Composite solid electrolyte for Li battery applications

SOLID STATE

Solid State Ionics67 ( 1993) 51-56 North-Holland

IONICS

Composite solid electrolyte for Li battery applications G. N a g a s u b r a m a n i a n , A.I. Atria, G. Halpert Jet Propulsion Laboratory, California Institute of Technology, Pasadena, CA 91109, USA

E. Peled Tel Aviv University, Ramat Aviv, 69978 Tel Aviv, Israel

Received 1 April 1993;acceptedfor publication 15 August 1993

The electrochemical,bulk and interfacial properties of the polyethyleneoxide (PEO) based compositesolid electrolyte (CSE) comprising Lil, PEO, and A1203have been evaluated for Li battery applications. The bulk inteffacial and transport properties of the CSEs seem to strongly depend on the alumina particle size. For the CSE films with 0.05 micron alumina while the bulk conductivityis around 10-4 (mho cm-1 ) at 103°C, the Li ion transport number seemsto be closeto unity at the same temperature. Compared to the PEO electrolyte this polymer compositeelectrolyte seems to exhibit robust mechanical and inteffacial properties. We have studied three different filmswith three different alumina sizesin the range 0.01-0.3 micron. Effectsof A1203 particle size on the electrochemical performanceof polymer composite electrolytewill be discussed. With TiS2 as cathode a 10 mAh small capacitycellwas chargedand dischargedat C/40 and C/20 rates respectively.

1. Introduction Recent activities in solid polymer electrolyte/Li batteries have centered around the use of polyethylene oxide (PEO) and other organic polymers complexed with lithium salts as the electrolyte. While considerable and significant advances have been made in terms of improving the bulk ionic conductivity and interfacial properties of the polymeric electrolyte contacting Li [ 1 ] some basic and fundamental problems remain to be solved. The problems include: (a) Low transference number (0.30.5) for Li cations leading to high concentration polarization and high interface resistance; (b) incompatible salt anions (BF~-, AsF~-, C10~-, CF3SO~etc.) contained in the polymer that cause the lithium to degrade, and (c) poor mechanical strength especially above 100 ° C. Other schools have used A12Oa [2] or alumina based [3,4] to improve the mechanical strength of polymers. But, the lithium transference number was low because of an inappropriate salt, low salt concentration and too large A1203 partides. To overcome some of the shortcomings LiI was used in our studies. This salt is totally compatible

with Li. The idea embodied in this work is to coat the alumina particles with a thin layer of LiI and to bond these particles together with PEO. It was expected that this composite solid electrolyte (CSE) would retain the vacancy conduction mechanism for Li + (responsible for near unity transference number of lithium) reported earlier [5,6] and in addition would exhibit the flexible nature of the polymer. In this study we report our electrochemical results on the CSE comprising LiI, PEO, A1203 and as a function of alumina particle size.

2. Experimental Initially there were problems dispersing uniformly the A1203 particles in the PEO matrix. Thus a modified procedure, developed at the Jet Propulsion Laboratory, was adopted. Appropriate amounts of LiI, A1203, and PEO were weighed separately. LiI was dissolved in 50 ml of acetonitrile, decanted and to the solution was added alumina. The solution was stirred well for 45 min. 80 ml of Isopropyl alcohol (IPA) was added and the solution again was stirred

0167-2738/93/$ 06.00 © 1993Elsevier SciencePublishers B.V. All rights reserved.

52

G. Nagasubramanian et a l . / L i battery applications

well. To this mixture was added approximately 120 ml of acetonitrile followed by another 80 ml of IPA. While the solution was being stirred vigorously, 1.6 g of PEO(M.W. 4 × 106) was added slowly . This procedure produced a uniform suspension of alumina in the solution. The mixture was stirred overnight to dissolve the PEO. This solution was cast into thin films as described elsewhere [7]. Our initial procedure to prepare thin films of CSE without the addition of IPA gave us only lumps of alumina covered with a coat of PEO. The thin films of CSE prepared by our modified procedure were subjected to a series of electrochemical measurements including ac and dc measurements. Both a symmetrical cell of the type Li/CSE/Li and an asymmetrical cell of the type Li/CSE/SS (stainless steel) were used for the electrochemical characterization of the CSE films. With TiS2 as the cathode a small capacity cell was made and charge/discharge studies were made.

BULK CONDUCTIVITY OF THE CSEs. EFFECT OF AI203 PARTICLE SIZE (0.05 MICRON/0.03 MICRON (~-AI203) 5O

I

]

I

0.05 MICRON E

I

I

0~

I

I 100 Z' (ohms)

3O

I 0.30 MICRON

F4

I

[]

rl []

[]

[]

090

1

150 Z' (ohms)

3. Results and discussion

Fig. I. Nyquist plot of the CSE.

3.1. Bulk conductivity and interfacial charge transfer resistance G-1 = PEO+AI203+Lil

Both the bulk conductivity (1/Rb) and the interfacial charge transfer resistance (Rot) of the electrolyte (CSE) were determined from the ac measurements. The ac measurements were made in the frequency regime 100 kHz-5 Hz. A typical Nyquist plot is shown in fig. l for CSE films containing 0.05 and 0.3 ~tm alumina. While the high frequency intercept on the x-axis is the bulk resistance of the electrolyte, the corresponding low frequency intercept gives the combination of the bulk resistance of the interracial layer (omnipresent on the Li surface) and the charge transfer resistance, which we defined earlier as R~t. The CSE film containing 0.3 ~tm A1203 exhibits three different regimes dominated by bulk processes at high frequencies followed by charge transfer processes at medium frequencies which in turn is followed by diffusional processes at low frequencies. However, the CSE films with 0.05 lim AI203 exhibit almost resistor like behavior where the contributions from the charge transfer and diffusional processes are insignificant. The ac characteristics of the CSE films with 0.3 ~tm A1203 are typical

1 E-2 7

f

I

I

t

~ 1E-3

~

"

~

•

Li/G-1/Li

1E-4

O 1E-5

I

z

O 1E-6 O 1 E-7 2.40

I 2.60

II 2.80

" I 3.00

"7 3.20

I 3.40

3.60

l/T, 103 K

Fig. 2. Conductivity versus 1 / T o f t h e CSE.

of systems where the transport number of the reversible ion is very low. Hence we have not studied this material in depth. The behavior of CSE films with 0.01 ~tm Al2Oa is similar to that of 0.05 lira A1203 film. In what follows we have described our results for CSE films with 0.05 A1203 only. In fig. 2 is shown the plot of the bulk conductivity of the CSE as a function of the reciprocal temperature. The data indicate that while the CSE exhibits a very modest

G. Nagasubramanian et al. / Li battery applications conductivity below 79 °C above this temperature the conductivity increases. Further, the temperature (79°C) at which the break occurs is higher than for PEO without the alumina. For PEO system without alumina the break in conductivity occurs around 60 ° C. The interfacial charge transfer resistance appears to be stable over a period of many days.

53

the mobility of positive and negative ions respectively and R b is the bulk resistivity of the electrolyte and Zd represents the impedance due to ion diffusion. This can be obtained as follows. The total impedance Zt (comprising Of Rb+Rct+Zd; where Rot is the charge transfer resistance), of the system, is computed from the potentiodynamic measurement. From this total impedance, Zt, subtracting Rb+Rct gives Zd. AC and potentiodynamic measurements were made on the cells of the type Li/CSE/Li at 105°C. While Rb is close to 73 fl Rot and Zd respectively are 2.5 and 2 ft. The cation transport number determined from these results is close to unity. This is a bench mark result that has not been reported in the literature yet. In table 1 our electrochemical data are compared with the data available in the literature for comparable systems. The data indicate that not only the transport number is higher but the Ret is lower for our system compared to stateof-the-art systems.

3.2. Transport number The ratio of the individual currents to the total current represents transport number. When an univalent salt comprising of two constituents M + and X - is dissolved in a solvent the salt dissociates in to M ÷ and X - in an ideal case. Under this condition the cation transport number corresponds to t+ and the anion transport number corresponds to t_. For the computation of the transport number we have made this assumption in our system. Several authors have developed different techniques to experimentally determine the transport number of the mobile ions [ 8-11 ]. The simplest method developed by Sorensen and Jacobsen [8], involves identifying the contributions of different processes to the total conductivity by ac measurement as a function of frequency in a symmetrical cell comprising two reversible electrode separated by an electrolyte. From the ac results the transport number of the reversible ion (in this study Li ion) can be computed as t+ =u~/ ( u l + u 2 ) = R b / ( R b + Z d ) where ul and u2 represents the mobility of positive and negative ions respec-

3.3. Thermal creep measurement The resistance of any material is directly proportional to its thickness. So by monitoring the resistance of the material as a function of time, in principle, at least qualitatively one can estimate the dimensional stability of the material using the formula c r e e p % = ( R - R i ) / R i X l O 0 , where creep% represents the percentage change in thickness and R i and R respectively are the initial resistance and resis-

Table I Comparison of electrochemicalpropertiesof CSE films. Materials compared

(LiI)t (PEO)3 (A1203)o.3 ( LiI ) t (PEO ) 1.65 (A1,%)o.39

Temp. (°C )

116 90 103

Bulk

t+

conductivity (mho cm- i ) 6XI0 -4 2X10 -4 10-4

Interface resistance

(~Icm2) 0.8_+0.05 0.9_+0.05 1_+0.05

2.5 10 25

Prior art electrolytematerials (PEO)aNaI, 10% A1203 [ 3 ] ( PEO ) aLiC104 10%A1203 [2] (PEO)4.5 LiSCN [8 ]

120

3× l 0 - 4

118

10- 3

0.22

25

115

10-4

0.5

72

G. Nagasubramanian et al. / L i battery applications

54

Table 2 Peak splitting as a function of state-of-charge. Temperature (°C)

Condition ~)

114 115 115 115 115 115

~(ffesh) fd pd fd pc ~

Initial OCV

Anodic peak voltage V.

Cathodic peak voltage Vc

(v)

(v)

(v)

2.700 2.305 2.215 1.810 2.450 2.800

2.143 2.106 2.087 1.696 2.069 2.139

2.620 2.635 2.688 2.300 2.578 2.584

pv.-Vcl

0.477 0.529 0.601 0.604 0.509 0.445

a) fc = fully charged; pc = partially charged; pd = partially discharged; fd = fully discharged.

100

f

CATHODE CAPACITY 10 mAh D-C CYCLIC VOLTAMMETRIC CHARACTERISTICS

I

CREEP % = { ( R - R i ) / R i } xl00 -500

80

I

I

]

^

[

Li/G5/Li; G5 = PEO, Lil o~

60 -360

Q. UJ UJ

o

40

-120 Li/G4/Li; G4 = PEO, Lil, AI20 3

20

(.3 O ~

~o I

40

v

-[

J

80

120

120

160

TIME (HRS.)

Fig. 3. Creep% versus time of PEO/LiI with and without A1203.

tance as a function of time. In fig. 3 we have shown a typical plot of creep% as a function of time for two different polymer electrolytes: one containing alumina (CSE) and the other without. Our results indicate that this CSE is much more dimensionally stable than the PEO/LiI electrolyte. 3.4. Studies on L i / C S E / T i S 2 cells

A 10 mAh small capacity cell was made with TiS2 as cathode and dc cyclic voltammetric measurements were made at several DOD as a function of open circuit voltages (OCVs). The cells were predischarged prior to CV measurements. In fig. 4 is shown a typical dc cyclic and in table 2 is shown the

360 Li/CSE/TiS2 AT 105°C OCV = 2.305

500 -1700

I -1960

I -2220

I -2480

I -2740

-3000

E (MV)

Fig. 4. D-cyclic voltammetric characteristics of Li/CSE/TiS2 cell. Scan rate 1 mV/s.

peak splitting as a function of OCV. The well defined cathodic and anodic peaks indicate that Li ÷ moves in and out of the TiS2 cathode (the cell can be charged and discharged). The peak splitting increases with decrease in OCV of the cell which may be related to the increase in resistance of TiS2 cathode with lithiation. In fig. 5 is shown the plot of apparent diffusion coefficient of Li + in the cathode and charge

G. Nagasubramanian et al. /Li battery applications DIFFUSSION COEFFICIENT (Do) AND Rct AS A FUNCTION OF OCV CELL CONFIGURATION: Li/CSE/TiS 2 10-6

I

I.Z LU

~

I

)

I

~

I

~

CHARGE/DISCHARGE STUDIES Li/CSE/TIS 2 (10 mAh) 2900

I

LOG DIFFUSION COEFFICIENT vs. OCV

2

$

I

I

I

2620

~J 10-7 0 0

C/20

.,--"-

2340

Z

_o ~ 10.8 o

W

2050

C/40

,

10-9

1780

50O

'

I

'

I

I

I

~

|

CHARGE TRANSFER RESISTANCE vs. OCV

150C

8 Z

I 3000

~< 400

I 6000

I 9000

I 12000

15000

SEC

03 iii

Fig. 6. Charge-discharge characteristics of Li/CSE/TiS2 cell.

rr 300 ,-r.
55

100

0

1.8

2.0

2.2 2.4 2.6 OPEN CIRCUIT VOLTAGE

2.8

3.0

Fig. 5. Diffusion coefficient and R~t versus OCV.

transfer resistance at the TiS2 electrode as a function of OCV. A brief description of the method used to compute the apparent diffusion coefficient (Davv)is given below. The D~pp was computed from ac measurement as a function of frequency at a fixed potential. The computational procedure is the same as described in ref. [ 12] and a brief description is given below. The slope (tr) of the linear portion of the plot of real component (resistance from ac) as a function of (09)- 1/2 (where 09 is the angular frequency) contains diffusion coefficient of the ions, (both cations and anions) and is given by the following equation: cr=RT/n2F2x/~{ 1/r~a/2r**/'vO"-'0+ 1/D1/2C~) where the symbols have the usual meaning. Since only the Li + moves in the TiS2 lattice the above equation reduces to a=RT/nEF2x/~{1/D~/2C~}. In this equation the only unknown is the Do (cm2/s) which represents the apparent diffusion coefficient of Li ions.

While the apparent diffusion coefficient goes through a maximum at around 50% state-of-charge the Rct values vary randomly with OCV. A similar observation was made earlier, where the apparent diffusion coefficient peaks at 50% state-of-charge, for TiS2 cathode in organic electrolytes. In fig. 6 is shown the charge-discharge characteristics of the above cell at 105°C. The cell was discharged at C/20 and charged at C/40 rates. Although the transport number for Li is close to unity the charge-discharge rates are very low. There could be two explanations for the observation: (a) the CSE bulk ionic conductivity is still very low by an order of magnitude than the required minimum of 10 -3 S c m - 1 to achieve higher rates and (b) high contact resistance developed between the solid electrodes and the CSE.

4. Conclusions At the Jet Propulsion Laboratory we have developed a new chemical technique to disperse alumina uniformly in the high molecular weight PEO matrix. Our composite solid electrolyte (CSE) exhibits the highest transport number reported yet in the polymeric electrolyte for Li+. The conductivity of the CSE at 103°C is 10 -4 mho c m - L Both the transport number and ionic conductivity are influenced by the particle size of alumina. Our thermal creep measurement studies show that the CSE is dimensionally

56

G. Nagasubramanian et aL / Li battery applications

stable much more than the PEO/LiI electrolyte. The cells could be charged and discharged only at very low rates although the cation transport number is very high.

Acknowledgement The work described here was carried out at the Jet Propulsion Laboratory, California Institute of Technology under contract with the National Aeronautics and Space Administration.

References [1]J.R. MacCallum and C.A. Vincent, eds., in: Polymer Electrolyte Reviews 1 and 2 (Elsevier, London, 1987 and 1989). [ 2 ] J.E. Weston and B.C.H. Steele, Solid State Ionics 7 ( 1982 ) 75.

[ 3 ] F. Croce, F. Bonino, S. Panero and B. Scrosati, Philos. Mag. 59 (1989) 161. [4 ] F. Croce and B. Scrosati, J. Power Sources 43-44 (1993) 9. [ 5 ] J.H. Kennedy, in: Comprehensive Treatise of Electrochemistry, eds. L O'M. Bockris, B.E. Conway, W. Yeager and R.E. White, Vol. 3 (Plenum Press, New York, 1981). [6] C.C. Liang, J. Electrochem. SOc. 120 (1973) 1289. [ 7 ] G. Nagasubramanian and S. DiStefano, J. Electrochem. SOc. 137 (1990) 3830. [ 8 ] P.R. Sorensen and T. Jacobsen, Electrochim. Acta 27 (1982) 1671. [9] M. Leveque, J.F. Le Nest, A. Gandini and H. Cheradame, Makromol. Chem. Rapid Commun. 4 (1982) 97. [ 10] A. Bouridah, F. Dalard, D. Deroo and M.B. Armand, Solid State Ionics 18/19 (1986) 287. [ 11 ] S. Bhattacharia, S.W. Smoot and D.H. Whitmore, Solid State lonics 18/19 (1986) 306. [ 12 ] A.J. Bard and L.R. Faulkner, in: Electrochemical Methods. Fundamentals and Applications (Wiley, New York, 1980) p. 327.