Casting and Solidification
Effect of Solute Elements on Boron Segregation in Boron-Containing Steels
2020 Volume 60 Issue 1 Pages 92-98
Details
2020 Volume 60 Issue 1 Pages 92-98
The addition of boron to steel alloys results in an increase in both hardenability and casting defects. The casting difficulties are predicted to stem from a metatectic reaction, δ + γ → L + γ, where a fully solidified material begins to locally remelt as the temperature decreases. Another possible source of casting defects is a boride-rich phase that is predicted to remain liquid at low temperatures. To experimentally determine which reaction is the likely source of the casting defects, the predicted reactions and the effect of solute elements on those reactions are investigated. Levitation zone melting is used to control segregation in a ternary Fe–C–B alloy and a commercial 22MnB5 alloy. Carbon segregation and a peritectic reaction result in a peritectic jump during directional solidification where the first directionally solidified (DS) zone undergoes δ-bcc solidification followed by a peritectic jump to steady state planar solidification of γ-fcc in the second DS zone. The presence of other solute elements in the zone melted 22MnB5 alloy lead to a breakdown in the planar solidification front before steady state solidification could be achieved in the second DS zone. With a cellular solid/liquid interface, boron-rich intercellular liquid formed low melting iron boro-carbide particles. The controlled solidification conditions in a levitation zone melter were unable to prevent ~0.003 wt% boron from segregating to high enough levels to form boride particles. Therefore, it is likely that during commercial casting, the formation of the low melting boride phase from interdendritic segregation is a key source of the casting issues.
Small amounts of boron, between 0.002 and 0.005 wt%, can increase the hardenability of steel alloys.1) However, the addition of boron also makes the steel difficult to commercially produce, specifically via continuous casting. The casting defects that develop include surface cracks, internal half-way cracks, and casting breakouts.2) The source of the casting defects are anticipated to stem from a metatectic reaction. In the binary iron-boron system, a metatectic reaction, δ → L + γ, occurs resulting in a solid piece of material locally remelting as it cools.3) Based on previous experimental work, the metatectic reaction occurs in the binary system between 0.0025 and ~0.06 wt% B at ~1385°C.4) When carbon is added to iron and boron, the metatectic reaction is predicted to occur as δ + γ → L + γ.5,6) With 0.05 wt% C, for example, the TCFE8 thermodynamic database (Thermo-Calc: Steels/Fe alloys database) predicts a metatectic reaction to occur from 0.002 to 0.0055 wt% B at temperatures between 1420°C and 1458°C, shown in Fig. 1(a).5,6) As the carbon content increases, the composition range of the predicted metatectic reaction shrinks, eventually disappearing around 0.2 wt% C, Fig. 1(b). Aside from remelting, another source of casting defects could be the presence of liquid at low temperatures from a low melting boride phase resulting from microsegregation during solidification.
Blazek et al. studied the effects of boron on the solidification behavior of steels using 45 kg ingots cast in a vacuum induction furnace.3) Examination of the microstructure on the ingot’s surface by optical microscopy revealed a phase with a white contrast at the prior austenite grain boundaries. However, this phase was not found in the boron free ingots. Furthermore, the amount of this white phase increased as boron content increased, leading Blazek et al. to conclude that boron segregates to the grain boundaries forming a boron rich phase that could undergo the metatectic reaction.3) Subsequent examinations of Fe–C–B alloys with differential scanning calorimetry (DSC)7) and confocal scanning laser microscopy2) produced results that deviated from commercial thermodynamic database predictions.
This study aims to experimentally investigate the source of the casting defects that form during commercial production of boron-containing steels. The predicted reactions and the effect of solute elements on those reactions are studied through levitation zone melting. Initially, zone melting is utilized to investigate solute segregation in a Fe–C–B ternary alloy. This successful experimental procedure is then applied to a commercially available boron-containing steel alloy, 22MnB5, to further investigate the source of casting defects through controlled solute segregation.
A simple ternary alloy containing nominally Fe, 0.23 wt% C, and 0.003 wt% B was cast at the East Chicago (USA) Center of ArcelorMittal Global R&D. Electrolytic iron and vacuum induction melting were utilized to limit impurities in the cast ingot. An as-cast sample of the commercially produced 22MnB5 steel alloy was also provided. The standard composition of 22MnB5 is given in Table 1.8) Both alloys were machined into ~20 mm diameter bars and levitation zone melted.
| C | Mn | P | S | Si | Al | Ti | B |
|---|---|---|---|---|---|---|---|
| 0.2–0.25 | 1.1–1.4 | ≤ 0.025 | ≤ 0.008 | 0.15–0.35 | ≥ 0.015 | 0.02–0.05 | 0.002–0.005 |
Levitation zone melting is a containerless processing method as shown in Fig. 2(a). This process is similar to vertical float zone melting; however, a 20 mm diameter liquid zone is too large to be supported solely by surface tension.9) Levitation zone melting utilizes induction coils and an eddy current plate to heat, levitate, and constrain the liquid zone to a height similar to the bar diameter. A melting solid/liquid interface is present at the top of the hot zone while a freezing solid/liquid interface is located at the bottom. During processing, the size of the liquid zone is visually monitored to keep the size stable. If the liquid zone appears to grow or shrink, the power is adjusted manually to return the liquid zone back to its original size and shape. To maintain a smooth solid/liquid interface, the bottom solid bar is rotated leading to a well-mixed liquid zone via turbulent flow when combined with induction stirring.10,11) In this study, the ingots were moved through the assembly at a rate of ~13 mm/h.
(a) The levitation zone melter at Purdue University. Induction coils are used to heat, levitate, and constrain the liquid while the ingot is rotated. A growth rate of ~13 mm/h pulls the ingot through the hot zone. (b) Depending on the partition coefficient of the alloying element, k, a single pass in the zone melter will result in a composition profile containing two distinct regions: the directionally solidified (DS) zone and the last zone to solidify. The composition profile can be calculated as described by Pfann.12) (Online version in color.)
The levitation zone melting process can result in a large range of compositions and microstructures in one ingot. Depending on the partition coefficient of the alloying element, k, a single pass in the zone melter, with a ratio of liquid zone length to total bar length of 0.125, will result in a composition profile such as those shown in Fig. 2(b). This profile can be broken down into two distinct zones, the directionally solidified (DS) zone and the last zone to solidify. As k approaches unity, the DS zone can be further broken down into a transient zone, where initial solidification occurs, followed by steady state planar solidification at the original alloy composition, Co. As k decreases, steady state planar solidification will only be achieved if the liquid zone length is small compared to the total length of the processed bar. The solid composition along the DS zone can be calculated as described by Pfann.12) The solute balance for a zone melted ingot resides in the last zone to solidify where the composition profile can be calculated from the Scheil equation.12)
While the last zone contains higher solute concentrations than the DS zone, it also solidifies faster as the induction power is reduced at the end of the run. The cooling rate for the last zone, as measured by a high temperature infrared pyrometer, was found to be to be ~55°C/min for the ternary alloy and ~30°C/min for the commercial alloy. The cooling rate of the DS zone was measured at ~3°C/min with a thermal gradient through the solid/liquid interface of ~15°C/mm. Before the reduction of induction power at the end of zone melting, the DS portion of the bar behind the molten zone should consist of high temperature γ-fcc phase. During powering down of the system, the cooling rate through the γ-fcc to α-bcc phase transition in the DS portion should occur at a similar rate as the cooling rate in the last zone.
After processing, the zone melted ingots were cross-sectioned, polished, and etched in ~3% nital solution. Various regions along the ingot were imaged under an optical microscope and a FEI Quanta 3D FEG Dual-beam Scanning Electron Microscope (SEM). For bulk composition measurements, spark optical emission spectroscopy (OES) was performed at the East Chicago (USA) Center of ArcelorMittal Global R&D. The detection limits of the system are 0.0005 wt% boron and 0.01 wt% carbon with a machine error of 0.0002 wt% B and 0.003 wt% C. The spatial resolution of the spark OES system is ~5 mm. Multiple profiles were performed on both ingots, and all measurements are included in each profile. For small scale qualitative chemical analysis, electron dispersive X-ray analysis (EDX) was utilized in the SEM. The system consists of an Oxford INCA Xstream-2 silicon drift detector with Xmax80 window allowing for the qualitative detection of light elements, such as boron.
The ternary alloy and the commercial 22MnB5 alloy were levitation zone melted, cross-sectioned, polished, and etched. Three distinct regions can be identified, as shown in Fig. 3. There are two directionally solidified (DS) zones, labeled 1 and 2, followed by the last zone to solidify, labeled 3. The two DS zones can be explained by carbon segregation and the presence of a peritectic reaction.
Macro images of the zone melted (a) Fe–C–B and (b) 22MnB5 alloys after cross-sectioning, polishing, and etching showing that the change in microstructure is evident on the macro scale. The two DS zones are labeled 1 and 2 while the last zone to solidify is labeled 3. (Online version in color.)
To examine carbon segregation, the composition profile across the zone melted Fe–C–B ingot was calculated as described by Pfann.12) For the calculations, the liquid zone length was 12.7 mm, and the average partition coefficients from the Thermo-Calc TCFE8 database are listed in Table 2.5,6) The calculated profiles are shown as solid lines in Fig. 4. The circles along the profile show experimental measurements obtained from spark OES with hollow circles representing measurements below the detection limit of the system. The three zones labeled in Fig. 3(a) are also labeled along each composition profile. The experimental measurements of the bulk carbon composition match with the calculated profile. The boron experimental measurements fall below the detection limit of 5 ppm B except measurements taken from the last zone to solidify. The calculated boron composition is also below 5 ppm B until it reaches the last zone where the average composition is 0.016 wt% B. The average calculated carbon composition in the last zone is 0.70 wt% C.
| Carbon | Boron | |
|---|---|---|
| kδ,avg | 0.17 | 0.016 |
| kγ,avg | 0.33 | 0.0025 |
The calculated and experimental composition profile for (a) carbon and (b) boron across the zone melted Fe–C–B alloy. The hollow circles show measurements that were below the detection limit of the spark OES system. The two DS zone are labeled 1 and 2 while the last zone is labeled 3. (Online version in color.)
The two DS zones shown in Fig. 3(a) correspond to a change from a planar δ-bcc growth front to γ-fcc as a result of a peritectic reaction. Based on the Fe–C binary phase diagram shown in Fig. 5, the zone melted alloy with Co of 0.23 wt% C is hyperperitectic. At the start of zone melting, the alloy will solidify primarily as δ-bcc. For a planar growth front, the solid/liquid interface temperature will decrease down the δ-solidus towards steady-state composition, C0, labeled “a” in Fig. 5. Once the peritectic temperature is reached with δ-bcc having composition Cδ, γ-fcc at composition Cp will overtake the growth front, covering the primary δ-bcc phase and stopping the peritectic reaction. This results in a discrete jump in composition from δ-bcc at Cδ to γ-fcc at Cp defined as a peritectic jump in composition, labeled “b” in Fig. 5.13) The peritectic jump is followed by γ-fcc growth as it solidifies down the γ-fcc solidus to reach steady state growth at Co, labeled “c” in Fig. 5. As the ingot freezes at Co, the liquid remaining in the last zone will have a composition of CL. This fully explains the carbon composition profile of the Fe–C–B zone melted ingot shown in Fig. 4(a). The first DS zone undergoes solidification of δ-bcc followed by a peritectic jump in composition as γ-fcc begins to grow at approximately steady state in the second DS zone while the last zone to solidify freezes with an average composition of CL.
Partial Fe–C binary phase diagrams illustrating planar solidification of a hyperperitectic alloy with original composition of Co. Initially, solidification will proceed down the δ-solidus towards steady state at Co, labeled “a”. At the peritectic temperature, γ-fcc will overtake the growth front resulting in a peritectic jump in composition from Cδ to Cp, labeled “b”. Finally, solidification continues down the γ-solidus towards steady state at Co, labeled “c”. (Online version in color.)
While the three distinct macroscopic regions of the zone melted ternary ingot can be attributed to carbon segregation, the presence of boron affects the development of the microstructure. Shown in Fig. 6 are SEM images of the polished and etched zone melted Fe–C–B ingot. The first DS zone, labeled 1, has a calculated average composition of 0.07 wt% C and 9.4E-5 wt% B. Based on the TCFE8 predicted phase diagrams and previous work in the Fe–B binary system, no remelting from a metatectic reaction should occur at these low boron compositions.4,5,6) The microstructure consists of ferrite and pearlite, labeled in Fig. 6(1). The second DS zone has the steady state composition of ~0.23 wt% C. No metatectic reaction is predicted from TCFE8 at this carbon level.5,6) As shown in Fig. 6(2), the second DS zone consists of a bainitic structure. At a carbon composition of ~0.23 wt% C, the Fe–C phase diagram predicts austenite to transform to pearlite and proeutectoid ferrite, as shown in Fig. 5. If the ingot was a binary Fe–C alloy, the second DS zone would be expected to look similar to the first DS zone with more pearlite. However, there is clearly a change in morphology as the sample cooled from γ-fcc in the second DS zone during shutdown of the zone melting system. Since boron is present in this ingot, even though it was below the experimental detection limit of 5 ppm B, it must play a role. The presence of boron below 5 ppm results in the formation of bainite instead of the ferrite and pearlite predicted from the Fe–C binary system. Also, clean prior-austenite grain boundaries are seen along the length of the second DS zone. The last zone contains carbon levels nearing the eutectoid composition, CL in Fig. 5, leading to the formation of a mostly pearlitic microstructure, Fig. 6(3).
SEM images taken from (1) the first DS zone, (2) the second DS zone, and (3) the last zone to solidify of the Fe–C–B zone melted ingot with regions of interest labeled. (Online version in color.)
The last zone also contains higher boron content which is predicted to no longer be soluble in the bulk solid.5,6) Investigating this zone in the SEM reveals an iron boro-carbide phase along the interdendritic regions where the last liquid would have solidified, Fig. 7. The EDX spectra shows the bulk of the last zone contains Fe and C, black spectra labeled 1, while the grain boundary phase contains Fe, C, and B, red spectra labeled 2. The boron-rich phase is most likely Fe23(B,C)6 that has been identified in other boron-containing steels.14) This is similar to the boron-rich white phase that Blazek et al. saw in their as-cast ingots.3) These iron boro-carbides are only found in the last zone to solidify of the ternary alloy.
SEM images and corresponding EDX spectra from the last zone to solidify in the zone melted Fe–C–B ingot. The bulk, black spectra labeled 1, contains Fe and C while the grain boundary phase, red spectra labeled 2, contains Fe, C, and B. (Online version in color.)
Similar to the ternary ingot, the zone melted 22MnB5 ingot also contained three zones that can be explained by carbon segregation and a peritectic jump. The composition profiles across the ingot calculated from Pfann and the TCFE8 database are shown in Fig. 8.5,6,12) The liquid zone length used in the calculations was 16 mm, and the average partition coefficients calculated from TCFE8 are listed in Table 3. The squares along the profile show experimental measurements from spark OES with hollow squares representing measurements that were below detection limit of the system.
Experimental and calculated composition profiles of (a) carbon and (b) boron along the zone melted 22MnB5 ingot with hollow squares representing measurements that were below the detection limit of the spark OES system. (Online version in color.)
| Carbon | Boron | |
|---|---|---|
| kδ,avg | 0.16 | 0.013 |
| kγ,avg | 0.31 | 0.0026 |
The composition profile along the first DS zone of the zone melted 22MnB5 ingot appears similar to the profile of the first DS zone of the zone melted Fe–C–B ingot. As explained in the previous section, δ-bcc is solidifying down the δ-solidus throughout the first DS zone of both ingots. There is no distinct change in carbon content, and the boron content remains below the detection limit of 5 ppm. The calculated profiles match the experimental measurements in the first DS zone. However, there is a change in microstructure in the first DS zone of the zone melted 22MnB5 alloy, shown in Fig. 9. The start of the 22MnB5 zone melt is comprised of proeutectoid ferrite and pearlite, labeled 1a. This is followed by a distinct change in microstructure to an acicular ferrite with pearlite, labeled 1b. Without a distinct change in carbon content, a change in boron likely plays a role in microstructure formation even though it remains below the detection limit. Recall that boron at these low levels affected the microstructure within the ternary ingot as well.
Optical images taken from the first DS zone of the 22MnB5 ingot. The image on the left, 1a, solidified prior to the image on the right, 1b. As can be seen, there is a distinct change in microstructure within the first DS zone. However, there is not a distinct change in carbon concentration along the length of this zone. (Online version in color.)
The solidification of δ-bcc is followed by a peritectic jump in composition as γ-fcc begins to grow in the second DS zone of both the ternary and 22MnB5 zone melted alloy. In a typical hyperperitectic alloy, the alloy would continuing solidifying down the γ-fcc solidus to reach steady state planar growth at the original alloy composition. As shown in the experimental composition measurements, this does not occur in the second DS zone of the 22MnB5 alloy, labeled 2 in Fig. 8. Instead, both carbon and boron composition profiles follow a similar trend in the second DS zone. The compositions increase to a maximum well above the original alloy composition before decreasing prior to the last zone to solidify. The deviation between the calculated and experimental measurements in the second DS zone is likely due to changes in the size of the liquid zone during processing. As previously explained, the liquid zone is visually monitored to keep the size stable. If the liquid zone begins to shrink, the power is manually adjusted to return the liquid zone back to its original size. As shown in the macro image of the zone melted 22MnB5 ingot, Fig. 3(b), the diameter of the processed bar initially decreases in the second DS zone, coming to a minimum, then increases until the last zone. The change in bar dimensions resulted from problems in keeping the liquid zone size constant during the zone melting run, explaining the reasons for the compositional deviations from the calculated values shown in Fig. 8. For example, if the liquid zone begins to shrink in size, the zone melted ingot is freezing at a faster rate than the original bar is melting. This will result in a liquid zone enriched with solute and an increased bulk composition. The subsequent decrease in composition of the second DS zone is due to an increase in liquid zone size from manual power adjustments.
Microstructural evidence found throughout the length of the second DS zone in the 22MnB5 ingot point to a breakdown of the planar solidification front as the cause of the apparent change in liquid zone size. Due to the increased solute content in 22MnB5, it is likely that the DS growth rate of ~13 mm/h resulted in the formation of a constitutionally supercooled solid/liquid interface leading to a breakdown of the planar growth front into a cellular growth front. The cellular solid/liquid interface allowed for the formation of a boron-rich intercellular liquid along the original austenite grain boundaries. The boron-rich liquid solidified into particles containing Fe, C, and B, most likely Fe23(B,C)6, which can be found along the prior austenite grain boundaries throughout the second DS zone of the 22MnB5 ingot, such as those shown in Fig. 10. The presence of a low melting boron-rich phase throughout this zone during processing likely resulted in an apparent change in the size of the liquid zone. The subsequent manual power adjustments based on the perceived change in liquid zone size resulted in the change in dimensions of the zone melted bar as shown in Fig. 3(b). On the other hand, the second DS zone in the simple Fe–C–B ternary ingot had clean prior austenite grain boundaries and composition profiles that behaved as expected for a hyperperitectic alloy. Given that the controlled solidification conditions of a levitation zone melter were unable to prevent boron from segregating to high enough levels to form boride particles during solidification of a commercial steel alloy containing ~0.003 wt% B, the formation of low melting borides would be expected in a conventional casting.
SEM image taken from the second DS zone of the 22MnB5 ingot showing iron boro-carbide particles along the prior-austenite grain boundaries. (Online version in color.)
This study aimed to experimentally investigate the source of casting defects in boron-containing steels through controlled solute segregation. Comparison of a zone melted Fe–C–B ternary alloy and a zone melted commercial 22MnB5 steel alloy allows conclusions to be drawn regarding the role of solute elements in the solidification behavior of these steels. Carbon segregation and a peritectic reaction result in a peritectic jump in the solid composition during directional solidification of hyperperitectic alloys with a planar growth front. However, the presence of other solute elements in the commercial 22MnB5 alloy resulted in a cellular γ-fcc solidification front. With a cellular solid/liquid interface, boron-rich intercellular liquid formed low melting iron boro-carbide particles which were found along the prior austenite grain boundaries. The controlled solidification conditions in a levitation zone melter were unable to prevent ~0.003 wt% boron from segregating to high enough levels to form boride particles. Therefore, it is likely that during commercial casting, the formation of the low melting boride phase from interdendritic segregation is a key source of the casting issues.
The authors would like to thank the East Chicago (USA) Center of ArcelorMittal Global R&D for the gift that supported this work along with the raw materials. We would also like to thank Keith Aumend at ArcelorMittal USA for his assistance with bulk composition measurements and Dr. Chris Gilpin at Purdue University for his assistance with EDX.
