A Coke Analogue for the Study of the Effects of Minerals on Coke Reactivity

Abstract

The properties of metallurgical coke are critical to the performance of the blast furnace for iron making. While extensive work has gone into understanding the relationship between the properties of coke and its behaviour in the blast furnace, the heterogeneous nature of coke makes understanding these relationships difficult. Herein we report the use of a laboratory produced coke analogue to examine the effects of minerals on the reactivity behaviour of coke in a pseudo-CRI test. The effects of a range of minerals on the reactivity of the coke analogue are demonstrated and the effects of binary combinations of silicon and iron bearing minerals are examined.

1. Introduction

The properties of the metallurgical coke used in the blast furnace for steel making have a direct effect on the quality and productivity of the steel making process. Coke provides three primary functions in the blast furnace.¹⁾ It is fuel for the furnace, the carbon source for the CO to convert iron oxide to metallic iron and provides the structural support for the blast furnace burden. The structural support of the burden during operation of the blast furnace is critical as collapse of the burden reduces gas and liquid permeability¹⁾ consequently reducing the efficiency of the blast furnace operation both in reducing through put and in increasing carbon dioxide emissions per unit of iron produced.

A number of factors affect the performance of metallurgical coke in the blast furnace. These include carbon forms (microtextures), pore network and catalytic effects on the reactivity due to minerals present in the coke.²⁾ It is desirable to predict the reactivity of coke from its key characteristics. Unfortunately this is in part limited by unknown or non-quantified effects of minerals on coke reactivity. Elucidation of the effects of minerals on coke reactivity has proved difficult due to the complex heterogeneous nature of coke and potential non-additive effects of minerals on coke reactivity.

Several different approaches have been taken to understanding the effects of minerals on the reactivity of metallurgical coke. Two principle methods have been utilised to study the effects of minerals on coke reactivity. Early attempts were made through doping of coke making coals with mineral additions^3,4) and studying the subsequent reactivity. Later techniques involved statistical analysis of a large range of cokes of known mineralogy and reactivity to attempt to identify the important mineral species with respect to coke reactivity.^5,6,7,8)

Early studies examining doped coke reactivates³⁾ indicated that the mineral cations present coke could be ranked as follows in terms of their effect on reactivity:   

$$\begin{array}{l}{\text{K}}_{\text{2}}{\text{CO}}_{\text{3}}>{\text{ Na}}_{\text{2}}{\text{CO}}_{\text{3}}>{\text{CaCO}}_{\text{3}}>{\text{MgCO}}_{\text{3}}=\text{MgO}>\\ {\text{FeCO}}_{\text{3}}>{\text{FeS}}_{\text{2}}>{\text{A1}}_{\text{2}}{\text{O}}_{\text{3}}={\text{SiO}}_{\text{2}}\text{(little or no change)}\end{array}$$

In addition it was show in this study and another^3,4) that coke reactivity was cation concentration dependent.

In a series of papers^5,6) a number of Australian cokes, selected for their range of iron barring minerals, were examined with respect to their initial apparent reactivities in CO₂. The results of these studies indicated that at low total iron contents a linear relationship is observed between iron content and reactivity which no longer holds at high iron content. It should be noted that in both these studies the initial apparent reactivities measured showed no correlation with micropore surface area measurements from the cokes.

In a major study on the effects of minerals on the reactivity of metallurgical coke^7,8) a statistical analysis of the correlations between the mineralogy and a suite of coke indices from a range of Australian cokes was examined. No strong correlations between the individual minerals or combinations of minerals could be established.

Progress in the area of assessing the impact of minerals on coke reactivity has been constrained by the inherent complexity of coke. Coke is a heterogeneous material comprised of different forms of carbonaceous material, mineral components, detrital, syngenetic and epigenetic,⁹⁾ and a pore structure dependent on the volatile mater in the source coal (and coking conditions). When exposed to high temperatures and reactive atmospheres, the aforementioned compositional and structural features, inherent in a given coke, render isolating the specific component effects on the behaviour of a coke difficult.

To overcome these issues a material, called coke analogue, has been developed for use as a tool in laboratory scale research into the mineral effects on reactivity of coke. The coke analogue was originally developed to mimic the dissolution behaviour of coke in liquid iron where it was demonstrated that the analogue behaved in a manner similar to that of industrial coke.^10,11,12,13) The use of the coke analogue is now being extended to the study of coke mineral reactivity at high temperature.¹⁴⁾ The analogue is made from a number of carbonaceous materials and can be doped with minerals to simulate an industrial coke’s mineralogy. In addition to being able to simulate the mineralogy of the coke, the coke analogue offers the advantages of having control over the porosity, carbon structures and mineral dispersion through the coke.

In this study the effects on coke analogue reactivity of individual and binary combinations of key minerals will be assessed in a pseudo-CRI test.

2. Experimental

Details of the preparation method for the coke analogue have been reported elsewhere.^{10,11,12,13,14)} The only difference for the current study was that the final firing temperature was reduced to 1473 K to better reflect metallurgical coke making conditions. Minerals for examination were added to the coke analogue during the preparation stage in a manner that allowed the cation concentration to be controlled relative to the amount of carbonaceous material remaining in the fired material.

The doped coke analogues were then reacted in a pseudo-CRI test. In each test a sample of coke analogue, approximately 18 mm diameter and 30 mm height, weighing approximately 8 grams, was placed in a perforated alumina crucible and suspended in a high temperature thermogravametric analysis (TGA) system using a platinum wire. The samples were heated under a flowing argon atmosphere (1 L/min) to 1373 K at 10 K per min and stabilised at temperature for 5 min. The gas flow was then switched to carbon dioxide at a flow rate of 2 L/min for two hours and the weight change logged to a PC. At the end of 2 hours the gas flow was returned to argon and the sample cooled to room temperature at 10 K/min. The pseudo -CRI test is designed to reflect conditions in a standard CRI test¹⁵⁾ with the exception that a much smaller mass of materials is examined (approximately 8 g in the pseudo -test versus 200 g in the standard test). Results from the pseudo-CRI test are expressed as fractional weight change (FWC). A total FWC is calculated based on the initial and final mass of the coke analogue sample, as shown in Eq. (1). Mass versus time data from the TGA test is then converted to FWC versus time data based on the total FWC over the test, as given by Eq. (2).   

$$\text{Total}\mathrm{\hspace{0.17em} }\text{FWC}=\frac{\left(\text{Final}\mathrm{\hspace{0.17em} }\text{sample}\mathrm{\hspace{0.17em} }\text{mass}-\text{Initial}\mathrm{\hspace{0.17em} }\text{sample}\mathrm{\hspace{0.17em} }\text{mass}\right)}{\text{Final}\mathrm{\hspace{0.17em} }\text{sample}\mathrm{\hspace{0.17em} }\text{mass}}$$(1)

  

$$\text{FWC}\mathrm{\hspace{0.17em} }\text{with}\mathrm{\hspace{0.17em} }\text{time}=\frac{\left(\text{Total}\mathrm{\hspace{0.17em} }\text{FWC}\right)\text{*}\left(\text{TGA}\mathrm{\hspace{0.17em} }\text{mass}\mathrm{\hspace{0.17em} }\text{at}\mathrm{\hspace{0.17em} }\text{time}\mathrm{\hspace{0.17em} }\text{t}\right)}{\text{Final}\mathrm{\hspace{0.17em} }\text{TGA}\mathrm{\hspace{0.17em} }\text{mass}}$$(2)

Fired coke analogues were sectioned and mounted in cold setting resin. Specific polishing procedures developed for the coke analogue were used to produce a mirror finish on the sectioned samples. Coke analogue porosities, in the pore size range 10–500 μm, were measured using optical microscopy and subsequent image analysis using the ImageJ software. Selected samples were subsequently carbon coated for examination in scanning electron microscopes (SEM). Two SEM’s were utilised, a JOEL 6490 for imaging and energy dispersive x-ray spectroscopy (EDS) and a JOEL FEGSEM 7001-F for imaging, EDS and automatic particle analysis using the INCA inclusion feature system. The INCA feature system provides the spatial distribution of particles in the selected area of a sample. In addition a number of metrics from the particles are recorded including area, perimeter and EDS measured composition.

2.1. Single Minerals

The coke analogue was doped with a series of different minerals of a particle size range of 25–38 μm at a level such that the mineral cation content was kept at a constant composition of 0.1 mol of cations per 100 grams of carbonaceous material after firing. The minerals selected for the current study were based on a review of the coal and coke^8,16) literature and what was readily available for purchase and are listed in Table 1. The mineral sources and an indication of the mineral purity are also given in Table 1.

Table 1. Details of the minerals, mineral form and mineral purity used to dope the coke analogue.

Mineral Formula	Form	Purity
Quartz SiO2	Refined Ore	>99.9%
Hematite Fe2O3	Unrefined Ore	~97%
Iron Fe	Laboratory mineral	>99.9%
Magnetite Fe3O4	Laboratory mineral	>99.9%
Pyrite FeS2	Geological specimen	Unknown
Troilite FeS	Laboratory mineral	>99.9%
Potash feldspar (K)[AlSi3O8]	Refined Ore	>95%*
Soda feldspar (Na)[AlSi3O8]	Refined Ore	>95%*
Gypsum CaSO4.2H2O	Refined Ore	>95%
Kaolinite Al2Si2O5 (OH)4	Laboratory mineral	>99.9%
Lime CaO	Refined Ore	>95%

*  Note the feldspars are substituted by up to 10% of the other feldspar alkali cation

Magnetite was subsequently selected as an iron mineral for detailed examination with regard to reactivity. The variables examined were: magnetite iron cation concentration, at a constant particle size vs. carbonaceous matter and; magnetite particle size at a constant iron cation concentration vs. carbonaceous matter. In the first case, magnetite concentration was set at 0.025, 0.05 and 0.1 mol iron cations per 100 grams fired carbonaceous material at a constant particle size range of 45–75 μm. In the second, the cation concentration was kept constant at 0.1 mol iron cations per 100 grams fired carbonaceous material, for particle sizes of 25–38 μm, 45–75 μm and 106–125 μm.

2.2. Binary Mineral Combination

A series of coke analogues were produced containing two concentrations of iron cations (based on magnetite) and a variety of silicon cation concentrations (based on quartz). These analogues were prepared using a size fraction of 45–75 μm for both minerals. The details of the mineralogy of these coke analogues are given in Table 2. Included in Table 2 is the silicon to iron cation ration of each of the coke analogues examined.

Table 2. Magnetite and Quartz binary mineral compositions (in moles of cations per 100 g of carbonaceous material).

Quartz	Magnetite	Si to Fe cation ratio
0	0.05	–
0.0125	0.05	0.25
0.025	0.05	0.5
0.05	0.05	1
0.075	0.05	1.5
0	0.1	–
0.025	0.1	0.25
0.05	0.1	0.5
0.1	0.1	1
0.15	0.1	1.5

3. Discussion

3.1. Single Minerals

Reactivity data from each of the doped coke analogues is presented as FWC versus time in Fig. 1. The total FWC over the entire test for each of the doped coke analogues is given in Table 3. Included in Fig. 1 and Table 3 are data from a base coke analogue (no added minerals) for comparison. The legend for the data in Fig. 1 has been arranged in order of the minerals effect on reactivity from lowest (top) to highest (bottom). Porosities, in the range 10–500 μm were measured to be between 39–43% for the coke analogues, indicating excellent control of porosity. An exception was the gypsum containing coke analogue with a porosity of 45%. This is believed to be due to dehydration of the gypsum during firing of the coke analogue.¹⁷⁾

Fig. 1.

FWC vs. time for each of the single mineral containing coke analogues, note the legend lists the minerals in their order of reactivity from lowest (top) to highest (bottom).

Table 3. FWC after 2 hours of single mineral containing coke analogues. Except for the base the minerals were added to give 0.1 mol of cation per 100 g of carbonaceous material.

Mineral	FWC	Mineral	FWC
Kaolinite, Al2Si2O5(OH)4	–0.32	Gypsum, CaSO4.2H2O	–0.60
Quartz, SiO2	–0.41	Iron, Fe	–0.63
Soda feldspar, (Na)[AlSi3O8]	–0.46	Troilite, FeS	–0.70
Potash feldspar, (K)[AlSi3O8]	–0.46	Magnetite, Fe3O4	–0.70
Base (no mineral matter)	–0.52	Lime, CaO	–0.81
Pyrite, FeS2	–0.59	Hematite, Fe2O3	–0.81

Minerals that reduce reactivity (relative to the base), in decreasing order of effect, are kaolinite, quartz, potash feldspar and soda feldspar. Minerals that increase reactivity (relative to the base), in increasing order of effect are pyrite, gypsum, metallic iron, troilite, magnetite, lime and hematite.

Alkali containing minerals: Of particular note is that in the present study the alkali containing minerals (feldspars) showed a reduction in reactivity of the coke analogue. This is contrary to what would be expected from the literature on doped cokes,^3,4) where coke reactivity was clearly increased by the addition of alkali oxides. However it has been previously noted that in real cokes that alkali metals incorporated in alumina-silicates display lower than expected reactivity’s.¹⁸⁾ Given in Fig. 2 are EDS maps from a potash feldspar particle in the fired potash containing coke analogue. Clearly evident is the retention of the alkali metals in the feldspar after firing. It is likely that the specific mineral form of the cations present in coke is of critical importance when considering their effect on coke reactivity. Previous studies examining the effects of alkalis have added sodium or potassium oxides.^3,4) In the current situation the alkalis are in solution in the feldspar and as such would have a lower Na₂O or K₂O activity than the pure alkali oxide.

Fig. 2.

Back scattered electron image (BSE) and Al, K, Na, O and Si EDS maps of a potash feldspar particle in fired coke analogue.

Metallic iron and iron oxide reactivity: The behaviour of metallic iron versus iron oxides is difficult to understand and is the subject of ongoing analysis. It is expected that during the coking process the iron oxides are reduced to iron (or iron carbide). Using a simple thermodynamic approach and considering the reaction given in Eq. (3)   

$${\text{<Fe}}_{\text{3}}{\text{O}}_{\text{4}}\text{>}+\text{4<C>}=\text{3<Fe>}+\text{4(CO)}$$(3)

where < > indicates solid and ( ) gas and the carbon in the form of coke is assumed to have the same activity as graphite. Using standard thermodynamic Gibbs free energy data (ΔG°)^19,20) a ΔG° for the reaction given in Eq. (3) can be obtained and is given in Eq. (4).   

$$\mathrm{\Delta }\text{G}°=633\mathrm{\hspace{0.17em} }460-656\text{T}\mathrm{\hspace{0.17em} }\text{J}$$(4)

At 1473 K the ΔG°= –332754 J. This highly negative value indicates that the reaction is highly favoured to progress from left to right. A similar result can be expected for all other iron oxides. Such a reaction would result in the iron particles potentially having a highly activated surface and as such a greater effect on the coke analogue reactivity. The reduction of magnetite has been confirmed and will be discussed later.

An alternative hypothesis is that during reduction of the iron oxides in the coking process the carbonaceous material around the mineral particles is modified in a manner that increases the carbon reactivity. Further research is currently being undertaken to assess the effects of iron minerals on carbon structures at or near the carbon mineral interface.

Gypsum, pyrite and troilite reactivity: During the coking process gypsum can be expected to be reduced to oldhamite (CaS) while pyrite will be reduced to pyrrhotite.^21,22) Note the iron oxides and calcium oxide behave very similarly, as do pyrite and gypsum. What is not clear is the behaviour of troilite (FeS) which falls in between the behaviour of pyrite (FeS₂) and the iron oxides. This may indicate troilite is undergoing partial reduction, as has previously been indicated, giving rise to a mixed phase mineral of metallic iron and troilite and as such an intermediate behaviour.

3.2. Single Mineral Concentration and Size Dependence

The effect of varying iron cation concentration (magnetite) of the observed FWC is given in Fig. 3. From Fig. 3 it can be seen that at low concentrations there is a linear effect on reactivity, while at high concentrations a nonlinear relationship is observed. This is consistent with what has been reported in the literature for metallurgical cokes doped with iron minerals and for cokes containing various amounts of iron minerals.^3,5,6) The effect of varying magnetite particle size, at a constant iron cation concentration on the coke analogue reactivity is shown in Fig. 4. It would appear that a nonlinear relationship is observed. However if the FWC is replotted against the total magnetite surface area (calculated based on the known mass of magnetite added to each sample, the known density of magnetite and assuming the particle size range can be characterised by the mean of the particle size range) a linear relationship between magnetite surface area and coke analogue reactivity is observed as shown in Fig. 5.

Fig. 3.

Effect of iron cation concentration (magnetite) per 100 g of carbonaceous material on coke analogue reactivity at a particle size range of 45–75 μm.

Fig. 4.

Effect of magnetite particle size (at constant Fe cation concentration) on the reactivity of coke analogue.

Fig. 5.

Effect of magnetite surface area (at constant cation concentration) on the reactivity of coke analogue.

3.3. Binary Minerals

Quartz and magnetite additions were examined at a variety of quartz concentrations and two magnetite concentrations. The distribution of quartz and magnetite was assessed using the INCA automatic particle analysis system for the 0.1 mol Si cation and 0.1 mol Fe cation (mol of cations per 100 g of carbonaceous material) containing coke analogue. The distribution of the silicon and iron containing particles for a 2 mm × 2 mm area at 5 mm depth from the surface of the sample is shown in Fig. 6. The maximum equivalent diameter for the Si and Fe inclusions is 33 μm and 37 μm respectively in the area shown. Both populations of particles are randomly distributed with no obvious association.

Fig. 6.

Distribution of iron and silicon containing particles in a 2 mm square area of the 0.1 mol Si cation and 0.1 mol Fe cation (mol of cations per 100 g of carbonaceous material) containing coke analogue.

Given in Fig. 7 are SEM/EDS maps showing both silicon and iron containing particles. The compositions of the particles, as determined from the areas indicated on the BSE image in Fig. 6 are given in Table 4 These data clearly indicate the Si containing particles are silicon oxide, while the Fe containing particles are oxygen free. While no obvious reduction of the silicon oxide particle has occurred, clear pickup of silicon in the iron particle has occurred. The mechanism through which this has occurred is unclear and is the subject of ongoing analysis. Also of note is the morphology of the Fe particle, which has as rough pock marked appearance. This is very different to the original powder which was fully dense and angular in nature.

Fig. 7.

Back scattered electron image and O, C, Fe and Si EDS maps of quartz and magnetite containing fired coke analogue.

Table 4. EDS measured compositions of regions shown in Fig. 7.

Spectrum 2	Atomic %	Spectrum 3	Atomic %
O	60.71	Si	2.77
Si	39.29	Fe	97.23

The total FWC results from each on the binary mineral containing coke analogues is given in Fig. 8. Included on the figure for reference are the total FWC results for the two Fe cation concentrations with no Si present. At both Fe cation concentrations reactivity is reduced with increasing Si cation concentration. In both cases the data shows a nonlinear reduction in reactivity with increasing Si cation concentration.

Fig. 8.

Effect of silicon cation content (quartz) on the coke analogue FWC at two iron cation contents (magnetite).

The data from Fig. 8 are replotted in Fig. 9 in terms of total FWC vs. the ratio of silicon cations (quartz) to iron cations (magnetite). The result demonstrates there is a near linear relationship between the ratio of silicon cations to iron cations and observed total FWC, independent of the total cation content. This may indicate quartz and magnetite (silicon and iron cations) are contributing to the coke analogue reactivity in a non-interacting manner and as such their contributions to the coke analogue reactivity are simply a direct addition of their individual effects on reactivity. An alternative hypothesis is that the reactivity effects of the minerals is saturating resulting in a linear relationship. In both cases this is supported by the non-associative distributions of the two particle distributions. The complex effects of binary minerals on the coke analogue reactivity is the subject of ongoing research.

Fig. 9.

FWC vs. silicon cation to iron cation ratio for two constant iron cation concentrations.

4. Conclusion

A coke analogue has been used to assess the effects of minerals on the reactivity of coke in a pseudo-CRI test. It was found that the analogue could discriminate the reactivity behaviour of different minerals.

• Specifically it was found that minerals that reduce reactivity, in decreasing order of effect, are kaolinite, quartz, potash feldspar and soda feldspar. Minerals that increase reactivity, in increasing order of effect are pyrite, gypsum, iron, troilite, magnetite, lime and hematite.

• The alkali containing minerals tested (soda feldspar and potash feldspar) did not show large increase in the reactivity of the coke analogue, contrary to what was expected from the literature. This is believed to be due to the form the in which the alkalis are present, i.e. in solution in the feldspars as opposed to as simple oxides and carbonates.

• The coke analogue reactivity was found to be linearly dependent on iron cation (magnetite) concentrations at low concentrations but not at high concentrations.

• The coke analogue reactivity is linearly dependent on magnetite particle surface area at a constant iron cation (magnetite) concentration.

• Coke analogues containing binary mixtures of quartz and magnetite showed a random distribution of the types two particles, indicating no association between the paticles within a type or between the two types.

• Magnetite was shown to be reduced to metallic iron during the coke analogue firing procedure. An apparent increase in the silicon content of the reduced magnetite is the subject of ongoing analysis.

• Iron (magnetite) and silicon (quartz) cations show a simple additive effect on the reactivity of the coke analogue indicating their contributions to the coke analogue reactivity are a simple combination of their individual effects.

• Future work will examine the effects of other individual minerals to determine if the same behaviour as that of magnetite is observed.

• A range of binary and ternary mineral combinations will be examined for their effect on coke analogue reactivity to determine if the additive effects observed with quartz and magnetite combinations can be generalised.

References

• 1)   A. K.  Biswas: Principles of Blast Furnace Ironmaking, Cootha Publishing House, Brisbane, Australia, (1981).
• 2)   P.  Bennett,  A.  Reifenstein,  G.  O’Brien and  B.  Jenkins: ACARP C12057, Australian Coal Association Research Program, Brisbane, (2008).
• 3)   W. W.  Gill,  N. A.  Brown,  C. D. A.  Coin and  M. R.  Mahoney: 44th ISS-AIME Ironmaking Conf., Warrendale, PA, (1985), 233.
• 4)   W. H.  Van Niekerk,  R. J.  Dippenaar and  D. A.  Kotze: J. S. Afr. Inst. Min. Metall., 1 (1986), 25.
• 5)   M.  Grigore,  R.  Sakurovs,  D.  French and  V.  Sahajwalla: ISIJ Int., 46 (2006), 503.
• 6)   B-c.  Kim,  S.  Gupta,  D.  French,  R.  Sakurovs and  V.  Sahajwalla: Energ. Fuel., 23 (2009), 3694.
• 7)   R.  Sakurovs and  D.  French: CRC for Coal in Sustainable Development, QCAT, Brisbane, (2005).
• 8)   R.  Sakurovs,  D.  French and  M.  Grigore: Int. J. Coal Geology, 72 (2007), 81.
• 9)   R. D.  Harvey and  R. R.  Ruch: Mineral Matter in Coal Ash and Coal, ed. by K. S. Vorres, American Chemical Society Symposium Series, Vol. 301, American Chemical Society, Washington DC, (1986), 10.
• 10)   B. J.  Monaghan,  M. W.  Chapman,  S. A.  Nightingale,  J. G.  Mathieson and  R. J.  Nightingale: High Temperature Processing Symp., University of Technology, Melbourne, High Temperature Processing Group, Melbourne, (2010), 41.
• 11)   B. J.  Monaghan,  N. W.  Chapman and  S. A.  Nightingale: Seetharaman Symp.: Materials Processing towards Properties, KTH, Stockholm, (2010).
• 12)   B. J.  Monaghan,  S. A.  Nightingale and  M. W.  Chapman: Steel Res. Int., 81 (2010), 829.
• 13)   R. J.  Longbottom,  B. J.  Monaghan,  M. W.  Chapman,  S. A.  Nightingale,  J. G.  Mathieson and  R. J.  Nightingale: Steel Res. Int., 82 (2011), 505.
• 14)   R.  Longbottom,  B. J.  Monaghan,  O.  Scholes and  M. R.  Mahoney: Scanmet IV, 4th Int. Conf. on Process Development in Iron and Steelmaking, Swerea MEFOS, Luleå, (2012), 147.
• 15)  ASTM D5341 - 99 (2010) Standard Test Method for Measuring Coke Reactivity Index (CRI) and Coke Strength After Reaction (CSR), ASTM, PA, (2010).
• 16)   C. R.  Ward: Int. J. Coal Geology, 13 (1989), 455.
• 17)   R. R.  West and  W. J.  Sutton: 55th Annual Meeting, The American Ceramic Society, New York, (1953), 221.
• 18)   P.  Arendt,  F.  Huhn,  H.  Kuhl and  G.  Sbierczik: Cokemaking Int., 1 (2000), 62.
• 19)   D. R.  Gaskel: Introduction to the Thermodynamics of Materials, Taylor and Francis, London, (2003), 581.
• 20)   E. T.  Turkdogan: Fundamentals of Steelmaking, Institute of Materials, London, (1996), 1.
• 21)   M.  Mahoney,  H.  Rogers,  N.  Andriopoulos and R. Gupta R: ACARP C9052, Australian Coal Association Research Program, Brisbane, (2002).
• 22)   D.  French,  R.  Sakurovs and  M.  Grigore: ACARP C14074, Australian Coal Association Research Program, Brisbane, (2009).