Electrochimica Acta 45 (2000) 1481 – 1490 www.elsevier.nl/locate/electacta
Transport and interfacial properties of composite polymer electrolytes G.B. Appetecchi, F. Croce, L. Persi, F. Ronci, B. Scrosati * Department of Chemistry, Electrochemical Section, Uni6ersity ‘La Sapienza’, P. le A. Moro 5, I-00185 Rome, Italy Received 2 November 1998; received in revised form 5 April 1999
Abstract Lithium polymer electrolytes formed by dissolving a lithium salt LiX in poly(ethylene oxide) PEO, may find useful application as separators in lithium rechargeable polymer batteries. The main problems, which are still to be solved for a complete successful operation of these materials, are the reactivity of their interface with the lithium metal electrode and the decay of their conductivity at temperatures below 70°C. In this paper we demonstrate that a successful approach for overcoming these problems, is the dispersion of selected ceramic powders in the polymer mass, with the aim of developing new types of composite PEO – LiX polymer electrolytes characterized by enhanced interfacial stability, as well as by improved ambient temperature transport properties. © 2000 Elsevier Science Ltd. All rights reserved. Keywords: Composite polymer electrolytes; Nanocomposite; Interfacial properties; Conductivity
1. Introduction The well known lithium polymer electrolytes formed by the dissolution of a lithium salt LiX in poly(ethylene oxide) PEO [1] are of practical importance, since they may find important application as separators in lithium rechargeable polymer batteries, LPBs. Indeed, large LPB research and development programs are presently in progress in North America and Europe [2–4]. However, although successful materials, PEO–LiX polymer electrolytes are still affected by some problems, namely (i) a reactivity towards the lithium metal electrode; and (ii) a low conductivity at ambient temperature. Thus, work is needed to find the proper approaches for solving or at least minimizing these drawbacks. Indeed, the optimization and control of the electrode/polymer electrolyte interface is a key requisite for success in LPB development. Similarly, the enhancement of conductiv* Corresponding author. Tel./fax: +39-6-4462866. E-mail address:
[email protected] (B. Scrosati)
ity to acceptable values at ambient temperature is of obvious importance for widening the application scopes of the polymer electrolytes. In this paper we demonstrate that the dispersion of selected ceramic powders in the polymer mass produces composite PEO – LiX polymer electrolytes showing consistent improvements in both interfacial and transport properties.
2. Experimental
2.1. Synthesis The procedures used for the synthesis of the composite polymer electrolytes were discussed in detail in previous works [5,6]. The basic goal was that of obtaining liquid-free, ceramic-added PEO-based composite electrolytes. This was achieved by a modification of the hot-pressing technique originally proposed by Vincent and co-workers [7]. Proper amounts of highly purified poly(ethylene oxide) PEO (either Aldrich Chemical Co.
0013-4686/00/$ - see front matter © 2000 Elsevier Science Ltd. All rights reserved. PII: S 0 0 1 3 - 4 6 8 6 ( 9 9 ) 0 0 3 6 3 - 1
1482
G.B. Appetecchi et al. / Electrochimica Acta 45 (2000) 1481–1490
or BDH Polyox WSR301, both 4 000 000 average molecular weight), LiCF3SO3 (Aldrich R.G Selectipur) or, alternatively, lithium tetrafluoborate, LiBF4 (Merck Selectipur, battery grade) and low particle size (about 4 mm) gLiAlO2 (Cyprus Foote Mineral Co., HSA-10) were intimately mixed, heated and hot-pressed at 110°C. After cooling in liquid nitrogen, the membranes were dried at 55°C under vacuum for at least 48 h and stored in an argon atmosphere dry box. This method produced homogeneous, mechanically stable electrolyte membranes of 200 mm average thickness and with a composition of either PEO20LiCF3SO3 +20w/o gLiAlO2 or PEO20LiBF4 +20w/o gLiAlO2 composition (w/o is the weight percent). The nanocomposite polymer electrolytes were obtained by combining PEO with LiClO4 and using TiO2, Al2O3 or SiO2, respectively, as the ceramic filler. The LiClO4/PEO concentration ratio was fixed to 1/8 concentration and the amount of added ceramic to 10% of the total PEO8 –LiClO4 weight. The preparation of the nanocomposite electrolyte samples involved first the dispersion of the selected ceramic powder and of the LiClO4 lithium salt in acetonitrile, followed by the addition of the PEO polymer component and by a thorough mixing of the resulting slurry. The homogeneous slurry was then cast between two glass plates provided by spacers to set thickness, to finally yield mechanically stable membranes of average thickness of about 150 mm. The same casting procedure was also used to prepare ceramic-free PEO–LiClO4 samples for comparison purpose.
The lithium electrode – polymer electrolyte interfacial stability was evaluated by monitoring the impedance response of cells again formed by sandwiching the given electrolyte sample between two lithium electrodes. These cells were stored under open circuit conditions and their impedance response was analyzed using the cited fitting program [8]. This procedure allows the separation of contributions from the various phenomena which are observed, to determine the interfacial resistance. The interfacial stability and the associated cyclability of the lithium electrode were further studied and evaluated by imposing galvanostatic pulses on a two-electrode cell using a finely polished nickel working (substrate) electrode, a lithium counter electrode, and the selected composite membrane sample as the electrolyte. First, a known amount of charge (deposition charge, QD) was passed through the cell in order to promote the ongoing of lithium deposition on the nickel substrate. Then, a fraction of this charge (cycling charge, QC) was alternately cycled across the cell to promote lithium deposition-stripping cycles and the lithium stripping overvoltage was monitored upon cycling. The test was assumed to be completed when a significant increase in overvoltage was detected. The mean value of the lithium electrode cycling efficiency, h, was finally calculated by using the equation: h=
nQC nQC + QD
(1)
where n represents the total number of cycles the cell has undergone. The experiment was run and controlled using a MACCOR 4000 battery cycler.
2.2. Characterization The various composite polymer electrolyte samples were characterized by conductivity, lithium ion transference number and compatibility with the lithium metal electrode. The conductivity was measured by placing the given samples in a two stainless-steel electrode cell of 1 cm2 active area The measurements were carried out both in heating and cooling scans using a Solartron Frequency Response Analyzer (Mod 1255) over a 0.1 Hz–100 kHz frequency range. The data were processed by using an appropriate fitting program [8]. The lithium transference number, T+ LI, was evaluated using two independent methods, i.e. the method proposed by Vincent and co-workers [9] and the refined method proposed by Abraham and co-workers [10]. According to these methods, the T+ Li values were determined by imposing ac and dc polarization pulses to cells of the Li/electrolyte sample/Li) type and by following the time evolution of the resulting current flow. To improve the accuracy of the results, we have developed a special software capable of acquiring a very large number of current data per second immediately after the application of the voltage pulse.
3. Results and discussion Fig. 1 shows the Arrhenius plot of the PEO20LiCF3SO3 + 20w/o gLiAlO2 and of the PEO20LiBF4 + 20w/o gLiAlO2 polymer electrolyte samples. It may be clearly seen that the conductivity curves of both samples show a break around 60 – 70°C due to the well-known crystalline-amorphous transition of the PEO component [1]. Useful conductivity (i.e. in the 10 − 4 S cm − 1 range) was generally obtained at temperatures higher than ambient and typically around 80 – 90°C. Thus, this was assumed as the testing temperature range for monitoring the stability of the lithium electrode/polymer electrolyte interface. As well known, the reactivity of the lithium electrode can affect this interface due to uncontrolled passivation phenomena which result in the formation of thick and not-uniform surface layers [11]. These layers may in turn cause uneven lithium deposition in the course of the charge process, this leading to dendritic growth and eventually to cell short-circuiting. Therefore, the investigation of the lithium interfacial characteristics is of
G.B. Appetecchi et al. / Electrochimica Acta 45 (2000) 1481–1490
particular importance for the practical evaluation of any given electrolyte. In the case here under study this investigation has been carried out by monitoring the impedance response of symmetric Li/sample/Li cells kept under an open circuit condition. Typical results, which refer to a cell using the two polymer electrolyte samples here considered, are shown in
1483
Figs. 2 and 3, respectively. One can first notice that the amplitude of the middle-frequency semiarc does not expand consistently by time and this is a preliminary indication of a good stability of the interface. By fitting the evolution of the impedance responses with a proper equivalent circuit [8], one can then refine the analysis to finally obtain the value of the
Fig. 1. Arrhenius plots for PEO20LiCF3SO3 + 20w/o gLiAlO2 and the PEO20LiBF4 +20w/o gLiAlO2 polymer electrolyte samples.
Fig. 2. Impedance response of a Li/LiCF3SO3 + 20w/o/Li cell at progressive storage times and at 90°C. Frequency range: 10 mHz–100 kHz. Electrode surface: 0.5 cm2.
1484
G.B. Appetecchi et al. / Electrochimica Acta 45 (2000) 1481–1490
Fig. 3. Impedance response of a Li/PEO20LiBF4 + 20w/o/Li cell at progressive storage times and at 90°C. Frequency range: 10 mHz–100 kHz. Electrode surface: 0.5 cm2.
Fig. 4. Charge transfer resistance Rct, and passivation layer resistance Rpl of the lithium/PEO20LiCF3SO3 +20w/o gLiAlO2 composite polymer electrolyte interface monitored upon time at 90°C. Data obtained by impedance spectroscopy.
interfacial resistance-parameters and their change with time. The latter are shown in Figs. 4 and 5. Clearly, the passivation layer resistance remains in both cases at
very low values over prolonged length of storage time, this confirming an unique stability of the lithium electrode interface.
G.B. Appetecchi et al. / Electrochimica Acta 45 (2000) 1481–1490
The explanation of this exceptional interfacial stability may be obtained by considering the particular structure and morphology of the composite polymer electrolytes developed in this study. In fact, it is well established that the type and growth of the lithium passivation layer are unpredictably influenced by the presence of liquid components and/or liquid impurities in the polymer electrolyte [12]. The liquid phase decomposes at the surface of lithium, thereby severely affecting the cyclability of the lithium electrode. Therefore, it is expected that consistent improvement in interfacial stability may be obtained using liquid-free, polymer electrolytes. The electrolytes developed in this work fulfill this condition since they are prepared by a dry hot-pressing technique. Furthermore, it has been shown in our and in other laboratories [13–16], that the addition of selected ceramic powders to the polymer electrolyte mass greatly reduces the growth rate of the lithium passivation, probably because the ceramic powders trap traces of residual impurities. These ceramics also favors the formation of compact thin passivation layers on the lithium electrode surface. All these beneficial effects have been exploited since our composite electrolytes have been prepared by dispersing finely divided gLiAlO2 powders in the PEO–LiCF3SO3 and in the PEO20LiBF4 matrices. Therefore, one may conclude that the combination of the solvent-free preparation procedure, so as to eliminate liquids, with the addition of ceramic powders, so as to trap any remaining traces liquid impurities, may
1485
account for the large enhanced interfacial stability of the new types of composite polymer electrolytes here discussed. It is also reasonable to assume that in these dry composite polymer electrolytes the passivation process may basically involve a reaction between the lithium metal and the anions of the lithium salt with the formation of a thin, compact, inorganic-type layer which should be the most favorable in assuring a high lithium cyclability. To confirm the latter assumption, we have tested the cycling efficiency of the lithium electrode in cells using a dry composite polymer sample as the electrolyte, nickel as the working substrate and lithium as both the counter and the reference electrode. Following the procedure described in Section 2, the cycling efficiency was evaluated by monitoring the overvoltage of the lithium stripping process upon lithium deposition-stripping cycles. Figs. 6 and 7 show typical results. The trend of Fig. 6 reveals that the lithium deposition-stripping processes in the PEO – LiCF3SO3-based electrolyte extend for more than 120 cycles with low overvoltage values. The lithium cycling efficiency, calculated by using Eq. (1), is of the order of 91 – 96%. Even higher efficiency values are obtained in the PEO – LiBF4-based electrolyte were the lithium deposition-stripping process may be extended to 700 cycles (see Fig. 7), this giving a lithium cycling efficiency of the order of 98.6%, i.e. to our knowledge the highest value so far ever reported for PEO-based polymer electrolyte cells. Although these dry composite electrolytes have an extraordinary high interfacial stability, their conductiv-
Fig. 5. Charge transfer resistance Rct, and passivation layer resistance Rpl of the lithium/PEO20LiBF4 +20w/o gLiAlO2 composite polymer – electrolyte interface monitored upon time at 90°C. Data obtained by impedance spectroscopy.
1486
G.B. Appetecchi et al. / Electrochimica Acta 45 (2000) 1481–1490
Fig. 6. Lithium stripping overvoltage upon galvanostatic cycling of a lithium metal electrode in a PEO20LiCF3SO3 +20w/o gLiAlO2 composite polymer electrolyte cell. Ni substrate. QD = 1C; QC =0.1C. Temperature: 90°C. Current density: 0.1 mA cm − 2.
Fig. 7. Lithium stripping overvoltage upon galvanostatic cycling of a lithium metal electrode in a PEO20LiBF4 +20w/o gLiAlO2 composite polymer electrolyte cell. Ni substrate. QD = 1C, QC =0.1C. Temperature: 90°C. Current density: 0.1 mA cm − 2.
ity reaches useful values only at medium-high temperatures (see Figs. 1 and 2), and this may limit their applicability. Therefore, it would be quite useful to lower to the ambient region the temperature of operation of the electrolytes. Large research efforts have been devoted in the past to reach this goal for the general class of the PEO–LiX polymer electrolytes. The most commonly used approach has been the addition of liquid plasticizers, e.g. low molecular weight
polyethylene glycols or aprotic organic solvents, to the PEO – LiX matrix. However, the addition of liquids beside resulting in a deterioration of the electrolyte’s mechanical properties, greatly increases its reactivity towards the lithium metal anode (see above). Therefore, the ideal achievement in electrolytes of the PEO – LiX type would be the enhancement of low temperature ionic conductivity by modifications which avoid any liquid contamination. This goal is not easily achievable
G.B. Appetecchi et al. / Electrochimica Acta 45 (2000) 1481–1490
since fast ion transport in PEO–LiX is a characteristic of the amorphous state which is intrinsically reached above 70°C or artificially induced at lower temperature by the addition of liquid plasticizers. We have considered to solve this problem by extending our composite electrolyte approach to the use of selected ceramic fillers, e.g. TiO2, Al2O3 or SiO2, at the nanoscale particle size. The idea is that these fillers may act as solid plasticizers, capable of enhancing the composite polymer electrolyte’s transport properties without affecting its mechanical and interfacial stability [6]. Fig. 8 shows the conductivity Arrhenius plots of a representative examples of these novel nanocomposite polymer electrolytes; also, the plot of a ceramic-free PEO8 –LiClO4 polymer electrolyte is reported for comparison purposes. The heating scan of the latter shows a break around 60°C, reflecting the above mentioned transition from the crystalline to the amorphous PEO state, which is accompanied by a relevant increase in ionic conductivity. The trend of the curve is reproduced in the following cooling scan, this confirming that, when cooled back below 70°C, the ceramic-free, common PEO polymer electrolyte tends to recrystallize and
1487
consequently, the conductivity to decay. The as-prepared nanocomposite electrolytes, e.g. the Al2O3-based one, have a room temperature conductivity and a first heating scan similar to that of the ceramic-free electrolyte. However, the behavior of the following cooling scan is quite different since no break occurs around 60 – 70°C and the conductivity remains consistently higher, i.e. comprised between 10 − 3 and 10 − 5 S cm − 1 (versus 10 − 4 and 10 − 8 S cm − 1) in the 80 – 30°C temperature range. This conductivity trend is reproduced in the following heating and cooling scans. These results are convincing in demonstrating the validity of the nanocomposite approach in enhancing the conductivity of the polymer electrolytes. Further support is provided by Fig. 9 which shows that the low temperature conductivity of the nanocomposite polymer electrolytes (here the TiO2-based sample) remains stable for a prolonged length of time. It is then reasonable to conclude that the favorable transport behavior is indeed an inherent feature of the nanocomposite structure. A possible explanation is that, once the composite electrolytes are annealed at temperatures higher than the PEO crystalline to amor-
Fig. 8. Arrhenius plots of the conductivity of the nanocomposite PEO8LiClO4·10w/o TiO2, PEO8LiClO4·10w/o SiO2 and PEO8LiClO4·10w/o Al2O3 polymer electrolytes. The plot of a ceramic-free PEO – LiClO4 polymer electrolyte is also reported for comparison purposes. Data obtained by impedance spectroscopy measurements.
1488
G.B. Appetecchi et al. / Electrochimica Acta 45 (2000) 1481–1490
Fig. 9. Conductivity of the nanocomposne PEO8LiClO4·10w/o TiO2 polymer electrolyte monitored upon time at 31°C. Data obtained by impedance spectroscopy measurements.
phous transition (i.e. above 70°C), the ceramic additive, due to its large surface area, prevents PEO chain reorganization with the result of freezing at ambient temperature a high degree of disorder which is likely to be accompanied by a consistent enhancement of the ionic conductivity. Accordingly, one may assume that the structural modifications are induced via Lewis acid–base reactions between the ceramic surface states and the PEO segments, as in fact already proposed by Wieczorek et al. [17,18]. According to this model, the Lewis acid character of the added ceramics would compete with the Lewis acid character of the lithium cations for the formation of complexes with the PEO chains. Thus, the ceramics would act as cross-linking centers for the PEO segments, this lowering the polymer chain reorganization tendency and promoting an overall structure stiffness. Such a structure modification would provide Li+ ions conducting pathways at the ceramics surface, this accounting for the improvement in ionic transport. We have undertaken a series of complementary tests to support this model. An important one was the determination of the lithium ion transference number T+ Li which, according to the model, should be consistently higher in the nanocomposite electrolytes than in ceramic-free electrolytes. In fact, if the action of the ceramic filler is that of promoting conducting pathway s as a result of its Lewis acid type interactions with the PEO chains, the lithium ions are expected to move freely along these ceramic surface pathways and thus, a consistent enhancement of their transference number is expected. Indeed, the results have been consistent in demonstrating an increase in T+ Li when passing from the ceramic-free to the nanocomposite polymer electrolytes [6]. In addition and to further confirm the model, the T+ Li values consistently vary according to the Lewis acid character of the added ceramic. For the most acidic
type, i.e. the TiO2 one, transference number values of the order of 0.5 – 0.6 in the 45 – 90°C temperature range were obtained. To our knowledge, such a high value has never been reported for common PEO-based electrolytes. Another important test was the control of the crystallization kinetics of the nanocomposite samples. Figs. 10 and 11 compare the differential scanning calorimetry, DSC, heating – – cooling traces of a ceramic-free polymer electrolyte with those of an Al2O3-based nanocomposite electrolyte. The heating scan of an as-prepared ceramic-free PEO8LiClO4 polymer electrolyte sample (Fig. 10), shows at 60 – 70°C, the peak series due to the crystalline to amorphous transition. The following trace is the cooling scan from 100°C to room temperature; no peak is observed since even ceramic-free electrolytes have a relatively slow recrystallization kinetics [1]. However, the electrolyte does recrystallize, as demonstrated by the peak shown in the trace obtained after 6 days of storage at room temperature. There is quite a different DSC response for the nanocomposite electrolyte (Fig. 11). The heating scan of a as-prepared sample also shows a peak due to the crystalline to amorphous transition around 60 – 70°C. However, in contrast with the previous case, in all the following cooling and heating traces no peaks are revealed even after prolonged storage times (i.e. exceeding 2 weeks). These results support the conductivity results (see Figs. 8 and 9) and, ultimately the transport model since they confirm that, after being annealed at temperatures above the PEO transition, the nanocomposite polymer electrolytes retain their amorphous state even if kept at room temperature for several days. A key question is whether the annealed nanocomposite sample may eventually recrystallize. Preliminary results suggest that after prolonged storage some recrystallization may occur. however, with kinetics which appear to be critically
G.B. Appetecchi et al. / Electrochimica Acta 45 (2000) 1481–1490
1489
Fig. 10. DSC traces of a ceramic-free PEO–LiClO4 nanocomposite polymer electrolyte as prepared and after 6 days of storage at room temperature. Heating–cooling rate: 10°C min − 1.
Fig. 11. DSC traces of a PEO–PEO–LiClO4·10w/o Al2O3 nanocomposite polymer electrolyte as prepared end after 4 and 14 days of storage at room temperature. Heating–cooling rate: 10°C min − 1.
dependent upon the annealing conditions, e.g. time and temperature.
4. Conclusion The results here reported demonstrated that new types of composite polymer electrolytes, prepared by a procedure which avoids any liquid-involving step and considers, the dispersion of selected ceramic powders have enhanced electrochemical properties. In particular, these dry composite electrolytes have an exceptionally high lithium metal electrode interfacial stability which allows to obtain a cycling efficiency approaching 99%, i.e. to our knowledge one of the highest values so far reported for PEO-based polymer electrolyte cell. Fur-
thermore, by extending the approach to the use of ceramic fillers at the nanoscale dimension, substantial enhancement of the transport properties may be obtained without affecting the electrolyte’s interfacial stability and its the mechanical properties. These improvements are considered to be of relevant importance in view of the most profitable application of these electrolytes, i.e. as separators in rechargeable lithium batteries [19].
Acknowledgements This work has been carried out with the financial support of ENEA, contract no. 2814.
1490
G.B. Appetecchi et al. / Electrochimica Acta 45 (2000) 1481–1490 [8] B.A. Boukamp, Solid State Ion. 20 (1986) 31. [9] J. Evans, C.A. Vincent, P.G. Bruce, Polymer 28 (1987) 2325. [10] K.M Abraham, Z. Jiang, B. Carroll, Chem. Mater. 9 (1997) 1918. [11] F. Croce, B. Scrosati, J. Power Sour. 43 – 45 (1993) 9. [12] I.M. Ismail, U. Kadiroglu, N.D. Gray, J.R. Owen, Fall Meeting of the Electrochemical Society, San Antonio, TX, 1996, abstract no. 63. [13] M.C. Borghini, M. Mastragostino, S. Passerini, B. Scrosati, J. EIectrochem. Soc. 142 (1995) 2118. [14] B. Scrosati, F. Croce, Poll. Adv. Technol. 4 (1993) 198. [15] F. Capuano, F. Croce, B. Scrosati, US Patent no. 5576115, 1996. [16] B. Kumar, L.G. Scanlon, J. Power Sour. 52 (1994) 261. [17] W. Wieczorek, Z. Florjancyk, J.R. Stevens, Electrochim. Acta 40 (1995) 2251. [18] J. Przyluski, M. Siekierski, W. Wieczorek, Electrochim. Acta 40 (1995) 2101. [19] B. Scrosati, Nature 373 (1995) 557.
References [1] F.M. Gray, Polymer Electrolytes, Royal Society of Chemistry Monographs, Cambridge, 1997. [2] (a) G. Kobe, Automotive Industries, 1996. (b) C. Donnelly, L. Christensen, D. Kuller, Electric and Hybrid Vehicle Technology, 1996. [3] P. Baudry, S. Lascaud, H. Majastre, D. Bloch, J. Power Sour. 68 (1997) 432. [4] G.B. Appetecchi, F. Croce, M. Mastragostino, B. Scrosati, F. Soavi, F. Zanelli, J. Electrochem. Soc. 145 (1998) 4133. [5] G.B. Appetecchi, F. Croce, G. Dautzenberg, M. Mastragostino, F. Ronci, B. Scrosati, F. Soavi, F. Zanelli, F. Alessandrini, P.P. Prosini, J. Electrochem. Soc. 145 (1998) 4126. [6] F. Croce, G.B. Appetecchi, L. Persi, B. Scrosati, Nature 394 (1998) 456. [7] F.M. Gray, J.E. MacCallum, C.A. Vincent, Solid State Ion. 18 – 19 (1986) 282.
.
