Electrodic characteristics of various carbon materials for lithium rechargeable batteries

Electrodic characteristics of various carbon materials for lithium rechargeable batteries

SYflTH|TIIC D|TRLS ELSEVIER Synthetic Metals 73 (1995) 9-20 Electrodic characteristics of various carbon materials for lithium rechargeable batterie...

1MB Sizes 0 Downloads 36 Views

SYflTH|TIIC D|TRLS ELSEVIER

Synthetic Metals 73 (1995) 9-20

Electrodic characteristics of various carbon materials for lithium rechargeable batteries Takashi Iijima a Kimihito Suzuki a, Yoshiharu Matsuda b aAdvanced Materials and Technology Research Laboratories, Nippon Steel Corporation, 1618 Ida, Nakahra-ku, Kawasaki 211, Japan b Department of Applied Chemistry and Chemical Engineering, Faculty of Engineering, Yamaguchi University, Tokiwa 2557, Ube 755, Japan Received 7 October 1994; revised 17 November 1994; accepted 9 December 1994

Abstract

Electrodic characteristics of various carbon materials have been investigated to study the correlation between structures of carbon materials and performances of negative electrodes of lithium rechargeable batteries. In the case of highly graphitized carbon materials, the discharge capacity was determined mainly by their crystallinity with no dependence on textures and natural graphite; the highest graphitization at about 360 mAh/g was stage-1 lithium-intercalated graphite, C6Li (theoretical maximum 372 mAh/g). The coulombic efficiency at the first cycle was strongly dependent on the textures of the carbon materials, and pitch-based carbon fiber of a radial structure showed an excellent coulombic efficiency over 96% by selecting appropriate electrolytes. The performances of the pitch-based carbon fiber were also excellent in the electrolytes consisting of mixed solvents containing propylene carbonate. On the other hand, the pitch coke heat-treated at 550 °C had an initial capacity over 550 mAh/g, which was beyond the theoretically maximum capacity of 372 mAh/g for C6Li, although the capacity decreased rapidly to less than 250 mAh/g within ten cycles. Polyacrylonitrile (PAN)-based carbon fiber showed a stable capacity with cycling over 350 mAh/g in spite of low graphitization. The initial coulombic efficiency seemed to increase in accordance with decrease of hydrogen and oxygen in the pitch coke, and oxygen and nitrogen in the polyacrylonitrile (PAN) fibers. These phenomena seemed to suggest that carbon materials of disordered structure would have higher capacity than that of the graphitic carbon materials. Keywords: Electrodic characteristics; Lithium; Batteries; Graphite intercalation compounds

I. Introduction

Lithium rechargeable batteries have been candidates for power sources of stationary electric power storage and electric vehicles as well as down sizing of commercialized rechargeable batteries because of their high energy density [ 1-4 ]. However, nonaqueous rechargeable lithium batteries using lithium metal as the negative electrode are still under development because of the significant problem of the short cycle life for deep-discharging and unsafe operating characteristics due to high reactivity of lithium metal. Therefore, investigation of new negative electrode materials that could replace lithium metal is one of the most important requirements in the development of more reliable and safer lithium rechargeable batteries. There have been several alternative materials proposed. Among these candidates carbon materials seem to be the most promising material from the viewpoint of large specific capacity, reversible cycling and safe operating characteristics [2]. Therefore, many efforts have been made to develop the better carbon materials for the negative electrode. 0379-6779/95/$09.50 © 1995 Elsevier Science S.A. All rights reserved SSDIO379-6779(95)O3290-Z

So far many types of carbon materials have been investigated and they are divided into two groups. One is highly graphitized carbon materials and the other is slightly graphitized carbon materials of disordered structure. As for highly graphitized carbon materials, natural graphites [5,6], artificial graphites [7], pitch cokes [8-14], pitch-based carbon fibers (PCFs) [ 15-20], chemically vapor-deposited (CVD) carbons [21,22], mesocarbon microbeads (MCMBs) [23,24], and carbon fibers produced through the CVD process [25-27] have been investigated. Especially natural graphites have been reported to demonstrate completely the same discharge capacity as C6Li, the maximum density of lithium-intercalated graphite [5 ]. As to slightly graphitized carbon materials, the electrodic characteristic.s of polyacrylonitrile (PAN)-based carbon fibers [28], various types of carbonized resins [ 29-33 ], slightly graphitized PCFs [ 16], MCMB [34,35], carbon black [36], and polyacenic semiconductor (PAS) materials [ 37 ] have been reported. Especially PAS has been reported to show discharge capacity over 1000 mAh/g [37], and a lithium density of C2Li between adjacent carbon layers has been suggested.

10

T. lijima et al. / Synthetic Metals 73 (1995) 9-20

In contrast to the experimental approach, fundamental studies to describe discharge capacity by using parameters determined through X-ray diffraction have been tried [38,39]. However, the key factors characterizing carbon materials, which could describe electrode properties, have not been obtained, because carbon materials have large variations in their chemical composition, crystallinity, microtexture, etc. The purpose of this work is to study electrodic characteristics, such as discharge capacity, initial coulombic efficiency and cycleability, of various carbon materials of amorphouslike structure and graphitic structure, and then to clarify the correlation between the structures of carbon materials and the performances as negative electrodes of lithium rechargeable batteries. As for carbon materials of disordered structure, electrodic characteristics are also discussed in relation to chemical composition. For these purposes, the authors selected carbon materials as follows: (1) graphitic carbon materials such as natural graphite, highly graphitized pitch cokes, PCFs and MCMBs, and (2) carbon materials of disordered structure such as pitch cokes heat-treated below 1200 °C, PAN carbon fiber, carbonized phenolic resin and furan resin. Furthermore, for PCFs which had the highest initial coulombic efficiency among the carbon materials above, the effects of electrolytes on the electrodic characteristics were investigated both to obtain higher coulombic efficiency and to examine the faculty of wider selection of solvents, especially the possibility of the electrolytes containing propylene carbonate (PC).

2. Experimental 2.1. Carbon materials used in this study 2.1.1. Natural graphites Two kinds of natural graphites were used: CSPE from Nippon Kokuen, Ltd., and BF5000 from Chuetsu Graphite Works Co., Ltd. The average diameter was about 4.5 and 4.0 /~m, and carbon content was over 98 and 99%, respectively. These natural graphites were used for electrode preparation as received. 2.1.2. Pitch cokes Two types of pitch cokes (Nippon Steel Chemical Co.) which had different texture were used. Pitch coke A had the texture of random directions, that is a mosaic-like texture, and pitch coke B had needle-like texture. For the investigation of highly graphitized carbon materials, both pitch cokes were used after a graphitization process at a temperature over 2800 °C, and the diameter was regulated between 20 and 30/xm. For the preparation of carbon materials of disordered structure, both were heat-treated below 1200 °C. The diameter was under 44/xm and the average diameter was about 30/~m.

2.1.3. Mesocarbon microbeads (MCMBs) Graphitized mesocarbon microbeads (MCMB-6-28) from Osaka Gas Co. were used as received. 2.1.4. Pitch-based carbon fiber (PCF) Mesophase PCF heat-treated over 2800 °C (Nippon Steel Chemical Co.) was used. The texture of the cross section of PCF was radial type with fine zigzag layers. 2.1.5. Carbonized resins Two kinds of phenolic resins, VF and HP, from Asahi Organic Chemicals Industry Co., Ltd., were prepared. VF is phenolic resin and HP is furan resin. Both were heat-treated at 500 °C for about 5 h under N2 atmosphere and then milled using a vibrating disc mill. After this process both were carbonized at temperatures of 700 and 800 °C for 1 h under N2 atmosphere. The average diameter was about 20/zm for both carbonized resins. 2.1.6. PAN-based carbon fibers T300B, T800HB and M46JB from Toray Industries, Inc., were used after washing the surface of the carbon fibers with tetrahydrofuran, HCI solution and acetone. Diameters were about 6.9, 5.1 and 5.1 /~m, respectively. To investigate the influence of electrolytes on the electrodic characteristics, T300B was heat-treated at 1000, 1300, 1500, 1700 or 2000 °C for 1 h under Ar atmosphere, after washing. 2.2. Electrochemical measurement In the case of carbon fibers, electrodes were prepared by bundling 10 mg of carbon fibers (about 30 mm long) with Ni string (0.1 mm diameter). Powder-type carbon material was mixed with 2 wt.% of polytetrafluoroethylene and kneaded to a flexible sheet; then it was cut into a square ( 1 cm 2) and pressed onto a copper foil (20 ~m thick). These electrodes were heated at about 100 °C for more than 1 h under vacuum before cell assembling. Typical three-electrode cells were constructed for electrochemical measurements using carbon fiber electrodes and lithium foil as the counter and reference electrodes. For powder-type carbon materials, the carbon sheet electrodes were sandwiched with a lithium foil separated with a porous polypropylene separator (from Ube Industries, Ltd.) in order to get the surroundings of carbon electrodes closer to that of the practical battery itself. To investigate the effect of electrolytes on the electrode properties of carbon materials, various kinds of solvents were examined as follows: ethylene carbonate (EC), PC, butylene carbonate (BC), 7-butyrolactone (BL), sulfolane (S), acetonitrile (AN), diethyl carbonate (DEC), 4-methyldioxisolane (4MeDOL), 2-methyltetrahydrofuran (2MeTHF), dimethoxyethane (DME), dimethylsulfoxide (DMSO). All the electrolytes used were 1 M solutions of LiCIO4 in one kind of solvent, or a mixture of two kinds of solvents (50/ 50 vol.).

T. lijima et al. /Synthetic Metals 73 (1995) 9-20

11

Table 1 Structural parameters measured by X-ray diffraction, surface area measured by the conventional N2 BET method, and electrode properties of natural graphite and various highly graphitized carbon materials Sample

doo2 ( n m )

Lc (nm)

L. ( n m )

Surf. area ( m 2 / g )

Init. cap. ( m A h / g )

Init. coul. eft. ( % )

Max. cap. ( m A h / g )

9.1 10.2

265 317

75.2 77.6

:283 317

352 " 300 317 269 289

76.3 a 54.7 66.8 87.3 96.3

359 " 1310 317 272 304

Natural graphite

CSPE BF5000

0.3354 0.3354

> 100 > 100

> 100 > 100

Pitch coke Pitch coke MCMB PCF

Ab Bc 6-28

0.3360 0.3358 0.3374 0.3371

> 100 > 100 30 45

> 100 > 100 80 78

l

0.22 0.76 4.0 0.31

a Current 0.1 mA/cm 2. b Pitch coke A, mosaic-like texture. c Pitch coke B, needle-like texture•

The electrochemical measurements were carried out using computer-controlled constant current cyclers. The charge and discharge were cycled at 0.1-0.3 mA for carbon fibers, and 0.1-0.5 mA for powder-type carbon materials at room temperature. The potential ranges of carbon electrodes were between 0 and 1.0 V versus Li/Li ÷ for highly graphitized carbon materials, and between 0 and 2.0 V for slightly graphitized carbon materials of disordered structure.

400 '7 e-

E

~

350

® 300

o

• o

o 250 0.335

3. Results and discussion

3.1. Electrodic properties of natural graphites and highly graphitized carbon materials For natural graphites, pitch cokes, MCMB and PCF, structural parameters determined through X-ray diffraction (XRD) measurement, surface area measured by the conventional N2 BET method and several electrodic properties are listed in Table 1. Parameters of d-spacing (doo2), Lc and L, correspond to the distance between adjacent carbon layers, average crystallite size along the c-axis and a-axis, respectively. The electrolyte of E C + D E C ( 5 0 / 5 0 vol.)/1 M LiC104 was used and the current density was controlled at 30 m A / g for PCF and 50 m A / c m 2 for the other carbon materials. With regard to BF5000, cycling tests at both 0.1 and 0.5 m A / cm 2 were carried out to investigate effects of the current density on the discharge capacity. Apparently, among these materials, natural graphites had the highest degree of graphitization and then pitch coke B, pitch coke A, PCF and MCMB in turn. The d-spacing of the natural graphites was almost the same (0.3354 nm) as that of the ideal graphite structure. Fig. 1 shows the relation between the graphitization (doo2) and the initial and maximum discharge capacities. It was recognized that discharge capacities of these carbon materials increased with decreasing doo2, and natural graphite (BF5000) showed about 360 m A h / g, just close to the capacity of C6Li, the maximum density in lithium-intercalated graphite. This result showed a good correspondence to our previous study [ 16], showing the linear relation between doo2 and the discharge capacities of pitch

i

i

0.336

0.337

0.338

doo2lnm

Fig. 1. Relation between discharge capacity and d-spacing of various graphitic carbon materials in EC+DEC (50/50 vol.)/LiCIO4: O, initial capacity; • : max. capacity. 1.5 +,,_1 - 1.2 .._1 0.9

r

i,

]

I 300

i

c 0.6 "5

(3..

0.3 s

I 100

i

I 200

i

400

Capacity/mAhg -1 Fig. 2. Discharging profiles of various graphitic carbon materials at the initial cycle in EC + DEC ( 5 0 / 5 0 vol. )/LiCIO4: , BF5000; . . . . . , pitch coke A ; - - - - , PCF; . . . . , MCMB.

cokes and PCFs heat-treated at various temperatures. Therefore, doo2 seems to be an excellent parameter to describe discharge capacities of graphitic carbon materials even though their textures are different from each other. Fig. 2 shows discharge profiles of these graphitic carbon materials at the first cycle. The common characteristics of all these carbon materials were long plateaux near the potential of Li/Li +, and it was recognized that the variation of discharge capacities resulted from the differences of length of the plateaux. Carbon materials of higher graphitization showed a longer plateau. The color of all the carbon materials

12

T. lijima et al./ Synthetic Metals 73 (1995) 9-20

'7,

highest among data reported previously. Including the reported data of MCMB (about 92%) and natural graphite (about 92%), the order of increasing efficiency could be summarized as follows: PCF > natural graphite, MCMB > pitch coke B > pitch coke A. This result indicates that the initial coulombic efficiencies seemed to be determined not by the surface area, but dominantly by the textures of carbon materials, because pitch cokes (20-30/xm in diameter) having the smallest surface area showed the lowest efficiency and natural graphites having the largest surface area showed a higher efficiency. In the case of PCF, MCMB or natural graphite, the textures are well defined and well controlled on the nanometer scale [40], that is, PCF of twodimensional radial structure, MCMB of three-dimensional radial structure and natural graphite of ideal graphite structure. On the other hand, the textures of the pitch cokes are rather complicated and can be recognized as a kind of composite consisting of lots of crystallites. Therefore, it was suggested that the high coulombic efficiency could be obtained by using carbon materials of well-defined and well-controlled textures on the nanometer scale. Because a high coulombic efficiency of graphitic carbon materials has a strong correlation with the faculty of preventing solvent co-intercalation, it was expected that PCF had a capability of wider selection of solvents for electrolytes. To investigate this capability, electrodic characteristics of PCF were examined in various kinds of electrolytes.

400

==

|iluilililil

E 300

8. 200

o 100 0

~ J 10 20 Cycle N u m b e r

30

Fig. 3. Change of discharge capacity with cycling of various graphitic carbon materials in EC +DEC (50/50 vol.)/LiCIO4: O, CSPE; [~, BF; i , BF at 0.1 mA; A, pitch coke A; O, pitch coke B; 0 , MCMB; A, PCF.

changed with charging from graphitic black to golden yellow, corresponding to the color of stage-1 lithium-intercalated graphite synthesized through chemical methods. XRD profiles also changed to that of stage 1, with a color change to golden yellow. These results, together with the capacity of natural graphite close to C6Li, indicated that the electrode process of graphitic carbon materials was essentially intercalation of lithium into graphite. Concerning the effect of current density during the cycling test on the discharge capacity, BF5000, for example, showed a large difference of discharge capacity between 359 and 317 mAh/g at the current density of 0.1 and 0.5 mA/cm 2, respectively. The other carbon materials showed a similar tendency of increasing discharge capacities with decreasing current density; however, the difference was less than 10 mAh/g. The rather large change of capacities for natural graphite seemed to be caused mainly by its high bulk density of electrodes: more than 1.5 g/cm 3 for natural graphite and less than 1.3 g/cm 3 for the other carbon materials. The high bulk density of the electrode was thought to prevent the diffusion of lithium into the space between the graphite particles. Fig. 3 shows cycle dependence of discharge capacities. All of the carbon materials used for electrodes showed stable discharge capacities with cycling within 30 cycles and, thus, it was suggested that graphitic carbon materials had a basic property of good reversibility in the lithium intercalationdeintercalation process. It was also important for practical use to compare these graphitic materials from the viewpoint of the initial coulombic efficiency. The initial coulombic efficiency of PCF was about 96%, by far superior to the others, and probably the

3.2. Effects of solvents used as the electrolyte on electrodic properties of PCFs The effect of electrolytes on the electrode performance of PCF was investigated, and the suitability of solvents for lithium-intercalation reaction through the electrodic properties of PCF was clarified. The electrodic properties on the first cycling were investigated in electrolytes consisting of one kind of solvent solution, then in electrolytes of mixed solvents using EC and another solvent, and finally in the electrolytes using PC with another solvent without EC, because PC is in general more suitable for a good performance at a temperature below 0 °C.

3.2.1. Effects of electrolytes consisting of single solvent/1 M LiCIO4 on the electrodic properties Electrodic properties of PCF in electrolytes consisting of single solvent/1 M LiC104 are summarized in Table 2. The electrolytes using PC, BC, 4MeDOL, 2MeTHF, DME and

Table 2 Electrodic properties of PCF in various electrolytes consisting of single solvent

Initial coulombic efficiency (%) Initial capacity (mAh/g) Max. capacity (mAh/g)

EC

BL

S

DEC

PC

BC

96.4 277 292

87.5 214 219

34.3 22 22

7.7 16 16

no intercalation

4-MeDOL

2MeTHF

DME

DMSO

T. l i j i m a e t al. / S y n t h e t i c M e t a l s 73 ( 1 9 9 5 ) 9 - 2 0

300 '7

0

1,5

000000000 +-

E

[] []

13

(a)

1.2

........

1st cycle 2nd cycle 3rd cycle

_J

200 []

~

0.9

~ o

0.6

[]

100

o._ 0.3

[]

t~

[]

5

10

15

i

l

I

2

4

6

Fig. 4. Change of discharge capacity with cycling of PCF in various electrolytes consisting of single solvent solution of LiCIO4: ©, EC; I-7, BL; A , S ; O , DEC.

1.5

(b)

10

1st cycle

t _ 1.2 ._J J

........ : : : : : :'

>~ 0,9

DMSO are not suitable for electrochemical lithium intercalation into PCF because PCF was destroyed to powder during charging and, therefore, no discharge capacity was obtained. The destruction of PCF was mainly due to the exfoliation caused by solvent co-intercalation with lithium. Fig. 4 shows cycle dependence of discharge capacity of PCF in various electrolytes. Only the EC electrolyte showed an excellent coulombic efficiency (96.3%) and a large and stable discharge capacity (292 mAh/g), which were almost the same as those in EC + DEC shown in Table 1. Figs. 5(a) and (b) show charge and discharge profiles of PCF in electrolytes EC and BL, respectively. PCF also demonstrated both a high initial efficiency and a large initial discharge capacity in the BL electrolyte, though the capacity decreased to less than half of the initial value within five cycles. Figs. 5(c) and (d) show charge and discharge profiles of PCF in the electrolytes S and DEC, respectively. In the case of DEC, lithium metal reacts with DEC easily, so activated carbon material (ACM) was used as both counter and reference electrodes. In the cases of S and DEC, discharging profiles at the first cycle indicated that lithium was intercalated into PCF smoothly; nevertheless little initial discharge capacity was obtained. It was recognized that, in the two electrolytes S and BL, the overpotential which became larger with cycling caused the rapid decrease of capacity. This suggested that film formation occurred on the surface of PCF with cycling, which prevented lithium intercalation. In fact, the PCF electrode after degradation in BL and S electrolytes showed similar electrodic properties in EC electrolyte to that of pristine PCF by washing the surface of degraded PCF with methanol, which allowed us to conclude that very little structural destruction of PCF was induced through charging and discharging processes. On the other hand, the small discharge capacity of PCF in the DEC electrolyte could be understood because a reaction between intercalated lithium near the surface and DEC caused very low discharge capacity. Little structural destruction of PCF through cycling in DEC was also confirmed. In any case, degradations of electrodic properties in electrolytes using BL, S and DEC essentially differed from the co-intercalation process of the solvent.

8

Time / hour

Cycle Number

c

0.6

o 13_

i b i J i

,

/

L

0.3

i

/

.;

2

3rd cycle 5 t h cycle

J

4

6

8

10

Time / hour 1,5

1 st cycle

C) t_ J J

2nd cycle

1.2

........

0.9

~

"5

3rd cycle

: ii .... ii .. /

0.6

I1.

0.3

~":~7--_ ...j 1.0

0.5

1.5

2.0

2.5

Time / hour -1.5

(d)

1st cycle 2nd cycte

- -

-1.9 O <

Y

o~ > -2.3

~ -2.7 a. -3.1

-3.5 0

I 1

[ 2

I 3

I 4

I 5

I 6

Time / hour

Fig. 5. Chargeand dischargeprofiles of PCF in (a) EC/LiC104, (b) BL/ LiCIO4, (c) S/LiC104 and (d) DEC/LiCIO4.

3.2.2. Effects of electrolytes consisting of mixed solvents: EC + X/1 M LiCl04 (EC/X=50/50 vol.) on the electrodic properties Electrodic properties of PCF in the electrolytes consisting of mixed solvents, EC + X/1 M LiCIO4, are summarized in Table 3, and the change of capacity with cycling is shown in Fig. 6. In the electrolytes except E C + D M E and EC + DMSO, PCF showed high initial efficiency and large discharge capacity, and also an almost stable capacity with

14

T. lijima et al. / Synthetic Metals 73 (1995) 9-20

Table 3 Electrodic properties of PCF in various electrolytes consisting of EC + X (50/50 vol. )

Initial coulombic efficiency (%) Initial capacity (mAh/g) Max. capacity (mAh/g)

BL

S

AN

DEC

PC

BC

4MeDOL

2MeTHF

DME

DMSO

94.4 288 289

98.8 267 270

96.4 300 301

96.3 289 304

91.0 270 279

90.2 209 237

89.6 287 290

85.8 274 282

1.4 1.8 4.1

1.8 4.1 8.5

350 "7, '~

300 - ~ O O O O O O O O o n o c ~ "A

@ @ @ ® *l l i i l

AAAAAAAAAAAAA AAAA AAAAA~

~. 250

8

iUuuUnu uUIu

~ 2oo

U

nnn

ii n l l l n u n ii m I l l

5 150

i

5

i

t

10 15 Cycle Number

i

20

25

Fig. 6. Discharge capacity of PCF in various electrolytes containing mixed solvents consisting ofEC + X (50/50 vol.)/LiCIO4: O, AN; 71, BL; II, BC; Zk, S; @, 2MeTHF; A, 4MeDOL.

cycling was confirmed. Especially in the electrolytes of EC + AN and EC + DEC, PCF showed the same initial efficiency as that in EC electrolyte, and showed larger initial capacity and larger maximum capacity. The other electrolytes decreased the initial coulombic efficiency compared with that in EC electrolyte. These changes of initial efficiency dependent on EC + X seemed to show both suitability of X for smooth intercalation of lithium into PCF and reactivity of X with lithium-intercalated PCF. It was assumed that reactivity of X would be negligible in a mixture EC + X to the first approximation because the affinity of EC with lithium was stronger than that of the other solvents. Therefore, if the initial efficiency in EC + X is the same as that in EC, as in EC + AN and EC + DEC, suitability of X for smooth intercalation is similar to EC. On the other hand, less efficiency in EC + X than in EC itself seemed to accord with a lower suitability of X for smooth intercalation of lithium than in EC. Thus, accounting for the value of the initial coulombic efficiency, the order of suitability for the lithium-intercalation reaction was determined as follows: EC > AN, DEC > BL > S > PC > BC, 4MeDOL > 2MeTHF > DME, DMSO.

3.2.3. Effects of electrolytes composed of mixed solvents: PC+X~1 M LiCl04 (PC/X=50~50 vol.) on the electrodic properties The electrolytes using mixed solvent based on EC are suitable for lithium intercalation into highly graphitized carbon materials, but an improvement of performance is still required for practical use, such as a good performance at a low temperature below 0 °C. PC is one of the candidates which could replace EC, because PC has a lower viscosity than EC and similar physical properties to EC.

Electrodic properties of PCF in electrolytes of PC + X are summarized in Table 4. The cycling stability of discharge capacity is shown in Fig. 7, and the charge and discharge profiles are drawn in Fig. 8. It was evident that PCF had the capability of a wider selection of solvents than EC for a smooth intercalation reaction. As for the initial efficiency, the order of suitability of solvents was reconfirmed with exception of DEC. PC + DEC electrolyte reacted with lithium metal, and its high reactivity with lithium metal seemed to cause the lessening of the initial coulombic efficiency and cycle stability. These results allowed us to conclude that PCF is one of the best materials for a negative electrode in lithium rechargeable batteries, because it had excellent high initial coulombic efficiency in many kinds of solvents for electrolytes. In particular, PC + X could be used, though the capacity is a little smaller than that in electrolytes using mixed solvents of EC +X.

3.3. Electrodic properties of pitch cokes A and B of disordered structures The effect of heat-treatment temperature (HTI') below 1200 °C was investigated in detail by using pitch cokes A and B as electrode materials because a previous study [ 16] on PCFs and pitch cokes heat-treated at various temperatures resulted in the suggestion that slightly graphitized pitch-based carbon materials heat-treated below 1000 °C were more capable of a higher capacity than C6Li. Tables 5 and 6 summarize the results of elementary analysis and electrodic properties of pitch cokes A and B, respectively. Fig. 9 shows the change of discharge capacities with cycling of pitch cokes A and B heat-treated at various temperatures; the HT'I" dependence of discharging profiles of pitch coke B is shown in Fig. 10. Both pitch cokes A and B indicated similar tendency in elementary analysis and electrodic properties. Initial discharge capacities showed the maximum on both pitch cokes heat-treated at 550 °C, although discharge capacity decreased rapidly with cycling. The maximum capacities over 500 mAh/g suggested that the electrode process differed essentially from that of highly graphitized carbon materials. This was also confirmed in discharge profiles of Fig. 10. Higher capacities than C6Li resulted from increasing capacity at the potential between 0.4 and 1.2 V, which was completely different from those at the potential range between 0 and 0.1 V of graphitic carbon materials. Cycling stability of discharge capacity could be improved by raising HTI's, and pitch cokes heat-treated higher than 800 °C showed stable capacity with

T. lijima et al. / Synthetic Metals 73 (1995) 9-20 300 "7,

== O

E .~., 200

AAAAAA

AA~A

A

8.

o

8 ~100

O ~5 i

I

3

i

O

i

12

6 9 Cycle Number

15

Fig. 7. Discharge capacity of PCF in various electrolytes containing mixed solvents consisting of PC + X (50/50 vol.)/LiCIO4: O, AN; I-1, BL; i , 4MeDOL; A, S.

t_ 1.2 1'5 I _J ....I

>~ 0.9

!I

._

>"

0.6

i ~,.

j

Y'. "6.

O

(a.

/

/

.:

i

i

0.3

o r~

I

i

0

50

100

q----7--~='-~.~ 150

200

250

300

15

an irreversible capacity, the initial value of which was over 300 mAh/g on the pitch coke heat-treated at 550 °C, and probably it was the main origin of higher capacity than that of C6Li. Recently, it was reported that slightly graphitized MCMB heat-treated at 700-800 °C indicated initial capacities over 700 mAh/g [ 33 ]. Larger discharge capacities than C6Li of slightly graphitized pitch-based soft carbons, such as PCFs, pitch cokes and MCMBs, heat-treated below 800 °C would be caused by the large irreversible capacity at the potential between 0.4 and 1,2 V. There seemed to be a good correlation between the contents of hydrogen and oxygen, and the initial electrode properties. The initial coulombic efficiency was improved by higher Hq-T, as well as by the cycling stability of discharge capacities. In particular, a sudden decrease of oxygen content in pitch cokes heat-treated above 700 °C had a good accordance with the improvement of cycling stability and the initial couIombic efficiency, and also with decrease of the irreversible capacity at potentials between 0.4 and 1.2 V. Although the detailed mechanism of larger capacity than that of C6Li has not been clarified, the results cited above showed the difficulty of larger stable cycling capacity by simple control of HTT for pitch-based soft carbons.

3.4. Electrodic properties of carbonized phenolic resin of disordered structures

350

Capacity/mAhg -1

Fig. 8. Charge and discharge profiles of PCF in electrolytes consisting of P C + X (50/50 vol.)/LiC104 at the initial cycle: - - , PC+BL; . . . . PC+DEC; . . . . . , PC+S.

cycling. However, the stationary capacities after cycling were almost the same (200-250 mAh/g) among all of the pitch cokes heat-treated at different temperatures, even if the initial capacities differed extremely. This implied that capacities of slightly graphitized pitch cokes consisted of two different capacities. One was a reversible capacity on cycling (200250 mAh/g), which was in good accordance with the capacity of pitch cokes reported previously [ 16], and the other was

Carbonized phenolic resin and furan resin were prepared as representatives of so-called hard carbon, and the charge and discharge profiles were measured. The results of elementary analysis and electrodic properties are summarized in Table 7. Charge and discharge profiles at the initial cycle are drawn in Fig. 11, and the cycle dependence of discharge capacities of both carbonized resins heat-treated at various temperatures are shown in Fig. 12. The electrodic properties of both carbonized resins were almost the same. Both resins heat-treated at 700 °C showed the largest discharge capacities of about 200 mAh/g. How-

Table 4 Electrodic properties of PCF in various electrolytes consisting of PC + X (50/50 vol.)

Initial coulombic efficiency ( % ) Initial capacity (mAh/g) Max. capacity (mAh/g)

BL

S

AN

DEC

BC

4MeDOL

2MeTHF

94.4 268 273

48.5 158 185

91.7 275 279

78.1 245 264

1.1 24 24

87.0 250 255

no intercalation

DME

DMSO

Table 5 Elementary analysis and electrodic properties of pitch coke A heat-treated at various temepratures HTT (°C)

C (wt.%)

H (wt.%)

O (wt.%)

Init. coul. eff. (%)

Init. cap. (mAh/g)

Cap. at 10th cycle (mAh/g)

550 650 700 1020 1200

92.6 92.6 93.7 98.7 99.1

1.95 1.19 0.95 0.14 <0.1

2.06 3.67 3.32 < 0.1 <0.1

63.1 67.1 65.2 81.2 84.8

564 387 297 239 221

103 294 220 205 214

T. lijima et al. / Synthetic Metals 73 (1995) 9-20

16

Table 6 Elementary analysis and electrodic properties of pitch coke B heat-treated at various temperatures HTT (°C)

C (wt.%)

H (wt.%)

O (wt.%)

lnit. coul. eff. (%)

Init. cap. ( m A h / g )

Cap. at 10th cycle ( m A h / g )

500 550 600 650 700 800

93.5 92.8 92.5 93.4 95.4 98.4

2.34 1.93 1.57 1.20 0.97 0.43

1.47 2.23 3.37 3.21 2.34 0.25

44.6 57.1 68.4 68.2 72.5 77.4

435 522 466 319 356 263

70 148 306 292 279 244

600

o

'7,

capacities with cycling, the discharge capacities of all the carbonized resins decreased monotonically with cycling and no stationary capacity was obtained. The initial coulombic efficiencies increased with rise of HTT, but the efficiencies of both resins were 55% at the highest. The different behaviors of cycling stability and initial efficiencies between carbonized resins and pitch cokes seemed to be caused by the differences in the contents of hydrogen and oxygen with the increase of HTT. A larger amount of oxygen remained in carbonized resins heat-treated at 800 °C. However, pitch cokes contained hardly any hydrogen and oxygen after heattreatment at 800 °C. Except for carbonized resins heat-treated at 500 °C, it was recognized from Fig. 11 that both carbonized resins had discharge profiles similar to those of pitch cokes heat-treated at 800 °C. These results lead to the conclusion that both phenolic resin and furan resin did not show a discharge capacity over 300 mAh/g by a simple heat-treating process.

(a)

: : 500 o E ~, 400 n 0 8. 300 m,,

~200

aaA

aaa~

lOO I

I

I

5

10

15

I

I

i

I

20 25 Cycle Number

30

35

600 [

,

sooi~ 400

IO:~:

(b)

i1~,

i3oo/x.~~.~.,....'... .o oo 100

A° D o o ~ I

0

5

,

L

I

10

L

I

3.5. Electrodic properties of PAN-based carbon fibers of disordered structures

I

15 20 25 Cycle Number

30

35

Fig. 9. Change with cycling of discharge capacity of pitch cokes heat-treated at various temperatures in EC + DEC (50/50 vol.)/LiCIO4: (a) pitch coke A; (b) pitch coke B.

•"~ 1.5 +-J

",=-~1.0

.

i i/ j Y

;

I

.-

.~

,

0

Q" O.5

0

/

I

I

I

I

1oo

200

300

400

I

500

600

Capacity/mAhg 1

Fig. 10. Discharge profiles of pitch coke B heat-treated at various temperatures in E C + D E C (50/50 vol.)/LiC104: , 550 °C; - - - - - , 600 °C; - - -, 700 °C; . . . . . ,800 °C.

ever, the capacities decreased within a few cycles, similar to the behavior of slightly graphitized pitch cokes. Although pitch cokes heat-treated at 800 °C showed stable discharge

Three kinds of PAN-based carbon fibers, T300B, T800HB and M46JB, were used in this work. T300B and T800HB were carbon fibers of high tensile strength, and M46JB was carbon fiber of high modulus. In Table 8 results of elementary analysis, diameter, structural parameters determined by XRD, and electrodic properties are summarized. EC + DEC/ 1 M LiC104 was used as electrolyfe. T800HB showed the largest capacity among the three samples, but no relation could be recognized between the structural parameters and discharge capacity. On the other hand, it was suggested that the high initial coulombic efficiency was in good accordance with the contents of oxygen and nitrogen. In order to investigate in detail the electrodic properties of PAN carbon fiber and clarify the key factors controlling electrodic performances, T300B was heat-treated under Ar atmosphere for 1 h at temperatures from 1000 to 2000 °C, and at the same time the effect of solvents in electrolytes on the electrodic properties was investigated for heat-treated T300B. EC, PC and BC were chosen as the solvents for the electrolytes because they have different suitability for smooth intercalation of lithium and relatively similar electrochemical stability. The current density was controlled at 10 mA/g in order to neglect

11 lijima et al. / Synthetic Metals 73 (1995) 9-20

17

Table 7 Elementary analysis and electrodic properties of phenolic resin heat-treated at various temperatures

HPI000

VF307

HTT (°C)

C (wt.%)

H (wt.%)

O (wt.%)

Init. coul. eff.(%)

Init. c a p . ( m A h / g )

Cap. at 10th cycle (mAh/g)

500 700 800 500 700 800

88.2 87.1 86.8 86.9 88.4 87.8

2.22 0.98 0.85 1.85 1.15 0.87

3.97 5.36 6.19 4.36 4.14 6.49

26.2 50.5 52.7 27.8 54.5 54.8

127 191 171 124 198 145

87 133 142 84 119 75

2.0

>~ + ~ 1.5 _.J

/::/i '.\

-~ 1.0

,,,,'/' "\ ,,' / ~% ,."/' "".X/

o ca- 0.5

.4

0

":'~",N

1O0

200

300

400

500

Capacity/mAhg -1 Fig. l 1. Charge and discharge profiles of carbonized resin HP heat-treated at various temperatures in EC + PC (50/50 v o l . ) / L i C I O 4 : - - , 500 °C; -, 700 °C; - . . . . , 800 °C. 200 '7, 150

~

100

0

0

10

20

30

40

50

Cycle Number 200 B-

(b)

'7

== 150

~

100

0

~ s0 0 10

20

30

40

5O

Cycle Number

Fig. 12. Change with cycling of discharge capacity of phenolic resin heattreated at various temperatures in EC + PC (50/50 vol.)/LiCIO4: (a) HP; (b) VF; O, no treatment; [2], 700 °C; A, 800 °C.

the effects of electric conductivity of the electrolytes on discharge capacity. Results of elementary analysis and structural parameters are listed in Table 9. With increasing HTF, the contents of

oxygen and nitrogen gradually decreased attd finally disappeared completely at 2000 °C. The initial electrode properties in three kinds of electrolytes are summarized in Table 10. At first it was recognized that smooth lithium doping and undoping processes occurred in all the electrolytes. This means that basically PAN carbon fibers had a wider selection of solvents for electrolytes than graphitic carbon materials, The initial discharge capacity decreased and the initial coulombic efficiency increased with rising HTT. Although the structural parameters did not change at all, T300B heat-treated at 1000 °C indicated a much larger discharge capacity than that of T300B after washing. Therefore, it was suggested that the difference of discharge capacity was affected by chemical functions on the surface of T300B. The discharge capacities of T300B heat-treated above 1000 °C seemed to be determined by structural change. It was observed that higher graphitization accorded with smaller discharge capacity, contrary to the results observed for highly graphitized carbon materials. T300B heat-treated at 1700 °C had maximum initial cycling efficiency, reaching about 90% in the electrolyte containing EC. The behavior of the initial coulombic efficiency with HIT seemed to have good accordance with the contents of hydrogen and oxygen. Decrease of the contents of hydrogen and oxygen accorded with increase of the initial efficiency, as for slightly graphitized pitch cokes. Fig. 13 indicates the stability of the discharge capacity with cycling of heat-treated T300B in EC electrolyte. The discharge capacity was stable with cycling, although a small decrease during the initial few cycles was observed in cases with heat-treatment at 1000 and 1300 °C. Compared with slightly graphitized pitch cokes and carbonized resin, these results indicated that stability with cycling required not only small contents of hydrogen and oxygen, but also a higher HTF than 1000 °C. Fig. 14 shows the initial charge and discharge profiles of T300B heat-treated at various temperatures. On increasing HTr, the discharge capacities decreased and the difference between charge and discharge capacities also decreased. It is worth noting that a plateau near 0-0.2 V, shown in the case of highly graphitized carbon materials, could be observed on T300B heat-treated at 1000 °C in spite of its low graphitization. This plateau decreased with increasing HT-F and disappeared at treatments higher than 1500 °C. This implied that high graphitization was not necessarily required to obtain a fiat discharge profile between 0 and 0.2 V versus Li/Li ÷. The space between the stacking of appropriate sizes of

T, lijima et al. / Synthetic Metals 73 (1995) 9-20

18

Table 8 Elementary analysis, structural parameters and electrodic properties of several kinds of PAN-based carbon fibers

T300B T800HB M46JB

Diameter (/zm)

Density ( g / c m 3)

C (wt.%)

H (wt.%)

O (wt.%)

N (wt.%)

doo2 (nm)

Lc (nm)

Init. eft. a (%)

Init. cap. b (mAh/g)

6.9 5.1 5.1

1.76 1.81 1.84

98.7 99.1 99.3

<0.1 <0.1 <0.1

0.59 <0.1 <0.1

5.11 4.56 <0.1

0.3538 0.3533 0.3456

1.3 1.6 4.1

44.3 67.2 87.6

193 281 123

a Initial coulombic efficiency at the 1st cycle. b Initial discharge capacity at the I st cycle. Table 9 Elementary analysis and structural parameters of PAN-based carbon fiber (T3OOB) heat-treated at various temperatures HTT (°C)

C (wt.%)

H (wt.%)

O (wt.%)

N (wt.%)

doo2 (nm)

Lc (nm)

1000 1300 1500 1700 2000

93.8 96.3 98.1 98.9 100

< 0.1 <0.1 <0.1 <0.1 <0.1

0.51 0.16 0.10 0.10 < 0.1

5.11 3.57 2.05 0.94 < 0.1

0.3538 0.3530 0.3512 0.3486 0.3474

1.3 1.4 1.7 2.0 2.8

Table 10 Electrodic properties of PAN-based carbon fiber (T300B) heat-treated at various temperatures in electrolytes composed of single solvent HTT (°C)

EC

1000 1300 1500 1700 2000

PC

BC

Init. eft. a (%)

Init. cap. b ( m A h / g )

lnit. eft. a (%)

Init. cap. b ( m A h / g )

lnit. eft. a (%)

Init. cap. h ( m A h / g )

71.9 79.0 84.0 90.1 85.2

360 332 245 211 162

70.9 77.1 83.4 87.3 85.2

339 308 247 207 160

71.5 76.8 82.8 88.0 84.6

306 282 229 198 157

a Initial coulombic efficiency at the 1st cycle. b Initial discharge capacity at the 1st cycle.

2.0

400

~OOo E

&

300

0

.

> .+__

oOOoo ooo 0000 OOO OOOO

DO0

1.5 .d

el

200

oeo

05 >

Ooooo

"~ 1.0

A •

~,

&A&A&&&&~AO&&&AA~

a. 0.5

100

i

i

i

I

5

10

15

20

0 25

Cycle Number

x

~

/ ',

"\

,"

100

,"

../

./

/

200 300 Capacity/mAhg "1

400

500

Fig. 13. Change of discharge capacity with cycling of PAN carbon fibers (T300B) heat-treated at various temperatures in EC/LiCIO4: (3, 1000 °C; El, 1300 °C; O, 1500 °C; Zk, 1700 °C; ~,, 2000 °C.

Fig. 14. Charge and discharge profiles of PAN carbon fibers (T300B) heattreated at various temperatures in EC/LiCIO4: ,1000 °C; - - - - - , 1500 °C; 2000 °C.

carbon layers is probably more important than small d-spacing. The initial discharge capacities and initial coulombic efficiencies of T300B heat-treated at various temperatures in three electrolytes are shown in Figs. 15 (a) and (b), respectively.

Though no difference in the initial cycling efficiencies was observed, discharge capacities showed interesting behaviors. With increasing HTI', the difference of discharge capacity between EC and PC electrolytes initially disappeared and then all the capacities in the three electrolytes became equal. An assumption can be presented to explain the results cited

T. lijima et al. / Synthetic Metals 73 (1995) 9-20 400

":,

o

,~

E

(a)

300

[]

0



[]

"6

,~

200

z t0C

i

800

1000

,

i

,

1200

i

,

i

e

1400 1600 H T T / °C

i

1800 2000

2200

100

0~

(b) o

90

tu

80

o

70

60 800

o

, , i 1000 1200

,

i , , , i 1400 1600 1800 H T T / °C

, 2000

2200

Fig. 15. Heat-treatment temperature dependence of (a) initial discharge capacity and (b) initial cycling efficiency of PAN carbon fibers (T300B) in various electrolytes: ©, EC; IS], PC; , , BC.

above, though there was no evidence that an average size of solvated lithium ions affected the discharge capacity through the pores existing on the surface of T300B. As the pores seem to shrink with increase of treatment temperature, the difference of discharge capacity in the three electrolytes would be smaller and would finally disappear for T300B heat-treated at 2000 °C. This can be understood by assuming that the average size of solvated lithium ions is in the order EC > PC > BC, and that the largest size of solvated lithium ions leads to the largest discharge capacity, although the reason has not been clarified. Also, it was assumed that the effect of the solvent on lithium intercalation is negligibly small compared with the effect of the average size of solvated lithium ions, because the initial cycling efficiencies indicated no difference for EC, PC and BC.

4. C o n c l u s i o n s

Electrochemical properties of various kinds of carbon materials were investigated from the viewpoint of practical application for a negative electrode of lithium rechargeable batteries. For highly graphitized carbon materials, d-spacing, which accords with the graphitization, was recognized as a good parameter to describe the discharge capacity, even if the textures of carbon materials were not the same. On the other hand, the initial coulombic efficiency seemed to be dominated by the texture of carbon materials. In particular, PCF of a radial texture showed excellent initial efficiency (96.3%) in the electrolyte consisting of EC +DEC. It was

19

also clarified that PCF showed cycling electrode performances in the electrolytes consisting of PC + X (50/50 vol.), although the other graphitized carbon materials could not be intercalated with lithium. This wide selection in solvents would make it possible to improve several performances of batteries. PC is one of the promising solvents for the electrolyte because of its good characteristics at a temperature below 0 °C. Thus, PCF is a promising material for lithium rechargeable batteries from the viewpoint of the electrolyte. Electrodic characteristics of carbon materials of disordered structure were investigated through slightly graphitized pitch cokes, carbonized resin and PAN carbon fibers by controlling the HTT. Pitch coke heat-treated at 550 °C showed larger discharge capacity (above 550 mAh/g) than C6Li, the highest density of lithium-intercalated graphite, although the capacity decreased to below a half of the initial value within ten cycles. Also, PAN carbon fiber heat-treated at 1000 °C demonstrated 350 mAh/g, in spite of its low graphitization. These results implied that slightly graphitized carbon materials were capable of a larger capacity than that of C6Li through the different mechanism of lithium intercalation into graphite. It was also recognized that a higher initial cycling efficiency had a good correlation with low contents of hydrogen and oxygen for pitch cokes, and oxygen and nitrogen for PAN carbon fibers.

References [ 1] M. Brousselly, J. Labat and J.M. Cocciantelli, J. Power Sources, 13 ( 1991 ) 429. [2] T. Nagaura, 4th Int. Rechargeable Battery Seminar, Deerfield Beach, FL, USA, 1990. [3] Z. Takehara and K. Kanamura, Electrochim. Acta, 38 (1993) 1169. 14] K. Abraham, Electrochim. Acta, 38 (1993) 1233. [5] M. Takahashi, T. Nohma, M. Fujimoto, R. Ohshita, K. Ueno, K. Nishio and T. Saito, Proc. 34th Battery Symp., Hiroshima, Japan, 1993, p. 73. [6] K. Higashimoto, K. Nakai, K. Hironaka, T. Hayakawa, A. Komaki, Y. Chong and N. Watanabe, Proc. 34th Battery Symp., Hiroshima, Japan, 1993, p. 75. [7] Z.X. Shu, R.S. McMillan and J.J. Murray, J. Electrochem. Soc., 140 (1993) L101. [8] J.R. Dahn, R. Fong and M.J. Spoon, Phys. Rev. B. 42 (1990) 6424. [9] J.R. Dahn, Phys. Reu. B, 44 ( 1991 ) 9170. [10] R. Fong, U. Von Sacken and J.R. Dahn, J. Electrochem. Soc., 137 (1990) 2009. [11] J.R. Dahn, U. Von Sacken, M.W. Juzkow and H. A1-Janaby, J. Electrochem. Soc., 138 ( 1991 ) 2207. [12] J.R. Dahn, J.N. Reimers, A.K. Sleigh and T. Tiedie, Phys. Rev. B, 45 (1992) 3773. [13] J.M. Tarascon and D. Guyomord, J. Electrochem. Soc., 138 (1991 ) 2864. [ 14] D. Guyomord and J.M. Tarascon, J. Electrochem. Soc., 139 (1992) 937. [ 15] N. Imanishi, H. Kashiwagi, T. Ichikawa, Y. Takeda, O. Yamamoto and M. Inagaki, J. Electochem. Soc., 140 (1993) 315. [ 16] T. lijima, K. Suzuki and Y. Matsuda, Denki Kagaku, 61 (1993) 1383. [ 17] M. lshikawa, M. Asao, M. Morita, Y. Matsuda, T. lijima and K. Suzuki, Denki Kagaku, 61 (1993) 1317. [ 18] T. Tamaki, Proc. 34th Battery Syrup., Hiroshima, Japan, 1993, p. 89.

T. lijima et al. / Synthetic Metals 73 (1995) 9-20

20

[19] M. Kikuchi, I. Tamura, Y. Ikezawa and T. Takamura, Proc. 34th Battery Symp., Hiroshima, Japan, 1993, p. 91. [20] J. Ebana, Y. Ikezawa and T. Takamura, Proc. 34th Battery Syrup., Hiroshima, Japan, 1993, p. 93. [21] M. Mohri, N. Yanagisawa, Y. Tajima, H. Tanaka, T. Mitate, S. Nakajima, M. Yoshida, Y. Yoshimoto, T. Suzuki and H. Wada, J. Power Sources, 26 (1991) 545. [22] T. Mitate, H. Tanaka, K. Yamada and M. Yoshikawa, Proc. 33rd Battery Symp., Tokyo, Japan, 1992, p. 199. [23] A. Mabuchi, C. Yamaguchi, H. Matsuyoshi, T. Kasuh and T. Maeda, Proc. 33rd Battery Symp., Tokyo, Japan, 1992, p. 191. [24] A. Mabuchi, K. Tokumitsu, H. Fujimoto, T. Kasuh and T. Maeda, Proc. 34th Battery Symp., Hiroshima, Japan, 1993, p. 77. [25] H. Abe, K. Zaghib, K. Tatsumi and S. Higuchi, Proc. 34th Battery

Syrup., Hiroshima, Japan, 1993, p. 7. [26] K. Zaghib, K. Tatsumi, H. Abe, S. Higuchi, T. Ohsaki and Y. Aawada, Proc. 34th Battery Symp., Hiroshima, Japan, 1993, p. 95. [27] K. Tatsumi, K. Zaghib, H. Abe, T. Ohsaki and Y. Sawada, Proc. 34th Battery Symp., Hiroshima, Japan, 1993, p. 97. [28] M. Sato, T. Iijima, K. Suzuki and K. Fujimoto, Proc. Syrup. Primary

and Secondary Lithium Batteries, Battery Division, Seattle, WA, USA, 1991, p. 407.

[29] R. Kanno, Y. Takeda, T. Ichikawa, K. Nakanishi and O. Yamamoto, J. Power Sources, 26 (1989) 535. [30] H. Azuma, A. Omaru and Y. Nishi, Proc. 51st Electrochemical Society of Japan, Fall Meet., Nagoya, Japan, 1991, p. 132. [ 31 ] M. Noguchi, K. Miyashita and M. Endo, Tanso, 155 (1992) 315. [32] N. Nakajima, A. Omaru, M. Nagamine, Y. Nishi and H. Azuma, Proc.

Electrochemical Society of Japan, FalI Meet., Yokohama, Japan, 1994, p. 225. [33] T. Ohsaki, A. Satoh and N. Takami, Proc. 34th Battery Symp., Hiroshima, Japan, 1993, p. 79. [34] K. Sato, M. Noguchi, A. Demachi, N. Oki and M. Endo, Science, 264 (1994) 556. [35] T. Takami, A. Satoh and T. Ohsaki, Science, 264 (1994) 81. [36] K. Takei, K. Kumai, T. lwahori, T. Uwai and M. Furusyo, Denki Kagaku, 61 (1993)421 (in Japanese). [37] S. Yata, H. Kinoshita, M. Komori, N. Ando, T. Kashiwamura, T. Harada, K. Tanaka and T. Yamabe, Synth. Met., 62 (1994) 153. [38] J.R. Dahn, A.K. Sleigh, H. Shi, J.N. Reimers, Q. Zhong and B.M. Way, Electrochim. Acta, 38 (1993) 1179. [39] H. Fujimoto, A. Mabuchi, K. Tokumitsu, K. Kasuh and N. Akuzawa, Carbon, 32 (1994) 193. [40] M. Inagaki, Tanso, 122 (1985) 114.