High-Spin cobalt(II) ions in square planar coordination: Structures and magnetism of the oxysulfides Sr2CoO2Cu2S 2 and Ba2CoO2Cu2S2 and their solid solution

Article abstract of DOI:10.1021/ja109553u

The antiferromagnetic structures of the layered oxychalcogenides (Sr _1-xBa_x)₂CoO₂Cu₂S ₂ (0 ≥ x ≥ 1) have been determined by powder neutron diffraction. In these compounds Co²⁺ is

Full text of DOI:10.1021/ja109553u

ARTICLE
pubs.acs.org/JACS
High-Spin Cobalt(II) Ions in Square Planar Coordination: Structures
and Magnetism of the Oxysulfides Sr CoO Cu S and Ba CoO Cu S
2 2
2
2
2 2
2
2
and Their Solid Solution
†
†
‡
‡
†
Catherine F. Smura, Dinah R. Parker, Mohamed Zbiri, Mark R. Johnson, Zolt ꢀa n A. G ꢀa l,
,†
and Simon J. Clarke*
†
Inorganic Chemistry Laboratory, Department of Chemistry, University of Oxford, South Parks Road, Oxford OX1 3QR, U.K.
Institut Max von Laue-Paul Langevin, 6 rue Jules Horowitz, BP 156, 38042 Grenoble Cedex 9, France
‡
S Supporting Information
b
ABSTRACT: The antiferromagnetic structures of the layered
oxychalcogenides (Sr1-xBa ) CoO Cu S (0 e x e 1) have
x 2
2
2 2
been determined by powder neutron diffraction. In these
2
þ
compounds Co is coordinated by four oxide ions in a square
plane and two sulfide ions at the apexes of an extremely
tetragonally elongated octahedron; the polyhedra share oxide
vertexes. The magnetic reflections present in the diffraction
√
√
patterns can in all cases be indexed using a 2a ꢀ 2a ꢀ c
2
þ
expansion of the nuclear cell, and nearest-neighbor Co
moments couple antiferromagnetically within the CoO planes.
The ordered magnetic moment of Co in Sr CoO Cu S (x = 0) is 3.8(1) μ at 5 K, consistent with high-spin Co ions carrying
2
2
þ
2þ
2
2
2 2
B
three unpaired electrons and with an additional significant unquenched orbital component. Exposure of this compound to moist air
is shown to result in copper deficiency and a decrease in the size of the ordered moment to about 2.5 μ ; there is a strong correlation
B
between the size of the long-range ordered moment and the occupancy of the Cu site. Both the tetragonal elongation of the CoO4S2
polyhedron and the ordered moment in (Sr Ba ) CoO Cu S increase with increasing Ba content, and in Ba CoO Cu S , which
1
-x x 2
2
2 2
2
2
2 2
2þ
has Co in an environment that is close to purely square planar, the ordered moment of 4.5(1) μ at 5 K is over 0.7 μ larger than
that in Sr CoO Cu S , so the unquenched orbital component in this case is even larger than that observed in octahedral Co
B
B
2
þ
2
2
2 2
systems such as CoO. The experimental observations of antiferromagnetic ground states and the changes in properties resulting
from replacement of Sr by Ba are supported by ab initio calculations on Sr CoO Cu S and Ba CoO Cu S . The large orbital
2
2
2 2
2
2
2 2
moments in these systems apparently result from spin-orbit mixing of the unequally populated d , d , and d orbitals, which are
z2
xz yz
reckoned to be almost degenerate when the CoO S polyhedron reaches its maximum elongation. The magnitudes of the ordered
4
2
2
þ
moments in high-spin Co oxide, oxychalcogenide, and oxyhalide systems are shown to correlate well with the tetragonal
elongation of the coordination environment. The large orbital moments lead to an apparently magnetostrictive distortion of the
crystal structures below the Ne ꢀe l temperature, with the symmetry lowered from tetragonal I4/mmm to orthorhombic Immm and the
size of the distortion correlating well with the size of the long-range ordered moment for all compositions and for temperature-
dependent data gathered on Ba CoO Cu S .
2
2
2 2
2
þ
’
INTRODUCTION
Oxychalcogenides are a relatively under-investigated class of
divalent M in square planar (MO2)2- layers which are two-
dimensional fragments of the perovskite structure. The two layer
2
þ
types stack alternately with A cations in between. The transi-
tion metal cation M is in the center of a highly tetragonally
distorted MO4S2 octahedron. Oxychalcogenides with these
layered structures provide a counterpoint to important perov-
skite-related oxide phases, such as the three-dimensional cubic
solid-state compounds compared with oxides and other solids
containing only one type of anion. Due to the different sizes and
coordination requirements of the oxide and the heavier chalco-
genide anions, oxychalcogenides tend to adopt layered struc-
tures. In the oxysulfides A MO Cu S (where A is an alkaline
earth metal and M is a transition metal; compounds known for
A = Sr, M = Mn, Co, Ni, Cu, and Zn, and for A = Ba, M =
Co, which were first described, along with a series of related
compounds, by Zhu and Hor and co-workers ), the more
2
2
2 2
perovskites AMO3-δ and the layered Ruddlesden-Popper
4
1
2
3
3
1
(
A
M O
) phases. It is well-established that the chalco-
nþ1
n
3nþ1
2
genide layers present in the oxysulfides can readily accept holes in
1
þ
2-
polarizable sulfide anion bonds to Cu in (Cu S ) antifluor-
ite-type layers, and the less polarizable oxide anion bonds to the
Received: November 2, 2010
Published: February 8, 2011
2
2
r 2011 American Chemical Society
2691
dx.doi.org/10.1021/ja109553u
|J. Am. Chem. Soc. 2011, 133, 2691–2705

Journal of the American Chemical Society
ARTICLE
the antibonding states at the top of a valence band that is
composed of well-mixed Cu 3d/S 4p orbitals and that these
holes are very mobile. So the band gap insulator Sr ZnO Cu S
the chalcogenide layer and an insulator-to-metal transition on
cooling. However, there is no clear consensus in the literature as
to the nature of the behavior of Sr CoO Cu S . This paper
2
2
2 2
2
2
2 2
may be doped to produce a p-type metal, Na Sr ZnO Cu S ,
shows that Sr CoO Cu S is subject to aerial oxidation, with
x
2-x
2
2 2
2 2 2 2
5
with x ≈ 0.1. When the valence band is not doped with holes
important implications for the composition and magnetic prop-
erties, which suggest that some re-evaluation of some of the
reported properties of this compound may be required. Here we
describe the crystal and magnetic structures of the members of
the solid solution Sr2-xBa CoO Cu S (0 e x e 2), which offer
and the metal M in the oxide layer is a mid-to-late transition
2,6
metal, the compounds are semiconducting, suggesting that the
oxide layer behaves as a Mott-Hubbard insulator. The interac-
tions between the two electronically very different layers is via
weak bonding interactions between M and S, which may in
principle be tuned by the size of the A cation. It is expected that
investigation of compounds with this structure type could lead to
unusual magnetic and electronic behaviors complementary to
those of the better-studied oxides. The oxysulfides, by virtue of
containing anions of less electronegative Group 16 elements,
also tend to exhibit lower oxidation states for the transition
metals than are found in perovskite-type oxides. For example, the
x
2
2 2
2
þ
very unusual examples of high-spin Co ions in a ligand field
that, especially in the Ba-rich members, is very close to being
purely square planar because the CoO S polyhedron is extre-
4
2
mely tetragonally distorted with very long Co-S distances. In all
2
þ
these compounds, the Co ion carries an unquenched orbital
contribution to its magnetic moment. We correlate the trends in
the size of the ordered magnetic moment and the details of the
low-temperature structures with the increasing tetragonal elon-
2
þ
homologous series of compounds Sr MnO Cu
S
(m = 1,
gation of the CoO S polyhedron as Sr is replaced by the larger
2
2
2m-δ mþ1
4 2
6
2þ
2
, 3; δ = 0.5) that we have described contains Mn in a mean
Ba cation, and we compare this behavior with that reported for
2
þ
oxidation state of þ2.5 instead of the oxidation states between
þ3 and þ4 that are more normally encountered in manganite
perovskites. Furthermore, these compounds offer high-symme-
try structures with transition metal ions in strictly planar oxide
layers, irrespective of the nature of the alkaline earth cation or
the chalcogenide anion, and may be good model compounds
that will contribute to the development of theories of low-
dimensional solids. Related oxychalcogenides include com-
other magnetically concentrated Co systems. Our experimen-
tal results are supported by ab initio computational investigations
of the end member compounds that enable electronic and
magnetic properties to be correlated with the crystal structures.
’ EXPERIMENTAL SECTION
Synthesis. In all syntheses of oxysulfides, the reagents were cold
pressed into pellets and contained in alumina crucibles inside sealed,
predried evacuated silica tubes. Due to the air sensitivity of many of the
reagents and the products, manipulations of the solids prior to synthesis
were always performed in a Glovebox Technology Ltd. argon-filled
pounds containing planar M O layers with M two-coordinate
2
by oxide and four-coordinate by chalcogenide ions, such as
7
3
8,9
La Fe Se O and the recently reported La Co Se O .
2
2
2
2
2
2 3
The synthesis and characterization of the first layered cobalt
oxysulfides, A CoO Cu S (A = Sr, Ba), containing square planar
glovebox with an O
products were also handled under argon, but see below. (Sr1-x
Ba CoO Cu (0 e x e 1) were generally synthesized by reacting
2 2
and H O content of less than 1 ppm. In general the
2
2
2 2
2
-
CoO layers, were reported by Zhu et al. It was reported that the
2
)
x 2
2
2 2
S
compound with A = Sr could also support a small copper
together ((1-x)SrS þ xBaS), CuO (Aldrich, 99.99%), and Co (Aldrich,
deficiency, and a pure phase with composition Sr CoO Cu S
2
2
1.75 2
9
9.995%) in the molar ratio 2:2:1 at a temperature of 850 °C (x = 2) or
25 °C (x < 2) for 48 h. In some syntheses of Sr CoO samples,
as a fourth reactant
(Alfa, 99.9985%) and Co in the molar ratio
2:1.85:0.0375:0.8875 for the composition Sr CoO ). In later
syntheses of Sr CoO Cu S samples, the reactants were SrO, Co, Cu
was described. All subsequent property measurements on
Sr CoO Cu S have been made on samples synthesized with
8
2
2 2
Cu2-δS
2
2
2 2
copper deficiency was introduced by using Co
i.e., SrS, CuO, Co
3 4
O
what we assume, in the absence of statements to the contrary, to
be full Cu site occupancy and handled at least partly in ambient
(
3 4
O
2
,10-12
2
Cu1.85S
2 2
air.
Magnetic susceptibility measurements on Sr CoO2-
2
2
2
2 2
Cu S and Ba CoO Cu S show broad maxima at about 220
2
2
2
2
2 2
2
,10
2
(Alfa, 99.99%) and S (Alfa, 99.999%) in a 2:1:2:2 ratio, and in these cases
the sealed ampules were heated to 400 °C for 16 h to ensure reaction of
sulfur with the elemental metals prior to heating to the synthesis
K
and about 230 K, respectively, and this, together with the
observation in Sr CoO Cu S of magnetic Bragg peaks in
neutron powder diffraction patterns obtained below 200 K,
suggests low-dimensional antiferromagnetic ordering of the
2
2
2 2
2
-
1
temperature of 825 °C at 1 °C min and maintaining this temperature
for 32 h. SrO was made by decomposing SrCO (Alfa, 99.994%) at
00 °C under vacuum, with a final firing at 1100 °C to ensure complete
decomposition. BaS and SrS were made by heating BaCO (Alfa,
9.95%) and SrCO at 800 °C in a flow of CS (Aldrich, 99.5%) vapor,
2
þ
3
Co ions. The magnetic structures of Sr CoO Cu S and
2
2
2 2
9
Ba CoO Cu S have not been reported, nor have the magni-
2
2
2 2
2
þ
3
tudes of the ordered moments of the Co sublattice. Sr CoO
2
2-
9
3
2
Cu S has been found to show a divergence in field-cooled (FC)
2
2
transported by bubbling argon through the liquid, for two periods of 5 h
with intermediate grinding.
and zero-field-cooled (ZFC) magnetic susceptibilities below 80
2
,10
K
that is not observed in the Ba analogue. This apparent spin-
Caution! CS
from the flow system was passed through hydroxide bleach to destroy
excess CS , and the flow system was purged with pure argon after the
2
is highly toxic and highly flammable. The gas emerging
glass-type behavior in Sr CoO Cu S appears to be nonconven-
2
2
2 2
tional because the rate of decay of the magnetization is slower at
0 and 75 K than at an intermediate temperature (45 K) and the
time dependence of the magnetization cannot be fit to a stretched
2
1
apparatus had cooled to room temperature; the entire apparatus was
contained in a fume hood.
X-ray Powder Diffraction (XRPD). XRPD was used to assess the
phase purity of reactants and products prepared as described above.
Measurements were made using a PANalytical X’Pert PRO diffract-
ometer operating in Bragg-Brentano geometry with monochromatic
1
3
exponential form. Sr CoO Cu S was also reported to show
2
2
2 2
semiconducting behavior, while the resistivity of the Ba analogue
4
2
is 10 times higher. Subsequent investigations of the dielectric
response and the transport properties of sintered powders of
1
2
Sr CoO Cu S seem to suggest a possible spin-carrier phase
Cu KR radiation and a multiangle X’Celerator detector. High-resolu-
2
2
2 2
1
separation below T : electron transfer from the S 3p-based states
tion measurements on Sr CoO Cu S samples were performed on
N
2
2
2 2
at the top of the valence band onto Co results in mobile holes in
Station ID31 at the ESRF, Grenoble, using X-rays with Si-calibrated
2
692
dx.doi.org/10.1021/ja109553u |J. Am. Chem. Soc. 2011, 133, 2691–2705

Journal of the American Chemical Society
ARTICLE
wavelengths of 0.4-0.45 Å and with the samples contained in 0.7 mm
diameter glass capillary tubes. The capillaries were filled with helium
exchange gas to ensure good thermal contact between the powder and
the He flow cryostat and to avoid heating of the sample by the intense
X-ray beam. In early experiments on ID31, we discovered that, when the
capillaries were sealed under argon, below about 50 K the peak shapes
broadened and became highly asymmetric and the lattice parameters
showed an anomalous expansion, reaching values as large as or even
exceeding their room-temperature values (see Supporting Information,
Figure S9). This behavior we attribute to beam-heating of the sample
when the exchange gas has condensed.
Magnetic Susceptibility Measurements. Measurements were
carried out using either a Quantum Design MPMS5 or MPMS-XL
SQUID magnetometer in the temperature range 5-330 K. Measure-
ments were made on warming in a field of 0.01 or 0.1 T, first after cooling
in zero field (ZFC) and then again after cooling in the measuring field
(FC). The measured moment of Ba CoO Cu S at room temperature
did not show a linear relationship with the applied field in the low-field
region, presumably due to the presence of minuscule amounts of an
unidentified ferromagnetic impurity. Measurement of the sample’s true
bulk susceptibility was made by recording the moment as a function of
temperature in fields of 3 and 4 T (i.e., in the region in which the
ferromagnetic impurity had saturated), allowing calculation of the
moment at an “effective” field of 1 T.
2
2
2 2
Neutron Powder Diffraction (NPD). Time-of-flight (ToF)
NPD data were collected at the ISIS Facility, Rutherford Appleton
Laboratory, UK, using the medium-resolution diffractometer POLARIS,
the high-resolution powder diffractometer HRPD, and the diffract-
ometer OSIRIS, which is optimized for high-resolution studies at long
d-spacings. On POLARIS, samples of 2-5 g mass were sealed inside
cylindrical thin-walled vanadium cans using indium gaskets and mea-
sured between 2 and 298 K in the d-spacing range 0.4 < d < 8 Å by means
Computational Investigations. Calculations were carried out
2
0
using the projector augmented wave (PAW) formalism of the Kohn-
Sham formulation of the spin-polarized density functional theory
2
1,22
(KSDFT)
at the generalized gradient approximation (GGA) level,
23,24
implemented in the Vienna Ab-initio Simulation Package (VASP).2
5,26
The GGA was formulated by the Perdew-Burke-Ernzerhof (PBE)
3
of three banks of detectors located at scattering angles, 2θ, of 35° ( He
density functional. Electron correlation plays a predominant role when
magnetic interactions in transition-metal-containing materials are stu-
died. Indeed, the strongly correlated character of the electrons in the
3
tube detector), 90° (ZnS scintillator), and 145° ( He tube detector;
-3
highest resolution bank, Δd/d = 5 ꢀ 10 ). Total integrated proton
2
þ
currents at the production target were typically 500-1000 μA h g for
open-shell 3d levels of Co should be described accurately, and the self-
interaction error related to the density functional description of the
magnetic unpaired Co 3d electrons should be corrected. In this context,
the Hubbard correction to the KSDFT has been considered employing
3
3
these measurements. On HRPD, a 10 g sample of Ba
2
CoO
2
Cu
2
S
2
(CFS044) was contained in an aluminum “slab” can with vanadium
windows, and a thin gadolinium foil was used to mask the aluminum.
Measurements were carried out at 5-10 K intervals between 4 and 250
2
7
the Dudarev approach. In the latter formalism, the term Ueff = U - J,
representing the difference between the on-site (U) and intersite (J)
interactions, is added as a penalty functional to the KSDFT Hamiltonian.
Thus, J was kept fixed at 0 eV, whereas different values were tested for
the U parameter. All results were obtained using the experimental
magnetic structures and adopting a fixed spin configuration correspond-
ing to that observed. Further, non-collinear magnetic calculations
considering the additional spin-orbit Hamiltonian were performed.
The effect of spin-orbit coupling (SOC) was investigated by adopting
different easy-axis orientations (in-plane and axial) to evaluate the
dependence of the orbital magnetic moment contribution and the
magnetic anisotropy.
K for an integrated proton current of 17 μA h (30 min runs). Four-gram
3
samples of samples of Sr2-xBa
x
CoO
2
Cu
2
S
2
with x = 0.6, 0.7, 0.8, and 0.9
were also measured at 4 K on HRPD, for a total integrated proton
current at the production target of 75 μA h. Neutrons were measured
3
using three detector banks at 168° (ZnS scintillator; highest resolution
-
4
3
bank, Δd/d = 4 ꢀ 10 ), 90° (ZnS scintillator), and 35° ( He tubes).
x 2 2 2
On OSIRIS, data were collected on 4 g samples of Sr2-xBa CoO Cu S
with x = 0, 0.4, 0.7, and 0.8 over the d-spacing range 0.5-8 Å in six
3
overlapping segments using a single detector bank at 160° ( He tubes;
-
3
Δd/d ≈ 4 ꢀ 10 ), for an integrated proton current of 40 μA h.
3
Data were also collected using monochromatic radiation between 5 K
and room temperature on 4 g samples of Ba
2 2 2 2 2
CoO Cu S and Sr CoO2-
Cu S using the D2B diffractometer at the Institut Laue-Langevin (ILL),
Grenoble. Measurements were made using neutrons of wavelength 1.59
2
2
’
RESULTS AND DISCUSSION
Å selected by diffraction from the (335) planes of a 28-crystal Ge
Composition and Room-Temperature Crystal Structure of
Sr CoO Cu S . The initial report of Sr CoO Cu S suggested
that copper-deficient samples were possible and the phase width
monochromator. Patterns were collected over about 4 h using an array of
2
2
2
2
2
2
2 2
3
1
28 high-pressure He detectors with a step size of 0.05° in 2θ. Further
2 2 2 2
measurements of Sr CoO Cu S samples were made on D1A at ILL
of Sr CoO Cu S apparently covered the range 0 e δ e
2
2
2-δ 2
2
using 1.9 Å neutrons selected using the (115) planes of a 30-crystal Ge
0
.25. No air sensitivity was reported. We carried out a number of
monochromator. All structural refinements against PND data were
preliminary synthetic investigations in which the samples were
handled in ambient air after the synthesis was complete. When
samples of stoichiometric composition were prepared from the
reaction between SrS, Co, and CuO, we discovered that ele-
mental copper often precipitated out of the reaction, appearing
on the surface of the pellets, suggesting the existence of copper-
1
4
carried out using the Rietveld profile refinement suite, GSAS. Refine-
ment against data collected on POLARIS was carried out using all three
detector banks simultaneously, and on HRPD the 90° and 168° banks
2
þ
were used. The magnetic form factor for Co used in the refinements
15
was obtained from standard tables.
X-ray Single Crystal Diffraction. Measurements were per-
2
deficient phases as originally reported. A powder neutron
formed a single crystal of Ba CoO Cu S with a Nonius Kappa CCD
2
2
2 2
diffraction investigation on POLARIS of such a sample, which,
apart from the existence of elemental copper, appeared very close
to single phase by laboratory XRPD, revealed a refined value of
the Cu-site occupancy of 0.961(2) at room temperature and
diffractometer using Mo KR (λ = 0.71073 Å) incident radiation. The
crystal was mounted on a nylon loop using paratone oil, and measure-
ments were made at room temperature and at 110 K using an Oxford
Instruments cooling system integral to the diffractometer. Data reduc-
1
6
0
.955(2) at 2 K. These two independent measurements agreed
tion was performed using the DENZO-SMN software package.
Structure solution (using Direct Methods) and full matrix least-squares
well at the two temperatures, showing that the wide d-spacing
range on POLARIS minimizes correlation of the fractional site
occupancy with the displacement parameters and sample absorp-
17
refinement via the WinGX suite were carried out using SHELXS97 and
18
SHELXL97, respectively. The data were corrected for Lorentz-polar-
ization, and a face-indexed numerical absorption correction was applied
tion. An actual stoichiometry of Sr
oxysulfide phase was thus concluded from these results because
CoO Cu1.91(1)S for the
2
2 2
19
through the PLATON interface.
2
693
dx.doi.org/10.1021/ja109553u |J. Am. Chem. Soc. 2011, 133, 2691–2705

Journal of the American Chemical Society
ARTICLE
it seemed consistent with both the amount of elemental Cu in the
sample and with the amount of Cu in the main phase. This led to
attempts to prepare a range of single-phase samples with a copper
2
8
deficiency. In the course of these attempts, we discovered that
the samples derived from copper-deficient mixtures of reagents
persistently contained impurity phases (SrS and other unidenti-
fied impurities) that could not be eliminated by further grinding
and heating cycles. We also discovered that samples of Sr CoO
2
2-
Cu S were rather unobtrusively air-sensitive: although cursory
2
2
inspection of laboratory X-ray diffractograms revealed no ob-
vious degradation of samples over time (i.e., no new Bragg peaks
were evident in the powder diffractograms), more detailed
measurements revealed that there was a significant decrease
(
(
by 0.2-0.5%) in the unit cell volume on prolonged air exposure
over days and weeks). This also resulted in changes in
the magnetic susceptibility, which we discuss further below.
Once the air sensitivity was taken into account, NPD investiga-
tions of two samples synthesized with a composition Sr CoO2-
2
Cu1 S and never exposed to air (samples CFS127C and
.85 2
CFS130B) showed refined fractional occupancies of copper of
at least 98-99%—i.e., much closer to stoichiometric Sr CoO
2
2-
Figure 1. Rietveld refinements against POLARIS data for two portions
Cu S than intended (Supporting Information, Table S1).
2
2
of a single sample of Sr
had been carefully sequestered from air (upper refinement), and the
other had received exposure to a flow of moist O as described in the text
2 2 2 2
CoO Cu S (quenched, DRP053). One portion
We carried out a new series of experiments in order to (i)
investigate the effect of air exposure on the magnetic properties
and on the composition and structure, (ii) investigate the
possible effect of the cooling rate from the synthesis temperature,
and (iii) obtain reliable data on the low-temperature magnetic
properties of Sr CoO Cu S . Two samples of stoichiometric
2
(lower refinement). The insets show the low d-spacing regions and the
appearance of broad reflections corresponding to the most intense
reflections from the CuO that forms on exposure to moist O .
2
2
2
2 2
material (i.e., the composition of the reactant mixture was
Sr CoO Cu S ) were prepared from the reactants SrO, Co,
Cu, and S using two slightly different thermal treatments. One
to elemental copper, and refinement of the phase fraction of
elemental copper in the sample produced a value of 1.6(1)
mol %, accounting, within the experimental uncertainty, for the
remaining copper used in the synthesis but unaccounted for in
the main phase. The second portion of this sample, which had
received exposure to moist oxygen, had a unit cell volume 0.8%
smaller than that of the unexposed portion of the same sample,
and refinement of the fractional site occupancy of copper
2
2
2 2
sample (DRP051) was cooled slowly from the synthesis tem-
-
1
perature of 825 at 0.5 °C min , and the other (DRP053) was
cooled rapidly by quenching the sealed silica tube from 825 °C
into iced water. Laboratory XRPD measurements of these
samples revealed extremely small amounts of elemental Cu but
no other obvious impurity phases. These two samples were each
split into two portions. The first portion of each sample was
stored in an argon-filled glovebox, while the second portion of
each sample was exposed for about 16 h to a flow of oxygen that
was bubbled through water upstream of the sample, in order to
simulate rapidly the exposure of the sample to moist air. The
resulting four samples were each investigated using NPD mea-
surements. The two portions of the quenched sample (DRP053)
were measured at room temperature on the POLARIS diffract-
ometer at the ISIS facility, where the high Q-range is optimal for
determining structural parameters and compositions. The two
portions of this quenched sample were also measured at 5 K on
the D2B diffractometer at ILL and at a series of temperatures
using OSIRIS at ISIS, principally in order to probe the long-range
magnetic ordering. The two portions of the slow-cooled sample
produced a refined composition of Sr CoO Cu1.890(2)S , sug-
2 2 2
gesting that 5% of the Cu present in the sample is lost through
exposure to “moist air” in the form of a flow of H O-laden O .
2
2
Note that the refined displacement parameters for Cu and the
other atoms in the two portions of the sample are similar
(Table 1), suggesting that the decrease in refined Cu content is
real and not merely a consequence of parameter correlations in
the refinement. Close examination of the NPD pattern of the
portion of DRP053 that had been exposed to moist O revealed
2
no change in the phase fraction of elemental copper, but very
broad additional scattering was evident, which was consistent
with the formation of poorly crystalline CuO (Figure 1). Refine-
ment of the phase fraction of CuO produced a mole fraction of
4.8(2)%, suggesting that on air exposure the oxysulfide sample is
oxidized: copper ions are removed from the structure and form
the oxide. We have previously shown that this oxidative de-
intercalation of copper on air exposure with the formation of
poorly crystalline CuO occurs to a larger extent (∼13%) in
stoichiometric samples of CeCuOS that have similar copper
(
DRP051) were also measured on D2B at 5 K and at tempera-
tures from 10 to 290 K on the high-resolution powder diffract-
ometer ID31 at ESRF to probe changes in structure at low
temperatures, which we discuss further below.
3
þ
29
Refinement against the room-temperature data on the two
portions of the quenched sample DRP053 collected on PO-
LARIS (Figure 1) produced the results summarized in Table 1
and the structural model shown in Figure 4 (below). The sample
that had received no air exposure was very close to stoichiometric
sulfide layers and contain the oxidizable Ce cation. The
results obtained on the same two portions of this quenched
sample DRP053 from the D2B diffractometer were quantita-
tively consistent with the results obtained from POLARIS and
are summarized in Table S1 in the Supporting Information. The
two portions of the slow-cooled sample (DRP051) analyzed on
D2B (Table S1) showed behavior qualitatively similar to those of
in copper, with a refined composition of Sr CoO Cu S .
2
2
1.980(2) 2
The neutron diffractogram contained very weak Bragg peaks due
2
694
dx.doi.org/10.1021/ja109553u |J. Am. Chem. Soc. 2011, 133, 2691–2705

Journal of the American Chemical Society
ARTICLE
Table 1. Rietveld Refinement Results against NPD Data for
Sr CoO Cu S
2
2
2-δ 2
radiation
instrument
T/K
neutron, ToF
POLARIS
298(2)
space group
sample
I4/mmm
DRP053 (quenched),
DRP053, air-exposed
portion
unexposed portion
3.99129(2)
17.71555(9)
282.216(2)
87
a/Å
c/Å
3.98354(2)
17.6458(1)
280.014(2)
94
3
V/Å
no. of variables
R
wp
0.0144
0.0116
R
F
2
2
0.0294
0.0272
χ
3.33
2.80
a
Sr z
0.40767(1)
0.17119(3)
0.990(1)
0.71(1)
0.40646(1)
0.17072(3)
0.945(1)
0.75(1)
a
S z
b
F(Cu)
Sr Ueq ꢀ 10 (Å )
2
2
2
2
Co Ueq ꢀ 10 (Å )
0.91(2)
0.96(2)
Figure 2. Rietveld refinements against ID31 data for two portions of a
2
2
Cu Ueq ꢀ 10 (Å )
1.73(1)
1.76(1)
single sample of Sr
carefully sequestered from air (upper), and the other had received
exposure to a flow of moist O as described in the text (lower). The
2 2 2 2
CoO Cu S (slow-cooled, DRP051). One had been
2
2
S Ueq ꢀ 10 (Å )
0.81(1)
0.84(1)
2
2
2
O Ueq ꢀ 10 (Å )
0.76(1)
0.80(1)
insets show the effect of air exposure on the representative (217)
reflection, which broadens significantly and moves to a shorter d-spacing
as the unit cell contracts.
c
Co-O [4] (Å)
1.99565(1)
3.0327(5)
2.5803(2)
3.1491(3)
2.4356(3)
110.04(2)
109.19(1)
1.99177(1)
3.0126(5)
2.5868(2)
3.1288(3)
2.4339(3)
109.84(2)
109.29(1)
Co-S [2] (Å)
Sr-O [4] (Å)
Sr-S [4] (Å)
air, show that the unit cell volume can be used as a proxy for the
Cu content. Comparison of portions of both sample DRP051
Cu-S [4] (Å)
S-Cu-S [2] (°)
S-Cu-S [4] (°)
(
synthesized as Sr CoO Cu S ) and sample CFS126B
2 2 2 2
(synthesized as Sr CoO Cu S ) that had received no air
2
2
1.85 2
a
Co, 2a(0,0,0); O, 4c(0.5,0,0); Cu, 4d(0.5,0,0.25); Sr, 4e(0,0,z); S,
e(0,0,z). Z = 2. Site occupancy factor of Cu. Numbers in square
exposure at 250 K on ID31 revealed lattice parameters of a =
3.98955(1) Å and c = 17.67726(7) Å for DRP051 and a =
b
c
4
brackets indicate the multiplicity of each bond or angle.
3
.98942(2) and c = 17.67198(9) Å for CFS126B, giving a
difference in cell volumes of just 0.04%, consistent with the
refined Cu occupancies of 0.985(7) (DRP051, D2B data) and
0.976(6) (CFS126B, D1A data), which are equal within the
experimental uncertainty (see Supporting Information, Table S1
and Figure S1). Indeed, all our attempts to intentionally synthe-
size Cu-deficient materials at high temperatures led to the
formation of oxysulfide phases that are very close to fully
stoichiometric in copper with accompanying impurity phases.
Significant copper deficiency is achieved only through low-
temperature oxidative de-intercalation, which we have achieved
DRP053. There is no significant difference in compositions of the
portions of the samples that had received no exposure to air
which could have arisen from the different thermal treatments
during synthesis (i.e., the compositions appear to be unaffected
by the rate of cooling from the synthesis temperature). On
exposure to moist O , DRP051 underwent a decrease in cell
2
volume of 0.5% and a decrease in Cu content of 3.5%
(
Sr CoO Cu
S ), i.e., less than the corresponding values
2
2
1.92(2) 2
of 0.8% and 5% on oxidation of DRP053. These results reveal a
good correlation with unit cell volume and Cu content. Using the
high resolution of ID31 at ESRF, comparison at 290 K of the two
portions of sample DRP051 (Figure 2) shows that air exposure
results in asymmetric peak broadening, with full widths at half-
maximum (fwhm) of all classes of reflection increasing by a factor
of about 1.7. This is qualitatively similar to the behavior in
CeCuOS on air exposure and might arise from changes in
particle size or particle strain or from inhomogeneity in the
distribution of Cu in the sample. In summary, our synthetic and
neutron diffraction investigations to probe the composition show
that samples of nominal composition Sr CoO Cu S appear to
here using moist O . Exposure of the sample CFS126B (prepared
2
with nominal composition Sr CoO Cu S ) to ambient labora-
2
2
1.85 2
tory air resulted in a decrease in the cell volume of 0.2%. Heating
of this exposed sample under vacuum at 250 °C for 36 h led to an
28
increase in the unit cell volume by 0.1%. We presume that this
indicates partial recovery of the stoichiometric material using
low-temperature annealing, and this is supported by the mag-
netic susceptibility data described below, but we have not
examined this behavior in further detail.
29
Consider the structural changes that occur on oxidative loss of
copper from the structure of Sr CoO Cu S . Table 1 lists the
2
2
2 2
2
2
2 2
be intrinsically slightly copper deficient by about 1%. A much
larger copper deficiency of about 5% is introduced by the
oxidative de-intercalation of Cu in moist oxygen (or moist air)
to produce CuO. The results of these experiments, in which
pristine samples are exposed to H O-laden O simulating moist
important bond lengths and angles obtained from refinement
against room-temperature POLARIS data on the two portions of
sample DRP053 (quenched from the synthesis temperature). In
the unexposed portion of this sample of Sr CoO Cu S , the
2
2
2 2
Cu-S bond length of 2.4356(3) Å is similar to the values of
2
2
2
695
dx.doi.org/10.1021/ja109553u |J. Am. Chem. Soc. 2011, 133, 2691–2705

Journal of the American Chemical Society
ARTICLE
Figure 3. Rietveld refinements against POLARIS data for Ba
2
CoO2-
1
Cu
2
S
2
(sample CFS028) at room temperature. The inset shows the low
d-spacing region.
2
.4410(9) and 2.4369 Å in the band insulators Sr ZnO Cu S
2
2
2 2
3
0
and LaCuOS, respectively, with similar CuS layers and which
Figure 4. Structures of Sr
2 2 2 2 2 2 2 2
CoO Cu S (left) and Ba CoO Cu S
must have filled Cu 3d/S 3p valence bands. In contrast, the value
(right) obtained from refinement against POLARIS NPD data at room
31
of the Cu-S distance in TlCu S , which has one hole per
2
2
temperature and shown on the same scale. 95% anisotropic displace-
ment ellipsoids are shown, and selected bond lengths are given in Å.
Note the increase of almost 11% in the Co-S distance on substituting Sr
by Ba. Further structural parameters are given in Table 2.
formula unit in the Cu 3d/S 3p valence band, is 1.7% shorter
2.3943 Å) than in the insulating compounds because the
(
depleted states at the top of the Cu 3d/S 3p band are
antibonding. In Sr CoO Cu S , loss of 5% of the Cu from the
4
2
2
2 2
structure on air exposure generates 0.1 hole per formula unit, and
in these systems, and we cannot rule out a small phase width. The
refinement results for the NPD and SXRD data and selected bond
lengths are given in Table 2. The refined structures of Sr CoO Cu S
2 2
the Cu-S distance contracts on oxidation from 2.4356(3) to
2
.4339(3) Å. While these distances could be considered similar at
2
2
the 3σ level, the change is only about half what would be expected
if solely the Cu 3d/S 3p antibonding states are depleted on
oxidation of Sr CoO Cu S and if a linear decrease in Cu-S
and Ba CoO Cu S at room temperature are compared in Figure 4.
2 2 2 2
Analysis of POLARIS NPD data (Supporting Information,
Figures S2 and S3, and Table 2) of the compounds (Sr1-
Ba ) CoO Cu S , synthesized assuming full occupancy of the
2
2
2 2
bond length with the number of holes per formula unit is
x
x 2
2
2 2
assumed. The decreases in the Co-O and Co-S distances of
tetrahedral sites in the sulfide layer by copper ions, revealed small
amounts of elemental copper as the only obvious impurity phase
in the NPD patterns. Analysis of the weight fraction of elemental
copper suggested Cu deficiencies of 1-2%—similar to the values
in the end members. Free refinement of the copper site occu-
pancies for the phases containing both Sr and Ba systematically
tended to produce values exceeding unity by 1-2%, presumably
reflecting the effect of correlations with the displacement para-
meters, which are also systematically larger in these materials
than in the end members of the solid solution (Tables 1 and 2).
We have not investigated the possibility of variable copper
deficiency in any of the Ba-containing compounds. In conclu-
sion: across the entire solid solution, samples prepared using
stoichiometric mixtures of reagents appear to be very slightly
deficient in copper, but the deficiency is in the range 0-2%.
Structural Trends. The trends in lattice parameters and bond
lengths for the series (Sr1-xBa ) CoO Cu S at room tempera-
0
.2% and 0.7%, respectively, are consistent with slight Co
oxidation. Comparative bond valence sum calculations using
32
literature parameters (for Co-O) and estimated parameters
3
3
(for Co-S) suggest a 0.03 increase in Co oxidation state—i.e.,
2þ
about half what would be expected if solely Co were oxidized.
The changes in Cu-S, Co-O, and Co-S bond lengths are
therefore consistent with partial oxidation of both Cu 3d/S 3p
and Co 3d states. Removal of Cu, with no corresponding
decrease in the Cu-S distance, would be expected to lead to
underbonding of the sulfide ions, and it is observed that the Sr-S
distance decreases significantly by 0.65% in order to maintain the
bond valence of the sulfide ion. The Sr-O distance increases
accordingly by 0.25%. The observation that the bonds to sulfide
þ
change most significantly is consistent with the loss of Cu , and
this is in line with the results on CeCuOS.
Composition and Room-Temperature Crystal Structure of
29
x 2
2
2 2
(
Sr1-xBa ) CoO Cu S (0 e x e 1). While Sr CoO Cu S is
x 2
2
2
2
2
2
2 2
ture are shown in Figure 5, obtained from refinements against
room-temperature NPD data from POLARIS. Complete sub-
stitution of Sr by Ba leads to a 7% increase in c, which is more than
rather unobtrusively air-sensitive, Ba CoO Cu S is very ob-
2
2
2 2
viously air-sensitive, decomposing readily in air to produce
products that we did not attempt to analyze. The samples were
kept in an inert atmosphere. The structure of Ba CoO Cu S ,
3
times greater than the 2% increase in a. Expansion of the a
2
2
2 2
2
þ
2þ
lattice parameter as Ba replaces Sr is limited by the max-
imum Co-O bond length that can be maintained. The c
parameter, on the other hand, can accommodate much greater
expansion. In order to accommodate the larger Ba ion between
the CoO planes and the puckered Cu S layers, the Co-S
refined without incident from room-temperature XRSCD data,
producing a refined composition of Ba CoO Cu S . Inde-
2
2
1.95(1) 2
pendent refinement of the copper site occupancy factor and the
phase fraction of excess elemental copper using POLARIS NPD
data (Figure 3) also produced a composition of Ba CoO Cu
1.95-
2
þ
2
2 2
2
2
(1)S2.
samples suggested compositions of Ba CoO Cu
1.98(2) 2
NPD data from HRPD and D2B collected on two further
distance increases by 11%, from its already large value of
3.0327(5) Å in Sr CoO Cu to 3.3721(9) Å in Ba CoO2-
Cu . The coordination environment for Co thus becomes
much more tetragonally elongated, and approaches a CoO
square plane, as x increases. The Cu-S length remains essentially
S and
2
2
2
S
2
2
2
2
2
þ
Ba CoO Cu S , respectively (Supporting Information, Ta-
S
2 2
2
2
1.98(1) 2
ble S2). The analysis suggests that, similar to the case of
4
Sr CoO Cu S , there is a small intrinsic 1-2% copper deficiency
2
2
2 2
2
696
dx.doi.org/10.1021/ja109553u |J. Am. Chem. Soc. 2011, 133, 2691–2705

Journal of the American Chemical Society
ARTICLE
Table 2. Results of Structural Refinements for (Sr1-xBa ) CoO Cu S (x > 0) at Room Temperature
x 2
2
2 2
radiation
neutron, ToF
Mo KR, λ = 0.71073 Å
instrument
x in (Sr1-xBa
POLARIS
0.6
Nonius FR590 KappaCCD
1
x
)
2 2
CoO Cu
S
2 2
0.2
0.4
0.8
1
sample
T/K
CFS047A
298(2)
CFS046A
298(2)
CFS043A
298(2)
CFS040A
298(2)
CFS028
298(2)
CFS037
298(2)
space group
a/Å
I4/mmm
4.01016(4)
17.9295(2)
288.332(8)
67
I4/mmm
4.02538(4)
18.1680(3)
294.390(9)
67
I4/mmm
4.04013(4)
18.4025(3)
300.378(10)
67
I4/mmm
4.05341(3)
18.6447(2)
306.334(4)
68
I4/mmm
4.06420(3)
18.8870(2)
311.970(5)
50
I4/mmm
4.0711(1)
18.9094(6)
313.40(1)
16
c/Å
3
V/Å
no. of variables
R1/Rwp
0.0196
0.0166
0.0133
0.0154
0.0209
0.0178
wR2/R
F
2
0.0629
0.0680
0.1090
0.0809
0.0417
0.0385
2
χ
1.555
1.086
1.042
0.967
0.843
1.163
a
Sr/Ba z
0.40735
0.17272(5)
1.012(2)
0.989(1)
0.65(1)
0.40688
0.17431(5)
1.017(2)
0.978(1)
0.65(1)
0.40614
0.40514
0.17705(4)
0.994(2)
0.978(1)
0.65(1)
0.40378(3)
0.17854(5)
0.975(2)
0.976(1)
0.60(1)
0.40386(1)
0.17822(5)
0.972(4)
-
a
S z
0.17577(5)
1.029(3)
0.988(1)
0.57(1)
b
F(Cu) refined
F(Cu) from wt fraction
2
2
Sr/Ba Ueq ꢀ 10 (Å )
0.829(6)
0.86(1)
2
2
Co Ueq ꢀ 10 (Å )
1.07(3)
0.97(4)
1.12(5)
1.07(4)
0.81(2)
2
2
Cu Ueq ꢀ 10 (Å )
1.87(2)
1.96(2)
2.06(2)
1.92(2)
1.62(1)
1.81(2)
2
2
S Ueq ꢀ 10 (Å )
0.86(3)
0.90(4)
0.85(4)
0.91(3)
0.91(2)
1.05(1)
2
2
O Ueq ꢀ 10 (Å )
0.81(2)
0.88(2)
0.87(2)
0.92(2)
0.89(1)
0.96(3)
c
Co-O [4] (Å)
2.00508(2)
3.0967(9)
2.6037(3)
3.1783(5)
2.4373(5)
110.71(4)
108.86(2)
2.01269(2)
3.1668(9)
2.6293(3)
3.2058(5)
2.4376(5)
111.31(3)
108.56(2)
2.02006(2)
3.2348(10)
2.6579(3)
3.2301(5)
2.4386(5)
111.86(4)
108.29(2)
2.02670(1)
3.3011(8)
2.6899(3)
3.2501(4)
2.4408(4)
112.27(3)
2.03210(2)
3.3721(9)
2.7261(3)
3.2675(5)
2.4395(5)
112.82(4)
107.82(2)
2.03560(10)
3.3700(5)
2.72915(14)
3.2705(4)
2.4465(5)
112.61(4)
107.924(18)
Co-S [2] (Å)
Sr/Ba-O [4] (Å)
Sr/Ba-S [4] (Å)
Cu-S [4] (Å)
S-Cu-S [2] (°)
S-Cu-S [4] (°)
108.09(2)
b
a
c
Co, 2a(0,0,0); O, 4c(0.5,0,0); Cu, 4d(0.5,0,0.25); Sr/Ba, 4e(0,0,z); S, 4e(0,0,z). Z = 2. Site occupancy factor of Cu. Numbers in square brackets
indicate the multiplicity of each bond or angle.
constant across the series: the Cu S layers respond to the expansion
cobalt sublattice. Ba CoO Cu S shows a second maximum at a
2 2 2 2
2
2
in a by a squashing of the CuS tetrahedra in the c direction as the Ba
lower temperature, dependent on the presence of a Curie-like
term in the susceptibility at low temperatures, which presumably
arises from paramagnetic impurities. The mixed A-site phases
4
content increases (Figure 5). This flexibility of the Cu S layer is
2
2
presumably important in enabling facile A-site substitution to occur
in this structure type. In the similar Fe As layers in the iron-based
(Sr1-xBa ) CoO Cu S show similar features: broad high-tem-
2
2
x 2 2 2 2
superconductors LnFeAsO, the As-Fe-As angle correlates with
perature maxima at about 250 K and, for x g 0.6, secondary
maxima between 140 and 170 K (Supporting Information, Figure
the lanthanide radius and seems to be one of the parameters that
34
controls the physical properties. The mean A-O and A-S bond
lengths obtained from the refinements do not vary linearly with
composition (Figure 5), presumably because we constrained the
mixture of Sr and Ba cations to lie on a single site. The refinements
are consistent with complete long-range disorder of Sr and Ba on the
A cation site, although anisotropic broadening terms in the profile
model were required in the refinements, which we assume to be a
consequence of short-range ordering of these cations or composi-
tional inhomogeneity (see Supporting Information, Figure S2, for
an example). The key structural change is the increasing tetragonal
S6). Sr CoO Cu S , which has been carefully sequestered from
2 2 2 2
air, also shows divergence of the ZFC and FC susceptibilities
below 80 K, consistent with a glassy component to the magnet-
ism. On exposure of Sr CoO Cu S to moist air, the behavior of
2 2 2 2
the magnetic susceptibility with temperature changes very sig-
nificantly, as shown in Figure 6, although the origin of the new
features, such as the sharp drop in the susceptibility below 130 K,
is not clear. This change, which is reproducible from sample to
sample (see Supporting Information, Figure S7), is diagnostic of
the degree of degradation. The samples previously reported in
the literature have magnetic susceptibilities similar to those of
our samples that had not been exposed to air, suggesting that the
other reports of the susceptibility were indeed likely measuring
2þ
elongation of the Co environment, which correlates with the
magnetic properties as described below.
Magnetic Properties. Magnetic Susceptibility Measure-
2
,10
ments. Magnetic susceptibility measurements on Sr CoO
material with almost full copper occupancy in the sulfide layer.
2
2-
Cu S and Ba CoO Cu S (Figure 6) produced results that
The change in the magnetic susceptibility is at least partially
reversible. Annealing of an air-exposed portion of sample
CFS126B at 250 °C under vacuum, which was found to increase
the cell volume by 0.1% (see above), or treatment under dilute
H (5% H in N ), largely restored the susceptibility of
2
2
2
2
2 2
2,10,11
were consistent with those previously reported.
Prior to air
exposure, Sr CoO Cu S shows a broad maximum at about 220
2
2
2 2
K, and Ba CoO Cu S shows a similar maximum at 235 K,
2
2
2 2
suggestive of low-dimensional antiferromagnetic ordering of the
2
2
2
2
697
dx.doi.org/10.1021/ja109553u |J. Am. Chem. Soc. 2011, 133, 2691–2705

Journal of the American Chemical Society
ARTICLE
Figure 6. (Top) Magnetic susceptibility of Sr CoO Cu S (quenched
2
2
2 2
sample, DRP053) in its pristine form and after exposure to a flow of
moist O . Zero-field-cooled (ZFC) and field-cooled (FC) measure-
2
ments made in a measuring field of 100 Oe are shown. (Bottom)
Magnetic susceptibilities of two samples of Ba CoO Cu S measured in
2
2
2 2
effective applied fields of 1 T.
Figure 5. Variation of lattice parameters and selected structural para-
meters at room temperature with composition in the solid solution
(
Sr1-xBa ) CoO Cu S . Lines are guides to the eye. Error bars lie
x 2 2 2 2
within the points.
the pristine material (Figure S7), although these processes also
introduced some impurities. This is further evidence that the
copper-deficient materials obtained by exposure to moist O are
2
not stable with respect to the stoichiometric material at elevated
temperatures. In the reduction experiment using dilute H , the
2
Figure 7. Refinement of the crystal and magnetic structure of Sr
Cu (quenched sample DRP053) at 5 K against D2B data. The data
red points), fit (green line), and difference (lower magenta line) are
shown together with tic marks indicating nuclear (black) and magnetic
(red) reflections. χ = 3.96, Rwp = 0.0629, and R(F ) = 0.0405. The inset
shows the model used for the magnetic structure.
2
CoO2-
CuO that formed on oxidation of the pristine material is
presumably reduced and the copper reinserted. This is consistent
2 2
S
(
29
with the observation that Cu can be reinserted into CeCu1-xOS
2
2
and suggests that the Cu deficiency is diminished on heating the air-
exposed and hence copper-deficient material to moderate tem-
peratures. Further measurements will be required to establish the
precise origin of the divergence in the unexposed and exposed
samples, and there are some reports of unusual behavior in the
magnetic structural model used for all the compounds described
in this paper is shown schematically in Figure 7. We used a
magnetic model in P1 symmetry with moments arranged colli-
nearly and nearest-neighbor moments antiparallel, but otherwise
unconstrained; other models are possible, such as a non-collinear
model in which the moments in adjacent antiferromagnetic
planes are rotated by 90° with respect to one another in the
12,13
regime below 80 K.
Long-Range Magnetic Ordering. Low-temperature NPD
measurements were performed using the constant wavelength
diffractometers D1A and D2B at ILL and the ToF diffract-
ometers HRPD, POLARIS, and OSIRIS at the ISIS facility. The
magnetic Bragg reflections observed for all samples were in-
35
basal plane, but these models are indistinguishable using a
powder diffraction measurement. The moments lie mainly in the
basal plane, with a small component directed along the z axis.
Sr CoO Cu S . The samples DRP051 (slow-cooled) and
2
dexed, as suggested previously for Sr CoO Cu S , using a
2
2
2 2
√
√
magnetic cell that is a 2a ꢀ 2a ꢀ c expansion of the nuclear
cell. The magnetic structure has antiferromagnetic coupling of
2
2
2 2
nearest-neighbor moments within the CoO planes, and the
spins in next-nearest-neighbor layers, which are separated by
translation by the lattice parameter c, are aligned parallel. The
DRP053 (quenched) described above were both measured on
the D2B diffractometer. Refinement of the ordered moment on
the cobalt ions (Figure 7) revealed that, in the portions of each
2
2
698
dx.doi.org/10.1021/ja109553u |J. Am. Chem. Soc. 2011, 133, 2691–2705

Journal of the American Chemical Society
ARTICLE
2
þ
Table 3. Refined Co Magnetic Moment Components from NPD Data at 5 K
compound
sample
instrument
M
x
B
(μ )
M
y
(μ
B
)
M
z
(μ
B
)
B
|M| (μ )
Sr
Sr
Sr
Sr
2
2
2
2
CoO
CoO
CoO
CoO
2
2
2
2
Cu
Cu
Cu
Cu
2
2
2
2
S
S
S
S
2
2
2
2
DRP051 (slow-cooled, unexposed)
DRP051 (slow-cooled, exposed to moist O
DRP053 (quenched, unexposed)
D2B
D2B
D2B
D2B
D2B
HRPD
2.94(7)
2.0(1)
2.39(8)
2.0(1)
0.4(5)
1.0(3)
0.9(2)
0.9(2)
1.2(2)
3.81(6)
3.0(1)
2
)
2.85(7)
1.6(1)
2.42(7)
1.6(1)
3.85(6)
2.44(6)
4.45(9)
4.53(1)
2
DRP053 (quenched, exposed to moist O )
Ba
2
2
CoO
2
Cu
Cu
2
S
2
CFS071B
CFS044A
3.05(9)
2.92(2)
3.01(9)
3.21(2)
Ba
CoO
2
2
S
2
1.32(4)
Figure 9. Refinement of the crystal and magnetic structure of Ba2-
CoO Cu (Immm) against D2B NPD data collected at 5 K. Positions
2
2 2
S
of nuclear and magnetic reflections are given by black and red tic marks,
Figure 8. Plot of refined Cu site occupancy against ordered magnetic
moment from neutron diffraction measurements on several “Sr CoO
respectively; blue and green correspond to scattering from elemental Cu
2
2
2
2-
wp
and the vanadium sample holder. χ = 1.947, R = 0.0326, and R(F ) =
Cu
synthesized stoichiometrically. Sample CFS127C (green) was synthe-
sized as “Sr CoO ”. For these three samples, measurements
were made both with no air exposure and after controlled exposure to a
flow of moist O
2 2
S ” samples. Samples DRP051 (red) and DRP053 (blue) were
0.0477. The inset shows the temperature dependence of the moment
obtained from refinements against D2B and HRPD data. The poor fit to
the low-temperature D2B data in I4/mmm is shown in the Supporting
Information, Figure S4.
2
2 2
Cu1.85S
2
.
3
þ
sample that had received no air exposure, the ordered moments
to Co should lead to a reduced moment per Co ion, even if the
2
þ
3þ
were both 3.8(1) μ . The refined components of the moments
ion remains high spin, and disorder between Co and Co
B
are listed in Table 3. The unexposed portion of sample DRP053
was also measured using the OSIRIS diffractometer, and the
species would also reduce the long-range ordered moment that is
measured in a diffraction experiment. It is also plausible that
some of the Co 3d electrons become itinerant on oxidation,
reducing the local moment, but our current measurements do
not allow us to disentangle these possible contributions.
refined moment from this data set was 3.5(1) μ , 8(5)% smaller
B
than the value from D2B. We presume that this discrepancy
arises from the correlation of the refined ordered moment with
sample absorption. On exposure to moist oxygen, simulating
moist air, which we have shown above results in a Cu deficiency
and significant changes in the magnetic susceptibility, the
ordered moments of the two samples DRP051 and DRP053
measured on D2B decreased significantly to 3.0(1) and 2.4(1)
μB, respectively. The magnetic model used remained unchanged,
and there is little effect on the orientation of the moment. In the
course of our investigations of the possibility of synthesizing
copper-deficient samples Sr CoO Cu S , and in investigating
Ba CoO Cu S . PND measurements on Ba CoO Cu S car-
2
2
2 2
2
2
2 2
ried out on D2B (Figure 9) and HRPD revealed magnetic
reflections visible below 210 K, which reached saturation below
about 80 K. In this case, the refined magnetic moment at 5 K was
4.5 μ for both sample CFS071B measured on D2B and sample
B
CFS044A measured on HRPD (Table 3). The moment obtained
for Ba CoO Cu S is at least 0.7 μ larger than that obtained for
2
2
2 2
B
unexposed samples of Sr CoO Cu S and, therefore, very much
2
2
2 2
larger than the spin-only moment, indicating that there is an even
2
2
2-δ 2
the sensitivity of Sr CoO Cu S to ambient air and to a moisture-
larger orbital contribution to the moment in Ba CoO Cu S ,
2
2
2 2
2
2
2 2
2
þ
laden O flow, we gathered several neutron diffractograms from
with a more tetragonally elongated Co environment. We note
here that the room-temperature I4/mmm structural model for
Ba CoO Cu S was found to fit the D2B data poorly at low
2
which we can refine both the ordered moment and the copper
site occupancy (see Table S1 in the Supporting Information for
further details of these samples). These results reveal an approxi-
mately linear dependence of the ordered moment against refined
copper site occupancy, as shown in Figure 8.
2
2
2 2
temperatures (Supporting Information, Figure S4), and a reduc-
tion in symmetry to Immm, but retaining the unit cell volume,
was required to model the data (a = 4.0466(3), b = 4.0602(3),
and c = 18.755(2) Å at 5 K; Supporting Information, Table S2).
We consider this structural distortion in more detail below.
(Sr1-xBa ) CoO Cu S . Several samples were measured on
The long-range ordered moment of about 3.8(1) μB in
Sr CoO Cu S phases that are no more than 1% deficient in
2
2
2 2
Cu shows that there is a very significant orbital contribution to
x 2
2
2 2
2
þ
the magnetic moment of Co (S = 3/2) in a highly tetragonally
the ToF diffractometers OSIRIS and HRPD at ISIS. Low-
temperature HRPD data, from which we obtained structural
information (see below), have a poor signal-to-noise ratio at low
Q but were used to define the nuclear structural model in the
distorted CoO S coordination environment, and we return to
4
2
this point below. A decrease in the ordered moment on oxidation
2
þ
by moist O is expected: oxidation of up to 10% of the Co ions
2
2
699
dx.doi.org/10.1021/ja109553u |J. Am. Chem. Soc. 2011, 133, 2691–2705

Journal of the American Chemical Society
ARTICLE
Figure 11. Cartoon showing the d orbital splitting for molecular high-
spin Co systems ranging from purely octahedral (a) to purely square
planar (d) via a slightly tetragonally elongated octahedron (b) and the
Figure 10. Variation of refined long-range ordered moment on Co as a
function of composition for the series (Sr1-xBa ) CoO Cu S . Three
x 2 2 2 2
samples were measured on two diffractometers, as indicated by the ovals
linking the points.
2
þ
highly tetragonally elongated arrangement appropriate to Sr CoO
2-
2
Cu S (c). (Note that, as far as we are able to ascertain, the purely
2
2
2
þ
square planar high-spin Co system has not been realized experimen-
tally in a molecular system.) The d and d orbitals are σ-antibond-
ing, and the d , d , and d orbitals are π-antibonding.
refinements against OSIRIS data on some of the same samples
x
2
-y
2
z
2
(Supporting Information, Figure S3). Data from OSIRIS offer
xy xz
yz
extremely high signal-to-noise ratios and resolution at low Q and
enable small CoO impurities that were not evident in the nuclear
refinements to be observed (Figure S3), but the high Q data are
not optimal for defining the structural model for these com-
pounds with small unit cells. The values of the refined magnetic
moments are somewhat correlated with the diffractometer used
for the refinement, as discussed for Sr CoO Cu S above, but
linearly with x in (Sr1-xBa ) CoO Cu S , as shown in Figures 10
x 2
2
2 2
and 5, respectively.
Computation. We have carried out first-principles electronic
structure calculations of the stoichiometric Sr CoO Cu S and
2
2
2 2
Ba CoO Cu S to gain insight into the magnetic behavior of
2
2
2 2
2
2
2 2
these nonmolecular compounds using an on-site Coulombic
correction in terms of the Hubbard term U to describe the 3d
electron-electron repulsion of the cobalt cation. Different values
of U were tested. Results for the case U = 3 eV are shown. The
inclusion of the U parameter was found to stabilize strongly the
observed antiferromagnetic ground state over an alternative
ferromagnetic ground state for both compounds, by 0.42 eV in
Sr CoO Cu S and 1.73 eV in Ba CoO Cu S . If this parameter
using realistic uncertainties in the values of the ordered moments,
analysis of several samples of (Sr1-xBa ) CoO Cu S revealed
x 2
2
2 2
that there was an approximately linear increase in the size of the
ordered moment with increasing Ba content, as shown in
Figure 10.
2
þ
Orbital Enhancements of the Moment in Co Compounds.
7
2þ
It is well-known that high-spin d Co ions in octahedral
coordination carry an unquenched orbital contribution to the
2
2
2 2
2
2
2 2
4
36
was not included in the calculations (i.e., U = 0), the ferromag-
magnetic moment arising from the T1 ground term. In the
g
netic state was favored by 0.53 eV for Sr CoO Cu S and 0.74 eV
2
2
2 2
case of a small distortion of a CoO octahedron from O
6
h
4
for Ba CoO Cu S .
2 2 2 2
symmetry by axial elongation, the octahedral T ground term
1
g
4
4
The total electronic densities of states (EDOS) for the two
compounds with U = 3 eV are compared in Figure 12 (results for
Sr CoO Cu S showing the opening of the gap at E as U is
splits into a lower-lying A ground term and a higher-lying E
term in D symmetry. The A ground term cannot carry a first-
2
g
g
4
4
h
2g
2
2
2 2
F
order orbital moment, but there will remain an orbital contribu-
4
increased are included in the Supporting Information, Figure
S8). The site-projected EDOS for Sr CoO Cu S is also shown
tion to the moment through the mixing-in of higher lying E
g
terms via spin-orbit coupling. The size of the orbital contribu-
tion to the moment will diminish as the size of the term splitting
increases relative to the spin-orbit coupling parameter. It is
perhaps convenient to consider the situation in terms of the
familiar 3d orbitals: axial elongation lifts the degeneracy of the
octahedral e and t sets of orbitals. Of the σ-antibonding (e in
2
2
2 2
in Figure 12. Clearly, the top of the valence band is composed of
well-mixed Co 3d/O 2p states and Cu 3d/S 3p states, so our
conjecture above—that the changes in Cu-S, Co-O, and Co-
S bond lengths when Sr CoO Cu S is subjected to mild
2
2
2 2
oxidation result from oxidation of the Cu 3d/S 3p states and
the mainly Co 3d states—is supported by the calculations. The
suggestion in ref 12 that there is a temperature-dependent
g
2g
g
octahedral symmetry) orbitals, d
x2-y2, which is σ-antibonding
with respect to the equatorial ligands, will be destabilized, while
dz2 will be stabilized. Of the π-antibonding (t2g in octahedral
electron transfer between the CoO and Cu S layers is also
2 2 2
symmetry) orbitals, dxy will be destabilized because it is π-
antibonding with respect to all four equatorial ligands, while
d and d , which are each antibonding with respect to only two
certainly not implausible from our calculations, although it is not
clear whether the sample in ref 12 is stoichiometric or slightly
copper-deficient as a result of air exposure.
xz
yz
of the equatorial ligands and to the two axial ligands, will be
Consider in more detail the projection of the Co 3d EDOS.
Figure 13 compares the results for Sr CoO Cu S and Ba CoO
2-
2
þ
7
stabilized. For a slightly axially elongated high-spin Co d ion,
2
2
2 2
2
4
1
the configuration is therefore expected to be (d ,d ) (d ) -
Cu S, which show that the effect of increasing the tetragonal
xz yz
xy
2
1
1
(dz
2
) (d
x2-y2
) , as shown schematically in Figure 11b. In the
elongation of the CoO S octahedron is profound. In Sr CoO
4 2 2 2-
(
Sr1-xBa ) CoO Cu S series, there are clearly large orbital
Cu S , the result is that the R spin EDOS lies below E , while the
x 2
2
2 2
2 2
F
enhancements to the magnetic moments, and the size of the
magnetic moment increases as the degree of tetragonal elonga-
tion of the Co environment increases: both the ordered
β components of d
2, d , and d
2
2
are unoccupied (above E ).
z
xy
x -y
F
This situation corresponds to the configuration derived above
2
þ
and shown in Figure 11b,c. In the case of Ba CoO Cu S , the
2
2
2 2
moment and the (Co-S)/(Co-O) ratio vary approximately
calculations show a slightly different picture (Figure 13). With
2
700
dx.doi.org/10.1021/ja109553u |J. Am. Chem. Soc. 2011, 133, 2691–2705

Journal of the American Chemical Society
ARTICLE
3
8,39
about 4 μ .
The spin-only moment expected for high-spin
B
2
þ
Co is 3 μ , but in a collective magnetic system, the value of the
B
ordered moment measured by neutron diffraction tends to be
40
reduced by covalency, so CoO has a very large orbital moment
in excess of 1 μ . A small axial compression of the octahedron in
B
CoO accompanying the cubic-to-monoclinic distortion below
3
8
2
90 K results in r(Co-O) /r(Co-O) = 0.99, and the
ax
eq
ground term should retain the orbital contribution to the
moment.
In La CoO , which has tetragonal symmetry at low tem-
2
4
35
peratures,2 there is distortion from octahedral symmetry
þ
4
about Co via axial elongation ( A2 ground term) such that
g
r(Co-O) /r(Co-O) = 1.05 (i.e., similar to Figure 11b). The
ax
eq
3
5
low-temperature ordered moment is 2.9 ( 0.1 μ , which
suggests that only a small orbital contribution is retained by
mixing-in of excited E terms. Similar ordered moments for
distorted octahedral Co ions are 3.0 ( 0.1 μ in both La -
Co O5 and La Co O (both orthorhombic), in which the
B
4
g
2
þ
42
B
2
41
2
4
3 9
CoO polyhedra are distorted to a slightly larger extent than in
6
La CoO , with four short distances of 1.93-2.03 Å and two
2
4
longer distances of about 2.25 Å. In the current oxysulfide
systems, there is a much larger axial anisotropy than in these
oxide systems (Figure 11c), and the coordination environment
Figure 12. (a) Total electronic density of states for Sr CoO Cu S and
2
2
2 2
Ba
Sr
2
CoO
CoO
2
Cu
Cu
2
S
2
for U = 3 eV. (b) Total and site-projected EDOS for
obtained for the antiferromagnetic magnetic spin con-
S
2
2þ
2
2
2
for Co approaches square planar, as noted in the title of the
figuration with U = 3 eV.
2
original communication. The data presented here show that, far
from the orbital moment being quenched completely by this
extreme distortion from octahedral symmetry, the orbital con-
tribution in Ba CoO Cu S , which has the most tetragonally
2
2
2 2
elongated (i.e., closest to square planar) coordination environ-
2
þ
ment we are aware of for a high-spin Co ion, is even larger than
that in CoO. The significant unquenched orbital moment in
similar systems is also noted by Knee et al. in the compounds
37
Sr CoO X (X = Cl, Br), which are formally related to the
2
2 2
oxysulfides by replacement of the chalcogenide ion by a halide
þ
and removal of the Cu ions from the nonoxide layer. These
compounds are reported to exhibit magnetic structures similar
to those of the oxysulfides described here. Our calculations (see
Figure 13) suggest that, in Ba CoO Cu S , we may have
2
2
2 2
approached the point indicated by Figure 11d, where the dz
2
is
the least antibonding orbital. Square planar systems are known
2
þ
43
2þ
for Co , and Ceulemans et al. considered Co -Schiff base
compounds, which are low-spin because of the higher ligand field
from N ligands. In those systems, spectroscopic data and reason-
able values for the Racah parameters B and C were used to
estimate the relative energies of the d orbitals. The analysis of this
Figure 13. Angular-momentum-resolved EDOS for the Co 3d levels
(
2 2 2 2 2 2 2 2
l = 2) for Sr CoO Cu S (top) and Ba CoO Cu S (bottom).
molecular system showed that the 3dz orbital was of lowest
2
energy and was estimated to be approximately nonbonding as a
the increased tetragonal elongation of the CoO S octahedron,
result of mixing with the Co 4s orbital; the π-type antibond-
4
2
-
1
the picture that emerges from the calculations is that now the dz
2
ing orbitals were located approximately 2000 cm (0.25 eV)
1
-
β spin component is occupied, while the dxz and d β compo-
(d ,d ) and 4000 cm (0.5 eV) (d ) higher, with the dx2-y2
xz yz xy
1
yz
-
nents are only half occupied. There is also greatly increased
σ-antibonding orbital approximately 20 000 cm (2.5 eV)
higher. In the oxysulfide and oxyhalide cases, the axial ligands
mixing of the dz
2
and d ,d spin orbitals compared with the
xz yz
44
Sr CoO Cu S case. The calculations suggest that, in our simple
are not completely removed. Demazeau et al., in considering
octahedral distortions in K NiF -type compounds, estimated
2
2
2 2
ligand field picture, the Ba CoO Cu S system is very close to
2
2
2 2
2
4
4
4
-5
the crossover from the A (Figure 11c) to the E (Figure 11d),
that the ligand field strength falls off roughly as r , where r is
2g
g
and the dz
2
is very close to degenerate with d and d .
the internuclear distance. In Ba CoO Cu S , the interatomic
xy
xz
2
2
2 2
2
þ
The environment for high-spin Co in oxide systems ranges
from almost purely octahedral in CoO to approximately square
planar (i.e., highly tetragonally elongated octahedral) in the
layered oxysulfides described here and in related oxyhalides such
distances are r (Co-S) = 3.37 Å and r (Co-O) = 2.03 Å. The
Ba
Ba
2- 2-
radius of S exceeds that of O by about r
according to standard tables, so the ratio of the axial and
equatorial ligand field strengths will be roughly σ /σ
[(r (Co-S) - r )/r (Co-O)] = [(3.37 - 0.4)/2.03]
≈ 0.15. Bearing in mind that sulfide is a weaker field ligand than
≈ 0.4 Å
S-O
45
=
ax eq
37
-5
-5
as Sr CoO X (X = Cl, Br). In CoO, which has the rock salt
2
2
2
Ba
S-O Ba
structure, the magnitude of the ordered magnetic moment is
2
701
dx.doi.org/10.1021/ja109553u |J. Am. Chem. Soc. 2011, 133, 2691–2705

Journal of the American Chemical Society
ARTICLE
oxide and that we are making several assumptions, we estimate that,
as a result of the weak σ-type axial ligand field of the sulfide, the dz
orbital will be raised relative to its energy in the purely square planar
case by about 10% compared with the d orbital. The net result,
2
2
2
x -y
43
following the result of Ceulemans et al., is that dz2 is expected to
be close to degenerate with dxz,yz, and these three orbitals will have
7
one hole between them in a high-spin d system. This is consistent
with the picture that emerges from our calculations. Spin-orbit
coupling mixes d
2
with d
to an extent that depends on the
z
xz,yz
energy separation of the levels, and since the orbitals are close to
degenerate in energy, we propose that this is the origin of the very
large orbital contribution to the moment in Ba CoO Cu S . It is
2
2
2 2
4
unclear whether the ground term in this case is E [corresponding
x2-y2) ] with a first-
order orbital moment or A [corresponding to the configuration
2þ
g
2
3
1
1
Figure 14. Changing magnetic moment of Co as a function of the Co
coordination geometry in various solid-state compounds from the
literature (red) and from this work (black). Literature values are used
to the configuration (dz
2
) (d ,d ) (d ) (d
xz yz
xy
4
2
g
4
1
1
1
4
3
5,38,39,41,42
37
(
d ,d ) (d ) (d ) (d
z2
x2-y2
g
) ] with the E close in energy. In
xz yz
xy
for the bond lengths in the oxides
and oxyhalides, with
Sr CoO Cu S and the oxyhalides Sr CoO Cl and Sr Co-
average values for the axial (ꢀ2) and equatorial (ꢀ4) Co-O bond
2
2
2 2
2
2
2
2
3
7
2þ 41,42
O Br , the anisotropy is smaller, and we estimate that σ /
lengths for systems with lower than D symmetry at Co
.
The line
2
2
ax
4h
-5
σeq = [(r (Co-S) - r )/r (Co-O)] = [(3.01 - 0.4)/
is a guide to the eye.
Sr
S-O Sr
-
5
43
1
.99] ≈ 0.25, which, following Ceulemans et al., would place
-1
2
dz
above d
by about 2000 cm (0.25 eV) and roughly
xz,yz
degenerate with d (i.e., Figure 11c), which seems to be borne
xy
out by our calculations (Figure 13). The many-electron spin-orbit
2þ
-1
coupling parameter, λ, for Co is around -200 cm (-25 meV),
so the mixing of the orbitals in Sr CoO Cu S will be significantly
2
2
2 2
less than in the case of Ba CoO Cu S , and the orbital moment will
2
2
2 2
be correspondingly reduced. Finally, delocalization onto the ligands
40
tends to reduce the spin and orbital moments. As the Co-O
distance increases, this reduction is expected to decrease further,
enhancing the ordered moment, and this may partially account for
the larger ordered moment in Ba CoO Cu S .
2
2
2 2
Calculations were also performed in which spin-orbit cou-
pling was included (Supporting Information, Table S3). These
showed two things: First, there is a strong in-plane magnetic
anisotropy in these systems, consistent with the experimental
observation that the magnetic moments have their major com-
ponent oriented in the CoO planes. Second, the calculations
2
showed qualitatively that the orbital component of the magnetic
moment should be larger in the case of Ba CoO Cu S than
2
2
2 2
Sr CoO Cu S , although the difference in the size of the experi-
2
2
2 2
mental moments of about 0.7 μ is about a factor of 5 larger than
B
that suggested by the calculations, as is commonly found in
2 2 2 2
Figure 15. Refinements of Ba CoO Cu S against HRPD NPD data
comparable computations due to the nature of the numerical
collected at 250 and 4 K (165° detector bank). Positions of nuclear and
magnetic reflections are given by black and red tic marks, respectively;
blue tic marks correspond to scattering from the small amount of
elemental Cu in the sample.
46
integration in these calculations. The calculations predict a
larger z component of the magnetic moment in Ba CoO Cu S
2
2
2 2
than in Sr CoO Cu S , and the data in Table 3 support this,
2
2
2 2
although the uncertainties are large in the M values.
z
2
þ
A survey of all the literature examples we can find with Co in
octahedral or axially distended environments reveals a sharp
decrease in the size of the long-range ordered magnetic moment
on distortion from octahedral symmetry, followed by a smooth
increase with the ratio of the axial Co-X (X = oxide, halide,
chalcogenide) distance and the equatorial Co-O distance. This
is shown in Figure 14.
measured the diffraction pattern of a second sample of Ba
Cu (CFS044A) between 4 and 250 K using the higher resolution
2
CoO2-
S
2 2
diffractometer HRPD at the ISIS facility, which revealed that the
distortion is evident only below about 155 K, somewhat below
the temperature of 210 K at which long-range 3D magnetic ordering
is apparent from the appearance of magnetic Bragg reflections.
The refinements were consistent with an orthorhombic distortion
of the crystal structure, and the splitting of, for example, the 200/
020 reflections was well-resolved using the highest resolution
backscattering detector bank on HRPD. The results of Rietveld
refinement against the HRPD data are shown in Figure 15 and
in the Supporting Information, Table S2. Attempts to establish
whether the symmetry was further reduced from Immm at low
temperatures were hampered by beam heating of the sample on
ID31.
Low-Temperature Structural Distortions. As indicated
above, the measurements of Ba CoO Cu S on D2B at 5 and
2
2
2 2
8
0 K revealed poorly resolved peak splittings in the high-angle
(
∼120° 2θ) region of the diffractogram (Supporting Informa-
tion, Figure S4), suggesting a structural distortion of the room-
temperature tetragonal cell. The D2B data were modeled adequately
using an orthorhombic cell in Immm with a volume similar to that of
the room-temperature tetragonal cell in I4/mmm. We subsequently
2
702
dx.doi.org/10.1021/ja109553u |J. Am. Chem. Soc. 2011, 133, 2691–2705

Journal of the American Chemical Society
ARTICLE
Figure 16. Size of the orthorhombic distortion in (Sr1-xBa
)
x 2
-
Figure 18. Temperature dependence of the lattice parameters of
Sr CoO Cu S (DRP051 not air-exposed) (blue) and Ba CoO Cu S
2 2 2 2 2 2 2 2
(CFS044) (red). The values of the lattice parameters (a, filled symbols;
b, open symbols) have been normalized to their values at 210 K
CoO
2
Cu
2
S
2
samples measured on ID31 (x = 0), D2B (one sample
with x = 1), or HRPD.
(
3.98789(2) Å for Sr
Cu ). The inset shows that the distortion (expressed as (b/a) - 1) is
almost an order of magnitude larger in Ba CoO Cu S than in
2 2 2 2 2 2
CoO Cu S and 4.05860(1) Å for Ba CoO -
2 2
S
2
2
2 2
2 2 2 2
Sr CoO Cu S .
analysis of the low-temperature structure of the exposed material,
although the orthorhombic model remains marginally superior
to the tetragonal model at low temperatures, so it seems that the
air exposure does not quench the structural distortion in this
sample. (Figure S5 in the Supporting Information shows that the
low-temperature broadening of the 114 and 200 reflections is
qualitatively similar to that shown in Figure 17.)
Figure 17. Rietveld refinement against ID31 data at 10 K of a sample of
Gathering together our results on samples of (Sr1-xBax)2-
Sr CoO Cu S (DRP051) that had not been exposed to air. The insets
CoO Cu S across the entire solid solution, we see a fairly
2
2
2
2
2
2 2
show the evolution of the orthorhombic distortion: the tetragonal (200)
splits at low temperatures, while the width of the (114) reflection is
temperature-independent.
smooth and approximately linear variation in the size of the
low-temperature distortion with x for samples that have received
no air exposure (Figure 16). Following our observation of a
similarly smooth variation of the size of the ordered moment with
composition for unexposed samples (Figure 10), we can plot the
size of the distortion, (b/a) - 1, determined from high-resolu-
tion diffraction measurements against the size of the ordered
moment determined from low-temperature neutron diffraction
measurements (Figure 19). Included in this plot are the results
for all the samples of “Sr CoO Cu S ” for which we have
We used HRPD to probe the presence of the low-temperature
structural distortion in the members of the (Sr1-xBa ) Co-
x 2
O Cu S series. For x = 0.9, the 200 and 020 reflections of the
2
2 2
orthorhombic Immm low-temperature cell are clearly resolved.
For smaller values of x, these reflections are not completely
resolved, even in backscattering, although the peaks are broader
than at ambient temperature, and stable refinements were
achieved using the orthorhombic model, showing a linear
decrease in the size of the distortion with decreasing Ba content
2
2
2 2
measurements of the ordered moment and diffraction data at
high enough resolution (ID31 or HRPD) to enable the distortion
to be discerned. Some of these had received some air exposure,
and others had been synthesized at the composition Sr CoO
(Figure 16), with an unusually low value of the distortion in the
2
2-
sample with x = 0.6.
Subsequently we analyzed samples of Sr CoO Cu S using
Cu1 S and contain impurities. For Ba CoO Cu S , we have
.85 2 2 2 2 2
measured the sizes of the distortion (Figure 18) and the ordered
moment (Figure 9) in our HRPD experiment, and these data are
also plotted in Figure 19. The onset of the distortion does not
coincide with the first appearance of magnetic Bragg peaks but is
evident once the long-range ordered moment has grown larger
than about 3.3 μB.
2
2
2 2
high-resolution diffraction data. The two portions of sample
DRP051 (slow-cooled sample, one portion unexposed to air, and
the other exposed to a flow of moist oxygen) were measured on
ID31 between 10 and 290 K. The high-resolution data showed a
clear orthorhombic distortion at 130 K and below in the
unexposed portion of the sample (Figure 17). At 150 K, the
Immm model and the alternative I4/mmm model gave fits that
were of similar statistical quality, although the orthorhombic
model failed to converge, so we conclude that the tetragonal
model is appropriate at 150 K. The variation of the basal lattice
parameters with temperature for Sr CoO Cu S is compared
When there is collinear ordering of magnetic moments that
have an orbital component, the effect of the spin-orbit coupling
is to induce a small cooperative crystallographic distortion at or
below the temperature at which long-range magnetic ordering
47
occurs. We propose that such a magnetostrictive distortion
occurs in (Sr1-xBa ) CoO Cu S . The distortion we observe is
2
2
2 2
x 2
2
2 2
with that of Ba CoO Cu S in Figure 18, where the lattice
evident at temperatures somewhat below T . The moments in
2
2
2 2
N
parameters have been normalized to their values at 210 K for
each compound. As described above (Figure 2), the exposure of
Sr CoO Cu S to moist O results in a broadening of all the
(Sr1-xBa ) CoO Cu S are oriented mainly in the basal plane,
x 2
2
2 2
so an in-plane magnetostrictive distortion is expected. If we
assume that the spin contribution to the magnetic moment of
2
2
2 2
2
2
þ
reflections that is somewhat asymmetric. This hampered the
Co in these systems remains constant, then the fact that the
2
703
dx.doi.org/10.1021/ja109553u |J. Am. Chem. Soc. 2011, 133, 2691–2705

Journal of the American Chemical Society
ARTICLE
2þ
magnetic moment in a range of Co solid-state compounds
correlates strongly with the coordination environment, ranging
from a slightly tetragonally elongated octahedron in La CoO to
2
4
highly elongated, and almost purely square planar, in Ba CoO
2
2-
Cu S (Figure 14). Furthermore, the fact that the ordered magnetic
2
2
moments in (Sr1-xBa ) CoO Cu S have their major component
x 2
2
2 2
oriented in the CoO plane results in a magnetostrictive distortion
2
of the CoO planes that is evident at temperatures below the long-
2
range ordering transition temperature. The size of this distortion
correlates well with the size of the ordered moment, and in
Sr CoO Cu S the distortion is close to the resolution limit of even
2
2
2 2
the highest resolution powder diffractometers (Figure 19). A final
important observation is that Sr CoO Cu S is rather unobtrusively
Figure 19. Dependence of the long-range ordered magnetic moment
on the size of the orthorhombic distortion for samples (Sr1-xBax)2-
CoO Cu S . The color coding indicates the diffractometer used for the
2
2
2 2
air-sensitive. We have shown that this compound is subject to
oxidative de-intercalation of copper through aerial oxidation, with
the formation of a 5% copper deficiency and poorly crystalline CuO
coexisting with the oxidized oxysulfide phase. This oxidation has a
significant effect on the size of the ordered magnetic moment, which
2
2 2
determination of the magnetic moment (black, D2B; blue, HRPD;
red, OSIRIS) as in Figure 10. Green symbols denote the variable-
temperature data on Ba CoO Cu S (HRPD), and orange symbols
2
2
2 2
indicate samples of nominal composition Sr
which had received exposure to moist air. The sample of (Sr1-
Ba CoO Cu with x = 0.6, which is an apparent outlier in Figures 10
and 16, is ringed.
2
CoO
2
Cu1.85
S
2
, some of
3þ
may be ascribed, at least in part, to the formation of Co ions that
2þ
carry a lower moment than Co ions. Whether partial frustration of
x
)
x 2
2
2
S
2
2þ
3þ
long-range magnetic order by Co /Co disorder or itinerant
behavior of some of the Co 3d electrons accompanies oxidation
remains to be investigated. In the absence of statements to the
orbital contribution to the long-range ordered moment increases
with increasing Ba content is consistent with the observation that
the size of the magnetostrictive distortion also increases with
increasing Ba content. Further evidence that the distortion is
driven by the magnetic ordering lies in the sample with x = 0.6.
This is an outlier in the plots of moment versus composition
contrary, we assume that the physical measurements that noted
12,13
unusual behavior of Sr CoO Cu S
were made on samples
2
2
2 2
exposed to air, which may be closer to Sr CoO Cu S or may have
2
2
1.9 2
changed their composition during their examination; we anticipate
that some re-evaluation of the previously reported physical proper-
ties of this compound may be required, not least because the partial
oxidative de-intercalation of copper evidently has a significant effect
on the magnetic susceptibility and on the size of the long-range
ordered moment, as we describe. The changes in other physical
properties, such as conductivity and thermopower, with composi-
tion may also be very significant.
(
Figure 10) and distortion versus composition (Figure 16).
However, this point (ringed in Figure 19) lies on the trend of
ordered moment against distortion exhibited by the other
samples. A distortion of magnitude similar to that in Ba CoO
2
2-
Cu S occurs below the charge-ordering Verwey transition in
2
2
2
þ
Fe O , in which the localized Fe ions carry an orbital magnetic
3
4
4
8
moment. In CoO there is a distortion below T that has as its
N
’
ASSOCIATED CONTENT
main component a tetragonal compression of the octahedron by
about 1.2%. A further smaller component of the distortion
S
Supporting Information. Further tables of refinement
38
b
renders the symmetry monoclinic. The largest orthorhombic
parameters, diffractograms showing Rietveld refinements, crys-
tallographic information files, magnetic susceptibility data, and
computational output. This material is available free of charge via
the Internet at http://pubs.acs.org.
distortion that we observe below TN in the (Sr1-xBa ) Co-
x 2
O Cu S series is similar in size to the smaller (trigonal)
2
2 2
component of the distortion in CoO.
’
CONCLUSIONS
’
AUTHOR INFORMATION
The series of compounds (Sr1-xBa ) CoO Cu S are rare
x 2
2
2 2
2
þ
examples of systems with high-spin Co ions in a coordination
environment that is a highly distorted CoO S octahedron, so
the environment for Co is close to being purely square planar.
The systems order antiferromagnetically with the moments
mainly oriented in the basal plane, as predicted by computational
studies. The size of the long-range ordered magnetic moment
Corresponding Author
simon.clarke@chem.ox.ac.uk
4
2
2
þ
’
ACKNOWLEDGMENT
We thank the UK Engineering and Physical Sciences Research
2
þ
Council for funding under grant EP/E025447 and for the award
of a studentship to C.F.S. We are grateful to Dr. R. I. Smith, Dr. K.
Knight, Dr. R. M. Ibberson, Dr. M. Telling (ISIS facility), Dr. A.
Hewat, and Dr. E. Suard (ILL, Grenoble) for assistance with the
PND measurements, and Dr. A. N. Fitch and Dr. C. Curfs for
assistance with measurements on ID31.
carried by the Co ion increases as the CoO S polyhedron
4
2
becomes increasingly tetragonally elongated with increasing x
increasing Ba content). Computational analysis of the end
member compounds (x = 0 and 1) and consideration of the
(
2
þ
calculated 3d manifold in molecular Co systems show that this
behavior is consistent with the decreasing separation with
increasing x of unequally occupied 3d levels that are coupled
by the spin-orbit interaction. Thus, the orbital component of
the moment carried by the Co ion increases as the Ba content
increases, and the Co environment becomes more tetragonally
elongated. The size of the low-temperature long-range ordered
’ REFERENCES
2
þ
(
1) Zhu, W. J.; Hor, P. H. J. Solid State Chem. 1997, 130, 319.
2
þ
(2) Zhu, W. J.; Hor, P. H.; Jacobson, A. J.; Crisci, G.; Albright, T. A.;
Wang, S.-H.; Vogt, T. J. Am. Chem. Soc. 1997, 119, 12398.
2
704
dx.doi.org/10.1021/ja109553u |J. Am. Chem. Soc. 2011, 133, 2691–2705

Journal of the American Chemical Society
ARTICLE
ꢀ
(
3) Otzschi, K.; Ogino, H.; Shimoyama, J.; Kishio, K. J. Low Temp.
(41) Hansteen, O. H.; Fjellvag, H.; Hauback, B. C. J. Solid State
Chem. 1998, 141, 41.
Phys. 1999, 117, 729.
ꢀ
(
(
4) Vajenine, G. V.; Hoffmann, R. Inorg. Chem. 1996, 35, 451.
5) Ueda, K.; Hirose, S.; Kawazoe, H.; Hosono, H. Chem. Mater.
(42) Hansteen, O. H.; Fjellvag, H.; Hauback, B. C. J. Mater. Chem.
1998, 8, 2089.
2
001, 13, 1880.
6) G ꢀa l, Z. A.; Rutt, O. J.; Smura, C. F.; Overton, T. P.; Barrier, N.;
Clarke, S. J.; Hadermann, J. J. Am. Chem. Soc. 2006, 128, 8530.
7) Mayer, J. M.; Schneemeyer, L. F.; Siegrist, T.; Waszczak, J. V.;
Van Dover, B. Angew. Chem., Int. Ed. Engl. 1992, 31, 1645.
8) Wang, C; Tan, M.; Feng, C; Ma, Z.; Jiang, S; Xu, Z.; Cao, G.;
Matsubayashi, K.; Uwatoko, Y. J. Am. Chem. Soc. 2010, 132, 7069.
9) Fuwa, Y.; Wakeshima, M.; Hinatsu, Y. Solid State Commun. 2010,
50, 1698.
10) Matoba, M.; Takeuchi, T.; Okada, S.; Kamihara, Y.; Itoh, M.;
Ohoyama, K.; Yamaguchi, Y. Physica B 2002, 312-3, 630.
11) Okada, S.; Matoba, M.; Yoshida, H.; Ohoyama, K.; Yamaguchi,
Y. J. Phys. Chem. Solids 2002, 63, 983.
12) Okada, S.; Terasaki, I.; Ooyama, H.; Matoba, M. J. Appl. Phys.
004, 95, 6816.
13) Okada, S.; Matoba, M.; Fukumoto, S.; Soyano, S.; Kamihara, Y.;
Takeuchi, T.; Yoshida, H.; Ohoyama, K.; Yamaguchi, Y. J. Appl. Phys.
002, 91, 8861.
14) Larson, A.; von Dreele, R. B. The General Structure Analysis
System; Los Alamos National Laboratory: Los Alamos, NM, 1985.
15) Prince, E., Ed. International Tables for Crystallography, Volume
(43) Ceulemans, A.; Dendooven, M.; Vanquickenborne, L. G. Inorg.
Chem. 1985, 24, 1159.
(44) Demazeau, G.; Buffat, B.; Pouchard, M.; Hagenmuller, P.
(
(
J. Solid State Chem. 1984, 54, 389.
(45) Shannon, R. D. Acta Crystallogr. A 1976, 32, 751.
(46) Xiang, H. J.; Wei, S.-H.; Whangbo, M.-H. Phys. Rev. Lett. 2008,
100, 167207.
(47) Goodenough, J. B. Magnetism and the Chemical Bond; John
Wiley & Sons: New York, 1963.
(
(
1
(
(48) Wright, J. P.; Attfield, J. P.; Radaelli, P. G. Phys. Rev. B 2002, 66,
214422.
(
(
2
(
2
(
(
C: Mathematical, physical and chemical tables; International Union of
Crystallography; Wiley: West Sussex, UK, 2004.
(
16) Otwinowski, Z.; Minor, W. In Methods in Enzymology, Macro-
molecular Crystallography, Part A; Carter, C. W., Jr., Sweets, R. M., Eds.;
Academic Press: New York, 1997; Vol.276.
(17) Farrugia, L. J. J. Appl. Crystallogr. 1999, 32, 837.
(18) Sheldrick, G. M. SHELX97, Programs for Crystal Structure
Analysis, Release 97-2; University of G €o ttingen: G €o ttingen, Germany,
1997.
(
(
(
(
(
(
(
19) Spek, A. L. Acta Crystallogr. A 1990, 46, C34.
20) Bl €o chl, P. E. Phys. Rev. B 1994, 50, 17953.
21) Hohenberg, P.; Kohn, W. Phys. Rev. 1964, 136, 864.
22) Kohn, W.; Sham, L. J. Phys. Rev. 1965, 140, 1133.
23) Kresse, G.; Furthm €u ller, J. Comput. Mater. Sci. 1996, 6, 15.
24) Kresse, G.; Joubert, D. Phys. Rev. B 1999, 59, 1758.
25) Perdew, J. P.; Burke, K.; Ernzerhof, M. Phys. Rev. Lett. 1996, 77,
3
1
865.
(
26) Perdew, J. P.; Burke, K.; Ernzerhof, M. Phys. Rev. Lett. 1997, 78,
396.
(
27) Dudarev, S. L.; Botton, G. A.; Savrasov, S. Y.; Humphreys, C. J.;
Sutton, A. P. Phys. Rev. B 1998, 57, 1505.
(28) Smura C. F. D.Phil. Thesis, University of Oxford, 2008.
(29) Pitcher, M. J.; Smura, C. F.; Clarke, S. J. Inorg. Chem. 2009, 48, 9054.
(30) Palazzi, M. C.R. Hebd. Seances Acad. Sci., Ser. C, Sci. Chim. 1981,
292, 7899.
(
(
(
31) Berger, R. J. Less-Common Met. 1989, 147, 141.
32) Brown, I. D.; Altermatt, D. Acta Crystallogr. B 1985, 41, 244.
33) Brown, I. D. , http://www.ccp14.ac.uk/ccp/web-mirrors/
i_d_brown/bond_valence_param/, 2006.
34) Lee, C. H.; Iyo, A.; Eisaki, H.; Kito, H.; Fernandez-Diaz, M. T.;
(
Ito, T.; Kihou, K.; Matsuhata, H.; Braden, M.; Yamada, K. J. Phys. Soc.
Jpn. 2008, 77, 083704.
(
35) Yamada, K.; Matsuda, M.; Endoh, Y.; Keimer, B.; Birgeneau,
R. J.; Onodera, S.; Mizusaki, J.; Matsuura, T.; Shirane, G. Phys. Rev. B
989, 39, 2336.
36) Figgis, B. N.; Hinchman, M. A. Ligand Field Theory and its
Applications; Wiley-VCH: New York, 2000.
1
(
(37) Knee, C. S.; Weller, M. T. Phys. Rev. B 2004, 70, 144406.
(38) Jauch, W.; Reehuis, M.; Bleif, H. J.; Kubanek, F.; Pattison, P.
Phys. Rev. B 2001, 64, 052102.
(
39) Jauch, W.; Reehuis, M. Phys. Rev. B 2002, 65, 125111.
(40) Hubbard, J.; Marshall, W. Proc. Phys. Soc. (London) 1965,
86, 561.
2
705
dx.doi.org/10.1021/ja109553u |J. Am. Chem. Soc. 2011, 133, 2691–2705