Resolving composition and structure of RE-Sb-O-C natural superlattice phases (RE = La, Ho)

Article abstract of DOI:10.1002/ejic.201100313

A family of rare earth antimonide oxycarbides have been prepared and structurally characterized. These superlattice phases are constructed from NaCl-type RESb slabs sandwiched between RE-O-C layers. Depending on the carbon content and synthetic conditions, three different RE-Sb-O-C structures can be obtained. At lower temperatures,RE_9-δSb ₅(O,C)₅ phases are obtained for RE = La, Ho. These phases adopt a stuffed Sc₂Sb-type structure with P4/nmm symmetry. An O/C mixture, in which the O/C ratio is larger than 4:1, is randomly distributed within the RE-O-C layers. The RE atoms are highly disordered within the oxide layer. At temperatures above the melting point of the samples, RE ₉Sb₅O₄C phases with P4/n symmetry are produced. The RE-O-C layers in RE₉Sb₅O₄ are fully ordered; the RE sites are well defined, and the O and C atoms occupy the tetrahedral and square-pyramidal voids, respectively. At high temperatures, a new ordered La₁₄Sb₈O₇C structure with P4bm symmetry was discovered. The La₁₄Sb₈O₇C phase is structurally similar to RE₉Sb₅O₄C and features orderedarrangements of La and O/C atoms in the La-O-C layer. The RE_9-δSb₅(O,C)₅, RE ₉Sb₅O₄C and La₁₄Sb₈O ₇C phases appear to be charge-balanced, and their compositions and structures are controlled by the O/C ratio. Parallel preparative experiments revealed the importance of carbon in the formation of these layered phases. In addition, it has been established that the purity of the rare earth metals influences the compositions and structures of the products.

Full text of DOI:10.1002/ejic.201100313

FULL PAPER
DOI: 10.1002/ejic.201100313
Resolving Composition and Structure of RE–Sb–O–C Natural Superlattice
Phases (RE = La, Ho)
Peng Wang,[a] Scott Forbes,[a] Volodymyr Svitlyk,[b] Ashur Aushana,[c] and
Yurij Mozharivskyj*[a]
Dedicated to Professor John D. Corbett on the occasion of his 85th birthday
Keywords: Solid-state structures / Solid-phase synthesis / Rare earths
A family of rare earth antimonide oxycarbides have been
prepared and structurally characterized. These superlattice
are well defined, and the O and C atoms occupy the tetra-
hedral and square-pyramidal voids, respectively. At high
phases are constructed from NaCl-type RESb slabs sand- temperatures, a new ordered La14Sb8O7C structure with
wiched between RE–O–C layers. Depending on the carbon
content and synthetic conditions, three different RE–Sb–O–C
structures can be obtained. At lower temperatures,
RE9–δSb5(O,C)5 phases are obtained for RE = La, Ho. These
phases adopt a stuffed Sc2Sb-type structure with P4/nmm
symmetry. An O/C mixture, in which the O/C ratio is larger
than 4:1, is randomly distributed within the RE–O–C layers.
The RE atoms are highly disordered within the oxide layer.
At temperatures above the melting point of the samples,
RE9Sb5O4C phases with P4/n symmetry are produced. The
RE–O–C layers in RE9Sb5O4 are fully ordered; the RE sites
P4bm symmetry was discovered. The La14Sb8O7C phase is
structurally similar to RE9Sb5O4C and features ordered
arrangements of La and O/C atoms in the La–O–C layer. The
RE9–δSb5(O,C)5, RE9Sb5O4C and La14Sb8O7C phases appear
to be charge-balanced, and their compositions and structures
are controlled by the O/C ratio. Parallel preparative experi-
ments revealed the importance of carbon in the formation of
these layered phases. In addition, it has been established that
the purity of the rare earth metals influences the composi-
tions and structures of the products.
2–
and the anionic Zn2Sb2 layer. Thermoelectric perform-
Introduction
ance is improved as the disordered cationic layer functions
as a phonon barrier to reduce the thermal conductivity,
whereas the electrical properties of the anion layer remain
unperturbed.
Recently, a new approach known as the natural superlat-
tice approach[1] has been employed in the design and prepa-
ration of new materials for thermoelectric applications. An
ideal thermoelectric material must have a large Seebeck co-
efficient α and a low thermal conductivity κ, while retaining
a high electrical conductivity δ.[2] These desired physical
properties, however, are rarely found simultaneously in a
single homogeneous material. By constructing a natural su-
perlattice, a material with desired electrical properties is
combined with a thermally insulating material in such a
way that both of their basic structural building blocks are
preserved and interwoven into a single structure.[3] Such
structural design has been realized in the CaxYb1–xZn2Sb2
phases.[4] The structure of the CaxYb1–xZn2Sb2 system con-
sists of two alternating layers: the cationic Ca2+/Yb2+ layer
Inspired by the natural superlattice approach, we have
undertaken systematic structural and compositional analy-
ses of the partially oxidized rare earth antimonide oxides
RE9Sb5O5 (RE = La, Ho).[5–7] Previous works[5–7] have re-
vealed some interesting structural features of these com-
pounds. As shown in Figure 1, their structure can be under-
stood as a variant of the Sc2Sb structure.[6] The unit cell of
Sc2Sb consists of a layer of Sc atoms sandwiched between
slightly puckered NaCl-type ScSb slabs. In the case of the
RE9Sb5O5 phases, the unit cells consist of [RE4O5] layers
and NaCl-type [RE5Sb5] slabs. The oxide layer has four RE
atoms occupying four out of five metal positions present in
the Sc2Sb structure, and the oxygen atoms fill the interstitial
sites.[5]
[a] Department of Chemistry and Chemical Biology, McMaster
University,
1280 Main Street West, Hamilton, Ontario L8S 4M1, Canada
[b] European Synchrotron Radiation Facility,
6 rue Jules Horowitz, P. B. 220, 38043 Grenoble, France
[c] Department of Chemistry, York University,
4700 Keele Street, Toronto, Ontario M3J 1P3, Canada
Supporting information for this article is available on the
WWW under http://dx.doi.org/10.1002/ejic.201100313.
The [RE5Sb5] slabs in the RE9Sb5O5 system are expected
to retain the electrical properties of the RESb binaries as
they share the same structure. The pristine RESb binaries
are semimetallic/metallic materials[8] containing some of the
heaviest elements in the periodic table. These binary phases,
Eur. J. Inorg. Chem. 2011, 3887–3895
© 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
3887

P. Wang, S. Forbes, V. Svitlyk, A. Aushana, Y. Mozharivskyj
FULL PAPER
phases related to the RE9Sb5O5/RE9Sb5O4C (RE = La, Ho)
phases. The early rare earth element La was chosen for this
investigation to establish a connection to previous studies.
Ho was selected to expand the series and to allow a com-
parison of the early and late rare earth systems. The aim of
this work was also to uncover the significance of carbon on
the structural, chemical, and physical properties of these
phases.
Results and Discussion
The experimental results revealed interesting composi-
tional and structural features as well as relationships be-
tween the various phases. Percentage yield (in wt.-%) of the
desired phase in the bulk samples is plotted against the
loaded amount of carbon in Figure 2. The products are
identified by their crystal structures. The different symbols
represent the purity of the La and Ho metals and annealing
conditions (table in Figure 2). For x = 0 (no carbon added),
the samples were prepared from commercial and purified
La and Ho. The graph in Figure 2 reveals two trends:
(1) the yield of the product depends on the carbon content
in the starting material; and (2) the synthetic conditions de-
termine the structure of the products. To explain the ob-
served trends, the crystal structure and composition of each
phase need to be discussed.
Figure 1. Structures of Sc2Sb and RE9Sb5O5.
adopting a simple NaCl-type crystal structure, are expected
to possess a relatively high thermal conductivity due to the
small number of atoms in their unit cells.[9,10] The [RE4O5]
layer sandwiched between two [RE5Sb5] slabs adopts a com-
pletely different structure and is expected to perturb
phonon propagation by increasing the structural complex-
ity.
It was also noticed that there are two dissimilar oxygen
positions within the RE4O5 layer. One of the sites (8g) is
surrounded by four neighboring rare earth elements in a
tetrahedral arrangement, whereas the other (2c) has a
rather unusual square-pyramidal arrangement surrounded
by five RE atoms. With oxygen atoms fully occupying
the two sites, the electron-rich formula RE9Sb5O5
[RE3+9Sb3–5O2–5(2e–)] results. An electron-balanced for-
mula RE9Sb5O4C (RE3+9Sb3–5O2–4C4–), in which C replaces
oxygen in the 2c site, was also suggested;[6] however, the
RE9Sb5O4C
According to the single-crystal X-ray diffraction analysis,
electrical resistivity measurements performed on the bulk the La9Sb5O4C and Ho9Sb5O4C phases adopt an
samples revealed a metallic-type behavior, thus supporting La9Sb5O5-type structure with the P4/n space group. As dis-
the electron-rich model.
cussed in the Introduction, in the La9Sb5O5 structure, oxy-
The metallic-type conductivity implies that the gen atoms occupy both the tetrahedral (8g) and square-py-
RE9Sb5O5 phases would possess a low thermopower α, un- ramidal (2c) interstitial sites. The La9Sb5O5 model was used
attractive for industrial applications. Nonetheless, the ob- to refine the La9Sb5O4C and Ho9Sb5O4C structures; how-
served electrical properties may also result from a partial ever, it was observed that the thermal parameters of the
substitution of C in the 2c site by O (if the 2c site is assumed oxygen atoms in the 2c site were abnormally large in com-
to be primarily occupied by C). The refinement results for parison with those of the 8g oxygen atom. A modified
the RE9Sb5O5 single crystals[6] yielded large thermal vi- model (Figure 3) with carbon instead of oxygen atoms at
brations for the 2c oxygen atoms, suggesting the presence the 2c site was tried. In the new model, the thermal vi-
of a lighter element at this site. The existence of pure bration parameters for the 2c site were significantly im-
RE9Sb5O5 may also be questionable considering the exis- proved (Table 1), whereas other atomic parameters re-
tence of the semiconducting RE3SbO3 and RE8Sb3–δO8 mained unaffected. As a result, the compositions of the La
antimonide oxides, in which oxygen atoms have a typical and Ho phases were assumed to be La9Sb5O4C and
tetrahedral RE environment.[11] The RE3SbO3 phases are Ho9Sb5O4C, respectively.
produced by direct high-temperature fusion of the RESb
and RE2O3 binaries, whereas RE8Sb3–δO8 can be derived port carbon rather than oxygen in the square-pyramidal in-
from RE3SbO3 through the loss of RE and oxygen. terstitial 2c site defined by five RE atoms. For example, in
Considering the above-discussed factors, it is likely that the rare earth dicarbides REC2
In addition to the refinement results, other factors sup-
[12,13]
with the CaC2 struc-
“RE9Sb5O5” is stabilized by light elements other than oxy- ture, the carbon atoms have five RE atoms in their first
gen. If a C atom can replace the O atom in the 2c site in coordination sphere, similar to the 2c site environment in
“RE9Sb5O5”, the electrical properties could be adjusted to RE9Sb5O4C. It has to be emphasized that in REC2, the
render these phases semiconducting and in favor of good square-pyramidal RE5C units share square faces, which re-
thermoelectric performance. This work focuses on the syn- sults in short C–C bonds, but no such C–C bonds are pres-
thesis, compositions, and structural characterization of ent in RE9Sb5O4C. In the RE2O3 oxides, oxygen atoms re-
3888
www.eurjic.org
© 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Eur. J. Inorg. Chem. 2011, 3887–3895

Resolving Composition and Structure of Natural Superlattice Phases
Figure 2. Yield of the desired phases in each sample plotted against the amount of added carbon.
Figure 3. Crystal structures of La14Sb8O7C, RE9Sb5O4C, RE9–δSb5(O,C)5 and their building blocks.
Eur. J. Inorg. Chem. 2011, 3887–3895
© 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.eurjic.org
3889

P. Wang, S. Forbes, V. Svitlyk, A. Aushana, Y. Mozharivskyj
FULL PAPER
Table 1. Thermal parameters of the O/C atoms in the 2c square-
pyramidal sites and resulting R1 values in the RE9Sb5O5/
RE9Sb5O4C models.
The RE deficiencies were always present irrespective of
the model chosen, and thus the general formula for these
phases is given as RE9–δSb5(O,C)5 [the refined compositions
were La8.80(1)Sb5(O,C)5 and Ho8.84(4)Sb5(O,C)5]. During the
refinement, the atomic site occupancy of the RE(2) atom
and the thermal vibration parameters of the C/O atoms in
the RE–O–C layers were correlated and, as a result, the RE
occupation and O/C ratio could not be refined simulta-
neously. Another interesting outcome is that the O/C ratio
in RE9–δSb5(O,C)5 is larger than in RE9Sb5O4C and likely
leads to the RE deficiencies as oxygen is divalent and car-
bon is tetravalent. The larger O/C ratio also follows from
the synthesis of the Ho9Sb5O4.5C0.5 sample, which yielded
the RE9–δSb5(O,C)5 phase with high purity (Figure 2, x =
0.5, open diamonds).
Refined composition
Atom on 2c site
Ueq
R1 value
La9Sb5O5
La9Sb5O4C
Ho9Sb5O5
Ho9Sb5O4C
O
C
O
C
0.024(5)
0.011(4)
0.050(8)
0.016(6)
0.0531
0.0531
0.0264
0.0263
side inside the RE tetrahedra,[14] and this environment is
analogous to that of the 8g oxygen site in RE9Sb5O4C.
The bond valence sum method[15] has been employed to
test site preferences in the La9Sb5O4C and Ho9Sb5O4C. The
Ho–C R0 parameter was derived from the bond lengths and
coordination geometry of Ho2C,[16] in which carbon is as-
sumed to be tetravalent. Unfortunately, the R0 parameter
for the La–C bond could not be extracted as no comparable
compound exists. The R0 parameters for the La–O and Ho–
O bonds were taken from the work by Brese et al.[17] The
calculated bond valence (B-V) sums for both oxygen and
carbon in the 2c site are listed in Table 2. The results sug-
gest the presence of a carbon atom in the 2c site, in agree-
ment with the X-ray structural analysis results.
Since the RE(2) disorder might be indicative of a super-
structure, we decided to check for superstructure reflections.
Analysis of the reciprocal space of the room-temperature
diffraction data both for La8.80(1)Sb5(O,C)5 and Ho8.84(4)
-
Sb5(O,C)5 revealed no signs of a superstructure. At 100 K,
however, streaks of weak diffused spots were observed
along the c-direction in the reciprocal space of La8.80(1)
-
Sb5(O,C)5 (see the Supporting Information). Such diffused
scattering can imply some degree of ordering on the 8i site.
From a structural perspective, the RE(2) disorder on the 8i
site allows the RE environment around the O/C atoms to
change from square-pyramidal into distorted tetrahedral
(Figure 4). Such a change may be preferred if a particular
site is occupied by oxygen. On the other hand, absence of
the distortion, that is, retention of the RE square-pyramidal
environment, can be associated with the presence of a car-
bon atom. It is difficult to establish what factors cause the
RE(2) disorder: the absence of the O/C ordering, the O/C
ratio, which is incommensurate with the atomic ordering
and prevents it, or other parameters.
Table 2. Bond valence sum values for atoms in the 2c site.
Atom in 2c
Expected B-V
sum
Calculated B-V
sum
R0 value
site
La9Sb5O4C
O
C
2.172 Å
–
2
4
1.693
–
Ho9Sb5O4C
O
C
2.023 Å
2.318 Å
2
4
1.801
3.997
RE9–δSb5(O,C)5
Similar to the RE9Sb5O4C structure, the RE9–δSb5-
(O,C)5 structure also consists of an RE–O–C layer sand-
wiched between NaCl-type RESb slabs. Initially, the
RE9–δSb5(O,C)5 crystals were solved in the Sc2Sb structure,
with full RE occupancies and O/C atoms filling the inter-
stitial sites between the RESb slabs. During the refinement,
the RE(2) metal atoms in the 2a position (¾,¼,0) in the
RE–O–C layer exhibited large disk-shaped thermal ellip-
soids in the ab-plane. Structural refinement of the low-tem-
perature (100 K) data of the La9–δSb5(O,C)5 crystal sug-
gested two alternative models, with the RE(2) atoms having
either an 8i or 8g position. The new models yielded im-
Figure 4. Structural models of RE9–δSb5(O,C)5 with RE(2) atoms
in the 2a and 8i sites. The possible metal environments around the
proved RE(2) thermal parameters and lower R values both oxygen interstitial sites are identified with the polyhedra.
for La9–δSb5(O,C)5 and Ho9–δSb5(O,C)5. The structural
model with the RE(2) atoms in the 8i site yielded the lowest
R values and reasonable thermal parameters for both the
RE(2) and interstitial O/C(1) atoms; therefore, it is believed
La14Sb8O7C
to be the best representation of the overall structure (com-
parison of the different models is given in the Supporting LaSb slabs and La–O–C layers, but it possesses a larger unit
Information). cell with P4bm symmetry. The LaSb slabs of La14Sb8O7C
The La14Sb8O7C structure also consists of NaCl-type
3890
www.eurjic.org
© 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Eur. J. Inorg. Chem. 2011, 3887–3895

Resolving Composition and Structure of Natural Superlattice Phases
are similar to those of La9Sb5O4C, but there are four rather loss is offset by the equiatomic oxygen gain. Due to the
than two C/O sites in the La–O–C layer. The O1 and O2 different coordination requirements of oxygen and carbon
sites have tetrahedral environments typical for oxygen atoms, the ordered atomic arrangement within the RE–O–
atoms in this family of compounds; therefore, these sites are C layer can be achieved only for specific RE and O/C stoi-
assumed to be fully occupied by oxygen atoms. The carbon chiometries. Whereas this condition is met in RE9Sb5O4C,
site has a square-pyramidal La coordination, which is sim- it is apparently not satisfied in RE9–δSb5(O,C)5. As a result,
ilar to that observed in RE9Sb5O4C. The single-crystal the RE–O–C layer of RE9–δSb5(O,C)5 contains both the RE
structural refinement performed against all collected reflec- disorder and the O/C statistical mixtures. On the local level,
tions yielded a smaller temperature factor on the carbon both of these features provide suitable environments for the
site in comparison with that of the tetrahedrally coordi- oxygen and carbon atoms. Although we could not pinpoint
nated oxygen atoms. The thermal vibration of the carbon the boundaries for the O/C ratio, based on the structural
atom, however, became normal when the structural model analysis and synthetic results, it appears that ratios 4:1 Ͻ
was refined against high-angle reflections (2θ Ͼ 30°). O/C Յ 9:1 are possible for RE9–δSb5(O,C)5.
Whereas some oxygen atoms can be present, based on the
The O/C ratio of 7:1 is of particular interest as it yields
structural analogy to La9Sb5O4C, we believe that this site the La14Sb8O7C structure. As expected from the O/C ratio,
is predominantly occupied by carbon atoms. There is also the La–O–C layer is depleted in La (compared with La9Sb5-
an additional oxygen site (O3), which has only three neigh- O4C) to maintain the charge balance. Interestingly, the
boring La atoms in the trigonal-planar arrangement. A La–O–C layer with the La6O7C composition now finds an
similar oxygen environment has also been observed in the atomic arrangement, which provides both the La ordering
RE8Sb3–δO8 structures.[11]
as well as proper environments for the oxygen and carbon
atoms. When compared with RE9Sb5O4C, the La layer in
La14Sb8O7C has larger voids that can be viewed as two
square voids from RE9Sb5O4C joined together. Since the
La14Sb8O7C phase could not be prepared in bulk, and a
Structural and Compositional Relationships
The common building blocks of the RE9–δSb5(O,C)5, similar Ho14Sb8O7C could not be verified to exist, it can be
RE9Sb5O4C, and La14Sb8O7C structures are NaCl-type assumed that even small deviations from the 7:1 O/C ratio
RESb slabs and RE–O–C layers. Although the NaCl-type results in the formation of the RE9–δSb5(O,C)5-type phases.
RESb slabs remain identical among the three structures, the
RE–O–C layers are different. The arrangement and order-
ing of the RE and O/C atoms within these layers appear to
Role of Temperature
dictate the unit cell size as well as its symmetry. The
For the same loading composition, the structure of the
RE9Sb5O4C structure can be viewed as a parent structure resulting phase depends on the annealing temperature. If
in this series. As discussed in the Introduction, this struc- the temperature is high enough to melt the samples, the
ture itself is a derivative of the Sc2Sb structure in which resulting structure is La14Sb8O7C (P4bm) or RE9Sb5O4C
the RE layer, sandwiched between RESb slabs, is deficient. (P4/n) with fully ordered RE–O–C layers. If the temperature
Another difference to the Sc2Sb structure is that the tetra- is below the sample’s melting point, however, the resulting
hedral and square-pyramidal voids between the RE layer structure is RE9–δSb5(O,C)5 (P4/nmm) with disordered RE–
and the RESb slabs in RE9Sb5O4C are filled with oxygen O–C layers. This trend can be clearly seen both for the
and carbon atoms, respectively. A more detailed description La9Sb5O4C and Ho9Sb5O4C samples prepared from the dis-
of the Sc2Sb/RE9Sb5O4C relationship can be found in the tilled metals (Figure 2). If the La9Sb5O4C sample is an-
work by Nuss et al.[6] One interesting point is that the stoi- nealed at 1400 °C, which is below its melting point, the
chiometric RE9Sb5O4C phases are charge-balanced. This major phase is La9–δSb5(O,C)5 (P4/nmm) (76 wt.-%, Fig-
fact may explain the “deficiency” of the RE layer, as com- ure 2, full sircles). But the same sample heated to 1550 °C,
pared with the Sc2Sb structure, and the overall phase com- above its melting point, contains more than 90 wt.-% of the
position. The NaCl-type (RE+3)5(Sb–3)5 slab (composition La9Sb5O4C (P4/n) phase (Figure 2, x = 1, open circles)
is per formula unit) is charge-balanced, so the (O–2)4C–4 without any traces of the La9–δSb5(O,C)5 phase. The
composition of the RE–O–C layer requires the presence of Ho9Sb5O4C sample could not be melted at 1650 °C, which
only four RE atoms to maintain the charge neutrality. This is the highest temperature currently achievable with our ex-
reduces the overall RE/Sb ratio from 2:1 expected for the perimental setup; at this temperature, the major phase was
Sc2Sb-type structures to 1.8:1. The arrangement of the RE Ho9–δSb5(O,C)5 (P4/nmm) (ca. 88 wt.-%, Figure 2, x = 1,
atoms is such that the square voids are created in the RE– open diamonds). Melting of the Ho9Sb5O4C sample was
O–C layers and, in addition, two types of interstitial voids, achieved by adding a 0.3 molar fraction of Ho5Sb3, which
tetrahedral and square-pyramidal, are formed with the acted as a low-temperature flux. After melting the sample,
neighboring RESb slabs.
the major phase was Ho9Sb5O4C (P4/n, 85 wt.-%, Figure 2,
The RE9–δSb5(O,C)5 phases can be considered as deriva- full diamonds). These experiments suggest that there is a
tives of RE9Sb5O4C, in which the carbon content is re- large energy barrier for the atomic ordering within the RE–
duced. Such a reduction would lead to a decrease in the RE O–C layers, and a full ordering is achieved only from the
concentration within the RE–O–C layer, even if the carbon liquid state.
Eur. J. Inorg. Chem. 2011, 3887–3895
© 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.eurjic.org
3891

P. Wang, S. Forbes, V. Svitlyk, A. Aushana, Y. Mozharivskyj
FULL PAPER
Impact of Carbon and Purity of RE Elements
La14Sb8O7C (RE = La, Ho) were discovered. All belong to
a family of natural superlattice phases based on the in-
tergrowth of NaCl-type RESb slabs and RE–O–C layers.
The pivotal role of carbon in the formation of these phases
has been clearly demonstrated. It has also been shown that
the purity of rare earth elements dictates the outcome of
the synthesis.
Although the O/C ratio and the positions of the oxygen
and carbon atoms could be unambiguously established in
the RE9Sb5O4C phases from the single-crystal X-ray and
bond valence analyses, the exact O/C ratio could not be
reliably refined for the other two phases. Still, it appears
that all phases tend to be charge-balanced and the O/C ra-
tio dictates both the RE content and the atomic arrange-
ment within the RE–O–C layers. The most carbon-rich
phases are RE9Sb5O4C, and further carbon incorporation
was unsuccessful. Lowering the carbon content results in
RE9–δSb5(O,C)5 phases with an O/C ratio larger than 4:1,
and in La14Sb8O7C with an O/C ratio of 7:1.
One can argue that the evidence from the structural re-
finement may not be sufficient to fully support the
RE9Sb5O4C composition. In addition, the X-ray diffraction
studies on the RE9–δSb5(O,C)5 single crystals could not
prove the presence of carbon atoms. A series of preparative
experiments was carried out to reveal the role of carbon
and purity of the RE elements on the formation of the
RE9Sb5O4C and RE9–δSb5(O,C)5 phases.
Commercial-grade late rare earth metals (Gd–Lu) usu-
ally contain light-atom impurities including oxygen (ca.
1 atom-%), nitrogen (ca. 0.5 atom-%), hydrogen (ca.
3 atom-%) and carbon (ca. 1 atom-%).[18,19] Although no
comparable literature data could be found for the light lan-
thanides, it is possible that the level of impurities is some-
what higher due to their stronger affinity to the light p-
elements. Samples with a loading composition of La9Sb5O5
(no added carbon) were prepared from both commercial-
grade (ca. 93 atom-%) and purified La metals (Ames Labo-
ratory, 99.9 atom-%). According to the X-ray powder analy-
sis, La9Sb5O4C was the major phase (ca. 75 wt.-%, Figure 2,
full triangles) in the sample made from the commercial La
metal, whereas its amount in the sample with purified La
metal was less than 14 wt.-% (Figure 2, x = 0, open circles).
Similar results were obtained for the Ho9Sb5O5 samples:
the commercial Ho yielded 58% (Figure 2, full triangles),
whereas the purified one gave only 13% of Ho9–δSb5(O,C)5
(Figure 2, x = 0, open diamonds). We do not know how
much carbon is present in the commercial lanthanum and
whether its amount is sufficient to yield 75 wt.-% of
La9Sb5O4C. If the amount of carbon is insufficient then
potentially other impurities, although very unlikely, may
stabilize these phases. Hydrogen is unlikely to be a factor,
as the high temperatures and dynamic vacuum make Ta
tubing transparent to hydrogen, thus allowing hydrogen to
escape from the sample.[20] We believe that accidental car-
bon contamination during handling and preparation may
be a factor.
The proof that carbon, and not other light elements, sta-
bilizes the targeted phases comes from the series of experi-
ments involving the purified La and Ho and controlled
amounts of carbon in RE9Sb5O5–xCx (Figure 2). The largest
yield for the RE9Sb5O4C phases (P4/n) was achieved for the
loading compositions with x = 1. Another important find-
ing regards the composition of the RE9–δSb5(O,C)5 phases.
The sintered Ho9Sb5O4.5C0.5 sample contained the
Ho9–δSb5(O,C)5 phase with a high yield (99 wt.-%), which
suggests that the O/C ratio in some RE9–δSb5(O,C)5 phases
may be close to 4.5:0.5 (or 9:1). As discussed before, the O/
C ratio is likely to be variable (4:1 Ͻ O/C ratio Յ 9:1). The
amount of carbon retained in this series of phases is, how-
ever, unlikely to exceed 1, since the samples with x Ͼ1 were
less pure, as shown in Figure 2.
The layered nature of these phases suggests that other
structures may be formed, for example, with REBi slabs in-
stead of RESb ones, or with thicker RESb slabs. Currently,
the research efforts are conducted to test these ideas.
Experimental Section
Synthesis: The rare earth antimonide oxide/oxycarbide samples
were prepared from antimony metal (99.999 wt.-%, CERAC Inc.),
La2O3 and Ho2O3 powders (99.99 wt.-%, Rhône–Poulenc) and car-
bon powder (99.995 wt.-%, Alfa Aesar). Rare earth metals with
different levels of purity, including the commercial-grade RE metal
(ca. 93 atom-%, CERAC Inc.) and the RE metals purified by Ames
Laboratory (99.9 atom-%), were used for parallel syntheses and
analyses. The general synthetic schemes involved three steps. As the
first step, RESb binaries were prepared by direct sintering of the
elements. Mixtures of rare earth metal filings and ground elemental
antimony in a 1:1 atomic ratio were pressed into 1 g pellets in a
glove box. The samples were sealed in evacuated silica tubes 10–
15 cm in length, then heated to 600 °C at a rate of 50 °C/h. The
sintering temperature was maintained at 600 °C to allow the anti-
mony to react with the rare earth metal. After 12 h, the temperature
was raised to 850 °C at a rate of 50 °C/h and kept for 48 h to drive
the reaction to completion. Black pellets were obtained after cool-
ing in air. The purity of these binaries was confirmed by X-ray
powder diffraction analysis. The second step involved preparing the
pseudo compounds “RE4O5” and “RE4O4C” from a stoichiometric
amount of RE metal filings, calcined RE2O3 powders, and carbon
powder. The mixtures were pressed into 0.5 g pellets under argon.
Subsequently, the sample pellets were sintered in evacuated silica
tubes at 1000 °C for 48 h. Uniform black pellets were obtained af-
ter cooling. The products were air-sensitive; therefore, they were
handled and stored in a glove box. Finally, the precursors were
mixed in different ratios according to:
5 RESb + (1 – x) “RE4O5” + x “RE4O4C” = RE9Sb5O5–xCx
(x = 0, 0.5, and 1)
Conclusion
Samples with a total weight of 0.3 g were pressed into pellets and
sealed in Ta tubes under argon. The Ta tubes were placed into a
In the process of resolving the compositions of the
RE9Sb5O4C phases, the novel phases RE9–δSb5(O,C)5 and molybdenum susceptor and heated in a high-frequency induction
3892 Eur. J. Inorg. Chem. 2011, 3887–3895
www.eurjic.org
© 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Resolving Composition and Structure of Natural Superlattice Phases
Table 3. Annealing temperature, carbon content, and product appearances of the RE9Sb5O5–xCx samples.
T [oC]
General formula
A (x = 0)
B (x = 0) C (x = 0.5) D (x = 1)
E (x = 1)
Commercial-grade RE metal
With extra 0.5 mol of carbon
1
2
3
4
1400
1550
1650
1575
La9Sb5O5–xCx
La9Sb5O5–xCx
Ho9Sb5O5–xCx
–
–
–
sintered
molten
sintered
molten
–
molten
sintered
–
molten
sintered
–
molten
sintered
–
molten
molten
–
Ho9Sb5O5–xCx + 0.3 Ho5Sb3
furnace under a dynamic vacuum below 10–5 Torr. The reaction
temperature was monitored by an optical pyrometer. All samples
were annealed at the desired temperature for 8 h, and then left to
cool to room temperature under dynamic vacuum. The products
obtained either remained in a pellet form or were solidified from a
melt. All samples appeared to be dark gray in color with a metallic
luster. The La samples were sensitive to air and decomposed into
a white/gray powder; the Ho samples were stable in air. Twelve
samples were prepared according to the general synthetic pro-
cedures. As shown in Table 3, these samples can be placed into five
groups according to the synthetic conditions and purity of the rare
earth elements. Samples labeled B, C, D, and E (Table 3) were pre-
pared from high-purity RE metals. Samples A were produced to
illustrate the impact of impurities in commercial-grade RE metals.
Lastly, samples E were synthesized with additional carbon to deter-
mine the maximum amount of carbon that can be retained in the
system.
data were collected with a STOE IPDSII diffractometer by using
Mo-Kα radiation in the whole reciprocal sphere. A numerical ab-
sorption correction was based on the crystal shape that was origi-
nally derived from the optical face indexing but was later optimized
against equivalent reflections by using the STOE X-Shape soft-
ware.[21] The data for the La8.80(1)Sb5(O,C)5 and La14Sb8O7C single
crystals were collected with a Bruker SMART Apex II CCD dif-
fractometer (Mo-Kα radiation) at 100(2) K in a reciprocal hemi-
sphere. Intensities were extracted and then corrected for Lorentz
and polarization effects through the SAINT program.[22] Numeri-
cal absorption correction was based on the crystal shape obtained
from the optical face indexing. Structural determinations and re-
finements of all the data sets were performed by using the
SHELXL program.[23] Crystal structures were determined for five
new phases. These new compounds are identified by their refined
compositions and space groups as follows: La14Sb8O7C with P4bm
symmetry, La8.80(1)Sb5(O,C)5 and Ho8.84(3)Sb5(O,C)5 with P4/nmm
symmetry, La9Sb5O4C and Ho9Sb5O4C with P4/n symmetry. The
crystallographic information of La14Sb8O7C, La8.80(1)Sb5(O,C)5,
Ho8.84(3)Sb5(O,C)5, La9Sb5O4C and Ho9Sb5O4C is given in Tables 4
and 5. Further details of the crystal structure investigations can
Single-Crystal X-ray Diffraction and Structure Refinement: Single-
crystal X-ray diffraction studies were performed on crystals ex-
tracted from the crushed samples. Room-temperature diffraction
Table 4. Crystallographic data and refinement results for the La14Sb8O7C, La8.80Sb5(O,C)5, La9Sb5O4C, Ho8.84Sb5(O,C)5, and Ho9Sb5O4C
single crystals (Mo-Kα radiation).
Sample code[a]
2B
1D
2D
3C
4D
[b]
[b]
Refined composition
Formula mass
Temperature [K]
Space group
La14Sb8O7C
3042.75
100(2)
La8.80(1)Sb5(O,C)5
1911.16
100(2)
P4/nmm
4.623(1)
9.388(2)
200.63(5)
0.40
La9Sb5O4C
1934.95
293(2)
P4/n
10.414(2)
9.334(2)
1012.2(3)
2
Ho8.84(3)Sb5(O,C)5
2148.38
100(2)
P4/nmm
4.3628(2)
8.8084(7)
167.659(7)
0.40
Ho9Sb5O4C
2169.13
293(2)
P4/n
9.782(1)
8.768(1)
838.94(1)
2
P4bm
Unit-cell dimensions [Å]
13.0788(1)
9.3886(2)
1605.97(4)
4
6.292
24.641
Volume [Å3]
Z
Dcalcd. [g cm–3]
Abs. coeff. [mm–1]
F(000)
6.327
24.878
319
6.349
25.063
1612
8.462
48.760
353
8.587
49.763
1792
2526
Crystal size [mm]
θ range [°]
Index ranges
0.062ϫ0.039ϫ0.015 0.090ϫ0.069ϫ0.054 0.023ϫ0.015ϫ0.006 0.027ϫ0.021ϫ0.019 0.048ϫ0.041ϫ0.029
3.09–52.22
–26 Յ h Յ 28,
–19 Յ k Յ 29,
–18 Յ l Յ 20
33821
2.17–45.21
–9 Յ h Յ 8,
–9 Յ k Յ 8,
–17 Յ l Յ 16
5866
3.52–34.74
–16 Յ h Յ 15,
–16 Յ k Յ 16,
–14 Յ lՅ 14
21750
4.63–44.94
–8 Յ h Յ 7,
–8 Յ k Յ 7,
–17 Յ l Յ 14
3101
2.32–29.23
–13 Յ h Յ 10,
–13 Յ k Յ 13,
–10 Յ l Յ 11
5452
Reflections collected
Independent reflections
Rint
8703
520
2112
456
1142
0.1029
0.0359
0.1163
0.2166
0.0859
Completeness to max. 2θ
Data/restraints/param.
Goodness-of-fit on F2
98.2%
8703/1/83
1.007
95.6%
520/0/17
1.044
96.4%
2112/0/47
1.083
99.8%
456/0/17
0.908
99.4%
1142/0/47
0.760
Final R indices [I Ͼ 2σ(I)] R1 = 0.0570
wR2 = 0.1036
R1 = 0.0184
wR2 = 0.0359
R1 = 0.0245,
wR2 = 0.0372
0.0046(5)
1.239/–1.860
R1 = 0.0531
wR2 = 0.0902
R1 = 0.1027,
wR2 = 0.1024
0.00046(6)
4.056/–3.991
R1 = 0.0481
wR2 = 0.1045
R1 = 0.0642,
wR2 = 0.1081
0.0040(9)
3.452/–6.895
R1 = 0.0263
wR2 = 0.0421
R1 = 0.0596,
wR2 = 0.0469
0.00103(4)
1.946/–2.557
R indices (all data)
R1 = 0.1084,
wR2 = 0.1233
0.00007(2)
Extinction coefficient
Diffr. peak/hole [eÅ–3]
6.200/–5.423
[a] Sample codes refer to the synthetic conditions listed in Table 3. [b] Composition refined with oxygen atoms occupying all interstitial
sites, since the ratio of O and C could not be determined unambiguously.
Eur. J. Inorg. Chem. 2011, 3887–3895
© 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.eurjic.org
3893

P. Wang, S. Forbes, V. Svitlyk, A. Aushana, Y. Mozharivskyj
FULL PAPER
Table 5. Atomic and isotropic temperature (U) parameters for the La14Sb8O7C, La8.80Sb5(O,C)5, La9Sb5O4C, Ho8.84Sb5(O,C)5, and Ho9S-
b5O4C single crystals.
Atom
Site
Occupancy
x/a
y/b
z/c
U [Å2]
La14Sb8O7C, 100(2) K
La(1)
La(2)
La(3)
La(4)
La(5)
La(6)
Sb(1)
Sb(2)
Sb(3)
Sb(4)
O(1)
2a
2b
4c
4c
8d
8d
2a
2b
4c
8d
2b
4c
8d
2a
1
1
1
1
1
1
1
1
1
1
1
1
1
1
0
0
0
1/2
0.25085(3)
0.60905(3)
0.24992(4)
0.88289(3)
0
0.0188(2)
0.0029(1)
0.00983(8)
0.3462(1)
0.69143(5)
0.3529(1)
0.6582(1)
0.6464(1)
0.65386(9)
0.04785(7)
0.239(1)
0.0040(2)
0.0047(1)
0.0042(1)
0.00448(8)
0.00427(7)
0.00418(5)
0.0044 (2)
0.0047(2)
0.0049(1)
0.0041(1)
0.008(2)
0.24915(3)
0.10905(3)
0.99027(4)
0.15480(3)
0
0
1/2
0.23992(4)
0.00157(4)
0
0.2387(5)
0.2390(4)
0
0.26008(4)
0.75143(4)
1/2
0.7387(5)
0.0271(4)
0
O(2)
O(3)
C(1)
0.262(1)
0.4393(7)
0.278(1)
0.007(1)
0.0066(8)
0.0013(2)
La8.80(1)Sb5(O,C)5, 100(2) K
La(1)
La(2)
Sb(1)
O(1)[a]
2c
8i
2c
2c
1
1/4
3/4
1/4
1/4
1/4
0.1810(2)
1/4
0.3405(1)
0.9972(2)
0.6974(1)
0.0889(7)
0.0075(1)
0.0175(2)
0.0074(1)
0.045(2)
0.1897(7)
1
1
1/4
Ho8.84(3)Sb5(O,C)5, 100(2) K
Ho(1)
Ho(2)
Sb(1)
2c
8i
2c
2c
1
1/4
3/4
1/4
1/4
1/4
0.1905(8)
1/4
0.33855(7)
0.999(2)
0.6933(1)
0.086(2)
0.0122(2)
0.025(1)
0.0122(2)
0.051(5)
0.192(1)
1
1
O(1)[a]
1/4
La9Sb5O4C, 293(2) K
1/4
La(1)
La(2)
La(3)
Sb(1)
Sb(2)
O(1)
2c
8g
8g
2c
8g
8g
2c
1
1
1
1
1
1
1
1/4
0.6663(2)
0.3460(1)
0.0020(1)
0.3070(3)
0.7018(1)
0.092(1)
0.0071(3)
0.0069(1)
0.0076(1)
0.0108(4)
0.0072(1)
0.010(2)
0.3535(1)
0.1692(1)
1/4
0.9536(1)
0.0219(1)
1/4
0.9472(1)
0.9727(9)
1/4
0.3471(1)
0.3722(8)
1/4
C(1)
0.929(3)
0.011(4)
Ho9Sb5O4C, 293(2) K
Ho(1)
Ho(2)
Ho(3)
Sb(1)
Sb(2)
O(1)
2c
8g
8g
2c
8g
8g
2c
1
1
1
1
1
1
1
1/4
1/4
0.6713(2)
0.3443(1)
0.0022(1)
0.3155(3)
0.6986(1)
0.090(1)
0.0104(3)
0.0096(1)
0.0094(1)
0.0125(4)
0.0094(2)
0.011(2)
0.3540(1)
0.1682(1)
1/4
0.3458 (1)
0.3699(6)
1/4
0.9547(1)
0.0233(1)
1/4
0.9455(1)
0.9740(8)
1/4
C(1)
0.931(4)
0.016(6)
[a] Atomic sites that were expected to be occupied by a mixture of O and C.
be obtained from the Fachinformationszentrum Karlsruhe, 76344
the Rietveld refinement. The La9–δSb5(O,C)5, Ho9–δSb5(O,C)5,
Eggenstein-Leopoldshafen, Germany (Fax: +49-7247-808-666; La9Sb5O4C, and Ho9Sb5O4C phases were identified in the bulk
E-mail: crysdata@fiz.karlsruhe.de) on quoting the depository num-
bers CSD-422097 (for La14Sb8O7C), -422098 [for La8.80(1)Sb5-
(O,C)5], -422099 (for La9Sb5O4C), -422095 [for Ho8.84(3)Sb5(O,C)5],
and -422096 (for Ho9Sb5O4C).
samples. The extracted lattice constants, purity, and corresponding
side products are listed in the Supporting Information with the
sample codes representing the synthetic conditions given in Table 3.
Supporting Information (see footnote on the first page of this arti-
cle): Crystal-structure details, evidence of structural anomalies, re-
sults from X-ray powder diffraction, and analyses of the formation
of side products.
X-ray Powder Diffraction: The sintered samples were subjected to
X-ray powder diffraction analyses. Between 20 and 50 mg of sam-
ple were used for each data collection. Diffraction data in the 20–
70° 2θ range were collected with a PANalytical X’Pert Pro dif-
fractometer with an X’Celerator detector and Cu-Kα1 radiation.
From the X-ray powder pattern, the desired phase in each sample
was identified along with the side products. The full-profile Riet-
veld refinement (Rietica program)[24] was used to refine the lattice
constants and the amount of side products. The structural models
obtained from single-crystal X-ray diffraction studies were used for
Acknowledgments
This work was supported by a Discovery Grant from the Natural
Sciences and Engineering Research Council of Canada and by a
grant from the ACS Petroleum Research Fund.
3894
www.eurjic.org
© 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Eur. J. Inorg. Chem. 2011, 3887–3895

Resolving Composition and Structure of Natural Superlattice Phases
[13] D. W. Jones, I. J. McColm, J. Yerkess, J. Solid State Chem.
1991, 92, 301.
[1]
K. Koumoto, I. Terasaki, R. Funahashi, MRS Bull. 2006, 31,
[14] Z. Sarbak, Binary Rare Earth Oxides (Eds.: G. Adachi, N. Im-
anaka, Z. C. Kang), Kluwer Academic Publishers, Dordrecht,
The Netherlands, 2005, vol. 84.
206.
[2]
[3]
[4]
F. J. Disalvo, Science 1999, 285, 703.
G. J. Snyder, E. S. Toberer, Nat. Mater. 2008, 7, 105.
F. Gascoin, S. Ottensmann, D. Stark, S. M. Haile, G. J. Snyder,
Adv. Funct. Mater. 2005, 15, 1860.
J. Nuss, M. Jansen, Acta Crystallogr., Sect. B 2007, 63, 843.
J. Nuss, H. G. von Schnering, Y. Grin, Z. Anorg. Allg. Chem.
2004, 630, 2287.
[15] I. D. Brown, D. Altermatt, Acta Crystallogr., Sect. A 1984, 40,
C444.
[16] M. Atoji, J. Chem. Phys. 1981, 74.
[5]
[6]
[17] N. E. Brese, M. O’Keeffe, Acta Crystallogr., Sect. B 1991, 47,
192.
[18] D. Fort, V. K. Pecharsky, K. A. Gschneidner, J. Alloys Compd.
1995, 226, 190.
[7] J. Nuss, M. Jansen, Z. Kristallogr. New Cryst. Struct. 2009,
224, 11.
[8] D. X. Li, Y. Haga, H. Shida, T. Suzuki, Y. S. Kwon, Phys. Rev.
B 1996, 54, 10483.
[9] G. S. Nolas, J. L. Cohn, B. C. Chakoumakos, G. A. Slack,
Therm. Conduct. 2000, 25, 122.
[10] R. Gaume, B. Viana, D. Vivien, J.-P. Roger, D. Fournier, Appl.
Phys. Lett. 2003, 83, 1355.
[11] P. Wang, S. Forbes, T. Kolodiazhnyi, K. Kosuda, Y. Mozhariv-
skyj, J. Am. Chem. Soc. 2010, 132, 8795.
[12] D. W. Jones, I. J. McColm, R. Steadman, J. Yerkess, J. Solid
State Chem. 1984, 53, 376.
[19] S. Y. Dan’kov, A. M. Tishin, V. K. Pecharsky, K. A. Gschne-
idner, Phys. Rev. B 1998, 57, 3478.
[20] E. A. Leon-Escamilla, J. D. Corbett, Chem. Mater. 2006, 18,
4782.
[21] X-Shape, STOE & Cie. GmbH, Darmstadt, 2004.
[22] SAINT, v6.02, Bruker AXS Inc., Madison, WI, 2002.
[23] G. M. Sheldrick, Acta Crystallogr., Sect. A 2008, 64, 112.
[24] B. A. Hunter, C. J. Howard, Rietica, Australian Nuclear Sci-
ence and Technology Organization, Menai, Australia, 2000.
Received: March 25, 2011
Published Online: August 2, 2011
Eur. J. Inorg. Chem. 2011, 3887–3895
© 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.eurjic.org
3895