Photocatalytic CO₂ Conversion Using Anodic TiO₂ Nanotube-Cu_xO Composites

Abstract

Nanosized titanium dioxide (TiO₂) is currently being actively studied by the global scientific community, since it has a number of properties that are important from a practical point of view. One of these properties is a large specific surface, which makes this material promising for use in photocatalysts, sensors, solar cells, etc. In this work, we prepared photocatalysts based on TiO₂ nanotubes for converting carbon dioxide (CO₂) into energy-intensive hydrocarbon compounds. Efficient gas-phase CO₂ conversion in the prepared single-walled TiO₂ nanotube-Cu_xO composites was investigated. Parameters of defects (radicals) in composites were studied. Methanol and methane were detected during the CO₂ photoreduction process. In single-walled TiO₂ nanotubes, only Ti³⁺/oxygen vacancy defects were detected. The Cu²⁺ centers and O₂⁻ radicals were found in TiO₂ nanotube-Cu_xO composites using the EPR technique. It has been established that copper oxide nanoparticles are present in the TiO₂ nanotube-Cu_xO composites in the form of the CuO phase. A phase transformation of CuO to Cu₂O takes place during illumination, as has been shown by EPR spectroscopy. It is shown that defects accumulate photoinduced charge carriers. The mechanism of methane and methanol formation is discussed. The results obtained are completely original and show high promise for the use of TiO₂-Cu_xO nanotube composites as photocatalysts for CO₂ conversion into hydrocarbon fuel precursors.

1. Introduction

An increase in the concentration of atmospheric CO₂ is considered one of the main sources of irreversible and dangerous climate change [1,2]. Therefore, conversion of CO₂ into useful products is a globally important objective. Different approaches are being developed for CO₂ conversion [3]. One promising approach is photocatalytic reduction of CO₂ [4]. The issue is that most of the known photocatalysts are too inefficient [5]. Among others, photocatalysts based on nanostructured TiO₂ have such important advantages as chemical stability and photocorrosion resistance, and suitable electron energy spectrum [6,7]. TiO₂ is also non-toxic material. Titania nanotube arrays (TiO₂ NTs) have large specific surface area. Due to their geometry and surface structure, they can promote faster diffusion of photocatalytic reaction products, directional charge transfer, and lower recombination rates of photoinduced charge carriers in comparison to TiO₂ nanoparticles [8,9,10]. Highly ordered TiO₂ NTs arrays can be formed by electrochemical oxidation of Ti [6,11,12]. TiO₂ NTs formed electrochemically in ethylene-glycol-based electrolyte contain an inner layer (IL) enriched with organic products, and an outer layer (OL) of mostly pure TiO₂ [13,14]. To improve catalytic activity and obtain crystalline TiO₂ structure, as-formed TiO₂ NTs arrays are annealed in air. After annealing, the inner layer of a nanotube has a mesoporous structure with a mixture of anatase and rutile nanocrystals with an average size of ~7 nm with the presence of carbon species [15,16]. The outer layer of a nanotube consists of practically stoichiometric anatase [15]. One of the issues of nanostructured TiO₂ catalysts is low absorption of visible light due to a wide band gap. Several ways have been proposed to improve the photocatalytic activity of TiO₂-based materials under visible-light illumination [6]. Creation of nano-heterojunctions with Cu_xO species is an effective method [17,18,19,20,21,22,23].

Photocatalytic reduction of CO₂ in water and alkaline solutions was shown in [24] using TiO₂-Cu_xO composites with different Cu content deposited onto a molecular sieve. It was demonstrated that these composites enhanced CO₂ photoreduction with selective formation of oxalic acid. Furthermore, methanol and acetic acid were observed on the molecular sieve combinate system. In [25], computational modeling and experimental studies were conducted to elucidate the role of Cu sites in promoting CO₂ reduction on TiO₂. The nanocomposites demonstrated significantly higher activity than bare TiO₂ in photocatalysis, and the activity was found to be related to the chemical state of surface Cu+ species. According to modeling studies, the highly dispersed Cu sites likely contributed to the improved photocatalysis by stabilizing surface adsorption of CO₂ on TiO₂. The experimental and modeling studies further confirmed the direct involvement of surface Cu sites in CO generation via CO₂ reduction. Additionally, using TiO₂-Cu_xO heterojunctions can promote photocatalytic decomposition reactions of phenol and acetone [26] and other organic compounds [27,28]. In [29], it was shown that an optimum CuO and TiO₂ ratio was necessary to achieve the highest efficiency of methanol conversion to methyl formate. In addition, in [30], the authors describe the TiO₂-protecting effect of Cu₂O from photocorrosion during CO₂ reduction in a saturated solution.

The review [31] has shown that, due to the smaller band gap of metal oxides, such as Cu_xO, the light absorption capacity of TiO₂ can be improved in heterojunctions. Methanol production was reported in CO₂ photoreduction process when TiO₂ is doped with CuO that provides the highest yield. Cu₂O is named in [31] as very promising and the most frequently used dopant candidate for the photoreduction of CO₂. It has a band gap ranging from 2 to 2.2 eV that absorbs visible light effectively.

From our point of view, the enhancement of photocatalytic activity in TiO₂-Cu_xO heterojunctions compared to nanostructured TiO₂ is due to effective absorption of visible light in Cu_xO nanoparticles, and charge separation at the heterojunction. The latter reduces the recombination rate of photoexcited electrons and holes.

Activity and selectivity of a TiO₂-Cu_xO catalyst depends on the method of its synthesis [17,18,19,20,21,22,23,32]. Moreover, the efficiency and selectivity of CO₂ conversion by TiO₂-Cu_xO catalysts varies during the conversion process [17,18,19,20,21,22,23,32]. One of the possible origins is the reduction of CuO to Cu₂O. Additionally, the properties of TiO₂-Cu_xO heterojunctions depend on the catalyst’s microstructure. The efficiency of methanol production on TiO₂-NTs-Cu_xO photocatalysts with different sizes and compositions of copper oxide particles has been studied in [32]. The output of methanol was significantly different for catalysts with different x and size of Cu_xO particles. The results were explained by the dependence of electron energy spectrum on copper oxide phase composition and by the dependence of methanol production rate on the size, shape and distribution of Cu_xO particles [32].

To our knowledge, there is a very limited number of articles reporting the study of the photocatalytic activity of TiO₂ NTs arrays obtained in ethylene-glycol-based electrolyte and TiO₂ NTs-Cu_xO arrays in the CO₂ conversion process in a gas phase in the presence of water vapor [7,9,33]. Moreover, the effect of multilayered structures and the role of inner layers was not discussed much in these publications.

In our previous publication, in [34] we reported a comparative study of the photocatalytic properties of TiO₂ multi-walled and single-walled nanotube arrays for photoinduced CO₂ conversion in the gas phase, their structures and their defect properties. Electron paramagnetic resonance (EPR) spectroscopy was applied to observe the structural defects in the outer and inner layers of TiO₂ nanotubes and analyze their effect on the photocatalytic properties of the nanotubes [35]. The present work aims to study the structural, electronic and photocatalytic properties of TiO₂ NTs/Cu_xO composites obtained from TiO₂ NTs arrays without IL by Cu_xO deposition applying successive ionic layer adsorption and reaction (SILAR). This section may be divided by subheadings. It should provide a concise and precise description of the experimental results, their interpretation and conclusions that can be drawn.

2. Results and Discussion

2.1. Samples Morphology and Composition

Morphology of SW TiO₂ NTs/Cu_xO has been studied using a scanning electron microscope equipped with energy-dispersion X-ray spectroscopy. SEM images of investigated samples are presented in Figure 1.

As can be seen from SEM images, the size of deposited Cu_xO particles increases with amount of deposition cycles. It is also noticeable that deposition in the described conditions leads to formation of islands and non-homogeneous coverage of the surface of SW TiO₂ NTs arrays by Cu_xO nanoparticles.

The size range of deposited nanoparticles were 9–21 nm, 11–36 nm and 14–40 nm for SW TiO₂ NTs/CuxO-10/30/60 samples, respectively. With an increase in the amount of deposition cycles, nanoparticles form agglomerates while the sizes of nanoparticles change slightly and do not depend significantly on the amount of cycles. A characteristic feature of the SW TiO₂ NTs/Cu_xO-60 sample is the presence of nanoparticles with the shape of needles or sheets with the length from 100 nm to 500 nm and 30 to 40 nm thickness. It is well noticeable that Cu_xO deposition inside the tube does not take place, i.e., only the top surfaces of the SW TiO₂ NTs arrays are modified during deposition.

As it is seen from the energy dispersive X-ray microanalysis (

Table 1. EDX microanalysis data for SW TiO₂ NTs arrays with different copper content.

	SW TiO2 NTs	SW TiO2 NTs/ CuxO-10	SW TiO2 NTs/ CuxO-30	SW TiO2 NTs/ CuxO-60
Ti, at. %	31.9	32.9	34.0	31.4
O, at. %	55.7	58.1	56.6	58.5
F, at. %	0.3	0.7	0.0	0.4
C, at. %	4.4	6.4	3.8	6.0
Cu, at. %	0.0	2.2	3.0	3.3
Al, at. %	0.5	0.3	0.2	0.3

), the amount of copper at the SW TiO₂ NTs arrays’ surface varies non-linearly depending on the amount of deposition cycles, and reaches about 3.3 at. % after 60 deposition cycles.

The obtained SW TiO₂ NTs and SW TiO₂ NTs/Cu_xO-60 samples were investigated by X-ray diffraction analysis (Figure S1). The copper oxide phase was not recorded by XRD, because of the amount of the CuxO is too small to be registered by this method.

2.2. Optical Properties

The diffuse reflectance spectra depend on the amount of copper oxide, as presented in Figure 2.

According to calculations performed using the Kubelka-Munk theory, the optical band gap is similar for all samples under study: 3.1 ± 0.1 eV for SW TiO₂ NTs samples and 3.3 ± 0.1 eV for NTs/Cu_xO samples. All NTs/Cu_xO samples showed higher light absorption in visible areas compared to pure SW TiO₂ NTs. The samples of SW TiO₂ NTs/Cu_xO-30 possess the most efficient visible light absorption from all samples under study.

2.3. Photocatalytic Conversion of CO₂

Prepared samples were investigated in situ during CO₂ conversion in gas phase in the presence of water vapor. The products of photoelectrochemical reactions were detected by chromatograph equipped with a plasma ionization detector (PID). The outcome of reaction products was calculated per unit of specific surface area for each sample 7.7 × 10⁻³ m²/cm². The specific surface area has been calculated in the frame of a simple geometric model using scanning electron microscopy data (Figure S2); the methodology was described in [34].

Three LEDs operating at 370 nm main wavelength were used for illumination. Illumination intensity at the sample was about 23 mW/cm². The measurement process was carried out in four stages: (1) pumping of helium through the reactor during 10 h; (2) cleaning of the sample surface from organic impurities in helium flux under LED illumination; (3) photoinduced conversion of CO₂ (helium flux replaced by CO₂ flux); (4) registration of relaxation after turning off illumination. Gas flux was set to 1.1 mL/min. Results are presented in Figure 3. Methane and methanol were detected as main reaction products. Acetaldehyde, ethanol, formic acid and acetone were detected in helium flux under illumination, but their amount did not increase after CO₂ supplement. The kinetics of acetaldehyde formation is presented in the Supplementary Material, Figure S3.

Only two compounds increased in concentration in the presence of CO₂ flux: methane and methanol. This is evidence of photoinduced CO₂ conversion. One can see that maximum methanol outcome was observed for the SW TiO₂ NTs sample while minimum methane output was observed for this sample. Deposition of copper oxide shifts the selectivity of the conversion process to methane production. Output of methane reaches its maximum for the SW TiO₂ NTs/CuxO-30 sample. Increasing the number of deposition cycles from 10 to 60 led to an increase in methanol output. However, the rate of methanol production for the samples with deposited copper oxide remains smaller than for the SW TiO₂ NTs sample.

Variation of CO₂ conversion selectivity can be explained by the difference of methane and methanol formation potentials and changing of conduction band potentials of TiO₂ in contact with CuO. Potentials of reactions of methane (1) and methanol (2) formations are equal to −0.24 V and −0.38 V versus normal hydrogen electrode (NHE), respectively [36].

$$\begin{array}{c}{\mathrm{CO}}_2+8{\mathrm{H}}^{+}+8{\mathrm{e}}^{-}\to {\mathrm{CH}}_4{+\mathrm{H}}_{2}O\\ {\mathrm{E}}_{\mathrm{Redox}}=-0.24\mathrm{V} \mathrm{vs}. \mathrm{NHE}\end{array}$$

(1)

$$\begin{array}{c}{\mathrm{CO}}_2+6{\mathrm{H}}^{+}+6{\mathrm{e}}^{-}\to {\mathrm{CH}}_3{\mathrm{OH}+\mathrm{H}}_{2}O\\ {\mathrm{E}}_{\mathrm{Redox}}=-0.38\mathrm{V} \mathrm{vs}. \mathrm{NHE}\end{array}$$

(2)

According to photoelectrochemical data, the conduction band bottom potential in TiO₂ is more negative compared to methanol reaction potential [37]. Therefore, reactions (1) and (2) can take place at the surface of TiO₂. Redistribution of charge carriers between two semiconductors occurs after deposition of copper oxide on the surface of anodic titania nanotube arrays. This leads to the bending of energy bands in semiconductors (Figure 4). The band edges positions of copper oxides are indicated according data in the literature [18,33,38].

As a result of the band-bending, the conduction band potential in TiO₂ becomes more positive than the reduction potential for the reaction (2). The latter can block the reaction. Illumination, in turn, shifts energy band potentials in TiO₂ to more negative values due to the migration of photoelectrons from CuO to TiO₂. This can lead to an increased probability of the reaction occurring (2). Variation (increase) of an amount of copper oxide leads in turn to the enhancement of the concentration of photogenerated charge carriers and thus to more significant flattening of energy band in TiO₂.

This mechanism makes the shift of CO₂ photoconversion selectivity towards methane production, caused by deposition of copper oxide on TiO₂ NTs arrays, understandable. One can suggest that the conduction band potential of TiO₂ in NTs/Cu_xO-10 sample is appropriate for CO₂ photoconversion to methane but not to methanol. For the NTs/Cu_xO-30 sample, the conduction band potential is also not negative enough for the reaction (2). However, methane output increases due to larger photoelectron concentration separated at the heterojunction of two semiconductors. Deposition of a larger amount of copper oxide (NTs/Cu_xO-60 sample) makes band flattening under illumination large enough for the reaction (2). However, efficiency of methanol production for this sample is smaller than for non-modified sample SW TiO₂ NTs. This fact can be linked to the impossibility of water decomposition on the surface of copper oxide due to insufficient positive potential of the top of valence band. As a consequence, hydrogen ions and hydroxyl ions can be formed only on the surface of TiO₂. The free surface of TiO₂ of NTs decreases with an increase in the amount of copper oxide deposition cycles resulting in a reduction in the rate of CO₂ photoconversion to methanol.

2.4. Electron Paramagnetic Resonance

The nanocomposites obtained were further investigated by electron paramagnetic resonance spectroscopy (EPR) since defects (spin centers) play an important role in photocatalytic processes [6]. The main type of defect in initial SW TiO₂ NTs is Ti³⁺/oxygen vacancy (g₁ = 1.9961, g₂ = 1.9697). Figure 5a shows the EPR spectra of SW TiO₂ NTs in the dark and under illumination. It can be seen (Figure 5a) that, upon photoexcitation, the intensity of the EPR signal (I_EPR) increases slightly.

Figure 5b shows an example of the EPR spectrum of the SW TiO₂ NTs/Cu_xO-30 nanocomposites. This spectrum is a superposition of several EPR signals. First, a strong EPR signal from copper ions Cu²⁺ (g = 2.1612) [39] is detected (Figure 5b), which indicates the presence of the CuO phase. We also observed on the right side of the EPR spectrum the signal from O₂⁻ radicals (g₁ = 2.029, g₂ = 2.009, g₃ = 2.003). The appearance of O₂⁻ radicals can be explained by the adsorption of oxygen molecules on oxygen vacancies on the surface of TiO₂ and, probably, on the surface of copper oxide nanoparticles, followed by the capture of electrons from the conduction band. Upon illumination, the EPR signal intensity from Cu²⁺ decreases while EPR signal intensity from O₂⁻ radicals increases (Figure 5b). We suppose that paramagnetic defects Cu²⁺ pass into a non-paramagnetic state (Cu²⁺ + e⁻ → Cu⁺), which leads to a diminution of corresponding EPR signal (Figure 5b). At the same time, the capture of a photoinduced electron by an oxygen molecule leads to an increase in the EPR signal from O₂⁻ radicals: O₂ + e⁻ → O₂⁻. The generation of Cu⁺ ions under illumination indicates the formation of Cu₂O phase.

Notice that we did not observe the EPR signal in SW TiO₂ NTs/Cu_xO nanocomposites from the Ti³⁺/oxygen vacancy centers, probably against the background of strong EPR signals from copper ions and oxygen radicals. Under illumination, they are also not recorded, as these defects probably capture the photoexcited hole and pass into a non-paramagnetic state: Ti³⁺ + h → Ti⁴⁺. Therefore, in SW TiO₂ NTs/Cu_xO nanocomposites the defects accumulate the charge necessary for redox reactions. We also investigated the behavior of spin centers after illumination. Figure 5b shows EPR spectrum of SW TiO₂ NTs/Cu_xO-30 one hour after illumination (spectrum 3). It can be seen that the EPR signal intensity from Cu²⁺ increases a little but does not reach the value before illumination, while EPR signal intensity from O₂⁻ radicals strongly decreases to the value before illumination. Therefore, a phase transformation of CuO to Cu₂O takes place during illumination. A reduction of O₂⁻ radicals after illumination can be due to spending radicals in redox reactions.

2.5. Secondary Ion Mass Spectroscopy

To understand the nature of copper oxide alternation on TiO₂ surface, two parts of the same sample were studied before and after illumination by means of secondary ion mass spectroscopy (SIMS).

It has been found that emission of CuO⁻ and CuO²⁻ (normalized to total amount of detected ions) is much higher from the sample before illumination compared to emission after illumination. Moreover, there is a difference in the ratio of amount of emitted CuO²⁻ and CuO⁻ ions before and after illumination. This difference can point to the different stoichiometry of copper oxides on the surface of the samples before and after illumination. One can suggest that the amount of oxygen bound to copper is larger before illumination. The data obtained provide evidence to suggest that copper oxide on the surface of the samples is most likely in the CuO phase before illumination. However, after illumination, copper oxide transforms to a less oxidized Cu₂O phase, which is consistent with the EPR data. We suppose that copper oxide operates as a source of photogenerated electrons. Their quantity controls potentials of energy bands in TiO₂.

2.6. Role of Cu₂O in Photocatalytic Conversion of CO₂

Taking into account transformation of CuO to Cu₂O phase, one can present an energy diagram of the structure as an energy diagram of heterostructure consisting of three semiconductors. Transformation of CuO to Cu₂O can take place at the TiO₂/CuO boundary and at CuO free surface (Figure 6). The change of copper oxide phase will not lead to significant shift of Fermi level position due to the small amount of Cu₂O formed compared to the amount of titania. For both possible positions of Cu₂O phase shown in Figure N, the amount of photoelectrons transferred to TiO₂ will be reduced due to their additional recombination at the CuO/Cu₂O boundary or accumulation in the CuO phase followed by CuO transformation to Cu₂O and to Cu [32]. Therefore, elevating of Fermi level caused by illumination will be smaller in the structure containing Cu₂O than in the structure with CuO phase only. Thus, the probability of methanol formation should be reduced due to CuO to Cu₂O phase transformation. The probability of methane formation can also be reduced but less significantly due to the less negative potential of methane formation.

Based on the data obtained, one can suggest that the efficiency and selectivity of CO₂ conversion in the Cu₂O-CuO-TiO₂ system is determined by the number of photoelectrons transferred to TiO₂. This number determines the rise of Fermi level in TiO₂ under illumination and its position with respect to potentials for CO₂ to methane or methanol conversion. At high concentrations of photoelectrons, the position of the Fermi level under illumination is high enough for effective conversion of CO₂ to methanol (2). At lower concentrations of photoelectrons, the position of the Fermi level is too low for effective methanol production. In this case, methane production (1) dominates.

3. Materials and Methods

Synthesis of single-wall (SW) nanotubes of anodic titania is described in [34]. SW TiO₂ NTs/Cu_xO nanocomposites were prepared by deposition of copper oxide on the surface of titania nanotubes arrays using successive ionic layer adsorption and reaction (SILAR). An aqueous solution of CuCl₂·2H₂O served as ionic source. An aqueous solution of 25% NH₄OH (Russian Specifications GOST 3760-79, Sigma Tec, Khimki, Russia) was added to increase the pH to 10. Mixture of ethanol with deionized water in 1:3 proportion heated up to 70 °C was used as the source of anions. The SILAR method was carried out in three stages. In the first stage, the sample was inserted into an aqueous solution of copper chloride containing [Cu(NH₃)₄]²⁺ ions for 30 s. In the second stage, the sample was inserted into a mixture of ethanol with deionized water for 7 s. In the third stage, the sample washed in deionized water for 30 s.

Variation of the amount of copper oxide Cu_xO deposited on SW TiO₂ NTs was performed by variation of cycles of ionic layer formation (10, 30 and 60 cycles were carried out for samples SW TiO₂ NTs/Cu_xO-10, SW TiO₂ NTs/Cu_xO-30 and SW TiO₂ NTs/Cu_xO-60, respectively). After deposition, the samples were heated up to 300 °C at a rate of 30 °C/min and kept at 300 °C in air for 60 min.

The surface morphologies and compositions were studied using a Helios G4CX (Thermo Fisher Scientific, Waltham, MA, USA) scanning electron microscope equipped with an EDS attachment (EDAX Octane Elite Super).

Spectra of diffuse light reflection from SW TiO₂ NTs and SW TiO₂ NTs/Cu_xO samples were recorded on LS-55 Perkin Elmer spectrometer (Waltham, MA, USA) which allows for the registration of the diffuse light scattering from the sample surface in a spectral range from 200 to 800 nm with spectral slit widths from 2.5 to 20 nm.

Electron paramagnetic resonance (EPR) spectra were recorded on a Bruker ELEXSYS E500 EPR spectrometer (Billerica, MA, USA) (X-band). The samples were illuminated directly in the cavity of the EPR spectrometer with the light of a BRUKER ELEXSYS ER 202 UV high-pressure mercury lamp (50 W). The photoexcitation intensity of the samples was 40 mW/cm².

Secondary ion mass spectroscopy was produced by TOF-SIMS IV manufactured by IONTOF GmbH, Muenster, Germany, and clusters of bismuth ion (Bi³⁺) were used.

The details of the methodology of the photocatalytic conversion of CO₂ were presented earlier in the work [34]. The photocatalytic CO₂ conversion was provided under the following conditions: reactor temperature 30 °C, relative humidity 5%, total CO₂ (ISO 14175-C1) gas flow 1.2 mL/min. The analysis of the gas products was carried out using an HP PLOT/Q capillary column Hewlett Packard HP 5890 Series II Gas Chromatograph-FID (Agilent Technologies, Santa Clara, CA, USA).

4. Conclusions

Single-walled TiO₂ nanotube-Cu_xO composites for converting carbon dioxide (CO₂) into hydrocarbon compounds were prepared. Methanol and methane were detected during the CO₂ photoreduction process. The production rate of gas-phase CO₂ conversion and photoinduced transformations of defects in the prepared photocatalysts were investigated. Only Ti³⁺/oxygen vacancy defects were detected by EPR spectroscopy in initial TiO₂ nanotubes. Cu²⁺ centers and O₂⁻ radicals were detected in the TiO₂ nanotube–Cu_xO composites. Under illumination, the capture of a photoinduced electron by oxygen molecules led to an increase in the EPR signal from O₂⁻ radicals. It was shown by EPR technique that Cu_xO nanoparticles are present in the TiO₂ nanotube-Cu_xO composites in the form of the CuO phase. Under illumination, a phase transformation of CuO to Cu₂O was observed. These results are consistent with data obtained by secondary ion mass spectroscopy. The shift of CO₂ photoconversion selectivity was observed towards methane production due to the deposition of copper oxide on the surface of TiO₂ nanotubes. The mechanism of methane and methanol formation in the TiO₂ nanotubes with different copper oxide content is discussed. The obtained results are completely novel and show a high level of importance for the development of energy-efficient photocatalysts for CO₂ conversion into hydrocarbon fuel precursors.