the reversible temperature coefficient.
We have derived the demagnetization curves of permanent magnets
in previous Sections, and have showed actual curves for some of the most
popular materials in Section 2A. It was apparent that, while the
characteristic shape of each curve is retained as the magnet's temperature
varies, the curve's position relative to each axis changes. In terms of a
magnet's intrinsic demagnetization curve, its magnetization
+Msat and its intrinsic
coercivity -Hci are both
changing with temperature. Because Br=µoMsat,
the magnet's "normal" B vs. H curve
will move by virtue of both its remanence Br
and its coercivity -Hc
changing with temperature.
Whatever flux density we expected the magnet to provide at
room temperature, this will be different at any other temperature,
directly affecting the performance of the device in which the magnet is
installed. Furthermore, in Section 1C we warned that the "knee"
in the demagnetization curves represents the onset of a reversal
of the material's magnetization, but now we see that the position
of the "knee", and hence the threshold for magnetization
reversal, is dependent upon temperature.
Change in Remanence
If a magnet delivers flux density B corresponding to a
point on its B vs. H curve which is below
the "knee", the reversal of some of the material's magnetization
will require a remagnetization of the magnet to restore its full
saturation magnetization Msat. In
this case, it is said that the magnet has suffered an irreversible
loss, because although the condition is in fact reversible
with full remagnetization, this is not usually something a user can
perform on an assembled device. However, if the magnetic properties vary
with temperature without causing the magnet to operate at a point
on the B vs. H curve which is below the "knee",
then the magnet is only suffering a reversible loss, since
the original operating condition is fully restored when the temperature
returns to its normal level. The latter reversible
condition is explained in this Section.
Magnetization M is the bulk material property, defined
in Section 1A as the sum of the magnetic dipole moments µm
per unit volume.
M
will assume its saturation value of Msat
when all the µm are aligned.
Increasing temperature thermally agitates the dipole moments, disrupting
their alignment and reducing M below its Msat
level. At a sufficiently high temperature, there is enough thermal
agitation to yield no net alignment of the µm,
so that M completely disappears. This critical value is
known as the Curie Temperature Tc,
which is a fundamental parameter of every magnetic material. This
general relationship between M and temperature T
is illustrated in the diagram to the right. Permanent magnet materials are
likely to have better thermal stability if they contain constituent
elements with higher values of Tc,
the most commonly used being:
| Ni: |
358oC |
| Fe: |
770oC |
| Co: |
1120oC |
Since Br=µoMsat
at room temperature, the remanence at elevated
temperature is the new value of flux density for the
condition when a magnet now develops no magnetizing force (H=0).
Therefore, the relationship between Br
and temperature follows the same form as the diagram above.
Furthermore, as a practical matter one doesn't want a permanent magnet
material to operate anywhere close to its Tc,
so Br will gradually
change over the normal range of operating temperature as illustrated. This
was seen in each set of demagnetization curves in Section 2A , by
looking at the characteristics' intercepts with the B-axes.
It is seen again in the demagnetization curves shown below, which
are for another grade of fully dense anisotropic neodymium-iron-boron.
By
plotting the intercepts of these curves with the B-axis,
we can see the change in Br
with temperature. The decline is approximately linear up to
some transition temperature (almost 200oC for
this neodymium-iron-boron grade), above which there is a rapid
degradation in Br. This linear
region is a reversible loss, whose slope is known
as the reversible temperature coefficient of Br
(α) for the material. Above the transition temperature, the
magnet is suffering an irreversible loss, such that
Br does not recover its original
value upon return to room temperature. This irreversible loss
is easily visualized from the set of demagnetization curves above,
which clearly show the "knee" approaching the B-axis
as the temperature rises towards 200oC.
The reversible temperature coefficient of Br (α) is defined as the percentage deviation in Br
from its value at room temperature (%/oC).
Typical values for the permanent magnets described in
Section 2A are:
|
ceramic ferrite |
sintered SmCo5 |
sintered Sm2Co17 |
fully dense Nd2Fe14B |
| reversible temperature coefficient of Br,
α |
-0.20 %/oC |
-0.045 %/oC |
-0.03 %/oC |
-0.10 %/oC |
| Curie Temperature, Tc |
450 oC |
750 oC |
825 oC |
310 oC |
From the Curie Temperatures of the constituent elements, it is
clear that cobalt-based alloys should have better temperature stability
than iron-based alloys, and indeed the Table above confirms this. The
problem is, of course, that cobalt is a much more costly element than
iron.
Change in Coercivity
The change in a magnet's coercivity Hc
with temperature will depend upon the changes occurring both in
Br and in Hci.
In Section 1B, we derived the intrinsic coercivity for a
magnet with magnetocrystalline anisotropy as:

We would therefore expect that, as temperature rises and M
falls below its Msat level,
Hci will increase. This
is indeed what happens with ceramic ferrites, whose actual
characteristics strongly adhere to magnetocrystalline anisotropy.
However, with the samarium-cobalt and neodymium-iron-boron
materials described in Section 2A, Hci
is seen to decrease with temperature. This is not
specifically due to nucleation or pinning complicating the
magnetocrystalline anisotropy mechanism. The cause is at the
atomic level, where magnetic moments µm
of the rare earth elements are themselves strongly temperature
dependent, and (since K1 = 8µoµmM) K1
is no longer constant
but will control the change in Hci
via the equation above.
While it was possible to assign an approximately constant value to the
reversible temperature coefficient of Br
(α), the changes in Hci and
hence Hc are not as easy to
predict, so the actual demagnetization curves should be
used to study these parameters.
Copyright © 1999 by Princeton
Electro-Technology, Inc.
All rights reserved.


