Synthesis and Structural Evolution of RuSb3, a New Metastable Skutterudite Compound

Article abstract of DOI:10.1021/ic030209o

A thin-layer synthesis technique was used to synthesize bulk amounts of the metastable phase, RuSb₃, a novel compound with the skutterudite structure. The compound crystallized at 350 °C and was stable to 525 °C. When annealed above 550 °C, it decomposed into RuSb₂ and Sb. Rietveld refinement of X-ray diffraction data showed the presence of excess Sb residing in the interstitial site in the skutterudite structure. X-ray diffraction and thermal analysis experiments allowed us to examine the evolution of the sample as a function of annealing and determine the reaction pathway. The activation energy for the crystallization of the compound was determined to be 3 eV/nucleation event, while the activation energy for decomposition was approximately 8 eV.

Full text of DOI:10.1021/ic030209o

Inorg. Chem. 2004, 43, 2486−2490
Synthesis and Structural Evolution of RuSb , a New Metastable
3
Skutterudite Compound
Arwyn L. E. Smalley, Michael L. Jespersen, and David C. Johnson*
Chemistry Department, 1253 UniVersity of Oregon, Eugene, Oregon 97403
Received June 25, 2003
A thin-layer synthesis technique was used to synthesize bulk amounts of the metastable phase, RuSb
compound with the skutterudite structure. The compound crystallized at 350 °C and was stable to 525 °C. When
annealed above 550 °C, it decomposed into RuSb and Sb. Rietveld refinement of X-ray diffraction data showed
3
, a novel
2
the presence of excess Sb residing in the interstitial site in the skutterudite structure. X-ray diffraction and thermal
analysis experiments allowed us to examine the evolution of the sample as a function of annealing and determine
the reaction pathway. The activation energy for the crystallization of the compound was determined to be 3 eV/
nucleation event, while the activation energy for decomposition was approximately 8 eV.
Introduction
predicting undiscovered compounds is to use analogous phase
diagrams as guides to identify potential local energy minima
for particular structures in related phases. We used this
The field of solid state chemistry is continually faced with
the challenge of how to make new, useful compounds. Unlike
the fairly gentle temperatures and techniques used in organic
chemistry, most solid state chemistry reactions take place at
high temperatures to overcome the diffusion difficulties
inherent in reacting solids. The resulting compounds, many
of them very useful, are the thermodynamic products of these
reactions. In recent times, the focus has changed from high
temperature, thermodynamic products, to searching for
metastable products, which are usually synthesized at lower
temperatures. Several schemes have been used to achieve
predictive technique to target and prepare RuSb
3
, a new
compound with the skutterudite structure.
Skutterudites have been of interest to the materials science
community for some time, since they have been identified
as potentially useful for thermoelectrics applications. The
skutterudite structure consists of a transition metal atom
octahedrally coordinated to six pnictide atoms. The octahedra
are shared by the metal atoms, creating a 1:3 metal-to-
pnictide ratio. The linking of the octahedra creates a 12
coordinate interstitial site, belted by four pnictides bound in
a ring. The interstitial site is empty in the thermodynamically
1
this, including flux and hydrothermal methods, high-pressure
2
3
techniques, and thin film techniques.
stable binary skutterudite compounds, CoSb
3 3
, RhSb , and
In addition to the reactivity challenges due to slow solid
state diffusion rates, and unlike in organic chemistry, there
are few guidelines in materials chemistry for predicting the
stability and structure of unknown solid state phases. In solid
state chemistry, the multiple coordination geometries possible
even within binary systems are so large that it makes
accurately predicting the coordination and structure of
IrSb . Although the electrical properties of these compounds
3
are highly favorable for thermoelectric materials, their
5
thermal conductivities are too high, and so there have been
consistent attempts to optimize their electronic and thermal
properties. Two main procedures have been suggested to
decrease the thermal conductivity in skutterudites. The first,
suggested by Slack, is to insert rare earth atoms into the void
in the structure. These interstitial atoms will “rattle” in the
4
unknown compounds very difficult. One approach to
6
*
To whom correspondence should be addressed. E-mail: davej@
void and scatter phonons. The second method is to use larger
oregon.uoregon.edu.
metal or pnictide atoms in the compound, which provides a
(
(
(
1) Stein, A.; Keller, S. W.; Mallouk, T. E. Science 1993, 259, 1558-
564.
2) Takizawa, H.; Miura, K.; Ito, M.; Suzuki, T.; Endo, T. J. Alloys Compd.
7
1
larger, more complex unit cell. As a result of Slack’s
1999, 282, 79-83.
(5) Caillat, T.; Borshchevsky, A.; Fleurial, J.-P. J. Appl. Phys. 1996, 80,
4442-4449.
(6) Slack, G. A. In CRC Handbook of Thermoelectrics; Rowe, D. M.,
Ed.; CRC Press: Boca Raton, FL, 1995; pp 407-440.
(7) Morelli, D. T.; Meisner, G. P. J. Appl. Phys. 1995, 77, 3777-3781.
3) Fister, L.; Novet, T.; Grant, C. A.; Johnson, D. C. In AdVances in the
Synthesis and ReactiVity of Solids; JAI Press Inc.: London, England,
1994; Vol. 2, pp 155-234.
(4) DiSalvo, F. J. Science 1990, 247, 649-655.
2486 Inorganic Chemistry, Vol. 43, No. 8, 2004
10.1021/ic030209o CCC: $27.50 © 2004 American Chemical Society
Published on Web 03/19/2004

3
RuSb , a New Metastable Skutterudite Compound
suggestions, the binary compounds have been modified to
make the related filled compounds, A Sb12 (A ) rare
earth; M ) Co, Rh, and Ir), where the A atoms occupy the
2 coordinate “void” site in the skutterudite structure.
The ternary compounds, A Sb12 (A ) rare earth; M )
x
M
4
1
x
M
4
Fe, Ru, Os), have also been extensively studied. In these
compounds, the atom A, in the interstitial site, is necessary
to donate electrons to the compound, stabilizing the struc-
8
ture. The related binary compounds, FeSb
3
, RuSb
, lack filler atoms and are not thermodynamically
stable. Although the binary compound FeSb has been
3
, and
OsSb
3
3
9
formed, it is metastable, and there have been no reports to
date of the synthesis of binary skutterudites formed from
the remaining elements in this column of the periodic table.
The antimony-rich part of the Ru-Sb system has been
previously investigated, but the synthesis of a binary skut-
terudite was not reported.10
Figure 1. Representative low-angle X-ray diffraction pattern of the
unannealed and annealed (to 335 °C) samples. The Bragg reflection has
disappeared during annealing, indicating complete interdiffusion.
Table 1. Compositional Data and Intended and Actual Thicknesses for
All Samples
intended
actual
actual tot.
layer thickness layer thickness thickness
There are extensive similarities among the phases formed
by Sb with the group 8 and 9 transition metals. All six
elements form 1:2 compounds with Sb, and all but Os form
name at. % Ru at. % Sb
(Å)
(Å)
(Å)
A1
A2
A3
A4
A5
A6
A7
A8
4.29
6.77
8.29
16.42
17.68
21.97
35.73
40.61
95.71
93.23
91.71
83.58
82.32
78.03
69.27
59.39
15
16
16
15
15
15
15
15
14.5
15.6
15.9
11.9
14.4
12.1
14.5
15.8
474
795
684
1045
no peaks
1051
1
:1 compounds. All the 1:2 compounds have the marcasite
FeS structure, FeSb, CoSb, and IrSb all have the NiAs
structure, and RuSb and RhSb both have the MnP struc-
2
1
1,12
ture.
These similarities have lead us to attempt to
, both in the hopes that it will prove to be
1145
no peaks
synthesize RuSb
3
useful for thermoelectric applications and to provide further
insight into the factors that control or affect metastable phase
formation.
and on both film and powder samples. The powder samples were
combined with NIST silicon powder standard reference material
number 604b to verify lattice parameters and for quantitative
analysis using Rietveld refinement. Unless indicated otherwise,
samples were annealed for 30 min at a time in a Thermolyne 1500
box furnace under nitrogen atmosphere. Annealing temperatures
ranged from 300 to 600 °C. Compositional studies were conducted
on a Cameca S-50 electron probe microanalysis (EPMA) instrument
using 20 nA beam current and beam voltages of 8, 10, and 15 keV.
Differential scanning calorimetry measurements were conducted
on a Netsch DSC 200 PC. 0.5-1 mg of powder material was used
for all DSC experiments, and the samples were heated from 50 to
550 or 600 °C, at a rate of 2, 5, 10, or 20 °C/min.
Experimental Section
All samples were synthesized in a custom-built ultrahigh vacuum
13
deposition system, described elsewhere. Samples were synthesized
-
6
-7
in a 10 -10 Torr atmosphere. Ruthenium was deposited from
an electron beam gun at a rate of 0.2 or 0.5 Å/s, and antimony was
deposited from an effusion cell at a rate of 1 Å/s. The ruthenium
and antimony layers were deposited alternately for 80 repeats,
resulting in repeat layer thicknesses of 10-15 Å and total sample
thicknesses of 800-1200 Å. A computer-controlled quartz crystal
monitoring system was used to control layer thickness. Samples
were deposited on a 4 in. diameter silicon wafer coated with poly-
Results and Discussion
(
methyl methacrylate) (PMMA), a silicon chip, and a 1 in. square
We synthesized a series of samples that varied in
composition from Sb-rich to Ru-rich, as presented in Table
1. Layer and total thicknesses were calculated from the Bragg
reflections and the Kiessig fringes, respectively. In some
cases the samples were sufficiently rough that no Kiessig
fringes were discernible, and we could not calculate the total
thickness (indicated by “no peaks” in Table 1). These
samples were synthesized with total thicknesses similar to
the other samples. The samples are completely X-ray
amorphous before annealing, indicating that no crystalline
material is present. Figure 1 shows a representative X-ray
reflectivity pattern for a sample before and after annealing
to 335 °C. The Bragg reflection in the unannealed sample
results from the modulation of electron density in the sample
due to sequential deposition of the elements rather than any
crystalline character. The Kiessig fringes are visible in both
scans, but the first Bragg reflection has disappeared during
annealing, indicating that the layers have completely inter-
piece of zero-background cut quartz. The coated wafers were soaked
in acetone to dissolve the polymer, which lifted the powder sample
off of the wafer. The powder sample was collected by vacuum
filtration using a Teflon filter.
X-ray diffraction studies were conducted on the quartz piece and
the silicon chip using a Philips X’Pert diffractometer and a Scintag
XDS-2000 θ-2θ diffractometer. X-ray reflectivity and diffraction
studies were conducted on both as-deposited and annealed samples
(
(
8) Nordstrom, L.; Singh, D. J. Phys. ReV. B 1996, 53, 1103-1108.
9) Hornbostel, M. D.; Hyer, E. J.; Thiel, J.; Johnson, D. C. J. Am. Chem.
Soc. 1997, 119, 2665-2668.
(10) Caillat, T.; Borshchevsky, A.; Fleurial, J.-P. J. Phase Equilib. 1993,
14, 576-578.
(
11) Villars, P.; Calvert, L. D. Pearson’s Handbook of Crystallographic
Data for Intermediate Phases, 2 ed.; ASM International: Materials
Park, OH, 1991.
(
(
12) Daams, J. L. C.; Villars, P.; van Vucht, J. H. N. Atlas of Crystal
Structure Types for Intermetallic Phases; ASM International: Materials
Park, OH, 1991.
13) Fister, L.; Li, X.-M.; McConnell, J.; Novet, T.; Johnson, D. C. Journal
Vac. Sci. Technol., A 1993, 11, 3014-3019.
Inorganic Chemistry, Vol. 43, No. 8, 2004 2487

Smalley et al.
Figure 3. Differential scanning calorimetry data for sample A6, with a
heating rate of 10 °C/min. The exotherm at 350 °C is due to crystallization,
and the exotherms between 475 and 550 °C are due to decomposition into
RuSb2 and Sb.
Figure 2. High-angle X-ray diffraction data collected from a powder that
was heated in the DSC to 550 °C showing three crystalline phases: RuSb3,
RuSb2, and Sb. The peaks were distinct enough to allow calculation of the
RuSb3 lattice parameter: 9.338(1) Å.
diffused. Before complete interdiffusion, the position of the
Bragg reflection shifted, reflecting a contraction in the layer
thickness from 12.1 to 11.8 Å. This layer contraction is
frequently observed during annealing of thin film samples
and is attributed to the increase in sample density as
14
vacancies and other defects are eliminated. A summary of
the total and layer thicknesses for the unannealed samples
can be found in Table 1.
The samples with greater than 90 at. % antimony did not
show any irreversible events in the differential scanning
calorimetry experiments. X-ray diffraction analysis of these
samples indicated that the only phase to crystallize was
antimony. The lack of an exotherm for the crystallization of
antimony is likely due to it crystallizing gradually, rather
Figure 4. Selected high-angle X-ray diffraction scans for sample A6,
showing the change in phases from pure RuSb (500 °C and below) to a
3
mixed phase of RuSb3 and RuSb2 at 550 °C and then a mixed phase of
RuSb2 and Sb at 575 °C.
than all at once, which has been observed in a number of
conducted, using the thermal analysis data as a guide. Figure
cases.1
4,15
Differential scanning calorimetry experiments
3
contains the DSC data for sample A6, which has the closest
conducted on nearly all the samples with less than 90 at. %
antimony showed one irreversible exotherm in the temper-
ature range under 450 °C. This peak occurred between 290
and 390 °C. In ∼10% of the samples, a second exotherm
with very small amplitude occurred within 50 °C of the first
exotherm. We were unable to identify any structural changes
that pertained to this exotherm. High-angle X-ray diffraction
studies were conducted on powder samples from the DSC
experiments, each of which had been annealed to 550 or 600
stoichiometry to the ideal 1:3 ratio. High-angle X-ray
diffraction data collected below the exotherm at 350 °C
revealed that the sample remained X-ray amorphous. Figure
4
summarizes the diffraction data collected as a function of
annealing temperature from 400 to 575 °C, with the sample
held at each temperature for 30 min. All scans through the
525 °C annealing (not shown) show a phase-pure skutterudite
diffraction pattern, which indicates that the large exotherm
is associated with the crystallization of the skutterudite phase.
The scan of the sample annealed to 550 °C reveals a mixed
°
C. When annealed to 550 °C, all samples where the
composition was less than 90% Sb showed mixed phases of
RuSb , Sb, and in some cases RuSb . The targeted skutteru-
phase of RuSb
3
and the thermodynamically stable phase
2
3
RuSb ; the skutterudite phase has completely disappeared
2
dite phase was visible in samples A4, A5 (see Figure 2),
and A7. The lattice parameter of the skutterudite phase of
A5 was calculated to be 9.338(1) Å, which is significantly
after the sample was annealed at 575 °C. This decomposition
is confirmed by exothermic peaks in the DSC experiments
where the sample was heated to 600 °C (Figure 3).
Exothermic decomposition peaks indicate that the compound
is metastable; this metastability confirms the findings of
higher temperature syntheses, in which this phase was unable
1
6
larger than the reported values for CoSb
3
(a ) 9.034 Å)
(a ) 9.2503 Å) and is similar to values for BaRu
Sb12 (a ) 9.315 Å).18
17
and IrSb
3
4
-
To determine the structural changes taking place during
the observed exotherms, detailed annealing studies were
10
to be formed. The differences between the temperatures
for the structural transitions viewed using X-ray diffraction
and the exothermic peaks observed in the thermal analysis
of these samples is due to several differences in the treatment
of the samples. First, the thermal analysis was conducted
on powder samples, while the annealing studies for X-ray
diffraction were conducted on thin films that had been
(
(
(
(
(
14) Maki, K. J. Cryst. Growth 1983, 65, 77-83.
15) Maki, K.; Ookawa, A. J. Cryst. Growth 1986, 79, 144-148.
16) Ackermann, J.; Wold, A. J. Phys. Chem. Solids 1977, 38, 1013-1016.
17) Slack, G. A.; Tsoukala, V. G. J. Appl. Phys. 1994, 76, 1665-1671.
18) Evers, C. B. H.; Boonk, L.; Jeitschko, W. Z. Anorg. Allg. Chem. 1994,
620, 1028.
2488 Inorganic Chemistry, Vol. 43, No. 8, 2004

3
RuSb , a New Metastable Skutterudite Compound
Figure 5. Representative Rietveld refinement. Red points are experimental
data, the black line is the calculated curve, and the blue is the difference
curve which is offset for clarity. The lattice size is 9.336 Å, Ru atoms are
located at (0.25, 0.25, 0.25), and Sb atoms are at (0, 0.16, 0.34) and (0, 0,
Figure 6. Superimposed differential scanning calorimetry data for the scans
used in Kissinger analysis. The black scan is 2 °C/min, the red is 5 °C/
min, the blue is 10 °C/min, and the green is 20 °C/min.
0
). The thermal parameters and occupancy for Ru and Sb1 were held
constant at U ) 0.01 and F ) 100%, respectively. Both parameters were
2
refined for Sb2 (interstitial), giving U ) 0.41 and F ∼ 57%. Ì ) 2.7, Rwp
Sb
previously reported filled RuSb
earth atoms (Ba, La, Ce, Pr, Nd, Eu) have been included to
δ
Ru
4
Sb12 to reflect the Sb in the interstitial site. In
)
0.094, and Rp ) 0.071.
3
compounds, several rare
Table 2. Sample Rietveld Refinement Dataa
1
8-20
U (Å2)
donate charge and stabilize the structure.
These filled
atom
x
y
z
occ.
skutterudites were designed to have 100% filling of the
interstitial site. The coincidental filling that we observe, of
Sb in the interstitial site, is therefore not surprising, since
Sb should be able to donate charge to stabilize the RuSb
3
structure. We are not aware of other cases in which Sb is
used as a filled atom, other than that which we observed
Ru
Sb1
Sb2
0.25
0
0
0.25
0.1607
0
0.25
0.3424
0
1
1
0.57
0.01
0.01
0.41
a
U values and occupancies for Ru and Sb1 are fixed; all other values
2
are refined. X ) 2.658. Rwp ) 0.0935. Rp ) 0.0714. Lattice size ) 9.325
Å.
3
being incorporated into the CoSb interstitial site in a recent
study.
deposited on off-cut quartz. In addition, the DSC samples
were heated at a constant rate through a range of tempera-
tures, while the X-ray samples were held at one constant
temperature for a longer period of time. Kissinger analysis
demonstrates the effect of heating rate on the temperature
at which a thermal event occurs, so it is not surprising that
the temperatures vary somewhat between the two methods.
A more careful look at the high-angle diffraction data
showed changes in the lattice parameter as a function of
annealing temperature. We suspect that this shift is a result
of defects being removed from the structure during annealing.
Rietveld refinement was employed to provide more in depth
structural information, including the atomic positions and
occupancies. The standard skutterudite crystal structure, with
Im 3h symmetry, 8 Ru atoms and 24 Sb atoms per unit cell,
and an empty interstitial site, was not able to reproduce the
peak intensities. Several structural and experimental variables
can affect the peak intensities, including crystallite size and
strain, sample transparency to the X-ray beam, diffractometer
alignment, and sample position in the X-ray beam. Modeling
these parameters moderately improved the fit but did not
correct the intensity mismatches. Since the samples had been
synthesized with excess Sb present, we added Sb to the
interstitial site (0, 0, 0), and this solved the intensity problem.
There are two interstitial sites/unit cell, giving a maximum
of two additional Sb atoms/unit cell. The refined occupancy
of Sb at (0, 0, 0) was approximately 55% and resulted in a
significantly improved fit for intensity (see Figure 5). The
large thermal parameter (U ) 0.41) is typical of the
interstitial atoms in skutterudites. Table 2 summarizes the
structural parameters. We conclude that although we have
2
1
In a broader context, it is instructive to examine the
structural evolution of the initially modulated reactants to
determine the factors that control metastable phase formation.
In bulk reactions, interdiffusion and interfacial nucleation
determine which compounds form. In contrast, the Ru-Sb
samples investigated here were deposited with angstrom-
thick layers, which allowed interdiffusion to be completed
during gentle heating, before nucleation could occur at the
evolving interfaces, as confirmed by the diffraction data in
Figure 1. After interdiffusion, only a small amount of
additional heating was required to induce the samples to
nucleate and crystallize. The composition of the amorphous
intermediate determined which compounds crystallized.
Samples with 75-85% Sb crystallized RuSb
energy of crystallization of RuSb , 3 eV/nucleation event,
was calculated from thermal data collected at scan rates of
, 5, 10, and 20 K/min, (see Figure 6) via a method first
3
. The activation
3
2
22
suggested by Kissinger. The presence of distinct, exother-
mic decomposition peaks allowed us to determine the
activation energy for transforming RuSb
dynamically favored products, RuSb and Sb. This energy
was calculated to be approximately 8 eV, significantly higher
than the activation energy to form RuSb . Since the tem-
3
into the thermo-
2
3
peratures of the two decomposition peaks did not correspond
exactly to the decomposition temperature we observed using
(
19) Evers, C. B. H.; Jeitschko, W.; Boonk, L.; Braun, D. J.; Ebel, T.;
Scholz, E. D. J. Alloys Compd. 1995, 224, 184-189.
(20) Takeda, N.; Ishikawa, M. J. Phys. Soc. Jpn. 2000, 69, 868-873.
(
21) Smalley, A. L. E.; Kim, S.; Johnson, D. C. Chem. Mater. 2003, 15,
847-3851.
(22) Kissinger, H. E. Anal. Chem. 1957, 29, 1702-1706.
3
3
synthesized binary RuSb , it would be better expressed as
Inorganic Chemistry, Vol. 43, No. 8, 2004 2489

Smalley et al.
Figure 7. This reaction pathway is specific to one in which
nucleation, not diffusion, is the rate-limiting step and in
which only enough heat is input to allow movement between
the kinetic steps in the reaction pathway.
Summary
Using a thin layer vacuum deposition technique, we have
synthesized bulk amounts of the novel, metastable, binary
skutterudite phase, RuSb
mixture crystallizes at low temperatures to selectively form
this phase. The lattice size of RuSb is even larger than the
filled A Ru Sb12 samples that have been previously synthe-
sized. This partially results from the presence of excess
Sb in the “voids” in the skutterudite structure. Differential
scanning calorimetry data indicate that this metastable phase
has a window of kinetic stability, decomposing if annealed
3
. The initially amorphous, layered
3
x
4
1
9
Figure 7. Schematic of proposed reaction pathway, from mixing of the
layered compound, to crystallization, and to decomposition.
X-ray diffraction, and at some scan rates we only clearly
observed one decomposition peak, we used the average
between the two energies calculated to provide an ap-
proximate decomposition energy.
past 525 °C. The exothermic decomposition of RuSb
RuSb and Sb demonstrates that this novel phase is meta-
stable at all temperatures.
3
into
2
The fact that this decomposition is exothermic and results
in two stable phases indicates that RuSb is thermodynami-
3
cally unstable at all temperatures. This confirms Caillat’s
Acknowledgment. We gratefully acknowledge support
from the National Science Foundation (DMR-9813726), and
one of us (A.E.L.S.) gratefully acknowledges the support of
the U.S. Department of Education Graduate Assistance in
Areas of National Need Program (P200A010222), and the
National Science Foundation GK-12 Program (DGE-0231997).
investigations into the Ru-Sb system, since their syntheses
10
were conducted at temperatures too high for RuSb
The formation of an amorphous phase, followed by crystal-
lization of RuSb and eventually decomposition into RuSb
and Sb, might follow a reaction pathway similar to that in
3
to form.
3
2
IC030209O
2490 Inorganic Chemistry, Vol. 43, No. 8, 2004