Magnetic Nanocomposite Materials for High Temperature Applications

Frank Johnson, Amy Hsaio, Colin Ashe, David Laughlin, David Lambeth, Michael E. McHenry
Department of Materials Science and Engineering
Carnegie Mellon University
Pittsburgh, PA 15213
Tel: 412-268-4037 Fax: 412-268-3113
Email: fj04@andrew.cmu.edu

Lajos K. Varga
Research Institute for Solid State Physics and Optics
Hungarian Academy of Sciences
H-1525 Budapest, P.O.B. 49
Tel: (36-1-) 392 2222 Fax: (36-1-) 392 2215
Email: varga@szfki.hu
Abstract
Magnetic nanocomposites comprised of nano-sized
magnetic crystals embedded in an amorphous matrix
have been shown to have excellent soft magnetic
properties. In particular, amorphous and nanocrystalline
materials have been investigated for various soft
magnetic applications including transformers and
inductive devices. The historical development of
nanocomposite soft magnetic materials will be
reviewed. In these materials it has been determined that
an important averaging of the magnetocrystalline
anisotropy over many grains coupled within an
exchange length is at the root of the magnetic softness
of these materials. The crystallization kinetics and the
chemical partitioning occurring during crystallization
will be described.
In particular, the role of the
amorphous phase in exchange coupling magnetic
nanoparticles at elevated temperatures will be
discussed. In HITPERM nanocomposites, developed
by our group, [1] nanocrystalline and (B2)-FeCo
are formed with significantly improved high
temperature magnetic properties than in its
predecessors,
NANOPERM
and
FINEMET.
HITPERM has been developed for applications with
high permeability requirements and with needs for large
inductions at high temperatures. Recent work in
characterizing the performance of HITPERM magnetic
nanocomposite systems will be described. Specifically
we have developed soft magnetic alloys that exhibit
high magnetic induction at temperatures above current
operating ranges for commercial devices.
The
applications for this material in high frequency (up to 1
MHz) inductive devices for high temperature power
electronics will be reviewed.
1.

Introduction

1.1

Key Principles of Soft Magnetic Materials

Soft magnetic materials face demanding

requirements from new, high-performance electronic
and power distribution systems. The new systems must
operate in high-temperature and high-frequency
regimes that are inaccessible to conventional
crystalline and amorphous magnetic materials. The
need for increased energy efficiency requires reduced
power
loss
from
inductive
components.
Nanocrystalline magnetic materials hold promise for
meeting these requirements without resorting to the
trade-offs needed when using conventional materials.
A material that maintains a remnant
magnetization, Mr, in the absence of an applied field is
known as a ferromagnetic material. Ferromagnets are
the most widely used type of magnet, and are typically
referred to as magnetic materials or magnets. A finite
field known as the coercive field, Hc, is used to reverse
the M r. When driven with an applied field a
ferromagnet reaches a saturation magnetization, MS,
that is an intrinsic property of the material. The
magnet's permeability, , reflects the ease with which
it is magnetized and is an extrinsic property that
depends on the microstructure of the material.
A magnet can be characterized by a plot
known as a hysteresis loop (Fig. 1). Hc is measured as
the width of the loop, Ms as the height of the loop, and
as the slope in the first quadrant. The area of the loop
corresponds to the energy stored in the material during
the magnetization process. If the magnetization is
cycled, as is the case in AC applications, then this
stored energy contributes to power loss. A soft
magnetic material is easily magnetized, and has a
narrow loop with a low Hc .

1.2

Figure 1 Example hysteresis loop illustrating

coercivity (Hc ), saturation magnetization (Ms ),
remnant magnetization (Mr ), and permeability ().
The coercive field is influenced by a property
known as magnetic anisotropy. This represents the
force required to rotate the magnetization vector of a
magnet out of it's equilibrium direction. High
anisotropy generally results in a high Hc, as the magnet
experiences a resistance to switching it's magnetization
direction. The magnetocrystalline anisotropy is
generated from the crystalline symmetry of the magnet.
Shape anisotropy results from the need to maintain a
closed flux path around and through a magnet. Stress
anisotropy results from residual or applied stress
applied to the magnet.
All magnetic materials lose their M r at a
temperature known as the Curie temperature (Tc. This
temperature corresponds to the phase transition
between the ordered ferromagnetic state and the
disordered paramagnetic state. This is typically a
second order phase transition, which means that
magnetization falls off gradually before the Tc is
reached. Additionally, Hc is reduced as temperature is
increased, due to a decrease in magnetocrystalline
anisotropy with temperature.

Historical Background

Nanocomposite magnetic materials have their

origins in the amorphous alloys that were brought to
market in the 1970's. Amorphous materials are
characterized by a lack of long range atomic order,
similar to that of the liquid state. Production techniques
include rapid quenching from the melt and physical
vapor deposition is another. The lack of crystallinity
causes amorphous materials to have a very low
magnetic anisotropy. METGLAS 2605 Fe78 Si13 B9 is
a common amorphous magnetic alloy, in which B acts
as a glass forming element.
The importance of anisotropy
suggests
searching for other materials with isotropic magnetic
properties. In magnetic materials the ferromagnetic
exchange length expresses the characteristic distance
over which a magnetic atom influences it's
environment, and has values on the order of 100 nm. If
the magnet has a structure with grain diameters smaller
than the exchange length, it becomes possible to
"average" the anisotropy of the grains to a low bulk
value. This phenomenon of randomized anisotropy will
be discussed in the next section. Such a material then
realizes the high saturation Ms of the crystalline state
and low Hc due to randomized anisotropy.
In 1988 Yoshizawa et al. [2] developed the
FINEMET alloy based on Fe73.5 Si13.5 B9 Nb 3 Cu 1 . This
was an extension of the common Fe-Si-B alloy with Cu
as a nucleation agent and Nb as a grain refiner. The
material is produced in the amorphous state and then
crystallized by annealing. Nb that segregates to the
grain boundaries acts a diffusion barrier preventing
grain growth. The structure is a nanocomposite of 10100 nm diameter bcc- FeSi grains embedded in an
amorphous intergranular matrix.
In 1990 Suzuki et al. [3] reported the
development of the Fe88 Zr7 B4 Cu 1 alloy which was
named NANOPERM . Zr and B act as glass forming
agents in this alloy and the microstructure consists of
-Fe grains embedded in an amorphous matrix. By
eliminating Si, higher saturation inductions are
achieved than in FINEMET, but the Hc are also higher.
The amorphous intergranular phase in both FINEMET
and NANOPERM have Curie temperatures lower than
that of the nanocrystalline grains.
In 1998 Willard et al. [1] reported the
development of HITPERM, an alloy based on the
composition Fe44 Co 44 Zr7 B4 Cu 1 . The key distinction is
the substitution of Co for Fe. HITPERM forms 'FeCo grains in a Co enriched amorphous matrix. The
amorphous matrix has a Curie temperature higher than
the primary crystallization temperature of the alloy.
This allows the '-FeCo grains to remain exchange
coupled at high operating temperatures. Due to the

presence of Co, HITPERM alloy has an Ms higher than

FINEMET or NANOPERM as well as a higher Hc.
1.3

anisotropy is now dependent on the volume fraction

and residual stress:

kt = x

Randomized Anisotropy

The notion of randomized anisotropy was

used by Harris, Plischke, and Zuckerman [4] to explain
amorphous rare earth-transition metal alloys. The rare
earth atoms have non-spherical electron orbitals with
high anisotropy. In crystalline materials, this leads to
high Hc and hard magnetic behavior. Alben, Becker
and Chi extended the model to amorphous soft magnets
[5]. The lower anisotropy of the soft magnetic atoms
allows the ferromagnetic exchange length to become
larger. This enables a larger volume to be sampled for
randomizing anisotropy. An energy expression can be
written that balances exchange energy vs. anisotropy
energy:
3

A
d 2
F 2 Ku
L
L
a

(1)

in which A is the exchange stiffness, L is the exchange

length, Ku is the anisotropy energy, and d is the particle
size. This expression can be minimized with respect to
particle size to find the equilibrium exchange length:

16 A 2
9K u d 3

(2)

Herzer [6] extended this model to

nanocrystalline materials. In a soft material, the
exchange length can be on the same order as a
nanocrystalline grain size. Thus, the anisotropy is
reduced to an effective anisotropy by averaging over
several grains by random walk considerations:

< K >=

Ku

(3)

in which N are the number of grains in the volume

defined by the exchange length. The Hc of the of a
nanocrystalline materials then has a d6 dependence as
expressed by:

HC =

< K >4 d 6
20 A3 M S

(4)

Hernando [7] and Suzuki [8] have extended

Herzer's model to take into account the two-phase
crystalline/amorphous nature of these materials. They
did this by modifying the exchange stiffness to account
for the penetration of the crystalline exchange field into
the amorphous intergranular layer. The randomized

ko 3
+ lef < s >
g3 2

(5)

where x is the particle volume fraction, e f is the

effective magnetostrictive constant of the alloy, and
is the phenomenological constant that modifies the
exchange stiffness.
2.0

Thermal Particle Decoupling

The amorphous and crystalline phases in

nanocrystalline magnetic alloys have distinct
behaviors. In particular, the Tc of the amorphous
phases can be different. In FINEMET and
NANOPERM alloys, the amorphous phase Tc is
approximately 400 K, whereas the crystalline grains
have their Tc in the vicinity of 1000 K. Such
decoupling could be observed as a peak in the Hc as a
function of temperature. Work by Varga [8] on
NANOPERM and FINEMET has observed decoupling
at temperatures higher than the Curie temperature of
the amorphous phase. This can indicate that the
particles are remaining coupled either by exchange
field penetration of the amorphous phase, or through
dipolar interactions of the particles.
Work is underway to characterize the thermal
decoupling behavior of HITPERM alloys. Ribbons of
the composition Fe44 Co 44 Zr3.5 Nb3.5 B4 Cu 1 (Fe:Co 50:50)
and Fe57.2 Co 30.8 Zr3.5 Nb 3.5 B4 Cu 1 (Fe:Co 65:35) were
synthesized in the amorphous state by the single roller
melt spinning method at the Hungarian Academy of
Sciences. These compositions are a modification of the
original HITPERM composition made by substituting
Nb for half the Zr. These ribbons were wound into
toroids and heated at a constant rate of while collecting
AC hysteresis loops at 10 Hz. From these loops Hc, Ms ,
, and Mr were measured as a function of temperature.
M s as a function of temperature was measured in DC
mode with a vibrating sample magnetometer equipped
with a high temperature oven.
Fig. 2 shows the evolution of coercivity with
temperature for the Fe:Co 50:50 alloy. A peak in Hc is
observed at 750 K for the 723 K annealing condition
and at 900 K for the 873 K annealing condition. Fig. 3
shows the Fe:Co 65:35 alloy. A peak in coercivity
appears at 750 K for the 723 K annealing temperature,
and at 900 K for the 873 K and 923 K annealing
temperatures. The measurements were made on one
toroid subjected to several heating runs, so the
microstructure is expected to be continually evolve
during the experiment.

estimated to be 967 K. The difference in amorphous Tc

is caused by the difference in Co composition of the
amorphous phase. It is important to note that the
estimated amorphous Tc will differ from the Tc of the
amorphous intergranular phase in nanocrystalline
materials due to segregation of the glass forming
elements during crystallization.

80

60
50
40

As-Cast
723 K
773 K
823 K
Annealing
873 K
Temperature
938 K

30
20
10

Measured Magnetization
0.08

0
300

400

500 600 700

Temperature (K)

800

900

Figure 2 Coercivity pl ot for Fe:Co 50:50

showing particle decoupling at 900K for the 873 K
annealing condition.

Magnetization (e.m.u.)

Coercivity (A/m)

70

0.06
Calculated
0.04

Amorphous
Magnetization

0.02

0.00
400

50

30
20
10

1200

Figure 4 Thermomagnetic plot for Fe:Co 50:50

showing crystallization at 725 K and an estimated
amorphous phase Tc of 1190 K.
As-Cast
723 K Annealing
773 K
Temperature
823 K
873 K
923 K

Measured Magnetization
0.4

0
300 400

500 600 700 800

Temperature (K)

900

Figure 3 Coercivity plot for Fe:Co 65:35 showing

particle decoupling at 900 K for the 873 and 923 K
annealing conditions.
Figure 4 shows the M s vs. T plot for the Fe:Co
50:50 alloy. The alloy shows a crystalline Tc of 1225
K. The plot shows evidence for primary crystallization
at 725 K. The Tc of the amorphous phase can be
estimated by fitting the amorphous magnetization to a
spin-only (J=1/2) Brillouin function and extrapolating
to high temperatures. In this way, the amorphous Tc is
calculated to be 1190 K. Figure 5 shows the Ms vs. T
plot for the Fe:Co 65:35 alloy. The crystalline Tc is
measured as 1239 K, and the amorphous Tc is

Magnetization (e.m.u.)

Coercivity (A/m)

40

600
800
1000
Temperature (K)

0.3
Calculated
0.2

Amorphous
Magnetization

0.1

0.0
400

600
800
1000
Temperature (K)

1200

Figure 5 Themomagnetic plot for Fe:Co 65:35

showing crystallization at 725 K and an estimated
amorphous phase Tc of 1239 K.
These results show that HITPERM have the
highest decoupling temperature yet observed for

3. Crystallization Kinetics
Nanocomposite materials are formed by
precipitating nanocrystalline grains from an amorphous
precursor. These crystallization events can be modeled
by using models of nucleation and growth from an
undercooled liquid. In HITPERM, two crystallization
events take place. In the primary crystallization event,
-FeCo grains form in the amorphous phase. As the
grains grow, glass forming elements such as Zr
segregate from the -FeCo grains into the amorphous
matrix.
At
higher
temperatures,
secondary
crystallization forms Fe(Co)B and Fe(Co)Zr grains
occurs in the amorphous intergranular phase. These
secondary phases have high magnetocrystalline
anisotropies and raise the Hc of the nanocomposite
material.
These crystallization events are diffusional
processes characterized by activation energies.
Kissinger [10] developed an analytical technique for
determining activation energies using differential
scanning calorimetry at constant heating rate. Peak
crystallization temperatures are collected at heating
rates. The peaks shift due to the finite crystallization
rate of the nanocrystalline grains. The peak
temperatures and heating rates are plotted against each
other by the equation:

ln
2
Tx
E
=
(6)
R
1

Tx

( ( ))
( )

where Tx is the peak crystallization temperature, is

the heating rate, R is the gas constant, and E is the
activation energy.
The HITPERM alloys were measured in a
Perkin-Elmer DSC-7 at heating rates of 5, 10, 20 and
40 K/min. Figure 6 presents the DSC curves collected
for the Fe:Co 50:50 sample. These curves have been
corrected by subtracting the baseline signal of DSC
system from the measured data. The peaks are
exothermic, reflecting the fact that the system is
moving from a high entropy amorphous state to a low
entropy crystalline state.

Heat flow (mW)

5.0
0.0
-5.0
5 K/min
10 K/min
20 K/min
40 K/min

-10.0
-15.0
-20.0
300

400

500
600
700
Temperature (K)

800

900

Figure 6 DSC scan for Fe:Co 50:50 showing

exothermic crystallization with peak temperatures
that increase with heating rate.
Figure 7 is a Kissinger plot of the
crystallization peaks for the Fe:Co 50:50 alloy. The
activation energy for this composition is 272 kJ/mol
(2.8 eV/atom) This is compared to a previously
reported value of 323 kJ/mol (3.4 eV/mol) for an
HITPERM alloy of composition Fe44 Co 44 Zr7 B4 Cu 1 [1].
An activation energy of 292 kJ/atom (3.0 eV/atom was
measured for the Fe:Co 65:35 composition. This is
compared to a previously reported value of 270 kJ/mol
(2.8 eV/atom) for HITPERM alloy with composition
Fe57.2 Co30.8 Zr7 B4 Cu 1 [11].
-13.5
-14.0

Ln(/Tp^2)

nanocomposite magnetic alloys. This is a consequence

of having Co as an alloying element. The decoupling
temperature decreases as Co content decreases,
reflecting behavior closer to NANOPERM and
FINEMET. Future work will examine the
ferromagnetic and dipolar nature of the interparticle
coupling. The evolution of anisotropy with processing
steps such as field annealing will also be studied. The
goal is better understand the phenomenon of
randomized anisotropy and how it contributes to the H c
of nanocomposite alloys.

-14.5
-15.0
-15.5
-16.0
0.00130

29.08 - 3.27E+004*x

0.00132

0.00134
1/Tp

0.00136

0.0013

Figure 7 Kissinger plot for Fe:Co 50:50 with an

activation E of 272 kJ/mol (2.8 eV/atom), R2 >99%.

Applications

FJ and MEM gratefully acknowledge support from the

ABB Corporation.

4.1

Transformers

References

Miniaturization and efficiency requirements

demand the reduction of size and mass of core
materials in transformer. Increasing the Ms and will
allow less magnetic material to be used for a given
transformer application. Decreasing Hc will reduce loss
in AC applications, improving efficiency. Operating
temperatures may increase as power electronic systems
become more densely populated with components. This
creates a need for magnetic materials with increased
operating temperatures. This can be achieved with
nanocrystalline materials with high crystalline and
amorphous Tc to prevent particle decoupling.
4.2

DC-DC power converters

DC-DC power converters offer the advantage

of reduced size and weight over conventional line
frequency transformer based power supplies. These
converters are high frequency devices that use
magnetic transformers and inductors, along with active
circuit elements, to convert voltage levels. Ferrite
materials are presently used to meet the frequency
requirements. The low Ms and Tc of ferrite materials
limits the miniaturization potential of converters. A
magnetic material that had the Ms of iron and an
operating frequency of 1 MHz could result in a factor
of 50 reduction in weight and volume. Nanocomposite
magnetic materials already have this Ms and have
operating frequencies of 100 kHz. New, more resistive
nanocomposite structures have been conceived that
will increase the operating frequency above 1 MHz.
5.

Conclusions

Nanocomposite magnetic materials have been

developed as an evolution of amorphous magnetic
alloys. These alloys display low Hc and power loss due
to randomized anisotropy of the two-phase structure of
nanocrystalline grains embedded in an amorphous
matrix. The HITPERM alloys, based on the FeCo
system, offer high Ms a n d Tc. HITPERM has been
shown to have the highest interparticle decoupling
temperatures yet observed. These properties show
promise for increasing efficiency and component
packing density in power electronic applications.
Acknowledgements
This work was supported in part by the Air Force
Office of Scientific Research, Air Force Material
Command, USAF, under Grant No. F49620-96-1-0454.

[1] M. A. Willard, M. -Q. Huang, D. E. Laughlin, M. E.

McHenry, J. O. Cross, V. G. Harris and C. Franchetti,
Magnetic properties of HITPERM (F 0.5 Co
0.5 )88 Zr7 B4 Cu 1 nanocrystalline magnets, J. Appl. Phys.,
vol. 85, pp. 4421-23, 1999.
[2] Y. Yoshizawa, S. Oguma, K. Yamauchi, New Febased soft magnetic alloys composed of ultrafine grain
structure, J. Appl. Phys., vol. 64, pp. 6044-6046,
1988.
[3] K.Suzuki, A. Makino, N, Kataika, A. Inoue, and T.
Masumoto, High saturation and soft magnetic
properties of bcc Fe-Zr-B and Fe-Zr-B-M (M=
transition metal) alloys with nanoscale grain size,
Matl. Trans., JIM, vol. 32, pp. 93-102, 1991.
[4] R. Harris, M. Plischke, and M. J. Zuckerman, "New
Model for Amorphous Magnetism," Phys. Rev. Lett.,
vol. 31, pp. 160-162, 1973.
[5] R. Alben, J.J. Becker, and M.C. Chi, "Random
Anisotropy in Amorphous Ferromagnets, " J. Appl.
Phys., vol. 49, pp. 1653-1658, 1978.
[6] G. Herzer, Grain size dependence of coercivity
and permeability in nanocrystalline ferromagnets,
IEEE Trans. Magn., vol 26., pp 1397-1402, 1990.
[7] A. Hernando, M. Vazquez, T. Kulik, and C. Prados,
"Analysis of the dependence of spin-spin correlations
on the thermal treatment of nanocrystalline materials,"
Phys. Rev. B, vol. 51, pp. 3581-3586, 1995.
[8] K. Suzuki and J. M. Cadogan, "Random
Magnetocrystalline
anisotropy
in
two-phase
nanocrystalline systems," Phys. Rev. B, vol. 58., pp.
2730-2739, 1998.
[9] L.K. Varga, L. Novak, and F. Mazaleyrat, Effective
magnetic anisotropy and internal demagnetization
investigations in soft magnetic nanocrystalline alloys,"
J. Mag. Mag. Mat'l., vol. 210, pp. L25-L30, 2000.
[10] Homer Kissinger, "Reaction Kinetics in
Differential Thermal Analysis," Anal. Chem., vol. 29,
pp. 1702-1706, 1957.
[11] F. Johnson, P Hughes, R. Gallagher, D.E.
Laughlin, M.E. McHenry, M.A. Willard, and V.G.
Harris, "Structure and Thermomagnetic Properties of
New FeCo-based Nanocrystalline Ferromagnets, IEEE
Trans. Mag., in press, 2001.