Feasibility of Solid-state Steelmaking from Cast Iron ...

© 2012 ISIJ 26

ISIJ International, Vol. 52 (2012), No. 1, pp. 26–34

Feasibility of Solid-state Steelmaking from Cast Iron-Decarburization of Rapidly Solidified Cast Iron-

Ji-Ook PARK, Tran Van LONG and Yasushi SASAKI

Graduate Institute of Ferrous Technology (GIFT), Pohang University of Science and Technology (POSTECH), Hyoja-dong,Pohang, 790-784 South Korea.

(Received on September 6, 2011; accepted on September 29, 2011)

To meet the unprecedented demand of environmental issues and tightened production cost, steelindustry must develop the disruptively innovative process. In the present study, totally new steelmakingprocess of ‘Solid State Steelmaking’ (or S3 process) without BOF process or liquid state oxidation processis proposed. The overview of the new process is as follows: (1) High carbon liquid iron from the ironmak-ing processes is directly solidified by using a strip casting process to produce high carbon thin sheets. (2)Then, the produced cast iron sheet is decarburized by introducing oxidizing gas of H2O or CO2 in a con-tinuous annealing line to produce low carbon steel sheets. The most beneficial aspect of the S3 processis the elimination of several steps such as BOF, and secondary refinement processes and no formation ofinclusions. To investigate the feasibility of S3 process, the cast iron strips with various high carbon contentproduced by a centrifugal slip casting method are decarburized at 1 248 K and 1 373 K by using H2O–H2

gas mixture and its kinetics of the decarburization is investigated. In the decarburization process, the car-bon diffusion through the decarburized austenite phase but not the decomposition of cementite is therate controlling step of the decarburizing process. It is found that 0.5 mass% C sheets can be producedfrom 3.89 mass% C sheets with the thickness of 1.0 mm within 30 min at 1 373 K. Based on theseresults, S3 process is confirmed to be feasible as an alternative low cost steelmaking process althoughthe further improvement of the process will be necessary.

KEY WORDS: solid-state steelmaking; strip casting; gaseous decarburization; white cast iron; carbon diffusion.

1. Introduction

In the steelmaking process, many efforts have been paidon how to eliminate inclusions or bubbles from liquid steels.The existence of inclusions and bubbles is simply due to theliquid phase oxidation process of pig iron melts. The solu-bility of oxygen in liquid Fe is quite high so that oxygen iseasily dissolved into the liquid iron during the liquid statede-carburization process. The solubility of oxygen in solidiron, however, is extremely small. The intake of oxygen dur-ing the decarburization process of the solid phase Fe–C canbe negligible.

Based on this consideration, as an alternative approach toavoid the inclusion formation, totally new process which iscalled ‘Solid State Steelmaking’ (or S3 process) has beeninvestigated. Differed from the conventional steel sheet pro-duction process, pig iron (high carbon of about 5 mass%)from the ironmaking process is directly solidified into thinsheets by a strip caster and served for continuous solid statedecarburization process that removes carbon from cast ironby gas-solid decarburization reaction. Namely, the new pro-cess eliminates the liquid state oxidation of conventionalBOF process. Hot metal treatment before the strip castingwill be preferably carried out to adjust final composition ofthe sheet product. The process image of S3 process is sche-matically shown in Fig. 1. The compositional paths for the

conventional steelmaking and the solid state steelmakingprocess are shown in Fig. 2. The pO2 changes of the con-ventional steelmaking process and S3 process are shown inFig. 3. It is noted that there’s no need to struggle for the oxy-gen control before the casting in S3 process since there’s nooxygen blowing for decarburization. Since the oxygen con-tent in molten pig iron is extremely low of about less than5 ppm, the oxygen content in strip cast sheet can be less than5 ppm. Thus, inclusion formation is practically negligible.

Fig. 1. Schematic process flow of Solid state steelmaking(S3) pro-cess.

ISIJ International, Vol. 52 (2012), No. 1

27 © 2012 ISIJ

Another significant benefit of S3 process is the elimina-tion of many steps such as BOF, CC, secondary refinementand reheating or hot rolling process. By eliminating thosesteps, the energy, cost, CO2 emission and time in steel pro-duction process can be drastically reduced.

For the successful implementation and development of S3

process, the clear understanding and verification of the fol-lowing two processes are important.

(1) Strip casting of cast iron.(2) High speed solid state decarburization process to

meet the mass production.Strip casting of cast iron was already investigated by sev-

eral researchers1–5) and cast iron sheets with the thickness of0.5 mm to 3 mm were successfully produced. Their mainaims were the production of ductile cast iron sheet. Namely,the produced cast iron sheet was annealed to make precipi-tations of fine spheroidal graphite in a ferrite matrix.Recently, the high carbon steel sheet with about 0.5 mass%C has been commercially produced by using strip castingprocess by Nucor.6) Although the carbon content is relative-ly low compared with that of cast iron, it supports the fea-sibility of the commercial production of cast iron sheet bystrip casting. Based on these results, it can be said that therewill be no essential problems for the production of the castiron sheet by strip casting. Thus, by using the current stripcasting technology, the cast iron sheets can be possibly pro-duced.

Therefore, in this study, the decarburization behaviors ofthe solid Fe–C alloy with high carbon have been investigat-ed. A large number of studies on the solid state decarburiza-tion of Fe–C steel has been carried out until now.7–11) Forexample, the decarburization process of the austenite of 0.6mass% C was studied by Nomura7) and he calculated decar-burizing ferrite depth considering chemical composition ofsteel and heating condition. Marder8) studied the effect ofcarbon content on the kinetics of decarburization in Fe–Calloys of 0.8 mass% C at 1 089 K. In the previous studies,however, the carbon contents were generally less than about1.0 mass% and they mainly focused on the phase transfor-mation of austenite to ferrite phase by decarburization reac-tion. In the case of the decarburization of cast iron, however,the decarburization behaviors of the mixed phases ofcementite and austenite must be investigated. Unfortunately,few studies on the decarburization of cast iron have beencarried out10,11) and the details of its mechanism have not yetbeen clearly understood.

2. Experimental

2.1. ApparatusIn this study, a centrifugal casting method was employed

to cast molten iron for high solidification rate. A horizontalfurnace was used for decarburizing the cast iron specimen.The centrifugal casting machine and the decarburizing reac-tion system are shown in Figs. 4 and 5, respectively. The gasflow was controlled by the mass flow controller (MFC). Thegases were purified by passing through a drying unit filledwith CaSO4 and then passed through the deoxidizing unitfilled with magnesium chips at 723 K. For the setting of thesuitable ratio of H2O/H2, Ar and H2 mixture was passedthrough a water bath with a particular temperature. Themethod was the same as that described earlier.12) The vaporpressure of water is determined by using the reported mea-sured values.13) To avoid the condensation of the watervapor, heating coils were used to heat the gas delivery lineafter the water bath. The pH2/(pH2O+pH2) was fixed to 0.78.An alumina tube inside the resistance furnace was sealed bywater-cooling copper end cap. The decarburizing gas wasinjected through a small alumina tube. Samples were easilytaken out of the furnace or moved to end side of the furnaceafter decarburization by pulling rod for the water quenchingor air cooling.

Fig. 2. Compositional paths in the conventional process and S3 process.

Fig. 3. The pO2 changes in the conventional steelmaking and S3

process.

© 2012 ISIJ 28


2.2. Experimental ProcedureCast iron specimen was prepared by the centrifugal cast-

ing machine. At first, electrolytic iron (high purity) wasmelted with graphite flakes by induction heating at 1 923 Kfor 1 hr and sampled by a quartz sampling tube and thenwater quenched. This master alloy was put into the centrif-ugal casting machine to form a 1 mm thick cast iron strip.70 g of this rod type master alloy was mixed with pure elec-trolytic iron to control the carbon concentration, and thisalloy was put into an alumina crucible with ejecting holeand put into the casting chamber together with a coppermold. Then cast iron strips with three different carbon con-tent of 3.89, 4.35 and 4.89 mass% C were prepared. Thecasting chamber was evacuated by a rotary pump for 3 min-utes and Ar flushed for 1 minute. After 1 additional cycleof evacuation and Ar flushing, the master alloy was heatedup to 1 896 K by induction heating. After fully melted, themolten alloy was cast. After 1.5 minutes, the mold was putinto the water for fast cooling. After casting, the strip wascut into pieces with 10 mm × 20 mm × 1 mm size for decar-burization experiments.

H2O/H2 gas mixture was used as a decarburizing gas. Theratio of pH2O/pH2 was set to avoid the oxidation of the ironstrip. A cast iron strip with 10 mm × 20 mm × 1 mm in analumina boat was put into the horizontal furnace and heatedup to 1 248 K or 1 327 K at the 2%H2/Ar atmosphere. 1 248K and 1 327 K were selected for decarburizing temperaturebecause at those temperatures, no ferrite phase and no liquidphase are formed during decarburization according to thephase diagram. Heating rate was fixed to 15 K per minutesin all experiments. When temperature reached to the targettemperature, H2O/H2 gas was introduced instead of 2%H2/Ar for decarburization. Gas flow rate was fixed to 300 cc/min in all cases. Decarburizing time was 5, 15, 30, 60 and120 minutes for 1 248 K decarburization and 5, 15, 30 and60 minutes for 1 373 K decarburization. After decarburiza-tion, sample was water quenched or air cooled. Initially, itwas supposed that the carbon diffusion may occur aroundthe interface during air cooling. To examine the effect ofcooling rate on the carbon diffusion, two different coolingprocesses were applied. From the preliminary experiment, itwas found that the carbon diffusion around the interface dur-ing the air cooling is negligible. The difference of the cool-ing rate is only the formed phase, and the interface positiondoes not changed by the cooling arte.

2.3. Analysis MethodsAfter decarburization, the sample was cut into small piec-

es by a shearing machine. The average carbon content of thesample was measured by the combustion infrared detectionmethod. Standard sample calibration was done before mea-suring for accurate results. For the microstructure investiga-tion, a mounting sample was polished and etched by 2%Nital solution. An optical microscope was used for micro-structure analysis. EBSD was used to identify phasesthrough cross section of the samples.

3. Results

3.1. Microstructure of Decarburized Cast IronThe microstructure of the as cast sample (4.35 mass% C,

3.89 mass% C) is shown in Fig. 6, and is consisted ofcementite and ferrite phases. The ferrite can be formed bythe phase transformation during the cooling process. Theaverage size of cementite may decrease with the solidifica-tion cooling rate. It is expected that the decarburization rateof the cast iron with finer cementite will be faster. Themicrostructure may have a strong effect on the decarburiza-tion rate of cast iron. In this study, the sample was preparedby using slip casting. To investigate the supposed decarbur-ization process in S3 process, the sample should have almostthe same microstructure produced by strip casting. The cool-ing rate of the cast iron is the most important factor whichdetermines microstructure. Average cooling rate of the cen-trifugal casting machine is calculated from the secondary armspacing of austenite in the matrix. From Yoshida’s results,5)

the average cooling rate can be obtained by equation

Fig. 4. Centrifugal casting machine.

Fig. 5. Experimental gas flow system.


29 © 2012 ISIJ

.................. (1)

where d2 is secondary dendrite arm spacing(um) and v isaverage cooling rate(K/min). By measuring the secondaryarm spacing of 3.89 wt% C as cast strip as shown in Fig.6(b), the average cooling rate of the centrifugal casting wascalculated to be 103~104 K/s. From Table 1, the centrifugalcasting’s cooling rate is comparable with that of the stripcasting. From these values, it can be expected that the pro-duced microstructures by the centrifugal casting will not beso different from those of the strip casting.

The cross sectional images of 3.89 mass% C strip as afunction of the decarburization time at 1 248 K are shownin Fig. 7. Samples were air cooled and etched with 2% Nitalfor 3 seconds. All specimens show clear interface with flatplane and this interface moves from the surface to the centerof the strip with the decarburization time. After the decar-burization of 120 min, the interface from the top surface andthat from the bottom surface meet together and disappear.

The observed structures at 1 373 K are almost the same tothat at 1 248 K, but the moving speed of the interface ismuch faster than that at 1 248 K. Within 30 minutes, theinterface disappeared by meeting together. The thin whitelayer observed just near the surface at 1 248 K as well as1 373 K is found to be ferrite phase without cementite basedon EBSD analysis. It means that the carbon concentration atthe surface is so small that the cementite formation is neg-ligible even after 5 min decarburization.

EBSD analysis shows a large fraction of ferrite phase andexistence of small amount of cementite at the decarburizedlayer. These phases were formed during cooling after thedecarburization. On the other hand, the center area is mainlycomposed of cementite. This cementite was formed duringsolidification. No notable graphite segregation or agglomer-ates were observed.

The cross sectional images of 4.35 mass% C strip as afunction of decarburization time at 1 248 K are shown inFig. 8. The samples were water quenched and etched with2% Nital for 3 seconds. Similar with that of the previoushypo-eutectic strip, the clear interface with flat plane wasobserved. The coarse cementite needles at the center regionmaintained their shapes until the interface merged. Due tothe fast cooling rate by water quenching, the phase of decar-burized zone is mainly martensite rather than ferrite orpearlite. The microstructures of 4.89 mass% C strips decar-burized at 1 248 K were found to be nearly the same withthat of 4.35 mass% shown in Fig. 8. The microstructureimages of 4.89 mass% C strips decarburized at 1 373 K withvarying time are shown in Fig. 9. Differed from the previousresults, the position of the interface was not clear at all andthe depth of decarburization layer varies with the position.Furthermore, voids and big agglomerates were observed. Itseems that voids are originated from coarse graphite gran-ules and its detachment during polishing. Also this coarsegraphite may cause the irregular shape of the interfaceobserved in 4.89 mass% C strip. This result indicates thatthe large amount of the initial carbon concentration of thestrip is not suitable condition for both kinetic analysis andpractical application.

Fig. 6. As cast structure of sample. (a) 4.35 mass% C, (b) 3.89 mass% C

Table 1. Comparison between strip casting and other continuouscasting process.

Casting Thickness Averagecooling rate

Totalsolidification

time

A Conventionalcontinuous casting 150~300 mm ~12 K/s 600~1 100 sec

B Thin slab casting 20~50 mm ~50 K/s 40~60 sec

C Strip casting 1~4 mm ~2 000 K/s ~0.15 sec

d%C

v20 35 0 28340

1 1

4 3= −(

.) /. .

Fig. 7. Microstructure of 3.89 mass% C strip decarburized at 1 248 K for (a) 5 min., (b) 15 min., (c) 30 min., (d) 60 min.and (e) 120 min.

© 2012 ISIJ 30


3.2. Interface Moving RatesThe depth of the interface for 3.89 mass% C and 4.35

mas% C strips with time were plotted respectively in Figs.10(a) and 10(b). The depth of the interface position wasdefined as a distance between the reacting surface anddecarburization interface. The decrease of carbon content(Creduced) during the decarburization was calculated from themeasured average carbon content (Cav) after the decarbur-ization reaction in mass percent unit.

Creduced = C0 – Cav............................ (2)

where C0 is the initial carbon content in the strip and Cav isthe average carbon content in the strip after the decarburiza-tion.

It’s hard to determine exact value of depth of decarbur-ization in case of 4.89 mass% C strip due to its irregular pat-tern as shown before, only 3.89 mass% C and 4.35 mass%C strip’s decarburization results are shown in the figure.From Fig. 10, the reaction temperature has a strong effecton the change of depth of decarburized layer, but the initialcarbon content seems to have negligibly small effects on

these. The depth of the interface position was also plottedas a function of the square root of the reaction time, and theresults are shown in Fig. 11. They show reasonably goodlinear relation. Namely, the depth of the interface positioncan be expressed by the following equation;

y = Ct0.5 ................................... (3)

where, C is a constant and t is reaction time, y is the depthof decarburization or amount of carbon decreased.

4. Discussion

4.1. Kinetics of the Cast Iron DecarburizationThe decarburized layer thickness follows the parabolic

relationship as shown in Fig. 10. It means that the gas filmdiffusion step is not the rate-limiting process for the decar-burization process. As already mentioned, the surface of thespecimen was covered with ferritic phase that was formed dur-ing cooling to a room temperature after the decarburizationprocess. It means that the removal rate of carbon at the surfaceis much faster than that of the carbon supply from the insideto maintain the austenitic phase without cementite in the sur-face zone. Thus, the surface chemical reaction step is also notthe rate controlling step in the decarburization process.

If the dissolution of carbon from the cementite is the ratecontrolling step, the decarburization rate around cementitecan be slower than the austenitic matrix. Typically shown inFig. 8(c), however, the decarburization interface plane isessentially flat in all the cases regardless of cementite posi-tion and size. From this result, the dissolution of carbonfrom cementite to austenite is also not the rate controllingstep. Therefore, the carbon diffusion in the austenite phasefrom the decarburizing interface to the reacting surface canbe confirmed to be the rate controlling step of the gaseousdecarburization reaction of the cast iron.

The apparent decarburization reaction at the interfacewith the gas mixture of H2O–H2 is described by the expres-sion:

H2O + C → CO + H2 ........................ (4)

The decarburization reaction (4) can be further broken downto the elementary steps:

Fig. 8. Microstructure of 4.35 mass% C strip decarburized at 1 248 K for (a) 5 min., (b) 15 min., (c) 30 min., (d) 60 min.and (e) 120 min.

Fig. 9. Microstructure of 4.89 mass% C strip decarburized at 1 373K for (a) 5 min., (b) 15 min., (c) 30 min. and (d) 60 min.


31 © 2012 ISIJ

H2O → Oad + H2 ........................... (5)

Oad + H2 → H2O ........................... (5)’

Oad + Cad → CO............................ (6)

C → Cad .................................. (7)

where Oad and Cad are the adsorbed oxygen and carbonrespectively. Under the chemical equilibrium condition ofthe reaction (4), the concentration of Oad is determined bythe balance between the oxygen supply rate by the reaction(5) and the oxygen removal rate by the reaction (5)’. Name-ly, the Oad is determined by pH2O/pH2 ratio, or by the oxy-gen potential of the gas mixture. Due to the fast reactionrates of the reactions of (5) and (5)’, the equilibrium con-centration of Oad is very rapidly established at the surface.The equilibrium carbon activity at the surface under theH2O–H2 gas mixture is negligibly small. In other words, theequilibrium concentration of Cad at the surface will be esti-mated as zero. The reaction rate of (6) is also known to beextremely rapid.15–17) Thus, the Cad is immediately removedby the reaction (6) as soon as the carbon is diffused to thesurface, or Cad should be essentially zero at all times duringthe decarburization reaction. This is the reason that the fer-ritic phase without cementite is formed at the surface layereven at the very early stage of the decarburization process.

4.2. Diffusion Coefficient of the Carbon Diffusion in theAustenite Phase

Since the carbon diffusion in the austenite phase controlsoverall reaction, it is important to know the carbon diffusioncoefficient in austenite layer. The diffusion coefficient isgenerally a function of temperature and average carbon con-tent of the matrix. In this estimation, however, the diffusioncoefficient was assumed to be constant as a first approxima-tion. It was also assumed that carbon content of center

region is the same as the initial carbon content of the castiron. The carbon concentration at the reacting surface wasfixed to be zero from the experimental results.

Figure 12 shows the carbon concentration profile modelpredicted base on the mentioned assumptions and the exper-imental results. M(x) moves from 0 to L/2 with time andphase changes to phase II (austenite phase) as interfacemoves. The equilibrium condition at M is simply describedby

CII (M,t) = CII*.............................. (8)

From the material balance at the interface, the followingrelation can be satisfied.

................ (9)

where the moving phase boundary is at x = M. C0 and CII*denote the average carbon content of as cast strip and solu-bility of carbon at target temperature in the phase II (auste-nite) respectively. Also, DII and Cs represent average diffu-sion coefficient of carbon in the phase II and surface carbonconcentration respectively. Under these conditions, the fol-

Fig. 10. (a) The depth of the decarburized layer as a function of the decarburization time for the samples with 3.89mass% C, and (b) that for the samples with 4.35 mass% C.

Fig. 11. (a) The depth of the decarburized layer as a function of the square root of the decarburization time for the sam-ples with 3.89 mass% C, and (b) that for the samples with 4.35 mass% C.

Fig. 12. Schematic cross sectional carbon concentration profile ofthe sample during decarburization.

− ∂∂

= −=DC

xC C

dM

dtIIII

x M II*( ) ( )0

© 2012 ISIJ 32


lowing relation is obtained. The details can be found else-where.18)

................. (10)

where β is a dimensionless parameter to be determined oncethe values of (Cs – CII*) and (CII* – C0) are known.18) Theposition of interface M (x,t) can be described by,

............................ (11)

Equation (11) simply suggests that the thickness of thedecarburized layer is proportional to the square root of thereaction time, t1/2. It is well agreed with the experimentalresults shown in Fig. 11.

This reasonable agreement means that the described mod-el in the present study can be practically used to evaluate theinterface moving rate during the decarburization process ofcast iron. In the Eq. (10), the right hand term can be calcu-lated from the experimental measurements. Once value is known, we can evaluate the value of β from the lit-erature.18) In case of 3.89 mass% C, value of β is calculatedas 0.50 at 1 248 K and 0.63 at 1 378 K respectively. In caseof 4.35 mass% C, value of β is calculated as 0.44 at 1 248K and 0.60 at 1 378 K respectively. Then, from Eq. (11) withevaluated β value, we can find out relation between M, DII

and t. By curve fitting with the results shown in Fig. 11, DII

can be calculated easily by using Eq. (11). The results of cal-culation are shown in Table 2.

As already mentioned, the diffusion coefficient generallydepends on temperature and carbon content of the matrix.

Many models which predict carbon diffusion coefficient inFe–C system as a function of temperature and average car-bon content of the matrix have been investigated by severalresearchers. Several reported carbon diffusion coefficientsas a function of carbon concentration19–21) are plotted in Fig.13. The reported diffusion coefficient formulae19–21) are alsoshown in Table 3. In the present carbon diffusion model,however, the diffusion coefficients were assumed to be con-stant. As a first approximation, this constant value may beassumed to the value at the average carbon concentration inthe surface layer. The average carbon concentration wasdefined by (Cs + CII*)/2 in this study. They are 0.7 mass%at 1 248 K and 0.95 mass% at 1 373 K, respectively. The dif-fusion coefficients at the average carbon concentrations arealso shown as closed circles in Fig. 13. The evaluated valuesare reasonably close to the reported values.

For the successful development of S3 process, the infor-mation of the removal rate of carbon from the solidified castiron strip as well as the interface moving rate is essentiallyimportant. The amount of removed carbon from 3.89 mass%C and 4.35 mas% C strips with time were plotted respec-

Table 2. Evaluated carbon diffusion coefficients at austenite phase.

1 248 K(CII

* ~1.45 mass% C)1 373 K

(CII* ~1.93 mass% C)

3.89 mass% CDII = 1.89 × 10–7(cm2/s) DII = 8.66 × 10–7(cm2/s)

(Co = 3.89 mass% C)

4.35 mass% CDII = 1.81 × 10–7(cm2/s) DII = 7.76 × 10–7(cm2/s)

(Co = 4.35 mass% C)

β βπ

βe erfC C

C Cs II

*

II*

2

0

( )( )

( )=

−−

M D t II= 2β

β ββe erf 2

( )

Table 3. Reported models of the carbon diffusion coefficient as afunction of temperature and carbon concentration.

EquationA119)

EquationA220)

EquationA321)

D C expRT

cm sC Fe( ) . . , /γ − = + ⋅( ) ⋅ −⎛⎝⎜

⎞⎠⎟

0 07 0 0632 000 2

R cal mol K C wt= ⋅1 99. / , , .%

D CC Fe( ) .γ − = − ⋅( ) ⋅1 0 23

expC

TC cm s

4 300 18 9002 63 0 38

1 51 5 2⋅ − − ⋅ −

⎛

⎝⎜

⎞

⎠⎟

... . , /

R J mol K C wt= ⋅8 31. / , , .%

D expT T

C cm sC Fe( ).. . , /γ − = ⋅ − + −⎛

⎝⎜⎞⎠⎟

⎡

⎣⎢

⎤

⎦⎥0 78

18 900 4 3002 63 1 5 2

R J mol K C wt= ⋅8 31. / , , .%

Fig. 13. The evaluated carbon diffusion coefficients with the previously reported results.


33 © 2012 ISIJ

tively in Fig. 14. The amount of removed carbon was alsoplotted as a function of the square root of the reaction time,and results are shown in Fig. 15. It shows good linearity.Thus, amount of removed carbon is also likely to follow theparabolic law.

In the present study, the diffusion of carbon in the γ phaseprocess is found to be the rate controlling step. Then the car-bon removal molar flux at the surface, JC can be equal to thecarbon diffusion flux at the surface;

........................ (12)

Since the interface moving rate is relatively small, we mayassume that the carbon removal process is carried out underthe pseudo steady state as a first approximation;

........ (13)

Then,

JC ........................ (14)

By integrated Eq. (14) from t0 to t,

...... (15)

where S(t) is the total amount of the removed carbon for thedecarburization time of t. Namely, the amount of the carbonremoval is proportional to the square root of the reactiontime under the assumption of pseudo steady state. The par-abolic dependency shown in Fig. 15 can be explained basedon the Eq. (15). In other words, Eq. (15) can be used to eval-uate the approximate amount of the carbon reduction fromthe cast iron strip by the decarburization reaction.

As already mentioned, 0.5 mass% C steel is commercially

produced by using the strip caster at Nucor.6) This result isused as a benchmark to investigate the feasibility of S3 pro-cess. Namely, the production of 0.5 mass% C sheets by S3

process is discussed. As shown in Fig. 14, it takes about 30min to obtain the strip with the average carbon content of0.5 mass% C from the 3.89 mass% C strip at 1 373 K. Thecommercial casting speed of high carbon strip casting isabout 60 to 100 m/min.6,14) Then, the required length for thecontinuous decarburization treatment line at around 1 373 Kmust be about 1.8 to 3 km at least. This is a quite demandingcondition to make S3 process in practice. Thus, to make S3

process in practice, the decarburization treatment timeshould be much shortened. The enhancement of the decar-burization rate will be the main target of the future study.

5. Concluding Remarks

In this study, white cast iron strips with three differentcarbon content (3.89, 4.35 and 4.89 mass% C) are preparedby centrifugal casting method and the gaseous decarburiza-tion behaviors of these cast iron strips by H2O–H2 gas mix-tures have been investigated at 1 248 K and 1 373 K toinvestigate the feasibility of “direct steel sheet productionfrom solid cast iron”. Based on the microstructures andaverage carbon content of the decarburized specimen, it hasbeen found that the rate of decarburization of cast iron stripsof 1 mm thickness is controlled primarily by slow diffusionprocess in the austenite phase. The dissociation rate of thecementite into the austenite and carbon is fast enough tomaintain the equilibrium concentration distribution betweenthe cementite and the austenite phases at the interface.Based on the kinetic of carbon diffusion process, the appar-ent carbon diffusion coefficient in the austenite phase is cal-culated and it shows reasonable agreement with the reported

Fig. 14. (a) The amount of the reduced carbon as a function of the reaction time for the samples with 3.89 mass% and (b)that for the samples with 4.35 mass%.

Fig. 15. (a) The amount of the reduced carbon as a function of the square root of the reaction time for the samples with3.89 mass% and (b) that for the samples with 4.35 mass%.

JC = − ∂∂ =DC

x II

IIx( ) 0

− ∂∂

≈ − ∂∂

= −= =DC

x D

C

xC C

dM

dt II

IIx II

IIx M II

*( ) ( ) ( )0 0

≈ −( )C CdM

dtII*

0

S t( ) ( ) ( )≈ − = −C C M C C D tII*

II*

II0 0 2β

© 2012 ISIJ 34


results. The amount of the reduced carbon from the cast ironstrip approximately increases with the square root of thedecarburization time. The required decarburization time toproduce 0.5 mass% C steel from 3.89 mass% C cast ironstrips with 1.0 mm thickness at 1 373 K is found to be about30 min. and this reaction time might be a little bit demand-ing for the industrial applications. Although the furtherenhancement of the apparent decarburization rate will becertainly required, it can be said the S3 process is feasible,rather promising as an alternative steelmaking process tomeet the demands from environmental issues. Despite of itsinnovative feature, S3 can be employed as the actual indus-trial process with some modifications of the conventionalprocesses.

REFERENCES

1) J. M. Song, T. S. Lui and L. H. Chen: Metall. Mater. Trans. A, 33A(2002), 1263.

2) G. R. Lee, H. Y. Ra and J. Kor: Foundrymen’s Soc., 20 (2000), 122.3) J. Y. Park, K. T. Choi, Jerzy A. Szpunar, K. H. Oh and H. Y. Ra:

Scr. Mater., 46 (2002), 199.4) C. Yoshida, H. Yasunaka and T. Nozaki: Trans. Iron Steel Inst. Jpn.,

27 (1987), 819.5) C. Yoshida, K. Taniguchi, T. Nakagawa, M. Sudo and T. Nozaki:

Testu-to-Hagané, 72 (1986), 2240.6) D. G. Edelman, P. C. Campbell, C. R. Killmore, K. R. Carprnter, H.

R. Kau, J. G. Williams and W. N. Blejde: Iron Steel Technol., 6(2009), 47.

7) M. Nomura, H. Morimoto and M. Toyama: ISIJ Int., 40 (2000), 619.8) A. R. Marder, S. M. Perpetua, J. A. Kowalik and E. T. Stephenson:

Metall. Trans. A, 16A (1985), 1160.9) R. J. Fruehan and L. J. Martonik: High Temp. Sci., 3 (1971), 244.

10) J. Burke, J. Bell and D. R. Bourne: Acta Metall., 8 (1960), 864.11) K. Nagasaki and N. Komuro: Tetsu-to-Hagané, 49 (1963), 14.12) I.-H. Jeong, J.-S. Lee and Y. Sasaki: ISIJ Int., 51 (2011), 805.13) J. M. Coulson and F. D. Richardson: Chemical Engineering, vol.1,

Fluid flow, heat transfer and mass transfer, Butterworth-HenemannLtd., Oxford, (1966), 691.

14) M. Ferry: Direct strip casting of metals and alloys, Woodhead Pub.Ltd., Canbridge, (2006), 87.

15) G. R. Belton: Metall. Tans. B, 24B (1993), 241.16) H-S. Kim, J.-G. Kim and Y. Sasaki: ISIJ Int., 50 (2010), 678.17) Y. Sasaki and K. Ishii: ISIJ Int., 44 (2004), 439.18) G. H. Geiger and D. R. Poireir: Transport Phenomena in Metallurgy,

Addison-Wesley, London, (1973), 495.19) J. I. Goldstein and A. E. Moren: Metall. Trans. A, 9 (1978), 1515.20) A. Ruck, D. Monceau and H. J. Grabke: Steel Res., 67 (1996), 240.21) E. T. Georgeand and A. H. Maurice: Steel Heat Treatment Handbook,

Marcell Dekker, NY, (1997), 678.

Feasibility of Solid-state Steelmaking from Cast Iron ...

Documents