Solution-Processable Redox-Active Polymers of Intrinsic Microporosity for Electrochemical Energy Storage

Redox-active organic materials have emerged as promising alternatives to conventional inorganic electrode materials in electrochemical devices for energy storage. However, the deployment of redox-active organic materials in practical lithium-ion battery devices is hindered by their undesired solubility in electrolyte solvents, sluggish charge transfer and mass transport, as well as processing complexity. Here, we report a new molecular engineering approach to prepare redox-active polymers of intrinsic microporosity (PIMs) that possess an open network of subnanometer pores and abundant accessible carbonyl-based redox sites for fast lithium-ion transport and storage. Redox-active PIMs can be solution-processed into thin films and polymer–carbon composites with a homogeneously dispersed microstructure while remaining insoluble in electrolyte solvents. Solution-processed redox-active PIM electrodes demonstrate improved cycling performance in lithium-ion batteries with no apparent capacity decay. Redox-active PIMs with combined properties of intrinsic microporosity, reversible redox activity, and solution processability may have broad utility in a variety of electrochemical devices for energy storage, sensors, and electronic applications.


General methods and equipment
Commercially available reagents were used without further purification. Multi-walled carbon nanotubes (outer diameter: 5-15 nm; length 10-30 µm) were purchased from Jiangsu XFNANO Materials Tech. Co.,Ltd. All reactions using air/moisture sensitive reagents were performed in oven-dried apparatus under a nitrogen atmosphere.
Scanning electron microscopy (SEM) was performed using a LEO Gemini 1525 FEGSEM for fresh electrodes or a Zeiss Gemini Sigma300 for post-cycling electrodes. Samples were coated with a thin layer of chromium or gold. Infrared spectroscopy was performed on a Perkin-Elmer Spectrum 100 FTIR spectrometer with polymer samples mounted on a zinc-selenium/diamond plate. Thermal analyses were performed using a PerkinElmer TGA 8000 thermogravimetric analyser. Samples were heated from room temperature to 1000 °C under flowing nitrogen at a heating rate of 10 °C min −1 . High-resolution mass spectrum (HRMS) was obtained in electron impact ionization (EI) mode on a Thermo Scientific Mat 900 XP double focus sector mass spectrometer. ToF-SIMS analysis was performed using an ION-TOF ToF-SIMS V instrument.
Lithium-ion battery cells were subjected to activation and one cycle of charging-discharging at 1C and then disassembled in a glove box to obtain electrode samples in their lithiated (discharged) state. To remove absorbed LiTFSI salts, the electrode samples were soaked in 3 mL DOL solvent three times with each time taking at least 4 hours, followed by drying under vacuum. Low-pressure gas physisorption was performed using a Micromeritics 3Flex surface characterization analyser. Each sample was degassed at 110 °C under vacuum for 12 h, and then loaded into the apparatus and in situ degassed at 110 °C for another 3 h. Nitrogen adsorption isotherms were measured at 77 K, and carbon dioxide adsorption isotherms were measured at 273 K. Dynamic methanal vapour sorption was performed using an IGA-001 gravimetric sorption analyzer (Hidden isochema) at 25 °C. Polymer samples (20-30 mg) were in situ dried in vacuo at 110 °C for at least 12 h until the mass became constant. UV-Vis spectroscopy was measured using a UV-Vis spectrometer (UV-1800, Shimadzu) with a wavelength range of 200-800 nm at an interval of 0.5 nm. High-powered decoupling magic angle spinning 13 C solid-state NMR spectra were collected using a Bruker Avance III 600 MHz instrument using an adamantane reference. A spinning rate of ~15,000 Hz was used with powder samples packed into a 3.2 mm zirconium rotor. Spectra were typically compiled from ~4,000 scans with a 6 s recycle delay. Electrode samples for solubility tests (Fig 5 a-b) were prepared by disassembling the electrodes and Celgard separator from battery cells at a given potential and soaking in 2 mL electrolyte solution overnight before UV-Vis measurements.
Single gas permeation was measured at a feed pressure of 4 bar and 22°C, using a homemade constant-volume-variable-pressure apparatus described elsewhere 1 . The Membrane sample was soaked in methanol overnight and then annealed at 110°C under vacuum for 12 h prior to measurements.

1,3,6,8-tetramethylanthracene
Anhydrous dichloromethane (40 mL, 0.63 mol) was added dropwise to a suspension of aluminium trichloride (60 g, 0.45 mol) in anhydrous m-Xylene (200 mL, 1.62 mol) at 0 o C over 1 h. The mixture was heated at 60 o C for 3 h and then at 80 o C for 5h, cooled and poured into crushed ice. The organic layer was extracted with diethyl ether and the solvent removed under vacuum at 80 o C to afford a green solid. The crude product was purified by flash chromatography over silica gel with petroleum ether (40-60 o C) to yield 1,3,6,8tetramethylanthracene (28.0 g, 0.119 mol, 19%, Lit 23% 2 ) as colourless crystals. 1   an azeotrope with toluene aided by the Dean-Stark apparatus. After cooling to room temperature, the precipitate was collected by filtration, washed with hot DMF, methanol and acetone, and then dried at 110 o C for 24 h to afford DPPMDI (3.43 g, 9.31 mmol, 93%) as yellow crystals. The chemical structure of the product was not characterized due to its poor solubility in common solvents tested, which has been observed previously 3 .

General procedure for the synthesis of polyimides
Diamine monomer and PMDA were stirred in dry m-cresol under N2 at room temperature for 15 min before being heated to 80 °C. Toluene and 1 mL iso-quinoline were added to the reaction mixture and reaction temperature was gradually raised to 200 °C in 1 h. The reaction mixture was kept at 200 o C for 4-5 h. Water formed from the cyclo-imidization reaction was removed by forming an azeotrope with toluene aided by the Dean-Stark apparatus. After cooling to room temperature, the mixture was added dropwise to methanol and the precipitate collected by filtration. After purification, the product was dried under vacuum at 110 o C for 24 h to afford the product.

PI-AQ
General procedure was followed using 2,6-diaminoanthraquinone (4.77 g, 20.0 mmol), PMDA (4.36 g, 20.0 mmol), 14 mL m-cresol, 0.2 mL iso-quinoline, and 10 mL toluene. The reaction mixture was heated at 200 o C for 5 h. After cooling to room temperature, the mixture was directly filtered to afford a dark brown powder and nearly colourless filter liquor. The crude product was purified by repeated stirring-filtration procedure in boiling DMF until the washing liquid changed from dark brown to colourless, followed by washing with methanol and acetone to yield PI-AQ (7.92 g, 18.8 mmol, 94%) as a red powder.

PIM-TMTrip-MQ/PI-Trip-MQ
Ceric ammonium nitrate (CAN, 6.0 g, 11 mmol) was added to a suspension of PIM-TMTrip-MHQ (0.5 g, 0.91 mmol) or PI-Trip-MHQ (0.50 g, 1.0 mmol) in acetonitrile (100 mL) and water (10 mL). The mixture was stirred at room temperature for 24 h, and then filtered and washed with water to afford a yellow powder in quantitative yield.

Electrochemical measurements
Ionic conductivity of the solution-cast PIM-Trip-MHQ film was measured by electrochemical impedance spectroscopy (EIS) using the potentiostat mode with a perturbation of 10 mV and a frequency range of 0.5 MHz-10 Hz. The membrane sample was soaked in 1 M LiTFSI in DOL/DME (2:1 by Vol.) for 24 h prior to measurements, and then sandwiched between two stainless steel electrodes and sealed in coin cells (Type 2032). The ionic conductivity was calculated according to the following equation: σ = L/(Rm × A), where Rm is ionic resistance, L is membrane thickness and A is the active membrane area (2.00 cm 2 ). Rm equals the intercept of the Nyquist plots with x-axis. Membrane thickness was measured by a micrometer.
Electronic conductivity of the polymers was measured by a two-probe technique using an electrochemical station (ModuLab MTS Systems). Polymer powders were well-ground into free-flowing fine powders in a mortar by a pestle, followed by drying under vacuum at 100 o C.
The powder (~0.1 g) was quickly mounted to a die (1.6 cm in diameter) and pressed using a hydraulic press equipment at 100 MPa for 1 min. The pellet was then sandwiched between two silver coated block electrodes and sealed in a coin cell (CR 2032). Chronoamperometry (CA) method was used to record the current and calculate ohmic resistance, R. The electronic conductivity was calculated according to the following equations: σ =1/ρ, R= (L/A)×ρ, where ρ is resistivity, L is pellet thickness and A is the active membrane area (2.0 cm 2 ). Pellet thickness was measured by a micrometer.
Electrode fabrication and cell assembly. Electrodes were fabricated via powder dispersion or solution processing. For the powder dispersion method, each polymer powder, Ketjen black and polyvinylidenefluoride (PVDF) binder (100 mg in total) were mixed using a pestle and a mortar with a ratio of 6:3:1 by weight (or 5:4:1 for CV measurements) with ~2 mL NMP added to form a well-dispersed slurry, which was subjected to ultrasonication for 5 minutes and then coated onto an Aluminium foil substrate. For the solution processing method, redox-active PIMs and PVDF binder were fully dissolved in NMP at 60 o C and Ketjen black was added to the polymer solution at a ratio of 6:3:1 by weight (60% of redox-active PIMs). When CNT was used, the weight ratio was 6:3:1:1 for PIM-TMTrip-MHQ, Ketjen black, CNT and PVDF. The mixture was stirred overnight and subjected to ultrasonication for 5 minutes before casting onto an Aluminium foil substrate. The electrodes were dried in an oven at 80 °C for 4 h and at 120 °C under vacuum for 24 h, and then cut into circular electrode discs with a diameter of 10 mm (mass loading: 0.3-0.7 mg cm -2 ). The electrodes were transferred into a glove box under an Ar atmosphere and assembled into half lithium-ion batteries using CR2032 coin cells.
Galvanostatic charging/discharging of the batteries was performed within a voltage window of 1.5-3.5 V (versus Li/Li + ) using a Land Battery Tester. Batteries were first activated by discharging, followed by cycling tests at 1 C.   PIMs that were initially designed as membrane materials for gas separation may contain redox-active structural units, but their theoretical capacity is too low to be useful.

This work Previous work
These PIMs were initially designed for gas separation applications  Fig. S10. Solid-state CV at varied scanning rates. All three polymers undergo reversible oxidation and reduction via a four-electron per repeat unit process. While the microporous PIM-TMTrip-MQ and PI-Trip-MQ showed well-resolved peaks, PI-AQ exhibited broad and overlapping peaks across the scanning rates. The anodic peak with the highest potential (~ 2.9 V vs. Li + /Li) was used to derive the log-log plot of peak current versus scanning rate (Fig 3c), as this anodic peak showed the least overlap with other peaks for all three polymers. Electrolyte: 1M LiTFSI in DOL/DME Fig. S11. Solid-state CV in acetonitrile electrolyte solutions. To link the micropore size of redoxactive PIMs with the size of counterions, we chose LiClO4, NaClO4, and TBAClO4 as the supporting electrolyte. The highest b values were found for the most porous PIM-TMTrip-MHQ while the nonporous PI-AQ showed the lowest b values, which is consistent with the observation when LiTFSI was used. It should be noted that we observed partial peeling-off of the coated composite electrode of PI-AQ from the working electrode in TBAClO4 electrolyte, which led to its lower peak intensity. With carbon black No carbon additives Figure S13. Prediction of lithiation pathways based on the minimum energy principle. All possible Li-binding geometries for the PIM-TMTrip-MQ-nLi (n=0, 1, 2, 3 and 4) isomers based on DFT calculations. Although they weren't included in our conformational search, optimised geometries for n=5 and n=6 repeat units are also included, as well as the energy of a lithium atom which was included in the computed binding energies. E denotes the DFT calculated energy for each geometry and BE denotes the binding energy for each possible lithiation step with respect to the preceding lowest energy lithiated isomer. Constitutiional isomers with the lowest total energy and strongest binding energy are predicted to be the preferred lithiation sites (highlighted in red and blue). The DFT calculations suggest that the first and second lithiation would take place on the triptycene carbonyls, whilst the third and fourth lithiation would take place on the PMDA unit. Atoms in grey, white, red, blue, and violet represent carbon, hydrogen, oxygen, nitrogen, and lithium, respectively. Table S1. Redox potential calculations. Computed Gibbs free energies (G) for the PIM-TMTripMQ-nLi (n=0, 1, 2, 3, 4, 5 and 6) isomers and a lithium cation are reported in the gas and solution phases. Gibbs free energy changes (∆Gred and predicted redox potentials (∆Ered) during each possible reduction are reported for both the gas and solution phases. Experimental redox potentials are also recorded for comparison.     Fabrication method: solution processing. Particle size was measured using ImageJ. The control sample is composed of carbon black and PVDF binder (3:1 by weight), while the solution-processed electrode is composed of PIM-TMTrip-MHQ, carbon black and PVDF binder (6:3:1 by weight). It should be noted the apparent particle size measured from SEM includes the intrinsic size of the particles and also that of the conductive coating layer that was sputtered prior to SEM tests to avoid sample charging.