An iron-based alloy system has been developed that exhibits impressive combinations of tensile strength and elongation that are not available with current steels used in the manufacture of automobiles. Furthermore, the heat treatments required to achieve these properties are consistent with practical production processes. The alloys rely on significant concentrations of ferrite-stabilizing solutes so that δ-ferrite which forms during solidification is retained in the microstructure.
There is a huge variety of steels available and many of these result in similar properties but are produced to satisfy particular requirements of cost, weldability and design criteria (Joo et al. 2009). This overlap of properties is evident also in the formable alloys designed for the automotive industries, although an examination of figure 1 shows that the distribution is not uniform; there are significant gaps at intermediate strength and ductility combinations and when the strength exceeds about 1200 MPa.
There already is research in progress on the stronger steels in order to enhance ductility and assess other engineering properties (Edmonds et al. 2006; Naderi 2007; Fan et al. 2009; Bhadeshia 2010). The same attention has not been paid to alloys that have a strength of 600–700 MPa and at the same time exhibit elongations in excess of 30 per cent since these materials are typically formed into components. Although the so-called TWIP1 steels fulfil these requirements, they are relatively expensive to manufacture and apply.
It is interesting in this context to explore the recent proposal of δ-TRIP steel in which the alloy is designed to retain a large quantity of δ-ferrite at all temperatures, with the residual microstructure consisting of a mixture of bainitic ferrite and retained austenite (Chatterjee et al. 2007; Yi et al. 2010a,b). The δ-ferrite substitutes for the allotriomorphic ferrite that is normally introduced into TRIP-assisted steels by intercritical annealing or continuous cooling transformation (Matsumura et al. 1987a,b; Sakuma et al. 1991; DeCooman 2004; Jacques 2004). There are advantages to this, for example, that a fully martensitic structure cannot be produced in the heat-affected zone of a resistance spot weld. The steel also relies more on aluminium rather than silicon to suppress the formation of cementite, which helps avoid the problem of the adherent fayalite scale (Fukagawa et al. 1994; Okada et al. 1995; Raman 2006) that forms on the surface during hot-rolling.
The δ-TRIP steel has been demonstrated in the cast state to have an ultimate tensile strength of about 1000 MPa and an elongation of 27 per cent (Chatterjee et al. 2007). Subsequent experiments highlighted the fact that the alloy system is sensitive to non-equilibrium solidification which meant that δ-ferrite could not be reliably retained in the microstructure (Yi et al. 2010a). Further intense research (Yi et al. 2010b) was necessary in order to characterize the transformation behaviour during and after hot-rolling in the two-phase field; the rolling deformation is necessary to apply the steel in sheet form for use in the manufacture of automobiles. In the present work we report for the first time, some exciting mechanical property data for what we believe is a complete alloy design of a type consistent with large-scale manufacture.
2. Alloys and experimental method
The alloy design procedures involving phase diagram calculations have been detailed elsewhere (Chatterjee et al. 2007; Yi et al. 2010a,b) and are not repeated here. They basically help ensure the presence of δ-ferrite under equilibrium conditions, at all temperatures in the solid state. Aluminium plays a key role in this and experience based on experiments (Yi et al. 2010a) has indicated that such calculations overestimate the amount of δ-ferrite expected when the real alloys solidify during casting under conditions which deviate from equilibrium. Therefore, the two alloys studied here (table 1) have larger aluminium concentrations than in the original work (Chatterjee et al. 2007).
The alloys were manufactured as 34 kg ingots 100×170×230 mm size using a vacuum melting furnace. The ingots were reheated to 1200°C for rough rolling to make 25–30 mm slabs followed by air cooling. These slabs were then reheated to 1200°C and hot-rolled to 3 mm in thickness; 1.2 mm thick sheets were then produced by cold-rolling. The heat treatments outlined in table 2 were conducted using 1.2×40×110 mm samples in a CCT–Ay (Ulvac–RIKO) simulator where the samples were heated at 20°C s−1 in a nitrogen atmosphere to allow some austenite to form, followed by cooling at −20°C s−1 to the temperature where bainitic ferrite is allowed to form, and finally at −10°C s−1 to ambient temperature. Samples for tensile tests were machined from these blanks to ASTM standard E8M–00 with elongation measured on a 10 mm gauge length over the 25 mm parallel length, following tension at 3.3×10−3 s−1. All the mechanical data presented in this paper represent the average of three tests; the reproducibility was excellent with the elongation varying by no more than ±2 per cent and the strength by ±5 MPa.
Microstructural evolution was studied using samples 2×3×10 mm on a dilatometer described elsewhere (Pak et al. 2008). X-ray diffraction was done on metallographically prepared samples etched using 2 per cent nital in order to remove any deformed surface, with CuKα radiation with the data subjected to Rietveld analysis (Hill & Howard 1987) and refinement (Rietveld 1967, 1969). The austenite lattice parameter thus obtained was used to estimate the concentration of carbon using a published equation (Dyson & Holmes 1970).
Most phases could easily be identified using scanning electron microscopy. The austenite regions stand proud on the surface, and the bainite is in the form of sheaves which appear at these magnifications as plates. The δ-ferrite is given away by its coarse scale whereas the allotriomorphic ferrite appears as finer equiaxed grains. Pearlite present in the hot-rolled samples (without subsequent heat treatment) appears with dark contrast because the cementite and ferrite phases are intimately mixed. Martensite that forms during deformation could not be resolved but its presence is deduced from the decrease in the retained austenite content as measured using X-ray diffraction.
The microstructures of the alloys in the hot-rolled condition are illustrated in figure 2a,b, where we distinguish δ-ferrite and α-ferrite simply to identify the former as having been retained from solidification and has persisted through the hot-deformation, whereas the α is a result of transformation from austenite. None of the δ-ferrite is retained in the microstructure of Alloy 8, which becomes fully austenitic during hot-rolling, resulting in a final microstructure of equiaxed grains of α and somewhat discontinuous pearlite. In contrast, Alloy 9, which contains more aluminium and less manganese, has the coarse elongated regions of ferrite that represent the δ phase. The influence of composition is also reflected in the quantitative data given in table 2 where the total fraction of (δ+α) is seen to be greater in Alloy 9 in the hot-rolled condition.
Metallography was conducted on all of the heat-treated samples listed in table 2 but only a few representative results are illustrated in figure 2c,d for the particular case where the intercritical annealing was at 850°C followed by isothermal transformation at 400°C for 600 s. For the discussion below, the terms VHR and VIA represent the fractions of (δ+α) ferrite present in the microstructure in the hot-rolled and intercritically annealed conditions, respectively.
Alloy 8, following heat treatment consists of fine and equiaxed allotriomorphic ferrite grains together with islands of retained austenite containing regions of bainitic ferrite. The intercritical annealing temperature is not high enough to fully austenitize the material, so the total fraction of allotriomorphic ferrite has decreased only a little from VHR=0.71±0.09 in the hot-rolled state to VIA=0.66±0.07—this general observation is essentially correct for both alloys and intercritical annealing temperatures studied, as can be seen in table 2. Retained austenite measurements are discussed in the next section.
Alloy 9 differs in that there is a mixture of the δ-ferrite together with allotriomorphic ferrite; the overall fraction of ferrite is larger by about 0.1 volume fraction, as expected from the composition difference relative to Alloy 8 (table 2). The quantity of bainite within the retained austenite is also much smaller, because of the greater stabilization of the residual austenite by carbon partitioned from the (δ+α). Most of the islands of austenite were featureless at high magnification, figure 2d.
A summary of the mechanical properties obtained is presented in figure 3 with the details listed in table 2. The strength–elongation combination is impressive for all the heat treatments applied, and approaches that of the high-Mn TWIP steels (Grässel et al. 2000; Frommeyer et al. 2003) which are more challenging to produce and apply. The data also fall in a domain which is not occupied by other automotive steels and furthermore, the ductility is comparable to that of the much weaker interstitial-free steels (Bayraktar et al. 2007; Hayat et al. 2009; Mukhopadhyay et al. 2009). The alloy also outperforms TRIP and dual-phase steels at the same strength level.
One reason for the large ductility observed is the presence of retained austenite in a form which transforms during deformation but has a stability which means that the transformation is gradual so that local stress concentrations which lead to damage can be accommodated to large strains (Chatterjee et al. 2007; Bhadeshia 2008). Figure 4 shows clearly that both the harder austenite (Furnemont et al. 2002), which also contains the bainite, and the allotriomorphic and δ-ferrite deform in a compatible manner until final fracture occurs; the micrographs are taken in the close proximity (≃1 mm) of the fracture surface and illustrate the elongation experienced by all the phases present.
Figure 5 shows the quantitative measurements of retained austenite both before and after deformation and it is evident that the austenite is reasonably stable given the large extent of plastic strain. The data also show that the austenite fraction is smaller in the case of Alloy 9, and its carbon concentration larger. This is because it contains a greater aluminium and lower manganese concentration than Alloy 8, which means that a larger amount of ferrite is present at the intercritical annealing temperature (86% compared with the 71% in Alloy 8). The residual austenite is therefore enriched to a correspondingly greater degree due to the partitioning of carbon between the ferrite and austenite.
It should also be noted that the stability of the austenite increases if it survives the initial plastic strains and stresses, because the defect structure introduced during deformation can mechanically stabilize it to martensitic transformation (Machlin & Cohen 1951; Fiedler et al. 1955; Leslie & Miller 1964; Chatterjee et al. 2006).
Alloy 8 in general shows better ductility than Alloy 9, presumably because of its finer structure. As pointed our earlier, the former becomes fully austenitic during hot-rolling so that the coarse regions of δ-ferrite which exist in Alloy 9 are absent. Even though Alloy 8 in its final state does not fit the general concept of δ-TRIP steels, the work suggests that the properties of steels which do retain the δ-ferrite could be improved by refining this phase. This could be achieved by controlling the solidification rate or by enhanced deformation during hot-rolling. This latter procedure would be practical when dealing with large castings than in the present work where small quantities of experimental alloys were fabricated.
Tensile curves for Alloy 8 are illustrated in figure 6 showing a form consistent with the requirements of automotive steels; the small initial elongation at constant stress did not lead to any observable Lüders bands on the tensile specimen surfaces. It is not clear whether this would become an issue in the context of industrial production as opposed to the present laboratory experiments. Note also the excellent reproducibility of the tests.
A detailed examination of the tensile data (table 2) shows that the time at the isothermal transformation temperature does not significantly influence the mechanical properties, probably because the amount of bainite produced is rather small given the large carbon concentration of the austenite. It would be interesting in future work to eliminate the isothermal step altogether and assess the resulting properties.
A novel alloy system based on the concept of stabilizing ferrite in the microstructure using aluminium as an alloying element has been found to exhibit promising combinations of tensile strength and elongation. These properties fall in a domain which is not represented by other commercially available or experimental steels designed for the automobile production.
The heat treatments necessary to achieve the microstructure involve intercritical annealing followed by isothermal transformation at temperatures and time periods which are consistent with large-scale production on the so-called continuous annealing facilities (Yanagishima et al. 1983). The isothermal transformation can even be conducted at 450°C, which is compatible with a final galvanizing treatment.
In future work it is hoped to investigate the spot-welding characteristics; the key feature of the alloy design in this respect is that δ-ferrite can be retained permanently in the microstructure so that fully martensitic regions are not produced in the heat-affected zone of the spot weld.
We acknowledge Prof. Hae-Geon Lee for the provision of laboratory facilities at GIFT, and support from POSCO through the Steel Innovation Programme. The authors are also grateful for support from the World Class University Programme of the National Research Foundation of Korea, Ministry of Education, Science and Technology, project number R32-2008-000-10147-0.
↵1 ‘TWIP’ stands for twinning-induced plasticity, and ‘TRIP’ for transformation-induced plasticity. Both mechanical twinning and displacive phase transformations have the ability to enhance ductility by delaying the onset of plastic instability during tensile deformation, and hence of ultimate fracture.
- Received March 2, 2010.
- Accepted April 13, 2010.
- This journal is © 2010 The Royal Society