O
pen
A
rchive
T
oulouse
A
rchive
O
uverte (
OATAO
)
OATAO is an open access repository that collects the work of Toulouse researchers and
makes it freely available over the web where possible.
This is an author-deposited version published in:
http://oatao.univ-toulouse.fr/
Eprints ID : 2613
To link to this article :
URL :
http://dx.doi.org/10.1038/nmat2245
To cite this version :
Gibot, Pierre and Casas-Cabanas, Montse and
Laffont-Dantras, Lydia and Levasseur, Stephane and Carlach, Philippe and Hamelet,
Stephane and Tarascon, Jean-Marie and Masquelier, Christian ( 2008)
Room-temperature single-phase Li insertion/extraction in nanoscale LixFePO4.
Nature
Materials, vol. 7 . pp. 741-747. ISSN 1476-1122
Any correspondence concerning this service should be sent to the repository
Room-temperature single-phase
Li insertion/extraction in nanoscale Li
x
FePO
4
PIERRE GIBOT
1, MONTSE CASAS-CABANAS
1, LYDIA LAFFONT
1, STEPHANE LEVASSEUR
2,
PHILIPPE CARLACH
2, ST ´EPHANE HAMELET
1, JEAN-MARIE TARASCON
1AND CHRISTIAN MASQUELIER
1*
1Laboratoire de R ´eactivit ´e et de Chimie des Solides, CNRS UMR 6007, Universit ´e de Picardie Jules Verne, 33 Rue St. Leu, 80039 Amiens Cedex 9, France 2UMICORE Research & Development, Kasteelstraat 7, B-2250 Olen, Belgium
*e-mail: christian.masquelier@u-picardie.fr
doi:10.1038/nmat2245
Classical electrodes for Li-ion technology operate by either single-phase or two-phase Li insertion/de-insertion processes, with single-phase mechanisms presenting some intrinsic advantages with respect to various storage applications. We report the feasibility to drive the well-established two-phase room-temperature insertion process in LiFePO4 electrodes into a single-phase one by modifying the material’s particle size and ion ordering. Electrodes made of LiFePO4 nanoparticles (40 nm) formed by a low-temperature precipitation process exhibit sloping voltage charge/discharge curves, characteristic of a single-phase behaviour. The presence of defects and cation vacancies, as deduced by chemical/physical analytical techniques, is crucial in accounting for our results. Whereas the interdependency of particle size, composition and structure complicate the theorists’ attempts to model phase stability in nanoscale materials, it provides new opportunities for chemists and electrochemists because numerous electrode materials could exhibit a similar behaviour at the nanoscale once their syntheses have been correctly worked out.
Crucial environmental issues together with the rapid advance and eagerness from the automotive industry towards the commercialization of hybrid electric vehicles and plug-in hybrid electric vehicles have combined to make the development of improved rechargeable lithium batteries a worldwide imperative1.
Whereas the Li-ion battery has conquered the portable market, its implementation into electric transportation is constantly postponed due to performance, cost and more importantly, safety issues. To address these issues, we need positive electrode materials that are naturally abundant and react with Li at voltages within the thermodynamic stability range of non-aqueous Li-based electrolytes.
The trade-off between specific capacity and physico-chemical characteristics of bulk insertion materials has significantly hampered the consideration of poorly conducting compounds (either electronically or ionically) as possible better electrodes. The initial report from Padhiet al.2 took almost three years to
capture the battery community, as it dealt with the reversible extraction/insertion of 0.6 Li ions at∼3.5 V versus Li+
/Li◦
in the mostly insulating phospho-olivine LiFePO4. It now stands as a
very promising positive electrode potential candidate for the high-power, safe, low-cost and long-life batteries required to power electric cars. To bypass the electrode kinetic issues linked to its low intrinsic electronic/ionic conductivity, carbon nanopainting was initiated by Ravet and co-workers3,4. It was demonstrated that
Li+
ions could be reversibly extracted out of LiFePO4, leading to
capacities of∼160 mA h g−1, that is, close to the theoretical capacity
of 170 mA h g−1. Improvement of matrix conductivity by coating
is done at the expense of an extra electrochemically inactive layer that tends to lower the electrode’s overall capacity, especially the volumetric one. Besides, such a carbon coating does not solve the problem of the low intrinsic ionic conductivity of LiFePO4that can
200 101 210 301 311 011 111 201 211 020 121 δ θ 2 20 Intensity (arb. units) 25 30 35 40 45 50 55 S140 V = 291 Å3 V = 289 Å3 S40 = 0.25° 2 (°), CoK θ a b α
Figure 1XRD patterns of S40and S140nanosized LiFePO4. a, Data collected using
Co Kα radiation. The unit-cell volumes given for each studied sample were determined by a Rietveld refinement in the Pnma space group. A noticeable decrease (1.3%) in cell volume is observed in favour of the 40 nm (S40) sized LiFePO4particles. b, Magnification of the (211/020) diffraction peaks of the diffraction patterns: an important shift towards higher angles is observed.
be addressed by downsizing the particles. Nuspl et al.5 reported
on a carbon-free powder with a narrow particle size distribution (PSD) (average 400–600 nm range) that could deliver 114 mA h g−1
at 8C rate. Soon after, LiFePO4crystalline powders (140–150 nm
range) with outstanding capacities (147 mA h g−1at 5C rate) were
prepared by Delacourt et al.6 through a novel low-temperature
precipitation process followed by 3 h annealing at 500◦
C. Generally speaking, many other groups put forward the advantages in moving to nanosized LiFePO4 (refs 7–9). The questions are: what makes
Particle size (nm) Cumula tive distribution (%) Volumetric partic le size distribution (%) 100 90 80 70 60 50 40 30 20 10 0 0 20 40 60 80 100 120 140 160 180 2000 1 2 3 4 5 6 7 8 9 10 200 Pnma 17 nm >50 nm 10 nm 100 nm SEI 20.0 kV X100,000 WD 7.7 mm b c a
Figure 2Characterization of the 40 nm nanosized LiFePO4sample. a, Volumetric PSD and cumulative distribution. b, SEM image. c, HRTEM image combined with the
Fourier transform of one box showing the orientation of the crystallite.
‘nano’ so different for chemical storage? Are improvements simply due to shorter ion/electron diffusion lengths? Is there a specific defect chemistry associated with the solution synthesis of LiFePO4
nanoparticles? Are the electrochemical responses (single- versus two-phase insertion processes) changing by adding the particle size as a third variable besides structure and composition?
In this regard, LiFePO4 is still the subject of very passionate
theoretical and experimental studies on how the material elaboration could impact single- versus two-phase processes10,11.
This was prompted by Delacourt et al.10, who demonstrated
the existence of a complete single-phase LixFePO4 (0 6x6 1)
solid solution at ∼350◦
C. Such a finding made us aware that such a system was presenting a miscibility gap, thus opening various possibilities to identify physical/chemical parameters to tune its temperature dependence. Yamadaet al.7,8have suggested
that reducing particle size would allow some deviation from the well-described two-phase Li extraction/insertion behaviour at room temperature. Materials showing small particle size (∼80 nm) exhibit some limited solid-solution domains with the existence at 298 K of Li-poor LiyFePO4(0.02< y <0.2) and Li-rich Li1−xFePO4
(0.02< x <0.3) as borders of the two-phase domain. More
recently, Chiang and co-workers9,12reported experimental evidence
for a size-dependent miscibility gap in nanoscale LixFePO4powders
(coexistence of two phases between 0.1< x <0.7 for the lowest 50 nm particles studied). Our goal was to push the limit of our recent low-temperature precipitation process further to trigger the threshold value beyond which we could make nanoscale LiFePO4 particles react as a solid solution with respect to the
extraction/insertion of lithium.
Our recently reported low-temperature precipitation process of making homogeneous 140 nm LixFePO4powders13was further
exploited by modifying the nucleation/growth parameters through fine tuning of the precipitation conditions. We obtained a panel of green–greyish precipitates; they were recovered by centrifugation and washed several times with distilled water and acetone before being oven dried (50◦
C) for one day. The obtained powdered
samples were single phase as determined by X-ray diffraction (XRD) (Fig. 1), with particle sizes ranging from 40 nm (S40)
(Fig. 2b) to 140 nm (S140) from one sample to another, as observed
by scanning electron microscopy (SEM). Figure 2 shows, for the S40
sample, monodisperse small crystalline particles with platelet shape in the 15–100 nm range, with a very narrow percentage cumulative volumetric distribution peaking at 40 nm. The high-resolution transmission electron microscopy (HRTEM) image in Fig. 2c shows a well-crystallized nanoparticle (width: 15 nm, length: 60 nm). Bright layers observed perpendicular to [100]Pnma correspond to FeO6 layers. Peak shape analyses of XRD patterns for both S140
and S40 samples indicate crystallite sizes averaging∼50 nm and
∼30 nm for S140and S40, respectively. Therefore, S40particles will
be considered as single nanocrystals, whereas S140particles will be
considered as polycrystalline.
Highly divided samples are sensitive to moisture surface absorption, and were checked for weight loss as a function of heating temperature by thermogravimetric analysis measurement. A weight loss of∼3% occurring over two domain ranges 80–120◦
C and 180–220◦
C was noted and found to correspond to the departure of water (for example, 0.25 H2O per unit formula). It was
confirmed by temperature-programmed desorption experiments coupled with a mass spectrometer as two significant removals at 80 and 200◦
C with a corresponding molecular weight of 18 g mol−1.
Infrared spectroscopy showed the presence of a broad single vibration band/infrared peak located in the 3,000–3,600 cm−1
range. No peak was observed near 800 cm−1(observed for tavorite
LiFePO4(OH)), suggesting the presence of surface rather than
lattice water. The samples were further characterized for their Li and Fe contents by atomic absorption analysis: Li/Fe atomic ratios of 0.82(3) and 0.92(5) were obtained for the S40and S140samples,
respectively. Energy dispersive X-ray analysis was carried out on many particles and the Fe/P ratio was measured to be close to 0.91(2). This suggests non-stoichiometry issues consistent with the existence of 22 at.% of Fe+3(n
Fe3+/(nFe3++nFe2+) =0.22) for S40
13 nm 41 nm 2 1.0 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 1.0 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 y / b z /c 1.0 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 0 0.10.20.30.40.50.60.70.80.91.0 x/a y/b 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 100 θ(º), Co K Intensity (arb. units) α
Figure 3Rietveld refinement of the 40 nm (S40) nanosized LiFePO4. Experimental X-ray powder diffraction pattern (dotted curve) compared with the Rietveld-refined
profile (solid line) and difference curve. The insets correspond to the average apparent shape and size of the crystallites’ coherent diffraction domains in the directions indicated as obtained with the SPH approach: apparent size of 13 nm along [100]Pnma, 41 nm along [010]Pnmaand 35 nm along [001]Pnma.
and Mossbauer spectroscopy. The significant decrease in the Li content linked to an increase in Fe+3content provides the first hint
of a more pronounced defect chemistry with decreasing the size of the LiFePO4particles.
Further information was obtained by comparing the XRD patterns for both S40and S140samples (Fig. 1). The most noticeable
differences are observed around the 2θCo Kα (33–37◦) range with
the 211–020 peaks being shifted by more than 0.2◦
towards larger angles for S40 as compared with the S140 sample. Nevertheless,
both XRD patterns were entirely indexed in the space groupPnma with ‘standard’ lattice constantsa =10.332(2)A,˚ b =5.998(1)A,˚ c =4.699(1)(V =291.2 ˚A3) for the S140sample as compared with
a =10.301(2)A,˚ b =5.967(1)A,˚ c =4.704(1) (V =289.1 ˚A3) for S40: this corresponds to a significant decrease ((0.7%)) in
the crystallographic cell volume, which we regularly noticed for many nanocrystalline precipitated samples. By linear extrapolation between the referenced unit-cell volumes of LiFePO4 and FePO4
(refs 14,15), such a unit cell of 289.1 ˚A3would blindly correspond to an x value of ∼0.85 in LixFePO4 (ref. 16). This implies
an Fe+3 content of 0.2, consistent with the aforementioned
chemical/physical analyses for Li and Fe+3. To gain further insight
into site occupancies and account for both this unusual and new finding (preparation of a single ‘Li1−xFe1−yPO4’ phase stable at
room temperature), careful Rietveld analyses of X-ray (Fig. 3) and neutron diffraction (see the Supplementary Information) data were carried out. Within the olivine-type structure of the triphylite LiFePO4, Li atoms are located in chains of edge-sharing LiO6
octahedra (M1 sites) separated by corner-sharing FeO6octahedra
(M2 sites), resulting in a slightly distorted hexagonal close-packed oxygen array14,15.
To extract appropriate microstructural information from the diffraction pattern, realistic models considering the anisotropic broadening of the diffraction peaks due to both the very small platelet-shaped crystallites and the presence of micro-strains were used. Owing to the weak contribution of Li to the XRD pattern, the Li/Fe ratio was constrained during all of the refinement to the value determined by atomic absorption spectroscopy (0.82) and, in the initial stages of the Rietveld refinement, Fe occupancy was constrained to its nominal value (1). At this stage, the Rietveld refinement was of poor quality (see Supplementary Information, Fig. S1). A substantial improvement in the fit was obtained when Fe partial occupancy was left to vary without constraints, resulting in a partial occupancy 0.964(4) for Fe atoms (and thus 0.790(4) for Li as Li/Fe=0.82) (see Supplementary Information, Fig. S2). The best result was achieved when a small degree of Fe disorder was introduced, allowing its presence in M1 sites: the total amount of Fe was constrained to a constant value of 0.96, leaving both M1 and M2 Fe occupancies to vary constrained (see Supplementary Information, Table S2), resulting in 6.2% of Fe in Li sites (M1 sites). Atomistic calculations have predicted that such defects might be intrinsic to LiFePO4 (ref. 17). The refined unit-cell and
atomic parameters reveal almost identical average Li–O and Fe–O inter-atomic distances (2.151(1) and 2.147(1) ˚A, respectively), shorter than those of a well-crystallized stoichiometric LiFePO4
(Li–O=2.150 ˚A, Fe–O=2.157 ˚A), consistent with the presence of FeIII–O bonds. The apparent size of the crystallites along
each lattice direction was calculated and plotted (Fig. 3, insets), revealing a platelet shape. Apparent sizes are fully consistent with HRTEM observations (Fig. 2c). Small (but significant) amounts of anisotropic strains (average maximum strain=32(12) ×10−4)
C/10 C 0 C/10 D/10 D 2D Rate nC Potential (V) versus Li +/ Li 3.8 3.6 3.4 3.2 3.0 2.8 0 40 80 120 160 Capacity (mA h g–1)
Discharge capacity (mA h g
–1 ) Capacity (mA h g –1 ) Capacity (mA h g–1) 0 20 40 60 80 100 120 140 160 0 10 20 30 40 50 60 Cycle number 40 80 120 160 0.2 0.8 1.0 0 0.4 0.6 Non-carbon-coated nano Carbon-coated nano 0 20 40 60 80 100 120 140 160 Potential (V) versus Li +/ Li 3.9 3.7 3.5 3.3 3.1 2.9 2.7 2.5 a b c
Figure 4Electrochemical characterizations of the 40 nm (S40) LiFePO4and carbon-coated nano-LiFePO4. a, Charge/discharge galvanostatic curves at C/10 (1 Li+in
10 h) for a Li/carbon-free 40 nm LiFePO4cell cycled between 2.5 and 4 V, together with its capacity retention over 60 cycles. Cycling was also done at a rate C (1 Li+in 1 h) to highlight the electrode sustained reversible capacity and the near-100% cycling capacity efficiency (as charge and discharge curves superimpose) of the 40 nm
LiFePO4-based electrodes, regardless of the cycling rate. b, Charge/discharge galvanostatic curves at various rates of a Li/carbon-coated nano-LiFePO4cell. c, Comparison of rate capabilities of both cells described in a and b. The discharge capacity is plotted as a function of the rate nC (discharge of the electrode in (1/n) h).
occur preferentially along the [100] and [010] directions, in full agreement with our previous results16. A very robust fit (R
f=2.9%)
was finally obtained. Overall, the combination of extensive chemical analyses and Rietveld refinement of XRD data of the S40
sample leads to the formula (0.15Li0.79Fe0.06)M1(0.10Fe0.90)M2PO4,
revealing the existence of both cation vacancies and inter-site mixing within the structure of our ‘LiFePO4’ nanopowders (S40).
A sample prepared in the same conditions as sample S40 and
with a similar composition (intermediate unit-cell parameters between those of S40 and S140) was measured using neutron
radiation diffraction at the Institute Laue–Languevin (ILL, Grenoble, France). The D1A powder diffractometer was specially chosen for its high resolution over a wide range of scattering angles to obtain accurate results on atomic occupancies, displacement factors and microstructure of the sample. The average particle size was determined from SEM to be 70 nm. The neutron powder diffraction Rietveld refinement details can be found in the Supplementary Information. Our best fit corresponds to the formula (0.07Li0.89Fe0.04)M1(0.08Fe0.92)M2PO4 for S70, thus
demonstrating once more the presence of Fe atoms in M1 sites, with cell parameters a =10.3237(4)A,˚ b =5.9824(2) and c =4.6989(1)A (˚ V =290.2 ˚A3) and significant amounts of Fe at the oxidation state +3. Similarly to the XRD refinement, average platelet-shaped crystallites were obtained from the profile analysis as well as a small amount of anisotropic strains (8(4) ×10−4),
somewhat lower than in sample S40 and mainly along the [010]
direction. The comparison between the observed and calculated patterns (see the Supplementary Information) witnesses the quality
of the fitting and provides convincing evidence for the Fe mixing/disorder suggested by XRD.
Such an Fe mixing/disorder, in contrast with the observed cationic deficiency, does not come as a total surprise. It was first put forward by Chen and Whittingham18after XRD Rietveld analysis of
hydrothermally formed submicrometre LiFePO4powders showing
extra electron density on the Li sites. However, the main difference with our finding is that an increase in the unit-cell volume was pointed out in their ‘stoichiometric’ LiFePO4, as opposed to the
significant decrease we report here.
To check the importance of such defect chemistry on the electrochemical performance, Li-half Swagelok cells were assembled. A typical voltage–composition trace for the non-carbon-coated S40 sample (Fig. 4a) shows that a reversible and
sustainable capacity of 120 mA h g−1 is obtained at a low rate
(C/10). The most striking feature is the presence of a sloping voltage curve on charge and discharge on cycling. The origin of the irreversible capacity loss on the first cycle is not clearly understood at this point. It might be rooted in some residual species remaining at the surface. The complete disappearance of such irreversibility for the carbon-coated 50 nm particles (Fig. 4b) supports such a hypothesis. It is worth mentioning that the irreversible capacity is present only during the first cycle, as nearly 100% charge/discharge efficiency (Fig. 4a, inset) was obtained on the subsequent cycles regardless of the rate used (C/10 orC) to cycle the cell. Strikingly, the data reported for this non-carbon-coated nano-LiFePO4contrasts with all the data so far reported for
Charge Discharge (311) (301) (121) (020) (211) 2θ (º), Cu K Voltage versus Li (V) Time (h) 70 0 10 20 30 40 50 60 80 2.8 3.0 3.2 3.4 3.6 29 30 31 32 33 34 35 36 37 38 a b α
Figure 5Electrochemical and structural characterization of the 40 nm (S40) LiFePO4. a, Charge/discharge galvanostatic curve at C/40 (1 Li+in 40 h) of the 40 nm
LiFePO4(red line) compared with the two-phase profile of the 140 nm LiFePO4(blue line). b, In situ XRD measurements; a continuous shift of the diffraction peaks during the charge/discharge of this 40 nm LiFePO4is characteristic of a solid solution. The red diffractogramm, which corresponds to the obtained material at the end of the first charge, exhibits a cell volume of 273 ˚A3, close to those of the fully delithiated FePO4phase. XRD patterns were recorded using Cu Kα radiation at a scan speed of 1
◦
min−1. Only one-fifth of the collected patterns are displayed here for clarity.
x Li + extracted x Li + extracted a ( Å ) b (Å) c (Å) V (Å 3 ) 9.9 9.8 10.0 10.1 0 0.2 0.4 0.6 0.8 1.05.80 5.85 5.90 5.95 6.00 4.69 4.70 4.71 4.72 4.73 4.74 4.75 4.76 4.77 270 275 280 285 290 0 0.2 0.4 0.6 0.8 1.0 10.3 10.2
Figure 6Unit-cell parameters as a function of x lithium extracted in 40 nm (S40) nanosized LixFePO4. These values were refined in the Pnma space group from the data
of Fig. 5. Also of interest, the unit-cell contraction is anisotropic, as a and b decrease (−4.2% and −2.5% respectively) while c increases (+1.4%).
three separated cells using powders from the same S40batch sample
were made. There was barely any deviation between them. Sloping voltage–composition curves are usually characteristic of a single-phase Li de-intercalation/intercalation mechanism. To check this possibility, we have carried out in situ XRD
measurements. As a S40/Li cell was being charged, we initially
observed (Fig. 5b) a slight and continuous shift in the Bragg peaks (mainly visible for the (401) peak), indicating a full solid solution with the end phase having lattice parameters (a =9.879(3)A,˚ b =5.800(1)A,˚ c =4.759(2)A and˚ V =273.1 ˚A3) close to those
of the fully delithiated FePO4 phase. On discharge, we further
observed a progressive shift of the Bragg peaks back to their initial position, unambiguously conveying that we not only have a solid-solution mechanism but a true fully reversible lithium insertion/de-insertion process. To further support this, we indexed the entire collected XRD patterns. The results of such fits (Fig. 6) indicate that the lattice parameter variations follow nicely Vegard’s law. This spectacular shift in the Li insertion/de-insertion mechanism from a two-phase process to a single homogeneous phase (SHP) with downsizing the particle size is rich in both fundamental and applied consequences. It also raises many other questions such as the narrowness of the threshold particle size and its dependence on synthesis conditions. Our finding indicates a SHP insertion process for 40 nm LiFePO4 particles in apparent contradiction
with the two-phase process reported by Chiang and co-workers12
and by Kim and Kim19 on similar size particles. This is a first
hint that particle size can be considered as a necessary but not sufficient condition to anticipate SHP versus two-phase behaviour in LiFePO4. Such a comparison bears important consequences, as
it highlights the feasibility of having either a SHP versus two phases for samples with the same particle size but different amounts of non-stoichiometry and cationic mixing. Our results agree with what has been frequently observed in the literature where, generally speaking, substitutions lead to monophasic behaviour. Among relevant examples are (1) the two-phase 4 V insertion/de-insertion process that becomes single phase on shifting from LiMn2O4 to
Li1+xMn2−xO4 with the extra Li moving to the B sites, (2) the
two-phase de-insertion process in LiFePO4that turns into a single
phase one by substituting Fe+2in the M
1site for Mn (for example,
Li(Fe,Mn)PO4) or (3) the reported single-phase versus two-phase
process in the Nasicon LiTi2(PO4)3 when Ti is substituted for
Fe (Lix(Ti,Fe)(PO4)3).
Our results set up new challenges, as downsizing a particle can affect both structure and composition. Needless to say that both the small particle size and the cationic mixing are the consequence of the precipitation conditions and more specifically of the nucleation/growth kinetics that depends on the nature of the precursors, synthesis parameters and so on. Obviously, trying to rationalize such an observation will call on a detailed examination of the system energetics that will require consideration of the molar free energy of mixing for one or two phases with respect to the relative contribution among others of (1) nanoparticle surface energy, stress and stoichiometry, (2) elastic energy related to phase transformation and (3) nucleation energy.
In summary, contrary to earlier perspectives, LiFePO4offers a
rich chemistry linked to the chemist’s ability to elaborate new low-temperature synthesis approaches. The main issue is that tuning the particle size and stoichiometry of ‘LiFePO4’ paves the way to new
chemical reactivity paths such as single- versus two-phase insertion processes. We are far from ascertaining which of the new variables injected by the size confinement (strain, composition/defects gradient concentrations, surface energetics, ion/electron diffusion lengths and so on) is the most determining. Answering such a question could greatly benefit from theoretical work aiming at revisiting chemical reactivity in conjunction with thermodynamics at the nanoscale as recently initiated by a few groups20–23. Whatever
the exact fundamental reasons for the single-phase reactivity of nano-LiFePO4 versus Li are, our finding offers numerous
chemical possibilities as long as this complex defect chemistry can be mastered. Moreover, the possibility of having single-phase extraction/insertion mechanisms (for example, a sloping voltage curve) presents some intrinsic advantages with respect to applications such as an easier and cheaper monitoring state of charge of the battery as compared with a flat constant voltage curve. Nevertheless, to take full advantage of this finding, at the
applied level, it remains crucial to pursue new approaches towards electrode nano-architecturing, so that the power rate performances of 40 nm ‘LixFePO4’ powders made at low temperature (108
◦
C) can compete favourably with today’s 140 to 600 nm carbon-coated particles prepared at 500◦
C. Such newly confectioned materials should sustain the great interest that LiFePO4holds among battery
manufacturers, and revitalize exchanges between experimentalists and theorists.
METHODS
SYNTHESIS
40 and 140 nm nanosized LiFePO4particles were synthesized by the
precipitation route in aqueous solution at low temperature and under atmospheric pressure. Equimolar amounts of 0.1M FeSO4·7H2O (99%—
Aldrich) and H3PO4(85%—Riedel-deHaen) were mixed in an aqueous-based
solution before addition, drop by drop, of a 0.3 M LiOH (98%—Aldrich) solution to give Li/Fe/P equal to 3:1:1. After homogenization, the reaction mixture was heated at reflux for 24 h under a nitrogen atmosphere. On cooling to room temperature, the green–grey precipitate was washed with water and acetone and dried in an oven at 50◦C overnight.
CHEMICAL ANALYSIS
Atomic absorption spectroscopy (Perkin Elmer AAnalyst 300) was used to determine the Li/Fe ratio (±2%) for each studied sample. The as-prepared LiFePO4samples were dissolved in an acidic solution
(H2O/HCl/HNO3,80 : 10 : 10 wt%) at room temperature under stirring.
Standard solutions of 1, 2, 2.5, 4 and 5 p.p.m. were used for the calibration of Fe. Calibration of Li was carried out using standard solutions of 1, 2 and 3 p.p.m. Electron dispersive X-ray was used to determine the Fe/P ratios (±0.02) in the pristine LiFePO4samples. The amount of iron (+) species
(±2%) in the pristine LiFePO4samples was determined from a titration
method based on a redox reaction between the following couples: Fe3+/Fe2+
and Cr2O2−7 /Cr3+. The tryphilite samples were dissolved in an acidic solution
(H2O/H2SO4/H3PO4,75 : 10 : 15 vol.%) before being titrated with a 0.0167 M
potassium dichromate (K2Cr2O7) solution. Sulphanilic acid was used as a
coloured indicator.
XRD AND RIETVELD REFINEMENT
X-ray powder diffraction patterns of the LiFePO4nanoparticles were collected
on a Bruker D8 diffractometer using Co Kα radiation (l1=1.78919 ˚A,
l2=1.79321 ˚A), equipped with a Gobel mirror, a Braun PSD detector and operating at 40 kV–30 mA in the range 2θ = 10–100◦
with a 2θ step size of 0.017◦
. The instrumental resolution function of the diffractometer was determined with an Al2O3standard (U = 0.018336 deg2,V = −0.012039 deg2,
W = 0.012975 deg2). Rietveld refinements were carried out with the program FullProf24,25(Windows version, March 2007) using the pseudo-Voigt profile
function of Thompsonet al.26. The background was refined by adjusting
the height of pre-selected points for linear interpolation. The refinements were carried out taking as initial cell parameters and atomic positions those obtained in the neutron powder diffraction study of Rousse et al.14. Anisotropic
size broadening was modelled in terms of spherical harmonics27, and the
crystallite average apparent size along each reciprocal lattice vector was calculated. An ‘average apparent shape’ of the crystallites that may suffer from some ‘ripples’ due to the spherical harmonics approximation can also be obtained using the GFOURIER 04.02 program28. An extra anisotropic strain
broadening model corresponding to the Laue classmmm was also used to
accurately describe the shape of the Bragg reflections, and to calculate the average maximum strain, of which the standard deviation is a measure of the degree of anisotropy29. The structural and microstructural refined values
can be found in the Supplementary Information. The structural evolution of LixFePO4on operation during Li+extraction/insertion was followed
byin situ XRD using a Bruker D8 diffractometer using Cu Kα radiation
(l1=1.54056 ˚A, l2=1.54439 ˚A), in Bragg Brentano geometry and equipped
with a Vantec ‘SuperSpeed’ PSD detector. The sample was placed right behind
an X-ray-transparent beryllium window also acting as a current collector. To bypass the oxidation of beryllium during high-voltage analysis, a thin layer of aluminium was deposited by evaporation onto the inner side of the window. Thein situ XRD patterns were collected for a complete charge/discharge cycle
NEUTRON POWDER DIFFRACTION AND RIETVELD REFINEMENT
Neutron powder diffraction data were obtained using the D1A high-resolution two-axis diffractometer at the Institute Laue–Languevin (ILL, Grenoble), at l=1.91 ˚A with a step size of 0.01◦
on a 2θ domain ranging from 1◦
to 157◦
. The instrumental resolution function was determined with a Na2Ca3Al2F14
standard (U = 0.11866 deg2,V = −0.21449 deg2,W = 0.16915 deg2). The
specimen (of about 3 g) was placed in a cylindrical vanadium cell. As in the XRD Rietveld analysis, the neutron data refinement was carried out using the atomic position set and space group of the olivine structure found in ref. 14. Likewise, the pseudo-Voigt profile function of Thompsonet al.26was used with
an asymmetry correction at low angle and an anisotropic broadening treatment of the diffraction peaks due to the small size of the crystallites and the presence of micro-strains. The nominal composition LiFePO4was used in the first stage
of the refinement. To improve our fit, Fe and Li partial occupancies were then allowed to vary, free of constraints. At this stage of the refinement, all atomic displacement factors were considered to be isotropic. A further improved fit was obtained when a simultaneous presence of Li and Fe was allowed in M1 positions, although no convergence was reached when a simultaneous presence of Li and Fe was considered in M2 sites. Owing to both the large 2θ range used and the high resolution characteristic of the D1A instrument, the anisotropic displacement parameters of M1 sites were also refined, leading to displacement ellipsoids elongated towards the theoretical curved one-dimensional lithium diffusion path. These results are in agreement with recent work that considered the shape and orientation of the ellipsoids to be the first experimental suggestion of one-dimensional lithium motion20. In the last stage, all parameters (profile
and structural) were refined simultaneously free of any constraints until convergence was reached to obtain correct estimated standard deviations (see Supplementary Information, Table S1).
ELECTROCHEMICAL INSERTION/EXTRACTION OF Li+
Li-half Swagelok cells were assembled. The obtained electrode containing 90 wt% of active material was used to manufacture Swagelok-type cells, using a typical loading of 6 mg cm−2of active material directly led onto the current
collector. Negative electrodes were made of metallic Li. Cells were cycled in LiPF6-based electrolyte between 2.5 and 4.0 V. Lithium de-insertion/insertion
was monitored either with Mac-Pile or Versatile Multiple Potentiostat cycling/data recording systems (Biologic SA), operating in galvanostatic mode. Power rate measurements were made by charging and discharging cells at various rates between 2.5 and 4.0 V. Between each cycle data set, the cells were recharged at a rate ofC/20 to recover their full capacity before applying a
subsequent discharge/charge cycle at a higher rate.
ELECTRON MICROSCOPY
SEM images were recorded with a Philips FEG XL-30 operating at 20 kV from samples previously sprayed with a 5-nm-thick layer of platinum. The PSD was calculated from the SEM images of 500 particles. TEM pictures were obtained using an FEI Tecnai F20 S-Twin electron microscope operating at 200 kV. Samples for electron microscopy analysis were prepared by depositing a dispersion of nanoparticles on copper grids coated with lacey-carbon film. The Fe/P ratio was determined using standards as references. The measurements were repeated in at least ten different spots of the samples, and the reported value is an average of these ten measurements. There was little deviation between them, consistent with the sample homogeneity.
References
1. Tarascon, J. M. & Armand, M. Issues and challenges facing rechargeable lithium batteries.Nature
414, 359–367 (2001).
2. Padhi, A. K., Nanjundaswamy, K. S. & Goodenough, J. B. Phospho-olivines as positive electrode materials for lithium batteries.J. Electrochem. Soc144, 1188–1194 (1997).
3. Ravet, N.et al. Improved iron-based cathode material. Abstract # 127,196th Meeting of the Electrochemical Society, Hawai(1999).
4. Armand, M., Gauthier, M., Magnan, J. -F. & Ravet, N. Method for synthesis of carbon-coated redox materials with controlled size. World Patent WO 02/27823 A1 (2002).
5. Nuspl, G., Wimmer, L. & Eisgruber, M. Lithium metal phosphates, method for producing the same and use thereof as electrode material. World Patent WO 2005/051840 A1 (2005).
6. Delacourt, C., Poizot, P., Levasseur, S. & Masquelier, C. Size effects on carbon free LiFePO4powders:
The key to superior energy density.Electrochem. Solid State Lett.9, A352–A355 (2006).
7. Yamada, A.et al. Room-temperature miscibility gap in LixFePO4.Nature Mater.5, 357–360 (2006).
8. Yamada, A., Koizumi, H., Sonoyama, N. & Kanno, R. Phase changes in LixFePO4.Electrochem. Solid
State Lett.8, A409–A413 (2005).
9. Meetong, N., Huang, H., Speakman, S., Carter, W. C. & Chiang, Y. M. Strain accommodation during phase transformations in olivine-based cathodes as a materials selection criterion for high-power rechargeable batteries.Adv. Funct. Mater.17, 1115–1123 (2007).
10. Delacourt, C., Poizot, P., Tarascon, J.M & Masquelier, C. The existence of a temperature-driven solid solution for 0≤x ≤1 in LixFePO4.Nature Mater.4, 254–260 (2005).
11. Zhou, F., Marianetti, C. A., Cococcioni, M., Morgan, D. & Ceder, G. Phase separation in LixFePO4
induced by correlation effects.Phys. Rev. B69, 201101(R) (2004).
12. Meetong, N., Huang, H., Carter, W. C. & Chiang, Y. M. Size-dependent lithium miscibility gap in nanoscale Li1−xFePO4.Electrochem. Solid State Lett.10, A134–A138 (2007).
13. Delacourt, C., Poizot, P. & Masquelier, C. Crystalline nanometric LiFePO4. World Patent,
CNRS-UMICORE, #WO 2007/0051 (2007).
14. Rousse, G., Rodriguez-Carvajal, J., Patoux, S. & Masquelier, C. Magnetic structures of the triphylite LiFePO4and of its delithiated form FePO4.Chem. Mater.15, 4082–4090 (2003).
15. Santoro, R. P. & Newmann, R. E. Antiferromagnetism in LiFePO4.Acta Cryst.22, 344–347 (1967).
16. Delacourt, C., Roriguez-Carvajal, J., Schmidt, B., Tarascon, J. M. & Masquelier, C. Crystal chemistry of the olivine-type LixFePO4system (0≤x ≤1) between 25 and 370
◦
C.Solid State Sci.7,
1506–1516 (2005).
17. Islam, S. M., Driscoll, D. J., Fisher, C. A. J. & Slater, P. R.Chem. Mater.17, 5085–5092 (2005).
18. Chen, J. & Whittingham, M. S. Hydrothermal synthesis of lithium iron phosphate.Electron. Commun.8, 855–858 (2006).
19. Kim, D. H. & Kim, J. Synthesis of LiFePO4nanoparticles in polyol medium and their electrochemical
properties.Electrochem. Solid State Lett.9, A439–A442 (2006).
20. Yamada, A.et al. Intermediate phases in LixFePO4.Mater. Res. Soc. Symp. Proc.972, 257–264 (2007).
21. Maier, J. & Amin, R. Defect chemistry of LiFePO4.J. Electrochem. Soc.155, A339–A344 (2008).
22. Fisher, C. A. J. & Islam, M. S. Surface structures and crystal morphologies of LiFePO4: Relevance to
electrochemical behaviour.J. Mater. Chem.18, 1209–1215 (2008).
23. Gaberscek, M., Dominko, R. & Jamnik, J. Is small particle size more important than carbon coating? An example study on LiFePO4cathodes.Electrochem. Commun.9, 2778–2783 (2007).
24. Rodr´ıguez-Carvajal, J. Recent advances in magnetic-structure determination by neutron powder diffraction.Physica B192, 55–69 (1993).
25. Rodr´ıguez-Carvajal, J.Recent Developments of the Program FULLPROF, inCPD Newsletter2001, 26, 12, available at<http://www.ill.eu/sites/fullprof/index.html>.
26. Thompson, P., Cox, D. E. & and Hastings, J. B. Rietveld refinement of Debye–Scherrer synchrotron X-ray data from Al2O3.J. Appl. Crystallogr.20, 79–83 (1987).
27. Jarvinen, M. Application of symmetrized harmonics expansion to correction of the preferred orientation effect.J. Appl. Crystallogr.26, 525–531 (1993).
28. Gonz´alez-Platas, J. & Rodr´ıguez-Carvajal, J. Graphic Fourier Program GFOURIER, Version 04.02. Univ. La Laguna, Tenerife, Spain, 2002.
29. Rodr´ıguez-Carvajal, J., Fern´andez-D´ıaz, M. T. & Mart´ınez, J. L. Neutron-diffraction study on structural and magnetic-properties of La2NiO4.J. Phys. Condens. Matter3, 3215–3234 (1991).
Supplementary Information accompanies this paper on www.nature.com/naturematerials.
Acknowledgements
We are grateful to C. Delacourt and D. W. Murphy for enlightening discussions and to J. Rodriguez Carvajal at ILL Grenoble for his help in collecting the neutron diffraction patterns.
Author information
Reprints and permission information is available online at http://npg.nature.com/reprintsandpermissions. Correspondence and requests for materials should be addressed to C.M.
