Austenite Grain Boundary Pinning during Reheating by Mixed AlN and Nb(C,N) Particles

Transcription

1 , pp Austenite Grain Boundary Pinning during Reheating by Mixed AlN and Nb(C,N) Particles Amrita KUNDU* School of Metallurgy and Materials, University of Birmingham, Birmingham B15 2TT, UK. (Received on July 2, 2013; accepted on October 18, 2013) The present study investigates the role of aluminium nitride (AlN) on grain boundary pinning during reheating in presence of niobium carbonitrides (Nb(C,N)) and subsequent evolution of grain size distribution. Three as continuously cast slabs of high strength low alloy steels containing different levels of Nb (between wt%) and Al (between wt%) have been characterized in terms of phase balance, ferrite grain size distribution and particle size distribution. Precipitate distributions have also been determined following simulated reheating schedules. The prior austenite grain size distributions after reheating have been correlated with precipitate distribution. Quantification of precipitate and prior austenite grain size distributions after reheating unveils the governing mechanisms for precipitate dissolution/ coarsening. It is observed that grain boundary pinning is controlled by AlN at temperatures below C, but by Nb(C,N) at higher temperatures. KEY WORDS: grain boundary pinning; microsegregation; bimodality; high strength low alloy steel. 1. Introduction High strength low alloy (HSLA) steels contain small amounts of Titanium, Niobium and/or Vanadium to allow higher strength levels to be achieved with lower carbon contents without any loss in toughness, weldability or formability. 1 6) Fine and uniform ferrite grain size distribution along with fine scale carbides and carbonitride particles of Niobium, Titanium and Vanadium ensures strength and toughness 5,7,8) in these steels. Formation of this microstructure largely depends on the processing history of the steel. 5) These steels are generally processed by continuous casting followed by thermomechanical control rolling. 5,8,9) Niobium is known to be the most effective microalloying element to achieve refinement of the final grain structure in HSLA steels. 5,8) Microalloying precipitates formed during and after solidification as the steel is continuously cast, dissolve either fully or partially during slab reheating (an intermediate stage in processing). 5,10 13) During slab reheating the austenite grain growth behaviour is affected by the presence of microalloy carbides or carbonitrides which restricts the austenite grain growth. Hence localised and/or partial dissolution of microalloy carbides/carbonitrides may lead to inhomogeneous austenite grain growth during reheating stage. 14,15) It has been found that some Nb and Nb-V microalloyed steel plates exhibit a duplex grain size. 10,14,15) Such evolution of bimodal grain structure is associated with the microsegregation of microalloying elements, particularly Nb, during solidification ) Niobium is known to partition to the * Corresponding author: akundu05@gmail.com DOI: interdendritic regions during solidification and also promote the partitioning of other elements such as carbon and nitrogen ) Such partitioning of the elements in the interdendritic regions modifies the local precipitation behaviour. Previous work 14 19) has also correlated the formation of bimodal grain structure with microsegregation of Nb during casting of the steel and subsequent variation in precipitate distribution and stability during reheating and deformation. The formation of a bimodal grain size is undesirable and may require the product to be downgraded due to the reduction in mechanical properties ) Deterioration in toughness due to the formation of a duplex, or bimodal, grain size has been observed. 20,21) It is, therefore, of prime importance to understand the evolution of austenite grain size and distribution in order to achieve a fine and uniformly distributed grain size in the final product (plate, strip or sections). The grain size of the final product is influenced by each of the stages of processing (casting, reheating and rolling). In order to understand the formation and severity of a duplex structure, as well as the average grain size, each of these processing stages needs to be considered. The segregation tendency of Al is much less as compared to that of Niobium. 15) Hence instead of Nb(C,N) particles, if AlN particles are used to pin the grain boundaries in regions of reduced Nb(C,N) volume fraction and stability during reheating the precipitate distribution will be similar in the dendritic and interdendritic regions. This is expected to result in the formation of unimodal grain structures on reheating. The present paper is concerned about the dissolution and coarsening of AlN and Nb(C,N) in dendritic and interdendritic regions during reheating stage and subsequent development of the austenite grain structure on cooling in three ISIJ

2 different continuously cast slabs of low carbon HSLA steel with varying Nb and Al content. 2. Experimental Three commercial low carbon ( 0.1 wt%) steels were supplied as sections of as-cast slabs 300 mm thick and mm wide. The same continuous casting process parameters were used for all three slabs. The steels contain variations in the amounts of Nb and Al along with smaller differences in the Ti, V, Ni, Si, Mn, P and S levels, Table 1. Thermo-Calc software (version Q) was used to predict the solidification sequences during casting; partitioning of elements within solid and liquid phases during solidification; precipitation sequence; and stability of precipitates. Samples ( mm) from the quarter thickness position of the as-cast slabs were sealed in evacuated silica tubing and then reheated (3 C/min) at 1125 C (for 1, 2 and 8 hours) and 1150 C (1 hour). After reheating the samples were water quenched. The as-cast and reheated specimens were mounted in Bakelite and then metallographically polished to 0.05 μm SiO 2 finish. The metallographically polished as-cast sample was etched with 2% nital solution. Etching the polished surface in a saturated aqueous picric acid solution at 60 C revealed the prior austenite grain structures of the reheated samples. The as-cast and reheated specimens were examined in a Leica DMRX microscope equipped with KS300 image analysis software to quantify secondary dendritic arm spacing (SDAS), cast and heat treated structures, coarse (> 1 μm) inclusion, grain size (equivalent circle diameter (ECD)) and fraction of second phase (pearlite/ bainite). About grains were measured in terms of individual grain area and ECD (obtained directly from KS300 software) of the as-cast and reheated samples to construct grain size distribution plot. Energy dispersive X-ray spectroscopy (EDS) area analyses covering an area of 50 μm 2 (10 μm length and 5 μm width) at each spot were carried out in a Jeol 7000 SEM (scanning electron microscope) to determine the spatial variation of Nb at a 10 μm spacing using an accelerating voltage of 10 kv in order to decrease the interaction volume. The K and L (for Nb) characteristic X ray lines were quantitatively analyzed using INCA software (version 17 B). Precipitates in the as-cast slabs and after reheating were characterized using JEOL 7000 Field emission gun SEM. The number density and area fraction of precipitates resolvable in the SEM (> 20 nm) were measured over 50 continuous fields of view (each field of view covers ~ 12 μm 2 area). The particle composition was determined using the Oxford INCA EDS system for an SEM operating voltage of kv and 10 mm working distance. 3. Results and Discussion 3.1. Thermodynamic Calculations of Solidification Sequence Microsegregation of microalloying elements, such as, Nb will lead to local variations in the response of the slab to reheating and rolling resulting scatter in the final plate mechanical properties. 15,20,21) Hence accurate prediction of the solidification sequence or, in other words, quantification of the solute-rich and solute-depleted fractions is essential to understand the evolution of subsequent microstructure. Table 2 shows the solidification sequence in all the three continuously cast slabs based on thermodynamic calculations. Due to lower carbon content the Slab 3 undergoes complete equilibrium solidification as δ -ferrite instead of first solidifying as primary δ -ferrite followed by a mixture of austenite and δ -ferrite (Slabs 1 and 2).Thermo-Calc predicted partition ratio (a high value indicating a stronger tendency to segregate) Nb can be as high as 7 between last liquid and first solid. Aluminum, on the other hand, shows very little tendency to segregate to the solid phase during solidification (either δ -ferrite or austenite) with a partition ratio of approximately 0.8 between the final liquid to solidify and the solid phase. From the above it follows that during continuous casting of thick slab the initial metal to solidify at the interface with the water-cooled Cu mould will be Nb depleted δ ferrite. As a result microsegregation of elements (e.g. Nb) to interdendritic or cellular boundary areas would coincide with the δ /δ phase boundaries. If full transformation from δ ferrite to austenite occurs prior to precipitation of Ti and Nb rich phases then the centre of the austenite grains will correspond to the solute rich region whilst austenite/austenite phase boundaries would remain as solute depleted. As cooling proceeds the higher solute content in the austenite grain centres will provide higher driving force for precipitation. It means that the precipitates would form at higher temperatures and the time available for their growth will be higher. Additionally, the precipitate volume fraction will be higher Table 2. Slab 1 Temperature ( C) Slab 2 Temperature ( C) Slab 3 Temperature ( C) Thermo-Calc predicted solidification sequence of the as cast slabs. Solidification Sequence L L+δ L+δ +γ δ+γ L L+δ L+δ +γ δ+γ L L+δ δ Table 1. Chemical compositions (wt%) of the as-cast materials. Sample C Si Mn P S Ni Al Nb Ti N V Slab Slab Slab ISIJ 678

3 due to the greater amount of solute. The number density of the precipitate particles depends upon the nucleation rate and so on the precise relationship between the cooling rate and the continuous cooling transformation behaviour of the prior austenite structure. Since the γ /γ grain boundaries will tend to transform to α-ferrite and the centre of the austenite grains will transform to pearlite it is expected that Nb(C,N) precipitate size and number density will be greater in pearlite, and in the ferrite grains close to the pearlite regions in case of low carbon steels with limited pearlite content. Area percent of the second phase materials represent the total solute-rich material present in the final structure. (c) (d) Fig. 1.. Microstructure from the quarter thickness position of the as cast Slab 2.. The ferrite grain size distribution from the quarter thickness position of the Slab 2. (c). SEM images for Slab 3 showing the line along which the EDS traces were taken. (d). EDS traces showing the weight % variation of Nb across the secondary dendrites Microstructure and Precipitates of As-Cast Slabs The microstructure of the as cast slabs at the quarter thickness position consists of ferrite and pearlite. The ferrite grain size distribution is unimodal. The area % of the pearlite in the Slab 1, 2 and 3 are 15, 16 and 15% respectively. The as cast microstructure of Slab 2 and the ferrite grain size distribution are shown in Figs. 1 and 1. EDS area analyses for Nb was carried out across regions containing both solute-rich and solute-depleted materials. One example is shown in Figs. 1(c) and 1(d). The corresponding elemental variation across the dendritic arm spacings, Fig. 1(d), indicates a steep rise in the Nb content on crossing from dendritic α to the second phase. Within the dendrite arms the profile for Nb is very level, albeit at low values. There is more apparent scatter through the interdendritic region, although the higher solute contents would make this more obvious. The measured partition ratio of Nb is 7 which matches with the Thermo-Calc predicted value. Figure 2 shows an array of Nb(C,N) precipitates decorating a prior boundary (δ ferrite or austenite) in the as-cast structure of Slab 2. The Nb(C,N) precipitate size distributions in the solute-rich and in the solute-depleted regions of Slab 2 show that Nb(C,N) precipitates are present in the size range of ~ nm (Fig. 2) in both regions. Table 3 shows the extent of inhomogeneity in the distribution of Nb(C,N) precipitates in Slabs 1, 2 and 3 quantified in terms of the average (local) precipitate number density (number/mm 2 ) in the solute-rich and solute-depleted regions, i.e., at the interdendritic and dendrite center regions. The number densities of Nb(C,N) precipitates are observed higher in Slab 1 as compared to Slab 2 and 3. This is due to the higher average Nb level in Slab 1. At the quarter- thickness position, the average precipitate density of interdendritic regions is ~ 4 to 6 times higher than that of the dendrite center regions (Table 3) which is consistent with the segregation tendency of Nb. The SEM investigation reveals that the AlN precipitates are faceted and irregular in shape as shown in Fig. 2(c). These precipitates were identified from their dark appearance in the back-scattered electron (BSE) image and from the presence of the Al peak in the EDS spectrum. The size distribution of AlN particles at the quarter-thickness position of the Slab 2 (mean particle size of 150 nm) is shown in Fig. 2(d). The AlN particles were found randomly distributed throughout the matrix. As Al shows limited segregation tendency compared to Nb the number density of the particles is expected to be similar both in the solute-rich and in the solute-depleted regions. However, a slightly higher number density of AlN particles was found in the solute-depleted regions. This happens because of greater availability of nitrogen in the solute-depleted regions. The total number density of the particles present in Slab ISIJ

4 (c) (d) Fig. 2.. Array of Nb-rich particles, probably Nb(C,N), with EDS trace showing the Nbpeakin the solute-rich and solutedepleted regions in the as cast Slab 2.. Number density distribution of Nb(C,N) particles in the solute-rich and solute-depleted regions in the as cast Slab 2. (c). Array of Al-rich particles, expected to be AlN, with EDS trace showing the Al peak in the solute-rich and solute-depleted regions in the as cast Slab 2. (d). Number density distribution of AlN particles in the solute-rich and solute-depleted regions in the as cast Slab 2. Table 3. Slab Slab 1 Slab 2 Slab 3 The measured area fraction, number density and average size of the Nb(C,N) and AlN particles in the solute-rich and in the solute-depleted regions of the as-cast slabs. Type of Particles Area fraction Number density (mm 2 ) Average size (nm) Solute-rich Solute-depleted Solute-rich Solute-depleted Solute-rich Solute-depleted AlN (14±4) 10 6 (22±9) 10 6 (2±1) 10 3 (3±1) Nb(C,N) (4±1) 10 4 (6 ±2) 10 5 (115±38) 10 4 (20±8) AlN (6±2) 10 5 (9±4) 10 5 (3±1) 10 4 (44±14) Nb(C,N) (18±7) 10 5 (42±14) 10 6 (84±24) 10 4 (18±7) AlN (32±12) 10 6 (4±1) 10 5 (12±5) 10 3 (16±6) Nb(C,N) (21±9) 10 5 (44±16) 10 6 (85±23) 10 4 (19±7) (i.e., Nb(C,N) and AlN) in the solute-depleted regions is mm 2, greater than that is present in Slab 1 ( mm 2 ) and 3 ( mm 2 ) (Table 3) in the solute-depleted regions. Due to the higher number density and area fraction of Nb(C,N) and AlN in the solute-depleted regions greater grain boundary pinning is expected in Slab 2 during reheating as compared to Slab 1 and ISIJ 680

5 3.3. Thermo-Calc Predicted Precipitate Stability The Thermo-Calc predicted stability of (Nb,Ti,V)(C,N) and AlN in the solute-rich and solute-depleted areas is shown in Fig. 3 for Slab 2. The same behavior is observed for Slab 1 and 3. The compositions of the last liquid and first solid were used as input to predict the precipitate stability in the solute-rich and solute-depleted regions respectively. The reheating temperatures were selected based on this diagram. Reheating in the temperature range of approximately C results in near complete dissolution of the (Nb,Ti,V)(C,N) precipitates in the solute-depleted regions, but retention of the majority of these carbonitrides in the solute-rich regions. Figure 3 predicts that a bimodal grain structure would be developed for reheating below C based on the Nb(C,N) particles stability alone. The presence of AlN in the solute-depleted region means that below ~ C a unimodal structure should form if AlN particles are effective in pinning the grain boundaries. In the present study reheating was carried out at and C (i.e. in the commercial reheat temperature range) to see whether the presence of AlN particles in the solute-depleted regions provides sufficient pinning in three slabs to stop grain growth in the solute-depleted regions and hence avoid the formation of bimodal grain size distribution in the commercial reheating temperature range. size distributions with the area percent of coarse grains (grain size class staring from μm to μm in Fig. 4, i.e. grain size classes present in the second peak area) 42, 29 and 37% respectively. This suggests that greater grain boundary pinning is available in the Slab 2 and this is consistent with the highest number density of pinning particles in the solute-depleted regions (Table 4). In all three slabs the total amount of coarse grains present in the distribution is less than the amount of solute depleted material (~ 84 85%) present in the slabs. This clearly suggests that some grain boundary pinning is available in the solutedepleted regions. The number density of Nb(C,N) and AlN particles in the solute-rich and solute-depleted regions in all the slabs are listed in Table 4. The particle size distribution of Slab 2 is shown in Figs. 5 and 5. Around 90% of the Nb(C,N) particles in the solute-depleted regions is dissolved after reheating at C (compared to that in the ascast condition) whilst the remaining ones are coarsened to a mean size of 300 nm. Finer particles with a greater area fraction are present in the solute-rich regions providing grain 3.4. Reheated Microstructures The microstructure and prior austenite grain size distribution of Slab 2 are shown in Fig. 4. The reheated samples of Slabs 1, 2 and 3 showed microstructures with bimodal grain Fig. 3. Thermo-Calc predicted stability of NbTiV(C,N) and AlN in Slab 2. Fig. 4.. Microstructure showing prior austenite grains after reheating at C of Slab 2.. Prior austenite grain size distribution after reheating at C of Slab 2. Table 4. Area fraction and number density of Nb(C,N) and AlN particles in the solute-rich and solute-depleted regions of Slab 1, 2 and 3. Slab Slab 1 Slab 2 Slab 3 Reheat temperature ( C) hour hour hour Type of Particles Area fraction Number density (mm 2 ) Solute-rich Solute-depleted Solute-rich Solute-depleted AlN (24±9) 10 7 (3±1) 10 6 (2±1) 10 2 (4±2) 10 2 Nb(C,N) (10±4) 10 5 (8±2) 10 6 (49±14) 10 4 (15±6) 10 3 AlN (42±15) 10 7 (27±11) 10 6 (10±3) 10 2 (62±21) 10 2 Nb(C,N) (76±21) 10 6 (2±1) 10 6 (29±11) 10 4 (10±2) 10 3 AlN (31±14) 10 7 (9±4) 10 6 (5±1) 10 2 (7±2) 10 2 Nb(C,N) (8±3) 10 5 (3±1) 10 6 (31±12) 10 4 (12±3) ISIJ

6 Table 5. Zener drag of Slab 1, 2 and 3 in the solute-rich and solutedepleted regions at C. Relative Zener drag Slab Solute-rich regions Solute-depleted regions Slab Slab Slab Fig. 5.. Number density distributions of Nb(C,N) particles in the solute-rich and solute-depleted regions after reheating at 1150 C, compared to that present in the as-cast slab of Slab 2.. Number density distributions of AlN particles in the solute-rich and solute-depleted regions after reheating at C, compared to that present in the as-cast slab of Slab 2. boundary pinning in these regions. The measured area percentage of AlN particles in the solute-depleted regions is (0.009 in as-cast condition) and the mean size of the particles present is 350 nm. In the solute-rich regions the mean size of the AlN particles is 350 nm with a reduced area percentage of The small number of coarse AlN particles are not effective in giving enough pinning to stop grain growth in the solute-depleted regions. This result indicates that the grain boundary pinning at C in both solute-rich and solute-depleted regions is controlled by Nb(C,N). In order to quantify the difference in pinning between solute-rich and solute-depleted regions the relative Zener drag value in both regions needs to be calculated. The average grain diameter (D) can be related to the average diameter (d) of the pinning particles and to the volume fraction (f ) of particles by the following equation: 23) D d = ξ. f... (1) where, ξ is a constant. Gladman 5,11,24) and Hillert 25) considered the inhomogeneity in grain size distribution and introduced the grain size π heterogeneity factor, Z. Gladman 5) 3 2 suggested, ξ =, 6 2 Z and compared his model with the experimental studies on the growth of austenite grains in the presence of NbC and AlN particles. 5,11,24) For the normally observed range of Z values ( ) excellent agreement was reported between Gladman s model and experimental data. 5,11,24) Considering Z=2, for which ξ =0.26, and using the experimentally measured volume fractions and average precipitate sizes, the limiting grain sizes can be predicted in the soluterich and solute-depleted regions of the reheated samples. The relative Zener drag in the solute-rich and solutedepleted regions has been calculated by taking the ratio of the limiting grain size in the solute-rich and solute-depleted regions. Zener drag results from the combined area fractions and sizes of Nb(C,N) and AlN particles and the relative levels of Zener drag in the solute-rich and solute-depleted regions are shown in Table 5. The large difference in Zener drag between solute-rich and solute-depleted regions results in faster coarsening in the solute-depleted regions and the bimodal grain size distribution. The largest bimodality in the Slab 1 is therefore linked with largest difference in Zener drag between solute-rich and solute-depleted regions. A lower reheating temperature of C retains a greater volume fraction of AlN in the solute-depleted regions as compared to 1150 C with approximately the same amount of (Nb,Ti,V)(C,N) in that region (Fig. 3). In case of Slab 2 the total volume fraction of particles is predicted to be in the solute-depleted regions and in the solute-rich regions. Assuming similar particle size distributions as seen in the as-cast condition it is expected that the grain boundary pinning would be more even between the solute-rich and solute-depleted regions and that should result in less coarsening of the particles. Figures 6 and 6 shows that the microstructure and prior austenite grain size distribution of Slab 2 after reheating for 1 hour C is bimodal with ~ 12% coarse grains by area (27% in case of Slab 1 and 19% in case of Slab 3). This reduced bimodality (12% coarse grains by area (i.e. area percent of grains present in the second peak region of the grain size distribution) are present after reheating at 1125 C as opposed to 29% after reheating at 1150 C, hence reduced bimodality) than that evolved after reheating at 1150 C indicates that the volume fraction of AlN particles are providing grain boundary pinning in the solute-depleted regions, but not completely as expected from equilibrium volume fraction values. Figure 7 shows the presence of AlN particles at the prior austenite grain boundaries. From the number density distribution of Nb(C,N) particles it is clearly observed (Fig. 7) that there is almost no change in number density of the Nb(C,N) particles in the solute-depleted regions compared to C as predicted by Thermo-Calc. However, a greater number of finer particles of AlN is present in the solute-depleted regions as compared to reheating at C (Fig. 7(c)). Therefore AlN particles should dominate grain boundary pinning at 1125 C in solute-depleted regions as there is no change in area fraction of Nb(C,N) 2014 ISIJ 682

7 Fig. 6.. Microstructure showing prior austenite grains after reheating at C of Slab 2.. Prior austenite grain size distribution after reheating at C of Slab 2. Table 6. Zener drag of Slab 1, 2 and 3 in the solute-rich and solutedepleted regions at C. Relative Zener drag Slab Solute-rich regions Solute-depleted regions Slab Slab Slab particles at C as compared to C, whereas AlN and Nb(C,N) should both act in solute-rich regions. The coarsening of AlN at 1125 C in the solute-depleted regions is also less as compared to 1150 C. The average AlN particle size after C is 175 nm, finer than that at 1150 C (350 nm). Using measured particle area fraction, mean size values and the limiting grain size, the relative grain boundary pinning forces in solute-rich and in solutedepleted regions were also determined (Table 6). The smaller difference in Zener drag between solute-rich and solutedepleted regions results in more uniform grain boundary movement i.e., reduced bimodality. Although AlN particles are providing greater grain boundary pinning in solutedepleted regions, parity between solute-rich and solutedepleted materials is not achieved as the particle distributions are not equilibrium ones after 1 hour holding at the reheating temperatures. As cast specimen from Slab 2 was reheated for 2 and 8 hours at C. Extended holding for 2 and 8 hours at 1125 C caused further dissolution of both Nb(C,N) and AlN particles. Figures 8 and 8 shows the variation in the number density distribution of AlN particles and prior austenite grain size distribution. The degree of bimodality is (c) Fig. 7.. SEM image of prior austenite grains after reheating at 1125 C of Slab 2 showing the presence of AlN particles at the prior austenite grain boundary providing grain boundary pinning.. Number density distributions of Nb(C,N) particles in the solute-rich and solute-depleted regions after reheating at C of Slab 2. (c). Number density distributions of AlN particles in the solute-rich and solutedepleted regions after reheating at C of Slab 2. increased for 2 hour (17% coarse grains in the distribution) reheating at 1125 C. The area fraction of the AlN particles in the solute-depleted regions after 2 hour holding is (12±4) 10 6 and after 1 hour holding it was (31±12) ISIJ

8 Fig. 8. This increase in area % of coarse grains is associated with the ~ 60% dissolution of the AlN particles within an hour. After 8 hours holding at 1125 C the grain size distribution is observed as a skewed unimodal distribution. This is consistent with the gradual dissolution of AlN particles in the solute-depleted regions. After 8 hours holding the area fraction of the AlN particles present in the solute-depleted regions is (48±14) 10 7, which is only ~ 4% of the particles present in the in the solute-depleted regions under as-cast condition, the mean size of the particles being 350 nm. This small number of coarse particles does not provide any significant pinning in the solute-depleted regions. As a result ~ 84% coarse grains is present in the distribution (i.e., total amount of solute depleted material). This would suggest that pinning for the equilibrium particle distributions in soluterich and solute-depleted regions should be uniform, but that, in practice, the pinning is greater due to undissolved particles in the solute-rich region resulting in differences in pinning between solute-rich and solute-depleted regions. Thus, formation of bimodal grain size distribution is consistent with AlN particles being present during reheating. Gradually it is dissolved to release grain boundaries and give a proportion of coarse-grained closer to 84%, i.e. the total solutedepleted materials present in the structure. 4. Conclusions. Number density distributions of AlN particles after reheating at C for 1, 2 and 8 hours of Slab 2.. Prior austenite grain size distribution after reheating at 1125 C for 1, 2 and 8 hours of Slab 2. Three continuously cast slabs with varying Nb and Al content have been investigated to see whether AlN particles are effective in pinning austenite grain boundaries during reheating to offset thedifference in grain boundary pinning between solute-rich and solute-depleted regions due to segregation of Nb. The following conclusions have been drawn. (1) The observed bimodality in grain size distribution is due to the difference in grain boundary pinning between solute-rich and solute-depleted regions. (2) Increased amounts of AlN particles can offset the difference in pinning behaviour due to Nb micro-segregation at C. At higher temperature (1 150 C) grain boundary pinning is controlled by Nb(CN) particles. (3) Quantitative correlation of coarse and fine grains with solute-depleted and solute-rich regions is not seen for shorter heating times due to incomplete dissolution of particles to equilibrium level. Prolonged holding at the reheating temperature results in greater particle dissolution and grain boundary pinning closer to that expected. Acknowledgements The author thanks Tata Steel Ltd. for the provision of test material, Professor Claire Davis and Dr. Martin Strangwood, School of Metallurgy and Materials, the University of Birmingham for useful discussions. Thanks are due to Professor Paul Bowen for the provision of research facilities at the University of Birmingham. The author is also grateful to The Universities, UK for awarding the scholarship to carry out her research in the UK. REFERENCES 1) W. B. Morrison: Mater. Sci. Technol., 25 (2009), ) A. J. DeArdo, M. J. Hua, K. G. Cho and C. I. Garcia: Mater. Sci. Technol., 25 (2009), ) A. Bakkaloglu: Mater. Lett., 56 (2002), ) P. D. Hodgson, H. Beladi and M. R. Barnet: Mater. Sci. Forum, (2005), 39. 5) T. Gladman: The Physical Metallurgy of Microalloyed Steels, The Institute of Materials, London, UK, (1997), ) M. J. Luton, R. Dorvel and R. Petkovic: Metall. Trans A, 11A (1980), ) L. J. Cuddy: Metall. Trans. A, 12A (1981), ) A. J. DeArdo: Int. Mater. Rev., 48 (2003), ) L. J. Cuddy: Metall. Trans. A, 15A (1984), ) E. J. Palmiere, C. I. Garcia and A. J. DeArdo: Metall. Mater. Trans. A, 25A (1994), ) T. Gladman and F. B. Pickering: J. Iron Steel Inst., 205 (1967), ) L. J. Cuddy and J. C. Raley: Metall. Trans. A, 14A (1983), ) M. Ali Bepari: Metall. Trans. A, 20A (1989), ) C. L. Davis and M. Strangwood: J. Mater. Sci., 37 (2002), ) C. L. Davis: Trans. Ind. Inst. Met., 59 (2006), 1. 16) C. L. Davis and M. Strangwood: Mater. Sci. Technol., 25 (2009), ) A. Kundu, C. L. Davis and M. Strangwood: Mater. Manuf. Process., 25 (2010), ) A. Kundu, C. L. Davis and M. Strangwood: Metall. Mater. Trans. A, 41 (2010), ) A. Kundu, C. L. Davis and M. Strangwood: Metall. Mater. Trans. A, 42 (2011), ) S. J. Wu and C. L. Davis: Mater. Sci. Eng. A., (2004), ) D. Bhattacharjee, C. L. Davis and J. F. Knott: Ironmaking Steelmaking, 30 (2003), ) D. Chakrabarti, C. L. Davis and M. Strangwood: Metall. Mater. Trans. A, 39A (2008), ) C. Zener, quoted by C. S. Smith: Trans. Met. Soc. AIME, 175 (1948), ) T. Gladman: Proc. R. Soc. (London), 294 (1966), ) M. Hillart: Acta Metall., 13 (1965), ISIJ 684