当前位置:首页 >> 天文/地理 >>

Effect of austenite microstructure and cooling rate on..in a low carbon Nb–V microalloyed stee


Materials Science and Engineering A 528 (2011) 2559–2569

Contents lists available at ScienceDirect

Materials Science and Engineering A
journal homepage: www.elsevier.com/locate/msea

Effect of austenite microstructure and cooling rate on transformation characteristics in a low carbon Nb–V microalloyed steel
M. Olasolo, P. Uranga, J.M. Rodriguez-Ibabe, B. López ?
CEIT and TECNUN (University of Navarra), P? de Manuel Lardizabal 15, 20018 Donostia-San Sebastián, Basque Country, Spain

a r t i c l e

i n f o

a b s t r a c t
Deformation dilatometry has been used to simulate controlled hot rolling followed by cooling of a Nb–V low carbon steel, looking for conditions corresponding to wide austenite grain size distributions prior to transformation. Recrystallization and non-recrystallization deformation schedules were applied, followed by controlled cooling at rates from 0.1 ? C/s to about 200 ? C/s, and the corresponding continuous cooling transformation (CCT) diagrams were constructed. The resultant microstructures ranged from polygonal ferrite (PF) and pearlite (P) at slow cooling rates to bainitic ferrite (BF) accompanied by martensite (M) for fast cooling rates. Plastic deformation of the parent austenite accelerated both ferrite and bainite transformations, displacing the CCT curve to higher temperatures and shorter times. However, it was found that the accelerating effect of strain on bainite transformation weakened as the cooling rate diminished and the polygonal ferrite formation was enhanced. Moreover, it was found that plastic deformation had different effects on the re?nement of the microstructure, depending on the cooling rate. An analysis of the microstructural heterogeneities that can impair toughness behavior has been done. ? 2010 Elsevier B.V. All rights reserved.

Article history: Received 17 November 2010 Received in revised form 24 November 2010 Accepted 25 November 2010 Available online 2 December 2010 Keywords: Transformation Recrystallized/unrecrystallized austenite Microstructure re?nement EBSD Ferrite unit Misorientation angle

1. Introduction Steels with a minimum strength balanced with a high degree of toughness and excellent weldability are required in a wide range of applications. This combination of properties is achieved by optimizing the chemical composition and by an appropriate thermomechanical processing (TMP) schedule. Both are needed to develop a ?ne-grained microstructure and a favorable texture in the hot rolled plate. As weldability limits carbon content [1,2], suitable combinations of microalloying additions (Nb, V, Ti etc.) contribute to increase strength both directly, through microstructural re?nement, solid solution and precipitation strengthening, as well as indirectly, through enhanced hardenability and associated modi?cation of the resultant transformation microstructures. The suitable design of thermomechanical processing schedules involves both the controlled rolling and the subsequent accelerated cooling [3]. The controlled rolling includes the re?nement of the austenite at high temperatures by successive recrystallizations followed by heavy deformation carried out in the non-recrystallization temperature region. The strain retained in the austenite will affect the transformation behavior and may produce an increase in the austenite to ferrite transformation tem-

? Corresponding author. Fax: +34 943213076. E-mail addresses: jmribabe@ceit.es (J.M. Rodriguez-Ibabe), blopez@ceit.es (B. López). 0921-5093/$ – see front matter ? 2010 Elsevier B.V. All rights reserved. doi:10.1016/j.msea.2010.11.078

perature and a decrease in transformation time [4–6]. Similarly, the ferrite fraction with non-polygonal morphology may change as a result of the retained strain [7,8]. The accelerated cooling step, designed to take advantage of the increased steel hardenability, aims to suppress the formation of polygonal ferrite (PF) and, instead, encourages non-equilibrium, non-equiaxed ferrite microstructures to be formed. The latter transformation products are known to contribute to increasing strength, through both small effective grain size and higher dislocation densities [9]. Concerning low-temperature toughness properties, the grain size re?nement is also the key factor. Nevertheless, in contrast to what happens with strength controlled by microstructural mean values, the weakest link behavior of the brittle fracture implies that local microstructural features are the key factors. At very low testing temperatures the brittle process is associated with microcrack nucleation events. As temperature increases, the propagation of the microcrack across the matrix becomes the controlling step [10–12]. In those cases, the nature and homogeneity of the matrix, de?ned by high-angle boundaries able to arrest the microcracks, become the controlling microstructural parameters. The microstructural unit size that controls strength may differ from that controlling toughness. Both effective grain sizes are de?ned in terms of crystallographic misorientations. Thresholds misorientations of 4? and 15? are typical values to de?ne the grain size effective for strength and toughness, respectively [13,14]. Low-angle misorientation (4? ) grain size is one of the factors that

2560 Table 1 Chemical composition of the steel (wt%). C 0.06 Mn 1.20 Si 0.29 S 0.007 Al 0.035

M. Olasolo et al. / Materials Science and Engineering A 528 (2011) 2559–2569

Nb 0.062

V 0.053

N 0.0081

controls the yield and tensile strength of steel because such misorientation is expected to be effective in opposing dislocation movement, while high-angle boundaries (15? ) provide effective barriers to cleavage fracture. In some thermomechanical processes, as for example in thin slab direct rolling technologies, the re?nement and conditioning of austenite prior to transformation may be limited. The initial as-cast coarse grain size present at the entry of the ?rst stand joined to the low total reductions makes it dif?cult to reach enough re?nement before transformation [15,16]. This problem is more notorious when ?nal gages thicker than 10 mm are produced. In addition, in the case of Nb microalloyed steels, if early Nb(C, N) strain induced precipitation occurs, recrystallization may be delayed and even stopped, leading to the presence of relatively large and isolated austenite grains at the end of the rolling process. The transformation behavior of these coarse grains will be completely different to that corresponding to the surrounding ?ner grains. Summarizing, in this technology, depending on the chemical composition and total reduction applied during rolling, combinations of wide austenite grain size distributions with limited accumulated strains can be present prior to transformation in the run out table. In a recent work it has been observed that a higher fraction of coarse grain sizes in the microstructure increases the chance of ?nding those grains at the cleavage origin, leading to a wider scatter in fracture stress in comparison to more homogeneous grain size distributions [17]. The aim of the present work is to investigate the transformation characteristics of a low carbon Nb–V microalloyed steel after the application of low deformations in the austenite, considering the microstructural features that intervene both in the strength and toughness behavior. The thermomechanical simulations were carried out by deformation dilatometry followed by cooling in a wide range of cooling rates. The objective of these simulations was to develop austenite microstructures showing a wide range of grain sizes prior to transformation, which could approach to the conditions present at thin slab direct rolling in cases where small total reductions are applied. 2. Material and experimental procedure The chemical composition of the steel used (in wt%) is given in Table 1. The steel was provided as 12.65 mm thick plate produced by EAF-CSP technology. Deformation dilatometry was performed using a B?hr DIL805A/D quenching and deformation dilatometer. Solid dilatometry specimens with a diameter of 5 mm and a length of 10 mm were employed. Different thermomechanical schedules were applied as shown in Fig. 1. Once reheated at 1250 ? C during 5 min, all deformation schedules included a simulated “roughing” step performed at a temperature of 1150 ? C using a strain of 0.3 followed by post-deformation holding at this temperature for 12 s to ensure a complete recrystallization, but avoiding excessive grain growth (schedule A). The simulated non-recrystallization rolling schedule contained an additional “?nishing” step performed at a temperature of 950 ? C (schedule B) or 900 ? C (schedule C), which were situated in the non-recrystallization region, using a strain of 0.4. The average cooling rates measured in the temperature interval between 800 and 500 ? C ranged between 0.1 and 200 ? C/s. The dilatometry samples were sectioned along their longitudinal axis selecting the region corresponding to a maximum area fraction of nominal strain and reduced strain gradient. Fig. 2 shows

Fig. 1. Schematic representation of simulated rolling schedules: A, recrystallization and B and C, non-recrystallization schedules.

the equivalent strain distribution in the central cross section calculated using Abaqus/Explicit [18] for a dilatometry sample deformed following schedule A. All the specimens were tested by Vickers hardness using 10 kg load and examined by light microscopy after being etched with 2 pct nital. Phase volume fractions were quanti?ed using the point count method. This characterization was carried out in the aforementioned cross section parallel to the specimen axis. The information obtained was used to construct continuous-cooling-transformation (CCT) diagrams. The lever rule was used for quanti?cation of the transformed phases [19]. In order to analyze the austenite state prior to transformation, samples were quenched immediately after the application of deformation schedules. The samples were polished and etched in a solution of saturated picric acid. After the application of schedule A the austenite grain size was measured by the mean equivalent diameter method. The samples were prepared for EBSD (electron backscattered diffraction) observations. The specimens were polished down to 1 m and the ?nal polishing was with colloidal silica. The scan step length was 0.75 m and the total scanned area was about 920 × 460 m2 . Orientation imaging was carried out on a Philips XL30cp scanning electron microscope with W-?lament, using TSL (TexSEM Laboratories, UT) equipment. Different OIM (Orientation

Fig. 2. Equivalent strain distribution in the central cross-section of the dilatometry sample for schedule A. The section at which the microstructural observation and hardness tests were performed is also marked (discontinuous line).

M. Olasolo et al. / Materials Science and Engineering A 528 (2011) 2559–2569

2561

Fig. 3. Stress–strain curves obtained after the application of the different simulated rolling schedules.

Imaging Microscopy) options were used to de?ne parameters that can be related to the microstructure, among others, the 4? and 15? misorientation crystallographic effective grain sizes and the misorientation angle distributions. The effective grain size was calculated as the equivalent circle diameter related to the individual grain area [20]. 3. Results and discussion 3.1. Microstructure analysis Fig. 3 shows the stress–strain curves obtained for the different deformation schedules. The microstructural analysis of the samples immediately quenched after deformation indicate that the austenite is completely recrystallized after schedule A, while pancaked austenite is observed after schedules B and C. This con?rms that the second deformation step is given in the non-recrystallization region. The degree of accumulated strain, resulting in schedules B and C, can be estimated by applying a microstructural model developed for Nb microalloyed steels described in [21] that considers the evolution of the austenite grain size and determines the amount of accumulated strain. The model predicts that in both deformation schedules the recrystallized austenite volume fraction is less than 5%. In consequence, it can be assumed that all the strain of 0.4 applied in the second pass is accumulated. The austenite grain size distribution obtained after schedule A, represented as area fraction, is illustrated in Fig. 4. In this case the austenite microstructure is completely recrystallized with a mean

Fig. 4. Austenite grain size distribution obtained after the application of schedule A.

grain size of Do = 53 ± 2 m, which corresponds to an austenite grain boundary area per unit volume of SV = 38 mm?1 , (SV = 2/Do , [22]). It is worth emphasizing that the presence of some coarse grains is observed resulting in about 20% of the area being occupied by grains larger than 100 m. After schedules B and C, the accumulated strain provokes elongated austenite grains prior to transformation. Considering that a strain of ε = 0.4 is accumulated over a mean recrystallized grain size of 53 m, it results in a speci?c grain boundary area of SV = 44 mm?1 (SV calculated with the following expression [22]: SV = 1/Do [exp(ε) + exp( ? ε/2)]). The mean grain size of Fig. 4 is signi?cantly coarser than that expected in industrial thin slab direct rolling for ?nal thicknesses of 10–12.5 mm (with an initial thin slab of 55 mm, the austenite mean grain size could be close to 20–25 m [23]), but the tail of coarse grains and the level of accumulated strain achieved after schedules B and C can be considered as representative of those predicted for steels with Nb contents of 0.05% [23]. In consequence, the austenite grain size distributions after schedules A to C can provide a good approach to analyze the transformation behavior when a signi?cant fraction of coarse austenite grains is present. Fig. 5 shows several micrographs of the microstructure observed at different cooling rates both for recrystallized (schedule A) and unrecrystallized (schedules B/C) conditions of the parent austenite phase prior to transformation. Depending on the cooling rate, the transformed microstructures are complex and may contain martensite (M), bainite (B) with different morphologies, polygonal (PF) and quasi-polygonal (QF) ferrite and pearlite (P). The volume fraction of each phase and the measured hardness values are indicated in Table 2. For the faster cooling rate of CR = 100 ? C/s, a mixture of bainite and martensite was obtained in both conditions, although the amount of martensite formed from the recrystallized austenite was larger as re?ected by the higher hardness value, indicating enhanced hardenability for recrystallized austenite. The opposite condition corresponds to the slowest cooling rate of 0.3 ? C/s in the microstructure formed by polygonal ferrite and pearlite. The signi?cant re?nement of the ferrite grain size when transformation has occurred from unrecrystallized austenite should be noted (see Fig. 5). When the cooling rate increases to 1 ? C/s, the microstructure continues being a mixture of ferrite and pearlite in the unrecrystallized austenite, but some bainite also forms in the recrystallized condition. Re?nement of the ferrite grain size is observed in both cases, comparing them with the 0.3 ? C/s cooling rate, but the difference in size between both austenite conditions remains. For intermediate cooling rates of 6 and 30 ? C/s, the microstructure is constituted by polygonal/quasi-polygonal ferrite and bainite, the amount of ferrite present for a given cooling condition being always greater for the unrecrystallized austenite (enhanced nucleation of ferrite grains given by the strain retained in the austenite before transformation). It was observed that the morphology of bainite changed as a function of the cooling rate and austenite microstructure. Increasing the cooling rate, the dominant morphology of bainite progressively changed from granular ferrite or granular bainitic ferrite (GF) to bainitic ferrite (BF) in accordance with the classi?cation of microstructures given in Refs. [24,25]. It should be noted that in the deformed austenite, the GF morphology predominated over the BF and that the grains were distributed in a more random manner, i.e., the prior austenite grain boundary networks were less evident in comparison to recrystallized austenite. Fig. 6 shows several FEG-SEM micrographs of these microstructural features. In Fig. 6(a), which corresponds to recrystallized austenite cooled at 100 ? C/s, a mixture of martensite (M) and bainite (mainly bainitic ferrite (BF)) is found. Fig. 6(b) shows a detail of this bainitic ferrite microstructure where the characteristic elongated ferrite grains separated by elongated M/A islands are observed.

2562

M. Olasolo et al. / Materials Science and Engineering A 528 (2011) 2559–2569

Fig. 5. Microstructure observed after deformation schedules A and B/C.

Fig. 6(c) shows a mixture of granular bainite (GF) and bainitic ferrite (BF) in a sample corresponding to recrystallized austenite cooled at 30 ? C/s. In Fig. 6(d) a detail of the morphology of granular bainite in a sample corresponding to recrystallized austenite cooled at 6 ? C/s, where a mixture of PF and bainite is obtained, is shown. The characteristic, more or less equiaxed shape of ferrite crystals and M/A islands in GF is observed.

3.2. CCT curves Based on the dilatometry data, the CCT diagrams corresponding to the different austenite conditions were constructed. These diagrams are presented in Fig. 7. It has been observed that the transformation data obtained after the application of schedules B and C, both leading to unrecrystallized austenite, can be represented in

M. Olasolo et al. / Materials Science and Engineering A 528 (2011) 2559–2569

2563

Table 2 Volume fractions of polygonal (PF) and quasi-polygonal ferrite (QF), pearlite (P), bainite (B) and martensite (M) and corresponding hardness measurements (HV10) as a function of the applied schedule and cooling rate (95% con?dence limits are indicated). Schedule A Cooling rate (? C/s) 0.3 1 6 30 100 0.3 1 6 30 100 PF/QF 0.94 ± 0.02 0.89 ± 0.02 0.45 ± 0.05 – – 0.95 ± 0.94 ± 0.79 ± 0.05 ± – 0.02 0.02 0.04 0.03 P 0.06 ± 0.02 0.05 ± 0.02 – – – 0.05 ± 0.02 0.06 ± 0.02 – – – B – 0.06 ± 0.02 0.55 ± 0.05 1 0.52 ± 0.06 – – 0.21 ± 0.04 0.95 ± 0.03 0.63 ± 0.05 M – – – – 0.48 ± 0.06 – – – – 0.37 ± 0.05 HV10 158 193 217 223 276 141 163 214 219 251 ± ± ± ± ± ± ± ± ± ± 3 2 2 3 2 2 2 4 2 2

B

the same diagram (Fig. 7(b)). Vickers hardness measurements are superimposed on the CCT diagrams of Fig. 7. In all cases it was observed that for cooling rates higher than about 50–60 ? C/s, the resulting microstructure is constituted by bainite and martensite, since the amount of martensite was larger when transformation occurred from recrystallized austenite. As a consequence, at similar cooling conditions, a signi?cant increase in hardness is produced in the latter case as is shown in the diagrams. At the highest cooling rate (≈200 ? C/s) the microstructure obtained after the application of schedule A is formed by about 80%M + 20%B, with a hardness of 282 HV, whereas the microstructure is constituted by about 56%M + 44%B when schedule C is applied, the hardness being of 259 HV. In Fig. 8 the CCT diagrams developed for both recrystallized and deformed are plotted together for comparison. In the ?gure, the cooling rates are also indicated. It is observed that the deformation of the prior austenite brought about an expansion of the ferrite transformation ?eld in the CCT diagram. An increase in the

ferrite transformation start temperature is observed (Ar3 ). Similarly, an increase of the bainite transformation start temperature was also observed in the range of high cooling rates, although at intermediate cooling rates (between 3 and 30 ? C/s approximately, where ferrite and bainite are formed) similar or even lower Bs bainite initial transformation temperatures are observed for deformed austenite. While the increase of Ar3 temperature with deformation is well documented, the behavior of Bs turns out to be more complex. It has been reported that, in contrast to what occurred with “reconstructive” type transformations, such as allotriomorphic ferrite, for displacive transformations, like bainite and martensite, plastic deformation retards the decomposition of the austenite, this effect being called “mechanical stabilization” [26]. The mechanism appears to be that the growth of bainite/martensite laths is retarded by the deformation debris in the austenite. However, Bhadeshia [26] pointed out that the transformation of deformed austenite might be accelerated at initial stages, although the over-

Fig. 6. FEG-SEM micrographs of microstructures corresponding to schedule A: (a) and (b) CR = 100 ? C/s, (c) 30 ? C/s and (d) 6 ? C/s.

2564

M. Olasolo et al. / Materials Science and Engineering A 528 (2011) 2559–2569

Similarly to what is observed in the present study, Kazimierz et al. recently reported that bainitic transformation was also accelerated in deformed austenite in the range of high cooling rates [27]. These authors suggested that a higher dislocation density in the deformed austenite leads to a faster nucleation of the bainite laths. In contrast at slower cooling rates, they observed a retardation of the bainitic transformation when compared to the kinetics found in non-deformed austenite. In the present work it is also observed that the amount of bainite formed in the intercritical range is lower when transformation takes place from unrecrystallized austenite. This could indicate a delaying in the bainitic transformation related to a signi?cant increment of the nucleation density of ferrite grains in this region promoted by deformation. 3.3. EBSD analysis Fig. 9 shows examples of the crystal orientations and misorientation grain boundaries as a function of the misorientation angle obtained after the application of schedule B for two different cooling rates: 1 ? C/s (ferrite–pearlite) and 30 ? C/s (nearly fully bainite, only about 5% ferrite), respectively. The increase in the number of low angle boundaries (between 4? and 15? ) for the bainitic microstructure compared to ferrite–pearlite is clearly evident. The EBSD analysis was extended to schedules A and B for a wide range of cooling rates and several microstructural features were quanti?ed. The in?uence of cooling rate on the mean effective grain size is shown in Fig. 10 for cases of recrystallized (schedule A) and deformed (schedule B) austenite microstructures, considering both aforementioned misorientation criteria of 4? and 15? , denoted as d4 and d15 , respectively. As expected, using a 15? threshold misorientation, larger effective grain sizes are quanti?ed under all conditions. On the other hand, it is found that the average grain size decreases while increasing the cooling rate. This is related to an increase in the nucleation rate and a decrease in grain growth while increasing the cooling rate [28–30]. For example, taking the 4? criterion, the d4 effective grain size decreases from 24.8 ± 1.7 to 7.6 ± 0.1 m in the case of recrystallized austenite and from 14.4 ± 0.5 to 6.6 ± 0.1 m in the case of unrecrystallized austenite. Nevertheless, the re?nement is more notorious for lower cooling rates. In the range where mainly ferrite predominates, the evolution of the mean grain size with cooling rate is close to that predicted by Beynon and Sellars [31], which suggested a relationship between the ferrite grain size (d ) and the cooling rate (CR) as follows: d ∝ CR?0.5 . In contrast, as bainite becomes the main phase this dependence decreases. The results of Fig. 10 indicate that the effect of cooling rate in re?ning the microstructure is more intense for the recrystallized microstructure. Consequently, for a given threshold misorientation, it is observed that the difference in effective mean grain size between the microstructures that result from both types of austenite reduces as the cooling rate increases. At a cooling rate of 100 ? C/s, which corresponds to a mixture of bainite and martensite, similar effective grain sizes were measured in both cases. On the other hand, it is also observed that plastic deformation leads to a re?nement of the microstructure (arrows in Fig. 10), but this effect weakens as the cooling rate increases. In fact, at the highest cooling rate, there are hardly any differences between both deformation schedules. Concerning the low cooling rates regime in which ferrite is the predominant phase, the re?nement observed in both 4? and 15? misorientation effective mean grain sizes agrees reasonably with the predictions obtained with the equation proposed by Beynon and Sellars [31] for Nb microalloyed steels assuming an accumulated strain of ε = 0.4, (d = d o (1–0.45ε0.5 ), where d and d o correspond to the ferrite grain sizes obtained from the unrecrystallized and recrystallized austenite microstructures, respectively).

Fig. 7. CCT diagrams for the (a) recrystallized and (b) deformed austenite.

all rate of transformation would be reduced when compared to non-deformed austenite. As a consequence, a smaller quantity of bainite may be formed in the former case, although this behavior seems to depend on the degree of deformation. Lightly deformed austenite transforms more rapidly relative to undeformed austenite because of the increase in the defect density (nucleation sites, higher SV ). However, although the nucleation rate will be greater in heavily deformed austenite, the overall rate of transformation will be reduced (growth is retarded).
1000 900
Dγ recrystallized Dγ unrecrystallized

Temperature (?C)

800 700 600 500 400 300 200 0.1
Ms= 475?C

PF+P PF+B M B+M
Cooling rate (?C/s)

15 30 60 6 3 1 0.3 0.6 0.15

210 100

1

10

100

1000

10000

Time (s)
Fig. 8. Comparison between the CCT diagrams obtained for recrystallized and unrecrystallized austenite.

M. Olasolo et al. / Materials Science and Engineering A 528 (2011) 2559–2569

2565

Fig. 9. Microstructural details obtained by EBSD: (a) inverse pole ?gure; (b) misorientation angle map.

Fig. 11(a) and (b) shows the misorientation angle distributions obtained for different cooling rates in the range from 0.3 to 200 ? C/s after the application of schedules A and B, respectively. It must be pointed out that the contribution of pearlite was eliminated for the analysis of EBSD scans when this constituent was present. The EBSD data clearly indicate that the change from ferrite to bainite alters the proportion of low and high angle grain boundaries. For example, in the case of cooling rates of 0.3 ? C/s for schedules A and B and 1 ? C/s for schedule B, the microstructure is mainly formed by polygonal ferrite with some amount of pearlite. In these three cases the percentage of high angle grain boundaries (HAGBs),

Fig. 10. In?uence of cooling rate on the average cell size for both recrystallized (schedule A) and unrecrystallized (schedule B) austenite using 4? and 15? threshold misorientation criteria.

Fig. 11. Misorientation angle distributions obtained for different cooling rates after transformation from: (a) recrystallized austenite; (b) unrecrystallized austenite.

2566

M. Olasolo et al. / Materials Science and Engineering A 528 (2011) 2559–2569

de?ned as misorientation angles higher than 15? , ranges between 89 and 90. In contrast, once bainite becomes the main constituent in the microstructure the percentage of HAGBs signi?cantly diminish. This is the situation of the cooling rate of 1 ? C/s in schedule A (HAGB percentage of 69) and schedule B at 6 ? C/s (65%). This decreasing tendency in the percentage of HAGBs continues as a higher bainite fraction is present in the microstructure (48% for the cooling rate of 30 ? C/s in schedule A, for example). In the case of diffusion controlled transformations, like polygonal ferrite, the distribution of misorientations is the result of a combination of a random distribution and relative orientations inherited from the Kurdjumow–Sachs (K–S) relationship in → [32–34], whereas a massive transformation like bainite or martensite would generate characteristic peaks in the misorientation distribution following more or less the K–S crystallographic relationships. The main characteristics of the K–S related distributions are the two peaks close to 50? and 60? , the lack of misorientations in the range 30–45? and a relatively high intensity in the range between 0? and 25? [35]. The results obtained with the fully bainitic (schedule A at 30 ? C/s) and bainite–martensite mixed (schedules A and B at 100 ? C/s) microstructures follow this tendency. Note the pronounced 60? peak found in Fig. 11 for the highest cooling rates of 100 ? C/s and 200 ? C/s after transformation from recrystallized austenite, this is probably related to the high volume fraction of martensite formed under these conditions [36]. This signi?cant increase in the high angle grain boundaries in this case may suggest a higher contribution of packet and block boundaries from the martensite [13]. In this context, a more pronounced peak at 60? could be expected after schedule B at 100 ? C/s and 200 ? C/s, as in these speci?c conditions a volume fraction of ?0.37 and 0.56 of martensite was quanti?ed, respectively. This could suggest that the limited number of OIM images considered to obtain the results in Fig. 11, as mentioned in other studies [37], provide a local scale analysis that can be very dependent on the local deviations from the averaged quantity. Independently of that, it can be considered that the tendencies observed with the misorientation angle distributions of Fig. 11 con?rm the quantitative measurements shown in Table 2 where there are some interpretation dif?culties. 3.4. Hardness tests Fig. 12 summarizes the mean hardness values measured on the specimens plotted as a function of either cooling rate (Fig. 12(a)) or austenite-transformation start temperature (Fig. 12(b)). An increase in the cooling rate or a decrease in the transformation start temperature lead to an increase in hardness. No signi?cant in?uence of the austenite condition on the form of these relationships was observed. Three different zones can be distinguished in both Fig. 12(a), for the cooling rate range, and in Fig. 12(b), for the transformation start temperature. At low cooling rates, between approximately 0.15 and 1 ? C/s, the lowest hardness values are observed, between 130 and 190 HV10, respectively. They correspond to the microstructures dominated by ferrite. In this range a relatively fast increment of hardness with increasing cooling rates was observed, although the transformation start temperature seemed to be less sensible to the cooling conditions. This temperature decreased from 794 to 763 ? C under recrystallized conditions, but remained nearly constant around 800 ? C for the unrecrystallized austenite. The increase in hardness observed in this cooling range can be associated with the reduction in the effective grain size, as quanti?ed in Fig. 10, and is probably also associated with the contribution of some V(C, N) precipitation. Conversely, the highest hardness values were displayed by the microstructures composed by bainite and martensite, forming at the lowest transformation start temperatures that correspond to

Fig. 12. Hardness characteristics plotted as a function of: (a) cooling rate and (b) transformation start temperature.

the highest cooling rates. At intermediate cooling rates, between about 1 and 40 ? C/s, the microstructure progressively changes from a mixture of ferrite and bainite to pure bainite with increasing cooling rates. An important decrease in the transformation start temperature was also observed in this range (close to 200 ? C), whereas the hardness did not change signi?cantly (from 190 to 223 HV10), showing a well de?ned hardness plateau. Cizek et al. also reported the existence of a hardness plateau extending over a wide range of cooling rates for transformed microstructures dominated by granular bainite in low and ultra-low carbon microalloyed steels [38]. As transformation takes place at lower temperatures, it can be expected that the contribution of V(C, N) precipitation will decrease [39], while the opposite will happen with the dislocation density. For a given cooling rate, the transformation microstructures formed from the recrystallized parent austenite were generally observed to display increased hardness values compared to those created from the deformed austenitic matrix. However, the opposite behavior is observed when the comparison between hardness values is made at similar microstructures rather than similar cooling rates. This is better illustrated in Fig. 12(b), where comparable microstructures may be roughly approximated by similar transformation initial temperatures, and hardness values for samples produced from deformed austenite are slightly higher than those from recrystallized austenite. 3.5. Microstructural heterogeneity The results shown in Figs. 10 and 11 de?ne the evolution of the mean values of several microstructural features, but they do not provide information concerning the possible in?uence of accumulated strain and cooling rate on microstructural heterogeneities

M. Olasolo et al. / Materials Science and Engineering A 528 (2011) 2559–2569

2567

Fig. 13. Grain size (15? ) distributions plotted in terms of accumulated area fraction: (a) schedule A; (b) schedule B.

that may affect the ductile–brittle behavior. In order to take this into account, the 15? effective grain size distributions obtained at different cooling rates have been represented in Fig. 13. In the ?gures, the grain area fractions have been considered in order to better identify the presence of coarse grains. Previously published studies suggest that to quantify the ductile–brittle transition, instead of the mean grain size, a parameter capable of properly catching the relevance of the coarse grain fraction is required [40,41]. Following that, in this study the grain size, for which about 0.20 of the area fraction of grains that have a size greater than that value, has been selected (denoted as d20 ). In the case of schedule A, the main characteristic is that for all the cooling conditions, the 15? crystallographic grain distributions show a wide range of sizes, with the coarser one corresponding to the lower cooling rate of 0.3 ? C/s (see Fig. 13(a)). In the ?gure it can be observed that there are no signi?cant differences between the tails of the distributions. In the case of schedule B, two different behaviors can be distinguished (Fig. 13(b)). At 30 and 100 ? C/s the effective grain size distributions are very similar to those found after schedule A, whereas a signi?cant re?nement in the coarse grains is observed at lower cooling rates (0.3–6 ? C/s) as well as at the highest cooling rate (200 ? C/s). In order to better differentiate these behaviors, the corresponding d20 values have been drawn in Fig. 14 as a function of the cooling rate (in the following, denoted as d20–15? ). In conditions corresponding to schedule A, Fig. 14(a) shows that there is a slight re?nement on d20–15? when the cooling rate increases from 0.15 and 0.3 ? C/s to 1 ? C/s, whereas this parameter remains nearly constant for higher cooling rates, with a slight increase observed for 100 ? C/s. Following a similar procedure, in the same ?gure the evolution of d20 corresponding to 4? crystallographic misorientation grain size distributions has been drawn (d20–4? ). In this case, as the cooling rate increases, d20–4? decreases continuously. Comparing

Fig. 14. Change of d20 value, de?ned as the grain size above which a 20% of grain area fraction is present in the distribution, as a function of the cooling rate. The crystallographic grains with misorientations of 4? and 15? have been considered: (a) schedule A; (b) schedule B.

d20–4? and d20–15? values, it turns out that there are small differences between both measurements for the microstructures obtained in the range of low cooling rates, but for CR > 1 ? C/s the differences become more and more notorious. While the d20–4? parameter follows the same tendency shown in Fig. 10 for the mean grain size values, i.e., re?nement while increasing the cooling rate, the behavior of d20–15? deviates signi?cantly for CR > 1 ? C/s. This indicates that as the cooling rate increases, the coarsest austenite grains transform into closely aligned low misorientation units with decreasing size (d20–4? ) that keep the dimensions of the high angle effective grains practically unaffected (d20–15? ). Similarly, the results of d20–4? and d20–15? corresponding to schedule B are drawn in Fig. 14(b). In the range of low cooling rates, smaller than 1 ? C/s, both parameters slightly differ, although deviations between them start becoming evident at 6 ? C/s, enlarging signi?cantly as the cooling rate increases. While a decreasing tendency is observed in d20–4? measurements, the evolution of the d20–15? parameter is completely different, increasing at 30 and 100 ? C/s, reaching levels close to those obtained with schedule A at similar cooling conditions, with a further decrease at the highest

2568

M. Olasolo et al. / Materials Science and Engineering A 528 (2011) 2559–2569

cooling rate of 200 ? C/s, leading to values similar to those measured at low cooling rates. The heterogeneity in transformed microstructures is partially inherited from the heterogeneity present in prior austenite structures. Therefore, the coarsest effective grains are probably the transformation products of the coarsest austenite grains (Fig. 4). At low and intermediate cooling rates (0.15–6 ? C/s), where the resulting microstructures are mainly constituted by ferrite, the accumulation of strain (schedule B), even at low levels, provides a higher density of nucleation sites and results in a re?ned and more homogeneous microstructure compared to the nondeformed material (schedule A), following a tendency similar to that observed for the mean effective grain size (Fig. 10). Under these cooling conditions, nucleation density is high enough to achieve a re?nement in the entire grain size distribution. However, in the case of higher cooling rates (30–100 ? C/s), where the microstructure is constituted mainly by bainite, the increase in the cooling rate or the accumulation of deformation in austenite, at least at the present low strain values, seems not to have any effect on the re?nement of the tail of the grain size distribution given by HAGBs (d20–15? ). In the case of high temperature bainite transformations, Furuhara et al. observed that a small undercooling promotes variant selection leading to coarse bainite packet sizes [42,43]. As undercooling increases, more bainite variants form at the austenite grain boundary leading to a re?nement of the packet size. Similarly, in the case of the bainite phase transformation in the heat-affected zone (coarse austenite grains), Lambert-Perlade et al. [44] identi?ed coarse upper bainite crystallographic packets constituted of several low crystallographic misorientation blocks. In this study, the CCT curves show that the bainite transformation start temperature, Bs , at the cooling conditions of 30 and 100 ? C/s, is relatively high (≈600 ? C). In consequence, it can be argued that the coarse austenite grains follow a behavior similar to that described by the aforementioned authors during transformation. Increasing the cooling rate to 200 ? C/s, bainite transformation start temperatures decrease significantly (see Fig. 7), which also increases the amount of martensite formed in comparison with the conditions described above. The higher undercooling might explain why, under these conditions, a greater re?nement is obtained. The in?uence of deformed austenite in microstructural heterogeneities at high cooling rates must be taken into account. The fact that deformed austenite expands the ferrite transformation regime in the CCT curves leads to some improvement at intermediate cooling rates. This is clearly evident at 6 ? C/s, where the higher amount of polygonal ferrite formed after schedule B (see Fig. 5) reduces signi?cantly the value of d20–15? when compared to that obtained from schedule A. This bene?cial effect completely disappears when bainite becomes the main phase (between 30 and 100 ? C/s). In contrast to what happens with ferrite, the results suggest that, for coarse austenite grains, the bainite packet re?nement is not enhanced by the small accumulated strain. Conversely, Kawata et al. observed that in ausformed upper bainite there was a packet re?nement as a consequence of bainite nucleation within austenite grains in addition to grain boundaries [45]. In the present case, the limited accumulated strain probably has not been suf?cient to promote additional intragranular bainite nucleation on austenite defects. The behavior of d20–15? for CR > 1 ? C/s described in Fig. 14 deviates from the tendency observed in Fig. 10 with mean size values, where an increase in the cooling rate or accumulating strain in austenite favor effective grain re?nement in the bainite regime. As there are also results in previous studies that con?rm this re?nement in microstructures with normal austenite grain sizes [46,47], and simultaneously, the results of this study with coarse grains follow the tendency described by Furuhara et al. [42] and LambertPerlade et al. [44], also under conditions corresponding to initial

coarse austenite grains, it seems that the austenite grain size can play an important role in the nucleation of bainite variants. Further work is required in this ?eld.

3.6. Industrial implications The achievement of a deformed austenite prior to transformation and the application of accelerated cooling rates are considered procedures to re?ne the ?nal microstructure and, in consequence, to improve both strength and toughness. This general rule is ful?lled if the transformation occurs on a rather homogeneous austenite microstructure. In order to accomplish that, it has been suggested to avoid ?nishing rolling temperatures in the region where both recrystallized and unrecrystallized austenite grains coexist, normally known as the interval between the recrystallization limit temperature (RLT, the lowest temperature above which recrystallization between passes is complete) and the recrystallization stop temperature (RST, the highest temperature at which recrystallization is completely absent) [48]. The results obtained in this study con?rm that, in addition to the aforementioned requirement, it is necessary to have a homogeneous and ?ne austenite microstructure before starting to accumulate strain, that is, before RLT temperature. In some industrial conditions this requirement is not always achieved. For example, in thin slab direct rolling of microalloyed steels, in those cases where high strength requirements (high level of microalloying) combined with low total reductions (gage thicknesses higher than 10 mm), it is possible to identify isolated coarse effective grains in the ?nal microstructure constituted by low temperature transformation products. An example is shown in [49]. Similarly, in conventional thermomechanical rolling, it is possible to have inappropriate austenite re?nement before strain starts accumulating, in some cases as a consequence of low soaking temperatures [50], or in others due to microalloying segregation [17]. In the aforementioned cases, besides the de?nition of proper rolling schedules to minimize the tail of coarse austenite grains before transformation, the above results indicate that for the present deformation conditions, where the level of retained strain is limited, better strength–toughness property combinations can be achieved for ferrite–bainite microstructures obtained at intermediate cooling rates. As shown in Fig. 12, the hardness levels are similar to those found at higher cooling rates, when fully bainitic structures form, but the presence of coarse grains that can impair toughness properties is signi?cantly reduced (Fig. 14).

4. Conclusions This study analyses, in a low carbon Nb–V microalloyed steel, the characteristics of phase transformations during continuous cooling in both recrystallized and deformed austenite with wide grain size distributions. The main conclusions are as follows: - The accumulation of strain in the austenite prior to transformation accelerates both the ferritic and bainitic transformations. In the case of bainite transformation this effect is greater in the range of high cooling rates. - Low (4? ) and high (15? ) crystallographic misorientation effective grain sizes, characterized by mean values and distributions, have been measured. Independently on the transformation products, both effective mean grain sizes diminish by increasing the cooling rate. This re?ning is more notorious in the range of low cooling rates corresponding mainly to ferrite–pearlite microstructures. Similarly, the re?nement is more intensive for the recrystallized austenite microstructure.

M. Olasolo et al. / Materials Science and Engineering A 528 (2011) 2559–2569

2569

- In the ferrite–pearlite regime, both low and high misorientation effective grain size distributions are re?ned as cooling rate increases or deformation is accumulated in the austenite. In contrast, when the microstructure is mainly bainitic different behaviors are observed. While an increase in the cooling rate or the accumulation of deformation in the austenite re?nes the low angle misorientation grain size distribution, there is no re?nement of the length of the grain size distribution tail given by high angle grain boundaries. These results suggest that the austenite grain size may have an important role in the nucleation of bainite variants, although further study is required in this ?eld. - For the present deformation conditions better strength–toughness property combinations can be achieved for ferrite–bainite microstructures obtained at intermediate cooling rates. Hardness levels are similar to those found at higher cooling rates, when fully bainitic structures form, but the presence of coarse grains deleterious for toughness properties is signi?cantly reduced. Acknowledgements Partial ?nancial support of this work by CICYT (MAT200909250) is acknowledged. References
[1] G.I. Garcia, Proc. Int. Conf. Microalloying 95, ISS, Warrendale, PA, 1995, pp. 365–375. [2] B.A. Graville, Proc. Int. Conf. Welding of HSLA Structure Steels, ASM, Metals Park, OH, 1978, pp. 85–101. [3] I. Tamura, H. Sekine, T. Tanaka, C. Ouchi, Thermomechanical Processing of HSLA Steels, Butterworth and Company, London, 1988. [4] D.N. Hanlon, J. Sietsma, S. van der Zwaag, ISIJ Int. 41 (2001) 1028–1036. [5] R. Bengochea, B. López, I. Gutiérrez, Mater. Sci. Forum 284–286 (1998) 201–208. [6] C. Calvo, I.-H. Jung, A.M. Elwazri, D. Bai, S. Yue, Mater. Sci. Eng. A 520 (2009) 90–96. [7] Y. van Leeuwen, J. Sietsma, Mater. Sci. Forum 539–543 (2007) 4572–4577. [8] J.L. Lanzagorta, D. Jorge-Badiola, I. Gutierrez, Mater. Sci. Eng. A 527 (2010) 934–940. [9] B.P. Wynne, P. Cizek, C.H.J. Davies, B.C. Muddle, P.D. Hodgson, in: T. Chandra, T. Sakai (Eds.), Proc. Int. Conf. Thermec 97, vol. 1, TMS, Warrendale, PA, 1997, pp. 837–843. [10] C.J. McMahon, M. Cohen, Acta Metall. 13 (1965) 591–604. [11] M.A. Linaza, J.M. Rodriguez-Ibabe, J.J. Urcola, Fatigue Fract. Eng. Mater. Struct. 20 (1997) 619–632. [12] J.H. Chen, G. Li, R. Cao, X.Y. Fang, Mater. Sci. Eng. A 527 (2010) 5044–5054.

[13] H. Kitahara, R. Ueji, N. Tsuji, Y. Minamino, Acta Mater. 54 (2006) 1279–1288. [14] W. Wang, W. Yan, L. Zhu, P. Hu, Y. Shan, K. Yang, Mater. Design. 30 (2009) 3436–3443. [15] P. Uranga, A.I. Fernández, B. López, J.M. Rodriguez-Ibabe, 43rd MWSP Conf., ISI, vol. XXXIX, 2001, pp. 511–529. [16] Q. Sha, Z. Sun, Mater. Sci. Eng. A 523 (2009) 77–84. [17] D. Chakrabarti, M. Strangwood, C. Davis, Metall. Mater. Trans. 40A (2009) 780–795. [18] Abaqus Reference Manuals, 2009, Providence, RI, USA. [19] R. Petrov, L. Kestens, Y. Houbaert, Mater. Charact. 53 (2004) 51–61. [20] ASTM E1382, Annual Book of ASTM Standards, vol. 03.01, ASTM International, 2006. [21] B. Pereda, J.M. Rodriguez-Ibabe, B. López, ISIJ Int. 48 (2008) 1457–1466. [22] R.T. De Hoff, F.N. Rhines, Quantitative Metallography, McGraw-Hill Book Co., New York, 1968. [23] P. Uranga, B. López, J.M. Rodriguez-Ibabe, Steel Res. Int. 78 (2007) 199–209. [24] G. Krauss, S.W. Thompson, ISIJ Int. 35 (1995) 937–945. [25] Atlas for Bainitic Microstructures, vol. 1, ISIJ, Tokyo, 1992. [26] H.K.D.H. Bhadeshia, Bainite in Steels, IOM Communications, 2001, pp. 208–213. [27] A. Kazimierz, J. Lis, Mater. Sci. Forum 539–543 (2007) 4620–4625. [28] M. Umemoto, Z.H. Guo, I. Tamura, Mater. Sci. Technol. 3 (1987) 249–255. [29] S. Zajac, T. Siwecki, B. Hutchinson, M. Attlegard, Metall. Trans. 22A (1991) 2681–2694. [30] R.W.K. Honeycombe, Steels Microstructures Properties, Edward Arnold Pub. Ltd., London, 1980. [31] J.H. Beynon, C.M. Sellars, High Strength Low Alloy Steels, ASM, Wollongong, 1984, p. 184. [32] G. Kurdjumow, G. Sachs, Z. Phys. 64 (1930) 325. [33] R.L. Ryder, W. Pitsch, Acta Metall. 14 (1966) 1437–1448. [34] E. Novillo, E. Cotrina, A. Iza-Mendia, B. López, I. Gutiérrez, Mater. Sci. Forum 500–501 (2005) 355–362. [35] A.F. Gourges, H.M. Flower, T.C. Lindley, Mater. Sci. Technol. 16 (2000) 26–40. [36] S. Zajac, V. Schwinn, K.H. Tacke, Mater. Sci. Forum 500–501 (2005) 387–394. [37] R. Petrov, L. Kesstens, A. Wasilkowska, Y. Houbaert, Mater. Sci. Eng. A 447 (2007) 285–297. [38] P. Cizek, B.P. Wynne, C.H.J. Davies, B.C. Muddle, P.D. Hodgson, Metall. Mater. Trans. 33A (2002) 1331–1349. [39] T.N. Baker, Mater. Sci. Technol. 25 (2009) 1083–1107. [40] A. From, R. Sandstr?m, Mater. Charact. 42 (1999) 111–122. [41] T. Hanamura, F. Yin, K. Nagai, ISIJ Int. 44 (2004) 610–617. [42] T. Furuhara, H. Kawata, S. Morito, T. Maki, Mater. Sci. Eng. A 431 (2006) 228–236. [43] T. Furuhara, N. Takayama, G. Miyamoto, Mater. Sci. Forum 638–642 (2010) 3044–3049. [44] A. Lambert-Perlade, A.F. Gourgues, A. Pineau, Acta Mater. 52 (2004) 2337–2348. [45] H. Kawata, K. Sakamoto, T. Moritani, S. Morito, T. Furuhara, T. Maki, Mater. Sci. Eng. A 438–440 (2006) 140–144. [46] K. Fujiwara, S. Okaguchi, H. Ohtani, ISIJ Int. 35 (1995) 1006–1012. [47] K. Fujiwara, S. Okaguchi, Mater. Sci. Forum 284–286 (1998) 271–278. [48] B. López, J.M. Rodriguez-Ibabe, New Developments on Metallurgy and Applications of High Strength Steels, TMS, 2008, pp. 843–856. [49] P. Uranga, B. López, J.M. Rodriguez-Ibabe, Ironmak. Steelmak. 36 (2009) 162–169. [50] H. Meuser, F. Grimpe, S. Meimeth, C.J. Heckmann, C. Tr?ger, Mater. Sci. Forum 500–501 (2005) 565–572.


相关文章:
更多相关标签: