Hydrated iron phosphates FePO4·nH2O and Fe4(P2O7)3·n H2O as 3 V positive electrodes in rechargeable lithium batteries

Article abstract of DOI:10.1149/1.1489686

Hydrated Fe¹¹¹ phosphates were investigated as positive electrode materials in lithium batteries. Reversible lithium insertion into amorphous and crystalline FePO₄·nH₂O and Fe₄(P₂O₇)_3<

Full text of DOI:10.1149/1.1489686

Journal of The Electrochemical Society, 149 ͑8͒ A1037-A1044 ͑2002͒
A1037
0
013-4651/2002/149͑8͒/A1037/8/$7.00 © The Electrochemical Society, Inc.
Hydrated Iron Phosphates FePO "nH O
4
2
and Fe „P O … "nH O as 3 V Positive Electrodes
4
2
7 3
2
in Rechargeable Lithium Batteries
a, ,z
b
a
a
Christian Masquelier, * Priscilla Reale, Calin Wurm, Mathieu Morcrette,
a
a
Lo ¨ı c Dupont, and Dominique Larcher
aLaboratoire de R e´ activit e´ et Chimie des Solides, Universit e´ Picardie Jules Verne,
0039 Amiens Cedex 9, France
8
b
Dipartimento di Chimica, Universita La Sapienza, 00185 Roma, Italy
Hydrated FeIII phosphates were investigated as positive electrode materials in lithium batteries. Reversible lithium insertion into
amorphous and crystalline FePO •nH O and Fe (P O ) •nH O compositions was found at potentials between 3.5 and 2.5 V vs.
4
2
4
2
7
3
2
ϩ
Li /Li. The roles of ͑i͒ specific surface area, (ii) amorphous vs. crystalline state, (iii) H O content, and (iv) electronic contact
between particles in the composite positive electrode, on the electrochemical performances of these materials are discussed. Very
2
stable cycling was obtained for optimized FePO •1.6H O and Fe (P O ) •4H O electrodes at an average voltage of 3.0 and 3.2
4
2
4
2
7
3
2
ϩ
V vs. Li /Li, respectively.
©
2002 The Electrochemical Society. ͓DOI: 10.1149/1.1489686͔ All rights reserved.
Manuscript submitted September 26, 2001; revised manuscript received February 13, 2002. Available electronically June 24, 2002.
1
2
13
Recently, polyanion three-dimensional ͑3-D͒ structures built of
ported
for
FeVO •nH O,
VOPO •nH O,
or
4
2
4
2
nϪ
14
(
XO4)
tetrahedra and MO octahedra have engendered much in-
6
MnO •xH O, the presence of constitutional and/or surface water
2
2
terest for their potential use as positive electrodes for lithium
in these compounds does not inhibit their electrochemical perfor-
1
batteries. Efforts toward this relatively novel class of intercalation
mances vis- a` -vis lithium insertion/extraction. Additionally,
2
hosts have focused on systems such as the olivine Li1ϪxFePO
Fe͑PO )•nH O compositions are extremely easy to prepare and are
4
4
2
3
and sodium superionic conductor compositions Li Fe (SO )
and
commercially available.
x
3ϩ
2
4 3
4
2ϩ
Li Fe (PO ) where reduction/oxidation of Fe /Fe occurs at
potentials close to 3.45, 3.55, and 2.8 V vs. Li/Li , respectively.
We report here our new findings on the reversible lithium inser-
tion into amorphous and crystalline FePO4•nH2O and
Fe4(P2O7)3•nH2O compositions at potentials between 3.5 and 2 V
3
2
4 3
ϩ
These materials take full benefit from the inductive effect of
nϪ
ϩ
(
XO4) tetrahedral groups that diminish the covalence of the Fe-O
vs. Li /Li. The roles of ͑i͒ specific surface area, (ii) amorphous vs.
3
ϩ
2ϩ
bonds and hence increase the operating voltage of the Fe /Fe
crystalline state, (iii) H2O content, and (iv) electronic contact be-
tween particles in the composite positive electrode, on the electro-
chemical performances of these materials are discussed.
5
redox couple compared to ‘‘simple’’ oxides. Such compounds stand
as serious candidates for the next generation of Li-based polymer or
Li-ion batteries.
The material of choice of the so-called polyanionic structures is
Experimental
without doubt the olivine LiFePO that, since the pioneer work of
4
The
amorphous
character
of
‘‘FePO •2H O,’’
4
2
2
6-11
Padhi, has gained much interest very recently.
Lithium is ex-
‘
‘FePO •4H O,’’ and ‘‘Fe (P O ) •4H O,’’ bought from Aldrich, is
4
2
4
2
7
3
2
tracted from the octahedral sites in a biphasic process which, al-
illustrated by the absence of Bragg peaks in their X-ray diffraction
though very attractive for providing a very flat voltage plateau at
͑
XRD͒ patterns ͑hence not reported here͒. This was fully confirmed
ϩ
3
.45 V vs. Li /Li, is penalizing through a kinetically limited front
through a selected area electron diffraction ͑SAED͒ study ͑Fig. 1͒
where only the transmitted beam is observed on the SAED pattern.
The exact stoichiometry of the hydrated powders was carefully
checked through thermogravimetric-differential thermal analysis
͑TG-DTA͒ experiments ͑Setaram 92-B͒, by heating and cooling the
samples under air flow at 5°C/min between room temperature and
phase migration between LiFePO and FePO . Since the work of
4
4
Padhi, recent optimization of LiFePO has focused on ͑i͒ increasing
4
the electronic conductivity of composite electrodes through carbon
coating on the active material particles6 and (ii) lowering the
synthesis temperature to ϳ400°C with the use of strongly reactive
,11
8
9
00°C. The experimental weight losses, due to water departure, are
of Ϫ16, Ϫ29, and Ϫ9%, respectively, to be compared with Ϫ19,
iron oxalate under N . The difficulty with LiFePO arises from its
2
4
insulating character and from the fact that iron is at the ϩ2 oxidation
Ϫ32, and Ϫ9% for the nominal compositions FePO •2H O,
state. This requires careful synthesis procedures, namely, controlled
4
2
III
FePO •4H O, and Fe (P O ) •4H O, respectively. Therefore, the
atmosphere, to prevent the formation of stable Fe phosphates such
4
2
4
2
7
3
2
III
exact H O content within the two amorphous hydrated forms of
as LiFeP O and/or Li Fe (PO ) and/or amorphous Fe phos-
2
2
7
3
2
4 3
phates if low temperature synthesis is used.
Bearing these considerations in mind, we undertook a general
study of possible lithium insertion into stable Fe phosphates. The
FePO4 investigated in this study are FePO4•1.6H2O and
FePO4•3.4H2O. From the shapes of the TG curves recorded at either
5 or 0.5°C/min, the H2O loss on heating occurs progressively in the
60 to 400°C temperature range, for both FePO •1.6H O and
III
theoretical specific capacities of simple iron phosphates operating on
4
2
3
ϩ
2ϩ
the Fe /Fe couple depends strongly, of course, on the Fe/P ratio,
as summarized in Table I. Among the compositions listed in Table I,
FePO and Fe (P O ) are the most attractive. Besides these com-
FePO •3.4H O. From both of these starting powders, the overall
4
2
product obtained at 400°C is amorphous and of ‘‘FePO ’’ overall
4
composition. Note that a small weight uptake of about ϩ2% is ob-
4
4
2
7 3
III
positions, there are only a few known crystalline Fe compounds
while many condensed hydrated iron phosphates exist, of which
Fe͑PO )•nH O should definitely deserve attention. As recently re-
served for these amorphous FePO powders cooled down and kept
4
at room temperature under ambient air.
For electrochemical comparison purpose, three crystalline iron
4
2
phosphates of compositions Fe (P O ) , ␣-FePO ͑low tempera-
4
2
7
3
4
ture quartz form͒, and FePO •2H O ͑monoclinic form, meta-
4
2
strengite͒, were also prepared and evaluated as positive electrodes in
*
Electrochemical Society Active Member.
E-mail: Christian.Masquelier@sc.u-picardie.fr
z
lithium batteries. Anhydrous crystalline Fe (P O ) , isostructural
4
2
7 3
Downloaded on 2014-12-14 to IP 169.230.243.252 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

A1038
Journal of The Electrochemical Society, 149 ͑8͒ A1037-A1044 ͑2002͒
Table I. Theoretical capacities of iron phosphates operating on
Table II. Space groups and lattice parameters of the three crys-
talline iron phosphates investigated in this paper.
3
¿
2¿
the Fe ÕFe redox couple at potentials between 4 and 2 V vs.
¿
Li ÕLi
a
b
c
␤
͑
Å͒
͑Å͒
͑Å͒
͑°͒
Space group
M
Th. Cap.
͑
g molϪ1͒ ͑mAh/g͒ References
Fe (P O )
7 3
FePO4
FePO4•2H2O
9.562͑2͒ 21.509͑1͒ 7.545͑1͒
5.033͑1͒ 5.033͑1͒ 11.247͑1͒
Pbnm
P321
4
2
Fe/P ϭ 1
LiFePO4
FePO4
157.8
150.8
186.8
222.8
417.4
45.2
817.2
236.7
170
178
143
120
128
144
131
113
1, 2
This paper
This paper
This paper
4
metastrengite 5.327͑2͒ 9.797͑1͒ 8.711͑1͒ 90.53͑3͒
P2 /n
FePO •2H O
1
4
2
FePO •4H O
Li Fe (PO )
3 2 4 3
4
2
calculations18 indicate total surfaces ͑poresϩexternal surface of the
Fe/P ϭ 2/3
Fe/P ϭ 1/2
2
particles͒ of 29 and 18 m /g for amorphous FePO •1.6H O and
Fe (P O )
7 3
7
22
4
2
4
2
2
Fe (P O ) •4H O
This paper
22
FePO •3.4H O, respectively, of which only 2 m /g correspond to
4
2
4
2
7
3
2
LiFeP O
micropores ͑Ͻ20 Å͒. Hence, most of the surface of the particles is
2
7
accessible for lithium intercalation as in the electrolyte used; sol-
ϩ
19
vated Li (Li (PC) was shown to have a diameter close to 12 Å.
4
2
0
The Barrett-Joyner-Halenda ͑BJH͒ method did not reveal any pre-
dominant distribution of mesopores sizes between 20 and 500 Å for
both powders. Note that the protocol of the physisorption experi-
1
5
with V (P O ) , was obtained by heating Fe (P O ) •4H O in a
4
2
7
3
4
2
7
3
2
Pt crucible under air at 900°C; the crystallization occurred at T
Ͼ 600°C. The lattice parameters, refined in the space group Pbnm
ments consists of a first thermal treatment under N flow of the
2
are a ϭ 9.652(2), b ϭ 21.509(1), and c ϭ 7.545(1) Å. ␣-FePO4
was prepared by evaporation of an equimolar aqueous solution of
powder between 120 and 150°C prior to placing it under vacuum for
the measurements. This procedure leads to some surface water de-
sorption that was indeed observed at the surface of the glassy
sample holder.
Fe͑NO ) •9H O ϩ NH H PO and heating under air up to 950°C
3
3
2
4
2
4
for 12 h. All the XRD peaks were indexed in the trigonal space
group P321 with a ϭ 5.033(1) and c ϭ 11.247(1) Å. A crystal-
lized form of FePO •2H O ͑metastrengite, named also
Electrochemical lithium insertion/extraction tests were per-
formed in Swagelok-type cells assembled inside an argon filled dry
box with oxygen and water contents below 10 ppm. The negative
electrode was a disk of lithium metal foil. A Whattman GF/D boro-
silicate glass fiber sheet, saturated with a 1 M LiPF6 salt dissolved in
1:1 ethylene carbonate/dimethyl carbonate solution, was placed be-
tween the two electrodes. The composite positive electrodes con-
tained 16.7 wt % of acetylene black and 83.3% of active material.
Intimate mixing was ensured by ballmilling for various durations,
between 15 and 120 min. The ratio R between the mass of the balls
used and the mass of powder milled was kept constant to a value of
4
2
16
Phosphosiderite͒ was prepared as small pink crystals from precipi-
tation in an aqueous solution of Fe O dissolved in concentrated
2
3
17
H PO . The metastrengite structure consists in a slight monoclinic
3
4
distortion of the strengite structure ͑orthorhombic͒ that belongs to
the wide variety of variscite isotypic series. The lattice parameters of
the metastrengite FePO •2H O of this study were a ϭ 5.327(2),
4
2
b ϭ 9.797(1), c ϭ 8.711(1), and ␤ ϭ 90.53͑3͒ Å in the mono-
clinic space group P2 /n. The lattice parameters of the three crys-
1
talline iron phosphates investigated in this study are listed in Table
II.
17. The electrochemical insertion/extraction was monitored with the
use of a VMP automatic cycling/data recording system ͑Biologic
SA, Claix, France͒ operating in galvanostatic or potentiostatic
modes.
The morphology of these powders were controlled by scanning
electron microscopy ͑SEM, FEG XL30͒ ͑Fig. 2͒ and transmission
electron microscopy ͑TEM, CM12͒. Very fine particles of 100 nm in
average were observed for the amorphous FePO •xH O and
Results and Discussion
4
2
Fe (P O ) •4H O powders. Crystalline FePO •2H O consists of
4
2
7
3
2
4
2
Crystalline FePO .—As shortly described above, the olivine
4
plate-like crystals of 1-5 ␮m in their largest dimension.
LiFePO4 possesses many peculiarities that make it a material of
choice for applications as a positive electrode in lithium recharge-
able batteries. It was therefore tempting to investigate how crystal-
line materials having related formula would behave in similar con-
ditions upon lithium insertion/extraction. Upon lithium extraction
from the olivine LiFePO , the end product is FePO that conserves
Specific surface area were investigated through physisorption
͑
adsorption͒ of N2 at 77 K. Brunauer-Emmett-Teller ͑BET͒
4
4
an olivine-like structure where, due to lithium departure, three-
fourths of the octahedral sites in the distorted hexagonal close-
packed ͑hcp͒ packing, are empty. This form of FePO cannot, to our
4
III
knowledge, be obtained through chemical synthesis via Fe precur-
sors.
Solid
state
reaction
between
iron
precursors
(
(
Fe O , Fe(NO ) •9H O, . . . ) and phosphate precursors
2
3
3
3
2
NH H PO , . . . ) at relatively high temperature lead to the so-
4
2
4
called quartz forms of FePO ͑␣ or ␤͒ where both iron and phos-
4
phorus are in tetrahedral coordination. Precipitation in aqueous so-
lutions under controlled pH leads to hydrated iron phosphates that
would lead, on further thermal annealing, to the same quartz forms
of FePO4 .
As seen in Fig. 3, 0.3 lithium can be inserted into ␣-FePO4
ϩ
during the first discharge at C/20 rate ͑current adjusted for 1 Li to
ϩ
be inserted within 20 h͒, between 3.5 and 2 V vs. Li /Li, out of
which only 0.1 lithium can be removed on the subsequent charge,
leading to a reversible capacity of only 20 mAh/g. The voltage com-
position curve has an S-type shape centered at an average voltage of
3
ϩ
2ϩ
3
V for the Fe /Fe redox couple. Despite this poor electrochemi-
3
Ϫ
Figure 1. SAED pattern of FePO •3.4H O.
cal behavior, the inductive effect of the (PO4) on the Fe-O bonds
4
2
Downloaded on 2014-12-14 to IP 169.230.243.252 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, 149 ͑8͒ A1037-A1044 ͑2002͒
A1039
Figure 3. Galvanostatic ͑C/20 rate͒ response between 4 and 2 V vs. Li for
crystalline ␣-FePO4 .
is once more observed as the position of the delivered voltage is
found at a pretty high value. Such limited insertion process is not
surprising since ͑i͒ the crystal structure adopted by ␣-FePO does
4
not permit rapid Li diffusion, (ii) Li insertion requires the reduction
III
II
of Fe to Fe in tetrahedral coordination that is unfavorable from
steric considerations, and (iii) crystalline ␣-FePO is mostly insu-
4
lating with particles of ϳ20 ␮m in average size ͑Fig. 2a͒.
Amorphous and crystalline FePO •2H O.—From the examina-
4
2
tion of the theoretical specific capacities of various Fe-based phos-
phates electrodes listed in Table I, one observes a 20% theoretical
capacity penalty on going from FePO4 ͑178 mAh/g͒ to
FePO •2H O ͑143 mAh/g͒ with still attractive capacity for the hy-
4
2
drated form. Then we investigated this hydrated form where iron is
in octahedral environment. Several crystalline forms of
1
7
FePO •2H O have been reported in the literature, generally ob-
4
2
tained by precipitation in aqueous media under controlled pH con-
ditions. We investigated the metastrengite form ͑monoclinic͒ of
FePO •2H O that appears as pinkish crystals of 1-5 ␮m in their
4
2
largest dimension ͑Fig. 2b͒. A commercially available ͑Aldrich͒
amorphous FePO •2H O, powder whose exact content of water was
4
2
1
.6 H O per Fe ͑see the Experimental section͒ was also investigated.
2
As described above, this amorphous powder is characterized by
small particles of 100 nm ͑Fig. 2c͒ and by a high accessible surface
2
area of ϳ26 m /g. Both powders were mixed with acetylene black
͑AB͒ carbon in the proportions 83/17 in wt %. 240 mg of these
mixtures were then ballmilled for 15 min in a stainless steel con-
tainer with a ball of 4 g to obtain the so-called composite electrodes
that were investigated for possible lithium insertion.
Preliminary room temperature voltage/composition curves ob-
tained for these two materials under equivalent cycling conditions
͑C/10 ϭ 1 Li in 10 h͒ are plotted in Fig. 4. Both materials show
electrochemical activity toward Li with an S-type shape voltage pro-
ϩ
file centered at around 3 V vs. Li /Li, similar to that previously
observed for crystalline ␣-FePO . Due to the mostly insulating
4
character of these materials, it is pretty difficult though from ‘‘stan-
dard’’ cell configuration and composite electrode preparation, to
reach theoretical capacities. The information behind the data of Fig.
4
is that, for iron phosphate materials in general, much care has to
be given to the crystallinity, particle size, and texture of the pristine
material particles and on the composite electrode preparation prior
to electrochemical measurements. On the basis of parallel studies
that we are pursuing on the electrochemical behavior of LiFePO
4
2
1
Figure 2. SEM photographs of selected powders investigated in this study.
and Li Fe (PO ) , optimization of ballmilling conditions to pre-
3 2 4 3
Downloaded on 2014-12-14 to IP 169.230.243.252 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

A1040
Journal of The Electrochemical Society, 149 ͑8͒ A1037-A1044 ͑2002͒
Figure 6. Discharge and charge curves of an optimized
ϩ
Figure 4. Galvanostatic ͑C/10 rate͒ response between 4.3 and 2 V vs. Li /Li
for crystalline FePO •2H O and amorphous FePO •1.6H O.
FePO •1.6H O/C composite electrode at different rates.
4
2
4
2
4
2
C/5 with only a small increase of the polarization of the cell. As the
ballmilling does not reduce the size of the pristine active material
nor reduces its crystallinity ͑it is already amorphous͒ it appears then
that its major influence is on producing a more efficient conductive
carbon coating at the surface of the iron phosphate particles. A better
way to evaluate the rate capabilities of electrode materials is to
perform so-called signature curves ͑Fig. 7͒. For three cells made of
composite positive electrodes ballmilled for 15, 60, and 120 min,
respectively, such curves were obtained by first realizing an entire
discharge/charge cycle at C/20 rate and then applying successive
pare composite positive electrodes of active material ͑AM͒ and AB
was undertaken.
Figure 5 shows the first discharge/charge profiles for a series of
AM/AB composites as a function of the ballmilling time. Note that
increasing the ballmilling time leads to an important increase of the
reversible capacities. For our optimized composite electrode, i.e.,
that obtained after 120 min of ballmilling, 90% of the theoretical
capacity was achieved at a cycling rate of C/20 at room temperature
for the first discharge. An irreversible capacity of approximately 0.1
Li was observed during the first charge when the cutoff voltage was
set to 4 V for limiting possible electrolyte decomposition. Note that
such irreversible capacity is constant whatever the milling time
while the reversible part increases. This contrasts favorably with
other known Li intercalation compounds that operate at higher volt-
ages and for which finely divided particles tend to promote electro-
lyte decomposition.
2
C, C, C/2, C/4, . . . , C/32 rates on discharge down to the same
ϩ
cutoff voltage of 2 V vs. Li /Li separated by rest periods of 30 min.
Immediately apparent is the higher rate capability for the 120 min
ballmilled sample compared to the two others. In light of these
results, capacity retention tests were performed for optimized com-
posite electrodes. Figure 8 shows the electrochemical performances
Figure
6 shows the response of such an amorphous
FePO •1.6H O/carbon optimized electrode for various rates of
4
2
discharge/charge. Although the theoretical capacity is not reached at
a slow rate, only 10% of the capacity is lost on going from C/20 to
Figure 7. Signature plots of three FePO •1.6H O/C composite electrodes.
discharge/charge cycle at C/20.
4
2
Figure 5. Effect of ballmilling on the performance of FePO •1.6H O/C
The discharge capacities, plotted vs. scan rates, were obtained after a first full
4
2
composite electrodes ͑83/17 in wt %͒, cycled at C/20 rate.
Downloaded on 2014-12-14 to IP 169.230.243.252 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, 149 ͑8͒ A1037-A1044 ͑2002͒
A1041
Figure 8. Electrochemical cycling between 4 and 2 V vs. Li, of optimized
FePO •1.6H O/C composite electrodes at C/5 scan rate at room temperature.
Figure 10. XRD patterns ͑Cu K␣͒ of the pristine crystalline
FePO4•2H2O and of the composite electrode obtained after 90 min of ball-
milling.
4
2
of such ballmilled composite electrodes. Note the excellent capacity
retention when cycled at C/5 regime at room temperature. Comple-
mentary tests in different cell configurations ͑coin cells, Bellcore
plastic films͒ at different temperatures are underway and will be
reported elsewhere.
maintained, although less pronounced, on subsequent cycling. This
suggested a partial amorphization of the active material: we con-
firmed such an hypothesis by in situ XRD performed on a cell op-
erated over an entire discharge/charge cycle, that has shown, besides
evidence for a reversible two-phases process, that the crystallinity of
the product obtained at the end of charge was significantly reduced.
For crystalline FePO •2H O, excellent cycling retention was also
At this point we wondered if we could enhance the electrochemi-
cal activity toward lithium of crystalline FePO •2H O as we did for
4
2
4
2
the amorphous FePO •2H O. A same sequence of experiments was
4
2
observed with, interestingly, a concomitant increase of both the dis-
charge and charge capacities, reaching 100 mAh/g after 60 cycles at
C/5 ͑Fig. 11͒. Such an increase in capacity on cycling may be remi-
niscent of an electrochemically driven grinding of the sample. As
the particles are broken, the Li diffusion path is shortened therefore
enhancing the full utilization of this material especially when ionic
conduction is limited.
undertaken and after a survey of various milling conditions; 90 min
was found to be the optimum time for our sample. The performance
of such crystalline FePO •2H O/C composite electrode cycled at
4
2
room temperature at C/20 is shown in Fig. 9. Note the drastic im-
provement from the data reported in Fig. 4 since 70% of the theo-
retical capacity is reached this time. From the scanning electron
microscope images of Fig. 2b and d and from the XRD patterns of
Fig. 10, it is clear that the ballmilling has decreased considerably the
particle sizes, down to values close to 500 nm, and that the sample,
however, still remains crystalline. The first discharge voltage profile
of crystalline FePO •2H O is somewhat different from that of amor-
Dehydration of amorphous FePO •nH O compositions.—
4
2
Besides the strong impact of the particle sizes and of the intimate
coating with conductive carbon on the electrochemical properties of
FePO •2H O, we also found through a series of experiments that
4
2
4
2
phous FePO •2H O with, for the former, the existence of an inter-
the presence of H O molecules on and or within the solid was an-
4
2
2
ϩ
calation plateau situated at 2.8 V vs. Li /Li that is more or less
other key parameter for their electrochemical activity toward
lithium. At first sight, it could be surprising that H O-containing
2
electrode materials can display such a steady capacity retention
ϩ
Figure 11. Electrochemical cycling between 4 and 2 V vs. Li /Li, of opti-
ϩ
Figure 9. Galvanostatic ͑C/20 rate͒ response between 4 and 2 V vs. Li /Li
mized crystalline FePO •2H O/C composite electrodes at C/5 scan rate at
4
2
of an optimized crystalline ͑metastrengite͒ FePO •2H O/C electrode.
room temperature.
4
2
Downloaded on 2014-12-14 to IP 169.230.243.252 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

A1042
Journal of The Electrochemical Society, 149 ͑8͒ A1037-A1044 ͑2002͒
Table III. TGA, XRD, and BET characterization of solids ob-
tained from annealing FePO "2H O at various temperatures.
4
2
BET
Temperature
°C͒
͑total surface͒
2
͑
Formula
XRD
͑m /g͒
Pristine
FePO •1.6H O
Amorphous
Amorphous
Crystalline
Crystalline
29
28
3
4
2
400
600
800
FePO •0.3H O
4
2
FePO
4
FePO4
Ͻ1
when cycled vs. Li. We remember however that these materials are
electrochemically active within the 2.8-3.4 V region which falls
within the lower voltage domain of water stability. Additionally, this
suggests that the nature of H O within the solid is more
2
constitutional-like than surface-like. It should be noted too that wa-
ter may act as a media to promote ionic conductivity through a
solvatation process.
ϩ
Figure 13. Galvanostatic ͑C/20 rate͒ response between 4 and 2 V vs. Li /Li
To get insight into how the presence of water would influence the
electrochemical activity with lithium of these materials, a series of
annealing experiments of amorphous FePO •1.6H O at tempera-
of an optimized amorphous FePO •3.4H O/C electrode.
4
2
4
2
tures ranging between 400 and 800°C was undertaken. As seen in
Table III, remaining H O is detected in and/or on the amorphous
2
solid we obtained after annealing at 400°C, due mostly to water
reuptake at room temperature from ambient moisture. The total sur-
firmed yet, the presence of H O within these powders seems to have
a positive effect on their electrochemical properties.
2
face area of amorphous FePO obtained at 400°C, calculated by the
4
This is further supported by the observed reversible capacity of
BET method, is still close to that of the pristine material while it has
amorphous FePO •3.4H O plotted in Fig. 13. For same rates of
decreased considerably for the crystalline FePO powders obtained
4
2
4
charge/discharge and composite electrode preparations ͑40 min in
this case͒, more Li can be inserted per Fe in FePO4•3.4H2O than in
FePO •1.6H O ͑see Fig. 5͒. However, the specific capacity, in
after annealing at 600 and 800°C. As mentioned in the Experimental
section, the surface area of the pristine material is actually that of a
partially dehydrated material: hence, the values of pristine amor-
phous FePO •1.6H O and amorphous FePO are fairly close. The
4
2
mAh/g, is slightly lower due to the increase in molecular weight on
going from 1.6H O to 3.4H O per Fe atom. Both amorphous com-
4
2
4
voltage profile curves of Fig. 12 were obtained for four composite
electrodes prepared the same way ͑1 h of ballmilling with 16.7% of
carbon black͒ cycled at C/20 at room temperature. The big crystal-
2
2
pounds show very similar signature curves. It would be interesting
to complete these observations by following the behavior of a whole
series of FePO •nH O compositions with controlled intermediate
values of n.
line particles of crystalline FePO ͑prepared at 800°C͒ behave con-
4
4
2
sistently much worse than the amorphous and highly divided pris-
tine amorphous FePO •1.6H O. More interestingly, a significant
4
2
Fe (P O ) and Fe (P O ) •4H O.—Possible electrochemi-
decrease in the discharge capacity is also observed on going from
4
2
7
3
4
2
7
3
2
cal insertion of lithium into Fe (P O ) had already been mentioned
the pristine amorphous FePO •1.6H O to the amorphous FePO ob-
4
2
7 3
4
2
4
and discussed in terms of the relatively high position of the
tained at 400°C. As indicated in Table III, the main difference be-
3
ϩ
2ϩ
22
Fe /Fe redox couple vs. Li. Both theoretical capacity ͑144
tween these two powders resides in the H O content, because the
2
Ϫ1
mAh g ͒ and operating voltage ͑3.1 V͒ make it an interesting posi-
specific surface area is essentially the same. Although not fully con-
III
tive electrode material if the 4Fe atoms per formula can be re-
II
duced to Fe . Only partial reduction was reached though, with 3 Li
reversibly inserted per amorphous Fe (P O ) which, nevertheless,
4
2
7 3
20
showed much better response than did crystalline Fe (P O ) . For
4
2
7 3
the present study, we used as the pristine material an amorphous
powder of Fe (P O ) •4H O ͑Aldrich͒ whose total surface calcu-
4
2
7
3
2
2
Ϫ1
lated by BET was of 5 m g and which, on annealing up to 900°C,
led to pure and well-crystallized Fe (P O ) particles, isostructural
4
2
7 3
1
5
with V (P O ) .
4
2
7 3
The general trends described in the previous sections for
the electrochemical behavior of FePO •nH O apply for
4
2
Fe (P O ) •nH O as well: ͑i͒ the electrochemical reactivity with Li
4
2
7
3
2
is much improved through ballmilling of the active material with
conductive carbon, as seen in Fig. 14 for amorphous
Fe (P O ) •nH O ballmilled for 15 and 90 min. (ii) On annealing
4
2
7
3
2
the amorphous Fe (P O ) •4H O powder, departure of water, and
4
2
7
3
2
then particle growth and crystallization occur that lead to a solid
with much less electrochemical activity than for the pristine powder
͑compare Fig. 14 and 15͒. (iii) The shapes of the voltage/
composition curves are of S-type with a smaller polarization as the
duration of the ballmilling increases. The difference between amor-
phous Fe (P O ) •4H O and crystalline Fe (P O ) is seen clearly
Figure 12. Effect of annealing temperature on the electrochemical response
of FePO •nH O/C composite electrodes at C/20 scan rate at room
4
2
4
2
7
3
2
4
2
7 3
temperature.
in the signature plots of Fig. 16.
Downloaded on 2014-12-14 to IP 169.230.243.252 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, 149 ͑8͒ A1037-A1044 ͑2002͒
A1043
Figure 14. Effect of ballmilling on the electrochemical behavior of amor-
phous Fe (P O ) •4H O/C composite electrodes ͑83/17 in wt %͒, cycled at
4
2
7
3
2
C/5 rate.
Figure 16. Signature plots of Fe (P O ) •nH O/C composite electrodes.
discharge/charge cycle at C/20.
4
2
7
3
2
The discharge capacities, plotted vs. scan rates, were obtained after a first full
Conclusions
Besides the widely investigated LiFePO and Li Fe (PO ) elec-
4
3
2
4 3
trode materials, no studies had been reported so far on the possible
reversible insertion of lithium into amorphous hydrated Fe phos-
III
eters such as powder crystallinity, morphology, texture, and elec-
trode preparation, all the materials investigated here, amorphous or
not, present a very similar operating voltage vs. Li, i.e., comprised
mainly between 3.4 and 2.6 V. As for crystalline LiFePO4 and
phates. Such compounds are easily prepared from precipitation in
aqueous media which leads to powders of submicrometer particle
size with high surface area. From the experiments developed in this
paper on a large series of compositions, amorphous or crystalline,
3
ϩ
2ϩ
Li Fe (PO ) , the position of the Fe /Fe redox couple is thus
3
2
4 3
III
hydrated or not, the poor electronic conduction of Fe phosphates
strongly modified by the inductive effect of the PO4 groups that
diminish the strength of the Fe-O bonds, even for amorphous com-
positions. Galvanostatic intermittent titration technique ͑GITT͒ plots
of amorphous FePO •2H O and Fe (P O ) •4H O are reproduced
may be partly compensated by selecting highly divided powders
intimately mixed, or ‘‘coated’’ ͑through ballmilling͒ with conductive
carbon. The beneficial role of constitutional water, that may act as a
promoter for faster ionic diffusion has been demonstrated. Within
the operating voltage window used ͑2-4 V vs. Li͒ these water mol-
ecules look unaffected as very good electrochemical stability was
demonstrated for extensive cycling. This stability suggests that, as
for the crystalline form of FePO •2H O, where the sixfold coordi-
4
2
4
2
7
3
2
in Fig. 17 and 18, respectively. Both show S-shape voltage profiles
without any well-defined intercalation plateau over the whole com-
3
ϩ
2ϩ
position ranges. The position of the Fe /Fe redox couples for
both materials were then determined, after a first full cycle in GITT
mode, by measuring the equilibrium potential after relaxing the bat-
tery for two days at half state of discharge. Under these conditions,
the Nernst global equation Eeq. ϭ E° ϩ (2.3RT/nF)•ln(x/1 Ϫ x)
becomes Eeq. ϭ E°. The obtained values are 3.0 and 3.2 V for
4
2
nation of Fe is ensured by four oxygens belonging to PO groups
4
ϩ2 oxygens belonging to H O groups, a similar local arrangement
2
occurs for the amorphous FePO •nH O powders. Despite a whole
4
2
panel of kinetic responses that depend strongly on extrinsic param-
Figure 17. Discharge and charge curves under GITT mode for amorphous
FePO •1.6H O. For each step, the current was set to a value corresponding
Figure 15. Effect of ballmilling on the electrochemical behavior of crystal-
line Fe (P O ) /C composite electrodes ͑83/17 in wt %͒, cycled at C/5 rate.
4
2
to a C/10 rate for 1 h, followed by a relaxation period of 2 h.
4
2
7 3
Downloaded on 2014-12-14 to IP 169.230.243.252 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

A1044
Journal of The Electrochemical Society, 149 ͑8͒ A1037-A1044 ͑2002͒
Universit e´ Picardie Jules Verne assisted in meeting the publication costs
of this article.
References
1
2
3
. J. B. Goodenough, A. K. Padhi, C. Masquelier, and K. S. Nanjundaswamy, U.S.
Pat. 8,840,523 ͑1997͒.
. A. K. Padhi, K. S. Nanjundaswamy, and J. B. Goodenough, J. Electrochem. Soc.,
1
44, 1188 ͑1997͒.
. S. Okada, K. S. Nanjundaswamy, A. Manthiram, and J. B. Goodenough, in Pro-
ceedings of the 36th International Power Sources Conference, Cherry Hill, NJ,
June 6-9, 1994, IEEE ͑1994͒.
4
. C. Masquelier, A. K. Padhi, K. S. Nanjundaswamy, and J. B. Goodenough, J. Solid
State Chem., 135, 228 ͑1998͒.
5
6
. A. Manthiram and J. B. Goodenough, J. Power Sources, 26, 403 ͑1989͒.
. N. Ravet, J. B. Goodenough, S. Besner, M. Simoneau, P. Hovington, and M. Ar-
mand, Abstract 127, The Electrochemical Society Meeting Abstracts, Vol. 99-2,
Honolulu, HI, Oct 17-22, 1999.
7
8
9
. A. Andersson, J. O. Thomas, B. Kalska, and L. Haggstr o¨ m, Electrochem. Solid-
State Lett., 3, 66 ͑2000͒.
. A. Yamada, S. C. Chung, and K. Hinokuna, J. Electrochem. Soc., 148, A224
͑
2001͒.
. N. Ravet, Y. Chouinard, J. F. Magnan, S. Besner, M. Gauthier, and M. Armand, J.
Power Sources, 97, 503 ͑2001͒.
1
0. F. Croce, F. S. Fiory, and B. Scrosati, Electrochem. Solid-State Lett., 4, A121
2001͒.
Figure 18. Discharge and charge curves under GITT mode for amorphous
Fe (P O ) •4H O. For each step, the current was set to a value correspond-
͑
4
2
7
3
2
11. H. Huang, S. C. Yin, and L. F. Nazar, Electrochem. Solid-State Lett., 4, A170
͑2001͒.
1
ing to a C/10 rate for 1 h, followed by a relaxation period of 2 h.
2. P. Poizot, E. Baudrin, S. Laruelle, L. Dupont, M. Touboul, and J. M. Tarascon,
Solid State Ionics, 138, 31 ͑2000͒.
1
3. N. Dupr e´ , J. Gaubicher, T. Le Mercier, G. Wallez, J. Angenault, and M. Quarton,
Solid State Ionics, 140, 209 ͑2001͒.
14. S. Bach, J. P. Pereira-Ramos, and N. Baffier, Solid State Ionics, 80, 151 ͑1995͒.
5. K. K. Palkina, S. I. Maksimova, S. I. Chibiskova, K. Schlesinger, and G. Ladwig,
Z. Anorg. Allg. Chem., 529, 89 ͑1985͒.
amorphous FePO •2H O and Fe (P O ) •4H O, respectively. This
4
2
4
2
7
3
2
difference may arise from different local environments around Fe
atoms. In this regard, it is interesting to note that the higher position
1
3
ϩ
2ϩ
of the Fe /Fe redox couple in Fe (P O ) •4H O ͑Fe/P ϭ 2/3͒
4
2
7
3
2
1
1
6. H. Strunz, Mineralogische Tabellen, Vol. 7, Auflage, Leipzig, Germany ͑1978͒.
may be related to the fact that Fe is statistically surrounded by more
´
´
7. P. Remy and A. Boulle, C. R. Acad. Sci. Paris, 253, 2699 ͑1961͒.
PO groups than it is in FePO •2H O ͑Fe/P ϭ 1͒: it is therefore
18. P. H. Brunauer, P. H. Emmett, and E. Teller, J. Am. Chem. Soc., 60, 309 ͑1938͒.
4
4
2
1
9. M. Endo, T. Maeda, T. Takeda, Y. J. Kim, K. Koshiba, H. Hara, and M. S. Dressel-
haus, J. Electrochem. Soc., 148, A910 ͑2001͒.
more affected by the inductive effect.
2
2
2
0. E. P. Barrett, L. G. Joyner, and P. C. Halenda, J. Am. Chem. Soc., 73, 373 ͑1951͒.
1. M. Morcrette, C. Wurm, and C. Masquelier, Solid State Sci., 14, 239 ͑2002͒.
2. A. K. Padhi, K. S. Nanjundaswamy, C. Masquelier, S. Okada, and J. B. Good-
enough, J. Electrochem. Soc., 144, 1609 ͑1997͒.
Acknowledgments
Many thanks to Professor J. M. Tarascon and Dr. P. Poizot for
fruitful discussions and constructive critiques of the manuscript.
Downloaded on 2014-12-14 to IP 169.230.243.252 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).