The crystal structure of FeSe0.44Te0.56

The crystal structure of FeSe0.44Te0.56
M. Tegel, C. Lo¨hnert, D. Johrendt ∗
Department Chemie, Ludwig-Maximilians-Universita¨t Mu¨nchen,
Butenandtstr. 5-13 (Haus D), 81377 Mu¨nchen, Germany
Abstract
The crystal structure of the superconductor FeSe0.44Te0.56 was redetermined by
high-resolution X-ray single crystal diffraction at 173 K (anti-PbO-type, P4/nmm,
a = 3.7996(2), c = 5.9895(6) A˚, R1 = 0.022, wR2 = 0.041, 173 F 2). Significantly
different z-coordinates of tellurium and selenium at the 2c site are clearly discernible
and were refined to zTe=0.2868(3) and zSe=0.2468(7). Thus the chalcogen heights
differ by 0.24 A˚ and the Fe–Se bonds are by 0.154 A˚ shorter than the Fe–Te bonds,
while three independent (Te,Se)–Fe–(Te,Se) bond angles occur. An elevated U33
displacement parameter of the iron atom is suggestive of a slightly puckered Fe layer
resulting from different combinations of Se or Te neighbors. Such strong disorder
underlines the robustness of superconductivity against structural randomness and
has not yet been considered in theoretical studies of this system.
Key words: A. Superconductors, C. Crystal structure
PACS: 74.70.Xa, 74.62.Bf
1 Introduction
β-FeSe with the tetragonal anti-PbO-type structure [1,2] can be considered as
the archetypal iron-based superconductor. Consisting solely of superconduct-
ing layers without separating atoms or building blocks, this seems to be the
ideal system to study the underlying physics, which turned out to have many
analogies to the iron arsenide superconductors [3,4,5]. However, soon after the
discovery of superconductivity in FeSe [6] with Tc = 8 K, a set of studies
revealed differences between iron selenide and pnictide materials. Examples
are the extreme sensitivity of superconductivity to the stoichiometry [7,8], the
∗ Corresponding author
Email address: johrendt@lmu.de (D. Johrendt).
Preprint submitted to Elsevier 13 January 2010
ar
X
iv
:0
91
2.
37
06
v2
[
co
nd
-m
at.
su
pr
-co
n]
1
3 J
an
20
10

huge increase of Tc under pressure [9,10], the absence of long-range magnetic
ordering and the orthorhombic or even monoclinic symmetry of the supercon-
ducting phase [11]. In contrast to the selenide, the binary iron telluride with
anti-PbO-type structure [12] is not superconducting even under pressure [13].
This may be due to the fact that stoichiometric FeTe does not exist, but only
Fe1+δTe with δ ≈ 0.05-0.15. The excess iron atoms between the layers may be
detrimental to superconductivity [14]. On the other hand, superconductivity
survives in the solid solution FeSe1−xTex with a maximum Tc of about 14 K
close to FeSe0.5Te0.5 [15].
The FeSe1−xTex system has been intensively studied with respect to the in-
terplay between structural or magnetic degrees of freedom and superconduc-
tivity. It is accepted that magnetic fluctuations play an important role, and
calculations revealed a strong sensitivity of the magnetic moment on the so-
called “chalcogen height”, i. e. the distance of the Se/Te atoms from the
plane of iron atoms [16]. According to this, the magnetic exchange parameter
J changes by a factor of 2-5 when the height varies by only 0.1 A˚. Such subtle
dependencies of the magnetic and superconducting properties are also known
from the iron pnictides [19], which re-emphasizes the importance of accurate
and correct structural data. However, against the background of several pub-
lished crystal structures of FeSe1−xTex[20], it is surprising that even simple
crystal chemical aspects have been completely disregarded. Selenium and tel-
lurium have quite different ionic radii: rSe2− = 1.98 A˚ and rTe2− = 2.21 A˚
[21]. Therefore, one cannot expect that both ions occupy the 2c position with
the same z-coordinate in the layered PbO-type structure and we wondered
why the structure is described with Se and Te in the same position in many
publications. Only in a recent preprint by Lehman et al. [22] were different
z-coordinates of FeSe0.5Te0.5 determined laboriously and with low precision
from an analysis of the pair density function (PDF) obtained from neutron
powder diffraction data.
In this paper we present a single-crystal X-ray diffraction study of FeSe0.44Te0.56,
which easily reveals the distinct z-coordinates of Se and Te with a one order of
magnitude better accuracy compared to the PDF method. We identify a quite
large degree of structural disorder, which has been neglected in recent studies
and emphasize the robustness of superconductivity against randomness.
2 Experimental details
The sample with the nominal composition FeSe0.4Te0.6 was synthesized by
heating the elements (purity > 99.8%) at 973 K for 40 h in an alumina cru-
cible, sealed in a silica tube under an atmosphere of purified argon. After
cautious homogenization using an agate mortar in an argon-filled glove box,
2

the sample was heated again at 1012 K for 30 h. The X-ray powder diffrac-
togram (STOE Stadi-P, Mo-Kα1 radiation) showed the expected pattern of
FeSe1−xTex. A Rietveld refinement using the TOPAS package [23] (see Fig-
ure 1) revealed only traces of impurity phases. The refined lattice parameters
are a = 3.8061(2), c = 6.0871(3) A˚. Superconductivity was confirmed by mea-
suring the AC susceptibility (inset in Figure 1). A relatively broad transition
with an onset of Tc at 14 K was found. Such broad transitions in FeSe1−xTex su-
perconductors have also been observed by other groups [24,20], and may be
due to a certain sample inhomogeneity. A small plate-like single crystal of
≈ 15 × 20 × 30 microns was selected from the polycrystalline sample and
checked by Laue photographs using white radiation from a Mo anode. Diffrac-
tion intensity data up to 2θ = 90◦ (0.5 A˚ resolution) were collected at 173 K
on an Oxford Xcalibur 4-circle diffractometer equipped with a CCD detector.
Graphite-monochromized Mo-Kα radiation from a conventional sealed tube
was used. The measured intensities were carefully corrected for absorption ef-
fects. The atom positions from [20] were used as starting parameters (zSe/Te
= 0.27388) and refined by the least-squares method using the JANA2006 pro-
gram package [25]. The positional and thermal parameters of Se and Te were
refined independently, while their occupation parameters were constrained to
unity.
Fig. 1. X-ray powder pattern (blue) and Rietveld fit (red) of FeSe0.4Te0.6. Inset: AC
susceptibilty measurement showing the onset of superconductivity at ≈ 14 K
3 Structure refinement
The initial refinement of the data using one z-coordinate already resulted
in small residuals of R1 = 0.029 and wR2 = 0.051. But the inspection of
the anisotropic displacement parameters showed a three times larger displace-
ment parameter U33 when compared with U11. It is known that such elonga-
tions can be artifacts from insufficient absorption corrections. For this reason
3

we performed several refinements using different absorption models. This re-
vealed only a small dependency, which cannot explain a factor of three in the
anisotropy. In the next step we enabled the independent refinement of the Se
and Te z-coordinates, which immediately converged to different values for Se
(≈ 0.246) and Te (≈ 0.286), while the residuals dropped to R1 = 0.023 and
wR2 = 0.05. Moreover, the anisotropy of the still combined (Se,Te) thermal el-
lipsoid was reduced from ≈ 3 to ≈ 1.3. The necessary independent refinement
of the anisotropic displacement is difficult, because the z-coordinates and the
U33 parameters are strongly correlated. However, due to the high-resolution
data (2θmax ≈ 90◦), the correlations remained acceptable and did not desta-
bilize the refinement. Final cycles converged to residuals R1 = 0.022, wR2 =
0.041 and GooF = 0.98. A careful check of the ∆F Fourier map revealed no
residual density from additional Fe atoms between the layers. Regarding the
Fe atom in the layer, all refinements showed an elongation of the thermal el-
lipsoid along z with an anisotropy of ≈ 2. We suggest that this is the response
of the iron atom to the different (Se,Te) coordinations, tantamount to a small
puckering of the Fe plane. A summary of the crystallographic data is compiled
in Table 1, the final atom position parameters and anisotropic displacements
are given in Table 2 and relevant bond distances and angles are collected in
Table 3.
Further details of the crystal structure investigation in CIF format may be ob-
tained from Fachinformationszentrum Karlsruhe, 76344 Eggenstein-Leopoldshafen,
Germany (fax: (+49)7247-808-666; e-mail: crysdata@fiz-karlsruhe.de, http://www.fiz-
karlsruhe.de/request for deposited data.html) on quoting the CSD number
421334.
Fig. 2. Crystal structure of FeSe0.44Te0.56 with Se/Te split positions. (space group
P4/nmm, thermal ellipsoids of 95% probability)
4

4 Discussion
The crystal structure of FeSe0.44Te0.56 is depicted in Figure 2. The redetermi-
nation clearly revealed the distinct positions of Se and Te as expected from
crystal chemical reasons and agrees with the assumption made in Ref. [22].
The resulting “chalcogen heights” are hSe = 1.478 A˚ and hTe = 1.718 A˚. In
terms of bond lengths, the Fe–Se bond length is by 0.154 A˚ shorter than
the Fe–Te bond. In comparison to the binary compounds, the Fe–Se bond
is slightly longer (by 1.6 %) than in FeSe [1] and the Fe–Te bond is slightly
shorter (by 1.8 %) than in Fe1+δTe [12]. This is plausible from crystal chem-
istry, in contrast to a mean Fe–(Se,Te) bond length if only one z-parameter
is used. Also the (Se,Te)–Fe–(Se,Te) bond angles depend significantly on the
z-coordinates. Iron occupies a position with 4m2 symmetry; therefore the an-
gle which is bisected by the c-axis (α) and the angle which is bisected by the
ab-plane (β) are related by cos(β) = −12 [1− cos(α)]. In the present case, three
independent angles α have to be considered, namely Te–Fe–Te, Se–Fe–Se and
Te–Fe–Se. We have also listed the dependent angles β in Table 3 for the sake
of completeness.
In the case of the iron pnictide superconductors, it has been argued that the
geometry of the FePn4 tetrahedra plays an important role. Tc is seemingly
maximized when the bond angles are close to the ideal tetrahedral angle of
109.47◦ [26]. The latter has not been observed in the FeCh systems, but the
large enhancement of Tc under pressure suggests again that details of the
structure are involved in the mechanism of superconductivity. Furthermore,
it is undoubted that the coordination of the iron atoms controls the Fermi
surface topology, which is very similar to the FePn superconductors [27]. Also
the magnetic moment depends strongly on the height of the pnictogen [28] and
chalcogen [16] atoms; thus the difference ∆hTe−Se = 0.24 A˚ as a consequence of
the split positions is a strong effect. With respect to the studies on structural
effects in the FeSe1−xTex system reported so far [20,16,29], we want to point
out that the changes in the geometry by the split position of Se and Te are
at least one magnitude larger than the effects induced by temperature or
pressure.
A further implication of our results concerns the discussion about the ro-
bustness of superconductivity in iron-based materials in the framework of the
s±-scenario [17]. The latter is expected to be rather fragile against impurities
or other kinds of randomness [18]. In this context, it is remarkable that Tc
increases from 8 K in FeSe to 14 K in FeSe1−xTex (x ≈ 0.5) notwithstanding
the strong geometrical disorder caused by the Te-doping at the very differ-
ent z-coordinates. Detailed calculations taking this into account are necessary
to fully understand the consequences of this disorder on the superconducting
properties of FeSe1−xTex.
5