In situ Observation of Reduction Behavior of Multicomponent Calcium Ferrites by XRD and XAFS

Abstract

Reduction behavior of various multi-component calcium ferrites at 900°C were investigated by using in situ X-ray diffraction (XRD) and X-ray absorption spectroscopy. Reaction products (intermediate components) were determined by XRD and change in X-ray absorption spectra of Fe and Ca K-edges were analyzed to determine reaction rate constants. SFCA-I (Ca₃(Ca,Fe)(Fe,Al)₁₆O₂₈) and SFCA (Ca₂(Fe,Ca)₆(Fe,Al,Si)₆O₂₀) consist of layered structure of spinel and pyroxene. Early stage of reduction reaction, diffraction peaks of spinel structure were observed which indicating SFCA-I and SFCA decomposed into these units at the first step of the reduction reactions. The spinel was reduced sequentially into FeOx then Fe. Intermediate component, Ca₂(Fe,Al)₂O₅ originated in pyroxene module was hard to reduce and reaction was controlled by decomposition of this phase. Reduction of SFCA-I started later than SFCA (with 5.7 mol% Al₂O₃) under hydrogen gas reduction condition at 900°C. SFCA with a high aluminum content indicated lower reducibility than that with a low one.

1. Introduction

Iron ore sinter is the main iron source for the blast furnace process in the Asia–Pacific region. Recently, the concentration of gangue minerals in iron ores has been increasing owing to resource degradation. Thus, the development of a process technology for low-grade iron ore or pellet feeds that also maintains the sinter yield and/or quality is required.

The microtexture of a sinter is composed of iron ore grains, a bonding layer, pores, and cracks. The bonding layer includes quasi-binary calcium ferrites (CaO–Fe₂O₃), multicomponent calcium ferrites (silico-ferrites of calcium and aluminum), calcium silicate glass (slag), and those solid solutions.¹⁾ Formation of the microtexture of a sinter as a final product is strongly affected by the various reactions and phase production occurring during the heating and cooling processes.^2,3)

Ca₂(Fe,Ca)₆(Fe,Al,Si)₆O₂₀ (SFCA) and Ca₃(Ca,Fe)(Fe,Al)₁₆O₂₈ (SFCA-I) phases are representative multicomponent calcium ferrite phases, which are included in a sinter. The crystal structures of SFCA and SFCA-I are related to the aenigmatite group minerals and can be described as M_14+6nO_20+8n (n = 0, 1), respectively.

The chemical formulae of the SFCA and SFCA-I phases are A₂M₆T₆O₂₀ and A₃BM₈T₈O₂₈ (A = Ca²⁺; B = Ca²⁺, Fe²⁺; M (octahedral site) = Fe³⁺; T (tetrahedral site) = Fe³⁺, Al³⁺, and Si⁴⁺). As shown in Fig. 1,⁴⁾ these phases have layered structures of spinel (S) and pyroxene (P), and the patterns in SFCA and SFCA-I are -S-P-S-P- and -S-S-P-S-S-P, respectively. Furthermore, the presence of an SFCA-II phase with an -S-S-P-S-P- pattern and an SFCA-III phase (M_14+6nO_20+8n (n = 2)) with an -S-S-S-P- pattern was reported.^5,6) Although an SFCA-II phase excluding Si was reported as a continuous solid solution of CaAl₃O₁₀–CaFe₃O₁₀ with an Fe₂O₃ range of 44.5–81.5 mol%,^7,8) its stable compositional range has not yet been clarified. In addition, the presence of SFCA-II and SFCA-III in a sinter has not yet been reported.

Fig. 1.

Crystal structure of SFCA, indicating layered structure of spinel (S) and pyroxene (P) modules.⁴⁾ (Online version in color.)

The above phases are solid solutions with a substitutional relationship of 2(Fe³⁺, Al³⁺) = (Ca²⁺, Fe²⁺) + Si⁴⁺ to maintain the charge valance and have a wide solution range.⁹⁾ However, their compositional ranges and chemical properties have not yet been fully clarified. In addition, single-crystal structure analyses of an SFCAM phase of a Mg-solute phase were conducted.⁹⁾

Generally, the reducibility of a sinter is evaluated using the reduction index (RI) (JIS M8713:2017 based on ISO 7215). In this method, the bulk sinter is isothermally reduced at 900°C for 180 min in a N₂–30 vol% CO atmosphere. Therefore, the indices are affected by not only the intrinsic reducibility of the individual phases by chemical reactions but also by various topological factors such as micropores and the morphology and/or microstructure of each phase in the sinter. On the other hand, a study on the intrinsic reduction behavior of individual phases in the sinter is required.

Regarding a quasi-binary system, the Baur–Glaessner diagram,¹⁰⁾ which indicates the phase equilibrium for the oxygen potential and the temperature, was experimentally studied well. Kimura et al.¹¹⁾ studied the changes in the valence states and local structure of Fe and Ca in quasi-binary calcium ferrites by X-ray absorption spectroscopy and derived their reduction rates. For multicomponent calcium ferrites, an experimental phase diagram around the SFCA region was reported;¹²⁾ however, the compositional region of SFCA-I has not been completely clarified yet.

Sugiyama et al.¹³⁾ reported that the reduction products of the SFCA-I and SFCA phases after 48 h of reduction under an oxygen pressure of 1 × 10^−9.1 Pa, were FeO+Fe₃O₄+Ca₂Fe₂O₅ and Ca₂SiO₄+FeO+Fe₃O₄+Ca₂Fe₂O₅, respectively. Maeda et al.¹⁴⁾ researched the relationship between the final reduction temperature of a quasi-ternary calcium ferrite (71.4Fe₂O₃–15.1CaO–5.7SiO₂–7.8Al₂O₃ (mass%)) and CO/CO₂ partial pressure. Maruoka et al.¹⁵⁾ studied the reducibility of the SFCA and SFCA-I phases with various compositions from 200°C to 800°C, and reported that they were not reduced at 800°C in CO–CO₂, whereas their reduction reactions were promoted under CO–CO₂–H₂–H₂O. In addition, the Fe concentration did not affect the reducibility of SFCA, whereas its increase promoted the reducibility of SFCA-I.

Multicomponent calcium ferrites with aenigmatite-type structures easily form twins owing to their crystallographic properties. In contrast, those in a sinter are crystallized from oxide melts and form various morphologies such as columnar, needle, and fine textures depending on the formation process. Furthermore, the surrounding texture is varied including silicate slag and micropores. It is considered that the surrounding gangue affects the composition and basicity of melts and the formation process of calcium ferrites.^16,17,18,19)

It is considered that the reducibility of calcium ferrites is affected by both the composition and fine texture. Sakamoto et al.²⁰⁾ investigated the relationship between the sinter structure and its reducibility quantitatively. They proved that a one-interface unreacted core model considering the diffusion chemical reactions in the gas boundary film and the diffusion process of the products was relatively in good agreement with the experimental results. They also tested the reducibility of a single mineral phase typically included in a sinter, and found that the reducibility of fine-type hematite and calcium ferrite was higher than that of a prismatic-type calcium ferrite in the slag melt. Cai et al.²¹⁾ compared the reducibility of columnar and needle calcium ferrites and reported that needle calcium ferrites surrounded by micropores present better reducibility than columnar calcium ferrites surrounded by a slag. Furthermore, Murakami et al.²²⁾ investigated the effects of coexistence of textures on reducibility and reported that reducibility decreases in the order of primary hematite > acicular calcium ferrite (ACF), secondary hematite > calcium ferrite, magnetite > ACF, magnetite > columnar calcium ferrite. Nicol et al.²³⁾ reviewed the crystal structure, morphology, and formation process of SFCA in detail.

The aim of this study is to clarify the effects of the crystal structures and compositions of multicomponent calcium ferrites on their reduction behavior. In situ observation and analysis of the high-temperature reduction process at 900°C of the multicomponent calcium ferrites with various Al₂O₃ solid solutions were conducted. Intermediate compounds were identified by X-ray diffraction (XRD), changes in the valence state and coordination number of Fe and Ca during the high-temperature reduction reaction were analyzed by in situ X-ray absorption fine structure (XAFS), and the reduction rates were determined. The reduction processes were studied by considering all results of high-temperature XRD, XAFS, and thermogravimetry (TG). Powdered single phases with few micropores prepared by the powder sintering method were adopted for the reduction experiments to avoid the effect of texture. The flux of the supplied gas to the sample chamber of the furnace for the reduction experiments was sufficiently high to avoid the gas supply-controlled reaction. The reduction reaction rate was obtained based on the assumption that it was a chemical reaction controlling process.

2. Experimental

2.1. Sample Preparation

Six types of calcium ferrites were synthesized by the conventional powder sintering method. The starting materials were α-Fe₂O₃ (Kojundo Chemical Lab., 4 N, 1 μm), CaCO₃ (Kanto Chemical, >99.5%, 12–15 μm), SiO₂ (quartz-type, Kojundo Chemical Lab., 99%, 1 μm), and α-Al₂O₃ (Kojundo Chemical Lab., 4 N, 1 μm). They were mixed with an agate pestle and mortar with the nominal compositions listed in Table 1. The samples were first pressed into pellets and calcined at 800°C for 2 h for decarboxylation, and subsequently crushed, pelletized, and sintered from 1180°C to 1230°C for 48 h. The obtained calcium ferrites were crushed and powdered for the reduction experiments. Except for the SFCA-I sample, which included small amounts of Fe₂O₃, the samples were almost single phases with less than 1% inclusions, based on XRD measurements.

Table 1. Chemical composition of calcium-ferrites for reduction observation (mol%).

Sample	Fe2O3	CaO	Al2O3	SiO2
CaFe2O4	50.0	50.0	–	–
Ca2Fe2O5	33.3	66.6	–	–
Ca2(Al0.5,Fe0.5)2O5	16.67	66.67	16.67	–
SFCA-I	67.69	25.59	6.72	–
SFCA (5)	56.09	30.35	5.74	7.82
SFCA (15)	50.37	27.34	17.11	5.17

*SFCA-I: Ca₃(Ca,Fe)(Fe,Al)₁₆O₂₈, SFCA: Ca₂(Fe,Ca)₆(Fe,Al, Si)₆O₂₀

For comparison, TG-differential thermal analysis (DTA) measurements of reagent grade α-Fe₂O₃ (Kojundo Chemical Lab., 4 N, 1 μm) were conducted. For Fe K-edge XAFS, iron foil and iron oxides were measured, and for Ca K-edge XAFS, CaO and CaSiO₃ (Aldrich, −200 mesh, 99%) were measured. The CaO sample was prepared by decarboxylation of reagent-grade CaCO₃ at 1100°C for 5 h immediately before the XAFS measurements. The CaSiO₃ sample was a mixture of crystalline and amorphous phases.

2.2. TG Measurements

A powder sample (20 mg) was filled in an Al₂O₃ pan, and TG-DTA measurements were conducted. Corrosion-proof sample holders with R-type thermocouple gage contacts in Al₂O₃-protecting tubes were used. The gas pressure and the flow rate were set as 0.1 MPa and 150 mL/min, respectively. The sample temperature was elevated to 900°C under an Ar–20 vol% O₂ atmosphere, and subsequently the flow gas was changed to Ar–3 vol% H₂ and isothermally reduced from 60 to 150 min by measuring the change in the sample weight.

2.3. In situ XRD Measurements

Each powder sample was filled in a black quartz glass sample holder with a depth of 0.3 mm and heated up to 900°C in an infrared heating attachment equipped on an X-ray diffractometer. The sample temperature was elevated to 900°C in air and subsequently isothermally reduced under a N₂–2 vol% H₂ atmosphere. The gas pressure and the flow rate were set as 0.1 MPa and 150 mL/min, respectively, and the reduction experiments were conducted under conditions in which the reduction gas was dispersed sufficiently rapidly.

Using a sealed X-ray tube with a Cu anode (tube voltage 40 kV, tube current 40 mA) as an X-ray source and a one-dimensional silicon strip detector as a detector, XRD measurements were repeatedly conducted in the range of 2θ = 10°–45° with a scan speed of 40°/min and 40 s cycles.

The sample temperature was monitored using a K-type thermocouple with a sheathe inserted into the sample holder.

2.4. In situ XAFS Measurements

In situ XAFS experiments were conducted at BL-9A of Photon Factory, Institute of Materials Structure Science, High Energy Accelerator Research Organization. The photon flux density of BL-9A at photon energy E = 6.5 keV is approximately 7 × 10¹¹ photons/s, for a ring energy of 2.5 GeV and a ring current of 450 mA.

For short-period measurements, a quick-XAFS method, in which a Si (111) double crystal monochromator was continuously scanned, was used, and Fe or Ca K-edge XAFS spectra under a high-temperature reduction atmosphere were collected. The flow gas-type cell used for in situ observation has been reported elsewhere;^11,24) therefore, only detailed measurement conditions are described in this paper.

For the Fe K-edge XAFS measurements, the starting material was diluted with 80 mg of BN powder to ensure the edge jump Δμt (= −lnI/I₀) of its spectra was approximately 1. Subsequently the starting material was filled and hand pressed within a cylindrical stainless steel tube holder with an inner diameter of 7 mm. It should be noted that μ is the linear absorption coefficient, t is the sample thickness, I is the transmitted X-ray intensity of the sample, and I₀ is the incident X-ray intensity.

For the Ca K-edge XAFS measurements, the amount of BN was set as 20 mg, and the adjusted differential of μt between the starting and stopping points did not exceed 4, owing to the large X-ray absorption of BN. A Ni-coated mirror was used to achieve higher harmonic reduction at the Ca K-edge.

The sample temperature was elevated up 900°C in an oxidization atmosphere and subsequently isothermally reduced under a He–20 vol% H₂ atmosphere. The gas pressure and the flow rate were set as 0.1 MPa and 200 mL/min, respectively, and the reduction experiments were conducted under conditions in which the reduction gas was dispersed sufficiently rapidly. XAFS spectra were repeatedly collected with a cycle of 21 s for the Fe K-edge and 51 s for the Ca K-edge.

XAFS measurements at the Fe K-edge were conducted in the energy range of 6911 eV < E < 7581 eV (wavenumber range of photon electrons 0 nm⁻¹ < k < 120 nm⁻¹) by angle scanning of a Si(111) monochromator (16.9° < θ < 14.0°) at a scan speed of 21s per spectrum. For high-temperature measurements at the Ca K-edge, only spectra around the X-ray absorption near edge structure (XANES) region (3990 eV < E < 4140 eV, 31.0° < θ < 26.5°) were collected to ensure sufficient S/N. Moreover, extended XAFS (EXAFS) spectra were measured at room temperature before and after reduction.

X-ray absorption spectroscopy data processing and analysis were conducted using the analysis software, ATHENA (Demeter 0.9.25).²⁵⁾ Background curves were fitted and corrected using Victreen’s approximation (Cλ³ − Dλ⁴ + const.). EXAFS oscillations were extracted by extrapolation using a spline function (Cook–Sayers). The calculated k³χ(k) was Fourier transformed in the range of 20 nm⁻¹ < k < 120 nm⁻¹ for Fe and of 20 nm⁻¹ < k < 120 nm⁻¹ for Ca, and radial distribution functions (RDFs) were obtained.

3. Results and Discussion

3.1. TG Curves

Figure 2 shows the reduction rate curves obtained from the TG curves measured after reduction is started at 900°C and the first derivative of each TG curve. In this study, the reduction rate was defined as the ratio of removed oxygen to oxygen bonds with iron before reduction, and iron before reduction was treated as trivalent. The starting time of the reduction could be determined from the form of the first derivative of the TG curve. The weight reduction of Fe₂O₃ started at 18 min after the change in the flow gas to Ar–H₂ gas, and it reduced via Fe₃O₄ and FeO into metallic Fe until 60 min. The reduction starting times of CaFe₂O₄ and SFCA-I were similar to that of Fe₂O₃. As for multicomponent calcium ferrite samples listed in Table 1, the reduction starting time of SFCA(5) was 16 min, which was earlier than those of SFCA-I and SFCA(15) (20 min for latter).

Fig. 2.

Reduction time dependence of degree of reduction and differential of thermogravimetoric curves of samples shown in Table 1 in Ar-3vol%H₂ atmosphere at 900°C. A triangle indicates presence of slopes on curves. Broken lines are for eyes. Numbers show the types of coexisting phases. See the text for the details. (Online version in color.)

CaFe₂O₄, SFCA-I, and SFCA underwent at least three steps of reaction, as determined from the several inflection points observed in the first derivates of the TG curves. The reduction process of Ca₂Fe₂O₅ was one step, and its starting time was similar to that of the third step of the reduction process of CaFe₂O₄. The reduction process of CaFe₂O₄ progressed as CaFe₂O₄ → (1) CaFe₃O₅ + Ca₂Fe₂O₅ → (2) CaFe₅O₇ + Ca₂Fe₂O₅ → (3) Ca₂Fe₂O₅+FeO → (4) Ca₂Fe₂O₅ + Fe → (5) CaO + Fe, which can be derived from the Baur–Glaessner diagram.¹⁰⁾ In this reaction process, the decomposition of Ca₂Fe₂O₅ was the reaction controlling step; therefore, a three-step reaction appeared in the first derivatives of the TG curves. The corresponding steps were as follows: first step CF → [(1)] → (2), second step (2) → (3), and third step: (3) → [(4)] → (5). Therefore, the third step of the reduction process of CaFe₂O₄ is equivalent to the reduction of Ca₂Fe₂O₅. This result corresponds to the time-dependent changes in the XRD patterns discussed in the subsequent section.

Ca₂(Al_0.5,Fe_0.5)₂O₅ was difficult to reduce, and its reduction starting time was 45 min. The reaction ends of CaFe₂O₄ and Ca₂Fe₂O₅ were 120 min, and the reaction of SFCA was considerably gradual, occurring at approximately 75 min.

3.2. In situ XRD

Figure 3 shows time dependence of the in situ XRD patterns of CaFe₂O₄ in a N₂–2 vol% H₂ atmosphere at 900°C. The intermediate products were identified as CaFe₃O₅, CaFe₅O₇, and Ca₂Fe₂O₅. The phases changed as Ca₂Fe₂O₅ + FeO → Ca₂Fe₂O₅ + CaO + FeO + Fe → CaO + Fe as the reduction progressed. Coexistence of Ca₂Fe₂O₅, FeO, and Fe was observed, which indicated that reduction of the hardly reductive Ca₂Fe₂O₅ was the reaction controlling step. The above reaction process corresponded well with the Baur–Glaessner diagram¹⁰⁾ and the results of the TG study.

Fig. 3.

In situ XRD patterns of CaFe₂O₄ in N₂-2 vol%H₂ atmosphere at 900°C with an interval of 1.5 min. Numbers show the types of coexisting phases. (Online version in color.)

Figures 4(a) and 4(b) show the time dependence of the in situ XRD patterns of SFCA-I and SFCA(5) in a N₂–2 vol% H₂ atmosphere at 900°C, respectively. Figure 5 shows the XRD patterns of CaFe₂O₄, SFCA-I, SFCA(5), and SFCA(15) after 3 min of reduction. During the reduction of the SFCA-I phase, Fe₃O₄ and Ca₂(Fe,Al)₂O₅ were observed as the intermediate products, and Fe₃O₄ was reduced to metallic Fe via FeO. The diffraction peak intensity of Ca₂(Fe,Al)₂O₅ was nearly unchanged after 90 min of reduction, and the formation of CaO was not observed, which was different from the case of Ca₂Fe₂O₄. Regarding the reduction of SFCA(5), the diffraction peaks of SFCA(5) almost disappeared after 3 min of reduction, whereas those of Fe₃O₄, FeO, and Ca₂(Fe,Al)₂O₅ were observed. SFCA(5) became a mixture of Fe, FeO, Ca₂(Fe,Al)₂O₅, and Ca₂SiO₄ for 90 min. The diffraction peak intensity of Ca₂(Fe,Al)₂O₅ tended to weaken. The X-ray scattering intensity at approximately 2θ = 35° was relatively high, indicating the possibility of the formation of amorphous Ca–Si–O. Although the reduction rate of SFCA(15) was lower than that of SFCA(5), its intermediate products were the same; therefore, the reduction process of SFCA(15) was considered to be the same as that of SFCA(5). It was considered that the crystal structure of the SFCA phase was more difficult to reduce owing to the higher Al content.

Fig. 4.

In situ XRD patterns of (a) SFCA-I and (b) SFCA in N₂-2 vol%H₂ atmosphere at 900°C with an interval of 1.5 min. (Online version in color.)

Fig. 5.

In situ XRD patterns of CaFe₂O₄, SFCA-I, SFCA(5) and SFCA(15) at 900°C, reduced for three minutes. (Online version in color.)

3.3. In situ XAFS

3.3.1. Change in Local Structure around Fe

Figure 6(a) shows the Fe K-edge XAFS spectra of CaFe₂O₄, SFCA-I, and SFCA(5) measured at room temperature. Red lines represent spectra before reduction and blue lines represent spectra after reduction for 80 min. The spectra of Fe and the Fe oxides are also shown as references. The absorption edge energy increased with the increase in the valence number of Fe, and the intensity of the white line (the peak immediately above the absorption edge) of the oxides tended to be higher than that of metallic Fe. The pre-edge peak energy position was Fe²⁺ < Fe³⁺, and the pre-edge peak intensity was four-coordinated > six-coordinated.

Fig. 6.

(a) Fe K-edge XAFS spectra of Fe, Fe oxides, CaFe₂O₄, SFCA-I and SFCA(5). Red lines indicate spectra before reduction and blue lines after reduction for 80 minutes. (b) Environmental radial distribution functions (RDFs) obtained from XAFS spectra shown in Fig. 6(a). (Online version in color.)

Before reduction, the Fe atoms included in both CaFe₂O₄ and SFCA were trivalent. SFCA contains a small amount of Fe²⁺; however, it could not be detected in the EXAFS measurements. Fe atoms in the SFCA structure are distributed both in oxygen four-coordinated and six-coordinated sites; therefore, its pre-edge peak intensity was stronger than those of CaFe₂O₄ and Fe₂O₃, in which all Fe atoms are six-coordinated with oxygen.

The spectrum form of each sample after reduction agreed well with that of metallic Fe. Figure 6(b) shows the environmental RDFs around Fe, as obtained by Fourier transformation of the XAFS spectra. In the RDFs of the samples after reduction, the correlation peak of the first neighbor of Fe–O, which was observed before reduction, almost disappeared and an Fe–Fe correlation peak emerged.

Figures 7(a)–7(c) show the time dependence of the Fe K-edge XAFS spectra of CaFe₂O₄, SFCA-I, and SFCA(5) in a He–20 vol% H₂ atmosphere at 900°C, respectively. After 20 min of reduction, the spectrum of CaFe₂O_s is similar to that of metallic Fe. In contrast, those of SFCA-I and SFCA are similar to the spectra of a mixture of oxides and metallic iron.

Fig. 7.

In situ Fe K-edge XANES patterns of (a) CaFe₂O₄, (b) SFCA-I and (c) SFCA(5) in He-20 vol%H₂ atmosphere at 900°C. Allows indicate isosbestic points at 7120 eV and 7143 eV and a peak position at 7125 eV. (Online version in color.)

Figure 8(a) shows the reduction time dependence of the normalized absorbances at 7120 eV, 7125 eV, and 7143 eV of SFCA(5), which are indicated by arrows in Fig. 7(c). The starting time of the H₂ gas flow was defined as 0 min on the time scale. At least three steps of intensity changes were observed from the start of the reduction gas flow to 19 min. Thus, SFCA(5) was reduced to FeO via several intermediate products including both Fe²⁺ and Fe³⁺. The spectra from 5 min to 10 min were similar to that of Fe₃O₄, which is in agreement with the in situ XRD results, in which diffraction peaks of Fe₃O₄ were observed. After 19 min of reduction, isosbestic points were observed at 7120 eV and 7143 eV.

Fig. 8.

(a) Reduction time dependence of normalized absorbance at 7120 eV, 7125 eV and 7143 eV of SFCA(5). A curve indicates the least square fitting result. (b) Reduction time dependence of normalized absorbance of white line of CaFe₂O₄, SFCA-I and SFCA(5). (Online version in color.)

The XAFS spectrum of a mixture is formed by adding the spectra of the including phases, and its absorbance is the summation of the products of the molar concentrations and absorbances of the including phases. An isosbestic point can be observed when a substance gradually changes into another substance, or when only two phases co-exist during the reduction.

Therefore, in this study, the reaction after 19 min was considered as a quasi-first order reaction: [Fe²⁺ ⇄ (1 − f)Fe²⁺ + f Fe⁰], and the reaction rate was calculated using the rate equation of the first-order reaction; [A] = [A]₀e^-k₁x. In the equation, [A] is the molar concentration of the reactant, [A]₀ is the initial molar concentration of the reactant, and x is the reaction time. For the reduction process of SFCA(5), [A] corresponds to the concentration of FeO. Specifically, least-square fitting using a function μt = y₀ + ae^k₁(x-x₀) was applied to a dataset of normalized absorbance μt at 7125 eV after 19 min, where μt is the absorbance, y₀ and a are constants, x is the reaction time, and x₀ is the reaction start time. In case of SFCA(5), x₀ = 19 min. Consequently, rate constant k₁ was calculated as 0.31(2) min⁻¹.

An isosbestic point was observed in the reduction of CaFe₂O₄ after 13 min. In this region, reduction reactions of Ca₂Fe₂O₅ → 2CaO + 2FeO and FeO → Fe occurred successively. Ca₂Fe₂O₅ → 2CaO + 2FeO controlled the reaction, and the formed FeO rapidly reduced to Fe. Therefore, under assumption of a quasi-first-order reaction, Ca₂Fe₂O₅ → 2CaO+2Fe progressed after 13 min, and the dataset of normalized absorbance μt at 7125 eV after 13 min was analyzed similar to as mentioned above. Consequently, rate constant k₁ was calculated as 0.147(2) min⁻¹.

Figure 8(b) shows the reduction time dependence of the normalized absorbances of the white lines of CaFe₂O₄, SFCA-I, and SFCA(5). The change in the absorbance of SFCA-I was slower than in those of CaFe₂O₄ and SFCA(5). The absorbance dataset after 33 min, in which isosbestic points were observed at 7125 eV and 7145 eV, was fitted by the rate equation of a first-order reaction, and rate constant k₁ was calculated as 0.037(2) min⁻¹. From the changes in the products observed in the XRD patterns, as shown in Fig. 4, this reaction corresponds to the reduction reaction, FeO → Fe. A possible cause of the different behavior of SFCA(5) is that SFCA-I produces a larger amount of Ca₂(Fe,Al)₂O₅, owing to which the solid solution of Ca into the spinel module is relatively higher. It is considered that the larger amount of the Ca and Al solution in FeO, which forms during the reduction of spinel, is one of the causes of the lower reduction rate of SFCA-I compared to that of SFCA(5).

3.3.2. Change in Local Structure around Ca

Figure 9(a) shows the Ca K-edge XANES spectra of CaFe₂O₄, SFCA-I, and SFCA(5) before and after reduction for 60 min. Figure 9(b) shows the environmental RDFs around Ca calculated by Fourier transformation of the EXAFS spectra in the range of 20 nm⁻¹ < k < 100 nm⁻¹. The spectra of CaO, CaSiO₃, and Ca₂Fe₂O₅ are also shown for comparison. During the reduction of the calcium ferrites by hydrogen, the valence of Ca is constantly divalent, and only the coordination structure of oxygen is changed. The Ca atoms in CaFe₂O₄ are oxygen eight-coordinated. The spectrum after reduction indicates that they are oxygen six-coordinated. In the environmental RDF, the profile up to 0.6 nm including the Ca–O correlation peak of the first neighbor and the Ca–Ca correlation peak of the second neighbor, agree well with those of CaO, proving the formation of CaO by the reduction.

Fig. 9.

(a) Ca K-edge XANES spectra of CaO, CaSiO₃, Ca₂Fe₂O₅, CaFe₂O₄, SFCA-I and SFCA(5). Red lines indicate spectra before reduction and blue lines after reduction for 60 minutes. (b) Environmental radial distribution functions (RDFs) obtained from XAFS spectra shown in Fig. 9(a). (Online version in color.)

In contrast, in the crystal structure of SFCA-I, Ca atoms are dispersed in six-coordinated and two distorted seven-coordinated sites and a part of six-coordinated Fe sites. In the crystal structure of SFCA, Ca atoms are dispersed at distorted eight-coordinated sites and a part of six-coordinated Fe sites. The spectra of the reduced SFCA-I and SFCA(5) present similar profiles to that of CaSiO₃ or Ca₂Fe₂O₅ before reduction. Although the correlation peaks in the RDFs are similar, they do not agree with those of Ca₂Fe₂O₅, suggesting the former are mixtures.

Figures 10(a)–10(c) show the time dependence of the Ca K-edge XAFS spectra of CaFe₂O₄, SFCA-I, and SFCA(5) in a He–20 vol% H₂ atmosphere at 900°C, respectively. In the spectrum of CaFe₂O₄ after 10 min, the peak shoulder at approximately 4040 eV disappeared and a feature of CaO emerged. Isosbestic points were identified at 4049 eV and 4053 eV from the initial stage of the reduction. It was suggested that the local structure around Ca changed by a first-order reaction. Two peaks at 4045 eV and 4050.5 eV observed in the spectra of SFCA-I and SFCA(5) disappeared, and a peak at 4045.7 eV grew until 10 min of reduction. Although isosbestic points occurred at 4048.5 eV and 4055 eV in both samples, the high noise in the spectra made it difficult to confirm.

Fig. 10.

In situ Ca K-edge XANES patterns of (a) CaFe₂O₄, (b) SFCA-I and (c) SFCA(5) in He-20 vol%H₂ atmosphere at 900°C. Arrows indicate isosbestic points at 4048.5 eV and 4055 eV. (Online version in color.)

Figure 11 shows normalized absorbances of CaFe₂O₄ at 4045.7 eV and of SFCA(5) and SFCA-I at 4045.7 eV. The Ca atoms in CaFe₂O₄ and the intermediate reaction products (CaFe₃O₅, CaFe₅O₇ and Ca₂Fe₂O₅) are oxygen eight-coordinated, and those in the final product, CaO, are oxygen six-coordinated structure. Assuming that local structure around Ca changes from eight-coordinated into six-coordinated by a first-order reaction, the rate constant was calculated by fitting of the normalized absorbance at 4047 eV.

Fig. 11.

Reduction time dependence of normalized absorbance of CaFe₂O₄ at 4047 eV, SFCA-I and SFCA(5) at 4045.7eV, respectively. Curves indicate the least square fitting results. (Online version in color.)

For SFCA-I and SFCA(5), the changing process of the coordination environment around Ca by the decomposition of the pyroxene and spinel modules in the initial reaction was considered as the first-order reaction. For SFCA(5), analyses of the initial reaction and the latter part of the reaction separately was considered to be effective. This was because Ca₂(Fe,Al)₂O₅ and Ca₂SiO₄ are formed in the initial stage, of which Ca₂(Fe,Al)₂O₅ was further reduced to (Fe,Al,Ca)O_x in the latter part of the reaction. However, because of the low quality of the spectra obtained in this study, this reaction was treated as a quasi-first-order reaction, and a fitting analysis was conducted. Consequently, the rate constants of the local structural change around Ca of CaFe₂O₄, SFCA-I, and SFCA(5) were determined as k₁ = 0.19(3), 0.16(3), and 0.26(3) min⁻¹, respectively.

3.4. Reduction Process of Multicomponent Calcium Ferrites

The crystal structure of the SFCA phase is a layered structure of alternating spinel (Fe₃O₄) and pyroxene (Ca(Fe,Ca)(Fe,Al,Si)₂O₆) modules. The oxide products including Ca, Al, and Si originated from the pyroxene structure. The results of the equilibrium experiments under a low oxygen partial pressure indicated that multistep reactions occurred during the initial reaction caused by the decomposition of the module structure and the progress of the reduction of the spinel structure.¹³⁾ The presence of diffraction peaks of the spinel structure (Fe₃O₄) at the beginning of the reduction reactions of SFCA-I and SFCA(5) in their high-temperature XRD indicated that decomposition of the module structure occurred at the earliest stage of the reduction reaction.

Table 2 lists the intermediate products and reaction controlling steps of the reduction of the calcium ferrites determined by the high-temperature XRD and XAFS analysis results. Although reduction speed of the spinel structure with a Ca and Al solution was lower than that of Fe₃O₄, it was higher than that of Fe in pyroxene. The pyroxene structure including Si decomposed into Ca–Si–O and Ca₂(Fe,Al)₂O₅. With the progress of the reduction of SFCA(5), the diffraction peak intensity of Ca₂SiO₄ increased, whereas that of Ca₂(Fe,Al)₂O₅ decreased.

Table 2. Reduction process of calcium ferrites at 900°C by hydrogen gas.

Starting materials	Intermediate products	Final products	Reaction control
CaFe2O4	CaFe3O5+Ca2Fe2O5 CaFe5O7+Ca2Fe2O5 Ca2Fe2O5+FeO Ca2Fe2O5+CaO+Fe	CaO, Fe	Decomposition of Ca2Fe2O5
SFCA-I	(Fe,Al,Ca)3O4+Ca2(Fe,Al)2O5 (Fe,Al,Ca)Ox+Ca2(Fe,Al)2O5 Fe+Ca2(Fe, Al)2O5	Fe, Ca2(Fe,Al)2O5	Decomposition of Ca2(Fe,Al)2O5
SFCA	(Fe,Ca,Al)3O4+Ca2(Fe,Al)2O5+ Ca2SiO4+Ca–Si–O (Fe,Al,Ca)Ox+Ca2(Fe,Al)2O5 Ca2SiO4+Ca–Si–O Fe+Ca2(Fe,Al)2O5+ Ca2SiO4+Ca–Si–O	Fe, Ca2SiO4, Ca2(Fe, Al)2O5	Decomposition of (Fe,Al,Ca)Ox and Ca2(Fe,Al)2O5

In contrast, the diffraction peak intensity of Ca₂(Fe,Al)₂O₅ observed during the reduction of SFCA-I, which has a higher Fe/lower Ca content than SFCA(5), did not change even after 90 min of reduction. (Ca,Fe²⁺)₂(Fe,Al)₂O₅ with the same crystal structure type of Ca₂(Fe,Al)₂O₅ could be present under a low oxygen partial pressure, and it may be more stable than Ca₂(Fe,Al)₂O₅ in SFCA(5), depending on the compositional ratio of Ca, Al, and Fe²⁺.

Based on these experimental results, the reduction reactions of the SFCA-I and SFCA phases were considered to start by the decomposition of the layered structure of the spinel–pyroxene modules, followed by rapid reduction of the Fe atoms in the spinel into metal. This is shown schematically in Fig. 12.

Fig. 12.

Schematic illustration of reduction route of SFCA. (Online version in color.)

Accordingly, the intermediate products of the reduction of the calcium ferrites could be determined by in situ observation by high-temperature XRD. The XAFS spectra analysis was effective for the characterization of the low-crystallinity materials and the discussion of the kinetics. Further qualitative and quantitative analyses of the elemental reduction reactions that progress in a blast furnace could be realized by combining in situ XAFS, XRD, and thermal analysis. It is also important to analyze the composition of the products by microtexture analysis and examine the distributions of Al and Ca in each phase; however, this has not been investigated in this study because only powdered samples were used.

4. Conclusion

In this study, the reduction behavior of multicomponent calcium ferrites was comprehensively analyzed by observation of oxygen desorption via TG, determination of the intermediate products by in situ powder XRD, and changes in the local structure around Fe or Ca by in situ XAS.

All results were quantitatively analyzed and combined to consider the reaction process. The reaction temperature was set as 900°C based on the testing temperature of the RI (JIS-RI) of an iron ore sinter. It was clarified that the reduction reaction of SFCA-I and SFCA started from the decomposition of the layered structure of the spinel–pyroxene modules, followed by rapid reduction of the Fe atoms in the spinel to metallic Fe. In the case of the SFCA-I phase, which does not include Si, Ca₂(Fe,Al)₂O₅ was formed and reduced gradually. The SFCA phase, which includes Si, was decomposed into the hardly reductive Ca₂(Fe, Al)₂O₅ and Ca–Si–O.

Analysis of the reduction behavior of complex oxides with complex crystal structures, such as the SFCA phase, consideration of the characteristics of the crystal structure makes it easier to understand the elemental reaction.

In this study, reduction at a fixed temperature of 900°C by hydrogen was studied; however, an actual blast furnace reaction progresses under a CO–CO₂–H₂–H₂O system, and both the oxygen partial pressure and temperature change from the initial to the final state of the reaction. It was reported that the reduction of SFCA-I starts at a lower temperature than that of SFCA under an intermediate oxygen partial pressure.¹⁵⁾ Further studies on the elemental reduction temperature and atmosphere dependence of each phase in a sinter are required to estimate the reactions in the blast furnace, which are future problems.

Acknowledgement

The X-ray absorption measurements were conducted under a research collaboration between Nippon Steel Corporation and the High Energy Accelerator Research Organization (Proposal Nos. 2012C202, 2013C209, and 2015C206). We thank Associate Professor Kenichi Kimijima and Dr. Yohei Uemura (currently belonging to Paul Scherrer Institute) of KEK and Mr. Kengo Noami and Mr. Yu Nemoto of Nippon Steel Technology Co. Ltd. for their technical support.

References

• 1)   T.  Inazumi: Iron Ore Sinter-Review of Steps Taken to Overcome the Challenges Posed Japan’s Lack of Iron Ore Resources-, ISIJ, Tokyo, (2000), 239 (in Japanese).
• 2)   F.  Matsuno and  T.  Harada: Trans. Iron Steel Inst. Jpn., 21 (1981), 318. https://doi.org/10.2355/isijinternational1966.21.318
• 3)   A.  Yazawa,  Y.  Takeda and  Y.  Waseda: Can. Metall. Q., 20 (1981), 129. https://doi.org/10.1179/cmq.1981.20.2.129
• 4)   R.  Murao,  T.  Harano,  M.  Kimura and  I.-H.  Jung: ISIJ Int., 58 (2018), 259. https://doi.org/10.2355/isijinternational.ISIJINT-2017-459
• 5)   W. G.  Mumme: Neues Jahrb. Miner. Abh., 178 (2003), 307. https://doi.org/10.1127/0077-7757/2003/0178-0307
• 6)   V.  Kahlenberg,  H.  Krüger and  V. S.  Goettgens: Acta Crystallogr. B, 75 (2019), 1126. https://doi.org/10.1107/S2052520619014380
• 7)   R. R.  Dayal and  F. P.  Glasser: Sci. Ceram., 3 (1967), 191.
• 8)   D. H.  Lister and  F. P.  Glasser: Trans. Br. Ceram. Soc., 66 (1967), 293.
• 9)   K.  Sugiyama,  A.  Monkawa and  T.  Sugiyama: ISIJ Int., 45 (2005), 560. https://doi.org/10.2355/isijinternational.45.560
• 10)   E.  Schürmann and  U.  Janhsen: Steel Res., 64 (1993), 331. https://doi.org/10.1002/srin.199301031
• 11)   M.  Kimura,  Y.  Uemura,  T.  Takayama,  R.  Murao,  K.  Asakura and  M.  Nomura: J. Phys. Conf. Ser., 430 (2013), 012074. https://doi.org/10.1088/1742-6596/430/1/012074
• 12)   T.  Patrick and  M.  Pownceby: Metall. Mater. Trans. B, 33 (2002), 79. https://doi.org/10.1007/s11663-002-0088-0
• 13)   K.  Sugiyama and  Y.  Ohgi: CAMP-ISIJ, 20 (2007), 17 (in Japanese).
• 14)   T.  Maeda and  Y.  Ono: Tetsu-to-Hagané, 75 (1989), 416 (in Japanese). https://doi.org/10.2355/tetsutohagane1955.75.3_416
• 15)   D.  Maruoka,  S.  Mataoka,  T.  Murakami and  E.  Kasai: 8th Int. Cong. on Science and Technology of Ironmaking (ICSTI 2018), (Vienna), ASMET, Leoben, (2018), 71. http://bestevent.management/event/28/overview, (accessed 2020-07-09).
• 16)   N. A. S.  Webster,  M. I.  Pownceby,  I. C.  Madsen and  J. A.  Kimpton: ISIJ Int., 53 (2013), 774. https://doi.org/10.2355/isijinternational.53.774
• 17)   N. A. S.  Webster,  D. P.  O’dea,  B. G.  Ellis and  M. I.  Pownceby: ISIJ Int., 57 (2017), 41. https://doi.org/10.2355/isijinternational.ISIJINT-2016-332
• 18)   N. A. S.  Webster,  M. I.  Pownceby,  R.  Pattel,  J. R.  Manuel and  J. A.  Kimpton: ISIJ Int., 59 (2019), 1007. https://doi.org/10.2355/isijinternational.ISIJINT-2018-701
• 19)   N. A. S.  Webster,  M. I.  Pownceby,  R.  Pattel,  J. R.  Manuel and  J. A.  Kimpton: ISIJ Int., 59 (2019), 263. https://doi.org/10.2355/isijinternational.ISIJINT-2018-573
• 20)   N.  Sakamoto,  H.  Fukuyo,  Y.  Iwata and  T.  Miyashita: Tetsu-to-Hagané, 70 (1984), 504 (in Japanese). https://doi.org/10.2355/tetsutohagane1955.70.6_504
• 21)   B.  Cai,  T.  Watanabe,  C.  Kamijo,  M.  Susa and  M.  Hayashi: ISIJ Int., 58 (2018), 642. https://doi.org/10.2355/isijinternational.ISIJINT-2017-552
• 22)   T.  Murakami,  D.  Maruoka and  E.  Kasai: CAMP-ISIJ, 32 (2019), 403, CD-ROM (in Japanese).
• 23)   S.  Nicol,  J.  Chen,  M. I.  Pownceby and  N. A. S.  Webster: ISIJ Int., 58 (2018), 2157. https://doi.org/10.2355/isijinternational.ISIJINT-2018-203
• 24)   M.  Kimura,  K.  Uemura,  T.  Nagai,  Y.  Niwa,  Y.  Inada and  M.  Nomura: J. Phys. Conf. Ser., 190 (2009), 012163. https://doi.org/10.1088/1742-6596/190/1/012163
• 25)   B.  Ravel and  M.  Newville: J. Synchrotron Radiat., 12 (2005), 537. https://doi.org/10.1107/S0909049505012719