Maintaining Glycerol-Based Hexagonal Structures by Crosslinkers for High Permeability Nanofiltration

Abstract

Hypothesis: Structural optimization of crosslinkers within a reactive glycerol-based hexagonal lyotropic liquid crystal (HLLC) system is proposed to enhance the interfacial stability of hexagonal mesophases and improve the hexagonal structure retention during polymerization. This targeted modification is anticipated to significantly improve the water filtration efficiency of HLLC-templated nanofiltration. Experiments: The effect of crosslinkers on the interfacial stability of glycerol-based hexagonal mesophases was studied by evaluating their concentration accommodation within the mesophases using ¹³C solid NMR, FTIR and SAXS. Findings: A hydrophilic crosslinker consisting of ten ethylene glycol units shows less interference with the interfacial stability of hexagonal mesophases, therefore contributing to a higher concentration accommodation compared to the one with three ethylene glycol units. This long-chain crosslinker, despite having a low content of reactive groups, effectively connects the cylinders and better retains the hexagonal structures during polymerization than the hydrophobic crosslinker with shorter ethylene glycol units but a higher content of reactive groups. The retained hexagonal nanofiltration membranes show a remarkable pure water permeability of 40 L m⁻² h⁻¹ bar⁻¹ µm, resulting from the strong hygroscopic effect of glycerol and the crumpled surface of membranes due to the flexible nature of the system plasticized by glycerol.

1. Introduction

Nanofiltration membranes synthesized through hexagonal lyotropic liquid crystals (HLLC) hold promise for bolstering membrane efficiency through their unique blend of superior selectivity and permeability [1,2,3,4,5]. Despite their ability to generate tunable nanopore geometries and high pore densities, lyotropic liquid crystals (LLC) face significant stability challenges in aqueous mesophases, limiting their implementation in water filtration membranes. Studies indicate that incorporating polar organic solvents can stabilize LLC mesophases in diverse surfactant and lipid systems [6,7,8,9]. Compared to other solvents, glycerol, a nontoxic water miscible solvent with low volatility (normal boiling point 290 °C; vapor pressure @ 20 °C < 1 Torr) [1] has been used as pure solvent or with water to engineer higher-stability LLC mesophases. However, because of the lower polarity of glycerol than water, mesophases, such as hexagonal, cubic or lamellar ones, usually need to be formed in temperature ranges much higher than room temperature [10,11,12,13,14]. This could increase the difficulties for later membrane fabrication, since many polymeric porous substrates used for supporting the LLC-templated active layer are sensitive to temperature.

Glycerol-based LLC systems have been developed at room temperature and showed better performance compared to water-based ones [15,16]. Nonionic ethoxylated phytosterol (BPS) surfactants with various oxyethylene unit lengths were reported to form lamellar mesophase with glycerol at room temperature. It was found that the BPS/glycerol systems possess wider concentration ranges of LLC mesophases and higher viscosity than BPS/water systems [15], which is because the H-bonding mean lifetime of glycerol is longer than that of water [17,18]. Shukla et al. reported the formation of hexagonal and lamellar phase by sodium dodecyl sulphate (SDS) and glycerol at room temperature [16]. Compared to water based-LLC mesophases, glycerol-based LLC mesophases are more ordered and compact in the literature due to the stronger solvophobic effect caused by the longer H-bonding mean lifetime [19].

Low-volatility glycerol-based LLC systems are promising for fabricating (solution-based processing) nanofiltration membranes with good LLC-templated structure retention [6,7]. Nanofiltration membranes with cubic mesophase (Q_I)-templated structures were fabricated at 70 °C using an alkydiene-imidazolium bromide-based gemini surfactant/glycerol system. This Q_I membrane shows a remarkable MgCl₂ rejection of 99% but a low thickness-normalized permeability of 0.066 L m⁻² h⁻¹ bar⁻¹ µm [9]. Osuji et al. fabricated glycerol-based HLLC-templated membranes formed by mono-acrylate quaternary surfactants at room temperature. These membranes showcase tunable pore sizes in a range of 0.6–1.5 nm and a water permeability higher than 2 L m⁻² h⁻¹ bar⁻¹ μm [1,7]. The synthesis of mono-acrylate quaternary surfactants involves only one step [2,7]. These kinds of surfmers were reported to form hexagonal mesophases with glycerol easily at room temperature [1,7], and the hexagonal structures can be retained with the assistance of crosslinkers [2].

The extensive chemical diversity of commercially available crosslinkers yields a substantial combinatorial space of potential molecular candidates [20,21]. However, molecular-level optimization strategies for glycerol-based HLLC template retention remain unexplored. Insight into the structural effect of crosslinkers on the interfacial stability of glycerol-based hexagonal structures before polymerization and the retention of hexagonal structures during polymerization is crucial to optimize the HLLC-based nanofiltration membranes.

Herein, the HLLC system of polymerizable surfactant 2-(acryloyloxy) ethyl dodecyl dimethyl ammonium bromide (AED) (Figure 1) and glycerol acts as the binary template [1]. It was found that the AED/glycerol system possesses a broad range of hexagonal phases. The pure hexagonal binary system with the highest concentration of reactive groups was filtered. Crosslinkers were strategically incorporated into the binary system to modulate macroscopic phase behavior and stabilize the template architecture during polymerization. Selection criteria prioritized (i) hydrophilic character, governing crosslinker localization within the mesophase, and (ii) molecular length, directly controlling crosslinking density. The subtle variations in molecular structures are responsible for the interfacial stability of the formed ternary hexagonal mesophases. HLLC structure retention was achieved by introducing a specific hydrophilic crosslinker, which mediated the structure retention mechanism. The variations in the ordering and unit cell sizes of HLLC structures are responsible for the transparent and rejection properties. The variations in the surface structures of HLLC-templated membranes are responsible for high water permeability.

2. Materials and Methods

2.1. Materials

The following reagents were sourced from Sigma-Aldrich and used as received (inhibitor was removed before using): (i) Crosslinkers: Poly (ethylene glycol) diacrylate (PEGDA250, Mₙ = 250 g/mol; PEGDA575, M_n = 575 g/mol), pentaerythritol tetraacrylate (PETA); (ii) Photoinitiator: 2-Hydroxy-2-methylpropiophenone (HMPP); (iii) Surfactant precursors: 2-(Dimethylamino)ethyl acrylate and 1-bromododecane (subsequently copolymerized to form quaternary ammonium surfactant AED). The quaternary ammonium surfactant AED was prepared using modified literature procedures, with full synthetic characterization provided in Section S1 [1,7].

2.2. LLC Formation and Crosslinking

Homogeneous LLC precursor mixtures containing surfactant, crosslinker, glycerol and photoinitiator (0.5 wt% relative to total solids) were prepared in amber vials. Thermal processing at 70–80 °C facilitated mesophase formation, with intermittent vortex mixing and centrifugation accelerating homogenization. LLC sample compositions are shown in Table S1.

A 365 nm UV source was employed to cure samples [22]. Bulk HLLC samples were UV-cured (20 min) between fluorinated ethylene propylene (FEP) spacers (0.15 mm thick) positioned 1 cm from the UV source. For thin-film composite (TFC) membranes, HLLC precursor gel was mechanically cast onto polyacrylonitrile ultrafiltration (UPAN) substrates using polyester films (silicon oil coated) as spacers, followed by N₂-purging (10 min) and UV-curing (365 nm, 20 min) under continuous N₂ flow. The TFC membranes were marked as AED76@PAN and AED76P2@PAN (AED76 and AED76P2 active layers after polymerization on the UPAN substrate).

2.3. Characterizations

SAXS measurements were conducted by using a Xeuss 3.0 system (Xenocs, Grenoble, France) to identify the lyotropic liquid crystal (LLC) phase structure pre and post polymerization. Around 50 mg of sample was loaded in a SAXS cell and covered by a thin Kapton film. A dual-source configuration was utilized: a GeniX3D Ga source (Deakin-CSIRO, λ = 1.34 Å) and a Cu source (ANSTO, λ = 1.54 Å). Data were collected at a sample-to-detector distance of 350 mm, covering a scattering vector range (q) of ~0.1 to 0.75 Å⁻¹. Exposure times were set between 300 s and 1800 s. Two-dimensional scattering patterns were acquired in line-eraser mode and subsequently calibrated using silver behenate, then radially averaged to yield one-dimensional intensity profiles. Bragg peaks were subjected to Gaussian fitting analysis using XSACT (v2.7) software (Xenocs) to extract scattered intensity (I) as a function of q.

The cross-sectional morphology of the HLLC layer was observed using scanning electron microscopy (SEM; Zeiss Supra 55VP, Overkochen, Germany). Samples were cryofractured in liquid nitrogen and sputter-coated with platinum prior to SEM imaging. Triplicate measurements were performed on three distinct specimens to ensure representative characterization.

The evolution of birefringent textures in the HLLC was characterized using polarized optical microscopy (POM; Nikon Eclipse 80i, Tokyo, Japan).

The tensile mechanical properties of polymerized LLC-templated materials were determined using an Instron 5967 universal testing system. Rectangular specimens (nominal dimensions: 18 mm × 4 mm) were tested at a crosshead speed of 5 mm/min, employing a 5 N load cell and mechanical grips.

¹³C solid-state nuclear magnetic resonance (NMR) spectra were acquired on a Bruker Avance III spectrometer (11.7 T) (Billerica, MA, USA) using a static 5 mm probe. Measurements were performed without ¹H decoupling employing the following parameters: recycle delay = 1 s, pw90 = 7 μs (at 100 W power). Samples were loaded into 5 mm outer diameter NMR tubes (length: 3–4 cm) and sealed with laboratory film and caps prior to analysis.

UV-Vis transmittance spectra of thin HLLC films were recorded on an Agilent Cary 5000 (Santa Clara, CA, USA) spectrophotometer over 400–800 nm.

Polymerization of the HLLC membranes was characterized by Fourier transform infrared spectroscopy (FTIR) using a Bruker Vertex 70 (Billerica, MA, USA) spectrometer equipped with an ATR accessory. Spectra were acquired at 2 cm⁻¹ resolution with 64 co-added scans across 4000–400 cm⁻¹.

Water permeability and ion rejection of TFC membranes were characterized using a crossflow nanofiltration system (Saifei SF-SA, Hangzhou, China) with an effective membrane area of 7.07 cm². Following standardized preconditioning (5.5 bar, 5 min), permeability measurements were conducted at 25 °C and 5.5 bar transmembrane pressure. Deionized water-wetted membranes were securely mounted in the test cell prior to all measurements. Water permeance (J) was calculated using Equation (1) [23]:

$$J={\displaystyle \frac{\Delta w}{\rho A\Delta tP}}$$

(1)

where permeability parameters are defined as follows: Δw: mass of permeate collected (g); Δt: collection time interval (h); A: effective membrane area (m²); ρ: permeate density (g/L); P: applied transmembrane pressure (bar).

Solute rejection (R) was computed using Equation (2) [23]:

$$R_1=\left(1-{\displaystyle \frac{C_{\mathrm{p}}}{C_{\mathrm{f}}}}\right)\times 100\%$$

(2)

where C_p and C_f represent permeate and feed conductivities, respectively.

3. Results and Discussion

3.1. The Effect of Crosslinkers on the Interfacial Stability of the HLLC Mesophases Before Polymerization

The structural optimization of crosslinkers can enhance the interfacial stability of hexagonal mesophases before polymerization, which determines the concentration accommodation of crosslinkers in HLLC systems. SAXS and polarized optical microscopy have been used to study the mesophase structure of AED/glycerol systems (Figure S3). The primary peak (q = ~0.2 Å⁻¹) corresponding to the HLLC phase is significant in most of the samples (from 55 to 94 wt% of AED), indicating that all of them possess hexagonal phase. The hexagonal phase formation for the same systems has been verified before [1]. However, the 1D SAXS patterns of the AED55 (AED/glycerol = 55/45, w/w), AED60 (AED/glycerol = 60/40, w/w) and AED65 (AED/glycerol = 65/35, w/w) samples show a two-phase system, where a broad isotropic peak appears with the primary peak of HLLC. Moreover, a two-phase system appears again for samples AED88 (AED/glycerol = 88/12, w/w) and AED94 (AED/glycerol = 94/6, w/w), where AED crystal peaks appear in 1D SAXS patterns. Therefore, AED71 (AED/glycerol = 71/29, w/w), AED76 (AED/glycerol = 76/24, w/w) and AED82 (AED/glycerol = 82/18, w/w) are pure hexagonal phases identified from SAXS patterns. Additionally, from POM, more birefringence textures corresponding to hexagonal phase appear from adding 55 to 66 wt%, indicating the increasing proportion of hexagonal phase in the system. Crystalline textures start to appear in the AED82 system, indicating the coexistence of hexagonal phase and crystal in this system. The AED crystal amount in AED82 may be so small that the corresponding Bragg peak could not be very significant in terms of SAXS pattern. Therefore, only AED71 and AED76 are single hexagonal phase systems.

Since higher concentrations of reactive groups in the HLLC precursor can better retain the HLLC structure during polymerization, the AED76 binary system has been used to form a ternary system with crosslinkers. Three crosslinkers (Figure 2A) with various molecular lengths and hydrophilicities were introduced into the AED76 system individually. The introduction of crosslinker exerts great effects on the stability of HLLC phase as shown in Figure 2 and Figure 3. It is found that with the increasing concentration of crosslinker, the intensity of primary peaks becomes lower and broad isotropic peaks become more significant, indicating the gradual destabilization of mesophase interfaces and the transition from a single hexagonal phase to a two-phase system. For the AED76 systems with PETA or PEGDA575, the intensity of hexagonal primary peaks decreases significantly and the broad isotropic peaks appear when adding more than 2 wt% of crosslinkers, while for the AED76 system with PEGDA250, this phenomenon occurs when adding 2 wt% of PEGDA250. Therefore, PEGDA575 and PETA can better stabilize the interfaces than PEGDA250 and the AED76 system with stable hexagonal mesophase formation can contain 2 wt% of PEGDA575 or PETA. For the AED76P2 and AED76PE2 systems, the lattice parameter has no change after adding PEGDA or PETA into the system, as shown in

Table 1. Dimensional elements of AED76, AED76P2 and AED76PE2 systems (D₁₀₀: lattice parameter, D_inter: the inter-channel distance between adjacent lipid cores, R: the radius of the lipid core, R_w: the thickness of water and head group, and S: the interfacial area per hydrophilic head group of the surfactant). The calculation of the parameters follows previously reported protocols [24].

Sample	Q100 (Å−1)	D100 (Å)	Dinter (Å)	R (Å)	Rw (Å)	S (Å2)
AED76 before UV	0.2022	31.0583	35.8631	16.7495	2.3639	76.6915
AED76P2 before UV	0.2022	31.0583	35.8631	16.7313	2.4003	76.7750
AED76PE2 before UV	0.2022	31.0583	35.8631	16.3392	3.1847	78.6177

. However, the radius of lipid core R decreases and the interfacial area per hydrophilic head group of the surfactant S increases after incorporating crosslinkers, especially for the system with PETA. These variations indicate the increasing interfacial curvature of cylinders after the addition of crosslinkers, thereby resulting in a looser and more unstable packing of the surfactant molecules. The much smaller R and larger S of sample AED76PE2 (AED/glycerol/PETA = 76/22/2, w/w) compared to AED76P2 (AED/glycerol/PEGDA575 = 76/22/2, w/w) suggests that low-polarity PETA exerts greater effects on the packing of surfactant molecules.

From the POM images shown in Figure 2, uniform birefringent textures can be found in the AED76PE2 and AED76P2 systems, while some areas become non-birefringent after adding more than 2 wt% of PEGDA575 or PETA into the AED76 system. These results are consistent with the SAXS results that AED76 can load 2 wt% of PEGDA575 or PETA with single hexagonal phase formation. Significant non-birefringence is found in the AED76PL2 (AED/glycerol/PEGDA250 = 76/22/2, w/w) system, meaning the destabilization of mesophase interfaces when loading 2 wt% of PEGDA250.

¹³C solid-state NMR can further detect the crosslinkers’ effect on the interfacial stability of the AED76 system by studying the interactions between the molecules in the systems, as shown in Figure 4 [5]. Peaks ‘a’ and ‘b’ (the peak splitting is due to the spin/spin coupling between ¹³C and ¹H, since ¹H decoupling was not possible during the experiment) represent carbons in the reactive alkenyl group, while peak ‘c’ corresponds to the carbon in the carbonyl group. It is interesting to find that all the peaks become narrower and more significant after adding either PETA or PEGDA575 crosslinker, indicating that the introduction of crosslinkers leads to more liquid-like and less ordered HLLC systems. Moreover, no significant peak intensity changes occurred when adding more than 2 wt% of crosslinkers for both crosslinker systems, indicating that the extra crosslinker amount of more than 2 wt% is partitioned in the isotropic phase and does not have more interactions with AED surfactants.

Additionally, it is seen that the intensity of peaks corresponding to the carbons around quaternary ammonium groups gradually decreases with the increasing concentration of crosslinkers, which is due to the decreasing concentration of glycerol after adding crosslinkers. Quaternary ammonium groups are in the hydrophilic/hydrophobic interfaces. The formation of hydrogen bonding among the quaternary ammonium groups of surfactants and the hydroxyl groups of the glycerol can de-shield the carbons near nitrogen atoms [25,26]. A lower concentration of glycerol leads to more significant chemical shielding of these carbons, which induces the gradual decrease in peak intensity.

FTIR measurements further confirm the interfacial stability of AED76 systems with various crosslinkers and indicate the positioning of the crosslinkers. Figure 4C,F present the FTIR spectra of AED76 systems with an increasing amount of PEGDA or PETA. The broad absorption band centered at ~3350 cm⁻¹ arises from O-H stretching vibrations involving intra- and intermolecular hydrogen bonds between AED molecules and glycerol.

For the system with PEGDA, the peak area of -OH decreases (Figure S4) when adding 2 wt% of PEGDA into the AED76 system, indicating that the hydrogen bond interaction between glycerol and the head group of AED is weakened. It is interesting to find that the peak intensity of -OH becomes higher again after adding 3 or 4 wt% of PEGDA, indicating that the hydrogen bond interaction becomes stronger again. The peak intensity decreases and weakened hydrogen bond interaction can be attributed to the fact that the mesophase system retains its hexagonal structure when adding 2 wt% of PEGDA. When the mesophase becomes isotropic, the hydrogen bond interactions between the head group of AED and glycerol become stronger after adding more than 2 wt% of PEGDA.

In the systems with PETA, the peak intensity of -OH becomes lower with increasing concentration of PETA. Compared to PEGDA, hydrophobic PETA is more likely to be distributed close to hydrophilic/hydrophobic interfaces. The hydrogen bond interactions between the AED head groups and glycerol gradually become weaker, since the interactions between the PETA and AED become stronger with the increasing concentration of PETA. The interactions between the PETA and AED become even more significant in an isotropic state due to the lower steric hindrance.

3.2. The Role of Crosslinkers in the Retention of Hexagonal Structures and Other Properties of the Glycerol-Based HLLC-Templated Membranes

The introduction of structural optimized crosslinker enhances the retention of hexagonal structures, which improves the mechanical properties, transparency and water filtration performances of the glycerol-based HLLC-templated membranes.

The optimized interfacial stability of hexagonal mesophases increases the concentration accommodation of crosslinkers in AOE75H system, which results in enhanced retention of hexagonal structures and improved water filtration performances of the HLLC-templated membranes. From the SAXS pattern (Figure 5), the AED76P2 system after polymerization shows q₁₀₀, q₁₁₀ and q₂₀₀ (1:3^1/2:2) Bragg peaks corresponding to hexagonal phase, while the other two systems AED76 and AED76PE2 only show q₁₀₀ and q₂₀₀ (1:2) peaks. The significant q₁₁₀ peak indicates that the AED76P2 can better retain the hexagonal structures after polymerization than AED76 and AED76PE2 [24,27,28]. The Gaussian fittings, center positions and FWHM of the scattering peaks are shown in Figure S5 and Table S3. High monomer conversion for AED76, AED76P2 and AED76PE2 during polymerization is shown in Figure S6.

Figure 6A shows the crosslinkers’ effect on the mechanical properties of the AED76 systems after polymerization. It is interesting to find that samples AED76 and AED76P2 have distinctly different stress/elongation curves. AED76 shows a low strength and very flexible behavior. AED76P2 also shows flexible behavior but a significant increase in mechanical strength due to the assistance of PEGDA575. The flexible behavior can be attributed to the plasticizing effect of glycerol, since there could be a small weight of glycerol remaining in the dead pores of both samples. The remaining glycerol in the membrane after polymerization and water swelling has been calculated (Section S2). If the glycerol is completely removed, the empty water channels in the membranes should be filled by water after water swelling. The experimental weight (0.0858 g) of the swelled AED76 system is apparently higher than the theoretical weight (0.0845 g), while these two weights are similar for the AED76P2 system (0.0741 vs. 0.0742 g). The higher experimental weight of the AED76 system indicates there is a considerable weight of unremoved glycerol in the system after water swelling, which is much less in AED76P2. This also suggests that the retained HLLC structure of the AED76P2 system contains more open pores. Furthermore, AED76P2 bulk material after polymerization becomes more transparent than AED76 (Figure 6B). The low visible light transmittance could be attributed to the phase rearrangement of the AED76 sample during UV-curing, which leads to different indexes of refraction in the system, while the better structure retention of the AED76P2 system results in uniform refraction indexes in the system.

The ion rejection performance of HLLC template active layers increased significantly after introducing the crosslinker, as shown in Figure 6C. The ion rejection rates of AED76P2@UPAN (NaCl: 68.3% and MgCl₂: 80%) are much larger than that of AED76@UPAN (NaCl: 14.9% and MgCl₂: 23.2%), which could be because the AED76P2 active layer (6.9 Å) has a much smaller estimated water pathway width than the AED76 active layer (13.09 Å) (Section S3). AED76P2 can theoretically reject all the hydrated Na⁺ (7.2 Å). However, the soft nature of the AED76P2 material and the possible defects within the membrane could limit the rejection performance. The AED76 active layer still shows a 14.9% rejection rate of Na⁺, although the estimated water pathway width surpasses the hydrated size of Na⁺. This can be attributed to the electrostatic repulsion of quaternary ammonium cations. The thickness-normalized permeabilities of AED76@UPAN and AED76P2@UPAN are ~190 L m⁻² h⁻¹ bar⁻¹ µm and ~40 L m⁻² h⁻¹ bar⁻¹ µm (active layer thicknesses are shown in Figure S7), respectively. It is anticipated that the AED76P2@UPAN system can achieve a high water permeance of 200 L m⁻² h⁻¹ bar⁻¹ when the active layer thickness is reduced to approximately 0.2 μm, which is significantly higher than the permeability typically found in commercial nanofiltration membranes (NF70-270, ~10–15 L m⁻² h⁻¹ bar⁻¹) [29,30]. This permeability also compares favorably to similar systems, such as 100 L m⁻² h⁻¹ bar⁻¹ reported by Osuji et al. [2]. The high water permeability can be attributed to the crumpled membrane surface with high specific surface area [23,31,32] as shown in the Figure 6D,E. This surface morphology can be caused by the flexibility of the AED76P2 material. The surface roughness of the AED76 membrane (Ra = 61.1 nm) is twice that of AED76P2 (30.2 nm), which is attributed to the weaker crosslinked network of and higher unremoved glycerol in the AED76 system. Additionally, the hygroscopic effect of glycerol can also enhance the permeability of the systems [33]. This results in the much higher permeability of the AED76 system than AED76P2.

4. Conclusions

Previous studies have made significant advancements in using crosslinkers to assist the retention of reactive glycerol-based hexagonal structures [6,7]. Osuji et al. [7] found that glycerol-based hexagonal mesophase structures formed by mono-acrylate quaternary surfactants can be well retained through enhancing the chain entanglement at hydrophobic domains by incorporating a hydrophobic crosslinker. In this study, we investigated the effect of crosslinker structure on the interfacial stability of reactive glycerol-based HLLC mesophases as well as the concentration accommodation of crosslinkers within the mesophase through tuning the molecular length and hydrophilicity of crosslinkers. The direct effect originating from these variations in concentration accommodation on the retention of hexagonal structure and the water filtration performance of the glycerol-based HLLC-templated nanofiltration membranes were further studied.

Our key findings reaffirm the following:

• The poly (ethylene glycol) diacrylate (PEGDA) with ten ethylene glycol units exhibits high hydrophilicity, which minimizes the interference with the interfacial stability of hexagonal mesophases and enhances the concentration accommodation of crosslinkers.
• The hydrophilic PEGDA, with a long chain but low concentration of reactive groups, better connects the cylinders and retains the hexagonal structures compared to the hydrophobic one with shorter ethylene glycol units but a high concentration of reactive groups.
• The well-retained glycerol hexagonal structured nanofiltration membranes exhibit a remarkable pure water permeability of 40 L m⁻² h⁻¹ bar⁻¹ µm, which is attributed to the strong hygroscopic effect of glycerol and the crumpled surface of membranes due to the flexible nature of the system plasticized by glycerol. The permeability of the membranes is significantly higher than that of commercial membranes [29,30] and compares favorably to similar systems [2].

This work provides critical mechanistic insights into how crosslinker architecture governs interfacial stabilization in glycerol-derived hexagonal mesophases, while demonstrating a rational molecular design pathway for optimizing HLLC-templated nanofiltration membranes.