One-Step Synthesis of CuxOy/TiO2 Photocatalysts by Laser Pyrolysis for Selective Ethylene Production from Propionic Acid Degradation

In an effort to produce alkenes in an energy-saving way, this study presents for the first time a photocatalytic process that allows for the obtention of ethylene with high selectivity from propionic acid (PA) degradation. To this end, TiO2 nanoparticles (NPs) modified with copper oxides (CuxOy/TiO2) were synthetised via laser pyrolysis. The atmosphere of synthesis (He or Ar) strongly affects the morphology of photocatalysts and therefore their selectivity towards hydrocarbons (C2H4, C2H6, C4H10) and H2 products. Specifically, CuxOy/TiO2 elaborated under He environment presents highly dispersed copper species and favours the production of C2H6 and H2. On the contrary, CuxOy/TiO2 synthetised under Ar involves copper oxides organised into distinct NPs of ~2 nm diameter and promotes C2H4 as the major hydrocarbon product, with selectivity, i.e., C2H4/CO2 as high as 85% versus 1% obtained with pure TiO2.


Introduction
Ethylene (C 2 H 4 ) is a key molecule in the chemical industry. About 75% of petrochemical products are obtained from this light olefin, including intermediates such as ethylene glycol and polymers such as polyethylene or polyvinyl chloride [1][2][3]. Currently, ethylene is mostly produced from the steam cracking of fossil fuels. However, besides relying on non-sustainable feedstock and being extremely polluting, such a method is one of the most energy-consuming processes in the chemical sector as it requires considerable working temperatures (T > 750 • C) [4,5].
Hence, researchers are developing new strategies for cleaner and lower-cost ethylene production without the use of hydrocarbons. Among them, heterogeneous photocatalysis attracts great attention for being an oxidation process operating at ambient temperature and atmospheric pressure, thus offering favourable perspectives for sustainable organic synthesis using solar light [6]. Nonetheless, very few research articles describe the photocatalytic production of ethylene [7][8][9][10][11][12]. Despite its high relevance, CO 2 photocatalytic reduction to C 2 H 4 is very challenging as it requires effective C-C coupling and a multi-electron reaction process [7]. Another strategy to form ethylene consists of the photocatalysis of carboxylic acids operating in an anaerobic medium. Photo-decarboxylation of such acids (acetic, propionic, butanoic, etc.) mainly produces CO 2 , H 2 , and alkanes [8][9][10][11][12][13] using TiO 2 or TiO 2 -based nanomaterials decorated with metal co-catalysts to promote quantum yield 2 of 12 maximization [14]. In some cases, alkenes were also detected. Kraeutler and Bard [11] identified ethylene traces from propionic acid decarboxylation with Pt/TiO 2, which were confirmed afterwards in the Betts et al. study using TiO 2 Evonik P25 [12]. Scandura et al. enhanced propene and ethylene selectivities from butyric acid and propionic acid photoreforming, respectively, with Au/TiO 2 compared to Pt/TiO 2 [10]. However, despite the use of onerous noble metals, alkenes were each time detected in trace amounts.
Herein, we report for the first time the selective photocatalytic production of ethylene from the valorisation of propionic acid using TiO 2 modified with copper nanoparticles under UVA light. Specifically, TiO 2 and Cu x O y /TiO 2 photocatalysts were elaborated using a laser pyrolysis technique, an industrial-scale process that enables a one-step and continuous synthesis of nanopowders at several grams per hour. The impact of the laser pyrolysis operating atmosphere (Ar, He) on the morphology and the material photocatalytic performances were studied. Propionic acid (PA) was selected as the model molecule; this volatile fatty acid can be obtained from the fermentation of organic wastes [15]. This work provides a new method for sustainable ethylene production with low energy consumption and low-cost photocatalysts.

Chemicals Synthesis of TiO 2 and Cu x O y /TiO 2 Nanoparticles
The synthesis of TiO 2 and Cu modified TiO 2 photocatalysts was performed with the laser pyrolysis technique under He or Ar gas. This method permits the continuous and direct production of nanopowders and is described elsewhere [16]. Briefly, its principle is based on the orthogonal interaction of an infrared CO 2 laser (10.6 µm) in continuous mode with a gaseous or liquid mixture of reagents. The laser was focused by means of a cylindrical lens, resulting in a horizontal beam of about 30 mm. For the synthesis of reference samples (TiO 2 ), the mixture consisted of 175 g of liquid titanium (IV) isopropoxide (TTIP) dissolved in a 150 mL o-xylene/ethylacetate solution (65 vol%/35 vol%). Dissolution of 4.13 g of Cu(acac) 2 (copper (II) acetylacetonate) into the previous mixture permitted us to obtain Cu x O y /TiO 2 particles with a theoretical Cu content equalling 2.0 wt% regarding the final nanomaterials. The resulting liquid was converted into microdroplets with a spray generator, regulating the mixture at 30 • C. The liquid droplets were transported into the reactor zone through a carrier gas (He or Ar) at 2000 cm 3 .min −1 flow. A sensitiser gas (C 2 H 4 , 800 cm 3 .min −1 ) was added to the carrier gas to enhance the laser absorption of the precursors. The reaction zone was confined by He or Ar inert gas flow. The CO 2 laser power (measured under inert gas) was set to 670 W and the pressure inside the reactor was regulated to 740 Torr. The powders were collected on filters located downstream.
The TiO 2 and Cu x O y /TiO 2 as-formed materials had the appearance of black, grey or brown-coloured powders due to the presence of carbon impurities. All samples were further annealed under air at 450 • C during 6 h in a tubular furnace (Carbolite) to remove residual carbon. The annealed particles were labelled TiO 2 -X and Cu/TiO 2 -X (X = He, Ar) depending on their synthesis environment. Powder images are provided in Figure S1.

Samples Characterization
The morphology of prepared photocatalysts was evaluated using transmission electron microscopy (TEM, Tecnai G2 F20 from Fei, Hillsboro, OR, USA).
In order to probe the presence of Cu element, energy-dispersive spectroscopy (EDS) analyses were performed in the scanning transmission electron microscope high-angle annular dark field (STEM-HAADF) imaging mode of a 200 kV Titan-Themis TEM/STEM electron microscope from Fei (Hillsboro, OR, USA) equipped with a Cs probe corrector and a ChemiSTEM Super-X detector (Fei, Hillsboro, OR, USA). These two accessories allow chemical mapping of light and heavy elements with a spatial resolution in the picometer range. The experimental conditions were set so that the total current within the probe used for the STEM-HAADF EDS chemical analysis was about 85 pA.
Samples specific surface areas were determined using N 2 adsorption according to the Brunauer-Emmett-Teller (BET) method with a Micromeritics (Norcross, GA, USA) 3Flex apparatus. Briefly, the BET method is based on the measurement of N 2 gas molecule physisorption onto the solid sample.
XPS measurements were performed on a K-Alpha spectrometer from ThermoFisher, Waltham, MA, USA equipped with a monochromated X-ray source (Al Kα, 1486.7 eV) with a spot size of 400 µm. The hemispherical analyser was operated in CAE (constant analyser energy) mode, with a pass energy of 200 eV and a step of 1 eV for the acquisition of survey spectra and a pass energy of 50 eV and 10 eV and a step of 0.1 eV for the acquisition of narrow scans. The spectra were recorded and treated by means of Avantage software (version 5.967). The binding energy scale was calibrated against the Ti 2p binding energy set at 258.5 eV.
Copper content was measured using inductively coupled plasma optical emission spectroscopy (ICP-OES) with a Jobin Yvon Activa instrument (Horiba, Edison, NJ, USA).

Photocatalytic Measurements
Photocatalytic runs were carried out in a 250 mL Pyrex photoreactor equipped with a mechanic agitator (500 rpm) and a gas inlet and outlet for gas sampling and purge. A 18W PL-L lamp (Philips, Amsterdam, The Netherlands), placed below the photoreactor, was chosen as the UVA source with a UV band of 350-410 nm centred at 370 nm (irradiance of 5 mW.cm −2 ).
We introduced 100 mL of a 1.0 vol% PA solution containing 50 mg of photocatalyst into the reactor. A continuous flow of argon (70 mL.min −1 ) was applied through a bubbler for 6 h into the system to fully remove the ambient air. After the purge, the photoreactor was sealed and irradiated with UVA light. Every 65 min, a 2 mL aliquot of photoproduced gases in the headspace was carried by vacuum pumping into a Clarus 500 GC FID-PDHID (PerkinElmer, Waltham, MA, USA) for analysis. Products were firstly separated though a RT-Q-Bond (Restek-30 m × 0.53 mm × 20 µm) column before FID with a Polyarc methanization module. For PDHID, a RT-M5A molecular sieve (Restek-30 m × 0.32 mm × 30 µm) was added. The carrier gas was helium N60 grade (Messer, Bad Soden, Germany). For each gaseous analysis, the oven temperature was set at 50 • C rising at 20 • C/min to reach a maximum temperature of 150 • C maintained for an additional 55 min. Tests were repeated twice to ensure reproducibility.

Synthesis
It was possible to observe the flame during synthesis experiments through a window. The flame seen in presence of helium was much less bright than under argon, which is an indication of a lower flame temperature under He atmosphere. The temperature appears to Nanomaterials 2023, 13, 792 4 of 12 be a major parameter due to the change of atmosphere. The specific heat capacity of helium is 10 times higher than argon, explaining why such a lower temperature was reached in helium atmosphere after laser absorption as well as the fast temperature decrease. As a consequence, the lower temperature observed under He atmosphere explains a lower grain size as well as a lower rutile content in nanoparticles synthesised under He compared to nanoparticles generated under Ar (Table 1 and Figure 1). Similar observations were reported by Pignon et al. [16], who explained these results as a cooling effect of He as well as less efficient confinement of the reactants. Indeed, He is a light gas and is therefore less efficient to confine the reaction. This is also apparent from the lower production rate (i.e., collected powder, Table 1) in He compared to Ar. 3, x FOR PEER REVIEW 5 of 12

Morphology and Structure
The morphology of the annealed samples was first investigated by TEM analysis (Figures 1 and S2). From the TEM images, all particles appear to present mostly a spherical shape arranged in chains typical of gas phase synthesis. The average diameters estimated from both BET and TEM measurements are reported in Table 1 and correspond well to each other.
The grain size is larger for nanoparticles synthetised under Ar rather than He environment. Particularly, the diameter of Cu/TiO2-Ar nanoparticles deduced from the TEM analyses seems to be 1.9 times larger than its bare TiO2-Ar reference. It seems that the addition of copper element promotes the formation of larger particles. This observation together with the highest rutile content observed in this sample confirms a higher synthesis temperature for Cu/TiO2-Ar. XRD analyses were performed on all the samples before and after annealing ( Figure S3). The samples consist of a mixture of a major anatase phase with a minor rutile phase, except for Cu/TiO2-Ar (Table 1). Finally, XRD  Table 1 summarises the characteristics of the obtained Cu x O y /TiO 2 samples and their TiO 2 references synthesised in similar conditions. In agreement with the more efficient decomposition of precursors, the production rate is higher under Ar atmosphere and carbon content increases in the as-formed TiO 2 -Ar and Cu/TiO 2 -Ar samples compared to samples obtained under He environment.
After annealing, TiO 2 samples turned white, indicating the efficient removal of carbon species, whereas Cu x O y /TiO 2 powders appeared pale green-coloured, suggesting the presence of copper. The associated mass losses are in agreement with the carbon content of the as-formed powders. This result was confirmed with elementary analyses revealing only traces of residual carbon ≤0.3 wt% in all the samples.

Morphology and Structure
The morphology of the annealed samples was first investigated by TEM analysis (Figures 1 and S2). From the TEM images, all particles appear to present mostly a spherical shape arranged in chains typical of gas phase synthesis. The average diameters estimated from both BET and TEM measurements are reported in Table 1 and correspond well to each other.
The grain size is larger for nanoparticles synthetised under Ar rather than He environment. Particularly, the diameter of Cu/TiO 2 -Ar nanoparticles deduced from the TEM analyses seems to be 1.9 times larger than its bare TiO 2 -Ar reference. It seems that the addition of copper element promotes the formation of larger particles. This observation together with the highest rutile content observed in this sample confirms a higher synthesis temperature for Cu/TiO 2 -Ar. XRD analyses were performed on all the samples before and after annealing ( Figure S3). The samples consist of a mixture of a major anatase phase with a minor rutile phase, except for Cu/TiO 2 -Ar (Table 1). Finally, XRD patterns do not reveal peaks associated with copper species, probably due to small crystallite sizes and low Cu content.
UV-visible DRS optical analyses ( Figure S4) revealed a band gap between 3.0 and 3.2 eV, in good agreement with anatase (3.2 eV) and rutile (3.0 eV) corresponding ratios in the annealed nanoparticles.

Chemical Analyses
The presence of copper was first confirmed by ICP-OES analysis (Table 1). Cu loadings are consistent regarding the 2.0 wt% Cu content in the liquid precursors. Regarding the copper content, it is slightly higher, with 1.91 ± 0.04 wt% (Cu/TiO 2 -Ar) vs. 1.76 ± 0.04 wt% (Cu/TiO 2 -He) when the argon atmosphere was used. The reaction temperature appears higher when working under argon atmosphere compared to helium atmosphere. This higher temperature induces a better decomposition of copper precursor in argon atmosphere and therefore a slightly higher Cu content in the powder.
High-angle annular dark-field imaging (STEM-HAADF) coupled with EDS chemical analysis was employed in order to evidence the presence as well as the repartition of Ti and Cu elements within the analysed samples. From the STEM-HAADF images illustrated in Figures 2 and S5, the presence of bright spots was detected on the surface of Cu/TiO 2 -Ar (Figure 2a,c). The corresponding STEM-HAADF EDS map recorded by considering the Cu Kα edge illustrated in Figure 2b suggests that these bright spots within the Cu/TiO 2 -Ar powders are Cu-based nanoparticles localised within the whole TiO 2 support. From these images, we deduce that the size of these NPs is of 1.8 nm ± 0.3 nm ( Figure S6 (Figure 2g,h) confirms that Cu element is homogeneously dispersed within TiO 2 , most likely due to the lower specific surface area of TiO 2 nanoparticles when synthetised under He. From this type of analysis, we can sustain that there are significant differences in terms of Cu dispersion and size within the TiO 2 support depending on the material synthesis conditions, i.e., the atmosphere in the present case. species across TiO2 particles. The STEM-HAADF EDS line scan analysis performed along the yellow arrow (Figure 2g,h) confirms that Cu element is homogeneously dispersed within TiO2, most likely due to the lower specific surface area of TiO2 nanoparticles when synthetised under He. From this type of analysis, we can sustain that there are significan differences in terms of Cu dispersion and size within the TiO2 support depending on the material synthesis conditions, i.e., the atmosphere in the present case. XPS analyses were conducted to investigate the surface chemical state of elements in both TiO 2 and Cu x O y /TiO 2 synthetised under different atmospheres. Figure S7 depicts the XPS full spectra of pure TiO 2 and Cu/TiO 2 , which shows the presence of, Ti, O, and C for all samples. In addition, Cu element appeared in both Cu/TiO 2 -He and Cu/TiO 2 -Ar powders. The Ti 2p core-level spectrum of TiO 2 (Table 2 and Figure S8a,b) consists of two peaks centred at 464.2 and 458.5 eV, corresponding to Ti 2p 1/2 and Ti 2p 3/2 of Ti 4+ , respectively [18]. A broadening of the Ti 2p doublet is observed for Cu x O y /TiO 2 samples in comparison to TiO 2 references (Table 2). Indeed, Ti 2p 3/2 FWHM corresponds to 1.1 eV for both TiO 2 -He and TiO 2 -Ar and increases to 1.3 eV and 1.5 eV for the Cu/TiO 2 -He and Cu/TiO 2 -Ar samples, respectively. This suggests an interaction between copper species and TiO 2 through the incorporation of Cu ions in TiO 2 lattice [19,20]. The Cu 2p 3/2 core-level spectrum ( Figure 3) reveals several co-existing Cu chemical states in Cu/TiO 2 -He and Cu/TiO 2 -Ar [18,21]. The peaks at 934.8 eV and 933.2 eV are assigned to the Cu 2+ 2p 3/2 signal of Cu(OH) 2 and CuO, respectively. Shake-up satellite peaks confirm the presence of Cu 2+ . Additionally, the component at about 932.5 eV refers to both the Cu 0 and Cu + 2p 3/2 signal, as the two oxidation states cannot be distinguished, having very close binding energies. By comparing Cu 2+ proportions, it seems that copper species are more reduced in the Cu/TiO 2 -He sample. XPS quantification reveals copper content of 4.1 wt% and 9.2 wt% for Cu/TiO 2 -He and Cu/TiO 2 -Ar, respectively. This is 2.3 and 4.9 times higher than bulk values provided by ICP-OES analyses, confirming a significant segregation of copper species on the surface of TiO 2 , clearly pronounced in the case of the Cu/TiO 2 -Ar sample. In addition, the Cu/Ti ratio is lower for Cu/TiO 2 -He, which could be attributed to a higher dispersion on the support [22,23], confirming our previous results.
Finally, the TiO 2 and Cu x O y /TiO 2 nanoparticles obtained by laser pyrolysis exhibit distinct morphologies: the Cu/TiO 2 -He sample consists of nanoparticles of 10 nm range, including very well dispersed Cu species on TiO 2 , whereas the Cu/TiO 2 -Ar particles are about 30 nm, with defined copper-based nanoparticles of about 2 nm size on the TiO 2 support.

Photocatalytic PA Degradation in Anaerobic Media
The main gases photoproduced from PA degradation under anaerobic conditions are presented in Figure 4 and a typical chromatogram is shown in Figure S9. The identified gases are CO 2 , ethylene C 2 H 4 , ethane C 2 H 6 and H 2 . Butane C 4 H 10 ( Figure S10) was detected as a minor product. The principal mechanisms for saturated carboxylic acid degradation in oxygen-free media with pure TiO 2 or decorated with noble metals have already been reported [11,13]. The first step consists of the decarboxylation of carboxylic acid through reaction with photo-generated trapped holes (Equation (1)):  Finally, the TiO2 and CuxOy/TiO2 nanoparticles obtained by laser pyrolysis distinct morphologies: the Cu/TiO2-He sample consists of nanoparticles of 10 nm including very well dispersed Cu species on TiO2, whereas the Cu/TiO2-Ar parti about 30 nm, with defined copper-based nanoparticles of about 2 nm size on t support.

Photocatalytic PA Degradation in Anaerobic Media
The main gases photoproduced from PA degradation under anaerobic condit presented in Figure 4 and a typical chromatogram is shown in Figure S9. The id gases are CO2, ethylene C2H4, ethane C2H6 and H2. Butane C4H10 ( Figure S10) was d as a minor product. The principal mechanisms for saturated carboxylic acid degr in oxygen-free media with pure TiO2 or decorated with noble metals have alread reported [11,13]. The first step consists of the decarboxylation of carboxylic acid t reaction with photo-generated trapped holes (Equation (1)): The CH3CH2• radical can react with a H• (H + + e − ) to further form C2H6 in m The coupling of two CH3CH2• is responsible for C4H10 formation, and two H• ca H2. According to Scandura et al. [10], traces of ethylene are formed by the reac CH3CH2• with h + . Hence, from Equation (1), selectivities are defined in te hydrocarbon (C2H4, C2H6) to CO2 ratio. The CH 3 CH 2 • radical can react with a H• (H + + e − ) to further form C 2 H 6 in majority. The coupling of two CH 3 CH 2 • is responsible for C 4 H 10 formation, and two H• can form H 2 . According to Scandura et al. [10], traces of ethylene are formed by the reaction of CH 3 CH 2 • with h + . Hence, from Equation (1), selectivities are defined in terms of hydrocarbon (C 2 H 4 , C 2 H 6 ) to CO 2 ratio.
Several differences can be seen from Figure 4. With pristine TiO 2 (black and blue curves, Figure 4), the main products are CO 2 and C 2 H 6 (C 2 H 6 /CO 2 = 73% and 68% for TiO 2 -He and TiO 2 -Ar, respectively). The slightly different photoactivities of TiO 2 -He and TiO 2 -Ar observed for CO 2 and C 2 H 6 productions could be correlated with different anatase to rutile ratios (i.e., presence of junction) as well as different BET surface areas. It is known that the presence of a junction enhances activity through more efficient electron transfer; see, for example [24,25]. However, in the present case, it is difficult to conclude if the slightly better efficiency of TiO 2 -He is related to its largest surface area (111 vs. 81 m 2 .g −1 ) or to the change in anatase to rutile ratio (87 vs. 76% of anatase). H 2 and C 2 H 4 (C 2 H 4 /CO 2 = 1%) are identified in trace amounts, in agreement with the literature in the absence of co-catalysts for H 2 [26][27][28] or with the use of noble metals (Pt, Au) for C 2 H 4 [10][11][12].
Surface modifications of TiO 2 photocatalysts with copper/copper oxides leads to drastic changes in terms of levels of photo-generated products. Notably, Cu/TiO 2 -He appears as the most efficient photocatalyst for CO 2 , C 2 H 6 , and C 4 H 10 production ( Figure S9), with C 2 H 6 as the major hydrocarbon (green curve, Figure 4). C 2 H 4 formation is enhanced (C 2 H 4 /CO 2 = 11%) until 200 min of reaction, then it slows down. On the contrary, H 2 production is increased from this same irradiation time (200 min). These phenomena could be attributed to the efficiency of the Cu x O y -TiO 2 heterojunction that decreases electron-hole recombination [29,30], but above all thanks to the reduction of copper oxides into metallic copper [31][32][33]. Indeed, well-dispersed copper species on the TiO 2 surface in the Cu/TiO 2 -He sample could be reduced after about 200 min of UV irradiation. Therefore, Cu 0 , acting as an escavenger, highly enhances H 2 formation. The CH 3 CH 2 • radical preferentially reacting with a H• formed on the Cu 0 site could explain how reduced copper species inhibit the formation of C 2 H 4 while enhancing C 2 H 6 production.   Figure  S11).
Several differences can be seen from Figure 4. With pristine TiO2 (black and blue curves, Figure 4), the main products are CO2 and C2H6 (C2H6/CO2 = 73% and 68% for TiO2-He and TiO2-Ar, respectively). The slightly different photoactivities of TiO2-He and TiO2-Ar observed for CO2 and C2H6 productions could be correlated with different anatase to rutile ratios (i.e., presence of junction) as well as different BET surface areas. It is known that the presence of a junction enhances activity through more efficient electron transfer; see, for example [24,25]. However, in the present case, it is difficult to conclude if the slightly better efficiency of TiO2-He is related to its largest surface area (111 vs. 81 m 2 .g −1 ) or to the change in anatase to rutile ratio (87 vs. 76% of anatase). H2 and C2H4 (C2H4/CO2 = 1%) are identified in trace amounts, in agreement with the literature in the absence of co-catalysts for H2 [26][27][28] or with the use of noble metals (Pt, Au) for C2H4 [10][11][12].
Surface modifications of TiO2 photocatalysts with copper/copper oxides leads to drastic changes in terms of levels of photo-generated products. Notably, Cu/TiO2-He appears as the most efficient photocatalyst for CO2, C2H6, and C4H10 production ( Figure  S9), with C2H6 as the major hydrocarbon (green curve, Figure 4). C2H4 formation is enhanced (C2H4/CO2 = 11%) until 200 min of reaction, then it slows down. On the contrary, H2 production is increased from this same irradiation time (200 min). These phenomena could be attributed to the efficiency of the CuxOy-TiO2 heterojunction that decreases  Figure S11).
Remarkably, Cu/TiO 2 -Ar (red curve, Figure 4) is the most efficient material for C 2 H 4 production: it is the major hydrocarbon product with a C 2 H 4 /CO 2 ratio in the range = 85-88%. Its formation evolves linearly at 260 ppmv/h, 130 times greater than the production rate reached from pure titania. This efficient production is obtained at the expense of C 2 H 6 (C 2 H 6 /CO 2 = 4-7%) and H 2 formation. This contrasting behaviour compared to Cu/TiO 2 -He may be related to less-dispersed, larger, and oxidised copper nanoparticles on TiO 2 support, as seen in STEM-HAADF/EDS and XPS characterizations. The larger size allows Cu x O y NPs to resist complete photo-reduction, as reported by Imizcoz et al. [31]. In addition, according to Yu et al. [29], a larger Cu x O y size limits the direct transfer of electrons from Cu x O y to H + due to a decrease in the conduction band. Consequently, H 2 and C 2 H 6 formations are not favoured. These particular conditions seem to enhance C 2 H 4 production, as observed in the present study.

Conclusions
This study reports for the first time the significant production of C 2 H 4 molecules from the photocatalytic degradation of propionic acid. TiO 2 nanoparticles modified with Cu x O y used as photocatalysts were prepared using the laser pyrolysis synthesis method. The atmosphere of synthesis (argon or helium) strongly modifies Cu x O y /TiO 2 morphologies as well as their photocatalytic properties. Cu x O y /TiO 2 obtained under helium consist of small TiO 2 nanoparticles with highly dispersed copper species that could be reduced into Cu 0 under UV irradiation. These reduced species promote the formation of H 2 and C 2 H 6 at the expense of C 2 H 4 . On the contrary, larger Cu x O y /TiO 2 NPs were obtained under argon atmosphere with defined Cu x O y nanoparticles (1-2 nm diameter) on the surface of TiO 2 . It seems that such less-dispersed Cu x O y nanoparticles inhibit H 2 and C 2 H 6 formation. The major hydrocarbon was C 2 H 4 , with an outstanding selectivity higher than 85%, outperforming by 130 times the ethylene production from pristine TiO 2 . Although the ethylene formation mechanism should be further examined, this study paves the way to a new energy-saving method to produce C 2 H 4 from organic acid without noble-metalbased photocatalysts.
Supplementary Materials: The following supporting information can be downloaded at: https://www. mdpi.com/article/10.3390/nano13050792/s1, Figure S1: Images of TiO 2 and Cu/TiO 2 samples before (a) and (b) after annealing; Figure S2: TEM images and associated histograms of size distribution of TiO 2 powders. Diameter measurements were performed on 100 particles with ImageJ software; Figure S3: XRD patterns of Cu/TiO 2 samples before and after annealing (a); TiO 2 and associated Cu/TiO 2 samples after annealing (b); XRD patterns of TiO 2 and Cu/TiO 2 as formed (c); Figure S4: Optical gap determination from UV-vis diffuse reflectance spectra using modified Kubelka-Munk function considering indirect gap of TiO 2 ; Figure S5: HAADF-STEM and Ti, O, Cu elemental maps performed on Cu/TiO 2 -Ar (a) and Cu/TiO 2 -He (b); Figure S6: Histograms of size distribution of Cu nanoparticles in Cu/TiO 2 -Ar. Diameter measures were realised on 40 particles with ImageJ software; Figure S7: XPS survey spectra of synthetised TiO 2 and Cu/TiO 2 ; Figure S8: Ti 2p (a,b), O 1s (c,d) and C 1s (e,f) core-level spectra of synthetised TiO 2 and Cu/TiO 2 ; Figure S9: FID (Flame Ionization Detector) gas chromatogram of after 910 minutes of irradiation from PA (1 vol%) photodecarboxylation with Cu/TiO 2 -Ar. Gas identifications were made with calibration cylinders of CO 2 alone (1000 ppm in nitrogen), C 2 H 6 alone (500 ppm in nitrogen) and a mixture of C 2 H 6 and C 2 H 4 (10 ppm in nitrogen). Retention times (min) are in green. Figure S10. C 4 H 10 formation from PA (1 vol%) photo-decarboxylation under UVA light. C 4 H 10 /CO 2 ratios are calculated at 910 min of irradiation, Figure S11. Carbon equivalent of each detected product by GC after 910 min of irradiation of PA (1 vol%). Calculations are made considering 1 CO 2 molecule for 1 C2 product (C 2 H 4 , C 2 H 6 , acetal-dehyde) and 2 CO 2 molecules for C4 product (C 4 H 10 ). Reference [34] is cited in supplementary materials.

Data Availability Statement:
The data that support the findings of this study are available from the authors on reasonable request.