View Online / Journal Homepage / Table of Contents for this issue
PCCP
Dynamic Article Links
Cite this: Phys. Chem. Chem. Phys., 2011, 13, 5171–5177
PAPER
www.rsc.org/pccp
Influence of particle size on solid solution formation and phase interfaces
in Li0.5FePO4 revealed by 31P and 7Li solid state NMR spectroscopy
Downloaded by University of Waterloo on 07 March 2012
Published on 07 February 2011 on http://pubs.rsc.org | doi:10.1039/C0CP01922D
L. J. M. Davis,a I. Heinmaa,b B. L. Ellis,c L. F. Nazarc and G. R. Goward*a
Received 23rd September 2010, Accepted 11th January 2011
DOI: 10.1039/c0cp01922d
Here we report the observation of electron delocalization in nano-dimension
xLiFePO4:(1 x)FePO4 (x = 0.5) using high temperature, static, 31P solid state NMR.
The 31P paramagnetic shift in this material shows extreme sensitivity to the oxidation state
of the Fe center. At room temperature two distinct 31P resonances arising from FePO4 and
LiFePO4 are observed at 5800 ppm and 3800 ppm, respectively. At temperatures near 400 1C
these resonances coalesce into a single narrowed peak centered around 3200 ppm caused by the
averaging of the electronic environments at the phosphate centers, resulting from the
delocalization of the electrons among the iron centers. 7Li MAS NMR spectra of nanometre sized
xLiFePO4:(1 x)FePO4 (x = 0.5) particles at ambient temperature reveal evidence of Li
residing at the phase interface between the LiFePO4 and FePO4 domains. Moreover, a new broad
resonance is resolved at 65 ppm, and is attributed to Li adjacent to the anti-site Fe defect.
This information is considered in light of the 7Li MAS spectrum of LiMnPO4, which despite
being iso-structural with LiFePO4 yields a remarkably different 7Li MAS spectrum due to the
different electronic states of the paramagnetic centers. For LiMnPO4 the higher 7Li MAS
paramagnetic shift (65 ppm) and narrowed isotropic resonance (FWHM E 500 Hz) is attributed
to an additional unpaired electron in the t2g orbital as compared to LiFePO4 which has
diso = 11 ppm and a FWHM = 9500 Hz. Only the delithiated phase FePO4 is iso-electronic
and iso-structural with LiMnPO4. This similarity is readily observed in the 7Li MAS spectrum of
xLiFePO4:(1 x)FePO4 (x = 0.5) where Li sitting near Fe in the 3+ oxidation state takes on
spectral features reminiscent of LiMnPO4. Overall, these spectral features allow for better
understanding of the chemical and electrochemical (de)lithiation mechanisms of LiFePO4 and the
Li-environments generated upon cycling.
Introduction
The olivine LiFePO4 has received a significant attention as a
cathode material for Li-ion battery applications due to its high
theoretical capacity, low cost and low environmental impact as
compared to many alternative cathode materials.1,2 LiFePO4
cycles with exceptional reversibility due to its high thermal and
electrochemical stability resulting from the strong covalent
bonding of oxygen within the [PO4]3 moieties.3 Despite the
enhanced properties seen in LiFePO4, the inherently low
conductivity of this material necessitates the use of nanocrystallite morphology and conductive additives.4–6 This has
motivated many researchers to unravel the limits to the
a
Department of Chemistry and Brockhouse Institute for Materials
Research, McMaster University, 1280 Main St. W. Hamilton, ON,
L8S 4M1 Canada. E-mail: [email protected];
Fax: +1 905-522-2509; Tel: +1 905-525-9140 x 24176
b
National Institute of Chemical Physics and Biophysics, Akadeemia
tee 23, 12618 Tallinn, Estonia
c
Department of Chemistry, University of Waterloo, 200 University
Ave. W. Waterloo, ON, N2L 3G1 Canada
This journal is
c
the Owner Societies 2011
conductivity and take steps toward creating a faster cycling
material while retaining the structural and economical
features that initially made LiFePO4 so attractive. To date,
the conductivity restrictions are ascribed to a one-dimensional
Li-ion diffusion pathway coupled with limited electronic
conduction owing to small polaron hopping conductivity.7–9
This results in a two-phase redox reaction during electrochemical cycling and a voltage plateau corresponding to the
phase transition between them.1,10–14 In order to enhance the
cycling capabilities of LiFePO4, generation of single phase or
‘‘solid-solution’’ properties at temperatures amenable to the
operation of the Li-ion battery has been investigated either
through chemical or mechanical manipulation of the parent
compound.13–18
Solid state NMR is a useful technique for investigating solid
solution formation in LixFePO4 as 7Li NMR is able to directly
monitor the mobile ion while 31P NMR can measure changes
to the host-framework upon Li removal and reinsertion.
The NMR spectra of these materials are complicated from
the strong through-bond (Fermi contact, measured through
Phys. Chem. Chem. Phys., 2011, 13, 5171–5177
5171
Downloaded by University of Waterloo on 07 March 2012
Published on 07 February 2011 on http://pubs.rsc.org | doi:10.1039/C0CP01922D
View Online
the hyperfine coupling) and through-space interactions
between the nucleus under investigation and the unpaired
electrons sitting on the paramagnetic Fe-metal center.19–21
These interactions prove to be invaluable for monitoring
subtle changes in the ionic and electronic environment in this
material. In particular, the 31P paramagnetic shift proves to be
an ideal tool for monitoring electronic changes taking place at
the Fe redox center as it has a very strong hyperfine coupling
to the unpaired electrons on the transition metal center.22
Previous NMR studies of LiFePO4 have characterized the 7Li
and 31P paramagnetic shifts of LiFePO4 and its delithiated
form FePO4, with particular attention paid to the hyperfine
interactions of the Li and P nuclei.22,23 NMR studies have also
focused on thermal treatments in air24 and solid solution
phases generated through anionic doping.25 The most recent
NMR report focused on solid solution formation in samples
with composition of Li0.6FePO4 and Li0.34FePO4 which were
subjected to different heat treatments to preserve solid solution
domains for 31P and 7Li MAS NMR experiments.26
The present work evaluates the temperature driven
formation of solid solution phases in samples of partially
delithiated LixFePO4 using static 31P variable temperature
NMR experiments. Hydrothermally prepared samples of
LiFePO4 are the focus of this investigation as this synthetic
method allows for the preparation of pure, carbon-free single
phase materials with nanometric particle sizes. The use of
static NMR measurements in this case provides access to a
broad temperature range, such that the phase transformation
can be performed in situ providing a complementary NMR
picture to the previous in situ diffraction and Mossbauer
studies of LixFePO4.14,15 For this study, Li0.5FePO4 was the
chosen composition as it allows for well-balanced 31P signal
intensity from the separate FePO4 and LiFePO4 domains.
Here we observed the transition from a room temperature
xLiFePO4 + (1 x)FePO4 (x = 0.5) (herein referred to
as biphasic Li0.5FePO4) compound to a high temperature
(400 1C) single-phase Li0.5FePO4 solid solution. Lastly, a 31P
static and 7Li MAS NMR analysis of the phase boundary
region between the FePO4 and LiFePO4 domains is presented.
This region is of particular interest for uncovering the limits to
ionic and electronic mobilities.
Experimental
Synthesis of LiFePO4 and biphasic Li0.5FePO4
Naturally abundant olivine LiFePO4 (92.58% 7Li, 7.42% 6Li)
samples were prepared via a modified hydrothermal synthesis
first described by Yang et al.27 Briefly, (NH4)2Fe(SO4)26H2O
(Alfa Aesar), H3PO4 (Fisher), LiOHH2O (Alfa Aesar) and
ascorbic acid (Sigma Aldrich) were combined in a
1 : 1 : 3 : 0.1 molar ratio and the reagents were stirred in a
sealed 45 ml Parr autoclave at 190 1C for 8 h. The ascorbic
acid acts as an in situ reducing agent.28 The solid product was
filtered and dried under vacuum at 90 1C for 18 h. The
as-prepared LiFePO4 was oxidized to various compositions
of xLiFePO4:(1 x)FePO4 (x = 0.5, 0.0) using xNOBF4. The
reagents were stirred in acetonitrile under an inert atmosphere
for one hour. X-Ray diffraction was performed on a Bruker
5172
Phys. Chem. Chem. Phys., 2011, 13, 5171–5177
D8-Advantage powder diffractometer using Cu-Ka radiation
(l = 1.5405 Å) from 2y = 10 to 701 at a count rate of
1 second per step of 0.021. The reflections were indexed using
DICVOL91 software. LiFePO4 samples were gold coated and
examined in a LEO 1530 field emission scanning electron
microscope (FESEM) instrument equipped with an energy
dispersive X-ray spectroscopy (EDX) attachment. Images
were recorded at 15 kV with a back-scatter electron detector.
Solid state NMR
All MAS NMR measurements were carried out at McMaster
University on a Bruker AV300 spectrometer. 31P and 7Li
MAS NMR spectra were acquired at Larmor frequencies of
121.8 MHz and 116.64 MHz, respectively. Room temperature
MAS measurements were made using a custom built 1.8 mm
probe with spinning speeds of 25 or 40 kHz. 7Li and 31P 1D
spectra were acquired using a Hahn-echo pulse sequence
(901 t 1801) with 901 pulses ranging from 2.0 to 2.5 ms
and recycle delays of 50–200 ms. Spin–lattice relaxation times
were acquired using a spin inversion recovery pulse sequence.
High temperature static 31P measurements were carried out
on a Bruker AMX 200 at the National Institute for Chemical
Physics and Biophysics (NICPB) in Tallinn, Estonia.
31
P spectra were acquired at Larmor frequencies of
81.28 MHz. A wide-bore, static, solid state NMR probe able
to reach temperatures up to 500 1C was used. Samples were
sealed in a quartz ampoule and heated under flowing nitrogen.
For all experiments, 7Li shifts were referenced to 1 M LiCl
(0 ppm) and 31P shifts were referenced to 85% H3PO4 (0 ppm).
Results and discussion
LiFePO4 is an olivine-type intercalation material with space
group Pnma. The Li ions sit in 1D chains coordinated by PO4
tetrahedra and slightly distorted FeO6 octahedra.1 Hydrothermal
synthesis is a fast, one-pot method of synthesizing nanocrystallites of the compound.27–29 Complete chemical and
electrochemical oxidation of triphylite produces orthorhombic
FePO4 (heterosite), while partial oxidation results in mixtures
of the two end-members. The diffraction pattern of biphasic
Li0.5FePO4, produced by the oxidation of LiFePO4 with a
stoichiometric amount of NOBF4, is shown in Fig. 1.
The lattice parameters of LiFePO4 were indexed to be
a = 10.323 Å, b = 6.002 Å, c = 4.696 Å with a unit cell
volume of 291.0 Å3, in good accord with several other
reports of LiFePO4 bulk crystallites. The lattice parameters
of the oxidized phase were determined to be a = 9.833 Å,
b = 5.801 Å, c = 4.771 Å with a unit cell volume of 272.1 Å3,
also in good accord with previous reports.
A SEM micrograph (inset, Fig. 1) shows the particle size of
the crystallites to be 200–500 nm in dimension.
In situ 31P NMR study of the thermally driven solid solution in
Li0.5FePO4
The sensitivity of the 31P resonance to changes at the redox
center allows for an in situ 31P NMR study of the thermally
driven phase transformation from biphasic Li0.5FePO4 to solid
solution Li0.5FePO4. A sample of half-delithiated, hydrothermally prepared, biphasic Li0.5FePO4 was first evaluated
This journal is
c
the Owner Societies 2011
Downloaded by University of Waterloo on 07 March 2012
Published on 07 February 2011 on http://pubs.rsc.org | doi:10.1039/C0CP01922D
View Online
Fig. 1 Powder X-ray diffraction pattern of biphasic Li0.5FePO4
prepared from chemical oxidation of LiFePO4. Selected reflections
of LiFePO4 (triphylite, T) and FePO4 (heterosite, H) are indexed.
Inset: SEM micrograph of the Li0.5FePO4 material which consists of
uniform sub-micron sized particles, with a narrow particle size
distribution.
Table 1 Summary of the 31P MAS NMR chemical shift and electronic
properties for LiFePO4 and FePO4
Sample
31
LiFePO4
FePO4
3800
5800
P MAS shift/ppm
Number of unpaired electrons
4
5
increase in the paramagnetic shift is understood by considering
the Fermi-contact interactions where the geometry dependent
electron spin density transfer from the paramagnetic center to
the nucleus of interest leads to resonances falling far outside
the diamagnetic range of chemical shifts.20,21 For the case
of FePO4, oxidation of the transition metal center and
subsequent unpairing of an electron in the t2g orbital leads to
an increased amount of unpaired electron spin density able
to delocalize onto the phosphorous nucleus and hence an
increased paramagnetic shift.
The sensitivity of the 31P paramagnetic shift to the electronic
environment at the transition metal center in biphasic
Li0.5FePO4 suggests that the electron delocalization associated
with the solid solution phase will be readily observed. In order
to reach temperatures where full electron delocalization occurs
(B400 1C), NMR experiments were acquired in the absence of
magic angle spinning. The static 31P NMR spectra (Fig. 3)
were collected as a series of frequency stepped sub-spectra and
added. The room temperature spectrum (Fig. 3, bottom 26 1C)
clearly shows the separate domains of phosphorous nuclei
near Fe3+ and Fe2+ similar to that observed in the MAS
NMR experiments. When the temperature is raised to 250 1C,
there is a loss of resolution between the two phases consistent
with the expected onset of electron and lithium delocalization
leading to a larger phase interface region.14,15 The migration of
the FePO4 and LiFePO4 resonances to lower frequency with
increased temperature results from Curie–Weiss behaviour of
paramagnetic resonances whereby the overall magnetic
susceptibility of each metal center decreases as the temperature
is raised. This leads to a decrease in the hyperfine coupling
constant and thus a lower chemical shift.30 When temperatures
of 400 1C are reached, a single resonance centered at 3200 ppm
Fig. 2 31P MAS NMR of FePO4 (top), LiFePO4 (middle) and
biphasic Li0.5FePO4 (bottom) all at room temperature. Isotropic shifts
for FePO4, LiFePO4 are found at 5800 and 3800 ppm respectively.
Asterisks denote spinning sidebands.
using 31P MAS NMR. Fig. 2 (bottom) shows the room
temperature 31P MAS NMR spectrum of biphasic Li0.5FePO4
where two distinct domains with resonances centered around
B3800 and 5800 ppm are observed. Fig. 2 also shows the 31P
MAS NMR spectrum of fully lithiated LiFePO4 and with an
Fe2+ electron configuration of t2g4eg2 giving a 31P isotropic
resonance centered at 3800 ppm. In contrast, for the 31P MAS
spectrum of FePO4, where the oxidation state of the iron
center is raised to Fe3+ giving a t2g3eg2 electron configuration,
the paramagnetic shift increases to 5800 ppm (Fig. 2, top)
(Table 1). This shift information is in agreement with previous
reports of the olivine LixFePO4 family.22,24,26 The cause of this
This journal is
c
the Owner Societies 2011
Fig. 3 31P static solid state NMR spectra under variable temperature
conditions. Frequency stepped sub-spectra (grey) added to yield total
spectrum (black).
Phys. Chem. Chem. Phys., 2011, 13, 5171–5177
5173
Downloaded by University of Waterloo on 07 March 2012
Published on 07 February 2011 on http://pubs.rsc.org | doi:10.1039/C0CP01922D
View Online
7
Table 2 Summary of FWHM values as temperature is decreased in
the 31P static spectra of single phase Li0.5FePO4
Li MAS NMR investigation of room temperature phase
boundary
Temperature/1C
FWHM/Hz
400
325
290
270
58 400
62 000
65 500
69 400
Here, detailed information regarding the FePO4/LiFePO4
phase interface is obtained from room temperature 7Li MAS
NMR spectrum of biphasic Li0.5FePO4 that has not undergone any heat treatment. For comparison, a 7Li MAS NMR
spectrum of pristine, fully lithiated, LiFePO4 is also presented
which shows a single isotropic resonance centered at 11 ppm
(Fig. 5a, top) consistent with previous reports.22,24 The
isotropic lineshape has a FWHM of 9500 Hz. Following
chemical delithiation, the 7Li MAS NMR spectrum of
biphasic Li0.5FePO4 (Fig. 5a, bottom) shows an isotropic
resonance centered at 12 ppm which deconvolutes to 5 peaks
labelled as A, B, C, D, and E (Fig. 5b). The deconvolution
resulting in the appearance of multiple peaks is somewhat
surprising as only the LiFePO4 domain should contain
Fig. 4 Static 31P sub-spectra of Li0.5FePO4 at 250 1C measured before
and after the solid solution phase was reached. The two phase features
observed before the solid solution was formed do not reappear in the
cooled sample on a timescale of 24 h following heat treatment.
appears. The 31P nuclei now ‘‘see’’ the Fe transition metal
in a single, average oxidation state due to the complete
delocalization of electrons of the solid solution phase.
The sample was then cooled in small increments to observe
the re-emergence of the two domains. Between temperatures of
400 1C and 270 1C, the sample retains single phase features but
with slight broadening (Table 2). At this stage it is difficult to
determine whether this is caused by the slowing of electron
mobility or increased paramagnetism at lower temperatures.
At 250 1C, where reappearance of the separate domains is
expected, the spectrum continues to appear as a single phase.
Fig. 4 shows two sub-spectra with the excitation frequency set
to 81.28 MHz collected at 250 1C before and after the solid
solution was formed. The pre-solid solution sub-spectrum
shows features that indicate that two domains are present.
For the same sub-spectrum collected 24 h after formation of
the solid solution, these features are no longer present and
only single phase Li0.5FePO4 is observed. Upon cooling and
allowing the sample to sit at room temperature for 72 h, the
separation into two domains is apparent (Fig. 3, top 26 1C).
This separation however, is not as well defined as the 26 1C
spectrum collected prior to heat treatment and thus suggests
that there is a considerable increase in the number of P nuclei
residing at the phase interface following treatment at high
temperature.
5174
Phys. Chem. Chem. Phys., 2011, 13, 5171–5177
Fig. 5 (a) 7Li MAS NMR spectra of LiFePO4 (top, MAS = 40 kHz)
and biphasic Li0.5FePO4 (0.5LiFePO4:0.5FePO4) (bottom, MAS =
40 kHz). Asterisks denote spinning sidebands. (b) Deconvolution of
biphasic Li0.5FePO4 with MAS = 40 kHz. Fitted spectrum (grey) sits
on top of experimental spectrum (black). Deconvoluted peaks are
shown in black.
This journal is
c
the Owner Societies 2011
View Online
Downloaded by University of Waterloo on 07 March 2012
Published on 07 February 2011 on http://pubs.rsc.org | doi:10.1039/C0CP01922D
Table 3 Summary of deconvolution parameters for the 7Li MAS
NMR spectrum of biphasic Li0.5FePO4 (Fig. 5b)
Peak
Chemical shift/ppm
FWHM/Hz
A
B
C
D
E
12
7
8
6
65
9998
114
427
2598
9827
observable Li. The FePO4 domain is not expected to give rise
to any Li signals. These newer peaks are therefore resulting
either from Li-ion disorder and/or Li belonging to the
LiFePO4 domain but sitting at the interface with the FePO4
domain. The deconvolution presented here is the minimum
number of peaks needed to obtain a goodness of fit of 95% in
the Topspin 2.1 program. Peak A is broad with parameters
similar to that of the 7Li MAS NMR spectrum of pristine
LiFePO4 (Table 3) and is thus confidently assigned as Li
sitting in the fully lithiated domain. Peak B is very narrow
and centered at 7 ppm. This peak does not contribute to the
lineshape of the spinning sidebands meaning it is a nonparamagnetic Li-environment (possibly a Li-salt impurity
arising from the oxidation of LiFePO4 with NO2BF4). Peaks
C, D, and E deconvolute with varying degrees of broadening
and also appear in the 1st order sidebands suggesting that
these resonances belong to paramagnetic phases. Peaks C and
D have chemical shifts slightly higher than the LiFePO4
domain (8 and 6 ppm, respectively). The narrowness and
appearance in higher order sidebands suggests that these
resonances arise due to their closeness to an Fe3+ environment
(i.e. proximity to the phase interface as purely Fe3+ regions
would have no Li-ions present). This conclusion is made in
consideration of the through-space nuclear–electron spin
interactions for paramagnetic materials. Increasing the
unpaired electron spin density significantly impacts not only
the paramagnetic shifts, but also the span of the chemical shift
anisotropy (CSA) and the broadness of the isotropic
resonance.19,20 Li sitting close to Fe3+ experiences a larger
electron-nuclear dipolar coupling as compared to Li sitting
close to Fe2+. This leads to a larger CSA which is now
observed for biphasic Li0.5FePO4 (span of the CSA is now
close to 7Li 8000 ppm). Furthermore, there is an increase in
the electric field gradient brought on by the extra charge of
the Fe3+ environment. This leads to an increase in the
quadrupolar coupling of the spin = 3/2 Li nucleus and thus
an increase in the sideband manifold.
There is also a characteristic narrowness to the 7Li
lineshapes that we have assigned to Li near Fe in oxidation
3+. This is justified based on the following relationships. In
paramagnetic systems, the linewidth (FWHM) has contributions
from both the hyperfine coupling value (A/h) and the spin–
lattice relaxation of the electrons (T1e) to which the nucleus is
coupled. Eqn (1) illustrates the dependence of the transverse
relaxation time of the nucleus (T2N) on the spin–lattice relaxation
time of the electron (T1e).31,32
FWHM ¼
This journal is
c
1
1
ðA=hÞ2 T1e
¼
þ
T2N T2a;b
8
the Owner Societies 2011
ð1Þ
The first portion of eqn (1), (1/T2a,b), is governed by the
transverse relaxation terms arising from effects other than
those induced by interaction with a paramagnetic metal center.
This has a significantly smaller contribution than the portion
containing terms associated with the unpaired electron spin
density sitting on the metal center (T1e, and the hyperfine
coupling constant, A/
h).32 In our system, the increase in the
chemical shift from 11 ppm for Li in a Fe2+ environment to
6 ppm for Li near Fe3+ is due to a modest increase in the A/
h
value. This increase alone would then lead to broadening of
the resonance for any Li near Fe3+. However, the opposite
trend is observed in this system (Table 3) thus indicating that
the changes in the linewidth are governed, rather, by the T1e
term. As the redox charge is raised from Fe2+ to Fe3+ (or as
the spin number goes from S = 2 to S = 5/2) the T1es
decrease, resulting in the observed narrower resonances. This
trend is also observed when comparing the 7Li MAS spectra of
LiFePO4 and LiMnPO4 (Fig. 7). In LiMnPO4, all Li nuclei are
surrounded by transition metals with a spin state of S = 5/2
(isoelectronic with Fe3+). The resultant 7Li MAS NMR
spectrum (with d = 75 ppm, FWHM of B1500 Hz) strongly
resembles the new, narrower resonances, assigned to Fe3+
environments in biphasic Li0.5FePO4. Previous studies have
measured the hyperfine coupling constants for both LiMnPO4
and LiFePO4 (0.86 106 rad s1 and 0.48 106 rad s1,
respectively).22 The larger A/
h value for LiMnPO4 would lead
to a less resolved spectrum according to eqn (1) however this
again is not observed meaning that T1e must be playing the
dominant role in the observed lineshapes. Therefore, the
greater resolution in the LiMnPO4 7Li spectrum is due to
the shorter T1e times of the spin density on the metal center as
compared to LiFePO4.
Overall, there is sufficient evidence to confirm that a fraction
of the Li nuclei are sitting close to a more paramagnetic Fe3+
(t2g3eg2) center. This thereby results in phase interface Li being
observable in these materials. However, due to the large
number of possible Li-sites that could be produced, caution
is emphasized when attempting site specific assignments of the
remaining deconvoluted peaks (C, D, and E). Fig. 6 shows the
local lithium environment in LiFePO4 where each Li-ion is
connected via O to 6 Fe atoms. Two of the Fe atoms have
Li–O–Fe overlap angles close to 901 while the other 4 Fe
atoms overlap at angles of either 1101 or 1201. Oxidation of an
Fe atom involved in the 901 orbital configuration (Fig. 6, left)
would lead to the greatest increase in the paramagnetic shifts
based on the geometry dependent delocalization of unpaired
electron spin density discussed previously. Alternatively,
oxidation of Fe atoms in the 1101 or 1201 configurations
(Fig. 6, right) would have a less significant impact on the
paramagnetic shift but through-space effects including
changes to the CSA would still be observed. These represent
the two extreme cases of numerous possible combinations
of Fe2+/Fe3+ neighbours to the crystallographic Li site.
Together these lead to a large number of Li-environments
with varying degrees of paramagnetic shift and broadness.
An exception is made for peak E which was deconvoluted
with a shift of 65 ppm and large FWHM of B9800 Hz. While
the paramagnetic shift is considerably higher, the linewidth is
virtually unchanged from the parent LiFePO4 thus suggesting
Phys. Chem. Chem. Phys., 2011, 13, 5171–5177
5175
Downloaded by University of Waterloo on 07 March 2012
Published on 07 February 2011 on http://pubs.rsc.org | doi:10.1039/C0CP01922D
View Online
Fig. 6 Schematic detailing the local Li environment with respect to Fe transition metal center proximity. Left: oxidation of Fe centers that have
orbital overlap close to 901 leading to Li-environments with high paramagnetic shift. Right: oxidation of Li-environments with orbital overlap of
1101–1221 leading to a smaller impact on paramagnetic shift but similar impact on lineshape and breadth of CSA.
Fig. 7 7Li MAS NMR (25 kHz) of isostructural LiMnPO4 (top) and
LiFePO4 (bottom). Asterisks denote spinning sidebands. This figure
illustrates the changes to the Li environment when the redox center
changes from a t2g4eg2 to a t2g3eg2.
sphere. This scenario is best represented by the anti-site defect
commonly generated in this system.33,34 Fig. 8 shows the local
Li environment generated if an Fe is found along the diffusion
channel. The Li ion now has 7, rather than 6, Fe2+ in its
coordination sphere where the defect Fe2+ (Fedefect) edge
shares with the Li-atom at two Li–O–Fe overlap angles
of 87.51 and 92.21. Due to the additive nature of the
Fermi-contact interaction, this would lead to the observed
increase in the Li-paramagnetic shift value but with minimal
changes to the lineshape. This anti-site defect was not observed
in the fully lithiated parent compound. This is attributed to the
difficulty in deconvolution of such a broad isotropic resonance
which is made worse by the increased concentration of Li in
the crystallographic, 11 ppm, environment. Future work will
look to overcome this obstacle through the use of 6Li MAS
NMR on enriched samples of Li1xFePO4.
Confirmation of the influence of particle size on the ability
to detect the phase boundary following delithiation is provided
through parallel studies of electrochemically cycled LiFePO4.
It is shown that observation of phase boundary lithium is
limited as the particle size increases. Fig. 9 shows 7Li MAS
spectra of large particle (1–1.5 mm) LiFePO4 before and after
Fig. 8 Schematic detailing the new Li-environment generated if a Fe
atom was found along the Li-diffusion channel. The Fedefect–O–Li
orbital overlap angles are very close to 901 resulting in a further
increase of the paramagnetic shift value.
that this Li site is still in an Fe2+ environment. This peak is
therefore assigned as a Li-environment generated where there
is a larger number of Fe2+ atoms in the Li-coordination
5176
Phys. Chem. Chem. Phys., 2011, 13, 5171–5177
Fig. 9 7Li MAS NMR (40 kHz) of micro-diameter LiFePO4 before
(top) and after (bottom) electrochemically cycling to a composition of
Li0.43FePO4. Asterisks denote spinning sidebands. Inset is the SEM
image of pristine particles.
This journal is
c
the Owner Societies 2011
Downloaded by University of Waterloo on 07 March 2012
Published on 07 February 2011 on http://pubs.rsc.org | doi:10.1039/C0CP01922D
View Online
cycling to a composition of Li0.43FePO4. The post-cycled
spectrum is unchanged relative to the parent material and no
new Fe3+ environment is observed. Therefore, although the
material is delithiated to an equivalent, or slightly greater degree
than the nanoparticles studied above, there is not sufficient
lithium in the interface region to allow detection by NMR.
Similarly in a recent report from Cabana et al.,26 a 6Li MAS
NMR spectrum of chemically delithiated, large particle
(>2 mm) LiFePO4 did not reveal features of phase boundary
Li. In both cases, there is certainly Li residing at the FePO4/
LiFePO4 interface, however, because of the larger particle
diameter, the fraction of this lithium (relative to Li in the fully
lithiated LiFePO4 domain) would be significantly smaller. The
interface regions are proposed to be only a few nanometres
thick.10 In smaller diameter particles, these few nanometres
carry greater significance than in particles of micrometre
diameter. From a 7Li NMR standpoint, signals arising from
Li residing in the fully lithiated domain would overtake the
NMR spectrum in the microdiameter samples making the small
region of interface Li unobservable. Due to the larger relative
impact of the interface in nanodimension particles, both the
fully lithiated domain as well as interface Li is observed.
Concluding remarks
Phase interface environments as well as electron dynamics are
probed in delithiated Li0.5FePO4. Room temperature MAS
NMR and static 31P spectra show the presence of two distinct
domains arising from the Fe3+PO4 and LiFe2+PO4 phases.
The P nuclei show extreme sensitivity to electronic changes
taking place at the transition metal center yielding 31P paramagnetic shifts of 5800 ppm and 3800 ppm, respectively. At
400 1C, these resonances coalescence into a single P-site centered
at 3200 ppm. This is due to complete electron delocalization
resulting in a single, average Fe oxidation state. Upon cooling
back to room temperature, the two domains of FePO4 and
LiFePO4 are not clearly resolved suggesting that some P-nuclei
continue to sit in an environment where the average Fe oxidation
state is between 2+ and 3+. 7Li MAS NMR spectra of
biphasic Li0.5FePO4 provide greater sensitivity to observe
phase interface regions as sites deconvolute into resonances
attributable to the distribution of Fe3+ and Fe2+ around a
lithium center. This is based on the extent of orbital overlap,
and the resulting magnitude of the chemical shift at one
extreme, and alternatively, the lack of orbital overlap, but
the significant influence of T1e on the narrowness of the
lineshape. These details allow for a clear distinction between
NMR spectra of samples with larger particle size from those
with submicron particle sizes. The latter uniform, small particles
have been shown to have excellent electrochemical performance,
and the utility of NMR in elucidating properties of the
lithiated/delithiated phase boundary will provide ongoing
insight into the development of similar phases.
References
1 A. K. Padhi, K. S. Nanjundaswamy and J. B. Goodenough,
J. Electrochem. Soc., 1997, 144, 1188–1194.
This journal is
c
the Owner Societies 2011
2 A. K. Padhi, K. S. Nanjundaswamy, S. Masquelier,
S. Okada and J. B. Goodenough, J. Electrochem. Soc., 1997,
144, 1609–1613.
3 A. S. Andersson, J. O. Thomas, B. Kalska and L. Haggstrom,
Electrochem. Solid-State Lett., 2000, 3, 66–68.
4 H. Huang, S. C. Yin and L. F. Nazar, Electrochem. Solid-State
Lett., 2001, 4, A170–A172.
5 B. Kang and G. Ceder, Nature, 2009, 458, 190–193.
6 Y. G. Wang, Y. R. Wang, E. J. Hosono, K. X. Wang and
H. S. Zhou, Angew. Chem., Int. Ed., 2008, 47, 7461–7465.
7 D. Morgan, A. Van der Ven and G. Ceder, Electrochem. SolidState Lett., 2004, 7, A30–A32.
8 A. S. Andersson and J. O. Thomas, J. Power Sources, 2001, 97–98,
498–502.
9 S. Nishimura, G. Kobayashi, K. Ohoyama, R. Kanno,
M. Yashima and A. Yamada, Nat. Mater., 2008, 7, 707–711.
10 A. S. Andersson, B. Kalska, L. Haggstrom and J. O. Thomas,
Solid State Ionics, 2000, 130, 41–52.
11 V. Srinivasan and J. Newman, Electrochem. Solid-State Lett.,
2006, 9, A110–A114.
12 C. Delmas, M. Maccario, L. Croguennec, F. Le Cras and F. Weill,
Nat. Mater., 2008, 7, 665–671.
13 L. Laffont, C. Delacourt, P. Gibot, M. Y. Wu, P. Kooyman,
C. Masquelier and J. M. Tarascon, Chem. Mater., 2006, 18,
5520–5529.
14 C. Delacourt, P. Poizot, J.-M. Tarascon and C. Masquelier, Nat.
Mater., 2005, 4, 254–260.
15 B. Ellis, L. K. Perry, D. H. Ryan and L. F. Nazar, J. Am. Chem.
Soc., 2006, 128, 11416–11422.
16 J. L. Dodd, R. Yazami and B. Fultz, Electrochem. Solid-State
Lett., 2006, 9, A151–A155.
17 A. Yamada, H. Koizumi, S.-I. Nishimura, N. Sonoyama,
R. Kanno, M. Yonemura, T. Nakamura and Y. Kobayashi, Nat.
Mater., 2006, 5, 357–360.
18 H. J. Tan, J. L. Dodd and B. Fultz, J. Phys. Chem. C, 2009, 113,
2526–2531.
19 A. Nayeem and J. P. Yesinowski, J. Chem. Phys., 1988, 89,
4600–4608.
20 C. P. Grey and N. Dupre, Chem. Rev., 2004, 104, 4493–4512.
21 D. Carlier, M. Menetrier, C. P. Grey, C. Delmas and G. Ceder,
Phys. Rev. B: Condens. Matter, 2003, 67, 174103.
22 M. C. Tucker, M. M. Doeff, T. J. Richardson, R. Finones,
J. A. Reimer and E. J. Cairns, Electrochem. Solid-State Lett.,
2002, 5, A95–A98.
23 S. L. Wilcke, Y. J. Lee, E. J. Cairns and J. A. Reimer, Appl. Magn.
Reson., 2007, 32, 547–563.
24 S. Hamelet, P. Gibot, M. Casas-Cabanas, D. Bonnin, C. P. Grey,
J. Cabana, J. B. Leriche, J. Rodriguez-Carvajal, M. Courty,
S. Levasseur, P. Carlach, M. van Thournout, J. M. Tarascon
and C. Masquelier, J. Mater. Chem., 2009, 19, 3979–3991.
25 N. Recham, M. Casas-Cabanas, J. Cabana, C. P. Grey,
J. C. Jumas, L. Dupont, M. Armand and J. M. Tarascon, Chem.
Mater., 2008, 20, 6798–6809.
26 J. Cabana, J. Shirakawa, G. Chen, T. J. Richardson and
C. P. Grey, Chem. Mater., 2010, 22, 1249–1262.
27 S. F. Yang, Y. N. Song, P. Y. Zavalij and M. S. Whittingham,
Electrochem. Commun., 2002, 4, 239–244.
28 B. Ellis, W. H. Kan, W. R. M. Makahnouk and L. F. Nazar,
J. Mater. Chem., 2007, 17, 3248–3254.
29 S. Franger, F. Le Cras, C. Bourbon and H. Rouault, J. Power
Sources, 2003, 119–121, 252–257.
30 B. Gee, C. R. Horne, E. J. Cairns and J. A. Reimer, J. Phys. Chem.
B, 1998, 102, 10142–10149.
31 T. J. Swift in NMR of Paramagnetic Materials, ed. G. N. La Mar,
W. DeW. Horrocks, Jr., and R. H. Holm, Academic Press,
New York, 1973, pp. 53-83.
32 F. H. Kohler in Magnetism: Molecules to Materials, ed.
J.S. Miller and M. Drillon, Wiley-VCH, New York, 2001, vol. 1,
p. 386.
33 R. Malik, F. Zhou and G. Ceder, 217th ECS Meeting, Vancouver,
Canada, 2010.
34 G. R. Gardiner and M. S. Islam, Chem. Mater., 2010, 22,
1242–1248.
Phys. Chem. Chem. Phys., 2011, 13, 5171–5177
5177