Ion–polyether coordination complexes: crystalline ionic conductors for clean energy storage

Abstract

Ion–polyether complexes are the solid state analogues of crown ether and cryptand complexes. They represent a fascinating class of coordination compounds in their own right, with the ability to support ionic conductivity and the potential to be used as electrolytes in all-solid-state rechargeable lithium batteries. Here the recent discovery of ionic conductivity in crystalline ion–polyether complexes, when for 30 years such materials were considered to be insulators, is described, along with their closely related structural chemistry.

---

Introduction

Since the pioneering discovery of crown ethers and cryptands by Pedersen, Cram and Lehn, coordination complexes based on these compounds have been studied in considerable detail.¹ The field of macrocyclic coordination chemistry has expanded significantly in recent years. By selecting chains composed of the same –CH₂–CH₂–O– repeat unit as used in the crown ethers and cryptands, such chains can wrap around ions forming particularly stable “crown-ether-like” multi-nuclear coordination complexes. If the chains are long then ion–polymer complexes are obtained, with the cations coordinated by the ether oxygens; the anions also exist within the polymer matrix, Fig. 1.

Fig. 1 Coordination (thin lines) of a Na⁺ ion in a 15-crown-5:NaI complex (left) and in PEO₃:NaI (right). Purple, sodium; brown, iodine; green, carbon; red, oxygen.

The ion–polymer coordination complexes not only represent a unique class of coordination compounds in the solid state, they also support ion transport. By combining solid yet flexible properties with ionic conductivity, they are ideally suited as solid electrolytes in all-solid-state electrochemical devices, such as rechargeable lithium-ion batteries, Fig. 2.

Fig. 2 Rechargeable lithium-ion battery.

Global warming and the finite nature of fossil fuels conspire to present one of the greatest threats to our planet in the 21st century. It is imperative that we reduce our use of fossil fuels. Some 30% of CO₂ emissions arise from burning fossil fuels for transportation. A realistic solution to address this problem over the next 20–40 years will be the hybrid electric vehicle, combining an internal combustion engine (and later a fuel cell) with a rechargeable battery. The engine/fuel cell charges the battery and the result is a significant reduction in CO₂ emissions per kilometre. Rechargeable lithium batteries offer the highest energy storage per unit weight and volume and are therefore the most attractive solution for future hybrid vehicles. The other major source of CO₂ rises from the burning of fossil fuels to generate electricity. Clean energy sources such as wind, wave and solar are inherently intermittent in supply and rechargeable lithium batteries can make an important contribution to storing the clean energy when it is available and supplying it to the consumer according to demand. This is especially true as we move from an energy supply based on a national grid to micro grids, with a diversity of clean energy generating sources. Taking the typical energy density of current lithium-ion batteries, 400 W h l⁻¹, a 10 m × 10 m × 10 m lithium battery installation could store 400 MW h of energy, comparable to the output of a wind farm!

Ion–polyether complexes or (polymer electrolytes) were first discovered by Fenton, Parker and Wright in 1973.² Since then, such complexes have been prepared with many different monatomic ions and indeed many different polymeric ligands, Fig. 3. More details on this and other general aspects of polymer electrolytes may be obtained in ref. 3. The polyether complexes remain the most stable and by far the most widely studied. The diversity of cations that form complexes include open-shell transition metals, lanthanides and actinides. Such complexes can display a wide variety of optical, magnetic and electronic properties, the vast majority of which have remained unexplored, in no small part because of the lack of knowledge concerning the structures of ion–polyether complexes. The outstanding exception to this lack of investigation is of course the Li⁺-polyether complexes, which Armand in 1979 recognised could be used as solid electrolytes in electrochemical devices.⁴

Fig. 3 Shaded elements form salts that are soluble in PEO.

Polymer electrolytes may be prepared as amorphous materials or, at certain specific cation to ether oxygen ratios, as crystalline complexes. For the past 30 years it has been universally accepted that ion transport through polymer electrolytes occurred only in amorphous materials above their glass transition temperature, T_g, and that the continuous motion of the polymer chain segments above T_g was essential for ions to move through the polymer.⁵ Segmental motion cannot, by definition, occur in the crystalline complexes and, as a result, such crystalline ion–polymer complexes were believed to be insulators. Recently however, this mantra has been overturned by the discovery of ionic conductivity in crystalline ion–polyether complexes formed between poly(ethylene oxide) and the salts LiXF₆; X = P, As, Sb, where the ether oxygen to cation ration is 6 : 1.^6–8 This discovery does not stand alone since others have also demonstrated that organisation and order can enhance the ionic conductivity in polymer membranes.^9,10 These results have opened up a new direction in the search for ion transport in the solid state. The ionically conducting crystalline ion–polyether complexes not only represent a new avenue in polymer electrolyte research but also constitute a unique class of solid ionic conductors in their own right, distinct from amorphous polymers or crystalline ceramic ionic conductors such as ZrO₂ or NaβAl₂O₃. Tremendous advances in the level of ionic conductivity in amorphous polymers have been made over the last 30 years by preparing materials with highly flexible polymer chains, resulting in low T_g's and hence high segmental motion, nevertheless the levels of ionic conductivity obtained remained too low for many applications.⁸ Within just a few years, research on crystalline ion–polyether complexes has raised the level of their conductivity to that of many amorphous polymers. It remains an intriguing possibility that yet higher levels of conductivity, suitable for applications, in for example in lithium-ion batteries, may be obtained.

This paper describes crystalline ion–polyether complexes, their structure, its relationship to conductivity and how the ionic conductivity of these materials may be raised by doping.

The structures of ion–polyether complexes

Modern chemistry is built on a knowledge of what are often complex structures, such knowledge is essential in order to understand function. Ion–polymer complexes are no exception, yet our knowledge of their structures remained minimal until recently. The origin of this lay in the poor quality of single crystal diffraction data from polymer electrolyte fibres. Unfortunately, despite good quality powder diffraction data being obtainable with careful sample preparation, the combination of significant overlap of reflections (ubiquitous with powder diffraction data) and the complexity of ion–polyether structures meant that previously successful methods for solving, ab initio, crystal structures from powder diffraction data proved inadequate. The severe lack of knowledge concerning the structures of ion–polyether complexes precluded the use of the well established method of Rietveld refinement, in which a reasonable model for the structure may be refined by fitting it to the diffraction data using non-linear least squares. Faced with these difficulties, we developed a different approach to the solution of flexible molecular structures ab initio from powder diffraction data that does not rely on deconvoluting overlapping reflections. The approach, based on simulated annealing, proved capable of solving the largest and most complex molecular structure to date from powder diffraction data.¹¹ Its role in unlocking the door to the structural chemistry of ion–polyether complexes and hence to the discovery of ion–polyether complexes that support ionic conductivity (PEO₆:LiXF₆, X = P, As, Sb) cannot be over estimated.

A number of the structures of ion–polyether complexes have now been solved. Space does not permit a detailed discussion here, instead some highlights will be given. More details may be obtained in refs. 11 and 12. The structures of crystalline ion–polyether complexes depend on the cation and anion sizes and charges, as well as the ratio of ether oxygens to cations. Ion–polyether complexes ranging in composition from ether oxygen to cation ratios of 8 : 1 (PEO₈:NaBPh₄) to 0.5 : 1 (PEO_0.5:LiCF₃SO₃) are known. A summary of the structures solved to date, based on PEO, is given in Fig. 4. Although still limited compared with the number of known structures, the structures summarised in Fig. 4 represent a significant advance in our knowledge of the structural chemistry of polymer electrolytes.

Fig. 4 Coordination (thin lines) of cations in PEO:salt complexes: (a) PEO₁:NaCF₃SO₃; (b) PEO₃:LiAsF₆; (c) PEO₄:KSCN; (d) PEO₈:NaBPh₄; (e) PEO₄:ZnCl₂. Light blue, lithium; light purple, sodium; dark purple, potassium; grey, zinc; yellow, sulfur; dark blue, nitrogen; orange, chlorine; magenta, fluorine; white, arsenic; green, carbon; red, oxygen.

The rich variety of structures and its corresponding sensitivity to ion type, size and ether oxygen to cation ratio is evident in Fig. 4. Considering first salts containing monovalent cations with radii up to and including that of Na⁺, and with anions ranging in size from I⁻ to N(SO₂CF₃)₂⁻, such salts form complexes with poly(ethylene oxide) in which the ether oxygens to cation ration is 3 : 1. The PEO chain describes a right-handed helix that wraps around the cations, resulting in the coordination of each cation by three ether oxygens and two anions. Each anion bridges between two neighbouring cations along the chain. The complete crystal structure consists of these infinite, one-dimensional and helical, ion–polyether coordination complexes interacting with their neighbours only through weak van der Waals forces. For salts containing monovalent cations larger than Na⁺, a right-handed helix is also observed but it presents a larger coordination environment to accommodate the larger cation, with five ether oxygens coordinating each cation and again each anion bridging between two neighbouring cations. The higher coordination number forces an increase in the ether oxygen to cation ratio from 3 : 1 to 4 : 1.

Starting with the PEO₃:LiAsF₆ complex and increasing only the PEO : cation ratio, results in the formation of the 6 : 1 complex, PEO₆:LiAsF₆, in which two polymer chains wrap around each Li⁺ ion, with three ether oxygens from one chain and two from the other coordinating each Li⁺ ion.^13,14 It would involve unfavourable strain if a single PEO chain were to wrap around a small ion such as Li⁺, in order to fulfil all its coordination requirements alone. The structure with the highest ether oxygen to cation ratio known so far is PEO₈:NaBPh₄. In this case a single chain wraps around the Na⁺ ions providing seven ether oxygens for coordination to each Na⁺ ion. One ether oxygen is not involved in coordinating Na⁺, its associated ethylene oxide unit links each of the coordination complexes along the chain.¹⁵ An example of a structure with the lowest PEO to cation ratio solved to date is PEO₁:NaCF₃SO₃.¹⁶ It consists of zigzag PEO chains, in which the cations are coordinated by only two ether oxygens, the coordination being completed by four anions that bridge between cations on neighbouring PEO chains. The 1 : 1 structures are the only known examples, so far, in which there is a three-dimensional network of ion–ether-oxygen–ion interactions. As a result of the 3D network these complexes exhibit relatively high melting points.

Although studies have been dominated by coordination complexes containing monovalent cations, we have recently solved the first structure of a multivalent cation polymer complex, PEO₄:ZnCl₂.¹⁷ The PEO chains form large loops within each of which a Zn²⁺ ion is located. Each Zn²⁺ ion is coordinated by two neighbouring ether oxygens along the chain and two Cl⁻ ions.

From structure to function: ionic conductivity in crystalline ion–polyether complexes

The systematic exploration of ion–polyether structures, enabled by the development of the simulated annealing approach to ab initio structure solution from powder diffraction data, led directly to the discovery of the 6 : 1 crystal structures PEO₆:LiXF₆, X = P, As, Sb, Fig. 5.¹⁴ The crystal structures of all three 6 : 1 complexes are broadly similar and involve the Li⁺ ions residing in polyether tunnels formed by two PEO chains each of which folds to form a half cylinder, with the two chains interlocking to form tunnels. The inner surface of the tunnels is lined by ether oxygens, which coordinate the lithium Li⁺ ions in a distorted square pyramidal geometry. The presence of Li⁺ ions in continuous tunnels that provide pathways for ion transport suggested that the 6 : 1 complexes might be the first examples of ionically conducting crystalline polymer electrolytes, this proved to be the case. One of the 6 : 1 complexes, PEO₆:LiSbF₆, may be quenched as an amorphous phase. As shown in Fig. 6, the conductivity of the equivalent crystalline 6 : 1 complex is higher than the amorphous phase. Furthermore, NMR evidence points to conductivity being dominated by Li⁺ ions, in contrast to the amorphous polymer electrolytes which generally show ion transport dominated by anions, something that is disadvantageous for applications in devices.⁶ Selectivity for Li⁺ transport is an anticipated consequence of ion transport in a crystalline phase.

Fig. 5 The structure of one 6 : 1 complex, PEO₆:LiAsF₆. (Left) View of the structure along the polymer chain axis showing rows of Li⁺ ions perpendicular to the page. (Right) View of the structure showing the positions and the conformation of the chains. Light blue, lithium; green, carbon; red, oxygen; white, arsenic; magenta, fluorine. Thin lines indicate coordination around Li⁺ ions.

Fig. 6 Ionic conductivity σ (S cm⁻¹) of crystalline (solid circles) and amorphous (open circles) PEO₆:LiSbF₆ complexes.

Conduction in the crystalline 6 : 1 complexes is envisaged to involve ion hopping between neighbouring five-coordinate ether oxygen sites and favoured by local motion of the polymer chains in these soft solids. Note that such motion is not the large-scale segmental movement of the chains in amorphous polymers. Computer modelling studies offer support for ion transport in the crystalline state, including the role of the local polymer chain motion.^18,19 It has also proved to be the case that lowering the molecular weight to 1000 not only increases the conductivity but ensures that the ion–polyether complexes are below the entanglement limit for this polymer (approximately 3000) ensuring a high degree of crystallinity.⁷

Perfect crystals, in which every Li⁺ site within the tunnels was occupied by a Li⁺ ion, could not support conductivity, since conduction must involve exchange of Li⁺ ions between energetically equivalent sites. Fortunately, crystals are imperfect and local defects, such as Li⁺ vacancies, give rise to the observed conductivity. However, it remains the case that such serendipitous defects are small in number and do not promote high levels of conductivity. How then may we increase the level of conductivity in these crystalline ion–polymer complexes? In contrast to amorphous polymers, where the chemical challenge is to design flexible chains and anions that lower T_g, thus increasing segmental motion, in crystalline polymer electrolytes the strategy we have adopted is to dope the materials, thus increasing the number of defects. Two approaches have already shown promise, isovalent and aliovalent doping, Fig. 7. This strategy is analogous to approaches that have been taken in ceramic ionic conductors, such as the oxide ion conductors used in solid oxide fuel cells. In the case of isovalent doping we have replaced up to 5 mole% of the AsF₆⁻ ions in PEO₆:LiAsF₆ by N(SO₂CF₃)₂⁻, i.e. we have formed the solid solution PEO₆:(LiAsF₆)_1−x(LiN(SO₂CF₃)₂)_x.⁸ The imide anion differs in size and shape, but not of course in charge, from the AsF₆⁻ that it replaces in the structure. This disrupts the local potential around the Li⁺ ions inducing more local disorder and leading to an increase in the conductivity by 1.5 orders of magnitude, Fig. 8(a). Beyond 5 mole% substitution a liquid phase begins to appear, which results in the slow rise in conductivity above x = 0.05, the material of course is no longer a solid ionic conductor.

Fig. 7 Illustration of (a) isovalent and (b) aliovalent doping. A possible position for an additional Li⁺ ion (dark blue) in a four-coordinate site of the PEO₆:(LiSbF₆)_1−x(Li₂SiF₆)_x is also shown.

Fig. 8 Conductivity isotherms as a function of x in (a) PEO₆:(LiAsF₆)_1−x(LiN(CF₃SO₂)₂)_x and (b) PEO₆:(LiSbF₆)_1−x(Li₂SiF₆)_x.

Aliovalent doping has been demonstrated by replacing SbF₆⁻ in PEO₆:LiSbF₆ with SiF₆²⁻, i.e. by a divalent anion of the same shape and marginally smaller size, forming the solid solution PEO_1−xLi(SbF₆)_1−x(SiF₆²⁻)_x. Electroneutrality demands that for each divalent anion substituting the monovalent, an additional Li⁺ must be incorporated somewhere in the crystal structure, i.e. an interstitial Li⁺ ion must be introduced. One possible location for the extra Li⁺ ion is a four-coordinate site that lies between each of the normal five-coordinate Li⁺ sites. This site was identified during solution of the 6 : 1 material, its size is such that it could accommodate a Li⁺ (acceptable Li–O distances). The presence of additional, interstitial, Li⁺ ions raises significantly the ionic conductivity, Fig. 8(b). Attempts to incorporate more than 1 mole% SiF₆²⁻ resulted in the appearance of a second phase, Li₂SiF₆. This is an insulator and as a result the conductivity does not continue to rise.

Some of the doped crystalline ion–polyether complexes have conductivities that make them suitable for use in electrochromic displays or smart windows.²⁰ More importantly, they demonstrate what can be achieved by modifying the crystalline ion–polyether complexes. They offer much scope for further doping, following similar or related strategies, to those described here. This is likely to define a future direction for research in the field and may lead to yet higher levels of ionic conductivity. It is also likely that increasing access to the crystal structures of ion–polymer complexes will lead to the synthesis of new complexes and to the exploration of their properties as ionic conductors and beyond.

Conclusions

Crystalline ion–polyether complexes represent a new class of solid ionic conductor, distinct from classical ceramic electrolytes or amorphous polymer electrolytes. They are soft solids that may be pressed or rolled into self-supporting films. They provide a new dimension within which to explore and understand ion transport in the solid state in general, as well as a possible route to new, more highly ionically conducting, polymer electrolytes. The achievements to date helped to define the scientific directions, there are of course still many important aspects of these materials that must be investigated if they are to be of technological use. The ionic conductivity must be increased further, their stability in contact with possible electrodes must be demonstrated, the mechanical properties and long term chemical stability must also be studied. However, they are composed of simple, low cost constituents and have already achieved conductivities comparable to many amorphous polymer electrolytes. They represent a fascinating area for further study that will undoubtedly reveal intriguing new scientific understanding and technological potential.

Acknowledgements

The author would like to acknowledge the members of his research group who have, over the years, contributed to the work described herein.

References

1. G. Gokel, in Crown Ethers and Cryptands: ed. J. F. Stoddart, RSC monographs, The Royal Society of Chemistry, Cambridge, 1991.
2. D. E. Fenton, J. M. Parker and P. V. Wright, Polymer, 1973, 14, 589.
3. Applications of Electroactive Polymers, ed. B. Scrosati, Chapman & Hall, 1993; Solid State Electrochemistry, ed. P. G. Bruce, Cambridge University Press, Cambridge, 1995; F. M. Gray, Polymer Electrolytes, RSC Materials Monographs, The Royal Society of Chemistry, Cambridge, 1997.
4. M. B. Armand, J. M. Chabango and M. J. Duclot, in Fast Ion Transport in Solids, ed. P. Vashishta, J. N. Mundy and G. K. Shenoy, North-Holland, Amsterdam, 1979.
5. M. A. Ratner, in Polymer Electrolytes Reviews-1, ed. J. R. MacCallum and C. A. Vincent, Elsevier Applied Science, London and New York, 1987; C. Berthier, W. Gorecki, M. Minier, M. B. Armand, J. M. Chabagno and P. Rigaud, Solid State Ionics, 1983, 11, 91; M. A. Ratner and D. F. Shriver, Chem. Rev., 1988, 88, 109.
6. Z. Gadjourova, Y. G. Andreev, D. P. Tunstall and P. G. Bruce, Nature, 2001, 412, 520.
7. Z. Stoeva, I. Martin-Litas, E. Staunton, Y. G. Andreev and P. G. Bruce, J. Am. Chem. Soc., 2003, 125, 4619.
8. A. M. Christie, S. J. Lilley, E. Staunton, Y. G. Andreev and P. G. Bruce, Nature, 2005, 433, 50.
9. P. V. Wright, Y. Zheng, D. Bhatt, T. Richardson and G. Ungar, Polym. Int., 1998, 47, 34.
10. S. H. Chung, Y. Wang, S. G. Greenbaum, D. Golodnitsky and E. Peled, Electrochem. Solid State Lett., 1999, 2, 553.
11. Y. G. Andreev, P. Lightfoot and P. G. Bruce, Chem. Commun., 1996, 2169; Y. G. Andreev, P. Lightfoot and P. G. Bruce, J. Appl. Crystallogr., 1997, 30, 294.
12. Y. G. Andreev and P. G. Bruce, J. Chem. Soc., Dalton Trans., 1998, 4071; P. G. Bruce and Y. G. Andreev, in Structure Determination from Powder Diffraction Data, ed. W. I. F. David, K. Shankland, L. B. McCusker and C. Baerlocher, Oxford University Press, Oxford, 2002.
13. I. Martin-Litas, Y. G. Andreev and P. G. Bruce, Chem. Mater., 2002, 14, 2166.
14. G. S. MacGlashan, Y. G. Andreev and P. G. Bruce, Nature, 1999, 398, 792; Z. Gadjourova, D. Martin, K. H. Andersen, Y. G. Andreev and P. G. Bruce, Chem. Mater., 2001, 13, 1282.
15. E. Staunton, A. M. Christie, Y. G. Andreev, A. M. Z. Slawin and P. G. Bruce, Chem. Commun., 2004, 148.
16. Y. G. Andreev, G. S. MacGlashan and P. G. Bruce, Phys. Rev. B, 1997, 55, 12011.
17. E. Staunton, A. M. Christie, I. Martin-Litas, Y. G. Andreev, A. M. Z. Slawin and P. G. Bruce, Angew. Chem., Int. Ed., 2004, 41, 2103.
18. P. Johansson and P. Jacobsson, Electrochim. Acta, 2003, 48, 2279.
19. D. Brandell, A. Liivat, H. Kasemagi, A. Aabloo and J. O. Thomas, J. Mater. Chem., 2005, 15, 1422.
20. A. Barnes, A. Despotakis, T. C. P. Wong, A. P. Anderson, B. Chambers and P. V. Wright, Smart Mater. Struct., 1998, 7, 752.

---