Graphene oxide membranes

Abstract

Disclosed herein is a graphene oxide film comprising a laminar arrangement of graphene oxide lamella having cyclodextrin intercalated therein, the laminar arrangement having a first interlayer spacing and a second interlayer spacing larger than the first interlayer spacing, wherein the graphene oxide film has a cyclodextrin to graphene oxide mass ratio of less than 1:1.

Description

GRAPHENE OXIDE MEMBRANES Field

[0001] The present invention is directed toward graphene oxide membranes having enhanced fluid permeance and host molecule selectivity. Background

[0002] In recent years there has been interest in exploiting two-dimensional nanosheets as platforms for fluid separation processes. The ordered arrangement of laminar 2D structures creates confined nanometre-scale interlayer space for fluid transport and separation.

[0003] Graphene Oxide (GO) is of particular interest given its ease of fabrication, exceptional theoretical transport properties i.e. slip flow, and the ability to manipulate interlayer distance to control solute permeability. Nevertheless, pristine GO membranes are typically associated with low water permeance. Swelling of GO membranes compromises stability and size exclusion effects resulting in decreased performance particularly with high solute concentrations.

[0004] Typical routes for improving the performance of GO membranes rely on narrowing the interlayer spacing through thermal or chemical reduction pathways or modulating surface charge for enhanced electrostatic repulsive interactions. GO membranes, with constricted interlayer spacing improve rejection with the trade-off that smaller and unfavourable channel chemistry drastically lowers water permeance.

[0005] Thermally reduced, cation-π stabilized, graphene oxide nanochannels have been shown to improve salt retention but at the cost of permeance. In particular, Yuan et al. (see Yuan, S., Li, Y., Xia, Y., Selomulya, C. & Zhang, X. Stable cation-controlled reduced graphene oxide membranes for improved NaCl rejection. J Memb Sci 621, (2021)) produced a reduced graphene oxide (rGO) membrane with nanochannels of 5.3 Å in size. This rGO membrane improved NaCl retention from 33 to 92.4 % but this was accompanied by a 90% reduction in membrane permeance of 0.6 L m-2h-1bar-1from 6.7 L m-2h-1bar-1.

[0006] To overcome this trade-off, various research groups have looked at incorporating various compounds within the rGO membrane. Wang et al. (Wang, Z., Ma, C., Xu, C., Sinquefield, S.A., Shofner, M. L. & Nair, S. Graphene oxide nanofiltration membranes for desalination under realistic conditions. Nat Sustain 4, 402–408 (2021)) incorporated π-conjugated polycyclic cations into GO to limit interlayer swelling and create tunable steric barriers within the ‘galleries’ of GO that increased the path tortuosity and narrowed the effective lateral spaces for hydrated solute transport thereby enhancing the intrapore energy barrier for diffusion leading to > 80% retention of NaCl with reduced loss at high ionic concentrations. However, low fluid permeance remained an issue.

[0007] It is desirable to provide a graphene oxide membrane having both enhanced selectivity and higher permeance. It is an object of the invention to address at least one shortcoming of the prior art and / or provide a useful alternative.

[0008] Reference to any prior art in the specification is not an acknowledgment or suggestion that this prior art forms part of the common general knowledge in any jurisdiction or that this prior art could reasonably be expected to be understood, regarded as relevant, and / or combined with other pieces of prior art by a skilled person in the art. Summary of Invention

[0010] In a first aspect of the invention, there is provided a graphene oxide film comprising a laminar arrangement of graphene oxide lamella having cyclodextrin intercalated therein, the laminar arrangement having a first interlayer spacing and a second interlayer spacing larger than the first interlayer spacing, wherein the graphene oxide film has a cyclodextrin to graphene oxide mass ratio of less than 1:1.

[0011] In an embodiment, the cyclodextrin to graphene oxide mass ratio is at least 1:1.1., preferably at least 1:1.15. Additionally or alternatively, the cyclodextrin to graphene oxide mass ratio is at most 1:5, preferably 1:4, more preferably 1:3, even more preferably 1:2.5. In one form, the cyclodextrin to graphene oxide mass ratio is about 1:1.2.

[0012] In an embodiment, the first interlayer spacing is provided by the laminar arrangement without the cyclodextrin and / or the second interlayer spacing is provided by the laminar arrangement having the cyclodextrin intercalated therein.

[0013] In an embodiment, the first interlayer spacing is from about 6 Å to about 8 Å and / or is represented by an X-ray diffraction peak from X-ray diffraction (XRD) analysis of the graphene oxide film at an angle of from about 12.5° to about 15.

[0014] In an embodiment, when measured in a dry state: (i) the second interlayer spacing is from about about 8.5 Å to about 11 Å and / or is represented by an X-ray diffraction peak from XRD analysis of the graphene oxide film at an angle of from about 8.5° to about 10.5°; or (ii) the second interlayer spacing is from about 15 Å to about 18 Å and / or is represented by an X- ray diffraction peak from XRD analysis of the graphene oxide film at an angle of from about 4° to about 6°.

[0015] In an embodiment, the graphene oxide film further comprises a third interlayer spacing larger than the second inter layer spacing.

[0016] In one form of the above embodiment, the first interlayer spacing is provided by the laminar arrangement without the cyclodextrin, and / or the second interlayer spacing is provided by the laminar arrangement having the cyclodextrin intercalated therein in a first orientation, and / or the third interlayer spacing is provided by the laminar arrangement having the cyclodextrin intercalated therein in a second orientation.

[0017] It is preferred that the first orientation corresponds cyclodextrin having a longitudinal axis projecting in a direction substantially out of a plane of the graphene oxide film, and the second orientation corresponds to the longitudinal axis projecting in a direction substantially within the plane of the graphene oxide film.

[0018] In one form of the above embodiment, when measured in a dry state: (i) the second interlayer spacing is from about 8.5 Å to about 11 Å and / or is represented by an X-ray diffraction peak of XRD analysis of the graphene oxide film at an angle of from about 8.5° to about 10.5°; or (ii) the third interlayer spacing is from about 15 Å to about 18 Å and / or is represented by an X-ray diffraction peak of XRD analysis of the graphene oxide film at an angle of from about 4° to about 6°.

[0019] In an embodiment, the graphene oxide film has a carbon to oxygen (C:O) ratio of the graphene oxide layer is less than 2.25 when measured using X-Ray Photoelectron Spectroscopy (XPS). Preferably, the C:O ratio is 2 or less. More preferably the C:O ratio is 1.95 or less.Alternatively, or additionally, the C:O ratio of 1.8 or greater. Preferably, the C:O ratio is 1.85 or greater.

[0020] In an embodiment, the graphene oxide film has a spectra, obtained using Fourier Transform Infrared Spectroscopy (FTIR) with: a O-C=O peak with a transmittance of greater than 10%; and / or a C-O peak with a transmittance of greater than 32%; and / or an OH peak with a transmittance of greater than 70%; and / or a -CH2- peak with a transmittance of greater than 85%; and / or an α-(1,4) glucopyranose peaks with a transmittance of greater than 35%; and / or a -C=O peak with a transmittance of greater than 70%. It is preferred that the OH peak has a transmittance of about 86% ±10% or ±5%. It is preferred that the CH2- peak has a transmittance of about 93% ±5%. It is preferred that the -C=O peak has a transmittance of greater than 80%, more preferably greater than 90%, and most preferably greater than 95%. These ranges generally cover reaction times between the cyclodextrin and graphene oxide of up to 150 minutes.

[0021] In an embodiment, at least a portion of the cyclodextrin is covalently bonded to the graphene oxide.

[0022] In one form of the above embodiment, the portion of the cyclodextrin that is covalently bonded to the graphene oxide is covalently bonded via an ester bond.

[0023] In one form of the above embodiment, the portion of the cyclodextrin that is covalently bonded to the graphene oxide is covalently bonded in the absence of a cross-linker.

[0024] In an embodiment, the cyclodextrin is selected from the group consisting of α- cyclodextrin, β-cyclodextrin, γ-cyclodextrin, and mixtures of two or more of the foregoing.

[0025] In an embodiment, the graphene oxide lamella is in the form of particles, such as plates, flakes, discs, or the like, having a long axis, and the long axis is substantially aligned with an axis in a plane of the graphene oxide film.

[0026] In an embodiment, the film has a permeance of at least 20 L m−2h−1bar−1determined at an applied pressure of 2 bar and ambient temperature. Preferably, the permeance is at least 25 L m−2h−1bar−1. More preferably, the permeance is at least 30 L m−2h−1bar−1. Most preferably, the permeance is at least 35 L m−2h−1bar−1.

[0027] In an embodiment, the film has a per- or poly-fluoroalkyl compounds (PFAS) retention of at least 80% determined at an applied pressure of 2 bar and at ambient temperature, the PFAS being selected from the group consisting of: perfluorobutanoic acid (PFBA), perfluoropentanoic acid (PFPeA), perfluorohexanoic acid (PFHxA), perfluorooctanoic acid (PFOA), and combinations thereof.

[0028] In a second aspect of the invention, there is provided a separation membrane comprising a membrane layer formed from a graphene oxide film according to the first aspect of the invention, and / or embodiments, and / or forms thereof.

[0029] In an embodiment, the separation membrane comprises a porous support layer with the graphene oxide film applied thereto.

[0030] In an embodiment, the graphene oxide film has a thickness of from about 20 nm to about 200 nm.

[0031] In an embodiment, the separation membrane has a permeance of at least 20 L m−2h−1bar−1determined at an applied pressure of 2 bar and ambient temperature. Preferably, the permeance is at least 25 L m−2h−1bar−1. More preferably, the permeance is at least 30 L m−2h−1bar−1. Most preferably, the permeance is at least 35 L m−2h−1bar−1.

[0032] In an embodiment, the separation membrane has a per- or poly-fluoroalkyl compounds (PFAS) retention of at least 80% determined at an applied pressure of 2 bar and at ambient temperature, the PFAS being selected from the group consisting of: perfluorobutanoic acid (PFBA), perfluoropentanoic acid (PFPeA), perfluorohexanoic acid (PFHxA), perfluorooctanoic acid (PFOA), and combinations thereof.

[0033] In a third aspect of the invention, there is provided a water treatment method comprising transporting water through a graphene oxide film according to the first aspect of the invention, and / or embodiments, and / or forms thereof or a separation membrane according to the second aspect of the invention, and / or embodiments, and / or forms thereof to remove one or more contaminants from the water and provide a filtrate substantially free of the one or more contaminants and / or with reduced concentration of the one or more contaminants.

[0034] In an embodiment, the one or more contaminants is selected from the group consisting of heavy metals and / or per- or poly-fluoroalkyl compounds (PFAS), such as perfluorobutanoic acid (PFBA), perfluoropentanoic acid (PFPeA), perfluorohexanoic acid (PFHxA), perfluorooctanoic acid (PFOA), and combinations thereof.

[0035] In a fourth aspect of the invention, there is provided an aqueous graphene oxide ink composition comprising cyclodextrin and graphene oxide in a mass ratio of less than 1:1.

[0036] In an embodiment, the graphene oxide is in an amount of at least 5 g / L. Preferably, the graphene oxide is in an amount of at least 6 g / L. More preferably, the graphene oxide is in an amount of at least 7 g / L. Most preferably, the graphene oxide is in an amount of at least 8 g / L. Preferably, the graphene oxide concentration is less than 15 g / L.

[0037] In an embodiment, the graphene oxide is in an amount sufficient that the aqueous graphene oxide ink composition is in the nematic phase.

[0038] In a fifth aspect of the invention, there is provided a method of forming a graphene oxide film, the method comprising solution casting the aqueous graphene ink composition according to the fourth aspect of the invention, and / or embodiments, and / or forms thereof onto a substrate surface and subjecting the aqueous graphene ink composition to shear forces sufficient to shear align the graphene oxide particles and form a graphene oxide film.

[0039] In an embodiment, the graphene oxide film is a wet graphene oxide film, and the method further comprises heating the wet graphene oxide film to a temperature of from about 60 C to about 99 C for a time sufficient to form a dry graphene oxide film.

[0040] As used herein, except where the context requires otherwise, the term "comprise" and variations of the term, such as "comprising", "comprises" and "comprised", are not intended to exclude further additives, components, integers or steps. Brief Description of Drawings

[0041] Further aspects of the present invention and further embodiments of the aspects described in the preceding paragraphs will become apparent from the following description, given by way of example and with reference to the accompanying drawings.

[0042] Figure 1: SEM surface images of GO, GO-βCD and GO-αCD films (A, C, and E) and cross-section images of GO, GO-βCD and GO-αCD films (B, D, and F).

[0043] Figure 2: X-ray diffraction (XRD) spectra for GO, GO-βCD and GO-αCD films in the dry and wet state.

[0044] Figure 3: Raman spectra and relative intensities of the IDand IGpeaks of GO, GO-αCD and GO-βCD.

[0045] Figure 4: FTIR spectra of GO, GO-⍺CD, GO-βCD, ⍺-CD and β-CD films.

[0046] Figure 5: Deconvolution of the HR-XPS C1s peaks of GO, GO-αCD and GO-βCD films.

[0047] Figure 6: Graph showing PFPeA (C5) retention and water permeance of GO, GO-⍺CD and GO-βCD membranes.

[0048] Figure 7: Graph showing PFPeA (50 ppm) transport rates through the GO-βCD membrane at 283.15K, 293.15K, and 303.15K (from bottom to top).

[0049] Figure 8: Arrhenius plot for PFPeA transport in GO-⍺CD, GO-βCD and GO membranes illustrating the dependency of the overall energy barrier on CD type.

[0050] Figure 9: Graph showing energy barriers to PFPeA transport vs thickness of the GO- βCD membrane.

[0051] Figure 10: Molecular dynamics calculation of the binding energy for interactions between the outer edges of αCD and βCD molecules with PFAS of chain lengths C4to C8.

[0052] Figure 11: Molecular dynamics calculation of the binding energy for interactions between the internal cavity of αCD and βCD molecules with PFAS of chain lengths C4 to C8.

[0053] Figure 12: Graph showing retention vs. molar mass for GO and GO-βCD to demonstrate dependence of retention on steric effects.

[0054] Figure 13: Graph showing PFBA, PFPeA, PFHxA, and PFOA concentrations in feed, retentate, and permeate for GO-βCD membrane.

[0055] Figure 14: Graph showing PFPeA retention vs feed concentrations for GO-⍺CD, GO- βCD and GO membranes.

[0056] Figure 15: Graph showing PFPeA retention vs pH for GO-⍺CD, GO-βCD and GO membranes.

[0057] Figure 16: Graph showing retention of 0.1ppm PFBA, PFPeA, PFHxA, and PFOA on GO-βCD

[0058] Figure 17: Graph showing PFPeA rejection vs time for GO-βCD membrane.

[0059] Figure 18: Graph showing PFBA, PFPeA, PFHxA, and PFOA retention as a function of molar mass for GO-βCD membrane. The inset shows PFBA, PFPeA, PFHxA, and PFOA concentrations in feed, retentate, and permeate for GO-βCD membrane.

[0060] Figure 19: Graph showing comparison of PFBA, PFPeA, PFHxA, and PFOA retention as a function of molar mass for GO-βCD membrane and commercially available NF270 membrane. The inset compares permeance of GO-βCD membrane and commercially available NF270 membrane

[0061] Figure 20: Graph showing performance comparison of permeance and selectivity of PFBA (C4) by representative polymeric membranes and GO-βCD.

[0062] Figure 21: Graph showing PFBA, PFPeA, PFHxA, and PFOA concentrations in feed, retentate, and permeate for commercially available NF270 membrane.

[0063] Figure 22: Graph showing performance of GO-βCD membranes with varied β-CD content.

[0064] Figure 23: Graph showing effect of βCD loading on XRD patterns of GO-βCD membranes with varied β-CD content.

[0065] Figure 24: Graph showing FTIR analysis of GO-βCD inks with reaction times varying from 30 to 150 min at a βCD content of 45 wt.%.

[0066] Figure 25: Graph showing retention of GO-βCD membranes at 0.1 ppm of C4-C8 (carboxylated) and C8 (sulfonated) PFAS, and C5 in the presence of 50 ppm of humic acid.

[0067] Figure 26: Graph showing PFPeA rejection over 24 hours of crossflow filtration on GO and GO-βCD membranes.

[0068] Figure 27: Graph showing PFPeA permeance over 24 hours of crossflow filtration on GO and GO-βCD membranes.

[0069] Figure 28: Graph showing PFAS rejection for GO–βCD, GO, and NF270 membranes during 24h crossflow operation.

[0070] Figure 29: Graph showing Humic acid rejection for GO–βCD, GO, and NF270 membranes during 24h crossflow operation.

[0071] Figure 30: Graph showing PFAS rejection for GO–βCD, GO, and NF270 membranes during 24h crossflow operation for 0.1 mg / L PFPeA, PFOA and PFOS in the presence of 20 mg / L Humic Acid.

[0072] Figure 31: Graph showing Ca2+rejection for GO–βCD, GO, and NF270 membranes during 24h crossflow operation for 2 mM Ca2+solution.

[0073] Figure 32: Graph showing PFAS rejection for GO–βCD, GO, and NF270 membranes during 24h crossflow operation for of PFPeA, PFOA, and PFOS in the presence of 1 mM Ca2+.

[0074] Figure 33: Graph showing permeance for GO–βCD, GO, and NF270 membranes during 24h crossflow operation in PFPeA, PFOA, and PFOS solutions.

[0075] Figure 34: Graph showing permeance for GO–βCD, GO, and NF270 membranes in binary PFAS-HA solutions during 24h crossflow operation for PFPeA, PFOA, and PFOS in the presence of Humic Acid.

[0076] Figure 35: Graph showing permeance for GO–βCD, GO, and NF270 membranes in binary PFAS-Ca2+solutions during 24h crossflow operation for PFPeA, PFOA, and PFOS in the presence of Ca2+.

[0077] Figure 36: Graph showing PFPeA rejection for GO–βCD membranes with varying Ca2+concentrations and corresponding permeance.

[0078] Figure 37: Graph showing PFPeA rejection for GO–βCD membranes with varying humic acid concentrations and corresponding permeance.

[0079] Figure 38: Graph showing 24-hour cross flow filtration rejection for GO–βCD membranes of carboxylated PFBA and PFOA and sulfonated PFBS and PFOS from mixtures at varying concentrations.

[0080] Figure 39: Graph showing long term cross flow rejection of PFAS from a mixture of PFPeA, PFOA, PFOS, Humic acid and Calcium for GO–βCD and GO membranes.

[0081] Figure 40: Graph showing permeances of GO-βCD and pristine GO membranes while treating a mixture of (i) PFPeA, PFOA, PFOS, and (ii) PFPeA, PFOA, PFOS, Humic acid and calcium.

[0082] Figure 41: Series of graphs illustrating fouling behaviour of GO-βCD membranes compared to that of pristine GO membranes. Description of Embodiments

[0083] The invention relates to graphene oxide membranes, such as reduced graphene oxide membranes, that have cyclodextrin intercalated therein. The inventors have found that the intercalation of cyclodextrin within or between graphene oxide lamellae advantageously provides a membrane that has both high fluid permeance and high selectivity for a guest molecule (e.g., a contaminant). In this way, the membrane is highly effective at removing the guest molecule from a fluid stream, such as water, without a reduction in fluid permeance.

[0084] Cyclodextrins are a family of cyclic oligosaccharides, consisting of a macrocyclic ring of glucose subunits joined by α-1,4 glycosidic bonds. Of particular interest are α-, β-, and γ-cyclodextrins which contain 6 to 8 glucose monomer units respectively that are arranged in a ring creating a cone shaped structure. Table 1 below summarises the physiochemical properties of α-, β-, and γ-cyclodextrins. Table 1.Cyclodextrins Physicochemical Properties

[0085] The internal cyclodextrin cavities are toroidal sugar-based macrocyclic compounds lined with multiple non-polar / semi-polar methylene groups (-CH2-) rendering the hollow interior cavity hydrophobic while the exterior is hydrophilic.

[0086] The inventors have found that the intercalation of cyclodextrin according to the disclosure herein causes the graphene oxide membranes to exhibit improved water permeance and selectivity properties. This is counter-intuitive since, generally, improving selectivity decreases permeance and increasing permeance decreases selectivity. These improvements are thought to occur due to both the physical interaction between the cyclodextrin and the graphene oxide and the chemical properties of the cyclodextrin itself.

[0087] Turning to the physical interactions between the cyclodextrin and the graphene oxide, the intercalation of graphene oxide physically spaces adjacent graphene oxide lamellae apart based on the size and orientation of the cyclodextrin molecule. This results in the graphene oxide membrane having multiple distinct interlayer spacings, e.g., a first interlayer spacing corresponding to the natural spacing between adjacent graphene oxide lamellae, and at least a second interlayer spacing arising from cyclodextrin located between graphene oxide lamellae physically spacing those graphene lamellae apart. In some embodiments, there may be anadditional interlayer spacing that is likewise caused by intercalation of cyclodextrin between graphene oxide lamellae, but in a different orientation to that in the second interlayer spacing (i.e., the cyclodextrin may be intercalated for example with its central axis parallel or tangential to the orientation of the graphene oxide lamellae). In either case, the larger interlayer spacing, arising from the intercalation of cyclodextrin, increases the fluid permeance of the membrane.

[0088] Turning to the chemical properties of cyclodextrin, cyclodextrin has a ring structure with a hydrophobic interior and external pendant hydroxyl groups. Because the interior of the cyclodextrin ring is hydrophobic, there is almost no interaction between water molecules and the inner surface of cyclodextrin. In this way, cyclodextrin is thought to enable fast transport of water molecules through its interior. Conversely, the hydroxyl pendant groups on the outer edges of the cyclodextrin molecule are thought to facilitate molecular interactions with guest molecules, e.g., via hydrogen bonding with polar components of guest molecules which then promotes hydrophobic-hydrophobic interactions. Both hydrogen bonding and hydrophobic- hydrophobic interactions are thought to hinder transport of guest molecules through the body of the cyclodextrin containing graphene oxide membrane.

[0089] In addition to the above, the interaction energy between cyclodextrin and potential guest molecules varies depending on the shape and strain of the cyclodextrin cavity. The steric interactions of the guest molecule with cyclodextrin is a factor in determining the binding strength therebetween. Given this, the selectivity for a target guest molecule and permeance of the membrane can be tailored depending on the cyclodextrin which is used. For example, with reference to Table 1 above, it can be seen that αCD has a smaller accessible cavity volume and has a distorted structure that turns symmetrical upon inclusion of a host that increases the relative energy of inclusion as compared to other cyclodextrin structures. In contrast, βCD can act as a chiral selector for separating racemates since it exhibits stronger affinity toward l- enantiomers compared to d-enantiomers, and furthermore, βCD has higher selectivity to lead (II) over other heavy metal ions through complexation with its deprotonated hydroxyl groups.

[0090] One class of guest molecules of particular interest are per- and polyfluoroalkyl substances (PFAS). PFAS are a group of toxic synthetic chemicals with fully or partially fluorinated carbon chains. PFAS have become a pollutant of global concern because of their pervasive and long-lasting presence in the environment, including ground and surface water, and its eventual physiological uptake by humans and animals. According to epidemiologicalresearch, exposure to trace amounts of PFAS is associated with higher rates of thyroid illness, cancers, and pregnancy induced hypertension. Thus, it is desirable to remove PFAS compounds from water, such as during wastewater treatment. However, removal of PFAS compounds has proven to be challenging with existing membranes. Other contaminants include humic acid, fulvic acid, tannins, phenolics, Pb, Cd, and Ni.

[0091] The inventors have found that the cyclodextrin intercalated graphene oxide membranes of the present invention are particularly well-suited to removing PFAS from aqueous solution. Thus, in another aspect, the invention is directed towards a method for removal of PFAS from aqueous solution using a cyclodextrin intercalated graphene oxide membrane.

[0092] The invention will now be described in relation to embodiments thereof. Example 1 Materials and membrane preparation

[0093] The GO reported in this example was obtained from Sixth Element Materials Technology Co. Ltd. in Changzhou, China, and prepared into a ~10 g L-1GO suspension with rheological characteristics consistent with the shear thinning behaviour required to create large area GO membranes.

[0094] Analytical grades of β-cyclodextrin (βCD) ≥ 97% and α-cyclodextrin (αCD) ≥ 98% were obtained from Sigma-Aldrich. Cyclodextrin was added to GO suspension at weight ratios of 1.2:1 (GO:CD) and the mixture was heated at 90°C under continuous stirring for 2h. The GO- CD solutions were cooled to room temperature, prior to coating on a fixed substrate.

[0095] For membrane performance testing, GO, GO-αCD and GO-βCD membranes were fabricated using shear alignment printing onto 13 ×14 cm2sheets of polyvinylidene difluoride (PVDF, HVF, 0.2 µm pore size, MDI, India). To improve stability in aqueous based filtration experiments, all the membranes were baked soon after coating, at 80°C for 1 hour in a convection oven. Performance of GO-βCD membranes was compared to a reference membrane fabricated using untreated GO suspension.

[0096] For membrane characterisation, 10 g / L GO, GO-αCD and GO-βCD suspensions were cast onto a glass substrate and dried at 40°C for 24 hours to create free-standing GO, GO-αCDand GO-βCD films. A razor blade was used to manually separate the dried film from the glass support, creating a thick freestanding film. After that, the free-standing films of GO, GO-αCD and GO-βCD received the same treatment as the fabricated membranes.

[0097] Scanning Electron Microscopy (SEM)

[0098] The surface and cross-sectional morphology of all membranes were studied in great resolution using scanning electron microscopy (SEM). All samples were sputter coated with a 2 nm carbon conductive coating in a Cressington 208 HR prior to imaging because GO is comparatively insulating. All SEM micrographs in this work were captured using ultra high resolution Thermo Scientific Verios 5 UC FEG-SEM operating at 5.0 keV and 13 pA.

[0099] Fourier Transform Infrared Spectroscopy (FTIR)

[0100] FTIR characterisations were acquired using a Perkin Elmer Spectrum 100. The spectra were captured over a 4-scan range, at wavenumbers between 4,000 and 400 cm−1and used to determine chemical structure of GO, GO-αCD and GO-βCD films through the occurrence and intensity of functional group specific peaks.

[0101] X-ray Photoelectron Spectroscopy (XPS)

[0102] X-ray photoelectron spectroscopy (XPS, Thermo Fisher, Nexsa Surface Analysis System) was used to analyse GO, GO-αCD and GO-βCD’s compositions. Survey scans, C 1s and O 1s were recorded using Al Kα radiation source, and survey scans were collected at 150 eV while C 1s and O 1s were collected at a pass energy of 50 eV. O 1s and C 1s peaks were deconvoluted through a multi-peak Gaussian fit paired with a Shirley background correction.

[0103] Raman spectroscopy

[0104] Raman spectroscopy was used to determine the structural changes that occurred after incorporation of αCD and βCD onto GO. A Reinshaw VIS Raman Spectrometer equipped with a HeNe (633 nm) laser running at 10% power was used to generate the Raman spectra. Extended scans (10 s) were carried out at wavenumbers between 200 and 3200 cm-1. GRAMS / AI software package was used for peak fitting, and the fitted curves were utilized to calculate ID / IGratios.Membrane characterisation

[0105] The CD modified GO membranes were systematically investigated using SEM, FTIR, XPS, and Raman spectroscopy as outlined above to determine changes in physiochemical properties of the GO nanochannel that may originate as a result of functionalization with cyclodextrin.

[0106] GO, GO-αCD and GO-βCD membranes showed a continuous and uniform structure as shown in Figure 1. Figure 1 are scanning electron microscopy (SEM) surface images of GO, GO-βCD and GO-αCD films (A, C, and E) and cross-section images of GO, GO-βCD and GO- αCD films (B, D, and F). The cross-sectional thickness of GO, GO-βCD and GO-αCD were determined to be 35.3 ± 4.7 nm, 37.9 ± 0.2 nm and 36.2 ± 0.5 nm, respectively. The similarity in thickness suggests that any differences in permeance are likely to be a result of the modified chemistry and expansion of the nanochannels and not changes in the thickness of the active layer of the membrane. The SEM images of the top surface of GO-βCD and GO-αCD membranes show an increase in heterogeneity of their structure in the form of increased wrinkling of the GO sheets as compared with the GO membrane which is relatively wrinkle- free.

[0107] Figure 2 shows X-ray diffraction (XRD) spectra for GO, GO-βCD and GO-αCD films in the dry and wet state. The XRD spectra confirms that this heterogeneity originates from the asymmetry of the CD-modified GO channels. The XRD spectra for GO shows the typical single, narrow, and high intensity peak at a 2θ value of 11.0° which indicates that unmodified GO has highly ordered channels with an interlayer spacing of 8.0 Å, consistent with literature. Upon addition of βCD, three new peaks appear centered at 5.3°(16.7 Å), 9.3°(9.5 Å) and 13.2°(6.7 Å) supporting the presence of an asymmetric channel structure comprised of two large channels, and a narrow channel. The narrow channel is formed from mildly reduced and stacked sp2graphitic lamella while the wider channels are a result of CD intercalation into the nanochannel. It is thought that the glucose units of CD deoxygenate GO’s hydroxyl and epoxy groups leading to the stacked domains, however, the relative orientation of CD within the nanochannel also influences the space between adjacent lamella. When the cavity of the βCD is orientated ‘upright’ with the axis of its cavity at 90° to the plane of GO, the channel is at its maximum width of 16.7 Å which correlates well with the maximum width of βCD (which has a width of ~ 15.4 Å as per Table 1). Similarly, the 9.5 Å channel can be correlated to the intercalation of βCD with oriented ‘lying flat’ with an apparent intercalation size of the axis of its cavity parallelto the basal plane of GO (i.e. the height of the CD torus structure ~7.9 Å as shown in Table 1). It is also noted that the CD-modified membranes showed little change of the XRD spectra in the dry and wet state, whereas the GO membrane interlayer distance enlarged to 21 Å in aqueous environment. The channels in GO-αCD and GO-βCD expand by only 2.7 and 0.7 Å, respectively in wet state, demonstrating the stable nature of GO-CD lamella. Furthermore, analysis of the wet state XRD illustrates that the smaller (6.7 Å in GO-βCD) nanochannels that result from reduction are non-swelling (Δd = 0 for both GO-αCD and GO-βCD). The results are summarised in Table 2 below. Table 2. Interlayer spacing of bare and modified GO in dry state and wet state calculated from XRD spectra

[0108] Figure 3 shows the Raman spectra and relative intensities of the IDand IGpeaks of GO, GO-αCD and GO-βCD. The Raman spectra for GO, GO-αCD and GO-βCD display a G-band at ~ 1600 cm−1corresponding to sp2stretching modes and D-band at ~ 1360 cm−1corresponding to sp3-hybridized carbon. The G-band peaks are related to vibration of sp2carbon atoms in the hexagonal lattice of GO, whereas the D-band peak is caused by sp3hybridised carbon atoms and is associated with defects and disorders in graphene structure. The slight increase in intensity ratios ID / IG of 0.91 for GO to 0.95 for GO-βCD suggests a decrease in average size of the sp2domains and a small increment in structural defects from incorporation of β-CD. The increase in intensity ratios (ID / IG) after incorporation of CD confirms the disruption of sp2hybridized carbon structure on both sides of GO nanosheets. An invariable intensity ratio of 0.92 for GO-αCD is observed after the addition of αCD to GO.

[0109] X-ray photoelectron spectroscopy (XPS) was used carry out surface elemental analysis of GO, GO-⍺CD and GO-βCD to further corroborate the composition and chemical construction of the samples. In order to determine the changing distribution of functional moieties of the modified membranes, the HR-XPS C1s spectra for GO, GO-⍺CD and GO-βCD were deconvoluted as shown in Figure 5, GO has peaks at 284.07 eV, 284.76 eV, 286.4 eV, 287.13 eV, and 288.38 eV which correspond to C−C, C═C, C−O, C═O and O−C═O bonds, respectively. There is an increase in peak areas O−C═O and C−O moieties, in both GO-⍺CD and GO-βCD, which suggests mild esterification between the carboxyl groups in GO with hydroxyl groups of CD.

[0110] Figure 4 is an FTIR spectra of GO, GO-⍺CD, GO-βCD, ⍺-CD and β-CD films, which shows different peaks due to the presence of specific functional groups. The spectrum of pristine GO shows a distinctive broad peak peaks between 3000-3700 cm-1from O-H stretching and vibrations of skeletal hydroxyl groups, at 1732 cm-1from C═O stretching of carboxylic acid groups1414, 1610 cm-1from aromatic C═C stretching of sp2 hybridized carbon bonds, 1385 cm-1from C-OH epoxy bending, and 1060 cm-1from C-O stretching of the alkoxy groups. A new peak at 2920 cm-1which is a characteristic of -CH2- stretching confirms the functionalization GO with CDs on the GO-βCD and GO-αCD curve. The near-complete disappearance of the peak at 1732 cm-1in GO-βCD and GO-αCD, compared to pristine GO, confirms the attachment of CDs onto GO sheets through ester bonds. This is evidenced by the formation of ester linkages between the carboxyl groups (-COOH) of GO and the hydroxyl groups (-OH) of βCD. Moreover the GO spectrum at 1385 cm-1weakens in the spectra of GO-CDs, due to partial deoxygenation associated with mild heating. This confirms reduction, and the emergence of additional between 500 cm-1and 900 cm-1inclusive of the α-(1,4) glucopyranose peak at 800 cm-1was indicative of the successful attachment of CDs onto GO. The key results are summarised in Table 3 below. Table 3. XPS C / O ratios and deconvoluted C1s of GO, GO-αCD and GO-βCD

[0111] GO has peaks at 284.81 eV, 285.11 eV, 287.0 eV and 288.74 eV which correspond to sp2(C=C) and sp3(C-C) hybridised carbons, alkoxy (C−OH and C-O-C), and carbonyl (C=O and O−C=O) bonds, respectively. There is an invariable change in the C=O and O−C=O groups (~11%) across GO and GO−CDs, due to the simultaneous depletion of the −COOH and creation of −COOR moieties from esterification reaction. A significant increase in C−OH and C−O groups for GO-βCD (56.0%) and GO-αCD (50.1%) compared to pristine GO (40.7%), is primarily attributed to the incorporation of CD’s glycosidic bonds during functionalization.

[0112] The C / O ratio changes from 2.26 for unmodified GO to 1.87 and 1.93 for GO-⍺CD and GO-βCD, respectively. CDs have an oxygen rich structure with an inherent C / O ratio of 1.5 which lowers the membrane C / O composition after incorporation. Membrane performance

[0113] Dead-end Filtration Experiments

[0114] The perm-selective characteristics of GO, GO-αCD and GO-βCD membranes were tested using four different Perfluoroalkyl carboxylic acids (PFCAs): perfluorobutanoic acid (PFBA), perfluoropentanoic acid (PFPeA), perfluorohexanoic acid (PFHxA), and perfluorooctanoic acid (PFOA).

[0115] The retention and permeance characteristics of the membranes were examined using a dead-end filtration setup. Dead-end filtration cells from Sterlitech (HP4750 Stirred Cell, Sterlitech, USA) with an effective surface area of 14.6 cm2were used for the testing. For all filtering experiments, the pressure was kept at 2 bar, and the cells' contents were constantly agitated at 400 rpm. Using Fluigent pressure pumps (MFCS-EX extended flow control, France) and Radwag precision balances (PS1000.R2, Poland), mass and pressure data were taken every second to display continuous permeability measurements. All membranes had their water permeance stabilised at 2 bar prior to testing. Three different samples were used for each test, and the results were used to compute the error. Membrane permeance, JP, (L m-2h-1bar-1, or LMH / bar) was calculated using the formula: Equation 1: JP= Vp / (A ×P×t )

[0116] Vp is the permeated volume after a given time, t, through a membrane with an effective surface area, A, at an applied pressure, P.

[0117] To determine membrane selectivity, 100 mL of PFAS prepared using distilled water was inserted inside the dead-end filtration cell. Using different PFCAs concentrations from 0.1 to 50 ppm, 75mL of the 100mL feed was permeated through the membrane under continuous agitation of 400rpm. Membrane retention, R (%), is calculated as follows: Equation 2: R = (1 -Cp / Cr)×100%

[0118] Cr and Cp are the concentrations of the retentate and the permeate, respectively.

[0119] Cross Flow Filtration Experiments

[0120] To demonstrate long term performance the membranes were tested on a cross flow setup unit. The crossflow setup comprises of a pump, four Sterlitech CF042 crossflow cells, a stainless steel 20 L feed reservoir, and a temperature control system. The Sterlitech cell accommodates a membrane with an active area of 42 cm2(9.21 x 4.57 cm). Flow rate in the unit is adjusted using a bypass valve and transmembrane pressure controlled by pressure regulators. Permeate and retentate-streams were recirculated back to the feed reservoir to maintain a constant feed volume and concentration. In crossflow filtration, the permeate and retentate streams are continuously recirculated through the feed tank, and, rejection Rj as defined as Equation 3: Rj=(1 -Cp / Cf)×100%

[0121] Cf and Cp are the concentrations of the feed and the permeate, respectively.

[0122] Energy barrier diffusion experiments

[0123] Temperature dependent diffusion experiments (between 283–315 K), were conducted in a 100 mL diffusion cell.100 ml of 50 ppm solution of PFAS is poured in the feed side and 100 ml is filled on the permeate side. The diffusion cell was placed on top of a hot plate stirrer to regulate temperature under stirring. Initially, 1 mL of permeate measurements for each temperature was collected after every 30 minutes for 3 hours to establish linearity of diffusion rates. Simultaneously, 1 mL of feed solution was pipetted out to maintain volume of each diffusion cell. Thereafter, PFAS permeation rate J (mol·m-2·h-1) was calculated using Equation 4 and the concentrations were determined using LCMS.Equation 4: J = ((CP - C0)∙V) / (A∙t)

[0124] Cp(mol·L-1) is the current PFAS concentration on the permeate side, C0(mol·L-1) is the initial concentration of PFAS on the permeate side, V (L) is the volume of the permeate side solution, A is the effective membrane area (1.96 × 10-2m2), and t (h) is the diffusion time. Permeation rates were calculated after every 3hrs for the thinner membrane (~35nm) and after every 24 hrs. for thicker membranes (88nm and 118nm). After completing a measurement, the device was washed with DI water to remove any residual PFAS before the next measurement.

[0125] PFAS analysis

[0126] PFAS samples were analyzed using an Agilent 6540 Q-TOF MS system with an ESI source (Santa Clara, CA, USA) - Liquid chromatograph mass spectrometer (LCMS). The LC consisted of an Agilent 1260 series solvent degasser, binary pump, column oven and sample manager. An Agilent EclipsePlusC18 RRHD 1.8 µM (2.1 x 50 mm) column was used for the LC / MS study.

[0127] The mobile phase consisted of acetonitrile and H2O, both containing 0.1% v / v formic acid, and delivered using a binary gradient 10 to 100% acetonitrile over 5 min followed by a hold at 100% for 1 minutes then re-equilibration to the starting conditions. The flow rate was 0.3 mL min−1, and the injection volume was 1 μL. LC / MS analysis was conducted in negative mode and elution of the analyte monitored using the following MS conditions: nebulizer pressure 45 psi, gas flow-rate 10 L / min, gas temperature 300°C, capillary voltage 3000 V, fragmentor 100 and skimmer 65 V. Instrument was operated in the extended dynamic range mode with data collected in m / z range 100–1000. Data analysis was conducted with Agilent Mass Hunter Workstation Qualitative Analysis Version 10. Peak areas were integrated from the extracted ion chromatogram of the deprotonated [M-H] target mass, and in the case of carboxylic acid compounds [M-H] and the [M-COOH] fragment masses.

[0128] Transport mechanisms in CD-GO membranes

[0129] To assess the effectiveness of binding interactions on the enhancement of membrane retention, GO-αCD and GO-βCD membranes were systematically tested in challenge water solutions containing PFAS with a range of molar mass and carbon numbers (Cn) as shown in Table 4 below.Table 4. PFAS structures, lengths and molecular weights

[0130] Pristine GO membranes, as characterized by their laminar structured nanochannel networks, have been reported to separate PFAS with a chain length > 8 such as PFOA based on size exclusion. Typical routes for improving GO’s ability to retain PFAS relies on narrowing the interlayer spacing through thermal or chemical reduction pathways or modulating surface charge for enhanced electrostatic repulsive interactions. GO laminar structured membranes, with constricted interlayer spacing should be expected to improve rejection with the trade-off that the smaller interlayer galleries and unfavorable channel chemistry will drastically lower water permeance.

[0131] To overcome this selectivity-permeability challenge, an asymmetric nanochannel structure is adopted by intercalation of CD with the GO laminar structure which provides larger channels to enable fast water transport and smaller channels to enable enhanced selectivity towards PFAS contaminants.

[0132] Figure 6 is a graph comparing the retention of PFPeA and water permeance through the pristine GO and GO-CD membranes. The water permeance of GO-CD membranes was significantly improved by ~ 6-fold, compared to GO. The inventors attribute the highly enhanced water transport through GO-CD membranes to the expanded lamella. Meanwhile, the retention of PFPeA (C5) by GO-βCD membranes is 24.9% and 13.5% higher than that of the pristine GO and GO-αCD membranes respectively (~ 80.8 % for GO-βCD), noting that the association constant for βCD:PFPeA complexation, is 25-fold greater than αCD:PFOA, as summarised in Table 5 below.Table 5. Average Association Constants for α-, β-, and γ-CD with PFCAs

[0133] Molecular dynamics was used to probe the interactions between PFAS and CD and to provide further context to the transport mechanisms. PFAS molecules of different chain lengths may interact with cyclodextrin in varying ways. For instance, while the optimal energy configuration for βCD with shorter PFAS (C4, C5, and C6) occurs near the cyclodextrin's pore entrance, longer PFAS molecules (C7 and C8) tend to bind strongly within the cyclodextrin's pore, effectively blocking it. The simulations also indicate that PFAS exhibit strong binding affinity to the outside periphery of cyclodextrin, following a pattern similar to the binding observed within the pores for C4 to C8 PFAS.

[0134] Transition-state theory (TST) describes molecular aspects of the transport across nanochannels that cannot be accessed by the solution-diffusion model. It portrays transport within a nanochannel as a series of enthalpic and entropic energy barriers, situated both at the pore entrance and then as multiple energy barriers within the nanochannel, that must be overcome for molecules to permeate the membrane. TST provides insights into the molecular mechanisms of transport in materials containing sub-nanometer structures, making it valuable for understanding the hindered transport of PFAS in GO-CD architectures. The thermodynamic properties of PFPeA transport through the GO-βCD membranes were analyzed by temperature dependent measurements. The calculation of the PFAS permeation rate J (mol m-2h-1) is given below: Equation 5: J = ((CP-C0)∙V) / (A∙t)

[0135] Cp (mol L-1) is the permeate side PFAS concentration at time t, C0 (mol L-1) is the permeate side initial PFAS concentration, V (L) is the permeate side solution volume, A is the effective membrane area (1.96 × 10-2m2), t (h) is the diffusion time.

[0136] GO-βCD nanochannels with favorable binding sites for PFPeA rejection displayed lower diffusion rates of 1.6 – 4.3 x 10-8mol m-2s-1over a 283 – 303 K temperature range as displayed in Figure 7, which reports PFPeA (50 ppm) transport rates through the GO-βCD membrane. In comparison, the wider interlayer spacing of GO (~ 21.4 Å) in wet state may account for the faster PFPeA diffusion given the PFPeA size of 6.6 Å and the reduced number of interactions impeding their movement. Equation 6: ln(J / (CF-CP)) = lnB - Ea / RT

[0137] CF and CP are the PFAS concentrations on the feed and permeate side, respectively, B is the preexponential factor, R is the gas constant, and T is the absolute temperature. The activation energy barriers of PFAS transport can be calculated according to the Arrhenius equation (Equation 6), which describes the overall energy required to permeate into the nanochannel as well as the process of overcoming the complexation with CDs within the nanochannel. This theory was applied to analyse the transport of PFPeA through the GO-βCD, GO-αCD and pristine GO membranes in a concentration gradient-driven process across varied temperatures (between 283 and 315 K). Figure 8 is an Arrhenius plot for PFPeA transport in GO-⍺CD, GO-βCD and GO membranes illustrating the dependency of the overall energy barrier on CD type. Energy barriers for PFPeA transport through GO, GO-αCD and GO-βCD were 3.4, 5.1 and 6.5 kcal mol-1, respectively. This indicates that the GO-βCD offers the highest resistance to diffusion compared to the other membranes. GO-βCD’s higher energy barrier for PFPeA is consistent with our retention results, aligning with the hierarchy of selectivity. Although the Eareflects the dependence of solute transport on temperature change, it is directly correlated to rejection, where high rejecting membranes have been associated with higher activation energy, and vice versa.

[0138] The same temperature varied technique was used to analyse transport through GO-βCD membranes with different thicknesses. The results are shown in Figure 9 which illustrates an exponential growth relationship between the energy barriers to PFPeA transport and the thickness of the GO-βCD membrane. The energy barriers jump from 6.5 kcal·mol-1(34.8 ±2.5nm) to 8.6 kcal·mol-1(88. ± 1.2 nm) and reach 13.76 kcal·mol-1at a thickness of 117.8 ± 2.5nm. Membrane thickness adds an exponential cumulative diffusional resistance to PFPeA transport. This aligns with the expected behavior based on Equation 7 below, which is derived from Equation 2 and Fick’s law. Equation 7: Ea=-R∙T∙ln{(DS(CF- CP)) / (B∙L)}

[0139] This behaviour indicates that longer pathways result in PFPeA molecules encountering exponentially more resistance to transport. Furthermore, the number of binding sites within the GO-βCD layers increases, leading to greater compounding effects of the interactions, which doubles the Eawhen the thickness is increased by 3.4 times. Hence, the contribution of PFPeA transport resistance is largely governed by intrapore diffusion within GO-βCD channels.

[0140] PFAS molecules that would potentially diffuse around the perimeter of cyclodextrin molecules will also feel the interaction given that the CDs create a region of influence as opposed to CD acting as a single point reaction coordinate for the interaction of PFAS. However, unlike the sharp increase in binding energy observed for C7 and C8 PFAS, binding inside the pores, the binding energy increase for interactions around the sides of cyclodextrin is nearly linear. Figure 10 shows the average binding energy for PFAS molecules around the perimeter of CDs, with error bars indicating the range of energy levels for optimized structures at various positions around the outside of the cyclodextrin molecule. The binding strength of PFPeA near the pore of βCD is approximately 20% stronger (6.1 eV) compared to that with αCD (5.1 eV) as shown in Figure 11. When examining the binding near the cyclodextrin sides, PFPeA shows a ~29% stronger binding to the side of βCD than to αCD, with the strongest binding energy recorded at 4.9 eV for βCD versus only 3.8 eV for αCD. Overall, this represents approximately a 25% increase in binding strength for βCD compared to αCD.

[0141] GO-based nanofiltration relies on size sieving, electrostatic repulsive interactions, and adsorptive effects to facilitate retention. The most reliable of which over long-term filtration is size exclusion. The retention of PFAS (e.g., PFBA, PFPeA, PFHxA and PFOA) with molar mass below 450 Da was used to assess the reliance of GO and GO-βCD membranes on size exclusion as a mechanism of retention. The estimated hydrodynamic diameters of PFAS are summarised in Table 4 above.

[0142] The results are summarised in Figure 12, Figure 13, Figure 14, and Figure 15 which illustrate the influence of traditional retention mechanisms in GO-βCD membranes. Figure 12 shows the dependence of retention on steric effects by βCD modified GO membranes compared to pristine GO membranes. Figure 13 shows the dependence of retention on adsorption, and in particular, it can be seen from Figure 13 that that PFAS retentate concentrations increase by more than twice the feed concentration. Figure 14 shows the dependence of retention on electrostatic effects and variation of PFPeA retention by GO, GO-αCD and GO-βCD membranes with change in concentration. Figure 15 shows the dependence of retention on electrostatic effects and variation of PFPeA retention by GO, GO-αCD and GO-βCD membranes with change in pH.

[0143] As expected, pristine GO membranes demonstrate a decay in retention from 88.5 ± 2.0 % for PFOA (chain length 10.4 Å) to 32.1 ± 3.1 % for PFBA (chain length 5.4 Å) which has a molecular size smaller than that of pristine GO (8.0 Å) and expanded GO’s interlayer spacing (~ 21.39 Å). This observation demonstrates the strong reliance of retention on steric effects as well as their limitation in retaining small-chain PFAS.

[0144] The GO-βCD membrane demonstrates retentions above 77 % for all tested PFAS at feed concentration of 50 ppm. Figure 16 illustrates similar behaviour in retention of PFAS (at concentrations of 0.1 ppm) on GO-βCD. This is a remarkable rise in retention of ~ 45 % for PFBA, ~ 25 % for PFPeA, ~ 26 % for PFHxA and ~ 1 % for PFOA compared to that of pristine GO membranes. If transport through the smaller nanochannels present in the asymmetric structure were a controlling factor on PFAS retention, it would be expected that the retention of PFPeA by GO-αCD be similarly improved (Figure 6) to that of GO-βCD. The observation validates the influence of the role of intra-pore energy-barrier that occurs to better effect in GO- βCD. It is also observed that concentration of the retentates of PFBA, PFPeA, PFHxA and PFOA increased significantly from a feed concentration of ~ 50 ppm to 95 ± 6.5, 106 ± 5.8, 118 ± 3.5 and 113 ± 6.7 ppm respectively, for GO-βCD membranes (Figure 13). This suggests that adsorption is reduced compared to that of pristine GO membranes (Figure 16).

[0145] The separation of anionic PFAS by GO membranes is also influenced by electrostatic repulsive forces with the negatively charged carboxylate functional groups. Anionic PFAS have pKa of 0.34 for PFBA and PFPeA, 0.32 for PFHxA and 0.3 for PFOA. Hence, they remain in deprotonated state and are charged in neutral pH. To uncover the role of electrostatics, thedependence of GO, GO-αCD, and GO-βCD membrane retention was evaluated through varying the feed concentration of PFPeA and the pH of the water matrix (Figure 14, Figure 15). The results in Figure 14 show that GO-βCD membrane retention of PFPeA was relatively higher (above 81%) over feed concentrations of 0.1 − 50 ppm. Moreover, GO-βCD shows a weaker dependence on concentration which can be attributed to higher energy barriers in GO-βCD (Figure 7). In contrast, there is sharp decline in retention for GO, decreasing from 81.7% to 55.9%, over the varied concentrations evaluated (Figure 14). Thus, the design of GO-CD channels is responsible for high selectivity and improves at concentrations relevant to drinking water, where retention is greater than 89%.

[0146] To investigate the effect of pH on PFPeA retention, the solution pH was adjusted between 3 and 11, for an initial feed of 50 ppm as shown in Figure 15. Fluctuation of solution pH was reported to show a negligible effect on PFAS charge, as these toxins have very low pKa, and as such they remain deprotonated at pH levels greater than 3.4. However, the membrane surface can be sensitive to changes in pH. When the pH is increased, only a slight increase (within 5 %) in GO-βCD retention is observed from pH 4 to 6 and maintained at 80% from pH 6 to 10. The slight increase can be attributed to minimal electrostatic repulsion, but overall, there is no significant effect of pH on PFPeA retention. In comparison, when the effect of the pH dependency of retention of the GO membrane is evaluated, there is an increase in retention from 56 ─ 93% over a 3.8 ─ 10.7 pH range. As pH increases above the pKa 4.3 of carboxylic acid groups, carboxyl groups in GO deprotonate, which imparts a negative surface charge on GO membrane and enhances electrostatic repulsion of PFPeA and thereby increasing retention gradually with pH (Figure 15). Moreover, at pH levels above 6.1, deprotonation of hydroxyl groups (pKa 6.1─10) increases negative charge already imparted on GO which sharps raises the retention to 86 and 93% at pH 9.1 and 10.7 respectively. This agrees with literature, that GO’s negative surface charge account for the electrostatic repulsion mechanism with negatively charged solutes, which results in selectivity increasing with pH. This suggests that at the same PFPeA ionic strength this 23% increase in retention on the GO-βCD membrane has little to do with effect of electrostatics. The results indicate that the improved PFAS retention by GO-βCD has significantly reduced dependency on steric hindrance, electrostatics and adsorption mechanisms.

[0147] Performance of membranes and comparison with commercial membranes

[0148] To demonstrate the feasibility of using GO-βCD in practical long-term operations, the membrane was tested in a crossflow filtration system at a pressure of 2 bar and a flowrate of 1.0 L min-1using a 0.1 mg L−1PFPeA feed solution. The results are summarised in Figure 17, Figure 18, Figure 19, and Figure 20. Figure 17 shows the long term cross flow performance of GO-βCD for removal of PFPeA at concentrations relevant to drinking water (0.1 ppm). Figure 18 shows retention of mixtures of PFAS with an inset displaying concentration of the various PFAS in the feed, retentate, and permeate. Figure 19 shows comparative performance of GO- βCD for removing PFAS of different molar masses compared to commercial NF270, and their respective water permeances (inset). Figure 20 provides a performance comparison of permeance and selectivity of PFBA (C4) by representative polymeric membranes and GO-βCD.

[0149] Figure 17 shows that the initial permeance was 48 L m-2h-1bar-1which stabilised at ~ 33 L m-2 h-1bar-1after 30 minutes, maintaining a highly stable rejection of ~ 75 % over the 24 h period. This performance aligns with the low-swelling property observed when comparing the wet and dry state XRD spectra (refer to Figure 2). Linear trends of PFPeA permeation against time at different temperatures was established indicating a lack of adsorption.

[0150] During long-term crossflow filtration, the GO-βCD membrane is exposed to a constant shear stress imposed by the applied crossflow velocity, which may lessen pollutants accumulation and adsorption effects that contribute to reduced membrane permeance. Nonetheless, over the 24 h there was a gradual decline in permeance to ~ 25 L m-2h-1bar-1. To further demonstrate the practical feasibility of the GO-βCD membrane, a challenge solution containing a mixture of PFAS (12.5 mg L-1of each PFBA, PFPeA, PFHxA, and PFOA) was permeated through the membrane. Practical separations requiring the removal of components from a mixture are subject to competitive sorption effects that negatively influence retention. The membranes were able to retain ~ 89 % PFBA, 93 % PFPeA, 95 % PFHxA, and 99 % PFOA from the mixture of the four PFAS (as shown in Figure 18). This exceptional selectivity of GO- βCD is consistent with the molecular dynamics displayed in Figure 11. The sharp increase in binding energy observed from C4−C8 PFAS correlates well with the increased retentions from 89%−99% respectively, for the mixtures of PFAS.

[0151] The performance of GO-βCD was compared to that of NF270, a commercially available NF membrane with a pore diameter of ~ 0.8 nm. NF270 has been demonstrated to have appropriate retention for PFOA (~ 89 %); however, exhibits a significant decline in its capacityto retain short-chain PFAS, with retention dropping to 35 % for PFBA. Mass balance analysis for tests performed on N270 membrane shows that the concentration of the retentate increases by a factor of 1.6 and 1.3 over the feed for PFPeA and PFBA respectively, as shown in Figure 21, indicating that the retention that is observed is dominated by adsorption affects. The permeance-selectivity performance of the GO-βCD membrane was compared to polymeric- based NF membranes that have been investigated to retain PFBA. Figure 20 illustrates that a majority of available membranes had PFBA retentions less than 80 %, with permeances below 10 L m-2h-1bar-1. Furthermore, the only work reporting a permeance above 20 L m-2h-1bar-1relies on electrostatic repulsion as the dominant retention mechanism which limits its viability at higher concentrations due to charge shielding. The ability of a membrane to operate at high solute concentrations is vital in PFAS retention as traditional PFAS destructive techniques require concentrations as high as 150 ppm for efficient operation. Together, these comparisons illustrate the practicality of introducing compensatory interactions to hinder solute transport within nanochannels as a viable design strategy for long-term and large-scale applications. The results are summarised in Table 6 below. Table 6. Rejection and water permeances of polymeric NF membranes on short chain PFBA and PFBA comparison including NaCl removal using GO nanofiltration membranes

[0152] In summary, by exploring the ability of CDs to bind with PFAS, the inventors designed rGO membranes with asymmetrically sized nanochannels with strategically embedded sites capable of hindering the transport of PFAS within the GO lamella. The membrane was fabricated using shear alignment printing which is effective for preparing GO films on a large scale. The use of CD, and in particular βCD, enables a favourable physical environment for transport of water and maintains an appropriate chemical environment that ensures PFAS exclusion. GO-βCD membranes demonstrated exceptional retention of short-chain PFAS of over 90 % while also improving water permeance to 22 L m−2h−1bar−1. This improvement in perm- selectivity of βCD was also illustrated through the removal of 4 different PFAS at concentrations relevant to drinking water. In comparison to commercial membranes, GO-βCD overcomes the common permeance-rejection trade off displayed by NF270 membranes. The GO-βCD membranes displayed stable rejection above 75% and permeance above 25 L m−2h−1bar−1over a 24 h period. Perm-selectivity results and molecular dynamics simulations serve as evidence that the GO-βCD membrane has inherently restrictive PFAS transport due to its highly interactive binding sites. The energy barrier to PFPeA transport in GO-βCD membrane was found to be twice that of the GO membrane, as determined by temperature varied diffusion tests. The strategy of employing host guest interactions in membrane separations is useful in concentrating and removing the toxic PFAS with concentrations comparable to that of landfills, industrial effluent streams and waste-water treatment plants. Example 2 Effect of βCD loading on PFPeA retention and permeance

[0153] The effect of βCD loading within GO ink ranging from 0-90 wt. % was investigated and optimised based on PFPeA retention and permeance. Membranes were generally fabricated according to the method discussed above where βCD powders were incorporated into nematicphase GO inks at various weight ratios and reacted at 90 °C for 2 hours. The membranes had a thickness of 148 nm.

[0154] The results are shown in Figure 22 where it can be seen that optimal membranes were prepared with a 1.2:1 weight ratio of GO to βCD content, which translates to a βCD content of 45 wt.% in the GO ink.

[0155] In order to analyse the effect of βCD on structure and performance of the GO-βCD membrane, βCD loading was systematically varied, and the optimal loading of 45% resulted in a PFPeA retention of 80% (see Figure 22). Beyond this point, as the βCD loading exceeds 50%, the XRD patterns exhibit sharp, well-defined peaks characteristic of crystalline βCD as shown in Figure 23. This suggests that excess βCD molecules are no longer incorporated into the GO structure but instead form crystalline aggregates due to the predominance of βCD in the sample. These aggregates disrupt the interlayer arrangement GO-βCD and significantly compromises the membrane's retention performance, as observed in Figure 22. FTIR analysis on GO-βCD inks and change in functional groups over different reaction times

[0156] Figure 24 reports the FTIR analysis of GO-βCD inks with reaction times varying from 30 to 150 min at a βCD content of 45 wt.%. The FTIR of GO-βCD demonstrates a new peak at 2920 cm-1which is a characteristic of -CH2- stretching and confirms the functionalization GO with βCDs. The near-complete disappearance of the peak at 1720 cm-1in GO-βCD from 90−150 minutes indicates that the carboxylic groups on GO are being consumed during the reaction with βCD. This confirms the attachment of the hydroxyl groups (-OH) of βCD onto carboxyl groups (-COOH) of GO sheets through ester linkages. Similarly, the O-H (3000-3700 cm⁻¹) peak shifts and decreases in intensity as the reaction progresses (from 30 to 150 minutes). This reduction in the -OH peak intensity can be attributed to covalent interactions, such as esterification, which consume βCD's hydroxyl groups, as well as mild dehydration effects responsible for reduction. The emergence of additional between 500 cm−1and 1000 cm−1inclusive of the α-(1,4) glucopyranose peak at 800 cm−1was indicative of the successful attachment of CDs onto GO. Retention of GO-βCD membranes at 0.1 ppm of C4-C8 (carboxylated) and C8 (sulfonated) PFAS, and C5 in the presence of 50 ppm of humic acid

[0157] Figure 25 reports the performance of the GO-βCD membranes with a βCD content of 45 wt.% at 0.1 ppm of PFAS with chain length C4-C8and carboxylated or sulfonated head groups and in the presence of 50 ppm of humic acid.

[0158] The GO-βCD membrane demonstrates rejection values above 80% for all tested PFAS at a feed concentration of 0.1 ppm. Retention in the presence of humic acid improved to 97% from 94% in the absence of humic acid indicating that competition between humic acid and PFPeA interacting with the βCD does not compromise the efficiency of the membrane. PFPeA rejection over 24 hours of crossflow filtration on GO and GO-βCD membranes

[0159] Figure 26 reports long term 24-hour cross flow filtration rejection of PFPeA by GO- βCD membranes with a βCD content of 45 wt% compared to pristine GO as fabricated by gravure printing. Over a continuous 24-hour operation, GO-βCD consistently achieves PFPeA rejection rates 10–15% higher than GO underscoring the role of βCD functionalisation. The hydrophobic and host–guest interactions endowed by βCD preferentially hinders PFAS transport through the lamella GO-βCD, thereby increasing the energetic barrier required for their transport through membrane nanochannels. The sustained high rejection across extended operation highlights that these mechanisms are robust even under shear stress imposed by cross flow conditions, which can typically challenge membrane integrity and selectivity. PFPeA permeance over 24 hours of crossflow filtration on GO and GO-βCD membranes

[0160] Figure 27 reports long term 24-hour cross flow filtration permeance of PFPeA by GO- βCD membranes with a βCD content of 45 wt% compared to pristine GO. Accounting for membrane compaction, GO-βCD effectively preserves nanochannel accessibility, resulting in only a ~10% permeance loss over 24 hours, compared to the ~40% decline observed for GO (figure 12). Therefore, βCD integration not only enhances PFAS rejection through molecular binding chemistry but also ensures more stable membrane performance during extended filtration by maintaining open and efficient water transport pathways. PFAS rejection for GO–βCD, GO, and NF270 membranes during 24h crossflow operation

[0161] Figure 28 reports long term 24-hour cross flow filtration rejection of single component PFPeA, PFOA and PFOS solutions at 0.1 mg / L by GO-βCD membranes with a βCD content of 45 wt% compared to pristine GO and commercial NF270. As can be seen, the performance of the GO-βCD membranes is generally superior for the PFPeA, PFOA and PFOS solutions.Humic acid rejection for GO–βCD, GO, and NF270 membranes during 24h crossflow operation

[0162] Figure 29 reports long term 24-hour cross flow filtration rejection of 20 mg / L Humic Acid by GO-βCD membranes with a βCD content of 45 wt% compared to pristine GO and commercial NF270. The influence of natural organic matter (NOM) was examined using Humic Acid (HA at 20 mg / L) as a representative model compound. In the absence of PFAS, the GO– βCD membrane exhibited near-complete rejection of HA, surpassing GO (95.6%) and NF270 (96.4%). This performance is attributed to the strong hydrophobic interactions and selective binding sites provided by βCD, which effectively hinders HA transport. In contrast, GO and NF270 rely primarily on steric hindrance within their narrow channels and pores, together with electrostatic repulsion between their negatively charged surfaces and the functional groups of HA, resulting in comparatively lower HA rejection. This is expected to apply to broad spectrum NOM such fulvic acids, tannins, phenolics, and more. Effect of humic acid (HA) on PFAS rejection for GO–βCD, GO, and NF270 membranes during 24h crossflow operation

[0163] Figure 30 reports long term 24-hour cross flow filtration rejection of 0.1 mg / L PFPeA, PFOA and PFOS in the presence of 20 mg / L Humic Acid by GO-βCD membranes with a βCD content of 45 wt% compared to pristine GO and commercial NF270. Humic acid and PFAS form complexes through hydrophobic interactions, which cause an increase in steric retention through pristine GO and GO-BCD. However, steric effects do not fully account for the increase in PFAS rejection through the larger GO-BCD channels, this instead is attributed to the increase in affinitive interactions between CD and the humic acid-PFAS complex. This effect results in an elevated rejection of 99% by GO-βCD membranes not achievable by pristine GO. Ca2+rejection for GO–βCD, GO, and NF270 membranes during 24h crossflow operation

[0164] Figure 31 reports long term 24-hour cross flow filtration rejection of 2 mM calcium salt by GO-βCD membranes with a βCD content of 45 wt% compared to pristine GO and commercial NF270. GO–βCD exhibits a much lower rejection of Ca²⁺ (22%) compared with GO (81%) and NF270 (51%). This disparity can be explained by the interplay between ion size, hydration properties, and membrane structure. Ca²⁺ has a hydrated radius (~4.12 Å), which normally hinders transport through narrow GO membranes nanochannels, leading to higher Ca²⁺ rejection. However, βCD functionalization in GO–βCD expands the interlayer spacing to about 17.8 Å, creating sterically accommodating nanochannels that allow hydrated Ca²⁺ ions topermeate more easily, indicating minimal direct interaction between βCD cavities and hydrated ions. This is performance trend is anticipated to apply to broad spectrum multivalent ions such magnesium, iron (II), cobalt, nickel, copper, aluminium, iron (III) and more. Effect of Ca2+on PFAS rejection for GO–βCD, GO, and NF270 membranes during 24h crossflow operation

[0165] Figure 32 reports long term 24-hour cross flow filtration rejection of PFPeA, PFOA, and PFOS in the presence of 1 mM calcium salt by GO-βCD membranes with a βCD content of 45 wt% compared to pristine GO and commercial NF270. Upon the addition of Ca²⁺ (1 mM), rejection of all PFAS compounds improved across all membranes, with the GO-βCD membrane reaching 100% rejection for PFPeA and PFOS and 99.7% for PFOA. The improvement in rejection was notably higher than that observed for the GO membrane, where rejections increased to 91% (PFPeA), 94% (PFOA), and 99% (PFOS), and the commercial NF270 membrane, which showed 89%, 93%, and 100% rejection for PFPeA, PFOA, and PFOS, respectively. Calcium was used as a representative divalent ion; similar trends are to be expected with other divalent and multivalent ions. Permeance for GO–βCD, GO, and NF270 membranes during 24h crossflow operation

[0166] Figure 33 reports long term 24-hour cross flow filtration permeances of PFPeA, PFOA, and PFOS through GO-βCD membranes with a βCD content of 45 wt% compared to pristine GO and commercial NF270. As can be seen, the GO-βCD membranes maintain significantly higher permeance than the GO and NF270 membranes.

[0167] Figure 34 reports long term 24-hour cross flow filtration permeances of PFPeA, PFOA, and PFOS in the presence of Humic Acid through GO-βCD membranes with a βCD content of 45 wt% compared to pristine GO and commercial NF270. As can be seen, the GO-βCD membranes maintain significantly higher permeance than the GO and NF270 membranes.

[0168] Figure 35 reports long term 24-hour cross flow filtration permeances of PFPeA, PFOA, and PFOS in the presence of Calcium through GO-βCD membranes with a βCD content of 45 wt% compared to pristine GO and commercial NF270. As can be seen, the GO-βCD membranes maintain significantly higher permeance than the GO and NF270 membranes. PFPeA Rejection and corresponding permeance

[0169] Figure 36 reports long term 24-hour cross flow filtration rejection and permeances of PFPeA in the presence of varying calcium concentrations through GO-βCD membranes with a βCD content of 45 wt%.

[0170] Figure 37 reports long term 24-hour cross flow filtration rejection and permeances of PFPeA in the presence of varying humic acid concentrations through GO-βCD membranes with a βCD content of 45 wt%. Rejection of PFAS from multicomponent mixtures

[0171] Figure 38 reports long term 24-hour cross flow filtration rejection of carboxylated PFBA and PFOA and sulfonated PFBS and PFOS from mixtures at varying concentrations through GO-βCD membranes with a βCD content of 45 wt%.

[0172] Figure 39 reports long term cross flow rejection of PFAS from a multicomponent mixture of PFPeA, PFOA, PFOS, Humic acid and Calcium through GO-βCD membranes with a βCD content of 45 wt% compared to pristine GO. As can be seen, the performance of the GO- βCD membrane is significantly better than the GO membrane. Permeances of optimal GO-βCD and pristine GO membranes while treating multicomponent mixture of PFAS, NOM and inorganic ions

[0173] Figure 40 reports long term cross flow permeance for GO-βCD membranes with a βCD content of 45 wt% and pristine GO membranes treating a multicomponent mixture of PFPeA, PFOA, PFOS, Humic acid and calcium. As can be seen, the performance of the GO-βCD membrane is significantly better than the GO membrane. Fouling behaviour

[0174] The surface roughness of pristine GO membranes and GO-βCD membranes with a βCD content of 45 wt% were measured and found to be Rav of 38.1 nm and 38.4 nm and RRMS of 49.7 nm and 45.2 nm respectively. The water contact angle was also measured and found to be 34° and 21.5° respectively. The smoother surface and more hydrophilic surface of the GO-βCD membranes correlates with improved anti-fouling performance, as it provides fewer sites for foulants to adhere to and accumulate.

[0175] Figure 41 reports results from the analysis of the fouling mode of pristine GO and optimal GO-βCD membranes. The fitted trends based on the Hermia model describe the foulingmode that occurs in each of the pristine GO and GO-βCD membranes and reveals that GO-βCD membranes display a distinct fouling behaviour compared to unmodified GO, reflective of their altered surface chemistry and morphology. GO-βCD’s fouling pattern aligns closely with cake filtration dynamics, suggesting that fouling occurs through the gradual build-up of fouling layers on the membrane surface rather than rapid channel entrance occlusion. This supports the conclusion from the surface roughness, and membrane hydrophilicity in the improved antifouling behaviour of GO-βCD membranes. Example 3

[0176] Example 3 reports the use of GO-βCD membrane for treatment of real ground water from Port Melbourne (Victoria, Australia).

[0177] Table 7 below provides a summary of the water chemistry of three ground water samples GW02, GW20, and GW25 subjected to membrane water filtration using a GO-βCD with a βCD content of 45 wt%. Table 7: Port Melbourne (Victoria, Australia) groundwater raw feed water quality

[0178] The GO-βCD membrane effectively removed PFAS from Port Melbourne groundwater under the water chemistry conditions presented in Table 7. The presence of smaller ions in the groundwater promoted complexation with PFAS, which enhanced their overall rejection. This interaction not only strengthened the hydrophobic associations of PFAS with the membrane but also increased the exclusion of the ions themselves. As a result, both PFAS and coexisting ions were more efficiently removed in the tested groundwater matrix. Table 8: Ability of GO- βCD membranes to treat real ground water from port Melbourne sample / site (GW02, GW20, and GW25) following 24-hour cross flow filtration operating at a low water recovery of 10%.B: PFAS was not detected in feed samples; C: carryover contamination; ND: No PFAS detected in permeate.

[0179] The GO-βCD membrane demonstrated effective removal of both short-chain PFAS (< 365 Da, 78.1 – 99.7% rejection) and long-chain PFAS (>414 Da, 98.1 – 99.2% rejection) during 24 hours of cross-flow operation at low permeate recovery (0.1), through hydrophobic interactions between PFAS fluorinated tails and β-cyclodextrin cavities. All rejections for which PFAS concentrations in the permeate were below the limit of reporting (LOR) are indicated as ND. These interactions are further enhanced in more realistic water chemistries containing 400– 600 mg / L of major cations, which reduce electrostatic repulsion and promote stronger hydrophobic aggregation. Importantly, the membrane was especially effective in removing PFAS when the total organic carbon (TOC) was below 15 mg / L. These findings underscore the suitability of GO-βCD membranes for treating complex waters, where realistic cation and DOM levels actually enhance PFAS rejection performance compared to pure water systems.

[0180] When operating in cross flow at a high recovery of 90%, following a pre-treatment filtration step through ultrafiltration PVDF membrane to remove high level NOM, GO-βCD membranes were able to significantly reduce PFAS concentration in the permeate as well as concentration PFAS undetectable in the raw feed. Three different groundwater feed from port Melbourne were analysed under these conditions (GW02, GW20 and GW25). In GW02 (Table 9), PFPeS and 6:2 FTS were undetected in the feed and were concentration up to 200 times using the optimal GO-βCD membranes. Similarly, in GW20 (Table 10), 6:2 FTS was similarly concentrated from below detection to ~100 times detection limits. Finally, in GW25 (Table 11), PFDA, EtFOSAA, and 6:2 FTS were all observed in the high recovery retentate and present at concentrations below detection in the feed. Across all 3 water matrices, wherein 13-15 distinct PFAS were detected, the optimal GO- βCD membrane was able to achieve average retentions between 78.6 to 86.6% while operating at 90% recovery with individual PFAS rejections ranging from 44% to 99.9%. Observed low retention by GO- βCD membrane were typically associated with non-detection in the permeate. Table 9: Ability of GO- βCD membranes to treat real ground water from port Melbourne sample / site (GW02) operating at a water recovery of 90%*ND: Not detected in optimal GO-βCD membrane permeate. C: Equipment contamination.Table 10: Ability of GO- βCD membranes to treat real ground water from port Melbourne sample / site (GW20) operating at a water recovery of 90% Detected PFAS Detection ( µ g / L ) Limit ( µ g / L )L L LPFPeS 0.0005 0.0165 <0.0005 0.0547 99.09 (ND) PFHxS 0.0005 0.0669 0.0003 0.0775 99.61*ND: Not detected in optimal GO-βCD membrane permeate. C: Equipment contamination. Table 11: Ability of GO- βCD membranes to treat real ground water from port Melbourne sample / site (GW25) operating at a water recovery of 90%Detected PFAS Detection GW2 Raw Feedβ C DPermeate Retentate ( µ g / L )Limit ( / L ) conc ( / L ) Retention (%)*ND: Not detected in optimal GO-βCD membrane permeate. C: Equipment contamination.

[0181] It will be understood that the invention disclosed and defined in this specification extends to all alternative combinations of two or more of the individual features mentioned or evident from the text or drawings. All of these different combinations constitute various alternative aspects of the invention.

Claims

CLAIMS 1. A graphene oxide film comprising a laminar arrangement of graphene oxide lamella having cyclodextrin intercalated therein, the laminar arrangement having a first interlayer spacing and a second interlayer spacing larger than the first interlayer spacing, wherein the graphene oxide film has a cyclodextrin to graphene oxide mass ratio of less than 1:

1.

2. The graphene oxide film of claim 1, wherein the first interlayer spacing is provided by the laminar arrangement without the cyclodextrin and / or the second interlayer spacing is provided by the laminar arrangement having the cyclodextrin intercalated therein.

3. The graphene oxide film of claim 1 or 2, wherein the first interlayer spacing is from about 6 Å to about 8 Å and / or is represented by an X-ray diffraction peak from X-ray diffraction (XRD) analysis of the graphene oxide film at an angle of from about 12.5° to about 15.

4. The graphene oxide film of any one of claims 1 to 3, wherein when measured in a dry state: (i) the second interlayer spacing is from about about 8.5 Å to about 11 Å and / or is represented by an X-ray diffraction peak from XRD analysis of the graphene oxide film at an angle of from about 8.5° to about 10.5°; or (ii) the second interlayer spacing is from about 15 Å to about 18 Å and / or is represented by an X-ray diffraction peak from XRD analysis of the graphene oxide film at an angle of from about 4° to about 6°.

5. The graphene oxide film of claim 1 to 3, wherein the graphene oxide film further comprises a third interlayer spacing larger than the second inter layer spacing.

6. The graphene oxide film of claim 4, wherein when measured in a dry state: (i) the second interlayer spacing is from about 8.5 Å to about 11 Å and / or is represented by an X-ray diffraction peak of XRD analysis of the graphene oxide film at an angle of from about 8.5° to about 10.5°; or(ii) the third interlayer spacing is from about 15 Å to about 18 Å and / or is represented by an X-ray diffraction peak of XRD analysis of the graphene oxide film at an angle of from about 4° to about 6°.

7. The graphene oxide film, wherein the graphene oxide film has a carbon to oxygen (C:O) ratio of the graphene oxide layer is less than 2.25 when measured using X-Ray Photoelectron Spectroscopy (XPS).

8. The graphene oxide film of any one of the preceding claims, having a spectra, obtained using Fourier Transform Infrared Spectroscopy (FTIR) with: a O-C=O peak with a transmittance of greater than 10%; and / or a C-O peak with a transmittance of greater than 32%; and / or an -OH peak with a transmittance of greater than 70%; and / or a -CH2- peak with a transmittance of greater than 85%; and / or an α-(1,4) glucopyranose peaks with a transmittance of greater than 35%; and / or a -C=O peak with a transmittance of greater than 70%.

9. The graphene oxide film of any one of the preceding claims, wherein at least a portion of the cyclodextrin is covalently bonded to the graphene oxide.

10. The graphene oxide film of claim 9, wherein the portion of the cyclodextrin that is covalently bonded to the graphene oxide is covalently bonded via an ester bond.

11. The graphene oxide film of any one of the preceding claims, wherein the cyclodextrin is selected from the group consisting of α-cyclodextrin, β-cyclodextrin, γ-cyclodextrin, and mixtures of two or more of the foregoing.

12. The graphene oxide film of any one of the preceding claims, wherein the graphene oxide lamella is in the form of particles, such as plates, flakes, discs, or the like, having a long axis, and the long axis is substantially aligned with an axis in a plane of the graphene oxide film.

13. The graphene oxide film of any one of the preceding claims, wherein the film has a permeance of at least 20 L m−2h−1bar−1determined at an applied pressure of 2 bar and ambient temperature.

14. A separation membrane comprising a membrane layer formed from a graphene oxide film of any one of the preceding claims.

15. The separation membrane of claim 14, wherein the separation membrane comprises a porous support layer with the graphene oxide film applied thereto.

16. A water treatment method comprising transporting water through a graphene oxide film according to any one of claims 1 to 13 or a separation membrane according to claims 14 or 15 to remove one or more contaminants from the water and provide a filtrate substantially free of the one or more contaminants and / or with reduced concentration of the one or more contaminants.

17. The water treatment method of claim 16, wherein the one or more contaminants is selected from the group consisting of heavy metals and / or per- or poly-fluoroalkyl compounds (PFAS), such as perfluorobutanoic acid (PFBA), perfluoropentanoic acid (PFPeA), perfluorohexanoic acid (PFHxA), perfluorooctanoic acid (PFOA), and combinations thereof.

18. An aqueous graphene oxide ink composition comprising cyclodextrin and graphene oxide in a mass ratio of less than 1:

1.

19. The aqueous graphene oxide ink composition, wherein the graphene oxide is in an amount of at least 5 g / L.

20. A method of forming a graphene oxide film, the method comprising solution casting the aqueous graphene ink composition of claim 18 or 19 onto a substrate surface and subjecting the aqueous graphene ink composition to shear forces sufficient to shear align the graphene oxide particles and form a graphene oxide film.

Images