A Review of the Single-Step Flame Synthesis of Defective and Heterostructured TiO₂ Nanoparticles for Photocatalytic Applications

Abstract

Titanium dioxide (TiO₂) is an excellent UV-photocatalytic material that is widely used in various applications, including clean energy production, environmental remediation, and chemical production. However, the use of TiO₂ is limited in the field of visible light photocatalysis due to its large bandgap and fast recombination rate between electron and hole pairs, which generally results in a low photocatalytic reaction. Defect/bandgap engineering by doping and the introduction of heterojunctions has been successfully employed to improve the photocatalytic activities of TiO₂ over a wide wavelength. To apply the unconventional structured TiO₂ with high photocatalytic performance to industries, the development of efficient methods for large-scale production is of high importance. Flame synthesis is a very promising method for the rapid production of nanoparticles. In this article, we summarize the latest reports on the synthesis of defective and heterostructured TiO₂ using the single-step method of flame synthesis. Fundamental understandings of reactor configurations, synthesis conditions, precursor preparation and their physicochemical properties are intensively discussed.

1. Introduction

Titanium dioxide (TiO₂) is a semiconductor material that is commercially used in construction (e.g., paints and coating materials), health care, cosmetics, environment and energy [1,2,3]. Concerning the global issues of environmental pollution, the use of TiO₂ as a photocatalyst has become more attractive due to its strong redox properties, nontoxicity and earth abundance. The demand of TiO₂ keeps increasing year by year. It was reported that the global market of TiO₂ was approximately USD 17.19 billion in 2020, and it is expected to grow over 6% CAGR from 2021 to 2028 [4]. Therefore, it is foreseen that the production of high-quality TiO₂ would be more competitive to take a share of this market. There are many world-class manufacturers (e.g., Chemours, Cristal, Evonik-Degussa, Huntsman, Ishihara Sangyo, Kronos, Lomon-Billions and Tronox [5]) commercially producing this material, mainly for paints and coatings, plastic, pulp and paper, and cosmetics. However, it is still very challenging to produce highly active TiO₂ as a photocatalyst, typically for visible-light photocatalysis.

Anatase and rutile are major crystalized phases of TiO₂, and they are highly stable and commonly known as active photocatalysts. In addition, the metastable phases of TiO₂, such as brookite and bronze (B), were also reported as active photocatalysts [6,7]. TiO₂ is a large-bandgap semiconductor (3.2 eV, anatase [8]), which activates under UV light only. When light energy that is equal to or higher than the bandgap energy of TiO₂ is radiated, electrons (e⁻) are excited to the conduction band (CB), leaving holes (h⁺) at the valence band (VB) (see Figure 1a). Electrons and holes are active species in photocatalysis for reduction and oxidation reactions, respectively. These photogenerated electrons and holes are active species for various chemical reactions. With a strong redox ability, TiO₂ is utilized for various applications, such as organic synthesis for value-added products/chemicals [9,10,11], environmental clean-up/pollutant removals (e.g., dye decomposition and microorganism de-activation [12,13,14]) and energy (e.g., solar cells and H₂ productions [15]) (see Figure 1a).

Unfortunately, the catalytic activities of pristine TiO₂ are still under limitations: (a) inactivity in visible light, which is abundant in solar spectrum and (b) rapid electron–hole pair recombination, which shortens the life time of the charge carriers for chemical reactions. To overcome these issues, various bandgap engineering methods have been developed to improve the catalytic activities of TiO₂. (1) Defect engineering: It has been widely reported that native defects (e.g., oxygen vacancies, Ti³⁺ and interstitial Ti⁴⁺ [16,17,18,19]) and doping defects largely affect the catalytic activities of TiO₂ [20,21]. The defects created by metal and non-metal doping are used to create energy levels inside the bandgap for narrowing the bandgap energy of TiO₂ to absorb the visible range and facilitate the photo-induced charge separation (Figure 1b). (2) Heterojunction: The recombination rate between electron and hole pairs can be suppressed by the construction of heterojunctions between different crystal phases or materials, resulting in the elongation of the life time of active species for chemical reactions [22,23,24]. A heterojunction can be created by combining two or more crystal phases, such as anatase–rutile [25,26,27], anatase–bronze [28], anatase–rutile–bronze [29] and anatase–brookite [30,31]. Alternatively, there are a number of semiconductors reported to have suitable band alignment with TiO₂ for making heterojunctions, including TiO₂-SiO₂ [32,33,34], SrTiO₃-TiO₂ [35,36], TiO₂-ZnO [37,38,39] and TiO₂-ZrO₂ [13,40]. With suitable band alignment, photo-induced electrons or holes transfer from one phase to others or from one semiconductor to others, which effectively separates these pairs and provides a higher chance for reactions (Figure 1c). Furthermore, constructing heterojunctions with metal cocatalysts such as noble metals (Pt, Rh, Ru and Pd [41,42,43]) and non-noble metals (Ni, Co and Cu [44,45]) was also successfully utilized for enhanced catalytic activities. It is well understood that loading metal nanoparticles improves the photo/catalytic reaction of semiconductors based on three main factors: (1) reduction of the activation energy (overpotential) on the surface of a semiconductor [46]; (2) improvement of the electron–hole separation [47] and (3) visible-light harvesting via plasmonic effects [48,49,50] (Figure 1d).

Therefore, defective and heterostructured TiO₂ nanoparticles exhibit higher catalytic activity over pristine ones. Many studies have been conducted to develop efficient methods to synthesize these unconventional structures of TiO₂. Wet chemical methods are generally used to synthesize various kinds of nanomaterials. However, wet chemical methods might not be preferred for large-scale productions for industrial purposes due to the complexity of scaled-up technologies, the large amount of aqueous waste and high-energy consumption by post-annealing [51,52]. Gas-phase synthesis has been highlighted as a potential method for the large-scale production of nanoparticles, from single to complex components [51,53]. Our research group has been developing various kinds of gas-phase reactors, such as chemical vapor synthesis (CVS) [25], flame synthesis (FS) [51,52,54], plasma-assisted spray/vapor synthesis (PSVS) [55] and spray pyrolysis synthesis (SPS) [56,57,58,59,60], for producing many kinds of nanomaterials used as catalysts, luminescent materials and batteries. Among these reactors, flame synthesis has shown to have a greater potential in the rapid production of nanomaterials, including TiO₂ nanoparticles. Summaries and intensive reviews have been reported in a few earlier publications [61,62,63,64], focusing on a broad range of nanoparticles. In this review, we summarize flame synthesis focusing only on TiO₂ production, typically used as a potential synthesis method for multijunction and defect-containing TiO₂ nanoparticles using a single-step process.

2. Flame Synthesis (FS)

FS is among gas-phase syntheses, and it is considered as a very promising route for the rapid production of nanoparticles, because it does not require complex processes, such as liquid–solid separation and washing and drying/annealing, as in wet chemistry methods [51,52,65]. A general schematic of FS is represented in Figure 2. Generally, the process starts from the generation of metal precursor mists or droplets from liquid precursors and/or precursor mixtures by an evaporator/aerosol generator. In another case, a gas-phase precursor generated from liquid precursor using an evaporating method is fed into flame. The common Ti metal precursors are titanium (IV) chloride (TiCl₄), titanium isopropoxide (TTIP, TiC₁₂H₃₂O₄) and titanium butoxide (TBT, TiC₁₆H₃₆O₄). Metal-precursor-containing mists/aerosols are carried by carrier gas (generally N₂ or Ar) together with reaction gas (O₂ or air) into the reaction zone, or high-temperature flame, where the particles are formed. The final particles are collected at the back filters in the collecting system. This process is able to accurately control the synthesis parameters, which enables the tuning of the shapes, sizes, chemical compositions and nanostructures. The particles are generated rapidly and continuously, which is suitable for large-scale production with the purpose of commercialization. FS is flexible for particulate powder production [66,67] or direct thin film depositing/coating [68,69,70]. The particle formation process in FS is represented in Figure 2b. First, the liquid precursor is evaporated beneath a flame to form the gas-phase precursor. Later, in the high-temperature flame, the gaseous precursor goes through an oxidation reaction or thermal deposition/hydrolysis and nucleates to form nanoclusters. Along the flame, particles grow by aggregation/agglomeration to form the final particle product [53].

The properties of the FS product can be tuned based on the simple controlling parameters, including the types of precursors, the precursor’s feeding rates/methods and the concentrations, reaction temperatures and gas flow rate. The different types of precursors (e.g., TTIP, TiCl₄ or TBT) affect the morphological structure, phase and size of the particles [71,72]. In addition, the precursor concentration is considered as an important operational parameter, which greatly impacts on the phases, sizes and structures of the final products. The primary particle sizes generally increase with an increase in the precursor concentration. An increase in the concentration leads to an enhanced nucleation rate and surface reaction [71,73]. In rare cases, an increase in the concentration leads to a decrease in the primary particle size due to the enlarged numbers of nuclei, which revolutionize to form individual particles (surface growth does not occur in this case). Contradictory to the primary particles, the secondary particle sizes always increase with increase in the concentration of the precursor due to the aggregation effects [71,73]. The temperature is a very important parameter for controlling the phases, structures and sizes of the particles because it affects the reaction, nucleation, collision and sintering rate. Generally, an increase in the temperature enhances the rutile formation. The flame is obtained from the combustion of fuel gas (e.g., CH₄, H₂ and C₂H₂) and oxidant (oxygen or air) [74,75,76]. In the flame reactor, the reaction zone’s temperature can be easily controlled by the selection of the fuel/oxidant types and the flow rate of their mixtures. The combustion flame can produce a large amount of heat, up to more than 2200 °C [75]. At this high temperature, a highly crystalline TiO₂ product can be obtained without post-annealing. Binary phases of rutile and anatase are commonly formed, which are considered a target for obtaining high photocatalytic activities [67,72,76,77]. However, it is a large drawback that this high temperature also leads to particle growth by the aggregation/agglomeration of the primary particles to form larger particles (i.e., secondary particles), as in Figure 2b. Therefore, poly-dispersed sizes of particles are commonly formed with a broad size distribution, ranging from several nanometers to hundreds of nanometers. Figure 3a shows the structure of TiO₂ particles with 3.6% WO_x synthesized from a C₂H₂/O₂ flame. The particles were in the range from 5 to 50 nm [78]. It is very challenging to produce TiO₂ with a unimodal size by the flame method.

Recently, the nonconventional phases of TiO₂, such as (B)-phase or α-PbO₂-type TiO₂ (TiO₂-II), were produced by a laminar premixed stagnation flame of C₂H₂/O₂ from TTIP precursor [74,75]. Wu et al. synthesized mixed phases of anatase, rutile and srilankite (TiO₂-II) by tuning the flame conditions, such as the gas flow rates of fuel, oxygen and argon carrier [74]. Fine particles of 10 to 20 nm were produced from this reactor (see Figure 3b). By using a similar reactor, Manuputty et al. were able to synthesize α-PbO₂-type TiO₂ (TiO₂-II) and other polymorphs, such as anatase, rutile and (B)-phase [75]. It is the first report on this unusual phase because TiO₂-II exists in high-pressure conditions or stabilizes under three-phase heterojunctions of anatase–rutile–TiO₂-II. TiO₂-II/rutile was obtained under oxygen-lean conditions with a high-temperature C₂H₂/O₂ flame (1800–2269 °C). By changing the ratio of oxygen/fuel, different crystallized phases of TiO₂, including anatase, rutile, (B)-phase and TiO₂-II, could be formed. The formation of TiO₂-II in atmospheric pressure in this flame reaction is explained by a kinetics force through the oxidation and solid-state transformation of TiO_x intermediate [75]. Very few cases have generated particles with a hollow-shell structure by introducing additives (e.g., urea, HNO₃ and water content) into the precursor solution, which is explained later in this paper [79,80].

Nanopowder with limited nanostructures is produced from the FS method. Unlike solution-based methods, which can produce a variety of nanostructures, including quantum dots (QDs) and one-dimensional (1D) (e.g., nanotubes, nanorods and nanowires), 2D (e.g., nanosheets and nanoflakes) and 3D structures (e.g., nanoflower, nanospheres and core–shell structures) [81,82,83], FS allows for the production of only a few structures, such as nanospheres, nanoparticulates or hollow-shell structures. Especially, using FS, it is hard to synthesize very fine particles with a narrow size distribution, because with a high-temperature flame, a particle’s growth by aggregation/agglomeration is unavoidable.

3. Defects and Doping

3.1. Native Defects

It is known that native defects, such as oxygen vacancies, Ti³⁺, Ti⁴⁺ interstitials and Ti⁴⁺ anti-sites (Ti⁴⁺_O), can lead to changes in the atomic and electronic structures of TiO₂ and influence the photocatalytic activities of TiO₂. In flame synthesis, in situ hydrogenation was developed to induce defects, such as oxygen vacancies and Ti³⁺ doping on the surface of TiO₂ [84,85]. Huo et al. developed an in situ hydrogenation prototype by supplying H₂ gas on the top of the flame inside the chamber by a homemade ring-shaped nozzle with 16 openings (diameter = 1 mm) placed at 15 mm from the nozzle flame of a H₂/O₂ mixture (H₂/O₂ = 1) (represented in Figure 4a) [84]. TiCl₄ was used as a starting precursor heated to 30 °C. TiCl₄ gas was carried by N₂ gas and passed through the flame. The TiO₂ product was composed of spherical particles with sizes of 20–40 nm, and composite phases of anatase and rutile were obtained in this work. With the in situ hydrogenation, the rutile phase was increased from 8.8% to 31.3 % due to the increase in the temperature by H₂ oxidation. The gray/blue color of the produced powder clearly indicated the defect-induced structure of the TiO₂, because pristine TiO₂ exhibits a white color. From the high-resolution transmission electron microscopy (HR-TEM) images in Figure 4b,c, shells with disordered layer on particle surface are clearly observable [84]. This kind of structure is always obtained with post-synthesis hydrogenation for producing defected TiO₂ (termed as black TiO₂) [16]. Electron paramagnetic resonance (EPR) spectroscopy and X-ray photoelectron spectroscopy (XPS) analysis confirmed the presence of Ti³⁺ species, the amount of which could be tuned by H₂ flow rates (up to 720 L·h⁻¹). Flame-synthesized Ti³⁺-doped TiO₂ was effectively used for photocatalytic methylene-blue degradation under a Xe lamp (>400 nm) [84]. A photoelectrochemical cell (PEC) fabricated from Ti³⁺ self-doped TiO₂ generated a high photocurrent density up to 1090 nA·cm⁻², while a PEC of pristine TiO₂ could generate a photocurrent up to only 428 nA·cm⁻². Ti³⁺ created a subenergy level below the conduction band of TiO₂, which reduced the bandgap of TiO₂, resulting in visible-light absorption. In addition, Ti³⁺ acts as the hole-trap and suppresses the electron–hole recombination. Under visible light radiation, excited electrons residing on the Ti³⁺ state react with dissolved O₂ on the particle’s surface to form superoxide anions (O₂^●−). At the valence band, holes oxidize water to form hydroxyl radicals (●OH). These superoxide anions (O₂^●−) and hydroxyl radicals (●OH) are further utilized for BM degradation. Table 1 shows the summary of recent works related to defective and heterostructured TiO₂ synthesized using FS.

Using a flame stabilized on a rotating surface (FSRS), TiO_2−x with different kinds of oxygen deficiencies could be synthesized [86]. A mixture of precursor gases of TiCl₄/Ar was delivered into the C₂H₄/O₂ flame. A rotating-plate-supported glass substrate was placed at the bottom of the flame for the particle deposition (the experimental setup is summarized in Figure 4d). Under an O₂-lean flame, the oxygen vacancy content and location could be simply controlled by the particle deposition time. At a short deposition time (6 min), defect-rich particles were produced, while the elongation of the deposition time produced TiO_2−x with fewer defects. The obtained product was composed of fine spherical particles (10–20 nm) with a major phase of rutile, along with minor phases of anatase and TiO₂-II. At a short deposition time, typically 6 min, defect-rich TiO_2-x was simply observed by its blue color (Figure 4d). The TEM/HR-TEM showed the nanostructure of the particles, which was highly crystalline at the core and amorphous/disordered at the shell (Figure 4e,f). By using the temperature programmed oxidation (TPO) method, the location of the oxygen vacancy defects through the particles could be analyzed. At a short deposition time, oxygen vacancies were formed throughout the particles, including both on the surface and in bulk. Since defects are not stable, particularly surface oxygen vacancy defects, the concentration of the defects decreased with the increase in the deposition time due to the oxidation of atmospheric O₂ at the substrate. Due to the high-temperature flame (~2103 °C) on the top of the substrate, the particles underwent thermal sintering after deposition on the substrate, which resulted in the formation of a grain boundary between adjacent nanoparticles. From this work, it was revealed that moderate oxygen vacancies and oxygen vacancies locked at the grain boundary facilitated the anchoring and reduction of the Pt cocatalyst during the photocatalysis reaction. The FSRS-made TiO_2−x showed exceptional photocatalysis for H₂ generation under visible light (>400 nm) [86]. The maximum production of H₂ from the methanol solution obtained from using TiO_2−x was ~960 µmol·g⁻¹·h⁻¹, which was 12 times higher than that proceeded by commercial P25. It was proposed that balancing the defects between the surface and bulk was the major factor attributing to this photocatalysis. Typically, a Pt metal cocatalyst (Pt⁰) is preferentially deposited at or near the grain boundary, which contains higher electron concentrations. Therefore, defects such as oxygen vacancies at the grain boundary enhanced the reduction of the Pt⁶⁺ to Pt⁰ metal. However, excessive defects might act as a recombination center, which is not beneficial for photocatalysis. That is why at a long deposition time (>15 min), the photocatalytic H₂ production was lower due to the excessive surface oxygen vacancies.

3.2. Impurity Doping

A high temperature in the flame synthesis is beneficial for doping foreign elements into TiO₂ crystals. Post-annealing is generally needed with wet chemical methods (e.g., precipitation) for successful doping. A numbers of works have been reported on doping TiO₂ nanoparticles with metal ions (e.g., V, Cr, Fe, Co, Mn, Mo, Ni, Cu, Y, Ce, Nb, Zr and Mg [67,87,88,89,90,91,92,93]) and non-metal ions (e.g., S, N and F [79,94,95,96]). Doping is very complicated phenomena, which improves or deteriorates the photocatalytic activities of TiO₂. The doping behaviors depend on a dopant’s size, valent states and the temperature of the flame.

Table 1. Summary of defective and heterostructured TiO₂ nanoparticles synthesized by FS.

Materials	Precursor/Concentration/Feeding Rate	Gas	Product’s Properties	Photocatalysis	Reference
Ti3+-TiO2	TiCl4/0.01~0.2 mol·h−1	H2/O2/N2	spherical; sizes (20–40 nm); A-R (A: 68.7%); bandgap (3.07 eV); SSA (42.3 m2·g−1); production rate~3.2 g·h−1	MB degradation and PEC under visible light (>400 nm); max. photocurrent: 1090 nA·cm−2	[84]
TiO2-x	TiCl4/3 mL·h−1	C2H4/O2/Ar/N2	spherical; sizes (10–20 nm); A-R-TiO2-II (R:70%); bandgap (2.8–3 eV); SSA (100–120 m2·g−1)	H2 generation under visible light (>400 nm) with Pt; max. H2 rate: 960 µmol·g−1·h−1	[86]
M- TiO2 (M = V, Cr, Fe, Co, Mn, Mo, Ni, Cu, Ce, Y and Zr)	TTIP (0.5M)/vanadium (V) tri-i-propoxy oxide/chromium (III) 2-ethylhexanoate/iron (III) naphthenate/cobalt 2-ethylhexanoate/manganese (III) naphthenate/molybdenum 2-ethylhexanoate/nickel (III) naphthenate/copper (II) 2-ethylhexanoate/zirconyl 2-ethylhexanoate/3 mL·min−1	CH4/O2	spherical/particulate; sizes (51–99 nm); A-R (A: 12–80%); bandgap (2.37–3.21 eV); SSA (60–108 m2·g−1)	acetonitrile conversion under visible light; max. rate constant of Cr-doped TiO2, k: 0.812 m−3·g−1·h−1	[88,89]
N-TiO2	TTIP-nitric acid-ethanol-DI-urea/3 mL·min−1	CH4/O2	spherical/particulate; sizes (50–300 nm); A-R (A: 47–66%); bandgap (2.47–2.95 eV); SSA (17–38 m2·g−1)	phenol degradation under visible light	[79,97]
N-TiO2	TBT-ethanol (0.5M)/NH3-H2O (28 wt.%, 2 mL·min−1)/5 mL·min−1	H2/Air/N2	spherical; sizes (10–30 mn); A-R (A: 90.8%); bandgap (2.90 eV); SSA (45.1 m2·g−1);	N/A	[98]
S-TiO2	TTIP-sulfuric acid-ethanol-DI (0.3M)/3 mL·min−1	CH4/O2	spherical; sizes (75–311 nm); A-R (A: 59–66%); bandgap 2.78–2.94 eV; SSA (5–13 m2·g−1);	acetaldehyde degradation under visible light	[95]
F-TiO2	TTIP -propionic acid-xylene/ hexafluorobenzene/4 mL·min−1	CH4/O2	A-R (A: 86–91%); SSA (113–117 m2·g−1); UV-vis abs < 400 nm	degradation of FA and TAOH under UV-light (>340 nm); max. rate constant k0FA: 1.89 × 10−7 M·s−1, k0TAOH: 14 nM·s−1	[94]
Pt/TO2	BTB-ethanol/platinum acetylacetonate/3 mL·min−1	CH4/O2/Ar	particulate; size (~10 nm); A-R (A: 83–89%); SSA (167 m2·g−1); bandgap (3.07–3.19 eV);	H2 production under Xe lamp (300 W), max. H2-rate: 552.39 µmol·h−1	[99]
Pt/TiOx	TTP/platinum acetylacetonate-xylene-acetonitrile/0.4 L·min−1 (N2 flow)	CH4/O2/N2/H2	particulate; size (20–50 nm); A-R (A: 69%); SSA (74 m2·g−1); bandgap (2.88 eV)	CO2 reduction under Xe lamp; max. AQY: 1.49%, CH4 selectivity: 81%	[85]
Pt/N-TiO2	TBT-ethanol/chloroplatinic acid/5 mL·min−1	H2/O2/N2/NH3	particulate; size (10–25 nm); A-R (A: 70.86%); SSA (61.4 m2·g−1);	N/A	[76]
Pt/F-TiO2	TTIP-propionic acid-xylene (0.6 M)/hexafluorobenzene/hexachloroplatinic acid hydrate/4 mL·min−1	CH4/O2	A-R (A: 86–91%); SSA (130–142 m2·g−1); UV-vis abs < 400 nm	methanol steam-reforming under UV-light (>350 nm); max. H2 rate: 22 mmol·h−1·g−1	[100]
Pd/TiO2	TTIP-ethylhexanoic-acetonitrile (0.159 M)/Pd-acetylacetonate/8 mL·min−1	CH4/O2	spherical/particulate; size (11–17 nm); A-R-amorphous (A: 74–86%); SSA (85–116); UV-vis abs < 550 nm	NO removal under solar light; max. NO removal: 67% after 5-h reaction	[101,102]
Au/TiO2	TTIP-xelene-pyridine (0.15 M)/1 % dimethyl-gold (III)-acetylacetonate/3.1 mL·min−1	CH4/O2	spherical/particulate; size (10–500 nm); A-R (A: 90 wt.%); size (10~500 nm); SSA (106 m2·g−1)	water-splitting reaction under Hg lamp (330–450 nm); max. H2 rate: 52 µmol·h−1·g−1	[103]
AuPd/TiO2	TTIP-xylene-acetonitrile (0.5 M)/gold chlorite hydrate/palladium acetylacetonate/5 mL·min−1	CH4/O2	spherical/particulate; size (10–30 nm); A-R (major A); SSA (99–152 m2·g−1)	N/A	[104]
AuPt/TiO2	TBT-xylene-ethanol (0.05 mol.)/chloroplatinic acid hexahydrate/chloroauric/5 mL·min−1	H2/O2	spherical/particulate; size (20–30 nm); A-R (major A); SSA (58–78 m2·g−1)	N/A	[105]
Cu/TiO2	TBT-ethanol/Cu(NO3)2.3H2O/5 mL·min−1	CH4/O2	spherical; A-R (A 90–80%); size (~10 nm); bandgap (3.09–3.15 eV); SSA (94–106 m2.g−1)	CO2 reduction under Xe lamp (300–400 nm); max. AQYCH4: 0.087% and AQYCO: 0.057%	[106]
CoPt/TiO2	TBT-ethanol/Co(NO3)2.6H2O/5 mL·min−1	H2/O2	spherical/particulate; size (5–25 nm); A-R (A 69.7%); size (5~25 nm); SSA (60.2 m2·g−1)	N/A	[107]

Table 1. Summary of defective and heterostructured TiO₂ nanoparticles synthesized by FS.

Materials	Precursor/Concentration/Feeding Rate	Gas	Product’s Properties	Photocatalysis	Reference
Ti3+-TiO2	TiCl4/0.01~0.2 mol·h−1	H2/O2/N2	spherical; sizes (20–40 nm); A-R (A: 68.7%); bandgap (3.07 eV); SSA (42.3 m2·g−1); production rate~3.2 g·h−1	MB degradation and PEC under visible light (>400 nm); max. photocurrent: 1090 nA·cm−2	[84]
TiO2-x	TiCl4/3 mL·h−1	C2H4/O2/Ar/N2	spherical; sizes (10–20 nm); A-R-TiO2-II (R:70%); bandgap (2.8–3 eV); SSA (100–120 m2·g−1)	H2 generation under visible light (>400 nm) with Pt; max. H2 rate: 960 µmol·g−1·h−1	[86]
M- TiO2 (M = V, Cr, Fe, Co, Mn, Mo, Ni, Cu, Ce, Y and Zr)	TTIP (0.5M)/vanadium (V) tri-i-propoxy oxide/chromium (III) 2-ethylhexanoate/iron (III) naphthenate/cobalt 2-ethylhexanoate/manganese (III) naphthenate/molybdenum 2-ethylhexanoate/nickel (III) naphthenate/copper (II) 2-ethylhexanoate/zirconyl 2-ethylhexanoate/3 mL·min−1	CH4/O2	spherical/particulate; sizes (51–99 nm); A-R (A: 12–80%); bandgap (2.37–3.21 eV); SSA (60–108 m2·g−1)	acetonitrile conversion under visible light; max. rate constant of Cr-doped TiO2, k: 0.812 m−3·g−1·h−1	[88,89]
N-TiO2	TTIP-nitric acid-ethanol-DI-urea/3 mL·min−1	CH4/O2	spherical/particulate; sizes (50–300 nm); A-R (A: 47–66%); bandgap (2.47–2.95 eV); SSA (17–38 m2·g−1)	phenol degradation under visible light	[79,97]
N-TiO2	TBT-ethanol (0.5M)/NH3-H2O (28 wt.%, 2 mL·min−1)/5 mL·min−1	H2/Air/N2	spherical; sizes (10–30 mn); A-R (A: 90.8%); bandgap (2.90 eV); SSA (45.1 m2·g−1);	N/A	[98]
S-TiO2	TTIP-sulfuric acid-ethanol-DI (0.3M)/3 mL·min−1	CH4/O2	spherical; sizes (75–311 nm); A-R (A: 59–66%); bandgap 2.78–2.94 eV; SSA (5–13 m2·g−1);	acetaldehyde degradation under visible light	[95]
F-TiO2	TTIP -propionic acid-xylene/ hexafluorobenzene/4 mL·min−1	CH4/O2	A-R (A: 86–91%); SSA (113–117 m2·g−1); UV-vis abs < 400 nm	degradation of FA and TAOH under UV-light (>340 nm); max. rate constant k0FA: 1.89 × 10−7 M·s−1, k0TAOH: 14 nM·s−1	[94]
Pt/TO2	BTB-ethanol/platinum acetylacetonate/3 mL·min−1	CH4/O2/Ar	particulate; size (~10 nm); A-R (A: 83–89%); SSA (167 m2·g−1); bandgap (3.07–3.19 eV);	H2 production under Xe lamp (300 W), max. H2-rate: 552.39 µmol·h−1	[99]
Pt/TiOx	TTP/platinum acetylacetonate-xylene-acetonitrile/0.4 L·min−1 (N2 flow)	CH4/O2/N2/H2	particulate; size (20–50 nm); A-R (A: 69%); SSA (74 m2·g−1); bandgap (2.88 eV)	CO2 reduction under Xe lamp; max. AQY: 1.49%, CH4 selectivity: 81%	[85]
Pt/N-TiO2	TBT-ethanol/chloroplatinic acid/5 mL·min−1	H2/O2/N2/NH3	particulate; size (10–25 nm); A-R (A: 70.86%); SSA (61.4 m2·g−1);	N/A	[76]
Pt/F-TiO2	TTIP-propionic acid-xylene (0.6 M)/hexafluorobenzene/hexachloroplatinic acid hydrate/4 mL·min−1	CH4/O2	A-R (A: 86–91%); SSA (130–142 m2·g−1); UV-vis abs < 400 nm	methanol steam-reforming under UV-light (>350 nm); max. H2 rate: 22 mmol·h−1·g−1	[100]
Pd/TiO2	TTIP-ethylhexanoic-acetonitrile (0.159 M)/Pd-acetylacetonate/8 mL·min−1	CH4/O2	spherical/particulate; size (11–17 nm); A-R-amorphous (A: 74–86%); SSA (85–116); UV-vis abs < 550 nm	NO removal under solar light; max. NO removal: 67% after 5-h reaction	[101,102]
Au/TiO2	TTIP-xelene-pyridine (0.15 M)/1 % dimethyl-gold (III)-acetylacetonate/3.1 mL·min−1	CH4/O2	spherical/particulate; size (10–500 nm); A-R (A: 90 wt.%); size (10~500 nm); SSA (106 m2·g−1)	water-splitting reaction under Hg lamp (330–450 nm); max. H2 rate: 52 µmol·h−1·g−1	[103]
AuPd/TiO2	TTIP-xylene-acetonitrile (0.5 M)/gold chlorite hydrate/palladium acetylacetonate/5 mL·min−1	CH4/O2	spherical/particulate; size (10–30 nm); A-R (major A); SSA (99–152 m2·g−1)	N/A	[104]
AuPt/TiO2	TBT-xylene-ethanol (0.05 mol.)/chloroplatinic acid hexahydrate/chloroauric/5 mL·min−1	H2/O2	spherical/particulate; size (20–30 nm); A-R (major A); SSA (58–78 m2·g−1)	N/A	[105]
Cu/TiO2	TBT-ethanol/Cu(NO3)2.3H2O/5 mL·min−1	CH4/O2	spherical; A-R (A 90–80%); size (~10 nm); bandgap (3.09–3.15 eV); SSA (94–106 m2.g−1)	CO2 reduction under Xe lamp (300–400 nm); max. AQYCH4: 0.087% and AQYCO: 0.057%	[106]
CoPt/TiO2	TBT-ethanol/Co(NO3)2.6H2O/5 mL·min−1	H2/O2	spherical/particulate; size (5–25 nm); A-R (A 69.7%); size (5~25 nm); SSA (60.2 m2·g−1)	N/A	[107]

Metal doping: Inturi et al. reported a wide range of metal ion doping into FS-made TiO₂ by feeding a mixed TTIP precursor solution into a CH₄/O₂ flame [87,88,89]. TTIP and metal ion precursors were dissolved in o-xylene/acetonitrile as a solvent to achieve a total concentration of 0.5 M. Metal precursors, such as chromium (III) 2-ethylhexanoate, vanadium (V) tri-i-propoxy oxide, iron (III) naphthenate, manganese (III) naphthenate, nickel (III) naphthenate, copper (II) 2-ethylhexanoate, cobalt 2-ethylhexanoate, molybdenum 2-ethylhexanoate, zirconyl 2-ethylhexanoate and yttrium (III) naphthenate, were selected for Cr, V, Fe, Mn, Ni, Cu, Co, Mo, Zr and Y doping, respectively. The doping concentration was fixed to 5% (M/Ti = 0.05). In general, doping with these metals promoted rutile phase formation. Among these metal dopants, Co and Ni induced the highest fraction of rutile phase (84 and 88%), while undoped TiO₂ contained only ~19% rutile [88]. It is commonly known that doping induces defects, such as oxygen vacancies or Ti³⁺ species, which should be an important nucleation center for rutile phase formation [77]. Therefore, the rutile phase increased with the increase in the doping contents. In addition to a phase change, a change in the bandgap of the TiO₂ was also observed with some metal dopants. A bandgap reduction was observed with Co, Cr, Cu, Fe, Mn, Ni and V doping. The undoped TiO₂ had a bandgap energy in the UV range, 3.08 eV. Ni doping induced the smallest bandgap of TiO₂ (2.37 eV). However, some of the other metal dopants, such as Zn, Y, Mo and Ce, enlarged the bandgap of TiO₂. These changes in the bandgap energies indicate that the doping behaviors were different from the individual metal ions. Some elements with ionic sizes similar to or smaller than a Ti⁴⁺ ion, such as Cr, Fe, Co, Mn, Ni, Mo, V, Cu and Zr, are more likely to be doped into crystal lattices of TiO₂ (interstitial doping and/or substitutional doping). Some elements with bigger ionic sizes than Ti⁴⁺ ions, such as Y and Ce, preferentially formed as metal oxides dispersed on the surface of TiO₂. However, in Inturi’s work, the metal oxide-related phases were not detected by the XRD analysis [88], which might be due to the lower content of the doping concentrations. The photocatalytic activity of the synthesized materials was evaluated by acetonitrile degradation under visible light. Although some dopants reduced the bandgap of TiO₂, which resulted in visible light absorption, not all of them improved the photocatalytic activities of TiO₂ in visible light. Only three dopants (Cr, Fe and V) enhanced the visible-light photocatalytic activities of the acetonitrile degradation. Among them, Cr doping showed the highest photocatalytic activities of the acetonitrile conversion due to the fact of its strong interaction between of the Cr–O–Ti bond [88,89]. The maximum decay constant of acetonitrile degradation was K_A = 0.812 m³·g⁻¹·s⁻¹ [89]. Doping Cr ions induced a substantial amount of Ti³⁺, which might account for the improving visible light photocatalysis [108,109,110,111,112]. The maximum Ti³⁺ concentration was up to Ti³⁺/Ti⁴⁺ = 1.14 at Ti/Cr = 40 [88,89]. The synchronization of the Cr ions and TiO₂ activated the visible light photocatalytic activities [108,109,110,111,112]. The Cr⁶⁺ species was the most active species in the Cr-doped TiO₂, because the synchronization of Cr⁶⁺ and TiO₂ took place by the Cr⁶⁺ = O²⁻ → Cr⁵⁺–O¹⁻ route, which is responsible for the visible light photocatalytic reaction [89,108].

In the Ce-doping case, Inturi et al. reported that the bandgap of TiO₂ increased (3.08 vs. 3.13 eV), and the photocatalytic degradation of acetonitrile under UV radiation was reduced [88]. However, other works conducted by Mikaeili et al. [67] and Chaisuk et al. [113] showed different results, where the bandgap was reduced by the Ce doping, and the photocatalytic degradation of methylene blue under visible light (>400 nm) also increased. Both Mikaeili and Chaisuk observed the existence of CeO₂, even at low doping concentration (3~5 at.%), while Inturi did not. This complicated phenomenon of the Ce-doping system was probably due to the fact of its unusual redox shift (Ce⁴⁺<–>Ce³⁺) [89]. Because the ionic size of Ce (Ce⁴⁺ = 101 pm; Ce³⁺ = 115 pm) is larger than Ti (Ti⁴⁺ = 74.5 pm; Ti³⁺ = 81 pm), it is hard to dope Ce ions into a TiO₂ lattice. However, it is possibly formed as a solid solution and composite (e.g., CeTi₂O₆ [114] and CeO₂-TiO₂ [115]). In an oxidative atmosphere, CeO₂-TiO₂ was stable, while in reductive atmospheres, other phases, such as Ce₂O₃·4TiO₂ and Ce₂O₃·2TiO₂, were reported [115]. In flame synthesis, the presence of CeO₂ was observed in the rutile phase only. Because of its larger radius, cerium was ejected outside the crystal lattice of the TiO₂ rutile and nucleated on the surface as CeO₂ clusters, later forming CeO₂ nanoparticles [114]. The existence of brannerite (CeTi₂O₆) at the interface of CeO₂ and TiO₂ is a typical characteristic of flame-made products. Brannerite is high-temperature entropy-stabilized compound, which is synthesized at a high temperature and subsequent quenching. A flame reactor is characterized by the suitable conditions of both high temperature and quenching process, which allow to obtain the brannerite phase. The formation of brannerite stabilizes oxygen vacancies at the interfaces of CeO₂-TiO₂ due to the existence of Ce³⁺ in the brannerite. Furthermore, the incorporation of Ce³⁺ into the TiO₂ lattice induces oxygen vacancies due to the charge compensation of Ce³⁺ and Ti⁴⁺, which enhances the rutile–anatase transformation [114]. Doping with Ce varies the band structures of TiO₂ depending on the doping concentrations, which alter the photocatalytic activities of TiO₂ [115]. In terms of heterogeneous catalysis, such as CO oxidation [116] and NO_x reduction [114], CeO_x-TiO₂ always shows a better performance due to the Ce-Ti-O interactions at the interface.

Non-metal doping: Anion doping is generally used to modify the valence band of TiO₂ for narrowing the bandgap. In contrast to cation doping, anion doping does not play the role of charge carriers and recombination center, which minimizes the side effects on photocatalysis. Only a few works have reported anion doping (e.g., N, S and F-doping) by flame synthesis.

N doping has been extensively studied for improving the visible light photocatalysis of TiO₂. However, the doping sites of N altered the photocatalytic activities of TiO₂ [117,118]. In flame synthesis, N doping was obtained by introducing nitrogen sources (e.g., urea and nitric acid) into the Ti-precursor solution [79,80,119] or supplying a NH₃ solution into the flame chamber [77,98]. Nitrogen-containing materials, such as aqueous solutions of nitric acid and urea, were premixed with a Ti-precursor solution [79,80,97,119] and sprayed into the flame nozzle. The precursor solution was simply prepared by dissolving TTIP into dilute nitric acid followed by the addition of water/ethanol and urea [79]. The amount of the N-doping concentration did not alter much with the increase in the nitric acid concentration. From the XPS analysis, the N-doping concentration merely increased from 0.11 at.% to 0.13 at.% with the increase in the nitric acid concentration from 0.8 to 3 M, respectively [80]. In addition, the bandgap of TiO₂ was slightly reduced from 3.12 (pristine state) to 2.95 eV upon the addition of 3 M nitric acid. It was expected that the majority of the nitric acid was oxidized in the flame to N₂, resulting in the limitation of the nitrogen-doping concentration. However, when the secondary N-source urea was added, the N concentration increased from 0.20 at.% to 1.63 at.% with the increase in the urea concentration from 0.25 to 2 M, and the bandgap largely decreased to 2.51 eV (at 2 M urea) [80]. Urea was thermally decomposed to NH₃ and reacted with an oxygen functional group of TiO₂, which subsequently formed intermediates resulting from ammonia pyrolysis (e.g., N, NH, NH₂ and N₂H_x, x = 1–4). These intermediates would react with TiO₂ to form N-doped TiO₂. Both substitutional and interstitial doping was observed in this preparation method. However, the addition of secondary N-source urea increased the interstitial N doping based on the XPS deconvoluting data [119]. In addition to changing the optical bandgap of TiO₂, the addition of nitric acid and urea suppressed the formation of anatase, where the percentage of anatase decreased from 82% in the undoped TiO₂ to 62% and 59% with the addition of 1 M nitric acid and 1 M nitric acid + 1 M urea [119]. The enhanced rutile phase formation with the addition of nitric acid and urea should be attributed to defect formation (e.g., oxygen vacancies) due to the N-doping effects and increased flame temperature due to the presence of additional organics, such nitric acid and urea, which acted as fuels.

The effects of the addition of HNO₃ and urea on the morphologies and nanostructures are depicted in Figure 5a,b, from Reference [95]. Without the addition of a nitrogen source, the TiO₂ particles were small (<20 nm). The addition of nitrogen sources, such as HNO₃ and urea, produced large spherical particles with sizes ranging from 30 to 150 nm with a hollow-shell structure. This hollow-shell structure formed from the low enthalpy flame due to the addition of N-containing sources, such as urea and HNO₃/water, which resulted in the enhanced precipitation of the reaction/intradroplet reaction [79,80,97,119]. The selected area electron diffractions of the corresponding particles informed the polycrystalline nature, and were indexed to the anatase and rutile phases. The inset in each figure shows images of TiO₂ powder with and without HNO₃-urea. The color of the TiO₂ powder changed from white to yellowish with HNO₃ and HNO₃-urea, inferring the rich N doping with urea as a secondary N-doping source. The optical absorption in the visible range was also enhanced with the addition of urea. From the same research group, it was also revealed that the addition of water improved the visible-light absorption of the N-doped TiO₂ (see the absorption spectra in Figure 5c) [79]. The nitrogen-doping concentration was increased from 0.43 at.% to 0.96 at.% when the water content was increased from 25 to 50%. Therefore, the addition of water improved the visible absorption of TiO₂, which demonstrates its potential as a photocatalyst candidate for solar harvesting. Alternatively, N doping was also prepared by spraying a NH₃/water solution onto the top of the flame, which allowed for the formation of N-surface-doped TiO₂ [76,98]. Figure 5d shows that a NH₃ aqueous solution was supplied on the top of the flame, and the generated TiO₂ nanoparticles were treated with NH₃ molecules and formed surface N-doped TiO₂, where the yellowish color of the powder indicated successful N doping. This procedure is similar to N doping prepared by the conventional NH₃-annealing method [98]. N doping was dominant in the interstitial sites at the surface of the TiO₂ particles. The photocatalytic degradation of phenol under visible light ( >420 nm) was studied to evaluate the photocatalytic activity of newly synthesized N-doped TiO₂. The 500 ppm phenol was significantly degraded within a 2 h reaction by N-doped TiO₂, while pristine TiO₂ showed negligible phenol degradation. Typically, when a secondary N source of urea was added, the photodegradation of phenol also increased, which indicates that the photocatalytic activity of N-TiO₂ under visible light was affected by the N-doping concentration. The dominant factor for the enhanced photocatalytic activities under visible light should be due to the reduced band gap energy of TiO₂ by incorporating a N atom into the TiO₂ lattice, resulting in the alteration of the electronic band structure. The N doping creates a mid-gap energy state of N 2p formed above the O 2p valence band and narrows the band gap of TiO₂ from 3.07 to 2.47 eV [97].

S doping: S element was successfully doped to TiO₂ material using one-step FS with TTIP and sulfuric acid as Ti and S precursors [95]. It was suggested that the preparation of the titanium oxysulfate precursor allowed for the S doping into the TiO₂ framework. The titanium oxysulfate precursor was prepared by mixing TTIP in water to form white precipitate. This precipitate was redissolved by the addition of sulfuric acid. The TTIP-water-sulfuric acid mixture was diluted into ethanol to form the FS precursor. The S-doping amount was controlled by the sulfuric acid concentration. Increasing the sulfuric concentration enhanced the S doping and the rutile phase formation. The percentage of the rutile phase increased from 19% to 34% upon the addition of 1 M sulfuric acid. A further increase in the sulfuric acid above 1 M slightly increased the rutile phase. The maximum doping concentration of S was 8.9 at.% with the addition of 2 M sulfuric acid. The major S species doped into TiO₂ were S⁶⁺ and S⁴⁺, as analyzed by XPS. S⁶⁺ doping was considered as the major factor for enhancing visible light utilization by inducing mid-energy level below the conduction band of TiO₂, which led to smaller bandgaps. Flame-made S-doped TiO₂ was utilized for acetaldehyde removal under visible light irradiation. The pristine TiO₂, as a control, showed inactivity for acetaldehyde degradation under visible light due to its large bandgap. The photodegradation of acetaldehyde was significantly improved with the S-doping concentration. The following factors may have affected the photocatalytic activity of the S-doped TiO₂: (1) a reduction in the bandgap by the formation of mid-gap energy levels induced by S⁶⁺ doping under a Ti 3d conduction band, which resulted in enhanced visible light absorption; (2) lattice defects formed by S doping enhanced the hydroxyl ion absorption and rapid generation of hydroxyl radicals, which are strong oxidation species; (3) formation of Ti-O-S bonds inhibit electron transition and reduce the electron–hole recombination, resulting in high photocatalytic activity [95].

F doping: F doping was carried out by mixing F sources (e.g., hexafluorobenzene (C₆F₆)) into a Ti precursor and spraying into the flame reactor [94,100]. The surface area of the obtained F-doped TiO₂ did not vary with the changes in the amount of the F source, which indicates that the morphological structures of F-doped TiO₂ were not much different from undoped TiO₂. However, F doping prohibited the rutile phase formation, which can be attributed to the inhibited condensation of spiral chains of the rutile TiO₆ octahedral at a high temperature [94]. The optical absorption revealed that F doping did not have an effect on changing the UV-vis absorption spectra; the UV-vis absorption spectra of both F-doped and undoped TiO₂ were <400 nm, which are typical spectra corresponding to the rutile–anatase mixed phase. It is already known that F doping creates new localized levels composed of F 2p states below the VB of TiO₂, but without inducing any variation in the optical absorption. The photocatalytic activity was evaluated by the formic acid degradation under UV light (>340 nm), but the degradation efficiency deteriorated with F doping. Physically absorbed fluorine ions hampered the absorption of formate anions on the surface of the catalyst, typically in an acidic condition, which resulted in the lower photocatalytic degradation of formic acid. However, the photocatalytic conversion of terephthalic acid (TA) into 2-hydroxyl-terephthalic acid (TAOH) increased with F doping, because fluorine ions can be replaced by hydroxyl group at a basic condition, which facilitated the photocatalytic oxidation [94].

4. Heterostructure

Based on the feeding methods of precursors, FS is considered a very promising strategy for the synthesis of multicomponent TiO₂-based compounds composited with other materials within a single step. Generally, multimetal precursors are premixed and sprayed/evaporated into a single flame (Figure 6a). To ensure a homogeneous solution of the mixed precursors, the selection of the types of metal precursors and the precursor’s solvents is very important. This feeding method is very facile, but there are some drawbacks to the method, as follows: Unintentional doping will possibly form and change the properties of the synthesized materials. The coverage of the active sites by the inactive phase of the secondary materials is also a main drawback in using premixed precursors. For example, active metal nanoparticles can be covered by TiO₂, which reduces the catalytic activities. Another method is the separate feeding of metal precursors into the same flame or a different flame (Figure 6b,c). These methods can minimize the alloy formation (and/or doping), and they are more promising in terms of synthesizing multicomponent nanoparticles and the facile control of individual components [86,104]. These feeding methods were successfully used to load metal/oxide nanoparticles onto TiO₂ [102,104].

4.1. TiO₂/Noble Metals/Oxides

Numerous works have reported using nanocomposites of TiO₂ with noble metals and metal oxides, such as Pt/TiO₂ [76,85,100,107,120,121], Pd/TiO₂ [65,101,102,122,123,124] and Au/TiO₂ [104,125].

Pt loading: FS was applied to load a Pt single atom, nanocluster or nanoparticle onto a TiO₂ support. In most cases, platinum acetylacetonate (C₁₀H₁₄O₄Pt) and platinum chloride hexahydrate (H₂PtCl₆·6H₂O) were used as the Pt metal precursor mixed with TTIP or TBT. The Pt precursor and Ti precursor were premixed as a binary metal feeding solution and sprayed into a single flame [99,126,127,128,129]. In some cases, the Pt precursors were fed separately by a sublimation unit into the same flame [130]. Gao et al. synthesized single-atom Pt catalysts anchored to a TiO₂ support at an ultralow concentration (<0.1 at.%), which is very important for activating the H₂ evolution reaction [99]. Due to the difference in the boiling points of PtO₂ and TiO₂ (450 vs. 2727 °C), in hot temperature of the flame (close to the burner), PtO₂ formed as a gas phase surrounding the TiO₂ nanoparticles. Downstream at a lower temperature, PtO₂ condensed onto the TiO₂ surface as a single atom and/or nanoclusters. It was observed that the Pt nanoclusters and nanoparticles were preferentially formed at a higher concentration (Figure 7b,c). At an ultralow concentration, premixed Pt precursor does not affect phase or bandgap energy of TiO₂. A good dispersion of Pt onto TiO₂ nanoparticles was achieved at both conditions. The Pt-loaded TiO₂ exhibited outstanding photocatalytic H₂ generation, with a production rate up to 552.39 μmol·h⁻¹, which was 107.5 times higher than the H₂ generated by P25 [99]. The FS enabled the loading of Pt metal at an ultralow concentration as well as well-dispersed single-atom Pt onto the surface of the TiO₂, which was mainly attributed to this extremely high H₂ evolution reaction. Moreover, the FS-made Pt/TiO₂ can potentially be utilized for various catalytic reactions, such as the mineralization of organic matters [127,129], catalytic hydrogenation of 3-nitrostyrene [126], SO₂ oxidation [130] and gas sensors [128].

Attempts to use FS to produce more complex structures of TiO₂ by co-inducing defects with Pt/TiO₂ heterostructures have also been successfully reported. The effects of surface-oxygen vacancies on heterostructures of Pt-supported TiO₂ nanoparticles were studied by in situ hydrogenation into flame chambers [85]. Lin et al. invented an in situ hydrogenation system. H₂ gas was supplied by a ring-shaped pipe into the flame chamber in order to provide a reducing atmosphere for inducing the surface oxygen vacancies onto TiO₂ nanoparticles [85]. The Ti-precursor vapor was carried by N₂, while the Pt precursor was separately sprayed into the same flame, which resulted in the formation of a Pt/TiO₂ composite. These composite nanoparticles passed through hydrogenation zone at the top of the flame, leading to the formation of Pt/TiO₂-V_o complex. Oxygen vacancies were successfully induced on the surface of TiO₂ supports and enhanced charge-transfer to Pt nanoparticles, which enhanced the photocatalytic properties for the CO₂ reduction. The electronic metal–support interaction also prevented the loss of metallic Pt⁰ and oxygen vacancies, which improved the lifetime of the catalyst. With a similar concept, a N-doped Pt/TiO₂ complex was synthesized by introducing a NH₃/water solution into the top of the flame [76]. This in situ N-doping procedure was also used to synthesize a composite of Pt/TiO₂ heterostructures, which showed high catalytic CO oxidation and high thermal stability due to Pt-N bond formation. F-doped Pt/TiO₂ was simply produced by mixing all metal precursors (TTIP, hexachloroplatinic acid hydrate, propionic acid and xylene) with a F source of hexafluorobenzene and sprayed into a single flame [100]. The F doping by substitution of the O sites stabilized the metal phase of Pt in the FS. F-TiO₂/Pt increased the hydroxyl moieties on the surface of the TiO₂, which resulted in higher photocatalytic activities of methanol stream-reforming to produce H₂ and CO₂. The maximum H₂ and CO₂ generation from the optimized samples were 22 and 3.3 mmol·g⁻¹·h⁻¹. At lower F doping (<10 at.%), a rich surface of hydroxyl moieties led to an indirect hydroxyl mediated mechanism, which was more selective toward the complete oxidation of methanol to CO₂ and generated H₂. However, at a higher F-doping concentration (>10 at.%), the increase in bulk defects, which acted as an electron–hole recombination center, could reduce the photocatalytic activity [100].

Pd loading: Fujiwara et al. synthesized Pd single-atom-, sub-nanocluster- and nanoparticle-supported TiO₂ composites from a solution precursor of 2-ethylhexanoic acid, acetonitrile, TTIP and palladium acetylacetonate in a CH₄/O₂ flame [101,102,122]. Sub-nanoclusters of Pd (<1 nm) were well distributed on the TiO₂ nanoparticles (see Figure 7d). In FS, substantial oxygen vacancies on the surface of TiO₂ play an important role in stabilizing the Pd nanoclusters. The interaction of the oxygen vacancies and Pd on the FS-made TiO₂ created an intermediate state between the metal and PdO. The Pd sub-nanoclusters were highly stable and resistant to thermal annealing up to 600 °C for 2 h [101]. In later reports, Fujiwara investigated the single atom of Pd-supported TiO₂ using the same synthesis procedure [102,122]. Isolated Pd atoms were observed with an ultralow concentration (<0.1 wt.%), while at a higher concentration, nanoclusters were preferentially formed. Isolated Pd atoms were shown to have higher activities for photocatalytic NO_x removal under solar light than that of the nanoclusters. Within only a 5 h reaction, 67% of the NO_x was decomposed when 1 wt.% Pd-TiO₂ was used, which is much higher than that degraded by commercial P25. The richness of surface defects on the FS-made TiO₂ anchoring the sub nanosized Pd was considered to be the main contribution to the high photocatalytic NO_x removal [101]. Different from Pt, the fact that Pd and TiO₂ have similar boiling points provides the possibility of Pd doping into TiO₂. However, Pd doping into TiO₂ had a minor impact on the photocatalytic activities in these studies [102,122].

Some other works focused on Pd-nanoparticle-anchored TiO₂ [123,124]. By using a similar FS system to Fujiwara [102,122], Mekasuwandumrong et al. obtained nanosized Pd (2–5 nm) decorated on TiO₂ nanospheres, which exhibited excellent catalytic activity for the hydrogenation of heptyne, with a high selectivity of heptane (>90%) [123]. Zong et al. used a stagnation swirl flame (SSF) with an ultrafine spray to produce Pd-nanoparticle-(<2.5 nm)-supported TiO₂ (7–8 nm) [124]. A typical feature of the obtained products from the SSF was a high specific surface area (210–220 m²·g⁻¹) due to its short residence time and fast quenching, which enabled the generation of very fine particles and a nonaggregated structure. The SSF-made Pd/TiO₂ was utilized for the combustion of CH₄, and it showed high catalytic activity at a low temperature (T₂₀ = 293 °C), and the optimized loading concentration of Pd was 15 wt.% [124].

Au loading: Au nanoparticles are among the active noble cocatalysts, and they can be loaded onto TiO₂ nanoparticles for various catalytic reactions. Mädler et al. used gold (III) chloride hydrate as a precursor for gold nanoparticles [125]. Gold precursor was mixed with Ti precursor (TTIP/Xylene) and sprayed into a CH₄/O₂ flame. The sizes of the Au nanoparticles formed varied with the change in the Au precursor concentration; the crystal size of the Au nanoparticles increased from 3.8 to 14.8 nm when the precursor concentration rose from 1 to 4 wt.%. However, the particle structure and crystal phase of TiO₂ were not affected by the addition of Au nanoparticles. It was proposed that the Au nanoparticles formed by heterogenous nucleation on the metal oxide support or homogenous nucleation of the gold itself [125]. In later work, Chiarello et al. premixed gold precursor (dimethyl-gold(III)-acetylacetonate) into TTIP/Xylene/pyridine solution as a precursor feed [103]. A mixed phase of rutile and anatase (90%) was produced with bimodal sizes, ranging from 100 to ~500 nm for the large spherical particles and 10 nm for the small particles. Au nanoparticles with an average particle size of 2 nm were well dispersed onto the TiO₂ nanoparticles. The FS-synthesized Au/TiO₂ showed its highly active photocatalytic activities for H₂ generation from pure water and methanol solution. The H₂ production rate from water and methanol solution was 52 and 7890 μmol·h⁻¹·g⁻¹, respectively. These production rates were superior to those obtained from P25/Au, which was prepared from solution method as a comparison sample [103]. By using a similar process for the single-step FS, Au-loaded TiO₂ was reported as an active catalyst for H₂ sensor application [131].

In addition to single loading, the co-loading of Au with other metals was also made possible using FS. Pongthawornsakun et al. studied various feeding methods to prepare Pd-Au-loaded TiO₂ nanoparticles by mix feeding and separate feeding with different compositions, such as PdAu/TiO₂, Pd/TiO₂ + Au, Au/TiO₂ + Pd, TiO₂ + AuPd and Au/TiO₂ + Pd/TiO₂ [104]. It was found that there was no significant difference in the TiO₂ structure of all samples in terms of the sizes and surface areas using cofeeding methods. In the prepared nanoparticles, the metallic nanoparticles were well distributed onto the TiO₂ supports. Among the feeding methods, the TiO₂ + AuPd composite catalyst produced by the two-nozzle FSP showed the highest photocatalytic activities of acetylene conversion with 50% efficiency and an ethylene selectivity greater than 95%. The possibility of bimetallic particles can be reduced by separate feeding. It is important that the cofeeding of metal precursor and Ti precursor produced a catalyst with a lower activity due to the coverage of the active metals by the Ti-O species [104]. Recently, Jiang et al. successfully prepared AuPt alloy cocatalyst-loaded TiO₂ by mixing chloroplatinic acid hexahydrate and chloroauric acid as the Pt and Au precursors, respectively [105]. The formation mechanism of AuPd-loaded TiO₂ particles and its nanostructures are presented in Figure 7a,e. The synergetic effects between Au and Pt reduced the agglomeration of metal particles, and CO poisoning was reduced. As a result, AuPt/TiO₂ could enhance catalytic CO oxidation by 20% at a low-temperature reaction of 100 °C in comparison to single-metal loading.

4.2. TiO₂/Non-Noble Metals/Oxides

Due to the high cost of noble metals, non-noble metals, such as Cu, Co and Ni, were alternatively used as cocatalysts to enhance the activities of TiO₂ [77,106,107,132,133,134]. In FS, these non-noble metals were loaded onto the TiO₂ via a single feed or a separated feed. Chen et al. used coflow FS to produce CuO/TiO₂ photocatalysts by mixing metal precursors composed of Cu(NO₃)₂•3H₂O, ethanol and TBT. This mixture was injected into the burner of a CH₄/O₂ flame by a syringe pump [77]. Contrary to noble metals, such as Pt or Au, the addition of a Cu precursor led to changes in the crystal phases of TiO₂. The surface area of TiO₂ powder was reduced with the increase in the copper contents; interestingly, the specific surface area (SSA) was reduced from 98.96 to 70.07 m²·g⁻¹ when the amount of Cu precursor was increased from 2% to 20%. However, the XRD data revealed that crystal size increased as the Cu content became higher. The rutile phase increased from 27.6% to 56.2% with the increase in the Cu content from 2% to 20%, respectively. Due to the fact of its small radius and lower valent state than Ti⁴⁺, Cu²⁺ preferentially doped into TiO₂ lattices and created oxygen vacancies by reducing the Cu²⁺. Since defective TiO_x species act as an important nucleation center, the percentage of the rutile phase increased as the Cu²⁺ doping content increased. However, at high concentrations, Cu could no longer dope into TiO₂. The excessive Cu²⁺ aggregated on the surface of the TiO₂ and formed CuO nanoparticles on the surface of the TiO₂. In the study, 20 wt.% CuO loading was the optimum composition for the combustion of lean CO, with a 95% conversion efficiency at 120 °C reaction [77]. Xiong et al. studied the synthesis of Cu/TiO₂ using FS and their CO₂ photoreaction under UV light [106]. They found that at a low Cu concentration, Cu⁺ was rich and well dispersed into TiO₂ crystals. The maximum CO and CH₄ yields were 43.5 and 16.7 μmol g⁻¹, respectively, after a 4 h reaction. It was believed that the Cu⁺ ions doped into the TiO₂ enhanced the charge separation and CO₂ absorption. Therefore, the Cu⁺ ions incorporated into TiO₂ lattices facilitated the CO₂ activation, which was considered as an active site for CO₂ photoreduction [106].

Another work by Zhao et al. studied codoping on Pt-loaded TiO₂ by FS using cofeeding methods under a H₂/O₂ flame [107]. A small amount (1 wt.%) of Co(NO₃)₂•6H₂O added resulted in elemental doping into TiO₂ crystal. Therefore, mixing Co, Pt and Ti precursors produced Pt-nanoparticle-loaded codoped TiO₂ heterostructures. The codoping did not affect the particle sizes of TiO₂ supports, but reduced the Pt-particle’s sizes from 2.47 nm to 0.72 nm with 1 wt.% codoping. With a high-temperature flame, Co²⁺ was oxidized to Co³⁺ and doped into the surface of the TiO₂. Doping with metals with different valence states induced more oxygen vacancies and suppressed the aggregation of Pt sub-nanoclusters. The catalytic activities of Pt/codoped TiO₂ for CO oxidation were much improved due to the rich oxygen absorbed on the surface of the defected codoped TiO₂. One hundred percent CO conversion was achieved by the Pt/codoped TiO₂ at a temperature of 70 °C, while it was achieved at 120 °C by Pt/TiO₂ [107].

4.3. TiO₂/Others

In addition to the above materials, FS was used for the synthesis of other composite structures with TiO₂, such as SnO₂/TiO₂ [135], Ga₂O₃/TiO₂ [136], MgO/TiO₂ [137], SiO₂/TiO₂ [138,139], Ni-CeO_x-TiO₂ [140], WO₃/CeO_x/TiO₂ [78,141], Cr-Si-TiO_x [87], PbS/TiO₂ [142] and carbon/TiO₂ [96,143].

5. Summary and Outlooks

Flame synthesis has been considered as a potential method for the synthesis of nanoparticles, ranging from the lab scale to the industrial scale. Nanopowders prepared from flame synthesis usually exhibit unprecedented physicochemical properties in comparison to conventional synthetic methods. In addition, the flame synthesis can be continuous and fast, and changes in the various parameters may result in different product properties, which affect the catalytic performance. The direct conversion of TiO₂ particles from titanium precursors without any separation process or post-thermal annealing allows this method to produce TiO₂ products at a low cost in comparison to wet chemical methods. The properties of TiO₂ nanoparticles were controllable based on simple operational parameters, such as precursors and feeding types, precursor concentrations and compositions, gas flow rates and gas compositions, reaction atmospheres, and reaction temperatures. So far, TiO₂ nanoparticles with various morphologies, including spherical and particulate forms, were synthesized, and their crystal phases of TiO₂ (e.g., anatase and rutile) could be easily tuned. Advanced structures have become current trends in flame synthesis to improve the properties of TiO₂ nanoparticles according to the target applications. Typically, to deliver TiO₂ nanoparticles as a promising candidate for a sustainable world, TiO₂ must effectively work under solar light, which is an abundant and clean energy on Earth. Defect and bandgap engineering, such as doping and heterostructures, are demonstrated as powerful ways to improve the activities of TiO₂. Primary works have already reported the use of flame synthesis for defective and heterostructured TiO₂ nanoparticles. However, scaling up the reaction is vital in order to demonstrate the feasibility for industrial-scale synthesis.

There is still room for the development of high-quality defective and heterostructured TiO₂ nanoparticles:

(1)

For achieving the industrial-scale production of flame-synthesized powder with well-defined characteristics, a deep understanding of the evaporation characteristics of the precursor solution, particle nucleation and growth, fluid–particle dynamics, etc., during flame synthesis is required for the design of reactors.

(2)

Since flame synthesis involves the instantaneous evolution of powder, the degree of defects and the amount of dopants introduced in each TiO₂ nanoparticle can vary slightly. Therefore, more research on homogenizing the particles when prepared at a larger scale should be conducted.

(3)

In terms of energy and environment, methods should be developed to clean or recycle gas that arises during the synthesis steps. In addition, efficient flame reactors should be designed in order to achieve low-energy consumption.

(4)

The diversification of precursors for obtaining heterostructured TiO₂ nanoparticles is necessary. There could still be more elements that can be introduced for the synthesis of heterostructured TiO₂ particles.