Synthesis of vanadium dioxide thin films and nanostructures

Vanadium oxides are ideal prototypical materials with strong electron–electron correlations discovered to demonstrate remarkable changes in their electrical properties from the insulating to the metallic state by applying external stimuli. A sudden change in the intrinsic electrical conductivity of vanadium oxides is normally related to a reversible semiconductor-to-metal transition (SMT), which is more widely known as a metal-to-insulator transition (MIT). Vanadium dioxide (VO₂), for example, undergoes fully reversible MIT between insulating monoclinic VO₂(M) to metallic rutile VO₂(R) phases or vice versa triggered by externally applied stimuli such as heat or stress. The phase transition corresponds to a four orders of magnitude dramatic change in resistivity and optical properties, which can be reversible via natural cooling. Since the discovery of MIT in VO₂(M) in the late 1950s,¹ vanadium oxides have attracted extensive recognition as resistance switching material for highly sensitive smart device applications. This heightened the interest in the other phases of vanadium oxides, and a series of vanadium oxide systems were realized to demonstrate MIT behavior. Non-stoichiometric vanadium oxides exhibiting MIT are categorized into two families, namely, Magneli-type vanadium oxides with a general formula of V_nO_2n−1, and Wadsley phases with structural compositions of V_nO_2n + 1.² The Wadsley phases are less oxidized compared with the Magneli phases and do not exhibit MIT, except V₆O₁₃.³ On the other hand, most Magneli phases, comprised of corundum and rutile-like layers with structural similarity to dioxide and sesquioxide, undergo reversible phase transition from metallic to insulating states at certain temperatures.⁴ Figure 1(a) shows typical examples of Magneli phase V_nO_2n−1 oxides and their corresponding MIT temperatures.

A Magneli-type VO₂ is an ideal textbook example of a system that exhibits abrupt structural transition from the metallic to the insulator state with an electrical conductivity change up to five orders of magnitude. To date, many VO₂ crystalline polymorphs have been reported, including monoclinic VO₂(M1) (P2₁/c),⁵ tetragonal VO₂(R) (P4₂/mnm),⁵ monoclinic VO₂(M2) (C2/m),⁶ tetragonal VO₂(A) (P4₂/nmc),⁷ triclinic VO₂ [P*(2)],⁸ VO₂(C),⁹ orthorhombic VO₂⋅H₂O (P222),¹⁰ tetragonal VO₂⋅0.5H₂O (I4/mmm),¹¹ monoclinic V₂O₄ (P21/c), and V₂O₄^.2H₂O.¹² In addition, the VO₂(D) phase, which is isostructural to monoclinic NiWO₄, with formation energies very close to VO₂(R),¹³ and orthorhombic paramontroseite VO₂(P),¹⁴ are also reported to exist. However, only rutile-type VO₂ phases exhibit fully reversible MIT phenomena near the phase transition temperature (T_c) of 68 °C.¹ In the rutile-type family, at 68 °C, the insulating M1 (monoclinic, P2₁/c) phase (low T) transforms into the tetragonal metallic (tetragonal, P4₂/mnm) R-phase (high T) under ambient pressure. Under applied hydrostatic pressure, the M1 phase can drive into another isostructural and more conductive M1′ phase and then toward a new metallic X (monoclinic) phase at room temperature.¹⁵ Additionally, the application of uniaxial stress induces the formation of a second metastable monoclinic (C2/m) M2 phase in VO₂ between the M1 and R phases.¹⁶ Unlike in the M1 phase, only half of the V-atom pairs dimerize in the M2 phase, and the other half of V atoms do not undergo dimerization, which results in more insulating characteristics than in the M1 phase.¹⁷ Hence, the M1 phase can be interpreted as the overlap of two M2 phases. A metastable M2 phase is reported to form either by metal doping or by applying mechanical stress.¹⁸ Furthermore, a triclinic (T) phase is reported to exist as an intermediate between M1 and M2.¹⁹ 

Due to the variable oxidation states of vanadium, it is highly complicated to sketch the phase diagram of VO_x compounds. In fact, a small deviation in the V–O stoichiometry results in the formation of different VO_x systems with variable crystalline structures. However, a phase diagram of the VO–VO_2.5 system constructed using thermodynamic models is shown in Fig. 1(b), suggesting that the correct stoichiometry for VO₂(M1) phase formation is tuned by adjusting the annealing temperature and mole fraction ratios of V and O.²⁰ 

Although VO₂ generates immense interest as multifunctional inorganic materials for energy-saving and smart device applications, the experimental synthesis of the pure VO₂(M1) phase is highly challenging owing to the presence of other stable oxide phases, and the formation of VO₂(M1) occurs only at a very narrow range of oxygen partial pressure. The current state-of-art preparation methods are still complicated for the large-scale production of the impurity free VO₂(M) phase. More often, VO₂ nanostructures grown on a selective substrate have limited size and poor yield, indicating the demand for a simplified and efficient large-scale method/process to produce a single crystalline VO₂(M) phase. It should be noted that there is still no MIT model that can clarify all of the experimental results, and the precise MIT mechanism in VO₂ can be best described as a dispute between the electron-driven Mott transition and the structure-driven Peierls transition, and as a cooperation of these two mechanisms. This tutorial summarizes the fundamentals of MIT in VO₂, current synthetic strategies, applications, and challenges are outlined for beginners working in the field of VO₂ synthesis.

MIT in VO₂ is a consequence of structural alteration from monoclinic to rutile states or vice versa. Nevertheless, the origin of MIT in VO₂ is still a topic of debate due to the controversies in electronic band descriptions in single particle theories and correlated electronic states in Mott–Hubbard theory. As given in Fig. 2(a), VO₂(R) has a tetragonal rutile crystal structure where V⁴⁺ ions occupy body center and corner positions, with each V-atom surrounded by six O²⁻ atoms and forming an octahedral unit cell. The infinite vanadium ions in tetragonal VO₂(R) are aligned in linear chains, where the adjacent V–V bond distance along the rutile c axis is ≈2.85 Å. Therefore, all metal V atoms in the V–V linear chain have equal shares of d-orbital electrons, resulting in metallic conductivity with a low resistivity of about 10⁻⁶ Ω m.²¹ In contrast, the monoclinic VO₂(M) crystal structure is characterized by a distorted rutile crystal structure, where the unit cell volume of VO₂(M) is double that of VO₂(R). The phase transition from a high symmetry tetragonal structure to a low symmetry monoclinic deviates from the position of the V atoms. In this case, dimerization creates zigzag V-atom chains with two different V–V distances of 3.19 and 2.60 Å between the nearest adjacent V atoms [Fig. 2(b)]. Dimerization patterns of V chains localize d-orbital electrons within the dimer, which gives VO₂(M) an insulating character with resistance in the order of 0.1 Ω m.²² 

According to the crystal field band structure model by Goodenough,²³ VO₂ adopts an unstrained stoichiometric monoclinic (M) crystal structure at a low temperature and a tetragonal rutile (R) phase at a high temperature. In the molecular orbital structure of VO₂, the V 3d orbital undergoes octahedral splitting into e_g and t_2g symmetries. Owing to the orthorhombic component, the t_2g state further splits into two d_π orbitals and one d|| orbital. At a high temperature (i.e., above T_c), the VO₂(R) phase forms a π-bond and π* anti-bond between the V and O orbitals along with a d|| non-bond between adjacent V orbitals. Although the VO₂(R) phase has an energy bandgap of about 2.5 eV, an overlap of partially filled π* and d|| bands, and the Fermi energy level near the overlap point of the π* and d|| bands, this material shows a metal-type conductivity [Fig. 2(c)]. Below T_c, the zigzag distortion changes the VO₆ octahedral position in VO₂(M), which increases the π overlap in V–O orbitals, thereby upwardly shifting the anti-bonding π* level. The significant structural band change further splits the d|| band into d|| bonding and d||* anti-bonding components. As a result, a narrow bandgap of 0.7 eV is generated between the π* and d|| bonds, which induces an insulating character to the VO₂(M) phase. Although Goodenough clarified the nature of MIT in VO₂, modern experimental and theoretical studies contradict the findings of this model.²⁴ 

A single crystal VO₂ nanobeam phase transition using a nanomechanical strain apparatus revealed that at zero strain, the MIT temperature of the insulating M1 and M2 phases is degenerate with the metallic R-phase [Fig. 3(a)].²⁵ In unstrained VO₂ or undoped VO₂, the free energies of all three phases are degenerate at its triple point. Figure 3(b) is a pressure–temperature (P–T) phase diagram of VO₂, including M1, M1′, R, O, and X phases constructed using optical reflectance, Raman, and data of electrical properties obtained under isothermal and/or isobaric conditions.¹⁵ The P–T phase diagram features a non-monotonic phase boundary between low temperature insulating phases (M1 and M1′) and high temperature metallic phases (R, O, and X). Along the pressure axis of the P–T phase diagram, the more insulating M1 progressively transforms into the less insulating M1′ phase and, finally, into the metallic X phases. For the transition from the insulating M1 to the metallic R-phase, the dT/dP phase boundary slope is determined to be 1.4 °C/GPa over an average pressure range of 0 to 16 GPa.

The extrinsic doping of VO₂ with another metal element can adjust the phases and the MIT temperature.²⁶ Figure 3(c) shows the substitution effect on the MIT temperature of VO₂. The substitution of the V-atom with dopants higher than a +4 valence of vanadium tends to decrease the MIT temperature.²⁷ The doping of VO₂ with W reduces the T_c up to 27 K/at. % of doping, and at 7.5 at. % of W doping, the lowest T_c of −50 °C was demonstrated.²⁸ On the other hand, dopants with valences lower than vanadium can increase the MIT temperature,²⁹ with the Ga-atom being the most effective to enhance the T_c at 8 K/at. %, but 1% Ga doping fraction is the saturation limit.³⁰ As was evident, Cr-doped Cr_0.2V_0.8O₂ demonstrated the highest T_c of 100 °C.³¹ Metal doping of VO₂ can work up to a maximum doping fraction, beyond which MIT might disappear or materials may have more complex phases than the M1 and R phases.

Despite six decades of MIT discovery, the synthesis of single phase VO₂ remains a challenging task. In contrast, VO₂ attracted great attention as a smart material for a number of electronic and thermochromic applications. The ultrafast resistive switching feature of VO₂ is extensively explored in smart electronic applications such as transistors, bolometers, ultrafast optical switching, and so on (Fig. 4). Some important applications of VO₂ are briefly summarized herewith.

A field-effect transistor (FET) is a three terminal device used to control current flow using an electric field. The three terminals of FET are source, gate, and drain. FET consists of a semiconductor channel to conduct electricity and a capacitor as a gate to amplify the carrier density of the channel by electrostatic charging effects. Despite groundbreaking and novel metal oxide semiconductor FETs (MOFET) in modern electronics, MOFET miniaturization is limited by the short channel effect and current leakage. Henceforth, new types of FETs, which work with a different operational mechanism, are of great interest. Strongly electron-correlated materials, such as VO₂ (Mott insulator), can be utilized as the channel in an FET and can operate with different mechanisms, namely, “Mott-transistor” (MT).³² MT supersedes MOFETs in switching speed, sharp threshold swing, and minimal fluctuation effect. Therefore, VO₂ has been explored as a channel material in MTs. However, the threshold carrier density of VO₂ MIT (10¹⁸–10¹⁹/cm³) is too high to attain by solid gating. As result, liquid gating using an electrolyte or ionic liquid is typically adopted in the construction of VO₂-based FETs [Fig. 5(a)].³³ To prevent the electrochemical reaction of electrolyte ions with oxides, Zhou et al.³⁴ applied monolayer graphene as an inert ion diffusion barrier at the oxide–electrolyte interface, which significantly suppressed the resistance modulation and enabled retention of the insulating phase [Fig. 5(b)]. Nakano et al.³⁵ developed ionic liquid-gated, VO₂-based electric double layer transistors, and the structural transition of VO₂ was attributed to the electrostatic gating effect. In addition, electrochemically driven doping of hydrogen ions or oxygen vacancies were attributed to the gating mechanism.³⁶ 

A bolometer detects and measures incident electromagnetic radiation. Ideal bolometers have an active material whose electrical resistance changes with the temperature induced by radiation absorption. Bolometer senses IR light intensity by screening the change in resistance on a detecting layer. For the high sensitivity of a bolometer, it is desired to have a large thermal coefficient of resistance (TCR) and low resistivity (ρ) component as the active material.³⁷ Typical bolometers employ small TCR materials such as nickel, titanium, and platinum as active materials with low sensitivities. Owing to large TCRs at the MIT region, VO₂ applied as an active material in microbolometers [Fig. 5(c)].³⁸ The MIT phenomenon of VO₂ suits its application as an IR sensing material in a bolometer. Nevertheless, the MIT of VO₂ in a narrow range hinders continuous temperature sensing, hysteresis during heating and cooling processes, and high resistivity (1 Ω.cm) in insulating state limits with VO₂ as a detecting material. Research efforts have been devoted to broadening the MIT temperature range, reducing the hysteresis by small temperature differences, and lowering the resistivity by doping. The undoped VO₂ nanowires display a rapid resistance switching around 67 °C, while the resistance of graded W-doped VO₂ nanowires progressively decreased from room temperature to 60 °C, without sudden resistance changes [Fig. 5(d)]. Particularly, graded W-doped VO₂ nanowires were reported to have a high TCR of 10%/K and a low resistivity of 0.001 Ω cm, which should be reliable for highly sensitive IR materials in microbolometers [Fig. 5(e)].³⁹ 

Recently, the main purpose of developing VO₂-based transparent thin films has shifted to energy conservation. The heat reflection and absorbing properties of VO₂ are pertinent as smart window materials. The energy efficiency of VO₂ thin films is estimated by building model houses with VO₂-coated glass substrates mounted as smart windows or roof. Gao et al.⁴⁰ fabricated a 2% tungsten-doped VO₂-coated float glass (30 × 40 cm²) and fixed it to a model house as a front room window. For comparison, a blank float glass was used as a front window for another identical model house. Both model houses were equipped with thermocouples to monitor the temperature, and IR lamps as light sources were kept at an equal distance. The model houses were irradiated with IR light, and the relative temperature change noted in the VO₂-coated model house was approximately 9 °C lower than the uncoated glass model house [Fig. 5(f)]. A composite of Cr₂O₃/VO₂/SiO₂ (CVS) coated glass substrates showed an ultrahigh solar modulation ability of 16.1% and excellent luminous transmittance of 54.0%.⁴¹ The model house equipped with CVS-coated glass exhibits rather slow temperature increments of 29.8%, whereas the blank uncoated house reflected rapid temperature increments by 84.8%. Polycrystalline VO₂ shows several magnitudes of conductivity change across the MIT, which led to an increased use of VO₂ for thermally conductive modulation applications as a thermal switch. The lattice thermal conductivity of single crystalline VO₂ nanowires measured by a suspended microdevice system was found to be almost a constant (∼6 W/mK) across the MIT [Fig. 5(g)].⁴² The total thermal conductivity of the VO₂ nanowires was typically lower than the value expected from the Wiedemann–Franz law. However, W-doped VO₂ shows an abrupt increase in the total thermal conductivity across the MIT, where such thermal conductivity changes normally suggest possible modes of thermal conduction switching using W_xV_1−xO₂ compounds. Additionally, the reversible phase transition of VO₂ drives its use in actuator applications. A flexible actuator composed of VO₂ and carbon nanotube biomorph produced high responses of ∼100 Hz and longer lifetime actuation cycles [Fig. 5(h)].⁴³ The power consumption of such microscopic bending in actuators can be lowered by reducing MIT via VO₂ doping.

Typically, synthetic conditions have a significant influence on the morphology and defects of VO₂ products, which directly affects their MIT behavior. A plethora of methods have been designed and developed to form the VO₂(M) phase. Based on the reaction medium, these methods are primarily categorized as gas phase synthesis and solution-based methods. Gas phase deposition techniques frequently used for the synthesis of VO₂(M) structures include chemical vapor deposition, physical vapor deposition, and sputtering methods. In contrast, solution-based approaches mainly consist of a sol–gel process, hydrothermal synthesis, and polymer assisted deposition.

Gas or vapor phase deposition techniques are commonly used for the fabrication of high-quality thin films, where one or more gaseous precursors are volatilized onto a target substrate to form a solid thin film. Thin film quality is highly dependent on the deposition rate, the distance between source and target, and the partial pressure of the system.

The first report of the chemical vapor deposition (CVD) of VO₂ films can be traced back to 1967.⁴⁴ Since then, CVD has been extensively used in the fabrication of high-quality VO₂ films. Depending on the synthetic configuration, CVD techniques used for the synthesis of VO₂(M) nanostructures are atmospheric pressure CVD (APCVD), aerosol assisted CVD (AACVD), metal-organic CVD (MOCVD), plasma enhanced CVD (PECVD), and so forth. The principle composition and quality of CVD-deposited VO₂ films are mainly determined by the vanadium precursor and the gas carrier type used. The APCVD system only utilizes volatile precursors, which is a limitation. Two types of vanadium precursors are used in the APCVD technique, namely, vanadium halides, such as VCl₄ and VOCl₃, and vanadium organometallics [VO(acac)₂ and VO(OC₃H₇)₃]. Maruyama and Ikuta used vanadium (III) acetylacetonate in the APCVD technique for the growth of high-quality polycrystalline VO₂ films on fused quartz and sapphire substrates.⁴⁵ Qureshi et al.⁴⁶ inspected the formation of VO₂ film on a glass substrate via APCVD reaction of VOCl₃ and water at 650 °C. The incorporation of TiCl₄ as a co-reactant affected the growth morphology of VO₂ thin films without titanium contamination. The morphology of VO₂ films deposited at 650 °C displayed an irregular ball shaped morphology, while decreasing the deposition temperature to 550 °C produced “warm-like” particles of VO₂ and spherical ball-like TiO₂ particles [Figs. 6(a) and 6(b)]. Interestingly, the introduction of TiCl₄ in the VO₂ films reduced the thermochromic switching temperature from 69 °C in bulk VO₂ to 49 °C. Gaskell et al.⁴⁷ used VCl₄ and water to fabricate smooth monoclinic VO₂ thin films with an optimized APCVD technique on an fluorine tin oxide (FTO) substrate. The deposited VO₂ films exhibited a lower surface roughness and good transmittance properties. Parkin and Manning developed W-doped VO₂ films on a glass substrate using an APCVD method with VOCl₃, H₂O, and WCl₆ precursors [Fig. 6(c)].⁴⁸ They found that tungsten incorporation into VO₂ films reduced the transition temperature by 19 °C per W at. % doping. The significant difference in VCl₄ and VOCl₃ systems are the redox reactions of VOCl₃⋅H₂O and the hydrolysis reaction of the VCl₄.H₂O system. Therefore, VCl₄ systems are easier to control than the VOCl₃ system and are suggested for APCVD VO₂ growth.

A pure monoclinic phase of VO₂ was grown on CMOS-compatible Si substrates using an MOCVD technique at a deposition temperature range of 520 °C to 550 °C.⁴⁹ The morphology of VO₂ films significantly varied with the deposition temperature; however, films retained the M1 phase of VO₂. Notably, the morphology of the thin film deposited at 535 °C showed sharp grain boundaries with an average grain size of 200 nm [Figs. 6(d)–6(f)]. The temperature dependent resistance behavior of a VO₂ film grown at 535 °C was better than that of the films grown at other temperatures of 520 °C and 550 °C. MOCVD deposition temperature induces various fractions of vanadium surface defects in the form of V³⁺, V⁴⁺, and V⁵⁺ oxidation states. The fractions of these components vary with the deposition temperature, and the fraction of V-defects-content regulates the MIT behavior of VO₂ thin films. Blackburn et al.⁵⁰ designed a single molecular precursor dichloro(oxo) vanadium (IV) diethyl malonate [[VOCl₂[CH₂(COOEt)₂]]₄] for the synthesis of the VO₂ monoclinic phase using AACVD [Fig. 6(g)]. The developed single molecular precursor has advantages over typical poly-molecular precursors including high solubility in AACVD-preferred solvents such as hexane, dichloromethane, and chloroform, whereas other precursor compounds, such as vanadium acetate, are sparingly soluble. Furthermore, single molecular vanadium precursors have high oxygen reactivity due to the presence of volatile chloride ligands. The morphology of VO₂(M) films deposited by AACVD using single molecular precursors at 550 °C was highly disordered with large rod-like assemblies of 300–500 nm [Fig. 6(h)], while W doping greatly altered its morphology into densely packed round vertical protrusions. The thermochromic characteristics of AACVD-deposited VO₂(M) films are competitive (a solar modulation of 15.9%), and small amounts of W doping into the molecular precursor reduced the MIT temperature from 70 °C in undoped VO₂ to 50 °C in W-doped VO₂.⁵⁰ 

Compared with other chemical deposition techniques, PECVD offers a low temperature synthesis of VO₂ thin films using either a strong electric field or microwaves to produce plasma that causes the rapid decomposition of vanadium precursors and produces thin films. Yet, there are very few reports on VO₂ synthesis using PECVD due to its complexity of deposition. Warwick et al.⁵¹ applied a potential difference between the top plate and bottom substrate to generate an electrical field with a positive bias at the end substrate to produce uniform VO₂(M) thin layers. The presence of the electrical field has led to substantial changes in the thin film microstructure, such as increased electrical field-reduced crystalline size and improved near-infrared transmittance property of VO₂(M). The limitation of the PECVD technique is that there are several oxidant and reductant by-products that prevent the growth of pure VO₂(M). Therefore, it is recommended that an appropriate single molecular vanadium precursor, such as VCl₄ and VO(acac)₂, be used for PECVD.

PVD is a well-known gaseous technique to fabricate vanadium oxide thin films, which usually involves evaporation, transportation, reaction, and deposition steps. In PVD, vanadium solid precursors are bombarded by high energy sources such as electron beam or ions under reduced pressure. PVD offers a low temperature synthesis of VO₂ thin films and an eco-friendly method with ease of multilayer deposition. Pulse laser deposition and sputtering have been the most common PVD deposition processes to fabricate VO₂ thin films. The optoelectronic characteristics of VO₂ films deposited by the PVD process are strongly influenced by the nature of the substrate materials. For instance, the morphology, structural phase, and optical properties of VO₂ thin films were investigated by changing substrate materials such as glass,⁵² silicon,⁵³ quartz,⁵⁴ and sapphire.⁵⁵ Fuls et al.⁵⁶ from Bell Labs first used reactive sputtering techniques to obtain VO₂ nanostructures under an argon–oxygen atmosphere. Thereafter, various sputtering developments were designed to obtain high-quality VO₂ films. To date, direct current (DC), radio frequency, and magnetron sputtering have been used for the fabrication of VO₂ films. The advantages of sputtering processes are good homogeneity, scalability over large substrate areas, and high-quality thin films. To standardize the sputter deposition of VO₂ thin films, deposition parameters such as temperature, oxygen partial pressure, and vanadium to oxygen ratio critically influence the structural, optical, and electrical properties of VO₂ thin films. Some of the notable findings on standard sputter deposition parameters can be tracked to the contributions of Jin et al.⁵⁷ and Razavi et al.⁵⁸ 

Recent developments in sputter deposition have accompanied great enhancements in VO₂ thin film thermochromic properties. Compared with conventional DC-sputtered VO₂, reactive pulsed DC-sputtered VO₂ films exhibited improved thermochromic behavior.⁵⁹ Indeed, asymmetric-bipolar pulsed DC power supply enabled an increase in sputter yield by periodically reversing the voltage of the electrode, thus neutralizing target surface poisoning. In addition, pulsed DC sputtering reduces working gas pressure and increases ion current density and ion bombardment, which improves the film density/crystallinity. Moreover, optimal Nb-doping into reactive pulsed DC sputter deposited VO₂ films decreased the phase transition temperature from 59 °C to 34 °C. However, the doping of Nb into VO₂ films degraded the optical transmittance of thin films due to the introduction of large defects in the crystal lattice. Chen et al.⁶⁰ developed VO₂ thin films using reactive ion beam sputter deposition with a low phase transition temperature of 35 °C [Fig. 7(a)]. The substantial suppression in the MIT temperature of VO₂ films for room temperature applications is ascribed to decreases in grain size and lower crystallite sizes of fabricated VO₂ thin films. The morphology of VO₂ thin films deposited on silicon substrates shows fine nanostructures with average crystallite sizes of 20 nm [Fig. 7(b)]. This study highlights that the phase transition temperature of VO₂ can be lowered by reducing the particle size of VO₂ rather than common doping methods. Later, Zhang et al.⁶¹ studied the relationship between grain size and hysteresis width in VO₂ thin films [Fig. 7(c)]. DC magnetron-sputtered VO₂ thin films with varied particle size were grown by controlling the annealing time. Extending the annealing time increased the VO₂ particle size and, after 45 min of annealing time, the grain sizes of VO₂ particles decreased. Although the authors did not mention the reasons for this phenomenon, the VO₂ films fabricated under optimal annealing time represented improved optical transmittance. In recent years, high power impulse magnetron sputtering (HiPIMS) has emerged as a modern PVD technique for growing high-quality VO₂ thin films. The HiPIMS technique offers extremely high peak power density and a high plasma density with very short plasma pulse durations. In HiPIMS, the high fraction of ionization and high plasma density and bias voltage are applied to control the bombardment momentum and improve the quality of VO₂ thin films. Lin et al.⁶² studied the effects of pulse bias voltage on the thermochromic properties of VO₂ films prepared by the HiPIMS technique. The crystalline orientation of VO₂ thin films changed with varied magnitude of applied bias, and films deposited with a bias voltage of −250 V produced smaller crystalline sizes of ∼12 nm. As a result, the MIT temperature of the VO₂ films was reduced from 54 °C to 31.5 °C when the applied bias voltage increased from −54 to −250 V.

Pulsed laser deposition (PLD) is a relatively new field of the PVD process for thin film deposition. PLD-assisted VO₂ thin films were first reported in 1993,⁶³ where a highly powered pulsed laser beam was focused on a metallic vanadium target that evaporated the source and deposited on the target substrate as a thin film inside a vacuum chamber with Ar/O₂ process gas. The PLD technique is highly efficient in controlling the stoichiometry and orientation of VO₂ thin films. The oxygen content of the VO₂ film was optimized by controlling the partial pressure of oxygen, which is critical for growing the pure VO₂ phase. Moreover, in PLD, oxygen flow rate, laser wavelength, and target substrate are vital parameters that influence the MIT characteristics of developed VO₂ thin films. The epitaxial grown VO₂ thin films on a sapphire substrate exhibit a lower MIT temperature than bulk or polycrystalline materials. Fu et al.⁶⁴ comprehensively presented a study of MIT in PLD epitaxial VO₂ thin films grown on a c-plane sapphire substrate. Their findings indicated that the growth direction of epitaxial VO₂ thin films on sapphire is oriented along [010]_M1, instead of [001]_M1. The electron effective mass of the metallic and insulating phase was determined by Hall and Seebeck measurements. Across the MIT region of PLD-grown VO₂, electrical conductivity changes were mainly accountable to the variation in electron density of the metallic and insulating states.⁶⁴ Pauli et al.⁶⁵ used synchronous PLD to grow VO₂ films on Si (001) and Al₂O₃ (0001) substrates. The VO₂ film grown on the Si (001) substrate at room temperature was amorphous and annealing these films produced crystalline VO₂ nanoparticles of hemispherical islands [Fig. 7(d)]. In contrast, VO₂ deposited on the Al₂O₃(0001) substrate produced preferentially oriented rod-like nanoparticles [Fig. 7(e)]. The optical measurements of VO₂ thin films grown on Si (001) substrates evidenced sharp thermal hysteresis and good switching characteristics compared with VO₂ deposited on the sapphire substrate [Figs. 7(f) and 7(g)].⁶⁵ The variation in the thermal switching of VO₂ films on different substrates was governed by the crystallinity of the VO₂ phase and strain relaxation flattening of the nanoparticles on the sapphire substrate. Hajlaoui et al.⁶⁶ used boron doping effects in reactive PLD-grown VO₂ thin films. A small amount of boron doping stabilized the VO₂(R) phase, and the (011) diffraction peak of VO₂ shifted toward higher wavelengths and stabilized the VO₂ R-phase. Boron doping drastically reduced the MIT behavior of PLD-grown VO₂ films as large as 31.5 °C per atomic percentage of boron, where similar behavior was reported for tungsten-doped VO₂.⁶⁷ Kumi-Barimah et al.⁶⁸ used femtosecond PLD for high-quality VO₂(M1) thin films on sapphire substrates with V₂O₅ as an ablation target. Thin films were deposited at 400 °C and 600 °C, representing the crystalline VO₂(M1) phase. The VO₂(M1) films deposited at 600 °C exhibited sharp and abrupt MIT behavior at a temperature of 66 °C with a narrow hysteresis width of 3.9 °C. The noticeable electrical properties of VO₂ films deposited at 600 °C were ascribed to the surface roughness of the films induced by the agglomeration of VO₂ nanoparticles.

The vapor transportation method (VTM) is widely used to synthesize VO₂ nanowires. VO₂ nanowires grown via VTM usually have rectangular cross sections with exposed [011] facets.⁶⁹ The length, width, and density of VO₂ nanowires grown using VTM can vary with the growth substrate, and this is mostly because of the difference in capillary interactions and crystallographic directions of the substrate. Phase selective growth of free-standing and substrate-bound VO₂ micro/nanowires with various shapes can be grown using liquid droplet condensation over a preferentially oriented substrate. Previously, one-dimensional VO₂ nanowires were thought to grow from a selective facet of VO₂ nuclei by condensing precursors of evaporated VO₂.⁷⁰ It was later discovered that the evaporated and condensed precursor are traces of V₂O₅,⁷¹ and this showed that VO₂ seeds are not required for the preferential growth of VO₂ nanowires. In situ optical and photoelectron emission microscopy analysis revealed that the evaporation of V₂O₅ at low temperatures (< ∼ 700 °C) forms V₂O₅ droplets on the substrate, which later transform into VO₂ nanowires at a high temperature (>800 °C) via reductive reactions in inert atmospheres (vacuum or Ar). Under these conditions, VO₂ nanowires directly crystallize from molten V₂O₅ droplets at high heating rates (ca. 500 °C/min), and when the heating rate is very low, the product undergoes multiple intermediate phase transformations such as V₂O₅, V₆O₁₃, and VO₂ (Fig. 8).⁷¹ More examples reporting the synthesis of VO₂ nanowires/microbeams via VTM are described in Sec. III A.

Solution phase synthesis of VO₂ nanostructures is relatively simpler than gas phase methods. The usual setup in solution phase methods is comparatively less complex and more cost-effective than gas phase methods. Sol–gel and hydrothermal synthesis are common solution phase routes to obtain VO₂(M) nanostructures.

Sol–gel is a wet-chemical process of producing solid nanoassemblies from a monomeric colloidal solution (sol), which acts as a precursor for a polymeric network (gel) of discrete particles. Typical precursors of the sol–gel process are metal alkoxides. Greenberg first deposited gel assisted VO₂ thin films using a VO(OC₃H₇)₃ precursor.⁷² Vapor deposition of metal-alkoxide precursors first produce metal hydroxide complexes, and subsequent annealing reduces the vanadium hydroxide complex into the VO₂ phase, which is shown as VO(OC₃H₇)₃ → VO(OH)₃ → V₂O₅ → V₂O₃ → VO₂. Although other phase formations over VO₂ are competitive, controlled reduction of V₂O₅/V₂O₃ results in the VO₂ phase. The doping of VO(OC₃H₇)₃ gel precursors with tungsten, molybdenum, and niobium shifted the transition temperature of the as-obtained VO₂ thin films below 68 °C. Similarly, Guo et al.⁷³ used sol–gel-derived VO(OC₃H₇)₃ precursors to fabricate VO₂ thin films onto a sapphire (0001) substrate. Variations in the annealing time of thin films produced greater influence on grain size and transition temperature. Prolonging the annealing time from 1 h to 7 h transformed V₂O₃ to V₃O₅ and then to VO₂, respectively. The most common precursors for sol–gel VO₂ thin films are tetravalent vanadium alkoxides, which are usually expensive and harmful and require high annealing temperatures for thermal polymerization. Hence, efforts have been focused on finding alternate vanadium precursors that are low cost, safe to handle, and can form a metal oxide network at lower temperatures.

Huang and co-workers used the cetyltrimethylammonium vanadate (CTAV) precursor for nanoparticle assembled VO₂ films.⁷⁴ As obtained, these VO₂ films exhibited a nano-hierarchical porous network induced by self-assembly of the long-chained CTAV. The visible light transmittance behavior of porous films was 25% higher than the continuous VO₂ films. The porosity of VO₂ films was largely tuned by adjusting the dip coating speed, which had great influence on the transmittance and switching efficiency of VO₂ films. Cao et al.⁷⁵ employed freeze drying methods to fabricate nanoporous VO₂(M) films with excellent thermochromic characteristics. For this, vanadium powder precursors were dissolved into solution by the addition of optimal hydrogen peroxide (H₂O₂) and then to gelation on fused silica substrates either by freeze drying or by solvent evaporation [Fig. 9(a)]. The VO₂ films fabricated by freeze drying with 7.5 ml of H₂O₂ produced a nanoporous film with an average mean pore size of ∼28 nm [Fig. 9(b)]. Importantly, the nanoporous nature of VO₂ films rendered a highly luminous transmittance (T_lum= 50%) and a large solar modulation ability (14.7%).⁷⁵ 

Induced doping in the sol–gel synthesis of VO₂ thin films is known to alter optical and thermochemical characteristics. For instance, sol–gel-derived Ce-doped VO₂ films on muscovite substrates exhibited an appreciable decrease in the phase transition temperature.⁷⁶ In addition, Ce doping narrowed the interfacial surface energy between the substrate and the VO₂ films, forming compact, (011) oriented films. The morphology of Ce-doped VO₂ films represented relatively small grain sizes than the undoped films [Figs. 9(c) and 9(d)] because Ce partially substituted V atoms with Ce⁴⁺ and Ce³⁺ ions in the crystal lattice, and, hence, reduced the particle size by suppressing the grain growth. After Ce doping, the MIT T_c of the VO₂ film decreased by 4.5 °C per 1 at. %, and 1.68 at. % Ce-doping T_c was reduced to 60 °C. However, the metal-insulator transition hysteresis width (T_w) of 0.43 at. % and 1.68 at. % Ce-doped VO₂ films was 14 °C and 16 °C, which was higher than that of the undoped VO₂ films (T_w: 12 °C). Likewise, sol–gel assisted Zr-doped VO₂ films fabricated on a glass substrate signified a compact and continuous growth of films with a relative decrease in the phase transition temperature [Fig. 9(e)].⁷⁷ A solution mixture of vanadium(IV)-oxy acetylacetonate, polyvinyl pyrrolidone, and Zr⁴⁺ ions were aged for 24 h to form a transparent uniform sol. The sol was coated on the treated glass substrate and subsequently annealed at 550 °C to obtain Zr-doped VO₂ films. Zirconium doping into VO₂ films induced lattice constant enlargements to the large ionic radii of Zr⁴⁺ ions. Zr-doping did not alter visible transmissivity; however, about a 2 wt. % increment in Zr-doping reduced the critical phase transition temperature to 50 °C.

Polymer assisted deposition (PAD) is similar to sol–gel synthesis where vanadium salts are mixed with a potential polymer to produce aqueous transparent solutions, and then mixed precursor solutions are transformed as VO₂ films by traditional spin or dip coating methods. Porous films with low optical constants and controlled thickness can have a strong influence on the optical properties of VO₂ thin films.^75,78 Kang et al.⁷⁸ demonstrated polyvinylpyrrolidone (PVP)-assisted deposition of nanoporous VO₂ films with low optical constants and enhanced luminous transmittance. PVP mixed aqueous solution of VOCl₂ was spin coated on a fused silica substrate, dried, and annealed at 500 °C to produce an interconnected VO₂ particle nanoporous network [Fig. 9(f)]. The VO₂ films with different thicknesses represented porosity variations, and increased porosity reduced the optical constants. The significant change in optical constants modulated the infrared transmittance properties of VO₂ films with a luminous transmittance of 43.3% and a solar modulation of 14.1%.⁷⁸ Analogously, Cao and group used poly(tetrafluoroethylene) (PTFE) as a self-templated structural material for the preparation of nanoporous VO₂ films.⁷⁹ The PTFE-assisted VO₂ films exhibited excellent optical performance with an ultrahigh luminous transmittance of 78.0% and a solar modulation ability of 14.1%. The developed films provide anti-reflection functionality with potential applications as smart window materials with high solar modulation ability. Laser sintering of VO₂ films processed by the PAD method demonstrated two orders of magnitude of improvement in SMT.⁸⁰ The main feature of this work is a replacement of a common PAD water solvent with propylene glycol (PPG). Replacing water with PPG offered control over the precursor solution viscosity, concentration, and environmental stability. The high viscosity-glycol based precursor solution enables highly reproducible thickness and properties that can be patterned using direct-write techniques, such as laser-based patterning and laser annealing. A precursor with 4:1 EDTA:NH₄VO₃ and polyethylene glycol dissolved in water solvents after water mixing was replaced with PPG by subsequent heating of solution. The resulting viscous precursor was found to be extremely stable, which was then spin coated and laser annealed. The room temperature optical transmittance of laser sintered VO₂ films was measured as 86% and the high temperature transmittance was measured at 42%, corresponding to a total transmittance of 44% [Fig. 9(g)].⁸⁰ These findings conclude that high viscosity precursors offer uniform thickness, and annealing of VO₂ at 500–550 °C yields high-quality films, whereas lowering the sintering temperature results in poor crystallinity and degradation in thermochromic properties. Lee and Kim⁸¹ investigated the effects of organic polymer additives in precursor solutions (VO²⁺ ion) and the physicochemical properties of thermally deposited VO₂ films on sapphire (0001) planes. Organic additive-(sucrose, polyvinyl alcohol [PVA], and polyvinylpyrrolidone [PVP]) added VO₂ films displayed an MIT, but a control, additive free VO₂ film revealed no abrupt resistance change [Fig. 9(h)]. The absence of MIT in the additive free VO₂ films is ascribed to the presence of the oxidized V₂O₅ phase, which may scatter the conduction process and act as a potential barrier for the VO₂ transition. In contrast, VO₂ films without V₂O₅ impurity and an organic additive polymer show phase transition properties with order of reduction potential of the added polymer.

Hydrothermal synthesis is a common solution method in synthesizing VO₂ nanostructures. However, the synthesis of a single VO₂(M) phase is challenging as most hydrothermal products have a mixture of other polymeric forms such as VO₂(B), VO₂(P), VO₂(D), and so on. Nevertheless, several polymeric VO₂ forms were transformed into pure VO₂(M) phase by hydrothermal heating. Hydrothermal heating of vanadium precursors commonly produces the VO₂(B) phase, and subsequent postheat treatment transforms it into the VO₂(M) phase. The paramount feature of hydrothermally transformed VO₂(M) is the retention of the parent phase morphology.⁸² There are only a few reports that claim morphology changes after hydrothermal transformation from VO₂(B) to VO₂(M).⁸³ The Pollet group⁸⁴ described rapid synthesis of the VO₂(B) phase using hydrothermal heating of V₂O₅ and citric acid precursors. The VO₂(B) phase is stable until 400 °C, and further heating above 400 °C resulted in irreversible phase transition into the VO₂(M) phase. In parallel, several studies report the indirect phase transition of other VO₂ (A,^13,85 D,⁸⁶ P,⁸⁷ C,⁹ and hydrate⁸⁸ polymorphs into VO₂(M) of various shapes by changing hydrothermal parameters such as annealing temperature and time.

Guo and co-workers⁸⁹ developed a symmetrically confined growth methodology to fabricate a geometrical pattern of the VO₂ nanonet. Single crystal-like VO₂ nanonet superstructures are realized by a controlled one-step hydrothermal heating. VO₂ films grown on rutile (001)-TiO₂ substrates reflect a uniform orthogonal nanostructure network with an average nano-width wall of 56 nm [Fig. 10(a)]. The interfacial mismatch between the surface energy of (001)-TiO₂ and VO₂ directs the horizontal growth of vanadium particles as a nanonet. The unique VO₂ nanonetwork displayed outstanding thermochromic properties.⁸⁹ The solar modulation ability of the VO₂ nanonet was up to 11.82% with visible light transmittance greater than 70%. Sun et al.⁸⁷ used hydrothermal methods to prepare paramontroseite VO₂(P) nanocrystals, and rapid heating of VO₂(P) nanocrystals at 400 °C for 40 s in an air atmosphere resulted in VO₂(M) phase nanocrystals. VO₂(M) nanocrystals of size 10, 20, and 30 nm were obtained by the direct heating of VO₂(P) phase nanocrystals [Fig. 10(b)]. The single-domain VO₂(M) nanocrystals exhibited single size dependent MIT behavior, and increases in particle size gave an increased MIT temperature but cooling broadened the hysteresis width.⁸⁷ Alternatively, W-doped VO₂(M) nanorods prepared by in situ stirring hydrothermal synthesis reflected improved infrared switching behavior. Mao et al.⁹⁰ proposed an in situ stirring approach for the hydrothermal synthesis of VO₂(M), verified phase composition, morphology, and the MIT temperature of different amounts of W-doped VO₂(M). The morphology of in situ stirred and hydrothermally heated undoped VO₂(M) exhibited thin nanobelt shapes, and W-doped VO₂(M) retained nanobelt or nanorod-like morphology [Fig. 10(c)].⁹⁰ The MIT temperature of W-doped VO₂(M) decreased with increasing the in situ stirring time. The reduction in MIT temperature with a longer stirring time is ascribed to the fine dispersion of W atoms into the VO₂ crystal lattice, which improves the doping efficiency. The MIT temperature of 1.0 at. % W-doped VO₂(M) nanorods was reduced to as low as 24.55 °C/at. %W doping in the range of 0.75–2.0 at. %. The W-doped VO₂(M) films coated on PET substrates represented high light transmittance and flexibility, which are desirable for practical thermochromic applications.⁹⁰ 

Zhou et al.⁹¹ synthesized boron-doped VO₂ powders by hydrothermal heating followed by a subsequent annealing step. A relative doping level of 9.0 at. % was realized for boron-doped VO₂. Boron doping induced reduction in the VO₂ particle size and endorsed uniform distribution and a compact doping effect. The VO₂ samples with a boron doping content of 6.0 at. % represented a fractured nanorod-like morphology of uneven size and length [Fig. 10(d)]. A DSC analysis of boron-doped VO₂ with different boron contents revealed clear variation in the endo- and exothermic peaks on heating and cooling ramps, which is an indication of first order phase transition. Doping induced decreases in VO₂ grain size, promoted reduction in the hysteresis loop, and with a boron doping content of 6.0 at. %, achieved an outstanding MIT temperature as low as 28.1 °C, which was very close to room temperature. The transmittance spectra of 6.0 at. % boron-doped VO₂ presented large differences in thermochromic properties at two different temperatures. The visible light transmittance and the solar modulation ability of the optimal 6.0 at. % boron-doped VO₂ were noted as 54.3% and 12.5%, respectively. Significant research efforts are focused on one-step hydrothermal syntheses of the pure VO₂(M) phase.⁹² More detailed examples are described in the nanoparticle section. However, longer times and higher temperatures during the hydrothermal reaction are needed, which lead to larger VO₂(M) particles than the products from the above-mentioned methods using phase transformation.

Directional growth of VO₂ nanostructures shows lattice deformation with more abrupt and intrinsic MIT than bulk VO₂. Since Morin's first report on the hydrothermal synthesis of 0.1 mm scale VO₂ crystals, the hydrothermal method has made great progress in the synthesis of various VO₂ nanostructures. The advantage of the hydrothermal method is that it provides control over the size, morphology, and phase structure of VO₂. However, pure thermochromic VO₂(M) nanostructures can hardly be produced by the hydrothermal method, due to the presence of multiple polymorphic phases, including VO₂(B), VO₂(A), VO₂(D), and VO₂(P). Moreover, the end product always has traces of VO₂(B) and so on. Recent research has focused more on the hydrothermal synthesis of the pure VO₂(M) phase for thermochromic applications.

In this regard, an indirect hydrothermal synthesis of pure VO₂(M) has been realized by simultaneous VO₂(B) formation and its rapid conversion into the VO₂(M) phase.⁸⁴ Banerjee and group⁸³ presented a stepwise scalable hydrothermal synthesis of free-standing VO₂(M) phase nanowires. The VO₂(M) nanowires were prepared by the initial oxalic acid reduction of V₂O₂ powders into V₃O₇.H₂O nanowires [Fig. 11(a)] and subsequent reduction using a 1:1 (v/v) mixture of water and 2-proponal, followed by postannealing [Fig. 11(b)]. These VO₂(M) nanowires exhibited almost four orders of magnitude change in electrical conductivity across the MIT. Similarly, Hou et al.⁹³ demonstrated a facile synthesis of VO₂(M1) nanorods through a low temperature hydrothermal reaction, followed by annealing for the transformation of VO₂(A) to the VO₂(M) phase. At first, metastable VO₂(A) nanorods [Fig. 11(c)] were synthesized by hydrothermal treatment of the VOSO₄ precursor, and monoclinic VO₂(M) nanorods were obtained by subsequent annealing of VO₂(A) nanorods at 500 °C under an inert atmosphere for 1 h [Fig. 11(d)]. Shi et al. prepared linear gradient MIT behavior, hydrogen doped single-domain VO₂(M) nanowires through hydrothermal synthesis, and hydrogen engineering via postannealing treatment. The intrinsic H-doping concentration in the VO₂(M) nanowires was adjusted by filling ratio and reducing agent concentration. Furthermore, the annealing treatment was shown to eliminate and redistribute dopant/vacancies in the hydric VO₂(M) nanowires.

VO₂ in the form of single crystal nanowires exhibit more abrupt and intrinsic MIT than polycrystalline based films/nanostructures. For various reasons (low cost and ease of processing), other nanostructures are also desired over thin films. A variety of high temperature synthetic approaches have been developed to grow doped VO₂ micro/nanowires on embedded substrates. Axially grading the tungsten (W) doping level in VO₂ nanowires from the tips to the center of the nanowire resulted in axial MIT behavior.³⁹ At high temperatures, the graded W_xV_1−xO₂ nanowires showed a gradual MIT by growth of the metallic phase from the tips into the center of the nanowire. Doping VO₂ with atomic hydrogen was also reported to effect the MIT.⁹⁴ Annealing VO₂ nanobeams in contact with catalytic metals (Au, Pd, Cu, and Ni) under a hydrogen atmosphere showed rapid hydrogen diffusion along the C_R axis.⁹⁴ A coexistence of metallic and insulating domains was observed in free-standing oxygen deficient/rich VO₂ nanobeams.⁹⁵ The deficiency of oxygen stabilized the rutile phase even at a low temperature of 103 K and suppressed phase transition, while oxygen-rich growth conditions stabilized metastable monoclinic M₂ and triclinic T phases. Cheng et al. reported ultra-long, highly dense, free-standing, single crystalline VO₂ nanowires of about 5 mm grown using VTM.⁹⁶ A low-melting point source V₂O₅ and rough quartz substrate facilitated the enhanced VO₂ nucleation and growth rate. Unidirectionally long VO₂ nanowires are typically fabricated by epitaxial growth. It is believed that epitaxial single crystalline VO₂ nanowires on a c-cut sapphire substrate are known to grow laterally on the basal plane along the $⟨1120⟩$ direction [Figs. 11(e) and 11(f)].⁶⁹ In contrast, r-cut and a-cut sapphire substrate VO₂ nanowires were grown out of r and a basal-planes [Fig. 11(g)].⁶⁹ Furthermore, the use of Si(100) and rough surface quartz as substrates in VTM produced highly dense and ultra-long VO₂ micro/nanowires [Figs. 11(h) and 11(i)].^96,97 VO₂ nanowires naturally grow to the [001] plane of a rutile structure, as it is a tetragonal system, and the surface energy of the (001) plane is larger than others.⁹⁸ Nevertheless, epitaxial growth yields long unidirectional VO₂ nanostructures usually clamped on the substrate, which makes postprocessing difficult and limits their further application. In a recent development, well-aligned vertical VO₂ nanowire arrays were grown on Si (001) nanoimprinted substrates. The size of the nanowires was smaller than 1 μm; however, complicated Si substrate preparation impedes the practical use of this process in real time. Recently, Zhao et al.⁹⁹ presented a simple and fast fabrication process for single crystal VO₂ microtube arrays as a thermal oxidation route. The synthesis of VO₂ microtube arrays includes a two-step procedure: first, high temperature heating of vanadium foil in air to approximately 1700 °C in 10 s with the aid of a direct current that produces V₂O₅ and then rapid successive cooling of as-synthesized V₂O₅ to below 250 °C enables the growth of single crystal VO₂ microtube arrays. The single crystal rod shaped VO₂ arrays grew vertically on the layered V₂O₅ substrate [Fig. 11(j)] and the interiors of the VO₂ rods were hollow with uniform rectangular cross sections [Fig. 11(k)]. Optical microscopy observations demonstrated the MIT phenomenon in VO₂ microtube arrays. Insulator-metal transition induced strain-free VO₂ microtubes and an obvious shrinkage was detected in reverse transition due to the lattice difference in both phases [Figs. 11(a) and 11(b)].

In general, vapor deposition and hydrothermal approaches are used in the deposition of VO₂ films, which are normally expensive and complicated. On the other hand, hydrothermal reactions are economical and offer control over morphology and particle size. Unfortunately, postheat treatment after a hydrothermal reaction at elevated temperature results in the agglomeration of VO₂ NPs into large size particles. Therefore, a recent advancement to the conventional hydrothermal process is the one-step hydrothermal synthesis, where pure VO₂(M) powders are obtained without an additional annealing step. Guo and co-workers¹⁰⁰ reported an annealing-free hydrothermal approach to prepare ultrafine VO₂(M) NPs. Hydrogen peroxide was used as the oxidizing and oxygen replenishing agent. A Teflon-lined autoclave was equipped with a quartz vessel and the gap between the quartz vessel and the Teflon-lined autoclave was filled with H₂O₂ solution [Fig. 12(a)]. During the hydrothermal process, H₂O₂ boils and slowly decomposes into water and oxygen, providing a strong oxidizing atmosphere favorable for the formation of VO₂(M) NPs. As obtained, VO₂(M) NPs show desirable dispersity with an average particle size of ∼30 nm [Fig. 12(b)]. These ultra-small VO₂(M) nanoparticles uniformly coated on poly(ethylene terephthalate) (PET) substrates exhibited high flexibility and good visible transmittance, which is propitious for light scattering [Fig. 12(c)] and provides opportunities for thermochromic applications. Likewise, Chen et al.¹⁰¹ developed a one-step hydrothermal heating up process to obtain W-doped VO₂(M) NPs ranging from 25 to 45 nm depending on the hydrothermal holding temperature [Fig. 12(d)]. Highly crystalline VO₂ NPs were dispersed in polyurethane and uniformly coated on PET substrates for flexible films [Fig. 12(e)]. In contrast to pure VO₂ NP based films, W doping into VO₂ NPs largely influenced the light transmittance behavior at 20 and 100 °C. W-doped VO₂ films represent a comparatively low reflectance (below 8%) than pure VO₂ films (above 30%).

The sol–gel method is another common technique to prepare VO₂ NPs. Ding and group⁷⁴ suggested cetyltrimethylammonium vanadate (CTAV) as a precursor for VO₂ NPs via the sol–gel route. Dip coating of the as-prepared CTAV precursors onto a fused silica substrate were followed by air annealing at 500 °C, and subsequent reduction under H₂/Ar gas resulted in VO₂ nanoparticle assembled (NPA) films [Fig. 12(f)]. The sol–gel assisted VO₂ NPA film exhibited enhanced visible light transmittance, and dip coating speed was used to adjust the film optical properties such as visible light transmittance and near-infrared switching efficiency. However, the disadvantage of the dip coating technique is that the deposited VO₂ films are rigid, which impedes their practical application [Fig. 12(g)].

As given in Table I, the main goal of developing VO₂(M) nanostructures is to modulate VO₂'s bulk electronic properties and reduce MIT temperature for near-room-temperature switching applications. The band structure of VO₂ varies with particle size and the underlying MIT behavior depends on the size of the VO₂ particle. The fine control growth of VO₂ particle size can be realized by preparation temperature, annealing time, and rate of VO₂ deposition as thin films. For instance, R → M1 transition temperature or vice versa of single-domain VO₂(M) nanocrystals with 10, 20, and 30 nm of particle size, increased with increasing particle size and a gradual broadening of hysteresis width.⁸⁷ The crystallinity of VO₂ particles was also controlled by adjusting the thickness of the precursor film and annealing time. A negligible hysteresis and a lower MIT temperature of 41 °C has been reported for a dispersed VO₂ particle film with an average particle size of 1.16 μm.¹⁰² The MIT behavior of large VO₂ particles is dominated by compressive stress, while tensile stress and defect density are prevalent in smaller sized particles. Furthermore, VO₂ nanoscale heterostructures integrated with other nanomaterials impart synergistic structure-functional properties, which are different from bulk VO2 or low-dimension VO₂. Nano-heterostructures with distinct size and particle–particle interface changes the MIT behavior of VO₂. The VO₂ nanorods with distinct VO₂–TiO₂–VO₂ interfaces revealed a notable difference in the phase transition temperature.¹⁰³ MIT variation of VO₂–TiO₂–VO₂ interfaces is due to the difference in the size of nanostructure domain interfaces and orientation of VO₂ seed growth. Therefore, the feasibility of incorporating VO₂(M) into hybrid nanoparticles opens several classes of VO₂-hybrid nanomaterials with characteristic switching properties. Hence, the nano-engineering of VO₂ aids in control over crystallinity, phase purity, and the lowering of MIT for integration into small-scale device architectures operational at near room temperature.

Hydrothermal heating induces abrupt changes in the morphology of VO₂ nanostructures and phase formation. Verma et al.¹⁰⁵ analyzed hydrothermal growth parameters to obtain various shapes of VO₂ nanostructures. In particular, increasing the hydrothermal growth time promoted the formation of the VO₂(M) phase with varied morphologies. Adjusting the hydrothermal growth time for 20, 40, 48, and 60 h, and postannealing produced spherical aggregates, platelike microstructures, 3D nanocrystals, and flower shaped VO₂(M) phase particles [Figs. 13(a)–13(d)].¹⁰⁵ In another report, Wu and co-workers¹⁰⁶ prepared VO₂(M) nanospherical assemblies by hydrothermal and calcination of metastable VO₂(B/D) powders at 300–500 °C for 2 h [Fig. 13(e)]. Rectangular plate-like VO₂(M)-SiO₂ structures were synthesized by the cumulative hydrothermal method and postannealing of VO₂(B) phase powder [Figs. 13(f) and 13(g)].¹⁰⁹ As obtained, nanoporous VO₂(M)@SiO₂ exhibited excellent optical performance for smart windows applications [Fig. 13(h)]. A report regarding the effects of precursor aging time on the shape and phase formation of VO₂ nanostructures demonstrated a vanadium oxyacetylacetone precursor assisted synthesis of pure VO₂(M) phase nanofacets with good thermochromic properties [Fig. 13(i)].¹¹⁰ 

Son et al.¹⁰⁷ reported a direct one-step hydrothermal synthesis of VO₂(M) micro- and nanocrystals. For the preparation of nanocrystals, a hydrous precipitate of VO²⁺ solution was realized by mixing optimized amounts of hydrazine and NaOH solution to the VO²⁺ precursor. The fresh amorphous VO²⁺, OH⁻ precipitate was directly applied in the hydrothermal reaction. Under optimal hydrothermal reaction parameters, asterisk-shaped microcrystals, small nanocrystals, and hexagonal crystals were obtained with the pure VO₂(M) phase [Figs. 13(j)–13(l)]. Wu et al.¹⁰⁸ reported a controlled oxidation reaction route for the synthesis of single-domain monoclinic VO₂(M) nanorods using solution-based synthesis. The developed method produced grain boundary free single crystalline VO₂(M) nanorods [Fig. 14(a)] with rapid magnetization change upon passing the phase transition. The first order transition of VO₂(M) to VO₂(R) brings a magnetic susceptibility change upon passing the MIT from the disordered to the ordered magnetic moment. The magnetic susceptibility curves of single-domain VO₂(M) nanorods obtained from solution synthesis exhibit sharper increases near the T_c than the VO₂ samples prepared from other methods [Fig. 14(b)].¹⁰⁸ Further, the differential magnetic susceptibility (∂x/∂T) values of VO₂(M) nanorods were comparatively higher than the solid-state samples. The large magnetic entropy changes of VO₂(M) nanorods near room temperature conjecture the magnetocaloric effects of VO₂ and the potential for magnetic refrigeration applications.

Developing appropriate synthesis methods for VO₂(M/R) is strongly demanded for studying the fundamental physics of the material and exploring commercial applications. A large-scale, low-cost universal method of obtaining the VO₂(M) phase is highly desirable. Industrially compatible protocols to develop free-standing single-domain VO₂(M) micro/nanowires and nanostructure growth remains a significant challenge. Even after decades of studies, gaseous and solution phase reactions mainly used to fabricate VO₂ suffer from by-products or non-stoichiometric VO₂ phases. In this context, the synthesis of desirable VO₂(M/R) films or powders for each study is the first essential step. The synthesis of single phase VO₂(M/R) with more uniform and smaller sizes, better dispersity, perfect stoichiometry, and higher crystallinity is still a task that remains to be achieved.

As discussed above, VO₂ has been a breakthrough material with strong electron–electron correlation; yet the real time use of VO₂ as a smart material requires the solving of several challenges.

1. First, following significant progress in VO₂ MIT, the exact mechanism of MIT still needs a thorough explanation. Purely Mott or Peierls scenarios are insufficient to elucidate the origin of MIT. Investigations to bridge Mott and Peierls theory can provide more insights into the MIT origin. The stimulant response of VO₂ to ultrafast optical fields, electrostatic fields, THz pulses, or stress can complement the current understanding of the MIT mechanism.

2. Second, the exploration of MIT in VO₂ is limited to certain (transistors, actuators, smart switches, and sensors) applications. The use of thermochromic VO₂ smart windows is gaining ground; yet, practical difficulties in uniform large area VO₂ coatings need to be precisely addressed. A balance between T_c, transmittance, and solar modulation ability can be extended with structure engineering, controlled porosity, and atomic level doping. Even though VO₂ serves as an energy-saving material, practical VO₂-based devices are relatively lacking. Accelerated research into the application of VO₂ to meet real-world demands can expand its scientific merit.

3. The modus operandi of exploring VO₂ for more commercial and industrially useful applications remains in combined simulations and experimental investigations. The coupling of VO₂ with other functional materials can alter its physicochemical properties and the developed hybrid VO₂ composites have potentially superior performance.

Given these problems, we hope that this Tutorial instigates the demand for novel approaches and ideas to accelerate the current understanding of VO₂ research and provide support to other inorganic materials.

This research was supported by the National Research Foundation of Korea (NRF) funded by the Ministry of Science, ICT (No. NRF- 2019R1A2C1084010).