13.1.4.1 Mo dopant and sol-gel VO₂

Hanlon et al.¹¹¹ investigated the effect of Mo doping on the thermochromic properties of VO₂ synthesized on a glass substrate using a sol-gel method. Figure 13.5 compares the electrical conductivity of undoped VO₂ to that of Mo-doped VO₂ films. The conductivity of both undoped and doped VO₂ films increases with temperature. The Mo dopant reduces the hysteresis width and decreases the T_trans, which reaches 24 °C for 7 at.% of Mo.

13.1.4.2 Co-doping VO₂ with W and F

Burkhardt et al.⁸⁶ studied the effects of doping VO₂ with W and F; Figure 13.6 shows the temperature dependence of the transmittance at a wavelength of 2 μm for undoped, W-doped, and F-doped VO₂ thin films deposited by RF sputtering from a vanadium target in background of O₂-Ar gas. In addition, Burkhardt et al. found that both dopants decrease T_trans and narrow the hysteresis at the cost of switching contrast (i.e., variation of the transmittance between the low-temperature and high-temperature states).⁸⁶

13.1.4.3 Ce-doped VO₂

Recently, Song et al.¹²⁰ synthesized Ce-doped VO₂ films on a muscovite substrate using a sol-gel method. The transmittance switching measurements at a wavelength of 2.5 μm showed that Ce decreased T_trans by about 4.5 °C/at.%. In addition, Song et al.’s results indicate that Ce may be the only VO₂ dopant that reduces the gain size of the VO₂ films (see Figure 13.1 in Song et al.¹²⁰).

13.1.4.4 Comparing W-doped and W-Ti Co-doped VO₂

Soltani et al.³ studied the effects of doping VO₂ with W and co-doping VO₂ with W and Ti. The results are shown in Figure 13.7, which compares the transmittance at a wavelength of 2.5 μm through both heating and cooling cycles of undoped, W(1.4 at.%)-doped, and W(1.4 at.%)-Ti(12 at.%) co-doped VO₂ thin films on quartz substrates. In this figure, we see that the transmittance drops (i.e., the samples become more opaque) with increasing temperature for undoped and metal-doped VO₂, but both W and Ti significantly affect the transmittance of the semiconducting state: the transmittance drops from 50% for the undoped film to about 40% for the W-Ti co-doped VO₂ film and to about 30% for W-doped VO₂. In the metallic state, all of the films are completely opaque. The hysteresis width (about 5 °C) of W-doped VO₂ remains identical to that of undoped VO₂, but co-doping VO₂ with W and Ti almost completely suppresses the hysteresis—a fact that can be exploited in the fabrication of optical modulators. T_trans is significantly affected by W doping, dropping to ~ 36 °C for 1.4 at.% W-doped VO₂ (cf. 68 °C for an undoped VO₂ film of the same thickness) and corresponding to 22.85 °C per at.% W dopant.³

In addition to the optical effects, doping VO₂ with W and Ti also affects electrical resistivity; Figure 13.8 shows the temperature dependence of the electrical resistivity of undoped, W(1.4 at.%)-doped, and Ti(12 at.%)-W(1.4 at.%) co-doped VO₂ thin films. Due to the SMT, the resistivity of all of the samples decreases as the temperature of the sample increases (i.e., when the sample switches to its metallic state, it is a better conductor). W doping has relatively little effect on the resistivity of the metallic state, but decreases the resistivity of the semiconducting state (i.e., doping with metals makes the semiconducting state more metallic) due to the increase in the number of charge carriers in the film resulting from the presence of the W donor dopant.

Compared to the W-doped VO₂, the electrical resistivity of W-Ti co-doped VO₂ is enhanced in both states. In the W-Ti co-doped film, the Ti concentration (12 at.%) is much higher than that of W dopant (1.4 at.%). As a result, the Ti acceptors compensate for the W donors, causing a decrease in the overall carrier charge and enhancing the resistivity in both states.

The change in resistivity as a function of temperature is exploited in IR bolometers that utilize the change resistivity to measure heating by IR radiation. The principal property that must be maximized in order to achieve maximum sensitivity for IR bolometers is the temperature coefficient of resistance (TCR), defined as the slope of the log of electrical resistivity R of the semiconducting state:

$\text{TCR}=\frac{\text{d}R}{R\text{d}T}$

The TCR values for VO₂ (− 1.76%/°C) and W-doped VO₂ (− 1.76%/°C) are comparable to that of VO_x (− 2%/°C) (which is currently used in commercially available uncooled IR micro-bolometers).¹²² Raytheon recently patented a neutron detector system combining a microbolometric sensitive VO_x layer with a neutron sensitive reaction layer (e.g., ¹⁰B or ⁶Li).¹²³

The results of Soltani et al.³ demonstrate that the Ti-W co-doped VO₂ results in a much higher TCR (− 5.12%/°C)—a value comparable to that of a single VO₂ crystal (− 6%/°C). The suppression of both the optical and electrical hysteresis combined with the higher TCR of Ti-W co-doped VO₂ films will open the way for development of innovative IR sensors based on these characteristics.

13.2.3.1 Tuneable metamaterial devices

Crunteanu et al.⁴ exploited the phase transition of VO₂ in the fabrication of thermally tuneable metamaterial devices operating between 0.1 and 1 THz; a sketch of the unit cell and an optical micrograph of their device are shown in the left and right panels of Figure 13.10, respectively.

The VO₂ hybrid metamaterial fabricated by Crunteanu et al. consists of a 500-μm thick c-cut sapphire substrate, a 120-nm thick active layer of VO₂, and a series of Au (200 nm)/Ti (10 nm) periodic resonator cells assembled using an electron-beam evaporation and lift-off process.⁴ This device exploits the thermally controllable THz signal transmitted through the active VO₂ layer.

Crunteanu et al.⁴ used THz time-domain spectroscopy to investigate the temperature dependence of the metamaterial devices. The normalized transmitted signals of the incident THz electric field were extracted from the ratio of the Fourier transform time traces of the transmitted THz signals through the device using the c-cut sapphire substrate as a reference; the results are shown in Figure 13.11.

In Figure 13.11, the two different configurations show that the THz transmitted signals are considerably attenuated when the VO₂ active layer switches to its metallic state at 100 °C (red curves exhibiting plateau). In the semiconducting state (at 20 °C; blue curves exhibiting resonant peaks), a resonant frequency of 0.65 THz is observed in the perpendicular configuration (left panel of Figure 13.11); this resonance is redshifted to 0.45 THz and a second resonant frequency at 0.8 THz (due to the coupling of the different branches of the metamaterial’s unit cell) is observed in the parallel orientation (right panel of Figure 13.11). These results show that simply rotating the device around the direction of the incident THz electric field can vary the resonant frequency at room temperature.

The fact that we can electronically stimulate the SMT of VO₂ implies that we can also fabricate VO₂-hybrid metamaterial devices that can be electrically tuned. Driscoll et al.¹²⁷ fabricated such a device by incorporating a patterned single layer of gold split-ring resonator (SRR) arrays onto 90-nm-thick VO₂ films deposited on a sapphire substrate using a sol-gel method (see panel A of Figure 1 in Driscoll et al.¹²⁷). This device exploits the memory effect of the wide hysteresis loop of the SMT of VO₂ (see panel B of Figure 1 in Driscoll et al.¹²⁷): the induced phase transition in VO₂ layer creates a large change in the permitivity of VO₂ that then increases the capacitance of the SRR. As a result, the resonant frequency decreases as the temperature increases (see panel C of Figure 1 in Driscoll et al.¹²⁷). In the semiconducting state, the resonance frequency is 1.65 THz; this frequency redshifts by ~ 20% as the temperature increases and induces the phase transition of VO₂.

The device’s electronic response control was demonstrated by maintaining its temperature at 338.6 K where the hysteresis is more pronounced (see the vertical dotted line in panel B of Figure 1 in Driscoll et al.¹²⁷), and by then applying a series of electric pulses to the device. Figure 2 of Driscoll et al.¹²⁷ shows the effects on the resonant frequency of the device due to the 1-s electric pulses with increasing power. In that figure, we see that the resonant frequency redshifts and the capacitance increases as the power of the electric pulses increases. The frequency remained redshifted even after stopping the electric pulses, thereby demonstrating the persistent change in the VO₂-hybrid metamaterial device.

Several groups have modeled these devices using a circuit model, in which each SRR array is modeled by an RLC circuit element with a resonance frequency ω = (LC)^− 1/2.¹²⁷ The inductance L is constant, and the capacitance of the SRR array changes with the frequency C/C₀ = (ω/ω₀)². The increase in conductivity introduces a memory capacitance C_m and a memory resistance R_m, which reduce the quality factor of the device. The total capacitance C_tot = C₀ + C_m increases since each electric pulse (V_ext) causes the conductivity of the VO₂ layer to increase.

13.2.4.1 VO₂-Au nanoparticle composite

Orlianges et al.¹²⁸ investigated the electrical resistivity and optical properties of VO₂ thin films containing Au nanoparticles (NPs). The VO₂-Au nanocomposite had a thickness of 200 nm that was synthetized by PLD onto c-cut sapphire substrates and alternatively ablating Au and vanadium targets using a KrF laser in an O₂ reactive gas atmosphere at a temperature of 920 K; an image of the resulting material is shown in Figure 13.12.

The temperature dependence of the electrical resistivity measurements showed the normal behavior of the SMT of VO₂: the resistivity decreased inversely with the temperature for both pure VO₂ films and the VO₂-Au composite. However, the VO₂-Au NP composite showed slightly less switching contrast (but within the same order of magnitude of plain VO₂ film) and had a slightly lower T_trans (341.85 K for VO₂-Au NPs compared to 345.34 K for the plain VO₂ film). Figure 13.13 shows the transmittance between 300 and 1500 nm for the VO₂-Au NPs composite, where we see that the transmittance in the NIR decreases with increasing temperature. This behavior is due to the thermally induced increase in free-carrier charge within the VO₂. The absorption peak near 650 nm is attributed to the LSPR of the Au NPs, and the position of the peak blueshifts as the temperature increases due to the variation of the dielectric properties of VO₂ at high temperatures (see, e.g., Verleur et al.¹²⁹). The transmittance hysteresis loops (see the inset in Figure 13.13) indicate that as the temperature increases, the transmittance at 750 nm increases, but the transmittance at longer wavelengths (e.g., 1550 nm) decreases.

13.2.4.2 Bilayer structure: Au-nanoparticles/VO₂ thin films

Xu et al.¹³⁰ created a bilayered system of Au NPs deposited directly on top of 25 nm of VO₂ on c-cut sapphire substrates (i.e., Au-NPs/VO₂/Al₂O₃). The VO₂ and Au NPs with equivalent mass thickness (d_m = 1, 2, 4, and 6 nm) were created using an RF magnetron sputtering process.

The bilayer structure was significantly more opaque in the NIR than the VO₂ reference material. As expected, both the VO₂ and bilayer composite samples were transparent in the infrared in the semiconducting state (at 30 °C), but the two samples were quite different once the VO₂ became metallic (at 80 °C). This difference was caused by the addition of the Au NPs and was very sensitive to the amount of Au in the sample. In addition, as more Au was added, the peak position of the LSPR continously redshifted (both as a function of the amount of Au and temperature). Figure 13.14 shows the absorption switching of the Au (6 nm)/VO₂(25 nm)/sapphire composite with its corresponding LSPR peak position: the redshift of this sample is 200 nm. Note that the LSPR of Au (6 nm)/quartz reference (without VO₂) is unaffected by the temperature increase, showing that this effect must be due to the interaction of the Au with the VO₂.

13.2.4.3 Au::VO₂ nano-arrays

Ferrara et al.¹²⁶ compared the optically induced SMT of VO₂ film with that of the plasmonic response of Au::VO₂ nanocomposites. In this case, the nanocomposites consisted of 60 nm of VO₂ on top of 140-nm diameter, 20-nm thick Au arrays. The Au was patterned onto a c-cut sapphire substrate using standard electron-beam lithography techniques, and the VO₂ layer was deposited onto the fabricated Au arrays with PLD. Figure 13.15a shows the SEM image of the fabricated Au array before deposition of the VO₂ layer. Figure 13.15b compares the extinction efficiency of the Au::VO₂ structure when the VO₂ switches from its semiconducting state at 22 °C to its metallic state at 100 °C. Similar to the prior examples, the LSPR redshifts with increasing temperature, but shifts as much as 250 nm using this construction. Figure 13.15c compares the temperature dependence of the absorption at 785 nm for Au::VO₂ structure and VO₂ film. In the semiconducting state, both samples are relatively transparent, but in the metallic state, the Au::VO₂ is significantly more opaque than plain VO₂ film.

Ferrara et al.¹²⁶ also investigated the effects on switching due to optically induced LPSRs using a standard pump-probe experiment. The pump beam (785 nm) with variable intensity I was used to induce the SMT, while the transmittance of the Au::VO₂ structure was probed by an 18 W/cm² diode laser (λ = 1550 nm). The temperature of the structure was maintained in the semiconducting state at 55 °C [i.e., T < T_trans(VO₂)], and for each measurement, the Au::VO₂ nanostructured array was irradiated by the pump beam for 5.3 s. Figure 13.15d shows the contrast of the switching C(t) (defined as the transmittance at 1550 nm normalized to that of the semiconducting VO₂ for both the Au::VO₂ nanostructured array and the VO₂ film), and as expected, the material becomes opaque when the VO₂ changes phase and the contrast is more dramatic as the laser power is increased.

The Au::VO₂ structure switches to the metallic state faster than plain VO₂ film; at the maximum intensity, the Au::VO₂ structure reaches 85 °C in ~ 1 s (cf. equilibrium of 5.3 s for the plain VO₂ film). Using first-principles thermodynamics,¹²⁶ Ferrara et al. show that the temperature increase ΔT of the interface between the substrate and the sample (either plain VO₂ film or a Au::VO₂ nanostructured array) can be described by:

$\Delta T=\frac{2q\sqrt{\alpha t}}{\kappa }\left[\frac{1}{\sqrt{\pi }}-i\text{erfc}\left(\frac{a}{2\sqrt{\alpha t}}\right)\right]$

in which q = I(1 − R)[1 − exp(− γ_effz₀)] is the energy absorbed in the sample, and α = 5.3 × 10^− 3 cm²/s and κ = 9.6 mW/(cm °C) are the thermal diffusivity and conductivity of the glass substrate, respectively. In Equation (13.2), R is the reflectivity at the VO₂-air interface, γ_eff is the effective laser absorption coefficient, and z₀ is the thickness of the film.

This basic model has a few flaws. It predicts that C(t) decreases more quickly and C_max increases more quickly than were experimentally observed. These discrepancies can be explained once we understand that this model overestimates the absorption in the metallic state, which then results in an overestimation of C_max. The model also ignores the fact that the granular nature of the VO₂ tends to preclude lateral heat diffusion out of the laser focal spot. Despite its shortcomings and relative simplicity, the model does support the main experimental results: the Au::VO₂ reaches a higher VO₂ metallic fraction and a higher C_max faster than the VO₂ film without the Au nanostructure (Figure 13.15d).

The dynamics of Au::VO₂ structure and VO₂ film were analyzed by fitting C_max(I) with a three-parameter sigmoid given by:

$C\left(I\right)=1-\frac{1-C_{\text{H}}}{1+exp\left(\frac{I-I_{\text{c}}}{I_{\text{w}}}\right)}$

in which I_c is the critical switching intensity defined at the midpoint of the transition, I_w is the width of the transition, and C_H is the high-temperature contrast. ¹²⁶ The fit to the experimental data (shown in Figure 13.16) indicates that both the I_c and I_w decrease by approximately a factor of 1.5 in the case of the Au::VO₂ structure, demonstrating that less pump laser power is required to drive the phase transition of the Au::VO₂ nanostructure than that of plain VO₂.