Effects of Minerals and Carbon Structures on the Dissolution of Coke in Liquid Iron

Abstract

The effects of coke mineralogy and carbon structures on coke dissolution in liquid iron were studied at 1500°C. Coke mineralogy was studied by measuring the dissolution of three cokes, with three different mineralogies, in liquid iron. To allow the change in carbon structures in the coke during dissolution to be determined, samples were quenched and characterised. The dissolution of coke analogue samples were also studied, which contained the ash from the cokes.

The three cokes were found to have distinctly different dissolution rates. The dissolution of the three coke analogue samples was found to closely replicate the dissolution of the three cokes. By using the coke analogue, carbon structure, porosity and particle size were largely eliminated as variables. Therefore, it was likely that the differences in coke minerals between the three cokes were predominantly responsible for the dissolution rates of the samples.

The differences in carbon structure between the cokes likely had little effect on the dissolution of coke in liquid iron. To help understand this, Raman spectroscopy of quenched samples was used to assess the changes in carbon structures in the coke samples during dissolution. The cokes became more graphitic with time at 1500°C. Further, though the coke started off with different carbon structures, the cokes assessed tended to a similar value dominated by the temperature effect on graphitisation.

1. Introduction

Coke will remain a key raw material in the blast furnace, even as low-CO₂ blast furnace practices are being adopted by steelmakers. Understanding how coke performs in the blast furnace is key for both improving productivity and reducing emissions. Coke is the reductant of the iron ore, the energy source providing the high temperatures required in the blast furnace, and also provides the structural support for the blast furnace burden.¹⁾ The focus in this study is on coke as the source of carbon in the liquid iron. Coke dissolution in liquid iron takes place in the lower regions of the blast furnace and is strongly affected by coke properties, in particular coke mineralogy, and how the properties change as coke descends through the blast furnace. An improved understanding of coke dissolution is key for improving blast furnace productivity, lowering the consumption of coke and decreasing the CO₂ emissions associated with the iron and steel industry.

Coke dissolution into liquid iron has been shown to be affected by the mineral matter^2,3,4,5,6) and the carbon structures of the coke.^7,8) Coke mineral matter can accumulate at the coke-liquid iron interface, forming a layer and potentially inhibiting the carbon transfer from the coke into the iron, i.e. slowing the rate of carbon dissolution. This layer often presented as a build-up of an alumina based mineral at the interface that ultimately limited the carbon transfer into the iron.^{2,9,10,11,12)} This alumina based mineral layer became increasingly enriched with calcium as the dissolution progressed, changing its structure (phase), with different mineral layer structures having different effects on the dissolution rate. The mineral matter in coke is also known to affect its reactivity. However, this has generally been studied and reported for the mineral effects on coke reactivity with CO₂.^13,14,15)

The carbon structures within the coke are also a focus of this study. The carbon structures within coke are formed from the macerals in the parent coal.¹⁾ In general, the carbon structures within coke are classified into two main groups, inert maceral derived components (IMDC) and reactive maceral derived components (RMDC). IMDC and RMDC within cokes have typically been identified and characterised by optical microscopy.^1,16) Raman spectroscopy in conjunction with microscopy has also been used to study the levels of graphitisation of IMDC and RMDC in coke.^{17,18,19,20,21,22,23,24,25,26)} Recent studies^17,27,28,29) assessed the change in carbon structures upon heat treatment and gasification at high temperatures (up to 2000°C) of coke in representative blast furnace gas compositions. It was found that while IMDC and RMDC both become more graphitic upon heating, they did so at different rates.^17,27,28,29) This led to the strength of high CSR coke and low CSR coke crossing over as the temperature increased.

A coke analogue has been developed as a tool to aid quantitative assessment of the effects of minerals and porosity on coke reactivity. The analogue is made from a number of carbonaceous materials and can be doped with minerals, including those simulating industrial coke mineralogy and porosity. The coke analogue has been used to study the effect of specific minerals and mineral combinations on coke reactivity with CO₂ at high temperatures,^14,30,31,32) has been confirmed to show similar reaction mechanistic behaviour to coke³³⁾ and has been previously used to help assess coke dissolution in liquid iron.^34,35) Using coke analogue to assess the relative effects of individual minerals on the reactivity of coke it was found that their effect of the rate of reaction, in terms of increasing reactivity, was lime > magnetite > base analogue (no mineral) > quartz > alumina. These relative reactivities were consistent with those expected for mineral reactivity effects in industrial coke.¹³⁾

The aim of this study was to assess how coke properties relate to high temperature coke dissolution behaviour in liquid iron, with a focus on the mineralogy and carbon structures within the coke. The objective was addressed through carburiser cover dissolution experiments of crushed coke and coke analogue samples on liquid iron. Quenched samples were characterised to understand the changes in the carbon structures in the coke, using Raman spectroscopy to assess the carbon structures.

2. Experimental

2.1. Industrial Coke and Coke Analogue Preparation

Two industrial cokes, 131-K-001 and 114-K-001, were sourced from the ACARP Coal Bank. These cokes will be referred to as coke 131 and coke 114 respectively. A third coke (138-K-001) was prepared from coal sourced from the ACARP Coal Bank. This coke was prepared using a sole heated oven (SHO) and will be referred to as coke 138. Coke 131 was chosen for its respective comparatively high iron ash level, and coke 114 for its high silica ash level. Selected data from the three cokes is given in Table 1.

Table 1. Coke analysis.

Property	Coke code
	114-K-001	131-K-001	138-K-001
Moisture (a.d. [%])	0.7	0.9	4.6
Ash (d.b. [%])	12.0	11.9	10.8
VM (d.b. [%])	0.3	0.3	–
Ash composition [%]
SiO2	55.8	45.9	45.8
Al2O3	30.3	28.8	20.8
Fe2O3	9	15.5	9.4
CaO	1.15	3.2	7.6
MgO	0.75	1.41	3.57
Na2O	0.47	0.23	1.25
K2O	0.71	0.81	2.23
TiO2	1.62	1.55	0.91
P2O5	0.62	1.05	0.52
SO3	0.24	1.5	5.7
Coke reactivity
CSR	67.1	46.3	–
CRI	25.6	42.7	–

The coke analogue was prepared from a mixture of graphite, Novolac resin, Bakelite and hexamethylenetetramine (HTMA). Full details of the preparation have been given elsewhere.^14,30,33) The composition, given as the raw materials and the proportions of each raw material, of the coke analogue is given in Table 2. This method produces analogue samples of cylindrical shape of ~8 g in mass, 18 mm in diameter and 30 mm in height. The porosity of the coke analogue samples was measured using an optical microscopy technique, which has been previously detailed.²⁶⁾ Samples needed to have a porosity of 29.3 vol% ± 2.3% to be acceptable for use.

Table 2. Composition (raw materials and amounts) of the coke analogue.

Material	Purity/composition (%)	Quantity
Coke ash	see Table 1 for composition data	10% of final analogue mass
Graphite, < 45 μm	> 99.99	28% of dry base*
Graphite, < 150 μm	> 99.99	28% of dry base
Phenolic resin (Bakelite)	**Note	44% of dry base
Phenolic resin (50% Novolac in liquid propylene glycol)	> 99.8	Mass ratio of 0.52 liquid/dry mix***
Hexamethylenetetramine, C6H12N4 (HTMA)	>99.5	3 mass% of liquid resin

*  Dry base consists of graphite and Bakelite powder.

**  Bakelite is a complex carbonaceous material containing a small amount of mineral filler.

***  Dry mix includes dry base and HTMA.

The mineral matter for the coke analogue was prepared by ashing the metallurgical cokes. Each coke analogue sample contained 10% by mass of coke ash. Approximately 100 g of a coke sample was placed in an alumina tray and placed in a muffle furnace at 520°C in air. The sample was stirred daily to help expose the entire sample to the air. The mass of the sample was measured daily, until the rate of mass change of the sample was less than 0.1 g/day.

Both the coke and coke analogue samples were crushed to a 0.5–2 mm size fraction for the carburiser cover experiments. The crushed coke was prepared using a laboratory-scale jaw crusher, followed by ring grinding. The crushed coke was sieved using a laboratory sieve shaker for 15 min to give the 0.5–2 mm size fraction. It has been previously established that crushing the coke did not change either the amount or composition of the mineral matter from that found in the original coke.³⁶⁾

2.2. Carburiser Cover Tests

A schematic of the experimental set-up for the carburiser cover experiments is given in Fig. 1. 572 g of Fe–C alloy (initial composition 2 mass% C) was prepared from high purity electrolytic Fe (99.98% Fe) and graphite rod (99.995% C). This was heated initially to 1550°C to ensure that the pure metallic Fe melted and the graphite dissolved to form the liquid Fe–C alloy. After cooling to 1500°C (the experimental temperature), the temperature of the liquid Fe–C alloy was allowed to stabilise for 20 minutes. After the 20 minutes thermal stabilisation, 35 g of coke or coke analogue was added using a vacuum funnel. Samples of the liquid Fe–C alloy were taken at regular intervals for 180 minutes using silica glass tubes. At the end of the 180 minutes, the sample was furnace cooled. The addition of coke and metal sampling were conducted using the same tube (see Fig. 1).

Fig. 1. Schematic of the experimental set-up for the carburiser cover experiments.

A vacuum funnel was used to allow the crushed coke or coke analogue to be added to the bath without introducing air to the system. The coke was added to the vacuum funnel which was then sealed. The funnel was evacuated to a roughing vacuum, then back filled with Ar. This evacuation and backfilling was carried out three times.

Sampling was conducted using silica glass tubes, with dimensions of 3 mm OD and 2 mm ID. A 5 ml syringe was used to draw the liquid Fe–C alloy into the tube. The target sample size was 0.5–1 g. The metal samples were analysed for their carbon levels using a LECO CS444-LS carbon-sulfur analyser. The nominal sample mass used for the analysis was 0.5 g.

Quenched carburiser cover experiments were conducted to characterise the changes in the carbon structures in the coke during dissolution. These experiments were similar to the carburiser cover experiments, with no sampling but with the liquid metal and coke quenched together at predetermined times. Smaller samples (164 g Fe-2%C alloy and 10 g coke/coke analogue) contained in a smaller crucible were used to aid quenching/rapid cooling at the desired time. The sample was held at the experimental temperature for the desired time, before being quickly (typically ~11 s) lowered into a water-cooled chamber.

The samples were mounted in epoxy resin (see Section 2.3 for more details) to maintain the spatial relation between the coke, mineral layer and iron. To prevent the coke particles from floating away from the iron during mounting, a layer of iron shot (diameter 2 mm) was added on top of the coke layer prior to pouring the resin onto the sample.

2.3. Raman Spectroscopy

Samples for Raman spectroscopy were vacuum impregnated using an epoxy mounting resin (Struers EpoFix), sectioned and polished to a 1 μm finish.

Raman spectroscopy data were obtained by a Renshaw inVia confocal Raman microscope (Gloucestershire UK), equipped with a helium-neon green laser (512 nm wavelength) and a diffraction grating of 1800 grooves/mm. All Raman data were recorded over the range 400–2000 cm⁻¹ (at a resolution of 1 cm⁻¹), laser power of 35 mW and a spot size of ~2 μm. The data analyses were performed using Renshaw WiRE v4.4 software and the spectra were calibrated against the silicon peak at ~520 cm⁻¹.

Three bands were measured from the Raman spectra, with the ratios of the intensities of the bands used to characterise the level of disorder in the carbon. The three bands are shown in the example Raman spectrum for coke shown in Fig. 2 and listed in Table 3.

Fig. 2. Example Raman spectra for unreacted coke 131, showing the three intensity values measured.

Table 3. The relationship between Raman band positions and carbon bonding.⁴¹⁾

Band name	Band position (cm−1)	Bond type
D	~1335	sp2
V	~1480	sp2,sp3
G	~1590	sp2

3. Results and Discussion

3.1. Dissolution of Coke and Coke Analogue in Liquid Fe

The Δ[%C] (as defined in Eq. (1)) of the liquid Fe–C alloy during the dissolution of coke containing coke ash at 1500°C is shown in Figs. 3(a)–3(c). The three cokes had different dissolution behaviour, which could be ranked in order of slowest to fastest as coke 114 (high Si) < coke 131 (high Fe) < coke 138 (high Ca).

  

$$\mathrm{\Delta }\left[\%C\right]={\left[\%C\right]}_t-{\left[\%C\right]}_0$$(1)

where Δ[%C] is the change in [%C] of the liquid Fe–C alloy, [%C]_t is the [%C] at time t, and [%C]₀ is the initial [%C].

Fig. 3. Comparison of the dissolution of coke and coke analogue containing coke ash in liquid Fe–C alloy at 1500°C, expressed as the change in the metal [%C] from the initial composition with time. Comparison of (a) coke 114 and analogue 114; (b) coke 131 and analogue 131 (high Fe); and (c) coke 138 and analogue 138.

It can be seen in Fig. 3 that the dissolution of all three cokes was very closely replicated by the coke analogue containing its coke ash. This result agrees with previous dissolution studies using the coke analogue.^34,35) In those analogue dissolution studies, an analogue was built to replicate (mimic) a specific industrial coke’s mineralogy and then its dissolution behaviour in iron was compared with the industrial coke. It was found that dissolution behaviour of the analogue was similar to that of the industrial coke the analogue was built to mimic.^34,35)

The close replication of the dissolution behaviour between the coke and coke analogues allows further understanding on the relative roles of the factors that affect coke dissolution. There are a number of factors that are generally considered to affect coke dissolution, including carbon structures and graphitisation, porosity, particle size, coke mineralogy, and contact area between the coke and liquid iron.

The effects of carbon structures and the graphitisation of carbon on coke dissolution in liquid iron have been widely studied.^{7,8,37,38,39,40)} Graphite has been shown to dissolve in liquid iron faster than coke, and coke analogue with no minerals.³⁴⁾ The effect of carbon structures in coke on the dissolution rate has been the source of some conjecture. Increasing structure of coke (higher L_c values) has been reported to related to increasing dissolution rate,⁷⁾ while other studies have shown that it had little effect.⁴⁰⁾

Porosity and particle size are physical coke properties that affect the contact area between the coke and liquid iron, and hence the coke dissolution rate. Increasing porosity may both increase or decrease the contact area depending on the wetting characteristics of the liquid iron on the coke, and the pore size. Decreasing the particle size will increase the number of particles (for a given mass of coke), increasing the total surface area of these smaller particles, for a given volume (or mass) of minerals.

The effect of coke mineralogy on the dissolution behaviour is often described in terms of the formation of a mineral layer at the coke iron interface.^2,3,4,5) The formation of a layer of minerals at the coke-iron interface decreases the contact area between the coke and iron, slowing the dissolution of the coke. In addition to the formation of a mineral layer, coke mineralogy is also known to strongly affect the reactivity of coke.^13,30,31,32) These mineral effects on coke reactivity have been widely studied for coke reactivity with CO₂, but less so for their effects on dissolution.

Through the use of coke analogue, the effects of several of these factors were eliminated or minimised. Coke analogue samples are prepared from high purity graphite and Bakelite, and undergoes a standard, repeatable firing regime. This has been shown to give coke analogue very similar carbon structures and types, which are largely repeatable between samples and batches.⁴¹⁾ The use of the coke analogue largely eliminates carbon structural effects on coke dissolution as a variable between the three samples (analogues 114, 131 and 138).

The porosity of the coke analogue is controlled at a constant value between batches (29.3 vol% ± 2.3%). Hence, by using the coke analogue for the dissolution tests, the porosity has largely been eliminated as a variable. Both the coke and coke analogue samples for the dissolution tests were crushed to the same size range (0.5 to 2 mm). Through the use of this common size fraction for all of the tests, the particle size variable should also have been largely eliminated.

This leaves coke mineralogy and the contact area as factors that were (or may have been) changed between the three coke analogue samples that may affect the dissolution behaviour. As the results from the coke analogue closely replicated the results from the coke, the differences between the three cokes are also due to the differences in their mineralogy and the effects of the different mineralogy on contact area, while the effect of the different carbon structures of the cokes is comparatively minor.

The results from the current study are not sufficient to distinguish between the two mineral effects on the dissolution of coke in liquid iron (i.e. the aforementioned surface area effect and the mineral effect on coke reactivity). The formation of a mineral layer at the coke-iron interface was not assessed in the current work, but will be a focus for further study. The effects of many of the components of the coke mineral matter on reactivity are well known. For the main components of the mineral matter in the three cokes studied, it is known that:

• SiO₂ decreases the reactivity of coke;¹⁴⁾

• Al₂O₃ has little effect on the reactivity of coke;^13,14) and

• Fe₂O₃ and CaO both increase the reactivity of coke, with CaO giving a larger increase.^{13,14,26,42,43)}

These effects match well with the dissolution behaviour of the three cokes. Coke 114 with the highest SiO₂ content in the mineral matter had the slowest dissolution (see Fig. 3(a)). Coke 138 with the highest CaO content had the fastest dissolution (see Fig. 3(c)). Coke 131 with the highest Fe₂O₃ content had faster dissolution than coke 114 (high Si), but slower than coke 138 (high Ca). The effects of the minerals were also investigated as the trends of the mineral matter composition against coke dissolution rate. In Fig. 4, the final (after 180 minutes) Δ[%C] for each coke has been used to distinguish between the dissolution rates of the three cokes and have been plotted against the major components of the coke mineral matter. In Fig. 4, the composition of the mineral matter is presented as the mass fraction of the minerals in the coke (or coke analogue) to normalise the slight differences in the amount of mineral matter in the coke samples (10.8–12.0 mass%, see Table 1) and the coke analogue samples (10 mass%). The mass fraction of the mineral components is defined in Eq. (2).

  

$$\begin{array}{l}mass\mathrm{\hspace{0.17em} } fraction\mathrm{\hspace{0.17em} } i=\\ \frac{\left(mass\%\mathrm{\hspace{0.17em} } ash\mathrm{\hspace{0.17em} } in\mathrm{\hspace{0.17em} } sample\right)}{100}\times \frac{\left(mass\%\mathrm{\hspace{0.17em} } i\mathrm{\hspace{0.17em} } in\mathrm{\hspace{0.17em} } ash\right)}{100}\end{array}$$(2)

where i is the ash component.

Fig. 4. Correlations between coke dissolution, represented by Δ[%C] in the liquid iron and mass fraction of ash components in the samples.

From Fig. 4, it can be seen that there are clear trends between coke dissolution and the SiO₂ and CaO content in the coke, but not for the Fe₂O₃ and Al₂O₃ contents of the coke. The effects of SiO₂ and CaO on coke dissolution were in line with that expected by their known effects on coke reactivity.^13,14,42,43) This was also true for Al₂O₃, which is known to have little effect on coke reactivity.^13,14) However, it may have been expected that increasing Fe₂O₃ in the coke minerals would have led to an increase in coke dissolution. From Fig. 4(c), it can be seen that this was not the case. It is known at “Fe₂O₃” in the coke minerals would be present as metallic Fe in the coke at the experimental temperature of 1500°C.²⁶⁾ Hence, the effect on reactivity of “Fe₂O₃” in the coke minerals is actually that of metallic Fe. The liquid Fe bath is in contact with the coke particles during dissolution and may also affect the coke reactivity in the same manner as iron present in the minerals. This contact with metallic iron in the bath may mask any effect of iron in the coke minerals on coke dissolution.

The composition range of the coke ashes examined was limited, and the trends found here in the dissolution rate with SiO₂, Al₂O₃, Fe₂O₃ and CaO may not continue much outside this composition range. It is possible, and perhaps likely, that ash compositions towards the extremities of SiO₂, Al₂O₃, Fe₂O₃ and CaO content in the ash may have different effects on the dissolution rates. However, as metallurgical cokes for use in the blast furnace are typically made from a limited number of coking coals, it should also be noted that range of the ash compositions in cokes would be expected to be restricted.

3.2. Change in Coke Structure during Dissolution

The dissolution of cokes in liquid Fe–C alloys was expected to be affected by the carbon structures within the cokes, with more graphitic carbon reported to dissolve more quickly in a liquid Fe–C alloy.^2,13) However, the coke structure was found to have little effect on the dissolution of the three cokes, despite coke 138 (high Ca) being formed from very different coal than cokes 114 (high Si) and 131 (high Fe).

Despite this, the carbon structures in the coke were studied, for two main reasons. The first was to attempt to understand why the carbon structures in the cokes seemingly had little effect on dissolution. The second was that the carbon structures in the coke and their changes during dissolution need to be characterised and understood if the mineral effect on coke dissolution is to be understood in the absence of carbon structural considerations.

• There are a number of factors that need to be considered to understand the effects of the carbon structures on coke on dissolution,

• IMDC in coke is typically more graphitic than RMDC in unreacted coke,^1,17)

• coke becomes more graphitic at high temperatures,^1,17,44)

• while both IMDC and RMDC both become more graphitic upon heating, they do so at different rates,^17,27,28,29)

• iron is known to catalyse the graphitisation of carbonaceous materials.^26,41) Does the presence of the liquid Fe–C bath affect the graphitisation of the coke, and subsequently the dissolution rate?

The carbon structures in coke 131 (high Fe) and coke 138 (high Ca) samples before and after dissolution were measured using Raman spectroscopy. The Raman data is summarised in Fig. 5, where the spectra were analysed and presented as a ratio of the intensities of the D and G peaks (I_(D)/I_(G)) plotted against the ratio of the intensities of the V valley and G peak (I_(V)/I_(G)). Lower values of I_(V)/I_(G) indicate more graphitic carbon, and lower values of I_(D)/I_(G) indicate more ordered carbon. In the post-reaction data, coke particles were analysed in two regions within the sample. The “near” region included coke particles <1 mm from the coke iron interface. The “away” region included coke particles >2 mm from the coke interface.

Fig. 5. Raman spectroscopy of coke after dissolution in Fe-2%C at 1500°C. Coke 131 (high Fe): (a) Unreacted; and (b) after 120 min dissolution. Coke 138 (high Ca, SHO): (c) Unreacted; and (d) after 120 min dissolution. The data are separated into points identified as IMDC and RMDC, and in parts (b) and (d), points near to (<1 mm) and away from (>2 mm) the coke-iron interface.

The two cokes characterised by the Raman spectroscopy were different in their unreacted state (Figs. 5(a) and 5(c)). In coke 131 (high Fe), the data for IMDC was spread out over a much larger range than the RMDC. For coke 138 (high Ca, SHO), both the IMDC and RMDC were tightly grouped, with the IMDC more graphitic than the RMDC. Overall, coke 131 (high Fe) was more graphitic than coke 138 (high Ca, SHO).

However, it can be seen from Fig. 5 that there was a general trend with time of both cokes becoming more graphitic (moving to the left on the x-axis) and more ordered (moving slightly down on the y-axis). This indicates that the coke had become more graphitic during the dissolution tests at 1500°C, with more of an sp² character to the carbon bonds. Increasing graphitisation of the samples was expected at 1500°C, as it is well known that cokes become more graphitic at high temperatures.^{17,18,19,20,21,22,23,24,25,26,44)} It was less expected that the carbon structures in the two cokes would tend towards a single range, with the data for both coke 131 (Fig. 5(b)) and coke 138 (Fig. 5(d)) mostly overlapping.

These changes in the graphitisation of the two cokes with time during the dissolution tests at 1500°C are summarised in Fig. 6. In Fig. 6, the level of graphitisation of the cokes was represented by the I_(V)/I_(G) ratio, where more graphitic carbon has a lower I_(V)/I_(G) value.

Fig. 6. Change in graphitisation of cokes with time. Coke 131 (high Fe): (a) IMDC; and (b) RMDC. Coke 138 (high CA, SHO): (c) IMDC; and (d) RMDC.

In both coke 131 (high Fe) and coke 138 (high Ca, SHO), the RMDC appeared to change (become more graphitic) more quickly than the IMDC. There was a change in the RMDC for both cokes going from the unreacted coke to that after 30 minutes at 1500°C. There was little further change in the RMDC going from 30 to 120 minutes. In addition, for the RMDC in both cokes, there was little difference between near to or away from the coke-iron interface at 30 minutes (Figs. 6(b) and 6(d)).

The IMDC in both cokes appears to have become graphitic more slowly than the RMDC. While the IMDC in the coke became more graphitic at 30 minutes (i.e. the I_(V)/I_(G) ratio in Figs. 6(a) and 6(c) decreased), it became further graphitised at 120 minutes. The behaviour of the IMDC and RMDC is consistent with the results of Xing et al.,^17,27,28,29) who reported that while IMDC and RMDC both become more graphitic upon heating, they do so at different rates.

The IMDC in both cokes also appeared to possibly show different behaviour depending on its proximity to the coke-iron interface. The IMDC away from the liquid iron appeared to have slower graphitisation than that near the surface of the liquid iron. It can be seen for both cokes that at 30 minutes, while the IMDC away from the liquid iron surface (the solid triangles in Figs. 5(a) and 5(c)) had become more graphitic, it had changed less than the IMDC situated near the liquid iron surface (the clear triangles). This indicates that the proximity of the coke particles to the liquid iron bath may have affected the graphitisation of the IMDC, but not the RMDC.

Metallic iron is known to promote the graphitisation of coke at elevated temperatures.^{20,26,41,45,46)} Coke particles near the coke-iron interface are more likely to have been contact with the liquid Fe bath during the dissolution test. It is possible that this contact with the liquid Fe-C alloy may have accelerated the graphitisation of the IMDC. It is also possible that as the changes to the RMDC occur more quickly, any effect caused by the proximity to the liquid Fe–C alloy had already occurred before 30 minutes, and thus not resolved in this study. Improving the resolution of the change in RMDC and any possible iron effects could be studied further in future work.

The coke carbon structures for both cokes appear to tend to similar values for both coke 131 (high Fe) and coke 138 (high Ca, SHO), despite the different starting points. This may mean that the effects of the carbon structures on coke dissolution have tended to a common starting value or normalised during the test, and the differences in the carbon structures between the two cokes (and the IMDC and RMDC within the two coke) were minimal. The similarity of the carbon structures after dissolution may be seen to approximate to the elimination of carbon structure as a variable, as through the use of the coke analogue. Given this phenomenon, a key point to make is that a wider range of carbon structures needs to be considered if carbon structural effects on coke dissolution are to be elucidated.

As the carbon structures within the IMDC and RMDC were normalised to a common value, it may be expected that these would both dissolve at similar rates as found during this study. However, as the carbon structures in the two cokes tended to a common value, the effects of the carbon structures on coke dissolution were effectively not assessed in the experimental programme. Whether this finding is applicable to a wider range of coke types is worth investigating in future studies.

The findings of this study were also consistent with other studies^7,47,48) that carbon structure had little effect on the dissolution behaviour of low-graphitised carbon materials such as coal, char or coke. Graphite was found to carburise liquid metal more quickly than these low-graphitised carbon materials. However, there was little difference in the dissolution behaviour between the low-graphitised carbon materials despite having different initial carbon structures.

4. Conclusions

To help better understand how coke performance in the blast furnace could be related to its properties, coke dissolution in liquid iron was studied, with a focus on the mineral and carbon structure effects on coke dissolution. The dissolution of three cokes with very different mineralogies in liquid iron was measured, while the effects of carbon structures were assessed by characterising quenched samples. In addition to the three cokes, three coke analogue samples each containing the ash from one of the cokes were also studied.

The three cokes were found to have distinctly different dissolution rates. The rate of dissolution of the cokes were ranked in order of increasing dissolution rate as analogue/coke 114 (high Si) < analogue/coke 131 (high Fe) < analogue/coke 138 (high Ca).

The dissolution of coke analogue was found to closely replicate the dissolution of the three cokes. Through the use of the coke analogue, carbon structure, porosity and particle size were largely eliminated as variables, meaning that the differences in the dissolution rates of the samples was primarily due to differences in the minerals of the three cokes. This finding also indicates that carbon structure appeared to have little effect on the dissolution of coke in liquid iron.

Assessment of the carbon structures in the cokes during dissolution at 1500°C showed that the cokes became more graphitic with time at high temperature. In addition, though the coke started off with different carbon structures they tended to similar values dominated by the temperature effect on graphitisation.

While the formation of any mineral layer was not studied, it is likely that mineral effects on carbon reactivity were at least partially responsible for the different dissolution rates of the cokes. It was found that decreasing SiO₂ and increasing CaO in the ash led to an increase in dissolution, having similar effects on reactivity as reported with their effects on coke reactivity with CO₂.

Acknowledgements

This study was funded by ACARP. Coke 138 was prepared at the University of Newcastle with the assistance of Arash Tahmasebi. The Raman data for the unreacted coke 131 samples were measured at the University of New South Wales with the assistance of Pramod Koshy.

References

• 1)   P.  Loison,  P.  Foch and  A.  Boyer: Coke quality and production, Butterworth & Co. Press, London, (1998), 1.
• 2)   M. W.  Chapman,  B. J.  Monaghan,  S. A.  Nightingale,  J. G.  Mathieson and  R. J.  Nightingale: Metall. Mater. Trans. B, 39B (2008), 418. https://doi.org/10.1007/s11663-008-9145-7
• 3)   M. W.  Chapman,  B. J.  Monaghan,  S. A.  Nightingale,  J. G.  Mathieson and  R. J.  Nightingale: ISIJ Int., 47 (2007), 973. https://doi.org/10.2355/isijinternational.47.973
• 4)   H. W.  Gudenau,  J. P.  Mulanza and  D. G. R.  Sharma: Steel Res., 61 (1990), 97. https://doi.org/10.1002/srin.199000309
• 5)   S.  Orsten and  F.  Oeters: Process Technology Proc., Iron and Steel Society, Washington DC, (1996), 143.
• 6)   M. B.  Mourao,  G. G.  Krishna Murthy and  J. F.  Elliot: Metall. Mater. Trans. B, 24B (1993), 629. https://doi.org/10.1007/BF02673178
• 7)   C.  Wu and  V.  Sahajwalla: Metall. Mater. Trans. B, 31B (2000), 215. https://doi.org/10.1007/s11663-000-0147-3
• 8)   V.  Sahajwalla,  F.  McCarthy,  S. T.  Cham,  J.  Hart,  R.  Sakurovs and  N.  Saha-Choudhury: Influence of Carbonaceous Materials on Carburisation of Liquid Iron, International Blast Furnace Lower Zone Symposium, AusIMM, Wollongong, Australia, (2002), 27.
• 9)   F.  McCarthy,  V.  Sahajwalla,  J.  Hart and  N.  Saha-Choudhury: Metall. Mater. Trans. B, 34B (2003), 573. https://doi.org/10.1007/s11663-003-0026-9
• 10)   F.  McCarthy,  R.  Khanna,  V.  Sahajwalla and  N.  Simento: ISIJ Int., 45 (2005), 1261. https://doi.org/10.2355/isijinternational.45.1261
• 11)   C. S.  Nguyen,  K.-I.  Ohno,  T.  Maeda and  K.  Kunitomo: ISIJ Int., 56 (2016), 1325. https://doi.org/10.2355/isijinternational.ISIJINT-2015-739
• 12)   M. M.  Sun,  J. L.  Zhang,  K. J.  Li,  H. T.  Li,  Z. M.  Wang,  C. H.  Jiang,  S.  Ren,  L.  Wang and  H.  Zhang: J. Met., 72 (2020), 2174. https://doi.org/10.1007/s11837-020-04048-0
• 13)   W. W.  Gill,  N. A.  Brown,  C. D. A.  Coin and  M. R.  Mahoney: 44th ISS-AIME Ironmaking Conference, ISS, Warrendale, PA, (1985), 233.
• 14)   M. H.  Reid,  M. R.  Mahoney and  B. J.  Monaghan: ISIJ Int., 54 (2014), 628. https://doi.org/10.2355/isijinternational.54.628
• 15)   M.  Grigore,  R.  Sakurovs,  D.  French and  V.  Sahajwalla: ISIJ Int., 46 (2006), 503.
• 16)   P.  Bennett,  A.  Reifenstein,  G.  O’Brien and  B.  Jenkins: Coke reactivity and Characterization, ACARP report C12057, ACARP, Brisbane, (2018), 22.
• 17)   X.  Xing,  H.  Rogers,  G.  Zhang,  K.  Hockings,  P.  Zulli and  O.  Ostrovski: ISIJ Int., 56 (2016), 786. https://doi.org/10.2355/isijinternational.ISIJINT-2015-704
• 18)   M.  Kawakami,  T.  Karato,  T.  Takenaka and  S.  Yokowama: ISIJ Int., 45 (2005), 1027. https://doi.org/10.2355/isijinternational.45.1027
• 19)   M.  Kawakami,  H.  Kanba,  K.  Sato,  T.  Takenaka,  S.  Gupta,  R.  Chandratilleke and  V.  Sahajwalla: ISIJ Int., 46 (2006), 1165. https://doi.org/10.2355/isijinternational.46.1165
• 20)   W.  Wang,  K. M.  Thomas,  R. M.  Poultney and  R. R.  Willmers: Carbon, 33 (1995), 1525. https://doi.org/10.1016/0008-6223(95)00097-W
• 21)   L.  Smedowski,  M.  Krzesinska,  W.  Kwasny and  M.  Kozanecki: J. Am. Chem. Soc., 25 (2011), 3142. https://doi.org/10.1021/ef200609t
• 22)   P. D.  Green,  C. A.  Johnson and  M. K.  Thomas: Fuel, 62 (1983), 1013. 10.1016/0016- https://doi.org/10.1016/0016-2361(83)90134-5
• 23)   S.  Dong,  N.  Paterson,  S. G.  Karazian,  D. R.  Dugwell and  R.  Kandiyoti: Energy Fuels, 21 (2007), 3446. https://doi.org/10.1021/ef7003656
• 24)   S.  Dong,  P.  Alvarez,  N.  Paterson,  D. R.  Dugwell and  R.  Kandiyoti: Energy and Fuels, 23 (2009), 1651. https://doi.org/10.1021/ef800961g
• 25)   G.  Rantitsch,  A.  Bhattacharyya,  J.  Schenk and  N. K.  Lunsdorf: Int. J. Coal Geol., 130 (2014), 1. https://doi.org/10.1016/j.coal.2014.05.005
• 26)   O. A.  Aladejebi,  B. J.  Monaghan,  M. H.  Reid,  M.  in het Panuis and  R. J.  Longbottom: Steel Res. Int., 88 (2017), 1. https://doi.org/10.1002/srin.201700039
• 27)   X.  Xing,  G.  Zhang,  H.  Rogers,  P.  Zulli and  O.  Ostrovski: Metall. Mater. Trans. B, 45B (2014), 106. https://doi.org/10.1007/s11663-013-0002-y
• 28)   X.  Xing,  H.  Rogers,  G.  Zhang,  K.  Hockings,  P.  Zulli,  A.  Deev,  J.  Mathieson and  O.  Ostrovski: Fuel Process. Technol., 157 (2017), 42. https://doi.org/10.1016/j.fuproc.2016.11.009
• 29)   X.  Xing,  H.  Rogers,  P.  Zulli,  K.  Hockings and  O.  Ostrovski: Fuel, 237 (2019), 775. https://doi.org/10.1016/j.fuel.2018.10.069
• 30)   R. J.  Longbottom,  B. J.  Monaghan,  A. A.  Chowdhury,  M. H.  Reid,  G.  Zhang,  M. R.  Mahoney and  K.  Hockings: ISIJ Int., 56 (2016), 1553. https://doi.org/10.2355/isijinternational.ISIJINT-2015-597
• 31)   R. J.  Longbottom,  Z.  Wang,  K.  Hockings and  B. J.  Monaghan: ISIJ Int., 61 (2021), 2745. https://doi.org/10.2355/isijinternational.ISIJINT-2021-336
• 32)   R. J.  Longbottom,  J.  Perkins,  G.  O’Brien and  B. J.  Monaghan: ISIJ Int., 63 (2023), 282. https://doi.org/10.2355/isijinternational.ISIJINT-2022-396
• 33)   A. S.  Jayasekara,  B. J.  Monaghan and  R. J.  Longbottom: Fuel, 154 (2015), 45. https://doi.org/10.1016/j.fuel.2015.03.053
• 34)   B. J.  Monaghan,  S. A.  Nightingale and  M. W.  Chapman: Steel Res. Int., 81 (2010), 829. https://doi.org/10.1002/srin.201000143
• 35)   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. https://doi.org/10.1002/srin.201100042
• 36)   R. J.  Longbottom and  B. J.  Monaghan: High Resolution Micro-CT 3D Assessment of Coke Gasification, ACARP report C29078, ACARP, Brisbane, (2021), 20.
• 37)   C.  Wu and  V.  Sahajwalla: Metall. Mater. Trans. B, 31B (2000), 243. https://doi.org/10.1007/s11663-000-0043-x
• 38)   Y.  Shigeno,  M.  Tokuda and  M.  Ohtani: Trans. Jpn. Inst. Met., 26 (1985), 33. https://doi.org/10.2320/matertrans1960.26.33
• 39)   V.  Sahajwalla and  R.  Khanna.: Scand. J. Metal., 29 (2000), 114. https://doi.org/10.1034/j.1600-0692.2000.d01-13.x
• 40)   S. T.  Cham,  V.  Sahajwalla,  R.  Sakurovs,  H.  Sun and  M.  Dubikova.: ISIJ Int., 44 (2004), 1835. https://doi.org/10.2355/isijinternational.44.1835
• 41)   O. A.  Aladejebi: Effect of Mineral on Coke Analogue Carbon Bonding and Reactivity, doctoral thesis, UOW, (2017), 110.
• 42)   A. S.  Jayasekara,  B. J.  Monaghan and  R. J.  Longbottom: Fuel, 225 (2018), 18. https://doi.org/10.1016/j.fuel.2018.03.138
• 43)   A. S.  Jayasekara,  B. J.  Monaghan,  R. J.  Longbottom,  M. R.  Mahoney and  K.  Hockings: Fuel, 264 (2020), Article 116818. https://doi.org/10.1016/j.fuel.2019.116818
• 44)   B. J.  Monaghan,  R.  Nightingale,  V.  Daly and  E.  Fitzpatrick: Ironmak. Steelmak., 35 (2008), 38. https://doi.org/10.1179/030192307X233377
• 45)   V.  Lyalyuk,  V.  Sokolova,  E.  Shmeltser,  D. Y.  Timofeeva and  V.  Beryeza: Coke and Chemistry, 57 (2014), 238. https://doi.org/10.3103/S1068364X14060052
• 46)   S.  Gornostayev and  J.  Harkki: Carbon, 45 (2007), 1145. https://doi.org/10.1016/j.carbon.2007.02.033
• 47)   M. M.  Sun,  J. L.  Zhang,  K. J.  Li,  H. T.  Li,  Z. M.  Wang,  C. H.  Jiang,  S.  Ren,  L.  Wang and  H.  Zhang: J. Met., 72 (2020), 2174. https://doi.org/10.1007/s11837-020-04048-0
• 48)   K.  Ohno,  A.  Babich,  J.  Mitsue,  T.  Maeda,  D.  Senk and  H. W.  Gudenau, ISIJ Int., 52 (2012), 1482. https://doi.org/10.2355/isijinternational.52.1482