14 014
P. BURLET et al.
56
with the ‘‘flux growth’’ method by slowly cooling (2° per
hour͒ a liquid of composition Mn15Te85 in the temperature
range 873–773 K.
125
¨
The
Te Mossbauer spectroscopy measurements ͑3/2 -
1/2, 35.6 keV transition͒ were performed using a sinusoidal
drive motion of a Mg3125mTeO6 source kept either at 4.2 K
or at room temperature. The MnTe2 absorber was maintained
at different temperatures between 4.2 K and room tempera-
ture. An absorber thickness of about 50 mg of natural Te/cm
2
was used. The 35.6 keV ␥ rays were detected with an
intrinsic Ge detector well suited to discern the intense Te K␣
¨
͑27.5 keV͒ and K ͑31.0 keV͒ x rays from the Mossbauer ␥
rays. The velocity calibration of the spectrometer was per-
formed using a 57Co/Rh source and a metallic iron absorber.
The experimental data were directly analyzed by a least-
squares method to obtain the hyperfine parameters by con-
straining the relative energies and intensities of the Lorentz-
ian lines to theoretical values. In the case of combined
quadrupole and magnetic interactions, the calculation of the
spectral shape is performed by diagonalization of the full
nuclear Hamiltonian. In that situation, the fitting routine pro-
vides the value of the transferred hyperfine field (Hhf), the
isomer shift (␦IS), the quadrupole coupling constants
(eQVzz , ), the relative orientation (, ) of the field Hhf
with respect to the z axis of the electric-field-gradient tensor
͑efg͒. This axis coincides with the local symmetry axis of
each anion pair Te22Ϫ which is along one of the four-body
FIG. 2. The different models of magnetic structures.
diagonal 111 directions.
͗
͘
between these models. Such a differentiation requires a con-
straint to be applied while cooling the sample, such as an
external magnetic field.
The neutron experiments were carried out at the high flux
reactor of the Institut Laue-Langevin ͑Grenoble͒. The pow-
der experiment was performed on the 400 cells position-
sensitive detector diffractometer D1B in the temperature
range 45–150 K. The wavelength ϭ2.522 Å was used and
was provided by a focusing pyrolitic graphite monochro-
mator. The sample was enclosed in a cylindrical vanadium
container 5 mm in diameter and 5 cm in height. The analysis
of the powder patterns was performed by Rietveld profile
refinement using the software FULLPROF.5 The experiment on
the single-crystal specimen was conducted in two steps on
the two-axis spectrometer D15, operating in the normal beam
mode. The wavelength ϭ1.17 Å was provided by a flat
copper monochromator. In a first step, the crystal was
mounted in a cryomagnet with its c axis roughly 10° from
the axis of the spectrometer, this axis being the direction
of the applied field. The orientation matrix was determined
by centering a set of 15 reflections. Several peaks were then
scanned while cooling down the sample ͑zero-field cooled͒.
Then, the sample was heated up, and the same reflections
were measured under the same temperature conditions, but in
an applied field Hϭ4 T ͑field cooled͒. This experiment was
carried out to determine whether domains exist in the crystal.
Then, in a second step, the crystal was mounted in a liquid-
helium cryostat. The orientation matrix was determined from
the same reflection set as before. Bragg intensities were then
collected at three different temperatures ͑4.2, Ӎ60, and
Ӎ80 K͒, in order to refine the magnetic structure. In both
cases, integrated intensities were measured by rotating the
crystal with the detector held in a fixed position. The inte-
gration of the peaks was done during experimental runtime
using the COLL5N program.6 The refinements of the data were
carried out using the MXD package.7 The scattering lengths
In the literature, one often encounters the terminology
single-k, double-k, and triple-k for these three models. This
terminology is abusive in the case of MnTe2 from a formal
point of view, since it refers to a pseudo-face-centered-cubic
lattice occupied by the manganese atoms and to a description
of the structure in terms of a propagation vector k ϭ (100).
This description is not true in the primitive cubic lattice Pa3
since, in this space group, (100) is a reciprocal-lattice vector
and therefore does not lie in the first Brillouin zone. The
structure corresponding to the observed additional peaks has
a propagation vector kϭ0, and one actually has to take into
account a site with four Bravais sublattices.
In order to obtain more information on the magnetic
structure of MnTe2, and on the possible spin reorientation at
¨
60 K, we have performed a set of Mossbauer investigations
as well as neutron experiments on a powder sample and on a
single crystal in zero and under an applied magnetic field.
II. SAMPLE PREPARATION AND EXPERIMENTAL
METHODS
MnTe2 was prepared from commercially available high-
purity elements: Mn ͑powder, 99.9%͒, and Te ͑pieces,
99.999%͒. A pellet of stoichiometric mixture was compacted
using a steel die and then introduced into silica tubes sealed
under argon ͑300 mm Hg͒. Preliminary homogenization
treatment was conducted at 773 K. The sample was then
annealed for 2 weeks at 873 K. The single phase nature of
the final product was checked by powder x-ray-diffraction
technique ͓curved position sensitive detector INEL CPS120
͑Cu K␣)͔. Single crystals of Ӎ1 mm3 size were obtained