Thermochemistry and Kinetics of Iron Melt Treatment

Thermochemistry and Kinetics of Iron Melt Treatment Simon N. Lekakh *, David G. C. Robertson * and Carl R. Loper Jr. ** * University of Missouri Rolla, ** University of Wisconsin-Milwaukee, U.S.A. Abstract Nodulization for producing spherical graphite in ductile iron (DI) and inoculation for increasing graphite nodule count are the result of chemical reactions of the additives with the melt. The multi-component equilibria and the kinetics of additive dissolution in the melt were computed and studied experimentally. Measurements of the oxygen activity confirmed the calculated features of the refining reactions. Possible reaction paths were identified, as ferrosilicon-based additives dissolved and reacted with the iron melts. Adding more nuclei to the melt, by C/SiC melt pretreatment and late additions of small amounts of oxides and sulfides, improved DI inoculation. Intensification of inoculant dissolution by changing the additive shape, and by argon stirring, were also investigated for increasing inoculation efficiency. Key words Ductile iron, nodulization, inoculation 68/1

Introduction Ductile iron (DI) treatment dramatically changes the structure and the properties of castings. Deep refining of the melt by Mg transforms the flake graphite to spheroidal graphite. Because liquid iron treated by magnesium has a tendency to undercool resulting in meta-stable cementite formation, DI treatment also includes inoculation by FeSiX additives which create additional graphite nuclei, resulting in a carbide-free structure even in thin wall castings. Therefore, the industrial DI treatment typically consists of two stage nodulization and inoculation. Nodulization. Magnesium is the main reagent used in nodulization. Other metals (Ca, Ba, and Ce) are often used to increase the efficiency of nodulization. These elements have a large affinity for both the sulfur and oxygen impurities in the iron melts and remove these soluble impurities. Also the remaining concentrations of Mg and Ce promote graphite spheroidization in DI. The processes which take place during DI treatment are often described as analogous to steel refining. However, the high carbon and silicon contents in irons significantly change the sequences of the refining reactions. In this study, the thermodynamic features of the refining reactions of alkali and rare earth metals with impurities of sulfur and oxygen were investigated by using computer simulation and experiments. These methods were also applied to the optimization of the additive compositions for the nodulizing stage of DI treatment. Inoculation. The nucleation of graphite nodules before the formation of carbides requires the presence of substrates that can initiate solidification [1]. A possible ranking of nucleants is as follows: graphite (highest least energy required), silicate, oxides, sulfides, carbides, nitrides, and austenite (lowest). The strong inoculation effect of FeSiX alloys can be considered to be due to the formation of graphite-containing substrates during dissolution of inoculants in the iron melt. When FeSiX dissolves, regions with high Si concentration will arise around the inoculant particles [2]. In these regions, favorable conditions for the formation of graphite-containing substrates are created. According to thermodynamic equilibrium [3, 4], graphite will be stable in liquid irons alloyed by Si >5-6% at 13-15ºC and SiC will be stable when Si >23-28%. A number of kinetic and thermodynamic factors influence these processes, including the rate of inoculant dissolution versus the rate of mixing, and the additional chemical reactions which may create extra non-metallic substrates for subsequent growth of graphite [5,6]. Experimental analysis of graphite nuclei compositions is given in recent research [7]. In this work, the thermodynamics and kinetics of nodulization and inoculation were studied using both computational and experimental methods. Some methods of improving DI treatments were developed on the basis of the additional understanding of the mechanism of the process. 68/2

Procedure Computing. In the actual DI treatments parallel reactions take place between the additives and the impurities already present in the melt. To predict which products will be formed as a function of the composition and the temperature, computer simulation (FACTSage software) based on the minimization of the Gibbs free energy was used. The parallel reactions were simulated step-by step as the additive amount was increased. It was possible to calculate both the products of the reactions as well as the composition of the remaining melt. The thermodynamic calculations allow us to predict the maximum possible degree of reaction, but in real processes the kinetics also play an important role in determining how the process actually proceeds. Unsteady heat transfer between the iron melt and the initially cold additive particle of FeSiX based nodulizers and inoculants was computed while taking into account the latent heat as well as the heat of possible exothermic reactions. The FLUENT software package was used for simulation of dissolution process with laminar melt flow. Experiments with high purity iron. High purity iron with 3.7%C, 1.8%Si and controlled concentration of impurities was used in this study. Special alkali and rare earth containing master alloys (with unreactive Fe, Cu or Ni) were used for the refining experiments, which were carried out under argon. These master alloys had high recovery in the small volume of the treated melts (.3 kg). An electrochemical method was used to qualitatively measure the oxygen activity in the melts. Experiments with industrial grade irons. Industrial grade irons were melted in a 5 kg induction furnace. Two types on FeSiX additives were used for DI treatment in the ladle. The first type were conventional additives with an equi-axed shape with 8-12 mm particles. The second type were rapidly cooled ribbon-shape additives with a thickness of.5-2. mm. These were produced directly from the melt by a continuous casting process with the use of a water-cooled copper wheel [8]. The dissolution of the additives was studied with thermocouples placed in the center of the additives submerged into the iron melt. In addition, two special reagents were tested for increasing the efficiency of the DI treatment. Reagent 1 was a treated graphite (75%C, 25%SiC), and was used as a pre-treatment agent before nodulization. Reagent 2 was an additive containing a 1:1 mixture of non-ferrous (Cu,Fe) oxides and sulfides, and was used together with the FeSiX inoculant after the nodulizing treatment. A step bar casting with thickness from 6 to 5 mm, a chill wedge, and a six-pin core mold with 12mm, 1mm, 8mm, 6mm, 4mm, and 2mm pin diameters, were used for evaluation of the effectiveness of these treatments. The samples were cut, polished, and etched. Structures were quantitatively analyzed using OPTIMAS software. 68/3

Results and Discussion Thermochemistry of nodulization. During nodulization treatment, the additives react with the impurities in the melt, and the sequence of the refining reactions depends on the type and the quantity of additive as well as on the melt composition. The computer simulations and the experiments determined these sequences by taking into account all these parameters at a particular temperature. The calculations showing the possible parallel reactions while increasing the amount of Mg added to the iron melts (with different initial sulfur contents) are given in Figure 1a. A small amount of Mg produces significant deoxidation of the cast iron melts. But then, with an increase in the amount of magnesium, desulfurization occurs and the two refining reactions take place in parallel. The oxygen potential in equilibrium with the level of free magnesium required for transformation of the flake graphite to spherical graphite (.2.3 wt.%) is shown by a dashed line. This predicted sequence of reactions was experimentally confirmed when the irons with different initial S were treated by Mg (Figure 1b). A small amount of Mg dramatically decreased a. In the liquid irons having larger initial sulfur contents the reaction with sulfur forces deoxidation to occur later. In this condition, two parallel reactions take place, and the concentration of sulfur in the melt and a decrease at the same time. When the magnesium addition is sufficient to react with effectively all the sulfur and to decrease the oxygen activity to a value less than 1x1-4 wt.%, the flake shape of graphite transforms to the spherical shape during graphite growth. In contrast to magnesium, calcium reacts first with sulfur in the liquid iron and only after thorough desulfurization will an increase in the amount of the calcium addition allow it to react with oxygen (Figure 2a). Also, the extent of deoxidization will be limited by calcium carbide formation in the liquid iron (dash dotted line). In the high purity Fe-C-Si alloy, the measured a decreased with an increase in the amount of calcium additive. When the initial melt contained.4% S, there was a negligible influence of small amounts of calcium additive on a. In this melt, the oxygen activity decreased only after desulfurization with larger amounts of the calcium additive, an a smaller than 1x1-4 wt. % was not reached, and the shape of flake graphite did not change. The experiments qualitatively confirmed the calculated prediction of the sequences of the refining reactions during iron nodulization. In contrast to to liquid steel, where they all first produce deep deoxidation and then further additions produce desulfurization, the behavior is more complex in cast irons, as shown in Figure 3a. The individual features of the reactive species can be exploited in the design of the complex refining additives to be used, for example, for increasing the effectiveness of DI treatment. An example of a 3-dimensional diagram which describes the interaction of [Mg] a and [Ca] a additives with impurities in the liquid iron melt is given in Figure 3b. The regions of the reactions of magnesium additives with oxygen and sulfur are indicated as [Mg] O and [Mg] S respectively. 68/4

Magnesium additions above those required to react with O and S create free magnesium in the melt [Mg] f, the quantity of which depends on the amount of calcium additive [Ca] a. If complex additives are used containing both Mg and Ca then the refining functions are divided between these elements. In general, most of the Ca is consumed in the reaction with sulfur, while the Mg reacts with oxygen. If the amount of calcium is not enough for desulfurization, then any excess of Mg left after deoxidation continues to react with sulfur. It is important to note, that Ca, Ba and Ce can decrease the critical values of the Mg consumption, which are necessary for ductile iron nodulizing treatment. Kinetics of additive dissolution in iron melts. The transfer of a solid additive to the iron melt can generally be assumed to occur either as dissolution by melting, which occurs when the melting temperature of the additive is below the temperature of the melt, or as dissolution by diffusion, which takes place when the additive has a melting point higher than the melt temperature. The differences between the dissolution mechanisms were studied when the quenched regions around a carbon raiser (dissolution by diffusion) and ferrosilicon (dissolution by melting) were analysed in the iron melts. Primary graphite phases adjacent to the raiser and ferrosilicon show the supersaturated dissolution regions. Also the large variations in the volume of the dissolution regions occur in these cases. During the dissolution of the ferrosilicon based additives, two important features are (a) the strong exothermic reaction (which could decrease the dissolution time) and (b) the formation of slag shells around the additive particles. These occur when the alkali earth metals (X) in FeSiX react with with sulfur and oxygen in the melt. These slag shells can have a negative effect on the melt treatment efficiency. These two effects were studied by computing the melting times of cylindrical and spherical shapes of ferrosilicon additives in the iron melt and by comparison with the experimentally measured melting times. Experimental temperature curves from thermocouples placed in the center of graphite, iron, and Fe75%Si cylinders (Ø25mm x 5mm) which were simultaneously submerged in the iron melt are given in Figure 4a. These materials were chosen because they exhibit the three principal different melting/dissolution mechanisms. Graphite, with its high thermal conductivity, heats up quickly in the iron melt but the process of diffusion dissolution is slow. The graphite cylinder did not dissolve measurably during the experimental time. Melting of an iron cylinder with the same chemistry as the iron melt is not accompanied by dissolution heat. Finally, the temperature rise of the less conductive Fe75%Si had a time delay relative to the iron, but then increased quickly when exothermic reaction started at the melting boundary. As a result, the temperatures of the centers of the iron and Fe75%Si cylinders approached the melt temperature practically simultaneously and then the dissolution heat 68/5

liberation further increased the temperature around the melted Fe75%Si specimen, when compared to the initial melt temperature. At the same time, the question arises, where is the energy of dissolution released: directly at the additive/melt boundary or partially distributed in the bulk melt? The fraction of the exothermic dissolution heat released at the melting boundary was evaluated by computing the melting time for different values of this fraction and comparing it to the experimental dissolution time. The values of these melting times were equal when approximately half of the dissolution heat was released at the melting boundary and half in the bulk melt. When a Fe75%Si additive begins to melt, silicon-rich liquid will be convected away from the boundary as it dissolves in the bulk liquid metal. This effect could be exploited for increasing the inoculation efficiency because the regions of high silicon liquid alloy create conditions for graphite nuclei formation directly in the bulk iron melt. The effects of argon stirring of the melt, additive compositions and shape of particles on rate of dissolution were experimentally studied. Additional mixing forces had significantly higher influence on the thermal behavior of Fe75%Si cylinders compared to carbon cylinders when both were simultaneously submerged in a ladle with a bottom porous plug for producing the active argon agitation of the melt (Figure 4b). Argon stirring intensified dissolution, not only by increasing heat transfer between the melt and the additive surface but also possibly by fragmentation of the mushy zone of the additive. The effects of the composition and the shape of additives on the dissolution time were experimentally evaluated. The dissolution rate of FeSiX complex additive alloyed by calcium was significantly reduced because the Ca component reacted with impurities in the melt with the formation of low thermal conductivity slag phases (oxy-sulfide type). As a result, the recovery of this type of additive may be significantly decreased. Changing the traditional equi-axed shape of the additive particles to a ribbon-shape [8] significantly increased the rate of dissolution and the recovery of elements from the additives with the same chemical compositions. Unlike traditional additives, the ribbon-shaped additives heated up quickly due to their high surface area to volume ratio. Also any slag shells that formed did not fully isolate the ribbons from the melt. Improvements of DI treatments. Improvements in nodulization and inoculation were suggested on the basis of the additional knowledge of thermochemistry and kinetics of DI treatments. Four examples are given bellow. 1. Computing of optimal nodulizer composition and particle shape. The optimal Fe5%Si5%Mg additionally alloyed with Ca nodulizers were computed while taking into account the thermochemistry of refining reactions and kinetics of dissolution (Figure 5a). Additional alloying by 68/6

calcium increased the concentration of free magnesium in the melt after treatment and decreased the consumption on nodulizer for DI treatment. Unfortunately, at the same time, calcium sulfides/oxides were formed around the spherical shape additives, decreasing the dissolution rate and the additive recovery. As a result, a function with a minimum is obtained - the position of which depends on the initial sulfur content in the initial melt. The optimum calcium contents of the nodulizer for the various initial sulfur contents are shown by the dotted line. Because calcium does not have a large negative influence on the dissolution rate of the ribbon-shape nodulizers, they could be used in the optimal range of high-calcium compositions (dashed line). 2. Pretreatment of DI by C/SiC additives. These experiments were conducted in a commercial DI foundry, using.1% of pretreatment agents (silicon treated graphite with 75%C and 25%SiC) placed on top of the FeSiMg nodulizer in the tundish ladle. The effectiveness of this treatment was evaluated by statistical analysis of uniformity of graphite nodule count distribution. It was found [9] that pre-treatment is a technique which enhances the nucleation of graphite and might not exhibit a significant influence on average nodule count but might be expected to result in more uniform graphite nodule distribution. 3. Improving DI inoculation. Thermodynamic analysis showed that there are differences in the processes when non-refined and refined-bymagnesium melts are inoculated. In the first case, reactive alkali and rare earth metals create the substrates, as a result of their reactions with the impurities of oxygen and sulfur in the melt. However, melts refined by Mg lose this possibility, since they will no longer contain significant O and S as impurities. As a result, the FeSiX inoculants with alkali and/or rare earth metals are not so effective for DI treatment of melts previously deeply refined by Mg in order to obtain the graphite nodules. On the other hand Mg-treated melts have the important potential possibility of in-situ nucleus formation when small amounts of active impurities S and O are introduced into the melt. Special additions of a small amount (.1-.2%) of a mixture contained copper/iron sulfides/oxides, which can react with Mg, intensifies inoculation of DI by the regular.2% addition of Fe75%Si. The possibilities of the formation of carbide-free structures in thin-wall castings and of a significant increase of graphite nodule count by using this technique have been experimentally confirmed (Figure 5 b). 4. Intensification by argon stirring in ladle. Because the process of inoculation significantly depends on the dissolution kinetics of inoculants in the melt, a new inoculation technique was suggested and tested under lab conditions. This technique combined inoculation in the ladle with argon stirring. The test showed, even in a small 1 lbs. laboratory scale ladle, that active argon stirring increased the inoculation effect and cast structure uniformity. 68/7

Conclusions Computer simulations and measurements of the oxygen activities were used to determine the sequences of the refining reactions during nodulizing treatment of DI. These calculations, together with the experimental data of additive dissolution, were used for optimization of the compositions of the complex additives for DI treatment. The effectiveness of DI treatment may be increased by realizing the advantages that individual active elements have, and by the use of ribbon-shape additions to give faster dissolution and increase recovery. Iron inoculation was analyzed from the point of view of the non-equilibrium dissolution of FeSiX in the melt. Dissolution kinetics were computed and compared with the experimentally measured melting times which confirmed the formation of the regions with the high silicon contents in the bulk melt. As a result, graphite nuclei would form in these regions directly from the melt. Some methods of improving nodulization and inoculation were discussed. Because non-equilibrium regions are responsible for nuclei formation, any methods which can increase the distribution of supersaturated regions in the melt may be used for inoculation improvement. These include in-stream inoculation and forced stirring of the melt. References 1. Loper C R, Inoculation of Cast Iron Summary of Current Understanding, AFS Transactions, vol. 17, 1999, pp 523-528. 2. Wang C H and Fredriksson H J, The Mechanism of Inoculation of Cast Iron Melts, Proc. 48 th Int. Foundry Congress, 1981, pp 16-26. 3. Lekakh S and Bestyzev N, Ladle Metallurgy of High Quality Cast Iron, Nauka&Tekhnika, Minsk, USSR, 1992. 4. Lekakh S and Loper C R Jr, Improving Inoculation of Ductile Iron, AFS Transactions, vol. 111, 23, paper 3-13. 5. Skaland T, Nucleation Mechanism in Dictile iron, Proc. AFS Cast Iron Inoculation Conference, 25, pp 13-3. 6. Igarashi Y and Senri Okada S, Observation and analysis of the nucleus of spheroidal graphite in magnesium treated ductile iron, Int. J. Cast Metals Res, 11, 1998, pp 83-88. 7. Riposan I, Chisamera M, Stan S and Skaland T, A new Approach to Graphite Nucleation Mechanism in Gray Irons, AFS Cast Iron Inoculation Conference, 25, pp 31-41. 8. Sverdlin A, Lekakh S, Kalinitchenko A and Sheinert V, Chipsprocess for cast iron inoculation, Foundry Management & Technology, vol. 5, 1994, pp. 31-34. 9. Loper C R, Winardi L and Lekakh S, Experiments in pretreatment of Ductile Iron, AFS Transactions, vol. 11, 22, paper 2-2. 68/8

(wt. %) log [a O]x1 4 (wt.%) (wt.%) log [a O ]x1 4 (wt.%) (wt.%) (wt. %) Figures.1 4.5.8.6.4 3.4.2.3.6.9.12.15 Magnesium additive (wt. %) 2.4%S.3-2.2-3 -4 1 DI.1-5 -6 DI.5.1.15.3.6.9.12.15 Magnesium additive (wt. %).5%S Magnesium additive (wt.%) a) b) Figure 1. Calculated interactions (a) of Mg in liquid iron with initial.1%s (dotted lines),.4%s (dashed lines), and.1%s (solid lines), and experimentally measured a (b).4 Cast Iron -2.2 4 3-4 2.4S.5S.5.4.3.2 CaC 2 1.1.5.1 Calcium additive (wt. %) -6.5.1.15.2 Calcium additive (wt.%) a) b) Figure 2. Calculated interactions (a) of Ca in liquid iron and experimentally measured a (b) 68/9

Consumption of additive (wt.%) Graphite particles/mm2 Temperature, o C Temperature, o C Cast iron Steel [Mg] f (wt.%) Ca.6 Mg Ce.3 [Mg] S Additions, wt.% Mg Ca Ce.1 [Ca] a (wt.%).5 [Mg] f [Mg] O.6 [Mg] a (wt.%) a) b) Figure 3. Sequences of reactions in liquid steel and iron (a) and interaction of Mg-Ca additives in iron melt with initial.4%s,.7%o (b) 14 12 1 14 12 1 8 8 6 iron 6 graphite in furnace 4 2 graphite Fe75Si 4 2 Fe75Si in furnace graphite in ladle Fe75Si in ladle 5 1 15 2 25 3 35 4 Time, sec 5 1 15 2 25 3 35 4 Time, sec a) b) Figure 4. Measured temperature of carbon, iron and Fe75%Si cylinders submerged in iron melt in furnace at 138 ºC (a) and influence of argon stirring in ladle on temperatures of graphite and Fe75%Si cylinders (b) 2.2 9.2 wt.% 2 Optimal 7 1.8.1 wt.% 5 1.6 3 1.4.5 wt.% 1 2 3 4 5 Calcium in Fe5%Si5%Mg (wt.%) 1 1 3 5 7 9 Thickness, mm a) b) Figure 5. Optimization of Fe5%Si5%Mg nodulizer alloyed by Ca (a) and improwing of.2% Fe75%Si inoculation by.2% oxy-sulfide additive (dushed line) treatment (b) 68/1