ISIJ International
Online ISSN : 1347-5460
Print ISSN : 0915-1559
ISSN-L : 0915-1559
Regular Article
Influence of Carbon Content on Toughening in Ultrafine Elongated Grain Structure Steels
Yuuji KimuraTadanobu Inoue
Author information
JOURNALS OPEN ACCESS FULL-TEXT HTML

2015 Volume 55 Issue 5 Pages 1135-1144

Details
Abstract

(0.2–0.6)%C-2%Si-1%Cr-1%Mo steels were quenched and tempered at 773 K and deformed by multi-pass caliber rolling (i.e. warm tempforming) with a rolling reduction of 78%, in order to obtain ultrafine elongated grain (UFEG) structures. The tensile and Charpy impact properties of the warm tempformed (TF) steels were investigated to determine the influence of the carbon content on toughening in the UFEG structures. The TF samples consisted of UFEG structures with strong <110>//rolling direction (RD) fiber textures. The transverse grain size and aspect ratio in the UFEG structure tended to reduce as the carbon content increased, whilst the carbide particle size became slightly larger. The increase in the carbon content resulted in an increase in the yield strength from 1.68 to 1.95 GPa at room temperature; however, it was accompanied by a loss of tensile ductility. In contrast to quenched and tempered samples exhibiting ductile-to-brittle transitions, the TF samples exhibited inverse temperature dependences of the impact toughness. This was due to delaminations, where cracks were observed to branch in the longitudinal direction (//RD) of the impact test bars. The upper-shelf energy of the TF sample was enhanced as the carbon content decreased, and higher absorbed energy was also achieved as delamination occurred at lower temperatures. The delamination was found to be controlled not only by the transverse grain size, the grain shape, and the <110>//RD fiber texture but also by carbide particle distribution in the UFEG structure.

1. Introduction

Ultrahigh-strength low-alloy steels usually exhibit poor toughness at a yield strength (σy) of 1.4 GPa or over.1) Additionally, the cold formability of ultrahigh-strength low-alloy steels is comparatively poor. Hence, the practical use of ultrahigh-strength low-alloy steels is restricted.

The working of tempered martensite was known as strain tempering, tempforming, and so on.2) Sekiguchi et al.3) developed a thermomechanical treatment, in which a quenched steel was rapidly heated to an elevated temperature and subsequently deformed during the tempering process. This thermomechanical treatment was applicable to secondary operations such as forging, and was called warm-temper-forging.

The present authors4,5,6) have developed a warm tempforming (WTF) process in order to achieve the toughening of ultrahigh-strength low-alloy steel bars. When the WTF was performed on medium-carbon low-alloy steels such as 0.39%C-2%Si-1%Cr-1%Mo steel4,5) and 0.6%C-2%Si-1%Cr steel,6) using multi-pass caliber rolling at 773 K with a rolling reduction of 78%, UFEG structures with strong <110>//RD fiber textures were formed. The steels with the UFEG structures demonstrated an inverse temperature dependence of toughness; the Charpy absorbed energy (vE) increased with decreasing test temperature in the subzero temperature range, in which ultrahigh-strength low-alloy steels usually undergo a ductile-to-brittle transition (DBT).

The inverse temperature dependence of toughness occurred as a consequence of delaminations, in which cracks branched and propagated in the longitudinal direction of impact bars.4,5,6) It was clear that the transverse grain size, the grain shape and the <110>//RD fiber texture in the UFEG structure were the dominant microstructural factors in controlling the delamination.4,5,6,7) Furthermore, the carbon content may influence the microstructural features of the UFEG structures and the occurrence of delamination for the warm tempformed steels; however systematic investigation on this subject has not yet been conducted.

In this study, the WTF using multi-pass caliber rolling at 773 K with a rolling reduction of 78% was applied to (0.2–0.6)%C-2%Si-1%Cr-1%Mo steels, and the microstructures, the tensile and Charpy impact properties were investigated in relation to the carbon content. Subsequently, the influence of the carbon content on the inverse temperature dependence of toughness was discussed for the UFEG structure steels.

2. Experimental Procedure

2.1. Materials and Warm Tempforming

100 kg ingots were prepared by vacuum melting and casting, and homogenized at 1473 K, and then hot-rolled into plates with a thickness of 4 cm, followed by air cooling. Table 1 lists the chemical compositions of the steels used. Blocks with a dimension of 12 × 4 × 4 cm were cut out of the plates and solution-treated at 1473 K for 1 h, and then hot-rolled using a caliber roll, into squared bars with a cross-sectional area of 9 cm2, followed by quenching. Note that the (0.2–0.43)% C steel bars were quenched in water, while the 0.6% C steel bars were quenched in oil to prevent quench cracking. Subsequently, the quenched bars were tempered at 773 K for 1 h and then subjected to multi-pass caliber rolling with a rolling reduction of 78%, into squared bars with a cross sectional area of 2 cm2, followed by air cooling (the TF sample). To prepare the sample without tempforming (r = 0%), some of the quenched bars were tempered at 773 K for 1 h and then air cooled (the QT sample). The principal axes of the squared bar in this study were defined as illustrated in Fig. 1. The axis which is coincident with the rolling direction is defined as the RD, the axis which is coincident with the direction of the main working force at the final pass is defined as the ND, and the axis which is normal to the RD and ND is defined as the TD. The SD denotes the striking direction of the impact test.

Table 1. Chemical compositions of the steels used (mass%).
SteelsCSiMnCrMoAlPSO
0.2 C0.201.950.211.011.020.0100.0010.0010.0007
0.43 C0.431.970.201.020.960.0210.0020.0010.0007
0.6 C0.601.950.211.001.000.0120.0010.0010.0014
Fig. 1.

Schematic drawing of a pair of rolls used for caliber rolling and the relationship between the position of a rolled bar and an impact bar.

2.2. Microstructural Characterization and Mechanical Testing

Electron back-scattering diffraction pattern (EBSP) analysis was performed using a scanning electron microscope equipped with a field emission gun (FE-SEM). The average intercept length (ILAV) was measured for the high angle grain boundaries (HABs) that had a misorientation angle of 15° or more. The length of the grid lines used in the ILAV measurement was 25 μm and the grid line spacing was 1.6 μm. The maximum intercept length (ILMAX) for the HABs was estimated by the statistics of extremes theory for the grid lines that were used in the ILAV measurement. The average dislocation density (ρ) was estimated by an X-ray diffraction (XRD) line profile analysis, using the modified Warren-Averbach method.8) The analysis conditions for the FE-SEM/EBSP technique and ρ estimation in the XRD are described in detail elsewhere.5,9) The integrated intensities of the (110) XRD peaks (Im) were measured on the RD planes for the TF samples, and the ratio of Im to the integrated intensity of the (110) XRD peak (Is), for a standard sample was measured to evaluate the development of the <110>//RD fiber texture. The carbide particle size was measured by transmission electron microscopy (TEM). The carbide particles were also extracted in carbon replicas, and were identified by TEM observations, coupled with energy dispersive X-ray spectrometry (EDS) and convergent beam electron diffraction (CBED).

Tensile tests were conducted for the JIS Z 2201-14A specimens (specimens had a diameter of 6 mm, machined in the RD (JIS-14A specimens)), at a crosshead speed of 0.85 mm/min. Small plate specimens with a parallel length of 4 mm, a width of 3 mm, and a thickness of 1 mm were machined from the squared bars at three different inclinations (0, 45, and 90°) to the RD. They were subsequently used for tensile testing at a crosshead speed of 0.11 mm/min to investigate the tensile anisotropy. The σy of 0.2% offset was reported. Charpy impact tests were performed for 2 mm V-notch specimens that were machined in the RD. The SD of the impact tests had an angle of ~45° to the TD and ND (Fig. 1).

3. Experimental Results

3.1. Microstructures

Figure 2 shows the microstructures for the TF samples. Following the WTF, UFEG structures with strong <110>//RD fiber textures were formed. The UFEG structures consisted of ribbon-like and rod-like grains that were aligned to the RD, although very fine equiaxed grains were partially formed. Spheroidized carbide particles were also dispersed and their long axes were observed to be almost aligned to the RD in many of the grain boundary carbide particles.

Fig. 2.

Grain boundary maps on the RD planes (┴RD) and ND planes (┴ND), inverse pole figures for the RD, and TEM bright-field images for the TF samples. The black and red lines represent a high angle boundary (HAB) with a misorientation angle of θ ≥ 15° and a low angle boundary with a misorientation angle of 2 ≤ θ < 15°, respectively.

Figure 3 summarizes the microstructural parameters for the TF samples as a function of carbon content. Firstly, the integrated (110) XRD peak intensity ratio (Im/Is) has a value within the range of 7.5–8.5, indicating that the influence of the carbon content on the development of the <110>//RD fiber texture is small. Secondly, the Karnel average misorentation (KAM) value, measured by EBSP analysis, reflects the density of the geometrically necessary dislocations (GNDs)10) associated with plastic strain gradients.11) The average KAM value is within the range of 0.6–0.7°, irrespective of the carbon content. This suggests that the TF samples may have almost the same density of the GNDs. Furthermore, the ρ, estimated by the XRD, is thought to correspond to the average density of the mobile dislocations and the statistically stored dislocations (SSDs).12) The ρ of the TF samples were estimated to be 3.6 × 1014 m–2 at 0.2% C and 2.9 × 1014 m–2 at 0.6% C, respectively. The influence of the carbon content on the ρ was also considered to be small. Thirdly, the transverse ILAV for the TF samples were measured to be 0.38 μm at 0.2% C, 0.28 μm at 0.43% C, and 0.27 μm at 0.6% C, respectively. Furthermore, the elongation ratios of the longitudinal ILAV to the transverse ILAV for the TF samples were 3.1 at 0.2% C, 2.5 at 0.43% C, and 2.3 at 0.6% C, respectively. Hence, in the same WTF condition, the transverse grain size and aspect ratio in the UFEG structure decrease as the carbon content increases, and then tend to level off at a carbon content of 0.43% or more.

Fig. 3.

Average intercept lengths for the HAB (ILAV), average KAM value for a 1st neighbor rank (KAM-1st ), and integrated intensity ratio (Im/Is) for the (110) plane as a function of carbon content. Closed and open symbols for the KAM-1st denote the data in the transverse planes (the RD planes) and longitudinal planes (the ND and SD planes), respectively.

Figure 4 shows Gumbel probability plots13) of the maximum intercept lengths (ILmax) in the grid lines (L0 = 25 μm). The return period (T = L/L0) and reduced variate (y = lnT) were calculated to be 400 and 6, respectively, when the prospective length (L) was 10 mm. The estimated maximum intercept lengths along the transverse direction (ILMAX_//SD) in the L = 10 mm, are 9 μm at 0.2% C, 7 μm at 0.43% C and 6 μm at 0.6% C, respectively. The estimated maximum intercept lengths along the longitudinal direction (ILMAX_//RD) are also estimated to be 26 μm at 0.2% C, 17 μm at 0.43% C, and 15 μm at 0.6% C. Thus, the ILMAX in addition to the ILAV for the TF sample tends to decrease as the carbon content increases. It also should be noted that the ILMAX is 20–30 times as long as the ILAV. This indicates that the UFEG structure has a wide grain size distribution.

Fig. 4.

Maximum intercept lengths (ILmax) in the TF samples plotted on a Gumbel probability paper: (a) transverse ILmax (┴RD) and (b) longitudinal ILmax (//RD).

Figure 5 shows the distributions of the carbide particles in the TF samples. Note that the average long-axis lengths (dAV) of the carbide particles are shown in this figure. The TF samples have bimodal carbide particle size distributions; the transgranular carbide particles are finer than the intergranular carbide particles. Such bimodal carbide particle distributions might be inherited from those in the tempered martensitic structure.5) The dAV of the transgranular and intergranular carbide particles were measured to be 15 nm and 37 nm at 0.2% C, 16 nm and 42 nm at 0.43% C, and 22 nm and 46 nm at 0.6% C, respectively; they became slightly larger as the carbon content increased. The respective average aspect ratios of the transgranular and intergranular carbide particles were approximately 1.3 and 1.6 in all the TF samples; the transgranular carbides were more spherical than the intergranular carbides. Figure 6 displays a TEM bright-field image and the CBED patterns of the extracted replicas for the 0.43% C sample. The CBED patterns identify the transgranular and intergranular carbides as cementite with an orthorhombic crystal structure. The EDS analysis also revealed that substitutional elements such as Cr, Mn and Mo dissolved into the cementite particles. Incidentally, it is possible for Mo carbides (such as Mo2C) to precipitate inside the grains, as a result of the tempering of the Mo bearing steels at 773 K or above; however, Mo carbides were not observed in the present TF samples. In addition, owing to the heterogeneous size distributions of the carbide particles, the volume fraction of the carbide particles could not be experimentally estimated. When all the carbon exists as cementite (with a density of 7.68 Mg/m3), the volume fraction of the carbide particles can be calculated to be 0.03 at 0.2% C, and 0.09 at 0.6% C; thus, the volume fraction of the carbide particles is proportional to the carbon content.

Fig. 5.

Histograms demonstrating the distributions of the long-axis lengths (d) of the carbide particles in the TF samples. The dAv indicates the average value of the long-axis of the carbide particles.

Fig. 6.

TEM bright-field image and diffraction patterns of extracted replicas showing cementite particles in a TF sample.

During the WTF, carbon may have two major effects on the evolution of the UFEG structure. The first effect is to refine the matrix grain structure. We have already observed that the UFEG structure evolved through the extension of blocks, packets and prior-austenite grains in the RD.5) Thus, the formation of the UFEG structure is largely influenced by variations in the crystallographic orientation and the geometric arrangement of blocks and packets. When the tempered martensitic structures have a high volume fraction of blocks with long-axes that are aligned to the RD and/or finer blocks, the UFEG structures are suggested to evolve at lower rolling reductions. Furthermore, it is recognized that blocks and packets become finer as the carbon content increases.14) In the QT samples, the ILAV for blocks5) were measured to be 0.98 μm at 0.2% C, 0.52 μm at 0.43% C, and 0.40 μm at 0.6% C, respectively. Therefore, an increase in the carbon content leads to the refinement of blocks, and this contributes to the formation of a finer grain structure under the same WTF condition. The second effect is the grain boundary pinning effect of carbide particles. The nanometer-size carbide particles that precipitated during the tempering might retard the recovery of martensitic structures, and thus the fine block structures were retained in the QT samples. Although the carbide particles grew slightly and were oriented in the RD during the WTF process, they might also play important roles in accumulating GN dislocations and retarding the migration of LABs and HABs through their pinning effect.5) As a result, the UFEG structures were evolved. The grain boundary pinning of the second phase particles is generally governed by the Zener equation.15) This can be expressed by D = βz × ds/f where D is the average grain size, βz is a constant, ds is the average second particle size, and f is the volume fraction of the second particles. Generally, the Zener equation cannot be applied to the UFEG structure, but ds/f can provide an indication of the grain boundary pinning effect of second particles. It is expected that the UFEG structure that is obtained through the WTF, will become finer as the value for f increases and the value for ds decreases. At the same initial carbide particle size, the inter-particle spacing of carbides becomes finer as the carbon content increases; however, this may cause an increase in the rate of Ostwald ripening of the carbide particles.16) Accordingly, an increase in the carbon content may lead to an increase in the grain boundary pinning effect of carbide particles and this effect is expected to level off. Therefore, it can be concluded that the evolution of the UFEG structure could be controlled by carbon content under the same WTF conditions. For example, the optimum carbon content for achieving effective grain refinement (using a minimized volume fraction of carbide particles) is determined to be 0.43% C, for the present WTF conditions.

3.2. Tensile Properties

Figure 7 shows the nominal stress-strain curves. All the QT samples exhibited continuous yielding behavior. By contrast, an obvious yield drop during discontinuous yielding was observed in the TF sample with a carbon content of 0.2%. The yield drop during the discontinuous yielding, however, tends to decrease with increasing carbon content, and is not pronounced in the TF sample with the carbon content of 0.6%. Such a yield drop was often observed in ultrafine grained materials, and it was usually accompanied by plastic instability.17,18,19,20) However, the TF samples exhibit yield drops followed by adequate uniform elongations. In these TF samples, the tensile strength (σB) was defined as the maximum tensile stress that was achieved following the yield drop.

Fig. 7.

Nominal stress-strain curves. The r indicates the rolling reduction.

Figure 8 summarizes the room temperature tensile properties as a function of the carbon content. The σy and σB increase with increasing carbon content in both the samples; the QT sample, however, shows a stronger carbon dependence of strength than the TF sample. The respective average values of the σy and σB for the TF samples were 1.68 and 1.63 GPa at 0.2% C, 1.87 and 1.87 GPa at 0.43% C, and 1.95 and 2.01 GPa at 0.6% C. Thus, the values tend to level off at a carbon content of 0.43% or above. Additionally, the σy is higher than the σB in samples with a low carbon content, up to 0.43% C. At a carbon content above this level, the relationship between the σy and σB is reversed because the carbon dependence of the σB is more significant than that of the σy. When compared to the QT samples, the TF samples demonstrate a higher σy at every carbon content. The difference in the σy is greater in samples with the lower carbon content range. There is no significant difference in the σB between the TF and QT samples at 0.6% C. At lower carbon contents, the σB is higher in the TF samples than in the QT samples. Thus, the effect of the WTF on the strength, especially the σy, becomes more significant with decreasing carbon content, in the carbon content range of 0.2 to 0.6%. Furthermore, the respective average values of the uniform elongation (εu), total elongation (EL), and reduction of area (RA) for the TF samples are 8.3%, 17.9%, and 56.7% at 0.2% C, whilst they are 6.0%, 11.2%, and 31.7% at 0.6% C; they are linearly reduced with increasing carbon content. At the same carbon content, the TF sample exhibits higher tensile ductility than the QT sample.

Fig. 8.

Yield strength (σy), tensile strength (σB), uniform elongation (εu), total elongation (EL), and reduction of area (RA) at room temperature, as a function of carbon content.

Figure 9 shows the σy, EL, and RA as a function of the carbon content and testing temperature for the TF sample. In all the TF samples, the σy increases linearly as the testing temperature decreases from room temperature to 123 K, below which it tends to increase abruptly. The increment in the σy between room temperature to 77 K is 0.4–0.5 GPa. Furthermore, the RA tends to decrease abruptly below a temperature of 123 K. The drop in the RA becomes more pronounced with increasing carbon content; the average value of the RA for the 0.2% C sample was 40.2% at 77 K and the value decreased to 4.4% for the 0.6% C sample. The EL also decreases in response to the decrease in local elongation, rather than the εu. The above result indicates that the ductile-to-brittle transition temperature (DBTT) for the TF sample is lowered with decreasing carbon content, when the sample is tensioned along the RD.

Fig. 9.

Yield strength (σy), total elongation (EL), and reduction of area (RA) as a function of carbon content and testing temperature in the TF samples.

3.3. Charpy Impact Properties

Figure 10 displays the representative fracture appearances of the TF samples. Figure 11 summarizes the Charpy impact properties. In Fig. 11, 1) the data for the TF sample that showed an almost complete ductile fracture (no delamination) at an elevated temperature is denoted by ○, 2) the data for the TF sample that exhibited a delamination where the macroscopic crack branching angle (β)5) was below 15° is denoted by ◇, and 3) the data for the TF sample where the β was 15° or over is denoted by △. In contrast to the QT samples, which exhibited DBTs above room temperature, all the TF samples exhibited the inverse temperature dependence of toughness; the vE markedly increased with decreasing temperature. This increase in the vE corresponds to the occurrence of delamination.

Fig. 10.

Fracture appearances of Charpy V-notched specimens after testing. The macroscopic crack branching angle (β) was defined as the angle between the crack path and longitudinal direction (//RD) of the impact bar. Arrows indicate the specimens that did not separate into two pieces during the impact test.

Fig. 11.

Charpy V-notch absorbed energy (vE) as a function of testing temperature. Data points with + indicate that the specimens did not separate into two pieces during the impact test.

According to the macroscopic fracture appearances and the variation in the vE, the Charpy impact behavior for the TF samples can be divided into three regions. At temperatures in the range of 500–573 K (Region 1), the average vE (Upper Shelf Energy, vEUS) for the TF samples exhibiting almost complete ductile fractures are 169 J at 0.2% C, 130 J at 0.43% C, and 59 J at 0.6% C, respectively. Thus, the vEUS, in addition to the RA, decreases with increasing carbon content. Additionally, the TF samples demonstrate much higher vEUS values than the QT samples, at the same carbon content, even though the σy of the TF samples is greater than that of the QT samples. At a temperature of approximately 500 K, delaminations start to occur, irrespective of the carbon content. In region 2, the delamination is pronounced. When the temperature is defined as the delamination finish temperature (TDF) (below this temperature, the vE for the TF samples decrease to the level of its vEUS value, or less), the respective TDF were measured to be 213 K at 0.2% C, 233 K at 0.43% C, and 253 K at 0.6% C. Thus, the TDF decreases with decreasing carbon content. As the carbon content decreased, it was observed that fewer impact bars broke into two pieces. Below the TDF (region 3), the β was observed to increase significantly with decreasing temperature. This resulted in the decrease of the vE. At 77 K, the β increased to 30° or greater, irrespective of the carbon content, and the vE dropped to approximately 10 J. Therefore, it can be concluded that in the samples with a carbon content of 0.2 to 0.6%; as the carbon content reduces, the vEUS for the TF sample increases. This increases the occurrence of delamination at lower temperatures, leading to a higher vE.

4. Discussion

4.1. Occurrence Mechanism of Inverse Temperature Dependence of Toughness

Our findings on the inverse temperature dependence of toughness for the UFEG structure with a <110>//RD fiber texture is summarized as follows.4,5,6,7,21) Brittle fracture occurs when the maximum tensile stress (σt) at the notch root or crack tip exceeds the brittle fracture stress (σBF). The σt is proportional to the σy. In bcc metals such as steel, the σy demonstrates a marked increase at lower temperatures, leading to an obvious DBT. In an equiaxed grain structure, grain refinement simultaneously increases the σBF and σy; however, the effect is much greater on the σBF than on the σy. This eventually leads to a lower DBTT.22) Furthermore, the increase in the σy for the UFEG structure with a strong <110>//RD fiber texture also becomes more pronounced at lower temperatures, as indicated in Fig. 9. However, the <110>//RD fiber texture provides many {100} cleavage planes on the longitudinal planes that are parallel to the RD and on the planes that are inclined at an angle of 45° to the RD (45° planes). The coherence length on the {100} cleavage planes corresponds to the effective grain size (Deff) for cleavage fracture.22) The Deff is larger along the RD than in the 45° directions, as a result of the elongated grain shape. If the cleavage fracture stress (σc) is proportional to the inverse square root of the Deff,22) the cleavage fracture stress along the 45° direction (σc//45°) is therefore expected to be larger than the cleavage fracture stress along the SD (σc//SD). The magnitude relation of σc//SD < σc//45° was confirmed by the static three-point bending test of the UFEG structure samples.7) Even when we consider an intergranular fracture in the UFEG structure, a similar magnitude relationship may be established for the intergranular fracture stress; the intergranular fracture stress in the SD is low, as a result of the elongated grain shape and the good grain boundary continuity along the RD.9,23) If the Tresca yield criterion is assumed in the process zone within the plastically yielding region of the notch root, the σt//RD and σt//SD in the process zone can be approximated as 2.6 × σy and 1.6 × σy,7) respectively. The condition of the σt//SD < σt//45° < σt//RD eventually holds. Figure 12 shows the schematic diagram illustrating the relationship between the σt and σc; namely, the Yoffee diagrams for the UFEG structure.5,6) The relationship between the vE and temperature is also schematically illustrated in this figure. Furthermore, a {100} pole figure is presented for the 0.6% C steel. From this pole figure, it is confirmed that many {100} cleavage planes exist on the SD planes that are normal to the SD and on the 45° planes that are inclined at an angle of 45° to the RD. We can consider that there are three separate relationships between the σt and σc in the UFEG structure; however, the cleavage crack will hardly propagate in the SD despite the highest σt//RD in the UFEG structure, since the {100} cleavage planes scarcely exist on the RD planes that are normal to the RD. Below T1, the σt//SD exceeds the σc//SD and thereby cleavage cracks initiate and propagate along the RD, leading to the occurrence of delamination. With decreasing temperature, the σt rises and the delamination becomes more pronounced. As crack arrester type delamination occurs, the notch and/or crack tip becomes blunt. Therefore, the stress state in the vicinity of the notch and/or crack tip changes from a highly triaxial tension state into an uniaxial one; the impact specimen virtually behaves as an un-notched specimen. Additionally, it can be expected that the stress fielding effect from the formation of microcracks along the RD will reduce the driving force for fracture21) As a result, the vE is enhanced. Furthermore, in addition to the delamination crack along the RD, the cleavage crack initiation and propagation along the 45° direction is facilitated below T2, in which the σt//45° exceeds the σc//45°. The occurrence of delamination is subsequently suppressed, leading to a reduction of the vE. Figure 13 presents examples of the fracture surfaces of a 0.6% C sample. Brittle cracks were observed to propagate along the RD in the TF sample that underwent a marked delamination. Furthermore, at a temperature of 77 K, where the vE markedly dropped, brittle cracks were observed to run along the planes that were inclined at approximately 45° to the RD (as indicated by the arrows) and the delamination crack propagation along the RD was suppressed. Such fracture morphologies were commonly observed in all the TF samples with the UFEG structures. Apparently the β in Figs. 10 and 11 might be related to the ratio of the length of the transverse cracks to that of the delamination cracks along the RD; as the transverse crack propagation along the 45° direction became significant below T2, the β might increase and get close to 45° at lower temperatures. From these observations, we can conclude that the occurrence of the delamination responsible for the inverse temperature dependence of toughness can be controlled by the transverse grain size and grain shape in the UFEG structure with a strong <110>//RD fiber texture, if the carbon content is the same. For example, in Fig. 12, in order to enhance the delamination toughening at lower temperatures, it is considered necessary to lower the T2 against the T1, by raising the σc//45° against the σc//SD. For this purpose, the reduction in the transverse grain size is found to be effective. The coherence length on the {100} cleavage planes along the 45° direction becomes shorter with decreasing the transverse grain size. Since the σy also increases as the average transverse grain size decreases,5,6) the transverse grain size is a key factor in controlling the strength and toughness of a steel with an UFEG structure. One of the effective methods to decrease the transverse grain size is to increase the rolling reduction; however, we also should consider that grain subdivision might be expected to become more significant at higher rolling reductions,5) leading to the evolution of ultrafine equiaxed grain structure. In this case, the anisotropy of mechanical properties for the TF sample will become smaller, and thereby the occurrence of delamination will be suppressed.

Fig. 12.

Yoffee diagram for UFEG structure with a strong {110} fiber deformation texture.5,6) Cleavage fractures on the longitudinal {100} planes (//RD) cause delamination toughening (curve i), while those on the {100} planes with the transverse components diminish the delamination toughening (curve ii). {100} pole figure is also shown for the 0.6 C sample.

Fig. 13.

SEM fractographs showing the fracture surfaces for the TF samples; impact tested at (a) 293 K and (b) 77 K. Arrows show cleavage cracks on the planes with transverse components.

4.2. Influence of Carbon Content on Inverse Temperature Dependence of Toughness

We verify the Yoffee diagram for the UFEG structure in Fig. 12, in relation to the influence of the carbon content. The average σy, true fracture stress (σF) and RA are plotted as a function of testing temperature and tensile direction in Fig. 14. Here, the σF was calculated by dividing the load at fracture by the minimum cross-sectional area of the ruptured specimen. As for the σy, a magnitude relation of σy//SD < σy//45° < σy//RD is observed for each TF sample. In all tensile directions, the σy increases with decreasing testing temperature, and its increment between room temperature and 77 K is 0.4–0.5 GPa, regardless of the carbon content and tensile direction. This increment in the σy also concurs with that in Fig. 9. Similarly, a magnitude relation of σF//SD < σF//45° < σF//RD is observed for each TF sample; however, the temperature dependence of the σF is variable, depending on the tensile direction. When the TF samples were tensioned along the RD, all samples exhibited significant necking prior to failure, leading to high RA//RD values. The σF//RD markedly increases with decreasing temperature. Furthermore, the TF samples ruptured without significant necking when the samples were tensioned along the SD. The increment in the σF//SD was much smaller than that in the σF//RD; the variation in the σF//SD values is also small for each TF sample. The σF//45° also tends to increase with decreasing temperature, but the increment was smaller in the σF//45° than the σF//RD; the σF//45° are plotted between the σF//RD and σF//SD. Here, it should be noted that most of the TF samples underwent brittle failure at a temperature of 77 K, in the tensile direction along the SD, and the σy//SD became almost equal to the σF//SD, regardless of the carbon content. Hence, in this case, it appears that T = 77 K may correspond to T1 in the Yoffee diagram (Fig. 12). This implies that there is little difference in the delamination starting temperature for the TF samples. This coincides with the results of the Charpy impact test in Fig. 11.

Fig. 14.

Yield strength (σy), true fracture stress (σF), and reduction of area (RA) as a function of testing temperature and the angle (α) between the tensile direction and RD for small plate specimens: ◆ 0°, ○ 45° and ▲ 90°.

If we assumed that the σc//SD was given by the average of the σF//SD for the TF samples that failed without necking (the RA of 1% or less), the σc//SD were calculated to be 1.84 GPa at 0.2% C, 2.04 GPa at 0.43% C, and 2.07 GPa at 0.6% C, respectively. Subsequently, a linear relationship between the σc//SD and the inverse square root of the ILMAX_//RD in Fig. 4, i.e. σc//SD = 1.09 + 3.84 × (ILMAX_//RD)–0.5 was confirmed. Here, we assumed that the ILMAX might reflect the longest {100} coherence length of UFEGs which might have minimum resistance to cleavage fracture (i.e. the lower limit for the σc). Although further investigations are required in order to determine the Deff for cleavage fracture in the UFEG structure, the quasi-cleavage facet size in the delamination (e.g. see Fig. 13) appears to be close to the ILMAX_//RD, rather than the ILAV_//RD in Fig. 3. Hence, it may be interpreted that the longitudinal grain size increased with decreasing carbon content and this resulted in a reduction of the σc//SD. Simultaneously, the σy decreased with decreasing carbon content (Fig. 8). As a result, there was little difference in the delamination starting temperature amongst the TF samples in Fig. 11.

Furthermore, we used a linear extrapolation based on the relationship between the σc//SD and the inverse square root of the ILMAX_//RD, in order to estimate the lower limit for the σc//45°. The lower limit for the σc//45° was estimated to be 2.2 GPa at 0.2% C, 2.3 GPa at 0.43% C, and 2.4 GPa at 0.6% C, by substituting the ILMAX_//45° (= √2 × ILMAX_//SD) for the ILMAX_//RD in the above equation. For each TF sample, the estimated value of the lower limit for the σc//45° is almost equal with the lower limit for the σF//45° in Fig. 14. The difference between the estimated lower limit for the σc//45° and average σy//45° tends to decrease with increasing carbon content; they were 0.18 GPa at 0.2% C, 0.10 GPa at 0.43% C, and 0.12 GPa at 0.6% C, respectively. Hence, on the basis of the Yoffee diagram in Fig. 12, the onset of brittle fracture along the 45° direction can be expected to be shifted to a lower temperature from the T1 (= 77 K), as the carbon content decreases.

However, the difference between the above estimated values for the σc//45° and σc//SD is small for each TF sample, and accordingly the difference between the T1 and T2 in Fig. 12 is small. Slight differences between the σc//45° and σc//SD were also demonstrated from the static three-point bending test of the 0.39% C steel.7) When the three-point bending test was performed for the specimens that were machined from the rolled bar with 0° (//RD), 45° (//45°), and 90° (//SD) rotation along the RD, the fracture energy (JF) of JF//RD, JF//45° and JF//SD were measured to be 5184, 142, and 44 kJ/m2, respectively. The difference between the JF//45° and JF//SD was thus small. Moreover, the JF//45° is not sufficiently large when compared to that of the QT sample (= 129 kJ/m2). Therefore, for delamination to occur in steels with the UFEG structure, it is considered necessary that the σc//45° is greater than the σc//SD. However, 1) the reduction in the T2 with decreasing carbon content in Fig. 11 and 2) the occurrence of delamination over a wide temperature range from 500 K to the subzero temperatures could not be adequately explained in terms of the magnitude relation between the σc//45° and σc//SD.

If we focus on the fact that the JF//RD of the TF sample is much larger than that of the QT sample, higher toughness and ductility for the RD plane (┴RD) in the UFEG structure can be considered a sufficient condition to enhance the delamination toughening at lower temperatures. Here, the TF samples were observed to be broken through stepwise crack propagation which consists of terraces (//RD) and steps (//SD) in the delamination fracture. In this case, the primary fracture mode was quasi-cleavage on the terraces, while it was ductile tearing on the steps. Once the notch and/or crack tip becomes blunt through the delamination, crack re-initiation is necessary to fracture a material. This occurs under conditions of nearly uniaxial tension, which is an unfavorable cleavage. When the re-initiation of ductile tearing along the SD is suppressed, this may suppress the onset of the brittle fracture along the 45° direction, in addition to the delamination cracking along the RD. This is because, as the re-initiation of ductile tearing is suppressed, it becomes difficult to generate a tri-axial state of stress at the crack tip again. Hence, it is reasonable to consider that higher toughness and ductility in the RD plane is a sufficient condition to enhance the delamination. The variations in the RA in Figs. 9 and 14 also determine that the reduction in the carbon content results in higher toughness and ductility in the RD plane at lower temperature. Therefore, as the carbon content decreases in the TF sample, there is an increase in the occurrence of delamination at lower temperatures, generating a higher vE.

As shown in Fig. 3, there are no significant differences in the development of the <110>//RD fiber texture, dislocation density, and transverse grain size amongst the TF samples. This quantitatively indicates that the toughness and ductility in the RD plane may depend on the distribution of the carbide particles. When a material fails by ductile fracture as a result of the growth and coalescence of voids nucleated at the interface between the hard carbide particles and the soft ferrite matrix, the toughness and ductility can be enhanced by reducing the volume fraction and size of the carbide particles which act as void nucleation sites. The reduction in the carbon content therefore resulted in the reduction in the volume fraction and size of the carbide particles in the TF sample, resulting in the high toughness and ductility.

When compared to the 0.6%C-2%Si-1%Cr steel,6,9) the 0.2%C-2%Si-1%Cr-1%Mo steel exhibited higher σy//RD, even though the average transverse grain size was comparable between these TF samples. This could be owed to the fact that the Mo bearing steel exhibits a denser distribution of fine carbide particles and higher average KAM values than the 0.6%C-2%Si-1%Cr steel. It was confirmed that the strengthening mechanism of the 0.6%C-2%Si-1%Cr steel that was processed by the WTF, was mainly as a result of grain refinement strengthening.6) Hence, it is suggested that dislocation strengthening and dispersion strengthening by carbides, in addition to grain refinement strengthening, contribute to the σy//RD of the Mo bearing steel. Since the GN dislocation density in the UFEG structures was estimated to be almost comparable, irrespective of the carbon content, the tendency for the yield drop to decrease with increasing carbon content (Fig. 7) appeared to be related to the increase in the dispersion strengthening effect, as a result of the carbide particles. As for the tension test along the transverse direction of the UFEG structures (//SD), the TF samples were observed to exhibit continuous yielding behavior, similar to that of the QT samples. Additionally, the UFEG structure is considered to have a relatively large grain size in the transverse tensile direction. Hence, it was suggested that the σy//SD in the UFEG structure depended on carbide particle dispersion strengthening and/or dislocation strengthening, rather than grain refinement strengthening.5) Therefore, the σy of the UFEG structure can be influenced by the carbide particle distribution in all tensile directions. However, the contribution of the dispersion strengthening by carbide particles to the σy is different, depending on the tensile direction. The inter-particle spacing (i.e. mean free path (MFP)) of the carbide particles decreases as the particle size decreases and as the volume fraction of carbide particles increases. The decrease in the MFP leads to an increase in the σy, when the microstructural parameters (except for the carbide particle distribution) are comparable in the UFEG structure. From the average carbide particle size in Fig. 5, it was suggested that the MFP of the carbide particles decreased as the carbon content increased from 0.2 to 0.43%, above which it tended to level off.

As described above, the distribution state of carbide particles can influence the strength, ductility, and toughness of the TF sample, and thereby the occurrence of delamination is varied. Therefore, it was clarified that the inverse temperature dependence of toughness in the UFEG structure with a strong <110>//RD fiber texture can be controlled not only by the transverse grain size and grain shape but also by carbide particle distribution.

Figure 15 summarizes the vE at room temperature as a function of the yield strength × total elongation (σy·EL) balance. Data for the conventional steels24,25) and ultrafine grained steels21) are also shown. The σy·EL balance is improved with decreasing carbon content in both of the QT and TF samples. However, the σy·EL balance is much greater in the TF samples than in the QT samples. The QT samples exhibited low vE, while the TF samples exhibited remarkably high vE by controlling the delamination. As the carbon content decreases, the vE of the TF samples increases. However, it must be emphasized that even with a high carbon content of 0.6%, the TF sample had an average σy of 1.95 GPa and vE of 159 J and exhibited an excellent combination of strength, ductility and toughness, comparable to those of JIS carbon and low-alloy steels.24)

Fig. 15.

Relationship between the yield strength × total elongation (σy·EL) balance and vE. Data for ultrafine-grained steels,21) JIS carbon and low-alloy steels,24) and 300 M steels25) are also presented for reference.

5. Conclusions

A WTF using multi-pass caliber rolling with a rolling reduction of 78% at 773 K was applied to (0.2–0.6)%C-2%Si-1%Cr-1%Mo steels. The microstructures, tensile and Charpy impact properties of the TF samples were investigated in relation to the carbon content. The results obtained are as follows;

(1) UFEG structures strong <110>//RD fiber textures were formed through the WTF. The transverse grain size and aspect ratio of the UFEGs tended to decrease as the carbon content increased from 0.2 to 0.43%, above which the influence of the carbon content on these microstructural factors levelled off. The volume fraction of carbide particles increased as the carbon content increased and the carbide particle size slightly increased.

(2) The TF samples exhibited a room temperature σy of 1.68–1.95 GPa. The σy increased as the carbon content increased from 0.2 to 0.43%, above which it tended to level off. The improved combination of σy and ductility was obtained in the sample with a lower carbon content.

(3) As the carbon content decreased, the vEUS of the TF samples increased and the inverse temperature dependence of toughness, resulting from delamination, became significant at lower temperatures, leading to higher vE.

(4) The inverse temperature dependence of toughness was found to be controlled by the distribution of carbide particles, in addition to the transverse grain size and grain shape in the UFEG structure with a strong <110>//RD fiber texture. The toughness and ductility for the RD planes that were normal to the RD should be improved to enhance the delamination. Reducing the volume fraction and size of carbide particles was found to be an effective for this.

Acknowledgement

The authors thank Mr. Kuroda and Mr. Taniuchi for materials processing with caliber-rolling and Ms. Hirota for her help with the microstructural observation. We also gratefully acknowledge Dr. Nie for her quantitative XRD analysis. The study was supported by the Japan Science and Technology Agency (JST) under Collaborative Research Based on Industrial Demand “Heterogeneous Structure Control: Towards Innovative Development of Metallic Structural Materials”.

References
 
© 2015 by The Iron and Steel Institute of Japan
feedback
Top