Transformation of Alumina Inclusions by Calcium Treatment

Transformation of Alumina Inclusions by Calcium Treatment MINNA LIND and LAURI HOLAPPAThe objectives of this study were to investigate reactions of calcium with Al2O3by differentmodel experiments both on the laboratory and on the industrial scale.Experiments with solidAl2O3and CaO were performed between1350°C and1600°C.Reaction rate constants weredetermined based on scanning electron microscopy/energy dispersive spectroscopy(SEM/EDS)observations of reaction products and weight measurements of the Al2O3reacted via dissolutionof the CaO bearing phases from the specimens after the annealing period.The results showedthat the formation of calcium aluminate phases proceeded rapidly at temperatures greater than1405°C when a liquid calcium aluminate was formed.In the lowest temperature range(1350°C–1405°C),when the formation of liquid phase ceased,the reaction rate was severalorders of magnitude lower.Industrial trials including Ca-alloy injection into steel,sampling andSEM/EDS analyses,as well as an inclusion rating in the samples show the concept of rapidtransformation of the alumina inclusions with Ca treatment.DOI:10.1007/s11663-009-9337-9ÓThe Minerals,Metals&Materials Society and ASM International2010I.INTRODUCTIOND URING the last30years,calcium treatment has become an established method for the desulfurization of aluminum-killed steels and to control the composition, morphology,and distribution of the remaining inclu-sions.A major benefit of this practice is that nozzle clogging in continuous casting can be eliminated;solid alumina inclusions are transformed to liquid calcium aluminates that do not clog the nozzle.Other benefits are realized in the mechanical properties of various steel grades.For example,the ductility and toughness of high-strength,low-alloy steel and high-quality structural steel are improved as the volume fractions of sulfides and of oxides are decreased.Inflat-rolled plate and sheet grades,manganese sulfide inclusions form string-ers or platelets when deformed,which results in poor through-thickness ductility and toughness.Calcium complexes the manganese sulfides into inclusions,which do not deform to elongated stringers.In free-machining grades,hard alumina inclusions cause excessive tool wear;calcium can transform these inclusions into softer calcium aluminates.[1]Liquid calcium aluminate inclu-sions act as cores for precipitating the sulfide shell.This concept was proposed based on experimental observa-tions in samples from industrial-scale heats.[2] Although the calcium treatment technology has been well established since the1980s,the basic mechanisms and kinetics of the calcium reactions with liquid steel are still poorly known.Clarification of these aspects was chosen as the objective of this work.Different types of ‘‘model experiments’’were designed and carried out.First,the reaction and dissolution rates between pure CaO and Al2O3were investigated at and below steel-making parable experiments have been reported by Kohatsu and Brindley[3]using sintered pellets of CaO and Al2O3.For1330°C and heating times of up to20days,they found all known com-pounds of the system CaO-Al2O3:C3A/C12A7/CA/CA2/ CA6.In these forms,C stands for CaO and A stands for Al2O3,respectively.Quantitatively,C12A7was the major phase in the reaction layer.The products C3A,C12A7, and CA had random orientations,whereas CA2and CA6had clear crystal orientation parallel to the reaction interface.More experiments with Pt markers indicated that the reaction occurred via Ca-ion diffusion into the Al2O3.[3]The observations are in accordance with the CaO-Al2O3phase diagram(Figure1[4,5]),which has (except for two pure oxide phases)four nearly stoichi-ometric compounds that can be called calcium alumi-nates.Jerebtsov et al.[6]studied the CaO-Al2O3system by using different analysis methods in the content interval from0to60mas pct CaO.They confirmed the absence of congruently melting compounds in an argon atmosphere and found an absence of12CaOÆ7Al2O3(C12A7)in a dry atmosphere.C12A7formed with less than1pct mass of the water absorbed in the solid. The existence of this compound in the presence of water traces was confirmed in their work by the appearance of C12A7,which contained eutectic peaks at1403±5°C on thefirst melting(synthesis)of the sample.CaO gained trace water during grinding in the air of the oxide mixture,but it was removed after melting.[6]In other experiments performed in this work,reac-tions and transformation rates of alumina inclusions to calcium aluminates were directly observed when a Ca-bearing addition was fed into the steel melt. Lu et al.[7]applied a similar principle to the one in this work to study kinetic phenomena in the calcium treatment of steel.They fed steel-covered wire(with calcium inside)into40kg heats of American Iron andMINNA LIND,Postdoctoral Researcher,and LAURI HOLAPPA, Emeritum Professor,are with the Laboratory of Metallurgy,Helsinki University of Technology,P.O.Box1000,FI-02015TKK,Otakaari1, FI-02150,Espoo,Finland.Contact e-mail:mlind@ Manuscript submitted July3,2009.Article published online January8,2010.Steel Institute1045steel with various sulfur and aluminum contents under an argon atmosphere.Sam-ples were taken frequently during and after the calcium injection for the chemical and the inclusion analyses. The authors assumed calcium vapor reacted and formed calcium oxide and sulfide in the melt.Calcium also dissolved into the melt and reacted with existing alumina inclusions to form calcium aluminates.The total Ca content reached a maximum after certain time period and then decreased because of the removal of oxide and sulfide inclusions as well as because of the vaporization of Ca.[7]Calcium can prevent nozzle blockage in aluminum-deoxidized steels.It has been suggested that the nozzle flow properties of the aluminum deoxidized steels did not improve until the ratio of pct calcium/pct aluminum in the tundish exceeded0.115.[8]In inclusions,the CaS content differed greatly because of the amount of added calcium as well as the time after it was added.The highest CaS content in the inclusions was only3minutes after the CaSi addition and with the calcium content in the steel analysis greater than 20ppm.Calcium dissolved in the melt and rapidly reacted with alumina inclusions to form calcium alumi-nates or calcium sulfide.The CaO and CaS contents in the inclusions decreased gradually as the dissolved calcium vaporized and decreased with the time.[9]A similar effect was calculated for when the oxygen content in steel was20ppm,and the CaS formed when the calcium content was20ppm(aluminum100ppm and sulfur250).With the same aluminum and sulfur contents,the CaS saturation occurred already when the calcium was around10ppm and the oxygen level was 10ppm.With an oxygen content of40ppm,it is only possible to form CaS with a calcium content greater than40ppm.[10]II.EXPERIMENTALThe reactions between Al2O3and CaO between 1350°C and1600°C were investigated in a resistance furnace.The furnace was heated by platinum-rhodium alloy windings.The temperature was controlled by a B-type thermocouple.The experiments were carried out in an air atmosphere.[11,12]CaO powder was prepared by calcinating CaCO3 powder at1050°C for2hours and afterwards pressing the powder into a cylinder shape with a diameter of 5.2mm.Alumina plates density were3790kg/m3, 3mm high,and had an Al2O3content greater than 99.5pct.The alumina plates were cut and ground into small cylinders with a similar diameter to the CaO samples.[11,12]When the furnace reached the desired temperature,a CaO cylinder(about0.140g)was put on an Al2O3 cylinder(about0.280g)and pushed into the high-temperature zone within2minutes.After annealing for a certain length of time,the sample was cooled by pulling it into the cooler end of the furnace.In low temperature experiments below1420°C,the CaO cyl-inder was pressed between two Al2O3cylinders with the aid of an alumina tube.The cylinders were lying on a platinum plate in the furnace.Two identical tests were made—one for scanning electron microscopy(SEM) analysis and another for the dissolution test.[11,12]Thefirst samples were cut and ground perpendicular to the original interface with silicon carbide polishing paper.Ethanol was used as a grindingfluid to avoid the dissolution of calcium aluminate inclusions.The pol-ished interface was observed by an optical microscope and an SEM to investigate the reaction mechanisms. For SEM/energy dispersive spectroscopy(EDS)analy-sis,the specimen was coated with carbon,which allowed for the composition of the reaction products(i.e., calcium aluminates)to be analyzed.The second set of samples were immersed in a10-wt pct HCl solution for 8hours.The un-reacted CaO as well as the calcium aluminates formed by the reaction between Al2O3and CaO all were dissolved into the acid solution,and only the unreacted alumina remained left.The mass of the alumina plate did not change after it was kept in the HCl solution for48hours.Therefore,based on the weight of the remaining residue,the mass of Al2O3(which had reacted and dissolved)was calculated.By combining the results of the dissolution experiments with the SEM observations,the reaction-rate constants were determined.[11,12]Experiments using the c-alumina powder and the similar CaO powder as in the previous experiments were performed.CaO and Al2O3powders of0.566g and 0.709g,respectively,were mixed and pressed into a cylinder with11.5mm in diameter.The specimen then was kept in a laboratory furnace in an air atmosphere for6hours at1400°C.Then,the specimen was cooled and polished for investigation of the structure.[12] Industrial samples were taken at Ovako Bar Imatra Steelworks after the Ca-treatment.The heat size was 77tons.Calcium was added via CaSi powder injection, which takes typically5minutes.Analysis of the steel just before casting was0.16pct C,0.23pct Si,1.22pct Mn, 0.03pct S,0.04pct Al,0.0021pct Ca,and0.0022pct O. Samples were taken with the standard lollypop-shaped sampler where the steel was sucked through a quartz tube into theflat oval-shaped steel container.Specimens Fig.1—CaO-Al2O3phase diagram.[4]were prepared by grinding and polishing and then were coated with carbon to get a good image without electrical charging.[12]Inclusions were analyzed with the SEM/EDS pro-gram,which automatically collected chemical and mor-phological data from the inclusions in steel.III.RESULTS AND DISCUSSIONA.Reaction Kinetics Between CaO and Al 2O 3The experimental results are discussed in the follow-ing two parts:a high-temperature range,1420°C–1600°C,in which the formation of liquid oxide was favored according to the CaO-Al 2O 3diagram,and a low-temperature range,below 1420°C,in which the reaction between the two oxides occurred in the solid state.The results of the high-temperature range are discussed first.[11]1.High-Temperature Range,1420–1600°CThe phases formed during the reaction between Al 2O 3and CaO were examined by SEM/EDS.The line-scan results of Ca and Al on the cross-section of a typical sample are shown in Figure 2.The sample was annealed at 1500°C for 10minutes.After which,a reaction layer of 1250l m was measured.Most of the layer seemed to have been liquid at the end of the trial.On either side of this apparently liquid layer,thin assumedly solid layers containing both aluminum and calcium oxides could be observed.The layer thicknesses calculated from the dissolution results were,in general,somewhat higher than those measured from the SEM image because the reaction layer was not necessarily a flat,uniform interface.Temperatures higher than 1500°C,the reac-tion was still faster,and at 1600°C,the CaO cylinderwas completely converted into calcium aluminate after 3minutes at temperature.[11]When considering the reaction between CaO and Al 2O 3,one of the following two steps could be rate-controlling during the interaction between the liquid calcium aluminate and the solid alumina:(1)diffusion of calcium and/or aluminum oxide in the liquid calcium aluminate layer,or (2)the chemical reaction at the alumina–calcium aluminate interface.If the rate was controlled by the diffusion in the liquid product layer,then significant concentration gradients should be in the liquid layer.However,calcium concentration in the liquid layer was homogenous.Numerous reaction layers were analyzed,but no clear gradients were detected.[11,12]Therefore,the chemical reaction (or dissolution)between the alumina and the calcium aluminate was assumed to be the rate-controlling step.However,by examining the cross section of the reacted samples,clear layers of CA 6,CA 2,or CA phases were not detected in the reaction product layers.These solid phases,if formed,were about 2l m thick.2.Low-Temperature Range Below 1420°CAs shown in Figure 1,the liquid composition region in the CaO-Al 2O 3phase diagram became narrow when approaching 1400°C,and at 1362°C no liquid can form.Observations of reaction kinetics showed an abrupt slow-down from 1390°C to 1370°C (Figure 3).At low temperatures (1370°C and below),extending the holding time did not greatly affect the thickness of the reaction layer,which must be because of the reduced content of the liquid calcium aluminate phase when the temperature decreased.In Figure 3,we can see that at the temperatures 1390(30minutes)and 1405°C (12minutes),the reacted amount (~1.5mg)of Al 2O 3was about the same,but the holding time was different.At 1405°C (12minutes)and 1420°C (10minutes),the holding time was about the same,but the reacted amount was different—1.55mg and 12mg,respectively.Line-scan analyses from the experiments at tempera-tures 1405°C for 12minutes is shown in Figure 4.Analyses were made with SEM/EDS to give the CaO and Al 2O 3contents at different points across the reaction layer.In this lower temperature range,differentFig.2—SEM (SE-image)showing a cross-sectional view of the inter-faces and the line scan results of Ca and Al concentrations in the reaction layer by EDS.Sample annealed at 1500°C for 10min.[11]Fig.3—Reacted amount of Al 2O 3as a function of time at five differ-ent temperatures.calcium aluminate phases could be determined from analysis of the EDS.From the forming phases,C12A7 was the dominant phase in the reaction layer.The next most common phase was C3A.Phases CA2,CA,and CA6could be detected clearly only in some samples less than3l m layers.3.Reactions Between CaO and Al2O3Powders Figure5shows results of the SEM/EDS analysis.The dark particle in the back scattered electron(BE)-image is an alumina grain that had reacted with CaO.The line scan shows the phase CA on the x-axis0-8l m,phase CA2on the x-axis8-13l m,and pure alumina on the x-axis13-22l m.Phases were verified by point-EDS analysis.The same samples also were analyzed by X-ray diffraction(XRD)to determine the relative amounts of the different phases present.The same calcium aluminate phases could be detected as with SEM/EDS analysis.The lowest melting phase C12A7was the major phase(41pct)followed by CA(36pct),CA2(11pct), CaO(2pct),Al2O3(9pct)and CaCO3(1pct).B.Industrial Experiments at SteelworksFigures7to10show the compositions and sizes of the 25biggest inclusions from different samples.The ana-lyzed area was2mm92mm,which then gave infor-mation on more than100inclusions.Data on the smallest inclusions,approximately below1.5l m,were not included in these bar-chartfigures.In thefigures,the inclusion analysis provided the oxide inclusions complex charter with the sulfur phase.The weight percentages of the inclusions are shown so that the eventual matrix(Fe) has been excluded from the analysis results.The fraction of the matrix(Fe)could be high in small inclusions.In bigger inclusions(>5l m),the matrix(Fe)had little or100µmCaO C3A C12A7 Calcium Ka1Oxygen Ka1Aluminum Ka1CA Al2O3k1 0.5050100µmFig.4—Reaction between CaO and Al2O3powders heated at 1405°C for12min.The upper picture shows a BE-image from the reaction layer.The lower picture shows the line scan analysis in detail.µm 032110CA CA2Al2O320Calcium Ka1Oxygen Ka1Aluminum Ka160µmFig.5—Reaction between CaO and Al2O3powders heated at 1400°C for6h.The upper picture shows the analysis line.The lower picture shows the line scan analysis in detail.no influence on the EDS analysis.Oxygen analysis also was seen in the bar-chart for Figures 7to 10;even EDS oxygen analysis did not have good certainty.Figure 6shows an element map of a heterogeneous inclusion with a calcium aluminate phase and a calcium sulfide phase.This inclusion was found in the same sample 6(27minutes,30seconds)but outside theanalysis area,which is presented in Figures 10and 11(sample 6).In Figure 7,the sample was taken only 55seconds after the CaSi injection.The Ca content was analyzed to be 37ppm.In this sample,the biggest 25inclusions analyses showed that more than half of the inclusions contained calcium in the form of either calcium alumi-nate or calcium sulfide.The rest of the inclusions showed the presence of aluminum oxides and manga-nese sulfides.This finding was clear evidence of a rapid calcium reaction and transformation of inclusions caused by the Ca injection.A few of the inclusions also contained some magnesium oxide.In Figure 8(2minutes,50seconds),calcium was present in each of the biggest inclusions.Inclusion characteristics in Figure 9(7minutes,20seconds)are similar in regardElectron image 1S Ka1Al Ka1Ca Ka1O Ka1Fig.6—One of the biggest inclusions (diameter 15l m)found from sample 6(27min 30s).The average analysis from the whole inclu-sion is:O 28wt pct,Al 14wt pct,Ca 33wt pct,and S 25wtpct.Fig.7—Industrial experiment.Sample was taken 55s afterCa-injection.Fig.8—Industrial experiment.Sample was taken 2min 50s afterCa-injection.Fig.9—Industrial experiment.Sample was taken 7min 20s after Ca-injection.to the inclusion size and the composition with the previous inclusion;the few biggest inclusions were larger with a higher oxygen content and a smaller manganese content.About 20minutes after the Ca addition,the alloying adjustments were made for Al,Mn,and Cr.The influence of the adjustments (shown in Figure 10,27minutes 30seconds)on the steel composition with greater values of Al and Mn.Even the inclusions were richer in aluminum than in the previous sample.It is also noteworthy that the Ca content gradually decreased from 37ppm (55seconds sample)to 21ppm (27min-utes,30seconds sample).Figure 11shows the inclusion paths in the same heat.For Figure 11,about 150inclusions were analyzed in each sample.The results were calculated so that the average amount of CaS,CaO,and Al 2O 3can be seen at different times after the Ca treatment.Although manganese sulfide formed first during the solidification and was not present in the inclusions at liquid-steel temperatures,it was taken into account when making the average inclusion analysis calculations.The calculation was done so that the amount of manganese sulfide was determined first from the manganese content of an inclusion.Second,the rest of the analyzed sulfur was assumed to form calcium sulfide.Third,the rest of the analyzed calcium was used to form calcium oxide.Finally,the amount of aluminum oxide was calculated from the aluminum content in the inclusion.Figure 11shows the inclusions analyses and average sizes presented at different times after the Ca treatment.According to Figure 11,it seems that the calcium reacted fast with both the oxides and the sulfur and formed calcium aluminates and sulfides.A reason for this strong tendency of CaS to form is the relatively high sulfur content (0.03pct)in the heat.The average inclusion size was small,under 2l m.At its lowest,it was about 1.3l m after the Ca addition (170seconds at sample 2).After that time,the inclusion size increased,so that the highest value (1.96l m)is in sample 5,whichwas taken 980seconds after the Ca treatment.After that time,two samples were taken,which both had average inclusion size smaller than 1.9l m.C.Evaluation of Reaction Kinetics and Mechanism In reference to the CaO-Al 2O 3reaction experiments (Figure 12),at temperatures 1350°C and 1370°C,only less than 10mg of alumina was reacted,although the holding time was more than 1hour.At temperatures of 1420°C and greater,the weight of alumina reacting with CaO was more than 100mg at a holding time of 1hour or less.By measuring the reaction area,reaction time,mass of product,and composition of the liquid product,it was possible to calculate the overall reaction rate constant k overall .This parameter for the reaction rate could be used to describe the kinetics of the alumina inclusion modification.The flux of CaO in the liquid calcium aluminate transferred through the boundary layer to the surfaceofFig.10—Industrial experiment.Sample was taken 27min 30s afterCa-injection.Fig.11—Industrial experiment.Change of inclusions composition after the Ca addition in the ladle.Each point represents the average of composition and size of about 150inclusions per sample.Samples from 1to 6arerepresented.Fig.12—Reacted amount of Al 2O 3as a function of time at seven different temperatures.alumina by diffusion.The reaction rate can be written as follows:r CaO ;1¼1S dN CaO dt ¼ÀD eff D x C lCaO ÀC i CaOÀÁ¼k D C l CaO ÀC i CaOÀÁ½1where C CaO l is the concentration of CaO in the liquid calcium aluminate and C CaO i is the concentration of CaO at the interface between liquid calcium aluminate and alumina.N is the number of moles in the reaction,and t is the reaction time.Assuming the reaction is of the first order with respect to CaO,the dissolution rate of CaO on unit surface S can be given as the following equation:r CaO ;2¼1S dN CaOdt¼k R C i CaO½2In our experiments,for the two cylinders in contact,the reacting surface area was 2.124910À5m 2.In a steady state,the flux to the surface was equal to the reaction rate at the surface.So,r CaO,1=r CaO,2,andfrom Eqs.[1]and [2],k D (C CaO l ÀC CaO i )=k R C CaO i,which results in the following equation:C i CaO ¼k Dk D þk RC l CaO½3Inserting Eq.[3]into either Eqs.[1]or [2],then elimi-nating C CaO i ,which cannot be measured,gives the fol-lowing equation:r ¼À11k D þ1k R C l CaO ¼k overall C lCaO mol m 2Ás !½4 Rate,r ,in Eq.[4]means that moles of CaO are reactingper unit volume per unit time.[13]When assuming that the liquid calcium aluminate is in the form of C 12A 7,the value of C CaO l is as follows:C l CaO¼q l CaO n CaO¼1338600ðg =m 3Þ56ðg =mol Þ¼23903:6ðmol =m 3Þ½5Table I shows the average values from several measure-ments at each temperature.Certain shifts of magnitudecan be seen in the temperature range between 1390and 1420°C.In this temperature region,the controlling mechanism of the reaction is evidently changing.The same phenomenon can be seen in Figure 13,where the results marked with a square (1405°C)and a rectangle (1390°C)are in this region.In this temperature region,the reaction rate is temperature sensitive,and the apparent activation energy has a high value,which indicates a transition in the controlling mechanism.[13]The rate-controlling mechanism in temperatures below this region is assumed to be Ca diffusion.[3]At temper-atures above the region,the chemical reaction between the alumina and the liquid calcium aluminate is assumed to be the rate-controlling mechanism.IV.CONCLUSIONSThe mechanism and kinetics of alumina inclusion modification were investigated by studying the reaction between CaO and Al 2O 3as well as between calcium and Al 2O 3inclusions in steel melts.Experiments were divided into two different groups that support each other;reactions between Al 2O 3and CaO were done in a laboratory resistance furnace,and Ca treatment was done on an industrial scale with melt sampling after the Ca addition.From the first type of experiments,it was possible to see how the reactions proceeded between CaO and Al 2O 3.The exact amount of reacted alumina was measured,and the reaction rate constants were calcu-lated between 1350°C and 1600°C.Reaction rate constants were drawn as a function of temperature so that the activation energies could be calculated from a straight line.The function showed that between 1370°C and 1420°C,a shift occurs in the controlling mechanism of the reaction,so linear dependence cannot be applied in this region.The result was interpreted to mean that,at lower temperatures,the reaction is diffusion con-trolled,but at higher temperatures,becomes chemical-reaction controlled.Concerning the interaction betweenTable I.Reaction Kinetics Between Solid CaO and Al 2O 3.Reaction Rate Constants Determined at Different Temperatures T(°C)135013701390140514201450150015501600k,m/s1.4910À101.9910À101.2910À95.8910À91.3910À62.0910À68.2910À61.4910À54.3910À5Fig.13—Logarithm of rate constant vs reciprocal temperature (1/K)that shows the temperature dependence of the reaction rate.Activa-tion energy E is shown between 1420°C and 1600°C.liquid calcium aluminate and an alumina inclusion,the chemical reaction between the solid alumina and the liquid calcium aluminate was assumed to be the rate-controlling step at temperatures greater than 1420°C.Also,experiments between mixed and pressed Al2O3 and CaO powders were performed.These samples were analyzed with SEM/EDS and XRD to identify the calcium aluminate phases formed.The phases formed at 1400°C in6hours were12CaO97Al2O3(41pct), CaO9Al2O3(36pct),CaO92Al2O3(11pct),and the rest were the initial materials CaO and Al2O3.The second groups of experiments were Ca treat-ments done in an industrial scale.The samples taken from the melt at certain intervals after the Ca addition were analyzed with SEM/EDS software,which auto-matically collected chemical and morphological data from inclusions in the steel.Figures showing the changes in the inclusion composition and size after the Ca treatment were produced.The results showed that even in industrial,77-ton,scale,calcium reactions take place rapidly;transformation of alumina inclu-sions to calcium aluminates was completed in a few minutes.During a longer delay time,calcium content gradually decreases.REFERENCES1.D.-Z.Lu,G.A.Irons,and W.-K.Lu:Scaninject VI,Lulea,Sweden,1992,pp.239–63.2.K.Ta htinen,R.Va ino la,and R.Sandholm:Scaninject II,Int.Conf.Injection Metallurgy,Lulea,Sweden,1980,pp.1–24.3.I.Kohatsu and G.W.Brindley:Zeitschrift fu¨r Physikalische Che-mie Neue Folge,1968,vol.60,pp.79–89.4.C.W.Bale,A.d.Pelton,W.T.Thompson,G.Erikson,K.Hack,P.Chartrand,S.Decterov,J.Melancon,and S.Petersen:FactSage5.4.1,.5.G.Eriksson and A.D.Pelton:Metall.Trans.B,1993,vol.24B,pp.807–15.6.D.A.Jerebtsov and G.G.Mikhailov:Ceram.Int.,2001,vol.27(1),pp.25–28.7.D.-Z.Lu,G.A.Irons,and W.-K.Lu:Ironmaking Steelmaking,1994,vol.21(5),pp.362–71.8.G.M.Faulring,J.W.Farrell,and D.C.Hilty:Iron Steelmaker,1980,vol.7(2),pp.14–20.9.Y.Higuchi,M.Numata,S.Fukagawa,and K.Shinme:ISIJ Int.,1996,vol.36,pp.S151–54.10.L.Holappa,M.Ha ma la inen,M.Liukkonen,and M.Lind:Iron-making Steelmaking,2003,vol.30(2),pp.111–15.11.Z.Han,M.Lind,and L.Holappa:Int.Conf.on Non-metallic Inclu-sions Control and Continuous Improvement of Processes based on Objective Measurement,STA˚L2004,Borla nge,Sweden,2004,p.13.12.M.Lind:PhD Dissertation,Helsinki University of Technology,Espoo,Finland,2006.13.O.Levenspiel:Chemical Reaction Engineering,3rd ed.,Wiley,New York,NY,1999,p.586.。