In Water High Yield and Selectivity of CH₄ and H₂ Production from CO₂ Using UVC Light and a SiO₂-surface-modified TiO₂ Photocatalysts

Oswaldo Núñez ^1,2,*, Mohammad Fereidooni ¹, Victor Márquez ¹, Duangthip Sattayamuk ¹, Piyasan Praserthdam ¹, Supareak Praserthdam ³

1. Center of Excellence on Catalysis and Catalytic Reaction Engineering. Department of Chemical Engineering, Faculty of Engineering, Chulalongkorn University, Bangkok 10330, Thailand

2. Laboratory of Physical Organic Chemistry and Environmental Chemistry, Department of Processes and Systems, Universidad Simón Bolívar (USB), Caracas 1080A, Venezuela

3. Performance Computing Unit (CECC-HCU), Center of Excellence on Catalysis and Catalytic Reaction Engineering (CECC), Chulalongkorn University, Bangkok 10330, Thailand

* Correspondence: Oswaldo Núñez

Academic Editor: Anna Zielińska-Jurek

Special Issue: Green Photocatalysis for Degradation of Emerging Contaminants and Hydrogen Production

Received: August 25, 2023 | Accepted: November 28, 2023 | Published: December 05, 2023

Catalysis Research 2023, Volume 3, Issue 4, doi:10.21926/cr.2304029

Recommended citation: Núñez O, Fereidooni M, Márquez V, Sattayamuk D, Praserthdam P, Praserthdam S. In Water High Yield and Selectivity of CH₄ and H₂ Production from CO₂ Using UVC Light and a SiO₂-surface-modified TiO₂ Photocatalysts. Catalysis Research 2023; 3(4): 029; doi:10.21926/cr.2304029.

© 2023 by the authors. This is an open access article distributed under the conditions of the Creative Commons by Attribution License, which permits unrestricted use, distribution, and reproduction in any medium or format, provided the original work is correctly cited.

1. Introduction

Production of methane and hydrogen through CO₂ photoreduction is of great interest not only from the environmental point of view but also due to the generation of combustible gases that can be converted into value-added products such as Methanol. The last conversion could be achieved via thermo-catalysis with a low performance/price ratio. However, this conversion can be performed via photo-catalytic oxidation of low-concentration methane over decorated graphitic carbon nitride [1,2]. H₂ and CH₄ are sources of energy storage that can also be produced from CO₂. As suggested, co-mixing H₂ and CH₄ may double the energy storage capacity compared with pure hydrogen or methane storage [3]. Therefore, more energy can be obtained by storing the gases produced by CO₂ photo reduction with UV/solar light. CO₂ may then be considered as a feedstock. Large surface areas of 2D materials and conjugated microporous polymers (CMPs) allow the storage of combustibles such as H₂ and CH₄ and their energy storage conversion [4,5].

In water, CO₂ photo reduction conversion to CH₃OH, HCOOH, and CH₃COOH using visible light and 2D/2D zinc vanadium oxide-reduced graphene oxide (ZnV₂O₆/RGO) nanosheets has been recently reported; up to 5154 μmolg_cat^-1 of CH₃OH is produced [6]. This efficiency is attributed to the trapping and transporting of electrons by reduced graphene oxide (RGO). Also, 3488 μmolg_cat^-1 CH₃OH production has been obtained using ZnV₂O₆/RGO/g-C₃N₄ composite [7].

Reports on CH₄ and CO production of 225 μmolg_cat^-1h^-1 from CO₂ UV-photo reduction using MXene/Clay (2D/2D/2D O-C₃N₄/Bt/Ti₃C₂T_x) composite and up to 145 and 133 μmolg_cat^-1 h^-1 with WO₃/g-C₃N₄ and visible light represent quite high yields results [8,9]. CO₂ photoreduction using pg/C3N4/Ti₃AlC₃/TiO₂ composite has been reported to yield 91.9 and 4.97 mmolg_cat^-1h^-1 [10]. MXene based cocatalyst: TiO₂A/Ti₃C₂/TiO₂R MXene yield85 and 18 mmolg_cat^-1h^-1 of CO and CH₄, respectively [11]. S-scheme has been proposed for producing 2103.5 μmolg_cat^-1h^-1 of CH₄ with 96.59% selectivity using g-C₃N₄/TiO₂/TiAlC₂2D/OD/2D composite [12]. However, results using carbon-based semiconductors as g-C₃N₄ to generate CH₄ from CO₂ must be carefully interpreted since some CH₄ formed may be produced by the reduction of carbon source in the catalyst. Therefore, control experiments without CO₂ must be conducted to verify the non-formation of Methane or Methanol under these conditions [13,14,15]. Other studies reported isotopic labeling (¹³C and ¹⁸O) to evaluate carbon and oxygen sources unequivocally [16,17,18,19].

Non-carbon-based photocatalysts such as ZnFe₂O₄/Ag/TiO₂ nanorods have been successfully used to produce CO, CH₄ and CH₃OH via UV/visible photo reduction of CO₂ [20]; up to 1566 μmolg_cat^-1 CO yield has been obtained using 3D MAX Ti₃AlC₂/TiO₂ in a monolith photo reactor where efficient formation of CH₄ via reforming is also reported [21].

SiO₂-TiO₂ materials are used in very different fields [22]. As a photocatalyst, CO₂ photo reduction enhancements have been reported. However, in the cases said TiO₂ is either embedded in SiO₂ matrices or anchored on glass [23,24]. In fact, in these cases, photoredox reactions are faster than TiO₂ since the band gap is relatively larger (up to 4.11 eV) than the TiO₂ one (3.2 eV); therefore, the CB (Conduction band) reduction and VB (Valence Band) oxidation potentials increase. Nevertheless, SiO₂ coating TiO₂ P25 has also been used to reduce CO₂ to CO. Only CO and H₂ were detected, and CO selectivity increased compared to unmodified P25 [25]. Toluene degradation using novel mesoporous SiO₂-doped TiO₂ nanofibers has been reported [26]. Zirconium-doped TiO₂/SiO₂ and visible light have been used to degrade VOC, including Toluene [27]. Simulated-sunlight-activated photocatalysis of Methylene Blue (MB) using cerium-doped SiO₂/TiO₂ nanostructured fibers has also been reported [28].

Crystalline TiO₂ modified with tailored nanostructures has been extensively studied [29]; however, many properties associated with semiconductors’ atom arrangements influence their photocatalytic activity. Therefore, using amorphous TiO₂ (am-TiO₂) may bring essential advantages due to the porosity (high specific area), its high light-harvesting capacity, and the atom disorder that might induce charge separation. C-doped am-TiO₂ has been synthesized and used in the photo reduction of CO_2, obtaining CH₄ production of 4.1 μmolg_cat^-1h^-1 [30]; this catalyst showed a high specific surface area and narrowed band gap of 2.1 eV. Additionally, it has been pointed out that introducing oxygen vacancies on the C-doped am-TiO₂ can efficiently accelerate charge separation and transportation of photo-generated hole-electron pairs, enhancing photocatalytic performance [31]. However, am-TiO₂ has been investigated less than its crystalline forms, applications that do not require high temperatures have been reported. For instance, its antibacterial capacity has been studied [32]. Density-functional theory (DFT) calculations of structural and electronic properties of am-TiO₂ as a photocatalyst for hydrogen production have been published [33]. Mesostructured SiO₂-TiO₂ mixed oxides with enhanced stability, textural properties, and surface area in the 290-346 m²/g range have been synthesized [34].

We have identified some relevant aspects that may help to increase the CH₄ formation rate from HCO₃^-/CO₂ UV light photo reduction [35]. For instance, CO₂/HCO₃^- adsorption and the corresponding adduct concentration increases at pH equal and above P25 point of zero charge (PZC = 6.2) [36]; however, electron transfer from Ti to form the radical anion Ti⁺⁴-OCO₂^- requires protonation, that is also needed to produce H₂ via water splitting. Therefore, a TiO₂ surface modification must be oriented to increase the negative charge on the surface, but at the same time, more protonation capacity than TiO₂ at the catalyst surface is required. CO formation contribution from the oxygen vacancy (Vo) sites must also be minimized to increase CH₄ formation during the photocatalytic reaction.

To fulfill the three previous aspects, we have modified am-TiO₂ by attaching SiO₂ to its surface at room temperature and pressure. Since SiOH is a more vital Bronsted acid than TiOH, PZC of the SiOH at the surface-modified TiO₂ must be lower than unmodified TiO₂; therefore, at neutral pH, the catalyst surface must be negatively charged, which may improve CO₂ adsorption and its photoreduction rate; furthermore, to orient the photoreduction to more CH₄ and H₂ formation, the proton must also be available (SiOH) at the catalyst active site; therefore, using this surface-SiO₂ modified TiO₂ catalyst, CH₄ formation rate is expected to improve as compared to unmodified TiO₂ at neutral pH. Further rate improvement is likely when pH changes to the new motivation PZC value. The amorphous condition of the synthesized catalyst may also improve the CO₂ photoreduction rate due to increased surface area, light harvesting, and decreased e/h recombination.

In this work, amorphous TiO₂ was synthesized via the sol-gel method, and its surface was modified by its addition to a tetraethyl orthosilicate acid solution. Complete characterization of the solid product obtained is reported. It includes SEM-EDX, TEM, DRS, FT-IR, BET, XPS, XRD, PZC measurement, and Raman spectroscopy. Additionally, catalyst (supported on ITO) current density, de-trapping capacity, surface potential, and electron transfer processes to the electrolyte are also reported using cyclic voltammetry (CV) and open circuit potential decay (OCPD) measurements.

Our approach is to introduce modifications of catalysts without using organic compound additives and as simple as possible to increase CH₄ and H₂ production and reduce the cost of catalyst preparation. Herein, we report essential CH₄ and H₂ production improvement from CO₂/HCO₃^- using silicate surface modified amorphous TiO₂ and UVC (200-280 nm) light. A necessary amount of H₂ is also produced from water splitting. Some insights on the reaction mechanism obtained using electrochemistry and photooxidation measurements are also reported, pointing to the role of the (d-p) π-bonding at the Si-O bond and the non-formation of hydroxyl radical at the catalyst valence band.

2. Materials and Methods

2.1 Materials

The following chemicals and solvents were used: Degussa (P25, Aeoxide); sodium hydroxide (NaOH, 99.0%, Emsure); hydrochloric acid (HCl, 37%, Ensure); sodium phosphate monobasic dehydrate (NaH₂PO₄ 2H₂O, 99.0%, Acros Organics); sodium hydrogen carbonate (NaHCO₃, 99.0%, Fluka Chemika); acetic acid (CH₃COOH, 100%, Merck); titanium (IV) tetraisopropoxide (Ti[OCH(CH₃)₂]₄, 98%, Aldrich); tetraethyl orthosilicate (Si(OC₂H₅)₄, 98%, Aldrich); methylene blue (Sigma-Aldrich); Ethyl Alcohol (C₂H₆O, 99.9% QRëC) and water (H₂O, 17.7 MΩ/cm, Nanopure). The following gases were used: methane (CH₄, 99.99%, Linde); hydrogen (H₂); carbon monoxide (CO, 99.99%, Boc Gases); carbon dioxide (CO₂, 99.99%, Linde); helium (He, >99.99%, Linde); argon (Ar, >99.99%, Linde); nitrogen (N₂, >99.99%, Linde); air (>99.99%, Bangkok Industrial Gas).

2.2 Characterization, Photo Reduction, and Gasses Concentration Measurements

Semiconductor morphologies and crystalline structures were obtained by Scanning electron microscopy and energy-dispersive X-ray Spectroscopy (SEM-EDX). An Amicus photoelectron spectrometer with an Mg Kα X-ray source (20 mA and 10 keV) controlled by Kratos Vision2 software. C 1 s peak at 284.6 eV was used for calibration. Using a Lamda 650 UV/Vis instrument. A JEOL JEM-2100 Electron Microscope was used to obtain the Transmission Electron Microscope (TEM) images. Diffuse Reflectance UV-Vis Spectra (DRS) was collected using UV Win Lab software from PerkinElmer. Using Tauc’s equation, (αhν)^1/n = A(hν-Eg), the semiconductors band gap (Eg) was obtained. In the last equation, α is the absorption coefficient, h is Planck’s constant, ν and λ, light frequency and wavelength, A is an optical constant, and n = 0.5 for direct transitions. FT-IR Reflectance Spectroscopy (FT-IR) spectra were obtained in a Nicolet 6700 FT-IR instrument. A Horiba 4P-Fluoromax spectrofluorometer with a Xenon lamp (200-800 nm) was used to obtain the corresponding Photoluminescence (PL) spectra. The specific surface area of the catalysts was estimated by the Brunauer-Emmett-Teller (BET) method using a model Asap 2020, Micrometrics instrument. The point of zero charge (PZC) was obtained by plotting a solid catalyst pHi-pHf Vs in water. pH_i where pH_i and pHf are initial and final-equilibrium pHs; a Metter Toledo Seven multi pH-meter was used. Surface quantitative analysis was obtained using X-ray Photoelectron Spectroscopy (XPS) using an Amicus spectrometer at a voltage of 15 kV and current of 12 mA with Mg Kα X-ray radiation (1253.6 eV) and Al Kα X-ray radiation (1486.6 eV). Crystal structures were determined by X-ray Diffraction (XRD) using a Bruker D8 Advance instrument using a data collection (0.05°s^-1) at a 20°-80° angle range. The crystallite size was calculated using the Debye-Scherrer equation. Raman spectra were obtained using a Horiba Scientific LabSpec 6 instrument.

Photoreduction was performed in a cylindrical quartz reactor surrounded by six 16-watt Philips Germicidal Ozone UVC (200-280 nm) bulbs. A Shimadzu GC-14A instrument was used to quantify CH₄ reactor headspace with a FID detector provided with a Shincarbon ST packed column at 70°C in 10 min of analysis and using He as carrier gas and air as make-up gas with injector and detector temperatures of 150°C. H₂ and CO were measured using a Shimadzu GC-8A instrument with a TCD detector and molecular sieve 5A column at 70°C for 18 min. Ar was used as carrier gas with injector and detector temperatures of 150°C.

2.3 Synthesis

2.3.1 Amorphous TiO₂ (Am-TiO₂)

Following the sol-gel method, 5 mL of acetic acid in 50 mL of Ethanol (95%) solution previously stirred for 30 min and 6.3 mL (0.021 mol) of titanium tetraisopropoxide were added dropwise while stirring for 15 min. Stirring was continued for one hour [31]. 10 ml of the above mixture was left to dry in the air at room temperature. White powder of TiO₂ was collected and used directly to obtain am-TiO₂-SiO₂.

2.3.2 Silicate-Surface Modified TiO₂ Catalysts (am-TiO₂-SiO₂)

While stirring to a solution of 35 ml of Ethanol and 35 ml of distilled water, microliters of HCl solution (35%) were added until pH = 3 was reached. Then, (5 mmol) of tetraethyl ortho silicate was added to the solution. Stirring was continued for 6 h. 2 g of catalyst (P25 or am-TiO₂) was added to the final mixture; stirring continued for 1 h. The solution was dried at room temperature for 24 h. Catalyst-SiO₂ was air-dried at 110°C for 15 min.

2.3.3 Reuse am-TiO₂-SiO₂

Catalysts am-TiO₂-SiO₂ used for CO₂ photo reduction (pH = 3) were removed from the reactor, and the water-catalyst mix evaporated at 90°C until the dry catalyst was obtained (0.2 g). The 0.3 g of catalyst needed for the reaction was completed with the catalyst used at pH = 7. The obtained powder was used directly as a reused catalyst for CO₂ photoreduction.

2.3.4 Catalysts Supported on ITO

The ITO (Indium Tin Oxide) electrodes were treated with a KOH 0.05% solution and rotated at 1000 RPM to remove the excess of the critical two times before the catalysts' coating. A suspension of P25-SiO₂ catalyst in ethanol was prepared using 20 g of catalysts and 20 mL of ethanol and sonicated for 3 minutes (6 rounds, 30 s each time). am-TiO₂-SiO₂ gel was directly added to ITO. The ITO Electrodes were entirely covered with the suspension/gel and rotated at 1000 RPM to remove the excess material. The covered electrodes were air-dried and then treated at 150°C for 10 minutes. The Catalysts supported on ITO were stored in a closed flask for later use.

2.4 CO₂/HCO₃^- Photo Reduction

A 250 mL three-way cylindrical quartz reactor provided with a sand sparger for bubbling gas into the solution was used. Into this flask, a solution of [CO₂]_w = 34 mmol/100 mL by bubbling CO₂ on 100 mL of water or [HCO₃^-] = 2.3 mmol/100 mL and 0.3 g silicate surface-modified TiO₂ catalyst was added. pH (3-8) was adjusted using HCl, NaOH or 100 mg/L of NaH₂PO₄ buffer solutions. The reaction was started by turning on the light while stirring. Irradiation was continued for 4 hours. The reactor temperature was maintained at 30°C. The resultant reactor headspace gas samples were taken every 0.5 hours using a 2.5 ml gas syringe (injecting 0.5 mL) and analyzed using GC-FID and GC-TCD. CH₄, CO, and H₂ calibration curves were used to obtain their concentration during the kinetic runs [36]. Figure S1 shows the reactor setup for CO₂ photo reduction, and Figure S2 is a [CH₄] calculation example.

2.5 Electrochemical Experiments

All experiments were performed using an Autolab^® potentiostat/galvanostat M240, a Metrohm^® Ag/AgCl_(sat) reference electrode, a Platinized Titanium mesh as the counter electrode, and the catalysts supported on ITO as working electrodes. The solutions were prepared using Deionized Water (DI) (18 MΩ/cm²) and pH adjusted to 7 using a 0.1 M Na₂HPO₄/NaH₂PO₄ Buffer solution. Experiments were performed in a conventional three-electrode single-chamber cell. The solution was purged with high-purity nitrogen for 30 min before the measures. The concentration of NaHCO₃ used was 0.1 M. To measure Semiconductors´ Open Circuit Potential Decay (OCPD), a 0.8 V potential was applied for 20 s until the current stabilized. After that, the circuit was opened and the working electrode's potential was followed for 10 minutes [37]. To obtain Cyclic Voltammetry (CV) graphs, a scan rate of 10 mV/s was used [38].

2.6 Methylene Blue Photooxidation

Methylene blue (MB) Absorption Vs. a concentration calibration curve (Abs = 0.184 [MB] + 0.0047; R² = 0.998, at 664 nm) of 4, 2, 1, and 0.5 mg/L were used to follow MB photooxidation using am-TiO₂-SiO₂ and P25 catalysts.

In each batch, 5 mg of the catalyst was dispersed in 100 ml of 8 ppm MB solution. Before photo-irradiation, the suspension was stirred for 30 min in the dark to reach adsorption/desorption equilibrium. Samples were taken at different time intervals during the photocatalytic reaction, and the catalyst was immediately separated from the solution by a syringe filter. UV-vis spectra was taken, and MB degradation followed at 664 nm. Photodegradation pseudo-first-order rate constants were obtained from plots of ln [MB] Vs. Time. Exponential [MB]/[MB]o Vs. t plots and % photodegradation = ([MB]o - [MB]t)/[MB]o x 100 Vs. t was also obtained. pH was measured at the beginning and the end of each kinetic run.

3. Results and Discussion

3.1 Catalysts Characterization

Figures 1a and 1b show absorption spectra (DRS) and band gaps (Eg) of the synthesized silicate-surface modified catalysts obtained according to Tauc’s equation. These band gaps are equal to or lower than the corresponding P25 one (3.20 eV). For instance, the P25-SiO₂ band gap does not change compared to P25. However, amTiO₂-SiO₂ moves down to 3.13 eV (396 nm). This value, together with the Cu-TiO₂-SiO₂ one (3.06 eV, 405 nm), is in the violet visible region. Therefore, visible light-photo reduction activity is expected for these catalysts. Cu-P25-SiO₂ and N-P25-SiO₂ have been previously synthesized in our laboratory, and their band gaps are included just for comparison.

Figure 1 a Absorption spectra and b Plots of (K.hv)² vs. hv according to Tauc’s equation: (αhν)1/n = A(hν-Eg), n = 0.5. Obtained band gaps of P25-SiO₂ and am-TiO₂-SiO₂, are shown. Band gaps are also shown for comparison previously synthesized in our laboratory N-P25-SiO₂ and Cu-P25-SiO₂.

In Table 1, characterization parameters of SiO₂-surface modified TiO₂ catalysts are shown. am-TiO₂-SiO₂ and P25-SiO₂ catalysts show larger surfaces than P25. These differences may contribute to improving these catalysts´ photo-reduction performances. Table S1 shows BET results between am-TiO₂ and am-TiO₂-SiO₂. The last catalyst shows a larger surface area, pore size, and pore volume than the unmodified am-TiO₂ one. The following characteristic IR signals were observed for the two silicate-surface-modified catalysts: 670 cm^-1, Ti-O-Ti asymmetric stretching; 1060 cm^-1, Si-O-Si symmetric stretching; 960 cm^-1 Si-O-Ti stretching; 3750 cm^-1, SiO-H extending and a broadband 3700-3500 cm^-1 corresponding to water adsorbed on catalysts. These signals have also been identified in SiO₂-TiO₂ hybrid spheres and sol-gel TiO₂-SiO₂ films [39,40]. FT IR spectrum of am-P25, am-P25-SiO₂, am-TiO_2, and am-TiO₂-SiO₂ are shown in Figure S3. O-Si-O stretching signal at 1022-1080 cm^-1 is present only in the -SiO₂-modified catalysts.

Table 1 Catalysts band gap (Eg), EDX-SEM weight percentage (Wt %), surface area, pore size/volume, and IR signals.

Figure 2a and Figure 2b show photoluminescence spectra of am-TiO₂-SiO₂ and P25-SiO₂ catalysts using exciting light at 350 and 250 nm, respectively. More light collection efficiency is shown for am-TiO₂-SiO₂, especially when exciting light of more energy (UV at 250 nm) is used. It can be ascribed to the removal of electron capture centers on the surface of am-TiO₂ or the removal of nonradiative decay channels due to the SiO₂ shell around am-TiO₂ [41]. This light harvesting efficiency may improve this catalyst photo activity. Figure 2c shows SEM-EDX results of am-TiO₂-SiO₂. Agglomerations of irregular particle sizes and homogeneous Si and O distribution (EDX) at the catalyst surface spot are observed. TEM images of am-TiO₂-SiO₂ and P25-SiO₂ are shown in Figure S4. Agglomerations and amorphous morphology are detected for amTiO₂-SiO₂. Meanwhile, crystalline morphology for P25-SiO₂ is revealed. According to EDX, prepared am-TiO₂-SiO₂ contains an average of 53% Ti, 5% Si, and 41% O wt% at the surface. We will use the active site of this surface structure to depict the catalytic aspects covered in this work. Figure 2d shows a PZC value of 4.1 for am-TiO₂-SiO₂ obtained from pH_i-pH_f Vs—pH_i plot.

Figure 2 a 320 nm excitation photoluminescence spectra of am-TiO₂-SiO₂ and P25-TiO₂ catalysts. b 250 nm excitation. c SEM-EDX of am-TiO₂-SiO₂. d am-TiO₂-SiO₂ PZC measurement.

The oxidation state of the constituent elements in the synthesized catalysts was determined using X-ray photoelectron spectroscopy (XPS), and the results are presented in Figures 3a, 3b, and 3c. The XPS spectrum of Ti 2p in Figure 3a was deconvoluted into two peaks with binding energies of 458 eV and 463.7 eV, attributed to Ti 2p3/2 and Ti 2p1/2, respectively. This confirms an oxidation state of +4 for Ti. Additionally, the Si 2p spectrum (Figure 3b) peaked at a binding energy of 102.2 eV, which was attributed to the Si-O-Ti bond. Furthermore, in the O 1 s XPS spectrum (Figure 3c), three deconvoluted peaks located at 529.8 eV, 530.9 eV, and 532 eV were attributed to Ti-O, Si-O-Ti, and surface hydroxyl groups, respectively.

Figure 3 Ti (a), Si (b), and O (c) XPS spectra of am-TiO₂-SiO₂. (d): XRD spectrum. (e) Raman spectrum.

In Figure 3d, the XRD spectrum of am-SiO₂-TiO₂ is shown. Only one broad signal at ca. 27.5 2θ is observed. This signal is typical of amorphous TiO₂, where no crystal TiO₂ patterns are observed. Due to the small amounts of Si (5%) present, the corresponding signal at ca. 58 2θ is barely detected in Figure 3d [40].

Figure 3e shows a sharp signal at ca. 3000 cm^-1 Raman shift corresponding to O-H stretching. This band is hidden in the corresponding FTIR due to the presence of water. Broad movements centered at 1400, 700, and 170 cm^-1 Raman shifts can be identified as titanium species' framework [42]. A signal at 80 cm^-1 Raman shift observed in Figure 3e could be assigned to the framework of Si since it is absent in the am-TiO₂ Raman spectrum (Figure S5).

3.2 Blanks and Alternative Carbon Sources

As shown in Figure 4, CH₄ is formed even without adding any CO₂ source (CO₂ or NaHCO₃). This alternative source could come from the catalyst synthesis or reactor walls' remaining organic materials. Special care is taken to eliminate any solvent from the catalyst by heating the solid catalyst while stirring at 80°C for 24 h before use. Likewise, the glass reactor is cleaned after use using successive 30 min washing of ultrasound (H₂O), KOH (EtOH), H₂SO₄ (H₂O₂), and HCl (30%). The maximum CH₄ observed after 1 h is ca. 0.5 nmol/mL H₂O g_cat. As observed in Figure 4, the contribution of this source is relatively low compared to the amount of CH₄ produced from CO₂ reduction. Using a new reactor, blanks with catalyst, CO₂ source, and lamp off and without catalysis, CO₂ source and lamp on were run. Under these conditions, [CH₄] < 0.05 nmol/mL H₂O g_cat was detected.

Figure 4 CH₄ photoreduction formation without and with CO₂ source at two different pHs.

3.3 HCO₃^-/CO₂ Photoreduction

P25 CO₂/HCO₃^- photo reduction proposed mechanism to form CH₄, CO, and H₂ in the reactor headspace consists of a fast formation of monodentate adduct followed by a rate-limiting adduct radical anion formation [35]. Since the Point of zero charge (PZC) of the silicate surface modified TiO₂ catalysts a ca. 2.9 [43] and 4.1 (Figure 2d) for the am-SiO₂-TiO₂ synthesized in this work, relatively lower values than the PZC (6.2) of P25, negative charge at the silicate surface must be present even at acid pH. Since a negative charge on one of the oxygen (-Si-O^-) remains, adduct formation is favored when carbonate or CO₂ is present (Scheme 1, center-top). Nevertheless, one -Si-OH is still prorogated at pH ca. 5; therefore, the proton required in the e^-transfer from Ti to produce the corresponding radical anion (Scheme 1, left-bottom) is available at the active site. Since this transfer has been pointed out as the photo reduction rate-determining step (rds), faster product formation may be observed when these silicate-surface-modified TiO₂ catalysts are used [35].

Scheme 1 TiO₂-SiO₂ carbonate adduct formation and electron transfer to form the corresponding adduct radical anion followed by fast product formation.

Figure 5a and Figure 5b show CH₄ and H₂ production in the photo reduction of HCO₃^- using the silicate-surface modified TiO₂ catalysts synthesized in this work. As expected, am-TiO₂-SiO₂ is the best catalyst, producing ca. 8 times more methane and H₂ than P25 in a non-buffered solution. Given this result in H₂ production, we performed water splitting using am-TiO₂-SiO₂ (Figure 5c). As shown in the figure, H₂ is formed with a rate of 2.3 nmol/mL_H2O·g_cat·h, that is 0.23 μmol/g_cat·h. This means that am-TiO₂-SiO₂ can also be used by adding water at pH = 3 to obtain H₂ at a relatively reasonable rate. Scheme 2 shows the proposed mechanism for water splitting and H₂ formation when the carbonate adduct is present at the catalyst surface.

Figure 5 a reactor headspace [CH₄] produced in water UV photo reduction of 100 mg of NaHCO₃ using 0.3 g of catalyst under non-buffered conditions. b [H₂] production. See Table 2 for the reaction's initial pH values. c [H₂] formation rate via water splitting, without any carbonate source added at pH = 3. d Reactor headspace [CH₄] and [H₂] produced in water UV photo reduction of 3.4 mmole/100 mL of CO₂ using 0.3 g of am-TiO₂-SiO₂ catalyst at pH = 3. Re-used catalysts results ([CH₄] and [H₂]) at pH = 4.6 are also shown.

Scheme 2 H₂ production proposed mechanism for am-TiO₂-SiO₂. With CO₂/HCO₃^- added (X = O - CO₂^-) and without carbonate source added (X = O) (water splitting).

Table 2 Production and yield of CH₄ and H₂ in the HCO₃^-/CO₂ photo reduction using am- TiO₂-SiO₂. P25 is included as a reference.

Figure 6 CH₄, H_2, and CO produced in the photocatalyzed UV reduction of CO₂ at pH = 3.

Since the reactor [CO₂]_w = 3.4 mmol/100 mL when using 1 atm of CO₂ pressure, faster CH₄ and H₂ productions than when NaHCO₃ is used (2.3 mmol/100 mL) are expected. However, as shown in Figure 5d, when am-TiO₂-SiO₂ is used, CH₄ is produced 80 times faster (8.9 μmol g_cat^-1h^-1) than P25. This is quite a remarkable result for a solid-liquid condition since this rate is even more significant than when Pt-TiO₂ is used; it is reported at 1.4 and 5.2 μmolg_cat^-1h^-1 using solid-liquid and solid-gas conditions [44]. For hollow anatase, single crystal and mesocrystal values of 0.2, 1, and 1.7 μmolg_cat^-1h^-1 have been reported for solid-gas needs [45]. Using carbon-doped amorphous TiO₂, 4.1 μmolg_cat^-1h^-1 has been reported [32]. A similar rate of 8.9 μmolg_cat^-1h^-1 has been obtained using 2%MgO-Pt-TiO₂ under solid-gas conditions [44]. This value is one of the best yields reported for CO₂ photo reduction for non-carbon-based photocatalysts. In the kinetics shown in Figure 5d, pH = 3, CH₄ is produced more significantly than H₂. This is quite an important result since at pH = 8 (Figure 5a and 5b), H₂ is favored instead. This means that acid pH favors CH₄ formation.

In Figure 5d, reuse am-TiO₂-SiO₂ results are also shown. As depicted, the CH₄ rate (3 μmolg_cat^-1h^-1) decreases, and the H₂ rate increases (0.6 μmolg_cat^-1h^-1) compared to catalysts used for the first time at pH = 3. This difference may be due to the difference in pH: 4.6 Vs. 3. As reported in Table 2 at pH = 5, CH₄ rate = 5 μmolg_cat^-1h^-1 and H₂ rate = 0.1 μmolg_cat^-1h^-1. However, catalyst deactivation cannot be discarded since the reused catalyst was obtained via evaporation of water at 90°C for several hours.

To study the H₂/CH₄ selectivity change with pH in more detail, we performed additional experiments in the 5-7 pH range using NaHCO₃ and NaH₂PO₄ as buffers. Two critical remarks can be pointed out from Figure S6a's and Figure S6b's results. As observed for pH = 5, without buffer there is more selectivity towards CH₄. It is the maximum rate of CH₄ produced using NaHCO₃. A 5 μmolg_cat^-1h^-1 is obtained. A relatively more significant rate than the reported values for other catalysts in solid-liquid or solid-gas conditions [32,44,45]. However, this rate is reduced if NaH₂PO₄ buffer is used at the same pH = 5. This means that an inhibition induced by the phosphate buffer occurs. Probably due to the additional consumption of CO₂ via the equilibrium: CO₂ + H₂O + Na₂HPO₄ = Na₂CO₃ + H₃PO₄. Therefore, adding phosphate buffer decreases CH₄ and H₂ formation rates, as experimentally observed. However, since H₂ is also formed without CO₂ (Scheme 1), CH₄ formation is more affected than H₂ (Figure S6c and Figure S6d) when buffer is added. These results can be pointed out as experimental evidence for the Scheme 2 mechanism in which H₂ and CH₄ are formed in two paths via partitioning competition of the radical anion formed in the rds. For practical purposes, selectivity CH₄/H₂ can be reversed by adding buffer to the solution and changing pH. Table 2 resumes the H₂ and CH₄ produced (μmolg_cat^-1h^-1) of am-SiO₂-TiO₂ catalysts under different photo reduction conditions. Water splitting is also shown.

As further evidence for the mechanisms proposed in Scheme 1 and Scheme 3, we measured [CO] and compared its formation rate with [CH₄] and [H₂] ones at pH = 3 (Figure 6). We have also measured product ratios ([CH₄]/[H₂]) with time at two different pHs to obtain the kinetic (t = 0) and thermodynamic main products using ratios extrapolation to t = 0 and the slope of the ratios Vs. Time plots (Figure S7 left and right).

Scheme 3 CH₄ Vs. H₂ formation in the CO₂/HCO₃^- reduction using am-TiO₂-SiO₂. CH₄ rate and selectivity increase by lowering the pH since more protons (4) are required to produce CH₄ than the one needed (2) to yield H₂.

As shown in Figure 6, mainly methane is produced using CO₂ and at pH ca 3. As depicted in the figure inset, some H₂ and CO are also formed in comparable ca. 8 nmol/mL H₂O gcat concentrations after 240 min. From the [CH₄]/[H₂] plot shown in Figure S7 (left), it can be interpreted that both products are formed from the same intermediate that cleaves preferentially to form CH₄ and H₂ (Scheme 3) with some cleavage through the CO path. Although, CH₄ and H₂ are produced from the cleavage of the same intermediate, CH₄ is thermodynamically (t = 240 min, [CH₄]/[H₂] = 40) and kinetically (t = 0, [CH₄]/[H₂] = 28) more favorable at pH = 3. However, at pH = 8, H₂ is favorably formed. As shown in Figure S7 (right), at t = 0, H₂ and CH₄ are formed in equal concentration, although H₂/CH₄ ratio increases up to 3 after 90 min, probably due to the limited amount of H⁺ at this pH. In fact, after 90 min, the ratio slope changes, indicating more than one mechanism for H₂ formation: water splitting (Figure 5c) and Scheme 3. Since in the water splitting mechanism CH₄ is not produced, this mechanism contributes more at the initial [H₂]/[CH₄] reaction times (positive slope, Figure S7 (right)): meanwhile, Scheme 3 one contributes at final reaction times (slightly negative slope, Figure S7 (right)).

3.4 Electrochemical Measurements

Figure 7a shows CVs of P25 and am-TiO₂-SiO₂ coated on ITO. The current intensity is larger for P25 than am-TiO₂-SiO_2, but its de-trapping (positive current) is slower [46]. Less current and slow de-trapping in am-TiO₂-SiO₂ is associated with the catalyst surface capacity to retain a charge. SiO₂ at the surface efficiently includes e; therefore, de-trapping to the solution becomes relatively slow. In the plot of E Vs. t embedded in Figure 7a, an OCPD with smoother decay is observed for am-TiO₂-SiO₂. Figure 7a (embedded) and reported values in Table 3 show that am-TiO₂-SiO₂ halftime is 122 s. Meanwhile, in P25, two slopes are observed with halftimes of 19 s and 103 s. The fast process in P25 corresponds to P25 electron transfer to the electrolyte from its CB and the slow one to electron transfer from oxygen vacancy (Vo) traps. We have reported these two mechanisms, and this OCPD measurement confirms our previous proposal [35]. Another aspect worth mentioning from Figure 7a is the equilibrium potential (surface potential [47]) reached after decay by the two catalysts; as observed in the figure and reported in Table 3, am-TiO₂-SiO₂ reaches a less negative potential than P25. Although the former catalyst´s PZC value (4.1) is relatively lower than the P25 one (6.2), there is more surface negative charge at pH = 7 for P25. This counterintuitive result can be explained in terms of better water solvation in am-TiO₂-SiO₂ compared to P25.

Figure 7 a CV of ITO-P25 and ITO-am-TiO₂-SiO₂. Inset: corresponding open circuit potential decays, after applying a potential of -0.8 V. b ln (E_∞-E_(t)) Vs. time of ITO-P25 and ITO-am-TiO₂-SiO₂. Rates (straight-line slopes) are shown. Values are summarized in Table 3. c CV of ITO-am-TiO₂-SiO₂ and ITO-am-TiO₂-SiO₂ adding NaHCO₃^- to the solution. Inset: corresponding open circuit potential decay after applying a potential of -0.8 V.

Table 3 Rate and half-life time values of potential decay ITO-P25, ITO-am-TiO₂-SiO_2, and ITO-am-TiO₂-SiO₂ with NaHCO₃ added.

These experiments were performed at pH 7 using NaH₂PO₄ buffer with reduced dissolved O₂ using previously blowing N₂ into the solution; therefore, electron from de-trapping are used to reduce water in the solution. If an additional electron acceptor is added, we expect to observe changes in the corresponding OCPD. As observed in Figure 7c, when NaHCO₃ is added to the solution, the capacity of am-TiO₂-SiO₂ to retain charge increases. Also, as shown in Figure 7c, the embedded plot of E Vs. t, the surface charge value obtained is more damaging than the am-TiO₂-SiO₂ and P25 ones (Table 3). This can be interpreted as an additional trap present when NaHCO₃ is added. The Adduct shown in Scheme 1 (top right) then acts as a trap that makes more electrons available for photo reduction at the CB. As the figure shows, the destabilization of the two negative charges is diminished since the distance between them in the adduct decreases. This charge accumulation capability is one of the factors contributing to the efficient methane and hydrogen production when am-TiO₂-SiO₂ is used.

3.5 Photo Reduction Vs. Photo Oxidation

We have used the bleaching induced by OH radical on methylene blue (MB) to test its production by photooxidation of water at the TiO₂ VB [49,50]. Figure S8a and Table 4 show that quenching effectively occurs when using P25 but does not proceed when am-TiO₂-SiO₂ and P25-TiO₂-SiO₂ are used. We conclude that OH^. is not an intermediate in the photooxidation of water. O₂ is then preferentially produced directly or produced from peroxide. The reduction potential for one-electron oxidation to produce hydroxyl radical from water is 2.81 V [51], a higher value than the VB potential of am-TiO₂-SiO₂ (1.93 V) and P25-SiO₂ (2.48 V) [52]. Therefore, one electron transfer is thermodynamically unfavorable; It is only the driven force of the 4 e oxidation to yield O₂ (0.81 V at pH = 7) [52] that allows the first electron transfer to occur, but the product (OH^-) does not accumulate at all. However, the 2 e and 4 e oxidation to produce H₂O₂ and O₂ are feasible since the corresponding potentials at pH = 7 are 1.34 V and 0.81 V, respectively [53]. Conditions for water splitting (H₂O = H₂ + 1/2O₂) without a cocatalyst or sacrificial reagent are then provided for am-TiO₂-SiO₂ since the conduction band reduction potential is -1.0 V [53]. Meanwhile, H⁺/H₂ at pH 3 and 7 are 0.17 V and -0.41 V, respectively. The non formation of hydroxyl radicals is an excellent advantage for water splitting and CO₂ photoreduction since a sacrificial reagent is unnecessary. Some MB degradation is observed when HCO₃^- is added to the MB solution (Figure S8b and Table 4). Adding HCO₃^- increases the negative charge on its surface, as shown in the surface potential values of Table 3. Increasing the negative charge on the surface induces down-bending on the valence band [54]. Therefore, the VB potential becomes more positive, allowing the one-electron transfer from water to produce a hydroxyl radical that MB quenches. Figure S9 describes the RedOx processes at the am-TiO₂-SiO₂ surface.

Table 4 Methylene Blue (MB) ([MB] = 8 ppm) photooxidation pseudo-first-order rate constants, half times and degradation efficiency after 30 min UV light irradiation. Adding [HCO₃^-] = 4 ppm, last column) significantly improves the activity of less active silicate surface-modified catalysts.

The more negative surface charge also favors positively charged organic adsorptions. Efficient MB removal and E. coli bacteria inactivation have been reported using P25-SiO₂. In both cases, an effective negative (P25-SiO₂)-positive (MB and protein) interaction is claimed as the leading cause for the improvement in photooxidation rate [55]. The adsorption process is then crucial in these photocatalytic reactions. Likewise, it may play a role in MB's binding inhibition of ACE2 (angiotensin-converting enzyme 2) to SARS-COV₂-RBD [56]. The more negative protein the ACE2 active site is, the more efficient MB antiviral activity is predicted.

4. Conclusions

We have prepared silicate-surface modified amorphous TiO₂ (am-TiO₂-SiO₂) and P25 (P25-SiO₂) catalysts. These catalysts yield more CH₄ and H₂ than P25 using UV light. In fact, in 4 h, using 3.4 mmol of CO₂ (1 atm)/100 mL H₂O and 0.3 g of am-TiO₂-SiO₂, 80 times more CH₄ (8.9 μmolg_cat^-1h^-1) than P25 at pH= 3 is produced and 8 times more H₂ (0.84 μmolg_cat^-1h^-1) at pH = 7 when 2.3 mmol of NaCO₃/100 mL is used with selectivity’s of 71% and 53%, respectively. To our knowledge, 8.9 μmolg_cat^-1h^-1 of CH₄ production in water from CO₂ photo reduction is one of the higher rates reported when using non-carbon-based photocatalysts. H₂ is also formed just by adding water to the am-TiO₂-SiO₂ catalyst. H₂/CH₄ selectivities can be modulated by changing pH and/or using a buffer. H₂ production is favored at neutral pH, and CH₄ at acid pH. Buffer NaH₂PO₄ addition diminishes CH₄ production and barely affects H₂ formation. Due to the low PZC value (4.1) of am-TiO₂-SiO₂ as compared to P25, at pH < 7, faster photo reduction rates are observed since there is a negative charge (-SiO^-) on the surface that promotes CO₂ adsorption, but there is also a proton available (-SiOH) at the active site. Therefore, once the carbonate-catalyst adduct is formed, the electron transfer from Ti to produce the radical anion can be catalyzed by -SiOH. Other factors may also be favoring am-TiO₂-SiO₂ performance as photo-reduction catalyst. Cyclic voltammetry and open circuit potential decay experiments show less intense current and less de-traping for am-TiO₂-SiO₂ than P25, and there is only one smooth decay slope for the former and two slopes for the last catalyst. Therefore, the attached Si at the surface also acts as an external intramolecular trap (EIT) that promotes electron transfer from Ti. When HCO₃^- is added to the solution and the carbonate-catalysts adduct is formed at the surface, a more negative surface potential is measured, meaning that the carbonate at the surface also facilitates the electron transfer from Ti. Additionally, due to the disordered distribution of atoms at the amorphous TiO₂, less e/h recombination occurs. For instance, this catalyst also shows a high light harvesting efficiency with P25-SiO₂. A larger surface area than P25 may also contribute. Meanwhile, CH₄ and H₂ further oxidations at the catalyst interface are avoided in the photo reduction reactions due to the relatively large hydrogen and methane gas-water coefficients, which allow them to accumulate in the reactor headspace.

P25 is a better catalyst than the silicate-modified TiO₂ catalysts for MB photooxidation. The less positive VB potential in the last catalyst makes the OH radical formation from water oxidation unfavorable. Therefore, there is no hydroxyl radical to be quenched by MB. However, when NaHCO₃ is added to the solution, these catalyst´s rates increase more than 10 times at neutral pH. The valence band down-bending induced by the more negative surface charge allows some OH radical formation and its corresponding quenching by MB.

The results and mechanism shown in this work have contributed to developing new procedures to increase CO₂ photo-reduction as is the case of coated films at the reactor headspace.

Acknowledgments

This research project is supported by the Second Century Fund (C2F), Chulalongkorn University.

Author Contributions

Oswaldo Núñez: Designing, directing, interpreting and writing manuscript. Mohammad Fereidooni: Synthesis, characterization, photo RedOx experiments and revising manuscript. Victor Márquez: Electrochemical experiments. Duangthip Sattayamuk: Synthesis and characterization. Piyasan Praserthdam: Directing and financing. Supareak Praserthdam: Preliminary calculations and revising manuscript.

Competing Interests

The authors declare no competing financial interest.

Additional Materials

1. Figure S1: Photo reactor setup and CO₂ photo reduction procedure.

2. Figure S2: GC-FID and concentration calculation example.

3. Figure S3: FT IR spectrum, top to bottom of am-P25, am-P25-SiO₂, am-TiO₂ and am-TiO₂-SiO₂. O-Si-O stretching signal at 1022-1080 cm^-1 is present only in the -SiO₂ catalysts.

4. Figure S4: 80000 x and 200000 x TEM images. Left: am-TiO₂-SiO₂. Right: P25-TiO₂-SiO₂.

5. Figure S5: am-TiO₂ Raman spectrum.

6. Figure S6: a and c Reactor headspace [CH₄] and [H₂] produced in water UV photo reduction of 100 mg of NaHCO₃ using 0.3 g of am-TiO₂-SiO₂ at different initial pH (without buffer); b and d Reactor headspace [CH₄] and [H₂] produced in water UV photo reduction of 100 mg of NaHCO₃ using 0.3 g of am-TiO₂-SiO₂ using NaH₂PO₄ buffer.

7. Figure S7: Left: [CH₄]/[H₂] ratio Vs. time at pH = 3. Right: [H₂]/[CH₄] ratio Vs. time at pH = 7.4.

8. Figure S8: a 8 ppm of MB photo oxidation using silicate surface-modified TiO₂ catalysts. b ln (C/Co) Vs. time plot of MB UV photo oxidation using P25-SiO₂ and am-TiO₂-SiO₂ catalysts. Pseudo-first order rate constants are found from the corresponding slopes and reported in Table 4.

9. Figure S9: am-TiO₂-SiO₂ RedOx photo-activity. Si acts as an external intramolecular trap (EIT) promoted by the (d-p) π-bonding at the Si-O bond. e/h recombination is then reduced. CH₄ is produced 80 times faster than when P25 is used and H₂ is produced 8 times faster. Only O₂ is produced at the valence band due to its relatively low potential with a value lower than the OH^-/H₂O one. However, when adding HCO₃^- the negative surface charge increases promoting VB down-bending and some OH is produced that is quenched by MB.

10. Table S1: am-TiO₂ Vs. am-TiO₂-SiO₂ BET results.