Boosting membrane carbon capture via multifaceted polyphenol-mediated soldering

Persistently high levels of atmospheric CO2 drive climate change and threaten targets of the Paris Agreement. Membrane-based gas separation is attractive for carbon capture due to cost-effectiveness, compact footprint, environmental friendliness, and ease of operation, but conventional polymer membranes face a permeability–selectivity trade-off that limits performance. Mixed-matrix membranes (MMMs) incorporating MOFs into polymers can surpass this trade-off, yet polymer–filler interfacial mismatch often creates nonselective voids that reduce selectivity. Strategies to improve compatibility include surface functionalization, morphology regulation, and ligand exchange, but in highly permeable, ladder-type PIMs, strong interactions can rigidify chains, improving selectivity and aging resistance at the expense of free volume and permeability. The authors propose a polyphenol-mediated molecular soldering strategy that simultaneously strengthens polymer/MOF interfaces and creates hollow MOF architectures to reduce mass transfer resistance, aiming to overcome the permeability–selectivity trade-off and improve stability against aging and plasticization.

The study builds on extensive work in MMMs and PIM membranes. Prior research has shown MOFs as promising MMM fillers due to tunable topology and microporosity, but poor interfacial compatibility with polymers often leads to nonselective voids and performance losses. Approaches such as covalent grafting, ligand exchange, and functionalization (e.g., UiO-66 derivatives, amine-functionalized ZIFs) improved adhesion but frequently rigidified the polymer matrix, increasing selectivity while sacrificing permeability, especially in highly permeable PIM-1. ZIF-8/PIM-1 is a widely studied system but suffers from interfacial defects and trade-offs at higher loadings. The concept of hollow MOF structures has emerged to provide internal free volume and reduced diffusion resistance. The authors integrate these ideas by using polyphenols (tannic acid) to both tailor MOF surfaces and induce hollowing, functioning as molecular solder to enhance adhesion (via hydrogen bonding, π–π interactions, metal coordination) and simultaneously lower mass transfer resistance.

Materials: ZIF-8 synthesized by mixing methanolic solutions of zinc acetate dihydrate (175 mg in 20 mL) and 2-methylimidazole (263 mg in 20 mL), stirring 5 min, aging at 30 °C for 24 h, centrifuging, washing with methanol, and activating at 100 °C overnight. HZIF-8 prepared by dispersing ZIF-8 (10 mg) in 5 mL water, adding 5 mL tannic acid (TA) solution (10 g/L), standing 5 min, centrifuging, and washing. PIM-1 synthesized via nucleophilic aromatic substitution of TTSBI (3.41 g) and TFTPN (2.01 g) in DMF (30 mL) with cyclohexane (150 mL) and K2CO3 (4.0 g) at 60 °C for 72 h under N2; product dissolved in CHCl3 and precipitated in methanol; dried at 80 °C for 24 h. Membrane fabrication: PIM-1-based MMMs prepared by priming. PIM-1 (0.1 g) dissolved in CHCl3 (9.9 g) to 1 wt% solution. Nanoparticles dispersed in CHCl3 by 1 h sonication, mixed with 1 mL PIM-1 solution and stirred 8 h; remaining PIM-1 solution added and stirred another 8 h; cast into a mold; solvent evaporation ≥2 days; membranes activated in methanol 24 h. Thin-film composite membranes formed by spin-coating 1 wt% PIM-1 on a polyimide substrate prepared by non-solvent phase inversion. MMMs denoted Z-X-Y or H-X-Y for ZIF-8 or HZIF-8, with PIM-1:MOF mass ratios from 5:0 to 5:5 (maximum workable loadings: ZIF-8 up to 5:3; HZIF-8 up to 5:5). Characterization: Morphology by SEM (ZEISS Sigma 300) and TEM (Tecnai T20). Crystallinity and phase via XRD (Bruker D8 ADVANCE, PANalytical X'Pert PRO). Surface chemistry via FT-IR (Thermo Nicolet iS50), XPS (Shimadzu AXIS Ultra DLD). Surface charge by zeta potential (MALVERN ZS-90). Porosity by N2 sorption at 77 K (Beishide BSD-660M A6B3M) with degassing at 100 °C for 450 min; BET surface areas calculated. Thermal analyses by TGA (PerkinElmer TGA400) and DSC. In-situ FT-IR from 30–150 °C (Bruker VERTEX 80). Microphase by SAXS. Free volume size/distribution by PALS using 22Na source and PATFIT analysis. Mechanical properties by nanoindentation (Young's modulus and hardness). Chain spacing by XRD and Bragg’s law. Gas transport testing: Pure-gas permeation by constant-volume, variable-pressure method at 35 °C and 3.5 bar for CO2, N2, and CH4. Sorption and diffusion coefficients derived via time-lag (solution–diffusion model). Plasticization resistance assessed by increasing CO2 feed pressure from 3.5 to 20 bar and monitoring normalized permeability. Aging assessed on thin-film composite membranes (≈780–800 nm selective layer) over 63 days, tracking CO2 permeance decay. Mixed-gas tests reported in Supplementary data.

• TA treatment converts solid ZIF-8 nanocubes (200–300 nm) into hollow ZIF-8 (HZIF-8) while preserving external morphology; TEM shows cavity formation with reaction time (0–5 min). FT-IR confirms presence of TA on HZIF-8 (OH 3300 cm−1, C=O 1704 cm−1), along with ZIF-8 bands (C=N 1146, C–N 995, Zn–N 420 cm−1). XPS shows increased O 1s and atomic O content 21.71% in HZIF-8; TA mass fraction ≈30.9 wt%.
• Porosity: ZIF-8 exhibits type I isotherm; BET area 1661 m²/g. HZIF-8 shows type II isotherm with evidence of lumen adsorption; BET area decreases to 266 m²/g due to removal of inner mesoporous structure while retaining shell crystallinity (XRD unchanged topology).
• Surface charge: ZIF-8 positively charged; HZIF-8 negatively charged due to phenolic OH; increased hydrophilicity improves dispersion and adhesion.
• Interfaces and morphology: SEM cross-sections show interfacial voids in PIM-1/ZIF-8 (Z-5-1) vs defect-free interfaces in PIM-1/HZIF-8 (H-5-1); HZIF-8 hollow structure intact in membranes.
• Interactions: Solid-state 13C NMR shows right-shift of aromatic ether peak for H-5-1; FT-IR shows red shift at 1108 cm−1 and temperature-dependent blue shift (30–150 °C), indicating hydrogen bonding between PIM-1 and HZIF-8. Proposed multiple interactions: hydrogen bonding, π–π interactions, and metal coordination.
• Polymer chain packing: Pristine PIM-1 XRD peak at 22.3° (d-spacing 3.98 Å). ZIF-8 increases chain spacing; HZIF-8 decreases chain spacing (acts as non-covalent cross-linker), densifying packing.
• Free volume: PALS r3 and r4 radii increase for Z-5-1 (from 2.84→3.01 Å; 4.52→4.61 Å), but decrease for H-5-1 (r3 2.84→2.82 Å), consistent with denser packing. N2 sorption-derived median micropore sizes: PIM-1 6.3 Å; Z-5-1 7.0 Å; H-5-1 5.6 Å.
• Mechanics: PIM-1/ZIF-8 mechanical properties decline beyond 5:1 loading and significantly at 5:3, indicating interfacial defects. PIM-1/HZIF-8 shows improved Young’s modulus and hardness from 5:0 to 5:3, only declining at 5:5, indicating better dispersion and interfacial soldering.
• Gas permeation at 35 °C, 3.5 bar: In PIM-1/ZIF-8, CO2 permeability rises from 6065 to 22046 Barrer as loading increases to 5:3, while selectivities drop (CO2/N2: 19.6→11.0; CO2/CH4: 14.6→8.2). In PIM-1/HZIF-8, low loading (H-5-0.5) keeps permeability nearly unchanged but increases selectivity (CO2/N2: 19.6→27.1; CO2/CH4: 14.6→19.7). Optimal H-5-1 achieves CO2 permeability 8268 Barrer with CO2/N2 25.1 and CO2/CH4 18.7. Even at H-5-5, selectivity remains similar to pristine PIM-1 (≈19.5 and 13.4). Mixed-gas tests show reduced absolute values due to competitive adsorption, but H-5-1 still outperforms pristine PIM-1.
• Transport mechanism (time-lag): Sorption coefficients change slightly with MOF addition; diffusivity changes dominate. ZIF-8 at higher loadings increases diffusivity but reduces diffusivity selectivity (CO2/CH4: 2.80→1.51; CO2/N2: 1.21→0.77), consistent with nonselective voids. HZIF-8 at low loading reduces CO2 diffusivity slightly but increases diffusivity selectivity (CO2/CH4: 2.80→3.80; CO2/N2: 1.21→1.60) due to chain rigidification; at moderate loadings (H-5-1, H-5-3) diffusivity rises with still higher selectivity than PIM-1 and ZIF-8 MMMs; at 5:5, agglomeration degrades selectivity.
• Plasticization resistance: Under CO2 pressure up to 20 bar, pristine PIM-1 shows 57% increase in CO2 permeability (1.57×), indicating plasticization; H-5-1 shows only 6% increase, indicating strong suppression of plasticization due to interfacial soldering and chain rigidification.
• Aging (thin films ~780–800 nm): Over 63 days, H-5-1 ages more slowly. First week CO2 permeance loss: PIM-1 42.5%, Z-5-1 28.5%, H-5-1 20.2%. Time to 50% permeance loss: PIM-1 ~8 days, Z-5-1 ~13 days, H-5-1 ~23 days. At 9 weeks: losses of 81% (PIM-1), 73% (Z-5-1), 60% (H-5-1).
• Benchmarking: PIM-1/HZIF-8 achieves simultaneous gains in permeability (up to 36%) and selectivity (up to 28%), surpassing the CO2/N2 Robeson upper bound and outperforming recent PIM-1/MOF MMMs, which often trade permeability for selectivity or vice versa. Strategy generalizes to Pebax and Matrimid with concurrent enhancements (Pebax: +126.2% CO2 permeability, +71.3% CO2/N2 selectivity; Matrimid: +33.5% permeability, +8.6% selectivity).

The work addresses the central challenge of MMMs—balancing permeability and selectivity while ensuring robust polymer–filler interfaces—by leveraging tannic acid as a multifunctional molecular solder. TA coordinates with ZIF-8 to form a hollow HZIF-8 shell and deposits phenolic groups on the surface, simultaneously improving adhesion to PIM-1 (via hydrogen bonding, π–π, and metal coordination) and reducing transmembrane mass transfer resistance within the filler. This dual action densifies and rigidifies local PIM-1 chains, raising diffusivity selectivity and suppressing physical aging and plasticization, while the hollow MOF architecture maintains or increases overall permeability. The resulting PIM-1/HZIF-8 MMMs surpass the permeability–selectivity upper bound, maintain performance under high CO2 pressures, and age more slowly than pristine PIM-1 and PIM-1/ZIF-8. Extending the approach to Pebax and Matrimid shows that polyphenol soldering is a general strategy for diverse polymer matrices, making it relevant beyond carbon capture to other separations where MMM interfacial defects and trade-offs limit performance.

A polyphenol-mediated soldering strategy using tannic acid creates hollow MOF (HZIF-8) fillers with phenolic-functionalized surfaces that strongly adhere to and rigidify polymer matrices (PIM-1), yielding defect-free interfaces and reduced mass transfer resistance. This enables synchronous improvements in CO2 permeability and CO2/N2 and CO2/CH4 selectivity, surpasses the polymeric upper bound, and significantly suppresses plasticization and physical aging. The method generalizes to different polymer classes (Pebax, Matrimid), offering a universal pathway for advanced MMMs for carbon capture and other separations. Future work could optimize filler loading and dispersion at higher contents, explore other polyphenols and MOF chemistries, and assess long-term mixed-gas performance and stability under industrially relevant conditions.

• High filler loadings approach mechanical limits: ZIF-8 MMMs become fragile beyond 5:3 loading; HZIF-8 MMMs show mechanical deterioration and agglomeration at 5:5, with loss of diffusivity selectivity.
• Selectivity decreases at higher HZIF-8 loadings as fillers connect and agglomerate, indicating a practical loading window is needed.
• TA modification reduces BET surface area of MOF (from 1661 to 266 m²/g), which may influence adsorption-driven separations.
• Mixed-gas performance shows reductions due to competitive adsorption (details in Supplementary), and long-term stability was assessed for 63 days; longer-term and process-scale validations are not reported.
• Maximum loading ratios constrained by membrane mechanical strength; the approach may require optimization for different polymers and processing conditions.