In 1988, Yoshizawa et al. [1,2] reported that excellent permeability is obtained when the crystal grain size is reduced to a nanometer scale by crystallizing the Fe-Si-B-Nb-Cu amorphous alloy. The principle of magnetic softening accompanied by the reduction of the grain size in the nanoscale range was later explained based on the random anisotropy theory by Herzer [3,4]. According to this model, the origin of the magnetic softness is ascribed to the reduced magnetocrystalline anisotropy due to the random distribution of nanoscale grains.
Figure 1 shows a typical electron micrograph of Fe-13.5Si-9B-3Nb-1Cu alloy produced by the primary crystallization process from the melt-spun amorphous ribbon. This consists of the nanocrystalline bcc phase with an average grain size of approximately 10 nm. There is no preferred orientations in the grains. In addition to grains of ~10 nm, smaller grains can be recognized as indicated by arrowheads. These are now known to be fcc Cu particles which precipitated out in the early stage of crystallization. For producing such nanocrystalline microstructure, it is reported that a combined addition of Nb and Cu is required [2], thus the role of Nb and Cu for nanocrystallization was of great research interest. Atom probe field ion microscopy was found to be an extremely effective technique to study solute clustering, precipitation, segregation and partitioning behaviors, in fact great contribution to the understanding of nanocrystallization mechanism were made by a conventional atom probe [5-8].
Figure 2 shows atom probe concentration depth profiles of the Fe73.5Si13.5B9Nb3Cu1
alloy with the optimum magnetic properties (annealed at 550 C for 60 min.).
The presence of three types of phases are clearly identified. One is enriched
with Si (~20-25 at.%) but contains little Nb, Cu and B. Complementary TEM
observations suggested that this region was the crystallized bcc a-Fe phase
containing Si. In addition to the a-Fe grains, a B and Nb enriched amorphous
phase is present with little Cu content but containing certain amount of
Si. In addition to these two phases, a Cu enriched particle is observed.
This phase was significantly enriched in Cu (~60%) but still contains appreciable
amounts of the other elements. Since the concentration of Fe is only about
30%, this phase is believed to be nonmagnetic. In fact, a separate nanobeam
electron diffraction study in a transmission electron microscope (TEM)
revealed that the Cu enriched particles were fcc Cu. This phase appears
as a grain having a diameter of approximately 5 nm as indicated by arrows
in Fig. 1.
By analyzing the as-quenched specimen by TEM, FIM and conventional AP,
it was confirmed that the as-quenched alloy was a structurally and chemically
uniform amorphous phase within the limitation of the spatial resolution
of these techniques. Since APFIM data does not contain information on the
nearest neighbor atoms, it is impossible to determine the chemical short
range ordering (CSRO), even if amorphous alloys are chemically or geometrically
short range ordered as suggested in many literature on the structure of
amorphous alloys [9]. On the other hand, APFIM plays a unique role in detecting
solute clusters. Fig. 3 shows the 3DAP analysis result of the same alloy
in the early stage of crystallization [5-8]. Distribution of Cu atoms in
a selected volume is shown in Fig. 3 (a) and iso-concentration surface
of 20at.%Cu is shown in Fig. 3 (b). The density of the particles is in
the order of 1024 m-3. Separate HREM observation suggests that the fringe
contrast of these particles in this stage already shows the feature of
fcc. One remaining question is whether or not the Cu clusters (or precipitates)
which are initially formed prior to crystallization provide heterogeneous
nucleation sites. One thought is that crystallization starts from the Cu
depleted region as proposed by Yoshizawa and Yamauchi, and the other thought
is that Cu cluster provide heterogeneous nucleation sites for a-Fe(Si)
primary crystallization. Fig. 4 shows 3D elemental maps near a Cu precipitate
in Fe-13.5Si-9B-3Nb-1Cu alloy annealed at 550 C for 10 min. Cu precipitate
is adjacent to the a-Fe(Si) particle, which suggests that Cu provides nucleation
sites for the a-Fe(Si) primary crystals. Similar observation was made in
Fe-7Zr-3B-1Cu alloy, in which Cu cluster form prior to the onset of nucleation
and these apparently provide heterogeneous nucleation sites for a-Fe. As
Cu has the fcc structure and a-Fe is bcc, they should have some orientation
relationships so that the atomic arrangement matches at the interface of
Cu/Fe. Direct determination of such a OR is now being attempted by HREM.
