Article pubs.acs.org/est

Deciphering Visible Light Photoreductive Conversion of CO2 to Formic Acid and Methanol Using Waste Prepared Material Qian Zhang,† Cheng-Fang Lin,† Bor-Yann Chen,‡ Tong Ouyang,*,§ and Chang-Tang Chang*,∥ †

Graduate Institute of Environmental Engineering, National Taiwan University, Taipei, 106, Taiwan Department of Chemical and Materials Engineering, National I-Lan University, Ilan, 260, Taiwan § College of the Environment and Ecology, Xiamen University, South Road of Xiangan, Xiamen City, 361102 Fujian, China ∥ Department of Environmental Engineering, National I-Lan University, Ilan, 260, Taiwan ‡

S Supporting Information *

ABSTRACT: As gradual increases in atmospheric CO2 and depletion of fossil fuels have raised considerable public concern in recent decades, utilizing the unlimited solar energy to convert CO2 to fuels (e.g., formic acid and methanol) apparently could simultaneously resolve these issues for sustainable development. However, due to the complicated characteristics of CO2 reduction, the mechanism has yet to be disclosed. To clarify the postulated pathway as mentioned in the literature, the technique of electron paramagnetic resonance (ESR) was implemented herein to confirm the mechanism and related pathways of CO2 reduction under visible light using graphene− TiO2 as catalyst. The findings indicated that CO−• radicals, as the main intermediates, were first detected herein to react with several hydrogen ions and electrons for the formation of CH3OH. For example, the generation of CO−• radicals is possibly the vital rate-controlling step for conversion of CO2 to methanol as hypothesized elsewhere. The kinetics behind the proposed mechanism was also determined in this study. The mechanism and kinetics could provide the in-depth understanding to the pathway of CO2 reduction and disclose system optimization of maximal conversion for further application. CO2 reduction.4−6 After this initiation, CO2−• radicals can be transformed to different radicals with or without the breaking of the C−O bond. The three main possible pathways were discussed elsewhere4,7−9 (i.e., Figure 1). In fact, some of the radicals involved in those pathways also had been confirmed. For example, ·C and ·H radicals have been directly detected during the formation of CH4, CH3OH, and CO.6 In addition, the existence of ·CH3 and ·OCH3 were confirmed via electron paramagnetic resonance (EPR) by Dimitrijevic et al.5 Although radicals of ·C, ·H, ·CH3, and ·OCH3 have been detected, some radicals such as CO−•,10 ·HOOC,10 (CH3)2C· OH,7,11 and HOC·11 still remained open for clarification. In particular, the existence of ·OC, the keystone species, can suggest the viability of the pathway 3 (Figure 1). However, such kinds of radicals have an unpaired electron and are very shortlived, leading to difficulties for analysis. On the other hand, due to the multiple routes of the reactions, the kinetic models were rarely reported for practical use in industry. In fact, the Langmuir−Hinshewood (L−H) kinetic model has been applied

1. INTRODUCTION As carbon dioxide is one of four major greenhouse gases, its reduction has raised considerable public concern for solving worldwide global warming problems. In fact, gradually increased levels of carbon dioxide emissions from fossil fuel consumption are considered one of the top reasons of causing the problem.1 Thus, seeking any alternative for minimization of CO2 emission has aroused significant attention to the public domain recently. Over 4 decades, methods of CO2 reduction have been popularly inspected since the first survey reported in 1979.2 Literature showed that using the light source to convert CO2 as chemical energy seems to be one of the economically feasible methods for CO2 utilization. However, for CO2 reduction, the mechanism is usually complicated since multiple reactions take place in parallel and/ or series simultaneously. As Koppenol and Rush3 indicated, the CO2 molecule can accept one electron to generate the anion CO2−• radical. The stable and linear geometry structure of CO2 was destroyed by the intrusion of such an electron. In other words, compared with CO2, CO2−• radicals are more likely to be broken due to the bending of chemical structure, repulsion of charge, and the destroyed balance with this intruded electron.4 Several studies mentioned that the formation of CO2−• radicals has been widely adopted as the initial step of the © XXXX American Chemical Society

Received: October 30, 2014 Revised: December 30, 2014 Accepted: January 22, 2015

A

DOI: 10.1021/es505301x Environ. Sci. Technol. XXXX, XXX, XXX−XXX

Article

Environmental Science & Technology

Figure 1. Possible mechanism and pathway of CO2 reduction summarized from the literature and this study.

to simulate the photoreduction rate of CO2.8,12 However, the intermediate generated in reactions was still seldom mentioned. To grasp more details of the mechanism, electron paramagnetic resonance (ESR) and GC-mass have been used to analyze the intermediate and final products, respectively. To the best of our knowledge, there are no reports in the literature revealing CO−• radicals as reacting species for CO2 reduction. In addition, the generation of methyl (−CH3) was also confirmed by Fourier transform infrared spectroscopy (FTIR) with verification of the final products (CH3OH). The detailed pathway and mechanism of CO2 reduction through visible light irradiation was discussed herein using graphene loaded TiO2 as catalysts. On the basis of this postulated mechanism, the kinetic modeling of the reaction was also developed to predict time courses of myriads of chemical species formed in reactions. This first-attempt study clarified the feasibility of methanol formation by verification of CO−• as the intermediate. With this better understanding of the CO2 reduction pathway in the graphene/TiO2-related photocatalytic system, the follow-up studies of maximal photoreduction for system optimization will be conducted.

2. MATERIALS AND METHODS 2.1. Preparation of Graphite Oxide. The waste pencil lead (denoted as waste PL carbon), which was previously grinded and sifted through a 200 mesh, was dispersed in 100 mL of NaOH solution via stirring for 24 h at 423 K to extract SiO2 and other impurities. Then, the solid was obtained after washing by a 1 mol/L HNO3 solution and filtering. Finally, the graphite oxide (GO) was synthesized using the modified Hummer’s method.13 The graphite powders were first oxidized by mixing with NaNO3 and H2SO4. Meanwhile, KMnO4 was added slowly, and the vessel was immersed in an ice bath to maintain a temperature below 5 °C. Then, the reaction was carried out for 30 min below 35 °C. Finally, the reaction temperature was increased to 95 °C for 30 min. Regarding the oxidation process, some oxygen containing groups were intruded into the space between two layers for stretching the layer spacing. After ultrasonication, the single layer graphite oxide with significant oxygen-containing groups could be separated from the graphite.13−15 2.2. Preparation of Graphene-TiO2. Graphite oxide was mixed with commercial TiO2 after a 1 h ultrasonication. The mixed solution was charged through a Teflon-lined autoclave B

DOI: 10.1021/es505301x Environ. Sci. Technol. XXXX, XXX, XXX−XXX

Article

Environmental Science & Technology and then oven-heated to 120 °C for 3 h. The final products were obtained by filtration, washing, and subsequent drying at 100 °C overnight. The postulated mechanism of the hydrothermal reduction in this process would be summarized as shown in the supplementary data (Figure S1, Supporting Information). 2.3. Characterization of GN-TiO2. A UV spectrometer (Hitachi, U-3900), scanned from 200 to 800 nm in wavelength, was used to analyze the optical characterization of the graphene-loaded TiO2 samples. The inner structure of the film was analyzed by high-resolution transmission electron microscopy (HR-TEM, JEOL, JEM-2000 FXΠ). The crystal phase of the catalysis was determined via an X-ray diffraction (XRD, X’Pert PRO) instrument using Cu Kα radiation (λ = 0.1541 nm). An elemental analyzer (EA, Elementar, Vario MICRO Cube) and X-ray fluorescence (XRF, Riguka, NEXCG) were used to analyze the percentages of different components. Infrared spectroscopy (FTIR, NiCoLET-iS10) was performed to identify the existing functional groups of materials. A solid-state nuclear magnetic resonance spectrometer (NMR, BRUKER, AVIII 600 WB) was used to identify the percentage of the functional groups in the catalysts. An energy dispersive spectrometer (EDS, OXFORD, Inca Energy 250) was used to confirm the results of the XRF and EA. The surface potential was analyzed by a nanoparticle analyzer (Horiba, SZ100). X-ray photoelectron spectroscopy (XPS, Auger, ESCA) was used to analyze the elements, bond, and valence of the material. Electron paramagnetic resonance (ESR, Bruker, E580 CW) was used to analyze the main radicals, and reaction mechanisms were then proposed. 2.4. Photoreduction. The reduction reaction took place using a double-wall cylindrical quartz reactor (Figure S2, Supporting Information). Recycled water was used in the reactor to maintain isothermal conditions at 25 °C. Before the reaction took place, CO2 was purged for ca. 30 min to exclude oxygen. During the reaction, the reaction solution was irradiated by a middle visible lamp (150 W; 1358 W/m2) under different photocatalyst loadings and pHs. The reduced products were determined by a gas chromatograph with a mass detector using helium as a carrier gas (HP, G1800A) with a fused silica capillary column (60 m × 0.25 mm × 0.25 μm). The heating process of the mass detector was separated to two stages. The first stage was from room temperature to 45 °C with the heating rate of 4 °C/min. The second stage increased the temperature from 45 to 150 °C with a heating rate of 30 °C/min.

Table 1. Comparison of Analyzed Components of Raw Material and Catalysts No.a

C (%)

SiO2(%)

CaO (%)

Fe2O3 (%)

Al2O3 (%)

TiO2 (%)

others (%)

1 2 3 4 5

65.39 77.14 1.38 6.60 13.64

14.5 8.89 1.11 1.55 5.06

0.6 0.23 0.52 0.43 0.91

12.24 8.36 0.182 0.30 0.92

4.9 2.64 0.25 0.48 0.98

0.56 0.42 96.0 90.3 77.46

1.81 2.32 0.55 0.34 1.03

a 1, raw material; 2, after pretreatment; 3, 5%GN-TiO2; 4, 20%GNTiO2; 5, 40%GN-TiO2.

2.2%, and 18.34%, respectively. Two peaks in the Fe 2p spectrum at 711.8 and 724.9 eV indicated the 2p3/2 and 2p1/3 of Fe3+ likely due to the low amount of Fe2O3 present in raw materials.16 The deconvolution of C 1s showed that the three Gaussian peaks centered near 284.8, 285.5, and 286.7 eV could be related to CC, C−C, and C−O, respectively.17,18 The CC bond corresponding to the aromatic sp2 structure suggested a triangular structure formed by four atoms attached together, which indicated the presence of graphene.17 However, the C−C bond assigned to some tetrahedron-structured sp3 carbon species also revealed the defective structure of graphene. This three-dimensional structured carbon confirmed the existence of multiple layered graphene as also verified by the TEM result. The C−O bond in the C 1s and the single bond of oxygen in the O 1s (532.4 eV) could be attributed to the epoxy (−O−) and hydroxyl (−OH) on the graphene surface.17 Correspondingly, the peak around 531 eV in the O 1s spectrum was assigned to be the double bond of oxygen, revealing the existence of carbonyl and carboxyl groups.17 For the impurity of the raw material, the peak in the Si 2p spectrum at 103.3 eV also suggested the presence of SiO2 as confirmed by XRF and EDS results.19 To quantify the total percentage of CC in the GN-TiO2, the nuclear magnetic resonance (NMR) has been utilized in this research. As shown in Figure 3A, the main bonds of carbon could be identified as CC, CH2−O−Ti, and CH3O, respectively.20,21 Two peaks in the 1H spectrum (Figure 3B− D) around 2 and 6 ppm could be attributed to the CH2−O/ CH3−O (C−O) and CC bond22 which also confirmed the result of the 13C spectrum and XPS result (Figure 2). The quantification of CC and C−O bonds in the GN-TiO2 indicated that the percentages of the CC bond in 5% GNTiO2, 20% GN-TiO2, and 40% GN-TiO2 were 73.79% ± 0.5%, 87.44% ± 0.7%, and 78.08% ± 0.4%, respectively. This phenomenon suggested that even most oxygen-containing functional groups were reduced during the hydrothermal process; the residuals of some oxygen containing groups (about 20%) remained in the catalysts which directly affect the superhydrophilic effect of the catalysts. As shown in transmission electron microscopy (TEM) micrographs of GN-TiO2 (Figure 4), although the images clearly illustrated the flake-like shape of the graphene, many layers still remained to be interactively linked as compared with pure graphite prepared catalyst. Moreover, three layers of graphene stacked together were clearly seen in images.23,24 This phenomenon suggested the existence of nonsingle-layered graphene as confirmed by XPS result. The defect of the structure may have been normally formed as the residue of the impurities in the catalyst; therefore, incomplete reduction of the hydrothermal process took place. For all the samples with

3. RESULTS AND DISCUSSION 3.1. Characteristics of the Catalysts. As shown in Table 1, the predominant components of raw materials (waste PL carbon) were carbon, SiO2, Fe2O3, Al2O3, CaO, and other trace impurities. After pretreatment, the amount of impurities was partially reduced and the percentage of carbon was significantly increased by ca. 12%. With the graphene loading increased, the amount of carbon, SiO2, and other impurities also increased. The impurities of 5% GN-TiO2, 20% GN-TiO2, and 40% GNTiO2 were 2.62%, 3.10%, and 8.94%, respectively. In fact, SiO2 was the major impurity of the raw material as clearly shown in the EDS results (Figure S3, Supporting Information). In addition, as indicated in Figure 2, the entire range of binding energy from 0 to 1400 eV was analyzed by XPS in order to identify the dominant atomic components. The main species of C 1s, O 1s, Si 2p, and Ti 2p were ca. 39.16%, 40.29%, C

DOI: 10.1021/es505301x Environ. Sci. Technol. XXXX, XXX, XXX−XXX

Article

Environmental Science & Technology

Figure 2. XPS spectra of 40% GN-TiO2.

various graphene loadings, TiO2 nanoparticles and a wrinkled edge were observed clearly (Figure 4). The TiO2 nanoparticles

with an average size of 10−20 nm were dispersed on the surface of graphene layers. With the increased content of TiO2, the D

DOI: 10.1021/es505301x Environ. Sci. Technol. XXXX, XXX, XXX−XXX

Article

Environmental Science & Technology

Figure 3. (A) 13C NMR of 40% GN-TiO2; (B) 1H NMR of 40% GN-TiO2; (C) 1H NMR of 20% GN-TiO2; (D) 1H NMR of 5% GN-TiO2; (E) 1H NMR of 40% GN-TiO2 after being recycled three times. E

DOI: 10.1021/es505301x Environ. Sci. Technol. XXXX, XXX, XXX−XXX

Article

Environmental Science & Technology

Figure 4. TEM images of GN-TiO2.

from the adsorbed water.25 The low-frequency absorption below 1000 cm−1 was assigned to the Ti−O−Ti vibration from the components of TiO2.23 A new absorption band cause by the skeletal vibration of the graphene sheets (1600 cm−1) was revealed in the curves of the GN-TiO2.25 The adsorption bands

graphene became less transparent and was almost sheltered by TiO2 nanoparticles as revealed in the images of 40% GN-TiO2. As shown in profiles of different graphene-loaded catalysts via FTIR (Figure 5A), the absorption peak at 3400 cm−1 was attributed to O−H stretching vibration of the surface hydroxyl F

DOI: 10.1021/es505301x Environ. Sci. Technol. XXXX, XXX, XXX−XXX

Article

Environmental Science & Technology

at 1720, 1220, and 1045 cm−1 were attributed to the groups of CO, C−OH, and O−C, respectively.26−28 The absorption band at 1385 cm−1 was assigned to the stretching vibration of the C−H bond. The details of the adsorption bands are summarized in Table S1, Supporting Information. As the content of the graphene increased (Figure 5A), the adsorption peak responding to CO was eliminated. The loss of some functional groups led to better conductivity, and the residuals of some oxygen containing groups (such as C−OH, C−O) were more favorable for maintaining a good dispersion in solution.29 As a matter of fact, the reductive efficiency of CO2 is independent of the degree of dispersion of the catalyst in water. At UV light irradiation, water molecules would cover most of the surface of TiO2 catalyst to form the superhydrophilic surface.30 As indicated in Figure 5B, the dispersed TiO2 could be capable of effective adsorption of different carbon species in water. The peaks around 1020 and 1500 cm−1 represent the structure of C−O appearing in all of the CO2, NaHCO3, and Na2CO3 samples. The adsorption band of bicarbonate (C−H), which appeared at 1427 cm−1, could be clearly observed in the TiO2-bearing NaHCO3 solution, suggesting the adsorption of HCO3−.31 Moreover, the signal peak at 1053 cm−1, appearing in the TiO2 after a 10 h reaction, was attributed to the formation of CH3− substituents.32 The existence of CH3− not only verified the adsorption of carbonate and bicarbonate on the catalytic surface but also confirmed the production of CH3OH. In general, the efficient dispersion of catalysts in water significantly enhanced adsorption of carbonate and bicarbonate in the aqueous solution. To inspect the surface polarity of composite materials, toluene and water were thus used as solvent. As shown in Figure S4, Supporting Information, the catalysts were totally dispersed in the aqueous phase rather than in the organic phase (i.e., toluene), which confirmed the superhydrophilic surface of graphene-modified materials. The superhydrophilic effect of composite materials could be attributed to the remaining oxygen containing groups on the polarity-rich graphene surface. As indicated in UV−vis absorption spectra of the pure TiO2 and GN-TiO2 composites (Figure S5, Supporting Information), the band gaps (band gap = 1240/wavelength)14,33,34 of the catalysts were 3.10, 2.95, 2.88, and 2.66 eV at the graphene loadings of 0%, 5%, 20%, and 40%, respectively. To grasp a glimpse of the catalysts in reaction, the crystal phases of TiO2 and GN-TiO2 were investigated by an X-ray diffractometer. Figure S6, Supporting Information, showed that the major crystal phase of TiO2 was relatively rutile, which proved to be the most appropriate crystal phase for both adsorption and reaction.35,36 To disclose the charge transfer capability of GNTiO2, the photoelectrochemical method was implemented (Figure S7, Supporting Information). Apparently, no significant current was observed except with the irradiation of UV light. For all catalysts, the photoresponse increases with the increased graphene loading could be ascribed to the increasing separation efficiency of the photogenerated electrons.31 The existence of graphene could act as a source of active sites to maintain photoreductive capabilities for CO2 reduction and exhibit a better photocurrent response.31 3.2. Performance of the Materials. As a matter of fact, the photocatalytic reduction of CO2 is a complex multistep process apparently affected by carbon species in the solution. As it significant affected the carbon species, inevitably, the pH level is vital to CO2 reduction. As shown in Figure 6A, the optimal yields of formic acid and methanol were obtained in

Figure 5. (A) Infrared spectroscopy spectra of different catalysts; (B) infrared spectroscopy spectra of TiO2 reacted with different reactants; (C) infrared spectroscopy spectra of recycled catalysts. G

DOI: 10.1021/es505301x Environ. Sci. Technol. XXXX, XXX, XXX−XXX

Article

Environmental Science & Technology

mobility of graphene is ca. 200 000 cm2 V−1 s−1, which is much larger than 15 000 cm2 V−1 s−1for graphite. As is known, a high efficiency of reduction required the adsorption of different carbon species upon the catalytic surface, which is of course affected by the surface potential of the catalyst materials. At pH levels of 3, 7, and 13, the predominant carbon species were H2CO3, HCO3−, and CO32−, respectively,37 as predicted by analytical chemistry. The surface potentials of the catalysts used herein were found to be less than zero in water for the negatively charged oxygen-containing groups remaining in the graphene. At acidic pHs, positively charged protons were more likely to be adsorbed on the surface of catalysts, as compared with uncharged H2CO3. The results of the potential analysis showed that the surface potential of 40% GN-TiO2 was ca. −8.8 eV under pH 3, evidently much less than the surface potential at pH 7 (−68.3 eV) or pH 13 (−63.0 eV) as confirmed by the adsorption of H+ under acidic pHs. At alkaline pHs, the negatively charged CO32− species was more likely to be expelled by the negatively charged surface of the catalyst, as compared with the HCO3− species. On the other hand, the dissolved CO2 in water also tended to affect reduction performance. Due to the interactive effect of pH, the carbon species of dissolved CO2 in water was totally different. As indicated by EA analysis, the dissolved CO2 levels in water were 0.0021, 0.033, and 0.108 mol/L at pH 4, 7, and 13, respectively. At acidic pHs, although H+ could be sufficiently supplied, the concentration of H2CO3 in the solution was still limited. Similarly, significant amounts of dissolved CO32− in aqueous phase at alkaline pHs also led to the lower reduction rate due to a small amount of hydrogen ions available for reaction. As shown in Figure 6A, the methanol yield was generally lower than the formic acid yield due to the higher energy and more electrons required. The methanol conversion was ca. 6 electrons/mol and 703 kJ/mol, which was evidently higher than 2 electrons/mol and 275 kJ/mol for formic acid. 38 Furthermore, formic acid is favorably generated, since it is a sole intermediate in the methanol formation process.7 The reduction path is HCOOH → CH3OH.7 Moreover, the effect of the graphene loading of 5%, 20%, and 40% on the performance of the catalysts was also determined. The results indicated that the photocatalytic efficiency increased with increases in the graphene loading, as shown in Figure 6B. With the graphene loading increased from 0% to 40%, the yields of formic acid and methanol increased simultaneously. This phenomenon resulted from the strong charge transfer capacity from TiO2 to graphene, which effectively prevented the recombination of the photogenerated electron and hole.39 In addition, the degree of dispersion of the catalyst also affected the efficiency. The yield efficiency was higher when the degree of dispersion was greater.39 Some oxygen containing groups remained on the graphene surface (e.g., C−OH and C−O) which were favorable for maintaining a promising dispersion in solution.40 The higher the graphene loading, the better is the dispersion efficiency of the catalyst.29 On the other hand, a larger surface area of GN-TiO2 with more graphene loading (Table S2, Supporting Information) offered more active sites of adsorption and reactive centers of photocatalysis.41 Due to aforementioned reasons, further increases in the graphene loading to 60% and 80% were also analyzed in this research. However, a further increase in the graphene content also led to a significant reduction of the

Figure 6. (A) Photocatalytic reduction of carbon dioxide at different pHs; (B) photocatalytic reduction of carbon dioxide using different content-bearing catalysts; (C) photocatalytic reduction of carbon dioxide of different recycle times.

the neutral pH and the lowest yields were obtained at alkaline condition (pH 13). As compared with the pure graphiteprepared GN-TiO2, the performance waste pencil lead prepared materials show worse efficiency (Figure 6A), which was attributed to the defective structure in the graphene as confirmed by TEM for the remaining impurities in the catalysts (also revealed as the result of the EA). The charge carrier H

DOI: 10.1021/es505301x Environ. Sci. Technol. XXXX, XXX, XXX−XXX

Article

Environmental Science & Technology Table 2. Quantum Efficiency for the CO2 Photo Reduction in This Research and the Literature light source

catalysts

Xe arc lamp (254 nm < λ < 400 nm)

0.5% Cu/TiO2−SiO2

8W Hg lamp (254 nm) UV−vis light 100 W halogen lamp visible light (λ ≑ 670 nm)

2.0% Cu/TiO2 Au−Cu loaded STO/TiO2 coaxial nanotube arrays NiO/InTaO4 Recycle 0 Recycle 1 Recycle 2 Recycle 3

⌀formic acid(%) =

6 × moles of methanol yield moles of photon absorbed by catalyst 2 × moles of formic acid yield moles of photon absorbed by catalyst

quantum efficiency (%) 0.85 0.56 10.02 2.51 0.063 2.202 2.307 1.45 1.82 1.44 2.35 1.34 2.27

ref. 43 44 45 46 this study

⎡ ⎛ μW ⎞ total photons absorbed by catalysts = ⎢absorbed photo flux ⎜ 2 ⎟ ⎝ cm ⎠ ⎣

photocatalytic performance possibly due to the scattering and shielding of the light.41 In addition, recycle numbers as a critical factor of utilization were investigated to determine long-term stability of the catalyst. For each cycle, a sample was collected and vacuumdried at room temperature overnight. As indicated in Figure 6C, the yields of both formic acid and methanol inevitably decreased in the second cycle. The slight decrease could be explained by the reduction of some oxygen-containing functional groups under light irradiation upon the catalysts.42 According to FTIR results of the recycled catalysts (Figure 5C), the C−O bonds remaining in the catalyst at 1045 cm−126 were nearly vanished in the third and fourth cycle. This phenomenon was also confirmed by an element analyzer; the percentage of carbon in the catalyst (40% GN-TiO2) changed slightly, from 13.64% to 14.72% and from 14.77% to 14.61%. The small increase in the carbon content was very likely a result of the disappearance of some oxygen groups on the surface of the catalysts. Furthermore, as the result of 1H NMR of 40% GNTiO2 after being recycled three times (Figure 3E) shows, the peak of the C−O bond, compared with that of 40% GN-TiO2, disappeared, which also confirmed the reduction of oxygencontaining functional groups during the photocatalytic process. As aforementioned, the oxygen-containing functional groups favorable for the superhydrophilic effect of the catalysts can effectively promote the efficiency of the reaction. As the recycle number increased, the worse dispersion in solution of the catalyst led to the slight decrease of the reduction efficiency. After 2 cycles, the yield of products tended to be favorably stabilized, which simply suggested the promising feasibility of catalyst reuse. To quantify the performance and stability of the catalysts to compare with literature to date, the quantum efficiency is calculated by eqs 1 and 2. Note that methanol and formic acid as the major products require two and six electrons for CO2 conversion, respectively. ⌀methanol(%) =

products CO CH4 CH3OH CH4 CH3OH HCOOH CH3OH HCOOH CH3OH HCOOH CH3OH HCOOH CH3OH

⎤ × surface area of reactor (cm 2) × radiation time (s)⎥ ⎦ /⎡⎣each photon energy (J) × 6.02 × 1023⎤⎦

E=

hc λ

(3)

(4) −34

where h = 6.626 × 10 J·s is a universal constant called Planck’s constant, c is the speed of light (3 × 108 m/s), and λ indicates the wavelength of the light source. The results of the quantum efficiency (Table 2) indicated that, even after being recycled, the quantum efficiency was still 1.45% and 1.82% for formic acid and methanol, respectively, which is still much better than many photocatalysts.43−46 3.3. Mechanisms. As a matter of fact, for decades, the reduction pathway of CO2 has yet to be disclosed. As literature revealed,4,7−9 the formation of CO2−• triggered the initiation of the reaction (Figure 1). The formation of CO2−• was first reported by Rasko and Solymosi in 1994,47 and then, the radical was widely proposed to explain the transfer of CO2. With a different e− and H+ involved in the reaction, the breakage of the C−O bond and formation of a new C−H bond can occur at different positions. Myriads of intermediates might also be generated, as well as the final products.48 To test the postulated mechanism of this reaction, PBN (N-tert-butyl-2phenylnitrone) and DMPO (5,5-dimethyl-1-pyrroline N-oxide) were used as the spin trapper in order to detect short-lived, low-concentration radicals formed in the system. The oxygen center radicals reacted with PBN to produce relatively longlived R2N−O· radicals.49 On the other hand, DMPO as a kind of widely used spin trapper can react with ·OH radicals to form stable adducts (DMPO−OH) which could be analyzed by ESR. Due to the interactions between electron spin and nuclear spin, ESR (νo (MHz) = 1.4·g·Ho) can have a several line hyperfine structure as shown in Figure 7A. The introduction of g factor and hyperfine structure is useful to understand the electronic state of atoms and molecules.50 At first, three significant peaks were shown from the result of the ESR spectrum (Figure 7A) by using PRN as trapper. When catalysts were absent, visible light irradiation caused no ESR signal with a 60 min irradiation. However, high levels of radicals were trapped in water by using GN-TiO2 to stimulate the reaction. With increases in the graphene loading, the intensity of ESR signal was also increased. This phenomenon indicated

(1)

(2)

To analyze the moles of photon adsorbed by catalyst (eq 3), the energy of a photon under visible light (main wavelength was 670 nm in this study) should be calculated first, as shown by the following eq 4:43,44 I

DOI: 10.1021/es505301x Environ. Sci. Technol. XXXX, XXX, XXX−XXX

Article

Environmental Science & Technology

Figure 7. (A) The profiles of ESR signal of the radicals in the reaction; (B) route based mechanism of the photoreduction taking place on the catalysts.

lower quasi-Fermi level of graphene, the photoirradiated electrons can be transferred from TiO2 to graphene, preventing the recombination of photoinduced electrons and holes. The separated electrons can then react with adsorbed CO2 to generate CO2−•. As Anpo et al. showed, when CO2−• attached to the surface of the catalyst by the carbon atom, the C−O bond breaking evidently required an electron and a hydrogen ion.51 The generated CO reacted with one electron to form a carbon monoxide radical. Due to aforementioned points, the

that the catalysts with more graphene loading must have higher capacity for CO2 reduction since more generated radicals were available for reaction. According to the g factor, which is a dimensionless quantity characterizing the magnetic moment and gyromagnetic ratio of a particle or nucleus, the radical formed in this study was very likely to be CO−•.49 Different from previous research, CO−• may be the major radical species of the rate-determining step of the reductive reaction. At the initiation of the reaction, for the J

DOI: 10.1021/es505301x Environ. Sci. Technol. XXXX, XXX, XXX−XXX

Article

Environmental Science & Technology mechanism of the reduction in this work could be modified as Figure 7B. At the initial step of the reaction, dissolved CO2 reacted with hydrogen ions and electrons to generate CO2−•. After the activation of dissolved CO2, the reduction took place with two pathways very possibly in parallel. In one pathway, the attachment of hydrogen ions to the oxygen in the CO2−• radicals led to the cleavage of the C−O bond. Then, the generated CO could accept an electron to form carbon monoxide radical. The CO−• radical could further combine with four hydrogen ions to form CH3OH. In the second pathway, some CO2−• radicals with no the breakage of the C− O bond favored the generation of carboxyl radicals (·COOH). The formation of HCOOH by combining with one H+ is the intermediate of CH3OH. This reduction path could then be summarized as CO2 (dissolved) → HCOOH → CH3OH. For the analysis of hydroxyl radicals, a typical EPR signal of DMPO−OH with an intensity ratio of 1:2:2:1 was observed in Figure 7A.52 With the increased graphene loading, the intensity for hydroxyl radicals was also increased, indicating the higher photoactivity under visible light. Furthermore, the highest response of the ·OH signal was also observed by using 40% GN-TiO2 as catalyst, which was confirmed to be the most effective for CO2 reduction. As reported in the literature, those ·OH radicals were generated by oxidation of water molecules and OH− ions were adsorbed at the particle surface (GNTiO2), as described in eqs 5 and 6.53 Due to the superhydrophilic property of the catalysts, water molecules are likely to be adsorbing on the catalytic surface for ·OH generation. Then, the generated ·OH radicals can react with CH3 to form CH3OH as identified in Figure 1. +

H 2O (ads) + hVB → ·OH + H

+

OH− + hVB+ → ·OH

(11)

k 3k5 k4 + k5

(12)

For the total generation rate of formic acid, one could obtain d[HCOOH] = k1[CO2 ] − k 2[HCOOH] dt

(13)

The concentration of formic acid can be expressed as follows: [HCOOH] =

k1[CO2 ](1 − e−k 2t ) k2

(14)

With the aforementioned kinetic models (6−14) to simulate the behavior of formic acid and methanol, the kinetic parameters for total generation rates were summarized in Table 3 (Figure 8). Table 3. Rate Constants Predicted from the Proposed Kinetic Model for the Photoreduction 5% GN-TiO2 20% GN-TiO2 40% GN-TiO2 pH = 4 pH = 7 pH = 13 Recycle 0 Recycle 1 Recycle 2 Recycle 3

(5) (6)

k1 (g−1h−1)

k2 (g−1h−1)

M (g−1h−1)

573 651 994 9854 994 198 994 630 630 630

0.0213 0.0452 0.0541 0.0769 0.0541 0.0323 0.0541 0.0486 0.0479 0.0515

112 181 247 3712 247 46.6 247 188 194 221

According to predicted kinetic constants, as graphene loading increased, the conversion rates of k1, k2, and M also increased. This finding suggested that more loaded graphene proportionally led to higher photoactivity of the catalysts. On the other hand, when pH changed from 4 to 13, both the rate constants of k1 and k2 and M decreased. That is, the concentration of H+ involved in the reaction also significantly affected the reaction rate. The large amount of protons under low pH apparently resulted in higher reaction rate. Furthermore, when with the recycle numbers increased, the kinetic parameters k1, k2, and M, as the indicators of the reaction, tended to be stabilized after a decrease of the first cycle, which showed the stability of the materials for long-term applications. As shown in Figure 8, the kinetic model seemed to be feasible for model prediction. This phenomenon confirmed the hypothesis of the abovementioned mechanism. As myriads of new pollutants still are continuously released to the atmosphere due to industrial development, CO2 elimination or utilization is urgent to develop an effective method to control the fate of CO2. The study on the grapheneTiO2 based photoreduction of CO2 under visible light

(8)

As the radical in the reaction is short-lived, the generated radicals cannot be significantly accumulated in aqueous solution. According to the quasi-steady state assumption, the concentration of CO−• at any instant in time was approximately constant as shown below:

k 3[CO·] k4 + k5

d[CH3OH] = k 2[HCOOH] − M[CO·] dt

M=

As aforementioned in Section 3.3, the intermediate of the reaction CO−• was generated from the surface of the catalysts and then transferred to methanol. Simultaneously, it could be interconverted to CO2. Therefore, the total generation of CO−• radicals could be expressed as

[CO·] =

(10)

where M is a combined rate constant of k3, k4, and k5 as

3.4. Kinetic Modeling. According to the above-mentioned mechanism, the conversion of methanol could be considered as two steps in the presence and the absence of generated CO−• radical. The formic acid was assumed to be produced from dissolved CO2 in solution. Therefore, the main reaction for CO2 conversion can be presented as

d[CO·] = k 3[CO2 ] − (k4 + k5)[CO·] dt

d[CH3OH] = k 2[HCOOH] − k5[CO·] dt k [CO·] = k 2[HCOOH] − k5 3 k4 + k5

(9)

Then, the formation rate of methanol can be shown as follows: K

DOI: 10.1021/es505301x Environ. Sci. Technol. XXXX, XXX, XXX−XXX

Article

Environmental Science & Technology

The significance of this study is to confirm the fate of CO2 during photoreduction and to suggest a basic guideline for effective CO2 removal with practical values.



ASSOCIATED CONTENT

S Supporting Information *

The mechanism of preparation of TiO2 and graphene composites by hydrothermal reaction; the structure of the reactor; the element test of the catalyst by EDS; the degree of dispersion of the catalysts in different solvents; UV−vis curves of catalysts; XRD patterns of various catalysts; the photoelectrochemical characterization of the catalysts; additional tables. This material is available free of charge via the Internet at http://pubs.acs.org.



AUTHOR INFORMATION

Corresponding Authors

*Tel.: +86 5922181613; fax: +86 5922185889. *Tel.: 886-3-9357400; fax: 886-3-9359674; e-mail: ctchang@ niu.edu.tw. Notes

The authors declare no competing financial interest.



REFERENCES

(1) Zhao, Z. H.; Fan, J. M.; Wang, Z. Z. Photo-catalytic CO2 reduction using sol−gel derived titania-supported zinc-phthalocyanine. J. Cleaner Prod. 2007, 15, 1894−1897. (2) Inoue, T.; Fujishima, A.; Konishi, S.; Honda, K. Photoelectrocatalytic reduction of carbon dioxide in aqueous suspensions of semiconductor powders. Nature 1979, 277, 637−638. (3) Koppenol, W. H.; Rush, J. D. Reduction potential of the carbon dioxide/carbon dioxide radical anion: A comparison with other C1 radicals. J. Phys. Chem. 1987, 91 (16), 4429−4430. (4) Habisreutinger, S. N.; Lukas, S. M.; Jacek, K. S. Photocatalytic reduction of CO2 on TiO2 and other semiconductors. Angew. Chem., Int. Ed. 2013, 52, 7372−7408. (5) Dimitrijevic, N. M.; Vijayan, B. K.; Poluektov, O. G.; Rajh, T.; Gray, K. A.; He, H.; Zapol, P. Role of water and carbonates in photocatalytic transformation of CO2 to CH4 on titania. J. Am. Chem. Soc. 2011, 133 (11), 3964−3971. (6) Anpo, M.; Chiba, K. Photecatalytic reduction of CO2 on anchored titanium oxide catalysts. J. Mol. Catal. 1992, 74 (1−3), 207− 212. (7) Dey, G. R.; Belapurkar, A. D.; Kishore, K. Photo-catalytic reduction of carbon dioxide to methane using TiO2 as suspension in water. J. Photochem. Photobiol., A 2004, 163 (3), 503−508. (8) Lo, C. C.; Hung, C. H.; Yuan, C. S.; Wu, J. F. Photoreduction of carbon dioxide with H2 and H2O over TiO2 and ZrO2 in a circulated photocatalytic reactor. Sol. Energy Mater. Sol. Cells 2007, 91, 1765− 1774. (9) Handoko, A. D.; Li, K.; Tang, J. Recent progress in artificial photosynthesis: CO2 photoreduction to valuable chemicals in a heterogeneous system. Curr. Opin. Chem. Eng. 2012, 2, 200−206. (10) Gattrell, M.; Gupta, N.; Co, A. A review of the aqueous electrochemical reduction of CO2 to hydrocarbons at copper. J. Electroanal. Chem. 2006, 594 (1), 1−19. (11) Cole, E. B.; Lakkaraju, P. S.; Rampulla, D. M.; Morris, A. J.; Abelev, E. B. A. Using a one-electron shuttle for the multielectron reduction of CO2 to methanol: Kinetic, mechanistic, and structural insights. J. Am. Chem. Soc. 2010, 132 (33), 11539−11551. (12) Tahir, M.; Amin, N. S. Photocatalytic CO2 reduction and kinetic study over In/TiO2 nanoparticles supported microchannel monolith photoreactor. Appl. Catal., A: Gen. 2013, 467, 483−496. (13) Arvidsson, R.; Kushnir, D.; Sanden, B. A.; Molander, S. Prospective life cycle assessment of graphene production by

Figure 8. Comparison of time-series profiles of model predicted and experimental data of formic acid and methanol concentrations during reaction.

irradiation presented herein effective CO2 conversion and an in-depth understanding to the pathway of CO2 reduction. The catalysts prepared from waste PL lead could utilize the visible light abundant in the daylight to generate renewable fuels. The intermediate of COCO2−• verified in the work of course fulfills the mystery of the unclear mechanism of CO2 reduction, which is very likely a rate-determining step for methanol formation as proposed in other research. Follow up study of the kinetic modeling can be used to determine the lowest reaction routes in order to enhance the reaction rate for system optimization. L

DOI: 10.1021/es505301x Environ. Sci. Technol. XXXX, XXX, XXX−XXX

Article

Environmental Science & Technology ultrasonication and chemical reduction. Environ. Sci. Technol. 2014, 48, 4529−4536. (14) Min, Y. L.; Zhang, K.; Zhao, W.; Zheng, F. C.; Chen, Y. C.; Zhang, Y. G. Enhanced chemical interaction between TiO2 and graphene oxide for photocatalytic decolorization of methylene blue. Chem. Eng. J. 2012, 193−194, 203−210. (15) Guo, S.; Dong, S. Graphene nanosheet: Synthesis, molecular engineering, thin film, hybrids, and energy and analytical applications. Chem. Soc. Rev. 2011, 40, 2644−2672. (16) Khalida, N. R.; Zhang, l. H.; Ahmedb, E.; Zhang, Y.; Chana, H.; Ahmad, M. Synergistic effects of Fe and graphene on photocatalytic activity enhancement of TiO2 under visible light. Appl. Surf. Sci. 2012, 258, 5827−5834. (17) Luisa, M.; Martínez, P.; Torres, S. M.; Likodimos, V.; Falaras, P.; Figueiredo, J. L.; Faria, J. L.; Silva, A. M. T. Role of oxygen functionalities on the synthesis of photocatalytically active grapheneTiO2 composites. Appl. Catal., B: Environ. 2014, 158−159, 329−340. (18) Xing, M.; Shen, F.; Qiu, B.; Zhang, J. Highly-dispersed borondoped graphene nanosheets loaded with TiO2 nanoparticles for enhancing CO2 photoreduction. Sci. Rep. 2014, 4, 6341−6348. (19) Fang, Q.; Meier, M.; Yu, J. J.; Wang, Z. M.; Zhang, J.-Y.; Wu, J. X.; Kenyon, A.; Hoffmann, P.; Boyd, I. W. FTIR and XPS investigation of Er-doped SiO2−TiO2 films. Mater. Sci. Eng. B 2003, 105, 209−213. (20) Cui, J. F.; Fang, X. W.; Schmidt-Rohr, K. Quantification of C C and CO surface carbons in detonation nanodiamond by NMR. J. Phys. Chem. C 2014, 118, 9621−9627. (21) Rockafellow, E. M.; Fang, X.; Trewyn, B. G.; Klaus, S.; Jenks, W. S. Solid-state 13C NMR characterization of carbon-modified TiO2. Chem. Mater. 2009, 21, 1187−1197. (22) Silverstein, R. M.; Webster, F. X.; Kiemle, D. J. Spectrometric identification of organic compounds, seventh ed.; Wiley: New York, 2011. (23) Zhao, J.; Wang, Z.; Zhao, Q.; Xing, B. Adsorption of phenanthrene on multilayer graphene as affected by surfactant and exfoliation. Environ. Sci. Technol. 2014, 48, 331−339. (24) Qiu, B.; Xing, M.; Zhang, J. Mesoporous TiO2 nanocrystals grown in situ on graphene aerogels for high photocatalysis and lithium-ion batteries. J. Am. Chem. Soc. 2014, 136, 5852−5855. (25) Hu, G.; Tang, B. Photocatalytic mechanism of graphene/titanate nanotubes photocatalyst under visible-light irradiation. Mater. Chem. Phys. 2013, 138, 608−614. (26) Zhang, Y.; Pan, C. TiO2/grapheme composite from thermal reaction of grapheme oxide and its photocatalytic activity in visible light. J. Mater. Sci. 2011, 46, 2622−2626. (27) Guo, J.; Zhu, S.; Chen, Z.; Li, Y.; Yu, Z.; Liu, Q.; Li, J.; Feng, C.; Zhang, D. Sonochemical synthesis of TiO2 nanoparticals on grapheme for use as photocatalyst. Ultrason. Sonochem. 2011, 5, 1082−1090. (28) Xu, Y.; Bai, H.; Lu, G.; Li, C.; Shi, G. Q. Flexible graphene films via the filtration of water-soluble noncovalent functionalized graphene sheets. J. Am. Chem. Soc. 2008, 130, 5856−5857. (29) Adel, A.; Ismail, R.; Geioushy, A.; Bouzid, H.; Saleh, A. A.; Hajry, A. A.; Bahnemann, D. W. TiO2 decoration of graphene layers for highly efficient photocatalyst: Impact of calcination at different gas atmosphere on photocatalytic efficiency. Appl. Catal., B: Environ. 2013, 129, 62−70. (30) Jeffrey, C.; Wu, S.; Lin, H. M.; Lai, C. L. Photo reduction of CO2 to methanol using optical-fiber photoreactor. Appl. Catal., A: Gen. 2005, 296, 194−200. (31) Wu, Y.; Lu, G.; Li, S. Micro fluidic reactor for the electrochemical reduction of carbon dioxide: The effect of pH. J. Photochem. Photobiol., A 2006, 181, 263−267. (32) Whipple, D. T.; Finke, E. C.; Kenis, P. J. A. The long-term photocatalytic stability of Co2+-modified P25-TiO2 powders for the H2 production from aqueous ethanol solution. Electrochem. Solid-State Lett. 2011, 13, B109−B111. (33) Hsieh, C. T.; Fan, W. S.; Chen, W. Y.; Lin, J. Y. Adsorption and visible-light-derived photocatalytic kinetics of organic dye on Codoped titania nanotubes prepared by hydrothermal synthesis. Sep. Purif. Technol. 2009, 67, 312−318.

(34) Mahmodi, G.; Sharifniaa, S.; Rahimpour, F.; Hosseini, S. N. Photocatalytic conversion of CO2 and CH4 using ZnO coated mesh: Effect of operational parameters and optimization. Sol. Energy Mater. Sol. Cells 2013, 111, 31−40. (35) Fujishima, A.; Rao, T. N.; Tryk, D. A. Titanium dioxide photocatalysis. J. Photochem. Photobiol., C 2000, 1, 1−21. (36) Indrakanti, V. P.; Kubicki, J. D.; Schobert, H. H. Photoinduced activation of CO2 on Ti-based heterogeneous catalysts: Current state, chemical physics-based insights and outlook. Energy Environ. Sci. 2009, 2, 745−758. (37) Ku, Y.; Lee, W. H.; Wang, W. Y. Photocatalytic reduction of carbon dioxide to methanol and formic acid by graphene-TiO2. J. Mol. Catal. A: Chem. 2014, 212, 191−196. (38) Liu, J.; Xu, S.; Liu, L.; Sun, D. D. The size and dispersion effect of modified graphene oxide sheets on the photocatalytic H2 generation activity of TiO2 nanorods. Carbon 2013, 60, 445−452. (39) Graetzel, M. Energy Resources through Photochemistry and Catalysis; Academic Press: New York, 1983; p 573. (40) Tu, W.; Zhang, Y.; Liu, Q.; Yan, S.; Bao, S.; Wang, X.; Xiao, M.; Zou, Z. An in situ simultaneous reduction-hydrolysis technique for fabrication of TiO2-graphene 2D sandwich-like hybrid nanosheets: Graphene-promoted selectivity of photocatalytic-driven hydrogenation and coupling of CO2 into methane and ethane. Adv. Funct. Mater. 2013, 23, 1743−1749. (41) Wang, Q.; Wu, W.; Chen, J.; Chu, G.; Ma, K.; Zou, H. Novel synthesis of ZnPc/TiO2 composite particles and carbon dioxide photocatalytic reduction efficiency study under simulated solar radiation conditions. Colloids Surf., A 2012, 409, 118−125. (42) Cai, X.; Ma, R.; Tadashi, C. O.; Nobuyuki, S.; Asami, F.; Takayoshi, S. Superlattice assembly of graphene oxide (GO) and titania nanosheets: Fabrication, in situ photocatalytic reduction of GO and highly improved carrier transport. Nanoscale 2014, 6, 14419− 14427. (43) Li, Y.; Wang, W. N.; Zhan, Z.; Woo, M. H.; Wu, C. Y.; Biswas, P. Photocatalytic reduction of CO2 with H2O on mesoporous silica supported Cu/TiO2 catalysts. Appl. Catal., B: Environ. 2010, 100, 386−392. (44) Tseng, I. H.; Chang, W. C.; Wu, J. C.S. Photoreduction of CO2 using sol−gel derived titania and titania-supported copper catalysts. Appl. Catal., B: Environ. 2002, 37, 37−48. (45) Kang, Q.; Wang, T.; Li, P.; Liu, L.; Chang, K.; Li, M.; Ye, J. Photocatalytic reduction of carbon dioxide by hydrous hydrazine over Au−Cu alloy nanoparticles supported on SrTiO3/TiO2 coaxial nanotube arrays. Angew. Chem., Int. Ed. 2014, 53, 1−6. (46) Wang, Z. Y.; Chou, H. C.; Wu, J. C.S.; Din, P. T.; Guido, M. CO2 photoreduction using NiO/InTaO4 in optical-fiber reactor for renewable energy. Appl. Catal., A: Gen. 2010, 380, 172−177. (47) Rasko, J.; Solymosi, F. Infrared spectroscopic study of the photoinduced activation of CO2 on TiO2 and Rh/TiO2 catalysts. J. Phys. Chem. 1994, 98, 7147−7152. (48) Wang, Z.; Ma, W.; Chen, C.; Ji, H.; Zhao, J. Probing paramagnetic species in titania-based heterogeneous photocatalysis by electron spin resonance (ESR) spectroscopyA mini review. Chem. Eng. J. 2011, 170, 353−362. (49) Grün, R. Dose determination on fossil tooth enamel using ESR spectrum deconvolution with Gaussian and Lorentzian peak. Ancient TL 1998, 16, 51−55. (50) Wertz, J. E.; Bolton, J. R. Electron spin resonance: Elementary theory and practical applications; McGraw-Hill: New York, 1972. (51) Anpo, M.; Yamashita, H.; Ichihashi, Y.; Ehara, S. Photocatalytic reduction of CO2 with H2O on various titanium oxide catalysts of special interest. J. Electroanal. Chem. 1995, 396, 21−26. (52) Han, S. K.; Hwang, T.; Yoon, Y.; Kang, J. W. Evidence of singlet oxygen and hydroxyl radical formation in aqueous goethite suspension using spin-trapping electron paramagnetic resonance (EPR). Chemosphere 2011, 84, 1095−1101. (53) Turchia, C. S.; Ollis, D. F. Photocatalytic degradation of organic water contaminants: Mechanisms involving hydroxyl radical attack. J. Catal. 1990, 122, 178−192. M

DOI: 10.1021/es505301x Environ. Sci. Technol. XXXX, XXX, XXX−XXX

Deciphering visible light photoreductive conversion of CO2 to formic acid and methanol using waste prepared material.

As gradual increases in atmospheric CO2 and depletion of fossil fuels have raised considerable public concern in recent decades, utilizing the unlimit...
7MB Sizes 0 Downloads 8 Views