三维形态形成的铁素体与夹杂物在低碳钢.pdf

,Universityof ferrite formed at intragranular inclusions in low-carbon steelMaterials Characterization 52 (2004)A good combination of strength and toughness oflow-carbon steel welds is achieved by so-called acic-ular ferrite microstructure, consisting of small inter-weaving ferrite plates formed within austenite grains[1–6]. The term acicular is used frequently in weldswhere ferrite plates are relatively small and cover alarge proportion of the matrix. The interweavingvariant ferrite plates at intragranular inclusions,growth thereof, and nucleation of secondary platesat the surface of preexisting ferrite plates [2,6,7].Since ferrite transformation occurs fast and the mi-crostructure is often observed after the weld is cooledto ambient temperature, studies of morphology andgrowth behavior of ferrite crystals at early stages arerelatively scarce in low-carbon steel.In recent years, both computer hardware and soft-1. Introduction microstructure is attributed to nucleation of multi-were studied by serial sectioning and computer-aided visualization. The specimens taken from the weld deposit wereaustenitized and isothermally reacted for varying times. The length, width, and thickness of ferrite plates were measured from3D-reconstructed images. From the length-to-width ratio, the morphology of ferrite is likely to be a lath rather than a plate inthis alloy. Multivariant ferrite plates (or laths) were observed to be nucleated at inclusions and grow in the directions close toh110igwith habit planes near {111}g, probably keeping a fixed orientation relationship with austenite.D 2004 Elsevier Inc. All rights reserved.Keywords: Low-carbon steel; Acicular ferrite; Widmansta¨tten ferrite; Three-dimensional reconstruction; InclusionThree-dimensional (3D) morphology and growth behaviorReceived 4 November 2003; accepted 10 April 2004AbstractThree-dimensional morphologyassociation with inclusionsK.M. Wua,1, Y. InagawaaJapan Space Utilization Promotion CenterbFaculty of Engineering, IbarakicDepartment of Materials Science, Ibaraki1044-5803/$ - see front matter D 2004 Elsevier Inc. All rights reserved.doi:10.1016/j.matchar.2004.04.004* Corresponding author. Tel.: +81-294-38-5058; fax: +81-294-38-5226.E-mail address: enomotom@mx.ibaraki.ac.jp (M. Enomoto).1Present address: Department of Applied Physics, WuhanUniversity of Science and Technology, Wuhan 430081, PR China.2Present address: Dietec Inc., Mibu-cho, Shimotuga-yun,Tochigi 321-0215, Japan.of ferrite formed inin low-carbon steelb,2, M. Enomotoc,*Nishi-Waseda, Tokyo 169-8624, Japan, Hitachi 316-8511, JapanUniversity, Hitachi 316-8511, Japan121–127ware for image processing and three-dimensional (3D)visualization have developed rapidly. These develop-ments prompted the application of 3D reconstructionto opaque metallic microstructures for the past decade,as reviewed by Kral et al. [8]. The characterization ofmicrostructure in three-dimensions often plays anessential role in identification of transformation mech-anism [8–14] and quantitative analysis of microstruc-ture [15]. In this report, low-carbon steel specimenswere reacted for very short times to obtain partiallytransformed microstructure and the microstructure was3D-reconstructed to observe the morphology of ferriteplates formed within the austenite matrix. Then, thelength, width, and thickness of individual ferrite crys-tals were measured from 3D-reconstructed images andthe growth behavior of ferrite plates was discussed.2. Experimental procedureTable 1 shows the compositions of the base plate(SM490), low-carbon weld wire and specimens takenfrom the weld deposit. The welding was conductedunder a shielding gas of 100% CO2. It is seen that theconcentrations of carbon and substitutional elementswere all reduced after welding, while the oxygenconcentration is large. The specimens were first aus-tenitized at 1200 jC for 20 min and furnace-cooled toobtain a grain size as large as 80 Am. They weresubsequently austenitized at 1350 jC for 150 s undera purified argon atmosphere, rapidly cooled (at a rateof f50 jC/s) and isothermally held at 570 jC for 1Table 1Chemical composition of the base plate, weld wire, and specimen(mass%)CSiMnTiONSBase plate 0.16 0.39 1.42 – – – –Weld wire 0.11 1.25 2.20 0.26 – – –Weld deposit 0.078 0.90 1.57 0.03 0.0320 0.0050 0.0059K.M. Wu et al. / Materials Characterization 52 (2004) 121–127122Fig. 1. Optical micrographs of specimen isothermally held (a) at570 jC for 1 s and (b) at 570 jC for 5 s.Fig. 2. 3D-reconstructed image of ferrite plates in the specimenand 5 s. These heat treatments were conducted in thehot deformation simulator. The second austenitizationand isothermal holding was also conducted in a saltbath to achieve fast quenching to the isothermalholding temperature.For serial sectioning, specimens were polishedwith an automatic grinder–polisher. Polished surfaceswere lightly etched with 3% nital. Microhardnessindents were applied to mark the area of interest(f120C290 Am2). They were used as fiducial marksfor horizontal image alignment and calibration of theisothermally held at 570 jC for 1 s.depth of removal. Typically, 0.32F0.04 Amwasremoved per section. After image alignment, maskingof objective ferrite plate(s) was made. Then, a stack of2D images of the masked object was transformed intoobtained in the heat treatment using a salt bath, whichindicates that at least a major proportion of these ferritethe specimen reacted for 1 s. A total of 13 plates areobserved in this image. Although not seen in thefigure, an oxide inclusion was observed at the inter-section of plates (arrowed). Some of these plates werethus nucleated at the inclusions (often denoted asprimary plates) and the other plates, probably at thesurface of preexisting ferrite plates (secondary plates).3.2. Size of platesThe length, width, and thickness of individualplates were measured from their 3D images. TheK.M. Wu et al. / Materials Characterization 52 (2004) 121–127 123Fig. 3. Distribution of plate lengths measured from 3D-recon-structed images in the specimen reacted at (a) 1 and (b) 5 s and (c)measured on the polished surface.plates was formed during isothermal holding.Fig. 2 shows a 3D image of ferrite plates formed ina 3D image using AVS software. These procedureshave been described elsewhere [14].3. Results and discussion3.1. Three-dimensional image of ferrite plateFig. 1a and b shows the light optical micrographs ofspecimens reacted at 570 jC for 1 and 5 s, respective-ly. It is seen in Fig. 1a that thin long plates arenucleated in the matrix, presumably at oxide inclu-sions. At this stage, the appearance of ferrite plates isvery similar to intragranular ferrite plate in the earlierclassification system of ferrite morphology [16].InFig. 1b, the number of ferrite plates increased andplates are considerably thickened. The microstructureevolved to a typical acicular ferrite microstructure atprolonged holding. Very similar microstructures wereTable 2Data on plate size measured from 3D-reconstructed images and onthe polished surfaceHeld at 570 jC for 1 s(number of plates=36)Held at 570 jC for 5 s(number of plates=18)Mean Range Mean RangeLength (Am)3D 10.8F4.5 3.4–24.0 18.1F7.0 10.3–36.52D 8.3F4.0 1.4–19.4 – –Width (Am) 2.8F1.4 0.8–7.0 4.6F1.8 1.4–7.8Thickness (Am) 0.7F0.2 0.3–1.2 1.9F0.6 0.8– 2.9distribution of lengths is shown in Fig. 3a and b.polished surface. The data on plate size are summa-rized in Table 2. The increase in plate thickness withholding time is considerably large in this time range.The width and thickness are plotted against thelength for each plate in Fig. 4a and b, respectively. Aweak correlation is observed between the two sizesplotted in the figures. In Fig. 4a, a line of constantK.M. Wu et al. / Materials Characterization 52 (2004) 121–127124The number of measured plates was 36 and 18 forholding time of 1 and 5 s, respectively. Fig. 3c showsthe distribution of the length of 106 plates for holdingtime of 1 s measured on the polished surface. Thepeak length on a 2D section is smaller than thatobtained from 3D images, as naturally expected.The corresponding data on width and thickness werenot secured because they are not distinguished on thelength-to-width ratio is drawn. Whereas the ratio isnear unity for a plate, the average is 4.1, whichFig. 4. (a) Plot of width (W) against length of ferrite plates (L), and(b) plot of thickness against length of ferrite plates measured from3D-reconstructed images.Fig. 5. Comparison of (a) measured plate length with calculationfrom Zener-Hillert and Trivedi models and (b) measured platethickness with calculation with carbon diffusivity at the a/gboundary concentration (Da/g) and the averaged carbon diffusivityin austenite (Dav).indicates that the shape of ferrite is more often a laththan a plate in this alloy. Earlier, the 3D shape ofWidmansta¨tten ferrite plates was studied by polishingthe specimen on two orthogonal planes or along {111}planes of austenite in medium-carbon steel, asreviewed by Reynolds et al. [17]. In these studies, theshape of ferrite plates was considered to be a plate [18]or a lath [19,20], although some of the plates wereactually formed at a free surface of the specimen.Recently, Spanos and Kral [13] observed by 3D recon-struction a ‘‘spike’’ morphology of grain boundaryferrite precipitates in an Fe–0.12% C–3.28% Ni alloy.In Fig. 5a and b, the plate length and thicknessmeasured from 3D images were compared with calcu-lation from the theory of carbon diffusion-controlledgrowth. The lengthening rate was calculated fromHillert [21] and Trivedi [22] theories. The calculatedtip radius, which tends to be smaller than the observedtip radius [23], was used in the calculation. It is seenthat the measured length of plates is somewhat smallerthan calculation at 5 s. In Fig. 5b, the measured platethickness compares reasonably well with calculationunder the assumption that thickening occurred by themotion of planar disordered boundaries. Although thenumber of data presently obtained is too few to drawdetailed comparison, the measured plates sizes arebasically in agreement with calculation.3.3. Habit plane and growth direction of ferriteOn the polished surface, the ferrite plates in Fig. 2may appear as a ‘‘Widmansta¨tten star’’ earlier notedisK.M. Wu et al. / Materials Characterization 52 (2004) 121–127 125Fig. 6. (a) Rotated view of Fig. 2, indicating that the growth directionindicating that the broad face of plates is near {111} of austenite.near h110i of austenite. (b) Enlarged view of plates 8, 9, and 11,by Aaronson and Wells [7]. The image was rotatedsuch that the broad face of plate 7 becomes perpen-dicular to the plane of the paper, as shown in Fig. 6.From the angles between the plates, it appears thattinguish sympathetic nucleation from impingement orintersection solely from the appearance on the pol-ished surface. Since nucleation may be suppressed atlater stages, impingement and intersection may largelybe responsible for the formation of characteristicinterlocking microstructure of acicular ferrite at pro-longed holding.4. SummaryThe 3D morphology and growth behavior of acic-ular ferrite at early growth stages in low-carbon steelwere studied by serial sectioning and computer-aided3D reconstruction. The specimen taken from the welddeposit was austenitized and isothermally reacted at570 jC for 1 and 5 s. Ferrite plates were observed tobe nucleated at intragranular inclusions, presumablyon oxides, and grew radially along specific directionsin the matrix austenite, thus forming an arrangementof ferrite crystals noted as Widmansta¨tten star onpolished surfaces [7]. The length, width, and thicknessK.M. Wu et al. / Materials Characterization 52 (2004) 121–127126they grew preferentially along h110igdirections. Thelongest direction can be indexed as [11¯0], [011], and[101], for plates 2 and 5, plates 4 and 11, and plates 6and 7, respectively. The growth direction of plates8 and 12 is the same as plate 2, and that of plate 13 isthe same as plate 4. Plate 1 (or 3) may have beenformed on the broad face of plate 2 (or 4), possibly byedge-to-face sympathetic nucleation [24]. Whereasthe longest direction of plates 1 (or 3) and 2 (or 4)is the same, the growth direction of plate 1 (or 3) isnot clear.3The growth direction of plate 9 is notidentified, either, and no plausible h110igvariantseems to exist for plate 10. These observations agreewith earlier observations that the growth direction ofgrain boundary Widmansta¨tten sideplates is parallel tothe h110i directions in the austenite matrix [18].Fig. 6b is an enlarged view of plates 8, 9, and 11from a different angle. The broad faces of these platesare apparently close to {111}gplanes. Although thebroad face of some plates is not well defined,assuming that the plate whose growth direction ish110igis lying in {111}gplanes, all plates exceptplate 10 appear to be lying on one of four {111}planes of austenite. According to earlier reports,however, a significant amount of deviation from{111}gplanes was reported [25,26]. It is noted that,even when the habit plane deviates considerably fromthe close-packed planes, ferrite plates are consideredto have either Kurdjumov–Sachs or Nishiyama–Wassermann orientation relationship within a fewdegrees [17].In specimens reacted for 5 s, many plates are incontact with other plates. The 3D-reconstructedimages of two examples are shown in Figs. 7a andb.InFig. 7a, plate A was presumably nucleated at aninclusion and impinged on plate B during growth. InFig. 7b, two plates growing in nearly perpendiculardirections intersect with each other. These platesmight have impinged at an early stage or came into3In Fig. 6a, plate 2 appears to consist of many fine plates. Thisis an artifact presumably because the depth of removal in one polishwas large for the angle of the plate to the specimen surface.contact at a late stage as they thickened. In both cases,the boundary between the two plates was not clearlydelineated by etching. Thus, one cannot always dis-Fig. 7. (a) 3D-reconstructed image of plate A presumably impingingon plate B. The red small circle is an inclusion. (b) 3D image ofintersecting plates A and B. In (a) and (b), the specimen was reactedat 570 jC for 5 s.of ferrite plates were measured from 3D-reconstructedimages. From the distribution of aspect ratio, theunder the auspices of the Japan Space Utilization1057–68.K.M. Wu et al. / Materials Characterization 52 (2004) 121–127 127[3] Grong O, Matlock DK. Microstructural development inmild and low-alloy steel weld metals. Int Met Rev 1986;31:27–48.[4] Bhadeshia HKDH. Modelling the microstructure in the fusionzone of steel weld deposits: keynote. 2nd International Con-ference on Trends in Welding Research, Gatlinburg (TN)1989;189–98.[5] Bhadeshia HKDH. Bainite in steels. London (UK): Institute ofMaterials; 1992.[6] Ricks RA, Howell PR, Barritte GS. The nature of acicularferrite in HSLA steel weld metals. J Mater Sci 1982;17:732–40.[7] Aaronson