An Overview of Chromic Transition Metal Oxide Thin Films

Abstract

Transition metal oxides constitute an important materials platform for chromic phenomena because their optical response is strongly coupled to the changes in electronic structure, phase state, carrier concentration, and defect chemistry. This review discusses selected transition metal oxide thin films, with emphasis on VO₂ and other vanadium oxides, WO₃, NiO, and TiO₂. The review summarizes the structural and electronic characteristics of these representative oxide systems and highlights the role of phase composition, crystal structure, oxygen non-stoichiometry, and defect chemistry in determining their optical response. The main thin film preparation routes, including pulsed laser deposition, magnetron sputtering, sol–gel and aerosol spray methods, atomic layer deposition, chemical vapor deposition, electrochemical routes, and molecular beam epitaxy, are reviewed with respect their influence on obtained thin films. Particular attention is given to applications in thermochromic VO₂-and electrochromic WO₃/NiO-based smart windows, and transition metal oxide-based gasochromic hydrogen sensors. Key challenges related to transition temperature tuning, luminous transmittance, solar modulation, optical contrast, cycling stability, ion transport and large-area integration are also discussed. Overall this review provides a comparative overview of selected transition metal oxide thin films by connecting material chemistry and physics, thin film preparation technology and functionality.

1. Introduction

Smart materials are materials capable of undergoing reversible and controllable changes in their physical or chemical properties under external stimuli such as temperature, light, electric field, mechanical stress, gas atmosphere, or chemical environment [1,2,3]. Unlike conventional materials, which are usually regarded as passive static systems with fixed functional parameters, smart materials behave as dynamic systems whose response can adapt to changing operating conditions. Their ability to modulate electrical, optical, magnetic, mechanical, or thermal properties makes them relevant to the transition from static material platforms toward responsive and adaptive systems.

Within this broader group, materials with adaptive optical response form a particularly important subgroup, since external stimuli can directly modify optical transmittance, reflectance, absorption, apparent color, or thermal emissivity. Such control over light–matter interaction is especially valuable because optical response is immediately linked to visual perception, daylight and solar energy management, thermal regulation, and signal readout. From a technological point of view, these changes can be readily monitored visually or spectroscopically, which facilitates their use in coatings, smart windows (SWs), sensors, and other optoelectronic devices [4,5,6,7,8]. Among these materials, chromic systems are of particular interest because they reversibly modulate the spectral response and/or visible appearance under external stimuli.

Chromic materials are actively investigated for smart and energy-efficient windows, building-integrated thermal-management coatings, optical and gas-sensing systems, visual indicators, adaptive camouflage, information-storage and display-related concepts, and thin film optoelectronic devices [5,8,9,10,11].

Transition metal oxides (TMOs) constitute a broad class of inorganic materials with a remarkable diversity of crystal structures and of structural, electronic, and optical properties [12]. Depending on composition, bonding geometry, and correlation strength, TMOs can behave as both band insulators (or semiconductors) and correlated metal. In addition, many TMOs exhibit metal-insulator (MIT) phase transitions, magnetism, multiferroicity, and other emergent phenomena [13]. Owing to this versatility, TMOs have become an important functional platform for electronics, optoelectronics, sensing, and energy-related technologies [14,15,16]. For chromic functionality, TMOs are especially attractive because their optical response is strongly coupled to their electronic and structural state [13]. Changes in carrier concentration, oxidation state, defect chemistry, and phase state appreciably transform the electronic structure and thereby modify optical properties [13,17]. As a result, external stimuli can induce pronounced and often reversible optical changes, making TMOs a particularly favorable platform for chromism [6,8] (Figure 1).

Although chromic behavior can also be investigated in bulk materials, thin films are the most relevant form for practical optical applications. Film geometry offers direct control over thickness, composition and stoichiometry, crystallinity, surface and interface structure, and defect concentration through deposition and post-treatment conditions, while also enabling multilayer architectures and large-area coatings [18]. For this reason, thin film TMOs dominate chromism applications. Accordingly, the emphasis of the present review is placed on selected chromic TMO thin films. The literature on individual chromic TMOs has expanded rapidly, whereas discussions of electronic/structural properties, preparation routes, and device integration often remain fragmented across separate material systems. This creates a clear need for a review that connects these subjects.

Accordingly, the present review aims to: summarize the key structural, electronic, and optical features of representative chromic TMOs; review the principal technological routes used to obtain selected TMO thin films and highlight the main application areas of chromic oxide thin films.

At the same time, this review does not seek to provide an exhaustive survey of all oxide materials, synthesis approaches, or device architectures. Rather, its focus is placed on chromism-related structural, electronic, and optical aspects of selected TMO systems. The material scope of this review is focused on selected TMOs thin films, namely VO₂ and other vanadium oxides, WO₃, NiO, and TiO₂. These materials were chosen as representative systems because they are among the most intensively studied and most frequently discussed chromic TMOs in the literature. In addition, they reflect distinct classes of chromic behavior and different physical routes to optical modulation, thereby providing a sufficiently broad yet coherent basis for comparative discussion. The selected oxides therefore do not represent an exhaustive list of chromic TMOs, but rather a set of benchmark material platforms through which the principal features of chromism in oxide thin films can be analyzed.

The article is organized into several major sections (Figure 2), each corresponding to a different level of analysis: the material platform section examines the structural and electronic proprieties of the selected TMOs; the technology section reviews the main preparation routes for TMO thin films; the applications section outlines the principal directions in which chromic properties of TMO thin films are utilized.

2. Materials Platform: Structural and Electronic Properties of Selected TMOs

2.1. Vanadium Oxides

2.1.1. V–O Binary Phase Diagram

The V–O phase diagram [19] reflects the thermodynamic equilibrium of phases over a wide temperature range and illustrates the complex phase relations that arise from the ability of vanadium to form oxides with different stoichiometries, oxidation states, and structural modifications. Despite the large number of phases shown in the diagram, the number of thermodynamically stable stoichiometric compounds is limited, whereas a significant fraction of the observed phases corresponds to solid-solution regions or metastable states. A characteristic feature of the V–O phase diagram is the presence of extended non-stoichiometric regions in both the V-rich and O-rich parts of the diagram. This behavior demonstrates the important role of oxygen content as a controlling parameter for phase stability in the V–O system.

In the V-rich region of the V–O phase diagram, four oxygen-containing solid solutions can be distinguished: $\alpha$-, $\beta$-, $\gamma$-, and $\delta$-V phases [20]. In the oxide region, five stoichiometric vanadium oxides are thermodynamically stable: V₂O₃ with a corundum-type trigonal structure ($R\overline{3}c$, No. 167), V₃O₅ with a monoclinic structure ($P2/c$, No. 13), high-temperature rutile-type VO₂ with a tetragonal structure ($P{4}_2/mnm$, No. 136), V₃${\mathrm{O}}_7$ with a monoclinic structure ($C2/c$, No. 15), and V₂O₅ with an orthorhombic structure ($Pmmn$, No. 59).

In addition to these stable stoichiometric compounds, the V–O system contains metastable and non-stoichiometric oxides that are typically formed under non-equilibrium, oxygen-deficient, or oxygen-rich conditions. Oxygen-deficient compositions are associated with mixed-valence Magnéli phases compositionally located between V₂O₃ and VO₂, whereas oxygen-rich compositions are associated with Wadsley phases located between VO₂ and V₂O₅. These homologous series describe how the V–O system accommodates oxygen deficiency or excess and can be represented as formal combinations of neighboring stable oxides.

The Magnéli phases are mixed-valence systems in which V³⁺ and V⁴⁺ ions coexist, and the average oxidation state of vanadium increases with increasing n. Their key structural feature is the presence of crystallographic shear planes, which arise from ordered shear transformations of the rutile-like VO₆ octahedral framework. The resulting disruptions of the VO₆ octahedral chains are accommodated by collective shifts of the octahedra, forming periodic shear planes whose spacing is determined by the parameter n:

$${\mathrm{V}}_n{\mathrm{O}}_{2n-1}={\mathrm{V}}_2{\mathrm{O}}_3+(n-2){\mathrm{VO}}_2,3\le n\le 9.$$

(1)

The Wadsley phases occupy the oxygen-rich side of the V–O phase diagram and can be represented as linear combinations of VO₂ and V₂O₅:

$${\mathrm{V}}_n{\mathrm{O}}_{2n+1}={\mathrm{V}}_2{\mathrm{O}}_5+(n-2){\mathrm{VO}}_2,n\ge 3.$$

(2)

The Wadsley phases are likewise mixed-valence systems involving V⁴⁺ and V⁵⁺ ions. Unlike Magnéli phases, in which oxygen deficiency is accommodated by crystallographic shear planes, excess oxygen in Wadsley phases is accommodated through rearrangements of the octahedral framework and the formation of more complex layered or block-like structures. Their periodicity and connectivity are governed by the structural parameter n.

2.1.2. Crystal Structure of Vanadium Dioxide VO₂

Below the MIT temperature, $T_c\approx 68°\mathrm{C}$, VO₂ exhibits insulating behavior and adopts a monoclinic structure. The M1 phase is the thermodynamically stable low-temperature phase of stoichiometric VO₂, whereas the M2 phase is metastable and can be stabilized by applied stress, chemical doping, or epitaxial strain.

The M1 phase has the monoclinic space group $P{2}_1/c$ (No. 14) and represents the stable low-temperature stoichiometric phase under ambient pressure. Its lattice parameters are $a_{\mathrm{M}1}=5.751\mathsf{\AA }$, $b_{\mathrm{M}1}=4.537\mathsf{\AA }$, $c_{\mathrm{M}1}=5.382\mathsf{\AA }$, $\alpha =\gamma ={90}^{\circ }$, and ${\beta }_{\mathrm{M}1}=122.{64}^{\circ }$ [21].

Structurally, the M1 phase is characterized by complete V–V dimerization along chains derived from the rutile $c_{\mathrm{R}}$ direction. Each vanadium atom participates in the formation of V–V dimers, leading to a doubling of the unit cell and a lowering of the lattice symmetry relative to the high-temperature rutile phase. In addition to dimerization, the vanadium atoms undergo lateral displacements accompanied by rotations of the surrounding VO₆ octahedra. These distortions break the equivalence of oxygen sites and result in two crystallographically distinct oxygen positions. The M1 structure therefore reflects a cooperative lattice distortion involving both V–V pairing and octahedral tilting, which plays a central role in stabilizing the insulating state of VO₂ below $T_c$.

The M2 phase is a metastable monoclinic phase with the space group $C2/m$ (No. 12). Its lattice parameters are $a_{\mathrm{M}2}=9.066\mathsf{\AA }$, $b_{\mathrm{M}2}=5.797\mathsf{\AA }$, $c_{\mathrm{M}2}=4.525\mathsf{\AA }$, $\alpha =\gamma ={90}^{\circ }$, and ${\beta }_{\mathrm{M}2}=91.{88}^{\circ }$ [22].

Unlike the M1 phase, the M2 phase is not generally observed in bulk stoichiometric VO₂ under ambient conditions but can be stabilized by chemical doping, applied stress, or epitaxial strain [22,23]. A defining feature of the M2 phase is the inequivalent behavior of vanadium chains. In this structure, only half of the vanadium chains exhibit V–V dimerization, while the remaining chains retain a non-dimerized configuration.

In the M1 phase, vanadium atoms and the two crystallographically distinct oxygen atoms occupy the Wyckoff position 4e. The corresponding symmetry-equivalent positions are given by $\pm (\mathrm{x};\mathrm{y};\mathrm{z})$ and $\pm (\mathrm{x};1/2-\mathrm{y};1/2+\mathrm{z})$. The fractional atomic coordinates are summarized in

Table 1. VO₂ M1 atomic parameters [21].

Atom	Wyckoff	Site Symmetry	x/a	y/b	z/c
V1	4e	1	0.23947	0.97894	0.02646
O2	4e	1	0.10616	0.21185	0.20859
O3	4e	1	0.40051	0.70258	0.29884

. Based on these atomic parameters, the crystal structure of VO₂ is shown in Figure 3a.

All crystal structure figures were generated using Diamond 4 software, based on crystallographic data reported in the corresponding cited references.

Above $T_c$, VO₂ transforms into a metallic phase with the rutile crystal structure. This high-temperature phase crystallizes in the tetragonal space group $P{4}_2/mnm$ (No. 136). Its lattice parameters are $a_{\mathrm{R}}=b_{\mathrm{R}}=4.556\mathsf{\AA }$, $c_{\mathrm{R}}=2.860\mathsf{\AA }$, and $\alpha =\beta =\gamma ={90}^{\circ }$ [22] (

Table 2. VO₂ rutile atomic parameters [22].

Atom	Wyckoff	Site Symmetry	x/a	y/b	z/c
V1	2a	m.mm	0	0	0
O2	4f	m.2m	0.305	0.305	0

, Figure 3b).

Across the MIT at $T_c$, neighboring vanadium atoms within each chain shift toward one another, leading to V–V pairing, while the surrounding VO₆ octahedra undergo accompanying rotations. This cooperative distortion produces a quasi-one-dimensional Peierls-like instability with contributions from electron correlations. It breaks the uniformity of the V chains, creates two distinct oxygen sites, and lowers the lattice symmetry from the rutile R phase to the monoclinic M1 phase.

The V–V distances become alternating, with a dimer bond length of $\sim 2.62\mathsf{\AA }$ and an interdimer separation of $\sim 3.12\mathsf{\AA }$. Upon heating, this distortion is reversed and the structure transforms back into the rutile phase [24] (Figure 4); this process is reversible.

2.1.3. Crystal Structure of Divanadium Trioxide V₂O₃

V₂O₃ is a strongly correlated TMO that exhibits an MIT accompanied by a structural phase transformation at $T_c\approx 155\mathrm{K}$. At higher temperatures, it adopts a corundum-type structure $R\overline{3}c$ (No. 167) with lattice parameters $a=b=4.951\mathsf{\AA }$, $c=14.003\mathsf{\AA }$, $\alpha =\beta ={90}^{\circ }$, and $\gamma ={120}^{\circ }$ [25] (

Table 3. V₂O₃ corundum structure atomic parameters [25].

Atom	Wyckoff	x/a	y/b	z/c
V1	12c	0	0	0.34629
O2	18e	0.3118	0	0.25

, Figure 5a).

In this phase, all vanadium atoms are crystallographically equivalent and are octahedrally coordinated by oxygen, forming a three-dimensional lattice of edge-sharing VO₆ octahedra. This high-symmetry structure corresponds to the metallic state of V₂O₃.

Upon cooling below $T_c$, V₂O₃ undergoes a structural transformation into a monoclinic insulating phase with space group $I2/a$ (No. 15) and lattice parameters $a=7.274\mathsf{\AA }$, $b=5.005\mathsf{\AA }$, $c=5.551\mathsf{\AA }$, $\alpha =\gamma ={90}^{\circ }$, and $\beta =96.{78}^{\circ }$ [26] (

Table 4. V₂O₃ monoclinic structure atomic parameters [26].

Atom	Wyckoff	x/a	y/b	z/c
V1	8f	0.3445	0.0025	0.3001
O1	8f	0.4043	0.8493	0.6459
O2	18e	0.25	0.3166	0.5

, Figure 5b). The monoclinic distortion lifts the equivalence of the vanadium sites and introduces subtle rearrangements of the VO₆ octahedra, which are closely associated with the onset of electronic localization in the insulating state.

2.1.4. Crystal Structure of Divanadium Pentoxide V₂O₅

V₂O₅ represents the most oxygen-rich compound and the highest oxidation state phase in the V–O system, with vanadium in the formal V⁵⁺ ($d^0$) electronic configuration. In contrast to VO₂ and V₂O₃, V₂O₅ does not exhibit an MIT and remains insulating over a wide temperature range.

V₂O₅ crystallizes in an orthorhombic structure with space group $Pmmn$ (No. 59) and lattice parameters $a=11.512\mathsf{\AA }$, $b=3.564\mathsf{\AA }$, $c=4.368\mathsf{\AA }$, $\alpha =\beta =\gamma ={90}^{\circ }$ [27] (

Table 5. V₂O₅ atomic parameters [27].

Atom	Wyckoff	x/a	y/b	z/c
V	4f	0.10118	0.25	0.8917
O1	4f	0.1043	0.25	0.531
O2	4f	−0.0689	0.25	0.003
O3	2a	0.25	0.25	0.001

, Figure 6). The crystal structure is highly anisotropic and consists of layers formed by distorted VO₅ square pyramids sharing edges and corners.

2.1.5. Electronic Structure of Vanadium Dioxide VO₂

Vanadium dioxide (VO₂) is a correlated oxide in which the electronic structure undergoes a pronounced reconstruction across the coupled structural–electronic MIT [28,29,30]. In VO₂, vanadium is formally in the V⁴⁺ state with a $3d^1$ configuration; therefore, the electronic states near the Fermi level are governed primarily by V-$3d$ states hybridized with O-$2p$ states [24,31].

In an approximately octahedral VO₆ environment, the five V-$3d$ orbitals are split by the crystal field into a low-energy $t_{2g}$ manifold ($d_{xy}$, $d_{xz}$, $d_{yz}$) and a high-energy $e_g$ manifold ($d_{z^2}$, $d_{x^2-y^2}$), with a characteristic splitting energy ${\Delta }_{\mathrm{O}}$ (occasionally denoted $10Dq$) (Figure 7a). Since the system hosts a single d electron, the $t_{2g}$ states primarily govern the electrical conductivity and the low-energy optical response. The actual local symmetry in VO₂ deviates from the ideal $O_h$ symmetry; therefore, the $t_{2g}$ manifold further splits into sublevels: $a_{1g}$, a singlet often denoted $d_{\Vert }$, with the largest projection along the rutile axis $c_R$ and strong overlap along the V–V chains; and $e_g^{\pi }$, a doublet consisting of two orbital components with a smaller projection along $c_R$ and predominantly $\pi$-type overlap with O-$2p$ states (p–d$\pi$ hybridization) [24,31,32] (Figure 7b).

In the high-temperature rutile phase, V atoms form quasi-linear chains along the $c_R$ axis. The $a_{1g}$ orbital exhibits the largest overlap along the V–V chains and gives rise to a relatively dispersive band, whereas the two $e_g^{\pi }$ components hybridize more strongly with the oxygen $2p$ states via $\pi$-channels. As a result, the $t_{2g}$ bands are partially filled, the Fermi level crosses the V–$3d$ states, and the system exhibits metallic behavior. Importantly, the metallic state of ${\mathrm{VO}}_2$ is correlated: the effective masses and spectral characteristics of quasiparticles differ markedly from the predictions of a simple single-electron picture due to substantial Coulomb correlations [29,30] (Figure 8). Upon lowering the temperature, VO₂ transforms into the monoclinic M1 phase, accompanied by two key lattice distortions: dimerization of V atoms along the chains and tilting of the VO₆ octahedra. Dimerization splits the $a_{1g}$ band into bonding and antibonding components due to the enhanced V–V overlap along $c_R$. Simultaneously, octahedral tilting modifies the $p-d$ hybridization and the energetic position of the $e_g^{\pi }$ states, typically shifting them upward relative to the Fermi level. As a result, the single d-electron occupies the lower $a_{1g}$ subband, while the upper $a_{1g}$ component and the $e_g^{\pi }$ states lie at higher energies, leading to the opening of a band gap and insulating behavior [24,29,31,32] (Figure 8).

2.1.6. Electronic Structure of Divanadium Trioxide V₂O₃

Unlike VO₂, where the MIT is strongly intertwined with a pronounced dimerization-driven band rearrangement, the MIT in V₂O₃ is commonly described as a Mott–Hubbard-type transition [33]. In stoichiometric V₂O₃, vanadium is formally in the V³⁺ state with a $3d^2$ configuration. Vanadium ions are coordinated by oxygen in an approximately octahedral VO₆ environment, so the V-$3d$ states are split by the crystal field into a low-energy $t_{2g}$ manifold and a high-energy $e_g$ manifold, separated by ${\Delta }_{\mathrm{O}}$. Because the relevant d electrons primarily occupy the $t_{2g}$ region, the low-energy electronic structure near the Fermi level is dominated by $t_{2g}$-derived states hybridized with O-$2p$ states.

In the corundum structure, the local symmetry deviates from ideal $O_h$ symmetry due to a trigonal distortion of the VO₆ octahedra, which further splits the $t_{2g}$ manifold into an $a_{1g}$ singlet and an $e_g^{\pi }$ doublet. This trigonal splitting and the resulting orbital polarization ($a_{1g}$ vs. $e_g^{\pi }$) play an important role in shaping the low-energy spectrum and the MIT [34,35,36].

In the paramagnetic metallic (PM) regime, realized, for example, at elevated temperatures or under conditions that effectively increase the electronic bandwidth W, the $t_{2g}$-derived bands are partially filled and the Fermi level crosses V-$3d$ states, yielding metallic transport. However, the metallic phase is strongly correlated: quasiparticle bands are substantially renormalized, with reduced bandwidth and enhanced effective masses, and the spectral function typically exhibits reduced coherent weight near $E_F$, accompanied by incoherent features at higher binding energies associated with Hubbard physics [37,38].

Upon entering an insulating regime, e.g., the antiferromagnetic insulating (AFI) phase at lower temperatures for stoichiometric material or a paramagnetic insulating state under specific tuning, the coherent spectral weight at the Fermi level is strongly suppressed and a gap opens. In the Mott–Hubbard representation, this corresponds to a redistribution of spectral weight from a coherent quasiparticle peak into lower and upper Hubbard bands, driven by the dominance of the on-site Coulomb repulsion U over the effective bandwidth W. The trigonal crystal-field splitting and orbital occupancy of the $a_{1g}$ and $e_g^{\pi }$ submanifolds can influence the detailed character of the transition, including the degree of orbital selectivity and the magnitude of the gap [33,36,37,38].

2.1.7. Electronic Structure of Divanadium Pentoxide V₂O₅

Stoichiometric V₂O₅ is an insulating, charge-transfer (CT) TMO. Vanadium in V₂O₅ is formally in the V⁵⁺ oxidation state with a $3d^0$ electronic configuration, i.e., the V-$3d$ shell is empty and there are no intrinsic carriers in a partially filled d band. The valence-band maximum is dominated by O-$2p$ states, whereas the conduction-band minimum has predominantly V-$3d$ character with substantial O-$2p$/V-$3d$ hybridization; therefore, the Fermi level lies inside the band gap [39,40,41]. The stable ambient-pressure phase of V₂O₅ is orthorhombic and layered [27,42]. The V-$3d$ states may still be approximately separated into $t_{2g}$-like and $e_g$-like groups. However, the low local symmetry and pronounced structural distortions substantially lift the degeneracy and modify the orbital energies compared with an ideal $O_h$ octahedron [39,42]. Unlike VO₂ and V₂O₃, V₂O₅ is not characterized by a temperature-driven MIT under ambient conditions [43].

2.2. Nickel Oxide

2.2.1. Ni–O Binary Phase Diagram

In the Ni–O system [44,45], nickel oxide $\mathrm{NiO}$ is the only oxide that remains thermodynamically stable over a wide range of temperatures and oxygen partial pressures, whereas other nickel oxides appear only under restricted conditions and are generally metastable, such as Ni₃O₄, Ni₃O₃ and NiO₂ [46]. Although the phase diagram indicates a strict Ni:O ratio of 1:1, in practice NiO almost always exists as a nickel-deficient oxide of the form ${\mathrm{Ni}}_{1-x}$O ($x\approx {10}^{-3}-{10}^{-1}$). Charge compensation in these non-stoichiometric states occurs through the formation of nickel vacancies, the presence of Ni³⁺ ions [47], and the emergence of localized hole polarons, which govern the p-type conductivity and significantly affect the optical properties of the material.

2.2.2. Crystal Structure of Nickel Oxide

Stoichiometric NiO crystallizes in the rock-salt structure, belonging to the space group $Fm\overline{3}m$ (No. 225). The lattice parameter of the cubic unit cell is $a=4.177\mathsf{\AA }$ [48] (

Table 6. NiO atomic parameters [48].

Atom	Wyckoff	x/a	y/b	z/c
Ni1	4a	0	0	0
O2	4b	1/2	1/2	1/2

, Figure 9). Each ${\mathrm{Ni}}^{2+}$ ion is octahedrally coordinated by six ${\mathrm{O}}^{2-}$ ions, forming NiO₆ octahedra, while each oxygen atom is similarly coordinated by six nickel ions.

2.2.3. Electronic Structure of Nickel Oxide

NiO is a prototypical strongly correlated TMO and a CT insulator. In stoichiometric NiO, nickel is formally ${\mathrm{Ni}}^{2+}$ with a $3d^8$ configuration. In an approximately octahedral NiO₆ environment, the Ni-$3d$ states can be described by crystal-field splitting into $t_{2g}$ and $e_g$ components. However, the insulating gap is not captured by a simple one-electron crystal-field picture and instead reflects substantial electron–electron correlations and strong $p-d$ hybridization [49]. The valence states near the band edge contain a significant O-$2p$ contribution hybridized with Ni-$3d$, while the unoccupied states have strong Ni-$3d$ character [50] (Figure 10).

2.3. Tungsten Oxides

2.3.1. W–O Binary Phase Diagram

The W–O system [51] exhibits one of the most complex phase diagrams among TMOs, characterized by a large number of reported oxide phases spanning a narrow range of oxygen concentrations. This apparent complexity does not arise from the stabilization of numerous independent stoichiometric compounds, but rather from the formation of extended families of structurally related oxygen-deficient and oxygen-rich phases derived from the WO₃ framework.

In the O-rich region of the W–O phase diagram, the thermodynamic stability is restricted to a limited number of oxide phases. The only fully stoichiometric and widely stable oxides are WO₂ and WO₃, which define the lower and upper bounds of oxygen content in this region. Between these two phases, a series of intermediate oxygen-deficient tungsten oxides is stabilized. These phases represent extended homologous series derived from the WO₃ framework via crystallographic shear mechanisms.

2.3.2. Crystal Structure of Tungsten Oxide WO₃

All polymorphs of WO₃ can be viewed as distortions of the cubic ${\mathrm{ReO}}_3$-type structure $Pm\overline{3}m$ (No. 221) (

Table 7. Ideal cubic WO₃ structure atomic parameters.

Atom	Wyckoff	x/a	y/b	z/c
W1	1a	0	0	0
O2	3b	1/2	0	0

, Figure 11a), which consists of a three-dimensional framework of corner-sharing WO₆ octahedra (Figure 11b).

The ideal cubic WO₃ structure is metastable and can be stabilized only under non-equilibrium conditions, such as chemical substitution or lattice constraint effects, since the A-site of the perovskite lattice remains vacant. Structural polymorphism in WO₃ arises primarily from cooperative tilting of WO₆ octahedra and off-centering of W atoms within the octahedral units [52]. As a result, WO₃ undergoes a sequence of symmetry-lowering phase transitions with decreasing temperature, exhibiting several stable crystallographic phases over the temperature range from 0 to 1170 K (

Table 8. ${\mathrm{WO}}_3$ stable phases.

Phase	Structure	Lattice Parameters	Stability Temperature Range
$\alpha$–WO3 [53]	Tetragonal $P4/nmm$	$a=5.250\mathsf{\AA }$	1010–$1170\mathrm{K}$
	(No. 129)	$c=3.910\mathsf{\AA }$
$\beta$–WO3 [54]	Orthorhombic $Pnma$	$a=7.340\mathsf{\AA }$	600–$1010\mathrm{K}$
	(No. 62)	$b=7.570\mathsf{\AA }$
		$c=7.750\mathsf{\AA }$
$\gamma$–WO3 [8]	Monoclinic $P{2}_1/n$	$a=7.300\mathsf{\AA }\beta =90.{88}^{\circ }$	290–$600\mathrm{K}$
	(No. 14)	$b=7.540\mathsf{\AA }$
		$c=7.690\mathsf{\AA }$
$\delta$–WO3 [55]	Triclinic $P\overline{1}$	$a=7.300\mathsf{\AA }\alpha =88.{81}^{\circ }$	230–$290\mathrm{K}$
	(No. 2)	$b=7.510\mathsf{\AA }\beta =90.{985}^{\circ }$
		$c=7.681\mathsf{\AA }\gamma =90.{985}^{\circ }$
$ϵ$–WO3 [56]	Monoclinic $Pc$	$a=5.277\mathsf{\AA }\beta =91.{76}^{\circ }$	<230 K
	(No. 7)	$b=5.155\mathsf{\AA }$
		$c=7.663\mathsf{\AA }$
H-WO3 [57]	Hexagonal $P6/mmm$	$a=7.298\mathsf{\AA }$	metastable
	(No. 191)	$c=7.798\mathsf{\AA }$

).

2.3.3. Electronic Structure of Tungsten Oxide WO₃

Tungsten trioxide WO₃ in stoichiometric form is a $d^0$ oxide: tungsten is formally in W⁶⁺ oxidation state with a $5d^0$ electronic configuration, so the conduction band is derived primarily from empty W-$5d$ states, whereas the valence band is dominated mainly by O-$2p$ states with significant hybridization [58,59]. In an octahedral environment, the low-lying W-d conduction states are often predominantly $t_{2g}$-like, but real structures exhibit sizable distortions that lift degeneracies and modify the band dispersion [60]. WO₃ exhibits multiple temperature-dependent polymorphs, yet it is generally treated as a semiconductor/insulator in its stoichiometric forms; the key changes are typically in lattice symmetry and quantitative band-edge positions rather than a temperature-driven MIT as in ${\mathrm{VO}}_2$ [58,60].

2.4. Titanium Oxides

2.4.1. Ti–O Binary Phase Diagram

The Ti–O system can be regarded as an intermediate case among TMOs, combining a relatively high solubility of oxygen in metallic titanium with the formation of numerous non-stoichiometric and shear-structure oxides at increasing oxygen content [61,62].

At intermediate oxygen concentrations, the Ti–O system is dominated by titanium suboxides based on TiO and ${\mathrm{Ti}}_{1-x}$O compositions. The nominal TiO phase crystallizes in a NaCl-type B1 structure and exhibits an exceptionally wide range of non-stoichiometry, which is accommodated by the simultaneous presence of vacancies on both the titanium and oxygen sublattices. As a result, TiO cannot be regarded as a strictly stoichiometric compound, but rather as a defect-stabilized phase with variable composition. Depending on temperature and oxygen chemical potential, TiO-based phases may undergo vacancy ordering, giving rise to lower-symmetry structures such as $\alpha$-TiO and $\beta$-TiO. These suboxides typically exhibit metallic or semimetallic conductivity and represent a transitional regime between metallic solid solutions and oxygen-rich shear-structure oxides. In addition to the nominal TiO composition, the Ti–O system contains several titanium suboxides in the Ti-rich and intermediate oxygen concentration range. Early suboxides such as Ti₃O, Ti₂O, and Ti₃O₂ are typically interpreted as oxygen-ordered derivatives of metallic titanium rather than distinct stoichiometric oxides.

In the O-rich region of the Ti–O system, phase stability is governed by stoichiometric titanium dioxide TiO₂ and a series of ordered oxygen-deficient Magnéli phases. The O-rich phases are stabilized by well-defined crystallographic rearrangements of the TiO₆ octahedral framework.

TiO₂ represents the fully oxidized and strictly stoichiometric endpoint of the Ti–O system. It is thermodynamically stable over a wide range of temperatures and oxygen partial pressures and crystallizes in several polymorphic modifications, with the rutile structure being the most stable under ambient conditions. All Ti sites in TiO₂ are occupied by Ti⁴⁺ ions in octahedral coordination, and large deviations from stoichiometry are energetically unfavorable.

Between TiO and TiO₂, a series of Magnéli phases with the general formula ${\mathrm{Ti}}_n$${\mathrm{O}}_{2n-1}$ forms a distinct homologous series. These compounds are thermodynamically stable phases rather than defect variants of TiO₂ and are characterized by the presence of periodic crystallographic shear planes within a rutile-derived structure. The introduction of shear planes accommodates oxygen deficiency without the formation of isolated point defects, resulting in well-defined compositions and long-range structural order. The Magnéli phases exhibit mixed-valence titanium states, Ti³⁺/Ti⁴⁺, and display electronic properties ranging from semiconducting to metallic, reflecting the progressive reduction of TiO₂. Structurally and electronically, they constitute a transitional regime between corundum-type Ti₂O₃ and rutile TiO₂, playing a central role in the O-rich phase chemistry of titanium oxides.

2.4.2. Crystal Structure of Titanium Oxide ${\mathrm{TiO}}_2$

TiO₂ crystallizes in three main polymorphic phases: anatase, rutile, and brookite. Anatase adopts a tetragonal structure with space group $I{4}_1/amd$ (No. 141) and lattice parameters $a=b=3.784\mathsf{\AA }$ and $c=9.514\mathsf{\AA }$ [63] (

Table 9. Anatase ${\mathrm{TiO}}_2$ atomic parameters [63].

Atom	Wyckoff	x/a	y/b	z/c
Ti1	4a	0	0	0
O2	8eb	0	0	0.20806

, Figure 12a). Rutile crystallizes in the tetragonal space group $P{4}_2/mnm$ (No. 136) with lattice parameters $a=b=4.594\mathsf{\AA }$ and $c=2.959\mathsf{\AA }$ [64] (

Table 10. Rutile ${\mathrm{TiO}}_2$ atomic parameters [64].

Atom	Wyckoff	x/a	y/b	z/c
Ti1	2a	0	0	0
O2	4f	0.306	0.306	0

, Figure 12b), while brookite exhibits an orthorhombic structure belonging to space group $Pbca$ (No. 61) [65,66]. Among these polymorphs, rutile is the thermodynamically stable phase under ambient conditions, whereas anatase and brookite are metastable and can be stabilized depending on synthesis conditions, particle size, and strain.

All polymorphs of TiO₂ are constructed from TiO₆ octahedra, in which Ti⁴⁺ ions are octahedrally coordinated by six ${\mathrm{O}}^{2-}$ ions. The different crystal structures arise from variations in the connectivity, distortion, and tilting of these octahedra. In the rutile structure, TiO₆ octahedra share edges along the crystallographic c-axis and corners in the perpendicular directions, forming a relatively dense and compact framework. In contrast, anatase exhibits a more open structure with stronger octahedral distortions, which is often associated with enhanced surface reactivity and a higher propensity for defect formation.

2.4.3. Electronic Structure of Titanium Oxide

Titanium dioxide (TiO₂) is also a prototypical $d^0$ oxide: titanium is formally in the Ti⁴⁺ oxidation state with a $3d^0$ electronic configuration. The electronic structure is therefore dominated by O-$2p$ states at the top of the valence band and Ti-$3d$ states at the bottom of the conduction band. In the approximately octahedral TiO₆ environment, the Ti-$3d$ states are split into lower-energy $t_{2g}$-like and higher-energy $e_g$-like groups, although the actual splitting depends on the polymorph, rutile, anatase, or brookite, and on local structural distortions [67,68]. The best-known polymorphs, rutile and anatase, have similar orbital character but differ in quantitative band-structure details, such as band-gap magnitude and band dispersion. For many optical and photochemical effects, a key role is often attributed to oxygen vacancies and related Ti³⁺-associated gap states; modern electronic-structure studies describe their electronic levels using beyond-semilocal DFT approaches, such as hybrid functionals [69].

2.5. Comparative Summary of Selected TMOs

Table 11. Comparative summary of key properties of selected TMOs.

Properties	Vanadium Oxides	Nickel Oxides	Tungsten Oxides	Titanium Oxides
Stoichiometry/ Oxidation state	– VO2; V4+; $3d^1$	– NiO; Ni2+; $3d^8$	– WO3; W6+; $5d^0$	– TiO2; Ti4+; $3d^0$
	– V2O3; V3+; $3d^2$	– ${\mathrm{Ni}}_{1-x}$O;	– ${\mathrm{WO}}_{3-x}$;	– ${\mathrm{TiO}}_{2-x}$;
	– V2O5; V5+; $3d^0$	Ni vacancies;	W5+/W6+ states	oxygen vacancies;
	– ${\mathrm{V}}_n$${\mathrm{O}}_{2n-1}$/	Ni3+ states		Ti3+ states
	${\mathrm{V}}_n$${\mathrm{O}}_{2n+1}$; mixed-valence phases			– ${\mathrm{Ti}}_n$${\mathrm{O}}_{2n-11}$; Ti3+/Ti4+ Magnéli phases
Crystal Structure/ Main Phases	VO2: monoclinic M1 ($P{2}_1/c$, No. 14);	NiO: rock-salt ($Fm\overline{3}m$, No. 225)	$\alpha$-WO3: tetragonal ($P4/nmm$, No. 129);	TiO2: anatase ($I{4}_1/amd$, No. 141);
	VO2: monoclinic M2 ($C2/m$, No. 12);		$\beta$-WO3: orthorhombic (Pnma, No. 62);	TiO2: rutile ($P{4}_2/mnm$, No. 136)
	VO2: rutile R ($P{4}_2/mnm$, No. 136);		$\gamma$-WO3: monoclinic ($P{2}_1/n$, No. 14);	TiO2: brookite ($Pbca$, No. 61)
	V2O3: corundum ($R\overline{3}c$, No. 167);		$\delta$-WO3: triclinic ($P\overline{1}$, No. 2);
	V2O3: monoclinic ($I2/a$, No. 15);		$ϵ$-WO3: monoclinic ($Pc$, No. 7);
	V2O5: orthorhombic ($Pmmn$, No. 59)		H-WO3: hexagonal ($P6/mmm$, No. 191)
Electronic Structure	VO2: correlated oxide; V-$3d$ $t_{2g}$ states split into $a_{1g}$ and $e_g^{\pi }$; temperature-driven MIT	CT/correlated insulator; O-$2p$ states dominate the valence band;	wide-band-gap $d^0$ oxide; O-$2p$ valence band and W-$5d$ conduction band;	wide-band-gap $d^0$ oxide; O-$2p$ valence band and Ti-$3d$ conduction band;
	V2O3: strongly correlated oxide; $t_{2g}$ states split into $a_{1g}$ and $e_g^{\pi }$ by trigonal distortion; Mott–Hubbard-type MIT	Ni-$3d$ states contribute to unoccupied states and multiplet/CT excitations	electron/ion insertion creates W5+/W6+ states	oxygen vacancies and Ti3+ states introduce defect levels
	V2O5: insulating CT oxide; O-$2p$ valence band and V-$3d$ conduction band;
Main Chromic Mechanism	VO2: Peierls-Mott MIT & V2O3: Mott MIT → thermochromism;	deintercalation/hole-polaron formation → anodic electrochromism > Alexandru V.:	intercalation/electron-polaron formation → cathodic electrochromism and gasochromism	photoinduced charge trapping/small-polaron formation → photochromism
	V2O5: intercalation/deintercalation → electrochromism/gasochromism

summarizes the key properties and chromic responses of the selected TMOs, providing the direct comparison between vanadium, nickel, tungsten and titanium oxides.

3. Technology of TMO Preparation

3.1. Pulsed Laser Deposition

Pulsed laser deposition (PLD) is a physical vapor deposition technique in which a high-power pulsed laser ablates a solid target inside a vacuum chamber, producing a plasma plume that expands toward a substrate where the ablated species condense. The process can be regarded as a sequence of strongly coupled stages: laser–target interaction and ablation, plume formation and expansion, including scattering and gas-phase reactions, and subsequent nucleation and growth on the substrate surface. As a result, the film thickness, microstructure, crystallinity, and phase stability can be tailored by adjusting the laser fluence, repetition rate, substrate temperature, target–substrate distance, and background gas pressure [70,71]. For TMO films, the oxygen background is particularly critical: plume–gas collisions and reactions can promote the formation of oxide species and precursors, enhance oxygen incorporation, and reduce the kinetic energy of arriving species, thereby mitigating surface damage and helping to stabilize the desired phase. In addition, oxygen pressure represents an effective processing parameter for tuning oxygen-related defect populations, such as oxygen vacancies, and consequently the functional properties of the grown oxide [72,73,74].

A schematic representation of the key PLD processes and main control parameters is shown in Figure 13, while representative deposition conditions for selected TMOs are summarized in

Table 12. PLD deposition conditions for selected transition-metal oxide films.

TMO	Substrate	PLD Parameters	Laser Type and Parameters
${\mathrm{VO}}_2$	Pt(111)/${\mathrm{TiO}}_2$ [75]	Substrate temperature: $600°\mathrm{C}$ Distance to substrate: $50\mathrm{mm}$ Working pressure $p_{{\mathrm{O}}_2}:\sim 1\mathrm{Pa}$ Target: ${\mathrm{V}}_2{\mathrm{O}}_5$	KrF excimer Frequency: $5\mathrm{Hz}$ Fluence: $1.0\mathrm{J}·{\mathrm{cm}}^{-2}$ Deposition time: –
	Si/${\mathrm{SiO}}_2$ [76]	Substrate temperature: $600°\mathrm{C}$ Distance to substrate: $31\mathrm{mm}$ Working pressure $p_{{\mathrm{O}}_2}:\sim 6.67\mathrm{Pa}$ Target: V-metal	KrF excimer Frequency: $10\mathrm{Hz}$ Fluence: $2.6\mathrm{J}·{\mathrm{cm}}^{-2}$ Deposition time: $15\min$
	Soda lime glass (SLG) [77]	Substrate temperature: 450–$600°\mathrm{C}$ Distance to substrate: $65\mathrm{mm}$ Working pressure $p_{{\mathrm{O}}_2}:\sim 2\mathrm{Pa}$ Target: V-metal	KrF excimer Frequency: $10\mathrm{Hz}$ Fluence: $1.7\mathrm{J}·{\mathrm{cm}}^{-2}$ Deposition time: $45\min$
	c-plane sapphire [78]	Substrate temperature: $600°\mathrm{C}$ Distance to substrate: $40\mathrm{mm}$ Working pressure $p_{{\mathrm{O}}_2}:0.5$–$3.6\mathrm{Pa}$ Target: ${\mathrm{VO}}_2$	KrF excimer Frequency: $10\mathrm{Hz}$ Fluence: – Deposition time: –
	Sapphire [79]	Substrate temperature: $600°\mathrm{C}$ Working pressure $p_{{\mathrm{O}}_2}:2.6$–$4\mathrm{Pa}$ Target: ${\mathrm{V}}_2{\mathrm{O}}_3$	ArF excimer Frequency: – Fluence: $2.0$–$3.0\mathrm{J}·{\mathrm{cm}}^{-2}$ Deposition time: –
	Si/${\mathrm{SiO}}_2$ [80]	Substrate temperature: $600°\mathrm{C}$ Distance to substrate: $35\mathrm{mm}$ Working pressure $p_{{\mathrm{O}}_2}:\sim 6.67\mathrm{Pa}$ Target: ${\mathrm{V}}_2{\mathrm{O}}_5$	KrF excimer Frequency: $10\mathrm{Hz}$ Laser energy: $200\mathrm{mJ}$ Deposition time: –
	Sapphire [81]	Substrate temperature: $600°\mathrm{C}$ Distance to substrate: $90\mathrm{mm}$ Working pressure $p_{{\mathrm{O}}_2}:\sim 1.33\mathrm{Pa}$ Target: ${\mathrm{V}}_2{\mathrm{O}}_5$ Flow $F_{{\mathrm{O}}_2}:10\mathrm{sccm}$	KrF excimer Frequency: – Fluence: – Deposition time: –
$\mathrm{NiO}$	${\mathrm{SnO}}_2$:$\mathrm{F}$ (FTO)-coated glass [82]	Substrate temperature: RT–$300°\mathrm{C}$ Distance to substrate: 40–$50\mathrm{mm}$ Working pressure $p_{{\mathrm{O}}_2}:0.1$–$10\mathrm{Pa}$ Target: $\mathrm{NiO}$	KrF excimer Frequency: $5\mathrm{Hz}$ Fluence: $1.0$–$2.0\mathrm{J}·{\mathrm{cm}}^{-2}$ Deposition time: 30–$60\min$
	${\mathrm{SnO}}_2$:$\mathrm{F}$ (FTO)-coated glass [83]	Substrate temperature: RT–$300°\mathrm{C}$ Distance to substrate: $55\mathrm{mm}$ Working pressure $p_{{\mathrm{O}}_2}:45\mathrm{Pa}$ Target: Ni-metal	KrF excimer Frequency: $20\mathrm{Hz}$ Fluence: $2.0\mathrm{J}·{\mathrm{cm}}^{-2}$ Deposition time: –
	ITO-coated glass [84]	Substrate temperature: RT–$400°\mathrm{C}$ Working pressure $p_{{\mathrm{O}}_2}:0.01$–$0.016\mathrm{Pa}$ Target: $\mathrm{NiO}$ Flow $F_{{\mathrm{O}}_2}:10\mathrm{sccm}$	KrF excimer Frequency: 5–$50\mathrm{Hz}$ Laser energy: $125\mathrm{mJ}$ Deposition time: –
${\mathrm{WO}}_3$	Si(100) [85]	Substrate temperature: 300–$400°\mathrm{C}$ Distance to substrate: $75\mathrm{mm}$ Working pressure $p_{{\mathrm{O}}_2}:2.67\mathrm{Pa}$ Target: ${\mathrm{WO}}_3$	ArF excimer Frequency: $10\mathrm{Hz}$ Fluence: $1.5\mathrm{J}·{\mathrm{cm}}^{-2}$ Deposition time: $30\min$
	${\mathrm{SrTiO}}_3$ [86]	Substrate temperature: 300–$800°\mathrm{C}$ Working pressure $p_{{\mathrm{O}}_2}:1$–$6\mathrm{Pa}$ Target: ${\mathrm{WO}}_3$	KrF excimer Frequency: $3\mathrm{Hz}$ Fluence: $0.6$–$1.2\mathrm{J}·{\mathrm{cm}}^{-2}$ Deposition time: –
${\mathrm{TiO}}_2$	ITO-coated glass, Si [87]	Substrate temperature: 25–$700°\mathrm{C}$ Distance to substrate: $80\mathrm{mm}$ Working pressure $p_{{\mathrm{O}}_2}:0.1\mathrm{Pa}$ Target: ${\mathrm{TiO}}_2$ (anatase and rutile)	Femtosecond laser $(800\mathrm{nm},100\mathrm{fs})$ Frequency: $1\mathrm{kHz}$ Fluence: $3.34\mathrm{J}·{\mathrm{cm}}^{-2}$ Deposition time: –
	Aluminum foil, Si(100) [88]	Substrate temperature: RT Distance to substrate: $30\mathrm{mm}$ Without oxygen Target: ${\mathrm{TiO}}_2$ (anatase)	Nd:glass Frequency: $10\mathrm{Hz}$ Laser energy: $2.8\mathrm{mJ}$ Deposition time: $180\min$

.

A major reason why PLD is widely used for the growth of TMO thin films is its combination of versatility and compositional control. Many complex oxides can be deposited from a single target with the desired composition, and PLD often enables near-stoichiometric transfer because the ablation and transport processes are highly energetic and occur on short time scales. However, deviations between the target and film composition may still occur due to incongruent ablation, preferential scattering in the background gas, and re-sputtering, particularly when the background pressure and kinetic energy of the plume species vary [89]. The energetic, ionized plume can also assist crystalline growth and, in some cases, reduce the substrate temperature required for crystalline or epitaxial film formation. The setup is comparatively clean and convenient for multilayer structures, since different targets can be switched during the same deposition run while the laser source remains outside the chamber [72,73].

At the same time, PLD has well-known limitations that are relevant to practical TMO devices. The plume is typically forward-directed and has a limited cross-sectional area, partly determined by the laser spot size, which complicates thickness uniformity and scaling to large-area substrates. For the same geometrical reason, conformal step coverage and in situ thickness control can also be challenging. Another persistent issue is the formation of droplets and particulates during ablation, which can be incorporated into the film surface and degrade performance or even cause device failure. Thus, PLD is often a method of choice for producing high-quality, compositionally faithful TMO films and multilayer structures, but it requires careful process optimization when thickness uniformity, droplet-free surfaces, and large-area manufacturing are required [71,90].

Overall, the representative deposition conditions summarized in Table 12 highlight the broad tunability of PLD for TMOs, where phase formation and defect populations can be adjusted by oxygen pressure, substrate temperature, and laser parameters. In oxide PLD, KrF excimer lasers ($\lambda =248\mathrm{nm}$) are used most frequently, and the employed laser fluence is commonly in the 1–$2\mathrm{J}{\mathrm{cm}}^{-2}$ range.

3.2. Magnetron Sputtering

Reactive magnetron sputtering is a widely used physical vapor deposition technique for growing TMO thin films with good thickness uniformity and strong industrial scalability. A metallic or suboxide target is bombarded by ions from a plasma, most commonly sustained in argon, ejecting atoms and clusters from the target surface. These sputtered species are transported through the gas phase and condense on a substrate to form a thin film. In the magnetron configuration, a magnetic field near the target confines electrons, enhances ionization, and thereby enables a stable discharge at relatively low pressures and high deposition rates (often with reduced thermal load on the substrate compared to non-magnetron operation) [91,92].

To deposit oxides, O₂ is introduced together with Ar, and the sputtered metal species react in the plasma and/or at the growing surface to form the oxide. Consequently, phase formation, stoichiometry, defect chemistry, and microstructure are governed by parameters such as total pressure, oxygen partial pressure (or O₂ flow), Ar:O₂ ratio, applied power (DC for conductive targets, RF for insulating conditions, and often pulsed-DC in reactive operation), substrate temperature, substrate bias, and the overall plasma composition. For many TMOs, oxygen partial pressure is especially critical because it directly affects oxygen vacancies and therefore electrical and optical properties [93,94].

One of the main advantages of reactive magnetron sputtering for TMO films is its ability to deliver uniformity and reproducibility over large areas. Thus, it is widely used for coatings and device fabrication. It is compatible with a broad range of substrates and can be operated at relatively low substrate temperatures. Multi-target sputtering and co-sputtering further enable composition tuning, doping, and multilayer architectures [93,95].

The method also has limitations. In reactive operation, the target surface can undergo oxidized target poisoning, reducing the deposition rate and introducing process hysteresis, which complicates stable control of oxygen incorporation—especially near the transition between metallic and compound modes [93,94,96].

In addition, because the typical kinetic energy of arriving species is lower than in highly energetic plume-based methods, high crystallinity in some TMOs may require elevated substrate temperatures, substrate biasing, or post-deposition annealing. Finally, plasma-driven ion bombardment can induce stress, cause re-sputtering of light elements, or create subsurface damage if pressure and bias conditions are not optimized. Overall, reactive magnetron sputtering is a robust and scalable route to functional TMO thin films, but it requires careful tuning of oxygen and plasma conditions when phase purity, defect engineering, and precise stoichiometry are critical [93,94]. Representative sputtering configurations (RF/DC reactive and HiPIMS) are summarized schematically in Figure 14, while representative growth conditions for selected TMOs are listed in

Table 13. Magnetron sputtering deposition conditions for selected transition-metal oxide films.

TMO	Magnetron Sputtering Technique	Deposition Parameters	Substrate and Target Parameters	Ambient Conditions
${\mathrm{VO}}_2$	RF [99]	Sputtering power: $70\mathrm{W}$	Substrate: Corning glass Substrate temperature: $450°\mathrm{C}$ Target: V	Working pressure: $1.8\mathrm{Pa}$
	RF [100]	Sputtering power: $350\mathrm{W}$	Substrates: Si(100), glass Substrate temperature: $500°\mathrm{C}$ Distance to substrate: $90\mathrm{mm}$ Target: V	Working pressure: $0.1$–$0.5\mathrm{Pa}$ Flow rate $F_{{\mathrm{O}}_2}:1.75\mathrm{sccm}$
	RF [101]	–	Substrates: ${\mathrm{SiO}}_2$/Si, c-plane sapphire Substrate temperature: $650°\mathrm{C}$ Target: V	$p_{{\mathrm{O}}_2}:56\mathrm{mPa}$ $p_{\mathrm{Ar}}:19.4\mathrm{mPa}$ Working pressure: $0.02\mathrm{Pa}$
	DC [102]	Sputtering power: 90–$100\mathrm{W}$	Substrates: ${\mathrm{SiO}}_2$/Si, quartz sapphire Substrate temperature: $650°\mathrm{C}$ Target: V	Working pressure: $0.02\mathrm{Pa}$ Flow rate $F_{{\mathrm{O}}_2}:1.3\mathrm{sccm}$ Flow rate $F_{\mathrm{Ar}}:100\mathrm{sccm}$
	Magnetron sputtering [103]	Power density on V-target: $1.1\mathrm{W}·{\mathrm{cm}}^{-2}$	Substrate: SLG Substrate temperature: 300–$400°\mathrm{C}$ Target: V	Working pressure: $0.02\mathrm{Pa}$ Flow rate $F_{{\mathrm{O}}_2}:1.8\mathrm{sccm}$ Flow rate $F_{\mathrm{Ar}}:40\mathrm{sccm}$
$\mathrm{NiO}$	DC [104]	Sputtering power: $600\mathrm{W}$	Substrates: Al, KCl, glass Substrate temperature: RT Distance to substrate: $75\mathrm{mm}$ Target: Ni	$p_{{\mathrm{O}}_2}:0.15\mathrm{Pa}$ $p_{\mathrm{Ar}}:0.35\mathrm{Pa}$ Working pressure: $0.5\mathrm{Pa}$
	DC [47]	Sputtering power: 30–$70\mathrm{W}$	Substrate: glass Substrate temperature: RT Distance to substrate: $75\mathrm{mm}$ Target: Ni	$p_{{\mathrm{O}}_2}/p_{\mathrm{Ar}}:2/1$ Working pressure: $1\mathrm{Pa}$
	RF [105]	Sputtering power: $90\mathrm{W}$	Substrate: SLG Substrate temperature: RT Target: NiO	Working pressure: $40\mathrm{mPa}$ Flow rate $F_{\mathrm{Ar}}:6\mathrm{sccm}$
${\mathrm{WO}}_3$	DC [106]	Sputtering power: $200\mathrm{W}$	Substrate: ${\mathrm{CaF}}_2$ Substrate temperature: $280°\mathrm{C}$ Target: W	$p_{{\mathrm{O}}_2}/p_{\mathrm{Ar}}:0.43$ Working pressure: $1.33$–$4\mathrm{Pa}$
	RF [107]	Sputtering power: $120\mathrm{W}$	Substrate: ITO-coated glass Substrate temperature: RT Target: ${\mathrm{WO}}_3$	Working pressure: $0.4\mathrm{Pa}$ Flow rate $F_{\mathrm{Ar}}:67\mathrm{sccm}$
${\mathrm{TiO}}_2$	DC [108]	Sputtering power: 0–$50\mathrm{W}$	Substrate: Ti/FTO Substrate temperature: RT Distance to substrate: $115\mathrm{mm}$ Target: Ti	$p_{{\mathrm{O}}_2}:46\mathrm{mPa}$ Working pressure: $0.14\mathrm{Pa}$ Flow rate $F_{\mathrm{Ar}}:10\mathrm{sccm}$ Flow rate $F_{{\mathrm{O}}_2}:5\mathrm{sccm}$
	RF [109]	Sputtering power: 80–$200\mathrm{W}$	Substrates: Si(100), glass Substrate temperature: RT–$150°\mathrm{C}$ Target: ${\mathrm{TiO}}_2$	Working pressure: 4–$8\mathrm{Pa}$ Flow rate $F_{{\mathrm{O}}_2}:0$–$30\mathrm{sccm}$
	RF [110]	Sputtering power: $500\mathrm{W}$	Substrate: p-Si(100) Substrate temperature: RT Distance to substrate: $44\mathrm{mm}$ Target: ${\mathrm{TiO}}_2$	Working pressure: $0.6\mathrm{Pa}$ Flow rate $F_{{\mathrm{O}}_2}:0$–$8.5\mathrm{sccm}$
	RF [111]	Sputtering power: $160\mathrm{W}$	Substrate: glass Substrate temperature: RT Target: ${\mathrm{TiO}}_2$	Working pressure: $0.6\mathrm{Pa}$ Flow rate $F_{\mathrm{Ar}}:30\mathrm{sccm}$

.

Overall, Table 13 emphasizes the practical versatility of magnetron sputtering for TMO thin films. In most studies, oxide layers are deposited in reactive mode using either DC power, for conductive targets, or RF power, commonly adopted to maintain stable operation when the target surface becomes oxidized. A further advantage is the broad compatibility with substrates of different thermal and structural constraints, including glass and SLG for scalable coatings, together with single-crystal substrates, such as sapphire, when improved crystallinity or texture control is needed. Together with the possibility of multi-target sputtering and co-sputtering, these features make magnetron sputtering an established technique for large-area TMO coatings and device fabrication, where uniformity, reproducibility, and compositional and defect tuning through oxygen control are the primary requirements.

3.3. Sol-Gel and Aerosol Spray Deposition

Sol–gel processing (Figure 15) is a solution-based chemical route for producing inorganic thin films and coatings, in which molecular precursors, commonly metal alkoxides or metal salts, are converted into an extended M–O–M network through hydrolysis and condensation reactions. The resulting colloidal sol evolves toward a gel, and subsequent drying and thermal treatment remove volatile species and organic residues, densify the layer, and, when required, promote crystallization of the final inorganic phase [112].

For thin film formation, the sol is typically deposited onto a substrate by dip-coating or spin-coating, with film thickness and uniformity governed by coupled fluid-dynamic and evaporation phenomena, including withdrawal or spin speed, viscosity, surface tension, and solvent volatility. At the same time, the developing gel network determines shrinkage, porosity, and mechanical integrity during drying. In practice, film properties are strongly influenced by chemical parameters, such as precursor type, solvent system, water-to-precursor ratio, pH, complexing or chelating agents, and aging time, together with processing parameters, including ambient humidity during drying, heating rate, annealing temperature, dwell time, and annealing atmosphere [113].

In aerosol-assisted variants, the precursor sol or solution is atomized into droplets and transported to the substrate, enabling coating of large areas and non-planar geometries and offering a practical pathway toward scalable manufacturing. Depending on temperature and chemistry, conversion to the final inorganic network can occur predominantly after deposition, as in aerosol-gel or aerosol-assisted sol–gel processing, or partially during flight, impact, and subsequent heating, as in spray-based routes [114,115,116]. Representative TMO thin film examples obtained using sol–gel routes and their processing conditions are summarized in

Table 14. Sol–gel deposition conditions for selected transition-metal oxide films.

TMO	Substrate	PLD Parameters	Laser Type and Parameters
${\mathrm{VO}}_2$	Spin coating  [117]	Substrate: Si Source of TM: $\mathrm{VO}{\left(\mathrm{acac}\right)}_2$ Sol: $100\mathrm{mM}$ of $\mathrm{VO}{\left(\mathrm{acac}\right)}_2$ in ethanol	Drying at $150°\mathrm{C}$ 1. humid ${\mathrm{N}}_2$: $1.8$–$2.3{\mathrm{g}}_{{\mathrm{H}}_2\mathrm{O}}/{\mathrm{kg}}_{\mathrm{dry}-\mathrm{air}}$ (11–$13\%$ RH at 22–$23°\mathrm{C}$) 2. dry air: $2.2$–$3.2{\mathrm{g}}_{{\mathrm{H}}_2\mathrm{O}}/{\mathrm{kg}}_{\mathrm{dry}-\mathrm{air}}$ (14–$18\%$ RH at 21–$23°\mathrm{C}$) 3. ambient air: $5.7$–$7.5{\mathrm{g}}_{{\mathrm{H}}_2\mathrm{O}}/{\mathrm{kg}}_{\mathrm{dry}-\mathrm{air}}$ (34–$45\%$ RH at $22°\mathrm{C}$) 4. humid air: $12.5$–$15.9{\mathrm{g}}_{{\mathrm{H}}_2\mathrm{O}}/{\mathrm{kg}}_{\mathrm{dry}-\mathrm{air}}$ (85–$90\%$ RH at 20–$23°\mathrm{C}$) Annealing at $550°\mathrm{C}$ for $1\mathrm{h}$ ($5°\mathrm{C}/\min$) under a $1.7\mathrm{L}/\min$ $F_{{\mathrm{N}}_2}$ flux
	Spin coating  [118]	Substrates: Si, quartz Source of TM: ammonium citrato-oxovanadate (IV) ${\left({\mathrm{NH}}_4\right)}_4{\left[{\mathrm{V}}_2{\mathrm{O}}_2{\left({\mathrm{C}}_6{\mathrm{H}}_4{\mathrm{O}}_7\right)}_2\right]}_2·{\mathrm{H}}_2\mathrm{O}$ (CA-V(IV)) Sol: CA-V(IV) + ethanol + cetyltrimethylammonium bromide (CTAB) + distilled water, with varying molar ratio of CTAB to V	Annealing at $500°\mathrm{C}$ for $1\mathrm{h}$ in Ar atmosphere
	Spin coating  [119]	Substrate: ${\mathrm{Al}}_2{\mathrm{O}}_3\left(0001\right)$ Source of TM: vanadyl triisopropoxide $\mathrm{VO}{\left({\mathrm{OC}}_3{\mathrm{H}}_7\right)}_3$ Sol: $50\mathrm{mL}$ of $\mathrm{VO}{\left({\mathrm{OC}}_3{\mathrm{H}}_7\right)}_3$, isopropanol and acetic acid with mass ratio of $1:20.1:3.2$, respectively	Drying at $250°\mathrm{C}$ for $3\min$ Annealing at 370–$520°\mathrm{C}$ for $7\mathrm{h}$ in ${\mathrm{N}}_2$ with $500\mathrm{mL}/\min$ flow rate
$\mathrm{NiO}$	Spin coating  [120]	Substrate: glass Source of TM: nanocrystalline NiO powder Sol: nanocrystalline NiO powder + m-cresol	Drying at $100°\mathrm{C}$ for $10\min$
	Spin coating  [121]	Substrate: – Source of TM: nickel acetate tetrahydrate $[\mathrm{Ni}{\left({\mathrm{CH}}_3\mathrm{COO}\right)}_2·4{\mathrm{H}}_2\mathrm{O}]$ Sol: $2.54\mathrm{g}$ of nickel acetate tetrahydrate + $100\mathrm{mL}$ of $1\mathrm{M}$ aqueous citric acid solution and $10\mathrm{mL}$ of ethylene glycol	Drying at $250°\mathrm{C}$ for $2\mathrm{h}$ at atmospheric pressure Annealing at $800°\mathrm{C}$ for $1\mathrm{h}$
	Spin & dip coating  [122]	Substrates: ITO-coated glass, Corning glass Source of TM: Nickel(II) 2-ethylhexanoate ${\left[{\mathrm{CH}}_3{\left({\mathrm{CH}}_2\right)}_3\mathrm{CH}\left({\mathrm{C}}_2{\mathrm{H}}_5\right){\mathrm{CO}}_2\right]}_2\mathrm{Ni}$ Sol: $42\mathrm{g}$ of Nickel(II) 2-ethylhexanoate + $4\mathrm{mL}$ isopropanol	Drying at RT in air for $10\min$ Annealing at $350°\mathrm{C}$ for $1\mathrm{h}$
${\mathrm{WO}}_3$	Spin coating  [123]	Substrate: ITO-coated glass Source of TM: tungsten (VI) chloride (${\mathrm{WCl}}_6$) powder Sol: $1\mathrm{g}$ of ${\mathrm{WCl}}_6$ $20\mathrm{mL}$ of absolute ethanol $2\mathrm{mL}$ of glacial acetic acid $2\mathrm{mL}$ of ${\mathrm{H}}_2{\mathrm{O}}_2$	Drying at $100°\mathrm{C}$ for $3\min$ Annealing at $250°\mathrm{C}$
	Dip coating  [124]	Substrate: glass Source of TM: ${\mathrm{H}}_2{\mathrm{WO}}_4$ Sol: $6.72\mathrm{g}$ of ${\mathrm{H}}_2{\mathrm{WO}}_4$ + $10\mathrm{mL}$ of ${\mathrm{H}}_2{\mathrm{O}}_2$	Drying at $100°\mathrm{C}$ for $10\min$ Air-annealing at 150–$400°\mathrm{C}$ for $1\mathrm{h}$ ($5°\mathrm{C}/\min$)
	Sol–gel  [125]	Substrates: FTO-coated glass, glass Source of TM: tungsten (VI) chloride (${\mathrm{WCl}}_6$) powder Sol: $1.1896\mathrm{g}$ of ${\mathrm{WCl}}_6$ $10\mathrm{mL}$ of ${\mathrm{H}}_2{\mathrm{O}}_2$ $30\mathrm{mL}$ of 2-methoxyethanol	Drying at RT for $5\min$ in air Air-annealing at $300°\mathrm{C}$ for $1\mathrm{h}$
${\mathrm{TiO}}_2$	Dip coating  [126]	Substrate: glass Source of TM: titanium tetra-isopropoxide (TTIP) $\mathrm{Ti}{\left[\mathrm{OCH}{\left({\mathrm{CH}}_3\right)}_2\right]}_4$ Sol: TTIP, 1-propanol, hydrochloric acid (HCl, $36\%$), and monoethanolamine	Drying at $100°\mathrm{C}$ for $20\min$ Preheated at $200°\mathrm{C}$ for $10\min$ Annealing at $500°\mathrm{C}$ for $2\mathrm{h}$
	Dip coating  [127]	Substrates: SLG, quartz Source of TM: titanium tetra-isopropoxide (TTIP) $\mathrm{Ti}{\left[\mathrm{OCH}{\left({\mathrm{CH}}_3\right)}_2\right]}_4$ Sol: TTIP, isopropanol ${\left({\mathrm{CH}}_3\right)}_2\mathrm{CHOH}$, glacial acetic acid ${\mathrm{CH}}_3\mathrm{COOH}$ and methanol ${\mathrm{CH}}_3\mathrm{OH}$	Drying at $100°\mathrm{C}$ for $15\min$ Annealing at 350–$950°\mathrm{C}$ for $1\mathrm{h}$
	Spin coating  [128]	Substrate: glass Source of TM: nanocrystalline ${\mathrm{TiO}}_2$ powder Sol: $0.4\mathrm{g}$ of nanocrystalline ${\mathrm{TiO}}_2$ powder $5\mathrm{mL}$ of ethanol $5\mathrm{mL}$ of diethylene glycol	Drying at $100°\mathrm{C}$ for $10\min$ Annealing at $400°\mathrm{C}$

.

Overall, Table 14 highlights the practical versatility of sol–gel processing for TMO thin films. Across the reported studies, spin coating and dip coating are the most common deposition modes, while a wide variety of substrates, including Si, quartz, glass, and transparent conductive glass such as ITO- or FTO-coated glass, can be employed depending on the targeted device architecture. The cited examples also illustrate that sol–gel chemistry provides multiple precursor options, including alkoxides, acetylacetonates, chlorides, and acetates, as well as straightforward composition tuning at the solution-processing stage. Importantly, the final oxide phase, crystallinity, and defect-related properties are largely determined by the drying and annealing steps, including atmosphere control, making thermal processing a central processing parameter for sol–gel prepared TMOs in functional coatings.

3.4. Other Deposition Methods

Beyond PLD, magnetron sputtering, and sol–gel, several other thin film technologies are frequently employed to access either (i) highly conformal coatings and precise thickness control, (ii) scalable conformal growth on complex geometries, or (iii) epitaxial films for fundamental structure–property studies. In the context of the present TMOs (VO₂, WO₃, TiO₂, NiO), particularly relevant routes include atomic layer deposition (ALD), chemical vapor deposition (CVD, including aerosol-assisted CVD), electrochemical growth (electrodeposition/anodization), and molecular beam epitaxy (MBE).

Atomic layer deposition (ALD) is especially attractive when thickness control and conformality on textured or 3D substrates are required. VO₂ films have been demonstrated by ALD using inorganic precursors, such as ${\mathrm{VCl}}_4$, with direct crystallization under suitable conditions [129]. NiO electrochromic coatings have also been prepared by ALD on transparent conductive substrates, illustrating the viability of ALD for anodic electrochromic oxides [130]. For WO₃, ALD processes, such as $\mathrm{W}{\left(\mathrm{CO}\right)}_6+{\mathrm{O}}_3$, have been reported as a route to controlled ultrathin tungsten oxide films [131]. TiO₂ is one of the best-established ALD oxides; comprehensive reviews summarize its wide precursor chemistry and broad application space [132].

CVD and aerosol-assisted CVD (AACVD) methods are widely used for scalable oxide coatings, often providing good step coverage and industrial compatibility. Thermochromic VO₂ films have been fabricated by low-pressure CVD (LPCVD), demonstrating controllable growth of crystalline VO₂ from volatile vanadium precursors [133]. AACVD has been used to prepare doped VO₂ films as well as WO₃ coatings on glass from solution-deliverable precursors [134,135]. For NiO, metal–organic methods have been reported for electrochromic NiO coatings, highlighting that gas-phase chemistry can also be leveraged for this anodic electrochromic oxide [136].

Electrochemical methods, including electrodeposition and anodization, are particularly common for electrochromic TMOs because they naturally integrate with conductive substrates and can yield porous or high-surface-area morphologies. NiO electrochromic films have been produced by electrodeposition onto FTO-coated glass [136]. WO₃ is likewise frequently prepared by electrodeposition for electrochromic applications, with multiple studies and reviews treating electrodeposition as a key low-cost method to obtain functional WO₃ coatings [137]. For TiO₂, anodization of Ti or Ti-coated substrates is a standard method to obtain self-organized TiO₂ nanotube arrays, enabling strongly nanostructured films after subsequent annealing [138,139].

Molecular beam epitaxy (MBE), including oxide or oxygen-assisted variants, is often used when epitaxial, well-defined model films are needed to probe intrinsic mechanisms and substrate/strain effects. VO₂ ultrathin films have been grown by reactive MBE on rutile TiO₂ substrates [140]. Plasma-assisted MBE has also been applied to epitaxial NiO films on semiconducting substrates [141]. For WO₃, single-crystalline and epitaxial WO₃ thin films grown by MBE have been reported and used to study orientation-dependent functional behavior [142,143]. Epitaxial rutile TiO₂ films on sapphire grown by MBE have been reported, illustrating MBE control over phase and microstructure [144].

Overall, other routes complement PLD, sputtering, and sol–gel processing by offering either superior conformality and thickness control (ALD), scalable conformal growth with strong industrial relevance (CVD/AACVD), device-friendly electrochemical fabrication of porous electrochromic layers (electrodeposition/anodization), or high-purity epitaxial model films for mechanistic studies (MBE).

3.5. Comparative Analysis of TMO Thin Film Preparation Technologies

The preparation route is a key factor that determines not only the phase composition, crystallinity, morphology, and stoichiometry of chromic TMO thin films, but also their technological relevance for practical devices. While laboratory-scale methods such as PLD and MBE are highly useful for obtaining well-controlled model systems and clarifying structure–property relationships, scalable techniques such as magnetron sputtering, sol–gel processing, spray deposition, ALD, CVD, and electrochemical deposition are more directly connected with device fabrication. Therefore, the choice of deposition method should be evaluated not only in terms of film quality, but also with respect to scalability, processing temperature, substrate compatibility, reproducibility, cost, and integration into multilayer or device architectures. A comparative overview of the main advantages, limitations, and device-integration aspects of the reviewed preparation technologies is summarized in

Table 15. Summary of TMO thin film preparation technologies.

TMO Preparation Technology	Main Advantages	Main Limitations	Device Integration Aspects
Magnetron Sputtering	Scalable to large-area coatings; industry-compatible; good thickness and uniformity control	Requires optimization of oxygen partial pressure, stress, crystallinity, and post-annealing; reactive sputtering may suffer from target poisoning and process hysteresis; phase control can be challenging for multivalent oxides	Highly relevant for SW and multilayer coatings; compatible with glass substrates and industrial coating lines
PLD	Good stoichiometry transfer; suitable for complex and multicomponent oxides; precise control of deposition parameters, including oxygen pressure, laser fluence, and substrate temperature	Limited scalability; relatively small deposition area; possible formation of particulates/droplets; relatively high cost and low throughput	Excellent for model films and mechanism studies; less attractive for industrial large-area devices
Sol–gel	Low cost; simple equipment; compositional flexibility; potentially scalable to large-area coatings	Cracking, porosity, thickness non-uniformity, organic residues, low oxygen control, and the need for thermal treatment	Useful for low-cost coatings, but device reproducibility and long-term stability require careful control

.

4. Devices and Applications

4.1. Thermochromic VO₂-Based Smart Windows

SWs are a critical component of the building envelope because they simultaneously manage (i) transmission heat losses driven by temperature differences, (ii) solar heat gains driven by short-wave irradiance, and (iii) daylight access, which affects both lighting energy demand and visual comfort.

In the energy balance of a building, the contribution of a window is commonly described by two main terms: a non-solar term, determined by the overall heat transfer coefficient (U-value), and a solar term, determined by the solar heat gain coefficient (SHGC, or g-value). At the same time, optical parameters such as visible transmittance ($T_{\mathrm{lum}}$) and solar transmittance ($T_{\mathrm{sol}}$) govern the quality of daylighting and solar-energy admission. Therefore, window performance is inherently multi-objective: a low U-value and an optimized SHGC are central design goals, but they must be balanced against daylight access, outside view, condensation resistance, and the long-term durability of the insulating glass unit (IGU) [145].

In more detailed analyses, the angular dependence of solar transmission and the spectral transmittance and reflectance over the solar spectrum are also taken into account. In typical double glazing with an air-filled cavity, a substantial part of the heat transfer across the gap may occur through long-wave radiative exchange between the panes, especially because uncoated glass has high thermal emittance. In addition, perimeter components such as spacers and edge seals may act as thermal bridges and strongly affect service life by limiting moisture ingress and fill-gas leakage. These edge-region effects are therefore not secondary: they directly influence the effective window-level $U_{\mathrm{w}}$ and determine whether the initial thermal performance is maintained over time [146].

A conventional window is typically characterized by standardized performance parameters, including $U\sim 0.35$, $\mathrm{SHGC}\sim 0.8$, $T_{\mathrm{vis}}\sim 0.9$, and $T_{\mathrm{sol}}\sim 0.8$.

A passive way to improve glazing performance is to combine low-emissivity (low-e) coatings with optimized cavity fills. Low-e coatings suppress long-wave radiative heat transfer by lowering the surface emissivity of the glazing, thereby reducing $U_{\mathrm{g}}$ without necessarily sacrificing high visible transmittance $T_{\mathrm{vis}}$ when spectrally selective designs are used. Reviews of commercially available low-e materials and products show that such coatings are already a widely implemented strategy for improving both the thermal and optical performance of windows [147]. In parallel, replacing air in the cavity with inert gases, most commonly Ar and, in higher-performance configurations, Kr, reduces conductive and convective heat transfer across the gap. However, the magnitude and long-term persistence of this benefit depend strongly on the quality of the edge seal and spacer design, since these components govern both gas retention and resistance to moisture ingress over service life [146].

An active way to further improve glazing performance beyond low-e coatings and gas fills includes triple glazing, warm-edge spacers, vacuum insulating glazing, and smart glazing technologies that dynamically tune solar and/or visible transmittance to match operational needs. Reviews show that electrochromic and related SWs can reduce cooling demand by modulating transmittance under external control. They can also lower lighting demand by maintaining a more favorable daylighting regime. In contrast, thermochromic systems, often based on VO₂, provide passive, temperature-driven solar control. The physics and optimization of thermochromic performance, particularly the trade-off between solar modulation and luminous transmittance, have been extensively reviewed and experimentally characterized [148].

In summer, an ideal SW (Figure 16) should maintain a bright indoor environment while limiting solar heat gains. It should transmit most of the visible light, about 60–80% in the 380–780 nm range, to preserve daylighting and a clear outside view, while blocking or reflecting a significant portion of the near-infrared radiation, about 780–2500 nm, which is mainly responsible for solar heating. Thus, an SW should remain highly transparent in the visible range, suppress NIR transmission, and, ideally, facilitate net radiative heat rejection from the indoor side to the outdoor environment.

In winter, an ideal SW should maximize useful solar heat gains while minimizing heat losses. It should still maintain high visible transmittance, about 60–$80\%$ in the 380–780 nm range, to ensure daylight access, but it should also exhibit high solar transmittance over a broad part of the solar spectrum to support passive heating. In the long-wave infrared, the desirable behavior is generally the opposite of the summer case: the window should reduce radiative heat loss by reflecting thermal radiation emitted by indoor surfaces back into the room. Thus, the winter-optimal state combines good visible transparency and enhanced solar admission with improved retention of indoor thermal radiation.

The application of vanadium oxide in SW is limited by its intrinsic properties, namely relatively low luminous transmittance ($T_{\mathrm{lum}}\sim 40\%$), relatively high absorption, and a limited solar modulation efficiency ($\Delta T_{\mathrm{sol}}\le 10\%$) for a single layer. Moreover, the critical temperature ($T_c\sim 68$ °C) is too high for most climatic regions. Therefore, lowering the critical temperature to $T_c\le 25°$C and improving $T_{\mathrm{lum}}$ are the main requirements for practical VO₂-based SWs. Another challenge is the long-term cycling stability of VO₂, since in the presence of oxygen it may undergo further oxidation to V₂O₅, which is the most stable oxidation state among vanadium oxides. These challenges can be addressed by doping, morphology engineering, and multilayer design.

Elemental doping is one of the conventional approaches used to tailor the MIT temperature and optical properties (

Table 16. Dopants for tuning the MIT temperature and optical performance of VO₂ films.

Dopant	${\mathit{T}}_{\mathbf{lum}}$	$\mathbf{\Delta }{\mathit{T}}_{\mathbf{sol}}$	$\mathbf{\Delta }{\mathit{T}}_{\mathit{c}}$	Mechanism
${\mathrm{W}}^{6+}$ [151,152]	$69.5\%$	$3.4\%$	Decrease in $\sim 20$–$26°\mathrm{C}/\mathrm{at}.\%$	Increased free electron concentration
${\mathrm{Ti}}^{4+}$ [153,154]	$53.0\%$	$17.2\%$	Increase	Smaller ionic radius
${\mathrm{Nb}}^{5+}$ [155]	22–$40\%$	–	Decrease in $\sim 2°\mathrm{C}/\mathrm{at}.\%$	Increased free electron concentration
${\mathrm{Mg}}^{2+}$ [156,157]	$82.1\%$	$4.8\%$	Decrease in $\sim 3°\mathrm{C}/\mathrm{at}.\%$	Increased free hole concentration
${\mathrm{Zr}}^{4+}$ [158]	$60.4\%$	$14.1\%$	Decrease in $\sim 0.4°\mathrm{C}/\mathrm{at}.\%$	Larger ionic radius
${\mathrm{Cr}}^{3+}$ [159]	$4.0$–$23.8\%$	$0.6$–$11.0\%$	Increase	Increased free hole concentration
${\mathrm{Sn}}^{4+}$ [160]	–	–	Increase in $\sim 1°\mathrm{C}/\mathrm{at}.\%$	Smaller ionic radius
${\mathrm{Eu}}^{3+}$ [161]	52–$56\%$	$5.3$–$6.7\%$	Decrease in $\sim 5°\mathrm{C}/\mathrm{at}.\%$	Larger ionic radius
${\mathrm{Mg}}^{2+}+{\mathrm{W}}^{6+}$ [162]	$81.3\%$	$4.3\%$	Decrease	Increased free hole and electron concentrations
${\mathrm{Zr}}^{4+}+{\mathrm{W}}^{6+}$ [158]	$56.4\%$	$12.3\%$	Decrease	Increased free electron concentration
${\mathrm{Mo}}^{6+}+{\mathrm{W}}^{6+}$ [163]	–	–	Decrease	Increased free electron concentration

). The choice of dopant is usually guided by two main considerations: increasing the carrier density to promote the electronic phase transition, or inducing lattice distortion to facilitate the structural phase transition [150].

Besides lowering ${\mathrm{T}}_c$, another major challenge for practical VO₂-based SWs is to achieve sufficiently high luminous transmittance ($T_{\mathrm{lum}}$) together with adequate solar modulation ability ($\Delta T_{\mathrm{sol}}$). One possible solution is related to the design and architecture of the glazing stack. A VO₂-based SW typically consists of a multilayer thin film structure, as shown in Figure 17. To reduce optical reflection losses and increase luminous transmittance, additional antireflection (AR) layers may be incorporated into the structure (Figure 17d). Candidate AR materials are listed in

Table 17. AR materials for improving the luminous transmittance of ${\mathrm{VO}}_2$-based SWs.

AR Materials	n at $\mathit{\lambda }$ = 550 nm	Design and Architecture
${\mathrm{SiO}}_2$ [166]	∼1.45	Top AR
${\mathrm{TiO}}_2$ [164]	$2.40$–$2.60$	Top and bottom AR, self-cleaning, photocatalyst
${\mathrm{SnO}}_2$ [167]	$1.9$–$2.0$	Top and bottom AR
$\mathrm{ZnO}$ [168]	$1.9$–$2.0$	Top and bottom AR
${\mathrm{Cr}}_2{\mathrm{O}}_3$ [169]	$2.40$–$2.60$	Bottom AR + UV shielding
${\mathrm{WO}}_3$ [170]	$2.00$–$2.10$	Top and bottom AR
${\mathrm{ZrO}}_2$ [171]	$2.05$–$2.15$	Top and bottom AR
$\mathrm{AlN}$ [172]	$2.00$–$2.10$	Top and bottom AR

and can be selected depending on the target optical and functional requirements of the device. The optimum thickness and the appropriate refractive index of these layers can be evaluated by means of the transfer matrix method (TMM), which is widely used for the optical design and optimization of multilayer coatings. Thus, multilayer engineering and the integration of AR coatings provide an effective route for improving the optical performance of VO₂-based methods.

Thus, a practical route toward high-performance thermochromic glazing is obtained by combining two complementary optimization strategies: chemical tuning of the active VO₂ by doping and optical tuning of the complete coating stack by AR, buffer, and protective layers. Doping mainly controls the intrinsic thermochromic response of VO₂, especially the $T_c$, visible absorption, and NIR modulation, whereas AR and buffer layers control reflection losses, optical interference, crystallization, interfacial strain, and long-term stability. Therefore, the most successful VO₂-based SW coatings are multilayer structures in which composition, thickness, morphology, refractive-index, and interface quality are optimized simultaneously [164,165,166].

Becker et al. [164] investigated the role of TiO₂ AR and buffer layer in undoped VO₂ film in order to separate the effect of optical multilayer engineering from doping of the VO₂ layer. The choice of TiO₂ is motivated by the following considerations. Rutile TiO₂ is structurally compatible with the high-temperature rutile phase of VO₂ and can therefore promote the formation of the desired thermochromic phase, whereas TiO₂ top layers can act as AR coatings for reducing reflection losses and increasing $T_{\mathrm{lum}}$. In this concept, the bottom rutile TiO₂ layer acts mainly as a buffer layer, while the top anatase TiO₂ layer acts as an AR layer. The coating was prepared by ion-beam sputter deposition (IBSD) on quartz glass, using a customized ceramic TiO₂ target for the deposition of both rutile and anatase TiO₂ layers and a metallic V target with a purity of $99.9\%$ for the VO₂ layer. This technique was selected because it allows the sequential deposition of oxide thin films under controlled energetic ion bombardment, which is important for stabilizing the desired TiO₂ and VO₂ phases and for controlling the morphology and interface quality of the multilayer stack. In the first step, R-TiO₂, VO₂, and VO₂/R-TiO₂ films were deposited on quartz glass at $550°$C, providing the reference structures required to evaluate the influence of the rutile TiO₂ buffer layer on the growth and thermochromic response of VO₂. The optical constants, namely the refractive index n and extinction coefficient k, were determined by spectroscopic ellipsometry at $25°$C and $90°$C, corresponding to the low- and high-temperature states of VO₂. These experimentally determined optical constants were then used in Macleod simulations to optimize the thicknesses of the R-TiO₂ buffer layer and the A-TiO₂ AR layer.

For a fixed VO₂ thickness of 50 nm, Becker et al. calculated the thermochromic parameters as a function of the rutile-TiO₂ buffer thickness and the anatase-TiO₂ AR thickness in the range 0–300 nm. The optimized ideal architecture was obtained for a rutile-TiO₂ buffer thickness of approximately 152 nm and an anatase-TiO₂ AR thickness of approximately 178 nm, giving predicted enhancements of about $8\%$ in $\Delta T_{\mathrm{sol}}$ and about $25\%$ in $T_{\mathrm{lum}}$ compared with a bare VO₂ layer on quartz glass. Experimentally, the realized trilayer coating improved both $T_{\mathrm{lum}}$ and $\Delta T_{\mathrm{sol}}$ by approximately 5–$8\%$. However, the authors showed that using rutile TiO₂ as the top AR layer is problematic, because the oxygen-rich and high-temperature conditions required for rutile growth may oxidize the underlying VO₂ toward higher vanadium oxidation states, including V₂O₅. For this reason, anatase TiO₂ was selected as the top AR layer, since it can be deposited under milder conditions while still providing AR and capping functions.

A key conclusion of Becker et al. is that ideal optical simulations overestimate the real performance if interface roughness is neglected. In the ideal case, the optimized multilayer stack gave a simulated $T_{\mathrm{lum}}$ of about $45\%$, whereas the inclusion of effective-medium interface layers with thicknesses of about 4–5 nm at the rutile-TiO₂/VO₂ interface and about 8–10 nm at the VO₂/anatase-TiO₂ interface reduced the calculated $T_{\mathrm{lum}}$ to approximately 35–$37\%$, in better agreement with experiment. Thus, Becker et al. demonstrated that TiO₂ buffer and AR layers can improve the optical response of VO₂-based coatings, but also that AR engineering alone is insufficient to reach the practical SW requirements of $T_{\mathrm{lum}}>60\%$ and a transition temperature close to room temperature [164] (Figure 17a).

A more complete approach was reported by Kaufman et al., who combined controlled W/Sr codoping of VO₂ with a second-order AR multilayer design [165]. The active layer composition was ${\mathrm{V}}_{0.855}$${\mathrm{W}}_{0.018}$${\mathrm{Sr}}_{0.127}$O₂ (Figure 17d). W was chosen because it is one of the most effective dopants for reducing the $T_c$ of VO₂, mainly by destabilizing the low-temperature semiconducting phase and increasing the electron concentration. Sr was chosen because it improves the visible optical properties by widening the visible-range optical gap and lowering the extinction coefficient in the visible spectral range. In particular, the incorporation of $12.7at.\%$ Sr into the metal sublattice increased the visible-range gap $E_{\mathrm{g}1}$ from $1.51$ to 1.75 eV, decreased $n_{550}$ from $2.84$–$3.13$ to $2.38$–$2.58$, and decreased $k_{550}$ from $0.45$–$0.47$ to $0.29$–$0.26$. These changes explain why Sr strongly increases $T_{\mathrm{lum}}$, while W keeps the transition temperature near room temperature.

The optimized coating architecture reported by Kaufman et al. [165] was YSZ (167 nm)/${\mathrm{V}}_{0.855}$${\mathrm{W}}_{0.018}$${\mathrm{Sr}}_{0.127}$O₂ (71 nm)/SiO₂ (280 nm) on 1 mm SLG. The coating was prepared by reactive high-power impulse magnetron sputtering (HiPIMS) at a substrate temperature of $320°$C and without substrate bias voltage. The top SiO₂ layer was selected because Sr lowers the refractive index of the active layer, making the optimum top-AR refractive index about 1.54–1.61, which is much closer to SiO₂ with $n_{550}\approx 1.46$ than to high-index oxides such as YSZ with $n_{550}\approx 2.21$.

Kaufman et al. deliberately used second-order interference in the AR layers rather than only maximizing the first-order visible-transmittance maximum. For a thermochromic layer thickness of 71 nm, optical modeling gave optimum AR thicknesses close to ∼170 nm for YSZ and ∼269 nm for SiO₂, which is consistent with the experimentally realized 167 nm/71 nm/280 nm structure. This design is important because the second-order maximum in the visible region can be combined with a favorable first-order maximum in the infrared region, thereby improving $\Delta T_{\mathrm{sol}}$ without sacrificing too much $T_{\mathrm{lum}}$. The same modeling also showed the trade-off associated with the thermochromic layer thickness: increasing the ${\mathrm{V}}_{0.855}$${\mathrm{W}}_{0.018}$${\mathrm{Sr}}_{0.127}$O₂ thickness from 40 to 100 nm decreases the predicted $T_{\mathrm{lum}}$ from $72.6\%$ to $49.6\%$, but increases the predicted $\Delta T_{\mathrm{sol}}$ from $7.41\%$ to $13.61\%$ (Figure 17e). Thus, the selected thickness of 71 nm represents a compromise between high visible transmittance and strong solar modulation.

The optimized W/Sr-codoped coating achieved $T_c=22°$C, $T_{\mathrm{lum}}=63.7\%$ in the low-temperature state, $T_{\mathrm{lum}}=60.7\%$ in the high-temperature state, $T_{\mathrm{sol}}=58.8\%$ in the low-temperature state, $T_{\mathrm{sol}}=47.6\%$ in the high-temperature state, and $\Delta T_{\mathrm{sol}}=11.2\%$. For comparison, the W-only reference coating YSZ (178 nm)/${\mathrm{V}}_{0.984}$${\mathrm{W}}_{0.016}$O₂ (73 nm)/SiO₂ (280 nm) showed only $T_{\mathrm{lum}}=47.0\%$ and $45.3\%$ in the low- and high-temperature states, respectively, with $\Delta T_{\mathrm{sol}}=8.4\%$. Therefore, Sr doping increased the low-temperature $T_{\mathrm{lum}}$ by $16.7\%$ and the high-temperature $T_{\mathrm{lum}}$ by $15.4\%$, while also increasing $\Delta T_{\mathrm{sol}}$ by $2.8\%$ (Figure 17c). This result shows that W/Sr codoping combined with YSZ/SiO₂ AR design can simultaneously satisfy the practical thresholds of $T_{\mathrm{lum}}>60\%$, $\Delta T_{\mathrm{sol}}>10\%$, and $T_c$ close to room temperature.

A further important contribution in this direction was reported by Vlček et al. [166], who demonstrated that the thermochromic performance of VO₂-based coatings can be significantly improved by replacing the compact active layer with W-doped VO₂ nanoparticles embedded in a SiO₂ matrix. W was again selected to reduce the $T_c$, while the nanoparticle morphology was selected to reduce the effective visible-range refractive index and extinction coefficient. The SiO₂ matrix and overlayer were chosen because SiO₂ has a low refractive index, $n\approx 1.46$ at $\lambda$ = 550 nm, negligible absorption with $k<{10}^{-3}$ at $\lambda$ = 550 nm, and can simultaneously act as a Na-diffusion barrier, low-index matrix, AR medium, and protective layer. In addition, the metallic VO₂(R) nanoparticles can enhance NIR absorption through localized surface plasmon resonance, which increases $\Delta T_{\mathrm{sol}}$ without excessively suppressing visible transmittance.

The coating reported by Vlček et al. was prepared on 1 mm SLG at a maximum substrate temperature of $350°$C. The structure consisted of a bottom ∼100 nm SiO₂ layer blocking Na diffusion from the glass substrate, four W-doped VO₂ nanoparticle layers separated by three ∼50 nm SiO₂ layers, and a top ∼160 nm SiO₂ AR/protective overlayer. Each nanoparticle layer was obtained by depositing a ∼10 nm V–W film, annealing it up to $350°$C for 1 h in pure oxygen at $p_{O_2}=15$ Pa, cooling below $50°$C, and depositing the next SiO₂ layer. The V–W films were deposited from a V–W target containing $4.0wt.\%$ W, corresponding to $1.14at.\%$ W, and the W-induced $T_c$ gradient was at least $-24°$C/at.% W.

TEM and HRTEM showed W-doped VO₂ nanoparticles embedded in the SiO₂ matrix, with representative vertical nanoparticle sizes of about 42 nm in the upper W–VO₂ layer and about 22 nm in the lower W–VO₂ layer (Figure 18a inset). SEM analysis showed that most nanoparticles had diameters of 41–55 nm and interparticle spacings of 7–21 nm, while some elongated particles reached lengths of 150–400 nm or even about 1 $\mathsf{\mu }$m.

The gradual increase in the number of W-doped VO₂ layers produced the expected trade-off between visible transmittance and solar modulation. For one, two, three, and four nanoparticle layers with approximately 50 nm SiO₂ overlayers, $\Delta T_{\mathrm{sol}}$ increased from $3.1\%$ to $4.7\%$, $6.3\%$, and $11.1\%$, respectively, while the low-temperature $T_{\mathrm{lum}}$ decreased from $76.3\%$ to $71.9\%$, $64.9\%$, and $56.8\%$. Replacing the ∼50 nm top SiO₂ overlayer in the four-layer coating with an optimized ∼160 nm top SiO₂ AR/protective layer strongly improved the performance. The final VWO-4 coating reached $T_{\mathrm{lum}}=65.4\%$ at low temperature, $T_{\mathrm{lum}}=60.1\%$ at high temperature, $T_{\mathrm{sol}}=66.1\%$ at low temperature, $T_{\mathrm{sol}}=50.8\%$ at high temperature, $\Delta T_{\mathrm{lum}}=5.3\%$, $\Delta T_{\mathrm{sol}}=15.3\%$, and $T_c=33°$C [166]. This is the highest $\Delta T_{\mathrm{sol}}$ among the three selected approaches while still maintaining $T_{\mathrm{lum}}>60\%$ in both thermochromic states (Figure 18).

Overall, these studies illustrate a clear progression in VO₂-based SW design. Becker et al. [164] demonstrated that buffer and AR layers can improve the optical performance of undoped VO₂, but the resulting $T_{\mathrm{lum}}$ remains limited by the intrinsic absorption of the active layer and by interface roughness. Kaufman et al. solved this limitation more effectively by combining W/Sr codoping with second-order YSZ/SiO₂ AR design, reaching $T_c=22°$C, $T_{\mathrm{lum}}>60\%$, and $\Delta T_{\mathrm{sol}}>10\%$ in a compact multilayer coating prepared at $320°$C [165]. Vlček et al. achieved the highest solar modulation by introducing W-doped VO₂ nanoparticles into a SiO₂ matrix, where the combination of reduced effective optical constants, localized near-infrared absorption, and optimized SiO₂ AR/protection produced $\Delta T_{\mathrm{sol}}=15.3\%$ while preserving $T_{\mathrm{lum}}>60\%$ [166]. Thus, the most promising direction is the simultaneous optimization of dopant chemistry, optical interference, layer morphology, and interface stability rather than the use of a single modification strategy.

In comparison with VO₂ and other inorganic thermochromic material systems, PNIPAm-based hydrogels generally show superior optical performance. PNIPAm-AEMA hydrogel microparticles have shown $T_{\mathrm{lum}}=87.2\%$, $\Delta T_{\mathrm{sol}}=81.3\%$, and a lower critical solution temperature (LCST) of around $32°\mathrm{C}$ [173]. HBPEC/PNIPAm hydrogels reached $T_{\mathrm{lum}}=87.5\%$, $\Delta T_{\mathrm{sol}}=71.2\%$, and a tunable transition range of $24.1$–$33.2°\mathrm{C}$ [174]. PNIPAm/HEMC hydrogels were also reported with $T_{\mathrm{lum}}\approx 85\%$ and $\Delta T_{\mathrm{sol}}\approx 75\%$ [149]. These values are substantially higher than those of most VO₂ coatings, for which $\Delta T_{\mathrm{sol}}$ typically remains below $20\%$. However, hydrogel systems face different device-integration challenges, including encapsulation, water retention, drying/freezing resistance, mechanical stability during swelling/deswelling, and long-term outdoor durability.

Emerging thermochromic systems occupy an intermediate position between VO₂ and hydrogel-based systems. Metal-halide perovskite SWs based on CH₃NH₃PbI₃ can exhibit visible transmittance above $85\%$ in the cold state and about $34.3\%$ in the hot state, with $\Delta T_{\mathrm{sol}}=25.5\%$ and a transition temperature of approximately $54.4°$C [175]. Hydrated ionic polymers, such as ${\left[{\left({\mathrm{C}}_2{\mathrm{H}}_5\right)}_2{\mathrm{NH}}_2\right]}_2{\mathrm{NiCl}}_4$@PVP-type systems, combine $T_{\mathrm{lum}}=87.7\%$, $\Delta T_{\mathrm{sol}}=30.5\%$, and a tunable transition temperature of 25–$42°$C [176]. Commercial IR-selective thermochromic coatings, such as SunSmart coatings, show more moderate but application-oriented values, with $T_{\mathrm{lum}}\approx 70$–$75\%$ and $\Delta T_{\mathrm{sol}}\approx 20$–$23\%$ [177]. Cholesteric liquid-crystal thermochromic films are promising for selective NIR modulation and composition-tunable optical response, although their performance metrics are less standardized than those of VO₂ or PNIPAm-based systems [178].

These comparisons indicate that the most effective SW designs will likely rely on combining the strengths of different material classes rather than selecting a single universal material system. VO₂ offers inorganic stability, compatibility with thin film processing, and strong NIR switching, but it still requires simultaneous improvement of $T_c$, $T_{\mathrm{lum}}$, $\Delta T_{\mathrm{sol}}$, color neutrality, interface quality, and scalable deposition. Hydrogels provide the highest optical performance, with $T_{\mathrm{lum}}>85\%$ and $\Delta T_{\mathrm{sol}}>70\%$, but their long-term durability and encapsulation remain unresolved. Perovskites and hydrated ionic polymers offer high contrast and tunable transition temperatures, but they require further validation with respect to toxicity, humidity resistance, UV stability, hysteresis, and cycling lifetime. Commercial IR-selective coatings are closer to real glazing implementation, but their optical modulation is lower than that of laboratory hydrogel systems.

4.2. Electrochromic WO₃- and NiO-Based Smart Windows

Among electrochromic materials for SW applications, TMOs remain the most extensively investigated class because they combine reversible optical modulation, chemical robustness, thermal stability, and compatibility with conventional thin film fabrication methods. In TMO-based electrochromic SWs, the optical response is governed by electrically driven redox reactions coupled with the insertion/extraction of ions, such as ${\mathrm{H}}^{+}$, ${\mathrm{Li}}^{+}$, ${\mathrm{Na}}^{+}$. Therefore, the electrochromic performance is determined not only by the intrinsic optical contrast of the active oxide, but also by ion-diffusion kinetics, CT reversibility, electrolyte compatibility, and long-term cycling stability.

WO₃ serves as the benchmark cathodic electrochromic oxide. Upon application of a suitable voltage, electrons are injected into the WO₃ layer while charge-compensating ions from the electrolyte are inserted into its structure, forming a colored tungsten-bronze-like state, generally described as ${\mathrm{M}}_x$WO₃, where M represents the inserted ion. When the voltage polarity is reversed, ions and electrons are extracted and the material returns to its bleached transparent state. In contrast, NiO is the prototypical anodic oxide counterpart, where coloration is associated with oxidation of Ni²⁺ toward higher-valence nickel states, such as Ni³⁺ and Ni⁴⁺. Consequently, the complementary integration of cathodic WO₃ and anodic NiO remains one of the standard configurations for high-contrast electrochromic SWs.

Besides the benchmark WO₃/NiO pair, other TMOs can also contribute to electrochromic functionality. V₂O₅ is attractive because its layered framework enables ion intercalation, multicolor switching, and infrared modulation, although structural stabilization is often required to improve cycling durability. TiO₂, especially in anatase or nanostructured form, is frequently employed either as a weak active electrochromic component or, more importantly, as an ion-storage, passivation, or interfacial layer in all-inorganic and multilayer electrochromic architectures. TMOs are commonly classified into cathodically coloring and anodically coloring materials, depending on the polarity and redox process responsible for coloration [179] (Figure 19).

A typical electrochromic SW consists of two transparent substrates coated with transparent conducting layers, between which an electrochromic layer, an ion-conducting electrolyte, and a complementary counter-electrode or ion-storage layer are arranged in a sandwich-type architecture. Under an applied voltage, ions migrate through the electrolyte toward one electrode, while electrons are supplied through the external circuit. In the colored state, ion/electron insertion into the cathodic layer, most commonly WO₃, increases optical absorption in the VIS and/or NIR region. At the same time, the anodic counter electrode, typically NiO or a Ni-containing oxide, balances the charge and can provide complementary coloration. Reversing the voltage extracts the inserted ions and electrons, restoring the bleached state. Therefore, the optical modulation of an electrochromic SW is a device-level response controlled by the active oxide layers, the transparent conductors, the electrolyte, and the stability of their interfaces (Figure 20).

The electrolyte controls ionic conductivity, switching speed, charge compensation, optical memory, leakage resistance, and interfacial degradation. Liquid and gel electrolytes generally provide high ionic conductivity and faster switching, but may suffer from leakage, evaporation, gas formation, and sealing problems. In contrast, solid polymer, inorganic, and hybrid electrolytes are more attractive for all-solid-state laminated windows because they improve mechanical integrity and device stability, although their ionic conductivity can be lower. Typical electrolyte types used in TMO-based electrochromic windows are summarized in

Table 18. Electrolytes used in TMO-based electrochromic windows.

Electrolyte Type	Typical Examples	Common Mobile Ions	Notes for TMO Electrochromic SW
Liquid/liquid-polymer electrolytes (LPEs) [180]	Propylene carbonate (PC), polyethylene glycol (PEG), polyethylene oxide (PEO) with salts such as LiClO4, LiI, LiTFSI, LiPF6	Mainly Li+	High ionic conductivity and easy processing, but limited by leakage, evaporation, bubble formation, and sealing/safety issues
Gel polymer electrolytes (GPEs)/quasi-solid electrolytes [180,181]	PVDF-HFP-based gels, PEGDA/PEO gels, UV-cured PMMA gels, ion gels	Mainly Li+	High ionic conductivity, better mechanical stability, and reduced leakage; common in WO3/NiO devices
Solid polymer electrolytes (SPEs) [180]	PMMA, gelatin, methyl-cellulose-based electrolytes, bio-based polymer electrolytes	Li+ or H+	Attractive for all-solid-state SWs because they eliminate leakage and simplify lamination/packaging, although conductivity can be lower than in liquid systems
Inorganic solid electrolytes [180,182]	Ta2O5, Li:Ta2O5, LiAlSiO4, phosphate glass, related oxide/ceramic ion conductors	H+, Li+, Na+ depending on composition	Important in all-solid-state SWs because they offer good film integration, durability, and device stability
Hybrid organic–inorganic polyelectrolytes [183]	Sol–gel-derived hybrid electrolytes, ORMOLYTE-type materials, siloxane-based hybrid polyelectrolytes	Mainly Li+	Designed to combine the mechanical robustness of inorganic networks with the processability and flexibility of polymers; reported in NiO/WO3 complementary devices

.

TMOs practical relevance cannot be assessed only from the maximum optical contrast. Recent WO₃ nanosheet/ITO nanocomposites reached optical modulation values up to $\Delta T\approx 88\%$ and a coloration efficiency (CE) of $154.16{\mathrm{cm}}^2{\mathrm{C}}^{-1}$ [184]. Rare-earth-modified WO₃ also shows strong performance: Tb-doped amorphous WO₃ exhibited an optical contrast of $78.34\%$, a CE of $138.82{\mathrm{cm}}^2{\mathrm{C}}^{-1}$, and coloration/bleaching times of $16.0/11.4\mathrm{s}$ [185]. These values are high enough to justify the continued dominance of WO₃ in electrochromic SW research, especially when compared with many other oxide systems. Nevertheless, WO₃ suffers from the classical limitations of cathodic oxide electrochromics: repeated H⁺ or Li⁺ insertion/extraction can produce ion trapping, defect accumulation, local structural disorder, incomplete bleaching, and gradual loss of optical reversibility. Therefore, a WO₃ film showing $\Delta T>70\%$ in the first cycles is not necessarily suitable for practical glazing unless the same modulation is retained over thousands of cycles.

Recent improvements in WO₃-based devices clearly show that interface engineering is more effective than simple single-layer optimization. For example, WO₃/TiO₂ electrochromic devices achieved $\Delta T=73.9\%$ at $600\mathrm{nm}$, CE up to $69{\mathrm{cm}}^2{\mathrm{C}}^{-1}$, coloration times of about 20–$21\mathrm{s}$, and stable operation over more than 1000 cycles [186]. More advanced TiO₂/WO₃/TiO₂ double-heterojunction structures achieved a much higher CE of $479.3{\mathrm{cm}}^2{\mathrm{C}}^{-1}$ and retained $94.72\%$ of their transmission modulation after 7000 cycles under recovery conditions [187]. This comparison is important because the CE of the TiO₂/WO₃/TiO₂ heterostructure is more than three times higher than that of the WO₃ nanosheet/ITO nanocomposite and more than six times higher than the WO₃/TiO₂ device. Although TiO₂ alone is usually a weak visible electrochromic material, in multilayer devices it acts as an ion-storage, passivation, or interfacial layer that suppresses irreversible ion accumulation and improves CT reversibility.

NiO is the most important anodic oxide counterpart to WO₃ in complementary electrochromic devices. Recent W⁶⁺-doped NiO films showed an optical modulation of $61.56\%$ at $550\mathrm{nm}$, a CE of $45.41{\mathrm{cm}}^2{\mathrm{C}}^{-1}$, fast switching times of $4.42/1.40\mathrm{s}$ for coloration/bleaching, and stable operation over 2000 cycles [188]. These figures indicate that NiO can switch faster than many WO₃ systems, but its CE remains lower than that of optimized WO₃, Prussian Blue, PANI, or PEDOT:PSS. Thus, NiO is not generally selected because it provides the highest CE, but because it is one of the most suitable anodic charge-balancing layers for WO₃-based full devices. Its main limitations are its relatively low electrical conductivity, moderate CT kinetics, and degradation during prolonged cycling, which explains why doping, defect engineering, and nanostructuring are required for practical SW integration.

Alternative Ni-based compounds can provide stronger optical modulation than conventional NiO. Electrodeposited ${\mathrm{NiHPO}}_4·$3H₂O films demonstrated an optical modulation of $90.8\%$ at $500\mathrm{nm}$ and a CE of $75.4{\mathrm{cm}}^2{\mathrm{C}}^{-1}$, while also enabling a large-area electrochromic device with an active area of $100{\mathrm{cm}}^2$ [189]. The $\Delta T$ value of $90.8\%$ is higher than that reported for the WO₃ nanosheet/ITO system, Tb-doped WO₃, W-doped NiO, V₂O₅, PANI, PEDOT:PSS, and viologen/PEDOT:PSS/Zn devices considered here. From a manufacturing perspective, electrodeposition is also attractive because it is lower-cost and more scalable than pulsed laser deposition or high-vacuum multilayer deposition. However, the long-term outdoor durability of nickel phosphate under humidity, UV exposure, thermal cycling, and realistic electrolyte conditions still requires further validation.

TiO₂ should be considered separately from WO₃ and NiO because its role in SWs is often interfacial rather than purely electrochromic. TiO₂-based heterostructures, such as 2D TiO₂/Ti₃${\mathrm{C}}_2{T}_x$ MXene systems, have been reported to provide high optical modulation, fast response, and good mechanical stability in flexible devices [190]. In Au nanoparticle-loaded TiO₂ nanofibrous layers, the optical modulation was approximately $\Delta T\approx 40\%$, the CE was about $20{\mathrm{cm}}^2{\mathrm{C}}^{-1}$, the switching time was around $6\mathrm{s}$, and the devices operated for more than 1000 cycles [191]. These values are lower than those of optimized WO₃ or conducting-polymer systems, but TiO₂ remains highly valuable because it can improve cycling stability, protect active layers, assist ion storage, and enable flexible or nanostructured device architectures.

V₂O₅ is attractive because it offers multicolor electrochromism and very fast switching. Annealed V₂O₅ films showed a bleached transmittance of $T_{\mathrm{b}}=71.89\%$, a colored transmittance of $T_{\mathrm{col}}=29.57\%$, an optical modulation of $\Delta T=42.32\%$, a CE of $34.93{\mathrm{cm}}^2{\mathrm{C}}^{-1}$, and very short coloration/bleaching times of $0.4/3.0\mathrm{s}$ [192]. The switching speed of V₂O₅ is therefore better than that of many WO₃-based devices and comparable to fast polymer systems. However, its optical contrast can degrade significantly after only about 100 cycles, which is far below the 2000 cycles reported for W-doped NiO, the 7000 cycles reported for TiO₂/WO₃/TiO₂, and the $\mathrm{10,000}$ cycles reported for PANI-based devices. This makes V₂O₅ promising for rapid or multicolor electrochromic functions, but still problematic for long-lifetime architectural glazing.

When TMOs are compared with non-oxide electrochromic materials, the differences become even clearer. Nanogranular Prussian Blue films reached optical modulation values of about $80\%$, CEs as high as $417.79{\mathrm{cm}}^2{\mathrm{C}}^{-1}$, and $91.4\%$ optical contrast retention [193]. Thus, the CE of Prussian Blue is approximately $2.7$ times higher than that of the WO₃ nanosheet/ITO system, about $9.2$ times higher than that of W-doped NiO, and about 12 times higher than that of V₂O₅. This demonstrates the strong charge-utilization advantage of Prussian Blue, although its dominant blue coloration and electrolyte sensitivity can limit its use in color-neutral architectural glazing.

Conducting polymers also show highly competitive electrochromic performance. Polyaniline-based dual-band electrochromic SWs demonstrated $\Delta T=65\%$ at $633\mathrm{nm}$ and $\Delta T=59\%$ at $1600\mathrm{nm}$, with CEs of $367.1{\mathrm{cm}}^2{\mathrm{C}}^{-1}$ in the visible region and $299.6{\mathrm{cm}}^2{\mathrm{C}}^{-1}$ in the NIR region, together with stability exceeding $\mathrm{10,000}$ cycles [194]. This is particularly important because dual-band modulation allows independent control of visible light and solar heat, a functionality that most single-component oxide systems cannot easily provide. PEDOT:PSS-based devices achieved an optical contrast of $47.9\%$ at $635\mathrm{nm}$, a CE of $352{\mathrm{cm}}^2{\mathrm{C}}^{-1}$, and very fast switching times of $1.6/0.6\mathrm{s}$ [195]. Although the optical contrast of PEDOT:PSS is lower than that of WO₃, Prussian Blue, or PANI, its switching speed and solution processability make it attractive for printed, flexible, and low-temperature electrochromic devices.

Viologen-based hybrid systems further illustrate the advantages of organic electrochromics. Viologen/PEDOT:PSS/Zn devices reached an optical difference of $64.23\%$, a CE of $389.44{\mathrm{cm}}^2{\mathrm{C}}^{-1}$, switching times of $1.6/0.8\mathrm{s}$, and $90.58\%$ retention of $\Delta T$ [196]. The CE of this hybrid system is higher than that of WO₃, NiO, nickel phosphate, TiO₂, and V₂O₅, and its switching is faster than most oxide systems. However, aqueous organic–metal hybrid devices may suffer from swelling, electrolyte leakage, dissolution, interfacial instability, and degradation of organic redox species during prolonged outdoor exposure. Therefore, their excellent short-term electrochemical performance must be balanced against environmental durability requirements.

A direct performance comparison therefore shows that each material class has a distinct advantage. WO₃ provides high optical modulation, with values up to $\Delta T\approx 88\%$, and remains the most reliable cathodic oxide. NiO provides fast anodic switching, with $4.42/1.40\mathrm{s}$ switching times and 2000-cycle stability, but lower CE. TiO₂ improves interfaces and cycling stability rather than acting as the strongest active electrochromic layer. V₂O₅ provides very fast switching, down to $0.4\mathrm{s}$ for coloration, but poor cycling stability. Nickel phosphate provides very high optical modulation of $90.8\%$ and scalable electrodeposition. Prussian Blue, PANI, PEDOT:PSS, and viologen systems provide much higher CE values, typically in the range of 352–$417.79{\mathrm{cm}}^2{\mathrm{C}}^{-1}$, but their environmental stability is generally more problematic than that of inorganic oxides.

From the viewpoint of deposition and manufacturing, TMOs are compatible with magnetron sputtering, PLD, ALD, sol–gel processing, and electrodeposition. Vacuum-based methods such as sputtering and ALD offer dense, uniform, and reproducible coatings, which are essential for large-area optical uniformity, but they increase equipment cost and may limit low-cost manufacturing. PLD is useful for high-quality research films but is less attractive for industrial-scale glazing. Sol–gel processing and electrodeposition are more scalable and cost-effective, as illustrated by the $100{\mathrm{cm}}^2$ nickel phosphate device, but they require strict control of film porosity, adhesion, crystallinity, and post-treatment. Processing temperature is also critical: high-temperature annealing can improve oxide crystallinity and cycling stability, whereas low-temperature processing is required for flexible substrates, polymers, and laminated window structures.

At the device level, electrolyte compatibility is as important as the active electrochromic material. H⁺, Li⁺, Na⁺, Zn²⁺, and other mobile ions can provide electrochromic switching, but they also determine diffusion kinetics, reversibility, charge compensation, interfacial stability, and degradation pathways. Dense WO₃ films may suffer from slow ion diffusion and ion trapping, whereas organic and hydrogel-like devices may suffer from swelling, leakage, or dissolution. A device with high CE is therefore not automatically superior if the electrolyte causes gas evolution, parasitic redox reactions, delamination, or rapid loss of optical memory. For practical SWs, the target is not simply the highest $\Delta T$ or CE, but the best combination of optical contrast, response speed, cycling lifetime, color neutrality, haze control, sealing, and compatibility with large-area laminated glazing.

These comparison shows that no single electrochromic material currently maximizes all relevant parameters. WO₃ remains the benchmark cathodic oxide because it provides high optical modulation and relatively strong durability. NiO remains the preferred anodic oxide because it enables complementary WO₃-based devices. TiO₂ is most valuable as an interfacial and stabilizing component. V₂O₅ is attractive for rapid and multicolor switching but is limited by cycling degradation.

For commercial applications, ITO, FTO, AZO, and PEDOT or PEDOT-based transparent conducting coatings are widely used.

At present, a relatively large number of both major and smaller companies are involved in the production of electrochromic SW.

Table 19. Commercial electrochromic glazing products and reported optical performance.

Manufacturer	Product Name	Maximum Size (mm)	${\mathit{T}}_{\mathbf{lum}}$ Range	${\mathit{T}}_{\mathbf{sol}}$ Range
SageGlass www.sageglass.com (accessed date: 19 May 2026)	SageGlass Clear DGU	$3095\times 1828$	$1\%\to 60\%$	$0.4\%\to 33\%$
	SageGlass Blue DGU	$3095\times 1828$	$0.5\%\to 40\%$	$0.3\%\to 21\%$
View https://view.com/ (accessed date: 19 May 2026)	View Gen 4 DGU–Clear	$1828\times 3048$	$1\%\to 53\%$	$0\%\to 29\%$
	View Gen 4 DGU–Clear + SN68 low-e	$1828\times 3048$	$1\%\to 45\%$	$0\%\to 19\%$
	View Gen 4 Laminated DGU–clear/${0.060}^{″}$ PVB/clear	$1828\times 3048$	$1\%\to 52\%$	$0\%\to 27\%$
	View Gen 4 TGU–Clear $\times 2$	$1828\times 3048$	$1\%\to 47\%$	$0\%\to 24\%$
	View Gen 4 TGU–Clear $\times 2$ + SN68 low-e	$1828\times 3048$	$1\%\to 40\%$	$0\%\to 16\%$
Vitrum Glass Group https://www.vitrum.ca/ (accessed date: 19 May 2026)	Halio	$3058\times 1508$	$2\%\to 65\%$	–
	Halio Black	$3058\times 1508$	$0.01\%\to 52\%$	–
ConverLight https://converlight.com/ (accessed date: 19 May 2026)	ConverLight Dynamic 75 2G	$1550\times 4400$	$36\%\to 67\%$	$21\%\to 41\%$
	ConverLight Dynamic 75 3G	$1550\times 4400$	$33\%\to 61\%$	$17\%\to 34\%$
	ConverLight Dynamic 65 3G	$1550\times 4400$	$14\%\to 56\%$	$7\%\to 31\%$
	ConverLight Dynamic 65 4G	$1550\times 4400$	$13\%\to 51\%$	$6\%\to 26\%$

summarizes selected commercial products from four companies chosen for this review, as they represent different technological, commercial, and manufacturing approaches within the SW sector. A common feature of these products is the use of a WO₃/NiO cathode–anode configuration.

However, the exact layer chemistry used in commercial electrochromic SWs is rarely disclosed in current product literature and is generally treated as proprietary. Publicly available information is therefore often fragmentary and must be reconstructed from patents, older technical brochures, and review papers. Taken together, these sources suggest that many commercially relevant inorganic electrochromic SW platforms rely on a tungsten-oxide-based cathodic layer combined with a nickel-oxide-based, nickel-containing, or mixed tungsten–nickel anodic/counter electrode, frequently in lithiated or otherwise compositionally modified form rather than as strictly stoichiometric WO₃ and NiO. Accordingly, the systems listed in Table 19 are best described as devices that, with a high degree of confidence, appear to belong to the broader WO₃/NiO materials family.

4.3. Optical Hydrogen Detection Using TMO-Based Gasochromic Sensors

Hydrogen leakage detection is a critical safety requirement because ${\mathrm{H}}_2$ [197]: (i) is flammable over a wide concentration range in air, typically ∼4–$75\mathrm{vol}.\%$; (ii) exhibits an extremely low minimum ignition energy, $0.017\mathrm{mJ}$; (iii) is colorless, odorless, and tasteless, making leaks difficult to notice without instrumentation; (iv) can cause hydrogen embrittlement of structural materials, thereby increasing the probability of failures and leakage.

Gasochromic hydrogen sensors [197] provide an optical route to leak detection by converting the gas–solid interaction into a reversible change of the materials’ optical response, such as transmittance, reflectance, absorbance, or perceived color. In TMO materials, the gasochromic response is commonly attributed to near-surface redox reactions, intercalation–deintercalation, which result in increased absorption in the VIS–NIR region. Gasochromic hydrogen sensors have several key features: (i) they can operate at room temperature and may not require external electrical biasing at the sensing element, reducing ignition risks; (ii) hydrogen exposure can induce a distinct optical contrast, including visible color changes in some systems, enabling direct visual indication and straightforward optical readout; (iii) the device concept can be implemented using simple thin film stacks and low-cost optical interrogation, e.g., photodiodes or cameras, which is attractive for scalable safety indicators.

Among TMOs, tungsten trioxide WO₃ is widely applicable as one of the most practically established materials for optical hydrogen detection [197]. This preference is largely driven by its pronounced and reversible optical modulation upon hydrogen exposure, typically a pale-yellow to deep-blue coloration [198,199]. In most device implementations, WO₃ is combined with a thin Pd or Pt catalytic overlayer to dissociate H₂ and promote spillover/insertion, enabling large changes in transmittance with good repeatability [200]. These features make Pd/Pt–WO₃ coatings a common choice for passive, optically readable hydrogen-leak indicators [199,201].

Vanadium oxides represent another important gasochromic TMO class for hydrogen optical sensing [197]. In particular, V₂O₅ thin films, often combined with a noble-metal catalyst such as Pt, exhibit a pronounced and reversible optical contrast upon H₂ exposure under ambient conditions, e.g., from yellow to gray/black, consistent with hydrogen-driven redox processes [202,203,204]. In parallel, VO₂ is frequently exploited in a distinct route, where the gas environment, hydrogenation, shifts the MIT and produces a large change in resistance [205,206]. The MIT in VO₂ is accompanied by a strong modulation of optical constants, especially in the NIR/MIR region, so a gas-induced MIT shift can, in principle, be read out optically as well, either as a gasochromic contrast or as an MIT-assisted transmittance/reflectance change [207].

In contrast to WO₃ and several vanadium oxides, NiO [208] and TiO₂ [209] are most commonly deployed as chemiresistive metal oxide (MOX) sensors, where the primary signal is a gas-induced change in conductivity rather than a pronounced, reversible optical contrast. Nevertheless, both oxides can still be integrated into optically readable sensor architectures at the device level. For NiO, its well-known anodic electrochromism under ion intercalation–deintercalation can be leveraged as a visual transducer in electrochromic sensor concepts, where a gas-driven electrochemical/potentiometric process is converted into a measurable optical modulation of the NiO layer [8,210]. For TiO₂, a practical route toward optical/low-temperature operation is photoactivated sensing, where UV illumination modifies surface charge chemistry and enables room-temperature or reduced-temperature gas detection, with optical control and straightforward optical interrogation schemes [209,211]. Finally, TiO₂ can develop Ti³⁺ or oxygen-vacancy centers under strongly reducing, hydrogenating treatments, producing blue/black optical states. This suggests a possible gas-induced optical pathway, although this typically requires elevated temperatures and careful engineering to ensure reversibility and selectivity [212,213].

5. Future Perspectives

The development of chromic TMO thin films is moving from the optimization of individual films toward the design of complete device architectures. The central issue is the simultaneous control of optical, ionic, electronic, mechanical, and environmental parameters within the same structure.

For thermochromic VO₂, the relevant performance window combines $T_c$ close to 20–$35°\mathrm{C}$, $T_{\mathrm{lum}}>60\%$, $\Delta T_{\mathrm{sol}}>10$–$15\%$, narrow hysteresis, and stable optical response during repeated heating and cooling. For electrochromic devices, the technologically meaningful range combines $\Delta T>60$–$80\%$, $CE>100{\mathrm{cm}}^2{\mathrm{C}}^{-1}$ where attainable, coloration/bleaching below 10–$20\mathrm{s}$ at device scale, and optical-modulation retention above $90\%$ after 5000–10,000 cycles. These combined targets are more restrictive than isolated record values and therefore represent a more realistic criterion for SW integration.

VO₂ demonstrates the balance between material improvement and optical trade-off. W/Sr-codoped VO₂ multilayers can reduce $T_c$ to $22°\mathrm{C}$ while maintaining $T_{\mathrm{lum}}=63.7\%$ in the low-temperature state and $60.7\%$ in the high-temperature state, with $T_{\mathrm{sol}}$ decreasing from $58.8\%$ to $47.6\%$ and $\Delta T_{\mathrm{sol}}=11.2\%$ [165]. W-doped VO₂/SiO₂ nanoparticle coatings reach $T_{\mathrm{lum}}=65.4\%$ and $60.1\%$ in the low- and high-temperature states, $\Delta T_{\mathrm{sol}}=15.3\%$, and $T_c=33°\mathrm{C}$ [166].

AI and machine-learning approaches provide a route for treating chromic oxides as multi-parameter systems. In NiO, recent machine-learning studies mainly concern optical-property prediction in RF-magnetron-sputtered or doped films [214,215]. Machine learning has been used to connect WO₃ sputtering parameters with electrochromic reversibility and blue-color persistence [216]. In an ammonium metatungstate/FeCl₂/D₂O a multilayer perceptron model trained on 25 devices enabled a device with $\Delta T=74\%$, coloration and bleaching times of $6.5$ and $13.5\mathrm{s}$, and stability above 1000 cycles [217]. Another machine-learning-guided all-liquid electrochromic device reached $\Delta T=62.6\%$ with response times of $5.7$ and $7.1\mathrm{s}$ [218]. Long short-term memory (LSTM) modelling has also been applied to cycle-life prediction, giving $\Delta T=43.95\%$, coloration/bleaching times of 7 and $8\mathrm{s}$, and $\Delta T=44.92\%$ after 1000 cycles [219]. Overall, AI is useful for device and multilayer structure optimization, and also for property prediction.

Interface engineering represents a decisive factor in both thermochromic and electrochromic oxide devices. In VO₂ multilayers, a few nanometers of interfacial roughness can change the optical response substantially. For TiO₂/VO₂/TiO₂ stacks, an ideal model predicted $T_{\mathrm{lum}}$ of about $45\%$, whereas the introduction of effective-medium interface layers of only 4–$5\mathrm{nm}$ and 8–$10\mathrm{nm}$ reduced the calculated $T_{\mathrm{lum}}$ to approximately 35–$37\%$ [164]. Thus, nominal layer thickness is not sufficient for describing real optical stacks. Interface roughness, interdiffusion width, oxidation-state gradients, and the optical constants of the actual multilayer determine the final coating performance [6,164]. In electrochromic devices, WO₃/electrolyte, NiO/electrolyte, TiO₂/WO₃, and transparent-conductor/oxide interfaces control CT resistance, ion trapping, incomplete bleaching, and long-term drift [186,220,221]. The high CE reported for TiO₂/WO₃/TiO₂ heterostructures shows that TiO₂ can operate as an active interfacial component that improves CT reversibility and suppresses irreversible ion accumulation [186,187].

Nanostructuring is effective only when optical gain, ion transport, and durability remain balanced. In W-doped VO₂/SiO₂ nanoparticle coatings, increasing the number of nanoparticle layers from one to four increased $\Delta T_{\mathrm{sol}}$ from $3.1\%$ to $11.1\%$, but decreased low-temperature $T_{\mathrm{lum}}$ from $76.3\%$ to $56.8\%$ [166]. After optimization of the SiO₂ AR/protective layer, the coating reached $T_{\mathrm{lum}}=65.4\%$, $\Delta T_{\mathrm{sol}}=15.3\%$, and $T_c=33°\mathrm{C}$ [166]. In electrochromic WO₃, NiO, and V₂O₅, nanosheets, porous films, nanofibers, and nanocomposites reduce ion-diffusion length and increase electrochemically active surface area [182,184,222,223,224]. At the same time, excessive porosity and uncontrolled open morphology can increase electrolyte penetration, local structural distortion, irreversible ion trapping, incomplete ion extraction, and cycling degradation [11,192,222]. V₂O₅ demonstrates this trade-off clearly: coloration and bleaching times of $0.4$ and $3.0\mathrm{s}$ are very fast, but degradation after about 100 cycles remains incompatible with long-life smart-window operation [192].

Flexible and large-area devices introduce constraints that are not visible in small-area cells. Commercial electrochromic glazing already reaches dimensions of several square meters (Table 19) In contrast, many laboratory TMO devices are tested on areas of only a few ${\mathrm{cm}}^2$. Scaling from ${\mathrm{cm}}^2$ to ${\mathrm{m}}^2$ changes the dominant limitations: sheet resistance, lateral voltage drop, electrolyte distribution, edge sealing, defect density, and non-uniform coloration become central performance factors [180,225]. Flexible devices add another set of variables, including bending radius, number of bending cycles, $\Delta T$ retention after bending, adhesion loss, electrolyte leakage, and change in sheet resistance [226,227]. Optical modulation measured before bending is therefore insufficient for evaluating flexible electrochromic devices, because mechanical stability, electrochemical reversibility, electrode conductivity, and encapsulation integrity must also be retained after repeated deformation [182,190].

Overall, AI-assisted optimization, flexible processing, interface engineering, and nanostructuring are connected strategies for achieving reproducible device performance under realistic optical, electrochemical, mechanical, and environmental conditions.

6. Conclusions

Chromic TMO thin films are promising materials for adaptive optical technologies, including SWs, optical switching, gas sensing, and multifunctional coatings. The systems discussed in this review, namely VO₂ and related vanadium oxides, WO₃, NiO, and TiO₂, show that chromic behavior is controlled by the interplay between crystal structure, electronic configuration, oxidation state, oxygen stoichiometry, defects, ion transport, and thin film microstructure. Thus, their optical response must be understood not only from chemical composition, but also from phase purity, morphology, interfaces, and preparation conditions.

VO₂ remains the reference thermochromic oxide because its MIT transition enables passive temperature-driven optical modulation, especially in the NIR region. WO₃ and NiO are key electrochromic oxides, with performance governed by ion insertion/extraction, charge compensation, crystallinity, porosity, electrolyte compatibility, and cycling stability. V₂O₅ extends chromic functionality toward multicolor electrochromism and gas sensing, whereas TiO₂ is important as a defect-active oxide and as an auxiliary layer in electrochromic, photocatalytic, antireflective, and optoelectronic architectures.

The preparation route is a decisive factor for chromic performance. Deposition and post-treatment conditions control thickness, stoichiometry, crystallinity, morphology, defects, and interfaces, which directly affect optical contrast, switching kinetics, stability, and reproducibility. Therefore, efficient chromic coatings require the simultaneous optimization of material composition, thin film processing, and device architecture.

Despite significant progress, several challenges remain. VO₂-based systems still require better control of transition temperature, hysteresis, visible transparency, solar modulation, and long-term stability. WO₃-, NiO-, V₂O₅-, and TiO₂-based systems are often limited by ion-transport constraints, charge trapping, slow switching, structural fatigue, and cycling degradation. These limitations are especially important for SW applications, where high luminous transmittance, strong solar modulation, color neutrality, durability, and large-area manufacturability must be achieved simultaneously.

Future research should focus on the following directions:

–

improving VO₂ thermochromic coatings through reducing the transition temperature while preserving high luminous transmittance, strong solar modulation, narrow hysteresis, and environmental stability through doping, strain engineering, multilayer design, protective layers, and nanostructuring;

–

enhancing WO₃/NiO-based electrochromic oxide devices, namely faster switching, lower operating voltage, higher optical contrast, better color neutrality, and longer cycling lifetime through improved ion transport, electrolyte compatibility, charge balance, and interface stability;

–

controlling defects and microstructure such as oxygen vacancies, mixed-valence states, grain boundaries, and porosity to tune optical modulation, ion diffusion, switching kinetics, and degradation;

–

expanding gasochromic WO₃ and V₂O₅ devices for hydrogen sensing and optical gas detection through improving sensitivity, selectivity, reversibility, humidity tolerance, and stable low-temperature operation.

Overall, future TMO films should be compatible with scalable deposition methods, large-area and flexible substrates, and multilayer architectures, while combining chromism with antireflection, photocatalysis, sensing, thermal regulation, or energy-related functions.