A new microwave ‘smart window’ based on a poly(3,4-ethylenedioxythiophene) composite

Abstract

The preparation and dc and microwave characterisation of a new microparticulate mixed polymeric conductor containing poly(3,4-ethylenedioxythiophene) (PEDOT) and copper metal in a poly(ethylene oxide)–Cu(BF₄)₂–LiBF₄ polymer electrolyte matrix is described. The composites exhibit large, rapid and reversible changes in their microwave transmission coefficients when small dc or ac fields (5–7 V) are applied radially across annular samples (13 mm o.d.) from the edges. At 1 GHz, changes of −6 dB (0 V) to −15 dB (5 V) with switching speeds of ≤250 ms were observed, and −2 to −12 dB at slower rates. Single cell experiments show that redox processes are confined to thin surface layers of the PEDOT particles. A mechanism for switching involving the polarisation of PEDOT particles under an applied field is proposed.

---

Introduction

Dispersions of conjugated polymers (CPs) such as poly(aniline) (PANi) and poly(pyrrole) (PPy) in insulating polymer matrices have been studied for their potential application in electromagnetic shielding (EMS) since the early 1980s.¹ However, these materials show only passive (non-switchable) absorption characteristics. There is also a need for controllable microwave surfaces, or microwave ‘smart windows’; a number of approaches to solving this problem have been described in the literature and these have been reviewed by Chambers and co-workers.^2–4

CPs generally exhibit at least two oxidation states, neutral or uncharged forms having low conductivity and charged (usually oxidised) forms being highly conducting. The low conductivity forms of CPs are highly transmissive to microwaves, whilst the conducting forms are highly reflective. Thus, a microwave ‘smart window’ (Fig. 1) capable of controlling the transmission of microwave radiation requires that the CP is able to undergo a redox reaction under the influence of an applied stimulus. We have previously published work in this field using either PANi⁵or PPy⁶ as the CP component and silver metal as a reducing agent in a polymer electrolyte matrix comprising AgBF₄ in poly(ethylene oxide) (PEO). Large, rapid and reversible changes in microwave transmission/reflection coefficients were observed when small (ca. 10 V cm⁻¹) dc/ac fields were applied across samples of the composites (1–3 cm) from the edges in either waveguide or coaxial line test sets. The electrodes must be applied at the edges of the composite film because they have very high microwave reflection coefficients; thus redox processes must propagate across the film by some kind of ‘cascade’ process.

Fig. 1 A ‘smart’ microwave window with fields applied from the edges.

The composites may be described in terms of a series of microcells at the CP–polymer electrolyte interfaces, with the reducing metal particles in close proximity to the CP particles. Each microcell may be represented by the electrochemical cell,

M|PEO–M(BF₄)_x–LiBF₄|CP (1)

where M is the reducing metal (Ag or Cu). When Cu/Cu²⁺ is used as the redox couple, the addition of LiBF₄ is also advantageous, since Cu²⁺ ions have only a limited mobility in the PEO matrix. The cell reaction occurring within each microcell may be represented by eqn. 1,where (n)x+ is the oxidation state of the CP or metal and superscript ‘0’ indicates the neutral state. In the absence of an applied field, the equilibrium shown in eqn. 1 is shifted to the left-hand side and the composite adopts the high resistance ‘off’ state. When the field is applied, the equilibrium shifts to the right-hand side of eqn. 1. However, in our previous investigations, the composites were found to have lifetimes of only 2 to 3 weeks due either to hydrolysis of the Ag salt or the formation of high resistance materials, presumably arising from CP degradation after storage and electrochemical cycling.

In this paper, we present the results of our investigations on a new composite composed of poly(3,4-ethylenedioxythiophene) (PEDOT, Fig. 2) and a Cu/Cu²⁺ redox couple in a PEO–Cu(BF₄)₂–LiBF₄ polymer electrolyte that shows much improved long term stability and reproducibility, presumably due to the enhanced thermal and environmental stability of PEDOT.⁷

Fig. 2 The structure of neutral PEDOT.

Experimental

Synthesis of poly(3,4-ethylenedioxythiophene)

3,4-Ethylenedioxythiophene (EDOT), sold under the trade name Baytron M^™, was generously donated by Bayer (UK) Ltd. and was used without further purification. All procedures were carried out at room temperature, unless otherwise stated.

PEDOT was prepared using a variation of the method described by Corradi.⁸ To a dispersion of EDOT (0.2 g, 1.4 mmol) in distilled water (20 cm³) was added anhydrous iron(III) chloride (4.4 g, 27 mmol). The solution was stirred for 72 h, after which time the product was collected by gravity filtration, washed with distilled water until the filtrate gave a negative response to sodium thiocyanate solution (10 wt%) and then dried in vacuo. The room temperature dc conductivity of the dried, compacted product was found to be 20 S cm⁻¹ (lit.⁸σ_rt = 20 S cm⁻¹); yield: 80 wt%.

Another sample of PEDOT was also prepared using similar quantities of starting materials, although in this case the PEDOT product was not washed, but used as collected from the first filtration; σ_rt = 30 S cm⁻¹. We hereafter describe samples from these two preparative procedures as ‘washed’ and ‘unwashed’, respectively.

Preparation of 26 wt% PEDOT microparticulate composites containing 0.85 wt% copper

The polymer electrolyte component of the composite material was prepared by dissolving PEO (0.57 g, 13 mmol; 60% molar mass = 5 × 10⁶, 40% molar mass = 3 × 10⁵) in acetonitrile (15 cm³) and stirring overnight. Copper(II) tetrafluoroborate hydrate (0.15 g, 0.63 mmol) in acetonitrile (2 cm³) was then added and stirring continued for a further 20 min.

PEDOT (prepared from 0.2 g EDOT) was suspended in acetonitrile (10 cm³) and copper metal (0.015 g; average particle size <10 µm), copper(II) tetrafluoroborate hydrate (0.15 g, 0.63 mmol) and lithium tetrafluoroborate (0.12 g, 1.3 mmol) in acetonitrile (3 cm³) added. The suspension was stirred overnight and then added to the polymer electrolyte component. After stirring the composite suspension for 4 h, the solvent was removed by rotary evaporation at 40 °C and the last traces removed on the vacuum line. Some composites were prepared using ‘washed’ PEDOT whilst others were prepared using the ‘unwashed’ product.

Films of the conducting polymer composite were prepared by pressing the dried product between two steel rollers until a sample thickness of about 1 mm was achieved. Annular discs of the composites (outer diameter = 13 mm, inner diameter = 5 mm) suitable for microwave transmission measurements using a coaxial line test-set were cut out of the film. Silver conducting paint was applied to the inner and outer perimeters to ensure good electrical contact.

Preparation of single cell

The cell Pt|composite PEO–LiBF₄–PEDOT|‘amorphous’ PEO–LiBF₄|Cu was constructed using a composite of 40 wt% PEDOT (no Cu salts or metal). The cell was prepared by pressing the composite film (1 cm × 1 cm × 0.1 cm) onto platinum foil and completed using amorphous PEO–LiBF₄ electrolyte and a Cu foil counter electrode. A branched amorphous polyether, ‘PEO/EM/AGE’, was generously supplied by the Daiso company and was used for the central polymer electrolyte electrode separator phase. The polymer electrolyte layer was deposited onto the copper electrode from an acetonitrile solution ([EO]∶[LiBF₄] = 5∶1) into a window (1 cm² ) of a 60 µm Sellotape™ spacer.

Optical microscopy. Optical micrographs of thin (ca. 10 µm) films of the composite material were recorded using an Olympus BX50 microscope equipped with digital Image Pro Grabber.

Electrochemical and microwave measurement techniques. Cyclic voltammetry and square wave fields (0.05 to 0.125 Hz) were applied to the composite discs using a Solartron 1287A electrochemical interface (ECI). Coaxial line microwave transmission measurements were recorded using an Agilent 8714ET RF network analyser over the frequency range 0.3–1600 MHz and carried out with the simultaneous application of the electric fields. The microwave impedance (Z) was calculated using eqn. 2,where Z₀ is the characteristic impedance of the apparatus (50 Ω) and τ is the microwave transmission coefficient (dB).

Results and discussion

The electrochemical properties of PEDOT composites

Fig. 3 shows a plot of log₁₀ conductivity (σ) versus composition for composites from both ‘washed’ and ‘unwashed’ PEDOT and with or without LiBF₄ present in the polymer electrolyte phase. All four plots are essentially sigmoidal in form. The regions of maximum gradients between ca. 10 and 20 wt% indicate the ‘percolation thresholds’ in the various materials, at which the transport of charge becomes facilitated by a critical proximity of adjacent particles (see Fig. 4). This observation is consistent with the Scarisbrick⁹ theory for random walks of conducting spheres with ohmic interparticle contacts. According to this theory, a decrease of about 8 orders of magnitude in resistivity is estimated between ca. 6 and 20 vol%. Assuming a density for PEDOT particles of approximately 1.5 g cm⁻³ and a density of approximately unity for the matrix, the experimental data in Fig. 3 suggest percolations are in the range 8–12 vol%. This figure may be considerably less in systems with interacting particles, such as carbon in hydrocarbon rubbers in which ‘concatenation’ or ‘necklaces’ occur. In these cases, the observed percolation threshold may be as low as ca. 5 vol%. This phenomenon has been attributed to de-wetting of the particles by the matrix.¹⁰

Fig. 3 Percolation threshold of microparticulate PEDOT composites: (■) ‘washed’ PEDOT; (□) ‘washed’ PEDOT + LiBF₄; (▲) ‘unwashed’ PEDOT; (△) ‘unwashed’ PEDOT + LiBF₄. Measurements were performed using 5 V square waves, 0.05 Hz.

Fig. 4 Transmission optical micrograph of ca. 10 µm thick composite film with 26 wt% PEDOT loading.

The Scarisbrick theory assumes that particles are non-interacting, i.e. are well ‘wetted’ by the matrix. The optical micrograph in Fig. 4 for a 26 wt% (∼17 vol%) PEDOT composite shows that the PEDOT particles are approximately 1 µm in dimension. There are no clear indications of concatenation and it is apparent that most particles are not in physical contact with their neighbours. However, there are larger particles (5–10 µm), some of which are Cu metal particles (as shown by SEM), and also clusters of PEDOT particles.

Thus, the presence of the LiBF₄ apparently has little influence on the critical percolation compositions of either ‘washed’ or ‘unwashed’ composites. However, LiBF₄ clearly has a significant influence on the conductivities of the ‘washed’ composites, raising the conductivities by a factor of ca. 10³ throughout the range of compositions. In the case of the ‘unwashed’ material, however, it is only below the percolation composition that LiBF₄ increases the conductivity. At percolation and above, the ‘unwashed’ composites, with and without the added salt, have similar conductivities. A plausible explanation for this observation is that the ‘excess’ FeCl₃ carried on the surface of the ‘unwashed’ particles migrates to the polymer electrolyte phase, promoting the conductivity of this phase. {From gravimetric analysis, it was estimated that for the ‘unwashed’ composite at a composition just above the percolation region (ca. 25 wt% PEDOT/PEO + PEDOT), the concentration of FeCl₃ in the electrolyte phase corresponds to a molar composition of [FeCl₃]∶[ether oxygen] ≅ 1∶7.}

Fig. 5 shows resistance versus voltage for annular samples (13 mm outer diameter, 5 mm inner diameter) of 26 wt% PEDOT composites prepared from ‘washed’ and ‘unwashed’ microparticulate PEDOT, both containing Cu, Cu(BF₄)₂ and LiBF₄. As shown in Fig. 5(ii), from which plot b in Fig. 5(i) was constructed, the current reaches a maximum in 0.25 s or less. Fig. 5(i) shows that both composites are non-ohmic for applied potentials of 0 to ca. 2 V, but greater field dependence is seen with the ‘unwashed’ material. Above 2 V, both materials have similar resistances.

Fig. 5 The dependence of resistance on applied field for the annular samples of PEDOT–Cu(BF₄)₂–LiBF₄–Cu composite films containing 26 wt% of PEDOT; molar ratio of EDOT∶Cu metal = 6∶1. (i) ‘Washed’ PEDOT composite (a), ‘unwashed’ PEDOT composite (b). (ii) Square waves of frequency 0.25 Hz, from which plot b for the unwashed sample was obtained.

According to Fig. 3, both composites are above the percolation thresholds, although there are differences in their resistance changes with potential. The composite with ‘unwashed’ PEDOT is more resistive at low fields and, so, shows the larger change between 0.1 and 2 V in Fig. 5(i). With further increases in applied voltage, the resistance decreases only slightly in both samples.

Application of PEDOT composites in microwave switching. Fig. 6 and 7 show microwave transmission measurements for annular discs of the ‘washed’ and ‘unwashed’ composites, respectively. Only small changes in transmission coefficient were observed for applied field changes between 0 and 2 V. With increasing applied field, the transmission changes become larger and optimum values are observed between 5 and 7 V. When a square wave potential is applied [Fig. 6(ii)], the microwave transmissions can be switched between the levels shown in Fig. 6(i) and 7 following the applied fields for more than a thousand cycles. The microwave switching speeds have not been independently measured, but they are apparently ‘instantaneous’ and as fast as the non-linear responses shown in Fig. 5(ii) (≤0.25 s). The magnitude of the change in transmission is frequency-dependent in both samples but for the square wave fields change in the ‘unwashed’ sample (−6 to −15 dB at 1 GHz) is larger than in the ‘washed’ sample (−4 to −10 dB at 1 GHz). These results correspond to changes in microwave impedances of 43 to 11.5 Ω for the ‘washed’ sample and 16.5 to 4.5 Ω for the ‘unwashed’ sample (see eqn. 2). Thus, the higher resistance of the ‘unwashed’ sample at 0 V in Fig. 5(i) is not evident from these data. However, a field applied more slowly to the ‘unwashed’ sample using cyclic voltammetry (50 mV s⁻¹) gave the response changes shown in Fig. 7(a). This indicates a more resistive material at 0 V, corresponding to a change from −2 dB (96 Ω) to −12 dB (8 Ω) at 1 GHz. This is the largest change in transmission that we have observed so far in a PEDOT composite and indicates that the ‘switching off’ process is the slower.

Fig. 6 Microwave transmission of ‘washed’ PEDOT composite film measured using a coaxial line analyser with square waves of 0 and 5 V applied: (i) change in microwave transmission coefficient; (ii) applied square wave.

Fig. 7 Microwave transmission of ‘unwashed’ PEDOT composite film measured using a coaxial line analyser: (a) field applied using cyclic voltammetry, 50 mV s⁻¹; (b) field applied using square wave, 0.05 Hz.

All of the experiments suggest greater changes at 300 MHz. In Fig. 7(a), for example, the change from ca. −0.3 to −16 dB corresponds to an impedance change from ca. 710 to 4.5 Ω. The origins of this frequency dependence are not clear. It may reflect features of the composite structure (particle size, matrix conductivity, etc.). Alternatively, modelling of the coaxial experiment suggests that the phenomenon may arise from inductive effects due to the creation of highly conducting pathways (‘wires’) having resistances significantly lower than most of the area of the disc.

Any change of temperature within the discs as a consequence of the currents passed are not considered to be responsible for the observed changes in resistivities and microwave transmission. This was tested by independently raising the temperature of a composite disc using warm air and measuring its resistance. A negligible temperature coefficient for the conductivity of the composite material over the range 20–55 °C was observed. The temperature of the sample in situ in the coaxial line was also measured with a square wave current of 0.5 A, and the temperature rise was found to be well within this range with a maximum of ca. 35 °C. Furthermore, the thermal capacity of the system should have ensured that the steady-state temperature was constant and could not follow the observed rapid changes in microwave transmission. Finally, it is noted from Fig. 5 that the largest changes in resistance take place at lower fields, becoming almost constant above 2 V, where larger increases in temperature might have been anticipated.

Single cell experiment

Fig. 8 shows a cyclic voltammogram for a cell incorporating a PEDOT composite electrode, Pt|composite PEO–LiBF₄–PEDOT/PEO + LiBF₄|Cu, at a scan rate of 2 mV s⁻¹. The polymer electrolyte in the composite has the same composition as that of the central ‘separator’ phase. Although oxidation and reduction peaks can be discerned, it is apparent that they are very indistinct and that the redox process (eqn. 1) takes place only to a minimal degree. From the area under the peaks and the dimensions of the electrode (1 cm × 1 cm × 0.1 cm) it may be estimated that only ca. 0.01% of the PEDOT in the composite electrode partakes in the exchange. Assuming that this is distributed uniformly throughout all the particles of the electrode, it must be concluded that only the surface layer of each PEDOT particle is involved. A solvent-free cell incorporating an electrodeposited polyaniline film versus Ag and using an “amorphous PEO” electrolyte was observed by Despotakis¹¹ to give a cyclic voltammogram having better defined redox peaks. However, the peak areas similarly demonstrated that only the surface layer of the film (ca. 20 Å) was involved in the exchange. This is to be anticipated in a solvent-free system in which the polymer electrolyte is incapable of swelling the conjugated polymer particles so that the migration of counterions into the interior of the particles is very slow. These observations suggest that the mechanism of rapid switching observed in the microwave transmission measurements involves only exchange of charge at the PEDOT–polymer electrolyte interface.

Fig. 8 Cyclic voltammogram of single cell Pt|composite|PEO + LiBF₄|Cu; scan rate 2 mV s⁻¹.

A proposed mechanism for switching

The composites described in this paper contain 26 wt% (∼17 vol%) of PEDOT. Although this composition is greater than the critical percolation compositions in Fig. 3, Fig. 4 suggests that the PEDOT particles are, for the most part, uniformly dispersed as ‘islands’ in the polymer electrolyte matrix, with little evidence for ‘necklaces’ of directly contacting particles. Indeed, composites with PANi^11,12 or PEDOT contents in excess of ca. 45 wt% (in which considerable particle–particle contacts might be anticipated) have high conductivities but do not allow field control of the microwave transmission/reflection. On the other hand, we have already reported significant changes in microwave impedance in PANi composites containing as little as 10 wt% of the CP in the PEO matrix.¹² This indicates that field-controlled microwave switching requires a low degree of particle–particle contact. However, local contacts could assist in propagating the applied field across the film. It may therefore be concluded that the process takes place by both particle polarisation and percolation processes.

The redox components M/Mⁿ⁺ are also essential for the observation of field-controlled behaviour and the presence of the supporting electrolyte, LiBF₄, is required for rapid, reproducible and sustainable cycling, particularly in the case of ‘washed’ composites. However, from the single cell experiments, we conclude that the redox processes may take place only on the surfaces of the particles, particularly during the higher frequency application of the alternating fields.

These experimental observations suggest a tentative schematic model for the composite material, as shown in Fig. 9. In the absence of the applied field, the core of each particle remains in the higher conductivity state, but the skin adopts the low conductivity form as a consequence of reduction by the metal [Fig. 9(a) and eqn. 1]. In the presence of the applied field, neighbouring particles become polarised in the direction of the field [Fig. 9(b)]. [It is significant that we have observed no microwave activity with the polymeric counterion polystyrene sulfonate, which would be unable to migrate as required by Fig. 9(b).] If the polarisation is sufficiently strong, this should result in electron transfer across the interface to the transition metal ion. The stabilisation of the conducting form in a given particle [Fig. 9(c)] in the CP skin layer should be promoted by the ready availability of counter ions in the polymer electrolyte phase, as well as by the corresponding polarisation within adjacent particles and in the surrounding medium. Such a process might commence in particles close to an electrode, but rapidly percolate across the material. Long sequences of similarly polarised particles with appropriately polarised double layers and hopping between redox states in the interparticle medium may allow uninterrupted long range displacements of charge by the incident microwaves. Following removal of the field, thermal disorder would allow random polarisation of the particles and the return of the electron to the conjugated polymer particles.

Fig. 9 Schematic representations of proposed mechanism for changes in composite impedance with applied field: (a) at zero field; (b) showing particle polarisation under applied field; (c) electron transfer from particle to redox metal ion under applied field.

The role of the surface skin of the particles is perhaps reflected in the differences in the behaviour of the ‘washed’ and ‘unwashed’ materials. The presence of Fe³⁺/Fe²⁺ in the polymer electrolyte and at the surface layer of the ‘unwashed’ particles should promote electron transfer across the interface, bringing about a more resistive surface skin in the ‘off’ state. With redox ions in place at the interface, the electron transfer may require only minimum displacements of counterions between their roles opposing the oxidised forms of the conjugated polymer or the transition metal ion. Electron hopping between redox states in the PEO matrix should also promote charge transfer in the interparticle medium.

The plausibility of the proposed scheme will be tested by observing the anisotropy of the change in microwave transmission when polarised microwaves are incident either parallel or normal to the direction of the applied field. The influence on performance of transport promoters in the PEO medium [e.g. aluminium tris(8-hydroxyquinolinate) and arylamines] will also be investigated. The results of these experiments will be reported in due course.

Conclusions

Electrochemical and microwave measurements on a PEDOT–Cu–{PEO–Cu(BF₄)₂–LiBF₄} composite containing 26 wt% PEDOT have been carried out. The results show that the application of a small dc or low frequency ac electric field across the edges of annular samples of the composite gives rise to large, rapid and reversible changes in the dc and microwave impedances. Electrochemical switching between the “off” and “on” states occurs at applied fields as low as 2 V and with switching times of ≤250 ms. Single cell experiments indicate that only a thin layer of the PEDOT particles may be electrochemically active within the PEO medium. The largest changes in microwave impedance were observed in samples with residual FeCl₃. The composites show reproducible switching for several hundred cycles. A schematic model for the mechanism of switching, involving polarisation of individual particles and the redox reaction of the thin surface layers, creating double or multiple layers of ions in the region of the particle surfaces, is proposed. Long sequences of similarly polarised particles may be stabilised by the formation of the ion layers, allowing long range charge displacements within the incident microwave fields.

Acknowledgements

The authors wish to thank Bayer (UK) Ltd. for the generous donation of the Baytron M™ monomer and the Daiso Company for the generous donation of the “amorphous PEO”. We also thank Mr E. Giri of the Department of Engineering Materials, University of Sheffield, for his assistance with the optical microscopy. We gratefully acknowledge the EPSRC and QinetiQ for financial support through grant number GR/M88754. We would also like to acknowledge the assistance of the ORS scheme for the award of a studentship to R. Z.

References

1. J. Roncali, Chem. Rev., 1992, 92(4), 711–738 and references therein.
2. B. Chambers, Smart Mater. Struct., 1997, 6, 521–529.
3. B. Chambers, Smart Mater. Struct., 1999, 8, 64–72.
4. B. Chambers and K. L. Ford, Electron. Lett., 2000, 36(15), 1304–1306.
5. A. Barnes, K. Lees, P. V. Wright and B. Chambers, Proc. SPIE-Int. Soc. Opt. Eng., 1999, 3675, 139–149.
6. A. Barnes, K. L. Ford, P. V. Wright, B. Chambers, C. D. Smith, D. A. Thompson and F. Pavri, Proc. SPIE-Int Soc. Opt. Eng., 2000, 4073, 109–120.
7. F. Jonas and G. Heywang, Electrochim. Acta, 1994, 39(8–9), 1345–1347.
8. R. Corradi and S. P. Armes, Synth. Met., 1997, 84, 453–454.
9. R. M. Scarisbrick, J. Phys. D, 1973, 6, 2098.
10. B. Wessling, Proceedings of the Conductive Polymers Conference 1992, University of Bristol, RAPRA Technology Ltd, Shrewsbury, UK, 1992, paper 4.
11. A. Despotakis, Mixed Polymer Conductors for Control of Microwave Reflective Surfaces, PhD thesis, University of Sheffield, UK, 1998.
12. P. V. Wright, B. Chambers, A. Barnes, K. Lees and A. Despotakis, Smart Mater. Struct., 2000, 9, 273–279.

---