A sol–gel derived, copper-doped, titanium dioxide–reduced graphene oxide nanocomposite electrode for the photoelectrocatalytic reduction of CO₂ to methanol and formic acid

Abstract

A classic Cu-RGO–TiO₂ photoelectrocatalyst was fabricated by a facile sol–gel method, deposited on ITO film via electrophoretic deposition and characterized by XRD, FESEM, UV-Vis and FT-IR spectroscopy. A uniformly distributed porous composite film was observed on the ITO substrate with an average particle size of 18 nm. A lower photoluminescence response of the Cu-RGO–TiO₂ sample indicates better electron/hole separation upon irradiation. A maximum 1.31 mA cm⁻² photocurrent density was observed at −0.61 V bias potential under solar simulator irradiation during CO₂ photoelectrocatalysis. Formic acid and methanol were the main products, but longer reaction times led to increased methanol formation. The estimated current efficiency of the production of formic acid and methanol was 32.47%, and the estimated rates of formation of formic acid and methanol were 255 μmol h⁻¹ cm⁻² and 189.06 μmol h⁻¹ cm⁻², respectively.

---

1. Introduction

The recycling of greenhouse gases (i.e., CO₂ and CH₄) offers long term solutions for energy and environmental related problems.¹ The presence of excess carbon dioxide (CO₂) in the earth's atmosphere has a great negative impact which is the warming of the planet. Various methods have been developed to efficiently convert and recycle CO₂. Renewable energy can be utilized in the CO₂ recycling process because it is a valuable raw material for the production of combustible chemicals such as formic acid, methanol and methane.^2,3 CO₂ photoconversion is challenging for several reasons: inefficient absorption of visible light by present day photocatalysts, low quantum efficiency due to rapid e⁻/h⁺ recombination upon irradiation, facile backward redox reactions that lead to product dissociation and lack of product selectivity. These disadvantages are the main obstacles for the commercial success of CO₂ recycling.⁴

However, current markets for CO₂-based products are developing, rather than considering it a waste chemical. Photoelectrocatalytic reduction of CO₂ is able meet this high energy conversion demand by utilizing free and abundant solar energy.⁵ Successful conversion of CO₂ depends on a highly active photocatalyst or photoelectrocatalysis that can facilitate high rates of conversion while drastically decreasing the necessary energy input.⁶

The earliest report of CO₂ photoconversion was published by Halmann in 1978.⁷ Honda and Fujishima first reported the use of titanium dioxide in the photoelectrolysis of water, but did not report the photoreduction of CO₂ with titanium until 1979.⁸ Since that time, TiO₂ has remained the most preferred photocatalyst and photoelectrocatalyst for CO₂ reduction. TiO₂ nanoparticles doped with N, Pt, Ag and Cu, among others, exhibited significant CO₂ photoconversion.^9–11 The wide band gap (i.e., 3.2 eV for the anatase phase) of TiO₂ is favorable due to its strong redox ability and high resistance to photocorrosion. Metal doping of TiO₂ could decrease the recombination process because the metal centers function as electron traps. Moreover, due to the work function difference, a Schottky barrier appears in the bimetallic TiO₂ structure. This phenomenon also helps to decrease the recombination of charge carriers.⁹ TiO₂ structures incorporated with crystalline carbon structures, such as graphene, reduced graphene oxide (RGO) and carbon nanotubes, exhibit high catalytic efficiencies.^11–13 Graphene–TiO₂ composite materials have been described as future generation photocatalysts due to the low toxicity and extended photoactivity.^14,15 Carbon structures can also promote simple reaction mechanisms during catalysis, which result in higher selectivity and desired yields.¹⁶ Thus, further modification of the TiO₂ nanostructures can enhance the overall photoconversion of CO₂.

Reduced graphene oxide (RGO) as a catalyst promoter with TiO₂ was reported in our previous work.¹⁷ Composite photocatalysts consisting of several elements are promising materials for visible light activity which enhance the kinetic properties of the desired reactions.¹⁸ Mohamed et al. reported RGO–TiO₂ composites for the photocatalytic reduction of CO₂ to methane gas.¹⁹ They reported that the incorporation of the RGO into the composite increased the photocatalytic performance of TiO₂ compared to pure TiO₂ by decreasing the TiO₂ band-gap into the visible region and prolonging the recombination time of the charge carriers.¹⁹ On the other hand, Cu is a representative d-block element with attractive catalytic properties for CO₂ electroreduction process. Copper electrodes have shown great success in hydrocarbon gas production such as hydrogen (H₂), methane (CH₄) etc. Moreover, the Cu atoms can act as a channel for the dispersion of photoelectrons, which suppress the e⁻/h⁺ recombination in the photoelectrocatalysis process. Hence, a Cu-RGO–TiO₂ nanocomposite photoelectrode was conceived and designed to convert CO₂ in aqueous methyl diethanolamine (MDEA) solution as the CO₂ absorption electrolyte.

2. Experimental methods

2.1 Materials

Graphite flakes with 7–10 μm nominal size were procured from Alfa Aesar. Tetra-n-butyl orthotitanate [Ti(OBu)₄] was procured from Sigma-Aldrich, and all other chemicals used in this work were of analytical grade. The indium tin oxide (ITO)-coated conducting glass plates (0.7 mm thickness) were procured from Osaka, Japan. Deionized water (18.2 MΩ cm) used in these experiments was from a Milli-Q system (Millipore, Bedford, MA; Mega-Pure System, Model MP-290). The carbon dioxide gas had 99.9% purity.

2.2 Preparation of reduced graphene oxide (RGO)

Graphene oxide (GO) was prepared by the modified Hummer's method.²⁰ Initially, 1 g of graphite flakes were mixed with sulfuric acid and phosphoric acid (120 mL : 13 mL). Then, 6 g of KMnO₄ was added gradually, and the mixture was stirred continuously for 3 days. This was followed by the addition of 7 mL of H₂O₂ solution (30%) and 135 mL of ice water until the mixture turned brown in color. The mixture was then centrifuged at 4000 rpm for 10 minutes. After centrifugation, the GO suspension was washed with 1 M HCl with sonication. The GO suspension was then dried in an oven and later subjected to hydrazine treatment to obtain the reduced graphene oxide (RGO).²¹ The GO powder (300 mg) was then added into 600 mL of ultrapure Milli-Q water and sonicated for 15 minutes. This was followed by the addition of 3 mL of hydrazine hydrate and the solution was stirred vigorously at 90 °C for 6 hours, precipitating the RGO as a black solid.²¹ Finally the solution was filtered and washed with ethanol and water several times to obtain the RGO powder. The powder was dried in an oven at 80 °C.

2.3 Preparation of the Cu-RGO–TiO₂ nanocomposite

To prepare the TiO₂ nanocomposite, 20 mL of Ti(OBu)₄ (TBOT) was diluted in absolute ethanol and stirred continuously. Then, 6 mL of ultrapure water, 6 mL of ethanol, 1 mol% Cu(NO₃)₂·5H₂O and 1 wt% RGO powder were mixed and sonicated for 1.5 hours. This mixture was added to the TBOT solution drop-wise for 30 minutes. Acetic acid was added to lower the pH of the solution to less than 3. The total mixture was vigorously stirred for 30 minutes to obtain a homogeneous solution. The dark colored solution was then stored at room temperature for 48 hours to complete the gel formation and was dried at 80 °C overnight. The dry porous gel was milled using a steel mortar and pestle. The fine powder catalyst was then calcined at 550 °C for 4 hours.

2.4 Fabrication of Cu-RGO–TiO₂/ITO photoanodes

The Cu-RGO–TiO₂ was deposited on the ITO substrate by electrophoretic deposition in a two-electrode cell. The prepared composite powder was dispersed in DI water with a concentration of 0.01 mg mL⁻¹. The pH of the solution was maintained by determining the zeta potential value. The composite powder mixture was sonicated in DI water to ensure a homogeneous dispersion before the electrophoretic deposition process.

2.5 Photoelectrochemical measurements

All photoelectrochemical measurements were performed in a cubic quartz photoelectrochemical cell using a potentiostat/galvanostat (Autolab PGSTAT30, Ecochemie, Netherlands). A standard three-electrode system was used, where the photoanode with an area of 1 cm² was the working electrode (WE). A platinum (Pt) wire and a saturated calomel electrode (SCE) were the counter (CE) and reference (RE) electrodes, respectively. The electrolyte in these experiments was 0.1 M Na₂SO₄. The potential in the voltammetric measurements was swept from 0.0 V to −1.5 V at 50 mV s⁻¹. Electrochemical impedance spectroscopy (EIS) was measured at the open circuit potential (OCP) both in dark and under solar irradiation. The EIS experimental data were simulated using analog circuits with NOVA 1.10 software on a computer interfaced with the potentiostat. The frequency range was from 10⁻¹ Hz to 10⁵ Hz with an acquisition of 10 points per decade and an amplitude of 5 mV around the OCP.

2.6 Constant potential photoelectrolysis

The photoelectrocatalytic conversion of CO₂ was performed in the quartz cell containing aqueous 10% methyl diethanolamine (MDEA). Fig. 1 shows a schematic diagram of the photoelectrocatalytic reduction process with a three-electrode system, where the CE and WE had the same surface area.

Fig. 1 Quartz cell photoelectrocatalytic (PEC) reactor.

The irradiation intensity was ∼10 mW cm⁻². Prior to the reaction, CO₂ was bubbled for 1 hour to saturate the solution, and the final pH of the CO₂-saturated solution was 7.6. The liquid phase product was analyzed by HPLC, and only the peaks for formic acid and methanol were considered.

2.7 Instrumentation

A scanning electron microscope (SEM, Quanta FEI 200) was used to study the morphology of the as-prepared composite samples. X-ray diffraction (XRD) was performed on a powder X-ray diffractometer (Bruker D8 Advance equipped with EVA diffract software, Germany) over a range of 10° ≤ 2θ ≤ 80°, at 40 kV and 30 mA with Cu Kα radiation (k = 1.5418 Å). UV-Vis diffuse reflectance spectra were obtained on a Lambda 35 series equipped spectrophotometer (Perkin Elmer) with a thin film slot. Fourier transformed infrared spectroscopy (FTIR) was recorded by a Bruker IFS 66 V/S. The light source, a 150 W (Xe arc lamp) solar simulator, was from OSRAM photo optic (Germany). It had a consistent spectrum profile of AM 1.5G (i.e., similar characteristic features to a 1.5 global radiation solar light under standard air conditions). X-ray photoelectron spectroscopy (XPS, Thermo Scientific K-alpha instrument) with an unmonochromatized Mg Kα radiation (photon energy 1253.6 eV) source, vacuum less than 10⁻⁹ Torr and spectral resolution of 0.1 eV, were carried out on flat gold (Si/10 nm Ti/200 nm Au) as the substrate and reference. The XPS core levels were aligned to the Au 4f 7/2 binding energy (BE) of 84 eV.

3. Results and discussions

3.1 Fabrication and characterization of the Cu-RGO–TiO₂/ITO composite electrode

The catalyst powder was characterized by crystallographic, imaging and optical techniques. Fig. 2 shows the XRD pattern of the prepared catalyst. From the JCPDS ref. code 01-086-1157, the characteristic peaks at 25.3 (101), 37.9 (103), 48.1 (200), 53.9 (105) and 75.1 (215) are attributed to the titanium tetragonal system of the anatase phase.

Fig. 2 XRD of TiO₂, RGO–TiO₂ and Cu-RGO–TiO₂.

The calcination temperature of 550 °C played an important role in the production of the crystalline nanocomposites in the TiO₂ anatase phase. The phase change from anatase to rutile was absent during the calcinations process. The Cu diffraction peaks at 2θ = 43° and 45° and Cu₂O peaks at 36° were not observed due to the low concentration and incorporation in the TiO₂ lattice structure. Fig. 2 shows the overlap of the RGO characteristic peak at 25.8° with the TiO₂ (101) plane. Cu could be incorporated into the TiO₂ lattice due to the similarity of the Cu and Ti atomic radii (132 pm and 160 pm, respectively). Furthermore, the peaks attributed to Cu or Cu₂O were not observed. The average crystallite size is given by the Scherrer's formula (eqn (1)):

where, D = crystallite size, λ = wavelength of the X-ray, β = full width at half maximum of the peak (in radians) and θ = angle of diffraction (in degrees). The average crystallite size of the Cu-RGO–TiO₂ composites (with respect to the anatase (101) peak) is between 15–28 nm. Thus, the doping of RGO did not introduce any crystalline phase changes, and we can conclude that RGO had no effect on crystallite size.

Fig. 3 shows the XPS spectra of the Cu-doped TiO₂ nanoparticles. The electron-binding energy (BE) of the Ti 2p photoelectron peak (Fig. 3a) illustrates the existence of Ti⁴⁺ in the TiO₂ nanostructures at 458.2 eV. The oxygen peak at 529.5 eV (Fig. 3b) confirms the formation of the metal–oxygen bond, while the Cu 2p spectrum indicates the co-formation of Cu²⁺ and Cu⁺ in the structure (Fig. 3c). The binding energy of 932.2 eV is a characteristic of Cu₂O. The shake-up satellite peaks at 940.7 and 943.3 eV (green dashed lines) are an indication of Cu(II) formation.^22,23 The existence of the shoulder (934.2 eV) in the copper spectrum also indicates the presence of Cu⁺ in the structure. The deconvoluted peaks in the copper spectrum indicate that 56% of the copper exists as Cu(I).

Fig. 3 XPS spectra of Cu-doped TiO₂ nanoparticles: (a) Ti 2p, (b) O 1s and (c) Cu 2p.

Fig. 4 shows the FESEM images of the thin film electrodes. The TiO₂ nanoparticles are quite homogeneous with sizes between 30–45 nm. The RGO–TiO₂ composites are also homogeneous, but more porous in the thin film form. The particle size varies within the same range as the TiO₂ nanoparticles, but the RGO–TiO₂ nanocomposites could be clearly distinguished from the single TiO₂ nanoparticles.

Fig. 4 FESEM images of (a) TiO₂, (b) RGO–TiO₂ and (c) Cu-RGO–TiO₂.

A different morphology was observed for the Cu-RGO–TiO₂ thin film. The film bed is dense and packed, and the particle size is larger than the TiO₂ and RGO–TiO₂ nanoparticles. This morphological change could be due to the formation of grain boundaries at high calcination temperatures. Moreover, RGO has a tendency to aggregate into larger particles because of the weak van der Waals interactions.²⁴ The RGO layers could not be clearly seen in the FESEM micrographs, which could be due to the charging effect of the electron-rich RGO.^17,25

Fig. 5 shows the HRTEM image of Cu-RGO–TiO₂. The estimated particle sizes (Fig. 5) are in good agreement with the XRD and SEM results. The relative d-spacing values resulted in successful Cu doping into the TiO₂ lattice. The Cu-RGO–TiO₂ film showed a slight red-shift in the absorbance edge and a significant enhancement of light absorption between 400–800 nm. Fig. 6 shows the corresponding UV-Vis diffuse reflectance spectra of the fabricated thin films. The doping of Cu into the TiO₂ lattice creates intermediate sites for the transfer of photogenerated electrons, hence facilitates the electron–hole pair separation.²⁶

Fig. 5 TEM image of Cu-RGO–TiO₂.

Fig. 6 UV-Vis diffuse reflectance spectra of the TiO₂, RGO–TiO₂ and Cu-RGO–TiO₂ photocatalysts (a) and the plot of αhν^1/2 vs. hν (b).

The incorporation of RGO, on the other hand, facilitates electron transfer and also increases the visible absorption of TiO₂. Thus, the doping of TiO₂ with Cu and incorporation of RGO are effective for enhanced photocatalytic activity. The optical band-gap was determined from the Kubelka–Munk equation as follows:

αhν = A(hν − E_g)ⁿ (2)

where, α = absorption coefficient, A = constant (∼1), v = frequency of light (3 × 10⁸ ms⁻¹), h = Plank's constant (4.136 × 10⁻¹⁵ eV), E_g = band gap energy, n = 2 for indirect allowed transitions.

From eqn (2) and the plot in 6(b), the E_g values of TiO₂, RGO–TiO₂ and Cu-RGO–TiO₂ are 3.23, 3.10 and 2.98, respectively. The lower band-gap of Cu-RGO–TiO₂ can be attributed to the formation of Ti–Cu–O, which lowers the conduction band level of TiO₂. The rapid recombination of the photogenerated electrons and holes emits photoluminescence (PL). A lower PL emission intensity indicates a decrease in the radiative recombination and better separation of the excitons.²⁷ Fig. 7 compares the PL spectra of the prepared samples. Broad peaks were observed in the TiO₂ and RGO–TiO₂ samples between 500–700 nm. This indicates direct emission after excitation and poor separation of the photogenerated electrons and holes. The lower emission intensity of the Cu-RGO–TiO₂ sample could be due to the lower radiative combination of excitons, which indicates a slower recombination process.

Fig. 7 Photoluminescence spectra of TiO₂, RGO–TiO₂ and Cu-RGO–TiO₂.

The inclusion of Cu nanoparticles generates considerable fluorescence quenching indicative of efficient electron transfer from TiO₂ to Cu, due to the favorable electrical conduction between them. Under light irradiation, the excited electrons will follow the Cu-assisted path towards the RGO structure, which eventually decreases the direct recombination of the charge carriers.

The FTIR spectra of the prepared catalyst samples are shown in Fig. 8. The prominent absorption bands are observed at 666, 1551 and 1626 cm⁻¹. In the RGO spectrum, most of the C=O bonds were eliminated, and the peak at 1720 cm⁻¹ was absent, which confirms a well-reduced RGO structure. The peak at 1551 cm⁻¹ is attributed to the graphene skeletal vibration. The Ti–O–Ti stretching vibrational peak is observed below 1000 cm⁻¹, and the broader peak in the 2400–3400 cm⁻¹ range is attributed to the H-bonding between the –OH groups present on the TiO₂ surface.²⁸ A smaller peak at 2358 cm⁻¹ could be due to the atmospheric CO₂ molecules adsorbed on the TiO₂ surface.

Fig. 8 FTIR spectra of TiO₂, RGO–TiO₂ and Cu-RGO–TiO₂.

The most critical drawback in photoelectrocatalysis is the recombination process.²⁹ EIS is a useful tool to investigate the charge transfer and recombination processes at the semiconductor/electrolyte interface.³⁰ The EIS responses of TiO₂/ITO, RGO–TiO₂/ITO and Cu-RGO–TiO₂/ITO in the dark and under illumination are shown in the Nyquist plots (Fig. 9a).

Fig. 9 Nyquist plots of the TiO₂, RGO–TiO₂ and Cu-RGO–TiO₂ film electrodes at open circuit potentials both in the dark and under visible irradiation (9a) and conceptualization of the equivalent circuit model at the electrode–electrolyte interface (9b).

The impedance values of the photocatalysts were measured at the open circuit potential in dark and illuminated conditions. The Cu-RGO–TiO₂ thin film showed lower resistance than TiO₂ and RGO–TiO₂ under both conditions. Upon irradiation, the semicircle diameter decreases, and the charge transfer resistance (R_CT), which is the resistance of electron transfer process across the electrode/electrolyte interface, also decreases.³¹ The inset of Fig. 9a shows the equivalent circuit model across the electrode/electrolyte interface, where a Warburg diffusion model is proposed. Here, R_e refers to the bulk resistance between the WE and RE at the high frequency intercept of the semicircle with respect to the real axis. The interfacial resistance is represented by R_CT in the equivalent circuit model.³² Fig. 9b shows the equivalent circuit model which represents the electrode–electrolyte interface. Table 1 gives the R_CT values for all three photoelectrodes in both dark and irradiated conditions.

Table 1 The charge-transfer resistance (R_CT) of the TiO₂, RGO–TiO₂ and Cu-RGO–TiO₂ photoelectrodes from simulation of the EIS results

Samples	Irradiation condition	Rct/Ω cm^2
TiO2	Dark	891.6
	Solar simulator	832.7
RGO–TiO2	Dark	488.3
	Solar simulator	465.1
Cu-RGO–TiO2	Dark	536.9
	Solar simulator	216.8

---

It can be observed that the Cu-RGO–TiO₂ electrode shows higher resistance (536.9 Ω cm²) in the dark. This is probably due to the incorporation of Cu and corresponding defects in regular octahedral titania structure. But the charge transfer resistance decreases significantly (∼217 Ω cm²) upon irradiation. This can be attributed to the electron channeling in the Ti–Cu–C structure which decreases the recombination of charge carriers and increases the photocurrent density. In fact, this proposed structure may have the rectifying characteristics due to the formation of Schottky barriers in the metal–semiconductor junction.³³

To investigate the effect of an applied bias on the modified TiO₂ photoelectrode, a Mott–Schottky plot was obtained at room temperature. The flat-band potential can be determined from the equation:

where, C_SC = space charge capacitance, V_fb = flat band potential, N_D = charge carrier conc. V_B = applied potential, ε is the dielectric constant of the semiconductor and ε₀ is the vacuum permittivity. Thus, the impedance measurement results can be plotted according to eqn (3) and are shown in Fig. 10. The estimated flat-band potential and the corresponding donor density are −0.76V and 5.76 × 10¹⁵ cm⁻³, respectively.

Fig. 10 Mott–Schottky plot for Cu-RGO–TiO₂/ITO.

It should be noted that the onset potential for the anodic current is −0.82 V, and is not necessarily equal to the flat band potential. In fact, the flat band potential is a region where the recombination process tends to suppress the threshold of the photocurrent. Some factors i.e. crystallographic structure of TiO₂, doping, thin film properties etc. play a vital role in determining the flat band region. Hence the applied bias was chosen a little higher (−0.61 V) than the estimated flat band potential to maintain the continuous flow of anodic photocurrent.

3.2 Photoelectrocatalytic reduction of carbon dioxide into formic acid and methanol

Fig. 11 shows the voltammetric response of the catalyst films under illumination by the solar simulator. Prior to this, CO₂ gas was bubbled in the 10% MDEA solution for 1 hour to saturate the solution. No significant photocurrent was observed at the TiO₂/ITO and RGO–TiO₂/ITO electrodes. Moreover, blank experiments (without CO₂) were also carried out on all three electrodes and no photocurrent was observed in the process.

Fig. 11 Voltammetry in 10% MDEA aqueous solution: (a) TiO₂, RGO–TiO₂ and Cu-RGO–TiO₂ electrodes under illumination and (b) the Cu-RGO–TiO₂ electrode in the dark and under illumination.

Cu-RGO–TiO₂ gives a maximum photocurrent of 1.31 mA cm⁻² at −0.61 V vs. SCE. Fig. 11b shows the effect of light and dark conditions of the Cu-RGO–TiO₂ photoelectrode in the CO₂-saturated solution. The photocurrent density was unsteady, and fluctuated during the reaction (Fig. 12a). In fact, CO₂ reduction process is a multi-electron transfer process and it is difficult to maintain the selectivity of a single product. Moreover, the competition between different electroactive species present in the electrolyte such as CO₂ and H⁺ decreases the selectivity of the overall reaction. The interaction and catalytic effect of Cu atoms towards CO₂ in the electro-reduction process has been reported in earlier reports.³⁴ But pristine TiO₂ are not reactive with CO₂ molecules. This is probably due to the difficulties of the angular CO˙₂ radicals to come in contact with the regular TiO₂ structure.³⁵ However, the initial photocurrent density is 4.56 mA cm⁻² but decreases to 0.63 mA cm⁻² at the end of the reaction. The slow diffusion process of CO₂ in the electrolyte may be responsible for the sudden decrease of photocurrent density and this can be understood from the impedance measurement Nyquist plots (Fig. 9a). This is evident by the appearance of the onset of a Warburg element at lower frequencies which denotes a diffusion limiting process for the Cu-RGO–TiO₂ electrode upon irradiation (Fig. 9a). A full photocurrent profile for a six-hour reaction period is shown in Fig. 12b.

Fig. 12 Cu-RGO–TiO₂ photoelectrode: (a) dependence of the photocurrent on reaction time in the photoelectrocatalytic reduction of CO₂ and (b) the photocurrent density profile for a six-hour reaction time.

3.3 Product analysis and efficiency

The products from the photoelectrolysis experiments were collected at one hour intervals. The saturated CO₂ solution in 10% MDEA contained an estimated concentration of 6950 ppm (∼0.16 M) dissolved CO₂. After a six-hour period, the solution was analyzed, and the final concentrations of formic acid (Fig. 13a) and methanol (Fig. 13b) were 98 ppm and 242 ppm, respectively. After 2 hours of reaction, the maximum formic acid concentration was 157 ppm, which decreased further (Fig. 13).

Fig. 13 HPLC detection of (a) methanol and (b) formic acid, and (c) the yield of formic acid and methanol with time.

This could be due to the reduction of formic acid to methanol during the course of the reaction. The current efficiency was determined using eqn (4):³⁶

where (IC)₀ and (IC)_t are the total inorganic carbon (g L⁻¹) at times 0 and t, respectively, I is the current (A), F is the Faraday constant (26.8 Ah), V is the volume (L) and t is the time of treatment (h). Thus, the estimated current efficiency for the process is 32.47%. Table 2 compares the literature results for the reduction of CO₂ to methanol and formic acid using TiO₂ photoelectrocatalysis and photocatalysis.

Table 2 Comparison of photocatalytic and photoelectrocatalytic reduction of CO₂ to methanol and formic acid on TiO₂ composites in the literature

No	Electrode material/light source/solution	Catalytic method	Products	Highest yield	Ref.
1	Cu doped TiO2, 8 W Hg lamp UVC (254 nm), CO2 saturated 0.2 M NaOH	Photocatalytic	Methanol	19.7 μmol g^−1 h^−1	37 and 38
2	Cu doped TiO2, 10 W UV lamp, CO2 saturated 1 M KHCO3	Photocatalytic	Methanol	450 μmol g^−1 h^−1	39
3	Cu(I)/TiO2/365 nm, 16 W cm^−2 UV, 1.2 wt%-Cu/TiO2 catalyst at 1.29 bar of CO2 saturated pure water	Photocatalytic	Methanol		40 and 41
4	CdSe quantum dot (QD)-sensitized TiO2/visible light > 420 nm, CO2 saturated pure water. 300 W Xe lamp	Photocatalytic	Methanol	3.3 ppm g^−1 h^−1	18
5	Cu–TiO2 on molecular sieve 5A/UV light, CO2 saturated \0.2 M NaOH. 250 W Hg lamp	Photocatalytic	Methanol	0.78 μg h^−1 g^−1	42
6	Pd nanoparticles bismuth titanate, 0.1 CO2 saturated H2SO4, solar simulator 300 W, 1.5 AM	Photocatalytic	Formic acid	110–160 μMol h^−1 g^−1	43
7	Nitrogen doped TiO2. Cu counter electrode, CO2 bubbled KHCO3 electrolyte, 2V vs. SCE. 100 W Xe lamp	Photo-electrocatalytic	Methanol formic acid	Faradaic efficiency lower than 8%	10
8	(CdS) and (Bi2S3) on TiO2 nanotube. 500 W Xe lamp, wavelength less than 400 nm, CO2 saturated 0.1 M NaOH, 0.1 M NaSO3	Photocatalytic	Methanol	44.9 μmol h^−1 cm^−2 (Bi2S3) 31.9 μmol h^−1 cm^−2 (CdS)	44
9	Cu–TiO2–RGO, 0.1 M Na2SO4, CO2 saturated 10% Methyl diethanolamine (MDEA), −0.61 V vs. SCE. 150 W Xe arc lamp, 1.5 AM	Photo-electrocatalytic	Methanol formic acid	255 μMol h^−1 cm^−2 189.06 μmol h^−1 cm^−2	This work

---

The photo-oxidation of Cu-RGO–TiO₂ releases electrons and produces reactive holes, which react with water to release oxygen:⁴⁵

Initial oxidation:

4 h⁺ + 2H₂O → O₂ + 4H⁺

Initial reduction:

e⁻ + H⁺ → H˙

and

e⁻ + CO₂→ ˙CO₂

Subsequent reactions:

˙CO₂ + 2H˙ + h⁺ → HCOOH

˙CO₂ + 6H˙ + h⁺ → CH₃OH

A simple mechanism is shown in Fig. 14.

Fig. 14 CO₂ photoelectrocatalysis mechanism in aqueous media.

4. Conclusions

Crystalline Cu-RGO–TiO₂ nanoparticles were prepared by sol–gel synthesis and deposited onto an ITO glass substrate via electrophoretic deposition. XRD and XPS analysis confirmed the incorporation of Cu dopant atoms into the TiO₂ lattice. Photo-absorption was observed in the visible region, and the band gap was 2.98 eV. The slow recombination of the photo-generated electrons and holes was confirmed by PL spectra. The composite WE was stable for more than 6 hours of reaction time. Formic acid (HCOOH) and methanol (CH₃OH) were the main products from the reduction of CO₂ with H₂O (or proton solvents) in MDEA solution, with rates of 255 μmol h⁻¹ cm⁻² and 189.06 μmol h⁻¹ cm⁻², respectively. HCOOH was initially produced (maximum concentration of 157 ppm), but it later became an intermediate for the formation of methanol and higher hydrocarbons. Cu doping directs the product selectivity towards HCOOH and CH₃OH and promotes the RGO–TiO₂ catalyst to produce larger amounts of CH₃OH (maximum 242 ppm) when the reaction time is more than 3 hours.

Acknowledgements

The authors would like to thank the University of Malaya for funding this work with High Impact Research (HIR-F-000032), RP005B 13AET and FP033 2013A research grants for their cordial support.

References

1. A. Goeppert, M. Czaun, J. P. Jones, G. K. S. Prakash and G. A. Olah, Chem. Soc. Rev., 2014, 43, 7957–8194.
2. W. N. Wang, Aerosol Air Qual. Res., 2014, 14, 533–549.
3. T. Yui, A. Kan, C. Saitoh, K. Koike, T. Ibusuki and O. Ishitani, ACS Appl. Mater. Interfaces, 2011, 3, 2594–2600.
4. A. Dhakshinamoorthy, S. Navalon, A. Corma and H. Garcia, Energy Environ. Sci., 2012, 5, 9217–9233.
5. Q. Zhai, S. Xie, W. Fan, Q. Zhang, Y. Wang, W. Deng and Y. Wang, Angew. Chem., Int. Ed., 2013, 52, 5776–5779.
6. Z. Jiang, T. Xiao, V. L. Kuznetsov and P. P. Edwards, Philos. Trans. R. Soc., A, 2010, 368, 3343–3364.
7. M. Halmann, Nature, 1978, 275, 115–116.
8. T. Inoue, A. Fujishima, S. Konishi and K. Honda, Nature, 1979, 277, 637–638.
9. Y. Wang, Q. Lai, F. Zhang, X. Shen, M. Fan, Y. He and S. Ren, RSC Adv., 2014, 4, 44442–44451.
10. Y. P. Peng, Y. T. Yeh, S. I. Shah and C. P. Huang, Appl. Catal., B, 2012, 123–124, 414–423.
11. Y. T. Liang, B. K. Vijayan, K. A. Gray and M. C. Hersam, Nano Lett., 2011, 11, 2865–2870.
12. Z. Li, B. Gao, G. Z. Chen, R. Mokaya, S. Sotiropoulos and G. L. Puma, Appl. Catal., B, 2011, 110, 50–57.
13. J. H. Yun, R. J. Wong, Y. H. Ng, A. Du and R. Amal, RSC Adv., 2012, 2, 8164–8171.
14. C. Chen, W. Cai, M. Long, B. Zhou, Y. Wu, D. Wu and Y. Feng, ACS Nano, 2010, 4(11), 6425–6432.
15. K. Zhou, Y. Zhu, X. Yang, X. Jiang and C. Li, New J. Chem., 2011, 35, 353–359.
16. L. Yuan, Q. Yu, Y. Zhang and Y. J. Xu, RSC Adv., 2014, 4, 15264–15270.
17. M. R. Hasan, S. B. A. Hamid, W. J. Basirun, Z. Z. Chowdhury, A. E. Kandjani and S. K. Bhargava, New J. Chem., 2014, 39, 69–76.
18. C. Wang, R. L. Thompson, J. Baltrus, C. Matranga and J. Phys, Chem. Lett., 2010, 1, 48–53.
19. L. L. Tan, W. J. Ong, S. P. Chai and A. R. Mohamed, Nanoscale Res. Lett., 2013, 8, 465.
20. D. C. Marcano, D. V. Kosynkin, J. M. Berlin, A. Sinitskii, Z. Sun, A. Slesarev, L. B. Alemany, W. Lu and J. M. Tour, ACS Nano, 2010, 4, 4806–4814.
21. S. Stankovich, D. A. Dikin, R. D. Piner, K. A. Kohlhaas, A. Kleinhammes, Y. Jia, Y. Wu, S. T. Nguyen and R. S. Ruoff, Carbon, 2007, 45, 1558–1565.
22. M. A. Mahmoud, W. Qian and M. A. El-Sayed, Nano Lett., 2011, 11, 3285–3289.
23. N. Pauly, S. Tougaard and F. Yubero, Surf. Sci., 2014, 620, 17–22.
24. M. Shi, J. Shen, H. Ma, Z. Li, X. Lu, N. Li and M. Ye, Colloids Surf., A, 2012, 405, 30–37.
25. M. R. Hasan, S. B. A. Hamid and W. J. Basirun, Appl. Surf. Sci., 2015, 339, 22–27.
26. N. R. Khalid, E. Ahmed, Z. Hong, M. Ahmad, Y. Zhang and S. Khalid, Ceram. Int., 2013, 39, 7107–7113.
27. J. Shi, J. Chen, Z. Feng, T. Chen, Y. Lian, X. Wang and C. Li, J. Phys. Chem. C, 2007, 111, 693–699.
28. J. Shen, B. Yan, M. Shi, H. Ma, N. Li and M. Ye, J. Mater. Chem., 2011, 21, 3415–3421.
29. X. Chen and S. S. Mao, Chem. Rev., 2007, 107, 2891–2959.
30. F. F. Santiago, E. M. Barea, J. Bisquert, G. K. Mor, K. Shankar and C. A. Grimes, J. Am. Chem. Soc., 2008, 130, 11312–11316.
31. W. H. Leng, Z. Zhang, J. Q. Zhang and C. N. Cao, J. Phys. Chem. B, 2005, 109, 15008–15023.
32. X. I. He, Y. Y. Cai, H. M. Zhang and C. H. Liang, J. Mater. Chem., 2011, 21, 475–480.
33. Z. Zhang and J. T. Yates Jr., Chem. Rev., 2012, 112(10), 5520–5551.
34. Y. Hori, Modern Aspects of Electrochemistry, ed. C. Vayenas et al., Springer, New York, 2008, vol. 42, pp. 89–189.
35. V. P. Indrakanti, J. D. Kubicki and H. H. Schobert, Fuel Process. Technol., 2011, 92, 805–811.
36. T. T. Guaraldo, S. H. Pulcinelli and M. V. B. Zanoni, J. Photochem. Photobiol., A, 2011, 217, 259–266.
37. I. H. Tseng, W. C. Chang and J. C. S. Wu, Appl. Catal., B, 2002, 37, 37–48.
38. I. H. Tseng, J. C. S. Wu and H. Y. Chou, J. Catal., 2004, 221, 432–440.
39. H. W. N. Slamet, E. Purnama, S. Kosela and J. Gunlazuardi, Catal. Commun., 2005, 6, 313–319.
40. J. C. S. Wu, H. M. Lin and C. L. Lai, Appl. Catal., A, 2005, 296, 194–200.
41. J. C. S. Wu and H. M. Lin, Int. J. Photoenergy, 2005, 7, 115–119.
42. B. Srinivas, B. Shubhamangala, K. Lalitha, P. A. K. Reddy, V. D. Kumari, M. Subrahmanyam and B. R. De, Photochem. Photobiol., 2011, 87, 995–1001.
43. K. S. Raja, Y. R. Smith, N. Kondamudi, A. Manivannan, M. Misra and V. Subramanian, Electrochem. Solid-State Lett., 2011, 14, F5–F8.
44. X. Li, H. Liu, D. Luo, J. Li, Y. Huang, H. Li, Y. Fang, Y. Xu and L. Zhu, Chem. Eng. J., 2012, 180, 151–158.
45. S. Ahmed, M. A. Mansoor, W. J. Basirun, M. Sookhakian, N. M. Huang, K. M. Lo, T. Söhnel, Z. Arifin and M. Mazhar, New J. Chem., 2015, 39, 1031–1037.

---