Electron Microscopy Study of the LiFePO4 to FePO4 Phase

Transition
Guoying Chen, Xiangyun Song and Thomas J. Richardson

Electrochem. Solid-State Lett. 2006, Volume 9, Issue 6, Pages A295-A298.

doi: 10.1149/1.2192695

Email alerting Receive free email alerts when new articles cite this article - sign up in the box at the
service top right corner of the article or click here

To subscribe to Electrochemical and Solid-State Letters go to:

http://esl.ecsdl.org/subscriptions

© 2006 ECS - The Electrochemical Society

Downloaded on 2012-10-09 to IP 128.143.23.241 address. Redistribution subject to ECS license or copyright; see www.esltbd.org
 Electrochemical and Solid-State Letters, 9 共6兲 A295-A298 共2006兲 A295
1099-0062/2006/9共6兲/A295/4/$20.00 © The Electrochemical Society

Electron Microscopy Study of the LiFePO4 to FePO4 Phase

Transition
Guoying Chen,* Xiangyun Song, and Thomas J. Richardson*,z
Environmental Energy Technologies Division, Lawrence Berkeley National Laboratory, Berkeley,
California, 94720, USA

The mechanism by which LiFePO4 is transformed into isostructural FePO4 has been elucidated using electron microscopy on
large, hydrothermally grown LiFePO4 crystals following chemical delithiation. Lithium is extracted at narrow, disordered transi-
tion zones on the ac crystal surface as the phase boundary progresses in the direction of the a-axis. The substantial lattice
mismatch along a 共ca. 5%兲 causes crack formation in the bc plane. Despite considerable disorder in the transition zone, the general
structural arrangement is preserved, leading to good crystallinity in the newly created FePO4 domains. Implications for improved
electrode performance are discussed.
© 2006 The Electrochemical Society. 关DOI: 10.1149/1.2192695兴 All rights reserved.

Manuscript submitted January 5, 2006; revised manuscript received February 10, 2006. Available electronically April 6, 2006.

The olivine-type lithium iron phosphate LiFePO4 has been the equal to 1:1:3. After stirring under nitrogen for about 5 min, the
subject of a great deal of experimental and theoretical interest since reaction mixture was transferred to a Parr reactor, which was purged
its utility as a positive electrode material in lithium and lithium ion with nitrogen and held at 220°C for 3 h. On cooling to room tem-
batteries was first reported by Pahdi et al.1 The “shrinking core” perature, the off-white precipitate was filtered and thoroughly
model for the charge and discharge processes advanced in that paper washed with deionized water and dried at 40°C for 48 h. Delithiated
has been generally accepted and promoted in one form or another2-4 crystals were obtained by stirring LiFePO4 for 1 h in a 0.05 M
to explain the poor kinetics and low capacities of electrodes contain- solution of bromine in acetonitrile in a molar ratio of 1:4 for
ing large particles of active material. In this model, lithium ions Li0.5FePO4 and 1:1.5 for FePO4.
percolate through LiFePO4 or its delithiated form, FePO4, toward or X-ray diffraction 共XRD兲 patterns were acquired in reflection
away from the phase boundary between these two line phases. Due mode using a Panalytical Xpert Pro diffractometer equipped with
to the exceedingly low mobility of lithium in both phases,5 it has monochromatized Cu K␣ radiation. The scan rate was 0.0025°/s
been suggested that one or more intermediate solid–solution phases from 10 to 70° 2␪ in 0.01° steps. Lattice parameters were deter-
may facilitate the conversion through enhanced Li diffusion and, mined by whole pattern refinement using Riqas Rietveld refinement
possibly, greater electronic conductivity in the presence of a mixture software 共MDI兲. Scanning electron microscopy 共SEM兲 images were
of Fe共II兲 and Fe共III兲 ions.4,6 Recently, Delacourt et al.7 demon- collected using a Hitachi S-4300 SE/N microscope at 25 kV accel-
strated the existence of solid solutions at elevated temperatures. The erating voltage. Transmission electron microscopy 共TEM兲 and high-
conductivity of these materials, however, was not found to be sub- resolution transmission electron microscopy 共HRTEM兲 experiments
stantially greater than those of the end member phases.8 In fact, on were carried out at the National Center for Electron Microscopy
quenching to room temperature, rapid demixing to line phases of 共NCEM兲 at LBNL, using a Philips CM200 field emission micro-
intermediate composition occurs, and these only slowly dispropor- scope and a JEOL 200CX electron microscope operating at 200 kV.
tionate to LiFePO4 and FePO4. Thus, it is unlikely that the mecha- Samples for TEM were ground under ethanol, and the resulting dis-
nism by which LiFePO4 is converted to FePO4 involves crystalline persion was transferred to a holey carbon film fixed on a 3 mm
solid solution intermediates. copper grid. Electron diffraction patterns were collected using the
The movement of lithium ions in the highly anisotropic LiFePO4 selected area electron diffraction 共SAED兲 technique.
and in its delithiated counterpart, FePO4, is confined to tunnels run- A high-quality LiFePO4 composite electrode supplied by Hydro-
ning along the b-axis 共space group Pnma兲.5 In addition, there is a Québec 共IREQ兲 comprised 82 wt % carbon-coated LiFePO4,
layered character to the FePO4 host parallel to the bc plane. These 4 wt % carbon black, 4 wt % graphite, and 10 wt % polyvinylidene
structural features, along with the highly restricted Li motion within difluoride 共PVDF兲 binder on a carbon-coated aluminum current col-
the bulk solids 共a process that is not assisted by any driving force lector. After drying under vacuum at 80°C for 24 h, electrodes
toward homogeneity兲, suggest that, rather than diffusing through the 共1.92 ⫻ 1.92 cm兲 were mounted in a single–compartment three-
crystals, Li is extracted or inserted only at the phase boundary, with electrode cell, with Li foil as reference and counter electrodes, and
Li ions moving in a direction parallel to the boundary. Given the 1 M LiPF6 in 1:1 propylene carbonate:ethylene carbonate electro-
preference for Li diffusion along b and the relative weakness of the lyte. Lithium was extracted potentiostatically at 3.35 V to 0.3% state
bonding between the bc layers relative to that in other directions, the of charge 共SOC兲, at 3.40 V to 0.6% SOC, and then at 3.45 V to 1
bc plane emerges as the preferred candidate for the orientation of the and 1.5% SOC. The remaining steps were carried out at 3.50 V.
phase boundary. This has been confirmed for the case of LiMnPO4.9
Here we present evidence from electron microscopy experiments in Results and Discussion
support of this mechanism.
Synthesis of LiFePO4, “Li0.5FePO4,” and FePO4 crystals.—
The dimensions of the plate-like crystals 共Fig. 1a兲 varied with the
Experimental
concentration of the precursor solution and reaction temperature.
Well-formed crystals of LiFePO4 were synthesized using the hy- Increasing temperature or decreasing concentration produced crys-
drothermal method pioneered by Whittingham et al.10 Equimolar tals of larger plate area and thickness. Reaction at 220°C for 3 h
amounts of FeSO4 共99%, Aldrich兲 and H3PO4 共85%, J. T. Baker兲 produced discrete crystals of uniform size and shape with dimen-
were mixed in deoxygenated and deionized water. A small amount sions 4 ⫻ 2 ⫻ 0.2 ␮m. Longer reaction times appeared to cause
of citric acid was added to the mixture to prevent iron oxidation. A aggregation and macroscopic twinning. Only LiFePO4 and/or FePO4
0.2 M LiOH 共Spectrum兲 solution was added slowly to give Fe:P:Li were detected in the XRD pattern of the fresh and oxidized crystals.
The refined lattice parameters were a = 10.334 Å, b = 6.002 Å, and
c = 4.695 Å for LiFePO4 and a = 9.826 Å, b = 5.794 Å, and c
* Electrochemical Society Active Member. = 4.784 Å for FePO4. The states of charge of the bromine-oxidized
z
E-mail: tjrichardson@lbl.gov samples as determined by XRD were close to those intended, i.e.,

Downloaded on 2012-10-09 to IP 128.143.23.241 address. Redistribution subject to ECS license or copyright; see www.esltbd.org
A296 Electrochemical and Solid-State Letters, 9 共6兲 A295-A298 共2006兲

Figure 2. Electron diffraction patterns of 共a兲 LiFePO4, 共b兲 Li0.5FePO4, and

共c兲 FePO4.

TEM observations.— The large faces of the plate-like crystals lie

in the ac plane, with the long axis parallel to c. Electron diffraction
patterns in this plane for both LiFePO4 共Fig. 2a兲 and FePO4 共Fig. 2c兲
crystals consisted of a single set of spots. Diffraction spots for two
phases were observed 共Fig. 2b兲 for the Li0.5FePO4 crystals, indexed
as LiFePO4 and FePO4. The relative intensity of these two sets
varied substantially with location on the mixed phase crystals. Al-
though the reaction is topotactic, a slight rotation around b 共ca. 2°兲
of the two phases was generally observed, indicative of deformation
and dislocation associated with delithiation. In some cases an addi-
tional set of spots, also corresponding to FePO4, was observed, ro-
tated in the opposite direction, representing a different crystalline
domain, possibly on the opposite crystal face.
Dislocation features were observed in both SEM 共Fig. 1兲 and
TEM images 共Fig. 3兲 of oxidized crystals. They run in the direction
of the c axis and appear to be aligned with the bc plane. The contrast
between alternating dark and light domains about 100 nm wide and
separated by dislocation lines was found to vary or even disappear
when the sample was tilted around the c axis. This suggests that
dislocations and/or rotations may also be present in this direction. In
order to obtain high resolution 共HRTEM兲 images, a much thinner
Li0.5FePO4 crystal 共Fig. 3b兲 was examined. The specimen was
highly crystalline everywhere except within narrow dislocations,
where disordered regions were observed between two crystalline
domains 共Fig. 4兲. Lattice spacings derived from Fourier transforms
共insets兲 of the images from the ordered areas corresponded to those
of LiFePO4 and FePO4. The transition region is not completely dis-
ordered, and probably supports a composition gradient that allows
the two phases to coexist despite the lattice mismatch, while retain-
ing sufficient short-range order to facilitate formation of a crystal-
line reaction product.
The LiFePO4 to FePO4 transition, a nucleation process driven
by the mismatch in lattice parameters.— Lattice mismatch within
crystals often leads to structural defects such as dislocations, which
can be either edge or screw type.12-14 Edge dislocation, the more
common case, generally develops perpendicular to the direction of
greatest mismatch.15 As Li ions leave the parent phosphate during
chemical delithiation, the formation of FePO4 generates strong dis-
location stress fields in the ac plane 共Fig. 5兲 and weaker ones in

Figure 1. SEM images of 共a兲 LiFePO4, 共b兲 Li0.5FePO4, and 共c兲 FePO4.

the Li0.5FePO4 sample had a LiFePO4 to FePO4 ratio of 51:49, and

the fully oxidized sample was 97% FePO4. The reflections due to
FePO4 were somewhat broader than those of the precursor LiFePO4.
Although the shape of the crystals was preserved after oxidation
共Fig. 1b and c兲, some cracks appeared in the partially oxidized crys-
tals, and more cracks developed with further oxidation. Chen et al.11 Figure 3. 共a兲 TEM image showing the domains in Li0.5FePO4 crystal,
observed cracked particles in a cycled LiFePO4 composite electrode, aligned along the c-axis, 共b兲 TEM image of thin Li0.5FePO4 crystal, showing
and these authors related their presence to capacity losses. crack in bc plane.

Downloaded on 2012-10-09 to IP 128.143.23.241 address. Redistribution subject to ECS license or copyright; see www.esltbd.org
 Electrochemical and Solid-State Letters, 9 共6兲 A295-A298 共2006兲 A297

Figure 6. Ex situ XRD patterns of an LiFePO4 electrode at various stages of

electrochemical extraction of lithium 共in percent at left兲.

electronic conductivity as well兲 over that in the crystalline line

phases. This should allow the transition zone to penetrate into the
crystal as it grows along the surface. Because Li is constrained to
move only parallel to b, only the ac plane is active for Li extraction
and insertion.
Phase behavior in LiFePO4 composite electrodes.— During the
early stages of charging a well-connected electrode, no reflections
due to FePO4 were detected until at least 10% of the Li had been
extracted 共Fig. 6兲. As the charging proceeded, however, the relative
Figure 4. HRTEM image of the disordered region at the end of the crack in amounts of the two phases approached their expected levels. Sau-
Fig. 3, with Fourier transforms of the indicated areas.
vage et al.16 have shown that extraction and insertion of lithium in
LiFePO4 thin films occurs only at three-phase boundaries where the
active material, the electrolyte, and the current collector meet. In a
other directions. The dislocation line is parallel to the c axis, and composite electrode, these conditions exist where the conductive
as the reaction proceeds at many points on the crystal surface, matrix is in direct contact with the phosphate phase. Electrodes con-
these dislocations begin to nucleate into an extended phase bound- taining carbon-coated phosphate particles exhibit superior rate per-
ary along c. Local bond stretching and bending in the phase bound- formance, possibly because the phase transition is nucleated at many
ary region may significantly enhance Li ion mobility 共and possibly locations on exposed ac faces. In the early stages, the domain size of
the newly formed phase is too small to be seen by XRD. At later
stages, these domains have grown together and become large
enough to be detected. Although the structure is highly ordered, the
crystallinity of the initial phase cannot be fully recovered.
This nucleation mechanism may also explain the efficacy of
charging LiFePO4 composite electrodes at a high potential.17 The
first nucleation sites will develop where the ac faces are well-
connected to the conducting matrix. If these are sufficient in number
to supply enough lithium to maintain the charging current, growth of
the new phase will be confined to those regions. As higher charging
rates demand more Li ions, however, the increased applied potential
will stimulate nucleation at less well-connected sites, providing an
increased flow of Li ions from the particles. Rapid charging may
also increase the stress at nucleation sites, thereby expanding the
disordered region through which the ions can move.
Implications for battery electrodes.— The electroactive ac faces
of the crystals used in this study represent ca. 85% of the surface
area, while for equiaxial or spherical particles, they represent only
about one-third. Isolated grains within the interior of secondary par-
ticles are often poorly connected to the conducting matrix and may
Figure 5. Illustration of the phase transition zone. be inaccessible to the electrolyte. Although extensive grinding and

Downloaded on 2012-10-09 to IP 128.143.23.241 address. Redistribution subject to ECS license or copyright; see www.esltbd.org
A298 Electrochemical and Solid-State Letters, 9 共6兲 A295-A298 共2006兲

other means of reducing particle sizes are beneficial, use of thin, Assistant Secretary for Energy Efficiency and Renewable Energy,
unagglomerated particles 共plates or needles兲 with large ac faces Office of FreedomCAR and Vehicle Technologies of the U. S. De-
would increase the active area and decrease the diffusion distance partment of Energy under contract no. DE-AC02-05CH11231.
for Li ions, thereby improving both rate capability and utilization of
Lawrence Berkeley National Laboratory assisted in meeting the publica-
the active material. This can be accomplished inexpensively via hy- tion costs of this article.
drothermal synthesis. Uniform carbon coating of the large, smooth
faces of these particles would be easier to achieve than for rough, References
agglomerated particles, and the amount of conducting additives 1. A. K. Padhi, K. S. Nanjundaswamy, and J. B. Goodenough, J. Electrochem. Soc.,
could be reduced relative to that required for fine, spheroidal grains. 144, 1188 共1997兲.
2. A. S. Andersson and J. O. Thomas, J. Power Sources, 97, 498 共2001兲.
Conclusions 3. V. Srinivasan and J. Newman, J. Electrochem. Soc., 151, A1517 共2004兲.
4. A. Yamada, H. Koizumi, N. Sonoyama, and R. Kanno, Electrochem. Solid-State
Large crystals of LiFePO4 were synthesized and the phase evo- Lett., 8, A409 共2005兲.
lution during oxidation of the material was studied by microscopic 5. M. S. Whittingham, Y. Song, S. Lutta, P. Y. Zavalij, and N. A. Chernova, J. Mater.
methods. In a partially oxidized sample, ordered domains of FePO4 Chem., 15, 3362 共2005兲.
spaced between the parent LiFePO4 domains were observed, a mor- 6. S.-Y. Chung, J. T. Bloking, and Y.-M. Chiang, Nat. Mater., 1, 123 共2002兲.
7. C. Delacourt, P. Poizot, J. M. Tarascon, and C. Masquelier, Nat. Mater., 4, 254
phology induced by the stress fields accompanying the lattice pa- 共2005兲.
rameter mismatch. The transformation proceeded in the direction of 8. C. Delacourt and C. Masquelier, Private communication.
the a-axis at dislocation lines running parallel to the c-axis. These 9. C. Delacourt, L. Laffont, R. Bouchet, C. Wurm, J.-B. Leriche, M. Morcrette, J.-M.
observations are consistent with movement of Li ions in the b di- Tarascon, and C. Masquelier, J. Electrochem. Soc., 152, A913 共2005兲.
10. S. Yang, P. Y. Zavalij, and M. S. Whittingham, Electrochem. Commun., 3, 505
rection at the phase boundary, which is disordered but not amor- 共2001兲.
phous. Although the widely accepted shrinking core model clearly 11. D. Wang, X. Wu, Z. Wang, and L. Chen, J. Power Sources, 140, 125 共2005兲.
does not apply to individual crystallites of LiFePO4, its predictions 12. J. W. Matthews and A. E. Blakeslee, J. Cryst. Growth, 27, 118 共1974兲.
regarding longer diffusion paths as charging or discharging proceeds 13. M. Albrecht, H. P. Strunt, P. O. Hansson, and E. Bauser, J. Cryst. Growth, 128,
327 共1993兲.
may be approximated in the case of electrodes containing large, 14. R. Hull, J. C. Bean, D. J. Weder, and R. E. Leibenguth, Appl. Phys. Lett., 52, 1605
agglomerated secondary particles. 共1988兲.
15. D. Hull and D. J. Bacon, in Introduction to Dislocations, 4th ed., Butterworth-
Acknowledgments Heinemann, Boston 共2001兲.
16. F. Sauvage, E. Baudrin, L. Gengembre, and J.-M. Tarascon, Solid State Ionics,
We thank Dr. Xiaofeng Zhang for helpful discussions. We are 176, 1869 共2005兲.
grateful to the National Center for Electron Microscopy at LBNL for 17. A. Yamada, M. Yonemura, Y. Takei, N. Sonoyama, and R. Kanno, Electrochem.
the use of advanced TEM facilities. This work was supported by the Solid-State Lett., 8, A55 共2005兲.

Downloaded on 2012-10-09 to IP 128.143.23.241 address. Redistribution subject to ECS license or copyright; see www.esltbd.org