ISIJ International
Online ISSN : 1347-5460
Print ISSN : 0915-1559
ISSN-L : 0915-1559
Regular Article
Gaseous Reduction of Titania-ferrous Solution Ore by H2–Ar Mixture
Zhenyang WangJianliang ZhangJinfang MaKexin Jiao
Author information

2017 Volume 57 Issue 3 Pages 443-452


The reductions of titania-ferrous solution ore (TFSO) ranging from 800°C to 1100°C temperature and 10 to 100 vol% H2–Ar gas mixture were conducted to account for the optimization of reduction parameters, the phase transition behaviors and solid solubility changes with Fe and Ti elements segregation and enrichment, and the micro morphology features in reduction process. 40 vol% of hydrogen in reducing gas, as well as 900°C reaction temperature was the suitable parameters when simultaneously considering the growth of reduction rate and the reasonable economy cost. The reduction of Fe3O4 contained in titanomagnetite (TTM) was proceeding earlier than that of ulvöspinel (Fe2TiO4), even though Fe2TiO4 was also the ingredient of TTM. The initial ilmenite was reduced quickly at primary stage, after that, there were also newly born ilmenite which experienced accumulation and vanishment, along with transformation of Fe2TiO4 and ferropseudobrookite (FeTi2O5). Although TiO2 was appeared as early as 5 min, the low valence titanium (Ti3O5 and Ti2O3) were just observed at the end of reduction, and it was mainly because of the existence of iron oxide, resulting in the weak reduction potential to the titanium oxide. For the non-homogeneous TFSO particles, the lamellas showed a harder reducibility than the substrates and a different priority reduction positions in terms of the central and the edge. For the homogeneous TFSO particles, the reduction was carried out in an outside-in way, which was preferential along some interphase strips that had a lower gangue elements distribution.

1. Introduction

Titania-ferrous solution ore (TFSO), generally called as ironsand for short, is gradually developed from the rapid cooling of volcanic lava and widely distributed in the coastal areas, for example New Zealand, Indonesia, South Africa and so on.1) Due to the shortage of high-quality hematite resource, TFSO have been widely investigated as an alternative source of conventional iron ore.2,3,4,5,6,7) The advantages of TFSO in large deposit, low mining cost, available iron grades, among others, caused industrial utilize ways of TFSO becoming focusing subjects such as the aspects in beneficiation,8,9) sintering,10,11,12,13) pelletizing,14,15,16) particularly the gaseous reduction in hydrogen which is beneficial to improve the dynamic condition and lower the adverse impact on environment at the same time.17,18,19,20,21,22,23,24,25,26)

Eungyeul P27) studied the influences of temperature and hydrogen contents on TFSO reduction at the constant condition of 25vol%H2-Ar and 1173 K respectively. In addition, the reduction path of TFSO by H2–Ar gas mixture at temperatures above 1173 K was proposed as Fe3-xTixO4 → FeO + Fe3-x-δTix+δO4 → Fe + Fe3-x-δTix+δO4 → Fe + xTiO2. However, in the research conducted by Jie D28) regarding the kinetics of TFSO reduced by H2, the reaction was divided into dual processes which occurred simultaneously, namely the transitions from titanomagnetite to wustite and ilmenite and from wustite and ilmenite to titanium-containing phase. Therefore, one purpose of this study is to define the phase transition further clearly based on the reaction time and solid-solubility. Besides, this paper extended the mechanism and experimental conditions of temperature and hydrogen partial pressure so as to choose the suitable reduction parameters from a large range considering both the growth of reduction rate and the reasonable economy cost. At last, the microcosmic morphology of TFSO particles during H2 reduction was investigated and three patterns with different reduction features were explored.

2. Materials and Methods

2.1. Raw Materials

The chemical composition and particle size distribution of TFSO used in this work were examined by neutralization titration and laser particle size analyzer respectively, as shown in Table 1. The total contents of iron and titanium oxides were about 87.59 wt%, which achieved the available standard. All TFSO particles were smaller than 420 μm while greater than 90 μm, with the percentage of dominant size scope between 110 μm and 300 μm reaching to 89.34 wt%. Two types of Fe–Ti solid solution occupied the main phase of TFSO. One was the titanomagnetite (TTM, Fe3-xTixO4) with a cubic spinel crystal structure. TTM was constituted by magnetite (Fe3O4) and ulvöspinel (Fe2TiO4) in a certain range proportion and formed the homogeneous particles and the substrate of non-homogeneous particles. The other was the titanohematite (TTH, Fe2-yTiyO3) with a rhombohedral crystal geometry, which was composed by hematite (Fe2O3) and ilmenite (FeTiO3) in a certain percentage. As the form of intersecting lamellas existing, TTH was usually dissolved from TTM substrates because of the atmosphere oxidation and its low solubility in TTM.29) Through testing the atomic percentages of three different micro-morphologies of TFSO particles by Electron Probe Micro Analyzer (EPMA), x and y could be approximately obtained, as expressed in Table 2. TTM in homogeneous particles and non-homogeneous particle substrates showed disparate solubilities between magnetite and ulvöspinel. It was mainly because of the oxidation by the atmosphere or by the dissolved oxygen in water flow for a long time after mineralization, resulting in the exsolution and further development of TTH. In other words, the titanium and oxygen were enriched into TTH lamellas from TTM substrates. As a consequence, the contents of Ti and O in the non-homogeneous particle lamellas were obviously more than other regions. And similarly because of the phase transition and the enrichment of elements discussed above, TTM in non-homogeneous particle substrates displayed a lower ulvöspinel contents and x value than TTM in homogeneous particle did, as illustrated in Table 2.

Table 1. Chemical composition and particle size distribution of TFSO.
Size/×103 μm0.09–0.110.11–0.210.21–0.300.30–0.42
Table 2. Atomic percentages of three different phase in the raw TFSO particles.
Phase CompositionFe/at%Ti/at%O/at%x or y
homogeneous particle43.024.0744.10x≈0.26
non-homogeneous particle substrate49.942.0345.10x≈0.12
non-homogeneous particle lamella31.9214.3650.73y≈0.62

2.2. Experimental Methods

The reduction of TFSO was conducted by hydrogen-argon mixture using a fixed bed reactor in a vertical tube electric resistance furnace. The schematic setup is shown in Fig. 1. After pre-dehydration through drierite and deep-dehydration through 3 A molecular, the hydrogen and argon were entered into the controller cabinet which controlled the volume ratio of H2/(H2+Ar) as 10%, 20%, 30%, 40%, 50%, 100% and the total flow rate to 5 L·min−1 by volumetric flowmeter. Considering the 45 mm inner diameter of the reaction tube, the linear velocity of the reaction gas can get to 0.052 m·s−1. The initial mass of TFSO in each experiment was 3 g and all of that was evenly distributed on a 40 mm diameter alumina plate. The plate was big enough so that there was just one layer of TFSO spreading on the bottom, which can ensure that every TFSO particle had a good contact with reduction mixture gas. The constant temperature zone of the furnace was controlled at every 50°C from 800°C to 1100°C. The sample was kept in the constant temp zone and the argon atmosphere for 30 min in order to reach the target temperature. After that the fixed ratio reduction mixture gas was blown into the furnace from the gas inlet tube in furnace bottom, and then the reduction began. The sample and alumina plate were hanging on the balance through a molybdenum wire, so that the weight of sample can be measured continuously and recorded by the connected computer for every 6 seconds. Thus, the reduction degree of TFSO at different time, temperature and atmosphere can be obtained. For investigating the phase transitions and microcosmic features during the reduction process, the samples were taken out from the furnace when achieving the scheduled times. Then they were immediately put into the cooling container and quenched in the shielding gas. The cooled samples were crushed for x-ray diffraction (XRD) measure on a Rigaku Ultima IV diffractometer operated at 40 kV and 250 mA with a scan speed of 0.03 degrees 2θ/second using Cu Kα radiation. Powder silicon was also detected at the same conditions and the acquired results were used for calibration. The phase compositions of TFSO powders in twelve reduced stages were determined by the JADE 6.5 software provided by Materials Data Incorporated. The characterization of reduced TFSO was also carried out using the scanning electron microscope (SEM) (FEI Quanta 250) equipped with energy dispersive spectrometer (EDS) (EDAX), which extended the micro morphology and elements distribution of samples.

Fig. 1.

Schematic setup for hydrogen-argon mixture reduction of TFSO. (Online version in color.)

3. Results and Discussion

3.1. Function and Mechanism of Temperature and Hydrogen Contents on TFSO Gaseous Reduction

The reduced levels of TFSO at a certain reaction time t are represented by reduction degree Rt, as expressed in Eq. (1), where ΔOt means the oxygen mass loss from iron-oxide at the reaction time t. Due to the negligible burn drop rate, ΔOt can be obtained by the mass loss of TFSO. And ΔO0 is the weight of oxygen contained in the raw iron oxide, which can be calculated based on the initial TFSO compositions.   

R t = Δ O t Δ O 0 (1)

The curve of reduction degree with reaction time in the temperature ranging from 800°C to 1100°C at different hydrogen contents (10%, 20%, 30%, 40%, 50%, 100%) are presented as Fig. 2. In the experimental temperature range, the TFSO were just partial reduced and difficult to approach to 100% reduction degree at the conditions of 10 vol% and 20 vol% H2/(H2+Ar). However, as the hydrogen partial pressure was improved to 0.3 bar (As the reduction was conducted in the ordinary pressure, it is approximately considered that gas partial pressure is in direct proportion to its volume fraction) and even higher, the full reduction status can be achieved at some certain temperatures. Under every fixed reducing atmosphere, the reduction rate and degree at the same reaction time increased with the enhancement of temperature. But the increase scope of reduction degree was not identical even though increasing the same temperature. For investigating the advantage scope and the stimulative principle of variational temperature to the reductive reaction, the reduction degrees at the same moment from 5 min to 50 min are described in Fig. 3.

Fig. 2.

Effect of temperature (every 50°C from 800°C to 1100°C) on the reduction of TFSO at different hydrogen contents. (H2/(H2+Ar): a=10 vol%; b=20 vol%; c=30 vol%; d=40 vol%; e=50 vol%; f=100 vol%). (Online version in color.)

Fig. 3.

Based on the same reaction moments from 5 min to 50 min, the variation trends of reduction degree with the reaction temperature were extracted to explore the advantage scope of variational temperature to the reductive reaction. (a=5 min; b=10 min; c=20 min; d=30 min; e=40 min; f=50 min). (Online version in color.)

In order to express the influence of temperature to the reduction quantitatively, the changes of reduction degree at the condition of improving the same temperature were introduced, as shown in Eq. (2).   

Δ R T, V H 2 = R T, V H 2 - R T-50°C, V H 2 (2)
where R represents reduction degree. T and VH2 mean the reaction temperature and hydrogen volume fraction respectively. Thus ΔR can be used to indicate the promotion effects of every 50°C improvement to the reduction degree within the experimental scopes.

In the incipient reduction stage as shown in Fig. 3(a), ΔR were almost uniform from 800°C to 1100°C for every fixed hydrogen concentration. However, for different reducing-gas proportions, this promotion (ΔR) was a bit higher along with the increase of hydrogen contents, which was more obvious when proceeding to 10 min and 20 min as indicated in Figs. 3(b) and 3(c). At these two moments, the turning points, which were identified by the increase rate of reduction degree along with the temperature, were appeared at the condition of 900°C and pure hydrogen atmosphere. It illustrated that the stimulative impacts of temp to the reduction slowed down when exceeding 900°C in the pure hydrogen. At 30 min, the turning points were also present to the curves of 40%, 50% as well as 100% hydrogen contents when the temperature reached to 900°C. Following the reaction conducting to 40 min and 50 min, the turning points turned up on the curve of 30 vol% hydrogen in 900°C and 20 vol% hydrogen in 850°C respectively. It was not hard to find that with the H2 ratio varied from 100% to 20%, the time when the turning points appeared became later from 10 min to 50 min in the reaction process. However, for 10% hydrogen reduction condition, ΔR was almost the same during the variation of the temperature and the proceeding of the reaction time. Thus, turning point did not appear in 10 vol% hydrogen and the promotion effects of every 50°C increase to the reduction were approximately identical. But overall, 900°C should be a suitable temperature when simultaneously considering the growth of reduction rate and the heating energy consumption.

For exploring the effect of hydrogen contents on the TFSO reduction at distinct temperatures, the reduction degree curves within the experimental ranges are presented in Fig. 4. At 800°C and 850°C, it was difficult to achieve 90% reduction degree in 60 min among 10 to 50 vol% H2. However, the 90% reduction degrees were received at 900°C and 950°C in 50 vol% H2 and it could be obtained more easily at 1000°C or higher, for example, 20 vol% H2 atmosphere in 1100°C could ensure a reduction degree above 90% in 1 hour. In order to reflect the effects of H2 contents more observably, the increase trends of reduction degree with the hydrogen concentration enhancement at constant temperature for different reaction moments were presented in Fig. 5. At 5 min, the impacts of H2 concentration from 10 vol% to 50 vol% to the reduction degree were uniform and stable at the experimental temperature range. However when the H2 concentration grew to 40% from 30% at 10 min, the increase range of reduction degree emerged a clear uptrend in the 1050°C and 1100°C. This phenomenon also happened at 20 min and 30 min in the range of 900°C to 1100°C. When the reaction extended to 40 min and 50 min, the improvement of hydrogen contents to 50 vol% from 40 vol% can promote the rise of reduction rate, but less obviously than the consequence of increasing to 40 vol% from 30 vol% or even lower. For example, at the moment of 40 min in 900°C, the reduction degree increased nearly 12.5% from 30 to 40 vol% H2, but it enhanced just 6.8% from 40 to 50 vol% H2. Therefore, 40% content of hydrogen in reducing gas should be a suitable proportion considering the reduction rate and reductant economy simultaneously.

Fig. 4.

Effect of hydrogen volume fractions (10 vol%; 20 vol%; 30 vol%; 40 vol%; 50 vol%) on the reduction of TFSO at the condition of different temperature. (a, b = 800°C; c = 850°C; d = 900°C; e = 950°C; f = 1000°C; g = 1050°C; h = 1100°C). (Online version in color.)

Fig. 5.

Based on the same reaction moments from 5 min to 50 min, the variation trends of reduction degree with the hydrogen contents were extracted to explore the acceleration scope of variational H2 concentration to the reductive reaction. (a=5 min; b=10 min; c=20 min; d=30 min; e=40 min; f=50 min). (Online version in color.)

With the reduction degree rising, the reaction rate showed a downward trend, especially in the middle-late stages (after 30 min), as shown in Figs. 2 and 4. It was mainly because the improving reduction degree caused thicker product layers, those, in this work, were mostly Fe and titanic oxide. The layers hindered the internal diffusion of the reducing gas and produced gas greatly. As a result, the rate-controlling steps changed and the reaction rate was becoming sluggish. The curve slope of reduction degree was certain to change at a time in the whole reaction process. Via the linear fitting and intersection point confirmation, the turning points (TPs) which resulted from the rate-controlling steps variation were confirmed in every reduction degree curve, as the example in Fig. 4(b).

Generally, the TP can be described and confirmed by the reduction degree RTP at the TP appeared moment tTP in reaction process. Besides, RTP and tTP were totally influenced by the reaction temperature and reducing-gas concentration based on the model of unreacted core. As expressed in Fig. 6, the thickness of chemical reaction controlled zone is represented by r, which actually determined the reduction degree where the turning point appeared (RTP). K means the reaction rate or the production layer formation rate and affect the time of turning point appeared (tTP). Theoretically, when the hydrogen concentration increased and the reaction temperature remain unchanged, r will not change because of the unvaried ability of molecule movement. However, the rise of reducing-gas partial pressure could promote the increase of K. Thus, it will bring an invariable RTP and a shortened tTP, as shown in Fig. 6(a).

Fig. 6.

Influence of reaction temperature and reducing-gas concentration to the direct reduction.

The constant hydrogen concentration and increased reaction temperature would result in a more intense molecular thermal movement, and of course the raised chemical reaction controlled zone and reaction speed. Consequently, both r and K have increased, which also lead to an increment of RTP. But tTP cannot be judged qualitatively due to the nondeterminacy of increasing amplitude of r and K, as is present in Fig. 6(b). Furthermore, the relationship between TP parameters (RTP, tTP) and reaction conditions can be explored as Fig. 4(b) pattern and more results are shown in Table 3.

Table 3. The reduction degree RTP and the time tTP where and when the turning points appeared at different conditions in terms of the temperature ranging from 800°C to 1100°C and the V(H2)/[V(H2) + V(Ar)] ranging from 10% to 100%.
V(H2)/[V(H2) + V(Ar)]/%1020304050100RTP/%

As previously mentioned, tTP was depending on the reaction rate and the chemical reaction controlled zone collectively, which means that a higher reaction temperature and a lower RTP correspond to a shorter tTP and vice versa. As shown in Table 3, with the increase of hydrogen concentration, and namely, the reaction rate K, tTP strictly showed up a rule of decline at the same temperature, which was obviously due to the identical RTP and the higher reaction speed K. However, once the reaction temperatures have risen at the condition of the same hydrogen contents, the increasing temperature improved the reaction speed, and at the same time enhanced the RTP. When ranging from 800°C to 900°C, the effect of temperature on RTP was stronger than that on the reaction rate, thus tTP grew up gradually from 800°C to 900°C and reached a maximum value at 900°C. As the temperature continued to increase from 900°C to 1100°C, although the chemical reaction controlled zone became wider which was proved by the higher RTP, the advancement of reaction rate held the dominant role and the tTP trended to decline in this temperature range.

3.2. Phase Transition during the Process of TFSO Hydrogen Reduction

XRD analysis graphics of raw TFSO and the TFSO which were reduced for different times from 2.5 min to 60 min at 900°C are shown in Fig. 7. For investigating the phase change without other factors interference, 100 vol% H2 was used as the reducing atmosphere. There were eight main phases appearing in TFSO reduction process. Through the XRD identification and reduction degree data, the phases at different reduction times from raw ore to 60 min are presented in Table 4.

Fig. 7.

XRD patterns of TFSO, which were isothermally reduced for different time from raw ore to 60 min in the 100 vol% H2 atmosphere at 900°C. (Online version in color.)

Table 4. Phases and reduction degree of TFSO reduced for different time from raw ore to 60 min by 100 vol% H2 atmosphere at 900°C.
Reduction Time/minPhase identified by XRDReduction Degree/%
0TTM; TTH; FeTiO3; Pyroxene0
2.5TTM; Fe; FeO; FeTiO3; Pyroxene24.98
5TTM; Fe; FeO; TiO2; Pyroxene51.71
10Fe2TiO4; Fe; FeO; FeTiO3; TiO2; Pyroxene82.18
15Fe2TiO4; Fe; FeTiO3; TiO2; Pyroxene94.12
20Fe2TiO4; Fe; FeTiO3; TiO2; Pyroxene96.97
25Fe2TiO4; Fe; FeTiO3; FeTi2O5; TiO2; Pyroxene97.58
30Fe2TiO4; Fe; FeTiO3; FeTi2O5; TiO2; Pyroxene98.20
40Fe2TiO4; Fe; FeTiO3; FeTi2O5; TiO2; Ti3O5; Pyroxene99.19
50Fe; TiO2; Ti3O5; Ti2O3; Pyroxene100
60Fe; TiO2; Ti3O5; Ti2O3; Pyroxene100

Obviously, TTM was the main phase in the raw TFSO. Besides, TTH existed among raw TFSO with a small quantity, which was formed from TTM owning to the intracrystalline solubility and atmospheric oxidation. A small peak corresponding to FeTiO3 can be observed around 32.72° in raw TFSO. In addition, oxides of silicon, magnesium, aluminum, and manganum with a catenulate pyroxene structure presented in TFSO throughout all the reduction process. When the sample was placed in the reducing gas for 2.5 min, the reduction degree reached to 24.98%. TTH was obliterated owning to its small content and rapid reducing of Fe2O3 (one ingredient in TTH) into Fe3O4. And Fe3O4, including the Fe3O4 in TTM and the Fe3O4 reduced from TTH, was partly reduced into wustite (FeO) and metallic iron (Fe). The raw FeTiO3 were fully reduced to Fe and TiO2 in 5 min and at this time the reduction degree exceeded 50%. However, the newly born FeTiO3 has appeared between 5 min and 10 min due to the start of Fe2TiO4 reduction in TTM and the rest ingredient (FeTiO3) in TTH, along with the production of another part Fe reduced from Fe2TiO4. The FeO peak was disappeared at 15 min, meaning that all of Fe3O4 contained in TTM, including the intermediate product FeO, were transformed into Fe before 15 min. At this moment, the 94.12% reduction degree illustrated that most of the iron oxide has been reduced before 15 min. From 15 min to 20 min, there were no changes about the phase species, except that the reduction degree experienced a stable rise, ranging to 96.97%. As an intermediate product from FeTiO3 to Fe and TiO2, the ferropseudobrookite (FeTi2O5) has not appeared until the reduction time came to 25 min, which represents that after accumulated from 5 min to 20 min, the newly born FeTiO3 from TTM and TTH starts to be reduced. The phase species did not change hovering from 25 min to 30 min. However, based on the disappearing of TTM peaks in 53.4°, 56.6° and 62.6°, the TTM content declined further. Titanium began to emerge low valence between 40 min and 50 min, for example, Ti3O5 could be found at 40 min and Ti2O3 was observed at 50 min. The reason why the low valence titanium state did not closely follow on the TiO2 generation at 5 min may be that the existence of iron oxide has weakened the reduction potential to the titanium oxide. Therefore, when the reduction degree achieved 99.19% at 40 min, it promoted the titanium reduction and the low valence state of titanium finally appeared. The reduction degree was closed to 100% at 40 min, which was according with the weak peaks of Fe2TiO4, FeTiO3 and FeTi2O5 still existing in XRD patterns. And after the reduction proceeding to 50 min, the FeTiO3 and FeTi2O5 were vanished and transformed into TiO2 and Fe completely. Moreover, the last TTM diffraction main peak around 35.3° was totally faded away, suggesting that the Fe2TiO4 in TTM was thoroughly reduced. At last, the metallic state of iron, three valences of titanium oxide and pyroxene were made up of the final TFSO samples after 50 min. The phase variation process based on the above analysis is shown in Fig. 8.   

TTM(Fe 3-x Ti x O 4 ){ (1-x) Fe 3 O 4 :(1-x) Fe 3 O 4 +(4-4x) H 2 =(3-3x)Fe+(4-4x) H 2 O x Fe 2 TiO 4 : Fe 2x Ti x O 4x +δ H 2 = Fe 2x-δ Ti x O 4x-δ +δFe+δ H 2 O    (0<δ<2x) [1]
TTH(Fe 2-y Ti y O 3 )   { (1-y) Fe 2 O 3 :(1-y) Fe 2 O 3 +(3-3y) H 2 =(2-2y)Fe+(3-3y) H 2 O y FeTiO 3 : Fe y Ti y O 3y +ε H 2 = Fe y-ε Ti y O 3y-ε +εFe+ε H 2 O (0<ε<y) [2]
Fig. 8.

Phase transitions during the TFSO reduction process from raw ore to 60 min in 100 vol% H2 atmosphere at 900°C.

The phase transformation and solution solubility changes of oxide in TTM and TTH during the TFSO reduction process are reflected in FeO–Fe2O3–TiO2 ternary components system, as shown in Fig. 9. Formulas [1] and [2] are indicating the reduction equations of ingredients in TTM and TTH, where x and y mean the solid solution ratio of Fe3O4 – Fe2TiO4 and Fe2O3 – FeTiO3; δ and ε represent the reduction degree of Fe2TiO4 and FeTiO3. According to the typical atomic ratio displayed in Table 2, the initial composition of TTM and TTH in raw TFSO could be confirmed as x = 0.26 or 0.12 for different particles and y = 0.62, which were labeled on the Fe3O4–Fe2TiO4 and Fe2O3–FeTiO3 solid solution lines by square and triangle samples respectively. Along with the hydrogen reduction, Fe3O4 and Fe2O3 dissolved in TTM and TTH decreased firstly, resulting in the composition points moving toward the FeO–TiO2 solid solution line. At the same time, x and y in formulas [1] and [2] gradually increased and finally reached to 1, which means Fe3O4 and Fe2O3 were totally transformed. The process, in terms of Fe2TiO4 in TTM reducing to newly born FeTiO3, was accompanied with δ tending towards x. Furthermore, with δ and ε approaching to 1.5x and 0.5y correspondingly, the FeTiO3 from two sources were reduced to FeTi2O5. And finally, Fe was completely reduced and separated with titanium oxide when the δ and ε were getting to 2x and y.

Fig. 9.

FeO–Fe2O3–TiO2 ternary components system showing phase transformation and solution solubility changes from the initial TTM and TTH in raw TFSO to the final reduction stage.

3.3. Microcosmic Features and Morphology Changes during the Process of TFSO Hydrogen Reduction

SEM and EDS analysis were carried out to reflect the phase morphology in terms of the lamellar zone and homogeneous regions, and to demonstrate the componential variance in different microstructures. The results displayed that there were three kinds of TFSO particles with different reduction mechanism even at the same reducing condition, as shown in Figs. 10, 11 and 12.

Fig. 10.

The microscopic composition change of one kind of TFSO particle (non-homogeneous particle with lamella TTH phase exsolved from substrate) during reduction process in 100 vol% H2 atmosphere at 900°C for different time. These kinds of lamellas were firstly reduced from the center position forming the pore and dot reduced Fe.

Fig. 11.

The microscopic composition change of another kind of TFSO particle (non-homogeneous particle with lamella TTH phase exsolved from substrate) during reduction process in 100 vol% H2 atmosphere at 900°C for different time. These kinds of lamellas were firstly reduced from the edge position forming the stretch Fe.

Fig. 12.

The microscopic composition change of another kind TFSO particle (homogeneous TTM with a smooth particle surface) during reduction process in 100 vol% H2 atmosphere at 900°C for 5 minutes. (Online version in color.)

Figsure 10(a) to 10(f) present morphology change of lamellas in TFSO particles reduced at 900°C for different time. When the substrates (TTM) were forming the iron whiskers and someplace even the connected metallic iron owning to the sintering effect, as shown in Fig. 10(a), only dot distribution metallic irons were observed on the lamella structure (TTH), showing its relative hard-reducibility. In addition, there were approximately circular and needle-like pores generated, which were mostly dispersed in the intermediate section of the lamella structure, the same sites as the dot metallic iron located, as examined in Fig. 10(b). The dot iron and pores were further connected into coarsened iron and seams respectively in the process of the second and third reduction stages (Figs. 10(c), 10(d)), which also illustrated that the lamella structures were preferentially reduced from the central and then developed into the lamella edges. Furthermore, after 40 min, although the non-homogeneous particles were adequately reduced, the initial phase boundaries between lamella structures and substrates were still visible in the final stage, as seen in Fig. 10(e). However when reduction were proceeding to 20 min, two forms of reduction processes in terms of lamella structures were obtained in Fig. 10(f). One was the process which has the priority reduction order from the central, as discussed above, the other one was reversed with the reduction starting from the edges, as explored from Figs. 11(a) to 11(f).

Figures 11(a) and 11(b) still suggested the properties of sluggishness reduction in terms of the lamellas, compared with the homogeneous substrates. The lamellas kept unreduced even though the connected metallic iron had been significantly observed in the nearby substrates before 10 min. In the second stage between 10 min to 20 min shown in Figs. 11(c) and 11(d), reduced Fe was dotted amongst the lamella center, while at the same time in the lamella edges, stretched Fe were generated, illustrating that the reduction of the edge was prior to the center in these lamellas, which was different from those in Fig. 10, where the lamella center sections were easier reduced. In the final stage after 40 min, the outside-in reduction order represented in Fig. 11(e) resulted in the fine string pyroxene and high-content titanium TFSO which were refractory reduced, while the metallic iron in the lamellas edges was significantly coarsened and sintered, as shown in Fig. 11(f).

As indicated in Fig. 12, the microscopic features of the homogeneous TTM with a smooth particle surface, which was another kind of TFSO particles, were investigated by SEM and EDS when the reduction process in 100 vol% H2 atmosphere at 900°C was proceeding for 5 min. Four different regions in the homogeneous TTM particle labeled as A, B, C, and D in Fig. 10(b) were directly detected by EDS. As more than 99 wt pct of the TFSO ingredients were occupied by the elements Fe–Ti–Mg–Al–Si–O, the atomic percent of six main elements were dealt with normalized calculation to 100% to ensure the consistent comparability among the four regions, as given in Table 5. After subtracting the oxygen which the gangue elements needed for the most stable valence, the other oxygen was used to approximately estimate the ferrous contents. Then the surplus iron could be considered as metallic state. And the metallization ratios were gained, as shown in Table 5.

Table 5. Atomic percentages of four different regions in the homogeneous TTM reflected in Fig. 12 reduced for 5 min by 100 vol% H2 at 900°C.
RegionsRelative atomic percent/at%Metallization ratio/%
FeTiOSiMgAlMetal Fe/Total Fe

The reduction of homogeneous TTM particles was carried out in an outside-in way with the obvious boundary lines distinguishing the reduced and unreduced areas. In the reduced areas, the bright regions (represented by region A) were mostly metallized with the metallization ratio approximating to 100%. However, there still had some relative insufficiency reduction regions even though the regions had already been located in the reduced areas, typically represented by region B with the metallization ratio reaching to 98.80%. Meanwhile, the contents of magnesium and silicon, particularly aluminum and titanium in region B were larger than those in region A.

Two different regions were also observed in the unreduced areas. The bright gray regions (represented by region C) were generated in strips and alternately distributed on the dark gray substrates (represented by region D). The distinctions between regions C and D were similar to those between regions A and B, showing the disparities in gangue elements contents.

Moreover, in the boundary districts, the bright gray regions (C) in the unreduced areas were strictly connected with the metallized regions (A) in the reduced areas, indicating that the proceeding of reduction and the formation of region A were selective and preferential along region C. It was mainly because of the nonuniform distributions of the gangue elements, such as magnesium, silicon, aluminum and titanium, formed during the TFSO original generation stage which had a rapid cooling speed through volcanic rock, resulting in occasional and inadequate diffusion. Therefore, the places which have more contents of the gangue elements in the forms of magnesia-alumina spinel, hercynite and ulvöspinel were easier to be reduced stragglingly and vice versa. Thus, the segregation reduction and alternately distributed strips in terms of regions A, B, C, D were generated during the reduction stage.

4. Conclusions

It was difficult to exceed 90% reduction degree in 60 min when setting the reaction temperature as 800°C and 850°C or the hydrogen partial pressure as 10% and 20%. Except that, higher experimental conditions would achieve over 90% reduction degree within 1 hour. Under most hydrogen partial pressures (30%, 40%, 50% and 100%), 900°C were the temperature points distinguishing the increase rate of reduction degree, and after that the increase of reduction degree with temperature became tardy. Moreover, 40% content of hydrogen in reducing gas was also a suitable parameter as 900°C when simultaneously considering the growth of reduction rate and the reasonable economy factor.

The increased reaction temperature would result in a more intense molecular thermal movement namely a raised reaction speed and a reduced internal diffusion controlled zone, while the increased hydrogen concentration would just be effective to the reaction rate enhancing but not the chemical reaction controlled zone. Therefore, the location (reduction degree) where the turning points occurred would be determined by the temperature only but the time when the turning points appeared would be depended on both the temperature and reducing-gas concentration.

TTH disappeared fast at the beginning due to the rapid transform of Fe2O3 (in TTH) into Fe3O4. The raw FeTiO3 could be fully reduced to Fe and TiO2 in the early stage. After that, the newly born FeTiO3 was appeared owning to the start reduction of Fe2TiO4 (in TTM) and the rest ingredient (FeTiO3) in TTH. The regenerated FeTiO3 experienced accumulation until FeTi2O5 emerged. TiO2 could be viewed early as the result of raw FeTiO3 reduction. However, it was not reduced into low valence until the later period. The reason why the Ti3O5 and Ti2O3 did not closely follow on the TiO2 generation at initial stage may be the existence of iron oxide weakened the reduction potential to the titanic oxide.

Three kinds of TFSO particles with different reduction microcosmic features were observed. For the non-homogeneous particles, considering that only dot irons were distributed on the lamellas when the connected iron occurred in the substrates, the lamellas showed a harder reducibility than the substrates. On the other hand, the reduction processes of lamella were different in the priority positions, namely starting from the central or from the edges. The reduction of homogeneous particles was carried out in an outside-in way with the obvious boundary lines distinguishing the reduced and unreduced areas. Furthermore, the proceeding of reduction was selective and preferential along some strips mainly because of disparate distribution of the gangue elements and the inhibition effects to the reduction.


This work was financially supported by the National Basic Research Program of China (No. 2012CB720401) and the Natural Science Foundation of China and Baosteel under Grant (No. 51134008).

© 2017 by The Iron and Steel Institute of Japan