A Study of Hold Time, Fade Effects and Microstructure in Ductile Iron

Transcription

1 Proceedings of the AFS Cast Iron Inoculation Conference, September 29-30, 2005, Schaumburg, Illinois A Study of Hold Time, Fade Effects and Microstructure in Ductile Iron Copyright 2005 American Foundry Society ABSTRACT E. Huerta Gregg Industries, El Monte, CA V. Popovski Elkem Metals, Inc., Pittsburgh, PA A series of treated ductile iron ladles were inoculated with various generic inoculants and at various addition rates. The ladles were held for up to 16 minutes after addition of inoculant and samples were extracted every two to five min. Samples were examined for chemistry, thermal analysis properties and microstructure evolution. The investigation has shown that there are significant metallurgical changes during holding of inoculated iron in a ladle. Time and temperature effects during holding are found to result in potential benefits to delayed pouring. The metal shows improved ATAS (Adaptive Thermal Analysis System) thermal analysis graphite factors 1 and 2, reduced recalescence and better nodule count and nodularity as a function of hold time in the ladle. INTRODUCTION Fade, in ductile irons, is considered to be the loss of nodularity with time (Gundlach, 1992). The concept of fade in ductile iron (DI) includes the processes of magnesium (Mg) fade as well as inoculant fade. These two terms both refer to the deterioration of microstructure ( loss of nodularity ) that accompanies the holding of treated and inoculated metal in the pouring vessel for some time prior to actual pouring. The loss of residual Mg (Mg fade) results in the reversion of spheroidal graphite to compacted or even flake graphite. This condition has been depicted graphically in studies (Frush, 1998). As a result, foundries that feature manual pouring almost always impose a time limit on how long iron can remain in the ladle prior to pouring. This time limit is often seven to twelve minutes and its existence assumes that the iron will have lost sufficient Mg content after such an interval as to require pigging of the metal. The caution that accompanies such a procedure is the natural result of the fear of producing a casting that fails in the field. These concerns can be depicted graphically as shown in Fig. 1. Immediately after treatment and inoculation, the structure is assumed by many operators to be near ideal. This is because Mg fade has just begun and the established fade limit is therefore far off in time. There is plenty of opportunity to pour off the ladle before the microstructure deteriorates and fades to gray iron (GI). The metal will inevitably fade to GI eventually, but this model completely neglects the effects of inoculation, inoculant fade and the possibility of negative effects of Mg content. Inoculant fade is a lesser acknowledged condition whose importance is reflected in inoculation practices. The widespread use of in-stream inoculant when possible, as opposed to ladle inoculant, may be as a result of fade of inoculation effect in the pouring vessel. Castings made with faded inoculant have irregular graphite structures, just as those suffering from Mg fade. Fuller writes, In ductile iron the fading of inoculants causes a reduction in the number of centres from which growth of the eutectic occurs... The lower degree of nucleation... is often accompanied by a deterioration in the shape of the nodules (Fuller, 1979). Fuller associates this effect with a loss in nodule count and views nodule count as a reveal(er) of decreased nucleation. The fading of inoculant is attributed to the coarsening and growth of micro-inclusions, also called the Ostwald Ripening Effect. The driving force for this coarsening is a reduction in the specific surface area of the inclusions, thus reducing the total energy of the system (Olsen, 2004). Castings made under conditions with inoculant fade are also more chill prone. A principal effect of fading is to cause greater undercooling to take place during eutectic solidification and to lead to a greater tendency for chilling in gray and ductile irons particularly in thin sections (Skaland, 1992). At the same time, it is also clear that over-treatment of DI is detrimental to microstructure, resulting in irregular graphite shapes, carbides, and slag defects (Goodrich, 1992). Over-inoculation can similarly be a problem in that pronounced early graphite growth depletes the iron of graphite units that are needed at the end of eutectic freezing to combat microshrinkage. Over-treatment and over-inoculation are also excessively costly. 43

2 Proceedings of the AFS Cast Iron Inoculation Conference, September 29-30, 2005, Schaumburg, Illinois All DI processes include some combination of Mg loss and inoculant fade over time. If the losses of inoculation effect and Mg content are accepted as being detrimental at some point, and it is also recognized that an excess of these same quantities is a problem, then it is logical to ask certain questions: How long can the foundry wait until the iron has reverted to compacted or flake graphite? How long can the foundry wait until the iron exhibits the effect of the loss of inoculation (nucleation)? When should the foundry pour the casting to assure the most carbide-free microstructure? When should the foundry pour the casting to assure the most shrinkage-free casting? Does one inoculant stand out from others in terms of fade resistance? How is this evaluated? Both Mg and rare earths contribute to nodularization, but does either one offer an advantage to fade resistance? Ultimately, the foundry is producing and selling spheroidal graphite and not residual Mg. This paper represents the culmination of two separate but related field studies that attempted to quantify the fade effects described above. Traditional Fade Limit Nodularity of Ladle of Iron Gray Threshold Time After Treatment, Minutes EXPERIMENTAL PROCEDURE Fig. 1. The graph illustrates a traditional fade model. An initial set of experiments was conducted over two days, designated as Gregg Study I. A similar set of experiments was done over a single day, four months later. This set of experiments was designated as Gregg Study II. In the first set of experiments (Gregg Study I), magnesium ferrosilicon (MgFeSi) was added to the pocket of a 1000-lb sandwich pocket ladle at an addition rate of 1.8 wt%. The sizing and analysis of the MgFeSi can be seen in Table 1. 44

3 Proceedings of the AFS Cast Iron Inoculation Conference, September 29-30, 2005, Schaumburg, Illinois Table 1. Typical Composition of the Magnesium Ferrosilicon Alloy Used in This Study Size 3/4 x 8 mesh Si 46.5% Mg 5.72% TRE 1.26% Ce 0.65% La 0.42% Ca 1.15% Al 0.64% Cover steel (punchings) was added at an addition rate of 1.5 wt%. DI treatment was then conducted by tapping 1000 lbs of iron into the ladle. The ladle was then split into two 500-lb pouring ladles. Inoculant was added on transfer to the pouring ladle. In total, eight pouring ladles were made. The different inoculants applied can be seen in Table 2. Table 2. Specifications for Inoculants Used in This Study Inoculant A Calcium-bearing 50% FeSi (addition rates 0.3 and 0.7 wt%) Inoculant B Barium, Calcium-bearing, 75% FeSi (addition rate 0.3 wt %) Inoculant C Cerium, Calcium-bearing, 75% FeSi (addition rates 0.2 and 0.3 wt %) Inoculant D Cerium, Calcium, Sulfur, Oxygen-bearing 75% FeSi (addition rates 0.2 and 0.3 wt %) These inoculants were used at varying addition rates. For the purpose of clarity, an example is provided in Table 3 to explain the nomenclature for the individual samples in Gregg Study I. Table 3. Example to Illustrate Sample Nomenclature Ladle 1B3-1, where: the first digit 1 designates the first day of the study the second digit B designates which inoculant was used the third digit 3 designates that the addition rate of the inoculant was 0.3% the fourth digit 1 designates that it was the first iteration of this scheme on that day Metallurgical samples were taken from the treated and inoculated ladle immediately after inoculation. The ladle was then covered in a refractory blanket and allowed to remain undisturbed for five to six min, whereupon a second sample was taken. This was repeated several times, at varying lengths of hold time. Table 4 details which metallurgical tests were conducted in Gregg Study I. Table 4. Metallurgical Tests Applied to Test Metal Thermal Analysis cups at 0, approx. 5, 10, and 14 minutes after inoculation Chemistry - X-Ray Fluorescence of all solidified cups Quantitative Metallography - Image analysis of all solidified cups Gregg Study II was similar to Gregg Study I, except that only one inoculant was used (Inoculant A). This inoculant was used at four different addition rates (0.2, 0.4, 0.6, and 0.8 wt %). Samples were taken at shorter time intervals in Gregg Study II. 45

4 Proceedings of the AFS Cast Iron Inoculation Conference, September 29-30, 2005, Schaumburg, Illinois RESULTS FROM FOUNDRY TESTING Gregg Study I Figure 2, generated from data in Gregg Study I, shows that the pouring ladle temperature drops over time. The temperature plotted here is the peak temperature logged by the thermal analysis unit. This drop in temperature appears to be linear. Gregg Study I, Pouring Temperature A7 Pour Temp 1B3 Pour Temp 1C3 Pour Temp 1D3 Pour Temp 2A3-1 Pour Temp 2C2 Pour Temp 2A3-2 Pour Temp 2D2 Pour Temp Time After Inoculation, Minutes Fig. 2. Graph illustrates the pouring temperature over time, Gregg Study I. Figure 3 shows that TeLow drops over the first six minutes of hold time. This represents a quantified measure of inoculant fade. The metal is measurably more chill-prone after six minutes than immediately after inoculation. The results are more erratic afterwards, with some evidence of a rise in TeLow over time. Gregg Study I, TeLow A7 TeLow 1B3 TeLow 1C3 TeLow 1D3 TeLow 2A3-1 TeLow 2C2 TeLow 2A3-2 TeLow 2D2 TeLow Time After Inoculation, Minutes Fig. 3. The graph illustrates TeLow over time, Gregg Study I 46

5 Proceedings of the AFS Cast Iron Inoculation Conference, September 29-30, 2005, Schaumburg, Illinois Figure 4 shows the change in recalescence over time. Here the change over time becomes more pronounced. Recalescence is markedly lower six minutes after inoculation. There is some minor scatter after the first six min, but the overall trend is for a minimal change in recalescence after the first six min. Gregg Study I, Recalescence A7 R 1B3 R 1C3 R 1R3 R 2A3-1 R 2C2 R 2A3-2 R 2D2 R Time After Inoculation, Minutes Fig. 4. The graph illustrates recalescence over time, Gregg Study I Figure 5 shows that there is clear improvement in Graphite Factor 1 (GRF1) six minutes after pouring, with a nominal decline thereafter. The graph suggests that a casting poured six minutes into a ladle will feature considerably more graphite precipitation than one poured immediately after inoculation. This is the case regardless of inoculant. Gregg Study I, GRF A7 GRF1 1B3 GRF1 1C3 GRF1 1D3 GRF1 2A3-1 GRF1 2C2 GRF1 2A3-2 GRF1 2D2 GRF Time After Inoculation, Minutes Fig. 5. The graph illustrates Graphite Factor 1 over time, Gregg Study I Figure 6 shows a dramatic improvement in Graphite Factor 2 (GRF2) over time, again, regardless of inoculant choice. Metal poured after a delay of six minutes will have considerably more late graphite precipitation and therefore be far more resistant to micro-shrinkage than one poured immediately after inoculation. 47

6 Proceedings of the AFS Cast Iron Inoculation Conference, September 29-30, 2005, Schaumburg, Illinois Gregg Study II, GRF A7 GRF2 1B3 GRF2 1C3 GRF2 1D3 GRF2 2A3-1 GRF2 2C2 GRF2 2A3-2 GRF2 2D2 GRF Time After Inoculation, Minutes Fig. 6. The graph illustrates the Graphite Factor 2 over time, Gregg Study I Figure 7 shows an improvement in nodule count, regardless of inoculant choice, for the entire period of the test. This is consistent with the results of Fig. 5. Late nodules are smaller, so the nodule count should be higher. Gregg Study I, Nodule Count A7 Nodule Count 1B3 Nodule Count 1C3 Nodule Count 1D3 Nodule Count 2A3-1 Nodule Count 2C2 Nodule Count 2A3-2 Nodule Count 2D2 Nodule Count Time After Inoculation, Minutes Fig. 7. The graph illustrates the nodule count over time, Gregg Study I. 48

7 Proceedings of the AFS Cast Iron Inoculation Conference, September 29-30, 2005, Schaumburg, Illinois Figure 8 shows that nodularity actually increases over time, regardless of which inoculant is used, for at least the first ten minutes of the test. This result is a logical extension of the result in Figs. 6 and 7. Late nodules are smaller and smaller nodules are a closer approximation of a circle than are larger, more irregular nodules. Gregg Study I, % Nodularity A7 Nodularity 1B3 Nodularity 1C3 Nodularity 1D3 Nodularity 2A3-1 Nodularity 2C2 Nodularity 2A3-2 Nodularity 2D2 Nodularity Time After Inoculation, Minutes Fig. 8. The graph illustrates the nodularity over time, Gregg Study I. Figure 9 shows that Mg dissipates (fades) faster than do rare earths. This graph was generated by averaging the residual contents of Mg and rare earth (cerium [Ce] + lanthanum [La]) over all eight heats over time. A trendline was plotted on that average. The slope of the Mg trendline is steeper than that of the rare earth trendline. Gregg Study I, Mg and Rare Earth Fade y = x y = x Time After Inoculation, Minutes 1A7 Mg 1A7 Ce+La 1B3 Mg 1B3 Ce+La 1C3 Mg 1C3 Ce+La 1D3 Mg 1D3 Ce+La 2A3-1 Mg 2A3-1 Ce+La 2C2 Mg 2C2 Ce+La 2A3-2 Mg 2A3-2 Ce+La 2D2 Mg 2D2 Ce+La mg ave Ce+La ave Li (C L Fig. 9. The graph illustrates magnesium and rare earth fade over time, Gregg Study I. After this study was complete, it was decided that a second study should be done with a few modifications, such as: Increased sampling frequency to better determine the timing of the effects demonstrated in Figs. 4, 5, and 6 and Single inoculant at varied levels to allow for a more focused study. 49

8 Proceedings of the AFS Cast Iron Inoculation Conference, September 29-30, 2005, Schaumburg, Illinois Gregg Study II Figure 10 shows the drop in pouring temperature associated with the individual ladles. Figure 10, like Fig. 2, shows that temperature is measurably falling. Gregg Study II, Pouring Temperature Pour Temp 0.4 Pour Temp 0.6 Pour Temp 0.8 Pour Temp Time After Inoculation, Minutes Fig. 10. The graph illustrates the pouring temperature over time, Gregg Study II. Figure 11 offers greater insight into the immediate inoculant fade effect on TeLow when compared to Fig. 3. The ladle made with an addition rate of 0.4 wt% remained relatively steady. Beyond that, regardless of addition rate, TeLow drops immediately after inoculation as reflected in the first two minutes of all four curves in Fig. 11. The nature of the sampling regimen in Gregg Study I, with a six-minute gap between the first two samples, was such that it was unclear how rapidly this deterioration occurs. Figure 11 therefore offers strong evidence of the potent effect of late inoculation on carbide tendency. Later in the curves, TeLow seems to rebound and rise again, similar to the effect seen in Fig. 4. Gregg Study II, TeLow TeLow 0.4 TeLow 0.6 TeLow 0.8 TeLow Time After Inoculation, Minutes Fig. 11. The graph illustrates the TeLow over time, Gregg Study II. 50

9 Proceedings of the AFS Cast Iron Inoculation Conference, September 29-30, 2005, Schaumburg, Illinois Unlike Fig. 11, Fig. 12 shows varying effects on recalescence depending on inoculation rate. The two ladles made with the higher inoculation rates of 0.6 and 0.8 wt% show rising recalescence over time while the two ladles made with lower inoculation rates of 0.2 and 0.4 wt% more closely repeated the results found in the earlier study. Gregg Study II, R R 0.4 R 0.6 R 0.8 R Time After Inoculation, Minutes Fig. 12. The graph illustrates recalescence over time, Gregg Study II. Figure 13 mostly affirms the data in Fig. 5, with the iron (Fe) displaying a rise in GRF1 over time. Three out of four curves suggest that that more graphite will be precipitated if the pourer waits four minutes to pour the metal. This is slightly counter to the data in Fig. 11; after all, high TeLow suggests high inoculation effect and should mean more graphite is precipitated. However, Fig. 13 shows a rise in GRF1 at the same time as the initial inoculation effect has faded in Fig. 11. This could be the result of a loss of Mg over this initial period such that more carbon (C) is available to form graphite. The subsequent gradual deterioration in GRF1 might be the result of colder metal simulating a thin section behavior and further inoculant fade impeding general graphitization. Gregg Study II, GRF GRF1 0.4 GRF1 0.6 GRF1 0.8 GRF Time After Inoculation, Minutes Fig. 13. The graph illustrates the Graphite Factor 1 over time, Gregg Study II. 51

10 Proceedings of the AFS Cast Iron Inoculation Conference, September 29-30, 2005, Schaumburg, Illinois Figure 14 consistently affirms the results shown in Fig. 6. Regardless of inoculation rate, more late graphite is precipitated after the initial inoculation has faded. Less early graphite again means more available carbon units for late graphite formation. Gregg Study II, GRF GRF2 0.4 GRF2 0.6 GRF2 0.8 GRF Time After Inoculation, Minutes Fig. 14. The graph illustrates the Graphite Factor 2 over time, Gregg Study II. Figure 15 is consistent with both Figs. 7 and 14. Nodule count rises with time regardless of inoculation rate. The formation of late nodules is consistent with lower GRF2 values. The data suggest that waiting approximately four minutest pour a casting will result in a casting with more late-forming nodules, characteristic high nodule count and more resistance to microshrinkage. Gregg Study II, Nodule Count Nod Ct 0.4 Nod Ct 0.6 Nod Ct 0.8 Nod Ct Time After Inoculation, Minutes Fig. 15. The graph illustrates the nodule count, Gregg Study II. Figure 16 offers insight into Mg fade that was not offered in Fig. 8. Because samples were taken for a longer time period, loss in nodularity is clearly visible in the samples taken toward the right side of the graph. Nodularity improves over time in 52

11 Proceedings of the AFS Cast Iron Inoculation Conference, September 29-30, 2005, Schaumburg, Illinois the first ten minutes of the test and then deteriorates. This suggests that excess Mg results in substandard nodule shape in the very beginning of the heat, and as Mg fades, nodule shape improves. At approximately 10 min, the Mg has dissipated to a point where the metal reverts to non-spheroidal graphite shape. This graph therefore suggests that DI improves in nodularity over time until a threshold at which it deteriorates (in this case, ten min). In other words, the graphite is most spheroidal right before it fades precipitously. Figure 16 also shows that the ladle made with the lowest addition rate of inoculant suffers the most loss in nodularity and also the earliest loss in nodularity, supporting the idea that inoculant addition rate is proportional to the ability to maintain spheroidal graphite over time. The remaining three ladles are difficult to separate in terms of timing of nodularity fade. Gregg Study II, Nodularity Nodularity 0.4 Nodularity 0.6 Nodularity 0.8 Nodularity Time After Inoculation, Minutes Fig. 16. Graph illustrates nodularity, Gregg Study II. Figure 17 affirms the results of Fig. 8. Mg fades at a rate faster than do rare earths. Gregg Study II, Mg and Rare Earth Fade y = x y = x Mg 0.2 Ce+La 0.4 Mg 0.4 Ce+La 0.6 Mg 0.6 Ce+La Gregg 0.8 Mg 0.8 Ce+La mg ave Ce+La ave Linear (Ce+La ave) Linear (mg ave) Time After Inoculation, Minutes Fig. 17. Graph illustrates magnesium and rare earth fade over time, Gregg Study II. 53

12 Proceedings of the AFS Cast Iron Inoculation Conference, September 29-30, 2005, Schaumburg, Illinois CONCLUSIONS Figure 3 suggests that the effect of inoculant fade is drastic and occurs within the first six minutes after inoculation. This might or might not be detrimental to the foundry. Because the TeLow number has dropped some measurable amount, the possibility exists that it could drop below the white eutectic temperature, causing carbides. If the white eutectic is still below the TeLow value at the time of freezing, then the iron will remain carbide free, and any such drop in TeLow will be irrelevant. The inoculant fade effect shown in Fig. 3 was largely affirmed in Fig. 11. It was further revealed to have occurred in the first two minutes after inoculation. This suggests that the resistance to chill in a similar DI will drop more quickly than was detectable in Gregg Study I. The data suggest that the contribution to chill resistance by inoculation is strictly time limited. Figure 4 suggests that the metal is far more resistant to primary shrinkage from mold wall movement six minutes after inoculation. The metal is also more likely to have more late graphite expansion after six minutes because less early graphite allows for the possibility of more late graphite. Figure 12 shows mixed results. Two of the four curves affirm findings in Gregg Study I, showing a decrease in recalescence over time, therefore suggesting the metal is more resistant to shrink, caused by mold wall movement after a delay in pouring. The other two curves do not display this condition; however, they affirm the general idea that over-inoculation is detrimental to maintaining a desirably low recalescence. Figures 5 8 should be read together. As Mg content falls, there will be fewer Mg units in solution during solidification. This means the metal will be less prone to carbide formation. Because of this, carbon units that would ordinarily form carbides or even pearlite are now free to form nodules (hence improved GRF1). These nodules will be late in forming (hence improved GRF2). Because they form later, they will be relatively small (Fig. 7). Because they are smaller, they will be rounder. Figures13 16 are consistent with these findings. It is true that colder pouring temperature in and of itself should, logically, cause a microstructure that simulates a thin section with characteristic high nodule count and nodularity. However, the effects seen in Figs. 5 8 and are likely not simply the result of colder metal. If so, the GRF2 curve in Figs. 6 and 14 should be more linear, such as that found in Figs. 1 and 10. Instead, there is a clear kink in the curve shown in Figs. 6 and 14. Also, GRF2 is not simply a function of pouring temperature (Udroiu, 2002). The increase in nodule count over time seen in these studies is in direct contradiction to Fuller. He found that, Nodule numbers were highest immediately after inoculation (Fuller, 1979). In Gregg Study I, Inoculant D showed the best long-term performance for GRF2, while Inoculant B showed the worst. In Gregg Study I, Inoculant B showed the best long-term performance for TeLow, suggesting a benefit with this inoculant for late chill control. This is consistent with the Frush s findings (Frush,1998). The fade limit in Gregg Study II was determined to be approximately eleven minutes. Mg fades faster than do rare earths as shown in Figs. 9 and 17. This suggests that a treatment alloy richer in rare earth content would be more resistant to nodularity fade than one with lower rare earth content. This could be the result of either the rare earths own nodularizing effect or a synergy between rare earths and inoculant. The latter condition is described by Fuller. The inoculating effect and fading of inoculants is changed significantly in the presence of cerium in magnesium treated irons. Its effect is to increase nodule numbers after inoculation and significantly reduce the extent to which fading occurs (Fuller, 1979). It is clear that rare earths play a role both in nodularization and inoculation. Regardless, their presence is documented in this study to be more persistent than Mg. This study sees no relationship between high nodule count and inoculant fade. This is likely the result of the inoculant effect being overwhelmed by the effect of relatively high Mg levels at the beginning of a ladle. It is difficult to state this as the certain cause because it is not realistic to hold Mg steady while one analyzes the effect of inoculant fade alone. The traditional model of fade, depicted in Fig. 1, needs to be replaced. This study shows that microstructure at the beginning of the heat is inferior to that generated some minutes later. This model is not in contradiction to Frush and others, but rather offers data from the time period immediately after inoculation (Frush, 1998). The exact values depicted graphically in such a model will likely be different for every foundry because of varied manufacturing conditions, including, but not limited to, rare earth content. It could look much like the depiction in Figure

13 Proceedings of the AFS Cast Iron Inoculation Conference, September 29-30, 2005, Schaumburg, Illinois Proposed Fade Model Nodularity of Ladle of Iron Gray Threshold Time After Inoculation, Minutes Fig. 18. This graph illustrates a proposed nodularity fade model. REFERENCES 1. Frush, T., Lerner, Y. and Fahmy, M., Inoculants Selection for Counter-Gravity Casting of Thin Wall Ductile Iron, International Inoculation Conference Proceedings (1998). 2. Fuller, A. G., Fading of Inoculants, Proceedings of the Conference on Modern Inoculating Practices for Gray and Ductile Iron, pp (1979). 3. Goodrich, George M., Ductile Iron Casting Defects, Ductile Iron Handbook, p 233 (1992). 4. Gundlach, Richard B., Loper, Carl R, Jr. and Morgenstern, Bernardo, Composition of Ductile Irons, Ductile Iron Handbook, p 87 (1992). 5. Lee, R. S., Extended Holding of Treated Nodular Iron, AFS Transactions, pp (1971). 6. Olsen, S. O., Skaland, T. and Hartung, C., Inoculation of Gray and Ductile Iron A Comparison of Nucleation Sites and Some Practical Advises, 66th World Foundry Congress, pp (2004). 7. Skaland, T., Fading of Inoculation in Cast Iron, 1992 Casting Congress, Czechoslovakia (1992). 8. Udroiu, Adrian, The Use of Thermal Analaysis for the Process Control of Ductile Iron, Novacast ATAS User Conference (2002). 55

14 56

15 Proceedings of the AFS Cast Iron Inoculation Conference, September 29-30, 2005, Schaumburg, Illinois A New Approach to Graphite Nucleation Mechanism in Gray Irons Copyright 2005 American Foundry Society ABSTRACT Iulian Riposan, Mihai Chisamera, Stelian Stan, Torbojorn Skaland Politehnica University of Bucharest, Romania, Elkem, Norway A complex research program involving the Scanning Electron Microscopy (SEM) has been undertaken in order to achieve a more detailed understanding of graphite nucleation in un-inoculated and High Purity (HP)-FeSi, Ca-FeSi and Sr-FeSi treated irons at different residual aluminum(al) levels ( % ) in iron melt. (Mn,X)S compounds, usually less than 5.0 µm in size, with an average µm well-defined core (nucleus), were found to be important sites for graphite nucleation in gray irons. Al contributes to the formation of Al 2 O 3 -based sites which act as nucleation sites for (Mn,X)S compounds, even at a very low Al level (<0.003%) in iron melt. Increasing the residual Al content in iron melt helps the initiation of graphite nucleation at a lower undercooling degree. A % Al content in melt appears to be beneficial, without the detrimental effect on pinhole occurrence in gray irons. Calcium (Ca) is present in most (Mn,X)S compounds and at similar levels in the core and shell (higher level in Ca-FeSi treated irons), while strontium (Sr) was found mainly in the core and only in Sr-FeSi treated irons. A three-stage model for the nucleation of graphite is proposed: 1) small Al 2 O 3 -based sites are initially formed in the melt; 2) complex (Mn,X)S compounds nucleate on these microinclusions and 3) graphite nucleates on the sides of the (Mn,X)S particles which have a lower crystallographic misfit with the graphite. In inoculated irons, the (Mn,X)S compound is more complex, at a lower Mn/S ratio and a higher compatibility to graphite nucleation, especially as inoculating elements (Ca, Sr, etc.) contribute. Keywords: gray irons, (Mn,X)S, graphite nucleant, Al-key role, optimum Al-range, Ca/Sr distribution. INTRODUCTION The initiation of graphite nucleation during the solidification of all commercial cast irons generally requires a nucleation site with specific peculiarities in each case. Residual graphite should be an ideal nucleant for the formation of graphite during solidification, but sulphide and oxide/silicate microinclusions, formed in the molten iron (Fig. 1), are also possible sites for the heterogeneous nucleation of graphite. Due to their hexagonal crystallographic symmetry, their favorable wetting properties, a sufficient lattice disregistry acceptance and their high thermal and chemical stability, silicate particles are apparently the favored nucleation sites. The ability of sulphides to nucleate graphite is hindered because these substrates belong mainly to the cubic system. However, the nucleation potency of sulphides can be enhanced by inoculating elements (such as calcium (Ca), strontium (Sr), rare earth (RE), barium (Ba) etc.) which can transform MnS into complex sulphides (Mn,X)S. These might have a better lattice matching to graphite, a low coagulation capacity, good stability and adequate interfacial energy (Loper, 1998; Skaland, 1993; Stefanescu, 1998; Chisamera, 1998 and 2000). EXPERIMENTAL PROCEDURE Iron melts, obtained in an acid lined induction furnace, were designed having different contents of carbon (C), silicon (Si), manganese (Mn), sulphur (S), oxygen (O) and aluminum (Al) at different specific ratios, covering the most representative foundry situations (Table 1). A very low level of other elements was ensured for all irons. Experimental gray irons with typical Si concentration (1.2%) were conventionally or excessively inoculated (0.2% and 1.0% inoculant, respectively). In addition, other irons with low Si level ( %) were over-inoculated by the addition of 2.0%wt inoculant in order to facilitate the detection of the inoculating elements (Ca, Sr) in the possible graphite nucleants. High Purity (HP)-FeSi and Sr-FeSi or Ca-FeSi were used for ladle inoculation. The last two materials contain lower (0.01%), normal ( %) and intentionally higher (2% ) Al content, at representative inoculating element levels ( % Ca or Sr). Up to 40 ppm Ca or 30 ppm Sr were ensured in the inoculated irons. Major experiments were the analysis of cooling curves, chill tendency, graphite size and morphology; carbides occurrence; pearlite/ferrite ratio; eutectic cell count and size and the characterization of microinclusions as possible graphite nucleation sites (Chisamera 2000, 2001 and 2004; Riposan 2001, 2003 and 2004). 31

16 Proceedings of the AFS Cast Iron Inoculation Conference, September 29-30, 2005, Schaumburg, Illinois Compound Type CaO Ce2O3 SrO ZrO2 BaO MgO Al2O3 TiO2 SiO2 MnO CeS CaS SrS LaS BaS MgS ZrS2 TiS MnS Zr N TiN LaN CeN AlN Si3N4 Ca3N2 Sr3N2 Mg3N2 Ba3N2 ZrC TiC SiC CeC2 LaC2 Mn7C3 CaC2 BaC2 Al4C3. O X I D E S SULPHIDES NITRIDES CARBIDES Standard Free Energy, G 0,KJ/mole(O,S,N,C) Fig.1. This chart illustrates the standard free energy ( G o ) of the reactions for the formation of the compounds (T=1723K). The main objective of the present paper is the detailed characterization of the graphite nucleants (microinclusions) and a proposition for a new view on the graphite nucleation mechanism in gray irons. RESULTS PHYSICAL CHARACTERISTICS OF MICROINCLUSIONS (GRAPHITE NUCLEANTS) Depending on the degree of contact between the inclusions and the graphite, several situations were observed: a) no visible contact with graphite (particle embedded in matrix); b) superficial contact between inclusion and graphite; c) partially encapsulated by graphite and d) totally encapsulated by graphite. Figure 2 presents SEM micrographs illustrating these cases. The majority of the microinclusions, assumed to play a role in the nucleation of graphite, range in size between 1 and 8 µm, but is usually less than 5.0 µm in diameter. The size of the microinclusions depends mainly on the cooling rate (lower size at higher cooling rate), while inoculant type exerts a complex influence, according to other possible influence factors. In addition, it was found that the choice of inoculant also exerts an influence on the morphology of these compounds. Thus, the microinclusions have mainly an oval shape in the Ca-FeSi treated iron, a regular polygonal shape in Sr-FeSi treated iron and mostly an irregular polygonal shape in the HP-FeSi treated iron (Fig.3). It can be easily observed that most of these microinclusions possess a well-defined core (nucleus), measuring µm in diameter but most often µm (Riposan, 2003). The type of inoculant affects the size of the cores of the graphite nucleants. Thus, there is a continuous decrease in core size from un-inoculated (2.0 µm) to HP-FeSi and Sr-FeSi treated irons and to Ca-FeSi treated irons, the last having the lowest core size (<0.5 µm). The difference (ratio) between the size of microinclusions (graphite nucleants, D max ) and their corresponding nuclei (d max ) is higher for irons treated with FeSi-base alloys than for un-inoculated irons. Moreover, the D max /d max ratio tends to be higher for irons inoculated with HP-FeSi than for those inoculated with Sr and Ca-FeSi. Several typical nuclei morphologies were identified, such as regular and irregular plates; ovoid or rounded, triangular, irregular star and square features. However, a visible dependence of the nucleus shape on the inoculant type was not observed. 32

17 Proceedings of the AFS Cast Iron Inoculation Conference, September 29-30, 2005, Schaumburg, Illinois Table 1. Experimental General Procedure Parameters VARIABLE PARAMETERS Parameters Level 1. CAST IRON CHEMISTRY Equivalent Carbon % Carbon % Silicon % Manganese % Sulphur % Mn/S Ratio Aluminum % Oxygen ppm Nitrogen ppm Calcium ppm Strontium ppm 2. INOCULATION *Inoculant Chemistry Si,% Al,% Ca,% Sr,% High Purity (HP)-FeSi <0.06 <0.05 <0.02 Ca bearing-fesi Sr bearing-fesi *Inoculant Addition (wt.%) TESTS *Iron Chemistry Base and Minor Elements POLYVAC-spectral analysis O,S,N LECO-analysis Ca, Sr Plasma Emission-analysis *Inclusions/Graphite Nucleation Sites SEM/TEM/EPMA Techniques -Specific Analyses Size and Morphological Features, Chemistry and Distribution Pattern of Elements *Cooling Curves Features NI-DAQ / NI 4350 Instrument *Chill Tests Wedge Samples and Chill Test Samples *Optical Metallography Graphite size and Morphology Carbides occurrence Pearlite/Ferrite Ratio Eutectic Cell Count and Size G G a) b G G c Fig.2. Micrographs represent typical examples of inclusions/graphite particles distribution in gray irons (X-ray) (G Graphite): a) Compo image, b) MnKα and c) SKα 33

18 Proceedings of the AFS Cast Iron Inoculation Conference, September 29-30, 2005, Schaumburg, Illinois Fig.3. These are typical morphologies of nucleants: G Graphite; N Nucleant CHEMISTRY OF GRAPHITE NUCLEANTS (MICROINCLUSIONS) Microinclusions found in superficial contact with the graphite flakes, as well as microinclusions, partially or entirely encapsulated into the graphite flakes, were analyzed by SEM. The chemical composition distribution in different regions of their section surface, such as the nucleus (core), compound body (shell), compound/graphite interface and compound/matrix interface, was also measured by EDXA (Energy Dispersive X-ray Analysis). Manganese and Sulphur Complex (Mn,X)S compounds (Fig.2) were identified as major nucleation sites for graphite in all tested gray irons, independent of iron melt chemistry and inoculant type and/or addition rate. Mn and S were found as the basic components of all identified microinclusions, with a Mn/S ratio varying in the range. These compounds are referred to as (Mn,X)S, where X = iron (Fe), Al, Ca, Sr, Si, zirconium (Zr), titanium (Ti), phosphorus (P), etc Figure 4 illustrates the variation in chemical composition along a line through the core of sectioned (Mn,X)S particles in eutectic field gray irons (carbon equivalent [CE] = %). The data covers a composition range of % Mn and %S (Mn/S = ), for lower ( %) and higher ( %) residual Al content. Uninoculated and HP-FeSi, Sr-FeSi and Ca-FeSi treated irons (1.0wt% addition) are analyzed for both ranges of residual Al. The compositions of the core (nucleus) and the shell (body) of complex (Mn,X)S compounds are visibly different the core has high levels of Al and O, while the inclusion body has high levels of Mn and S. Other elements, such as Ca, Sr and Si present peculiar distribution. Aluminum, Silicon and Oxygen Strong deoxidizers, such as Al and Si, were found to be important parts of the first formed, very small, microinclusions in the iron melt, which was believed to be nucleation sites for precipitation of (Mn,X)S particles. As shown in Fig. 5, a visible Si-Al relationship was identified in these early formed microinclusions. At very low levels (<0.002 wt% Al) in the iron melt the Al and Si contents are similar (up to 2.0wt%). In contrast, at more than 0.005wt% residual Al in the iron, the Al content in the cores of (Mn,X)S compounds is more than ten times higher than that of Si (Fig.5a). On the other hand, for more Al or Si levels, more O content in this area of graphite nucleants was found. The highest O (>14wt%) content is typically for lower Al and higher Si (Fig.5b and 5c). 34

19 matrix matrix core matrix matrix matrix core matrix Proceedings of the AFS Cast Iron Inoculation Conference, September 29-30, 2005, Schaumburg, Illinois wt.% Al wt.% Al U.I Fe, Mn, S, wt.% Fe matrix Si Mn Si Fe S O Al Si S Mn Fe shell core Mn Al Sr S Distance, µm O shell O S graphite matrix O, Al, Si; wt.% Fe, Mn, S, wt.% matrix Si O Al Si S Mn Fe shell Fe Mn core O shell Distance, µm Al S graphite O, Al, Si; wt.% HP - FeSI Fe, Mn, S, wt.% O Al Si S Mn Fe O 80 shell shell Fe Mn 50 Mn Fe matrix S Al 5 10 Si core Distance, µm graphite O, Al, Si; wt.% Fe, Mn, S, wt.% O Al Si S Ca Mn Fe Fe shell core shell shell matrix Fe O O matrix Sr Mn Al Mn Al 5 20 Si 10 Si S Ca S Si Distance, µm graphite O, Al, Si, Ca; wt.% Sr FeSi Fe, Mn, S, wt.% matrix Mn O Al Si S Ca Mn Fe Sr shell Fe core Sr Si S Al Distance, µm O shell graphite O, Al, Si, Ca, Sr; wt.% Fe, Mn, S, wt.% Fe matrix O Al Si S Mn Fe Sr shell Sr core shell O Mn Distance, µm S Al graphite Si matrix O, Al, Si, Sr; wt.% Ca - FeSi Fe, Mn, S, wt.% Fe shell O Al Si S Ca Mn Fe O shell Fe O matrix Mn graphite 50 Sr 40 Mn S 5 30 Si 20 Ca 10 Si S Al Distance, µm 15 O, Al, Si, Ca; wt.% Fe, Mn, S, wt.% Fe matrix Si Mn shell O Al Si S Ca Mn Fe Fe Si Ca core Al O shell Mn Sr S Distance, µm O S graphite matrix O, Al, Si, Ca; wt.% Fig.4. Change of chemical composition of sectioned (Mn,X)S particles (wt%) are graphed. 35

20 Proceedings of the AFS Cast Iron Inoculation Conference, September 29-30, 2005, Schaumburg, Illinois Fig.5. Aluminium, silicon and oxygen relationships in the core of (Mn,X)S compounds are shown. : Key: Figs. 5 and 6 Fig.6. Al (shell) Al (core) relationships in (Mn,X)S compounds are shown. 36

21 Proceedings of the AFS Cast Iron Inoculation Conference, September 29-30, 2005, Schaumburg, Illinois It can be concluded that Si-Al-O and Al-O containing compounds are at the bases of the cores (nuclei) of (Mn,X)S compounds for the very low ( wt%) and the higher ( wt%) residual Al content in the iron melt, respectively. Important Al contents were detected in all microinclusions found to have a role in the nucleation of graphite. The Al is concentrated mainly in the core (nucleus) of these nucleants (ten times higher than in the shell), even at very low levels of Al ( %) in the iron melt. The Al (shell) /Al (core) ratio exceeds 0.2 (and can reach up to 0.8) only at very low residual Al in the iron melt, when important Si quantities were identified in the cores of (Mn,X)S compounds. In this case, the typical levels of Al are less than 2.0wt% in the core and 0.4wt% in the shell. Increased Al content in the molten iron leads to increased Al content in microinclusions, in both the core and shell sections up to 23 wt% Al in the core and 1.0wt% in the shell, for an Al (shell) /Al (core) ratio less than 0.2. Generally, lower levels of Al were found in Sr-FeSi inoculated irons than in Ca-FeSi treated irons (Fig.6). Calcium and Strontium As inoculating elements, Ca and/or Sr were detected with peculiar distribution pattern in different sections of (Mn,X)S compounds (Fig. 7 and 8). Ca is present in most (Mn,X)S compounds, including Sr-FeSi and HP-FeSi treated irons, while Sr has a detectable presence only in the Sr-FeSi inoculated irons. It was found previously that Ca had a quasi-homogeneous distribution over the whole particle cross-section, apparently at a higher level in the shell than in the core. In contrast, Sr was present mainly in the core and possibly in small amounts as solid solution in the (Mn,X)S compound body (Riposan, 2001). A more thorough analysis, using Environmental Scanning Electron Microscope-Energy Dispersive Spectrometry (ESEM-EDS), revealed a clearer difference between the shell and core concentrations and the shell/core distribution ratios for Ca as compared to Sr (Fig.7). Thus, Ca appears to be similarly distributed in the shell and core, with a Ca (shell) /Ca (core) ratio of In contrast, the Sr distribution pattern presents two specific fields. Approximately 75% of the compounds analyzed had less than 2.0wt% Sr in the core and up to 1.5wt% Sr in the shell, while others had up to 20wt% Sr in the core and less than 1.5wt% Sr in the shell. It was found that the Al concentration in the core is an important influencing factor on the Sr and Ca levels in different sections of the (Mn,X)S compounds (Fig.8). Higher Al content in the cores led to a decreased Sr concentration in the same region, resulting typically in less than 2.0wt% Sr. In contrast, more Sr is usually present at very low Al concentrations in the iron melt and the first formed microinclusion as well. At the same time, a higher Al content in the core is accompanied by a higher concentration of Sr in the shell (body) of (Mn,X)S compounds. In Ca-FeSi inoculated irons, the Ca concentration in both the core and the shell regions is favored by higher Al levels in the iron melt and in the core of (Mn,X)S compounds. Increasing the residual Al level in the iron melt appears to favor higher concentrations of Sr or Ca in the (Mn,X)S compounds. GRAPHITE NUCLEATION Based on the analysis of a large number of inclusions, the following three-stage model for graphite nucleation is proposed, for both un-inoculated and inoculated gray irons. Stage 1: Microinclusions containing strong deoxidizing elements, such as Al, Si, Mn, Ti, Zr etc, are formed in the melt. Stage 2: (Mn,X)S compounds nucleate at these oxide-type microinclusions. In un-inoculated irons, X is mainly Fe and to a lesser extent X = Ca, Al, Ti, etc., while in inoculated irons X = Ca, Sr, Al, RE, etc. A thin layer of silicate may form at the surface of the (Mn,X)S compound in the inoculated irons. Stage 3: Graphite nucleates at one or more (Mn,X)S compound sides. In un-inoculated iron, the (Mn,X)S compounds are simple and the crystallographic misfit between the graphite and the compound is relatively large. In inoculated iron (Mn,X)S compounds are more complex and the crystallographic misfit between the graphite and the compound is lower; thus, these compounds are better suited to graphite nucleation. Figures 2 and 3 show representative nucleation sites for flake graphite. Typical model for graphite nucleation in gray irons is illustrated in Fig.9. A peculiar behavior is shown in Fig.3d for Ca-FeSi inoculated iron graphite grew on the right side of an ovoid (Mn,X)S compound, while on the left side this process was only at an incipient stage in a porous area. 37

22 Proceedings of the AFS Cast Iron Inoculation Conference, September 29-30, 2005, Schaumburg, Illinois Fig.7. Distribution pattern of (a, b) Ca and (c,d) Sr in (Mn,X)S compounds are shown. Key: Figs. 7 and 8 Fig.8. Influence of Al on Sr (a,b) and Ca (c,d) distribution in (Mn,X)S compounds is shown. 38

23 Proceedings of the AFS Cast Iron Inoculation Conference, September 29-30, 2005, Schaumburg, Illinois Fig.9. This is a typical model for graphite nucleation in gray irons: (1) Al 2 O 3 nucleus core; (2) (Mn,X)S shell compound and (3) graphite flakes. SURFACE TENSION, σ, [Dynes/cm] Optimum Level OPTIMUM LEVEL %Al % Al H [Dawson, Smith] 0.6 σ [Hernandez, Wallace] H PINHOLES σ (σ<600) HYDROGEN PICKUP, H, [ppm] Aluminium in iron melt, % Fig.10. Optimum Al level ( wt.%) according to surface tension (σ) of gray irons (normalized to 1400 o C) (Hernandez 1979) and hydrogen (H) pick-up in green sand mold (Dawson 1956, Wallace 1989) is illustrated. AL KEY ROLE IN GRAPHITE NUCLEATION IN GRAY IRON As a residual element in molten iron, Al ( %), could have two effects in gray irons a) traces of Al can produce pinhole gas defects in iron castings (critical range % Al) and b) the presence of Al should influence the graphite nucleating capacity. It is generally considered that Al has virtually no inoculating effects as such and it is not included in the inoculating elements group. However, this research pointed out a key role that Al plays in the graphite nucleation process. It forms Al 2 O 3 -based sites, which act as nucleants for (Mn,X)S compounds. The later, in turn, are the major nucleation sites for the growth of graphite flakes. In addition, Al appears to favor the presence of Sr or Ca in the shell of the (Mn,X)S compounds (Fig.8). This also lowers the crystallographic misfit between the compound and the graphite. Under specific conditions, residual Al also causes the decreasing of the undercooling degree and the increasing of the eutectic cell count, and favors the formation of A-type graphite (Chisamera, 2004: Riposan, 2004). 39

24 Proceedings of the AFS Cast Iron Inoculation Conference, September 29-30, 2005, Schaumburg, Illinois In order to minimize the pinholing in green sand molds and to obtain an efficient graphitization process, an Al content of % is recommended (Fig.10). At this level, the surface tension of molten iron is high enough (Hernandez 1979). The hydrogen (H) pick-up from green sand mold is limited (Dawson,1956; Wallace, 1989). This optimum level of residual Al can be reached through all of the tested procedures as charge materials, Al-addition as a preconditioning agent or through FeSi-based inoculants. Raising the Al content in an iron is useful for un-inoculated gray irons or for inoculated gray irons when the inoculation response is inadequate (other process parameters at normal levels). If the inoculant performance is satisfactory and consistent at lower Al levels in the iron, an adjustment of the Al content could be considered to further improve iron quality or cost effectiveness. CONCLUSIONS Based on the SEM analysis of a large number of microinclusions, which are possible graphite nucleation sites, the following main conclusions can be drawn. (Mn,X)S compounds (where X = Fe, Al, O, Si, Ca, Sr, Ti etc) with different morphologies (regular or irregular polygonal shape or ovoid shape) are major sites for graphite nucleation in gray irons. These compounds are 1-8 µm in size (usually <5.0 µm) and have a well-defined core (nucleus), measuring µm (usually µm). It was found that Al has the ability to contribute to the formation of Al 2 O 3 -based sites suitable for the nucleation of (Mn,X)S compounds. This was observed even at very low Al (<0.003wt%), contents in the iron and especially in un-inoculated irons or under less efficient inoculation conditions. Higher Al levels in the iron result in higher concentrations of Al in the nuclei of these compounds and at a lesser extent also in their shells (bodies). Increasing the residual Al level in iron melt is useful to initiate the graphite nucleation process. A wt% Al range appears to be beneficial for graphite formation in gray irons (type A-graphite formation, no carbides, higher eutectic cell count, etc.) without the detrimental effect on pinhole occurrence. Under specific conditions (higher cooling rate, very low graphitizing potential of iron melt, etc.), Al between 0.01wt% and wt% appears to be necessary. Ca is present in most (Mn,X)S compounds, at similar levels in the core and shell. The Ca content is promoted by the presence of Al, especially for Ca-FeSi inoculated irons. Sr is present mainly in the core of (Mn,X)S compounds (more Sr in the core, lower Sr (shell) /Sr (core) ratio) only in Sr-FeSi inoculated irons. At higher Al content, the Sr concentration is lower in the core but higher in the shell. A three-stage model for the nucleation of graphite in both un-inoculated and inoculated gray irons is proposed: 1) small Al 2 O 3 -based sites are formed in the melt; 2) complex (Mn,X)S compounds nucleate at these microinclusions; 3) graphite nucleates on the sides of the (Mn,X)S compounds with lower crystallographic misfit with graphite. In inoculated irons, (Mn,X)S compound is more complex, at lower Mn/S ratio and at higher compatibility to graphite nucleation, especially as inoculating elements (Ca, Sr, etc.) contribute. REFERENCES 1. Chisamera, M., Riposan, I. and Barstow, M., The Importance of Sulphur to Control Graphite Nucleation in Cast Iron, AFS International Inoculation Conference, paper no. 3, Chicago (1998). 2. Chisamera, M., Riposan, I., Stan, S. and Skaland, T., Undercooling Chill Size-Structure Relationship in the Ca/Sr Inoculated Gray Irons under Sulphur/Oxygen Influence, 64 th World Foundry Congress, paper RO 62, Paris (2000). 3. Chisamera, M., Riposan, I., Stan, S. and Skaland, T., Cooling Curve Analysis of the Ca/Sr Over-Inoculated Gray Iron, at Lower Initial Silicon Content, International Conference on the Science of Casting and Solidification, pp , Brasov (2001). 4. Chisamera, M., Riposan, I.,Stan, S. and Skaland, T., Effects of Residual Aluminium on Solidification Characteristics of Un-Inoculated and Ca/Sr Inoculated Gray Irons, AFS Transactions, vol 112, pp (2004). 5. Dawson, J.W. and Smith, L.W.L, Pinholing in Cast Iron and its Relationship to the Hydrogen Pickup from Sand Mold, BCIRA, J.Res. and Dev. 6, p 226 (1956). 6. Hernandez, B. and Wallace, J.F., Mechanisms of Pinhole Formation in Gray Irons, AFS Transactions, vol. 87, pp (1979). 40

25 Proceedings of the AFS Cast Iron Inoculation Conference, September 29-30, 2005, Schaumburg, Illinois 7. Loper, C.R and Gundlach, R.B., Inoculation What Is It and How Does Inoculation Work? AFS International. Inoculation Conference, paper no. 1 Chicago (1998). 8. Riposan, I., Chisamera, M., Stan, S., Skaland, T., and Onsoien, M.I., Analyses of Possible Nucleation Sites in Ca/Sr Over-Inoculated Gray Irons, AFS Transactions, vol 109, pp (2001). 9. Riposan, I., Chisamera, M., Stan, S. and Skaland, T., Graphite Nucleant (Microinclusion) Characterization in Ca/Sr Inoculated Irons, International Journal of Cast Metals Research, vol 16, no.1 3, pp (2003). 10. Riposan, I., Chisamera, M., Stan, S., Gadarautanu, C. and Skaland, T., The Key Role of Residual Aluminum in Chill Tendency and Structure Characteristics of Un-Inoculated and Ca/Sr Inoculated Gray Irons, 66 th World Foundry Congress, pp , Istanbul (2004). 11. Skaland,T., Grong,O and Grong, T., A Model for the Graphite Formation in Ductile Cast Irons, Metallurgical Transactions, 24A, 2321 (1993). 12. Stefanescu, D.M., Inoculation of Thin-Wall Castings, AFS International Inoculation Conference, paper no. 16, Chicago, (1998). 13. Wallace, J.F. and Wieser, P.F., Pinholes in Gray Iron, Elemental Effects on Gray Irons, AFS Conference, paper II A 1 - A 14 Chicago (Sept 19-20, 1989). 41

26 42

27 RECOVERY OF MAGNESIUM IN A DUCTILE IRON PROCESS. S.O.Olsen and C.Hartung Elkem Foundry Products, Kristiansand, Norway Abstract. Residual magnesium and magnesium recovery have always been subjects for discussions amongst foundry people. This presentation summarises the most important factors that will influence the recovery and addition rate of magnesium in ladle treatment processes. Factors influencing the Magnesium Recovery and Addition. Sulphur and Oxygen in Base Iron. In order to evaluate the basics of ductile iron production the growth mechanism of the graphite has to be considered. It is proposed that growth normally occur along the pole of the plane with the lowest interfacial energy in contact with the melt. This will be the plane with the highest packing density and this will have the highest growth rate. With surface-active elements like O and S present the prism plane will grow fastest, but these elements are neutralised the basal plan will again have the highest growth rate. The highest growth rate from the basal plane will result in ductile iron and from the prism plane in grey iron. Hence magnesium is added in order to neutralise surface-active elements such as sulphur and oxygen. This means that increased content of sulphur and oxygen in base iron require higher addition of magnesium. (Fig. 1.) An example showing the effect of higher sulphur level in the base iron without increasing the addition rate of MgFeSi is presented in figure 2. Y:\Presentasjoner\New Presentation Folder\Mg Treatment\English\Papers\Recovery of Mg in a Ductile Iron Process.doc/ /egh

28 17. July 2003 S & O content in base iron %Mg Foundry Products Division %S & %O In Base Metal High S & O Base Metal Increased Mg- Addition S & O are surface and interface active elements and have to be neutralised Elkem Presentation Technical Information Sheet 23 Figure 1: Increased content of sulphur and oxygen in base iron require higher addition of magnesium. 17. July 2003 Example: Base Metal Sulphur Content Foundry Products Division Elkem Base S = % 1.0 wt% MgFeSi Final S = % Nodularity 50 % Presentation Technical Information Sheet 23 Magnesium vs.. Sulphur Base S = % 1.0 wt% MgFeSi Final S = % Nodularity 80 % Figure 2: Shows an example of the effect of 2 different S contents in base iron at the same addition rate of MgFeSi-alloy. The 0,018% S gives 50 % nodularity while the 0,010 % S gives 80 % nodularity.

29 Figure 3 shows the relation between S in base iron and the minimum residual Mg that is required to give ductile iron. Figure 3. Schematic representation of the relation between base iron sulphur content and required residual magnesium to produce ductile iron. Tapping Temperature. Tapping or treatment temperature should be kept as low as possible in order to avoid excessive reaction violence. The higher the temperature, the more vaporisation and lower recovery of Mg. The boiling point of pure Mg is about 1110 C and the normal treatment temperature in a foundry will be close to 1500 C. Nodulariser/MgFeSi addition to ladle. The time between magnesium addition and tapping should be minimised to prevent preheating and oxidation of the alloy. At the same time there should be no liquid

30 metal residual from previous treatments in the ladle as this may start to react with the alloy and lead to lower recovery of Mg. Slag. Slag carry over from melting and holding furnaces into the treatment ladle should be avoided. Slag that is transferred from furnace to the ladle will react with magnesium and reduce recovery. Proper separation procedures to minimise slag carry over need to be in place. Normally the furnace is properly skimmed before the first treatment, but the next treatments will suffer from slag contamination. Ladle Design. The ratio of internal Height : Diameter should be at least 2 : 1 and there should be an alloy pocket big enough to carry the alloy and covering material. The Mg recovery will increase with the height of the ladle because we increase the ferrostatic head before the reaction takes place. 17. July 2003 Ladle Design A good designed ladle gives the following advantages: Foundry Products Division 50-80% Mg - Recovery! High Recovery Consistent Reproducability Reliable No Flare 90% Fume Reduction No Metal splashing Minimum C and Temperature losses Good Economy Elkem Presentation Technical Information Sheet 23 Figure 4: Ladle design criteria with the advantages of a good ladle design indicated.

31 The ladle should also be properly insulated to minimise heat losses and consequently the required treatment temperature. A tundish cover lid is also highly recommended for magnesium and temperature yield reasons. Alloy Cover. An alloy cover (sandwich cover) in the ladle will delay the reaction start and give better absorption of magnesium into the liquid iron. As cover a fine sized FeSi are often used, but also steel plates or dry cast iron turnings could be used. The important thing is to use a cover and not so much what type of cover is used. Filling Time. Filling rate should be high in order to achieve a high ferrostatic head in the ladle before the reaction starts. A short filling time will also lead to reduced temperature loss and evaporation. Chemical Composition of Nodulizer. High magnesium content in the alloy will give a more violent reaction and reduced recovery. High calcium content will reduce the reactivity and increase the magnesium recovery, but it will also increase the tendency to slag formation. The rare earth metals (cerium) will assist in giving a better recovery because it allows for working at lower magnesium in the alloy and lower residual magnesium in the iron. Aluminium should be kept low in the alloy in order to reduce tendency to slag and dross formation. Alloy Sizing. A wide alloy sizing gives dense bulk packing in the ladle chamber. The alloy will then fuse and react slowly in a controlled manner with a minimum of pieces escaping. Pieces floating and burning on the surface are a waste. However grain segregation should be avoided since this can cause inconsistency in the production. Recommended sizing for small ladles is 1 10 millimetres and for bigger treatments 4 32 millimetres.

32 Pouring Time. Long pouring times require higher initial residual magnesium in order to compensate for fading losses during time. This means increased alloy addition and again reduced magnesium recovery. Higher addition in general will give lower recovery. Inoculation. With a good inoculation less residual magnesium is required to give good nodularity, and as a result a lower alloy addition can be used and thereby a better alloy recovery can be achieved. In many cases poor nodularity or degeneration of graphite nodules is incorrectly attributed to low residual Mg-levels when the cause is really insufficient inoculation. Slag in Ladle and Alloy Pocket. Slag building up in the ladle and alloy pocket leads to reduced magnesium recovery, probably due to reactions between the slag and magnesium. Overspill of alloy will occur if pocket is allowed to fill with slag and this can lead to floatation of alloy. Ladles should be kept tilted when empty to avoid slag clogging the alloy pocket and ladle walls. Storage of Foundry Alloys. All foundry alloys will contain a certain level of reactive elements. These reactive elements are necessary in order to give the wanted effect. This means that foundry alloys will oxidise if exposed to moisture. Oxidised alloys will give a lower recovery than fresh materials and heavy oxidation can result in up to 50 % reduction in the magnesium recovery. Containers of alloy should be stored in a dry place and not be opened until required at the treatment station. Large changes in temperature should also be avoided in order to minimise risk of condensation and transportation should be done in closed and watertight units.

33 Examples of how different Factors can influence the MgFeSi Addition. Required MgFeSi addition wt % S = 0.01 S = Sulphurcont. S = Tapping temp C 1480 C 1460 C lot of some clean furnace Slag in furnace. Time between MgFeSi add. and tapping 5 min 2 min 30 s Alloy Cover Ladle design diameter : height none some good 1:1 1:1,5 1:3 Ca= 0,5, Re = 0,5 Ca = 0,5 Ca = 1,0 Ca = 2,5 Ca = 2,5 Re = 1,5 Chemical composition of MgFeSi Oxidised Mg-alloy. heavy some fresh Figure 5: The diagram shows some examples how different factors can influence the MgFeSi addition in a ductile iron ladle treatment process. Residual Magnesium and Fading. The total analytical or residual magnesium content of liquid iron immediately after treatment is comprised of: Dissolved magnesium Micro inclusions of magnesium compounds Larger magnesium containing slag particles

34 Figure 6: Fading of magnesium during holding of treated ductile iron (left), and schematic representation of magnesium losses from a treatment ladle (right). These contributions to total (residual) magnesium will react in different ways during subsequent holding of the iron. The slag particles will float according to Stokes Law and move to the surface. Analytically we will see this effect as a fading of residual magnesium. In order to use residual magnesium as criteria for acceptance there should be a very strict sampling procedure. Only if samples are taken the same way every time will it be possible to compare the results. References. 1. T.Skaland: A model for the graphite formation in ductile iron, Ph.D. Thesis 1992: 33, The Norwegian Institute of Technology, Trondheim, Norway (1992). 2. R.Elliott : Cast Iron Technology, 1988, London, UK, Butterworths. 3. S.I.Karsay: Ductile Iron I Production, QIT M.Onsøien: Microstructure evaluation in ductile cast iron containing rare earth metals, Ph.D. Thesis 1997:115, The Norwegian Institute of Technology, Trondheim, Norway. 5. J.Nilsson: Slagginnesluninger i segjarn och gråjarn, Rapport , Svenska Gjuteriforeningen, Jønkøping, Sweden. 6. Gjutfelsanalys (Handbok). Mekanpublikasjon Sveriges Mekanforbund, Elkem Technical Information Sheets

35 1 An Alternative Route for the Production of Compacted Graphite Irons C.M.Ecob and C.Hartung, Elkem ASA, Norway Abstract. This paper reviews the properties desired from a compacted graphite iron (CGI), these being an intermediate between grey and ductile irons. The two commonly accepted production methods are discussed, undertreatment with magnesium and a Mg + Ti addition, which creates and then suppresses nodule formation. The advantages and disadvantages of these methods are reviewed and compared to a new alloy based process specifically designed for CGI production. The new process gives a wider production window and does not have the disadvantages of returns contaminated with titanium. Other considerations in the production of CGI, such as base metal composition, oxygen levels, inoculation and preconditioning are discussed. Introduction Changes to environmental legislation, predominantly in Europe, are affecting the way automotive manufacturers are planning car and truck design. By the year 2008, average emissions must be reduced from 180 grams of carbon monoxide per kilometre to 140 g/km and average fuel consumption must be reduced from 7 litres per 100 km to 5 l/100km. This means that fuel has to be burned more efficiently to generate increased power per litre of fuel. Inevitably this means that engines will have to burn fuel at hotter temperatures and therefore the engine must intrinsically have increased thermal stability and strength. One solution to meet these standards is to introduce a higher proportion of turbocharged diesel engines. Both grey iron and aluminium would struggle to meet the properties required, aluminium in terms of thermal stability, grey irons in terms of thermal conductivity and strength. Compacted graphite will certainly meet the desired mechanical and physical properties, provided that historic production difficulties, particularly section sensitivity, can be overcome. Compacted Graphite Iron (CGI) has been known now for many years, although it is only recently that the material has become accepted as a serious engineering material.

36 2 The production of ingot moulds, slag pots and casting trays has for many years been in compacted graphite irons, although these are all thick section castings without the intricacy of automotive casting design. In recent times, cylinder blocks, heads, brake drums and discs, manifolds, turbochargers and even piston rings have been produced in compacted graphite irons. Most of the major automotive manufacturers have either produced components in CG iron or are at the prototyping stage, although this varies across the globe and some producers are considering alternative methods of increasing vehicle performance. The list of components above have mainly been traditional grey iron castings, yet the foundries most suited to the production of CGI are ductile iron foundries, where low sulphur levels are normally more easily achieved. Until now undertreatment with MgFeSi or subverting the nodularity of a ductile type iron with additions of titanium, have been the most common production methods for CGI. The former has a dangerously narrow production window and the latter can lead to contamination of ductile/grey iron returns with undesirable titanium. In this paper, the desired properties of CGI are reviewed and a third method of production is described which give a greater production window and more consistency in the production process. Properties of CGI The properties of ductile iron are controlled by the matrix, whereas the graphite flake form and size govern the properties of grey iron. This determines such properties as the ductility in nodular irons and the thermal conductivity of grey iron. Compacted graphite irons make use of the vermicular graphite form to give properties intermediate between grey and ductile. An example of this is the thermal conductivity and is shown in Figure 1. Figure 1: Comparison of heat conductivity of grey, compacted (CGI) and ductile iron as a function of operating temperature. [1]

37 3 With CGI having a higher strength than grey iron, this has enabled thinner wall sections to be produced, partially explaining why the majority of castings switching to CG iron come from the grey iron sector. A general guide to the mechanical properties achievable compared to grey and ductile irons are given in Table 1. Relative damping capacity of grey, compacted and ductile iron is presented in Figure 2. Table 1: Comparison of mechanical properties of grey (GI), compacted (CGI) and ductile iron (DI). Matrix Tensile [MPa] Hardness [HB] Elongation [%] GI CGI DI Pearlitic Ferritic Pearlitic Ferritic Pearlitic Relative Damping Capacity Ductile CGI Grey 0 0,5 1 Figure 2: Comparison of the relative damping capacity of grey, compacted (CGI) and ductile iron. Production Challenges The principal challenge in producing a satisfactory compacted graphite iron remains the problem of section sensitivity. Most foundries will gauge the structure of the iron from a standard test bar, however in CGI; this is unlikely to reflect the actual properties of the casting. This is due to the unfortunate fact that if, for example, the test bar contains 100% of compacted graphite forms, thinner sections in the actual casting will contain a proportion of graphite nodules, whereas a thicker section may contain graphite flakes as found in a grey iron. This also assumes that the test bar is cast in the same moulding medium as the actual castings, again an influence often overlooked in foundries.

38 4 Figure 3 shows some microstructures containing various proportions of graphite nodules and the effect of increased nodule proportions is given in Figures 4, 5 and 6. Factors influencing the graphite shape formed during casting are discussed later in this paper. 5% 15% 30% 55% Figure 3: Various proportions of graphite nodules in CGI microstructures. Figure 4: Effect of increasing nodularity on some of the properties of Compacted Graphite Iron.

39 5 Figure 5: Effect of increasing nodularity on the mechanical properties of CGI. [2] Figure 6: Effect of increasing nodularity on the thermal conductivity of CGI for various operating temperatures. [2] Generally, CGI producers, and those starting in the production of CGI, will measure the percentage of nodules in the matrix and this method is widely accepted. An assumption is always made that flake graphite is not present other than at the surface. A new ISO standard is currently nearing completion where the structure will be classified based on nodularity. It may seem strange to describe a material based on unwanted features, but since all mechanical and physical properties have been linked to nodularity the standard will also be based on nodularity. The standard will cover 5 grades of CGI with tensile strength from 300 to 500 N/mm 2 and elongation from 2.5 to 0.5%. When only looking at the nodule count the shape, distribution and thickness of the compacted graphite is not fully taken into consideration when classifying CGI. Shape, distribution and thickness of the compacted graphite will however have a significant influence on the thermal and mechanical properties.

40 Elkem has developed a programme for measuring the compactness of an iron using image analysis techniques and this is shown in Figure 7. Whatever method is used to classify CGI it is however important to remember that this material will depend on visual control by trained personnel as there is no good method to differentiate between compacted graphite and degenerated graphite forms as chunky, exploded and flake just using image analysis. Roundness or aspect ratio is often used to classify the different graphite structures, but still a visual control would be needed to rule out flake graphite since this graphite form would be classified as compacted graphite using roundness. 6

41 Figure 7: Elkem Microstructure Report for CGI. 7

42 8 Production Routes. Compacted graphite irons may be produced from any one of several treatment methods. The most common are undertreatment by magnesium (compared to a ductile iron process) and by a Mg + Ti treatment which creates and then suppresses the nodules to the compacted form. In theory, treatment with cerium or nitrogen is possible, but the authors are unaware of any foundry making commercial castings by these processes and, for the purposes of this paper, they are ignored. Undertreatment with magnesium. Typical ductile iron will have a residual magnesium of % Mg, depending on the casting section thickness and the type of castings being made. For compacted graphite irons, the residual magnesium tends to be in the range of % and standard MgFeSi alloys can be used. This is, however, a difficult process to control with only a narrow residual Mg window giving satisfactory compacted structures, too high a Mg will give an excess of nodules whilst too low Mg will lead to the formation of grey iron flake structures, particularly in thicker sections. In castings of multiple section thicknesses, this method is practically impossible to control and is not widely used. The process becomes even more difficult with pure magnesium treatments, particularly wire, and often involves expensive trimming treatments, which incur royalty or licence fees. Magnesium plus titanium treatments. In this case, the iron is treated similarly to a ductile iron in terms of the magnesium addition, the difference being an addition of titanium to the process, either as an addition of FeTi or using the Ti as an integral part of the MgFeSi. A residual % Ti would be typical. This method gives a wider production window than the Mg undertreatment and reasonable CG structures can be obtained in both thin and thicker sections. The major disadvantage of this method, apart from the high costs of the treatment alloys, is the exceptionally poor machinability of the castings. A further concern is the contamination of returns with titanium making them unsuited for use in either grey (promotion of type D graphite) or ductile irons. Alternative methods for producing CGI are summarised in figure 8.

43 9 Figure 8: Different production methods for CGI. The Elkem alternative, CompactMag alloy Elkem has developed a new alloy for the production of CGI, which is free from harmful elements such as titanium and yet retains a good production window. The alloy contains 5-6% Mg and % Rare Earth (RE) in a normal ferrosilicon nodularising base alloy. It has previously been noted that magnesium is the common factor in the two described methods of making CGI and research by Elkem has shown that rare earth s have a beneficial effect on the section sensitivity, resulting in less variation of microstructure between thin and thicker sections. Rare Earth s are also easier to control than magnesium in that better and more predictable recoveries are obtained. Using this Mg + RE alloy, good CGI structures can be obtained using an alloy addition rate of only % as either a ladle or in-the-mould treatment. This compares to 1-2% MgFeSi + Ti in the Mg+Ti method or % MgFeSi in the undertreatment method using standard commercially available alloys. In either case this represents a substantial alloy cost saving. Whilst this figure will vary from foundry to foundry, an example of the treatment cost is given in Table 2. Table 2: Comparison of treatment cost per ton treated iron (2003 numbers). 1.3 wt% MgFeSi US$ wt% CompactMag TM alloy TM US$ wt% FeTi US$ 6 US$ 0.3 wt Inoculant US$ wt% Inoculant US$ 3 Total US$ 24 Total US$ 8 It is interesting to compare the Mg+Ti route and the undertreatment method to the Mg+RE (CompactMag TM alloy) route.

44 10 Figure 9 shows microstructures obtained with a 0.35% addition of a standard MgFeSi alloy (6%Mg, 1% RE) and CompactMag TM alloy. This shows that a 5mm section made with the standard MgFeSi contains predominantly nodules whilst a 35mm section from the same casting is mainly grey iron. The same addition of CompactMag TM alloy (0.35%) gives mainly compacted graphite in both sections, although inevitably more nodules are seen in the 5mm section. This is in line with the earlier discussion on the narrow production window with the magnesium undertreatment. 5 mm Section 35 mm Section MgFeSi with 1% RE Addition rate: 0.35 wt% MgFeSi with 1% RE Addition rate: 0.35 wt% CompactMag TM alloy CompactMag TM alloy Addition rate: 0.35 wt% Addition rate: 0.35 wt% Figure 9: Example comparison Mg undertreatment and CompactMag TM alloy. Table 3 shows a comparison of properties obtained in a foundry that ran the Mg+Ti and Mg+RE (CompactMag TM alloy) systems. The Mg + Ti additions were 1.3% MgFeSi and 0.5% FeTi compared to 0.35% CompactMag TM alloy in a sandwich ladle process.

45 11 Table 3: Example comparison Mg + Ti and CompactMag TM alloy. Example Example Property Yield Strength [MPa] Tensile Strength[MPa] Elongation [%] Grey Iron (ISO 100) Compacted by Titanium Compacted by CompactMag TM Ductile Iron (ISO ) min. 250 min min. 400 ca min. 15 Better yield strengths and tensile strengths are noted, whilst the lower alloy addition rates not only gave a huge cost saving, but far less slag on the surface of the metal as shown in Figure 10. MgFeSi with 1% RE: 1.5 wt% CompactMag TM alloy: 0.35 wt% FeTi: 0.25 wt% Figure 10: Slag on the surface of treatment ladle.

46 12 It is clear from the examples shown that a third commercially acceptable method of producing CG iron is available and this offers several advantages over the MgFeSi - undertreatment and MgFeSi+Ti route. No Ti contamination of returns. CGI returns can be safely mixed with ductile returns. Less Si introduction to the ladle allows for higher Si in the furnace leading to improved lining life. Minimum of slag and dross. Better machinability by avoiding hard titanium carbide and titanium carbonitride inclusions. Lower treatment costs. Greater flexibility due to wider production window compared to MgFeSi - undertreatment. Less section sensitivity compared to MgFeSi undertreatment. No royalty or licence fees normally found with wire adjustment systems.

47 13 Other considerations. Whilst a correctly applied Mg/RE treatment process offers the best option to the foundryman for the treatment process, some other factors have to be considered before good CGI can be produced. These include the iron composition, preconditioning and inoculation on the metallurgical concerns chill, shrinkage, microstructure and section sensitivity. Iron Composition As with the successful production of ductile irons, probably the most important consideration is the preparation of the base iron for subsequent treatment. Whilst the compacting process and inoculation are important, many foundries needlessly waste alloy and time trying to correct the iron after treatment when this can be done before the metal enters the treatment/casting cycle. When looking at the base iron composition it is most important to control the following three elements: % C % Si % S (preferred) All other elements have less importance, but should not be significantly higher than for ductile iron production. Generally a higher level of pearlite and carbide promoting elements can be tolerated, as long as the S-level in the base iron is kept low and the CompactMag TM alloy addition is kept below 0.40 wt%. After treatment the final iron composition should be in the following range: % C % Si % S (preferred) % Mg % Ce Typically the Mg- and Ce-content will be in the same range in the final iron. It is recommended to aim for low C- and Si-content in the final iron, because this will give a more consistent process although inoculation may be needed. Experimentation has shown that nodule count increases with increasing Si-content and it becomes difficult to get a good compacted graphite structure. Many foundries wish to use the same base charge for both ductile and compacted graphite irons, however it should be noted that the low addition rates of

48 14 CompactMag and Foundrisil inoculant required may necessitate additional silicon units being added to the base charge. Preconditioning Some foundries pre-treat their iron to create the same conditions prior to every treatment. This preconditioning can be a controlled introduction of either S and/or O. The most successful foundries now measure the base iron oxygen levels with an oxygen probe to determine active oxygen. Through additions of low stability oxygen source the level is maintained between ppm of total oxygen. One of the popular preconditioners is the Elkem product Ultraseed inoculant, which provides both oxygen and sulphur in addition to other nucleating elements at this vital stage of the process. The preconditioning with Ultraseed inoculant can be done either through addition to the furnace or to the stream as the iron is poured from the furnace to the tundish ladle or other treatment vessel. Care should, however, be taken as preconditioning can provide sufficient nuclei to generate excess nodules. It must be remembered that the use of Mg/RE CompactMag TM alloy is not as violent as typical ductile reactions and the destructive effect on potential nuclei is not as great. Sulphur Content in the Base Iron The base metal sulphur content plays a critical role in the production of CG iron. In many cases, castings being converted to CGI have traditionally been made in grey iron, hence the interest of some grey iron foundries in producing CG irons. This does require a change of thinking in such foundries to produce a base iron satisfactory for CGI. The sulphur content in the base iron should be in the range %. It is possible to produce good CGI with a base iron sulphur level as high as 0.02%, but generally the process becomes harder to control as the base iron sulphur level increases. While there seems to be a linear correlation between base iron sulphur and addition of CompactMag TM alloy needed for the sulphur range %, this is not the case for sulphur levels above 0.02%. Here it looks like there is an exponential correlation, hence more compacting agent will be needed and the compacting process becomes unpredictable. Figure 11 shows that CompactMag TM alloy will give highly satisfactory structures at the higher sulphur levels and that the addition rate of the alloy does not have to be increased significantly.

49 15 a) 0.30 wt% CompactMag TM alloy -addition b) 0.35 wt% CompactMag TM alloy -addition c) 0.35 wt% CompactMag TM alloy -addition d) 0.40 wt% CompactMag TM alloy -addition 0.01 wt% S a) and b) wt S c) and d) 0.02 wt% S e) e) 0.45wt% CompactMag TM alloy -addition Figure 11: CGI structure at different sulphur levels with CompactMag TM alloy. It is suggested that, due to alloy consumption, slag generation, chill promotion and process control concerns, irons of above 0.020% base sulphur are not suited to production of CGI.

50 16 Foundrisil Inoculant Sandwich Cover In the production of ductile iron, many foundries use a steel scrap cover in the treatment ladle to retard the onset of the magnesium reaction and enable them to get a greater head height of metal into the ladle. The low addition rate of CompactMag TM alloy, as previously described, has two major benefits when compared to higher addition rates of MgFeSi based alloys. Not only does the low quantity of alloy give a low reactivity, but the knock-on effect of not destroying the nuclei inherent within the melt. Replacing the steel scrap cover with a moderately powerful inoculant, such as Foundrisil inoculant, provides sufficient additional nucleation to minimise or eliminate the need for subsequent inoculation. Experience has shown that a 0.3% by weight addition of Foundrisil inoculant is an optimum addition. Under standard conditions, the 0.3% Foundrisil inoculant cover has been found to decrease the tendency to chill formation and to give better graphite compacts than the use of a steel cover. In some cases, particularly in thicker section castings, the need for post inoculation can be eliminated. While for chill prone section the addition of Foundrisil inoculant as cover for CompactMag TM alloy may need to be adjusted upwards or additional post inoculation has to implemented. Post-inoculation Inoculation of CGI has to be considered carefully. Whilst it is desirable to produce a structure free from iron carbides, inoculation will tend to promote the formation of graphite nodules. It is therefore recommended that inoculants of moderate potency be used. Interestingly, it has been found that inoculants generally associated with grey iron, e.g. Superseed inoculant, are very effective in CGI, as are the more common ductile inoculants. Typically, addition rates are intermediate between grey iron and ductile iron and are generally % for ladle applications, depending on metal temperature, fade time and casting thickness. With good preconditioning and perhaps the use of an inoculant base material as sandwich cover in the treatment ladle, it is often possible to eliminate totally the post-inoculation process. Fade time and treatment temperature Depending on casting and casting condition the fade time may vary from 5 to 20 minutes without a negative influence on the microstructure obtained with CompactMag TM alloy and Foundrisil inoculant cover. Treatment temperatures in the range 1400 C to 1520 C have been tested without any negative effect on the microstructure, but, as with all castings, the choice of post inoculant may have to be adjusted dependent on the final pouring temperature.

51 17 PQ-CGI - Process Control technology for production of Compacted Graphite Iron Elkem & NovaCast is a single-source supplier of technology, equipment, software, training and alloys for efficient, safe and economical production of compacted graphite iron. Why is a process control system needed? Compacted graphite (also called vermicular graphite) is an intermediate form between lamellar and spheroidal graphite. The requirements on process control are therefore much more pronounced because both a minimum and a maximum limit have to be considered. The process window for magnesium is very narrow, typically less than +/ %. However the crystallisation of graphite into a compacted shape is not only dependent of the level of magnesium but also on the metallurgical status of the base iron. Especially important is the nucleation status, the level of total oxygen, sulphur, nitrogen and the active carbon equivalent. Controlling the iron by spectrometer analysis is therefore not sufficient as it only shows the amount of each element but does not reveal anything about the metallurgical status. The illustration shows how the graphite shape varies with the magnesium level. If the amounts of nodules should be within 10-30% then the total process window represented as active magnesium is 0.003%. The probability to be within the CGI-window with normal variations in oxygen, nitrogen, sulphur, ACEL and nucleation and with traditional control methods is less than 80%. Thus in order to be able to produce high quality castings in CGI it is essential to have a metallurgically based process control system. Figure 14: Process window for magnesium by the production of compacted graphite iron. NovaCast has developed a special patented process control system for production of castings in compacted graphite iron. The system which is called PQ-CGI (abbreviation for Prime Quality Compacted Graphite Iron) has been enhanced in co-operation with ELKEM for use in combination with alloys specially developed for CGI. The system uses a combination of advanced quantitative

52 18 thermal- and chemical analysis to determine the metallurgical status of the base iron. Based on the analysis, which includes total oxygen, the computer system suggests how to condition the base iron in order to keep its status within a predetermined window. Once the base iron is within that specification, the system produces a recipe for additions of alloys to the treatment ladle. The recipe is thus optimised for the specific castings to be made and for the current status of the base iron. Several treatments can be made using the same recipe as long as the base iron remains the same. Thus it is a true one-step process. The PQ-CGI system also includes a thermal analysis system for verification (quality assurance) of the treated iron. The PQ-CGI Ladle system for batch treatment The PQ-CGI Ladle system uses two sampling stations. One is for metallurgical analysis of the base iron. The other is for testing and verifying the treated iron. The metallurgical analysis, combined with chemical analysis, is used in order to recommend conditioning of the base iron if needed until it is within a predetermined process window. Once the base iron is properly prepared then the PQ-CGI system recommends optimal additions to the treatment ladle in order to obtain the desired CGI structure and physical properties. The additions are specific to the current base iron. That means that the recipe can repeated until the furnace has been emptied or until there is a change in the iron. The system is adaptive which means that it is gradually fine-tuned for each specific casting by means of a learning algorithm. The one-step process means that the time from treatment to start of pouring is very fast. This minimises the need for over-heating and the time wait for iron after a stop in the moulding line. As the nucleation with the process is carefully controlled no extra inoculation is needed. Combined with special Elkem alloys such as CompactMag the fading is practical non-existing within 15 minutes after treatment. The section sensitivity is less than with conventional alloys. Figure 15: Example showing the process control window for PQ-CGI system.

53 19 The main technical advantages of the PQ-CGI Ladle process are: Compacted graphite iron can be produced with high repetition accuracy The process is controlled by a computer system, which logs all events The process allows very low magnesium levels to be used, which reduces the risk for shrinkage and dross problems The one step-process improves machinability of the castings The process reduces the section sensitivity and produces a more homogeneous structure The process shows very low fading which allows long times (<20 minutes) between the treatment with magnesium and pouring Extra inoculation is not needed The quality of a melt can be verified before pouring Figure 16: Flow-sheet for PQ-CGI Ladle process control system.

54 20 Summary. Although CGI has been known for a number of years, it is only now that castings are being produced in commercially interesting quantities. There are three principle routes for the production of CGI; a) Undertreatment with Mg, normally MgFeSi b) Suppression of the nodules to a compact form by using Mg + Ti c) The use of CompactMag TM alloy Mg/RE system. The latter system has been shown to have some significant advantages over the alternative methods, Greater production window and more flexibility. Low reactivity in the ladle, thus reducing the need for subsequent post inoculation. Low residual Mg and RE levels which reduces susceptibility to chill. Can be used over a range of sulphur levels within the normal limits for CGI production. Low slag generation. No contamination of returns with Ti. Used in conjunction with Foundrisil inoculant cover in the treatment ladle minimises the need for post inoculation. Long fade time. For further information on, please contact your local Elkem representative who will be able to demonstrate this total CG iron production package. Elkem & Novacast can help you set up a technical solution for production of CGI, which gives you an advantage when it comes to quality, environment and economy. Reference List [1] Mekanpublikation Mekanresultat Mars 1985 [2] BCIRA Broadsheet 253

55 Preconditioning of Gray Iron Melts using Ferrosilicon or Silicon Carbide M.I. Onsoien SINTEF, Norway Copyright 2001 American Foundry Society T. Skaland Elkem ASA, Norway ABSTRACT The scope of the current work is to quantify eventual effects of preconditioner choice on microstructure, mechanical properties and casting performance in gray cast iron melts produced under carefully controlled laboratory conditions. The experimental approach involves preconditioning of gray iron melts, produced according to the same charging/time and temperature/time program, with either high purity ferrosilicon (HP-FeSi), standard ferrosilicon (Std. FeSi), abrasive grade silicon carbide (Abr. SiC) or metallurgical grade silicon carbide (Met. SiC) as the preconditioning agent. In general there are only minor differences in microstructural features and mechanical properties in the experimental irons produced using the four different preconditioning agents. The carbon and silicon yield was found to be lowest for the irons produced with the abrasive grade SiC and highest for the irons produced using standard grade FeSi. The amount of graphite in thick section size specimens was found to be highest in the irons produced with metallurgical grade SiC and lowest in the irons produced with standard FeSi. In thin section specimens the amount of graphite was highest in the irons produced with FeSi and lowest in the irons produced with SiC. All examined thick plate specimens had type A graphite, size 3, with some minor type D graphite areas. The graphite was also mainly type A, size 4, in the thin plate specimens, small areas of type B and type D graphite were also present. The eutectic cell number is slightly higher in the irons produced with SiC than in those produced with FeSi. INTRODUCTION In cast iron carbon precipitates during solidification by a eutectic reaction either as the thermodynamically stable graphite phase (gray iron) and/or the metastable cementite phase (mottled or white iron). Whether the stable or the metastable phase forms depends on the nature and treatment given to the liquid, in particular, its graphitization potential, preconditioning, inoculation treatment and cooling rate. The achievement of the required properties in all cast irons, both as-cast and heat treated, depends largely upon adequate control of temperature, chemical composition and metal processing. This control, in turn, is primarily determined by the correct choice of raw materials for both the initial furnace charge mix and the subsequent preconditioning of the melt. Preconditioning is here regarded as a pretreatment of the gray iron melt in order to increase the response to the subsequent inoculation.(hugot 1988) Silicon increases the graphitization potential strongly and is always present in higher concentrations in gray irons. Silicon is normally added to the iron during melting either as ferrosilicon or as silicon carbide. Silicon carbide (SiC) is used as a preconditioning agent in both cupola furnaces as well as in induction furnaces for industrial production of gray cast iron. SiC is claimed to have some specific advantages as compared to other commonly used silicon and carbon sources. These advantages include higher eutectic temperature, higher liquidus temperature, higher number of eutectic grains and lower shrinkage porosity.( Scubert 1984, Venkateswaran 1989, Benecke 1994) Silicon carbide dissolves endothermly in the iron melt at a rate that is slower than that of ferrosilicon which dissolves exothermly.(benecke 1987, Kimstach 1992, Deng 2000) Benecke and co-workers have shown that metallurgical grade SiC performs better as a preconditioner than does the more pure abrasive grade SiC. They further reported that FeSi did not perform as good as SiC as preconditioner of small (0.5 kg) cast iron melts.(schubert 1984, Venkateswaran 1989, Benecke 1994, Benecke 1987) The scope of the current work is to quantify eventual effects of preconditioner choice on microstructure, mechanical properties and casting performance in larger gray cast iron melts (80 kg) produced under carefully controlled laboratory conditions. The experimental approach involves preconditioning of gray iron melts, produced according to the same charging/time and temperature/time program, with either high purity ferrosilicon (HP-FeSi), standard ferrosilicon (Std. FeSi), abrasive grade silicon carbide (Abr. SiC) or metallurgical grade silicon carbide (Met. SiC) as the preconditioning agent. In addition to the eventual effects caused by selection of ferrosilicon or silicon carbide this experimental matrix will also reveal how the changes in purity of the preconditioning agents will influence the properties of the produced gray irons. 1

56 MATERIALS AND EXPERIMENTAL PROCEDURE The experimental gray cast irons were produced in batches of 80 kg using a melt based on 70% steel and 30% recycled iron. The target analysis for the casting experiments was 3.3 %C, 2.0 %Si, 0.75 %Mn and 0.05 %S. Prior to the first experimental cast iron melt a dummy charge was melted in order to preheat the furnace and T-pot ladle. Insulating kaowool plates were kept on the top of the furnace and the ladle at all times during the experiments to limit the heat loss to the surrounding atmosphere. During charging, heating and melting the induction furnace followed the same power-time program as a means to keep the thermal history of the iron as controlled as possible in the in-furnace phase. The iron had a temperature of 1500 C when it was poured into the T-pot ladle. The iron was then transferred, in time intervals of 20s and in batches of 15 kg, into insulated cups made of alumino-silicate fibers for inoculation. Inoculant was placed in the cups prior to filling with molten iron. Irons that were not inoculated were also transferred in the same amount into cups in order to maintain similar thermal history in the two cases. After a holding time of two minutes the iron was poured into a sand mould for casting of specimens for metallographic examination as well as into a sand mould for casting of tensile bars. Coin shaped samples for analysis of chemical composition were quenched in a copper mould, after filling of the sand forms, using some of the remaining melt in the inoculation cups. Table 1 gives an overview of the casting experiments performed in the present work. Table 1. Overview of casting experiments. Experiment no Furnace Standard FeSi x x High Purity FeSi x x Metallurgical SiC x x Abrasive grade SiC x x Inoculation cup No inoculant 1a 2a 3a 4a 5a 6a 7a 8a Foundry grade FeSi 1b 2b 3b 4b 5b 6b 7b 8b RAW MATERIALS AND CHARGE ADDITIONS Ordinary St37 steel bars with a cross section of 100mm by 100mm were cut into pieces with a weight of 26.5 kg. Two of these bars were used in each furnace charge together with 23 kg iron returns. After charging the furnace with steel and iron returns, ferrosilicon or silicon carbide was added in an amount sufficient to reach the target composition. Desulco graphite was also added to the furnace in order to reach the desired carbon level. Both the silicon source as well as the graphite was added together with the steel and iron returns prior to turning on the furnace power. When the furnace charge was molten a small amount of FeS was added to the melt in order to reach the target sulfur concentration in the iron. Table 2 summarizes the chemical composition of the raw materials used in the experiments. Table 2. Chemical composition of raw materials. Material Element (in wt%) C Si Mn P S Al Ca Ti Fe Steel Bal. Iron returns Bal. Standard FeSi Bal. High purity FeSi Bal. Foundry grade FeSi Bal. Compound/Element (in wt%) Material SiC C Si+Si Si SiO2 Fe2O Al2O CaO O Metallurgical SiC Abrasive grade SiC CASTING CONDITIONS The experimental gray irons were cast into a sand mould to produce one thick plate (90mm long, 25mm wide and 20mm thick), one thin plate (90mm long, 25mm wide and 5mm thick), a standard chill wedge sample, a cross shaped sample (90mm long and 64mm wide) and a cylinder used for machining of tensile test bars. Figure 1 show photos of the model used to produce the mould. 2

57 Top view Figure 1. Model used to produce moulds for test casting of gray irons. Side view CHEMICAL ANALYSES Coin shaped samples for chemical analyses using XRF were quenched in a copper mould using melt remaining in the inoculation cups after casting of the samples for evaluation of the microstructure and mechanical properties. Carbon and sulfur were analyzed using a standard Leco procedure on samples extracted from the outer end of the 20mm thick plate. METALLOGRAPHIC EXAMINATION Samples for metallographic examination of the gray irons were extracted from a cross section cut at the center of both the thin plate and the thick plate. The metallographic samples were prepared according to standard metallographic techniques, i.e. polished to a 1µm diamond spray finish, for characterization of the graphite flake shape and size according to ASTM A247 plate II and III, respectively. Subsequently the polished samples were etched in 2% nital for quantification of the microstructure constituents such as ferrite, pearlite, carbide and graphite by means of point counting (minimum 500 points at a magnification of 200x) in the light microscope. The microstructure were also documented by micrographs taken at the same magnification. The eutectic cell count was determined after etching of the re-polished metallographic specimens using Stead s reagent (10 g. cupric chloride, 40 g. magnesium chloride, 20 ml hydrochloric acid, 1000 ml ethanol) for approximately 3 hours. Micrographs were taken at a magnification of 12.8x and the eutectic cell count is measured from the micrographs by comparison to the BCIRA Broadsheet 94-2 Comparator charts for counting eutectic cells. Chill formation The cast chill wedge samples were broken at the center and the size of the white iron zone formed at the tip of the sample is used as a measure of the chill forming propensity of the produced gray irons. Macro photos were taken for documentation purposes and also in order to directly compare the chill formation in the different specimens. Shrinkage porosity The cross shaped specimens were cut in a cross section through the center of the cross to evaluate pore forming tendency in the irons. The cutting of the specimens is illustrated in Figure 2. The specimens were ground down to 2400 grit paper finish. A cross is scratched into the polished surface to mark the middle of the specimen. Micrographs were taken at a 4x magnification such that each micrographs represent an area of 2.3x3.1 mm. Shrinkage porosity is measured by point counting in the light microscope using small grid which represents an area of 0.81x0.81 mm. Ten fields per specimen are measured giving a total measured area of 6.56 mm 2. cutting Figure 2. Cutting of cross shaped samples to evaluate porosity forming tendency. 3

58 MECHANICAL TESTING AND HARDNESS MEASUREMENTS Tensile testing of cylindrical bars with dimensions given in Figure 3 were performed in an Instron testing machine using a cross head speed of 4.0 mm/min. Brinell hardness was measured on each metallographic sample using a hardness tester with a 2.5mm steel ball indenter and a load of 62.5kp. R=20mm d=22mm d=16mm About 40mm Figure 3. Dimensions of tensile test specimen. RESULTS AND DISCUSSION CHEMICAL COMPOSITION With a carbon concentration in the range from 3.25% to 3.47% and a silicon concentration in the range from 1.93% to 2.19% carbon equivalents in the range from 3.9% to 4.2% were reached. This corresponds to slightly hypoeutectic to eutectic compositions, thus in some cases it can be expected to see traces of primary austenite dendrites in the microstructure. It seems from the analysis of the chemical composition that reproducible concentration of silicon is achieved when silicon is added both as ferrosilicon or as silicon carbide. There is however a weak tendency for lower silicon recovery when silicon carbide is used as the silicon source. Table 3 summarizes the composition of the produced gray irons. Table 3. Chemical composition of produced gray irons. Sample Si source Elements (in wt%) 1 C Si Mn S P Al Cu 1 Std. FeSi HP FeSi Met. SiC Abr. SiC Abr. SiC Met. SiC HP FeSi Std. FeSi Ce<0.005, La<0.005, Cr<0.022, V<0.005, Ni=0.05, Mo<0.004, Nb<0.005, Ti<0.010, Sn<0.004, Zr=0.008 MATRIX MICROSTRUCTURE The microstructure found both in the thick and thin plate consists mainly of graphite flakes in a pearlittic matrix. Very little ferrite was found, i.e., less than 6 percent in the thick plates and less than 8 percent in the thin plates. The graphite content in the thick plates varies within the range 8 to 14 volume percent, while the graphite content in the thin plates is a little bit lower and varies from 5 to 11 volume percent. Only minor differencies in the microstructure is seen when comparing the irons made with ferrosilicon and silicon carbide, also the use of high purity ferrosilicon and abrasive grade silicon carbide gives no significant change in the microstructure. Table 4 gives the detailed results from the quantitative microstructure analysis and Figures 4 to 7 show typical micrographs from the produced irons. 4

59 Table 4. Microstructure in thick and thin plates of gray cast iron (in vol. pct.). Sample Si-source Thick plate Thin plate Graphite Ferrite Pearlite Graphite Ferrite Pearlite 1a Std. FeSi a HP FeSi Inoculated Un-inoculated 3a Met. SiC a 1 Abr. SiC a 1 Abr. SiC a Met. SiC a HP FeSi a Std. FeSi b Std. FeSi b HP FeSi b Met. SiC b Abr. SiC b Abr. SiC b Met. SiC b HP FeSi b Std. FeSi No data available for thin plate due to poor filling of mould. Std. FeSi Met. SiC HP FeSi Abr. SiC Figure 4. Micrographs showing the microstructure in un-inoculated gray iron thick plate specimens. 5

60 Std. FeSi Met. SiC HP FeSi Abr. SiC Figure 5. Micrographs showing the microstructure in inoculated gray iron thick plate specimens. Std. FeSi Met. SiC HP FeSi Figure 6. Micrographs showing the microstructure in un-inoculated gray iron thin plate specimens. 6

61 Std. FeSi Met. SiC HP FeSi Abr. SiC Figure 7. Micrographs showing the microstructure in inoculated gray iron thin plate specimens. GRAPHITE CHARACTERIZATION In the thick plate specimens the graphite is of type A, i.e., the graphite flakes are randomly distributed and oriented throughout the matrix. However, most of the specimens also have colonies of type D graphite which is undercooled graphite where small flakes are located between austenite dendrite remainings. The graphite flakes are in general of size 3 according to ASTM A247 plate III, this means that the largest flakes have a length of 0.25mm to 0.5mm. In some cases, where type D graphite exist together with the type A graphite the graphite size for the type D graphite is also classified. Mostly the size of the type D graphite is classified as size 6, i.e., the largest flakes of the type D graphite is 0.03mm to 0.06mm. Some of the type D graphite are of size 7 where the largest flakes are 0.015mm to 0.03mm. Type A graphite is also the most predominant graphite type in most of the thin plate samples. Type B graphite is, in some cases, the major graphite type. Type B graphite is formed in irons of near-eutectic composition that solidify with greater undercooling than associated with type A graphite. Rosettes containing fine graphite, which are characteristic of type B, precipitate at the start of solidification. The heat of fusion associated with their formation increases the temperature of the surrounding liquid, thus decreasing the undercooling and resulting in the formation of type A graphite between the type B rosettes. Type B graphite is formed in one parallel of the un-inoculated iron made with standard ferrosilicon and in one parallel (both inoculated and un-inoculated) of the irons made with high purity ferrosilicon as well as in one of the uninoculated irons made with metallurgical SiC. Table 5 gives the detailed results from the graphite characterization. EUTECTIC CELLS The eutectic cell count in the thick plates are in the range from 169 to 311 cells per square centimeter. The iron produced with abrasive grade silicon carbide addition gives the highest cell count. The cell count in the thin plate are about five times higher than the cell count of the thick plates and ranges from 1221 to 1807 cell per square centimeter. Highest cell count in this case is obtained for the irons produced with abrasive grade silicon carbide (one parallel) and high purity ferrosilicon (one parallel). Inoculation gives as expected a systematically increasing cell count for all examined samples. Table 6 and Figure 8 summarize the data from the measurement of eutectic cell numbers in the produced irons. CHILL FORMATION The formation of chill on standard chill wedge samples does not seem to be sensitive to the selection of silicon source. The data given in Table 7 and Figure 9 shows that there is some scatter between the two parallel samples produced using the four different silicon sources. 7

62 Table 5. Graphite type and size in thick and thin plates of gray cast iron. Sample Si-source Thick plate Thin plate Graphite type Graphite size Graphite type Graphite size 1a Std. FeSi A (D) 3 (6) B 4 2a HP FeSi A (D) 2 (6) B (D) 3 3a Met. SiC A (D) 3 B (A,D) 3 4a 1 Abr. SiC A (D) 3 (6) - - 5a 1 Abr. SiC A (D) a Met. SiC A (D) 3 (6) A (D) 4 (6) 7a HP FeSi A (D) 3 (6) A (B) 3 8a Std. FeSi A (D) 3 (7) A (B,D) 3 1b Std. FeSi A 3 A (B) 3 2b HP FeSi A 3 B (A) 3 3b Met. SiC A (D) 3 A (B,D) 3 4b Abr. SiC A (D) 3 (7) A (B,D) 3 5b Abr. SiC A 3 A (B) 3 6b Met. SiC A (D) 3 (6) A (B) 3 (6) 7b HP FeSi A (D) 3 A (B) 3 8b Std. FeSi A (D) 3 A (B) 4 1 No data available for thin plate due to poor filling of mould. Un-inoculated Inoculated Table 6. Eutectic cell count in gray iron samples. Un-inoculated Inoculated Cell count [#/cm 2 ] Cell count [#/cm 2 ] Si-source Sample Thick plate Thin plate Sample Thick plate Thin plate Std. FeSi 1a b HP FeSi 2a b Met. SiC 3a b Abr. SiC 4a b Abr. SiC 5a b Met. SiC 6a b HP FeSi 7a b Std. FeSi 8a b No data available for un-inoculated thin plate sample due to poor filling of mould Cell count (#/cm 2 ) Un-inoculated Inoculated 0 Std. FeSi HP FeSi Met. SiC Abr. SiC Thick plate Thin plate Abr. SiC Met. SiC HP FeSi Std. FeSi Std. FeSi HP FeSi Preconditioner Met. SiC Abr. SiC Abr. SiC Met. SiC HP FeSi Std. FeSi Figure 8. Eutectic cell count in samples of thick and thin plates of gray iron. Table 7. Results from chill measurements on chill wedge samples. Un-inoculated Inoculated 8

63 Si-source Sample Chill (mm) Sample Chill (mm) Std. FeSi 1a 4.5 1b 2.0 HP FeSi 2a 5.5 2b 4.0 Met. SiC 3a 4.0 3b 1.0 Abr. SiC 4a 3.5 4b 3.0 Abr. SiC 1 5a - 5b 1.0 Met. SiC 6a 5.5 6b 2.0 HP FeSi 7a 3.5 7b 2.0 Std. FeSi 8a 4.0 8b No data available for un-inoculated sample due to poor filling of mould. SHRINKAGE POROSITY The shrinkage porosity of the experimental gray irons, measured as an area fraction on a polished surface at the center of the cross shaped specimen, is very low. Some pores were, however, found in all examined specimens, the highest amount porosity was 1.4 per cent and was observed in the iron produced with high purity ferrosilicon. Table 8 summarizes the results from the porosity measurements. 6 5 Chill (mm) Un-inoculated Std. FeSi HP FeSi Met. SiC Abr. SiC Abr. SiC Met. SiC HP FeSi Std. FeSi Inoculated Std. FeSi HP FeSi Met. SiC Abr. SiC Abr. SiC Met. SiC HP FeSi Std. FeSi Preconditioner Figure 9. Amount of chill on chill wedge samples from experimental gray irons. Table 8. Results from porosity measurements. Un-inoculated Inoculated Si-source Sample Average porosity (area%) Sample Average porosity (area%) Std. FeSi 1a 1-1b 1 - HP FeSi 2a b 0.75 Met. SiC 3a 1.4 3b 0.7 Abr. SiC 4a 0.5 4b 1 - Abr. SiC 5a 1-5b 0.5 Met. SiC 6a 0.9 6b 0.5 HP FeSi 7a 0.7 7b 0.95 Std. FeSi 8a 0.7 8b No results available due to poor filling of mould. MECHANICAL PROPERTIES The Youngs modulus of the produced experimental gray irons varies from 89 GPa to GPa. There is no systematic variation of the modulus with varying silicon source of the iron, even inoculation of the iron seems not to give any systematic change of the Youngs modulus. The yield strength varies from 135 to 172 MPa and the tensile strength varies from 172 to 9

64 209 MPa. There is not observed any dependency of tensile properties on silicon source of the gray iron. However, there is a slight tendency for better reproducibility of the tensile test data for the irons produced with high purity ferrosilicon and abrasive grade silicon carbide as silicon source as compared to the standard ferrosilicon and metallurgical silicon carbide grade. The strength of the irons, both yield strength and tensile strength, as well as the hardness are higher for the inoculated irons than for the not inoculated irons. This behavior can probably be ascribed to the increase of the cell count due to inoculation. The data from the tensile testing and hardness measurements are summarized in Table 9 and Figure 10. Table 9. Mechanical properties of produced gray cast iron. Sample Si-source Youngs Yield Tensile Hardness (BHN) modulus GPa strength MPa strength MPa Thick plate Thin plate 1a Std. FeSi a HP FeSi a Met. SiC a Abr. SiC a Abr. SiC a Met. SiC a HP FeSi a Std. FeSi b Std. FeSi b HP FeSi b Met. SiC b Abr. SiC b Abr. SiC b Met. SiC b HP FeSi b Std. FeSi No results available due to poor filling of mould. Un-inoculated Inoculated 250 Strenght or Modulus Std. FeSi Un-inoculated HP FeSi Met. SiC Abr. SiC Abr. SiC Youngs Modulus [GPa] Yield strength [MPa] Tensile strength [MPa] Met. SiC HP FeSi Std. FeSi Std. FeSi HP FeSi Preconditioner Inoculated Met. SiC Abr. SiC Abr. SiC Met. SiC HP FeSi Std. FeSi Figure 10. Mechanical properties of produced gray cast iron. SUMMARY The results from an analysis of the microstructure in gray irons produced with different silicon sources show that the iron matrix consists of pearlite with a small amount of ferrite (1-4 vol. pct.). In both in the inoculated and uninoculated samples of the thick plate the graphite is of type A with some type D graphite. In the inoculated thin plate the graphite is mainly of type 10

65 A with some type B and D graphite, while the graphite in the uninoculated thin iron is of type A and B with some D graphite. The eutectic cell count in the thick plates is in the range from 247 to 311 cells/cm 2 for the inoculated irons and in the range from 220 to 273 cells/cm 2 in the uninoculated irons. In the thin plates the cell count for the inoculated irons is in the range from 1495 to 1707 cells/cm 2, and in the not inoculated irons the cell count is in the range from 1302 to 1607 cells/cm 2. The yield strength is in the range from 153 to 171 MPa and the tensile strength is in the range from 191 to 202 MPa for the inoculated irons, while lower yield strength values (148 to 162 MPa) and tensile strength values (185 to 199) are obtained for the uninoculated irons. There seems to be no systematic changes in the microstructure and mechanical properties with changing silicon source. A summary of the experimental data is given in Tables 10, 11 and 12. Table 10. Summary of data for the thick plate specimens (average numbers). Chemistry Microstructure Graphite Si source %C %Si G F P Type Size Cell count BHN Std. FeSi A (D) 3 (6) HP FeSi A (D) 3 (6) Inoculated Met. SiC A (D) 3 (6) Abr. SiC A (D) 3 (6) Std. FeSi A (D) HP FeSi A (D) Met. SiC A (D) 3 (6) Abr. SiC A (D) 3 (7) Table 11. Summary of data for the thin plate specimens (average numbers). Chemistry Microstructure Graphite Si-source %C %Si G F P Type Size Cell count BHN Std. FeSi A,B (D) HP FeSi A,B (D) Inoculated Met. SiC A,B (D) 4 (6) Abr. SiC Std. FeSi A (B) HP FeSi A,B Met. SiC A (B,D) 3 (6) Abr. SiC A (B,D) No results available due to poor filling of the mould. Table 12. Summary of data for chill, porosity and strength measurements (average numbers). Si source Chill mm Porosity area% Youngs modulus GPa Yield strength MPa Tensile strength MPa Uninoculated Uninoculated Uninoculated Inoculated Std. FeSi HP FeSi Met. SiC Abr. SiC Std. FeSi HP FeSi Met. SiC Abr. SiC

66 CONCLUSIONS Based on the examination of gray cast iron specimens produced using either standard ferrosilicon, high purity ferrosilicon, metallurgical silicon carbide or abrasive grade silicon carbide as preconditioning agent the following main conclusions can be drawn with respect to microstructure, mechanical properties and casting performance: In general there are only minor differences in microstructural features and mechanical properties in the experimental irons produced using the four different preconditioning agents. The carbon and silicon yield was found to be lowest for the irons produced with the abrasive grade SiC and highest for the irons produced using standard grade FeSi. The amount of graphite in the thick section size specimens was found to be highest in the irons produced with metallurgical grade SiC and lowest in the irons produced with standard FeSi. In the thin section specimens the amount of graphite was highest in the irons produced with FeSi and lowest in the irons produced with SiC. All examined thick plate specimens had type A graphite, size 3, with some minor type D graphite areas. The graphite was also mainly type A, size 4, in the thin plate specimens, here small areas of type B and type D graphite were also present. The eutectic cell number is slightly higher in the irons produced with SiC than in those produced with FeSi. The chill formation in a standard chill wedge sample is virtually similar for irons produced using FeSi and SiC as preconditioning agent. No significant amount of shrinkage porosity was found in any of the cast specimens. The mechanical properties (Yield and tensile strength and hardness) is slightly higher for the irons produced using SiC. This is due to the lower Carbon recovery with SiC and the fact that lower C gives improved tensile properties. REFERENCES Benecke, T., Ta, A.T., Kahr, G., Schubert, W.D., and Lux, B., Auflöseverhalten und vorimpfeffekt von SiC in Gusseisenschmeltzen, Geisserei, vol. 74, no. 10, pp (1987). Benecke, T., Venkateswaran, S., Scubert, W.D., and Lux, B., The investigation of the influence of silicon carbide in the production of ductile cast iron, The Foundryman, Oct., pp (1994). Deng, D.X., Liu, W.T., Chen, M.J. and He, H.J., Experiment method of observing melting and dissolving process of ferrosilicon block used for in-mold inoculation, Foundry (China), vol. 49, no. 3, pp (2000). Hugot, A., Preconditioning Cast Iron Melts, Fonderia, vol. 37, no. 5-6, pp 31-32, (1988). Kimstach, G.M., Drapkin, B.M. and Zhabrev, S.B., Effect of ferrosilicon on structural transformations in cast irons, Metallovedenie i termicheskaya obrabotka metallov (Russia), no. 10, Sep.-Oct., pp (1992). Schubert, W.D., Ta, A.T., Kahr, G., Benecke, T. and Lux B., Influence of SiC additions on the microstructure of gray iron in Proc. of the third Int. Symp. on the physical metallurgy of cast iron, Stockholm, Sweden, Aug , pp (1984). Venkateswaran, S., Wilfing, J., Schubert, W.D., Lux B. and Benecke, T., Influence of SiC and FeSi additions on the microstructure, cooling curve and shrinkage porosity of ductile iron in Proc. of the fourth Int. Symp. on the physical metallurgy of cast iron, Tokyo, Japan, Sep. 4-6, pp (1989). 12

67 Proceedings of the AFS Cast Iron Inoculation Conference, September 29-30, 2005, Schaumburg, Illinois Copyright 2005 American Foundry Society Nucleation Mechanisms in Ductile Iron T. Skaland Elkem ASA, Foundry Products, Kristiansand, Norway ABSTRACT The present paper reviews different mechanisms for graphite nucleation in ductile iron (DI) and how these are affected by the inoculation process. Theories describing the fundamentals of graphite formation are given and the strengths and weaknesses of each theory discussed. Effects of key elements in the nucleation process, such as silicon (Si), calcium (Ca), strontium (Sr), barium (Ba), aluminum (Al), magnesium (Mg), cerium (Ce), sulfur (S), oxygen (O) and nitrogen (N) are described and discussed, and the importance of non-metallic heterogeneous compounds in the iron such as sulphides, oxides, nitrides and silicates are considered. Studies of nucleation and growth of graphite are shown, and the complex interaction between the Mg treatment and the inoculation process is described. The importance of the crystal structure and the stability of the nuclei to become a potent site for graphite formation are reviewed, and examples of potent and non-potent nucleation sites are shown. The paper arrives with a more comprehensive understanding of graphite nucleation in DI, and explains the key difference between Mg and the other three elements Ca, Sr and Ba in this respect. INTRODUCTION Ductile irons (DIs) are iron-carbon-silicon alloys where the chemical composition is adjusted to ensure that carbon (C) will precipitate as graphite spheroids during solidification. The C content is typically between 3-4% and the silicon (Si) content between 2-3%, which gives a eutectic solidification temperature of about 1165ºC (2129ºF). It is evident that one of the most important stages of the iron founding process is the economic production of liquid iron and its metallurgical treatments in preparation for pouring into the mold. This involves maintaining compositional and temperature control over the liquid during melting and holding in order to achieve the correct condition of the iron, the correct graphitizing potential and the correct state of the nodularizing and inoculation processes in order to ensure a sound casting of the desired structure and the required properties. Magnesium (Mg) is the most common spheroidizing element used in the DI production, and it is usually added in multicomponent alloy form with Si, calcium (Ca), rare earths, etc. Such alloys are balanced to reduce the reaction violence, to promote graphite spheroidizing, to neutralize the effect of impurities on graphite morphology and to control the matrix structure. The most common materials for nodularizing DI are ferrosilicon alloys containing about 45% Si, from 3-12% Mg and various levels of Ca and rare earths (cerium [Ce], lanthanum [La], etc.). Inoculation is a means of controlling the structure and properties of cast iron by minimizing undercooling and increasing the number of graphite nucleation events during solidification. An inoculant is a material added to the liquid iron just prior to casting that will provide a suitable phase for nucleation of graphite nodules during the subsequent cooling (Patterson, 1978). Traditionally, inoculants have been based on graphite, ferrosilicon or calcium silicide. The most common inoculants today are ferrosilicon based alloys containing small and controlled quantities of elements such as Ca, aluminum (Al), barium (Ba), strontium (Sr), zirconium (Zr), Ce, titanium (Ti), bismuth (Bi), etc. (Elliott, 1988). HETEROGENEOUS NUCLEATION THEORY Heterogeneous nucleation of graphite is an important aspect of cast iron metallurgy (Minkoff, 1983). The classic model for heterogeneous nucleation is shown schematically in Fig. 1. Here the graphite phase (G) grows from the nucleant (N), and the geometry of the graphite phase is a segment of a sphere of radius (r) and an angle of contact (θ). The interfacial energies between the three phases graphite (G), nucleant (N), and liquid (L) are γ GN, γ GL, and γ NL, respectively. The following relationship exists between the interfacial energies: γ cos θ + γ = γ Equation 1 GL GN NL 13

68 Proceedings of the AFS Cast Iron Inoculation Conference, September 29-30, 2005, Schaumburg, Illinois Fig.1. This is a schematic representation of heterogeneous nucleation. The change in free energy, G, accompanying the formation of a graphite nucleus with this configuration is given by: G = VG GV + AGLγ GL + AGNγ GN AGNγ NL = f ( θ ) πr G V + 4πr γ GL Equation 2 3 where V G is the volume of solid graphite, G V is the free energy of graphite formation, A GL and A GN are the area of the graphite-liquid and graphite-nucleant interfaces, respectively, and f (θ) is the so-called shape-factor, defined as: ( θ ) ( 2 + cosθ )( 1 cosθ ) 2 f = Equation 3 4 The critical radius of the stable nucleus, r*, is found by differentiating equation 2 with respect to r and equating to zero: 2γ GL sinθ r* = Equation 4 GV The corresponding value of the critical free energy barrier, G*, is then given by: 3 16πγ GL C1 G* = f ( θ ) = f ( θ ) Equation GV ( T ) where T is the undercooling, and C 1 is a kinetic constant which is characteristic of the system under consideration. When θ = 0 the graphite nucleus will completely wet the substrate, which implies that there is no energy barrier to nucleation. The nucleation rate N (the number of graphite nuclei formed per unit time and volume) is, in turn, interrelated to G* through the following equation (Elliott, 1988): ( G + G *) D N = ν N exp V Equation 6 kt where v is a frequency factor, N V is the total number of heterogeneous nucleation sites per unit volume, and G D is the activation energy for diffusion of atoms across the interface of the nucleus. Since G D is negligible compared with G* in liquids, the nucleation rate of graphite is determined by G*. The value of G* (or T) depends, in turn, on the crystallographic disregistry between the substrate and the nucleated solid. The disregistry can be defined as δ = ( a 0 /a 0 ) where a 0 is the difference between the lattice parameter of the substrate and the nucleated solid for a low-index plane, and a 0 is the lattice parameter for the nucleated phase. A mean factor representing planar lattice disregistry can be calculated as follows (Bramfitt, 1970): δ1 + δ 2 + δ 3 δ (%) = 100 Equation 7 3 where δ 1, δ 2, and δ 3 are the disregistries calculated along the three lowest-index directions within a 90º quadrant of the planes of the nucleated solid and the substrate. In practice, the undercooling, T, increases in a parabolic manner with increasing values of the planar lattice disregistry (δ) (as shown in Fig. 2) (Turnbull, 1952). Since the undercooling during solidification of DI varies typically from 2-10 C ([36-50ºF] depending on the section size) the results in Fig. 2 suggest that planar lattice disregistry between the inoculant and the graphite is in the order of 3-10%. (Minkoff, 1983). Such low values are characteristic of coherent/semi-coherent interfaces. 14

69 Proceedings of the AFS Cast Iron Inoculation Conference, September 29-30, 2005, Schaumburg, Illinois Fig. 2. The characteristic undercooling vs. planar lattice disregistryi is graphed. (From D. Turnbull and R. Vonnegut, Industrial Engineering Chemistry, vol 44, 1952) Some scientists have proposed that graphite formation in cast iron may be resulting from homogeneous nucleation. Turnbull, however, investigated the magnitude of undercooling, T, required for homogeneous nucleation by means of small droplet experiments (Turnball, 1952). The magnitude of required undercooling is found to be about 20% of the melting temperature before homogeneous nucleation occurs. This means that undercooling in excess of 250 C (482ºF) would be required in DI for homogeneous nucleation of graphite to be initiated. Consequently, homogeneous nucleation is rarely encountered in the solidification processing of such liquid irons. If homogeneous nucleation occurs in cast iron, this would happen anyway at undercoolings well below the metastable iron-carbide equilibrium temperature, thus resulting in fully carbidic microstructures. In conventional DI production, there will always be a number of non-metallic inclusions present in the treated (deoxidized and desulphurized) liquid iron as dispersed heterogenities throughout the metal volume. As described above, heterogeneous nucleation sites having the best planar lattice fit to graphite nucleate at only a very few degrees undercooling. Even heterogenities having a very poor crystallographic fit to graphite as well as non-crystalline (amorphous) heterogenities, eventually act as nucleation sites according to the Bramfitt model (Bramfitt, 1970). This occurs anyway at some 30 to 50 degrees undercooling at the maximum for the worst possible mismatch between the graphite and the heterogeneity. THEORIES FOR GRAPHITE NUCLEATION MECHANISMS Traditionally, cast iron inoculants are based on ferrosilicon, graphite or calcium silicide, the former being the most common (Patterson, 1978; Hughes, 1980). Since pure Si and ferrosilicon are found to be ineffective as inoculants, their nucleation potency depends on the presence of minor elements such as Ca, Al, Zr, Ba, Sr, Ti, etc. in the alloys (Dawson, 1961; Dawson, 1966; Kanetkar, 1984; Lownie, 1963; McClure, 1957; Mickelson, 1967). At present, the role of these minor elements are partly understood, but still complex matters related to formation of different types of nucleation sites in DI remains to be understood completely. Several theories that exist in the literature explain the phenomena of heterogeneous nucleation of graphite in solidifying cast iron. In the following, some of the most established theories are described and discussed. THE GAS BUBBLE THEORY According to Karsay, graphite tends to crystallize onto any given surface or imperfections such as cracks, pinholes, inclusions, etc. (Karsay, 1976). The gas bubble theory states that graphite can form only if its crystallization is protected by the presence of some sort of phase boundary. The needed phase boundaries are provided by the presence of carbon monoxide bubbles in the melt. The carbon monoxide bubbles are very finely dispersed in the melt and their size is less than 10 µm. Karsay presented the gas bubble theory as illustrated in Fig. 3. Karsay s gas bubble theory is in principle based on the presence of carbon monoxide bubbles (Karsay,1976). However, in industrial DI heats, strong deoxidizers, such as Mg, rare earths (RE), Ca, etc., are added that will effectively tie up and neutralize any oxygen (O) in the form of dissolved O or as CO gas. Various gases such as hydrogen (H), nitrogen (N), and CO are however found in DI castings as internal defects and voids. These are often covered on the inside by graphite linings. However it is highly unlikely that a complete graphite nodule will extend into the entire volume of a gas bubble, since this eventually would have to involve diffusion of C through the graphite shell. Under normal conditions, there should be no driving force for C diffusion through solid graphite. Partly solidified and quenched irons should then also reveal partly filled gas bubbles, which normally would never be observed in DI under any circumstances. 15

70 Proceedings of the AFS Cast Iron Inoculation Conference, September 29-30, 2005, Schaumburg, Illinois Fig. 3. Karsay s gas bubble theory is illustrated (A) gas bubble,(b) graphite, (C) melt and (D)austenite (From S. I. Karsay, Ductile Iron I: Production, Quebec Iron & Titanium Corp.,1976). THE GRAPHITE THEORY The early theories for heterogeneous graphite nucleation are based on the assumption that the graphite nucleation occurred epitaxially from other graphite particles contained in the iron melt (Boyles, 1947). Eash extended these ideas to Si-based inoculants by proposing that their effectiveness is due to the formation of Si-rich regions around the dissolving particles within which the solubility of C is sufficiently reduced to promote graphite precipitation (Eash, 1941). Later, Feest showed that this assumption is not correct, since the dissolution time of ferrosilicon in liquid iron is just a matter of seconds, and that graphite tends to form at the interface between the dissolving particle and the liquid (Feest, 1983). They therefore modified Eash s model by proposing that these seed crystals will be preserved in the melt down to the eutectic temperature, provided that Sr or Ba is present in sufficient amounts to prevent redissolution of the graphite ((Eash, 1941; Kayama, 1979). One weakness of the graphite theory and the assumption of small crystalline graphite particles, being preserved in the liquid iron for extended times, is the conflict with the well established fact that graphite in the form of crystalline recarburizers readily dissolves in liquid iron. Graphite recarburizers are typically added in sizes of millimeters, and will dissolve within seconds or a few minutes. Graphite based nucleation sites in a solidifying iron would be in the sizes of microns, and their dissolution time would consequently be very short. There is no question that graphite would be the ideal nucleation site for graphite itself. However, it can be argued whether the thermodynamic stability of micron sized graphite particles above the liquidus temperature would withstand its own dissolution characteristics for the entire fading time of inoculation. THE SILICON CARBIDE THEORY Following the dissolution of ferrosilicon in liquid iron, Wang and Fredriksson observed that silicon carbide crystals and graphite particles are formed in the melt close to the dissolving ferrosilicon particles (Wang, 1981; Fredriksson, 1984). They also observed that these transient particles redissolve readily after the inoculation treatment. No oxide or sulphide particles are detected. Based on their experimental observations, a theory developed and calculations were performed in order to explain the nucleation of graphite and the fading mechanism. A salient assumption in Wang and Fredriksson s model is the existence of an inhomogeneous distribution (local supersaturation) of C and Si in the melt subsequent to the SiC dissolution which provides the necessary driving force for homogeneous nucleation of graphite (Wang, 1981; Fredriksson, 1984). The fading effect is thus explained by a homogenization of the melt with respect to Si and C through convection and diffusion. One weakness of the SiC theory for graphite nucleation is that the recognized critical role of elements like Ca, Sr and Ba in the FeSi inoculant cannot be explained by this theory. Another weakness of the SiC theory is the assumption of local supersaturation of C and Si due to restricted convection and diffusion. Both C and Si are recognized for having very high diffusivity in liquid iron, and heat convection in hot metal is also recognized for being quite significant. It is therefore unlikely that dissolving SiC particles in liquid iron would be capable of maintaining a supersaturation of C and Si throughout liquid metal processing and into the solidification. Furthermore, the observation of SiC and graphite, surrounding a partly dissolved FeSi inoculant particle, is most likely resulting from the experimental quenching technique itself, forcing transient SiC and C out of solution in the Si-rich metal during quenching. 16

71 Proceedings of the AFS Cast Iron Inoculation Conference, September 29-30, 2005, Schaumburg, Illinois THE SALT-LIKE CARBIDE THEORY In a classical paper on the nature of the graphite nuclei, Lux considers both homogeneous and heterogeneous nucleation of graphite (Lux, 1964). He concludes that the elements Ca, Sr and Ba form salt-like carbides of the CaC 2 type in liquid iron, and that a direct epitaxial transition from the CaC 2 -lattice to the graphite lattice is possible without major changes in the lattice dimensions. Under such conditions, the interfacial energy between the nucleus and the substrate is sufficiently low to allow for extensive graphite nucleation at small undercoolings during solidification. The concept of the salt-like carbide nucleation theory by Lux is illustrated in Fig. 4 (Lux, 1964). Fig. 4. Epitaxial growth of graphite on a CaC 2 -crystal is illustrated.(from B. Lux, Modern Casting, vol 45, 1964). However, particles of CaC 2 have never been observed in the microstructure of inoculated DI. The thermodynamic stability of CaC 2 crystals as heterogeneous substrates having to survive in sulphur (S) and O containing liquid iron throughout holding and pouring is also highly questionable. In competition with available S and O in commercial irons, it is unlikely that the inoculating active elements, such as Ca, Sr and Ba, prefer to combine with C, forming such salt-like carbides. The sulphides and oxides of these elements are significantly more stable and thus more favorable than forming compounds with C. The salt-like carbide theory still offers an interesting approach from a crystallography point of view. It also attempts to give an explanation to the important role of Ca, Sr and Ba in the inoculation process. The theory is however questionable from a thermodynamic standpoint. THE SULPHIDE/OXIDE THEORY Several investigators have suggested that the graphite nucleation occur on sulphide, oxide or nitride particles, which are formed after the addition of the inoculant (Gadd, 1984; Jacobs, 1974; Muzumdar, 1972; Muzumdar, 1973; Naro, 1970; Sun, 1983). Lalich and Hitchings confirmed this hypothesis by demonstrating the importance of non-metallic inclusions (Lalich and Hitchings, 1976). They found that compounds of magnesium calcium sulphide act as nucleation sites for graphite nodules in DI treated with Mg ferrosilicon alloys. They concluded that the majority of nodules in DI are associated with nonmetallic inclusions and that graphite growth in some instances is also related to the shape and distribution of these inclusions. Inclusions in graphite nodules extracted from cast iron have been investigated by different techniques in order to determine the identity of the catalyst particles. These techniques include both electron diffraction pattern analysis and X-ray microanalysis (Deuchler, 1962; Rosenstiel, 1964; Zeedijk, 1965). The investigation by Jacobs is directed to determine the nature of nuclei and detect possible changes in their chemical composition and crystal structure after treatment of iron with Mg ferrosilicon (Jacobs, 1974). The subsequent inoculation treatment included the use of commercial Sr-FeSi alloy. Different series are carried out in order to clarify the effects of elements such as Al and Sr on the inclusion characteristics. These results are interesting for cast iron in general, since the examination revealed evidence of a duplex substrate structure consisting of a sulphide core surrounded by an oxide shell. The different constituent phases are found (Ca,Mg)- and (Sr,Ca,Mg)-sulphides in the core, and (Mg,Al,Si,Ti)-oxides in the outer shell. Moreover, Jacobs observed that inclusions embedded in the iron matrix contained the same constituent elements as those detected in the nodule centers, and that the typical size of the particles is about 1 µm (Jacobs, 1974). THE SILICATE THEORY In an investigation of the inoculation mechanisms in DI, Skaland put particular emphasis on the aspects of heterogeneous nucleation of graphite at inclusions (Skaland, 1993). It showed that the majority of the inclusions in ductile cast iron are primary or secondary products of the Mg treatment (e.g. MgS, CaS, MgO SiO 2, and 2MgO SiO 2 ). After inoculation with (X,Al)-containing ferrosilicon (X denotes Ca, Sr or Ba), hexagonal silicate phases of the XO SiO 2 or the XO Al 2 O 3 2SiO 2 type form at the surface of the oxide inclusions, probably through an exchange reaction with MgO. The presence of these phases enhances the nucleation potency of the inclusions with respect to graphite. In particular, the (001) basal planes of the 17

72 Proceedings of the AFS Cast Iron Inoculation Conference, September 29-30, 2005, Schaumburg, Illinois crystals are favorable sites for graphite nucleation, since these facets allow for the development of coherent/semi-coherent low-energy interfaces between the substrate and the nucleus. Figure 5a shows a TEM (Transmission Electron Microscope) examination of a silicate nucleus in DI. Figure 5b shows X-ray mapping images of the Mg, Ca, Al and Si distribution in an inclusion, while Fig. 5c shows a schematic representation of a heterogeneous nucleation site for graphite in DI. (a) (b) (c) Fig. 5. These depict (a) TEM examination of silicate nucleus in DI, (b) STEM X-ray images showing the distribution of Mg, Ca, Al and Si in inclusions after inoculation with a (Ca,Al) containing ferrosilicon and (c) schematic illustration of heterogeneous nucleation site in D. (From T. Skaland, Metallurgical Transactions A, vol 24A, 1993). Skaland also gives a theory for the fading mechanisms of inoculation (Skaland, 1993). This is explained by a general coarsening of the inclusion population with time, which reduces the total number of catalyst particles for graphite in the melt. A theoretical analysis of the reaction kinetics gives results which are in close agreement with experimental observations. NATURE OF NON-METALLIC INCLUSIONS Non-metallic inclusions of varying composition have been observed in the iron matrix and at the centers of graphite nodules by a number of investigators. Table 1 gives a summary of different element combinations and phases detected in DI inclusions. In the periodic table of elements, the group IIA-elements Mg, Ca, Sr and Ba are of specific interest in DI production, since they are all strong sulphide and oxide formers and are typically added deliberately through ferroalloys. In the following, possible reactions between these elements and C, Si, S, O and N are discussed. SULPHIDES The pure sulphides of the group IIA-elements are all of the face center cubic NaCl-structure type, and are characterized by similar lattice parameters and high melting points. In cast iron melts these sulphides are among the most stable non-metallic compounds. Hence, sulphides should form in preference to oxides. This conclusion is in close agreement with the results of Jacobs who found that the inclusions consisted of a sulphide core surrounded by an oxide shell (Jacobs, 1974). Sulphides of the group IIA-elements are also found to be a vital ingredient in the nucleus of graphite nodules by several researchers, as shown in Table 1. Table 2 gives a summary of crystal structures, melting points and standard free energies of formation for the group IIA-sulphides. From the numerous literature sources, there should be no question that sulphides of the group IIA-elements do exist in the core of graphite nodules in DI. Several scanning electron microscope (SEM) investigations have revealed in particular the presence of Mg and S in the core of graphite nodules. It is therefore reasonable to expect that MgS and also CaS and other sulphides are important ingredients in the heterogeneous nucleation sites for graphite. It is also recognized in the foundry industry that the addition of Mg to cast iron is contributing to desulphurizing and subsequently to the growth of nodular graphite morphologies. Table 2 shows that MgS and the other sulphides of the group IIA-elements, i.e. CaS, SrS, and BaS, have very similar crystal structures, lattice parameters and stability. Mg additions to cast iron are however not recognized for contributing to the nucleation of graphite nodules, while the other three elements Ca, Sr and Ba are very well recognized for being nucleating elements when added through ferrosilicon inoculants. Because of the physical similarities between the group IIA-sulphides, it is therefore unlikely that the existence of sulphides alone can explain the remarkable difference between Mg and the other three elements in the graphite nucleation process. 18

73 Proceedings of the AFS Cast Iron Inoculation Conference, September 29-30, 2005, Schaumburg, Illinois Table 1. Summary of Element Combinations and Phases Detected in Inclusions Phase Literature Reference MgS Askeland (1969), Askeland (1970), Jacobs (1976), Lalich (1976), Mercier (1969), Warrick (1966), CaS Jacobs (1976), Lalich (1976), Mercier (1969) SrS Jacobs (1976) CeS Warrick (1966) LaS Warrick (1966) MgO SiO 2 MgO SiO 2 2MgO SiO 2 (Mg,Al) 3 O 4 (Mg,Al)SiO 3 (Mg,Ca,Al)SiO 3 CaO Al 2 O 3 2SiO 2 Fe 2 O 3 Fe 2 SiO 4 Mg-Al-Si-Ti-O CeO 2 MgSiN 2 Mg 3 N 2 Mg 2.5 AlSi 2.5 N 6 Askeland (1969), Askeland (1970), Askeland (1972), Francis (1979), Heine (1966) Askeland (1970), Askeland (1972), Heine (1966), Heine (1966) Askeland (1970), Askeland (1972), Skaland (1993) Askeland (1969), Askeland (1970), Askeland (1972), Skaland (1993) Latona (1984) Latona (1984) Latona (1984) Skaland (1993) Askeland (1972), Francis (1979) Trojan (1968) Jacobs (1976) Francis (1979) Mercier (1969) Wittmoser (1952) Igarishi (1998), Solberg (2001) Mg 3 P 2 Wittmoser (1952) Table 2. Summary of Crystal Structures, Melting Points, and Standard Free Energies of Formation for the Group IIA-Sulphides (From I. Barin, Thermochemical Data of Pure Substances, VCH Verlagsgesellshaft, 1989) Phase Space group Crystal system Lattice parameter (Å) Melting point, T M ( C) Free energy, G F *) MgS Fm3m cubic CaS Fm3m cubic SrS Fm3m cubic BaS Fm3m cubic *) Standard free energy of formation at 1327C (1600K). CARBIDES The carbides CaC 2, SrC 2 and BaC 2 also reveal the NaCl-structure type and have similar lattice parameters. They are supposed to be metastable in liquid iron, but it is uncertain whether these phases can actually be formed from ferrosilicon alloy additions in liquid iron. In fact, and in opposition to sulphides and oxides, such carbides have never been detected experimentally as nucleation sites for graphite, although they, from a theoretical standpoint, are considered to be favorable (Lux, 1964). It is however interesting to note that Mg does not form any known compound of the MgC 2 -type. This might explain why Mg is only used as a spheroidizing agent in DI and not as an inoculant. Table 3 gives a summary of crystal structures, melting points and standard free energies of formation for the group IIA-carbides. Table 3. Summary of Crystal Structures, Melting Points, and Standard Free Energies of Formation for the Group IIA-Carbides (From I. Barin, Thermochemical Data of Pure Substances, VCH Verlagsgesellshaft, 1989) Phase Space group Crystal system Lattice parameter (Å) Melting point, T M ( C) Free energy, G F *) MgC CaC 2 Fm3m cubic SrC 2 Fm3m cubic BaC 2 Fm3m cubic *) Standard free energy of formation at 1327C (1600K). 19

74 Proceedings of the AFS Cast Iron Inoculation Conference, September 29-30, 2005, Schaumburg, Illinois OXIDES The group IIA-elements also form stable oxides in liquid iron; therefore, nodularizers and inoculants based on these elements are known to be effective deoxidizers. Table 4 gives information on crystal structures, melting points and standard free energies of formation for the pure oxides. As with the sulphides of group IIA, the oxides also show very little difference in crystal structure and stability between Mg and the other elements Ca, Sr and Ba. Thus, the pure oxides alone do not offer any good explanation as to why Mg is only active as a spheroidizing agent in DI, while the other elements are effective also as inoculating agents. Table 4. Summary of Crystal Structures, Melting Points, and Standard Free Energies of Formation for the Group IIA-Oxides (From I. Barin, Thermochemical Data of Pure Substances, VCH Verlagsgesellshaft, 1989) Phase Space group Crystal system Lattice parameter (Å) Melting point, T M ( C) Free energy, G F *) MgO Fm3m cubic CaO Fm3m cubic SrO Fm3m cubic BaO Fm3m cubic *) Standard free energy of formation at 1327C (1600K) In the following, different types of more complex oxide inclusions that contain Mg, Ca, Sr or Ba as constituent elements are considered. Since these elements are typically added via ferrosilicon alloys, a convenient basis for the discussion of oxide inclusions is the ternary systems, XO Al 2 O 3 SiO 2 where X denotes Mg, Ca, Sr or Ba. The specific interest in this respect is the MgO Al 2 O 3 SiO 2 and the CaO Al 2 O 3 SiO 2 ternary systems, since a variety of different phases (silicates and aluminates) may form, depending on the deoxidation and inoculation practices applied. THE MgO AL 2 O 3 SiO 2 SYSTEM Non-metallic inclusions containing MgO as one component may form during the Mg treatment. The pure MgO Al 2 O 3 SiO 2 system is similar to the MnO Al 2 O 3 SiO 2 and FeO Al 2 O 3 SiO 2 systems. A projection of the liquidus surface of the former system is given in Fig. 6. Fig. 6. Phase diagram for the system MgO Al 2 O 3 SiO 2 includes also projections of different liquidus surfaces. (From R. Kiessling, Non-Metallic Inclusions in Steel, book no. 194, the Metals Society, 1978). 20

75 Proceedings of the AFS Cast Iron Inoculation Conference, September 29-30, 2005, Schaumburg, Illinois It is evident that a number of silicate phases may form as a result of reactions between MgO and SiO 2, including: Mg + SiO2 MgO SiO 2 (enstatite) 2MgO + SiO2 2MgO SiO2 (forsterite) Enstatite can exist in three different modifications, and is a common reaction product in Mg treated DI slags (Askeland, 1972; Kiessling, 1978). MgO in MgO SiO 2 can be completely substituted by FeO, but pure FeO SiO 2 is not stable at normal pressures. MgO can also be replaced by CaO up to about 50 wt%. In addition, enstatite may dissolve as much as 10 wt% Al 2 O 3. Because of a faceted growth morphology, enstatite tend to form characteristic angular shaped inclusions in DI, which means that they can easily be identified by means of optical microscopy. The other Mg silicate type, forsterite, may also exist in liquid iron. This phase may contain varying amounts of other oxides in solid solution. For example, CaO is widely soluble in forsterite, since the crystal structure of 2MgO SiO 2 is similar to that of γ-2cao SiO 2. In addition to enstatite and forsterite, a variety of other phases have been detected in DI, including (Mg,Al) 3 O 4, (Mg,Al)SiO 2 and (Mg,Al,Ca)SiO 3 together with complex Ca and aluminum silicates and pure silica (Latona, 1984). THE CaO-Al 2 O 3 -SiO 2 SYSTEM Ca is, from a technical standpoint, insoluble in liquid iron. Nevertheless, a small solubility has been reported by Sponseller and Flinn, who found that pure iron could dissolve up to wt% Ca at 1600C (2912F) (Sponseller and Flinn, 1964). Slightly higher values were observed in the presence of Al, C, nickel (Ni) and Si. Ca is the most common trace element in ferrosilicon inoculants. Consequently, due to the low solubility and the high affinity to O, Ca could play a role in the graphite nucleation process by entering the deoxidation products at some later stage of the process. However, in steel inclusions, CaO is not present as a separate phase, since it reacts readily with other oxides to form complex calcium silicates and aluminates. This may also be the case in DI. Figure 7 shows the ternary CaO Al 2 O 3 SiO 2 phase diagram and projections of the different liquidus surfaces. Fig. 7. Phase diagram for the system CaO Al 2 O 3 SiO 2 includes also projections of different liquidus surfaces. (From R. Kiessling, Non-Metallic Inclusions in Steel, book no. 194, The Metals Society, 1978). Referring to the phase diagram, CaO can combine with silica according to the following reactions: CaO + SiO CaO SiO (wollastonite) 2 2 3CaO + 2SiO2 3CaO 2SiO2 (rankinite) 2CaO + SiO2 2CaO SiO2 (bredigite) 3CaO + SiO2 3CaO SiO2 (alite) Several modifications of the calcium silicates are also known, but the transformations between the different polymorphic forms are complex and not fully understood. 21

76 Proceedings of the AFS Cast Iron Inoculation Conference, September 29-30, 2005, Schaumburg, Illinois CaO SiO 2 may dissolve varying amounts of MnO, FeO and Al 2 O 3, but not MgO. Inclusions with a composition corresponding to CaO SiO 2 are commonly observed in steel deoxidized with CaSi. These inclusions may contain up to about 10 wt% Al 2 O 3 in solid solution (Kiessling, 1978). CaO can also combine with alumina to form a number of different phases, including: CaO Al 2O3 CaO Al2O3 Al2O3 CaO 2Al2 3 Al2O3 CaO 6Al2 3 2O3 3CaO Al2 3 CaO + 2 O CaO + 6 O 3CaO Al O Oxide inclusions containing CaO Al 2 O 3 and CaO 2Al 2 O 3 are all common deoxidation products. Some of the group IIA silicates and aluminates are summarized in Table 5. This table also includes data for the inclusion crystal structures, lattice parameters, and melting points (if known). Table 5. Selected Crystallographic and Thermodynamic Data for Some Possible Silicates and Aluminates in Liquid Iron Containing Mg, Ca, Sr and Ba (From I. Barin, Thermochemical Data of the Pure Substances, VCH Verlagsgesellshaft, 1989; R. Kiessling, Non-Metallic Inclusions in Steel, book no. 194, The Metals Society, 1978; R. H. Rein and J. Chipman, Transactions of the Metals Society, AIME, vol 233, 1965) Phase Space Group Crystal System Lattice Parameters [Å] T M [ C] G F *) [kj/mol] MgO SiO 2 Pbca orthorhombic 18.2/8.86/ MgO SiO 2 Pmnb orthorhombic 4.76/10.20/ CaO SiO 2 P1 hexagonal 6.82/ SrO SiO 2 hexagonal 7.127/ BaO SiO 2 hexagonal 7.500/ MgO Al 2 O 3 2SiO CaO Al 2 O 3 2SiO 2 P6 3 /mcm hexagonal 5.113/ SrO Al 2 O 3 2SiO 2 hexagonal 5.25/7.56 BaO Al 2 O 3 2SiO 2 hexagonal 5.304/7.789 (1380) MgO 6Al 2 O CaO 6Al 2 O 3 hexagonal 5.54/ SrO 6Al 2 O 3 P6 3 /mmc hexagonal 5.589/ BaO 6Al 2 O 3 P6 3 /mmc hexagonal 5.607/22.90 (1400) *) Standard free energy of formation at 1327C (1600K). Four intermediate ternary phases exist in the ternary CaO Al 2 O 3 SiO 2 system. Their stoiciometric compositions are as follows: CaO Al 2 O 3 2SiO 2 (anorthite) 2CaO Al 2 O 3 SiO 2 (gehlenite) 2CaO 2Al 2 O 3 5SiO 2 (corderite) 3CaO Al 2 O 3 3SiO 2 (grossularite) Bruch has studied the composition of Ca inclusions in steel after CaSi-deoxidation (Bruch, 1965). The mean composition of these CaO-containing inclusions is situated within the dotted area of Fig. 8. Crystalline ternary phases in the CaO Al 2 O 3 SiO 2 system are rarer than glassy phases. The most common of the crystalline phases is anorthite (CaO Al 2 O 3 2SiO 2 ), which is the only ternary phase within the dotted area of Fig. 8. Anorthite undergoes four different transformations down to room temperature and the stable high temperature modification is the hexagonal α-anorthite. Crystallographic data and melting point for the α-anorthite phase are given in Table 5. Note that crystalline ternary phases of the gehlenite, corderite and grossularite type are not common deoxidation products in liquid steel or cast iron. 22

77 Proceedings of the AFS Cast Iron Inoculation Conference, September 29-30, 2005, Schaumburg, Illinois Fig. 8. A summary of the mean composition (in wt%) of different calcia containing inclusions found in steel after CaSi-deoxidation is illustrated. (From R. Kiessling, Non-Metallic Inclusions in Steel, book no. 194,The Metals Society, 1978). NITRIDES In a study by Igarishi and Okada, the existence of a nitride phase, containing Mg, Al and Si taking part in the graphite nucleating process, is demonstrated (Igarishi and Okada,1998). In a recent study by Solberg, the crystal structure and composition of nuclei for graphite spheroids in DI containing small amounts of Mg and traces of Al are studied (Solberg, 2001). The particles are identified as Al Mg Si nitrides, having a trigonal super-lattice crystal structure derived from a hexagonal Bravais lattice, with parameters a = nm and c = nm. The parameters of the fundamental cell are a f = nm and c f = nm, deviating only 1-3% from the parameters of hexagonal AlN. Based on the compositional analysis, the chemical formula of the nitride is suggested to be Mg 2.5 AlSi 2.5 N 6. No such nitride phase containing all the elements Mg, Al and Si are known from earlier literature. Figure 9 shows a bright field image from TEM, a typical EDX (Energy Dispersive X-ray)-spectrum from the particle and the crystallographic unit cell for this nitride phase. Solberg found, however, that apart from the fact that graphite also has a hexagonal lattice, there is no obvious crystallographic similarity between the nitride particle and graphite (Solberg, 2001). Thus, the nucleating power of the nitride does not seem to be associated with its crystal structure, but with the fact that it is a heterogeneity in the melt. The simple existence of this Mg containing nitride phase may also explain why N gas defects are rarely observed in treated DI. The Mg addition assists in effectively neutralizing N. (a) (b) (c) Fig. 9. These illustrate (a) bright field image of nitride particle from TEM, (b) EDX-spectrum and (c) four super-lattice unit cells viewed along the (001)]-axis. (From J. K. Solberg and M. I. Onsoien, Materials Science and Technology, vol 17, 2001). 23

78 Proceedings of the AFS Cast Iron Inoculation Conference, September 29-30, 2005, Schaumburg, Illinois NUCLEATION OF GRAPHITE AT INCLUSIONS There seems to be general agreement in the literature that nucleation of graphite in DI occurs heterogeneously from particles contained in the melt. The problem has mainly been to determine the nature of these heterogenities, i.e. their origin, composition, surface characteristics, stability, etc. The SEM micrograph in Fig. 10 shows a graphite nodule associated with a non-metallic inclusion that contains a magnesium sulphide core and an outer shell of complex magnesium silicates. It should be noted that the presence of these phases is not a sufficient criterion for graphite formation, since Mg treatment of DI generally is not regarded as an efficient means of nucleating high numbers of graphite nodules. Modification of the inclusion surface chemistry by additions of minor elements through the inoculants is always required to achieve a high nodule density. Consequently, the key to a better understanding of the microstructure evolution in DI lies primarily in the recognition of the important difference between nodularizing (Mg treatment) and inoculation when it comes to graphite nucleation. GRAPHITE SULPHIDE NUCLEUS SILICATE (a) (b) Fig. 10. SEM micrographs show evidence of graphite nucleation at a complex duplex magnesium sulphide and silicate inclusion: (a) graphite nodule with nucleus in the core and (b) larger magnification of core. (From T. Skaland, Metallurgical Transactions A, vol 24A, 1993). BRAMFITT S PLANAR LATTICE DISREGISTRY MODEL As stated earlier in this paper, the interfacial energy at the nucleating interface (γ GN ) is the controlling factor in heterogeneous nucleation. For fully in-coherent interfaces, γ GN would be expected to be of the order of J/m 2. However, this value will be greatly reduced if there is epitaxy between the inclusions and the graphite nucleus, which results in a low lattice disregistry between the two phases. In general, assessment of the degree of atomic misfit between the graphite (G) and the nucleant (N) can be done on the basis of Bramfitt s planar lattice disregistry model (Bramfitt, 1970). ( d ) i cosα d i [ uvw] [ uvw] 3 1 N G δ = 100% 1 3 [ ] Equation 8 i= d i uvw N where [uvw] N = a low-index direction in (hkl) N ; [uvw] G = a low-index direction in (hkl) G ; d [uvw]n = the inter-atomic spacing along [uvw] N ; d [uvw]g = the inter-atomic spacing along [uvw] G ; α = the angle between the [uvw] N and the [uvw] G. In practice, the undercooling T (which is a measure of the energy barrier against heterogeneous nucleation) increases monotonically with increasing values of the planar lattice disregistry δ, as shown earlier in Fig. 2. This means that the most potent catalyst particles are those that provide a good epitaxial fit between the nucleant and the graphite embryo. The characteristic irregular shape of the inclusion in Fig. 5a indicates faceted growth morphology. Faceted growth occurs as a result of anisotropy in the growth rates between high-index and low-index crystallographic planes (Kurz and Fisher, 1989). If the former type of planes are the fastest growing ones, these planes grow out soon, leaving a faceted crystal delimited solely by low-index planes. 24

79 Proceedings of the AFS Cast Iron Inoculation Conference, September 29-30, 2005, Schaumburg, Illinois NUCLEATION OF GRAPHITE AFTER MAGNESIUM TREATMENT The probable crystal growth morphologies of enstatite (MgO SiO 2 ) and forsterite (2MgO SiO 2 ), as the main deoxidizing substrates formed after Mg treatment, are shown in Fig. 11. Included in Fig. 11 is also a sketch of the lattice arrangement at the interface between the (100)-plane of MgO SiO 2 and the (001)-plane of graphite. This orientation relationship conforms to growth of graphite along the pole of the basal plane perpendicular to the inclusion surface, which is the normal growth mode of graphite in DI. By only considering the position of the corner atoms in the orthorhombic unit cell (see sketch in Fig. 11c, the planar lattice disregistry between graphite, enstatite and forsterite is calculated from Equation 8 for a wide spectrum of orientation relationships. The results from these computations are summarized in Table 6. (a) (b) (c) Fig. 11. Diagrams illustrate nucleation of graphite at enstatite (MgO SiO 2 ) and forsterite (2MgO SiO 2 ) (a) crystal form of MgO SiO 2, (b) crystal form of 2MgO SiO 2 (From P. Ramdohr and H. Strunz, Lehrbuch der Mineralogie, Ferdinand Enke Verlag, 1978), (c) details of lattice arrangement at graphite/ MgO SiO 2 interface (From T. Skaland, Metallurgical Transactions A, vol 24A, 1993). Table 6. Calculated Planar Lattice Disregistry between Enstatite, Forsterite and Graphite for Different Orientation Relationships (From T. Skaland, Metallurgical Transactions A, vol 24A, 1993). Inclusion Phase Orientation Relationship *) Lattice Disregistry (100) I (001) G 10.2 % Enstatite (010) I (001) G 8.7 % MgO SiO 2 (001) I (001) G 5.9 % (110) I (001) G 12.3 % (111) I (001) G 10.1 % (100) I (001) G 9.9 % Forsterite (010) I (001) G 24.3 % 2MgO SiO 2 (001) I (001) G 15.5 % *) I = inclusion, G = graphite (101) I (001) G 25.5 % (111) I (001) G 29.7 % It is evident from the data in Table 6 that chances of obtaining a small planar lattice disregistry between graphite, MgO SiO 2 or 2MgO SiO 2 are rather poor, which means that the energy barrier against heterogeneous nucleation is correspondingly high. Hence, these phases, which are primary reaction products of the Mg treatment, would not be expected to act as favorable nucleation sites for graphite during solidification. This is also in agreement with general experience of Mg treatment not providing efficient nucleation of graphite. The Mg treatment however, provides an important basis for the subsequent inoculation. Formation of a high number of small magnesium sulphides, oxides, silicates and nitrides dispersed throughout the iron is expected to facilitate the formation of heterogeneous and potent nucleation sites that settle onto the surface of the Mg treatment reaction products. A calm and gentle Mg treatment is recognized to provide better conditions for the subsequent inoculation than a more violent and reactive treatment process. The treatment process reactivity decides whether numerous small and dispersed nucleation sites or, on the contrary, larger agglomerates of slag clusters are to be formed after the treatment. Thus, even though the Mg treatment does not provide potent nucleation sites on its own, it forms a very important basis for the subsequent inoculant to settle potent phases onto a highest possible number of sites initially produced during the Mg treatment. 25

80 Proceedings of the AFS Cast Iron Inoculation Conference, September 29-30, 2005, Schaumburg, Illinois NUCLEATION OF GRAPHITE AFTER INOCULATION During inoculation with Ca, Sr or Ba containing ferrosilicon, faceted hexagonal silicate phases of the XO SiO 2 or the XO Al 2 O 3 2SiO 2 type (X denotes Ca, Sr or Ba) may form at the surface of the Mg treatment inclusion products. In particular, the (001) basal planes of the crystals will be favorable sites for graphite nucleation, since these facets allow for formation of coherent/semi-coherent low energy interfaces between the nucleant and the graphite, as illustrated by the examples in Fig. 12. In fact, nearly all of the hexagonal silicate phases of the XO SiO 2 and the XO Al 2 O 3 2SiO 2 type, which may form at the surface of inclusions after the inoculation treatment, are effective catalyst substrates for graphite. Table 7 gives examples of calculated planar lattice disregistry between graphite and the different silicates that may form during inoculation. The low lattice disregistry explains why commercial inoculants for cast iron typically are based on either calcium, strontium or barium as the critical reactive ingredients of the ferrosilicon inoculant. (a) (b) Fig.12. These illustrate the details of lattice arrangement at a nucleus/substrate interface (a) coherent graphite/bao SiO 2 interface and (b) coherent graphite/cao Al 2 O 3 2SiO 2 interface. (From T. Skaland, Metallurgical Transactions A, vol 24A, 1993). Table 7. Calculated Planar Lattice Disregistry between Graphite and Different Inclusion Constituent Phases That May Form During Inoculation (From T. Skaland, Metallurgical Transactions A, vol 24A, 1993). Inclusion Phase Orientation Relationship *) Lattice Disregistry CaO SiO 2 (001) I (001) G 7.5 % SrO SiO 2 (001) I (001) G 3.5 % BaO SiO 2 (001) I (001) G 1.5 % CaO Al 2 O 3 2SiO 2 (001) I (001) G 3.7 % SrO Al 2 O 3 2SiO 2 (001) I (001) G 6.2 % BaO Al 2 O 3 2SiO 2 (001) I (001) G 7.1 % *) I = inclusion, G = graphite. EFFECTS OF Al, Ti, and Zr IN GRAPHITE NUCLEATION Smickley reported that Al additions had a mild effect as an inoculant and a stronger effect as a chill reducer (Smickley, 1981). McClure conducted experiments with gray irons (GIs) exploring the inoculating effect of two ferrosilicon qualities, one containing low levels of Al and Ca and the other one high levels of both elements (McClure, 1957). From these experiments it is found that the alloy with low levels of Al and Ca had, for all practical purposes, no inoculation effect, while the alloy with high levels of Al and Ca produced an iron with improved mechanical properties and reduced chill forming propensity. A variety of proprietary inoculants are developed in order to produce more effective inoculants and more consistent inoculation. These alloys normally, in addition to Al and Ca, contain controlled amounts of other elements. Higher levels of Al, up to 4% in ferrosilicon based inoculants, is also found to improve inoculation effectiveness in DIs and increase the ferrite content. Titanium is a commonly known additive to GIs for N scavenging. Titanium is also observed to cause a change in the hardness of iron castings (Merchant, 1972). Titanium has a very high affinity to O and S. Therefore, when adding Ti to cast iron for N control, an element that has a higher affinity for S and O than Ti does, should also be added. Narasimhan found that inoculation of GI with Ti effectively lowered the chill depth in low C equivalent irons (Narasimhan, 1969). Titanium bearing inoculants are generally not recommended for DI due to the risk of interfering with the formation of graphite nodules. The detrimental effect of Ti on the graphite morphology in DI is more pronounced as section size increases (Lownie, 1956; Watmough, 1971). Additions of Ce or mish metal to the iron may reduce or eliminate the detrimental effects of Ti on the 26

81 Proceedings of the AFS Cast Iron Inoculation Conference, September 29-30, 2005, Schaumburg, Illinois graphite structure in DI (Pearce, 1962). Both the graphite morphology and the mechanical properties of the DI may be restored and even improved when compared to irons without either Ti or rare earths (Naro, 1969; Sawyer, 1968). Several investigations have reported that N in high carbon flake graphite irons may make the graphite shorter, round the end of graphite flakes, hinder the growth of eutectic cells and stabilize the graphite, which are of benefit to some of the mechanical properties (Hu, 1994, Koshel, 1971, Mountfort, 1966, Naro, 1970, Ruff, 1976, Wallace, 1965, Wallace, 1975). Since the affinity between Zr and N is quite strong, very fine and dispersed nitrides are formed by adding Zr-bearing inoculants (Quian, 1985). These Zr-nitrides may act as nuclei for graphite during the solidification and thus prevent chill formation. The matrix of the iron will be strengthened by the dispersion of the hard ZrN particles as well. Dissolved N in GIs is controllable by the deliberate introduction of Zr, thus reducing the risk for N fissure defects in the castings. The possible beneficial effects of Zr in the DI nucleation process is however not well understood, since it is expected that the Mg treatment itself will tie up and neutralize free N in the base iron forming complex Mg-Al-Si nitrides as described by Solberg (Solberg, 2001). The formation of stable Zr-oxides as potent and additional nucleation sites for graphite, cannot however be ruled out. EFFECTS OF RARE EARTH METALS (REM) IN GRAPHITE NUCLEATION The effects of Ce and other rare earth metals (REM) on microstructure and mechanical properties of DI are widely studied. These studies generally involve the incorporation of REM in the MgFeSi nodularizer alloy. However, Amin conducted experiments with Ce added either as a constituent in pre-alloyed MgFeSi, added separately with the MgFeSi, added with the inoculant or added with both the MgFeSi and the inoculant (Amin, 1978). It was concluded from this extensive study of the behavior of REM in DI that the observed effects are independent of the method of addition. Previous investigations have shown that REM such as Ce and La can either have a beneficial or a detrimental effect on the microstructure and properties of DI, depending on experimental conditions and additions. For example, small additions of REM are frequently used to restore the graphite nodule count and nodularity in DIs containing subversive elements, such as antimony (Sb), lead (Pb), Ti, etc. (Bofan, 1984, Stefanescu, 1986, Udomom, 1985). On the other hand, REM in excessive concentrations may lead to problems with chill formation in thin cast products and chunky graphite in larger section irons, with subsequent degradation in the mechanical properties (Itofuji, 1990; Liu, 1989; Pan, 1994). Several investigators reported an optimum level of REM with respect to a high nodule count and reduced carbide forming propensity. However, the optimum rare earth content varies significantly according to different investigators. For example, Lalich concluded that the optimum Ce level is about % for low Ce rare earths, and about % for high Ce rare earths (Lalich, 1974). Kanetkar found a maximum nodule count at the following residual contents of REM: % praseodymium (Pr), 0.017% neodymium (Nd), 0.018% La, 0.02% yttrium (Y) and 0.032% Ce (Kanetkar, 1984). Onsøien reported an optimum Ce level of 0.035% and an optimum La level of 0.017% with respect to an optimum nodule count in low sulphur DI (Onsøien, 1997). It is expected that the formation of stable rare earth sulphides and oxides, and also oxy-sulphides play a very important role in the heterogeneous nucleation of graphite in DI. HIGH PURITY IRON CONDITIONS GRAPHITE NUCLEATION IN HIGH PURITY MELTS Heterogeneous nucleation of graphite at foreign particles is also reported in melts containing low levels of S and O (less than 0.2 ppm and 7 ppm, respectively). Dhindaw and Verhoeven studied vacuum melted high purity Fe C Si alloys produced from ultra-pure zone-refined iron (Dhindaw and Verhoeven, 1980). They found from extensive SEM examinations that impurity atoms are never detected in the nodule centers, which suggests that the graphite nucleation is not associated with sulphide inclusions. However, commercial Si/Ca/Al-inoculants effectively increased the nodule count in the ultra-pure zonerefined iron. It should however be noted that the maximum nodule count of about 26 per mm 2, as reported by Dhindaw and Verhoeven is rather low compared with that normally observed in commercial DI where the nodule number density often exceeds 300 to 400 per mm 2 (Dhindaw and Verhoeven, 1980). Consequently, their investigation is not conclusive in that it excludes the possibility of graphite nucleation on non-metallic inclusions. GRAPHITE GROWTH IN HIGH PURITY MELTS In a paper by Sadocha and Gruzleski, Fe C Si alloys, having various purities higher than those found for commercial irons, are investigated (Sadocha and Gruzleski, 1974). It is shown that with increasing purity there is a transition from a plate-like to a nodular morphology. Impurities present in commercial irons are thought to suppress nodular growth because they do not allow curved crystal growth of graphite to occur, and they change the nature of the austenite-liquid mushy zone so that 27

82 Proceedings of the AFS Cast Iron Inoculation Conference, September 29-30, 2005, Schaumburg, Illinois graphite cannot grow physically unhindered by austenite. Their conclusion is that nodules appear to be the basic form of graphite growth in pure Fe C Si alloys, and that the nodularizing elements act solely as scavengers to remove deleterious impurities such as S and O from the melt. In a recent paper by Nakae, the findings of Sadocha and Gruzleski were confirmed by applying special inert atmosphere and crucibles for melting cast iron (Nakae, 2004; Sadocha and Gruzleski, 1974)). The effect of cooling rate and S level on the graphite morphology is investigated. It is confirmed that the formation of spheroidal graphite needs a critical cooling rate and S level in Mg free base iron conditions. A specimen with 86 mass ppm S has no spheroidal graphite even at the cooling rate of 1000 K/min. Spheroidal graphite appeared in a specimen with 11 mass ppm S at the cooling rate of 100 K/min, while for specimens with 1.5 mass ppm S spheroidal graphite easily formed at the cooling rate of 40 K/min. This confirms that nodularizing additions are not required to form nodular graphite as long as the cooling rate is high or the S content is kept ultra low, below 11 mass ppm. However, Nakae also found that cementite structure prevailed for the ultra low S containing irons at higher cooling rates due to the difficulty of forming graphite nucleus (Nakae, 2004). This confirms the theory that removing impurities such as S and O are decisive in controlling nodular graphite growth, and that Mg additions as such are not directly required for this purpose. The key role of the nodularizing addition is to neutralize S and O and thus facilitate nodular graphite growth in an essentially S and O free environment. The findings by Nakae and Sadocha also confirm that S and O have a role in improving the nucleation of graphite by the formation of heterogeneous sulphide and oxide nucleation sites (Nakae, 2004; Sadocha, 1974). Figure 13 shows the 3-dimensional growth morphology of graphite from SEM investigations (Nakae, 2004). With reducing concentrations of S from 98 to 1.5 mass ppm, the graphite morphology changes from flake to nodular growth. Fig. 13. Growth morphology of graphite with reducing sulphur concentrations shows the transition to nodular growth without introduction of any nodularizer to the iron. (From H. Nakae, S. Jung, H. Inoue, H. Shin, Proceedings of the 66 th World Foundry Congress, 2004). CONCLUSIONS The following summary is given from the present paper: Although the subject of nucleation and growth of graphite nodules in DI has long been a topic of considerable discussion, conflicting views are still held about the major controlling mechanisms. This situation calls for a closer review and examination of the existing theories. Several theories have been developed in the past to explain the nucleation of graphite nodules during solidification of DI, including the gas bubble theory, the graphite theory, the silicon carbide theory, the salt-like carbide theory, the sulphide/oxide theory, the nitride theory and the silicate theory. These theories are mostly based on the assumption that the graphite is formed as a result of heterogeneous nucleation events occurring during solidification and that minor elements such as Ca, Ba, Sr, Al, Zr, Ti, and Ce play an important role in this nucleation process. The effectiveness of a substrate in promoting heterogeneous nucleation depends on the crystallographic disregistry between the substrate and the graphite to be nucleated. In practice, the undercooling, T, increases with increasing values of the planar lattice disregistry. Since the undercooling during solidification of DI is very small, the lattice disregistry between the nucleus and the graphite phase must also be small and comparable with that of coherent/semicoherent interfaces. A variety of different inclusions (sulphides, oxides, nitrides, and silicates) can form in the liquid state. The sulphides and oxides of the group IIA-elements (Mg, Ca, Sr and Ba) are all of the face centered cubic NaCl-structure type, and are 28

83 Proceedings of the AFS Cast Iron Inoculation Conference, September 29-30, 2005, Schaumburg, Illinois characterized by very similar lattice parameters and high melting points. The similarity between these phases does not explain why Mg is not acting as a potent nucleating element for graphite in DI, while the other three elements Ca, Sr and Ba will provide potent nucleation effects. Additions of Mg, Ca, Sr, or Ba to Si-rich iron melts may also result in the formation of complex silicates with different stoiciometric compositions. After Mg treatment, a wide spectrum of inclusions is present in the melt, including Mg silicates of the enstatite and forsterite type. These phases, however, will not act as potent nucleation sites for graphite during solidification because of their non-hexagonal (orthorhombic) crystal structures and large planar lattice disregistries with the hexagonal graphite. After inoculation with Ca, Sr or Ba (and Al), containing ferrosilicon, faceted hexagonal silicate phases of the XO SiO 2 or the XO Al 2 O 3 2SiO 2 type may form at the surface of existing inclusions from the Mg treatment. Such phases will be favorable sites for graphite nucleation, since the hexagonal facets allow for the formation of coherent/semi-coherent low energy interfaces between the nucleant and the graphite. REFERENCES 1. Amin, A. S., Loper, C.R., AFS Transactions, vol 86, pp (1978). 2. Askeland, D. R., Trojan, P.K., AFS Transactions, vol 77, pp (1969). 3. Askeland, D. R., Trojan, P.K., Flinn, R.A., AFS Transactions, vol 78, pp (1970). 4. Askeland, D. R., Trojan, P.K., Flinn, R.A., AFS Transactions, vol 80, pp (1972). 5. Barin, I., Thermochemical Data of the Pure Substances, VCH Verlagsgesellshaft, Weinheim, Germany (1989). 6. Bofan, Z., Langer, E.W., Scandinavian Journal of Metallurgy, vol 13, p 15 (1984). 7. Boyles, A., The Structure of Cast Iron, ASM, Metals Part, OH (1947). 8. Bramfitt, B. L., Metallurgical Transactions, vol 1, pp (1970). 9. Bruch, et al, Rheinstahl Technologie, no. 2, pp and no. 36, pp (1965). 10. Dawson, J. V., BCIRA Journal, vol 9, pp (1961). 11. Dawson, J. V., Modern Casting, vol 49, pp ((1966). 12. Deuchler, W., Giesserei Technische Wissenshaft Beihefte, vol 14, pp (1962). 13. Dhindaw, B, Verhoeven, J. D., Metallurgical Transactions A, vol 11A, pp (1980). 14. Eash, J. T., AFS Transactions, vol 49, pp (1941). 15. Elliott, R., Cast Iron Technology, pp 79 85, Butterworths & Co. Ltd., London (1988). 16. Feest, G. A., McHugh, G., Morton, D. O., Welch, L. S., Cook, I. A., Proceedings of Solidification Technology in the Foundry and Casthouse, Metals Society, pp (1983). 17. Francis, B., Metallurgical Transactions A, vol 10A, pp (1979). 18. Fredriksson, H., Materials Science and Engineering, vol 65, pp (1984). 19. Gadd, M. A., Bennett, G.H.J., Physical Chemistry of Inoculation in Cast Iron, 3 rd International Symposium on the Physical Metallurgy of Cast Iron, Stockholm (1984). 20. Heine, R. W., Loper, C. R., AFS Transactions, vol 74, pp , pp (1966). 21. Hu, H., Gu, H., Pan, Z., AFS Transactions, vol 102, pp (1994). 22. Hughes, I. C., Proceeding of Solidification Technology in the Foundry and Casthouse, Institute of Metals, Warwick (1980). 23. Igarishi, Y., Okada, S., International Journal of Cast Metals Research, vol 11, pp (1998). 24. Itofuji, H., Uchikawa, H., AFS Transactions, vol 98, p 429 (1990). 25. Jacobs, M. H., Law, T. J., Melford, D. A., Stowell, M. J., Metals Technology, vol 1, pp (1974). 26. Jacobs, M. H., Law, T. J., Melford, D. A., Stowell, M. J., Metals Technology, vol 3, pp (1976). 27. Kanetkar, C. S., Cornell, H. H., Stefanescu, D. M., AFS Transactions, vol 92, pp (1984). 28. Karsay, S. I., Ductile Iron I: Production, Quebec Iron & Titanium Corporation, Quebec, Canada (1976). 29. Kayama, N., Suzuki, K., Report Casting Research Laboratory, Waseda University, vol 30, pp (1979). 30. Kiessling, R., Non Metallic Inclusions in Steel, book no. 194, The Metals Society, London (1978). 31. Koshel, V, Palestin, S., Russian Castings Production, pp (April, 1971). 32. Kurz, W., Fisher, D.J., Fundamentals of Solidification, 3 rd Edition, Trans Tech Publications, Switzerland (1989). 33. Lalich, M. J., Proceedings of the 2 nd International Symposium on the Metallurgy of Cast Iron, Geneva, Switzerland, p 561 (1974). 34. Lalich, M. J. and Hitchings, J. R., AFS Transactions, vol 84, pp (1976). 35. Latona, M. C., Kwon, H. W., Wallace, J. F., Voss, J. D., AFS Transactions, vol 92, pp (1984). 36. Liu, P. C., Li, T. X., Li, C. L., Loper, C. R., AFS Transactions, vol 97, p 11 (1989). 37. Lownie, H. W., AFS Transactions, vol 64, pp (1956). 38. Lownie, H. W., Foundry, vol 91, pp (1963). 39. Lux, B., Modern Casting, vol 45, pp (1964). 40. McClure, N. C., Khan, A. V., McGrady, D., Womochel, H. L., AFS Transactions, vol 65, pp (1957). 29

84 Proceedings of the AFS Cast Iron Inoculation Conference, September 29-30, 2005, Schaumburg, Illinois 41. Merchant, H. D., Journal of the Iron and Steel Institute, vol 210, pp (1972). 42. Mercier, J. C., Fonderia, no. 277, pp (1969). 43. Mickelson, R. L., Foundry, vol 95, pp (1967). 44. Minkoff, I., The Physical Metallurgy of Cast Iron, pp 55 63, John Wiley & Sons Ltd., New York (1983). 45. Mountfort, F., The British Foundryman,, pp (April, 1966). 46. Muzumdar, K. M., Wallace, J. F., AFS Transactions, vol 80, pp (1972). 47. Muzumdar, K. M., Wallace, J. F., AFS Transactions, vol 81, pp (1973). 48. Narasimhan, M. C., Chaudhari, R. D., Indian Journal of Technology, vol 7, pp (1969). 49. Nakae, H., Jung, S., Inoue, H., Shin, H., Proceedings of the 66 th World Foundry Congress, pp , Istanbul, Turkey (2004). 50. Naro, R. L., Wallace, J. F., AFS Transactions, vol 77, pp (1969). 51. Naro, R. L., Wallace, J. F., AFS Transactions, vol 78, pp (1970). 52. Onsøien, M. I., Grong, Ø, Skaland, T, Olsen, S. O., AFS Transactions, vol 105, pp (1997). 53. Pan, E. N., Lin, C. N., Chiou, H. S., Proceedings of the 2 nd Asian Foundry Congress, Japan Foundrymen s Society, p 36 (1994). 54. Patterson, V. H., Lalich, M. J., AFS Transactions, vol 86, pp (1978). 55. Pearce, J. G., AFS Transactions, vol 70, pp (1962). 56. Quian, L., Modern Castings, no. 3, pp (1985). 57. Ramdohr, P, Strunz, H., Lehrbuch der Mineralogie, Ferdinand Enke Verlag, Stuttgard, Germany (1978). 58. Rein, R. H., Chipman, J., Transactions of the Metals Society AIME, vol 233, pp (1965). 59. Rosenstiel, A. P., Bakkerus, H., Giesserei Technische Wissenshaft Beihefte, vol 16, pp (1964). 60. Ruff, G. F., Wallace, J. F., AFS Transactions, vol 84, pp (1976). 61. Sadocha, J. P., Gruzleski, J E., Proceedings of the 2 nd International Symposium on the Metallurgy of Cast Iron, pp Geneva, Switzerland (1974). 62. Sawyer, J. C., Wallace, J. F., AFS Transactions, vol 76, pp (1968). 63. Skaland, T., Metallurgical Transactions A, vol 24A, pp (1993). 64. Smickley, R. J., Rundman, K. B., AFS Transactions, vol 89, pp (1981). 65. Solberg, J. K., Onsøien, M. I., Materials Science and Technology, vol 17, pp (2001). 66. Sponseller, D. L., Flinn, R. A., Transactions of the Metals Society AIME, no. 230, pp (1964). 67. Stefanescu, D. M., Biswal, S. K., Kanetkar, C., Cornell, H.H., Proceedings of Advanced Casting Technology, p 167 Kalamazoo, MI (November, 1986). 68. Sun, G. X., Loper, C. R., AFS Transactions, vol 91, pp (1983). 69. Trojan, P. K., Guichelaar, P. J., Bargeron, W. N., Flinn, R. A., AFS Transactions, vol 76, pp (1968). 70. Turnbull, D. and Vonnegut, R., Industrial Engineering Chemistry, vol 44, pp (1952). 71. Udomom, U. H., Loper, C.R., AFS Transactions, vol 93, p 519 (1985). 72. Wallace, J. F., Wieser, P. F., AFS Research Report (1965). 73. Wallace, J. F., AFS Transactions, vol 85, pp (1975). 74. Wang, C. H. and Fredriksson, H., Proceedings of the 48 th International Foundry Congress, Varna, Bulgaria (1981). 75. Warrick, R. J., AFS Cast Metals Research Journal, vol 2, no. 3, pp (1966). 76. Watmough, N. L., Shaw, R., Bock, J., AFS Transactions, vol 79, pp (1971). 77. Wittmoser, A., Giesserei Technische Wissenshaft Beihefte, no. 6 8, pp (1952). 78. Zeedijk, H. B., Journal of Iron and Steel Institute, vol 203, pp (1965). 30

85 A New Approach to Ductile Iron Inoculation T. Skaland Elkem ASA, Research Kristiansand, Norway Copyright 2001 American Foundry Society ABSTRACT The objective of the present paper is to describe a new approach to inoculant design that has proven successful in improving casting performance and properties. The described inoculant represents a unique new generation of products developed for powerful cast iron inoculation. The ferrosilicon alloy contains levels of Calcium and Cerium that are adjusted to minimize chill formation and neutralize subversive trace elements in the iron. In addition, the inoculant contains small and controlled amounts of Sulphur and Oxygen in a form that make them available for reaction with the Calcium and Cerium during introduction into liquid iron. This special composition is designed to give highly powerful graphite nucleation conditions in ductile irons along with very effective chill and shrinkage reduction. Examples from foundry testing are reviewed, and the unique characteristics of this new inoculant concept described. The new inoculant is found to be more powerful than conventional ferrosilicon based inoculants, and give rise to very effective reduction in the shrinkage tendency of ductile irons. Special effectiveness has been observed in irons of low sulphur or irons of a dead nature from prolonged holding times. Also, results show improvements in both tensile properties as well as machinability for ferritic ductile irons. The new inoculant concept represents a patent protected design (Int.Pat.No.WO99/29911), and is available under a special trade name, see footnote 1. INTRODUCTION Based on years of laboratory work with experimental inoculants of various compositions, a new concept for inoculation of ductile iron has been developed. The concept is novel in the sense that it involves not only an alloyed ferrosilicon-based material, but also introduction of non-metallic powders with the ferrosilicon alloy to give its special characteristic. The background for development of this product has been based on new fundamental understanding of graphite nucleation mechanisms in ductile iron, where the main body of nucleation sites was found to be comprised of complex but very stable sulphides and oxides (Skaland 1992). Figure 1 shows an example of such nucleation site in ductile iron both as a high magnification micrograph and a schematic representation of its phase composition. In conventional ductile iron production the availability of such sulphide and oxide nucleation sites are determined by the purity of base metal and its additives, holding times and temperature as well as metallurgical treatment processes and additives. Traditionally, commercial inoculants have been based on ferrosilicon alloys containing metallic additives such as Calcium, Barium, Strontium, Aluminum, Zirconium, Rare Earth s, etc., with the main objective of these reactive elements to combine with Sulphur and Oxygen in the iron and form potent heterogeneous nucleation sites for graphite. However, with restricted availability of Sulphur and Oxygen in the iron, the metallic inoculant additives may reach a performance limit where their effectiveness are restricted by the number of potent nucleation sites that can be formed after treatment. Thus, the primary objective of the new inoculant concept has been to introduce controlled concentrations of non-metallic elements such as Sulphur and Oxygen with the metallic inoculant. From balanced and controlled inclusion engineering, this will deliberately produce a higher number density of nucleation sites for graphite from a reaction taking place between the highly reactive metallic ingredients (Ca and Ce) and the non-metallic ingredients (S and O) of the inoculant. These additional nucleation sites will then perform in parallel to the traditional sites formed during reactions between nodulizer, inoculant and the base metal. The outcome will be a remarkable improvement in conditions for controlled graphite precipitation and growth, with all possible benefits this may introduce to the final iron quality. Several researchers have proven the benefits of Sulphur to graphite nucleation (Chisamera 1994, Lalich 1976, Mercier 1969). Also, it has been proposed that Oxygen may play a vital role in the inoculation process (Tartera 1980, Nakae 1992, Podrzucki 2000). However, the combined use and performance of both elements through post-inoculation is a novel approach that has been designed to get the benefits from both Calcium, Cerium, Sulphur and Oxygen simultaneously in the graphite nucleation process. Calcium is used as the primary reactive element in inoculation, and has proven crucial for eutectic graphite 1 The new inoculant is available under the trade name Ultraseed inoculant, produced by Elkem. 1

86 nucleation (Bilek 1972). Cerium is introduced for several reasons. First, Cerium will contribute in neutralizing subversive trace elements in the base iron, forming stable inter-metallic compounds (Park 2000, Udomon 1985). Cerium will also have strong affinity to Sulphur and Oxygen, resulting in the formation of highly stable Cerium oxides, sulphide, and oxy-sulphides (Kozlov 1991, Warrick 1966). These Cerium compounds appear to be very beneficial in the inoculation process, resulting in improved nucleation effectiveness throughout the entire solidification range. (a) (b) Figure 1. (a) Example of duplex sulphide/oxide nuclei particle in ductile iron at large magnification in a transmission electron microscope (70,000X). (b) Schematic representation of a nucleus particle containing complex sulphide and oxide phases after nodularizing and inoculation of the iron (Skaland 1992). EFFECTS OF INOCULATION ON CAST IRON PROPERTIES The principal effects of cast iron inoculation can be described as follows: Avoid the formation of hard carbides (cementite) Promote the formation of graphite and ferrite Reduce the segregation tendency of alloying and trace elements Reduce the solidification shrinkage tendency Improve the machinability of castings Reduce the hardness Increase the ductility Give more homogeneous structures and properties in different sections of complex castings The new inoculant concept is found to improve most all of these properties to a greater extent than other ferrosilicon based inoculant alloys. Especially, improvements in ferrite formation, shrinkage minimizing, machinability, and microstructure homogeneity have been observed through extensive testing in various foundry conditions. In the following, some of the unique features will be described in more detail, also including examples from foundries. A series of case studies will be reviewed to illustrate the performance characteristics through realistic examples from the industry. UNIQUE FEATURES OF THE NEW INOCULANT CONCEPT The new inoculant concept provides formation of extra nucleation sites in ductile iron in addition to those initially generated by the magnesium treatment. This will increase nodule count and improve nodularity thus reducing carbide and shrinkage tendency. The balanced cerium content also neutralizes subversive elements that may prevent the formation of nodular graphite. Due to the higher nodule count obtained, the inoculant also provides formation of more ferrite in ductile irons. This can be an advantage when producing the higher ductility and impact resistant grades of ferritic iron (e.g. grade 40.3). The powerful nucleation characteristics are based on the formation of special cerium-calcium-sulphides and oxides that will act as effective nucleation sites for graphite during solidification of the iron. These nucleation sites will contribute together with the primary magnesium-silicon oxides to give powerful graphite nucleation with the outcome being a very high nodule number density. The inoculant is found to be especially potent in ductile irons of relatively low sulphur content and in irons treated with magnesium metal in a converter or wire injection process. The introduction of Cerium (Rare Earth metal) through the inoculant can also replace the need for Rare Earth s to be added through the nodularizing process. The inoculant has also proven highly successful in providing fresh nucleation sites to ductile irons of long holding time where the base iron 2

87 or magnesium treated iron have been held for prolonged times before addition of the post inoculant. Such long hold times are well known to reduce the overall nucleation capabilities of the iron prior to inoculation resulting in so-called dead iron. The new inoculant concept will thus re-install good nucleation effectiveness from reactions with its deliberate Sulphur and Oxygen content forming additional, new nucleation sites. Due to the powerful effects on raising nodule count and improving chill protection, it has been found that the tendency to shrinkage cavity formation is also greatly reduced with this inoculant. Especially, the type of shrinkage that often occur as small porosities in hot-spot sections of complex castings appear to be effectively reduced or even eliminated by this inoculant concept. It has been found that a characteristic bi-modal size distribution of nodules often will occur from a secondary, late precipitation event in the last part of the solidification sequence. Such late graphite expansion effects will effectively counteract shrinkage contraction in the last part of solidification, when risers have stopped functioning and graphite expansion is most needed to counteract shrink. It appears that the new inoculant concept is effectively distributing the graphite nucleation and growth phenomena throughout the entire solidification range. Conventional inoculants, however, have a tendency to give massive and early expansion effects and very little contribution in the last part of solidification when really most needed. Strong nucleation effect and high nodule count is also a prerequisite to maximize the ferrite content when producing as-cast ferritic grades of ductile iron. Particularly when there are limitations on the final silicon content of the iron, the high nodule count obtained with this inoculant has proven effective to ensure the required minimum content of ferrite in such castings. The bi-modal size distribution of nodules, and the fact that the smaller and late precipitated nodules are formed in the last liquid to freeze, also aid in formation of more ferrite by acting as effective carbon sinks in these segregated areas enriched in pearlite promoting elements. This is also indirectly improving the machinability of ductile iron, and the inoculant should therefore be the preferred choice when good machinability is an important requirement. The special inoculant composition including additions of finely dispersed oxides and sulphides with the ferrosilicon based alloy, causes a specific appearance of this product. Figure 2 shows a comparison of the new inoculant design and a conventional ferrosilicon based alloy. It is clear that the new concept appears with characteristic black particle surfaces due to its sulphide and oxide content. Table 1 gives the specifications and typical composition of the new inoculant concept. (a) (b) Figure 2. Physical appearance of (a) conventional ferrosilicon inoculant with metallic, glinsing surface characteristics, (b) new inoculant concept with black surface characteristics. Table 1. Specifications and typical composition of the new inoculant concept. % Silicon % Calcium % Cerium % Aluminum % Sulphur % Oxygen Specifications Max. 1.0 Max. 1.0 Typicals Trace Trace 3

88 RESULTS FROM FOUNDRY TESTING The new inoculant concept has now been tested and implemented in numerous foundries Globally. More than 80 foundries have conducted testing so far, and of these above 60% have reported some kind of successful results. Foundries have different criteria and objectives in testing, and are looking for individual types of property improvements. The following specific features may be mentioned from foundry testing: Especially effective to reinstall powerful inoculation conditions in dead base irons Improved performance both in pure Magnesium and Magnesium ferrosilicon treatment applications Especially effective used as a late in-stream inoculation Especially effective in eliminating shrinkage porosity in complicated hot-spot sections Reduces the section sensitivity of nodule structures in castings of variable thickness May not be that efficient as an all-round gray iron inoculant Successful applications also observed in compacted graphite iron (reduced section sensitivity) In the following, four case studies from foundry testing will be reviewed. The intention with these case studies is to show some typical examples of performance characteristics observed in different foundry conditions with the new inoculant concept. CASE STUDY 1 This foundry uses electric induction melting and a tundish ladle process for preparing ductile iron. The treated ductile iron is transferred into a fairly large channel induction holding and pouring unit where iron may sit for a while. The foundry has experienced problems with carbides and excess shrinkage in complex castings of very thin sections. A key challenge is to avoid big shrinkage porosity in a distant hot spot knob section that has to be drilled and need a smooth inner hole surface. The foundry normally uses an in-stream late post inoculation addition of a (Zr,Mn,Ca)-bearing ferrosilicon inoculant. When first testing the new inoculant concept, this was directly compared to the (Zr,Mn,Ca)-bearing ferrosilicon alloy as an in-stream addition. Table 2 shows the typical average nodule number densities obtained for the two inoculants, while Figures 3 through 5 shows examples of the effects on nodule structure, carbide formation, and hot-spot shrinkage formation tendency. It is evident from the results that a pronounced difference in nodule number density and size distribution occur for the two inoculants. The new inoculant concept gives about double the nodule number density of the (Zr,Mn,Ca)-bearing inoculant, and the nodule size distribution shows a shift to numerous smaller and better shaped nodules. Figure 4 also shows the effects on carbide formation tendency in a very thin, 2-3 mm section of the same casting. The (Zr,Mn,Ca)-bearing inoculant with its lower nodule count reveals the appearance of intercellular carbides, while the (Ca,Ce,S,O)-inoculant has effectively eliminated these thin section carbides. Finally, Figure 5 shows the effects of inoculation on the critical and difficult hot-spot section. The large region of micro-shrinkage occurring with the (Zr,Mn,Ca)- bearing inoculant has been effectively minimized with the new (Ca,Ce,S,O)-inoculant. The extension of micro-porosity in Figure 5b has proven small enough avoiding rough inner surfaces after drilling of this critical knob section. Table 2. Average nodule number density for the (Zr,Mn,Ca)-bearing ferrosilicon inoculant and the new inoculant concept. (Zr,Mn,Ca)-inoculant New inoculant concept 315 nodules per mm nodules per mm 2 Conclusively, it can be said that this foundry is very happy with the new inoculant performances. A complete transition from the (Zr,Mn,Ca)-bearing inoculant to the (Ca,Ce,S,O)-inoculant has been taking place for ductile iron production. Significant improvements in reject rate and final casting quality has been obtained. The present addition rate of the new inoculant concept is about 25% less than the previous (Zr,Mn,Ca)-bearing alloy, still providing the significant improvements. 4

89 (a) (b) (c) (d) Figure 3. Examples of microstructure results from Test Foundry 1. (Zr,Mn,Ca)-containing inoculant, (a) polished condition, (b) etched in Nital. New (Ca,Ce,S,O)-containing inoculant, (c) polished condition, (d) etched in Nital. (100X) (a) (b) Figure 4. Examples of carbide conditions in thin 2 mm flange from Test Foundry 1. (Zr,Mn,Ca)-containing inoculant revealing intercellular carbides. (Ca,Ce,S,O)-containing inoculant showing carbide free conditions. (a) (b) Figure 5. Examples of hot-spot micro-shrinkage porosity formation tendency from Test Foundry 1. (a) (Zr,Mn,Ca)-containing inoculant causing massive shrinkage porosities. (b) (Ca,Ce,S,O)-containing inoculant giving only traces of micro-shrinkage porosity. 5

90 CASE STUDY 2 This foundry is also an induction melting, sandwich treatment operation, and here the objective has been to test a series of different generic inoculants in order to find the optimum product for their autopouring and in-stream inoculation on a DISA molding line. Output parameters evaluated include nodule- and microstructure, mechanical properties and shrinkage tendency for the different inoculants. (a) (b) (c) (d) Figure 6. Examples of graphite nodule structure in plate castings from Test Foundry 2. 5 mm section size: (a) Sr-containing inoculant, (b) (Ca,Ce,S,O)-containing inoculant. 40 mm section size: (c) Sr-containing inoculant, (d) (Ca,Ce,S,O)-containing inoculant (100X). This foundry normally uses a Strontium-bearing ferrosilicon inoculant, and Figure 6 shows an example of the effects of Sr- FeSi and (Ca,Ce,S,O)-FeSi inoculation on the final nodule structure in a thin 5 mm and a thick 40 mm plate section casting. A quite remarkable difference is observed, especially for the heavier 40 mm plate section, where the (Ca,Ce,S,O)-inoculant appears to give a very strong increase in nodule count. Figure 7 shows a quantitative comparison of nodule counts for the two inoculants in sections of 5, 10, 20 and 40 mm thickness. The Sr-FeSi inoculant gives a quite normal and expected behavior with a falling nodule count from about 300 per mm 2 to 150 per mm 2 when the section thickness increases from 5 through 40 mm. This behavior is normally observed for most commercial inoculants when increasing the section thickness. The (Ca,Ce,S,O)-containing inoculant on the other hand, shows a complete different and quite remarkable behavior. The histograms in Figure 7 and also the micrographs in Figure 6, shows an almost unaffected nodule count for the spread in section thickness. About 300 nodules per mm 2 are measured for both the 5 and 40 mm sections of the same test casting. In fact, the 40 mm section contains an even higher nodule number density of 340 per mm 2 versus only 312 per mm 2 for the much thinner 5 mm section. This is clearly confirmed by the micrographs in Figures 6b and 6d. 6

91 N/mm Ce, S, O Sr Section size (mm) Figure 7. Example of nodule count in various section thicknesses from 5 to 40 mm plate castings from Test Foundry 2 for (Ca,Ce,S,O)-containing inoculant and Sr-containing inoculant. This unusual behavior opens up for some interesting performances. First, it is evident that the section sensitivity for complex castings can be greatly reduced and microstructure and properties controlled and equalized for a large span in section thickness. This in itself may offer advantages to the homogeneous production of difficult castings with demanding properties in different locations. Further, the observation of an elevated nodule count in heavier sections also offer some additional benefits in relation to solidification contraction and shrinkage tendency. Figure 8 shows examples of crossbar hot spot shrinkage conditions using three different inoculants in Test Foundry 2. The inoculants are Ba-containing, Sr-containing, and the new (Ca,Ce,S,O)-containing FeSi. As the Figure shows, shrinkage tendency differs greatly for the different inoculants. Both the Ba-containing and Sr-containing alloys give massive contraction effects and large cavities in the hot-spot cross bar. The (Ca,Ce,S,O)-containing inoculant on the other hand, shows an almost complete elimination of shrinkage porosity with only one very small cavity revealed in the section cut through the parting line of the experimental cross bar casting. This dramatic effect on shrinkage tendency can be directly related to the nodule formation and the rate of graphite growth throughout the entire solidification sequence. As shown in Figures 6 and 7, the conventional inoculant behavior, represented by the Sr-inoculant, is to give fairly uniform nodule sizes and a reduction in nodule count as the section size increases. With the (Ca,Ce,S,O)-inoculant concept, there is an effect causing a bi-modal nodule size distribution and numerous smaller nodules that are precipitated very late during solidification. This late graphite expansion effectively counteract shrinkage contraction, as can be clearly seen in Figure 8c for the (Ca,Ce,S,O)-containing inoculant. Both the Sr- and Ba-containing inoculants give low nodule counts for heavier sections, around per mm 2, while the (Ca,Ce,S,O)-containing inoculant gives about 350 nodules per mm 2 for the similar section. Since this phenomenon predominantly occur for heavier sections, this is where the bi-modal size distribution is most clearly observed and also where shrinkage control is mostly needed. There exist no clear understanding of the mechanisms of the late graphite formation and the resulting bi-modal nodule distribution effect. However, it is expected that the introduction of Cerium in combination with sulphur and Oxygen in the inoculant, will introduce more nucleation sites, and possibly also a second type of sites that are activated later during solidification. Cerium oxides and oxy-sulphides will behave very different from the traditional Ca,Ba,Sr-type oxides and silicates know to nucleate primary graphite in the early stages of solidification (see Figure 1b). 7

92 (a) (b) (c) Figure 8. Examples of shrinkage porosity formation in crossbar castings from Test Foundry 2. (a) Ba-containing inoculant, (b) Sr-containing inoculant, (c) (Ca,Ce,S,O)-containing inoculant. Since the phenomenon is most evident in heavier sections, it is likely that this second type of beneficial and late activated nucleation sites only show their characteristic effects in slower cooling conditions. This is, the second type nucleation sites need more time to become activated, and will thus only give maximum benefits in heavier sections of a casting. The net outcome appears as a uniform nodule count in different sections, combined with effective shrink elimination in the heavier and slower cooled sections. The conclusion from this extensive inoculant testing has been that the Test Foundry 2 now has converted to use the new (Ca,Ce,S,O)-containing inoculant for all their ductile iron production. Great improvements in especially shrinkage reduction have been experienced, and the inoculant addition rate also reduced to a minimum. CASE STUDY 3 The third case study represents a foundry producing very heavy ductile iron castings. Induction melting and tundish treatment is applied also here. In this case, the foundry suffers from the classical problems of heavy section castings such as graphite flotation, segregation, shrinkage and relatively poor nodularity. The foundry uses manual transfer inoculation to large pouring ladles, and the present inoculant material applied is a Barium-containing ferrosilicon alloy. Barium inoculants are traditionally recommended for heavy casting and slow cooling applications, since Ba is recognized for its minimum of fading tendency during prolonged hold and solidification times. The new (Ca,Ce,S,O)-type inoculant was tested in parallel to the Ba-inoculant as a ladle addition, and effects on microstructure, machinability and shrinkage tendency evaluated. Figure 9 shows examples of typical microstructures obtained with the two different inoculants in a fairly heavy 50 mm section. The Ba-inoculant shows the expected relatively large and uniformly sized nodules in a ferritic/pearlitic matrix (see Figures 9a and 9b). The (Ca,Ce,S,O)-inoculant, on the other hand, shows a much wider spread in nodule sizes, and the characteristic bi-modal distribution effect is again clearly revealed. Figures 9c and 9d shows the effect of the (Ca,Ce,S,O)- inoculant on nodule distribution and ferrite/pearlite ratio. Table 3 also gives the quantified microstructure data for the two respective inoculants in this heavy section application. From Figure 9 and Table 3 it is evident that the heavy section impact on microstructure for the (Ca,Ce,S,O)-inoculant is significant. The bi-modal nodule distribution effect was found to effectively minimize difficult and massive shrinkage formation in large castings. The formation of smaller nodules also gave a general improvement of 10% in the nodularity from about 80 to 90 %. Further, Figures 9b versus 9d clearly show a significant reduction in intercellular pearlite with the new inoculant concept. The reduction is quantified from 25% down to 13% pearlite. The interconnected network of pearlite at 25% and higher is broken down into only minor fragments of pearlite in a predominantly ferritic matrix. This is again due to the numerous smaller nodules arising in the segregated intercellular regions, acting as carbon sinks during the eutectoid transformation. Raising nodule counts by a general increase in the primary formed larger nodules, typically will not influence the pearlite ratio to the same great extent. This is because segregation patterns and profiles will still remain the same. When the grain boundary nodules are included, this will have a pronounced effect on scavenging the matrix for carbon, thus effectively reducing the risk for harmful segregation phenomena and formation of intercellular carbides, phosphides, and other unwanted microconstituents. A general improvement in tensile and impact properties was also experienced with the bimodal and homogeneous nodule distribution. Also, an improved tool life during machining of up to 50% was experienced with this new situation. 8

93 (a) (b) (c) (d) Figure 9. Examples of microstructure in heavy section castings from Test Foundry 3. Ba-containing inoculant: (a) polished condition, (b) etched in Nital. (Ca,Ce,S,O)-containing inoculant:(c) polished condition, (d) etched in Nital. Table 3. Effects of Ba- and (Ca,Ce,S,O)-containing inoculants on microstructure characteristics in heavy section casting at Test Foundry 3. Nodule count Per mm 2 Nodularity % Pearlite % Shrinkage tendency Relative machinability Ba-inoculant Significant Medium (Ca,Ce,S,O)-inoculant Much less Good CASE STUDY 4 The final case study included here is from an automotive foundry using electric induction melting and sandwich magnesium treatment. Molding is done on BMD and DISA lines, autopouring through Junker units. Late in-stream inoculation is applied on all pouring lines, and typically Zirconium-bearing inoculants have been applied. The foundry is suffering from some serious shrinkage problems, and the type shrink can be described as massive cavities in critical sections. Extensive risering has been implemented to try and overcome the problems, but still large shrinkage cavities are found to occur even adjacent to the risers. The new inoculant concept using a (Ca,Ce,S,O)-bearing ferrosilicon inoculants was tested out on the DISA line for a special test pattern involving a square cubic test piece attached to a fairly large riser. The cube and riser are cut in half, and evaluated for degree of shrinkage porosity formation and distribution of cavities in test casting and riser. Figures 10 and 11 shows examples of such test castings cut through the middle for evaluation of shrink. The examples compare two different Zirconium-containing inoculants, one (Zr,Mn,Ca)-type and one (Zr,Ca)-type, to the new (Ca,Ce,S,O)- 9

94 concept inoculant. Figure 10 shows conditions for the three inoculants, where the (Zr,Mn,Ca)-type to the left reveals a large cluster of micro-shrinkage porosity in the body of the cube test casting. The riser is virtually sound. The (Zr,Ca)-inoculant in the middle shows one large cavity in the cube sample and another in the neck of the riser, while the (Ca,Ce,S,O)-inoculant to the right shows only a very small porosity in the cube test piece. Conditions shown in Figure 11 compares the (Zr,Ca)-inoculant to the (Ca,Ce,S,O)-inoculant. The test piece for the (Zr,Ca)- inoculant to the left shows again massive shrinkage in the cube sample and still quite widespread porosities also in the riser. The (Ca,Ce,S,O)-inoculant to the right has pushed the shrinkage void all the way back to the top of the riser, leaving the lower part of the riser and the cube casting itself sound and completely free from porosities. The foundry has solved a severe shrinkage problem using the new concept inoculant. Complete conversion from Zr-bearing to (Ca,Ce,S,O)-bearing inoculation has been implemented at this foundry. Figure 10. Example of cubic test castings with attached riser from Test Foundry 4. Inoculants tested are: Left: (Zr,Mn,Ca)-bearing, middle: (Zr,Ca)-bearing, and right: (Ca,Ce,S,O)-bearing. There exist numerous other cases showing similar improvements in chill situation, nodule structure, and shrinkage formation tendency using the new concept (Ca,Ce,S,O)-containing inoculant. However, from space limitations this paper is restricted to only cover a selected handful of classical situations experienced in small/thin and heavy/thick casting conditions. As shown above, effects are found to be most pronounced for nodule count and size distribution, as well as carbide restriction and shrink control. Other foundries have also reported great improvements in machinability conditions, and one foundry even eliminated their heat treatment operation after converting inoculants. Double tool life at half the addition rate of inoculant versus the previous regular calcium-bearing ferrosilicon has also been reported. Some foundries report of increased tensile strength even with a significant increase in the ferrite content. This paper most of all shows that the choice of inoculant material is not a trivial thing, and that different commercial inoculants may have dramatic effects on the final ductile iron quality. It is important to keep in mind, that sometimes the least expected inoculant is the one to actually perform best. Systematic and thoroughly controlled foundry testing will be the only sound and safe way to ensure that the optimum cost efficient alternative inoculant is being used in the individual foundry. There are too many unknown and uncontrollable factors affecting inoculation to give a general recommendation, and testing to solve specific challenges will at the end be the only safe way to find the improved or optimized inoculant solution. 10

95 Figure 11. Example of cubic test castings from foundry 4. Inoculants tested are: Left: (Zr,Ca)-bearing, right: (Ca,Ce,S,O)-bearing. Note the differences in shrinkage cavity distribution for the two inoculants. When testing inoculants in cast iron it is always important not only to look for a quick micrograph. Very attractive effects may then be overseen. A thorough evaluation of all the above discussed factors must be considered, since great savings may be found in reduced scrap rate from shrink, poor structures, machinability or even tensile and impact properties. CONCLUSIONS The following main conclusions can be given from the present investigation: A new approach to ductile iron inoculant design has been described. The new design has proven successful in improving casting performance and properties. The new ferrosilicon based inoculant material contains levels of Calcium and Cerium that are adjusted to minimize chill formation and neutralize subversive trace elements in the iron. The new inoculant design also contains small and controlled amounts of Sulphur and Oxygen in a form that make them available for reaction with the Calcium and Cerium during introduction into liquid iron. The special composition is designed to give highly powerful graphite nucleation conditions in ductile irons along with very effective chill and shrinkage reduction. Experience from foundry testing has proven that the new inoculant concept is especially effective in re-installing powerful inoculation conditions in irons of a dead nature. Also, especial effectiveness in minimizing shrinkage porosity in complicated hot-spot sections has been observed. The new (Ca,Ce,S,O)-containing inoculant is also reducing the section sensitivity of nodule structures in castings of variable thickness. A bi-modal size distribution of graphite nodules is often observed. This nodule distribution is instrumental in minimizing shrink and intercellular pearlite and carbide formation. 11

96 REFERENCES Bilek, P.J., Dong, J.M., McCluhan, T.K., The Role of Ca and Al in Inoculation of Gray Iron, AFS Transactions, pp , (1972) Chisamera, M., Riposan, I., Proc. 5th Int. Symp. On the Physical Metallurgy of Cast Iron, Nancy, France, Sept. (1994) Kozlov, L.J., Vorobyev, A.P., The Role of Rare-earth Metals in the Process of Spheroidal Graphite Formation, Cast Metals, vol.4, no.1, (1991) Lalich, M.J., Hitchings, J.R., Characterization of Inclusions as Nuclei for Spheroidal Graphite in Ductile Cast Iron, AFS Transactions, pp , (1976) Mercier, J-C., Paton, R., Margerie, J-C., Mascre, C., Inclusions dans les spheroides de graphite, Fonderie, April (1969) Nakae, H., Koizumi, H., Takai, K., Okauchi, K., Nucleation of Graphite in Inoculated Cast Iron, Trans. Japan Foundrymen s Society, vol.11, pp , (1992) Park, J., Loper, C.R., Neutralizing of Lead in Gray Iron Melts Using Misch Metal, AFS Transactions, (2000) Podrzucki, C., Fras, E, Lopez, H.F., The Inoculation of Cast Iron: Role of Oxygen, AFS Transactions, (2000). Skaland, T., A Model for the Graphite Formation in Ductile Cast Iron, Ph.D. Thesis 1992:33, The University of Trondheim, NTH, Department of Metallurgy, Norway, (1992) Tartera, J., Cast Iron Inoculation Mechanisms, AFS Int. Cast Metals J., pp.7-14, December (1980) Udomon, U.H., Loper, C.R., Comments Concerning the Interaction of Rare Earths With Subversive Elements In Cast Irons, AFS Transactions, pp , (1985) Warrick, R.J., Spheroidal Graphite Nuclei in Rare Earth and Magnesium Inoculated Irons, AFS Cast Metals Research J., pp , Sept., (1966) 12

97 Proceedings of the AFS Cast Iron Inoculation Conference, September 29-30, 2005, Schaumburg, Illinois Metallurgical Comparisons between Operating Conditions, Inoculant Types and Fade Effects in Gray Iron Copyright 2005 American Foundry Society ABSTRACT V. Popovski Elkem Metals, Inc., Pittsburgh, PA C. Misterek John Deere Foundry, Waterloo, IA L. Kaiser Dalton Foundries, Warsaw, IN Inoculant testing has been conducted in two operating foundries with different melting systems (electric vs. cupola). A series of gray iron (GI) ladles were inoculated with various generic inoculants and at different addition rates. The ladles were held for up to 22 minutes and samples were extracted approximately every 2 minutes. Samples were examined for thermal analysis properties and microstructure evolution. The investigation has shown that inoculants fade differently and that the various inoculants behave differently in the two melting systems. INTRODUCTION The primary objective of gray iron (GI) inoculation is to produce a shrink-free casting with a microstructure with a maximum of Type A graphite, no carbides and excellent machinability. It is established that inoculant effectiveness degrades (fades) in GI as the pouring ladle is held over time. Skaland writes, The principal effects of fading are to cause greater undercooling to take place during eutectic solidification and to lead to a greater tendency for chilling in grey and ductile irons, particularly in thin sections (and) to reduce the number of eutectic cells growing in flake graphite irons resulting in less uniform size and distribution of graphite in the castings and a reduction in mechanical properties (Skaland, 1992). Olsen believes the fading of inoculant to the coarsening and growth of microinclusions, also called the Ostwald Ripening Effect. The driving force for this coarsening is a reduction in the specific surface area of the inclusions, thus reducing the total energy of the system (Olsen, 2004). Fading of inoculant has been demonstrated in various measurable ways. Inoculant fade was demonstrated in tensile strength by Datta, who states, Prolonged holding of iron after inoculation is detrimental to tensile strength of iron (and) the drop in tensile strength is primarily due to progressive increase in Type D graphite and ferrite in the structure with increase in holding time. Datta also observes, Besides formation of undercooled graphite and ferrite, prolonged holding of iron also promotes formation of carbide as measured by casting hardness (Datta, 1977). Fuller measured drops in cell count over time, among other parameters. However, Fuller goes on to write, There is not a common relationship between eutectic cell number and chill for all inoculating materials (Fuller, 1979). Skaland agrees with this assessment, An inoculant which gives a high eutectic cell number is not necessarily the most effective in reducing chill (Skaland, 1992). Inoculation effectiveness depends on many operating conditions, including chemistry. Skaland writes, The effects of inoculants may vary according to the composition of the iron, particularly if it has a low sulfur content (Skaland, 1992). Also, Chisamera writes, Eutectic Undercooling and Recalescence degrees as main solidification parameters were connected to aluminum level (Chisamera, 2004). Based on those remarks, one should also expect different fading characteristics of the same inoculant under different operating conditions, such as iron chemistry. It is firmly established that inoculation mechanisms vary widely between inoculants. Comparing nuclei (complex inclusions of the type MnS) of irons inoculated with Sr- and Ca-FeSi, Riposan writes, Important elements such as Ca and Sr was (sic) found to distribute differently in the inclusion volume (Riposan, 2001). Because the active ingredients in inoculants vary in behavior so widely, it is logical to expect different fading characteristics in the same iron for different inoculants; these inclusions are the same ones referred to by Olsen (Olsen, 2004). 57

98 Proceedings of the AFS Cast Iron Inoculation Conference, September 29-30, 2005, Schaumburg, Illinois Other studies have compared the fading of different inoculants. Skaland writes, The barium containing inoculant produces a high initial number of nucleation sites. Compared to other inoculants, it maintains a high number of nucleation sites throughout the holding period (Skaland, 1992). Fuller concurs in a different study, With the exception of barium containing inoculants the fading rate decreased with holding time for the first ten minutes. Thereafter fading almost ceased. However, Fuller also adds that Irons inoculated with strontium containing ferrosilicon which always had the lowest cell number also had the lowest chilling tendency and acknowledged that cerium (Ce) provided better chill reduction than did barium (Fuller, 1979). Ultimately, the effectiveness of an inoculant is dependent on the desires of the foundry. While one foundry might desire chill prevention first and foremost, another foundry might never have a chill problem and therefore they seek to maximize tensile strength with inoculation. The relevance of inoculant fade is likewise unique to the foundry. EXPLANATION OF THERMAL ANALYSIS TERMS Thermal analysis of iron samples was done in this study with the ATAS (Adaptive Thermal Analysis System) method. In thermal analysis, a sample of iron is poured into a sand cup containing a thermocouple. The metal cools and begins to solidify. There is a thermal arrest when the first austenite nucleates (Fig. 1, [Sillen, 2003]). This is the liquidus temperature (TL). The sample continues cooling, austenite continues nucleating and the liquid iron grows richer in carbon content until the sample begins eutectic freezing. This is followed by another arrest, the eutectic temperature or TElow. Graphite forms, releasing the latent heat of solidification, and the temperature rises to a peak. This peak temperature is called TEhigh. The difference between them is called the recalescence (R). The sample continues cooling until it is solid. The last temperature of liquid iron is called the solidus (TS). Further analysis of the curve provides Graphite Factor 1 (GRF1), which is an indication of overall eutectic graphite precipitation during eutectic freezing after TEhigh. The angle of the first derivative of the cooling curve at the end of freezing represents Graphite Factor 2 (GRF2). This technology can be useful in predicting defects such as carbides and shrinkage; for example, if TElow or TS fall below the white eutectic temperature, then carbides will, by definition, form. Sillen explains further, A low (TElow) temperature indicates that the nucleation is less effective (and) also means increased risk for chill and macro-shrinkage If TS is too low it can (identify a risk for) inverse chill Recalescence is a measure of the initial crystallisation rate of eutectic the precipitation of eutectic causes the temperature to increase high recalescence (may indicate excessive) precipitation rate. The consequence is an expansion... (that) can cause mould wall movement and also contribute to penetration problems. A good inoculant should reduce recalescence (Sillen, 2003). A high GRF1, indicating more eutectic graphite value, means the iron is more shrinkage resistant. A lower value of GRF2 indicates more precipitation of graphite at the very end of freezing, leading to an iron that is more resistant to microshrinkage. Fig. 1. Example of a cooling curve in hypoeutectic iron is illustrated. (From R. Sillen, ATAS Manual, 2003). 58

99 Proceedings of the AFS Cast Iron Inoculation Conference, September 29-30, 2005, Schaumburg, Illinois EXPERIMENTAL PROCEDURE Table 1 lists the inoculants used in this study. Inoculant A Strontium-Bearing 75% FeSi Table 1. Specifications for Inoculants Used in This Study Inoculant B Calcium-Bearing inoculant (note: inoculant based on 50% FeSi at Dalton and 75%FeSi at John Deere) Inoculant C (Strontium, Zirconium)-Bearing 75% FeSi Inoculant D (Manganese, Zirconium, Calcium, Barium)-Bearing Inoculant Inoculant E (Barium, Calcium)-Bearing, 75% FeSi At Dalton-Kendallville (Dalton)(from now on referred to as Foundry 1), GI was melted in a cupola and held in a 40-ton channel induction holding furnace. It was then tapped into a 2000-lb ladle and inoculated during tapping with an addition of 4.1 lbs of Inoculant A per 2000 lbs (0.205wt%) of iron. Immediately after inoculation, the ladle was sampled with chill wedges, test bars (type B), and thermal analysis cups. The ladle was then covered and set aside. Thermal analysis samples were then taken every two minutes for approximately 22 minutes. Additional chill wedges and test bars were poured as well, though only at (approximately) 12 and 22 minutes after inoculation. This was repeated with Inoculants B-E. The actual solidified thermal analysis samples themselves were then examined for cell counts and flake size. This study was repeated at John Deere (from now on referred to as Foundry 2). The actual tests were largely the same, with only a few notable differences: The metal was melted in a coreless induction melting furnace and the holding furnace is an 80-ton channel furnace. Pouring ladles were 6000 lbs. Inoculant was added at a rate of 19.8 lbs per 6000 (0.330wt%). Chill wedges and test bars were poured at different intervals (immediately after inoculation, 4 minutes after inoculation, and 12 minutes after inoculation). The length of fade was shorter due to loss of pouring temperature. Inoculant B was based on 75% ferrosilicon instead of 50% ferrosilicon. Iron chemistry was different as shown in Table 2. Table 2. Chemical Analysis of Dalton (Foundry 1) Iron Compared to John Deere (Foundry 2) Iron Foundry 1 Foundry 2 C Si Mn S P The actual chill samples taken at each foundry also differed slightly. They are depicted in Fig. 2 (not to scale). In an effort to reconcile the two tests, the area of chilled iron was calculated for each sample. 59

100 Proceedings of the AFS Cast Iron Inoculation Conference, September 29-30, 2005, Schaumburg, Illinois Foundry 2 Wedge Foundry 1Wedge Fig. 2. The comparison of chill wedge geometry, used in study, is illustrated. RESULTS FROM FOUNDRY TESTING Due to technical difficulties, this study was unable to draw significant thermal analysis data from the second half of many cooling curves at Foundry 2. As such, determining trends in TS, GRF2 and other such parameters is impossible. Figure 3 shows the drop in pouring temperature over time for both trials. This data represents the very first temperature registered by the thermal analysis cup. The trends show mostly linear drops in pouring temperature. Foundry 1 experiences more heat loss than does Foundry 2. This is likely the result of a smaller pouring ladle at Foundry 1 that could necessitate higher starting temperatures. Figure 4 shows the change in tensile strength over time for all inoculants in both foundries. The ultimate tensile strength (UTS) mostly drops over time at both foundries. This confirms the results found by Datta (Datta, 1977). Inoculant fade appears to be linear with Inoculant A at both foundries. Inoculant E displays a more gradual decline in UTS than Inoculant A with no rise in the middle of the ladle at both foundries. However, Inoculants B, C and D mostly display an improvement in UTS between the first and second samples, followed by deterioration in UTS. This could be a function of inoculant dissolution and distribution within the metal. Alternatively, it could be the result of a loss in pouring temperature causing the tensile bar to approximate a thinner section after a few minutes of hold time, but inoculant fade seems to overwhelm this effect in the long run, resulting in a final drop in UTS. The end result is that Inoculants C and D display the least deterioration in both foundries, though this does not mean their initial or average strengths were the highest. 60

101 Proceedings of the AFS Cast Iron Inoculation Conference, September 29-30, 2005, Schaumburg, Illinois Peak Temperature vs. Time After Inoculation Peak Temperature, Degrees C A B C D E Time After Inoculation, Minutes Fdry 1 Fdry 2 Fig. 3. Peak temperature over time is graphed. Tensile Strength (MPa) vs. Time After Inoculation Tensile Strength, MPA A B C D E Time After Inoculation, Minutes Fdry 1 Fdry 2 Fig. 4. Tensile strength over time is graphed. Figure 5 shows the changes in TL as well as tensile strength over time with each of the six inoculants tested in both foundries. TL and UTS at Foundry 1 appear to reflect that Inoculant B provides less graphitization at Foundry 1, resulting in an artificially low carbon equivalent (CE) condition. It is logical to believe that this would be revealed in GRF1 and this is 61

102 Proceedings of the AFS Cast Iron Inoculation Conference, September 29-30, 2005, Schaumburg, Illinois indeed the case (Fig. 6). The T statistic, based on a 95% confidence level, shows that GRF1 is significantly lower with Inoculant B compared to Inoculant A. These results are further supported by the difference in ACEL (Active Carbon Equivalent) between these two ladles (Fig. 7). The difference in ACEL is only partially the result of the fact that Inoculant B was based on 50% FeSi instead of the more common 75% FeSi; the balance of the difference in ACEL can only be the result of a difference in graphitization between inoculants. Overall, there appears to be a clearer correlation between TL and UTS between inoculants at Foundry 1, with little to no such correlation at Foundry 2 (Fig. 5). Figure 6 shows that GRF1 seems to have no relationship with time after inoculation. Figure 7 shows clearly that inoculant choice impacted ACEL significantly at Foundry 1, with little to no impact at Foundry 2. It also shows that ACEL clearly varies more with inoculant choice at Foundry 1 that at Foundry 2. Maximum flake size was studied because of the impact of this parameter on tensile strength. As stated by Bates, strength was found to be a function of the reciprocal of the square root of the flake length (Bates, 1991). However, this study showed no clear correlation between maximum flake size and tensile strength (Fig. 8). Inoculants A, D and E only displayed flakes size getting smaller with hold time at Foundry 1. This is likely the result of colder pouring temperatures and this result is not reflected in UTS. Furthermore, there was no evidence of systematic changes in maximum flake size between inoculants or over time. Figure 9 shows the changes in eutectic cell count over time. It suggests that the Foundry 1 iron dropped in cell count over time while the Foundry 2 iron remained mostly unchanged. Both of these observations are independent of inoculant choice. The correlation between cell count and properties (such as chill) is problematic already, and this study does not suggest any relationship between cell count and any other characteristic. TL & Tensile (MPa) vs. Time After Inoculation TL, Degrees C A B C D E Tensile, MPa Time After Inoculation, Minutes Fdry 1 TL Fdry 2 TL Fdry 1 Tensile Fdry 2 Tensile Fig. 5. TL (liquidus) and tensile strength over time is graphed. 62

103 Proceedings of the AFS Cast Iron Inoculation Conference, September 29-30, 2005, Schaumburg, Illinois GRF1 vs. Time After Inoculation GRF A B C D E Time After Inoculation, Minutes Fdry 1 Fdry 2 Fig. 6. GRF1 over time is graphed. ACEL vs. Time After Inoculation ACEL A B C D E Time After Inoculation, Minutes Fdry 1 Fdry 2 Fig. 7. Active carbon equivalent (ACEL) over time is graphed 63

104 Proceedings of the AFS Cast Iron Inoculation Conference, September 29-30, 2005, Schaumburg, Illinois 1000 Maximum Flake Size & Tensile (MPa) vs. Time After Inoculation 300 Maximum Flake Size, microns A B C D E Tensile Strength, MPa Time After Inoculation, Minutes Fdry 1 Flake Size Fdry 2 Flake Size Fdry 1 Tensile Fdry 2 Tensile 0 Fig. 8. Maximum flake size and tensile strength over time are graphed. Eutectic Cell Count vs. Time After Inoculation Cell Count, Cells per sq. mm A B C D E Time After Inoculation, Minutes Fdry 1 Fdry 2 Fig. 9. Eutectic cell count over time is graphed. 64

105 Proceedings of the AFS Cast Iron Inoculation Conference, September 29-30, 2005, Schaumburg, Illinois Figure 10 illustrates the relationship between TElow and chill level at Foundry 1. Chill level clearly rises over time regardless of inoculant. Inoculant A provides the best chill resistance regardless of time interval with Inoculant C in second place. Inoculants B, D and E show inferior chill resistance in the short run as well as the long run. In general terms, Inoculant D provides the highest TElow and Inoculant B provides the lowest TElow at Foundry 1. Over time, TElow shows the most change over time when using Inoculants A and D and very little change with the other three inoculants. This discrepancy between TElow and chill level suggests a section size effect between the chill wedge and the thermal analysis cup. There is some scatter in both data sets at Foundry 2. A rise in chill level corresponds with a drop in TElow for Samples B C, and E (Fig. 11). Again, Inoculant A showed the lowest tendency to chill both in the long and short runs. Inoculants B and C provided the least chill resistance. Inoculant E provides the highest TElow followed by inoculant A. Figure 12 shows the relationship between recalescence and time. Inoculant B appears to display an increase in recalescence at both foundries, as compared to the other inoculants studied. Inoculant D displays a decrease in recalescence over time at both foundries. No firm trends are apparent with the other inoculants. Figure 13 represents a scatter chart made with data points from both foundries. The heats selected for this chart, by definition, were heats in which a valid chill wedge and a valid thermal analysis sample were both taken. Despite the discrepancy between TElow and chill discussed above, Fig. 13 suggests that a larger data sample might generate a clearer relationship between TElow and chill level independent of inoculant choice. This relationship is clearly stronger at Foundry 2 than at Foundry 1. TElow & Chill Area vs. Time After Inoculation - Foundry Chill, sq. mm A B C D E TElow, C Time After Inoculation, Minutes Fdry 1 Chill Fdry 1 TElow Fig. 10. This graph illustrates TElow and chill over time at Foundry 1. 65

106 Proceedings of the AFS Cast Iron Inoculation Conference, September 29-30, 2005, Schaumburg, Illinois TElow & Chill Area vs. Time After Inoculation - Foundry 2 Chill, sq. mm TElow, C 5.0 A B C D E Time After Inoculation, Minutes 1130 Fdry 2 Chill Fdry 2 TElow Fig. 11. This graph illustrates TElow and chill over time at Foundry 2 R vs. Time After Inoculation Recalescance, Degrees C A B C D E Time After Inoculation, Minutes Fdry 1 Fdry 2 Fig. 12. Recalescence over time is graphed. 66

107 Proceedings of the AFS Cast Iron Inoculation Conference, September 29-30, 2005, Schaumburg, Illinois Area of Chill vs. TElow TElow, C CONCLUSIONS Chill, sq. mm Fdry 2 Fdry 1 Fig. 13. Scatter chart of TElow Chill, independent of time and inoculant, is illustrated. This study has clearly shown that the selection of the correct inoculating system has an influence on the fading characteristics of the nucleation level in the iron. Two sets of conditions have been studied and these have given significantly different results. One of the main variables between the foundries was the base sulfur (S) level of the iron. Many factors affecting inoculation have been documented. However, advances in measurement techniques, such as thermal analysis methods, have enabled more accurate measurement to be made. Much more work needs to be done industry wide in this field, although it is apparent from this study that each individual foundry has it s own set of unique circumstances which dictate the choice of an inoculant. Foundry 1 experiences more temperature loss than does Foundry 2, but no definitive conclusion can be readily drawn from the effect of pouring temperature on inoculant fade. This is because pouring and inoculation temperature, while quantified here, is only one of the different operating conditions between the plants. This parameter could be an avenue of further research. At both the medium and high S levels, the rate of loss of UTS appears to be fairly consistent. Inoculant C shows the lowest rate of UTS deterioration. With three of the inoculating systems, including inoculant C, there appears to be a slight increase in the UTS during the first few minutes following the addition of the inoculant. While one could postulate that this is due to the greater rate of heat loss from the iron during this period, no conclusion may be drawn from the number of results available. Further work in this area needs to be done. Inoculant B provided a lower level of graphitization relative to the other inoculants at Foundry 1. This is reflected in the values seen for TL, GRF1 and ACEL. This artificially low ACEL led to a higher UTS value for Inoculant B. However, Inoculant B also generated higher chill levels and recalescence. The Foundry 1 iron displayed a clearer correlation between TL and UTS than did the Foundry 2 iron. ACEL is clearly impacted by inoculant choice at Foundry 1, and this impact appears to be smaller at Foundry 2. This study did not show a meaningful correlation between maximum flake size and UTS, even though the study by Bates found otherwise (Bates,1991). Conclusions about changes in flake size between inoculants, foundries and over time are uncertain. As stated earlier, the correlation between cell count and properties (such as chill) is problematic already and this study does not suggest any relationship between cell count and any other characteristic. 67

108 Proceedings of the AFS Cast Iron Inoculation Conference, September 29-30, 2005, Schaumburg, Illinois Inoculant A performed best in chill resistance at both shops. There is a relationship between TElow and chill level in both foundries, though this is much clearer at Foundry 2 than at Foundry 1. Recalescence is a good measure of the effectiveness of an inoculant. As explained by Sillen earlier in this paper, too high a recalescence can result in mold wall movement and subsequent penetration and/or shrinkage problems (Sillen, 2003). In the examples examined in this study, inoculants A, C and E showed lower recalescence values. Inoculant E gave a good set of values at the lower S level. Foundries need to decide on their priorities, such as chill, UTS, shrinkage and fade as the main charateristics normally in deciding the choice of inoculant. This study, while limited, indicated that fading is to be taken seriously because chill, UTS and other properties will alter over time to a greater or lesser degree dependant on the choice of inoculant. Consistency in castings poured from a single ladle will vary as demonstrated and it is the job of the foundry engineer to minimize these differences to reduce costs and variation in post foundry operations. REFERENCES 1. Bates, C. E., Tucker, J. R., Starrett, Composition, Section Size, and Microstructural Effects on the Tensile Properties of Pearlitic Gray Cast Irons, American Foundry Society Research Report Number 5 (1991). 2. British Cast Iron Research Association, Comparator Charts for Counting Eutectic Cells, BCIRA Broadsheet 94-2 (1974). 3. Chisamera, M., Riposan, I., Stan, S., Skaland, T., Investigation of Effect of Residual Aluminum on Solidification Characteristics of Un-Inoculated and Ca/Sr Inoculated Gray Irons, AFS Transactions (2004). 4. Datta, N. K., Influence of Ladle Inoculation and Holding Time on Structure and Mechanical Properties of Gray Iron Melted in Channel Furnaces, AFS Transactions, pp (1977). 5. Fuller, A. G., Fading of Inoculants, Proceedings of the Conference on Modern Inoculating Practices for Gray and Ductile Iron, pp , Rosemont, IL (1979). 6. Olsen, S. O.; Skaland, T.; Hartung, C., Inoculation of Grey and Ductile Iron A Comparison of Nucleation Sites and Some Practical Advises, 66th World Foundry Congress, pp Istanbul, (2004). 7. Riposan, I., Chisamera, M., Stan, S., Skaland, T., Onsoien, M. I., Analyses of Possible Nucleation Sites in Ca/Sr Over- Inoculated Gray Irons, AFS Transactions (2001). 8. Rundman, K. B., Some Observations on the Effect on Inoculation on the Tensile Properties of Gray and Ductile Cast Iron, International Inoculation Conference Proceedings (1998). 9. Sillen, R., ATAS Instruction 22 - Testing Inoculation, ATAS Manual (2003). 10. Sillen, R., The Active Carbon Equivalent, ATAS Newsletter (2003). 11. Skaland, T., Fading of Inoculation in Cast Iron, Casting Congress, Prague, Czechoslovakia (1992). 68

109 INOCULATION OF GREY AND DUCTILE IRON A COMPARISON OF NUCLEATION SITES AND SOME PRACTICAL ADVISES Svein Oddvar Olsen*, Torbjørn Skaland*, Cathrine Hartung* Elkem ASA, Foundry Products Division, NORWAY ABSTRACT The objective of this paper is to review some important aspects related to cast iron inoculation. Important conditions in the production of cast iron are described and characteristic microstructures and mechanical properties exemplify the difference between inoculated and un-inoculated irons. Principal mechanisms of inoculation and graphite nucleation in grey and ductile irons are described. The findings are based on advanced electron microscopy studies of micro-particles as heterogeneous nucleation sites for graphite. Effects of minor alloying elements such as Ca, Ba, Sr, and Al are explained as well as the critical role of oxygen and sulphur in the graphite nucleation process. (1) Finally, the mechanisms of inoculant fading are explained and some practical advises for optimized and reproducible inoculation given. Keywords: Cast iron, inoculation, graphite nucleation, fading INTRODUCTION In the production of quality cast irons the inoculation process is of vital importance. When comparing un-inoculated and inoculated irons, differences in microstructure are easily revealed, which again will strongly affect the final mechanical properties of the casting. Through inoculation the graphite nucleation 1

110 and eutectic undercooling of the iron can be controlled and this will be of crucial assistance in giving the iron its required service properties. WHAT IS INOCULATION? Inoculation is a means to control and improve the microstructure and mechanical properties of cast iron. The inoculation process will provide sufficient nucleation sites for the dissolved carbon to precipitate as graphite rather than iron carbides (cementite). The most common inoculant is a ferrosilicon based alloy with small and defined quantities of either Ca, Ba, Sr, Zr, rare earth s, and Al. Examples of un-inoculated and inoculated irons are shown in Figure 1 and the influence of inoculation on mechanical properties in Figure 2. Consequently, the effects of grey and ductile iron inoculation are improved machinability, increased strength and ductility, reduced hardness and section sensitivity and a more homogeneous microstructure. Typically, inoculation also reduces the tendency for solidification shrinkage formation. GREY IRON INOCULATION The grey iron microstructure is normally determined by the base iron composition, the solidification cooling rate and the inoculation process. Figure 3 shows different grey iron microstructures as a function of solidification undercooling. Controlled undercooling promote the normally desired type A flake graphite, characterised by randomly distributed graphite flakes in a fully pearlitic matrix. The role of inoculation is to provide sufficient nucleation sites for graphite that is activated at low undercooling, thus promoting the formation of good type A graphite structures. Hence, inoculation is a means to change the otherwise undesired graphite forms into a more desired form. It has been found that balancing manganese and sulphur is important for the machinability of grey iron. Experiences have also resulted in a recommended ratio between manganese and sulphur in grey iron. Manganese should be adjusted to balance the residual sulphur level according to the following relationship: %Mn = %S x [1] Table 1 shows the influence of Mn:S ratio on eutectic cell count and chill tendency in un-inoculated condition. This relationship also suggests that MnS inclusions could act as nucleation sites for graphite flakes. The crystal lattice match between cubic MnS and hexagonal graphite is actually quite good. It is also known that if the sulphur content is less than about 0.03%, although balanced properly by Mn, the number of MnS inclusions will be insufficient to produce effective nucleation of good type A graphite structures. 2

111 Further, scanning electron microscope (SEM) investigations has shown that in uninoculated and inoculated irons the number of MnS inclusions are about the same, but the distribution tends to be somewhat different. In un-inoculated iron, MnS inclusions are predominantly found between the primary austenite dendrites while in inoculated iron these inclusions are found to be more randomly distributed throughout the iron matrix. This suggests that inoculation is affecting the formation sequence of MnS particles during cooling and solidification. Figure 4 shows an example of an inclusion that has acted as nuclei for graphite flake. The figure shows the distribution of relative intensity (X-ray mapping) of the different constituent elements. From this analysis it can be seen that a Mn(X)S compound with a core of Al/Ca oxides is present as graphite nucleation site. Further studies show that Ba and Sr can act the same way as Ca and Al. This means that the active elements in the inoculant, Ca-Ba-Sr-Al, primarily will form stable oxides that can act as nuclei for the Mn(X)S phase to precipitate on. The sulphide particle will again be the preferred nuclei for graphite flakes to grow from upon solidification. For the foundry it is therefore very important that the Mn:S ratio is adjusted to the right level and that some oxygen is also available for the inoculating elements to combine with in the production of grey iron. (3, 6, 7, 8) DUCTILE IRON INOCULATION Figures 5 shows examples of microstructure in inoculated and un-inoculated ductile irons. The extensive chill (carbides) in un-inoculated condition will destroy the mechanical properties of this iron and make it very difficult to machine such castings. Hence, inoculation is a crucial requirement for most ductile iron processes simply to make machinable castings. In ductile iron the nodularising treatment will influence inoculation efficiency and therefore it is important to select the correct treatment process and magnesium bearing material. Formation of a high number of small micro-inclusions during magnesium treatment is an advantage, and Figure 6 shows how nodularising provides the basis for an effective subsequent inoculation. Also, Figure 7 shows how investigations of micro-inclusions at different magnifications have led to the discovery of the nucleation site for graphite in ductile iron. During nodularising, numerous inclusions are formed with a sulphide core and an outer shell containing complex magnesium silicates. Such micro-inclusions will however not provide effective nucleation of graphite because the crystal lattice structure of magnesium silicates does not match well with the lattice structure of graphite. However, after inoculation with a ferrosilicon alloy containing Ca, Ba or Sr, the surface of the magnesium silicate micro-particles will be modified and other complex Ca, Sr, or Ba silicate layers will be produced (see Figure 8). Such silicates have the same hexagonal crystal lattice structure as graphite, and due to very good lattice mach will therefore act as effective nucleation sites for graphite nodules to grow from during solidification. (1) 3

112 FADING OF INOCULATION EFFECT The gradual loss of inoculation effect during liquid metal holding is well known to the foundry people, and this fading of inoculation will eventually result in carbide formation and poor graphite structures if the iron is held for prolonged times before pouring. The reason for this fading loss is coarsening and growth of micro-inclusions, also called the Ostwald Ripening Effect. The driving force for this coarsening is a reduction in the specific surface area of inclusions, thus reducing the total energy of the system. The volume fraction of non-metallic inclusions will however remain unchanged due to the high particle phase stability. (10) This fading effect is very fast just after inoculation when distances between micro-particles are short, and is much more severe to the iron quality than fading losses of residual magnesium. Figures 9 and 10 show this inoculation fading effect by particle coarsening and a reduction in the number density of potential nucleation sites during time. The fading rate of inoculation is directly related to the diffusion rate of reactive elements through the liquid metal. INOCULATION METHODS The required addition rate of an inoculant to liquid iron is very much depending on where and when it is to be introduced. Figure 11 shows an example of substantial reductions in addition rate when going from an early addition to the transfer ladle to a late addition to the metal stream. At transfer, the required inoculant addition rate may be as high as 1 wt%, while the alternative late instream inoculation may require only 0.1 wt% addition still providing sufficient or even better inoculation effectiveness. This is primarily due to the late addition giving much less time available for particle coarsening and fading effects. (2, 4, 5) INOCULATION ELEMENTS The main finding from studies of micro-inclusions as nucleation sites for graphite is that the key nucleating elements in the inoculant are Ca, Ba, Sr and Al. The ferrosilicon alloy itself is only the carrier material of these critical active elements, but is also needed in order to give these minor elements the right concentration and solubility for an optimum inoculation performance. COMPARISON OF ACTIVE MICRO-INCLUSIONS In grey iron it is found that small oxide particles will acts as the nuclei for Mn(X)S that again will be the decisive nuclei for graphite flakes to grow from at small undercoolings. In ductile iron however, a stable sulphide core is found to be the nuclei for complex silicates that again will be modified by the active elements in the inoculant before it can act as a potent nuclei for graphite. However, the 4

113 same specialty ferrosilicon inoculant materials are still being used for both grey and ductile irons and the main reason is that key elements are highly reactive and can form various types of micro-inclusions, some of them being favourable sites for graphite to grow from during solidification. SUMMARY The principal inoculation mechanisms are quite different in grey and ductile irons. In grey iron, a stable oxide will be the primary nuclei for manganese sulphide precipitation that again will nucleate graphite flakes of good type A form. In ductile iron, a sulphide is the nuclei for complex silicates that again will nucleate a high number of graphite nodules. The same inoculant materials can however be used successfully in both type of irons, since the reactive elements such as Ca, Ba, Sr and Al are all strong oxide, sulphide and silicates formers in both grey or ductile irons. The inoculant fading effect is connected to diffusion rate, growth and coarsening, and a general reduction in the number density of micro-inclusions as nucleation sites for graphite. In order to obtain a sound and reproducible iron production process some critical inoculation factors will have to be controlled properly. For grey iron one should pay special attention to the following factors: 1) The Mn:S ratio should be maintained at the same level every time and sulphur should preferentially be kept at minimum 0.05%. 2) Aluminium is found to be an important part of the nucleus core and should be adjusted and kept at controlled levels every time. Recommended residual Allevel in grey iron is 0.005% % for optimum inoculation effectiveness. 3) There should be a certain oxygen level in the base iron from fresh metal processing. The use of some rusty raw materials may assist in providing a good oxygen potential. 4) Pouring time after inoculation should be minimized in order to keep fading losses under control. 5) Use an inoculant with defined chemical composition and sizing. For ductile iron, the following factors must be controlled: 1) The magnesium treatment process reactivity should be controlled and minimized. A violent treatment process will provide less potential nucleation sites and more difficult conditions for powerful inoculation effectiveness. 5

114 2) There should be a certain oxygen level in the base iron from fresh metal processing. The use of some rusty raw materials may assist in providing a good oxygen potential. 3) The sulphur content should be kept low and constant. Preferential range for ductile iron is to 0.015% base iron sulphur content. 4) Pouring time after inoculation should be minimized in order to keep fading losses under control. 5) Use an inoculant with defined chemical composition and sizing. REFERENCES 1) T.SKALAND A model for graphite formation in ductile iron. Ph.D Thesis 1992 : 33, The Norwegian Institute of Technology, Norway (1992) 2) R.ELLIOTT Cast Iron Technology, 1988, London, UK, Butterworths 3) I.RIPOSAN, M.CHISAMERA, S.STAN, T.SKALAND, M.ONSOIEN Analysis of possible nucleation sites in Ca/Sr over-inoculated grey irons. AFS Transactions vol. 109, 2001, pp ) S.I.KARSAY Ductile Iron Production, QIT, ) Elkem Technical Information Sheets No ) I.RIPOSAN, M.CHISAMERA, S.STAN, T.SKALAND Graphite nucleants (micro-inclusions) characterization in Ca/Sr inoculated grey irons. SPCI 7 Science and Processing of Cast Iron International Conference, Barcelona, Spain, ) J.K.SOLBERG, M.ONSOIEN Nuclei for heterogeneous formation of graphite spheroids in ductile cast iron. Material Science and Technology, vol 17, October 2001, pp ) F.NEUMANN Theorien über das Impfen. Giesserei, No.14, July 1996, pp. 9) ASM Metals Handbook, vol 1, tenth edition, 1990, pp. 6 10) J.D.VERHOEVEN, Fundamentals of Physical Metallurgy, Chapter 8 and 10, John Wiley & Son, Inc,

115 Grey Iron Un-inoculated Inoculated Ductile Iron Figure 1: Examples of structures in un-inoculated and inoculated irons. (5) Un-inoculated Inoculated Example: Tensile: 200 MPa Elongation: 0 % Hardness: 700 HB Example: Tensile: 450 MPa Elongation: 10 % Hardness: 180 HB Control of structure and properties by minimizing undercooling and providing nucleation of graphite during solidification Figure 2: Effects of inoculation on typical mechanical properties of ductile iron. (5) 7

116 Figure 3: Graphite structures as a function of eutectic undercooling in grey iron. Table 1: Experimental results showing effects of Mn and S contents and Mn:S ratio on eutectic cell count and chill level in grey iron. % Mn % S Mn:S Cell Count [mm] Chill [mm]

117 a) SEM micrograph of (Mn,X)S compound and graphite flake b) Distribution of Carbon c) Distribution of Manganese d) Distribution of Sulphur e) Distribution of Aluminium f) Distribution of Calcium g) Chemical composition along a cross line through the (Mn,X)S compound. Figure 4: X-ray mapping showing composition of micro-inclusion as nuclei for graphite flake in grey iron. (3, 6) 9

118 Poor Inoculated Inoculated Good Inoculation Improved Recovery Reduced Mg-Addition Property Uninoculated Inoculated Proof Strength R p0.2 Not detected MPa Tensile Strength R m < 300 MPa MPa Elongation A 5 Not detected 3-30 % Brinell Hardness HB > Nodule Count 10 mm section < 50 per mm 2 > 150 per mm 2 M icrostructure ASTM Classification Carbidic Ferritic and/or Pearlitic Figure 5: Examples of microstructure and mechanical properties in un-inoculated and inoculated ductile irons. (5) Nuclei Slag Treatment Reactivity 5 µm Size Distribution Figure 6: Schematic representation of size distribution of inclusions as micronuclei and slag in treated ductile iron. 10

119 a) 100x (optical) b) 1,000x (SEM) XO SiO 2 or XO Al 2O 3 2SiO 2 Where X = Ca, Sr or Ba c) 70,000x (TEM) d) Schematic composition Figure 7: Ductile iron micro-inclusions at different magnifications and the schematic composition of nucleation sites for graphite. (1) Mg-treatment Inoculation Major constituent phases: Shell: MgO SiO 2 2MgO 2SiO 2 XO SiO 2 or XO Al 2O 3 2SiO 2 Core: MgS CaS Where X = Ca, Sr or Ba Figure 8: Schematic representation of micro-inclusion composition in treated ductile iron before and after inoculation. (1) 11