Origin of ferromagnetism in polycrystalline Ca1+δB 6 (-0.05<δ<0.05) ceramics

Article abstract of DOI:10.1016/j.physb.2005.08.001

We have investigated systematically the relationship of ferromagnetic properties of polycrystalline Ca_1+δB₆ ceramics with Ca content, as δ ranges from -0.05 to 0.05. We demonstrated that the origin of the high-temperature ferromagnet

Full text of DOI:10.1016/j.physb.2005.08.001

ARTICLE IN PRESS
Physica B 369 (2005) 39–43
www.elsevier.com/locate/physb
Origin of ferromagnetism in polycrystalline Ca1+dB6
(ꢀ0:05odo0:05) ceramics
Ã
Ã
Minghe Caoa,b, , Jun Jiangb, Hanxing Liua, Jun Yuanb,
aState Key Lab of Advanced Technology for Materials Synthesis and Processing, Wuhan University of Technology,
Wuhan 430070, PR China
bDepartment of Materials Science & Engineering, Tsinghua University, Beijing 100084, PR China
Received 14 April 2005; received in revised form 1 August 2005; accepted 2 August 2005
Abstract
We have investigated systematically the relationship of ferromagnetic properties of polycrystalline Ca1+dB6 ceramics
with Ca content, as d ranges from ꢀ0.05 to 0.05. We demonstrated that the origin of the high-temperature
ferromagnetism observed in CaB6 can be ascribed to the existence of Fe containing impurity phases. For do0 (Ca
deficiency) Fe exists as boride, whereas for d40 (excess Ca) it is present as oxide. The dependence of magnetization with
the nominal stoichiometry of Ca1+dB6 can be understood from the evolution of Fe boride and oxide phase in the
materials.
r 2005 Elsevier B.V. All rights reserved.
Keywords: CaB6; Ferromagnetism; Fe impurity
1. Introduction
orbits in CaB6 [3]. In single crystal CaB6, it is
reported that only for do0 Ca1+dB6 is ferromag-
netic; for d40 (excess Ca) no trace of ferromagnet-
ism is found [4,5]. It has been suggested by
Moriwaka that Ca vacancies may induce this
ferromagnetism [2]. More recent works [6–8]
indicate that the effect is due to the presence of
iron in samples, although the nature of iron in the
materials has not been properly investigated. In
some studies [6] treatment of the CaB6 samples with
hydrochloric acid eliminated ferromagnetism,
which was correlated with removal of iron from
the samples. Matsubayashi [8] suggested that iron
A recent report that La-doped CaB6 is a
ferromagnet with a Curie temperature of 600 K
has generated a great deal of interest [1]. Later,
polycrystalline CaB6 is also found to have this effect
[2] even though it has no partially filled d- or f-
Ã
Corresponding authors. Materials Science and Engineering,
State Key Lab of Advanced Technology for Materials Synthesis
and Processing, Wuhan 430070, PR China.
Tel.: +8602787864681; fax: +8602787879468.
E-mail addresses: caominghe@tsinghua.org.cn (M. Cao),
yuanjun@tsinghua.edu.cn (J. Yuan).
0921-4526/$ - see front matter r 2005 Elsevier B.V. All rights reserved.
doi:10.1016/j.physb.2005.08.001

ARTICLE IN PRESS
40
M. Cao et al. / Physica B 369 (2005) 39–43
was present as the borides FeB and Fe2B, whereas
Meegoda [7] believes that the iron was present as
Fe–O phase. In a previous paper [9], we have shown
that Fe is present as boride in Fe-doped CaB6.
Here, we present a study of the evolution of Fe
containing phases with the relative Ca content in
Ca1+dB6. We showed that the dependence of
magnetization with Ca content may be accounted
for by the evolving Fe-containing secondary phase.
2. Experimental
We synthesized calcium hexaboride ceramic
materials according to the following oxidation
reduction reaction:
Fig. 1. Saturation moment per gram at T ¼ 300 K as
a
function of Ca content in Ca1+dB6.
ð1 þ dÞCaO þ ð7 þ dÞB ! Ca1þdB6 þ ð1 þ dÞBO:
(1)
Starting materials include 99.99% pure CaO
and B powders. They were mixed and milled for
2 h in alcohol, then dried and granulated with
glycerine as the binder. The mixture was pressed
into a cylindrical-shaped tablet in a pressure cell.
The tablets, with typical dimensions of 10 mm in
diameter and 2 mm in thickness were placed on an
Al2O3 plate and heated in an Ar atmosphere at
1450 1C for 4 h. They were then cooled down to
room temperature in the furnace. The atomic
structure of the sample is characterized by X-ray
powder diffractometry (Automated D/Max-RB)
using CuKa radiation. The chemical composition
of these samples was analyzed with an inductively
coupled plasma (ICP) method using a OPTIMA
3300RL atomic emission spectrometer. The mag-
netization was measured by an alternating gradi-
ent magnetometer (AGM) of MicroMagTM 2900.
Using a JEOL 2010F high-resolution transmission
electron microscope operating at 200 keV com-
bined with electron energy-loss spectroscopy
(EELS), the mappings of Ca or B elements in the
samples are determined.
Fig. 2. Fe content in Ca1+dB6 samples measured by ICP.
denoted Ca1+dB6 (d from ꢀ0.05 to 0.05). The
results indicate that the saturated magnetic mo-
ment of these samples depends sensitively on the
nominal Ca content, being strongest for a slightly
‘Ca-deficient’ sample (d ¼ ꢀ0:01). This is compar-
able to the result found in single crystals [1,5].
Using ICP, we measured the Fe content in our
samples. The results are shown in Fig. 2. It
indicates that there is ꢁ400 ppm Fe in these
samples. The results show that the Fe content is
almost the same in all samples. According to the
analyses mentioned above, the ferromagnetism of
Ca1+dB6 originates from the Fe impurity, which
indicates that the form of Fe in Ca1+dB6 must be
3. Results and discussion
Fig. 1 shows the saturation moment per gram of
samples which have
a nominal composition

ARTICLE IN PRESS
M. Cao et al. / Physica B 369 (2005) 39–43
41
Fig. 3. X-ray diffraction pattern of Ca1+dB6 samples whose
nominal composition are (a) Ca0.96B6 (b) CaB6 and (c) Ca1.04B6.
Fig. 4. Lattice constant of Ca1+dB6 samples measured by
X-ray diffraction.
different from that in Ca1ꢀdB6 samples, as the
results of the saturation moments in Ca1ꢀdB6 are
larger than that in Ca1+dB6 samples. Fig. 3
presents the X-ray diffraction patterns of the
samples with nominal composition Ca0.96B6,
CaB6 and Ca1.04B6, respectively. From these
patterns, we can see that the host material for
the present samples Ca1+dB6 is the divalent
alkaline-earth hexaboride CaB6. All the peaks of
CaB6 in the Ca1+dB6 were indexed to a cubic
structure and the lattice constant was determined
with (3 1 0) and (3 0 0) peaks using a least-squares
fitting program. The lattice constants of all the
samples are almost the same (as shown in Fig. 4).
So, it is inferred that the structure of the CaB6
phase in Ca1+dB6 samples are the same. At the
same time, a small amount of a secondary phase
was found in the Ca0.96B6 and Ca1.04B6 samples.
The secondary phase in Ca0.96B6 is indexed
to B7O, while that in Ca1.04B6 is indexed to
CaB2O5. As the host product is CaB6 phase, we
think that the ferromagnetism cannot originate
from these secondary phases. As we mentioned
above, the magnetic properties vary significantly
with Ca content. If the origin of the magnetic
property is due to the CaB6 phase, the maximum
moment should be in the sample with d ¼ 0. In
fact, we find the sample with deficient Ca at 0.01
(e.g. Ca0.09B6 sample) has the largest magnetic
moment.
To find out the nature of Fe in ceramics, we
examined the microstructure of calcium hexabor-
ide. Fig. 5 is a typical element mapping of Ca and
B elements in Ca0.96B6 measured by HREM
combined EELS. Fig. 5a is the mapping of B in
samples with nominal composition Ca0.96B6. The
contrast of grain C is bright, which indicates that
there is B element in this grain. Fig. 5b is the
mapping of Ca. The contrast of grain C is black,
which indicate that there is no Ca element in this
grain. It can be seen that grains A and B are
composed by Ca and B elements, while grain C is
composed mainly by B element. So, it is confirmed
that B atoms are segregated in the samples with Ca
deficiency. In the samples with excess Ca, we did
not find any evidence of boron segregation. We
think that the surplus Ca atoms in Ca1+dB6
samples may be lost through evaporation or are
combined to form Ca–B–O phase. As the result,
we can see that the reaction products of deficient
CaO with B are composed with CaB6 and surplus
B, while that of excess CaO with B are composed
with CaB6 and the surplus Ca may be bound in a
Ca–B–O phase. It is difficult to form the Ca
vacancy or B vacancy defects in CaB6 samples.
According to the results, the oxidation–reduction
(1) should be written as follows:
ð1 þ dÞCaO þ ð7 þ dÞB !
CaB6 þ BO þ dðCa ꢀ B ꢀ OÞ ðd40Þ,
ð2Þ

ARTICLE IN PRESS
42
M. Cao et al. / Physica B 369 (2005) 39–43
Fig. 5. B and Ca element mapping of a Ca0.96B6 sample: (a) B mapping and (b) Ca mapping.
ð1 þ dÞCaO þ ð7 þ dÞB !
ð1 þ dÞCaB6 þ ð1 þ dÞBO ꢀ 6dB ðdo0Þ.
ð3Þ
Given that Fe is known to be related to
ferromagnetic properties in nearly stoichiometric
calcium hexaboride, we speculated about the
origin of the ferromagnetic variation in Ca1+dB6
samples. In the samples with Ca deficiency, as
there is surplus B segregation, the Fe–B phase will
be formed easily in the sample. However, with
increasing Ca deficiency, the ferromagnetic prop-
erties are diluted by more and more surplus B
segregated. Through Fe diluting into Boron and
sintering at 1450 1C in a Ar atmosphere to form
FeBx, we find the magnetic moment per Fe atom
in FeBx reducing with the increase of B content; it
is consistent with the results in Fig. 1 that the
saturation moment per gram in Ca1ꢀ|d|B6 samples
is reducing with the reduction of Ca content in
CaB6. However, in the Ca1+|d|B6 samples, because
there is no surplus B phase, it is more difficult to
form Fe–B phase. As the other role of excess B is
to reduce the CaO, B-deficiency may lead to
incomplete reduction of the oxide. Fe impurities
may form Fe–O phases, many of which are either
non-magnetic or antiferromagnetic. This might
explain the observation that one does not see a
significant ferromagnetic effect in nominal Ca-rich
boride samples.
Fig. 6. Dependence of the saturation moment MS of 0.05%
Fe-doped Ca1+dB6 samples on d.
ICP, we measured the Fe content in these samples
is about 1000 ppm. The variation of the saturated
magnetic moment with x in Fig. 6 has the same
trend as that in Fig. 1. Among those results, the
saturated magnetic moment of the sample of
Ca0.9B6 is the largest. To the Ca1ꢀdB6 or Ca1+dB6
samples, the saturate moment reduces with the
increasing x, in which the saturate moment of
Ca1ꢀdB6 is larger than that of Ca1+dB6. Fig. 7
shows the X-ray diffraction patterns of the
samples (CaB6 Ca0.9B6 Ca1.1B6 and Ca1.15B6) with
0.05% Fe. From the results, we can see that the
CaB2O4 phase can be present in CaB6, Ca1.1B6 or
Ca1.15B6 but that it is absent in the Ca0.9B6 sample.
With the increase of d in the Ca1+dB6 samples with
To test out this model, we prepared a series of
Ca1+dB6 (ꢀ0:2odo0:2) samples using the same
starting materials doping into a constant but small
amount of Fe (0.05%) and the ferromagnetic
properties of the samples were analyzed. Using

ARTICLE IN PRESS
M. Cao et al. / Physica B 369 (2005) 39–43
43
4. Conclusions
In summary, we have investigated the saturation
magnetization of Ca1+dB6 samples. B containing
phases are segregated in samples with Ca defi-
ciency, while oxides are segregated in samples with
excess Ca. We suggest that Fe–B phases are
formed in the samples with Ca deficiency, and
that these phases are the origin of ferromagnetism
in CaB6. Because no Fe–B phase can be formed
easily in Ca excess samples, and the Fe oxides
formed in that region have no (or a smaller)
magnetic saturation, the Ca-rich samples exhibit
no ferromagnetism, or—at higher Fe concentra-
tions, a less significant ferromagnetic behavior to
that of the Ca-poor samples.
References
Fig. 7. XRD pattern of Ca1+dB6 samples: (a) CaB6 (b) Ca0.9B6
(c) Ca1.1B6 and (d) Ca1.15B6.
[1] D.P. Young, D. Hall, M.E. Torelli, Z. Fisk, J.L. Sarrao,
J.D. Thompson, H.R. Ott, S.B. Oseroff, R.G. Goodrich, R.
Zysler, Nature (London) 397 (1999) 412.
[2] T. Moriwaka, T. Nishioka, N.K. Sato, J. Phys. Soc. Jpn. 70
(2001) 341.
[3] H.J. Tromp, P. van Gelderen, P.J. Kelly, G. Brocks, P.A.
Bobbert, Phys. Rev. Lett. 87 (2001) 016401.
[4] D.P. Young, Z. Fisk, J.D. Thompson, H.R. Ott, S.B.
Oseroff, R.G. Goodrich, Nature (London) 420 (2002) 144.
[5] P. Vonlanthen, E. Felder, L. Degiorgi, H.R. Ott, Phys. Rev.
B 62 (2000) 10076.
excess Ca, the amount of CaB2O4 phase is
increased. Due to the fact that the Fe content is
small, the peaks of phases with Fe cannot be
detected. Owing to the Ca–B–O phase in the
Ca1+dB6 samples, the Fe atoms may partially be
bound into the Fe–O phase. As the magnetization
of Fe–O phases is weaker than that of the Fe–B
phases which may be formed in the samples with
Ca deficiency (due to the surplus B in these
samples), the saturation moment of the samples
with excess Ca decreases faster than that in the
samples with Ca deficiency.
[6] K. Matsubayashi, M. Maki, T. Tsuzuki, T. Nishioka, N.K.
Sato, Nature 420 (2002) 143.
[7] C. Meegoda, M. Trenary, T. Mori, S. Otani, Phys. Rev. B
67 (2003) 172410.
[8] K. Matsubayashi, M. Maki, T. Moriwaka, T. Tsuzuki, T.
Nishioka, C.H. Lee, A. Yamamoto, T. Ohta, N.K. Sato,
J. Phys. Soc. Jpn. 72 (2003) 2097.
[9] M.H. Cao, J. Jiang, H.X. Liu, J. Yuan, Physica B 364
(2005) 150.