Fluorine-induced gradient electric field in mesoporous covalent organic frameworks for efficient separation of polarized perfluorinated gases

Abstract

The challenge of effectively capturing and separating perfluorinated gases is critical due to environmental concerns and the potential to recover these gases as high-value products in the silicon semiconductor industry. Various strategies have been developed to enhance the mass transfer capabilities of microporous adsorbents, which are vital for producing high-performance adsorbent materials. However, slow mass transfer within micropores significantly limits their performance in applications. In this study, we introduce two isostructural mesoporous covalent organic frameworks (COFs) characterized by high crystallinity, porosity, and stability. Compared to non-fluorinated COFs, the highly fluorinated COF demonstrates superior storage capacity for octafluoropropane (C₃F₈) and perfluorocyclobutane (c-C₄F₈). It also achieves remarkable separation efficiencies for C₃F₈ (or c-C₄F₈)/N₂ (or Ar, H₂, O₂) mixtures under ambient conditions, establishing a standard in the field. This study highlights the significance of strategically modifying pore surface chemistry based on the polarizability differences of guest molecules. Such modifications enable the efficient separation of mixed gas molecules in mesoporous materials.

Capturing and separating perfluorinated gases is critical due to their environmental concerns. However, microporous adsorbents show slow mass transfer within micropores limiting their performance in applications. Here, the authors report fluorinated mesoporous covalent organic frameworks for the storage of octafluoropropane and perfluorocyclobutane and separation from other gases mixtures.

Introduction

Global warming is a complex phenomenon resulting from the accumulation of greenhouse gases (GHGs) in the Earth’s atmosphere, and it has garnered significant attention as a major environmental challenge^1,2. The primary driver of this warming trend is the excessive use of fossil fuels by humans, leading to elevated concentrations of GHGs, particularly carbon dioxide (CO₂) and methane (CH₄), which are critical heat-trapping gases³. In recent decades, researchers have made considerable efforts to mitigate global warming by capturing and storing these gases or converting them into valuable chemicals^4,5. However, greenhouse effects are influenced by various GHG mixtures. Among these, fluorocarbons are less recognized in mainstream media^6,7. Due to the unique properties of fluorine, the carbon−fluorine (C−F) bond ranks among the strongest covalent single bonds in organic chemistry⁸, with a bond dissociation energy of up to 485 kJ mol⁻¹. Consequently, the prolonged atmospheric lifetimes of perfluorinated gases present significant environmental challenges that require increased attention.

Octafluoropropane (C₃F₈) and perfluorocyclobutane (c-C₄F₈) are critical perfluorinated gases widely used in the semiconductor industry. In recent years, plasmas of C₃F₈ and c-C₄F₈ have increasingly replaced conventional tetrafluoromethane (CF₄) plasmas for the selective dry etching of SiO₂ films in ultra-large-scale integrated circuit manufacturing^9,10. This transition is attributed to the higher radial density of CF_x species in C₃F₈ and c-C₄F₈ plasmas, which provides better etching selectivity for SiO₂ over Si^11,12. However, during practical etching processes, the conversion rates of these perfluorinated gases are notably low, reaching only 30−40%. This inefficiency results in the emission of significant quantities of unutilized C₃F₈ and c-C₄F₈, along with other protective gases (e.g., N₂, Ar, H₂, or O₂), causing resource wastage and environmental pollution (Table 1)^13–15. Furthermore, C₃F₈ and c-C₄F₈ are classified as highly potent GHGs, with global warming potentials (GWPs) 8830 and 9540 times greater than CO₂, respectively^16,17. Projections suggest that the emission rates of these perfluorinated gases will further increase within the next one to two years¹⁸. Given their long atmospheric lifetimes, these gases exert a persistent impact on Earth’s radiative balance, necessitating immediate reduction measures. Developing effective recycling technologies to separate, purify, and reuse C₃F₈ and c-C₄F₈ is therefore crucial. Such advancements would substantially mitigate environmental pollution and support sustainable development in the semiconductor industry^19–22.

Table 1.

Binary gas mixtures of C₃F₈ and c-C₄F₈ with application scenarios in semiconductor etching processes

Gas mixtures	Application scenarios	Substrate materials
c-C4F8/N2	Low-k material etching with reduced damage	SiO2, porous SiCOH
c-C4F8/Ar	High-aspect-ratio dielectric etching	SiO2, low-k dielectrics
c-C4F8/O2	Selective etching of Si3N4	Si3N4
C3F8/H2	Selective tungsten (W) etching	W, Si, SiO2
C3F8/O2	Bulk etching of poly-Si or metal layers	Poly-Si, Aluminum
C3F8/Ar	Shallow etching of SiO2	SiO2

In recent years, various technological approaches have been implemented to prevent the release of perfluorinated gases into the atmosphere and to enable their economically feasible capture and reuse^23,24. Compared to traditional low-temperature distillation and liquefaction methods, adsorptive separation technology using porous materials offers advantages such as energy savings and high operational feasibility. This approach is an effective strategy for capturing and recovering C₃F₈ and c-C₄F₈^25–29. Although adsorptive separation technology is up-and-coming, most of the literature in this field focuses on non-perfluorocarbon gases, with few studies addressing the adsorption and separation of perfluorinated gases using porous materials as adsorbents³⁰, especially for macro-sized perfluorinated compounds.

Covalent organic frameworks (COFs)^31–35, as a well-known class of crystalline porous organic solids, have experienced rapid development over the past two decades. These materials have found extensive applications in gas storage and separation^36–38, catalysis^39–42, energy storage^43–46, and biomedicine^47–50, among other fields^51–55. COFs are constructed by linking carefully designed organic building blocks through reversible covalent bonds, which confer high crystallinity and good thermochemical stability, enabling them to withstand harsh environments. Given the presence of toxic and corrosive impurities such as HF and NO_x in fluorine-containing gas emissions, COFs present significant potential as high-performance adsorbents for perfluorinated gas separations.

To enhance the adsorption capacity for perfluorinated gases and improve selectivity over N₂, Ar, H₂, and O₂, we employed electrophilic fluorination to incorporate perfluorinated gas-polarizing groups into pre-synthesized COFs. To the best of our knowledge, the use of COF-based adsorbents for the adsorption and separation of C₃F₈ (or c-C₄F₈)/N₂ (or Ar, H₂, O₂) mixtures has not been previously reported. Furthermore, fluorination introduces an induced electric field gradient along the pore surface of COFs due to the high electronegativity of fluorine. This gradient enhances interactions with highly polarizable perfluorinated gases while maintaining minimal interactions with N₂, Ar, H₂, and O₂ (Supplementary Table 1). This differential affinity significantly improves separation performance^56–58. On the other hand, considering that mesoporosity facilitates rapid and high-throughput mass transfer, the development of mesoporous-based advanced separators is of great significance for practical applications^59,60.

In this study, we synthesized two isostructural mesoporous two-dimensional (2D) COFs: fluorinated CPOF-6 and hydrogenated CPOF-7. The introduction of abundant hydrophobic fluorine groups in CPOF-6 enhanced crystallinity, specific surface area, and chemical stability. The uneven distribution of strong electron-withdrawing fluorine atoms in CPOF-6 induces electric field gradients on its pore surface, leading to strong interactions with highly polarizable perfluorinated gases. Compared to CPOF-7, CPOF-6 exhibits superior capture capacities for C₃F₈ and c-C₄F₈, along with higher selectivity for C₃F₈ (or c-C₄F₈)/N₂ (or Ar, H₂, O₂) mixtures. Breakthrough experiments confirm that CPOF-6 facilitates more efficient mass transfer of C₃F₈ and c-C₄F₈ than traditional microporous adsorbents. Further theoretical analysis using density functional theory (DFT) calculations revealed strong interactions between c-C₄F₈ and the fluorinated frameworks. These findings underscore the importance of surface property modulation in enhancing the gas separation performance of mesoporous adsorbents.

Results and discussion

CPOF-6 and CPOF-7 were synthesized through Schiff base condensation reactions involving 1,3,5-tris(2,3,5,6-tetrafluoroaniline)benzene (TTFAB) or 1,3,5-tris(4-aminophenyl)benzene (TAPB) with 3,3′′,4,4′′,5,5′′-hexafluoro[1,1′:4′,1′′-terphenyl]−2′,5′-dicarboxaldehyde (HFTPDA) or [1,1′:4′,1′′-terphenyl]−2′,5′-dicarbaldehyde (TPDA) in a 1:1.5 molar ratio (Fig. 1a). The formation of imine bonds was confirmed using Fourier-transform infrared (FT-IR) spectroscopy and solid-state carbon-13 cross-polarization/magic angle spinning nuclear magnetic resonance (¹³C CP/MAS NMR). FT-IR spectra exhibited peaks at 1617 cm⁻¹ and 1620 cm⁻¹, corresponding to the characteristic absorption of C=N bonds in CPOF-6 and CPOF-7, respectively (Supplementary Figs. 5 and 6). Additionally, the absence of stretching vibration bands associated with the vN−H bond of the amine group (3420 cm⁻¹ for TAPB, 3496 cm⁻¹ for TTFAB) and the vC=O bonds of the aldehyde groups (1673 cm⁻¹ for TPDA, 1682 cm⁻¹ for HFTPDA) in the FT-IR spectra indicated the complete conversion of amine and aldehyde groups. The ¹³C CP/MAS NMR spectra further confirmed the imine bonds formation, with peaks observed at 165 ppm and 153 ppm corresponding to the chemical shifts of carbons in the imine bonds of CPOF-6 and CPOF-7, respectively (Supplementary Figs. 7 and 8). Scanning electron microscopy (SEM) analysis revealed a meshed morphology for both COFs, characterized by entangled nanofibers (Supplementary Figs. 9 and 10). Transmission electron microscopy (TEM) images showed distinct crystalline lattice fringes, indicating an ordered structural arrangement (Supplementary Figs. 11 and 12). Hydrophobicity was evaluated through water contact angle (WCA) measurements. CPOF-6 demonstrated a WCA of approximately 105°, higher than that of CPOF-7 (Supplementary Figs. 13 and 14), suggesting greater hydrophobicity. This observation was further supported by water adsorption experiments, which confirmed the hydrophobic nature of CPOF-6 (Supplementary Figs. 15 and 16). The pronounced hydrophobicity of CPOF-6 can be attributed to the high incorporation of fluorine functional groups within its framework. Notably, the theoretical fluorine content of CPOF-6 is 36.11 wt.%, significantly surpassing that of previously reported fluorinated COFs^61–64.

Fig. 1. Synthesis routes and structural characterization of 2D COFs.

a Topology-directed synthesis of imine-linked CPOF-6 and CPOF-7. b The electrostatic potential (ESP) distribution of CPOF-6. The color scale for the ESP map is shown on the right (the gradation on the scale bar is in Hartree/e). Red and blue colors represent the positive and negative parts of ESP, respectively. c The ESP distribution C₃F₈ and c-C₄F₈ molecules. The color scale for the ESP map is shown on the right (the gradation on the scale bar is in Hartree/e). d Experimental (black circle) and Pawley refined (red) PXRD patterns of CPOF-6. The difference plot between the observed and the refined PXRD patterns is presented in dark blue. Reflection positions are shown in blue-grey. e PXRD patterns of CPOF-6 after treatments under different chemical environments for 24 h. f Nitrogen adsorption isotherms of CPOF-6 measured at 77 K. Adsorption (closed symbols) and desorption curves (open symbols). Inset: pore size distribution for CPOF-6.

The chemical stability of CPOF-6 and CPOF-7 was evaluated by exposing the samples to organic solvents with varying polarities, such as N,N-dimethylformamide, ethanol, and n-hexane, as well as to aqueous solutions at different pH levels, including 6 M HCl, water, 12 M NaOH, for 24 h. Powder X-ray diffraction (PXRD) analysis revealed that CPOF-6 maintained its crystal structure under all tested conditions, significantly outperforming CPOF-7 (Fig. 1e and Supplementary Figs. 17–19). Furthermore, nitrogen adsorption and FT-IR experiments confirmed that the structure of CPOF-6 remained unchanged (Supplementary Figs. 20 and 21). To our knowledge, the chemical stability of CPOF-6 exceeds that of most existing Schiff base COFs. Thermogravimetric analysis (TGA) further showed that both COFs are stable up to 400 °C in a nitrogen atmosphere without decomposition. (Supplementary Figs. 22a and 23a). These results highlight the promising potential of CPOF-6 for practical applications.

The crystal structure and unit cell parameters of the synthesized COFs were determined through PXRD analysis, complemented by structural simulation and Pawley refinement (Fig. 1d). Their potential stacking modes were optimized using the Forcite module of the Materials Studio 7.0 software package. The theoretical PXRD patterns generated based on the AA stacking mode closely matched the experimentally observed PXRD profiles. Consequently, the optimized structures of both COFs were assigned to the P6 space group, with unit cell parameters as follows: for CPOF-6, a = b = 37.250 Å, c = 4.574 Å, α = β = 90°, and γ = 120°; for CPOF-7, a = b = 33.572 Å, c = 4.577 Å, α = β = 90°, and γ = 120°. Based on the experimental PXRD results, full-profile pattern matching (Pawley refinement) was employed to fit the proposed models for CPOF-6 and CPOF-7 (Fig. 1d and Supplementary Figs. 24–27), yielding favorable agreement factors (R_p = 4.62% and R_wp = 7.94% for CPOF-6; R_p = 4.94% and R_wp = 6.22% for CPOF-7). Peaks at 2θ = 2.75, 4.76, 5.52, 7.31, and 8.29° for CPOF-6 were attributed to the (100), (110), (200), (210), and (300) facets, respectively. These findings suggest that CPOF-6 and CPOF-7 are likely to possess similar theoretical pore diameters of approximately 2.8 nm in a honeycomb-like network. Notably, the inner surface of CPOF-6 pores is almost entirely occupied by fluorine atoms.

Nitrogen adsorption-desorption isotherms were used to evaluate the permanent porosity of activated CPOF-6 and CPOF-7 at 77 K. As shown in Fig. 1f and Supplementary Figs. 28 and 29, both COFs exhibited characteristic reversible type IV isotherms, confirming their mesoporous nature. The Brunauer−Emmett−Teller (BET) method determined specific surface areas of 1680 m² g⁻¹ for CPOF-6 and 1054 m² g⁻¹ for CPOF-7. This notable difference in BET surface areas can be attributed to enhanced interlayer stacking and structural ordering induced by fluorine functional groups, consistent with previous observations in fluorinated porous materials^65,66. Pore size distributions, calculated using DFT, revealed pore sizes of 2.8 nm for CPOF-6 and 2.2 nm for CPOF-7, in good agreement with simulation results.

Given the permanent porosity and distinct pore surface environments of the synthesized COFs, we conducted a detailed investigation into their gas adsorption and separation properties. Pure component adsorption isotherms for C₃F₈, c-C₄F₈, N₂, Ar, H₂, and O₂ were collected up to 1 bar at 273 and 298 K. Both COFs displayed significantly different adsorption profiles for the perfluorinated gases compared to N₂, Ar, H₂, and O₂, with higher uptake capacities for C₃F₈ and c-C₄F₈. Specifically, at 1.0 bar, CPOF-6 exhibited adsorption capacities of 3.37 mmol g⁻¹ and 1.79 mmol g⁻¹ for C₃F₈ at 273 and 298 K, respectively, and 4.62 mmol g⁻¹ and 3.59 mmol g⁻¹ for c-C₄F₈ at these temperatures (Fig. 2a, b). These results are higher than those of F-Cage (4.41 mmol g⁻¹ at 273 K and 2.02 mmol g⁻¹ at 298 K)³⁰. In stark contrast, the uptake of N₂, Ar, H₂, and O₂ by CPOF-6 is negligible at both 273 K and 298 K (Supplementary Fig. 30). For CPOF-7, the adsorption capacities for both C₃F₈ and c-C₄F₈ were significantly lower than those of CPOF-6 (Fig. 2c, d and Supplementary Table 2), likely due to the hydrogen-occupied pore surfaces and the lower BET surface area of CPOF-7. Nonetheless, the adsorption capacity of CPOF-7 for C₃F₈ and c-C₄F₈ remained higher than those for N₂, Ar, H₂, and O₂ (Supplementary Fig. 31).

Fig. 2. Gas sorption properties.

a N₂ and C₃F₈ adsorption isotherms of CPOF-6 at 273 K (red) and 298 K (blue). b N₂ and c-C₄F₈ adsorption isotherms of CPOF-6 at 273 K (red) and 298 K (blue). c N₂ and C₃F₈ adsorption isotherms of CPOF-7 at 273 K (red) and 298 K (blue). d N₂ and c-C₄F₈ adsorption isotherms of CPOF-7 at 273 K (red) and 298 K (blue). e Calculated isosteric heat of adsorption (Q_st) as a function of loading for C₃F₈, c-C₄F₈, and N₂ on two COFs. f Predicted selectivity for C₃F₈/N₂ (10/90, v/v) and c-C₄F₈/N₂ (10/90, v/v) on two COFs at 298 K and 1 bar.

To determine the binding preferences of each COF for perfluorinated gases relative to N₂, Ar, H₂, and O₂, we calculated the isosteric heats of adsorption (Q_st) for C₃F₈, c-C₄F₈, N₂, Ar, H₂, and O₂ using the virial fitting method based on the adsorption isotherms (Fig. 2e and Supplementary Figs. 32–37). Due to the relatively low adsorption amounts of Ar, H₂, and O₂, which resulted in unreasonable fitting parameters, these gases are excluded from the analysis. At near-zero coverage, CPOF-6 exhibited Q_st values of 37.2 kJ mol⁻¹ for C₃F₈ and 33.6 kJ mol⁻¹ for c-C₄F₈, while CPOF-7 showed slightly lower Q_st values of 28.7 kJ mol⁻¹ and 28.9 kJ mol⁻¹, respectively. Notably, the Q_st values for N₂ on CPOF-6 and CPOF-7 were significantly lower, at 9.0 kJ mol⁻¹ and 11.9 kJ mol⁻¹, respectively, compared to those for perfluorinated gases. These results suggest that the introduction of fluorine functional groups through pore surface engineering can effectively enhance the adsorption affinity for perfluorinated gases.

To further evaluate the separation potential of the two COFs for C₃F₈/N₂ and c-C₄F₈/N₂ binary gas mixtures, we calculated adsorption selectivities using ideal adsorbed solution theory (IAST). We fitted single-component adsorption isotherms with single- or dual-site Langmuir-Freundlich (DSLF) models to derive the adsorption parameters required for IAST. As depicted in Fig. 2f and Supplementary Tables 3-10, at 1 bar and 298 K, CPOF-6 exhibited selectivities of 148 for C₃F₈/N₂ (10/90, v/v) and 418 for c-C₄F₈/N₂ (10/90, v/v), significantly higher than the selectivities of 91 for C₃F₈/N₂ (10/90, v/v) and 167 for c-C₄F₈/N₂ (10/90, v/v) observed for CPOF-7. Remarkably, the c-C₄F₈/N₂ (10/90, v/v) selectivity of CPOF-6 represents the highest value reported to date³⁰. We also calculated IAST selectivities for C₃F₈/N₂ (10/90, v/v) and c-C₄F₈/N₂ (10/90, v/v) gas mixtures at 273 K for both COFs. The results, presented in Supplementary Fig. 38 and Supplementary Table 11, confirm that CPOF-6 consistently achieves higher selectivity than CPOF-7 across different temperatures. Consequently, CPOF-6 demonstrates an exceptional combination of high adsorption capacity and selectivity for perfluorinated gases through pore surface fluorination (Supplementary Table 12), setting a potential standard in the design of adsorptive materials.

Building on the high selectivity and strong binding affinity of CPOF-6 for C₃F₈ and c-C₄F₈, we conducted an in-depth evaluation of its gas separation performance for C₃F₈/N₂ and c-C₄F₈/N₂ binary mixtures through dynamic breakthrough column experiments. A 10:90 C₃F₈/N₂ or c-C₄F₈/N₂ gas mixture was introduced into an activated CPOF-6 packed column at 298 K with a total flow rate of 10.0 mL min⁻¹, using helium (50 vol%) as the carrier gas. Figure 3a, c demonstrate that both gas mixtures were effectively separated in the column, with N₂ eluting rapidly. Meanwhile, C₃F₈ and c-C₄F₈ achieved breakthrough times of 20.3 min g⁻¹ and 45.2 min g⁻¹, respectively, resulting in substantial intervals of 16.8 min g⁻¹ and 40.6 min g⁻¹ relative to N₂. This separation aligns with the differences observed in their adsorption isotherms, indicating that CPOF-6 effectively separates these mixtures under dynamic flow conditions. During the breakthrough process, the total dynamic adsorption capacities of CPOF-6 for C₃F₈ and c-C₄F₈ were calculated to be 1.06 mmol g⁻¹ and 2.15 mmol g⁻¹, respectively, close to their saturation adsorption capacities at 298 K and 1 bar. Impressively, CPOF-6 sustained high c-C₄F₈/N₂ separation performance even at 60% relative humidity, highlighting its structural stability and practical viability in humid industrial environments (Supplementary Figs. 40). To assess reusability, multiple-cycle breakthrough tests under identical conditions exhibited stable breakthrough curves for C₃F₈/N₂ and c-C₄F₈/N₂ mixtures after five cycles (Fig. 3e, f), confirming the excellent stability and reusability of CPOF-6. Additionally, breakthrough experiments with various binary mixture ratios were conducted, with results shown in Fig. 3b, d, indicating robust dynamic separation performance across a wide range of perfluorinated gas ratios (Supplementary Figs. 41 and 42). PXRD and N₂ adsorption characterization of the used CPOF-6 verified its structural integrity, consistent with the original sample (Supplementary Figs. 43 and 44). For comparison, CPOF-7 underwent identical breakthrough experiments but failed to effectively separate C₃F₈/N₂ and c-C₄F₈/N₂ mixtures (Supplementary Figs. 45 and 46). Since mesoporous COFs lack a size-sieving mechanism for these gases, the superior separation performance of CPOF-6 is primarily attributed to the fluorinated pore surfaces. Thus, CPOF-6 can be regarded as a promising benchmark material for perfluorinated gas purification under room-temperature conditions.

Fig. 3. Breakthrough curves of CPOF-6 for C₃F₈/N₂ and c-C₄F₈/N₂ mixtures at 298 K and 1 bar, using He as the carrier gas.

a 10/90, v/v and b 20/80, v/v for C₃F₈/N₂. c 10/90, v/v and d 20/80, v/v for c-C₄F₈/N₂. Cycling breakthrough experiments of CPOF-6 for e c-C₄F₈/N₂ (10/90, v/v) and f c-C₄F₈/N₂ (20/80, v/v).

Under the same experimental conditions, we further investigated the gas separation performance of C₃F₈ (or c-C₄F₈)/Ar (or H₂, O₂) binary mixtures on CPOF-6. The dynamic breakthrough curves are shown in Fig. 4. In all experiments, Ar, H₂, or O₂ breakthrough the packed column first, while C₃F₈ or c-C₄F₈ gases showed slower penetration. For example, the breakthrough times for c-C₄F₈ in the c-C₄F₈/Ar, c-C₄F₈/H₂, and c-C₄F₈/O₂ systems were 53.9, 29.4, and 51.6 min g⁻¹, respectively, with corresponding breakthrough windows of 50.3, 26.5, and 48.7 min g⁻¹. Similar sequences were observed in the C₃F₈/Ar and C₃F₈/O₂ systems (Supplementary Fig. 47). These findings highlight the effectiveness of CPOF-6 in separating these mixtures under dynamic flow conditions, emphasizing its potential application in recovering fluorine-containing waste gases from etching processes. Additionally, repeated breakthrough tests confirmed the excellent reproducibility of CPOF-6 (Fig. 4d–f). These results suggest that the fluoride-induced gradient electric field strategy facilitates the efficient separation of perfluorinated gases from N₂, Ar, H₂, or O₂, with the gradient electric field enhancing the interaction with high-polarization gases and having minimal effect on low-polarization gases. Based on our findings, we propose a promising adsorption-based technology for the recovery of fluorinated electronic specialty gases, which could provide a more sustainable alternative to incineration methods. (Supplementary Figs. 48 and 49).

Fig. 4. Breakthrough curves of CPOF-6 for c-C₄F₈/Ar, c-C₄F₈/O₂, and c-C₄F₈/H₂ mixtures at 298 K and 1 bar, using He as the carrier gas.

a c-C₄F₈/Ar (10/90, v/v), b c-C₄F₈/O₂ (10/90, v/v) and c c-C₄F₈/H₂ (10/90, v/v). Cycling breakthrough experiments of CPOF-6 for d c-C₄F₈/Ar (10/90, v/v), e c-C₄F₈/O₂ (10/90, v/v) and f c-C₄F₈/H₂ (10/90, v/v).

To explore the microstructure of the fluorinated COF and its correlation with the selective adsorption of perfluorinated gases over N₂, DFT calculations were performed. The c-C₄F₈ and N₂ mixture served as a model system to identify adsorption sites and binding energies on CPOF-6. As shown in Fig. 5, two primary adsorption sites (sites I and II) are located on the inner surface of CPOF-6 pores. At site I, c-C₄F₈ binds to the framework through cooperative interactions involving C−H···F−C hydrogen bonds and C−F···F−C halogen bonds, resulting in a calculated static binding energy of 24.6 kJ mol⁻¹. At site II, due to the abundance of positively charged hydrogen atoms, c-C₄F₈ forms multiple, shorter C−H···F−C hydrogen bonds (2.755−3.011 Å) with the framework, leading to a higher static binding energy of 29.6 kJ mol⁻¹. In contrast, the static binding energies of N₂ at these sites are significantly lower, calculated at 6.8 kJ mol⁻¹ for site I and 10.6 kJ mol⁻¹ for site II. These findings suggest a greater affinity of the adsorption sites for c-C₄F₈, aligning with experimental adsorption isotherms and breakthrough curves. Analysis of the calculated electrostatic potential and DFT results indicates that the significant difference in binding energy can be attributed to charge transfer at the adsorption phase interface, induced by the highly electronegative fluorine groups on the framework. This results in the formation of positive and negative charge regions at the adsorption interface, creating a unique electric field channel that favors the adsorption of highly polarizable c-C₄F₈ over N₂. Based on these findings, we propose a separation mechanism leveraging the electric field gradient effect, whereby channels with electric field gradients can achieve the separation of mixed gases with varying polarizabilities. This mechanism exemplifies an advanced thermodynamically driven separation technology, offering the potential for both large adsorption capacity and high separation efficiency.

Fig. 5. Adsorption configurations and static binding energies of c-C₄F₈ and N₂ in CPOF-6.

a Optimized configuration of c-C₄F₈ at Site I showing one C−H···F (3.087 Å) and two C−F···F−C (2.548 Å and 2.819 Å) interactions (ΔE = 24.6 kJ mol⁻¹). b Optimized configuration of c-C₄F₈ at Site II showing three C−H···F (2.755−3.011 Å) and one C−F···F−C (2.919 Å) interactions (ΔE = 29.6 kJ mol⁻¹). c Adsorption of N₂ at Site I showing weak interactions with low static binding energy (ΔE = 6.8 kJ mol⁻¹). d Adsorption of N₂ at Site II with slightly stronger binding than Site I (ΔE = 10.6 kJ mol⁻¹), still much weaker than c-C₄F₈. Color code: C, grey (in CPOF-6) and yellow (in c-C₄F₈); N, blue; F, green; H, white.

It is important to note that previous studies have reported enhancements in adsorption and separation performance by incorporating fluorine atoms into framework materials, enabling applications such as the capture of perfluorinated pollutants from water and the capture of low-concentration CO₂. However, these studies primarily rely on mechanisms such as F−F interactions, hydrophobic effects, anion exchange, or electrostatic attraction^67–71, which differ fundamentally from the mechanism presented in this work. In addition, conventional gas separation strategies based on size-sieving or kinetic effects typically require precise matching between pore dimensions and molecular sizes of target gases. In contrast, the fluorine-induced gradient electric field mechanism transcends microporous confinement limitations, achieving efficient separation even within mesoporous structures while significantly enhancing mass transfer efficiency. We postulate that this separation paradigm may demonstrate significant potential for broader gas mixture separations, particularly those involving components with substantial disparities in molecular polarizability. This breakthrough establishes a versatile and industrially relevant pathway for high-flux gas separation applications.

Besides the separation performance, the environmental sustainability of the material must also be considered. Perfluorooctanoic acid (PFOA), perfluorooctanesulfonic acid (PFOS), and related per- and polyfluoroalkyl substances (PFAS) have prompted intensified regulatory scrutiny worldwide due to their environmental persistence, bioaccumulative toxicity, and global distribution^72,73. Although PFAS-based polymers have not yet been explicitly banned, increasing concerns have emerged regarding their potential to cause serious and irreversible effects throughout their lifecycle. Most current knowledge about the adverse impacts of PFAS pertains to their non-polymeric forms, while the degradation behavior of polymeric PFAS over time remains poorly understood⁷⁴. Notably, CPOF-6 addresses environmental concerns by enabling the efficient capture and recovery of high-GWP gases (C₃F₈ and c-C₄F₈), aligning with sustainability goals. In contrast to non-polymeric PFAS, CPOF-6 is a crystalline, robust framework where fluorine atoms are stably integrated, leading to non-volatility, insolubility, and biological inertness under environmental conditions. This structural stability greatly mitigates risks associated with environmental leaching or degradation. Even at the end of its service life, CPOF-6 can be safely disposed of through established industrial waste management procedures⁷⁵. Specifically, TGA of CPOF-6 under air was conducted to simulate potential incineration conditions and provide direct insights into its thermal decomposition behavior in oxidative environments. The results showed that decomposition begins at around 390°C, indicating the onset of framework breakdown, and complete degradation occurs near 610°C with negligible residue. These findings confirm that CPOF-6 can be fully decomposed during incineration (Supplementary Figs. 22b and 23b), allowing for safer end-of-life management.

In summary, we have designed and synthesized two isostructural mesoporous COFs. Our findings confirm that the electric field gradient generated by the fluorinated pore surfaces significantly enhances the affinity for perfluorinated gases. This results in notable capture capacities of CPOF-6 for C₃F₈ (1.79 mmol g⁻¹) and c-C₄F₈ (3.59 mmol g⁻¹) at 298 K and 1 bar. IAST calculations indicate high separation selectivities for C₃F₈/N₂ (10/90, v/v) and c-C₄F₈/N₂ (10/90, v/v) mixtures, reaching up to 148 and 418 at 298 K and 1 bar, respectively. This positions CPOF-6 as one of the benchmark materials reported to date. The selectivity for c-C₄F₈/N₂ (10/90, v/v) is particularly noteworthy, representing the highest value reported to date. Furthermore, dynamic breakthrough experiments demonstrate that CPOF-6 effectively separates C₃F₈ (or c-C₄F₈)/N₂ (or Ar, H₂, O₂) mixtures. This study underscores the effectiveness of tailoring pore surface chemistry by considering the polarizability differences of guest molecules to achieve highly selective and efficient separations using mesoporous materials. Our study paves an alternative pathway for the development of efficient adsorption-based separation technologies.

Methods

Materials

All materials were reagent grade and used as received, unless stated otherwise. 1,3,5-tribromobenzene (Bide Pharmatech Ltd., ≥97.0%), 2,3,5,6-tetrafluoroaniline (Bide Pharmatech Ltd., ≥98.0%), 2,5-dibromo-1,4-benzenedicarboxaldehy (J&K Scientific Ltd., ≥98.0%), 3,4,5-trifluorophenylboronic acid (J&K Scientific Ltd., ≥97.0%), 1,3,5-tris(4-aminophenyl)benzene (Jilin Chinese Academy of Sciences-Yanshen Technology Co., Ltd., ≥98.0%), [1,1′:4′,1′′-terphenyl]-2′,5′-dicarbaldehyde (Jilin Chinese Academy of Sciences-Yanshen Technology Co., Ltd., ≥98.0%), dichloro[1,4-bis(diphenylphosphino)butane]palladium (II) (J&K Scientific Ltd., ≥97.0%), potassium acetate (Bide Pharmatech Ltd., ≥98.0%), N,N-dimethylacetamide (Sinopharm Chemical Reagent Co., Ltd., ≥99.0%), 2,5-dibromo-1,4-benzenedicarboxaldehy (Bide Pharmatech Ltd., ≥97.0%), 3,4,5-trifluorophenylboronic acid (Bide Pharmatech Ltd., ≥98.0%), tetrakis(triphenylphosphine)palladium(0) (J&K Scientific Ltd., ≥99.0%), sodium carbonate (Sinopharm Chemical Reagent Co., Ltd., ≥99.8%), toluene (Sinopharm Chemical Reagent Co., Ltd., ≥99.5%), sodium sulfate (Sinopharm Chemical Reagent Co., Ltd., ≥99.0%), dichloromethane (Sinopharm Chemical Reagent Co., Ltd., ≥99.5%), n-hexane (Sinopharm Chemical Reagent Co., Ltd., ≥97.0%), ethyl acetate (Sinopharm Chemical Reagent Co., Ltd., ≥99.5%), tetrahydrofuran (Sinopharm Chemical Reagent Co., Ltd., ≥99.0%), acetone (Sinopharm Chemical Reagent Co., Ltd., ≥99.0%), HCl 32% (Sinopharm Chemical Reagent Co., Ltd.), acetic acid (Sinopharm Chemical Reagent Co., Ltd., ≥99.5%), NaOH (Sinopharm Chemical Reagent Co., Ltd., ≥96.0%). The solvents were purified and dried according to the standard techniques: N,N-dimethylacetamide was distilled from CaH₂. All flash chromatography was performed using silica gel, 60 Å, 300 mesh. TLC analysis was carried out on glass plates coated with silica gel, 0.2 mm thickness. The plates were visualized using a 254 nm ultraviolet lamp.

Instrumentation

¹H NMR spectra were measured on a Bruker Fourier 400 or 600 MHz spectrometer. Unless otherwise stated, all spectra were measured at ambient temperature. CP/MAS solid-state ¹³C-NMR spectra were recorded at ambient pressure on a Bruker Fourier 600 MHz spectrometer using a standard CP pulse sequence probe with 3.2 mm (outside diameter) zirconia rotors at a spinning frequency of 12 kHz. The FTIR spectra (KBr) were obtained using a SHIMADZU IRAffinity−1 Fourier transform infrared spectrophotometer. A SHIMADZU UV−2450 spectrophotometer was used for all absorbance measurements. PXRD patterns were carried out in reflection mode on a Bruker D8 advance powder diffractometer with Cu Kα (λ = 1.5418 Å) line focused radiation at 40 kV and 40 mA from 2^θ = 1.0° up to 40° with 0.020481 increments by Bragg−Brentano. The powdered sample was added to the glass and compacted for measurement. TGA was recorded on a SHIMADZU DTG-60 thermal analyzer under nitrogen or air. Nitrogen sorption experiments were performed at 77 K up to 1 bar using a nanometric sorption analyzer. The adsorption-desorption isotherms of N₂ were obtained at 77 K using a BELSORP MAX gas sorption analyzer. Gas sorption isotherms of C₃F₈ and c-C₄F₈ were obtained from a PhysiChem iPore400 instrument, and the adsorption tube was kept at a constant temperature of 273 K and 298 K by an ethanol bath. Scanning electron microscopy (SEM) was performed on a Zeiss Gemini SEM 300 microscope instrument. Samples were prepared by dispersing the material onto conductive adhesive tapes attached to a flat aluminum sample holder and then coated with gold. High-resolution transmission electron microscopy (HR-TEM) analysis was performed using a JEOL JEM-2100 instrument operated at 200 kV. The synthesized sample was dispersed in ethanol to form a well-dispersed suspension, and a droplet of the suspension was subsequently deposited onto a carbon film-coated TEM grid.

Column breakthrough experiments

Breakthrough experiments were conducted using a custom-built breakthrough column apparatus to evaluate the separation performance of mixed gases. The flow rate of the gas mixture was controlled using a mass flow controller and calibrated with a soap film flowmeter. Pre-activated sample powders were packed into a stainless-steel column (4.65 mm inner diameter, 100 mm length) and subjected to helium (He) purging at 20.0 mL min⁻¹ and 100 °C for 4 h, ensuring complete activation. After cooling to room temperature (298 K), C₃F₈ (or c-C₄F₈)/N₂ (or Ar, H₂, O₂) mixtures were introduced at a flow rate of 10.0 mL min⁻¹. The outlet gas composition was monitored using a Pfeiffer GSD 320 mass spectrometer (Germany). After the breakthrough experiment, the sample was regenerated by purging with He at a flow rate of 20 mL min⁻¹ at 100 °C for 4 h to ensure complete removal of adsorbed gases.

Density functional theory calculations

To gain deeper insight into the guest−host interactions at the molecular level, density functional theory (DFT) calculations were performed using the Vienna Ab Initio Simulation Package (VASP, version 5.4.4)^76–79. The Perdew−Burke−Ernzerhof (PBE) functional within the generalized gradient approximation (GGA)⁸⁰ was employed to describe the exchange-correlation energy. The projector augmented-wave (PAW)^81,82 method was used to represent the interaction between electrons and ions, with a plane-wave cutoff energy of 450 eV. Grimme’s D3 dispersion correction was applied to account for long-range van der Waals interactions^83,84. During the geometry optimizations, all atomic positions were fully relaxed until the forces on each atom were smaller than 0.01 eV Å⁻¹, and the electronic self-consistent field (SCF) energy convergence threshold was set to 1.0 × 10⁻⁵ eV. Due to the large size of the simulation cell, a 1 × 1 × 1 k-point grid was used for Brillouin zone sampling⁸⁵. The adsorption energies (E_ad) of c-C₄F₈ and N₂ on the CPOF-6 surface were calculated to evaluate the interaction strength between the gas molecules and the framework. The adsorption energy was obtained using the equation:

$$E_{\mathrm{ad}}=E_{\mathrm{total}}-E_{\mathrm{gas}}-E_{\mathrm{cof}}$$

where E_total is the total energy of the optimized guest−host complex, E_gas is the energy of the isolated gas molecule (either c-C₄F₈ or N₂) in the gas phase, and E_cof is the energy of the pristine CPOF-6 structure.

Synthesis of 1,3,5-tris(2,3,5,6-tetrafluoroaniline)benzene (TTFAB)

1,3,5-tribromobenzene (2.20 g, 7.00 mmol, 1.00 eq.) was added to a Schlenk flask with 4 equiv of 2,3,5,6-tetrafluoroaniline (4.62 g, 28 mmol, 4.00 eq.), along with potassium acetate (1.37 g, 14.00 mmol, 2.00 eq.) and dichloro[1,4-bis(diphenylphosphino)butane]palladium (II) (84.40 mg, 0.14 mmol, 0.02 eq.). The flask was purged with nitrogen, and 21 mL of anhydrous DMAc was added. The solution was heated to 150 °C and stirred rapidly for 20 h. After cooling to r.t., EA (180 mL) was added to the solution, which was then washed with H₂O (200 mL × 3) and brine (200 mL × 3) and dried over anhydrous Na₂SO₄. The solvent was removed, and the residue was purified via column chromatography using EA/n-hexane [4:6] as eluent to afford the product as a light pink solid (1.23 g, 31%). ¹H-NMR (400 MHz, DMSO-d₆): δ_H [ppm] = 7.54 (s, 3H), 6.23 (s, 6H). ¹³C NMR (151 MHz, DMSO-d₆): δ_C [ppm] = 143.80 (dm, J = 240.3 Hz), 135.86 (dm, J = 233.3 Hz), 131.60, 128.50, 128.32, 103.13 (t, J = 17.3 Hz).

Synthesis of 3,3′′,4,4′′,5,5′′-hexafluoro[1,1′:4′,1′′-terphenyl]-2′,5′-dicarboxaldehyde (HFTPDA)

2,5-dibromo-1,4-benzenedicarboxaldehy (0.5 g, 2.00 mmol, 1.00 eq.), 3,4,5-trifluorophenylboronic acid (0.88 g, 5.00 mmol, 2.50 eq.), tetrakis(triphenylphosphine)palladium(0) catalyst (80.0 mg, 0.07 mmol, 0.035 eq.) and toluene (25 mL) were added to a 100 mL Schlenk flask equipped with a magnetic stirrer bar. The mixture was degassed by bubbling N₂ for 30 min, at which point degassed aqueous Na₂CO₃ solution (2 M, 4 mL) was added. The reaction was stirred at 80 °C under an N₂ atmosphere for 48 h. After cooling to r.t., 100 mL deionized water was added and extracted with CH₂Cl₂. The resulting organic extract was washed with water (100 mL × 3) and brine (100 mL × 3), dried over anhydrous Na₂SO₄, and then concentrated under reduced pressure. The crude solid was recrystallized from CHCl₃ to afford the product as a white solid (0.51 g, 75%). ¹H-NMR (600 MHz, DMSO-d₆): δ_H [ppm] = 10.03 (s, 2H), 8.01 (s, 2H), 7.64 (dd, J = 8.4, 6.5 Hz, 4H). ¹³C NMR (151 MHz, DMSO-d₆): δ_C [ppm] = 190.98, 150.90, 149.34, 140.86, 139.95, 138.29, 136.06, 133.20, 131.23, 114.98 (dd, J = 16.9, 4.6 Hz).

Synthesis of CPOF-6

A Pyrex tube measuring o.d. × i.d. = 10 × 8 mm² was charged with TTFAB (28.4 mg, 0.05 mmol), HFTPDA (29.6 mg, 0.075 mmol) in a mixed solution of DMAc (1.0 mL) and trifluoroacetic acid (0.5 mL). The Pyrex tube was flash-frozen in a liquid nitrogen bath sealed under a vacuum. Upon warming to room temperature, the tube was placed in an oven at 120 °C for 7 days. The yellow solid was isolated by filtration and washed with THF (3 × 15 mL), acetone (3 × 15 mL), and n-hexane (3 × 15 mL). The powder was dried at 80 °C under vacuum overnight to afford the CPOF-6 as a yellow crystalline solid (42.3 mg, Yield: 73%). Elemental analysis for the calculated: C, 58.71%; H, 1.37%; N, 3.81%. Found: C, 57.15%; H, 1.56%; N, 4.14%.

Synthesis of CPOF-7

A Pyrex tube measuring o.d. × i.d. = 10 × 8 mm² was charged with TAPB (17.6 mg, 0.05 mmol), TPDA (21.5 mg, 0.075 mmol) in a mixed solution of DMAc (1.0 mL) and 6 M aqueous acetic acid (0.2 mL). The Pyrex tube was flash-frozen in a liquid nitrogen bath sealed under a vacuum. Upon warming to room temperature, the tube was placed in an oven at 120 °C for 3 days. The yellow solid was isolated by filtration and washed with THF (3 × 15 mL), acetone (3 × 15 mL), and n-hexane (3 × 15 mL). The powder was dried at 80 °C under vacuum overnight to afford the CPOF-7 as a yellow crystalline solid (29.7 mg, Yield: 76%). Elemental analysis for the calculated: C, 89.26%; H, 4.95%; N, 5.78%. Found: C, 85.56%; H, 4.79%; N, 6.07%.

Author contributions

T.B. and Y.Z. initiated and led the research project. M.C. and D.W. performed the synthesis and characterization of the samples. W.G. and Y.C. conducted the dynamic breakthrough experiments. Y.G. carried out the theoretical calculation. Y.Y., G.X., Y.B.Z., and W.Z. were involved in data analysis and participated in revising the manuscript. T.B., Y.Z., and M.C. organized the results and wrote the paper with discussion and revision of all authors.

Peer review

Peer review information

Nature Communications thanks Heping Ma, Feng He,and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Data availability

The data that support the findings detailed in this study are available within the article and supplementary information as well as on figshare: 10.6084/m9.figshare.29108999. All data are available from the corresponding author upon request. Source data are provided with the paper. Source data are provided with this paper.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41467-025-61333-9.