Reduction of fine iron ore via a two-step fluidized bed direct reduction process

Powder Technology 254 (2014) 1–11

Contents lists available at ScienceDirect

Powder Technology journal homepage: www.elsevier.com/locate/powtec

Reduction of fine iron ore via a two-step fluidized bed direct reduction process Tao Zhang a,b,⁎,1, Chao Lei a,c,1, Qingshan Zhu a,⁎ a b c

State Key Laboratory of Multiphase Complex Systems, Institute of Process Engineering, Chinese Academy of Sciences, Beijing 100190, China Dalian National Laboratory for Clean Energy, Dalian Institute of Chemical Physics, Chinese Academy of Sciences, Dalian 116023, China Graduate University of Chinese Academy of Sciences, Beijing 100049, China

a r t i c l e

i n f o

Article history: Received 10 July 2013 Received in revised form 14 November 2013 Accepted 6 January 2014 Available online 13 January 2014 Keywords: Fluidized bed Direct reduction process Fine iron ore Defluidization Carbon precipitation

a b s t r a c t The industrial application of fluidized bed direct reduction (DR) process for fine iron ore is hampered by the sticking of direct reduction iron (DRI) particles. In the present study, the carbon precipitation reaction is coupled with the reduction reaction of fine iron ore to modify the cohesive force among DRI particles. The competition between the reduction reaction of fine iron ore and the carbon precipitation reaction leads to three types of fluidization behaviors: fluidization, unstable fluidization and defluidization. The carbon precipitation reaction is dominant at the temperature below 600–675 °C, and the presence of H2 could retard the growth of iron whiskers and promote carbon precipitation. A new type of DRI particle covered with carbon shell is therefore constructed and named as DRIc particle. The growth of iron and carbon gasification can destroy the carbon shell, and lead to the increase of stickiness; however, the presence of CO can retard or prevent the destroying. The C/Fe mass ratio on the surface has significant influence on the stickiness and also the fluidization behavior of DRIc particles. The lower limit of C/Fe mass ratio, below which defluidization occurs, increases sharply with increasing temperature. Based on these findings, a two-step fluidized bed DR process for fine iron ore is proposed and proved feasible, and the operating lines of the fluidization zones are indicated as maps. © 2014 Elsevier B.V. All rights reserved.

1. Introduction The world's iron and steel market and its industrial structures have been changing for the last several decades due to rapidly growing demands for direct reduction iron (DRI) and the rise in costs for iron and steel production. In direct reduction (DR) processes, the characteristics and cost of available iron ores play a very important role. The selection of suitable raw materials and DR process will optimize productivity, energy consumption and the overall economy of industrial plants. Driven by the high cost of raw materials (i.e., lump ore and pellets) and the environmental protection, several fluidized bed DR processes have been developed [1,2]. The main advantage of this technology is that, the fine iron ore, which accounts for about two-thirds of world's iron-ore production, can be charged directly without prior treatment to the process [1]. The bottleneck in the development of fluidized bed DR process is appearance of the sticking of DRI particles. The sticking is intrinsically caused by the sintering of freshly precipitated iron grains, and it can spread out over the whole fluidized bed during very short time. This unintentional agglomeration of particles, which results in that the gas drag force becomes too small to balance the gravitational

⁎ Corresponding authors. Tel.: +86 411 84379638; fax: +86 411 84379289. E-mail addresses: [email protected] (T. Zhang), [email protected] (Q. Zhu). 1 These authors contribute to this work equally. 0032-5910/$ – see front matter © 2014 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.powtec.2014.01.004

and buoyancy forces acting on the agglomerations, is the reason for the problem of defluidization that appears in fluidized bed DR processes [3–6]. The sticking tendency of DRI particles is influenced by the adhesive force and the breakage force. The adhesive force is directly proportional to adhesive strength and contact area, while the breakage force is a function of the momentum of the particles, the time of collision and the drag force. A great deal of effort has been made to increase the breakage force and/or reduce the adhesive force, such as increasing the particle velocity [7,8], adding inert components to coat the iron ore particles [4,6,9–12], and limiting the size range of fine iron ore [1]. The addition of inert components, such as MgO, CaO, SiO2, Al2O3, coal and coke particles, may be the most practical and effective method of preventing sticking, but it is not effective for fine iron ore with small size (e.g. less than 0.25 mm) and high metallization degree because of lower kinetic energy and more contact area [13,14]. For example, the mean diameter of the fine iron ore used in the commercial fluidized bed DR process (FINEX process) should be larger than 0.2 mm [1]. Additionally, there are problems with adding inert components into the process, such as high processing cost, high portion of inert material, and separation between DRI and inert material [4]. Since the adhesive feature of DRI particles is the root cause of sticking and defluidization, reducing the stickiness by modifying DRI particles is of key interest, and it may be a general method to avoid defluidization for fluidized bed DR process. It is well known that active iron atom

2

T. Zhang et al. / Powder Technology 254 (2014) 1–11

generated from the reduction of iron oxides, especially α-Fe2O3, is a catalyst for Boudouard reaction [15,16]. This catalytic reaction provides a way to reduce the stickiness of DRI particles via carbon precipitation. An investigation of Neuschütz [13] suggests that carbon deposition as a counter-measure against sticking of fine iron ore seems to be effective with different ores and even for extremely small particle fractions, and pellet feed coated with carbon can be completely metallized without sticking at the temperatures of 850–900 °C. An investigation of Zhang et al. [17] suggests that the carbon element deposited on the surface of Fe2O3 particles is in the forms of graphitic and Fe3C, and the prevention of sticking at 900 °C is attributed to graphitic carbon. However, the fluidization behavior and reduction behavior of fine iron ore, where the reduction reaction and the carbon precipitation reaction are coupled, are scarcely studied. For now, this method has not been applied in fluidized bed DR process because the control mechanism of fluidization behavior is not well understood. The aim of this present work is to investigate the modification of DRI particles by carbon precipitation and its influence on the fluidization behavior and the reduction behavior of fine iron ore particles in fluidized bed. More specifically, the experimental parameters affecting carbon precipitation rate, including gas composition, operating temperature and metallization degree, are investigated in detail. The morphology, structure and phase change during reduction are obtained, through which fundamental insights into the fluidization behavior and the reduction behavior of modified DRI particles are provided. Finally, a process for DRI production based on the reduction reaction and the carbon precipitation reaction is proposed and is experimentally proven. The present paper reports our first findings. 2. Experiment 2.1. Raw materials The chemical compositions of the fine iron ore supplied by Vale of Brazil are given in Table 1. The fine iron ore with diameter range of 106–150 μm [(− 45 + 60) mesh TYLER size fraction] was used in all the reduction experiments. Its true density is about 4939 kg/m3 and the bulk density is about 2209 kg/m3. The H2 and N2 gases used for reduction were of 99.999% purity, and the CO gas was of 99.99% purity. All the gases were supplied by Beijing Huayuan Gas Chemical Industry Co. Ltd. 2.2. Experimental apparatus The experimental apparatus is shown in Fig. 1. The bubbling fluidized bed reactor with inner diameter of 16 mm was made of quartz, and there were holes for measuring the temperature of the reaction zone and the pressure drop across the bed. The facility was surrounded by an electric resistance furnace, and the bed temperature control was achieved by a PID controller. The gas flow rates and pressure drop across the bed were measured by several digital mass flow controllers and a differential pressure sensor, respectively. 2.3. Experimental procedure The fluidized bed reactor, which was loaded with approximately 8 g of ore, was first purged with N2 gas at fluidized condition, and then placed into the furnace. When the aimed reaction temperature was

Table 1 Chemical composition of Brazil iron ore in mass%. Composition

TFe

Fe2O3

FeO

SiO2

CaO

MgO

Al2O3

[Mass%]

68.94

96.80

0.72

1.98

0.10

0.10

0.30

attained, the purging stream was switched to the reducing gas so as to reduce the iron ore. The fluidization was carried out at atmospheric pressure. The total gas (N2 or H2 –CO mixture) flow rate was set to 1.5 l/min at the standard state, and the superficial velocity, which depended on the experimental temperature, varied between 0.3 and 0.5 m/s. The bed was operated at bubbling fluidization regime. At the predetermined time, or at the time when defluidization occurred, the reducing gas stream was switched back to N2 gas, and the reactor was removed from the furnace. Then the reduced iron ore sample was cooled down under N2 atmosphere to the ambient temperature, and transferred into zipper seal sample bag full of N2 gas in an effort to prevent reoxidation. The carbon contents of the reduced samples were measured by a carbon–sulfur analyzer (LECO CS-344, USA). The degree of metallization (MET) is calculated on the basis of the chemical analysis of the reduced iron ore sample for the moles of total iron (Fetot) and metallic iron (Fe0) as follows [2]: 0

MET ¼ Fe =Fetot  100 in ½% 

ð1Þ

where, both Fe0 and Fetot are determined by the titrimetric method according to National Standard GB 223.7-2002 of China, and the iron of cementite is included in Fe0. Since the MET value of 100% can hardly be achieved, the value of 90% is adopted as the sign that the reduction of iron ore is nearly complete. For the cases of defluidization, the experiments were ended at the defluidization time, and for the other cases, the experiments were ended until the reduction is nearly complete. The phase compositions were characterized using X-ray diffractometer (XRD, X' Pert MPD Pro, PANalytical, the Netherlands) with the Cu Kα radiation (λ = 1.5408 Å). The microstructure of the iron ore and reduced samples was observed by a field emission scanning electron microscope (FESEM, JSM-7001F, JEOL, Japan), and the elemental concentration on the surface was analyzed by an associated energy dispersive X-ray spectroscopy (EDS, X-Max, Oxford Instruments, England). An investigation of Schouten and Van den Bleek [18] demonstrates that if defluidization occurs, part of the bed mass will no longer be involved in the fluidization process, which will lead to a lower bedpressure drop. The easiest way to detect defluidization seems to be the recording and evaluation of the average bed-pressure drop. To distinguish the feature of different fluidization behaviors, the raw data of bed-pressure drop were smoothed using the ‘smooth’ function in MATLAB. The time where defluidization occurred was recorded as the defluidization time (tdef). 3. Results 3.1. Reduction behavior of fine iron ore The metallization curves of the iron ore reduced by pure H2 at different temperatures are plotted in Fig. 2. Defluidization occurred at all the experimental temperatures during metallization of iron ore because of sticking. As the temperature increased, the adhesive strength among DRI particles increased sharply, so the MET of defluidized iron ore at higher temperature is lower. It is clear that the reduction was generally enhanced by the rise in temperature. Thus, the defluidization time decreased sharply with increasing temperature. The metallization curves of the iron ore reduced by pure CO and 80% H2–20% CO mixture, and the corresponding carbon precipitation curves at different temperatures are plotted in Figs. 3 and 4, respectively. The metallization curves indicate that the reduction was faster at the early stage and slowed down at the later stage. The reduction rate seemed to be sensitive to the gas composition, and the reaction proceeded faster with H2 than CO. For this series of experiments, the thermodynamic

T. Zhang et al. / Powder Technology 254 (2014) 1–11

3

Fig. 1. Schematic diagram of the fluidized bed apparatus: 1—gas cylinder; 2—shutoff valve; 3—mass flow controller; 4—fluidized bed reactor; 5—differential pressure sensor; 6—thermocouple; 7—display instrument; 8—data acquisition system; 9—electric resistance furnace.

conditions were met for Boudouard reaction, and CO gas could decompose according to the following reaction: 2CO ¼ CO2 þ C:

ð2Þ

Although this reaction was a high exothermic reaction, the temperature fluctuation could be controlled within ±2 °C due to the good heat releasing characteristics of the reactor. It is of interest to note that carbon was not detected in all the samples obtained during the initial 4 min. It might be because the reduction process dominated and the carbon precipitation rate was very low at the early stage. Additionally, defluidization occurred only at the early stage where no carbon could be detected. As to the cases of fluidization, the carbon precipitation reaction was promoted since the iron phase was formed, and then the mass ratio of carbon-element to iron-element (C/Fe) of the bed material increased linearly. The carbon precipitation rate of pure CO at 600 °C was about 0.0027 gC/(gFe·min), while that of 80% H2–20% CO mixture at 600 °C or 700 °C was about 0.0084 gC/(gFe·min) or 0.0047 gC/ (gFe·min), respectively. As the reaction proceeded, the C/Fe mass

Fig. 2. Metallization curves of the iron ore reduced by pure H2.

ratio could be up to 45%. The carbon content in iron carbide (Fe3C), which is about 6.67% [19], is much lower than the carbon content in the samples, and it can be therefore inferred that there must be a large amount of elemental carbon in the samples. The freshly precipitated iron showed good catalytic effect, and the carbon precipitation rates were in the same order of magnitude compared to the values obtained by Turkdogan and Vinters [20]. It can be found that the presence of H2 would actually promote carbon precipitation and the carbon precipitation reaction was significantly retarded at 700 °C compared to that at 600 °C. The catalytic effect of iron has been systematically studied by previous researchers, and it is thought that the fastest carbon precipitation rate is at 500–600 °C [16,21] and the promoting effect of H2 is generally attributed to the following reaction [21–24]: CO þ H2 ¼ C þ H2 O:

ð3Þ

Although the present work was concerned with the reduction of iron ore, these previous findings were also observed.

Fig. 3. Metallization curves of the iron ore reduced by pure CO and the corresponding carbon precipitation curves.

4

T. Zhang et al. / Powder Technology 254 (2014) 1–11

Fig. 4. Metallization curves of the iron ore reduced by 80% H2–20% CO mixture and the corresponding carbon precipitation curves.

During the reaction, a certain amount of fine carbon particles formed due to abrasion. The entrainment rate of the fine carbon particles was measured through collecting the particles with a bag filter, and the value varied between 0.0003 and 0.0009 gC/(gFe·min), which was about one order of magnitude less than the carbon precipitation rate on the bed particles. It is needed to note that the overall carbon precipitation rate is the sum of the carbon precipitation rate on the bed particles and the entrainment rate of the carbon particles, and approximately equal to the carbon precipitation rate on the bed particles. In this paper the carbon precipitation rate refers to the carbon precipitation rate on the bed particles. Fig. 5 presents the X-ray diffraction (XRD) patterns of the raw iron ore and the reduced ores. The raw iron ore appeared only hematite peaks. After the reduction with 80% H2–20% CO mixture at 600 °C for 35 min, hematite could not be detected, and the iron phases were in the forms of wustite, iron and cementite, while the carbon phases were in the forms of graphite and cementite. For carbon, graphite is the stable state and cementite is the metastable state [19]. It seems that a certain amount of cementite could exist when the iron ore was reduced at 600 °C. The non-sticking graphite and cementite should have significant influence in preventing sticking of the particles. The researches of Ruston et al. [25] and Turkdogan and Vinters [20] on Fe catalyst suggest that cementite forms as an intermediate reaction product, and it then decomposes to iron and graphite. The surface SEM micrographs of the raw iron ore and the reduced ores are displayed in Fig. 6, and the results of EDS analysis for the

Fig. 5. X-ray diffraction patterns of the raw iron ore and the reduced iron ores.

selected regions are listed in Table 2. The raw ore showed a dense structure, and the O/Fe atomic ratio was about 1.5. It is clear that the surface of the ore reduced by CO showed a whisker structure, while that reduced by H2 showed a very porous structure with no iron whiskers. For the sample reduced by pure CO, a number of iron (or cementite) whiskers grew toward the exterior of the iron ore particles, and some graphite grains could be found on the whiskers and among the whiskers. For the sample reduced by 80% H2–20% CO mixture, the surface seems to be covered with an intact carbon shell, which is constituted of graphite grains with mean size less than 1 μm. Its cross-section is displayed in Fig. 7, and the results of EDS analysis for the selected four spots are also listed in Table 2. From this figure, it can be seen that the characteristic sizes of carbon shell, pore and also iron (or ferrous oxide) grain are less than 1 μm. Since the penetration depth of the electron beam, which is larger than 1 μm, is larger than the characteristic sizes, the EDS analysis of the phases may have influence on each other. In this experiment the epoxy resin was used to fix the sample, and the oxygen concentration in it (Spectrum 5) was about 18%. This feature could be used to distinguish the epoxy resin and the carbon deposited on the particle. It is shown that the particle was surrounded by a shell with a thickness less than 1 μm. The oxygen concentration in the shell (Spectrum 6) was about 5%, and it might be in the form of ferrous oxide. From the elemental compositions and the XRD results, it could be inferred that the shell might be composed by graphite and small amounts of iron, ferrous oxide and cementite. Since the O/Fe atomic ratio was about 0.74, the white phase (Spectrum 7) might be a mixture of iron (the relatively lighter region) and ferrous oxide (the relatively darker region). There were lots of pores in the particle, and most of the pores were occupied by graphite grains as revealed by the EDS analysis result (Spectrum 8). During the reduction of iron ore particle with pure CO or H2–CO mixture, the carbon precipitation rate in the pores may be smaller than the rate on the surface due to diffusion limitation and lack of active iron grains. Thus the C/Fe mass ratio on the surface is much higher than the average C/Fe mass ratio, which means that carbon is not evenly distributed in the particle. Especially for the sample reduced by 80% H2–20% CO mixture, this feature is very obvious, and the C/Fe mass ratio on the surface is about 6 times of the average C/Fe mass ratio. This new type of DRI particle covered with carbon shell was named as DRIc particle. Since the graphite grains, which are strongly adhered to the iron particles, are noncohesive at the reaction condition, the stickiness of DRIc particles is therefore ignorable compared to DRI particles. Carbon precipitation under such circumstances has the technological merit of reducing stickiness. 3.2. Fluidization behavior of fine iron ore The fluidization behavior of fine iron ore was investigated systematically in the temperature range of 600–800 °C during reduction by pure CO, pure H2 or H2–CO mixtures. The experiments with 11 different gas compositions and 9 different temperatures were carried out, thus 99 independent data points were obtained to indicate the field of different fluidization states. As illustrated in Fig. 8, when the iron ore was reduced by pure H2, defluidization occurred during metallization of iron ore for all the experiments. After defluidization, most of the particles adhered together, and a hole formed in the bed. It can be inferred that the bed might go through three stages: fluidized bed, channeling fluidized bed and fixed bed. However, when the iron ore was reduced by pure CO or H2–CO mixtures, the fluidization behavior seemed to be more complex. It is known that the weight of fluidized particles in fluidized bed could be measured by pressure drop. The pressure drop curves and their baselines of three typical fluidization behaviors, which are named as fluidization, unstable fluidization and defluidization, are illustrated in Fig. 9. As to the case of fluidization, the baseline of pressure drop curve maintained approximately horizontal. As to the case of unstable fluidization,

T. Zhang et al. / Powder Technology 254 (2014) 1–11

Fig. 6. SEM micrographs of (a, b) the raw iron ore and the ore reduced by (c, d) H2, (e, f) CO and (g, h) 80% H2–20% CO at 600 °C for 35 min.

5

6

T. Zhang et al. / Powder Technology 254 (2014) 1–11

Table 2 Elemental analysis results of the raw ore and the reduced ores.a Spectrum

1 2 3 4 5b 6 7 8 9 10 11 12 a b

Element [mass%] Fe

O

C

68.96 96.86 82.37 42.48 1.36 37.95 73.54 46.49 97.85 47.07 63.73 51.37

30.63 2.84 1.74 6.06 18.42 5.09 15.52 9.21 0.24 0.91 0.52 0.82

– – 15.73 51.03 80.03 56.64 10.68 44.16 1.74 51.75 35.36 47.51

C/Fe mass ratio [%]

O/Fe atomic ratio

– – 19.09 120.13 – 149.25 14.52 94.99 1.78 109.94 55.48 92.49

1.55 0.10 0.07 0.50 – 0.47 0.74 0.69 0.01 0.07 0.03 0.06

Other elements, such as Al, Si Mg and Ca, in trace amounts were also detected. Epoxy.

there existed a ‘V’ type change in its baseline. It means that a local partial defluidization occurred temporarily, but then the defluidized particles were gradually fluidized during further reduction. As to the case of defluidization, defluidization occurred suddenly with a sharp decrease of pressure drop. All of the above phenomena were also confirmed by direct observation. It is needed to note that the influence of weight change on the pressure drop curve due to the loss of oxygen and carbon precipitation was ignorable compared to that due to defluidization or local partial defluidization, and was therefore not obviously revealed in the curves. For the cases of fluidization and unstable fluidization, the samples were reduced for more than 3 h to be metallized completely. As the reaction proceeded, more carbon precipitated on the particles, the bed expanded further (after the temporarily partial defluidization for the cases of unstable fluidization), and the fluidization quality seemed to be improved. The fluidization and unstable fluidization behaviors were only observed when reduced by H2–CO mixtures or pure CO, accompanying with carbon precipitation. It is generally believed that once an iron phase is formed, carbon precipitation will proceed, and the reduction and the carbon precipitation processes will take place in parallel [26]. However, the carbon precipitation reaction is also significantly influenced by thermodynamic condition. The equilibrium CO2/CO ratio of Boudouard reaction decreases sharply with increasing temperature, and the values are about 2.86, 0.62 and 0.12 at 600, 700 and 800 °C [27], respectively. For this series of experiments, the CO2/CO ratio, which was about 0 at the inlet of the fluidized bed reactor, increased along the axial direction, and might exceed the equilibrium CO2/CO ratio at the outlet at the early stage of the reaction, especially for the

Fig. 8. Fluidization behavior of fine iron ore and the operating line of the fluidization zone.

reaction at higher temperature. For example, when the iron ore was reduced by 80% H2–20% CO at 800 °C, the CO2/CO ratio of the off gas, which was analyzed by gas chromatograph, was higher than 0.5 throughout the reaction. It can be therefore inferred that the thermodynamic condition was not met for Boudouard reaction in most of the reaction zone because the CO2/CO ratio of the off gas was far higher than the equilibrium CO2/CO ratio. However, as the reaction proceeded (for the fluidization cases), at the later stage, the CO2/CO ratio of the off gas tended to be close to that at the inlet because of the sharp decrease of the reduction rate, thus the thermodynamic condition might be met for Boudouard reaction throughout the reaction zone. It is clear that the carbon precipitation reaction might be retarded by the reduction reaction of iron ore, which consumed a large portion of CO preferentially. As to the cases of defluidization, the temperature was relatively higher, the carbon precipitation was therefore significantly retarded due to the severe thermodynamic limit, and defluidization occurred at the early stage. As to the cases of fluidization, the temperature was relatively lower, and the thermodynamic condition might be met for Boudouard reaction throughout the reaction zone even at the early stage. The carbon precipitation reaction could proceed at a fast rate once a little amount of iron precipitated on the particle surface, thus the iron ore could be well fluidized throughout the reaction. As to the cases of unstable fluidization, the reduction process overwhelmed the carbon precipitation process at the early stage, leading to a local partial defluidization, but the carbon precipitation process overwhelmed the reduction process at the later stage because of the sharp decrease of the reduction rate, therefore, the defluidized particles were gradually fluidized.

Fig. 7. The cross-section SEM micrographs of the ore reduced by 80% H2–20% CO at 600 °C for 35 min.

T. Zhang et al. / Powder Technology 254 (2014) 1–11

7

Fig. 9. Typical pressure drop curves of the three fluidization behaviors obtained with 80% H2–20% CO mixture.

The unstable fluidization was the transition from fluidization to defluidization, and was observed in a broad temperature range. From Fig. 8, it can be seen that the transition temperature range is significantly influenced by the gas composition, and both its upper bound (Tub) and lower bound (Tlb) seem to increase linearly as the concentration of H2 increases from 0 to 0.9. The Tub or Tlb was derived through linear fitting the data points near the upper bound or the lower bound. For each gas composition, one data point above the bound, and one data point below the bound were used to fit Tub or Tlb. The results revealed that Tub = 637.9 + 143.5[H2] and Tlb = 610.4 + 99.2[H2], respectively. The increase of Tub and Tlb with increasing H2 concentration is caused by two main reasons. One reason is the promoting effect of H2 on carbon precipitation, which promotes the formation of carbon shell. And the other reason is that the iron whiskers, which increase the tendency to sticking because of hooking mechanically the particles together, tend to precipitate on the particle surface, when the iron ore is reduced by the gas mixture with higher CO concentration. It is needed to note that Tlb is the operating line of the fluidization zone. It can be concluded that the iron ore tends to defluidize during reduction at higher temperature for two main reasons: (a) the stickiness of iron is higher at higher temperature; (b) as the temperature rises, the reduction reactions are enhanced, while the carbon precipitation reactions are retarded mainly because of the thermodynamic limit. The most attractive finding implied in Figs. 8 and 9 is that the DRIc particles, which are noncohesive, can be constructed via carbon precipitation reaction at the temperature lower than 600–675 °C without defluidization or local partial defluidization. Since the production efficiency is very low at low temperature, e.g. to achieve the MET of 90% at 600 °C, the reaction time required is more than 2 h. The commercial DR processes are usually operated at the temperature higher than 800 °C [2] to obtain higher production efficiency. Therefore, the fluidization behavior and reduction behavior of DRIc particles at high temperature should be investigated.

the N2 gas was switched to reducing gases with different compositions, and the reduction behavior and the fluidization behavior of DRIc particles were investigated at 800 °C. The MET and C/Fe mass ratio were measured during the process of constructing DRIc particles and the following processes. The data, which are plotted in Fig. 10, show that in step (b), the DRIc particles could be reduced by the deposited carbon under nitrogen atmosphere, and the MET increased from 82.3% to 88.3% while the C/Fe mass ratio decreased from 20.6% to 19.5%. It has now been accepted that the direct reduction in Fe–O–C system is mainly carried out via gaseous intermediates and the overall rate is controlled by the gasification of carbon [28,29]. When CO forms and diffuses through the DRIc particles, favorable condition for the reduction of iron oxide inside the particles is obtained. The reduction and carbon gasification reactions are represented by the following reactions: FeO þ C ¼ Fe þ CO

ð4Þ

FeO þ CO ¼ Fe þ CO2

ð5Þ

C þ CO2 ¼ 2CO:

ð6Þ

3.3. Reduction behavior and fluidization behavior of DRIc particles The DRIc particles were prepared at 600 °C, and then the reduction behavior and the fluidization behavior were investigated at 800 °C. The experiment included three steps: (a) the iron ore was reduced by 80% H2–20% CO mixture at 600 °C for 35 min to construct DRIc particles; (b) the reducing gas was switched to N2 gas, and then the reactor temperature was raised to 800 °C under N2 atmosphere in 8 min; and (c)

Fig. 10. Reduction curves of the noncohesive DRIc particles at 800 °C.

8

T. Zhang et al. / Powder Technology 254 (2014) 1–11

As the temperature rose from 600 to 800 °C, these reactions were intensified, especially between 700 and 800 °C, and it was confirmed by the increase of the partial pressures of CO and CO2 in the off gas. In step (c), the DRIc particles were further reduced at 800 °C, and the MET increased from 88.3% to more than 93% in 10 min for all of the three reducing gas compositions. The reduction rate decreased significantly when the MET was over 93%, which was mainly attributed to the increase of diffusion resistance of the iron layer. When the particles were reduced by pure H2, the C/Fe mass ratio decreased linearly from 19.5% to 3.6% in 37 min mainly due to carbon gasification. The XRD pattern shows that after reduction at 800 °C, cementite, which was metastable at this temperature, was decomposed nearly completely, and both the amounts of cementite and carbon were below the lower detection limit of XRD. The off gas was analyzed by gas chromatograph to investigate the carbon gasification reaction, and the result showed that methane was the primary product gas. The reaction may produce methane through the following reaction: C þ 2H2 ¼ CH4 :

ð7Þ

An investigation of Gilliland and Harriott [30] on Fe catalyst suggests that the deposited carbon is reactive toward hydrogen. This is consistent with the present study. As shown in Fig. 11(a) and (b), the carbon shells disappeared because of carbon gasification, and iron grew toward the exterior of the particles with a finger-like morphology. The EDS analysis of the selected region (Spectrum 9) indicates that the iron element concentration on the surface was 97.85%, but the carbon element concentration was only 1.74%. This result further confirmed that the surfaces of the particles were nearly completely occupied by iron. The stickiness of DRIc particles increased due to the gasification of carbon shells, and

finally led to defluidization, which is illustrated in Fig. 12, at the time of 37 min. To avoid destroying the carbon shells, H2–CO mixtures were selected to perform the experiments. It can be seen that carbon gasification was inhibited by increasing CO concentration. In detail, the C/Fe mass ratio was stabilized at the level of about 19% when the CO concentration was 13.3%, and had a net increase of about 16% in 60 min when the CO concentration was 20.0%. Additionally, cementite seemed to be decomposed to iron and graphite nearly completely, as revealed by XRD pattern given in Fig. 5. During further reduction of DRIc particles by 86.7%H2–13.3%CO mixture at 800 °C, iron grew toward the exterior of the particles, however, the carbon shells could still remain almost intact because of the dynamic modification of the particles by carbon precipitation, as shown in Fig. 11(c) and (d). From the elemental analysis results listed in Table 2, it is proved that the white spots (Spectrum 11) contained a relatively higher concentration of iron element than the dark regions (Spectrum 10), and the average C/Fe mass ratio on the surface was about 0.92 (Spectrum 12), which was approximate to the C/Fe mass ratio of the dark regions. Because of the dynamic modification of the precipitated iron, the stickiness of DRIc particles was limited at a low level, and the bed was well fluidized during further reduction. The corresponding pressure drop curve is also illustrated in Fig. 12. It is clear that besides temperature, the parameter of C/Fe mass ratio also has significant influence on the stickiness of DRIc particles. The experimental result above reveals that if the C/Fe mass ratio of DRIc particles is lower than 3.6%, the particles will defluidize at 800 °C, and it can be therefore inferred that the lower limits of C/Fe mass ratio, below which defluidization occurs, may be different at different temperatures. To obtain the lower limits, the experimental temperatures of step (c) were changed to 850, 900 and 950 °C, respectively, and the H2–CO

Fig. 11. SEM micrographs of the noncohesive DRIc particles reduced by (a, b) pure H2 and (c, d) 86.7%H2–13.3%CO mixture at 800 °C.

T. Zhang et al. / Powder Technology 254 (2014) 1–11

9

Fig. 12. Pressure drop curves of the noncohesive DRIc particles.

mixtures with CO content of 0%, 5% and 10% were selected to carry out the experiments. The lower limits of C/Fe mass ratio of DRIc particles at different temperatures are depicted in Fig. 13. The gas composition seems to have little influence on the lower limit of C/Fe mass ratio, and the value at the same temperature varies irregularly in a narrow range as the CO content in the gas mixture increases. These variations may be resulted from the experimental errors. The stickiness of DRIc particles depends on its properties, such as temperature, C/Fe mass ratio and morphology. As revealed by the experimental results in Section 3.1, when the iron ore particles are reduced by pure CO, the iron whiskers would be formed, but when the iron ore particles are reduced by H2–CO mixture with a little amount of CO (even the CO content is as high as 20%), a very porous structure with no iron whiskers would be formed. It means that the variation of CO content in the range of 0 to 10% has little influence on the morphology of reduced particles. Maybe this is the reason why the gas composition has little influence on the lower limit of C/Fe mass ratio of DRIc particles. In spite of the variation, the operating line of fluidization zone can be estimated through linear fitting and indicated as a map using two parameters: temperature and C/Fe mass ratio, as illustrated in Fig. 13.

Fig. 13. Operating line of the fluidization zone for noncohesive DRIc particles.

Since the stickiness of iron increases as the temperature rises from 800 to 950 °C, the lower limit of C/Fe mass ratio increases sharply from ~ 4% to ~ 16%. It is of interest to note that the C/Fe mass ratio of DRIc product could be adjusted above the lower limit, and the carbon could provide chemical energy for smelting the DRIc product in the electric arc furnace. 4. Discussion The structure change of an iron ore particle during reduction is illustrated by diagrammatic sketches in Fig. 14. When the noncohesive iron ore particle is reduced by pure H2, the cohesive DRI particle is the final product. The iron oxide is reduced from the outer surface to the inner core. When the channels are formed, the iron grains will also precipitate on the channel surfaces. As shown in the white phase in Fig. 7(b), the ferrous oxide phase (the relatively darker region) is surrounded by the iron phase (the relatively lighter region). The stickiness increases monotonously with increasing MET, and will lead to defluidization finally. When the noncohesive iron ore particle is reduced by H2–CO mixture, iron precipitates on the particle surface and the channel surfaces firstly and then the carbon precipitation reaction is induced by the freshly precipitated iron at appropriate conditions, such as low temperature and low CO2/CO ratio, thus the proportion of the particle surface occupied by iron increases firstly and then decreases. Since Fe2O3, FeO and C are noncohesive but Fe is cohesive, the stickiness of the particle varies following the change of the iron amount on the surface. This tendency is very obvious especially in the transition temperature range, where a local partial defluidization may occur. The carbon precipitation reaction can be dominant through adjusting the reaction conditions as suggested by Fig. 8, such as reduced by the gas containing a certain amount of CO at a relatively lower temperature, therefore, the carbon shell can be formed efficiently in fluidized state, and the channels can also be filled with graphic grains as shown in Fig. 7. This dynamic modification by carbon precipitation leads to the formation of noncohesive DRIc particles. But at a relatively lower temperature, the iron oxide is hard to be reduced completely, thus, the noncohesive DRIc particles should be further reduced at a relatively higher temperature to complete the reduction. During further reduction of DRIc particles, because of the growth of iron and carbon gasification, some iron grains may penetrate the carbon shell and precipitate on the particle surface as shown in Fig. 11. Thus to avoid the severe destruction of carbon shell, the

10

T. Zhang et al. / Powder Technology 254 (2014) 1–11

Fig. 14. Diagrammatic sketches of the structure change of an iron ore particle during reduction by (a) H2 and (b) H2–CO mixture.

reducing gas should have the ability to repair the carbon shell. At such a circumstance, the carbon shell could remain almost intact. The method of coating iron ore particles with inert components has been suggested by some researchers to avoid defluidization [2,4,10–12]. But during reduction, abundant iron grains or whiskers may penetrate the coating and precipitate on the surface [12], and also the iron ore particles may disintegrate [4], thus the coating will be destroyed and lose effectiveness, and as a result, the stickiness will increase and defluidization will occur. The dynamic modification of the precipitated iron grains by carbon precipitation can construct the carbon shells and also repair the destroyed carbon shells in real-time, so it can overcome the above shortcoming. The speed of constructing or repairing the carbon shells depends on the carbon precipitation rate, while the speed of destroying the carbon shells mainly depends on the carbon gasification rate, the abrasion rate of surficial carbon and the growth rate of iron grains or whiskers from the inner to the surface. The competition among these reactions plays a key role in controlling the stickiness and the fluidization behavior. In the practical application, the stickiness can be limited by adjusting the rates of these reactions, which are significantly influenced by temperature and gas composition, to avoid defluidization. It is clear that the reduction process can be divided into two stages. At the early stage, the aim is to modify the cohesive surface and construct DRIc particles, thus the speed of constructing carbon shell should be as fast as possible. At the later stage, the aim is to reduce the noncohesive DRIc particle efficiently to achieve high MET in fluidized state, thus the reduction rate and the fluidization quality are both of importance. The present work reveals that the reaction conditions are quite different between the two stages. Therefore, a two-step fluidized bed DR process for fine iron ore is proposed, and it includes the following two steps: (1) constructing the noncohesive DRIc particles by H2–CO mixture at the temperature lower than 600–675 °C, where the carbon precipitation reaction is dominant, and (2) further reducing the noncohesive DRIc particles at the temperature higher than about 800 °C, where the reduction is dominant, to obtain high MET. For step (1), the operating line of the fluidization zone is indicated as a map using two parameters: temperature and H2 concentration. And for step (2), the operating line of the fluidization zone is indicated as a map using two parameters: temperature and C/Fe mass ratio of DRIc particles. This type of DRI product may have the advantage of added stability due to its carbon content, and the combined carbon can provide chemical energy to the electric arc furnace. This proposed two-step fluidized bed DR process provides a promising way to overcome the problem of sticking and defluidization. However

the reducing gas in industrial plant always contains certain amounts of CO2, H2O, H2S and etc., and they might hamper the reduction and the carbon precipitation. The iron ore may contain certain amounts of SiO2, Al2O3, TiO2, V2O5 and etc., and they might hamper or promote the reduction and the carbon precipitation. An interesting work in future would be to investigate the influence of the components in the reducing gas, the ore components as well as the temperature on the morphology, structure, reduction characteristics and fluidization behavior of DRIc particles in this proposed two-step fluidized bed DR process.

5. Conclusions The cohesive force among DRI particles was modified by carbon precipitation, and defluidization was therefore prevented. The competition between the reduction reaction of iron ore and the carbon precipitation reaction led to three types of fluidization behaviors: fluidization, unstable fluidization and defluidization. At the temperature below 600–675 °C, the carbon precipitation reaction was dominant, thus the noncohesive DRIc particles, which were covered with carbon shells, could be constructed from iron ore particles under H2–CO atmosphere in fluidized state. At the temperature above 650–775 °C, the reduction of iron ore was dominant, and the DRIc particles could be reduced by H2–CO mixture nearly completely without defluidization. The C/Fe mass ratio on the surface, which can be controlled by the dynamic modification of precipitated iron grains, has significant influence on the stickiness of DRIc particles and plays a key role in controlling the fluidization behavior. It increases due to carbon precipitation, and decreases mainly due to carbon gasification and the growth of iron. The lower limit of C/Fe mass ratio of DRIc particles, below which defluidization occurs, increases sharply with increasing temperature. Based on constructing noncohesive DRIc particles, a fluidized bed DR process for fine iron ore is proposed, and it contains two steps: (1) constructing the DRIc particles by H2–CO mixture at the temperature lower than 600–675 °C, and (2) further reducing the DRIc particles at the temperature higher than about 800 °C to obtain high MET. For step (1), the operating line of the fluidization zone is indicated as a map using two parameters: temperature and H2 concentration. The presence of H2 can retard the growth of iron whiskers and promote carbon precipitation. And for step (2), the operating line of the fluidization zone is indicated as a map using two parameters: temperature and C/Fe mass ratio of DRIc particles. The growth of iron and carbon gasification can destroy the carbon shells, but the presence of CO can retard or prevent the destroying.

T. Zhang et al. / Powder Technology 254 (2014) 1–11

Acknowledgments The authors wish to gratefully acknowledge the financial support from the State Key Development Program for Basic Research of China (973 Program) under grant no. 2013CB632603. References [1] F.J. Plaul, C. Böhm, J.L. Schenk, Fluidized-bed technology for the production of iron products for steelmaking, J. S. Afr. Inst. Min. Metall. 109 (2009) 121–128. [2] J.L. Schenk, Recent status of fluidized bed technologies for producing iron input materials for steelmaking, Particuology 9 (2011) 14–23. [3] Q. Zhu, R. Wu, H. Li, Direct reduction of hematite powders in a fluidized bed reactor, Particuology 11 (2013) 294–300. [4] M. Komatina, H.W. Gudenau, The sticking problem during direct reduction of fine iron ore in the fluidized bed, Metalurgija 10 (2004) 309–328. [5] J.F. Gransden, J.S. Sheasby, The sticking of iron ore during reduction by hydrogen in a fluidized bed, Can. Metall. Q. 13 (1974) 649–657. [6] H.W. Gudenau, J. Fang, T. Hirata, U. Gebel, Fluidized bed reduction as the prestep of smelting reduction, Steel Res. 60 (1989) 138–144. [7] S. Hayashi, S. Sayama, Y. Igushi, Relation between sulfur pressure and sticking of fine iron ores in fluidized bed reduction, ISIJ Int. 32 (1992) 962–971. [8] B. Zhang, Z. Wang, X. Gong, Z. Guo, A comparative study of influence of fluidized conditions on sticking time during reduction of Fe2O3 particles with CO, Powder Technol. 225 (2012) 1–6. [9] P.L.M. Wong, M.J. Kim, H.S. Kim, C.H. Choi, Sticking behaviour in direct reduction of iron ore, Ironmak. Steelmak. 26 (1999) 53–57. [10] J. Shao, Z. Guo, H. Tang, Effect of coating MgO on sticking behavior during reduction of iron ore concentrate fines in fluidized bed, Steel Res. Int. 84 (2013) 111–118. [11] Y. Zhong, Z. Wang, Z. Guo, Q. Tang, Defluidization behavior of iron powders at elevated temperature: influence of fluidizing gas and particle adhesion, Powder Technol. 230 (2012) 225–231. [12] Y. Zhong, Z. Wang, Z. Guo, Q. Tang, Prevention of agglomeration/defluidization in fluidized bed reduction of Fe2O3 by CO: the role of magnesium and calcium oxide, Powder Technol. 241 (2013) 142–148. [13] D. Neuschütz, Sticking prevention during fine-ore metallization in two-stage smelting-reduction processes, Steel Res. 62 (1991) 333–337.

11

[14] D. Neuschuetz, T. Hoster, Concept and present state of a coal-based smelting reduction process for iron ore fines, Steel Res. 60 (1989) 113–119. [15] R. Schenck, F. Zimmermann, Decomposition of carbon monoxide and chemical equilibrium in the blast furnace, Ber. Dtsch. Chem. Ges. 36 (1903) 1231–1251. [16] M.-W.P. Xu, J.B. Jesse Jr., Mechanism of iron catalysis of carbon monoxide decomposition in refractories, J. Am. Ceram. Soc. 72 (1989) 110–115. [17] B. Zhang, Z. Wang, X. Gong, Z. Guo, Characterization of precipitated carbon by XPS and its prevention mechanism of sticking during reduction of Fe2O3 particles in the fluidized bed, ISIJ Int. 53 (2013) 411–418. [18] J.C. Schouten, C.M. Van den Bleek, Monitoring the quality of fluidization using the short-term predictability of pressure fluctuations, AIChE J. 44 (1998) 48–60. [19] G. Krauss, Steels: Processing, Structure, and Performance, first ed. ASM International, 2005. [20] E.T. Turkdogan, J.V. Vinters, Catalytic effect of iron on decomposition of carbon monoxide I. Carbon deposition in H2–CO mixtures, Metall. Trans. 5 (1974) 11–20. [21] M.I. Nasr, A.A. Omar, M.M. Hessien, A.A. El-Geassy, Carbon monoxide reduction and accompanying swelling of iron oxide compacts, ISIJ Int. 36 (1996) 164–171. [22] T.F. Berry, R.N. Ames, R.B. Snow, Influence of impurities and role of iron carbides in deposition of carbon from carbon monoxide, J. Am. Ceram. Soc. 39 (1956) 308–318. [23] P.L. Walker, J.F. Rakszawski, G.R. Imperial, Carbon formation from carbon monoxide–hydrogen mixtures over iron catalysts. I. Properties of carbon formed, J. Phys. Chem. 63 (1959) 133–140. [24] R.G. Olsson, E.T. Turkdogan, Catalytic effect of iron on decomposition of carbon monoxide: II. Effect of additions of H2, H2O, CO2, SO2 and H2S, Metall. Trans. 5 (1974) 21–26. [25] W.R. Ruston, M. Warzee, J. Hennaut, J. Waty, The solid reaction products of the catalytic decomposition of carbon monoxide on iron at 550 °C, Carbon 7 (1969) 47–57. [26] N. Towhidi, J. Szekely, The influence of carbon deposition on the reduction kinetics of commercial grade hematite pellets with CO, H2, and N2, Metall. Trans. B Process Metall. 14B (1983) 359–367. [27] P.L. Walker Jr., F. Rusinko Jr., L.G. Austin, Gas reactions of carbon, in: D.D. Eley, P.W. Selwood, B.W. Paul (Eds.), Advances in Catalysis, Academic Press, New York, 1959, pp. 133–221. [28] H.A. Mashhadi, A.R. Rastgoo, J.V. Khaki, An investigation on the reduction of iron ore pellets in fixed bed of domestic non-coking coals, Int. J. Iron Steel Soc. Iran 5 (2008) 8–14. [29] R.J. Fruehan, The rate of reduction of iron oxides by carbon, Metall. Trans. B 8 (1977) 279–286. [30] E.R. Gilliland, P. Harriott, Reactivity of deposited carbon, Ind. Eng. Chem. 46 (1954) 2195–2202.