Furuday Discuss. Chern. SOC., 1989, 88, 103-111 Cation-Oxygen Geometry in Polymer Electrolytes: Interpretation of EXAFS Results Roger J. Latham,* Roger G. Linford and Walkiria S. Schlindwein Department of Chemistry, School of Applied Physical Sciences, Leicester Polytechnic, P.O. Box 143, Leicester LE1 9BH Two issues are of current interest in the field of ionically conducting polymers (polymer electrolytes): these are ion pairing and possible interference of the polymer-cation interaction by water. EXAFS was chosen as a suitable technique to probe local structure surrounding the cations. The systems studied were PEO,, : ZnXz, where n = 6-15 and X = C1, Br or I. They were chosen in order to ascertain the reliability of information pertaining to oxygen neighbours when the system under investigation contains heavy counterions.The results reveal, as expected, that the information about numbers of oxygen nearest neighbours is qualitative rather than quantitative, and firmer conclusions can be drawn for the lighter counterions. Cations and anions were found to be in close proximity, thus confirming ion pairing in PEO-zinc polymer electrolytes; this is in accord with recent observations of zinc diffusion. Polymer electrolytes can be considered to be concentrated solutions, which are special in the sense that the solvent is immobile. A given ion within this type of electrolyte can interact not only with its counterions, but also with mobile ligands present as impurities (e.g. water) and with the immobile matrix solvent. The interaction between a chosen mobile ion and its counterions leads to ion pairing; this is important in concentrated electrolytes such as these polymer electrolyte materials.’ The interaction between ion and ligand is similar to that found in normal electrolyte solutions.The solvent sheath has an effect on such parameters as the mobility, polariz- ability and ionic size of the potentially mobile species. The unusual feature in these electrolytes, however, is the interaction between the ion and its immobile matrix, which mimics that found in ionic complexes with crown ethers. If the cation interaction with the polymer host is strong, then the only potentially mobile ion is the anion. If on the other hand the cation is weakly bound, then the polymer electrolyte solution may be unstable and this leads to ‘salting-out’ effects in which the anions and cations recombine to form phase-separated crystallites.It is because of these factors that an understanding of the environment immediately surrounding the anions and cations within a polymeric electrolyte system is particularly important. The system chosen for this study is PEO, : ZnX,, where n = 6-15 and X = CI, Br or I. This was selected for two reasons. The first is that the Leicester group is among several who are interested in exploring the new avenues of behaviour revealed by non-monovalent systems. The complementary reason is that zinc is conveniently access- ible for the EXAFS technique that has been used for the exploration of the local structure. EXAFS is a structural technique that can be used on materials in a variety of forms, including crystalline, amorphous, liquid and glassy states.The absorption of a high- intensity X-ray beam by the sample is recorded over an energy range from ca. l00eV below to ca. 1000 eV above an absorption edge (usually the K-edge) of the chosen target element. The incident X-ray beam is usually provided from a synchrotron storage ring. For dilute samples, it is helpful to record an indirect function of absorbance, such as 103104 EXA FS of Polymer Electrolytes the fluorescent X-ray signal. The EXAFS phenomenon is seen as a series of oscillations on the high-energy side of the edge. It is in fact divided into several sub-regions. Nearest the edge is seen the so-called XANES region which contains information about the number, distance and geometric arrangement of second and third nearest neighbours.Beyond this is the EXAFS region proper which provides similar information about the first-nearest-neighbour shell. At still higher energies, oscillations due to so-called atomic EXAFS are seen which result from interference of the outgoing photoelectron wave with the outer electron shells of the target atom. Finally, a background region is obtained which is free from EXAFS effects. The local structural information provided by EXAFS lies within a spherical region of up to 6 , 4 of the target atom. In contrast with diffraction techniques, there is no long-range structural information and samples in which long-range order is lacking can, therefore, be studied. It is this feature that makes EXAFS so appropriate for the study of polymer electrolytes, in which the region of electrochemical interest is the amorphous phase.' The information contained in the raw EXAFS oscillations is not immediately acces- sible.In order to deconvolute this information, it is necessary to subtract two types of background: pre-edge and post-edge. Unreliable conclusions can be drawn from data for which the background subtraction has been improperly carried out. The background- subtracted data are then Fourier-transformed into a probability function, akin to a radial distribution function. This necessitates the use of phase shifts which are appropriate for both the target atom and the backscattering species. There are tests, which include the use of model compound data, to ascertain the suitability of the selected phase shifts.The aim of the deconvolution procedure is to provide a good fit between the experimental data and that obtained from an optimised theoretical local structure. This fit is carried out in both k space and real space. A region which has become of interest in the study of polymer electrolytes is the effect of small to trace amounts of water. The presence of water can cause remarkable changes in structure, conductivity behaviour and mechanical properties for some electrolytes. These effects could arise from the water acting as a plasticizer or because the water molecules are coordinated with the potentially mobile cations or for other reasons. A way of discriminating between these possibilities is to obtain information about the local structure, and such a study forms the basis for this paper.In earlier work, water was often uncritically and unintentionally incorporated into the films in the form of hydrated salts or imperfectly dried reagents. It is more common in current practice to ensure that the film is dry and this is achieved in one of two essentially different ways. In the first approach, normal quality reagents are used and after casting the film is subjected to a drying regime, typically involving heating to above 100°C for many hours under vacuum. This drying procedure is also unavoidably an annealing process and consequently the morphology of the film is affected. The alterna- tive method, which is being carefully developed to avoid undesirable structural modifications, involves the use of pre-dried reagents under scrupulously dry conditions.The entire procedure is carried out under high-integrity glove-box conditions. It is unusual for water to be employed as a reagent in the preparation of electrolyte films. Since PEO is water soluble, however, some workers are starting to investigate the properties of water-cast films and the samples studied in this investigation were of this type. When water is the chosen casting solvent, it is clearly impossible to use the pre-dried reagent approach to produce dry films. In this work, we have used a post- casting drying regime similar to that employed in non-aqueous systems but with the difference that the film after casting is heated under vacuum for seven days at 50°C.This is to ensure that the PEO melting temperature is not exceeded and so the as-cast morphology is maintained, and the formation of high-melting complexes is not encouraged.R. J. Latham, R. G. Linford and W. S. Schlindwein 105 Both methods give films that are dry, within the detection limits of the normal laboratory. Typical detection methods include thermogravimetric analysis, Karl- Fischer titrations and F.t.i.r. spectroscopy. None of these methods is sufficiently sensitive nor, with the exception to some degree of F.t.i.r. spectroscopy, do they discriminate between water in different sites. For example, it is not easy to separate the contribution of adventitious water, water of crystallisation and water that is present within the film as a plasticizer.The probable lower limit of convenient detection of water is 0.03%. This is a reflection of the character of water itself; it is too light to be easily weighed, and its all-embracing chemical reactivity does not assist in its identification. The aim of the experiments described in this paper was to elucidate the arrangement of the PEO matrix around the mobile cations. The methodology has involved the use of EXAFS to determine oxygen nearest neighbours. Clearly there is a complication in that the PEO-oxygen environment may be influenced by the presence of moisture. A secondary focus was to establish whether or not ion pairing occurred in these divalent systems. Contrary suggestions have been obtained from earlier work on rubidium’ and ca~cium.~ Experimental Film Preparation The films used in this investigation were prepared using water as a casting solvent.PEO of 4x lo6 relative molar mass (BDH) and zinc halides of the highest available purity (BDH) were dissolved in triply distilled water with stirring for 48 h. ‘Wet’ films were cast by allowing the water to evaporate at room temperature and these were then dried at 50°C for 7 days. All subsequent storage and processing was carried out in a high-integrity, sub-ppm-water-level dry box. EXAFS Experiments The EXAFS studies were carried out using the facilities of the synchrotron radiation source at the S.E.R.C. Daresbury Laboratories, Cheshire. Station 7.1 was used and the ring was operated at 2 GeV with a beam current in the region of 50-200 mA in multibunch mode and high-brilliance configuration.This station uses an Si ( 1 11) double-crystal monochromator and 70% harmonic rejection was used. Samples were typically exposed to the X-ray beam for 40 min during each run, and deconvolution was performed using the Daresbury EXAFS suite of programs. The dried films were protected from humid ambient air during transit by prior encapsulation at Leicester into special sample mounts; these consisted of Mylar sheets sealed with a knife-edge seal and have been described fully elsewhere.s The sample mounts were carried in a desiccator to Daresbury Laboratories, where they were stored within a high-integrity dry box prior to use on the EXAFS station. Results and Discussion Polymer electrolytes are likely to behave in a similar fashion to concentrated electrolyte solutions and as such there may be ion pairing in addition to interactions between the cation and the heteroatom from the host polymer backbone.In EXAFS experiments it is difficult to obtain conclusive information about oxygen nearest neighbours as oxygen atoms are weak back-scatterers. This is particularly the case when there may be heavier nearest neighbours, as provided by ion pairing in polymer electrolytes. In order to stu y these effects, electrolytes of the general composition PEO,, : ZnX, (where X = C1, Br, t , ) were selected. Thus, it might be expected that the evidence for oxygen bdfkscatteririg becomes less apparent as the EXAFS spectra are examined for samples OF increasirlg106 EXAFS of Polymer Electrolytes -5 t h L v 4 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0.0 - I ? Y Fig.1. PEO,: ZnC1, ‘dried’ film. Best fit using chlorine nearest neighbours only; ( a ) background subtracted EXAFS spectrum; ( b ) Fourier transform. (-), experiment; (- - -), theory. mass or size of the possible paired ion. When the experimental and theoretical EXAFS spectra are compared together with their corresponding Fourier transformation, it is clear that a satisfactory interpretation of the data cannot be made on the basis of a single back-scattering halide species. This is demonstrated by the typical result shown in fig. 1, which shows the background subtracted EXAFS spectrum and its accompanying Fourier transform.R. J. Latham, R. G. Linford and W. S. Schlindwein 107 Acceptable fitting of the data can only be obtained when oxygen is also considered as a nearest-neighbour species.The lightness of oxygen as a back-scatterer in comparison with the halide means that the information obtained is less well defined. The simplified expression for the EXAFS function does not take into account low energies, otherwise the plane-wave approximation cannot be applied. This is a particular problem when dealing with scattering species that are light (e.g. nitrogen, carbon and oxygen) since they only strongly scatter low-energy electrons. Two features of the EXAFS suite of programs available at the S.E.R.C. Daresbury Laboratories, however, provide the user with the facility to enhance the interpretation of the data. These are ( a ) the plotting of individual theoretical EXAFS spectra and Fourier transforms for each of the near-neighbour species so that the contribution to the total fit can be examined, and (6) a statistical significance test which can be applied to the fitting of the theoretical data.Fig. 2 shows the EXAFS spectrum and accompanying Fourier transform for the same polymer electrolyte film with fitting of the data for both oxygen and halide nearest neighbours. Fig. 3 shows the same spectrum with individual contributions from the oxygen and halogen back-scattering species. As expected at the high-energy end of the spectrum the main contribution is from halide back-scattering species, and the contribu- tion from oxygen is at low energies. For the systems studied the effects of oxygen backscatter are more clearly seen in the presence of the lightest halide species.The distances of the back-scattering species from the target zinc species suggest that the fit corresponds to a ‘split’ shell of two different nearest neighbours rather than a first- and second-nearest-neighbour situation. The results demonstrate that the halogen provides the most substantial contribution towards the observed EXAFS behaviour, and the distances and numbers of the nearest neighbours are shown in table 1 . This provides evidence that ion pairing is present in these polymer electrolytes. In contrast, whilst it is possible to say that there is certainly an interaction with oxygen, it is much more difficult to state precise coordination numbers. This is supported by the use of the statistical tests of Joyner et al.’ which show that there is much greater confidence in the coordination numbers obtained for halide nearest neighbours.There are two essentially different approaches that can be taken to the deconvolution of the experimental data. In the first, the best possible fit with the lowest fit index is sought, regardless of the acceptability or otherwise of the Debye-Waller factor that is finally evaluated. This approach is rationalised on the basis that the samples are expected to be more disordered (and consequently to have a higher Debye-Waller factor) than is the case for model compounds. In contrast, an alternative approach follows the phil- osophy of Joyner el al.’ in which fits where the Debye-Waller factor is not within the suggested range of 0.005-0.025 are rejected for samples where no static disorder is expected.The validity of this approach for polymer electrolytes where some disorder could be expected may be questioned. However, as is demonstrated in fig. 4, a relatively large envelope is obtained at the 1% level for the fits for oxygen nearest neighbours in comparison to the halide. This confirms that the majority of the EXAFS is due to the heavier and larger halide nearest neighbours and demonstrates that ‘intuitive’ fits obtained with high Debye- Waller factors and higher coordinations of oxygen could be less satisfactory. A further difficulty encountered with oxygen relates to the moisture effects encountered with polymer electrolyte systems. The results referred to in this paper are for ‘dried’ films and the films were specially mounted for the EXAFS experiments.If water is present in a polymer electrolyte film it may be difficult, using the EXAFS technique, to distinguish between oxygen backscatterers from the PEO backbone, and those from either water of hydration surrounding the cation or adven- titious water within the film. This more careful analysis has revealed that information on oxygen nearest neigh- bours is qualitative rather than quantitative. In earlier investigations we reported: ( a )108 EXAFS of Polymer Electrolytes c I ? Y Fig. 2. PEO,:ZnCI, ‘dried’ film. Best fit using chlorine and oxygen nearest neighbours; ( a ) background subtracted EXAFS spectrum; ( b ) Fourier transform. (-), experiment; (- - -), theory . the cation in PEO,: CaI, prepared under wet conditions was surrounded by ca.ten nearest n e i g h b ~ u r s , ~ in accord with the results of Enderby.‘ (6) The large number of oxygen nearest neighbours was also found for the same material prepared under dry conditions. Additional studies of other stoichiometries again showed large coordination numbers, and a small dependence of coordination number on overall film stoichiometryR. J. Latham, R. G. Linford and W. S, Schlindwein 109 1.1 t 1 2 3 4 5 6 7 8 9 10 r / 'A Fig. 3. PEO,: ZnC1, 'dried' film. Data from Fourier transform in fig. 2. Shown as individual contributions. (-), experiment; (- - -), theory (oxygen); (- - -) theory (chlorine). Table 1. PEO, : ZnX, 'dried' films nearest neighbour fit sample atom R I A N h cr2/A2 " EIeV index PEO, : ZnI, 0 PEO, : ZnBrz 1 .O- 1.4 0.005-0.007 (1 .O) (0.006) I 2.514 0 PEO, : ZnClz 2.0-3.0 0.009-0.012 (2.4) (0.010) Br 2.324 0 1.0-2.0 0.005-0.018 7.41 1.12 ( 1.40) (0.01 0) 2.054 0.6- 1.1 0.002-0.007 2*196 (0.9) (0.004) c1 Distance of nearest neighbours, coordination number, '' Debye- Waller factor110 EXA FS of Polymer Electrolytes 0.014 0.012 0.010 0.008 0.006 0.OOL 0.002 0 .o o o 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.1 1.6 1.8 2.0 coordination no. 0.020, , ,, , , , ,, , , ,, , 1 k 0.008 D A 0.6 0.8 1.0 1.2 1.C 1.6 1.8 2.0 2.2 2.6 coordination no. Fig. 4. PEOb : ZnC1, ‘dried’ film. Statistical plots using the approach of Joyner et al.’ ( a ) chlorine backscatterers; ( b ) oxygen backscatterers. was observed. ( c ) In contrast, for dry (i.e. those prepared using stringently dry conditions) PEO, : ZnX, films,5 oxygen coordination numbers were smaller than for calcium, showed no dependence on n and apparently were different for bromide and iodide.It remains clear that in these materials calcium is surrounded by more oxygen nearest neighbours than zinc. This implies that the mobility of the cation in calcium electrolytes is likely to be lower than that for zinc. Further, the qualitative nature of the data does not invalidate the conclusion that there is a modest oxygen coordination numberR. J. Latham, R. G. Linford and W. S. Schlindwein 111 dependence for calcium but not for zinc, and that the oxygen coordination varies with the nature of the anion. The only implication of the present work that affects the previously reported studies is that the earlier numerical values may be less precise than quoted.The present studies have reinforced the clarity with which statements about ion- pairing in these systems can be made. Calcium ions do not appear to be ion-paired in a nearest neighbour sense but zinc clearly is. This is in accord with many recent studies of zinc systems in which for zinc, but not Zn*+, transport exists. The authors thank Dr S. Gurman, IJniversity of Leicester, for helpful discussions about the data deconvolution process. References 1 M. A. Ratner and A. Nitzan, Solid Stare lonics, 1988, 28-30, 3. 2 C . Berthier, W. Gorecki, M. Minier, M. B. Armand, J. M. Chabagno and P. Rigaud, Solid Srate lonics, 1983, 11, 91. 3 C. R. A. Catlow, A. V. Chadwick, G. N. Greaves, L. M. Moroney and M. R. Warboys, Solid Stare lonics, 4 K. C. Andrews, M. Cole, R. J. Latham, R. G. Linford, H. M. Williams and B. R. Dobson, Solid Stare 5 M. Cole, M. H. Sheldon, M. D. Glasse, R. J. Latham and R. G. Linford, Appl. Phys. A, 1989, 49, 249. 6 N. A. Hewish, G. W. Nelson and J. E. Enderby, Narure (London), 1982, 297, 138. 7 R. W. Joyner, K. J. Martin and R. Meehan, J. Phys. C, 1987, 20, 4005. 8 H. Yang and G. C. Farrington, Extended Abstracts, The Electrochemical Society 174th Meeting, Abstract 9 G. C . Farrington and R. G. Linford, in Polymer Elecrrolyre Reviews 11, ed. J. R. MacCallum and C. A. 1983, 9-10, 1107. Ionics, 1988, 28-30, 929. no. 728, 1988. Vincent ( Elsevier Applied Science Publishers, London), in press. Paper 9/02162K; Received 22nd May, 1989