J O U R N A L O F C H E M I S T R Y Materials Lithium insertion in two tetragonal tungsten bronze type phases, M8W9O47 (M=Nb and Ta) Sagrario M. Montemayor, A. Alvarez Mendez, A. Martý�nez-de la Cruz, Antonio F. Fuentes* and Leticia M. Torres-Martý�nez Facultad de Ciencias Quý�micas, Divisio�n de Estudios Superiores, Universidad Auto�noma de Nuevo Leo�n, Apartado Postal 1625, Monterrey, Nuevo Leo�n, Mexico.E-mail: afernand@ccr.dsi.uanl.mx Received 10th June 1998, Accepted 21st September 1998 A study of lithium insertion in two tetragonal tungsten bronze (TTB) type phases of general formula M8W9O47 (M=Nb and Ta), is presented. The electrochemical insertion of up to 20 lithium atoms per formula unit in Nb8W9O47 (Li/SM=1.2) proceeds through a reversible reaction with several single phase and one two-phase domains, while in Ta8W9O47 the reversibility of lithium insertion is limited to 15 atoms (Li/SM=0.9).Structural changes on Nb8W9O47 as a function of the number of lithium atoms inserted have been studied by X-ray powder diVraction. dal coordination are preferentially occupied by niobium while Introduction in the corner sharing metal–oxygen octahedra, metal atoms The study of the chemistry of tungsten bronzes and related are on average (0.4 Nb+0.6 W).phases has attracted considerable attention because of their While only one polymorph for Nb8W9O47 has been reported potential use as electrodes, catalysts and in optical displays. to exist, two have been found for Ta8W9O47, a low (tetragonal ) Additionally, these phases were found to present interesting and a high temperature form (orthorhombic), both exhibiting optical and ferroelectric properties.In this context and continu- TTB type structures, although only in the high temperature ing with work started recently in our research group on one (1400 °C) is it possible to observe threefold TTB-type insertion reactions in some niobium–tungsten mixed oxides,1–3 superstructure.8 A similar cation distribution to that found a study of lithium insertion in two tetragonal tungsten bronze for the niobium–tungsten mixed phases was also proposed (TTB) type phases of general formula M8W9O47 (M=Nb and here.It is important to point out that although M8W9O47 is Ta), has been carried out. the ideal composition, a small range of non-stoichiometry may From the solid state chemistry point of view, the exist in both niobium and tantalum TTB type phases, due to Nb2O5–WO3 and Ta2O5–WO3 systems behave in a fairly disorder in the tunnels occupancy.similar way in the region rich in WO3. Phases containing Compounds presenting structures containing PCs as main between 50 and 80 wt.% of WO3, present a structure similar building units, such as W18O49, have been already studied as to the so-called tetragonal tungsten bronzes (TTB).These host materials for reversible lithium insertion reactions.9 In phases consist in a framework of MO6 octahedra sharing this work a study of the electrochemical lithium insertion in corners, linked in such a way that three, four- and five-sided M8W9O47 has been carried out.In order to evaluate the tunnels are formed. Decreasing oxygen-to-transition metal influence of lithium insertion on the structure of the parent ratio (3 in WO3 vs. 2.76 in M8W9O47) is achieved in these oxide, some lithiated phases were synthesised by indirect TTB type phases by filling a certain number of five-sided chemical reaction and characterised by X-ray powder diVractunnels with oxygen and metal atoms thus forming the so- tion experiments.As the main diVerence between both mixed called pentagonal columns (PCs for short):4aMO7 pentagonal tungsten oxides is the presence of Nb5+ instead of Ta5+, it bipyramid sharing equatorial edges with five MO6 octahedra. would be interesting to compare their behaviour versus lithium In Nb8W9O47 (Fig. 1) one third of the pentagonal tunnels are insertion reactions. No attempt is made in this work on testing M8W9O47 as cathode materials in lithium ion batteries.occupied in this way.5–7 Cation sites with pentagonal bipyrami- Fig. 1 Idealised structure of Nb8W9O47. When joining the diVerent polyhedra, five-, four- and three-sided tunnels are formed where additional ions can be inserted.J. Mater. Chem., 1998, 8, 2777–2781 2777the time of observation, intercalated samples were covered by Experimental a polyethylene film while recording data on the diVractometer Preparation and characterisation of pristine materials, to prevent sample oxidation. M8W9O47 The tantalum–tungsten and niobium–tungsten mixed oxides Results and discussion used in this study were synthesised by solid state reaction.The Electrochemical lithium insertion in Nb8W9O47 starting materials, WO3 (Aldrich Chem. Co, 99+%), Nb2O5 (Alfa Products 99.5%) and Ta2O5 (Aldrich Chem. Co, 99.99%), Electrochemical lithium insertion in Nb8W9O47 was carried were weighed in the appropriate stoichiometric ratio (4 out by discharging a cell with the following configuration: M2O559 WO3) and thoroughly mixed by grinding in an agate Li|1 mol dm-3 LiClO4 in (50% DEE+50% EC)|Nb8W9O47 mortar using Analar grade acetone.Powders were then pressed into 10 mm diameter pellets, placed in a platinum crucible and Results obtained from SPECS experiments run down to 1.1 V fired at 1250 °C in an electrical furnace. During firing samples vs. Li+/Li using ±10 mV (2 h)-1 potential steps, are shown were periodically extracted from the furnace and ground to in Fig. 2 as an E vs.composition (x in LixNb8W9O47) plot. favour reaction before being finally quenched at room tempera- As can be seen in this graph, Nb8W9O47 can reversibly ture. Phase identification was carried out by X-ray powder incorporate up to 20 lithium atoms per formula unit which diVraction in a Siemens D-5000 diVractometer using Cu-Ka correspond to about 1.2 Li per metal atom.Almost all lithium radiation (l=1.5418 A° ). A typical diVraction experiment for atoms inserted are removed after completing a charge– determining cell parameters was run with a step size of discharge cycle. The main features of this plot are two plateaux, 0.12° min-1 using KCl as internal standard.A and B, of approximately constant E values around 2.1 and 1.7 V vs. Li+/Li separating three regions where a continuous variation of E with composition is observed. In order to Electrochemical lithium insertion determine the existence of continuous transformations or to Electrochemical experiments were carried out with a multi- multiphase regions, potentiostatic experiments were analysed channel potentiostatic-galvanostatic system MacPile II,10 using in more detail.Representing these results as DQ/m vs. E, Fig. 3, a SwagelokTM type cell11 with metal lithium acting simul- it is possible to see that the electrochemical lithium insertion taneously as negative and reference electrode. Positive elec- in this material goes really through three very well defined trodes were prepared by mixing the phase being tested, reduction steps labelled here as A, B and C.The electrochemical Nb8W9O47 or Ta8W9O47, with carbon black and a binder, lithium extraction from LixNb8W9O47 follows a similar mech- (0.5% ethylene–propylene–diene terpolymer, EPDT, in cyclo- anism and three analogous steps, labelled as A¾, B¾ and C¾, are hexane) either in a 8951051 ratio (wt.%) for niobium contain- also present.ing samples or in a 5954051 ratio for the tantalum compound It was observed that the nature of the first step ( labelled as (this material is not a good electronic conductor so a larger A and A¾) was always better defined on oxidation and second proportion of carbon has to be used). The electrolyte used was a 1 mol dm-3 solution of LiClO4 in a previously dried 50550 mixture of ethylene carbonate (EC) and diethoxyethane (DEE).Cell assemblage was carried out in a MBraun glove box under an argon atmosphere with continuous purge of water vapour and oxygen ensuring an inside concentration for both compounds of <1.5 ppm. Two diVerent electrochemical experiments were carried out on these in a current or in a potential-controlled mode.Potentiodynamic titrations were carried out by a stepwise technique also known as step potential electrochemical spectroscopy, SPECS.12 In this technique, the potential is stepwise increased or decreased while recording charge increments vs. time at each potential level and therefore allowing the study of insertion reaction kinetics.The simultaneous determination of incremental capacities and Fig. 2 Evolution of cell voltage versus composition (x in observation of the kinetics of the redox process helps in LixNb8W9O47) obtained during a SPECS experiment run at determining the succession of single phase and two phase ±10 mV (2 h)-1 potential steps showing a complete charge– domains which might take place on insertion/deinsertion.discharge cycle. Typical experimental conditions were set at ±10 mV (2 h)-1 potential steps with a charge recording resolution of 5 mA h and a strict temperature control. Lithium chemical insertion Lithium insertion was also carried out by chemical reaction of Nb8W9O47 with an appropriate reducing agent, n-butyllithium in n-hexane. A given volume of this reagent (1.6 mol dm-3) was added to a known weight of host material previously placed in a glass reaction vessel.The mixture was stirred for 7 days. Reaction products were thoroughly washed with n-hexane, dried and stored in the glove box. Lithium content in LixNb8W9O47 was analysed by AAS using a Varian SpectrAA5 spectrometer after extraction of the inserted ions using concentrated nitric acid.Phase identification was carried Fig. 3 DQ/m vs. E plot obtained from a SPECS experiment run at out by X-ray powder diVraction as described above. Although ±10 mV (2 h)-1 potential steps showing the diVerent reduction and oxidation steps observed on lithium insertion and extraction. lithium inserted Nb8W9O47 proved to be stable at least during 2778 J. Mater. Chem., 1998, 8, 2777–2781Fig. 6 I/m and E vs.time plot obtained when crossing at Fig. 4 Evolution of cell current (,,) and voltage (———) with time during a SPECS experiment run using+10 mV (2 h)-1 potential steps. -10 mV (12 h)-1 potential steps the voltage region where peak B was observed. reduction than on the first discharge. Therefore, we shall start of the amount of lithium atoms inserted.14 A similar behaviour examining a chronoamperogram obtained when crossing at of the current decay versus time is observed on the third +10 mV (2 h)-1 potential steps, the potential region where the reduction peak labelled in Fig. 3 as C making for a second oxidation peak A¾ was observed, Fig. 4. The I vs. time plot continuous transformation between 1.3 and 1.6 V vs. Li+/Li shows a profile obviously not governed by a simple diVusion on lithium insertion in Nb8W9O47.process which would give a monotonic tendency towards I= A confirmation of these observations is presented in Fig. 7 0, but it is typical of a first order transition. In this type of which shows a comparison between two voltamperograms process, the response to a potential step will depend on the obtained when running SPECS experiments ±10 mV (2 h)-1 mobility of the interface between the two phases, relative to (filled circles) and ±10 mV (12 h)-1 (open squares) potential lithium diVusivity in both phases and on the kinetics of lithium steps.While steps labelled as A and A¾ show hysteresis in both transfer at the interface with the electrolyte assuming that experiments (initial slopes can be aligned) independently of there is no electronic conductivity limitation.13 Thus, it could the voltage scanning rate used, confirming our previous be assumed that the first reduction step A which corresponds assumption of dealing with a first order phase transition at to the first plateaux (at higher E values) also labelled as A in this potential range, a diVerent situation is observed for steps Fig. 2, can be associated with a multiphase domain separating labelled B and C. In these latter cases and for the experiment two solid solution regions, I and II. The phase I–phase II run at ±10 mV (12 h)-1, hardly any hysteresis eVect is noticed equilibrium potential can be given as the intercept at the slope between reduction and oxidation peaks. On the contrary, extrapolations at zero current, i.e. 2.12 V. Fig. 5 shows the peaks B¾ and C¾ are now found located on top of peaks B and evolution of the cell voltage versus composition obtained when C supporting the idea of crossing two consecutive continuous cycling a similar cell to the one described elsewhere in this phase transitions between 1.8 and 1.3 V vs. Li+/Li. Therefore, work, between 3 and 1.75 V vs.Li+/Li. As can be seen in this a complete phase diagram Li–Nb8W9O47 can be now proposed graph, phase I transformation to phase II is a reversible process as follows: above 2.12 V vs. Li+/Li, a solid solution domain with all the atoms initially incorporated being extracted after (I) from x=0 to x#2. At 2.12 V vs. Li+/Li, a two phase completing a charge–discharge cycle.Since potentiostatic equilibrium between phase I and phase II. Below 2.12 V vs. experiments run at ±10 mV (2 h)-1 potential steps did not Li+/Li, a solid solution domain (II) from x=6. Around 1.7 V reveal unambiguously the exact nature of reduction steps vs. Li+/Li, an incremental capacity peak probably due to a labelled in Fig. 3 as B and C and their corresponding oxidation tendency to local ordering in phase II around 11<x<12.steps B¾ and C¾, it was necessary to run experiments at much Below 1.7 V vs. Li+/Li, continuation of phase II with another lower scan rates [±10 mV (12 h)-1]. Time dependence of the possible tendency to local ordering giving rise to an increase current when crossing at -10 mV (12 h)-1 potential steps, the of the incremental capacity of the cell around 1.4–1.5 V vs. voltage region where the reduction peak labelled in Fig. 3 as Li+/Li. B was found, is shown in Fig. 6. In this case, the monotonic tendency towards I=0 observed in the current decay vs. time plot, can be associated with a continuous transformation. In homogeneous solutions and assuming that there is no ion–ion interaction, the system behaves in a similar way independently Fig. 7 Comparison between two DQ/m vs. E plots obtained Fig. 5 E (V vs. Li+/Li) vs. lithium content (x) in LixNb8W9O47 plot when running SPECS experiments at ±10 mV (2 h)-1 and ±10 mV (12 h)-1 potential steps. obtained when cycling a cell within a limited potential window. J. Mater. Chem., 1998, 8, 2777–2781 2779Fig. 8 E vs. lithium content (x) in Lix Nb8W9O47 plot obtained after completing three charge–discharge cycles.Fig. 10 E vs. lithium content (x) in LixTa8W9O47 plot obtained from Lithium insertion reaction reversibility in this material can a SPECS experiment run at ±10 mV (2 h)-1 potential steps showing be better appreciated in Fig. 8 where results obtained after two complete charge–discharge cycles within limited potential completing three charge–discharge cycles in a cell configured windows.as mentioned above are presented. Cycling is carried out with minimal capacity losses. Data shown were obtained from a galvanostatic experiment run at a cycling rate greater than plateaux indicating multiphase regions. As can be seen in this C/76 (C/76 corresponds to a charge or a discharge within 76 h).plot, Ta8W9O47 incorporates a larger number of lithium atoms According to the structure shown in Fig. 1, a total of 26 per formula unit (27, Li/SM=1.6) than its niobium analogue diVerent tunnels are found in a unit cell of this phase which (20 Li/formula unit) between 3.1 and 1.1 V. However, almost will also contain two formula units of Nb8W9O47. These 17% of those atoms remained in the structure after completing cavities are distributed as follows: 8 five-sided tunnels; 6 four- a charge–discharge cycle showing therefore the existence of sided and 12 three-sided channels.In order to be able to larger structural changes on insertion. Fig. 10 shows the evolaccommodate 40 lithium atoms (20 per formula unit) in 26 ution of the cell voltage versus composition for a similar cell tunnels of a unit cell, multiple lithium occupancy has to be to the one described above but cycled up to 1.85 (open dots) assumed in some cavities (ratio of lithium atoms intercalated and 1.35 volts (solid dots) showing this time that the insertion in a unit cell to available cavities =1.54). Based on tunnel of up to 15 lithium atoms in Ta8W9O47 (Li/SM=0.9), is a size, it is most likely that only one lithium atom would enter reversible reaction. into three-sided tunnels.This assumption would leave 28 Both ions Nb5+ and Ta5+ have similar ionic radii (0.64 A ° , lithium atoms to be distributed in 14 tunnels which makes a six-coordinate),15 but contain a diVerent number of electrons lithium atoms to available cavities ratio of 2.Thus, It is (36 and 68 respectively). Therefore, Ta5+ has a larger electronic reasonable to suggest that two lithium atoms would enter in density than Nb5+ which makes lithium diVusion more diYcult each of the four (coordination number 12) and five-sided in Ta8W9O47 relative to that in Nb8W9O47 and proving that tunnels (coordination number 15). With the available data it these metals play an important role in the insertion reaction.is not possible to give a filling sequence of these cavities. An attempt was made to prepare lithiated phases by chemical reaction with n-butyllithium but these proved to be extremely Electrochemical lithium insertion in Ta8W9O47 unstable upon exposure to the atmosphere. A preliminary study of the electrochemical lithium insertion Chemical lithium insertion in Nb8W9O47 in Ta8W9O47, an isostructural phase with Nb8W9O47 containing tantalum instead of niobium, was carried out by discharg- In order to study the influence of lithium insertion on the ing a similar cell to the one described above. Results obtained structure of the parent oxide, diVerent inserted compositions from a SPECS experiment run at ±10 mV (2 h)-1 potential included in the solid solution regions detected during the steps, are shown in Fig. 9 as the evolution of cell voltage electrochemical study, were prepared by chemical reaction of versus composition (x in LixTa8W9O47). Only slight slope Nb8W9O47 and n-butyllithium. Inserted materials prepared changes are observed with apparently no existence of voltage were characterised as previously described and resulted in the following compositions: Li1.8Nb8W9O47, Li7.3Nb8W9O47 and Li18.8Nb8W9O47. Phase characterisation was carried out by Xray powder diVraction using KCl as internal standard.Powder patterns, which are shown in Fig. 11, were indexed using the same orthorhombic cell described in the literature for the starting material by using a least-squares cell refinement program.Results are shown in Table 1. Powder patterns showed a gradual change in position and intensity of some reflections as the insertion reaction proceeded which can be related with small displacements of the atoms in Nb8W9O47 in order to accommodate the lithium ions. The most important changes are observed between diVraction patterns b and c which correspond to phase I transition to phase II.A colour change for these materials was observed as the number of lithium atoms incorporated increased, going from pale yellow for the pristine phase to blue developing to Fig. 9 Evolution of cell voltage vs. composition (x in LixTa8W9O47) black for phase II which corresponds to an increase in the obtained during a SPECS experiment run at±10 mV (2 h)-1 potential steps showing a complete charge–discharge cycle.electronic conductivity of Nb8W9O47 as the insertion reaction 2780 J. Mater. Chem., 1998, 8, 2777–2781(11<x<12) and between 1.40 and 1.50 V probably indicate the existence of a tendency to local ordering in phase II. These results are supported by X-ray diVraction data of inserted materials which show important changes on the transition from phase I to II with additional variations as the number of lithium atoms inserted increased.In order to accommodate such a number of lithium atoms and taking into account the number of tunnels available per unit cell, multiple lithium occupancy has to be considered in some cavities. Electrochemical lithium insertion was also studied in Ta8W9O47 where the reaction seems to follow a diVerent mechanism confirming that the second transition metal in these TTB type structures (Nb and Ta) plays an important role in the insertion reaction.Lithium insertion reversibility in Ta8W9O47 was observed for a smaller number of atoms than its niobium analogue (Li/SM=0.9). Acknowledgements Fig. 11 X-Ray powder diVraction patterns of (a) Nb8W9O47, (b) Li1.8Nb8W9O47, (c) Li7.3Nb8W9O47 and (d) Li18.8Nb8W9O47. Financial support from CONACYT (Project 3862P-A9607) is gratefully acknowledged. The authors are also especially grate- Table 1 Cell parameters and cell volume values obtained from X-ray ful to Dr.Yves Chabre for useful discussions and his critical powder diVraction data for LixNb8W9O47 remarks on the original manuscript.Compound a/A° b/A° c/A° Cell volume/A° 3 References W9Nb8O47 a 36.692 12.191 3.945 1764.65 1 A. F. Fuentes, A. Martý�nez de la Cruz and L. M. Torres-Martinez, W9Nb8O47 36.675(8) 12.186(2) 3.9447(8) 1763.0±0.7 Solid State Ionics, 1996, 92, 103. Li1.8Nb8W9O47 36.71(1) 12.201(4) 3.942(1) 1766.1±0.9 2 A. F. Fuentes, E. Briones Garza, A. Martý�nez de la Cruz and Li7.3Nb8W9O47 36.9(1) 12.29(3) 3.93(1) 1787.8±9.0 L.M. Torres-Martý�nez, Solid State Ionics, 1997, 93, 245. Li18.8Nb8W9O47 36.9(1) 12.28(5) 3.94(1) 1790.0±12.7 3 A. F. Fuentes, A. Martý�nez de la Cruz and L. M. Torres-Martinez, aRef. 5. Mater. Res. Soc. Symp. Proc., 1997, 453, 659. 4 M. Lundberg, Chem. Commun. Univ. Stockholm, 1971, No. XII. 5 R. S. Roth and J. L. Waring, J. Res. Nat. Bur.Stand., Sect. A, 1966, 70, 281. proceeds. Powder patterns obtained after Li removal were 6 A. W. Sleight, Acta Chem. Scand., 1966, 20, 1102. very similar to those of the pristine phase confirming the 7 D. C. Craig and N. C. Stephenson, Acta Crystallogr., Sect. B, reversibility of this reaction. 1969, 25, 2071. 8 F. Krumeich and T. Geipel, J. Solid State Chem., 1996, 124, 58. 9 A. Martý�nez de la Cruz, F. Garcý�a-Alvarado, E. Mora�n, Conclusions M. A. Alario-Franco and L. M. Torres-Martý�nez, J. Mater. Chem., 1995, 5, 513. A study of lithium insertion in M8W9O47 (M=Nb and Ta), a 10 C. Mouget and Y. Chabre, Multichannel Potentiostatic and tetragonal tungsten bronze type phase (TTB) has been carried Galvanostatic System MacPile, Licensed from CNRS and UJF out. The electrochemical insertion of up to 20 lithium atoms Grenoble to Bio-Logic Corp., 1 Av. de l’Europe, F-38640, Claix, France. per formula unit of Nb8W9O47 (Li/SM=1.2) is a reversible 11 J. M. Tarascon, J. Electrochem. Soc., 1985, 132, 2089. reaction which proceeds through three reduction steps: a first 12 Y. Chabre, J. Electrochem. Soc., 1991, 138, 329. order transition at around 2.12 V and two continuous trans- 13 Y. Chabre and J. Pannetier, Prog. Solid State Chem., 1995, 23, 1. formations at around 1.70 and between 1.40 and 1.50 V vs. 14 C. J. Wen, B. A. Boukamp, R. A. Huggins and W. J. Weppner, Li+/Li. These transformations originated at least two solid J. Electrochem. Soc., 1979, 126, 2258. solution regions of general formula LixNb8W9O47 with the 15 R. D. Shannon, Acta Crystallogr., Sect. A, 1976, 32, 751. following approximate composition limits: I: 0x2; II: 6x20. Two incremental capacity peaks observed at 1.70 V Paper 8/04410D J. Mater. Chem., 1998, 8, 2777