Tungsten-Based Catalysts for Environmental Applications

Abstract

This review aims to give a general overview of the recent use of tungsten-based catalysts for wide environmental applications, with first some useful background information about tungsten oxides. Tungsten oxide materials exhibit suitable behaviors for surface reactions and catalysis such as acidic properties (mainly Brønsted sites), redox and adsorption properties (due to the presence of oxygen vacancies) and a photostimulation response under visible light (2.6–2.8 eV bandgap). Depending on the operating condition of the catalytic process, each of these behaviors is tunable by controlling structure and morphology (e.g., nanoplates, nanosheets, nanorods, nanowires, nanomesh, microflowers, hollow nanospheres) and/or interactions with other compounds such as conductors (carbon), semiconductors or other oxides (e.g., TiO₂) and precious metals. WO_x particles can be also dispersed on high specific surface area supports. Based on these behaviors, WO₃-based catalysts were developed for numerous environmental applications. This review is divided into five main parts: structure of tungsten-based catalysts, acidity of supported tungsten oxide catalysts, WO₃ catalysts for DeNO_x applications, total oxidation of volatile organic compounds in gas phase and gas sensors and pollutant remediation in liquid phase (photocatalysis).

1. Introduction

Tungsten is the metal having the highest melting point (3410 °C) and a density (19.3 g cm⁻³) close to that of gold (19.32 g cm⁻³). α-W (lattice type: body-centered cube) is the crystallographic stable form of tungsten (lattice parameter: 0.3165 nm) [1]. β-W is a metastable cubic (type A15) form of tungsten, first found in products of WO₃ reduction by hydrogen [2]. γ-W is an fcc form of tungsten only detected in thin film. Owing to its unique physical properties, tungsten is widely used in refractory alloys. While group 8–11 elements are currently used as nanoparticles of metals in catalysis, W-based heterogeneous catalysts contain tungsten in the form of oxides, sulfides, carbides or heteropolytungstates [3]. Although less employed than molybdenum sulfides, tungsten sulfides are essential components of hydrotreating catalysts [4,5]. More recently, catalysts and photocatalysts based on tungsten sulfides were developed for application in hydrogen production by water splitting [6,7,8]. Utilization of tungsten carbides for electrochemical applications was also developed in the domain of hydrogen evolution reaction [9,10]. Due to their high solubility in water, heteropolytungstates serve as precursors of tungsten in the preparation of W-based catalysts [11,12]. They are also utilized in many organic syntheses, especially in oxidation reactions [13]. Associated with metals such cobalt, heteropolytungstates are also excellent water-splitting catalysts [14].

Tungsten-based catalysts are currently utilized for environmental applications (e.g., DeNO_x, oxidation in gas and in liquid phases, sensors, photocatalysis). In these catalytic systems, tungsten is nearly always in an oxidized form, either as tungsten oxide particles or in strong interaction with specific supports or dopants. The structures of tungsten oxides, supported or not on other oxides, will be reviewed in Section 2. One of the specificities of tungsten oxides is their acid properties that are examined in Section 3. Catalytic applications are then reviewed in the following sections: selective catalytic reduction of NO_x by ammonia over WO₃-V₂O₅ and other WO₃-supported catalysts (Section 4), other DeNO_x applications (Section 5), total oxidation of pollutants in gas phase and gas sensors (Section 6) and total oxidation of pollutants in liquid phase with special insight into photocatalytic processes (Section 7). Selective oxidation in gas or liquid phase (for instance, in the cyclohexanol-to-ε-caprolactone process), as well as the area of heteropolytungstates, will not be reviewed in these sections. Readers interested in these topics are invited to read the recent review by Dai et al. [15]. Due to the high number of works published on the catalytic properties of tungsten-based catalysts, Section 3, Section 4, Section 5, Section 6 and Section 7 will be restricted to the 2014–2020 period, occasionally including some earlier reference papers.

2. Structure of Tungsten-Based Catalysts

Although it is important to stress that the bulk crystal structure may have a limited impact on catalysis since reactions occur at the surface of the crystalline solids, which may not exhibit the same chemical composition as the bulk, the structures of tungsten-based catalysts are summarized in this section to offer an in-depth introduction to the materials presented in this review.

2.1. Structure of Unsupported Tungsten Oxides

No less than seven tungsten oxides were identified in the tungsten–oxygen system [1,16]. However, three oxides are currently more often encountered in catalysis and in most applications [3]: WO₃ (yellow oxide), WO_2.9 or β-tungsten oxide (blue oxide) and WO_2.72 or γ-tungsten oxide (violet oxide).

2.1.1. Tungsten Trioxide

Tungsten trioxide has several allotropic forms, mainly two monoclinic structures (α-WO₃ and γ-WO₃), a triclinic structure (β-WO₃) and a hexagonal form (h-WO₃). All these forms of WO₃ consist of corner-sharing octahedra of WO₆ units. Monoclinic and triclinic structures form a quasicubic arrangement of the octahedra. They differ by the angles of distortion between the adjacent octahedra. Triclinic WO₃ is stable only at low temperature (T < 17 °C) [17]. Ab initio calculations showed that the ideal cubic arrangement of WO₆ blocks is not stable and would tend to tetragonal structure [18]. Sun et al. recently compared the potentialities of monoclinic WO₃ (m-WO₃) and hexagonal WO₃ (h-WO₃) in electrochemical and gas adsorption applications [19]. The structures of the two forms of WO₃ are represented in Figure 1.

While monoclinic WO₃ possesses arrangements of small rhomboid tunnels whose great diagonals are alternatively along x and y axes, hexagonal WO₃ has a network of large hexagonal tunnels alternating with small triangular ones. Accessibility of internal surface is then better for h-WO₃ than for m-WO₃. This was evidenced in the work of Sun et al. by a higher specific capacitance of h-WO₃ with facilitated proton transportation along the tunnels.

Szilágyi et al. investigated the behavior of m- and h-WO₃ for the photocatalytic reaction of methyl orange decomposition [20]. The two materials were prepared by annealing hexagonal ammonium tungsten bronze in air or N₂. Hexagonal WO₃ is formed by annealing the tungsten bronze (NH₄)_xWO_3−y at moderate temperature (470–550 °C), while monoclinic WO₃ is formed at higher temperatures (600–650 °C). Decomposition processes of the hexagonal ammonium tungsten bronze were earlier investigated by the same group, who showed that residual ammonia and/or ammonium ions are vital for stabilizing the hexagonal structure [21]. Blue oxides are formed when (NH₄)_xWO_3−y is decomposed in N₂, which shows the presence of reduced tungsten oxides. On the contrary, yellow oxides are obtained by decomposing the tungsten bronze in air. However, XPS of the oxidized materials shows that small amounts of W⁵⁺ and W⁴⁺ are still present in the h-WO₃ ox, while tungsten in m-WO₃ ox is entirely in the form of W⁶⁺. Monoclinic WO₃ ox is more active than hexagonal WO₃ ox in the photocatalysis reaction because the electron–hole recombination is less probable in WO₃ having 100% W⁶⁺ (no defect sites). In addition to XRD, Raman spectroscopy is a good technique to distinguish hexagonal and monoclinic WO₃ phases (Figure 2). XRD spectra are consistent with m-WO₃ (ICDD file 43-1035) and h-WO₃ (ICDD file 85-2460). Peak width broadening observed on reduced samples is indicative of distortion and creations of defects in the structure. O-W-O stretching bands allow to clearly distinguish m-WO₃ (two bands at 714 and 806 cm⁻¹) and h-WO₃ (three bands at 651, 693 and 786 cm⁻¹). By contrast, O-W-O deformation vibrations, in the 260–330 cm⁻¹ range, are virtually recorded at the same wavenumber for the two forms of WO₃. For that reason, deformation vibration bands cannot be used for the characterization of m- and h-WO₃.

If the presence of ammonium ions allows h-WO₃ to be stabilized, it is also a key factor in controlling the shape of nanorod crystals [22]. NH₄⁺ ions are preferably adsorbed on the (001) facets and block the crystal growing along this axis (Figure 3). Hexagonal nanowires of WO₃ were also prepared as anodic materials of Li-ion batteries [23]. Hydrothermal treatment of lithium tungstate–lithium sulfate mixed solutions in HCl medium leads to typical nanowires, as shown in Figure 4.

Crystal structures, phase transition and physical properties of nanostructured WO₃ oxides have been reviewed by Zheng et al. [24]. Numerous synthesis methods were reported for the preparation of films, ribbons, nanowires, nanotubes, nanorods, etc. In many cases, hexagonal WO₃ is observed in these nanostructures. For instance, leaflike nanoplatelets were prepared by dehydration of orthorhombic tungstite H₂WO₄ to monoclinic WO₃ [25].

2.1.2. Tungsten Trioxide WO_2.9 Blue Oxide (β-Tungsten Oxide W₂₀O₅₈) and Other Suboxides WO_3−δ (δ < 0.13)

Tungsten suboxides phases with formulae W_nO_3n−1 and W_nO_3n−2 were first described by Magnéli in 1950 [26,27]. They were fully described by Tilley in 1995 [28]. The lattice (symmetry P2/m) consists of blocks of WO₆ octahedra joined by sharing corners (as in m-WO₃) with certain octahedra sharing edges along a zig-zag dislocation line, as illustrated in Figure 5.

WO_3−x suboxides are formed when WO₃ is treated in H₂ or alcohol at moderate temperature (>400 °C). First, oxygen vacancies are created without a definite compound being formed [30]. β-Tungsten oxide is usually prepared by heating a mixture of WO₃ and tungsten powder at high temperature (~1000 °C), by reduction of WO₃ in hydrogen around 650 °C [1,27] or by reduction of ammonium paratungstate ((NH₄)₁₀ H₂W₁₂O₄₂, 4H₂O) in hydrogen from 430 to 650 °C [31]. It should be noted that the color of ultrafine “WO₃” powders may change from yellow for the greatest particle sizes (>50 nm) to blue for the finest particles (<10 nm) [32]. In the meanwhile, the structure of ultrafine powders changes from triclinic (>50 nm) to cubic (<10 nm). A cubic phase of blue oxide nanosheets was synthetized by direct decomposition of H₂WO₄·H₂O in H₂/N₂ [33]. The cubic phase is stable up to 280 °C and tends to form a monoclinic phase, complete at 410 °C. Structures similar to those shown in Figure 5 are described as crystallographic shear (CS) by Tilley [28]. By varying the number and orientation of WO₃ blocks sharing edges, it seems possible to reach WO_2.889 (W₁₈O₅₂) composition.

2.1.3. WO_2.72 Violet Oxide (γ-Tungsten Oxide W₁₈O₄₉) and Other Suboxides WO_3−δ (δ > 0.2)

More complex structures are obtained when the O/W ratio decreases to less than 2.88. Solonin et al. described a suboxide WO_2.8 directly derived from hexagonal WO₃ [34]. They named it h-WO_2.8; it is formed at the initial stage of h-WO₃ reduction. Mesostructured WO_2.83 (W₂₄O₆₈) was recently prepared by Cheng et al. by using KIT6 silica as a hard template [35]. XRD confirmed the formation of monoclinic WO_2.83 (PDF # 36-0103). This result contrasts with the formation of WO_2.9 and WO_2.72 by reduction of bulk WO₃ (vide infra). Cheng et al. showed that the reduction of meso-WO₃ to meso-WO_2.83 occurred at low temperature due to a fast hydrogen diffusion through the mesoporosity (Figure 6).

Violet oxide is generally formed by reduction of WO₃ in hydrogen at 900 °C in the presence of water vapor [1]. Sarin showed that WO_2.72 is the most stable oxide formed by reduction of WO₃ [36]. Intermediary blue oxide WO_2.90 is difficult to observe. As shown in Figure 7, violet oxide tends to form whiskers [37] or nanoneedles [38]. This 1D morphology seems to be characteristic of WO_2.72 oxide. While ultrasmall nanoparticles of WO₃ are not easily prepared, Soultanidis et al. succeeded in preparing 1.6 nm particles by thermal decomposition of ammonium metatungstate in oleyamine [39]. XRD revealed that these particles are mainly composed of the W₁₈O₄₉ phase (JCPDS 05-0392).

Interestingly, XPS analysis can be performed to assess the chemical composition of tungsten oxide nanopowders. The x value in WO_3−x can be determined by means of the ratio between W⁵⁺4f and W⁶⁺4f-states. For instance, three series of WO_3−x were synthesized by Shpak et al. at atmospheric pressure by electric explosion of wires (EEW) with different proportions between argon and oxygen [40]. In Figure 8, WO_3−x, WO_2.9 and WO_2.72 nanopowder compositions were determined using the W⁶⁺ response (comps. d-d, W4f_7/2 at 35.7 eV; comps. e-e, W4f_7/2 at 36.1 eV for hydroxide) and the W⁵⁺-response (comps. c-c, W4f_7/2 at 34.8 eV).

2.1.4. Tungsten Dioxide (WO₂)

Tungsten dioxide is a rutile-like oxide, but contrary to TiO₂, WO₂ has a monoclinic structure with distorted octahedron of oxygen ions (C⁵_2h, P2₁/c). This was confirmed by Ben-Dor and Shimony, who succeeded in preparing WO₂ monocrystals [41]. The monoclinic structure was later refined by powder neutron diffraction (PDF# 32–1393) [42]. An orthorhombic form of WO₂ was also reported (PDF# 82–728); it was evidenced in sonochemical preparation [43]. Most preparations of WO₂ were carried out by annealing of WO₃ at high temperatures (up to 1000 °C). A facile preparation at low temperature (500 °C) was reported by Coşkun and Koziol by hydrolysis of WCl₆ in the presence of NaBH₄ [44]. Compared to other tungsten oxides, WO₂ presents relatively high conductivity properties. The energy bands and density of states from theoretical calculations confirm a metal-like behavior of WO₂ [45].

2.2. Structure of Supported Tungsten Oxides

The structure of tungsten species dispersed on oxide supports depends on both the nature of support and the concentration of tungsten. In 2007, Knowles et al. reported a detailed overview of the surface chemistry of supported oxide, especially tungsten oxide [46]. WO₃ is more easily dispersed on alumina than on silica, on which WO₃ tends to form agglomerated crystallites [47]. The surface structure of tungsten oxide on various oxide supports (Al₂O₃, TiO₂, Nb₂O₅, ZrO₂, SiO₂ and MgO) was investigated by Raman spectroscopy by Kim et al. [48]. WO₃ on Al₂O₃, TiO₂, Nb₂O₅ and ZrO₂ exhibits strong bands at 1005–1020 cm⁻¹ characteristic of terminal W=O mono-oxo species. Bands in the 800–960 cm⁻¹ spectral region are also observed. They are attributed to W-O-W species whose intensity increases with the tungsten surface coverage. Bands at 1005–1020 cm⁻¹ are not observed in Raman spectra of WO₃/SiO₂ and WO₃/MgO. Two intense bands at 802 and 715 cm⁻¹ are visible on the spectra of 5%WO₃/SiO₂. These bands are characteristic of the formation of crystalline WO₃, due to the poor dispersion of WO₃ on this support. Raman spectra of WO₃/MgO are more complex and suggest the formation of crystalline MgWO₄ (as well as CaWO₄ due to Ca impurity in MgO). In situ Raman spectroscopy coupled to ¹⁸O/¹⁶O exchange allowed Lee and Wachs to precisely identify the surface structure of different oxides (including WO₃) dispersed on silica [49]. In this study, WO₃/SiO₂ showed Raman bands at 968-985-1014 cm⁻¹ shifted to 920-935-963 cm⁻¹ in totally ¹⁸O-exchanged solids. Monoxo and dioxo tungstate species were formed, while crystallized WO₃ was virtually absent (contrary to the previous study). Decisive information on the structure of supported tungsten catalysts can be obtained by UV–Vis. Ross-Meedgarden and Wachs showed that the electronic edge energy (Eg) can be correlated to the nature of W surface species, especially the number of nearest cations surrounding the central M cation [50]. The number of covalent W-O-W bonds around the central W(VI) cation would be given by Equation (1):

N_W-O-W = 11.89 − 2.37Eg

(1)

with Eg varying from 5.5 eV for isolated WO₄ species (as bi-grafted, di-oxo WO₄ surface sites [51]) to 2.7 eV for crystallized WO₃ 3D structures.

Though WO₃ tends to form nanocrystals on silica, there remains an amorphous fraction more or less dispersed. Chauvin et al. developed a methodology coupling XRD and Raman studies for the quantification of W species: surface dispersed WO_x (monomeric and oligomeric), amorphous WO₃ and crystallized WO₃ [52]. On silica, only dispersed W species (monomeric and polymeric) would be formed below 1 W nm⁻². For higher concentrations, WO₃ nanoparticles (amorphous and crystallized) are evidenced. This contrasts with other supports (Al₂O₃ [53,54], TiO₂ [55,56], ZrO₂ [57,58,59]) for which tungsten impregnation up to 4–5 W nm⁻² leads to well-dispersed catalysts. On these supports, WO₃ nanocrystals are generally observed in the 4.5–9 W nm⁻² loading range, while large bulklike WO₃ crystals can be observed beyond 9 W nm⁻² [60]. In the last decades, the WO₃-ZrO₂ catalytic system has been the subject of a huge number of investigations owing to its excellent performances in reactions requiring strong acid sites [61,62]. Recent investigations by HAADF [63] and HRTEM [64] allowed obtaining detailed pictures of the various tungsten species populating the zirconia surface (Figure 9a,b).

It should be noted that WO₃ clusters (crystallized or not) may contain Zr atoms from the support (Figure 9a). Tungstate species depicted in Figure 9 are in agreement with the model of growth mode of W species when the surface density exceeds 5 W nm⁻² [65]. Preparation of tunsgtated zirconia catalysts from Lindqvist-type complexes containing (W₅O₁₈Zr)²⁻ moieties confirmed the structure of such surface species containing both W and Zr atoms, as well as their role in acid catalysis [66]. Structures of oxo-W(VI) species deposited on TiO₂ solids were extensively studied by Tribalis et al. by Raman spectroscopy [56]. Molecular configuration of W surface species depends on the temperature of calcination in O₂. Raman features shifted from 930 cm⁻¹ (Ti₂OH···OWO₃, main species at 25 °C) to 950 cm⁻¹ (Ti-O-WO₃ at 100 °C), then 970 cm⁻¹ ((Ti-O)₂-W(=O)₂ at 120–250 °C) and finally 1030 cm⁻¹ for the trisubstituted mono-oxo (Ti-O)₃-W=O at 430 °C.

Though W dispersion is not easily obtained on amorphous silica, recent developments of mesostructured silicas (MCM 41 [67,68], KIT-5 and KIT-6 [69,70], SBA-15 and SBA-16 [71,72,73]) has allowed inserting significant amounts of tungsten while limiting the formation of WO₃ nanocrystals. This technique has been extended to spongelike silicate TUD-1 discovered at Delft University [74]. Isolated WO₄^2- species could be anchored on Ti-modified TUD-1 up to 30% WO₃ [75]. Other techniques including aerosol-assisted sol–gel processes [76] or grafting of tungsten-containing molecules on silica [77,78] were also developed to obtain highly dispersed “single-site” W species.

As a final statement, critical insights of tungsten-based materials reveal no less than seven tungsten–oxygen systems, with three arrangements mainly encountered in the catalysis field: WO₃, WO_2.9 (β-tungsten oxide) and WO_2.72 (γ-tungsten oxide). WO₃ presents several allotropic forms: monoclinic, triclinic and hexagonal structure. The structure impacts the photocatalysis activity since the electron–hole recombination is favored by the absence of defects. Tungsten suboxide phase (WO_2.9) is formed under H₂ treatment and exhibits oxygen vacancies. More complex structures are obtained when the O/W ratio decreases to less than 2.88, and WO_2.72 is the most stable oxide formed by reduction of WO₃.

3. Acidity of Supported Tungsten Oxide Catalysts

Tungsten oxide WO₃ is among the most acidic transition metal oxides [79]. Very few studies have been devoted to the acid properties of unsupported tungsten oxides. Kanan et al. studied the change of acid site concentration upon dehydroxylation/dehydration of the surface of monoclinic WO₃ [80]. They concluded that the changes in Brønsted/Lewis (B/L) acid site concentration were not related to the degree of dehydroxylation but rather to a mild reduction of the tungsten oxide surface. Choi et al. investigated surface acidity of mesoporous WO₃ synthesized using KIT-6 as a hard template [81]. Acid site characterization (NH₃-TPD, FTIR of adsorbed pyridine) showed that mesoporous WO₃ exposes predominantly Lewis acid sites associated with W⁶⁺ and probably cus-W⁶⁺ (cus = coordinatively unsaturated sites). A similar study performed by Kasian et al. showed that acid site concentration of mesoporous tungsten oxide was increased by a mild reductive treatment in hydrogen at 250 °C [82]. Yue et al. studied the reactivity of m-WO₃, h-WO₃, hexagonal tungsten bronze (HATB) and (NH₄)_0.33−xWO_3−z for the hydroconversion of n-heptane reaction (Pd was added to the W oxides as hydrogenation function) [83]. The tungsten oxides were reduced in H₂ at 440 °C, 35 bar. The n-C7 reactivity revealed the formation of strong Brønsted acid sites, certainly linked to the reduction of W⁶⁺ to lower valence states, the protons allowing the neutralization of negative charges localized on W-O-W formed with W^(6-x)+ ions. Li et al. investigated the behavior of MoO₃ or WO₃ clusters in the conversion of labeled ethanol (CH₂CH₃OD) [84]. Dehydration and dehydrogenation can occur. The overall alcohol conversion would be correlated to Lewis acidity, while the selectivity (dehydration vs. dehydrogenation) would reflect the redox properties of the oxide (i.e., the propensity of Mo or W to reduce from VI to V state in the reaction medium). The reaction was carried out with (MO₃)₃ trimers supported on graphene. Other supports and other reactions were tested with contrasted differences between Mo and W oxides depending on the substrate (see the review [85]).

3.1. General Overview

Barton et al. reported that supported tungsten oxide catalysts are able to generate Brønsted acid sites when W⁶⁺ species are replaced by tungsten cations with lower valency [86]. A general overview of the acidity of tungstated oxides was performed by Zaki et al., who compared the relative strength of acid sites of 10% WO₃ samples deposited on Al₂O₃ (69 m² g⁻¹), TiO₂ (34 m² g⁻¹) and SiO₂ (136 m² g⁻¹) [87]. All samples were calcined at 500 °C. Well-dispersed monotungstates and polytungstates were the most abundant surface species on WO₃/Al₂O₃, while 3D polytungstates were observed on WO₃/TiO₂. On WO₃/SiO₂, tungstosilicates and crystallized WO₃ would be the dominant surface structures. Acid sites were monitored by pyridine adsorption, which can lead to four species: LPy (Lewis acid bound Py), BPy (Brønsted acid bound Py), HPy (hydrogen-bonded Py) and PPy (physically adsorbed Py). FTIR spectra of adsorbed pyridine were recorded at room temperature (Figure 10) and after desorption at 100, 200 and 300 °C.

Hydrogen-bonded pyridine is the main species at RT on WO₃/SiO₂, but it is unstable and tends to disappear upon heating. Lewis acid sites (LPy) and hydrogen-bonded pyridine (HPy) are clearly observed on WO₃/Al₂O₃, while both BPy and LPy sites are identified on WO₃/TiO₂. Brønsted sites would be very strong on WO₃/TiO₂ (still intense upon heating at 300 °C). The work of Zaki et al. [87] clearly underlines the critical role of the support of tungsten oxides, both in the nature of W species and in their respective acidity.

3.2. Silica-Supported Tungsten Oxides

Due to the quasiabsence of acid sites on silica [88], WO₃/SiO₂ materials were often used to investigate the acidity of tungsten oxides. In their study on WO₃/SiO₂ catalysts (0 to 11.7 wt.% W on a silica of 200 m² g⁻¹, i.e., 0 to 2.4 W nm⁻²), Chauvin et al. monitored the acidity of the samples by FTIR spectroscopy of adsorbed 2,6-dimethylpyridine (lutidine) [52]. The spectra reported in Figure 11a show the presence of Brønsted sites (bands at 1643 and 1630 cm⁻¹). Lewis sites that should give bands at 1620–1600 cm⁻¹ are clearly absent from the samples. Amounts of Brønsted sites linearly increase with the WO₃ surface density up to 1.5 W nm⁻² (Figure 11b).

The acidity of WO₃/SBA-15 oxides was characterized by Hu et al. using pyridine adsorption, solid-state NMR and quantum chemistry calculation [89]. The study was completed by measurements and calculations of ¹⁵N NMR chemical shift tensors of pyridine interacting with tungsten oxides and silica. It was shown that W-OH-W dimers are the principal Brønsted acid sites. W-OH monomers and silanols are very stable with minimal Brønsted acidity. Contrasting with the work of Hue, Gonzalez et al. found quasiexclusively Lewis acid sites in a series of WO₃/SBA-15 materials [90]. Acidity was characterized by pyridine adsorption monitored by FTIR. Bands at 1595 and 1445 cm⁻¹ attributed to pyridine coordinated to Lewis acid sites are much more intense than the 1545 cm⁻¹ band characteristic of protonated pyridine bound to Brønsted acid sites. However, even though it is relatively weak, the Brønsted site density seems very stable when the temperature of pyridine adsorption is increased from 50 to 100 °C, while the number of Lewis sites is strongly affected (

Table 1. Acid site density on SBA-15 and WO₃/SBA-15 measured by pyridine adsorption at 50 and 100 °C. The number of sites is calculated from the integrated area of the band at 1545 cm⁻¹ for Brønsted sites and at 1445 cm⁻¹ for Lewis sites. From González et al. [90].

Sample	T (°C)	Brønsted Acid Sites	Lewis Acid Sites
		(µmol g−1)	(µmol g−1)
SBA-15	50 °C	0	152
	100 °C	0	67
5%WO3/SBA-15	50 °C	24	920
	100 °C	24	92
15%WO3/SBA-15	50 °C	34	1102
	100 °C	31	350
25%WO3/SBA-15	50 °C	15	1246
	100 °C	12	458

).

The reasons for the discrepancy between the work of Hu et al. [89] and that of González et al. [90] are not clear. They may originate from the nature of the tungsten precursor (WCl₆ in toluene for Hu et al. and ammonium metatungstate for González et al.), the final temperature of calcination (400 °C for Hu et al. and of 600 °C for González et al.) or the method of acid site detection (NMR for Hu et al. and FTIR for González et al.). This question should certainly be reconsidered in the future.

WO₃/SiO₂ catalysts were prepared by Kulal et al. for liquid phase nitration of aromatics [91]. A sol–gel technique using ammonium metatungstate and ethyl silicate 40 was employed for the preparation. FTIR of adsorbed pyridine revealed the presence of both Lewis and Brønsted sites. Kulal et al. attributed the formation of Brønsted sites to the presence of polytungstate species, while Lewis sites would require rather isolated W ion centers. The B/L ratio increased with the W loading in the catalysts. At low loading (<5% W), there were virtually no Brønsted sites. Bhaumik et al. developed silica-supported WO₃ and Ga₂O₃ catalysts for lignocellulosic biomass to furfural processes [92]. Although the acidity was not characterized in detail, the authors of this work stated that Lewis sites are the most abundant acid sites on the catalysts.

Tungsten species are generally added by impregnation on the silica support. They may also be incorporated during the preparation of the mesoporous silica. Zhu et al. prepared WZr-KIT6 with various Zr/W ratios by addition of pluronic triblock copolymer to a mixture of tetraethoxysilane (TEOS), ammonium tungstate and ZrOCl₂ [93]. Acidity of the materials was measured by adsorption of pyridine monitored by NMR. Both Brønsted and Lewis acid sites coexisted on the catalysts. Lewis acid sites would be mainly linked to the presence of zirconium while tungsten is necessary to generate strong Brønsted acid sites. The materials were tested in ethanol dehydration. Interestingly, it was shown that calcination of the coke catalysts regenerated Brønsted sites, which became even stronger than those on the fresh materials.

Attempts to modify the activity of WO₃/SiO₂ catalysts by means of gold nanoclusters revealed that the acidity of tungsten oxide is not modified by Au (methanol transformation tests) [94]. WO₃/SiO₂ is highly selective for dimethyl ether, while the presence of gold is necessary for the formation of methyl formate and dimethoxymethane. This result suggests that pure WO₃ possesses only acid sites and virtually no redox sites. The addition of WO₃ was recently used to give acidity to Pt/SiO₂ for glycerol hydrogenolysis to 1,3-propanediol [95]. Tungsten oxide would be present as W₂₅O₇₃ with a high proportion of W⁵⁺. It also maintains a high dispersion of platinum.

3.3. Alumina-Supported Tungsten Oxides

Acidity of WO₃/Al₂O₃ was studied in the 1980s and 1990s by Soled et al. [96] and Zhang et al. [97]. While alumina possesses virtually no (or very weak) Brønsted sites, the addition of tungsten oxide generates both Brønsted and Lewis sites. Most probably, WO₃ titrates the strongest Lewis sites on alumina to form Brønsted sites. The number and strength of these Brønsted sites increase with the temperature of calcination [96]. The proportion of Brønsted sites also increases with the tungsten loading (

Table 2. Effect of W loading on the number of Lewis and Brønsted sites of WO₃/Al₂O₃ (152 m² g⁻¹). Acid site distribution was determined by a modified Benesi colorimetric technique. From Zhang et al. [97].

WO3 Loading	Lewis Sites	Brønsted Sites	Lewis Sites	Brønsted Sites
(wt.%)	(µmol g−1)	(µmol g−1)	(%)	(%)
0	190	14	93	7
2.5	207	69	75	25
10	220	125	64	36
30	242	182	57	43

).

Combining lutidine adsorption and other spectroscopic investigation, Chen et al. obtained precise information about the change in acid site concentration with respect to the W surface density [54]. Their results are summarized in Figure 12, which shows that monomeric tungsten species contain quasiexclusively Lewis acid sites, while Brønsted sites are relatively abundant on polymeric species appearing for W density above 1.4 W nm⁻².

The acid properties of WO₃-Al₂O₃ catalysts were recently revisited by Kitano et al., who showed that the formation of Brønsted sites depended on the temperature of calcination [98]. They proposed that Brønsted sites are predominantly located at the boundaries between WO₃ monolayer domains (Figure 13).

Potassium-doped WO₃/Al₂O₃ catalysts are able to transform gas mixtures such as CO₂/H₂S/H₂, methanol/H₂S or CH₄/H₂S/CO₂ into valuable products such as methyl mercaptan (CH₃SH) [99,100,101]. All these reactions were reviewed by Taifan and Baltrusaitis [102]. Structural and acid–base properties of K₂O-WO₃-Al₂O₃ catalysts were recently investigated by Wang et al. [103], Zhu et al. [104] and Kiani et al. [105]. Raman spectroscopy was used to differentiate and detect crystalline WO₃ (bands at 273, 719 and 809 cm⁻¹), crystalline K₂WO₄ (bands at 324 and 927 cm⁻¹) and dispersed surface oligomeric WO_x species on alumina (strong band at 1021 cm⁻¹). Conformation and position of most of these Raman bands were confirmed by DFT calculation [105]. Surface acidity was probed by NH₃-TPD [103] or NH₃ adsorption monitored by FTIR [104], while basicity was probed by CO₂-TPD. K₂O suppressed both Brønsted and Lewis acidity from the surface WO_x species or exposed Al₂O₃ sites [104]. Adsorption of CO₂, nil on unpromoted WO₃/Al₂O₃ (no basic sites), significantly increases on K₂O-promoted catalysts. Residual acidity could play a major role in catalytic activity, while basicity would be essential for a good selectivity for methyl mercaptan [103].

3.4. Zirconia-Supported Tungsten Oxides

The system WO₃-ZrO₂ was earlier reported to possess strong acid sites with H₀ < -14.7 [106]. This “super” acidity makes the catalyst able to replace concentrated sulfuric acid for alkylation and isomerization reactions of alkanes [62,106] or aromatics [61,107]. Acidic sites certainly play a great role in these reactions, even though redox properties of WO₃-ZrO₂ can be put forward for explaining the enhanced activity of reduced materials [108]. The effect of the preparation method on the acidity of WO₃-ZrO₂ was studied by Santiesteban et al. [109]. Acid sites were titrated by adsorption of 2,6-dimethyl pyridine. Brønsted sites (0.054 meq g⁻¹) were more abundant than Lewis sites (0.013 meq g⁻¹), but only 20% of Brønsted sites were strong acid sites (0.011 meq g⁻¹) in proportion close to Lewis sites. The most efficient preparation consists of reflux/impregnation of ZrO(OH)₂ and H₂O with ammonium metatungstate. The acid site structure (Scheme 1) is similar to the model proposed by Afanasiev et al. in 1994 [110].

Brønsted/Lewis (B/L) site ratio in WO₃-ZrO₂ catalysts was measured by Baertsch et al. for various tungsten loadings and in various conditions of pretreatment [111]. Acid sites were evaluated by adsorption/desorption of NH₃ or pyridine. Complementary information about the redox sites was obtained by adsorption of O₂ and H₂, while the number of hydroxyl groups was measured by D₂/OH isotopic exchange. Brønsted sites were virtually absent on low-loaded catalysts where tungsten was in the form of monomeric species. The B/L ratio gradually increased with the tungsten loading. The presence of H₂ during NH₃ adsorption led to the generation of strong Brønsted sites. Platinum nanoparticles used in bifunctional Pt/WO₃-ZrO₂ catalysts would also be a source of Brønsted site by H spillover between Pt and WO₃ [112]. The surface chemistry and acidity of WO₃-ZrO₂ catalysts for skeletal isomerization of alkanes were discussed by Di Gregorio and Keller, who suggested that a condensation phenomenon between Lewis and Brønsted sites can occur during the calcination treatment [113] (Figure 14).

A combined theoretical–experimental study on the acidity of tungstated zirconia was performed by Galano et al. [114]. They confirmed that the Lewis acidity (band at 1444 cm⁻¹ for pyridine coordinated to L sites) and Brønsted acidity (band at 1539 cm⁻¹ for protonated pyridine) is a function of the degree of WO₃ polymerization. The number of Brønsted sites increased while that of Lewis sites decreased as the W centers shifted from tetrahedral to octahedral coordination, i.e., when the degree of WO_x condensation increased. In parallel, the strength of the Brønsted acid sites tended to decrease. A maximum of Brønsted acidity was observed for a W loading of 7 W nm⁻². Recent investigations of tungstated zirconia designed for acid-driven catalytic reactions showed that active sites are three-dimensional distorted Zr-WO_x clusters of 0.8−1 nm size [59]. Optimum structure for a better activity is obtained when W is in a distorted octahedral environment in close contact with Zr cations. Calcination temperature (up to 800 °C) and nature of zirconium oxyhydroxide used as support play a major role in obtaining higher Brønsted acidity [115]. Though acid properties are generally characterized by adsorption of basic molecules (e.g., NH₃, pyridine) monitored by FTIR, other techniques are available. Li et al. showed that ³¹P-NMR using TMPO (trimethyl phosphine oxide) as a probe molecule reveals the presence of Brønsted sites on WO₃-ZrO₂ prepared by a sol–gel method with Zr(OBu)₄ and WCl₆ as reagents [107]. TMPO gives a broad band from 40 to 90 ppm, while chemical shift above 57 ppm is characteristic of Brønsted sites.

Yttrium doping of zirconia is often used to stabilize the tetragonal structure of ZrO₂. The effect of yttrium on the generation of acid sites on WO₃-ZrO₂ was investigated by Yamamoto et al. [116]. Acidity of the catalysts (15% WO₃) was characterized by means of model reactions: n-butane skeletal isomerization, alkylation of anisole with benzyl alcohol, and 2-butanol decomposition. Acidity increased with yttrium content up to 4% Y. In the meanwhile, the tetragonal-to-monoclinic ratio of zirconia increased to reach 85% of tetragonal phase for 4% Y. This proves that t-ZrO₂ is more suitable than m-ZrO₂ for stabilizing WO_x species with the highest acidity. Above 4% Y, the catalyst acidity decreased due to the excessive formation of inactive Y-W-O species. The beneficial role of tetragonal zirconia was demonstrated for different applications of WO₃-ZrO₂ catalysts, such as hydrolysis of cellobiose [117] or viscoreduction of heavy crude oil [118]. The mode of preparation of the zirconia support can greatly affect the acidity of the materials. Sol–gel synthesis and extraction of the solvent in supercritical conditions give the most active catalysts for organic reactions of industrial interest: acylation of veratrole with acetic anhydride and acylation of anisole with benzoic anhydride [119]. Doping zirconia with phosphorus was extensively studied by Miao et al. [120,121]. WO₃ was deposited on mesoporous zirconium oxophosphate [120] or directly incorporated in the one-pot preparation of meso-ZrPOx to synthesize mesoporous WZrP materials [121]. The catalysts were tested in benzylation of anisole. Unfortunately, the effect of phosphorus could not be clearly established due to the absence of reference WO₃/ZrO₃ catalyst in these studies. However, the mode of preparation and the mesoporous texture give the WZrP materials a high activity and a high stability. The nature of acid sites is similar to the acidity spectrum characterized in WO₃/ZrO₂: W-free materials possess Lewis sites (83 µmol g⁻¹) and very few Brønsted sites (19 µmol g⁻¹). Brønsted site density increases up to 55 µmol g⁻¹ for W/Zr ratio of 0.2.

3.5. Titania-Supported Tungsten Oxides

WO₃-TiO₂ is also an acid catalyst with activity for n-alkane isomerization [55]. These mixed oxides possess both Lewis sites ascribed to surface Ti⁴⁺ species and strong Brønsted sites generated by WO₃ nanoclusters. WO₃-TiO₂ was proven to have excellent performances in the reaction of sorbitol transformation into biofuels [122]. The reaction proceeds via successive steps of dehydration and hydrogenation, which requires metal–acid bifunctional catalysts. WO₃-TiO₂ (acid function) associated with Pt/ZrO₂ (metallic function) gave the best performance for the production of C5-C6 hydrocarbons. The main application of WO₃-TiO₂ (associated with V₂O₅) is for the NO_x abatement by the NH₃-SCR process [123,124,125]. While the chemical state of vanadium and tungsten seems to be crucial for a good DeNO_x activity [126], acid properties could also play a primary role in the reaction [123,127]. The characterization of V₂O₅-WO₃/TiO₂ is examined in Section 4.

3.6. Conclusions

Evaluation of the acidity of supported WO_x obviously depends on the used probe molecule because of its own basic strength. For instance, NH₃, lutidine and pyridine, the most popular probe molecules for acidity evaluation, display pKa values of 9.23, 6.65 and 5.23, respectively. Consequently, they do not all exhibit the same sensibility toward the Brønsted (or acidic hydroxyl group) and Lewis acid sites (coordinatively unsaturated sites).

In addition, the acidity of WO_x species clearly depends on the W surface coverage. Indeed, dispersed phase surface sites (WO₄, WO₅), poorly crystalline WO₃ nanoparticles and bulklike crystalline WO₃ all show unique acidity. For instance, crystalline WO₃ exhibits Brønsted acidity, not encountered on dispersed WO₄ sites. Consequently, the acidity of catalysts strongly depends on the obtained WO_x surface species, which depend on the preparation protocol and the tungsten loading. Brønsted acid sites are reported to be mostly dependent on the presence of polytungstate polymeric species. Lewis acid sites are associated with coordinatively unsaturated W⁶⁺ sites or isolated W ions centers. The B/L ratio depends on the support. Lewis sites are the most abundant acid sites on WO₃/SiO₂ catalysts. Addition of tungsten oxide on alumina generates both Brønsted and Lewis sites, and the proportion of Brønsted sites increases with the tungsten loading. The stronger acid sites are encountered on the WO₃-ZrO₂ system, where a condensation phenomenon between Lewis and Brønsted sites can occur during the calcination treatment. New surface tungstate species appear with various degrees of WO₃ polymerization, leading to octahedral coordination that promotes an increase in the amount of Brønsted acid sites.

4. WO₃ Catalysts for DeNO_x Applications: WO₃-V₂O₅/TiO₂ and Other Tungsten-Based Materials

Nitric oxide abatement in nitric acid plants was initially carried out by NH₃-SCR over V₂O₅-TiO₂ catalysts. Very soon, it was proven that adding WO₃ (or MoO₃) to the catalyst formulation led to a dramatic improvement of the performance. These V-W/TiO₂ catalysts were also implemented in DeNO_x autocatalyst processes, especially for heavy-duty engines, working with NO₂/NO_x inlet ratio of 0 or 0.5, related to the standard and fast SCR stoichiometry, respectively. The role of WO₃ was summarized by Chen and Yang in 1992 [123]: (i) it increases the activity and widens the temperature window; (ii) it increases the resistance to various poisons (alkalis, arsenic); (iii) it reduces ammonia oxidation by O_2, which is a nondesired reaction (especially possible at elevated temperature, T > 400 °C). Kompio et al. also showed by different techniques (e.g., Raman spectroscopy, EPR, H₂-TPR) that tungsten allowed an increase in V-O-V species by confining vanadia in small clusters [128]. Most of these effects were confirmed and detailed in a recent review by Lai and Wachs [129], who stressed three important features of the reaction: (i) the role of V⁵⁺ surface species as active sites: WO₃ is not active per se but promotes the reaction by vanadia by formation of oligomeric vanadia (V₂O₅) sites or crowding; (ii) the specific role of Brønsted acid sites in NH₃ activation: most of the Lewis sites are converted to Brønsted sites in the presence of moisture at 250 °C; (iii) the detailed mechanisms reported by Topsoe and Dumesic et al. in the 1990s for V₂O₅-TiO₂ [130,131] (see Figure 15) seemed to be still valid in the 2010s for V₂O₅-WO₃/TiO_2, as also recently summarized by Han et al. [132].

This mechanism, however, contrasts with the amide–nitrosamide mechanism proposed by Lietti et al. [133] in which NH₃ is predominantly activated on Lewis sites as amide species (Equation (2)). The main steps of this mechanism lead to the formation of NH₂NO (nitrosamide) (Equation (3)), which further decomposes into N₂ + H₂O.

H₃N:→V⁵⁺ − O²⁻ ⇒ V⁴⁺ − NH₂ + OH⁻

(2)

V⁴⁺ − NH₂ + NO ⇒ V⁴⁺ − NH₂NO

(3)

Another important feature to consider is the formation of N₂O. Nitrous oxide is a potent greenhouse gas with a global warming potential of 295–300 times that of CO₂ [134,135]. Its concentration in exhaust gases should be controlled. Djerad et al. reported that N₂O is formed at high temperature (T > 300 °C) in the NH₃-SCR process over V₂O₅-WO₃/TiO₂ [136]. Nitrous oxide concentration increases with the vanadium loading in the catalyst and is formed via NH₃ oxidation (Equation (4)).

2NH₃ + 2O₂ → N₂O + 3H₂O

(4)

Liu et al. showed that the use of titania supports with high surface area would be a key to reduce N₂O formation [137]. Vanadium being more dispersed on these supports, ammonia oxidation to N₂O could be favored on polymerized vanadyl species.

4.1. Mechanisms and Surface Intermediates

The nature of active sites and the surface intermediates were investigated by Zhu et al. by in situ IR spectroscopy [138]. The first step of the mechanism would be the adsorption of ammonia on acid sites either as NH_3ads (on Lewis (L) sites) or NH₄⁺ (on Brønsted (B) sites). The surface VO_x sites (V⁵⁺_L and V⁵⁺_B) are the active sites for the reaction, tungsta sites (W⁶⁺_L and W⁶⁺_B) being significantly less active. Both B and L sites participate in the reaction. However, if Brønsted sites are more abundant, Lewis sites possess the highest turnover frequency. Overall, ammonia adsorption is part of a complex NO_x removal mechanism and results in hydrogen abstraction to provide –NH₂ amide-type species. Hydrogen migration is involved in both acidity conversion for Lewis to Brønsted acid site transformation at T > 300 °C and reoxidation for the redox loop of the DeNO_x reaction [139]. In addition, coordination of ammonia on Lewis acid sites does not prevent the dissociative H₂O chemisorption and the formation of Brønsted acid sites, leading to adsorbed NH₄⁺ species [140].

Surface acidity and redox properties, as well as their effect on SCR activity, were investigated by Zhao et al. on a series of V_nW-TiO₂ with various vanadium loadings (1–13 wt.%), while keeping the tungsten loading constant (8 wt.% WO₃) [141]. The main characteristics of the series are reported in

Table 3. Vanadium state and acid and redox properties of a series of VnW-TiO₂ catalysts (n: vanadium loading (%), 1 ≤ n ≤ 13; [WO₃] = 8%). TiO₂ (120 m² g⁻¹) mainly as anatase. Acid site concentrations were measured by pyridine adsorption monitored by FTIR (bands at 1580 and 1633 cm⁻¹ for Brønsted sites; bands at 1597 and 1438 cm⁻¹ for Lewis sites). H₂-TPR peaks are between 535 and 569 °C, while O₂-TPO gives peaks around 450–460 °C. The mean rate of reduction (νTPR) and oxidation (νTPO) are given in the table. SCR activity was measured between 150 and 420 °C in a flow of 1000 ppm NO + 1000 ppm NH₃ in 5% O₂ (WHSV: 20,000 h⁻¹). NO conversion at 200 °C is representative of the activity order. From Zhao et al. [141].

Catalyst	V4+/V5+	B Sites	L Sites	νTPR	νTPO	SCR Activity: NO Conv. at 200 °C
	(XPS)	(µmol m−2)	(µmol m−2)	(nmol m−2 s−1)	(nmol m−2 s−1)	(%)
V1W/TiO2	0.45	28	131	2.4	0.95	19.5
V5W/TiO2	0.99	57	153	4.6	1.4	38.0
V7W/TiO2	1.47	149	292	7.5	2.1	41.5
V9W/TiO2	1.73	164	318	8.6	2.9	60.5
V11W/TiO2	0.88	140	325	8.8	7.6	51.5
V13W/TiO2	0.50	98	485	1.1	1.0	44.0

. The highest activity observed for the V9W-TiO₂ sample is correlated with both higher V⁴⁺/V⁵⁺ ratio and higher Brønsted concentration sites. Redox properties do not seem to play a major role. Additional conclusions resulting from experimental analyses and DFT calculations show that the structure of vanadyl species plays a crucial role in DeNO_x performances; polymeric species demonstrate higher NH₃-SCR activity than monomeric vanadyl compounds [142]. Besides, based on XPS results, low-valence vanadium species (V⁴⁺ and V³⁺) were recently claimed to enhance SCR activity compared to the V⁵⁺ form [143]. Both low-valence vanadium species and polymeric surface VO_x species drive the NH₃-SCR activity at low temperature (T < 350 °C), as previously mentioned in [142].

The crystallinity of WO₃ has a significant effect on vanadia dispersion [127]. Two-dimension VO_x moieties have a tendency to anchor onto the titania surface in the vicinity of WO_x. Small WO₃ crystallites are preferable to WO_x amorphous layer occupying a large surface of titania. Nevertheless, a compromise should be found because bulk tungsten oxide is less acidic than the dispersed oxide.

Interactions between tungsten oxide and vanadia were reinvestigated by Kompio et al. to better understand the promoter effect of tungsten on SCR activity during heat treatments and thermal stress [144]. The study was carried out on a V1.5-W10/TiO₂ catalyst (anatase, 140 m² g⁻¹ after low-temperature calcination at 350 °C). Heat treatments generally provoke catalyst deactivation. However, several maxima of activity were observed during the global decrease in activity when the support BET area reached 42 m² g⁻¹ and 20 m² g⁻¹. New surface sites were then created because isolated vanadia species, less active, were replaced by more active bridged V-O-V sites. Similar active species were proposed by Kwon et al. for non-tungstated V/TiO₂ catalysts calcined at 600 °C [145]. Marberger et al. underlined the fact that the presence of water is required to judge the thermal stability of the catalysts [146]. Water accelerates sulfate group removal (generally present in commercial titania), which changes the acidity of the catalysts. Water also accelerates the loss of BET surface area and V volatility. However, curiously, it tends to increase the vanadium dispersion (

Table 4. Comparison of dry (10% O₂ in N₂) and wet (10% O₂ + 10% H₂O in N₂) aging of a commercial V₂O₅ (2%)-WO₃(10%)/TiO₂ catalyst. Vanadium emissions and VO_x surface coverage (based on 7.9 VO_x nm⁻² for one monolayer). From Marberger et al. [146].

	Gas Feed	550 °C	600 °C	650 °C
V release (µg m−3)	dry	1.0	1.4	34.2
	wet	2.1	49.3	201
VOx coverage (%)	dry	24	32	42
	wet	27	37	44

). Authors recently confirmed the mobility of the VO_x and WO_x species of the catalyst exposed to hydrothermal aging together with various structural changes of surface acidity [146]. Additionally, Liang et al. [147] observed that gas flow containing SO₂ and H₂O decreased the number of NH₃ adsorption sites, active component content, specific surface area and pore volume over F-containing V₂O₅-WO₃/TiO₂ catalysts.

Another possible cause of deactivation is the transformation of anatase to rutile, a poorer support of vanadia and tungsta for the SCR reaction. Promotion of titania by silica can prevent rutile formation above 600 °C and give more stable catalysts [148,149]. A commercial catalyst named VSCR1 has been compared to a catalyst doped with 4–5% SiO₂, named VSCR2. Figure 16 shows the global behavior of the two catalysts when they are calcined at 600, 700 and 800 °C.

A dramatic improvement of the catalyst stability is obtained by Si doping. Maintaining the anatase phase is correlated with a decrease in the WO₃ and V₂O₅ phase segregation into monoclinic tungsten oxide and vanadia crystallites. It seems that SiO₂ tends to segregate as bulky silica crystallite at high temperature (>600 °C) [149]. Only small polymeric entities formed by a minute amount of silica can prevent catalyst deactivation. Deactivation is often observed after hydrothermal treatments at high temperature. However, this is not the case in all circumstances. Chen et al. prepared V-W/TiO₂ catalysts in two steps: (i) WO₃/TiO₂ was first prepared by mixing metatitanic acid and ammonium paratungstate; (ii) the solid, dried and calcined at 550 °C, was then impregnated with 1% V₂O₅ using metavanadate dissolved in monoethanolamine [143]. The resulting catalyst was finally hydrothermally treated at 750 °C. This treatment induced a decrease in surface area but resulted in considerably higher SCR activity. In fact, more V⁴⁺ and V³⁺ species were found on the catalyst treated at 750 °C, which explains its better SCR performances.

WO₃-TiO₂ catalysts, without vanadia addition, were investigated by Shin et al. [150]. Tungsten oxide was well dispersed in the interlayer between the grains of titania, avoiding TiO₂ sintering. It created surface acid sites (mainly Brønsted), giving the material a good SCR activity.

4.2. Influence of the Preparation Method: Powder and Monolithic Catalysts

Powder V-W/TiO₂ catalysts are commonly prepared by impregnation of anatase with aqueous solutions of ammonium metavanadate (NH₄VO₃) and ammonium metatungstate hydrate (NH₄)₆(H₂W₁₂O₄₀)·xH₂O or paratungstate hydrate (NH₄)₁₀(H₂W₁₂O₄₂)·4H₂O precursors. He et al. evaluated the role of the synthesis method and showed that coprecipitation of the three elements—V, W and Ti—gave the best results in terms of SCR activity [151]. Coprecipitated catalysts possess new O-VO₃ and O-WO₄ sites that enhance the ammonia adsorption capacity. The origin of the best results obtained with coprecipitated catalysts is not clear. Titania is present as poorly ordered anatase with many defects, while acidity is not strengthened with respect to impregnated catalysts. Tungsten–titanium pillared clays were shown to be excellent supports of vanadia for the NH₃-SCR reaction [152]. A great advantage of this preparation method is the better control of the intimacy of contact between W and Ti, which are both inserted in clay pillars. The addition of S or N compounds during the preparation of TiO₂ clearly enhanced the performances.

SCR catalysts should be deposited on monoliths before practical use in DeNO_x processes. Differences in color, structure and local activity of the V-W/TiO₂ deposit depending on its location (center, monolith periphery) were studied by Wang et al. on commercial honeycomb catalysts [153]. As these catalysts were designed for denitrification of flue gases, NH₃-SCR was tested in presence of SO₂. In the monolith, initially yellow, some grey areas may appear in used catalysts. This due to changes of the vanadium valence state from +5 (V₂O₅ is bright yellow) to +4 (VO₂ is yellow darker, almost brown) and +3 (V₂O₃ is brown). This study confirms that the V-W/TiO₂ is tolerant to a certain amount of SO_x.

Monoliths pretreated in an acidic medium seem to give good impregnation characteristics [154]. The fabrication process is illustrated in Figure 17. Best results in NH₃-SCR were obtained with the monolith impregnated three times with 3% V + 10% W.

In the NH₃-SCR technology dedicated to automotive applications, ammonia is generally provided by urea decomposition/hydrolysis, even though some applications using liquid ammonia have been envisaged [155]. Combining WO₃-V₂O₅/TiO₂ (upstream) and Cu-zeolite (downstream) catalysts was proven to give excellent performances in urea-SCR technology [156]. Deterioration of activity of the W-V/TiO₂ catalyst above 270 °C was compensated by the high performance of the zeolite catalyst. Lower N₂O selectivity was observed in the dual catalyst. Similar results were obtained by combining WO₃-V₂O₅/TiO₂ and Fe-zeolite catalysts [157].

4.3. Poisoning of V₂O₅-WO₃/TiO₂ Catalysts

Apart from the physical deactivation of WO₃-based catalysts due to its working time (e.g., sintering, volatilization of active elements), the system can suffer from chemical poisoning [158]. In 2008, Kröcher et al. reported a detailed study on the deactivation of V₂O₅-WO₃/TiO₂ catalysts by inorganic impurities of lubricants, biodiesel or urea solutions [159,160]. Among all the poisons tested, potassium had the strongest effect on both catalytic activity and N₂O selectivity (at 500 °C). The poisons can be ranked as follows: K >> Ca >> Mg > Zn > P. These results prompted researchers to focus their investigations on K and Ca poisoning (Section 4.3.1). Another poison present in exhaust gases or flue gases is sulfur dioxide. The effect of SO₂ was also investigated by several authors, as presented in Section 4.3.2; poisoning by arsenic is presented in Section 4.3.3.

4.3.1. Potassium and Calcium Poisoning

Potassium is a poison of acid sites and inhibits NH₃ adsorption. Xie et al. investigated the poisoning of a W-V/TiO₂ catalyst exposed to the flue gas of a coal-fired power plant [161]. They showed that alkali contaminants (mainly K) contained in the flue gas preferentially poison the vanadium sites (V⁵⁺–OH and/or V⁵⁺=O) rather than the sites associated with tungsten oxide or TiO₂. A detailed kinetic analysis reveals that NH₃ adsorption/desorption, NH₃ oxidation and DeNO_x activity are all affected by K poisoning. Similar tendencies were demonstrated by Siaka et al. [162], who claim that potassium preferentially neutralizes strong acid sites and alters V=O redox sites, leading to stabilization of well-dispersed VO_x species. Contrasting with the results of Xie et al., Chen et al. concluded that tungsten oxide can serve as a sacrificial agent protecting vanadia from severe K poisoning [163]. The combined effect of potassium and chloride ions showed that KCl is a more severe poison than KOH [164]. A higher amount of potassium was fixed by the catalyst when KCl was used as a potassium precursor. Better performance and especially higher potassium tolerance were obtained by doping the V₂O₅/TiO₂ catalyst with 15% HPA (H₃PW₁₂O₄₀, H₄SiW₁₂O₄₀ or H₃PMo₁₂O₄₀) instead of 10% WO₃ [165]. HPA does alter vanadium dispersion and does not increase N₂O formation. A higher concentration of acid sites could explain the higher K tolerance. By studying multielement poisoning systems, Mia et al. [166] found that phosphorous–potassium combination results in lower deactivation than single potassium poisoning. The advanced explanation is that new active sites generated by phosphorous react with potassium to liberate V-OH acidic sites. Concomitantly, alkali species may provide additional basic sites for NO₂ adsorption, enabling gaseous ammonia to react with adsorbed NO_x compounds for fast SCR stoichiometry [167]. Finally, potassium resistance may be strengthened by Ce and Cu addition over V₂O₅-WO₃/TiO₂ SCR materials due to an enhancement of V⁵⁺ amount and active oxygen species [168].

Calcium poisoning was investigated in detail by Li et al. [169,170]. The degree of poisoning follows the order CaCO₃ > CaO > CaSO₄. It is linked to the ability of the precursor to form calcium tungstate [170]. CaCO₃ decreases oxygen availability, vanadium reducibility and acid site concentration. If calcium sulfate is a less severe poison than CaO and CaCO₃, it tends to increase the N₂O selectivity. The catalysts can be regenerated by using specific treatment based on 1-hydroxy ethylidene-1,1-diphosphonic acid (HEDP) liquid under weak acid environment [169]. Odenbrand investigated the effect of CaSO₄ on kinetic parameters of NH₃-SCR over W-V-TiO₂ catalysts [171]. He concluded that Ca introduced into the pore system by impregnation strongly poisons the catalyst, but in a different way from that when it is coming from the engine during operation. CaSO₄ affects both activation energy and NH₃ heat of adsorption. A moderate effect on N₂O selectivity was observed.

4.3.2. Sulfur Poisoning

Tungsten-promoted V₂O₅/TiO₂ catalysts are not resistant to SO₂ for the NH₃-SCR reaction. Antimony oxide [172], molybdenum oxide [173] or iron oxide [174] are better promoters of these catalysts for sulfur resistance. Xu et al. confirmed the superiority of V₂O₅-Sb₂O₃/TiO₂ compared with the commercial V₂O₅-WO₃/TiO₂ catalyst for SO₂ resistance [175]. Sulfur deactivation is due to the formation of a surface layer of NH₄HSO₄ [176]. Compared to tungsten, Sb-doped catalysts have a lower activity for SO₂ oxidation to SO₃, which leads to weaker deposit of ammonium sulfate. Moreover, it seems that NH₄HSO₄ is more reactive with NO when it is deposited on V₂O₅-Sb₂O₃/TiO₂ [175]. The sulfur resistance of iron-promoted WO₃/TiO₂ is also reported in [177], where non-vanadium-based catalysts present thorough regeneration. Nobia-doped materials also exhibit significant sulfur resistance [178]. Nb₂O₃-CeO₂/WO₃-TiO₂ (NbCeWTi) catalyst presents mainly ammonium hydrogen sulfate species after H₂O and SO₂ exposition that protect Ce^x+ active sites.

SO₂ resistance of V₂O₅-WO₃/TiO₂ catalysts can be improved by the addition of WO₃-graphene nanocomposite [179]. WO₃ remains well dispersed on graphene, which tends to decrease the SO₂ oxidation activity. Moreover, WO₃ acidity is preserved on graphene, essential for a good SCR activity. In another attempt, sulfur resistance was successfully increased by the addition of barium sulfate [180]. BaSO₄ contributed to reinforcing the acidity of the catalysts while blocking the SO₂ oxidation activity.

4.3.3. Arsenic Poisoning

Arsenic is present in the gas phase of power plants and may reach concentrations of up to thousands of micrograms of As₂O₃ per cubic meter [181]. Power plant gases can be treated with limestone to reduce arsenic concentration. The compound thus formed is (Ca₃(AsO₄)₂), and it may be condensed or adsorbed on ashes, decreasing the final As concentration to less than 100 µg m⁻³. However, even at this concentration, As remains a severe poison for NH₃-SCR catalysts. Deactivation of V₂O₅-WO₃/TiO₂ catalysts was investigated by Peng et al. [182,183] and Kong et al. [184]. Arsenic oxides decrease Lewis acidity and tend to form very unstable As-OH Brønsted sites. At a certain concentration, As₂O₃ is oxidized to As₂O₅, much less active for NH₃-SCR but more active for NH₃ oxidation, leading to N₂O in significant amounts. As₂O₅ forms a dense layer at the catalyst surface, which reinforces the deactivation effect of arsenic. It was proven that catalysts promoted by Mo instead of W are more resistant to As poisoning.

Synergetic poisoning effect of arsenic with potassium was evaluated in [185,186]. Results indicated that the coeffect of As + K was more dramatic than the additive effect of a single poison. As previously discussed, arsenic loading leads to the formation of As-OH acid sites, which can be neutralized by potassium deposits. In addition, the effect of charge-compensating anions was also investigated, with the following enhancement ranking sequence: SO₄²⁻ < Cl⁻ << NO₃⁻. Finally, catalysts poisoned by arsenic could be almost regenerated by sulfuric acid treatment, leading to the recovery of Lewis and Brønsted acid sites. However, the main drawback remains the formation of chelating bidentate sulfates and the loss of vanadium species [187]. Alkali solution treatment by single sodium carbonate [188] or with sodium hydroxide combined with sulfuric acid [189] also demonstrated interesting regeneration results.

4.3.4. Metal Release and Reuse of V-W-TiO₂ Catalysts

A cause of slow but irreversible deactivation of V₂O₅-WO₃/TiO₂ catalysts is the metal release at high temperature (T > 600 °C) [190,191]. Vanadium VO(OH)₃ and tungsten WO₂(OH)₂ oxo-hydroxides are responsible for the greatest part of the metal release. This effect is more important on agglomerated vanadium and tungsten particles, especially when the support undergoes severe sintering. Therefore, metal release is significantly reduced with catalysts supported on stabilized titania. Tungsten can be recovered from spent catalysts by a chemical treatment consisting of an alkaline leaching–ion exchange method [192].

Spent catalysts were also used to prepare new W-V catalysts. Huo et al. reported a feasible preparation method for a visible-light-sensitive BiVO₄/Bi₂WO₆ heterojunction photocatalyst from waste SCR catalysts [193].

4.4. Cerium-Promoted WO₃ Catalysts for NH₃-SCR

Promotion by ceria was studied in detail in the last decade. Redox properties [194,195] and oxygen mobility on ceria [196,197] could have a major impact on the NH₃-SCR reaction [198]. Ceria could be added to V₂O₅-WO₃/TiO₂ catalysts, WO₃/TiO₂ catalysts (without vanadia) or WO₃ catalysts (without vanadia and titania).

4.4.1. Ceria Added to V₂O₅-WO₃/TiO₂

The addition of ceria to V₂O₅-WO₃/TiO₂ for the NH₃-SCR reaction was investigated by Chen et al. [199]. The great merit of ceria is the enhancement of the DeNO_x activity at low temperature (below 300 °C). For instance, with a catalyst having typical composition close to that of commercial materials (1% V₂O_5–9% WO₃), a NO_x conversion of 52% is obtained at 200 °C in the following conditions: 500 ppm NO, 500 ppm NH₃, 3% O₂ and 28,000 h⁻¹. Addition of 5% CeO₂ increases NO conversion up to 88%, while manganese oxide has virtually no effect (NO conversion = 50%), and iron oxide decreases NO conversion down to 15%. Vanadium oxide being responsible for the formation of N₂O at high temperature, attempts were made by Chen et al. to significantly reduce the vanadium loading down to 0.1% while maintaining a good activity by doping the V_0.1W₆Ti with ceria [199]. Above 5% CeO₂, activity of V_0.1W₆CexTi becomes equal or superior to that of V₁W₉Ti at 200 °C. In the meanwhile, N₂O formation is virtually suppressed on the low-loaded vanadium catalysts. This effect is still more marked on Ce-doped catalysts. Addition of ceria leads to complex features since the basicity of titania seems to be reinforced while there are stronger and more active Brønsted acid sites on VWTi.

Series of CeO₂-V₂O₅-ZrO₂/WO₃-TiO₂ catalysts with different loadings of ceria were prepared by Wang et al. [200]. They confirmed the specific role of ceria in improving both the low-temperature activity and the N₂ selectivity at high temperature (less N₂O). However, the zirconia-free catalyst (CexV1/W8Ti) has a poor hydrothermal stability (750 °C, 10% H₂O, 12 h). Addition of zirconia (10%) significantly reinforces the stability of the ceria-promoted catalyst. Ceria also has a great influence on the reaction mechanism. Ce-doped catalysts (CeVZr/WTi) are extremely sensitive to the presence of NO₂ in the reaction gases and are much more active in fast SCR conditions (Equation (5)) than in standard conditions (Equation (6)). The influence of NO₂ is limited on the cerium-free catalysts (VZr/WTi), suggesting that NO₂ does not readily react via the fast SCR reaction in absence of ceria.

NO + NO₂ + 2NH₃ → 2N₂ + 3H₂O

(5)

NO + ½ O₂ + 2NH₃ → 2N₂ + 3H₂O

(6)

The group of Tianjin University investigated the method of introduction of cerium and vanadium on the commercial 10% WO₃/TiO₂ support [201]. Ceria was introduced either by impregnation of cerium nitrate (IMP) or by deposition–precipitation of cerium nitrate in the presence of ammonia (DP). Characteristics and performances of the catalysts are reported in

Table 5. Comparison of VCe/WTi DP, VCe/WTi IMP and cerium-free catalyst in NH₃-SCR. Vanadium was introduced by impregnation on the support (VWTi) or on the Ce-doped support using the VO(CO₂)₂ complex (prepared by reaction of V₂O₅ powder with oxalic acid). SCR conditions: 500 ppm NO, 500 ppm NH₃, 5% O₂ and 4% H₂O. From [201].

Catalyst	BET Area	Ce3+ (%)	NO Conv.	N2O
	(m2 g−1)	(from XPS Data)	at 250 °C (%)	at 550 °C (ppm)
VWTi	80	-	35	14
VCeWTi IMP	80	44	42	8
VCeWTi DP	83	53	67	5

.

The catalyst prepared by deposition–precipitation shows higher performances for the NH₃-SCR reaction. This is not due to changes in the textural properties but to specific properties of ceria, more reducible in VCeWTi DP. This catalyst also possesses more acidic sites (mainly Lewis sites) associated with surface wolframyl and ceria species. Youn et al. studied the role of the order of impregnation of vanadium and cerium–tungsten [202]. They showed that vanadium impregnated first (CeW/V/Ti) gave better performances. Similar topics were recently explored by Liu et al. [203], who demonstrated that the impregnation sequence of W and Ce influence the NH₃ adsorption capacity and the Ce³⁺/(Ce⁴⁺ + Ce³⁺) ratio in correlation with catalytic results. Addition of copper to ceria (Cu-Ce-W-V/Ti) reinforces the beneficial effect of ceria at low temperature [204].

The specific role of ceria on the sulfur resistance of V₂O₅-WO₃/TiO₂ catalysts was investigated by Liang et al. [205]. Ceria helps to improve DeNO_x activity and sulfur resistance. A 3% loading of ceria is optimal for better performances. Higher loadings of ceria are detrimental: they provoke a decrease in surface area and partial covering of vanadia by ceria. Complex formulae combining Ce and Mn doping were described by Zhao et al. Ce and Mn coaddition confers excellent H₂O and SO₂ resistance on V₂O₅-WO₃/TiO₂ catalysts [206]. SO₂ is mainly adsorbed on Mn⁴⁺ and Ce⁴⁺ cations, making the active surface of V₂O₅-WO₃ free. A multiplicity of valence states of the elements would be beneficial to a high NO conversion. Additionally, codoping of Ce⁴⁺ and Zr⁴⁺ enhances the tolerance to alkali metals of V₂O₅-WO₃/TiO₂ catalysts, as K-poisoning resistance occurs by blocking potassium in the form of Ce-O-K structure [207].

4.4.2. Ceria Added to WO₃/TiO₂ (Without Vanadium)

A number of works were carried out on vanadium-free catalysts to better study the interaction between ceria and tungsten and its role in the SCR reaction. Chen et al. prepared CeO₂/TiO₂ and CeO₂-WO₃/TiO₂ catalysts by an ultrasonic method and investigated the reactivity of adsorbed ammonia with NO + O₂ by DRIFT [208]. Comparison of CeO₂/TiO₂ and CeO₂/WO₃/TiO₂ catalysts shows that tungsten has a dramatic effect on the SCR reaction: (i) it allows conversion of NO at lower temperatures (50% conversion is reached at 165 °C on CeWTi instead of 225 °C on CeTi); (ii) it increases the selectivity for N₂ at high temperature (less N₂O). Tungsten increases the number of Brønsted sites (virtually absent on CeTi) and tends to accelerate the cerium reduction (more Ce³⁺ in CeWTi). Another report by Chen et al. confirmed the important role of tungsten by increasing the oxidation activity of NO to NO₂, allowing the catalyst to work in fast SCR conditions [209]. Interaction between ceria and tungsta seems to play a decisive role in SCR activity and selectivity. Geng et al. compared WTi and CeWTi catalysts and showed that ceria is effective for the inhibition of N₂O formation when it is added to WO₃-TiO₂ [210].

Great efforts were made in the last decade to improve the preparation of cerium-based tungsten catalysts. Michalow-Mauke et al. developed a preparation method based on flame-spray (FS) pyrolysis of Ce, W and Ti precursors in tetrahydrofuran [211,212]. Compared to catalysts prepared by impregnation, catalysts prepared by FS synthesis show superior performances. This is due to a better dispersion of ceria and tungsta and to a greater interaction of these elements with titania leading to highly active Ce–O–W (especially Ce³⁺–O–W⁶⁺) and Ce–O–Ti sites [212]. The fact that FS synthesis mainly produces rutile instead of anatase does not seem to hamper the performance of the catalysts prepared by this technique. Sol–gel techniques were also used by Jiang et al. to prepare CeO₂-WO₃/TiO₂ catalysts [213]. Butyl titanate was dissolved in a solution of ethanol/water/nitric acid and then mixed with cerium nitrate and ammonium metatungstate. Different catalysts were prepared with 10% WO₃ and various loadings of ceria (from 0 to 40%). DeNO_x activity of these materials is visualized in Figure 18.

The catalyst with 20% CeO₂ shows the best performance in NH₃-SCR. It has also a good resistance to SO₂/H₂O when 500 ppm SO₂ and 10% H₂O are added to the reactant mixture. The same optimal composition (20% CeO₂) had already been observed on a series of Ce_xW₂₀TiO₂ catalysts prepared by homogeneous precipitation with urea aqueous solution [214]. In every case, these CeO₂-WO₃-TiO₂ catalysts show exceptional resistance to SO₂.

Improved methods of preparation of CeO₂-WO₃/TiO₂ catalysts were recently reported, either using H₂O₂ as a promoter of active sites [215] or leading to 2D materials (Ce_0.184W_0.07Ti_0.748O_2−δ nanofibers prepared by electrospinning) [216]. H₂O₂ modification improves both the BET surface area and the number of Brønsted sites, while nanofibers seem ideal for a good performance at low temperature owing to the formation of a great number of oxygen vacancies. Salazar et al. investigated the method of impregnation of Ce and W on TiO₂ [217]. They showed that successive impregnation (with intermediary calcinations) of Ce on W/Ti led to higher performances than when catalysts were prepared by coimpregnation. Successive impregnation favors the formation of Ce-O-W bridges, which play an essential role in the SCR reaction. No clear correlation was observed between the SCR activity and the degree of Ce reduction or the number of Brønsted sites, even though these parameters are probably important for a good DeNO_x activity.

Monolithic catalysts were prepared and tested by Cao et al. [218]. A TiO₂-SiO₂ powder was mixed with aqueous solutions of ammonium paratungstate and cerium nitrate. Ammonia was added up to pH = 10 to form a slurry which was extruded, dried and calcined at 550 °C. Silica was added to titania to increase its thermal stability. Excellent performances were obtained, superior to those of powder catalysts especially at high temperatures.

4.4.3. Ceria Added to WO₃ (Without Vanadia or Titania)

In order to avoid complex interactions between Ce, W, V and Ti, CeO₂-WO₃ catalysts were prepared and tested in NH₃-SCR. Zhang et al. compared the behavior of ceria doped with various acid promoters: phosphotungstic, silicotungstic and phosphomolybdic acids or ammonium sulfate [219]. Ceria doped with phosphotungstic (CeO₂-P-W) or silicotungstic acid (CeO₂-Si-W) showed the highest NO_x conversion (Figure 19). However, CeO₂-P-W exhibited the best N₂ selectivity over the whole temperature range, especially at T > 450 °C where N₂O could be formed.

The mode of preparation of ceria for the synthesis of P-W/CeO₂ catalysts was further investigated by Song et al. [220]. They compared hydrothermal (cerium nitrate + glucose + acrylic acid aged at 160 °C in autoclave), sol–gel (cerium nitrate in citric acid) and precipitation (cerium nitrate solution + ammonium carbonate) techniques. The P-W/CeO₂ catalyst with ceria prepared by hydrothermal technique gave the highest performances. It combined the highest BET area, the highest Ce³⁺ concentration and an adequate balance between Brønsted and Lewis acid sites. Wang et al. showed that the ceria morphology (cubes, particles, rods) would have a great impact on the performance of W-CeO₂ catalysts for the NH₃-SCR reaction [221]. Ceria supports were prepared by hydrothermal methods according to the synthesis procedures developed by Peng et al. [222]. Ceria nanorods were prepared from cerium acetate, while nanoparticles and nanocubes were synthesized using cerium nitrate. Nanocubes expose preferentially (100) planes while nanorods expose both (110) and (100) planes. Contrasting with these surface structures, (111) planes are essentially found with nanoparticles. Nanoparticles are more active than nanocubes and nanorods for the NH₃-SCR reaction, which tends to prove that WO₃ attached to (111) faces of ceria gives the best performances for NO_x abatement.

Modification of ceria by manganese shows that N₂O can be avoided by the incorporation of WO₃ into the catalyst. Better performances, with a good SO₂ and CO₂ resistance, were obtained with W_0.1Mn_0.4Ce_0.5 composition [223]. One of the roles of ceria is to promote redox properties by oxygen mobility improvement. As cerium–zirconium oxides have superior properties of reduction [194,224,225], it was logical to replace pure ceria with cerium–zirconium oxides in the preparation of supported WO₃ catalysts. These formulations were explored by Ning et al. [226]. Methods of preparation of CeO₂-ZrO₂-WO₃ (CZW) catalysts were similar to those developed by Song et al. for tungsten catalysts supported on ceria [220]. Catalysts prepared by hydrothermal methods show the best performance for the NH₃-SCR reaction. This is due to a higher tungsten dispersion and to the coexistence of Brønsted and Lewis acid sites, while CZW catalysts prepared by other techniques possess only Lewis sites. Acid properties of CeO₂-ZrO₂-WO₃ catalysts are greatly influenced by the state of tungsten oxide [227]. Moreover, the formation of amorphous W species resulted in the abundance of Ce³⁺ and oxygen vacancies with a correlative increase in the NO oxidation activity. Recently, great efforts were devoted to reinforcing the stability of SCR catalysts. Liu et al. showed that doping WO₃/Ce_0.68Zr_0.32O₂ with silica significantly improved the thermal stability of the catalyst (10% H₂O, 800 °C) [228]. Silica allows maintaining the acidity of the catalyst after thermal treatment. Moreover, it inhibits the formation of cerium tungstate Ce₂(WO₄)₃, which is detrimental to the catalyst performances. Similar effects were obtained by doping the CeZrOx support with alumina in monolithic catalysts [229]. The Ce_xZr_1−xO₂ composition also impacts the WO₃/Ce_xZr_1−xO₂ behaviors. The characterization of 9% WO₃/Ce_xZr_1−xO₂ catalysts with various ZrO₂ weight ratios in Ce_xZr_1−xO₂ (30%, 42%, 60% and 80%) indicated that the increase in the zirconium content enhanced the acidity (number and strength of acidic sites). Accordingly, the NH₃-SCR activity also increased [230]. Note that after WO₃ addition, the basic NO_x storage sites of Ce_xZr_1−xO₂ were fully altered while the oxygen storage capacities (OSC) were dramatically decreased for all samples.

4.5. Iron-Promoted WO₃ Catalysts for NH₃-SCR

Iron catalysts being promising SCR catalysts, especially when Fe is inserted in zeolites [231], it was logical that researchers attempt to associate Fe and W for the DeNO_x reaction. Fe-W mixed oxides were studied for this application by Li et al. [232] and Wang et al. [233]. Catalysts were prepared by ammonia precipitation of ammonium paratungstate (APT) and iron nitrate [232], or by urea precipitation of ammonium metatungstate (AMT) and iron nitrate [233], which may explain some differences in performances for the two series of catalysts. Activity maximum was observed by Li et al. with the FeW5 sample (i.e., having a Fe/W molar ratio of 5), while the optimal performances were found by Wang et al. for a Fe/W ratio of 3 (Fe_0.75W_0.25Ox sample). Location of acid sites would also be different: B sites on tungsten (W-OH) and L sites on iron (Li et al.), or B sites on FeWO₄ and L sites on Fe₂O₃, i.e., Fe oxide not associated with tungsten (Wang et al.). In fact, Wang et al. proposed that the good performances of Fe_0.75W_0.25Ox would be due to a fine interaction between Fe₂O₃ and FeWO₄ with an easier electron transfer from W⁶⁺ sites to Fe³⁺ sites, which favors the formation of NO₂. Wang et al. also reported that their mixed oxide catalysts were extremely resistant to SO₂ poisoning, especially at high temperature (T > 300 °C) [234]. Similar conclusions were reported for tungsten-free catalysts, where Fe_0.1V_0.1TiO_x catalyst showed the optimal NH₃-SCR performance and excellent SO₂ resistance [174].

Direct impregnation of tungsten on hematite Fe₂O₃ was reported by Liu et al. [235]. Hematite was first prepared by urea precipitation of Fe nitrate. The solid (dried and calcined at 500 °C) was then impregnated with ammonium metatungstate solution in oxalic acid. Optimal performances were observed with the 5% WO₃/Fe₂O₃ sample. The reverse impregnation (iron on WO₃ nanorods) was studied by Li et al. [236]. It is difficult to compare the two methods of preparation. Nevertheless, both led to comparable DeNO_x activity with a 50% NO conversion around 275 °C, even though the catalyst evaluation was carried out in somewhat different conditions.

Iron–tungsten was also supported on zirconia [237] or cerium–zirconium oxide for monolith preparation [238]. In absence of iron, supported tungsten shows a good ammonia oxidation activity but virtually no SCR activity. The presence of iron is required to create SCR catalysts whose activity is linked to the formation of Fe³⁺ Lewis sites. The main role of these sites would be to promote NO oxidation to NO₂, the first step in the SCR mechanism. These studies on ZrO₂ and CeZrO_x were carried out with relatively high W loading (Fe/W molar ratio close to 1). Magnetic iron oxides doped with tungsten and cerium were also prepared as SCR catalysts [239]. In these materials, iron is mainly in the form of γ-Fe₂O₃ and α-Fe₂O₃, but other forms can be present since W and Ce tend to create highly dispersed iron species. Compared to previous catalysts synthesized over zirconia or CeZrO_x, higher iron loadings were used to prepare FeCeWO_x catalyst (most active material: Fe_0.90Ce_0.05W_0.05 and Fe_0.85Ce_0.10W_0.05). Copper being an active component of SCR catalysts, it was tempting to promote iron catalysts with copper. Ma et al. reported a one-pot preparation of Cu_0.02Fe_0.2W_xTiO₂ materials (x varying from 0.01 to 0.03), which were tested in the SCR reaction [240]. A moderate amount of tungsten significantly improves stability and acid and redox properties, leading to superior performances for the SCR reaction, even in the presence of water and SO₂ (best formula: Cu_0.02Fe_0.2W_0.02TiO₂). Complex catalysts including several promoters (Fe, V, Mn, W, Ce) were also developed [241]. It appears that multiple redox pairs (Fe²⁺/Fe³⁺, V⁴⁺/V⁵⁺, Mn²⁺-Mn³⁺/Mn⁴⁺) would play an important role in the reaction.

4.6. Manganese-Promoted WO₃ Catalysts for NH₃-SCR

Titania-supported manganese tungstate catalysts were prepared by Kong et al. in order to replace the standard V-W/TiO₂ catalyst for the NH₃-SCR reaction [242]. The solution combustion method with Mn, W and Ti precursors mixed with glycine is suitable for preparing high-surface-area materials (up to 280 m² g⁻¹). Mn/TiO₂ without W showed a high activity at low temperature, which declined rapidly above 300 °C. Tungsten allowed enlarging the activity window up to 400 °C. The highest performance was obtained with the Mn_0.1W_0.05Ti_0.85O_2−δ catalyst. Similar materials were reported by Shin et al. [243] and Wang et al. [244]. In these studies, high tungsten contents were used (15% in [243] and 25% in [244]), which seems beneficial to improve the performance at low temperature. For instance, Wang et al. obtained a 50% NO conversion at 40–50 °C and 100% NO conversion from 80 to 280 °C over the W_0.25Mn_0.25Ti_0.5O₂ catalyst. The role of tungsten is to facilitate Mn redox capacity, leading to high concentration of Mn⁴⁺ and reactive oxygen species (Figure 20).

Additionally, a study investigating the vanadium loss from V-W/TiO₂ catalysts was conducted in [245] and reports that manganese is the best candidate to make up for the loss of SCR activity caused by the decrease in V₂O₅ loading (50%), in opposition to other transient metals as Nb, Co, Cr, Cu or Ce. Note that similar behaviors are denoted for tungsten-free catalysts: Mn-loaded catalyst (Mn₅V₁Mo₃Ce₇/Ti) exhibits the optimal SCR performance associated with a large number of acid sites and high redox properties [246].

Many studies were devoted to the promotion by cerium of manganese-based SCR catalysts. Nie et al. investigated the effect of several acidic oxides (Nb₂O₅, WO₃ and MoO₃) on the performance of MnO_x-CeO₂ catalysts [247]. NbCeTi and WCeTi are very active at 200 °C, while MoCeTi is active only at high temperature. WO₃-promoted catalyst has the broadest operation window with significant NO conversion above 300 °C. Ma et al. investigated the effect of WO₃ doping on the performances of MnO_x-CeO₂ catalysts for the SCR reaction [223]. The concentration of manganese in ceria was kept constant (Mn/Ce = 0.4) while the concentration of tungsten was varied from 0.03 to 0.2. Undoped Mn_0.4Ce was active only at low temperature: its activity decreased sharply above 200 °C. WO₃-doped catalysts showed far better activity above 200 °C, with an optimum for the W_0.1Mn_0.4Ce sample. Mn_0.4Ce has only Lewis acid sites. Doping with tungsten generates Brønsted sites allowing activity at high temperature. Moreover, High SO₂ resistance is also achieved on W_0.1Mn_0.4Ce by suppression of SO₂ oxidation activity. Promotion by tin of MnO_x-CeO₂ support seems to favor low-temperature SCR activity [248]. High NO conversion is obtained on the SnMnCeO_x support (50% at 60 °C but only 50% at 300 °C). Addition of tungsten shifts the conversion profile to higher temperatures with good activity and excellent N₂ selectivity up to 300 °C.

In several studies by the group of Hong (South Korea), WMnCe-based catalysts were supported on titania [249,250,251]. In the first two studies, WMnCeTi samples were prepared by wet impregnation without any control of the pH, while in the third study, a strict control of the pH between 2.8 and 1.7 was applied by addition of oxalic acid. Controlling the pH of the slurry around 1.7 leads to catalysts more active for the NH₃-SCR. This result is linked to the fact that an acidic pH increases the formation of surface Mn⁴⁺ and Ce³⁺, while NO would be less strongly adsorbed. A good synergetic effect was observed with zirconia as support for manganese. Increasing the concentration of tungsten on MnZrO_x allows obtaining catalysts more active and more selective for N₂ up to 400 °C [252]. XPS results and DRIFT studies on the most active catalyst (15%W/MnZrO_x) showed that the redox cycle (Equation (7)) would promote the electron transfer between W and Mn, contributing to NH₃ activation.

Mn⁴⁺ − O²⁻ − W⁵⁺ ⇔ Mn³⁺ − O²⁻ − W⁶⁺

(7)

Developing active catalysts over a wide range of temperatures is a challenge for the NH₃-SCR reaction. Combining two catalysts (MnO_x-CeO₂/TiO₂ (MnCeTi) and V₂O₅-WO₃/TiO₂ (VWTi)), Zhang et al. were able to obtain high NO conversion from 150 to 400 °C [253]. The best configuration was obtained when VWTi was set at the fore part and MnCeTi at the rear part of the catalyst bed (Figure 21, CC-B curve). MnCeTi is much more active than VWTi for the NO oxidation reaction, allowing work in fast SCR conditions over a wide range of temperatures. MnCeTi is also more active for the NH₃ oxidation reaction, which has a less detrimental effect in CC-B configuration (VWTi + MnCeTi).

5. Other DeNO_x Applications of WO₃-Doped Catalysts

5.1. Tungsten Catalysts for the NO_x Trap–SCR Coupled System

One of the technologies for NO_x abatement in diesel engine exhausts is the NO_x trap system, also called NO_x storage reduction (NSR) or lean NO_x trap (LNT) [254]. The catalyst (typically Pt/BaO-Al₂O₃) works according to a sequential operation: (i) during one or two minutes, NO_x from the exhaust gases is stored on the catalyst as nitrate and nitrite species; (ii) during a few seconds, hydrocarbons are added to the exhaust gases for the reduction step, and adsorbed nitrates and nitrites are reduced to N₂. However, some ammonia may also be produced (ammonia slip), which should be eliminated on an ammonia oxidation catalyst. An alternative solution would be to use this in situ produced ammonia on an SCR catalyst to transform the residual NO_x not yet converted. This is the NO_x trap–SCR coupled system [255]. Several studies considered the coupling of Pt-BaO-Al₂O₃ (NSR catalyst) with Cu-zeolites (SCR catalyst) [256,257,258,259] or with Fe-zeolites [260,261]. Can et al. investigated the use of WO₃/CeZrOx (as SCR material) coupled to a usual Pt/BaO/Al₂O₃ catalyst [230]. The effect of adding an SCR catalyst (WO₃/Ce_0.58Zr_0.42O₂) downstream of the NSR catalyst is depicted in Figure 22. SiC being inert for all the reactions, NSR + SiC represents the performance of the NSR catalyst alone. The effect of adding an SCR catalyst to the NSR material is clearly visible in the figure: more NO_x is converted and more NH₃ produced on the first bed is consumed. N₂O is never produced on the NSR catalyst alone or on the coupled system NSR+SCR. In fact, a part of the ammonia produced on the NO_x-trap catalyst reacts with oxygen (NH₃ oxidation reaction or SCO). Fortunately, both SCR and SCO are selective for N₂ on these materials.

Other tungsten catalysts (WO₃/Al_0.2Ce_0.4Ti_0.4, WO₃/Al_0.2Ce_0.16Zr_0.32Ti_0.32 and WO₃/Al_0.1Si_0.1Ce_0.16Zr_0.32Ti_0.32, selected among 30 formulations) were also tested in the NO_x trap–SCR coupled process [262]. Ammonia release and its use in oxidation (SCO) or reduction (SCR) is shown in Figure 23 for four configurations of NO_x trap–SCR coupled systems. Configuration 4 with silica-containing SCR catalyst offers the best performances, especially at 300 °C. Presence of silica increases Lewis acidity and more strongly increases Brønsted acidity, which can explain the good behavior of the Si-doped catalyst.

5.2. Tungsten Catalysts for NO_x Reduction by Other Reductants

5.2.1. Reduction by Hydrogen (H₂-SCR)

NO_x reduction by hydrogen is well adapted to depollution of sites where hydrogen is available (e.g., refineries, H₂ plants) [263]. The authors of [263] investigated the H₂-SCR over WO_x-ZrCe (zirconium rich) and WO_x-CeZr (cerium rich) catalysts. Reaction conditions were 520 ppm NO + NO₂ (NO/NO₂ = 9), 5% O₂ and 10% CO₂. The reaction can be written as follows (Equation (8)):

NO + H₂ → ½ N₂ + H₂O

(8)

However, formation of nitrous oxide (Equation (9)) and undesired hydrogen oxidation (Equation (10)) may be observed:

2NO + H₂ → N₂O + H₂O

(9)

H₂ + ½ O₂ → H₂O

(10)

Maximum activity is obtained at 250 °C over WO_x-CeZr and at 300 °C over WO_x-ZrCe. The Zr-rich catalyst WO_x-ZrCe shows better performances if the whole 150–600 °C temperature range is considered. This could be due to a higher concentration of acid sites on WO_x-ZrCe than on WO_x-CeZr (about twice). Selectivity for N₂ is close to 80–90%, with maximum formation of N₂O around 20 ppm over WO_x-CeZr and 15 ppm over WO_x-ZrCe. Addition of 7% water slightly decreases the NO_x conversion but improves the N₂ selectivity.

Platinum is one of the most active metals for the NO reduction by H₂ at low temperature [264,265,266]. Platinum activity and selectivity are extremely sensitive to the nature of support. For instance, Pt/ZrO₂ is superior to Pt-Al₂O₃ (more active and more selective for N₂), and promotion by WO₃ improves the performances of Pt/ZrO₂ [267]. NO₂ and nitrate species are thought to be essential intermediates in the reaction [265]. However, other authors found that NO adsorption and decomposition on Pt would be the most important step of the reaction [268,269]. Maximum NO conversion is observed between 90 and 130 °C. Zhang et al. reported a NO conversion of 91% at 110 °C over 0.1%Pt-1%W-HZSM-5 in the following reaction conditions: 910 ppm NO + 90 ppm NO₂ + 5000 ppm H₂ + 10% O₂. Tungsten oxide is thought to maintain Pt in the metallic state even in highly oxidizing conditions. It suppresses NO₂ adsorption on Pt and inhibits the formation of nitrate species. It accelerates the dissociation of NO and H₂ on Pt. The main drawback of H₂-SCR is the lack of N₂ selectivity due to ammonia formation. Platinum being a poor NH₃-SCR catalyst, this ammonia cannot be used for increasing the selective conversion of NO to N₂. By contrast, Pt-WO₃ is an excellent catalyst for selective oxidation of NH₃ (SCO), but only at higher temperatures (250–300 °C) [270]. Replacing Pt by Ru does not significantly improve the SCO reaction at lower temperatures [271], which does not allow selective conversion of ammonia formed in H₂-SCR.

5.2.2. Reduction by Ethanol (C₂H₅OH-SCR)

Alumina-supported silver catalysts are very active for the NO_x reduction by oxygenated compounds, especially ethanol or acetone [272]. It seems that the formation of isocyanate species (-NCO) is a key step of the ethanol-SCR reaction over Ag/Al₂O₃ [273,274] even though many other species may be detected [275,276]. This remarkable activity of silver catalysts for ethanol-SCR prompted Barreau et al. to imagine a combination between Ag/Al₂O₃ and a NH₃-SCR catalyst (WO₃/CeZrO₂) using C₂H₅OH-NH₃ mixture for NO_x reduction [277,278]. The effect of silver alone for the C₂H₅OH-SCR and for the C₂H₅OH-NH₃-SCR and finally the effect of adding WO₃/CeZrO₂ are illustrated in Figure 24.

Silver alumina alone possesses a significant activity for the ethanol-SCR reaction (blue curve). The activity is enhanced by the presence of ammonia (red curve). In the dual bed (Ag + W) system (green curve), an increased conversion of NO_x is observed, and simultaneously the outlet concentration of ammonia is decreased.

The respective roles of Ag/Al₂O₃ and WO₃/CeZrO₂ in this complex reaction were summarized in the review by Barreau et al. [279]. Figure 25 illustrates the reaction pathway identified in the ethanol-NH₃ SCR reaction over the Ag/W dual bed. Note that WO₃/CeZrO₂ is then more suitable than a copper-exchanged zeolite (2.5% Cu–FER) as an SCR catalyst. Despite a significantly higher ammonia conversion rate using the zeolite, NO_x abatement is lower because ammonia and ethanol strongly interact together on Cu2.5–FER [278].

5.3. Conclusions

Tungsten-based catalysts are highly referenced as active samples in NO_x reduction abatement for a wide range of reducers or processes. Commonly used in V-W/TiO₂ catalysts for NH₃-SCR application, WO₃ is known to increase the activity, widen the temperature window, improve the resistance to various poisons and lower ammonia oxidation activity by O₂. Tungsten is also involved in V-O-V species formation by confining vanadia in small clusters leading to oligomeric vanadia (V₂O₅) sites, demonstrating higher NH₃-SCR activity than monomeric vanadyl compounds. WO₃ is also associated with ceria as redox support, with or without vanadia or titania, for low-temperature activity and N₂O limitation emission. Tungsten increases the number of Brønsted sites, favors the formation of Ce-O-W bridges and tends to accelerate the cerium reduction.

6. Total Oxidation of Volatile Organic Compounds in Gas Phase and Gas Sensors

Owing to the multiple oxidation states of tungsten and rapid diffusion of surface oxygen, WO₃ is also a candidate for oxidation reaction in gas phase [280], VOC oxidation (Section 6.1) and gas sensor application (Section 6.2).

6.1. VOC Oxidation in Gas Phase on Tungsten Catalysts

The first part of this section is focused on the non-photocatalyzed reactions. However, WO₃ is response-sensitive to light up to 480 nm, which also makes it a good candidate for photoassisted VOC oxidation (Section 6.1.2). Note that WO₃-based photacatalysts for liquid phase applications are depicted in Section 7.

6.1.1. VOC Oxidation on Tungsten Catalysts (Non-Photocatalyzed Reactions)

Balzer et al. showed that WO₃ alone (7.5 m² g⁻¹) can efficiently catalyze BTX oxidation [281]. The reactivity of the different hydrocarbons is as follows: benzene (T₅₀ = 250 °C) > toluene (T₅₀ = 340 °C) > m-xylene (T₅₀ = 420 °C) ≈ p-xylene (T₅₀ = 430 °C). Oxidation activity of WO₃ is linked to the presence of W⁶⁺, W⁵⁺ and W⁴⁺ surface species generating reactive oxygen species.

However, WO₃ is generally used in supported catalysts. Pansare et al. investigated NH₃ and toluene decomposition on tungsten carbide (WC) and tungstated zirconia (WZ) [282]. Both WC and WZ catalysts were active for the simultaneous decomposition of NH₃ and toluene at 700 °C in the presence of H₂, CO, CO₂ and H₂O. Benzene is formed by steam dealkylation of toluene. The WO₃-V₂O₅-TiO₂ catalyst usually employed for NH₃-SCR reaction (see Section 4) was proven to possess good oxidation activity for VOC abatement. Debecker et al. studied the total oxidation of benzene and chlorobenzene on a catalyst containing 3% WO₃ (or MoO₃) supported on titania variously loaded with vanadia (3 to 10% V₂O₅) [283]. The main results of this study are summarized in

Table 6. Total oxidation of benzene and chlorobenzene (CB) at 300 °C on V₂O₅-TiO₂, 3%WO₃-V₂O₅-TiO₂ or 3%MoO₃-V₂O₅-TiO₂ catalysts. Reaction conditions: 100 ppm benzene or chlorobenzene, 20% O₂ in He. VVH = 37,000 h⁻¹. Adapted from [283].

	Benzene Conversion @ 300 °C (%)			CB Conversion @ 300 °C (%)
Vanadium Loading	3% V2O5	5% V2O5	10% V2O5	10% V2O5
No promoter	23	71	86	75
WO3-promoted	34	77	98	93
MoO3-promoted	43	83	95	93

.

Vanadium oxide was then the active phase for VOC oxidation. However, tungsten and molybdenum oxides significantly increased V₂O₅ activity. It is worth noting that promotion by WO₃ allowed reaching almost 100% conversion of benzene with the 10% V₂O₅ catalyst. WO₃ and MoO₃ promotion were also beneficial for chlorobenzene conversion, even though total oxidation cannot be reached at 300 °C. Benzene oxidation over WO₃-V₂O₅-TiO₂ catalysts was also investigated by Lu et al. [284]. Traces of benzene (1–10 ppm) were treated in a gas containing CO₂, CO, O₂ and H₂O, simulating a flue gas issued from a methane burner. HCl (50 ppm), SO₂ (400 ppm), NO (300 ppm) and NH₃ (360 ppm) were added to simulate a waste incineration flue gas atmosphere. Due to the very low concentration of benzene, its conversion was measured by resonance-enhanced multiphoton ionization time-of-flight MS (REMPI-TOFMS), a technique well adapted to analyze minute traces of hazardous air pollutants [285]. The most active catalyst was supported on a high-surface-area TiO₂ (166 m² g⁻¹, quasipure anatase) with a low loading of vanadia (0.8 wt.%) promoted by 6 wt.% WO₃. About 80% of benzene could be eliminated with little variation of efficiency when other pollutants (NO, NH₃, HCl, SO₂) were present.

Other tungsten-based catalysts were employed for VOC abatement. For instance, magnesium tungstate (MgWO₄)-based catalysts were evaluated by Gancheva et al. in CO and hydrocarbon oxidation [286]. Four catalysts were tested: pure MgWO₄ (5.2 m² g⁻¹), MgWO_4–3%WO₃ (4.0 m² g⁻¹) and these materials promoted by 0.5% Pd. Without palladium, pure MgWO₄ was more active than the same support enriched in WO₃. With Pd-promoted catalysts, two opposite behaviors were observed: MgWO₄/3%WO₃/0.5%Pd was the most active catalyst in hydrocarbon combustion (toluene, n-hexane) while MgWO₄/0.5%Pd (with no tungstate) was much more active for CO oxidation. Gancheva et al. proposed that CO and HC would be activated on MgWO₄ and WO₃ by two different mechanisms depending on the nature of the reactant CO or HC. The role of palladium would be to favor O₂ adsorption and activation.

Catalytic abatement of trichloroethane (TCE) was performed on complex materials consisting of W-Mo bronzes (W-Mo/Nb/V/P) [287]. The best performances were obtained on the mixed W-Mo bronze with a W/Mo ratio close to 1. Complete composition (based on W + Mo = 1) was Mo(0.54)/W(0.46)/Nb(0.41)/V(0.20)/P(0.08). A 50% decomposition of TCE was achieved at 300 °C on this material, while a reference zeolite (HMOR) gave 50% conversion at 470 °C. The performance of this bronze catalyst was linked to its acidity (B sites only present on W-containing materials) and to its high oxygen mobility in the bulk, as revealed by ¹⁶O/¹⁸O exchange.

Tungsten oxide is also a good promoter of diesel oxidation catalyst (DOC) allowing increasing activity of noble meta—alumina catalyst: a gain of 20 °C on the light-off conversion of HCs was attained with 1% WO₃ on the reference PtPd/Al₂O₃ catalyst [288].

For ambient air purification, the very mild operating conditions require the development of fine structures/morphologies to improve the number of surface oxygen vacancies. With this aim, formaldehyde oxidation can be performed at room temperature by 1% Pt-doped WO₃ nanoflakes assembled into hollow microspheres (23 m² g⁻¹), in which the porous architecture promotes diffusion and adsorption of HCHO [289]. However, photoassisted catalysis appears more suitable for the very mild operating conditions.

6.1.2. WO₃-Based Photocatalysts for VOC Oxidation

WO₃-based materials as photocatalysts are mostly dedicated to reactions occurring in liquid phase (Section 7, in which general information about WO₃-based photocatalysts is provided). However, few recent studies reported that such catalysts also demonstrated intersecting behaviors in gas phase, especially for ambient air purification. To overcome drawbacks of bare WO₃, such as photocorrosion and unsuitable bandgap structure for the reduction of molecular oxygen, improvement can be obtained by structure/morphology control and/or doping. For instance, WO₃ nanoparticles obtained by gas phase method and annealing at 600 °C were reported to be much more active than commercial WO₃ in acetaldehyde oxidation. Oxidation rate can be further significantly improved by addition of ZrO₂, which acts as a sorbent for the acetic acid intermediate species to release the WO₃ surface. Moreover, Pt or Ru addition allowed the total mineralization into CO₂ and H₂O [290,291]. A physical mixture of WO₃ and CeO₂ also exhibited excellent photocatalytic activity in acetaldehyde oxidation, which was attributed to the electron scavenging property of CeO₂ aiding charge carrier separation [292]. WO₃ can be also associated with the other usual photocatalyst, namely TiO₂. To improve the visible light sensitivity of WO₃ nanoparticles impregnated into a commercial TiO₂ powder, Balayeva et al. successfully added Fe(III) nanoclusters. The observed enhancement in acetaldehyde photooxidation was then attributed to the promotion of multielectron reduction processes [293]. However, Caudillo-Flores et al. pointed out that a cautious approach must be adopted when interpreting the results. They prepared TiO₂/WO₃ samples with various W/Ti atomic ratios from 0 to 0.5, leading to various structures ranging from truly doped samples in which tungsten was exclusively located in lattice positions of the anatase structure to composite catalysts where nanosized tungsten species were supported over TiO₂. The authors highlighted that both the reaction rate (in toluene and styrene photo-oxidation) and the apparent quantum efficiency can lead to misleading results in terms of the most active TiO₂/WO₃ sample(s) as well as the (positive/negative) magnitude in comparison with bare titania reference [294].

Implementation of photocatalysts in building windows is of major interest in air purification. Li et al. proposed a g-C₃N₄@Cs_xWO₃ heterostructure as a coating for a multifunctional smart window for UV-isolating, Vis-penetrating, NIR-shielding and photocatalytic activity. These composites display excellent formaldehyde and toluene decomposition properties. The shielded NIR light is used instead of wasted as heat, while the C₃N₄@Cs_xWO₃ structure promotes the separation of charge carriers and then enhances photocatalytic oxidation. Moreover, the small polaron can jump from localized states to the conduction band of Cs_xWO₃ under NIR irradiation (730–1100 nm), resulting in an NIR-catalytic reduction [295]. Note that this kind of structure was also proposed as a photocatalyst for water purification (see Section 7.6.3 and Section 7.6.4).

6.2. Gas Sensors Using Tungsten-Based Catalytic Materials

Since the mid-1950s, gas sensors have experienced great development for environmental and safety applications. Gas sensors are based on the measurement of electrical behaviors caused by chemical changes. Expected properties are sensibility, selectivity, stability, repeatability and response time. To build low-cost sensors, the measurement of the resistivity of heated semiconductor oxides is suitable thanks to the cost of raw materials and the convenience of microelectronic integration. Details of the internal structure of the sensors, including the heater, electrodes and external circuit, are out of the scope of this review.

The sensing is related to changes in the electrical conductivity, which are mainly attributable to changes in the oxygen concentration at the oxide surface. Many oxides exhibit conductivity changes in presence of gases, but most of the commercialized sensors are based on SnO₂, WO₃ or ZnO. As reported in the recent review by Dong et al., a huge number of works have been devoted to tungsten oxide based sensors [296]. Tungsten oxide is sensitive to many gases such as O₂, O₃, CH₄, CO, H₂, NH₃, C₃H₈, NO, NO₂ and H₂S, with operating temperature in the 250–450 °C range [297,298]. WO₃ is an n-type oxide; i.e., the adsorption of surface oxygen atoms form an electron depletion region (Equations (11)–(13)), depending on the temperature of the material, leading to a potential barrier. Consequently, the adsorption of an oxidizing gas increases the resistivity of WO₃ [299,300].

O₂ + e^– → O₂^– (low temperature)

(11)

O₂^– + e^– → 2O^– (medium temperature)

(12)

O^– + e^– → O^2– (high temperature)

(13)

The formation of oxygen anions (Equation (13)) should be avoided (via the temperature control of the sensor) because of their tendency to be incorporated in the bulk of the material.

Recent studies about WO₃-based sensors are mainly dedicated to NO₂ detection (Section 6.2.1), but other gases are also considered (Section 6.2.2).

6.2.1. WO₃-Based Sensor for NO₂ Detection

Since WO₃ is an n-type semiconductor, NO₂ adsorption leads to anionic adsorbates on the surface of WO₃ according to reaction (14):

NO₂(gas) + e − (from surface) ⇆ NO₂⁻

(14)

Han and Yin performed density functional theory (DFT) calculations to study the adsorption characteristics and electron transfer of nitrogen dioxide on O- and WO-terminated WO₃ (001) surfaces with oxygen vacancies [301]. It was found that NO₂ is (i) oxidized into nitrate on the bridging oxygen atom from an oxygen defect of the O-terminated WO₃ (001) surface and (ii) dissociated on a WO-terminated (001) surface: one oxygen atom from NO₂ fills the oxygen vacancy, and the resulting NO fragment is adsorbed onto a W atom. Both of these adsorption models are responsible for an increase in the electrical resistance of WO₃. In a very recent study, Yang et al. showed by in situ DRIFT spectroscopy that NO₂ and NO exhibit similar interaction with the surface of tungsten oxide; both nitrogen oxides were detected as oxidizing gases [302].

The structure/morphology of WO₃ is a key factor in the sensor response, and this topic is particularly developed in the recent literature. For instance, comparison of Ni-doped WO₃ nanowires and nanosheets shows that nanowires exhibited a rapid response time (66 s) but a slow recovery time (204 s) due to a low NO₂ desorption rate from the internal porous structure of nanowires. On the contrary, the recovery time over nanosheets was shorter (126 s), thanks to a lower surface area and a less porous structure [303].

The effect of the structure on the sensor response for undoped WO₃ was also evidenced by Li et al. [304]. An enhancement of the gas sensing properties was obtained by the formation of self-assembled hierarchical hollow spheres. Such structure enhances the sensitivity to NO₂. At 140 °C, a response was recorded for NO₂ concentration of 18 ppb. This was attributed to a high specific surface area (7 m² g⁻¹) and to a wide range of pore size distribution (from 3 to 60 nm) that promotes the gas diffusion.

Table 7. Latest developments in the synthesis of controlled undoped WO₃ structures for high-sensitivity NO₂ gas sensors (papers published in 2019–2020).

WO3 Structure/Morphology	Detection Limit @ Optimal T	Remarks	Ref.
Nanoparticles (~40 nm)	100 ppb @ 25–50 °C	Ionic liquid-assisted synthesis, samples calcined at 500 °C for 2 h.	[305]
Microflowers assembled from nanoplates	125 ppb @ 105 °C	Outer diameters of ~2 µm, composed of nanoplates with the average pore size of 10.9 nm	[306]
Nanowire-assembled WO3 nanomesh	50 ppb @ 160 °C	WO3 nanomesh, assembled from single-crystalline WO3 nanowires	[307]
Nanowires (2 nm)	930 ppb @ 100 °C		[308]
Nanosheets	300 ppb @ 100 °C	Comparison of monoclinic, triclinic and hexagonal WO3 nanosheets; best performances obtained with triclinic WO3	[309]
Mesoporous WO3	60 ppb @ 500 °C	Diameter: 7 nm; specific surface area: 209 m2 g−1, prepared from SBA-15 as the hard template	[310]

summarizes the latest developments in the synthesis of controlled undoped WO₃ structures for high-sensitivity NO₂ gas sensors (papers published in 2019–2020).

Improvement in NO₂ detection can also be obtained by both tuning the tungsten morphology and creating interactions with a support. For instance, Ma et al. showed that tungsten oxide nanorods (diameters of 50–150 nm and lengths of 5–20 mm) supported on porous silicon (PS) had a good response to NO₂ and good recovery characteristics at room temperature. The lowest detected NO₂ concentration was 250 ppb [311]. The same authors also synthesized nanowires with diameters of 20–30 nm and lengths of 1–2 μm which were grown directly on the porous silicon through thermal annealing of tungsten film. The sensor responded well to NO₂ compared to other gases since the NO₂ response was three times higher for 2 ppm NO₂ than for 50 ppm NH₃, 100 ppm ethanol or 100 ppm acetone at 150 °C [312]. Modulations of the potential barriers at both homo- and heterojunctions between porous silicon and tungsten oxide were proposed to be responsible for the good sensibility at a low operating temperature (100 °C) [313]. Excellent gas-sensing behaviors were obtained with n-WO_3−x/n-PS nanocomposite prepared by sputtering tungsten oxide films on a high specific surface area porous silicon substrate [314]. This sensor exhibited an anomalistic p-type semiconducting behavior, which was supposed to improve the amplification effect of the heterojunction between WO₃ nanowires and the PS composites (PS forms p–n heterointerface with n-type WO₃ nanowires). The XPS analysis indicated the presence of large amounts of surface oxygen vacancies that directly contributed to the sensor response. NO₂ concentration as low as 30 ppb was then detected, and Figure 26 illustrates that the NO₂ selectivity vs. other gases was especially interesting. To improve the control of the WO₃ nanowire synthesis on silicon microelectromechanical systems (MEMS), Lee et al. recently developed a fabrication process by stress-induced method, in which the growth position of the WO₃ nanowires can be controlled by patterning of the WO₃ seed film [315]. Finally, note that competition between different gases may influence the recovery time. For instance, WO₃ film demonstrated a longer recovery time towards NO₂ compared to NH₃ [316].

Doped WO₃ sensors: Sensor improvement can also be obtained by WO₃ doping. The addition of precious metals to WO₃ is performed to improve the gas–sensor interaction, as well as the response and recovery times, and decrease the operating temperature. Gold is the most represented metal in the recently reported works, probably because Au-doped WO₃-based sensors exhibit overall enhancement in NO₂ sensing performances (response, detection limit and response/recovery times) [317,318] but also because they are only slightly affected by humidity. In fact, measurement in a humid atmosphere is one drawback of the semiconductor oxide-based sensors, and WO₃ is no exception [319]. Water adsorption leads to the oxidation of the WO₃ lattice because water fills the oxygen vacancies (operando DRIFT experiments), and consequently the resistance of the WO₃-based sensor increases [320]. The selectivity of the sensor is then affected, with lower signals toward NO₂ and higher sensor signals to CO. Sevastyanov et al. showed that water dependency can be virtually avoided by gold addition in the bulk and on the surface of WO₃. The conductivity in pure air of such Au/WO₃:Au material increased only 1.1–1.2 times when the absolute humidity raised from 2 to 16 g m³ (it increases 6–7 times for Pt/SnO₂:Sb films in the same humidity range). Measurements with 0.45–10 ppm NO₂ showed that the Au/WO₃:Au sensor response did not depend on humidity [321]. This behavior toward humidity was confirmed on Au-WO₃ core–shell-structured nanospheres, associated with excellent NO₂ selectivity and long-term stability [322]. The significant performance improvement in NO₂ detection of Au/WO₃ compared to bare WO₃ was attributed by Hang et al. to a combined sensing mechanism: the surface Au nanoparticles dominated chemical sensitization while interbedded Au nanoparticles induced electronic sensitization [323]. An activation of Au/WO₃:Au thin films by laser diode radiation instead of constant heating was proposed by Almaev et al. [324] to reduce the response time to NO₂ by photodesorption. Moreover, holes generated in the near-surface region of WO₃ film by optical transitions were supposed to favor the photodesorption of chemisorbed O₂^- species, leading to the absence of response to reducing gases and change in oxygen concentration.

Results obtained with platinum-doped WO₃ appear possibly contradictory. Chmela et al. reported that platinum addition to WO₃ nanowires (<100 nm) caused detrimental effects, with lower sensitivity and selectivity toward NO₂ compared with the unfunctionalized systems, associated with better sensing properties toward C₂H₅OH [325]. On the opposite, with nitrided WO₃ (WO_xN_y nanofibers), platinum addition enhanced the gas sensing characteristics by (i) lowering the operating temperature; (ii) enhancing the sensor reversibility at 50 °C; and (iii) exhibiting an exceptional selectivity toward NO₂ against interfering molecules such as C₂H₅OH, C₇H₈, CH₄, CO, NH₃ and NO [326].

Silver was also recently reported as a promising dopant. The optimal calcination temperature was found at 500 °C to obtain the larger response, better selectivity, faster response/recovery time and better long-term stability for NO₂ [327]. With 0.5%Ag-WO₃ [328], the sensor shows higher NO₂ sensing response at 200 °C and higher selectivity for NO₂ in the presence of different interfering gases (NH₃, acetone, SO₂, methanol, CO₂, NO). Significant decreases in gas responses were observed with high silver loading (5 and 10 mol%) because the silver crystallite size became too large and hindered their catalytic effects [329].

Comparing Ag-, Pd- and Pt-doped WO₃ nanoplates, Li et al. showed that Pd-WO₃ exhibited the highest response to NO₂ while Ag-WO₃ exhibited the fastest response speed [330]. The promotional effect of palladium was attributed by Liu et al. to the Schottky barrier between Pd and WO₃ [331].

WO₃-based sensors doped with other oxides have attracted little attention in the recent literature. Only a few recent studies report improvements of WO₃-based sensors by doping with oxides such as Sn, Sb and Fe. The main characteristics of these studies are summarized in

Table 8. Oxide-doped WO₃-based sensors for NO₂ detection.

Sensor Material	Detection Limit @ Optimal T	Remarks	Ref.
RuO2/WO3 nanowires		Improved selectivity by enhancement of the electron depletion layer due to the formation of RuO2/WO3 Schottky junctions	[332] (2014)
Fe-doped WO3 hollow nanospheres.	10 ppb @ 120 °C	The light distortion in the WO3 crystal lattice by Fe doping produced interesting defects for gas sensing, with more oxygen vacancies	[333] (2018)
WO3-In2O3 nanocomposites	@ 140 °C	Sol–gel preparation method; also active for CO detection at 240 °C	[334] (2019)
3D hierarchical structured Sb-doped WO3	@ 30 °C	Improvement in NO2 detection attributed to abundant structural defects derived from Sb doping, reduced bandgap and the 3D hierarchical microstructure	[335] (2018)
Sn-doped WO3 nanoplates	5 ppb NO2 @ 100 °C	Optimal loading: 2 wt.% Sn; introduction of Sn ions resulted in shorter response and recovery times, attributed to the increased number of oxygen vacancies on the sensing surface	[336] (2018)

. The observed improvements were generally attributed to an increase in the oxygen vacancies on the sensing surface.

6.2.2. WO₃-Based Sensors for Detection of Gasses Other Than NO₂

WO₃-based sensors were also investigated for the detection of gases other than NO₂. With this aim, only a few studies have dealt with undoped WO₃. Nanostructured WO₃ was evaluated for H₂S [337] or Cl₂ [338] detection, while thin film was found to be sensitive to acetone [339]. However, for this latest gas, a comparative study showed that SnO₂ is the best thin film compared to tungsten oxide or tin-doped tungsten oxide [340].

As described in Section 6.2.1, water adsorption on semiconductor oxides usually affects the measurements. However, this behavior can be used to develop sensors for relative humidity measurement. With this aim, WO₃ is usually associated with TiO₂ in the recently published studies. Zanettia et al. examined the influence of the WO₃ loading (0–10 mol%) on nanopowdered samples prepared by the polymeric precursor method [341]. Best results in humidity measurements (in the 15–85% relative humidity range) were obtained with 2 mol% WO₃. This was attributed to the best compromise between the increase in acid sites and the mean pore size and pore size distribution. Faia et al. studied the influence of V₂O₅ doping for TiO₂-WO₃ sensors [342]. A p- to n-type transition still occurred for the doped sensors with the lower V₂O₅ content, while the sensor with the higher V₂O₅ content exhibited a typical n-type behavior with humidity increase. In addition, the electrical response to the relative humidity (10–100%) depended on the changes in the fabrication route (sintering temperature), which influences the final structure.

Section 6.2.1 also shows that NO₂ detection can be improved by the addition of precious metals to WO₃-based sensors. Nevertheless, these modifications are rather dedicated to the detection of reductant gases such as NH₃, H₂, H₂S and CO.

Palladium–WO₃ samples were prepared by spray pyrolysis by Gobole et al. The optimal palladium loading was found to be 3 wt.% Pd in WO₃ for the detection of NO₂, SO₂ and NH₃ at 100, 200 and 225 °C, respectively. For all of the studied gases, the response and recovery times for concentrations up to 750 ppm were fast, in the ranges of 0.5–1.25 s and 1–6.7 s respectively [343]. Tungsten trioxide nanowires decorated with iridium oxide nanoparticles resulted in remarkable changes in the morphology and defects of tungsten oxide nanowires. Such a sensor was found to be sensitive towards ethanol, NH₃, H₂, H₂S and NO₂ [344]. Tungsten oxide functionalized with gold or platinum nanoparticles was synthesized by Vallejos et al. using a single-step method via aerosol-assisted chemical vapor deposition. The metal additions allow the discrimination of C₂H₅OH, H₂ and CO gases, which are present in proton-exchange fuel cells. Particularly, Pt-functionalized tungsten oxide films allow H₂ detection at 250 °C, whereas nonfunctionalized tungsten oxide films detected low CO concentration (100 ppm) at a lower temperature (150 °C) [345]. Platinum–tungsten oxide can also be used to improve the CO electro-oxidation activity employed in the electrochemical sensor [346]. The interaction between platinum and tungsten oxide was enhanced by a reductive heat treatment, leading to a significant negative shift in CO oxidation potential. A portable CO sensor device built with Pt/WO_x/C exhibited a higher sensitivity, faster response time and good linearity within 50 ppm CO, compared with a usual Pt/C-based sensor.

Xu et al. showed that triethylamine vapor is efficiently detected by Ag/Pt/W₁₈O₄₉ hybrid nanowire sensor, with a detection limit of 71 ppb. The ternary hybrid showed better behavior in detecting triethylamine than the binary Ag/W₁₈O₄₉ and Pt/W₁₈O₄₉ nanowires [347]. This excellent performance was attributed to the dual sensitization mechanism, i.e., a synergy of both electronic and chemical interactions.

Pt-catalyst-loaded tungsten oxide is also a good candidate for optical hydrogen gas sensor applications due to its gasochromic behavior (Pt/WO₃ turns blue in H₂ atmosphere while its electrical conductivity also changes). Yamaguchi et al. investigated the influence of partial pressures of hydrogen and oxygen gases on a Pt/WO₃-based sensor prepared by a sol–gel method [348]. Unfortunately, the optical absorbance of the film exhibited a nonlinear relationship with the H₂ concentration in absence of oxygen. However, the absorbance and electrical conductivity increased proportionally with the H₂ concentration in presence of oxygen, but both parameters strongly depended on the oxygen partial pressure. According to the relationship between the gasochromism and the oxygen concentration, the authors demonstrated that Pt/WO₃ is able to detect hydrogen gas concentrations in a low oxygen gas concentration atmosphere.

6.3. Conclusions

In summary, WO₃ is active in catalytic or photocatalytic oxidation of volatile organic compounds. Catalytic oxidation activity of WO₃ is linked to the presence of W⁶⁺, W⁵⁺ and W⁴⁺ surface species generating reactive oxygen compounds. In supported catalysts, tungsten oxides promote the activity of the V₂O₅ active phase. Many pollutants are reported in studies considering the catalytic abatement of VOCs, such as formaldehyde, hexane, trichloroethane, chlorobenzene, benzene, toluene and xylene. W-based samples are also largely studied for gas sensors, mainly in NO₂ detection. WO₃ is an n-type semiconductor, and its electrical conductivity varies with the oxygen concentration at the oxide surface. In addition, great recent efforts have been made in the development of the synthesis of controlled WO₃ structure/morphology for high-sensitivity gas sensors.

7. Pollutant Remediation in Liquid Phase (Photocatalysis)

The use of WO₃-based catalyst for pollutant remediation in liquid phase mainly concerns the oxidation of organic compounds via photocatalytic processes. Only a few recent studies have dealt with WO₃-based catalysts for non-photocatalytic application, such as H₂O₂ electrogeneration over WO_2.72/Vulcan XC72 gas diffusion electrode [349], hydrolysis of waste bottle PET in supercritical CO₂ assisted by acidic catalysis over WO₃-TiO₂ [350] or heavy metal ion adsorption on inorganic–organic hybrid WO_x–ethylenediamine nanowires [351]. However, this section is focused on the main use of tungsten-based catalysts in liquid phase, namely photocatalysis.

Compared to TiO₂, which responds only to UV light, WO₃ is visible-light-responsive. The bandgap between the valence band (VB) and the conduction band (CB) is 2.6–2.8 eV. However, bare WO₃ is not very active due to the fast recombination of photogenerated electrons and holes. Fortunately, the photocatalytic activity of WO₃ can be improved in five main ways: (i) the control of the WO₃ structure/morphology to obtain nanoporous structures with large surface area and fast pollutant diffusion; (ii) surface hybridization with graphene to obtain large specific surface areas and improvement in the charge transfer; (iii) coupling with other semiconductors (TiO₂) to enhance the photoinduced charge separation efficiency; (iv) doping by noble metal, which works as an electron pool and catalyzes O₂ activation.

WO₃ as a photocatalyst has been the subject of three recent reviews (“WO₃-based photocatalysts: morphology control, activity enhancement and multifunctional applications” by Dong et al. (2017) [352]; “Adsorptive removal and photocatalytic degradation of organic pollutants using metal oxides and their composites: A comprehensive review” by Gusain et al. (2019) [353]; “Tungsten oxide-based visible light-driven photocatalysts: crystal and electronic structures and strategies for photocatalytic efficiency enhancement” by Quan et al. (2020) [354]). Consequently, the section is mainly focused on the more recent studies. Evaluated molecules are mainly dyes, but other organics that are of major interest for water treatment are also considered, such as active molecules (medicinal and plant protection products) or pathogens.

7.1. Undoped WO₃ Photocatalysts

As previously introduced, bare WO₃ is poorly active as a photocatalyst, but structured/hierarchized WO₃ exhibits interesting behaviors. Monoclinic WO₃ phase was often reported as a suitable structure for photocatalysis. Noticeably, Xie et al. showed in 2012 that monoclinic WO₃ exhibits preferentially the suitable high-surface-energy {002} facet [355]. Indeed, the simple thermodecomposition of H₂WO₄ leads first to the cubic phase, but the monoclinic structure is formed with increasing time and/or temperature. After 30 min at 450 °C, the resulting WO₃ oxide shows 90% monoclinic phase [356]. The phase transformation is associated with a decrease in the specific surface area and an increase in oxygen vacancies. The electron transport is then improved, favoring the photocatalytic activity (ibuprofen removal by ozonation under visible-light radiation).

Various preparation methods have been developed to improve the WO₃ photocatalyst. Chen et al. evaluated different fuels (glycine, urea, urea and citric acid) to synthesize nanoscale tungsten oxides (nanoparticles, nanorods and nanoneedles) by the solution combustion synthesis method. Monoclinic WO₃ was obtained with urea, while a mixture of monoclinic W₁₈O₄₉ and WO₃ structures was obtained with glycine when the molar ratio glycine/ammonium paratungstate was 10 or higher. Such W₁₈O₄₉ + WO₃ mixture was also obtained with improved porosity using a mixture of urea and citric acid because the reactants generated more gases and the combustion reaction was more vigorous (production of larger holes). Accordingly, this sample exhibited the best activity in methylene blue degradation [357]. Nanoparticles of monoclinic WO₃ can be also synthesized from Na₂WO₄ and PVP 70,000 (polyvinylpyrrolidone) as surfactant and template [358]. The synthesis was carried out at 180 °C in a Teflon autoclave. The best morphology for the photocatalytic efficiency in degradation of rhodamine B (RhB) under visible light was the block-shaped morphology. The authors proposed that the high photocatalytic efficiency could be attributable to the reducing activity of PVP which led to the formation of oxygen vacancies, beneficial for the capture of photoelectrons and the generation of superoxide radicals.

Monoclinic WO₃ nanoplates (100–170 nm in side length and 30–50 nm in thickness) can be also obtained by a one-step template-free hydrothermal route, from a mixture of aqueous solutions of Na₂WO₄· and HCl maintained at 180 °C for 12 h in an autoclave [359]. The optimal calcination set-up was found to be 600 °C for 2 h. The resulting catalyst allowed the degradation of rhodamine B with an activity 5 times higher than that of a commercial WO₃ powder. It was proposed that the high concentration of -OH species was responsible for the enhancement of photocatalysis.

However, the superiority of the monoclinic phase has been questioned very recently by Zhang et al. [360]. They studied the photocatalyzed degradation of rhodamine B (RhB) over monoclinic (m-WO₃) and hexagonal (h-WO₃) tungsten oxide. Samples were synthesized via acid precipitation process and decomposition of H₂WO₄ at different temperatures in air or N₂ to obtain oxidized or partially reduced oxidation states. After irradiation, the RhB removal efficiency classification was as follows: m-WO₃ (22%) < m-WO_3−x (48%) < h-WO_3−x (76%) ≤ h-WO₃ (80%). It is important to note that the specific surface areas of the hexagonal samples (52–55 m² g⁻¹) were approximately twice those of m-WO₃. It also appeared that the impact of the oxidation state of tungsten was more significant for the monoclinic phase than for the hexagonal one. In fact, the impact of the hexagonal/monoclinic-WO₃ ratio (h/m) in the photocatalytic activity was clearly pointed out by Lu et al. [361]. To tune the h/m ratio, they used a sol–gel preparation method with K₂SO₄ as stabilizing agent and controlled calcination time and temperature. The observed optimal h/m ratio was close to 70/30, for which the activity in rhodamine B degradation was multiplied by a factor of 7.4 compared to pure m-WO₃. This improvement was attributed to the formation of a phase junction between h-WO₃ and m-WO₃, which exhibits high efficiency for the separation and transfer of photoexcited electron–hole pairs (electrochemical impedance measurements).

New innovative designs are also a way to produce efficient WO₃ photocatalysts. For instance, monoclinic nanocuboids, synthesized by hydrothermal treatment of commercial WO₃ with H₂O₂, showed high amounts of coordinative unsaturated W sites suitable for the photocatalytic degradation of methylene blue [362]. Other morphologies such as disk-shaped WO₃ (D-WO₃, monoclinic phase) were recently reported as attractive for acetic acid mineralization (after 0.1% Pt addition). To obtain such samples, aqueous solution of (NH₄)₁₀W₁₂O₄₁· was mixed with HNO₃ at 75 °C. After maturation, the resulting powder was calcined 3 h at various temperatures to vary the physical properties. As expected, the crystallinity of D-WO₃ increased with the calcination temperature while the specific surface area decreased. The highest photocatalytic activity was obtained with the sample calcined at 600 °C [363]. Another recent approach was to design a single tungsten atom oxide photocatalyst, supported on poly(ethylene oxide), as shown in Figure 27a [364]. In this case, the photogenerated electron transfer process is enabled by an electron in the spin-up channel excited from the highest occupied molecular orbital to the lowest unoccupied molecular orbital +1 state, which can only occur with single tungsten atom oxide with W⁵⁺ (Figure 27b). The resulting catalyst exhibited a degradation rate of 0.24 s⁻¹ for dyes (methyl orange, methyl red, dimethyl yellow; mechanism illustrated in Figure 27c for methyl orange), which is two orders of magnitude higher than those of available photocatalysts. Additionally, the reactions between oxygen with photogenerated electrons (e⁻ + O_2ads → O₂⁻•) and between adsorbed water and photogenerated holes (h⁺ + H₂O_ads → H⁺ + OH_ads•) produce reactive oxidizing species such as superoxide anions (O₂⁻•), hydrogen peroxide (H₂O₂) and hydroxyl radicals (OH•). Reactive oxidizing species are responsible for the degradation of aqueous organic compounds, besides the direct decomposition of organic compounds with photogenerated holes.

It is important to note here that dye decolorization and mineralization have to be distinguished. This point is highlighted for instance in the paper of Cheng et al. about the treatment of effluent from oil palm agroindustry with tungsten oxide photocatalysts [365]. With the optimum catalyst loading (at 0.5 g/L), the decolorization reached 96.2% but the mineralization reached only 51.1%.

To improve the efficiency of treatment by photocatalysis, tungsten oxide can be also used as a photoelectrocatalyst, i.e., implemented at the surface of a photoanode. The idea is to promote the photoelectron transfer to dissolved oxygen, which is a drawback of WO₃ photocatalysis. To enhance charge separation, electrochemical and photocatalytic techniques can be used together. Zheng et al. synthetized nanoporous, monoclinic WO₃ photoanodes from a tungsten foil (0.1 mm thickness) that was anodized at 50 V in 0.25 wt.% HF electrolyte and subsequently annealed at 500 °C. It was showed that methyl orange mineralization rate was higher in photoelectrocatalysis (at E = 1 V vs. SCE in NaH₂PO₄ electrolyte) than in photocatalysis or electrocatalysis [366], as illustrated in Figure 28. The high efficiency of the WO₃ photoelectrocatalyst was also demonstrated for the abatement of pathogens such as Escherichia coli [367].

7.2. WO₃–Carbon Photocatalysts

As previously mentioned, the fast electron–hole recombination is a major drawback of WO₃ for photocatalysis application. In addition to the synthesis of structured WO₃ with controlled morphology, another way to improve the photocatalysis efficiency is to associate WO₃ with carbon-based materials that exhibit interesting high specific surface areas. Among the various possible structures of carbon-based materials, graphene appears the most popular. Preparation of WO₃–graphene composite can be performed by different methods. Recent studies reported techniques such as the electrospinning technique to obtain a porous tungsten oxide nanoframework with graphene film [368]; a one-pot synthesis, first mixing sodium tungstate and nitric acid and then adding tetraphenylporphyrinn, graphene and citric acid (final calcination in the 350–550 °C temperature range) [369]; or a method based on pulsed laser ablation in liquid phase: WO₃ nanoparticles suspended in water were mixed with a suspension of monolayer graphene and irradiated for 30 min by the pulsed laser beam of 355 nm. During the laser irradiation, the photoinduced electrons in the WO₃ reduced graphene, and WO₃ nanoparticles were anchored on the graphene sheets [370].

The synthesized film reported in [368] showed the ability to adsorb aromatic molecules, extensive light absorption range, significant light trapping and efficient charge carrier separation properties. Consequently, this sample allowed a high activity in the photodegradation of rhodamine B compared to bare WO₃ and TiO₂ nanomaterials.

The in situ prepared nanocomposites made of tetraphenylporphyrin/WO₃/reduced graphene [369] showed specific surface areas around 450 m²g⁻¹ and spherically shaped nanoparticles of monoclinic WO₃ when the appropriate tetraphenylporphyrin loading and calcination temperature of 350 °C were used. The visible radiation absorption was confirmed with a bandgap energy of 2.14 eV. The photocatalytic degradation rate of 20 ppm Acid Blue 25 was 85% in 3 h at pH = 4 (UV–Vis and TOC analyses). This behavior was attributed to the visible-light absorption properties and to the high separation rate of photogenerated charge carriers. Similar conclusions were obtained with the samples prepared by the pulsed laser ablation evaluated in methyl blue degradation [370]: WO₃-reduced graphene samples showed much better visible-light absorption and less photogenerated charge recombination than pure WO₃.

Other carbon-based materials were also recently developed. Amorphous carbon-coated tungsten oxide was obtained via pyrolysis of hybrid polyoxometalates (hybrid POMs) in nitrogen atmosphere [371]. Compositions, microstructures and concentrations of oxygen vacancies were closely related to the species of organic amines in the hybrid POMs. The presence of defects (oxygen vacancy) was responsible for the improvement in the degradation of dyes (methylene blue, methyl orange and rhodamine B) compared with WO₃ or TiO₂ photocatalysts.

WO₃/carbon nanotube (CNT) nanocomposites were proposed by Isari et al. for the abatement of tetracycline (antibiotic) and other pharmaceutical wastes in water [372]. Samples were synthesized via a sol–gel method (aqueous sodium tungstate + lactic acid + CNT+ HCl; the obtained gel was then transferred in an autoclave for 30 h at 190 °C). The characterization techniques demonstrated the incorporation of CNTs into the WO₃ framework and an efficient reduction in charge carrier recombination rate compared to the corresponding WO₃ catalyst without CNT. The pollutant remediation was performed coupling visible light and ultrasound (US) irradiations (sono-photocatalysis). Ultrasonic irradiation favors HO• formation, mainly via water decomposition. It was found that 60 mg/L tetracycline could be perfectly degraded with the following set-up: WO₃/CNT: 0.7 g L⁻¹; pH = 9; US power: 250 W m⁻²; light intensity: 120 W m⁻²; duration: 60 min. Trapping experiment results verified that HO• radicals and h(+) were the main oxidative species.

7.3. WO₃ Photocatalysts Doped by Precious Metals or Silver

As previously mentioned, the restricted application of tungsten oxide (WO₃) is due to its low conduction band. Tungsten trioxide (WO₃) is an n-type semiconductor with a small bandgap (Eg 2.6–2.8 eV). Consequently, WO₃ has a limited ability to react with electron acceptors such as oxygen and has a high recombination rate of the photogenerated electron–hole pairs. Therefore, great attention has been paid to develop novel visible-light-driven WO₃-based photocatalysts, and special attention has been paid to the crystal facet engineering of WO₃ nanocrystals (Section 7.1). Additionally, doping with noble metal (Section 7.3.1) or silver (Section 7.3.2) is an effective way to enhance the absorption of sunlight and improve the photocatalytic efficiency together with material phase/morphology design to generate oxygen vacancies.

7.3.1. WO₃ Photocatalysts Doped by Precious Metals

Abe et al. [373] proposed the enhancement of photocatalytic properties of WO₃ by loading Pt, which can trap electrons photogenerated from WO₃. The use of Pt as an electron scavenger on WO₃ nanorods was also investigated in [374]. Platinum nanoparticles were loaded on WO₃ nanorods with various mass ratios (0.1, 0.2 and 0.3) via a photoreduction process (PRP). The photocatalytic activity in aerobic oxidation of alcohols reached up to 98% for Pt/WO₃ and 69% for WO₃, while no oxidation was observed in the absence of light. The highest photocatalytic performance was obtained for the mass ratio of 0.2. This enhancement in the photocatalytic activity after platinum loading was attributed to an extended lifetime of the generated electron–hole pairs compared to WO₃ support. Additionally, Gunji et al. [375] reported the importance of controlling the metal deposition site on the photocatalyst surfaces to optimize the use of the photoexcited electrons and holes without the recombination of photogenerated carriers during the photochemical decomposition of organic pollutants. The authors proposed a cocatalyst based on the site-selective deposition of PtPb nanoparticles on oxidation sites and deposition of Pt nanoparticles on reduction sites. The synthesized PtPb/Pt/WO₃ exhibited higher photocatalytic performance for the decomposition of acetic acid under visible-light irradiation (λ > 420 nm) than that observed using the conventional photodeposited Pt/WO₃ or chemically deposited PtPb/WO₃.

Platinum doping was also evaluated on tungsten trioxide for catalyzed oxidation reaction [289] or as an electrode for direct methanol fuel cells (DMFCs) [376]. Pt/WO₃ catalyst layers were prepared by electrosynthesis of WO₃ on a graphite (Gr) support. It was reported that the corresponding Pt/WO₃/Gr electrode has higher catalytic activity for methanol oxidation than a commercial Pt/C catalyst. Interestingly, methanol oxidation is enhanced under visible-light illumination, which is attributed to both a synergy between Pt and WO₃ active sites and the simultaneous occurrence of methanol photooxidation at WO₃ sites.

7.3.2. WO₃ Photocatalysts Doped by Silver

The surface plasmon resonance (SPR) effect greatly contributes to enhancing the visible-light absorbance of catalysts by allowing resonant photons at the metal–dielectric interface to resist the restoring force of positive nuclei [377]. The SPR effect of silver was investigated by Ding et al. [378]. By studying the relationships between the locations of silver nanoparticles on different facets of hexagonal WO₃ nanorods and the photocatalytic performance of the photocatalyst, Ding et al. reported that both the intrinsic nature of charge separation on the {001} facets of WO₃-110 nanorods and the SPR effect contribute to the enhancement of visible-light absorption and the decrease in the recombination of the photogenerated electron−hole pairs. The highest efficiency in the degradation of methyl orange (MO) was obtained with 4.5 wt.% Ag/WO₃ photocatalyst presenting dominant exposed {001} facets compared to {100} or {010} ones. A mechanism was proposed where photons are absorbed under visible-light irradiation and photogenerated electrons and holes are subsequently produced by a facet surface transfer, as summarized in Figure 29.

The crucial role of oxygen defects was investigated by Wei et al. [379]. Tungsten oxide (WO_3−X) is a transition metal oxide that has rich substoichiometric compositions and possesses oxygen defects involved in photon–electron interactions. In combination with silver nanowires (Ag NWs) presenting surface plasmon resonance properties, the incident photon-to-electron conversion efficiency is enhanced together with the methylene blue (MB) photodegradation performance. In fact, the electron concentration in WO_3−X depends mainly on the stoichiometric defect concentration. This means that oxygen deficiencies play a critical role in the decay of localized SPR phenomena.

Morphology and properties of Ag-WO₃ hierarchical materials were also investigated by Capeli et al. [380] for the degradation of rhodamine B (RhB) dye under 467 nm LED light irradiation. Three-dimensional WO₃ catalysts decorated with silver nanoparticles (Ag NPs) were prepared by a one-step hydrothermal method in the absence of surfactant (WO₃•Ag). The amount of Ag NPs is an important factor in the formation of various novel and complex WO₃ 3D hierarchical architectures, from 3D irregular-platelet-like building blocks, which evolve into 3D hexagonal building blocks, to three-dimensional hexagonal-football-like and finally to 3D multibranched spiky ball-like microcrystals. The WO₃•0.20Ag 3D hierarchical structure presented higher photodegradation of RhB dye solution compared to the individual WO₃ 3D material. This result was assigned to an engineering heterojunction between Ag NPs and WO₃ semiconductor, which could enhance the light absorption (more photoexcited electrons) and suppress the photogenerated electron–hole pair recombination. Ag NPs act as an electron reservoir. Under light, photogenerated electrons (e^–) in the valence band (VB) are excited to the conduction band (CB), and electrons from the reservoir on the Ag NPs could be trapped by O₂ molecules to generate superoxide radical (•O₂^–) species. Concurrently, a similar number of holes (h⁺) are generated in the valence band (VB), which are able to react with H₂O or OH^– molecules to generate OH• radicals. Similar results were achieved by Gao et al. [381], who used silver to improve the properties of the WO₃. Ag nanoparticles (Ag NPs) were reported to enhance the production of negative oxygen ions (•O₂^– radicals) under visible light by the prepared Ag/WO₃ catalysts supported on wood. More precisely, Ag NPs redistribute the charge carriers, which could trap the photogenerated electrons and inhibit the recombination of excited electrons and holes (Figure 30).

Promising plasmonic Ag/AgCl photocatalysts have been recently investigated. The SPR effect of silver is responsible for the broadened absorption in the visible-light region, and these properties were expanded to WO₃. For instance, Ma et al. [382] prepared Ag–AgCl/WO₃ hollow spheres with a flowerlike structure that had superior visible photocatalytic activity because of their unique morphology. Adhikari et al. [383] also observed a photocatalytic activity enhancement of Ag/AgCl/WO₃ powder prepared by a microwave-assisted hydrothermal method. However, most WO₃/Ag/AgCl photocatalysts reported were in powder form, thus having limited use in practical applications because tedious regeneration is required. Consequently, Fang et al. [384] prepared WO₃/Ag/AgCl films on a conventional glass substrate to reduce the regeneration cost. The WO₃ film was prepared on a glass substrate by calcination of spin-coated W precursor. Ag/AgCl particles were then deposited on WO₃ film by an impregnation–precipitation–photoreduction method. Excellent photocatalytic performances were obtained in the degradation of methyl orange (MO) and rhodamine B (RhB) under visible light. It is proposed that Ag and AgCl greatly promoted the separation of photogenerated electron–hole pairs and improved the charge transfer efficiency of WO₃. Based on the photoelectrochemical test and radical trapping measurement, a Z-scheme mechanism for WO₃/Ag/AgCl is proposed where •O₂^– and h⁺ play the major roles in photodegradation, while the effect of OH• could be neglected (Figure 31a). Consequently, Z-scheme composite photocatalysts have to be considered as an effective way to enhance the photocatalytic performance of catalysts. Li et al. [385] proposed plasmonic Ag/Ag₂WO₄/WO₃ Z-scheme visible-light composite photocatalyst for the degradation of rhodamine B, methylene blue and methyl orange. A possible Z-scheme mechanism of the ternary composite was proposed under visible light where Ag particles produce SPR effect but also work as the charge transmission bridge (Figure 31b).

Sahoo et al. [386] designed a Z-scheme WO_3−X-Ag-ZnCr layered double hydroxide (LDH) photocatalyst for tetracycline degradation, based on the SPR effect of metallic Ag as redox electron mediator. The defects created by surface oxygen vacancy in WO_3−X and the existence of Ag as electron transfer conductor facilitate the charge pair separation efficiency and enhance the photocatalytic activity.

7.4. WO₃–TiO₂-Based Photocatalysts

Because of the poor activity of WO₃ due to its rapid electron–hole recombination, coupling with other semiconductors to use the sunlight as a free light source is an attractive research area. Concomitantly, titanium dioxide (TiO₂) is an interesting material for applications in the purification of air and water by removal of recalcitrant organic and inorganic pollutants such as synthetic dyes, phenols and chlorophenols, volatile organic compounds, detergents, solvents, heavy metals or pharmaceutical antibiotics.

Unfortunately, due to the size of TiO₂ bandgap (Eg 3.0–3.2 eV), bare TiO₂ only works in conjunction with irradiation of limited wavelength (λ_irr < 385 nm) corresponding to ultraviolet (UV) light that accounts for only 6.8% of the solar spectrum, thus limiting its photocatalytic activity. The visible (Vis) range (400–760 nm) accounts for about 38.9%, and infrared radiation (IR, 760–3000 nm) makes up most of the remaining 54.3% [387]. In addition to the ability of materials to absorb solar spectrum, the effectiveness of photocatalytic systems also depends on the sample capacity to separately collect photogenerated electrons and holes. The high electron–hole recombination rate limits the photocatalytic performance of bare TiO₂. Consequently, several approaches were devoted to improving the photocatalytic performance of titania, especially by coupling with other semiconductors. The objective is to increase the separation of charge carriers by doping titania with metal oxide presenting both conducting band (CB) and valance band (VB) of higher or equal energy compared to TiO₂. Among the many possibilities, the addition of crystalline tungsten oxide (WO₃) has interesting advantages. Its shorter bandgap (Eg 2.6–2.8 eV) compared to TiO₂ (Eg 3.0–3.2 eV) requires longer wavelengths for the excitation and extends the photoresponse of TiO₂ to the visible light region. The coupling with WO₃ also induces energy levels enabling the electrons photogenerated in the conducting band of TiO₂ to transfer into the CB of WO₃. Consequently, the photopromoted holes can diffuse from the CB of WO₃ into the valence band of TiO₂. The surface acidity of the TiO₂ catalyst can also be increased by adding WO₃, which facilitates the adsorption of OH⁻or H₂O molecules and targeted molecules on TiO₂ and thereby improves the photocatalytic activity of TiO₂. WO₃–TiO₂ combination also induces dark activity due to the energy storage ability of tungsten oxide.

WO₃–TiO₂-based systems exist as (i) tungsten oxide doping of supported or hybrid materials (Section 7.4.1) and (ii) metal doping of WO₃–TiO₂ materials (Section 7.4.2).

7.4.1. WO₃–TiO₂-Based Systems as Supported and Composite Photocatalysts

The photocatalytic activity of a WO₃–TiO₂ system is largely determined by its structure, which is significantly influenced by the preparation method of the catalyst. Consequently, both WO₃ supported (WO₃/TiO₂) and composite (TiO₂/WO₃) materials are described.

Supported photocatalysts. Dyes such as methylene blue, rhodamine B or methyl orange are often chosen as model pollutants, but the degradation of several other contaminants has also been evaluated. For instance, Gao et al. [388] obtained WO₃/TiO₂ material with high surface area and unique morphology by synthesizing heterostructured photocatalysts from wood fibers through a two-step hydrothermal method and a calcination process. The wood fibers acted as carbon substrates, allowing the recombination probability of photoexcited charge carriers to be reduced, and also increased the transport of charges. It results in high performance of the WO₃/TiO₂–wood fibers as a UV-light or a visible-light photocatalyst for degradation of rhodamine B, methylene blue, methyl orange or phenol. Ding et al. [389] also observed a photoinduced electron–hole separation effect over tungsten oxide (WO₃)/TiO₂ core–shell nanowires for the degradation of rhodamine B. The degradation of RhB was also undertaken over WO₃/TiO₂ and MoO₃/TiO₂ composites [390]. The reduction of the electron–hole recombination rate by coupling TiO₂ with tungsten oxide or molybdenum oxide led to high photocatalytic activity. The 5 wt.% WO₃/TiO₂ composite was the more efficient sample due to the more efficient separation of charge carriers. The electron migration from the conduction band (CB) of TiO₂ to the WO₃ CB can be offset by O₂ and produces superoxide radical (O₂•) and hydrogen peroxide (H₂O₂) and finally hydroxyl radical (OH•) to decompose RhB (Figure 32).

Yang et al. [391] synthesized porous WO₃/TiO₂ hollow microspheres by a spray drying method. The authors observed that the tungsten oxides mainly existed in a highly dispersed amorphous form on anatase when the loading amount of tungsten oxide was below 3 mol%. The improved photocatalytic activity in methylene blue and phenol photodegradation under UV-light irradiation over the WO₃/TiO₂ catalyst mainly arises from the enhanced charge separation efficiency provided by the acidity induced by WO₃ addition, rather than the improved light absorbance by highly dispersed amorphous tungsten oxides.

WO₃/TiO₂ materials, designed as nanocomposites, mixed oxides or supported photocatalysts, were also used to degrade malathion pesticide [392], imazapyr [393], sulfamethoxazole [394] or formic acid [395]. For instance, 2%WO₃/TiO₂ prepared by sol–gel method showed excellent photocatalytic performance, achieving complete malathion degradation after 2 h [392]. Synthesis of mesoporous WO₃–TiO₂ nanocomposites presented interesting behavior for imazapyr herbicide degradation under visible light and UV illumination. The overall photocatalytic efficiency of the 3% WO₃–TiO₂ nanocomposite was 3.5 higher than for mesoporous TiO₂ [393]. The WO₃–TiO₂ nanocomposite showed both monoclinic and triclinic WO₃ structures. Tungsten-promoted composite catalysts were also envisaged by Ioannidou et al. [394]. Experiments showed that the 4% W-TiO₂ catalyst calcined at 700 °C was the most active for sulfamethoxazole degradation (350 µg/L) under simulated solar irradiation, enhancing the rate of pristine TiO₂ by 50%. Tungsten is believed to act as a trap of electrons, thus reducing the rate of electron–hole recombination and, consequently, increasing degradation rates. The addition of electron acceptors, such as hydrogen peroxide and sodium persulfate, in the reaction mixture also improved the catalyst activity. Photoactivity of WO₃–TiO₂ mixed oxides prepared by a sol–gel method was tested under UV–visible irradiation in both the mineralization of formic acid in aqueous suspension and the gas phase oxidation of acetaldehyde [395]. The catalyst containing a 3% W/Ti molar ratio gave the best results, due to the formation of an intimately mixed oxide resulting in a better charge separation due to the migration of photoproduced holes from WO₃ to TiO₂. WO₃ supported samples were also synthesized using liquid phase plasma process for degradation of diethyl phthalate [396]. The best degradation performance was observed for bare TiO₂ photocatalyst under UV light source, but modified TiO₂ photocatalysts showed a 1.7–6.2 times higher degradation rate under blue light.

Composite photocatalysts. Composite (or hybrid) TiO₂/WO₃ materials are another strategy to obtain an enhancement in the photon absorption and a dark activity due to both the storage of photogenerated electrons and a redox process. Khan et al. [397] prepared microsized hybrid TiO₂/WO₃ samples (TWx; x is 0.025, 0.5, 0.075, 0.1 molar ratio of W precursor) by sol–gel and crash precipitation methods followed by spray drying. Evaluated in the photocatalytic degradation of methylene blue (MB), the hybrid TW_0.075 was reported as the most active material, with 90% of dye degradation reached in 100 min with dark runs. It was reported that TiO₂ and WO₃ acted in synergy to increase the lifetime of electron–hole pairs and to decrease the recombination rate by electron diffusion from CB of TiO₂ to WO₃. The authors proposed a mechanism for the transfer pathways of the photogenerated charge carriers in the hybrid powders (TWx) (Figure 33) where both hydroxyl radicals (OH•) and superoxide anions (O₂^–) play a major role.

TiO₂/WO₃ systems (TW) were also studied by Rimaldi et al. for ethanol and tetracycline (TC) degradation [398]. Analyses supported the parallel occurrence of several TC degradation pathways in the case of TW samples in which OH• radicals are involved, as previously mentioned in [397] for MB conversion. TiO₂-WO₃ composites were also evaluated for bisphenol A (BPA) oxidation under simulated solar light [399]. A core–shell system composed of a layer of TiO₂ nanoparticles on the surface of WO₃ nanowires was also evaluated as an effective photocatalyst in the degradation of RhB [389]. Again, the photoinduced electron–hole separation effect between WO₃ and TiO₂ contributed to the improvement of the photocatalytic activity.

WO₃-TiO₂ photocatalytic systems can be also associated with other processes to obtain overall improvements. With the aim to associate photocatalysis with electrochemistry, WO₃ was associated with a TiO₂ nanotube array to obtain a photoanode designed to improve photoelectrocatalytic (PEC) performance for the degradation of methylene blue [400]. A synergistic vacancy-induced self-doping effect and localized surface plasmon resonance (LSPR) effect were observed, confirming the importance of oxygen vacancies in improving PEC performance.

WO₃–TiO₂-based catalysts were also evaluated for photocatalytic-assisted ozonation of pollutants. A 4% WO₃/titanate nanotube composite catalyst allowed the complete removal of emerging contaminants such as caffeine, metoprolol and ibuprofen in municipal wastewater in less than 40 min with TOC removal up to 64% after 2 h. Compared to bare TiO₂, WO₃ addition induced a higher activity under visible-light radiation and an increase in the adsorption capacity of organic compounds [401].

7.4.2. Metal-Doped WO₃–TiO₂-Doped Systems

To decrease the fast electron–hole recombination and improve the interfacial charge transfer process of WO₃–TiO₂ materials, the addition of noble metals or heteropolyoxometallates (POMs) has been successfully explored.

Noble metals act as a sink of conduction band photoinduced electrons, avoiding their recombination and enhancing the photocatalytic process. Thus, TiO₂–WO₃–Au systems were evaluated by several authors. X-ray diffraction (XRD) data and Raman spectroscopy measurements usually indicated that TiO₂ is mainly in anatase structure, while the WO₃ can be amorphous or crystallized. Gold nanoparticles improve the material’s response in the visible domain [402]. Kovács et al. [403,404] studied the structural peculiarities of TiO₂/WO₃/Au materials. It was shown that each minor structural change in bulk or surface has a significant impact on the photocatalytic activity and intermediate formation dynamics, using phenol as a model pollutant. However, the gold deposition mode was found crucial over structural peculiarities, especially for W species. The resulting sample presented three WO₃ forms in different proportions: amorphous, crystalline and doped. These forms manifested in different species (W⁶⁺, W⁴⁺, W_defects), which critically influenced the activity and the intermediate formation during the photodegradation. Consequently, gold-containing samples resulted in a modification of structural particularities, but the correlation between structural features and PCA or intermediate formation was strongly dependant on the synthesis pathway [404].

Platinum addition is reported to enhance the electron storage ability, the visible activity and the electron–hole separation of photocatalysts by the creation of Schottky-type junctions, thus facilitating the charge transfer at the catalyst–environment interface [405]. By varying the platinum amount from 0.2 to 1 wt.% in Pt/TiO₂-WO₃ materials prepared through the spray drying method, the optimum amount for the degradation of methylene blue (MB) was found at 0.8 wt.% Pt. Hydrogen peroxide (H₂O₂) and hydroxyl radicals (OH•) were found as the main species responsible for the aqueous degradation of the model pollutant. TiO₂ maintained its tetragonal anatase phase and WO₃ maintained its monoclinic structure. The corresponding Pt_0.8TW sample presented the lowest electron–hole recombination rate (photoluminescence measurements) and the highest amount of surface hydroxyl groups (FTIR experiments).

The use of heteropolyoxometallates (POMs) is another interesting way to decrease the fast electron–hole recombination on TiO₂. POMs act as effective traps of photoinduced electrons, leading to visible-light-absorbing materials with high photocatalytic activity. Tungstosilicic acid (TSA; H₄SiW₁₂O₄₀) was successfully used by Rengifo-Herrera et al. [406] to obtain the formation of a surface complex between the Keggin structure of TSA and the TiO₂ surface, or the formation of Keggin–TSA/TiO₂ composites. It resulted in visible-light absorption of modified catalysts. The photocatalytic activity in the degradation of 4-chlorophenol depends on the TSA amount. FT-IR and FT-Raman characterizations indicated that the main heteropolyoxometallate species presented in the composites was the [SiW₁₂O₄₀]⁴⁻ anion, which exhibited a strong interaction with the TiO₂ surface.

7.5. WO₃ Associated with Transition Metal Oxides Other Than TiO₂ (Zn, Fe, Sn, Mn, Ni, Mo, Co, Nb)

In view of photocatalytic activity enhancement of WO₃ catalysts, doped materials have attracted much attention, and the present section focuses on the modification of tungsten(VI) oxide by transition metal oxides (MOs). In fact, by coupling WO₃ with another semiconductor, an improvement of charge separation in photoinduced charge transfer processes of n-type semiconductor is expected, leading to a drop in electron–hole (e⁻/h⁺) pair recombination and an increase in photocatalytic activities.

7.5.1. ZnO–WO₃

Among the multitude of oxides considered, zinc receives great attention owing to its high light sensitivity and wide bandgap energy (3.37 eV). However, the photocatalytic activity of zinc oxide is limited to the ultraviolet light range of the solar energy spectrum, representing only 5–7% of the total sunlight energy. Another limitation is the rapid recombination of the electron–hole (e⁻/h⁺) cavity. To overcome these issues, surface modification of the structures of the ZnO nanoparticles has been proposed. The aim is to alter the bandgap energy, thus shifting the absorption band to the visible region in semiconductor systems. Doping ZnO by WO₃ enables the electrons produced from ZnO to be trapped, thus preventing the fast recombination of electron–hole pairs and increasing the photocatalytic activity. WO₃ has the advantage of narrow bandgap energy (2.2–2.8 eV), which could activate the zinc oxide. Such ZnO-WO₃ materials can act as good candidates for the degradation of Direct Blue 15 (DB15) or methylene blue (MB) dye, organophosphorus pesticides such as diazinon, diclofenac and bacterial or fungal strains.

Zinc oxide nanoparticles doped with 0.5% to 2% WO₃ were synthesized by hydrothermal method to obtain samples where W was added into ZnO structure [407]. Photocatalytic activity for diazinon removal was reported to reach 99% for 2% WO₃-doped ZnO catalyst after 180 min of contact time. This result was attributed to a reduction in the network constant. Similar tungsten oxide doped ZnO nanoparticles were evaluated in the photocatalytic degradation of Direct Blue 15 [408]. Numerous parameters, namely pH, light intensity, dopant percentage, dye concentration and contact time, affect the process. It appeared that acidic pH resulted in higher efficiencies of the photocatalytic process. In addition, increasing the concentration of the WO₃ dopant percentage from 1 to 5% w/v increased the degradation rate from 30.7 to 73.1%. Tungsten acted as a recombinant mediator of interfacial charge transfer, which resulted in a change in the photocatalytic efficiency of the doped ZnO. Sajjad et al. [409] revealed that the formation of Zn-O-W linkage was responsible for a redshift of the absorption peak to the visible region, resulting in a lower bandgap and a promotion of the separation of photogenerated carriers. In fact, the proposed energy state configuration showed dissimilar locations of conduction and valence bands of both semiconductor oxides in ZnO/WO₃ composites. Due to a narrowing of the bandgap, the charge separation is facilitated and the rate of recombination is significantly reduced, resulting in improved photodegradation efficiency. In addition, the transfer of charge carriers due to the narrow bandgap can be confirmed by OH• radical measurement as proposed by Mugunthan et al. [410] for the degradation of diclofenac with ZnO-WO₃ mixed oxide catalysts. Authors observed higher hydroxyl radical (OH•) generation when coupling ZnO and WO₃ compared to bare ZnO. The benefit of the coupled system is mainly due to optimal loading of WO₃ since good dispersion of WO₃ in coupled oxides promotes the formation of heterojunction structures between the two coupled oxides. The optimal molar ratio of Zn:W = 10:1 was found for the degradation of diclofenac.

Photocatalytic activity of Zn-doped WO₃ (Zn-d-WO₃) nanoparticles for antifungal and antibacterial degradation was studied by Arshad et al. [411]. Zn loading from 0.1 to 0.3% had a significant impact on both antibacterial/antifungal photocatalytic degradation and structural/morphological properties. Characterizations showed that the Zn-d-WO₃ structure was triclinic, the particle shape was a mix of spheres and rods and the particle size was decreased as the Zn concentration increased from 0.1 to 0.3%. Photocatalytic activity of Zn-WO₃ was evaluated in methylene blue (MB) dye degradation. Again, Zn loading impacted the activity. The maximum dye degradation rate was achieved with Zn(0.3%)-WO₃ sample, at 78% and 92% with 120 min of contact time for visible and UV irradiation, respectively. Again, it was proposed that after the generation of electron–hole pairs (e⁻/h⁺), the electronic vacancies or holes in the valence band are promoted due to the migration of electrons from the valence band to the conduction band. As a consequence, electron–hole pairs generate hydroxyl radicals (OH•) and superoxide ions (•O₂^–) that act as powerful oxidizing agents to mineralize the dye into CO₂ and H₂O.

Novel Pd/ZnWO₄ nanocomposite materials for photocatalytic degradation of atrazine were evaluated by Al-Amshany et al. [412]. Photocatalytic activity of ZnWO₄ was highly affected by the insertion of Pd into the composition. Palladium was reported to replace W in the ZnWO₄ lattice leading to a sample presenting low bandgap, low electron–hole pair (e⁻/h⁺) recombination rate and high BET surface area. Consequently, high photocatalytic activity for atrazine degradation was obtained for 1.65 wt.% Pd/ZnWO₄ nanocomposite, with 100% of degradation after 60 min of reaction time.

7.5.2. MO–WO₃ (MO: Metal Oxides = Fe, Sn, Mn, Ni, Mo, Co)

Other transition metal oxides such as Fe, Sn, Mn, Ni, Mo and Co were evaluated in the past few years to improve the photocatalytic activity of WO₃. Metal oxides act as electron sinks to capture the photoinduced electrons through their low energy trapping sites; they then reduce the electron–hole recombination and enhance charge separation.

Fe₂O₃ is a visible-light-driven material with a bandgap energy of 2.0–2.3 eV and a CB of 0.46–0.77 eV. Wang et al. [413] proposed a new active FeWO₄/Fe₂O₃ di-modified WO₃ for the degradation of methylene blue (MB), toluidine blue (TB), azure I (AI) and acridine orange (AO) under visible-light irradiation. Beneficially, Fe²⁺, which is classically involved in the Fenton reaction favors the generation of OH• radicals and subsequent organic dye degradation. Over FeWO₄/Fe₂O₃-modified WO₃ sample, FeWO₄ and Fe₂O₃ acted as electron traps. Consequently, a valence decrease of Fe³⁺ to Fe²⁺ led to the capture of photoexcited electrons of WO₃ and enabled the Fenton reaction in the presence of trace H₂O₂. Consequently, di-modified WO₃ samples showed high photocatalytic activity compared to bare WO₃. This assisted Fenton process was also described by Mwangl Ngigi et al. [414] for photocatalytic degradation of Methylparaben (MeP) over Fe-doped WO₃ nanoparticles in presence of H₂O₂. The best formulation was 5 wt.% Fe-WO_3, with 50.8% MeP degradation in the presence of H₂O₂. A hydroxyl radical mechanism was advanced to explain the efficiency and enhancement of iron-doped WO₃ on visible-light activity of the organic pollutant removal.

SnO₂ is also used for bandgap tuning of nanomaterials to transform UV-light-active catalysts into visible-light-driven catalysts. Nevertheless, tin dioxide has a wider bandgap of 3.6 eV (UV-light-driven catalyst) with a high recombination rate of its photogenerated electron–hole pairs, which limits its use. These drawbacks can be overcome by tungsten doping to serve as a photogenerated charge trap, thus decreasing the recombination rate of e⁻/h⁺ pairs by the reduction of the bandgap. In this way, Ullah et al. [415] reported that tungsten-doped SnO₂ (W@SnO₂) nanospheres can be a preferred choice for visible light photocatalytic degradation compared to standard TiO₂ (Degussa-P25). With the increase in the W content from 0 to 6 _mol%, the bandgap of SnO₂ was tuned from 3.6 to 2.8 eV, and the degradation of crystal violet dye reached 100% after 40 min of contact time.

The formation of a nanocomposite of WO₃ with manganese oxides is also a way to reduce the bandgap of tungsten oxide. The constitution of composite with Mn, indigo dye and reduced graphene oxide (RGO) was investigated by Ahmad et al. [416] as visible light photocatalyst for methylene blue (MB) dye removal. A synergetic association between WO₃ and RGO was observed which increased the rate of charge transfer and limited the recombination of electron–hole pairs in the nanosized composite. As a consequence, the so-called Mn/indigo/RGO/WO₃ presented improved photocatalytic performance compared to Mn/indigo/WO₃.

Well-crystalline NiO-WO₃ nanoparticles were synthesized by Rosaline et al. [417] for the photocatalytic degradation of eosin yellow (EY) dye. A photocatalytic degradation rate of 95% was obtained after 180 min under visible-light irradiation. The authors proposed that photogenerated electrons of NiO possibly migrate into the conduction band of WO₃. Simultaneously, the holes move in the opposite direction, which greatly decreases the recombination of electron–hole pairs.

Mo-doped WO₃ nanowires for adsorbing methylene blue (MB) dye from wastewater were evaluated by Silveira et al. [418]. Samples presented hexagonal structure (h-WO₃), and the Mo doping led to a change in morphology of h-WO₃ nanowires from large bundles to narrow bundles when Mo loading increased from 4 to 15 _at%. In fact, Mo⁶⁺ and W⁶⁺ have similar ionic radii and electron structures, which enables Mo to directly incorporate the crystalline lattice of WO₃ to form a Mo_xW_1−xO₃ structure. As a consequence, the bandgap of WO₃ can be reduced by substituting W by Mo in order to improve the photocatalytic activity with visible light, as demonstrated in [419]. In addition, Mo-doped WO₃ samples also exhibited a considerable improvement in MB adsorption capacity [418].

The reduction in electron and hole recombination of WO₃ can also occur with nanocomposites of cobalt(II, III) oxide and tungsten(VI) oxide. The synthesis of p–n heterojunction catalysts, such as Co₃O₄/WO₃, was claimed to enhance the photoactivity under UV- or visible-light radiation [420]. The strong reduction in photoexcited electrons and holes resulted in the enhancement in diclofenac sodium degradation. The highest degradation efficiency was obtained for 0.02 M of cobalt acetate (CW2), with 98.7% achieved for the degradation of 15 ppm diclofenac sodium at pH 10.7. Higher generation of hydroxyl radicals (OH•) associated with the enhancement of effective charge separation was still advanced to explain the high photocatalytic activity of the CW2 sample. Figure 34 shows the charge transfer proposed by Malefane et al. [420] and the p–n heterojunction formed to minimize photoexcited electron and hole recombination in CW2 nanocomposite.

7.6. WO₃ Associated with Other Systems (Post-Transition Metal Oxides and Metalloids, Rare Earths, Nitrides)

7.6.1. Post-Transition Metal or Metalloid–WO₃ Systems

Among the post-transition metals, bismuth and gallium were evaluated in recently published works.

Bi₂WO₆/perylene diimide (PDI) composite photocatalyst was prepared and evaluated in phenol oxidation (and water decomposition to produce oxygen) under visible light [421]. Compared to the self-assembled perylene diimide and Bi₂WO₆, the composite showed an enhancement in the phenol degradation because the positions of conduction and valence bands between Bi₂WO₆ and perylene diimide favored the separation of the photogenerated carriers. The Bi₂WO₆/PDI composite exhibited an n–n-type heterojunction.

BiFeWO₆/WO₃ nanocomposites were studied by Priya et al. [422]. Various BiFeWO₆ loadings from 1 to 3 wt.% deposited on WO₃ nanorods were prepared by simple coprecipitation and hydrothermal treatment. Pure BiFeWO₆ bandgap was measured at 2.0 eV, and the best photodegradation of rhodamine B under visible-light illumination was obtained with 1% BiFeWO₆/WO₃ This sample exhibited a bandgap of 2.3 eV, while the bandgap increased to 2.9 eV for 3% BiFeWO₆/WO₃.

The influence of the partial substitution of gallium by tungsten in a Ga₂Zr₂O₇ fluorite-type nanosized material was recently examined in crystal violet degradation [423]. Ga₂Zr_2−xW_xO₇ samples with x = 0, 0.05, 0.1, 0.15 and 0.2 were prepared using the citrate technique. W_0.15 substitution (optimal loading) allowed the shift of the absorption range to the visible-light range by decreasing the bandgap from 4.95 eV for Ga₂Zr₂O₇ to 1.7 eV for Ga₂Zr_0.85W_0.15O₇. The crystal violet photocatalytic degradation rate was then multiplied by 20 (optimum operating conditions: pH = 9, 1 g L⁻¹ catalyst, reaction duration of 300 min). Complete degradation was obtained with the addition of 25 mmol L⁻¹ hydrogen peroxide. O₂ center dot and holes were found to have a more important role in Ga₂Zr_2−xW_xO7 systems compared with HO center dot.

The excellent semiconductor tungsten selenide WSe₂ was recently used as quantum dots (particle size: 7–8 nm) and associated with nitrogen-doped graphene oxide to prepare a composite photocatalyst by the mechanical stripping method [424]. The degradation of methylene blue under visible light with the composite (optimal WSe₂ loading: 4.7 wt.%) was better than WSe₂ and nitrogen-doped graphene oxide (N-GO) alone. The photocatalytic degradation rate of 50 mL at 1 × 10^–4 mol·L⁻¹ dye with only 0.01 g catalyst reached 93% after 1 h light exposure (300 W Xe lamp with a cutoff wavelength of 420 nm; reaction was monitored by a total organic carbon analysis).

7.6.2. Rare Earth–WO₃ Systems

Cerium is a very popular component in redox catalysis thanks to its ability to vary between Ce^III and Ce^IV. Impregnation of 1 to 25 wt.% of Ce³⁺ ions onto WO₃ led to a better absorption cross-section and a redshift in the band edges [425]. The decrease in the photoluminescence emission intensity and the suppression of the Raman active bands of WO₃ indicated the recombination quenching ability of Ce surface states. For low cerium loading (≤5%), cerium was mainly found as Ce^III, while Ce^IV was predominant for higher loadings. An enhancement in the charge retention ability was observed with the Ce loading increase. In comparison to pure WO₃, Ce-doped catalysts exhibited superior activity for 2-nitrophenol and 2-chlorophenol removal in natural sunlight exposure. For low cerium loadings, the surface oxygen bonded with Ce^III serves as electron trapping and transfer centers, while the synergic composite mechanism is the dominating mode for high Ce loading [425].

Tahir et al. evaluated various other rare earths (REs) by doping WO₃ with La, Gd and Er (RE molar contents of 2, 4 and 6%). Samples were obtained by coprecipitation at 80 °C in acidified aqueous solution. Photocatalytic activity was measured in degradation of dyes (methyl orange, methyl blue, crystal violet), antibiotic (tetracycline) and antimetabolite (methotrexate) and compared to a 10% carbon nanodots–WO₃ reference sample [426]. La, Gd and Er did not enter into the crystal lattice of WO₃ (monoclinic and hexagonal phases) but were positioned in the interstitial sites, with La-O-W, Gd-O-W and Er-O-W bonds. The photocatalytic activity order was as follows: 2%Gd-WO₃ > carbon nanodots–WO₃ > 4%Er-W > 4%La-W > undoped WO₃. The best photocatalytic performance achieved with 2% Gd-WO₃ could be attributed to its higher surface area, the inhibition of the charge-carrier recombination, the high surface hydroxyl content and the extended visible-light absorption region.

7.6.3. Nitride–WO₃ Systems

The graphite-like carbon nitride (g-C₃N₄) is a metal-free polymer that is an n-type semiconductor. Its structural and physiochemical behaviors give it interesting properties in electrical and optical fields. The g-C₃N₄-based nanostructures are emerging materials for energy and environmental photocatalytic applications (e.g., photocatalytic water reduction and oxidation, degradation of pollutants, carbon dioxide reduction) [427].

Xiao et al. prepared Z-scheme WO₃/g-C₃N₄ composite hollow microspheres by controlled in situ hydrolysis and a polymerization process (Na₂WO₄ in deionized water with the addition of dicyandiamide C₂H₄N₄ and glucose; the homogenized solution was then transferred into an autoclave and heated 200 °C; for 20 h; after centrifugation and washing, the product was finally annealed at 550 °C; for 3 h in air (Figure 35)) [428]. The shell of the hollow microspheres consisted of well-distributed WO₃ and g-C₃N₄ nanoparticles, favoring heterojunctions with numerous interfaces and highly exposed oxidation–reduction active sites. Various contents of g-C₃N₄ were evaluated, and the best results were obtained with a molar loading of 4% dicyandiamide with respect to W. The lifetime of the charge carriers reached 2.23 ns, which is obviously prolonged compared to WO₃. This optimized Z-scheme system maintained the redox potential of the components. Combined with the long lifetime of holes and electrons, this photocatalyst showed enhanced degradation rates towards tetracycline hydrochloride (antibiotic agent) and ceftiofur sodium (antibacterial agent).

g-C₃N₄/WO₃ materials can be also shaped as thin films for water purification [429]. A bilayered structure has to be formed to ensure an enhanced charge carriers’ separation. Trapping experiments performed under UV illumination revealed that the activation of the photocatalytic oxidation of methylene blue (MB) came from both holes and superoxide radicals.

Other nitrides such as tantalum nitride (Ta₃N₅) were also recently evaluated in dye degradation. Ta₃N₅ is one of the few visible-light-absorbing photocatalysts capable of overall water splitting (OWS). Ta₃N₅/W₁₈O₄₉ nanocomposite fibers were prepared by the solvothermal technique to obtain a catalyst made of orthorhombic Ta₃N₅ nanoparticles and monoclinic W₁₈O₄₉ nanowires. This sample showed a doubled degradation rate of rhodamine B under white light compared to Ta₃N₅ nanoparticles alone [430]. In fact, W₁₈O₄₉ allowed dye adsorption before illumination. Using tert-butanol and p-benzoquinone as scavenger agents, it was demonstrated both hydroxyl and superoxide radicals contributed to the cleavage of rhodamine B. The superoxide species were involved in N-deethylation of the dye. The generation of both radical species on W₁₈O₄₉ was due to the electronic transfer between W₁₈O₄₉ and Ta₃N₅. Taking into account the recycling ability of the catalyst and the simplicity of the synthesis process, the authors propose that this catalyst is a good candidate for the scale-up.

7.6.4. Photoreduction over WO₃ Based Systems

Photocatalytic water treatment is mainly focused on the oxidation of organic pollutants. However, due to its high toxicity, Cr^VI species reduction into Cr^III is also a major concern. g-C₃N₄/WO₃ thin-film materials presented in the previous section also exhibited Cr⁶⁺ elimination behavior under UV illumination [429]. The Cr⁶⁺ reduction rate was 6% and 93% after 120 min illumination for WO₃ and g-C₃N₄/WO₃, respectively. In opposition with photocatalytic oxidation of methylene blue for which active species came from both holes and superoxide radicals, the main active species for chromium reduction are electrons.

For the same chromium reduction, Thwala et al. evaluated the doping of WO₃ by 1, 3 and 5 _mol% of magnesium (one-pot preparation method) [431]. A slight increase (0.0069 nm) of d spacing was observed by HRTEM imagery after Mg addition, according to a decrease in the bandgap energy and a shift in the edge positions, also confirmed by XPS analysis (0.08 eV). The rate of recombination was greatly reduced upon doping (photoluminescence analysis). DFT calculations supported the recombination reduction rate due to the introduction of the Mg orbital. With 500 mL at 9.5 ppm Cr^VI maintained at pH = 1 and a catalyst weight of 0.1 g, 97% Cr^VI reduction rate was reached with 3 _mol% Mg-WO₃ after one hour under visible-light irradiation (990 W Xe lamp).

7.7. Conclusions

WO₃-based photocatalysts benefit from a visible-light response due to a 2.6–2.8 eV bandgap between the valence band (VB) and the conduction band (CB). The challenge is to avoid the fast recombination of the photogenerated electrons and holes of bare WO₃ by doping or promotion (with carbon, metal(s), TiO₂, etc.). Depending on the reaction conditions, the formation of reactive oxidizing species such as superoxide anions (O₂⁻•), hydrogen peroxide (H₂O₂) and hydroxyl radicals (OH•) is reported. The WO₃-TiO₂ system is of great interest for extending the photoresponse of TiO₂ (Eg 3.0–3.2 eV) to the visible light region since the electrons photogenerated in the conducting band of TiO₂ can transfer into the CB of WO₃. The photopromoted holes can thereafter diffuse from the CB of WO₃ into the valence band of TiO₂.

8. Conclusions

Tungsten oxide materials exhibit suitable behaviors for surface reactions and catalysis such as acidic properties (mainly Brønsted sites), redox and adsorption properties (due to the presence of oxygen vacancies) and a photostimulation response under visible light. The bandgap of WO_x is commonly stressed to 2.6–2.8 eV for crystallized WO₃ 3D structures but grows up to 5.5 eV for isolated WO₄ species (such as bi-grafted, di-oxo WO₄ surface sites) or supported WO_x catalysts depending on the loading. From the operating conditions of the catalytic process, each of these behaviors is tunable by (i) controlling structure and morphology (nanoplates, nanosheets, nanorods, nanowires, nanomesh, microflowers, hollow nanospheres, etc.), (ii) thermal treatment under controlled atmosphere and (iii) interactions with other compounds (such as conductors (carbon), semiconductors (e.g., TiO₂) and precious metals); WO_x particles can be also dispersed on high specific surface area support. Based on these behaviors, WO₃-based catalysts were developed for environmental applications. One of the major advantages of the acidic and redox behaviors of WO_3−x is their suitability for the abatement of NO_x from stationary sources, associated mainly with TiO₂ and V₂O₅ (nitric acid plants first; treatment of incinerator and other combustion gases nowadays). WO₃-based catalysts also promote the abatement of some VOCs. Moreover, surface oxygen vacancies are also useful in the development of gas sensors, especially for NO₂ detection. Modifications are suitable to improve sensibility, selectivity, stability, repeatability and response time. Besides, the recent literature shows a high interest in WO₃-based materials for photocatalysis applications. Bare WO₃ is not very active due to the fast recombination of photogenerated electrons and holes, but the photocatalytic activity of WO₃ can be significantly improved by numerous associations to obtain suitable reactivity. Generally evaluated in dye degradation, WO₃-based photocatalysts are also valuable for the treatment of wastewater polluted with medicinal molecules and plant protection products. This topic should continue to retain considerable attention in the near future.