at 1075°C 72 From the dynamic corrosion results for the MC group, we can easily distinguish between corrosion behaviors and to a certain extent correlate them with the brick properties. Using the corroded area as a corrosion parameter, there is a clearly better behavior of the rebonded fused grain type (RFG) than the direct bonded type (DB). From all the results for the corroded area as a corrosion parameter, the MC2 had the lowest corrosion, while the highest belongs to MC5. The corrosion rating, based on the corroded area data, gave us the following order for all 9 bricks in the MC group: MC2 < MC9 < MC4 < MC8 < MC3 < MC7 < MCI < MC6 < MC5 Trying linear correlation of the corrosion data with the brick properties (MgO, Cr203, Fe203, apparent porosity, CCS, MOR, HMOR), no apparent explanations could be found regarding their corrosion behavior, but there were some similar properties among the top 3 bricks (MC2, MC9, MC4) in the corrosion rating scale which were believed to be responsible for better corrosion performance: • Low silica content (0.6%, 1.0% and 0.5% respectively) • Low apparent porosity (14.0%, 14.5% and 14.0% respectively) • High bulk density (3.28, 3.25 and 3.28 g/cm3 respectively) • High crushing strength (78.0MPa, 90.0MPa, - respectively) • Rebonded fused grains 73 From the bricks at the end of the corrosion scale, MC5 and MC6 were the most corroded. It was observed that both bricks were direct bonded, with seawater (MC5) or dead-burned (MC6) magnesite-chrome clinker as aggregates (Table 8). Especially, the MC6 brick had the highest open porosity (20.6%), highest silica (4.9%) (Table 2). And these factors were believed to be responsible for the poor corrosion behavior of MC5 and MC6. When taking a close look at the best (MC2) and the worst (MC5 and MC6) specimens (Figure 33), we found that MC5 and MC6 had much coarser textures than MC2. These coarser textures which have larger pores allowed the molten slag to penetrate them more easily. This is an another factor that contributes to the poor corrosion resistance of the bricks. Microstructural studies by SEM/EDS on the MCI specimen (MC1/KS/RSF) shed some light on the corrosion process, see Figure 34. In Figures 34a, 34c and 34d, two different kinds of structures could be identified in the slag layer adherent to the brick surface. One, containing coarse grains of magnesia-chromia clinker (Field 03, Figure 34d) surrounded by the reacted slag and another the reacted slag (Field 02, Figure 34c). This could be explained that the slag penetrated into the brick and destroyed the matrix, and as a result the aggregate grains were washed away into the reacted slag. 74 The EDS analyses in Table 9 showed the slag (Point B) as a white phase and fine precipitates with very high iron content (42.1%, while the chromium was only 15.22%). Also, in Field 01 in Figure 34a and 34b, at 0.5mm from the slag-refractory interface, the slag penetrated through the pores and intragranular fractures and formed different compositions of complex spinels. Point F in Field 01 was very rich in chromia (34.14%, while magnesia was 13.09%) and Point G was very rich in magnesia (41.41%, while chromia was 6.22%). The complex spinel (Mg, Fe)0-(Cr,Al)203 richer in chromia than a stoichiometric magnesia-chromia spinel, within the periclase or a spinelic phase richer in magnesia, are specific to high-fired magnesia-chrome bricks, regardless if they are of RFG or DB type [40,41]. Point A (Figure 32a) and its vicinity on the right hand side were also of similar formation. 75 Figure 34a. SEM image on MCI brick after RSF test with KS slag Figure 34b. SEM image on MCI brick after RSF test with KS slag (Field 01) 77 Figure 34c. SEM image on MCI brick after RSF test with KS slag (Field 02) Figure 34d. SEM image on MCI brick after RSF test with KS slag (Field 03) 78 Table 9. EDS analyses of MCI brick after RSF test with KS slag. Element Field Point Field 02 Field 03 Field 01 01 A Point C Point B Point D Point E Point F Point G 0 38 37 40 32 39 37 36 39 Mg 31 13 5.0 3.6 11 14 13 41 Al 3.7 12 3.1 4.0 7.6 13 10 1.4 Si 3.1 0.7 19 0.2 8.9 0.2 0.0 0.4 Ti 0.1 0.4 0.2 0.1 0.7 0.4 0.4 0.0 Cr 11 29 0.6 15 14 30 34 6.2 Fe 8.2 7.0 13 42 4.3 6.1 6.0 8.4 Na 1.9 0.0 0.0 0.0 0.0 0.0 0.0 1.7 Ca 2.5 0.0 16 0.0 14 0.0 0.0 0.2 Zn 0.0 0.0 1.2 2.7 0.0 0.0 0.0 1.1 5.4.2 Alumina-Chromia (AC) Bricks with KS Slag The corrosion test results (Table 10, Figure 31b), using both dynamic and static methods, showed for the AC group of bricks, a better corrosion behavior than for the MC group, when tested in identical conditions. In the RSF test, all the AC bricks lost amounts comparable with MC2 brick, which was the best according to any rating presented above. 79 Table 10. Results of corrosion tests of alumina-chromia (AC) bricks with KS slag Brick Sample Dynamic (RSF) Static (SCT) Penetration Corroded Penetration Corroded Depth Area Depth Area (mm) (mm2) (mm) (mm2) AC1 3.6 30.1 3.3 0.8 AC2 2.8 23.8 0.0 0.5 AC 3 4.5 27.0 0.0 0.6 AC4 7.5 14.9 0.0 2.9 AC 5 5.2 39.0 6.8 2.5 AC6 2.0 29.6 0.0 2.1 AC7 3.1 41.3 0.0 1.0 AC 8 2.5 39 7.7 0.7 AC9 - - 5.5 2.0 The static corrosion results were comparable to MC group with respect to corrosion, but the penetration for the majority of AC bricks was practically zero (Figure 35). Because of such uniform results for the whole group of nine bricks, we have not used any ratings regarding the corrosion behavior. 80 5.4.3 Magnesite-Chrome Bricks with TBRC Slag The results of corrosion tests with TBRC slag were presented in Table 11 and Figure 32b. Surprisingly, the TBRC slag - considered potentially more corrosive than KIVCET slag due to its high content of PbO, Sb203 and Bi203 - did not corrode the MC and AC group very much, even in the dynamic test. We do not know at this time if that was due to a tendency of this slag to be reduced to a metallic phase, or due to a factor in our experimental conditions, or to a clearly observed non-wetting behavior of all the MC bricks in contact with the TBRC slag at 1075°C. The interesting fact was a very deep penetration by the metallic phase, rich in bismuth. SEM on MC4 after the dynamic test with TBRC slag (Figure 36), showed that the liquid phase penetrated through the pores and grain boundaries very deep in the brick, reaching as deep as 35 mm . 82 Table 11. Results of corrosion tests of MC and AC bricks with TBRC slag at 1075° C Brick Dynamic Sample Penetration Depth Corroded Area (mm) (mm2) MCI 0.5 0.6 MC2 1.3 1.1 MC3 0.4 0.6 MC4 2.0 1.3 MC5 2.3 5.1 MC6 0.1 0.1 MC7 0.2 0.8 MC8 0.7 0.3 MC9 0.4 0.3 AC1 0.7 0.5 AC2 0.0 0.0 AC3 0.7 0.2 AC4 0.1 0.4 AC 5 0.1 0.1 AC6 0.7 0.3 AC 7 0.0 0.0 AC 8 0.1 0.1 AC9 0.1 83 Figure 36a. SEM image of MC4 brick with TBRC slag (interface) Figure 36b. SEM image of MC4 brick with TBRC slag (35mm from interface) The bismuth associated with lead penetrated as deep as 25mm, but bismuth continued its way down to 35mm (Table 12). The presence of Te, Pb, and Sb could be detrimental to the refractory if they would penetrate and react in their oxidic state. Table 12. Results of EDS analyses of MC4 brick after RSF test with TBRC slag. Element Elemental concentration (%) at different depth of penetration 0 mm 5 mm 15 mm 25 mm O 34 34 35 35 Mg 30 31 31 33 Cr 12 13 13 14 Fe 6.2 6.5 6.9 7.4 Sb 5.8 3.0 0.0 0.0 Pb 0.0 0.8 0.0 0.5 Al 2.8 2.7 3.1 3.1 Si 1.1 0.8 0.7 0.8 Cu 0.0 o.o 0.0 0.0 Te 0.0 0.0 0.0 o.o Bi 8.1 7.8 7.3 5.0 Ag 0.0 0.0 0.0 0.0 As 0.0 o.o 1.7 0.0 . No sign of any reactions at the interface metallic phase refractory could be identified. Further study is needed towards the corrosion mechanisms in this respect. 85 5.5 Summary Although refractories are indispensable in lead-smelting processes, there is no systematic study on the corrosion of refractories in literature so far. The selection of refractories is based either on the trial and error method or on the knowledge accumulated in other pyrometallurgical processes such as iron- and steel-making. Corrosion of refractories in lead-smelting reactors is quite a complex phenomenon. Many factors can be attributed to the failure of refractories, such as their physical and chemical properties, nature and intensity of the interactions with slag, slag properties, service temperature, etc. Corrosion by slag, depending on the particular system, involves various processes, such as chemical wear (corrosion) and physical or mechanical wear (erosion), which may act synergistically. No single model can explain all cases of corrosion nor can it explain all corrosion mechanisms of a particular refractory in different environments. Knowledge of the microstructure and the chemistry of the refractory-slag system are necessary to understand the corrosion mechanism of a refractory material. In this work, the interfacial phenomena at the slag-refractory interfaces were investigated using different analytical methods, such as sessile drop technique, and 86 microstructural changes on refractories and slags due to their chemical reactions at the surface were studied using SEM/EDS and XRD. The wetting behaviors of KS slag on zero porosity substrates (pure AI2O3 and MgO) and porous substrates (MC2, AC4 AC7) were different. The contact angles of molten KS slag on pure A1203 and MgO substrates were temperature-dependent, ranging from 10° to 70° when the temperature changed from 1400° C to 1250° C. They also remained practically constant at 25-28° after the first 30 seconds at 1250° C. On the porous MC2 substrate, due to the phase separation of the KS slag, the observed contact angle didn't represent the real wetting of the KS slag. It was pointed out that porous specimens are not suitable to be used as substrates to measure the contact angle under these conditions. At 1250° C, the KS slag contains two phases, a crystalline phase identified as mainly zinc ferrite (ZnFe204) and a glassy phase containing Si02, CaO, ZnO and PbO. The glassy phase could penetrate the refractories and then react with the refractories, leading corrosion. For the alumina-chromia refractories, PbO, CaO and Si02 in the glassy phase could react with AI2O3 and Cr203 and form low temperature eutectics or low-melting temperature compounds. No such damages were identified in our study. For magnesite-chrome refractories, it was found that the ZnO from the KIVCET slag reacts with Cr203 and Fe203 from the magnesite-chrome brick and reacted slag to form spinel-type phases: zinc chromite (ZnCr204) and zinc ferrite (ZnFe204) respectively. The volume changes (-0.55% and +7.1% respectively) accompanying 87 these reactions could lead to microcracks in the brick matrix, and eventually cause the failure of the brick. Laboratory evaluation of the refractory corrosion by industrial slags from KIVCET and TBRC furnaces was performed using both dynamic and static corrosion tests. The results showed that the magnesite-chrome brick (MC2), which was already declared the best in industrial conditions, had the best corrosion behavior when tested with KS slag. The magnesite-chrome bricks (MC5, MC6) were the worst. The rebonded fused grain type of magnesite-chrome bricks had superior performance compared with direct bonded type when used in contact with KS slag. The alumina-chromia bricks performed better than magnesite-chrome bricks when tested with KS slag. 88 6 Conclusions Following are conclusions of this research work: 1 The KIVCET slag separates into two parts: a liquid and a solid, above 1150° C. The liquid mainly consists of SiCh, CaO, and PbO, while the solid primarily contains Fe203 and ZnO which can forms zinc ferrite spinel (ZnFe204). 2 ZnO in the KIVCET slag reacts with Cr203 and Fe203 in the magnesite-chrome (MC) refractories and reacted slag to form spinels: zinc chromite (ZnCr204) and zinc ferrite (ZnFe204). The volume changes accompanying these reactions are - 0.55% (contraction) and +7.1% (expansion) respectively. 3 The contact angle of molten KIVCET slag on pure AI2.O3 and MgO substrates: o is temperature-dependent, ranging from 8 = 10° to G = 70° when the temperature changes from 1400° C to 1250° C. • remains practically constant with time (0 = 25-28°) after initial 30- second spreading at 1250° C. 4 The results of laboratory corrosion experiments proved that the MC2 brick, which Was already declared the best in industrial conditions, also showed the best corrosion behavior against KS slag. 5 The results also indicated that the rebonded fused grain type of magnesite- chrome bricks has superior corrosion performance compared with direct bonded type when used in contact with KIVCET slag. 89 6 According to both static and dynamic corrosion experiments with KS slag, the alumina-chromia bricks performed better than magnesite-chrome bricks. 7 The results of laboratory corrosion experiments showed that magnesite-chrome bricks which had the properties of low silica content, low apparent porosity, high bulk density and high crushing strength had good corrosion performance. 90 7 Future Work As a result of the present study, several directions for further investigation on the corrosion of refractories in lead-smelting reactors are recommended: 1 In order to measure the real contact angle between KS slag and MC or AC refractories, zero porosity substrates are required. 2 Although the phase separation was observed in KS slag at high temperature, the mechanism of crystallization remains unclear. Therefore more studies of the CaO-ZnO-PbO-Fe203-Al203-Si02 system are needed. 3 It is likely that reactions took place between the KS slag and the alumina- chromia substrate. More studies are recommended to verify these reactions. 91 8 References 1. D. Sarkar, "Development of Slag-Resistant Refractory Linings", M.A.Sc. Thesis, The University of British Columbia, July 1984. 2. R. E. Fisher, L. P. Krietz, L. Leung and J. G. Beetz, "Refractory Usage in Non Ferrous Metallurgical Application", UNITECR'1993. Proceeding of Unified International Technical Conference on Refractories, Sao Paulo, Brazil, ppl31- 172, 1993. 3. M. L. Jaeck (editor), Primary and Secondary Lead Processing: Proceedings of the International Symposium on Primary and Secondary Lead Processing, Halifax, Nova Scotia, August 20-24, 1989, Elmsford, N.Y. Pergamon Press, 1989. 4. DW. Goosen and M.T. Martin, "Application of the KIVCET Smelting Technology at Cominco", Mining Millennium 2000, Metro Toronto Convention Center, March 5-10, 2000. 5. G.Cuthbert, R. White, B. Martin, C. Doyle et al., "The Effect of Slag Composition Control on the Performance of Magnesite-chrome Refractory Linings in Top Blown Rotary Converter", CIM Bulletin, Vol.90, No.6, pp87-90, 1997 6. M. Brothers, G. Oprea, L. Wei, and T. Troczynski, "Corrosion of Refractories in Lead Smelting Reactors", UN1TECR'99 Proceeding of Unified International Technical Conference on Refractories, Berlin, Germany, pp391-394, 1999. 7. J.H. Chesters, "Refractories Production and Properties", The Iron and Steel Institute, London 1973. 92 8. Z. Guo, "Investigation and Applications of Cr203-Al2CVZr02 Refractories for Slagging Coal Gasifiers", China's Refract. Vol.6, No.4, ppl8-22,1997. 9. D. Harris and G. Oprea, "Cryolite Penetration Studies on Barrier Refractories for Aluminum Electrolytic Cells" , Minerals, Metals and Materials Society/AIME, Light Metals 2000, Nashville, TN, USA, pp 419-427, Mar. 2000. 10. W. E. Lee and S. Zhang, "Melt Corrosion of Oxide and Oxide-Carbon Refractories, International Materials Reviews", Vol.44, No.3, pp77-104, 1999. 11. K.Goto, B. B. Argent and W. E. Lee, "Corrosion of MgO-MgAl204 Spinel Refractory Bricks by Calcium Aluminosilicate Slag", Journal of American Ceramic society, Vol.80, No.2, pp 461-471, 1997. 12. W. D. Kingery, H.K. Bowen and D.R. Uhlmann, "Introduction to Ceramics", Second Edition, John Wiley & Sons, Inc. 1976. 13. J. L. Bates, Heterogeneous Dissolution of Refractory Oxides in Molten Calcium- Aluminum Silicate, Journal of American Ceramic Society, Vol.70, ppC55-C57, 1987. 14. S. Rose and T. D. McGee, "Corrosion of MgO Single Crystals by BOF Slags", Bulletin of American Ceramic Society, Vol. 57 , pp674-679, 1978. 15. S. M. Kim, P.S. Nicholson and W. K. Lu, "Attack of Magnesite Refractories by Alumina-containing Slag", Bulletin of American Ceramic Society, Vol.57, pp652-656,1978. 93 16. P. Zhang and S. Seetharaman, "Dissolution of MgO in CaO-"FeO"-CaF2-Si02 Slags under Static Conditions", Journal of American Ceramic Society, Vol.77 , pp970-976, 1994. 17. E. A. Thomas and E. L. Manigault, "The Application of Bonded Alumina - Chromia Refractories in the Glass Industry", Interceram 35, (Spec. Issue), 50, 1986 18. S. A. Cooper and PS. Nicholson, "Influence of Glass Redox Conditions on the Corrosion of Fusion-Cast Chrome-Alumina Refractories", Bulletin of American Ceramic Society, Vol.59 pp715-717, 1976. 19. J. A. Bonar, C. R. Kennedy and R. B. Swaroop, "Coal-Ash Slag Attack and Corrosion of Refractories", Bulletin of American Ceramic Society, Vol.59, pp473-478, 1980. 20. D. J. Chakins, V. Gilbert and J. M. Saccomano, "Refractory Wear in the Argon- Oxigen-Decarbarization Process", Bulletin of American Ceramic Society, Vol.52, pp570-574, 1980. 21. H. Kyoden, T. Onoye, Y. Satoh and K. Taniguchi, "Texture and Slag resistance of Commercial Fused Magnesia-chrome Clinkers", Taikabutsu Overseas Vol.6, No.l, pp3, 1986. 22. H. Takahashi, T. Kawakami, Y. Oguchi, I. Tsuchiya et al, "High Corrosion Resistant Magnesite-chrome Refractories for RH Degassing Lower Vessel", Taikabutsu Overseas Vol.11, No.3, pp.37-41, 1991 23. K. Ichikawa and K. Minato, "Corrosion Resistance of Magnesia-chrome Brick for Low Basicity Slag", Refractories (Tokyo) 41, No. 10, 1989, p.26-28. 94 24. K. Hiragushi and K. Tamaki, "Influence of Impurity in Chrome ore on Corrosion of Magnesia-chrome Brick", Refractories (Tokyo) 41, No. 10, 1989, p. 24-26. 25. K. Cherif, 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, Vol.66, No.3 Aug. 1997. 26. S. M. Wielderdon, R. F. Krause and J. Sun, "Effect of Coal Slag on the Microstructure and Creep Behavior of a Magnesium-chromite Refractory", Bulletin of American Ceramic Society, Vol.67, ppl201-1209, 1988. 27. J. C. Berg (editor), "Wettability", University of Washington, Seattle, Washington 1993. 28. J. R. Donald, H. Fukuyama and J. M. Toguri, "The Evaluation of novel refractory materials for the non-ferrous industry", Proceeding of the International Symposium on Advances in Refractories for the Metallurgical Industries, Montreal, Quebec, Aug. 24-29, 1996. 29. J. E. Coneforo and R. K. Hursh, "Wetting of Al203-Si02 Refractories by Molten Glass: I Measurement of Wetting", Journal of American Ceramic Society , Vol.35, No.5 , 1952. 30. H. Fukuyama, J. R. Donald and J. M. Toguri, "Wetting Behavior Between Fayalite-type Slags and solid Magnesia", Journal of American Ceramic Society, Vol.80 No. 9, pp2229-36, 1997. 31. H. Towers, "Wetting of Dense Alumina by Calcium Aluminosilicate Slag", Transaction of British Ceramic Society, Vol.53, ppl80-96, 1954. 95 32. D. A. Weirauch, J. E. Lazaroff and P. D. Ownby, "Wetting in an Electronic Packaging Ceramic System: 11 Wetting of Alumina by a silicate glass melt under controlled P02 conditions", Journal of American Ceramic Society, Vol.78 , No. 11, pp2923-28, 1995. 33. L. Evrard, A. Vanderlinden and C. Van Riet, "Mineralogy: A Tool for Processes Development in the Non-ferrous Extractive Metallurgy", UNITECER'89 Proceeding of Unified International Technical Conference on Refractories, Anaheim California, USA, pp895-905, 1989. 34. M. Makipaa and P. Taskinen, "Refractory Wear in Copper Converter Part II. Blister Copper-refractory interaction", Scandinavian Journal of Metallurgy, No.22, pp203-212, 1993. 35. E. N. Selivanov, "Impregnation of a Periclase-Chromite Lining with Nickel Matte", Refractories, Vol.36, No. 5-6, ppl 91-193, 1995. 36. ASTM C874-85 (Reapproved 1995), Standard Practice for Rotary Slag Testing of Refractory Materials, 1999 Annual Book of ASTM Standards, Volume 15.01. 37. S. Itoh and T. Azakami, "Phase Relations and Activities of the FeO-ZnO- ZnFe204-Fe304System", Shigen-to-Sozai, Vol. 109, ppl79-184, 1993. 38. M. Romero and J. M. Rincon, "Surface and Bulk Crystallization of Glass- ceramic in the Na20-CaO-ZnO-PbO-Fe203-Al203-Si02 System Derived from a Goethite waste", Journal of American Ceramic Society , Vol.82, No. 5 1313-17, 1999. 39. R. S. Roth, J. R. Dennis and H. F. McMurdie (editors) "Phase Diagrams for Ceramists, Volume VI", The American Ceramic Society Inc., 1987. 96 40. G. Buchebner, T. Molinari and D. Rumpf, "Developing Basic High - Performance Products for Furnaces in the Nonferrous Metals Industries", JOM, No.2, pp68-72, 2000. 41. K. Goto, W. E. Lee, "The Direct Bond in Magnesia-chromite and Magnesia Spinel Refractories, Journal of American Ceramic Society", Vol.78 , No.7 , ppl753-1760, 1995. 42. E. R. Segnit and A. E. Holland, "The system MgO-ZnO-Si02 ", Journal of American Ceramic Society, Vol. 48, No. 8 pp409-413, 1965. 43. J. F. Sarver, F. L. Katnack and F. A. Hummel, "Phase Equilibria and Manganese-Activated Fluorescence in the System Zn3(P04)2-Mg3(P04)2", Journal of Electrochemistry Society, Vol. 106 No. 11 pp960-963, 1959. 44. F. Toolenaar, "The Formation of Zinc Ferrite", Journal Materials Science Vol.24, No.3 pp 1089-1094 1989. 45. L. J Davies and J. M. McMollum, "Refractory Selection for Non-ferrous Smelting Application", Proceeding of Refractories for Metallurgical Industries, Winnipeg, August 23-26, 1987. 46. W. G. Davenport, "Flashing Smelting: A Look Back and a Look Ahead, Metallurgical Review of MMIJ" Vol.4, No. 2 1987. 97