2.1 Materials

The experimental steels used in this study were low-carbon pipeline steel produced by TMCP process. The detailed chemical composition was measured by inductively coupled plasma-atomic emission spectroscopy (ICP-AES) and the results are shown in Table 1.

Three steel plates were fabricated by TMCP process with different cooling conditions, and the parameters are shown in Table 2. After austenitizing at 1190 ℃ for an hour, rolling was started at about 1100 ℃ and finished in the austenite region of 840 ℃. A grain refinement effect was expected by controlled rolling with rolling reduction ratio of about 85% in the non-recrystallized region of austenite. The thickness of plates was reduced from 200 mm to 31.75 mm. Some rolled plates were air-cooled to different start cooling temperatures of 780 ℃ (C1), 740 ℃ (C2) and 700 ℃ (C3) to regulate the volume fraction of microstructures and stabilize retained austenite. The steel plates were then water-cooled to 480 ℃ and finally air-cooled to room temperature.

2.2 Microstructure characterization

After the controlled rolling and controlled cooling process, the specimens were taken from the 1/4 thickness of steel plates. The specimens were further processed by the standard mechanical polishing procedures with dimensions of 10 mm× 10 mm×8 mm. The polished samples were then chemically etched by 4 vol% nital solution for 10-14 s to reveal microstructure. The microstructure of each sample was observed by optical microscope (OM, LEICA DMI3000M) and scanning electron microscope (SEM, FEI Quanta-200). The crystallographic orientation information of samples was also characterized by electron backscattered diffraction (EBSD, Nordlysmax2) experiments. The samples were cut from the 1/4 thickness of steel plates. And they were ground and electrolytically polished in 10 vol% perchloric acid alcohol solution. Then, the experiments were put forward by scanning electron microscope (SEM, SIRION200) with electron backscattered diffraction probe. The EBSD experiments were carried out at accelerating voltage of 20 kV and step size of 5 μm. The obtained statistics were analyzed by HKL Channel 5 software. More detailed microstructural information was observed by transmission electron microscope (TEM, Talos F200x). The samples for TEM experiments were cut and mechanically ground to 50 μm thickness, and then were twin-jet electropolished in mixed 10 vol% perchloric and 90 vol% ethanol solution at -25 ℃ until observable thin area was obtained. The statistical information of the microstructure was evaluated by Image Pro Plus software. Then, the fraction of matrix phases and the average size of MA islands were obtained.

2.3 Low-temperature toughness test

According to ISO 148-1-2016, the standard Charpy V-notch impact specimens were machined into dimension of 10 mm×10 mm×55 mm. The impact tests were performed in a temperature range of -160 ℃ to -10 ℃ and were carried out three times in each condition. After low-temperature impact tests, the fracture surfaces of the three experimental samples were characterized by SEM.

3.1 Microstructure

SEM micrographs of steel plates under different cooling conditions are shown in Figure 1. According to the ISIJ Bainite Committee classification [21], the microstructure of GB consists of elongated ferrite sheaves with roughly equiaxed carbon-enriched MA islands. The classification defines BF as a microstructure consisting of highly dislocated ferrite laths separated by MA constituents. The BF defined by classification is like the well-known upper bainite, but cementite is replaced by MA constituents for the low-carbon content available. As shown in Figure 1, the microstructure of the three samples consisted of two parts: one was the QF microstructure formed at a higher temperature during the slow cooling process, and the other was the mixture of GB and BF microstructure formed at a lower temperature during fast cooling process. From OM photographs, the number fractions of QF, GB and BF microstructure were collected from OM diagram and the size distribution of MA was characterized from OM diagram. The results are shown in Table 3 and Figure 1(d) respectively. From C1 to C3, QF and BF increased and GB decreased, the average size of MA islands was 1.27, 1.44 and 1.56 μm, respectively.

Figure 1全屏查看

Microstructure results of steels (a₁, a₂, a₃) C1, (b₁, b₂, b₃) C2 and (c₁, c₂, c₃) C3, (a₁, b₁, c₁) OM micrographs, (a₂, b₂, c₂) SEM micrographs and (a₃, b₃, c₃) EBSD micrographs; (d) Statistics of average MA size; (e) Effective grain size; (f) HAGBs

The EBSD results of grain boundary misorientation information for three samples are also presented in Figure 1. As shown in Figures 1(a₃)-(c₃), the green lines are low-angle grain boundaries (2°<LAGBs<15°) and the black lines are high-angle grain boundaries (HAGBs>15°). It was noted that the effective grains of the three samples can be divided into two types, where some grains contained low density of LAGBs and the others had high density of LAGBs. The former type of grain is related to ferrite formed at high temperature and the latter type of grain is related to low-temperature transformation microstructure (GB and BF). The grain boundary orientation information of the three samples was characterized, and the distribution maps of grain boundary orientation and effective grain size were drawn. As shown in Figures 1(e)-(f), the average effective grain size of C1, C2 and C3 was 3.5, 2.8 and 3.0 μm, and their proportion of HAGBs were 30.3%, 36.3% and 34.2%. C2 had the finest grain and highest density of HAGBs, which is beneficial to its low-temperature toughness.

Detailed microstructural characterization of samples is performed by TEM analysis and the results are shown in Figure 2. Figure 2(a) presents the microstructure of C1, where the typical morphology of QF and GB is shown in the figure. The QF grains were bright wight, possessing low-density dislocation. The GB grains contained relatively high-density dislocation and blocky MA islands, where MA islands were distributed along grain boundaries or inside grains. Figure 2(b) shows the microstructure of C2, where QF, GB and BF are all presented in the selected region. The BF microstructure contained a higher dislocation density than GB and QF, and strip-like MA islands were distributed among elongated ferritic lath bundles inside BF grain. As shown in Figure 2(c), the microstructure of QF and BF can be observed in C3. The QF microstructure of C3 was the brightest among the three samples, indicating its lowest dislocation density.

Figure 2全屏查看

TEM micrographs of experimental steel: (a) C1; (b) C2; (c) C3

As shown in Figure 3, the three samples had similar crystallographic orientation with a strong <110> orientation and a weak <001> orientation parallel to TD. But the content of <110> and <001> oriented grains in the three samples were different. C2 sample had higher content of <110> oriented grains and lower content of <001> oriented grains compared with the other two samples. Because the specimens for Charpy impact test were prepared along the TD, the <110> orientation parallel to TD should represent {110} slip planes and the <001> orientation should represent {001} cleavage planes. The content of <110> and <001> oriented grains parallel to TD may be responsible to the variation of low-temperature toughness of the samples.

Figure 3全屏查看

Distribution of <110> orientation parallel to the TD for (a) C1, (b) C2 and (c) C3; Distribution of <001> orientation parallel to the TD for (d) C1, (e) C2 and (f) C3

Figures 4(a)-(c) illustrate the φ₂=45° section of the orientation distribution function (ODF) for the experimental steel. To help analyze texture components, some ideal orientations along RD, TD and ND fibers are exhibited in Figure 4(d). As shown in ODFs, C1 exhibited high intensities around components of rotated cube {001}<110>, {112}<110> and low intensities around {332}<113>. By contrast, C2 possessed low intensities around {112}<110> and high intensities around {332}<113> components. The intensities of C2 around rotated cube {001}<110> were too low to be found. For C3, the intensities around rotated cube {001}<110> and {332}<113> components were high, but the intensities around {112}<110> components could hardly be characterized.

Figure 4全屏查看

ODF (φ₂=45° section) of the (a) C1, (b) C2, (c) C3; (d) φ₂=45° section of Euler space

3.2 Low-temperature toughness

As shown in Figure 5, Charpy impact tests were performed at temperatures from -160 to -10 ℃ to evaluate the fracture behavior of the three samples. The ductile-to-brittle transition temperature (DBTT) of samples was obtained by the energy method where the temperature was defined as the average value of upper shelf energy (USE) and lower shelf energy (LSE). After the tests, curves of Charpy impact absorbed energy versus temperature of three samples were fitted as shown in Figure 5. On the whole, the low temperature impact energy decreased with the reduction of the experimental temperature. The DBTT of C1, C2 and C3 was -79, -86 and -82 ℃ respectively, and the USEs of them were 261 J, 298 J and 278 J respectively. In general, C2 had the highest low-temperature toughness, while C1 had the lowest low-temperature toughness.

Figure 5全屏查看

Low-temperature toughness of the samples

3.3 Fracture behavior

Figure 6 exhibits the SEM observations for the fracture morphology of three samples after impact tests at -20, -80 and -120 ℃. When the test temperature was -20 ℃, ductile fracture morphology was predominantly observed in the three samples, where brittle fracture morphology was scarcely observed. The dimples of C2 were large, deep and evenly distributed, showing enhanced toughness characteristics. On the contrary, the dimples of C1 were small and shallow, and the dimples of C3 were unevenly distributed, both of which showed weakened toughness characteristics. For the fracture surface obtained at -80 ℃, some cleavage fractures appeared. The three samples tested at -80 ℃ all exhibited mix-mode (ductile/brittle) fracture behavior. C2 had the largest ductile area and C1 had the smallest ductile area. For the fracture surface obtained at -120 ℃, typical patterns of brittle fracture behavior with cleavage facets were observed. The crack propagation direction on the cleavage planes can be indicated by arrows in Figure 6. Microcracks observed in the morphology of C1 were large in number and deep in depth, while the microcracks of C2 were small and shallow. It can be found that the propagation paths of cracks were more tortuous in C2 than C1 and C3. The front end of cleavage facet was developed into several branches along the propagation direction of cracks. The characteristics of fracture morphology were consistent with the results of toughness variation of steels.

Figure 6全屏查看

SEM observations of the impact fracture morphology for (a, d, g) C1, (b, e, h) C2 and (c, f, i) C3: (a, b, c) Tested at -20 ℃; (b, e, h) Tested at -80 ℃; (c, f, i) Tested at -120 ℃

4.1 Effect of start cooling temperature on microstructural evolution

The continuous cooling process of experimental steel is composed of slow cooling process after hot rolling and fast cooling process after start cooling. During slow cooling process, QF preferentially nucleates at prior austenite grain boundaries by diffusion mechanism at high temperature. Ferrite growth is characterized by long-range diffusion of carbon atoms and rapid migration of replacement atoms, resulting in different composition between ferrite and austenite [22, 23]. During the process, austenite-stabilizing elements, especially carbon, aggregate in austenite, stabilizing austenite grains. When experimental steels are cooled to the start cooling temperatures, a rapid cooling process occurs. Then, the microstructure of GB and BF is formed by mixed shear and diffusion transformation mechanism under a high thermodynamic driving force. As the start cooling temperature decreases, the austenite stability is enhanced and the transformation of undercooled austenite is carried out at lower temperatures. According to research, GB preferentially nucleates and grows at the prior austenite grain boundaries, where BF usually nucleates intra-granularly and grows inside retained space, GB transformation needs a higher thermodynamic driving force than BF and is carried out at higher temperature [24].

The matrix microstructure of the three samples was affected by the start cooling temperature. As for C1, due to the high start cooling temperature, the atomic exchange and diffusion process before rapid cooling can only be carried out in a limited time, so only a small amount of ferrite nucleated along the austenite grain boundaries, and distributed discontinuously. Untransformed austenite was unstable and bainite transformation was carried out at relatively high temperatures, and the formed bainite contained a large amount of GB and a small amount of BF. As for C3, on the contrary, the atoms had enough time to exchange and diffuse before rapidly cooling to low start cooling temperature. The amount and size of ferrite increased and the formed ferrite continuously distributed at the prior austenite grain boundaries. The untransformed austenite of C3 was more stable than that of C1, and bainite transformation was carried out in a lower temperature range. So, the formed bainite contained a larger amount of BF and a smaller amount of GB. For C2, the start cooling temperature was between and C1 and C3, so the content and size of the formed ferrite were moderate, resulting in moderate stability of the untransformed austenite. The amount of BF in C2 is between C1 and C3. According to the results, the illustration of the microstructure evolution of experimental steels with different cooling rates is schematically summarized, as shown in Figure 7. The low-temperature toughness and ductile-to-brittle behavior of steels are closely related to the microstructures which are affected by the start cooling temperatures. The relationship between microstructure characteristics and toughness will be discussed in the next section.

Figure 7全屏查看

Illustration of microstructure evolution

4.2 Effect of start cooling temperature on DBTT

HAGBs play an important role in the crack propagation process of steels [25-29]. It is demonstrated that the crack propagation can be influenced by the misorientation between adjacent grains. In general, the increase of misorientation means higher grain boundary energy. When the misorientation exceeds 15°, the energy of grain boundaries reaches the maximum value. In the impact test, the crack propagation will select a suitable path to achieve the minimum energy consumption. It is believed that the cracks can be deflected or be arrested by HAGBs. According to Figure 1, from C1 to C3, the proportion of HAGBs was 30.3%, 36.3% and 34.2% respectively, the effective grain size was 3.5, 2.8 and 3.0 μm respectively. C2 possessed the highest amount of HAGBs and the smallest effective grain size, which became one of the main reasons for its lowest DBTT value. On the contrary, C1 had the lowest amount of HAGBs and the largest effective grain size, contributing to its highest DBTT value.

The DBTT of the three samples is also closely related to the MA islands in the matrix [30]. MA islands are considered brittle phases embedded in the ferrite matrix and can affect impact toughness. The hardness and brittleness of MA constituents are significantly different from ferrite matrix. During impact tests, the soft matrix in steel undergoes plastic deformation first, leading to local stress concentration on or around the hard phases [31]. When the local stress concentration exceeds critical stress, nucleation and propagation of cracks from the MA islands may occur. DEVIS and KING [32] found that large size of MA would reduce the crack initiation stress and have serious degradation effects on toughness of steel. As reported before, with the decrease of start cooling temperature, the average size of MA islands was 1.27, 1.44 and 1.56 μm respectively. C3 had the largest MA islands, which reduced its critical stress for crack nucleation and deteriorated its low-temperature toughness.

Crystallographic orientation characteristics can also affect the toughness of steels [18, 33]. Since the {001} plane is favorable for the cleavage fracture of BCC metal, grains with large amounts of {001} planes parallel to fracture surfaces always result in large areas of cleavage fracture in low-temperature impact tests of experimental steels. On the contrary, {110}<111> slip system is considered to be the main cause of plastic deformation in BCC metal, and the {110} planes parallel to fracture surfaces contribute to the toughness of the steels. In this study, Charpy impact specimens were cut along TD, and the grains with <001> and <110> orientation components parallel to TD can represent the grains with {001} cleavage planes and {110} slip planes parallel to impact fracture surfaces. As shown in Figure 4, C2 possessed the highest content of <110> oriented grains and the lowest content of <001> oriented grains. The results were consistent with the fracture morphology variation shown in Figure 6. The texture components derived from the austenite texture after being deformed and recrystallized. The rotated cube {001}<110> component is mainly resulted from the cube {001}<100> of the recrystallized austenite texture and always makes the steel brittle [34]. In order to obtain an ideal toughness of steel, the intensity of {001}<110> components needs to be reduced in most cases. On the contrary, the {332}<113> component is the most desirable in the transformation texture components and contributes to the improvement of toughness [35]. The intensity of {332}<113> needs to be increased to improve the toughness of samples. Sample C1 possessed a high intensity of rotated cube {001}<110> and a low density of {332}<113>, which should be one of the reasons for the lowest toughness among three samples. Sample C2 was opposite to C1 that the intensity of rotated cube {001}<110> was low and the intensity of {332}<113> was high, contributing to the highest toughness of steel. C3 sample had a high intensity of rotated cube {001}<110> and {332}<113>, contributing to its toughness between C1 and C2.

4.3 Prediction model of DBTT

The fracture morphology changed from void coalescence to cleavage with reduced temperature in impact tests. DBTT variation is due to competition between plastic flow and cleavage fracture, and is defined as the point where yield stress is the same as cleavage fracture stress. Therefore, the value of DBTT can be determined by microstructural parameters that can influence cleavage fracture or yield stress. According to previous research, the value of DBTT (T_DBT) is related to some factors as below [36]:

(1)

where the , k₁, k₂, k₃, k₄ and k₅ are constants and is the mass fraction of Mi element; (V_Pearlite+V_M-A) is the volume fraction of pearlite and MA components in the microstructure; Δσ_y is the yield strength corresponding to precipitation and dislocation; D_avg is the average grain size of the microstructure; D_20pct/D_avg describes the heterogeneity of effective grain size, and D_20pct is the cut off value of grain size at 80% area fraction in the grain size distribution histogram; D_M-A is the average size of MA components.

According to previous studies, the yield strength of samples contributed to precipitation and dislocation can be calculated as below [36, 37]:

(2)

where Δσ_Dis is dislocation strengthening, and Δσ_Orown represents the precipitation strengthening; σ₀, Δσ_ss and Δσ_GB are the friction stress of ferritic matrix, the solid solution strengthening and the grain boundary strengthening, respectively.

can be calculated by Eq. (3) and represents the average size of ductile phase.

(3)

The microstructural parameters of steel in the equations are summarized in Table 4.

Table 4全屏查看

Values of the microstructural parameters determining DBTT

Sample		Δσy	Davg	D20pct/Davg	DM-A
C1	6.9	30	3.5	1.3	1.3
C2	7.1	29	2.8	1.4	1.4
C3	7.2	29	3.0	1.4	1.6

展开更多

According previous studies, the constants from k₁ to k₅ are selected. The Eq. (1) is then modified as follows [19, 38, 39]:

(4)

The contribution of chemical composition and microstructural factors to DBTT can be calculated and the results are shown in Figure 8. Effective grain size is the most important factor in determining the DBTT of experimental steel. MA island diameter and microstructural heterogeneity also play key roles in DBTT. From Figure 8, it can be found that the calculated value of DBTT was relatively in agreement with the experimental results for steel.

Figure 8全屏查看

Contribution of chemical composition and microstructural characteristics to DBTT by calculation