Reversible Electrochemical Charging of n-Type Conjugated Polymer Electrodes in Aqueous Electrolytes

Conjugated polymers achieve redox activity in electrochemical devices by combining redox-active, electronically conducting backbones with ion-transporting side chains that can be tuned for different electrolytes. In aqueous electrolytes, redox activity can be accomplished by attaching hydrophilic side chains to the polymer backbone, which enables ionic transport and allows volumetric charging of polymer electrodes. While this approach has been beneficial for achieving fast electrochemical charging in aqueous solutions, little is known about the relationship between water uptake by the polymers during electrochemical charging and the stability and redox potentials of the electrodes, particularly for electron-transporting conjugated polymers. We find that excessive water uptake during the electrochemical charging of polymer electrodes harms the reversibility of electrochemical processes and results in irreversible swelling of the polymer. We show that small changes of the side chain composition can significantly increase the reversibility of the redox behavior of the materials in aqueous electrolytes, improving the capacity of the polymer by more than one order of magnitude. Finally, we show that tuning the local environment of the redox-active polymer by attaching hydrophilic side chains can help to reach high fractions of the theoretical capacity for single-phase electrodes in aqueous electrolytes. Our work shows the importance of chemical design strategies for achieving high electrochemical stability for conjugated polymers in aqueous electrolytes.

synthesized as previously reported. 1 The synthesis of monomer (bg4N) and polymers p(bg4Ng3N) and p(bg4NC16N) are also reported in section 2 of the supporting information.
Thickness measurements using AFM: Atomic force microscopy (AFM) images were taken using an Asylum Research Cypher-ES. The thicknesses of the dry films were measured by imaging across a razor blade scratch in tapping mode using 325 kHz AFM probes (HQ:NSC15/Pt, µMasch). Hydrated thickness was determined by depositing 100 µL of degassed 100 mM NaCl solution onto the sample and imaging over the same scratch in contact mode using ContGB-G tips (BudgetSensors) with a force set point of 6 nN. Line profiles (4 µm wide) were taken perpendicular to the scratch for plotting. All thicknesses were averaged from at least 6 locations on at least 2 different films.
Swelling measurements were performed as follows. First, we recorded the QCM-D response of the bare Au sensors in the air, followed by injection of a 0.1 M NaCl aqueous solution into the chamber. This resulted in large shifts in the frequency (f) and dissipation of energy (D), due to the density differences between air and electrolyte, which must be excluded from the swelling percentage calculation. The measurements were then stopped, the sensors were removed, and the polymer layers were spin-coated directly on the same sensors from 4 mg/mL solutions in chloroform at 1000 rpm. The absolute f value for each polymer coated sensor was obtained both in air and in a 0.1 M NaCl aqueous solution, after the f signal was perfectly flat (i.e., Δf <0.5 Hz/2 min), assuring that the system is in equilibrium. We then compared the absolute difference in f for multiple overtones between the bare sensor and the polymer coated sensors, both in air and in the electrolyte, by using the function "stitched data" of Q-soft software. This function compares the selected datasets based on the raw frequencies measured and excludes the effect of the different densities between the two media. Thus, the difference of the f values of the stitched data is directly analogous to the mass absorbed by the polymer (m) in both media, which is calculated by using the Sauerbrey equation: The thickness of the films in air and the electrolyte is calculated accordingly ( Figure S6). For with M being the moelcualr weight of one repeat unit and n = 2.
Spectroelectrochemical measurements: The electrochemical cell was a quartz cuvette with transparent windows which enabled simultaneous optical spectra acquisition through a UV-vis spectrometer (OceanOptics USB 2000+) which collected the transmitted light through the sample from a tungsten lamp used as probe light source. 2 The electrochemical measurements were carried out with an IviumStat potentiostat, using a three electrode setup with a Ag/AgCl reference electrode (eDAQ leakless Ag/AgCl reference electrode) and a Pt-mesh as the counter electrode.
OECT measurements OECT test chips were prepared following previously reported microfabrication techniques. 5 The OECT channels containing the p([g7:a]NDI-g3T2) series were fabricated by drop casting from chloroform (5 mg/mL) at room temperature, while the p([g7:a]NDI-T2) series were spincoated. This was followed by patterning via peeling a sacrificial parylene layer and a rinse in deionized water. OECTs were gated with aqueous 0.1 M NaCl using an Ag/AgCl pellet as the gate electrode. Electrical characterization (output, transfer, and pulsed stability) of the OECTs were carried out using NI source-measure units controlled by custom LabView code. μC* was calculated from the slope of the transfer curves, which is the gate transconductance (gm = ∂Id/∂Vg), using the relation gm = Wd/L × μC* × (Vt -Vg).

Density functional theory calculations
All quantum chemical calculations (Density functional theory (DFT) and time-dependent DFT (TDDFT)) in this study were performed using Gaussian16. 6  Further computational details are included in the Supporting Information, section 9.  10 , the synthesis is summarised in Figure S4

13-(iodomethyl)-2,5,8,11,15,18,21,24-octaoxapentacosane. A 2-neck oven dried 250 mL RBF
was charged with NaI (4.88 g, 32.6 mmol, 2.0 eq.) and dispersed in acetone (100 mL) under a nitrogen atmosphere. 13-(2,5,8,11-tetraoxadodecyl)-2,5,8,11-tetraoxatetradecan-14-yl 4methylbenzenesulfonate (which was synthesized according to previously reported protocols) 11 (9 g, 16.3 mmol, 1.0 eq.) was added dropwise via a syringe. The reaction was heated to 55 °C and stirred overnight. Upon cooling to room temperature, the mixture was poured into water, washed with brine, and extracted with DCM. The organic layer was separated, dried over MgSO4 and the solvent was removed under reduced pressure to yield a crude pale-yellow oil. The crude was purified by column chromatography on silica using a solvent mixture of ethyl acetate:methanol (9:1) as the eluent to afford the title compound as a colorless viscous oil (7.9 g, 95%).   g, 10.13 mmol, 4.0 eq.) was injected into the reaction in a single portion and the mixture was left to stir for an additional 4 hours. Upon cooling to room temperature, the reaction was poured into water and acidified to pH using aqueous 2M HCl. A saturated ammonium chloride solution was added to help separate the organic layer upon the addition of DCM. Once separated the organic phase was dried over MgSO4 and solvent removed to yield a deep purple sticky solid. The crude product was columned on silica twice with a DCM:Acteone (9:1) eluent system and a third time using DCM:Acetone (8:2) to isolate the desired blue product.
The title compound was isolated as a blue solid (347 mg, 13%).      with M being the molecular weight of one repeat unit as shown in Table S1.

Comparison of charging of a neat and pre-cycled polymer thin film
To investigate the changes of the cyclic voltammogram and absorption spectrum during the first and second charging/discharging cycle, the electrochemical and spectroelectrochemical properties of a freshly prepared polymer thin film were compared to the same film after precycling. The pre-cycled polymer film was washed with with DI water and dried in ambient conditions (no heating). A similar response between the first and second scan was observed for the pristine polymer and the pre-cycled polymer.

eQCM-D measurements
We attempted conducting measurements at low oxygen concentrations to limit faradaic side reactions between the charged polymer and molecular oxygen, however only with limited success, as only low coulombic efficiencies are observed during the cycling ( Figure S24 and Figure S25).

eQCM-D measurements of the p([g7:a]NDI-g3T2) series to <-0.4 V vs. Ag/AgCl
We also conducted measurements at potential < -0.4 V vs. Ag/AgCl and observed a large dissipation of energy values of more than 1500 *10 -6 J for the low glycol content polymer, indicative of the formation of a hydrogel-like state of the polymer under high degree of charging. We believe these changes are more drastic for the polymers containing more glycolated side chains and one reason why irreversible changes in the eQCM-D signal are observed at high degree of charging.

Method
Molecules in their neutral, singly and doubly reduced states were optimized at the DFT level (B3LYP /6-31g(p,d)). Excited stated TD-DFT calculations were performed on optimized geometries of neutral and charged (reduced) molecules. All calculations shown in this study were performed using Gaussian16. For the polymer series p([g7:a]NDI-g3T2) and p([g7:a]NDI-T2), we approximated the polymer chain to be dimers (spectroscopy calculations) and the calculated structures are shown in Figure S36 and S37.

Results
The calculated spectra of the charged species supported the interpretation of the changes of the absorption spectrum of the polymers during polaron and bipolaron formation. We also present calculations for the excited states of (aNDI-gT2)2 and (aNDI-T2)2 as a comparison to (gNDI-gT2)2 and (gNDI-T2)2 to demonstrate that replacing glycol side chain with alkyl side chains on the NDI unit has a little impact on the calculated absorption spectra.       Table S6. Energies and oscillator strengths calculated for (aNDI-T2)2, (aNDI-T2)2 -2 and (aNDI-T2)2 -4 in water (PCM) up to 3.1eV (400nm). The collective entry in the columns indicates group of states in the given energy interval and sum of oscillator strengths of these states.