ISIJ International
Online ISSN : 1347-5460
Print ISSN : 0915-1559
ISSN-L : 0915-1559
Regular Article
Effect of Silica on Reduction of Calcium Ferrite with CO–N2 Gas Mixtures
Chengyi DingXuewei LvSenwei XuanJie QiuYun ChenChenguang Bai
Author information
JOURNALS OPEN ACCESS FULL-TEXT HTML

2017 Volume 57 Issue 4 Pages 634-642

Details
Abstract

Silica is a significant chemical component in the formation of SiO2–Fe2O3–CaO–Al2O3 from CaO·Fe2O3 (CF). The influence of silica on the reducibility of the CaO–Fe2O3–SiO2 system was fully examined in this study. The isothermal reduction kinetics of CF, CF2S, CF4S, and CF8S were investigated through thermogravimetric analysis at 1123, 1173, and 1223 K with 30% CO and 70% N2 gas mixtures. CF2S and CF4S presented a slight degrader in reduction degree but activated the reaction faster than CF reduction. The reduction of the samples with 8% silica was not only highly accelerated but also proceeded easily. Rate analysis revealed that CF and CF8S reduction occurs in two stages, whereas CF2S and CF4S reduction occurs in three stages, the Fe3O4–to–FeO stage overlaps with the previous Fe2O3–to–Fe3O4 stage and tends to approach the following FeO–to–Fe stage with an increase in silica content. The apparent activation energy values of CF, CF2S, CF4S and CF8S reduction are 46.89, 24.48, 44.84, and 8.71 kJ·mol−1. CO diffusion is the rate–controlling step of the entire process of CF8S reduction, whereas the rate–controlling step of CF, CF2S and CF4S reduction are performed as inner gas diffusion, inner gas diffusion and interface chemical reaction mixed control then interface chemical reaction in turns with reduction degree increasing. Sharp analysis indicated that CF and CF8S reduction was expressed by the Avrami–Erofeev equation presenting a 2D shrinking layer reaction, whereas CF2S and CF4S reduction was expressed initially by a 2D shrinking layer reaction and then by a 2D phase boundary–controlled reaction.

1. Introduction

The use of low–purity iron ore is significantly promoted as the cost of iron ore gradually increases. Sintering prior to a blast furnace produces superior raw materials by enhancing strength and metallurgical performance. The SiO2–Fe2O3–CaO–Al2O3 (SFCA) phase is regarded as a desirable bonding phase for fluxed sinter, and silica content influences the phase composition and bonding phase mass of SFCA. Edstrom1) in 1986 showed that low–SiO2 (<5 wt%) sinter exhibits good metallurgical performance and reduces blast furnace slag in the iron–making process. Silicate (CaO–SiO2) and calcium ferrite (CaO–Fe2O3) systems are the two main bonding phases formed in high–basicity sintering. As the liquidus temperature increases, the silicate system generates less mass than the calcium ferrite system but plays a significant role in the formation of SFCA nevertheless. SiO2–Fe2O3–CaO (SFC) is believed to be the transition phase in the SFCA formation process and has been widely investigated.2,3) Hyunsik4,5) explored the effect of silica on the reduction kinetics of carbon composite pellets and found that silica decreases the reduction rate by lowering the surface area of samples. Taguchi et al.6) analyzed phase transition in the CaO–2Fe2O3 system with silica addition. However, the reduction mechanism especially described by thermal kinetics with the silica content changing in the CaO–Fe2O3–SiO2 system has rarely been investigated. Reduction kinetics measured through thermogravimetric analysis (TGA) in terms of iron oxides has been developed but is almost not involved in the description of the multistep of the reduction mechanism. In this study, the reducibility of the CaO–Fe2O3–SiO2 system with series of silica addition was compared. The reduction mechanism was also fully discussed.

In sintering production, the effect of silica on sinter quality and on the determination of blast furnace slag has been examined by various studies. Energy consumption in a blast furnace is determined by the reducibility of raw materials. The phases formed in the solid state of the CaO–Fe2O3–SiO2 ternary system during the sintering process have been explored, and the common conclusions obtained are that CaO·Fe2O3 (CF) and 2CaO·SiO2 (C2S) are generated from the solid reactions of Fe2O3 with CaO and SiO2 with CaO. Fe2O3 can be decomposed or reduced into Fe3O4, FeO, and Fe in the sintering process, and FeO, CaO, and SiO2 can be bonded into iron olivine and further solidified in the sinter phases accompanied by CF and C2S, as shown in Fig. 1. However, the phase composition and reducibility in the solid reaction with the CaO–Fe2O3–SiO2 ternary system have not been fully discussed, and the reduction of samples in previous studies was chemically regarded as the solid–gas reduction of iron oxide.7,8,9,10) The reduction route of iron oxide has been explored by many studies, but explanations with the model function derived by thermal kinetics are lacking. The current study attempts to improve the understanding of these aspects.

Fig. 1.

Reduction route of the solid reaction in the sintering process with silica addition.

Silica and alumina are two necessary composition in the formation process from CF to SFCA, ternary system of CaOFe2O3x system (x: SiO2, Al2O3) obviously needed further investigation, the reduction kinetics study focus on the reduction rate, activation energy and model function of reduction process. Other paper discussing on the system of CaO–Fe2O3–Al2O3 will be also investigated to rich the reduction research of ternary calcium ferrite system.

2. Experimental

2.1. Materials

Samples were prepared from CaCO3 (≥99.99%) and Fe2O3 (α–Fe2O3, ≥99.99%) at a molar ratio of 1:1 with the addition of 0, 2, 4, and 8 wt.% silica, as shown in Table 1. The powdery raw materials were uniformly mixed and then pressed into cylinder–shaped samples with 10 MPa pressure. The samples were roasted in a Si–Mo furnace at 1173 K for 1 h to decompose CaCO3 to CaO. Then, the temperature was increased to 1473 K for 10 h to allow the complete reaction of CF, CF2S, CF4S, and CF8S [CaO/Fe2O3=1:1, wt.% (SiO2)=0, 2, 4, 8] generation in solid state. Solid state roasting can avoid the reaction of samples with alumina crucible and the generation of wustite in the reacted samples.11,12) The samples were crushed and sieved into powder (<74 μm) for subsequent tests.

Table 1. Chemical composition of samples CF, CF2S, CF4S, and CF8S (wt.%).
SamplesCaCO3Fe2O3SiO2
CF38.4661.540
CF2S37.6960.312
CF4S36.9259.084
CF8S35.3856.618

X–ray diffraction (XRD) analysis with Rigaku D/max2500/PC (Cu Kα) was conducted to confirm the phase composition of the roasted samples. Scanning was carried out at an angular range of 10° to 90° and scan rate of 4°/min. MDI Jade 5.0 was utilized to analyze the intensity data obtained by XRD analysis. Figure 2 shows the XRD patterns of samples CF, CF2S, CF4S, and CF8S.

Fig. 2.

XRD patterns of samples CF, CF2S, CF4S, and CF8S.

The reaction of the CaO–Fe2O3–SiO2 system in the roasting process can be described as follows:   

F e 2 O 3 +CaO+Si O 2 =CaOF e 2 O 3 +2CaOSi O 2 +3( CaOSi O 2 ) F e 2 O 3 +F e 2 O 3 , (1)
where 3(CaO·SiO2)·Fe2O3 is equal to the phase SFC indicated in the XRD pattern shown in Fig. 2. As the silica content added to the CaO–Fe2O3–SiO2 system increases from 2% to 8%, the peak corresponding to CF weakens gradually, whereas that of C2S and Fe2O3 strengthens. This result indicates that the CF content in the roasted phase decreases and SiO2, C2S and Fe2O3 increase. Phase transform was also proven in Taguchi’s study.6) The increase in SFC content is not obvious. The Fe2O3 in samples CF, CF2S, and CF4S has a much lower content than that in the CF8S sample.

CaO–Fe2O3–SiO2 phase diagram shown in Fig. 3, the results indicated that samples with 2% and 4% silica mainly comprise CF and C2S, whereas sample with 8% silica mainly comprises CF, C2S and Fe2O3.

Fig. 3.

CaO–Fe2O3–SiO2 phase diagram of samples CF2S, CF4S and CF8S at 1473 K.

2.2. TGA

Isothermal analysis was carried out via thermogravimetric (TG) measurement in a Setaram analyzer (Model Setsys Evo TG–DTA 1750), as shown in Fig. 4. The samples (20 mg) in an alumina crucible were heated to 1123 K, 1173 K, and 1223 K in N2 (≥99.999%) atmosphere at 15 K/min from room temperature. Gas mixtures of 30% CO (≥99.999%) and 70% N2 were blown into the furnace at 20 ml/min for 150 min to enable complete reaction with the samples at the isothermal stage. To exclude the influence of the system error from the thermal analyzer and the buoyance force from the gas mixtures, a blank test was conducted under the same elimination conditions with no samples in alumina crucibles. Weight loss data were obtained during the isothermal reduction stage, from which the TG data of the blank test were deduced.

Fig. 4.

Schematic of the TG analyzer.

2.3. Thermal Analysis Kinetics

Reduction degree is defined as the ratio of removed oxygen mass at a fixed time t to the theoretically removed oxygen mass from iron oxide; it can be expressed as   

α= Δ m t Δ m 0 , (2)
where α is the reduction degree and ∆mt and ∆m0 refer to the removed oxygen mass at fixed time t and the theoretically removed oxygen mass from iron oxide, respectively. In the reduction of the CaO–Fe2O3–SiO2 system, the removed oxygen is only from phases containing iron oxides. For instance, ∆m0 accounts for 30% and 22.22% of the total mass of Fe2O3 and CF, respectively.

The basic kinetic equation13) that describes the relationship between reduction rate and time can be expressed as   

dα dt =k( T ) f( α ) , (3)
where dα/dt is the reduction rate and k(T) and f(α) are the rate constant and model function of the reduction reaction, respectively. f(α) is influenced by the reaction mechanism. k(T) is determined by the Arrhenius equation as follows:   
k( T ) =Aexp( -E RT ) , (4)
where A is the pre–exponential factor, E is the apparent activation energy, and R is the gas constant [8.314 J/(mol·K)]. Equation (3) can be further expressed as   
dα dt =Aexp( -E RT ) f( α ) . (5)
Given that reduction degree α is a constant, ln f(α) remains unchanged. Therefore, activation energy can be calculated as   
E=-R d( ln dα dt ) d( 1 T ) . (6)
The method to calculate activation energy can eliminate the limit from the model function and is called the model–free method.14) f(α) is usually not easily obtained. G(α), which is the integral function of f(α), can be described as   
G( α ) = 0 α dα f( α ) = 0 t Aexp( -E RT ) dt=k(T)t. (7)

3. Results and Discussion

3.1. Reduction Degree and Reduction Rate

The reduction degree of samples CF, CF2S, CF4S, and CF8S with the change in time during the isothermal reaction stage is shown in Fig. 5. Minimal distinction was observed in the reduction degree among temperatures during the initial stages (before 40 min) for CF, CF2S, and CF4S, but the reduction degree increases with the increase in temperature. The reduction process completes early as temperature increases. The maximum reduction degree and its corresponding time for samples CF, CF2S, CF4S, and CF8S at 1123, 1173, and 1223 K are shown in Table 2, where αm and tm are the maximum reduction degree and its reduction time, respectively. The results indicated that the reduction time for samples CF2S, CF4S, and CF8S at the same temperature increases initially and then decreases. Reduction degree analysis revealed that the increase in silica content from 2% to 4% in the CaO–Fe2O3–SiO2 system prevents sample reduction compared with the pure CaO–Fe2O3 binary system. Meanwhile, adding 8% of silica to the system promotes the reduction.

Fig. 5.

Reduction degree of samples CF, CF2S, CF4S, and CF8S at 1123, 1173, and 1223 K.

Table 2. Maximum reduction degree of samples CF, CF2S, CF4S, and CF8S at 1123, 1173, and 1223 K and the corresponding time.
Samplesαmtm/min
1123 K1173 K1223 K1123 K1173 K1223 K
CF0.920.930.9413710084
CF2S0.880.890.90116115114
CF4S0.900.910.91132122122
CF8S0.940.940.97696862

As shown in Fig. 2, the phases in the roasted samples mainly comprise SiO2, CF, C2S, CFS, and Fe2O3. Removed oxygen originates only from the iron oxides in CF, CFS, and Fe2O3. The reducibility of Fe2O3 is better than that of CF, and that of CFS is the worst among the three phases.6) When the silica content increases from 2% to 4%, the mildly increasing of CFS content slightly degrades sample reducibility. As the silica content increases from 4% to 8%, the rapid increase of Fe2O3 improves sample reducibility. In other words, with increasing SiO2 content, the reducibility of samples is mainly influenced by CFS content in the sample with 4% silica and by Fe2O3 content in the sample with 8% silica. The presence of Fe2O3 significantly improves sample reducibility. The reduction rates of CF, CF2S, CF4S, and CF8S at three temperatures are shown in Fig. 6. The reduction routes of iron oxides can be expressed as   

F e 2 O 3 ( CF,SFC ) F e 3 O 4 FeOFe. (8)
Fig. 6.

Reduction rate of samples CF, CF2S, CF4S, and CF8S at 1123, 1173, and 1223 K.

The theoretical reduction degree is fixed at 0.11 and 0.33 when Fe2O3 reduces to Fe3O4 and Fe3O4 reduces to FeO, respectively. The Gauss peak fitting of the reduction rate for samples CF, CF2S, CF4S, and CF8S at 1173 K is shown in Fig. 7. The reduction processes of samples CF and CF8S present two stages, and those of CF2S and CF4S present three stages. The Fe3O4–to–FeO stage overlaps with the previous Fe2O3–to–Fe3O4 stage and tends to approach the following FeO–to–Fe stage with an increase in silica content. Figure 7 also shows that the reduction rate of Fe2O3 to Fe3O4 rapidly increases as silica contents increase from 2% to 8%. The peak rates of Fe2O3 to Fe3O4 are 0.012, 0.015, 0.017, and 0.022 min−1 corresponding to the reduction of CF, CF2S, CF4S, and CF8S, respectively. In Fig. 2, the Fe2O3 content increases in the CaO–Fe2O3–SiO2 system with the increase in the addition of silica, especially for CF8S. As a result, the first peak rate corresponding to the reduction of Fe2O3 to Fe3O4 gradually improves when the silica content increases. The reduction rate of Fe2O3 to Fe3O4 is much higher than that of Fe3O4 to FeO, which was also proven by Edstrom.15) However, the peak rates of FeO to Fe gradually decrease as the silica content increases from 0% to 4%. Thus, the addition of silica from 0% to 4% in the samples promotes the initial stage of reduction (Fe2O3 to Fe3O4) but prevents the final stage (FeO to Fe) in the CaO–Fe2O3–SiO2 system.

Fig. 7.

Peak fitting of the reduction rate for samples CF, CF2S, CF4S, and CF8S at 1173 K.

3.2. Apparent Activation Energy for Sample Reduction

Apparent activation energy was calculated by the slope of the plots of ln(dα/dt) against 1/T shown in Fig. 8, and the results are presented in Table 3. Given that sample reduction was combined with several reaction stages and gas diffusion, activation energy was regarded as the apparent value reflecting the comprehensive influence on the reduction process. Gas switching from pure N2 to N2 and CO mixtures led to the samples not being reduced by the target CO content during the initial stage. Consequently, the reduction stage before α=0.3 was not considered in this study for the calculation of apparent activation energy. Instead, the reduction during α=0.3 to 0.8 was considered for the activation energy calculation. The apparent activation energy values of CF, CF2S, CF4S, and CF8S reduction are 46.89, 24.48, 44.84, and 8.71 kJ·mol−1. Sample reduction tends to be easy with the addition of silica in the Fe2O3–CaO system. Moreover, the reduction of CF8S is activated the most easily among all the samples.

Fig. 8.

ln(dα/dt) against 1/T of samples CF, CF2S, CF4S, and CF8S.

Table 3. Apparent activation energy of samples CF, CF2S, CF4S, and CF8S.
E/( kJ·mol−1)
α0.30.40.50.60.70.8Avg
CF20.3223.1130.8552.8567.5186.6846.89
CF2S6.716.026.3516.2258.2053.3924.48
CF4S−4.8220.4653.7058.6672.7768.2444.84
CF8S8.586.245.183.787.4621.018.71

3.3. The Rate–controlling Step of Samples Reduction

The relationship of rate–controlling step with activation energy was investigated by Nasr16) shown in Table 4. The change of activation energy and rate–controlling step with the increasing of reduction degree was shown in Fig. 9. As for the CF8S reduction process, the E values are all less than 29 kJ·mol−1. It indicated that CO diffusion is the rate–controlling step of the entire process of CF8S reduction. The reduction rates of CF8S appeared higher than other three samples and it caused CO diffusion limited reduction process. The activation energy of CF, CF2S and CF4S reduction lies within 20–86 kJ·mol−1 with reduction degree increasing. Therefore, the rate–controlling step of CF, CF2S and CF4S reduction are performed as inner gas diffusion, inner gas diffusion and interface chemical reaction mixed control then interface chemical reaction in turns.

Table 4. Connection of rate–controlling step to activation energy of iron oxide.
E/(kJ·mol−1)Rate-controlling step
8–16Inner gas diffusion
29–42Inner gas diffusion and interface chemical reaction mixed
60–67Interface chemical reaction
>90Solid diffusion
Table 5. Model function G(α) for normal solid–state reactions.
MechanismFunctionsG(α)
One-dimensional diffusionD1(α)α2=kt
Two-dimensional diffusion (bidimensional particle shape)D2(α)(1−α)ln(1−α)+α=kt
Three-dimensional diffusion (tridimensional particle shape) Jander equation20)D3(α) [ 1- (1-α) 1/3 ] 2 =kt
Three-dimensional diffusion (tridimensional particle shape) Ginstling-Brounshtein equation21)D4(α)(1−2/3α)−(1−α)2/3=kt
Bimolecular decay law (instantaneous nucleation and unidimensional growth)F1(α)−ln(1−α)=kt
Phase boundary controlled reaction (contracting area, i.e., bidimensional shape)R2(α)1−(1−α)1/2=kt
Phase boundary controlled reaction (contracting volume, i.e., tridimensional shape)R3(α)1−(1−α)1/3=kt
Random instant nucleation and two-dimensional growth of nuclei (Avrami-Erofeev equation22,23,24))A2(α)[−ln(1−α)]1/2=kt
Random instant nucleation and three-dimensional growth of nuclei (Avrami-Erofeev equation)A3(α)[−ln(1−α)]1/3=kt
Fig. 9.

Change of activation energy and rate–controlling step with the increasing of reduction degree.

3.4. Model Function Results

3.4.1. Modified Sharp Analysis

Model function G(α) was obtained to describe the relationship of reaction degree with time. Nine normal functions17) expressing the solid–state reaction are listed in Table 4. Functions A2 and A3 were obtained from the nucleation process (crystallization) and can also be applied to the shrinking layer reaction. Considering that the model function cannot be directly obtained, Sharp18,19) defined non–dimensional parameter y(α) to target model function.   

y( α ) = G( α ) G( 0.5 ) = kt k t 0.5 = t t 0.5 , (9)
where G(0.5) represents G(α) when α=0.5 and t0.5 refers to the time when α=0.5. The curves of plots of α against t/t0.5 derived by the nine solid reactions are called standard curves. Experimental data [ti/t0.5(i=1,2,...l), αi] were obtained from the TG curves of sample reduction. Target G(α) is clear, given that the experimental data match one of the corresponding standard curves. Sharp analysis was conducted when the transition of the reaction rate peak approximately occurred at α=0.5. According to the rate change of CF, CF2S, CF4S, and CF8S reduction, y(α) was set to t/t0.1, t/t0.2, t/t0.2, and t/t0.3 for CF, CF2S, CF4S, and CF8S, respectively.

Experimental data [ti/t0.1(i=0.1,0.2,...0.8,0.9), αi], [ti/t0.2(i=0.1,0.2,...0.8,0.9), αi], [ti/t0.2(i=0.1,0.2,...0.8,0.9), αi], and [ti/t0.3(i=0.1,0.2,...0.8,0.9), αi] of CF, CF2S, CF4S, and CF8S reduction were obtained and are shown in Fig. 10. The results indicate that the experimental data y(α) values for CF and CF8S reduction mostly lie on the standard curve corresponding to function A2. The y(α) values for CF2S and CF4S reduction initially lie on the standard curve corresponding to function A2, followed by between the two standard curves determined by functions A2 and R2. The reduction of CF and CF8S can only be described by the mechanism of 2D shrinking layer reaction, and the reduction of CF2S and CF4S can be described by the mechanism of 2D shrinking layer reaction in the initial reduction stage and the 2D phase boundary–controlled reaction in the later stage.

Fig. 10.

Standard curves and experimental data based on Sharp analysis for CF, CF2S, CF4S, and CF8S.

The reduction of powdered iron ores is expressed by the porous reaction model,25) which indicates that the samples in the crucible can be regarded as layers formed by numerous fine particles, among which many pores exist having the CO diffused onto the interface of the unreacted layer with the product layer. The diffusion distance of CO and the thickness of the unreacted layer are defined as d and D, respectively, as shown in Fig. 11. Given that the flow rate of gas mixtures was set to 20 ml/min and that a large rate is not required to have the samples reduced, layer thickness D is much larger than diffusion distance d, similar to the 1 m depth going to 100 m well digging. Accordingly, the reduction of CF, CF2S, CF4S, and CF8S is described by the 2D shrinking layer reaction (for CF2S and CF4S reduction, the mechanism of the 2D phase boundary–controlled reaction is targeted in the middle and end stages). The main reduced phases are CF and Fe2O3 in the samples CF and CF8S, but a larger content of CFS phase exists in CF2S and CF4S, as shown in Fig. 12. The samples with a complicated phase composition proceed as a complicated reduction mechanism. Reduction rate analysis also indicated that the reduction of CF and CF8S presents two stages, whereas that of CF2S and CF4S presents three stages. This finding can be evidence of the phenomenon in which the reduction of CF and CF8S is controlled by a single mechanism, whereas that of CF2S and CF4S is controlled by two reduction mechanisms.

Fig. 11.

Reaction mechanism of sample reduction.

Fig. 12.

Main reduced phases in samples CF, CF2S, CF4S, and CF8S.

4. Conclusions

The isothermal reduction kinetics of CF, CF2S, CF4S, and CF8S was investigated via TG measurement with 30% CO and 70% N2 gas mixtures. The reducibility of the phases in the Fe2O3–CaO system was implied by revealing the reduction rate and apparent activation energy. The samples with 8% silica addition showed superior reducibility, and the reduction mechanisms of CF, CF2S, CF4S, and CF8S were examined through Sharp analysis. The following conclusions were obtained.

(1) The rate analysis revealed that reduction processes of samples CF and CF8S present two stages, and those of CF2S and CF4S present three stages. The Fe3O4–to–FeO stage overlaps with the previous Fe2O3–to–Fe3O4 stage and tends to approach the following FeO–to–Fe stage with an increase in silica content.

(2) The apparent activation energy values of CF, CF2S, CF4S and CF8S reduction are 46.89, 24.48, 44.84, and 8.71 kJ·mol−1. Sample reduction tends to be easy with the addition of silica in the Fe2O3–CaO–SiO2 system, especially for CF8S.

(3) CO diffusion is the rate–controlling step of the entire process of CF8S reduction, whereas the rate–controlling step of CF, CF2S and CF4S reduction are performed as inner gas diffusion, inner gas diffusion and interface chemical reaction mixed control then interface chemical reaction in turns with reduction degree increasing.

(4) Sharp analysis implied that reduction of CF and CF8S can only be described by the mechanism of 2D shrinking layer reaction, and the reduction of CF2S and CF4S can be described by the mechanism of 2D shrinking layer reaction in the initial reduction stage and the 2D phase boundary–controlled reaction in the later stage.

Acknowledgment

The authors are grateful for the financial support provided by the Natural Science Foundation of China (51544203).

Nomenclature

α: reduction degree [–]

mt and ∆m0: removed oxygen mass at a fixed time t and the theoretically removed oxygen mass from iron oxide [mg]

dα/dt: reduction rate [min–1]

k(T): rate constant [min–1]

f(α) and G(α): model function

E: apparent activation energy [kJ/mol]

A: pre–exponent [min–1]

R: gas constant, 8.314 [J·(mol·K)–1]

n: Avrami exponent [–]

y(α): a defined non–dimensional parameter [–]

d and D: diffusion distance of CO, thickness of the unreacted layer, [mm]

References
 
© 2017 by The Iron and Steel Institute of Japan
feedback
Top