Highly Efficient Photothermal Reduction of CO₂ on Pd₂Cu Dispersed TiO₂ Photocatalyst and Operando DRIFT Spectroscopic Analysis of Reactive Intermediates

Abstract

The photocatalytic conversion of CO₂ to fuels using solar energy presents meaningful potential in the mitigation of global warming, solar energy conversion, and fuel production. Photothermal catalysis is one promising approach to convert chemically inert CO₂ into value-added chemicals. Herein, we report the selective hydrogenation of CO₂ to ethanol by Pd₂Cu alloy dispersed TiO₂ (P25) photocatalyst. Under UV-Vis irradiation, the Pd₂Cu/P25 showed an efficient CO₂ reduction photothermally at 150 °C with an ethanol production rate of 4.1 mmol g⁻¹ h⁻¹. Operando diffuse reflectance infrared Fourier transform (DRIFT) absorption studies were used to trace the reactive intermediates involved in CO₂ hydrogenation in detail. Overall, the Cu provides the active sites for CO₂ adsorption and Pd involves the oxidation of H₂ molecule generated from P25 and C–C bond formation.

1. Introduction

An anthropogenic climate dramatically increases the CO₂ level in the atmosphere, which can lead to the deterioration of the environment. The utilization of CO₂ is an alternative route to reduce it in the environment, and it can also be converted into valuable fuels and chemicals [1,2,3,4,5]. The photocatalytic hydrogenation of CO₂ is an efficient technique for eco-friendly conversion, but CO₂ activation and selective hydrogenation remain a challenge [6,7]. This is due to the high thermodynamic stability and chemical inertness of CO₂ [8,9]. Therefore, the activation and hydrogenation of CO₂ require enormous energy to break the O=C=O bond and implement the insertion of the C-H bond. Previous research studies show that the hydrogenation of CO₂ resulted in numerous other products. Mostly, the CO₂ reduction processes resulted in the production of carbon monoxide (CO), a more extremely poisonous gas than CO₂. Moreover, it is very rare to produce value added chemicals, such as methane, methanol, acids, and hydrocarbons [10,11,12,13,14,15,16,17]. Therefore, the selective hydrogenation of CO₂ to ethanol has received great attention, and this can be achieved via electrochemical and thermochemical processes [18,19]. However, no appreciable development in the field of photo- and photothermal catalysts has been observed, which offers a sustainable method for CO₂ to ethanol conversion.

Currently, palladium (Pd) and platinum (Pt) based catalysts are demonstrated for the selective hydrogenation of CO₂ to ethanol [19,20,21]. In such processes, the Pd metal is cheaper than Pt, and Pd-based catalysts have worked well in the C–C coupling reactions as well [22]. Besides, the C–C coupling reaction step is the most important step to control the yield of ethanol [23]. Copper (Cu) is the cheapest and most easily available material, and it has greater affinity for CO₂ gas. Moreover, oxides of Cu represent an efficient photocatalyst with excellent redox potential for water splitting (water oxidation) in the photocatalytic hydrogen evaluation reaction [24]. The Cu/Fe₂O₃ catalyst is a promising catalyst for water-splitting in thermochemical reactions, to produce H₂ and capture CO₂ at the same time [25]. TiO₂ has been proven to be a most promising material in photocatalytic water splitting, pollutant degradation, self-cleaning agents, and many other fields [26,27]. The effect of reaction temperature on photodegradation performance was studied over Pd/TiO₂ and Cu/TiO₂ photocatalysts. The result revealed that Pd/TiO₂ performed well during the photodegradation of methylene blue. [28] Therefore, the combination of two attractive elements is an efficient way to produce ethanol. Previous studies have demonstrated the various combinations of Pd-Cu nanoparticles that are highly selective and active catalysts for the production of ethanol from CO₂ by thermochemical reaction [19]. Although selective hydrogenation produces valuable chemicals, the mechanism behind this selectivity is still an uncovered area. Therefore, the mechanism of selective CO₂ reduction was probed by operando diffuse reflectance infrared Fourier transform absorption spectroscopy (DRIFTS).

The DRIFT study is used to detect the molecules and ions at their place of origin where the chemicals and substances are being produced in a chemical reaction. DRIFT analysis is a widely used technique for in-situ studies concerning the photodegradation of adsorbed dye and tracing the gas adsorbed reaction mechanism on the catalytic surface. It has also been used in agro-food industries due to its analytical compatibility with physicochemical properties of food commodities [29,30]. Moreover, DRIFT analysis was applicable to test non-transparent and indispersable materials. In our previous work, we have demonstrated photocatalytic efficiency and the reaction mechanism using DRIFT in the oxidation of ethanol [31]. The solid-state photodegradation mechanism was evaluated by DRIFT for the surface adsorbed AB-9 dye on titanium dioxide (TiO₂) [32]. This technique was used to investigate the electron transfer and energy conversion processes at the interface, adsorbed species, and dynamics of photo-generated electrons on nitrogen-doped TiO₂ in dye-sensitized solar cells (DSSC) [33]. Moreover, the DRIFT analysis was used to identify the reaction mechanism of CO₂ hydrogenation [19,20,34].

In this work, we investigate the selective hydrogenation of CO₂ to ethanol by Pd₂Cu/P25 photocatalyst. A facile one-pot synthesis procedure was used to obtain the Pd₂Cu alloys on TiO₂ nanoparticles. The physicochemical properties of the as prepared Pd₂Cu/P25 catalysts were analyzed using various analytical techniques. The hydrogenation of CO₂ was tested using as-prepared Pd₂Cu/P25 catalysts following the photothermal method. Moreover, the reactive intermediates and reaction mechanism of hydrogenation of CO₂ to ethanol were determined from operando DRIFT analysis.

2. Materials and Methods

2.1. Chemicals

Palladium (II) bis(acetylacetonate) (Pd(acac)₂), 99%), Oleylamine (70%), ethyl alcohol (99.9%), and acetone (99.9%) were purchased from Sigma Aldrich. Copper chloride dihydrate (CuCl₂·2H₂O, 99%), triphenylphosphine (TPP, 99%), and commercial P25 (99.9% grade) were obtained from Nacalai Tesque, Acros Organics, and Degussa, respectively. The gases deployed in this research, such as high purity nitrogen (N₂, 99.999%), carbon dioxide (CO₂, 99.5%), and helium (He, 99.99%), were procured from Yusheng gas company, Taipei, Taiwan.

2.2. Preparation of Pd₂Cu/P25 Photocatalyst

For the synthesis of Pd₂Cu/P25 photocatalyst, 10 mg of Pd(acac)₂ and 55 mg of CuCl₂·2H₂O were dissolved in 12 mL of Oleylamine in a three-neck flask. Additionally, 90 mg of triphenylphosphine (TPP) was added. The flask was heated to 80 °C under continuous stirring in an N₂ atmosphere for about 20 min until the chemicals were thoroughly dissolved. The temperature was then gradually raised to 170 °C, and the system was kept at this temperature for 4 h. Then, 100 mg of TiO₂ (P25) was added to this suspension and kept for 20 min, before being naturally cooled down to room temperature. The precipitate was separated and washed with a mixture of ethyl alcohol and acetone. The obtained precipitate dried overnight at 85 °C, followed by calcination at 450 °C for 4 h.

2.3. DRIFT Analysis

The DRIFT experiments were conducted by Perkin Elmer FT-IR spectroscopy. The custom-designed Harrick DRIFT cell was utilized for this study. Approximately 40 mg of the photocatalyst was placed in a DRIFT cell sample holder. The sample was covered by ZnSe/CaF₂ windows for IR radiation and a light source was passed through respectively. The DRIFT chamber was filled with N₂ gas for 20 min to remove impurities, before the background spectrum of the sample was acquired. The temperature of the sample was controlled using a thermocouple attached to the DRIFT cell. The wavenumber resolution was fixed at 4 cm⁻¹, and the experiments were carried out in the range of 500–4000 cm⁻¹. For the CO₂ reduction experiment, the CO₂ was purged into the chamber with the water vapor at 10 cc/s for 120 min. An outlet of the reaction chamber was connected with the mass spectrum during the collection of DRIFT spectra. Meanwhile, the mass spectra wer also collected (QMS 100 Series, capillary; PEEK, and glass-lined plastic inlet type, pressure 10 mbar to 1 bar, Quadrupole type spectrometer, Faraday cup standard electron multiplier optional detector) [31].

2.4. Photothermal CO₂ Reduction by the Reactor

The photothermal CO₂ reduction was performed in a custom-made reactor. Firstly, 40 mg of Pd₂Cu/P25 was dispersed in 50 mL of deionized (DI) water by ultrasonicator, then poured into the reactor. Therein, the magnetic stirrer continuously stirred the dispersion. Subsequently, the CO₂ was purged into the reactor and saturated for 30 min, and then the reactor was closed perfectly. The UV-Vis (350 to 750 nm) light source (with an intensity of 380.8 μW) was used to irradiate the reactor. The reaction was kept at 80 °C by the hotplate. During each measurement, 0.1 μL of the reaction mixture was collected from the reaction and analyzed by gas chromatography (GC Model 6890 N/MS Model 5973, Agilent Technologies, Poway, CA, USA).

2.5. Characterizations

The crystal structures of the as-prepared samples were investigated by the X-ray diffractometer (XRD) with Cu-Kα radiation (λ = 1.5406 Å). The Raman spectra and UVDRS spectra of the samples were recorded using confocal Raman spectroscopy (ACRON Technology, Taipei, Taiwan) and UV-Vis-NIR spectroscopy (Agilent Technology, Santa Clara, CA, USA), respectively. The back scattered images of the as-synthesized samples were recorded by scanning electron microscopy (FESEM, JEOL-JSM-7610F, Tokyo, Japan). High-resolution transmission electron microscopy (HRTEM) images were obtained from JEOL TEM (JEM-2010) at 200 kV. The element composition was recorded on the energy dispersive spectrum EDS (HORIBA, X-act, Tokyo, Japan). The JEOL JPS-9030 (Tokyo, Japan) X-ray photoelectron spectrometer (XPS) was deployed to analyze the surface properties of the prepared samples. The dissolved solution of the as prepared photocatalyst was analyzed by the inductively coupled plasma-atomic emission spectrometer (ICP-AES, Shimadzu ICPE-9000, Kyoto, Japan) for the quantitative elemental contents in each sample.

2.6. Dark Adsorption by CO₂ with Transmission FTIR

The dark adsorption experiment was carried out in the transmission mode by a spectrometer (JASCO FT-IR 4600, Tokyo, Japan) equipped with a Mercury Cadmium Telluride detector. The sample was pressed into a self-supporting disk (20 mm diameter, ca. 30 mg) and placed in an IR cell attached to a gas-phase closed circulatory system for observation. The sample was pretreated at 723 K for 1 h under vacuum and then probed by a fixed amount of CO₂ vapor from 5 to 500 Pa at ambient temperature. After the dark adsorption at 500 Pa, the system was evacuated to vacuum again to remove free CO₂ and other free carbonyl species, to investigate the chemically adsorbed CO₂ species on the samples.

3. Results and Discussion

3.1. Material Characterizations

Figure 1a shows the X-ray diffraction patterns of Pd₂Cu alloy on P25 along with the diffraction pattern of pristine P25. The observed crystalline peaks at 41.11, 48.46, and 71.02° are attributed to the (111), (200), and (220) planes of PdCu alloy that are matched with the ICDD No. 00-048-1551 [35]. The obtained crystalline parameters indicate that the Pd₂Cu alloy is crystallized in the body-centered cubic crystal. The crystal structure of Pd₂Cu is illustrated in the inset of Figure 1a.

Moreover, the remaining peaks at 25.38, 38.23, 48.47, 54.71, 56.01, 27.58, 36.90, 41.60, and 55.0° are assigned to the (101), (004), (200), (105), (211), (110), (101), (210), and (211) planes of TiO₂ (P25) respectively [31]. It can be seen from XRD peaks that the intensities of Pd₂Cu peaks are dominated by the P25 crystalline peaks. This is because of the low loading level and high dispersion of Pd₂Cu on P25. Figure 1b shows the Raman spectra of Pd₂Cu/P25 and P25. The typical peaks appearing at 142, 397, 533, and 651 cm⁻¹ correspond to the Eg, B1g, A1g, and Eg modes of TiO₂ (P25), respectively. The additional peaks that appeared at 218 and 279 cm⁻¹ are correlated to the specific interactions between PdCu alloys and P25. The PdCu alloys have no vibrations. However, the corresponding oxides of Pd and Cu revealed significant peaks. The vibrations at 398 and 650 cm⁻¹ are increased for the Pd₂Cu/P25 when compared with P25. This might be attributed to the stretching vibrations of Pd-O and Cu-O. Moreover, the light absorption properties of Pd₂Cu/P25 were investigated by UV-Vis spectra. As shown in Figure 1c, the UV-visible absorption spectra of Pd₂Cu/P25 exhibited stronger absorption in the visible range when compared to the P25. The indirect bandgap of Pd₂Cu/P25 is estimated from the plots of (αhv)^1/2 vs. hv (energy of light), which shows 3.0 eV for both Pd₂Cu/P25 and P25 (Figure 1c, inset). Furthermore, the FESEM backscattered electron (BSE) image gives the topography of the P25 supported Pd₂Cu alloys.

Figure 1d shows the BSE image of Pd₂Cu/P25, wherein the Pd₂Cu alloy shows a stronger backscatter electron signal, which can be seen as dark spots. P25 exhibits a weaker electron signal. This BSE image can help to distinguish the Pd₂Cu alloy and P25 in the highly dispersed Pd₂Cu/P25 catalyst. Figure 2a displays the TEM images of Pd₂Cu alloy on the P25 nanoparticles, wherein the 5 ± 1 nm-sized Pd₂Cu nanodots are highly dispersed on 20 ± 5 nm-sized P25 nanoparticles. The high magnification images (Figure 2b,c) clearly show the mono dispersion of Pd₂Cu nanodots and their lattice fringes (Figure 2d). Figure 2e shows the SAED patterns of Pd₂Cu/P25, which are taken at a random angle to view clear ED patterns. The obtained ring patterns are indexed to the (101), (200), and (111), (311) planes, which are in accordance with the XRD patterns of P25 and Pd₂Cu, respectively. The uniform distribution and the ratio of elemental percentage are investigated using the EDX and mapping studies. Figure 2f reveals 82% of P25 and 18% of Pd₂Cu alloy in the photocatalyst, respectively. Moreover, elemental mapping of Pd₂Cu/P25 was performed and the results are shown in Figure 2g–j. Thus, it confirmed that the Pd₂Cu nanodots are uniformly distributed on the P25 nanoparticles, supporting better photothermal CO₂ conversion.

The X-ray photoelectron spectra (XPS) were used to explore the surface properties of Pd₂Cu/P25. The narrow scan Pd 3d spectrum shows two peaks for spin-orbits 3d_5/2 and 3d_3/2 at 339 and 345 eV, corresponding to the metallic Pd⁰ respectively (Figure 3a). Moreover, the deconvolution spectra of Pd 3d show an additional peak for the Pd²⁺, which is attributed to PdO on the surface [36]. Figure 3b shows the narrow scan spectrum of Cu 2p. The binding energies at 937 and 957 eV correspond to the 2p_3/2 and 2p_1/2 spin-orbits, respectively. The Cu 2p spectrum was deconvoluted into four peaks. The two main peaks denote Cu in the metallic state, while the minor peaks denote Cu in +2 oxidation states. Similar to Pd, Cu also exhibited CuO on its surface. Moreover, the peak intensity of Cu 2p was lower than that of Pd, which is in accordance with EDX results. Additionally, Figure 3c,d shows the narrow scan peaks of Ti 2p and O 1s, which are indexed to the typical peaks of TiO₂ (P25). The Ti 2p spectra shows two peaks at the binding energies of 462.1 and 467.8 eV belonging to the 2p_3/2 and 2p_1/2 orbitals, respectively. The deconvoluted O 1s peak reveals two binding energies at 533.5 and 535.3 eV, corresponding to the surface hydroxyl groups and lattice oxygen atoms respectively. Overall, these characterizations confirm the formation of Pd₂Cu alloy on the P25 support. The amount of metals proportion in the photocatalyst were further confirmed by ICP-AES. For the analysis, the powder of the as-prepared catalyst was soaked in the aqua regia solutions for 6 hours and then mixed with 30wt% HF for overnight. The results were summarized in

Table 1. ICP-AES results of Pd₂Cu/P25 photocatalyst.

ICP Analysis	Pd	Cu	Ti
mg L−1	0.0416	0.0129	0.232
Molecular Weight	106.42	63.54	47.867
mMolL−1	3.9 × 10−4	2.0 × 10−4	4.8 × 10−3
Molar ratios	~2	1	24

.

3.2. Photothermal CO₂ Reduction

The photothermal hydrogenation of CO₂ over a Pd₂Cu/P25 catalyst was evaluated using the photoreactor. The detailed experimental procedure is presented in the experimental section and the schematic diagram of the reactor setup is shown in Figure 4a. The reaction was continued for up to six hours, while the sample was taken from the reaction mixture and subjected to the GCMS analysis. In the first hour, no products were seen in the reaction mixture. Afterwards, the GCMS shows the peaks for ethanol, as shown in Figure 4b. The typical fragmentation of mass 45 and 46 m/z correspond to the [C₂H₅O]⁺ and [C₂H₅OH]⁺ respectively [37]. The ethanol yield is shown in

Table 2. The comparison of ethanol yield on Pd₂Cu/P25 with various catalysts.

Catalysts	Gas (COx)	Temperature (°C)	C2H5OH Yields (mmol g−1 h−1)	Reference
1% Pt/Co3O4	CO2	200	0.07	[38]
1.2% RhMnLiFe	CO	320	3.5	[39]
Rh/SiO2	CO	250	0.37	[40]
Au/TiO2	CO2	200	2.82	[41]
8.7% Pd-10% Cu/SiO2	CO2	250	1.12	[42]
Pd2Cu/P25	CO2	150	4.1	This work

, and the yield was compared with other reported catalysts. The photocatalytic CO₂ reduction of Pd₂Cu/P25 was also compared with pristine P25. The P25 showed no signal for ethanol production. Herein, the P25 (TiO₂) oxidized the water at its valence band, and the electrons were utilized by Pd₂Cu at the conduction band. Further, the Pd₂Cu has reduced the CO₂ by capturing the H⁺ from aqueous media. Therefore, without Pd₂Cu, it is difficult to produce ethanol by P25 alone. Figure 4c shows the schematic representation of the overall reaction process. Furthermore, the reaction mechanism of CO₂ reduction and the intermediate products were explored by the DRIFT analysis.

3.3. The FTIR Spectra of the Dark Adsorption of CO₂ on Photocatalysts

An airtight gas-cell FTIR accessory with 15 cm pathlength for transmission mode measurements was used to analyze the extent of interactions between adsorbates and photocatalysts. Figure 5 illustrates the FTIR spectra of Pd₂Cu/P25 probed by CO₂ from 5 to 500 Pa at ambient temperature. After the CO₂ adsorption at 500 Pa was completed, the system was evacuated to vacuum again to remove the free CO₂ and free carbonate species. The evacuated spectrum is symbolized as “Evac.” in Figure 5. This information provides strong evidence of the formation of chemical adsorption bonding between our sample and carbonate species. The peaks are assigned following the reference [43,44,45,46,47]. These spectra feature two major species, namely carbonates and bicarbonates, that were adsorbed on the catalysts by a few different kinds of connections. Starting with bicarbonates, the band at 1215 cm⁻¹ is related to the (OH) vibrational signals of monodentate and bidentate bicarbonates (m-HCO₃⁻ and b-HCO₃⁻). The band at around 1625 cm⁻¹ can be ascribed to the asymmetric vibrational signals (v_as(CO₃)) of both monodentate and bidentate bicarbonates (m-HCO₃⁻ and b-HCO₃⁻) [43,44,45,46]. However, the significant bands at 1425 cm⁻¹, identified as the symmetric vibration signal (_vs(CO₃)) of the bidentate bicarbonates (b-HCO₃⁻), implies that the bicarbonate groups on the Pd₂Cu/P25 were dominated by bidentate bicarbonates (b-HCO₃⁻).

The bands related to carbonate species were also investigated. The broad bands at 1675 cm⁻¹ and 1325 cm⁻¹ represented the asymmetric vibrational signals (v_as(CO₃)) and the symmetric vibration signal (v_s(CO₃)) of the adsorbed bridged carbonates (br-CO₃²⁻), respectively [43,44,45,46]. The significant bands at 1525 cm⁻¹ and 1350 cm⁻¹ exhibited the asymmetric vibrational signals (v_as (CO₃)) and the symmetric vibration signal (v_s(CO₃)) of the adsorbed bidentate carbonates (b-CO₃²⁻), respectively. Besides, a small shoulder at 1500 cm⁻¹ (asymmetric vibrational signals (v_as (CO₃)) indicated that there was a trace amount of poly-dentate carbonate species (p-CO₃²⁻) observed by IR. After the sample reached equilibrium at 500 Pa of CO₂, the system was then turned in to vacuum for 30 min. As shown in Figure 5, the remaining signals mostly belonged to carbonates. This result can provide further insights in DRIFT analysis when the reaction was applied.

Thus, the dark adsorption FTIR analysis probed by CO₂ was carried out using Pd₂Cu/P25. This analysis gave us a closer look at the micro-environment around the active sites and the adsorbent, which helped us rationalize the photocatalytic performance of each sample in the next section.

3.4. In-Situ DRIFT Investigation of Photothermal CO₂ Reduction

The mechanisms of CO₂ reduction and resultant intermediate products were monitored by in-situ DRIFT analysis. Figure 6a shows the time-dependent DRIFT absorbance spectra of Pd₂Cu/P25. The pristine Pd₂Cu/P25 shows peaks of around 3500–3000 cm⁻¹ and 1320 cm⁻¹ for the surface hydroxyl group (moisture). These surface hydroxyl groups were initially removed by thermal treatment at 150 °C for 20 m in a nitrogen atmosphere. After removing the moisture, the CO₂ and gas-phase H₂O were passed into the drift cell at 150 °C. The spectra were collected in time intervals (t) of 20, 40, and 60 min in darkness (absence of light) and in the presence of light. The peaks appearing at 2346 cm⁻¹ and 3700–3500 cm⁻¹ correspond to the ʋ₃ and ʋ₃ + ʋ₁ stretching vibrations of adsorbed CO₂, respectively. Additionally, the monodentate carbonate CO₃²⁻ peak appeared at 1085 cm⁻¹ for the metal carbonate (m-C–O–) [18], which confirmed the CO₂ adsorption on Pd₂Cu/P25 in dark.

This metal carbonate (m-C–O–) peak and the CO₂ peaks (3700–3500 cm⁻¹) were slowly decreased when irradiated with UV-Vis light, and at the same time, the C–H peaks are increased at 2962 cm⁻¹ (Figure 6a). In addition, the *CO peak arises at 1640 cm⁻¹ during the light irradiation, as shown in Figure 6b [19]. The obtained DRIFT spectra were derived to visualize the significant changes. Thereby, the obtained IR spectra were subtracted from the initial absorbance spectra. The T₀, T2₀, T4₀, and T₆₀ refer to the spectra taken in darkness at t = 20, 40, and 60 min respectively. Similarly, TL₀, TL₂₀, TL₄₀, and TL₆₀ refer to the spectra under UV-Vis light irradiation at t = 20, 40, and 60 min respectively. As shown in Figure 6c, the specific peaks at 2937, 2873 cm⁻¹ (asymmetric and symmetric C-H stretching frequency), and 3250 cm⁻¹ (–OH symmetric stretching) are increased with the reaction time [48,49,50]. The strong peaks at 1640 and 1473 cm⁻¹ were also increased, which correspond to the *CO and CH₂ (scissoring), respectively. The small peaks at 1266 and 1163 cm⁻¹ refer to the formation of C–C (stretching frequency) and CO₂^δ−, respectively [19]. The peaks at 1556 and 1425 cm⁻¹ correspond to the CO₃²⁻ and CHCOH respectively. These two peaks are gradually decreased by further increasing the reaction time. From these observations, we have illustrated the mechanistic pathway of CO₂ reduction given below.

In the above reaction steps, the CO₂ gas-phase is adsorbed as a monodentate metal carbonate (m-C–O–) CO₃²⁻. On the catalytic (Cu) surface, the CO₃²⁻ to *CO conversion occurs via π–π interaction by the excited electrons [46]. The local concentration of CO on the catalyst (Pd) undergoes the C–C coupling step, and then the OC–CO reacts with water and produces the intermediates of CHCOH. Furthermore, the addition of water with CHCOH produces the final product of ethanol [18]. Overall, DRIFT studies have demonstrated the reaction pathways wherein Cu was assumed as a CO₂ adsorption site. It is very difficult to separate the active sites in DRIFT spectra where the CO₂ is adsorbed. However, previous studies have demonstrated that CO₂ adsorption is more favorable on the Cu site [25].

On other hand, Pd was engaged in the C–C coupling reaction, as well as the highly selective hydrogenation of CO₂ to C₂H₅OH. The production of ethanol was confirmed by the mass spectra. The CO₂ reduction on Pd₂Cu/P25 was compared with P25. The corresponding DRIFT results are shown in Figure 6d. Hence, the Pd₂Cu/P25 is a more selective and active catalyst that produces ethanol when compared to P25. Moreover, the photothermal hydrogenation of CO₂ over the Pd₂Cu/P25 catalyst was evaluated at various temperatures under UV-Vis light irradiation (Figure 6e). The DRIFT difference spectra reveal that the C₂H₅OH formation (identified from C-H peaks arising) increased with an increase in temperature. In contrast, the reaction was continued without light irradiation at 150 °C, exhibiting no C–H peak, and with no ethanol formation in dark (Figure 6f). Therefore, the light provides the driving force for CO₂ reduction (0.611 V) and water oxidation (1.23 V). Overall, the reaction begins with the oxidation of water on P25 and then the electrons that evolved during water oxidation are utilized by Pd₂Cu. The H₂ molecules are generated from the P25 as a result of photocatalytic water oxidation. The Cu provides the active site for CO₂ adsorption and Pd oxidizes the generated H₂ molecules and involves the C–C bond formation. The key step of the CO₂ reduction is the hydrogenation of *CO to *HCO, which is performed by the Pd active sites.

4. Conclusions

This work reports the photothermal reduction of CO₂ and the oxidation of H₂O over the Pd₂Cu/P25 photocatalyst. The formation of ethanol was confirmed from GCMS and online mass spectroscopy. The Pd₂Cu/P25 showed an efficient CO₂ reduction photothermally at 150 °C with an ethanol production rate of 4.1 mmol g⁻¹ h⁻¹ under UV-Vis irradiation. Other products were not visible on GCMS spectra due to their low quantity. However, operando DRIFT absorption spectroscopy traces the intermediate products. The mechanism of CO₂ reduction on the Pd₂Cu/P25 catalyst was monitored by DRIFTS. The reaction begins with the oxidation of water on P25 and then the electrons involved in water oxidation were utilized by Pd₂Cu. The Pd active sites utilize those electrons and hydrogenate the *CO to *HCO, which is a key step for the reduction of CO₂ to ethanol. Moreover, the Pd₂Cu/P25 catalyst demonstrated the highly active and selective hydrogenation of CO₂ to C₂H₅OH.