Self-propagating High-temperature Synthesis of Ca1.24Co 1.62O3.86thermoelectric powders

Article abstract of DOI:10.1016/j.jallcom.2010.05.018

Thermoelectric oxide, Ca_1.24Co_1.62O_3.86, was prepared by Self-propagating High-temperature Synthesis (SHS) followed by a short heat treatment. Prepared powders were characterized by XRD for phase purity. Thermal analyses (TG/DSC) were used to establish the reaction mechanism of the process and study the product thermal stability under different gas atmospheres. Temperature history of the SHS process was measured and the activation energy of the reaction was calculated from the temperature history. Electrical conductivity, thermal diffusivity, and Seebeck coefficient of the prepared power were measured.

Full text of DOI:10.1016/j.jallcom.2010.05.018

Journal of Alloys and Compounds 503 (2010) 402–409
Contents lists available at ScienceDirect
Journal of Alloys and Compounds
journal homepage: www.elsevier.com/locate/jallcom
Self-propagating High-temperature Synthesis of Ca1.24Co1.62O3.86
thermoelectric powders
Sidney Lin∗, Jiri Selig
Dan F. Smith Department of Chemical Engineering, Lamar University, Beaumont, TX 77710, USA
a r t i c l e i n f o
a b s t r a c t
Article history:
Thermoelectric oxide, Ca1.24Co1.62O3.86, was prepared by Self-propagating High-temperature Synthesis
(SHS) followed by a short heat treatment. Prepared powders were characterized by XRD for phase purity.
Thermal analyses (TG/DSC) were used to establish the reaction mechanism of the process and study the
product thermal stability under different gas atmospheres. Temperature history of the SHS process was
measured and the activation energy of the reaction was calculated from the temperature history. Electrical
conductivity, thermal diffusivity, and Seebeck coefficient of the prepared power were measured.
© 2010 Elsevier B.V. All rights reserved.
Received 25 April 2010
Received in revised form 1 May 2010
Accepted 3 May 2010
Available online 10 May 2010
Keywords:
Thermoelectric materials
Calcium cobaltate
Combustion synthesis
Oxide materials
Thermal analyses (TG/DSC)
X-ray diffraction (XRD)
1. Introduction
From Eq. (1), a good thermoelectric material should have a large
Seebeck coefficient, a high electrical conductivity, and a low ther-
A tremendous amount of waste heat is generated from US
industries (5.04 × 1018 J) and transportation (2.21 × 1019 J) [1]. The
released waste heat not only reduces the fuel efficiency but also
is an important factor in climate change. Thermoelectric devices
provide a solution to reduce heat pollution and increase fuel effi-
ciency at the same time. In general, thermoelectric devices are used
in either electricity generation directly from a temperature differ-
ence or the creation of a temperature difference (cooling) from
electricity. No moving mechanical parts are needed in these appli-
cations. The Seebeck effect describes the generation of electrical
current/potential from a temperature difference (power genera-
tors) and the Peltier effect describes the creation of a temperature
difference using an electrical current (solid-state coolers) [2].
A thermoelectric material is characterized by its figure of merit,
ZT, defined by Eq. (1):
mal conductivity in order to maximize the ZT value. This implies
that a viable thermoelectric compound should be both electron
conductor and heat insulator at the same time. Slack suggested
the best thermoelectric material would behave as a “phonon-
glass/electron-crystal” (PGEC) [3]. In an ideal PGEC thermoelectric
material, the thermal resistance increases due to phonons scatter at
a shorter mean free path and the electrons move through the mate-
rial without increasing resistance due to their longer mean free
paths. It has been reported that the nanostructure inside thermo-
electric materials could scatter mid- to long-wavelength phonons
and thereby reduce the lattice thermal conductivity and increase
the figure of merit, ZT.
Most of the state of the art thermoelectric materials are alloys
of different metals and semiconductors, such as PbTe, SiGe, and
Bi2Te3 because of their high electrical conductivities [4]. However,
these materials have a tendency to oxidize at high temperatures
and are not stable in an oxidizing atmosphere, such as inside the
automobile exhaust lines where temperatures can reach as high
as 800 ◦C. To recycle the waste heat from automobile exhaust, it is
needed to develop thermoelectric materials other than alloys.
Unlike metallic and semiconducting thermoelectric materials,
oxide thermoelectric materials do not oxidize or vaporize at high
temperatures, retain their thermoelectric performance in oxidizing
environments, andare suitable forusein severeenvironments, such
as the exhaust line of an automobile. Several oxide materials have
been reported to have good thermoelectric properties and are good
˛2ꢀT
ZT =
(1)
ꢁ
where ˛ is Seebeck coefficient, ꢀ is electrical conductivity, T is tem-
perature, and ꢁ is thermal conductivity.
∗
Corresponding author at: Dan F. Smith Department of Chemical Engineering,
Lamar University, 4800 MLK Blvd., Beaumont, TX 77710, USA. Tel.: +1 409 880 2314;
fax: +1 409 880 2197.
E-mail address: sidney.lin@lamar.edu (S. Lin).
0925-8388/$ – see front matter © 2010 Elsevier B.V. All rights reserved.
doi:10.1016/j.jallcom.2010.05.018

S. Lin, J. Selig / Journal of Alloys and Compounds 503 (2010) 402–409
403
candidates for automobile applications. For example, NaCo2O4 is a
bronze-type compound expressed as AxBO2 [5] having alternating
layers of CoO2 and Na along the c-axis and being highly anisotropic
in the c-axis is a good thermoelectric oxide. Single crystal NaCo2O4
[6,7] have thermoelectric properties comparable to those of Bi2Te3,
but are difficult to produce and not economically feasible. Recently,
highly textured polycrystalline NaCo2O4 samples, grown by tem-
plated grain growth (TGG) [8] and reactive templated grain growth
(RTGG) [8,9] method, have shown to approach the thermoelectric
properties of single crystal NaCo2O4.
Spark plasma sintering (SPS) can be used to improve thermo-
electric properties of bulk oxides [10–12]. It was reported that SPS
could enhance power factor and electrical conductivity. The advan-
tage of SPS method is its rapid heat and mass transfer which results
textured ceramics [12].
Fig. 1. Sketch of SHS reactor (not to scale).
2. Experimental
Complex calcium cobalt oxides were used to improve the
thermoelectric performance of Na–Co–O by reducing the sodium
vaporization at high temperatures [13]. Various calcium cobaltates
have been reported with good thermoelectric properties, espe-
cially at high temperatures. They are good thermoelectric materials
due to their layered structure, combining high mobility layers and
phonon-glass layers [3]. All cobaltates have a similar structure
of alternating ordered metallic CoO2 blocks and insulating disor-
dered blocks [14]. In the case of Ca1.24Co1.62O3.86, its [CoO2] blocks
are alternating with its rocksalt-type [Ca2CoO3]0.62 blocks [15].
This mismatched structure leads to a reduced thermal conductiv-
ity of layered cobaltates due to a high phonon scattering at the
disordered boundaries of the different building blocks [16] and
provides a good combination of “phonon-glass/electron-crystal”
(PGEC).
Reported processes to synthesize oxide thermoelectric powders
include sol–gel or co-precipitation of metal salts followed by a high-
temperature sintering, and direct solid-state sintering of mixtures
of oxide, chloride, carbonate, or nitrate powders. Good ZT values
were reported for Ca3Co4O9 prepared by co-precipitation from
metal nitrates-hydrates followed by a spark plasma sintering (SPS)
[11,12], solid-state sintering followed by a solvent treatment to
increase grain size [17], and Ca1.24Co1.62O3.86 prepared by the pre-
cipitation method followed by templated grain growth [8]. These
processes use costly or hazardous raw materials, generate haz-
ardous gases (NOx), acids (HCl and HNO3), and greenhouse gas
(CO2) during the sintering. Furthermore, their long sintering peri-
ods at high temperatures make them energy intensive and increase
the overall production cost.
Stoichiometric proportions (Ca:Co = 1.24:1.62) of calcium peroxide powders
(Alfa Aesar, 65% pure balanced with Ca(OH)2), cobalt powders (Alfa Aesar, 99.5%,
−325 mesh), and Co3O4 (Alfa Aesar, 99.7%, −400 mesh) were mechanically mixed
for 4 h in a ball mill (U.S. Stoneware, Mahwah, NJ) and pressed into pellets 2.22 cm
(7/8 in.) in diameter under a load of 1 metric ton. In some experiments, a solid oxi-
dizer, NaClO4 (Alfa Aesar, 98.0–102.0%, anhydrous) was used to provide additional
oxygen for the synthesis. Reactant pellets were placed inside a custom made SHS
reactor (Fig. 1) and ignited under an oxygen atmosphere by a graphite igniter con-
trolled by a variable autotransformer (Staco Inc., Dayton, OH). The ignition period
was about 2 s. After the ignition, the igniter was turned off and the combustion front
moved forward and converted the reactants to the products as shown in reaction
(2).
1.24CaO2 + (1.62 − 3x)Co + xCo3O4 + 0.18NaClO4 + (0.33 − 2x)O2
→ Ca1.24Co1.62O3.86 + 0.18NaCl ꢂHr = −462 kJ/mol (for x = 0)
(2)
In this SHS reaction scheme, cobalt metal was the fuel to provide heat needed to
sustain the combustion front movement and CaO2 served as both the solid oxidizer
and the filler. The oxygen decomposed from CaO2, and in some experiments NaClO4,
oxidized the adjacent cobalt powder making the reaction less dependent on the
oxygen diffusing from the surrounding atmosphere. Temperature history during
the SHS reaction was measured by two K-type thermocouples (Omega CHAL-020,
Stamford, CT) inserted into the centerline of the pellet at a known distance apart.
After the SHS reaction, products were analyzed by XRD (Bruker D8 Discover GADDS)
for their phase purity.
Thermal behaviors of the reactants and products were studied using a Netzsch
STA 449C TG/DSC. In all thermal analyses alumina powders were used as a reference.
To understand the reaction mechanism during the SHS reaction, a loose mixture of
CaO2 and Co (Ca:Co = 1.24:1.62) was analyzed by TG/DSC under an oxygen atmo-
sphere. Loose powders were used to minimize oxygen diffusion resistance. Oxygen,
nitrogen, or air of a flow rate of 40 ml/min was used to study product stability under
different atmospheres. Sample heating rate was set to be 5 or 50 ◦C/min. The fast
heating rate (50 ◦C/min) was used to simulate reaction conditions of SHS condi-
tions and the slow heating rate (5 ◦C/min) was used to study the thermodynamic
behaviors of synthesized products.
In this work, we used a low cost solid-state combustion process,
Self-propagating High-temperature Synthesis (SHS) to produce
oxide thermoelectric powders. Self-propagating High-temperature
Synthesis process was developed in Russia in the late 1960s and
has been used to synthesize various ceramic materials, including
oxides, nitrides, carbides, and metal hydrides [18,19]. In an SHS,
reactants are mixed and pressed into a pellet. One end of the pellet
is then ignited by an external heat source. The reaction process is
highly exothermic, and reaction heat released from the combus-
tion is sufficient to sustain the reaction front movement at room
temperature. This process does not need a high-temperature fur-
nace, and the only external heat needed is for the ignition. The fast
combustion front movement (0.5–100 mm/s) enables a large-scale
production in a short period of time. In addition, its fast cooling after
the reaction allows the formation of metastable compounds and
nanostructured ceramic powders with very fine grains that could
further decrease the thermal conductivity of the products. The fea-
sibility of a continuous SHS production of Perovskites to further
reduce the production cost as well as capital investment has been
proven [20]. The motivation of this work is to use the economical
SHS process to produce calcium cobalt oxides with nanostructure
to improve its thermoelectric performance.
Some SHSed powders were pressed into pellets 1/2 in. in diameter under a load
of 2 metric tons and heat treated in a high-temperature furnace (Zircar Hot Spot 110)
at 850 ◦C for different times to understand the thermal stability of the SHS products
and post-treatment in some experiments.
Post-treatment on product was further studied in a tube furnace (Blue M,
TF55035A) in air or oxygen. The samples were placed into the tube furnace which
was preheated to 850 ◦C and removed immediately after the sintering to minimize
possible reactions during the heating up and cooling down processes.
The thermoelectric properties of synthesized Ca1.24Co1.62O3.86 powders were
measured during the feasibility study at Oak Ridge National Lab. The samples were
hot pressed at 850 ◦C under a load of 20 MPa. The Seebeck coefficient and electric
resistivity were measured using ULVAC ZEM-3 system.
3. Results and discussion
3.1. TG/DSC analyses
Thermal gravity analysis of CaO2 heated at a rate of 50 ◦C/min
in O2 shows two mass losses between 390 ◦C and 700 ◦C (Fig. 2).
The abrupt mass loss starting from 390 ◦C is attributed to the
decomposition of CaO2 to form CaO and O2. This decomposition
is endothermic as indicated by a negative DSC peak. The endother-
mic peak reaches a minimum at 425 ◦C. The mass loss caused by

404
S. Lin, J. Selig / Journal of Alloys and Compounds 503 (2010) 402–409
Fig. 2. TG/DSC of CaO2 in oxygen heated at a rate of 50 ◦C/min.
Fig. 4. TG/DSC of Co in oxygen heated at a rate of 50 ◦C/min.
decomposition of CaO2 is 15%, which agrees for sample initially
containing 65% CaO2 and 35% of Ca(OH)2 as reported by the man-
ufacturer. Another noticeable drop in the TG measurement occurs
at 702 ◦C indicates the decomposition of Ca(OH)2 to form CaO and
water. Again, this decomposition is endothermic with a negative
DSC slope change. Reactions (3) and (4) show the decompositions
of the CaO2 and Ca(OH)2 in O2 during the SHS reaction.
NaCl(l) → NaCl(g) (≥ 950 ◦C)
(7)
TG curve for cobalt metal heated at a rate of 50 ◦C/min in O2
(Fig. 4) shows oxidation between 450 ◦C and 998 ◦C followed by a
decomposition. Several small peaks on DSC curve between 450 ◦C
and 998 ◦C suggest that the oxidation of Co to Co3O4 is a multiple
step reaction. The first exothermic peak on DSC curve is observed
at 520 ◦C and corresponds to the formation of CoO and second peak
is observed at 620 ◦C and corresponds to the formation of Co3O4.
The mass gain during the oxidation was approximately 33% which
is in agreement with complete oxidation of cobalt to Co3O4. After
oxidation, a sudden drop in mass is observed above 1000 ◦C. This
drop is concurrent with a large endothermic peak on DSC curve
at 1038 ◦C. This mass drop is 8%, which represents decomposition
of Co3O4 to CoO and oxygen. Oxidation of cobalt and its subse-
quent decomposition can be expressed in series of three reactions
(8)–(10).
CaO2 → CaO + (1/2)O2 (425 ◦C)
Ca(OH)2 → CaO + H2O (702 ◦C)
(3)
(4)
Sodium perchlorate which is used as a solid oxidizer in some
experiments decomposes exothermically to oxygen and sodium
chloride at 594 ◦C when heated at 50 ◦C/min in O2 (Fig. 3). Before
decomposition, NaClO4 melts as indicated by the endothermic peak
at 504 ◦C. Sodium chloride melts at 832 ◦C and begins to vaporize at
950 ◦C. Vaporization of NaCl is complete at 1180 ◦C. Because NaCl
vaporize at a relatively low temperature the final product of SHS
reaction should be NaCl free. Any leftover NaCl can be easily washed
away by warm DI water. Reactions (5)–(7) show the decomposi-
tion of NaClO4 during the SHS reaction and subsequent melting
and evaporation of NaCl.
Co + (1/2)O2 → CoO (450–520 ◦C)
CoO + (1/6)O2 → (1/3)Co3O4 (520–998 ◦C)
(1/3)Co3O4 → CoO + (1/6)O2 (1038 ◦C)
(8)
(9)
(10)
NaClO4 → NaCl + 2O2 (594 ◦C)
NaCl(s) → NaCl(l) (832 ◦C)
(5)
(6)
From reactions (3) and (5), it is clear that the oxygen from the
decomposed CaO2 and NaClO4 can be used for oxidization of Co
metal immediately as designed for this SHS reaction. Because of
the oxygen from decomposition has a higher temperature than that
from the surrounding gas phase it increases the SHS reaction rate.
For sample containing mixture of reactants heated at 50 ◦C/min
three main reaction steps are observed during the analysis (Fig. 5).
The first step is the decomposition of CaO2 to CaO and O2 between
384 ◦C and 460 ◦C which is similar to the thermal behavior of CaO2
alone (Fig. 2). The second step is Co oxidation and subsequent
Ca1.24Co1.62O3.86 formation, between 460 ◦C and 988 ◦C. It is clear
that cobalt oxidation starts at a lower temperature when CaO2 is
present, which proves CaO2 is acting as an oxidizer as well as filler
in our reaction. A sudden slope change of TG curve around 690 ◦C
which is the same Ca(OH)2 decomposition observed in Fig. 5. How-
ever, no net weight loss was detected. The DSC curve reaches a
maximum at 754 ◦C and the slope of TG curve slowly decreases after
that temperature. This suggests that the reaction slows down once
the temperature is lower than 754 ◦C. The last major step is the
decomposition of the Ca1.24Co1.62O3.86/Co3O4 mixture beginning
at 998 ◦C to form CaO, CoO and O2. Sample is fully decomposed at
1035 ◦C. The endothermic decomposition is indicated by a negative
DSC peak at 1015 ◦C.
Fig. 3. TG/DSC of NaClO4 in oxygen heated at a rate of 50 ◦C/min.

S. Lin, J. Selig / Journal of Alloys and Compounds 503 (2010) 402–409
405
Fig. 7. Proposed reaction mechanism of 1.24CaO2 + (1.62 − 3x)Co + xCo3O4
0.18NaClO4 + (0.33 − 2x)O2 → Ca1.24Co1.62O3.86
+
.
a larger mass increase between 450 ◦C and 950 ◦C when the sam-
ple was heated at 5 ◦C/min which suggests more product is formed
during the slow heating analysis.
Fig. 5. TG/DSC of a mixture of CaO2 and Co (Ca:Co = 1.24:1.62) in oxygen heated at
a rate of 50 ◦C/min.
TG/DSC analysis suggests that during the SHS reaction, six major
reaction steps occur. The reaction mechanism of reaction (2) is pro-
posed in Fig. 7. The SHS reaction starts with the decomposition
of calcium peroxide to form calcium oxide and oxygen at 425 ◦C.
At the same time, cobalt metal begins to oxidize at 450 ◦C which
results in large amount of heat release. The oxidation of cobalt is
the main reaction that results in reaction front propagation. Oxi-
dation of cobalt occurs in two steps – formation of CoO at 520 ◦C
followed by the formation of Ca3O4 at 620 ◦C. As Co3O4 is formed
it reacts with calcium oxide to form Ca1.24Co1.62O3.86. When tem-
perature reaches 594 ◦C sodium perchlorate decomposes to sodium
chloride and oxygen which accelerates the oxidation of cobalt. At
702 ◦C calcium hydroxide decomposes to water and forms addi-
tional calcium oxide which also reacts with Co3O4. If a too high
temperature is reached during SHS reaction Ca1.24Co1.62O3.86 may
decompose back to CoO and CaO. The dominant reaction steps for
Ca1.24Co1.62O3.86 formation are summarized in Table 1.
If reaction (2) reaches a 100% conversion a 10% increase in
mass should be observed. However, our TG results in Fig. 5
show a significantly less mass increase which suggests the con-
version of reactants to Ca1.24Co1.62O3.86 is not complete during
the thermal analyses. The product at the highest conversion
(highest mass of the TG curve) at 998 ◦C is most likely a mix-
ture of Ca1.24Co1.62O3.86, CaO, and Co3O4. Oxidation of cobalt
metal and formation of Ca1.24Co1.62O3.86 can be confirmed by the
decomposition at 998 ◦C (mass decrease in the TG curve). Simi-
lar decomposition of Ca1.24Co1.62O3.86 was reported at 956 ◦C [13].
However, because the decompositions of calcium cobalt oxide and
Co3O4 occur at close temperature ranges it is not clear from the
TG/DSC results how much of Co3O4 or Ca1.24Co1.62O3.86 is present
at 998 ◦C before the decompositions take place.
Fig. 6 shows the TG/DSC results when different heat rates were
used in analyzing the reactant mixture. It can be seen when a faster
heating rate (50 ◦C/min) was used, the reaction temperature of each
step shifted to a higher temperature due to the larger temperature
gradient inside the sample as predicted by Kissinger [21]. In addi-
tion, a slower heating rate (5 ◦C/min) results lower amplitude of
peaks on DSC curve (Fig. 6a) and decreases temperature ranges for
phase transitions and reactions by approximately 25 ◦C. A signif-
icant difference in TG results is also observed (Fig. 6b). There is
3.2. SHS of Ca1.24Co1.62O3.86
The thermodynamic properties [22] of reactants and products
of reaction (2) were listed in Table 2 to calculate the reaction heat
and the adiabatic temperature. Eq. (11) was used to calculate the
Fig. 6. TG of a mixture of CaO2 and Co (Ca:Co = 1.24:1.62) in oxygen (a) heated at a rate of 5 ◦C/min and (b) heated at a rate of 50 ◦C/min.

406
S. Lin, J. Selig / Journal of Alloys and Compounds 503 (2010) 402–409
Table 1
Summary of reaction steps during SHS of Ca1.24Co1.62O3.86
.
Reactions
ꢃH (kJ/mol)
Temperature range (◦C)
1.24CaO2 → 1.24CaO + 0.62O2 (65%)
1.24Ca(OH)2 → 1.24CaO + 1.24H2O (35%)
0.18NaClO4 → 0.09O2 + 0.18NaCl
19.9
92.3
−28.4
−237.7
−177.9
−456.5
384–460
690–702
594
1.62Co + 0.81O2 → 1.62CoO
450–520
520–998
650–800 [14]
≥998
1.62CoO + 0.27O2 → 0.54Co3O4
1.24CaO + 1.62Co3O4 + 0.23O2 → Ca1.24Co1.62O3.86
Ca1.24Co1.62O3.86 → 1.24CaO + 1.62Co3O4 + 0.23O2
1.62Co2O3 → 1.62Co + 0.27O2
≥1038
or Ca1.24Co1.62O3.86 intermediate compounds of a different lattice
constant or structure than the reported one [8].
Fig. 9 shows the temperature history at the centerline of the
reaction pellet during the SHS process (reaction (2)). The tem-
perature increases from room temperature to 1200 ◦C in 65 s and
then cools down below 650 ◦C (the minimum temperature for
Ca1.24Co1.62O3.86 formation) in 200 s. As seen from the TG/DSC anal-
yses in Fig. 6, even at a much slower heating rate (5 ◦C/min), the
reactants do not have enough time to form the desired product
completely before temperature reaches the decomposition tem-
perature of Ca1.24Co1.62O3.86 (1038 ◦C). The maximum temperature
reached during the SHS is 1200 ◦C, which is 160 ◦C higher than the
decomposition of Ca1.24Co1.62O3.86. However, the duration of tem-
perature above the decomposition temperature is only 60 s, which
is not sufficient for Ca1.24Co1.62O3.86 to decompose completely but
instead a mixture of undesired Ca–Co–O may be formed. A temper-
ature plateau at 1175 ◦C lasting for 20 s in Fig. 9 is caused by the
vaporization of sodium from NaClO4 decomposition.
Fig. 8. XRD pattern of product of SHS reaction. Reference peaks for Ca1.24Co1.62O3.86
are taken from Itahara and Tani [8].
adiabatic temperature.
ꢁ
ꢃ
ꢀ
T
ad
ꢂ
niCpi
dT = −ꢃHr
(11)
298
i
3.3. Post-treatment of SHS products
In this equation ni and Cpi are the stoichiometric coefficient and
specific heat of product (Ca1.24Co1.62O3.86 and NaCl). To simplify the
calculation a heat capacity of Ca1.24Co1.62O3.86 (220 J/mol K) esti-
mated from that of CaO and Co3O4 and specific heat of NaCl at
1500 K are used. Values of specific heats are in Table 2. The calcu-
lated adiabatic temperature is 2265 K.
In our previous work [23], it has been found that SHSed powder
can be converted to Ca1.24Co1.62O3.86 by sintering at 850 ◦C for 2 h in
air. This suggests that the product obtained after SHS is a kinetically
preferred metastable compound and can be converted to a ther-
modynamically preferred product (Ca1.24Co1.62O3.86) by sintering
at a temperature lower than its decomposition temperature. Dur-
ing the SHS, reaction pellet heats up rapidly to almost 1200 ◦C and
it takes approximately 5 min for the sample to cool down below
450 ◦C (Fig. 9). No reaction should occur bellow 450 ◦C according
to TG/DSC analyses. This fast heating and cooling processes may
not enable the sample to achieve the thermodynamically preferred
state but some metastable compound is formed.
Ignition of the sample occurs in about 2 s after the igniter
is turned on. Reaction front propagates through the pellet from
the ignition region towards the unreacted remainder of the pel-
let. Sometimes, a pulsating reaction front movement is observed.
Pulsating reaction front often indicates reaction instability or the
reaction front movement is barely self-sustainable. If NaClO4 solid
oxidizer is used a white smoke (vaporized NaCl) is observed dur-
ing the reaction. After the reaction front propagates through the
whole pellet, the center part of sample glows for several minutes
before slowly cools down. During the SHS reaction, the color of the
pellet changes from light grey to dark grey. Synthesized sample is
porous and easy to grind. As-synthesized samples were analyzed by
XRD to identify their phases (Fig. 8). No detectable amount of unre-
acted reactants nor cobalt oxides or calcium oxide is found by XRD
in the product. The XRD pattern shows unidentified peaks which
are believed to be calcium cobalt oxide of unknown stoichiometry
Table 2
Heats of formation and heat capacities of reactants and products in the proposed
SHS reaction.
Compound
ꢃHf (kJ/mol)
Cp (J/mol K)
Co
0
CaO2
NaClO4
CaO
Ca3O4
Ca1.24Co1.62O3.86
NaCl
−654.8
−382.6
−382.6
56
228
−382.6
−1268.4 (estimated)
−411.1
220 (estimated)
75
Fig. 9. Temperature history of SHS reaction at the center of reactant pellet.

S. Lin, J. Selig / Journal of Alloys and Compounds 503 (2010) 402–409
407
Fig. 10. XRD patterns of SHS powder (a) as-synthesized, (b) sintered at 850 ◦C for
30 min in O2, (c) sintered at 850 ◦C for 60 min in O2, and (d) sintered at 850 ◦C for
90 min in O2. Reference peaks for Ca1.24Co1.62O3.86 are taken from Itahara and Tani
[8].
To understand and improve the post-treatment process, sam-
ples were heated in a tube furnace in an oxygen or air atmosphere
to study the impact of sintering time on the final product. Fig. 10
shows XRD patterns of samples post-treated in a tube furnace for
various time periods at 850 ◦C in oxygen. It can be seen from Fig. 10b
that after sintering at 850 ◦C in oxygen for 30 min, desired product
(Ca1.24Co1.62O3.86) starts to form, although there is an unidenti-
fied XRD peak at 2ꢄ = 32.5◦. Samples sintered for 60 and 90 min in
the same condition have the same XRD patterns and contain only
Ca1.24Co1.62O3.86 (Fig. 10c and d). It can be concluded that highly
pure Ca1.24Co1.62O3.86 can be obtained by a 1 h post-treatment at
850 ◦C in oxygen.
X-ray diffraction of as-synthesized products post-treated in a
tube furnace at 850 ◦C in air shows that a 15-min post-treatment
in air can produce pure Ca1.24Co1.62O3.86 (Fig. 11b). There is no
change in composition when post-treatment time is increased to
30 and 60 min (Fig. 10c and d). These results suggest that powders
obtained from our SHS are very good precursor for bulk thermoelec-
tric devices and can be easily converted to pure Ca1.24Co1.62O3.86
during a hot pressing process.
Fig. 12. TG curves of Ca1.24Co1.62O3.86 heated at a rate of 10 ◦C/min (a) Al2O3 blank,
(b) in N2, (c) in air, and (d) in O2.
For sample analyzed in N2 the mass decreases 7% of its ini-
tial mass between 850 ◦C and 1000 ◦C indicates the decomposition
of Ca1.24Co1.62O3.86 to CaO, CoO, and O2. The decomposition of
Ca1.24Co1.62O3.86 occurs in two steps with first step starting at
850 ◦C and second step beginning at 920 ◦C and finishing at 1000 ◦C.
Similar decomposition in air and oxygen atmospheres occurs
between 950 ◦C and 1100 ◦C and between 1020 ◦C and 1200 ◦C
respectively. These results suggest that the decomposition temper-
ature and thus the stability of Ca1.24Co1.62O3.86 is increased with
increasing amount of partial pressure of oxygen.
3.5. Activation energy
Activation energy of the SHS reaction can be calculated from
the maximum temperatures and the velocities of the reaction front
movement when different amounts of diluent are added to the SHS
reaction system using the equation below [24,25]:
3.4. Thermal stability of Ca1.24Co1.62O3.86 synthesized by SHS
ꢄ
ꢅ
−E
u = k exp
(12)
TG analysis was conducted to study the thermal stability of
the product. Fig. 12 shows the results of TG curves of pure
Ca1.24Co1.62O3.86 (SHS followed by a 10 h 850 ◦C post-treatment) in
N2, O2, and air separately. Samples were placed above a thin layer
of alumina to prevent possible reaction between the product and
sample crucible. When a high enough temperature (at least 900 ◦C,
depending on atmosphere) is reached, Ca1.24Co1.62O3.86 decom-
poses to CaO, CoO, and O2. The release of oxygen results is 7.7%
mass loss.
2RT
where u is reaction front propagation speed, k is a pre-exponent, E
is the activation energy, R is the gas constant, and T is the maximum
combustion temperature reached during SHS reaction. The maxi-
mum combustion temperature and the reaction front movement
velocity can be calculated from the temperature history measured
at two points on the centerline of the reaction pellet. Activation
energy is calculated from the slope of the straight line in the Arrhe-
nius plot. Fig. 13 is the Arrhenius plot of reaction front movement
and maximum combustion temperature for x = 0, 0.025, 0.05, and
0.075 dilution is used. From Fig. 13, the activation of reaction (2) is
calculated to be 71 kJ/mol.
3.6. Thermoelectric properties
SHS powders followed by a 15 min post-treatment were hot
pressed and analyzed for their thermoelectric properties in a tem-
perature range of 25–500 ◦C. Fig. 14 shows the Seebeck coefficients
of our sample. SHS sample has an almost constant Seebeck coeffi-
cient of 165 ␮V/K in this temperature range which is higher than
the reported data of samples prepared by other processes includ-
ing conventional solid-state sintering (CS) [14] and spark plasma
sintering (SPS) [14,15]. However, the electrical conductivity of our
sample is lower than those prepared by spark plasma sintering but
higher than the reported data from solid-state sintering (Fig. 15).
Fig. 11. XRD patterns of SHS powder (a) as-synthesized, (b) sintered at 850 ◦C for
15 min in air, (c) sintered at 850 ◦C for 30 min in air, and (d) sintered at 850 ◦C for
60 min in air. Reference peaks for Ca1.24Co1.62O3.86 are taken from Itahara and Tani
[8].

408
S. Lin, J. Selig / Journal of Alloys and Compounds 503 (2010) 402–409
Fig. 13. Arrhenius plot of experimental data used to calculate activation energy of
SHS of Ca1.24Co1.62O3.86
Fig. 16. Power factor compared with Ca3Co4O9 prepared by different methods
[11,12].
.
Fig. 17. ZT values compared with Ca3Co4O9 prepared by different methods [12].
Fig. 14. Seebeck coefficient compared with Ca3Co4O9 prepared by different meth-
ods [11,12].
Because of the lower thermal conductivity, the thermal power of
our sample is also lower than those prepared by spark plasma sin-
tering process (Fig. 16). The room temperature thermal diffusivity
of our sample (0.0088 cm2/s) is used to calculate the figure of merit.
From Fig. 17, it can be seen that SHS powders consolidated by hot
press has figure of merit close to samples prepared by spark plasma
sintering process.
4. Conclusion
In this work Ca1.24Co1.62O3.86 was successfully synthesized
by SHS followed by a short post-treatment at 850 ◦C. Powders
obtained from SHS reaction contain an unidentified intermediate
compound and can be easily converted to Ca1.24Co1.62O3.86 by a
15 min post-treatment in air. The hot pressed SHS powders have
a high Seebeck coefficient and a figure of merit close to sam-
ples prepared by spark plasma sintering process. Thermal analyses
(TG/DSC) were used to study the reaction network. TG/DSC analysis
of pure Ca1.24Co1.62O3.86 suggests that the decomposition temper-
ature increases with oxygen partial pressure. Temperature history
and velocity of combustion front movement were measured. The
activation energy of the SHS reaction is calculated to be 75 kJ/mol.
Fig. 15. Electrical conductivity compared with Ca3Co4O9 prepared by different
methods [11,12].

S. Lin, J. Selig / Journal of Alloys and Compounds 503 (2010) 402–409
409
Acknowledgements
[8] H. Itahara, T. Tani, R&D Rev. Toyota CRDL 39 (2004) 63–70.
[9] S. Tajima, T. Tani, S. Isobe, K. Koumoto, Mater. Sci. Eng. B 86 (2001) 20–25.
[10] A. Kikuch, L. Zhang, N. Okinaka, T. Tosho, T. Akiyama, Mater. Trans. 50 (2009)
2675–2679.
[11] Y. Liu, Y. Lin, Z. Shi, C. Nan, Z. Shen, J. Am. Ceram. Soc. 88 (2005) 1337–1340.
[12] Y. Zhang, J. Zhang, J. Mater. Process. Technol. 208 (2008) 70–74.
[13] S. Li, R. Funahashi, I. Matsubara, K. Ueno, H. Yamada, J. Mater. Chem. 9 (1999)
1659–1660.
[14] A.C. Masset, C. Michel, A. Maignan, M. Hervieu, O. Toulemonde, F. Studer, B.
Raveau, J. Hejtmanek, Phys. Rev. B 62 (2000) 167–175.
[15] J. Sugiyama, A. Mystery in Thermoelectric Layered Cobaltites, R&D Rev. Toyota
CRDL 39 (2004) 50–62.
This work was partially supported by Lamar University Research
Enhancement Grants under project no. CM1006 and Texas Haz-
ardous Waste Research Center (THWRC) under contract no.
068LUB0969. The authors would like to express their special thank
to Dr. Hua-Tay Lin and Dr. Hsin Wang of the Oak Ridge National
Laboratory for measuring the thermoelectric properties during the
feasibility test. The authors also wish to thank Dr. Andrew Gomes,
Mr. Doanh Tran, and Ms. Lauren Griffin for assistance in sample
analyses and helpful discussion.
[16] Q. Li, Mater. Res. Soc. Symp. Proc. 886 (2006), 0886-F01-05.1–0886-F01-05.12.
[17] M. Mikami, E. Guilmeau, R. Funahashi, K. Chong, D. Chateigner, J. Mater. Res. 20
(2005) 2492–2497.
[18] A.S. Mukasyan, K.S. Martirosyan, Combustion of Heterogeneous Systems: Fun-
damentals and Applications for Materials Synthesis, Transworld Research
Network, Kerala, India, 2007.
References
[19] A. Varma, G. Cao, M. Morbidelli, AIChE J. 36 (1990) 1032–1038.
[20] S.C. Lin, J.T. Richardson, D. Luss, Physica C 260 (1996) 321–326.
[21] H.E. Kissinger, Anal. Chem. 29 (1957) 1702–1706.
[22] C.L. Yaws, Yaws’ Handbook of Thermodynamic Properties for Hydrocarbons
and Chemicals. Knovel. http://knovel.com/web/portal/browse/display? EXT
KNOVEL DISPLAY bookid=2380&VerticalID=0.
[1] U.S. Energy Flowchart 2008, Lawrence Livermore National Laboratory. https://
publicaffairs.llnl.gov/news/energy/energy.html.
[2] D.M. Rowe (Ed.), CRC Handbook of Thermoelectrics, CRC Press LLC, Boca Raton,
FL, 1995.
[3] G.A. Slack, in: D.M. Rowe (Ed.), CRC Handbook of Thermoelectrics, CRC Press,
Boca Raton, FL, 1995, p. 407.
[23] S. Lin, J. Selig, J.A. Gomes, and D. Tran, The Materials Science & Technology
2009 Conference and Exhibition proceedings (2009) 241-252, ISBN 13: 978-1-
61503r-r006-4.
[4] G.J. Snyder, E.S. Toberer, Nat. Mater. 7 (2008) 105–114.
[5] R. Ray, A. Ghoshray, K. Ghoshray, S. Nakamura, Phys. Stat. Sol. (b) 215 (1999)
703–707.
[24] A.B. Zeldovich, The Mathematical Theory of Combustion and Explosions, Con-
sultants Bureau, New York, NY, 1985 (translated from Russian).
[25] K.K. Kuo, Fundamentals of Solid-Propellant Combustion, American Institute of
Aeronautics and Astronautics, New York, NY, 1984.
[6] G. Mahan, Proceeding of Int. Conference on Thermoelectrics, Dresden, Ger-
many, 1997.
[7] G. Peleckis, T. Motohashi, M. Karppinen, H. Yamauchi, Appl. Phys. Lett. 83 (2003)
5416.