Artificial photosynthesis converts CO₂ into hydrocarbon fuel under solar irradiation, and is attracting increasing interest. It exploits a photocatalyst medium, in contrast to natural photosynthesis where this is achieved within green plants. Artificial photosynthesis can potentially overcome the shortage of sustainable energy, and solve the environmental problems associated with CO₂ emission, but is hindered by low conversion efficiencies. Considerable progress is required to make the photocatalytic reduction of CO₂ practical and economical ^{[1, 2, 3, 4, 5, 6]}.

The photocatalyst is the key factor in the photocatalytic reduction of CO₂, and high conversion efficiency requires an efficient photocatalyst. Semiconductors are attractive photocatalysts because their bandgap is comparable to the energy of solar light, so they have been widely studied ^{[7, 8]}. In 1979, Inoue et al. ^[9] reported the photocatalytic reduction of CO₂ into organic compounds in the presence of a semiconductor, under irradiation by a Xe lamp. Many subsequent studies have used various semiconductors to convert CO₂ with water into hydrocarbons under irradiation ^{[10, 11]}. TiO₂ has been the most widely studied semiconductor, because of its high stability, strong oxidation activity, favorable optical and electronic properties, and efficient charge transport ^{[12, 13, 14, 15]}. CO₂ has a high thermodynamic stability, and significant energy is required to break its C=O bond ^[1]. An efficient photocatalyst should ideally harvest the maximum possible amount of solar light. TiO₂ has a bandgap of ~3.2 eV, which corresponds to the ultraviolet region. This constitutes only 3%‒5% of the solar spectrum, which limits the photocatalytic efficiency of TiO₂ ^[16]. The favorable recombination of photo-generated charge carriers also suppresses the photocatalytic efficiency of TiO₂ ^[17]. The potential of the TiO₂ valence band occupied by photo-generated holes is highly positive, which may oxidize the previously reduced products back to CO₂ ^{[10, 18]}.

Much effort has focused on addressing these limitations ^[19]. Modifying TiO₂ with a narrow bandgap semiconductor can extend its absorption into the visible region, and facilitate photo-generated electron-hole separation. Cu₂O is a semiconductor with a bandgap of 2.0‒2.2 eV ^{[20, 21]}. It is widely applied as a photocatalyst, because it is nontoxic, widely available at low cost, and has a high absorption coefficient ^[22]. The optimum design of the band configuration is important for enhancing the photocatalytic reduction efficiency ^[23]. Both the conduction and valence bands of Cu₂O are at higher energy level than their respective bands in TiO₂. Thus, excited electrons accumulate in the TiO₂ conduction band, while holes accumulate in the Cu₂O valence band. The TiO₂/Cu₂O system has appropriate band edge alignments that facilitate the separation of photo-generated charge carriers ^{[24, 25]}. The oxidation potential of holes from TiO₂ also decreases upon transferring into Cu₂O, which can hinder re-oxidation, and achieve the same effect as sacrificial hole scavengers ^{[26, 27]}. Many studies have proved that Cu₂O-modified TiO₂ exhibits excellent performance in the photocatalytic production of hydrogen and the photocatalytic degradation of pollutants ^{[28, 29, 30]}.

To the best of our knowledge, no studies have probed Cu₂O-modidied TiO₂ in the photocatalytic reduction of CO₂. Much recent progress has been made on controlling the photocatalysts morphology, and forming an appropriate nanostructure can enhance its photocatalytic efficiency ^[31]. One- dimensional vertically-arrayed TiO₂ nanotubes arrays (TNTs) have been widely studied in many photocatalytic reactions, because of their high specific surface area, favorable electronic properties, and rapid photoelectron transport ^{[13, 32, 33, 34, 35]}. Herein, we prepared TNTs by conventional electrochemical anodization, and then modified the TNTs with Cu₂O by electrochemical deposition. The morphologies, compositions, and crystallinities of the products were characterized. Cu₂O-modified TNTs with varying amounts of Cu₂O were exploited for the photocatalytic reduction of CO₂ in the presence of water vapor under visible and simulated solar irradiation. The photocatalyst containing an optimized Cu₂O content exhibited a significantly higher hydrocarbon production efficiency than that of pure TNTs. The light absorption and photoelectrochemical behaviors of the photocatalysts were investigated to explain the high photocatalytic activity, and a mechanism for the photocatalytic process was proposed.

Ti foil and all chemicals were purchased from Sinopharm (Shanghai, China) and used as purchased. Highly-ordered TNTs were synthesized by conventional electrochemical anodization. Ti foil (3 x 7.5 x 0.1 mm, 99.6%) was polished with various abrasive papers, and then ultrasonically washed for 10 min in acetone, alcohol, and distilled water. The clean foil was immersed into ethylene glycol electrolyte containing 0.4 vol% NH₄F, 2 wt% citric acid and 10 vol% water. Anodization was performed in a two-electrode cell, with the Ti foil as the anode and carbon rod as the cathode, at an applied potential of 50 V. Anodization was conducted for 4 h in ice water. After reaction, the resulting TNT samples were immediately ultrasonically washed for 5 s, to remove any surface debris. The samples were then rinsed with deionized water and dried in air. They were then crystallized by annealing at 450 °C in air for 3 h.

Cu₂O nanoparticles were then deposited on the crystalline TNT samples by electrochemical deposition. This was performed in a conventional three-electrode cell, with the TNT sample as the working electrode, Pt foil as the counter electrode, and Ag/AgCl as the reference electrode. The deposition electrolyte was 0.4 mol/L CuSO₄, 3 mol/L lactic acid, and its pH was adjusted to 11 with 5 mol/L NaOH. The as-prepared TNT sample was immersed in the electrolyte solution in a 60 °C water bath, and −12 mA was applied to the TNT working electrode. The resulting Cu₂O/TiO₂ sample was named TCx, where x represents the deposition time (min). Thus, t four prepared samples TC5, TC15, TC30 and TC45 were prepared using Cu₂O deposition times of 5, 15, 30 and 45 min, respectively. A pure TNT sample was similarly characterized and probed as a control. The freshly-prepared composites were immersed in 10 vol% ammonium hydroxide for 10 min, to remove the Cu₂O oxide layer and any surface impurities which may influence their photoactivity. The resulting samples were dried in a vacuum oven for 12 h.

Morphologies were investigated by scanning electron microscopy (SEM, JEOL JSM-6700F). Crystal structures were characterized by X-ray diffraction (XRD) using a PANalytical diffractometer (D/max 40 kV) with Cu K_α radiation (λ = 0.154598 nm). Ultraviolet-visible (UV-Vis) diffuse reflectance spectra (DRS) were recorded using a PerkinElmer Lambda 35 Spectrophotometer, at a wavelength range of 200-800 nm. Photoluminescence (PL) spectra were measured on a Perkin-Elmer 4500 fluorescence spectrophotometer, at an excitation wavelength of 310 nm.

The photoelectrochemical (PEC) behavior of the TCx samples was investigated using a three-electrode configuration, with the TCx sample as the working electrode, Pt foil as the counter, and Ag/AgCl electrode as the reference. The electrolyte was 0.1 mol/L Na₂SO₄ with a pH of 7. A 350 W Xe lamp (Shanghai Lansheng Lighting Appliance Co. Ltd., China) with and without an UV cutoff filter (λ < 420 nm) was used to simulate AM 1.5 solar irradiation. Impedance measurements were performed using a PARSTAT-2273 Advanced Electrochemical System (Princeton Applied Research), which was controlled by a computer and equipped with an impedance analyzer.

The photocatalytic reduction of CO₂ by the samples was examined in the presence of water vapor without co-catalysts or sacrificial agents. The prepared sample (3 x 5 cm²) was loaded into a stainless steel chamber (170 ml) with two valves for evacuation and gas supply, a side port with a rubber plug for gas sampling, and an O-ring-sealed quartz window at the top for irradiation. A 350 W Xe lamp with and without an UV cutoff filter (λ < 420 nm) was used as the light source. After the sample was loaded into the chamber with approximately 0.5 ml of deionized water, the chamber was evacuated and filled with CO₂ (99.9999%), which had been passed through a bubble containing deionized water, and was finally sealed. The inlet and outlet valves of the chamber were open during CO₂ gas flow for at least 30 min, and were then closed to retain a chamber pressure of approximately 103 kPa. The reaction products were collected with a gas-tight syringe inserted through a septum, and then immediately analyzed by gas chromatography with FID detector (GC-2014, Shimadzu, Japan).

The morphology of the pure TNTs sample was first investigated, to ensure well-aligned TiO₂ nanotubes prior to Cu₂O deposition. Fig. 1(a) and (b) shows that the TiO₂ layer formed on the Ti foil consisted of an array of vertically-aligned nanotubes. The average nanotube diameter and length was 100 nm and ~5 μm respectively. Brief ultrasonication after electrochemical anodization was necessary to form the openings of individual tubes ^{[32, 33, 34, 35]}.

Fig. 1. SEM images of the prepared samples: top (a) and side views of TNTs (b), and top view of TC5 (c), TC15 (d), TC30 (e), and TC45 (f).

Cu₂O nanoparticles were then deposited onto the TNTs. Fig. 1(c)‒(f) shows SEM images of TNTs after modification with Cu₂O for different times. Fig. 1(c) shows a few Cu₂O nanoparticles of size ~150 nm on the TNTs after electrochemical deposition for 5 min, and that these nanoparticles had an octahedral-like structure. Increasing deposition time resulted in the octahedral nanoparticle size increasing to ~270 nm (Fig. 1(d)). Increasing the deposition time to 30 min significantly increased the size and quantity of the Cu₂O nanoparticles, and the average particle size was 480 nm (Fig. 1(e)). Increasing the deposition time to 45 min resulted in the entire TNT surface being largely covered with Cu₂O nanoparticles of ~540 nm in size. These Cu₂O nanoparticles were aggregated, had partly lost their octahedral structure, and partly obstructed tube openings. The SEM images show that the Cu₂O nanoparticles were generally well-distributed across the surface, and that some were embedded within tubes.

Figure 2(a) shows the XRD patterns of the TNTs before and after modification with Cu₂O. Peaks in the pattern of pure TNTs were attributed to elemental Ti from the underlying substrate, and well-crystallized anatase TiO₂ (JCPDS 21-1272). After electrochemical deposition of Cu₂O, new peaks characteristic of Cu₂O were observed (JCPDS 05-0667). No impurities including Cu or CuO were observed. Cu₂O contained five distinct peaks, which were attributed to the (110), (111), (200), (220), and (311) planes, as shown in Fig. 2(a). In the XRD patterns of TC5, TC15 and TC30, the (111) diffraction peak of Cu₂O was much sharper than the other peaks, indicating that (111) was the dominant exposed face. The intensity of the (200) peak increased more quickly than the other peaks with increasing deposition time. At 45 min deposition time (TC45), the intensity of the (200) plane was stronger than that of the (111) plane. From the SEM images of Cu₂O, we speculated that the exposed face of Cu₂O tended towards (200) with increasing deposition time.

Fig. 2. XRD patterns of TNTs and Cu2O-modified TNTs (a). XPS survey spectrum (b) and the Ti 2p (c) and Cu 2p (d) spectral regions of TC15.

Sample TC15 was characterized by XPS to identify its chemical composition and elemental states. Fig. 2(b)‒(d) shows XPS peaks corresponding to C, O, Ti and Cu. The binding energies of the Ti 2p state were located at 464.6 and 459.1 eV, indicative of Ti⁴⁺. The XPS peaks for the Cu 2p_3/2 and 2p_1/2 states were located at 952.9 and 932.7 eV, respectively. The characteristic Cu²⁺ satellite peak at ~942 eV was not observed, further suggesting that Cu existed as Cu₂O and not CuO ^{[25, 36]}.

The pure and Cu₂O-modified TNTs were then applied in the photocatalytic reduction of CO₂ in the presence of water vapor. We first tested their photocatalytic activities under visible light irradiation to investigate the role of the Cu₂O nanoparticles. Fig. 3(a) shows that the pure TNTs exhibited no photocatalytic activity in the reduction of CO₂ under visible light irradiation, because of the wide bandgap of TiO₂. After modifying with octahedral Cu₂O, a clear evolution of hydrocarbon product was observed. One role of Cu₂O in the catalyst was to enhance visible light harvesting, to produce more photo-generated electrons and holes for further reaction. The hydrocarbon product was predominantly methane, indicating selective production by the catalyst during photoreduction. The methane yield increased with increasing Cu₂O content of the catalyst, but began to decrease when the Cu₂O deposition time reached 45 min. Thus, the highest methane yield under visible light irradiation was observed for the catalyst prepared at a Cu₂O deposition time of 30 min.

Fig. 3. Photocatalytic reduction of CO2 in the presence of water vapor over the prepared samples, under visible (a) and simulated solar irradiation (b).

our ultimate aim was to use sunlight to convert co₂ into hydrocarbon fuel, so we carried out the photocatalytic reduction of co₂ under simulated solar irradiation (fig. 3(b)). at the optimized amount of cu₂o, the methane evolution yield of tcx was significantly higher than that of the pure tnts. tc15 exhibited the highest photocatalytic activity, at approximately three times higher than that of the pure tnts. the methane production of tc45 was lower than that of pure tnts.

The samples were further characterized to understand their photoreduction behavior. Fig. 4(a) shows the UV-Vis DRS of the samples. The bandgap of Cu₂O is 2.0‒2.2 eV, so modifying TNTs with Cu₂O extended its absorption into the visible region. This is why the Cu₂O-modified TNTs exhibited methane evolution under visible light irradiation. We then examined the photocurrent response of the samples, to study the separation and transportation of photogenerated electrons and holes. Higher light harvesting and faster electron-hole separation lead to higher photocurrent. Fig. 4(b) and (c) show that no significant current response was observed for any of the samples in the absence of irradiation. The rise and fall in current corresponded to the irradiation on/off cycles. Irradiation resulted in prompt photocurrent, which immediately returned to zero when irradiation stopped. Under visible light irradiation, the photocurrent density increased with increasing Cu₂O deposition time from 5 to 30 min, and then decreased with further increasing deposition time (Fig. 4(b)). Under simulated solar irradiation, the photocurrent densities of the samples exhibited a similar trend, except that TC15 exhibited the highest photocurrent density (Fig. 4(c)).

Fig. 4. UV-Vis DRS (a) and short-circuit photocurrent density verses time plotted (b); (c) for electrodes of samples in 0.1 mol/L Na2SO4 under visible (b) and simulated solar irradiation (c).

Treating the TCx samples with ammonium hydroxide also greatly influenced their photoactivity. Figs. 5(a), 5(b) and 1(d) showed no obvious change in morphology and crystallinity of the TNTs/Cu₂O composites upon treatment with ammonium hydroxide. Prior to treatment, the photocurrents of the TCx composites were much lower than those of pure TNTs. Fig. 3(b) shows that immersing the samples in ammonium hydroxide for 10 min greatly enhanced their photocurrent. Active sites on the TiO₂ surface are largely Ti³⁺ sites arising because of oxygen vacancies ^[37]. The enhanced photoresponse probably resulted from the removal of impurities at active sites. Previous studies indicated that CO₂ did not readily adsorb at the Cu₂O surface ^[26]. Treatment of TNTs/Cu₂O with ammonium hydroxide may have modified the Cu₂O surface, recovering active sites on the surface of TNTs and Cu₂O, thus improving the photocatalytic activity of the TNTs/Cu₂O.

Fig. 5. SEM image (a) and XRD pattern (b) of TC15 after treatment with ammonium hydroxide.

The UV-Vis DRS and photoelectrochemical results accounted for the observed trends in the photocatalytic activities of the samples. Under visible light irradiation, Cu₂O acted as the photocatalyst, and the vertically aligned TNTs acted as channels to facilitate rapid charge carrier transfer and separation ^{[1, 38, 39]}. Increasing deposition time led to more Cu₂O participating in the photoreduction, so more methane was evolved. TC30 exhibited the highest photocurrent density, reflecting its higher generation and faster charge carrier separation, so TC30 exhibited the highest methane evolution. The TiO₂ tube openings of TC45 were obstructed by aggregated Cu₂O particles (Fig. 1(f)), which hindered charge transfer and separation, and decreased the photocatalytic efficiency. Smaller particle sizes reportedly benefit charge transfer ^{[40, 41]}. The Cu₂O particle size increased with increasing deposition time, so the largest Cu₂O particle size for TC45 resulted in its decreased photocatalytic efficiency. The (111) face of Cu₂O reportedly exhibits higher photocatalytic activity than the other faces ^{[42, 43]}. XRD indicated that the (200) face was the predominant exposed face of TC45, which may also have contributed to the lower photocatalytic efficiency of TC45.

The photocatalytic activity differed under simulated solar irradiation, where the TNTs and Cu₂O nanoparticles both generated electron-hole pairs. The conduction and valence bands of Cu₂O are at higher energy level than their respective bands of TiO₂, so photogenerated electrons accumulated within TNTs and rapidly participated in the photocatalytic reaction. The holes accumulated in Cu₂O particles, which reduced the recombination of charge carriers. Fig. 6(a) shows a high-resolution TEM image of TC15. The lattice fringes of d = 0.351 and 0.247 nm matched the TiO₂(101) and Cu₂O(111) planes, respectively, in accordance with the XRD and XPS results. The TiO₂ and Cu₂O crystal grains were well connected, which facilitated charge transfer. The PL spectra in Fig. 6(b) show that the emission intensity of pure TNTs was much higher than that of the Cu₂O-modified TiO₂ samples. PL emission arose from the combination of photo-generated electrons and holes, with stronger emission indicating more recombination ^[44]. An optimized Cu₂O deposition amount extended the absorption range, facilitated charge transfer, and suppressed the recombination of electrons and holes. TC45 contained excess Cu₂O which blocked the UV light from reaching the TiO₂. UV-Vis spectra in Fig. 4(a) showed a dramatic decrease in the absorption of TC45 in the UV region compared with the other TNT samples. This explained why TC15 exhibited the highest photocatalytic activity under simulated solar irradiation.

Fig. 6. High-resolution TEM image of TC15 (a) and PL emission spectra of the samples (b).

Electrochemical impedance spectroscopy (EIS) measurements were recorded for pure TNTs and TC15 in the dark and under simulated solar irradiation, to confirm the enhanced charge separation in the Cu₂O-modified TiO₂ samples. A smaller semicircle diameter in the Nyquist plot at high frequency corresponds to faster electron-hole separation and charge transfer ^{[45, 46, 47]}. Fig. 7 shows that modifying the TNTs with Cu₂O promoted charge separation and transfer in the dark and under simulated solar irradiation. Thus, TC15 exhibited a photocatalytic efficiency three times higher than that of pure TNTs under simulated solar irradiation.

Fig. 7. Nyquist plots of TNTs and TC15 electrodes in aqueous 0.1 mol/L Na2SO4 in the dark and under simulated solar irradiation.

From the above results, a mechanism for the high photocatalytic reduction efficiency is proposed, as shown schematically in Fig. 8. Three major factors contributed to the high activity. First, the TNT nanotube structure provided abundant active sites for reactant adsorption, and the vertically-aligned TNTs provided a direct pathway for electron transfer, thus promoting charge carrier separation. Second, modifying with an optimum amount of Cu₂O nanoparticles enhanced light harvesting, so the catalyst used a greater portion of the solar spectrum. Third, the well-matched band edges of TiO₂ and Cu₂O reduced recombination, and reduced the oxidation potential of photo-generated holes from TiO₂, which had a similar effect to sacrificial agents.

Fig. 8. Schematic of the proposed photocatalysis mechanism for the Cu2O-modified TNTs under visible light.

Cu₂O-modified TiO₂ catalysts were prepared, and used in the photoreduction of CO₂. The photocatalysts comprised octahedral Cu₂O nanoparticles deposited on TNTs. The catalyst with an optimized Cu₂O loading exhibited a high photocatalytic efficiency and selectivity for methane. The high photocatalytic efficiency was ascribed to a combination of the vertically-aligned nanotubes, high visible light absorption of Cu₂O, and the well-aligned band edges of TiO₂ and Cu₂O. TiO₂ and Cu₂O are abundant and low cost materials. This study provides an economic and promising catalyst for the photoreduction of CO₂ under solar irradiation.