High-performance thermochromic VO2-based coatings with a low transition temperature deposited on glass by a scalable technique

We report on high-performance thermochromic ZrO2/V0.982W0.018O2/ZrO2 coatings with a low transition temperature prepared on glass by a low-temperature scalable deposition technique. The V0.982W0.018O2 layers were deposited by a controlled high-power impulse magnetron sputtering of V target, combined with a simultaneous pulsed DC magnetron sputtering of W target to reduce the transition temperature to 20–21 °C, at a low substrate surface temperature of 330 °C in an argon–oxygen gas mixture. ZrO2 antireflection layers both below and above the thermochromic V0.982W0.018O2 layers were deposited at a low substrate temperature (< 100 °C). A coating design utilizing a second-order interference in the ZrO2 layers was applied to increase both the luminous transmittance (Tlum) and the modulation of the solar transmittance (ΔTsol). The ZrO2/V0.982W0.018O2/ZrO2 coatings exhibit Tlum up to 60% at ΔTsol close to 6% for a V0.982W0.018O2 thickness of 45 nm, and Tlum up to 50% at ΔTsol above 10% for a V0.982W0.018O2 thickness of 69 nm.

www.nature.com/scientificreports/ be mentioned that HiPIMS techniques are compatible with existing magnetron sputtering systems utilized in industrial deposition devices 26,27 .
In this work, we report on high-performance three-layer thermochromic ZrO 2 /V 0.982 W 0.018 O 2 /ZrO 2 coatings with a low transition temperature (20-21 °C) prepared on soda-lime glass using a low-temperature (330 °C) magnetron sputter deposition without any substrate bias voltage. We present basic principles of this new solution for a low-temperature scalable deposition of high-performance durable thermochromic VO 2 -based coatings for smart-window applications.

Methods
Coating preparation and elemental composition. The V 0.982 W 0.018 O 2 layers were deposited by controlled HiPIMS of V target, combined with a simultaneous pulsed DC magnetron sputtering of W target, at a low substrate surface temperature T s = 330 °C and without any substrate bias voltage in an argon-oxygen gas mixture. The argon flow rate was 60 sccm corresponding to an argon partial pressure of 1 Pa, while the oxygen flow rate (Φ ox ) was not fixed but pulsing between 0.6 and 1.5 sccm (see Fig. 1), and the duration of the Φ ox pulses (injecting oxygen in front of the sputtered V target toward the substrate 24 ) was determined during the deposition by a programmable logic controller 29 using a pre-selected critical value of the average discharge current on V target in a period of the power supply ( − I d ) cr = 0.43 A. The basic principle of the pulsed oxygen flow control is illustrated in Fig. 1, which shows the time evolution of the magnetron voltage (U d (t)) and the target current density (J t (t)), averaged over the total target area, for both targets at the minimum and maximum value of the oxygen partial pressure in the vacuum chamber corresponding to the minimum and maximum − I d , respectively, during the deposition. Here, it should be mentioned that the effective pulsed oxygen flow control makes it possible to utilize two benefits of the reactive HiPIMS deposition 23,24 . The first benefit is highly ionized fluxes of particles with many V + and V 2+ ions onto the substrate and enhanced energies (up to 50 eV relative to ground potential) of the ions bombarding the growing films, allowing us to achieve the VO 2 crystallinity at a low T s and without any substrate bias voltage. The second benefit is a very high degree of dissociation of the O 2 molecules injected into the high-density plasma in front of the V target, allowing us to achieve the required VO 2 stoichiometry at a Figure 1. Waveforms of the magnetron voltage (U d ) and the target current density (J t ) for preset depositionaveraged target power densities of 12.9 W cm −2 and 33 mW cm −2 for V and W target, respectively, during a deposition of the V 0.982 W 0.018 O 2 films (the J t values for W target are magnified 1,000 times). Time evolution of the average discharge current on V target in a period of the power supply ( − I d ) during the deposition is shown in the inset. A pre-selected critical value ( − I d ) cr = 0.43 A determining the switch between the oxygen flow rates Φ ox = 0.6 sccm and Φ ox = 1.5 sccm is marked by dots. Reprinted from the work 28  www.nature.com/scientificreports/ low compound fraction in the target surface layer. This is of key importance for reduced arcing, increased sputtering of V atoms, and low production of O − ions at the target 29 . The depositions were performed in an ultra-high vacuum multi-magnetron sputter device (ATC 2200-V AJA International Inc.) using two unbalanced magnetrons with planar V and W targets (99.9% purity, diameter of 50 mm and thickness of 6 mm in both cases). The magnetron with a V target was driven by a high-power pulsed DC power supply (TruPlasma Highpulse 4002 TRUMPF Huettinger) 24 . The voltage pulse duration was 50 µs at a repetition frequency of 200 Hz (duty cycle of 1%) and the deposition-averaged target power density was 12.9 W cm −2 . The magnetron with a W target was driven by a pulsed DC power supply (IAP-1010 EN Technologies Inc.). The voltage pulse duration was 16 µs at a repetition frequency of 5 kHz (duty cycle of 8%) and the deposition-averaged target power density was 33 mW cm −2 . Under these conditions, the W content in the metal sublattice of V 1−x W x O 2 , as measured on a dedicated 285 nm thick film in a scanning electron microscope (SU-70, Hitachi) using wave-dispersive spectroscopy (Magnaray, Thermo Scientific), was 1.8 ± 0.6 at.% (i.e., x = 0.018).
Both bottom and top ZrO 2 antireflection layers were deposited by reactive mid-frequency AC magnetron sputtering without ohmic heating (T s < 100 °C) and without any substrate bias voltage in an argon-oxygen gas mixture. The argon partial pressure was 1 Pa and the oxygen partial pressure was 0.35 Pa (oxide mode). The depositions were performed using two strongly unbalanced magnetrons with planar Zr targets (99.9% purity, diameter of 100 mm and thickness of 6 mm) driven by a mid-frequency AC power supply (TruPlasma MF 3010, TRUMPF Huettinger) 30 . The oscillation frequency was close to 85 kHz and the deposition-averaged target power density was 15.5 W cm −2 .
The thickness of individual layers was measured by spectroscopic ellipsometry using the J. A. Woollam Co. Inc. VASE instrument 31 .
The presented deposition technique for preparation of the thermochromic ZrO 2 /V 0.982 W 0.018 O 2 /ZrO 2 coatings is, just like the deposition of low-emissivity coatings, compatible with the existing magnetron sputtering systems in glass production lines.
Coating structure and properties. For structural investigation of the films, X-ray diffraction (XRD) measurements were carried out using a PANalytical X´Pert PRO diffractometer working with a CuKα (40 kV, 40 mA) radiation at a glancing incidence of 1°. The average size of coherently diffracting regions of the VO 2 (R)/ VO 2 (M1) phase was estimated from the full width at half maximum of the main VO 2 (R)/VO 2 (M1) diffraction peak, corrected for instrumental broadening, using the Scherrer's equation.
The surface morphology of the films was determined by atomic force microscopy (AFM) using a SmartSPM Microscope (AIST-NT) with a diamond tip (nominal radius below 10 nm) in a semicontact mode. The rootmean-square roughness of the surface, R rms , was computed from a randomly selected square area of 1 × 1 μm 2 . The AFM images were processed by Gwyddion 2.41 software 32 , and an implemented "watershed" method was used for grain analysis. The grains identified were approximated by an equivalent disc diameter with the same projected area as the grain.
The hardness of VO 2 (without the ZrO 2 overlayer) was measured using a Hysitron TI 950 triboindenter with a cube corner tip at a maximum load of 100 μN.
The normal-incidence coating transmittance was measured by spectrophotometry using the Agilent CARY 7000 instrument equipped with an in-house made heat/cool cell. Spectroscopic measurements were performed in the wavelength range λ = 300 to 2,500 nm at the temperatures T ms = − 5 °C (semiconducting state below T tr ) and T mm = 60 °C (metallic state above T tr ). Hysteresis curves were measured at λ = 2,500 nm in the temperature range T m = − 10 to 60 °C. The luminous transmittance (T lum ) and the solar transmittance (T sol ) are defined as follows where φ lum is the luminous sensitivity of the human eye and φ sol is the sea-level solar irradiance spectrum 33 at an air mass of 1.5. The modulation of the luminous transmittance (ΔT lum ) and of the solar transmittance (ΔT sol ) are defined as Using relation (2) it can be written

Results and discussion
Design and transition temperature of Zro 2 /V 0.982 W 0.018 o 2 /Zro 2 coatings. The three-layer structure of ZrO 2 /V 0.982 W 0.018 O 2 /ZrO 2 coatings, formed by an active layer in the middle and two antireflection (AR) layers, is shown in Fig. 2. Let us emphasize the combination of properties which makes ZrO 2 a proper candidate for the AR-layers. First, ZrO 2 has a refractive index (n) close to the required geometric mean of refractive indices of V 0.982 W 0.018 O 2 and glass (bottom AR-layer) or V 0.982 W 0.018 O 2 and air (top AR-layer). Second, ZrO 2 has almost zero extinction coefficient (k) for visible and infrared wavelengths (λ), allowing one to utilize higher-order ARlayers without concessions in terms of absorption. Third, crystalline structure of the bottom ZrO 2 layer can be achieved even at a low deposition temperature, which in turn improves the V 0.982 W 0.018 O 2 crystallinity and the process reproducibility. Fourth, ZrO 2 is a hard (for an oxide) and stable material, which allows the top AR-layer to provide a mechanical protection and environmental stability for the active V 0.982 W 0.018 O 2 layer. The hardness of ZrO 2 prepared by the present technique is 15-17 GPa 30 , compared to the hardness of VO 2 of only 12 GPa. Note that ZrO 2 layers are being increasingly applied in architectural glass as a protective overcoat for advanced low-emissivity stacks 34 . These properties cannot be matched by many other potential or occasionally used ARlayer materials due to their, e.g., non-zero k (Cr 2 O 3 ), lower hardness (SiO 2 , Ta 2 O 5 ), high deposition temperature of the hardest phase (α-Al 2 O 3 ), too low n for the bottom AR-layer (SiO 2 ) or usable but too high n (rutile TiO 2 ). We examined the effect of smoothly varied h b and h t in our recent work 31 and thereby identified the optimum value h b = h t = 180 nm leading to a second-order interference maximum of T lum (consistently with the optimization of h t in the work 35 ). This choice constitutes a crucial part of the efforts to maximize T lum and ΔT sol (at a given h) in parallel: while the frequently used first-order AR-layers (λ/4-layers; see e.g. the work 36 for a first-order ZrO 2 AR-layer) lead to a high transmittance modulation only in the far infrared (where it is weighted by weak solar irradiance when calculating ΔT sol ), second-order AR-layers (3λ/4-layers) lead to a high transmittance modulation mainly in the near infrared (where it is weighted by much higher solar irradiance; see below for a graphical example). Furthermore, we use two different h values of 45 nm or 48 nm (leading to higher T lum ) and 69 nm (leading to higher ΔT sol ) in order to demonstrate the corresponding tradeoff. Here, it should be mentioned that the thickness of the V 0.982 W 0.018 O 2 layer deposited onto amorphous soda-lime glass was 48 nm while it was 45 nm for the same layer deposited using the same discharge conditions (Fig. 1) Table 1). The T tr value is reproducible (almost the same for two different h values) and in agreement with the requirement for smart-window applications 14 . It is very important that using the present deposition technique, we did not experience any tradeoff (indicated in the literature 19,37,38 ) between lowering T tr by W doping and optimizing the other optical properties: the differences in the V(W)O 2 optical constants at λ = 550 nm were within the measurement error and reproducibility noise and did not exhibit any systematic dependence on the W content. The present W content of 1.8 ± 0.6 at.% and the transition temperature T tr = 57 °C achieved for undoped VO 2 prepared by the same technique 23 (Fig. 2). The transition temperatures (T tr ) are also given. Adapted from the work. 28 Table 1. Thermochromic properties of different configurations of the VO 2 -based coatings on 1 mm thick glass substrates. Here, h is the thickness of the V 0.982 W 0.018 O 2 layer, and h b and h t are thicknesses of the bottom and top ZrO 2 layers, respectively. Adapted from the work 28 .  Fig. 5. The figure constitutes an independent confirmation of the fact that while the present deposition technique allows a lowtemperature crystallization of VO 2 -based layers on amorphous glass, their crystallinity on crystalline ZrO 2 is even better. Most importantly, the grains identified by the "watershed" method make up 80% of the projected surface area for the V 0.982 W 0.018 O 2 layer (R rms = 1.1 nm) deposited onto the bare soda-lime glass (Fig. 5a), while they make up 94% of the projected surface area for the V 0.982 W 0.018 O 2 layer (R rms = 1.2 nm) deposited onto the crystalline ZrO 2 AR-layer (Fig. 5b). Furthermore, the latter V 0.982 W 0.018 O 2 layer exhibits also a narrower distribution of the horizontal grain sizes (Fig. 5c). thermochromic properties of Zro 2 /V 0.982 W 0.018 o 2 /Zro 2 coatings. Figure 6 shows in detail the aforementioned role of second-order AR-layers in optimizing T lum and ΔT sol , given by Eqs. (1) and (5), respectively, in parallel. On the one hand, Fig. 6a,b show that T lum depends only on a narrow range of wavelengths: the transmittance T(λ, T m ) is weighted by a narrow function φ lum (λ). There is an easily explainable increase of T(λ, T m ) in the corresponding narrow visible λ range, resulting from using only the bottom AR-layer, only the top AR-layer (stronger increase than the previous one) and both AR-layers (the strongest increase). This phenomenon is almost independent of T m , which means that the low ΔT lum (Table 1) is almost independent of the coating design. Furthermore, Fig. 6a,b confirm that owing to the absorption in V 0.982 W 0.018 O 2 , T(λ, T m ) is generally higher at h = 45 nm and 48 nm than at h = 69 nm.

h (nm) h b (nm) h t (nm) T tr (°C) T lum (T ms ) (%) T lum (T mm ) (%) ΔT lum (%) T sol (T ms ) (%) T sol (T mm ) (%) ΔT sol (%)
On the other hand, Fig. 6c shows that the dependence of ΔT sol on the coating configuration is much more difficult to explain, because the transmittance modulation ΔT(λ) is weighted by a wide and complicated function φ sol (λ) and there is no coating configuration leading to the highest ΔT(λ) in the whole λ range shown. Indeed, while ΔT(λ) in the far infrared above ≈1,600 nm (weighted by relatively low φ sol ) is actually the highest without any AR-layer, ΔT(λ) in the near infrared below ≈1,600 nm (weighted by relatively high φ sol ) is the highest when using both second-order AR-layers or at least the top one. The reason is that the second-order interference maxima on both AR-layers in the visible are followed by lower-order interference minima and maxima in the www.nature.com/scientificreports/ infrared, and that the overall improvement of the near infrared transmittance by this interference is more significant below than above T tr . The fact that this kind of effect cannot be achieved by thinner first-order AR-layers www.nature.com/scientificreports/ is discussed in more detail in our recent work 31 . Furthermore, Fig. 6c confirms that ΔT(λ) is generally higher at h = 69 nm than at h = 45 nm or 48 nm. The transmittance-based integral quantities, given by Eqs. (1)- (5), are summarized in Fig. 7 and Table 1. In agreement with the discussion of the transmittance in itself (Fig. 6), it can be seen that the transition from (1)   www.nature.com/scientificreports/ leads to a gradual improvement of the optical performance (average T lum and ΔT sol ). The performance of the best coating configuration (h b = h t = 180 nm) is characterized by T lum = 59.4% and ΔT sol = 5.5% (at h = 45 nm) and by T lum = 48.0% and ΔT sol = 10.4% (at h = 69 nm), in both cases accompanied by low ΔT lum and the aforementioned T tr = 20-21 °C. It is possible to state that our results are close to the requirements (see the introductory part and the gray area in Fig. 7) for smart-window applications.
In addition to comparing the thermochromic properties of the present coatings with the industrial requirements, it is worth comparing them with the properties of coatings reported in the literature (Table 2). We focus on coatings on glass substrates and on plastic foils 42,43 , which can be pasted on the glass, with an at least somewhat lowered T tr ≤ 40 °C.
The ZrO 2 /V 0.988 W 0.012 O 2 /ZrO 2 coating 31 was deposited using the same method as in the present work. The V 0.958 Tb 0.031 W 0.011 O 2 coating 17 was fabricated on a glass substrate from Tb-and W-codoped VO 2 nanopowders  Table 2. Comparison between this work and previously reported studies on T lum and ΔT sol of VO 2 -based coatings with a transition temperature T tr ≤ 40 °C prepared on glass substrates or polyethylene terephthalate (PET) foils. (T s ) max is the maximum substrate temperature during the preparation of the coatings and h is the thickness of the active VO 2 -based layer. Here, ACMS, RFMS and DCMS denote the AC, RF and DC magnetron sputtering, respectively.  2 43 nanoparticles dispersed in polyurethane. These nanoparticles were produced by complex hydrothermal reactions. Here, it should be mentioned that the F-doping and W-doping of these nanoparticles resulted in the required reduction in the transition temperature, but it led also to a decrease in the modulation of the solar transmittance. The ΔT sol value decreased from 13.1% for the coating with pure VO 2 nanoparticles to 10.7% (see Table 2) for the coating with 2.93 at.% F-doped VO 2 nanoparticles 42 . In case of the coating with 1 at.% W-doped VO 2 nanoparticles 43 , the ΔT sol value decreased to 12.7% (see Table 2) from 22.3% for the coating with pure VO 2 nanoparticles. Note that the transition temperature T tr = 36 °C of the highperformance thermochromic coating with the V 0.99 W 0.01 O 2 nanoparticles (see Table 2) was determined as a mean value from the temperature of 46 °C, related to an endothermic peak, and 26 °C, related to an exothermic peak, detected using differential scanning calorimetry during the heating-up and cooling-down period, respectively.

(T s ) max (°C) T tr (°C) T lum (T ms ) (%) T lum (T mm ) (%) ΔT sol (%) h (nm) Preparation method
As can be seen in Table 2, we achieved an excellent combination of the required characteristics: the lowest maximum glass temperature during the preparation of the coatings (T s ) max = 330 °C, an appropriate transition temperature T tr = 20-21 °C, and T lum up to 60% at ΔT sol close to 6% or T lum up to 50% at ΔT sol above 10%. These optical properties are comparable with those achieved for the thermochromic VO 2 -based coatings which were prepared using long and too complicated chemical processes on flexible PET foils 42,43 at a very low (T s ) max ≤ 100 °C, but with too high transition temperatures T tr = 35 °C and 36 °C, respectively. conclusion High-performance thermochromic ZrO 2 /V 0.982 W 0.018 O 2 /ZrO 2 coatings with a low transition temperature were prepared on soda-lime glass by a low-temperature scalable deposition technique which is compatible with the existing magnetron sputtering systems in glass production lines. The V 0.982 W 0.018 O 2 layers were deposited by controlled HiPIMS of V target, combined with a simultaneous pulsed DC magnetron sputtering of W target (doping of VO 2 by W to reduce the transition temperature to T tr = 20-21 °C without any degradation of thermochromic properties), at a low substrate surface temperature T s = 330 °C in an argon-oxygen gas mixture. The effective pulsed oxygen flow control of the reactive HiPIMS deposition makes it possible to utilize the enhanced energies of the ions bombarding the growing V 0.982 W 0.018 O 2 layers for the support of the crystallization of the thermochromic VO 2 phase in them at the low T s = 330 °C and without any substrate bias voltage. Our design of the three-layer VO 2 -based coatings utilizes the second-order interference in two antireflection ZrO 2 layers to increase both the luminous transmittance and the modulation of the solar transmittance. The ZrO 2 /V 0.982 W 0.018 O 2 /ZrO 2 coatings exhibit the optical properties which are relatively close to the requirements (T lum > 60% and ΔT sol > 10%) for smart-window applications. For applications in large-scale systems, it is important that the presented controlled deposition of the active VO 2 -based layers can be performed also at prolonged duty cycles (up to 5%). This results in up to 5 times lower target power density in a pulse at the same deposition-averaged target power density (approximately 13 W cm −2 in this work). Moreover, the deposition rate of these layers is higher.

Data availability
All experimental deposition conditions and characterization procedures, methods and data are provided in the text. Any clarifications will be available by contacting the corresponding author.