Abstract
The hot ductility of twin induced plasticity (TWIP), 0·6 wt-C steels containing 18–22 wt-Mn with N levels in the range 0·005–0·023 wt- and the Al additions either low (<0·05 wt-) or high (1·5 wt-) has been examined. Little change in ductility occurred in the temperature range 1100–650°C as the structure was always fully austenitic. Ductility was generally poor (<40), reduction of area values, the best ductility at the higher Al level being given by the steel with the lowest N and S levels. Because the steel is fully austenitic, the ductility is solely dependent on that for unrecrystallised austenite. Therefore, to avoid transverse cracking the volume of second phase particles should be kept to a minimum, i.e. the N should be low to reduce the amount of AlN that can be precipitated out and the S level should be as low as possible to limit the amount of MnS inclusions. Metallographic and TEM studies were carried out and the poor ductility was found to be due to extensive precipitation of AlN at the austenite grain boundaries. Increasing the cooling rate from the melting point to the test temperature from 60 to 180°C min−1 or introducing an undercooling step both led to even worse ductility.
Keywords
Introduction
Twin induced plasticity (TWIP) steels are being developed for automotive industry. 1 These steels have an exceptional combination of strength and ductility and POSCO (Korea) are currently developing such a steel with a tensile strength of 940 MPa and elongation of 60. 1
The mechanism which gives the TWIP steels their exceptional ductility and high strength, depends on having a fully austenitic structure which has a low stacking fault energy. Extensive twinning then occurs in deformation, giving a high work hardening rate, thus preventing ‘necking’ from occurring during the forming operation. This is best achieved by having a high Mn level.1,2
In order to achieve these properties the stacking fault energy must be controlled. If it is too high, twinning will not occur and if too low martensite will form.
Adding 1–2 wt-Al as well as reducing the stacking fault energy also improves the low temperature toughness. More importantly they have been shown to prevent delayed fracture after the steel has been deep drawn. 1 However, Al is known when added in small amounts, ⩾0·04 wt-, to cause transverse cracking problems in peritectic C level steels due to AlN precipitation at the austenite grain boundaries. 3 The influence of these very much higher Al levels on hot ductility in these TWIP steels has not been explored and forms the basis of this paper.
In the simple hot ductility test, conditions are kept as close as possible to those pertaining to the commercial continuous casting operation. The tensile specimen is heated to a high temperature to dissolve all the microalloying elements or as in the present case melted. It is then cooled at the average cooling rate which the strand experiences when cooled down to the test temperature range of 1000 to 700°C; this is the temperature range in which poor ductility occurs in a tensile test and in which cracking occurs when the strand is unbent at the straightener. This is followed by straining the tensile specimen to failure at the same strain rate used in the bending operation and the reduction of area (RA) is taken as a measure of the hot ductility. It has been found that using a standard testing procedure RA values >40 will generally avoid this form of cracking. 3 Although the simple hot ductility test is a good simulator of the continuous casting operation, there are nevertheless important differences, one of the most important being that the continuous casting cooling conditions are quite complex and very different from those used in the simple test. The temperature during cooling from the melting point is observed to fall rapidly, reach a minimum and then increase again and this is followed by the temperature cycling: first rising as the strand goes into the guide rolls and then falling as it exits and the water sprays impinge on the surface. A typical cooling sequence is shown in Fig. 1. 4 Experimentally, it is difficult to carry out this complex cycling treatment. However, it has been found that by introducing an under cooling step, the agreement between the experimental work and commercial experience is much enhanced. 5
Predicted strand surface temperature during thin slab casting (TSC): midbroad face, 90 mm strand cast at 5 m min−1 (Ref. 4)
It is proposed in the present examination that to simulate the continuous casting operation more accurately, an under cooling step should be introduced. The simple programme without an under cooling step will also be carried out. The influence of cooling rate will also be explored.
Experimental
The steels were produced as experimental 50 kg vacuum melt ingots, the dimensions of which after removing the shrinkage cavity at the top of ingot is 170×220×100 mm. The tensile samples (110 mm in length and 7·94 mm in diameter) were machined from the as cast ingots and the ends of the samples were threaded so that they can be screwed into the grips of the tensile test machine. They also had a 2 mm diameter hole drilled in the centre that ran up to the midpoint of the length of the sample in order to allow a thermocouple to enter into the melt zone. A schematic diagram of the samples is shown in Fig. 2.
Details of tensile test sample
The melt zone was ∼22 mm long (either side of the midpoint of the sample was 11 mm long). The molten steel was contained in a silica tube that had 0·2 mm diametrical clearance of the sample. Protection from oxidation was achieved by surrounding the sample with a large silica glass tube through which argon was passed, as shown schematically in Fig. 3.
Schematic diagram of testing rig
The samples were initially heated at 150°C min−1 to 1420°C, the melting point at which they were held for 2 min. They were then either cooled at 60°C min−1 to the required test temperature or under cooled at the same rate to 100°C below the test temperature. Once under cooling was carried out, the samples were then immediately reheated at 500°C min−1 to the test temperature. The sample was held at the test temperature for 3 min before straining to failure. The test temperatures were in the range of 650 to 1100°C. The tensile specimens were strained to failure at a strain rate of 3×10−3 s−1 based on a gauge length of 22 mm. The two temperature profiles used without and with an under cooling step are shown in Fig. 4a and b respectively. In addition, samples from selected steels, given a faster cooling rate of 180°C min−1, without under cooling, were examined.
Temperature profiles used a without and b with under cooling step
The compositions of the TWIP steels are given in Table 1.
Compositions of TWIP steels, wt-
All the steels contain high C (0·6 wt-) and 18 wt-Mn except for steel 2 which contains ∼22 wt-Mn. Two levels of Al were examined: low, <0·05 wt-, steels 1–2; high, 1·50 wt-, steels 3–6. The N level was generally high, 0·01–0·02 wt-. Steels 1–5 were laboratory melts, whereas steel 6 was a commercial melt and it had both a low N level and a lower S level (0·002 wt-S compared to 0·006 wt-S for the laboratory made steels), as well as having Cr and a small amount Ni.
Results
The hot ductility curves for a low Al containing steels (steels 1 and 2), and the high Al containing steel TWIP steels (steels 3–5) are shown in Figs. 5 and 6 respectively and that for the commercially cast steel (steel 6) are shown in Fig. 7 (top curve). It can be seen from Figs. 5 and 6 that the addition of 1·5 wt-Al causes the hot ductility to deteriorate in the temperature range 800–900°C from ∼38 RA (Fig. 5) to ∼15 RA as the N level increased from 0.1, steel 3 to 0.023 steel 5 (Fig. 6). The commercially cast high Al containing steel with low N and S level (steel 6 in Fig. 7), had improved ductility (∼39 RA in the temperature range 800–900°C) similar to that given by the low Al containing steels.
Hot ductility curve for low Al steels, steels 1 and 2
Hot ductility curves for high Al steels, steels 3, 4 and 5
Hot ductility curves for 1·42 wt-Al, 0·0043 wt-N TWIP steel, steel 6, after cooling at 1 and 3°C s−1 and after introducing under cooling step
In the laboratory cast high N and 1·5 wt-Al containing steels (Fig. 6), there was a tendency for the ductility to be the worst in the temperature range 800–850°C and to recover slowly with further increasing temperature.
Influence of cooling rate and introducing under cooling step
Introducing an under cooling step as in Fig. 4b to steel 6, results in a deterioration in ductility, with the RA decreasing by ∼20 (Fig. 7). A similar deterioration in ductility occurred for this steel when the cooling rate was increased from 60 to 180°C min−1 without the under cooling step (Fig. 7).
A smaller decrease in ductility was observed on increasing the cooling rate from 60 to 180°C min−1 for the higher N steel (steel 5), the RA being on average 7 lower at the faster cooling rate (Fig. 8).
Hot ductility curves for high Al and high N steel (steel 5), at two cooling rates, 60 and 180°C min−1
Microstructure of steels
Two test temperatures (800 and 900°C) were examined for steels 5 and 6, containing ∼1·5Al, with the former having a high N content and the latter low. The microstructures for these steels at the two test temperatures are given in Figs. 9 and 10 for the low N steel and in Figs. 11 and 12 for the higher N steel.
Micrograph of TWIP steel 6, (1·42 wt-Al, 0·0043 wt-N) melted and tested at 800°C showing precipitation at austenite grain boundaries
Micrograph of steel 6 (1·42 wt-Al, 0·0043 wt-N) melted and tested at 900°C showing precipitation at austenite grain boundaries
Micrograph of TWIP steel 5, (1·53 wt-Al, 0·023 wt-N) melted and tested at 800°C showing longer precipitates at boundaries compared to lower N steel 7 (Fig. 10)
Micrograph of TWIP steel 5 (1·53 wt-Al, 0·023 wt-N) melted and tested at 900°C showing extensive precipitation at austenite grain boundaries
The structures are fully austenitic and the grain boundaries are seen to contain many precipitates. In the low N and 1·5 wt-Al containing steel, the precipitates were of shorter length than in the higher N steel.
Scanning electron microscope work
Scanning electron microscope work identified the precipitates at the grain boundary as AlN precipitates (Fig. 13a and b ). In the higher N containing steel the precipitates were coarser (Fig. 13b).
Copious precipitation of AlN at γ grain boundaries in a low N, TWIP steel 6 (1·43 wt-Al, 0·0043 wt-N) showing very thin films of AlN at grain boundaries on fracture surface and b coarse AlN particle at boundary in high N, TWIP steel 3 (1·53 wt-Al, 0·01 wt-N): both steels were tested at 800°C after melting and fracture surfaces examined
Discussion
Because the steels are fully austenitic, ductility would not be expected to improve at the low temperature end as it does in plain C–Mn and high strength low alloy steels, when ferrite that has excellent ductility forms in substantial amounts below the Ar3. 6 Ductility would be expected to improve at the high temperature end when dynamic recrystallisation occurs, however for these steels dynamic recrystallisation does not take place until very high temperatures. Previous work 7 has found that as the C level increases dynamic recrystallisation becomes more difficult and that it is clear from Fig. 5 that even without the presence of large volume fractions of AlN precipitation that the high C and probably high Mn levels make it increasingly difficult for dynamic recrystallisation to occur.
Previous work 8 on low Al steels tested in the temperature range in which the steels are fully austenitic but unrecrystallised has also found that the ductility remains approximately constant with temperature. Presumably the increase in grain boundary sliding which occurs with increasing temperature, which would result in a deterioration in the hot ductility, is approximately balanced by the increased rate of recovery. This also seems to be the case for the presently examined low Al steels, with the RA remaining constant within temperature range 700–1000°C (Fig. 5); no dynamic recrystallisation having been observed to take place. For the laboratory produced high Al containing steels, although dynamic recrystallisation was not observed, there is a slow improvement in ductility as the temperature increases, suggesting that some grain boundary migration may be taking place (Fig. 6).
The temperature for the maximum rate of precipitation of AlN is 815°C, 9 and the curves for the higher Al containing steels indicate that this test temperature gives worse ductility (Figs. 6 and 7). It can be seen from Table 2 that taking the depth of the hot ductility curves at 800°C from Figs. 5 and 6, increasing the Al and N contents or increasing the amount of AlN precipitated out, assuming equilibrium conditions and all the N is taken out of solution by Al as AlN, results in the decrease in ductility (Table 2). Ductility will be dependent on both the Al and N levels. Higher N levels favour more AlN precipitation and higher Al levels favour a greater driving force for AlN precipitation. Thus, both high Al and N levels should be detrimental to ductility.
AlN precipitation and its influence on minimum depth of trough
It therefore appears that from Table 2, the ductility of the high Al, high N, 1·5 wt-Al laboratory cast steels is worse than the low Al containing steels because of the greater volume of AlN particles presented at the austenite grain boundaries. The marked AlN precipitation at the austenite boundaries is clearly shown in Figs. 9–13. Previous work 10 on TRIP assisted steels in which the microstructure is ferrite/austenite in the test temperature range 700–900°C has also found that 1 wt-Al is detrimental to ductility. In this case, the poor ductility is due to a combination of AlN precipitation and the presence of the thin film of ferrite at the austenite grain boundaries.
Under cooling led to the worst ductility. In the previous work5,11 on low Al containing steels, under cooling also caused deterioration in ductility and this was ascribed to the under cooling causing more AlN to precipitate out. Similar reasoning may apply to the presently examined steels but replicas from the fracture surface need to be taken to confirm this.
Increasing the cooling rate again led to worse ductility. Previous work 12 on low Al containing steels has also found this behaviour and it was shown to be due to both a finer AlN particle and MnS inclusion size. This may also be the case for the higher Al containing steels and again further examination is required to confirm this.
As the ductility of these TWIP steels depends solely on the ductility of unrecrystallised austenite, low S and N levels are needed to keep the ductility sufficiently high to avoid transverse cracking and this is probably the reason for steel 6 giving the highest ductility of all the steels examined for the temperature range 700–1000°C.
Castability is a concern with any high Al grade since the chemical conditions are favourable for the formation of the spinel MgO.Al2O3 and CaO6.Al2O3 type inclusions that are well known for clogging of the caster pouring system refractories. 13 However, it should be noted that high Al (1–2 wt-) TRIP steels are also being commercially produced and have no problems. Becker et al. 13 have reported two production trials at their US Steel Gary Works on high Al TRIP steels. In the first one, longitudinal cracks were prevalent while by altering the mould flux they were absent in their second trial. Transverse cracking was avoided by calculating the hot ductility curve and avoiding the low ductility zone which from their calculation was in the temperature range 780–900°C when ductility was <40. Detailed inclusion analysis of liquid steel provided some insight as why the clogging behaviour in their trials was less than expected. The major inclusion was CaS which is believed to be formed via a reaction with the ladle slag under highly reducing conditions imposed by the high Al level in the steel and these Ca rich inclusions have a lower tendency to cluster than Al2O3.
Commercially, POSCO have successfully cast TWIP steels, with its composition given in Table 1 using a new molten mould flux feeding process which gives improved lubrication between the solidifying steel shell and the copper mould. 1
Conclusions
1. Twin induced plasticitysteels containing 0·6 wt-C and 1·5 wt-Mn have relatively flat hot ductility curves, with the ductility gradually increasing with temperature. Because the thin film of ferrite is not present and dynamic recrystallisation does not take place, recovery in ductility at the high and low temperature ends of the trough does not occur.
2. Ductility was generally <40 RA. The 1·5 wt-Al containing steels had worse ductility than lower Al containing steels because of the presence of large amounts of AlN precipitated at the γ grain boundaries.
3. Ductilty is dependent on unrecrystallised austenite and therefore to avoid transverse cracking, low N and low S levels are recommended so that the volume fraction of second phase particles (AlN and MnS) is low.
4. Introducing an under cooling step to the cooling program to better simulate the commercial process led to worse ductility.
5. Increasing the cooling rate from 60 to 180°C min−1 similarly caused a deterioration in ductility.
Footnotes
Acknowledgement
The authors would like to thank POSCO for permission to publish and for the financial support for the program.