Solid State Ionics 177 (2006) 2699 – 2704 www.elsevier.com/locate/ssi
Polyphosphazene based composite polymer electrolytes N. Kaskhedikar a,b , J. Paulsdorf a , M. Burjanadze a , Y. Karatas a,b , B. Roling c , H.-D. Wiemhöfer a,b,⁎ a
University of Münster, Institute of Inorganic and Analytical Chemistry and SFB 458, Corrensstr. 30, 48149 Münster, Germany b NRW International Graduate School of Chemistry, Germany c University of Marburg, Department of Chemistry, Physical Chemistry, Hans-Meerwein-Str., 35032 Marburg, Germany Received 14 July 2005; received in revised form 2 May 2006; accepted 2 May 2006
Abstract Composite salt-in-polymer electrolyte membranes were prepared from poly[(bis(2-methoxyethyl)amino)1−x(n-propylamino)x-phosphazene] (BMEAP) with dissolved LiCF3SO3 and dispersed Al2O3 nanoparticles (40 nm). Membranes with good mechanical stability were obtained. Low ionic conductivities were found in particle free membranes with maximum conductivities at 10 wt.% LiCF3SO3 ranging from 3.1 × 10− 7 S/cm at 30 °C to 1.8 × 10− 5 S/cm at 90 °C. For the composite membranes, addition of 2 wt.% Al2O3 nanoparticles leads to a steep increase of the conductivity by almost two orders of magnitude as compared to the homogeneous membranes. The highest room temperature conductivity for the investigated BMEAP–LiCF3SO3–Al2O3 composite systems was 10− 5 S/cm. © 2006 Elsevier B.V. All rights reserved. Keywords: Composite membrane; Polymer electrolytes; Polyphosphazenes; Ionic conductivity; Transference number; Salt-in-polymer
1. Introduction The first report on the favorable effect of fillers on the mechanical and electrochemical properties of polymer electrolytes was given by Weston and Steele [1]. The expectation for enhanced conductivities has thereafter induced many investigations on composite polymer electrolytes. Early activities on composite polymer electrolytes concerned polyethylene oxide (PEO) and ion conducting fillers such as NASICON, β- and β″alumina, and glassy fillers [2–6]. Even ferroelectric particles such as BaTiO3 and carbon particles have been tested as additives [7,8]. Wieczorek et al. showed for in PEO–NaI electrolytes that dispersed oxide particles with diameters below 4 μm increased the ionic conductivity substantially [9]. The increase in ionic conductivity was attributed to modifications of the local structure in the polymer due to interactions
⁎ Corresponding author. University of Münster, Institute of Inorganic and Analytical Chemistry and SFB 458, Corrensstr. 30, 48149 Münster, Germany. Tel.: +49 251 83 33115; fax: +49 251 83 33193. E-mail address:
[email protected] (H.-D. Wiemhöfer). 0167-2738/$ - see front matter © 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.ssi.2006.05.003
between the polymer molecules and surface functions of the particles. Another series of investigations on the ionic conductivity of nanocomposite polymer electrolytes were carried out by Croce et al. [10,11]. They studied systems with TiO2, Al2O3 and SiO2 as ceramic fillers and observed an increase of the ionic conductivity by almost two orders of magnitude. For PEO based electrolytes, the most common explanation is that ceramic fillers decrease the crystallization temperature of the polymers thereby favoring the amorphous structure which shows enhanced ionic conductivity [11]. Later on, Wieczorek et al. argued about the Lewis acid–base type interactions at the interface between polymer and particles. The presence of these interactions leads to a distribution of ions between the coordination sites of polymer and particle surface of the filler and hence to a modified conductivity in the vicinity of the particles [12]. Most investigations on composite polymer electrolytes up to now concerned PEO and related polyethers. Very few results are available on the effect of ceramic fillers in polymer electrolytes based on polyphosphazenes and polysiloxanes [13,14]. In this work, we investigate the composite effect on salt-in-polymer electrolytes made of poly[(bis(2-methoxyethyl)amino)1−x(n-
2700
N. Kaskhedikar et al. / Solid State Ionics 177 (2006) 2699–2704
membranes were stored in a desiccator over Sicapent® to avoid the uptake of moisture. 2.1. GPC
Scheme 1. Poly[(bis(2-methoxyethyl)amino)1−x(n-propylamino)x-phosphazene] (BMEAP): Note that the polymer investigated here (x was nearly 0.4) is a random copolymer of three different momomer units –N f PR′2− , –N f PR′2− and –N f PRR′–according to the three possible substitution patterns with two substituents at the phosphorous.
propylamino)x-phosphazene] (BMEAP, cf. Scheme 1) with incorporated Al2O3 nanoparticles. 2. Experimental The precursor polymer poly[dichlorophosphazene] was synthesized by the living cationic polymerization as described in detail previously [15,16]. Poly[(bis(2-methoxyethyl)amino)1−x (n-propylamino)x-phosphazene] (= BMEAP in the following) was synthesized from the precursor polymer by nucleophilic substitution using bis(2-methoxyethyl)amine (Fluka, 97%) followed by a second substitution of the remaining unreacted chlorine with n-propylamine as described elsewhere [16]. The ratio of the two different side groups, namely bis(2-methoxyethyl)amino to propylamino, was determined as 60:40. Alumina powder (40 nm particle size, Aldrich) was used as received after drying without any additional surface treatment. The characterization of the parent polymers was done with NMR, FT-IR, GPC and DSC. The molecular structure was checked and verified by 1H, 13C and 31P NMR spectra and by FT-IR spectroscopy. Details are described elsewhere [16]. The polymer electrolyte membranes were prepared from THF solutions of BMEAP and 2.5, 5, 10, 12.5 and 15 wt.% lithium triflate by a solution casting technique. A typical procedure was as follows: 0.3 g BMEAP was dissolved in 3 mL of freshly distilled THF to obtain a clear solution. The desired amount of the lithium salt was added, and the solution was stirred for 24 h to ensure a homogeneous mixture of polymer and salt. The resultant solution was then poured into Teflon® moulds and sealed with Parafilm® to slow the rate of solvent evaporation. After 3 days, the evaporation of the solvent was complete and the membrane was removed from the mould, dried in an oven at 60 °C for 48 h and further dried in vacuum for 24 h to remove traces of solvent left. Composite membranes with dispersed alumina nanoparticles (40 nm) were prepared analogously. We used compositions of between 2 wt.% and 10 wt.% Al2O3 with respect to the polymer weight. The amount of lithium triflate was fixed in the BMEAP–Al2O3 composites at 10 wt.% (maximum conductivity for the particle free sample). All membranes obtained in this way were homogeneous, free standing, transparent and about 100 μm in thickness. The
The molecular weight distribution of BMEAP was determined by a gel permeation chromatograph, equipped with two PSS-SDV linear XL columns (8 mm × 300 mm, 5 μm) from Polymer Standards Service (PSS, Mainz), a RI-detector (agilent) and a η-1001 viscosimeter (WGE Dr. Bures). The column was calibrated with polystyrene ReadyCal standard (PSS, Mainz). The samples were eluted with a 0.1 wt.% solution of tetra-n-butylammonium bromide in THF. The weight average molecular weight was 1.36 × 105 Da with a polydispersity index (PDI) of 4.2. In order to assure complete substitution of the chlorine atoms in the polymer, a second small nucleophilic substituent propylamine was reacted in presence of triethylamine with the polyphosphazene partially substituted by the secondary amine. From the peak integration in 1H NMR of the completely substituted polymer, the amount of secondary and primary amine groups in the polymer was estimated. It was found that 60% substitution was with the secondary amine i.e., bis(bis(2-methoxy)ethyl)amine and remaining 40% chlorine were substituted with propylamine. 2.2. DSC Differential Scanning Calorimetry (DSC) was used to obtain the glass transition temperatures (Tg) and melting temperatures of the polymers and composites. The DSC analysis was carried out using a Netzsch DSC 204 at a heating rate of 10 K/min between − 150 °C and +200 °C. Three measurement cycles were done in each case to remove thermal hysteresis. 2.3. Impedance measurements The ionic conductivities of the polymer electrolyte samples were determined by impedance spectroscopy in the frequency range of 0.1 Hz–1 MHz. The as-prepared membranes were dried in vacuum at 60 °C for 24 h. The membranes were placed between two ion blocking silver electrodes. Prior to the measurement, the set-up was heated in the conductivity cell at 100 °C for about 30 min. All measurements were then carried under a constant flow of dry nitrogen. We recorded the impedance spectra between 0 and 90 °C using a Novocontrol α-S high-resolution dielectric analyzer. The phase angle of the impedance usually reached values near zero (about 2–5°) over an extended frequency range between 10 Hz and 104 Hz. In that range, the measured impedance corresponds to the bulk ohmic resistance and was used to calculate the ionic conductivities. Li+ transference numbers were measured at 90 °C with a standard potentiostat using the symmetrical cell Li∣polymer electrolyte∣Li which was assembled in a argon filled glove box. The average thickness of the measured membranes was around 100 μm. The area of the polymer electrolyte membranes and of the lithium foils was 0.25 cm2. The DC current was monitored
N. Kaskhedikar et al. / Solid State Ionics 177 (2006) 2699–2704
Fig. 1. Enlarged view of the DSC results for pure BMEAP and BMEAP + 2.5 wt.% LiSO3CF3 depicting the temperature range around the transitions at T1 and T2.
with time after a stepwise change of the cell voltage from 0 mV to 10 mV. 3. Results and discussion The DSC diagrams of the pure polymer BMEAP and of solutions of LiCF3SO3 in BMEAP are shown in Figs. 1 and 2. Pure BMEAP is thermally stable up to 250 °C, however it becomes very soft and flows above 150 °C. As seen in Fig. 1, BMEAP shows a broad glass transition with onset around − 56.9 °C and offset at − 26 °C. In the enlarged view of Fig. 2, there is no appearance of any crystalline melting indicating that BMEAP as well as the salt-in-polymer systems are completely amorphous in the relevant temperature range above 0 °C. The statistical distribution of the two different side groups at the phosphorous atoms favors disorder in the polymer. The broadening of the glass transition range can be expected for a randomly substituted chain. It may also be influenced by the relatively high polydispersity. As Fig. 1 shows, after the dissolution of LiCF3SO3 salt, the broad glass transition below − 26 °C disappears and is replaced by a dispersed transition range consisting of up to three overlapping peaks. Two peak maxima are clearly distinguishable at about T1 = − 50 °C and T2 = − 40 °C. This multi-peak
Fig. 3. DSC results of BMEAP–LiSO3CF3–Al2O3 composites.
structure remains also for the materials with additional dispersed Al2O3 nanoparticles, as seen in Fig. 3. The change in the appearance of the DSC curves from pure BMEAP to the salt-in-polymer systems clearly indicates a disturbance of the polymer structure and of the interactions between the segments. The remaining transition peaks which seem to be superimposed by the broad glass transition feature in pure BMEAP mark the remaining molecular interactions of the polymer segments. It is very probable that these are at least partly due to hydrogen bonding between the amino groups at neighbouring phosphorous atoms. The values of the transition temperatures T1 and T2 are compiled in Table 1 as a function of the concentrations of salt and Al2O3. The peak at the higher temperature (denoted T2) does not shift with changing salt concentration. After addition of Al2O3, the peak height at T2 clearly increases, although its temperature practically remains constant. The first peak at T1 shows a weak dependence on the Al2O3 content, but also a negligible influence of changing salt concentrations. These observations indicate an interaction of the Al2O3 particles with the polar groups of the polymer. Nevertheless, the predominantly amorphous character of the polymer membranes remains in the presence of dissolved salt and dispersed oxide particles.
Table 1 Transition temperatures T1 and T2 obtained from thermal analysis (DSC) of pure BMEAP and corresponding polymer electrolytes with and without dispersed Al2O3 Sample
x (wt.%)
Pure BMEAP, BMEAP + x wt.% LiSO3CF3
Broad glass transition with onset at −56.9 °C and offset: at − 26 °C 2.5 − 49.4 −39.8 5 − 49.4 −39.8 10 − 49.0 −39.4 12.5 − 48.9 −39.4 15 − 49.2 −39.4 0 − 49.0 −39.4 2 − 47.5 −39.2 4 − 48.3 −39.0 6 − 49.3 −38.9 8 − 49.8 −39.5 10 − 50.2 −38.7
BMEAP + 10 wt.% LiSO3CF3 + x wt.% Al2O3
Fig. 2. DSC results of pure BMEAP and BMEAP–LiSO3CF3 polymer systems.
2701
T1 (°C)
T2 (°C)
2702
N. Kaskhedikar et al. / Solid State Ionics 177 (2006) 2699–2704
Fig. 4. Arrhenius plots of some BMEAP based polymer electrolyte membranes.
Fig. 6. Arrhenius plots of BMEAP–LiSO3CF3–Al2O3 composite polymer electrolytes.
The conductivities of homogeneous salt-in-polymer membranes with BMEAP are shown in Fig. 4. A non-Arrhenius behaviour is clearly seen which indicates the coupling of ion transport to the segmental motions of the polymer. The maximum room temperature conductivity is low, for the 10 wt.% sample it is about 3.1 × 10− 7 S/cm increasing to 1.8 × 10− 5 S/cm at 90 °C. The dependence on the salt concentration is shown in Fig. 5. The ionic conductivity first increases and reaches a maximum at 10 wt.% LiSO3CF3. For higher concentrations, it decreases due to an increasing formation ion pairs and higher ion clusters. The temperature dependence of the ionic conductivity follows the empirical Vogel–Tamman–Fulcher (VTF) equation and was already discussed earlier [17]. For the BMEAP–Al2O3 composites, the corresponding Arrhenius plots are shown in Fig. 6. It is evident that the ionic conductivity in the composites is greatly increased as compared to the oxide free BMEAP polymer electrolytes (being about two orders of magnitude higher). This effect was observed for all investigated Al2O3 concentrations. The maximum ionic conductivity is about 10− 5 S/cm at 4 wt.% Al2O3. The conductivity enhancement is of the same order of magnitude as recently reported for PEO–LiClO4–Al2O3/TiO2
composites where the authors found a conductivity enhancement from 10− 7 S/cm to 10− 5 S/cm at room temperature [10]. We also determined transference numbers of lithium cations on homogeneous and composite membranes using the symmetric cell: Li∣polymer electrolyte∣Li. The electrical current response was monitored after a voltage step. A smooth, gradual decay was observed as the typical result of Fig. 7 shows. In analogy to the experiments by Shriver et al. [18,19], the effective lithium ion transference number (tLi+) was calculated from the ratio of the steady state current (It→∞) to the initial current (It→0) according to:
Fig. 5. Salt concentration dependence of the ionic conductivity of the investigated salt-in-polymer electrolytes.
tLiþ ¼
IðtYlÞ rLiþ ¼ IðtY0Þ rtotal
ð1Þ
The above relationship is valid only for small concentration gradients as discussed by Evans et al. [20]. This condition was fulfilled in our experiments. For BMEAP + 10 wt.% LiSO3CF3 at 90 °C, we found tLi+ = 0.23 indicating a minor contribution of cations to the total conductivity. In the case of composites with Al2O3, however, tLi+ was found to be 0.38 for samples with 4 wt. % and 6 wt.% Al2O3. The presence of oxide particles apparently increases the relative contribution of cations to the total
Fig. 7. Electrical current vs. time for the cell Li∣BMEAP + 10 wt.% LiSO3CF3∣Li cell after switching-on a DC voltage at time zero (90 °C, initial current = 2.06 μA, steady state current = 0.49 μA).
N. Kaskhedikar et al. / Solid State Ionics 177 (2006) 2699–2704
conductivity. Without applying further analytical techniques, one cannot give a definite reason for this difference. For PEO, the conductivity enhancement in the presence of fillers is often explained by the suppressed crystallisation of the polymer and a higher volume fraction of the fast ion conducting amorphous phase near the ceramic particles. The polyphosphazene investigated here, however, is free of microcrystallinity. Hence, a different explanation has to be found. As already mentioned, some authors explained conductivity enhancements in polymer electrolyte composites on the basis of Lewis acid– base type interactions at the polymer/oxide interface under participation of both the anions and the coordinating sites of the polymer [12,21]. Increased transference numbers of lithium ions have also been observed already by other authors in composite polymer electrolytes. Explanations concerned an increased mobility of the cations at or near the boundaries of the ceramic particles [10]. The large increase of the conductivity by almost two orders of magnitude basically concerns both the anions and the cations, although there is a slight preference in favor of the cations. Detailed investigations have been done recently by Bhattacharyya et al. on heterogeneously doped salt solutions in liquid and polymeric electrolytes [26,27]. Using a local conductivity measurement based upon AFM in contact mode, the authors proved that the conductivity increase of a polymer composite is indeed located directly in front of the particle surface. According to their discussion, their published results as well as the conductivity increase in this work is most probably due to an increase of the degree of salt dissociation and possibly to an accompanying preferential adsorption of anions at the surface of the oxide particles. But in addition to the changes in the degree of ion dissociation, it is quite probable that an interaction between polymer molecules and the particle surfaces also leads to disturbances of the random coil configuration of the amorphous polymers in the vicinity of oxide particles. This may result in a zone with a different mean free volume in the polymer around the particles. As a result, a more open structure of the polymer near the particles may enhance the segmental mobility of the polymer and, thus, also the ionic mobilities which are coupled to them. With regard to the transference numbers, an interesting consequence comes into the discussion from the recent findings of Stolwijk et al. [24,25]. Using tracer diffusion techniques, they could prove that anion diffusion predominates in polymer electrolytes based on polyethers with dissolved NaI over the entire temperature range and that the cation transport is largely due to the transport of ion pairs. Accordingly in these systems, the cation transport in principle occurs by a vehicle mechanism with the anions as vehicles. A net transport of a cation in the steady state then corresponds to the transport of one ion pair and the coupled transport of one anion in the reverse direction. Such a model is analogous to the transport of protons by water molecules in proton conducting polymers. If the transport of free lithium ions can be neglected for BMEAP based electrolytes, too, then the measured lithium transference numbers are mainly determined by the contribution of ion pairs to the total charge transport.
2703
Using Maier's treatment of conservative ensembles [22,23], the total lithium ion current (charge current) consists of the sum of the free lithium ion current and that of the ion pairs according to jLiþ ;total ¼ jLiþ ;free þ jLiþ ;pairs
ð2Þ
Neglecting the contribution of free lithium ions, the net lithium ion current is given by the following expression where the effective conductivity σLiX of lithium ions due to ion pairs is defined l jLiþ ;total c jLiþ ;pairs ¼ FDLiX d grad cLiX r−rLiX d grad LiX F ð3Þ In the steady state, the chemical potential gradient grad μLiX is equivalent to the Faraday constant F times the electrical potential gradient grad ϕ. Hence, under these assumptions, the transference number is given by tLiþ ¼
rLiX e2 cLiX DLiX with : rLiX ¼ rLiX þ rx− kT
ð4Þ
Taking the experimentally determined transference numbers, one obtains a value of 3.3 for the ratio σX−/σLiX of free anion to pair transport coefficients in the case of the homogeneous BMEAP + LiSO3CF3 membrane and 1.6 in the case of the composites. These numerical results closely resemble those inferred from tracer measurements on PEO– NaI for the ratio DI−/DNaI [24,25]. Nevertheless, in order to clarify the role of ion pairs in cation transport, the mechanisms of cation and anion transport have to be analysed further. If coupled transport of cations and anions by ion pairs turns out to play a decisive role, it would imply that, in addition to a fast pair transport, an efficient and fast anion transport is desirable for salt-in-polymer electrolytes to achieve a fast cation transport. 4. Conclusion The salt-in-polymer electrolytes prepared from the polyphosphazene BMEAP show a strong enhancement of the ionic conductivities by two orders of magnitude upon addition of Al2O3 nanoparticles. The ionic conductivity of the resulting composites is comparable to results reported for good polymer electrolytes based on poly[bis(methoxyethoxyethoxy)phosphazene] (MEEP) [28]. However, the mechanical stability of the BMEAP membranes at ambient temperatures is far superior to MEEP (if not cross-linked). Furthermore, if ion pair transport is more important for the cations than assumed before, one should lay more emphasis on polymer hosts that enable fast anion transport. Acknowledgement We are thankful to Wilma Pröbsting for carrying out the DSC measurements. Financial support from Deutsche Forschungsgemeinschaft (DFG) within the framework of the Sonderforschungbereich SFB 458 is gratefully acknowledged. N.K.
2704
N. Kaskhedikar et al. / Solid State Ionics 177 (2006) 2699–2704
and Y.K. are grateful to the NRW Graduate School of Chemistry for doctoral fellowships. References [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14]
J.E. Weston, B.C.H. Steele, Solid State Ionics 7 (1982) 75. J. Plocharski, W. Wieczorek, Solid State Ionics 28 (1988) 979. W. Wieczorek, Mater. Sci. Eng., B 15 (1992) 108. W. Wieczorek, K. Such, J. Przyluski, Z. Florianczyk, Synth. Met. 45 (1991) 373. F. Croce, F. Capuano, A. Selvaggi, B. Scrosati, G. Scibona, J. Power Sources 32 (1990) 381. F. Croce, F. Bonino, S. Panero, B. Scrosati, Philos. Mag., B 59 (1989) 161. H.Y. Sun, H.J. Sohn, O. Yamamoto, Y. Takeda, N. Imanishi, J. Electrochem. Soc. 146 (1999) 1672. G.B. Appetecchi, S. Passerini, Electrochim. Acta 45 (2000) 2139. W. Wieczorek, K. Such, H. Wysiclik, J. Plocharski, Solid State Ionics 36 (1989) 255. F. Croce, G.B. Appetecchi, L. Persi, B. Scrosati, Nature 394 (1998) 456. F. Croce, R. Curini, A. Martinelli, L. Persi, F. Ronci, B. Scrosati, R. Caminiti, J. Phys. Chem., B 103 (1999) 10632. W. Wieczorek, J.R. Stevens, Z. Florjanczyk, Solid State Ionics 85 (1996) 67. Y.W. Chen-Yang, H.C. Chen, F.L. Lin, Abstr. Pap. - Am. Chem. Soc. 222 (2001) 282. Y.W. Chen-Yang, H.C. Chen, F.J. Lin, C.W. Liao, T.L. Chen, Solid State Ionics 156 (2003) 383.
[15] H.R. Allcock, C.A. Crane, C.T. Morrissey, J.M. Nelson, S.D. Reeves, C.H. Honeyman, I. Manners, Macromolecules 29 (1996) 7740. [16] J. Paulsdorf, M. Burjanadze, K. Hagelschur, H.-D. Wiemhöfer, Solid State Ionics 169 (2004) 25. [17] J. Paulsdorf, N. Kaskhedikar, M. Burjanadze, S. Obeidi, N.A. Stolwijk, D. Wilmer, H.-D. Wiemhöfer, Chem. Mater. 18 (2006) 1281. [18] D.F. Shriver, S. Clancy, P.M. Blonsky, L.C. Hardy, in: F.W. Poulsen, N. HasselAndersen, K. Clausen, S. Skaarup, O.T. Sørensen (Eds.), International Symposium on Metallurgy and Materials Science, Risø National Lab., Roskilde, 1985. [19] P.M. Blonsky, D.F. Shriver, P. Austin, H.R. Allcock, Solid State Ionics 18–19 (1986) 258. [20] J. Evans, C.A. Vincent, P.G. Bruce, Polymer 28 (1987) 2324. [21] J. Przyluski, M. Siekierski, W. Wieczorek, Electrochim. Acta 40 (1995) 2101. [22] J. Maier, J. Am. Ceram. Soc. 76 (1993) 1218. [23] J. Maier, J. Am. Ceram. Soc. 76 (1993) 1223. [24] N.A. Stolwijk, S. Obeidi, Phys. Rev. Lett. 93 (2004) Art. No. 125901. [25] S. Obeidi, N.A. Stolwijk, S.J. Pas, Macromolecules 38 (2005) 10750. [26] A.J. Bhattacharyya, M. Dolle, J. Maier, Electrochem. Solid-State Lett. 7 (2004) A432. [27] A.J. Bhattacharyya, J. Fleig, Y.G. Guo, J. Maier, Adv. Mater. 17 (2005) 2630. [28] P.M. Blonsky, D.F. Shriver, P. Austin, H.R. Allcock, Solid State Ionics 18–19 (1986) 258.
