Acid-base bifunctional porous liquids

Abstract

Basic solutions are essential for gas-involved catalytic transformations, yet their ultralow free volume limits gas solubility and diffusivity. Engineering permanent porosity in such reactive liquids has remained a longstanding challenge. Here, we introduce a surface-sacrifice strategy that incorporates Brønsted-acidic zeolite nanocrystals into strong organic bases to create permanently porous reactive liquids. Stereochemically controlled acid–base neutralization at the external zeolite surface forms a thin ionic solvation shell that stabilizes the dispersion while fully preserving the internal microporosity and crystallinity of the zeolite framework. The resulting basic media exhibit persistent microporosity, confirmed by inert-gas sorption, ¹²⁹Xe nuclear magnetic resonance, neutron scattering, and theoretical simulations. These porous liquids show significantly enhanced catalytic performance in CO₂ conversion and hydrogenation reactions. The coexisting acidic and basic sites further enable antagonistic cascade catalysis within a single liquid phase. This surface-sacrifice approach provides a general route to introduce permanent porosity into reactive media, enabling boosted gas transformations.

Here authors create permanently porous basic liquids from acid zeolite nanocrystals, enabling stable microporosity, improved gas-phase catalysis, and acid–base cascade reactions within a single liquid phase.

Introduction

Gas–liquid reactions play a central role in many catalytic transformations in chemical manufacturing and energy conversion, including hydrogenation, CO₂ utilization, and electrocatalytic processes^1,2. However, their efficiency is often limited by the inherently low solubility and slow diffusion of gaseous reactants in liquid media, which restricts mass transport to catalytic sites and frequently necessitates elevated pressures or complex reactor designs³. In such systems, physicochemical parameters, including polarity and particularly available free volume, strongly influence gas solubility, diffusion, and transport efficiency^4,5. Strategies such as introducing bulky substituents or disrupting molecular packing can modestly increase free volume, yet conventional liquids fundamentally lack the permanent porosity characteristic of rigid porous frameworks such as metal–organic frameworks (MOFs) and zeolites, which exhibit superior gas storage and transport capabilities^6,7. Improving gas uptake and transport remains particularly challenging in strongly reactive liquid media, where the absence of permanent and accessible porosity often limits the amount of dissolved gas available near catalytic sites and may require elevated pressures to achieve sufficient reactant concentrations^8,9. Such limitations affect many base-mediated catalytic reactions involving gaseous substrates. For example, transformations such as the isomerization of terminal alkynes or the hydrogenation of aromatic nitriles rely on sufficient dissolved H₂ concentrations to control catalytic efficiency and selectivity^10,11. Developing strategies that enhance gas solubility in reactive basic liquids while preserving their chemical functionality, therefore, remains an important and largely unexplored challenge.

Porous liquids (PLs), which integrate permanent porosity into fluid media^12–14, have advanced considerably since their conceptual proposal in 2007 and first synthesis in 2015^15–17, with applications spanning gas storage, separation, and catalysis^16,18–29. For example, microporous water, generated by dispersing nanocrystals with hydrophobic cavities (e.g., silicalite-1) in water, enhances electrocatalysis by increasing local O₂ concentration²⁵. Macroporous water constructed from surface-modified hollow carbon spheres exhibits over sixfold higher gas uptake and largely improves the oxygen reduction reaction²⁶. In hydrogenation²⁸ and CO₂ conversion reactions²⁹, PLs could enhance local H₂ and CO₂ concentration near catalytic sites, leading to markedly improved gas diffusion and catalytic efficiency. Despite these advances, current PL systems are largely restricted to neutral media, including bulky organic solvents²⁷, ionic liquids (ILs)^24,29, and water^25,26, limiting their utility under strongly acidic or basic conditions. Developing robust PLs that retain permanent porosity in harsh reactive environments remains a grand challenge.

Here, we report a surface-sacrifice strategy to introduce permanent porosity into dense organic base media by incorporating Brønsted-acidic zeolite nanocrystals, creating a class of porous reactive liquids with markedly enhanced performance in gas-involved transformations (Fig. 1). Selective acid–base neutralization occurs at the zeolite external surface, forming a thin ionic solvation shell that stabilizes dispersions of negatively charged zeolites, prevents pore blockage, and preserves both internal porosity and acidic protons. This structural integrity is confirmed by density measurements, inert-gas sorption, ¹²⁹Xe nuclear magnetic resonance (NMR) spectroscopy, neutron scattering, titration, and theoretical modeling. Catalytically, the porous liquid base delivers over fourfold improvement in benzonitrile hydrogenation and CO₂ conversion. The coexistence of Brønsted-acidic and basic sites further enables antagonistic cascade catalysis, exemplified by a one-pot deacetalization–Knoevenagel condensation in a single liquid phase. The surface-sacrifice strategy is broadly applicable to diverse basic media and rigid porous frameworks with tunable porosity. This work establishes a generalizable approach for embedding permanent porosity into dense reactive liquids, providing a versatile platform to substantially enhance gas-involved reactions under mild conditions. The central hypothesis of this work is that selective surface acid–base neutralization can create a stabilizing interfacial solvation shell while preserving the internal porosity of rigid porous hosts in highly reactive liquid media. To test this hypothesis, we systematically combine structural characterization, molecular-level analysis, and catalytic studies to establish the formation, accessibility, and functional consequences of permanent porosity in reactive liquids.

Fig. 1. Schematic depiction of the design concept to engineer permanent porosity in a liquid organic base.

Schematic illustration of a surface-sacrificial approach, in which the external acid sites of a porous host are neutralized by an organic base while the internal micropores remain inaccessible and unreacted, preserving permanent porosity in the liquid phase.

Results and discussion

Design principles for engineering acidic zeolites in basic solutions

To evaluate this hypothesis, we investigated whether stable PL formation could be achieved in a strongly basic medium while preserving the structural integrity of the porous host. As a representative system, 1,8-diazabicyclo[5.4.0]undec-7-ene (DBU, pK_a in CH₃CN = 24.54), a widely used strong organic base and solvent, was selected as the liquid medium^30,31. Constructing a DBU-based PL is challenging due to its strong basicity, low viscosity, and lack of ionic functionality, which can lead to degradation and aggregation of the porous host^23,32. Two design criteria were employed: (1) choosing porous hosts with pore sizes smaller than the DBU kinetic dimensions (9.31 Å × 7.37 Å × 5.68 Å, Supplementary Figs. 1, 2), establishing strong liquid–solid surface interactions to ensure uniform and long-term stable dispersion. Therefore, protonated Zeolite Socony Mobil–5 (H-ZSM-5) was selected to provide permanent porosity (Supplementary Fig. 2). Its micropores exclude DBU from entering the cavities while surface Brønsted-acidic protons can undergo neutralization with the DBU base, forming ionic pairs at the external surface that stabilize the zeolite dispersion (Fig. 1). H-ZSM-5 was synthesized via hydrothermal crystallization followed by ion exchange, with tailored modifications to achieve a uniform particle size around 200 nm and high dispersity. The resulting material exhibited a Brunauer-Emmett-Teller (BET) surface area of 397 m² g⁻¹, predominantly from micropores around 0.76 nm (Supplementary Figs. 3–5), and high crystallinity as confirmed by X-ray diffraction (XRD) pattern (Supplementary Fig. 6). Scanning electron microscopy (SEM) and transmission electron microscopy (TEM) images further confirmed uniform particle size of H-ZSM-5 around 200 nm (Supplementary Fig. 7). The PL was prepared by dispersing H-ZSM-5 (30 wt%) in an acetonitrile solution of DBU, followed by solvent removal to yield porous base solution (denoted as PL-DBU) (Fig. 2a). Thermogravimetric analysis (TGA) under air indicated a final zeolite loading amount of 28.7 wt% in PL-DBU (Supplementary Figs. 8, 9). The viscosity of the PL-DBU system increases with zeolite loading, from 0.035 Pa·s (20 wt%) to 0.073 Pa·s (30 wt%) and 0.102 Pa·s (40 wt%), compared to 0.014 Pa·s for pure DBU (Supplementary Fig. 10). The dispersions exhibit nearly shear-independent behavior, indicating stable flow characteristics. A 30 wt% loading was therefore selected as a balance between dispersion stability and manageable viscosity. The resulting PL-DBU (30 wt%) exhibits good storage stability over at least one month under ambient conditions, with no observable phase separation and unchanged XRD patterns (Supplementary Fig. 11), confirming that the crystallinity of H-ZSM-5 is well preserved without structural degradation.

Fig. 2. Local structure of PL-DBU.

a An optical image of PL-DBU and structures of DBU (base) and H-ZSM-5 (porous host). b DLS profile of PL-DBU. c XRD patterns of H-ZSM-5 and PL-DBU. d X-ray PDFs of H-ZSM-5, DBU and PL-DBU. e SANS profiles of H-ZSM-5, DBU and PL-DBU. The inset shows the magnified SANS profiles in the high-q region, highlighting the characteristic peaks at q = 0.56, 0.64 and 1.71 Å⁻¹. f Measured and predicted density of DBU/H-ZSM-5 mixture. Predicted upper and lower boundaries correspond to non-porous and dry pore-containing fluids. The density values obtained from three parallel measurements are presented with corresponding error bars. Source data for (b–f) are provided as a Source Data file.

Dynamic light scattering (DLS) analysis of PL-DBU showed a narrow particle size distribution around 200 nm (Fig. 2b), indicating uniform zeolite dispersion with minimal aggregation. Upon PL-DBU formation, the crystalline structure of H-ZSM-5 was well preserved, as evidenced by the XRD pattern of PL-DBU, which retained characteristic peaks at 2θ = 7.98°, 8.82°, and 23.14°, corresponding to the (011), (020), and (051) planes of the Mobil Five (MFI) framework (Fig. 2c)³³. These reflections correspond to interplanar spacings of 11.1, 10.0, and 3.8 Å, as calculated using Bragg’s Law. X-ray scattering further confirmed that the H-ZSM-5 framework remained well-retained within the liquid basic medium (Supplementary Fig. 12). The local structures of H-ZSM-5, DBU, and PL-DBU (Fig. 2d and Supplementary Figs. 13−15) were probed via synchrotron X-ray atomic pair distribution function (PDF) analysis³⁴. For pristine H-ZSM-5, the PDF peak at 1.47–1.74 Å corresponds to the average Si–O and Al–O bond lengths, while the peak at ~2.5 Å arises from O–O correlations (Fig. 2d), consistent with simulated radial distribution functions from the crystallographic model (Supplementary Fig. 16). In PL-DBU, phase and amplitude changes between 2–5 Å indicate strong interactions between H-ZSM-5 and DBU, with peak shifts reflecting perturbations in the local chemical environment of DBU, possibly arising from neutralization of DBU with acidic hydroxyl groups on the zeolite surface.

Small-angle neutron scattering (SANS) was employed to probe the real-space mesoscale structure, nanoparticle dispersion and interfacial interactions in DBU, H-ZSM-5 and PL-DBU (Fig. 2e)^35,36. Pure DBU exhibited a featureless profile, consistent with a homogeneous liquid lacking ordered local structure. H-ZSM-5 displayed sharp high-q peaks at 0.56, 0.64, and 1.71 Å⁻¹, corresponding to characteristic distances of 11.2, 9.8, and 3.8 Å, in agreement with XRD data (Fig. 2c) and reflective of its crystalline microporous framework. A low-q slope of –3.85 indicated relatively smooth granule surfaces of H-ZSM-5 (Supplementary Table 1, Supplementary Fig. 17). Upon formation of PL-DBU, the high-q features were retained, confirming preservation of the intrinsic zeolite microporous framework, while the low-q slope changed to –4.17 (Supplementary Table 1 and Supplementary Fig. 17), suggesting formation of a modified interfacial region due to DBU interactions with zeolite. The combination of change of power law decay and well-preserved high-q peaks indicates strong DBU–zeolite surface interactions, which stabilize the dispersion and suppress zeolite particle aggregation. Given that the PL with air–filled pores contributed a larger volume than a solvent-filled system²³, density measurements were employed as an initial indicator of preserved pore volume in PL-DBU (Fig. 2f). Mixtures of DBU with varying H-ZSM-5 content (0–15 vol%) were measured, with density values obtained from three parallel measurements and presented with corresponding error bars. The observed density trends with increasing zeolite loading remain consistent with predictions assuming air-filled (no solvent filling) micropores, calculated from the intrinsic density and pore volume of the zeolite. Although density measurements alone do not conclusively establish pore accessibility, these results provide supporting evidence that the zeolite micropores are likely not occupied by the liquid phase and are consistent with PL formation.

Probing permanent porosity and interfacial acid–base interactions

Having established the structural stability of PL-DBU, we next sought to determine whether the internal pores remain accessible and identify the interfacial interactions responsible for maintaining stable dispersion. To directly probe the origin and accessibility of the internal porosity, we next employed molecular dynamics (MD) simulations to examine interfacial structure and pore exclusion at the molecular level. MD simulations were performed using a model system comprising an H-ZSM-5 slab and 120 DBU molecules in a periodic box to probe interfacial behavior. As shown by the representative snapshot (Fig. 3a and Supplementary Fig. 18) of the MD trajectory after reaching equilibrium, DBU molecules preferentially accumulate near the external surface of H-ZSM-5, while steric constraints prevent their entry into the internal micropores (Fig. 3b). In addition, the DBU molecules do show inhomogeneous distribution at the interface with the H-ZSM-5 surface, not due to the intermolecular H-bonding and cluster formation^37,38, but because of the interaction with the surface. Some layering of DBU molecules at the interface can be clearly seen from the sharp peaks in the number density of DBU at the H-ZSM-5 surfaces.

Fig. 3. Characterization of permanent porosity and surface neutralization in PL-DBU.

a A partial view of the MD simulated structure of PL-DBU (partial DBU structures are omitted for clarity). b The position of DBU relative to the H-ZSM-5 center. c ¹²⁹Xe NMR for Xe gas in DBU and PL-DBU. d Solid-state ¹³C NMR spectra of PL-DBU (SP and CP) and liquid ¹³C NMR of DBU and [DBUH]Br. e Solid-state ¹H NMR of PL-DBU and liquid ¹H NMR of DBU and [DBUH]Br. Source data for (b–e) are provided as a Source Data file.

The accessibility of these preserved micropores was further validated experimentally using ¹²⁹Xe NMR spectroscopy, a highly sensitive probe of local free volume and confinement. ¹²⁹Xe NMR exploits the pronounced chemical-shift sensitivity of Xe to its confinement environment^39,40. In dense liquids, Xe experiences restricted mobility, whereas the formation of accessible cavities, as in PLs, induces an upfield shift due to reduced steric confinement. For Xe-saturated DBU, a resonance at 195 ppm was observed (Fig. 3c). Introduction of H-ZSM-5 to form PL-DBU shifted the Xe signal upfield to 190 ppm, with only a single resonance observed because Xe undergoes rapid exchange between dissolved Xe in the liquid phase and Xe occupying accessible pore environments on the NMR timescale. Under this fast-exchange regime, the measured chemical shift represents an averaged Xe environment rather than separately resolved liquid- and pore-confined Xe populations. The observed upfield shift indicates that Xe experiences a less sterically constrained average environment, consistent with Xe sampling accessible free volume within the porous framework and providing evidence for enhanced internal free volume in PL-DBU.

Having established that the internal pores remain accessible, we next probed the interfacial interactions between DBU and the zeolite framework to elucidate the mechanisms underlying stable dispersion. The acidity of zeolite was probed by ammonia temperature-programmed desorption (NH₃-TPD) (Supplementary Fig. 19). After PL-DBU formation, solid-state NMR analysis reveals that PL-DBU contains both free and protonated DBU species. To distinguish these environments, single-pulse (SP) and cross-polarization (CP) ¹³C magic-angle spinning (MAS) NMR were employed due to their complementary sensitivity to molecular mobility (Fig. 3d). SP ¹³C MAS NMR quantitatively detects all carbon species regardless of their motional state, allowing observation of both mobile DBU molecules in the liquid phase and more rigid species that are either interacting with the solid surface or present in the solid state. In contrast, CP ¹³C MAS NMR relies on efficient ¹H–¹³C dipolar coupling and therefore selectively enhances signals from motion-restricted species, whereas highly mobile molecules whose dipolar interactions are averaged to near zero, such as those in the liquid phase, contribute little or no signal to the CP spectrum. Consequently, the CP spectrum preferentially detects DBU species immobilized through interactions with the zeolite surface, including the 165 ppm C = NH⁺ resonance of protonated [DBUH]⁺ formed at external zeolitic Brønsted acid sites (Fig. 3d). Complementary ¹H NMR identifies a 9.7 ppm C = NH⁺ proton, further confirming proton transfer (Fig. 3e). The combined SP and CP measurements therefore distinguish freely mobile DBU in the liquid phase from surface-associated or immobilized DBU species, providing direct spectroscopic evidence for DBU–zeolite interfacial interactions.

A control experiment was performed to assess the role of surface acid–base interactions in stabilizing H-ZSM-5 dispersions in DBU (Supplementary Fig. 20). Both Na-ZSM-5 and H-ZSM-5 initially formed stable dispersions in acetonitrile after ultrasonication, and the addition of DBU had minimal immediate effect. Comparatively, after 1 h of standing, the Na-ZSM-5/DBU dispersion showed pronounced white precipitation, indicating zeolite aggregation, whereas H-ZSM-5/DBU remained homogeneous. Therefore, unlike conventional PLs, which are limited to neutral conditions, PL-DBU leverages interfacial interactions between H-ZSM-5 Brønsted sites and basic DBU to form abundant ionic pairs at the external surface, as supported by the formation of protonated [DBUH]⁺ species identified by solid-state ¹³C and ¹H NMR, acting as a robust solvation shell surrounding H-ZSM-5 nanoparticles and maintaining dispersion stability in highly basic environments.

Creation and tuning of permanent porosity across diverse basic media

After successfully constructing DBU-derived PLs via the surface-sacrificing strategy, we next examined the generality of this approach by extending it to basic media with different molecular structures, basicity, and polarity, to further tailor permanent porosity across various basic media. Diverse basic media possessing different basicity (see details in Supplementary Table 2), size, and polarity with DBU such as 1,1,3,3-tetramethylguanidine (TMG, pK_a (in CH₃CN) = 22.48), 1,5-diazabicyclo[4.3.0]non-5-ene (DBN, pK_a = 23.49), phosphazene base P₁-t-Bu-tris(tetramethylene) (PtBu, pK_a = 25.46), primary amine (poly(propylene glycol) bis(2-aminopropyl ether) (PPG-2NH₂, pK_a = 17.04)³⁰, and basic (trihexyl(tetradecyl)phosphonium hydroxide) ([P₆₆₆₁₄][OH]) were explored (Fig. 4a), and the resulting solution with H-ZSM-5 (30 wt%) was denoted as PL-TMG, PL-DBN, PL-PtBu, PL-PPG-2NH₂, and PL-[P₆₆₆₁₄][OH] respectively. Then, XRD patterns, SANS profiles, gas uptake isotherms, and NMR detection were subsequently employed to assess the structural integrity of the PLs and to probe their crystallinity, porosity, and interfacial ionic characteristics. Optical images of these PLs revealed a homogeneous and stable colloidal solution formation (Supplementary Fig. 21). XRD analyses show that the crystalline framework of H-ZSM-5 remains intact across a broad spectrum of basic media—from moderately basic TMG and DBN to strongly basic PtBu—as well as in linear amines (PPG-2NH₂) and highly ionic environments such as [P₆₆₆₁₄][OH] (Fig. 4b). Complementary SANS measurements confirm that zeolite-derived porosity is preserved in all PL formulations, with high-q peaks at 0.56, 0.64, and 1.71 Å⁻¹ corresponding to the characteristic micropore features of H-ZSM-5 (Fig. 4c). Notably, PL systems exhibiting nearly identical low-q slopes (Porod exponents approaching 4; Supplementary Table 1 and Supplementary Fig. 22) display high dispersion stability, consistent with suppressed aggregation and well-defined particle interfaces. This behavior is attributed to interfacial interactions between the base and zeolite surface, likely involving ionic-pair formation, which stabilizes the dispersed particles and inhibits mesoscale clustering.

Fig. 4. Creating and tuning permanent porosity in diverse basic media.

a Chemical structures and pKa values of diverse bases for PL construction. b XRD patterns of PLs combining H-ZSM-5 or H-MCM22 with diverse basic media. c SANS profiles of diverse PLs. The upper inset shows an enlarged view of the high-q region for H-ZSM-5 and the corresponding PL samples, highlighting the characteristic peaks at q = 0.56, 0.64 and 1.71 Å⁻¹. The lower inset shows the enlarged high-q region for H-MCM22 and PL-DBU/H-MCM22, highlighting the characteristic peaks at q = 0.25, 0.52, 0.58 and 0.72 Å⁻¹. Source data for (b, c) are provided as a Source Data file.

The porosity tunability in basic media was further examined using a layered Mobil Twenty-Two (MWW) framework type zeolite (H-MCM22) combined with DBU, forming the PL-DBU/H-MCM22 system (Supplementary Fig. 23). H-MCM22 exhibited a BET surface area of 386 m² g⁻¹, microporosity of 0.58 nm (Supplementary Fig. 24), and characteristic PXRD reflections at 2θ = 7.2°, 8.2°, and 9.8°, corresponding to the (100), (101), and (102) planes, respectively (Fig. 4b). The H-MCM22 framework maintained its crystallinity after PL formation. SANS profiles of both solid H-MCM22 and liquid PL-DBU/H-MCM22 displayed distinct peaks at q ≈ 0.25, 0.52, 0.58, and 0.72 Å⁻¹, corresponding to real-space distances of 25.1, 12.1, 10.8, and 8.7 Å (Fig. 4c). These features reflect the ∼25 Å unit-cell-level cage and ∼13 Å intralayer cavity of the H-MCM22 framework, demonstrating that its intrinsic microporosity and intralayer order are preserved after PL formation (Supplementary Fig. 25). The decrease in low-q intensity is attributed to the reduced contrast between the zeolite particles and the surrounding medium (Supplementary Fig. 26). The preserved high-q features further indicate the high stability of the PLs, even in the presence of strong interfacial acid–base interactions between DBU and H-MCM22.

Probing the accessibility of internal acidic sites in basic environments

After demonstrating preserved framework integrity and pore accessibility, we next investigated whether internal Brønsted acidic sites remain chemically accessible within the basic PL environment. Given that SANS profiles, density measurements, ¹²⁹Xe NMR, and molecular simulations (Fig. 2e, f and 3a–c) show steric exclusion of basic media such as DBU from internal micropores of H-ZSM-5, it is reasonable to conclude that the internal Brønsted acidic sites within H-ZSM-5 are preserved and chemically intact in PL-DBU. Conventional acidity probes such as NH₃-TPD and pyridine IR are unsuitable for PL-DBU because the liquid environment and strong DBU–zeolite interactions obscure the characteristic signatures of adsorbed pyridine⁴¹. As confirmed by MD simulations, pyridine (6.5 Å × 6.5 Å × 3.4 Å, Supplementary Fig. 27) exhibited a ~ 1.0 nm penetration depth (Fig. 5a, Supplementary Figs. 28 and 29) in H-ZSM-5, providing sufficient accessibility to probe internal Brønsted acidic sites in PL-DBU. ¹⁵N-labeled pyridine (¹⁵N-Py) serves as a sensitive NMR probe of internal Brønsted acid sites, as protonation of pyridine generally induces a characteristic downfield ¹⁵N chemical shift reflecting local acidity⁴². While free ¹⁵N-pyridine resonates at 294.9 ppm and remains unchanged in neat DBU, PL-DBU induces an 8 ppm downfield shift (Fig. 5b), reflecting alteration of the sp²-hybridized nitrogen electronic environment and confirming the presence and accessibility of internal Brønsted sites within the zeolite micropores.

Fig. 5. Exploiting the internal acidic sites in organic bases-derived PLs.

a Partial view of the MD-simulated structure of pyridine with H-ZSM-5 (partial pyridine structures are omitted for clarity). b The ¹⁵N NMR spectra of free ¹⁵N-Py, ¹⁵N-Py in DBU and PL-DBU respectively. c The solid-state ¹H NMR spectra of H-ZSM-5, PL-[P₆₆₆₁₄][OH] and PL-DBU. Source data for (b, c) are provided as a Source Data file.

To directly probe the internal Brønsted acid sites, we employed solid-state ¹H NMR spectroscopy, which is highly sensitive to protons associated with the canonical Brønsted acid sites (Si–OH–Al) in zeolites⁴³. As shown in Fig. 5c, dehydrated H-ZSM-5 shows the characteristic 1.7 ppm silanol signal and a 4.3 ppm resonance from Si–OH–Al groups. These features are absent in neat DBU and [P₆₆₆₁₄][OH], confirming that the liquid bases do not obscure the zeolite signatures (Supplementary Figs. 30 and 31). After removing excess base, both PL-DBU and PL-[P₆₆₆₁₄][OH] retain the 4.3 ppm signal, demonstrating that internal Brønsted acid sites remain accessible within the basic medium. PL-DBU additionally shows the 9.7 ppm [DBUH]⁺ resonance (Fig. 3e), while PL-[P₆₆₆₁₄][OH] exhibits a broad 5.0–5.5 ppm feature attributed to water formed via surface neutralization through a surface-sacrificing mechanism (Fig. 5c). Furthermore, NMR results revealed that the formed solvation shell, composed of [DBUH]⁺ and ⁻[O–ZSM-5], further modifies the zeolite surface and suppresses water diffusion (Supplementary Figs. 32 and 33).

Enhanced gas solubility in basic media via porosity engineering

Having confirmed preserved porosity and internal accessibility, we next examined the functional consequences of porosity engineering on gas uptake behavior. By introducing porosity into the liquid phase of PLs, the free volume of the media is significantly increased, thereby greatly enhancing gas solubility. In PL-DBU, the strong basicity of DBU renders acidic gases unsuitable (e.g., CO₂), so inert N₂ was selected for its kinetic diameter (3.64 Å) smaller than the H-ZSM-5 micropores (0.76 nm, Supplementary Fig. 5), allowing unhindered diffusion into the internal framework. Pressure-swing adsorption up to 80 bar shows that PL-DBU (30 wt% H-ZSM-5) exhibits a > 10-fold increase in N₂ uptake relative to neat DBU, reaching 32.6 cm³ g⁻¹ at 80 bar, with 93–96% of adsorption consistently arising from the zeolite cavities (Fig. 6a and Supplementary Fig. 34). Similar behavior is observed for PL-PPG-2NH₂, where the N₂ uptake increases linearly with zeolite loading and reaches 49.2 cm³ g⁻¹ at 40 wt% (Fig. 6b). Across the 20–40 wt% loading range, the zeolite cavities consistently account for approximately 99.99% of the total uptake (Fig. 6c). At a common loading of 30 wt%, both PL-DBU and PL-PPG-2NH₂ achieve comparable capacities (~33 cm³ g⁻¹). Porosity engineering in PPG-2NH₂ is further verified using CH₄ as a probe: while pristine PPG-2NH₂ adsorbs virtually no CH₄, the corresponding PL-PPG-2NH₂ displays 10- to 27-fold enhancements at 0.5 and 1 bar (Fig. 6d), exceeding reported COF-based PLs⁴⁴. ¹²⁹Xe NMR spectroscopy further confirmed the increased free volume and the successful incorporation of permanent porosity in PL-PPG-2NH₂ (Supplementary Fig. 35).

Fig. 6. Characterization of permanent porosity in liquid amine-derived PL.

a N₂ uptake isotherm of DBU and PL-DBU at 298 K. b N₂ uptake isotherms of PPG-2NH₂ and PL-PPG-2NH₂ at 298 K. c N₂ Uptake of PPG-2NH₂ with varying zeolite loadings and corresponding zeolite contributions. d CH₄ uptake isotherms of PPG-2NH₂ and PL-PPG-2NH₂ at 298 K. Source data for (a–d) are provided as a Source Data file.

Boosted gas conversion and antagonistic cascade catalysis

To determine whether the enhanced gas uptake and preserved multifunctional environments translate into practical advantages, catalytic reactions involving gaseous reactants and cascade transformations were evaluated. The introduction of permanent porosity into organic base media establishes these PL systems as powerful platforms for gas-involved catalysis by enhancing gas solubility and mass transport¹⁸. This capability is clearly demonstrated in nitrile hydrogenation, a key route to primary amines⁴⁵, where the benzonitrile-to-benzylamine transformation typically relies on elevated H₂ pressures and strong bases¹¹. PL-DBU exhibits substantially higher N₂ solubility than neat DBU and, given the smaller kinetic diameter of H₂, is expected to further facilitate H₂ uptake. PL-DBU exhibits substantially higher N₂ solubility than neat DBU (Fig. 6a) and, given the smaller kinetic diameter of H₂, is expected to further enhance H₂ uptake. Under Ru-catalyzed conditions (H₂, 2 MPa, 100 °C, 3 h), PL-DBU with 30 wt% H-ZSM-5 achieves 94% conversion and 80% yield of benzylamine, significantly outperforming pure DBU (30% conversion, 19% yield) (Fig. 7a and Supplementary Table 3). Notably, the hydrogenation in the PL-DBU system proceeds under 2 MPa H₂ and a 3 h reaction time, compared with previously reported studies employing higher H₂ pressures (5 MPa) and/or longer reaction times (5.5 h)^11,46, while delivering superior catalytic performance, highlighting the efficiency of PL-enabled gas utilization. In contrast, DBU containing 30 wt% Na-ZSM-5 shows only 41% conversion and 24% yield, due to phase separation that limits interfacial mass transport and suppresses catalytic efficiency. These results demonstrate that the preserved microporosity and stable dispersion in the PL system are consistent with improved H₂ accessibility and transport behavior, thereby potentially reducing gas transport limitations and enhancing hydrogenation performance under milder conditions.

Fig. 7. Catalytic performance of PL systems in three representative reactions.

a A hydrogenation of benzonitrile in DBU or PL-DBU. b Reaction of CO₂ and 2-amino-3-fluorobenzonitrile catalyzed by PPG-2NH₂ or PL-PPG-2NH₂. c One-pot deacetalization–Knoevenagel condensation catalyzed by [P₆₆₆₁₄][OH]-PL containing 30 wt% H-MCM22. Source data for (a–c) are provided as a Source Data file.

Beyond increasing free volume, the surface-sacrifice strategy stabilizes zeolite dispersions via ionic-pair formation, preserving coexisting acidic and basic sites within a homogeneous phase. This bifunctionality enables multistep transformations such as CO₂ carboxylation⁴⁷. The PL-PPG-2NH₂ system, combining high CO₂ uptake, surface ionic pairs, and residual H-ZSM-5 acidity, catalyzes the carboxylation of 2-amino-3-fluorobenzonitrile under CO₂ (1 MPa, 100 °C), achieving 92% conversion and 78% yield, compared to 43% conversion and 20% yield in neat PPG-2NH₂ (Fig. 7b and Supplementary Table 4). Notably, these enhancements are achieved under identical gas pressures, indicating that improved performance may arise from improved local gas accessibility and transport within the PL system, rather than increased external pressure. The coexistence of accessible acid–base sites and confined microenvironments may contribute to the observed catalytic enhancement, highlighting the promise of PL platforms for sustainable and selective catalysis.

Furthermore, harnessing zeolite acid sites in basic media enables one-pot cascade catalysis. In the deacetalization–Knoevenagel reaction (Fig. 7c and Supplementary Table 5), nanosheet H-MCM22 (Si/Al = 15) combined with [P₆₆₆₁₄][OH] to form PL-[P₆₆₆₁₄][OH] achieves 98% conversion and 95% yield of benzylidene malononitrile, whereas H-MCM22 alone (98% conversion, 10% yield), [P₆₆₆₁₄][OH] alone (24% conversion, 19% yield), and their physical mixture (93% conversion, 56% yield) show significantly inferior performance. Based on these catalytic results and control experiments, the reaction is proposed to proceed across multiple spatial domains (Supplementary Fig. 36). First, in the zeolite micropores (acid catalysis), Brønsted acid sites catalyze the deacetalization step, affording up to 96% yield of benzaldehyde from benzaldehyde dimethyl acetal. Second, in the bulk liquid phase (base catalysis), the dissolved superbase ([P₆₆₆₁₄][OH]) promotes the Knoevenagel condensation, yielding benzylidene malononitrile in up to 98% yield from benzaldehyde. Third, at the solid–liquid interface (cooperative region), the close proximity of acid and base sites facilitates intermediate transfer and enhances overall catalytic efficiency. As a result, this spatially separated yet cooperative mechanism, involving micropores, bulk liquid, and interfacial regions, may contribute to efficient tandem catalysis and accounts for the superior performance of the PL system.

Liquid-phase media are indispensable in catalysis, yet their intrinsic dense nature restricts gas dissolution and transport, imposing fundamental limits on gas-involved reactivity. By integrating acidic zeolite nanocrystals into strong organic bases via a surface-sacrifice strategy, we generate reactive liquids that maintain their chemical functionality while acquiring permanent, accessible porosity. Multimodal characterization confirms enhanced gas uptake and preservation of active sites, translating into substantially improved catalytic performance. The concurrent presence of acidic and basic sites further enables antagonistic cascade transformations within a single fluid phase. Owing to its modular design, this approach can be extended to diverse liquid matrices and porous frameworks, offering a general route to engineer mobile media with solid-like porosity. This strategy broadens the design space for next-generation catalytic liquids, enabling more efficient gas conversion and multistep transformations.

Methods

Characterizations

DLS measurements were performed on a NanoBrook Series instrument to determine the hydrodynamic size distribution of particles in the PL. Samples were loaded into clean quartz cuvettes and equilibrated at 25 °C prior to analysis.

Total X-ray scattering measurements for PDF analysis were conducted at beamline 1-ID-C of the Advanced Photon Source (APS), Argonne National Laboratory. Data were collected in transmission geometry using a monochromatic 71.6 keV (λ = 0.173112 Å) beam confined to 200 × 200 µm and an amorphous-silicon area detector (Varex, 2880 × 2880 pixels, 150 µm pitch) positioned ~500 mm downstream of the sample. Powder specimens were loaded into 1.1 mm OD polyimide capillaries (Cole-Parmer), sealed with epoxy, and measured at room temperature. For each sample, 180 diffraction frames were recorded with 1 s exposures; scattering from an empty capillary was collected separately for background removal. Instrument geometry, including detector distance and alignment, was refined using a CeO₂ standard in GSAS-II⁴⁸. The integration of the raw images yielded intensity profiles in q-space after application of a detector mask. Total scattering structure functions, S(q), were obtained with PDFgetX3⁴⁹ by subtracting container scattering, incorporating sample composition, and applying standard area-detector corrections⁵⁰. Real-space pair distribution functions, G(r), were generated by Fourier transforming S(q) (q_max = 20.0 Å⁻¹) (Eq. (1)):

$$\mathit{G}\left(\mathit{r}\right)=\frac{\mathbf{2}}{\mathit{\pi }}{\int }_{{\mathit{q}}_{\mathit{min}}}^{{\mathit{q}}_{\mathit{max}}}\mathit{q}\left[\mathit{S}\left(\mathit{q}\right)-\mathbf{1}\right]\mathbf{sin}\left(\mathit{qr}\right)\mathit{dq}$$ 1

SANS is a powerful technique to probe the dispersion stability, porosity retention, and molecular-level interactions in PL systems. SANS measurements were carried out on the BL-6 (EQ-SANS) beamline at the Spallation Neutron Source (SNS), Oak Ridge National Laboratory³⁵ to investigate the structural characteristics of the pure DBU, H-ZSM-5, and the resulting PL-DBU system, as well as extended PL-DBN/PL-TMG/PL-PtBu/PL-PPG-2NH₂/PL-[P₆₆₆₁₄][OH]/PL-DBU/H-MCM22, with the corresponding scattering profiles presented in Figs. 2e and 4c. The 1D scattering data were stitched together from datasets collected with three configurations. The combined q range of 0.005 Å⁻¹ < q < 3Å⁻¹ was achieved by using neutrons with a wavelength range of 1-4.6 Å at 1.3 m detector distance, a wavelength range of 2.5–5.65 Å at 4 m detector distance and a wavelength range of 10–13.2 Å at 4 m detector distance. The samples were loaded in 1 mm pathlength demountable cells. The background of the empty cell was subtracted, and the scattering intensities were corrected for detector sensitivity and transmission values. The sample temperature was set to 25 °C with a Quantum Northwest Peltier-driven sample changer with a stability of 0.1 °C. The SANS data reduction followed standard procedures implemented in drtsans³⁶. A power-law form factor combined with a correlation-length term and a constant term for incoherent scattering background provides an excellent fit to the data, and the corresponding fitting parameters are summarized in Supplementary Table 1.

Viscosity measurements were performed using a rotational rheometer (ARES-G2 instrument) equipped with a 40 mm parallel-plate geometry and a Peltier temperature-controlled steel plate. The measurements were conducted in flow sweep mode, where the shear rate was systematically varied to determine the shear-rate-dependent viscosity of the sample at 298 K. The tests were carried out at a controlled temperature using the Peltier plate system, and the viscosity was recorded as a function of shear rate. Additional rheological procedures, including frequency sweep and time sweep measurements, were also conducted to evaluate the viscoelastic stability of the system.

The density of the samples was measured using a Mettler Toledo Easy D30 densimeter. The data was collected at 298 K by injecting 3 mL of the porous liquid sample into the port of the densimeter slowly and continuously to make sure that there were no bubbles. Three parallel runs were performed, and an averaged density value was deployed. Theoretical densities²³ were calculated for comparison with the experimental values (Fig. 2f). The solution density was expressed as the sum of contributions from the bulk fluid outside the pores, the porous nanoparticles, and the fluid occupying the nanoparticle micropores (Eq. (2)):

$$\mathrm{\rho }\mathrm{solution}=\mathrm{\rho }\mathrm{fluid},\mathrm{bulk}\left(1-{\mathrm{C}}_{\mathrm{np}}\right)+{\mathrm{\rho }}_{\mathrm{np}}{\mathrm{C}}_{\mathrm{np}}+\mathrm{\rho }\mathrm{fluid},\mathrm{pore}{\mathrm{V}}_{\mathrm{pore}}{\mathrm{\rho }}_{\mathrm{np}}{\mathrm{C}}_{\mathrm{np}}$$ 2

where ρ_fluid, _bulk is the density of the bulk fluid (e.g. DBU, 1.02 g/cm³),C_np is the nanoparticle volume fraction, ρ_np is the crystallographic density of H-ZSM-5 (1.79 g/cm³), ρ_fluid,pore is the density of the pore-confined fluid, and V_pore is the micropore volume (V_pore = 0.1036 cm³/g was obtained from t-plot analysis of the BET isotherm).

The ¹²⁹Xe NMR spectra were acquired on a Varian INOVA 500 spectrometer. Prior to measurement, each sample was saturated with Xe gas by bubbling for 10 min.

High-pressure gas sorption isotherms (N₂ up to 80 bar) were measured using a Micromeritics HPVA II volumetric analyzer.

CH₄ adsorption isotherms were measured using an Intelligent Gravimetric Analyser (IGA, Hiden Isochema), which couples precise pressure and temperature control with high-accuracy gravimetric detection. PPG-2NH₂ and PL-PPG-2NH₂ samples were degassed at 60 °C prior to measurement.

In situ ¹H magic-angle spinning (MAS) NMR measurements were conducted on a Varian Agilent wide-bore 300 MHz spectrometer equipped with a variable-temperature heating stack. Spectra were collected using a 7.5 mm MAS probe operated at a spinning rate of 5.5 kHz. A single-pulse sequence with a 22.5° flip angle (1 µs) and a 10 s recycle delay was used, and 512 scans were collected for each measurement. All experiments were conducted under a dry nitrogen flow to minimize moisture uptake during heating. To facilitate H₂O desorption at elevated temperatures, a small needle hole was drilled in the center of the Teflon plug sealing the H-ZSM-5 or PL samples.

Molecular dynamics simulation

Classical MD simulations were carried out with the GROMACS package⁵¹. DBU were modeled with the general Amber force field (GAFF)⁵². The H-ZSM-5 zeolite model from MFI P1 structure was parameterized based on the INTERFACE force field⁵³ with 2 T14 site Si atoms (the Si-Si distance is around 13 Å) replaced by Al to reach a Si/Al ratio of 47:1⁵⁴. Bronsted acid H atoms were added to Bronsted acid O sites. Surface cleavage along the (1 0 0) direction leads to a sinusoidal channel and along the (0 1 0) direction to a straight channel. Dangling Si–O bonds and under-coordinated Si sites were saturated with H and OH groups, respectively. A 2 × 2 × 2 H-ZSM-5 slab supercell was employed to model the interface between DBU and the zeolite surface. A total of 120 DBU molecules were inserted between slabs to generate a sufficiently thick liquid region and establish bulk-like fluid behavior away from the interface under periodic boundary conditions. The model was designed to probe local interfacial interactions and pore accessibility rather than reproduce the macroscopic composition of the porous liquid system. Following geometry optimization, the system was equilibrated for 100 ns under the NPT ensemble at 300 K and 1 bar, followed by an additional 100 ns production simulation under the NVT ensemble for structural analysis.

Synthesis of H-ZSM-5

TPAOH solution (25 wt%, 6.1 g) was initially mixed with deionized water (9.6 g), followed by the addition of tetraethyl orthosilicate (TEOS, 6.25 g). The mixture was stirred at 80 °C for 24 h to form a silica precursor. In parallel, aluminum nitrate nonahydrate (Al(NO₃)₃ 9H₂O, 0.23 g) and NaOH (0.12 g) were dissolved in water (2 g) to generate the alumina source, which was introduced dropwise into the silica precursor under stirring. The resulting gel was transferred to a Teflon-lined autoclave and crystallized at 170 °C for 24 h. The solids were isolated by centrifugation, washed thoroughly with water and ethanol, and dried under vacuum at 100 °C overnight. Calcination at 550 °C for 6 h (1 °C min⁻¹) yielded Na-ZSM-5. Proton-exchange with aqueous NH₄Cl followed by a second calcination at 550 °C afforded the H-ZSM-5 form.

Synthesis of [P₆₆₆₁₄[OH]

An anion-exchange resin in the hydroxide form (Amber Rec UP550 OH) was packed into a column. Trihexyl(tetradecyl)phosphonium chloride [P₆₆₆₁₄]Cl (5.09 g, 10.0 mmol) was dissolved in absolute ethanol (50 mL) in a 100 mL round-bottom flask to obtain a clear solution. The solution was then passed through the hydroxide-form resin column at room temperature at a moderate flow rate. After loading, the column was rinsed with additional ethanol (20–30 mL), and the combined eluates were collected. The solvent was removed under reduced pressure at 50 °C, and the resulting viscous residue was dried under high vacuum at 50 °C overnight, affording [P₆₆₆₁₄[OH] IL. The absence of chloride in the final product was confirmed by the AgNO₃ test (no AgCl precipitation).

Synthesis of porous basic solutions

The as-prepared H-ZSM-5 (300 mg) was dispersed in acetonitrile (3 mL), stirred for 1 h, and then sonicated for another 1 h. DBU (700 mg) was added and sonicated for 1 h to form a uniform solution, which was then combined with the H-ZSM-5 dispersion. The mixture was further stirred overnight at room temperature to yield a stable suspension. The resulting homogeneous mixture was concentrated using a rotary evaporator, and the porous liquid was obtained after drying under vacuum at 60 °C overnight. The final product was designated as 30 wt% H-ZSM-5 in DBU. PL-TMG, PL-DBN, PL-PtBu, PL-PPG-2NH₂, and PL-[P₆₆₆₁₄][OH] samples were prepared following analogous procedures.

Synthesis of [DBUH][Br]

In a 50 mL round-bottom flask, DBU (1.52 g, 10.0 mmol) was cooled in an ice bath and stirred with a magnetic bar. A 48 wt% aqueous HBr solution (≈1.1 mL, 10.0 mmol) was added dropwise over 5–10 min under vigorous stirring. After complete addition, the mixture was allowed to warm to room temperature and stirred for 1 h to ensure full protonation. The resulting material was dried under high vacuum at 60 °C overnight to afford [DBUH][Br], which was used without further purification.

Synthesis of H-MCM22

MCM-22 was synthesized following a gel composition of H₂O: NaOH: NaAlO₂: hexamethyleneimine: dicyclohexylamine: SiO₂ = 18: 0.1: 1/15: 0.3: 0.6: 1. Deionized water, NaOH, hexamethyleneimine, and dicyclohexylamine were added sequentially under stirring until fully dissolved, after which NaAlO₂ was introduced and dissolved completely. Silica sol was added dropwise to the precursor mixture, which was stirred at room temperature for 4 h and subsequently transferred to a Teflon-lined autoclave. Crystallization proceeded at 150 °C for 5 days. The solids were collected by centrifugation, washed with deionized water, and dried under vaccum at 110 °C overnight. Calcination at 550 °C for 6 h (1 °C min⁻¹) yielded Na-MCM-22. Ion exchange with NH₄Cl and a second calcination produced the H-form MCM-22.

Author contributions

S.D. and Z.Y. conceived the project, designed the study, and organized the manuscript. L.Q. performed the primary experiments, including synthesis and catalytic testing, conducted data analysis, and contributed to manuscript preparation. T.W. guided the zeolite synthesis. H.L. and D.J. performed MD simulations, demonstrating that DBU is sterically excluded from zeolite micropores, thereby preserving intrinsic porosity. A.G., E.L., and S.M. conducted high-pressure gas adsorption measurements (N₂ and CH₄), providing direct evidence of accessible and permanent porosity in the PL system. L.H. and G.N. carried out and analyzed SANS measurements, revealing nanoscale dispersion, interparticle correlations, and the absence of mesoscale aggregation. K.S. performed NH₃-TPD, establishing the presence and strength of Brønsted acid sites in H-ZSM-5. J.H. and Y.W. conducted solid-state NMR (¹³C and ¹H), distinguishing free versus protonated DBU species, probing interfacial ionic-pair interactions, and confirming the persistence of internal acidic sites of zeolite in a strongly basic environment. A.I., X.W., L.G., and V.C. performed PDF analysis, providing insight into the local structural integrity of the zeolite framework within the liquid phase.

Peer review

Peer review information

Nature Communications thanks Kecheng Jie 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 of this study have been included in the main text and Supplementary Information. The data that support the findings of this study are available from the corresponding author on requests. 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 https://doi.org/10.1038/s41467-026-76573-6.