In situ SEM-EBSD observations of the hcp to bcc phase ...

G.G.E. Seward et al. / Acta Materialia 52 (2004) 821–832 829transformation, the interface velocities are very different.The plates have an approximate interface velocity of10 2 m/s based on the observation that it took 30 s toelongate to the size of 60 lm shown in Fig. 1. This is farless than the interface mobility of martensitic transformations,which can approach the speed of sound in thematerial (i.e. 10 3 m/s) [22]. For truly martensitic behaviour,the interface structure must meet certain restrictivecriteria, otherwise the transformation willrequire long-range diffusion of material to the disconnections,crystal defects at the interface, or disconnection/crystal-defectintersections [23,24]. This diffusionwould ensure accommodation of the misfit strainsparallel to the moving interface. Nevertheless, transformationby the motion of interfaces can still exhibitshear-like features because of the shear component ofthe disconnection, i.e. the dislocation component of theinterfacial defect, while the kinetics is limited by diffusionalaccommodation to specific sites at the interface.4.2. Grain boundary allotriomorphsThe grain boundary allotriomorphs exhibit no systematicsurface topography or morphological features ofmilitary transformations like that of the intragranularplates, though they show the same oriented nucleation.After nucleation (regardless of whether this is on, ornear to the grain boundary), a b allotriomorph growsalong an a–a grain boundary and so its morphology isdominated by the orientation of the a–a grain boundary.At this stage it is unlikely that the interfaces willhave the particular {3 3 4} habit plane that is requiredfor the military transformation mechanism. Growth ofthe b phase is not then expected to result from themotion of disconnections (thus displaying no surfacedisplacement), but to proceed by a ÔcivilianÕ or disorderlytransport of atoms across the interface [16] (i.e.similar to the migration of a high angle grain boundary).The term massive transformation could be used toclassify this type of transformation.Summarising, while there is no difference in thecompositions between the a and b phases, the characteristicsof the intragranular plates and the grainboundary allotriomorphs show that each form grows bydifferent mechanisms, the former with a shear or militarycomponent, and the latter with a civilian mechanism.Next it will be discussed how the difference ingrowth kinetics between these two mechanisms affectsthe final b phase texture.4.3. Final textureThe dominant {0 0 1}h100i texture component of theb phase is consistent with transformation from thestarting a texture according to the Burgers OR withvariant selection. Direct observation of b textures hasbeen made by Zhu et al. [9] using high temperatureXRD, three components reported are {0 0 1}h100i,{0 0 1}h110i and {1 1 2}h110i. Gey and Humbert [10],using a restitution method to calculate b textures frompost-transformation a textures, have calculated{0 0 1}h100i and {1 1 2}h111i components. However,both studies suggested that variant selection occurs inthe a–b transformation only after rolling. These observationsare consistent with the ideas of Furuhara andMaki [7] that suggest that the nucleation/selection of aparticular variant is determined by dislocations andother crystal defects, even if the growth mechanism is acivilian process. In the present study, we are able toobserve transient microstructures and quantify thecrystallography during the phase transformation to assistin our understanding of variant selection processes.Our in situ observations of intragranular plates showthat several crystallographically distinct variants (atleast 4 in Fig. 1(a)) are nucleated within a single parent agrain, suggesting there is no variant selection duringnucleation in this material. Furthermore, for the intragranularplates there was no significant texture componentwith h100i in the ND (see Fig. 5(b)) as there wasfor the grain boundary allotriomorphs (Fig. 5(a)) andfor the final b texture (see Fig. 8(b)). This implies thatthe variant selection we observe in the final b texture isoccurring by means of preferential nucleation and/orgrowth of the grain boundary allotriomorphs. Comparisonof the texture symmetry for the grain boundaryallotriomorphs in Fig. 5(a) with the final b textureFig. 8(b) shows that further variant selection has takenplace. While the b allotriomorphÕs texture has cylindricalsymmetry with the h100i parallel to the ND, the finalb texture has in addition to this the h100i parallel to theRD to give a cube texture. An explanation for the variantselection that causes the observed texture (alsonoted in [9,10] ) is yet to be found.For the grain boundary allotriomorphs to dominatein the final b texture, they must prevail over the intragranularplates during the growth stage. The interpretationof the observations in Fig. 4 is that the boundariesof the b allotriomorphs have greater mobility than thoseof the intragranular plates during the final stages of thetransformation. Furthermore, since intragranular platesare not observed in the final b microstructure, thesemust have been engulfed by the surrounding allotriomorphs.This is shown schematically in Fig. 11. Sincethe intragranular plates within the allotriomorphs arethe same b phase, the driving force for removing theplates cannot come from differences in chemical potential.However, the high curvature at the end of the plates(labelled ÔrÕ in Fig. 11) will make such microstructuralfeatures unstable due to interface energy driving forces,once they are surrounded by b phase and causethese b–b grain boundaries to migrate as shown in theschematic.