See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/257974158

Classical eddy current losses in Soft Magnetic Composites

Article  in  Journal of Applied Physics · March 2013

DOI: 10.1063/1.4795744

CITATIONS READS
16 569

6 authors, including:

Carlo Appino Olivier de la Barrière

INRIM Istituto Nazionale di Ricerca Metrologica French National Centre for Scientific Research
123 PUBLICATIONS   966 CITATIONS    40 PUBLICATIONS   440 CITATIONS   

SEE PROFILE SEE PROFILE

Fausto Fiorillo M. Lobue

INRIM Istituto Nazionale di Ricerca Metrologica Ecole normale supérieure de Cachan
280 PUBLICATIONS   3,940 CITATIONS    78 PUBLICATIONS   995 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Measuring methods of magnetic materials. View project

Magneto-electrics View project

All content following this page was uploaded by Carlo Appino on 16 May 2014.

The user has requested enhancement of the downloaded file.

 Classical eddy current losses in soft magnetic composites
C. Appino, O. de la Barrière, F. Fiorillo, M. LoBue, F. Mazaleyrat et al.

Citation: J. Appl. Phys. 113, 17A322 (2013); doi: 10.1063/1.4795744

View online: http://dx.doi.org/10.1063/1.4795744
View Table of Contents: http://jap.aip.org/resource/1/JAPIAU/v113/i17
Published by the AIP Publishing LLC.

Additional information on J. Appl. Phys.

Journal Homepage: http://jap.aip.org/
Journal Information: http://jap.aip.org/about/about_the_journal
Top downloads: http://jap.aip.org/features/most_downloaded
Information for Authors: http://jap.aip.org/authors

Downloaded 01 Oct 2013 to 193.204.114.64. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://jap.aip.org/about/rights_and_permissions
 JOURNAL OF APPLIED PHYSICS 113, 17A322 (2013)

Classical eddy current losses in soft magnetic composites

C. Appino,1 O. de la Barrière,2,a) F. Fiorillo,1 M. LoBue,2 F. Mazaleyrat,2 and C. Ragusa3
1
Instituto Nazionale di Ricerca Metrologica (INRIM), Torino, Italy
2
SATIE, ENS Cachan, CNRS, UniverSud, 61 av du President Wilson, F-94230 Cachan, France
3
Dipartimento Energia, Politecnico di Torino, C.so Duca degli Abruzzi 24, 10129 Torino, Italy
(Presented 15 January 2013; received 2 November 2012; accepted 12 December 2012; published
online 22 March 2013)
This paper deals with the problem of loss evaluation in Soft Magnetic Composites (SMCs),
focusing on the classical loss component. It is known that eddy currents can flow in these granular
materials at two different scales, that of the single particle (microscopic eddy currents) and that of
the specimen cross-section (macroscopic eddy currents), the latter ensuing from imperfect
insulation between particles. It is often argued that this macroscopic loss component can be
calculated considering an equivalent homogeneous material of same bulk resistivity. This
assumption has not found so far clear and general experimental validation. In this paper, we discuss
energy loss experiments in two different SMC materials, obtained using different binder types, and
we verify that a classical macroscopic loss component, the sole size-dependent term, can be
separately identified. It is also put in evidence that, depending on the material, the measured
sample resistivity and the equivalent resistivity entering the calculation of the macroscopic eddy
currents may not be the same. A corrective coefficient is, therefore, introduced and experimentally
identified. This coefficient appears to depend on the material type only. An efficient way to
calculate the macroscopic classical loss in these materials is thus provided. V C 2013 American

Institute of Physics. [http://dx.doi.org/10.1063/1.4795744]

I. INTRODUCTION are negligible. In this paper, this limitation is overcome,

Soft Magnetic Composites (SMCs) are of interest in
modern electrical engineering applications. Their isotropic
magnetic and thermal behavior provides a clear advantage
for machines with 3D flux paths.1
The extension of the loss separation concept to SMC
would be suitable to predict the loss in applications, but is not
obvious, because one has to deal with an inhomogeneous
granular structure, where eddy currents (e.c.) flow at two dif-
ferent scales: the scale of the single particle (microscopic
e.c.), and the scale of the whole sample (macroscopic e.c., due
to intergrain conductivity).2,3 The correspondingly measured
bulk resistivity is often considered in the macroscopic eddy
current calculations, assuming an equivalent homogeneous
material. It has been suggested that microscopic and macro-
scopic e.c. can be associated with microscopic and macro-
scopic classical loss components, respectively.2 Although a
certain dependence of the total dynamic loss on the bulk resis-
tivity has been shown,4 no clear experimental evidence of the
separation between macroscopic and microscopic classical
losses has been provided so far. At the same time, the assumed
link between the measured material resistivity and the macro-
scopic e.c. has not been supported by experimental
observations.
It was previously shown5 that loss separation accord-
ing to STL could be carried out in small and highly resis-
tive SMC samples, where the macroscopic eddy currents
a)
Author to whom correspondence should be addressed. Electronic mail:
barriere@satie.ens-cachan.fr
by considering different SMC samples of various sizes and
nature.
II. EXPERIMENTAL
A. Samples
The experiments presented in this paper have been car-
ried out on several samples of two SMC materials, herein
called SMC1 and SMC2, produced from a high purity iron
powder ATOMET 1001HP (Ref. 6) provided by Quebec
Metal Powders (QMPs). The particles in the SMC1 and
SMC2 materials are insulated by means of organic and inor-
ganic binders, respectively. The SMC1 material is heat-
treated at low temperature (1 h at 160  C), so as to improve
the mechanical properties (e.g., fracture strength) without
damaging the organic insulator.7 A higher-temperature treat-
ment (1 h at 425  C), as permitted by the inorganic insulator,
is applied to the SMC2 material.7
The samples are delivered as rings with rectangular
cross-section (outside diameter 52.6 mm, inside diameter
43.8 mm). Three different ring thicknesses have been consid-
ered: t1 ¼ 5 mm, t2 ¼ 9 mm, and t3 ¼ 13 mm. Type and geom-
etry of each sample are here identified as SMC1-tj (i ¼ 1, 2
and j ¼ 1, 2, 3). The compaction pressure (600 MPa), permits
one to reach densities in the order of 7100 kg/m3 in all sam-
ples and materials. The resistivity of all samples has been
measured, and is shown in Table I. Some dispersion (due to
non-perfect reproducibility of the manufacturing process) is
observed.

0021-8979/2013/113(17)/17A322/3/$30.00 113, 17A322-1 C 2013 American Institute of Physics

V

Downloaded 01 Oct 2013 to 193.204.114.64. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://jap.aip.org/about/rights_and_permissions
 17A322-2 Appino et al.
TABLE I Obtained resistivities q for all materials (SMC1 and SMC2) and
thicknesses (t1, t2, and t3).
Axial thickness (mm)
Material t1 ¼ 5 t2 ¼ 9
SMC1 q ¼ 1590 lX  m q ¼ 911
SMC2 q ¼ 48 q ¼ 43
III. MODELING THE MACROSCOPIC EDDY CURRENT
LOSSES
A. Macroscopic eddy currents: evidence and
computation
The magnetic characterization of the ring specimens is
performed under controlled sinusoidal polarization (peak
value Jp ¼ 1 T) from DC to 10 kHz. The experiments show
(see Fig. 1 for SMC2, a similar behavior being observed in
SMC1) that the specific loss in SMC materials depends on
the material cross-sectional area (i.e., ring thickness). To
explain this phenomenon, it is often assumed2 that the
observed losses in SMC samples are due to physical effects
occurring upon two different scales: (a) the microscopic loss,
due to the e.c. circulating within the individual iron particles;
(b) the macroscopic classical loss, due to the e.c. flowing
from particle to particle thanks to imperfections in particle
insulation and describing macroscopic patterns. However, no
clear experimental evidence for effective role of these eddy
currents has been provided so far and there is no consensus
on the underlying assumptions.8,9
In order to single out the contribution to the specific loss
depending on the sample cross-section (WMAC) from the one
occurring upon the scale of the single particle (the micro-
scopic loss WMIC), the loss difference measured in samples
differing only for their size is considered. We thus write,
considering two sizes (a) and (b): DW(a,b) ¼ W(a)  W(b)
¼ WMAC(a)  WMAC(b). Figs. 2 and 3 show the differences
DW measured between SMC1-t3 and SMC1-t1, and SMC2-t3
and SMC2-t1, respectively. Similar results are obtained in
other samples. DW linearly depends on frequency, thereby
showing that the macroscopic loss contribution is classical in
nature.
FIG. 1. Specific loss in SMC2 samples (inorganic binder) as a function of
frequency, for three different thickness values t1, t2, and t3 (sinusoidal polar-
ization, Jp ¼ 1 T).
Downloaded 01 Oct 2013 to 193.204.114.64. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://jap.aip.org/about/rights_and_permissions
J. Appl. Phys. 113, 17A322 (2013)
t3 ¼ 13
q ¼ 1170
q ¼ 45
FIG. 2. Measured and calculated energy loss difference DW (sinusoidal
polarization, Jp ¼ 1 T) between the SMC1-t3 and SMC1-t1 samples. The pre-
dicted DW behavior is obtained from Eq. (3) using the loss coefficient
Q(loss) ¼ 1.
DW ða;bÞ ¼ W ðaÞ  W ðbÞ ¼ Wclass;MAC ðaÞ  Wclass;MAC ðbÞ : (1)
We can thus generally write for the total specific loss W(Jp, f)
¼ Wclass,MAC(Jp, f) þ WMIC(Jp, f). The microscopic part WMIC
was previously analyzed in the framework of STL (Ref. 5)
and was shown to be equal to the sum of an hysteresis contri-
bution, an excess component, and a classical loss term deriv-
ing from the eddy currents circulating within the particles.10
It is frequently assumed that the measured material re-
sistivity can be directly used for the macroscopic loss com-
putation, assuming an equivalent homogeneous material.2
But the link between sample resistivity and macroscopic
classical loss is not obvious, because, as shown in Ref. 10,
percolation due to random contacts between particles plays a
role in highly compacted samples and interpretation of the
experiments calls for a specific model of conduction by ran-
dom contacts.10 A simpler approach is proposed here by
introducing the notion of equivalent resistivity for the loss
q(loss), i.e., the resistivity which would produce, in an homo-
geneous sample, the same macroscopic loss observed in the
SMC. Due to percolation, q(loss) is expectedly different from
the measured resistivity q, but we assume that proportional-
ity exists, so that we can write q(loss) ¼ Q(loss)  q, with Q(loss)
a phenomenological coefficient. It is verified that Q(loss)
depends only on the type of material and can be obtained
FIG. 3. As in Fig. 2 for the samples SMC2-t3 and SMC2-t1. The theoretical
DW is obtained using Q(loss) ¼ 1.56 in Eq. (3).
 17A322-3 Appino et al. J. Appl. Phys. 113, 17A322 (2013)

comparing two samples with different cross-sectional area. independent of the cross-sectional area in sufficiently big
we consider a ring sample with rectangular cross-section samples. This is an important point in the practical use of
(thickness t, width DR, and cross-sectional area Sc ¼ t  DR) this model. We observe in Figs. 2 and 3 the close behaviors
and obtain the macroscopic classical loss as of the experimental and the so calculated loss differences
DW versus f in the ring samples SMCi-t3 (i ¼ 1 or 2) and
Wclass;MAC ðJp ; f Þ SMCi-t1 (the experimental DW observed at f ¼ 0 being
 
1 2 1 DR related to the uncertainty associated with the determination
¼ 2p ðlossÞ
Kshape  Sc  Jp2 f ½J=kg; (2) of this quantity for the hysteresis loss component). The coef-
dQ q t
ficient Q(loss) can then be simply obtained, for a given mate-
where the parameter Kshape, which depends only on the rial, from the loss difference measured on two differently
width-to-thickness ratio DR/t, is computed using a finite ele- sized samples, an important result in view of loss prediction
ment method (the skin effect is neglected). Comparing two in practical cores.
samples (a) and (b) of the same material, Eq. (2) can be writ-
ten as
IV. CONCLUSION
ða;bÞ ðaÞ ðbÞ
DW ¼W W We have put in evidence the link between sample resis-
( ðaÞ ðaÞ ðbÞ ðbÞ
)
1 Kshape Sc Kshape Sc tivity and macroscopic classical loss in two different classes
¼ 2p2  Jp2 f ½J=kg: of commercial soft magnetic composites. An equivalent re-
QðlossÞ dðaÞ qðaÞ dðbÞ qðbÞ
sistivity for the magnetic losses, taking into account the
(3) effect of random interparticle contacts and percolation, has
been introduced besides the measured resistivity. It is a ma-
B. Validation of the macroscopic loss model terial related quantity, independent of the sample size.
In order to validate the macroscopic loss model, we
show that the dimensionless coefficient Q(loss) is independent 1
L. Li et al., IEEE Trans. Magn. 46(2), 574–577 (2010).
2
of sample shape and size and is only material dependent. M. Anhalt and B Weidenfeller, J. Magn. Magn. Mater. 304(2), e549–e551
Q(loss) is identified looking at the experimentally observed 3
(2006).
A. H. Taghvaei et al., Mater. Des. 30(10), 3989–3995 (2009).
loss difference between SMCi-t2 (i ¼ 1 or 2) and SMCi-t1 4
L. P. Lefebvre et al., J. Magn. Magn. Mater. 176(2), L93–L96 (1997).
samples. We find Q(loss) ¼ 1 for the material SMC1 and 5
O. de la Barrière et al., J. Appl. Phys. 109, 07A317 (2011).
Q(loss) ¼ 1.56 for SMC2. Since Q(loss) ¼ 1, the conventional
6
L. P. Lefebvre, S. Pelletier, B. Champagne, and C. Gelinas, “Effect of
resin content and iron powder particle size on properties of dielec-
approach invoking an equivalent homogeneous material2 is
tromagnetics,” Adv. Powder Metall. Part. Mater. 6, 20.47-20.61 (1996).
acceptable in calculating Wclass,MAC(Jp,f) in the material 7
C. Gelinas, S. Pelletier, P. Lemieux, and L. Azzi, “Properties and process-
SMC1. On the other hand, in the material SMC2 q(loss) is ing of improved smc materials,” in Proceedings of the 2005 International
higher than the measured resistivity.10 8
Conference on Powder Metallurgy and Particulate Materials, 2005.
M. De Wulf et al., J. Appl. Phys. 91, 7845–7847 (2002).
That the coefficient Q(loss) is, to good approximation, 9
H. Shokrollahi and K. Janghorban, J. Mater. Process. Technol. 187(1),
material dependent only can be understood in terms of local 1–12 (2007).
character of the random interparticle contacts, making Q(loss) 10
C. Appino et al., IEEE Trans. Magn. 48(11), 3470–3473 (2012).

View publication stats

Downloaded 01 Oct 2013 to 193.204.114.64. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://jap.aip.org/about/rights_and_permissions