Lowering insulator-to-metal transition temperature of vanadium dioxide thin films via co-sputtering, furnace oxidation, and thermal annealing Open Access

Vanadium dioxide (VO₂) is a thermochromic material with a reversible insulator-to-metal phase transition (IMT) around its transition temperature of 68 °C.^1–3 The phase change between monoclinic insulator and rutile metal is not instantaneous and depends on the fabrication method of the vanadium dioxide as well as the heating and cooling rate.^4–9 This thermochromic property allows vanadium dioxide to be utilized in many radiative heat transfer applications where passive thermal control is desired. Thermochromic smart windows regulate the incoming solar radiation in order to minimize radiative heat transfer when temperatures are high while allowing this heat to transmit through the window when temperatures are low, allowing for temperature control of an indoor space without additional power requirements.^7,10–13 Modulating the infrared transmittance through a thermochromically switchable filter can also be used in other applications, such as thermal camouflaging or energy dissipation techniques.^14,15 Another popular use of vanadium dioxide is in infrared emitters, where the emissivity of the device is high at high temperatures, promoting heat rejection away from the device and into the space, but at low temperatures the emissivity is low to prevent heat loss.¹⁶ These devices are accomplished by incorporating vanadium dioxide into Fabry–Pérot metafilm designs^17–21 to take advantage of resonance effects of the metallic vanadium dioxide in the design, or using vanadium dioxide-based metamaterials^22–27 where nanostructures are used to excite different effects depending on the phase of the vanadium dioxide. The deposition of vanadium dioxide on flexible substrates has also been a topic of recent interest and allows for vanadium dioxide-based devices to find even broader application.^28,29

However, one challenge of these devices is the relatively high transition temperature of vanadium dioxide. For many terrestrial thermal control applications, the desired transition temperature should be at or near room temperature to promote a comfortable thermal environment for humans. For extraterrestrial spacecraft applications, the desired transition temperature may be even lower depending on the surrounding temperature of a particular space mission. Additionally, the thermal hysteresis of the phase transition should be narrow, such that the device can quickly switch between its high and low emittance/transmittance states. While the transition temperature and phase hysteresis of vanadium dioxide are affected by other factors such as fabrication method, and thin-film layer parameters such as thickness, crystal structure, mechanical stress or deformation, and non-stoichiometry, doping the vanadium dioxide with other elements has been a popular and effective method in lowering the transition temperature.

Substitutional dopant elements, which could alter the crystalline and electronic structures of VO₂, can be used to lower the transition temperature.^30–34 Tungsten dopants can reduce the transition temperature by 20–30 °C per at. % of tungsten,^30,33,34 indicating that around 2 at. % of tungsten would be effective in lowering the transition temperature from 68 °C to room temperature. There are quite a few fabrication methods that can be used to introduce tungsten dopants into vanadium dioxide thin films, among them physical vapor deposition and wet chemistry methods are two popular fabrication methods. Physical vapor deposition involves a solid vanadium source material being converted to a vapor, transferred to a substrate, and then condensed back as a thin film. The thin film on the substrate could be a precursor vanadium film to be oxidized with a later process,⁴ or through reactive methods, could be directly fabricated as vanadium dioxide.^35,36 Wet chemistry methods utilize a precursor mixture of varying powders and acids to develop a vanadium dioxide thin film.^37,38 The doping material is introduced to the precursor film or solution with the molar ratio between vanadium and dopant atoms carefully controlled while additional annealing for further diffusion of the dopant atoms is usually required.³² However, most of fabrication methods in fabricating high-quality tungsten-doped VO₂ thin films suffer from expensive operations, slow deposition rate, and highly selective substrate materials.

This study aims to develop a scalable and cost-effective process for the fabrication of tungsten-doped vanadium dioxide thin films with lowered transition temperature via co-sputtering, furnace oxidation, and thermal annealing. Tungsten-doped vanadium dioxide thin film samples with different doping levels up to 4.1 at. % are fabricated, and materials characterizations are carried out to confirm the tungsten doping concentrations. Parametric studies on temperature and time are performed to determine optimal conditions in furnace oxidation and thermal annealing. The insulator-to-metal phase transition behaviors of fully annealed samples are comprehensively studied with temperature-dependent infrared spectroscopy in a wide temperature range from −60 to 100 °C. The dependence of tungsten doping level on the phase transition temperature reduction and thermal hysteresis is quantitatively discussed.

The co-sputtered tungsten-vanadium thin films were first oxidized in a tube furnace (Thermco Minibrute) with carefully controlled gas flow, temperature, and time and then thermally annealed to fully diffuse tungsten atoms. Following our previous successful furnace oxidation process,^4,5 the N₂ and O₂ gas flow rates were, respectively, set to 7 SLPM (standard liter per minute) and 0.5 SLPM, while the oxidation temperature was fixed at 300 °C. The oxidation time is critical, as the VO₂ thin film could be under-oxidized with inadequate time, or over-oxidized with excessive time. For 30 nm-thick vanadium thin films deposited on a silicon substrate by e-beam evaporation, it requires 3 h to fully oxidize the film into VO₂. However, sputtered vanadium film of the same thickness could have different grain size, porosity, and surface roughness than that prepared by e-beam evaporation. Therefore, an oxidation time study needs to be conducted for determining the optimal oxidation time for the sputtered vanadium thin films. The thermal annealing was performed in the same tube furnace with only N₂ gas flowing at 7 SLPM and O₂ gas flow shut off to avoid over-oxidizing the VO₂ film to other vanadium oxides like V₂O₅. A comprehensive thermal annealing study on the temperature and time was conducted to determine the optimal condition to fully anneal the tungsten-doped VO₂ thin films at different doping levels.

In order to study the insulator-to-metal phase transition behaviors of undoped and doped VO₂ thin films on undoped silicon wafer substrate, temperature-dependent infrared transmittance was characterized using a Fourier-transform infrared spectrometer (Thermo Fisher, Nicolet iS50 FTIR) in the spectral range from 2 to 25 μm in wavelength at a resolution of 8 cm⁻¹ with each spectrum averaged over 32 scans. To facilitate the oxidation and thermal annealing studies, a home-built ambient heating stage with an 8 mm clear aperture, whose temperature can be precisely controlled from 25 to 120 °C with a PID temperature controller and polyimide thin-film heater, was used to mount the sample for the temperature-dependent FTIR transmittance measurement in ambient. To characterize fully annealed tungsten-doped VO₂ thin films whose phase transition could go below room temperature, an optical cryostat (Janis VPF-800-FTIR) custom-built to be mounted inside the FTIR sample compartment with optical pass through ZnSe windows was used to vary the sample temperatures in a wider range from −60 to 100 °C. At each sample temperature, 5 min was given to ensure the steady state was achieved before the spectrum was taken. The temperature-dependent transmittance of WVO₂ thin film was measured at 5 °C intervals in both heating and cooling processes to study the thermal hysteresis behavior during the phase transition.

X-ray photoelectron spectroscopy (XPS, Kratos Axis Supra⁺) was conducted with an Al kα 1486.6 eV x-ray source in ultrahigh vacuum. V2p elemental scans from binding energies 510.0–540.0 eV were performed to identify the oxidation state of vanadium, and W4f elemental scans from binding energies 30.0–50.0 eV were carried out to measure the atomic concentration of tungsten in the fully annealed W_xV_1−xO₂ samples. Raw data were analyzed and fitted with CasaXPS software, while Shirley algorithm was utilized to draw background for both V2p and W4f elemental scans. A Lorentzian Asymmetric curve was used to fit the V2p, V3p, and O1s peaks, while Gaussian Lorentzian peaks were used to fit the peaks of W4f.

It is not trivial to obtain high-quality VO₂ thin films by oxidizing vanadium thin films. In our previous work,^{4, 5} optimal oxidation conditions such as temperature, gas flow rates, and time have been identified for fully oxidizing 30 nm vanadium thin film on silicon wafers prepared from e-beam evaporation. Among them, the film quality is sensitive to the oxidation time as it could be under or over oxidized with insufficient or excessive time. As the vanadium and doped vanadium thin films were prepared by the sputtering process, it is necessary to check the oxidation time again to achieve full oxidization of vanadium films into stoichiometric VO₂. To this end, 30 nm vanadium films were deposited onto undoped silicon wafers, respectively, with e-beam evaporation and sputtering, which were oxidized at 300 °C in the furnace for different time periods.

As shown in Fig. 2(a), after 3 h furnace oxidation, the evaporated vanadium film exhibits a full oxidation state of VO₂ with maximum infrared transmission around 55% at 25 °C in its insulating phase and lowest infrared transmission around 15% at 100 °C in its metallic phase, which is consistent with our previous work.⁴ However, the sputtered 30 nm vanadium thin film shows much less infrared transmission particularly about 38% at 25 °C in its insulating phase. This clearly indicates that the vanadium film is not completely oxidized, and more oxidation time is required. With 4 h oxidation, as shown in Fig. 2(b), the sputtered vanadium thin film achieves the same infrared transmission around 55% at 25 °C as expected for the VO₂ in insulating phase. Moreover, the oxidized film also exhibits a similar infrared transmission spectrum at 100 °C for the VO₂ in metallic phase as the fully oxidized vanadium from e-beam evaporation, regardless of 3%–5% difference, which might be due to differences in grain sizes and porosity from the different deposition methods of vanadium. Note that the high transmission peaks around 10 μm wavelength observed in the metallic phase stems from the over oxidation at the film surface due to high oxygen concentration. It is reasonably assumed that slight doping of tungsten (less than 5 at. %) would not significantly affect the oxidation time. Therefore, in the following studies, all the 30 nm doped vanadium films were oxidized at 300 °C for 4 h into WVO₂ thin films.

In order to ensure that tungsten atoms are completely diffused and bonded with VO₂ to lower its IMT temperature, thermal annealing was carried out for fully oxidized WVO₂ thin films. Comprehensive studies were performed on the thermal annealing temperature from 400 to 650 °C and annealing time from 1 to 2 h. Infrared transmittance above room temperature upon heating (25 °C up to 100 °C) and cooling (100 °C down to 25 °C) was measured with the FTIR spectrometer along with the ambient heating stage after each WVO₂ film was thermally annealed.

As shown in Fig. 3, the spectral transmittance at λ = 8 μm wavelength at various sample temperature was selected to illustrate the phase transition behavior upon heating and cooling for the W_0.013V_0.097O₂ thin film after different annealing conditions. With 400 °C and 1 h annealing, the WVO₂ film with 1.3 at. % exhibits gradual IMT mainly occurring between 40 and 80 °C. However, when the annealing temperature increases to 500 °C, the IMT becomes abrupt with lowered temperature ranges from 25 and 65 °C. No noticeable changes were observed from thermal annealing with an additional hour, which suggests one hour is sufficient for thermally annealing the fabricated WVO₂ thin films.

To get a better idea on the thermal annealing effect on the IMT of the W_0.013V_0.097O₂ thin film, the derivative of the infrared transmittance at λ = 8 μm wavelength with respect to sample temperature, i.e., Δτ_λ/ΔT_s, is plotted in Fig. 4. The W_0.013V_0.097O₂ thin film changes its infrared transmittance at most by −0.75%/K after 400 °C annealing, which increases up to −1.75%/K upon heating or even −2.65%/K upon cooling after 500 °C annealing. The corresponding temperature for the largest rate upon heating drops from 72 to 42 °C for the thermal annealing from 400 to 500 °C. This confirms that thermal annealing at 500 °C is sufficient for the W_0.013V_0.097O₂ thin film to effectively lower its IMT temperature by 30 °C.

Figure 5 shows the spectral transmittance at λ = 8 μm wavelength at temperatures from 25 to 100 °C for the W_0.021V_0.079O₂ thin film thermally annealed at 500, 550, 600, and 650 °C after 1 h, where apparent changes in transmittance with higher annealing temperature are observed. The phase transition upon heating and cooling shifts to lower sample temperatures with the film annealed at a higher temperature. In particular, the difference between 600 and 650 °C becomes marginal, indicating that the thermal annealing is effective at these higher temperatures. Therefore, for the W_1−xV_xO₂ films with tungsten doping levels greater than 2 at. %, thermal annealing at 650 °C is adopted to effectively lower IMT temperatures.

Note that the spectral transmittance at 25 °C is only 18% upon heating and 12% upon cooling, while the difference between heating and cooling indicates that the phase transition could continue at sub-ambient temperatures. This is more clearly observed with the rate of transmittance change as Δτ_λ/ΔT_s shown in Figs. 6(a) and 6(b), respectively, upon heating and cooling for the W_0.021V_0.079O₂ thin film thermally annealed at different temperatures. The rate becomes flat at temperature above 50 °C with a large drop around room temperature after 650 °C annealing, indicating the lowered IMT temperatures toward room temperature and sub-ambient IMT. The W_0.021V_0.079O₂ film has similar behaviors upon cooling but the rate is smaller compared to the heating due to thermal hysteresis. The observation with sub-ambient IMT phase transition with the W_0.021V_0.079O₂ film is expected because the IMT temperature of VO₂ could be lowered by 20–30 °C per at. % tungsten doping as reported by other fabrication methods.^30,33,34

In order to fully characterize the IMT behaviors of W_1−xV_xO₂ films, particularly with tungsten doping level (x) greater than 2 at. %, whose phase transition could go to sub-ambient temperatures, an FTIR cryostat was used to measure the infrared transmittance at a much wider temperature range from −60 to 100 °C. Figure 7 shows the spectral infrared transmittance at steady sample temperatures upon heating and cooling with 5 °C intervals for undoped VO₂ (not annealed) and several fully annealed W_1−xV_xO₂ films with tungsten doping levels of 1.3, 2.6, 3.3, 3.7, and 4.1 at. %, fabricated from the aforementioned co-sputtering, furnace oxidation, and thermal annealing processes. It can be observed that the transmittance in the insulating phase decreases from ∼60% to ∼40% with tungsten doping from 0 to 4.1 at. %. In the metallic phase, the annealed WVO₂ thin films exhibit much flatter infrared transmission spectra around 10% compared to that of undoped VO₂ without annealing. It is believed that this is due to the much lower oxygen trace in the furnace during the annealing process with only N₂ gas flow, which reduces most of the surface over-oxides (i.e., V₂O₅) to VO₂.

To better understand the IMT behaviors of fully annealed WVO₂ films, spectral transmittance at λ = 8 μm is plotted as a function of sample temperature upon both heating and cooling processes for all the samples in Fig. 8(a). Clearly, with more tungsten doping, the insulator-to-metal phase transition successfully shifts to lower temperatures along with the thermal hysteresis between heating and cooling becoming smaller. Figures 8(b) and 8(c) present the derivative of the infrared transmittance at λ = 8 μm wavelength with respect to sample temperature, i.e., Δτ_λ/ΔT_s, for heating and cooling processes, respectively. The IMT derivative valley moves to lower temperatures along with a smaller magnitude with higher tungsten doping. In particular, the IMT midpoint temperatures upon heating, T_m,heat, at which there exists the largest derivative magnitude, are 72.5, 44.5, 20, −2.5, −12.5, and −32.5 °C for unannealed VO₂ and fully annealed W_1−xV_xO₂ films with doping levels of x = 0, 1.3, 2.6, 3.3, 3.7, and 4.1 at. %.

Figure 9(a) reveals the dependence of the IMT midpoint temperature upon heating, T_m,heat, on the tungsten doping level as a linear correlation with a slope of −24.5 °C per at. % in lowering the IMT temperature with tungsten doping. Thermal hysteresis behavior during IMT is also known for VO₂ thin films. By taking the difference in the IMT midpoint temperature between the heating and cooling processes, i.e., ΔT_m = T_m,heat − T_m,cool, the thermal hysteresis is 25, 10, 5, 1, and 0 °C for unannealed VO₂ and fully annealed W_1−xV_xO₂ films with doping levels of x = 0, 1.3, 2.6, 3.3, 3.7, and 4.1 at. %. As shown in Fig. 9(b), a linear fitting suggests that the thermal hysteresis decreases with tungsten doping level at a slope of −5.5 °C per at. % from the W_1−xV_xO₂ films with x from 0 to 4.1 at. % fabricated from the co-sputtering, furnace oxidation, and thermal annealing processes.

Finally, the tungsten doping level or the atomic concentration was confirmed with XPS characterization for the fully annealed W_xV_1−xO₂ sample with x = 3.7 at. % estimated from the sputtering deposition rates of tungsten and vanadium targets according to Eq. (1). Figure 10(a) shows the XPS spectra as a function of binding energy at the film surface, where high W4f_7/2 and W4f_5/2 peaks are observed at 34.9 and 37.1 eV, confirming the successful thermal annealing. However, the measured atomic concentration of tungsten at the film surface is 14.8 at. %, which is much higher than 3.7 at. % from the estimation. Note that this cannot represent the tungsten concentration within the film, as it could be affected by surface termination conditions, and XPS characterization within the film was required for a more accurate measurement of tungsten concentration. Therefore, about 25 nm of the W_0.037V_0.063O₂ film was etched away with 5 keV Ar⁺ ions for 10 min to reach the center of the film. As shown in Fig. 10(b), the W4f peaks become much lower. With another 10 min of etching, Fig. 10(c) presents the XPS spectrum at the bottom of the W_0.037V_0.063O₂ film, which is almost the same as that at the center of the film, indicating excellent uniformity within the film from the co-sputtering and thermal furnace processes. The measured tungsten concentration was 3.88 at. % at the center and 3.62 at. % at the bottom of W_0.037V_0.063O₂ film, or averaged to be 3.75 ± 0.18 at. %, which is consistent with the estimated value of 3.7 at. %. This confirms the accurate estimation of the tungsten doping levels from the sputtering deposition rates.

In summary, we have successfully demonstrated that the phase transition temperature is lowered with tungsten-doped vanadium dioxide thin films fabricated via co-sputtering, furnace oxidation, and thermal annealing processes. By varying the sputtering power for the tungsten target, vanadium thin films were doped with tungsten at different concentrations. For 30 nm sputtered vanadium thin films on undoped silicon wafer, optimal oxidation time was found to be 4 h in an oxygen-rich furnace at 300 °C. Also, lightly tungsten-doped vanadium dioxide thin films with a doping level of 1.3 at. % could be fully annealed at 500 °C, while higher annealing temperature up to 650 °C was required to achieve full annealing for heavily doped ones with an atomic concentration greater than 2 at. %. Comprehensive temperature-dependent infrared transmittance measurements of fully annealed tungsten-doped vanadium dioxide thin films up to 4.1 at. % revealed the phase transition temperature decreases at 24.5 °C per at. % of tungsten doping along with 5.5 °C per at. % shrinkage of thermal hysteresis between heating and cooling. The results in this study would facilitate the research and development of thermochromic coatings for wide energy and thermal control applications.

This work was supported in part by the National Science Foundation (No. CBET-2212342) and the National Aeronautics and Space Administration (No. 80NSSC21K1577).

The authors have no conflicts to disclose.

Vishwa Krishna Rajan and Jeremy Chao contributed equally to this paper.

Vishwa Krishna Rajan: Data curation (equal); Formal analysis (equal); Investigation (equal); Methodology (equal); Validation (equal); Visualization (equal); Writing – original draft (equal); Writing – review & editing (equal). Jeremy Chao: Data curation (equal); Formal analysis (equal); Investigation (equal); Methodology (equal); Validation (equal); Visualization (equal); Writing – original draft (equal); Writing – review & editing (equal). Sydney Taylor: Conceptualization (supporting); Funding acquisition (supporting); Writing – review & editing (supporting). Liping Wang: Conceptualization (lead); Funding acquisition (lead); Project administration (lead); Resources (lead); Supervision (lead); Visualization (lead); Writing – original draft (lead); Writing – review & editing (lead).

The data that support the findings of this study are available from the corresponding author upon reasonable request.