High-Pressure CO₂ Photoreduction, Flame Spray Pyrolysis and Type-II Heterojunctions: A Promising Synergy

Abstract

In this work, three catalysts, TiO₂, WO₃ and TiO₂/WO₃, have been synthesized through flame spray pyrolysis synthesis (FSP) and have been tested for CO₂ photoreduction. The catalysts were fully characterized by XRD, DRS UV–Vis, N₂ physisorption and SEM. Experimental tests were performed in a one-of-a-kind high-pressure reactor at 18 bar. TiO₂ P25 was used as a benchmark to compare the productivities of the newly synthetized catalysts. The two single oxides showed comparable productivities, both slightly lower than the P25 reference value (ca. 17 mol/kg_cat·h). The mixed oxide, TiO₂/WO₃, instead showed an impressive productivity of formic acid with 36 mol/kg_cat·h, which is around 2.5 times higher than both of the single oxides alone. The formation of a type-II heterojunction has been confirmed through DRS analysis. The remarkable productivity demonstrates how FSP synthesis can be a crucial tool to obtain highly active and stable photocatalysts. This approach has already been successfully scaled up for the industrial production of various catalysts, showcasing its versatility and efficiency.

1. Introduction

Energy is indispensable, fulfilling fundamental needs and sustaining a comfortable and enhanced standard of living [1]. Over recent decades, energy consumption has escalated due to global population expansion and uncontrolled industrialization [2]. This has led to significant consumption of fossil fuels, resulting in energy shortages and elevated CO₂ emissions. Emissions of CO₂ and other greenhouse gases from carbon-based energy sources have caused numerous environmental and social challenges. CO₂ emissions are responsible for more than 60% of global warming, with annual emissions exceeding 30 gigatons, primarily from fossil fuels [3]. The adoption of renewable energy is essential for reducing CO₂ emissions, alleviating climate change and preserving a clean environment through sustainable energy solutions. Various strategies have been developed to mitigate the impact of CO₂, such as CO₂ capture, storage and reduction and its conversion into value-added chemicals. Among these, the conversion of CO₂ into value-added chemicals is particularly appealing, as it also offers the potential to decrease fossil fuel consumption. CO₂ utilization can be accomplished through several approaches, including chemical conversions, electrochemical conversions, biological conversions and reforming processes. However, each of these technologies faces some challenges, including requirements for high temperatures or electrical voltages to break the stable CO₂ molecule, limitations in raw materials, high operational costs and issues related to sustainability [4]. Given these factors, one of the most promising approaches for the reutilization of CO₂ is artificial photosynthesis. Despite the intermittency and variability in its intensity, sunlight is an inexhaustible energy source that is freely accessible. Natural photosynthesis is a process that transforms atmospheric CO₂ into sugar-based biomass. Artificial photosynthesis emulates this process as a model to attempt the storing of solar energy in chemical form. The generation of solar fuels under sunlight can be achieved through photocatalysis. In heterogeneous photosynthetic processes, solid semiconductors are employed, activated by the absorption of photons with energy exceeding the bandgap [5]. When irradiation reaches the semiconductor surface, an electron–hole pair is generated; the excited electron moves from the valence band to the conduction band, leaving a positive hole that facilitates oxidation and reduction with the redox pairs present in the reaction medium. The efficiency of the reaction is negatively impacted by the possibility of electron–hole recombination. In recent years, a variety of materials have been adopted as efficient and environmentally sustainable photocatalysts for CO₂ photoreduction, including TiO₂, WO₃, ZnO, CdS, CuO and g-C₃N₄. Among these, TiO₂ stands out as one of the most promising due to its abundance, high chemical stability, low price and robust catalytic properties [6,7,8,9,10,11]. However, it has a limitation due to its wide bandgap of approximately 3.2 eV, making it responsive only to ultraviolet (UV) light, which constitutes approximately 3% of the solar spectrum [12,13]. Over recent decades, numerous studies have been published aiming to reduce the bandgap, enhance the light absorption and decrease the charge transfer and recombination rates of TiO₂ by coupling it with other semiconductors that have narrower bandgap values, such as WO₃, CdS, MoO₂ [8,14,15,16,17,18]. WO₃ has received considerable attention due to its relatively narrow bandgap of approximately 2.6–2.8 eV, non-toxicity, resistance to photo-corrosion, chemical stability and capability to exist in a wide range of oxidation states (2+, 3+, 4+, 5+ and 6+), which facilitates the storage of photogenerated electrons [19].

It is challenging to meet all of the needs, using only a semiconductor, for photocatalysis, such as having a wide visible-light response, long-term stability and a strong redox capacity, as well as preventing fast electron–hole recombination. A beneficial means that can be applied to enhance the properties of semiconductors is the type-II heterojunction, shown in Figure 1 [20]. The latter is formed when two semiconductors capable of absorbing light with different energy bands are put in contact. In this arrangement, the valence band (VB) of photocatalyst-I (PC-I) is more positive than the VB of PC-II, while the conduction band (CB) of PC-II is more negative compared to the CB of PC-I, as reported in Figure 1. When light irradiates the semiconductor surfaces, each generates an excited electron that transitions from the VB to the CB. In type-II heterojunctions, photo-promoted electrons tend to accumulate in the material with a more positive CB potential. On the other hand, holes tend to accumulate in the semiconductor with a more negative VB [21]. This kind of heterojunction differs from the so-called Z-scheme ones, where after two excited electrons and holes are generated, the electrons from the lower conduction band can recombine with the holes in the highest energy band, following the Z-scheme and making the potential differences between the holes and electrons higher than each of the materials alone [20,22].

Flame spray pyrolysis (FSP) is a promising route, but other well-established techniques have been widely explored in the literature to synthetize type-II heterojunctions with tunable properties as well. For the sake of completeness, some papers on different techniques that appeared of interest to the authors are given below. Lei et al. reported the synthesis of Zn_xIn_yS_z by the hydrothermal method, and the relative Zn_xIn_yS_z/ZnO composites with internal type-II heterojunctions prepared by a solvent-assisted interface, highlighting how the latter significantly enhanced the separation of the photocarriers [23]. Lately, electrospinning has gained attention as a method for fabricating one-dimensional (1D) nanostructured photocatalysts. This technique allows for the formation of nanofibers with high ratios, as well with large surface areas, for catalytic reactions. Experiments performed on TiO₂/CdS nanofiber heterojunctions have reported significant improvements as H₂ production photocatalysts, due to the formation of type-II heterojunctions that facilitate charge separation [24].

Despite the higher redox potential of the Z-scheme, type-II heterojunctions are able to improve photocatalytic efficiency when two materials are placed in intimate contact. It is well known that TiO₂’s performance as a photocatalyst is highly affected by its extremely short e lifetime [25]. In this sense, the WO₃/TiO₂ type-II heterojunction can play a crucial role in enhancing charge separation and supporting CO₂ reduction. De Castro et al. investigated WO₃/TiO₂ heterostructures towards organic contaminant degradation, reporting a particular improvement in the catalytic activity of the material for the 40 wt% WO₃/TiO₂ sample due to the larger interface and contact between the two phases [26]. In this study, we present a novel TiO₂/WO₃ hybrid material, synthetized though the flame spray pyrolysis (FSP) technique, that has been characterized in-depth and tested for CO₂ photoreduction in a high-pressure photoreactor. The FSP technique allows us to place different materials in intimate contact as a strategy to favor the electron–hole transfer, improving the performances of the single photocatalysts.

2. Results and Discussion

2.1. Material Characterization

In the WO₃ sample (Figure 2a,b), a petal-like structure was observed. The latter was due to the non-isotropic growth of the WO₃ crystals, where certain crystal facets grew faster than others, leading to the formation of these morphologies. The whole surface of the sample resulted in small spherical particles of WO₃. In contrast, on the TiO₂ FSP sample (Figure 2e,f), this petal-like structure was not observed, but instead, a structure characterized by a well-defined porous network and particle aggregation was noted. Fewer spherical TiO₂ particles can be observed on the surface of the sample. On the WO₃/TiO₂ hybrid material (Figure 2c,d), as reported in Figure 2d, it can be seen how the petal-like structure of the WO₃ and the more classical porous structure due to the TiO₂ joined with each other. The non-isotropic growth was still present, but to a lower extent. The WO₃ deposited on top of the titania in the hybrid sample did not adversely affect the surface area of the latter, as reported in

Table 1. Characterization data for tested catalysts.

Sample	BET SSA (m2/g)	Total Pore Volume (cm3/g)	BJH ads. Pore Width (nm)	Crystallite Size (nm)	Phase %	BG (eV)
P25	52.7	0.225	18.7	15(A); 26(R)	78(A); 22(R)	3.22
FSP-TiO2	29.7	0.109	14.5	20(A); 24(R)	48(A); 52(R)	3.10
FSP-WO3	3.4	0.015	35.0	42	/	2.73
TiO2/WO3 60/40 wt	29.2	0.077	9.8	22(A); 16(R)	45(A); 65(R)	2.7

, allowing for excellent contact between the phases, as shown by the DRS analysis.

2.2. XRD Analysis

The XRD diffractograms are reported in Figure 3. In both the FSP-TiO₂ and FSP- TiO₂/WO₃ 60/40 samples, one can observe the main reflections at 25.3° (2θ) and 27.5° (2θ) due to the anatase (JCPDS 21-1272—anatase) and rutile (JCPDS 21-1276—rutile) phases, respectively. The FSP-WO₃ diffractogram showed correspondence with the monoclinic WO₃ across all ranges of measurement. These results are in accordance with the fact that this phase is the most stable at room temperature. Furthermore, authors have reported that WO₃ is generally unable to retain alternative phases such as β-WO₃ (orthorhombic) and α-WO₃ (tetragonal) that are more stable at high temperatures [27]. The main reflection of the WO₃ phase was still visible at 23° (2θ) in the FSP-TiO₂/WO₃ 60/40, though much less crystalline and possibly dispersed into the titania phase, whose crystallinity was also decreased with respect to the bare titania sample. The crystallite sizes of the different catalysts are reported in Table 1, together with the calculated anatase/rutile compositions of the titania-containing compounds. The TiO₂ prepared through FSP showed a significant difference with commercial P25 regarding the anatase/rutile composition. The FSP sample showed an anatase/rutile ratio of 48/52, compared to the 78/22 ratio of the commercial P25 sample. As reported by some authors, a higher rutile content can improve charge separation due to the formation of a homojunction with anatase [28].

2.3. N₂ Physisorption Analysis

The results from the N₂ physisorption are reported in Figure 4. All of the samples displayed a type-IV isotherm. The BET SSA, total pore volume and BJH adsorption pore width results obtained from the N₂ physisorption curve elaboration are reported in Table 1. The BET SSA area of the FSP-TiO₂ was one order of magnitude higher than that of the pure WO₃, with values of 29.7 and 3.4 m²/g, respectively. The commercial P25 showed a higher surface area of 52.7 m²/g. The pore volume and pore width also varied greatly between the WO₃ and TiO₂, with the latter showing roughly half of the total pore volume compared to the sample of P25 (0.109 cm³/g vs. 0.225 cm³/g). The composite catalyst showed slight decreases in the SSA and the total pore volume compared to the FSP-TiO₂. The BJH calculated pore width displayed was also lower (9.8 nm).

2.4. DRS UV–Vis Analysis

Both the DR UV–Vis spectra of all of the samples and the Tauc plot elaboration are shown in Figure 5. The calculation of the bandgap was obtained starting from the Tauc equation and implementing the Kubelka–Munk function. Both TiO₂ and WO₃, as explained, are semiconductors with an indirect transition. The values of the bandgaps can be obtained from the graph, plotting (F(R∞)hν)^γ against the energy. The results obtained are reported in Table 1. The P25 and FSP-TiO₂ bandgaps were estimated, respectively, as 3.22 and 3.10 eV. The titanium dioxide and tungsten trioxide bandgaps obtained by FSP were in agreement with the literature, which attributes a value of 3.2 eV for the former [29] and 2.7 eV for the latter [30]. The bandgap of the TiO₂ FSP sample was slightly lower compared to that of the P25, which can be attributed to the higher rutile content. Anatase and rutile have bandgaps of, respectively, 3.2 eV and 3.0 eV [31], and the TiO₂ FSP bandgap of 3.10 eV is in line with the anatase/rutile composition that was calculated through XRD. All of these characteristics align with the higher flame temperature and longer residence time, which enhance the crystallinities of both the anatase and rutile phases while reducing defects. This is advantageous in photocatalysis, as a lower defect density improves charge separation, reduces recombination and, combined with a high surface area for reactant adsorption and efficient light exposure, enhances the photocatalytic performance.

The bandgap of the mixed oxide is interesting because its value, 2.70 eV, was lower than that of both of the pure oxides synthetized, and resembles that of the pure WO₃. This behavior is typical of type-II heterojunctions, where a small reduction in the measured bandgap is observed by DRS. This means that better sunlight harvesting can be exploited by this substance: a lower bandgap can be translated into a wider range of wavelengths capable of being absorbed by the substance considered. Being that the sunlight spectrum shifted to visible light, rather than UV radiation, this improvement is relevant for photocatalyst application in real sunlight.

2.5. Photoreduction of CO₂

Tests were performed at a pH of 14, 18 bar and 6 h of reaction time, with 1.66 g/L of sodium sulfite as a hole scavenger (HS) and 31 mg/L of catalyst concentration on the prepared catalysts, as previously optimized. The productivity results and HS conversions are reported in Figure 6. Starting from the productivity, the total stored energy was calculated considering the (Lower Heating Value) values of the primary reduction products [32], using Equations (1) and (2). Both the productivity and the calculated total stored energy are reported in

Table 2. Productivity and stored energy of tested catalysts—18 bar—pH 14.

Catalyst	HS Conv. (%)	HCOOH (mol/kgcat h)	HCOOH (g/kgcat h)	Total Stored Energy (J) in 6 h
P25	47	16.9	778	195
FSP TiO2	40	13.9	640	160
FSP WO3	43	14.7	677	169
FSP TiO2/WO3 60/40	77	36.5	1680	406

.

$$n_i\left(\mathrm{m}\mathrm{o}\mathrm{l}\right)={\mathrm{p}\mathrm{r}\mathrm{o}\mathrm{d}\mathrm{u}\mathrm{c}\mathrm{t}\mathrm{i}\mathrm{v}\mathrm{i}\mathrm{t}\mathrm{y}}_i\left({\displaystyle \frac{\mathrm{m}\mathrm{o}\mathrm{l}}{{\mathrm{k}\mathrm{g}}_{\mathrm{c}\mathrm{a}\mathrm{t}}\times \mathrm{h}}}\right)\times t\left(\mathrm{h}\right){\times m}_{cat}\left(\mathrm{k}\mathrm{g}\right)$$

(1)

$$\mathrm{T}\mathrm{o}\mathrm{t}\mathrm{a}\mathrm{l} \mathrm{s}\mathrm{t}\mathrm{o}\mathrm{r}\mathrm{e}\mathrm{d} \mathrm{e}\mathrm{n}\mathrm{e}\mathrm{r}\mathrm{g}\mathrm{y}\left(\mathrm{J}\right)=\sum \left(n_i\left(\mathrm{m}\mathrm{o}\mathrm{l}\right){\times \mathrm{L}\mathrm{H}\mathrm{V}}_i\left({\displaystyle \frac{\mathrm{J}}{\mathrm{m}\mathrm{o}\mathrm{l}}}\right)\right)$$

(2)

The performance of our benchmark, namely, a commercial TiO₂ P25 produced by Evonik (formerly Degussa) and supplied by EIGENMANN & VERONELLI S.p.A, was ca. 15% better than that of the titania sample prepared by FSP. This can be attributed to the combined effects of the higher anatase content and higher surface area of the P25 sample. The performances of the two bare oxides prepared by FSP were comparable, with the FSP titania showing a slightly lower productivity of HCOOH (13.9 mol/kg_cat·h) compared to that of the FSP WO₃ (14.7 mol/kg_cat·h). This result is attributed to the lower bandgap of the WO₃, which suggests a better utilization capacity of the absorbed light when provided the use of the same UV lamp. The WO₃/TiO₂ 40/60 showed a two-fold increase in HCOOH productivity compared to the P25 with an astonishing value of 36.5 mol/kg_cat·h, among the highest reported in the literature, as reported in Table 3.

The hole scavenger conversion followed the productivity of the different catalysts, with the mixed oxide reporting a conversion of 77% compared to the 43% of the WO₃ and the 40% of the TiO₂. The greater HS conversion is in line with the higher productivity recorded. It should be underlined that HS conversion should be maintained well below 90 percent to avoid the photoreforming of the obtained products [33].

As shown by the DRS measurements, the TiO₂ FSP sample showed a slightly lower bandgap compared to the P25, which can be attributed to the higher rutile content. When TiO₂ and WO₃ are put in intimate contact, such as through FSP synthesis, a type-II heterojunction is formed, as confirmed by the DRS measurements. The latter showed a bandgap slightly inferior even to that of the pure WO₃. In type-II heterojunctions, the charge recombination is reduced by spatially separating the photogenerated electrons and holes, with the electrons migrating to the lower conduction band (WO₃) and the holes to the higher valence band (TiO₂). In this case, a small reduction in the measured bandgap was observed, as reported in Figure 1. Some authors have also reported that the greater acidic character of WO₃ should interact more with the carbonate ions present in the solution, due to its greater acidity [34].

As previously mentioned, CO₂ photoreduction is now a well-known and growing field. The photocatalytic reduction of CO₂ to formic acid represents a small fraction of the research carried out, despite the high added value of the formic acid produced. Recently, an interesting review was published on this topic [35]. A quick overview of the relevant productivities recorded in the literature is shown in Table 3.

Table 3. Literature comparison—catalyst, reaction medium, light source and productivity.

Catalyst	Reaction Medium	Light Source	Productivity (mol/kgcat h)	Reference
Cu2O	Aqueous suspension in 0.3 M NaHCO3, 2 M glycerol as HS	Arc Lamp 1000 W Xenon	2.76	[36]
Ag/TiO2	Aqueous suspension in 0.2 M NaHCO3	Arc Lamp 1000 W Xenon	3.89	[37]
Co3O4	Aqueous suspension in water	LED lamp 21 W, λ = 510	HCOOH 4.53 (mmol/kgcat·h) HCHO 0.62 (mmol/kgcat·h)	[38]
ZnO-Cu2O NPs	Aqueous suspension in water and ZnO-Cu2O 300 W Xe lamp 0.2 M Na2CO3	300 W Xe lamp	CH4 1080 (mmol/kgcat·h) CO 1.4 (mmol/kgcat·h)	[39]
TiO2 FSP	Aqueous suspension in water, HS 1.66 g/L of Na2SO3	Medium Hg-pressure UV lamp 250 W	13.9	This work
WO3/TiO2 (40/60) FSP	Aqueous suspension in water, HS 1.66 g/L of Na2SO3	Medium Hg-pressure UV lamp 250 W	36.5	This work

Table 3. Literature comparison—catalyst, reaction medium, light source and productivity.

Catalyst	Reaction Medium	Light Source	Productivity (mol/kgcat h)	Reference
Cu2O	Aqueous suspension in 0.3 M NaHCO3, 2 M glycerol as HS	Arc Lamp 1000 W Xenon	2.76	[36]
Ag/TiO2	Aqueous suspension in 0.2 M NaHCO3	Arc Lamp 1000 W Xenon	3.89	[37]
Co3O4	Aqueous suspension in water	LED lamp 21 W, λ = 510	HCOOH 4.53 (mmol/kgcat·h) HCHO 0.62 (mmol/kgcat·h)	[38]
ZnO-Cu2O NPs	Aqueous suspension in water and ZnO-Cu2O 300 W Xe lamp 0.2 M Na2CO3	300 W Xe lamp	CH4 1080 (mmol/kgcat·h) CO 1.4 (mmol/kgcat·h)	[39]
TiO2 FSP	Aqueous suspension in water, HS 1.66 g/L of Na2SO3	Medium Hg-pressure UV lamp 250 W	13.9	This work
WO3/TiO2 (40/60) FSP	Aqueous suspension in water, HS 1.66 g/L of Na2SO3	Medium Hg-pressure UV lamp 250 W	36.5	This work

The performances of both of the bare oxides prepared by FSP were comparable, with high productivity compared to what has been reported in the literature, although it is worth noticing that different conditions, such as lamps and hole scavengers, were used. Authors have reported a Ag/TiO₂ catalyst which showed a remarkable productivity of 3.89 mol/kg_cat·h when tested in an aqueous suspension (0.2 M NaHCO₃) under irradiation by a 1000 W Xenon arc lamp [37]. As previously said, the hybrid oxide being prepared through FSP synthesis allowed it to obtain an astonishing productivity of 36.5 mol/kg_cat·h, emphasizing that FSP represents a fundamental tool for the preparation of oxide-based catalysts, and all the more so in the field of photocatalysis, where the intimate contact obtained through this technique allows for simple and fast preparations of type-II heterojunctions.

3. Materials and Methods

3.1. Materials Used

Commercial TiO₂ P25, produced by Evonik (formerly Degussa; Essen, Germany) and supplied by EIGENMANN & VERONELLI S.p.A. (Rho, Italy), was used as a benchmark. Carbon dioxide (CO₂), methane (CH₄), oxygen (O₂) and helium (He), all with a purity >99.99%, were purchased from AirLiquide (Milano, Italy). Ammonium metatungstate hydrate (AMT) (Sigma Aldrich, St. Louis, MO, USA, >99.99% purity) and titanium tetraisopropoxide (TTIP) (Sigma Aldrich, >97% purity) were used as precursors for the preparation of the metal oxides through flame spray pyrolysis. Dimetilformamide (DMF) (Sigma Aldrich, >99.9% purity) was used as a solvent for the preparation of the precursor solution. Sodium sulfite (Na₂SO₃, Sigma Aldrich, >98% purity) was used as a hole scavenger (HS) during the photocatalytic test. NaOH (Sigma Aldrich, 98% purity) was used to achieve a basic pH (pH = 14) during the experiments. The operating conditions, including the use of a basic pH, were optimized in previous work to maximize the catalyst productivity [40].

3.2. Flame Spray Pyrolysis

Flame spray pyrolysis, often called FSP, is a method that allows us to obtain single or mixed metal oxides which are thermally stable, with a high degree of purity and an interesting specific surface area. The metal or metals whose oxide is to be obtained are contained in organometallic salts, called precursors. The precursors are dissolved in a combustible organic solvent, and the solution thus obtained is fed by means of a syringe pump through a needle in a vertical nozzle placed in the center of the burner. A scheme of the FSP apparatus is reported in Figure 7. The end of the capillary tube is coaxially lapped by a flow of O₂, which acts as a comburent and at the same time has the function of dispersing the solution into very fine droplets. The heterogeneous O₂/solution mixture is ignited by twelve CH₄/O₂ flames, arranged in a cone around the nozzle. The instantaneous dispersion and vaporization of the solution droplets and the decomposition (pyrolysis) of the precursor take place in the flame, giving rise to the first oxide nuclei, which increase by coalescence and condensation. The primary nanoparticles thus obtained aggregate in the final part of the flame, forming the desired powders. The powders are collected from the dust collection glass chamber on which they are deposited. The size of the formed particles depends on the temperature of the flame, on the residence time and on the concentration of particles in the flame. When the values of these parameters increase, the probability that sintering phenomena will occur increases as well, leading to the formation of larger particles with lower surface areas. The FSP process is affected by many parameters, including the nature of the solvent, the concentration and flowrate of the precursor solution, the flow of the oxygen and methane and the overpressure of the oxygen used for the dispersion of the solution. The latter can play a significant role in the final surface area of the catalyst, and for this reason, a manometer is used to constantly control this value.

FSP can be defined as one of the so-called non-conventional synthesis techniques. Several advantages over traditional synthesis methods like thermal annealing, sol-gel and wet impregnation are present, but it also comes with some challenges. One of its main strengths is the ability to produce nanoparticles in a single step, without the need for multiple drying and calcination stages. This translates as well to better control over the particle size and dispersion. The FSP process is highly scalable, making it suitable for industrial applications as well, where large amounts of material need to be synthetized. The main downsides for this technology when compared to more conventional ones such as thermal annealing include the use of specialized equipment, including a high-temperature flame and oxygen supply, which cannot be carried out with conventional lab equipment. In comparison, thermal annealing is the most straightforward method, but typically results in larger particles and lower surface areas, which can negatively impact catalytic performance. The powder collection efficiency varies greatly based on the filters and collection systems that are put in place. In the case of setups such as the one used in the present work, the dust collection is about 25% of the theoretical maximum amount. This occurs because only the powder deposited on the glass wall of the collection chamber is collected. When considering the synthesis of catalysts on a pilot or industrial scale, the powder collection efficiency improves greatly due to the combination of baghouse filters and HEPA porous stainless steel filters, reaching values close to 100%.

Figure 7. FSP schematic representation. Adapted from [41].

Figure 7. FSP schematic representation. Adapted from [41].

3.3. Catalyst Preparation

All of the catalysts tested in this work, except for the commercial P25 that was used as a benchmark, were synthesized via FSP. In order to synthetize the pure WO₃ and TiO₂ and the mixed oxide WO₃/TiO₂, ammonium metatungstate hydrate (AMT) and titanium tetraisopropoxide (TTIP) were used as precursors, respectively, for WO₃ and TiO₂. AMT is a water-soluble compound which cannot be dissolved in alcohols. On the other side, TTIP is strongly hygroscopic, and requires the use of an anhydrous organic solvent to avoid hydrolysis directly in the flask. DMF (dimetilformamide) was used as a solvent, since it is an anhydrous organic solvent able to solubilize AMT and due to its boiling point (153 °C). Both a low-boiling and a very high-boiling solvent must be avoided to prevent, respectively, the sintering of the produced particles or the formation of egg-shell phenomena, which lead to the formation of hollow particles. Solutions with a concentration of the organic precursor equal to 3% wt. compared to the solvent were used for injection in the FSP apparatus. This concentration was taken as a reference from a previous work [40] to avoid the vaporization of the solvent and the precursor directly into the tip of the needle, which can lead to clogging and to a twisted flame. Furthermore, increasing the concentration above a certain value again incurs the phenomenon of particle sintering, due to the higher probability of collision of the formed particles. All of the instrumentation was well dried in the oven to eliminate any moisture. The prepared solutions were injected into the FSP apparatus with a flow of 2.7 mL/min. The flowrates for the methane, external and the internal oxygen flows were, respectively, 0.5 L/min, 1 L/min and 5 L/min. During all of our syntheses, an overpressure of 1.5 bar of oxygen used for dispersion was maintained. After synthesis, all of the powder obtained was collected from the wall of the glass chamber. The composite catalyst obtained had final mass and molar ratios of the two oxides equal, respectively, to TiO₂/WO₃ = 60/40 (wt/wt) and TiO₂/WO₃ = 81.3/18.7 (mol/mol).

3.4. Catalyst Characterization

Images from the scanning electron microscopy (SEM) were captured using a Hitachi TM 1000 tabletop scanning electron microscope (Marunouchi, Japan). Carbon tape was used as a substrate for the SEM analysis to prevent charging effects without requiring metallization.

The X-ray diffraction (XRD) analyses were performed with a Rigaku Miniflex-600 horizontal-scan powder diffractometer (Tokyo, Japan) using Cu-Kα radiation, with a graphite monochromator on the diffracted beam. Scherrer’s Equation (3) was used to calculate the size of the crystallites.

D = (K∙λ)/(β∙cosθ)

(3)

where D is the crystal size, λ is the X-ray wavelength (0.154 nm with the Cu Kα generator), K is the shape factor (0.9), β is the width at the half maximum of the peak (i.e., FWHM) and θ is the Bragg angle.

The N₂ physisorption analyses were performed with an ASAP 2020 instrument by Micromeritics (Norcross, GA, USA). The pretreatment before the actual analysis included the degassing of the samples, performed under a vacuum at 150 °C for 4 h. This pretreatment allowed us to remove any contaminants or traces of water present in the catalysts. This non-destructive analysis was performed on 200 mg of nano powder at a temperature of −196 °C. The data obtained were processed following the BET theory, BJH theory and t-plot analysis method.

The diffuse reflectance (DR) UV–Vis spectra of the catalysts were recorded on a Shimadzu UV−3600 Plus (Kyoto, Japan) in the range of 200–800 nm, using an integrating sphere and BaSO₄ as reference standards. Using the reflectance spectra as the input data, a Kubelka–Munk elaboration was performed, according to Equation (4) [42].

F(R∞) = (1 − R∞)²/2R_∞

(4)

The calculated (F(R)hν)1/r (with r = 2 or ½ for the direct and indirect bandgaps) was plotted versus hν to obtain the bandgap of each sample [43]. To prevent the saturation of the samples that absorb too much, all of the photocatalysts were diluted one-to-one or one-to-two with BaSO₄, which was the reflectance standard. The shapes of the curves were not affected by blending with the BaSO₄, so a comparison of the absolute value of the reflectance could not be performed. The final form of the equation is reported in Equation (5).

(F(R∞)hν)^γ = A(hν − E_g)

(5)

with the parameter F(R∞) corresponding to the Kubelka–Munk function, A being a constant, E_g being the bandgap energy of the semiconductor and hν being the energy of the incident radiation calculated using Planck’s constant and the frequency of the radiation itself. The exponent γ can assume two values: 2 or ½. In the specific case of the synthetized FSP catalysts, the value of γ considered was ½, because both TiO₂ and WO₃ (monoclinic) are semiconductors with an indirect transition [44].

3.5. Photoreduction Setup

The tests with the synthetized catalysts were performed in a special high-pressure photoreactor previously developed by our group and described elsewhere [45]. This reactor allows us to perform experiments in the liquid phase up to 20 bar and 100 °C. An AISI 316 stainless steel jacketed reactor (cylinder shaped) with an internal capacity of 1.7 L was used. A quartz sleeve able to withstand pressures up to more than 20 bar was placed in the middle of the reactor. Housed inside the quartz jacket was the UV lamp, consisting of two medium Hg-pressure UV bulbs, each one with a power of 125 W. The lamp irradiation was in the range of 254–365 nm, with the peak emission located at 365 nm. The irradiance was checked periodically using a photoradiometer (Delta OHM HD2102.2 equipped with a LP 471A-UVeff probe, producer Senseca Italy Srl (ex Delta OHM Srl), Caselle di Selvazzano (PD), Italy), active in the range of 250–400 nm and resulting in an average of 120 W/m². Sodium sulfite (Na₂SO₃) was used as a hole scavenger (HS). To test the catalysts, 1.2 L of water were loaded into the reactor, together with 1.66 g/L of Na₂SO₃ and 31 mg/L of the catalyst dispersed in it [25,46,47,48]. All of the photocatalytic tests were performed at a basic pH (pH = 14), 80 °C and 18 bar under intense and constant stirring to ensure the homogeneous dispersion of the catalyst. NaOH was used to obtain a basic environment. Each experiment lasted 6 h.

The gaseous products were analyzed with a GC Agilent 7890A (producer Agilent Technologies, Santa Clara, CA, USA) equipped with two different columns, HP PlotQ and MS, respectively, used for the analyses of the gaseous product and permanent gases. The GC analysis was performed at 70 °C, He was used as a carrier gas and the quantification was performed by means of a TCD. The two columns were placed in a series, with the second (MS) being able to be bypassed and sent into isolation. At the beginning of the analysis, the two columns were connected in a series, which allowed the permanent gases (H₂, N₂, O₂, CH₄ and CO), which were the first to leave the HP PlotQ, to flow directly into the MS column. The MS column was then set into bypass mode, and the CO₂ coming out from the HP PlotQ was directly detected. After this, the bypass of the MS column was removed, and the permanent gases were quantified.

The liquid products were evaluated through an HPLC analysis (Jasco, Tokyo, Japan), equipped with a PS-DVB sulphonated gel column (SepaChrom Benson Polymeric BP-OA BL0053 2000–0, Rho, Italy), a UV–Vis (Jasco 4070) set at 210 nm and a Refractive Index (RI) (Jasco 4030) detector. An aqueous H₂SO₄ solution (0.05 mM) was used as an eluent, with a flowrate of 0.4 mL/min. The column was held in a bath at 45.5 °C. An iodometric titration was performed after the tests to evaluate the sodium sulfite conversion.

4. Conclusions

The present study demonstrates that FSP is a highly effective synthesis method for producing photocatalysts with well-defined properties to enhance photocatalytic performance. Three semiconductors were synthetized: TiO₂, WO₃ and a type-II heterojunction, WO₃/TiO₂ (40/60 weight ratio). These three semiconductors were subsequently tested for their efficacy in CO₂ photoreduction. TiO₂ is one of the most common photocatalysts used in the world. Of all commercial products, the TiO₂ (P25) produced by Evonik (formerly Degussa) is not only one of the best known in the world, but is also produced industrially in a flame with a similar technique to FSP, using titanium tetrachloride (TiCl₄) in an oxygen-rich flame. For these reasons, TiO₂ P25 (Evonik) was therefore used as a reference catalyst for benchmarking purposes.

All catalysts tested were characterized by X-ray diffraction (XRD), nitrogen physisorption, scanning electron microscopy (SEM) and diffuse reflectance spectroscopy (DRS) UV–Vis. The latter returned values that align with those of the literature. The slightly lower bandgap of the TiO₂ FSP compared to P25 can be attributed to the higher rutile content, as confirmed by the XRD analysis.

The experimental tests were conducted using a unique high-pressure reactor operating at 18 bar. The results showed that the two single oxides, TiO₂ and WO₃, despite their significant difference in bandgap energy (3.10 eV vs. 2.73 eV), exhibited comparable productivities in CO₂ photoreduction, with each slightly underperforming relative to the P25 benchmark, which had a productivity of 16.9 mol/kg_cat·h.

A key advantage of FSP emerged in the synthesis of the mixed oxide WO₃/TiO₂, where intimate contact between the phases was obtained, leading to the formation of a type-II heterojunction. This heterojunction is evidenced by the slight reduction in the bandgap compared to the pure WO₃, as observed in the DRS measurements. As shown in Figure 1, the type-II alignment facilitated charge separation by directing the photogenerated electrons to the conduction band of the WO₃ and the holes to the valence band of the TiO₂, suppressing recombination and improving the photocatalytic efficiency. The WO₃/TiO₂ (40/60 weight ratio) catalyst exhibited a remarkable two-fold increase in formic acid productivity compared to P25, reaching 36.5 mol/kg_cat·h, which is among the highest values reported in the literature.

To further reinforce the advantages of FSP, a comparative benchmark with sol-gel or other synthesis techniques would be beneficial. However, the results presented herein already indicate that FSP enables intimate contact between semiconductor phases, leading to superior photocatalytic performance.

By achieving an outstanding productivity of 36 mol/kg_cat·h, this method leveraged the synergy between flame spray pyrolysis (FSP), high-pressure photoreduction and the type-II heterojunction. FSP proved to be an efficient and scalable synthesis technique for the synthesis of type-II heterojunctions, where phase interaction plays a crucial role in enhancing activity. The integration of these techniques not only allows us to optimize catalytic performance, but also paves the way for innovative applications in industrial catalysis.