Infrared Electrochromic Devices Based on Thin Metal Films

Tunable emissivity technology is promising for the dynamic regulation of infrared radiation. Herein, infrared electrochromic devices based on thin metal films that operate via a novel hydrogen‐induced metal–insulator transition are demonstrated. The use of thin magnesium–nickel (MgxNi) alloy films as both a variable emissivity material and top conductive electrode simplifies the device structure and ensures that large changes in emissivity can be achieved. The constructed sandwich‐structured electrochromic devices also have polyethyleneimine (PEI) as a middle proton‐conducting electrolyte layer and hydrogen tungsten bronze (HxWO3)/indium tin oxide (ITO) as a bottom ion‐storage layer. Upon application of a voltage of ±2.6 V, the emissivity of the MgxNi/Pd/PEI/HxWO3/ITO device can be reversibly regulated, with emissivity changes of 0.48 and 0.43 in the 3–5 and 7.5–14 µm atmospheric windows, respectively. Under open‐circuit conditions, the high‐emissivity state of the device can be stably maintained for 3 h. The emissivity change is affected by the composition and thickness of the MgxNi film and the device failure mechanism involves the breakage and oxidation of this film after cycling. Corresponding flexible devices that exhibit electrochromism in the visible region have great potential for adaptive thermal camouflage, smart thermal management, and dynamic information displays.


Infrared Electrochromic Devices Based on Thin Metal Films
can be achieved by tuning the temperature [2,6] or the emissivity. [10][11][12] Compared with temperature control, tunable emissivity has the advantages of feasibility and energy efficiency. Emissivity is commonly modulated using infrared electrochromic devices, [13,14] which exhibit reversible changes in infrared emissivity under an applied voltage. Most such devices are based on reversible metal electrodeposition [15,16] or ion injection/extraction in metal oxides, [17] conducting polymers, [18] and graphene. [19] The magnitude of the emissivity change is an important device property and research efforts have focused on realizing large emissivity changes in a wide spectral window.
In previous studies, [15] we achieved substantial emissivity modulation using metal electrodeposition. The effectiveness of this metal manipulation approach suggested that metal-insulator transitions or large changes in electrical conductivity may also produce substantial changes in infrared optical properties, which could be used to dynamically modulate the infrared radiation properties. Typically, infrared electrochromic devices adopt a sandwich structure in which the top electrode is both conductive and infrared transparent (i.e., top conductive electrodeelectrolyte-counter electrode). The properties of the top conductive electrode, especially the infrared transmittance, greatly affect the emissivity change of a device. Common top conductive electrodes include Si and Ge electrodes, [20] metal grid electrodes, [21] and ultrathin indium tin oxide (ITO) electrodes. [22] Although Si and Ge electrodes are highly conductive, they tend to reflect infrared light from the surrounding environment, which reduces the magnitude of the emissivity change. In contrast, metal grid electrodes do not reflect surrounding infrared light, but they are expensive and require complicated preparation processes. Despite having good electrical conductivity, the infrared transmittance of ultrathin ITO electrodes is low. Thus, it is difficult to obtain top electrodes that combine good electrical conductivity and high infrared transmittance.
To overcome these issues, it may be possible to integrate the variable emissivity layer with the top conductive electrode. Recently, it was found that graphene-based infrared electrochromic devices do not require a top conductive electrode owing to the good conductivity of multilayer graphene with different doping levels. [23] Therefore, materials that exhibit large changes in electrical conductivity and can be used as electrodes are of interest. Notably, various thin metal films retain some electrical conductivity after complete hydrogenation [24] and are Tunable emissivity technology is promising for the dynamic regulation of infrared radiation. Herein, infrared electrochromic devices based on thin metal films that operate via a novel hydrogen-induced metal-insulator transition are demonstrated. The use of thin magnesium-nickel (Mg x Ni) alloy films as both a variable emissivity material and top conductive electrode simplifies the device structure and ensures that large changes in emissivity can be achieved. The constructed sandwich-structured electrochromic devices also have polyethyleneimine (PEI) as a middle proton-conducting electrolyte layer and hydrogen tungsten bronze (H x WO 3 )/indium tin oxide (ITO) as a bottom ion-storage layer. Upon application of a voltage of ±2.6 V, the emissivity of the Mg x Ni/Pd/PEI/ H x WO 3 /ITO device can be reversibly regulated, with emissivity changes of 0.48 and 0.43 in the 3-5 and 7.5-14 µm atmospheric windows, respectively. Under open-circuit conditions, the high-emissivity state of the device can be stably maintained for 3 h. The emissivity change is affected by the composition and thickness of the Mg x Ni film and the device failure mechanism involves the breakage and oxidation of this film after cycling. Corresponding flexible devices that exhibit electrochromism in the visible region have great potential for adaptive thermal camouflage, smart thermal management, and dynamic information displays.

Introduction
In recent years, technology for dynamically regulating infrared radiation has attracted extensive research attention owing to its application prospects in adaptive thermal camouflage, [1][2][3][4][5][6] smart thermal management, [7][8][9] and dynamic information displays. [2,10] According to the Stefan-Boltzmann law, infrared radiation from a surface is proportional to both the surface emissivity and the fourth power of the surface temperature. Therefore, effective dynamic modulation of infrared radiation www.advmatinterfaces.de suitable as electrodes. Furthermore, the resistivity of a thin metal film increases dramatically following hydrogenation, [25] which may cause a large change in infrared emissivity.
Rare earth metals, transition metals, and their alloys can undergo metal-insulator transitions upon reaction with hydrogen or hydrogen ions, which can induce large changes in infrared absorption. In 1996, Huiberts et al. discovered the hydrogen-induced discoloration of yttrium film during hightemperature superconductivity experiments. [26] Further studies have shown that other rare earth metals, such as La, Sm, Ce, and Gd, have similar characteristics. Initially, metal hydride films were mostly used for the safe detection of hydrogen concentrations. [27] Subsequently, research on metal hydrides expanded to smart windows, and Mg-rare earth metal alloy films were developed to meet the requirements of this application. Compared with pure rare earth metal films, Mg-rare earth metal films are colorless in the hydrogenated state and have higher transmittance in the visible region, making them more suitable for smart window applications. [28][29][30] However, most previous research on the spectral modulation of metal hydrides has focused on the visible region rather than the infrared region.
In this work, an infrared electrochromic device that integrates a variable emissivity layer and the top conductive electrode was developed based on the hydrogen-induced metalinsulator transition of a Mg-Ni alloy film. In this device, the emissivity was regulated by applying a voltage to realize the dynamic modulation of infrared radiation. Large emissivity changes of 0.48 and 0.43 were achieved in the 3-5 and 7.5-14 µm regions, respectively. Under open-circuit conditions, the highemissivity state of the device remained stable for 3 h. The introduction of Ni promoted the reaction of Mg with hydrogen to improve the response speed of the device. Failure of the device was caused by defects formed during the reaction between Mg and hydrogen. In addition, flexible devices with visible electrochromism could be fabricated.

Device Design and Structure
The necessity of electric field excitation typically requires that electrochromic devices have a battery-like structure (i.e., electrode-electrolyte-electrode), from which traditional sandwichstructured devices have evolved. The devices constructed in this work had a sandwich structure consisting of a variable emissivity part, electrolyte, and ion-storage part (Figure 1a). A barium fluoride (BaF 2 ) lens was used as an infrared-transparent substrate because of its high infrared transmittance. Mg x Ni alloy and 5 nm Pd layers were successively deposited on the BaF 2 lens. Pd was used as a catalytic layer to provide a channel for hydrogen transport. Owing to its ability to form mixedvalence compounds with cations (H + , Li + , Na + ), WO 3 can be used to store ions. Therefore, a 600 nm WO 3 film was deposited www.advmatinterfaces.de on the ITO/glass substrate as an ion-storage layer. The WO 3 / ITO/glass substrate was immersed in sulfuric acid solution to inject a 0.1 C charge into the WO 3 film by electrolysis, which converted WO 3 into hydrogen tungsten bronze (H x WO 3 ). Polyethyleneimine (PEI) has good cationic conductivity, is neutral, does not corrode metal films, and can block electron conduction, thus avoiding short circuits. Therefore, PEI was chosen as the electrolyte for the infrared electrochromic device. The final device was assembled by coating PEI on the H x WO 3 /ITO/ glass substrate and then placing BaF 2 /Mg 3 Ni/Pd on top of this structure. The production process is shown in Figure S1 in the Supporting Information, and the appearance of the constructed device is shown in Figure S2 in the Supporting Information.
To obtain a large emissivity change, we investigated the effects of the thickness and composition of the Mg-Ni alloy layer. As shown in Figure 1b,c, when the proportion of Ni is low (Mg 4.48 Ni), the emissivity change of the device is small. This is because hydrogen atoms entering the magnesium-rich metal film generate MgH 2 locally. As MgH 2 hinders the diffusion of hydrogen atoms, the complete reaction of Mg becomes difficult. [31] The device with the 50 nm Mg 3 Ni alloy layer exhibits a large emissivity change of 0.35 in the 3-5 µm region, whereas the device with the 50 nm Mg 1.47 Ni alloy layer has a large emissivity change of 0.31 in the 7.5-14 µm region. The emissivity curves of devices with metal layers containing different Mg/Ni ratios are shown in Figure S3 in the Supporting Information. The thickness of the alloy layer also affected the device performance, with the change in emissivity first increasing and then decreasing as the alloy layer thickness increased (Figure 1d,e). This phenomenon is mainly due to the reflective properties of the metal being more pronounced for thicker alloy layers and the emissivity of the initial state of the device being low. The device with a 100 nm Mg 3 Ni film has a small emissivity change because the amount of protons stored in the WO 3 layer is not sufficient for complete transformation of the alloy layer. In this case, the alloy layer remains partly metallic after hydrogenation, resulting in the device having a lower emissivity in the highemissivity state. The emissivity curves of devices with Mg 3 Ni layers of different thicknesses are shown in Figure S4 in the Supporting Information.
The device with the structure of Mg 3 Ni (80 nm)/Pd (5 nm) has a large emissivity change. As shown by the cross-sectional morphology of the Pd/Mg 3 Ni (80 nm) film in Figure 1f, a clear interface exists between the Pd layer and the underlying metal film. The good continuity of the Pd layer provides a complete channel for the longitudinal transport of hydrogen atoms. Energy-dispersive spectroscopy (EDS) mapping (Figure 1g) confirmed that the main components of the alloy layer are Mg and Ni. Although the Mg-Ni alloy film is crystalline, many dislocations are present (Figure 1h), and the lattice defects in this disordered structure can provide channels to facilitate the diffusion of hydrogen atoms. The presence of numerous dislocations is consistent with the absence of distinct diffraction spots in the selected area electron diffraction pattern (Figure 1i).

Infrared Electrochromic Performance and Mechanism
After applying a voltage of −2.6 V for 150 s, the device with the structure of Mg 3 Ni (80 nm)/Pd (5 nm) switched completely from a low-emissivity state to a high-emissivity state, with emissivity changes of 0.48 and 0.43 in the 3-5 and 7.5-14 µm regions, respectively (Figure 2a). After removing the voltage, the high-emissivity state of the device was stably maintained for a long period of time. The emissivity change of the device in 2.5-14 µm region decreased by 7.3% after 3 h and 21.1% after 6 h (Figure 2b and Figure S5, Supporting Information). Thus, we assume that the device remains stable in the high-emissivity state for 3 h under open-circuit conditions. MgH 2 and Mg 2 NiH 4 have good thermal stability and decompose slowly at room temperature. During the process of regulating the device emissivity, the apparent temperature of the device increased gradually, whereas it decreased again when a voltage of 2.6 V was applied. The change of device state was reversible, and the maximum change in apparent temperature was 7.2 °C, as shown in Figure 2c,d. This change in the apparent temperature of the device in the 7.5-14 µm region is shown in Video S1 in the Supporting Information, whereas the actual temperature of the device remains constant (42.5 °C) throughout the entire experimental process. Similarly, as shown in Figure 2e,f, the apparent temperature change of the device in the 3-5 µm region was ≈9.3 °C and the response time was no more than 160 s. The effective working area of the device is about 5.8 cm 2 . The modulation of the apparent temperature in this region is shown in Video S2 in the Supporting Information. These results confirm that the prepared infrared electrochromic device with a simple structure enables substantial emissivity modulation over a wide wavelength range.
The reaction process between the Mg 3 Ni alloy film and hydrogen is faster than that of the pure Mg film owing to the presence of more grain boundaries in the alloy phase after the introduction of Ni. As metal hydrides nucleate at grain boundaries, [32] the introduction of Ni increased the number of nucleation sites during the reaction and enhanced the reaction rate. Cyclic voltammetry tests revealed the oxidation and reduction potentials required to obtain a reasonable drive voltage (Figure 3a). The anode current peak (2.4 V) represents the conversion of hydrogen atoms into protons, whereas the cathode current peak (−2.4 V) represents the reverse process. To ensure that the state of the device can be fully transformed in a short time, we chose ±2.6 V as the excitation voltage. Upon complete hydrogenation, the Mg-Ni alloy film became semi-absorbing and semi-transmissive in the infrared band ( Figure S6, Supporting Information), whereas the electrolyte was highly absorbing in this band. Therefore, the infrared transmission of the Mg 3 NiH 6 /Pd film could be converted into infrared absorption. In this way, the device underwent a transition from a high infrared reflecting state to a high infrared emitting state, as shown in Figure 3b.
Previous studies have only given a general reaction equation for the emissivity change process of such devices without any in-depth analysis. Herein, we meticulously analyzed the reaction mechanism during one cycle of the device. Upon applying a voltage of −2.6 V to the device, electric field excitation caused protons to leave the WO 3 layer and reach the surface of the Pd layer via conduction through the PEI electrolyte. Because the Mg 3 Ni alloy and Pd layers work together as the top conductive electrode of the device and reduction reactions usually occur on the cathode surface, [33] the protons are electronically www.advmatinterfaces.de reduced to hydrogen atoms on the surface of the Pd layer and subsequently cross the Pd layer into the Mg 3 Ni layer. Once the diffusion process reaches saturation, the hydrogen atoms react (Mg 2 Ni + 4H ad → Mg 2 NiH 4 , Mg + 2H ad → MgH 2 ). [34] When Mg 3 Ni has completely reacted, Mg 2 NiH 4 and MgH 2 decompose spontaneously (Mg 2 NiH 4 → Mg 2 Ni + 4H ad , MgH 2 → Mg + 2H ad ). The hydrogen atoms accumulate at the interface between the Pd layer and the electrolyte layer until a certain concentration is reached and then diffuse into the hydride of the Mg 3 Ni alloy layer. In this way, the formation and decomposition of the hydride of the Mg 3 Ni alloy reach a dynamic equilibrium. Therefore, the high-emissivity state of the device remains stable under the continuous application of a negative voltage. When reverse energized, hydrogen atoms lose electrons at the interface between the Pd and electrolyte layers and are oxidized to protons, which then return to the WO 3 layer under the electric field to complete the cycle. A schematic diagram of the operating principle of the device is shown in Figure 3c. As the variable emissivity layer (Mg 3 Ni) is not under the action of the electric field, the formation and decomposition of metal hydrides are not affected by the electric field. Thus, we propose the concept of dynamic equilibrium. With the applied voltage, the device can always maintain the stability of the high-emissivity state. And in the open-circuit state, the

www.advmatinterfaces.de
high-emissivity state of the device can only remain stable for 3 h due to the continuous decomposition of the Mg 3 NiH 6 .

Cyclability and Failure Mechanism
As the number of cycles increased, the initial current of the device tended to decrease (Figure 4a). This behavior was due to the presence of a proton deficit caused by the spontaneous decomposition of a small amount of metal hydride followed by the spillover of a small amount of hydrogen in the form of a gas. [35] With continued cycling, the magnitude of the apparent temperature change of the device decreased (Figure 4b), indicating that the performance of the device deteriorates. Device failure is considered to occur when the change in the apparent temperature is less than 50% of the first change. [36] Thus, the device could be effectively cycled seven times before failing. X-ray photoelectron spectroscopy (XPS) analysis of the Mg 3 Ni alloy layers of the device was used to investigate the failure mechanism. The Mg 1s spectrum exhibits peaks at binding energies of 1303.4 and 1305.3 eV, corresponding to Mg and Mg 2+ , respectively. [37,38] Figure 4c shows the XPS results for devices in different states. In the initial state, the alloy layer does not appear oxidized, and the ratio of Mg to Mg 2+ is 3.72. After one cycle, this ratio becomes 2.20, which indicates an increase in the oxidation of the alloy layer. After eight cycles, the Mg to Mg 2+ ratio in alloy layer decreases to 0.42, suggesting greater oxidation of the alloy layer. Thus, oxidation of the device alloy layer becomes more pronounced with cycling.
To further investigate the cause of device failure, we characterized the Mg 3 Ni alloy layer of the device. The surface of the Mg 3 Ni alloy layer in the initial state was flat and unbroken (Figure 4d), whereas wrinkles caused by transverse stress appeared on the alloy layer surface after one cycle (Figure 4e).
After eight cycles, the surface of alloy layer exhibited defects caused by stress accumulation (Figure 4f,g). The main cause of film peeling or breakage was the 32% change in lattice volume that occurred when Mg or Mg 2 Ni completely reacts with hydrogen. [39] Repeated expansion and contraction of the lattice caused stress accumulation, resulting in breakage or peeling of the metal film. The interface between the alloy layer and the BaF 2 substrate was abrupt, and the transverse stress from alloy lattice deformation was confined to a small area, which could easily lead to interface fracture. Magnification of a relatively intact area of the broken alloy layer surface showed the presence of some smaller defects. Such defects could provide channels for water and oxygen to enter, which explains why the oxidation of the alloy layer became more pronounced as cycling progressed. Therefore, the defects caused by stress in the alloy layer are the main cause of device failure.

Device Multifunctionalization
Flexible infrared electrochromic devices can be constructed by replacing the infrared-transparent substrate and counter electrode. A polypropylene film, which has good flexibility and high infrared transmittance, was used to replace the BaF 2 lens as an infrared-transparent substrate in the flexible infrared electrochromic device. In addition, flexible polyethylene terephthalate/ITO was used as the counter electrode in the device. The device could be attached to the outer wall of a beaker, as shown in Figure 5a. By applying a voltage of −2.6 V, the device gradually changed from a lowemissivity state to a high-emissivity state (Figure 5b) with an apparent temperature change of 5.1 °C in the central region ( Figure 5c); this process is shown in Video S3 in the Supporting Information.

www.advmatinterfaces.de
The infrared electrochromic device could be modified for visible light camouflage while retaining its variable infrared emissivity performance. Colored devices were produced by adding a Cr 2 O 3 layer of various thicknesses between the BaF 2 substrate and the Mg 3 Ni alloy layer, its structure is shown in Figure 6a. The device with a 30 nm Cr 2 O 3 layer appears earthy yellow and dark green in the low-and high-emissivity states (Figure 6b). The reflectivity curves of this device in the visible region are shown in Figure 6c Thus, a thicker Cr 2 O 3 layer in the device results in a smaller emissivity change in the 3-5 µm region, as the transmittance in this region gradually decreases as the thickness of the Cr 2 O 3 layer increases. The device also becomes darker in color and the change in reflectance in the visible region decreases because of the reduced reflectance of the initial state of the device in the 492-577 nm region and the lower infrared transmittance. Nevertheless, the visible reflectance spectra of the devices in the two states differ significantly, demonstrating that the emissivity modulation process can also induce color changes.

Conclusion
A new infrared electrochromic device based on a hydrogeninduced metal-insulator phase transition from metallic Mg x Ni alloy to dielectric hydrides of Mg x Ni was developed. Unlike traditional ion injection/extraction devices, this device employs a new material system that does not require a top conductive electrode, which can effectively improve the emissivity change while reducing costs. Using applied voltages of ±2.6 V, the infrared emissivity of the device was reversibly modulated, with emissivity changes of 0.48 and 0.43 in the 3-5 and 7.5-14 µm regions, respectively. The high-emissivity state of the device was stably maintained for 3 h under open-circuit conditions. The main cause of device failure was defects caused by stress accumulation and subsequent oxidation of the Mg x Ni alloy. The poor cycling stability of the device could potentially be addressed by adding a buffer or transitional layer. In addition, the addition of a Cr 2 O 3 layer allowed the device to also exhibit visible color changes, mainly yellow or green, making it suitable for visible light camouflage of military vehicles. The infrared electrochromic device described herein is advantageous owing to its simple structure, large emissivity change, compatibility, and flexible nature. Consequently, this device has good application prospects in a variety of fields, including adaptive thermal camouflage, smart thermal management, and dynamic information displays. The use of thin metal films for dynamic infrared emissivity modulation provides a new approach for constructing infrared electrochromic devices, and the potential of other metals or alloys to provide superior performance should be explored.

Experimental Section
Preparation of Thin Metal Films: ITO (Huanan Xiangcheng Technology Co., Ltd., China) measuring 30 mm × 30 mm × 1 mm was used as the substrate for preparing WO 3 films. First, the ITO substrate was cleaned with deionized water and acetone for 15 min each and then heated at 100 °C for 1 h in a vacuum drying oven. The WO 3 film was deposited on the ITO substrate using an MSP-300BTI high-vacuum four-target magnetron sputtering system (Beijing Genesis Weiner Technology Co., Ltd., China) with a base vacuum of ≈2 × 10 −4 Pa. Next, the Mg x Ni/Pd films were deposited on the substrate. For the deposition of Mg x Ni (x = 1.47, 3, 3.67, and 4.48), a pressure of 0.8 Pa was used with a deposition power of 50, 65, 75, or 100 W for Mg and 30 W for Ni with 3 in. Mg and Ni targets (Beijing Yipin Material Technology Co., Ltd., China). To deposit the Pd film, a pressure of 1.2 Pa was used with a deposition power of 35 W and a 3 in. Pd target (Zhongnuo Advanced Material Technology Co., Ltd., China).
Device Assembly: WO 3 was injected with protons using an electrochemical method. The counter electrode was a 304 stainless steel sheet and the electrolyte was a 0.1 mol L −1 sulfuric acid solution. A voltage of −2 V was applied to the ITO/WO 3 electrode to inject a charge of 0.1 C into the WO 3 film. PEI was dripped onto the WO 3 film to completely cover the surface. Then, the Mg x Ni/Pd films were placed on the PEI layer. Finally, the edges were encapsulated with tape to complete device assembly.
Characterization: The atomic ratios in the alloys were determined by inductively coupled plasma mass spectrometry (ICP-MS 2000, Jiangsu Skyray Instruments Co., China). Infrared spectra were collected using a VERTEX70 Fourier transform infrared spectrometer (Bruker Technology Co., Ltd., China) with a sampling time of 16 s. Scanning electron microscopy (SEM) images were obtained using a Hitachi SU8010 fieldemission scanning electron microscope (Hitachi Co., Ltd., Japan), and transmission electron microscopy (TEM) images were obtained using a Japan Electron Optics Laboratory (JEOL) JEM 2100F transmission electron microscope (JEOL Ltd., China). XPS data were collected using an Escalab 250xi instrument (Thermo Fisher Technology Co., Ltd., China). Infrared images were recorded using an Flir Systems, lnc (FLIR)T1050sc infrared thermal imager (Foresight Infrared Photoelectric Technology Co., Ltd., China) with a shooting distance of 1 m, reflection temperature of 20 °C, and relative humidity of 50%. The infrared images were captured from the videos, and the apparent temperatures were determined from the videos using the FLIR Tools and FLIR Altair software packages. Electrochemical tests were performed using a Parstat 4000 electrochemical workstation (AMETEK Co., Ltd., USA). For cyclic voltammetry measurements, the potential scan range was −6 to 6 V and the scan rate was 0.2 mV s −1 . In the cycling experiments, each cycle involved applying a voltage of −2.6 V to the device for 180 s and then applying a voltage of 2.6 V for 200 s.

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