Thermal Decomposition Behaviour of Fine Iron Ore Particles

Transcription

1 , pp Thermal Decmpsitin Behaviur f Fine Irn Ore Particles Yingxia QU, 1) * Yngxiang YANG, 2) Zngshu ZOU, 1) Christiaan ZEILSTRA, 3) Ken MEIJER 3) and Rb BOOM 2) 1) Schl f Materials and Metallurgy, Nrtheastern University, Bx 312, NO. 3-11, Wenhua Rad, Heping District, , Shenyang, P. R. China. 2) Department f Materials Science and Engineering, Delft University f Technlgy, Mekelweg 2, 2628 CD Delft, the Netherlands. 3) Tata Steel, 1970 CA IJmuiden, the Netherlands. (Received n April 9, 2014; accepted n May 28, 2014) In the smelting cyclne f HIsarna prcess, bth thermal decmpsitin and gaseus reductin f irn re cntribute t the expected pre-reductin degree abut 20%. Hwever, the fine re reductin and melting prcess in the smelting cyclne is extremely fast and it is very difficult t differentiate between the thermal decmpsitin and gaseus reductin. This study fcused n the thermal decmpsitin mechanism f the fine irn re under different cnditins. Firstly, the theretical evaluatin has been cnducted based n the thermdynamics, and then the labratry investigatin was cnducted in three stages with three reactrs: the TGA-DSC, the electrically heated hrizntal tube furnace and the High-temperature Drp Tube Furnace (HDTF). Accrding t the experimental results f the first tw stages and the theretical evaluatin, it was fund that the temperature f intensive thermal decmpsitin f Fe 2O 3 in the inert gas envirnment is in the range f K, while the thermal decmpsitin f Fe 3O 4 culd be sped up when the temperature is abve K in the inert gas. Temperature plays an imprtant rle in the thermal decmpsitin degree and reactin rate. Finally, it was fund that the thermal decmpsitin f the individual irn re particles tk place very rapidly in the HDTF and n significant influence f the particle size and residence time (t ms) n the equivalent reductin degree culd be bserved, when the particle diameter was smaller than 250 μm in the CO 2 gas. KEY WORDS: HIsarna prcess; smelting reductin; fine irn re particles; high temperature; thermal decmpsitin. 1. Intrductin Nwadays the blast furnace prcess is still the primary irn making technlgy. Fr achieving a mre efficient peratin frm an energy and ecnmic pint f view, varius revlutinary technlgies are nw develped as alternative irnmaking prcesses such as COREX and FINEX already in cmmercial peratin, Circsmelt and HIsarna 1,2) under develpment. HIsarna, as shwn in Fig. 1, is ne f these prmising technlgies under intensive develpment, based n smelting reductin principles. It is the technlgy which has the ptential t reduce emissins f carbn dixide (CO 2 ) cmpared t the blast furnace steelmaking rute by mre than 50%. 3) The cre f the prcess cnsists f three reactrs, a smelting cyclne, a smelting reductin vessel (SRV) and a cal pyrlyser. Due t the high temperatures in the furnace, fine irn re particles are melted in the smelting cyclne. The melt is cllected n the water-cled sidewalls f the cyclne sectin, runs dwn the wall and drps int the liquid bath in the SRV where the final reductin is cmpleted. The riginal irn re particles are injected int the cyclne reactr and pre-reduced t a reductin degree f abut 20% thrugh thermal decmpsitin and reductin by the pst-cmbustin gases arising frm the SRV. Up t nw, * Crrespnding authr: quyingxia800@163.cm DOI: mst f the previus studies 4 6) fcused n the mechanisms f the gas-hematite reactin at lw temperature and the gaswüstite reactin at high temperature with the backgrund f the blast furnace prcess. The hematite decmpsitin was usually neglected because f the requirement f the higher temperature than gas-hematite reductin. The smelting cyclne f the HIsarna prcess is a cmplicated high temperature reactr. In the smelting cyclne, the temperature is extremely high and the size f the irn re particles is very small. The smelting cyclne prvides a suitable envirnment fr the thermal decmpsitin f irn re particles. Althugh xygen is injected int the cyclne tgether with irn re particles, it preferably reacts with carbn mnxide and hydrgen rapidly and generates sufficient energy t prduce a large amunt f energy which heats the particles and the gas quickly. The fine irn xide particles in the smelting cyclne experience a series f physical and chemical changes including rapid heating up, thermal decmpsitin, gas-slid particle reductin, melting, and gas-mlten particle reductin. Fr that all the prcesses are extremely fast and it is very difficult t differentiate between the thermal decmpsitin and gaseus reductin. It is a fundamental questin whether the thermal decmpsitin f hematite in the cyclne culd be an imprtant factr f determining the pre-reductin degree (abut 20%) f irn re. Frm the thermdynamics pint f view, the prducts f the thermal decmpsitin and reductin are the same 2014 ISIJ 2196

2 Fig. 1. Schematic diagram f the HIsarna prcess. 7,8) because bth f the prcesses remve xygen frm the high xygen cncentratin irn xides and generate the lw xygen cncentratin irn xides. Hwever, frm the kinetics pint f view, the required reactin cnditins and reactin rates are different. In rder t btain a cmprehensive and clear understanding f the irn re behaviur in the smelting cyclne, the labratry experiments n thermal decmpsitin f irn re particles has been carried ut. On the ther hand, as that the smelting cyclne and SRV have a clse cntact and the reactins in the smelting cyclne wuld directly affect the peratin parameters in the SRV. Therefre, the kinetic study f the thermal decmpsitin f irn re under the cnditins f the smelting cyclne will be helpful t ptimize the whle HIsarna prcess. 2. Theretical Evaluatin There are three main frms f irn xides: hematite (Fe 2 O 3 ), magnetite (Fe 3 O 4 ) and wüstite (FeO), and their melting pints are K, K and K, respectively. Accrding t the literature, the thermal decmpsitin f hematite re has been mentined in sme labratry experiments. Fr example, Gilles 9) stated that the highest xide frm is unstable abve K and it decmpses int gaseus xygen and an xide cntaining 71.6% irn which is clse t the cmpsitin f magnetite. This reactin was als mentined in the study f Nakamura et al. 10) in A labratry-scale test was made, in which irn xide cntained in a water-cled crucible was melted and reduced by using 10 15% H 2 Ar transferred arc plasma. It was stated that at high temperatures, the xygen remved by thermal decmpsitin befre the start f the reductin crrespnds t a degree f reductin f abut 18%. The equatin fr calculating the equivalent reductin degree R due t thermal decmpsitin 11) and thermal decmpsitin reactins f irn xides are as fllws. m R = Δ m xygen tt xygen... (1) 6Fe2O3(s) = 4Fe3O4(s) + O2(g)... (2) ΔG = T 2Fe3O4(s) = 6FeO(l) + O2(g)... (3) ΔG = T FeO(l) = Fe(l) + O2 (g)... (4) ΔG = T The equivalent reductin degree R f the irn re is defined as the rati f weight lss f xygen Δm xygen t the ttal initial mass f xygen in the irn xides in the irn re sample m tt-xygen as shwn in Eq. (1). The reversible reactins and their standard Gibbs free energy (J/(ml K)) are shwn in Eqs. (2) (4). 12) Accrding t the definitin f Eq. (1), it can be calculated that when Fe 2 O 3 decmpses t Fe 3 O 4 cmpletely, the equivalent reductin degree is 11.11%. When Fe 2 O 3 decmpses t FeO cmpletely, the equivalent reductin degree is 33.33%. Finally when Fe 2 O 3 decmpses t Fe cmpletely, the equivalent reductin degree is 100%. At a given temperature, the directin f a reactin can be judged by calculating the Gibbs free energy ΔG (J/(ml K)) as shwn in Eq. (5). If ΔG = 0, the reactins achieve an equilibrium state. The equilibrium state at each temperature has an equilibrium xygen partial pressure. In ther wrds, the start temperature f thermal decmpsitin is dependent n the partial pressure f xygen in the reactin system. The equilibrium partial pressure f xygen f the three irn xides can be calculated frm Eq. (6) which is derived by substituting Eq. (2) r (3) r (4) int Eq. (5). p 2 Δ G = ΔG + RgT ln... (5) p ln p G 2 = Δ... (6) p RT g Where, ΔG the standard Gibbs free energy (J/(ml K)) p the reference pressure in the system (Pa) p 2 the partial pressure f xygen (Pa) R g the universal gas cnstant (R g =8.314 J/(ml K)) Since the reactin furnace in this study is an pen system, the reference pressure (p ) f the system is abut 1 atm. Figure 2 was made accrding t Eq. (5) based n the reference pressure f 1 atm. It shws that the equilibrium partial pres ISIJ

3 p 2 sure f xygen ( ) as a functin f temperature fr hematite, magnetite and wüstite, increases with the increase f temperature. The calculatin results fr the case f hematite shw that when the p 2 in the system is 0 Pa, the reactin wuld take place at rm temperature. It was fund that at K, the thermal decmpsitin f hematite can take place nly if the p 2 is less than 82 Pa which is abut 0.08% f 1 atm. At K, the thermal decmpsitin f hematite can take place if the p 2 is less than Pa which is abut 1.1% f 1 atm. Figure 2 reveals that abve abut K, the equilibrium p 2 increases rapidly with the increase f temperature. It means that if the p 2 in the reactin system is cntrlled clse t be less than equilibrium partial pressure, the thermal decmpsitin f hematite will accelerate at abve K. It can als be calculated that the biling temperature f the thermal decmpsitin f hematite in an pen system is K, at which temperature hematite culd decmpse intensively with p 2 f 1 atm. The thermal decmpsitin f magnetite is much mre difficult than that f hematite. Althugh the thermal decmpsitin f magnetite can als take place when the p 2 is 0 Pa, accrding t the thermdynamic calculatin, the equilibrium p 2 nly has slight increase between K and K. Hwever, it has an bvius increase frm K t K. The biling temperature f the thermal decmpsitin f magnetite in an pen system is K. The decmpsitin f wüstite is the mst difficult prcess as shwn in Fig. 2. Belw the temperature f K, the equilibrium partial pressures are almst always kept at the value f zer. If the temperature is K, the equilibrium p 2 can reach 1 atm, which is much higher than hematite (1 783 K) and magnetite (2 186 K). 3. Experimental 3.1. Experimental Strategy The actual equivalent reductin degree f hematite re caused by thermal decmpsitin in the cyclne furnace is difficult t estimate because f the cmplicated envirnment. The labratry experimental study f the thermal decmpsitin f hematite re has been cnducted in three methds with three different reactrs, and 4 influential factrs (temperature, gas cmpsitin, particle size, and residence time) were investigated. Methd 1: Thermal decmpsitin f hematite re with a TGA-DSC analyser with an inert gas flw (small sample size: 80 mg; K; p 2 0 Pa); Aim: bserve the thermal decmpsitin at elevated temperature. Methd 2: Thermal decmpsitin f hematite re at different temperatures with an electrically heated hrizntal furnace with inert carrier gas (larger sample size: 10 grams; K, K, K; p 2 0 Pa); Aim: cnfirm the results with TGA-DSC analyser and btain the accurate equivalent reductin degree. Methd 3: Thermal decmpsitin f hematite re in different inert gas cmpsitins, with different particle sizes, and residence time in a HDTF (individual particles: ttal sample size f 3 grams; K, K, K, K, K, K; p 2 0 Pa); Aim: study the in-flight thermal decmpsitin behaviur and prvide mre direct infrmatin fr the HIsarna cyclne reactr Experimental Set-up and Cnditins Materials The raw material f the dry irn re used in the experiment was prvided by Tata Steel in IJmuiden, the Netherlands as shwn in Table 1. The hematite cntent is as high as 94.9%. The raw material was prepared by sieving t get different size grups f particles and then dried t remve the misture Experiments with the TGA-DSC The first stage f the experiments has been carried ut by using a thermgravimetric analyzer TGA-DSC (NETZSCH Thermal Analysis STA 409). The inert gas f N 2 (purity: %) was cntrlled by a mass flw meter t prtect the sample. The sample hlder is a cylindrical crucible which has the vlume f 85 μl. This study fcused n the thermal decmpsitin behaviur f hematite re at elevated temperature. The experimental cnditins are shwn in Table Experiments with the Hrizntal Tube Furnace The sample size in the TGA-DSC analyser is t small (~80 mg) t d chemical analysis r ther tests. The experiments were further carried ut in the hrizntal tube furnace at pre-selected isthermal temperatures as shwn in Table 1. Hematite re cmpsitin. Cmpsitin Mass% Cmpsitin Mass% Al 2O Mn CaO SiO Fe 2O Rest MgO Table 2. Experimental cnditin in the TGA-DSC analysis. Fig. 2. Equilibrium f xygen partial pressure as a functin f temperature fr hematite, magnetite and wüstite. Experimental cnditin Operating parameters Sample Hematite re: Fe 2O % Hlding time (h) 4 Heating rate (K/min) 10 Particle size (μm) Hematite re: 45 53, Temperature (K) ISIJ 2198

4 Fig. 3. A larger sample size culd be used t btain an accurate equivalent reductin degree. The experimental cnditins are listed in Table 3. The temperature at the isthermal zne was measured by a type S thermcuple. Frm this experiment, the maximum equivalent reductin degree f hematite re at a certain temperature and in a certain hlding time (keeping p 2 0) has been btained. Only the small particle size f hematite re was used in the experiment. When the hlding time is lng enugh, the particle size desn t play an imprtant rle in the final equivalent reductin degree. During the experiment, high purity f N 2 gas has been used as prtectin gas. After the experiment, the Fig. 3. Table 3. Experimental cnditin in the hrizntal furnace. Experimental cnditin Operating parameters Sample Hematite re: Fe 2O % Hlding time (h) 1, 2, 3 Heating rate (K/min) 10 Particle size (μm) Hematite re: Temperature (K) 1 673, 1 773, Schematic diagram f the electrically heated hrizntal tube furnace. sample was grinded int pwder fr the further analysis with chemical titratin Experiments with the HTDF In this study, the thermal decmpsitin behaviur f the hematite re has been further explred with the HDTF. The schematic drawing f the experimental set-up is shwn in Fig. 4. The labratry experimental set-up f the HDTF was nrmally used in the study f cal cmbustin behaviur ) The cre f the experimental set-up mainly cnsists f 6 facilities: an electrically heated tube furnace, a syringe pump particle feeder, a particle injectin prbe, a sampling prbe, a gas pre-heater and a sample cllectr. The injectin prbe and sampling prbe are made f high temperature resistance stainless steel. The alumina tube with inside diameter f 60 mm, utside diameter f 70 mm and length f 1100 mm was used as the reactr tube (vlume: 3.1 L). After the starting f the experiment, a cntinuus flw f inert gas frm the gas cylinder was regulated by a mass flw cntrller. The typical gas flw rate was 4 L/min. A small part f the gas was separated as carrier gas which was measured by a mass flw meter. The carrier gas flwed int the glass tube f the particle feeder and brught several particles tgether with it, and then flwed int the water cled injectin prbe. The remaining gas was preheated t 773 K, and then intrduced int the reactr directly frm the tp f the furnace. The thermal decmpsitin tk place in the ht zne f the furnace which starts frm the tip f the injectin prbe and ends at the tp f the sampling prbe. The reacted particles and gas were received by the water cled and gas quenched sampling prbe. Finally, the particles were cllected in the sample cllectr. The temperature distributin at the ht zne is shwn in Fig. 5 with three examples: K, K and K. The experimental cnditins are shwn in Table 4. The effects f gas, temperature, residence time, and particle size n the irn re thermal decmpsitin have been studied in Fig. 4. Schematic diagram f the High-temperature Drp Tube Furnace (HTDF) ISIJ

5 the HTDF. The inert gas (purity: %) was changed frm the symmetric mlecule gas N 2 t mnatmic gas Ar, and then t the asymmetric mlecule gas CO 2. The temperature was changed frm K t K. The residence time was varied frm 210 ms t ms, which was calculated by applying Newtn s secnd law f mtin and Stkes law t the particles mtin thrugh fluids. It was assumed that there were three frces acting n a single particle: gravity frce, buyancy frce and drag frce. The six particle sizes have been tested, which are in the verall range f μm. T prevent the thermal shck and extend the service life f the reactr tube and heating elements, the heating rate f the furnace was set t 3 K/min and the cling rate was set t 2 K/min. 4. Results and Discussin 4.1. Experiments in the TGA-DSC Firstly, a series f experiments have been carried ut with Table 4. Experimental cnditins in the HTDF. Experimental cnditin Operating parameters Sample Hematite re: Fe 2O % Particle feed rate (g/h) 1 Gas cmpsitin 100% N 2; 100% Ar; 100% CO 2 Particle size (μm) 38 45, 45 53, 53 75, 75 90, , Residence time (ms) 210, 970, Temperature (K) 1 550, 1 600, 1 650, 1 700, 1 750, Fig. 5. Temperature prfiles at the ht zne. hematite re with different particle sizes and heating rates. Figure 6 shws the TG and DSC prfiles f ne test. The weight f the sample was 80.2 mg and the particle size was in the range f μm. The highest temperature achieved was K and the heating rate was set at 10 K/min. In Fig. 6(a), the temperature histry in the furnace is presented by the red line. It was linearly increased t the maximum temperature and then held fr 4 hurs. There are three steps n the TG curve and tw trughs n the DSC curve befre the hlding time. Due t that the device was nt stable at the beginning f the test (0 30 min.), the TG value was a little mre than 100%. The ttal weight lss f the sample is abut 4.43%. A slight weight lss appears at the first stage f the heating time ( min.) due t the thermal decmpsitin f a small quantity f hematite. The thermal decmpsitin f hematite is endthermic reactin. It is the reasn why the DSC curve ges dwn belw zer. The slight weight lss at the beginning f the heating time ( min.) is fllwed by the sharp weight lss (at abut 130 min.). The ttal weight lss at the end f the stage f sharp weight lss is arund 3.55%. Firstly, it indicates that hematite is nt stable with the increase f temperature in the inert gas, which is in agreement with the theretical evaluatin perfectly. Hematite started t decmpse slwly with the increase f temperature in the first stage (heating time: min. and temperature: K), while the decmpsitin rate was accelerated when temperature was raised abve arund K and the sharp weight lss was btained. Secndly, the stage f sharp weight lss is crrespnding t the first big trugh n the DSC curve which means that the mass lss was caused by an endthermic reactin f the thermal decmpsitin f hematite. Further mre, the reactin enthalpy f thermal decmpsitin f 1 ml Fe 2O 3 is abut kj/ml (ΔH 298) at 298 K and it increases t kj/ml (ΔH 1 473) at K. The reactin heat nly has a small increase with the increase f temperature. Therefre, the big trugh n the DSC curve at the sharp weight lss stage is caused by the high decmpsitin rate. Thirdly, it can be calculated that if Fe 2O 3 decmpses t Fe 3O 4 cmpletely, the weight lss is abut 3.3%. It means that the hematite was prbably cmpletely transferred t magnetite at the end f the sharp weight lss. The weight lss f 0.25% (3.55% 3.3%) is prbably caused by the thermal decmpsitin f magnetite. The secnd trugh shws that the heat absrptin f the reactin during this Fig. 6. TG-DSC analysis f hematite re (Heating rate: 10 K/min) (a) as a functin f time, (b) as a functin f temperature ISIJ 2200

6 perid is less than the heat absrptin f the reactin which results in the first trugh. The weight lss during the secnd trugh is als nt bvius. By the end f the secnd trugh, the weight lss is abut 3.74% (3.55% %). Fr that the TGA-DSC analyser is an pen system, the small amunt f generated O 2 culd be taken away by inert gas cntinuusly. The xygen partial pressure in the reactin system can be kept at a very lw level. Therefre, the secnd trugh wuld be caused by the decmpsitin f magnetite r the evapratin f the ther impurities in the irn re sample. The ttal weight lss during the rest time is nly 0.69% (4.43% 3.55% 0.19%), althugh it has a 4 hurs hlding time. The result shws that the weight lss decreases slwly after 200 minutes. If all the weight lss is caused by thermal decmpsitin f irn xides, the equivalent reductin degree f hematite re can be rughly estimated t be 14.8%. Frm Fig. 6(b), it can be clearly seen that the sharp weight lss stage is exactly marked by the big trugh n the DSC curve. Accrding t the analysis f the DSC curve, the nset and end temperatures f the first trugh are K and K, respectively. The pint f K can be seen as the start temperature f thermal decmpsitin f the hematite re at the sharp weight lss stage. In ther wrds, the thermal decmpsitin rate f hematite is accelerated abve K. The nset and end temperatures f the secnd trugh are K and K. The start temperature f intensive thermal decmpsitin f hematite btained frm experiments is in agreement with the range f the theretical evaluatin. The same cnclusin can be btained that thermal decmpsitin f magnetite is much mre difficult than the hematite. The larger particle sized hematite re was als tested with the TGA-DSC analyser as shwn in Fig. 7. The particle size is the grup f μm. The heating rate is 10 K/min. The results cnfirmed that the thermal decmpsitin f hematite has a sharp weight lss stage during the heating time. The weight lss f the sharp weight lss stage is abut 3.4% and the ttal weight lss is abut 4.6% in the 4 h hlding time. In Fig. 7(b), the start temperature f the sharp weight lss stage f hematite is abut K and the start temperature f the secnd trugh n the DSC curve is abut K. The tw temperatures are slightly higher than the results in Fig. 6(b). It prbably because that the larger particles were heated up mre slwly than the small particles in the furnace Experiments in the Hrizntal Furnace The experiments in the hrizntal furnace differ frm the TGA-DSC analysis by the amunt f the sample and the way f heating (temperature prfile). TGA-DSC analysis culd give the descriptin f the nging prcess f thermal decmpsitin, which shws the intensive reactin temperature, exthermic reactin r endthermic reactin, and the weight lss at each mment. Hwever, the mre accurate equivalent reductin degree caused by thermal decmpsitin f hematite re culd nt be analysed by chemical analysis, while the experimental study in the hrizntal furnace culd slve this prblem. The larger sample size has been tested in the hrizntal furnace. Chemical titratin was first carried ut with a certified reference material JK 29 cnsisting f 90.11% magnetite and 7.1% hematite, the cmpnents f which have been analyzed by a Swedish institute. Thereafter the samples in this experiment were analyzed with the accuracy f ± 0.2%. Mre imprtantly the thermal decmpsitin experiments in the hrizntal tube furnace were cnducted at a cnstant temperature, and the time t reach the reactin temperature is very shrt (TGA-DSC tests have a cnstant heating rate f 10 K/min.). The temperature f the sample befre pushing t the ht zne f the reactr tube is rm temperature. It is similar t the actual situatin. Hematite re was suddenly psitined at the isthermal zne f the hrizntal furnace. Accrding t the thermdynamic thery, fr an pen system, the released O 2 is cnstantly remved by the flwing inert gas and maintain the real partial pressure f xygen lwer than the equilibrium value in the reactr, s the thermal decmpsitin reactins wuld nt stp as lng as the samples are kept in the isthermal zne. Hwever, accrding t the results f TGA-DSC analysis, the reactin rate becmes very lw after tw hurs hlding time and the sharp weight lss stage was bserved during the heating time (befre the hlding time). In rder t btain the sharp weight lss stage in the hrizntal furnace, different hlding time f 1 hur, 2 hurs and 3 hurs at K were tested t make sure that the ttal experimental time is lng enugh. The results shwn in Fig. 8 indicate clearly that the sharp weight lss already takes place in the experiment with 1 hur hlding time, and the equivalent reductin degree (R) f hematite re after 2 hurs hlding time increases slwly. Therefre, the hlding time was fixed t 2 hurs fr the ther experiments. The effect f temperature n the equivalent reductin Fig. 7. TGA-DSC analysis f hematite re (Particle size: μm) (a) as a functin f time, (b) as a functin f temperature ISIJ

7 degree is shwn in Table 5. In the table, T Fe% dentes weight percentage f ttal irn in the sample, Fe 2+ %, and Fe 3+ % are the weight percentages f Fe 2+ and Fe 3+ in the sample, respectively. Accrding t the cmpsitin f the raw material, the elemental Fe is cmpletely in the frm f Fe 2O 3 in the raw material. Therefre the riginal xygen percentage in the sample can be calculated with the weight percentage f the ttal irn in the sample. On the ther hand, the xygen weight lss by decmpsitin can be calculated with the weight percentage f Fe 2+ in the sample. Finally, the equivalent reductin degree can be evaluated by Eq. (1). The rugh prduct cmpsitin is calculated by assuming that FeO is generated until the transfrmatin frm Fe 2O 3 t Fe 3O 4 is cmpleted. Accrding t the assumptin and theretical evaluatin, the partially decmpsed hematite re is mainly cmpsed f magnetite and wüstite and the percentage f wüstite ges up with increasing temperature. Generally, the equivalent reductin degree increases with the increase f temperature. The equivalent reductin degree at K is slightly higher than that at K by 0.87%. Hwever, it increases strngly frm the temperature f K t K. At K, the decmpsitin reactin f Fe 2O 3 takes place intensively resulting in 16.4% equivalent reductin degree. It indicates that the thermal decmpsitin f magnetite becmes faster abve K. It cnfirms the results f theretical evaluatin as shwn in Fig. 2. The hematite re at K were melted dwn cmpletely, while the samples were still in slid state at K and K Experiments in the HTDF With TGA-DSC and hrizntal tube furnace, a sharp weight lss stage has been detected during the thermal decmpsitin f hematite. At the end f the sharp weight lss stage, hematite decmpsed t magnetite cmpletely at the experimental temperature abut K f smelting cyclne. It wuld play an imprtant rle in the ttal prereductin degree in the smelting cyclne. Hwever, the residence time f the irn re particles in the cyclne reactr is much shrter than the time investigated in the TGA-DSC and hrizntal tube furnace. Whether the sharp weight lss stage is present in the smelting cyclne and hw fast the thermal decmpsitin will prceed during the sharp weight lss stage were nt clear and culd nt be deduced frm the abve results. Therefre, the thermal decmpsitin behaviur f individual particle was further studied with the HDTF, in which the hematite re particle was injected int the reactr by the injectin prbe and reactin tk place during the flying time in the reactr. The reactin gas in the cyclne reactr mainly cnsists f CO, CO 2, H 2, H 2O and N 2. Thrugh analyzing the ff-gas cmpsitin frm the cyclne reactr, it was fund that CO 2 is ver 60% and N 2 is arund 5 10%. In this study, bth CO 2 and N 2 were used as inert gas t study the thermal decmpsitin behaviur f hematite re. In rder t verify the effect f the structure f the gas mlecular n the equivalent reductin degree, mnatmic gas f Ar was als used. The equivalent reductin degree f the cllected samples was als calculated based n the analysis by chemical titratin. Fig. 8. Irn re equivalent reductin degree in different hlding time at T = K Effect f Temperature and the Type f Gas Figure 9 shws the effect f temperature and the type f gas n the thermal decmpsitin f irn re particles. The experiments f thermal decmpsitin f hematite re in CO 2 gas were carried ut at K, K, K, K and K. The particle size was in the range f μm. The particle residence time in the ht zne was maintained at ms. In rder t study the influence f different inert gas n the equivalent reductin degree f irn re, the experiments have als been carried at three selected temperature: K, K and K in N 2 gas and Ar gas. The results shw that the equivalent reductin degree f hematite re increases with the increase f temperature fr all inert gases used. But the difference f thermal decmpsitin between the temperature f K and K is quite small. At the three temperatures f K, K and K, the equivalent reductin degrees f hematite re particles in CO 2 gas (slid circle) are much higher than Table 5. Equivalent reductin degree f hematite re in hrizntal furnace. T (K) t (h) Chemical titratin result (wt%) Cmpsitin (wt%)# Physical R (%) T state Fe Fe 2+ Fe 3+ Fe 2O 3 Fe 3O 4 FeO Slid Slid Liquid # calculated based n titratin analysis ISIJ 2202

8 Fig. 9. Thermal decmpsitin f hematite re in Ar, CO 2 and N 2 at different temperatures. Fig. 10. Cmparisn f thermal decmpsitin f hematite re in the hrizntal furnace and the HDTF, residence time: ms in HDTF, 2 hurs in hrizntal tube furnace. that in N 2 gas (slid triangle). The difference f the equivalent reductin degrees in the tw gases (CO 2 and N 2 ) is abut 3 4%. By estimatin f the cmpsitin f the irn re sample, it shws that the remaining hematite cntent in the sample with N 2 gas is much higher than that in the sample with CO 2 gas. Fr example, at K, the Fe 2 O 3 in the prduct which was decmpsed in N 2 gas was abut 31.2%, while the Fe 2 O 3 in the prduct which was decmpsed in CO 2 gas was nly abut 2.3%. The experimental results in Ar gas (slid square) as shwn in the figure are almst the same with the results in N 2 gas. It is because CO 2 is an asymmetric diatmic mlecule, while N 2 is a symmetric mlecule and Ar is a mnatmic mlecule. The structure f mlecules is imprtant fr mlecular radiatin - emissin and absrptin which ccurs nly when an atm makes a transitin frm ne state with a certain amunt f energy t a state with lwer (higher) energy, respectively. Hmnuclear diatmic mlecules like N 2 and O 2 and mnatmic mlecules like Ar d nt have radiative capability fr its symmetrical distributin f charges. Cmpared t N 2 and Ar, CO 2 has strng thermal radiatin capacity. That is als why CO 2 gas is the mst imprtant greenhuse gas. Therefre a fine particle is heated up mre quickly in CO 2 gas thrugh thermal radiatin than in N 2 and Ar gases. The experiments carried ut at K, K and K in the hrizntal furnace were selected t cmpare with the experiments carried ut at K, K, K and K in the HTDF as shwn in Fig. 10. It was fund that the sharp weight lss stage still exists in the HTDF during the particle flying time thrugh the ht zne. Hwever, the sharp weight lss stage was nt cmpleted at K. The equivalent reductin degree f irn re in CO 2 gas is 7.5% and in N 2 gas is 3.5% which are far belw the result f 11.7% in the hrizntal tube furnace. At K, K and K, the equivalent reductin degrees f 10.8%, 10.9% and 11.0% f irn re in the HTDF with CO 2 are quite clse t the result f 12.6% at K in the hrizntal furnace. It indicates that almst the whle sharp weight lss stage bserved in TGA-DSC tests was achieved at the temperature frm K t K in the HTDF, althugh the residence time f irn re particles were much shrter than the hlding time in the hrizntal furnace. This may be caused by tw reasns. Firstly, the hematite re in HTDF is mving individual particles, having much better and mre favrite kinetic cnditins than in the hrizntal tube furnace where they are packed and steady with lwer relative mving velcity frm the gas. Secndly, the thermal decmpsitin rate f irn re at the sharp weight lss stage is extremely fast Effect f Residence Time Hw fast the thermal decmpsitin rate f irn re is at its sharp weight lss stage culd nt be estimated frm the results with TGA-DSC analysis and the hrizntal furnace. Althugh TGA-DSC analysis prvided the nline reprt f weight lss as a functin f time, the heating rate (10 K/min.) in the TGA-DSC analyzer was different frm the heating rate in the smelting cyclne. Therefre, further experiments were carried ut in the HTDF with different residence time. The results f chemical analysis f the sample are shwn in Table 6. Bth N 2 gas and CO 2 gas were separately used. The residence time f the flying irn re particle was adjusted by varying the gas flw rate. In the first part, the temperature in the ht zne was set t K and K with the inert gas f N 2. The particle size was in the range f μm. Tw residence time f ms and 970 ms was tested. It was fund that the decmpsitin degree f hematite re in the N 2 gas ges up with the increase f residence time. The equivalent reductin degree in ms (e.g. 5% at K) is abut 1 1.5% higher than that in 970 ms (e.g. 3.5% at K) in N 2 gas. In the smelting cyclne the bulk gas is mainly cmpsed f CO 2 gas which has a strng thermal radiatin capacity. Therefre, mre experiments have been carried ut in CO 2 gas. It was fund that the equivalent reductin degree did nt change by changing the residence time. Fr example, at K, the residence time f irn re particles was verified frm 210 ms t ms, but the equivalent reductin degree f irn re was all arund 10.8% which is very clse t 11% (the cmplete decmpsitin f Fe 2 O 3 ). Therefre, the residence time in the HTDF has influence n the equivalent reductin degree f irn re with N 2, but ISIJ

9 Table 6. Thermal decmpsitin f hematite re with different residence time in HTDF (particle size: μm). Gas T (K) t (ms) N Chemical titratin result (wt%) Cmpsitin (wt%)# R (%) T Fe Fe 2+ Fe 3+ Fe 2O 3 Fe 3O 4 t du (h) m sample (g) ~ ~ ~ ~ CO ~ ~ ~ *m sample is weight f cllected samples; t du means duratin f each experiment at the reactin temperature. # calculated based n titratin analysis. Table 7. Thermal decmpsitin f hematite re in different particle size. Gas d p (μm) t (ms) Chemical titratin result (wt%) Cmpsitin (wt%)# R (%) T Fe Fe 2+ Fe 3+ Fe 2O 3 Fe 3O 4 FeO t du (h) m sample (g) ~ CO ~ ~ ~ *m sample is weight f cllected samples; t du means duratin f each experiment at the reactin temperature. # calculated based n titratin analysis. has n influence n the equivalent reductin degree f irn re with CO 2. It is mainly caused by the gas radiatin prperties, which makes the CO 2 gas very effective fr heat transfer t the particles. In ther wrds, the particle heating rate in N 2 gas wuld be the rate cntrlling step f the re thermal decmpsitin. In the CO 2 gas, the studied range f residence time did nt affect the equivalent reductin degree, and what s mre the sharp weight lss stage culd be partly achieved in 210 ms and n further change f equivalent reductin degree culd be detected during the residence time between 210 ms t ms. Fig. 11. XRD pattern f reacted irn re sample at K in CO Effect f Particle Size The experiments have been cnducted with different particle sizes f the hematite re in CO 2 gas. The average temperature f the ht zne was cntrlled t K. The experiments were divided int tw grups: small particle and large particle. The residence time f the smaller particles was 970 ms and that f the larger particles was 210 ms. The particle sizes in the first (small) grup were in the range f μm and μm and in the secnd (large) grup were in the range f μm and μm. The results are listed in Table 7. It can be seen that in all the studied particle size grups f μm, the equivalent reductin degree f irn re is in the range f %. The small deviatin can be neglected. Therefre, the particle size belw 250 μm des nt have significant influence n the equivalent reductin degree f the hematite re in the HDTF when the inert gas is CO 2. On the ther hand, fr the difference f particle gravity, the residence time f the larger particles was shrter than the smaller particles. The results prved again that with CO 2 gas, residence time is nt a significant factr fr the thermal decmpsitin behaviur f fine irn re particles in the HTDF. The sample f μm in Table 7 was analysed by XRD t identify the phases in the prduct as shwn in Fig. 11. The figure gives the peak psitins and intensities f the identified phases. This cnfirms fully the analytical results f chemical titratin. The main phase in the reacted irn re sample is magnetite and the phases f hematite and wüstite are either nt present (wüstite) r t small t be detected (hematite). 5. Cnclusins In this paper, the thermal decmpsitin behaviur f hematite re has been studied thrugh theretical and experimental methds. The experimental study has been carried 2014 ISIJ 2204

10 ut with the TGA-DAC analyzer, the hrizntal tube furnace and the drp tube furnace (HTDF). The equivalent reductin degree f cllected samples frm hrizntal furnace and HTDF were determined by chemical titratin. The results were helpful fr the further study f the kinetics f the melting and reductin mechanism f irn re in the next stage. The cnclusins are as fllws: Frm the theretical calculatin, Fe 2O 3 is the mst unstable irn xide amng the three irn xides, while FeO is the mst stable ne. In an pen system with flwing inert gas (the released O 2 is remved cntinuusly), the intensive thermal decmpsitin f Fe 2O 3 is likely t take place abve K and the thermal decmpsitin f Fe 3O 4 is pssible t speed up abve K. Frm TGA-DSC analysis, it is fund that a sharp weight lss stage appeared n the TG curve in the range f temperature f K. A big trugh was fund in the DSC curve at the same time. It was caused by the intensive decmpsitin f hematite. The result is in accrdance with the theretical estimatin. Frm the experimental results in the hrizntal furnace, the accurate equivalent reductin degree f hematite at a given temperature and hlding time was btained. The sharp weight lss stage bserved in TGA-DSC test was achieved in all the studied temperatures in the hrizntal tube furnace in 2 hurs. Generally, the equivalent reductin degree f hematite increases with the increase f temperature. N significant difference was bserved between the results at K and K. Hwever, the equivalent reductin degree f hematite at K was much higher than that at K. It cnfirms that the thermal decmpsitin f magnetite culd be accelerated abve K. Frm the study f in-flight thermal decmpsitin within HTDF, it was fund that sharp weight lss stage which was bserved in the TGA-DSC experiments culd be partly achieved in the HTDF at high temperatures especially in the CO 2 gas. The fine irn re particles culd be heated up faster in CO 2 gas than in N 2 and Ar gas due t the strng radiatin prperties (emissin and absrptin) f CO 2 gas. N significant influence f particle size and residence time n the equivalent reductin degree culd be bserved in the HTDF, when the particle diameter is smaller than 250 μm in the CO 2 gas. At K in CO 2 gas, the equivalent reductin degree f irn re in HTDF is arund 10.8% which is slightly lwer than the value 12.6% btained in hrizntal furnace (hlding time 2 hurs). Acknwledgments This research was carried ut at Delft University f Technlgy and was financially supprted by the Materials innvatin institute M2i ( under the prject M The authrs wuld like t express their thanks t Mr. Jan van der Stel and Mr. Jeren Link frm Tata Steel Eurpe (IJmuiden) fr fruitful discussins and prviding prcess data fr this study. The first authr acknwledges the China Schlarship Cuncil (CSC) fr prviding the schlarship during this research at Delft University f technlgy. REFERENCES 1) I. O. Lee, M. K. Shin, M. Ch, H. G. Kim and H. G. Lee: ISIJ Int., 42 (2002), S33. 2) A. Orth, N. Anastasijevic and H. Eichberger: Miner. Eng., 20 (2007), ) W. Peter and B. Christpher: A Sectral Apprach, Agreement and Mechanism (SAAM) fr the Mitigatin f Greenhuse Gas Emissins in Japan s Irn and Steel Industry, Climate Strategies 2011, UK, (2011), 13. 4) C. Feilmayr, A. Thurnhfer, F. Winter, H. Mali and J. Schenk: ISIJ Int., 44 (2004), ) K. Pitrwski, K. Mndalb, T. Wiltwski, P. Dydd and G. Rizeg: Chem. Eng. J., 137 (2007), 73. 6) F. Tsukihashi, K. Kat and T. Sma: ISIJ Int., 22 (1982), ) J. Link: Rev. Métall., 106 (2009), ) K. Meijer, C. Guenther and R. J. Dry: METEC Cnf., Verlag Stahleisen, Düsseldrf, (2011), 1. 9) H. L. Gilles and C. W. Clump: Ind. Eng. Chem. Prc. DD., 9 (1970), ) Y. Nakamura, M. It and H. Ishikawa: Plasma Chem. Plasma Prcess., 1 (1981), ) A. Habermann and F. W. H. Hfbauer: ISIJ Int., 40 (2000), ) Y. K. Ra: Stichimetry and Thermdynamics f Metallurgical Prcesses, Cambridge Univ. Press, Cambridge, (1985), ) R. M. Charles and J. G. Geffrey: Energ. Fuel., 7 (1993), ) B. Richelieu, C. Michael and L. Edward: Fuel, 82 (2003), ) M. Sait, M. Sadakata, M. Sat and T. Sutme: Cmbust. Flame, 87 (1991), ISIJ