Chevrel-phase solid solution Mo6Se8-xTex. Study of its superconducting, magnetic and NMR properties

Article abstract of DOI:10.1016/S0921-4526(99)02286-3

The Chevrel-phase solid solution Mo₆Se₈-Mo₆Te₈ was studied by X-ray diffraction, AC and DC magnetic susceptibility and ⁷⁷Se and ¹²⁵Te NMR spectroscopy. From the smooth evolution of the lattice parameters and superconducting critical temperatures, a progressive substitution of selenium atoms by tellurium is shown, on the whole range of composition 0≤x≤8, in the formulation Mo₆Se_8-xTe_x: the unit-cell volume increases linearly because of the larger ionic size of tellurium, while T_c decreases rapidly (from 6.45 down to 0 K) because of the different formal oxidation states of the anions and a probable evolution of the Fermi level in the density of states. Results of magnetic susceptibility support this model and suggest the inhibition of the intrinsic metallic behavior with increasing x. The NMR spectra of the binaries Mo₆Se₈ and Mo₆Te₈ reveal two significant features, attributed to two different chalcogen positions in the R3 symmetry. At low Se contents in Mo₆Se_8-xTe_x (x = 7.5, 7 and 6), selenium first fills the two X(2) sites along the three-fold axis (2c positions), and then it becomes statistically distributed over the general 6f positions, leading to broad ⁷⁷Se NMR lines. On the other hand, substitution of Te atoms in Mo₆Se₈ seems to occur in a random way, creating large perturbations on the ¹²⁵Te NMR spectra, over the whole range of x. Theoretical analysis based on the presence of two anisotropic lines (of axial and non-axial symmetries, respectively) allowed us to estimate their anisotropy factors and to perfectly simulate the frequency response of both Mo₆Se₈ and Mo₆Te₈ binaries. Analysis of the Knight shift anisotropy leads us to conclude about the importance of the molybdenum z² molecular orbital contribution which controls the Mo-X dipolar interactions.

Full text of DOI:10.1016/S0921-4526(99)02286-3

Physica B 291 (2000) 339}349
Chevrel-phase solid solution Mo Se Te . Study of its super-
ꢁ
ꢂ\V V
conducting, magnetic and NMR properties
C. Hamard!ꢃꢀ, V. Au!ret!, O. Pen
ꢀ
a!ꢃ*, M. Le Floch", B. Nowak#, A. Wojakowski#
!
Laboratoire de Chimie du Solide et Inorganique Mol e& culaire, UMR 6511, Universit e& de Rennes 1, 35042 Rennes Cedex, France
"
Laboratoire de Verres et C e& ramiques, UMR 6512, Universit e& de Rennes 1, 35042 Rennes Cedex, France
#
W. Trzebiatowski Institute of Low Temperature and Structure Research, Polish Academy of Sciences, 50-950 Wroclaw 2, Poland
Received 8 November 1999
Abstract
The Chevrel-phase solid solution Mo Se }Mo Te was studied by X-ray di!raction, AC and DC magnetic suscepti-
ꢁ
ꢂ
ꢁ
ꢂ
bility and ꢄꢄSe and ꢀꢅꢆTe NMR spectroscopy. From the smooth evolution of the lattice parameters and superconducting
critical temperatures, a progressive substitution of selenium atoms by tellurium is shown, on the whole range of
composition 0)x)8, in the formulation Mo Se
Te : the unit-cell volume increases linearly because of the larger
ionic size of tellurium, while ¹ decreases rapidly (from 6.45 down to 0 K) because of the di!erent formal oxidation states
ꢁ
ꢂ\V
V
#
of the anions and a probable evolution of the Fermi level in the density of states. Results of magnetic susceptibility
support this model and suggest the inhibition of the intrinsic metallic behavior with increasing x. The NMR spectra of the
binaries Mo Se and Mo Te reveal two signi"cant features, attributed to two di!erent chalcogen positions in the R3
ꢁ
ꢁ
ꢂ
ꢁ
ꢂ
symmetry. At low Se contents in Mo Se
Te (x"7.5, 7 and 6), selenium "rst "lls the two X(2) sites along the three-fold
ꢁ
ꢂ\V
V
axis (2c positions), and then it becomes statistically distributed over the general 6f positions, leading to broad ꢄꢄSe NMR
lines. On the other hand, substitution of Te atoms in Mo Se seems to occur in a random way, creating large
ꢁ
ꢂ
perturbations on the ꢀꢅꢆTe NMR spectra, over the whole range of x. Theoretical analysis based on the presence of two
anisotropic lines (of axial and non-axial symmetries, respectively) allowed us to estimate their anisotropy factors and to
perfectly simulate the frequency response of both Mo Se and Mo Te binaries. Analysis of the Knight shift anisotropy
leads us to conclude about the importance of the molybdenum zꢅ molecular orbital contribution which controls the
Mo}X dipolar interactions. ( 2000 Elsevier Science B.V. All rights reserved.
ꢁ
ꢂ
ꢁ
ꢂ
PACS: 81.05.Bx; 61.46.#w; 74.70.Dd; 75.20.En; 76.60.Cq
Keywords: Chevrel phases; Crystal structure; Superconductivity; ꢄꢄSe NMR; ꢀꢅꢆTe NMR
1
. Introduction
The ternary molybdenum chalcogenides
ML>Mo X (X"S, Se, Te), known as Chevrel
V
ꢁ ꢂ
*
Corresponding author. Tel.: 33-2-99-28-67-57; fax: 33-2-99-
3-57-04.
E-mail address: pena@univ-rennes1.fr (O. Pen
Part of the thesis work presented at the University of
Rennes, october 1999.
phases [1], present very interesting superconduct-
ing properties, such as high transition temperatures
and extremely important critical "elds (e.g., 15 K
and &60 T, for PbMo S [2]). In addition to this,
6
ꢀ
a).
ꢀ
ꢁ
ꢂ
0
921-4526/00/$- see front matter ( 2000 Elsevier Science B.V. All rights reserved.
PII: S 0 9 2 1 - 4 5 2 6 ( 9 9 ) 0 2 2 8 6 - 3

3
40
C. Hamard et al. / Physica B 291 (2000) 339}349
when M is a rare-earth (RE) ion, the Chevrel-phase
compounds may show unusual physical properties,
in particular, the coexistence of magnetism and
superconductivity [3].
levels. This charge transfer scales down when going
from sulfur to selenium and tellurium: the number
of valence electrons which can be transferred is
lower in the tellurides than in the selenides and
sul"des, and it becomes much more di$cult to
insert divalent or trivalent cations in the Mo Te
In view of their similar superconducting behav-
ior we can consider the binary chalcogenides
Mo X themselves as very interesting compounds
ꢁ
ꢂ
sublattice.
ꢁ
ꢂ
(
[
for instance, ¹ "6.45 K, H &8 T, for Mo Se
Although many studies have been performed on
the sul"de and selenide compounds, the Chevrel
phase tellurides have not been the subject of a sys-
tematic research. In order to give a further insight
into the telluride Chevrel phases, we started a sys-
tematic work on the pseudo-binary solid solution
#
#ꢅ
ꢁ
ꢂ
2,4]). Their structure is based on the same network
of Mo X clusters as in the ternary Chevrel phases.
ꢁ
ꢂ
Contrary to the selenide and telluride binaries, the
sul"de compound Mo S is a metastable phase
ꢁ
ꢂ
which cannot be synthesized by direct combination
of the elements; however, it can be prepared from
a ternary phase (e.g., Cu Mo S ) by desintercala-
Mo Se
Te , looking as an ultimate goal for the
ꢁ
ꢂ\V
V
possibility to insert rare-earth ions in a tellurium-
based network in the Chevrel-phase structure. Our
NMR investigation is a "rst contribution, which
may bring new elements to understand these sys-
tems.
V
ꢁ ꢂ
tion methods, for instance, leaching the M ion by
using a hydrochloric acid solution or by electro-
chemical oxidation [5,6]. Solid solutions between
binaries have been well studied in the past: for
instance, Chevrel et al. showed that the solid solu-
tion Mo Se }Mo S can be prepared by direct
ꢁ
ꢂ
ꢁ ꢂ
ꢁ
synthesis from Mo Se up to Mo Se S [7], while
2. Structural description
ꢂ
ꢁ
ꢇ ꢇ
Shelton studied the evolution of the critical temper-
ature ¹ of the binary selenide Mo Se partially
Since the discussion of our NMR data will be
based mostly on structural features, we recall here
below some basic facts concerning the crystal struc-
ture of these materials. The reader is referred to Ref.
[5] for more details.
#
ꢁ
ꢂ
substituted either by S or by Te [8]. Very recent
results have been published in single crystals of the
Mo Se
nickel in Ni Mo Se
The valence-electron concentration (VEC) of the
S series obtained by desintercalation of
ꢁ
ꢂ\V V
S crystals [9,10].
W
ꢁ
ꢂ\V V
Molybdenum chalcogenides Mo X crystallize
ꢁ
ꢂ
Mo cluster plays an important role in the stability
in a hexagonal}rhombohedral symmetry (space
group R3). These phases can be described as
ꢁ
of the Chevrel phase. With only 20 electrons per
ꢁ
Mo cluster (3.33 electron per Mo cluster-atom) in
a stacking of Mo X cluster units, each unit being
a slightly distorted cube: the chalcogen atoms sit at
ꢁ
ꢁ ꢂ
the Mo S compound, the Mo}d bonding level is
ꢁ
ꢂ
located just below a forbidden energy gap. In the
case of the selenide and telluride binaries, the
covalent bondings are largely modi"ed and the
formal charges of Se and Te become smaller than
the one of sulfur. This fact a!ects the number of
the corners and the molybdenum atoms in the
middle of the faces. All Mo atoms and 6 chalcogen
atoms (hereafter called peripheral or X(1)) are in 6f
positions, while the other two chalcogen atoms
(axial or X(2)) are located at 2c positions on the
electrons localized on the Mo octahedron and
three-fold axis. The Mo X clusters are turned by
an angle of about 253 around the ternary axis,
resulting in close contacts between a Mo atom of
one unit and the chalcogen atom of a nearby unit
(Fig. 1).
ꢁ
ꢁ ꢂ
causes the VEC to increase by ꢀ and ꢅ electrons/
ꢁ
ꢆ
cluster-atom, respectively [11]. As a consequence,
the Mo-d band becomes more "lled and the struc-
ture will stabilize; therefore, these binaries may
form by direct synthesis. In the case of the ternary
compounds, the structure is stabilized by the cation
sublattice: valence electrons are transferred from
the M atom towards the Mo X cluster, increasing
Each Mo atom of a Mo octahedral cluster is
ꢁ
surrounded by "ve chalcogen atoms constituting
a square-based pyramid. The base of the pyramid is
in fact a face of a Mo X unit, while the top is
ꢁ
ꢂ
ꢁ ꢂ
the degree of "lling of the metal}metal bonding
a chalcogen atom belonging to a neighboring

C. Hamard et al. / Physica B 291 (2000) 339}349
341
Fig. 1. View of 4 Mo X units, showing axial (X(2)) and peri-
ꢁ
ꢂ
pheral (X(1)) chalcogens, and strong (Mo}Mo) and (Mo}X(1))
bondings between nearby clusters. Units are rotated by &253
with respect to the three-fold ( 3ꢁ ) axis.
Mo X unit (Fig. 1). The X(1) atoms of a unit
ꢁ
ꢂ
belong to the square faces and are also summits of
six pyramids, while the X(2) atoms belong only to
the square faces of the pyramid.
Fig. 2. Lattice parameters (a) , c)%ꢀ) and unit-cell volume (<)%ꢀ
%ꢀ
in the hexagonal symmetry for the Mo Se
as a function of the tellurium x content (open symbols from Ref.
)
Te solid solution
ꢁ
ꢂ\V
V
[7]). Lines serve as a guide to the eye.
3
. Experimental techniques and sample
deoxygenated argon atmosphere. The sintering re-
action was "rst carried out at 8003C for a day,
followed by a heat treatement at 12003C for 3 days
and "nally quenched in water. A series of samples
characterization
3
.1. Synthesis and crystallochemistry
of general formula Mo Se
Te , for x"0, 0.5, 1,
ꢁ
ꢂ\V
V
In a "rst step, molybdenum diselenide MoSe
2, 4, 6, 7, 7.5 and 8 was prepared. The resulting
products were controlled by X-ray powder analysis.
X-ray powder di!raction patterns were obtained
ꢅ
and molybdenum ditelluride MoTe were prepared
ꢅ
in a quartz tube by direct reaction between metallic
Mo (previously reduced under a hydrogen #ow at
at room temperature using Cu-K radiation, and
were indexed on the basis of a hexagonal unit cell
a
9
003C) and the corresponding chalcogen. The reac-
tion temperature was increased slowly during
days and stopped at 6503C (for MoTe ) or 8503C
similar to the well-known Mo Se [12,13] and
ꢁ
ꢂ
3
Mo Te [12] crystal structures (space group R 3ꢁ ).
ꢅ
ꢁ
ꢂ
(for MoSe ).
The lattice parameters are plotted in Fig. 2 as
ꢅ
In the next step, the binary compounds Mo Se
a function of the substitution rate x. Contrary to
ꢁ
ꢂ
and Mo Te and their solid solution Mo Se
Te were prepared by sintering the corresponding
molybdenum dichalcogenides with molybdenum
the metastable pseudo-binaries of Mo S
Se
ꢁ
ꢂ
ꢁ
ꢂ\V
ꢁ ꢂ\V
V
type [5], the substitution of selenium by tellurium
results in an almost linear increase of the unit-cell
volume: in the former case, the weakness of the S}S
intracluster bondings are responsible for an unsta-
ble Mo S lattice, whereas in our case, close Se}Te
V
metal. Before heating, the powders were thoroughly
mixed, compacted into pellets, put inside alumina
crucibles and sealed under secondary vacuum in
a quartz tube. To avoid oxidation, all operations
were performed in a glove box under dried and
ꢁ
ꢂ
contacts explain the linear increase of the lattice
with the substitution rate. In the region of low x,

3
42
C. Hamard et al. / Physica B 291 (2000) 339}349
a small decrease of the values of c
is observed,
)
%ꢀ
while the a) parameter seems to increase faster
%ꢀ
than an average linear behavior. At "rst sight, these
deviations, also observed by other authors [7], may
appear to be of no signi"cance; however, as we shall
see later (Section 4.2), it may also imply some real
di$culties to insert a large chalcogen atom (like Te)
in an otherwise stable Se cage.
ꢂ
3
.2. Superconducting properties
The solid solution Mo Se
Te was also char-
ꢁ
ꢂ\V
V
acterized from its superconducting point of view.
A home-made AC susceptometer equipped with
a mutual-inductance bridge [14,15] allowed us to
measure the inductive transitions for samples of
well-de"ned shape and volume. Full transitions
were observed for all samples with 0)x)4. Sam-
ples richer in tellurium could not be measured since
Fig. 3. Superconducting critical temperature ¹ as a function of
#
x for the solid solution Mo Se
Te (open symbols from Ref.
ꢁ
ꢂ\V
V
[8]).
their ¹ꢂ s [8] were beyond our temperature capabil-
#
ities. Fig. 3 shows our values compared to those
e!ect due to the larger size of tellurium compared
to selenium. In this case, the position of the Fermi
level may change and produce similar e!ects as in
the doped binaries [4]. At the same time, the Mo}d
bands become more "lled because of the di!erent
charge transfers between Se and Te, leading to
obtained by Shelton [8]; here again, a small in-
crease of ¹ at low x (within our error bars) may be
#
of no signi"cance. However, we should recall our
recent results on slightly doped Mo Se binaries,
ꢁ
ꢂ
where a de"nite increase of ¹ , correlated with the
#
ionic radius of the dopant, was systematically ob-
a decrease of ¹ . These two e!ects may tend to
#
served [4,16].
compensate each other at low content of tellurium,
It is well known that superconductivity in the
Chevrel phases is mostly due to the existence of
a high Mo}d band density of states at the Fermi
level, and that their relative positions may change
with the oxidation state of both cations and anions.
Calculations by Cai et al. [17] showed indeed, that
the Fermi level in Mo Se is situated at or very
resulting in quite constant ¹ values in the range
#
0)x)1.
At this point, it is good to recall results obtained
in mixed clusters of the type Mo Se
Y [18],
ꢁ
ꢂ\V V
where the chalcogen (Se) was partially replaced by
a Y halogen anion (Br or I). In that case, charge
transfer e!ects (smaller valence of the anion) and
halogen occupation on special crystallographic
sites (axial Se(2) site) were invoked to explain
an increase of ¹ (up to 7.1 K for Mo Se Br and
ꢁ
ꢂ
close to a peak of the density of states, while in the
sul"de and telluride cases, it is situated near a min-
imum. In addition, in the case of Mo Te , the large
ꢁ
ꢂ
#
ꢁ
ꢄ
increase of the VEC number provokes almost full
occupation of the bonding states and, therefore,
a semi-metallic behavior may be expected. As a re-
sult of these arguments, among the three binary
chalcogenides, Mo Se is at the optimum position
7.6 K for Mo Se I). In our case, tellurium occupies
ꢁ
ꢄ
random positions among the 8 corners of the
pseudo-cubic cage (see Section 4.2) and it will pro-
duce a slight decrease of ¹ for the same substitu-
#
tion rate (¹ "6.2 K for Mo Se Te).
ꢁ
ꢂ
#
ꢁ
ꢄ
from the point of view of their superconducting
properties (¹ "1.8, 6.45 and &0 K, for Mo S
3.3. Magnetic susceptibility
#
ꢁ ꢂ
[
5], Mo Se [4] and Mo Te [8], respectively).
ꢁ
ꢂ
ꢁ
ꢂ
Two e!ects may explain the ¹ behavior at
DC magnetic susceptibility was measured in a
SHE-VTS SQUID susceptometer, between 5 and
#
xP0, as observed in Fig. 3: from one side, a steric

C. Hamard et al. / Physica B 291 (2000) 339}349
343
imum above room temperature. The last sample,
Mo Se Te , with equal amounts of selenium and
ꢁ
ꢇ
ꢇ
tellurium anions, shows an intermediate tem-
perature variation between these two behaviors,
with an almost #at susceptibility in the range
[
100}300 K]. From these results it is quite tempt-
ing to suggest a smooth shift of ¹ꢁ (that is, the
!ꢀ
temperature at which the magnetic susceptibility
goes through a maximum) as a function of the
tellurium concentration but we have no clear ex-
planation at the moment for this particular e!ect.
The Pauli paramagnetism, as de"ned in our case
by the room-temperature value s(300 K), rapidly
decreases with the increase of tellurium. This tend-
ency suggests that the intrinsic metallic behavior
observed for the Mo Se binary is inhibited by the
ꢁ
ꢂ
Fig. 4. Magnetic susceptibility, measured at 5 kOe, for 5 sam-
presence of tellurium, in agreement with our pre-
vious supposition of a semi-metallic behavior in the
non-superconducting Mo Te . Preliminary NMR
ples of the solid solution Mo Se
Te .
ꢁ
ꢂ\V
V
ꢁ
ꢂ
3
00 K under an applied "eld of 5 kw. At each mea-
data obtained in the ꢈꢆMo nucleus con"rm these
sured temperature, the sample holder's contribu-
tion (of the order of 65}80% of the total signal)
was subtracted, and the results were plotted as
a function of temperature for 5 characteristic sam-
ples (Fig. 4). In most of the curves, a small rise of the
paramagnetic signal was observed below 50 K,
which was neglected in our subsequent discussion.
Two types of behaviors are apparent in Fig. 4,
one for selenium-rich samples (Mo Se and
assumptions [20,21].
3.4. NMR experimental conditions
The as-sintered samples of the Mo Se
Te
ꢁ
ꢂ\V
V
(0)x)8) solid solution were thoroughly crushed
to allow RF "eld penetration; about 3 grams of
powder was then put into a pyrex tube and placed
at the center of the RF coil. Room-temperature
ꢄꢄSe and ꢀꢅꢆTe NMR spectra were obtained on
a BRUKER ASX 300 MHz spectrometer using the
well-known Hahn spin-echo technique, that is
a p/2}q}p sequence (p/2 pulse of about 4 ls) which
produces an echo signal in the receiving coil. The
external applied magnetic "eld was 7 T and the
expected resonance frequencies for the ꢄꢄSe and
ꢀꢅꢆTe nuclei (I"1/2) were 57.2 MHz and
97.8 MHz, respectively. The accumulated echo sig-
nal was Fourier-transformed to give the absorption
spectra, taking as time origin the top of the echo.
At this point we should stress that, due to the
very broad lines observed in this work, several
spectra had to be recorded for each compound at
slightly di!erent resonance frequencies (shifts of
about 10 kHz each time): the resulting convoluted
spectrum represents then the full response of the
corresponding nucleus [22,23]. More details will be
given in Section 4.
ꢁ
ꢂ
Mo Se Te ), the other for tellurium-rich samples
ꢁ
ꢁ
ꢁ
ꢅ
ꢅ
ꢁ
(
Mo Se Te and Mo Te ). In the "rst group, the
ꢁ
ꢂ
magnetic susceptibility is rather temperature-inde-
pendent, with a tendency to increase linearly with
decreasing ¹, in the range 100 K)¹)300 K.
This behavior can be interpreted in the same way as
in some high-¹ high-H ternary Chevrel phases,
#
#ꢅ
in which the Fermi level is situated near a peak in
the density of states [2] (for instance, in PbMo S
or LaMo Se ). Mo Se could be assigned to such
ꢁ
ꢂ
ꢁ
ꢂ
ꢁ
ꢂ
a group because of the relative position of the Fermi
level [17] and also because of its ¹ and H (0)
#
#ꢅ
(
6.45 K and 8 T, respectively), not far from those
observed in the above-mentioned compounds (e.g.,
1 K and 55 T, respectively, in LaMo Se [19]).
1
ꢁ
ꢂ
The tellurium-rich samples (Mo Te
and
ꢁ
ꢂ
Mo Se Te ) showed a di!erent thermal variation
of the susceptibility, increasing with increasing
temperature and suggesting the presence of a max-
ꢁ
ꢅ
ꢁ

3
44
C. Hamard et al. / Physica B 291 (2000) 339}349
Standard materials used as chemical shift refer-
ences for the ꢄꢄSe and ꢀꢅꢆTe nuclei, were liquid
solutions of H SeO and K TeO , respectively.
ꢅ
ꢇ
ꢅ
ꢉ
Independent NMR analyses were made on pure
selenium and tellurium elements, together with the
two dichalcogenides (MoSe and MoTe ) used as
ꢅ
ꢅ
starting materials for synthesis. In this way, any
spurious peak which could appear in the NMR
spectra could be identi"ed as extrinsic or intrinsic
features.
4
. NMR results and discussion
4
.1. ꢄꢄSe NMR in Mo Se and ꢀꢅꢆ¹e NMR in Mo Te
6
8
6
8
Figs. 5 and 6 show the ꢄꢄSe and ꢀꢅꢆTe NMR
spectra obtained at room temperature for the two
binaries Mo Se and Mo Te . Because of the ex-
ꢁ
ꢂ
ꢁ
ꢂ
Fig. 6. Room-temperature ꢀꢅꢆTe NMR spectra of Mo Te
taken at slightly di!erent resonant frequencies. Actual spectrum
corresponds to the envelope.
tremely broad lines, as stated in Section 3.4, each
gure corresponds to the superposition of several
ꢁ
ꢂ
"
spectra taken at di!erent applied frequencies. In
this way, the resulting convolution will better de-
scribe the expected characteristics, in particular the
pronounced discontinuities in the low- and high-
frequency regions, of each of the studied com-
pounds.
The overall widths of the NMR lines of the pure
binary phases are of comparable values (several
tens of kHz for both Mo Se and Mo Te ). Sev-
ꢁ
ꢂ
ꢁ
ꢂ
eral mechanisms may be responsible for such an
important line width: "rst of all, the usual X}X and
Mo}X dipolar interactions, which depend on the
internuclear distances. Since the Mo}X distances
are shorter than the X}X distances, we may expect
that the Mo}X dipolar interaction would be domi-
nant. However, the dipolar mechanism being also
proportional to the gyromagnetic ratio (larger for
Se and Te than for Mo) and to the natural abund-
ance of the respective nuclei, the relative di!erences
compensate and both Mo}X and X}X dipolar
mechanisms retain the same order of magnitude. In
any way, whatever the interaction may be, the dipo-
lar mechanism is too weak to explain such broad
lines. Another broadening mechanism, related to
the distribution of powder grain shapes can also be
considered. Here again, the low magnetic suscepti-
bility of these compounds (Section 3.3) cannot
justify any demagnetization broadening of the
experimental lines. In fact, the main broadening
mechanism expected in these compounds comes
from the anisotropy of the paramagnetic (Knight)
Fig. 5. Room-temperature ꢄꢄSe NMR spectra of Mo Se taken
ꢁ
ꢂ
at slightly di!erent resonant frequencies. Actual spectrum cor-
responds to the envelope.

C. Hamard et al. / Physica B 291 (2000) 339}349
345
Table 1
Parameters of simulated spectra of Mo Se and Mo Te !
ꢁ
ꢂ
ꢁ
ꢂ
Line d*ꢂꢃ (ppm) g
D (ppm)
S (%)
ꢄꢄSe in Mo Se
A
B
A
B
2690
3530
3040
4119
0.06 !510
0.85 770
0.06 #370
0.9 !1150
28.25
71.75
24.60
75.40
ꢁ
ꢂ
ꢀ
ꢅꢆTe in Mo Te
ꢁ
ꢂ
!
Note: (d*ꢂꢃ) isotropic position; (D) anisotropy; (g) asymmetry;
(
S) surface under each peak.
ical parameters obtained from these "ts, where
(in units of ppm) describes the line's isotropic
d
*
ꢂꢃ
position, D and g are the anisotropy and asymmet-
ric parameters, respectively, and S (in percentage)
corresponds to the area under each peak. Both lines
get closer at lower temperature as d decreases
more rapidly than d , which suggests an overall
ꢊ
lattice contraction and a deformation of the
Mo X unit with a slight elongation along the
ꢁ
ꢂ
three-fold axis [24].
A few other remarks should be made. Firstly, the
Fig. 7. Simulated spectra of Mo Se and Mo Te , showing the
positions of lines A and B and their superposition.
ꢁ
ꢂ
ꢁ
ꢂ
Mo Te spectrum is broader than the Mo Se one
ꢁ
ꢂ
ꢁ
ꢂ
since [d }d ]+ꢃ
ꢁ
2%
ꢂ
'[d }d ]+ꢃ
ꢁ
1%
, as it also oc-
ꢂ
ꢊ
ꢊ
curs for the corresponding anisotropy factors. Sec-
ondly, line A of axial symmetry is qualitatively
di!erent in both compounds: while, in the selenide
case, the anisotropy points toward decreasing
values of d (i.e., in the direction of negative fre-
shift resulting, in a polycrystalline sample, from the
molecular orientation with respect to the applied
"
eld. Since the electronic p-orbitals of the chalco-
gens are very sensitive to the local symmetry, as we
shall see later, important shifts are then observed.
The experimental spectra obtained for the
Mo Se and Mo Te binaries were "tted by stan-
*
ꢂꢃ
quencies), in the telluride compounds it points in
the opposite direction. This type of inversion, al-
ready encountered in other metals, has been ex-
plained by electronic arguments due to di!erent
"llings of the conduction band and the Fermi level
position [25,26]. Such arguments may also apply to
our case since, as we already discussed in Section
3.2, the e and a bands at the Fermi level may be
ꢁ
ꢂ
ꢁ
ꢂ
dard techniques (WINFIT program, from
BRUKER), in which two lines of di!erent symmet-
ries were assumed: line A, of axial symmetry, was
assigned to the X(2) atoms situated on the ternary
axis, while an anisotropic line B was ascribed to the
six peripheral X(1) atoms. Integration of both lines
was in the ratio 25 : 75, that is, on the expected
relative occupancies for both crystallographic sites.
Fig. 7 gives details about each of the simulated lines
and their corresponding convolution. A perfect
identi"cation of the simulated spectra with the ex-
perimental data con"rms the validity of our as-
sumptions and, in particular, the formation of
a `pseudo peaka at about 3100 ppm in the case of
Mo Te . Table 1 gives details about the mathemat-
'
ꢄ
basically di!erent when going from the Mo Se
ꢁ
ꢂ
case to the Mo Te compound. Finally, we should
ꢁ
ꢂ
recall that the paramagnetic shift anisotropy is re-
lated to the hyper"ne "eld due to the dipolar coup-
ling between the nuclear and conduction electrons
spins. In the case under consideration, of the two
molybdenum molecular orbitals present at the
Fermi level, (xꢅ}yꢅ) and zꢅ, the latter is the most
important one since it points toward the chalcogen
atoms of the nearby unit. In this way, the Mo}X
ꢁ
ꢂ

3
46
C. Hamard et al. / Physica B 291 (2000) 339}349
bonding and its strongest covalent character in the
case of Mo Te , may be the principal reason for its
4.2.1. ꢄꢄSe NMR
Fig. 8h and g show such spectra when only 0.5
and 1 atom are, on the average, substituted in the
Mo Te structure: only the narrow line A of axial
ꢁ
ꢂ
largest anisotropy, compared to the Mo Se case.
ꢁ
ꢂ
ꢁ
ꢂ
4
.2. Solid solution
symmetry is observed, centered at about the same
position as the line attributed to the Se(2) atoms in
the structure. From this result we can conclude that
selenium atoms are "rst inserted at a preferential
site in the 2c positions (i.e., along the three-fold
axis), this occurring up to the composition
Mo Se Te (Fig. 8f ). It is interesting to note that,
Figs. 8 and 9 show the ꢄꢄSe and ꢀꢅꢆTe NMR
spectra for the 9 samples of the solid solution
Mo Se
Te , x"0, 0.5, 1, 2, 4, 6, 7, 7.5 and 8. For
ꢁ
ꢂ\V
V
the sake of clarity, we have reproduced only the
most relevant spectrum for each sample, and not
the ensemble of spectra, as shown before in the case
of the pure binaries.
ꢁ
ꢅ
ꢁ
at the lowest selenium contents, the anisotropy is of
the Mo Te -type, that is, it points toward positive
ꢁ
ꢂ
The simplest way to describe the e!ects of partial
frequencies. Indeed, at these low concentrations of
selenium, the conduction band and the position of
the Fermi level are essentially similar to the
Mo Te case and, therefore, the sign of the Knight
chalcogen substitution in the Mo Se
Te solid
ꢁ
ꢂ\V
V
solution is to look at the NMR spectra for very low
concentration of substituent atoms. Of course, due
to the very small amounts of the resonating nuclei
ꢁ
ꢂ
shift anisotropy is imposed by the tellurium envi-
ronment and the corresponding VEC. At larger
selenium content, the anisotropy must change its
sign, to become negative, as explained before for the
case of pure Mo Se .
(
ꢄꢄSe and ꢀꢅꢆTe), the signal-to-noise ratio at both
limits (xP0 and xP8) may look poor but good
enough to perfectly describe the substitution mech-
anism between chalcogens.
ꢁ
ꢂ
Fig. 8. ꢄꢄSe NMR spectra for the solid solution Mo Se
x"7.5.
Te . (a) x"0, (b) x"0.5, (c) x"1, (d) x"2, (e) x"4, (f) x"6, (g) x"7, (h)
ꢁ
ꢂ\V
V

C. Hamard et al. / Physica B 291 (2000) 339}349
347
Fig. 9. ꢀꢅꢆTe NMR spectra for the solid solution Mo Se
Te . (a) x"8, (b) x"7.5, (c) x"7, (d) x"6, (e) x"4, (f) x"2, (g) x"1,
ꢁ
ꢂ\V
V
(h) x"0.5.
When the two axial sites X(2) of the Mo Te
spectrum is already very broad, as if a random
ꢁ
ꢂ
structure have been fully occupied by selenium
atoms, other atoms start to occupy the X(1) sites at
the general positions 6f of the structure. Since these
distribution of Se and Te takes place. Broad reson-
ance lines with undistinguishable patterns continue
to occur up to the composition Mo Se Te (Fig.
ꢁ
ꢇ
ꢇ
6
new sites are strictly equivalent, the progressive
9e). At further Te increment, lines start to split,
giving place to the well-de"ned discontinuities of
the Mo Te binary; during this process, the overall
occupation is done in a random way, leading to very
broad lines (Mo Se Te and Mo Se Te , Fig. 8e
ꢁ
ꢇ
ꢇ
ꢁ
ꢁ
ꢅ
ꢁ
ꢂ
and 8d, respectively), although their centers of grav-
ity remain unchanged. The statistical character of
the occupation of the 6 peripheral sites may also
explain the approximate gaussian shape of the ob-
served spectra. Finally, at composition Mo Se Te
line width stays quite constant.
Additionally, the presence of a noticeable peak in
the tellurium resonance, centered at&8000 ppm, is
seen in Fig. 9h, which progressively disappears
when the amount of Te increases. Di!erent trials to
identify such a peak by using the starting materials
ꢁ
ꢄ
ꢀ
(
Fig. 8c) lines start to separate out, until the well-
resolved ꢄꢄSe spectrum of Mo Se is attained.
(Te metal or MoTe , for instance) were unsuccess-
ful. On the other hand, if this peak was due to some
ꢁ
ꢂ
ꢅ
4
.2.2. ꢀꢅꢆ¹e NMR
The parallel situation, that is, the progressive
increment of Te in the binary matrix Mo Se , is
shown in Fig. 9. However, as seen in Fig. 9h for the
lowest substitution rate (x"0.5), the ꢀꢅꢆTe NMR
extrinsic telluride phase, then it becomes di$cult to
understand why it disappears at higher concentra-
tion of tellurium. Therefore, our current hypothesis
favors an intrinsic mechanism related to the inser-
tion of a very big atom (i.e., Te) in the stable binary
ꢁ
ꢂ

3
48
C. Hamard et al. / Physica B 291 (2000) 339}349
cluster formed by Mo and Se: the presence of 1 Te
atom per Mo X unit will considerably deform its
almost perfect cubic shape, pushing the Te atom
slightly outside the 2c or 6f crystallographic posi-
tion. With further tellurium insertion, atoms will
the paramagnetic shift anisotropy as being mainly
due to the Mo zꢅ molecular orbitals. In addition,
the relative position of the Fermi level with respect
to the Mo-d band density of states, which explains
important changes in the physical properties of this
ꢁ
ꢂ
"
nd more easily their expected atomic positions
series as a function of x (e.g., ¹ , magnetic suscepti-
#
since the X cage would have already adopted
a deformed geometry. In this way, the `interstitiala
bility, electrical transport, etc.) may also take into
ꢂ
account the di!erent signs of the anisotropy be-
position occupied by the Te atom, characterized by
the NMR peak at &8000 ppm, will progressively
disappear at higher Te concentration. This mecha-
nism will also explain the deviations to linearity
observed in the evolution of the lattice parameters
discussed in Fig. 2 and Section 3.1. Single-crystal
structural re"nement is in progress to investigate
this point.
tween Mo Se and Mo Te .
ꢁ
ꢂ
ꢁ
ꢂ
Further work is in progress, including the ꢈꢆMo
NMR, the thermal evolution of the anisotropic
shifts or the importance of the cation lattice (i.e.,
M"RE). Evidently, a parallel characterization of
these materials, as done in this work, is essential to
verify their quality, but also to understand other
important NMR phenomena, such as [Knight
shift-magnetic susceptibilty] correlations or spin-
lattice relaxation mechanisms.
5
. Conclusion
NMR spectroscopy is particularly interesting to
References
study the microscopic aspects of the interplay be-
tween magnetism and superconductivity in the
Chevrel-phase materials. The local modi"cations
produced by the insertion of a third element (i.e.,
M Mo X ), or by the partial substitution at the
chalcogen site can be monitored by the use of local
probes. Up to now, NMR techniques have not been
much used to study the Chevrel-phase compounds,
despite the fact that many of the constituent atoms
are appropriate nuclei for magnetic resonance (e.g.,
[1] R. Chevrel, M. Sergent, J. Prigent, J. Solid State Chem.
3
(1971) 515.
[
[
2] " Fischer, Appl. Phys. 16 (1978) 1.
3] ". Fischer, M.B. Maple (Eds.), Topics of Current Physics,
Vols. 32 and 34, Springer, Berlin, 1982.
V
ꢁ ꢂ
[
[
4] F. Le Berre, D. Tshimanga, A.-L. Guilloux, J. Leclercq,
M. Sergent, O. Penꢀ a, R. Horyn, A. Wojakowski,
Physica B 228 (1996) 261.
5] R. Chevrel, M. Sergent, in: ". Fischer, M.B. Maple (Eds.),
Topics of Current Physics, Vol. 32, Springer, Berlin, 1982,
pp. 25}86.
ꢄꢄSe, ꢀꢅꢆTe, ꢈꢆMo, ).
2
[6] R. Schollhorn, M. Ku
Bull. 12 (1977) 781.
7] R. Chevrel, M. Sergent, ". Fischer, Mater. Res. Bull. 10
1975) 1169.
K mpers, J.O. Besenhard, Mater Res.
In this work, we have chosen the simplest
example in this family, that is, the binaries Mo Se
[
ꢁ
ꢂ
(
and Mo Te and their solid solution, paying
ꢁ
ꢂ
[8] R.N. Shelton, in: D.H. Douglass (Ed.), Superconductivity
in d- and f-Band Metals, Plenum Press, New York, 1976,
pp. 137}158.
special attention to the way in which way the par-
tial substitution SePTe takes place. An appropi-
ate modelization allowed to identify the two
di!erent crystallographic sites, the peripheral X(1)
and the axial X(2). When Se replaces Te in
Mo Te , the substituent "rst "lls the high-sym-
[9] S. Belin, Thesis, Universite
10] S. Belin, L. Burel, R. Chevrel, M. Sergent, Mater. Res. Bull.
5 (2000), to appear.
H de Rennes I (France), 1997.
[
[
3
11] K. Yvon, in: E. Kaldis (Ed.), Current Topics in Materials
Science, Vol. 3, North-Holland, Amsterdam, 1979, pp.
ꢁ
ꢂ
metry sites X(2) and then it becomes statistically
distributed on the 6f positions. On the other hand,
when Te replaces Se in Mo Se , it occupies ran-
5
3}129.
[12] O. Bars, D. Grandjean, A. Meerschaut, M. Spiesser, Bull.
Soc. Fr. Mineral. 93 (1970) 498.
13] O. Bars, J. Guillevic, D. Grandjean, J. Solid State Chem.
ꢁ
ꢂ
[
domly all the 8 sites of the Mo X structure, cre-
ꢁ
ꢂ
6 (1973) 48.
ating large perturbations of the NMR spectra over
the whole range of x. The evaluation of all the
parameters used in these "ts allowed us to interpret
[
[
14] O. Pen
15] R. Horyn, F. Le Berre, A. Wojakowski, O. Pen
cond. Sci. Technol. 9 (1996) 1081.
ꢀ a, Meas. Sci. Technol. 2 (1991) 470.
ꢀ
a, Super-

C. Hamard et al. / Physica B 291 (2000) 339}349
349
[
[
[
[
[
[
16] F. Le Berre, O. Pen
A. Wojakowski, J. Alloys Compounds 262}263 (1997) 331.
17] S.H. Cai, C.W. Liu, J. Chem. Soc. Faraday Trans. 91 (1995)
ꢀ
a, C. Hamard, A. Corrignan, R. Horyn,
[22] T.J. Bastow, Solid State Nucl. Magn. Reson. 3 (1994)
17.
[23] D. Massiot, I. Farnan, N. Gautier, D. Trumeau, A.
Trokiner, J.P. Coutures, Solid State Nucl. Magn. Reson.
4 (1995) 241.
4
79.
18] M. Sergent, ". Fischer, M. Decroux, C. Perrin, R. Chevrel,
J. Solid State Chem. 22 (1977) 87.
[24] C. Hamard, B. Nowak, M. Le Floch, O. Pen
ꢀ a,
19] O. Pe nꢀ a, F. Le Berre, J. Padiou, T. Marchand, R. Horyn,
A. Wojakowski, to be published.
[25] H. Ebert, J. Abart, J. Voitlaender, J. Phys. F: Met. Phys. 16
(1986) 1287.
[26] N. Winzek, F. Hentsch, M. Mehring, H. Mattausch,
R. Kremer, A. Simon, Physica C 168 (1990) 327.
A. Wojakowski, J. Solid State Chem. 135 (1998) 160.
20] C. Hamard, Thesis, Universite
H
de Rennes I (France), 1999,
unpublished
21] C. Hamard, O. Pen
ꢀ
a, M. Le Floch, Solid State Commun.,
to be published.