Prakt. Met. Sonderband 30 (1999) 123
4. Discussion
Combining both, the results of growth kinetics and microstructure investigations, offers the unique
possibility to reconstruct a scheme of the solidification pathway of undercooled Ni-Si melts. In Fig.
1 the calculated phase diagram, the metastable extensions of liquidus and solidus lines and the To-
line of the respective phases are shown.
The solidification of the undercooled eutectic Ni-Si melts reflects two different solidification modes
inferred from the in situ recalescence measurements. According to the microstructure investigations
there is coupled eutectic growth below the critical undercooling level AT. < 50 K. Beyond that limit
primary growth of weakly segregated metastable dendrites occurs along with the supersaturated o.-
Ni and minor amounts of the B;-phase. Only adjacent to the chill surface the metastable phase is
retained. The thickness 20 pum of the metastable phase region, which displays nanometer sized
grains, is much smaller than 200 um found in undercooled eutectic Nb-Al samples under similar
quenching conditions (8). One obvious reason is the concentration deviation of the layer from the
initial melt composition. This leads to a Si depletion in the melt ahead of the melt-solid interface
and finally to a breakdown of that solidification mode. In more distant regions we observed a segre-
gated NizsSio + o-Ni + B3 microstructure and finally a transformed a-Ni + 8; microstructure of the
two equilibrium phases in the regions distant from the substrate.
Previous authors only have proven a shift in critical undercooling for the formation of the anoma-
lous eutectic microstructure by higher cooling rates. But even liquid metal immersion failed to re-
tain any metastable phases in bulk eutectic Ni-Si samples (6, 12). The present method allows for
much higher cooling rates because the solder layer on the chill substrate improves the substrate-
sample heat transfer. The question whether the molten layer of solder can affect the sample compo-
sition was carefully checked by EPMA and by secondary ion mass spectrometric analyses, too. No
contamination with Sn exceeding 0.05 at.% could be detected. Therefore, we argue that the ten-
dency toward metastable Ni,sSio phase formation is an inherent property of the undercooled Ni-Si
melt itself, The relatively low transformation temperature (compared to the melting point) of 500
°C, points to the poor thermal stability and the possible decay on moderate post-solidification cool-
ing.
The anomalous eutectic microstructure itself, which consists of a three-phase mixture of equiaxed
grains, may evolve from the as-grown metastable dendrites by a diffusion mechanism in the semi-
solid state. The mechanism resembles the formation grain-refined (single-phase) equiaxed micro-
structures from undercooled melts reported by Schwarz et al. (13). That microstructure is formed by
remelting and coarsening of primarily formed dendrites. The microstructure shown in Fig. 3d is
closely related to that reported in (13). For slower cooling the diffusion in the semisolid state might
finally result in the anomalous eutectic two-phase structure. A transformation from the weakly seg-
regated solid phases toward the equilibrium phases solely via solid state transformation is less prob-
able because of the slow diffusion processes.
S. Conclusions
In situ observations of the recalescence process of undercooled melts and microstructure investiga-
tions of quenched samples enable the following conclusions concerning the solidification pathway
of eutectic Ni-Si alloy:
(i) the coupled eutectic growth is replaced by a dendrite solidification mode beyond a critical un-
dercooling level of 30 - 50 K.
(ii) different from previous authors a primary nearly partitionless solidification of NizsSiy metasta-
ble phase dendrites was detected in substrate quenched samples beyond a critical undercooling.
(iii) the Si concentration of the metastable phase rises with undercooling. Its decomposition tem-
perature is near 500 °C.
