## The structure of the α″/β Martensite interface in 10% Mo-Ti

Interfacial Defects. The HREM images show that the interface has step/terrace features (Fig.4). The terraces were parallel to {110}a-(also parallel to{112}p) and the step height was usually about two {110}α· planes. Each of these features corresponds to an interfacial disconnection (interfacial line defect exhibiting step and dislocation character (Hirth and Pond 1996)). The most important question is whether these dislocations are responsible for the martensitic transformation or not. Material flux of the defects and nucleus. The transformation process is envisaged to begin with the nucleation of a favourably oriented but homogeneously strained embryo (Nixon 1999). The nucleus was chosen to exhibit an orthorhombic lattice similar to the a", but homogeneously distorted and rotated to [1 iO]nuci=[l 1 l]p and [ 1 10]p=[001]nUci. This lattice represents a nucleus, which satisfies the exact OR conditions and has a zero defect content on the interfacial plane. There is a number of possible interfacial defects, but only defects which exhibit a finite step can lead to transformation of the crystal structure (Hirth and Pond 1996). The diffusionless nature of the transformation means that the material flux arising due to motion of a disconnection should be zero (Nixon and Pond 1997). It can be shown, that this is equal to the following condition 1ι(μ)/1ι(λ)=Χ(λ)/Χ(μ), where h is the step height, Χ(λ/μ) represents the density of atomic sites in the λ or μ crystal. The final expression can be re-expressed for the nucleus, because the projections of the unit cells onto the interfacial plane are equal: 1ι(μ)/1ι(λ^(λ)Λ1(μ), where d is the interplanar distance of the crystallographic planes parallel to the interface. This result means that the disconnection should be based on step heights proportional to the interplanar distances, and can be written 1ι(λ) = 2n*d(X) and h(p) = 2n-d(p), where n is the same integer. The Burgers vector length of the disconnection is limited, because the Burgers vector length increases with the number n of the atomic planes and for certain n the Burgers vector of the disconnection with the material flux, had smaller length and correspondingly lower energy. This n can be defined by the condition η < (αη2-1)/(2an(tfPcosa-an)) ~ 3.84, where ap is the unit cell parameter of the β phase, an = *j3aa'Cifi / +ba· is the unit cell parameter of the nucleus and a is the angle between [100]a-and [001]p. Thus, the Burgers vectors with the step 2, 4 and 6 interplanar distances are energetically favourable and can produce a diffusionless transformation by their motion along the interfacial plane.