Reactivity of Coke Ash on Aluminosilicate Blast Furnace Hearth Refractories

Abstract

Blast furnace hearth refractories are a key component in achieving long furnace lives. These refractories can be degraded by among other things reactions with coke ash products. Recent studies have shown that these coke ash products could be calcium aluminate based. To understand and characterize the effects of these calcium aluminates on hearth refractories a study has been carried out that investigates the reaction kinetics of CaO.Al₂O₃, CaO.2Al₂O₃ and CaO.6Al₂O₃ in contact with an aluminosilicate blast furnace hearth refractory. The experimental program covered the temperature range 1450° to 1550°C. The temperatures were chosen to represent the hot face temperatures of the hearth refractories.

From this study it was found that the rate of reaction with the aluminosilicate refractory and CaO.6Al₂O₃ was much slower than that of CaO.Al₂O₃ and CaO.2Al₂O₃. The prevailing kinetics of the aluminosilicate refractory with CaO.Al₂O₃ and CaO.2Al₂O₃ was found to be consistent with the linear rate law. The slow rate of reaction of the refractory with CaO.6Al₂O₃ prohibited identification of the prevailing kinetic regime.

In characterizing the reaction interface between the aluminosilicate and the calcium aluminates it was found that there was significant reaction between the refractory and CaO.Al₂O₃ and CaO.2Al₂O₃ but little reaction with the CaO.6Al₂O₃. The reaction layers formed at the interface between the couples were found to consist of CaO.2Al₂O₃, CaO.6Al₂O₃, corundum (Al₂O₃), plagioclase (CaO.Al₂O₃.2SiO₂) and melilite (2CaO.Al₂O₃.SiO₂). The formation of a layer with these phases could result in spalling/wear of the hearth refractory.

1. Introduction

The iron blast furnace can be typically characterised as a shaft reduction furnace. It is generally operated continuously and can have a furnace campaign life in excess of 20 years. A key component in achieving this long furnace operating life is the condition of the hearth refractories.¹⁾ Blast furnace hearth refractories are principally carbon or carbon based. The hot face may also be lined with aluminosilicate and alumina–carbon refractories.¹⁾ These refractories can be degraded by thermal cycling of the furnace, wear from the flow of hot corrosive liquids (iron and slag) in the furnace, abrasion with coke (generally other phases in the hearth are either liquid or gas), reaction with the hot metal or liquid slag and reaction with the coke ash formed during the dissolution of coke in iron. The ash in blast furnace feed coke can be up to 15 mass%, the majority of which is silica.¹⁾ Given this composition it was reasonable for plant operators to develop hearth refractory protection strategies to accommodate or account for this siliceous ash. Unfortunately, it was found in a blast furnace dissection study²⁾ that the SiO₂ level in the coke decreased with increasing temperature (lower position in the furnace), see Fig. 1.²⁾

Fig. 1.

Change in SiO₂ content in coke with increasing temperature in a blast furnace.²⁾

The loss of silica from coke has also been observed in recent work by Chapman et al.^3,4,5,6) on coke dissolution in liquid iron. The SiO₂ loss from the coke in the Ono²⁾ and Chapman et al.^3,4,5,6) studies was explained in terms of SiO(g) formation by the reaction given in Eq. (1).   

$${\text{SiO}}_2\left(\text{ash}\right)+\text{C}\left(\text{coke}\right)=\text{SiO}\left(\text{g}\right)+\text{CO}\left(\text{g}\right)$$(1)

The SiO(g) formed by this reaction is then free to leave with the furnace off gas or react with the liquid slag or hot metal.

The Chapman et al.^3,4,5,6) studies also showed that the ash products that formed on dissolution of the coke into liquid iron were found to be comprised of alumina and calcium aluminates. Generally it can be expected that any ash formed will be readily taken up by the slag phase. In the hearth area, below the liquid iron-slag interface, there is no slag to absorb the coke ash. It is therefore able to react with the aluminosilicate and alumina-carbon hearth refractories. These reactions are not well understood and are likely to have significant consequences for hearth wear and blast furnace campaign life. Understanding the nature of the coke ash reactions and in particular the nature of calcium aluminate reactions with hearth refractories is important if blast furnace hearth refractory life is to be prolonged. The focus of this investigation is to characterize the reactions and reaction interface between selected calcium aluminates and an aluminosilicate blast furnace hearth refractory.

2. Experimental

An experimental study into the reactions between calcium aluminates and aluminosilicate refractories has been carried out at temperatures between 1450 and 1550°C using the experimental apparatus given in Fig. 2. The temperatures were chosen to be representative of what might be expected of the blast furnace hearth refractory hot face.

Fig 2.

A shemctic of the experimental set up used.

Reaction couples made from various calcium aluminates and an aluminsilate refractory were placed in the hot zone of the furace. These reaction couples were then heated under a high purity (99.998%) argon atmosphere to the experimental temperature. The high purity argon gas was scrubbed prior to entering the furnace by passing it through ascarite, drierite and then copper turnings at 300°C.

The calcium aluminates used in this reaction study were CaO.Al₂O₃, CaO.2Al₂O₃ and CaO.6Al₂O_3, noted as CA, CA2 and CA6 respectively. These were chosen based on the findings of Chapman et al.^3,4,5,6) Reaction times for the CA, CA2 and CA6 with the aluminosilicate refractory at 1500°C were 4, 8, 12, 18 and 24 hours. Reaction temperatures at 4 hours for the couples were 1450°C, 1500°C, 1530°C and 1550°C (with the exception of CA at 1550°C and CA6 at 1530°C). At the end of the experiment the samples were furnace cooled at 5°C min^–1, then prepared for scanning electron microscopy (SEM) and energy dispersive spectroscopy (EDS) analysis. SEM/EDS sample preparation involved mounting the aluminosilicate-calcium aluminate reaction couple in resin and sectioning/polishing it perpendicular to the reaction interface, as shown in Fig. 3.

Fig. 3.

A schematic showing the analysis plane for SEM/EDS analysis.

The thinkness of the reaction layer at the refractory-aluminate interface was measured using image analysis. An example of a measurement is given in Fig. 4 showing the reaction layer thickness using image analysis software. This measurement was repeated across the length of the layer 90 times and the average reported as the layer thickness.

Fig. 4.

A typical example of the thickness measurement. The sample is an aluminosilicste CA reaction couple reacted at 1500°C for 4 hours. The scalebar is 200 μm.

2.1. Materials Preparation

The aluminosilicate material is an industrial refractory that was supplied by BlueScope and is representative of what is used in a blast furnace hearth. The calcium aluminates used are laboratory produced synthetic coke ash compositions CA, CA2 and CA6. The use of synthesised calcium aluminates minimizes the variability typical of industrial coke ash compositions and phase distributions allowing for control of the compositions of the reactants and improving the experimental reliability. Full details of the method used to produce the aluminates are detailed elsewhere.^3,4)

The aluminosilicate refractory was comprehensively characterised to check for homogeneity.⁷⁾ This involved taking a number of samples from different locations (see in Fig. 5) in the refractory brick and inspecting the samples under the SEM. It was found that the microstructure was independent of the position within the refractory brick, indicating that experimental samples could be made from anywhere within the brick. No large defects or variations in microstructure (e.g. laminations) were observed throughout the material. The refractory consisted of mullite and corundum phases, see Table 1 for details. Both the calcium aluminates and the aluminosilicate refractory were polished to achieve a surface roughness of less than 3 μm prior to use as a reaction couple in the experimental setup.

Fig. 5.

a) An aluminosilicate brick showing the postion of core sampling b) a core sample showing the sectioning c) a typical microstrucre of the aluminosilicate refractory.

Table 1. The phases and phase compostions of the aluminosilicate refractory. The EDS positions refer to numbers noted on Fig. 4.

EDS Position	Al2O3 (Mass%)	SiO2 (Mass%)	FeO (Mass%)	Phase
1	66.7	31.7	1.6	Mullite
2	94.0	6.0		Corundum

3. Results and Discussions

For each of the reaction couples tested SEM micrographs were produced showing the reaction interface and surrounding areas (see Fig. 6). Selected EDS elemental maps are given in Fig. 7. The EDS spot analysis compositions at the points indicated in Fig. 6 were used to identify the phases at the interface in the reaction couples. The compositions and identified phases are given in Tables 2, 3, 4, 5, 6. Phases were assigned to the EDS spot analysis compositions based on the MTDATA⁸⁾ predicted phase compositions at the experimental temperature. MTDATA is a thermodynamic modeling package that can be used to calculated complex multicomponent phase equilibria.⁸⁾

Fig. 6.

SEM micrographs showing the reaction interface of the calcium alimina-aluminosilicate refractories tested for a) the temperature series 1450–1550°C at 4 hours and b) the time series 4–24 hours at 1500°C. The numbers on the micrograph represent postions of EDS spot analysis. The scalebar is 200 μm.

Fig. 7.

EDS mapping for aluminium, calcium and silicon of the SEM micrograph of the aluminosilicate – CA reaction couple after heating for 4 hours at 1500°C. The scalebar is 100 μm.

Table 2. The EDS spot analysis of the aluminosilicate-calcium aluminate couples for the temperature series experiments at 4 hours shown in Fig. 6 a). The compositions are in mass%. Note only the 1450°C, 1500°C and 1550°C data are shown.

	Spot No.	1450°C				1500°C				1550°C
		Al2O3	CaO	SiO2	Phase	Al2O3	CaO	SiO2	Phase	Al2O3	CaO	SiO2	Phase
CA	1	64.6	35.4		CA	77.2	15.4	7.4	Hibonite (CaAl12O19)	CA not tested at 1550°C
	2					63.3	4.6	32.1	Mullite
	3					64.5	35.5		CA
	4					67.5	1.75	30.8	Mullite
CA2	1	78.8	21.2		CA2	77.9	22.12		CA2	91.6	8.4		CA6
	2	87.9		12.0	Corundum	100			Alumina Grain	75.2		24.8	Mullite
	3					67.0		32.9	Mullite	78.0	22.0		CA2
CA6	1	90.7	9.3		CA6	76.7		23.3	Mullite Grain	90.4	9.6		CA6
	2	73.8		26.2	Mullite Grain	91.0	9.0		CA6	73.1		26.9	Mullite

Table 3. The EDS spot analysis of the aluminosilicate-calcium aluminate couples for the time series experiments at 1500°C shown in Fig. 6 b) at 8 hours. The compositions are in mass%.

	Spot number	Al2O3	CaO	SiO2	Phase
CA	1	65.3	34.7		CA
	2	73.8	26.2		CA2/Grossite
	3	38.2	22.5	39.3	Gehlenite (Melilite)
	4	83.4	11.3	5.3	Corundum
	5	65.4	33.1	1.6	Mullite
CA2	1	57.8		42.2	Mullite
	2	91.9	8.2		CA6
	3	77.8	22.2		CA2
CA6	1	93.3	6.7		CA6/corundum
	2	91.2	8.8		CA6/corundum
	3	77.1		23.0	Mullite

Table 4. The EDS spot analysis of the aluminosilicate-calcium aluminate couples for the time series experiments at 1500°C shown in Fig. 6 b) at 12 hours. The compositions are in mass%.

	Spot number	Al2O3	CaO	SiO2	Phase
CA	1	63.5	36.5		CA
	2	76.8	5.9	17.3	Corundum
	3	71.3		28.7	Mullite
	4	78.8	21.2		CA2
	5	79.0	21.0		CA2
CA2	1	78.13	21.9		CA2
	2	89.01	2.4	8.6	Corundum
	3	61.81	12.1	26.1	CA2/Hibonite
	4	75.55	24.4		Mullite
	5	70.73	29.3		Mullite
CA6	1	68.2	0.4	31.4	Mullite
	2	89.5	8.5	2.0	Corundum

Table 5. The EDS spot analysis of the aluminosilicate-calcium aluminate couples for the time series experiments at 1500°C shown in Fig. 6 b) at 18 hours. The compositions are in mass%.

	Spot number	Al2O3	CaO	SiO2	Phase
CA	1	67.8	32.2		CA
	2	67.8	32.2		CA
	3	79.5	20.5		CA2
	4	100.0			Corundum
	5	67.8		32.2	Mullite
CA2	1	79.5	20.5		CA2
	2	92.3	7.7		CA6
	3	92.3	7.7		CA6

Table 6. The EDS spot analysis of the aluminosilicate-calcium aluminate couples for the time series experiments at 1500°C shown in Fig. 6 b) at 24 hours. The compositions are in mass%.

	Spot number	Al2O3	CaO	SiO2	Phase
CA	1	100.0			Corundum
	2	37.7	18.3	44.0	Anorthite (plagioclase)
	3	80.3	19.6	0.1	CA2/Grossite
	4	66.8	33.1	0.1	CA
	5
CA2	1	91.4	7.8	0.8	Corundum / CA6
	2	100.0	0.0		Corundum
	3	100.0	0.0		Corundum

Phase stability diagrams (see Figs. 8, 9 and 10) for the aluminosilicate-calcium aluminate reaction couples were also calculated using MTDATA.⁸⁾ A key with the phase names associated with the phase numbers in these graphics is given in Table 7. The calculations were based on a total system mass of 1 kg at 1 atm (101325 Pa). As such the y-axis given in mass (phase)/kg can also be read as mass fraction. Figures 8, 9 and 10 represent the phase stability of different mass ratio mixtures of the aluminosilicate refractories and the calcium aluminates at temperatures from 1450°C to 1550°C. The left hand sides of the figures represent the pure aluminosilicate refractory and the right hand side the pure calcium aluminate. The phases in-between represent possible phases formed from reaction during the experiment. From Figs. 8, 9 and 10 it can be seen that for the mass ratios considered that the majority of stable phase fields contain two or three phases. It is also worth noting that a liquid oxide phase is possible (no. 8 on the figures) for all the aluminosilicate-calcium aluminate couples and not surprisingly, the amount and stability range of this phase increases with temperature (see the change in the phase distribution in these figures going from a) to c). Further, when reviewing Figs. 8, 9 and 10 it can be seen that there is a greater potential to form a liquid in the aluminosilicate-CA and CA2 couples (Figs. 8 and 9) than that of the couple containg CA6 (Fig. 10). Generally the amount of liquid phase increased with increasing alumina content of the CAx phase as evidenced by the the decrease in mass fraction of this phase going from Fig. 8 through to Fig. 10. If a liquid phase formed during the experiments this is likely to have a profound effect on the rate of reaction of the couples. The phases identified via EDS spot analysis given in Tables 2, 3, 4, 5, 6 were compared with those predicted by the MTDATA (Figs. 8, 9 and 10). Not all the phases, or rather the compositions representing the phases, predicted by MTDATA were found in the reaction couples. This is at least in part a kinetic issue. The MTDATA predictions are at equilibrium and the reaction couple experiments represent a transient (kinetic) position. There is also a general issue with EDS analysis itself. EDS compositions can have significant uncertainties as a result of the electron beam generating characteristic spectrums of “unseen” elements or phases due to sub surface beam penetration or the beam overlapping a phase boundary of a two phase region. This uncertainty combined with compositions that represent a phase near a phase boundary transition can result in a mis-identification of the phase. Ideally these phase identification problems would be overcome by carrying out x-ray diffraction analysis. The nature of the sample in general, and the reaction interface in particular, precluded such analysis.

Fig. 8.

Mass of phase predicted to be stable for aluminosilicate-CA reaction couples at different aluminosilicate CA mass ratios at a) 1450°C, b) 1500°C and c) 1550°C. Note the left hand side of each graphic represents 100% aluminosilicate, the right hand side 100% CA.

Fig. 9.

Mass of phase predicted to be stable for aluminosilicate-CA2 reaction couples at different aluminosilicate CA2 mass ratios at a) 1450°C, b) 1500°C and c) 1550°C. Note the left hand side of each graphic represents 100% aluminosilicate, the right hand side 100% CA2.

Fig. 10.

Mass of phase predicted to be stable for aluminosilicate-CA6 reaction couples at different aluminosilicate CA6 mass ratios at a) 1450°C, b) 1500°C and c) 1550°C. Note the left hand side of each graphic represents 100% aluminosilicate, the right hand side 100% CA6.

Table 7. A key of phase names representing the phase numbers shown in Figs. 8, 9, and 10.

Phase Number	Phase Name	Phase Number	Phase Name
1	CA6	10	Melilite
5	CA2	12	Plagioclase H
6	CA	13	Spinel
7	C12A7	14	Mullite
8	Liquid oxide	15	Corundum

The time dependency for the formation of the various equilibrium phases predicted in the phase stability diagrams (Figs. 8, 9, and 10) was tested by comparing phases formed at different reaction times (Tables 3, 4, 5, 6). It was found that some phases take a significant amount of time to form. The formation of CA6, plagioclase (CaO.Al₂O₃.2SiO₂), melilite (2CaO.Al₂O₃.SiO₂) and corundum (Al₂O₃) were all found to be time dependent. Each of these phases were only observed at longer reaction times e.g. 8 hours for melilite and 24 hours for plagioclase.

The formation of some phases was also found to be temperature dependent. The formation of CA2 and melilite in particular was found to be temperature dependent; both were only observed after 4 hours above 1530°C reaction temperature. However each of these phases were predicted to be thermodynamically stable at lower temperatures. This is most likely due the higher reaction rates at higher temperatures. The rate of reaction for most high temperature refractory reaction processes follow simple kinetics where the rate is in part proportional to a rate constant k.⁹⁾ This k may be representative of mass transfer or chemical reaction control kinetics. Temperature effects on k can be predicted by the Arrhenius relation given in Eq. (2)⁹⁾   

$$k=k_O{e}^{\left(-\frac{Q}{RT}\right)}$$(2) ⁹⁾

where k₀ is a pre-exponential constant, Q an activation energy, R is the gas constant and T the thermodynamic temperature. From Eq. (2) it can be seen that k has an exponential relationship with temperature, whereby increasing temperature increases k and hence increases the rate of reaction.

As stated previously in reviewing Figs. 8, 9, and 10 it can be seen that for all the reaction couple systems studied there is the potential for liquid phase formation, particularly in the higher CaO calcium aluminates and high temperature reaction couples. Given the reaction couple analysis is post experiment at ambient temperatures no liquid oxide was observed directly. Indirect evidence of liquid formation could be local densification of the reaction couple, pore penetration by the liquid, curvature of phases formed and/ or the surrounding pores and compositions of phases representative of what would be expected of a liquid at the experimental temperature. There is some evidence of these in Fig. 6. There are rounded or curved pores that may indicate that a liquid has been present. This is not definitive though as such pores can be caused by solid state reaction that causes volume change. There are also compositions corresponding to phases of what might be expected to have been a liquid oxide (plagioclase and melilite in Tables 3 and 6) at the experimental temperature.

The liquid oxide phase, as predicted by MTDATA⁸⁾ has a composition of approximately 35 mass% SiO₂, 32 mass% Al₂O₃ and 33 mass% CaO. This composition could readily solidify to form melilite and plagioclase. Since both these were identified in Tables 3 and 6 this may indicate that a liquid oxide phase was present at the experimental temperature.

3.1. Reaction Kinetics of the Aluminosilicate-calcium Aluminate Couples

Using the reaction layer thicknesses the aluminosilicate-calcium aluminate reaction couples were assessed for logarithmic, linear and parabolic kinetics. Equations representing these kinetic regimes are given in Eqs. (3), (4), (5),^10,11)   

$$x^2=2k_{par}t+B$$(3) ^10,11)

  

$$x=k_{lin}t+A$$(4) ^10,11)

  

$$x=k_{log}logt+C$$(5) ^10,11)

where x is the reaction layer thickness, k is a rate constant, t is time, A, B and C are constants and the subscripts par, lin and log denote parabolic, linear and logarithmic respectively.

Graphics representing logarithmic, linear and parabolic kinetics are given for reaction couples of aluminosilicate and CA, CA2 and CA6 are given in Figs. 11, 12, 13 respectively. The solid line represents a best fit regression line and the R² value is a measure of degree of fit.

Fig. 11.

The thickness of the reaction layer of the aluminosilicate – CA reaction couple at 1500°C anlaysed for a) logarithmic b) linear and c) prarbolic kinetics. The solid line represents a best fit regression line and the R² value is a measure of degree of fit.

Fig. 12.

The thickness of the reaction layer of the aluminosilicate – CA2 reaction couple at 1500°C anlaysed for a) logarithmic b) linear and c) prarbolic kinetics. The solid line represents a best fit regression line and the R² value is a measure of degree of fit.

Fig. 13.

The thickness of the reaction layer of the aluminosilicate – CA6 reaction couple at 1500°C anlaysed for a) logarithmic b) linear and c) prarbolic kinetics. The solid line represents a best fit regression line and the R² value is a measure of degree of fit.

The R² values were used as the discriminating parameter for deciding best fit in Figs. 11, 12, 13. The R² values of 0.9454 and 0.9396 for the CA and CA2 reaction couples indicates a good fit of the experimental data to the linear rate law (Eq. (3)).^10,11) The CA6 reaction couples had poor fits to the logarithmic, linear and parabolic rate laws with R² values of 0.672, 0.682 and 0.687 respectively. The kinetics of reaction of the CA6 couples were much slower than that of CA and CA2. This meant that very little reaction thickness data was generated over the experimental timescales. The poor fit associated with the CA6 reaction couples is most likely due to slow kinetics and the resultant small data set.

Linear reaction kinetics in refractory systems occur when the reaction layers have high porosity, cracking, spalling or consist of a liquid phase. These defects prevent the reaction (product) layer from slowing the reaction kinetics. The CA and CA2 reaction couples were observed to form reaction layers with high porosity at temperatures of 1500°C or greater (Fig. 6). This reaction mechanism is consistent with phenomena observed in the micrographs shown in Fig. 6. In this Figure there is evidence of spalling and cracking of the reaction layer at reaction times >12 hours. Also as previously discussed there is evidence that a liquid may have been present at the experimental temperature.

There is a general trend of increasing reaction layer thickness with greater Ca²⁺ content of the calcium aluminate. This indicates that Ca²⁺ is playing a key role in the reaction kinetics. Though not explored in this study, it may be expected that there would be a greater thermodynamic driving force for reaction or calcium transfer in the higher Ca²⁺ calcium aluminates.

3.2. Influence of Reaction Products Properties on Refractory Degradation

The possible formation of the liquid oxide phase, as predicted in the thermodynamic modeling of the aluminosilicate-calcium aluminate couples shown in Figs. 8, 9, and 10, may accelerate the rate of refractory loss via removal of the liquid phase and reduce wear resistance of the high porosity region remaining at the refractory surface. Table 8 provides the unit cell volume, density and thermal expansion coefficients of selected solid phases observed to be present in the reaction layer.

Table 8. Unit cell volume, density and thermal expansion coefficients of phases identified in the aluminosilicate-calcium aluminate reaction couples.^{12,13,14,15,16,17)}

Phase	Unit Cell Volume (Å3)	Density (g/cm3)	α (10–6/K)
CA (CaO.Al2O3)	1069.4	2.92	~26
CA2 (CaO.2 Al2O3)	297.7	2.9	~24
CA6 (CaO.6 Al2O3)	588.1	3.77	~21
Corundum (Al2O3)	84.9	3.99	19
Plagioclase (CaO. Al2O3.2SiO2)	697.64	2.76	1.6–2.38
Melilite (2CaO. Al2O3.SiO2)	299.05	2.98	~7.2
Mullite (3 Al2O3.2SiO2)	169.7	3.08–3.17	~6

It can be seen from Table 8 that the mullite thermal expansion coefficient is significantly different to the reaction product phases (6×10^–6/K compared to 1.6×10^–6/K and 26×10^–6/K). This difference will increase the susceptibility of the refractories to degradation by thermal spalling when exposed to cyclic temperatures.

Volume changes due to the formation of reaction products can induce local stresses causing subsequent spalling of the refractory materials. There is some evidence of this in Fig. 6. The increase in unit cell volume in the reaction layer is between 75% (formation of CA2 in the refractory) and 411% (formation of plagioclase). These values represent significant volume changes and would be difficult for a refractory or a ceramic to absorb without spalling.

The calcium aluminates, plagioclase and melilite all have lower liquidus temperatures than the refractory phases (mullite and corundum). The calcium aluminates soften between 1500°C and 1600°C, which is within the range of hearth temperatures (1450°C–1550°C). Therefore the reaction layer is likely to have a lower resistance to mechanical wear and deformation than the refractories.

3.3. Refractory – Coke Ash Reaction Kinetics in the Blast Furnace Hearth

The experimental methodology used for this study represents the situation in the hearth in which coke ash comes into contact with the hearth refractory without any molten iron present at the interface. Due to the non-wetting behavior of liquid iron with calcium aluminates¹⁸⁾ and aluminosilicates¹⁹⁾ the absence of liquid iron at the interface is believed to be a good assumption for localised conditions within the hearth. The liquid iron in the blast furnace hearth also contains [Si] up to 0.9% at 1550°C in some cases.²⁰⁾ This [Si] may react with the calcium aluminates and refractory and assist in the formation of low melting point phases such as plagioclase and melilite. The experimental setup used did not simulate the cyclical coke bed float/sink conditions that can occur in a blast furnace hearth.²¹⁾ These cyclical conditions would increase the removal rate of any reaction products via high iron flows or coke bed movement and subsequently increase the wear rate.

4. Conclusions

The focus of this investigation was to characterize the reactions and reaction interface between selected calcium aluminates and an aluminosilicate blast furnace hearth refractory. It was found that the kinetics of the refractory reaction with the CA and CA2 were consistent with the linear rate law. This is typical of a material that forms a non-protective reaction layer perhaps as a result of the reaction product layer spalling or having high porosity or being liquid.

The kinetics of reaction of the aluminosilicate refractory-CA6 reaction couples was slow, much slower than that of the CA and CA2 with the aluminosilicate. This limited the amount of data generated for aluminosilicate-CA6 reaction couple in the experimental program and made the delineation of its associated kinetic regime difficult. Given this, little weight can be given to analysis of the kinetic regime of this reaction couple.

The rate of reaction was observed to be dependent on temperature and the CaO composition of the synthetic coke ash. It was not clear whether a liquid formed during reaction of the refractory- CAx couples. Thermodynamic analysis indicated this was possible and there was some micro-structural and composition evidence that indicated it may have occurred.

The combined effects of the volume and thermal expansion changes of the reaction products formed are likely to increase the susceptibility of the blast furnace hearth refractory to spalling during furnace operation. This is significant as spalling releases fresh material for further reaction/degradation with the coke ash.

Acknowledgment

The support of BlueScope and the use of the Australian Research Council funded JEOL–JSM6490 LV SEM at the UOW Electron Microscopy Centre is acknowledged.

References

• 1)   A. K.  Biswas: Principles of Blast Furnace Ironmaking, Cootha Publishing House, Brisbane, Australia, (1981), 1.
• 2)   K.  Ono: PhD Thesis, Tohoku University, Sendai, Japan, (1979).
• 3)   M. W.  Chapman: PhD Thesis, University of Wollongong, Wollongong, Australia, (2009).
• 4)   M. W.  Chapman,  B. J.  Monaghan,  S. A.  Nightingale,  R. J.  Nightingale and  J. G.  Mathieson: Metall. Mater. Trans. B, 39 (2008), 418.
• 5)   M. W.  Chapman,  B. J.  Monaghan,  S. A.  Nightingale,  J. G.  Mathieson and  R. J.  Nightingale: ISIJ Int., 47 (2007), 973.
• 6)   M. W.  Chapman,  B. J.  Monaghan,  S. A.  Nightingale,  J. G.  Mathieson and  R. J.  Nightingale: Metall. Mater. Trans. B: Process Metall. Mater. Process. Sci., 42 (2011), 642.
• 7)   P. B.  Drain: Honours Thesis, University of Wollongong, Wollongong Australia, (2010).
• 8)   H.  Davies,  A. T.  Dinsdale,  J. A.  Gisby,  J. A. J.  Robertson and  S. M.  Martin: Calphad, 26 (2002), 229.
• 9)   P.  Kofstad: High Temperature Oxidation of Metals, John Wiley & Sons, London, UK, (1966).
• 10)   A. S.  Khanna: Introduction to High Temperature Oxidation and Corrosion, ASM International, OH, USA, (2002), 9.
• 11)   D. R.  Poirier and  G. H.  Geiger: Transport Phenomena In Materials Processing, TMS, Warrendale, PA, USA, (1994), 444.
• 12)   C. B.  Carter and  M.  Norton: Ceramic Materials Science and Engineering, Springer, London, UK, (2007).
• 13)  Slag Atlas, ed. by VDE (VDEh), VerlagStahleisen GmbH, Dusseldorf, Germany, (1995), 39.
• 14)   E. M.  Levin,  C. R.  Robbins and  H. F.  McMurdie: Phase Diagrams For Ceramists, The American Ceramic Society, Columbus, OH, USA, (1964).
• 15)   A. M.  Hofmeister: Geochimica et Cosmochimica Acta, 68 (2004), 4721,
• 16)   W. T.  Lankford,  N. L.  Samways,  R. F.  Craven and  H. E.  McGannon, Eds.: Making, Shaping and Treating of Steel, Herbick and Held, Pittsburge, USA, (1985), 367.
• 17)   M.  Tribaudino,  R. J.  Angel,  F.  Cámara,  F.  Nestola,  D.  Pasqual and  I.  Margiolaki: Contrib. Mineral. Petr., 160 (2010), 899.
• 18)   B. J.  Monaghan,  M. W.  Chapman and  S. A.  Nightingale: ISIJ Int., 50 (2010), 1707.
• 19)   E.  Kapilashrami,  V.  Sahajwalla and  S.  Seetharaman: ISIJ Int., 44 (2004), 653.
• 20)   I.  Kequin: Chin. J. Met. Sci. Tech., 8 (1992), 435.
• 21)   R. J.  Nightingale: PhD thesis, University of Wollongong, Wollongong, Australia, (2000).