Characterization of Compound Particles Formed during Thin Slab Direct Rolling of Ti-added Nb HSLA Steel

Abstract

The absence of a reheating stage in thin slab direct rolling of Ti-added Nb HSLA steel results in the formation of compound two-phase particles prior to and during rough rolling in an in-line strip processing line. The compound two-phase particles are composed of a cuboid Ti-rich (Ti_xNb_1–x)N (0.76≥x>0.72) core and a Nb-rich cap-shaped epitaxial deposit of (Ti_xNb_1–x)C (0.29≥x>0.09) or NbC formed on one of the {100}-type faces of the cuboid (Ti_xNb_1–x)N (0.76≥x>0.72) core. At the interface between the cuboid core and the cap-shaped deposit, the Ti/(Nb+Ti) atomic ratio was found to increase gradually from a low value of Ti/(Nb+Ti)≈0, on the cap-side of the particles, to a high value of Ti/(Nb+Ti)≈0.6, on the cuboid core side of the precipitate. The fact that the compound two-phase particles are present at 1200°C indicates that they have a greater thermodynamic stability compared to NbN or NbC. A kinetic precipitation model was used to evaluate three possible mechanisms for the formation of the compound particles: a low cap/cuboid interfacial energy and a high matrix/cuboid interfacial dislocation density.

1. Introduction

Thin slab direct rolling (TSDR) technologies are promising processing routes to maintain steel as the material of choice in many technological applications. Their major advantage is to offer a cost effective alternative to the standard hot strip mill (HSM) process. The essential difference between TSDR and conventional HSM technologies is that in TSDR the entire casting and rolling process is run continuously. In addition to the obvious advantage of having lower capital investments, using less energy, having significantly lower operational costs and a smaller size, TSDR offers a more flexible processing which can more easily adapt to market requirements. As CO₂ emissions are reduced due to the use of liquid metal produced from scrap metal in an electric arc furnace (EAF) and the reduced energy consumption, TSDR technologies are also more environment-friendly.¹⁾ There are also essential differences between TSDR and conventional HSM technologies in terms of the steel microstructure. The centerline segregations during solidification are smaller in TSDR and the inclusion size is also generally smaller due to a higher solidification rates and a faster post-solidification cooling.²⁾ In terms of metallurgical characteristics, TSDR hot strip production has a number of significant drawbacks. The as-cast austenite grains prior to hot rolling are very coarse as there is no austenite-to-ferrite and ferrite-to-austenite transformation related to slab cooling and reheating as is the case in conventional HSM processing. These two transformations refine the austenite grain size prior to rolling in a conventional HSM. The use of long reheating times in the HSM also guarantees the effective dissolution of micro-alloying elements and a more homogeneous slab composition.²⁾ In TSDR, two stages should be considered when defining the rolling sequence. Ideally the complete elimination of the initial as-cast microstructures should be achieved in the first rolling stands and an optimum austenite conditioning should be achieved prior to transformation. The activation of dynamic recrystallization and subsequent meta-dynamic recrystallization (MDRX) softening can be used to induce a faster microstructural refinement.^3,4,5)

In Nb-added high strength low alloy (HSLA) steels the strengthening is achieved mainly by grain refinement and, to a lesser extent, precipitation strengthening. The grain refinement is obtained by thermo-mechanical controlled processing (TMCP) during hot rolling in the non-recrystallization temperature range in the finishing tandem mill section of a conventional HSM. During TMCP, a large amount of strain accumulation is achieved by the presence of Nb in solid solution in the deformed austenite. Both solute drag and strain-induced precipitation of NbC contribute to the austenite conditioning, i.e. the suppression of the interpass recrystallization of deformed austenite during TMCP. The transformation of strained, rather than recrystallized, austenite results in a high ferrite nucleation rate and a considerably reduced final grain size. The application of TMCP to thin slab processing lines such as Compact Strip Production (CSP), In-line Strip Production (ISP) and Endless Strip Production (ESP) has proven to be challenging. This is due to four factors, (a) the steel composition (b) the absence of slab cooling to room temperature, (c) the lower amount of thickness reduction and (c) the thermal cycle which is inherently different from the thermal cycle of a conventional HSM. The molten metal used in thin slab lines is produced in electric arc furnaces (EAF). The EAF steel, which uses scrap metal, is characterized by a higher level of tramp elements (Cu, Sn, etc.) and a high N content compared to the molten metal produced by the basic oxygen furnace (BOF) process which is usually used in a conventional HSM.²⁾ As the thin slabs are not cooled to room temperature prior to reheating and hot rolling, no grain-refinement due to phase transformation takes place and a coarse grained as-cast microstructure is directly hot rolled. These issues are addressed, for example in the ISP line, by the addition of the soft reduction equipment in the continuous casting, and the application of a rough rolling stage carried out in a 2–3 stand tandem mill. Ti-additions are often made in TSDR to simultaneously reduce the solute N content and form very fine TiN precipitates. The TiN precipitates are formed during thin slab solidification and post-solidification cooling. The TiN precipitates exert a strong grain boundary pinning effect and limit the high temperature austenite grain growth very effectively.⁶⁾ The main impact of a high solute N content is the formation of Nb-containing TiN with the Nb-alloying additions needed for the TMCP of HSLA steel. A high N content is known to disturb the usual strain-induced NbC precipitation during TMCP.⁷⁾ Previous research has explained the pronounced absence of the solute drag effect of Nb in the presence of high levels of N by the formation of Nb-nitride solute atom complexes,⁸⁾ but little is currently known about the precipitation in Nb HSLA steels with high N contents. The present contribution is the first report on the precipitation sequence in Ti-added Nb-HSLA steel during the rough rolling stage in TSDR processing using ISP process. It is shown that the presence of a high N content results in low levels of solute Nb prior to the finish rolling, even when most of the N is stabilized as TiN. This is due to the fact that the addition of Ti to Nb micro-alloyed steel, significantly changes the precipitation behavior. The addition of Ti results in the formation of compound precipitates, consisting essentially of a Nb-containing TiN-type cuboid core and and a NbC-type cap-shaped deposit. The premature nucleation and growth of theses NbC-type cap-shaped deposit on a TiN-type cuboid core during the rough rolling stage in ISP, results in a decrease of the solute Nb content and renders the TMCP much less effective during finishing rolling. A similar mechanism, not involving deformation, is also believed to lead to Nb-precipitate formation during the equalizing time in the tunnel furnace of the CSP lines. The problem is exacerbated by the fact that the total rolling reduction during hot rolling in TSDR technologies is less than the reduction given in the case of conventional HSM.

2. Experimental

Table 1 shows the chemical composition of the experimental alloys used in the present study. Ingots 30 kg in weight were prepared by vacuum induction melting and reheated to 1200°C for hot rolling to 25 mm thick plates with a finishing temperature higher than 900°C. Torsion specimens with a gage section of diameter 6.5 and 13 mm in length were prepared and laboratory processed by simulation of the roughing mill reduction stage of the ISP process. The specimens were heated to 1200°C for 300 s in Ar gas to increase the initial grain size and dissolve the pre-existing precipitates. This was followed by hot torsion test at 980°C at a strain rate of 1 s^–1 with total amount of true strain of 0.42. After deformation the specimens were directly water quenched. The replica extraction method was used to investigate the composition of the precipitates by transmission electron microscopy (TEM). The specimens were mechanically polished and chemically etched for 40 minutes with a 5% nital solution. A carbon film was evaporated onto the surface and divided into squares 2×2 mm² in size. A second etching was applied for 5 minutes using a 10% nital solution. After the dissolution of the matrix, the carbon film containing extracted precipitates was floated off and picked up on Cu grids for TEM observation.⁹⁾ TEM observations were carried out on a JEOL 2100F field emission (FE) electron source TEM operating at 200 kV. The TEM was equipped with an Oxford EDX (Energy Dispersive X-ray) analyzer and a Gatan DigiPEELS spectrometer, which uses a photodiode array detector for Electron Energy Loss Spectroscopy (EELS) analysis. In order to reduce the carbon contamination of the TEM samples, the extraction replicas were cooled to –170°C. A detailed description of the quantitative analysis of the precipitates by EELS can be found elsewhere.¹⁰⁾ The kinetics of the precipitation were modelled using the MatCalc software, which allows for the study of nucleation, growth and coarsening of precipitates in multi-component ferrous alloys during complex time-dependent thermo-mechanical cycles.¹¹⁾

Table 1. Chemical composition of the steel used in the present work (in mass-%).

C	Mn	Si	P	S	Al	Cr	Ti	Nb	N
0.040	0.894	0.049	0.02	0.003	0.01	0.03	0.0100	0.040	0.0099

3. Results

Isolated NbC or NbN precipitates were not observed in samples quenched prior to and after the rough rolling during the simulation of ISP processing. Instead, two-phase compound particles, consisting of a cube-shaped core and a cap-shaped deposit, were found before and after rough rolling stage of the ISP process. The High-Angle Annular Dark-Field (HAADF) scanning TEM image of a typical compound particle consisting of a cube-shaped core and a cap-shaped deposit is shown in Fig. 1(a). As HAADF contrast is sensitive to the atomic number (Z-contrast), the bright contrast of the cap-shaped deposit and the dark contrast of the cube-shaped core of the precipiates indicate a considerably different composition, with the core being likely Ti-rich, whereas the deposit is Nb-rich. A high resolution TEM (HR-TEM) lattice image of a typical compound precipitate is shown in Fig. 1(b). The image clearly shows that the cap is an coherent epitaxial deposit on one of the {100}-type surface facets of the cube-shaped precipitate. The HR-TEM image shows that the interface is free of interface dislocations and that there is a perfect lattice match between the substrate and the cap-shaped deposit.

Fig. 1.

Precipitate analysis after rough rolling. (a) High-Angle Annular Dark-Field (HAADF) STEM image of a compound particle consisting of a cube-shaped core and a cap-shaped deposit. (b) High resolution TEM micrograph of the interface between a cube-shaped core and a cap-shaped deposit. (c) EELS spectra analysis of a cap-shaped deposit and a cube-shaped core showing that N is only present in the cube-shaped core. (d) Change of the lattice parameter across the interface between the cube-shaped (Ti_xNb_1–x)N (x≈0.6) core and the (Ti_xNb_1–x)C (x≈0) cap-shaped deposit. (e) Lattice mismatch as a function of position relative to the interface; the lattice mismatch region is 9.57 nm thick. (f) Change of the chemical composition and the lattice parameter at the interface between the cube-shaped (Ti_xNb_1–x)N (x≈0.6) core and the (Ti_xNb_1–x)C (x≈0) cap-shaped deposit. The lattice parameter is calculated on the basis of the chemical composition using the data in (f) and in Table 3. (g) Lattice parameter mismatch based on the composition measurements; the lattice mismatch region is 9.93 nm thick. (h) Schematic of the structure of the compound particles.

Table 3. The relation between compound composition and calculated lattice parameter according to reference.¹³⁾

Compound	Structure	Lattice parameter (nm)
NbC	NaCl	0.4506
(Nb0Ti4)N4	d-NaCl	0.4256
(Nb1Ti3)N4	d-NaCl	0.4309
(Nb2Ti2)N4	d-NaCl	0.4363
(Nb3Ti1)N4	d-NaCl	0.4404
(Nb4Ti0)N4	NaCl	0.4453

*d-NaCl indicates a defective NaCl type structure.

EELS analysis considering the Ti-L_2,3, Nb-M_4,5 and N-K ionization peaks was used to analyse the composition of the cap-shaped deposits in more detail. Figure 1(c) shows the EELS spectra of a cap-shaped deposit and cube-shaped core. The EELS spectrum of the cube-shaped core had clear Ti-L_2,3, Nb-M_4,5 and N-K ionization peaks but the EELS spectrum of the cap-shaped deposit had only Nb-M_4,5 ionization peak. It is therefore reasonable to identify this cap-shaped deposit as NbC.

The {111} interplanar distance was measured across the compound particle from the cube-shaped core to the cap-shaped deposit directly from the HR-TEM image. The lattice parameter data derived from these measurements shown in Fig. 1(d) indicate a gradual increased of the lattice parameter in the interfacial region from the cap-shaped deposit to the cube-shaped precipitate. A similar increase of the lattice parameter form a cube-shaped carbonitride substrate particle to a cap-shaped carbide deposit has been reported previously.¹²⁾ The highest value and the lowest value reported for the experimentally measured lattice parameter of NbC, NbN and TiN are listed in Table 4. The measured lattice parameter change measured in the present study matches a change in lattice parameter corresponding to a change from a TiN-like core to a NbC cap-shaped deposit.

Table 4. Reported values for the lattice parameters of NbC, NbN and TiN. The experimental data is taken from the references indicated in parentheses.

Compound	Maximum value Lattice parameter, nm	Minimum value Lattice parameter, nm
NbC	0.44729)	0.43613)
NbN	0.44430)	0.43831)
TiN	0.42732)	0.42331)

Figure 1(e) shows the local value of the lattice mismatch across the core/deposit interface. The mismatch was calculated using the data of Fig. 1(d) for points 0.427 nm apart and the resulting data points were fitted to a Gaussian curve. The interface, defined as the region over which there is a change in lattice parameter mismatch within a distance of 0.427 nm, was found to have a thickness of approximately 9.6 nm. The maximum lattice mismatch was 0.12%.

The results of the EDS analysis shown in Fig. 1(f) indicates that there is a gradual, rather than an abrupt, change in the chemical composition at the interface between the cube-shaped core and the cap-shaped deposit. The EDS analysis revealed that the composition of the cube-shaped cores of the compound particles was (Ti_xNb_1–x)N with x≈0.6. The composition of the cap-shaped deposits was (Ti_xNb_1–x)C with x≈0. Assuming that the chemical composition changed from (Ti_xNb_1–x)N in the cube-shaped core to (Ti_xNb_1–x)C (x<0.1) in the cap-shaped deposit, the lattice parameter was calculated from the measured chemical composition using the composition-lattice parameter relation given by Tirumalasetty et al. (Table 4).¹³⁾ Using the same approach as the one described to obtain Fig. 1(f), the lattice mismatch variation across the interface was computed. A maximum lattice mismatch of 0.26% was found and the width of the mismatch region was 4 nm in Fig. 1(g).

Figure 1(h) shows a schematic for the structure of the compound particles with a NbC cap-shaped epitaxial deposit on one of the {100}-type surface facets of a TiN-like cube-shaped precipitate. In the dark gray region there is a transition in composition and lattice parameter from the cuboid core to the cap-shaped deposit.

Typical EDS spectra for the core region and the cap-shaped deposit of the compound precipiates collected in samples prior to and after rough rolling are shown in Fig. 2(a). The EDS spectra reveal that the Ti content of the cap-shaped deposit increased during rough rolling. Figure 2(b) shows the compsotional analysis of the cube-shaped core and the cap-shaped deposit. The composition of the cuboid cores was rich in Ti, with small amounts of Nb, (Ti_xNb_1–x)N (0.76 ≥ x>0.72), resulting an atomic Ti/(Nb+Ti) ratio in the range of 0.72–0.76. The composition the cap-shaped deposits was rich in Nb, with a very small amount of Ti, (Ti_xNb_1–x)C (0.29 ≥ x>0.09), resulting an atomic ratio in the range of 0.09–0.29. The Ti/(Nb+Ti) ratio of the cube-shaped cores decreased during rough rolling from 0.76 to 0.72. In contrast, the Ti/(Nb+Ti) ratio of the cap-shaped deposit increased during rough rolling from 0.09 to 0.29.

Fig. 2.

Typical EDS spectra of the cube-shaped core regions and the cap-shaped deposits in samples quenched (a) prior to and (b) after the rough rolling. The Ti/Ti+Nb ratio distribution based on the analysis of the EDS spectra (c) prior to and (d) after the rough rolling.

The exact composition of the cap-shaped deposits is difficult to determine due to experimental difficulties. The deposit are believed to be essentially NbC, but this cannot ascertained with absolute certainty due to the presence of the (Ti_xNb_1–x)(C_yN_1–y) transition region between the core and the cap, which was mentioned earlier, and the difficulty in avoiding even extremely small amounts of sample drift during the collection of the EDS and the EELS spectra. It can therefore not be entirely excluded that the composition of the cap-shaped deposits might be (Ti_xNb_1–x)(C_yN_1–y), with the Ti and N content being very small. These difficulties are also apparent in previous work. Whereas similar cap-shaped deposits have been identified as (Ti_xNb_1–x)C and (Ti_xNb_1–x)(C_yN_1–y) in a few publications, others have clearly indentified the cap-shaped precipitates as NbC.^{12,14,15,16,17)} In the discussion section, it was assumed that NbC cap-shaped deposits are formed on TiN cube-shaped cores.

4. Discussion

Table 2 reviews the calculated minimum, maximum and median value of the dissolution temperature of carbide and nitride precipitates in austenite for the steel composition given in Table 1. The dissolution temperatures are based on solubility data taken from the literature. Based on the solubility temperature range listed in Table 2 it is possible to predict the fromation of NbC, NbN, TiC and TiN precipitates. As the solubility temperature of TiC is too high to precipitate during rough rolling at 980°C, no TiC precipitates were expected and none were observed. The solubility temperature of TiN and NbN are however low enough to result in precipitate formation during rough rolling at 980°C. NbN precipitates were however not observed in this study. In the case of NbC it is difficult to make any precise prediction about its precipitation during rough rolling at 980°C based on the available solubility data.

Table 2. Calculated minimum, maximum and median value of the dissolution temperature of carbide and nitride precipitates in austenite for the steel composition given in Table 1. The values are based on solubility data taken from the references indicated in parentheses.

	Maximum dissolution temperature, °C	Minimum dissolution temperature, °C	Median dissolution temperature, °C	Precipitation during Rough Rolling at 980°C
TiN	164619)	133520)	145321)	Yes
NbN	124722)	107923)	110124)	Yes
NbC	112825)	94826)	102324)	Possible
TiC	95727)	85921)	92428)	No

The premature formation of NbC during the ISP processing of Ti-added Nb HSLA steel with a high N content implies that, in the absence of enough solute Nb, the TMCP processing will not be possible. The observations reported in the present study and in other studies on TSDR suggest that there is a fundemental difference with conventional HSM processing of Nb HSLA steel in which the premature formation of NbC does not occur.

Two mechanisms may facilitate the nucleation and growth of a NbC cap-shaped deposit on a {100} facet a TiN cube shaped precipitate:

• A NbC cap is formed on a cube-shaped precipitate because the interfacial energy for homogenous nucleation is considerably reduced by the heterogeneous nucleation.¹⁸⁾

• A large density of geometrically necessary dislocation is accumulated at the matrix/TiN interface because the TiN particles do not deform during rough rolling. This locally high dislocation density provides preferred nucleation sites for the nucleation of Nb precipitates at the matrix/TiN interface.¹⁶⁾

Figure 3 shows the temperature dependence of the volume fraction of NbC particles and solute Nb content in austenite during ISP processing, for different values of the surface energy for nucleation and dislocation density as calculated by MatCalc.

Fig. 3.

Model calculation of the temperature dependence of the NbC precipitate volume fraction and the solute Nb content. In (a) the interfacial energy is increased from 0.067 J/m² to 0.603 J/m² and the dislocation density is kept constant at 10¹¹ m^–2. In (b) the interfacial energy is increased from 0.067 J/m² to 0.603 J/m² and the dislocation density is kept constant at 10¹⁴ m^–2. In (c) the dislocation density is increased from 10¹¹ m^–2 to 10¹⁵ m^–2 and the interfacial energy is kept constant at 0.201 J/m².

In Figs. 3(a) and 3(b) show the effect of a change of the surface energy on the kinetics of precipitation. The dislocation density was kept constant at 10¹¹ m^–2 in Fig. 3(a) to simulate the precipitation in the absence of deformation. The dislocation density was kept constant at 10¹⁴ m^–2 in Fig. 3(b), to simulate the deformation-enhanced precipitation during rough rolling. The simulations show that when the surface energy is lowered, the NbC precipitation start temperature increases. An increase of the dislocation density from 10¹¹ m^–2 to 10¹⁴ m^–2, results in a higher volume fraction of precipitates. The increase of the dislocation density had only a minor influence on the precipitation start temperature.

In Fig. 3(c) shows the effect of an increase of the dislocation density on the kinetics of precipitation. The surface energy was kept constant at a value of 0.201 J/m². The result show that an increased dislocation density leads to a increased volume fraction of precipitates and a pronounced reduction of the solute Nb content.

The simulations are in agreement with the observations of compound particle formation during the rough rolling stage of the ISP process. They predict that both NbC and NbN may form in the austenite. The EELS spectrum of Fig. 1(c) shows that the cap-shaped precipitates were NbC. Therefore, the main reason for the reduction in solute Nb content is the precipitation of NbC on prior TiN particles in the rough rolling temperature range. This co-precipitation is only possible when the interfacial energy between the two compounds is low enough. A prerequisite for a low interfacial energy is a small lattice mismatch at the interface between the two precipitates. The NbC/TiN lattice mismatch is in the range of 2.1% to 5.7%, dependening on the choice of lattice parameters used to compute the mismatch. Both mismatch value are considerably larger than the values measured experimentally, 0.12% and 0.26%, which are due a gradual change of the interfacial composition. The formation of a transition layer results in an effective reduction of the interfacial energy, which is believed to be the main reason for the formation of the two-phase (Ti_xNb_1–x)N (0.76≥x>0.72) and (Ti_xNb_1–x)C (0.29≥x>0.09) particles during the rough rolling stage of the ISP process. In addtion, the formation of a high density of geometrically necessary dislocations at the TiN/austenite interface during rough rolling provides energetically favorable nucleation sites. Both effects result in a premature formation of NbC and this greatly reduces the solute Nb content prior to finish rolling.

Not only is the NbC precipitation on TiN accelerated during the rough rolling deformation by an increased dislocation density but the Ti/(Nb+Ti) ratio of the cap-shaped deposit is also increased as shown in Fig. 2(d). In this study, this compositional change appeared to occur during the growth rather than being the result of the diffusion of Ti from the cuboid core to the cap-shaped deposit.

5. Conclusions

The main results of the present study can be summarized as follows:

Two-phase compound particles were formed before and during the rough rolling of a Ti-added Nb HSLA steel during the simulation of the ISP process. The particles were composed of a TiN-type cube-shaped core and a NbC-type cap-shaped deposit. The deposit formed epitaxially on one of the {100} facets of the cube-shaped precipitate.

The composition of the cube-shaped core was Ti_xNb_1–xN (0.76≥x>0.72). The most likely composition of the cap-shaped deposit was NbC.

A gradual increase in the lattice parameter was measured in the transition region from the cube-shaped core to the cap-shaped deposit. This lattice parameter increase was associated with the change of the Ti/(Nb+Ti) ratio in the interface region.

The results also suggest that during rough rolling the Ti content of the cube-shaped core decreased during the hot deformation and that in the transiton region the Ti-content of the cap-shaped deposit increased due to Ti diffusion from the cuboid core to the cap-shaped deposit.

The results of model calculations show that premature nucleation and growth of NbC during rough rolling in the ISP process results from a lower interfacial energy due to the gradual change of the lattice parameter and the composition at the interface between the TiN-type core and the cap-shaped NbC-type deposit.

Acknowledgement

The authors gratefully acknowledge the support of the POSCO Technical Research Laboratories, Gwangyang, South Korea.

References

• 1)   N.  Zapuskalov: ISIJ Int., 43 (2003), 1115.
• 2)   J. M.  Rodriguez-lbabe: Thin Slab Direct Rolling of Microalloyed Steels, Vol. 33, Trans Tech Publications, Switzerland, (2007), 2.
• 3)   A.  Najafi-Zadeh,  J. J.  Jonas and  S.  Yue: Metall. Trans. A, 23 (1992), 2607.
• 4)   J. H. P.  De Bresser,  J. H.  Ter Heege and  C. J.  Spiers: Int. J. Earth Sci., 90 (2001), 28.
• 5)   N.  Fujita,  T.  Narushima,  Y.  Iguchi and  C.  Ouchi: ISIJ Int., 43 (2003), 1063.
• 6)   F. B.  Pickering: The Institute of Materials, 10 (1997).
• 7)   R. M.  Poths,  R. L.  Higginson and  E. J.  Palmiere: Scr. Mater., 44 (2001), 147.
• 8)   N.  Maruyama and  G. D. W  Smith: Mater. Sci. Eng. A, 327 (2002), 34.
• 9)   B.  Soenen,  A. K.  De,  S.  Vandeputte and  B. C.  De Cooman: Acta Mater., 52 (2004), 3483.
• 10)   E.  Courtois,  T.  Epicier and  C.  Scott: Micron, 37 (2006), 492.
• 11)   E.  Kozeschnik,  J.  Svoboda and  F. D.  Fischer: Calphad, 28 (2004), 379.
• 12)   S. G.  Hong,  K. B.  Kang and  C. G.  Park: Scr. Mater., 46 (2002), 163.
• 13)   G. K.  Tirumalasetty,  M. A.  Van Huis,  C. M.  Fang,  Qiang Xu,  F. D. Tichelaar, D. N. Hanlon, J. Sietsma and H. W. Zandbergen: Acta Mater., 59 (2011), 7406.
• 14)   S. V.  Subramanian,  G.  Zhu,  H. S.  Zurob,  G. R.  Purdy,  G. C.  Weatherly,  J.  Patel,  C.  Klinkenberg and  R.  Kaspar: Int. Conf. on Thermomechanical Processing, The University of Sheffield, England, (2002), 148.
• 15)   A. J.  Craven,  K.  He,  L. A. J.  Garvie and  T. N.  Baker: Acta Mater., 48 (2000), 3857.
• 16)   Y.  Qin,  H.  Luo,  B.  Kooi and  J. Th. M.  De Hosson: Int. Conf. on HSLA Steel, CSM, Sanya, Hainan, China, (2005), 1.
• 17)   H. J.  Jun,  K. B.  Kang and  C. G.  Park: Scr. Mater., 49 (2003), 1081.
• 18)   D. A.  Porter,  K. E.  Easterling and  M. Y.  Sherif: Phase Transformations in Metals and Alloys, CRC Press, New York, (2009), 147.
• 19)   J.  Kunze: Met. Sci., 16 (1982), 217.
• 20)   W. J.  Liu: PhD Thesis, McGill University, (1987).
• 21)   S.  Akamatsu,  M.  Hasebe,  T.  Senuma,  Y.  Matsumura and  O.  Akisue: ISIJ Int., 34 (1994), 9.
• 22)   T. H.  Hoogendoorn: Micro Alloying 75, Bureau of Standards, Washington, DC, (1975), 61.
• 23)   K.  Narita: ISIJ Int., 15 (1975), 145.
• 24)   B.  Aronsson and  H.  Nordberg: J. Iron Steel Inst., 206 (1968), 1263.
• 25)   T. H.  Johansen,  N.  Christensen and  B.  Augland: Trans. Am. Inst. Min. Engrs., 239(1967), 1651.
• 26)   J.  Copreaux and  P.  Delanneau: ECSC Report 7210-EC/203, European Commission, Luxembourg, (1993).
• 27)   R.  Fountain and  J.  Chipman: Trans. AIME, 212 (1958), 737.
• 28)   H.  Ohtani,  T.  Tanaka,  M.  Hasebe and  T.  Nishizawa: Calphad, 12 (1988), 225.
• 29)   S.  Nagakura and  S.  Oketani: ISIJ Int., 8 (1968), 265.
• 30)   R. E.  Treece,  M. S.  Osofsky,  E. F.  Skelton,  S. B.  Qadri,  J. S.  Horwitz and  D. B.  Chrisey: Phys. Rev. B, 51 (1995), 9356.
• 31)   P.  Haas,  F.  Tran and  P.  Blaha: Phys. Rev. B, 79 (2009), 085104.
• 32)   K.  Aigner,  W.  Lengauer,  D.  Rafaja and  P.  Ettmayer: J. Alloys Compd., 215 (1994), 121.