Investigation of the transport properties and compositions of the Ca2RE7Pn5O5 series (RE=Pr, Sm, Gd, Dy; Pn=Sb, Bi)

Article abstract of DOI:10.1016/j.jssc.2016.04.015

The Ca₂RE₇Pn₅O₅ phases (RE=Pr, Sm, Gd, Dy; Pn=Sb, Bi) were successfully prepared from high temperature reactions at 1225–1300?°C. These phases maintain the same structure types as the parent RE₉Pn₅O₅ phases, except for a Ca/RE mixing. The study and preparation of these phases was motivated by the desire to shift the metallic type properties of the parent RE₉Pn₅O₅ phases to a level more suitable for thermoelectric applications. Electrical resistivity measurements performed on pure, bulk samples indicated all phases to be narrow band gap semiconductors or semimetals, supporting the charge balanced electron count of the Ca₂RE₇Pn₅O₅ composition. Unfortunately, all samples are too electrically resistive for any potential usage as thermoelectrics. Electronic band structure calculations performed on idealized RE₉Pn₅O₅ structures revealed the presence of a pseudogap at the Fermi level, which is consistent with the observed electrical resistivity and Seebeck coefficient behavior.

Full text of DOI:10.1016/j.jssc.2016.04.015

Author’s Accepted Manuscript
Investigation of the transport properties and
compositions of the Ca RE Pn O series (RE = Pr,
2
7 5 5
Sm, Gd, Dy; Pn = Sb, Bi)
Scott Forbes, Fang Yuan, Kosuke Kosuda, Taras
Kolodiazhnyi, Yurij Mozharivskyj
www.elsevier.com/locate/yjssc
PII:
S0022-4596(16)30139-6
http://dx.doi.org/10.1016/j.jssc.2016.04.015
YJSSC19361
DOI:
Reference:
To appear in:
Journal of Solid State Chemistry
Received date: 29 January 2016
Revised date: 8 April 2016
Accepted date: 11 April 2016
Cite this article as: Scott Forbes, Fang Yuan, Kosuke Kosuda, Taras
Kolodiazhnyi and Yurij Mozharivskyj, Investigation of the transport properties
and compositions of the Ca RE Pn O series (RE = Pr, Sm, Gd, Dy; Pn = Sb
2
7 5 5
B i ) , Journal
of
Solid
State
Chemistry
http://dx.doi.org/10.1016/j.jssc.2016.04.015
This is a PDF file of an unedited manuscript that has been accepted fo
publication. As a service to our customers we are providing this early version o
the manuscript. The manuscript will undergo copyediting, typesetting, and
review of the resulting galley proof before it is published in its final citable form
Please note that during the production process errors may be discovered which
could affect the content, and all legal disclaimers that apply to the journal pertain

Investigation of the transport properties and compositions of the
Ca2RE7Pn5O5 series (RE = Pr, Sm, Gd, Dy; Pn = Sb, Bi)
Scott Forbesa, Fang Yuan, Kosuke Kosudab, Taras Kolodiazhnyib, and Yurij
Mozharivskyja*
aDepartment of Chemistry and Chemical Biology, McMaster University, 1280
Main Street West, Hamilton, Ontario, L8S 4M1, Canada
bInstitute for Materials Science, 1-1 Namiki, Tsukuba, Ibaraki, 305-0044, Japan
*Corresponding author, email: mozhar@mcmaster.ca
Abstract
The Ca2RE7Pn5O5 phases (RE = Pr, Sm, Gd, Dy; Pn = Sb, Bi) were successfully prepared
from high temperature reactions at 1225-1300°C. These phases maintain the same structure types
as the parent RE9Pn5O5 phases, except for a Ca/RE mixing. The study and preparation of these
phases was motivated by the desire to shift the metallic type properties of the parent RE9Pn5O5
phases to a level more suitable for thermoelectric applications. Electrical resistivity
measurements performed on pure, bulk samples indicated all phases to be narrow band gap
semiconductors or semimetals, supporting the charge balanced electron count of the
Ca2RE7Pn5O5 composition. Unfortunately, all samples are too electrically resistive for any
potential usage as thermoelectrics. Electronic band structure calculations performed on idealized
RE9Pn5O5 structures revealed the presence of a pseudogap at the Fermi level, which is consistent
with the observed electrical resistivity and Seebeck coefficient behavior.
Keywords: rare-earth antimonide oxides; rare-earth bismuthide oxides; crystal structure;
thermoelectric properties; electronic structure.
Introduction
One of the reemerging areas of research in the development of alternate forms of energy
is the study of new and more efficient thermoelectric materials. While most of the early
thermoelectrics obtained the figure-of-merit ZT ~ 1, theoretical studies performed in the 1990s
suggested improvements in ZT through the decoupling of the electrical conductivity and Seebeck
coefficient by forming quantum nanostructures with reduced dimensionality.1-3 Although not
1

intended, the major advancements in thermoelectric efficiency have instead arisen from the
reduction of thermal conductivity by nanostructuring.4 Such notable gains in efficiency have
ushered in a new age of research into thermoelectrics, and a doubling of efficiency (ZT ~ 2)
among the state of the art materials, such as PbTe-SrTe.5 As such, there is a great deal of
motivation to discover new materials with intrinsically attractive properties that may push the
boundaries of thermoelectric efficiency even further.
Devices containing thermoelectric materials are capable of performing heating and
cooling, depending on the direction of the applied current, as well as being able to convert waste
heat into a usable form of electricity.6 Thermoelectrics offer several advantages over other
competing energy sources including long operation hours, scalability, quietness, and versatility.
To obtain the most suitable balance of these properties in a single phase, narrow band gap
semiconductors and semiconductors are targeted for thermoelectric applications as they possess
electrical properties corresponding to an optimized power factor α2σ. The lattice thermal
conductivity κl of a thermoelectric material must also be kept as low as possible, so structures
containing soft bonds, heavy atoms, and site disorder are popular choices.Unfortunately,
thermoelectrics suffer from low efficiencies due to the tradeoffs between Seebeck coefficient α,
electrical conductivity σ, and charge carrier thermal conductivity κe (Equation 1).
 2
kl  ke
ZT 
(1)
A number of methods are then applied to reduce thermal conductivity further, mostly by doping
and nanostructuring.
One of the more noteworthy strategies for the design of a good thermoelectric is the
design of a natural superlattice structure, in which electrically conductive and phonon scattering
properties are accomplished in separate portions of the lattice. This approach is advantageous for
thermoelectricity since it allows for the tuning of one property without affecting the other. As
such, a natural superlattice structure may display a phonon-glass electron-crystal (PGEC)-type
behavior, as postulated by Slack.7 Several successful thermoelectric materials, such as Ca1-
8
xYbxZn2Sb2 and NaxCo2O49, 10 display natural superlattice type structures, and have been shown
to achieve good ZT values.
As per the natural superlattice approach, we have turned our attention on rare-earth
pnictide oxide phases. Ideally, we may achieve good thermoelectric properties from a natural
2

superlattice type structure containing the electrically conductive and thermally insulating
frameworks of rare-earth pnictides and rare-earth oxides, respectively. The presence of heavy
rare-earth and pnictogen elements and the added structural complexity introduced by chemical
fusion of different sublattices would also serve to heavily reduce thermal conductivity. In
particular, the RE9Pn5O5 series is of special interest, since it displays natural superlattice-type
structures, with alternating slabs of RE5Pn5 and RE4O5 stacked along the c direction.11, 12 The
pristine RE9Pn5O5 phases have been shown to behave as metals, owing to their unbalanced
electron counts, which renders them unsuitable as thermoelectrics. Additionally, these phases are
very air sensitive, requiring great precautions for usage and handling. Our group has studied the
incorporation of carbon into these phases in the past in an effort to stabilize their structures and
form charge balanced phases for the purpose of thermoelectricity, but we were unable to achieve
a pure, stable sample.13
Our studies on the CaRE3SbO4 and Ca2RE8Sb3O10 phases14 motivated us to explore
calcium substitution in the RE9Pn5O5 phases. By a simple electron counting scheme, a charge
balanced (Ca2+)2(RE3+)7(Pn3-)5(O2-)5 formula would exist if the RE9Pn5O5 structure would
tolerate calcium/rare-earth mixing. Verily, we have shown that the Ca2RE7Pn5O5 phases do in
fact exist, demonstrating that calcium substitution simultaneously improves the stability and
maintains the structure of the parent RE9Pn5O5 phase while also eliminating its metallic-type
conduction. In this work, we present the details of the synthetic approach, structural and
compositional analyses, physical property measurements, and electronic structure calculations of
the Ca2RE7Pn5O5 phases.
Experimental
Synthesis
Samples were made using high-purity RE metals (RE = Pr, Sm, Gd, Dy; 99.9 wt.% or
better, SmartElements), calcium metal (99.98 wt.% Alfa Aesar), antimony metal (99.999 wt.%,
CERAC Inc.), RE2O3, (99.99 wt.%, Rhône-Poulenc for RE = Sm, Gd, Dy; 99.9 wt.% Alfa Aesar
for RE = Pr), and calcium oxide (99.99 wt. %, CERAC). The RESb and REBi binary phases were
first prepared as precursors to avoid the use of elemental antimony and bismuth, which are
volatile at high temperatures. These samples were prepared in an argon-filled glove box by
mixing stoichiometric amounts of RE and Sb/Bi. The RESb samples were pressed into pellets
3

while the REBi samples were not. All samples were sealed below 10-4 torr in evacuated silica
ampoules of 10 to 15 cm in length. The silica tubes were carbon coated to avoid reaction
between RE and silica. The RESb samples were heated in a box furnace at 600°C for 12 hours,
then at 850°C for 48 hours. The REBi samples were heated at 235°C for 12 hours, then at 350°C
for 12 hours, and finally at 850°C for 48 hours. The resulting powder was ground and pressed
into a pellet within an argon-filled glove box. The pellets were then sealed in silica tubes and
heated at 850°C for 48 hours. Black, solid chunks of rare-earth antimonides and rare-earth
bismuthides were obtained in this manner, and the purity of each sample was confirmed by X-ray
powder diffraction experiments.
Preparation of the Ca2RE7Sb5O5 and Ca2RE7Bi5O5 samples (RE = Pr, Sm, Gd, Dy) was
accomplished by mixing stoichiometric amounts of calcium oxide, rare-earth sesquioxide
(RE2O3), and RESb/REBi precursors in an argon-filled glove box. All samples were consolidated
with the use of a hydraulic press, and were subsequently sealed in tantalum ampoules with the
use of an arc melter. This method allows for high temperature reactions with no oxidation,
sample loss, or reaction with the container. High-temperature reaction conditions were
accomplished through the use of an induction furnace. The tantalum ampoules containing
samples were placed in a molybdenum susceptor, which was heated under dynamic vacuum
below 10-4 torr by radio frequency induction from a water-chilled copper coil. The Ca2RE7Sb5O5
and Ca2RE7Bi5O5 samples were heated for 18 hours at 1300°C and 1225°C, respectively. Upon
completion of heat treatments, samples were allowed to cool for one hour before removal. The
Ca2RE7Sb5O5 and Ca2RE7Bi5O5 samples were obtained as shiny, molten chunks. Decomposition
of the Ca2RE7Sb5O5 and Ca2RE7Bi5O5 samples due to moisture after at least 48 hours was
confirmed by X-ray powder diffraction. Thus, all samples were stored in an evacuated dessicator
filled with drierite, which was observed to maintain sample purity for several months. A
summary of all samples prepared is presented in Table 1.
Table 1. Ca2RE7Pn5O5 samples prepared by high-temperature synthesis in an induction furnace
(Pn = Sb, Bi).
Temperature
System
RE used
Loading composition
State
used
4

Light gray,
molten
Ca2RE7Sb5O5
Ca2RE7Bi5O5
Pr, Sm, Gd
Gd, Dy
2CaO + 5RESb + RE2O3
2CaO + 5REBi + RE2O3
1300°C
1225°C
Dark gray,
molten
X-ray Single Crystal Diffraction
Single crystals picked up from the samples were analyzed on a Bruker SMART Apex II
diffractometer using MoKα radiation. Intensity corrections for Lorentz and polarization effects
were performed with the SAINT program.15 A multi-scan absorption correction was applied
based on the crystal shape determined by optical face indexing. Due to an ordering of vacancies,
a five-fold supercell for each structure is created (when compared to the parent Sc2Sb-type
structure), which in turn resulted in twinning for all crystals except Ca2Dy7Bi5O5. All data sets
except Ca2Dy7Bi5O5 were detwinned using the TWINABS program to eliminate intensity bias
due to overlapping reflections. Crystal structures were determined and solved using the SHELX
software.16 The final refinements of the twin crystals were performed using the hkl5-type files
containing intensities of both twinned components. A summary of the refinement results is
presented in Table 2. All Ca2RE7Pn5O5 phases crystallize in the P4/n space group, much like the
analogous RE9Sb5O5 phases. Upon inspection, it was observed that there was an unusually large
amount of electron density on the normally vacant 2a sites in each structure from the twinned
crystal (except for the Ca2Dy7Bi5O5 crystal that was not twinned). Most likely, this extra electron
density stems from the twinning interfaces that place RE/Ca atoms almost exactly in the center of
square voids (i.e. the 2a site), and such phenomenon has been observed in the parent RE9Pn5O5
structure12. This conclusion is supported by the fact that an additional electron density is not
found in the non-twinned Ca2Dy7Bi5O5 crystal. Additionally, the presence of Ca or RE atoms on
the 2a site would offset the stoichiometry and charge balance of the Ca2RE7Pn5O5 phases.
However, we could not prove with absolute certainty that the extra electron density is the artifact
of the twinning, or it is real and in fact leads to twinning. As a result, we report the crystal
structures below with the artifact peak as a residual electron density peak. Further information on
the crystal structures can be found in the Supporting information and may be obtained from the
Fachinformationszentrum Karlsruhe, 76344 Eggenstein-Leopoldshafen, Germany (fax (49)
5

7247-808-666; email crysdata@fiz.karlsruhe.de), by quoting the CSD depository numbers
429690 – 429692 for Ca2RE7Sb5O5 and 429693 – 429694 for Ca2RE7Bi5O5.
Table 2. Crystallographic and refinement data for the Ca2RE7Pn5O5 crystals.
Ca2Pr7Sb5O5
Ca2Sm7Sb5O5
Ca2Gd7Sb5O5
Ca2Gd7Bi5O5
Ca2Dy7Bi5O5
Refined
composition
Space
Ca2.11(5)Pr6.89(5)Sb5
O5
Ca2.01(4)Sm6.99(4)Sb5
O5
Ca1.98(2)Gd6.98(2)Sb5
O5
Ca2.39(5)Gd6.61(5)Bi5
O5
Ca2.09(3)Dy6.91(3)Bi5
O5
P 4/n
P 4/n
P 4/n
P 4/n
P 4/n
group
Mo Kα (0.71073 nm)
Radiation
Scan mode
ω and φ
Temperatur
e
296(2) K
Crystal
dimensions
(mm)
0.080 x 0.066 x
0.053
0.072 x 0.064 x
0.058
0.077 x 0.070 x
0.062
0.083 x 0.068 x
0.061
0.115 x 0.094 x
0.049
10.223(1)
9.138(2)
10.067(1)
9.011 (2)
9.992(1)
8.976(2)
10.102(1)
9.027(2)
9.9996(1)
8.9341(2)
a (Å)
c (Å)
Volume
(Å3)
Twin
955.1(3)
913.2(3)
896.2(3)
921.2(3)
893.34(2)
N/A
0.531/0.469(3)
0.512/0.488(2)
0.506/0.494(2)
0.501/0.499(3)
fractions
pcalc (g/cm3)
6.015
2
6.622
2
6.938
2
8.147
2
8.668
2
Z
-13 ≤ h ≤ 14
0 ≤ k ≤ 20
0 ≤ l ≤ 18
0 ≤ h ≤ 20
-13 ≤ k ≤ 14
-18 ≤ l ≤ 0
0 ≤ h ≤ 20
-13 ≤ k ≤ 14
-17 ≤ l ≤ 0
0 ≤ h ≤ 20
-13 ≤ k ≤ 14
-18 ≤ l ≤ 0
-19 ≤ h ≤ 17
-20 ≤ k ≤ 19
-17 ≤ l ≤ 17
Index
ranges
2θ max
90.80°
23235
3314
90.76°
44532
3850
90.82°
28980
3786
90.90°
35183
3934
90.64°
36098
3757
Measured
reflections
Unique
reflections
Reflections
used
Number of
parameters
Twinning
artifact el.
densitya,
e/Å3
3918
50
9464
51
5120
51
6235
51
1848
50
15.45
11.475
5.424
21.497
N/A
Max/min el.
densityb,
e/Å3
4.29/-3.32
1.020
3.65/-4.13
1.024
4.01/-2.72
1.048
4.29/-4.47
1.138
4.35/-2.84
1.017
Goodness-
of-fit on |F2|
6

R1 (>4σ) =
0.0543
wR2 = 0.1195
R1 (all data) =
0.1547
R1 (>4σ) =
0.0533
wR2 = 0.897
R1 (all data) =
0.1183
R1 (>4σ) =
0.0383
wR2 = 0.0822
R1 (all data) =
0.0682
R1 (>4σ) = 0.0467 R1 (>4σ) = 0.0326
wR2 = 0.0881
R1 (all data) =
0.0938
wR2 = 0.0637
R1 (all data) =
0.0711
R indices
a
The largest residual electron density peak in these structures, except for Ca2Dy7Bi5O5, is an
artifact of the twinning these crystals display.
b
The second largest electron density peak, except for Ca2Dy7Bi5O5 where the largest peak is
reported.
X-ray Powder Diffraction
Each sample under study was analyzed by X-ray powder diffraction on a PANalytical
X’Pert Pro instrument using CuKα1 incident radiation and an X’Celerator detector.
Approximately 50 mg of sample was ground up using a mortar and pestle and deposited on a
zero-background Si holder. The data were collected in the 2θ range between 20° and 70° and
analyzed using the Rietveld refinement method (Rietica program17) to determine sample purity
and lattice constants. The structural parameters obtained from the single crystal solutions were
used for the refinements, and site occupancies for calcium and rare-earth were fixed.
Electron Microprobe Analysis (EPMA)
The samples of interest were mounted in discs of epoxy resin one inch in diameter and
approximately 10 mm in thickness. Polishing was accomplished using a Struers Rotopol-31 unit
with a Struers RotoForce-4 revolving sample holder. The final step of polishing used a solution
of diamond dust of approximately 1 µm in size. Quantitative elemental analysis of the samples
was performed by electron probe microanalysis (EPMA) using wavelength-dispersive (WDS) X-
ray spectroscopy (model JXA-8500F, JEOL). CaSiO3, PrB6, SmB6, Gd3Ga5O12, DyP5O14,
antimony metal, and bismuth metal were used as standards to determine the concentration of
oxygen, Ca, Pr, Sm, Gd, Dy, Sb, and Bi in the samples, respectively.
Thermoelectric Measurements
Pure samples of Ca2RE7Pn5O5 (RE = Pr, Sm, Gd, Dy; Pn = Sb, Bi) were prepared,
pressed, sealed in evacuated silica ampoules, and annealed at 500°C for three days. Powder X-
ray diffraction analysis confirmed no decomposition or generation of impurities by these
methods. Once annealed, samples were cut into bars of approximately 8  2  2 mm in
dimensions using a kerosene lubricant to avoid sample oxidation, and were subsequently stored
in an argon atmosphere. Thermoelectric measurements were carried out on a Quantum Design
7

Physical Property Measurement System (PPMS). Platinum leads were attached on the
rectangular bars using Epo-Tek H20E silver epoxy. The contacts for all samples were cured
under a stream of ultra-high purity Ar inside a tube furnace.
Electronic Band Structure Calculations
The electronic structures of all samples for which a crystal could be obtained and solved
were calculated using the tight-binding, linear-muffin tin orbital method18 with the atomic sphere
approximation (TB-LMTO-ASA) as implemented in the Stuttgart program.19 All 4f electrons
were considered as core electrons. Exchange and correlation were treated by the local density
approximation (LDA).20 A scalar relativistic approximation21 was employed to account for all
relativistic effects except spin-orbit coupling. Overlapping Wigner-Seitz cells were constructed
with radii determined by requiring the overlapping potential to be the best approximation to the
full potential, according to the atomic sphere approximation (ASA). Automatic sphere
generation22 was performed to construct empty spheres to be included in the unit cell in order to
satisfy the overlap criteria of the TB-LMTO-ASA model. The X-ray single crystal data,
including lattice parameters and atomic coordinates, were used as the model for calculations. The
Ca/RE sites in Ca2RE7Sb5O5 and Ca2RE7Bi5O5 were set to be fully occupied by rare-earth atoms,
resulting in idealized RE9Sb5O5 and RE9Bi5O5 structures with no changes in atomic parameters
(such approach neglects a size difference23 between Ca2+ and RE3+).
Results and Discussion
Phase Formation and Stability
In a previous study by our group, the RE9Sb5O5 phases were reacted with carbon in an
effort to yield a charge balanced, stable phase (assuming C4- was present). In most cases,
however, a pure sample could not be obtained, and all samples were observed to be highly air
sensitive.13 Instead, we have determined that stabilization by calcium addition is more practical.
Our experiments revealed that sample compositions of Ca2RE7Pn5O5 may yield air stable, pure
phases through high temperature reaction conditions. Site mixing of calcium and rare-earth is
accomplished in all Ca2RE7Pn5O5 phases due to similar atomic radii23; this strategy was also
employed in the discovery of the CaRE3SbO4 and Ca2RE8Sb3O10 phases.14
The stable nature of the Ca2RE7Pn5O5 phases most likely results from the charge
balanced electron formula achieved by substitution of trivalent RE3+ with divalent Ca2+. Due to
8

the mixing of calcium and rare-earth, as well as the potential for off-stoichiometry, a Ca5-
xRE5+0.667xSb5O5 compositional series with varying amounts of calcium and rare-earth may exist.
However, our studies concluded that all samples of this series share the same Bragg peak
positions in their powder X-ray diffraction data, suggesting that there is an optimal composition
of calcium and rare-earth that is most stable. We then determined that the Ca2RE7Pn5O5
stoichiometry yielded the most pure samples, indicating this system is thermodynamically driven
to a charge balanced phase. Thus, our synthetic approach was able to yield pure Ca2RE7Sb5O5
phases for the RE = La - Dy, and pure Ca2RE7Bi5O5 samples for late members of the RE = Gd -
Dy series (Ca2RE7Bi5O5 samples containing early RE were observed to form (Ca,RE)2BiO2).
High temperature reactions in an induction furnace are necessary to form pure
Ca2RE7Pn5O5 samples. Phase formation was observed to occur for all temperatures between
1200°C and 1600°C, although temperatures of 1300°C and 1225°C were determined to be the
most suitable for obtaining pure samples of Ca2RE7Sb5O5 and Ca2RE7Bi5O5, respectively. Gray,
molten products are obtained for the Ca2RE7Pn5O5 samples, with the Ca2RE7Bi5O5 samples
being slightly darker in tone. A set of experiments performed on the pristine RE9Pn5O5 samples
confirmed their instability, with samples tarnishing in air after a few minutes and full
decomposition in air after 24-48 hours. In contrast, the Ca2RE7Pn5O5 phases were observed to be
stable in dry air or argon, although still reactive in moist air, with noticeable decomposition
occurring after 48 hours. Thus, storage in an argon-filled glovebox or evacuated desiccator was
deemed to be necessary in order to prevent the loss of sample integrity.
Structural determination and analysis
The basic frameworks of the Ca2RE7Pn5O5 phases are identical to the RE9Pn5O5 series.
Based on the single crystal data, all Ca2RE7Pn5O5 phases adopt the tetragonal P4/n symmetry,
assuming an oxygen-stuffed, rare-earth deficient derivative of the Sc2Sb-type structure.24, 25 Each
Ca2RE7Pn5O5 phase contains site mixing of calcium and rare-earth. These phases consist of
NaCl-type (Ca/RE)5Pn5 slabs stacked along the c direction and separated by (Ca/RE)4O5 layers
containing (Ca/RE)4O and (Ca/RE)5O polyhedra (Figure 1). One fifth of all calcium/rare-earth
sites with respect to the Sc2Sb structure are vacant in the parent RE9Pn5O5 phases; these ordered
vacancies form the basis of the superstructure with respect to the original P4/nmm cell and occur
in the (Ca/RE)4O5 layers, forming ordered, empty channels along the c direction.12 The existence
of the P4/n phase was also confirmed by X-ray powder diffraction experiments based on the
9

presence of characteristic weak (220) and (221) reflections, which are observed for the P4/n
superstructure but not the P4/nmm structure.
Figure 1. The Ca2RE7Pn5O5 structure viewed along the b and c direction (left and right,
respectively). Twinning interfaces result in electron density located in the empty channels along
the c direction.
According to the single crystal diffraction data, each phase (except Ca2Dy7Bi5O5)
1
displays reticular merohedral twinning with /5 of all reflections shared between domains. In
order to obtain reliable diffraction intensities, each data set (except for Ca2Dy7Bi5O5) had to be
detwinned with the TWINABS program, yielding corrected intensities that could be used to
solve the crystal structure of one twin component. The resulting data sets were solved with good
R factors and yielded atomic parameters comparable to those in the original RE9Pn5O5 structures.
However, one major difference was observed: each of the crystal solutions, apart from the non-
twinned Ca2Dy7Bi5O5, had residual electron density on the normally vacant 2a site. Since the
pristine RE9Pn5O5 phases do not contain rare earth atoms on these sites, we assume these sites to
be indeed vacant. As mentioned before, this electron density can be attributed the twinning
interfaces in the crystal, and is supported by the fact that the non-twinned Ca2Dy7Bi5O5 crystal
does not have this property.
Compositional analysis
Unfortunately, our crystallographic models of the Ca2RE7Pn5O5 phases make it difficult
to obtain an accurate composition due to the calcium and rare-earth mixing. Thus, we performed
electron microprobe analysis on all samples in order to verify their compositions. The results are
presented below in Table 3.
10

Table 3. Comparison of experimentally determined compositions of Ca2RE7Sb5O5 and
Ca2RE7Bi5O5 samples derived from X-ray diffraction and EPMA experiments.
X-ray diffraction results
Phase
Loading composition
EPMA results
Ca2Pr7Sb5O5
Ca2Sm7Sb5O5
Ca2Gd7Sb5O5
Ca2.11(5)Pr6.89(5)Sb5O5
Ca2.13(6)Pr6.89(4)Sb5.00(4)O6.92(6)
a
Ca2RE7Sb5O5
Ca2.01(4)Sm6.99(4)Sb5O5 Ca2.03(5)Sm7.00(3)Sb5.00(3)O6.47(5)
Ca1.98(2)Gd6.98(2)Sb5O5
Ca2.16(6)Gd6.92(3)Sb5.00(3)O5.75(6)
Ca2Gd7Bi5O5
Ca2Dy7Bi5O5
Ca2.39(5)Gd6.61(5)Bi5O5
Ca2.09(3)Dy6.91(3)Bi5O5
Ca2.07(5)Gd6.95(3)Bi5.00(3)O5.99(6)
Ca2.11(6)Dy6.93(4)Bi5.00(3)O5.75(6)
a
Ca2RE7Bi5O5
a The Ca2RE7Sb5O5 and Ca2RE7Bi5O5 samples were normalized to compositions containing five
antimony/bismuth atoms.
The EPMA data for all samples revealed atomic amounts no greater than three standard
deviations from the loading compositions. The calcium amounts were consistently higher than
expected, which is perhaps due to the fact that the Kα/Kβ peaks of calcium and the Lα/Lβ peaks of
antimony overlap. The oxygen amounts were also well above the expected values, which can be
attributed to oxidation from surface polishing. No tantalum signals were observed in any of the
data sets. Based on the EPMA, we believe that all samples under study maintain the loading
compositions without any significant fluctuations.
Physical Properties and Electronic Structure
The pure Ca2RE7Sb5O5 and samples displayed an Arrhenius-type decrease in electrical
resistivity with temperature, suggesting that these phases are semiconductors (a bump in the
Ca2Pr7Sb5O5 data at lower temperatures occurs due to an antiferromagnetic transition).
Conversely, the Ca2RE7Bi5O5 phases displayed a linear decrease in electrical resistivity with
temperature (Figure 2). These types of behavior directly contrast with the parent RE9Pn5O5
phases, which were reported to have metallic-type conduction.11 Thus, by reducing the electron
count, we did succeed in removing the metallic-type electrical resistivity behavior with respect to
the parent RE9Pn5O5 structures and achieved activation energies on the order of about 0.04-0.06
eV in Ca2RE7Sb5O5 (Table 4). In the case of Ca2RE7Bi5O5, the electrical resistivity decreases
11

with temperature (similar to semiconductors) but the decrease is linear as opposed to exponential
in semiconductors. We could not establish the origin of such non-Arrhenius behavior, but a
similar trend was observed in some disordered semimetallic systems due to the electron-electron
correlations.26,27 Additionally, electrical resistivity at low temperatures is much lower for
Ca2RE7Bi5O5, suggesting a larger metallic character, likely due to the greater orbital overlap of
bismuth versus antimony. Overall, the Ca2RE7Pn5O5 phases display similar or lower electrical
resistivity than most other rare-earth pnictide oxide phases studied by our group so far. However,
these electrical resistivities (around 0.1-2 Ω cm at room temperature) are still too great for any
thermoelectric applications.
Figure 2. Electrical resistivity data for the Ca2RE7Pn5O5 samples: a) Ca2Pr7Sb5O5, b)
Ca2Sm7Sb5O5, c) Ca2Gd7Sb5O5, d) Ca2Dy7Sb5O5, e) Ca2Gd7Bi5O5, and f) Ca2Dy7Bi5O5.
Table 4. Room temperature electrical resistivity and activation energy data for selected
Ca2RE7Pn5O5 samples.
Room temperature
electrical resistivity
Sample
Activation energy
Ca2Pr7Sb5O5
Ca2Sm7Sb5O5
Ca2Gd7Sb5O5
Ca2Dy7Sb5O5
1.92 Ω cm
0.058 eV
0.048 eV
0.042 eV
0.033 eV
1.51 Ω cm
1.52 Ω cm
1.53 Ω cm
12

Ca2Gd7Bi5O5
Ca2Dy7Bi5O5
0.151 Ω cm
0.108 Ω cm
n/a
n/a
Seebeck coefficient measurements (α) on the Ca2RE7Sb5O5 and Ca2RE7Bi5O5 samples
revealed a near-linear dependence of α on temperature, with α approaching zero as T → 0 for all
samples (Figure 3), which is indicative of heavily doped (degenerate) semiconductors and
semimetals. The Ca2RE7Sb5O5 samples display larger Seebeck coefficients than the Ca2RE7Bi5O5
samples, which correlates well with their electrical resistivity. The Ca2Pr7Sb5O5 and
Ca2Sm7Sb5O5 phases also display a curvature in their Seebeck coefficient data with increasing
temperature, which could arise from contributions from electrons and holes in a complex
multiband structure. Small variations between samples with respect to calcium/rare-earth
amounts could also greatly affect the dominant charge carrier type. The nature of this type of
conduction may also explain why Ca2Gd7Sb5O5 is p-type while Ca2Gd7Bi5O5 is n-type (albeit
weakly). Unfortunately, the Seebeck values obtained from our experiments are far lower than
those observed from benchmark materials (α ≥ 150 μV K-1), making thermoelectric applications
unlikely.
The thermal conductivity for Ca2RE7Pn5O5 is dominated by the lattice contribution, since
even for the electrically most conductive Ca2Gd7Bi5O5 the electronic thermal conductivity is
below 10-2 Wm-1K-1 at the highest measured temperature of 375 K. At low temperatures, the
thermal conductivities of all Ca2RE7Pn5O5 samples rapidly increase with temperature, which is
due to the T3 dependence in their heat capacities (Figure 4). After reaching the maximum the
thermal conductivities of the Ca2RE7Sb5O5 phases slowly decline due to the Umklapp scattering.
The Ca2RE7Bi5O5 phases have less pronounced peaks in their thermal conductivities at low
temperatures, which is likely due to a short phonon free path resulting form the small particle
size and significant grain boundaries. This short phonon free path can also explain a relatively
small effect of the Umklapp scattering on the thermal conductivity of Ca2RE7Bi5O5 at higher
temperatures.
13

Figure 3. Seebeck coefficient data for the Ca2RE7Pn5O5 samples: a) Ca2Pr7Sb5O5, b)
Ca2Sm7Sb5O5, c) Ca2Gd7Sb5O5, d) Ca2Dy7Sb5O5, e) Ca2Gd7Bi5O5, and f) Ca2Dy7Bi5O5.
Figure 4. Thermal conductivity data for the Ca2RE7Pn5O5 samples: a) Ca2Pr7Sb5O5, b)
Ca2Sm7Sb5O5, c) Ca2Gd7Sb5O5, d) Ca2Dy7Sb5O5, e) Ca2Gd7Bi5O5, and f) Ca2Dy7Bi5O5.
Electronic band structure calculations performed on the idealized RE9Pn5O5 structures
reveal the presence of a noticeable pseudogap (density of states ≠ 0) at the Fermi level,
corresponding to a charge balanced electron formula (80 e/f.u. for Ca2RE7Pn5O5). Such a
distribution of states would be expected to yield semimetallic-like behavior with no activation
energy in the electrical resistivity. However, the Ca2RE7Sb5O5 samples display semiconductor
behavior, which suggests the presence of Ca in their structure leads to a small band gap. In case
of Ca2RE7Bi5O5, the Ca incorporation does not open a band gap and this effect combined with an
atomic disorder likely leads to an unusual behavior of the electrical resistivity. Presence of the
14

pseudogap at 80 e/f.u. indicates that the system is thermodynamically driven to a charge
balanced Ca2RE7Pn5O5 formula when calcium is introduced. For all phases, the valence band is
dominated by the Pn p states immediately below the Fermi level, and the states in the conduction
band are dominated by the RE d states. As with many other rare-earth pnictide oxide structures,
the density of states approaching the Fermi level becomes quite small, which may explain the
relatively high electrical resistivity values.
Figure 5. Calculated electronic band structures for the Ca2Gd7Sb5O5 structure (left) and
Ca2Gd7Bi5O5 structure (right).
Conclusions
Substituting rare earth with calcium in RE9Pn5O5 was shown to yield pure, charge-
balanced Ca2RE7Pn5O5 phases upon high temperature reactions. Samples of Ca2RE7Pn5O5 were
observed to be much more stable versus the corresponding RE9Pn5O5 phases when prepared by
similar methods, although the former are still moisture sensitive. X-ray powder and single crystal
diffraction experiments confirmed that the Ca2RE7Pn5O5 phases crystallize in the same P4/n
structure type as the parent RE9Pn5O5 phases, and in most cases display the same twinning laws.
The acquired X-ray single crystal data suggested residual electron density on the 2a site, which
we argue is an artifact of the reticular merohedral twinning.
Electrical resistivity and Seebeck coefficient measurements performed on pure
Ca2RE7Pn5O5 samples yielded semiconducting behavior for Ca2RE7Sb5O5 phases and a linear
decrease in resistivity for Ca2RE7Bi5O5 phases, in contrast to the metallic nature of the parent
RE9Pn5O5 structures. Unfortunately, the electrical resistivity and Seebeck coefficient values are
15

not ideal for thermoelectric applications. The calculated electronic structures of idealized
RE9Pn5O5 suggest semimetallic densities of states, which appear to be modified by the Ca
incorporation as seen from the experimental data.
Acknowledgements. This work was supported by a Discovery Grant and a CREATE HEATER
Grant from the Natural Sciences and Engineering Research Council of Canada.
Supporting Information. Additional crystallographic data, magnetic data, and electronic
structures are presented.
References
1.
Balandin, A.; Wang, K. L., J. Appl. Phys. 1998, 84 (11), 6149-6153.
2.
Hicks, L. D.; Dresselhaus, M. S., Phys. Rev. B: Condens. Matter 1993, 47 (19), 12727-
31.
3.
4.
Hicks, L. D.; Harman, T. C.; Dresselhaus, T. C., Appl. Phys. Lett. 1993, 63 (23), 3230-2.
Li, J. F.; Liu, W. S.; Zhao, L. D.; Zhou, M., NPG Asia Mater. 2010, 2, 152-158.
Biswas, K.; He, J.; Blum, I. D.; Wu, C.-I.; Hogan, T. P.; Seidman, D. N.; Dravid, V. P.;
5.
Kanatzidis, M. G., Nature (London, U. K.) 2012, 489 (7416), 414-418.
6.
7.
8.
Tritt, T. M., Science (Washington, D. C.) 1996, 272 (5266), 1276-1277.
Slack, G. A., CRC Handbook of Thermoelectrics 1995.
Gascoin, F.; Ottensmann, S.; Stark, D.; Haile, S. M.; Snyder, G. J., Adv. Funct. Mater.
2005, 15 (11), 1860-1864.
9.
10.
Terasaki, I., Int. Conf. Thermoelectr. 2002, 21st, 185-191.
Terasaki, I.; Sasago, Y.; Uchinokura, K., Phys. Rev. B: Condens. Matter 1997, 56 (20),
R12685-R12687.
11.
Nuss, J.; Georg von Schnering, H.; Grin, Y., Z. Anorg. Allg. Chem. 2004, (630), 2287-
2291.
12.
Nuss, J.; Jansen, M., Acta. Cryst. 2007, B63, 843-849.
13.
Wang, P.; Forbes, S.; Svitlyk, V.; Aushana, A.; Mozharivskyj, Y., Eur. J. Inorg. Chem.
2011, (26), 3887-3895.
14.
Forbes, S.; Yuan, F.; Saparov, B.; Sefat, A. S.; Kosuda, K.; Kolodiazhnyi, T.;
Mozharivskyj, Y., Chem. Mater. 2014, 26 (7), 2289-2298.
15.
16.
17.
Bruker Analytical X-Ray Systems 2002, Madison, WI, USA.
Sheldrick, G. M. SHELXL; University of Gottingen, Germany, 1997.
Hunter, B. A.; Howard, C. J., Rietica Australian Nuclear Science and Technology
Organization: Menai, Australia 2000.
18. Andersen, O. K.; Pawlowska, Z.; Jepsen, O. Phys. Rev. B: Condens. Matter Mater. Phys.
1986, 34, 5253–69.
16

19.
Jepsen, O.; Burkhardt, A.; Andersen, O. K. The TB-LMTO-ASA Program, version 4.7;
Max-Planck-Institut für Festkörperforschung: Stuttgart, Germany, 1999.
20.
21.
Andersen, O. K.; Jepsen, O. Phys. Rev. Lett. 1984, 53, 2571-2574.
Andersen, O. K.; Jepsen, O.; Glotzel, D. In Highlights of Condensed Matter Theory;
Bassani, F.; Fumi, F.; Tosi, M. P., Eds.; Elsevier: Amsterdam, 1985.
22.
23.
24.
25.
26.
476.
27.
Jepsen, O.; Andersen, O. K. Phys. B.: Consens. Matter 1995, 97, 35-47.
Shannon, R. D., Acta Crystallogr., Sect. A 1976, A32 (5), 751-67.
Berger, R., Acta Chem. Scand., Ser. A 1977, A31 (6), 514-16.
Nuss, J.; Jansen, M., Z. Kristallogr. - New Cryst. Struct. 2002, 217 (1), 19-20.
Besson J. M.; Makharti E. H.; Gonzalez J.; Weill G., Phys. Rev. Lett. 1987, 59 (5), 473-
Kaveh M; Mott N. F., J. Phys. C. 1982, 15, L707-716.
Highlights
 The RE9Pn5O5 structure may be stabilized with calcium substitution in the form of
Ca2RE7Pn5O5.
 The Ca2RE7Pn5O5 phases maintain the parent P 4/n structure, albeit with Ca/RE mixing.
 The Ca2RE7Sb5O5 phases behave as semiconductors while the Ca2RE7Bi5O5 phases are
semimetal with electron-electron correlations.
 Electronic structure calculations yield a semimetal-like density of states for both
Ca2RE7Sb5O5 and Ca2RE7Bi5O5.
17