Phase Diagram of Lix  ( Mny Fe1 − y  )  PO 4  ( 0 ⩽ x ,  y ⩽ 1 ) 

, and

Published 11 September 2001 © 2001 ECS - The Electrochemical Society
, , Citation Atsuo Yamada et al 2001 J. Electrochem. Soc. 148 A1153 DOI 10.1149/1.1401083

1945-7111/148/10/A1153

Abstract

The room temperature (x, y) two-dimensional phase diagram of the olivine-type solid-solution, orthorhombic, is determined. The x-dependent changes in the unit cell dimensions at various fixed Mn contents y are analyzed in detail. The manganese substitution for iron in the octahedral 4c sites induces 1, the two-phase redox region with a potential of 4.1 V vs. 2, a partial conversion of the form of the redox reaction (3.4 V vs. from two-phase to single-phase; and 3, phase instability at the composition region close to the point (x, y) = (0, 1) caused by the Jahn-Teller effect of The conversion, 2, is complete at around The phase instability, 3, makes the Mn-rich phase unsuitable for battery applications. The local lattice deformation around is severe enough to induce significant selective damping in the extended X-ray absorption fine structure for © 2001 The Electrochemical Society. All rights reserved.

Export citation and abstract BibTeX RIS

Since the demonstration of reversible electrochemical lithium insertion-extraction for in 1997,1 lithium transition metal phosphates with an ordered olivine structure, Ni, Mn, Fe, Cu), have attracted much attention as promising new cathode materials for rechargeable lithium batteries.2 3 4 5 6 7 8 9 10 The initially present single lithium ion, per transition metal ion, can be extracted in the first charge process, compensating for the oxidation of to and transferred to the carbon anode through the nonaqueous electrolyte

The oxidation (charge) of to induces a reduction in the unit cell volume. This shrinkage compensates for the volume expansion of the carbon anodes in the charge process and contributes to efficient use of volume in a practical lithium-ion cell. The opposite movement of lithium ions and electrons occurs during the discharge process, while the transition metal M is reduced from trivalent to divalent.

The crystal has an orthorhombic unit cell space group which accommodates four units of 11 As a typical example, has unit-cell dimensions of and 7 Graphic representations of the crystal structure have been given in many references.1 3 4 11 Both the Li and M atoms are in octahedral sites with Li located in the 4a and M in the 4c positions. The oxygen atoms are nearly hexagonal closed-packed and the M atoms occupy zigzag chains of corner-shared octahedra running parallel to the c axis in alternate planes. These chains are bridged by corner- and edge-sharing polyanions to form a host structure with strong three-dimensional bonding. The ions in 4a sites form continuous linear chains of edge-shared octahedra running parallel to the c axis in the other planes, which makes two-dimensional motion possible.

The charge-discharge reaction of the presently used materials, such as the layered rock salt systems, (space group, and a spinel framework system (space groups: are all based on the couple in edge-shared octahedra in the closed-packed oxygen array which generates ca. 4 V12 13 14 15

Therefore, the presence of large tetrahedral polyanions corner-shared octahedra, and the use of the redox couple are distinctive features of olivine class cathodes. The linkage in the structure induces the superexchange interaction that tunes the redox energy to useful levels (3.4, 4.1, and 4.8 V for and respectively).1 5 The strong covalency stabilizes the antibonding state of primarily cationic origin to generate an appropriately high voltage, which is called the "inductive effect."1 6 The stable nature of the olivine-type structure having a polyanion with a strong P-O covalent bond provides not only excellent cycle-life but also a safe system. When the battery is fully charged, the reactivity to the combustion reaction with the organic electrolyte is low.6 Safety issues are of paramount importance in the design of consumer batteries, and this makes olivine-type materials particularly attractive as cathodes for lithium-ion battery systems.

The energy density of olivine-type is equal to that of presently used materials, based on the theoretical charge-discharge capacity of ca. 170 mAh/g obtained from the one-electron redox reaction of 6 and the high voltage of >3.4 V vs. However, the inherent low conductivity for both electrons and lithium ions must be overcome by optimized powder engineering to realize uniformly small particles, or utilization is limited by slow diffusion of these charge carriers. We have recently observed a reversible capacity of >160 mAh/g at room temperature for optimized powder sintered at the moderate temperatures of 500-600°C,6 as shown in Fig. 1a. In the optimization process, we identified two main obstacles to achieving optimum charge/discharge performance: (i) undesirable particle growth at higher temperatures and (ii) the presence of residual at lower temperatures in the sintering and/or milling process.6

Figure 1.

Figure 1. Typical charge-discharge curves measured at room temperature for the optimized (a, from Ref. 6) and (b, from Ref. 7 and 8).

In the olivine-type family, and their solid solution system, 1 look promising because they operate at 3.4-4.1 V vs. This is providential, because it is not so high as to decompose the organic electrolyte but is not so low as to sacrifice energy density. The positions of redox couples were confirmed in our laboratory by first-principle calculations.6 The use of is of particular interest because the position of the couple, 4.1 V vs. 1 is compatible with present lithium-ion batteries and generates high energy density. However, it has been shown that the capacity at 4.1 V is not achieved without Fe coexisting with Mn at the octahedral 4c site1 Padhi et al. have performed systematic experiments on the electrochemical charge and discharge characteristics of 0.50, 0.75, 1.0) and reported that the width of the 4.1 V plateau relative to that of the 3.4 V plateau increases as Mn content y is increased, but that the total capacity rapidly decreases at .1

In our study, the strong electron lattice interaction (Jahn-Teller effect) in the charged state and the resultant limit to lattice distortion have been suggested to be intrinsic obstacles to generating the charge-discharge of the Mn-rich phase 7 The solubility limit in the binary system lies in the composition range of For this reason, was initially chosen as a promising composition for the 4 V cathode material and a capacity equal to the theoretical value (170 mAh/g), more than half of which generates 4 V, was demonstrated using the optimized powder.7 See Fig. 1b. The charge-discharge reaction mechanism of was also investigated in detail.8 We found a flat two-phase region with an open-circuit voltage (OCV) of 4.1 V (region I: and an s curved single-phase region with (region II: 8 This is significantly different from the in which the whole reaction proceeds in a two-phase manner with a flat voltage profile at 3.4 V.1

Our principal concern at this stage is why the redox reaction in the same olivine-type host structure proceeds in a single-phase mechanism when 60% of the manganese coexists in the same octahedral 4c sites, 8 while there is a two-phase reaction over the entire compositional range of 1 In this paper, the (x, y) two-dimensional phase diagram of the olivine-type solid-solution, is determined. We show below the results of a survey of the phase change in the (x, y) plane and how the manganese substitution for iron changes the charge-discharge reaction mechanism, with special consideration of the two-phase to single-phase transformation of the redox reaction. The severe local lattice deformation around the Jahn-Teller active which induces phase instability around is also highlighted.

Experimental

The (six samples: 0.2, 0.4, 0.6, 0.8, 1.0) compounds were prepared by solid-state reaction of and They were dispersed into acetone, then thoroughly mixed, and reground by ballmilling. After evaporating the acetone, the olivine phase was synthesized in purified gas flow (800 cm3/min) to prevent the formation of trivalent compounds as impurities. The mixture was first decomposed at 280°C for 3 h to disperse the gases and firmly reground again, then sintered for 24 h at 600°C.

Chemical oxidation to obtain was performed by reacting with nitronium tetrafluoroborate in acetonitrile.16 The redox potential of is ca. 5.1 V vs. and is effective for oxidizing with redox potentials of 3.4 V and 4.1 V vs. 1 7 8 16 The reaction is written as

Equation (1)

To ensure a complete reaction, the amount of added was double the amount estimated from Reaction 1. After 8.5 g of was dissolved into 300 mL acetonitrile, 5 g of was added and the mixture was stirred for 24 h at room temperature under bubbling purified Ar gas. The product was filtered and washed several times with acetonitrile to remove impurities before it was dried under vacuum at 70°C.

Lithiated samples of were prepared by reacting with various amounts of LiI (High Purity Chemicals, 99.9%) in acetonitrile. The hygroscopic LiI powder was treated in a dry atmosphere, and the ratio of acetonitrile to was set at 200 mL to 0.2 g. The solution was stirred for 24 h at room temperature, and the products were filtered and washed several times with acetonitrile to remove impurities before they were dried under vacuum at 70°C. The chemistry of LiI as a reducing agent in acetonitrile has been studied previously in detail.17 The oxidation of to occurs in two reversible steps

Equation (2)

Equation (3)

Reaction 2 was performed successively to form with a redox potential of 2.9 V.18 In the present case, for the 4.1 V region of both steps 2 and 3 shift completely to the right at room temperature, and the reaction can be assumed to incorporate all of the in acetonitrile supplied by LiI into 8 whereas, for the 3.5 V region of only Reaction 1 was induced and excess LiI was added to produce the desired composition.8 All the products were analyzed using inductively coupled plasma spectroscopy-atomic emission spectroscopy (ICPS-AES, Shimazu Co., ICPS-8000) to determine their exact composition.

All of the characterizations by X-ray powder diffraction (XRD), Mössbauer spectroscopy, and X-ray absorption spectroscopy (XAS) were based on chemically prepared samples of to ensure the data is for clean samples. X-ray powder diffraction (RINT-2500v, Rigaku Co.) with radiation was used to identify the phases and analyze the structure. The lattice constants were calculated using the least squares method with a Si standard. The diffraction profiles were measured in the slow-scan mode (0.5°/min). The Mössbauer spectra in transmission geometry were collected using a γ-ray source. Velocity calibration was made with α-Fe data at room temperature. The sample thickness was adjusted so that the Fe content was ca. 8 mg/cm2.

X-ray absorption measurements were performed with the Industrial Consortium Beamline BL16B2 in SPring-8 using synchrotron radiation from the electron storage ring at an electron energy of 8 GeV. A white X-ray beam was monochromatized by a Si(111) double-crystal monochromator. A half-cylindrical Rh-coated total-reflection mirror was used to eliminate the harmonics and converge the beam in the horizontal direction. X-ray absorption spectra were taken in the transmission mode near the Mn and Fe K edges on samples diluted in boron nitride powder at proportions calculated to give an edge jump in of ca. 1. The intensities of incident and transmitted X-ray beams were measured using ionization chambers filled with 100% nitrogen gas and nitrogen gas mixed with 15% Ar gas, respectively. The data was analyzed using the REX1 program (Rigaku Co.)

Results and Discussion

Samples with small Mn content .—

The compositional analysis from ICP-AES revealed that for the samples with the amount of residual lithium in the oxidized products from Reaction 1 for 24 h is less than 0.02 mol per formula unit. In addition, the amount of lithium after partial lithiation by LiI for 24 h based on Reactions 2 and 3 was close to the expected values. Therefore, for our samples with having uniformly small particle size, 24 h is long enough for the reaction to go to completion. The delithiation-lithiation reactions were confirmed by X-ray diffraction to be topotactic; the structure of the host was changed only by atomic displacements with no diffusive rearrangement. The X-ray diffraction (XRD) profiles for are shown in Fig. 2. Variations in the three lattice constants of the orthorhombic Pmnb lattice a, b, and c with the lithium composition x are summarized in Fig. 3.

Figure 2.

Figure 2. XRD patterns measured at fixed y values 0.2, 0.4, 0.6) with various Li composition x for Broken lines are for the single-phase region and solid lines for the two-phase region.

Figure 3.

Figure 3. Variations in the three lattice constants of the orthorhombic Pmnb lattice a, b, c with lithium composition x at fixed y values 0.2, 0.4, 0.6) for The single-phase regions are shaded to show the general tendency. The lines are for visual guidance.

A typical two-phase reaction proceeds in 1 The Mn substitution for Fe in the octahedral 4c sites results in (i) the formation of a two-phase region at (4.1 V vs. where all of the three orthorhombic lattice constants show discrete elongation by Li insertion, and (ii) a partial conversion of the reaction for the region of (3.4 V vs. from two-phase to single-phase, where the a and b lattice constants increase, and the c lattice constant decreases. As a result of this two-phase to single-phase conversion in and the growth of the region, the two-phase region in disappears at The tendency for changes in unit cell dimensions (lithium insertion induces an elongation of the a and b axis and a slight contraction of the c axis) is common for all samples with any Mn content y.

The change in the valence state of Fe was analyzed using Mössbauer spectroscopy, and showed the and redox reactions in the compositional regions and respectively. The boundaries at were distinct. A typical example, the spectrum for can be seen in a previous report.8 One of the advantages of Mössbauer spectroscopy is its ability to detect short-range symmetry and the local electronic state, and hence to identify the number of phases, including amorphous, nanoparticle, and/or ions with different valence states.6 7 8 However, no X-ray indiscernible nonolivine phase (amorphous or nanoparticle) was detected for any sample with This, in turn, strongly suggests that the reaction is exclusively topotactic.

Samples with large Mn content .—

The (local) lattice distortion induced by the Jahn-Teller effect of 19 makes unstable when the content y is large. For the samples with reversible topotactic delithiation-lithiation reactions of occurred within the limited Li composition x but had very slow kinetics; the phases close to the point (x, y) = (0, 1) are not stable enough to maintain the olivine framework.7 As demonstrated in our previous reports, the fully delithiated forms, and could not be isolated in a stable equilibrium state.7

Although an accurate compositional analysis of Li content x in the olivine phase was not possible, the overall trend of the reaction within the limited lithium composition x is dominated by the two-phase character of XRD profiles measured for with nominal composition 0.2, 0.4, 0.6, 0.8, 1.0, are shown in Fig. 4. These lithium compositions are, of course, not accurate because they are based on the assumption that LiI reacts with pure which includes an X-ray indiscernible nonolivine phase.7 However, we can conclude from Fig. 4 that the reaction mechanism is dominated by the two-phase reaction of For the situation was almost the same.

Figure 4.

Figure 4. XRD patterns measured for These samples include the X-ray indiscernible nonolivine phase, which could be detected only by using Mössbauer spectroscopy. The compositions x in parenthesis are based on the assumption that LiI reacts with pure

Local structures.—

For short-range characterization, EXAFS provides independent information on the local structures (coordination number, bond length, etc.) around Fe and Mn. Of particular interest here is the local structure around in the charged state, In our previous paper,7 we reported the existence of -induced anisotropic lattice distortion in the binary system, and the resultant equilibrium solubility-limit in the compositional range However, the analysis was based on the average long-range dimensional change of the orthorhombic lattice found using XRD. In phospho-olivines, the edge-shared arrangement of the oxygen-octahedra as well as the localized character of the 3d electrons will enhance the selective local lattice distortion around the Jahn-Teller active ions.

The normalized extended X-ray absorption fine structure (EXAFS) χ weighted by as shown in Fig. 5, are consistent with this. The EXAFS, are extracted from the absorption spectra after the pre-edge background is subtracted, and are normalized by the absorption spectrum obtained assuming that absorbing atoms are isolated, The function was determined using the cubic spline method as follows: was separated into several parts; was independently estimated in each part and was smoothly connected at the boundary between the neighboring parts. The energy, E, is converted to the photoelectron wavenumber, where is the threshold energy of the absorption edge and m is the free-electron mass. The difference in amplitude is evident for of the significant damping suggests the enhancement of local disorder by the Jahn-Teller effect of 20 21 22 23 24 In addition, the spectrum shapes for and (Fig. 5c and d) are similar, while those for and are completely different (Fig. 5a and b). These experimental results strongly suggest that, in the process of oxidation (charging) of to changes in the atomic position are induced mainly around

Figure 5.

Figure 5. Normalized EXAFS χ weighted by on the K edge of (a) in (b) in (c) in and (d) in measured at 300 K. (a) and (c) were charged; (b) and (d) were discharged.

The Fourier transform of shows a radial atomic distribution-like function in real space about the emitting metal ions (Fig. 6). Note that these peaks are at shorter distances than the actual interatomic distance, because the phase shift in the electron-scattering process is not considered. Attention should also be paid to the peak around Fe-O at ca. 1.1 A, which is a false peak not indicating any actual chemical bonding. A nonessential ghost peak in the short bond-length region often appears, it is dependent on the unrestricted processing in the REX1 program. However, all other peaks were well defined, as is shown in Fig. 6. The numbers in parentheses in this figure are the coordination numbers for the emitting metal ions. As is shown in Fig. 6c and d, for and have similar distributions, indicating the local environments are almost identical.

Figure 6.

Figure 6. Fourier transform of of (a) in (b) in (c) in and (d) in measured at 300 K. (a) and (c) were charged; (b) and (d) were discharged.

On the other hand, the amplitude of the peak in is unusually low compared to the peak in and is significantly reduced as the content y increases (Fig. 6a and b). The peak height is related to the backscattering of photoelectrons by the coordinated atoms; the atoms arranged in coherent coordination shells have a larger contribution.22 When the distances between the absorber atom and the coordination atoms are not uniform, as in the case of a distorted oxygen octahedron, the related peak shows a broadening and an apparent decrease in the peak height due to interference between the real and imaginary part of the spectrum.23 Consequently, the local lattice distortion around the Jahn-Teller active is much more severe than that expected from the XRD data, and this situation promotes phase destabilization close to the point

(x, y) Two-dimensional phase diagram.—

Based on the results of the preceding three sections the (x, y) two-dimensional phase diagram of the system shown in Fig. 7 establishes some general trends for the phase change. The (x, y) phase map in Fig. 7 is divided into four areas: (a) the unstable region close to the point (b) the two-phase region caused by (d) the two-phase region caused by (a part of and (c) the single-phase region caused by connecting (b) and (d).

Figure 7.

Figure 7. The (x, y) two-dimensional phase diagram of the system obtained using XRD and Mössbauer spectroscopy. Information is simply given on the single-phase or two-phase matter together with the valence states of Mn and Fe. The map is divided into four areas: (a) the unstable region close to the point (b) the two-phase region by (closed circles; (d) the two-phase region by (open circles; a part of and (c) the single-phase region by connecting (b) and (d) (open triangles).

It is noteworthy that the redox reaction (region of proceeds entirely in a two-phase manner. This can be readily understood by the first-order transition between the Jahn-Teller active phase with cooperative elastic deformation and the Jahn-Teller-inactive phase. Another important aspect is that the single-phase region (c) appears only in in Mn/Fe solid solution system but not in The origin of the appearance of this single-phase region c is not clear, but it is reasonable to speculate conceptually that the random distribution of divalent manganese in region c may dilute the weak cooperative interaction that discretely adjusts the framework to lithium insertion/extraction and makes the inherent redox reaction in a two-phase type.

The phase diagram in Fig. 7 leads us to the intuitive idea that is not entirely inactive in the electrochemical charge-discharge reaction at 4.1 V vs. and may be useful as a stable yet compatible 4 V cathode for and However, our electrochemical investigations of gave negative results even though the uniform small particles were carefully prepared. showed an open-circuit voltage of 4.1 V, but with much smaller capacity than the theoretical value of 170 mAh/g and very large polarization in the galvanostatic charge/discharge mode. In comparing with as practical 4 V cathodes, several intrinsic obstacles in must be overcome, such as the competitive materials cost, the limited capacity, the much slower reaction kinetics, the need to synthesize in inert gas with a divalent source, and a much lower true density (3.4 g/cm3 vs. 4.2 g/cm3 Thus, at the moment, olivine-type as well as Mn-rich remain far away from a practical application.

Conclusion

A room-temperature (x, y) two-dimensional phase diagram of the olivine-type solid-solution, orthorhombic, D: Pmnb) is presented. The manganese substitution for iron in the octahedral 4c sites induces 1, the two-phase redox region with a potential of 4.1 V vs. 2, a partial conversion of the reaction form in the redox region 3.4 V vs. from two-phase to single-phase; and 3, phase instability in the composition region close to the point caused by the Jahn-Teller effect of Conversion 2 is complete at around and the phase instability 3 makes the Mn-rich phase unsuitable for battery applications. The local lattice deformation around is severe enough to induce significant selective damping in the EXAFS for

Acknowledgments

The authors thank K. Ishikawa for ICP measurements. Helpful and enlightening discussions with Dr. S. C. Chung and Dr. M. Hosoya are greatly appreciated.

Sony Corporation assisted in meeting the publication costs of this article.

Please wait… references are loading.