Classification of Displacive Transformations : What is a Martensitic Transformation ?

The displacive transformation classification proposed at ICOMAT 79 [I] is reviewed in light of recent progress in mechanistic understanding. Issues considered include distinctions between shuffle transformation vs. self-accommodating shear, dilatation vs. shear-dominant transformation, and nucleated vs. continuous transformation. 1. HISTORICAL INTRODUCTION Cohen, Olson and Clapp (COC) at ICOMAT-79 defined a martensitic transformation as a sub-category of a wider field of displacive changes [I]. In the intervening 16 years, there has been appreciable progress in understanding martensitic and related transformations, and it now seems appropriate to re-examine the COC criteria in the light of recent experimental and theoretical work on microscopic mechanisms of transformation. To some metallurgists, martensite will always be a particular microconstituent in suitably heat-treated ferrous alloys, whilst others may use the apparently tautological definition, "Martensite is the product of a martensitic reaction." The emphasis on the mechanism of transformation rather than the properties of the product phase began with an important paper by Troiano and Greninger [2], who showed that various kinetic and crystallographic characteristics of a martensitic change are markedly different from those observed in other "nucleation and growth" reactions. Solid state reactions were thus divided into two main groups, but Kurdjumov [3] and Kurdjumov and Maximova [4] soon showed that some of the Troiano-Greninger features (e.g., no isothermal transformation) are not applicable to all martensitic transformations. It was later proposed [5,6] that a change of shape of the product crystals should be an experimental test for a martensitic (or more generally a "displacive" [7] or "military" [8]) reaction. Phase transformations which involve long range diffusion are regarded as reconstructive rather than displacive, but the early work of Garwood [9] showed, and more recent work [lo-1 11 has confmed, that shape changes occur in some plate-shaped precipitates with substitutional solute contents different from those of their parent solid solutions. These experimental results imply that some transformations requiring diffusion also have displacive character; they are sometimes called "nonferrous bainites," but this can be misleading. To emphasize their mixed characteristics, the term "diffusional-displacive transformations" has recently been suggested [12]. The COC classification scheme for separating martensitic from other displacive transformations is reproduced in Figure 1. Displacive transformations dominated by atomic shuffles rather than by lattice deformation are first eliminated, and only those lattice distortions in which shear strains are more important than dilatational strains are accepted as martensitic. Finally, a division is suggested between true martensites, in which the strain energy has a major influence on the transformation kinetics and product morphology, and quasimartensites with very small lattice deformations, in which transformation may be continuous. Whilst this scheme attempted to minimize the need for detailed mechanistic information, emphasizing instead kinematic and morphological information deducible from relatively macroscopic observation, it has since been suggested that nucleation is an essential characteristic of a martensitic reaction [13]. A concise definition of a martensitic transformation is then a "shear dominant, lattice distortive, diffusionless transformation occurring by nucleation and growth" 1131. Some points worthy of further consideration are (a) the definition of a shuffle and the validity of the distinction between shuffle-dominant and strain-dominant mechanisms, (b) self-canceling microshears which give Article published online by EDP Sciences and available at http://dx.doi.org/10.1051/jp4:1995801 (3-4 JOURNAL DE PHYSIQUE IV 1 DlSPLAClVVDlFFUSlONLESS PHASE TRANSFORMATIONSI