IEEE Electrification Magazine - March 2017 - 20

The adoption of
grain-boundarydiffused NdFeB
magnets has come
with some
challenges for the
end users as far as
the characterization
of and design with
these anisotropic
magnetic materials.

One way to reduce the use of HRE
elements offered by many magnet
manufacturers has been to develop
Dy or Tb grain-boundary-diffusion
(GBD) methods, which result in higher Br and Hcj at elevated temperatures
while using approximately one-half,
or in some cases one-third, of the Dy
or Tb amounts utilized in traditionally
produced NdFeB magnets with similar magnetic properties. The adoption
of grain-boundary-diffused NdFeB
magnets has come with some challenges for the end users as far as the
characterization of and design with
these anisotropic magnetic materials.
It is our goal here to review some of
the items that should be considered
when designing EV motors with grainboundary-diffused magnets.
To produce traditional sintered magnets, the measured raw materials are melted and strip-cast to produce
small flakes of NdFeB. These are further broken up in size
using hydrogen decrepitation and fine milling, resulting
in a fine powder with a particle diameter of 2-10 nm. The
powder is then compacted into small, magnetically
aligned blocks that are sintered and annealed. The sintered blocks are cut into magnets of the desired size. After
surface grinding and corner chamfering, the magnets are
coated for corrosion protection. In conventional processing, the Dy and Tb are added at the initial melting step,
resulting in the HRE elements being uniformly distributed
within the finished magnet, and therefore in uniform

coercivity throughout the magnet
cross section.
In the mid-2000s, some magnet
manufacturers introduced a grainboundary step following the cutting
and grinding step. In this step, a Dyor Tb-rich compound is applied on
the surface of the magnet. The HRE
element is then diffused into the
magnet through another annealing
process, which can take from 24 to
36 h. Since the HRE element diffuses
much faster in the grain boundaries
than in the grains, the resultant magnet contains a layer located right below
the magnet surface where the grain
boundaries are rich with the diffused
HRE element while very little HRE element reaches the grain boundaries in
the geometrical center of the magnet
(see Figure 1). It is also evident that the corners and edges
have an increased concentration of HRE element due to
diffusion from two (in case of the edges) or three (corners)
perpendicular sides of the magnet. The resultant magnet
contains much fewer HRE elements, as the amount in the
base alloy can significantly be reduced and in some cases
eliminated and is often in the range of 0-4 weight percent
(wt%) while diffusion is usually achieved with less than
1 wt% of Tb or Dy.
The level of demagnetization in a permanent magnet can be estimated using finite element analysis
(FEA). The shaded plot of Figure 2 shows the magnitude
of the normal (along the direction of magnetization)

109
102
95
88
80
73
66
59
51
44
37
30
22
15
8
0

78
74
69
64
59
54
49
44
39
34
29
24
19
14
9
4

50 µm Dy 20 kV 20 nA
(a)

1,000 µm Dy 20 kV 80 nA
(b)

Figure 1. Electron-probe-microanalysis (EPMA) maps of a Dy-diffused magnet. (a) A closer view of the grain boundaries, which are richer with Dy
compared with the grains. (b) A Dy map of half of the magnet.

I E E E E l e c t r i f i c ati o n M agaz ine / march 2017

Table of Contents for the Digital Edition of IEEE Electrification Magazine - March 2017