Metallurgical and Materials Transactions A

, Volume 44, Issue 1, pp 560–576

Influence of Warm Tempforming on Microstructure and Mechanical Properties in an Ultrahigh-Strength Medium-Carbon Low-Alloy Steel

Authors

    • National Institute for Materials Science
  • Tadanobu Inoue
    • National Institute for Materials Science
Article

DOI: 10.1007/s11661-012-1391-2

Cite this article as:
Kimura, Y. & Inoue, T. Metall and Mat Trans A (2013) 44: 560. doi:10.1007/s11661-012-1391-2
  • 592 Views

Abstract

A 0.4 pct C-2 pct Si-1 pct Cr-1 pct Mo steel was quenched and tempered at 773 K (500 °C) and deformed by multi-pass caliber rolling (i.e., warm tempforming). The microstructures and the mechanical properties of the warm tempformed steels were investigated as a function of the rolling reduction. At rolling reductions of more than 28 pct, not only extension of the martensite blocks and/or the packets in the rolling direction (RD) but also a grain subdivision became more significant, and an ultrafine elongated grain (UFEG) structure with a strong 〈110〉//RD fiber deformation texture was formed after 78 pct rolling. The tensile deformation behavior became significantly anisotropic in response to the evolution of UFEG structure. The longitudinal yield strength (σy) of the quenched and tempered sample increased from 1480 to 1860 MPa through the 78 pct rolling, while the transverse σy leveled off at around 1600 MPa up to 28 pct rolling. The transverse true fracture stress was also markedly degraded in contrast to the longitudinal one. Charpy impact properties were enhanced at a rolling reduction of 52 pct or more. The 52 pct-rolled sample underwent a ductile-to-brittle transition in the temperature range from 333 K to 213 K (60 °C to −60 °C), while the 78 pct-rolled sample showed an inverse temperature dependence of the impact toughness because of brittle delamination. The tensile and Charpy impact properties are discussed in association with the microstructural evolution.

1 Introduction

Stronger and tougher structural materials are always needed to reduce weight and improve safety in transportation, enhance architectural flexibility in construction, and improve performance in heavy industry. Low-alloy steels are workhorse materials in most industries due to their low cost and high performance. However, ultrahigh-strength low-alloy steels with yield strength (σy) exceeding 1400 MPa typically exhibit low impact toughness; their V-notch Charpy absorbed energy (vE) is 10 to 40 J at room temperature,[1,2] which limits their structural applications.

There are two important aspects to improve the notch toughness of structural materials: (1) increase of the intrinsic fracture resistance in solid materials, and (2) relaxation of stress concentration at crack tips.[3] The well-known techniques for raising the intrinsic fracture resistance of steels are (1) the reduction of impurity elements such as P[4] and S,[5] and inclusions causing embrittlement,[68] (2) the reduction of C,[9] (3) the addition of alloying elements such as Ni,[9,10] and (4) grain refinement.[1012] A much better combination of σy and vE was obtained in a high-purity 18 wt pct Ni maraging steel strengthened by precipitating nanometer-size intermetallic compounds on a matrix of tough lath martensite; however the vE decreased to 60 J at the σy of 1700 MPa.[13] In the latter aspect to improve the notch toughness of structural materials, delaminations are known to relax the triaxial tension generated by the localized plastic constrain at the notch and/or the crack tip ahead of advancing crack tips. In certain Al-Li alloys[14,15] and thermomechanical treated steels,[1620] the notch toughness at low temperatures was reported to be improved by controlling the occurrence of delamination cracks (i.e., delamination toughening). For these techniques, grain refinement and delamination toughening are considered to be effective in lowering the ductile-to-brittle transition temperature (DBTT) in low-alloy steels.[19,20]

We recently discovered that delaminations in ultrafine elongated grain (UFEG) structures resulted in the remarkable enhancement of the impact toughness of ultrahigh-strength low-alloy steel bars.[2123] In a 0.4 pct C-2 pct Si-1 pct Cr-1 pct Mo steel with a UFEG structure,[22] the σy of 1840 MPa and the vE of 226 J were obtained at room temperature. In addition, the developed UFEG structure steel showed a significant inverse temperature dependence of the vE in response to the delamination at the temperature range from 333 K to 213 K (60 °C to −60 °C), where ultrahigh-strength steels typically undergo ductile-to-brittle transition (DBT).[1,2] A static three-point bending test also demonstrated that the static fracture toughness of the developed UFEG structure steel was about 40 times higher than that of the tempered martensitic steel.[24]

The delaminations occur due to the microstructural anisotropy of the UFEG structures formed through multi-pass caliber rolling of tempered martensitic structures at 773 K (500 °C) with a total reduction in area of about 80 pct.[2125] The dominating factors controlling the delamination toughening are considered to be the ultrafine grain size, the grain shape and the 〈110〉//RD fiber deformation texture,[2125] which vary depending on the starting microstructure[25] and the caliber-rolling conditions such as the deformation temperature[23] and rolling reduction. So far, we have shown that the delamination toughening became more pronounced for a 0.6 pct C-2 pct Si-1 pct Cr steel with decreasing the caliber-rolling temperature from 973 K to 773 K (700 °C to 500 °C) at a rolling reduction of 80 pct.[23]

The deformation of tempered martensitic structures is referred to as tempforming.[2123] The aim of the present study is to investigate the influence of warm tempforming on the microstructure and the tensile and the Charpy impact properties as a function of the caliber-rolling reduction. A 0.4 pct C-2 pct Si-1 pct Cr-1 pct Mo steel was quenched and tempered at 773 K (500 °C) and then subsequently tempformed using multi-pass caliber rolling up to a rolling reduction in area of 78 pct. The microstructural factors controlling the tensile and impact properties were also discussed in association with the evolution of the UFEG structure.

2 Experimental

2.1 Material and Thermomechanical Treatment

A steel with a chemical composition of 0.39 C, 1.98 Si, 1.04 Cr, 1.05 Mo, 0.21 Mn, <0.001 P, <0.001 S, 0.041 Al, 0.002 N, <0.005 O and the balance Fe (all in mass pct) was used in this study. A 100 kg ingot was prepared by vacuum melting and casting, homogenized at 1473 K (1200 °C), and then hot-rolled to a plate with a thickness of 4 cm. A 12 × 4 × 4 cm block was cut out of the plate, solution-treated at 1473 K (1200 °C) for 60 minutes to reduce undissolved carbides, and then hot-rolled using a caliber roll[26] to a squared bar with a cross-section area of 9 cm2, followed by water quenching to obtain the martensitic structure. The squared bars were tempered at 773 K (500 °C) for 60 minutes and then subjected to multi-pass caliber rolling to squared bars with cross sections of 7.9, 6.4, 4.3 or 2.0 cm2 followed by air cooling. The rolling reduction in area (r) by multi-pass rolling was 12 pct in 2 passes, 28 pct in 3 passes, 52 pct in 5 passes and 78 pct in 9 passes, which corresponds to the equivalent strain (ε = 2/√3ln (1/(1 − r/100)) of 0.15, 0.40, 0.85, and 1.75, respectively. The samples were held for 5 minutes in a furnace after every three passes during rolling and passed through twice for the final groove to control the cross-sectional shape of the bars. The surface temperature of the present tempformed bar was measured to rise to around 873 K (600 °C) after every three passes during the tempforming up to a rolling reduction of 78 pct. To prepare the sample without tempforming (r = 0 pct), parts of the quenched bars were tempered at 773 K (500 °C) for 60 minutes and then air cooled (the QT sample).

The principal axes of the squared bar in this study are defined as follows. The axis that is coincident with the rolling direction (RD) is defined as the RD, that which is coincident with the direction of the main working force at the final pass is defined as the ND, and that which is normal to the RD and the ND is defined as the TD.

2.2 Microstructural Characterization and Mechanical Testing

The microstructures of the cross sections of the squared bars were observed by optical microscopy, scanning electron microscopy (SEM) and transmission electron microscopy (TEM). Electron back-scattering diffraction pattern (EBSP) analysis was performed using the SEM equipped with a field emission gun (FE-SEM, JEOL JSM-6500F, JEOL, Tokyo, Japan). The EBSP measurements were conducted on a hexagonal grid with a step size of 50 or 200 nm, and TexSEM Laboratories, Inc. (TSL, Utah) software was used to analyze the data. All orientation data were subjected to a clean-up procedure that comprised confidence index standardization and grain dilation single iteration. The pole figure was calculated using harmonic series expansion with series rank 16, and orthotropic sample symmetry was enforced. The internal strain state was evaluated by analyzing the local orientation gradients within a given region by calculating the kernel average misorientation (KAM).[27,28] Values above a predefined threshold (here, 2 deg) were excluded from the KAM calculation. The average dislocation density (ρ) was estimated by an X-ray diffraction (XRD) line profile analysis using the modified Warren-Averbach method.[29]
$$ { \ln }A(L) \cong { \ln }A^{\text{s}} (L) - \rho BL^{2} { \ln }(\text{Re} /L)(K^{2} C) + O(K^{4} C^{2} ) $$
(1)
where A(L) is the real part of the Fourier coefficients, As is the size Fourier coefficient, B = πb2/2 indicates the magnitude of the Burgers vector of dislocations, Re is the effective outer cut-off radius, and O stands for higher-order terms in K2C. L is the Fourier length: L = na3, where a3 = λ/2(sinθ2 − sinθ1), n is an integer starting from zero and (θ2 − θ1) is the angular range of the measured XRD line profile. The details of the analysis procedure were described elsewhere.[30,31] Since the tempformed samples had 〈110〉//RD fiber deformation textures, the XRD line broadening analysis was performed on the electrically polished surfaces of the plates that were machined from the squared bar at an angle of 45 deg to the RD. The XRD line profiles of the (110), (200), (211) and (220) planes were measured using a Rigaku RINT-2200 X-ray diffractometer (RINT-2200, Rigaku, Tokyo, Japan) with a rotating Co target. To determine the carbide particle size from the TEM images, the sizes of 300 to 500 particles were measured in two to three fields of view for each sample.
The tensile tests were conducted at room temperature at a crosshead speed of 0.85 mm/min for round specimens with a gage length of 30 mm and a diameter of 6 mm machined in the RD (JIS-14A specimens). 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 deg) to the RD and then used for tensile testing at room temperature at a crosshead speed of 0.11 mm/min to investigate the tensile anisotropy. A yield strength of 0.2 pct offset was reported. True fracture stress (σf) was calculated by
$$ \sigma_{\text{f}} = P_{\text{f}} /A_{\text{f}} $$
(2)
where Pf is the load at fracture and Af is the minimum cross-sectional area of the ruptured specimen. For the small plate specimen with an original rectangular cross section, the Af was measured using SEM. Charpy impact tests[32] were carried out at a temperature range from 77 K to 500 K (−196 °C to 227 °C) for full-size 2 mm V-notch specimens that were machined in the RD. The striking direction (SD) of the impact tests had an angle of ~45 deg to the TD and the ND.

3 Experimental Results

3.1 Microstructures

Figure 1 shows the microstructures of the QT sample (r = 0 pct). The sample before the tempforming has a typical tempered martensitic structure, in which equiaxed prior-austenite grains are divided into packets that are subdivided into blocks of martensite laths. The average prior-austenite grain size was measured to be approximately 50 μm by optical microstructural observations. When the grain boundaries with a misorientation angle (θ) of 10 deg or over were defined as the high-angle boundaries (HABs), the average intercept length (AIL) for the HABs was measured to be 0.49 μm. Since variants with the Kurdjumov-Sachs (K-S) relationship[33] have a minimum misorientation angle of 10.53 deg among the six possible block boundaries with the same habit plane in the martensitic structure,[34] the AIL for the HABs mainly reflects the size of the thin martensite blocks. The average number fraction of the HABs was calculated to be as high as 0.82 in the EBSP analysis. Besides, the TEM observation reveals the precipitation of nanometer-size carbide particles in the highly dislocated matrix. Such microstructural features of the tempered martensitic structure are advantageous for the evolution of the UFEG structure through subsequent tempforming treatment at a lower rolling reduction.
https://static-content.springer.com/image/art%3A10.1007%2Fs11661-012-1391-2/MediaObjects/11661_2012_1391_Fig1_HTML.jpg
Fig. 1

IQ map (a) and TEM image (b) in a sample that was quenched and tempered at 773 K (500 °C) (the QT sample, rolling reduction of 0 pct). The black and the red lines represent a HAB with a misorientation angle of θ ≥ 10 deg and a LAB with a misorientation angle of 2 ≤ θ < 10 deg, respectively in the IQ map. The white arrows in (a) and (b) show a triple junction of prior-austenite grain boundaries and carbide particles, respectively

Figure 2 presents the longitudinal-section inverse pole figure (IPF) maps for the SD and the pole figures of the samples that were tempformed by multi-pass caliber rolling in the different rolling reductions. The scanning step size of 200 nm was used. The tempered martensitic structure with the equiaxed prior-austenite grains remains after 28 pct rolling (Figure 2(a)). As the rolling reduction increases, the prior-austenite grains are elongated in the RD, and the prior-austenite grain boundaries become difficult to recognize after 78 pct rolling. If the prior-austenite grains with an average grain size of about 50 μm were simply deformed in the RD, the average transverse grain size should be reduced to a few 10 μm by 78 pct rolling. The pole figures also indicate the development of strong 〈110〉//RD fiber deformation textures in the 52 and 78 pct-rolled samples. The texture intensity for the 78 pct-rolled sample is more pronounced than that for the 52 pct-rolled sample. Bourell[16] reported that the degree of {100}〈110〉 texture became stronger with the rolling strain in multi-pass warm plate rolling of low-carbon steels and was independent of the rolling temperature between 813 K and 923 K (540 °C and 650 °C). Moreover, it must be emphasized here that the warm deformation behavior of the martensite blocks and/or the packets strongly depends on their variations in the crystallographic orientation and their geometric arrangement (Figure 2(b)), resulting in the evolution of a heterogeneous UFEG structure (Figure 2(c)). For example, block-shaped grains appear in the 52 pct-rolled sample, as indicated by a white arrow. Such block-shaped grains are suggested to be formed through the deformation of martensite blocks whose long axes are aligned at high angles to the RD. The kinetic of the microstructure evolution from the block-shaped grains into the UFEGs appears to be slow, and, as a result, relatively large, elongated grains remain even after 78 pct rolling. Figure 3 shows transverse- and longitudinal-section image quality (IQ) maps in the 52 and 78 pct-rolled samples. The scanning step size was 50 nm. The horizontal axis of images (b) and (d) is parallel to the RD. The 52 pct-rolled sample is characterized by rectangular plate-like grains that might be inherited from the martensite blocks, and their long axes are not fully parallel to the RD in the many parts. On the other hand, the matrix grain structure of the 78 pct-rolled sample shows a UFEG structure consisting of ribbon-like and rod-like grains that are aligned to the RD, although very fine equiaxed grains are partially formed. Such ribbon-like and rod-like grains with strong 〈110〉//RD fiber deformation textures are similar to those of heavily cold-drawn and/or swaged steel wires.[35,36] Langford and Cohen[35] reported that the dislocation cells in the cold-drawn iron wire started with equiaxed shapes, became a ribbon-like cross section at a strain of about 1.5, and then tended toward an equiaxed cross section again at higher strain. Furthermore, Ueji et al.[37] reported that there were three kinds of typical microstructures when a martensite was cold-worked with a plate rolling mill; (1) the lamellar dislocation cell (LDC) structure mainly composed of the lamellar boundaries elongated to the RD, (2) the irregularly bent lamella (IBL) structure, and (3) the kinked lath (KL) structure, in which the martensite lath is kinked by shear bands. The area fraction of the LDC increased with increasing the rolling reduction, and almost all the area consisted of the LDC structure after 70 pct (ε = 1.5) cold rolling.
https://static-content.springer.com/image/art%3A10.1007%2Fs11661-012-1391-2/MediaObjects/11661_2012_1391_Fig2_HTML.gif
Fig. 2

IPF maps for the SD and pole figures in the samples that were quenched and tempered at 773 K (500 °C) and then caliber-rolled, i.e., tempformed at rolling reductions of 28 pct (a), 52 pct (b), and 78 pct (c). The white arrow shows a block-shaped grain

https://static-content.springer.com/image/art%3A10.1007%2Fs11661-012-1391-2/MediaObjects/11661_2012_1391_Fig3_HTML.jpg
Fig. 3

IQ maps in the samples that were tempformed at rolling reductions of 52 pct (a, b) and 78 pct (c, d). The black and the red lines represent a HAB with a misorientation angle of θ ≥ 10 deg and a LAB with a misorientation angle of 2 ≤ θ < 10 deg, respectively

The changes in the averages of transverse and longitudinal intercept length (AIL) and the grain boundary length per unit area (GBL) are summarized in Figure 4 as a function of the rolling reduction. Changes in the number fractions of the HABs and the average misorientation angle (θAv) are also presented for reference. At low rolling reductions up to 28 pct, low angle boundaries (LABs) with misorientation angle of 2 ≤ θ < 10 deg increase and then level off, while there are no apparent changes in the HABs. With increasing the rolling reduction from 28 to 78 pct, the transverse AIL for the HABs with a misorientation angle of 10 deg or over decreases from 0.44 to 0.26 μm, whereas the longitudinal AIL for the HABs increases from 0.48 to 0.73 μm. The HABs increase gradually and then increase significantly in the rolling reduction range from 52 to 78 pct, mainly owing to refinement in the transverse grain size. Such a significant increase in the HABs indicates that a grain subdivision accompanied by strong texture evolution[38] becomes more pronounced in the final stage of tempforming, leading to the formation of the UFEGs and the equiaxed grains (Figures 3(c) and (d)).
https://static-content.springer.com/image/art%3A10.1007%2Fs11661-012-1391-2/MediaObjects/11661_2012_1391_Fig4_HTML.gif
Fig. 4

Changes in averages of the transverse and longitudinal intercept lengths for HABs with a misorientation angle of θ ≥ 10 deg (AIL), average misorientation angle (θAv), GBL for the LAB and the HAB, and number fraction of HABs as a function of the rolling reduction. Transverse (closed rhombus) and longitudinal (open rhombus) intercept lengths for HABs with a misorientation angle of θ ≥ 15 deg are also shown in the figure. Closed and open symbols for the θAv, the GBL and the number fraction of HABs denote the data in the transverse and the longitudinal planes that are normal and parallel to the RD, respectively

Figure 5 presents the TEM images showing the typical substructures of the tempformed samples. The TEM observation reveals that relatively large carbide particles exist on the LABs and/or the HABs (i.e., intergranular carbide particles), while finer spherical carbide particles are homogeneously dispersed inside the matrix grains (i.e., transgranular carbide particles) in all of the rolled samples (Figure 5(a) through (d)). Such bimodal carbide particle distributions might be inherited from the tempered martensitic structure. However, the number fraction of the intergranular carbide particles appears to increase with increasing the rolling reduction and the long axes of the carbide particles are nearly aligned to the RD after 78 pct rolling. The histograms (Figure 5(e)) of the carbide particle-size distributions show that the number fraction of larger carbide particles increases and the bimodal distribution becomes more significant through the tempforming treatment. Similar bimodal carbide particle distribution was observed in ultrafine-grained 0.15 pct C-0.3 pct Si-1.4 pct Mn steel, fabricated with a warm caliber rolling[30]; the size of cementite particles on the grain boundaries was 100 to 200 nm, while the size of those within ferrite grain was quite smaller. The average long-axis lengths of the transgranular and the intergranular carbide particles in the QT sample are 20 and 36 nm, respectively, and increased to 24 and 46 nm in the 78 pct-rolled sample (Figure 5(f)). The average aspect ratios of the transgranular and the intergranular particles in Figure 5(f) were respectively measured to be about 1.5 and 1.6 in the QT sample but 1.2 and 1.6 in the 78 pct-rolled sample. The transgranular carbides thus tend to be more spherical during tempforming. In addition, the TEM images of the 78 pct-rolled sample (Figures 5(c) and (d)) indicate a rearranged dislocation structure and become much clearer, as seen in that of the QT sample in Figure 1(b), indicating the decrease in the internal dislocations during rolling.
https://static-content.springer.com/image/art%3A10.1007%2Fs11661-012-1391-2/MediaObjects/11661_2012_1391_Fig5_HTML.gif
Fig. 5

TEM images in the samples that were tempformed at rolling reductions of 28 pct (a), 52 pct (b), and 78 pct (c, d) and histograms (e) showing distributions of the long-axis lengths of carbide particles. Averages of the long-axis lengths and aspect ratio for intergranular and transgranular carbide particles are summarized in (f) as a function of the rolling reduction

3.2 Tensile Properties

Figure 6 shows the nominal stress-strain curves and the tensile properties for JIS-14A specimens as a function of the rolling reduction. The QT sample (r = 0 pct) exhibits typical continuous yielding behavior without a sharp yield point in the stress-strain curve[39] and its average yield strength (σy) and tensile strength (σB) are 1444 and 1730 MPa, respectively. With increasing the rolling reduction, the σy and the σB increase and reach the values of 1844 and 1832 MPa after 78 pct rolling, respectively. The effect of warm tempforming is especially pronounced for the σy, and the 78 pct-rolled samples exhibit a sharp yield-point phenomena followed by an adequate elongation where the upper yield point is the maximum force achieved during the test. Sharp yield-point phenomena are commonly observed in ultrafine grained metals; however there is a significant loss in uniform elongation.[37,4043] When the uniform elongation of the 78 pct-rolled samples is determined at the highest force (defined here as the σB), occurring just prior to necking, the average uniform elongation (εu) is measured to be 6.7 pct; it tends to be higher than that of the QT samples (the εu = 5.0 pct). The average tensile strength-total elongation balance also shows a tendency to increase with the tempforming treatment; it is 23.7 GPa pct for the QT sample, while it is 26.9 GPa pct for the 78 pct-rolled sample. Although the tempforming treatment had no effect on the reduction of area (δ), the true fracture stress (σf) was confirmed to tend to increase with tempforming; the average σf was 2567 MPa for the QT sample, while it was 2686 MPa for the 78 pct-rolled sample.
https://static-content.springer.com/image/art%3A10.1007%2Fs11661-012-1391-2/MediaObjects/11661_2012_1391_Fig6_HTML.gif
Fig. 6

Changes in nominal stress-strain curves (a) and tensile properties (b); yield strength (σy), tensile strength (σB), uniform elongation (εu), total elongation (εt), and reduction of area (δ) with tempforming treatment for JIS-14A tensile specimens

Figure 7 displays the typical nominal stress-displacement curves showing tensile anisotropy for the 78 pct-rolled sample. The longitudinal σy and σB of the small plate specimens agree well with those of the JIS-14A specimens in Figure 6 and the sharp yield-point phenomena followed by the adequate uniform elongation are demonstrated as well. In contrast, the strength and the elongation are reduced with increasing the angle (α) between the tensile direction and the RD. In the transverse tensile direction (α = 90 deg), the sample shows a continuous yielding behavior; the stress-displacement curve was observed to be similar to that seen in the 28 pct-rolled sample or the QT sample, except for its small necking elongation. These trends are summarized in Figure 8 as a function of the rolling reduction and the α. The tensile properties are almost isotropic at low rolling reductions up to 28 pct. The σy, especially, shows a rapid increase in all tensile directions with increasing the rolling reduction. However, in the 52 and 78 pct-rolled samples, the tensile anisotropy becomes stronger in response to the evolution of the UFEG structure. The transverse σy and σB show almost constant values of 1590 and 1710 MPa, respectively, in the rolling reduction range from 28 to 52 pct and then decrease slightly to 1560 and 1670 MPa at a 78 pct rolling reduction, in contrast to the longitudinal σy and σB, which increase to 1860 and 1870 MPa, respectively. The average transverse total elongation (εt) and reduction of area (δ) decrease significantly at rolling reductions of 52 and 78 pct, in contrast to the longitudinal ones. The decrease in the transverse εt is mainly owing to the loss in the transverse necking elongation. Figure 9 plots the true fracture stress (σf) as a function of the angle (α) between the tensile direction and the RD in the QT, 52- and 78 pct-rolled samples. The σy is also shown in this figure. The σf decreases markedly up to 45 deg and then gradually to 90 deg in both the 52- and 78 pct-rolled samples. The longitudinal σf of the 78 pct-rolled samples is higher than that of the 52 pct-rolled samples, while there is no significant difference in the σf between the two samples at 45 or 90 deg. When compared to the QT samples with the isotropic microstructure, these tempformed samples show higher σf in the longitudinal direction (//RD), while the σf is lower in the transverse direction (//SD).[44]
https://static-content.springer.com/image/art%3A10.1007%2Fs11661-012-1391-2/MediaObjects/11661_2012_1391_Fig7_HTML.gif
Fig. 7

Typical nominal stress-displacement curves as a function of the angle (α) between the tensile direction and the RD for small plate specimens machined from a sample that was tempformed at a rolling reduction of 78 pct

https://static-content.springer.com/image/art%3A10.1007%2Fs11661-012-1391-2/MediaObjects/11661_2012_1391_Fig8_HTML.gif
Fig. 8

Changes in (a) yield strength (σy), (b) tensile strength (σB), (c) uniform elongation (εu), total elongation (εt), and (d) reduction of area (δ) as a function of the rolling reduction and the angle (α) between the tensile direction and the RD for small plate specimens

https://static-content.springer.com/image/art%3A10.1007%2Fs11661-012-1391-2/MediaObjects/11661_2012_1391_Fig9_HTML.gif
Fig. 9

Change in true fracture stress (σf) as a function of the rolling reduction and the angle (α) between the tensile direction and the RD. The average yield strength (σy) is also plotted in the figure

3.3 Charpy Impact Properties

Figure 10 shows the V-notch Charpy absorbed energy as a function of the rolling reduction and the test temperature for the QT and the tempformed samples. The Charpy transition curve of the 28 pct-rolled sample is similar to that of the QT sample exhibiting a DBT from room temperature to 173 K (−100 °C).[44] The warm tempforming up to 28 pct rolling reduction hence has a minimal effect on the Charpy impact properties. The effect of the warm tempforming on the absorbed energy, on the other hand, occurs in the 52 and 78 pct-rolled samples with the strong microstructural anisotropies. As expected, the 78 pct-rolled sample shows much greater enhancement in the Charpy impact properties than the 52 pct-rolled sample. First, the average absorbed energy (vE) of the 78 pct-rolled samples (≈140 J) is nearly twice as high as that of the 52 pct-rolled sample in the high temperature range from 423 K to 500 K (150 °C to 227 °C), where ductile fracture was observed to be the primary mode of failure (i.e., upper-self energy). Secondly, the 78 pct-rolled sample exhibits significant inverse temperature dependence of the vE in the temperature range from 333 K to 213 K (60 °C to −60 °C), whereas the vE of the 52 pct-rolled sample shows the highest value of 94 J at 333 K (60 °C) followed by an abrupt decrease. When compared to the QT samples, the 52 pct rolling reduction is seen to have significant effect on the upper-self energy but not on the DBTT.
https://static-content.springer.com/image/art%3A10.1007%2Fs11661-012-1391-2/MediaObjects/11661_2012_1391_Fig10_HTML.gif
Fig. 10

Change in the V-notch Charpy absorbed energy as a function of the testing temperature and the rolling reduction. Data points with arrows indicate that the specimens did not separate into two pieces during the impact test

Figure 11 shows representative appearances of the Charpy impact bars after testing. There is a large difference in the macroscopic crack path between the samples with and without delamination toughening. For example, in the 78 pct-rolled sample, delamination, where cracks branch parallel to the longitudinal direction (//RD) of the impact bars, apparently occurs at 333 K (60 °C). The delamination fracture was observed to be more striking at the temperatures of 293 K to 213 K (20 °C to −60 °C), where the vE remarkably rose. On the other hand, the 78 pct-rolled samples showed zigzag fracture features in the temperature range from 193 K to 77 K (−80 °C to −196 °C), where the vE shows a sharp drop. When the crack branching angle (β) was defined as the angle between the crack path and longitudinal direction of the impact bar, it was measured to be ~15 deg at 193 K (−80 °C) and increased to ~30 deg at 77 K (−196 °C). A similar trend is observed in the 52 pct-rolled samples. The crack branching angle was measured to be 15 to 25 deg at 333 K (60 °C), where the 52 pct-rolled sample tended to show delamination toughening; however, it increased to 30 to 45 deg at 253 K (−20 °C) or less, where the vE was as low as those of the QT and the 28 pct-rolled samples. Figure 12 is a summary of the relationship between the vE and the crack branching angle in the 52 and 78 pct-rolled samples. Data for the 28 pct-rolled and QT samples that were tested at 253 K (−20 °C) or less are also shown in this figure. There seems to be a correlation between the vE and the crack branching angle. The vE decreases as the crack branching angle increases to 30 deg. This result also suggests that macroscopically the crack branching with an angle of about 15 deg or over leads to diminishing the delamination toughening in the present tempformed samples.
https://static-content.springer.com/image/art%3A10.1007%2Fs11661-012-1391-2/MediaObjects/11661_2012_1391_Fig11_HTML.jpg
Fig. 11

Fracture appearances of V-notch Charpy impact specimens after testing. The crack branching angle (β) was defined as the angle between the crack path and the longitudinal direction (//RD) of the impact bar

https://static-content.springer.com/image/art%3A10.1007%2Fs11661-012-1391-2/MediaObjects/11661_2012_1391_Fig12_HTML.gif
Fig. 12

Relation between the V-notch Charpy absorbed energy and the crack branching angle (β)

Figure 13 shows the representative fracture surfaces of the 52 and 78 pct-rolled samples that were tested at 253 K (−20 °C), where a remarkable difference in the vE was observed between the two samples. In the 78 pct sample, “terraces” are formed on the delamination fracture surface roughly parallel to the longitudinal direction (//RD), and “steps” are formed on the surfaces roughly parallel to the SD. The stepwise crack propagation was also demonstrated on the static bending test of the 78 pct-rolled samples.[24] Such stepwise crack propagation is a common feature in steels exhibiting similar delamination toughening behavior.[45,46] The fracture mode for the terraces is observed to occur primarily by a quasi-cleavage (Figure 13(e)), whereas a ductile fracture mode with fine dimple patterns takes place for the steps (Figure 13(f)) in the present 78 pct-rolled sample. The presence of many steps provides evidence for the extensive local plastic flow. While, in the 52 pct-rolled sample, cracks are seen to propagate stepwise, the crack propagation in the longitudinal direction is less pronounced. The terraces of the 52 pct-rolled sample are wider and shorter (Figure 13(b)) than those of the 78 pct-rolled sample and are primarily linked with quasi-cleavage planes (Figure 13(c)). The linkage of the cleavage cracks in the transverse directions might, thus, result in the degradation of vE. Furthermore, it must be emphasized that the 78 pct-rolled sample shows narrow, elongated quasi-cleavage facets that are aligned in the RD, while the 52 pct-rolled sample is characterized by wide, less-elongated facets. Hence, there are visible differences in the morphology of the quasi-cleavage facets on the terraces of the 52 and 78 pct-rolled samples, depending on the microstructure.
https://static-content.springer.com/image/art%3A10.1007%2Fs11661-012-1391-2/MediaObjects/11661_2012_1391_Fig13_HTML.jpg
Fig. 13

SEM fractographs showing fracture surfaces for the samples that were tempformed at rolling reductions of 52 pct (a through c) and 78 pct (d through f) after V-notch Charpy impact testing at 253 K (−20 °C). (b) and (e) are magnified images of “terraces” that are formed on the surfaces roughly parallel to the RD, and (c) and (f) are magnified images of “steps” that are formed on the surfaces roughly parallel to the SD. The black arrows in (a) and (d) indicate the steps

4 Discussion

4.1 Microstructural Evolution During Warm Tempforming

The HABs in the UFEG structure are suggested to be generated during the warm plastic deformation not only from the extension of original boundaries in the tempered martensite, such as the martensite block, packet and prior-austenite grain boundaries but also from the evolution of LABs, which involve dislocation boundaries (Figures 3 and 4). Figure 14 is a plot of the KAM average values and the average dislocation density (ρ), which were estimated by EBSP and XRD, respectively. There are two types of dislocations in the tempered martensitic steel; mobile and immobile dislocations. The immobile dislocations can be further divided into two types of stored dislocations[47]; the statistically stored dislocations (SSDs) that accumulate mainly as a result of random trapping processes of mobile dislocations inside the grains and the geometrically necessary dislocations (GNDs) that are necessary to maintain continuity between the grains. The KAM quantifies the average misorientation around a measurement point with respect to a defined set of nearest or nearest plus second-nearest neighbor points and is used to evaluate the GND density.[28] The variations in the KAM average values indicate that the density of the GNDs might increase at the early stage of tempforming and show a maximum in the rolling reduction range from 28 to 52 pct, followed by a slight decrease. This tendency is similar to that in the LABs with a misorientation angle of 2 ≤ θ < 10 deg in Figure 4. In contrast, the ρ of the QT samples was estimated to be 1.9 × 1015 m−2, while it decreased to 3.6 × 1014 m−2 after 78 pct rolling. This result can be supported by the TEM observation in Figure 5. The XRD analysis also gives information regarding the dislocation arrangement parameter,[48] M = Reρ0.5, which shows a sharp drop at a low rolling reduction up to 12 pct and then tends to increase. The decrease in M is assumed to indicate the increase in both the strength of the dipole character and the screening effect in the displacement fields of dislocations.[48] In other words, the large value of M in the QT sample suggests that the present sample might contain a large number of mobile dislocations and/or the SSDs that might be remobilized through warm tempforming. Nakashima et al.[49] reported, in an ultra-low-carbon martensite (Fe-18 pct Ni alloy), that excess mobile dislocations that were introduced during the martensitic transformation were easily annihilated during the early stage of cold-working and then the retained dislocations were trapped in the form of a tangled dislocation cell structure. In 9 and 12 pct Cr steels, mobile dislocation density was observed to decrease during long term tempering and creep; however, the dislocation density was still high as compared to other metals.[50] In addition, the decrease in the ρ during deformation of metals has been observed in heavy cold working of metals at large true strains exceeding 2.[36,51,52] In the case of the ferritic stainless steel that was subjected to cold rolling/swaging, the number of internal dislocations was reported to increase to its maximum (~1.4 × 1015 m−2) upon processing to total strains of about 1 to 2 and then to smoothly decrease to 6 × 1014 m−2 at strains of about 2 to 4, in which the development of ribbon-like microstructures attends a remarkable increase in the misorientations between deformation (sub) grains.[36] Hence we suggest that the decrease in the ρ at the early stages of the present tempforming can be related to the annihilation and rearrangement of the mobile dislocations and/or SSDs in the tempered martensite, although most of the excess mobile dislocations that were introduced from during the martensitic transformation might be annihilated during tempering at 773 K (500 °C) for 60 minutes. In the later stages, the retained SSDs as well as mobile dislocations in the tempformed samples decrease during the evolution of the UFEGs. On the other hand, the GNDs accumulate to form LABs, which eventually developed into HABs, leading to grain subdivision during the tempforming.
https://static-content.springer.com/image/art%3A10.1007%2Fs11661-012-1391-2/MediaObjects/11661_2012_1391_Fig14_HTML.gif
Fig. 14

Changes in the KAM average value for a 1st neighbor rank that was estimated by EBSP analysis, and average dislocation density (ρ) and dislocation arrangement parameter (M) estimated by XRD line profile analysis using the modified Warren-Averbach method.[29] Closed and open symbols for the ΚΑM average value denote the data in the transverse and the longitudinal planes that are normal and parallel to the RD, respectively

From the findings above, the microstructural evolution during warm tempforming using multi-pass caliber rolling can be divided into the following three stages. At low rolling reductions below 30 pct (ε ~ 0.4), there are marked changes in the dislocation substructures of the tempered martensite: (1) the annihilation and rearrangement of mobile dislocations and/or SSDs that can be remobilized through warm tempforming and (2) the increases in the GND density. At middle rolling reductions between 30 and 50 pct (ε = 0.4 to 0.9), the heterogeneous deformation can be recognized in the matrix grain structures, such as the martensite blocks, packets, and prior-austenite grains on the micron scale. At large rolling reductions from 50 to 80 pct (ε = 0.9 to 1.8), the grain subdivision accompanied by the evolution of strong texture evolution becomes more significant. The immobile dislocations as well as the mobile dislocations might decrease through grain subdivision. In addition, we must pay attention to the fact that the temperature rose to around 873 K (600 °C) during warm tempforming using multi-pass caliber rolling. Such deformation heating is considered to influence the microstructural evolution through accelerating the processes of the recovery, recrystallization, and spheroidization of carbide particles, in addition to the rolling reduction. In such a case, the nanometer-size carbide particles that are thermally stable apparently play important roles in accumulating dislocations and retarding the migration of LABs and/or HABs through their pinning effect, although they grow slightly and are oriented in the RD during the tempforming. As a result, the UFEG structure with strong 〈110〉//RD fiber deformation texture evolves.

4.2 Dominating Microstructural Factors Controlling Tensile Properties

It has been reported that a small degree of cold working results in a marked increase in the yield strength of as-quenched martensite although the dislocation density is reduced.[49] This is thought to be due to the annihilation and immobilization of the excess mobile dislocations because the mobile dislocations can lower the yield strength through the generation of plastic strain.[49] The accumulation of immobile dislocations, such as GNDs and SSDs, on the other hand, can contribute to the increase in the flow stress of deformed metals.[47] Yin et al.[30] investigated the dislocation structures in warm-rolled low-carbon steels with ultrafine ferrite grains (~1 μm) by means of TEM and XRD line profile analysis using the modified Warren-Averbach method.[29] They pointed out that not only the dislocation density, but also the dislocation interaction behavior controlled the changes of yield strength of the ultrafine-grained low-carbon steels; an increase in the dislocation interaction as well as the dislocation density might lead to an increase in yield strength. There are no significant changes in the carbide structure, the HABs, and the texture at the early stage of the present tempforming up to rolling reductions of 28 pct. From these findings, the rapid increase in the σy at the early stage of the present tempforming up to rolling reductions of 28 pct can be attributed to the reconfiguration in the dislocation substructure: (1) the annihilation and rearrangement of the mobile dislocations and SSDs in the tempered martensite and (2) the increase in both the GND density and the LABs (Figures 4 and 14). The sharp drop in the dislocation arrangement parameter, M[48] at a low rolling reduction up to 12 pct is suggested to be due to the sharp increase in the dislocation interaction through annihilation and rearrangement of the mobile dislocations and/or SSDs in the QT sample. Since the development of the dislocation substructure has similarities in almost all directions, the tempformed samples show almost isotropic strengthening behavior. However, in the large rolling reduction range from 52 to 78 pct, the yielding behavior of the tempformed sample becomes significantly anisotropic in response to the evolution of the UEFG structures, and the strengthening mechanism should be considered separately for each tensile direction. Here, let us consider the microstructural factors controlling the longitudinal and the transverse σy in the final stage of tempforming. For the longitudinal direction (//RD), the significant increase in the σy coincides with the increase in the amount of the HABs and the development of the strong 〈110〉//RD fiber deformation texture. The texture hardening effect[53,54] is, however, considered to be small in the present tempformed sample. This is because the post-annealing at 973 K (700 °C) for 60 minutes of the 78 pct-rolled sample resulted in the reduction of the longitudinal σy to 1000 MPa even though the strong 〈110〉//RD fiber deformation texture remained. The post-annealed samples had a transverse AIL of about 0.5 μm,[22] and a Hall-Petch relationship was observed between the transverse grain size and the longitudinal σy in the post-annealed sample and the 0.6 pct C-2 pct Si-1 pct Cr steels that were tempformed at temperatures of 773 K to 973 K (500 °C to 700 °C) using a rolling reduction of 80 pct.[23] Although the longitudinal σy of the present 78 pct-rolled sample tended to deviate upward from the Hall-Petch relation,[23] it may be concluded that the increase in the longitudinal σy in the final stage of the tempforming predominantly arises from the refinement in the transverse grain size. On the other hand, the change in the transverse σy is observed to correspond to the changes in the GNDs, the LABs, and the carbide particle distribution rather than changes in the HABs, as seen in Figures 4 and 5. If the transverse σy depended on the longitudinal AIL, the transverse σy should decrease significantly as the longitudinal AIL increased to 0.8 μm through the 78 pct tempforming. The contribution of grain refinement strengthening to the transverse σy is hence considered to be small in the present UFEG structure. Similarly, Ohmura et al.[55] suggested that, by using the nanoindentation technique, the grain size effect on the macroscopic strength was significantly reduced by coarsening of the martensite block that was analogous to effective grain in martensite when Fe-0.4 pct C was tempered at 723 K (450 °C) for 90 minutes. As indicated in Figure 7, the present tempformed samples exhibit continuous yielding behavior in the transverse tensile directions. From these findings, the strengthening mechanism of the transverse σy in the UFEG structure is similar to that of the tempformed samples with a low rolling reduction of 28 pct or less. The transverse σy of the tempformed samples is hence considered to depend on the substructure of the matrix grains including immobile dislocations, the LABs, and the nanometer-size carbide particles. Such anisotropic yielding behavior of the tempformed samples can influence the delamination behavior.

Ultrafine grained metals often show relatively large necking elongation but little uniform elongation, resulting in low tensile elongation.[37,4043] Such little uniform elongation is well explained in terms of the plastic instability condition to bring necking in tension; the relationship of σ < dσ/dε should be maintained to produce uniform elongation, where σ and ε are true stress and true strain, respectively.[37,4043] The grain refinement increases the yield strength especially in the ultrafine grain size below 1 μm, but it does not significantly affect the work-hardening rate (dσ/dε). This eventually causes plastic instability in the early stage of tensile deformation. Takaki et al.[56] reported that work hardening of ferritic steel after yielding at room temperature became smaller with decreasing the grain size. They also suggested that work-hardening might not take place in ferritic steels with finer grain size below 0.35 μm because dislocation strengthening could not exceed the initial high yield stress (=1.1 GPa) obtained by grain refinement strengthening. However, the present tempformed samples as well as the QT sample have shown adequate uniform elongation despite their ultrafine transverse grain size and high σy (Figure 6). Figure 15 shows the change in the true σ-ε curve and the work-hardening rate (dσ/dε)-true ε curve as a function of the rolling reduction for the JIS 14A specimen that was machined in the RD. The dσ/dε in the QT and 28 pct-rolled samples decreases monotonously with increasing the ε. Unique work-hardening behavior is observed in the 52 and 78 pct-rolled samples. In particular, the dσ/dε of the 78 pct rolled sample approximates the σ just after yielding and shows a slight increase up to a true ε of 0.07, followed by a rapid decrease. Such dσ/dε-ε curves have been observed in the transformation induced plasticity (TRIP)[57] and twining induced plasticity (TWIP)[58,59] steels. Takemoto and Senuma, on the other hand, observed similar dσ/dε-ε curves in the tensile deformation of bcc single crystals for a Fe-3 pct Al alloy. For most of tensile orientations, including 〈001〉, the dσ/dε-ε curves changed from concave to convex with increasing the ε.[60] The dσ/dε for the 〈101〉 tensile orientation also showed a sharp drop after reaching a maximum value. Since the Fe-3 pct Al alloy dose not undergo any deformation-induced transformations, such a dσ/dε-ε curve is considered to be due to the transition of the dislocation structure during deformation. Furthermore, they reported that the dσ/dε was observed to show monotonous decrease similar to that observed in the tensile deformation of polycrystalline Fe-3 pct Al alloy, when the tensile axes were on the line connecting 〈102〉 and 〈101〉 in the stereo triangle. In the present 78 pct-rolled sample, the work-hardening behavior in the transverse tensile direction appears to be similar to that seen in the 28 pct-rolled sample or the QT sample (Figure 7). From these findings, the strong 〈110〉//RD fiber deformation textures in the 52 and 78 pct-rolled samples are suggested to influence their dσ/dε-ε curves; however, we need further study to confirm it by considering the influences of the ultrafine grain size and elongated grain shape. In addition, we consider that nanometer-size carbides that are dispersed in the UFEG matrix can play an important role in producing adequate uniform elongation at the ultra-high σy. This is because none of cold-drawn pure iron wires were reported to exhibit more than 1 pct uniform elongation despite their UEFG structures with strong 〈110〉//RD fiber deformation textures.[35] Dispersion of fine second phase particles has been reported to improve the εu of ultrafine grained metals.[41,61,62] Finely dispersed particles, such as carbides[41,61] and oxides,[62] are thought to promote the accumulation of GNDs during tensile testing and to improve the work-hardening capacity. Furthermore, it must be emphasized that the transverse εu of the 78 pct-rolled samples is comparable to those of the QT samples. This indicates that the dispersion strengthening by the nanometer-size carbide particles is effective even in the transverse tensile direction of the UFEG structure.
https://static-content.springer.com/image/art%3A10.1007%2Fs11661-012-1391-2/MediaObjects/11661_2012_1391_Fig15_HTML.gif
Fig. 15

Change in the work-hardening rate (dσ/dε) as a function of the true strain and the rolling reduction for the JIS 14 A specimen. True stress-strain curves are also presented as a function of the rolling reduction

The average transverse necking elongation (εn), the reduction of area (δ), and the true fracture stress (σf) become much lower in the 52 and 78 pct-rolled samples in contrast to the longitudinal ones, which tend to increase (Figures 8 and 9). This can be rationalized by the development of the strong 〈110〉//RD fiber deformation textures and the elongated grain shape because lots of {100} cleavage planes[11] are provided on the longitudinal planes parallel to the RD (Figure 2). This is also supported by the fact that the degradation in the εn, δ and σf is significant in the angle (α) between the tensile direction and the RD of 45 deg as well. The 〈110〉//RD fiber deformation textures also provide numerous {100} cleavage planes on the transverse planes with the angle of ~±45 deg to the RD. SEM fractography of the fracture surfaces for the 78 pct-rolled samples after tensile testing is shown in Figure 16. The fracture surface in the longitudinal tensile direction (//RD) shows primarily a ductile fracture mode by microvoid nucleation, growth and coalescence (Figure 16(a)). On the other hand, the fracture mode on the fracture surfaces at α = 45 and 90 deg (//SD) is characterized by a mixture of quasi-cleavage and ductile dimple rupture (Figures 16(b) and (c)). A similarity was observed in the 52 pct-rolled sample. Such quasi-cleavage fracture is, hence, considered to occur primarily because of the alignment of {100} planes and degrades the εn, δ, and σf.
https://static-content.springer.com/image/art%3A10.1007%2Fs11661-012-1391-2/MediaObjects/11661_2012_1391_Fig16_HTML.jpg
Fig. 16

SEM fractographs showing fracture surfaces for the samples that were tempformed at a rolling reduction of 78 pct after tensile testing at the angle (α) between the tensile direction and the RD of 0 deg (a), 45 deg (b), and 90 deg (c)

4.3 Dominating Microstructural Factors Controlling Impact Properties

Thermomechanically treated steels,[1625,45,46,63,64] laminates[3] and laminated composites[65] were reported to exhibit delaminations due to their anisotropic microstructures. Examples of laminate geometries[3] that show delamination toughening are shown in Figure 17. There are two basic geometries, “crack divider” and “crack arrester.” In both cases, weak interfaces are present parallel to the longitudinal direction (//RD) of the impact test bar, and the interaction between the weak interfaces and the stress, σt//z or σt//SD, that is generated by the localized plastic constrain at the notch and/or the crack tip can cause delaminations. In the crack divider orientation, delamination at the weak interfaces divides the crack into a series of cracks, and, thereby, the stress condition at the crack tip is relaxed from a triaxial stress condition and becomes a biaxial planar stress condition, interrupting the propagation of the main crack. As a result, the DBTT is lowered.[3,16,19,20] However, the ductile fracture that is accompanied by the delamination results in the reduction of upper-shelf energy because the effective width of the specimen across the notch decreases.[3] On the other hand, delamination in the crack arrester orientation can change the stress state from a triaxial tension state towards a uniaxial tension state and blunt the crack tip. In this case, crack re-initiation is necessary to fracture a material and occurs under conditions of nearly uniaxial tension, which is an unfavorable cleavage; hence, higher vE is obtainable.[3,45,46,63,65] In the present 78 pct-rolled sample, the delamination fracture is similar to that seen in the crack arrester type, and the stepwise crack propagation in Figure 13(d) shows the trace of crack re-initiations. The remarkable enhancement in vE at a temperature range from 333 K to 213 K (60 °C to −60 °C) is, therefore, a result of the delaminations in the crack arrester type.
https://static-content.springer.com/image/art%3A10.1007%2Fs11661-012-1391-2/MediaObjects/11661_2012_1391_Fig17_HTML.gif
Fig. 17

Schematic illustration of laminate geometry showing delamination toughening

Delamination frequently occurs in association with the texture[1620] and the weak interfaces between the matrix and the inclusions, such as the elongated MnS[46] and the carbides[63] that were aligned to the RD. McEvily and Bush[45] observed an inverse temperature dependence of the vE in the vicinity of around 473 K (200 °C) for an ultrahigh-strength 0.2 pct C-3 pct Ni-3 pct Mo steel processed by an ausforming treatment. In the ausformed steel, delamination occurred along the boundaries of the elongated, coarse prior-austenite grains, and the vE showed an extremely high value of 325 J; however, delamination did not occur in the vicinity of room temperature, and the vE decreased to 33 J. On the other hand, in the 78 pct-rolled sample, the primary fracture mode on the delamination planes is quasi-cleavage, as shown in Figure 13(e), and the delamination, thus, occurred in association with the UFEG structure.

Let us consider the dominating microstructural factors to control the delamination fracture in the UFEG structure. Firstly, the 〈110〉//RD fiber deformation textures provide considerable {100} cleavage planes on the longitudinal planes parallel to the RD and on the transverse planes with the angle of ~±45 deg to the RD and the SD of the impact bar, while {110} planes are present on the transverse planes normal to the RD, as shown in Figures 9 and 16. Hence, the 〈110〉//RD fiber deformation texture is one of the principal microstructural factors for crack branching. Secondly, the grain shape is also an important factor. Cleavage fracture occurs when the peak tensile stress in the process zone of a crack tip (σt) exceeds the cleavage fracture stress (σcf).[11,12] The σt is proportional to the σy, and is of the order (3 to 5) σy.[11] Effective grain size (deff) for the cleavage fracture is the coherence length on {100} planes, which corresponds to the cleavage crack length in bcc steel, and the σcf is proportional to the reciprocal of the square root of the deff; σcf = Kdeff−1/2.[11] For the elongated grain structure, the coherence length on {100} planes is longer in the RD than in the transverse directions with an angle of ~±45 deg to the RD due to the elongated grain shape. The elongated grain with a 〈110〉//RD fiber deformation texture, hence, gives a condition of σcf//SD < σcf@45 < σcf//RD.[44] Thirdly, according to the Yoffee diagram, the σy of bcc metals, such as low-alloy steel, increases sharply, especially in the lower-temperature range, and this eventually causes a clear DBT.[11] The refinement of the grain size increases the σcf as well as the σy; however, the grain refinement effect is much larger on the σcf than on the σy, leading to the lowering of the DBTT. This means that finer transverse grain size is needed to suppress the occurrence of brittle fracture on the transverse planes and cause delamination toughening at lower temperature. Clearly, the ultra refinement of the transverse grain structure is the key to enhance both the yield strength and delamination toughening at lower temperature in the UFEG structure.[23] Therefore, in conclusion, the combination of the microstructural factors, i.e., the transverse grain size, grain shape, and 〈110〉//RD fiber texture, can control delamination toughening in UFEG structure steel.

Figure 18 is an illustration of the Yoffee diagram[23,44] for the UFEG structure with a strong 〈110〉//RD fiber deformation texture. For the triaxial stress system associated with the notch in a Charpy specimen, the σt is expressed by a relation of σt//SD < σt@45 < σt//RD by considering the yield strength anisotropy in Figure 8 and the effect of the notch geometry of the impact bar.[16,18] Hence, there are three separate σtvs temperature curves; however, the cleavage crack will hardly propagate in the SD despite the highest σt because the {100} cleavage planes are not present on the transverse planes in the UFEG structure. In fact, it was confirmed that the QT sample lost most of its tensile ductility when the σy increased to 1.78 GPa at 77 K (−196 °C), while good tensile ductility was retained for the UFEG structure sample although the σy increased to 2.29 GPa.[22] Below T1, in which the σt//SD exceeds the σcf//SD, cleavage fractures on the longitudinal {100} planes preferentially occur, resulting in delamination. A thermal increment in the σy with decreasing temperature can eventually promote delamination, and this leads to the inverse temperature dependence of vE (curve 1). Below T2, in which the σt@45 exceeds the σcf@45, the cleavage fracture is also induced on the {100} cleavage planes with the angle of ~±45 deg to SD, leading to a diminishing of delamination toughening (curve 2). When the difference between the σcf//SD and the σcf@45 is smaller, the delamination toughening becomes less pronounced. In addition, the fact that there is a great difference in delamination toughening between the 52 and 78 pct-rolled samples (Figure 10) requires attention, even though the relation between the σcf//SD and the σcf@45 in the 52 pct-rolled sample appears to be similar to that in the 78 pct-rolled sample from the result in Figure 9. This is probably related to the microstructural inhomogeneity in the matrix grain structure. In comparison to the 78 pct-rolled sample, the 52 pct-rolled sample contains more wide and less elongated grains that are not fully aligned to the RD (Figures 2 and 3). These grains provide preferable paths for cracks to propagate to the transverse directions during an impact test, and linkage of the transverse cracks may lead to the suppression of crack growth along the longitudinal direction. In other words, the occurrence of brittle delamination is suggested to be sensitive to transverse crack propagation (Figure 13), and microstructural inhomogeneity to produce transverse cracking needs to be minimized in the UFEG structure to achieve delamination toughening. Torizuka et al.[66] investigated the effect of multi-pass caliber rolling at 773 K (500 °C) on the microstructure and the Charpy impact properties of a low-carbon steel with initial microstructure of (ferrite + pearlite) or bainite. They reported that Charpy impact properties deteriorated up to a rolling reduction of 51 pct and then drastically improved beyond a rolling reduction of 79 pct, where the pearlitic or bainitic structure could be fully replaced by an ultrafine grain structure with a distribution of spheroidized carbide particles.
https://static-content.springer.com/image/art%3A10.1007%2Fs11661-012-1391-2/MediaObjects/11661_2012_1391_Fig18_HTML.gif
Fig. 18

Yoffee diagram for UFEG structure with a strong {110} fiber deformation texture. Cleavage fractures on the longitudinal {100} planes (//RD) cause delamination toughening (curve 1), while those on the {100} planes with the transverse components diminish the delamination toughening (curve 2)

As shown in Figure 10, the upper-self energy in the temperature range from 423 K to 500 K (150 °C to 227 °C) increases in response to the evolution of the UFEG structure. Enhancement of the upper-self energy was commonly observed through the evolution of UFEG structures with strong 〈110〉//RD fiber deformation textures.[22,23,25,67] The microstructural factors affecting the upper-self energy of the UFEG structures were considered to be the transverse matrix grain size, the texture and the carbide particle distribution.[23] When compared to the 52 pct-rolled sample, the 78 pct-rolled sample has finer transverse grain size and stronger 〈110〉//RD fiber deformation texture. Hence, the ultrafine transverse grain structure with considerable {110} ductile planes is responsible for the enhancement in the upper-self energy. This also may be supported by the fact that steel wires with UFEG structures with strong 〈110〉//RD fiber deformation textures were observed to fracture in a ductile manner despite the ultra-high tensile strength of 3 GP or over.[68,69] In addition, finer particle size can be expected to lead to higher void nucleation resistance.[8] Alignment of carbide particles with an average aspect ratio of 1.6 in the RD can provide finer transverse particle size on the transverse plane in the UFEG structure, as shown in Figure 5. Hence, the spherical carbide particle distribution and the alignment of the intergranular carbide particles in the UFEG structure are suggested to be convenient to suppress the degradation of the upper-self energy.

5 Conclusions

The influence of warm tempforming on the microstructure, tensile, and Charpy impact properties was investigated as a function of the caliber-rolling reduction for a 0.4 pct C-2 pct Si-1 pct Cr-1 pct Mo steel that was quenched and tempered at 773 K (500 °C) and, subsequently, tempformed using multi-pass caliber rolling. The following results were obtained.
  1. (1)

    At the large rolling reduction range from 52 to 78 pct, not only extension of the martensite blocks and/or the packets in the RD but also grain subdivision became more significant. As a result, an UFEG structure with a transverse grain size of about 0.3 μm and a strong 〈110〉//RD fiber deformation texture evolved after 78 pct rolling. Through tempforming, the dislocation density decreased and the nanometer-size carbide particles grew slightly, becoming more spherical.

     
  2. (2)

    The room-temperature tensile deformation behavior was almost isotropic at low rolling reductions up to 28 pct, above which it became significantly anisotropic. The longitudinal yield strength (σy) of the quenched and tempered (QT) sample increased from 1480 to 1860 MPa through 78 pct rolling, whereas the transverse σy leveled off at around 1600 MPa up to 28 pct rolling. Although the effect of tempforming on the anisotropy of the uniform elongation (εu) was small, the transverse necking elongation (εn), reduction of area (δ), and true fracture stress (σf) were markedly degraded in the 52 and 78 pct-rolled samples in contrast to the longitudinal ones.

     
  3. (3)

    The warm tempforming significantly enhanced the Charpy impact properties at the large rolling reduction range from 52 to 78 pct. The upper-self energy of the 52 and 78 pct-rolled samples rose about 2 and 4 times as high as that of the QT sample (the average absorbed energy, vE = 33 J), respectively. The vE of the 52 pct-rolled sample decreased significantly at the temperature range from 333 K to 213 K (60 °C to −60 °C), while the vE of the 78 pct-rolled sample remarkably increased because of brittle delamination.

     
  4. (4)

    The transverse σy was suggested to be influenced by the substructure of the UFEGs, while the longitudinal σy was influenced by the transverse grain size. Although the work hardening rate curves in the 52 and 78 pct-rolled samples appeared to be influenced by the 〈110〉//RD fiber deformation texture, dispersion strengthening by nanometer-size carbide particles with an average particle size of less than 50 nm was considered to be effective to produce εu in the UFEG structure. The εn, δ, and σf were influenced by the 〈110〉//RD fiber deformation texture, and the degradations in the transverse εn, δ, and σf might be related to the alignment {100} cleavage planes.

     
  5. (5)

    It was confirmed that the combination of the microstructural factors, i.e., the transverse grain size, grain shape, and 〈110〉//RD fiber texture, could control delamination toughening in the UFEG structure. The comparison between the microstructures of the 52 and 78 pct-rolled samples suggested that the microstructural heterogeneity in the matrix grain structure to produce transverse cracking should be minimized to achieve delamination toughening.

     

Acknowledgments

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 Drs. Nie and Yin for their quantitative XRD analysis. The study was partly supported by KAKENHI 21560763 and partly 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”.

Copyright information

© The Minerals, Metals & Materials Society and ASM International 2012