摘要：制備可以同時高效且高通量地處理納米乳液的超浸潤材料仍然具有挑戰(zhàn)。為此，本文提出了一種通過在超分子骨架納米片上修飾氧化石墨烯以增強親水性的策略。通過將兩種具有片狀形態(tài)的材料連續(xù)抽濾于商業(yè)基質(zhì)上，可制備得到氧化石墨烯輔助的超分子骨架復(fù)合膜，并用于分離具有納米尺寸液滴的水包油乳液。骨架一方面通過均勻的納米孔攔截乳液中分散的微小液滴，另一方面也通過帶負(fù)電的表面提供靜電相互作用來驅(qū)動破乳過程發(fā)生。具有良好親水性的氧化石墨烯賦予膜材料改善的親水能力和水合層。該復(fù)合膜具有納米級的截留尺寸、帶負(fù)電的表面和水下疏油性，并且還獲得了高的水通量和耐油污染性。基于尺寸篩分和破乳效應(yīng)，該復(fù)合膜可有效地去除分散在水包油乳液中由非離子、陰離子和陽離子表面活性劑穩(wěn)定的納米油滴。特別是對于離子型乳液，在分離后動態(tài)光散射未檢測出殘留液滴。濾液中總有機(jī)碳含量小于10 ppm，對應(yīng)著大于99.9%的分離效率，優(yōu)于許多國家和組織的標(biāo)準(zhǔn)。在各種乳液的分離過程中，復(fù)合膜表現(xiàn)出較高的分離滲透性，約為原始骨架膜的3.5倍。此外，具有防污效果的復(fù)合膜獲得了較高的通量回收率，通過簡單的水洗處理即可實現(xiàn)5次具有穩(wěn)定分離性能的循環(huán)。該復(fù)合膜在重復(fù)使用過程中沒有組分損失，在150 °C內(nèi)具有熱穩(wěn)定性，并能抵抗腐蝕性化學(xué)環(huán)境。在本工作中，我們試圖將具有不同結(jié)構(gòu)特性和表面特性的兩種組分結(jié)合，通過簡單的方法制備復(fù)合膜，并在功能協(xié)同作用下實現(xiàn)水包油型納米乳液的高性能分離。

關(guān)鍵詞：超分子骨架；氧化石墨烯復(fù)合膜；破乳；納米乳液；高效分離

中圖分類號：O648

GO-Assisted Supramolecular Framework Membrane for High-Performance Separation of Nanosized Oil-in-Water Emulsions

Abstract： Intercepting tiny droplets in nano-emulsions is crucial for thedevelopment of membrane materials with pore diameters smaller than the dropletsize， as per the size screening mechanism. While this method achieves highseparation efficiency， it results in a decrease in separation flux. On the one hand， theuse of macro-porous materials can increase the flux， but it does not guarantee highefficiency on the other hand. Fabricating superwetting materials that exhibit both highefficiency and flux in separating nanosized emulsions provides opportunities forovercoming the bottleneck yet how to extend the applicable range with high efficiencyremains a challenge. To address this issue， we propose a strategy to enhance thehydrophilicity of supramolecular framework nanosheets by modifying hydrophilicgraphene oxide （GO）. By incorporating GO into the supramolecular framework （SF）composite membrane through a sequential pumping process onto a commercial matrix， we create a GO-assisted SFcomposite membrane capable of separating oil-in-water （O/W） emulsions containing nanosized droplets. The frameworkintercepts the dispersed tiny droplets in the emulsions through uniform nanoscale pores while also facilitating thedemulsification process through electrostatic interaction on its negatively charged surface. The hydrophilic GO modificationon the composite membrane enhances its water affinity and promotes the formation of a hydrated layer on the membranesurface. The resulting composite membrane exhibits a nanoscale cut-off size， a negatively charged surface， andoleophobicity under water. Importantly， it achieves high water flux and resistance to oil contamination. By leveraging thesize screening and demulsification effects， the composite membrane efficiently removes nanosized oil droplets dispersedin O/W emulsions stabilized by non-ionic， anionic， and cationic surfactants. Particularly for emulsions containing ionicsurfactants， no residual droplets are detected through dynamic light scattering （DLS） after separation. The filtrate exhibitsa total organic carbon （TOC） content of less than 10 ppm， corresponding to a separation efficiency greater than 99.9%，which surpasses the standards of many countries and organizations. Furthermore， compared to the original SF membrane，the composite membrane demonstrates approximately 3.5 times higher separation permeation during the separationprocess of various emulsions. Additionally， the composite membrane exhibits an anti-fouling effect and achieves a highflux recovery rate， ensuring stable separation performance for 5 cycles through simple water washing treatment. Thecomposite membrane retains its components throughout repeated use， exhibits thermal stability up to 150 °C， and canwithstand corrosive chemical environments， including 1 mol·L?1 HCl， 0.01 mol·L?1 NaOH， and 1 mol·L?1 NaCl. In this study，we realize the combination of two components with distinct structural and surface characteristics to fabricate a compositemembrane through a simple method and achieve high-performance separation of nanosized O/W emulsions throughsynergistic functionality.

Key Words： Supramolecular framework; Graphene oxide composite membrane; Demulsification;Nanosized emulsion; High-performance separation

1 Introduction

Emulsified oily water separation is essential for high-puritywater preparation and wastewater treatment in delicateelectronic devices， bioindustry， and sewage disposal 1–4. Withthe extensive application of nanosized emulsions inpharmaceutical， food， and personal care products， the treatmentof nano-emulsions contained in industrial emissions has receivedincreasing attention 5–7. Compared with conventional emulsiontreatment methods， such as flotation 8， flocculation 9， andadsorption 10， the superwetting membrane separation has beenstudied by numerous researchers due to its simple process， lowenergy consumption， no secondary pollution， and otheradvantages 11–13， and some works have focused on the treatmentof nanosized emulsion 14–16. However， for intercepting theminute droplets in the emulsion， the membrane pore diameterneeds to be decreased according to the size screeningmechanism， which brings high separation efficiency butsacrifices the flux 17，18. Using membrane materials with largepores to improve the flux does not guarantee high efficiency.Therefore， the trade-off effect makes it challenging to realizehigh-performance nano-emulsion separation with simultaneouslyhigh separation efficiency and permeability.

The demulsification material breaks the emulsion by inducingthe dispersed droplets to coalesce under the provided force， suchas physical deformation 19， hydrophilic effect 20， light effect 21，and electrostatic interaction 22. During the demulsificationprocess， the membrane traps the oil drops and breaks thesurfactant layer. Then， minute oil droplets merge under the demulsification force from the media. At last， the obtained largeroil droplets leave the medium， and demulsification occurs 23.There are some advantages of introducing demulsificationseparation. Moderate pores separate tiny droplets， ensuring highefficiency and flux simultaneously. And the coalescence of oildroplets prevents the formation of the cake layer， easing thefouling and benefiting from high permeability 24–26. However， itis difficult to demulsify emulsions containing small droplets，especially droplets with diameters less than several nanometers.What’s more， to realize the demulsification of nano-emulsionsunder various conditions， both chemical micro-environmentcontaining demulsification sites and supplementary means arerequired. At present， the reported demulsification materialsmainly realize demulsification through a single function andonly meet a limited application scope. Most of them are onlyused to deal with micron-sized emulsions， resulting in lowdemulsification efficiency 27，28.

As one of the porous assemblies， the SF shows uniformnanoscale pores and high porosity 29. Constructed by dynamicnon-covalent interactions 30， SF with high structural flexibilitymakes it available for further processing 31. And designablebuilding blocks provide the framework for various features onsurface wettability and electrical property， which extends thefunctional application of SF. By introducing polyoxometalate（POM） into the building units 32， some SFs realize thecombination of organic and inorganic components 31，33–35. Thecoexistence of hydrophilic and hydrophobic sites enriches thewettability of the framework but also improves the compatibility and processability of the material by alternating the rigid andflexible parts 36，37. Furthermore， the POM in some SF remainsmodifiable， providing a platform for further functionalapplications. Importantly， the introduction of covalentlymodified POM in the connection part of the frameworkpreserves the multiple electrostatic sites of the polyanion cluster，so the resulting assembly surface shows negative charges 38.

In this context， the SF with covalently modified POM ischosen for its suitable properties， in which the uniform nanoscalepores and high porosity assure the cut-off of the tiny droplet， andthe negatively charged surface helps demulsification under theelectrostatic interaction. Although this water-permeable SFmembrane is accessible for aqueous phase separation， the waterflux and contamination resistance still require to be improved forhigh-performance emulsion treatment. For membrane separationtechnology， modification by hydrophilic material to enhancewater affinity is a popular way to better improve permeation andthe anti-fouling property. We herein propose a strategy fordecorating the hydrophilic GO onto the SF nanosheets to get anapplicable membrane with separation capability for nanosizedemulsions. The SF-GO composite membrane is preparedfollowing the stepwise filtration of SF dispersion and GOdispersion on the commercial porous substratepolytetrafluoroethylene （PTFE） （Scheme 1a）. The obtainedmembrane exhibits nanosized pores， a negatively chargedsurface， enhanced hydrophilicity， oleophobicity under water，and oil resistance. With these characteristics， the SF-GOcomposite membrane can be utilized in separating the O/Wemulsions containing nanosized droplets （Scheme 1b）. The highseparation efficiency is achieved with the low total organiccarbon （TOC） content in the filtrate. On the premise ofintercepting the nanoscale droplets， the separation process isperformed with a pretty high flux under pressure. And thecomposite membrane with an anti-fouling effect shows a highflux recovery rate （FRR）， resulting in the maintained separationperformance for recycled multiple times.

2 Experimental and computational section

2.1 Materials

General chemicals， N，N-diisopropylethylamine （AR），trifluoromethanesulfonic acid （AR）， 1，2-dibromoethane （AR），1，4-dimethoxybenzene （AR）， paraformaldehyde （AR）， 5-bromopentanenitrile （AR）， phthalimide （AR）， and boron fluorideethyl ether （AR） are the products of Jamp;K Scientific Co.， Ltd. GO（AR） was purchased from Jiangsu XFNANO Materials Tech.Co.， Ltd. Tween 80， sodium dodecyl sulfate （SDS， AR），hexadecyl trimethyl ammonium bromide （CTAB， AR）， 4-methoxyphenol （AR）， 4-cyanophenol （AR）， NaOH （AR）， NaCl（AR）， HCl （AR）， and solvents （AR） were purchased fromSinopharm Chemical Reagent Co.， Ltd. All compounds wereused without further purification. Doubly distilled water wasused in the experiments. Silica gel （300–400 mesh） was appliedfor column chromatography. Preparation of host doublypillar[5]arene-grafted cluster MnMo6 （PCT）， guest moleculetripentanenitrile-triazine （TT）， and SF followed procedures inthe literature 38， and detailed synthesis processes are presented inSupporting Information， S1 （Figs. S1–S8）. Si nanoparticles（SiNPs） were prepared according to a one-step approach withslight modifications 39. The zeta potential of SiNPs is +0.7 mVin water （pH = 7）.

2.2 Preparation of GO-assisted SF membrane

The SF dispersion （1 mg?mL?1） was prepared by sonicatingthe above obtained SF in the mixed solvent of dimethyl sulfoxide（DMSO） and water （1 ： 4， in v/v） for 30 min. The GO dispersion（5 μg?mL?1） was prepared under the sonication for 30 min. Aseries of GO-assisted SF membranes were prepared throughstepwise filtration of SF dispersion and GO dispersion with thecommercial porous substrate PTFE （220 nm， pore size） as thesupport. When the diameter of the membrane was 15 mm， 3 mLof SF dispersion was added， and the content of GO was tuned bycontrolling the volumes of GO dispersion. When GO masseswere 5， 10， 15， and 20 μg， the membranes were named SF-GO5，SF-GO10， SF-GO15， and SF-GO20. Membranes prepared withsole SF or GO dispersion served as controls and were named SFand GO membranes.

2.3 Emulsion separation experiments

The surfactant-free O/W and corrosive O/W emulsions wereprepared by mixing oil （hexane） with pure water or corrosiveaqueous solution （1 mol?L?1 HCl， 0.01 mol?L?1 NaOH， and 1mol?L?1 NaCl） （1 ： 100， in v/v） and sonicating under a power of100 W for 20 min， and were named H/W， H/HCl， H/NaOH， andH/NaCl， respectively. The surfactant-stabilized O/W emulsions were prepared by mixing hexane and water （1 ： 100， in v/v） withthe addition of 0.02 mg?mL?1 surfactant （Tween 80， SDS， andCTAB）. Then， the emulsions were intensively stirred for 4 h at800 r?min?1 and named Tween H/W， SDS H/W， and CTAB H/W.

Tests of the pure water flux and emulsion separationperformance were applied by a filtration cell with the membranesandwiched in the middle. The contact area of the emulsion andmembrane is 1.8 cm2， and the separation was carried out under apressure of ?0.8 bar. The separation efficiency R （%） wasobtained by the following equation 40：

R = （1 ? Cf/Ce） × 100% （1）

where Cf and Ce are the oil concentrations in the filtrate and emulsion， respectively. The data were collected from the average of three parallel experiments.

The flux J （L?m?2?h?1?bar?1） of pure water and filtrate was calculated by following the equation 17：

J = V/（t × A × P） （2）

where V is the filtrate volume （L）， A is the effective filtration membrane area （m2）， t is the separation time （h）， and P is the pressure value （bar）. The data were collected from an average of five parallel experiments.

The anti-fouling capacity of the membrane was evaluated by measuring the FRR by following the equation 40：

FRR = JR/JW × 100% （3）

where JW is the flux of pure water， and JR is the recovered water flux after the separation test with water cleaning.

2.4 Characterization

1H NMR spectra were recorded on a Bruker （Germany）Avance 500 MHz spectrometer using tetramethylsilane as theinternal reference. Matrix-assisted laser desorption ionizationtime-of-flight （MALDI-TOF） mass spectra were recorded on aBruker （Germany） AutoflexTM speed TOF/TOF massspectrometer equipped with a nitrogen laser （337 nm， 3-nspulse）. Elemental analysis （EA） was implemented on anElementar （Germany） Vario microcube. Fourier transforminfrared （FT-IR） spectra （KBr pellet） were collected on a Bruker（Germany） Vertex 80 V spectrometer equipped with a DTGSdetector （32 scans） at a resolution of 4 cm?1. X-ray diffraction（XRD） data were recorded on a Rigaku （Japan） SmartLab 3 Xraydiffractometer using Cu Kα1 radiation at a wavelength of 1.54?. The N2 sorption experiment was carried out on Quantachrome（U.S.A.） instruments ASiQwin. Bruker （Germany） DimensionFastScanTM atomic force microscope （AFM） was used to takeimages under ambient conditions. Transmission electronmicroscopic （TEM） images were obtained on a field emissionelectron microscope （JEOL JEM-2100F， Japan） with anaccelerating voltage of 200 kV without staining. Scanningelectron microscopic （SEM） measurement was performed on aJEOL （Japan） JSM-6700F field emission scanning electronmicroscope. Contact angles （CAs） were measured on an OCA20contact angle measurement system （Dataphysics， Germany），using pure water droplets （3 μL） for water contact angle （WCA）and isooctane droplets （3 μL） for oil contact angle （OCA）. The final CA was the average of three different positions.Thermogravimetric analysis （TGA） was performed on aTGA550 Thermal Analyzer （Waters Corporation， U.S.A.） usinghigh-purity nitrogen as the carrier gas with a heating rate of10 °C?min?1 in the temperature range of 30–800 °C. DLS andzeta potential measurements were performed on a ZetasizerNano ZS （Malvern Panalytical， Britain）. Optical microscopeimages were obtained using the Zeiss （Germany） Axioskop 40microscope. Oil contents in the water phase were determinedusing the Vario TOC cube （Elementar， Germany）.

3 Results and discussion

3.1 Preparation of SF

The preparation of SF follows the synthetic route in previousliterature 38. 1H NMR experiments （Fig. S9） are performed bydissolving the complex and building units in the mixed solventCDCl3/DMSO-d6 （v/v = 50 ： 1）， respectively. Consistent with theliterature results 38， the corresponding proton peaks of themixture exhibit distinct moving and broadening compared to theindependent units， indicating the host-guest interaction in theformed complex. The integral values of the peak area of thecomplex reveal that the molar ratio of PCT and TT is 3 ： 2 in theobtained assemblies. The EA result （Table S1， SupportingInformation） further confirms the stoichiometric chemicalcomposition of [PCT]3[TT]2. SEM and TEM images of the[PCT]3[TT]2 assemblies display the rectangular sheetmorphology in micron size （Fig. S10a，b）. Furthermore， the AFMimage points out that the sheet is stacked by multilayers （Fig.S10c）. The thickness of the sheet is ca. 79 nm， and the heightsteps of the smallest distances are ca. 1.4 nm （Fig. S10d）. Afterultrasonic dispersion in mixed H2O/DMSO （v/v = 4 ： 1）， singlelayers are exfoliated with a thickness of ca. 1.4 nm （Fig. S10e，f）.The multiple diffraction peaks in the XRD pattern （Fig. S11）show that the d value in the assembly is ca. 4.9 nm， giving thecalculated pore size in the periodic 2D structure of ca. 4 nm. Theobtained results on composition and morphology are inaccordance with those reported in the previous literature 38，demonstrating the successful preparation of SF.

3.2 Fabrication and characterization of GO-assistedSF membrane

The composite membranes are fabricated by successivelyfiltering SF and GO dispersions on the porous PTFE substrate.Balancing the separation efficiency and flux of pure SFmembrane with different thicknesses in size retention given bythe literature 38， an optimal value is taken to fix the addition ofSF. A series of composite membranes SF-GO5， SF-GO10， SFGO15，and SF-GO20 with increasing GO contents are preparedby adding different masses of GO. According to the digitalphotos of the as-prepared membranes， the surface of the SF-GOmembrane is smooth without visible cracks （Figs. 1a and S12），and the membrane remains well under bending （Fig. 1b）. Theseresults indicate the good processibility of the membranes， whichis ascribed to the high flexibility from SF and GO. Chemical compositions of the composite membrane are investigated byFT-IR spectra as demonstrated in Fig. 1c. Because of the largeproportion of SF in the membrane， the spectrum of SF-GO10 issimilar to that of the sole SF. Nonetheless， the appearance of theC＝O peak of carboxyl groups at 1740 cm?1 attributed to GOillustrates the successful load of GO on SF. The same series ofdiffraction peaks as SF is presented in the XRD pattern of theSF-GO10 （Fig. S11）， indicating that the layered porous structureof SF still exists in the presence of GO.

Surface morphologies of the SF-GO membranes are recordedby SEM. Compared to the smooth SF membrane （Fig. 1d）， thecomposite membranes with rising GO contents present anincreasing fold structure （Figs. 1e and S13a–c）， which is similarto the original GO morphology （Fig. S13d）. The resultsdemonstrate that wrinkled GO deposits on the surface of themembrane via the successive filtration of SF and GO dispersionsduring the membrane fabrication. The elemental mappingimages （Fig. 1f） of SF-GO10 show homogeneous distributionsof N elements from PCT and Mo and Mn elements from MnMo6in SF， indicating the uniform SF layer in the membrane. C andO elements are concentrated in the wrinkle， corresponding to theGO sheets. Cross-sectional SEM images （Fig. S14a–e） reveal astructure without obvious interlayer gaps. Considering the fixedamount of SF in this series of SF-GO membranes， the factoraffecting the membrane thickness should be the loading amountof GO. In the membranes， the single-layer GO with nanosizedthickness is used in low added quantities， then the obtained SFGOmembranes exhibit thicknesses of around 10 μm with littledifference （Fig. S14f）.

To evaluate the effect of GO addition on the pore size in thecomposite SF-GO membrane， the effective cut-off size of themembrane with or without GO is tested via the filtrationseparation experiment of SiNPs. Since the SiNPs are electricallyneutral in an aqueous solution， the influence of electrostaticinteraction during the separation process can be ignored. Asshown in the DLS result and TEM photo （Fig. 2a）， the feed SiNPsolution displays a size distribution within 2–16 nm. Afterseparation by SF， only small particles less than 4 nmoverwhelmingly leave in the filtrate （Fig. S15）. Taking SFGO10as an example of the SF-GO membrane to sieve theSiNPs， a similar distribution below 4 nm of particle size isobserved in the filtrate （Fig. 2b）， indicating the close cut-off sizeof SF and SF-GO10 ca. 4 nm. In addition， according to thecalculation by the Barrett-Joyner-Halenda model， the nitrogenadsorption and desorption process of SF-GO10 points out anaverage pore diameter of ca. 3.8 nm （Fig. S16）， which is similarto the effective cut-off size obtained in the SiNP separationexperiment. These results reveal that the pore size in themembrane remains unchanged after loading GO， proving that theaddition of GO does not affect the SF pores of ca. 4 nm that playa role in size interception during the separation process.

Except for pore size distribution， the surface charge of thematerial also has an important influence on the separationperformance of emulsions 41. The surface zeta potential of thepure SF， pure GO， and the composite SF-GO over the range frompH 1–12 is presented in Figs. 2c and S17. The membranes show close negative charges， and their zeta potential decreases withthe increasing pH value. For SF-GO10 （Fig. 2c）， the surfacepotential of the membrane is ?14.5 mV at pH 1. When the pH isgreater than 3， the values are less than ?35.0 mV， and theminimum potential even gets ?61.0 mV at pH 12. The graduallydownward trend of the charge values may be due to the shift ofthe movable electron cloud surrounding the POMs in SF and thedeprotonation of abundant carboxyl groups in GO 42.

3.3 Wettability and anti-fouling properties of themembrane

The surface wettability of the material is critical for oil-wateremulsion treatment. The wettability of the membranes withdifferent GO-loaded amounts is checked according to the CAtowards water and oil. First， the effect of GO addition onhydrophilicity is studied by testing the WCA. In the air， the pureSF membrane shows a WCA of ca. 125° and takes around 35 sto let the water droplet completely spread （Figs. 3a and S18a，b）.Since the rich oxygen-containing functional groups on thesurface of GO， such as hydroxyl and carboxyl， exhibits betterhydrophilicity than SF， with a smaller WCA of ca. 52° and lesswater infiltration time below 10 s. In the variation curves ofWCA with time corresponding to the SF， composite SF-GO， andGO （Fig. S18c）， when the composite membrane contains ahigher GO loading amount， the WCA on the membrane surfacein the air gradually decreases， and water droplets spend less timeon the membrane surface to spread. For SF-GO20， the WCAreduces to 62°， and the time of WCA transferred to 0° drops toless than 10 s. The results indicate that the composite SF-GOmembranes possess improved hydrophilicity when more GO isadded. With the increase of the GO addition， the proportion ofGO in the chemical composition of the membrane surfacegradually rises. Considering the water affinity of GO， the spreading and infiltration process of the water dropletsaccelerates.

In the air， the dried SF and a series of composite membranesshow the oil contact angle （OCA） of ca. 0°， and the OCA of theGO membrane is ca. 9.5° （Fig. S19）， revealing the oleophylicityof these membranes in the air. However， the underwater oilcontact angles （UWOCAs） of SF-GO are above 150° （Fig. 3b），demonstrating that composite membranes have underwatersuperoleophobicity. Based on the variation of the Youngequation， the oleophobic wetting behavior of the amphiphilicSF-GO membrane is inferred 42. When water contacts themembrane surface， the hydrophilicity of the membrane promotesthe rapid diffusion of water， forming a stable water layer. For anoil-water-solid three-phase interface， the UWOCA follows theequation 43：

cosθOW = （γOAcosθO ? γWAcosθW）/γOW （4）

where θOW， θO， and θW represent the inherent CA of oil underwater， oil in the air， and water in the air， while γOA， γWA， and γOWmean the interface tensions of oil in the air， water in the air， andoil under water， respectively. θO and θW of the membranes are0°， hence cosθO and cosθW are both equal to 1. From commonsense， γWA is generally higher than γOA due to the higher surfacetension of water than oil. And γOW must be a positive value. So，cosθOW is smaller than 0 with the θOW ranges from 90° to 180°，proving that it is reasonable for a hydrophilic interface to beoleophobic under water.

Considering that the underwater oleophobic nature of SF-GOis beneficial to the resistance to oil pollution in water and thatthe GO loaded on the membrane is a recognized anti-foulingmaterial 44， the anti-fouling property of the composite membraneis evaluated by taking SF-GO10 as an example. Dynamic contacttest is performed by successively contacting， preloading， anddetaching the oil droplet on the membrane under water. Duringthe relaxation process， the oil droplet is elongated on the pureSF， while there is almost no deformation on SF-GO10 （Fig. 4a），indicating that SF-GO10 has smaller adhesion to oil dropletsthan the original SF. The oil jet test is implemented by injectingoil dyed red on the membrane surface underwater to further testthe anti-adhesion property （Fig. 4b）. When the tilt angle is 10°，the injected oil droplet adheres to the surface of the SFmembrane. However， for SF-GO10， the oil droplet immediatelydetaches from the surface and does not adhere to the membrane.These results reveal a higher repulsion towards the oil of SFGO10than the original SF. Moreover， to be closer to the actualemulsion separation scene， the wettability of SF and SF-GO10is checked after immersion in emulsion （H/W） for 2 h. SFpresents a low oil resistance ability， with the UWOCA droppingfrom 152° to 118° after soaking （Fig. S20）. However， theUWOCA of SF-GO10 after immersion remains at 153°，indicating the unchanged underwater superoleophobic feature ofthe composite membrane. The above experiments demonstratethat the introduction of GO could improve the anti-foulingproperty of the membrane， which may profit from the oilblockingeffect of GO. Because of the strong water affinity ofoxygen-containing groups， hydrophilic GO could absorb waterto form a stable hydration layer to resist oil pollutants.

3.4 Membrane separation performance

To elucidate the contribution of the loaded GO to themembrane separation ability， pure water flux tests and emulsionseparation experiments are performed using a filtrationseparation device （Fig. S21a）. As shown in Fig. 5a， the initial SFexhibits a pure water permeability of 2774.7 L?m?2?h?1?bar?1.With the rise of GO content， the flux first increases and thendecreases， and SF-GO10 has the highest one of 5203.4L?m?2?h?1?bar?1. The first rapid growth of the permeability isattributed to the improvement of surface hydrophilicitybenefitted from a load of superhydrophilic GO. According toLaplace equation 45：

ΔP = ?2γcosθ/r （5）

where ΔP means the pressure change experienced by the liquid，γ and θ is the liquid surface tension and CA on the surface， andr represents the radius of the pores. Hydrophilic surface bearinga WCA θ less than 90° shows a positive cosθ， which means thatwater is subjected to downward pressure with ΔP less than 0.And for a surface with better hydrophilicity， the smaller θ leadsto a higher cosθ， demonstrating a bigger ΔP corresponding to thestronger downward force. Hence， the modification of GO withexcellent water affinity could provide a dragging capillarity force to promote the rapid transfer of water drops. The stablehydrated layer formed on the hydrophilic surface also offersassistance for the water phase to penetrate. Therefore， SF-GO10owns a high level of water permeability. However， the purewater flux of the SF-GO composite membrane has graduallydecreased since SF-GO15， which may be due to the coverage ofthe porous assembly surface by the excessive GO without poresin the membrane.

The emulsion separation performance of the SF-GOcomposite membranes with different GO loading is evaluated byseparating surfactant-free O/W emulsion H/W. Duringseparation， the same device is used for the pure water flux test（Fig. S21a）， and the emulsions are poured onto the water-wettingmembranes. The cloudy emulsion feed turns transparent after theseparation by a series of membranes （Fig. S21b）. Under theoptical microscope， the oil droplets smaller than 20 μm can beseen dispersed in the continuous phase of the emulsion （Fig.S22a）， while no droplet is observed in the filtrates （Fig. S22b–f）.The above results indicate that these membrane materials arecapable of intercepting the dispersed oil droplets in thesurfactant-free O/W emulsion.

TOC tests are performed to measure the oil content in feed andfiltrates. Before separation， the oil concentration in the H/Wemulsion is 6290 ppm， and the contents in the filtratecorresponding to the SF-GO membranes with different GOcontent are all near 0， illustrating the separation efficiency ofaround 99.99% for these membranes （Fig. 5b）. Notably， with theincrease of the GO content in SF-GO membranes， the emulsionseparation flux of this series of membranes first increases andthen decreases， which is consistent with the trend of pure waterflux （Fig. 5a）. And SF-GO10 shows the highest separation fluxof 1546.6 L?m?2?h?1?bar?1. The above results reveal that both theoriginal SF and SF-GO composite membranes can treat thesurfactant-free O/W emulsion with high separation efficiency.However， an obvious difference exists in the permeability， andliquid passes through SF-GO10 fastest. This phenomenon is dueto SF-GO10 possessing the more appropriate amount of GO，which greatly balances the improvement of the hydrophilicity ofthe surface and the coverage of the SF pores. Thus， SF-GO10shows a better separation ability with a higher water penetrationspeed.

To further assess the separation performance towardssurfactant-stabilized O/W emulsions， Tween H/W， SDS H/W，and CTAB H/W are prepared with the addition of surfactantsbearing non-ionic， anionic， and cationic heads. Considering thatSF-GO10 performs the optimal separation performance amongthe SF-GO composite membranes and single componentmaterials in the separation process of the surfactant-freeemulsion H/W， SF-GO10 is further applied to treat the threetypes of surfactant-stabilized emulsions mentioned above. DLStests precisely present the size distribution of the oil droplets infeed and filtrate. As shown in Fig. 6a， the droplet sizedistribution in Tween H/W is around 2–7 nm， and the dropletsize decreases to 1–4 nm after separation by SF-GO10. For SDSH/W and CTAB H/W， the size of the oil droplet distributes overa wide range of 10–70 nm， and it should be noted that no dropletis detected in the filtrates （Fig. 6b，c）. It can be seen from theabove results that the composite membrane SF-GO10 blocks theoil droplets above 4 nm in the O/W emulsion stabilized by nonionicsurfactants， while for those stabilized by anionic andcationic surfactants， the composite membrane nearly blocks alldispersed phase droplets.

Measured by TOC， the oil contents in the filtrates of TweenH/W， SDS H/W， and CTAB H/W are 9.2， 2.2， and 6.0 ppm （Fig.6d）， respectively， which are less than the standard proposed bymany countries and organizations （Table S2）. The calculatedretention efficiency of SF-GO10 for the emulsions is higher than99.9% （Fig. 6d）. The permeability of SF-GO10 is at a high levelof 749.2， 1094.1， and 895.6 L?m?2?h?1?bar?1 for the H/Wemulsion stabilized by Tween 80， SDS， and CTAB， respectively（Fig. 6e）. The separation efficiency of SF for these threesurfactant-stabilized emulsions reaches 99.8% （Table S3）.However， the permeability for Tween H/W， SDS H/W， andCTAB H/W is only 226.6， 308.3， and 239.0 L?m?2?h?1?bar?1，respectively. The flux of SF-GO10 is ca. 3.5 times thecorresponding value of SF， which is consistent with theirperformance for the separation of H/W emulsion. These resultsfurther prove that on the premise of ensuring high separationefficiency， the introduction of appropriate GO can improve theseparation flux of the composite membrane for O/W emulsions containing nanosized oil droplets. The difference in separationefficiency and flux of SF-GO10 for the three emulsions may bedue to the difference in the electrification of the surfactants.Compared to various separation materials towards nanosizedO/W emulsion （Table S4） 23，28，46–49， the SF-GO compositemembrane intercepts almost all liquid droplets， resulting in amuch lower TOC in the filtrate.

3.5 Reusability and stability of the membrane

To illustrate the anti-fouling capacity of the GO-assistedmembrane in recycling， cycle experiments toward emulsionseparation are carried out. During the separation process， theseparation flux of SF-GO10 for three emulsions decreasesbecause the filter cake formed by the accumulation ofcontaminants blocks water from passing through （Fig. 7a）. Aftereach separation process， the permeability could recover to someextent via washing with pure water to wipe off the pollutantsfrom the membrane. The FRR of SDS H/W， CTAB H/W， andTween H/W remains at 96.6%， 93.9%， and 88.1%， respectively（Fig. 7b）. On the contrast， the flux of the original SF decreasesmore obviously after each cycle （Fig. S23a）， and the FRRtowards the H/W emulsion stabilized by SDS， CTAB， andTween 80 is 88.4%， 82.7%， and 78.8%， which is lower than thecounterpart of SF-GO10 （Fig. S23b）. These results show thatcompared to the original SF membrane， the compositemembrane modified with GO demonstrates a higher recoveryproportion on the emulsion separation performance， so it isreusable in recycling.

The stability under different conditions of the membrane iscrucial for application in practical separation. The firmness ofthe loaded GO on the composite membrane is tested in repeatedseparation experiments. After 5 separation cycles， SF-GO10remains a similar zeta potential with the as-prepared one at thepH of 1–12， demonstrating the stable surface negative charge（Fig. S24）. FT-IR spectrum of SF-GO10 after 5 cycles showsunmoved characteristic peaks with the as-prepared membrane，especially the peaks corresponding to GO have no apparentchange （Fig. S25）. In the SEM image， wrinkled GO nanosheetscan still be found on the membrane （Fig. S26）. These resultsdemonstrate the stable composition of SF-GO10 with no loss ofGO in repeated use. TGA results reveal that SF-GO10 has noweight loss below 150 °C （Fig. S27）， and SF-GO10 with atreatment at the same temperature for 3 h maintains a similarseparation flux and efficiency for H/W emulsion with theoriginal one （Fig. 7c，d）. These results indicate that the membraneis stable below 150 °C. To characterize the resistance of themembrane to corrosive environments， SF-GO10 is immersed inacid （1 mol?L?1 of HCl）， alkali （0.01 mol?L?1 of NaOH）， and salt（1 mol?L?1 of NaCl） solutions for 12 h. In the FT-IR spectra （Fig.S28）， no peak shifting before and after immersion illustrates theunchanged chemical composition in the membrane. And forcorrosive emulsions， the composite membrane exhibits the samelevel of separation performance as H/W （Fig. 7c，d）. The aboveresults show that the membrane is stable to repeated use， heating，and corrosion.

3.6 Possible emulsion separation mechanism

The above separation results of SF-GO10 for three emulsions stabilized by anionic SDS， cationic CTAB， and non-ionic Tween80 reveal that the TOC content in the corresponding filtrategradually increases， the separation flux and the FRR decrease inturn， indicating the separation performance of gradualdeterioration. The difference in separation performance for threeemulsions may be caused by the different charges of thesurfactants in the emulsion. The demulsification experiments arecarried out by leaving the emulsions standing with or without themembrane for 20 min. Compared to the control group， theemulsions stabilized by SDS and CTAB become clear （Fig.S29a，b）， while the Tween 80 stabilized one shows almost nochange （Fig. S29c）. The results indicate that SF-GO10 has thedemulsification ability for emulsions stabilized by anionic andcationic surfactants but not for emulsions stabilized by non-ionicones.

The SF-GO10 membrane with underwater oleophobicitypossesses a pore size of ca. 4 nm， which contributes to theinterception of nanosized oil droplets in the continuous waterphase of the emulsion， depending on the size-sieving effect.Moreover， because of the coexistence of SF and GO， the surfaceof the composite membrane bears negative charges， whichprovides electrostatic interaction to promote demulsification.The emulsion separation process of SF-GO10 can be performed.For the emulsion containing uncharged oil droplets， such asTween H/W， according to the size retention mechanism， themembrane intercepts oil droplets larger than the pore size（Scheme 2a）. The oil droplets that are not subject todemulsification accumulate on the membrane， forming a cakelayer. The contaminated layer hinders the liquid flow anddestroys the separation performance in repeated use. Therefore，SF-GO10 shows the lowest flux and FRR to Tween H/W. Foremulsion droplets with negative charge on the surface stabilized by SDS， when the oil droplets are close to the membrane surface，the membrane with the same negative charges provideselectrostatic repulsion to the oil droplets （Scheme 2b）. The SDSmolecules on the droplet surface rearrange， and the emulsionwith damaged stability breaks. For CTAB H/W emulsiondroplets that are positively charged， the membrane with oppositeelectrical nature promotes the oil droplets to approach throughthe electrostatic attraction （Scheme 2c）. The system becomesunstable with the rearrangement of the CTAB molecules on thedroplet surface， and demulsification occurs. Because of someadsorption of CTAB molecules on the membrane surface underelectrostatic attraction， the pores are a little blocked. The fluxand FRR of SF-GO10 when separating CTAB H/W emulsion areslightly lower than those for the separation of SDS H/Wemulsion.

To verify the emulsion separation process proposed above， thesurfactant contents in the filtrate and membrane washingsolution of the surfactant stabilized emulsion are characterizedby 1H NMR （Fig. S30）. The filtrate of Tween H/W contains 2.1%of Tween 80， and 85.9% is detected on the membrane. Theremaining portion may be lost during operation. SDS or CTABis not detected in the corresponding filtrates of the SDS H/W andCTAB H/W emulsion， and the membranes contain 93.1% and90.2% after separation， respectively. These results support theabove analysis of the separation mechanism. For Tween H/W，SF-GO10 blocks the oil droplets according to the size sieving.Although most droplets larger than the pores of the membrane at4 nm are intercepted， some small aggregates containingsurfactants pass through the membrane， so a small amount ofTween 80 is detected in the filtrate. For SDS H/W and CTABH/W stabilized by the ionic surfactants， when the compositemembrane intercepts oil droplets based on pore size， thenegatively charged membrane surface can also provideelectrostatic interaction to promote demulsification. Afterdemulsification， surfactants are still at the two-phase interface ofoil and water to reduce the interfacial energy. Due to its abilityto intercept large oil droplets， SF-GO10 prevents oil andsurfactants from entering the filtrate. The content of CTABdetected on the membrane is slightly higher than the counterpartof SDS， which may be due to the electrostatic attraction thatcauses part of CTAB to be adsorbed on the membrane. Thisresult is consistent with the slightly lower flux and FRR ofCTAB H/W than that of SDS H/W during the emulsionseparation process.

4 Conclusions

In conclusion， SF-GO composite membranes are developedby successively filtering the dispersions of SF and GOnanosheets on the substrate. The introduction of GO has noeffect on the cut-off size from the framework nanopores andnegatively charges from the POMs units in SF. Under anoptimized amount of hydrophilic GO， the composite membraneshows improved water affinity with underwater oleophobicproperty， which can be well explained by the Young equation.The stable hydration layer formed by GO endows theantiadhesion effect of the membrane to resist oil pollutants.Benefiting from these features， the SF-GO composite membraneperforms the treatment of nanosized droplets in O/W emulsionsstabilized by non-ionic， anionic， and cationic surfactants.According to the size screening process， the compositemembrane for the non-ionic emulsion shows a cut-off size of 4nm， consistent with the membrane pore size. Combining thesize-sieving and demulsification effects， the compositemembrane exhibits complete interception of dispersed dropletsin ionic emulsions. The TOC contents in the obtained filtratesare below 10 ppm， with a calculated separation efficiency of over99.9%， better than the standards of many countries andorganizations. In addition， the improved hydrophilicity increasesthe water flux through the membrane driven by the downwardpressure under the analysis of the Laplace equation. As a result，high permeability is obtained even for surfactant-stabilizedemulsions. More importantly， through simple water cleaningtreatment， because of its enhanced anti-fouling performance， themembrane is suitable for emulsion separation in multiple cycleswith high FRR. It is expected that the present method can leadto a simultaneously high efficiency and flux treatment towardsnanosized emulsions in medicine， food， and personal caresewage.

Author Contributions： Conceptualization， Methodology，Measurement， Investigation， Verification， Writing OriginalDraft， Zhang， Y.; Analyze data， Review amp; Editing， Li， B.;Conceptualization， Analyze data， Review amp; Editing， Wu， L. X.

Supporting Information： available free of charge via theinternet at https：//www.whxb.pku.edu.cn.

References

（1） Peterson， C. H.; Rice， S. D.; Short， J. W.; Esler， D.; Bodkin， J. L.;Ballachey， B. E.; Irons， D. B. Science 2003， 302， 2082.doi： 10.1126/science.1084282

（2） Schrope， M. Nature 2010， 466， 304. doi： 10.1038/466304a

（3） Cai， Q.; Zhu， Z.; Chen， B.; Zhang， B. Water Res. 2019， 149， 292.doi： 10.1016/j.watres.2018.11.023

（4） Jiang， Y.; Xian， C.; Xu， X.; Zheng， W.; Zhu， T.; Cai， W.; Huang， J.;Lai， Y. J. Membr. Sci. 2023， 667， 121166.doi： 10.1016/j.memsci.2022.121166

（5） Solans， C.; Izquierdo， P.; Nolla， J.; Azemar， N.; Garciacelma， M.Curr. Opin. Colloid Interface Sci. 2005， 10， 102.doi： 10.1016/j.cocis.2005.06.004

（6） Mason， T. G.; Wilking， J. N.; Meleson， K.; Chang， C. B.; Graves， S.M. J. Phys. Condens. Matter. 2006， 18， 635.doi： 10.1088/0953-8984/18/41/r01

（7） Tadros， T.; Izquierdo， P.; Esquena， J.; Solans， C. Adv. ColloidInterface Sci. 2004， 108， 303. doi： 10.1016/j.cis.2003.10.023

（8） Zouboulis， A. I.; Avranas， A. Colloid. Surf. Physicochem. Eng. Asp.2000， 172， 153. doi： 10.1016/S0927-7757（00）00561-6

（9） Rattanapan， C.; Sawain， A.; Suksaroj， T.; Suksaroj， C. Desalination2011， 280， 370. doi： 10.1016/j.desal.2011.07.018

（10） Pitakpoolsil， W.; Hunsom， M. J. Taiwan Inst. Chem. Eng. 2013， 44，963. doi： 10.1016/j.jtice.2013.02.009

（11） Li， W.; Yong， J.; Yang， Q.; Chen， F.; Fang， Y.; Hou， X. ActaPhys. -Chim. Sin. 2018， 34， 456. [李文濤， 雍佳樂， 楊青， 陳烽，方瑤， 侯洵. 物理化學(xué)學(xué)報， 2018， 34， 456.]doi： 10.3866/PKU.WHXB201709211

（12） Zheng， W.; Huang， J.; Li， S.; Ge， M.; Teng， L.; Chen， Z.; Lai， Y. ACSAppl. Mater. Interfaces 2020， 13， 67. doi： 10.1021/acsami.0c18794

（13） Liang， Y.; Yang， E.; Kim， M.; Kim， S.; Kim， H.; Byun， J.; Yanar， N.;Choi， H. Chem. Eng. J. 2023， 452， 139710.doi： 10.1016/j.cej.2022.139710

（14） Gao， S. J.; Zhu， Y. Z.; Zhang， F.; Jin， J. J. Mater. Chem. A 2015， 3，2895. doi： 10.1039/c4ta05624h

（15） Hu， M.-X.; Niu， H.-M.; Chen， X.-L.; Zhan， H.-B. Colloids Surf. A2019， 564， 142. doi： 10.1016/j.colsurfa.2018.12.045

（16） Naik， N. S.; Padaki， M.; Déon， S.; Karunakaran， G.; Dizge， N.;Saxena， M. J. Water Process Eng. 2019， 32， 100959.doi： 10.1016/j.jwpe.2019.100959

（17） Zhu， Y.; Xie， W.; Zhang， F.; Xing， T.; Jin， J. ACS Appl. Mater.Interfaces 2017， 9， 9603. doi： 10.1021/acsami.6b15682

（18） Zhan， B.; Liu， Y.; Li， S.-Y.; Kaya， C.; Stegmaier， T.; Aliabadi， M.;Han， Z.-W.; Ren， L.-Q. Appl. Surf. Sci. 2019， 496， 143580.doi： 10.1016/j.apsusc.2019.143580

（19） Wang， J.; He， B.; Ding， Y.; Li， T.; Zhang， W.; Zhang， Y.; Liu， F.;Tang， C. Y. ACS Appl. Mater. Interfaces 2021， 13， 4731.doi： 10.1021/acsami.0c19561

（20） Zeng， X.; Qian， L.; Yuan， X.; Zhou， C.; Li， Z.; Cheng， J.; Xu， S.;Wang， S.; Pi， P.; Wen， X. ACS Nano 2017， 11， 760.doi： 10.1021/acsnano.6b07182

（21） Kwon， G.; Panchanathan， D.; Mahmoudi， S. R.; Gondal， M. A.;McKinley， G. H.; Varanasi， K. K. Nat. Commun. 2017， 8， 14968.doi： 10.1038/ncomms14968

（22） Zhu， X.; Zhang， J. Q.; Zhu， L.; Wang， R.; Gan， S.; Xue， J. W.; Liu，X.; Li， H.; Xue， Q. Z. Sep. Purif. Technol. 2022， 280， 119984.doi： 10.1016/j.seppur.2021.119984

（23） Zuo， J.; Liu， Z.; Zhou， C.; Zhou， Y.; Wen， X.; Xu， S.; Cheng， J.; Pi， P.J. Hazard. Mater. 2020， 403， 123620.doi： 10.1016/j.jhazmat.2020.123620

（24） Zolfaghari， R.; Fakhru’l-Razi， A.; Abdullah， L. C.; Elnashaie， S. S. E.H.; Pendashteh， A. Sep. Purif. Technol. 2016， 170， 377.doi： 10.1016/j.seppur.2016.06.026

（25） Liang， H.; Esmaeili， H. Environ. Technol. Innovation 2021， 22，101498. doi： 10.1016/j.eti.2021.101498

（26） Xu， X.; Zhu， T.; Zheng， W.; Xian， C.; Huang， J.; Chen， Z.; Cai， W.;Zhang， W.; Lai， Y. Chem. Eng. J. 2023， 451， 137879.doi： 10.1016/j.cej.2022.137879

（27） Mao， X.; Zhao， Z.; Yang， D.; Qiao， C.; Tan， J.; Liu， Q.; Tang， T.;Zhang， H.; Zeng， H. Sep. Purif. Technol. 2022， 285， 120382.doi： 10.1016/j.seppur.2021.120382

（28） Hu， Y.-Q.; Li， H.-N.; Xu， Z.-K. J. Membr. Sci. 2022， 648， 120388.doi： 10.1016/j.memsci.2022.120388

（29） Zhang， K.-D.; Tian， J.; Hanifi， D.; Zhang， Y.; Sue， A. C.-H.; Zhou， T.-Y.; Zhang， L.; Zhao， X.; Liu， Y.; Li， Z.-T. J. Am. Chem. Soc. 2013，135， 17913. doi： 10.1021/ja4086935

（30） Zhang， X. Acta Phys. -Chim. Sin. 2020， 36， 2004004. [張希. 物理化學(xué)學(xué)報， 2020， 36， 2004004.] doi： 10.3866/PKU.WHXB202004004

（31） Yue， L.; Wang， S.; Zhou， D.; Zhang， H.; Li， B.; Wu， L. Nat. Commun.2016， 7， 10742. doi： 10.1038/ncomms10742

（32） Guan， W.; Wang， G.; Li， B.; Wu， L. Coord. Chem. Rev. 2023， 481，215039. doi： 10.1016/j.ccr.2023.215039

（33） Zhou， Y.; Zhang， G.; Li， B.; Wu， L. ACS Appl. Mater. Interfaces 2020，12， 30761. doi： 10.1021/acsami.0c05947

（34） Duan， F.; Liu， X.; Qu， D.; Li， B.; Wu， L. CCS Chem. 2021， 3， 2676.doi： 10.31635/ccschem.020.202000498

（35） Li， B.; Wu， L. Polyoxometalates 2023， 2， 9140016.doi： 10.26599/pom.2022.9140016

（36） Zhang， G.; Li， B.; Zhou， Y.; Chen， X.; Li， B.; Lu， Z.-Y.; Wu， L. Nat.Commun. 2020， 11， 425. doi： 10.1038/s41467-019-14227-6

（37） Zhang， Y.; Zhang， G.; Li， B.; Wu， L. Small Methods 2023， 7，2201455. doi： 10.1002/smtd.202201455

（38） Zhang， G.; Li， X.; Chen， G.; Zhang， Y.; Wei， M.; Chen， X.; Li， B.;Wu， Y.; Wu， L. Nat. Commun. 2023， 14， 975.doi： 10.1038/s41467-023-36684-w

（39） Ma， S.-D.; Chen， Y.-L.; Feng， J.; Liu， J.-J.; Zuo， X.-W.; Chen， X.-G.Anal. Chem. 2016， 88， 10474. doi： 10.1021/acs.analchem.6b02448

（40） Gao， S.; Zhu， Y.; Wang， J.; Zhang， F.; Li， J.; Jin， J. Adv. Funct. Mater.2018， 28， 1801944. doi： 10.1002/adfm.201801944

（41） An， Y. P.; Yang， J.; Yang， H. C.; Wu， M. B.; Xu， Z. K. ACS Appl.Mater. Interfaces 2018， 10， 9832. doi： 10.1021/acsami.7b19700

（42） Eda， G.; Chhowalla， M. Adv. Mater. 2010， 22， 2392.doi： 10.1002/adma.200903689

（43） Liu， M.; Wang， S.; Wei， Z.; Song， Y.; Jiang， L. Adv. Mater. 2009， 21，665. doi： 10.1002/adma.200801782

（44） Feng， L.; Gao， Y.; Xu， Y.; Dan， H.; Qi， Y.; Wang， S.; Yin， F.; Yue， Q.;Gao， B. J. Hazard. Mater. 2021， 420， 126681.doi： 10.1016/j.jhazmat.2021.126681

（45） Kim， B.-S.; Harriott， P. J. Colloid Interface Sci. 1987， 115， 1.doi： 10.1016/0021-9797（87）90002-6

（46） Lu， T.; Deng， Y.; Cui， J.; Cao， W.; Qu， Q.; Wang， Y.; Xiong， R.; Ma，W.; Lei， J.; Huang， C. ACS Appl. Mater. Interfaces 2021， 13， 22874.doi： 10.1021/acsami.1c05667

（47） He， H.; Liu， Y.; Zhu， Y.; Zhang， T. C.; Yuan， S. Sep. Purif. Technol.2022， 293， 121089. doi： 10.1016/j.seppur.2022.121089

（48） Zheng， Y.; Zhang， C.; Wang， L.; Long， X.; Zhang， J.; Zuo， Y.; Jiao， F.Sep. Purif. Technol. 2021， 272， 118893.doi： 10.1016/j.seppur.2021.118893

（49） Zhu， Y.; Wang， J.; Zhang， F.; Gao， S.; Wang， A.; Fang， W.; Jin，J. Adv. Funct. Mater. 2018， 28， 1804121.doi： 10.1002/adfm.201804121

國家自然科學(xué)基金（22271117， 22172060）資助項目