Photocatalytic Reduction of CO₂ Over Me (Pt, Pd, Ni, Cu)/TiO₂ Catalysts

The TiO₂ powder (Degussa, P-25) was purchased from a commercial supplier. The TiO₂ supported catalysts were prepared by traditional wetness impregnation method using aqueous solutions of nitrate salts (for Ni and Cu catalysts), chloroplatinic acid, and palladium(II) chloride. The estimated quantities of the precursor solutions were added dropwise to TiO₂ powder and, then, the obtained mixture was left at ambient overnight. Next, the excess of the solvent was evaporated under a vacuum. Finally, the impregnated samples were dried at 110 °C for 4 h and, afterward, they were calcined at 300 °C for 4 h in an air atmosphere. The nominal metal loading was as follows: 0.5–5% wt. Cu and Ni, 0.5–1% wt. Pt and Pd. Prior the physicochemical measurements (except reducibility) and photocatalytic activity tests, the obtained samples were reduced in hydrogen stream (5% H₂–95% Ar) for 4 h at 120 °C for copper catalyst, at 300 °C for nickel catalyst, and at 200 °C for palladium and platinum samples.

Surface area, pore volume, and pore size distribution were determined from the nitrogen adsorption/desorption isotherms at − 196 °C, using a Sorptomatic 1900 (Carlo Erba Instruments). The samples were previously outgassed at 300 °C and equilibrated under vacuum for at least 4 h before measuring the adsorption–desorption isotherm. Specific surface areas were evaluated from the measured monolayer capacity (Brunauer–Emmett–Teller method) using the range of relative pressure from ~ 0.05 to 0.33 and the value for nitrogen cross-section 0.162 nm².

The phase composition of pristine TiO₂ and TiO₂-supported catalyst was investigated by X-ray diffraction analysis. The diffraction patterns were collected using a PANalyticalX'Pert Pro MPD diffractometer in Bragg Brentano reflection geometry. The diffractometer was equipped with Cu Kα radiation source (λ = 1.5418 Å). Data was collected in the 2θ range of 5°–90° with a step size of 0.0167° and exposure per step of 27 s. Because the raw diffraction data contain some noise, the background during the analysis was subtracted using the Sonneveld, E. J. and Visser algorithm. The data were then smoothed using a cubic polynomial function.

The secondary ions mass spectra were recorded with a TOF-SIMS IV mass spectrometer manufactured by IONTOF GmbH, Muenster, Germany. The instrument is equipped with Bi liquid metal ion gun and high mass resolution time of flight mass analyzer. Secondary ion mass spectra were recorded from approximately 100 µm × 100 µm area of the spot surface. During measurement analyzed area was irradiated with the pulses of 25 keV Bi³⁺ ions at 10 kHz repetition rate and an average ion current 0.3 pA. The analysis time was 30 s for both positive and negative secondary ions giving an ion dose below a static limit of 1 × 10¹³ ions/cm². Secondary ions emitted from the bombarded surface were mass separated and counted in time of flight (TOF) analyzer. During the analysis, a low-energy electron flood gun was used to neutralize the charge on the sample surface. The samples were pressed into pellets before measurements

The reducibly of samples was studied in an automatic TPR system AMI-1 in the temperature range of 25–800 °C with a linear heating rate of 10 °C min⁻¹. Samples (weight about 0.05 g) were reduced in hydrogen stream (5% H₂–95% Ar) with a volumetric flow rate of 40 cm³ min⁻¹. Hydrogen consumption was measured as a function of time using a thermal conductivity detector (TCD).

Surface concentrations of acidic and basic sites were determined by temperature-programmed desorption of ammonia (NH₃-TPD) and carbon dioxide (CO₂-TPD). Before TPD experiments, the samples were plugged with helium at 600 °C for 60 min in order to remove any contaminations. After cleaning, the samples were cooled and saturated for 20 min. in flow of pure NH₃ at 50 °C or alternatively CO₂ at 50 °C. In both cases, the total flow rate was 25 ml/min. Then, the samples were purged in helium flow until a constant baseline level was attained. TPD measurements were performed in the temperature range 100–600 °C at a rate of 10 °C/min using helium as carrier flow. The evolved ammonia or carbon dioxide were detected by an on-line TCD calibrated by the peak area of known pulses of NH₃ or CO₂.

Infrared spectra were recorded with an IR Tracer-100 FTIR (Shimadzu) spectrometer equipped with liquid nitrogen cooled MCT detector. A resolution of 4 cm^-1 was used in collecting all spectra. 128 scans were taken in order to achieve a satisfactory signal to noise ratio.

Photocatalytic measurements were carried out in a continuous flow reactor equipped with a cooling jacket and a quartz window (Fig. 1). The amount of catalyst used for the photocatalytic reaction was 0.2 g for each run. The rector was also equipped with two paired mercury lamps POLAMP LRF 400W. The temperature of the reactor was maintained at 35 °C and the total flow of reaction mixture (5% vol. H₂O/95% vol CO₂) was equal to 0,8 ml/min. The first steady-state activity measurements with and without ultraviolet irradiation were taken after at least 2 h on stream.

The analysis of reaction products was performed by means of an on-line gas chromatograph equipped with a flame ionization detector and a VF-5 ms column. The yield of organic compounds was calculated according to the following equations:

The reducibility of modified TiO₂ photocatalysts was investigated with the aim to determine the optimum temperatures at which the metal oxides nanoparticles are being reduced. This is necessary to avoid excessive heating during the reduction process which can lead to the agglomeration of deposited metal particles. According to literature, the reduction of pristine TiO₂ occurs at a temperature of 400–720 °C. However, in our case, no detectable hydrogen consumption peaks were detected in this temperature range (Fig. 2).

The TPR profiles of Cu/TiO₂ catalysts showed two overlap peaks (P1 and P2) located in the temperature range from 200 to 250 °C which can originate either from stepwise reduction of CuO to metallic Cu via formation of Cu₂O intermediate or reduction of highly dispersed copper oxidizes (P1) and bulk-like CuO phases (P2) [10, 11].The reduction of nickel oxide spices occurs at a higher temperature of around 250–350 °C. Similarly, as in the case of copper samples, two peaks (P3 and P4) can be distinguished with the maxima at 270 and 300 °C, respectively. The observed hydrogen consumption peaks can be due to either different interaction of nickel oxide species with TiO₂ support or kinetic limitation of the process itself. In the case of palladium and platinum catalysts, no hydrogen consumption peaks were observed in the investigated temperature range. The negative peak observed at about 80 °C in the TPR profile of Pd/TiO₂ is attributable to the decomposition of palladium-hydride formed at room temperature at the beginning of the analysis [12, 13]. It is worth mentioning that the absence of hydrogen consumption peak, in this case, does not exclude the presence of the oxidized palladium species as their reduction can occur below room temperature in reducing atmosphere [14].

Based on the results obtained from TPR-H₂ measurement, it was proposed to perform the mild reduction of copper and nickel catalysts at 120 and 300 °C, respectively. At such conditions, some amount of metal nanoparticles remain in the form of oxide which is also photocatalytically active [15, 16]. Also, it was reported that the coexisting of both Cu⁺ and Cu⁰ facilitates the separation of electron–hole pairs and promotes electron transfer to adsorbed CO₂ [5]. The palladium and platinum samples were reduced at 200 °C to eliminate the chloride species, originating from metal precursors. The oxidation state of investigated samples before and after reduction was evaluated using the TOF-SIMS technique. It is assumed that differences in the oxidation state of metal nanoparticles affect the emission intensity of MeO⁻ ions containing metal atom Me bonded to oxygen. Negative ions were chosen for analysis as they mainly form by fragmentation of the structures present on the surface of the catalyst whereas positive ions can be formed by recombination of the atoms emitted from distant sites on the catalyst surface [17]. Table 1 presents intensities of MeO⁻ secondary ions recorded for the as-received and reduced catalysts. From the obtained result, it can be seen that the emission intensity of MeO⁻ ions decreased for Cu, Pt and Pd catalyst after reduction in the hydrogen atmosphere thereby indicating that these catalysts were almost completely reduced at given conditions. Since no differences between NiO⁻ ion emission intensity were observed for as-received and reduced Ni/TiO₂ catalysts, it can be concluded that the nickel nanoparticles did not undergo reduction.

The XRD measurements did not confirm the presence of copper and nickel oxides probably due to their low concentration (Fig. 3). The XRD patterns of the investigated samples showed the presence of characteristic bands of rutile and anatase phases of TiO₂. The estimated ratio of anatase to rutile was approximately 5:1. Generally, among the various crystalline phase of titania, anatase exhibits a better photocatalytic activity. However, it is worth mentioning that anatase TiO₂, with a small addition of either rutile or brookite phases, demonstrates enhanced photocatalytic activity compared to pure anatase due to the improved electron and hole separation [11, 18].

The results of BET measurement showed that the pristine TiO₂ had a specific surface area (SSA) of about 57 m²/g which is similar to that reported in the literature [19]. The SSA of modified TiO₂ catalysts was found to vary insignificantly depending on the type of metal and its amount (Table 2).

The catalytic properties of catalyst are also influenced by the presence of acidic and basic surface sites. Within the Lewis theory framework, Ti⁴⁺ ions are potential electron charge acceptors, i.e. acids, while O²⁻ anions play the role of conjugated bases. Acid–base interactions between TiO₂ surface and incoming gaseous mixture may lead to activated complexes and promote further reactions. At the same time, stable surface species can be formed that can block the catalytic process. Therefore, the investigation of acid-base properties of TiO₂ and modified TiO₂ samples is of vital importance to understand their catalytic activity.

The strength of acidic and basic sites can be determined by deconvolution of TPD peaks, considering both peak area magnitude and central temperature of each peak. The following types of strength of surface sites were proposed: weak (desorption in the temperature range from 100 to 300 °C), medium (300–450 °C), and strong (450 − 600 °C). According to the results of TPD-NH₃/CO₂ measurements, the TiO₂ catalysts are acidic in nature (Table 3). The pristine TiO₂ was characterized by the highest amount of acidic sites (527.2 NH₃µmol/g_cat) most of which were of high strength. The concentration of these surface sites was found to decrease after the deposition of metal particles. This may be attributed to the partial coverage of TiO₂ surface by these particles. In the case of the catalysts with high metal loading (of about 1%), the deposited particles are likely to be agglomerated which leads to larger support surface exposed to NH₃ molecules and, thus, to a higher amount of acidic sites.

The similar results were obtained for basic sites. Upon deposition of metal on the surface of TiO₂, the concentration of basic centers decreased, and the biggest decrease was observed for the catalysts with metal loading of 1%. As discussed earlier, such result might be due to partial masking of TiO₂ surface. It’s also worth mentioning that the decrease of existing basic sites in TiO₂ was likely to be offset by the generation of new oxygen vacancies during the reduction of investigated catalysts [20]. As shown in Table 4, the reduction of pristine TiO₂ in the hydrogen atmosphere led to an increase in the total basicity of its surface from 59.6 to 94.8 CO₂ µmol/g_cat.

The key steps of the photocatalytic reduction of carbon dioxide are the adsorption and subsequent activation of reaction reagents. According to Benkoula et al. the water vapor can be adsorbed via four different mechanisms among which the dissociative adsorption at O²⁻vacancies is recognized as a dominant path [21]. Similarly, the adsorption of CO₂ first proceeds at oxygen vacancies and then, after filling all the oxygen vacancies, at Ti⁴⁺ sites [22]. Therefore, the reduction in the number of basic sites due to deposition of metal can be considered as a negative effect due to decrease of number of centers available for reactants adsorption.

The analysis of the post-photocatalytic reaction mixtures revealed the presence of methanol and small amounts of methane. The other organic compounds and CO could also be formed during the process but were not detected due to apparatus limitations. Therefore, the methanol was chosen as a marker compound to evaluate the photocatalytic activity of TiO₂ and metal modified TiO₂ samples.

The photocatalytic activity of semiconductor depends on numerous factors, including its energy band gap, recombination of formed electron–hole pairs, adsorption properties, crystallinity, and so forth. The results of photocatalytic measurements showed the TiO₂ itself was almost inactive in the process, the methanol formation rate was about 0.027 µmol/g h. In comparison to pristine TiO₂, the metal modified samples displayed higher photocatalytic reaction rate for methanol formation (Fig. 4). Among the catalyst with metal loading of 0.5%, the most active samples were platinum (0.59 µmol/g h) and copper (0.17 µmol/g h) samples. Such results are in contradiction with those reported previously. Particularly, Nasir Shehzad et al. reported that Cu/TiO₂ photocatalyst is more efficient in liquid-phase reduction of carbon dioxide while Ni/TiO₂ in the gas-phase process [23].

In order to verify the obtained results, the additional series of copper and nickel catalysts were prepared with higher metal concentrations. In general, for all investigated samples, the photocatalytic activity gradually decreased with an increase in the content of metal phase (Fig. 5).The exceptions were only 1.5% and 2% Ni/TiO₂ that demonstrated a relatively high rate of methanol formation above 0.24 µmol/g h. It is well known that the metal deposited on the surface of semiconductor can serve as an efficient electron trap preventing electron–hole recombination and thereby favoring the photocatalytic activity. However, above an optimum concentration of metal, the completely opposite effect is observed that significantly limits the photocatalytic activity of samples.

As discussed earlier, the water vapor can compete with CO₂ for surface site adsorption, thus hampering the photocatalytic reduction process. Therefore, it was attempted to perform long-term stability tests of investigated samples for 3.5 h under the same operating conditions as those previously reported. The obtained results revealed the gradual decrease in photocatalytic activity of samples with increasing reaction time (Fig. 6).

For instance, the photocatalytic rate of methanol formation over 2% Ni/TiO₂ decreased form 0.27 (2 h on reaction stream) to 0.17 µmol/g h (3.5 h). The regeneration of catalysts, performed at 300 °C for 2 h in the reaction mixture, showed that they exhibit comparable activities as those observed at the beginning of the photocatalytic process. Such results indicate that the reaction products are likely responsible for the reversible deactivation of photocatalysts. In order to confirm this assumption, the in situ IR measurements of 2% Ni/TiO₂ were performed during adsorption of methanol (0.1% methanol in Ar mixture) followed by subsequent purging with Ar either under UV light or upon heating up to 300 °C (Fig. 7).

The adsorption of methanol on the surface of investigated sample resulted in an appearance of characteristic bands at 2951, 2926, 2897, 2847, and 2824 cm⁻¹. According to the literature, the bands at 2926, 2897, 2824 cm⁻¹ can be assigned to the methoxy groups while at 2847 and 2951 cm⁻¹—to formate species [24, 25]. The irradiation of the sample with UV light led only to a slight attenuation in this frequency range. Also, significant intensification of bands at 2359 and 2341 cm⁻¹ were observed which originate from asymmetric stretching vibration of gaseous CO₂. Such result implies that only small amount of methanol was photodegraded to carbon dioxide. The more prominent decrease in absorption intensity of high-frequency bands was observed only upon heating the sample indicating the decomposition and/or desorption of methanol. Therefore, it can be assumed that the formed methoxy species during the photocatalytic process remain mostly on the surface of the catalysts which lead to their reversible deactivation.