Improving artificial photosynthesis in carbon nitride by gas-liquid-solid Interface management for full light-induced CO2 reduction to C1-C2 fuels and O2.

Abstract The activity and selectivity of simple photocatalysts for CO2 reduction remain limited by the insufficient photophysics of the catalysts, as well as the low solubility and slow mass transport of gas molecules in/through aqueous solution. In this study, these limitations are overcome by constructing a triphasic photocatalytic system, in which polymeric carbon nitride (CN) is immobilized onto a hydrophobic substrate, and the photocatalytic reduction reaction occurs at a gas–liquid–solid (CO2–water–catalyst) triple interface. CN anchored onto the surface of a hydrophobic substrate exhibits an approximately 7.2‐fold enhancement in total CO2 conversion, with a rate of 415.50 μmol m−2 h−1 under simulated solar light irradiation. This value corresponds to an overall photosynthetic efficiency for full water–CO2 conversion of 0.33 %, which is very close to biological systems. A remarkable enhancement of direct C2 hydrocarbon production and a high CO2 conversion selectivity of 97.7 % are observed. Going from water oxidation to phosphate oxidation, the quantum yield is increased to 1.28 %.

The activity and selectivity of simple photocatalystsf or CO 2 reductionr emain limited by the insufficient photophysics of the catalysts, as well as the low solubility and slow mass transport of gas molecules in/through aqueous solution.I nt his study, these limitations are overcome by constructingatriphasic photocatalytic system, in which polymeric carbon nitride (CN) is immobilized onto ah ydrophobic substrate, andt he photocatalytic reduction reaction occurs at ag as-liquid-solid (CO 2water-catalyst) triple interface. CN anchored onto the surface of ah ydrophobic substrate exhibits an approximately 7.2-fold enhancement in total CO 2 conversion, with ar ate of 415.50 mmol m À2 h À1 under simulated solarlight irradiation. This value corresponds to an overall photosynthetic efficiency for full water-CO 2 conversion of 0.33 %, which is very close to biological systems. Aremarkable enhancement of direct C2 hydrocarbon production and ah igh CO 2 conversion selectivity of 97.7 %a re observed. Going from water oxidation to phosphate oxidation, the quantum yield is increased to 1.28 %.
Ever-increasing consumptionoffossil fuels along with the massive emission of carbon dioxide (CO 2 )h as generated an energy crisis and resulted in climate change. [1] Artificial photosynthesis throughp hotocatalytic CO 2 conversion into valuable chemicals (e.g.,C Oo rH 2 ,a nd, preferably,C H 4 ,C 2 H 4 ,e tc.) in the presence of H 2 Oh as been recognized as ap otentially promisingw ay to resolve these issues. [2] The transfer of photogenerated charge carriers and mass transportp lay crucial roles in determining the kinetics of catalysts and CO 2 photoreduction efficiency. [3] Meanwhile, the competitive reactiono fp hotocatalytic hydrogen evolution also diminishes the generation of hydrocarbons, resultingi nl ow selectivity and activityo fC O 2 reduction of most currents ystems. To overcome kinetic limitations and suppress the hydrogen evolution reaction, numerous efforts have focusedo nt he improvement of pristine photocatalysts, by methods such as loading cocatalysts, [4] tailoring morphologies, [5] adjusting defectd ensities, [6] and constructing heterojunctions. [7] At the same time, the reaction interfacet hat governs the solid-liquid contact and mass transfer is also of vital importance to the photocatalytic CO 2 conversion process. Previous studies found that the availability of excess protons (H + ) and low concentrationo fC O 2 at the reaction interfacel ead to unsatisfactory activity ands electivity of the photocatalytic CO 2 reduction system. [8] In ac onventional liquid-solid diphase system for CO 2 photoreduction, the availability of CO 2 at the reactioni nterface is dependent on its mass transfer through the water phase. [9] The low concentration and slow diffusion rate of CO 2 molecules in water thereby strongly hindert he surface catalytic process of CO 2 photoreduction.
In this study,t oo vercome the limitations of the conventional liquid-solid diphase system of CO 2 photoreduction, as imple and sustainable approach is developed by constructing at riphase (gas-liquid-solid) interfacial photocatalytic system. The photocatalysts are immobilized on the surfaceo facarbon fiber substrate ( Figure 1). The concentration of CO 2 molecules at the interface can be controlled by adjustingt he surfacea dsorptiono nt he substrate. Particularly,ahydrophobic substrate surfacep romotes CO 2 localization from the gas phase and helps to rapidly deliver CO 2 molecules to the contact area of gas (CO 2 ), liquid (water), and solid (catalyst). Such ar eaction systemt hen allows the continuous delivery of CO 2 molecules from the gas phase to the reactioni nterface via its hydrophobic channels, instead of the slow diffusion through the liquid phase. As ar esult, the accessibility of CO 2 molecules to the photocatalyst is greatly increased, which subsequently enhances the rate of the reaction between CO 2 and photogenerated electrons, thereby diminishing electron-hole recombination and increasing charge utilization.F inally,t he activitya nd selectivity of photocatalytic CO 2 conversion is remarkably improved.
As ap roofo fc oncept, we immobilizedc arbon nitride (CN) nanosheets onto the surfaceo facarbon fiber (CF) fleece with porouss tructure and different wettability.T he contact angles (CAs) of the superhydrophilic, hydrophilic, and hydrophobic CF substrates are approximately 08,3 7.58,a nd 1128,r espectively ( Figure 2A-C). After CN immobilization on one side of the substrate, the side with the CN layer becomes superhydrophilic or hydrophilic with CAs of approximately 08,0 8 and 108 (see the Supporting Information, FigureS1; the correspondings amples are denoted as CN/CF1, CN/CF2, and CN/ CF3, respectively). In this case, water can wet the hydrophilic photocatalyst layer whileg as-phase CO 2 is directed through the hydrophobic substrate up to the CN particles, resultingi n the formation of ag as-liquid-solid triphase boundary zone. Such af ramework then secures the supply of both abundant water and CO 2 molecules to driveo verall CO 2 photoreduction. Field-emission scanning electronm icroscopy( FESEM) allows direct observation of the interface between immobilizedC N photocatalysts and carbon fiber substrates ( Figure 2D,E).
The powder X-ray diffraction (XRD) patterns of CF,C N/CF1, CN/CF2,C N/CF3, and CN are shown in Figure 3A.T he pattern of CF exhibits ab road peak at around 26.58,i ndexed as the (002) plane of ag raphitic carbons tructure, the fleece. [10] Two distinct diffraction peaks appeared at 13.18 and 27.38,c orresponding to the (100)a nd (002) planes of CN, respectively.T he former corresponds to the repeateds tructuralp acking of tri-striazine heterocycles in the conjugated planes, and the latter can be ascribed to the regular graphite-like interlayers tacking. [11] After the immobilization of CN onto the surfaceo ft he carbon fiber substrate, the XRD patterns of CN/CF show the characteristic peaks of both CN and CF.T he peak intensity of CF in CN/CF samples gradually decreasesf rom CN/CF1 to CN/ CF3. This is due to the differentsurface energy of CF with varying surface wettability,r esulting in differentf ilm thicknesses of the CN layer, [12] although the primary loading of CN is identical. The optical absorption properties of all samples were then measured by UV/Vis diffuse reflectance spectroscopy( DRS; Figure 3B). For CN, the absorption edge at 450 nm corresponds to its intrinsic band gap of 2.76 eV.C N/CF1, CN/CF2, and CN/ CF3 show similar absorptione dges but enhanced light-absorption intensity in visible-light regions, owing to the strong broad absorption of CF. [13] Particularly,t he absorptioni nt his regiono fh ydrophilic CF-supported CN coated is stronger than that of hydrophobic CF-supported CN, because of the different film thickness of photocatalyst layer,c onsistent with the XRD results. The differentt hicknesses of CN films affected the scattering of light among the texture and pore structure in CF substrates, which led to differences in the light absorption in the visible-light region over CN/CF samples. FT-IR spectroscopy was used to further investigate the surfaces tructure of CN and CN/CFs amples ( Figure 3C). The broad peaks between 3000 and 3500 cm À1 can be ascribed to the adsorbed hydroxy groups and the amino groups in CN, whereas the peaks at around8 03, 1211, 1402, 1531, and1 639 cm À1 are the typical stretching vibrations of the s-triazine ring system, C=Na nd C-N heterocycles, respectively. [14] All CN/CF samples show the same  characteristic peaks as those of CN, suggesting that the coating of CN ontot he surfaceo fC Fh as no obvious effect on the structure of CN, and that CF only serves ap latform for the immobilization of CN. In order to reveal the effect of surface wettability of CF on the charges eparation efficiency for CN, photoluminescence (PL) spectra werem easured ( Figure 3D). The intensity of all PL spectra is similar, indicating that the charge recombination rate in the samples of CN immobilized on CF with different surface wettability is similar.N amely,t he expected performance difference of photocatalytic CO 2 reduction could not be relatedt ov ariation in the charge transfer dynamics within the series of catalysts.
The photocatalytic activity and selectivity of asprepared CN/CF photocatalysts were investigated under simulated sunlight irradiation in the absence of any sacrificial agent and co-catalyst. The novel reaction environment of gas-liquid-solid (G-L-S) triphase reactioni nterface wase mployed ( Figure S2). As ar esult,as pectrum of product molecules is found ( Figure 4A and Ta ble S1);t he main products of CO 2 reduction are CH 4 ,C O, C 2 H 4 and H 2 over CN/ CF1, CN/CF2, and CN/CF3. On increasing the hydrophobicity of the CF substrate, the generation of hydrocarbon fuels from CO 2 reduction reactionm assively increases, whereas that of H 2 from water splitting is suppressed. In particular, CN/CF3s hows the best performance with at otal CO 2 conversion of 415.50 mmol m À2 h À1 ,a nd ac orresponding quantum yield of 0.33 %, which is highert han or comparable to reported results. [15] This is of the order of natural photosynthesis, albeit hered escribed with am uch simpler synthetic system,f ree of further cocatalysts to increaset he rate of hydrocarbon formation, as well as oxygen liberation. The CO 2 conversion selectivity is as high as 97.7 %, as compared to that of H 2 evolution. that is, 97.7 %o fa ll electrons end up in carbon products. The rate of generation of C2 hydrocarbon product (C 2 H 4 ), as well as that of CH 4 , over the hydrophobic substrate is significantly higher than that over the hydrophilic substrate. It can be concluded that CN immobilizedo nto the hydrophobic substrate experiences ac ontinuous supply of CO 2 molecules, thus making CO 2 reduction processes much more effective than H 2 generation under the given conditions of no metal cocatalyst. Moreover,i ncreasing the loading amount of CN would further enhance the photocatalytica ctivity while maintain the high selectivity (see data of CN/CF4i nT ableS1).
Furthermore,t he good photocatalytic stability of CN/CF3 was demonstrated by ac yclingt est ( Figure 4B). The XRD pattern and contact angle of CN/CF3w ere measured after the cycling test ands howed no obvious change ( Figure S3). To understandt he unique interfacial effect, ac ontrolled experiment was conducted in which the hydrophobic substratei mmobilizing with CN (CN/CF3) was completely immersed in the water, which is similar with the conventional liquid-solid diphasic system ( Figure S4). In this case, there was no direct contact between the substrate and CO 2 atmosphere.H ere, the necessary CO 2 could only be supplied from the liquid phase. This test was conducted under the same reactionc onditions as the triphase system.T he main products of CO 2 reduction in this liquid-solid diphase system are CH 4 ,C O, C 2 H 4 and H 2 ,r espectively.T he total CO 2 conversion reached 265.16 mmol m À2 h À1 ( Figure 4C). Indeed,b yc omparing the two experiments, we find the CO 2 conversion ratei nt he triphase system to be twice that in the diphase system.
In addition, the production of O 2 from water oxidation is the sole reaction to consume the photogenerated holes. Thus, we also monitored O 2 evolution, which occurred at ar ate of 711.95 mmol m À2 h À1 ( Figure S5) by CO 2 photoreduction over Figure 4. A) Photocatalytic activityo fC N/CF1, CN/CF2,and CN/CF3i nthe triphase system with CO 2 atmosphereconnection.B )Cyclingt est of photocatalytic CO 2 reduction over CN/CF3int he triphase system. C) Photocatalytic activity of CN/CF3i nthe diphase system completely immersedint he water.D,E)GC-MS analysis of reactionp roducts CH 4 (D) and CO (E) over CN/CF3i nthe triphase systema fter irradiation for several hours with 12 Ca nd 13 Casc arbon sources.F ,G)GC-MS analysis of reactionp roducts 13 C 2 H 4 (F) and 13 C 2 D 4 (G) overC N/CF3int he triphase system after irradiationf or severalh ours with 13 CO 2 as carbon sourcea nd D 2 Oand H 2 Oash ydrogen sources. ChemSusChem 2020ChemSusChem , 13,1730ChemSusChem -1734 www.chemsuschem.org 2020 The Authors. Publishedb yWiley-VCH Verlag GmbH &Co. KGaA, Weinheim CN/CF3 in the triphase system, which is higher than the stoichiometric yield calculated from the given product distribution (644.34 mmol m À2 h À1 ). This deviation is reasonable within experimental error and might be relatedt ou ndetectable species, such as methanol and ethanol, which end up dissolved in the water phase.
To confirm the carbon source of the photocatalytic products, isotope-labeled CO 2 was employed, and the hydrocarbon products carrying the isotopes 13 C, 12 C, and 2 H( D) were detected by GC-MS for the triphase system ( Figure 4D-G and Figure S6). The products labeled by 12 Ca ppear at earlier retention times than those labeled by 13 C( Figure S6A). Clearly,t he signals m/ z = 16.1 ( 12 CH 4 )a nd m/z = 17.1 ( 13 CH 4 )w ere dominant in the GC-MS spectra forp hotocatalytic 12 C-labeled CO 2 and 13 C-labeled CO 2 reduction,r espectively ( Figure 4D). The signal at m/ z = 28.1 can be attributed to 12 CO and 12 C 2 H 4 ,a nd that at m/ z = 29.1 can be attributed to 13 CO when using 12 CO 2 and 13 CO 2 , respectively ( Figure 4E). Subsequently, the isotope-labeling experiments with carbon and hydrogen sourcesl abeled with 13 C and 2 H( D) were performed for photocatalytic CO 2 reduction over CN/CF3( Figure S6B). The signals for molecular ethylene are at m/z = 30.1 and m/z = 34.1 when 13 CO 2 reacts with H 2 O and D 2 O, respectively.T he signals at m/z = 29.1 and m/z = 32.1 are stronger those at m/z = 30.1 and m/z = 34.1, owing to the higher stability of the molecular ions ( Figure 4F,G). [16] In conclusion,t hese resultss trongly indicate that the photocatalytic products originate solelyf rom CO 2 reduction.
In light of the above analysis,i ti sr easonable to propose a model to analyze the excellent CO 2 photoreduction performance of CN/CF samples in the triphase reaction system ( Figure 5A,B). When the hydrophilic substrate was used to anchort he CN catalyst, at inyp roportion of CO 2 molecules is supplied directly from the gas phase, and the majority is supplied from liquid phase to participate in photocatalytic CO 2 reduction. Thus, the CO 2 consumption and supply is unbalanced, resultingi nalower CO 2 conversion rate. In contrast, when the hydrophobic substrate was used to anchort he CN catalyst, the main part of CO 2 molecules is supplied directly from the gas phase with ah igh transport rate, thus ac onstant and higher interfacial CO 2 concentration is maintained. As ar esult,t he CO 2 conversion rate and selectivity,a sw ell as the amount of C2 molecules, are significantly higher over hydrophobic substrates than over hydrophilic substrates. To further demonstrate the significance of the continuous gas access, ad iphasic system was analyzed as ar eference, with the substrate immobilized with CN completely immersed in water ( Figure 5C). When the trapped CO 2 molecules were isolated by the liquid phase, the diphase system disabledt he continuous supply of CO 2 .T hus, the decreased concentration of interfacial CO 2 resulted in lower conversion efficiency.
There are stillan umber of means to improvet he activity that were not explored in these first-generation experiments. We still consider the rate-determining step to be the four-electron oxidationo fw ater to dioxygen, which here proceedsi n the absence of acocatalyst. Akinetically much simpler reaction than O 2 formation is the direct oxidation of phosphates to perphosphates. [14a, 17] It was excitingt oo bserve that CN/CF3 showed much highera ctivity when Na 3 PO 4 was added to the water phase. In the triphase system,aC O 2 conversion of 1413.85 mmol m À2 h À1 with aC O 2 conversions electivityo f 95.5 %( Figure S7 and Ta bleS2) and aq uantum yield of 1.28 % were achieved. In this case, the synthetic photosynthesis even outperformed biological photosynthesis, in spite of its simplicity.P erphosphates therefore representavaluable route of investigation,b ut also can be thermally or catalytically decomposed to liberate dioxygent ot erminate product formation. [17,18] In summary,atriphase interfacial photocatalytic CO 2 reduction system based on the gas-liquid-solid reaction interface allows efficient andc ontinuous delivery of CO 2 molecules to the catalysts urface and inhibits the hydrogen evolution reaction. The photogenerated chargecarriers are efficiently utilized, resultingi ns ignificantly enhanced activity and selectivity in the photocatalytic CO 2 reduction. In particular, the CN anchored onto the surfaceo fahydrophobic substrate (CN/CF3) exhibits about 7.2-fold enhancement in the total CO 2 conversion with 415.50 mmol m À2 h À1 as compared to the CN anchored onto the surface of superhydrophilic substrate (CN/CF1). This is accompanied by aC O 2 conversion selectivity of 97.7 %, the remainder being the otherwise dominant H 2 evolution. This product yield for photocatalytic CO 2 reduction is also clearly superior to that with the conventional diphase system, and could be furthere nhanced by simplifying the photooxidation process from four-electron dioxygen generation to perphosphate formation, with aC O 2 conversion of 1413.85 mmol m À2 h À1 and aq uantum yield of 1.28 %. Interestingly,ahigher CO 2 -based reactant flux also improved C 2 hydrocarbon production, which reflects the higherc hance of intermediate C 1 speciesr ecombining at the catalyst surface. This work provides ap latform to explore furtheri nterfaciala rchitectures in system engineering of highly actives emiconductor photocatalysts.