Layered Li_xMn_1−yLi_yO₂ intercalation electrodes: synthesis, structure and electrochemistry

Abstract

Layered Li_xMn_1−yLi_yO₂ materials with the O3 (α-NaFeO₂) structure have been synthesised by a low temperature ion exchange route from the corresponding sodium compounds. The effects on electrochemical performance (ability to store reversibly large quantities of lithium and hence charge) and crystal chemistry by partially substituting some of the Mn ions by Li have been investigated by structural and electrochemical techniques as well as chemical analysis. The materials deliver high discharge capacities; in excess of 200 mA h g⁻¹ at 25 mA g⁻¹ or C/8. On the first charge once all Mn is in the 4+ oxidation state further Li⁺ removal occurs by two unconventional mechanisms, one involving oxidation of O²⁻ with subsequent O loss (effective removal of Li₂O) and the other electrolyte oxidation generating H⁺ which exchanges for Li⁺ in the electrode. Although both mechanisms appear to occur at 30 and 55 °C, the former dominates at both temperatures. A mechanism for O loss involving O release at the surface with Mn migration into the bulk in order to fill up vacant octahedral sites in the transition metal layers is proposed, consistent with the neutron diffraction data. Evidence for tetrahedral Li and Li loss from the octahedral sites in the transition metal layers is also presented. The compounds in this study irreversibly transform to spinel-like materials on extended cycling. This is not, however, detrimental to their electrochemical performance and is analogous to the behaviour of other O3 layered lithium manganese oxides.

---

Introduction

Rechargeable lithium batteries are one of the major technological success stories of the past decade.^1,2 Intercalation compounds based on lithium manganese oxides such as LiMn₂O₄ spinel, orthorhombic or layered LiMnO₂ are receiving intense interest as potential alternatives to LiCoO₂ for use as the positive electrode in rechargeable lithium batteries.^1–8 In fact, LiMn₂O₄ is already used in cells. Such interest is driven by the lower cost, lower toxicity, higher safety and, in the case of layered LiMnO₂ for example, superior capacity to store charge on cycling.^5,8–11 Layered LiMnO₂ is structurally analogous to LiCoO₂, possessing the α-NaFeO₂ structure, and was first reported in 1996.^5,6 The synthesis involves low temperature ion exchange from the corresponding sodium phase. The initial reports of stoichiometric LiMnO₂ showed that it was possible to remove almost all the lithium during the first charge, corresponding to 280 mA h g⁻¹, which is twice that of LiCoO₂.^5,6 The subsequent discharge capacity was significantly less and indeed, on further charge–discharge cycling, the amount of charge delivered was comparable to LiCoO₂ and at a lower voltage. Later studies have shown that it is possible to prepare non-stoichiometric Li_xMn_yO₂ which contains vacancies in both the alkali and transition metal layers.¹⁰ Varying the synthesis conditions can alter the number of transition metal vacancies and, to a certain extent, the lithium content. Preparing the sodium phase under more oxidising conditions and/or using milder ion exchange conditions favours more transition metal vacancies. It is possible to substitute some of the Mn ions with a range of ions to form solid solutions of the type: Li_xMn_y−zM_zO₂ (M = e.g. Ni and Co).^9,11–15 The doped materials offer higher discharge capacities, in excess of 200 mA h g⁻¹, and excellent cycling stability with fade rates of only 0.08% per cycle. Recently we and others have discussed the processes accompanying the layered to spinel transformation.^16,17 We have also addressed the fact that the first charge capacity of the non-stoichiometric materials is lower than the subsequent discharge.¹⁸

We have reported the synthesis of Li_xMn_1−yLi_yO₂ materials (0 ≤ y ≤ 0.2) by an analogous ion exchange route.¹⁹ These compounds exhibit unconventional lithium extraction on the first charge beyond the point at which all the Mn is in the +4 oxidation state. There are two mechanisms involved in this unconventional removal of Li; these are the loss of oxygen (effective removal of Li₂O) as proposed by Dahn et al. for Li(Mn,Ni,Li)O₂ and Thackeray et al. for acid leaching of Li₂MnO₃,^20–23 and proton exchange for Li⁺, in which the protons are generated by the oxidation of the non-aqueous electrolyte.^24,25 The interplay between these mechanisms is subtly dependent on the conditions, particularly temperature. In our earlier work we focused on the synthesis and the electrochemical behaviour on the first charge; here we discuss the electrochemical performance of such Li_xMn_1−yLi_yO₂ materials on extended cycling and show that it compares favourably with other singly-doped Li_xMn_1−yM_yO₂ systems but with the advantage of containing manganese as the only transition metal ion. We also present further information on the structural changes occurring on the first cycle for these compounds.

Experimental

Na_xMn_1−yLi_yO₂ materials were prepared by mixing Na₂CO₃ (Aldrich 99.5+%), Li₂CO₃ (Aldrich 99+%), and Mn(CH₃CO₂)₂·4H₂O (Aldrich 99+%) in distilled water in stoichiometric ratios corresponding to x = 0.67 and y = 0.025, 0.05, 0.1 and 0.2. After rotary evaporation the mixtures were heated in air at 250 °C for 12 h to decompose the acetates. The resulting powder was then fired at 600 °C in air for 1 h followed by quenching. The sodium phases were ion exchanged by refluxing in ethanol at 80 °C with a seven- to eight-fold molar excess of LiBr for up to 8 h. The products were filtered, washed with distilled water and ethanol and dried.

Chemical analyses for sodium and lithium were carried out by flame emission and for manganese using atomic absorption spectroscopy. The average manganese oxidation state was determined by redox titration using ferrous ammonium sulfate/KMnO₄. Further information on the manganese oxidation state was obtained from XPS measurements on as-prepared and charged Li_xMn_1−yLi_yO₂ composite electrodes. Binding energies were charge corrected using the C 1s peak (284.4 eV).

Powder X-ray diffraction data were collected on a Stoe STADI/P diffractometer operating in transmission mode with Fe Kα₁ radiation (λ = 1.936 Å) to eliminate manganese fluorescence.

The structures of the materials were further characterised using neutron diffraction. Time-of-flight powder neutron diffraction data were collected on the Polaris and GEM high intensity, medium-resolution instruments at ISIS, Rutherford Appleton Laboratory. Since lithium, and to a lesser extent manganese, are neutron absorbers, the data were corrected for absorption. The structures were refined by the Rietveld method using the program Prodd based on the Cambridge Crystallographic Subroutine Library (CCSL).^26,27 Neutron scattering lengths of −0.19, −0.373, and 0.5803 (all 10⁻¹² cm) were assigned to Li, Mn, and O respectively.²⁸ Powder neutron diffraction experiments were also carried out on electrochemically charged samples on the GEM diffractometer. The samples were contained in 2 mm quartz capillaries.

In order to evaluate the performance of the layered materials, composite electrodes were constructed by mixing the active material, Kynar Flex 2801 (a copolymer based on PVDF) and carbon, in the weight ratios 85 : 5 : 10. The mixture was prepared as a slurry in THF and spread onto aluminium foil using a Doctor Blade technique. Following evaporation of the solvent and drying, electrodes were incorporated into an electrochemical cell in which the second electrode was formed from lithium metal and the electrolyte was a 1 molal solution of LiPF₆ (Hashimoto) in ethylene carbonate–diethylene carbonate (V/V 1 : 2 (Merck)). Electrochemical measurements were carried out using a Biologic MacPile II. We define C-rate here as 1C being equal to a cell discharging in 1 h. The C-rate may be calculated for any cell at any cycle number by dividing the rate (in mA g⁻¹) by the capacity (in mA h g⁻¹) e.g. if the rate is 50 mA g⁻¹ and the capacity of the cathode is 200 mA h g⁻¹ then the C-rate is C/4.

Combined thermogravimetric analysis–mass spectrometry (TGA-MS) measurements were performed using a Netzsch STA449 Jupiter instrument coupled with a Pfeiffer Vacuum Thermostar GSD300T. The TGA heating rate was 5 °C min⁻¹ up to 400 °C under an argon atmosphere.

Results and discussion

Composition and crystal structure

Results of detailed compositional analysis for the materials after ion exchange are presented in Table 1 (note that the composition of the undoped material is included for comparison). The Li doping is introduced during the solid state synthesis of the Na_xMn_1−yLi_yO₂ phase and, given the size difference of Na⁺ and Li⁺, is unlikely to be located anywhere other than in the transition metal layers. We apportion the total Li content, determined from the chemical analysis and Rietveld refinement (see later), as shown in Table 1. The total alkali metal content in the alkali metal layers is approximately 2/3, consistent with the alkali metal content of the Na phases prior to exchange and with the parent Li_2/3Mn_zO₂ phase. A small amount of Na remains after ion exchange although this decreases continuously with increasing levels of lithium doping. There is a clear trend towards a reduced vacancy content in the transition metal layers with increased lithium doping. The reduction in Na content and vacancy concentration is in keeping with our observations on other systems (particularly undoped Li_xMn_yO₂).¹⁰

Table 1 Compositions and oxidation states for various Li_xMn_1−yLi_yO₂ materials obtained from chemical analysis and Rietveld refinement. Ions in alkali metal layers ( ), ions in transition metal layers [ ]

Li doping level (y)	Mn valence	Formula
0.0	3.55	(Li0.59Na0.069)[Mn0.94]O2
0.025	3.61	(Li0.61Na0.047)[Mn0.92Li0.025]O2
0.05	3.64	(Li0.60Na0.041)[Mn0.91Li0.05]O2
0.10	3.70	(Li0.61Na0.032)[Mn0.88Li0.1]O2
0.20	3.83	(Li0.66Na0.015)[Mn0.82Li0.18]O2

---

Let us now consider the structures of the doped Li_xMn_1−yLi_yO₂ phases in more detail; fitted neutron diffraction patterns for the compositions y = 0.05 and 0.20 are shown in Fig. 1. Similar quality fits were obtained for the other compositions. In each case the structure was refined using a rhombohedral layered model in space group R3̄m, as has been reported previously for Li_xMn_yO₂ and its derivatives.^9–14 The structure is essentially that of layered LiCoO₂ with oxygen atoms occupying the 6c sites, [Mn, Li and vacancies] the 3a sites and (Li, residual Na and vacancies) the 3b sites of space group R3̄m. In the refinements the sodium contents of the 3b sites were fixed at the values obtained from chemical analysis (Table 1). The Mn occupancies of the 3a sites were also fixed at the values obtained from the compositional analysis, whilst the Li occupancy was allowed to vary. The occupancy of the oxygen site was fixed at unity. Crystallographic parameters for y = 0.2 are shown in Table 2. Excellent agreement was obtained between observed and calculated patterns as can be seen from Fig. 1 and the very good weighted profile R-factors quoted in Table 2.

Fig. 1 Fits to neutron powder diffraction data for Li_x[Mn_1−yLi_y]O₂ phases; (a) y = 0.05; (b) y = 0.20.

Table 2 Refined crystallographic parameters for Li_x(Mn_1−yLiy)O₂, y = 0.2. Space group R3̄m. Lattice parameters a = 2.8522(3) Å, c = 14.4151(7) Å. R-factors: R_exp = 1.10%, R_wP = 2.21%, R_P = 3.35%

Atom	Wyckoff symbol	x/a	y/b	z/c	B 11 = B22 (Biso)	B 33	B 12	Site occupancy
Li/Na	3b	0	0	0.5	1.02(7)	—	—	0.73(2)/0.015
Mn/Li	3a	0	0	0.0	0.41(3)	0.79(5)	0.20(2)	0.82/0.18(2)
O	6c	0	0	0.26051 (6)	0.83(2)	0.82(2)	0.42(1)	1

---

Electrochemistry: cycling performance

Fig. 2 shows the cycling data at 30 °C between potential limits of 2.5 and 4.6 V for y = 0.025, 0.05, 0.10 and 0.20. Initial discharge capacities in excess of 200 mA h g⁻¹ at C/8 are achieved for each of these Li doped systems. Each of these materials exhibits very good capacity retention with low fade rates. For y = 0.2 a fade rate of 0.08% per cycle is observed between cycle 50 and 150. These values compare very favourably with some of the best performances reported previously for related layered compounds e.g. the single-doped Li_xMn_y−zCo_zO₂ (z = 0.025, M = Co) which also has a fade rate of 0.08% per cycle under these conditions.^9,15 The capacities are delivered above 3 V; these are not 4 V materials.

Fig. 2 Discharge capacity as a function of cycle number for Li_x[Mn_1−yLi_y]O₂ at 30 °C. Rate = 25 mA g⁻¹ (approximately C/8). Voltage limits = 2.5–4.6 V.

The effect of elevated temperatures on cycling stability is important for any potential electrode material and is particularly significant for Mn-rich electrodes because it is known that deleterious reactions occur with LiPF₆-based non-aqueous electrolytes resulting in Mn dissolution and increased capacity fade.^29,30Fig. 3 shows cycling data at 55 °C for y = 0.1 at high (1C) rates. The data show reasonable stability at 55 °C for over 200 cycles, such that the loss in capacity is only about 0.11% or 0.21 mA h g⁻¹ per cycle. It should be pointed out, however, that at these high rates cycling is completed quickly (in this case 1 h per cycle) and inferior stability may be expected at lower rates due to Mn dissolution. Surface modification by, for example, coating the particles would doubtless improve further the capacity retention at elevated temperature, as would an electrolyte not based on LiPF₆.

Fig. 3 Discharge capacity as a function of cycle number for Li_xMn_0.9Li_0.1O₂ at 55 °C. Rate = 200 mA g⁻¹ (approximately 1C). Voltage limits = 2.5–4.6 V.

We have investigated the cycling performance at a variety of charge–discharge rates, i.e. the ability to intercalate substantial quantities of lithium reversibly at different rates. Although LiCoO₂ has a relatively low capacity (130 mA h g⁻¹) it has good rate capability due to a high lithium diffusion coefficient and fast kinetics for transformation between the different ordered phases encountered as a function of Li content. Fig. 4 illustrates the rate capability at 30 °C of a typical Li-doped material, in this case y = 0.05. Increasing the rate by a factor of eight results in a loss in capacity of around 50% initially but dropping to only 10% after 50 cycles, such that a capacity of 175 mA h g⁻¹ is delivered at the 1C rate. This value is almost 40% higher than LiCoO₂. The initial rise is associated with transformation of the layered structure to that of spinel (discussed below). Although this is relatively facile, sufficiently so to have little effect on the capacity at low (C/8) rate, it clearly does become a factor at these higher rates.

Fig. 4 Discharge capacity as a function of cycle number for Li_xMn_0.95Li_0.05O₂ at various charge–discharge rates of 25–200 mA g⁻¹. Voltage limits = 2.5–4.6 V.

Layered O3 lithium manganese oxide materials have been shown to convert irreversibly to spinel-like materials upon cycling with concomitant formation of a nanometre-scale nanostructure.^7,10,15,31 This nanostructure contains a mosaic or nanodomain structure within micron-sized particles of largely the same dimensions as the parent compound. Unlike lithium manganese oxide spinels prepared by conventional solid state reaction at high temperatures, the layered materials continue to show excellent reversibility when cycled over both the 3 and 4 V plateaux. This necessitates overcoming lattice strain caused by a co-operative Jahn–Teller distortion, which occurs when approximately 50% of the octahedral transition metal sites are occupied by Mn³⁺. In spinel this induces a first order phase transformation from cubic to tetragonal symmetry, with an anisotropic volume change. In the case of the nanostructured spinels derived from layered compounds, it is believed that slippage at the defect-rich domain wall boundaries of the nanostructure accommodate the strain caused by the transformation between the undistorted and distorted phases and allow far more facile phase transitions with consequently much higher cycling stability.

All of the Li doped samples convert to spinel-like materials upon cycling. Fig. 5 shows the incremental capacity plots for the y = 0.2 material. Double 4 V peaks, characteristic of spinel, are evident after 50 cycles. This result is consistent with previous observations for non-stoichiometric Li_xMn_yO₂ layered materials prepared by ion exchange in ethanol at 80 °C.^9–11,14,15 The rate of conversion to spinel is dependent on the number of transition metal vacancies/residual sodium ions and the ratio of c/a lattice parameters. Fewer vacancies/less sodium and a lattice parameter ratio closer to 4.9 cause the conversion to spinel to be more rapid. Other factors that affect the rate of conversion to spinel are the cycling rate and temperature. Higher temperatures and lower charge–discharge rates cause the conversion to occur at lower cycle numbers.¹⁵

Fig. 5 (a) Incremental capacity plots for Li_xMn_0.8Li_0.2O₂ as a function of cycle number. Cycling was carried out at 25 mA g⁻¹ (approximately C/8). Voltage limits = 2.5–4.6 V. (b) Variation of potential (vs. Li⁺[1M]/Li) on charging then discharging the Li_x[Mn_1−yLi_y]O₂ electrodes at 25 mA g⁻¹ (C/8) and 30 °C.

Structure/electrochemistry: the first cycle

For these Li deficient/highly oxidised phases the charge extracted on the first cycle is less than the discharge. We have discussed this phenomenon and how it may be addressed elsewhere.¹⁸ The incremental capacity plots, Fig. 5a, also show that there is an irreversible process close to 4.5 V on the initial charge which decreases in magnitude with increasing cycle number. The capacity at 4.5 V increases with the amount of Li in the transition metal layers as shown in the load curves for the first cycle, Fig. 5b.^19,32 The stoichiometric compound Li₂MnO₃ also shows a significant irreversible capacity at 4.5 V on the initial charge.^24,25 We have discussed this phenomenon for the Li_x(Mn_1−yLi_y)O₂ materials elsewhere,¹⁹ so the results will only be summarised here to place the structural studies in context. It has been shown that in all of these materials lithium extraction at 30 °C and 4.5 V occurs by simultaneous oxygen loss, i.e. effective Li₂O removal.^19,25,32 For compositions with extensive Li-doping, a second mechanism involving Li ⇔ H exchange takes over as the dominant mechanism at high levels of lithium extraction.^19,25 The processes involved in charging Li_xMn_1−yLi_yO₂ are as follows. The initial Li extraction is accompanied by conventional oxidation of manganese until all Mn is in the +4 oxidation state. Thereafter, further Li is removed at 4.5 V which lies within the oxygen valence band and will result in oxidation of O²⁻. This is unstable in the lattice leading to oxygen evolution (overall Li₂O removal). In addition, especially at higher states of charge and elevated temperatures, oxidation of the alkyl carbonates in the electrolyte yields H⁺, as shown previously for other Li–Mn–oxides.^24,25,33,34 The H⁺ will then exchange for Li⁺.

Powder neutron diffraction studies for samples halted at various states of charge on the first cycle have been performed for y = 0.2. Rietveld refinement of data obtained from a sample charged to the start of the 4.5 V plateau (i.e. prior to any O loss or H⁺ exchange) gave an excellent fit to a single phase adopting the O3 structure (Fig. 6a) with lattice parameters a = 2.850(1) Å, c = 14.446(2) Å. This is consistent with the proposed conventional oxidation of Mn³⁺ to Mn⁴⁺. Grey et al. have recently reported a combined NMR/theoretical study on the system Li[Li_(1−2x)/3Mn_(2−x)/3Ni_x]O₂ which also exhibits a 4.5 V plateau.³⁵ Here it was observed that Li removal from the transition metal sites occurs before the start of the plateau via tetrahedral sites in the lithium layers and is a reversible process. Extended cycling was found to induce disorder, presumably due to migration of the transition metal ions. We have observed similar tetrahedral site occupancy on charging layered LiMnO₂.¹⁶ Accordingly, further refinements were conducted incorporating such a tetrahedral Li site. This yielded some Li occupancy for this site and depletion of Li from the transition metal layers, albeit with only a slight improvement in the fit. Refined parameters are shown in Table 3. We do not rule out the possibility of some Mn migration to tetrahedral sites as well, but decoupling Mn from Li is beyond the capability of the neutron data.

Fig. 6 Profile fits for Rietveld refinement of Li_xMn_0.8Li_0.2O₂. (a) Charged to 4.5 V at 30 °C (single O3 phase model), (b) charged to 4.8 V at 30 °C, model incorporating two O3 phases, (c) charged to 4.8 V then discharged to 3.3 V at 30 °C, model incorporating two O3 phases. Dots indicate observed data, solid line calculated profile. The lower box shows difference/esd.

Table 3 Refined crystallographic parameters for Li_x(Mn_1−yLi_y)O₂, y = 0.2, charged to 4.5 V. Space group R3̄m. Lattice parameters a = 2.8501(7) Å, c = 14.444(2) Å. R-factors: R_exp = 2.67%, R_wP = 3.07%, R_P = 2.67%

Atom	Wyckoff symbol	x/a	y/b	z/c	B iso	Site occupancy
Li/Na	3b	0	0	0.5	0.6(4)	0.54(4)/0.015
Mn/Li	3a	0	0	0.0	0.55(14)	0.82/0.05(3)
Li	6c	0	0	0.118(5)	0.5	0.09(2)
O	6c	0	0	0.2622 (2)	0.64(5)	1

---

A sample charged to 4.8 V (end of charge) gave data which could not be fitted to a single phase model. Instead a model involving two O3 phases (Fig. 6b) having lattice parameters corresponding to: (1) similar to those obtained at 4.5 V and (2) a phase with similar a parameter (2.852(2) Å) and longer c parameter (14.646(6) Å), gave an excellent fit. The observed similarity in a parameters is to be expected since this has a strong correlation with Mn oxidation state (+4 in both cases). From our previous diffraction studies of a wide range of doped and undoped layered lithium manganese oxides, both as-prepared and after electrochemical cycling, the elongated c parameter in phase 2 points to a material of the form Li_xMn_yO₂, where y ∼ 1.^9–11,16 Such a material is exactly what would be expected to result from loss of Li₂O. The pattern obtained after subsequent discharge to 3.3 V again is best fitted using two phases (Fig. 6c) suggesting that complete removal of Li from the Mn layers does not occur on the first charge or is partly reversible. This is supported by the incremental capacity plots (Fig. 5) which continue to show the 4.5 V process on extended cycling.

Combining the first charge load curves and the structure refinements we can propose the following processes on first charge. The charge curve below 4.5 V shows some structure (see for example the y = 0.1 composition in Fig. 5b) suggesting that this is not a single process. It is likely that Li is first removed from the alkali metal layers, the creation of octahedral lithium vacancies in the 3b sites enables lithium (and possibly Mn) to migrate from the 3a sites within the Mn layer to tetrahedral 6c sites. This may be associated with the inflection in the charging curve at ∼4 V, most evident in the y = 0.1 and y = 0.2 data (Fig. 5b). On further charging the considerable vacancy content is eliminated by loss of Li₂O from the surface of the particles along with Mn migration from the surface to the bulk to fill vacant sites. This is consistent with the formation of a 2nd phase in which the Mn content of the transition metal layers approaches more closely to 1, as observed in our neutron data of a highly charged sample. Examination of the load curves (Fig. 5b) reveals that on the first discharge there is a Li insertion process occurring close to 4 V for y = 0.025 and 0.05. This suggests that for these compositions the lower vacancy concentration stabilises the structure in the charged state, and enables the tetrahedral (4V) sites to be repopulated on Li insertion.

Electrochemistry: cycling to 4.3 V

It was decided to investigate the cycling performance for these materials reducing the upper voltage cut-off to 4.3 V, thereby avoiding the 4.5 V plateau. Cycling data are presented in in Fig. 7a and incremental capacity plots for y = 0.1 are shown in Fig. 7b. As expected the capacities are lower for the more highly doped compositions, but these also show a pronounced rise in capacity with cycle number. Examination of the incremental capacity plots reveals that despite the lower voltage cut-off, these compounds still transform to spinel at a comparable rate. The discovery that Li can leave the transition metal layers and occupy tetrahedral sites at potentials up to 4.5 V is consistent with this observation.

Fig. 7 (a) Discharge capacity as a function of cycle number for Li_x[Mn_1−yLi_y]O₂ at 30 °C. Rate = 25 mA g⁻¹. Voltage limits = 2.5–4.3 V. (b) Incremental capacity plots for Li_xMn_0.9Li_0.1O₂ as a function of cycle number. Cycling was carried out at 25 mA g⁻¹, voltage limits 2.5–4.3 V.

Acknowledgements

PGB is indebted to the EPSRC, the EU, and The Royal Society for financial support.

References

1. P. G. Bruce, Chem. Commun., 1997, 1817.
2. J.-M. Tarascon and M. Armand, Nature, 2001, 414, 359.
3. M. M. Thackeray, W. I. F. David, P. G. Bruce and J. B. Goodenough, Mater. Res. Bull., 1983, 18, 461.
4. J.-M. Tarascon, W. R. McKinnon, F. Coowar, T. N. Bowmer, G. Amatucci and D. Guyomard, J. Electrochem. Soc., 1994, 141, 1421.
5. A. R. Armstrong and P. G. Bruce, Nature, 1996, 381, 499.
6. F. Capitaine, P. Gravereau and C. Delmas, Solid State Ionics, 1996, 89, 197.
7. H. Wang, Y.-I. Jang and Y.-M. Chiang, Electrochem. Solid-State Lett., 1999, 2, 490.
8. J. M. Paulsen, C. L. Thomas and J. R. Dahn, J. Electrochem. Soc., 1999, 146, 3560.
9. A. D. Robertson, A. R. Armstrong and P. G. Bruce, Chem. Commun., 2000, 1997.
10. A. R. Armstrong, A. J. Paterson, A. D. Robertson and P. G. Bruce, Chem. Mater., 2002, 14, 710.
11. A. D. Robertson, A. R. Armstrong, A. J. Fowkes and P. G. Bruce, J. Mater. Chem., 2001, 11, 113.
12. A. R. Armstrong, R. Gitzendanner, A. D. Robertson and P. G. Bruce, Chem. Commun., 1998, 1833.
13. A. R. Armstrong, A. D. Robertson, R. Gitzendanner and P. G. Bruce, J. Solid State Chem., 1999, 145, 549.
14. T. E. Quine, M. J. Duncan, A. R. Armstrong, A. D. Robertson and P. G. Bruce, J. Mater. Chem., 2000, 10, 2838.
15. A. D. Robertson, A. R. Armstrong and P. G. Bruce, Chem. Mater., 2001, 13, 2380.
16. A. R. Armstrong, A. J. Paterson, N. Dupre, C. P. Grey and P. G. Bruce, Chem. Mater., 2004, 16, 3106.
17. J. Reed, G. Ceder and A. van der Ven, Electrochem. Solid-State Lett., 2001, 4, A78.
18. A. J. Paterson, A. D. Robertson and P. G. Bruce, Electrochem. Solid-State Lett., 2004, 7, A331.
19. A. R. Armstrong and P. G. Bruce, Electrochem. Solid-State Lett., 2004, 7, 1.
20. Z. Lu and J. R. Dahn, J. Electrochem. Soc., 2002, 149, A815.
21. Z. Lu, D. D. MacNeil and J. R. Dahn, Electrochem. Solid-State Lett., 2001, 4, 191.
22. Z. Lu, L. Y. Beaulieu, R. A. Donaberger, C. L. Thomas and J. R. Dahn, J. Electrochem. Soc., 2002, 149, A778.
23. Y. Paik, C. P. Grey, C. S. Johnson, J.-S. Kim and M. M. Thackeray, Chem. Mater., 2002, 14, 5109.
24. A. D. Robertson and P. G. Bruce, Chem. Commun., 2002, 2790.
25. A. D. Robertson and P. G. Bruce, Chem. Mater., 2003, 15, 1984.
26. J. C. Matthewman, P. Thompson and P. J. Brown, J. Appl. Crystallogr., 1982, 15, 167.
27. P. J. Brown and J. C. Matthewman, Rutherford Appleton Laboratory Report, RAL-87-010, 1987.
28. V. F. Sears, Neutron News, 1992, 3, 26.
29. G. G. Amatucci, C. N. Schmutz, A. Blyr, C. Sigala, A. S. Gozdz, D. Larcher and J. M. Tarascon, J. Power Sources, 1997, 69, 11.
30. M. Yoshio, Y. Y. Xia, N. Kumada and S. H. Ma, J. Power Sources, 2001, 101, 79.
31. P. G. Bruce, A. R. Armstrong and R. L. Gitzendanner, J. Mater. Chem., 1999, 9, 193.
32. A. R. Armstrong, A. D. Robertson and P. G. Bruce, J. Power Sources (submitted).
33. A. Du Pasquier, A. Blyr, P. Courjal, D. Larcher, G. Amatucci, B. Gérand and J.-M. Tarascon, J. Electrochem. Soc., 1999, 146, 428.
34. M. Moshkovich, M. Cojocaru, H. E. Gottlieb and D. Aurbach, J. Electroanal. Chem., 2001, 497, 84.
35. C. P. Grey, W.-S. Yoon, J. Reed and G. Ceder, Electrochem. Solid-State Lett., 2004, 7, A290.

---