Structure of electro-polymerized PEDOT
The SEM images (Fig. 1) of the PEDOT surfaces prepared by potentiostatic electro-polymerization in 0.01 mol dm−3 EDOT and 0.01 to 1.0 mol dm−3 KCl show that PEDOT forms a granular and porous structure on vitreous carbon. These porous grains have a high surface area and consequently a high number of accessible positive charged doping positions for chloride ions.
The structure of the PEDOT films, prepared in an increasing concentration of KCl in the electro-polymerization solution, strongly depends on the doping anion concentration. The density of the grains increases significantly with increasing chloride concentration up to 0.1 mol dm−3 KCl in the electro-polymerization solution (Fig. 1a–c). The structure and grain density remains similar for electrodes prepared in solutions with 0.1 to 1.0 mol dm−3 KCl (Fig. 1c, d).
The distribution of the grains, which were polymerized in aqueous solution containing 0.1 mol dm−3 KCl and 0.01 mol dm−3 EDOT, is very uniform at the surface of the vitreous carbon substrate (Fig. 1c). The randomly distributed agglomerates of PEDOT have an average size of 0.2 to 2 μm. The formation of three-dimensional grains on top of an initial uniform layer of the conductive polymer is typical for the material because of the higher adhesion of the agglomerates on a first thin layer of PEDOT than on the native vitreous carbon substrate [40,41,42]. The average ratio of the proportion of sulfur, which represents one monomer unit of EDOT per one chloride ion, was measured by EDX. The ratio corresponds to a degree of doping of 0.33 and is in accordance with literature value .
Undoped PEDOT (Fig. 1a), which was polymerized in 0.01 mol dm−3 EDOT and in the absence of KCl, is characterized by a smooth surface with very small and separated grains. A small amount of chloride ions (0.01 mol dm−3) (Fig. 1b) increases the growth of the grains, covering the whole surface (Fig. 1b–d). Agglomerated grains as well as flat structures were formed over the whole surface of the carbon electrode. The diffusion zones around the hemispherical nuclei have started to overlap, forming the agglomerated grain structure until a critical height of the layers is reached. The height might depend on the concentration of the doping anion. Therefore, the growth continues laterally and flat structures are formed. The formation of a dense structure of grains is attributable to the fact that the conductivity of the solution increases with the chloride concentration. The structures polymerized at a low chloride concentration (0.01 mol dm−3 KCl) show presumably the initial stage of the PEDOT layer formation. However, the polymerization continues if the oxidation potential remains sufficiently high to oxidize further EDOT to the radical cation. With an insufficient conductivity, the diffusion zones increase until the oxidation current cannot be maintained and the polymerization stops.
Cyclic voltammetry of PEDOT
The cyclic voltammograms (Fig. 2) of the PEDOT films, prepared in different concentrations of the doping component, show a clear influence of the amount of doping component on the capacity. Each cyclic voltammogram shows a constant capacity beyond cycle 10. Therefore, cycle 10 was used to compare peak positions and capacity.
The cyclic voltammogram of the PEDOT film prepared in the lowest KCl concentration of 0.01 mol dm−3 (Fig. 2 (a)) shows neither anodic nor cathodic peaks. Thus, under these conditions, the PEDOT film seems to be electrochemically inactive. The PEDOT film polymerized in 0.1 mol dm−3 KCl (Fig. 2 (b)) is characterized by significant anodic and cathodic peaks as well as high capacity. A further increase of the doping component, to 0.5 mol dm−3 (Fig. 2 (c)) and 1.0 mol dm−3 KCl (Fig. 2 (d)) solution, shows lower currents compared to the film prepared in 0.1 mol dm−3 (Fig. 2 (b)) KCl. The shapes and the currents and consequently the capacities of the films deposited in 0.5 and 1.0 mol dm−3 KCl (Fig. 2 (c, d)) do not show significant differences. From these results, it can be suggested that the amount of chloride ions incorporated in the PEDOT film reaches a maximum. Regarding the decreasing current and capacity from the solutions containing 0.5 and 1.0 mol dm−3 chloride, it is assumed that an increasing saturation of the positive charges with doping anions in the polymer backbone transforms PEDOT into a less conductive state. The degree of doping, determined by Mohr titration [37, 38], increases with an increasing chloride concentration in the electro-polymerization electrolyte from 0.1 to 0.33. For the PEDOT film polymerized in 0.01 mol dm−3 EDOT and 0.1 mol dm−3 KCl, a degree of doping of 0.27 was determined, along with the highest capacity in the cyclic voltammogram (Fig. 2 (b)). The PEDOT films polymerized in 0.01 mol dm−3 EDOT and 0.5 mol dm−3 as well as 1.0 mol dm−3 (Fig. 2 (c, d)) have a degree of doping of 0.33, which agrees with the EDX results.
Performance of PEDOT in Lewis acidic ionic liquid
The cyclic voltammograms of PEDOT electrodes (prepared with 0.01 mol dm−3 EDOT and 0.1 mol dm−3 KCl) in a Lewis acidic ionic liquid from 0 to 2.7 V vs Al/Al(III) were recorded at 10, 25, and 50 cycles (Fig. 3). A vitreous carbon electrode was used as counter electrode and an aluminum wire as reference electrode. The oxidation and reduction reactions of the film in the Lewis acidic ionic liquid are characterized by four anodic and three cathodic waves, indicating different levels of doping and de-doping of the polymer. The anodic broad waves arise at 0.7 V (I), 1.45 V (II), 1.9 V (III), and 2.2 V (IV) vs Al/Al(III). The doping at 2.2 V (IV) occurs with the decomposition of the ionic liquid to chlorine gas. The cathodic wave around 1.85 V (V) correlates with the anodic wave at 1.9 V (III) vs Al/Al(III) and indicates a reversible doping and de-doping reaction. The second cathodic wave appears at 1.5 V (VI) and the third wave arises at 0.7 V (VII) vs Al/Al(III).
With an increasing cycle number, the anodic and cathodic currents decrease. A swelling of the PEDOT film was observed in the ionic liquid during doping and de-doping, which appears as blue cloud-like film around the electrode that causes the decrease in current, as there are less available active sites to accommodate chloride ions in the PEDOT backbone. The polymer film loses its adhesion on the vitreous carbon substrate and is partially detached. The stability of the polymer on the substrate can be improved by using rougher surfaces and three-dimensional carbon substrates to deposit PEDOT or by using pure porous PEDOT electrodes, as they are more stable due to the higher number of doping positions in contrast to flat thin films of PEDOT on the vitreous carbon surface. This approach will be a focus of future work.
Determination of the anion doping species
PEDOT films doped with chloride ions were formed on a gold-coated crystal quartz electrode of 0.2 cm2 area by polymerization of 0.01 mol dm−3 EDOT in 0.1 mol dm−3 KCl aqueous solution at a constant potential of 1.2 V vs Ag/AgCl for 600 s. The quartz with the PEDOT film were dried and transferred into a Lewis acidic ionic liquid and cycled from 0 to 2.0 V vs Al/Al(III) (Fig. 4 (a)). Simultaneously, the changes of the resonance frequency  (different resonance frequencies of the loaded and unloaded quartz crystal) (Δf) of the quartz crystal were measured (Fig. 4 (b)).
PEDOT could also been doped with AlCl4
− and Al2Cl7
− ions, which are present in the ionic liquid electrolyte, during cycling. The changes in mass (Δm), calculated by the changes in resonance frequency (Δf) with the Sauerbrey equation (Eq. 6), and the transferred anodic and cathodic charges (ΔQ) (Table 1) provide information about the doping species.
The regions I and II (Table 1) indicate the doping potential window (ΔE) of PEDOT. The mass change (Δm) of the polymer, determined by quartz crustal microbalance measurement (Fig. 4 (b)), in these regions, refers to doped anion species. The transferred charge (ΔQ) was measured simultaneously with cyclic voltammetry (Fig. 4 (a)). The specific charge (Q
spec) was calculated by the ratio of ΔQ and Δm. The cyclic voltammogram (Fig. 4 (a)) shows anodic waves from 0.4 to 0.8 V (I) and 1.0 to 1.9 V (II) vs Al/Al(III). The frequency (Δf) decreases approximately linearly at these anodic potentials (Fig. 4 (b)), which indicates the anion doping (p-doping) . The subsequent cathodic waves from 1.4 to 1.8 V (III) and 0.35 to 0.65 V (IV) vs Al/Al(III) show the de-doping reaction, which also represents the increase of the frequency. PEDOT can be also doped with cations (n-doping) when the polymer carries negative charges. It is assumed that the cationic species EMIm+ is inserted in the PEDOT film from 0 to 0.4 V vs Al/Al(III). The oxidation shows an increase of the frequency in the same potential range, indicating a decrease of mass or removal of the cationic species.
The doping and de-doping reactions occur in the potential window 0.4 to 1.9 V vs Al|Al(III) (Table 1). As the damping changes (Δw) (Fig. 4 (c)) of the quartz crystal in this potential window are in the same order of magnitude as the frequency changes (Fig. 4 (b)), we can use the Sauerbrey equation (Eq. 6) to calculate the change of mass from the change of frequency. The calculated molar mass (M
doped anion) for the doped anion species are 140 and 87 g mol−1 (Table 1), indicating that the doping proceeds via a mixture of Cl− (M
theo = 35.6 g mol−1) and AlCl4
theo = 169.4 g mol−1) (Table 1). It is assumed that bulky chloroaluminate ions are partially trapped and irreversibly inserted in the polymer film and causing a damage with subsequent detachment of the PEDOT film, which was observed as blue cloud-like film in the electrolyte (“Performance of PEDOT in Lewis acidic ionic liquid” section). The selective doping of PEDOT with smaller chloride ions might be realized in a Lewis basic ionic liquid, which contains an excess of free chloride ions.
Performance of aluminum dissolution and deposition in Lewis acidic ionic liquid
A Lewis acidic ionic liquid of [EMIm]Cl and AlCl3 contains an excess of AlCl3, which enables the electro-deposition of aluminum. The cyclic voltammogram (Fig. 5) shows the tenth cycle of the deposition and dissolution of aluminum on aluminum plate (4.5 cm2) between −0.5 V and 0.5 V vs Al/Al(III).
The reduction of the complex aluminum ion (Al2Cl7
−) starts at −0.175 V vs Al/Al(III). The deposition occurs with some over potential due to the nucleation process, which has been reported for many metals in ionic liquids . The aluminum oxidation is characterized by a wide peak with a maximum around 0.25 V vs Al/Al(III). The stripping peak shows an anodic charge of 5.3 mC cm−2 whereas the charge of the cathodic electro-deposition is 7 mC cm−2 giving a stripping efficiency of 76%. The stripping efficiency remains approximately constant with the cycles of the cyclic voltammetry. It is assumed that the reduced stripping efficiency is caused by the formation of a porous deposit of ionic liquid compounds on the aluminum electrode surface (Fig. 6), inhibiting the oxidation of aluminum. However, the deposition and dissolution of aluminum in a Lewis acidic ionic liquid is acceptable and can be improved  and therefore suitable for the charging and discharging reactions of the PEDOT-aluminum system.
The deposition of aluminum in the Lewis acidic ionic liquid shows a very porous surface (pore diameter ~0.4 to 1.4 μm) (Fig. 6a) on the aluminum electrode and circular deposits with a diameter of ~28 μm (Fig. 6b). It is proposed that the circular deposits are organic decomposition products of the ionic liquid, because the cyclic voltammetry was carried out until −0.5 V vs Al|Al(III). The aluminum deposition occurs at potentials lower than −0.2 V vs Al|Al(III) and the ionic liquid decomposition process starts at −0.5 V. Furthermore, the porous structure may be a thin film of organic depositions or decomposition products as well, which are permeable for aluminum ions.
Charge and discharge behavior of PEDOT-aluminum battery
The reference sample, prepared in 0.01 mol dm−3 EDOT and 0.1 mol dm−3 KCl supported on vitreous carbon plate (4.5 cm2), was introduced as the positive electrode and metallic aluminum plate (4.5 cm2) was used as negative electrode in the cell. The open circuit potential was measured before the galvanostatic cycling and before every charge and discharge cycle (Eqs. 1 and 3). The OCP before the first cycle is between 2.1 and 2.3 V and before cycle 2 to 10 between 1.3 and 1.6 V. The battery was galvanostatically discharged (Fig. 7 (a, c)) at −0.02 mA cm−2 and charged (Fig. 7 (b)) at 0.2 mA cm−2.
The average cell voltage of the first discharge is 2 V for 124 mA s (Fig. 7 (a)). The first charge voltage is around 2.2 to 2.3 V for 1600 mA s (Fig. 7 (b)). The beginning of the charging curve after 1500 s shows a steep slope at 1.8 to 1.9 V. Assuming that the aluminum deposition is at 0 V vs Al/Al(III), the slope and the voltage plateau correlate with the peaks at 1.9 to 2.0 V and 2.2 V vs Al/Al(III) in the cyclic voltammogram (Fig. 3 potential regions III and IV), indicating the doping of chloride ions in the polymer. The charging cycle (Fig. 7 (b)) was terminated at 2.35 V because the electrolyte started to decompose. The subsequent discharge curve (Fig. 7 (c)) is characterized by a decreasing cell voltage from around 1.2 to 0.8 V for around 170 mA s and terminated at 0.5 V. The suggestion is that the first discharge is accompanied by the de-doping of chloride ions and the subsequent charge is characterized by the doping of AlCl4
− in the polymer, influencing the OCP, discharge voltage, and capacity. In addition, the insertion of bulky ions like AlCl4
− in the polymer could change the structure of the polymer film and cause a damage of the film and a loss of active mass.
The battery-related characteristics (Table 2), which are based on the deposited mass of PEDOT, the active mass of aluminum, applied discharge current, and discharge time, show a specific capacity of 84 Ah kg−1. This is 2.3 times higher than the maximum theoretical value. This discrepancy can be explained if it is assumed that the PEDOT electrode behaves like a capacitor and battery (also known as hybrid-capacitor). The sum of the non-faradaic cathodic-specific charges (52 Ah kg−1), obtained from the second cycle of the cyclic voltammogram (Fig. 3) of PEDOT in Lewis acidic ionic liquid, and the maximum theoretical capacity (36 Ah kg−1) equals approximately to the measured specific capacity (84 Ah kg−1). This additional capacity is probably provided by the capacitive double layer of anion species (Cl−, AlCl4
−) and cationic species (EMIm+) of the electrolyte.
Influence on the charge and discharge behavior with porous electrolyte support
In order to limit the loss of PEDOT mass due to swelling and detachment of the film from the electrode, a porous support for the electrolyte was used. The aim was to keep the PEDOT film as close as possible to the vitreous carbon electrode. The pore size was between 0.03 and 0.8 mm to allow the ions in the ionic liquid to move freely and maintain the ionic conductivity. In this new configuration, the PEDOT-aluminum system in Lewis acidic ionic liquid was galvanostically charged and discharged at 0.2 and −0.02 mA cm−2, respectively. The second discharge cycles without (Fig. 8 (a)) and with (Fig. 8 (b)) electrolyte support were compared.
The discharge reaction for the cell with the porous electrolyte support is characterized by a significant longer discharge plateau over 600 mA s between 1.1 and 0.9 V (Fig. 8 (b)). The discharging was terminated at 0.9 V because of the low practical use of this cell potential. The cell with electrolyte support was charged between 2.2 and 2.3 V for 1600 mA s similar to the charging cycle of the cell without electrolyte support (Fig. 7 (b)). The coulombic efficiency is 38% for the cell with electrolyte support. The cell shows a discharge time, specific charge and capacity that are approximately three times larger (Table 2) than the cell without electrolyte support. The calculated energy densities, based on the active mass of aluminum and deposited mass of PEDOT, are 84 Wh kg−1 without electrolyte support and 228 Wh kg−1 with electrolyte support, which is in the order of nickel-metal-hydride batteries (60–100 Wh kg−1) and lithium-based battery systems (100–256 Wh kg−1), respectively . The improved performance is due to the fact that the film was kept in place by the electrolyte support and the behavior of PEDOT as hybrid capacitor. A more porous conductive polymer electrode reaches a higher capacity due to proportional capacitance to sample volume. Future work will aim at a deeper understanding of whether the measured values were caused by the battery system or are an effect of the porous electrolyte support.