(b) Austenite grain boundaries act as Zn diffusion
paths. The boundary region transforms to a-
Fe(Zn) since Zn is a strong ferrite stabilizer.
This further enhances Zn grain boundary diffusion,
as the substitutional solute diffusivity in
ferrite is higher than in austenite.
(c) Grain boundary cracking of the thin layer of
diffusion-transformed region, where the austenite
grain boundary region has been replaced by
the low-strength a-Fe(Zn).
(d) After crack formation, Znliq advances by capillarity
and the grain boundary at the crack tip
is again diffusion-transformed by Zn to a-
Fe(Zn). The crack propagates when the grain
boundary is weakened and unable to withstand
the applied stresses magnified locally at the
crack tip. This leads to crack propagation by
repetition of stages (b) and (c).
(e) During quenching, Fe-saturated Znliq transforms
to C-Fe3Zn10, making it possible to map
precisely the position of the liquid in the matrix.
The austenitic matrix transforms to a0 martensite.
The a-Fe(Zn) surrounding the crack remains
unchanged.
Old [4] noted that diffusion mechanisms were unlikely
to cause LME as cracks have a high velocity. The fast
diffusion of small amounts of Zn in the austenite grain
boundaries can result in the formation of a very thin
a-Fe(Zn) layer and its subsequent cracking as the Znliq
continuously follows the advancing crack. The phenomena
observed in the present study are similar to the
embrittlement of pure Fe reported by Suzuki et al.
[20]. They showed that embrittlement by intergranular
fracture was caused by formation of a pro-eutectoid ferrite
film at austenite boundaries.
In conclusion, it is postulated that LME crack formation
is caused not by Znliq, but by the presence of a thin