2.1.

Thin Film Deposition and Device Fabrication

The distance of the target-to-substrate was about 8 cm. Prior to deposition, the vacuum chamber was evacuated to $\sim 6\times {10}^{-4}\mathrm{Pa}$, and the target was pre-sputtered for 15 min in pure argon gas to remove surface oxides and other impurities. The substrate was rotated at a speed of 20 times per minute during the deposition process. After deposition, the samples were kept in the chamber for 15 min for the thermal equilibration. Subsequently, these samples were extracted or subjected to the deposition of the next layer. The inorganic multi-color EC devices fabricated in this study consist of ${\mathrm{WO}}_3$ as the EC layer, ${\mathrm{Li}}_3{\mathrm{PO}}_4$ (LPO) as the electrolyte, WVO as the ion storage layer (i.e., EC inactive), and copper (Cu) as the current collector. The device is deposited on ITO-coated glass, and the sheet resistance is $6\mathrm{\Omega }$ per square ($\sim 60\mathrm{nm}$) with a visible light transmittance exceeding 85%. The final devices demonstrate a practical dimension of $4\times 4{\mathrm{cm}}^2$ and possess the structure of $\text{g}\text{lass}/\mathrm{ITO}/\mathrm{WVO}/\mathrm{LPO}/{\mathrm{WO}}_3/\mathrm{Cu}$ through a sequential deposition. Detailed sputtering parameters for the ${\mathrm{WO}}_3$, ${\mathrm{Li}}_3{\mathrm{PO}}_4$, and Cu layers are listed in Table 1.

Table 1

Deposition of the WO3, WVO, Li3PO4, and Cu layers.

2.2.

Characterizations

Electrochemical characterizations of single and multi-layers were performed in a 1 M ${\mathrm{LiClO}}_4\text{-}\mathrm{PC}$ electrolyte with a three-electrode geometry on two electrochemical workstations (CH Instruments, Inc.: CHI660E and Ivium Technologies B.V.: Ivium-N-STAT). For electrochemical measurements inside the glovebox, lithium foils served as the counter and reference electrodes in an argon-filled glovebox. For the measurement outside the glovebox, platinum foil served as the counter, and Ag/AgCl was the reference electrode.

Digital photographs of samples were taken with a Canon 850D camera. The chromaticity and reflectivity of the samples and devices are measured by a spectrophotometer (Shenzhen Threenh Technology: Benchtop Spectrophotometer TS8296). The in-situ optical transmission spectra under the electrochemical operations and the reflectivity of the devices were in-situ recorded by the fiber optical spectrometer (Avantes B.V.: AvaSpec-ULS2048CL-EVO).

X-ray diffraction (XRD) for structure analysis was conducted by a Rigaku Smartlab with a Cu $\mathrm{K}\alpha$ X-ray source operating at 40 kV. Multiple scans were collected between $2\theta =10\mathrm{deg}$–80 deg with a scan rate of 9 deg per min. Scanning electron microscope and energy dispersive spectroscopy (SEM-EDS) for morphography element analysis was measured by Hitachi (Regulus 8230) under a voltage of 20 kV. The optical indexes were measured and fitted by an ellipsometer (J.A. Woollam Company: M-2000 ellipsometers).

3.1.

Electrochromism-Based F–P Cavity for Dynamic Color Modulation

By incorporating the F–P cavity, traditional transmissive EC devices can be changed to be reflective. In traditional EC devices, the transmissive mode is controlled by absorption in the EC layer to achieve color variation.¹⁹ The varied intensity of absorption in different wavelengths leads to color variations. The F–P cavity can achieve color variation under the combined changes of refractive index and absorption. This leads to variations in the optical path difference in the cavity, thereby resulting in a reduction or enhancement of a certain wavelength range.

Inside the F–P cavity (i.e., it is ${\mathrm{WO}}_3/\mathrm{Cu}$ in our case), the incident light repeatedly refracts, reflects, and interferes, generating coherent enhancements or subtraction. The F–P cavity can be taken as a three-layered construction consisting of layer 1 (air), layer 2 (EC layer), and layer 3 (metallic reflective layer), as shown in Fig. 1(a). The total reflectivity of F–P cavity is given as^20–22

Fig. 1

Optical characterizations of the F–P cavity. (a) Schematic diagram of the F–P cavity constructed by ${\mathrm{WO}}_3/\mathrm{Cu}$. (b) Digital photos of the as-deposited F–P cavities with varied thicknesses of ${\mathrm{WO}}_3$. (c) Reflectance spectrum of the F–P cavity as a function of ${\mathrm{WO}}_3$ thickness. (d) Color coordinates for the F–P cavity with different thicknesses of ${\mathrm{WO}}_3$. (e) Digital photos of F–P cavities working under marked potentials. The potential range was $-1.7$ to 0.3 V (versus Ag/AgCl) with a sweep rate of $1\mathrm{mV}{\mathrm{s}}^{-1}$.

However, the thickness ($\mathit{d}$) and the optical index ($\mathit{n},\mathit{k}$) of the substrate (${\mathit{n}}_{\mathbf{3}}$) are normally fixed once the device is fabricated. Benefiting from our employed EC material, the optical index of the EC layer can be dynamically regulated in the ion intercalation process. Therefore, the modulated optical index leads to the variation of the optical path difference and absorption. In transmissive EC devices, ion insertion commonly induces an enhancement of the absorption of a certain wavelength range by EC layers, which means that the rest of the spectra can selectively pass through. Here, the dynamic color modulation is achieved through the utilization of an EC layer with a variable refractive index in the F–P cavity, i.e., the structure of ${\mathrm{WO}}_3/\mathrm{Cu}$. According to Eqs. (1) and (4), the more significant the disparity between ${\mathit{n}}_2$ and ${\mathit{n}}_3$ is, the greater the attainable reflectance. By combining the broad absorption formed by the F–P cavity, a higher color saturation is obtained.

As mentioned, our F–P cavity constructed by the ${\mathrm{WO}}_3/\mathrm{Cu}$ plays a core role in dynamic color modulation. This electrode is shown in Fig. 1(a), and the reflector (Cu) and dielectric layer (${\mathrm{WO}}_3$) are sequentially deposited on the ITO/glass. Cu was selected as the reflective layer, not only due to its significant optical mismatch with the EC layer but also due to its high reflectance of $>90\%$ beyond the wavelength of 600 nm. According to Eqs. (1) and (4), the more significant the disparity between ${\mathit{n}}_2$ and ${\mathit{n}}_3$ is, the greater the attainable reflectance is. By combining the broad absorption formed via the F–P cavity, a higher color saturation is obtained. By varying the thickness of the ${\mathrm{WO}}_3$ layer, the structural colors of magenta (red), orange, yellow, cyan (green), blue, and purple are obtained, as presented in Figs. 1(b) and 1(c). Among these colors, cyan, magenta, and yellow compose the three primary colors for the CMYK color model, and red, green, and blue comprise the tricolors for the RGB color model. In addition, the saturation of the cyan, purple, and magenta colors is beyond the sRGB color gamut, showing a very vibrant color [Fig. 1(d)]. The coordinates of the initial structural colors of six samples are revealed by modified-Bezier curves in the CIE coordinate system, demonstrating the immense potential to encompass the entire color spectrum in the F–P cavity [Fig. 1(b)]. Comparing with previous studies on ${\mathrm{WO}}_3/\mathrm{W}$ structures,^11,12 ${\mathrm{WO}}_3/\mathrm{Cu}$, with varied thicknesses of the ${\mathrm{WO}}_3$ layer, exhibits improved color saturation and brightness in forming the structural colors.

When the thickness of ${\mathrm{WO}}_3$ is fixed at 233 nm, the color of the F–P cavity changes from the initial yellow to cyan and magenta as ions are persistently inserted [Fig. 1(e)]. The color transition is less pronounced when the thickness is relatively thin, for example, 128 and 151 nm [Fig. 1(e)]. This is due to the decreased thickness of the EC layer failing to support the substantial absorption during the ion intercalation process.²³ Moreover, it is discernible that the reflectivity decreases with increasing the thickness of ${\mathrm{WO}}_3$, whereas the quantity of absorption peaks demonstrates a proportional escalation with increasing the thickness. The interference of the lower order primarily occurs with a thickness of about ${\mathit{m}}_{\text{low}}·\lambda /4{\mathit{n}}_2$ (where ${\mathit{m}}_{\text{low}}=1$). When the thickness is relatively thick ${\mathit{m}}_{\text{high}}·\lambda /4{\mathit{n}}_2$ (where ${\mathit{m}}_{\text{high}}$ is an odd number greater than 1), a higher-order interference immediately emerges in the spectrum. The number of absorption peaks in the spectrum gradually increases with the increase of the interference order, eventually yielding a low reflectance. This tendency can also be observed in the reflectance spectrum in that the absorption was split from a single-peak to a double-peak in the range of 130 to 230 nm [Fig. 1(c)]. Thus, the thickness should be appropriately selected in the potential device to balance the brightness of the structural color and the color variation range.

3.2.

Design for Focused Bragg Reflection

The Bragg reflector, a quarter-wave stack with a one-dimensional array in high and low refractive indexes, is strategically employed to construct the initial structural color. The Bragg reflector is constructed by the current collector (ITO), ion storage electrode (WVO), electrolyte (${\mathrm{Li}}_3{\mathrm{PO}}_4$), and EC layer (${\mathrm{WO}}_3$), all of which contribute to the high saturation in a quarter-wave stack. The reflectance peaks of the Bragg reflector can be varied according to thickness and variation of refractive indexes. The reflector was stacked based on the periodic laws of high and low refractive indexes interleave, and the parameters are described by the following equations:^22,24

According to Eq. (6), the arrangement of high and low refractive index layers is designed. The refractive indexes ($\mathit{n}$) of ITO, WVO, and ${\mathrm{WO}}_3$ are around 2.0, which are categorized as high refractive index parts. By contrast, LPO, with a refractive index close to 1.5, is selected as the low refractive index layer. The detailed parameters of the reflective layers are listed in Table 2. This leads to a refractive index mismatch at the interface of the high and low refractive layers. According to Eqs. (6) and (7), a more substantial distinction between ${\mathit{n}}_L$ and ${\mathit{n}}_H$ indicates a more pronounced mismatch. Therefore, a stacking approach with refractive indexes of approximately 2 and 1.5 is designed to form structural colors. It is worth noticing that the stacking number $\mathit{N}$ in Eq. (6) determines the intensity of the reflection peaks: a more considerable $\mathit{N}$ results in a sharper peak, which implies a higher contrast. However, considering the sandwich structure of an EC device, we determined the number of reflective layers at 3 for the EC layer, electrolyte, and counter electrode, respectively.

Table 2

Measured refractive index and thickness of each layer.

The correlation between the stacked functional layers and the quarter-wave can be validated from the experiments, as discussed below. The center wavelength of the Bragg reflector was determined based on Eq. (5), and the reflectivity was simulated with the assistance of the finite-difference time-domain method (FDTD). The thickness of each layer is fixed by aligning their refractive indexes. We demonstrate a Bragg reflector with a center wavelength of 600 nm as a representative. The refractive index and thickness of each layer are shown in Table 2, where a series of thicknesses are strictly controlled to follow a quarter-wave format. According to this design, the construction of the Bragg reflector ($\mathrm{ITO}/\mathrm{WVO}/\mathrm{LPO}/{\mathrm{WO}}_3$) was deposited layer by layer [Fig. 2(a)]. Primarily, the ITO-coated glass is transparent, and the WVO layer on ITO presents a light green color with a broad peak in the reflectance spectrum [Fig. 2(c)]. According to Eq. (2), the formation of broad reflection peaks is limited by the number of stacked layers, and the intensity of the reflection peak gradually increases with the number of stacking bilayers [Figs. 2(c) and 2(d)]. The initial green structural color gradually switches to yellow along with an increase in the number of stacked layers, as shown in the CIE coordinate in Fig. 2(b). Moreover, the optical indexes ($\mathit{n}$ and $\mathit{k}$) in the visible spectra used in our optical simulation are shown in Figs. 2(e) and 2(f).

Fig. 2

Bragg reflector design. (a) Schematic diagram of our Bragg reflector ($\mathrm{ITO}/\mathrm{WVO}/\mathrm{LPO}/{\mathrm{WO}}_3$), with $\mathit{H}$ and $\mathit{L}$ representing the high and low refractive indexes, respectively. (b) Color coordinates for the monolayer of WVO, WVO/LPO, and $\mathrm{WVO}/\mathrm{LPO}/{\mathrm{WO}}_3$ in CIE 1931 color space. (c) and (d) The reflectance of WVO, WVO/LPO, and $\mathrm{WVO}/\mathrm{LPO}/{\mathrm{WO}}_3$ from experiments and simulations. The solid line and dashed line represent the experiments and the simulations data, respectively. (e) and (f) The refractive index and extinction coefficient of ITO, WVO, ${\mathrm{Li}}_3{\mathrm{PO}}_4$, and ${\mathrm{WO}}_3$ in the visible range.

3.3.

Counter Elelctrode Selection

Traditional EC devices typically present a low reflectance or transmittance due to the simultaneous absorption of the two EC electrodes upon ion intercalation. By contrast, in devices consisting of a F–P cavity, the color variation is primarily achieved by altering the optical index of the EC layer during the EC process. To avoid excessive and non-gain absorption, we choose to employ an EC inactive material (WVO)²⁵ as the counter electrode.

The WVO film is co-sputtered from the pure W and V targets in the mixed atmosphere of Ar and ${\mathrm{O}}_2$. Its primary atomic ratio of W : V is 46.08 : 53.92 characterized by SEM-EDS, close to 1:1. Only the diffraction peaks from ITO were identified for the deposited WVO films on ITO/glass [Fig. 3(a)], indicating an amorphous nature of the as-deposited WVO films. The transmittance of WVO exceeds those of ${\mathrm{WO}}_3$ and ${\mathrm{V}}_2{\mathrm{O}}_5$ beyond the 500 nm wavelength [Fig. 3(b)], and the extinction coefficient ($\mathit{k}$) of WVO is almost zero as the wavelength is beyond 450 nm [Fig. 3(c)]. The high transmittance and low extinction coefficient enable the reduction of the negative optical impact, and it can therefore eliminate the influence on the reflectance in an assembled device. Compared with the light blue (${\mathrm{WO}}_3$) and yellow (${\mathrm{V}}_2{\mathrm{O}}_5$), the WVO exhibits a light green color that is stretched from the two colors and is close to the origin (i.e., the white point) in the CIE color space [Fig. 3(d)]. All figure-of-merits of WVO play a positive role for improving the brightness and saturation. Hence, the neutral light green colors are more suitable for the devices.

Fig. 3

Characterizations of mixed-tungsten-vanadium oxide (WVO). (a) XRD patterns of the WVO layer on ITO/glass (the black vertical line is the label of PDF#97-005-0848 for crystal ITO). (b) The reflectance of ${\mathrm{WO}}_3$, ${\mathrm{V}}_2{\mathrm{O}}_5$, and WVO. (c) Optical index of the WVO film. (d) Color coordinates for ${\mathrm{WO}}_3$, ${\mathrm{V}}_2{\mathrm{O}}_5$, and WVO. (e) Cyclic voltammetry curve of the WVO film. (f) The in-situ transmittance variation of the WVO film upon the CV cycle in the range of 2.0 to 4.0 V (versus $\mathrm{Li}/{\mathrm{Li}}^{+}$) with a sweep rate of $20\mathrm{mV}{\mathrm{s}}^{-1}$.

The WVO films were subjected to cyclic voltammetry in the potential window of 2.0 to 4.0 V versus $\mathrm{Li}/{\mathrm{Li}}^{+}$, and the results showed a maximum charge capacity of $22.5\mathrm{mC}{\mathrm{cm}}^{-2}$ [Fig. 3(e)] and nearly a non-variation of the optical spectra [i.e., EC in-active, Fig. 3(f)]. The obtained charge capacity of WVO matched well with amorphous ${\mathrm{WO}}_3$ ($\sim 20\mathrm{mC}{\mathrm{cm}}^{-2}$)²⁶ in the same potential range, enabling mutual charge balancing and stable potential windows for ion intercalation. The non-regulation in the spectra also proves that the operation of the WVO film itself rarely alters the optical modulation of a device.

3.4.

Full EC Device Demonstration

Finally, an all-solid-state EC device based on the combination of the Bragg reflector and F–P cavity was fabricated. The structure of the device is illustrated in Fig. 4(a), which is composed of $\text{glass}/\mathrm{ITO}/\mathrm{WVO}/\mathrm{LPO}/{\mathrm{WO}}_3/\mathrm{Cu}$. The device can be considered as two optical stacks: a Bragg reflector consisting of $\mathrm{ITO}/\mathrm{WVO}/\mathrm{LPO}/{\mathrm{WO}}_3$ and an F–P cavity formed by ${\mathrm{WO}}_3/\mathrm{Cu}$. The formed structural colors are facilitated by the designed Bragg reflection. All reflection peaks were focused at the designed wavelength ($\sim 600\mathrm{nm}$) in the reflective device, resulting in a uniform absorption in the device that is beneficial for achieving color purity. Meanwhile, the F–P cavity primarily undertakes the function of color variation. By incorporating of EC material (${\mathrm{WO}}_3$), the reflector with a fixed structure enables the transition from static structural colors to dynamic color modulation in the solid-state EC device. Based on the aforementioned experiments and FDTD simulations in Figs. 2(c) and 2(d), the thickness of each layer is fixed [Fig. 4(a)] to yield a center absorption wavelength at 600 nm of the EC device for demonstration. The direction of the color shift and its position in the CIE space are illustrated in Fig. 4(b), transitioning from an initial grass-green to the dark brown [Fig. 4(c)]. So far, the limited magnitude of this transition is attributed to the constraints of solid/solid interface during the fabrication process; for example, the ion intercalation between ${\mathrm{WO}}_3$ and LPO is insufficient for varying the optical constants in ${\mathrm{WO}}_3$ due to poor interface contact. We believe that the desired outcome can be achieved by improving the fabrication in the future. However, a high reflectance (37%) in the device [Fig. 4(c)] was still maintained due to the utilization of the WVO electrode and our optical stacking strategies. The color difference ($\mathrm{\Delta }{E}_{ab}^{*}$) is utilized to quantify the degree of color variation. $\mathrm{\Delta }{E}_{ab}^{*}$ is the separation between two colors and can quantify the small change of color variation. It is described in CIE coordinates of $L$, $a$, and $b$ (termed as CIELAB) as²⁷

Fig. 4

Optical performance of the solid-state EC device. (a) Schematic diagram of the device structure composed of a Bragg reflector and an F–P cavity. (b) The dynamic transition of color coordinates in the 1931 CIE color space upon ion intercalation. (c) The reflectance in the visible range of the device. Inset: digital photos of the color variation. (d) Variation of the color difference under the external electric field ($-2\mathrm{V}$) in different duration times.

To clarify the difference between the traditional device and our current design, we performed FDTD simulations. First, when the multilayer stacking is a random selection of the material and its thickness (i.e., traditional design), for example, $\mathrm{ITO}(300\mathrm{nm})/\mathrm{NiO}(400\mathrm{nm}){/\mathrm{LiNbO}}_3(200\mathrm{nm}){/\mathrm{WO}}_3(500\mathrm{nm})/\mathrm{Al}(300\mathrm{nm})$, as shown in Fig. 5(a), the irregular arrangement of refractive indexes and the arbitrary thickness design cause a discord resonance. Thus the structural color is hard to form, as revealed by the redundant interference fringe in the reflectance spectrum [Fig. 5(c)]. Such a chaotic optical stacking causes irregular enhancement and extinction, which hinders the reflection of an object from showing a pure color. By combining the Bragg reflector and the F–P cavity design into the device, optical interference can be taken as an effective approach for reflective EC devices to form the multicolor variation. This Bragg reflector is designed according to the alternating arrangement rule of high and low refractive indexes, and the thickness of each layer strictly coincides with the quarter-wave, with the primary objective of attaining consistent optical enhancement or suppression. This deliberate design seeks to optimize color saturation while circumventing irregular interference, which could otherwise lead to a reduction in the reflectance characteristics of the reflector. The configuration of $\mathrm{ITO}(65\mathrm{nm})/\mathrm{WVO}(105\mathrm{nm}){/\mathrm{Li}}_3{\mathrm{PO}}_4(220\mathrm{nm}){/\mathrm{WO}}_3(170\mathrm{nm})/\mathrm{Al}(300\mathrm{nm})$ was taken as our example [Fig. 5(b)], by carefully selecting the material of each layer and its thickness. In this device combining the Bragg reflector and F–P cavity, the resonance absorption peak was designed at 500 nm, showing a remarkable orientation in color [Fig. 5(d)].

Fig. 5

Structural design for the traditional device and our device combined of the Bragg reflector and the F–P cavity. (a) and (b) Schematic diagram of the traditional device and our device. (c) and (d) Simulated the reflectivity of the traditional device and our device.