Electrochromic devices (ECDs), which control light transmittance through electrochemical redox reactions, have been widely used in electrochemical reflective displays or smart windows for energy efficient buildings1,2,3,4,5. Recently, the functionality of ECDs has been extended to include energy storage, which is referred to as electrochromic supercapacitors (ECSs)6,7,8,9. ECSs have been increasingly studied as next-generation electrochemical components that can not only change their own optical properties but also store the energy supplied for coloration. In particular, their optical characteristics, such as color intensity, directly reflect the real-time levels of energy stored in these devices10,11,12,13,14,15.

One key parameter for achieving high-performance ECSs is an EC chromophore. Among EC materials, transition metal oxides such as tungsten trioxide (WO3) have been widely employed to exploit their superior electrochemical properties16,17,18, large optical modulation19,20, and high coloration efficiency21,22. Recently, a few research groups reported ECSs based on WO3 films23,24,25,26. Mai et al fabricated ECSs with WO3 films prepared by thermal evaporation. However, the diffusivity of Li+ ions across the film was low because a dense WO3 layer was formed by the thermal evaporation process27. Additionally, the diffusion distance for sufficient intercalation was relatively long from the surface of the WO3 film, thereby resulting in a slow switching speed and unsatisfactory optical modulation. To overcome these drawbacks arising from the dense WO3 structure, various WO3 nanostructures, such as nanosheets, nanoparticles, and macroporous films, have been suggested28,29,30,31,32. However, the ultrafast dynamics needed for practical applications, such as displays, cannot be achieved with these nanostructured WO3 because the interfacial area where ion intercalation occurs is not sufficiently large and the ion diffusion distance is still relatively long. Therefore, mesoporous structures consisting of fully interconnected small pores (diameters of 2–50 nm) are considered an ideal nanostructure for realizing the ultrafast response of devices33,34.

Hard templating methods have often been used to fabricate mesoporous structures, but a multistep process is necessary, and the range of tunable pore sizes is limited35,36,37,38,39,40,41. On the other hand, evaporation-induced self-assembly (EISA) based on sol-gel chemistry and the self-assembly of amphiphilic molecules (e.g., block copolymers) is a powerful technique capable of producing various mesoporous metal oxides42,43,44,45. The pore size of the resulting mesoporous metal oxides is easily controlled by adjusting the volume ratio of hydrophobic and hydrophilic parts in amphiphilic molecules46,47. Moreover, this method can be combined with versatile solution processes, such as spray-coating, spin-coating, dip-coating, and printing48,49,50,51,52.

Herein, we report ultrafast response ECSs by exploiting the mesoporous structure of an amorphous WO3 film prepared by EISA. Compared to compact-WO3 film-based devices, mesoporous WO3 devices showed outstanding electrochromic/supercapacitor properties, for example, a large optical modulation (ΔT = 76%), ultrafast sub-1 s switching speeds (tc = 0.8 s and tb = 0.4 s), and excellent coloration/bleaching and charging/discharging stability under fast switching conditions. By exploiting the ultrafast response, we successfully demonstrated functional ECS displays (ECSDs) showing user-customized images and storing the energy supplied for coloration. To this end, the EISA protocol and printing technology were combined for patterning mesoporous WO3, which is referred to as printing-assisted EISA (PEISA). The energy stored in the ECSDs can be extracted for the operation of other electronic devices, such as LEDs. These highly functional energy-storing ECSDs have great potential for use in next-generation smart electrochemical components.

Results and discussion

Figure 1a illustrates the fabrication process of a mesoporous WO3 film through EISA. First, a mixed solution of tetrahydrofuran (THF) including polystyrene-block-polyethylene oxide (PS19k-b-PEO6.5k) and ethanol-containing WCl6 as a WO3 precursor was spin-coated on FTO-coated glass. The resulting film was comprised of inorganic-organic composites. The THF evaporated first due to its low boiling point, and then PS19k-b-PEO6.5k formed spherical micelles where hydrophobic PS blocks were surrounded by PEO shells containing hydrolyzed hydrophilic inorganic tungsten species53,54. The composite was calcined at 350 °C for 1 h, in which the organic parts were partially removed and the remaining ones were transformed to amorphous carbon. On the other hand, the inorganic parts formed WO3 through condensation. Thus, the resulting composite film became a mesoporous amorphous carbon/WO3 structure. Finally, the amorphous carbon/WO3 was exposed to O2 plasma to eliminate amorphous carbon. The Raman spectrum of amorphous carbon/WO3 (Fig. S1a) shows two broad peaks at approximately 1375 and 1595 cm−1 corresponding to the D and G peaks arising from amorphous carbons55,56. After the O2 plasma treatment, the amorphous carbons are fully removed, which is confirmed through the disappearance of the D and G peaks shown in Fig. S1b. It is noted that the O2 plasma treatment is an effective method to selectively remove the amorphous carbon without promoting further WO3 crystalline growth because the activation energy of the grain growth of WO3 is higher than that of the plasma57. However, when the thermal treatment (e.g., 400 °C) is performed to remove amorphous carbons, crystalline WO3 phases can be developed, which prevents effective ion diffusion through the WO3 film (see the X-ray diffraction (XRD) pattern and scanning electron microscopy (SEM) image in Figs. S2a, b, respectively). The SEM image supports the mesoporous structure of the WO3 film with small pores having an average diameter of ~30 nm (Fig. 1b). The thickness of the mesoporous structure of the WO3 film is determined to be ~250 nm (see the cross-section SEM image in the inset of Fig. 1b).

Fig. 1: Fabrication and characterization of mesoporous WO3 films.
figure 1

a Schematic showing the fabrication process of the mesoporous WO3 film. b SEM image (top view) of the mesoporous WO3 film (inset: cross-sectional view). c W 4f XPS spectra of the mesoporous WO3 film. d XRD patterns of the mesoporous WO3 film and bare FTO glass. e HR-TEM image of the mesoporous WO3 film (inset: SAED pattern).

Figure 1c shows the X-ray photoelectron spectroscopy (XPS) of the W 4f spectrum of the mesoporous WO3 film. Two peaks at 35.5 and 37.7 eV are observed, corresponding to W 4f7/2 and W 4f5/2 of the electronic state of W6+ in WO3, respectively58,59. It is known that a high EC performance can be obtained when amorphous WO3 is employed because of the easier ion penetration into the film60,61,62,63. Therefore, we investigated the crystalline structure of the mesoporous WO3 film using XRD, high-resolution transmission electron microscopy (HR-TEM), and selected area electron diffraction (SAED) (Fig. 1d, e). Figure 1d shows the XRD patterns of the mesoporous WO3 film on FTO glass and bare FTO glass. Distinct diffraction peaks are not observed from the WO3 film, implying its amorphous nature64. The amorphous nature of WO3 is further supported by the HR-TEM image and SEAD pattern, in which no distinct lattice fringe is shown (Fig. 1e), but an amorphous halo in the isotropic SAED pattern is shown (inset of Fig. 1e). Accordingly, a high electrolyte-ion accessibility to the WO3 surface and high ion diffusivity into the WO3 film are anticipated based on the large surface area of the mesoporous structure and the amorphous nature of WO3, respectively. As a result, the ultrafast dynamics of mesoporous WO3-based ECSs (denoted meso-WO3 ECS) are expected.

Figure 2a depicts the schematic of the meso-WO3 ECS with an electrolyte (1 M LiClO4 in propylene carbonate (PC)) and NiO film (ion-storage layer). The characterizations of the NiO film using XRD, XPS, and SEM are shown in Fig. S3a–c, respectively. For comparison, we also fabricated devices denoted as compact-WO3 ECS, in which WO3 films were prepared by WO3 nanoparticles (see the XRD pattern, XPS spectra, and SEM image of the compact-WO3 film in Fig. S4a–c). The thickness of the compact-WO3 film (~260 nm) is quite similar to that (~250 nm) of the mesoporous WO3 film. To investigate the electrochromic (EC) behavior of these ECSs, we recorded the UV–vis transmittance spectra of the meso-WO3 ECS at various applied voltages (Fig. 2b). As the applied voltage is increased, the transmittance gradually decreases over the entire range of visible light wavelengths, which is caused by the formation of reduced WO3 (W5+) and oxidized NiO (Ni3+). On the other hand, when +2.3 V is applied, the transparent bleached state is recovered by producing oxidized WO3 (W6+) and reduced NiO (Ni2+). While the device is highly transparent in the bleached state, a blue-colored state is observed in the colored state (see also the inset of Fig. 2b). A similar optical transition is observed from the compact-WO3 ECS (Fig. S5).

Fig. 2: EC performance of meso- and compact-WO3 ECSs.
figure 2

a ECS structure employed in this work. b Transmittance spectra of the meso-WO3 ECS as a function of applied voltages (inset: photographs of the colored and bleached states). c In situ optical responses of the compact-WO3 ECS (red line) and the meso-WO3 ECS (blue line). d Nitrogen adsorption-desorption isotherm curves of the mesoporous and compact WO3. e Pore size distribution of mesoporous WO3. f Optical density variation as a function of injected charge density. g Coloration/bleaching cycling stability of the two ECSs for 1000 cycles.

However, there is a significant difference in the dynamic responses of the device, which is one crucial metric for evaluating ECS performance. The response times of coloration (tc)/bleaching (tb) are defined as the time at which 90% of the maximum transmittance contrast (ΔTmax) is achieved. To compare the EC dynamic responses of the above two devices, we recorded the transmittance profiles at 700 nm at alternating potentials (+2.3 V, 5 s and ‒2.3 V, 10 s) (Fig. 2c). The meso-WO3 ECS exhibits large optical modulation (ΔTmax = 76%) and ultrafast coloration (tc = 0.8 s) and bleaching time (tb = 0.4 s), which is notably faster than previously reported WO3-based ECDs (Table S1). In the case of the compact-WO3 ECS, under the same operating conditions as those for meso-WO3, a stable state of coloration and bleaching is not obtained and shows a lower optical modulation (69.2%), indicating a slower dynamic response. After a sufficient coloration (30 s) and bleaching times (30 s), ΔTmax is determined to be ~78% similar to that (~76%) of the meso-WO3 ECS (Fig. S6). However, a much slower response is observed with tc and tb values of ~6.8 and ~9.2 s (Fig. S6), respectively, compared to those (~0.8 and ~0.4 s) of meso-WO3. To elucidate the origin of such different behaviors, we conducted a nitrogen adsorption-desorption experiment to investigate the larger surface area of mesoporous WO3 (Fig. 2d). Similar to conventional mesoporous samples44,45, mesoporous WO3 shows a Type-IV isotherm, and the BET surface area (SBET) is measured to be 32.14 m2/g. The pore size distribution derived from the adsorption branch was also investigated using the BJH method, resulting in an average diameter of ~29.3 nm (Fig. 2e). On the other hand, compact WO3 shows a much smaller surface area (SBET with 6.4 m2/g).

Another factor for evaluating the performance of ECSs is the coloration efficiency (η), which is calculated from the slope of the linear regime of the optical density variation as a function of injected charge density. The coloration efficiencies of meso-WO3 and compact-WO3 ECSs are calculated to be 682 and 188 cm2/C, respectively (Fig. 2f). Accordingly, the meso-WO3 ECS consumes less energy than the compact-WO3 ECS. Cycling stability under fast response conditions is required for practical ECSs such as smart windows and electrochemical reflective displays. Figure 2g shows the transmittance responses of both ECSs during a 1000-cycle test. Under quick switching conditions (e.g., coloration/bleaching of 1 s/5 s), the meso-WO3 ECS maintains 85.5% of the original optical modulation after 1000 cycles, whereas the drop in optical modulation of the compact-WO3 ECS is 70.7% (see also the changes in ΔT/ΔT0 of both ECSs in Fig. S7). Li+ ions are repeatedly inserted and extracted in the WO3 and NiO films during the cycling test with alternating potentials. However, Li+ ions are gradually trapped in both films. As a result, unchangeable and transparent W6+ and Ni2+ are present in the forms of Li2xW2O6+x and Li2yNi2O2+y, respectively65. These species do not participate in the redox reactions, so the color intensity is reduced while the bleached state is not affected. The excellent cycling stability of the meso-WO3 ECS even with fast switching is attributed to their mesoporous WO3 structure with a large surface area. Therefore, we conclude that the meso-WO3 ECS is more suitable for dynamic applications (e.g., displays) that require a fast response.

Considering the electrochemical reactions (Eqs. (1) and (2)) during coloration/bleaching, ECSs can store energy via ion insertion/extraction processes. Therefore, we investigated and compared the capacitive performances of the meso-WO3 and compact-WO3 ECSs.

$${\mathrm{WO}}_3 + x{\mathrm{Li}}^ + + x{\mathrm{e}}^ - \leftrightarrow {\mathrm{Li}}_x{\mathrm{WO}}_3\left( {{\mathrm{EC}}\,\,{\mathrm{layer}}} \right),$$
$${\mathrm{NiO}} + x{\mathrm{ClO}}_4^ - \leftrightarrow {\mathrm{NiO}} \cdot {\mathrm{ClO}}_4 + x{\mathrm{e}}^ - \left( {{\mathrm{ion}}{\mbox{-}}{\mathrm{storage}}\,\,{\mathrm{layer}}} \right).$$

Electrochemical impedance spectroscopy (EIS) characterizations were conducted to compare the charge transfer and ion kinetics of the meso-WO3 and compact-WO3 ECSs. From the Nyquist plots (Fig. S8), the meso-WO3 ECS shows a lower intercept at the real axis and a smaller semicircle than those of the compact-WO3 ECS, indicating a smaller contact resistance and charge transfer resistance. Additionally, a larger slope in the Warburg region of the meso-WO3 ECS compared to that of the compact-WO3 ECS represents a lower ion diffusion resistance. All the EIS fitting results are summarized in Table S2. Figure S9 shows the cyclic voltammetry (CV) curves of the meso-WO3 and compact-WO3 ECSs at various scan rates from 200 to 1500 mV/s. All the CV curves exhibit pseudocapacitive behaviors with redox peaks. However, particularly at high scan rates (e.g., 1500 mV/s), the meso-WO3 ECS maintains its characteristic CV curves, while the CV curves of the compact-WO3 ECS are severely distorted. The fast electrochemical kinetics of meso-WO3 may be attributed to its fast charge transfer and ion kinetics. Figure 3a displays the galvanostatic charging/discharging (GCD) curves of the meso-WO3 ECS at various current densities (see also the GCD curves of the compact-WO3 ECS in Fig. S10). For quantitative analysis, the areal capacitance (C) is calculated from discharging curves using23,66.

$${C} = \frac{{{{I}}{\Delta} {{t}}}}{{{{S}}{\Delta} {{V}}}}.$$
Fig. 3: Energy storage characteristics of meso- and compact-WO3 ECSs.
figure 3

a GCD curves of the meso-WO3 ECS at various current densities. b Dependence of the capacitance retention of the compact- and meso-WO3 ECSs on the charging/discharging current densities. c Charging/discharging cycling stability of the ECSs at a current density of 1.0mA/cm2. d GCD curve at 1.0mA/cm2 and the corresponding in situ transmittance profile at 700nm for the meso-WO3 ECS. e Photographs of the meso-WO3 ECS during the charging and discharging process. f Schematic illustration of the ion intercalation in meso- (left) and compact-WO3 (right).

For example, the C values of the meso-WO3 ECS and compact-WO3 ECS are determined to be 2.57 and 2.51 mF/cm2, respectively, at the lowest current density in this study (i.e., 0.02 mA/cm2). However, there is a significant difference in the charge-storing capability as the operating current density increases (Fig. 3b). The capacitance of the compact-WO3 ECS is greatly decreased compared with that of the meso-WO3 ECS. For instance, the capacitance retentions of the meso-WO3 and compact-WO3 ECSs are 68 and 34%, respectively, when the current density is changed from 0.02 to 1.0 mA/cm2. The areal capacitance values at various current densities are shown in Fig. S11. Furthermore, from the Ragone plots of both ECSs (Fig. S12), the meso-WO3 ECS exhibits an energy density of 15.8 Wh/kg at a power density of 0.2 kW/kg. Even at a high power density of 9.6 kW/kg, the energy density is measured to be 10.6 Wh/kg, which is much higher than that (4.9 Wh/kg) of the compact-WO3 ECS at a similar power density of 8.8 kW/kg. This result implies that the meso-WO3 ECS is more promising for quick charging/discharging devices. Indeed, the long-term charging/discharging stability of the meso-WO3 ECS at 1.0 mA/cm2 is remarkably outstanding (Fig. 3c). Only a 6.2% reduction in the first run is detected after 1000 cycles, whereas the compact-WO3 ECS deteriorates considerably under the same conditions. For example, only 14% of the initial areal capacitance is maintained after 1000 cycles.

Another fascinating function of ECSs was also examined, in this case, the direct visualization of the stored energy level. We simultaneously recorded the GCD curve at 1.0 mA/cm2 and the corresponding in situ transmittance profile at 700 nm of the meso-WO3 ECS (Fig. 3d). As the device is charged to ‒2.3 V, the transmittance decreases (i.e., colored). In contrast, during discharging, the transmittance increases and eventually returns to its initial transmittance. Photographs of the meso-WO3 ECS at six representative stages during charging/discharging are shown in Fig. 3e, where the device indicates the real-time level of the stored energy through its color intensity. Regardless of the current density, the meso-WO3 ECS exhibits a symmetrical GCD profile shape and no significant degradation of optical contrast, which is attributed to its effective and fast ion transport (Fig. S13). In the case of the compact-WO3 ECS, the GCD profiles become more asymmetrical in shape, and the optical modulation dramatically decreases as the current density increases (Fig. S14). These results suggest that the compact-WO3 ECS is not proper for high-rate operations due to its inefficient ion transport and slow charge transfer. Similar to the origin of its outstanding kinetic EC performance, the availability of the quick charging/discharging of meso-WO3 ECS arises from the large surface area of the mesoporous WO3 and its amorphous nature, in which the fast ion diffusion and simultaneous intercalation/deintercalation of large amounts of ions can occur (Fig. 3f (left)). In contrast, the behavior of the compact-WO3 ECS is related to its low ion and charge transfer because of its very small surface area and its crystalline nature, which causes incomplete ion deintercalation (Fig. 3f (right)).

Last, we demonstrated highly functional energy-storing ECS displays (ECSDs) by combining printing and EISA protocols, referred to as printing-assisted EISA (PEISA) (Fig. 4a). When the EISA solution is printed out of the nozzle, a micellar structure is directly produced through evaporation. When a sequential calcination and O2 plasma treatment are performed, patterned mesoporous WO3 is prepared. Figure 4b shows the optical microscopy (OM) and SEM images of the printed mesoporous WO3. Well-defined mesoporous WO3 is clearly observed in the printed lines. This result implies that user-defined patterns consisting of mesoporous WO3 can be fabricated by PEISA, which can then be applied to energy-storing ECSDs. To complete device fabrication, we also patterned NiO with the same shape as WO3, followed by sandwiching the electrolyte layers with mesoporous WO3. The resulting ECSD successfully functions as both an electrochemical reflective display and an energy storage device (Fig. 4c). When the device is charged, the patterns turn dark blue, indicating the charged state. To observe the utilization of the charged energy, the ECSD was connected to a white light-emitting diode (LED). Initially, the LED emits light, and then the ECSD returns to its original transparent state when the stored energy is completely consumed.

Fig. 4: Application of PEISA for the fabrication of functional ECSDs.
figure 4

a Schematic illustration of PEISA. b OM (left) and SEM (right) images of mesoporous WO3 fabricated by PEISA. c Photographs of the ECSD during the reversible charging (coloration of the pattern) and discharging (LED light on and bleaching) test. For this application, two ECSDs were connected in series.


In this study, multifunctional ECSs were fabricated based on amorphous mesoporous WO3 films produced by EISA. Compared to the compact-WO3 ECS, the meso-WO3 ECS exhibited superior EC and supercapacitor performance based on its large surface area and the amorphous WO3. For example, the meso-WO3 ECS exhibited a large optical modulation (~76%), ultrafast dynamics (tc = 0.8 s, tb = 0.4 s), and high areal capacitance (2.57 mF/cm2), even at a high current density. Moreover, the meso-WO3 ECS exhibited excellent coloration/bleaching and charging/discharging cycling stability under rapid operating conditions (frequent switching or high current operations). Additionally, we developed PEISA to fabricate patterned mesoporous WO3 for ultrafast ECSDs, which could serve as electrochemical reflective displays that could also store electrical charge. The stored energy was used for powering other electronic devices, while the color intensity of the pattern indicated the level of stored energy. These results show that the ECSD in this work has tremendous potential for use in next-generation smart electronics. Our ongoing work aims to improve the performance of these ECSDs through the optimization of the pore diameter and film thickness along with the doping of various other metals (e.g., Ti, Mo, Sb, etc.) into the mesoporous structure. Since the PEISA method introduced in this study can be easily employed to prepare various mesoporous metal oxides, such as TiO2, Al2O3, V2O5, and ZnO, this method provides a versatile foundation for producing printed electronics.

Experimental section


All chemicals were purchased from Sigma-Aldrich (St. Louis, MO, USA) except tert-butyl alcohol (>99.0%, Tokyo Chemical Industry Co., Ltd., Japan) and polystyrene-block-polyethylene oxide (PS-b-PEO, polydispersity index (PDI) = 1.09, Mn (PS) = 19,000 g mol−1, Mn (PEO) = 6500 g mol−1, Polymer Source Inc., Canada). Indium tin oxide (ITO)-coated (sheet resistance: 15 Ω/sq) and fluorine-doped tin oxide (FTO)-coated glasses (sheet resistance: 15 Ω/sq) were purchased from Asahi Glass Co (Tokyo, Japan). The ITO and FTO-coated glasses were cleaned with 2-propanol (15 min) under sonication. Then, further cleaning with a UV-ozone treatment was conducted for 15 min before use.

Fabrication of the mesoporous and compact-WO3 films

Mesoporous WO3 films were prepared using amphiphilic PS19k-b-PEO6.5k as the structure directing agent and WCl6 as the precursor. PS19k-b-PEO6.5k (50 mg) was dissolved in THF (1 g), while WCl6 (0.2 g) was separately dissolved in EtOH (0.4 g). The two above solutions were mixed in a weight ratio of 2 (PS19k-b-PEO6.5k solution):1 (WCl6 solution), and then the solution was kept at 45 °C. After 24 h, the obtained blue solution was spin-coated onto FTO glass at 1000 rpm for 5 s. Subsequently, the thin film-coated FTO glass was placed in an oven at 45 °C for 24 h to form an inorganic-polymer composite. Next, the inorganic-polymer composite film was calcined in a preheated 350 °C furnace for 1 h, which resulted in an amorphous carbon/WO3 film. Last, an O2 plasma treatment was conducted to remove amorphous carbon while maintaining the amorphous nature of WO3. Additionally, we prepared compact-WO3 films for comparison. Compact-WO3 films were prepared by spin coating a nanoparticle WO3 suspension on FTO glass at 5000 rpm for 20 s, followed by thermal annealing at 60 °C for 10 h in vacuum 23.

Fabrication of the NiO film

NiO nanoparticles were synthesized by the solvothermal method according to a previous report67. The resulting NiO nanoparticles (0.2 g) were dispersed in a mixture of DI-water and 2-propanol as the solvent (DI-water:2-propanol = 1:1, 2 g). Thereafter, the obtained suspension was spin coated onto ITO glass at 1000 rpm for 20 s, followed by annealing at 200 °C for 2 h.

Device assembly

ECSs were fabricated by assembling the mesoporous WO3 (or compact-WO3 film) and NiO film as an active layer and ion-storage layer, respectively, in which a precut thermoplastic (Parafilm) spacer was employed. Then, 1 M LiClO4 in a propylene carbonate (PC) electrolyte was filled into the assembled ECSs and sealed with epoxy glue.

Printing-assisted EISA

The EISA solution was filled into a syringe, which had a nozzle diameter of 160 μm. An electrostatic field was applied between the nozzle and the substrate during printing, and the electrostatic force allowed the polar EISA solution to form a well-defined pattern. The movement of the sample stage was controlled by a computer, by which the printing speed and the distance between the nozzle and substrate were adjusted. The obtained patterned films were calcined at 350 °C, followed by an O2 plasma treatment before use.

Film characterization

The fabricated films in this work were characterized by OM (ECLIPSE LV100ND, Nikon), FE-SEM (S-4800, Hitachi) at an acceleration voltage of 10 kV, HR-TEM (JEM-2200FS, JEOL), XRD (D/MAX2500 VL-PC, Rigaku) with a Cu Kα radiation source (λ = 0.154178 nm), XPS (Vgescalab 250, Thermo Fisher Scientific) equipped with a monochromatic Al-Kα radiation source (hv = 1486.8 eV) and Raman spectroscopy (Alpha 300 R, WITEC) using a He–Ne laser with an excitation wavelength of 632.8 nm. Nitrogen physisorption was performed at 77 K using a Micrometrics ASAP 2010 system. The Brunauer-Emmett-Teller (BET) surface area was calculated using the data obtained over the relative pressure range of 0.05 ≤ p/p0 ≤ 0.2. The cell and window size distributions were calculated by the Barrett–Joyner–Halenda (BJH) method from the adsorption and desorption branches of the isotherm.

Performance characterizations

DC voltage and square wave voltages were supplied from a potentiostat (Wave Driver 10, Pine Instrument). A UV–vis spectrophotometer (V-730, Jasco) was used to record the UV–vis spectra (360 to 1100 nm at a scan rate of 400 nm/min) at various applied voltages and the transient transmittance changes at a fixed wavelength (700 nm). CV, EIS, and galvanostatic charge–discharge (GCD) measurements were conducted using a battery cycler system (WBCS3000L, WonATech). EIS was conducted from 1000 kHz to 0.01 Hz at −0.5 V.