Jun Cheng*,
Xiaoxu Xuan,
Xiao Yang,
Junhu Zhou and
Kefa Cen
State Key Laboratory of Clean Energy Utilization, Zhejiang University, Hangzhou 310027, China. E-mail: juncheng@zju.edu.cn; Fax: +86 571 87951616; Tel: +86 571 87952889
First published on 18th September 2018
To increase the reaction productivity and selectivity of the CO2 photoelectrochemical reduction reaction, a Cu (benzene 1,3,5-tricarboxylic acid [BTC])-reduced graphite oxide (rGO) catalyst was prepared by using a facile hydrothermal method and used in a CO2 photoelectrochemical cell (PEC) as a cathode catalyst. Characterization of the catalyst proved that successfully bonding of rGO to Cu(BTC) not only facilitated faster transfer of electrons on the surface of the catalyst but also created more active sites. CO2 photoelectrochemical reduction experimental results showed that the total carbon atom conversion rate was up to 3256 nmol h−1 cm−2 which was much higher than when pure Cu(BTC) was used as a cathode catalyst. The liquid product's selectivity to alcohols was up to 95% when −2 V voltage was applied to the system with Cu(BTC)-rGO used as the cathode catalyst.
Given the large specific surface area and extremely high electron transition rate of graphene,8,9 using graphene as a CO2 reduction catalyst is a topic of great interest in the research field.10 Fan Yu Jia et al.11 studied the role that graphene played in the CO2 and H2 hydrocarbonylation reaction and graphene obviously helped increase the absorption of H2 and CO2, as well as the CO2 conversion rate. Moreover, selectivity toward methanol was increased to 85%. However, high pressure should be applied to ensure the proceeding of the reaction and this requires good resistance of the reactor. Geioushy R. A. et al.12 synthesized a Cu2O/graphene catalyst and used it in the CO2 electrochemical reduction reaction. The product analysis result showed that this catalyst had good selectivity toward ethanol generation. However, the energy used in this system totally came from electricity; renewable energies such as solar energy were not used. Cheng Jun et al.13 brushed the mixture of Nafion and reduced graphite oxide (rGO) on a Ni foam and used it as the cathode catalyst to reduce CO2 to valuable chemicals. However, the major products of Nafion and rGO are formic and acetic acids, which cannot be used as fuel directly. Moreover, the selectivity toward liquid fuels such as ethanol is low. Cu foam, which has 3D structure, is a good cathode base in electrochemical reactions. Cheng Jun et al.14 used Cu foam as a support base and loaded graphene on the matrix to reduce CO2 in a photoelectrochemical cell. However, the selectivity toward a specific product is still low.
Metal organic frameworks (MOFs), as an emerging porous material, have been reported in many researches.15–21 Numerous researchers have reported its outstanding performance in gas separation and catalytic reduction or oxidation.22–25 Bohan Shan et al.26 synthesized a novel Co-based MOF to separate CO2/N2 and CO2/CH4. Moreover, the CO2 separation rates have reached 61.4% and 11.7%, respectively. However, this Co-based MOF in the gas separation process is only a physical adsorption and desorption process but not in a chemical reaction. R. Senthil Kumar et al.27 used an electrochemical method, which successfully synthesized a highly active Cu-based MOF material and used it as a catalyst to reduce CO2 with an electrolyte consisting of a dimethylformamide solution of tetrabutyl ammonium tetra fluoroborate saturated with CO2. The result showed that this catalyst is an effective catalyst to reduce CO2 to organics. However, the organic synthesized was mainly oxalic acid and liquid fuels like ethanol and methanol were not obtained. Guoqiang Song et al.28 fabricated a porous Cu-benzene 1,3,5-tricarboxylic acid (BTC) silica monolith heterogeneous catalyst for the selective oxidation of alkylbenzenes. However, the usage of Cu(BTC) in CO2 photoelectrochemical reduction reactions has never been reported.
In this paper, a facile hydrothermal method was implied to synthesize Cu(BTC)-rGO catalyst, characterizations of the catalyst proved faster electron transfer rate and existence of more reaction active sites on the surface. These features facilitated the improvement of productivity and selectivity of CO2 photoelectrochemical reduction reaction.
The Cu foam was deposited with Pt by using an electrochemical deposition method in a 0.1 g L−1 H2PtCl4 with an applied voltage of −0.2 V and deposition time of 180 s.
Then, the cathode was fabricated using the following method: 90 mg of as synthesized Cu(BTC)-rGO was mixed with the Nafion solution; afterward, the solution was placed on a Pt-deposited Cu foam using the drop–dry method.
The CO2 photoelectrochemical reduction reaction was conducted in an H-shaped double chamber reactor. Then, a Nafion 117 membrane (Shanghai Hesen Co., Ltd.) was used to separate the two chambers. During the reaction process, light was applied to the anode Pt-TNT, and voltage was applied between the anode and cathode. The photocurrent was detected with CHI660D electrochemical workstation (CH instruments Co., Ltd.), and sunlight was simulated using Perfectlight PLS-SXE300CUV xenon lamp.
The working area of cathode was 1 cm.2 The anolyte was 0.5 M H2SO4, while the catholyte was dimethylformamide (DMF). A water decomposition process happened on the anode Pt-TNT under light. The H+ generated by the oxidation of H2O under the catalytic effect of Pt-TNT crossed the Nafion membrane and took part in the CO2 reduction reaction on the cathode. At the same time, the photo-generated electrons went to the cathode through the external circuit. Afterward, CO2 and these electrons, as well as H+, reacted and produced organics under the catalytic effect of the cathode catalyst. The schematic of the photoelectrochemical cell for CO2 reduction is shown in Fig. 1. After the reaction, the liquid and gas products were collected and detected.
Fig. 2 Cu 2p XPS spectra of Cu(BTC)-rGO (a) and Cu(BTC) (b); C 1s XPS spectra of Cu(BTC)-rGO (c) and Cu(BTC) (d); O 1s XPS spectra of Cu(BTC)-rGO (e) and Cu(BTC) (f). |
The quantitative analysis can be conducted by fitting the XPS spectra. The element atom number ratio in the samples was calculated by using the following equation:
n(E1)/n(E2) = [A(E1)/S(E1)]/[A(E2)/S(E2)] | (1) |
The calculation result showed that in Cu(BTC)-rGO, the element atom number ratios of Cu, C, and O were 4.9%, 23.7%, and 71.5%, respectively. Moreover, in Cu(BTC), the element atom ratios of Cu, C, and O were 5.7%, 18.4%, and 75.8%, respectively. The C atom ratio in Cu(BTC)-rGO was much higher than that in Cu(BTC). Further mining of the C 1s XPS spectra showed that three characteristic peaks were observed at 284 eV, 286 eV, and 289 eV, which correspond to C–C/CC, C–O and OC–O respectively. The weight ratios of these three functional groups were calculated. In Cu(BTC)-rGO, the ratios were 81.8%, 7.3%, and 10.9%, respectively. Moreover, in Cu(BTC), the ratios were 57.6%, 22.2%, and 20.2%, respectively. Thus, the successfully bonding of rGO in Cu(BTC)-rGO was proved; rGO bonding in the sample can accelerate the transfer of electron on the surface of the catalyst, leading to easier adsorption of the intermediates such as CO2* in the CO2 reduction reaction.31,32
Successfully bonding of rGO into the Cu(BTC) crystal lattice in Cu(BTC)-rGO was also proven by the XRD pattern. As shown in Fig. 3, similar characteristic peaks in Cu(BTC)-rGO and Cu(BTC) were observed. However, the peaks in the Cu(BTC)-rGO XRD pattern shifted to the right due to rGO bonding. Both of the two samples consisted of a face-centered cubic crystal lattice of the Fmm space group and had typical peaks at 2θ = 9.5° (220), 13.5° (400), 14.7° (331), 16.5° (422), 17.5° (511), and 19.1° (440). This result is in agreement with that of the previous reports.33 In addition to the characteristic peaks of the Cu(BTC) crystal, several small peaks in the range of 35°–45° were observed, which were the characteristic peaks of CuO, Cu2O, and Cu. This result indicated the synthesis of Cu and CuxO in Cu(BTC)-rGO and Cu(BTC). According to a previous study,12 the mixture of Cu and CuxO is an efficient catalyst to selectively reduce CO2 to ethanol; thus, the presence of these particle can promote the reaction selectivity to ethanol.
The morphology of cathode catalyst was shown in Fig. 4. Accordingly, Cu foam evidently had a 3D porous structure that can minimize the diffusion resistance for mass transport. This property of Cu foam favored the efficient transfer of CO2 and fast emission of products, which would accelerate the reaction rate.
Fig. 4 SEM patterns of Cu(BTC)-rGO/Pt–Cu foam cathode ×150 (a), Cu(BTC)-rGO catalyst ×2500 (b), Cu(BTC) catalyst ×1100 (c), TEM pattern of Cu(BTC)-rGO catalyst (d). |
Fig. 4(b) observed evidently wrinkled morphology, which was one of the typical characteristic of rGO. Many octahedral shaped Cu(BTC) particles were distributed uniformly on the rGO surface. Fig. 4(c) showed that morphology of Cu(BTC). TEM pattern of Cu(BTC)-rGO catalyst was given in Fig. 4(d), many octahedral shaped Cu(BTC) can also be observed bonded to the surfaced of rGO.
According to BET measurement results, the specific surface area of Cu(BTC)-rGO and Cu(BTC) were 938 m2 g−1 and 1322 m2 g−1 respectively. Though rGO bonding lead to the decrease of specific surface area of the catalyst, it successfully changed the pore structure of the catalysts. Fig. 5 showed the nitrogen sorption isotherms of Cu(BTC)-rGO (a) and Cu(BTC) (b). Accordingly, Cu(BTC)-rGO nitrogen adsorption plot showed a small hysteresis between adsorption and desorption branches suggesting the existence of few mesopores in the catalyst. However, Cu(BTC) nitrogen adsorption plot showed there only existed micropores in Cu(BTC). Pore size distribution [Fig. 5(c) and (d)] further confirmed this conclusion. The existence of mesopores better connected the micropores thus improving mass transfer rate.
Fig. 5 Nitrogen sorption isotherms: Cu(BTC)-rGO (a), Cu(BTC) (b); pore size distribution: Cu(BTC)-rGO (c), Cu(BTC) (d). |
Pt nanoparticles were supposed to be observed on the Cu foam. However, as the SEM pattern shows, Pt particles on the Cu foam matrix were hard to be observed mainly due to the small size. To identify the Pt content on the cathode, EDX analysis was conducted. EDX analysis showed that the Pt content on the cathode was about 0.56 wt%. The existence of Pt nanoparticles also contributed to facilitate charge transfer and intermediates adsorption on the surface of the catalyst.34,35
Fig. 6 showed the current densities of the system when Cu(BTC)-rGO was used as catalyst (a), and Cu(BTC) was used as catalyst (b). The distance between TiO2 anode and light resource was 2 cm and the average incident light intensity provided at the electrode surface was measured to be 10 mW cm−2. Accordingly, the average current density under CO2 purging condition with Cu(BTC)-rGO used as catalyst was 2.13 mA cm−2. However, the average current density decreased to 1.75 mA cm−2 with Cu(BTC). This is a strong evidence to prove the improving of electron transfer rate on the surface when rGO is bonded to Cu(BTC). Moreover, when CO2 was bubbled into the solution, the current density showed sharp increase compared to that when N2 was bubbled.
Fig. 6 Electric current densities with different cathodes used: Cu(BTC)-rGO/Pt–Cu foam cathode (a), Cu(BTC)/Pt–Cu foam cathode (b). |
The J–V curves of Cu(BTC)-rGO catalyst and Cu(BTC) catalyst obtained from CO2 electrochemical reduction in DMF (three electrode configuration, without photoanode) were shown in Fig. 7 and clearly, the current density of the system using Cu(BTC)-rGO catalyst was much higher than that when Cu(BTC) was used as catalyst.
Fig. 7 J–V curves of Cu(BTC)-rGO catalyst and Cu(BTC) catalyst obtained from CO2 electrochemical reduction in DMF (three electrode configuration, without photoanode). |
The results of the CO2 photoelectrochemical reduction experiment were shown in Fig. 8 Accordingly, the Cu(BTC)-rGO loaded on the Pt-deposited Cu foam cathode showed the best catalytic effect to reduce CO2 when the applied voltage on the system was −2 V. Moreover, the total carbon atom conversion rate reached 3255.87 nmol h−1 cm−2, and the liquid products selectivity towards alcohols was up to 95%. The Faraday efficiency was measured to be 34.1% with −2 V voltage applied.
Fig. 8 Carbon atom conversion rates and Faraday efficiencies with various applied voltage (−1.4 V, −1.7 V, −2 V, −2.3 V, −2.6 V) using Cu(BTC)-rGO/Pt–Cu foam cathode. |
The catalytic effect of Cu(BTC) under the condition with −2 V voltage applied was also studied. The result showed that the total carbon atom conversion rate was 2561.76 nmol h−1 cm−2, which was much lower than that when Cu(BTC)-rGO was used as the cathode catalyst. This result mainly due to the reason that lacking of rGO in the material will lead to lower transfer rate of electrons on the surface of the catalyst, resulting in the decrease of the carbon atom conversion rate.
This journal is © The Royal Society of Chemistry 2018 |