Conditions for the Electrochemical Synthesis of the CoPn3 (Pn = P, As, Sb) Skutterudites

Article abstract of DOI:10.1149/1.1389342

The binary pnictide skutterudites, CoPn₃ (Pn = P, As, Sb), have been successfully synthesized for the first time by molten salt electrolysis. The melt for depositing CoP₃ contained CoO and (NaPO₃)₃, and graphite

Full text of DOI:10.1149/1.1389342

Journal of The Electrochemical Society, 148 ͑9͒ D109-D111 ͑2001͒
D109
0013-4651/2001/148͑9͒/D109/3/$7.00 © The Electrochemical Society, Inc.
Conditions for the Electrochemical Synthesis
of the CoPn3 „Pn ϭ P, As, Sb… Skutterudites
R. C. DeMattei,a A. Watcharapasorn,b and R. S. Feigelson,a,b
aGeballe Laboratory for Advanced Materials, bDepartment of Materials Science and Engineering, Stanford
University, Stanford, California 94305, USA
The binary pnictide skutterudites, CoPn3 (Pn ϭ P, As, Sb͒, have been successfully synthesized for the first time by molten salt
electrolysis. The melt for depositing CoP3 contained CoO and (NaPO3͒3, and graphite rods were used for both anode and cathode.
All experiments were run in a flowing N2 gas atmosphere. By adjusting the melt composition, single phase CoP3 could be
prepared. Attempts were also made to deposit CoP3 using platinum and silver strip electrodes. Although CoP3 powder was
obtained in the crucible, no CoP3 deposit was found on the Pt cathode, which was attacked by the phosphorus present in the melt
and produced platinum phosphide (PtP2). A low deposition rate was observed for Ag cathodes, and as a result, a very small
amount of material was deposited. This made it difficult to determine the type and number of phases formed. Using graphite
electrodes, two other cobalt skutterudites were prepared. CoAs3 was synthesized from a CoO ϩ NaAsO2 melt and CoSb3 from a
CoO ϩ Na2CO3 ϩ Sb2O3 melt. While the skutterudite phases were formed in these systems, the optimal melt compositions and
other experimental conditions necessary to prepare single phase deposits still need to be determined.
© 2001 The Electrochemical Society. ͓DOI: 10.1149/1.1389342͔ All rights reserved.
Manuscript received February 21, 2001. Available electronically August 9, 2001.
Skutterudite compounds have been identified as potential candi-
dates for thermoelectric applications.1,2 One of the target materials
within this class is CoPn3 (Pn ϭ P, As, Sb͒ and its derivative com-
pounds obtained by cation and/or anion substitution. The common
methods for synthesizing bulk quantities of the cobalt skutterudites
are either direct synthesis from the elements ͑reaction temperature
cathode compartments. All experiments were conducted under a
flowing nitrogen atmosphere at temperatures ranging from 650 to
700°C, depending on the type of melts under investigation. Current
vs. voltage ͑I-V͒ plots were made before each deposition in order to
be sure that the deposition potential was above the last plateau. The
applied voltages ranged from 3 to 5 V with currents as high as 1 A.
The products were recovered by dissolving the melts in a recirculat-
ing water stream.
3-8
range ϳ600 to 1000°C͒ or solution growth, usually from molten
tin for cobalt triphosphide or antimony for cobalt triantimonide ͑pro-
cessing temperature range ϳ780 to 1000°C͒.8-10 Both of these tech-
niques require extended reaction time ͑up to 2 weeks͒ and in the
case of the phosphides and arsenides involve high vapor pressure of
group V components. The tin solution process requires acid etching
to remove the tin. In addition, the presence of tin in the skutterudite
lattice and its role in the thermoelectric performance of the material
is not known.
The reactions taking place at the electrodes during electrolysis
can be described by the equations
Mϩ2 ϩ 2eϪ → M cathode
POϪ3 ϩ 5eϪ → P ϩ 3 OϪ2 cathode
Mϩ2 ϩ 3POϪ3 ϩ 17eϪ → MP3 ϩ 9 OϪ2 overall cathode
OϪ2 → 1/2 O2 ϩ 2eϪ anode
A possible alternative to direct synthesis or solution growth of
the pnictide skutterudites is electrochemical synthesis. Chene11 elec-
trochemically synthesized a series of cobalt-phosphorous com-
pounds from melts containing cobalt oxide and sodium metaphos-
phate. However, he did not report synthesizing CoP3. Other
investigators have reported on the electrochemical synthesis of some
2Mϩ2 ϩ 6POϪ3 → 2MP3 ϩ 17/2 O2 ϩ OϪ2 overall
The potential ͑E͒ for driving the overall reaction can be ex-
pressed by the Nernst equation
Ϫ
6
8.5
2
E ϭ ⌬E0 Ϫ ͑RT/34F͒ ln͓͑O2͔ ͓OϪ2͔/͓Mϩ2͔ ͓PO3 ͔ ͒
technologically important materials such as GaP12,13 and GaAs.14
These earlier results led to the current investigation.
where ⌬E0 is the difference in standard electrode potentials and
the oxygen is included since its partial pressure ͑or more pro-
perly fugacity͒ is less than 1 atm. Unfortunately, the standard elec-
Electrochemical synthesis offers several advantages over direct
synthesis and solution growth techniques. First, it is a relatively low
temperature technique and thus avoids problems of thermal decom-
position and excessive vapor pressure. Second, it is relatively unaf-
fected by small changes in temperature. Third, the technique does
not require that either the starting material or the final product be
congruently melting. ͑CoPn3 compounds are peritectics.15,16͒ Fourth,
the total reaction time is 2 to 4 days. Finally, electrochemical syn-
thesis is controlled by electrical parameters, which can be more
precisely controlled than temperature.
This investigation represents the first electrochemical synthesis
of the cobalt pnictide skutterudites. The conditions for the synthesis
of these compounds are reported.
Experimental
All electrodepositions were performed in a closed Inconel 600
cylinder fitted with a removable cover sealed with an O-ring. The
cover had fittings for two electrodes ͑anode and cathode͒, a gas inlet
and outlet, a thermocouple, and a view port ͑Fig. 1͒. The cylinder
was externally heated by a resistance furnace controlled by a pro-
portional integral differential controller. The melts were contained in
a graphite crucible fitted with a baffle to separate the anode and
Figure 1. Schematic drawing of the experimental setup for electrochemical
synthesis of CoPn3 (Pn ϭ P, As, and Sb͒ skutterudites.
Downloaded on 2014-11-30 to IP 138.251.14.35 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

D110
Journal of The Electrochemical Society, 148 ͑9͒ D109-D111 ͑2001͒
Figure 3. Experimental I-V curves for deposition of CoAs3 and CoSb3 using
graphite electrodes. An I-V curve for CoP3 is also shown for comparison.
Figure 2. Experimental I-V curves for the CoO ϩ Na͑PO3͒3 melt using dif-
ferent cathode/anode pairs for the synthesis of CoP3.
changes that lead to the formation of CoP3. When a lower concen-
tration of sodium trimetaphosphate was used, the phases present in
the products were Co2P and CoP. Increasing the amount of sodium
trimetaphosphate resulted in mixtures containing various ratios of
CoP3 and CoP. The final melt composition used was 99.7 mol %
metaphosphate and 0.3 mol % Coϩ2. Electrolysis of this melt com-
position yielded phase pure CoP3 with slight traces of graphite con-
tamination. An X-ray diffraction ͑XRD͒ pattern of the electrochemi-
cally synthesized CoP3 is shown in Fig. 4a.
Preliminary studies of the deposition of a CoP3 film on platinum
sheet yielded a dark gray deposit on both sides of the cathode. These
areas were very brittle, and X-ray analysis showed that they con-
sisted mostly of the compound platinum phosphide (PtP2). No sign
of CoP3 was observed on the cathode. A portion of the material
retrieved from the crucible, however, contained the skutterudite
phase. It seems that Pt, similar to graphite, does not provide a good
surface for cobalt triphosphide adhesion. When the Ag cathode was
used, a very small amount of material was deposited. The low depo-
sition rate corresponds to the low current observed in the I-V curve
͑Fig. 2͒. Preliminary energy dispersive X-ray analysis indicated that
there were different Co and P ratios in the deposits but the actual
phases formed have yet to be determined.
The results with CoP3 are rather interesting in light of Chene’s
work.11 He reported obtaining CoP2 as one of the products of his
electrochemical studies and there is also an XRD pattern for this
compound from another source. Even though compounds ranging
from Co2P to CoP3 were electrochemically synthesized in the cur-
rent experiments, there was no evidence for the formation of the
CoP2 phase. There are several reasons that may explain these dis-
crepancies. First, the voltage and current used in Chene’s work were
much higher than those in our experiments, and such a large driving
force may shift the experimental conditions very far from equilib-
rium conditions, thereby leading to metastable phase formation. Sec-
ond, Chene used chemical analysis rather than XRD to determine
trode potentials for the molten salt systems used for skutterudite
deposition were not known and, therefore, the Nernst equation could
not be used to determine either the reactant concentrations for a
given potential or the potential to apply for a given concentration of
reactant species. As a result, the proper metal to metaphosphate ratio
must be determined empirically. This is one of the few disadvan-
tages of the molten salt electrochemical synthesis method.
The following are the experimental conditions for each of binary
pnictide skutterudites.
CoP3.—The results of Chene’s investigations11 were used to plot
the ratio of PO3 ͑metaphosphate͒ to Coϩ2 in the melt vs. the ratio of
P to Co in the product. This was then extrapolated to a P:Co ratio of
3:1 which gave a starting point for a melt composition of 93.6:1.0
(POϪ3 :Coϩ2). Electrolysis was carried out using a graphite anode
and cathode in a melt containing CoO and (NaPO3͒3 ͑sodium tri-
metaphosphate͒. The initial result was a mixture of cobalt phosphide
compounds. A systematic study of the effects of melt composition
on the number and stoichiometry of the phases present in the syn-
thesized product was used to find the optimum melt composition for
the production of single phase CoP3. Attempts were also made to
deposit CoP3 on platinum and silver strip electrodes using the same
melt composition. In these experiments, a Pt anode was used. Figure
2 shows experimental I-V plots for these melts with different elec-
trode materials.
CoAs3.—There is no commercially available metaarsenate, but
experience with GaAs14 demonstrated that arsenides can be elec-
trodeposited from melts containing sodium metaarsenite (NaAsO2).
The only melt that was tried for the synthesis of CoAs3 contained 95
mol % metaarsenite. An I-V curve for this melt system with graphite
electrodes is shown in Fig. 3.
CoSb3.—Initially, we were not able to find any commercially
available metaantimonate or metaantimonite so the melt was pre-
pared from sodium carbonate (Na2CO3) and antimony oxide
(Sb2O3) which would react to form the antimonite. The melt was
nominally 95% metaantimonite and, like the CoAs3 experiment, was
electrolyzed using graphite electrodes. The I-V characteristics for
the deposition of CoSb3 are plotted in Fig. 3 along with that of CoP3
and CoAs3 experiments.
Table I. Products vs. melt composition for the Co-P system.
Amount
͑%, estimated
from X-ray͒
Melt composition
Average P content
͑mol % POϪ3 ͒
Products
͑atom %͒
97.6
99.0
99.3
99.7
5
95
50
50
12
88
100
48.8
66.7
73.4
75.0
Co2P
CoP
CoP
CoP3
CoP
CoP3
CoP3
Results and Discussion
CoP3.—The synthesized materials did not adhere well to the
graphite electrodes and the products were recovered from the solidi-
fied melts in the form of a black powder and tiny crystalline par-
ticles. Table I shows the results of the systematic melt composition
Downloaded on 2014-11-30 to IP 138.251.14.35 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

Journal of The Electrochemical Society, 148 ͑9͒ D109-D111 ͑2001͒
D111
CoSb3.—The products of the electrolysis of the nominally 95
mol % metaantimonate were metallic antimony ͑70%͒ and CoSb3
͑29%͒. Again, the melt composition needs to be adjusted to produce
phase pure material. Changing to another source of antimony, such
as potassium hexahydroxyantimonate, may also be useful.
The above results show that all three pnictide skutterudites can
be synthesized by molten salt electrolysis. While cobalt triphosphide
was obtained as a single phase, the arsenide and antimonide have
not yet been obtained in pure form. The principal barrier to obtain-
ing these materials as a single phase is the adjustment of the melt
composition.
Conclusions
It was demonstrated for the first time that the binary skutterudite
compounds, CoPn3, could be synthesized electrochemically. How-
ever, there is some anomalous behavior in the electrodeposition of
the phosphorous compounds that remains to be understood. Using
platinum metal as electrodes for the electrodeposition of CoP3 did
not work well because it reacted with phosphorous to form a plati-
num phosphide phase. Ag cathodes gave poor deposition rates under
the experimental condition investigated, and resulted in a very small
amount of the synthesis products. Running more experiments at
longer time will be useful to determine whether Ag is a suitable
electrode material. In the case of CoAs3 and CoSb3, the optimum
melt compositions for the synthesis of phase pure compounds need
to be determined.
There are other interesting areas of research that may follow
from these results. These include the synthesis of pure or doped
mixed pnictide skutterudites, and the electrodeposition of thin films
on suitable substrates. The latter would greatly facilitate the study of
their thermoelectric properties.
Figure 4. XRD patterns of electrochemically synthesized skutterudites: ͑a͒
CoP3, ͑b͒ CoAs3, and ͑c͒ CoSb3.
Acknowledgments
This research was supported by the office of Naval Research
through Grant no. N00014-97-1-0524 and was carried out in the
facilities of the Geballe Laboratory for Advanced Materials ͑former
Center for Materials Research͒ at Stanford University.
the composition of the synthesis products. Since the former tech-
nique only gives the relative amount of elemental constituents, it
could not have been used to distinguish between the pure CoP2
phase or a 50:50 mixture of CoP3 and CoP. In our experiments, the
XRD patterns did not reveal any sign of CoP2 formation and it is
possible that this phase may not form under normal experimental
conditions.
The difficulty in obtaining CoP2 was also mentioned by
Donohue17 who reported the presence of CoP2 phase only when
prepared at high pressure. Since the data in the Co-P phase
diagram18 are incomplete, especially in the high phosphorus region,
it cannot be precisely determined whether CoP2 is a thermodynami-
cally stable phase. It is possible, however, that this compound may
form peritectically under certain conditions, similar to the behavior
of the di- and tripnictides compounds in Co-As15 and Co-Sb16 sys-
tems
Stanford University assisted in meeting the publication costs of this ar-
ticle.
References
1. B. C. Sales, D. Mandrus, and R. K. Williams, Science, 272, 1325 ͑1996͒.
2. J-P. Fleurial, T. Caillat, A. Borshchevsky, D. T. Morelli, and G. P. Meisner, in
Proceedings of the 15th International Conference on Thermoelectrics, p. 91, Insti-
tute of Electrical Engineers ͑1996͒.
3. W. Biltz and M. Heimbrecht, Z. Anorg. Allg. Chem., 241, 349 ͑1939͒.
4. S. Rundqvist and N.-O. Ersson, Ark. Kemi, 30, 103 ͑1968͒.
5. H. D. Lutz and G. Kliche, J. Solid State Chem., 40, 64 ͑1981͒.
6. A. Kjekshus and T. Rakke, Acta Chem. Scand., A28, 99 ͑1974͒.
7. J. W. Sharp, E. C. Jones, R. K. Williams, P. M. Martin, and B. C. Sales, J. Appl.
Phys., 78, 2, 1013 ͑1995͒.
8. G. Kliche, Solid State Commun., 80, 1, 73 ͑1991͒.
9. P. Jolibois, Comptes rendus Acad. Sci. (Paris), 150, 106 ͑1910͒.
10. A. Watcharapasorn, R. C. DeMattei, R. S. Feigelson, T. Caillat, A. Borshchevsky,
G. J. Snyder, and J.-P. Fleurial, J. Appl. Phys., 86, 11, 6213 ͑1999͒.
11. P. M. Chene, Ann. Chim. (Paris), 15, 187 ͑1941͒.
12. J. J. Cuomo and R. J. Gambino, J. Electrochem. Soc., 115, 755 ͑1968͒.
13. R. C. DeMattei, D. Elwell, and R. S. Feigelson, J. Cryst. Growth, 44, 545 ͑1978͒.
14. R. C. DeMattei, D. Elwell, and R. S. Feigelson, J. Cryst. Growth, 43, 643 ͑1978͒.
15. K. Ishida and T. Nishizawa, Binary Alloy Phase Diagrams, H. Okamoto, P. R.
Subramanian, and L. Kacprzak, Editors, Vol. 1, p. 270, ASM International, OH
͑1990͒.
CoAs3.—The product of the electrolysis of the 95 mol %
metaarsenite melt contained approximately 90% CoAs3 and 10%
CoAs2 as estimated from the XRD pattern ͑Fig. 4b͒. There was
evidence that most of the metaarsenate had evaporated from the melt
during the two day experiment. Although there have been no further
attempts to synthesize phase pure CoAs3, it is clear in which direc-
tion the melt composition must be modified. Not only will the ratio
of metaarsenite to cobalt ion have to be changed, but other compo-
nents will have to be added to reduce the melting temperature of the
charge and therefore its vapor pressure. This was done with the
melts for the electrochemical synthesis of GaAs.14
16. K. Ishida and T. Nishizawa, Binary Alloy Phase Diagrams, H. Okamoto, P. R.
Subramanian, and L. Kacprzak, Editors, Vol. 2, p. 1234, ASM International, OH
͑1990͒.
17. P. C. Donohue, Mater. Res. Bull., 7, 943 ͑1972͒.
18. K. Ishida and T. Nishizawa, Binary Alloy Phase Diagrams, H. Okamoto, P. R.
Subramanian and L. Kacprzak, Editors, Vol. 2, p. 1219, ASM International, OH
͑1990͒.
Downloaded on 2014-11-30 to IP 138.251.14.35 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).