All-Solid-State Proton-Based Tandem Structure Achieving Ultrafast Switching Electrochromic Windows

Zewei Shao Shanghai institute of Ceramics, Chinese Academy of Sciences Aibin huang Shanghai institute of Ceramics, Chinese Academy of Sciences Chen Ming Shanghai Institute of Ceramics John Bell The Queensland University of Technology Pu Yu Tsinghua University https://orcid.org/0000-0002-5513-7632 Yi-Yang Sun Shanghai Institute of Ceramics Liangmao Jin State Key Laboratory of Advanced Technology for Float Glass Liyun Ma State Key Laboratory of Advanced Technology for Float Glass Hongjie Luo Shanghai University Ping Jin Shanghai institute of Ceramics, Chinese Academy of Sciences Xun Cao (  cxun@mail.sic.ac.cn ) Shanghai institute of Ceramics, Chinese Academy of Sciences


Introduction
Electrochromic smart windows, which can regulate the transmittance of solar radiation by applying a voltage, have been regarded as a promising method for enhancing building energy efficiency. [1][2][3][4] Overall consideration on production cost, coloration efficiency, response speed and cycle durability suggests that tungsten trioxide (WO3) is the most preferred candidate for large-scale application of electrochromic devices (ECDs). [5][6][7] Previous works suggested that when small ions such as H + , Li + , Na + or Al 3+ are inserted into the WO3 lattice, transitions between small-polaron states associated with W ions of different valences could result in optical absorptions that are responsible for coloration of WO3. [8][9][10][11] The reversible injection and extraction of the extrinsic ions as controlled by electric field form the basic mechanism of a WO3-based ECD.
The ionic radius of the insertion ions is a critical factor determining the coloration speed of an ECD, as well as its durability because of the damage to the lattice during cycling. As shown in Fig. 1a, protons with the smallest radius and mass migrate much faster than other ions and yield higher electrochromic performance. [12][13][14][15][16] However, in previous studies, protons are exclusively introduced by liquid electrolytes. 15,[17][18][19] Despite the superiority of the proton-based ECDs, the use of liquid electrolyte renders them not appealing for many practical applications. [20][21][22] Another challenging issue using protons as insertion ions is the generation of H2 gas if the applied potential is over the electrochemical potential of the H + /H2 pair, which incurs serious safety issue. 23 Recently, studies on neuromorphic materials reveal that the organic conductive polymer poly (3,4-ethylenedioxythiophene):polystyrene sulfonate (PEDOT:PSS) can release protons under an electric field. This finding has been used to fabricate memristive devices. 24,25 In the polymer mixture, PEDOT functions as an organic semiconductor while PSS functions as a dopant and provides hole carriers to PEDOT by removing a portion of H atoms from the -SO3H groups. Because of the transparent and conductive nature of PEDOT:PSS, previous studies have explored the use of PEDOT:PSS in ECDs. [26][27][28][29] But in these studies, instead of a proton source PEDOT:PSS serves as an electrode or an electrochromic layer only. [30][31][32][33] A detailed comparison of these previous works has been provided in the Supporting Information (SI , Table S1).
In this paper, we design an all-solid-state tandem structure for the ECDs, which is composed of an electrochromic WO3 layer, a PEDOT:PSS layer as the proton source, and a solid polymeric electrolyte (SPE) layer on top of the PEDOT:PSS layer. This design is not merely a proof of concept using solid-state proton source in ECDs, but it also demonstrates excellent performance in coloration speed, optical modulation, and durability. As will be shown, the tandem structure manifests its multiple functions: 1) the SPE provides Na + ions to PEDOT:PSS and pumps protons to WO3 realizing a relay of the insertion ions; 2) the insulating SPE layer carries most of the applied voltage so that H2 formation in PEDOT:PSS is avoided during high-voltage operation; 3) PEDOT:PSS as an electrochromic material supplements the light absorption of WO3 and enhances the optical modulation of the overall device. Based on this design, we have fabricated large-area as well as flexible devices to demonstrate its great potential for large-scale applications.

Results and discussion
Design of the tandem-structure ECD. To achieve an all-solid-state design and use protons as insertion ions, we employed a PEDOT:PSS film as proton source. We fabricated an ECD composed of five layers: ITO/WO3/SiO2/PEDOT:PSS/ITO, where the indium tin oxide (ITO) layer serves as a transparent conducting layer and the SiO2 layer between the WO3 and PEDOT:PSS layers serves as an ion-conductive and electron-insulating layer. The upper panel of Fig. 1b shows the transmittance changes of this device in response to voltage pulses with the width varying from 40 to 5/16 s.
The wavelength of the illumination light is 670 nm. Under voltage pulse of 40 s, the transmittance change (i.e., the difference between the maximum and minimum transmittances, or ΔT) is 15%. Even when the pulse width is reduced to 5/8 s, the device is still able to response by a ΔT of 13%. When the pulse width is further reduced to 5/16 s, ΔT quickly drops to 8% suggesting the switching time is in between 5/8 and 5/16 s. This is an encouraging result as it shows that 1) PEDOT:PSS can serve as a proton source in an ECD and 2) it can yield comparable switching time to the liquid proton sources, which is typically in the sub-second level. 16,17 Even though the switching speed by using protons as insertion ions is fast, the small ΔT of 15% is not sufficient for electrochromic applications. We attributed this small ΔT to the limited amount of removable protons in the PEDOT:PSS layer. To solve this problem, we design a tandem structure by adding an extra SPE layer containing NaClO4 and ferrocene on top of the PEDOT:PSS layer with an aim to use Na + ions from the SPE layer to pump more protons to the WO3 layer. Here, ferrocene was added into the electrolyte as a redox mediator in the ECD to compensate the charge lost due to Na + ion extraction. 34 This design is motivated by the fact that NaClO4-containing SPE is a widely used Na + source in WO3-based ECDs and meanwhile PSS is a well-known ionexchange material. [35][36][37] Before fabricating a full device, we first examined the response time of the SPE/PEDOT:PSS junction to voltage pulses of different widths. We used a device structure of ITO/SPE/PEDOT:PSS/ITO. The lower panel of Fig. 1b shows the ΔT as a function of voltage pulse width. It can be seen that this device without the WO3 layer also exhibits electrochromic behavior. At pulse width of 40 s, ΔT is 30%. When the pulse width is reduced to 5/4 s, ΔT still maintains at 28%. When the pulse width is further reduced to 5/8 s, ΔT drops to 18%, suggesting that the SPE/PEDOT:PSS junction also has a switching time at sub-second level and is suitable for use in a tandem structure with the PEDOT:PSS/WO3 junction. Fig. 1c shows a schematic diagram for the tandem structure with five layers, i.e., ITO/SPE/PEDOT:PSS/WO3/ITO. The expected working mechanism is that the protons are the main insertion ions to the WO3 layer and responsible for the coloration. The SPE layer only provides Na + ions to the PEDOT:PSS layer, but the Na + ions do not diffuse into WO3 in a significant amount. Instead, the Na + ions mainly play the role of pumping protons out of PEDOT:PSS through an ion-exchange process.
To justify this expected mechanism, we first carried out density functional theory calculations. We compared the cases with either a Na + ion or a proton from the PSS model injected into WO3, as shown in the Fig. 1c. We found that the insertion of Na + into WO3 is 0.46 eV higher in energy than a proton. The high coordination of Na atoms by O atoms from two side chains of PSS (see Fig. 1c and Fig. S1 in the SI) plays the key role of stabilizing a Na + ion in PSS over a proton.
We further verified this process by carrying out X-ray photoelectron spectroscopy (XPS) experiments on three different samples as shown in Fig. 1d, where both the SPE/WO3 and SPE/PEDOT:PSS/WO3 samples were prepared after the coloration by ion insertion. Before the XPS experiments, all capping layers above WO3 were removed.
Consistent with the calculation results, the contents of Na show obvious differences among these samples. Compared to SPE/WO3, the significantly less Na content in the colored WO3 from the SPE/PEDOT:PSS/WO3 sample indicates that the coloration of WO3 is caused by proton insertion.  where D is in the unit of cm 2 •s -1 , ip is the peak current (in mA), A is the working electrode area (in cm 2 ), c is the concentration of active ion (mol•cm -3 ), v is the scan rate (mV•s -1 ), n is the number of electrons assumed to be 1. In the case without the PEDOT:PSS layer, the diffusion current is contributed to Na + insertion from SPE to WO3. In the case with the PEDOT:PSS layer, the diffusion current is mainly contributed by proton insertion from PEDOT:PSS to WO3, which should be accompanied by Na + insertion from SPE to PEDOT:PSS to fill the H vacancies, as shown in Fig. 1c. As expected, the kinetics of proton diffusion is indeed significantly faster than Na + diffusion, which explains the enhanced performance by using the PEDOT:PSS layer.  Fig. 3a. It can be seen that after 3000 cycles the device still retains over 90% of its initial optical modulation (i.e., less than 10% degradation). The inset of Fig. 3a shows the transmittance spectra of the 1000th, 2000th, and 3000th cycles confirming the consistency over the cycling. In contrast, the traditional ECD based on the simple SPE/WO3 structure not only has a slow switching speed (~3.9 s for coloration and ~9.8 s for bleaching as shown in Fig. S7 in the SI), but also poor cycling stability suggesting that protons introduce less damage to the lattice than the larger Na + ions during the cycling. From the CV measurement (the red-colored arrow in Fig. 2g), it can be seen that the peak current is nearly unchanged for high scan rates indicating that the second stage involves a slow ion diffusion process. From Fig. 2g, it can also be seen that the coloration also involves two peaks. But these two peaks show similar scan-rate dependence, i.e., having similar ion diffusion rates, which explains that the coloration in Fig. 3b does not exhibit two stages. As a comparison, the result for the traditional SPE/WO3 structure without the PEDOT:PSS layer is also shown in Fig. 3b. The significant improvement by introducing the PEDOT:PSS layer can be clearly seen.
It is worth noting that the PEDOT:PSS layer has an optimized thickness. If the layer is too thick, the intrinsic light absorption will reduce the transmittance in the off state as shown in Fig. S8a-f. If the layer is too thin, it cannot provide sufficient protons to the WO3 layer. We fabricated a number of devices with varying thickness of the PEDOT:PSS layer. The thickness was controlled by the times of spin coating. The thickness of the WO3 layer was fixed at about 300 nm, while the thickness of the SPE layer is usually much thicker (see Fig. 2b) and therefore does not affect the device performance significantly. As shown in Fig. 3c, the thicker the PEDOT:PSS layer is, the faster the coloration speed is until reaching the intrinsic limit of proton diffusion, which is estimated from our results to be 0.7-0.9 s (specific process can be seen in Fig.   S9). In the best-performed device, the PEDOT:PSS layer is of the order of 50 nm as shown in Fig. 2b, which matches with the 300-nm WO3 layer.

Another important performance indicator for an ECD is the coloration efficiency
(CE), defined as the optical density changes (ΔOD) per unit of charge injection (ΔQ).
A high CE value indicates a large optical modulation with a small energy input. As determined from Fig. 3d, the CE for the tandem structure with PEDOT:PSS is 109 cm 2 C -1 at 670 nm, while the CE for the device without PEDOT:PSS is 93.5 cm 2 C -1 . As a comparison, a typical LiTaO3-based ECD shows a CE of 73.5 cm 2 C -1 (see Fig. S10 in the SI). initial ΔT with pulse width of 20 s is ~87% and ~70% for the tandem and traditional structure, respectively. As the pulse width is reduced to 10, 5, 5/2, 5/4, 5/8, and 5/16 s, the ΔT of the tandem structure decreases to 86%, 81%, 71%, 65%, 58%, and 48%, respectively, and the ΔT of the traditional structure decreases to 65%, 60%, 45%, 35%, 25%, and 15%, respectively (Fig. S11 in the SI). This result further demonstrates the superiority of the tandem structure over the traditional structure from the aspect of optical modulation. The 17% increase in the CE can be attributed to the second coloration stage. It also suggests that the main contributor to coloration in the tandem structure is the WO3 layer, while the PEDOT:PSS mainly serves as an intermediary layer for the relay of protons and Na + ions. This is consistent with the results in Fig. 1 that without the synergy between the three layers in the tandem structure, the optical modulation is rather low.
A high-performance ECD is typically characterized by the switching time and the optical modulation (i.e., the maximum ΔT between on/off states). We summarize the  To test the scalability, we fabricated 10×10 cm 2 windows based on the tandem structure.
Photographs in Fig. 4a show the uniformity of the coloration at different applied voltages as inspected by naked eyes. The transmittance changes at 670 nm recorded at the center and the edge of the window are shown in Fig. 4b. Full spectra are shown in irradiance converted from the transmittance spectra is shown in Fig. 4c and Fig. 4d, which demonstrate high optical modulation for the full solar spectrum. Besides, we also fabricated a 30×40 cm 2 window. As shown in Fig. 4e, uniform coloration can also be achieved (more detailed results are given Fig. S13 in the SI).
The tandem structure discussed above could be easily adapted to make flexible  Center a relay process of Na + ions and protons in the PEDOT:PSS layer, i.e., using Na + ions from the SPE layer to pump more protons from PEDOT:PSS to WO3. Overall, the ECD based on the all-solid-state tandem structure achieved ultrafast response (coloration to 90% in 0.7 s and bleaching to 65% in 0.9 s and 90% in 7.1 s), high optical modulation (about 90%) and excellent durability (<10% degradation after 3000 cycles). We also demonstrated scale-up capability of the all-solid-state tandem structure by fabricating large-area ECDs up to 30×40 cm 2 with appealing performance. A two-stage electrochromism was observed in the tandem structure, which were dominated by proton insertion to WO3 and Na + insertion to PEDOT:PSS, respectively. This two-stage ion insertion not only helps enhance the optical modulation of the ECD, but may also find applications in, e.g., information displays and triple-state optical devices.
Transparent indium tin oxide (ITO) glass electrode and flexible Ag NW-PET were purchased from CSG Holding Co., Ltd.

Material and device characterization
UV-vis transmittance/absorption spectra were measured using a Hitachi U-4100 spectrophotometer. FT-IR spectra were measured using a Thermo Fisher Scientific Nicolet iS10 spectrophotometer. Cyclic voltammograms and EIS were measured using a Zahner IME6 electrochemical work station. Wetting contact angles were measured using a JY-82A contact angle meter. The EC devices were analyzed by scanning electron microscopy (SEM) using a Hitachi SU8220 equipped with an energy dispersive X-ray spectrometer (EDS) from Oxford X-Maxᴺ.

Preparation of WO3
The amorphous WO3 layer was deposited by re-active DC magnetron sputtering for 30 min at a power of 70 W, an Ar/O2 ratio of 94:6 and a pressure of 2 Pa.

Preparation of solid polymer electrolyte (SPE)
NaClO4/PC (1 mol/L), ferrocene and TTA21 in a ratio of 1:0.1:1 g were mixed with 2 mL of PMA and then stirred for 30 min. After that, 0.01 g iodonium salt (1,2-Dimethyl-3-Propylimidazolium Iodide) was added into the mixtures and stirred for another 5 min.
The pristine electrolyte precursor was a gel-like mixture (2 mL), which was spin-coated (1000 rpm, 20 s) on the ITO/quartz electrode and then capped by another electrode.
After curing for 30 seconds by a 300-W UV lamp, the electrolyte is solidified and meanwhile binds the two ITO layers. Here, NaClO4 dissolved in PC was used as a Na + cation source. Ferrocene was used as a reducing (charge-compensating) agent in the electrochromic process. TTA21 and iodonium salt (1,2-Dimethyl-3-Propylimidazolium Iodide, Adamas) were used as UV-curing monomer and initiator.

Preparation of PEDOT: PSS
PEDOT: PSS dispersion was spun onto the bottom ITO/WO3 electrode at 1000 rpm for 60 s, followed by 3000 rpm for 20 s. Then the substrates were heated to 80 ℃ with a rate of 10 ℃/min followed by annealing under vacuum condition.

First-principles calculation
The first-principles calculation was implemented in VASP program. 39 Projector augmented wave (PAW) potentials were used to describe the core-valence interaction. 40 Plane waves with kinetic energy up to 408 eV were used as the basis set. Perdew-Burke-Ernzerhof (PBE) functional 41 was used in the calculation. All the structures were relaxed until the force on each atom was less than 0.03 eV/Å and then the energy was calculated. Structural models are described in Fig. S1 in SI.  square root of scan rates using the results from (f) and (g). For the case with PEDOT:PSS, the peaks marked by blue arrows in (g) were used.