AC impedance study of the lithum electrode in propylene carbonate solutions

J EIectroanaL Chem. 180 (1984) 609-617 Elsevler Sequom S.A., Lausanne - Prmted in The Netherlands

AC IMPEDANCE STUDY CARBONATE SQLUTIONS

OF THE

PART

SURFACE

I. EFFECT

OF THE

LI.T-

605

ELECTRODE

PREPARATION

IN PROPYLENE

ON THE

INTl-XAL

IMPEDANCEPARAMET’ERS*

R.V. MOSHTEV

and B PURESHEVA

Cenrral Lnboraiory of Electrochemrcal Powea Sources. BulgarranAcademy of Sc*e-~ce.s,Softa iO30 (BuigarIa) (Recetved

4th Apn!

1984)

ABSTRACT The effect of several surface preparation methods (mechanical Impedance parameters of Lt electrodes in 1 M LiCIO,-propylene

and electrochetical) on the imttal ac carbonate soluttons bas been studted

with the aim of obtammg

rehable kmettc parameters of the Ll/Lt electrode reactton Mzchamcal cleaning of the Lt surface m a dry box brings about htgher reststdnce and lower capacttance owing to the passtve film on Lt formed m the box atmosphere Anodtc stripping at 4.6 mA/cm’ for 40-50 mm produces a very thin “amxhc” film, but at the same time strongly activates the crystal Li surface The low reststance m thts case ts assumed to approach the charge transfer resistance across the metal/solution interface The formation of a fresh Li surface by cut&g the electrode under the test solution provtdes a smeared, but film-free, surface, which reacts rapidly wtth the test solutton. The high capacitance value (15 pF/cml) obtamed after thts treatment IS assumed to approach the double-layer capacttance at the metal/solut:on interface The transfer coeffictent Q = 0 7 has been determmed and the effect of traces of water (5-50 ppm) in the test solutton on the Impedance parameters IS assessec.

INTRODUCTION

The kinetics of the Li,/Li+ electrode reaction in various non-aqueous electrolyte solutions is a basic topic in the electrochemistry of lithiux Owrng to its sigxxflcance in practical Li cells, the LiClO,/propylene

carbonate (PC) solutron has been mos: often used in such investigations [l-6]. A review of the pubhshed data on the kinetic parameters of the above electrode reaction in this solution [l] revealed a considerable dispersion. Inasmuch as the various electrochemical methods applied are quite reliable, this dispersion should be attributed to the different state of the Li surface at the moment of measurement. Since Li is a highly reactive metal, the preparaticn of a clean Li surface is a critrcal moment in any investrgation. The ideal preparation technique should provide complete elimination of the surface layers, a smooth metal * Dedicated to the memory of Professor Dr Dr. h. c Kurt Schwabe. 0022-0728/84/SO3.00

0 1984 E;sevter Sequoia S A

610

surface with a definite area, and a minimum lapse of time between the end of the preparation and the beginning of the measurement_ In some cases, the formation 01‘ active sites on the lithium crr’stallites is also desirable. Still, the early investigators of the simple electrochemical reaction of the active LJ electrode m the widely used LiCIO,/PC solutions Li + S = Li’ (S) + e-

(11 realized that the apparent kinetic parameters are affected significantly by the rapidly forming passive film [l-3]. Consequently, a number of methods, mechamcal and electrochemical, were applied to remove the film prior to the electrochemicaz measurements A comparison of the potentialities of the various electrochemical methods for the study of both the active and the passive Li electrode indicates a preference for the ac impedance method in view of its small perturbation signal fast performance and the posstbXty of assessing simultaneously the various processes involved in the electrode reactton. The first ac impedance measurement of a fresh LI surface in PC solutions was carried out by Epelboin et al. [S], later by Thevenin [S] and more recently by Garreau et al [9]. The authors, however, applied only anodic stripping for the Li surface preparation and stud& mainly the initial impedance parameters. In a recent paper, Povarov et al. [lo] studred the evolutron of the ac impedance of LI m the same solution. Their first measurenents, following mechanical cleaning in a dry box, were performed only after 6 h of storage at the open circuit potential_ In Part I of this serves, the effect of dtft?rent surface preparation methods on the initial impedance parameters ~111 be studied, which m the best case could reflect the kinetics of reaction (1) on a “Film-free” surfac?. In Part II [ll], the evolutton of the impedance parameters at the open circuit potential will be followed and the temperature dependence of these parameters till be studied in an attempt to eluctdate the properties of the passive film. EXPERIMENTAL

All-glass jacketed cells were employed in the experiments_ The test electrode holder enabled a smooth Li surface wrth a definite surface area (0.125 cm’) to be prepared. The Li wire (Merck) was strongly pressed in a composite tube made of a stainless-steel upper part and a lower insulating part of polypropylene, which is inert both to Li and the solvent. The reference electrode was of the same destgn. The counter-electrode was a Li foil pressed agamst a Ni screen cylinder with a geometric surface area of ca. 70 cm’. The test solutrons were prepared with vacuum-dried AR glade LICIO, and TBACIO, (Fluka and Merck respectively). Propylene carbonate (PC) (Merck) was dned over molecular sieves. The solutions were kept For 2 weeks over Li chips, which collect not only the residual water but also traces of reducrble contaminants, including oxygen. The moisture level, analysed by an automattc moisture meter (Mitsubishi CA-2), was usually kept below 10 ppm. All manipulations with salts, solutions, electrodes and cells were carried out in an Ar-flushed dry box, monitored by a Betatest HM 155 drgrtal hygrometer (Hz0 < 10 ppm).

611

The following surface preparation methods of the Li electrode were applied in the present study: (i) scraping with a stainless-steel blade in the dry box atmosphere and rapid (1-3 s) immersion in the test solution (SB); (ii) cutting in the dry box atmosphere with a specral gurllotine and rapid immersion in the test solution (CB); (iii) cutting under the solution in the test cell with a special device (CS); (iv) anodic stripping by galvanostatic polarrzation at 2.3 mA/cm’ for 75-90 mm or at 4.6 mA/cm2 for 45-50 mm, corresponding to lo-12 C/cm’ (AS); (v) cathodrc deposition of a thin Li coating at 0.46 r&/cm2 for 40-50 rain (ca. 3 pm) (CD). The ac Impedance measurements were perfolmed with a Solartron 1174 Transfer Functron Analyzer connected to a plotter and a printer. Measurements were carried out m the galvanostatrc mode with a two-electrode configuration. The cell was connected to the ac output via a selected resistor of 100 kfi, which also served as the standard resistance_ The impedance diagrams were plotted in the frequency range from 100 kHz to 0.5 Hz and in some cases down to 0.01 Hz, at 6-10 points per frequency decade. Most of the measurements were done at the open circuit potential (OCP). When necessary, galvanostatic polarization was applied by a small battery. RESULTS

AND

DISCUSSION

Frgures 1 and 2 present six typical mitral ac Impedance diagrams in the complex plane of LI electrodes m 1 M LKIOJPC solutrons at the OCP obtained after various surface preparation methods of the test electrode. It can be seen that in all cases a middle frequency arc (MF) appears, with its respective parameters: the resrstance on the real axrs R,, the characteristrc frequency fz*, the capacitance C, = (27if,R2)-l, and the depression angie a2_ The solution resistance R, found from the intercept of the MF arc at f = 00 on the real axis is subtracted from the real component values. According to refs. 5, 8 and 9, the MF arc reflects the charge transfer resistance R ,, coupfed in parallel with the double-layer capacitance C, at the solid/liquid interface. Following the AS preparatron (Fig. 2c), a well-defined low frequency (LF) arc appears, which has been recently interpreted by Thevenin [8] and Garreau et al. [9] as reflecting the slow diffusion of Li‘ ions in a porous polymer film. Contrary to the finding of Garreau et aL [9] for Lr electrodes stored for 1 day in the same solution, no Warburg line was observed in the low frequency range down to G-01 Hz, even after prolonged storage (200 h). Actually, the LF arc increased only moderately with time and was finally suppressed by the larger MF arc. A false Warburg line, due obviously to the impedance of the counter-electrode, was recorded in our early experiments, where instead of a LI foil cylinder we used a Ni screen cylinder as the counter-electrode. The latter was evrdently passivated in the course of the experiment. The mitial impedance spectra in the LiClO,/PC solutions do not reveal any high

612 TABLE

1.

Average mutualImpedance parameters of the La/Li+ a’ter various surface treatments Surface VW1 k’ f2* Q cm’ kHz treatment = ppm SB CB cs AS-2,3 mA/cm’ AS-4.6 mA/cm* CD-O.5 mA/cm’

5-25 5-25 5 5-15 5-15 10

75 11.2 118 12 65 11.8

42 3.3 090 1.8 22 1.4

electrode in 1 M LICIO,/

PC solEtio& at the OCD

G/

9/

RI/

A*/

5.1 4.3 15 74 11 9.6

14 14 17 13 11 16

<2 =G2 G2 4-5 2-3 2

30-50 30-50 30-40 5-10 5-7 14

pFcmb2

o

Stem*

Hz

a See text for notations

frequency (HF) arcs associated with the bulk resistance and the geometric capacl!snce of the film It appears, however, that vestiges of a HF arc tire exhibited in most cases (Figs. 1 and 2) by the asymmetry in the high frequency domains of the MF arcs, except for the arc in Fig. 2b.

IP= 23 mA/cmP .

.

AS

.

la=4.6mA/cm2 e *

..J. m

L=46

mA/cm2

R. Q .cmZ Fig. 1 Typical ac unpedance diagrams m 1 M LiCIO,/PC soluuons of LI electrodes after various surface preparatrons. (a) scrapmg m the dry box (SB); (b) cuttmg by a gudlotme m the dry box (CB); (c) cuttrng by a guillotme under the solution in the test cell (CS). Fag 2. Typical ac Impedance diagrams m 1 M LiCIO,/PC solutions of LI electrodes after anodi srnppmg (AS) in the fest cell, (a) After AS at 2 3 mA/cm’; (5) at the end of AS at 4 6 mA/cmz, (c) ai* current interruption

.

613

_The effect of the surface preparation method on the initial impedance parameters is demonstrated in Figs. 1 &d 2 and Table 1, which presents the average values from at least three independent measurements. As revealed by hght microscopic observations, the LI electrode surface is smoother after CB than after SB, this resulting in higher C, and lower R2 values in the case of CB (Table i). The C, values after both these treatments are, however, much lower than that expected-for the double-layer capacitance at a clean meta! surface in a solvent with a considerable. dielectric constant (e = 65). It was also established that increasing the water content in the solution from 2 to 20 ppm had practically no effect on the Initial unpedance parameters after both CB and SB. All these fmdings film, most probably of Ll,O, inevitably indicate the presence of a thin “native” formed on the Li electrode surface during its short exposure to the box atmosphere. The CS method is expected to yield the most reliable initial impedance parameters, inasmuch as the measurement can be started 1-2 s after the production of the fresh Li surface, avoiding its exposure to the box atmosphere. Consequently, in the solution contaming 5 ppm of water a considerably higher C, value is obtained (Table l), which is the closest approach to the double-layer capacitance at an almost film-free surface. On the other hand, the higher R, value obtained after CS needs some comment One possible reason is the smoother surface produced by the lubncating effect of the solution when the metal is cut. It might be argued that the larger R2 value is due to the hgher resistivity of the film obtained by the interaction of the virgin Li surface with the solution. As shown in Fig. 3, the initial characteristic

frequency fzO IS, _ in this case, strongly dependent on the insignificant rise in the concentration of water m the solution, which probably reflects the effect of water on

IM

I

1

I

LICLO~,/PC

a a .,-_

.

-

I

-1

I1

Fig 3. Dependence of the charactenstic frequencyfi* of Li electrodes prepared by CS on the concentration of water in the test solution

614

the specifr, conductivity of the film. All these findings suggest that the nature of the film after CS is different from that of the “native” one. The AS method of surface cleaning IS convenient yielding reproducible-results under defined potentrostatic or galvanostattc conditions. It allows for measurements immediately after, as well as during, the treatment_ A detailed evaluation of this method was performed by Garreau et al. [4], who pointed out that efficient anodic cleaning is achieved at anodic overpotentials below 200 mV at a charge 10 C/cm2. Higher over-potentials lead to anodrc film formation, while larger anodic charges c the metal surface. These conclusions are might result in appreciable etching 01 supported by SEM observations, which revealed that Li surfaces carefully treated by AS are smooth, film-free and exhrbit the features of Li crystallites [4]_ We studied the effect of current density (cd) applied during AS at over-potentials below 100 mV on the evolution of the impedance parameters. At lower cd (2.3 mA/cm2; Fig. 4), there is an initial rise of R2 and na followed by a slow decline_ The flat maximum of R, could result from the superposition of several processes: anodic dissolutron; native film disruptron; surface actrvation by revealing faces, edges, comers and point defects of the Li crystallities; all these opposed by film formation due to the interaction of the fresh metal surface with the solution_ It appears that during AS at 2.3 n-&/cm2 the last process is still prevailing over the first three, as a result of which the cleaning process is inefficient. This can be seen by companng the initial value of R2, 8 52 cm2, with the final one, 12 D cm2, after 10 C/cm2 (Fig. 4). The evolution of the ac impedance parameters during AS at 4.6 mA/cm’ is quite different (Fig. 5). There is a steep initial decay of R, and qa. which reach almost steady values after 10 C/cm2. The capacitance C, increases accordingly to attain 11

1M tiCLOb/ PC , H,O = 10 ppm Ia = 2.3 mA/cmz

1 2.5

I

I

5

7.5 4ajC

4. Evolution of the anodlc overvoltage test solution at 2 3 r+.,‘cmL

1 10

cm+

va and ac Impedance

parameters

R2 and C, dunng AS in tke

615

,uF/cm’ after the same charge_ In this case the rate of film formation is evidently exceeded by those of the three other processes, this resulting in a .-ore efficient anoidic cleaning. The evolution of R,, C, and qn after 10 C/cm2 is insigmficant ar?d

1

1

1

1M LIC104/PC

, H,O s 10ppm

r,=G6mA/cm2

5

10 q. /C

cmw2

15

Fsg. 5. Evolution at the anodic overvoltage q, and ac impedance parameters R, and c2 test solution at 4 6 rnA/cm2_

1

I

I

during AS in the

71

I

1=4.6mA/cm~

0

i

I

e

I 25

I 5.0

I

I

7.5

10

i

q,,/C cmm2 Fig. 6. Evolution at ac unpedance parameters R, and f,* during AS at 2 3 mA/cm2. were recorded 1 mm after the current interruption_

Impedance diagrams

616 reflects some additional activation or/and surface roug’leipgg. The relatively more complete elimination of the film at 4.6 &/cm’ is demonstrated by the shape of the MF arc at the end of the AS treatment, Just before the current interruption (Fig 2b). This arc is completely symmetrical, in contrast to all the other initial MF arcs in Figs. I and 2, which exhibit and appreciable asymmetry in their high frequency domains. It was found that the increase of water concentraZion From 5 to 50 ppm has practically no effect on the mitial impedance parameters after AS, contrary to what was established for electrodes prepared by CS (Fig. 3). The lower values of a2 after AS are probably related to a smoother and more homogeneous Ll surface after this treatment (Table 1). The low frequency (LF) arc after AS is larger than that after mechanicai cleaning of the electrode (Fig. 2~). Since the LF arc is not observed during anodic polarlzatlon, the ac impedance measurements had to be performed 1 min after the current interruption, during which the solution was vigorously stirred by dry Ar bubbling. Figure 6 presents the evolution of R, and f, * during the anodic treatment at 2.3 mA/cm’ revealing that R, reaches 4 Q cm2 after 1.5 h. For comparison, a mechanically cleaned electrode attains the same value of R, at the OCP after about 100 h. This result suggest that the the anodic polarization promotes the growth of thk LF arc and consequently the growth of an anodic film. Hence even under this mild anodic treatment ( qln= 40 mV) film formation is also operative. An attempt to interpret the LF arc more closely is presented in Part 2 of this Series [II]. Preliminary ac impedance measurements, carried out during cathodic deposition (CD) of Ll on the Li electrode in the test solution at a low cd (0.46 mA/cm2) revealed another possibility of preparing a clean metal surface. Observation of the Li electrde under a light microscope proved that under these conditions .t smooth, fine-grained Li coatkg is obtained. There is a rise of R2 dunng the initial stage of CD (up 70 1 pim), wluch is then followed by a linear decrease down to 12 Q cm2. The value of Cz remains invariable at 10 p’F/cm’, signifying the absence of -ouglmess effects. More systematic experiments are needed to explore this possibility of producing, by CD, a fresh and smooth Li surface for electrochemical measurements directly m the cell. The effect of ihe concentration of Li+ ions in the solution could give evidence as to how the initial impedance parameters reflect the electrode kinetics of reaction (1). Table 2 presents the experimental data for a Li electrode prepared by CB in two LiClO, solutions with tetrabutylammomum (TBA) perchlorate as the supporting

probably

TABLE 2 Concentration

dependenceof the rnltralimpedanceparametersof the LI/LI’

electrode

R2/

IO/

c2/

M

M

Sl cm2

m/i cm-’

FF cm”

0.1

09 0

38.0 7.7

0.68 33

8.3 7.3

CKIO.

1.0

/

cTBACIO,/

617

electrolyte. The considerable nse of R, wrth the ten-fold decrease of the concentration of Li+ ions is evidently not related to the formation of a thicker him since the change in C, is insignificant. The exchange cd, z,,. is esttmated frolrn R, on the assumption that the latter is equal to the charge transfer resistance R, = RT/Fz,. The approximate value of the transfer coefficient estimated from these data, (Y = (2, log i,/A log c> = 0.705 is m good agreement wrth previously reported values [1,3]_ These findings indicate that although the initial impedance parameters are more or less affected by the Impedance of the thin initial f&n. they still reflect the electrode kinetics of reaction (1). In conclusion rt appears that all the surface preparation methods explored in the present study have advantages as well as drawbacks. In the case of SB, CB and AS, thin films are formed during the treatment. The mechanical methods (CB and CS) produce smooth, but smeared, metal surfaces. The electrochemical treatment (AS, on the other hand, develops active centres, offering sites for a faster ion exchange at the interface, but at the same time promotes the growth of an anodic film. The CS method yields a clean Li surface, but the latter Interacts very raprdly with the solution. Nevertheless, it deserves to be explored further by coupling it with a very fast electrochemical measuring method. REFERENCES

1 J. Butter, P. CogIey and J. Synnot.J Phgys Chem , 73 (1969) 4026. 2 3 4 5 6 7

8 9 10 11

R Starr, J Electrochem Sot., 117 (i970) 295. S Merbuhr, J EIec:rochem. Sot .117 (1970) 56 M Garreau, J. Thevenm and D Warin, Progr Batt , Solar Cells, 2 (19793 54 1. Epelbom, M Garreau and J. Thevenin, Electrochem Sot Meetmg, Atlanta, Electrochem. Sac . Pennmgton, 1977. Extended Abstract No. 3. I A. Kednnsku, T.V. Kuznetsova, V P. Plekhanov. V A. Borsukov. and A.a Lysenko. 18 (1982) 965. R. Moshtev, J. Power Sourees, 1: (1984) 93. J. Tbevenm, C R Acad Set Paris, 235 (1982) 971. M Garreau, J. Thevenin and B Milandou, Electrochem Sot Meeting, Washmgton, Abstract No. 58 J.M. Povarov. L A Beketaeva and LA Vorobyova, Elektrokhtmtya, 19 (1983) 586 R.V Moshtev and B Puresheva, m preparatton

Vol

77-2.

The

Eiektrokhtmtya,

1983.

Extended