Electrochemical and Solid-State Letters, 6 共8兲 A162-A165 共2003兲
A162
0013-4651/2003/6共8兲/A162/4/$7.00 © The Electrochemical Society, Inc.
Reversible Lithium Uptake by FeP2
D. C. C. Silva,a O. Crosnier,a G. Ouvrard,b,* J. Greedan,c A. Safa-Sefat,c
and L. F. Nazara,z
a
Department of Chemistry, University of Waterloo, Waterloo, Ontario N2L 3G1, Canada
Institut des Matériaux Jean Rouxel, 44322 Nantes Cedex 3, France
c
Department of Chemistry and Brockhouse Institute for Materials Research, McMaster University, Hamilton,
Ontario L8S 4M1, Canada
b
The electrochemical reaction of lithium with the metal phosphide FeP2 is reported. Li uptake in this material was investigated
using electrochemical studies combined with X-ray diffraction, X-ray absorption spectroscopy, and magnetic measurements. The
uptake of six Li during the first discharge leads to a specific capacity of 1365 mAh/g, and an amorphous phase retaining substantial
Fe-P bonding that displays superparamagnetic behavior and an extremely low transition temperature. Upon charge, 5.5 lithium can
be extracted from this metastable ternary ‘‘Li-Fe-P’’ phase, leading to a reversible capacity of 1250 mAh/g. The low 共ca. 10-15%兲
irreversibility exhibited by FeP2 makes it an interesting candidate in the search for new negative electrode materials.
© 2003 The Electrochemical Society. 关DOI: 10.1149/1.1588112兴 All rights reserved.
Manuscript submitted January 30, 2003; revised manuscript received April 8, 2003. Available electronically June 3, 2003.
The development and study of negative electrode materials for
rechargeable lithium batteries remains a major topic in the area of
energy storage. Although intermetallics and silicides have been most
actively examined recently,1 reports of Li uptake in transition metal
compounds of group 15 such as nitrides2-4 and antimonides,5 have
also received much attention. Their low intercalation potential results from the lower formal oxidation state of the metal, and strong
covalent character of the M-pnictogen bond, leading to high-lying
mixed anion-metal bands, and a high degree of electron
delocalization.6,7 Transition metal phosphides, such as CoP3 and
MnP4 have been investigated recently as possible candidates for
anode materials in lithium-ion batteries.8-10 These materials show a
good initial gravimetric capacity, and interesting transformations
characterized by anion-based redox chemistry, due to the high degree of covalence in the metal-phosphorus bond. In MnP4 , a quasitopotactic intercalation process is observed. The P-P bonds in the
binary layered phosphide are cleaved on reduction to form crystalline Li7 MnP4 and reformed on oxidation when MnP4 is recrystallized. Here we present the results for the reaction of lithium with
FeP2 , which displays different behavior.
FeP2 was synthesized using a tin flux technique. Powders of iron
共99.98%, Alfa Aesar兲, red phosphorous 共99.8%, Aldrich兲, and tin
共99.9%, Aldrich兲 in molar ratio of 1:10:40 were mixed inside an
argon filled glove box and transferred to a quartz tube and sealed
under vacuum. The tube and its contents were then heated at 700°C
for one week and quenched to room temperature. The resulting solid
was treated with three portions of 6 N HCl aqueous solution to yield
air-stable black crystals. The powder diffraction pattern obtained for
the material 共Siemens D500 diffractometer equipped with diffracted
beam monochromator using Cu K␣ radiation兲, showed that singlephase marcasite-type FeP2 was obtained.11 The electrochemical performance of the material was evaluated using lithium metal as both
pseudoreference and counter electrode in Swagelok-type cells. A
slurry of the material, previously ground and sieved to less than 20
␮m, was mixed with 20 wt % carbon black and 5 wt % polyvinylidene fluoride 共PVDF兲 in cyclopentanone and then cast on
nickel disks and dried under vacuum for at least 2 h. The cells were
assembled inside an argon-filled glove box using Celgard as separator and 1 M solution of LiPF6 in 1:1 ethylene carbonate 共EC兲/
dimethyl carbonate 共DMC兲 as electrolyte, and examined using a
MacPile controller 共Biologic S.A., Claix, France兲.
Figure 1 shows the first cycle of the potentiodynamic voltage vs.
composition curve acquired at an equivalent galvanic rate of C/4 共1
Li⫹ per 4 h兲. The flat intercalation plateau at 0.25 V corresponding
* Electrochemical Society Active Member.
z
E-mail: [email protected]
to the uptake of 6 Li⫹ is associated with a bell-shaped current response, characteristic of a two-phase transition. The initial reversibility upon charge is unusually high, approaching that of graphite.
The 5.5 Li⫹ that can be extracted from the structure correspond to a
reversible capacity of 1250 mAh/g 共or 6350 mAh/cm3兲, with an
irreversible component of only 12%. By comparison, oxides such as
CoO which react with lithium to form Li2 O and Co, exhibit irreversible capacities of up to 25% on the first cycle.12 Intermetallic compounds exemplified by FeSn2 , which has the same stoichiometry
but a different structure than FeP2 , uptake lithium with the dissociation of the starting material into metallic Fe and Li/Sn alloys. On
charge, partial reformation of FeSn2 occurs, leading to a specific
capacity of 800 mAh/g with 20% irreversibility during the first
cycle.13 Long-term cycling of FeP2 共see Fig. 1 inset兲 causes a
gradual loss of reversible capacity, partly due to reaction with the
electrolyte at the low potentials employed, and/or due to structural
changes within the material. Ex situ X-ray diffraction 共XRD兲 experiments of the cycled material revealed substantial loss of crystallinity
during the two-phase transition. Compared to the crystalline pattern
of FeP2 共Fig. 2a兲, only two very broad features appear in the XRD
pattern of the discharged material 共Fig. 2b兲, along with small peaks
ascribed to traces 共⬍5%兲 of residual FeP2 that did not fully react. A
lithium intercalation reaction like that exhibited by MnP4 共which
possesses a thermodynamically stable ternary phase, Li7 MnP4 ) is
not expected since only a lithium-poor ternary phase LiFeP14 exists
in the Li-Fe-P phase diagram. This phase does not correspond with
either the amount of lithium electrochemically ‘‘inserted’’ or the
broad features on the observed XRD patterns. The d-spacings of the
two broad features are close to some reflections of both Li3 P and
LiP,15 but the correspondence is poor, as many characteristic intense
peaks of these phases are also missing 共Fig. 2兲. Neither are any
features of metallic iron observed in the diffraction pattern, suggesting that simple decomposition of FeP2 into bulk Li3 P and Fe particles does not explain the reduction process. On oxidation, the XRD
pattern becomes completely featureless 共Fig. 2c兲.
To further characterize this complex Li uptake process, we used
X-ray absorption spectroscopy 共XAS兲 at the Fe K edge to probe the
nature and environment of the Fe atoms in the material before and
after cycling. X-ray absorption spectra were recorded at LURE 共Orsay, France兲 using X-ray synchrotron radiation emitted by the DCI
storage ring 共1.85 GeV, average intensity 250 mA兲. Data were collected in the transmission mode at the Fe K edge on the XAS 4
spectrometer, using a double-crystal monochromator Si共111兲 for extended X-ray absorption fine structure 共EXAFS兲 analysis and
Si共311兲 for X-ray absorption near-edge structure 共XANES兲 analysis.
XANES spectra were recorded by steps of 2 eV with 1 s of accumulating time per point within the 7060-7090 and 7140-7310 eV
Downloaded 16 Mar 2012 to 129.97.80.189. Redistribution subject to ECS license or copyright; see http://www.ecsdl.org/terms_use.jsp
Electrochemical and Solid-State Letters, 6 共8兲 A162-A165 共2003兲
Figure 1. Voltage-composition curve under potentiodynamic conditions corresponding to an equivalent C/4 rate for FeP2 . Inset: voltage vs. time curve
on cycling under galvanostatic conditions 共0.455 A/g兲 corresponding to a 2C
rate.
energy range, and 0.25 eV between 7090 and 7140 eV. For each
sample, three EXAFS spectra were recorded between 7000 and 8100
eV with an accumulating time of 2 s for each point split by 2 eV
steps. The analysis of the data was performed using WINXAS
code,16 and the amplitude and phase functions were extracted from
the Feff tables.
Consistent with the loss of crystallinity in the cycled samples, the
oscillations in the XAS spectra of the cycled materials were significantly reduced compared to those of FeP2 . The XANES spectra 共not
shown兲 reveal that the bandedge of FeP2 at half-height 共7118.2 eV兲
shifts to lower energy by only 0.5-1 eV after Li uptake. The edge
energy is similarly raised only marginally on oxidation by 0.2 eV.
These are small changes compared to the ⬎4 eV shift observed in
the reduction of MoO3 and other oxides upon Li uptake.17 As the
position of the main transition energy is correlated to the oxidation
state of the element,18 we conclude that the formal iron oxidation
state in FeP2 does not change significantly on cycling. The general
edge shape of the FeP2 derivatives also differs strongly from el-
A163
Figure 3. Pseudoradial distribution functions 共not corrected for phase shifts兲
derived from the EXAFS signals of FeP2 , 共
兲.; FeP2 on full discharge
共—兲; FeP2 on charge 共䊏兲; Fe metal 共• • • •兲. First and second shell Fe-P
and Fe-Fe distances are shown for FeP2 . Inset: Projection of the FeP6 octahedra along the 关001兴 axis in the structure of FeP2 .
emental iron. This is especially true for the discharged phase and
rules out the formation of such elemental iron upon lithiation.
The EXAFS portion of the XAS experiment reveals information
on the changes in the local coordination environment. The pseudoradial distribution functions 共RDFs兲 obtained from Fourier transformation of the EXAFS signals are shown in Fig. 3. The RDFs of
pristine FeP2 , and the material after full reduction and oxidation
show interesting differences. In FeP2 , the fits of the EXAFS spectra
for the first and second coordination shells of Fe yield 6 P atoms at
2.25 Å and 1.6 Fe at 2.71 Å 共Table I兲. These values are in perfect
accord with the average values of the Fe-P 共2.26 Å兲 and Fe-Fe 共2.72
Å兲 distances obtained from crystallography 共see structure, Fig. 3兲.
After the uptake of 6 Li, a diminution of the intensity of these Fe-P
interactions in the structure is apparent, but not their complete disappearance 共Fig. 3b兲. Surprisingly, 4.6 neighbors are maintained in
the first Fe-P shell at about the same distance 共2.29 Å兲 relative to the
starting material. The rather low coordination number for an expected octahedral environment could be due to a significant distribution in the Fe-P distances. Long range order, however, is destroyed by Li uptake. The lack of features at higher distances also
proves that the existing Fe-P interactions cannot result from traces of
Table I. Summary of EXAFS fits at the Fe K edge for the first
shell of FeP2 .
Material
FeP2
Crystal data
FeP2 pristine
FeP2 discharge
FeP2 charge
Neighbor
Na
P
Fe
P
Fe
P
Fe
P
Fe
6
2
6.0
1.6
4.6
0
4.7
2.0
R 共Å兲a ␴ 2 共Å2兲a ⌬E o 共eV兲a Re 共%兲b
2.26
2.72
2.25
2.71
2.29
—
2.25
2.68
—
—
0.0032
0.0023
0.0080
—
0.0064
0.0101
a
Figure 2. Ex situ XRD patterns of FeP2 electrode 共a兲 pristine FeP2 before
cycling; 共b兲 stopped at 0.1 V during first discharge; 共c兲 after the first charge.
The dotted lines are for reference and represent LiP 共
兲 and Li3 P
共• • • •兲.
—
—
1.34
—
—
7.00
2.13
—
0.41
16.44
—
12.73
N, R, ␴, and ⌬E o are the fitted values of the number of surrounding
atoms, their distance from absorbing Fe, the Debye-Waller factor, and
the energy shift, respectively.
b
The residual Re is defined as
N
N
Re ⫽ 兺 i⫽1
兩 y exp(i) ⫺ y theo共i) 兩 / 兺 i⫽1
兩 y exp(i) 兩 .
Downloaded 16 Mar 2012 to 129.97.80.189. Redistribution subject to ECS license or copyright; see http://www.ecsdl.org/terms_use.jsp
A164
Electrochemical and Solid-State Letters, 6 共8兲 A162-A165 共2003兲
The behavior in this system can be contrasted to MnP4 , where
the existence of a crystalline ternary phase (Li7 MnP4 ), allows for
reversible structural transformation between the two phases on electrochemical Li uptake and extraction. A phase transition of this nature is not possible for FeP2 , however, due to the lack of thermodynamic stability of the Li-Fe-P phase with the corresponding
stoichiometry. From a thermodynamic view point, only two events
can occur on Li uptake at room temperature: either phase separation
into Li3 P ⫹ Fe; or formation of a metastable ternary phase. We
have shown here that the latter takes place. The sum of the XAS,
magnetization, and Mössbauer data unequivocally demonstrate formation of a metastable Li-Fe-P phase which displays a high level of
disorder. This nanostructured material contains Fex P-rich magnetic
domains and Li-P rich regions in which the iron atoms retain coordination to P
discharge
charge
Li ⫹ FeP2 → Fex -P-Liy P-Fex → ‘‘FePq ’’ ⫹ Li
Figure 4. Temperature dependence of magnetization 共white and black symbols for FC and ZFC measurements, respectively兲 of FeP2 before cycling
共squares兲, stopped at the end of discharge 共circles兲, and at the end of the
charge 共triangles兲; inset: field dependence of the magnetization of the discharged material at different temperatures.
residual unreacted FeP2 , since the entire RDF would be diminished
in intensity. The first or second shell cannot be fit with Fe-Fe interactions, showing that metallic iron is not formed, as evidenced by
the RDFs 共Fig. 3兲.
The formation of particles of bulk metallic Fe is furthermore
ruled out by the magnetic behavior of the materials. These experiments were conducted between 2 and 350 K using a Quantum Device MPMS SQUID 6T magnetometer and the results are shown in
Fig. 4. In agreement with previous reports,19,20 FeP2 is essentially
diamagnetic with a weak, nearly temperature independent paramagnetic contribution, allowing changes in the magnetic behavior on
cycling to be readily detected. After Li uptake 共discharge curve兲, the
magnetic susceptibility increases by a factor of ⬃102 over the temperature range studied and a strong temperature dependence develops. This is consistent with the presence of small particle magnetism, i.e., superparamagnetism. No obvious blocking temperature
appears but data collected in field cooled 共FC兲 and zero field cooled
共ZFC兲 modes never overlap. Magnetization vs. field data collected at
350 and 100 K show behavior typical of superparamagnetism with
no hysteresis while a very weak remanent moment appears at 2 K
共see inset兲. It is not possible to distinguish between the formation of
nanoscopic Fe, or Fex P (x ⫽ 1, 2, 3) as these are all magnetic
materials that would exhibit properties different from their bulk behavior. Mössbauer studies of the material, however, suggest the Fe is
in an environment in close contact with phosphorus, whose isomer
shift resembles ‘‘FeP’’, with no magnetic splitting observable even
at 4 K.21
On charge, the superparamagnetic component in the magnetization curve is dramatically reduced, albeit not to zero 共Fig. 4兲. Since
the electrochemical lithium uptake process is not completely reversible, the remaining paramagnetic signal may be due to a small inactive fraction of the material that has reacted irreversibly during
lithium insertion. The pseudo-RDF of the charged material shows
that changes occur in the local environment vis-à-vis the reduced
sample. A modest increase in the number of Fe-P nearest neighbors
is evident, although the inaccuracy of the EXAFS method in determining coordination numbers makes it difficult to quantify the
changes. Most important, some of the long-range structure at distances ⬎4 Å that exists in FeP2 reappears. An increase in both the
Fe-Fe and the long-range interactions at 4.5 and 6.1 Å shows that a
more ordered environment is reassembled, despite an overall longrange amorphous structure.
The size of the magnetic domains, as revealed by the magnetization
and Mössbauer data is in the nanoregime 共⬍10 Å兲, approaching that
of Fe-P clusters.
That a metastable disordered ternary phase can be formed is, we
believe, due to the similarity in the P atom ordering in FeP2 and
Li3 P, as depicted in Scheme 1
Scheme 1. Large circles correspond to phosphorus atoms, small circles to Li
atoms 共in Li3P) and Fe atoms 共in FeP2).
Both structures have been oriented with a view perpendicular to two
close-packed 共cp兲 phosphorus layers. In Li3 P the phosphorous atoms
form an ideal hexagonal cp arrangement, whereas in FeP2 , a small
distortion from ideal cp behavior is observed. To illustrate the topographical correspondence in the two structures, three atoms that
form the upper trigonal cp surface are lighter in color, and three
atoms that span the lower trigonal plane are in black. The major
differences between the two structures lie in 共i兲 the occupation of the
additional cation sites in Li3 P; (ii) an expansion of the P-P distances in FeP2 共2.7/3.1 Å兲 to 4.27 Å in Li3 P. Thus we propose Li
uptake occurs by expansion of the phosphorus lattice, and incorporation of the additional lithium to form a disordered, stuffed metastable ‘‘Li3 Fe0.5P’’ lattice that still maintains Fe within the structure.
That process would disrupt the P-P bonding by the probable occupancy of antibonding P-P levels by the first extra electron. Second,
the FeP6 octahedra would be isolated from each other by an expansion of the P-P distance. Such a transformation would suppress
Fe-Fe interactions but maintain Fe-P bonding, consistent with the
observed EXAFS data. Then on charge, Li is extracted from the
disordered phase to yield a material that although amorphous, contains some of the local coordination between FeP6 octahedra. The
electrochemical formation of metastable ternary phases has also
been very recently suggested in the transformation of LiMnSb into
Li3 Sb via a metastable solid solution Li1⫹2x Mn1⫺x Sb. 22
In summary, we note that for oxides, owing to the higher electronegativity difference between the components, the reduction to
the metallic state is driven by a large free energy of reaction. In
phosphides, the high degree of covalent bonding and low valence of
the metal induces a transformation more akin to intercalation, either
a structural rearrangement (MnP4 -Li7 MnP4 ) or the transformation
Downloaded 16 Mar 2012 to 129.97.80.189. Redistribution subject to ECS license or copyright; see http://www.ecsdl.org/terms_use.jsp
Electrochemical and Solid-State Letters, 6 共8兲 A162-A165 共2003兲
to a disordered state observed here. This behavior is also in contrast
to the lower members of group 15 共i.e., antimonides兲, as these materials are characterized by complete phase separation into the
lithium antimonide and the metal, as occurs for FeSb2 . 21,23
Acknowledgments
D.C.C.S. thanks CAPES/Brazil for their financial support in the
form of a scholarship. L.F.N. gratefully acknowledges the support of
NSERC for funding this work.
University of Waterloo assisted in meeting the publication costs of this
article.
References
1. S.-W. Song, K. A. Striebel, R. P. Reade, G. A. Roberts, and E. J. Cairns, J. Electrochem. Soc., 150, A121 共2003兲; K. D. Kepler, J. T. Vaughey, and M. M. Thackeray, Electrochem. Solid-State Lett., 2, 307 共1999兲.
2. M. Nishijima, T. Kagohashi, M. Imanishi, Y. Takeda, O. Yamamoto, and S. Kondo,
Solid State Ionics, 83, 107 共1996兲.
3. T. Shodai, S. Okada, S. Tobishima, and J. Yamaki, Solid State Ionics, 86-88, 785
共1996兲.
4. Y. Takeda, M. Nishijima, M. Yamahata, K. Takeda, N. Imanishi, and O. Yamamoto,
Solid State Ionics, 130, 61 共2000兲.
5. J. T. Vaughey, J. O’Hara, and M. M. Thackeray, Electrochem. Solid-State Lett., 3,
13 共2000兲; L. M. L. Fransson, J. T. Vaughey, K. Edström, and M. M. Thackeray, J.
Electrochem. Soc., 150, A86 共2003兲.
6. I. Lefebvre-Devos, M. Lassalle, X. Wallart, J. Olivier-Fourcade, L. Monconduit,
and J.-C. Jumas, Phys. Rev. B, 63, 125110 共2001兲.
A165
7. L. Monconduit, M. Tillard-Charbonnel, and C. Belin, J. Solid State Chem., 156, 37
共2001兲.
8. D. C. S. Souza, V. Pralong, A. Jacobson, and L. F. Nazar, Science, 296, 5575
共2002兲.
9. V. Pralong, D. C. S. Souza, K. T. Leung, and L. F. Nazar, Electrochem. Commun.,
4, 516 共2002兲.
10. R. Alcantara, J. L. Tirado, J. C. Jumas, L. Monconduit, and J. Olivier-Fourcade, J.
Power Sources, 109, 308 共2002兲.
11. E. Dahl, Acta Chem. Scand., 23, 2677 共1969兲.
12. P. Poizot, S. Laruelle, S. Grugeon, L. Dupont, and J. M. Tarascon, Nature (London), 407, 496 共2000兲.
13. O. Mao, R. A. Dunlap, and J. R. Dahn, J. Electrochem. Soc., 146, 405 共1999兲.
14. R. Juza and K. Langer, Z. Anorg. Allg. Chem., 361, 58 共1968兲.
15. G. Brauer and E. Zintl, Z. Phys. Chem. Abt. B, 37, 323 共1937兲.
16. T. Ressler, J. Phys. IV, C2, 269 共1997兲.
17. F. Leroux, G. R. Goward, W. P. Power, and L. F. Nazar, Electrochem. Solid-State
Lett., 1, 255 共1998兲; C. Rossignol, G. Ouvrard, and E. Baudrin, J. Electrochem.
Soc., 148, A869 共2001兲.
18. A. N. Mansour, X. Q. Yang, X. Sun, J. McBreen, L. Croguennec, and C. Delmas,
J. Electrochem. Soc., 147, 2104 共2000兲; M. Giorgetti, S. Passerini, W. H. Smyrl, S.
Mukerjee, X. Q. Yang, and J. McBreen, J. Electrochem. Soc., 146, 2387 共1999兲.
19. G. Brostigen and A. Kjekshus, Acta Chem. Scand., 24, 2993 共1970兲.
20. G. Boda, B. Stenstrom, V. Sagredo, O. Beckman, B. Carlsson, and S. Rundqvist,
Phys. Scr., 4, 132 共1971兲.
21. O. Crosnier, D. C. C. Silva, G. Ouvrard, M. Danot, and L. F. Nazar, In preparation.
22. L. M. L. Fransson, J. T. Vaughey, K. Edström, and M. M. Thackeray, J. Electrochem. Soc., 150, A86 共2003兲.
23. O. Crosnier, D. C. Souza, J. Greedan, A. Safa-Sefat, and L. F. Nazar, Abstract
EE8.3, Materials Research Society Symposium, Dec 2-6, 2002.
Downloaded 16 Mar 2012 to 129.97.80.189. Redistribution subject to ECS license or copyright; see http://www.ecsdl.org/terms_use.jsp