J Mater Sci DOI 10.

1007/s10853-012-6346-x

Analysis of martensiteaustenite constituent and its effect on toughness in submerged arc welded joint of low carbon bainitic steel
Liangyun Lan Chunlin Qiu Dewen Zhao Xiuhua Gao Linxiu Du

Received: 2 October 2011 / Accepted: 13 February 2012 Springer Science+Business Media, LLC 2012

Abstract Martensiteaustenite (MA) constituent formed during welding is generally recognized as an important factor to decrease the toughness of welded joint. In this article, the morphology and chemical composition of MA constituent in the low carbon bainitic steel welded joint was analysed in detail by means of optical microscope, transmission electron microscope and scanning electron microscope with electron probe microanalysis. The experimental results show that the MA constituent formed in the different sub-zones presents different morphologies and different amounts. The maximum amount of MA constituent occurs in the coarse grained heat affected zone (HAZ). It is evident that the carbon atoms segregate on the MA constituent and carbon concentration on the slender MA constituent is higher than that on the massive MA constituent. Meanwhile, the distribution prole of silicon on the MA constituent shows an obvious inhomogeneity. Most of MA constituents have a twinned structure and/or a high dislocation density. According to impact testing results, the crack initiation energy in the HAZ specimens deteriorates signicantly because the large MA constituent can assist the formation of cleavage crack. On the other hand, the coarse prior austenite grain in the HAZ lowers the crack propagation energy.

Introduction In recent decades, the submerged arc welding technique has been widely applied in the welding eld of high strength

L. Lan (&) C. Qiu D. Zhao X. Gao L. Du State Key Laboratory of Rolling and Automation, Northeastern University, P.O. Box 105, No.11, Lane 3, Wenhua Road, HePing District, Shenyang 110819, Peoples Republic of China e-mail: [email protected]

structural steel, e.g., pipeline, pressure vessels, box columns, offshore structure [1, 2]. However, although the base metal (BM) has an excellent combination of high strength and toughness, the deterioration phenomenon of toughness of welded joint always occurs after welding, especially for large heat input processing. It is generally accepted that the deterioration in toughness of welded joint has a close relationship with microstructural characteristics, such as martensiteaustenite (MA) constituents, non-metallic inclusions, prior austenite grain size as well as the fraction of acicular ferrite in the matrix microstructure [18]. Considerable research has focused on the deleterious effect of MA constituent on simulated heat affected zone (HAZ) toughness. For example, Davis and King [5] have shown that cleavage crack initiation mainly depends on the distribution and morphology of MA constituent in the simulated intercritically reheated HAZ (ICHAZ) and concluded several possible micromechanisms of MA constituent causing a reduction in HAZ toughness. Li et al. [6] have shown that the ICHAZ toughness does not deteriorate drastically for the API X70 pipeline steels because the size of MA constituent is small and the content of MA constituent is relatively little for the effect of accelerating cooling. The alloying addition (e.g., Nb) had a signicant effect on the amount and size of the MA constituent and those constituents further deteriorated the toughness of simulated ICHAZ [7]. In addition, Moeinifar et al. [8] have studied the variation of MA particles in the ICHAZ by means of Gleeble simulation technique and real tandem submerged arc welding and concluded that the fraction and morphology of MA constituent as well as the prior austenite grain size are important factors on the Charpy impact energy in the ICHAZ. As stated above, it is evident that the MA constituents in the welded joint play an important role in decreasing the toughness, especially for crack initiation energy. Most of these

123

J Mater Sci Table 1 Chemical compositions and as-rolled mechanical properties of the low carbon bainitic steel Chemical compositions (wt%) C 0.053 Si 0.27 Mn 1.35 Nb ? V ? Ti 0.082 B (ppm) 12 Others Cr, Mo, Ni Pcm 0.185 Mechanical properties Rt0.5/(Mpa) 640 Rm/(Mpa) 760 A (%) 18 Toughness (-20 C) 161 J

studies were carried out by the welding thermal cycle simulation technique because it can be used to generate a relatively large region for special sub-zone [7]. However, this technique also has its own limitations, such as single simulation microstructure cannot reect completely the continuous variation of real HAZ microstructure and the weld metal (WM) columnar microstructure is hard to be simulated because it cools directly from the liquid [9]. The characters of MA constituent in different sub-zones of the real welded joint must be different because the welded joint is subjected to uneven welding thermal cycle. Therefore, the MA constituent should be studied in detail to further understand its effect on the cleavage fracture micromechanism. Meanwhile, the formation of MA constituent is associated with the segregation of many elements, such as carbon and silicon [10]. These segregated elements may have different distribution proles on the MA constituents formed in different sub-zones, which, in turn, may inuence the amounts and morphologies of MA constituents. However, very few studies have focused on the microsegregation of composition on the MA constituent. Therefore, in this study, more attention has been paid to the morphology and chemical composition segregation of MA constituent in different sub-zones of real welded joint. In addition, the effect of MA constituent on the impact toughness has been also discussed to further understand the cleavage fracture micromechanism in the welded joint.

thermo-mechanical control process (TMCP). The chemical compositions and mechanical properties of this steel are shown in Table 1. To protect the welded joint from welding cold crack, the carbon content and welding crack susceptibility index Pcm designed are 0.053 and 0.183%, respectively, and Pcm was given by Eq. 1 [11]. Si Mn Cu Cr Ni Mo V + 5B 30 20 60 15 10 1 According to the standard of EN ISO 9692 [12], a single V-type butt weld groove was designed for the experimental steel plate with size of 20 (thickness) 9 100 (width) 9 400 (length) mm which was cut along the longitudinal direction of hot rolled steel. Four-pass welding procedure was conducted to generate a relatively wide HAZ region. The main welding parameters were: current 600 A, voltage 30 V, welding speed 40 cm/min, inter-pass temperature range 180200 C and the corresponding heat input was about 27 kJ/cm. Low alloy steel electrodes (H08MnMoNiA) with 4.0 mm in diameter were used for submerged arc welding according to Chinese welding standard of GB/T 12470 [13]. An agglomerated commercial ux was selected which contained Al2O3, SiO2, MnO, CaO, CaF2, TiO2. After welding, impact specimens with standard size were prepared along the transversal direction of welding passes. In order to x the right position of Charpy V notch, the impact specimens were macro-etched with 3% nital rst to exhibit the outline of HAZ. According to ISO 9016 standard [14], the Charpy V notch position was xed at different regions of welded joint to examine the toughness of these regions as shown in Fig. 1. Because of the narrow Pcm C

Experimental procedure The low carbon microalloyed steel with the thickness of 20 mm was produced in the as-rolled condition by

(a)

(b)

WM

HAZ

BM

WM

HAZ

BM

Fig. 1 Schematic diagram of Charpy V notch positions at the WM (a) and HAZ (b)

123

J Mater Sci

width of sub-zones in HAZ, the Charpy V notch in the HAZ was across the whole of coarse grained HAZ (CGHAZ) and a part of ne grained HAZ (FGHAZ). The impact testing was carried out at -20 C and each region was tested at least ve times. An instrumented drop-weight impact tester was employed to record the impact loaddeection curves. The impact absorbed energy can be divided into crack initiation energy and crack propagation energy based on these impact curves [15, 16]. The metallographic specimen was cut from the welded plate on transverse section with respect to the welding direction, which was parallel to the rolling direction. The polished specimen was etched with 3% nital for optical microscope and scanning electron microscope (SEM) observation. To identify the morphology and amount of MA constituent in different sub-zones of HAZ, the specimen was repolished and etched with LePera reagent. The segregation of composition on the MA constituent was examined using electron probe micro-analysis (EPMA) attached to a eldemission SEM. Thin-foil specimens for transmission electron microscope (TEM) were prepared to observe the renement microstructure at different regions in welded joint. Fracture surface morphology of impact specimens were also examined with the SEM. The statistical analysis on the amount and size of MA constituents was conducted using an image analyzer based on optical and SEM micrographs.

CGHAZ has lath or granular bainite with large prior austenite grain size (Fig. 2e).The maximum size of prior austenite grain measured can reach about 300 lm. Acicular ferrite and proeutectoid grain boundary ferrite are present at regular intervals in the WM region and some of inclusions are signied by arrows (Fig. 2f). The WM microstructure begins to solidify by epitaxial growth of columnar delta-ferrite from the parent plate at the fusion line, and then the austenite grains form during further cooling. The grain boundary ferrite forms at the columnar austenite grain boundaries by a carbon diffusion mechanism and subsequently, acicular ferrite nucleates and grows with a displacement mechanism attached to granular inclusions inside the columnar austenite grain [17]. TEM study of detailed microstructures in different sub-zones The detailed morphology of microstructures in welded joint studied by TEM is presented in Fig. 3. Figure 3a shows that acicular ferrite plates in the WM are arranged in an interlocking pattern due to hard impingement and some acicular ferrite plates seem to nucleate on the granular inclusions arrowed in Fig. 3a. These granular inclusions have the characteristics of complex oxides as they are composed of many elements including Ti, Al, Mn, Si, O, and S shown by the EDS in Fig. 3b. As these inclusions can greatly inuence the formation of acicular ferrite which contributes to enhancing toughness due to its high angle boundaries, many researchers have concentrated on the role of inclusions in stimulating the nucleation of acicular ferrite [3, 4, 18]. Thewlis et al. [18] proposed that the amount of acicular ferrite transformed mainly depends on the inclusion type in WM because an absence of inclusions TiN or MnS can lead to reduce acicular ferrite. Fig. 3c reveals the coarse parallel laths of bainite structure in the CGHAZ, accompanied with many small MA constituents distributing among the bainite lath boundaries. These MA constituents are further discussed later. The bainite laths contain a high dislocation density mainly resulting from uneven phase transformation by a shear mechanism [19]. The TEM micrographs in FGHAZ consisting of polygonal ferrite (or massive ferrite) with small grain size are presented in Fig. 3d. The polygonal ferrite has curved irregular grain boundaries and relatively lower dislocation density because of microstructure occurring completely recrystallization in the FGHAZ. Morphology, micro-composition, and structure of MA constituents in different sub-zones After the metallurgical specimen was etched with the LePera reagent for about 20 s, the MA constituents appear

Results and discussion Microstructural characteristics in different sub-zones Figure 2 shows the microstructural characteristics at different regions in the welded joint after subjected to multipass welding process. The overall microstructure at the low magnication condition is shown in Fig. 2a, which contains ve different sub-regions: BM, ICHAZ, FGHAZ, CGHAZ, and WM designied with number square grids. The corresponding higher magnication microstructures are showed in the following gures from Fig. 2bf. The BM microstructure consists mainly of granular bainite and a small amount of quasi-polygonal ferrite with the dispersive distribution of second phase (Fig. 2b) due to subjected to low welding temperature (peak temperature below Ac1). The ICHAZ has a mixture of untransformed granular bainite and quasi-polygonal ferrite as well as retransformed martensite islands or MA constituents which form at the grain boundaries (Fig. 2c). The microstructure of FGHAZ is predominantly quasi-polygonal ferrite with very small grain size (Fig. 2d), mainly due to complete reaustenitisation at relatively low peak temperatures and for short periods of time. The mean grain size in FGHAZ is about 5.4 lm roughly measured by the line intercept method. The

123

J Mater Sci

Fig. 2 Optical micrographs showing the microstructures in different regions of the welded joint a low magnication overall microstructure, b BM (or tempered zone), c ICHAZ, d FGHAZ, e CGHAZ,

f WM (GB granular bainite, QPF quasi-polygonal ferrite, M martensite, MA martensiteaustenite constituent, PAGB prior austenite grain boundary, AF acicular ferrite, GF grain boundary ferrite)

white, while inclusions and carbides are black, as shown in Fig. 4. In the WM (Fig. 4a), only a few MA constituents have a relatively large size (maximum length more than 2 lm) and most MA constituents have a very small size with dot shape (less than 1 lm). Because of the formation of acicular ferrite by a shear mechanism, carbon enrichment always occurs at the boundaries of acicular ferrite. Therefore, most of the MA constituents are distributed

along the acicular ferrite boundaries. As shown in Fig. 4b, many MA constituents in the CGHAZ have a conspicuous character of slender shape with the high aspect ratio (the ratio of maximum length to width). The maximum length of MA constituent measured can reach about 25 lm. Those slender MA constituents delineate the bainite lath boundaries inside the large prior austenite grain. Several large massive MA constituents as well as many relatively

123

J Mater Sci Fig. 3 TEM micrographs showing renement microstructures at different regions in the welded joint a WM, b chemical composition of inclusions in the WM, c CGHAZ, d FGHAZ

small irregular MA constituents are represented in the FGHAZ, as shown in Fig. 4c. It can be found that the irregular MA constituents are also prone to form along the boundaries of massive ferrite because the formation of massive ferrite is controlled by carbon diffusion. Optical micrographs were processed with an image analyzer as per the method proposed by Bonnevie et al. [20]. All the micrographs were transformed into binary images to calculate the area fraction of MA constituent easily. As shown in Fig. 4d, the average area fraction of MA constituent in the WM, CGHAZ and FGHAZ is about 0.8 0.3, 4.0 0.9, and 3.8 0.7%, respectively. It is evident that the amount and morphology of the MA constituent formed in different sub-zones is directly determined by welding thermal cycle experienced. Moreover, the formation of MA constituent is also inseparable from segregation of chemical composition in these micro-zones. Some researches have shown that some micro-elements added in steel can obviously enhance the amount of MA formation [7, 10, 20]. Therefore, micro-elements segregated on the MA constituent were analyzed with EMPA and discussed in the following paragraph.

The EPMA line scanning method can be used for qualitative and semi-quantitative analysis of chemical composition segregated on the MA constituent as shown in Fig. 5. Figure 5a shows the SEM microstructure in the WM at high magnication. The scanning line signied with a black arrow line in Fig. 5a is across three acicular ferrite grains and two small MA constituents and the corresponding elemental distribution proles are shown in Fig. 5b, in which the relative counts of the element represent qualitatively the concentration distribution of element along the scanning line. It can be seen that the carbon concentration in the MA constituent is at least twice that in the acicular ferrite matrix. On the contrary, the segregation of silicon in the MA constituents seems to be controlled by a short range diffusion mechanism, which shows that the peak concentration of silicon occurs in the MA constituent near the one side of constituent boundary and the lowest concentration appears at the other side of the MA constituent/matrix interface, as shown in Fig. 5b. Meanwhile, the concentration of substitutional elements, such as Mn, Ni, does not exhibit any obvious change on the MA constituents and their distribution proles are not

123

J Mater Sci Fig. 4 Morphology and area fraction of MA constituent in different sub-zones a WM, b CGHAZ, c FGHAZ, d area fraction of MA constituent

(a)

(b)

(c)

(d) 6
M-A constituent area fraction (%)
5 4 3 2 1 0 WM CGHAZ FGHAZ

Sub-zone

given. Figure 5c shows the slender MA constituents in the lath bainite matrix of the CGHAZ and corresponding carbon and silicon concentration distributions along the scanning line are shown in the Fig. 5d. It can be found that the carbon concentration is higher and the larger concentration gradient of silicon is still presented on the slender MA constituents compared with the massive MA constituent. The SEM micrograph and corresponding carbon and silicon distribution proles from the massive MA constituents in the polygonal ferrite matrix of the FGHAZ are shown in Fig. 5e and f, respectively. A similar concentration distribution trend of carbon and silicon can be obtained from the FGHAZ massive MA constituent, the carbon concentration at the periphery of the massive MA constituent seems to be higher than that in its middle region and the silicon concentration in the center of the massive MA constituent equals the matrix silicon concentration. The occurrence of carbon segregation is a necessity to the formation of MA constituent because carbon enrichment in the residual austenite at the last stage of phase transformation can notably enhance the stability of residual austenite and decrease the marstenite start temperature (Ms) at these local zones. Biss and Cryderman [10] have reported that the high carbon peaks on the MA constituent often exceeded 0.5 wt% and occasionally exceeded 1 wt%. Santomia et al. [21] have proposed that the carbon concentration ranged from value close to zero in ferrite to up to 1.2 wt% in the MA constituents after the low carbon steel was subjected to intercritical annealing treatment.

According to the EMPA line scan analysis, after welding, carbon is clearly concentrated on the MA constituent compared with the matrix. The concentration of carbon in the slender MA constituent seems to be higher than that in the massive MA constituent. Meanwhile, the carbon distribution prole at the massive MA constituents in the FGHAZ shows that the carbon concentration at the periphery is usually higher than that in the center. As regards the silicon distribution of the MA constituent, because of the lower diffusion coefcient of silicon compared to carbon atoms, the segregation of silicon on the MA constituent is not as obvious as the carbon segregation. However, there is a repulsive force between silicon and carbon atoms. High silicon content in ferrite can promote the carbon enrichment of the MA constituent. On the other hand, silicon segregated at the interface of untransformed austenite can greatly retard the formation of carbides in austenite as well as in ferrite [22]. Therefore, the increase of silicon content in steel can slow down the decomposition of retained austenite and increase the amount of MA constituent [20, 22]. According to the EPMA line scan results, it can be found that the carbon and silicon atoms are concentrated in the MA constituent in the welded joint. This is consistent with the results of Pereloma et al. [23]. They have studied the distributions of carbon and silicon in the various matrix phases of TRIP steels by using three dimensional atom probe analysis and concluded that the highest concentrations of carbon and silicon occur on the residual austenite phase.

123

J Mater Sci Fig. 5 SEM micrographs showing morphology of MA constituent and corresponding EMPA analysis of element segregated on the MA constituent in different subzones a WM microstructure, b corresponding carbon and silicon distribution proles in the WM, c CGHAZ microstructure, d corresponding carbon and silicon distribution proles in the CGHAZ, e FGHAZ microstructure, f corresponding carbon and silicon distribution proles in the FGHAZ
Si relative counts
500 400 300 200 100 500 400 300 200 100 0 2 4 6 8 10 12 0 2 4 6 8 10 12

(a)

(b)

C relative counts

AF

M-A

AF

M-A

AF

Distance (m)
C relative counts Si relative counts

(c)

400 300 200 100 500 400 300 200 100

(d)

10

12

14

LBF

M-A

LBF

M-A

10

12

14

Distance (m)

C relative counts Si relative counts

(e)

400 300 200 100 500 400 300 200 100

(f)

10

12

14

PF

M-A

PF

10

12

14

Distance ( m)

As stated above, the MA constituents formed at different sub-zones have some differences in the distribution of segregated elements and morphology, which may reect the wide variety of MA constituents in nature. Some researchers have classied the MA constituent in low carbon steels into three groups [20, 24]: one group is fully martensite constituents with microtwins and/or high dislocation density, the second group is fully retained austenite with pile-up defects, and the third group is mixed constituents associating with austenite and martensite. In the mixed MA constituents, austenite is distributed at the periphery, while martensite is located at the center reported by Lambert et al. [24], which seems to correspond to the carbon gradient on the massive MA constituents in the FGHAZ because carbon content is higher at the periphery of the massive MA constituents (Fig. 5f). However, it is generally accepted that fully twin martensite is often obtained in the HAZ not only

because of inhomogeneity of carbon concentration in the austenite grains but also due to the various cooling rates and thermal stresses in the HAZ [25]. According to the selected area diffraction (SAD) analysis as shown in the Fig. 6, the typical fully martensite phase occurred in the CGHAZ and the cycled area showed the microzone of SAD analysis (Fig. 6a). The corresponding SAD pattern, with a [113] zone axis, showed the twinning relationship between the parallel martensite laths and the matrix (Fig. 6b). These transformation twin martensites can relieve high thermal, solidication induced and transformation induced inside stresses [25]. Meanwhile, the amount of MA constituent existing in the form of fully retained austenite should be very little mainly because of low carbon content in this steel. Thus, most MA constituents in the welded joint have a character of twinning structure and/or high dislocation density [8, 25].

123

J Mater Sci

Fig. 6 TEM micrograph showing a twinning structure of MA constituent in the CGHAZ a the morphology of MA constituent with typical micro-twins signied by arrows and a twin region cycled for SAD analysis, b corresponding SAD pattern of twinned martensite from (a)

Toughness and fracture morphologies in different subzones of welded joint Impact loaddeection curves as well as impact absorbed energy-deection curves can be recorded by means of the instrumented Charpy impact tester [15, 16, 26]. Figure 7 shows the typical instance of impact curves of the HAZ impact specimen. It can be seen that the impact load decreases sharply after it reach the peak value and the total impact absorbed energy is about 58 J. Fig. 8 shows that the toughness of the welded joint deteriorated signicantly compared with the toughness of the BM specimens. Both the crack initiation energy and the crack propagation energy in the welded joint decrease simultaneously. The lowest values of crack initiation and propagation energies are obtained for the HAZ specimens. In general, the amount of acicular ferrite and inclusions in the WM are mainly responsible for the WM toughness.
20 60

High volume fraction of acicular ferrite in the WM matrix can obviously enhance the impact toughness because of its high angle grain boundaries and interlocked arrangement in nature as shown in Fig. 3a. However, the amount of acicular ferrite formed in the WM can be inuenced by various factors, e.g., welding parameters, alloying elements, and inclusions [1, 3, 17, 18, 27]. Viano et al. [1] concluded that the volume fraction of acicular ferrite decreased as the heat input increased and heat input also had a signicant effect of the acicular ferrite lath width. Thus, high heat input adversely affected the impact toughness. Under the present welding parameters, the acicular ferrite as well as grain boundary ferrite can be formed simultaneously in the WM. The corresponding fracture surface is shown in Fig. 9a. It can be seen that the WM fracture surface exhibits the typical quasi-cleavage fracture features which contains a mixed morphologies of ductile and brittle fracture. Figure 9b shows magnied ductile fracture morphology
180 160 140 30

Peak Load
50 15 40

crack propagation energy crack initiation energy peak load

25

Load (KN)

120

20

10

30

100 15 80 60 40 10

20 5 10

5 20

0 0 2 4 6 8 10

BM

WM

HAZ

Deflection (mm)

Specimens

Fig. 7 Typical curves of impact loaddeection and impact absorbed energy-deection of the HAZ specimen

Fig. 8 Average impact absorbed energy and peak load of specimens with different Charpy V notch positions

123

Peak load (KN)

Energy (J)

Energy (J)

J Mater Sci

Fig. 9 SEM micrographs showing the fracture surface morphology of specimens with different Charpy V notch positions a WM fracture surface morphology, b magnied ductile fracture morphology

corresponding to the square grid region in (a). c CGHAZ cleavage fracture morphology in the HAZ specimen, and d FGHAZ cleavage fracture morphology in the HAZ specimen

corresponding to the square grid region in Fig. 9a. According to fracture surface features, there are many inclusions inside the dimples and on the cleavage facets, which indicate that both microvoid coalescence and cleavage fracture are strongly inuenced by the nonmetallic inclusions in the WM. Tweed et al. [27] suggested that cracked inclusions above a certain size (about 1 lm) lead to the initiation of ferrite cleavage which propagates preferentially through grain boundary ferrite in the WM. Avazkonandeh-Gharavol et al. [17] proposed similar conclusions, i.e., that if inclusions do not provide efcient sites for nucleation of acicular ferrite, they are detrimental to toughness especially for inclusions greater than 1 lm in diameter. The MA constituent is also an important factor in inuencing the toughness in the WM. However, the small size and low volume fraction of MA constituents in the WM as shown in Fig. 4 may reduce their deleterious effect. According to impact results as shown in Fig. 8, the HAZ toughness deteriorated signicantly compared to the BM toughness. It is generally recognized that the MA constituents formed in the HAZ have a signicant negative effect on the toughness of the welded joint [58, 17, 23,

24]. The fractions of MA constituent formed in the CGHAZ and FGHAZ are about 4.0 and 3.8%, respectively. Most of these MA constituents have a twinned martensitic structure with high contents of carbon and silicon, as shown by the EPMA and TEM analyses. Thus, the hardness of MA constituents is very high compared with the matrix. When an impact load is applied to the welded joint containing MA constituents having a large size, an excessive internal stress will be generated in the matrix near the MA constituent because of the concentration of plastic strains resulting from the mismatching of their hardness [28]. The concentration of plastic strains near the MA constituent plays an important role in the initiation of cleavage crack. The fracture surface morphologies of the CGHAZ and FGHAZ in the HAZ specimen are shown in Fig. 9c and d, respectively. It can be seen that the cleavage crack initiates near the MA constituents attached to prior austenite boundaries arrowed in the Fig. 9c. Because of the large prior austenite grain size in the CGHAZ, the unit cleavage facet with a large size present on the fracture surface contributes to lowering the crack propagation energy. The cleavage facets in the FGHAZ seem to initiate near the MA constituent along the grain boundaries,

123

J Mater Sci

although the size of unit cleavage facet is rather small (Fig. 9d). However, the lowest crack initiation energy for the HAZ specimens is mainly attributed to large MA constituents in the CGHAZ, because the cleavage fracture micromechanism of coarse bainite is largely controlled by the crack initiation stage according to previous research [26]. In addition to the high volume fraction of MA constituent which can signicantly lower the toughness, the morphology of MA constituent is also an important factor inuencing the toughness. Davis and King [5] reported the lowest impact toughness of simulated ICHAZ was associated with a particular microstructure containing a nearconnected grain boundary network of blocky MA particles. Due to the MA particles being signicantly harder than the internal grain microstructure, the crack initiates easily between two closely spaced MA constituents. From a fractographic examination, Li and Baker [29] concluded that the elongated stringer MA particles were more likely to debond from the matrix while the blocky MA particles were more likely to crack. Moreover, Moeinifar et al. [30] who studied the microstructure evolution of welding HAZ with different cooling rates by the multiple tandem submerged arc welding process and thermal simulated welding technique, concluded that the size of MA constituents, such as the mean diameter and length, is an important factor inuencing Charpy impact properties of CGHAZ and the microcrack nucleation may occur from MA particles at the intersection of prior austenite grain boundaries. The large slender shape MA constituents in the CGHAZ distributing along the prior austenite grain boundaries or lath packet boundaries can notably promote the formation of microcrack (Fig. 9c). The MA constituents with massive shape in the FGHAZ can also assist the initiation of cleavage crack (Fig. 9d). Therefore, the crack initiation energy of HAZ samples is obviously lower than that of BM. On the other hand, the large prior austenite grain size can greatly decrease the crack propagation energy because of the increasing unit cleavage crack facet size.

(2)

(3)

(4)

Carbon concentration on the slender MA constituent is higher than that on the massive MA constituent and the carbon content at the periphery of massive MA constituent is slightly higher than that in the center of massive MA constituent. Meanwhile, the distribution prole of silicon on the MA constituent shows an obvious inhomogeneity. Most MA constituents formed in the welded joint are fully martensite constituents with a twinned structure and/or a high dislocation density. The MA constituents with large size can greatly decrease the crack initiation energy in the HAZ because they can stimulate the formation of cleavage crack, while the deterioration effect of MA constituent on the WM toughness may weaken because of their small size.

Acknowledgements The present study is supported by the National Natural Science Foundation of China (No. 51074052) and the Fundamental Research Funds for the Central Universities (N100607001).

References
1. Viano DM, Ahmed NU, Schumann GO (2000) Sci Technol Weld Join 5(1):26 2. Kojima A, Yoshii K, Hada T, Saeki O, Ichikawa K, Yoshida Y, Shimura Y, Azuma K (2004) Nippon Steel Tech Report 90:39 3. Koseki T, Thewlis G (2005) Mater Sci Technol 21:867 4. Ricks RA, Howell PR, Barritte GS (1982) J Mater Sci 17:730. doi:10.1007/BF00540369 5. Davis CL, King JE (1994) Metall Mater Trans A 25:563 6. Li C, Wang Y, Chen Y (2011) J Mater Sci 46:6424. doi: 10.1007/s10853-011-5592-7 7. Li Y, Crowther DN, Green MJW, Mitchell PS, Baker TN (2001) ISIJ Int 41:46 8. Moeinifar S, Kokabi AH, Hosseini HRM (2011) Mater Des 32:869 9. Harrrison PL, Farrar RA (1989) Int Mater Rev 34:35 10. Biss V, Cryderman RL (1970) Metall Trans 2:2267 11. De Meester B (1997) ISIJ Int 37:537 12. EN ISO 9692-2 (1998) Welding and allied processes-joint preparation: Part 2: submergered arc welding of steels 13. GB/T 12470 (2003) Low-alloy steel electrodes and uxes for submerged arc welding (Chinese Standard) 14. ISO 9016 (2001) Destructive tests on welds in metallic materialsimpact tests-test specimen location, notch orientation and examination 15. Wiesner CS (1996) Int J Pres Ves Pip 69:185 16. Cvetkovski S, Adziev T, Adziev G, Sedmak A (2002) Eur Struct Integr Soc 30:95 17. Avazkonandeh-Gharavol MH, Haddad-Sabzevar M, Haerian A (2009) J Mater Sci 44:1902. doi:10.1007/s10853-009-4141-0 18. Thewlis G, Chao WT, Harrison PL, Rose AJ (2008) Mater Sci Technol 24:771 19. Bhadeshia HKDH, Christian JW (1990) Metall Trans A 21A:767 20. Bonnevie E, Ferriere G, Ikhlef A, Kaplan D, Orain JM (2004) Mater Sci Eng A 385:352 21. Santomia MJ, Kwakernaak C, Sloof WG, Zhao L, Sietsma J (2010) Mater Charact 61:937

Conclusions (1) After the welded joint was subjected to multi-pass submerged arc welding process, most MA constituents formed in the WM have a small size and are always distributed along the acicular ferrite boundaries. On the contrary, quite a number of MA constituents exhibit long slender shape in the CGHAZ, which delineates the bainitic ferrite lath boundaries inside the large prior austenite grains. In the FGHAZ, some MA constituents with large massive shape are distributed in the polygonal ferrite matrix.

123

J Mater Sci 22. Taillard R, Verrier P, Maurickx T, Foct J (1995) Metall Mater Trans A 26A:447 23. Pereloma EV, Timokhina IB, Miller MK, Hodgson PD (2007) Acta Mater 55:2587 24. Lambert A, Drillet J, Gourgues AF, Sturel T, Pineau A (2000) Sci Technol Weld Join 5:168 25. Poorhaydari K, Patchett BM, Ivey DG (2006) Mater Sci Eng A 435436:371 26. Lan L, Qiu C, Zhao W, Gao X, Du L (2011) Mater Sci Eng A 529:192 27. Tweed JH, Knott JF (1987) Acta Metall 35:1401 28. Matsuda F, Ikeuchi K, Fukada Y, Horii Y, Okada H, Shiwaku T, Shiga C, Suzuki S (1995) Trans JWRI 24:1 29. Li Y, Baker TN (2010) Mater Sci Technol 26:1029 30. Moeinifar S, Kokabi AH, Madaah Hosseini HR (2011) J Mater Process Technol 211:368

123