Phospho-Olivines as Positive-Electrode Materials for Rechargeable Lithium Batteries

Abstract

Reversible extraction of lithium from

(triphylite) and insertion of lithium into

at 3.5 V vs. lithium at 0.05 mA/cm2 shows this material to be an excellent candidate for the cathode of a low‐power, rechargeable lithium battery that is inexpensive,
nontoxic, and environmentally benign. Electrochemical extraction was limited to ∼0.6 Li/formula unit; but even with this restriction
the specific capacity is 100 to 110 mAh/g. Complete extraction of lithium was performed chemically; it gave a new phase,

, isostructural with heterosite,

. The

framework of the ordered olivine

is retained with minor displacive adjustments. Nevertheless the insertion/extraction reaction proceeds via a two‐phase process,
and a reversible loss in capacity with increasing current density appears to be associated with a diffusion‐limited transfer
of lithium across the two‐phase interface. Electrochemical extraction of lithium from isostructural

(M = Mn, Co, or Ni) with an

electrolyte was not possible; but successful extraction of lithium from

was accomplished with maximum oxidation of the

occurring at x = 0.5. The

couple was oxidized first at 3.5 V followed by oxidation of the

couple at 4.1 V vs. lithium. The

interactions appear to destabilize the

level and stabilize the

level so as to make the

energy accessible.

Full text

1188 J. Electrochem. Soc., Vol. 144, No. 4, April 1997 The Electrochemical Society, Inc.
Depolymerization
Cathodic Process
the drawbacks previously mentioned for DMcT electrodes
have been overcome by using DMcT/PVP composite films.
Acknowledgment
The present work has been financially supported by the
Japanese Ministry of Education. K.N. would like to thank
Mitsubishi Chemicals for supplying high quality PC solvent.
Manuscript submitted June 19, 1996; revised manuscript
received about Dec. 5, 1996.
Tokyo University of Agriculture and Technology assist-
ed in meeting the publication costs of this article.
REFERENCES
1. 5. J. Visco and L. C. DeJonghe, in Handbook of Solid
State Batteries and Capacitors, M. Z. A. Munshi,
Editor, Chap. 22, World Scientific, NJ (1995).
2. K. Naoi, M. Menda, H. Ooike, and N. Oyama, J.
Electroanal. Chem., 318, 395 (1992).
3. A. Kaminaga, T. Tatsuma, T. Sotomura, and N. Oyama,
This Journal, 142, L47 (1995).
4. N. Oyama, 11 Tatsuma, T. Sato, and P Sotomura,
Nature, 373, 598 (1995).
5. J. M. Liu, S. J. Visco, and L. C. DeJonghe, This
Journal, 136, 2570 (1989).
6. J. M. Liu, S. J. Visco, and L. C. DeJonghe, ibid., 138,
1891 (1991).
7. J. M. Liu, S. J. Visco, and L. C. DeJonghe, ibid., 138,
1896 (1991).
8. T. Sotomura, H. Uemachi, K. Takehara, K. Naoi, and
N. Oyama, Electrochim. Acta, 37, 1851 (1992).
9. Y. Iwamizu, V. Oura, and K. Naoi, in Proceedings of
36th Japanese Battery Meeting, p. 51 (1995).
10. I. Sekine, K. Kohara, P Sugiyama, and M. Yuasa, This
Journal, 139, 3090 (1992).
11. K. Naoi, Y. Oura, I Iwamizu, and N. Oyama, ibid., 142,
354 (1995).
12. 5. Picart and E. Genies, J. Electroanal. Chem., 53, 408
(1996).
13. G. D. Thorn, Can. J. Chem., 38, 1439 (1960).
14. B. Stanovnik and M. Tisler, Croat. Chem. Acta, 37, 17
(1965).
15. A. J. Bard and L. R. Faulkner, in Electrochemical
Methods, Fundamentals and Applications, Chap. 3,
John Wiley & Sons, Inc., New York (1980).
16. E. M. Genies and S. Picart, Synth. Met., 69, 165 (1995).
17. M. Mori, Y. Naruoka, WI Kokubo, and K. Naoi, in 1996
Japanese Electrochemical Society Autumn Meeting,
p. 16 (1996).
Phospho-olivines as Positive-Electrode Materials for
Rechargeable Lithium Batteries
A. K. Padhi, K. 5. Nanjundaswamy,** and J. B. Goodenough
Center for Materials Science and Engineering, The University of Texas at Austin, Austin,Texas 78712-1 063, USA
ABSTRACT
Reversible extraction of lithium from LiFePO4 (triphylite) and insertion of lithium into FePO4 at 3.5 V vs.lithium at
0.05 mA/cm2 shows this material to be an excellent candidate for the cathode of a low-power, rechargeablelithium bat-
tery that is inexpensive, nontoxic, and environmentally benign.Electrochemical extraction was limited to —0.6 Li/f or-
mula unit; but even with this restriction the specific capacity is 100 to 110 mAh/g. Completeextraction of lithium was
performed chemically; it gave a new phase, FePO4, isostructural with heterosite, Fe0 65Mn035PO4.The FePO4 framework of
the ordered olivine LiFePO4 is retained with minor displacive adjustments. Nevertheless the insertion/extraction reaction
proceeds via a two-phase process, and a reversible loss in capacity with increasingcurrent density appears to be associ-
ated with a diffusion-limited transfer of lithium across the two-phase interface.Electrochemical extraction of lithium
from isostructural LiMPO4 (M =Mn, Co, or Ni) with an LiC1O4 electrolyte was not possthle but successful extraction of
lithium from LiFe2MnPO4 was accomplished with maximum oxidation of theMn3VMn + occurring at x = 0.5. The
Fe34fFe2t couple was oxidized first at 3.5 V followed by oxidation of the Mn3/Mn24 coupleat 4.1 V vs. lithium. The Fe34-
0-Mn24 interactions appear to destabilize the Mn24 level and stabilize the Fe34 level so asto make the Mri3/Mn24 energy
accessible.
Introduction
Since the demonstration of reversible lithium intercala-
tion between the layers of Ti53,1 considerable effort has
been devoted to the identification of other lithium-inser-
tion compounds that can be used as the cathode for a sec-
ondary lithium battery. The desired material would have a
relatively flat open-circuit voltage over a large lithium
solid solution within the voltage range of 2.5 .c 170< 4.0 V
and be inexpensive, easy to fabricate, environmentally
*Electrochemical Society Student Member.
** Electrochemical Society Active Member.
benign, and safe in handling and operation. Reversible
lithium insertion/extraction has been performed on a vari-
ety of compounds containing different transition-metal
cations and structural architectures. The sulfides have too
low a V and the halides too low an electronic conductiv-
ity, so particular attention has been given to transition-
metal oxides. These efforts have resulted in the develop-
ment of rechargeable lithium batteries that now serve as
state of the art power sources for consumer electronics.
Among the known Li-insertion compounds, the layered
rock salt systems Li2_4Co03, 2 Li1_4Ni03, and the man-
ganese-spinel framework system Li1_EMn2l04 are now
(A)
I#L,4oç1
L'fJPolymerization
Anodic Process
Fig. 5. Proposed sfructural change of DMcT in PVP film during the
oxidation-reducHon cycle.
) unless CC License in place (see abstract). ecsdl.org/site/terms_use address. Redistribution subject to ECS terms of use (see 17.209.11.199Downloaded on 2015-06-11 to IP

J. Electrochem. Soc., Vol. 144, No. 4, April 1997 The Electrochemical Society, Inc. 1189
used commercially as 4.0 V positive-electrode materials in
rechargeable lithium batteries. However, the voltages in
excess of 4.0 V on higher charge in these oxides can lead to
the decomposition of the electrolytes, and the fully
charged compounds are metastable: [Mn2]04 converts to e-
Mn02 at 190°C5 while fully charged phases of Li,Co02
and Li1_Ni02 lose 02 above 180 and 250°C, respectively.6
Moreover, a lower lithium mobility within the [Mn2]04
spinel framework limits its power capability. Although the
Li9[Mn2]O4 system has a flat V,, 3.0 V vs. lithium,
which is attractive,7 structural changes associated with a
cooperative Jahn-Teller deformation of the framework
tend to reduce the capacity irreversibly on repeated
cycling.8 In addition, the availability and cost of the tran-
sition metals used in these compounds are unfavorable as
the Wh/$ is a more important figure of merit than Wh/g in
the case of large batteries to be used in an electric vehicle
or a load-leveling system. These consideration have moti-
vated the investigations of iron-based oxides.
The iron-based oxides containing 02- as the anion pose
a problem for the cathode designer; in these oxides the
Fe47Fe3 redox energy tends to lie too far below the Fermi
energy of a lithium anode and the Fe3/Fe2 couple too
close to it. Layered LiFeO2 prepared by ion-exchange from
a-NaFeO2 has been investigated.'0 It would operate on
the Fe/Fe3* redox couple, but it is metastable and gives
unimpressive battery performance. The other iron-based
compounds proposed, viz. FePS3," FeOC1,'2 and FeOOH,'3
have a relatively poor rechargeability and/or too low a dis-
charge voltage. On the other hand, the use of polyanions
such as (SO4)2, (PO4)3, (As04)3, or even (Mo04)2 or
(W04)2 have been shown to lower the Fe3/Fe2 redox
energy to useful levels. Among the compounds with NASI-
CON framework, for example, the open-circuit voltages
vs. lithium are 3.6 V for LiFe2(SO4)3,'4 2.8 V for
Li3Fe2(PO4)3,'5 and 2.75 V for Li2FeTi(P04)3 16; each of these
materials has a specific capacity of about 100 mAh/g.
Tuning of the energy of the Fe3*/Fe2* couple is accom-
plished through the choice of the countercation within the
polyanion. Polarization of the electrons of the O ions into
strong covalent bonding within the polyanion reduces the
covalent bonding to the iron ion, which lowers its redox
energy. The stronger the covalent bonding within the
polyanion, the lower is the Fe3/Fe2 redox energy and the
higher the V vs. lithium for that couple.
The open NASICON framework allows fast Li-ion dif-
fusion, but a separation of the Fe06 octahedra by polyan-
ions reduces the electronic conductivity, which is polaron-
ic in the mixed-valent state. In this paper we report the
cathode performance of an iron phosphate having an
ordered olivine structure in which the FeO6 octahedra
share common corners.
The M2X04 olivine structure has M atoms in half of the
octahedral sites and X atoms in one-eighth of the tetrahe-
dral sites of an hexagonal close-packed (hcp) oxygen
array; it is the hexagonal analog of the cubic normal spinel
X[M2]04. Olivine crystallizes in preference to spinel for
certain small X ions such as Be2, B3, Si4, p5 and occa-
sionally Ge4. Unlike spinel, the two octahedral sites in
olivine are crystallographically distinct and differ in size,
which favors ordering in MM'X04 olivines containing M
and M' ions of different size and charge. The LiMPO4 com-
pounds, with M = Fe, Mn, Co, or Ni, have the ordered
olivine structure.
Figure 1 shows the crystal structure of olivine: an ideal
hcp model and the actual structure. In the actual struc-
ture, the M(l) site has I symmetry, the M(2) octahedron has
mirror symmetry with average M-O distances greater than
that in the M(1) octahedron. The M(1) sites form linear
chains of edge-shared octahedra running parallel to the
c-axis in the alternate a-c planes; the M(2) sites form zig-
zag planes of corner-shared octahedra running parallel to
the c-axis in the other a-c planes (see Fig 6). Each M(1) site
shares its edges with two M(2) sites and two X sites; there
is one edge shared by an M(2) site with an X site.'7
Distortion of the hcp oxygen array has been related to the
cation-cation coulomb repulsion across the shared edges.
In the LiMPO4 (M = Mn, Fe, Co, or Ni) compounds, the
lithium moieties occupy M(1) sites and the M atoms M(2)
sites. With Li in the continuous chain of edge-shared octa-
hedra on alternate a-c planes, a reversible extraction/
insertion of lithium from/into these chains would appear
to be analogous to the two-dimensional extraction or
insertion of lithium in the LiMO2 layered oxides with M =
Co or Ni. On the other hand, the XO4 tetrahedra bridge
between adjacent M(2) planes in the olivine structure,
which constrains the free volume in which the Li-ions
move; only the Li-O bonding constrains the spacing
between MO2 layers in the LiMO2compounds.
Our attempts to delithiate LiMnPO4, LiCoPO4, and
L1N1PO4 proved unsuccessful with the LiClO4 electrolyte
used. However, we could use our ability to delithiate
LiFePO4 to initiate delithiation in the solid-solution sys-
tem LiFe1.MnO4; we report a Mn37Mn2 couple at
(a) (b)
Fig. 1. Olivine crystal sfruc-
tare: (a) ideal HCP model, (b)
actual shjcture.
) unless CC License in place (see abstract). ecsdl.org/site/terms_use address. Redistribution subject to ECS terms of use (see 17.209.11.199Downloaded on 2015-06-11 to IP

1190 J. Electrochem. Soc., Vol. 144, No. 4, April1997 The Electrochemical Society, Inc.
4.1 eV below the Fermi energy of a lithium anode where
there are Fe3-O-Mn' interactions.
Experimental
LiMPO4 (M = Mn, Fe, Co, or Ni) compounds were pre-
pared by direct solid-state reaction of stoichiometric
amounts of M(II)-acetates, ammonium phosphate, and
lithium carbonate. LiFePO4 and LiFe,Mn04 (x = 0.25,
0.50, and 0.75) were synthesized in inert atmosphere to
prevent the formation of Fe3 compounds as impurities.
The intimately ground stoichiometric mixture of the start-
ing materials was first decomposed at 300 to 350°C to
drive away the gases. The mixture was then reground and
returned to the furnace at 800°C for 24 h before being
cooled slowly to room temperature. The x-ray powder dif-
fraction technique was used to identify the phases. The
unit-cell parameters were obtained with a least squares
refinement to the diffraction peaks. Rietveld refinement of
the x-ray diffraction (XRD) data was performed to obtain
the structure.
The electrochemical extraction/insertion of lithium and
characterization of the performance of the phospho-
olivines as cathodes were made with coin-type cells (Type
2320). After the materials were ground to fine particles
with a milling machine, they were mixed/blended with
acetylene black and polytetrafluoroethylene (PTFE) in the
weight ratio 70:25:5. This cathode mixture, after being
kept at 140°C for 2 h, was rolled into thin sheets of uni-
form thickness and cut into pellets of required size for
coin-cell fabrication. The electrolyte was 1 M LiC1O4 in a
1:1 mixture of propylene carbonate (PC) and dimethoxy-
ethane (DME). A lithium foil was used as the anode. The coin
cell was fabricated in a glove box under argon atmosphere.
Chemical delithiation to obtain Li,1MP04 (0 < x<1)
was performed by reacting the materials with nitronium
hexafluorophosphate (NO2PF,) in acetonitrile under inert
atmosphere. Reaction of LiFePO4 with bromine in acetoni-
trile was also used to extract lithium chemically. Chemical
lithiation was carried out by reacting the material with
lithium iodide. The products were washed several times
with acetonitrile to ensure the purity of the solid phase
before it was dried in vacuum. Atomic absorption spec-
trosopy was performed on intermediate compositions to
obtain the exact lithium content with a Perkin-Elmer 1100
spectrometer.
The thermal stability of the phases was monitored from
50 to 500°C by TGA and DSC techniques on a Perkin-
Elmer Thermal Analysis 7 instrument. These experiments
were performed in both oxygen and an inert atmosphere.
Results and Discussion
Electrochemical charge and discharge curves for
LiFePO4, Fig. 2, show that approximately 0.6 lithium
atoms per formula unit can be extracted at a closed-circuit
voltage of 3.5 V vs. lithium and the same amount can be
reversibly inserted back into the structure on discharge.
The extraction and insertion of lithium ions into the struc-
ture of LiFePO4 is not only reversible on repeated cycling;
the capacity actually increases slightly with cycling.
The placement of the Fe3/Fe2 redox energy at 3.5 eV
below the Fermi level of lithium in Li,FePO4 is to be
compared with 2.8 eV found'5'16 in Li3+e2(PO4)3 and
Li2+FeTi(PO4), and at 3.6 eV in LiFe,(SO4),.'4 A difference
of 0.8 eV between the redox energies in the isostructural
NASICON frameworks of Li3+Fe,(PO4), and LiFe,(SO4),
can be attributed to the inductive effect, the oxygen form-
ing a stronger bond within (SO4)2- than in (PO4fpolyan-
ions. On the other hand, all the oxygen of both Li1FePO4
and Li3Fe,(PO4)3 form strong covalent bonds within a
(P04)3- complex, so the 0.7 eV difference in the Fe3/Fe2
redox energies of these two compounds must have anoth-
er origin than the inductive effect. For the origin of this
difference, we turn to the ionic component of the bonding.
In an ionic compound, the position of the electron ener-
gy levels depends critically on the Madelung potential at
the different atoms, which depends on both the structure
Fig. 2. Discharge/charge curves vs. lithium at 2.0 mA/g
(0.05 mA/cm9 for Li,..,,FePO4.
andthe degree of covalence in the bonding. The Madelung
electric field raises the electron energies of the cations and
lowers those of anions; in an ionic crystal, the Madelung
fields are strong enough to overcome the energy required
to create the ionic species, and the redox states are anti-
bonding states of primarily cationic origin. Reference to
an Fe37Fe2 redox energy implies a substantial ionic com-
ponent to the bonding; and the stronger the Madelung
electric field at the cation site, the higher is the Fe3/Fe2
redox energy. The Madelung sum of coulomb energies can
account qualitatively for a lower Fe3 7Fe2 redox energy,
and hence a higher V0. vs. lithium in Li,0FeP04 than in
Li3+e,(P04)3. In the NASICON framework, the Fe06
octahedra share no edges with other cation polyhedra,
which reduces the cation-cation coulomb repulsions con-
tributing to the Madelung sum, whereas considerable edge
sharing occurs in the ordered olivines. The cation-cation
repulsive forces distort the hcp anion array of an olivine,
as noted above, but the repulsion is not sufficient to screen
the reduction by these forces of the total Madelung electric
field that raises the Fe'/Fe' redox energy above the
(P04)3- energies. Therefore, the Fe3/Fe2 level lies lower in
the ordered olivine structure.
The V(x) curves for Li1_ePO4 in Fig. 2 show a voltage
that is independent of x over a large range of x, which
indicates, by Gibb's phase rule, that the extraction/inser-
tion reactions proceed by the motion of a two-phase inter-
face. To establish the existence and structure of the second
phase, a partial chemical delithiation was performed by
reacting LiFePO4 with varying amounts N0,PF6 in ace-
tonitrile. Chemical delithiation allows XRD patterns to be
taken on clean samples. The XRD patterns in Fig. 3 show
the emergence and growth of a second phase at the
expense of LiFePO4 as more and more lithium is extract-
ed. With total chemical delithiation, the second phase
could be identified by both chemical analysis and Rietveld
refinement to XRD data to be FePO4. XRD patterns for
chemical lithiation of FePO4, Fig. 4, show the emergence
and growth of LiFePO4 at the expense of FePO4 on more
lithiation. Electrochemical characterization of a cathode
made from the FePO4 obtained by total chemical delithia-
tion of LiFePO4 gave the V(x) curves of Fig. 5; they are
similar to those of Fig. 2, thus confirming that FePO4 is the
second phase that is present on electrochemical extraction
of lithium from LiFePO4. Therefore the extraction of lithi-
um from LiFePO4 to charge the cathode may be written as
LiFePO4 — xLi—xe -* xFePO4 +(1 —x)LiFePO4
and the reaction for the insertion of lithium into FePO4 on
discharge as
FePO4 + xLi0 +xe xLiFePO4 + (1 —x)FePO4
Capacity [mAh/gJ
>4)aCt0>4)C.)
y in Li1.FePO4
) unless CC License in place (see abstract). ecsdl.org/site/terms_use address. Redistribution subject to ECS terms of use (see 17.209.11.199Downloaded on 2015-06-11 to IP

J. Electrochem. Soc., Vol. 144, No. 4, April 1997 The Electrochemical Society, Inc. 1 191
Fig. 3. Chemical delithiation of LiFePO4. XRD patterns showing
the emergence and growth of the second phase FePO4.
The excellent reversibility of the cells on repeated
cycling is due to the striking similarity of the LiFePO4 and
FePO4 structures, which are compared in Fig. 6. FePO4 is
isostructural with heterosite, Fe065Mn035PO4, for which
several bond lengths have been refined.'8 The lattice
2-Theta (Cu Kcx)
Fig. 4. Chemical lithiation of FePO4. XRD patterns showing the
emergence ond growth of the second phase L1FePO4.
0.0 0.2 0.4 0.6 0.8
x in LiFePO4
Fig. 5. Discharge/charge curves vs. lithium at 2.0 mA/g
(0.05 mA/cm2) for LiFePO4.
parameters and the space group of both LiFePO4 and
FePO4 phases are listed in Table I; both LiFePO4 and
FePO4 have the same space group. On chemical extraction
of lithium from LiFePO4, there is a contraction of the a
and b parameters, but a small increase in the cparameter.
The volume decreases by 6.81% and the density increases
by 2.59%. Although the changes in the FePO4 framework
are displacive, not diffusional, a first-order transition
between LiFePO4 and FePO4 prevents the continuous
insertion reaction
LiFePO4 — xLix —xe -* Li,FePO4
A first-order transition would seem to require a coopera-
tive elastic deformation of the FePO4 framework. It is
therefore of interest that the principal change in the
framework on delithiation is a cooperative adjustment of
the framework to the coulombic repulsion between the
O2ion sheet interfacing the delithiated planes.
Insertion of lithium into FePO4 was reversible over the
several cycles investigated. LiFePO4 represents a cathode
of good capacity, and it contains inexpensive, environmen-
tally benign elements. However, a nearly close-packed
hexagonal oxide-ion array that is bonded strongly in three
dimensions provides a relatively small free volume for Lit-
ion motion, so the electrode supports only relatively small
current densities at room temperature. Nevertheless,
increasing the current density does not lower the open-cir-
cuit voltage V0; rather it decreases, reversibly, the cell
capacity. Reducing the current restores the capacity. This
observation indicates the loss in capacity is a diffusion-
limited phenomenon associated with the two-phase char-
acter of the insertion process.
As is illustrated schematically in Fig. 7, lithium inser-
tion proceeds from the surface of the particle moving
inward behind a two-phase interface, a LiFePO4/
Li1 FePO4 interface in this system. As the lithiation pro-
ceeds, the surface area of the interface shrinks. For a con-
stant rate of lithium transport per unit area across the
interface, a critical surface area is reached where the rate
of total lithium transported across the interface is no
longer able to sustain the current; the cell performance
becomes diffusion-limited. The higher the current, the
greater is the total critical interface area and, hence, the
smaller the concentration x of inserted lithium before the
cell performance becomes diffusion-limited. On extraction
of lithium, the parent phase at the core of the particle
grows back toward the particle surface, which is why the
parent phase is retained on repeated cycling and the loss
in capacity is reversible on lowering the current density
delivered by the cell. This loss of capacity is not due to
a breaking of the electrical contact between particles as
a result of volume changes, a process that is normally
irreversible.
>'U,a)Ca)>
a)
>0)C)0>0)C)
Capacity [mAh/gl
020 40 60 80 100 120 140 160
SCurrent Density 0.05 mA/cm2, 1.85 mAIg
4uwLlt% .&¼i
3.
2• lsICycle
•2nd Cycle
£10thCycle
o20th Cycle
2-Theta (Cu Ku)
>.U,Ca)Ca)>
a)
15 20 25 30 35 40 45
) unless CC License in place (see abstract). ecsdl.org/site/terms_use address. Redistribution subject to ECS terms of use (see 17.209.11.199Downloaded on 2015-06-11 to IP

I i j. tiecrrocnem. soc., vol. 144, No. 4, pru ii cc ne iectrocMemicai society, inc.
Fig. 6. Crystal sfructures of (a)
LiFePO4 and (b) FePO4.
(a)
bL(b)
The thermal stability of the fully charged state of
LiFePO4, FePO4, is shown in the TGA curves of Fig. 8. On
thermal treatment of FePO4 in nitrogen atmosphere up to
350°C, ther? was no appreciable change in the weight. A
weight loss of 1.6% is observed when the sample was heat-
ed up to 500°C. There was very little difference in the TGA
curves when the sample was heated in oxygen atmosphere.
No appreciable change could be found in the XRD pat-
terns taken after thermal treatment; there was no trace of
impurity. Since the FePO4 for these experiments was pre-
pared by treating LiFePO4 with bromine several times in
acetonitrile, there could be a small amount of LiBr in the
sample even after washing the products several times with
acetonitrile, which decomposes at 350°C. The DSC curve
shows a small reversible peak at 300°C of unknown origin.
In order to locate the Mn34/Mn2, Co3/Co2', and
Ni37Ni2 redox energies with respect to the Fermi energy
of lithium, we tried to extract lithium electrochemically
from other LiMPO4 compounds with M = Mn, Co, or Ni.
Since LiC1O4 with 1:1 by volume mixture of PC and DME
was used as the electrolyte, the upper voltage limit used in
our experiments were 4.3 to 4.4 V. Higher upper voltages
resulted in oxidation of the electrolyte, and we could not
initiate access to the Mn3jMn2, Co37Co2, and Ni3/Ni2°
redox couples in these compounds. However, we could
access the Mn37Mn2 couple in the presence of some iron
atoms in the structure. Solid-solutions LiFe1 Mn2.PO4
with x = 0.25, 0.50, and 0.75 were synthesized. Figure 9
shows linear increases of the lattice parameters with
increasing Mn content in the structure, in accordance with
Vegard's law.
Figure lOa-d show the electrochemical charge and dis-
charge curves for coin-type cells with LiFe12.MnPO4 (x =
0.25, 0.50, 0.75, and 1.0) as the cathode and lithium as the
anode. The charging curve for LiFe0 75Mn025PO4, Fig. lOa,
shows a small plateau at 4.1 V, which is not very distin-
Table I. The space group and lattice parameters of LiFePO4 and
delithiated phase FePO4.
L1FePO4 FePO4
SpaceGroup Pb mn Pb nm
o (A)
b(A)(A)
Volume (As)
6.008 (3)
10.334 (4)
4.693 (1)
291.392 (3)
5.792 (1)
9.821 (1)
4.788 (1)
272.357 (1)
guishable in the discharge curve. For LiFe0 1Mn0 5P04, the
charging curve Fig. lOb shows two distinct plateaus of
almost equal width, and these plateaus are reproducible
on discharge and over repeated cycling. As the Mn content
is increased in the structure, the amount of lithium that
can be electrochemically extracted by charging decreases
as is evident in Fig lOc for LiFe095Mn075PO4. With all the
Fe atoms replaced by Mn atoms as in LiMnPO4, lithium
could not be extracted either electrochemically, Fig. lOd,
or chemically by reacting with NO2PF6 in acetonitrile.
From these observations, we conclude that the Mn' /Mn2
redox couple in phospho-olivines lies 4.1 eV below the
Fermi energy of lithium if the Mn atoms have an Fe atom
as a nearest neighbor. Destabilization in the presence of
iron of the Mn3/Mn2' redox couple from over 4.3 to 4.1 eV
below the Fermi energy of lithium could reflect the Fe3-
O-Mn2 superexchange interaction; the Mn2* level would
be antibonding and the Fe3° level bonding with respect to
this interaction. In LiMPO4 with M = Co and Ni, the
M31M2 redox energies lie well below the highest occupied
molecular orbital of our electrolyte, with the Ni3 '/Ni2
redox couple lying around 0.6 eV below the Co3/Co2
redox couple as in the case of the inverse spinels V[LiM]04
Fig. 7. Schematic representation of the motion of LiFePO4/FePO4
interface on lithium insertion to a particle of FePO4.
C
Li FePO4
— INTERFACE
) unless CC License in place (see abstract). ecsdl.org/site/terms_use address. Redistribution subject to ECS terms of use (see 17.209.11.199Downloaded on 2015-06-11 to IP

J. Electrochem. Soc., Vol. 144, No. 4, ApriL 1997 The Electrochemical Society, Inc. 1193
TGA DSC
100.2
100.0
99.8
Cl)Cl) 99.6
99.4
_ 99.299.098.898.698.498.2
1002
100.0
—99.8
99A
99.2
99.0
98.8
98.6
98.4
TEMPERATURE (°C)
Fig. 8. The thermal stability of delithiated phase FePO4: to) TGA
and DSC curves in nitrogen atmosphere and (b) TGA curve in oxy-
gen atmosphere.
with M Co or Ni. 'Itappears that the greater covalence
of the P04 tetrahedron relative to that of the V04 tetrahe-
dron not only favors the olivine as against the spinal struc-
ture; it also stabilizes4he redox couples at the octahedral
sites by at least 0.4 eV, lowering the Mn3/Mn2 couple
from 3.7 eV below the lithium-anode Fermi energy in
V[LiMnlO4 to 4.1 eV in LiFe0 5Mn0 5po4, 20
Conclusion
Onextraction of lithium from LiFePO4, a flat closed-cir-
cuit voltage (CCV) curve at 0.05 mA/cm2 of 3.5 V vs. lithi-
um is obtained for the Fe3/Fe2 redox couple due to the
presence of two phases, LiFePO4 and FePO4. These phases
belong to the same space group with a variation of the
FePO4 host only in the unit-cell parameters. This material
is very good for low-power applications; at higher current
densities there is a reversible decrease in capacity that, we
suggest, is associated with the movement of a two-phase
interface, a feature characteristic of cathodes that traverse
a two-phase compositional domain in the discharge cycle.
The intercalation of only 0.6 Li atom/formula unit of
LiFePO4 may be an extrinsic problem since the same V(x)
curves are obtained starting with FePO4 and essentially all
the lithium can be extracted chemically from LiFePO4.
The deintercalation of lithium from the solid solution
LiFe, MnO4 (x = 0to 1) allows location of the position
of the Mn3/Mn' redox couple. When x = 0, we get a
plateau at 3.5 V; but as the manganese content is
increased, a plateau at 4.1 V appears. Maximum charging
at 4.1 V is accomplished for x = 0.5. As expected, we
observe that the oxidation of Mn2 occurs only after the
oxidation of Fe'. We were unable to take out any lithium
from LiMnPO4 while charging up to 4.3 V. Moreover, the
width of the 4.1 V plateau decreases with increasing x>
0.5, which suggest that the 4.1 V plateau is associated with
the Mn atoms having Fe near neighbors. It appears that
the Mn-0-Fe interactions raise the energy of the
10.46
10.44
10.38
10.36
19.34
6.10
6.08
I______ 11
0.25 0.50 0.75
x in LiFe1MnPO4
Fig. 9. Variation of lattice parameters of LiFe1_,M;P04 with
increasing Mn content in the structure.
Mn37Mn' couple and slightly lower the Fe3*/Fez+ couple.
We were unable to extract Li from isostructural LiCoPO4
and LiNiPO4 with the electrolyte LiC1O4 in PC and DME
due to the stability of the Co3'/Co2 and Ni3/Ni2 couples.
Acknowledgment
We thank the Robert A. Welch Foundation, Houston,
Texas, for financial support.
Manuscript submitted Sept. 5, 1996; revised manuscript
received Dec. 15, 1996.
The University of Texas at Austin assisted in meeting the
publication costs of this article.
REFERENCES
1. M. S. Whittingham, This Journal, 123, 315 (1976).
2. K. Mizushima, P. C. Jones, P. J. Wiseman, and J. B.
Goodenough, Mater. Res. Bull., 15, 783 (1980),
3. M. G. S. R. Thomas, W. I. F. David, andJ. B. Goodenough,
ibid., 20, 1137 (1985).
4. T Ohzuku, M. Kitagawa, and T. Hirai, This Journal,
137, 769 (1990).
5. J. M. Tarascon and D. Guymard, Electrochim. Acta,
38, 1221 (1993).
6. J. R. Dahn, E. W. Fuller, and U. von Sacken, Solid
State lonics, G9, 265 (1994).
7. M. M. Thackeray, P. J. Johnson, L. A. de Piciotto, P. G.
Bruce, and J. B. Goodenough, Mater. Res Bull., 19,
179, (1984).
8. M. M. Thackeray, A. de Kock, M. H. Rossouw, D. Liles,
R. Bittihu, and D. Hoge, This Journal, 139, 363,
(1992).
9. V. B. Nalbandyan and I. L. Shukaev, Russ. J. Inorg.
Chem, 32, 3, (1987)
10. Y. Takeda, K. Nakahara, M. Nishijima, N. Imanashi, 0.
Yamamoto, M. Takano, and R. Kanno, Mater. Res.
Bull., 29, 659 (1994)
11. A. Le Mehaute, G. Ouvrard, R. Brec, and J. Rouxel,
ibid., 12, 1191 (1977).
12. M. S. Whittingham, Prog. Solid State Chem., 12, 41
(1978)
13. R. Brec and A. Dugast, Mater. Res. Bull., 15, 619
(1980).
E0-JU.I.-4UiI
302300298296294292
.4>
.4C.)
TEMPERATURE (°C)
.4'5 6.06
6.04
6.02
50 100 150 200 250 300 350 400 450 500 550 0.00 1.0
) unless CC License in place (see abstract). ecsdl.org/site/terms_use address. Redistribution subject to ECS terms of use (see 17.209.11.199Downloaded on 2015-06-11 to IP

1194 J. Electrochem. Soc., Vol. 144, No. 4, April 1997 The Electrochemical Society, Inc.
14. S. Okada, K. S. Nanjandaswamy, A. Manthiram, J. B.
Goodenough, H. Ohtsuka, H. Arai, and J. Yamaki, in
Proceedings of 36th International Power Sources
Symposium, p. 110, Cherry Hill, NJ (1994).
15. C. Masquelier, A. K. Padhi, K. S. Nanjandaswamy, and
J. B. Goodenough, J. Solid State Ch.em., Submitted.
16. A. K. Padhi, K. S. Nanjandaswamy, C. Masquelier, and
J. B. Goodenough, in Proceedings of 37th Interna-
tional Power Sources Symposium, p. 180, Cherry
Hill, NJ (1996).
17. G. E. Brown, Ph.D. Thesis, Virginia Polytechnic
Institute, Blacksburg, VA (1970).
18. W. Evantoff, R. Martin, and D. R. Peacor, Am. Mm., 57,
45 (1972).
19. G. T. Fey, Wu Li, J. R. Dahn, This Journal, 141, 2279
(1994).
20. A. K. Padhi, MT. B. Archibald, K. S. Nanjundaswamy, and
J. B. Goodenough, J. Solid State Chem., In press.
020 40Capacity [mAh/g]
60 80 100 120
4.5
3.5
0>3.0
02.5
2.0
Capacity [mAh/gJ
(a)
U
• ••jf
•chagig I
.istCycle
ID 2ndCycI(
3rdCycLe
>
C)t,)0>
C)0
.0 0.1 0.2 0.3 0.4 0.5
x in Li1÷Fe75Mn25PO4
Capacity [mAh/g]
0.6 0.7
x in Li1+Fe05Mn05PO4
Capacity [mAh/g]
C)0)
0
C)0
>C)0)a0>
C)0
Fig. 10. Discharge/charge curves vs. lithium at current densities 2.0 mA/g, (0.05 mA/cm2) for (a) LiFeO75Mnb2SPO4, (b) LIFeD5MnOSPO4, (c)
LiFe0 25Mn0 nPO4, and (dJ LMnPO4.
x in Li1+Fe25Mn75PO4 x in Li1+MnPO4
) unless CC License in place (see abstract). ecsdl.org/site/terms_use address. Redistribution subject to ECS terms of use (see 17.209.11.199Downloaded on 2015-06-11 to IP