A Parametric Study of the Crystal Phases on Au/TiO₂ Photocatalysts for CO₂ Gas-Phase Reduction in the Presence of Water

Abstract

Au/TiO₂ photocatalysts were studied, characterized, and compared for CO₂ photocatalytic gas-phase reduction. The impact of the nature of the TiO₂ support was studied. It was shown that the surface area/porosity/TiO₂ crystal phase/density of specific exposed facets and oxygen vacancies were the key factors determining CH₄ productivity under solar-light activation. A 0.84 wt.% Au/TiO₂ SG (Sol Gel) calcined at 400 °C exhibited the best performance, leading to a continuous mean CH₄ production rate of 50 μmol.h⁻¹.g⁻¹ over 5 h, associated with an electronic selectivity of 85%. This high activity was mainly attributed to the large surface area and accessible microporous volume, high density of exposed TiO₂ (101) anatase facets, and oxygen vacancies acting as reactive defects sites for CO₂ adsorption/activation/dissociation and charge carrier transport.

1. Introduction

Amongst the huge variety of semiconductors and composite materials investigated for CO₂ photocatalytic reduction in gas-phase, TiO₂ often prevails due to its chemical stability, moderate cost, and resistance toward corrosion [1,2,3,4]. However, its relatively fast electron-hole pair recombination rate and hindrance of visible light harvesting are considered the main limitations for solar light-driven gas phase CO₂ photocatalytic reduction in the presence of water. In the recent years, several strategies have been tested in order to overcome these limitations [5,6], including loading with metal nanoparticles, acting as electron sink, co-catalyst, or inducing surface plasmon phenomena [3,7,8,9,10]; combining TiO₂ with other semiconductors [11,12,13] or other elements by means of mono-doping [14,15,16] or co-doping approaches [17,18]; or modifying the morphology [19,20,21,22].

Among the different reaction products that could be obtained from CO₂ reduction, and also knowing that the energy content of oxygenated compounds decreases as the oxygen content increases (due to the relative decrease in C-H bond storage energy), CH₄ is one of the most interesting, together with methanol. Nonetheless, as methanol’s photo-reactivity is much higher than that of CO₂, leading to easy over-oxidation to formaldehyde and formic acid, high selectivity towards methane formation is preferred [3]. It has already been reported that TiO₂ anatase crystal facets can impact the performances towards CO₂ photoreduction, the {010} and the {001} facets leading to the highest and lowest selectivity toward methane formation, respectively [23]; whereas the exposure of both facets leads to the highest yields [24]. Furthermore, it has been established that, even if amorphous phases can be active, the crystallinity and defects of the semiconductor also play a crucial role, influencing the mobility of electron-hole pairs [15,24]. Theoretical studies have confirmed that the CO₂ adsorption, activation, and dissociation steps were highly influenced by oxygen-deficiency/defects disorders in TiO₂ anatase, rutile, or brookite crystal phases [25,26,27]. The role of oxygen vacancies as reactive defect sites in TiO₂ photocatalysts, even at low concentrations, is a key parameter in air-free photocatalytic properties, impacting the surface adsorption, and charge carrier transport [28,29]. Thus, the presence/introduction of oxygen vacancies may result in favorable charge transport phenomena, thus improving the charge carrier lifetime [30,31], and their role in photocatalysis has been extensively studied [32,33]. The TiO₂ particle size, crystallinity, presence of vacancies, density of exposed facets, porosity, and surface area, consequently play crucial roles in CO₂ photoreduction [34].

Between the different metal nanoparticles (NPs) used as co-catalysts and exhibiting surface plasmon resonance (SPR) properties in the visible range, Au can be considered one of the most interesting for driving CH₄ production from CO₂ gas-phase photocatalytic reduction in the presence of water vapor as a reducing agent. Due to interest in their use for catalysis, Au/TiO₂ materials have already become a reference as excellent photocatalysts [35,36]. Furthermore, as TiO₂ (P25 Evonik) is a commercially available reference, Au/TiO₂ P25 is often prepared by the so-called “deposition–precipitation” method [37].

In this paper titania (TiO₂ SG) was prepared by a home-made sol-gel method inspired by the literature [38]. Its structure and surface were tuned by varying various synthesis parameters, including the temperature of the calcination step. Au NPs were subsequently loaded using sodium borohydride reduction of chloroauric acid in the presence of titania. The structural, surface, porosity, and optical properties, as well as the presence of oxygen vacancies, are discussed and compared to those of synthesized TiO₂ brookite, commercially available (TiO₂ P25, TiO₂ UV-100)-based materials, and correlated to photocatalytic activity.

2. Results

2.1. Structural Characterization (XRD)

The XRD pattern of the TiO₂ SG-X (X stands for the post-synthesis temperature) are presented in Figure 1a and compared to those of bare TiO₂ P25, TiO₂ brookite, TiO₂ UV100, and TiO₂ UV100-X (X stands for the calcination temperature, Figure 1b). The corresponding mean crystallite sizes, calculated from the Debye-Scherrer equation, are summarized in

Table 1. TiO₂ mean crystallite size, surface area, porosity measurements, band-gap value. Experimental error of ^a 10%, ^b 8%, and ^c 5%.

	Mean Crystallite Size (nm) a			SBET (m2/g) b	Vpore (cm3/g) b	Mean Pore Diameter (nm) b	Eg (eV) c
	Anatase	Brookite	Rutile
TiO2 P25	17	/	23	55	0.20	≈30	3.20
TiO2 brookite	/	19	/	150	0.18	3–5	3.20
TiO2 UV100	11	/	/	315	0.33	˂3	3.25
TiO2 UV100-350	17	/	/	135	0.24	4.5	3.25
TiO2 UV100-550	19	/	/	77	0.28	8.0	3.25
TiO2SG	6	7	n.d	295	0.19	<4	2.85
TiO2SG-400	6	10	n.d	200	0.22	<4	2.96
TiO2SG-500	10	11	19	75	0.11	<4	2.95
TiO2SG-600	/	/	29	n.d	n.d	n.d.	2.90

. One can observe that the uncalcined TiO₂ SG is relatively well crystallized and exhibits a major peak at 25.36°, assigned to (101) planes in anatase phase, in agreement with JCPDS file 00-021-1272 [39,40]. Traces of brookite and rutile could be observed but are difficult to analyze and quantify. Increasing the calcination temperature to 400 °C did not significantly improve the degree of crystallinity of anatase, but rather increased the relative concentration of the rutile phase (from 500 °C) at 27.42° (JCPDS file 00-004-0551, [39]). Calcining further at 500 °C resulted in further crystallization of the anatase and rutile phases, increase in the rutile phase content, with no significant change to brookite (JCPDS file 00-029-1360, [40]). A complete transformation into rutile, accompanied by better crystallinity (29 nm) of the resulting phase occurred after calcination at 600 °C. We can also mention that calcination of TiO₂ UV-100 at 350 °C led to an increased anatase mean crystallite size, which reached a value close to the one determined for TiO₂ P25, without formation of rutile.

2.2. Surface Characterizations

2.2.1. BET Surface Area and Porosity Measurements

From the Brunauer, Emmett, and Teller (BET) measurements (Table 1, Figure 2), it can be seen that the TiO₂ P25 sample exhibited type-IV adsorption-desorption isotherm profiles with H₂ hysteresis, which are characteristic of mesoporous/macroporous solids with a mean pore diameter of ca. 30 nm. TiO₂ UV100 approximated type I behavior, which is characteristic of microporous solids with H3 hysteresis, attributed to rather uniform slot-type pores in accordance with its lower pore diameter (Table 1) and distribution (Figure 2d, pore size essentially below 3 nm). Increasing its calcination temperature to 350 and 550 °C yielded a mesoporous type-IV-like behavior, in line with the shift of the mean pore size from <3 nm to 4.5 and 8 nm. As for TiO₂ UV-100, brookite revealed a microporous type-I isotherm (Figure 2c) associated with type-H4 hysteresis, related to a non-uniform slot-type porosity, as confirmed by the corresponding broad pore size distribution (Figure 2d). From Figure 2a, one can observe that the TiO₂ SG-based materials (except for TiO₂ SG-600, whose porosity collapsed) showed a type-I microporous behavior, evidencing micropores with H2-hysteresis and revealing rather cylindrical channeled porosity. Figure 2b underlines that theTiO₂ SG-X samples exhibited the sharpest pore size distribution, centered at ca. 2.5–3.0 nm, with exclusive contribution of micropores. The corresponding pore volume increased with calcination at 400 °C, then diminished, to reach a very low value and surface area.

2.2.2. Au NPs Deposition and UV-Vis Absorption Properties

Kubelka–Munk function determination from the different TiO₂ supports (Figure 3) showed the expected behavior for the commercial TiO₂ P25 and UV-100 samples (Figure 3b). Calcination of TiO₂ UV100 seemed not to affect these properties. The synthesized TiO₂ brookite revealed the same absorption trend and edge, in accordance with band gap values of 3.20–3.25 eV. The as-synthesized TiO₂ SG (Figure 3a) exhibited, in addition to the standard anatase contribution, a visible light absorption contribution, whose origin could be attributed either to precursor residues and/or to oxygen vacancies. These hypotheses were confirmed by the large attenuation of the corresponding contribution tail by calcining at 400 °C. It must, however, be mentioned that calcining TiO₂ SG further induced a shift of the absorption edge towards higher wavelengths (as confirmed by the band-gap determination via a Tauc plot, using Kubelka–Munk formula, insert Figure 3a, correlated with the increase in rutile phase proportion (Table 1)).

The targeted Au content was 0.86 wt.% for all Au/TiO₂ photocatalysts. ICP-AES analyses showed that the Au deposition yield on some TiO₂ supports was high, varying from 92% to 98%, leading to Au/TiO₂ materials that all contained 0.82 ± 0.03 wt.% gold (

Table 2. Au content and deposition yield determined from ICP. Experimental error of 6%.

Support	Au(Th.) (wt.%)	Au(Real) (wt.%)	Au Deposition Yield (%)	λSPR (nm)
TiO2 P25	0.86	0.79	92	548
TiO2 Brookite	0.86	0.82	95	550
TiO2 UV100				556
TiO2 UV100-350	0.86	0.81	94	548
TiO2 UV100-550				562
TiO2 SG				561
TiO2 SG-400	0.86	0.84	98	556
TiO2 SG-500				559
TiO2 SG-600				541

). From the Au/TiO₂ absorbance spectra (Figure 4), two contributions could be distinguished, assigned respectively to TiO₂, as discussed previously, and to Au NPs. The second contribution was indeed characteristic of the localized surface plasmon resonance (LSPR) of Au NPs. It was positioned between 541 and 562 nm, in line with Au NPs in interaction with a titania matrix [41]. The variations in terms of position, broadness, and intensity could be linked to variations in Au content, particles size, and morphology, and Au-TiO₂ interactions. The same shape of SPR as observed on Au/TiO₂ P25 and Au/TiO₂ UV100-350 (Figure 4b) confirmed the same Au content as found by ICP-AES and also suggested that both samples exhibited the same gold particle size and interaction with the support. Even if those samples showed almost the same Au content as Au/TiO₂ brookite, one could observe a more intense LSPR signal on the last one, which may have been due to the larger Au NP size, thus suggesting that the dominant brookite phase may have provided less nucleation sites for Au NPs, hence the extended growth over the few sites available. Calcining TiO₂ UV-100 at 550 °C resulted in a more intense signal, but also in a slight red-shift in the LSPR position, suggesting the presence of a broader size distribution of Au NPs, i.e., the occurrence of sintering and aggregation, and associated LSPR coupling. The slight shift towards higher wavelengths observed in the TiO₂ UV100 material may have been correlated with the presence of a larger amount of amorphous phase.

Despite the similar Au contents of Au/TiO₂ P25, Au/TiO₂ UV 100-350, and Au/TiO₂ SG-400, the LSPR of Au/TiO₂ SG-400 was more intense and somewhat broader (Figure 4), suggesting the presence of a much broader size distribution of Au NPs. This may have arisen from the particularly low TiO₂ anatase mean crystallite size (6 nm), which was the lowest observed among the different TiO₂ supports and which was also close to the Au NP size generally produced by the method of deposition used, i.e., 3–5 nm [42,43]. The LSPR intensity was further enhanced after calcination at 500 °C (Figure 4a), as the anatase mean crystal size slightly increased, the surface area diminished, and the rutile phase appeared. Finally, after calcination at 600 °C, the LSPR signal was drastically lowered, in correlation with the huge decrease in surface area and the presence of an exclusively rutile phase.

2.3. Gas-Phase CO₂ Photoreduction Activity

Considering both the CH₄ productivity and electronic selectivity, Au/TiO₂ SG-400 was the best performing photocatalyst, leading to a mean rate of CH₄ production of 50 μmol h⁻¹g⁻¹ over 5 h (Figure 5a). This activity was higher than those reported by the Au/TiO₂ commercial-based reference materials (Figure 5b). Amongst these latter samples, one could observe that the TiO₂ P25 and TiO₂ UV-100 supports led to the same CH₄ productivity. However, the selectivity was much higher for the former. Applying and increasing the thermal treatment on TiO₂ UV-100 resulted in decreasing CH₄ productivity. Au/TiO₂ brookite revealed a moderate CH₄ production and selectivity. From Figure 5a, it can clearly be observed that an optimum calcination temperature of 400 °C was required for TiO₂ SG-based supports. Further increasing the calcination temperature is detrimental for both CH₄ production and selectivity, as is the absence of calcination. Comparing all the samples from a CH₄ productivity criteria point of view, the following order was obtained: Au/TiO₂ SG-400 > Au/TiO₂ UV100 ≈ Au/TiO₂ P25 > Au/TiO₂ UV100-350 > Au/TiO₂ SG > Au/TiO₂ SG-500 > Au/TiO₂ brookite > Au/TiO₂ UV100-550 > Au/TiO₂ SG-600.

As all those photocatalysts exhibit large differences in surface area, normalized reaction rates were calculated (Figure 6), allowing comparison of the intrinsic activities. Considering the normalized CH₄ production rates, the following order of activity was observed: Au/TiO₂ P25 > Au/TiO₂ SG-400 > Au/TiO₂ SG-500 > Au/TiO₂ UV100-350 >Au/TiO₂ UV100> Au/TiO₂ brookite > Au/TiO₂ SG ≈ Au/TiO₂ SG-500 > Au/TiO₂ SG-600, suggesting that the surface area parameter is an impacting parameter but not the only one.

2.4. TRMC Measurements

The lifetime of charge-carriers is completely described by the TRMC decay (I(t)) [44]. This decay is complex and may be due to different mechanisms that govern free electron mobility (considered as the most mobile charge carriers on TiO₂-based materials), such as recombination and trapping. At a 360 nm excitation pulse (Figure 7), one could clearly observe that the TiO₂ P25 support showed a higher charge carrier generation and free electron lifetime compared to the TiO₂ SG-X supports. Furthermore, in this latter series, neither an increase in calcination temperature from 400 to 500 °C nor Au NPs deposition seemed to induce any differences in terms of the charge carrier generation and lifetime (Figure 7b). Nevertheless, concerning Au/TiO₂ UV-100-based materials, calcining at 350 °C resulted in limiting both the charge carrier production and free electron lifetime decay, which may be attributed to trapping by Au NPs. Under visible light pulses at 450 nm (Figure 8), only very few charge carriers were generated on the Au/TiO₂ SG-X and Au/TiO₂ UV100-350 samples.

2.5. EPR Measurements

Electron paramagnetic resonance (EPR) measurements were conducted for the TiO₂ SG-400 support, which outperformed all the other studied samples regarding the total CH₄ productivity criteria, in order to detect potential active sites/defects in TiO₂ (Figure 9). Three types of EPR active centers were identified on TiO₂: the signals at 336, 332.5, and 330.0 mT (*) were assigned to the NO species captured at the surface of the porous TiO₂ [45,46]. The other visible signal for TiO₂ was a combination of (1) unpaired electrons trapped on an oxygen vacancy V_o⁺, at g = 2.00 (332 mT; Δ), a common defect created in various metal oxides [47,48]; (2) the substitutional N (the weak signal indicated by I); and (3) the paramagnetic N^2- nitrogen species in [O-Ti⁴⁺-N^2--Ti⁴⁺] units (signal marked II) [49]. This last signal might have originated from the nitric acid used for the sol gel synthesis and not eliminated due to the mild annealing temperature used for the TiO₂ SG-400.

3. Discussion

In order to obtain more insights into the structure/activity correlation, based on the aforementioned characterizations and solar-light photocatalytic CO₂ reduction activity, one may assume that the main impacting parameters driving CH₄ productivity are linked to the surface area/porosity type and volume/access to porosity, size/density of specific exposed TiO₂ facets, and presence/dispersion/size of Au NPs, as well as the presence/density of surface vacancies. Indeed, Au/TiO₂ SG-400 was the material exhibiting the highest productivity towards CH₄ formation (in terms of production and electronic selectivity). It was observed that its activity was first related to its high surface area and large and exclusive microporosity, presumably leading to larger amounts of adsorbed CO₂ molecules. Regarding charge carrier generation, it was seen from the TRMC measurements that TiO₂ P25 was the semiconductor generating the largest amount of long-life time charge carriers from UV-A activation, although some other TiO₂ supports also exhibited minor charge carrier generation at 450 nm. Among the other influencing factors, anatase crystallite size and orientation may play a determining role. Indeed, TiO₂ SG-400 and TiO₂ SG showed the smallest crystallite size (6 nm) and the largest surface area, even if the latter exhibited lower pore volumes, probably resulting from obstruction of part of the microporosity (hindering CO₂ adsorption) by the residues of precursors, which could be removed by applying a thermal treatment. Consequently, knowing that anatase (101) facets are the most stable [50], it can be assumed that these two TiO₂ SG-materials exposed a high density of anatase (101) facets. In addition, it may be assumed, comparing TiO₂ SG and TiO₂ SG-400, that the former was characterized by a larger contribution of amorphous phase, which was detrimental to the charge carrier mobility and thus the separation. One can also mention that oxygen vacancies, and more precisely the density of oxygen vacancies, confirmed for the TiO₂ SG-400 material, also played a determining role in both the CO₂ adsorption and activation step and in the charge carrier transport, as previously reported in the literature [27,28,29,30]. In addition, the deposited Au NPs likely acted as supplementary electron traps, as well as co-catalysts. It seems that the intensity of the corresponding TiO₂ SG-based SPR signal increased with the temperature of the post-treatment, up to 500 °C, leading to a modification of the Au-TiO₂ interaction, which obviously depended of the surface area, crystallinity, and exposed phases of TiO₂.

4. Materials and Methods

Titanium dioxide (TiO₂) Aeroxide^® P25 was purchased from Evonik Industries (Essen, Germany) and TiO₂ Hombikat UV-100 was purchased from Sachtleben Chemie GmbH (Duisburg, Germany). Both were used without further purification. TiO₂ Brookite was grown hydrothermally, following a previously reported procedure [51]. Briefly, 3.75 mL of titanium(IV) bis (ammonium lactate) dihydroxide (50 wt.%, Sigma-Aldrich, St. Louis, MO, USA) was mixed with 33.75 mL of 6.0 M urea aqueous solution in a 100 ml-hydrothermal reactor Teflon cup. Hydrothermal synthesis was carried out at 160 °C for 24 h. After cooling, solid product was washed with bi-distilled water several times and finally dried in a lyophilizer overnight. Titanium(IV) butoxide (Ti(OBu)₄) [Bu = CH₂CH₂CH₂CH₃], purum, ≥97%, Sigma Aldrich), nitric acid (HNO₃, ACS reagent 70%, Sigma Aldrich), ethanol (EtOH, CH₃CH₂OH, ≥98%, Sigma Aldrich), chloroauric acid (HAuⁱⁱⁱCl₄.3H₂O, ≥99.9% trace metal basic, Alfa Aesar, Haverhill, MA, USA), and sodium borohydride (≥98%, Sigma Aldrich) were used without any further purification.

4.1. TiO₂ Sol Gel (SG) Synthesis and Hombikat UV-100 Treatment

TiO₂ SG was obtained via a sol-gel process, following the protocol below. In a first beaker, 5 mL of Ti(OBu)₄ was mixed with 20 mL EtOH and heated at 40 °C in an oil bath. In a second beaker, 5 mL of H₂O and 5 mL of EtOH were mixed with 0.9 mL of HNO₃. After 1 h under magnetic stirring, the solution containing HNO₃ was added drop by drop into the solution containing the Ti precursor. The whole solution was kept at 40 °C (oil bath) under magnetic stirring for 2 h. Then, the gel was dried in an oven under static air at 80 °C for 48 h. After that, the sample was crushed into a fine powder and calcined in a tubular furnace at 400 °C for 2 h, with a heating ramp of 5 °C/min under air flow (100 cm³/min). A parametric study was carried out regarding the temperature of calcination (400, 500, and 600 °C).

The TiO₂ UV-100 was used as purchased but also thermally treated at 350 °C or 550 °C in static air for 4 h, with a heating rate of 10 °C/min, in order to remove the amorphous phase.

The resulting samples were labelled TiO₂ SG-X and TiO₂ UV100-X (X stands for the calcination temperature).

4.2. Au Nanoparticles (NPs) Deposition

Gold metal nanoparticles (Au NPs) (Au: 0.86 wt.% theoretical value) were loaded onto TiO₂ via an impregnation–reduction method [52] using HAu^IIICl₄ as the precursor of gold and NaBH₄ as the reducing agent. In a 100 mL flask, 400 mg of TiO₂ was mixed with 40 mL of H₂O for 5 min. Then, 80 µL of an aqueous solution of HAu^IIICl₄.3H₂O (2.2 × 10⁻¹ M) was added and the mixture was left under magnetic stirring (1000 rpm) for 45 min while the precursor is impregnated. Then, 1 mL of a fresh solution of NaBH₄ was added to the mixture. The NaBH₄ concentration solution was adjusted, in order to have a ratio NaBH₄/M_pr. = 5. After the reduction, the mixture was stirred for 15 min (1000 rpm) and then the obtained material was filtered and washed with 1 L of distilled water. To finish, the resulting Au/TiO₂ was dried in an oven at 100 °C overnight (air). The same deposition method was applied to all the different TiO₂ supports.

4.3. Characterization Methods

The X-ray diffraction apparatus used was a Bruker D8 Advance (Billerica, MA, USA) diffractometer equipped with a Lynxeye XE detector operating at 40 kV and 40 mA in θ/θ mode. The source of the X-Rays was a copper anticathode using the Kα line at 1.5418 Å. In order to obtain normalized diffractograms, a fixed amount of 50 mg of the sample was placed on a plastic holder. The acquisition of diffractograms was performed in scanning mode in steps of 2θ = 10° to 2 θ = 90° with a step of 0.05 and a counting time of 5 s per step. The mean crystal thickness was calculated using the Debye–Scherrer equation based on the width at half maximum and the position of the most intense peak.

The nitrogen adsorption–desorption isotherms were obtained using a Micromeritics Asap 2420 porosimeter (Norcross, GA, USA). The analysis was carried out at the nitrogen liquefaction temperature (77 K). The materials were degassed at 150 °C under primary vacuum for 5 h with a temperature rise of 10 °C/min, in order to desorb water and adsorbed molecules from their surfaces. The specific surfaces of the materials were calculated from the adsorption isotherm using the BET method. The pore size distribution could be calculated from the desorption data, using the BJH method specifically for mesoporous solids.

UV-vis absorption spectra were recorded on a Perkin Elmer 950 spectrophotometer (Waltham, MA, USA) fitted with a Labsphere 100 nm integrating sphere. The spectra were acquired in reflection mode (diffuse reflectance). The diffuse reflectance spectra were converted to Kubelka–Munk units via the equation: F(R) = $\frac{{(1-\mathrm{R})}^2}{2\mathrm{R}}$. To obtain the band gap (Eg) of the semiconductor (SC), the Tauc equation (F(R)∗hν)^S = (hν − Eg) was used, where h is the Planck constant, v is the frequency, and S is a coefficient, which is ½ in the case of an indirect gap SC and 2 in the case of a direct band gap SC (½ for TiO₂).

The charge-carrier lifetimes under illumination were determined through microwave absorption experiments using the time resolved microwave conductivity method (TRMC) [42]. The incident microwaves were generated by a Gunn diode of the K_α band at 30 GHz. The pulsed light source was an OPO laser (EKSPLA, NT342B, Vilnius, Lithuania) tunable from 225 to 2000 nm. It delivered 8 ns fwmh pulses with a frequency of 10 Hz.

Elemental analyzes of the samples were performed using inductively coupled plasma atomic emission spectrometry (ICP-AES, Varian 720 ES, Palo Alto, CA, USA) with a detection sensitivity for gold of 0.1 mg/mL.

EPR measurements were performed on a Bruker ESP300 (Billerica, MA, USA) apparatus using the X band (10 GHz) at 100 K, and 40 mg of sample was inserted in a quartz tube of 4 mm diameter.

4.4. Photocatalytic Tests

Photocatalytic setup has previously been described elsewhere [37]. During the test, a continuous flowing reaction mixture (CO₂ + H₂O) flow (0.3 mL.min⁻¹) passes through a light-transparent photoreactor (6 mL) equipped with a Hg lamp, simulating artificial solar light (150 W Ceramic-Metal-Halide Lamp, total photon flux for the whole lamp spectrum: 0.028 mol.s⁻¹.m⁻²). The irradiated surface was 19.6 cm². For the photocatalytic tests, 50 mg of the materials are mixed in ethanol and deposited on a 50 mm diameter glass disk by evaporation at 100 °C. The surface concentration of photocatalyst on the glass disc was ≈ 25 g.m⁻². The glass disk was then put in the reactor and the pilot was purged for 5 min with CO₂ flow (>100 mL.min⁻¹), to remove air and other gas impurities. Before starting the test, the pilot was run with a CO₂ gas flow (0.3 mL.min⁻¹) without illumination, to control the absence of products and the quantity of CO₂. Finally, products from the reactor were analyzed using an online micro-GC (Agilent 3000A SRA instrument, Santa Clara, CA, USA). Finally, the lamp is switched on and the duration of the test was 5 h. Production rates were calculated according to the following equation:

$${ \mathrm{r}}_{\mathrm{X}}\left(\mathrm{mol}.{\mathrm{h}}^{-1}.{\mathrm{g}}^{-1}\right)=\frac{\left[\mathrm{X}\right]\times {10}^{-6}\times \left(\mathrm{flow} \mathrm{rate}\right)\times 60}{{\mathrm{V}}_{\mathrm{m}}\times {\mathrm{m}}_{\mathrm{photocat}}}$$

where [X]: Concentration in ppm, m_photocat: mass of photocatalyst (g), the flow rate was 0.0003 L.min⁻¹, and V_m = 24.79 L.mol⁻¹ (STP conditions).

Electronic selectivity was calculated using the following equations:

$${\mathrm{CH}}_4 \mathrm{Selectivity} =\frac{8\left[{\mathrm{CH}}_4\right]}{8 \left[{\mathrm{CH}}_4\right]+2 \left[{\mathrm{H}}_2\right]}$$

$${\mathrm{H}}_2 \mathrm{Selectivity} =1-{\mathrm{CH}}_4 \mathrm{Selectivity}$$

5. Conclusions

To conclude, in the present paper, Au/TiO₂ photocatalysts were studied, characterized, and compared for CO₂ photocatalytic gas-phase reduction, with regards to the nature of the TiO₂ semi-conductor support. It was shown that the surface area, porosity TiO₂ crystallographic phase, density of specific exposed facets, and oxygen vacancies were the key factors driving CH₄ productivity under solar-light activation. Understanding the main impacting factors made it possible to design, via simple sol-gel synthesis, an optimal TiO₂-based photocatalyst, exhibiting the best compromise between these impacting factors. In this work, 0.84 wt.% Au/TiO₂ SG calcined at 400 °C exhibited the best performance, leading to a continuous mean CH₄ production rate of 50 μmol.h⁻¹.g⁻¹ over 5 h and an electronic selectivity of 85%, mainly due to UV-driven activity. This highest activity was attributed to the high surface area and accessible microporous volume, high density of exposed TiO₂ (101) facets, and oxygen vacancies acting as reactive defects sites for CO₂ adsorption/activation/dissociation and charge carrier transport.