Photochemical Precharging of Tungsten Trioxide for Enhanced Transmittance Modulation in Flexible Electrochromic Devices

UV irradiation is used to precharge sputtered tungsten trioxide (WO3) on polyethylene terephthalate enhancing the photochromic response with organic solvents. A comparison between the optical and electrochemical properties of photochemically and electrochemically charged WO3 results in a correlation of the transmittance to the respective charge density. This allows for a precise control of the charge density in the precharging process monitored by UV–Vis spectroscopy. A proof‐of‐concept flexible electrochromic device combining precharged WO3 (charge density: 20 mC cm−2) and Prussian blue (15 mC cm−2) exhibits a superior change in the visible light transmittance (τv) from 8% (dark) to 79% (bleached) and a coloration efficiency of 139 cm2 C−1 at 716 nm.


Introduction
The importance of windows in the façades of buildings is indisputable. They provide daylight and visual contact to the outside, increasing well-being and comfort. Nevertheless, windows offer low insulation against solar irradiance and heat gain. [1] Static or mechanically driven blinds offer some control over the amount of sunlight entering the building, however, they also obstruct the view to the outside. Other factors to be considered for shading technologies are aesthetics, the total energy consumption of the building, glare, and levels of illumination. [2] Glare and solar heat gain can be partially reduced by tinted windows, while the heat gain caused by incident solar irradiation can also be limited by an IR reflective coating. [3] Electrochromic smart windows allow for a dynamic control of illumination levels while offering high transparency in the dark/bleached states and intermediate states, respectively. [4] In addition, most of the commercially available systems containing WO 3 and nickel oxide (NiO) only require voltage during switching, making them highly attractive as a low-power technology for energy-efficient buildings. [5] Several mechanisms for the photochromic effect of WO 3 are described in the literature. [16][17][18][19][20][21][22] In general, the incident electromagnetic radiation generates excitons (electrons, e − and holes, h + ), resulting in a measurable photocurrent. [23] This effect was first discovered by Deb in 1973 and the blue color was attributed to this excited state trapped in so-called color centers. [16] Further studies have shown that this first theory is incomplete. Now, the most widely accepted proton injection model (a-c in Figure 1) starts with a first excitation step and the generation of excitons. The generated electrons are not trapped, but reduce W 6+ to W 5+ and charge balance is attained by the diffusion of protons from the surface into the material. [18][19][20][21][22] These protons are abstracted from adsorbed water molecules on the surface.
Gavrilyuk describes a similar pathway for the holes, involving small organic molecules containing oxygen (d-f in Figure 1). [17] The oxygen atom coordinates on W 6+ centers. In the next step, the hole reacts with the organic molecule, resulting in a proton and a radical. Since carbon radicals are more stable than hydroxo radicals formed from water, the photochromic response in solvent-containing atmospheres is enhanced. [17] Finally, the radical reacts with another hole, creating a volatile molecule that can easily desorb from the surface. This proton insertion mechanism suggests that the irradiation of WO 3 with UV light prior to the ECD assembly can be used as a precharging method for WO 3 /PB-based ECDs.
In this study, a flexible WO 3 /PB ECD on PET-ITO with outstanding transmittance modulation is demonstrated by precharging WO 3 prior to the cell assembly. We report for the first time the enhancement of the photochromic effect of WO 3 by low-hazardous organic solvents as a precharging method for WO 3 . By comparing photochemical and electrochemical charging of WO 3 , a precise adjustment of the charge density is possible. Since the photochemical precharging method does not require direct contact with the WO 3 electrode, it can be implemented in a high-throughput roll-to-roll production process in contrast to the state-of-the-art electrochemical precharging methods.

Results and Discussion
First, the enhancement of the photochromic response is investigated as a precharging method for the WO 3 . A detailed characterization of photochemically charged WO 3 and a comparison to electrochemically charged WO 3 is given. Finally, the charge balance in the WO 3 /PB ECDs is adjusted and the performance of the resulting ECD is characterized.

Enhancement of the Photochromic Effect of WO 3 by Low-Hazardous Organic Solvents
It has been reported that the vapor of water and organic solvents can enhance the photochromic response in WO 3 . The toxic solvents methanol and formaldehyde show the best results in terms of the precharging time and the intensity of photochromic effect but less hazardous solvents like acetone or ethanol would be preferable for a large-scale production. [17] To evaluate the effect of these solvents on the photochromic response under UV irradiation, a setup with the WO 3 electrodes encapsulated between two silica glass panes (see Figure 2) with 10 µL of water, ethanol or acetone for the irradiation in a daylight simulation chamber was chosen.   [17][18][19][20][21][22] a) Generation of excitons by UV irradiation (electrons, e − , and holes; h + , in the bulk material), diffusion of e − and h + to different reaction sites. b) Reduction of W 6+ to W 5+ . c) Reaction of h + with surface water. d)-f) Detailed reaction mechanism with formaldehyde: d) Adsorption of H 2 CO at the surface, coordination to W 6+ . e) Reaction of coordinated H 2 CO with h + , resulting in a radical and a proton. Protons diffuse into the bulk for charge balance. f) Reaction of the radical with another h + , desorption of the resulting carbon monoxide, and diffusion of the proton into the bulk. with λ being the wavelength, D λ the standard illuminant D65, T(λ) the transmittance, and V(λ) the spectral luminous efficiency of a standard observer at 10°. The τ v change of the WO 3 electrodes over time for all tested conditions is shown in Figure 3a: without encapsulation and encapsulated with water, ethanol, and acetone (WO 3-air/no encapsulation, WO 3 -air/water, WO 3 -air/ethanol, WO 3 -air/ acetone). The photochromic response of WO 3 is enhanced by all studied solvents in comparison to WO 3-air/no encapsulation. For the sample WO 3-air/no encapsulation, τ v decreased from initially 80% to 53% after 231 h of irradiation, where a plateau is reached, indicating an equilibrium in the photochemical reaction. This plateau can be explained as follows: The WO 3 reacts with ambient water and is photochemically reduced (see Figure 1). In addition, the reverse reaction with ambient oxygen can occur as well. Without an enhancement of the photochromic response, the equilibrium lies at a state in which the WO 3 is not completely colored (reduced). Although the photochromic response of WO 3 is enhanced by water, opaque spots (Figure 3b), and a haze of ≈30% are observed, making this solvent unsuitable for precharging. The highest photochromic effects were reached with ethanol and acetone, where τ v values decreased to 14% and 10%, respectively, after 231 h of irradiation. According to the mechanism in Figure 1, a crucial step in the photochromic reaction is the formation of the solvent radical in the reaction with a hole. Thus, the slight advantage of acetone over ethanol can be explained by the stabilities of the corresponding radical species. [24] The best results were obtained  with WO 3 -air/acetone and therefore all further precharging experiments with WO 3 were carried out under these conditions.

Comparison of Electrochemically Reduced and Photochemically Precharged WO 3
In ECDs, WO 3 can be reversibly reduced (colored) and oxidized (bleached) by electrochemical methods. In order to use the photochemically induced coloration of WO 3 as a precharging method for the fabrication of ECDs, it is crucial that the photo chromic coloration is related to a reduction of the WO 3 and that the electrochemical reversibility is given. Thus, the photochemically charged WO 3 is compared with electrochemically charged WO 3 . The electrochemical charging was carried out in 1 M lithium bis(trifluoromethanesulfonyl)imide (LiTFSI) in propylene carbonate (PC) as electrolyte with lithium as counter electrode (CE) and reference electrode (RE), respectively. When comparing the transmission spectra of the electrochemically charged WO 3 electrode (Figure 4a, corresponding optical data in Table 1) with those of the photochemically charged WO 3 -air/acetone (Figure 4b, for WO 3 -air/no encapsulation, WO 3 -air/water, WO 3 -air/ethanol see Figure S3, Supporting Information), an increasingly pronounced absorption band resulting in a decrease of the transmittance reaching from the visible to the near-infrared region of the spectrum is observed. It can be assigned to an intervalence charge transfer between W 5+ and W 6+ . [25] The transmission spectra mainly differ in the position of the interference peaks and the overall transmittance values, which can be explained by the different refractive indices of air and electrolyte. Thus, in both cases, a (partial) reduction of WO 3 occurs.
For the production of WO 3 /PB ECDs, it is necessary to precisely control the charge densities of both electrodes. Thus, the precharging process of WO 3 by UV irradiation has to be monitored. The charge density can be determined by correlating it with the τ v value using UV-Vis spectroscopy. From the spectroelectrochemical measurement shown in Figure 4c, the charge density q of electrochemically charged WO 3 was correlated with τ v and fitted with the asymptotic Equation (2) This indicates that 1.29% is the lower limit for τ v. The values calculated with Equation (3) provide a rough estimation for WO 3 electrodes measured in air or argon, while τ v was determined in the electrolyte (1 m LiTFSI/PC). The asprepared WO 3 electrodes on PET-ITO are almost colorless and exhibit a τ v value in air and argon of 80%, in contrast to 91% in the electrolyte due to a different scattering behavior between the electrolyte/WO 3 interface. This has to be considered comparing the data for samples measured in air and in the electrolyte.
The WO 3 electrode (active area: 1 cm × 1 cm, encapsulated with acetone) was irradiated until a τ v value of 29% was reached. Subsequently, the photochemically charged electrode was discharged in a spectroelectrochemical experiment. This procedure was carried out in order to prove the reversibility of the photochemical reduction and to correlate the resulting τ v with the charge density. As shown in Figure 4d, it is possible to fully bleach the precharged WO 3 by electrochemical oxidation indicated by the plateau in the τ v value. The τ v value of the precharged electrode immersed in the electrolyte was 40%. According to Equation (3), the expected charge density for the τ v value is 9.9 mC cm −2 . The obtained charge density induced by the photochromic effect was 9.8 mC cm −2 and is in agreement with the estimated value from Equation (3). There-fore, the change in the τ v value shows that the charge density induced by UV irradiation is comparable to the electrochemical charging. These findings coincide with the proton insertion mechanism explained above. [18][19][20][21][22] The long discharging time indicates that the inserted protons (H + ) into the WO 3 electrode are trapped, which is a well-known phenomenon observed with Li + in WO 3 . [26] Since the color in reduced WO 3 is caused by an intervalence charge transfer between W 5+ and W 6+ , the influence of the cation on the optical properties of the reduced WO 3 appears to be small. [25] To further investigate the differences between photochemically and electrochemically reduced WO 3 , scanning electron microscopy (SEM), X-ray powder diffraction (XRD), and X-ray photoelectron spectroscopy (XPS) measurements were carried out for both WO 3 samples precharged to approx. 20 mC•cm −2 . As shown in Figure 5a, the surface morphology is typical for sputtered WO 3 and does not depend on the precharging method. [27] The XRD spectra ( Figure S4, Supporting Information) of the charged and pristine WO 3 samples show very broad peaks due to the low crystallinity. Thus, there are no structural changes observable compared to pristine WO 3 during the reduction of the sputtered WO 3 with low crystallinity. However, structural changes during the reduction of crystalline WO 3 were detected in the literature. [28] The XPS measurements in the W4f regions of pristine WO 3 (Figure 5b) show peaks for W 6+ , whereas the XPS spectra of the samples precharged by UV irradiation and electrochemical reduction display additional W 5+ peaks, as it is expected for partially reduced WO 3 (Figure 5c,d)  Thus, for both samples an irreversible reduction to tungsten bronze can be excluded.
The pristine WO 3 was electrochemically characterized in detail (Figure 6). A comparison of the cyclic voltammograms ( Figure 6a) with literature values confirms the position of the oxidation peak at ≈2.75 V vs. Li/Li + at 20 mV s −1 and the general shape of the curves. [26,30] In the charging/discharging curves (Figure 6b), a maximum charge density of 35 mC cm −2 was obtained. During galvanostatic cycling between 2 V and 4 V versus Li/Li + (Figure 6c), a drastic loss of charge density from initially 22.6 to 5.3 mC cm −2 after 100 cycles is observed. This can be assigned to the formation of tungsten bronze due to overreduction. [31] Asuitable approach to avoid such fast charge density decrease is to charge the WO 3 electrode only up to 15 mC cm −2 at a current density of 25 µA cm −2 (Figure 6d). The lower current density of 25 µA cm −2 in this measurement was chosen due to the charging/discharging behavior (Figure 6b). The coulombic efficiency, i.e., the ratio of the discharge density to the charge density, is ≈80% at 25 µA cm −2 and 70% at 50 µA cm −2 , respectively. Higher coulombic efficiencies >99% are obtained after the initial formation step over 100 cycles. Thus, a balanced EC cell with WO 3 photochemically precharged to 15 mC cm −2 and PB with 15 mC•cm −2 is targeted. According to Equation (3), WO 3 has to be irradiated until a τ v value of 27% is reached.
The corresponding irradiation time depends on the solvent according to Figure 3a and Table 2. Without the enhancement of the photochromic response by solvents, charge densities >6.4 mC cm −2 cannot be obtained, independent of the irradiation time.

ECDs Comprising Photochemically Precharged WO 3 and PB
Proof-of-concept ECDs (active area: 3 cm × 3 cm) with WO 3 (target charge density: 15 mC cm −2 ) and PB (15 mC cm −2 ) as electrode materials were assembled (ECD-15/15, Figure 7). The WO 3 electrodes were irradiated in an acetone atmosphere until a τ v value of ≈27% was reached. A LiTFSI-containing polymer electrolyte was used. The optical and (spectro-) electrochemical characterization of the PB electrode can be found in the supporting information ( Figure S5   state with τ v = 9% (Figure 7a,b). However, in the bleached state at 2.5 V, a residual coloration is observed (Figure 7b). In the completely bleached area, a τ v value of 76% is reached. From the transmission spectrum, the origin of the residual coloration cannot be determined due to the broad absorption bands of both the reduced H x WO 3 and the oxidized PB. Thus, the ECD was disassembled in its bleached state to analyze the EC electrodes by UV-Vis spectroscopy (Figure 7a). The residual electrolyte on the surface of the electrodes was not removed to avoid damaging the EC layers. The disassembled electrodes have a lower transmittance than the pristine ones due to light scattering on this electrolyte surface (Figure 7c). Nevertheless, when comparing the transmission spectra it is obvious that the WO 3 is in its bleached state, whereas the PB is partially colored. The photographs of the electrodes in Figure 7d confirm that. The charge density of WO 3 with 15 mC cm −2 was not sufficient to fully bleach the PB layer. Although two electrodes with equal charge densities are used, the reason could be a loss of charge density due to side reactions with the electrolyte. [32] In order to reduce the residual coloration in the balanced configuration (ECD-15/15) the WO 3 electrode was precharged to a charge density of ≈20 mC cm −2 (τ v = 18%) prior to the ECD assembly. The ECD-20/15 (Figure 8)  with ΔOD being the difference between the optical densities of the dark and bleached states at a specific wavelength, q the exchanged charge density, and T d and T b, the transmittance in the dark and bleached states, respectively. The charge density during switching between 2.5 V and −1.5 V was 15.9 mC cm −2 , T d = 0.49%, and T b = 77.46% resulting in a CE of 139 cm 2 C −1 .
In addition, the CE was also determined as the slope of ΔOD plotted against the charge density (see Figure S6, Supporting Information). With this method, a CE of 167 cm 2 C −1 at 716 nm was obtained. The ECD-20/15 was cycled over 230 cycles between 2.5 and −1.5 V while continuously measuring the τ v value (Figure 8b, raw data in Figure S7, Supporting Information). During the first 70 cycles, the visible light transmittance decreases from 8/79% (dark/bleached) to 13/67% at −1.5 and 2.5 V, respectively. After 130 cycles the ECD partially recovers and stabilizes at visible light transmittances of 11% (dark) and 69% (bleached). The degradation can be attributed to the high cell voltage in the bleached state. [33] By lowering the voltage from 2.5 to 1.5 V, the ECD-20/15 still reaches a τ v value of 75% in the bleached state. [33] Thus, the ECD-20/15 was cycled over 110 potentiostatic cycles at 1.5 V and at −1.5 V. However, this ECD ( Figure S8, Supporting Information) showed a loss of visible light transmittance to 21/58%. The switching times at the fifth cycle determined at 90% of the maximum τ v change are 2 min 20 s and 13 min 40 s for coloring at −1.5 V and bleaching at 1.5 V, respectively.
In Table 3, the performance of the ECD-20/15 is compared to state-of-the-art ECDs based on WO 3 and PB. The transmittance of the bleached state of the ECD-20/15 is exceptionally high (79%) while the transmittance of the dark state (8%) is comparable with literature values. [14,34,35] The visible light transmittance modulation (Δτ v = 71%) measured for the ECD-20/15 is among the highest reported in the literature. Jeong et al.
reported a transmittance modulation of 80% at the specific wavelength of 633 nm for an ECD based on nanoparticular WO 3 and PB, precharged by UV treatment of the assembled ECD, [13] which is slightly higher than the 77% in this study, but the CE with 123 cm 2 C −1 is lower than the value of 139cm 2 C −1 for the ECD-20/15. An ECD from the same group using electrochemically charged WO 3 with a CE of 139 cm 2 C −1 has a lower transmittance modulation of 65% at 633 nm. [35] However, a direct comparison of the different ECDs is difficult since the transmittance in the dark and bleached states or the transmittance change are measured at different wavelengths. Moreover, the CE can be determined either as the slope of ΔOD vs. the charge density or from single data points according to Equation (4), making a comparison difficult, as described above. In   [34] ed ed Tajima et al 2020 [14] np

Conclusion and Outlook
The enhancement of the photochromic response in WO 3 thin films on PET-ITO by acetone has been demonstrated. For flexible ECDs with PB as the ion storage layer it is promising as an alternative precharging method compared to electrochemical precharging in a liquid electrolyte. The precharging process can be monitored by a correlation of the charge density of the charged WO 3 electrode and the transmittance measured by UV-vis spectroscopy. In proof-of-concept ECDs an exceptionally high difference in the visible light transmittance τ v between the dark and bleached state is observed. A WO 3 electrode photochemically precharged with 20 mC cm −2 combined with PB (15 mC cm −2 ) results in a transmittance change from 8% (colored) to 79% (bleached) at −1.5 V and 2.5 V, respectively. A coloration efficiency (CE) at λ max = 716 nm of 139 cm 2 C −1 was reached. The cycle stability of the ECD was tested over 230 cycles. After formation (≈130 cycles) the visible light transmittance stabilizes at 13/67% (dark/bleached) at −1.5 V and 2.5 V, respectively. Further studies will focus on improving the electrochemical cycle stability and switching times of precharged WO 3 /PB ECDs. The photochemical precharging of the WO 3 does not require any direct contact with the electrode. By decreasing the illumination times it can be used in a continuous roll-to-roll process making the production of ECDs more cost-efficient, e.g., for smart window applications. The precharging method of WO 3 on PET-ITO might also be transferrable to WO 3 thin films on glass substrates.

Experimental Section
Chemicals and Substrates: PET-ITO was purchased from Eastman Chemical Company (Flexvue OC50, sheet resistance: ≈50 Ω □ −1 ). WO 3 deposited on PET-ITO by reactive dc magnetron sputtering was provided by ChromoGenics AB (layer thickness: ≈350 nm). PB on PET-ITO was produced by roll-to-roll slot-die coating (layer thickness: ≈440 nm). [37] SEM cross-section images of the WO 3 and PB electrodes are depicted in Figure S1 (Supporting Information). Silica glass Borofloat with a thickness of 3 mm was supplied by Prinz Optics GmbH. Modified polyisobutylene sealing tape (Helioseal PVS 101 Tape) was received from Kämmerling chemische Fabrik GmbH. Lithium, LiTFSI, and PC were purchased from Sigma-Aldrich. Ethanol and acetone were purchased from CSC Jäklechemie and Avantor, respectively. All chemicals were used without further purification.
UV Irradiation: The Atlas SUNTEST Cps+ with a xenon lamp and daylight filter at an intensity of 765 W m −2 was used as the UV light source. The bottom of the test chamber was covered with a black cardboard to prevent reflection. The irradiation was carried out under ambient atmosphere. During illumination, the temperature inside the test chamber was between 40 and 50 °C and the relative humidity was in the range of 10-20%.
Encapsulation of the WO 3 electrodes: WO 3 electrodes were encapsulated between two silica glass panes using a modified polyisobutylene sealing tape. The sandwich was put under metal plates and sealed by heating on a preheated electric heating plate for 6 min at 120 °C. For a solvent vapor atmosphere, 1.8 µL solvent cm −2 of the encapsulated area were dropped onto the lower glass pane with a syringe before sealing.
ECD Preparation: ECDs were prepared with an active area of 3 cm × 3 cm. The ECDs were assembled under an argon atmosphere (O 2 < 5 ppm, H 2 O < 1 ppm). Double-sided adhesive tape was used as spacer. A proprietary UV-curable polymer electrolyte (thickness ≈100 µm) containing LiTFSI was coated by using a doctor-blade onto the PB electrode and the WO 3 electrode was assembled on top. An adhesive copper tape was used for contacting the ECD. The curing of the electrolyte was carried out with a Hönle UVA HAND LED UV lamp with a wavelength of λ = 365 nm for 90 s. The distance between the lamp and the ECD was 4 cm.
Optical Measurements: Transmission spectra were recorded with a CCD-based Ava-Spec-3648 Standard Fibre Optic Spectrometer (Avantes). CIE L*a*b* color coordinates were calculated with the Avasoft 8 software with the standard illuminant D65 and the standard observer V(λ) at 10°.
(Spectro)Electrochemical Measurements: All measurements were performed with a Solartron Multistat 1470E potentiostat/galvanostat. Transmission spectra were recorded with a CCD-based Ava-Spec-3648 Standard Fibre Optic Spectrometer (Avantes). The measurements of the EC electrodes were carried out in a three-electrode setup with Li as CE and RE and 1 m LiTFSI in PC as electrolyte. Electrical contact was provided by an adhesive copper tape.
Potentiostatic discharging experiments of precharged WO 3 were carried out on samples with an active area of ≈1 cm × 1 cm in the spectroelectrochemical setup at 5 V versus Li/Li + .
The switching times of the ECDs were determined by monitoring the transmittance change. The τ v value reached a plateau when the switching was completed. The switching time was calculated at 90% of transmittance change. [38] SEM: The SEM images were recorded on a field emission scanning electron microscope Zeiss Auriga 60 with an SE2 detector.
XRD: The XRD was carried out on a Rigaku Smartlab 3 kW using Cu K α radiation.
XPS: XPS measurements were performed on a Surface Science Instruments S-Probe with upgraded detector using Al Kα radiation at 10 kV and a detector angle of 0°. The data were evaluated with the Hawk Data Analysis 7 software.

Supporting Information
Supporting Information is available from the Wiley Online Library or from the author.