Electrochimica Acta 56 (2010) 592–599 Contents lists available at ScienceDirect Electrochimica Acta journal homepage: www.elsevier.com/locate/electacta Influence of the tetravalent cation on the high-voltage electrochemical activity of LiNi0.5 M1.5 O4 spinel cathode materials My-Loan-Phung Le a,b,c , Pierre Strobel a,∗ , Fannie Alloin b , Thierry Pagnier b a b c Institut Néel, CNRS/Université Joseph Fourier, BP 166, 38042 Grenoble Cedex 9, France Laboratoire d’Electrochimie et de Physicochimie des Matériaux et Interfaces (LEPMI), Grenoble Institute of Technology, BP 75, 38402 Saint-Martin-d’Hères, France VietNam National University, VNU-HCM, Ho Chi Minh City, VietNam a r t i c l e i n f o Article history: Received 28 May 2010 Received in revised form 31 August 2010 Accepted September 2010 Available online 15 September 2010 Keywords: Lithium batteries High-voltage cathodes Spinel Lithium–manganese–nickel oxide a b s t r a c t The electrochemical properties of substituted LiNi0.5 Mn1.5−x Mx O4 spinels at high potential (>4 V vs Li+ /Li) have been investigated for M = Ti and Ru, in order to determine the role of the tetravalent cation in such systems where nickel is a priori the only electroactive species These systems are found to form extended solid solutions (up to x = 1.3 and x = 1.0 for Ti and Ru, respectively) that were characterized by X-ray diffraction and Raman spectroscopy Titanium substitution induces a drastic decrease in high potential electrochemical capacity, whereas the capacity is maintained and the kinetics are even improved in the presence of ruthenium These results are completed by new results on the Li4−2x Ni3x Ti5−x O12 spinel system, which shows not any high potential activity in spite of the presence of up to 0.5 Ni2+ per spinel formula unit on the octahedral site Taking into account previous data on LiNi0.5 Ge1.5 O4 , we clearly show that even if the tetravalent cation does not participate in the overall redox reaction, electrochemical activity is only possible when nickel is surrounded by tetravalent cations able to accept a local variation of valence (Mn, Ru), whereas full-shell cations such as Ti4+ and Ge4+ block the necessary electron transfer pathways in the spinel oxide electrode © 2010 Elsevier Ltd All rights reserved Introduction The ability of manganese-based based oxides to insert reversibly lithium at high voltage (>4.7 V) was discovered in the late 1990s [1–3] This property opened the way for new perspectives for lithium batteries, such as an increase in power density with respect to LiCoO2 and other “4V” materials, and the possibility to use negative electrodes working significantly above V while preserving a high power density The high-voltage plateau is mainly present in spinels where the B-site (often conventionally indicated by square brackets) contains a combination of Mn and another transition metal M, as in formula Li[Mx Mn2−x ]O4 (M = Cr, Fe, Co, Ni, Cu), where x is close to the maximum value x = 0.5 [4,5] The high-voltage capacity depends on the nature and concentration of M; numerous studies showed that best performances are achieved for M = Ni in compositions equal or close to LiNi0.5 Mn1.5 O4 [3–5] This compound contains Ni2+ and Mn4+ ions exclusively The superior performances of LiNi0.5 Mn1.5 O4 compared to other compositions was analyzed considering energy diagrams, showing that the Ni2+/3+ and Ni3+/4+ 3d bands have the least overlap with the O2− 2p band [6] As a result, the high-potential redox reaction ∗ Corresponding author Tel.: +33 476 887 940; fax: +33 476 881 038 E-mail address: Pierre.strobel@grenoble.cnrs.fr (P Strobel) 0013-4686/$ – see front matter © 2010 Elsevier Ltd All rights reserved doi:10.1016/j.electacta.2010.09.004 in LiNi0.5 Mn1.5 O4 does not involve oxygen loss and allows stable cycling, unlike most other compositions, including those involving the hypothetical Mn4+/5+ redox couple [7,8] There is indeed strong experimental evidence (mostly from X-ray absorption near edge spectroscopy studies) that the highpotential redox mechanism involves only nickel via the Ni2+/3+/4+ redox states [9–11] The generally accepted reaction mechanism is: LiMn1,5 4+ Ni0,5 2+ O4 ⇔ Li + Mn1,5 4+ Ni0,5 4+ O4 (I) A confirmation is provided by the existence of a double step at ca 4.7 V vs Li/Li+ corresponding to successive steps involving the Ni2+/3+ and Ni3+/4+ redox couples Within this reaction scheme, the Mn4+ cation does not participate in the reaction; in fact, manganese manifests itself mainly in an additional V plateau appearing when the material contains a fraction of Mn3+ due to non-stoichiometry [2,3,12] According to some authors, the electrochemical inactivity of manganese could be a factor of stability of the structure on cycling [12] Assuming reaction I to be correct, similar properties should be expected if Mn4+ is replaced by another tetravalent cation in the LiNi2+ 0.5 M4+ 1.5 O4 formula Kawai et al [2] reported the absence of electrochemical activity in LiNi0.5 Ge1.5 O4 and LiNi0.5 Ti1.5 O4 , although the latter showed a complex XRD pattern that could not be indexed as a single-phase spinel Two recent studies on the M.-L.-P Le et al / Electrochimica Acta 56 (2010) 592–599 LiNi0.5 Mn1.5−x Tix O4 system (0 < x < 1.5) showed a severe drop in reversible capacity for large titanium contents [12,13] Another peculiarity of LiNi0.5 Mn1.5 O4 is the occurrence of Ni:Mn cation ordering on the spinel B-site, that results in a lowering of symmetry from face-centered cubic (space group Fd-3m) to primitive cubic (space group P43 32) [14] This effect is present in some LiM0.5 Mn1.5 O4 phases (M = Cu, Ni) and not in others (M = Co), depending on the ionic radius difference between M2+ and Mn4+ [15] It seems not to be a major factor in electrochemical properties [3,16] In this paper, we re-examine the role of the tetravalent cation on the high-voltage electrochemical behaviour in LiNi0.5 Mn1.5−x Mx O4 and LiNi0.5 M1.5 O4 (M = / Mn) compounds This study encompasses the LiNi0.5 Mn1.5−x Tix O4 and LiNi0.5 Mn1.5−x Rux O4 spinel solid solutions In view of the non-existence of spinel-type “LiNi0.5 Ti1.5 O4 , a third system has been studied in an attempt to reach this stoichiometry by introducing nickel in Li4 Ti5 O12 , i.e making the solid solution Li4−2x Ni3x Ti5−x O12 , which yields “Li3 Ni1.5 Ti4.5 O12 ” = “LiNi0.5 Ti1.5 O4 ” for x = 0.5 The Li–Ni–Ti–O spinel solid solution has not been studied previously to our knowledge Although ruthenium compounds are not realistic choices for battery materials because of the cost of ruthenium, Ru4+ is one of few substitution candidates as a tetravalent cation in spinels We are aware of only one recent study of the Li–Ni–Mn–Ru–O spinel system, where substitution by ruthenium was attempted not on tetravalent manganese, but on nickel, assuming questionable cation-deficient formulas “LiNiy Mn1.5 Ru0.05 O4 ” with y = 0.35 and 0.4 [17] No higher doping level was attempted The availability of electrochemical data on all these systems will allow drawing systematic conclusions about the role of the “inactive” tetravalent cation on the high-voltage electrochemical reaction in LiNi0.5 M1.5 O4 spinel oxides Experimental 2.1 Synthesis procedure All compositions were prepared by solid state reaction Starting reagents were Li2 CO3 or LiCH3 COO (Merck), Ni(CH3 COO)2 4H2 O, TiO2 , RuO2 and MnCO3 (Aldrich) Samples with appropriate stoichiometry ratios were mixed with mortar and pestle and fired initially in air at 600–700 ◦ C for 20 h to decompose all of the carbonate, nitrate and acetates The materials were then re-grounded, pressed into pellets and reacted repeatedly 24 h at 900 ◦ C, with furnace cooling (ca 200 ◦ C/h) For titanium-rich and rutheniumcontaining compositions, higher temperatures were necessary: 1000 ◦ C (Ti) or 1200 ◦ C (Ru) for the final annealing 593 Fig XRD patterns in the LiNi0.5 Mn1.5−x Tix O4 series as a function of titanium content x (values of x are indicated on left side) Inset: variation of the spinel cell parameter with x and temperature regulation The powder was placed on the substrate of a laboratory-made cell under vacuum and cooled by liquid nitrogen [18] 2.3 Electrochemical measurements Electrochemical test was carried out in liquid electrolyte at room temperature using Swagelok-type batteries Cathodic pastes were prepared by intimately mixing the oxide powder with carbon black and PTFE emulsion in weight ratio 70:20:10 This paste was rolled down to 0.1 mm thickness, cut into pellets with diameter 10 mm and dried at 130 ◦ C under vacuum Typical active material masses used were 15–20 mg/cm2 The electrolyte was a M solution of LiPF6 in EC-DMC 1:2 (Merck Co.) Negative electrodes were 200m thick lithium foil (Metall Ges., Germany) Cells were assembled in a glove box under argon with ≤2 ppm H2 O Electrochemical studies were carried out using MacPile Controller (Bio-Logic, Claix, France) in the potential window 3.50–4.85 V vs Li/Li+ , in either galvanostatic mode at C/20–C/25 regime or by step-potential electrochemical spectroscopy (SPES) [19], using typically 10–20 mV/1 h steps Results and discussion 3.1 Structural characterization 2.2 Structural characterization Samples were analyzed by powder X-ray diffraction (XRD) using a Siemens D-5000 diffractometer with Co K␣ radiation, 0.02◦ step and 20 s/step counting time to minimize noise Lattice parameters were determined by a least squares method (CELREF software) The morphology and the distribution of grain size were determined using a LEO S440 scanning electron microscopy (SEM) instrument equipped with EDX spectroscopy analysis Raman spectroscopy (RS) measurements were carried out with a Renishaw’s InVia Raman Spectrometer Spectra were obtained with the red line of a laser (785 nm) in micro-Raman configuration (objective x 50) In the Li–Ni–Ti–O case, the structure stability and cation arrangement were also probed at low temperature and low frequency The spectra were measured on a Jobin-Yvon T64000 Raman spectrometer equipped with the green line of an Ar-ion laser (514.5 nm) as excitation source, a vacuum bench apparatus 3.1.1 LiNi0.5 Mn1.5−x Tix O4 solid solution Ten different titanium contents covering the whole range of compositions (0 ≤ x ≤ 1.5) were prepared by solid-state reactions at 900–1000 ◦ C Spinel phases are found to form up to x = 1.3 (see Fig 1) The terminal “LiNi0.5 Ti1.5 O4 ” compositions, however, does not yield a spinel phase, but is multiphase with a new hexagonal compound Li4 Ni3 Ti8 O21 as major component; the crystal structure of this new phase has been solved recently using electron diffraction [20] In the spinel solid solution, the cubic lattice parameter a increases linearly as a function of titanium content x, in agreement with previous results [12,21] (see Fig 1, inset) This is consistent with the increase in octahedral ionic radius between Mn4+ and Ti4+ ˚ respectively [22]) A detailed structural analysis (0.53 and 0.61 A, is given elsewhere, showing more complex features such as the disappearance of Ni:Mn cation ordering with titanium substitution, partial cation inversion and the presence of a second, minor 594 M.-L.-P Le et al / Electrochimica Acta 56 (2010) 592–599 phase identified as Ni1−x Lix O [23] The latter seems to be a rather systematic property of LiNi0.5 Mn1.5 O4 , and was reported for various syntheses routes [1,2,16,24–28] The expulsion of a fraction of nickel from the spinel phase induces a nickel deficiency in the spinel phase, as in formula LiNi0.5−ı (Mn,Ti)1.5+ı O4 A significant consequence is the presence of a fraction of Mn3+ : charge compensation yields n(Mn3+ ) = 2ı per spinel formula Values of ı have been estimated as 0.06 ± 0.02 from structural analysis, corresponding to 0.12 ± 0.04 Mn3+ per spinel formula unit [23] The morphology of LiNi0.5 Mn1.5−x Tix O4 samples was studied by scanning electron microscopy (SEM) Unsubstituted LiNi0.5 Mn1.5 O4 prepared by solid-state reaction exhibits a wide distribution of particle size (Fig 2) The presence of titanium seems to promote grain growth, as shown by the fairly large, well-faceted octahedral particles for x = 0.3–0.45 (see Fig 2b and c) A similar effect has already been reported in titanium-substituted LiMn2 O4 [13,29] 3.1.2 LiNi0.5 Mn1.5−x Rux O4 solid solution As shown in Fig 3, single-phase spinels were obtained in this system up to x = At higher contents, unreacted RuO2 remains present, even after repeated firings at 1200–1300 ◦ C Samples were furnace-cooled (ca 100 ◦ C/h) to ensure a stoichiometric oxygen content Electrochemical measurements (see Section 3.3) will indeed show that the Mn3+ content in ruthenium-substituted samples, that is directly correlated to the V capacity, is similar in Ti- and in Ru-containing spinels Attempts to prepare the terminal phase “LiNi0.5 Ru1.5 O4 failed, yielding mixtures of the rocksalt-type phase Li(Mn,Ru)O2 and of RuO2 The structural features are very similar to those of the titanium case: disappearance of B-site cation ordering, presence of the minor rocksalt-type phase for all compositions and lattice parameter increase with ruthenium content The increase in a, however, is much smaller than in the titanium case (compare y-scales in insets of Figs and 3), in spite of the neighbouring ionic radii of Ti4+ ˚ and Ru4+ (0.62 A) ˚ [22] This could be due to differences in (0.61 A) cation distribution, especially partial inversion detected in the titanium case We also note a broadening of X-ray reflections that may reflect a lower cristallinity and/or a higher cationic disorder in the ruthenium case 3.1.3 Li4−2x Ni3x Ti5−x O12 solid solution The syntheses in this system started from the Li[Li1/3 Ti5/3 ]O4 spinel formula and aimed at introducing nickel on the octahedral sites Charge balance imposes a double substitution of 3Ni2+ for 2Li+ + Ti4+ The extent of this solid solution is theoretically limited by the amount of octahedral lithium (1/3 per spinel formula or per 4-5-12 formula unit), yielding a maximum x value of 0.5, where all octahedral lithium atoms are substituted XRD diagrams and the evolution of cell parameter (Fig 4) show that the solid solution range is limited to x ≈ 0.3 Up to this value, the lattice parameter a increases as expected For x > 0.3, a decreases and additional diffraction peaks appear The terminal composition (x = 0.5) is confirms the non-existence of the “LiNi0.5 Ti1.5 O4 ” spinel phase Fig SEM micrographs of the different spinels (initial magnification 4000× (5000× for x = 0); x = (a), x = 0.3 (b), 0.45 (c) 3.2 Raman spectroscopy Raman spectroscopy is a local probe that is very sensitive to local symmetry changes It is used here to detect possible octahedral cation ordering that is known to occur in LiNi0.5 Mn1.5 O4 [14] In this stoichiometric compound, the 1:3 ratio of Ni2+ :Mn4+ cations with different ionic radii induces an ordered cation distribution of Ni2+ and Mn4+ in two different crystallographic sites As a result, the symmetry is lowered from space group (SG) Fd-3m to P43 32 Such ordering is difficult to detect by X-ray diffraction because of the weak chemical contrast between Mn and Ni We showed previously [14,15] that this symmetry change can be easily detected by neutron diffraction and by vibration spectroscopy Group theory predicts five Raman active modes for a normal AB2 O4 spinel with SG Fd-3m (Oh ), while the lower-symmetry P43 32 structure has 42 Raman active modes [26,30] As showed previously [15,31], cation ordering shows up in Raman spectra by a narrowing of Raman bands and important changes such as the appearance of additional bands at 166 and 411 cm−1 and a doublet near 600 cm−1 These features are illustrated in Fig 5, where the two upper spectra correspond to cation-ordered LiNi0.5 Mn1.5 O4 and disordered LiNi0.4 Mn1.6 O4 In addition, we checked that low-temperature Raman spectra (acquired at 128 K) show no narrowing of the cation- M.-L.-P Le et al / Electrochimica Acta 56 (2010) 592–599 Fig Raman spectra of different Li–Ni–Mn–O, Li–Ni–Mn–Ru–O spinels (formulas indicated) Fig XRD patterns in the LiNi0.5 Mn1.5−x Rux O4 series as a function of nominal ruthenium content x (from bottom to top: x = 0.25, 0.50, 0.75, 1.0) Inset: variation of the spinel cell parameter with x disordered Raman bands This clearly indicates that disorder is static, probably due to local variations of the cation–cation interactions Fig shows that substitution by titanium, and to a lesser extent by ruthenium, induces a considerable broadening of Raman lines and suppresses the 600 cm−1 doublet, i.e present clear evidence of cation disorder This has been confirmed by neutron diffraction on Ti-substituted samples [23] Regarding the Li4−2x Ni3x Ti5−x O12 system, Raman spectra for different compositions (0 ≤ x ≤ 0.5) are shown in Fig For x ≤ 0.25, a single phase is observed and the Raman spectrum is that of a cation disordered spinel [32] For x = 0.375, however, Raman spectra recorded at several sample positions are different, showing the presence of two phases in this sample These spectra can be considered as combinations of the x = 0.25 and x = 0.50 spectra The solid solution limit lies therefore between x = 0.25 and x = 0.375, in agreement with the crystallographic data For x = 0.50, a single Raman spectrum is obtained whatever the point illuminated This strongly suggests that the second phase, observed in X-ray diffrac- Fig XRD patterns in the Li4−2x Ni3x Ti5−x O12 series as a function of titanium content x Inset: variation of the spinel cell parameter with x 595 Li–Ni–Mn–Ti–O and tion experiments, is minor and/or that the Raman spectrum of the second phase is weak or absent, as expected for a nearly cubic perovskite We checked that the Raman spectrum of “LiNi0.5 Ti1.5 O4 obtained in this series for x = 0.5 is identical to that of the sample prepared by Gemmi et al [20] We can thus attribute this Raman spectrum to the new phase Li4 Ti8 Ni3 O21 Fig Raman spectra of different Li4−2x Ni3x Ti5−x O12 compositions (x values indicated) Samples are single-phase spinel up to x = 0.25 596 M.-L.-P Le et al / Electrochimica Acta 56 (2010) 592–599 Table Main oxidation and reduction potentials (maximum current peak values vs Li/Li+ ) measured by slow scanning voltammetry (10 mV/h) Formula Ered /V (±0.01) Eox /V (±0.01) LiNi0.5 Mn1.5 O4 LiNi0.5 MnRu0.5 O4 LiNi0.5 Mn1.35 Ti0.15 O4 LiNi0.5 Mn1.2 Ti0.3 O4 LiNi0.5 Mn1.05 Ti0.45 O4 LiNi0.5 Mn0.9 Ti0.6 O4 4.70 4.72 4.76 4.77 4.76 4.78 4.77 4.75 4.79 4.79 4.78 4.80 The LiNi0.5 Mn1.5 O4 main redox peak is split into two components attributable to the Ni2+/3+ and Ni3+/4+ redox couples [33] The Rusubstituted peak has a very similar shape, with sharper peaks giving a better resolution of the two components, and a lower potential difference between anodic and cathodic current peak maxima On the contrary, in the Ti-substituted case, the main current peak is considerably broader In addition, the reduction peak maximum is shifted to higher potential: from 4.69 to 4.76 V vs Li+ /Li This value (4.76 ± 0.01 V) is found to be constant in the whole range of titanium compositions (0.15 ≤ x ≤1.30) Peak maxima data are summarized in Table 1, showing that the tetravalent cation is not as “inert” as expected in the electrochemical reaction The fact that the Ti or Ru substitution modifies the current peak width, the reaction potential and the reduction–oxidation peak difference indicates that the nature of the tetravalent cation influences both reaction kinetics and octahedral cation–anion bond strength The remarkable influence of ruthenium on kinetics confirms the recent report by Wang et al [17] The effect of substitution on electrochemical capacity can be seen in Fig The effect of titanium substitution is drastic and unexpected in view of reaction I cited in the introduction: the capacity drops steeply with titanium substitution beyond ca 0.3 Ti per spinel formula, in spite of the fact that the nickel content is unchanged The oscillations measured at the end of discharge for x(Ti) ≥ 0.6 indicate a severe deterioration of intercalation reaction kinetics in these materials In addition, a detailed analysis of the charge–discharge curves shows that the capacity decrease is entirely due to the highvoltage reaction: the V plateau capacity remains constant (see Fig 9), meaning that the Mn3+/4+ redox process is not influenced by substitution, whereas the Ni2+/3+/4+ is The capacities are stabilized after a few cycles and the effect of titanium is constant on extended cyclings (Fig 9b) These results confirm previous reports Fig Slow-scanning voltammetry (step-potential electrochemical spectroscopy) of LiNi0.5 Mn1.5 O4 and LiNi0.5 Mn1.35 Ti0.15 O4 (3rd cycle, sweeping rate 10 mV/h; data normalized to 0.1 mmol active material) 3.3 Electrochemical behaviour Fig shows SPES voltammetry scans of unsubstituted LiNi0.5 Mn1.5 O4 (a), and of samples substituted by 0.5 Ru (b) and 0.15 Ti (c) vs lithium metal, all measured at same potential sweeping rate All show a main, reversible, high-potential reaction around 4.75 V vs Li/Li+ and a slight capacity around V that is ascribed to the Mn3+ fraction resulting from non-stoichiometry in the spinel phase Fig 3rd cycle galvanostatic charge-discharge curves of LiNi0.5 Mn1.5−x Tix O4 spinels for various titanium contents x M.-L.-P Le et al / Electrochimica Acta 56 (2010) 592–599 597 Fig 10 3rd cycle charge–discharge curves of LiNi0.5 Mn1.5−x Mx O4 spinels for similar contents of Ti and Ru Inset: Evolution of the capacity of LiNi0.5 MnRu0.5 O4 with cycle number (conditions as in Fig 9) with a theoretical capacity of 0.5Li per spinel formula Yet experimental results show the absence of any significant high potential capacity in this material (see Fig 11) This negative result is not due to experimental artefacts, since this material behaves quite normally in the potential range 1–2 V, where the Ti4+/3+ redox couple is electrochemically active and gives a clean, reversible plateau at 1.5 V vs Li/Li+ (see Fig 11, inset) 3.4 Discussion: high-potential reaction mechanism XANES studies of LiNi0.5 Mn1.5 O4 showed that nickel is the only electro-active species in the high-potential (“5 V”) reaction Assuming this model, where Mn4+ is inactive, it should be possible to modify the tetravalent cation composition without interfering with the nickel redox behaviour Titanium substitution presents the additional interest to slightly lower the spinel molar mass, resulting in a higher theoretical specific capacity Our study showed that, contrary to expectations, substitutions on “inert” Mn4+ affect the electrochemical performances, in a positive and negative way Fig (a) V and V capacity of LiNi0.5 Mn1.5−x Tix O4 spinels as a function of x (3rd cycle data, galvanostatic cycling at C/20 between 3.75 and 4.85 V (b) Evolution of the capacity with cycle number for different titanium contents about the detrimental effect of titanium substitution [12,13] and will be discussed in more detail in Section As shown in Fig 10, the capacity drop is much smaller in the Li–Ni–Mn–Ru–O system, in spite of an increase in molar mass that decreases the specific capacity at constant lithium intercalation level in the ruthenium case On the contrary, it could be expected that the lower atomic mass of titanium with respect to manganese would induce a gain in specific capacity, but the opposite trend is observed The inset in Fig 10 also shows the high stability on cycling of the ruthenium-substituted spinel Turning now to the Li4−2x Ni3x Ti5−x O12 system, we showed in Section 3.1.3 that single-phase spinels were obtained up to x = 0.25, corresponding to a spinel composition Li3.5 Ni0.75 Ti4.75 O12 = Li[Li0.167 Ni0.25 Ti1.583 ]O4 A priori, the reaction mechanism I should apply, at least partially (since the spinel formula contains only 0.25 Ni2+ ), as follows: Li[Li0.167 + Ni0.25 2+ Ti1.583 4+ ]O4 ⇔ 0.5Li + Li0.5 [Li0.167 + Ni0.25 4+ Ti1.583 4+ ]O4 Fig 11 Charge–discharge cycles on Li3.75 Ni0.375 Ti4.875 O12 in the potential window 3.75–4.85 V (C/22 regime) Inset: charge–discharge behaviour in the 1–2 V range 598 M.-L.-P Le et al / Electrochimica Acta 56 (2010) 592–599 Fig 12 Environment of a Ni cation in the LiNi0.5 Mn1.5 O4 structure Table Electrochemical activity and electronic properties of tetravalent cations found in Li–Ni spinel oxides M4+ High-potential activity Electron configuration Mixed-valence possibility (3–5 V) Ti Mn Ge Ru (Sn) Poor/none Yes None Yes (Spinel not obtained) d0 d3 d0 d4 No Yes No Yes Conclusions with ruthenium and titanium, respectively The detrimental effect of titanium is confirmed by studies in the Li–Ni–Ti–O spinel system, which does not give any electrochemical activity in the high potential range (Fig 12) If we add to the set of available data in this study the LiNi0.5 Ge1.5 O4 case, that was found to be electrochemically inactive in the range 3.5–5 V [3], the electrochemical properties of various LiNi0.5 M1.5 O4 spinels are summarized in Table For the sake of completeness, we may add that we attempted to synthesize a Sn4+ -containing spinel with hypothetical formula “LiNi0.5 Sn1.5 O4 All syntheses of such a compound, including attempts under high pressure (6 GPa), failed, and produced mixtures of LiNiO2 , NiSnO3 and SnO2 Table shows an obvious correlation between the electronic configuration of the tetravalent cation and the high-potential electrochemical activity: the electrochemical reaction seems to require the possibility of mixed valence on the tetravalent cation This result can be understood considering the intercalation–deintercalation mechanism of lithium in the spinel host structure This requires three main steps: (1) migration of lithium in or out of the host lattice, (2) electron exchange (redox reaction) on nickel, (3) transfer of these electrons from/to the surface of the spinel electrode In predominantly ionic compounds such as transition metal oxides, this last step is expected to take place via an electron hopping mechanism Considering now the environment of a nickel atom in the spinel structure (see Table 3), structural data show that, in the cationordered structure of LiNi0.5 Mn1.5 O4 , the first coordination shell of Table Interatomic distances (in nm) in ordered and disordered LiNi0.5 Mn1.5 O4 spinels Coordination shell Ordered spinel (P43 32) Disordered spinel (Fd-3m) 1st neighbours 2nd neighbours 3rd neighbours Next Ni neighbour 6O Mn Li >0.5 6O 6M Li 0.205 0.292 0.334 nickel is oxygen, and the second one consists of tetravalent cations √ only at a distance a/ 8; nickel–nickel distances are exceedingly high (>0.5 nm) In the absence of cation ordering (as in Ti- and Ru-substituted spinels), this scheme is not strictly applicable, but electrostatic and steric considerations still favour a B-site cation distribution with Ni2+ locally surrounded by M4+ cations The major hopping conduction pathways will then be Ni–O–M–O– ., and such a transfer is only possible if M cations can accept a local variation of valence, whereas full-shell cations such as Ti4+ or Ge4+ will act as blocking elements for the electron transfer This scheme is fully confirmed by experimental results: the best electron transfer takes place in the presence of Ru, a 2nd-transition row element with larger electron delocalization than 1st transition row elements; ruthenium is indeed notorious for forming numerous conducting (and even superconducting) oxides [34,35] In the Li–Ni–Mn–Ti–O system, which is the most studied to date, an abrupt capacity decrease with increasing Ti content at high potential was already noted, in spite of chemical diffusion determinations (from GITT measurements) giving an increase in the lithium chemical diffusion coefficient with Ti substitution [12] This confirms that the blocking effect of the d0 Ti4+ cations is the major factor limiting the redox intercalation–deintercalation mechanism 0.196 0.289 0.334 In this study, we investigate the influence of different tetravalent cations M4+ on the high-potential redox reaction involving nickel in oxide spinels Most possible M elements have been considered: titanium manganese, germanium, ruthenium and tin Ti and Ru give rise to extended solid solutions in the nickel-containing LiNi0.5 (Mn1.5−x Mx )O4 systems, although this solid solution does not extend to the terminal phases ‘LiNi0.5 Ti1.5 O4 ’ and ‘LiNi0.5 Ru1.5 O4 ’ Although the tetravalent cation is a priori “inert” in the redox reaction, a clear correlation is found between the feasibility of reversible lithium electrochemical intercalation involving Ni2+/4+ oxydo-reduction and the ability of the neighbouring tetravalent cations in the structure to allow the necessary electron transfer to and from nickel atoms This parameter proves to be a key requirement for the use of nickel as an active redox species in lithium-containing spinels Best results are obtained with mixedvalence neighbouring elements such as Mn and Ru, with the latter improving electrochemical kinetics probably due to its larger electron delocalization Acknowledgment The authors thank région Rhône-Alpes for financial support of My Loan Phung Le References [1] Q.M Zhong, A Bonakdarpour, M.J Zhang, Y Gao, J Dahn, J Electrochem Soc 144 (1997) 205 [2] K Amine, H Tukamoto, H Yasuda, Y Fujita, J Power Sources 68 (1997) 604 [3] H Kawai, M Tabuchi, M Nagata, H Tukamoto, A.R West, J Mater Chem (1998) 1273 [4] H Kawai, M Nagata, H Tukamoto, A.R West, J Power Sources 81/82 (1999) 67 [5] T Ohzuku, S Takeda, M Iwanaga, J Power Sources 81/82 (1999) 90 [6] Y Shin, A Manthiram, Electrochim Acta 48 (2003) 3583 [7] Y Shin, A Manthiram, Electrochem Solid State Lett (2003) A249 [8] W Wen, B Kumarasamy, S Mukerjee, M Auinat, Y Ein-Eli, J Electrochem Soc 152 (2005) A1902 [9] S.J Hwang, H.S Park, J.H Choy, G Campet, J Phys Chem B 105 (2001) 335 [10] Y Terada, K Yasaka, F Nishikawa, T Konishi, M Yoshio, I Nakai, J Solid State Chem 156 (2001) 286 [11] H Shigemura, H Sakaebe, H Kageyama, H Kobayashi, A.R West, R Kanno, S Morimoto, S Nasu, M Tabuchi, J Electrochem Soc 148 (2001) A730 [12] J.H Kim, S.T Myung, C.S Yoon, I.H Oh, Y.K Sun, J Electrochem Soc 151 (2004) A1911 M.-L.-P Le et al / Electrochimica Acta 56 (2010) 592–599 [13] R Alcántara, M Jaraba, P Lavela, J.L Tirado, Ph Biensan, A.D Jordy, J.P Peres, Chem Mater 15 (2003) 2376 [14] P Strobel, A Ibarra-Palos, M Anne, F Le Cras, J Mater Chem 10 (2000) 429 [15] P Strobel, A Ibarra-Palos, M Anne, C Poinsignon, A Crisci, Solid State Sci (2003) 1009 [16] D Pasero, N Reeves, V Pralong, A.R West, J Electrochem Soc 155 (2008) A282 [17] H Wang, H Xia, M.O Lai, L Lu, Electrochem Commun 11 (2009) 1539 [18] O Sze Nga Sum, E Djurado, T Pagnier, N Rosman, C Roux, E Siebert, Solid State Ionics 76 (2005) 2599 [19] A.H Thompson, J Electrochem Soc 126 (1979) 608 [20] M Gemmi, H Klein, A Rageau, P Strobel, F Le Cras, Acta Cryst B 66 (2010) 60 [21] D Gryffroy, R.E Vandenberghe, J Phys Chem Solids 53 (1992) 777 [22] R.D Shannon, Acta Cryst A 32 (1976) 751 [23] M.L.P Le, P Strobel, C.V Colin, T Pagnier, F Alloin, J Phys Chem Solids, submitted [24] S.T Myung, S Komaba, N Kumagai, H Yashiro, H.T Chung, T.H Cho, Electrochim Acta 47 (2002) 2543 [25] [26] [27] [28] [29] [30] [31] [32] [33] [34] [35] 599 S.H Park, Y.K Sun, Electrochim Acta 50 (2004) 431 M Kunduraci, G.G Amatucci, J Electrochem Soc 153 (2006) A1345 L Yu, Y.L Cao, H.X Yang, X.P Ai, J Solid State Electrochem 10 (2006) 283 G.Q Liu, H.W Xie, L.Y Liu, X.X Kang, Y.W Tian, Y.C Zhai, Mater Res Bull 42 (2007) 1955 L Hernan, J Morales, L Sanchez, J Santos, J Solid State Ionics 118 (1999) 179 C.M Julien, F Gendron, A Amdouni, M Massot, Mater Sci Eng B 130 (2006) 41 M Kunduraci, G.G Amatucci, J Power Sources 165 (2007) 359 C.M Julien, M Massot, K Zaghib, J Power Sources 136 (2004) 72 K.A Striebel, A Rougier, C.R Horne, R.P Reade, E.J Cairns, J Electrochem Soc 146 (1999) 4339 R.J Cava, Dalton Trans (2004) 2979 P.M Woodward, J Goldberger, M.W Stoltzfus, H.W Eng, R.A Ricciardo, J Am Ceram Soc 91 (2008) 1796 ... width, the reaction potential and the reduction–oxidation peak difference indicates that the nature of the tetravalent cation influences both reaction kinetics and octahedral cation anion bond strength... was attempted The availability of electrochemical data on all these systems will allow drawing systematic conclusions about the role of the “inactive” tetravalent cation on the high-voltage electrochemical. .. Inset: variation of the spinel cell parameter with x disordered Raman bands This clearly indicates that disorder is static, probably due to local variations of the cation cation interactions Fig