Temperature Dependence of Core Loss in Cobalt Substituted Ni-Zn-Cu Ferrites
Temperature Dependence of Core Loss in Cobalt Substituted Ni-Zn-Cu Ferrites
Temperature Dependence of Core Loss in Cobalt Substituted Ni-Zn-Cu Ferrites
Ferrites
A. Lucas 1,2, R. Lebourgeois 1, F. Mazaleyrat 2, E. Labour 2,3
1. THALES R&T, Campus Polytechnique, 1 av. Augustin Fresnel, 91767 Palaiseau,
France
2. SATIE, ENS de Cachan, 61 av. du Prsident Wilson, 94235 Cachan, France
3. LGEP SUPELEC, Plateau de Moulon, 11 rue Joliot-Curie, 91192 Gif Sur Yvette
Keywords : NiZnCu ferrites, cobalt substitution, core loss versus temperature, permeability
1 Introduction
Nickel-zinc-copper ferrites are essential materials because of their high permeability in MHz
range. Moreover, their low sintering temperature makes them suitable for the realization of
integrated components in power electronics. As for nickel-zinc ferrites, cobalt substitution is an
efficient technique to decrease the permeability [1] and the magnetic losses of nickel-zinc-copper
ferrites [2]. The effect of cobalt is to allow the pinning of the domain wall by inducing a magnetic
anisotropy [3]. The consequence is a decrease of tan at high frequency [4] and also an
improvement of core loss [2]. The aim of this paper is to study the effect of cobalt substitution on
core losses versus temperature. A lot of papers investigated core losses of spinel ferrites versus
temperature but it was mostly for Mn-Zn power ferrites [5] and not at high frequency. There are
very few papers concerning core losses versus temperature of Ni-Zn or Ni-Zn-Cu ferrites [6][7],
moreover, they only investigated losses of a single composition measured at low frequency (50
kHz). In order to understand how the ferrite composition influences the core loss, ferrites with
three Ni / Zn ratios were studied (Ni / Zn = 0.43, 1 and 3) with, for each ratio, cobalt substitutions
up to 0.035 mol per formula.
2 Experimental procedure
2.1 Sample preparation
Ferrites were synthesized using the conventional ceramic route. The raw materials (Fe2O3, NiO,
ZnO, CuO) were ball milled for 24 hours in water. Co3O4 was then added before the calcination
around 800C in air for 2 hours. The calcined ferrite powder was then milled by attrition for 30
min. The resulting powder was compacted using axial pressing. The sintering was performed at
935C for 2 hours in air. Magnetic characterizations were done on ring shaped samples with the
following dimensions : outer diameter = 6.8 mm; inner diameter = 3.15 mm; height = 4 mm.
Ni0.24Zn0.56Cu0.20CoFe1.98O4-
Ni0.40Zn0.40Cu0.20CoFe1.98O4-'
Ni0.60Zn0.20Cu0.20CoFe1.98O4-''
Four formulations were done for each series, with cobalt rate of 0, 0.014, 0.028 and 0.035 mol
per formula. The copper rate was 0.20 mol to allow the densification below 950C. Three Ni / Zn
ratios were studied in order to investigate the influence of cobalt substitution on ferrite with
different magneto-crystalline anisotropy. Indeed, K1 of the host crystal increases with nickel
content [8].
The samples were sintered at 935C for 2 hours in air. Figure 1 shows the X-ray diffraction
pattern of a Ni0.40Zn0.40Cu0.20Fe2O4 ferrite sintered at 935C. Only the spinel structure can be
observed, none of the precursor oxides are present in the sintered materials.
The bulk densities of the sintered ferrites are shown in figure 2. All the ferrites have a low
porosity ( > 92 % of the theoretical density). The cobalt substitution does not seem to have a
significant effect on densification, a slight increase of the densification can however be noticed.
This phenomenon is probably a consequence of the slight increase in the iron default of the ferrite.
Microstructure of the ferrites were also observed, SEM pictures are shown in figure 3. The
grain size is not affected by the Ni / Zn ratio. Even if the nickel is known to lead to a more difficult
densification [9], it is not observed here. The cobalt rate does not change the microstructure. All
the ferrites exhibit the same kind of microstructure, with a low porosity and an average grain size
between 1.5 and 2 m. The grain size repartition is relatively homogenous, although one can see a
few larger grains with defaults inside.
3 M s2
s 1 = .
.D for the domain wall motion contribution [13]
16 K 1
2 M s2
s 1 .
3 K1
With D the average grain size and K1 the magneto-crystalline anisotropy of the ferrite. In this
study, we consider that the grain size is similar whatever the composition, so this parameter does
not affect the permeability. The variation of permeability versus temperature is then only
depending on the variations of Ms and K1 versus temperature.
As the saturation magnetization of these ferrites is monotonically decreasing versus
temperature, such a change in permeability is therefore due to a change in magnetic anisotropy. K1
of the Ni-Zn-Cu spinels is weak and negative (between 2 and 6,5.103 J/m3, for NiFe2O4,
ZnFe2O4 and CuFe2O4 [8]). The cobalt ferrite has a high and positive K1 (close to 300.103 J/m3 at
room temperature [8]). Addition of a small amount of cobalt will thus lead to a compensation of
K1. Looking at the previous relations, if there is a magneto-crystalline anisotropy compensation,
the permeability will go through a maximum. This explains the apparition of a local maximum in
s(T) curves of cobalt substituted ferrites at a compensation temperature (T0) increasing with the
cobalt content and K1 of the Ni-Zn-Cu host crystal.
The compensation temperature of the three series of ferrites are shown in table 2, these results
are established from s(T) curves (Figures 4.A1, B1 and C1). For a given Ni /Zn ratio, T0 increases
with the cobalt content. For the same cobalt rate, T0 increases with the nickel content, i.e. with
magneto-crystalline anisotropy of the Ni-Zn-Cu host crystal. At low temperature, the magnetocrystalline anisotropy is positive due to Co2+ ions contribution. At high temperature, Ni-Zn-Cuhost crystal contribution becomes preponderant and K1 is negative. The measurements are thus
coherent with the single-ion model theory.
The cobalt substitutions are also known to pin the domain walls and to lower the domain wall
displacements contribution to the permeability. This results in a decrease of s with the increase of
the cobalt content, as can be seen in figures 4.A1 and 4.B1. For the third series (Ni / Zn = 3, figure
4.C1), the magneto-crystalline compensation is particularly marked. The permeability of cobalt
substituted ferrites can then become higher than that of the cobalt-free ferrite (curve B, figure
4.C1). It seems that for these ferrites, a slight cobalt substitution leads to a smaller K1 than that of
the cobalt-free ferrite.
Temperature dependence of core loss
Core loss variations mainly depend on saturation magnetization and effective anisotropy (since
we consider that the microstructures are similar for all the materials). Indeed, it is known that
larger grain size can increase core loss [15]. For the cobalt-free ferrites these two parameters
decrease versus temperature.
The Ni0.24Zn0.56Cu0.20Fe1.98O4 ferrite (curve A of the figure 4.A2) has core loss decreasing until
20C and rapidly increasing above 120C. To understand this behaviour, we have to look at the
temperature dependence of the permeability. Until 80C, the anisotropy decreases faster than the
saturation magnetization, which results in an increase of the permeability and a decrease of the
core loss. Above 80C, the initial static permeability reaches 1000, shifting the resonance
frequency towards 3 MHz. As the measurements are done at 1.5 MHz, relaxation losses become
4
therefore predominant and lead to a fast increase of the core loss at high temperature.
When the nickel content increases, the Curie temperature rises. For the
Ni0.40Zn0.40Cu0.20Fe1.98O4- ferrite (curve A, figure 4.B2), Tc is around 340C. The core loss
measurements between 50 and 150C are continuously decreasing. This is again related to the
variation of permeability which constantly increases while maintaining a resonance frequency far
enough from 1.5 MHz.
For the Ni0.60Zn0.20Cu0.20Fe1.98O4 ferrites (curve A, figure 4.C2), one can note that the
permeability is almost constant in the studied temperature range. The core losses are thus nearly
stable between 50 and 150C, a slight decrease can be observed at high temperature. Concerning
the core loss values of the Ni-Zn-Cu ferrites, the higher the magneto-crystalline anisotropy, the
higher the core loss. The power losses at room temperature increase from 600 mW/cm3 to 2800
mW/cm3 when nickel content goes from 0.24 mol to 0.60 mol per formula.
Cobalt substituted ferrites have a completely different behaviour. The core loss goes through a
minimum around T0 (magneto-crystalline compensation temperature) and their values are also
highly reduced near this compensation compared to the cobalt-free ferrites. This phenomenon
appears to be similar to the K1 compensation observed in Mn-Zn spinels [12]. This compensation,
due to Fe2+ ions, also leads to a minimum in the core loss. If this is well known for Mn-Zn ferrites,
it is the first time that it is observed in Ni-Zn based ferrites.
Such low core losses around T0 are not only due to the K1 compensation but are also the result
of the cobalt induced anisotropy (Ku). Small amounts of Co2+ ions in Ni-Zn spinels are known to
pin the domain walls, however, the temperature dependence of this induced anisotropy is not well
known. In order to study this parameter, Core loss versus magnetic induction of the
Ni0.40Zn0.40Cu0.20Co0.028Fe1.98O4+ ferrite was measured at different temperatures (figure 5). The
cobalt substituted ferrites have not a classical behaviour of core loss versus induction. When
induction increases, the core loss increases almost linearly until a threshold induction
corresponding to the required energy to unpin the domain walls. Above this threshold induction
(which depends on K1 and Ku), the anisotropy induced by the cobalt is no longer efficient so the
core loss increases fast. This phenomenon appears clearly on figure 5 for the measurement at
20C, the core loss increases slightly up to 35 mT and then rises rapidly. When the temperature
increases, the threshold induction decreases and, for the measurement done at 100C, the core loss
nearly recovers a classical behaviour with an evolution proportional to the square of the magnetic
induction. This ferrite has a T0 close to 10C, therefore the measurements done on figure 5 tend to
prove that the cobalt induced anisotropy is more efficient around the magneto-crystalline
anisotropy compensation.
Consequently, around T0 the magnetic configuration is doubly favourable in order to have low
power loss with a weak effective anisotropy and an induced anisotropy that improve the linearity
of the ferrite. This can then lead to very low core loss of respectively 150 and 200 mW/cm3 (at 1.5
MHz, 25 mT and 20C) for Ni / Zn = 0.43 and Ni / Zn = 1 ferrites with cobalt content of 0.028
mol. The third series of ferrites have higher core loss with a minimum of 400 mW/cm3. It seems
that K1 of the Ni-Zn-Cu host crystal (which increases due to high nickel content) is too high to
have very low core loss.
3 Conclusion
The study of the temperature dependence of core loss of cobalt-substituted Ni-Zn-Cu ferrites
highlighted that the core losses were minimum around the magneto-crystalline anisotropy
compensation. This is extremely interesting for adapting the ferrite to the operating range
temperature for power applications. Today, power electronics need materials that can work at high
frequency and high temperature (> 80C). The ferrites studied in this paper can perfectly answer to
these issues. Indeed, for the Ni0.40Zn0.40Cu0.20CoFe1.98O4+ ferrites (figure 4.B2), core loss at 80C
can be divided by a factor 4 thanks to cobalt substitution. These materials represent also a real
improvement compared to the state of art Ni-Zn ferrite for radiofrequency, with core loss 4 times
lower than a commercial ferrite with s=120 (figure 6).
References
[1] T. Y. Byun, S. C. Byeon, K. S. Hong, Factors affecting initial permeability of Co-substituted
Ni-Zn-Cu ferrites, IEEE, vol 35, Issue 5, Part 2, pages : 3445-3447 (1999)
[2] R. Lebourgeois, J. Ageron, H. Vincent and J-P. Ganne, low losses NiZnCu ferrites (ICF8),
Kyoto and Tokyo, Japan (2000)
[3] L. Nel, J. Phys. Radium 13, 249 (1952)
[4] J. G. M. De Lau, A. Broese Van Groenou, J. De Phys. IV 38, C1-17 (1977)
[5] V. T. Zaspalis, E. Antoniadisa, E. Papazogloua, V. Tsakaloudia, L. Nalbandiana and C. A.
Sikalidisb, The effect of Nb2O5 dopant on the structural and magnetic properties of MnZn
ferrites, Journal of Magnetism and Magnetic Materials Volume 250, Pages 98-109, (2002)
[6] H. Su, H. Zhang, X. Tang, Y. Jing, Y. Liu, Effects of composition and sintering temperature
on properties of NiZn and NiCuZn ferrites Journal of Magnetism and Magnetic Materials 310,
p1721, (2007)
[7] Y. Matsuo, M. Inagaki, T. Tomozawa, and F. Nakao, IEEE Transactions On Magnetics, Vol.
37, n. 4, (2001)
[8] J. S. Smit and H. P. J. Wijn, Ferrites, Philips technical library (1961)
[9] J. Ageron ,Thesis (1999)
[10] G. K. Joshi, A. Y. Khot and S. R. Sawant, Magnetisation, curie temperature and Y-K angle
studies of Cu-substituted and non substituted Ni-Zn mixed ferrites, Solid State Communications,
vol 65 n12, 1593-1595 (1988)
[11] E. Rezlescu, L. Sachelarie, P. D. Popa, and N. Rezlescu, Effect of Substitution of Divalent
Ions on the Electrical and Magnetic Properties of NiZnMe Ferrites, IEEE transactions on
magnetics, vol 36 n6 (2000)
[12] H. Pascard, Basic concepts for high permeability in soft ferrites, J. Phys. IV France, 377-384
(1998)
[13] R. Lebourgeois, C. Le Fur, M. Labeyrie, M. Pat, J-P. Ganne, Permeability mechanisms in
high frequency polycrystalline ferrites, Journal of Magnetism and Magnetic Materials, Volume
160, Pages 329-332,(1996)
[14] J. L. Snoek, Physica 14, 207 (1948)
[15] R. Lebourgeois, S. Duguey, J.-P. Ganne, J.-M. Heintz, Influence of V2O5 on the magnetic
properties of nickelzinccopper ferrites, Journal of Magnetism and Magnetic Materials, Volume
312, Issue 2, Pages 328-330, (2007)
Ms
(emu/g)
65.5
Ferrites
Ni/Zn=0.43
Co=0
Ni/Zn=0.43
Co=0.014
65.5
Ni/Zn=0.43
Co=0.028
64.4
Ni/Zn=0.43
Co=0.035
65.2
Ni/Zn=1
Co=0
79.8
Ni/Zn=3
Co=0
72.4
% Cobalt
T0
Ni /Zn = 0.43
T0
Ni /Zn = 1
T0
Ni /Zn = 3
0.014
-45C
-50C
-28C
0.028
-7C
5C
30C
0.035
3C
17C
50C
Figure 2 : Density of the three series of ferrites sintered at 935C for 2 hours in air.
Ni/Zn=0.43 _ Co=0
Ni/Zn=1 _ Co=0
Ni/Zn=1 _ Co=0.028
Figure 3 : SEM Micrographics of three materials sintered at 935C for 2 hours in air.
Ni0.24Zn0.56Cu0.20CoFe1.98O4+
(4.A1)
(4.A2)
Ni0.40Zn0.40Cu0.20CoFe1.98O4+
(4.B1)
(4.B2)
Ni0.60Zn0.20Cu0.20CoFe1.98O4+
(4.C1)
(4.C2)
Figure 4 : Initial static permeability versus temperature and core loss at 1.5 MHz and 25 mT versus
temperature
of
Ni0.24Zn0.56Cu0.20CoFe1.98O4+,
Ni0.40Zn0.40Cu0.20CoFe1.98O4+
and
Ni0.60Zn0.20Cu0.20CoFe1.98O4+ ferrites
10
Figure 5 : Core loss of the Ni0.40Zn0.40Cu0.20Co0.028Fe1.98O4+ ferrite measured at 1.5 MHz versus magnetic
induction for different temperatures.
Figure 6 : Core loss versus temperature measured at 1.5 MHz and 25mT. Comparison between a commercial NiZn ferrite ( s=120) and the Ni0.40Zn0.40Cu0.20Co0.028Fe1.98O4+ ferrite ( s=130)
11