Effects of Blast Furnace Representative Temperatures and Gas Compositions on Coke Reactivity

Abstract

The effects of coke mineralogy on coke reactivity under conditions representative of the blast furnace were studied. Coke samples were reacted under coke reactivity index (CRI) like conditions (1100°C, 100% CO₂) and at higher temperatures and atmospheres designed to replicate conditions lower in the blast furnace.

The effects of the minerals on reactivity changed as the temperature increased and the atmosphere was modified. At the lower temperatures investigated (1100–1350°C) with CO₂ present in the gas, gasification of the coke by CO₂ dominated. The effects of minerals in the coke on gasification by CO₂ under these conditions were similar to their reported effects on reactivity in the CRI test. At the highest temperature investigated, (1600°C), with no CO₂, the mineral-carbon reactions dominated. The main reaction was the reduction of the silica in the coke. These results show that when coke-gasification is dominant, CRI data can be related to conditions beyond the temperature and gas environment the data were obtained, to the higher temperatures and less oxidising conditions of the blast furnace. At higher temperatures, mineral-carbon reactions are dominant, and more data in addition to that of the CRI, may be required to understand coke behaviour in the blast furnace.

IMDC and RMDC within the coke were identified with the aid of the CGA technique, and the changes in the carbon structures within the coke studied using Raman spectroscopy. The carbon structures within the coke became more graphitic at 1600°C, with the change in RMDC greater than that in IMDC.

1. Introduction

Metallurgical coke is a key raw material in blast furnace ironmaking, fulfilling three roles. 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.^1,2) As such, improving key properties of coke will improve blast furnace productivity, lower the consumption of coke and decrease the CO₂ emissions associated with the iron and steel industry.

The performance of metallurgical coke in the blast furnace is related to its hot strength and reactivity. The coke reactivity index (CRI) and the coke strength after reaction (CSR) tests are the principal tests used by industry to assess coke reactivity.^3,4) However, it is known that using CRI/CSR values to assess coke performance at temperatures well in excess of 1100°C can be problematic.

There are a number of coke characteristics that affect its performance in the blast furnace performance, such as the carbon structures, coke texture, porosity (total porosity, pore size distribution, pore connectivity and morphology) and coke mineralogy.¹⁾ The focus of this study is on the coke minerals, and their effects on the reactivity of coke. Relating the behaviour of minerals and their effects on coke properties at high temperatures representative of a blast furnace to the effects at the CRI/CSR test is difficult. This is due to not only the lower temperature of the CRI/CSR test, but also the reactive atmosphere used. The atmosphere used in the CRI/CSR test (100% CO₂) is far more oxidising than what the coke would be exposed to in the lower regions of the blast furnace.^4,5,6,7,8,9) To deal with this, the gas compositions at higher temperatures need to be informed by the gas composition-temperature profile of a blast furnace.

The carbon structures within the coke area 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.^1,2) Raman spectroscopy in conjunction with microscopy has been used to the levels of graphitisation of IMDC and RMDC in coke.^{6,10,11,12,13,14,15,16,17,18)} Recent studies^6,7,8,9) 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. The gas compositions were based on van der Velden et al.⁵⁾ gas composition profile of a blast furnace. It was found that while IMDC and RMDC both become more graphitic upon heating, they did so at different rates.^6,7,8,9) This led to the strength of high CSR coke and low CSR coke crossing over as the temperature increased. The impacts of this changing graphitisation of the coke with increasing temperature on its reactivity has yet to be fully understood.

The aims of this study are the assessment of key coke-mineral interactions at high temperatures:

1. the reactions between minerals and coke,

2. the effects of minerals on coke-gas reactions, and

3. the change in the carbon types and structure within coke during reaction.

These objectives will be addressed by characterising coke samples before and after reaction at three different temperature regimes and conditions, representing both the CSR/CRI test and high temperature conditions in the blast furnace. Characterisation of the minerals will include X-ray diffraction (XRD) and scanning electron microscopy (SEM) with energy dispersive spectroscopy (EDS). Assessment of the carbon structures will use optical microscopy using the coal grain analysis (CGA) technique with both plain and polarised light, and Raman spectroscopy.

2. Experimental

2.1. Industrial Coke 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. 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 industrial cokes used is given in Table 1. Both cokes were supplied in the form of CRI lumps (19–21 mm size range).

Table 1. Data for the two coke samples.

	coke 114	coke 131
ash	12.0	11.9
volatile matter	0.3	0.3
CSR	67.1	46.3
CRI	25.6	42.7
ash composition
SiO2	55.8	45.9
Al2O3	30.3	28.8
Fe2O3	9.0	15.5
CaO	1.2	3.2
MgO	0.8	1.4
Na2O	0.5	0.2
K2O	0.7	0.8
P2O5	0.6	1.1
SO3	0.2	1.5

2.2. High Temperature Reaction Tests

Both coke samples were studied in a series of high temperature reactivity tests. Experiments up to 1350°C were conducted in a large sample TGA set up (maximum temperature 1400°C), while experiments at 1600°C were conducted in a high temperature vertical tube furnace. A schematic of the TGA set up for the fractional weight change (FWC) gasification test in the high temperature TGA equipment is given in Fig. 1. The vertical tube furnace set up for the 1600°C tests was very similar to the schematic in Fig. 1, but with the crucible supported from underneath rather than being suspended from the balance. The inner diameter of the furnace tube in each set-up was 70 mm. The temperatures and gas conditions for these experiments is summarised in Table 2. In each case, the samples were heated and cooled at 10°C/min, and held at the experimental temperature for 120 min.

Fig. 1.

Experimental arrangement for high temperature TGA tests.

Table 2. Experimental conditions used in the high temperature reaction tests for coke and coke analogue.

Isothermal reaction T (°C)	Gas at reaction T
1100	• CO2, 2 L/min
1350	• 39% CO-2% CO2-59% Ar, 2 L/min
1600	• 40% CO, 60% Ar, 2 L/min

The gases used for the high temperature tests were high purity argon (>99.999% purity, cleaned by passing it through drierite and ascarite), carbon monoxide (>99.7% purity, cleaned with drierite) and industrial grade carbon dioxide (>99.9%, cleaned with drierite). The samples were typically 1–2 pieces CRI lumps, with a total mass in the range 8–10 g. The alumina crucible was Ø35 mm, height 65 mm and wall thickness 2 mm. 12×Ø6 mm holes were drilled around the bottom of the crucible wall in 2 layers of 6 holes each to allow gas to access the sample.

The reactivity tests of the coke samples were carried out in a thermogravimetric analysis (TGA) system, used to log the temperature and weight loss of the sample during the experiment with time. Results from the high temperature tests are expressed as fractional weight change (FWC), calculated by Eq. (1),   

$$FWC=\frac{m_t-m_i}{m_i}$$(1)

where m_t is the mass of the sample at time t, and m_i is the initial sample mass.

2.3. Characterisation of Samples

Unreacted samples, and those reacted at 1100°C and 1600°C were characterised using X-ray diffraction (XRD), optical microscopy (both plain and polarised), scanning electron microscopy (SEM) with energy dispersive spectroscopy (EDS) and Raman spectroscopy. The samples reacted at 1350°C were only characterised after reaction using XRD.

Samples for microscopy and Raman spectroscopy were vacuum impregnated using an epoxy mounting resin (Struers EpoFix), sectioned and polished to a 0.25 μm finish. The samples prepared for optical microscopy/Raman were subsequently carbon coated using a SPI carbon coater for examination in the scanning electron microscopy (SEM).

2.4. XRD

The coke ashes and selected analogues before and after gasification reaction were examined by XRD. XRD patterns were collected on a GBC-MMA X-ray diffractometer using Cu-Kα radiation in the 2θ range 20–80° at a power of 1 kW and a nickel filter. The patterns were subsequently analysed using the Traces X-ray processing software V6.

2.5. Optical and Electro-optical Microscopy

For the plain and polarised light CGA microscopy, reflectance calibrated photomicrographs of the entire sample surface were collected in air for each sample in reflected white light (LED light source) at 200× magnification with a digital pixel resolution of 0.32 μm using a Zeiss Axio-Imager microscope fitted with a Zeiss HrC 14-bit colour camera. Zeiss ZEN Lite software was used to process and export captured images for further processing.

SEM images and EDS composition analysis were obtained using a JEOL JSM 6490 LV SEM with an Oxford Instruments X-Max^N 80 EDS system.

2.6. Raman Spectroscopy

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.

3. Results and Discussion

3.1. Choice of Reaction Conditions

Coke and coke analogue samples were reacted at range of temperatures and gas compositions as listed in Table 2. Given the importance of the CRI test in characterising coke reactivity, the test conditions that replicated these were used as a reference in understanding and evaluation of the effects of the high temperature experimental test conditions on coke. The high temperature experimental test conditions (gas composition and temperature) were chosen to replicate the lower zone of a blast furnace. Using this CRI comparison with the high temperature work approach offers the possibility of better understanding and prediction of the behaviour of cokes in use in a blast furnace.

The temperatures of 1350°C and 1600°C were chosen as they can be considered representative of the cohesive zone and hearth areas of the blast furnace respectively. 1100°C was chosen as it represents the temperature of the standard CSR/CRI test. The experimental conditions at 1100°C (100% CO₂) were similar to those in the standard CSR/CRI test. However, the experimental set-up is different. As such, the experimental conditions used in this study are referred to as pseudo-CRI conditions.

In a blast furnace, in general it can be stated that coke is exposed to higher temperatures and less oxidising atmospheres the lower it is in a blast furnace. A notable exception to this is the raceway area, were coke is combusted with oxygen enriched air. In this study, the less oxidising atmosphere is accounted for by a lower CO₂ in gas in the higher temperature experiments.

The atmospheres used at the high temperature conditions were informed by the work of Xing et al.^6,7,8,9) and van der Velden et al.⁵⁾ blast furnace/coke performance studies, with the substitution of Ar for N₂. As both gases can be considered to be inert (true for Ar, largely true for N₂ at the experimental temperatures) this substitution was considered not to be significant in the study of the behaviour of the minerals in the cokes.

3.2. Mineral Reactions under Blast Furnace Conditions

Minerals in coke under blast furnace conditions have several effects on the reactions that take place. It is well known that minerals in coke have an effect on the coke (carbon)-gas reactions, especially under more oxidising conditions.^19,20,21) The minerals can also react with the carbon and/or the gas directly. These all play a part in determining the effect of the coke minerals on the reactivity of coke under blast furnace conditions.

The reactivity of the coke samples under blast furnace conditions may be summarised by their FWC, as summarised in Fig. 2(a). A general trend in the FWC shown in Fig. 2(a) is that there was less weight change for the conditions representing the lower zones of the blast furnace than in the pseudo-CRI conditions. This may be expected based on the how oxidising the atmospheres used in each case were, with less weight change at the less oxidising conditions.

Fig. 2.

(a) FWC for coke samples after the high temperature experiments; (b) FWC curves for coke under pseudo-CRI conditions (1100°C, 100% CO₂).

Another trend is that coke 131 was, in general, more reactive (more negative FWC) than coke 114, especially at the lower temperature, more oxidising conditions. To better illustrate the difference between the two cokes, the FWC curves for the tests performed on the cokes and coke analogues with coke ash at the pseudo-CRI conditions (1100°C, 100% CO₂) are shown in Fig. 2(b). Coke 131 was much more reactive than coke 114 under these conditions.

The mineral phases that were formed during the reactions were characterised by XRD, with any changes in morphology checked by SEM-EDS. The XRD patterns from the coke samples before and after reaction are shown in Fig. 3. One feature immediately noticeable in the XRD patterns of the two coke samples is that the main graphite peak (002 peak, 2θ = ~26°) became stronger and sharper as the temperature increased. As the peak became sharper, it also shifted position, becoming closer to the expected position for graphite at 2θ = 26.7° for the Cu-Kα wavelength. Such a change in the graphite peak is associated with increasing graphitisation of the coke. Increasing graphitisation of the coke with increasing temperature was expected,^6,11,18,22) and was also investigated further in this using Raman spectroscopy.

Fig. 3.

XRD patterns of coke samples after reaction at high temperatures. (a) Coke 131; and (b) coke 114. Phase key: G – graphite (C); Q – quartz (SiO₂); C – cristobalite (SiO₂); M – mullite (Al₆Si₂O₁₃); α – metallic iron (Fe, ferrite); E – cementite (Fe₃C); N – magnetite (Fe₃O₄); S – silicon carbide (SiC).

Both cokes showed similar changes in their mineralogy. The main change was observed in the silicon bearing minerals. Quartz was found in both unreacted coke samples. Upon heating to 1100°C, some of the quartz changed into cristobalite (one of the high temperature polymorphs of silica). Further heating to 1350°C led to the formation of small amounts of SiC. Increasing the temperature to 1550°C and 1600°C increased the amount of the SiC formed.

After heating the two coke samples in the oxidising CRI conditions, the mineral peaks were more pronounced, especially in the case of mullite. There was also evidence of oxidation of the iron-bearing minerals, with magnetite peaks found in the XRD patterns. These magnetite peaks appeared to be stronger in the XRD pattern for coke 131 (with the higher Fe content in the ash).

To further investigate the changes to the minerals in the cokes, coke 131 was extensively characterised using SEM-EDS. The microstructures observed within coke 131 samples before and after reaction are described in Table 3. Typical SEM micrographs of the coke 131 samples are shown in Fig. 4. In the unreacted coke 131 sample, the minerals were a predominantly aluminosilicates, with some Fe-bearing minerals, as shown in Fig. 5. In general, at lower temperatures (up to 1100°C) despite the reactions of some of the minerals there was little change to the distribution of the minerals in the coke samples.

Table 3. Summary of optical and electro-optical characterisation of coke samples.

Sample	Mineral characterisation by SEM-EDS	Carbon structure characterisation by CGA
unreacted	Mineral matter was a mixture of < 20 μm and > 100 μm particles. There was no relation between the mineral size or distribution with the IMDC and RMDC.	The microstructure was large IMDC (> 500 μm) in a matrix of RMDC. There was little difference between the edge, annulus and centre of the unreacted sample.
	Most minerals were aluminosilicates, often with a complex composition (containing Mg, P, S, Na, K).	The RMDC regions appeared to contain much of the porosity, with a large size range (10–1000 μm) observed. Finer pores were present in the IMDC regions.
	The aluminosilicate particles often encompassed other minerals, such as quartz particles and Fe bearing mineral particles. Fe bearing minerals were found to be either metallic Fe or sulfides.
	Ca bearing sulfides and phosphates were also observed.
1100°C	The mineral distribution showed little change with reaction.	The interior of the sample was similar to the unreacted sample.
	In IMDC regions nearer the periphery of the coke the minerals were more randomly oriented than in the unreacted coke. This is likely related to carbon loss from the coke.	Carbon loss was apparent in the IMDC regions nearer the periphery of the coke.
	There was little change in the mineral composition in the sample.
1600°C	Large changes in the composition of the mineral phases were found after reaction at 1600°C.	The carbon structural features appeared similar to that in the unreacted sample.
	Si-rich and Al-rich minerals were often separate, rather than appearing as aluminosilicates. The Si-rich minerals appeared to be in part spread along the pore walls (e.g. Fig. 7).	There were no obvious changes in the observed structures.
	Fe-rich particles were largely metallic Fe*, with a rounded morphology, indicating that they may have been molten at some point.
	Ca-rich minerals were small and appeared largely separate from the other mineral particles.
	Mg was coincident with the Al-rich minerals.

*It should be noted that due to saturation of the back-scattered detector, in some figures (e.g. Figs. 4(e) and 4(f), Fig. 7), the Fe-rich particles are shown as dark features, rather than as bright features.

Fig. 4.

Typical back-scattered electron micrographs of coke 131 samples before and after reaction. (a) and (b) Unreacted sample; (c) and (d) after reaction at 1100°C in CO₂; and (e) and (f) after reaction at 1600°C in Ar–CO. (a), (c) and (e) used contrast-brightness settings that emphasised the carbon structures and porosity in the coke; and (b), (d) and (f) used contrast-brightness settings emphasised the minerals in the coke.

Fig. 5.

EDS elemental maps of a typical region within an unreacted coke 131 sample. The field of view matches the white box given in Fig. 4(a). (Online version in color.)

At higher temperatures (up to 1600°C), in addition to the reaction of the minerals, their distribution within the cokes had also significantly changed (Figs. 4(f) and 6). One noted change was the separation of the Al- and Si-bearing minerals. The Si-bearing minerals showed possible agglomeration and also appeared to have possibly distributed around pore surfaces. The Fe-bearing minerals show evidence (rounding) or having been liquid or partially liquid at the experimental temperature.

Fig. 6.

EDS elemental maps of a typical region within a coke 131 sample after reaction at 1600°C in CO–Ar. The field of view matches the white box given in Fig. 4(e). (Online version in color.)

In addition to the characterisation of the minerals, the carbon structures within the coke, and their changes, were characterised. This was carried out using high resolution light optical microscopy over the whole sample (using the CSIRO CGA technique^23,24)) and by Raman spectroscopy of specific features within the samples. Raman spectroscopy is an established technique for examining carbon structure.^{6,7,8,9,10,11,12,13,14,15,16,17,18)} Raman spectroscopy was used to assess the carbon structures in the samples and changes with reaction at high temperature. The Raman spectroscopy results for an unreacted coke 131 sample, and after reaction at 1100°C and 1600°C are shown in Fig. 7. In Fig. 7, I_(D), I_(G) and I_(V) are the intensities of the D peak (band position ~1335 cm^–1) and G peak (band position ~1590 cm^–1) and the V “valley” (band position ~1480 cm^–1) respectively in the Raman spectra. The relative intensities of the D and G peaks and the valley V are indicative of the proportions of sp² and sp³ carbon bonds within the coke. Typical CGA microscopy images of the same samples are shown in Fig. 8. The CGA data was used to help identify the carbon structures in the coke analysed by Raman spectroscopy as IMDC or RMDC.

Fig. 7.

Raman data from coke 131 sample, presented as plots of I_(D)/I_(G) against I_(V)/I_(G). (a) Unreacted; (b) after reaction at 1100°C with CO₂; and (c) after reaction at 1600°C with CO–Ar.

Fig. 8.

Mosaic CGA micrographs of coke 131-K-001 samples. Unreacted at (a) low magnification; and (b) high magnification. After reaction at 1100°C with CO₂ at (c) low magnification; and (d) high magnification. After reaction at 1600°C with CO–Ar at (e) low magnification; and (f) high magnification. (Online version in color.)

In the unreacted coke 131 sample, all points were grouped in a small I_(D)/I_(G) range, with a gentle increase in I_(D)/I_(G) with increasing I_(V)/I_(G) (Fig. 7(a)). The differences in the IMDC and RMDC also appeared to be small. The IMDC appears to be grouped at a lower I_(V)/I_(G) range than the RMDC regions, indicating that the IMDC was slightly more graphitic than the RMDC. There was little difference in the Raman data with position (i.e. the edge and middle of the coke piece were the same).

In the coke 131 sample reacted at 1100°C with CO₂, there was little change from the unreacted sample. It is possible that some graphitisation of the coke had occurred, as indicated by the small number of points to the left of the dataset in Fig. 7(b).

The I_(D)/I_(G) against I_(V)/I_(G) plots for the coke 131 sample reacted at 1600°C with the CO–Ar atmosphere (Fig. 7(c)) were very different to those for the unreacted sample and the sample reacted at 1100°C. The data points were in general grouped quite tightly in the I_(V)/I_(G) and had a much lower I_(V)/I_(G) and slightly lower I_(D)/I_(G) ratios than in the lower temperature samples, indicating that these samples had undergone significant graphitisation.

From the CGA micrographs (Fig. 8), there was little change in the carbon structures after reaction. After reaction at 1100°C in CO₂ (Figs. 8(c) and 8(d)), the main change observed was the increase in porosity near the surface of the coke particle, related to the gasification of the carbon due to reaction with CO₂. While the focus of this study was not on determining the relative reactivities of IMDC and RMDC for coke 131, it appeared that the increase in porosity on reaction was greater in the IMDC than the RMDC regions. Similar increases in porosity of IMDC compared to RMDC have been reported for other cokes.²⁾ After reaction at 1600°C in CO–Ar (Figs. 8(e) and 8(f)), the main change observed was an increase in the reflectivity of the coke. It has been reported that the reflectivity of graphitic materials increases as the graphitisation increases.²⁵⁾ It is possible that the increase in the reflectivity of the coke after reaction at 1600°C may support the Raman results showing an increase in the graphitisation of the coke after reaction at 1600°C.

3.3. Assessment of Reactions Occurring under Simulated Blast Furnace Conditions

In addition to the characterisation of the samples, the changes to the minerals under the experimental conditions were assessed through thermodynamic calculations of the carbon-mineral-gas system using MTDATA. Phase stability was assessed over at temperature range of 800–1600°C, with the coke ash composition and CO₂ representing the oxidising conditions of the CRI test and carbon reducing conditions at higher temperatures. Additional isothermal calculations were carried out at 1100°C, 1350°C and 1600°C using the same compositional endpoints. The isothermal calculations were used to establish the compositions of the phases, with particular focus on the composition of the gas phase at each temperature. These data are summarised with the observed changes in the mineralogy of the coke samples in Table 4.

Table 4. Summary of changes to minerals during the high temperature experiments.

Sample	Temperature/gas conditions	Observed changes to mineralogy (XRD)	Predicted changes to mineralogy from thermodynamics
coke 131	1100°C, 100% CO2	• Little change to mineralogy.	• Transformation of quartz to tridymite.
		• Limited oxidation of metallic Fe to Fe3O4.	• Oxidation of metallic Fe to FeO and Fe3O4.
		• Some transformation of quartz to cristobalite.
	1350°C, 39% CO- 2% CO2-59% Ar	• Some reduction of SiO2 (quartz and/or cristobalite) to SiC.	• Tridymite is stable.
	1600°C 40% CO-60%Ar	• Further reduction of SiO2 (quartz or cristobalite) to SiC.	• Reduction of SiO2 to SiC and SiO(g).
		• Fe-bearing minerals at least partially liquid.
		• Possible transformation of crystalline aluminosilicates to amorphous.
coke 114	1100°C, 100% CO2	• Similar to coke 131.	• Similar to coke 131.
	1350°C, 39% CO- 2% CO2-59% Ar	• Similar to coke 131.	• Similar to coke 131.
	1600°C 40% CO-60%Ar	• Similar to coke 131, but SiC peaks stronger (likely caused by more SiO2 in ash).	• Similar to coke 131.

From the data summarised in Table 4, the reactions that took place in the coke can be proposed. The reactions that took place in the coke under the different blast furnace conditions can be separated into three categories, gas-carbon reactions, mineral-gas reactions and mineral-carbon reactions. These reactions will be proposed, and temperature, atmosphere and mineral conditions/effects on the reactions will be discussed.

3.3.1. Gas-carbon Reactions

The Boudouard reaction (reaction 2) was found to be the dominant reaction in terms of the weight change under the pseudo-CRI conditions (1100°C, 100% CO₂) and at 1350°C in the 39% CO-2% CO₂- 59% Ar atmosphere.   

$${\mathrm{C}}_{(\mathrm{s})}+\mathrm{C}{\mathrm{O}}_{2(\mathrm{g})}=2\cdot \mathrm{C}{\mathrm{O}}_{(\mathrm{g})}$$(2)

Under the pseudo-CRI conditions, the minerals within the coke affected the reactivity of the coke with CO₂, as would be expected.^19,20,21) The reactivity of the two coke samples was also in the order that would be expected from their ash compositions, with coke 131 having a higher reactivity than coke 114 (Fig. 2(b)). Coke 131 had more iron-bearing minerals in the ash, which assuming all other factors equal, would be expected to increase the reactivity with CO₂.^18,21) Coke 114 had more silica in its ash, which would be expected to decrease the reactivity.^19,20,21) Neglecting any carbon structure/type effects here, it is likely that these differences in the iron and silica content of the ash of the two cokes at least in part led to the different reactivities of the two cokes.

At 1350°C with only 2% CO₂ in the reaction gas the reactivity of the samples was much lower. However, the order of the reactivity of the two coke samples was the same as at 1100°C with 100% CO₂. The relative reactivities of the samples was similar at 1100°C and 1350°C, indicating that the effects of the minerals on reactivity at these two conditions were similar. This is likely a result of the fact that

• there was not a large change in the minerals present in the samples (only a small amount of the reduction of silica had occurred); and

• there was still CO₂ in the reaction gas for the Boudouard reaction (Eq. (2)) to occur.

This finding does show that results from the standard CSR/CRI test is likely applicable to higher temperature conditions within the blast furnace, as long as these two conditions are applicable.

There was some evidence that the IMDC within the coke reacted preferentially to the RMDC, with qualitative analysis of the micrographs showing that less was IMDC in the periphery of the coke particles after reaction at 1100°C with CO₂ than in the unreacted sample. In Fig. 7, it can be seen that the IMDC was slightly more graphitic than the RMDC in unreacted coke 131 and in the sample after reaction at 1100°C. This more graphitic nature of the RMDC in the sample after reaction at 1100°C is seen in the large grouping of data for the RMDC to the right of the IMDC in Fig. 7(b). This agrees with previous characterisation of coke samples after reaction with CO₂, where IMDC was reported to have preferentially reacted with CO₂ in comparison to RMDC.²⁾

3.3.2. Mineral-gas Reactions

Also found to occur at the lower temperature, more oxidising pseudo-CRI conditions (1100°C, 100% CO₂) was the formation of iron oxides. The XRD patterns of the coke samples after reaction (Fig. 3) showed the formation of magnetite in the high iron coke (coke 131). The step-wise formation of magnetite from metallic iron and iron sulfide is given in reactions 3–7.   

$$\mathrm{F}{\mathrm{e}}_{(\mathrm{s})}+\mathrm{C}{\mathrm{O}}_{2(\mathrm{g})}=\mathrm{F}\mathrm{e}{\mathrm{O}}_{(\mathrm{s})}+\mathrm{C}{\mathrm{O}}_{(\mathrm{g})}$$(3)

  

$$3\cdot \mathrm{F}{\mathrm{e}}_{(\mathrm{s})}+2\cdot \mathrm{C}{\mathrm{O}}_{2(\mathrm{g})}=\mathrm{F}{\mathrm{e}}_3{\mathrm{O}}_{4(\mathrm{s})}+4\cdot \mathrm{C}{\mathrm{O}}_{(\mathrm{g})}$$(4)

  

$$3\cdot \mathrm{F}\mathrm{e}{\mathrm{O}}_{(\mathrm{s})}+\mathrm{C}{\mathrm{O}}_{2(\mathrm{g})}=\mathrm{F}{\mathrm{e}}_3{\mathrm{O}}_{4(\mathrm{s})}+\mathrm{C}{\mathrm{O}}_{(\mathrm{g})}$$(5)

  

$$2\cdot \mathrm{F}{\mathrm{e}}_3{\mathrm{O}}_{4(\mathrm{s})}+\mathrm{C}{\mathrm{O}}_{2(\mathrm{g})}=3\cdot \mathrm{F}{\mathrm{e}}_2{\mathrm{O}}_{3(\mathrm{s})}+\mathrm{C}{\mathrm{O}}_{(\mathrm{g})}$$(6)

  

$$\mathrm{F}\mathrm{e}{\mathrm{S}}_{(\mathrm{s})}+\mathrm{C}{\mathrm{O}}_{2(\mathrm{g})}=\mathrm{F}\mathrm{e}{\mathrm{O}}_{(\mathrm{s})}+\mathrm{C}\mathrm{O}{\mathrm{S}}_{(\mathrm{g})}$$(7)

Iron oxides were only observed to have formed in the edge regions of the coke 131 where significant carbon loss had occurred. Reactions 3–6 all represent an increase in mass, while reaction 7 represents a decrease in mass (as sulfur is heavier than oxygen). However, as previously noted, the gasification of carbon under the pseudo-CRI (1100°C, CO₂) conditions dominates the mass change of the sample and as expected all samples lost mass under these conditions.

In summary, at 1100°C in a CO₂ atmosphere, the oxidising nature of the conditions meant that oxidation of iron bearing minerals and iron sulfide were the most prominent mineral-gas reactions that took place.

3.3.3. Mineral-carbon Reactions

The reduction of the minerals (mainly silica) by the coke carbon was found to occur at the higher temperatures (starting at 1350°C, but more prominent at 1600°C). The other reaction that occurred was the dissociation of the aluminosilicates in the coke to separate alumina and silica. As the majority of the iron bearing minerals in the unreacted coke were metallic iron (Fig. 3), the reduction of iron oxides by carbon was not considered here. These reactions are shown in Eqs. (8), (9), (10).   

$$\mathrm{S}\mathrm{i}{\mathrm{O}}_{2(\mathrm{s})}+{\mathrm{C}}_{(\mathrm{s})}=\mathrm{S}\mathrm{i}{\mathrm{O}}_{(\mathrm{g})}+\mathrm{C}{\mathrm{O}}_{(\mathrm{g})}$$(8)

  

$$\mathrm{S}\mathrm{i}{\mathrm{O}}_{2(\mathrm{s})}+3\cdot {\mathrm{C}}_{(\mathrm{s})}=\mathrm{S}\mathrm{i}{\mathrm{C}}_{(\mathrm{s})}+2\cdot \mathrm{C}{\mathrm{O}}_{(\mathrm{g})}$$(9)

  

$$\mathrm{A}{\mathrm{l}}_6\mathrm{S}{\mathrm{i}}_2{\mathrm{O}}_{13(\mathrm{s})}=3\cdot \mathrm{A}{\mathrm{l}}_2{\mathrm{O}}_{3(\mathrm{s})}+2\cdot \mathrm{S}\mathrm{i}{\mathrm{O}}_{2(\mathrm{s})}$$(10)

At 1350°C a small amount of silica was carbothermally reduced in the coke samples (Fig. 3) as given by reactions 8–9. While these reactions were not predicted by the thermodynamic modelling (Table 4), the issue was that the calculations were set-up in such a way that the conditions were slightly different from the experiments. In the experiments, CO generated by reaction would be removed by the flowing gas such that the partial pressure of CO (pCO) would be dominated by the inlet gas and not by the products of any reaction. A simple consideration of Le Chatelier’s Principle for Eqs. (8) and (9) shows that as the CO content of the gas decreases, the two reactions would be promoted.

Although these reactions occurred at 1350°C, the characterisation of the samples indicates that only a small amount of SiO₂ was reduced to SiC and SiO_(g) (Fig. 3). As such, the mass decrease expected from the reduction of SiO₂ would be small. As the mass loss due to the mineral reactions was small, it is likely that the mass loss in the coke samples (Fig. 2) was dominated by the Boudouard reaction (Eq. (2)) rather than the reduction of silica. The Boudouard reaction is possible under these conditions due to the presence of 2% CO₂ in the reaction gas.

In addition to the chemical reactions outlined above, physical changes to the minerals were also likely to have occurred within the samples. At 1350°C, melting of the iron bearing minerals within the coke samples may expected at the experimental temperatures. It is expected that the carbon content of the metallic iron in contact with the coke or coke analogue would be high, so that liquid is formed at the eutectic temperature of 1153°C.²⁶⁾ Iron sulfide was also expected to melt as a liquid sulfide phase (predicted for temperatures above ~1000°C).²⁷⁾

The reactions at 1600°C were similar to those at 1350°C, but due to either kinetic or thermodynamic considerations (or both) the reactions progressed much further. Silica was largely not present in the samples after the experiments, while SiC was found (see XRD data in Fig. 3). SiC was also predicted by the thermodynamic modelling to occur, even up to a pCO of ~1 atm.

It is also possible that silica was reduced to SiO_(g). However, this is gaseous species and is not present in the sample after reaction. To consider whether this reaction also occurred, simple mass balances was carried out. The XRD patterns in Fig. 3 show that SiC was present in the sample, while the silica (quartz or cristobalite) had largely disappeared. It can be assumed that all the silica in the sample was converted. The amount of quartz originally in the sample was 0.1 mol/100 g. For a nominal 8 g coke analogue sample, there was 0.008 mol SiO₂. The mass change for the total conversion of SiO₂ to SiC would be given by a mass balance of the solid phases (using the stoichiometry of reaction 9)   

$$\begin{array}{l}\mathrm{\Delta }\mathrm{m}\left(\mathrm{S}\mathrm{i}{\mathrm{O}}_2\to \mathrm{S}\mathrm{i}\mathrm{C}\right)=\\ \mathrm{n}\left(\mathrm{S}\mathrm{i}{\mathrm{O}}_2\right)\cdot \mathrm{M}\left(\mathrm{S}\mathrm{i}\mathrm{C}\right)-\left[\mathrm{n}\left(\mathrm{S}\mathrm{i}{\mathrm{O}}_2\right)\cdot \mathrm{M}\left(\mathrm{S}\mathrm{i}{\mathrm{O}}_2\right)+\mathrm{n}\left(\mathrm{S}\mathrm{i}{\mathrm{O}}_2\right)\cdot \mathrm{M}\left(\mathrm{C}\right)\right]\end{array}$$(11)

where Δm(SiO₂→SiC) is the mass change from the conversion of silica to SiC, n(i) is the number of moles of species i, and M(i) is the molar mass of species i. Calculating Δm(SiO₂→SiC) gives −0.448 g, which equates to an FWC of −0.056. A similar mass balance for the formation of SiO can be carried out (using the stoichiometry of reaction 9)   

$$\begin{array}{l}\mathrm{\Delta }\mathrm{m}\left(\mathrm{S}\mathrm{i}{\mathrm{O}}_2\to \mathrm{S}\mathrm{i}\mathrm{O}\right)=\\ -\left[\mathrm{n}\left(\mathrm{S}\mathrm{i}{\mathrm{O}}_2\right)\cdot \mathrm{M}\left(\mathrm{S}\mathrm{i}\mathrm{C}\right)+3\cdot \mathrm{n}\left(\mathrm{S}\mathrm{i}{\mathrm{O}}_2\right)\cdot \mathrm{M}\left(\mathrm{C}\right)\right]\end{array}$$(12)

This gives Δm(SiO₂→SiO) as −0.769 g, or an FWC of −0.096. The measured FWC of the coke samples were −0.07 for coke 131 and −0.08 for coke 114. Taking into account the identification of SiC in the samples through XRD and the mass balance calculations, it is likely that much of the silica in the sample was reduced to SiC, while only a comparatively small amount was reduced to SiO(g) and was lost.

In addition to the reduction of silica, In the XRD patterns (Fig. 3), peaks associated with aluminosilicates in the coke samples (as well as in the coke analogue samples) had largely disappeared after reaction at 1600°C. The characterisation of the samples by SEM showed that there was clear separation of the Al-rich minerals from the Si-rich minerals (Fig. 6) which was not present in the unreacted coke 131-K-001 sample or coke analogue with 131-K-001 ash. This decomposition of the aluminosilicates in the coke minerals is represented by reaction 10, using mullite (the main aluminosilicate identified in the coke) as an example.

Physical changes in the minerals were also observed in the samples, with the formation of liquid iron in the cokes after reaction at 1600°C with the CO–Ar. After melting of the iron, it was found to have subsequently spread through the coke (with smaller and more particles distributed through the sample). An iron-carbon alloy melts anywhere from the (Fe–C) eutectic temperature at 1153°C to the melting point of pure Fe at 1535°C.²⁷⁾ As this temperature range for the formation of liquid is below 1600°C, liquid iron would be expected in the samples at the experimental temperature.

In summary, the most prominent mineral-carbon reactions that occurred in the samples in conditions representing those in the lower zone of the blast furnace were the carbothermal reduction of silica to both SiC and SiO_(g). The mineral reactions that occurred at 1350°C in the CO–CO₂–Ar atmosphere were similar to those that occurred at 1600°C in the CO–Ar atmosphere.

3.4. Carbon Structure Changes during Reaction under Blast Furnace Conditions

The Raman spectroscopy of the coke samples was shown in Fig. 7. Clear changes in the Raman data can be seen with increasing temperature, with the coke becoming increasingly graphitic with increasing temperature.

After reaction with CO₂ at 1100°C, there was little change in the Raman spectra of the coke 131 sample, for both IMDC (Fig. 7(b)) and RMDC (Fig. 7(e)). As the graphitisation of coke is largely a thermally driven process,^6,11,18,22) little change in the sample after reaction at 1100°C may be expected. The coke 131 sample had likely been at temperatures in the range 1000–1200°C during firing, making further changes to the graphitisation of the sample at 1100°C unlikely.

However, after the reaction at 1600°C with the CO–Ar atmosphere, the coke 131 sample was much more graphitic, with more of an sp² nature to the carbon bonds. Increasing graphitisation of the samples was expected at 1600°C, it is well known that cokes become more graphitic at high temperatures.^{11,15,22,28,29,30,31)} This increasing graphitic nature of the coke 131 sample during reaction at higher temperatures was supported by the XRD results (Fig. 3), where the main graphite peak became stronger and sharper as the temperature increased.

The IMDC (Fig. 7(c)) and RMDC (Fig. 7(f)) has also tended to similar values, with their ranges overlapping in comparison to the unreacted coke sample. This indicates that the RMDC that was initially less graphitic had changed more than the IMDC. This finding agrees with the results of Xing et al.⁷⁾ who reported that while IMDC and RMDC both become more graphitic upon heating, they do so at different rates. This is significant and may have profound implications of how a coke will perform in a blast furnace. For example, low CSR cokes have been reported to become stronger than high CSR cokes at high temperatures.⁷⁾

4. Conclusions

An investigation into how coke mineralogy affects coke reactivity was conducted with a view to better understand the use of CRI data in predicting coke performance in a blast furnace. The aim of this study was to better understand the impact of mineralogy on the reactivity of coke at temperatures much higher than 1100°C and how this can or cannot be related back to the behaviour under CRI conditions. This was carried out by reacting coke samples under pseudo-CRI conditions (1100°C, 100% CO₂) and at higher temperatures under conditions designed to replicate the blast furnace. Two cokes from the ACARP coal bank were studied, a high iron coke (131) and a high silica coke (114). Extensive characterisation before and after reaction was carried out to assess the changes to the mineralogy and carbon structures in the coke and coke analogue samples.

The main conclusion of the study was that the effects of minerals on reactivity changed as temperature increased under blast furnace conditions. The effects of minerals on the reactivity under the pseudo-CRI conditions (1100°C, 100% CO₂) and at 1350°C with 39% CO-2% CO₂-59% Ar were found to be similar, while at higher temperatures (1600°C with gas composition 40% CO-60% Ar) the effect of minerals on the reactivity of cokes was very different. This confirms that the CRI test has applicability out with the temperature and gas environment the data were obtained and has relevance for the elevated temperatures and less oxidising conditions of the blast furnace. At higher temperatures, when the reactivity of the sample changed from being dominated by the gas-carbon reactions (and the mineral effects therein) to being dominated by the mineral-carbon reactions as the temperature increased and gas composition became less oxidising, more data, in addition to that of CRI, are required to understand and predict coke behaviour in a blast furnace.

Although the reactivity of the coke under the pseudo-CRI conditions (1100°C, 100% CO₂) was much higher than at 1350°C with 39% CO-2% CO₂-59% Ar, the effects of the minerals on the reactivity of coke were similar. The high iron coke 131 had higher reactivity than the high silica coke 114. The main cause of mass decrease under CSR/CRI conditions and also under the CO–CO₂–Ar atmosphere at 1350°C was carbon gasification by CO₂, which is widely known to be effected by the minerals within the coke. The effect of the minerals on the reactivity of the coke can be stated as: quartz decreases the reactivity of the coke, alumina has little effect, while iron bearing minerals increase the reactivity of the coke at 1100°C with CO₂.

At 1600°C with a 40% CO-60% Ar atmosphere, the main reaction of the coke switched from its reaction with CO₂ to reactions with the minerals themselves. Silica in coke reacted with the coke carbon to form both SiC and SiO(g). This resulted in the high silica coke 114 having a higher mass loss at 1600°C than the high iron coke 131.

The study on the carbon structures and types within the coke showed that, as expected, the coke samples became more graphitic at 1600°C. The change in the RMDC appeared to be greater than for the IMDC, as might be expected from previous studies.

Acknowledgements

This study was funded by ACARP. The authors acknowledge use of SEM JEOL JSM-6490LV within the UOW Electron Microscopy Centre.

References

• 1)   R.  Loison,  P.  Foch and  A.  Boyer: Coke: Quality and Production, Butterworth & Co. (Publishers) Ltd, London, (1998), 158.
• 2)   P.  Bennett,  A.  Reifenstein,  G.  O’Brien and  B.  Jenkins: ACARP Report, C12057, ACARP, Brisbane, (2008), 9.
• 3)  ASTM D5341-99: 2010, Standard test method for measuring coke reactivity index (CRI) and coke strength after reaction (CSR).
• 4)  ISO 18894: 2006, Coke - Determination of coke reactivity index (CRI) and coke strength after reaction (CSR).
• 5)   B.  van der Velden,  J.  Trouw,  R.  Chaigneau and  J.  van den Berg: 58th Ironmaking Conf., ISS, Warrendale, PA, (1999), 275.
• 6)   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
• 7)   X.  Xing,  G.  Zhang,  H.  Rogers,  P.  Zulli and  O.  Ostrovski: Metall. Mater. Trans. B, 45 (2014), 106. https://doi.org/10.1007/s11663-013-0002-y
• 8)   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
• 9)   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
• 10)   M.  Kawakami,  T.  Karato,  T.  Takenaka and  S.  Yokoyama: ISIJ Int., 45 (2005), 1027. https://doi.org/10.2355/isijinternational.45.1027
• 11)   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
• 12)   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
• 13)   Ł.  Smędowski,  M.  Krzesińska,  W.  Kwaśny and  M.  Kozanecki: Energy Fuels, 25 (2011), 3142. https://doi.org/10.1021/ef200609t
• 14)   P. D.  Green,  C. A.  Johnson and  K. M.  Thomas: Fuel, 62 (1983), 1013. https://doi.org/10.1016/0016-2361(83)90134-5
• 15)   S.  Dong,  N.  Paterson,  S. G.  Kazarian,  D. R.  Dugwell and  R.  Kandiyoti: Energy Fuels, 21 (2007), 3446. https://doi.org/10.1021/ef7003656
• 16)   S.  Dong,  P.  Alvarez,  N.  Paterson,  D. R.  Dugwell and  R.  Kandiyoti: Energy Fuels, 23 (2009), 1651. https://doi.org/10.1021/ef800961g
• 17)   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
• 18)   O. A.  Aladejebi,  B. J.  Monaghan,  M. H.  Reid,  M.  in het Panhuis and  R. J.  Longbottom: Steel Res. Int., 88 (2017), 1700039. https://doi.org/10.1002/srin.201700039
• 19)   W. W.  Gill,  N. A.  Brown,  C. D. A.  Coin and  M. R.  Mahoney: 44th Ironmaking Conf., ISS, Warrendale, PA, (1985), 233.
• 20)   M. H.  Reid,  M. R.  Mahoney and  B. J.  Monaghan: ISIJ Int., 54 (2014), 628. https://doi.org/10.2355/isijinternational.54.628
• 21)   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
• 22)   B. J.  Monaghan,  R.  Nightingale,  V.  Daly and  E.  Fitzpatrick: Ironmaking Steelmaking, 35 (2008), 38. https://doi.org/10.1179/030192307X233377
• 23)   G.  O’Brien,  Y.  Gu,  B. J. I.  Adair and  B.  Firth: Miner. Eng., 24 (2011), 1299. https://doi.org/10.1016/j.mineng.2011.04.024
• 24)   K.  Warren,  G.  Krahenbuhl,  M.  Mahoney,  G.  O’Brien and  P.  Hapugoda: Int. J. Coal Geol., 152 (2015), 3. https://doi.org/10.1016/j.coal.2015.09.012
• 25)   B.  Kwiecińska and  H. I.  Petersen: Int. J. Coal Geol., 57 (2004), 99. https://doi.org/10.1016/j.coal.2003.09.003
• 26)   W. D.  Callister: Materials Science and Engineering: An Introduction, 6th ed., John Wiley and Sons, Hoboken, NJ, (2003), 356.
• 27)   G.  Aylward and  T.  Findlay: SI Chemical Data, 3rd ed., Jacaranda, Melbourne, (1994), 14.
• 28)   A.  Oberlin: Carbon, 22 (1984), 521. https://doi.org/10.1016/0008-6223(84)90086-1
• 29)   H.  Marsh: Introduction to Carbon Science, Butterworth & Co. (Publishers) Ltd, London, (1989), 1.
• 30)   T.  Hilding,  S.  Gupta,  V.  Sahajwalla,  B.  Bjorkman and  J.  Wikström: ISIJ Int., 45 (2005), 1041. https://doi.org/10.2355/isijinternational.45.1041
• 31)   J.  Goma and  M.  Oberlin: Thin Solid Films, 65 (1980), 221. https://doi.org/10.1016/0040-6090(80)90256-4