Lithium incorporation in tungsten oxides

Lithium incorporation in tungsten oxides

Solid State Ionics 1 (1980) 151-161 © North-HoUand Publishing Company LITHIUM INCORPORATION IN TUNGSTEN OXIDES Kent H. CHENG and M. Stanley WHITTINGH...

476KB Sizes 0 Downloads 24 Views

Solid State Ionics 1 (1980) 151-161 © North-HoUand Publishing Company

LITHIUM INCORPORATION IN TUNGSTEN OXIDES Kent H. CHENG and M. Stanley WHITTINGHAM Corporate Research, Exxon Research and Engineering Company, Linden, NJ 07036, USA Received 30 January 1980

The incorporation, both electrochemically and chemically, of lithium at ambient temperatures into WO 3 has been studied. The phases formed, together with their lattice parameters, are compared to the known high temperature LixWO 3 phases as well as to amorphous WO 3. The cubic phase was found to be extended to higher lithium contents, 0.67, at 25°C than at 650°C, 0.5; this is approaching the value 0.75 calculated from the crystal structure. In addition, a number of other tungsten oxides have been studied for their capability of incorporating lithium. A strong interrelationship of crystal structure and degree of lithium incorporation was also evident in these oxides.

INTRODUCTION Although the alkali metal tungsten bronzes, of general formula MxWO 3 wher~ M is a metal and x < i, have been known for many years (1,2), little work has been reported on their synthesis at ambient temperatures. The structures of these bronzes and many other oxides contain partially filled tunnels, whose occupancy can vary over wide limits without significant change of the host crystalline lattice. Such minimal changes in the structure often allows for the chemically reversible incorporation of the M cations into the structure (3) and in some cases for the application of this class of compound in ambier temperature secondary batteries (4). WO 3 itself has been described as a potentially useful electrochromic electrode (5) when protons are the incorporated cation. In this paper we discuss the chemical and electrochemical intercalation of lithium at ambient temperature into various tungsten oxide host lattices with emphasis on the simple perovskite structure. The phases formed are compared with those obtained using the normal high temperature synthetic techniques. It will be shown that in the perovskite case a much higher lithium content can be obtained at 25°C than at 650°C. EXPERIMENTAL The WO 3 was prepared by heating reagent grade ammonium paratungstate (Alf~ Products) in the air at 500°C for 14 hrs; n-butyl lithium in hexane solution (1.617 M) was used as supplied by Alfa Products; LiCI04, dioxolane solution (2.0 M) was used as the electrolyte in the electrochemical experiments, and for sample homogeneity equilibration.

K.H. Cheng, M.S. lChittingham / Lithium incorporation in tungsten oxides

152

5.210

3,840

{..+o+

o~

co

o

~x 5.200 .:.5 0

{f"~}'~l

._x 3.830 .C

._1

,{

0

(a) 5.190

,

0.2

,

0.3

..

I

0.4

°°

(b)

3.820

X in LixWO3

i

i

0.2

0.3

1 0.4

X in LixWO3

t 7.460 ¢o

o ×

.L.3 °~ 0

0

7.440

(c) I

0.2

I

0.3

I

0.4

I

0.5

I

0.6

0.7

X in LIxWO3 Fig. i Cell dimensions of tetragonal, (a) and (b), and cubic (c) phases in the mixture of overall lithium content x of LixW03, before equilibration - - -, after e q u i l i b r a t i o n - - . The composition, x, represents the overall lithium content of the ~etragonal plus the cubic phase and is therefore not necessarily an accurate value for the individual phase. Thus the true tetragonal phase region is much narrower than indicated in this figure.

K.H. Cheng, M.S. IChittingham /Lithium incorporation in tungsten oxides

153

The maximum lithium content was obtained by adding an excess of n-butyl lithium hexane solution (4.00 ml, 6.468 mmol) to powdered WO 3 (0.5041 gm, 2.174 mmol) at room temperature under helium in a vacuum atmospheres drybox. The color of the solid changed from yellow to light blue to dark blue to brown. ' Reaction seems to occur rapidly; however, the mixture was allowed to stand overnight in a dry box to insure completion of reaction. The product was filtered and washed with hexane. The amount of unreacted n-butyl lithium was determined, after combining the filtrate and washings, by back titration (6), (5.01 mmol). The composition was Li.67WO 3. Other tungsten oxides were reacted and analyzed in a similar manner. Compounds with lower lithium contents were prepared by using the calculated amount of n-buty lithium, and checking for completion of reaction by titrating the liquid phase as above; no unreacted n-butyl lithium was found for x ! .60. Analysis of lithium content by atomic absorption was also used to confirm the compositions. The samples were then immersed in a LiClO4/dioxolane solution (2.0 M) for 1 w e e k for equilibration to remove any inhomogeneity in lithium content among the LixWO 3 particles because of the non-reversibility of the synthesis reaction: WO 3 + xC4H9Li

>

LixWO 3 + C8H18

Powder x-ray data was recorded on a Philips diffractometer using CuK~ radiation both before and after equilibration. Samples were mounted under dry nitrogen between Sellotape and Kaptan foil to avoid exposure to moisture. ~ quartz was used as an internal standard. After indexing, the lattice parameters were refined using an iterative least squares procedure. The open circuit voltage (o.c.v.) of the two electrode cell Li(s)|2.0 M LiCI04, dioxolaneILixWO3(s) was monitored as a function of x. The lithium anode was sufficiently large to preclude the possibility of any polarization at that electrode; polarization due to the poor conductivity of the pure W O 3 was minimized by ball milling with 40 wt.% graphite prior to construction of the cathode following our standard hot pressing techniques (4). The value of x was changed galvanostatically by stepping the cell voltage in 20 mV increments; the current flow was allowed to decay to 33 ~ A between each step leading to a voltage composition relationship which approximates equilibrium (thermodynamic) values for fast kinetics. This stepping procedure was performed from the initial open circuit value of 2.71 volts down to 1.25 volts, b e l o w w h i c h electrolyte side reactions can complicate the data. This follows the procedure described earlier (7). RESULTS AND DISCUSSION Samples of LixWO 3 over the range 0 < x < 0.67 at intervals of Ax = 0.I were synthesized using n-butyl lithium. The highest lithium content was found to be 0.67 using excess butyl lithium; atomic absorption analysis indicated an x value slightly higher, 0.75. As another check on the value of x, the magnetic susceptibility was measured at 298OK and found to be x e = 19.3 x 10 -6 . This compares with a value of 20-24 x 10 -6 found for hydrogen and sodium tungsten bronzes in the composition range 0.6 to 0.75 (8); this suggests that the true value x is closer to 0.67 than to 0.75. The color varied from light blue at x = 0.i, blue at x = 0.2 and 0.3, dark blue at 0.4 and 0.5, to gold-brown at 0.6 to 0.7. The measured lattice parameters, before and after equilibration in the electrolyte solution are plotted as a function of the lithium content in Fig. i That a marked change is observed between the two sets suggests that the as-reacted samples exhibit a rather broad range of inhomogeneity for a given overall x-value. In the cubic phase, equilibration results in an apparent decrease in the slope of a o vs x line for x > 0.4. This is due to the presence of some low lithium content particles in the unequilibrated mixture

K.H. Cheng, M.S. Wl~ittingham /Lithium incorporation in tungsten oxides

154

which moves the center of gravity of the diffraction peaks disproportionately to lower 2g values. The elimination of such particles upon equilibration then leads to the observed narrower range in cell dimensions. Below x ~0.4 a two phase region is observed similar to that previously reported for the high temperature preparations (9). However the homogeneity ranges of these phases changes toward higher x-values as the temperature is decreased. This is shown in Fig. 2. Thus the phase boundaries of the cubic phase change to .4 < x < .67 at 25°C from .27 < x < .50 at 650°C. On annealing the 0.67 composition at 650°C, the sample disproportionates to Li2WO4, lower oxides of tungsten, and a cubic bronze of presumed lower lithium content (as Fig. i shows, the lattice parameter is not a sufficiently strong function of x to allow the precise determination of x from ao). In addition, when a mixed cubic/tetragonal compound of overall lithiu~ content 0.32 was annealed, a compound with a single cubic phase was formed consistent with the expected phase behavior. In agreement with the trends of high temperature data (9), the width of the tetragonal phase is very narrow at 25°C; in these studies this phase was not obtained in a pure state which places an upper limit of A x < 0.1. The lattice parameters of the cubic phase obtained in this work are compared with the earlier high temperature work (i0,ii) in Fig. 3. The agreement is good except at low x-values, where the data reported here have too low an x-value due to incomplete equilibration in the two-phase region.

650°C (HAGENMULLER, et al (9)) p CUBIC ! [ TET TETRAGONAL k.

CUBIC

~ii ,E, i

k

. . . . . .

Fig. 2 Comparison of phase domains at 25°C and 650°C; approximate phase boundaries are drawn with dashed lines. M, monoclinic; O, orthorhombic; TET, tetragonal.

i

ROOM TEMP. {THIS WORK) 0.1

0.2

0.3 0.4 XlnLixWO 3

0.5

o< Fig. 3 Comparison of the data obtained at 25°C, ~ , with that at higher temperatures,

._X ...I

--

0

(10),

~

(11).

0.6

0.7

7.460

.E 7,440 i 0.2

I 0.3

I 0.4

I 0.5

X in LixWO3

I

0.6

0.7

K.H. Cheng, M.S. Whittingham /Lithium incorporation in tungsten oxides

155

The thermodynamics of incorporation of lithium into tungsten oxide were determined in an electrochemical cell. Samples of LixWO 3 prepared using n-butyl lithium were equilibrated in LiClO4/dioxolane solutions, and then their potentials were measured relative to a pure lithium anode. The results, calculated as earlier (12) for NaxW03, are given in Table i and compared with the corresponding values for crystalline and amorphous NaxWO 3 and HxWO 3 in Fig. 4. As expected the free energies of formation of the lithium compounds are slightly higher than for the corresponding sodium materials. Just as in the case of sodium the change in free energy with composition or phase is very small so that there is little driving force for conversion of one phase to another; that is, the true equilibrium phases will not be easily obtained just as noted in the earlier x-ray discussions. The data on the amorphous compositions will be discussed later.

Table i Thermodynamic Properties of LixWO 3 x

E, Volts

•i .2 .3 .4 .5 .6 .67

-AGLi , kJ/mole

2.806 2.711 2.674 2.640 2.611 2.616 2.470

-AG, kJ/mole

270.8 261.6 258.0 254.7 252.0 252.4 238.4

270.8 262.0 258.9 256.3 254.5 255.5 247.4

3.° ,

,

,

,

,

axWO 150 0 ~ =Ex ~' ZOO 0 z o

1.o 0 "~I

50

z, - - 4 - - ~ ' v . " . " . ~ , , , .1 -0.

.2

.4 .6 X in MxWO3 a)

.8

.0

.2

.3

,, - ' v - . - - ~ , .4

.5

.6

.7

, .8

X in MxWO3

b)

Fig. 4 a) EMF values of crystalline and amorphous bronzes, eLi (this work), cNa (12), aLl (13), a'Li (14), aNa (13), aH (15), where c is crystalline and a is amorphous; and b) free energy of formation per mole of bronze.

156

K.H. Cheng, M.S. Whittingham /Lithium incorporation in tungsten oxides

When the lithium incorporation is carried out in an electrochemical cell the cell emfs are significantly less than the equilibrium values, as shown in Fig. 5. This is true both on lithium insertion and subsequent removal and suggests that the actual lithium composition in the solid phase at the electrolyte/conductor interphase is much higher than the overall value. Equilibration of high x value electrodes even for a week show little change in the emf. This presumably is a result of a very low lithium diffusion coefficien~ at these high lithium concentrations,

2.75

vt'vv

o > v

, •

~i~AIA~

z

0

UJ

1.25

0.0

,

,

0.2

0.4

i

0.6

X in L i x W O 3

Fig. 5 Electrochemical incorporation (w) and removal (A) of Li in crystalline LixWO^j gavanostatically stepped at Imi n-- 33uA. EMF values of the equilibrated bronzes are also shown (e).

x > 0.67. X-ray studies of these electrodes, however, showed no evidence for phases other than those found in Fig. i. This high polarization has been observed previously; thus, LixVS 2 can be readily prepared for 0 < x < I using n-butyl lithium, but in electrochemical studies the maximum x-values obtained were around 0.4 (16). In this case, the behavior may be associated with distortions in the structure which inhibit further lithium uptake; presumably, during the chemical reduction by n-butyl lithium the uptake is much more rapid so that these distortions do not occur. In contrast, the chemical oxidation of the bronze, Li.67W03, by reaction with iodine, aeetonltrile solution is rather slow. Complete oxidation back to the monoclinic WO 3 occurs in about 12 hours. In order to better understand the polarization behavior, a series of tungsten trioxides were studied. These differed from one another in the degree of erystallinlty, in the addition of carbon conductive diluents, and in preparation technique. In Fig. 6 curves A and B used commercially

K.H. Cheng, M.S. Whittingham /Lithium incorporation in tungsten oxides

2.75 ,o.,'

'.



157



o

._1

Z 0

1.75

~.2 o

°%~

u-

,,

0 1.25

,

0.0

,

0.2

0.4

.

L

0.6

X in LixWO 3 Fig. 6 Step discharge curves for WO 3 electrodes. A (O), 20% C, Imin = 33~A; B (v), 20% C, Imln = 99~A; C (@), 40% C, Imin=33~A; D (A), 20% C, Imin = 33~A; D (I), 40% C, constant I at .i ma/cm 2.

available WO 3 annealed at 600°C in air and of high crystallinity, whereas the other three curves used WO 3 formed by the thermal decomposition of ammonium paratungstate in air at 500°C of slightly lower crystallinity as shown by the broader diffraction peaks. However, the latter had the larger particles as measured by the BET surface area, 0.97 m2/gm compared to 3.3 m2/gm. Clearly the more crystalline material had a significantly lower lithium uptake which is suggestive of a low lithium diffusion coefficient, which may be associated with a lower concentration of lattice defects. However, this is not the only factor as the electrode, B, with the higher current cut-off in the galvanostatically stepped discharges had the slightly higher capacity. Moreover, the highest capacity of all was in the electrode intercalated at a continuous rate of 0.i ma/cm 2. These results suggest that the final product may be obtained by at least two different mechanisms. The slow discharge of the cathode allows the crystalline structure to approach closer to an equilibrium state where the lithium ion mobility is restricted more than in a non-equilibrium condition where the lithium ions may be thought of as being in more of a random configuration. By optimizing crystallinity and rate of lithium uptake complete lithiation could presumably take place; however, this sensitivity of uptake on rate of reaction combined with the low diffusion coefficient precludes the effective use of WO 3 in energy storage systems (17). In the mechanism of reaction discussed above, one would expect less polarization in a disordered form of WO 3 than in the crystalline WO 3. Amorphous WO 3 has recently been described (18) as being built-up of a disordered network of cornersharing octahedra containing a wide range of ring sizes not only the four membered rings found in crystalline WO 3. A study of an amorphous WO 3 was therefore made. This material was prepared by the decomposition of ammonium paratungstate at 350oc for one hour. As shown in

158

K.H. Cheng, M.S, lChittingham / Lithium incorporation in tungsten oxides

Fig. 7, a larger lithium uptake is indeed found above a voltage cut-off of 1.40 volts; the lower overall emf, relative to the crystalline material, is in part due probably to the less ideal coordination around the lithium. The lithium here was not removable electrochemically in contrast to the crystalline case. Also as shown in Fig. 4 the "equilibrium" emf values are lower for the amorphous compounds. It has been proposed (14) that in contrast to the crystalline materials, where the interstitual sites may be well defined structurally and in energetics, the amorphous materials present a large spread in the site energies; only a small fraction of these sites have energies comparable to those in the crystalline case. No definitive evidence is however available for this hypothesis.

2.75

-6

2.25'

"-





mm • 0

~ qn ~ moral U •

1.75

1 .25 0.0

• •



,

I 0.2

,

I 0.4



,



n 0.6

X in L i x W O 3

Fig. 7 Electrochemical incorporation of Li in amorphous WO 3 at constant current. • cathode is 40% carbon, 10% Teflon, I= .i ma/cm2; • cathode is 0% carbon, 10% Teflon, I= .Sma/cm 2.

Another feature of the curves in Fig. 6 are the breaks that occur around and greater than x = 0.2. These suggest that the lithium uptake in the non-cubic phase is easier than in the cubic phase; it is in the region of the cubic phase that the greatest polarization is noted. The structure of the cubic phase, formed at high temperatures, has been determined using neutron diffraction (ii). It has the basic perovskite structure but with the W• 6 octahedra tilted about each of the cubic axes giving a number of differently coordinated sites for the lithium. The degree of tilt has been shown to be a function of the size of the ion incorporated into the W• 3 matrix. Presumably the lack of any apparent change in the lattice parameters of a o = 7.446 ~ for x > 0.4 in LixWD 3 whilst the lithium content and electron concentration (or tungsten oxidation state) change markedly must be related to a balancing change in the tilting of the octahedra. Such an increase tilt of the octahedra could make for even more disparate site environments for the lithium, which in turn could both increase the degree of ordering of the

K.H. Cheng, M.S. Whittingham /Lithium incorporation in tungsten oxides

159

lithium and decrease the ionic diffusion as x increases. The preferred sites for lithium in the WO 3 matrix number 0.75 per tungsten (ii), which would place this as the upper composition limit of the cubic phase. In this ambient temperature study the upper limit, 0.67, was close to this value. In an effort to push the lithium content still higher, WO 3 was refluxed with n-butyl lithium. The product, however, was non-cubic and of substantially higher lithium content, x & 2.5. The x-ray powder pattern has not been indexed, but is compared with those of WO 3 and Li0.67WO 3 in Fig. 8. This compound may be analogous to the compound Li2.3ReO 3 which has a orthorhombic structure (19).

i~ ' ~

60

50

I

I

40

30

l

20

2e

Fig. 8 X-ray powder diffraction patterns of W03, cubic Li.67W03, and Li2.5WO 3 of unknown cell dimensions.

Structure is clearly an important factor in the reaction of lithium with the tungsten oxide matrix. This study of pure WO 3 has been extended to a number of other tungsten oxide structures (2), and the results are given in Table 2.

160

K.H. Cheng, M.S. Whittingham / Lithium incorporation in tungsten oxides

Table 2 Lithium Uptake in Tungsten Oxides Compound WO 3 Na0.7WO 3 Nao.4WO 3 K0.33WO 3 (NH4)O.33WO 3 W18049

Structure

Li/W "(n-Butyl Li)

Li/W Electrochem.

Perovskite Perovskite Tetrag. Tunnel Hexag. Tunnel Hexag. Tunnel Tunnel

0.67 0 0.6 0.67 2.0 1.44

0.57 0 0.5 2.0 0.9

For Na0.7W03, which has the perovskite structure, no reaction with lithium was observed. This is readily accounted for by the very low diffusion coefficient of the sodium ions, 10 -15 cm2/sec (20), which will essentially prevent the remaining empty sites from being reached by the lithium ions. However, when the sodium content is reduced, a tetragonal tunnel structure is formed which contains vacant sites in triangular, square and pentagonal tunnels. These sites are interconnected so that the lithium ions may diffuse around the relatively immobile sodium ions and occupy all the empty sites to give Li0.6Nao.4WO 3. In the hexagonal tungsten bronze structure of KO.33WO 3 there are two major sets of sites, those associated with the hexagonal tunnels and those in the triangular tunnels. One ion in each set of these sites would lead to the composition Lio.67Ko.33WO3 with all the lithium in the smaller site; this is the same as that observed experimentally. Such a model is in agreement with a high temperature study of KxLiyWO 3 (21). In contrast to the isostructural KxWO~ , (NH4)xWO 3 reacts with two lithium so that some of the lithium must reside along the walls of the hexagonal tunnels. The ammonium is believed to be reduced to NH 3 and possibly even to amide, and the lithium ions in the large channels can then be solvated by ammonia or residual water molecules. There is ample volume in this structure for two lithium per tungsten when there are no larger positively charge species in the tunnels. In both these hexagonal structures only 0.67 Li can be deintercalated electrochemically, suggesting that the lithium ions in the large tunnels are essentially immobile. Just as in the ammonium case, the tunnel compound W18049 both reacts to give a formal oxidation state of W 4+ and must have multiple occupancy of the large tunnel sites. More detailed crystal and electrochemical studies of these and other tungsten oxides will be discussed elsewhere (22). CONCLUSIONS Lithium can be incorporated into tungsten trioxide at ambient temperature either electrochemically or chemically giving a significantly higher lithium content than has been reported by the previously used high temperture techniques. These reactions could be reversed either chemically or electrochemically. However, the electrochemical reaction was not so readily accomplished due to polar~izatlon related to the formation of intermediate surface phases which inhibited lithium diffusion. Lithium incorporation in other tungsten oxides was also found to be a strong function of the crystal structure.

K.H. Cheng, M.S. Whittingham /Lithium incorporation in tungsten oxides

161

ACKNOWLEDGMENTS We wish to thank A. J. Jacobson for numerous helpful discussions and his kind encouragement, A. H. Thompson' for the electrochemical instruments, D. C. Johnston for the magnetic susceptibility measurements, J. A. Panella for technical assistance, and E. Frey for manuscript preparation. REFERENCES (i) (2) (3) (4) (5) (6) (7) (8)

(9) ~0) (ii) (12) ~3) (14) (15) (16) (17)

F. Wohler, Ann. Chim. Phys. 29 (1823) 43. P. G. Dickens, M. S. Whittingham, Quart. Rev. Chem. Soc. 22 (1968) 30. M. S. Whittingham, J. Electrochem. Soc. 123 (1976) 315. M. S. Whittingham, Science 192 (1976) 1126. B. Reichman, A. J. Bard, J. Electrochem. Soc. 126 (1979) 583. M. B. Dines, Mater. Res. Bull. i0 (1975) 287. A. H~ Thompson, J. Electrochem. Soc. 126 (1979) 608. P. G. Dickens, R. J. Hurdltch, in: The Chemistry of Extended Defects in Non-Metallic Solids, Ed. L. Eyring and M. O'Keefe (North-Holland, Amsterdam, 1970). J. M. Reau, C. Fouassier, G. LeFlem, J. Barraud, J. Doumerc, P. Hagenmuller, Rev. Chim. Minerale 7 (1970) 975. P. L. Mart, N. J. Clark, Mater. Res. Bull. 13 (1978) 1199. P. J. Wiseman, P. G. Dickens, J. Solid State Chem. 17 (1976) 91. M. S. Whittingham, J. Electrochem. Soc. 122 (1975) 713. M. Green, Thin Solid Films 50 (1978) 145. S. K. Mohapatra, S. Wagner, J. Electrochem. Soc. 125 (1978) 1603. P. G. Dickens, in: Intercalation Chemistry, Ed. M. S. Whlttingham, A. J. Jacobson (Academic Press, New York, 1980). D. W. Murphy, J. N. Carides, F. J. DiSalvo, C. Cross, and J. Waszczak, Mater. Res. Bull. 12 (1977) 825. M. S. Whittingham, in: Fast Ion Transport in Solids, Electrodes and Electrolytes, Ed. P. Vashishta, J. N. Mundy, G. K. Shenoy (NorthHolland, Amsterdam, 1979).

(18) H. R. Zeller, H. U. Beyeler, Appl. Phys. 13 (1977) 231. (19) D. W. Murphy, P. A. Christian, Science 205 (1979) 651. (20) M. S. Whittingham, R. A. Huggins, in: Fast Ion Transport in Sollds, Ed. W. VanGool (North-Holland, Amsterdam, 1973). (21) E. Banks, A. Goldstein, Inorg. Chem. 7 (1968) 966. (22) K. H. Cheng, A. J. Jacobson, M. S. Whittingham, to be published.