Magnetic disorder in the Cu0.995Fe0.005O solid solution

Article abstract of DOI:10.1016/S0304-8853(98)00353-9

The magnetic hyperfine field, measured by Moessbauer spectroscopy, of a Cu(Fe)O solid solution displays a spin-glass-like behaviour that undergoes two transitions. The samples were produced by a 48 h ball-milling and 40 h successive annealing treatments at 650, 700 and 800 K with 0.25 mol% of α-⁵⁷Fe₂O₃ and CuO. The signal shows a magnetic splitting that develops at temperatures lower than ca. 150 K. The measured distribution of hyperfine fields broadens at lower temperatures and its behaviour down to 15 K extrapolates to a saturation field of ≈29 T. The second transition takes place at temperatures between 4.2 and 15 K. The observed magnetic behaviour is interpreted in terms of magnetic disorder and canted local states of the system of magnetic moments.

Full text of DOI:10.1016/S0304-8853(98)00353-9

Journal of Magnetism and Magnetic Materials 192 (1999) 77—82
Magnetic disorder in the Cu
Fe
O solid solution
ꢀꢁꢂꢂꢃ ꢀꢁꢀꢀꢃ
S.J. Stewart, R.A. Borzi, R.C. Mercader*
Departamento de Fı&sica, Universidad Nacional de La Plata, C.C. 67, 1900 La Plata, Argentina
Received 23 June 1998
Abstract
The magnetic hyperfine field, measured by Mossbauer spectroscopy, of a Cu(Fe)O solid solution displays a spin-glass-
like behaviour that undergoes two transitions. The samples were produced by a 48 h ball-milling and 40 h successive
ꢃꢄ
annealing treatments at 650, 700 and 800 K with 0.25 mol% of a- Fe O and CuO. The signal shows a magnetic
ꢅ
ꢆ
splitting that develops at temperatures lower than ca. 150 K. The measured distribution of hyperfine fields broadens at
lower temperatures and its behaviour down to 15 K extrapolates to a saturation field of +29 T. The second transition
takes place at temperatures between 4.2 and 15 K. The observed magnetic behaviour is interpreted in terms of magnetic
disorder and canted local states of the system of magnetic moments. ( 1999 Elsevier Science B.V. All rights reserved.
PACS: 76.80.#y; 75.30.!m; 75.30.Hx
Keywords: Magnetic ordering; Cupric oxide; Spin canting; Reentrant behaviour; Mossbauer spectroscopy
1. Introduction
temperatures higher than 230 K, but diverging re-
sults have been reported about its low-temperature
behaviour, and, in addition, no coherent explana-
tions are found about the results obtained [1—3].
Towards understanding the magnetic behaviour,
in a previous work [4] we have reported the suc-
cessful insertion of a hyperfine probe that senses the
magnetic ordering of tenorite (CuO). In this work
we report further Mossbauer measurements on
Due to the close connection of CuO with the
superconducting planes in high ¹ cuprates, a signi-
ficant number of investigations have been per-
formed to disclose the phenomena responsible for
the unusual magnetic properties of cupric oxide.
There is consensus on the low-dimensional origin
of the magnetic interactions that lead to the non-
Curie—Weiss dependence of its susceptibility at
#
ꢃꢄ
Fe probes in intimate contact with the Cu and
O atoms in a Cu—Fe—O solid solution displaying
magnetic disorder features. These are discussed in
terms of a model that contemplates the formation
of canted local states in the system. The possible
influence of impurity atoms on the magnetic prop-
erties is considered.
*Corresponding author. Tel.: #54-21-24-6062; fax: #54-
21-25-2006; e-mail: mercader@venus.fisica.unlp.edu.ar.
0304-8853/99/$ — see front matter ( 1999 Elsevier Science B.V. All rights reserved.
PII: S 0 3 0 4 - 8 8 5 3 ( 9 8 ) 0 0 3 5 3 - 9

78
S.J. Stewart et al. / Journal of Magnetism and Magnetic Materials 192 (1999) 77—82
2. Experimental
broad doublet that can be fitted assuming a distri-
bution of quadrupole sites. An average D value of
1.16 mm/s is obtained. A slight broadening in the
spectra is noticed when lowering the temperature.
The Cu(Fe)O solid solution was obtained by
milling for 48 h powders of tenorite (CuO, 99.999%
purity) and 0.25 mol% of hematite (a-Fe O ,
It gets more noticeable at a temperature ¹ of ca.
ꢅ ꢆ
ꢀ
ꢃꢄ
99.99% purity) enriched in Fe to +95%, with
150 K, indicating the onset of a magnetic tran-
subsequent thermal treatments in air up to 800 K.
The milling was carried out in air in a cylindrical
steel vial with one steel ball using a 24 Hz vibratory
mixer mill (Retsch MM2). The ball-to-powder
weight ratio was 14 : 3. The annealings were per-
formed cumulatively in air at 650, 750, and 800 K.
The heat-treatment followed a programmed cycle
of three steps. In the first one, the sample was
heated at a constant rate of 8 K/min; the second
step was an annealing for 40 h; and the third one
was a cooling stage at a rate of 1 K/min down to
room temperature.
sition. This was confirmed measuring the number
of counts at zero velocity as a function of temper-
ature. The transition is not sharp but develops in
a range of tens of degrees.
The ¹ +150 K value is within the range of
ꢀ
other transition temperatures observed in pre-
vious Fe-doped cupric oxide studies [5—7].
Indeed, Sohma et al. [5], in Fe-doped CuO thin
films, estimated an ordering temperature in the
150—180 K range. Using Mossbauer spectroscopy,
Masterov et al. [7] found a non-Lorentzian
broadening in their lines at 150 K, in 0.6 at% Fe-
doped CuO obtained by co-precipitation.
The Mossbauer transmission measurements
ꢃꢄ
were performed using a 50 mCi Co source in a
When dealing with spectra that show both site
distributions and relaxation, there is no fitting pro-
cedure that can deconvolute the contributions from
each phenomenon. This is the situation that seems
to be taking place in our spectra at temperatures
lower than 150 K, where the magnetic splitting sets
in (Fig. 1). For simplicity we have assumed a distri-
bution of static hyperfine fields over the whole
range of temperatures (Fig. 2).
Rh matrix, with a multiscaler of 512 channels in
constant acceleration mode. A Displex helium
closed-cycle cryogenic refrigerator was used for
measurements in the 15—300 K range. The 4.2 K
spectrum was taken in a liquid helium cryostat
both with the source and the sample at the same
temperature and a sinusoidal velocity waveform.
The isomer shifts are referred to a-Fe at room
temperature. Thermal scanning at zero velocity was
measured with the drive stationary after discon-
nection of the servo-amplifier.
Fig. 3 shows the average hyperfine field B and
)&
its standard deviation (p) obtained from the fits in
the range from 4.2 to 80 K, where the paramagnetic
contribution is minimal. Below 80 K the distribu-
tions become extremely broad, as can be seen in
Figs. 2 and 3. We can appreciate that there are no
3. Results and discussion
marked changes in B nor in the width of the
)&
We showed earlier [4] that after milling and
distribution in the 15—37 K range.
further annealing of CuO and 0.25 mol% of a-
At 4.2 K a much narrower distribution with an
Fe O up to 800 K, the Fe atoms form mainly
a Cu(Fe)O solid solution. The residual ca. 5% of
average B "42.6 T is obtained. This sudden
ꢅ ꢆ
)&
change in the rate increase of B together with the
)&
the Fe signal corresponds to hematite particles that
undergo the Morin transition and belong therefore
to particles of a size greater than 200 A that have
not reacted. This signal is observed for the spectra
at all temperatures and its contribution is a con-
stant proportion (within the statistical errors of our
data) of the total Mossbauer spectra.
decline in p as the temperature decreases has been
previously observed in semi-disordered magnetic
systems [8]. Because of the almost constant value
of p at temperatures between 15 and 80 K (see Fig.
3b), and the drastic drop between 4.2 and 15 K, the
second magnetic transition can be estimated to
occur within this range.
The room temperature Mossbauer spectrum as-
A model that can explain the emergence of the
ꢆ>
sociated with the solid solution displays an Fe
second transition in the B (¹) curve, is that of
)&

S.J. Stewart et al. / Journal of Magnetism and Magnetic Materials 192 (1999) 77—82
79
serted and collinearity is broken, the canted spins
can rotate around z. A certain long range order
remains along this direction, while the transversal
spin components average to zero within the experi-
mental characteristic time. At sufficiently low tem-
peratures the transverse spin components freeze in
random directions due to the interaction between
CLS [10], and the total spin value is recovered.
The existence of non-satisfied bonds is an essen-
tial but not sufficient condition for these semi-
disordered states to arise. Magnetic frustration has
been mentioned to be present in CuO by several
authors [3,11,12] as a result of its magnetic struc-
ture and topology. Indeed, Kondo et al. [11]
contemplate a non-collinear structure due to
competing exchange interactions along the [1 1 0]
and [1 1 0] directions. Muraleedharan et al. [13]
considered that the existence of randomly canted
states in the CuO matrix, similar to that found in
disordered spinel ferrites [9], could explain their
results. These authors linked the canted state with
the breaking of antiferromagnetic correlations be-
ꢅ>
tween Cu neighbour ions because of crystal de-
fects. Furthermore, the low-dimensional magnetic
character of CuO [2,11,12,14,17] could favour the
arising of a spin canted state, since this one is more
difficult to emerge in 3D systems because of the
greater number of magnetic nearest neighbour
atoms [10].
Many works have found magnetic semi-disorder
features in doped YBa Cu O
and La CuO
ꢅ
ꢆ ꢇ>V
ꢅ
ꢈ>V
[15]. In such systems, localised holes generate a lo-
cal ferromagnetic exchange interaction between the
otherwise antiferromagnetically coupled Cu ions,
yielding frustration and spin-glass-like behaviour
[15,16]. These compounds are frequently cited in
CuO papers because of their similar magnetic
properties (see, e.g., [17,18]), that initially worked
as a motivation to study CuO magnetic properties
[13,19].
Fig. 1. Mossabauer spectra of Cu
ꢃꢄFe
O solid solution
ꢀꢁꢂꢂꢃ
ꢀꢁꢀꢀꢃ
at temperatures between 4.2 and 295 K. The solid line is the
least-squares fit to the data performed as described in the text.
The contributions of hematite and of the distribution of hyper-
fine fields are shown as dotted lines.
canted local states (CLS). These can appear, as
observed in several substituted compounds [8,9],
when a magnetic atom is substituted by non-mag-
netic atoms, or by magnetic atoms of a different
effective moment, or by the presence of interstitial
impurities in an antiferromagnet [10]. In this
model the system has, in the absence of impurities,
a conventional long range order with the spins
along the axis of easy magnetisation z (if a collinear
picture is supposed). When the impurities are in-
In the Cu
Fe
O solid solution,the frus-
ꢀꢁꢂꢂꢃ ꢀꢁꢀꢀꢃ
tration effects might be enhanced by the presence of
impurities in the CuO structure (if unsatisfied
bonds were already present in the undoped sample)
or might be created by the doping. As mentioned
above, in the model of Villain [10], a small number
of substitutional or interstitial impurities can trans-
form an originally collinear ground state into

80
S.J. Stewart et al. / Journal of Magnetism and Magnetic Materials 192 (1999) 77—82
Fig. 2. Hyperfine field distributions obtained after fittings performed as described in the text for spectra between 4.2 and 160 K.
a canted one. Indeed, iron doping has been claimed
to be the source of semi spin-glass behaviour con-
currently with a random freezing of Fe spins in
to be entirely dictated by the host. Using Barcs
et al.’s notation [22], these assumptions allow us to
express the hyperfine field as B "B B (z),
)&
ꢀ ꢃꢋꢅ
YBa (Cu
Fe
)O by Mirebeau et al. [20] in
where B and z"gꢀcꢀ/gc are fitting parameters,
ꢅ
ꢀꢁꢉꢉ ꢀꢁꢊꢅ ꢇꢁꢃ
ꢀ
neutron diffraction and Mossbauer spectroscopy
experiments.
B is the Brillouin function, g and gꢀ the g-factors,
and c and cꢀ are the molecular constants at the Cu
In the collinear region, we can roughly estimate
and Fe sites, respectively. The best fit is achieved
the longitudinal spin value S and the strength of
the Fe—Cu interaction compared with that of
with B "29 T and z"0.1. Assuming that g and gꢀ
ꢁ
ꢀ
are similar, the z value obtained would represent
Cu—Cu. To this end, we formulate the Jaccarino
et al.’s [21] mean field model in the same way as
Barcs et al. [22] and Niedermayer et al. [23], where
the magnetisation at the impurity site is supposed
a likely loose superexchange linkage between the
iron ion and the tenorite lattice. This can be ex-
plained by an indeed low J
or by competing
!ꢂ—$%
interactions leading to a small Weiss field at Fe

S.J. Stewart et al. / Journal of Magnetism and Magnetic Materials 192 (1999) 77—82
81
been proposed by Viglin et al. [1] for CuO, and by
Bocquet et al. [24] for goethite, as responsible for
cluster isolation. Both oxygen [1,13,18,19,25] and
cation vacancies [7,26—28] have been suggested to
be present as tenorite natural defects. Furthermore,
for Fe-doped samples, Masterov et al. [7] asso-
ꢆ>
ciate the existence of cationic vacancies with Fe
doping.
4. Conclusions
A semi-disordered spin state that freezes at some
temperature between 4.2 and 15 K has been ob-
served by Mossbauer hyperfine field measurements
in Fe-doped cupric oxide. Such behaviour is ana-
logous to that of several similar systems of localised
spins. The present results do not allow to conclude
unambiguously if this kind of spin ordering is in-
nate to CuO, as suggested by previous works, or
has been introduced by the iron doping.
Fig. 3. (a) Average hyperfine field B versus temperature, and
(b) standard deviation, p versus temperature, in the range
4.2—80 K. The solid line is the fit obtained as described in the text
using the model of Ref. [22]. The dashed lines are a guide for the
eye.
)&
positions. From the B and B (4.2 K) values, a ra-
ꢀ
)&
tio of S /S+0.68 can be crudely estimated.
Acknowledgements
ꢁ
The Fe spin freezing occurs at some temperature
between 4.2 and 15 K. The B "42.6 T measured
Partial economic support by CONICET, Argen-
tina, through PICT0201 and TENAES is acknow-
ledged. R.A.B. is fellow of CONICET and R.C.M. is
member of Carrera del Investigador Cientıfico,
CONICET. We thank Prof. Rosa Scorzelli, Centro
Brasileiro de Pesquisas Fisicas, Rio de Janeiro,
Brazil, for the measurement at 4.2 K.
)&
ꢆ>
at 4.2 K corresponds to Fe in a high spin state.
Although this value agrees with that observed in
Fe-doped thin films, the broad lines at 4.2 K are
indicative of different Fe environments in the CuO
lattice. This might involve an Fe site distribution
caused by low crystallinity induced during the
sample preparation and/or, in addition, a contribu-
tion caused by the small size of the particles in
which the system is divided [4].
References
There is another model [8] that could explain
the present results exhibiting two magnetic tran-
sitions. It is an inhomogeneous model based on
magnetically isolated clusters coexisting with an
ordered matrix. Impurity atoms, or vacancies,
might act as centres that disturb the superexchange
paths and isolate magnetically small clusters of
atoms. The Fe—Cu—O system studied in this work
has a very dilute concentration of iron atoms. For
this reason, we think that within this model, it is
unlikely that Fe atoms would operate as impurities
causing the observed results. However, the model
cannot be totally ruled out, since vacancies have
[1] N.A. Viglin, S.V. Naumov, A.A. Samokhvalov, Sov. Phys.
Solid State 38 (1996) 706.
[2] Y.G. Raydugin, V.E. Naish, E.A. Turov, J. Magn. Magn.
Mater. 102 (1991) 331.
[3] F. Marabelli, G.B. Parravicini, P. Wachter, Solid State
Commun. 86 (1993) 131.
[4] S.J. Stewart, R.A. Borzi, G. Punte, R.C. Mercader, Phys.
Rev. B 57 (1998) 4983.
[5] M. Sohma, K. Kawaguchi, Solid State Commun. 79
(1991) 47.
[6] P. Shah, A. Gupta, Phys. Rev. B 45 (1992) 483.
[7] V.F. Masterov, F.S. Nasredinov, G.T. Daribaeva, V.F.
Kovelev, P.P. Seregin, N. Tritskaya, Sov. Phys. Solid State
33 (1991) 1525.

82
S.J. Stewart et al. / Journal of Magnetism and Magnetic Materials 192 (1999) 77—82
[8] J.L. Dormann, Hyperfine Interactions 68 (1991) 47.
[9] R.A. Brand, H. Georges-Gilbert, J. Hubsch, J.A. Heller,
J. Phys. F 15 (1985) 1987.
[20] I. Mirebeau, C. Bellouard, M. Hennion, G. Jehanno, V.
Caignaert, A.J. Dianoux, T.E. Phillips, K. Moorjani,
Physica C 184 (1991) 299.
[10] J. Villain, Z. Phys. B 33 (1979) 31.
[11] O. Kondo, M. Ono, E. Sugiura, K. Sugiyama, M. Date,
J. Phys. Soc. Japan 57 (1988) 3293.
[12] J.B. Forsyth, P.J. Brown, B.M. Wanklyn, J. Phys. C 21
(1988) 2917.
[21] V. Jaccarino, R.L. Walker, G.K. Wertheim, Phys. Rev.
Lett. 13 (1964) 752.
[22] A. Barcs, L. Bottyan, B. Molnar, D.L. Nagy, N.S.
Ovanesyan, H. Spiering, Hyperfine Interactions 55 (1990)
1187.
[13] K. Muraleedharan, C.K. Subramaniam, N. Ven-
kataramani, T.K. Gundu-Rao, C.M. Srivastava, V. San-
karanarayanan, R. Srinivasan, Solid State Commun. 76
(1990) 727.
[14] A. Junod, D. Eckert, G. Triscone, J. Muller, W. Reichard,
J. Phys.: Condens. Matter 1 (1989) 8021.
[15] I.Y. Korenblit, A. Aharony, Phys. Rev. B 49 (1994) 13291.
[16] A. Aharony, R.J. Birgeneau, A. Coniglio, M.A. Kastner,
H.E. Stanley, Phys. Rev. Lett. 60 (1988) 1330.
[17] U. Kobler, T. Chattopadhyay, Z. Phys. B- Condens. Mat-
ter 82 (1991) 383.
[23] C. Niedermayer, A. Golnik, E. Recknagel, M. Ros-
samanith, A. Weidinger, X.S. Chang, A. Kleinhammes,
N. Rosov, J. Saylor, R. Schuhmann, L. Tackacs, A. Teh,
G. Zhang, C. Hohenemser, J.I. Budnick, Phys. Rev. B 38
(1988) 2836.
[24] S. Bocquet, R.S. Pollard, J.D. Cashion, Phys. Rev. B 46
(1992) 11657.
[25] C.B. Azzoni, A. Paleari, G.B. Parravicini, J. Phys.: Con-
dens. Matter 4 (1992) 1352.
[26] M.S. Seehra, Z. Feng, R. Gopalakrishnan, J. Phys. C 21
(1988) L1051.
[18] E. Gmelin, Indian J. Pure Appl. Phys. 30 (1992) 596.
[19] T.V. Chandrasekhar Rao, V.C. Sahni, J. Phys.: Condens.
Matter 6 (1994) L423.
[27] Y.K. Jeong, G.M. Choi, J. Phys. Chem. Solids 57 (1996) 81.
[28] S. Asbrink, A. Waskowska, J. Phys.: Condens. Matter
3 (1991) 8173.