Effect of Silica on Reduction Behaviors of Hematite-carbon Composite Compact at 1223–1373 K

Abstract

The influence of compositional changes in the SiO₂–Fe₂O₃ binary oxide system on the reduction behavior of carbon-bearing compact was investigated at 1223–1373 K. Density functional theory (DFT) calculations were employed to investigate the adsorption behaviors of O–Si–O on various FeO surfaces in order to explore the mechanism of the conversion of FeO to Fe–Si–O phase. Chemical analysis, SEM-EDX, and XRD were carried to characterize the conversion mechanisms. According to experimental results, silica causes the decrease of the metallization ratio by hindering the reduction of FeO to Fe and facilitating the reaction of FeO to Fe–Si–O phases. The size of metallic iron granules diminishes gradually and the boundary between the iron phase and the slag phase becomes less obvious with increasing silica content, which greatly increases the contact probability of SiO₂/liquid phases and FeO. For the three adsorption sites, the adsorption energies of Si atoms onto FeO surfaces are all obviously larger than that of the reducing gases, making it extremely difficult for the reducing gases to remove O atoms from FeO surfaces, as the DFT results show. Moreover, it is easier to form the Fe–O–Si phase on the FeO (110) surface since the adsorption energy of O–Si–O onto this surface is greater than that onto the FeO (100) and FeO (111) surfaces. Additionally, the charge of the O (FeO) 2p orbital, the Si 3p orbital, as well as the 3s, 3p, and 3d orbitals of the Fe atom take charge of the formation of Si–O–Fe bond.

1. Introduction

With the increasing shortage of high-quality raw materials around the world, various beneficiation methods are being developed for iron ores, especially for the high-silica and low-iron ores, to meet the rapidly growing demand of iron ores in the iron making industry.^1,2) In particular, China has more than 55 billion tons of this iron ore in the iron grade range of 20–50%, accounting for 89% of total iron ore reserves in China.^3,4) Regarded as one of the most refractory iron ore sources, this large resource is difficult to be beneficiated effectively with existing mineral processing technologies due to its complex mineral composition, high silica content and other reasons.⁵⁾

SiO₂-containing minerals found in iron ore are usually regarded as detrimental by many ironmaking and steelmaking operators. The studies relevant to the effects of SiO₂ are summarized below. El-Geassy et al.⁶⁾ found that the various forms of SiO₂, such as the crystalline (α -SiO₂) or amorphous forms, play an important role in determining the materials’ reducibility during the reduction of Fe₂O₃ and FeO. N Shigematsu et al.⁷⁾ found that with a small amount of SiO₂ dissolved in the FeO phase, the reduction is promoted at temperatures above 1073 K and retarded at temperatures below 1003 K. In addition, above 1473 K,⁸⁾ reduction of FeO is not accelerated by SiO₂, but retarded by the formation of the layer of dense iron on wüstite or by the dense fayalite phase. M Bahgat, et al.⁹⁾ observed that the addition of silica decreases the porosity of wüstite compact owing to the formation of iron silicate phases. Hyunsik Park et al.¹⁰⁾ investigated the influence of alumina and silica on the reduction behavior of carbon composite pellet at 1273 K, 1373 K¹¹⁾ and 1473 K¹²⁾ by thermogravimetric analysis. Wanho K, et al.¹³⁾ indicated that the reduction rate and reducibility under H₂ are significantly higher than those under CO when SiO₂-containing (0–30 wt.%) compacts get reduced at 1273 K. Huang Z, et al.¹⁴⁾ investigated the reduction behavior of low-iron and high-silica ore-coal composite pellets at 1143–1263 K. Although the effect of SiO₂ has been well documented for many research techniques, the exact underlying reasons for various performances are often more difficult to access.

As a robust and accurate method, density functional theory (DFT) is becoming more and more prevalent in the modeling of adsorptions of different phase interfaces, including those relevant to the chemical industry and metallurgical engineering.^15,16) Du M F et al.¹⁷⁾ carried out a study of fayalite with a high content of iron in the initial layer of coal ash using the GGA and PW 91 based on quantum chemistry. In this study regarding the characteristics of high silica, it is easy for the SiO₂–Fe₂O₃ system to form some low melting phases such as Fe₂SiO₄, Fe₂SiO₄–SiO₂, Fe₂SiO₄–FeO, and Fe₂SiO₄–Fe₃O₄. The soft melting points of Fe₂SiO₄, Fe₂SiO₄–SiO₂, Fe₂SiO₄–FeO, and Fe₂SiO₄–Fe₃O₄ are only 1478 K (1205°C), 1451 K (1178°C), 1451 K (1178°C) and 1415 K (1142°C), respectively, leading to a ring formation in the rotary kiln and a disruption of normal operation.^18,19) Another major characteristic of this high-silica and low-iron ore is high fine-grained dissemination. Within the iron ore used in our previous studies, process mineralogy tests showed that 95.73% of the diameters of hematite grains were smaller than 37 μm and 39.94% diameters of the hematite particulates were smaller than 10 μm.^14,20,21) In summary, all of these characteristics bring a great difficulty to the beneficiation of iron resources and provide an opportunity for silica and hematite particulates to concatenate tightly with each other. As we know, elementary reaction steps, that is, the dissociation of reactants and association to products, take place at the solid-gas or solid-liquid interface. However, surface and electronic properties of solids are often also significantly altered as particle sizes decrease.²²⁾ Thus, hematite and silica powder with average particulate sizes of 7.172 μm and 7.920 μm, respectively, were adopted to make briquettes for reduction roasting in the experimental part of this study in order to simulate the grain distribution states of fine-grained dissemination of high-silica and low-iron ore to understand the influence of different silica contents. For the theoretical experiments, the ‘O–Si–O’ segment, a shared structure between SiO₂ and 2FeO·SiO₂, was employed for the DFT calculations according to Fig. 1. The unit cell of SiO₂ (as depicted in Fig. 1(a)) shows that silicon ions are tetrahedrally coordinated to oxygens.²³⁾ Meanwhile, the 2FeO·SiO₂ unit cell (as shown in Fig. 1(b)) contains four formula units with 28 atoms (8 Fe, 4 Si, and 16 O). Silicon ions are tetrahedrally coordinated to oxygens, whereas iron-ions occupy the centers of distorted oxygen octahedra.^17,24,25) Thus, DFT calculations based on the periodic surface model has been performed in this paper, which investigate the adsorption behaviors and properties of O–Si–O onto three different FeO surfaces.

Fig. 1.

The side view of unit cell structure of SiO₂ (a) and 2FeO·SiO₂ (b). (Purple, red and green balls represent the Fe, O and Si atoms, respectively.).

2. Experimental Procedure

2.1. Apparatus and Procedure

For the sample preparation, hematite powder was obtained by roasting a magnetite concentrate from Columbia in air at 1173 K (900°C) for 1 hour to oxidize. The chemical compositions of hematite powder and pulverized anthracite (reductant) are shown in Tables 1 and 2, respectively. The amount of coal addition in the mixture was determined according to C/Fe mass ratio of 0.3 (i.e., mass ratio of fixed carbon in the anthracite to reducible iron in hematite powder was 0.3) as this was confirmed as adequate from the previous studies.^14,20,21) Besides, the bentonite content was fixed at 1.0% of the total mass of the compacts. The silica contents were correspondingly set to 1.5 wt.%, 6.0 wt.%, 30.0 wt.%, and 50.0 wt.% for the investigation of its impact on reduction behavior. The carbon composite compacts used in this study were composed of hematite powder, anthracite, bentonite, and analytical SiO₂ reagent chemicals (≥ 99.0%), which were cylindrical in shape, 10 mm in diameter, 8–13 mm in height, and 2.5 g (± 0.05 g) in weight. Then, the samples were dried in air at 378 K (105°C) for 3 hours to eliminate moisture. The mean particulate sizes of the hematite and silica powders were 7.172 μm and 7.920 μm, respectively.

Table 1. Chemical composition of hematite powder (mass, %).

Fe2O3	FeO	SiO2	Al2O3	CaO	MgO	MnO	Na2O	K2O	S	P
95.13	1.52	1.53	0.78	0.69	0.11	0.05	0.06	0.03	0.01	0.02

Table 2. Proximate analysis of pulverized anthracite (mass, %).

Category	FC, ad	M, ad	A, ad	V, ad	S
Anthracite	82.54	3.14	10.57	6.68	0.21

(FC, ad: Fixed carbon (air dried basis); M, ad: Moisture (air dried basis); A, ad: Ash (air dried basis); V, ad: Volatile matter (air dried basis))

The experimental procedure can be described as follows: 10–12 dried compacts, coupled with 60 g of coal powder beneath and another 120 g of coal on the top of the compacts, were used each time for carbothermic reduction. Reductive roasting of the compacts was performed in a shaft furnace in the temperature range of 1223–1373 K for 30–90 min. Reduced compacts were then cooled to ambient temperature in a nitrogen atmosphere. The metallization ratios of the reduced compacts were measured after each reductive roasting experiment. The metallization ratio (M_R) was evaluated with Equation:   

$${\mathrm{M}}_{\mathrm{R}}={\mathrm{M}}_{\mathrm{F}\mathrm{e}}/{\mathrm{T}}_{\mathrm{F}\mathrm{e}}\times 100\%$$(1)

where M_Fe is the metallic iron content and T_Fe is the total weight of iron in the reduced compacts. The measurements of M_Fe and T_Fe were corresponding obtained by FeCl₃–K₂Cr₂O₇ and TiCl₃–K₂Cr₂O₇ volumetric methods.²⁶⁾

After reduction, phase transfers of the reduced compacts were analyzed by chemical phase dissolution measurements and X-ray diffraction (XRD). The morphological evolution and phase evolution were investigated by scanning electron microscope (SEM) equipped with EDX.

2.2. Computational Method and Models

In this work, all the calculations were performed with the Cambridge Sequential Total Energy Package (CASTEP)²⁷⁾ with a pseudopotential plane-wave approach based on DFT. The electron ion-core interactions were described by an OTFG ultra-soft pseudopotential, the most accurate calculation method at present, supplied with VASP of Si 3S²3P², Fe 3s²3p⁶3d⁶, and O 2s²2p⁴ states as valence states.^28,29) The generalized gradient approximation (GGA)³⁰⁾ with the Perdew-Burke-Ernzerhof (PBE) exchange-correlation functional³¹⁾ applied to deal with the electron exchange and correlation interaction. The solutions of Kohn-Sham equations were represented by the plane-wave-basis with a kinetic energy cutoff of 280 eV.³²⁾ The convergence tolerance energy, self-consistent iterative convergence accuracy, max displacement, max force and convergence of internal stress among atoms were 2.0×10⁻⁵ eV/atom, 2.0×10⁻⁵ eV/atom, 0.002 Å, 0.05 eV/Å and 0.1 GPa, respectively. The Brillouin zone integration was performed in the Monkhorst-Pack scheme using the k-point set of 2 × 2 × 1 for adsorption studies.^33,34) For the Brillouin zone integrations, a 12 × 12 × 12 Monkhorst-Pack grid of special k-points was used.

As a crystal with face-centered cubic structure, the solid FeO used in this study has a cell parameter of 4.307 Å,³⁵⁾ in which the Fe and O atoms adopt an octahedral coordination. All Fe–O bond lengths were found to be 2.166 Å (Fig. 2(a)), which coincides well with Zhong H’s theoretical result of 2.166 Å.³⁶⁾ As shown in Figs. 2(b), 2(c) and 2(d), FeO (100), FeO (110) and FeO (111) surfaces were chosen. 1 × 1 cells³⁸⁾ with five relaxed atomic layers were adopted.³⁷⁾ To avoid the interaction between neighboring slabs, the vacuum space was set to 15 × 15 × 15 Å, which had been proven to be large enough in previous studies.³⁸⁾ A structurally optimized O–Si–O segment with Si–O bond lengths of 1.549 Å and a bonding angle of 180° was obtained in the same vacuum space, as depicted in Fig. 2(e), before adsorbing onto the FeO surface.

Fig. 2.

The side view of 1 × 1 unit cell structure of FeO (a), FeO (100) (b), FeO (110) (c), FeO (111) (d) and O–Si–O (e). (Purple, red and yellow balls represent the Fe, O and Si atoms, respectively.).

The adsorption energies (E_ad) of an O–Si–O fragment approaching the clean (100), (110) and (111) FeO surfaces were each calculated as the total energy difference of the energy of an isolated O–Si–O segment (E_O–Si–O) and the energy of the FeO surface before adsorption (E_slab),   

$${\mathrm{E}}_{\mathrm{a}\mathrm{d}}={\mathrm{E}}_{\mathrm{s}\mathrm{y}\mathrm{s}\mathrm{t}\mathrm{e}\mathrm{m}}-{\mathrm{E}}_{\mathrm{O}-\mathrm{S}\mathrm{i}-\mathrm{O}}-{\mathrm{E}}_{\mathrm{s}\mathrm{l}\mathrm{a}\mathrm{b}}$$(2)

where E_system, E_O–Si–O, and E_slab are the energy of the adsorption system, the isolated O–Si–O segment and the clean surface, respectively.

3. Results and Discussion

3.1. Influence of Silica on Metallization at 1223-1373 K

The metallization ratios of the reduced compacts with SiO₂ contents of 1.5–50 wt.% in the temperature range of 1223–1373 K (950–1100°C) for 30 min are given in Fig. 3. It can be observed from the line chart that when the compact with a SiO₂ content of 1.5% was roasted at 1223 K, the metallization ratio was lower than 60%. This value greatly increased when the temperature rose to 1373 K, reaching 93.7%, and increases by 35.2% compared to 1223 K. Additionally, increasing the SiO₂ content to 50 wt.% led to a sharp diminution of the metallization ratio, even when the compact was roasted at 1373 K. This value decreased to 60.9% for a SiO₂ content of 50% at 1373 K, indicating that SiO₂ produces a tremendous inhibition effect on the metallization ratio of reduced compacts. Moreover, when a compact with a 50 wt.% of SiO₂ was reduced at 1223 K, the metallization ratio was extremely low, only 44.3%. Therefore, the results imply that silica within the iron oxide and solid carbon significantly influences the formation of metallic iron both at higher temperatures and at lower temperatures.

Fig. 3.

Effect of temperature and time on metallization ratio of reduced compact.

A more comprehensive study of the distribution of valence states within the iron phase in the reduced compacts at 1323 K was carried out via chemical phase dissolution, presented in Fig. 4. As can be seen, the Fe⁰ state accounts for most of the total iron of the reduced compact with different silica contents, followed by Fe²⁺ and Fe³⁺. It is quite clear that the percentage contents of both Fe²⁺ and Fe³⁺ are gradually promoted with increasing silica content, while the Fe⁰ changes in the opposite trend, simultaneously.

Fig. 4.

Valence state distribution of reduced compact with different silica content at 1323 K.

3.2. Influence of Silica on Phase and Morphology Evolution at 1323 K

Compacts with different contents of silica reduced at 1323 K for 30 min were characterized by XRD to understand the phase transformation upon increasing the silica content, as presented in Fig. 5. The XRD pattern showed the major phase of metallic iron coexists with minor silica in the reduced compact with 6 wt.% silica, implying that the repercussions for 6 wt.% silica usage in compact are almost negligible to the reduction process. A further 30 wt.% rise of silica content leads to the formation of cristobalite and fayalite, which inevitably hinders the separation of elemental Fe and Si. Furthermore, the increase of silica content from 30% to 50 wt.% significantly strengthens the intensity of cristobalite and fayalite diffraction peaks. Meanwhile, the diffraction peak corresponding to metallic iron weakens with the increase of silica content, which corresponds directly with the results in Fig. 4. Thus, the reason for the phenomenon of diminution of the metallization ratio is considered to be the formation of fayalite (Fe₂SiO₄), which is more difficult to be reduced by carbon.

Fig. 5.

XRD patterns of reduced compact with different silica content at 1323 K.

The microstructure (SEM) and energy spectrum analysis of the reduced compact with different silica contents reduced at 1323 K for 30 min were characterized, as shown in Fig. 6, to reveal the intrinsic morphological evolution and its relationship to silica content. A compact platy structure can be widely observed in a reduced compact with a silica content of 6.0%, as shown in Fig. 6(a). It is mainly composed of metallic iron and wüstite after being identified by EDX. In addition, analysis of mineralogical constitution and the structure of the compact showed that larger iron phase granules (about 30–50 μm) and a distinct boundary between the iron phase and the silicon-rich slag phase form (Fig. 6(a)). Compared to Figs. 6(b), 6(c), it created a favorable condition for subsequent milling-magnetic separation. Silica content from 6 to 30 wt.% contributed to the fragmentary structure with fine iron granules (about 10 μm), as shown in Fig. 6(b). Additionally, the Fe–Si–O phase (FeO·3.7SiO₂; FeO·3.2SiO₂; 2FeO·SiO₂) and the independent silica particulates confirmed by EDX analysis became more and more apparent. A further 50 wt.% rise of SiO₂ content led to a sharp diminution of the iron phase dimension (only 5 μm) and a sharp increase of Fe–Si–O phase (FeO·20.1SiO₂; 1.6FeO·SiO₂), as shown in Fig. 6(c). Moreover, the subsidence holes in the reduced compact diminished gradually with the increase of silica content. To summarize, it is clear that silica plays a crucial role in the microstructure and phase evolution of these materials.

Fig. 6.

SEM-EDX analysis of reduced compact with different silica content at 1323 K: 6% silica (a), 30% silica (b), and 50% silica (c).

3.3. Influence of Silica on Compressive Strength of Reduced Compact

The compressive strength of the reduced compacts with different silica contents at 1223–1323 K are shown in Fig. 7.

Fig. 7.

Compressive strength of reduced compact with different silica content at 1223–1323 K.

It can be seen that the compressive strength of the reduced compact with a silica content of 1.5% is relatively low (400–1100 kPa) at all of the roasting temperatures. Moreover, the value reached 3215 N at 1323 K when 6.0% silica is added and increases by 2731 N compared to 1223 K. A further 30 wt.% rise of silica content brought about a sharp enhancement of the compressive strength (2000–3000 kPa) of the compact, even when roasted at 1223 K, as depicted in Fig. 7. Furthermore, it is obvious that the value (4000–8000 kPa) continued to increase with increasing silica content, demonstrating that melts of low-temperature eutectic point formed abundantly. Therefore, the results suggest that the compressive strength of the reduced compact is enhanced effectively by silica, particularly when the silica content is above 30%, making the diffusion resistance of the reducing gases larger and larger, which is another reason for the diminution of the metallization ratio.

3.4. Mechanism Analysis of Effect of Silica on Reduction

3.4.1. Reaction Mechanism of Iron Ore Reduction with Silica

The main reduction of carbon composite compact takes place through the reaction between wüstite and anthracite in the solid state at 1223–1373 K (950–1100°C). The reaction mechanism has been studied by numerous researchers.^39,40,41,42)   

$$\begin{array}{l}\mathrm{F}\mathrm{e}\mathrm{O}+\mathrm{C}=\mathrm{F}\mathrm{e}+\mathrm{C}\mathrm{O}\\ \mathrm{\Delta }{\mathrm{G}}^{\mathrm{\theta }}=147\mathrm{\hspace{0.17em} }900-150.24\mathrm{T}\mathrm{\hspace{0.17em} }(\mathrm{J}/\mathrm{m}\mathrm{o}\mathrm{l})\end{array}$$(3)

  

$$\begin{array}{l}\mathrm{F}\mathrm{e}\mathrm{O}+\mathrm{C}\mathrm{O}=\mathrm{F}\mathrm{e}+\mathrm{C}{\mathrm{O}}_2\\ \mathrm{\Delta }{\mathrm{G}}^{\mathrm{\theta }}=-22\mathrm{\hspace{0.17em} }800+24.26\mathrm{T}\mathrm{\hspace{0.17em} }\text{(}\mathrm{J}/\mathrm{m}\mathrm{o}\mathrm{l})\end{array}$$(4)

  

$$\begin{array}{l}\mathrm{C}+\mathrm{C}{\mathrm{O}}_2=2\mathrm{C}\mathrm{O}\\ \mathrm{\Delta }{\mathrm{G}}^{\mathrm{\theta }}=172\mathrm{\hspace{0.17em} }423-175.69\mathrm{T}\mathrm{\hspace{0.17em} }(\mathrm{J}/\mathrm{m}\mathrm{o}\mathrm{l})\end{array}$$(5)

  

$$\begin{array}{l}2\mathrm{F}\mathrm{e}\mathrm{O}+\mathrm{S}\mathrm{i}{\mathrm{O}}_2=2\mathrm{F}\mathrm{e}\mathrm{O}·\mathrm{S}\mathrm{i}{\mathrm{O}}_2\\ \mathrm{\Delta }{\mathrm{G}}^{\mathrm{\theta }}=-76\mathrm{\hspace{0.17em} }415+33.79\mathrm{T}\mathrm{\hspace{0.17em} }\text{(}\mathrm{J}/\mathrm{m}\mathrm{o}\mathrm{l})\end{array}$$(6)

  

$$\begin{array}{l}2\mathrm{F}\mathrm{e}\mathrm{O}·\mathrm{S}\mathrm{i}{\mathrm{O}}_2+2\mathrm{C}=2\mathrm{F}\mathrm{e}+\mathrm{S}\mathrm{i}{\mathrm{O}}_2+2\mathrm{C}\mathrm{O}\\ \mathrm{\Delta }{\mathrm{G}}^{\mathrm{\theta }}=353\mathrm{\hspace{0.17em} }922-338.96\mathrm{T}\mathrm{\hspace{0.17em} }\text{(}\mathrm{J}/\mathrm{m}\mathrm{o}\mathrm{l})\end{array}$$(7)

  

$$\begin{array}{l}1/2\left(2\mathrm{F}\mathrm{e}\mathrm{O}\cdot \mathrm{S}\mathrm{i}{\mathrm{O}}_2\right)+\mathrm{C}\mathrm{O}=\mathrm{F}\mathrm{e}+1/2\mathrm{S}\mathrm{i}{\mathrm{O}}_2+\mathrm{C}{\mathrm{O}}_2\\ \mathrm{\Delta }{\mathrm{G}}^{\mathrm{\theta }}=6\mathrm{\hspace{0.17em} }261-5.01\mathrm{T}\mathrm{\hspace{0.17em} }\text{(}\mathrm{J}/\mathrm{m}\mathrm{o}\mathrm{l})\end{array}$$(8)

The whole reduction process is a perfect cycle: First of all, although the solid-solid interaction (Eq. (3)) contributes only a small part to the total reduction process at 1223–1373 K, the reaction cycle is initiated by this reaction. Then, the gas immediately reacts with wüstite as shown in Eq. (4), followed by the generation of CO in Eq. (3). Finally, this solid-gas reaction generates CO₂ gas that is used for a source of the Boudouard reaction seen in Eq. (5).¹¹⁾ However, the reaction of FeO to Fe, which should take place, seems to be substantially inhibited by the added amount of silica which is known to form high concentration in FeO–SiO₂ melt.¹²⁾ The concentrations of CO used as a reducing gas needed for the reduction reactions are demonstrated in Fig. 8. It can be observed from the line chart that, it is harder for the FeO/Fe reaction to proceed forward compared with reactions of Fe₂O₃/Fe₃O₄ and Fe₃O₄/FeO on account of a far higher CO concentration requirement.⁴³⁾ When considering Figs. 3 and 7, we found that the metallization ratio of the reduced compacts with 50% silica are extremely low, only 44.27–60.94% when roasted at 1223–1373 K. Meanwhile, the compressive strengths were fairly high, 3947–8111 N. The results confirm that the CO concentration around some FeO particulates in the compacts is most probably lower than the required level for the FeO particulates that are surrounded by massive, fine-grained disseminated silica particulates (shown in Fig. 9) and the newly formed silicate melts. Thus, the intermediate FeO reduction will inevitably react with silica/silicate, particularly when the silica content is high, to form the Fe–Si–O phase, shown in Eq. (6). P. Semberg⁴³⁾ and A. A. Elgeassy⁴⁴⁾ also reported the same observation. Furthermore, Eqs. (7), (8) and Fig. 8 indicate that, once FeO transformed into Fe₂SiO₄, it can hardly be reduced to metallic iron again.

Fig. 8.

Concentration of reducing gas needed for reduction using CO.

Fig. 9.

Schematic diagram of particulate distribution in silica-hematite compact.

3.4.2. Adsorption Behaviors of O–Si–O on FeO Surface

As depicted in Fig. 9, the volume ratio of silica to hematite increases gradually with the increase of silica content. By computation, the values are 49:51 and 69:31, respectively, when the silica content in the SiO₂–Fe₂O₃ system is 30% and 50%. Moreover, the d₅₀ of silica and hematite used in this study is correspondingly 7.920 μm and 7.172 μm identified by particulate size analysis. Thus, the granule count ratios of silica to hematite are 41:59 and 62:38 when the silica content is 30% and 50%, respectively. Consequently, silica and hematite particulates are in tight conjunction with each other in compacts, creating an opportunity for the adsorption and dissolution especially when melts of low-temperature eutectic points form.

(1) Adsorption of O–Si–O onto the FeO (100) Surface

The adsorption of O–Si–O on the FeO (100) surface was calculated for the Si–O (FeO) top site, as shown in Fig. 10. The results implied that the bonding angle of the ‘O–Si–O’ segment changed (from 180° to 102.948°) distinctly for the configuration of the Si atom adsorbing directly to the O top site. The ‘O–Si–O’ segment was chemisorbed onto the surface, forming a Si–O bond with a bond length of 1.889 Å and an adsorption energy of −6.25 eV. Compared with Figs. 10(b) and 10(c), these results show that a slight relaxation occurs in the adsorbed surface. Moreover, the relaxation of the second and the third inter-lattice spacing is found to be quite different from that of the pure optimized FeO (100) surface, with an expansion instead of a contraction of atomic layers. As shown in Fig. 10(c), the distance between the adsorbed O atom and the second layer Fe atom is elongated from 1.831 Å to 3.015 Å, demonstrating the formation of a new Si–O bond and the breakage of an original Fe–O bond. Meanwhile, it is worth noticing that the O atom on the FeO (100) surface is raised slightly due to the adsorption of the ‘O–Si–O’ segment. Also, the distance between the adsorbed O atom and the adjacent Fe atom on the first layer is shortened from 1.903 Å to 1.890 Å, indicating that the binding energy between Fe and O on the first layer is increased after adsorption, meaning the reducing deoxygenation process will become more difficult. In addition, the Si–O bond of O–Si–O is activated and elongated, which are correspondingly 1.701 Å and 1.710 Å in length and longer than that of the optimized O–Si–O (1.549 Å). Hence, the present study confirms that the interaction between the O atom of the first layer and Si intensifies after adsorption, while the interaction between the O atom of the first layer and the Fe atom of the second layer weakens substantially.

Fig. 10.

Electron distribution and Mulliken charges of atoms for adsorption of O–Si–O on the FeO (100) surface: top view before adsorption (a), side view before adsorption (b), and side view after adsorption (c) (Purple, red and yellow balls represent the Fe, O and Si atoms, respectively).

The total atomic charges from the natural bond orbital charge analysis for O–Si–O adsorbed on FeO (100) at the Si–O top site are presented in Table 3. After adsorption, the delocalization of the electron cloud of the outermost layers of Si and Fe atoms makes the O atoms of the O–Si–O segment and the O atom of the first layer negative. From the results in the table, we can see that the charges of the 3p orbital of the Si atom and the 3d orbital of the Fe1 atom change in the adsorption process. The main reason for the adsorption is the interactions between the 2p orbital of the O3 atom and the 3p orbital of the Si atom consorting with the 3d orbital of the Fe1 atom. It can be concluded that the adsorption of Si on the FeO (100) surface causes charge transference from the 3p orbital of the Si atom and the 3d orbital of the Fe1 atom to the 2p orbital of the O atom.

Table 3. Orbital charge population analysis of O–Si–O adsorbed on FeO (100) surface.

Site	Species	s	p	d	Total
Free	O1	1.95	5.04	0.00	6.99
	O2	1.95	5.04	0.00	6.99
	Si	0.72	1.29	0.00	2.02
Clean	O3	1.92	4.66	0.00	6.58
	Fe1	0.56	0.10	6.68	7.35
	Fe2	0.58	0.22	6.71	7.51
System	O1	1.87	5.15	0.00	7.02
	O2	1.86	5.16	0.00	7.02
	O3	1.91	4.75	0.00	6.66
	Si	0.76	1.17	0.00	1.93
	Fe1	0.38	0.29	6.57	7.25
	Fe2	0.66	0.20	6.74	7.60

Further investigation into the type of bonding mechanism for adsorption of the O–Si–O segment onto the FeO (100) surface can be obtained by analyzing the partial density of states (PDOS) of O–Si–O and FeO (100) surface before and after adsorption, as shown in Fig. 11. The Fermi energy level was set as zero when plotting. In Figs. 11(a) and 11(b), the electronic state locates in the energy range of −2.5 to 0 eV and 5.0 to 12.5 eV, which is the main contribution of the 3p orbital of the Si atom before adsorption, while the 3p band peak splits into three peaks and shifts downward towards the Fermi level after adsorption. Evidently, the state density of Si atom weakens. As a comparison, the results of Figs. 11(c) and 11(d) lead to a similar conclusion that the 2p orbital peak of the O atom splits into two peaks and shifts downward towards the Fermi level as well, making the new Si–O bond more stable. Thus, it indicates clearly that the 3p band of Si and the 2p band of O are responsible for the Si–O bonding, corresponding to the results of orbital charge population analysis.

Fig. 11.

PDOS of the Si before adsorption (a), the adsorbed Si (b), the clean O of FeO before adsorption (c), the adsorbed O (d), and the adsorbed Si/O (e).

(2) Adsorption of O–Si–O onto the FeO (110) Surface

Adsorption of the O–Si–O segment onto the FeO (110) surface was investigated as well by approaching the Si atom onto the O top site, as shown in Fig. 12. In Fig. 12(c), the O–Si–O is chemisorbed onto the surface via Si–O bonding after adsorption, making the surface relaxation more apparent. Additionally, the calculation results imply that the Si–O bond length is 1.686 Å and the adsorption energy is −7.01 eV, which is stronger than that of Si adsorbing onto the FeO (100) surface. Meanwhile, the bonding angle of the O–Si–O changed from 180° to 150.148°, and the Si–O bond of O–Si–O is activated and elongated, which are correspondingly 1.706 Å and 1.739 Å in length and longer than that of the optimized O–Si–O (1.549 Å). In addition, the Fe–O bond length had been stretched slightly from 1.779 Å to 1.862 Å, with the raising of the O atom on the surface being due to the adsorption of Si, indicating a new Si–O–Fe bond formation.

Fig. 12.

Electron distribution and Mulliken charges of atoms for adsorption of O–Si–O on FeO (110) surface: top view before adsorption (a), side view before adsorption (b), and side view after adsorption (c) (Purple, red and yellow balls represent the Fe, O and Si atoms, respectively).

The total atomic charges from the natural bond orbital charge analysis for O–Si–O adsorbed on FeO (110) at Si–O top site are presented in Table 4. From the results in the table, the delocalization of the electron clouds of outermost layers of the Si and Fe atom makes the O atom of the first layer negative. Additionally, the charge of the 2p orbital of the O atom, the 3p orbital of the Si atom, as well as the 3s, 3p, and 3d orbitals of the Fe atom are responsible for the new Si–O–Fe bonding. As a comparison, PDOS of the Si atom, the O atom and Si–O are shown in Fig. 13. Orbital hybridization and bonding characteristics are analyzed according to the interaction between Si atoms and O atoms. In Fig. 13(e), there is a strong resonance between the O 2p band and the Si 3p band in the energy range of −10.0 to −2.0 eV, implying that the interaction between the O 2p orbital and the Si 3p orbital is relatively strong, meaning a hybridized orbital has formed.

Table 4. Orbital charge population analysis of O–Si–O adsorbed on FeO (110) surface.

Site	Species	s	p	d	Total
Free	O1	1.95	5.04	0.00	6.99
	O2	1.95	5.04	0.00	6.99
	Si	0.72	1.29	0.00	2.02
Clean	O3	1.92	4.67	0.00	6.59
	Fe	0.45	0.23	6.75	7.43
System	O1	1.88	5.10	0.00	6.98
	O2	1.90	4.91	0.00	6.82
	O3	1.89	4.96	0.00	6.85
	Si	0.81	1.22	0.00	2.03
	Fe	0.33	0.14	6.54	7.01

Fig. 13.

PDOS of the Si before adsorption (a), the adsorbed Si (b), the clean O of FeO before adsorption (c) and the adsorbed O (d) for adsorption of O–Si–O on FeO (110) surface.

(3) Adsorption of O–Si–O onto the FeO (111) Surface

Here, just like the adsorptions of O–Si–O onto the (100) and (110) FeO surfaces, adsorption of the O–Si–O onto the FeO (111) surface was investigated by approaching the Si atom onto the O top site, as shown in Fig. 14. In Fig. 14(c), the O–Si–O is chemisorbed to the O top site by Si–O bonding. After adsorption, the new Si–O bond length is 2.390 Å and the Si–O bond lengths of the O–Si–O are elongated from 1.549 Å to 1.962 Å and 1.967 Å, respectively. The adsorption energy of this configuration is −5.72 eV according to calculation, which is the smallest adsorption energy value among that of the above three configurations.

Fig. 14.

Electron distribution and Mulliken charges of atoms for adsorption of O–Si–O on FeO (111) surface: top view before adsorption (a), side view before adsorption (b), and side view after adsorption (c) (Purple, red and yellow balls represent the Fe, O and Si atoms, respectively).

(4) Different Adsorption Behaviors of O–Si–O onto FeO Surfaces

Comparing the adsorption behaviors of O–Si–O onto three different FeO surfaces, three different characteristics are obtained. Firstly, the adsorption energies of O–Si–O onto FeO surfaces are all larger than that of H₂ (−0.65 eV) and CO (−1.41 eV),³²⁾ implying that O–Si–O is more easily and steadily adsorbed onto FeO surfaces than reducing gases (H₂ or CO). Thus, it is difficult for H₂ and CO to remove O atoms from FeO surfaces when it is adsorbed by O–Si–O, which is corroborated by the experimental results of Fig. 4. Secondly, the first layer O (FeO) atoms on different FeO surfaces are all raised slightly after adsorption. Also, the bond lengths between the adsorbed O atoms and the second layer Fe atoms are clearly elongated, demonstrating the formation of a new Si–O bond and the breakage of the original Fe–O bond. Thirdly, the configuration of O–Si–O adsorbing to the FeO (110) surface is the most stable state (−7.01 eV) among the three examined sites, indicating that it is easier for FeO (110) to form the Fe–O–Si phase. Moreover, the surface relaxation on the FeO (110) surface is the most obvious due to a stronger interplay between Si atoms and the surface.

4. Conclusions

The effect of silica on the reduction behaviors of a hematite-carbon composite compact at 1223–1373 K (950–1100°C) was investigated. DFT calculations have been employed to investigate the adsorption behaviors of the ‘O–Si–O’ segment onto three FeO surfaces for further understanding of the mechanism. The following conclusions can be drawn from this study.

(1) The metallization ratio of carbon composite compact is influenced by silica contents. Silica decreases the metallization ratio by hindering the process of reducing FeO to Fe and facilitating the reaction of FeO to Fe–Si–O phases (FeO·20.1SiO₂; FeO·3.7SiO₂; FeO·3.2SiO₂; 1.6FeO·SiO₂; 2FeO·SiO₂) according to the results of XRD, SEM-EDX and chemical analysis. Moreover, the size of metallic iron granules in the reduced compacts diminishes gradually and the boundary between the iron phase and the silicon-rich slag phase becomes less and less obvious with increasing silica content, which greatly increases the contact probability of SiO₂/liquid phases and FeO.

(2) For the three adsorption sites, the adsorption energies of Si atoms onto FeO surfaces are all obviously larger than that of reducing gases, making it very difficult for reducing gases to remove O atoms from FeO surfaces, especially when Si–O–Fe bonding forms. Moreover, the charge of the 2p orbital of the O atom, the 3p orbital of the Si atom, as well as the 3s, 3p, and 3d orbital of the Fe atom, are responsible for the new Si–O–Fe bond according to the orbital charge population analysis.

(3) It is easier for the FeO (110) surface to form a Fe–Si–O phase since the adsorption energy of O–Si–O onto this surface is greater than that onto the Fe (100) and FeO (111) surfaces. In addition, a strong resonance between the O 2p orbital and the Si 3p orbital and their corresponding hybridized orbital in the energy range of −10.0 to −2.0 eV are observed when a Si atom adsorbs onto the FeO (110) surface, which takes charge of the formation of a Si–O bond.

Acknowledgments

The authors would like to express thanks to the National Science Fund for Distinguished Young Scholars of China (Grant No. 50725416) for financial support of this research.

References

• 1)   U.  Srivastava and  S. K.  Kawatra: Miner. Process. Extr. Metall. Rev., 30 (2009), 361.
• 2)   S. X.  Song,  E. F.  Campos-Toro,  Y. M.  Zhang and  A.  Lopez-Valdivieso: Int. J. Miner. Metall. Mater., 20 (2013), 113.
• 3)   Y.  Wang,  S. W.  Xing,  Z. J.  Zhang,  Y. B.  Ma and  Y.  Zhang: Multipurp. Util. Miner. Resour., 10 (2014), 15.
• 4)   Y. F.  Yu: Conserv. Util. Miner. Resour., 12 (2005), 43.
• 5)   Y. S.  Sun,  Y. X.  Han,  Y. F.  Li and  Y. J.  Li: Int. J. Miner. Metall. Mater., 24 (2017), 123.
• 6)   A. A.  Elgeassy,  K. A.  Shehata and  S. Y.  Ezz: Iron Steel Int., 50 (1977), 329.
• 7)   N.  Shigematsu and  H.  Iwai: Tetsu-to-Hagané, 73 (1987), 1329.
• 8)   N.  Shigematsu and  H.  Iwai: Tetsu-to-Hagané, 79 (1993), 920.
• 9)   M.  Bahgat,  K. S. A.  Halim,  M. I.  Nasr and  A. A.  El-Geassy: Ironmaking Steelmaking, 35 (2008), 205.
• 10)   T.  Inami and  K.  Suzuki: ISIJ Int., 43 (2003), 314.
• 11)   H.  Park and  V.  Sahajwalla: ISIJ Int., 54 (2014), 1256.
• 12)   H.  Park and  V.  Sahajwalla: ISIJ Int., 54 (2014), 49.
• 13)   K.  Wanho,  L.  Youngseok and  S.  Inkook: ISIJ Int., 52 (2012), 1463.
• 14)   Z. C.  Huang,  R. H.  Zhong,  L. Y.  Yi,  T.  Jiang,  L. M.  Wen and  Z. K.  Liang: Metall. Mater. Trans. B, 49 (2018), 411.
• 15)   H.  Zenia,  K.  Lounis,  E. H.  Megchiche and  C.  Mijoule: Comput. Mater. Sci., 124 (2016), 428.
• 16)   J. P.  Lu,  M. R.  Albert,  S. L.  Bernasek and  D. J.  Dwyer: Surf. Sci., 215 (1989), 348.
• 17)   M. F.  Du,  G. L.  Lu and  T.  Zhou: Adv. Mater. Res., 156 (2010), 1430.
• 18)   J. Y.  Fu,  T.  Jiang and  D. Q.  Zhu: Sintering and Pelletizing, Central South University of Technology Press, Changsha, (1996), 36.
• 19)   D. Q.  Zhu,  T. J.  Chun,  J.  Pan and  J. L.  Zhang: Int. J. Miner. Process., 125 (2013), 51.
• 20)   Z. K.  Liang,  Z. C.  Huang,  L. Y.  Yi and  R.  Zhong: 9th Int. Symp. on High Temperature Metallurgical Processing, The Minerals, Metals & Materials Society, Pittsburgh, PA, (2018), 465.
• 21)   Z. C.  Huang,  L. M.  Wen,  R. H.  Zhong and  T.  Jiang: 7th Int. Symp. on High Temperature Metallurgical Processing, The Minerals, Metals & Materials Society, Warrendale, PA, (2016), 461.
• 22)   S. E.  De,  M. M.  van Schooneveld and  F.  Cinquini: Angew. Chem. Int. Ed., 50 (2011), 1584.
• 23)   B. W.  Van Beest,  G. J.  Kramer and  R. A.  Van Santen: Phys. Rev. Lett., 64 (1990), 1955.
• 24)   M.  Cococcioni,  A. D.  Corso and  S. D.  Gironcoli: Phys. Rev. B, 67 (2003), 552.
• 25)   N. P.  Hou,  M. F.  Du,  J.  Li,  G. H.  Wu and  X. J.  Wu: Mater. Rev., 24 (2010), 76.
• 26)   R. J.  Saikkonen and  I. A.  Rautiainen: Bull. Geol. Soc. Finl., 65 (1993), 59.
• 27)   M. C.  Payne,  M. P.  Teter,  D. C.  Allan,  T. A.  Arias and  J. D.  Joannopoulos: Rev. Mod. Phys., 64 (1992), 1045.
• 28)   G. J.  Kresse and  D. P.  Joubert: Phys. Rev. B, 59 (1999), 1758.
• 29)   J. P.  Perdew,  J. A.  Chevary and  S. H.  Vosko: Phys. Rev. B, 46 (1993), 6671.
• 30)   J. A.  White and  D. M.  Bird: Phys. Rev. B, 50 (1994), 4954.
• 31)   T.  Ossowski and  A.  Kiejna: Surf. Sci., 637 (2015), 35.
• 32)   H.  Zhong,  L. Y.  Wen,  J. L.  Li,  J.  Xu,  M. L.  Hu and  Z. Q.  Yang: Powder Technol., 303 (2016), 100.
• 33)   A.  Stukowski and  K.  Albe: Modell. Simul. Mater. Sci. Eng., 18 (2010), 15.
• 34)   S. W.  Mai,  G. D.  Zhou and  W. J.  Li: Advanced Inorganic Structural, Chemistry Peking University Press, Beijing, (2006), 292.
• 35)   H. S.  Cheng,  D. B.  Reiser and  S.  Dean: Catal. Today, 50 (1999), 579.
• 36)   J.  Song,  X.  Niu,  L.  Ling and  B.  Wang: Fuel Process. Technol., 115 (2013), 26.
• 37)   C.  Dong,  S.  Sheng,  Q.  Wu,  Q.  Lu,  Y.  Zhao,  X.  Wang and  J.  Zhang: Appl. Surf. Sci., 257 (2011), 8647.
• 38)   Y. K.  Rao: Metall. Trans., 2 (1971), 1439.
• 39)   E. T.  Turkdogan and  J. V.  Vinters: Metall. Trans., 3 (1972), 1561.
• 40)   J.  Moon and  V.  Sahajwalla: Metall. Mater. Trans. B, 37 (2006), 215.
• 41)   J.  Yang,  T.  Mori and  M.  Kuwabara: Trans. Iron Steel Inst. Jpn., 47 (2007), 1394.
• 42)   N. S.  Srinivasan and  A. K.  Lahiri: Metall. Trans. B, 8 (1977), 175.
• 43)   P.  Semberg,  A.  Rutqvist and  C.  Andersson: ISIJ Int., 53 (2013), 391.
• 44)   A. A.  Elgeassy,  M. I.  Nasr,  A. A.  Omar and  E. A.  Mousa: ISIJ Int., 48 (2008), 1359.