A NEW METHOD FOR THE PREPARATION OF FINE-GRAINED SnO2 AND WO3 POWDERS: INFLUENCE OF THE CRYSTALLITE SIZE ON THE ELECTROCHEMICAL INSERTION OF Li + IN SnO2 AND WO3 ELECTRODES

We propose an unconventional method to obtain fine-grained SnO 2 and WO 3 powders. It uses as precursors, polymer complexes between polyethylene oxide (POE) and SnCI 4 or WCI 6 respectively. By pyrolysis of these complexes in the 350-5500C temperature range, metal-oxide powders possessing small crystallite sizes are obtained. They are free from water and hydroxyl group contaminations, which is an added advantage where the application of these materials to Li-batteries is concerned. We have, indeed, demonstrated that these powders show good ability to insert reversibly lithium ions in the Li / Li + / SnO (WO 3) cells.


INTRODUCTION
SnO 2 is an n-type semiconductor with a tetragonal rutile structure and a large indirect band energy gap (2.7 eV)2. It has attracted considerable attention to the variety of applications related to its unique electrical, optical and catalytic properties. Among its applications, to mention only a few, are in transparent heaters for windshield defrosting, in anti-reflection coating for solar cells, as a transparent electrode for electrochromic devices, as a sensing material for combustible gas sensors, and as an electrocatalyst for organic oxidation3-8. In the last two applications mentioned above, fine powders of SnO 2 are used. These powders have generally been obtained through two different methods: one involves the oxidation of elemental tin with acids (HNO3, HESO4 etc.) and the other utilizes the wellknown sol-gel route. The sol-gel method involves the dispersion of stannic hydrate in aqueous ammonia to give a sol. The sol is later 'solidified' through stages of stiffening and polymerization to give a gel (gelation). The gel so obtained is *Author for correspondence *Present address: Institut de Chemic de la Matib.ra Condens6e de Bordeaux, CNRS, Universit6 Bordeaux I, Chateau Brivazac, Ave du Dr. A. Schweitzer, 33600 Pessac, France. 40 S. HAN et al.
thoroughly washed with distilled water, filtered, dried and finally heated to high temperatures to obtain the required material9. The sol-gel route has been found to give finer crystallites compared to the former method. However, samples free from water and hydroxyl groups are hardly achievable when low calcination temperatures are used. WO 3 is an n-type semiconductor that displays three crystal polytypes: monoclinic, hexagonal and orthorhombic .WO 3 powders can be obtained using the same preparation procedure used for SnO 2. They have also been prepared using other techniques. Cheng et al. 1'12 obtained WO 3 powders from the thermal decomposition of ammonium paratungstate while Zhong et al. 3 used a similar process with H eWO4. These authors were primarily concerned with making cathode electrodes for secondary lithium batteries where high surface area is an important parameter. The specific surface areas of the WO 3 powders obtained from the thermal decomposition of H 2WO4 were between 12.2 and 4.3 me/g13, and so were of limited use for this purpose. Low specific surface areas are obtained not only for the WO3 powders but also for the SnO 2 powders obtained using the preparation procedures quoted above. There is, therefore, a need to develop a method to produce WO 3 and SnO 2 powders with higher surface areas for use as cathode materials in lithium electrochemical cells. For this purpose the powders will also have to be free of water and hydroxyl groups.
We propose here an unconventional and easy-to-carry-out method for the preparation of fine particles of SnO 2 and WO3 powders. It uses as precursors complexes of the respective halides SnC14 and WCI 6 with polyethylene oxide (PEO)14. The method has been shown to give not only very fine particles, obtained after eliminating the polymer by heating, but also particles free from water and OH group impurities.
The conventional and unconventional sol-gel methods will be compared here; they will be referred to as method 1 (for the conventional one) and method 2. We also investigated the ability to insert reversibly lithium ions into the fine-grained SnO 2 and WO 3 powders in Li/Li + / SnO 2 (WO 3) electrochemical cells. Related to that, we have reported elsewhere that electrodes based on fine-grained transitions metal oxides (LixFe2Oa, Li2_xNiO 2 etc.) in the form of thin films exhibit highly efficient electrochemical (de)insertion of Li + ions into 15 17 lithium conducting electrolytes The electrode materials were symbolized as NCIMs (nanocrystallite-insertion-materials). Indeed, by minimizing the crystallite size, we favor the formation of dangling and weak bonds at the surface and, thereby, the rate of reversible lithium insertion. Since the previously studied NCIMs were based on transition metal oxides, we were more concerned in earlier publications in the cationic "d" orbitals that were involved in the electrochemical 15 17 processes In the case of SnO 2 based electrodes, it is the cationic "5s" orbitals that would be of concern; as for WO3, the "5d" orbitals are involved. Therefore, it is worthwhile to have an insight into the respective influences of "s" and "d" orbitals on the electrochemical response using SnO 2 and WO 3   The preparation procedure involved in methods 1 and 2 are schematically illustrated in Fig. 1. Let us note that the time required for the synthesis of the powders using the method 2 is much shorter (about 2/3 shorter) than that required for method 1: indeed, the filtering and milling processes intervene only in method 1.

Sample analysis:
Thermogravimetric analyses, using a Leco TGA-501 display, were performed in dry air between 22C and 600C with a Setaram thermobalance. X-ray diffraction measurements were obtained with a Phillips PW 1050 spectrometer and CuK,, Jl/2 is the corrected width of the main diffraction peak at half height, A is the X-ray wavelength, and 0 the diffraction angle. The specific surface area of the samples was measured using the single point Brunauer Emnett and Teller (BET) method with a Micromeretics Accu Sorb 2100 E. The samples were outgassed at 180C for 5 hrs. The adsorbate gas was nitrogen. The IR spectra in the adsorbance mode were recorded on a Perkin-Elmer 983G spectrometer between 4000 and 200 cm-1 with an average resolution of 5 cm-1. The experiments were performed on SnO 2 and WO 3 powders dispersed in Nujol and sandwiched between two cesium iodide disks. Conductivity experiments using the Van der Pauw four probe technique were carried on using samples that were pressed at 5 tons/cm 2 in a steel die of diameter 13 mm. The analysis of the carbon content and the composition of the gases released during the sample pyrolysis were achieved using the TGA RECHARGEABLE LITHIUM BAITERIES 43 apparatus equipped either with an elemental analyser (LECO, CHN-1000) or with a gas chromatograph (Varian 3700).
Electrochemical measurements: Electrochemical (de)insertion of lithium was realized in bottle-type cells having two electrodes. The cathode consisted of 35 mg of SnO or WO3 powder and 6 mg of carbon black (i.e. 15w%). The cathode was prepared by intimately mixing the powders (previously outgassed at 180C for 5 hrs) and pressing in the form of a disk in a 1.3 cm diameter stainless steel die. Lithium metal was used as both reference and aaode. The electrolyte was a 1M LiCF3SO3-propylene carbonate solution impregnated into a glass filter paper. The propylene carbonate (Aldrich 99 + %) was further dried by fractional distillation under 4 molecular sieves.
The lithium triflate (Aldrich 97%)was kept under vacuum at 150C during 72 hrs. The experiments were then carried out inside a glove box maintained under argon atmosphere and containing less than 1 ppm H20 and O 2. The presence of adsorbed water hydroxyl groups in samples 1-(150) and 1-(350) is also revealed in the IR spectra ( fig. 2). Adsorbed water is best characterized by its deformation mode t$OH 2 occurring near 1620 cm-1 18,19. This adsorption band is only observed in sample 1-(150). The broad features between 3500 and 3100 cmobserved for samples 1-(150) and 1-(350) can involve stretching modes of water, hydroxyl groups. Traces of ammonium ions arising from the synthesis might also occur and contribute to the previous broad adsorption (, and v 2 modes of NH)2'21. However, nothing is detected around 1420 cm -(v 4 mode of NH4+).

RESULTS AND DISCUSSION
On the other hand, a weak and broad feature is observed at about 1200 cm-1 in both 1-(150) and 1-(350) samples. It can be attributed to a hydroxyl bending mode. It can be attributed to a hydroxyl bending mode. Therefore, for sample 1-(350), which presents no water adsorption at 1620 cm -, it can be concluded that the protons are essentially involved in hydroxyl groups and the broad high frequency adsorption reflects mainly the stretching vibrations of these hydroxyl groups.
Finally, sample 1-(550) exhibits neither water nor hydroxyl group vibrations. It is known that the presence of hydroxyl groups and adsorbed water in fine-grained SnO 2 powders favors protonic conductivity 22'23. The latter can be evidenced from d.c. conductivity measurements, as shown below. First of all, the d.c. electric induced by an applied potential of 1 V across a disk of SnO 2 (1 mm thickness, 13 mm diameter) is shown in fig. 3 gold electrodes that are blocking to ionic motion insure the electrical contacts. The term "blocking to ionic motion" means the absence of any ionic transfer via the interfaces between the electrodes and the electrolyte. With the blocking electrodes, the protonic conductivity causes the "polarization effect" illustrated in fig. 3 for samples 1-(150) and 1-(350): it can be noted that the current measured immediately after the potential has been applied is time dependent, accounting for an ionic (i.e., protonic) conductivity. The large concentration of the mobile enables, indeed, the progressive accumulation of a large number of charges within a thin thickness at the interfaces, causing the observed decrease of the current between 0 and 150 s (fig. 3). We are dealing, in fact, with the well known phenomena of the double electrical layer that which is produced at the blocked interface between an electronic conducting electrode and an ionic conductor. When t > 150s, the interfaces are built and the observed residual steady current ( fig. 3) account for a residual electronic conductivity. The occurrence of protonic conductivity in samples heated only up to 350C has also been observed by others25-26. The constant current, observed uniquely for sample 1-(550)within    (table 1), therefore IR spectroscopy measurements cannot be carried out. However TGA measurements reveal no measurable traces of water and hydroxyl groups in all samples. Consequently, the SnO 2 and WO 3 powders are likely to behave as the (quasi) water free and hydroxyl group free NCIMs whose electrochemical Li / (de)insertion efficiency has already been demonstrated by some of us5-7. To prove the occurrence of the similarity in the behavior of the SnO 2 and WO 3 powders to those of the NCIMs, we have to examine first the crystalline structure and texture prior to any electrochemical investigation.  (a) this study: let us quote that the surface areas reported here are related to the particle size and pore volume (the accepted definition being that many crystallites make a particle). RECHARGEABLE LITHIUM BATI'ERIES 49 and particle growth. The crystallite and particle growths are, indeed promoted only when the organic part is almost eliminated. As expected, the specific area increases as the crystallite size is reduced, from 2-(550) to 2-(450) on the one hand, and from 2-(550)-W to 2-(450)-W, on the other hand (table 2). Finally, the low surface area values reported in table 2 for samples 2-(350) and 2-(350)-W are due to the incomplete elimination of the polymer-residuals at 350C; the carbon content (measured using atomic adsorption spectroscopy) is indeed as high as 15% (table 1). It may be possible that the SnO 2 and WO3 crystallites are bound together through O-C bonds; consequently very small surface areas are observed for 2-(350) and 2-(350)-W.
Electrochemistry: influence of grain and particle sizes on the reversible electrochemical insertions of lithium in SnO z and WO 3 NCIM electrodes: As pointed out above in the experimental part, the obtained samples were intercalated with lithium in the following electrochemical cells: Li/LiCF3SO3 in propylene carbonate/ShOe and Li/LiCF3SO 3 in propylene carbonate /WO 3.
The measurements were carried out at room temperature. The lithium insertion process was conducted, using the well known cyclic manner, by (dis)charging the cells with a constant current of 50/A/cm 2. Fig. 6 illustrates the reversible 10 th discharge curves of the cells for the samples free from contamination by water, hydroxyl groups, and residues of PEO (table 1). In agreement with our model15-17, the highest rate of lithium insertion occurs for the NCIMs 2-(450) and 2-(450)-W, which possess the highest specific surface area.
Our results concerning WO 3 are also in argreement with Zhang 31 who showed that powdery LiWO3 electrodes having surface areas of 4.3me/g and 12.2me/g reversibly insert x 0.28 and x 0.48 lithium, respectively, between 1.8 and 3.2 V vs Li. In our work, the rate of lithium reversibly inserted is higher (x 1.2 for 2-(450)-W) because of the higher specific surface area. Within the same voltage range ( fig. 6), the amount of Li inserted is larger in the case of WO 3 compared to SnO2: it is indeed related to the electron affinity of the W6+/W 5+ couple, which is larger (in absolute value) than that of the Sn 4+/Sn 2+ couple. Fig. 7 illustrates the good cycling reversibility observed for 2-(450) and 2-(450)-W electrodes.
x in LixMOy

CONCLUSIONS
In this paper we have distinguished two methods for preparing fine-grained tin oxide powders. The method 2, to our knowledge, seems to be original and leads to SnO 2 powders free from water and/or hydroxyl groups and with small crystallite and particle sizes. Method 2 also has been used to prepare powdery WO 3 samples having the finest crystallites. Related to the fine-grained texture of the samples, we have shown that they are able to sustain long-term electrochemical cyclability. In fact, they behave like other NCIMs that we have recently investigated17. Indeed, by minimizing the size of the crystallites, the formation of defect bonds is favored, particularly at the crystallite surface, acting as reversible grafting sites for Li +.
Moreover, the cation-anion bonding would be weakened not only in the grain boundary region, but also within a grain close to its surface. Therefore, the electrochemical insertion of Li + would also occur through an easy bonding rearrangement.
Let us point out that for WO 3 these "non-conventional" insertion mechanisms should occur in addition to the well known intercalation mechanisms of Li within the crystallites and allowed by their tunnel type structure. However, in order to check the validity of the discussions and conclusions made here, further investigations of in situ IR and Raman spectroscopy and of the evolution of the electrodeequilibrium potentials as a function of the Li insertion rate are necessary.