Photocatalytic reduction of carbon dioxide to methanol and formic acid by graphene-TiO₂

Abstract

Graphene-TiO₂ was obtained by reduction of graphite oxide by the hydrothermal method. Using photocatalytic activity to reduce carbon dioxide to methanol and formic acid was investigated in this study. The results show that the graphene loading affects the absorption of light in the visible light region. A larger surface area can also improve the catalytic activity. The largest yield of methanol and formic acid, under light of 365 nm, can reach 160 and 150 μmol g⁻¹, respectively, with 8.5% graphene loading. An increase in graphene loading can enhance photocatalytic performance, but too much will also decrease the reduction efficiency by shielding the light from reaching the catalytic surface. The effect of pH was also investigated. The mechanism of the reaction was also discussed in this study.

Implications: 

Graphene-TiO₂ hybrids were prepared by the hydrothermal method. Surface area and visible light adsorption increased with the graphene loading. Increased graphene loading improved the methanol and formic acid production. Too much graphene loading will decrease the reduction efficiency. The effect of pH shows that the HCO₃⁻ species prefers the formation of formic acid. The mechanism of the reaction was also discussed in this study.

Introduction

Society’s current use of fossil and nuclear fuels has many adverse consequences. One of those problems is global climate change caused by emissions of greenhouse gases from fossil fuel combustion. The surface temperature of the earth is maintained at an equilibrium level by receiving energy from the sun, balancing the outgoing infrared energy from the surface back into the space. If the earth had no atmosphere, the surface temperature of the earth would be −18 °C. This natural “greenhouse effect” maintains the surface at a temperature suitable for life, around 15 °C (Godfrey, 2004). Since the industrial revolution, however, human activities have been adding extra greenhouse gases to the atmosphere. The principle contributor to these increased emissions is carbon dioxide from the combustion of fossil fuels.

Scientists estimate that these human-induced emissions have caused a rise in the earth’s global mean surface temperature of 0.6 °C (Intergovernmental Panel on Climate Change [IPCC], 2001) during the 20th century. For the half million years before the 20th century, atmospheric concentration of CO₂ has varied between 200 and 300 ppm by volume (Godfrey, 2004). However, it reached 370 ppm by 2001, and 399.55 ppm by January 2013 (National Oceanic and Atmospheric Administration National Climatic Data Center [NOAA-NCDC], 2013).

Several methods for capturing and storing CO₂ released into the atmosphere due to human activity have been investigated over the years. This kind of technology, carbon capture and storage (CCS), involves collecting and generating a highly concentrated steam of almost pure CO₂. CCS is used in a range of technologies and techniques that can capture up to 90% of future CO₂ emissions from power stations and industrial sites (Richter et al., 2013).

Captured pure CO₂ may be turned into energy rather than merely being stored at the bottom of the earth and ocean. In 1979, the photoreduction of CO₂ to formaldehyde and methanol in purified water using the semiconductors TiO₂, ZnO, CdS, GaP, SiC, and WO₃ was reported (Inoue et al., 1979). The key to this method is to find sustainable and economical catalysts to convert CO₂ to fuels.

Graphene is a very attractive material for its unique electronic properties, high transparency, flexible structure, and large theoretical specific surface area (Štengl et al., 2011; Y. Zhang et al., 2011; Jiang et al., 2011). One of its many applications is the use of graphene in the preparation of highly photoactive composite materials based on titanium oxides. Graphene is an ideal nanostructure product for pairing with titanium dioxide, increasing its photocatalytic activity (Štengl et al., 2011; Y. Zhang et al., 2011). As graphene is a metal-like conductor (Guo and Dong, 2011), the electrons freed after the titanium dioxide activation are easily transported to the graphene nanosheets and recombination of e⁻ and h⁺ is strongly reduced, which increases the process yield (Štengl et al., 2011; Y. Zhang et al., 2011).

In this study, we will use graphene-loaded TiO₂ to catalyze carbon dioxide reduction. This type of catalysts is usually used to oxidize organic pollutants, but their reduction efficiency has scarcely been studied. The effects of graphene loading and pH of the solution are also discussed in this study. The feasibility of energy generation by graphene-TiO₂ (GN-TiO₂) is revealed in this paper.

Experimental Methods

Preparation of graphite oxide

Graphite oxide (GO) was synthesized by the modified Hummer’s method. The graphite powders were first oxidized by mixing with NaNO₃ and H₂SO₄. During this step, KMnO₄ was added slowly, and the vessel was immersed in ice bath to keep the temperature under 5 °C. Then, the reaction was allowed to go on for 30 min under 35 °C. Finally, the reaction temperature was increased to 95 °C, and the reaction was maintained for 30 min. After reaction, 25 mL H₂O₂ was added and the GO obtained was washed with 1 mol L⁻¹ HCl and deionized water.

Preparation of GN-TiO₂

Graphite oxide was mixed with commercial TiO₂ after ultrasonication for 1 hr. Then, the mixed solution was charged into a Teflon-lined autoclave. The autoclave was then oven-heated at 120 °C for 3 hr. The final products were obtained by filtration and washing, with subsequent drying at 100 °C overnight.

Characterization of GN-TiO₂

A ultraviolet (UV) spectrometer (Hitachi, U-3900), scanned from 200 to 800 nm in wavelength, was used to collect the reflectance spectra of graphene-loaded TiO₂ samples. High-resolution transmission electron microscopy (HR-TEM; JEOL, JEM-2000 FXΠ) was used for morphological observations of the catalysts. The crystal phase of the catalysts was tested by X-ray diffraction (XRD; X’Pert PRO) using Cu K radiation (λ = 0.1541 nm). And the elemental analyzer (EA; Elementar, Vario MICRO Cube) was used to analyze the percentage of graphene loading. Infrared spectroscopy (FTIR; NiCoLET-iS10) was performed to estimate the functional groups. The surface area of samples was determined from nitrogen adsorption-desorption isotherms at liquid nitrogen temperature using Brunauer-Emmett-Teller methods (Micromeritics, BET ASAP 2020N).

Photoreduction

Photocatalytic reduction was carried out using a double-wall cylindrical quartz reactor. This reactor was water-jacketed to maintain the solution temperature at 25 °C. Then, 2 mg L⁻¹ of the photocatalyst was loaded into a 1 mol/L NaOH solution and stirred vigorously to form a suspension. After purging CO₂ for 30 min, the solution was irradiated by a 150-W visible lamp. The methanol yield was determined by gas chromatograph (HP, G1800A) with fused silica capillary column (60 m × 0.25 mm × 0.25 μm). The detector was a mass detector and the carrier gas was He.

Results and Discussion

Characteristics of the materials

FTIR spectra are shown in Figure 1. The peak at 1617.5 cm⁻¹ in GO is attributed to the C=C bond. The adsorption bands at 1726.6, 1250, and 1045 cm⁻¹ are attributed to the COOH, O–C–O, and O—C groups, respectively (Y. Zhang et al., 2011; Guo and Dong, 2011). The adsorption bands that appear between 3000 and 3500 cm⁻¹ indicate the surface absorbed water and OH groups in water (Štengl et al., 2011; Y. Zhang et al., 2011; Guo and Dong, 2011). The peak around 1600 cm⁻¹ is contributed by the skeletal vibration of grapheme (Y. Zhang et al., 2011; Fan et al., 2011; H. Zhang et al., 2010). The broad band under 1000 cm⁻¹, in 2.21%, 8.50%, and 16.39% GN-TiO₂, is attributed to the vibration of Ti–O–Ti bond in TiO₂ (Y. Zhang et al., 2011; Guo et al., 2011; H. Zhang et al., 2010).

Figure 1. Infrared spectra of various catalysts.

The crystal phases of the P25 and P25/graphene were measured by X-ray diffraction. The results in Figure 2 show that all the materials exhibit similar XRD patterns, and obvious graphene diffraction peaks were observed in different graphene loading.

Figure 2. XRD patterns of various catalysts.

The obvious peak at 27.4° can be corresponding to the rutile phase of TiO₂, which can’t be observed in commercial TiO₂. The XRD results demonstrate that anatase TiO₂ was exclusively obtained regardless the percentage of graphene loading. The peak around 36°, attributed to rutile TiO₂, shows that the crystal phase of TiO₂ was changed by reaction with graphene during hydrothermal process (Deng et al., 2010). The peak around 43° that appear in graphene-loaded TiO₂ is the crystal plane (111) of rutile, which also can’t be observed in commercial TiO₂ (Xiao et al., 2008). The major crystal phase of TiO₂ is rutile, the useful crystal phase both for adsorption and reaction. The research of Fujishima et al. showed that CO₂ adsorption mainly existed on the TiO₂ (110) single crystals, which was attributed to the rutile phase (Fujishima et al., 2000). Indrakanti et al. also confirmed the result of Fujishima et al. and proved that the anatase phase of TiO₂ could not lead to charge transfer and form CO₂^⊸ (Indrakanti et al., 2009). The small change of the crystal may have been resulted from the formation of the C–O–Ti bonds.

The transmission electron microscopy (TEM) images of GN-TiO₂ are shown in Figure 3. The images clearly illustrate the flake-like shape of graphene with wrinkles (Y. Zhang et al., 2011; Chen et al., 2010; Wang et al., 2008; H. Zhang et al., 2010). On the other hands, the TiO₂ nanoparticles can also be observed. The figure shows that the 16.39% GN-TiO₂ sheets are almost decorated by the TiO₂ nanoparticals, compared with the 8.50% GN-TiO₂ loading.

Figure 3. TEM images of GN-TiO₂.

Figure 4 shows the ultraviolet-visible (UV-Vis) absorption spectra of pure TiO₂ and GN-TiO₂ composite. It is obvious that GN-TiO₂ not only has a red shift, but also has a stronger absorption in the visible light region. This enhanced light absorption can be explained by the formation of chemical bonding between TiO₂ and graphene (Štengl et al., 2011; C. Zhao et al., 2012). In the heat treatment reaction, the functional groups on the surface of GO (–COOH, –OH, and so on) disappeared. The π electron of the carbon atom can’t entirely bond with others to form the π bonds, so some unpaired π electrons bond with the free electrons on the TiO₂ surface to form C–O–Ti bonds (Zhang and Pan, 2011). The C–O–Ti bonds favor charge transfer upon light excitation (Guo et al., 2011; H. Zhang et al., 2010).

Figure 4. UV-Vis spectra of the TiO₂ catalysts.

The specific surface area and total pore volume of the prepared GN-TiO₂ are listed in Table 1. The pore-size distribution plot and nitrogen adsorption-desorption isotherms are shown in Figure 5. According to the results, the surface areas increased with an increase in graphene content. The average pore size lies between 8 and 25 nm. With decreasing graphene content, the pore size distribution of the GN-TiO₂ narrows.

Table 1. Surface areas, pore volumes, and pore widths of various GN-TiO₂ samples

Sample	Specific surface area (m^2 g^−1)	Total pore volume (cm^3 g^−1)	Average pore width (nm)
2.21% TiO2-GN	64.92	0.38	23.74
8.50% TiO2-GN	101.05	0.29	11.53
16.39% TiO2-GN	115.84	0.24	8.29

Figure 5. N₂ adsorption-desorption isotherms of the catalysts.

The effect of the graphene loading

Figure 6 shows the dependence of methanol and formic acid yields on illumination time. Various yields of formic acid and methanol were obtained depending on the graphene loading. The formation efficiencies of formic acid and methanol were found to be higher for GN-TiO₂ than TiO₂. The activity of catalysts followed the order of 8.50% GN-TiO₂ > 16.39% GN-TiO₂ > 2.21% GN-TiO₂ > TiO₂. The largest yields of formic acid and methanol, under a light of 365 nm, were 167.79 and 155.78 μmol g⁻¹, with 8.5% graphene loading. The results clearly indicate that the photo reduction performance of TiO₂ greatly improves with the addition of graphene. The higher efficiency can be attributed to two factors. First, the conducting band of TiO₂ is around 4.21 eV (Xu and Martin, 2000). Thus, the work function of graphene of ≈4.42 eV (Czerw et al., 2002) allows the charge transfer from TiO₂ to graphene (Štengl et al., 2011). The excited-state electrons transporting from the TiO₂ upon UV illumination into the graphene can be shuttled freely along the conducting network of the graphene, and subsequently transferred to the surface to react with CO₂ (Tu et al., 2013). The longer lifetime and mean free path for electrons on graphene implies that energetic electrons will cover a larger area of the graphene surface, thereby increasing the likelihood of interaction with the adsorbed reactants. Another factor is the larger surface area of GN-TiO₂. It offers more active adsorption sites and photocatalytic reaction centers (Tu et al., 2013). However, a further increase in the graphene content leads to a deterioration of photocatalytic performance, which is likely due to the increased scattering and absorbance of photons through excess graphene in the photocatalytic system, shielding the light from reaching the surface of the TiO₂ photocatalysts (Tu et al., 2013). As a result, an appropriate graphene loading is important in optimizing the photocatalytic activity of the GN-TiO₂ nanocomposites.

Figure 6. Photocatalytic reduction of carbon dioxide by the various catalysts.

The effect of pH

The effect of pH was also explored in this study. The results are shown in Figure 7. The results show that the formation of methanol and formic acid are different at different pH levels because of the complications between carbonate species (Whipple et al., 2010). In view of the mass balance, the percentage of different species can be calculated by eqs 1 and 2:

Figure 7. Photocatalytic reduction of carbon dioxide under various pH.

1

2

We may conclude that H₂CO₃ is the dominant carbonate species at a pH of 6.3. The predominant species changes to HCO₃⁻ when the pH is between 6.3 and 10.3. CO₃²⁻ will be the main species if the pH exceeds 10.3 (Chen et al., 2007). The percentage of different carbonate species under different pH can be calculated by eqs 3 and 4:

3

4

Thus, the concentration of H₂CO₃ and HCO₃⁻ may be expressed by CO₃²⁻ (as shown in eqs 5 and 6).

5

6

The percentage of carbonate species can be expressed by [H⁺] (related to pH), as shown in eqs 7, 6, and 7:

7

8

9

The results show that the species of HCO₃⁻ accounts for 97.6% of the carbonate species at a pH of 8, and formic acid is the main product of the reaction (about 90% of all products). This means that the HCO₃⁻ species prefers the formation of formic acid. The formation of methanol is very low at a pH of 8. At all the pH levels in this study, the formation of methanol was lower than formic acid. This is because methanol cannot be formed directly and must be reduced stepwisely to HCOOH and HCHO first. The formation of methanol under alkaline condition was higher than under acid and neutral conditions. The result is similar to the findings of Liu and colleagues (Liu et al., 2009). This phenomenal was explained by that it was most likely of CO₃²⁻ to adopt the photo-generated electrons and give rise to methanol after protonation (Liu et al., 2009).

The mechanism of the reaction

The mechanism of this reduction can be summarized as shown in Figure 8. When a photon with an energy that matches or exceeds the bond gap energy of the semiconductor is incident on a semiconductor surface, a conductive band electron (e⁻) will jump to its valence band, leaving a positively charged hole behind (h⁺) (Chen et al., 2007; Lee et al., 2003). Because of a notable synergetic effect between TiO₂ nanosheets and graphene, the composite cocatalyst has several advantages, including suppression of charge recombination, improvement of interfacial charge transfer, and increase in the number of active adsorption sites and photocatalytic reaction centers (Xiang et al., 2012a, 2012b). The photo-generated electron can be transferred to the surface of the graphene to inhibit recombination (Morel and Hering, 1993). The separated electron shows a strong reducing ability in the presence of H₂O. Depending on the different electrons transferred, the CO₂ can be reduced to generate HCHO, CO, HCOOH, CH₃OH, CH₄, and other hydrocarbons, as shown in eqs 10–21) (C. Zhao et al., 2012). The summaries of other authors show that the formation of CO₂^•− was the start of the reaction, as shown in eq 11 (Kohno et al., 1998; Liu et al., 1998; Subrahmanyam et al., 1999). The CO₂^•− can react with H⁺ to generate •CH₃ and •CH₃O, as shown in eqs 13– 15. These species can react with H⁺ and H₂O to generate organic products.

Figure 8. The mechanism of carbon dioxide reduction.

10

11

12

13

14

15

16

17

18

19

20

21

Conclusion

Using the hydrothermal method, we succeed in making TiO₂ and graphene hybrids. This synthesis process is simple and efficient. The photocatalytic activity of GN-TiO₂ is superior to that of commercial TiO₂ in the reaction of carbon dioxide reduction. The results of this study show that an increase in graphene loading can increase the surface area of the catalyst, but more is not necessarily better. Although a larger surface area affords more active adsorption sites and photocatalytic reaction centers, too much will prevent light from reaching the surface of the TiO₂ photocatalysts. In this study, an 8.50% graphene loading displayed the best performance. The effect of pH was also discussed in the study. The results show that the HCO₃⁻ species prefer the formation of formic acid. The mechanism was also discussed in this study.

Additional information

Notes on contributors

Qian Zhang

Qian Zhang is a Ph.D. student and Cheng-Fang Lin is a professor in the Graduate Institute of Environmental Engineering, National Taiwan University, Taiwan, China.

Cheng-Fang Lin

Qian Zhang is a Ph.D. student and Cheng-Fang Lin is a professor in the Graduate Institute of Environmental Engineering, National Taiwan University, Taiwan, China.

You Hai Jing

You Hai Jing is a professor in the College of the Environment and Ecology, Xiamen University, Xiamen City, Province of Fujian, China.

Chang-Tang Chang

Chang-Tang Chang is a professor in the Department of Environmental Engineering, National I-Lan University, I-Lan City, Taiwan, China.