/Energy storing bricks for stationary PEDOT supercapacitors

Energy storing bricks for stationary PEDOT supercapacitors

Conversion of a fired brick’s α-Fe2O3 to a PEDOT coating

Deposition of PEDOT nanofibers is initiated by dissolving α-Fe2O3 at 160 °C with HCl vapor; this process liberates Fe3+ ions, promotes hydrolysis and initiates precipitation of colloidal 1D FeOOH nuclei (Fig. 1a). As previously reported, partially dissolved FeOOH nuclei serving as templates oxidize 3,4-ethylenedioxythiophene (EDOT) monomer vapor and control oxidative radical polymerization13. Control of both stoichiometry and reaction time leads to PEDOT coatings with different thicknesses (Fig. 1b). We advance previous findings by demonstrating here that the synthesis utilizes two potential polymerization initiators, i.e., oxidant (Fe3+) and acid (HCl) where the former leads to oxidative radical polymerization and the latter, to acid-catalyzed polymerization (Supplementary Fig. 1). We readily control the polymerization mechanism as an acid-catalyzed polymerization typically produces nonconductive oligomers stemming from active chain termination16. Only PEDOT synthesized via oxidative radical polymerization exhibits long conjugation length, ordered chain packing, low electrical resistance, as well as high chemical and physical stability13,14,17,18.

Fig. 1: Deposition of a nanofibrillar PEDOT coating on brick.
figure1

a In a one-step reaction, a brick’s α-Fe2O3 microstructure is partially dissolved by acid vapor to liberate Fe3+, promote hydrolysis and precipitation of FeOOH spindles that control oxidative radical polymerization. As previously reported, monomer vapor reacts with partially dissolved FeOOH nuclei resulting in preferential directional growth of high aspect ratio PEDOT nanofibers13. b The thickness of a PEDOT coating is controlled by reaction time and stoichiometry enabling a reaction to generate surface-polymerized PEDOT-coated bricks (core/shell architecture) or fully polymerized bricks (monolithic PEDOT architecture). c A reaction diagram shows the competition between acid-catalyzed polymerization and oxidative radical polymerization mechanisms present in our reactions.

In our polymerization mechanism, the acid concentration determines both dissolution rate and synthetic pathways. A high H+ concentration facilitates the liberation of Fe3+ as well as oxidative radical polymerization while promoting acid-catalyzed polymerization that consumes EDOT and results in PEDOT of low conjugation length (Fig. 1c). To investigate the effect of H+ concentration, different volumes of concentrated HCl are added to the reactor and the HCl vapor concentration is calculated by dividing moles of HCl with reactor volume (assuming total evaporation). Using a HCl vapor concentration less than 4.8 mM leads to an incomplete reaction because both oxidative radical polymerization and acid-catalyzed polymerization are impeded (Supplementary Fig. 2a). Increasing concentration to 14 mM liberates Fe3+ and promotes oxidative radical polymerization resulting in PEDOT of low electrical resistance, whereas concentrations above 14 mM activate the acid-catalyzed polymerization pathway resulting in uncontrolled reactions (Supplementary Fig. 2b). When oxidative radical polymerization dominates, the EDOT vapor concentration determines the thickness of a polymer coating and its electrical resistance (Supplementary Fig. 2a, c).

Synthesis starts when a brick and chemical reactants are heated together in a sealed vessel; chemical and physical changes on a brick are monitored by collecting sample aliquots at different time intervals. Notably, there is no color change on a brick’s red surface during the first 4 h of reaction because brick dissolution is the rate-limiting step in our mixed synthetic mechanism that consists of (1) evaporation, (2) dissolution, (3) hydrolysis, and (4) polymerization. A blue PEDOT coating is visible on a brick 4 h after initiating a reaction and its thickness increases inversely proportional with electrical resistance until the end of reaction at 14 h. An extended polymerization time increases the polymer coating’s two-point probe electrical resistance (Supplementary Fig. 2d) because PEDOT loses dopant during heating18; fortunately, post-synthetic doping lowers the electrical resistance. Our synthesis produces a 400 µm thick nanofibrillar PEDOT coating (2.8 wt%) exhibiting 2Ω two-point probe electrical resistance and nanofibers characterized by a ~30 µm length and ~190 nm diameter (Supplementary Fig. 3a–d). Polymer, purified by repeated rinses in methanol, is comprised of S, C, O, and doped by Cl in situ during polymerization as shown in energy-dispersive X-ray spectra (Supplementary Fig. 3e).

Our synthesis is general and applicable to different types of bricks. Here, three types of bricks (type 1–3) with different gravel (SiO2) sizes and porosities are investigated (Fig. 2). Type 1 brick shows the most open microstructure (Fig. 2a) that facilitates reagent vapor diffusion, therefore utilized for the synthesis study in Fig. 1 and Supplementary Figs. 13. Powder X-ray diffraction of pulverized type 1 brick shows that SiO2 is the major phase while α-Fe2O3 and Al2O3 are minor phases (Fig. 2c). These three types of bricks possess similar inorganic composition and concentration as shown by powder X-ray diffraction patterns (Fig. 2d). Porosity differences between brick types stem from various gravel sizes and manufacturing variables such as water content before sintering, sintering temperature and duration. Vapor-phase synthesis produces a contiguous polymer coating over the entire brick surface because dissolution generates an aqueous Fe3+ layer that coats inert gravel sites. Under identical reaction stoichiometries and time, PEDOT nanofibers of high aspect-ratio coat all types of bricks homogeneously (Fig. 2b). Our polymer coating technology is scalable (Fig. 2e) and patternable (Supplementary Fig. 4a) as demonstrated using type 1 brick. Nanofibrillar PEDOT coatings on all types of bricks show linear current–voltage curves with slopes that indicate ohmic behaviors and similar electrical resistances (~7Ω) (Fig. 3a). Note that a two-point probe multimeter measurement leads to a lower value (2Ω) because of a wide tip diameter that lowers contact resistance.

Fig. 2: Nanofibrillar PEDOT-coatings on different types of bricks.
figure2

a Three types of red fired bricks are utilized for synthesis. All types show homogeneous red-orange color under the optical microscope, indicating uniform α-Fe2O3 distribution. White spots on bricks (left top inset) show the gravel (SiO2) sizes increasing from type 1 to type 3 brick. Scanning electron micrographs exhibit different porosity for these three types of bricks with pore sizes decreasing from type 1 to type 3. b After synthesis, all types of bricks are homogeneously coated by PEDOT nanofibers of similar aspect ratio. c Powder X-ray diffraction pattern for type 1 brick shows a composition comprised of SiO2, α-Fe2O3, and Al2O3. d Type 2 and type 3 bricks show similar patterns and crystalline components. The sharper and stronger peak at 2θ = 28° from type 2 brick is due to its higher Al2O3 content or crystallinity. e Synthesis is scalable to decimeter-sized bricks.

Fig. 3: Analysis of PEDOT coatings on various bricks.
figure3

a Linear current–voltage curves show ohmic behaviors of nanofibrillar PEDOT coatings possessing comparable resistances. b Water absorption experiments on bricks enables studying their open pore structure and all bricks show insignificant changes in water absorption after PEDOT coatings with type 1 brick absorbing the most water due to a higher porosity. c A nanofibrillar PEDOT coating exhibits superior adhesion on type 1 brick versus other types during Scotch tape tests. Unfortunately, the commercial product PEDOT:poly(styrene sulfonate) delaminates completely after coated on type 1 brick. d Cross-sectional scanning electron micrographs show the embedded PEDOT network in type 1 brick leading to robust PEDOT adhesion. Type 2 (e) and type 3 (f) bricks show distinct boundaries between PEDOT and brick that are prone to delamination.

In order to study the effect of polymer coatings on the porosity of bricks, water absorption experiments are carried out before and after synthesis. A PEDOT-coated type 1 brick absorbs the most weight of water among the three types of polymer-coated bricks due to its open pore structure and large pore size and volume (Fig. 2a). A nanofibrillar PEDOT coating on type 1 brick exhibits minimal delamination via Scotch tape tests whereas a coating of the commercial product PEDOT:poly(styrenesulfonate) peels off readily and completely. Note that both type 2 and type 3 bricks show partial delamination (Fig. 3c) of our polymer coating because of a semi-closed brick microstructure (Fig. 2a) that impedes vapor reactant diffusion (Fig. 3e, f). Cross-sectional scanning electron micrographs reveal poor PEDOT interpenetration into the brick’s pores, this leads to a surface localized coating with minimal anchoring that is prone to delamination. Fortunately, type 1 brick’s open microstructure enables reagent vapor diffusion producing in situ PEDOT grafting; a nanofibrillar polymer coating is embedded as a network throughout pores resulting in strong adhesion (Fig. 3d). The addition of α-Fe2O3 particles enables deposition of PEDOT coatings on customized substrates such as concrete road pavers (Supplementary Fig. 4b) and Portland-based white concrete (Supplementary Fig. 4c).

Nanofibrillar PEDOT-coated brick electrochemical electrode

We present a summary of geometries and mass loadings for electrodes and devices in Supplementary Table 1 while areal, gravimetric, and volumetric capacitances, energy and power densities are shown in Supplementary Table 2 (calculations in detail are included in “Supplementary Methods” section). We choose areal metrics for evaluating our electrodes and devices because area is a practical parameter for evaluating our wall-like brick-based supercapacitor and it is directly related to the thin polymer coating (Supplementary Fig. 5) present on a wall. Areal metrics facilitate the estimation of capacitance and energy density delivered by a brick wall (Supplementary Discussion).

A 1 cm × 0.5 cm × 0.28 cm nanofibrillar PEDOT-coated type 1 brick (0.14 cm3) weighing 249 mg and carrying 6.97 mg PEDOT is fabricated into an electrode with one 1 cm × 0.5 cm face exposed (Fig. 4a). When calculating areal metrics, we use the area of the face exposed to electrolyte (0.5 cm2). The electrode exhibits a quasi-rectangular shaped three-electrode cyclic voltammogram and a capacitance of 2.61 F cm−2 (187 F g−1 based on PEDOT’s mass) at 2 mV s−1 in 1 M H2SO4 (Fig. 4b). The low scan rate of 2 mV s−1 enables enough time for charge transfer in the thick electrode to calculate maximum capacitance and energy density. We summarize capacitances obtained at scan rates from 2 to 100 mV s−1 in Supplementary Fig. 6a. The Fe3+/Fe2+ redox pair at 0.37 V and 0.49 V (vs. Ag/AgCl) arise due to iron species in brick and disappears as scan rate increases to 25 mV s−1 because Faradaic processes in PEDOT are faster than those occurring from solid Fe2O3 (Fig. 4c inset)15,19. Rate performance tests demonstrate capacitive behavior as scan rate increases to 100 mV s−1; however, due to limited charge transport the curve changes to fusiform shape20 (Fig. 4c). Capacitance is also dependent on the aqueous electrolyte and sulfuric acid leads to greater capacitance (1.64 F cm−2) than sodium sulfate (0.878 F cm−2) at 25 mV s−1 (Fig. 4d). This drastic difference is plausibly due to higher ionic mobility in H+ (36.23 × 10−8 m2 V−1 s−1) vs. Na+ (5.19 × 10−8 m2 V−1 s−1) or a lower electrical resistance caused by doping at low pH14. We probed this behavior further using electrochemical impedance spectroscopy. Nyquist plots and an equivalent circuit diagram demonstrate a significantly lower ion diffusion resistance for H2SO4 (1.7Ω) vs. Na2SO4 (4.6Ω) and minimal change in electrode material electrical resistance at low pH (Fig. 4e). To test the effect of an acidic electrolyte on a brick’s inorganic species, we pulverize and homogenize the entire polymer-coated brick electrode using mortar and pestle. Powder X-ray diffraction is carried out before and after electrodes are cycled in 1 M H2SO4 using sequential scan rates of 2, 5, 10, 25, 50, and 100 mV s−1 (10 cycles each). Diffraction patterns remain unchanged demonstrating that most inorganic species in a brick remain unaffected by synthesis and electrochemical cycling (Fig. 4f).

Fig. 4: Three-electrode characterization of nanofibrillar PEDOT-coated bricks.
figure4

a Nanofibrillar PEDOT-coated type 1 brick (1 cm × 0.5 cm × 0.28 cm) is connected to a Pt current lead using polyimide tape that exposes a 1 cm × 0.5 cm face to the electrolyte. b Cyclic voltammogram at 2 mV s−1 shows a quasi-rectangular shape stemming from PEDOT’s capacitive behavior with Fe3+/Fe2+ redox pair peaks at 0.37 and 0.49 V (vs. Ag/AgCl). c The Fe3+/Fe2+ redox peaks disappear when scan rate increases to 25 mV s−1. d The electrode shows a smaller curve area for 1 M Na2SO4 electrolyte compared to 1 M H2SO4, indicating lower capacitance. e Nyquist plots collected using different electrolytes (inset shows equivalent circuit diagram) show fitted (solid lines) versus experimental data (segregated points). f Powder X-ray diffraction shows identical patterns for pristine type 1 brick vs. PEDOT-coated type 1 brick after cycling in 1 M H2SO4 between −0.2 and 0.8 V (vs. Ag/AgCl) at scan rates of 2, 5, 10, 25, 50, and 100 mV s−1 (10 cycles each).

To quantify dissolution of a brick’s α-Fe2O3 and Al2O3 by 1 M H2SO4, we perform inductive coupled plasma mass spectrometry on the electrolyte after cycling experiments (Supplementary Fig. 6b, c). For a precise calculation, we control electrolyte volume (5 mL) and electrode mass (249 mg). Analysis shows negligible concentrations of Fe (4.44 µg mL−1) and Al (1.97 µg mL−1) in the electrolyte after cycling, these concentrations are equivalent to mass losses (based on the entire brick) of 0.0127 and 0.0075 wt% for α-Fe2O3 and Al2O3, respectively. These results confirm that inorganic species in a brick are preserved after electrochemical cycling.

Symmetric nanofibrillar PEDOT-coated brick supercapacitor

Two nanofibrillar PEDOT-coated bricks (1 cm × 0.5 cm × 0.28 cm) serve as electrodes in a symmetric supercapacitor (1 cm × 0.5 cm × 0.5625 cm) using 1 M H2SO4 aqueous electrolyte (Fig. 5a upper row and Supplementary Fig. 7a). A device consists of a volume (0.28 cm3) encompassing two PEDOT-coated bricks and a separator and its total mass (499 mg) includes 13.94 mg of PEDOT. Nyquist plot shows an aggregated internal resistance of 3Ω and a line with a ~45° slope between semicircle and low-frequency domain (Warburg region) (Supplementary Fig. 7b). This line is characteristic of thick electrodes where a tortuous path stifles ion diffusion. At the low-frequency region, the curve tends to form an arc because of Nernst diffusion impedance of H+ and SO42− in the electrolyte21; this plausibly stems from an increased ion diffusion distance caused by a separation gap between anode and cathode. Cyclic voltammogram shows a quasi-rectangular shape between 0 and 1 V (collected at 2 mV s−1) leading to a device areal capacitance of 1.59 F cm−2 calculated using the electrode area directly in contact with separator (0.5 cm2) (Fig. 5b, black curve and Supplementary Fig. 7c). In our device, a nanofibrillar PEDOT coating covers all six faces of a brick, two of the larger faces (1 cm × 0.5 cm) are directed towards each other and during cycling, ions travel through the inner brick pores resulting in an electrochemical contribution by the other faces. The areal metrics of this aqueous electrolyte supercapacitor are thus qualitative.

Fig. 5: Nanofibrillar PEDOT-coated bricks for supercapacitors.
figure5

a Schematic illustration of aqueous electrolyte supercapacitor and quasi-solid-state supercapacitor shows different charge storage sites. The black dash line in the aqueous electrolyte device represents a separator. b Cyclic voltammograms for symmetric supercapacitors using 1 M H2SO4 aqueous electrolyte and poly(vinyl alcohol)/H2SO4 gel electrolyte. c Galvanostatic charge–discharge profiles for quasi-solid-state device at current densities ranging between 0.5 and 25 mA cm−2; curves at 1, 5, 10, and 25 mA cm−2 are horizontally expanded 2×, 10×, 15×, and 20×, respectively. Inset shows IR drop at current densities of 5, 10, and 25 mA cm−2. d Quasi-solid-state supercapacitor charge-discharge curves after 10,000 cycles at 5 and 25 mA cm−2 exhibit 87% and 90% capacitance retention, respectively (coulombic efficiency is ~100%). e Cyclic voltammograms for a single quasi-solid-state supercapacitor are collected at voltage windows of 1 and 1.2 V; a tandem device (comprised of three supercapacitors connected in series) withstands a 3.6 V window. The tandem device is waterproof after coated by epoxy and exhibits a stable cyclic voltammogram. f Photograph shows a supercapacitor module lighting up a green light-emitting diode. This tandem device (4 cm × 3 cm × 1 cm) contains three supercapacitors (4 cm × 1 cm × 1 cm) connected in series; the core–shell structure of an electrode is also shown.

Our supercapacitor possesses low internal resistance resulting in a low IR drop (0.01 V) during galvanostatic charge-discharge experiments at 0.5 mA cm−2 current density in a 1 V window (Supplementary Fig. 7d). These curves demonstrate a device areal capacitance of 1.60 F cm−2 (2.84 F cm−3 for volumetric) as well as areal energy and power densities of 222 and 0.25 mW cm−2, respectively (394 µWh cm−3 and 0.44 mW cm−3 for volumetric). High power density (12.5 mW cm−2) is obtained at a current density of 25 mA cm−2 albeit with lowered capacitance (0.706 F cm−2) and energy density (98 µWh cm−2) because ion transport in our thick electrode is limited (shown by high IR drop of 0.4 V). Our device works in an extended voltage window (1.2 V) resulting in cyclic voltammograms and galvanostatic charge–discharge curves that retain shape (Supplementary Fig. 7e, f). However, the increased IR drop (0.07 V) at 1.2 V decreases coulombic efficiency as well as long term cycling stability leading to lower capacitance versus 1 V (Supplementary Fig. 7g, h). The energy densities at 1.2 V are higher than at 1 V as shown in a Ragone plot (Supplementary Fig. 7i) because a wider voltage window allows a device to store more charge (Supplementary Eq. 7 from “Supplementary Methods” section). Connecting three devices in series increases the voltage window to 3.6 V, this also triples the internal resistance and reduces output current to one-third (Supplementary Fig. 8a). A tandem device reaches an output voltage of 2.685 V (charged at 4.5 V for 15 s) and lights a white light-emitting diode for 11 min. The output voltage decreases to the same level as the light-emitting diode’s turn-on voltage (2.546 V) after discharging for 214 s (Supplementary Fig. 8b–d and Supplementary Movie 1).

Quasi-solid-state PEDOT-coated brick supercapacitor

To minimize electrolyte leakage, we develop a symmetric supercapacitor (1 cm × 0.5 cm × 0.63 cm) using a poly(vinyl alcohol)/1 M H2SO4 gel that binds PEDOT-coated bricks and serves as electrolyte and separator (Fig. 5a lower row and Supplementary Fig. 9a). The gel electrolyte layer (0.7 mm thick) prevents bricks (1 cm × 0.5 cm × 0.28 cm) from short-circuiting and leads to enhanced adhesion between electrodes. This sandwich-type supercapacitor is 0.32 cm3 in volume, 518 mg in weight (includes mass of electrodes and electrolyte) containing 13.94 mg of PEDOT. In a tensile test, our electrode–gel–electrode structure withstands a shearing force equal to 1000 times the device’s weight20 (Supplementary Fig. 9b). Intimate contact between gel and PEDOT nanofibers enhances charge transfer resulting in low internal resistance (2.5Ω) and a linear Nyquist plot (Supplementary Fig. 9c). Device areal capacitance (0.868 F cm−2) and areal energy density (121 µWh cm−2) originating from PEDOT-coated brick electrodes are calculated from galvanostatic charge–discharge curves (1.38 F cm−3 and 192 µWh cm−3 for volumetric, collected at 0.5 mA cm−2); cyclic voltammograms also show capacitive behavior (Fig. 5b, c, Supplementary Fig. 9d–g). The absence of redox peak from cyclic voltammogram indicates a minimal contribution from the α-Fe2O3 present in a brick to energy storage. A gel electrolyte leads to 50% lower areal capacitance and energy density than an aqueous electrolyte because of stifled gel permeation kinetics throughout the electrode preventing access of ions to all non-sandwiched PEDOT-coated faces (Fig. 5a, Supplementary Fig. 9h).

Outdoor exposure is inevitable for a stationary supercapacitor and epoxy encapsulation affords a cost-effective, mechanically robust and waterproof housing. An epoxy-coated supercapacitor retains ~90% of original capacitance and exhibits ~100% coulombic efficiency after 10,000 charge–discharge cycles (collected at 25 mA cm−2) (Fig. 5d, red curves). This 5-min epoxy coating prevents water evaporation from the gel’s hydrated ionic percolation network22 (Supplementary Fig. 10a) enabling 10,000 charge–discharge cycles at 5 mA cm−2 (640 h of continuous operation) with ~87% capacitance retention (Fig. 5d, black curves). Gel electrolyte and encapsulation enables operation at temperatures between −20 and 60 °C (this range covers most outdoor temperatures) as shown by cyclic voltammograms and Nyquist plots (Supplementary Fig. 10b, c). The capacitance increases proportionally with temperature due to enhanced ionic transport22; PEDOT remains capacitive after repeated reversible heating-cooling cycles (Supplementary Fig. 10d). Temperatures below −20 °C or above 60 °C cause significant freezing or evaporation of water from the gel electrolyte leading to unstable electrical performance and breakage of epoxy seal.

Epoxy renders a stationary supercapacitor module waterproof. Three quasi-solid-state supercapacitors are sealed in epoxy pucks (Supplementary Fig. 9a) and connected in series by electrical wires. This entire module (including connection points) is coated by another layer of epoxy that leaves only cathode and anode exposed. After epoxy curing, the entire module is immersed in water except for two electrode-leads and is electrochemically cycled (electrical circuit shown in Fig. 5e inset and Supplementary Fig. 10e). The module’s cyclic voltammograms (Fig. 5e) collected while immersed demonstrate stable behavior identical to pre-immersion tests. A device charges to 3 V in 10 s while immersed in water and lights up a green light-emitting diode (2.155 V forward voltage) for ~10 min (Supplementary Fig. 10e, f). A gel electrolyte and our deposition technology enable scale up as demonstrated by connecting six large nanofibrillar PEDOT-coated brick electrodes (2 cm  1 cm  1 cm) in series resulting in a supercapacitor module that charges to 3 V in 5 s readily lighting up a green light-emitting diode (Fig. 5f, Supplementary Fig. 11). Notably, a brick wall constructed using our nanofibrillar PEDOT-coated bricks holds the potential to deliver a maximum device capacitance of 11.5 kF m−2 and an energy density of 1.61 Wh m−2 (Supplementary Fig. 12; Supplementary Discussion).