Direct liquefying organic cages into porous liquid molecules for enhanced near-infrared photothermal conversion and catalysis

Abstract

The direct liquefaction of molecular cages by incorporating alkyl chains as sterically hindered fluids, without compromising porosity due to self-filling, presents a significant challenge. Here, we demonstrate that transforming hydrophobic amine cages into hydrophilic ammonium cages via quaternization with poly(ethylene glycol) bearing a terminal carboxylic acid produces a series of targeted type I porous liquid molecules featuring a porous ammonium cage as the cation and multiple carboxylate ions as anions on a kilogram scale. The hydrophobic-hydrophilic incompatibility between the cation and anion prevents alkyl chain interpenetration, preserving porosity and liquidity. Notably, photoirradiation induces stable radical generation (lasting over a year) and a red-shift in absorption toward the near-infrared region for photothermal conversion—an unexpected phenomenon in porous liquids. Utilizing this unique property, we further enhance solvent-free photothermal catalytic performance by encapsulating Au clusters within the cage cavities. This study provides new insights into the straightforward synthesis of porous liquids, akin to conventional chemical synthesis of targeted molecules through precise precursor stoichiometry. It also facilitates the extension of their functions and applications from traditional sorption to smart photothermal conversion/catalysis, promising significant advancements in these fields.

The direct liquefaction of molecular cages without compromising porosity due to self-filling presents a significant challenge. Here, the authors report a series of type I porous liquid molecules featuring a porous ammonium cage as the cation and multiple carboxylate ions as anions on a kilogram scale and demonstrate application in NIR photothermal conversion and catalysis.

Introduction

The concept of “porous liquids” was first proposed by James and colleagues in 2007, referring to liquids with permanent porosity¹. Nowadays, four types of porous liquids are classified: type I being pure fluids containing empty cavities; while type II and type III are dissolved empty hosts or homogeneously dispersed framework materials in sterically hindered solvents, respectively; type IV represents the neat microporous hosts that form transient, strongly associated liquids^2–9. Porous materials such as hollow silicas¹⁰, porous carbons¹¹, zeolites^12,13, zeolitic imidazolate frameworks^14–16, microporous metal-organic frameworks^17–19, macroporous metal-organic frameworks²⁰, metal-organic cages^21–24, covalent organic frameworks²⁵, hydrogen-bonded organic frameworks²⁶, and macrocycles^27,28 have since been employed as pore generators to prepare porous liquids. Among them, porous organic cages (POCs) with permanent pores and molecular solubility stand out as excellent candidates for constructing porous liquids^29–35. In 2015, James and Cooper et al. pioneered cage-based type II porous liquids using a POC functionalized with six crown ether substituents, followed by dissolution in 15-crown-5³⁶. While a few POC-based porous liquids have been reported to date, their synthesis has been extensively impeded by inherent drawbacks such as tedious synthetic procedures, complexity in the functionalization of the cage skeleton, a limited choice of sterically hindered solvents, and issues with volatility. In particular, the extensive use of excessive reagents throughout the prolonged synthesis and purification process presents a significant hurdle to scaling up production and achieving widespread applications of these porous liquids^37–42. Developing a straightforward synthetic procedure for controlling precursor stoichiometry to produce porous liquid molecules, akin to conventional chemical synthesis of targeted molecules, is essential for advancing this field. A promising strategy involves the direct liquefaction of cages by incorporating organic alkyl chains onto the skeleton as sterically hindered solvents to ensure fluidity. In this case, a rigid and highly stable cage is considered necessary to maintain structural integrity during the demanding synthetic and purification procedures. However, self-filling of the mobile alkyl chains inevitably occurs due to the strong hydrophobic interaction between the hydrophobic cage cavity and aliphatic chains, leading to the loss of porosity^{7,8,33,39,43,44}. Thus, exploring an easily operated and efficient synthetic methodology for porous liquids remains a grand challenge.

On the other hand, the unique properties of porous liquids, characterized by their permanent porosity and fluidic nature, have opened up a widespread application scope in various fields, including gas sorption and separation^45–48, catalytic conversion^49–53, thermal switch²⁰, and chiral separation²⁷. However, the potential of porous liquids remains largely unexplored. Recently, Dai and co-workers developed a bifunctional zeolitic porous liquid featuring incompatible Lewis pairs, which exhibited superior catalytic activity due to the synergy between the two distinct components⁵⁴. Such an approach not only broadens the horizon for investigating the cooperative interplay between sterically hindered fluids and porous skeletons but also holds the promise of achieving enhanced performance and unlocking entirely new functionalities. Nonetheless, the field of porous liquids remains unripe with uncharted territory, offering vast opportunities for groundbreaking discovery.

We report here our discovery in the preparation of type I porous liquids by liquefying amine cages, which is previously deemed unsuitable for constructing such materials due to the susceptibility of their pore structure to distortion in a solid state⁵⁵. Our approach involves a straightforward acidification of the chemically reduced CC3 cage (rCC3) with poly(ethylene glycol) (PEG) that contains a terminal carboxylic acid, resulting in targeted porous liquid molecules (PEGX-Cage, PEGX represents different PEG chains) at the kilogram scale (Fig. 1). The key to our success lies in the quaternization reaction that transforms the hydrophobic amine cage into a hydrophilic ammonium cage (Cage⁺), which not only significantly enhances the charge density of cage skeleton to retain the cavity configuration due to intracage cation-cation repulsion, but also increases cage window polarity to effectively minimize the interpenetration of hydrophobic alkyl chains, ultimately leading to permanent porosity in a liquid state. The mechanism behind the porosity preservation, driven by the incompatibility between hydrophobicity and hydrophilicity, markedly differs from previously reported size-exclusive and Coulombic repulsion effects between the cage skeleton and sterically hindered fluids^21,36,37. The porous structure of the resulting ionic cage liquid is confirmed through molecular simulation, positron (e⁺) annihilation lifetime spectroscopy (PALS), and guest sorption experiments. The systematic rheological analysis identifies the essential liquidity difference of the porous liquids by varying the molar ratio of the PEGX⁻ counteranions to Cage⁺. Unexpectedly, we discover the rarely observed generation of a radical state of the porous liquid upon exposure to external photostimulation (UV light at 365 nm), due to the spontaneous electron transfer from the carboxylate to the ammonium moiety of the cage complex. The radical state could last for at least one year in the air owing to delocalization of the unpaired electrons to neighboring phenyl rings in the cage skeleton to achieve a long-lasting durability. Interestingly, the formation of the charge-separated state represents a promising platform for near-infrared (NIR) photothermal conversion (808 nm) due to its absorption in the NIR region. This property can be harnessed to boost catalytic performance when Au clusters are encapsulated within the cage cavities.

Fig. 1. Schematic illustration depicting the synthetic methodologies for designing and fabricating cage-based type I porous liquids.

The previous “bottom-up” synthesis resulted in compromised porosity and was prone to generating a non-porous state due to the interpenetration of alkyl chains into hydrophobic cage cavities caused by hydrophobic-hydrophobic interactions. The current ionic complexation strategy modulates the polarity of the cage skeleton to be more hydrophilic, preventing the interpenetration of alkyl chains and utilizing electrostatic interactions with the negatively charged carboxylate head of PEGX⁻, thereby maintaining the inner pore of the cage cavity. Additionally, a photograph of the scalable synthesized PEGL-Cage with >1 kg in quantity is also displayed.

Results

Characterization of PEGX-Cage porous liquids

The synthesis of PEGX-Cage involves directly mixing acid-monoterminated PEG (PEGLH, PEGOH, or PEGBH) with an amine-like rCC3 cage in deionized water at stoichiometric ratios (Supplementary Figs. 1–3). After completing of quaternization reaction, the solvent can be removed via a freeze-drying procedure to obtain viscous liquids (Supplementary Fig. 4). As an example, we present here the characterizations of PEGL-Cage porous liquid (chemical structure in Fig. 2a), while the detailed characterizations for other porous liquids, such as PEGO-Cage and PEGB-Cage, are provided in the Supplementary Information (Supplementary Figs. 5–10, Supplementary Movies 1, 2). The formation of PEGL-Cage is initially confirmed by ¹H and ¹³C nuclear magnetic resonance (¹H-NMR and ¹³C-NMR) spectroscopy and Fourier transform infrared (FT-IR) spectroscopy. The chemical shift of typical signals from 7.14 and 2.27 ppm (for rCC3) to 7.42 and 2.98 ppm (for PEGL-Cage) in the ¹H-NMR spectra, and the shifting of peaks from 142.35 and 126.65 ppm (for rCC3) to 138.30 and 129.53 ppm (for PEGL-Cage) in the ¹³C-NMR spectra respectively, support the quantitative quaternization of the amine groups (Fig. 2b and Supplementary Fig. 11). Additionally, the slight downfield-shift of peak from 173.85 to 176.34 ppm in the ¹³C-NMR spectra further proves the generation of carboxylate anion. Moreover, the broad band at 3150-3400 cm⁻¹ in the FT-IR spectrum of PEGLH, which can be assigned to the stretching vibration of O–H, is significantly narrowed, indicating the consumption of carboxylic acid groups. The red-shift of band from 1747 cm⁻¹ (stretching vibration of C=O) to 1460 cm⁻¹ (symmetric stretching vibration of C···O as sharp peak) and 1610 cm⁻¹ (asymmetric stretching vibration of C···O as broad peak), which can be attributed to the transformation of C=O bond to corresponding homogenized C···O bond, further indicates the formation of PEGL⁻ carboxylate anion (Fig. 2c)⁵⁶. However, the spontaneous transfer of ammonium proton between ionic bonding and hydrogen bonding results in a residual absorption peak slightly shifted to 1735 cm^−156–58.

Fig. 2. Characterizations for the chemical and porous structure of PEGL-Cage liquid.

a Molecular structure of PEGL-Cage. b ¹H-NMR spectra (400 MHz) in CD₃OD solvent with TMS as internal standard, and c FT-IR spectra of rCC3, PEGLH, and PEGL-Cage. d MD simulation for PEGL-Cage liquid to reveal the preserved intrinsic cage cavity after quaternization reaction, with PEGL⁻ counteranions constrained and located around the Cage⁺ cavity. Cage⁺: VDW model (C: gray, N: blue, H: white), PEGL⁻ chain: CPK model (C: gray, O: red, H atoms are omitted for clarity). e Estimated pore size distribution of PEGL-Cage liquid with an averaged size of 0.63 nm from the MD simulation. f PALS spectrum of PEGL-Cage porous liquid. The fitted lifetime of τ3 (2.69 ns) corresponding to the cage cavity of 0.68 nm is shown in the spectrum with a red circle. g CO₂ sorption isotherm of PEGL-Cage at 0 °C, 1 bar. h Iodine vapor adsorption of PEGL-Cage at 75 °C for 36 h. Inset: photos showing the color change before (left, light yellow) and after (right, dark brown) iodine vapor adsorption. i UV-Vis absorption spectra for the iodine extraction from PE solution (iodine concentration: 1 mg mL⁻¹) using PEGL-Cage liquid. Inset: photos showing the iodine solution color change from pink to colorless.

In order to validate our assumption on pore structure formation, a molecular dynamics (MD) simulation was carried out first. The result reveals that the electrostatic interaction between the carboxylate head of PEGL⁻ and the ammonium group of Cage⁺ in PEGL-Cage can constrain and locate the PEG chain around the cage cavity. Meanwhile, the cationized cage windows with high polarity and hydrophilicity indeed prevent the penetration of hydrophobic terminals of PEGL⁻ into the cage cavity (Fig. 2d and Supplementary Movie 3). The calculated pore size distribution reveals an average size of 0.63 nm for PEGL-Cage liquid, indicative of the intrinsic porous structure in the liquid (Fig. 2e). The preservation of cage cavity is further confirmed by PALS measurement. The o-Ps lifetime (τ3) in free space of PEGL-Cage is measured to be 2.69 ns, corresponding to the average void diameter of 0.68 nm, which is in good agreement with the simulated data (Fig. 2f and Supplementary Table 1). Guest binding experiments provide further valuable insight and evidence into the porosity assessment for liquid-like porous materials, where CO₂ and iodine are selected as probe molecules. At 0 °C, PEGL-Cage displays an almost linear increment in the adsorption curve of CO₂ to achieve a moderate uptake capacity of 0.38 mmol g⁻¹ at 1 bar pressure, which abnormally rises to a maximum value of 0.41 mmol g⁻¹ initially, then declines as the relative pressure decreases (Fig. 2g). The unusual CO₂ sorption behavior is probably attributed to the nature of liquid-like absorbents, which lack rigid interconnecting pore channel for the rapid diffusion of the gas molecules. The CO₂ molecules require a longer diffusion time to satisfy the adsorption equilibrium (a similar result is observed for toluene uptake by ZIF-67-PLs-10 porous liquid)⁴⁶. However, the reversible physical sorption of CO₂ undoubtedly suggests the presence of permanent porosity within the PEGL-Cage liquid. Iodine adsorption experiments exhibit similar results. The PEGL-Cage displays a convenient and remarkable iodine uptake of 110 wt% within 36 h, accompanied by a color change from light yellow to dark brown (Fig. 2h). Moreover, the immiscibility of porous liquid with petroleum ether (PE) allows for the examination of the intrinsic porosity of PEGL-Cage through extraction of iodine from its solution state, characterized by the weakened UV-Vis absorption peak and the color change of iodine solution from pink to colorless (Fig. 2i). In summary, the close relationship between the simulated pore diameter of the PEGL-Cage liquid and the experimental value derived from PALS as well as the guest binding experiments strongly supports the hypothesis that the inherent cage cavity is intact and preserved in the resulting neat liquid state. In contrast, the alkyl chains of PEGL⁻ (with Na⁺ as countercation) actively occupy cavities of neutral rCC3 cage, thus leading to the loss of porosity as verified by the MD simulation and guest binding experiments (the detailed characterizations are provided in the Supplementary Information, Supplementary Fig. 12, Supplementary Movie 4). Impressively, the direct liquefying strategy through a quaternization reaction is facile and highly efficient, which could be easily scaled up to 1 kg in quantity in the one-pot procedure in our lab to produce PEGL-Cage porous liquids (Fig. 1 and Supplementary Fig. 13).

Rheological property investigation

The PEGX-Cages exhibit fluidity at room temperature (25 °C), prompting an investigation into their thermal properties through differential scanning calorimetry (DSC) measurements. Our findings indicate that all PEGX-Cages demonstrate low glass transition temperatures, ranging from −15 to 23 °C, confirming their liquid-like behavior at room temperature (Supplementary Fig. 14). This observation raises questions regarding the essential differences between these liquids and the key factors influencing their fluidity. To address these queries, we conducted an elaborate characterization of the rheological properties of the porous liquids (all of which were performed at 25 °C except for the temperature sweep tests). Our study commences with dynamic oscillatory measurements, encompassing a strain sweep test and a frequency sweep test. The strain sweep analysis unveils apparent linear viscoelastic region (LVR) for all the porous liquids, indicating their intrinsic viscoelastic nature (Supplementary Fig. 15). Subsequently, in the frequency sweep test, a 1% strain within the LVR is employed, revealing that all porous liquids exhibit viscoelastic liquid behavior across the applied frequency range of 0.1 to 100 rad s⁻¹. This is evidenced by significantly higher G” (loss modulus) values than G’ (storage modulus) values (Fig. 3a), consistent with the DSC results. Furthermore, the apparent viscosity of the PEGX-Cages was determined through steady shear experiments, unveiling a typical non-Newtonian liquid feature characterized by shear thinning behavior under the applied conditions (Supplementary Fig. 16). In contrast, pure PEGXH liquids display Newtonian behavior (under the same applied conditions), underscoring the significant influence of the cage component (Supplementary Fig. 17).

Fig. 3. Investigation of the rheological properties of PEGX-Cage porous liquids at 25 °C.

a Frequency sweep of various PEGX-Cage porous liquids, with the strain controlled at 1% within the LVR. b Strain sweep and c frequency sweep of PEGL-Cage porous liquids with different PEGLH to rCC3 molar ratios (12:1, 9:1, 6:1, and 4:1), where the frequency in (b) is controlled at 10 rad s⁻¹ and the strain in (c) is controlled at 1% within the LVR. d G” and G’ values, as well as the G”/G’ ratio at 0.1% strain and 10 rad s⁻¹ frequency, as a function of the PEGLH to rCC3 molar ratio. e Flow curve of PEGL-Cage porous liquids with different PEGLH to rCC3 molar ratios (12:1, 9:1, and 6:1) to reveal the apparent viscosity change during the steady shear experiments. f Zero-shear viscosity and critical shear rate as a function of the PEGLH to rCC3 molar ratio (12:1, 9:1, and 6:1). g Micro-physical picture for the PEGL-Cage porous liquids with varied molar ratios (PEGLH to rCC3 as 12:1, 9:1, 6:1, and 4:1) to exhibit the increased cage interaction by reducing inter Cage⁺ distance.

To gain a deeper understanding, we select PEGL-Cage as a representative for detailed investigation. By varying the mixing ratio of PEGLH to rCC3 cage, we synthesized a series of PEGL-Cage complexes with tailored molar ratios from 12:1 to 9:1, 6:1, and 4:1 (Supplementary Figs. 18, 19). Based on the strain sweep test, we observe that as the PEGLH content increases, the LVR lasts longer, indicative of greater fluidity (Fig. 3b)⁵⁹. Furthermore, when a 1% strain within the LVR is applied, all samples exhibit liquid-like behavior (G” > G’) across the frequency range of 0.1–100 rad s⁻¹ (Fig. 3c). To quantify these observations, we plot the combined curves of G” and G’ values, as well as the G”/G’ ratio at 0.1% strain and 10 rad s⁻¹ frequency, as functions of the PEGLH to rCC3 molar ratio (Fig. 3d). The data shows that as the cage content increases, there will be a convergence of G” and G’ values, indicating a gradual transition from a viscoelastic liquid to a more viscoelastic solid state (e.g., PEGL-Cage at a 4:1 ratio exhibits gel-like behavior with a G”/G’ value of 1.78)⁶⁰. In the steady shear experiments, we observe a significant increase in the apparent viscosity of the PEGL-Cage liquids by more than one order of magnitude and shear thinning behavior at a high shear rate, indicating a more viscoelastic behavior that is in consistent with the dynamic oscillatory results (Fig. 3e). Notably, the PEGL-Cage at a 4:1 ratio exhibits high viscosity and gel-like behavior, making it challenging to be applied with the steady shear experiments. Furthermore, we extract the zero-shear viscosity and critical shear rate to quantify the effect of the cage content on the rheological properties (Fig. 3f). The zero-shear viscosity increases from 3.2 to 10.3 and 162.3 Pas with higher cage content, indicating increased internal friction and higher viscosity of the porous liquids. Simultaneously, the critical shear rate decreases from 286.7 to 133.5 and 23.3 s⁻¹, suggesting that the porous liquids become more sensitive to the shear rate changes. This phenomenon can be attributed to the slowed relaxation dynamics of the PEGL-Cage complex (from 3.5 to 7.5 and 42.9 ms), which is influenced by the incorporation of more cage molecules, enhancing interactions between cage complexes⁶¹. To provide a better understanding, we propose a micro-physical picture for the PEGL-Cage porous liquids with varying molar ratios (Fig. 3g). By gradually increasing the amine cage content, the surrounding PEGL⁻ counteranions of each Cage⁺ center is reduced which subsequentially shortens the inter Cage⁺ distance (as evidenced by the MD simulations, Supplementary Fig. 20). This shortening in inter Cage⁺ distance consequentially increases the obstruction between neighboring PEGL-Cage complexes and strengthens their dynamic friction, ultimately leading to reduced relaxation dynamics. This micro-physical picture can also be employed to explain the higher viscosity observed for PEGB-Cage liquid (two orders of magnitude higher than the other two PEGX-Cages, Supplementary Fig. 16), in which the shorter PEGB⁻ counteranions facilitate interactions between PEGB-Cage complexes to reduce relaxation dynamics. Besides, it is worth noting that when the molar ratio of PEGLH to rCC3 is controlled at 3 or even smaller, the resulting PEGL-Cage appears as a solid at room temperature (Supplementary Fig. 21). Therefore, we can conclude that the species of PEGX⁻ counteranions and their molar ratio to Cage⁺ play crucial roles in determining the intrinsic liquidity nature of these materials and significantly impact their apparent rheological properties.

Additionally, the PEGX-Cage porous liquids demonstrate remarkable structure recovery under strain cycle tests and exhibit favorable thermal stability during temperature sweep tests. Both viscosity and stress of the porous liquids showcase nearly complete restoration for at least three strain cycles, as observed in the strain cycle tests ranging from 1% to 100% at a frequency of 10 rad s⁻¹ (Supplementary Figs. 22, 23). Furthermore, in the temperature sweep experiments, the decreases in G” value and apparent viscosity not only indicate an enhanced fluidity for the porous liquids due to the accelerated molecular motion with increased temperature but also confirm their superior thermal stability within the temperature range of 25–100 °C (Supplementary Fig. 24). Interestingly, the temperature ramp cycle displays a slight discrepancy between the increasing and decreasing curves, suggesting a time-dependent restructuring behavior for the porous liquids⁶².

Photostimulation-induced charge-separation radical state

To our surprise, the PEGL-Cage liquid exhibits interesting photostimulation behavior under UV light at 365 nm. An apparent color evolution from light yellow to brown is observed, which is further evidenced from the UV-Vis diffuse reflectance spectra revealing an increase in the absorption bands in a broadening range (Fig. 4a). This process occurs while maintaining PEGL-Cage structure, as evidenced by NMR spectra (Supplementary Fig. 25a). The distinct photochromic phenomenon is reminiscent of the generation of charge-separated state through the photoinduced electron transfer reaction (Fig. 4b)^63–65. This process is associated with the formation of free radicals (denoted as PEGL-Cage^•), as demonstrated by electron spin resonance (ESR) spectroscopy measurements with a strong single-line signal at g = 2.003 (Fig. 4c). In contrast, no radical signal is observed in the raw materials (rCC3 and PEGLH) and pristine PEGL-Cage, indicating the crucial role of electrostatic complexation and photostimulation. Besides, the new absorption bands in the ranges of 290–355, 360–470, 490–650, and 670–860 nm, characteristic of the generated PEGL-Cage^• radical, closely align with the calculated absorption bands (Supplementary Fig. 25b). The electron transfer mechanism finds further supports through the X-ray photoelectron spectroscopy (XPS) studies. Prior to UV irradiation, PEGL-Cage exhibits the characteristic peaks of N 1s at 400.9 eV (corresponding to the quaternary ammonium on cage skeleton) as well as peaks of O 1s at 531.8 and 533.2 eV (corresponding to the carboxylate oxygen and C–O–C on PEG chain, respectively) (Fig. 4d, e). However, after photostimulation, the intensity of the quaternary ammonium N 1s peak becomes weaker with a slight red-shift (to 400.8 eV) accompanied by the appearance of a new peak at 398.8 eV, suggesting the occurrence of partial reduction for the ammonium N atoms (on cage skeleton) as electron acceptors to form the radical state (Fig. 4d). On the other hand, an obvious blue-shift for O 1s binding energy position from 531.8 to 533.8 eV is observed, implying an electron donation from the carboxylate group to decrease the resulting electron density on PEG chain (Fig. 4e).

Fig. 4. Formation mechanism and characterization of PEGL-Cage^• liquid in the radical state.

a UV-Vis diffuse reflectance spectra of PEGL-Cage under UV light irradiation for 60 min (λ = 365 nm, 100 mW cm⁻²). Inset: photos revealing the color change in the porous liquid. b Schematic illustration showing the electron transfer process from the carboxylate group of PEGL⁻ to the ammonium group of Cage⁺ to form the PEGL-Cage^• radical state. c ESR spectra (at 9.5 GHz) of rCC3, PEGLH, PEGL-Cage, and PEGL-Cage^•. d N 1s and e O 1s core-level spectra of PEGL-Cage before (up part) and after (down part) UV light irradiation, with apparent peaks red-shift for N 1s and blue-shift for O 1s, indicating the potential electron transfer mechanism to generate PEGL-Cage^• radical state. f Structural fragment of PEGL-Cage containing the anionic carboxylate group and cationic ammonium group. g Representation of the HOMO, LUMO, and SOMO in the PEGL-Cage fragment with B3LYP level and 6–31 + G** basis set.

To gain deeper insights, density functional theory (DFT) calculations were conducted on the representative fragments of PEGL-Cage complex for the comprehensive investigation into their electronic structure and molecular orbital (Fig. 4f). Our findings reveal that the highest occupied molecular orbital (HOMO, −6.79 eV) primarily locates on the carboxylate group of the PEGL⁻ counteranion, whereas the lowest unoccupied molecular orbital (LUMO, −0.70 eV) comprises the quaternary ammonium group and phenyl ring in the cage skeleton (Fig. 4g). This highly separated HOMO-LUMO level with considerable energy gap is theoretically responsible for the typical photoinduced electron transfer process to generate the PEGL-Cage^• radical state, which agrees well with the XPS analysis. It should be noted that the generation of organic ammonium radicals via a photoinduced redox mechanism is indeed feasible, as reported in the literature, but normally exhibits a relatively short lifetime in air^66,67. Nevertheless, our PEGL-Cage^• radicals are highly stable and can persist in air for at least one year, as verified by the comparable intensity of the ESR signal (Supplementary Fig. 26). To explore the origin, the spin density distribution in the singly occupied molecular orbital (SOMO) of the PEGL-Cage^• radical was calculated, revealing that the unpaired electron mainly delocalizes to neighboring phenyl ring and, to a lesser extent, remains on the quaternary ammonium group in the current PEGL-Cage^• (Fig. 4g). The delocalization effect significantly enhances the durability of the PEGL-Cage^• radical. More solid evidences were obtained from elaborately designed control experiments, in which small amine molecules, including dimethylamine (DMA) and N-methylbenzylamine (DMBA), possessing a similar chemical structure to the cage skeleton fragments, were utilized to react with PEGLH acid precursor to generate PEGL-DMA and PEGL-DMBA complexes (the detailed characterizations are provided in the Supplementary Information, Supplementary Figs. 27–29). As consequences, we find that PEGL-DMA (without phenyl ring) does not give any significant radical signal under the external UV light stimuli (Supplementary Fig. 30). However, upon introducing a phenyl ring into the fragment, the resulting PEGL-DMBA complex exhibits a detectable signal but a less stable radical state, lasting for up to two weeks upon photoirradiation (Supplementary Fig. 31). In the case of the PEGL-Cage^• radical, the unique 3D topological structure with confined nanospace may contribute to prolonging their radicals’ lifetime^68,69. The results above strongly suggest that the photoinduced electron transfer process is responsible for the generation of the PEGL-Cage^• radicals, and the electron delocalization effect plays a crucial role in enhancing the radical stability under ambient air conditions.

NIR photothermal conversion and catalysis

Beyond our expectation, a significant red-shift in the UV-Vis diffuse reflectance spectra extending to the NIR region is observed upon the generation of a stable PEGL-Cage^• radical state (Fig. 4a), which indicates the great potential for photothermal conversion under NIR irradiation. Exposure of PEGL-Cage^• under 808 nm laser at a power density of 0.5 W cm⁻² leads to a rapid temperature rise to 62 °C within 90 s (Fig. 5a). In contrast, no obvious temperature increase occurs for the original rCC3, PEGLH, and pristine PELG-Cage samples under the same irradiation conditions, clearly proving the fascinating photothermal conversion properties of PEGL-Cage^• radical. The photothermal behavior of PEGL-Cage^• is found to be highly stable and repeatable for at least six cycles without significant performance decay (Fig. 5b). Moreover, the photothermal temperature of PEGL-Cage^• is directly proportional to the laser power, indicating the excellent thermal control performance (Fig. 5c). The photothermal conversion efficiency (η) of the PEGL-Cage^• porous liquid is calculated to be 50.8%, which outperforms among the state-of-the-art supramolecular photothermal materials^70,71 indicating a great potential in the field of photothermal application (the detailed calculation is provided in the Supplementary Information, Supplementary Fig. 32). To assess the stability of porous liquids, thermal gravimetric analysis was further conducted, which revealed a much higher thermal decomposition temperature of over 200 °C (Supplementary Fig. 33), ensuring their excellent stability under photothermal conditions.

Fig. 5. Photothermal properties and catalytic performance of the porous liquids and their metal cluster hybrids.

a Photothermal conversion curves of rCC3, PEGLH, PEGL-Cage and PEGL-Cage^• under irradiation with 808 nm laser (0.5 W cm⁻²). Photothermal conversion curves of PEGL-Cage^• b for six consecutive cycles under 808 nm laser irradiation at 0.5 W cm⁻² and c at different irradiation intensities for 90 s. d Schematic illustration for the NIR (808 nm) photothermal catalytic hydroamination reaction with AN and PA as substrates by employing the Au⊂PEGL-Cage^• catalyst. e HAADF-STEM image of the encapsulated Au clusters in Au⊂PEGL-Cage^• catalyst, scale bar: 20 nm. Inset: corresponding statistical size distribution histogram of Au clusters, data calculated from 100 counts. f XPS spectrum showing metallic Au in Au⊂PEGL-Cage^• catalyst with binding energies at 84.0 eV for 4f_7/2 and 87.7 eV for 4f_5/2, respectively. g Photothermal conversion curve of the catalytic media with Au⊂PEGL-Cage^• as catalyst, AN and PA as substrates under irradiation with an 808 nm laser (0.5 W cm⁻²) for 10 min. Inset: photos of the reaction media under 808 nm laser irradiation (0.5 W cm⁻²) at different time intervals. Comparison of the h substrate conversion at different times and i pseudo-first-order kinetic fitting curves for the model hydroamination reaction by using Au⊂PEGL-Cage^• catalyst under 808 nm laser irradiation (0.5 W cm⁻²) and external heating (oil bath at 80 °C) conditions, as well as physically mixed catalyst of Au⊂Cl-Cage/PEGL-Cage^• under 808 nm laser irradiation condition (0.5 W cm⁻²).

With essential liquefying features and considerable photothermal conversion capability, these porous liquids are expected to be ideal reaction matrix for the solvent-free photothermal catalysis of hydroamination reaction (which is considered to be a convenient and powerful approach with atom-efficiency toward industrial aminated products for medicinal chemistry, total synthesis and material science)⁷² between phenylacetylene (PA) and aniline (AN) after encapsulating Au clusters as catalytic sites (Fig. 5d). In this study, the ultrasmall Au clusters were first incorporated into the cage cavity through the electrostatic complexation strategy via a previously reported methodology, forming the Au⊂PEGL-Cage^• catalyst eventually^73,74. The narrow size distribution of the Au clusters, averaging 0.62 ± 0.13 nm (in consistence with the cage cavity size)⁷⁵, is clearly observed in the high-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM). This strongly implies the success of the encapsulation process (Fig. 5e). The metallic state as well as Au content are further confirmed by the XPS analysis and inductively coupled plasma optical emission spectroscopy, respectively (Fig. 5f and Supplementary Table 2).

Initially, the catalytic reaction was carried out by utilizing the Au⊂PEGL-Cage^• catalyst under NIR (808 nm) light irradiation (0.5 W cm⁻²) conditions. The homogeneous catalytic matrix (containing a liquid-like catalyst and substrate) could efficiently convert light into thermal energy, maintaining within 10 min the whole system at an elevated temperature of 80 ± 1.3 °C (Fig. 5g). Consequentially, the conversion of substrates to the desirable product of N-(1-phenylethylidene)benzenamine was feasibly realized, achieving a yield of >99% within 24 h (Fig. 5h and Supplementary Fig. 34). In contrast, when the catalytic reaction was carried out under conventional external heating condition (oil bath at 80 °C), a significantly lower conversion of 66% was observed (Fig. 5h). This discrepancy in catalytic performance prompts us to further explore the kinetic behavior of the Au⊂PEGL-Cage^• catalyst. The preliminary kinetic studies reveal that the hydroamination reaction follows a pseudo-first-order kinetics with proximate linear fitting plots of the ln(C/C₀) vs. reaction time (t) for both photothermal and external heating conditions, for which the kinetic constants are estimated to be k = 0.251 and 0.044 h⁻¹, respectively (Fig. 5i). Obviously, the substrate conversion is 4.7 times faster under NIR light irradiation than external heating, suggesting an unusual pathway of thermal energy transduction from reaction media to catalytic center, which could eventually promote the activation of substrates by the encapsulated Au clusters. To support our hypothesis, a control experiment was designed where a physical mixture comprising the Au⊂Cl-Cage catalyst (the detailed characterizations are provided in the Supplementary Information, Supplementary Figs. 35 and 36, Supplementary Table 2) as active sites and PEGL-Cage^• as the photothermal converter was prepared and subjected to the NIR-irradiated hydroamination reaction (the molar amounts of Au and PEGL-Cage^• are equal to their corresponding components within Au⊂PEGL-Cage^•). As expected, the catalyst mixture exhibits a lower reaction conversion (95%) and a smaller kinetic constant (k = 0.125 h⁻¹) than the Au⊂PEGL-Cage^• catalyst (Fig. 5h, i). Therefore, a reasonable explanation can be inferred that the local heating effect induced by the surrounding radical cage complex may greatly mitigate the energy loss^76–79. Moreover, the viscous ionic liquids are expected to enhance the utilization of incident light by confining it within the nanofluid system⁸⁰. This ultimately converts it into heat energy, thereby directly activating substrates to effectively accelerate the catalytic reaction (Supplementary Fig. 37). Additionally, when either the Au cluster or NIR irradiation is absent, only a negligible amount of the target product is obtained, highlighting the indispensable roles of both factors (Supplementary Fig. 38). Furthermore, the Au⊂PEGL-Cage^• catalyst exhibits excellent stability and recyclability, maintaining >99% conversion even after three catalytic cycles with almost unchanged Au cluster size and photothermal conversion property (Supplementary Figs. 39–42, Supplementary Table 2).

The versatility of Au⊂PEGL-Cage^• porous liquid catalyst was subsequently examined by expanding the substrate scope during the hydroamination reaction, in which reactants with either electron-withdrawing (–Cl, –Br, –NO₂) or donating (-OCH₃) groups bonded to the aromatic ring were investigated. All the selected substrates can be efficiently transformed to corresponding substituted imine products, with >99% conversion, >97% selectivity, >91% isolated yields and >99:1 E/Z ratios under the same photothermal condition (808 nm laser, 0.5 W cm⁻²), demonstrating the universally excellent photothermal conversion property and catalytic performance (Table 1, Supplementary Figs. 43–56, Supplementary Table 3).

Table 1.

Comparison of conversion and selectivity of the photothermal catalytic hydroamination reaction with different substrates over Au⊂PEGL-Cage^• catalyst

3a	3b		3c		3d		3e
99.9%[b], 99.1%[c]	99.8%[b], 99.9%[c]		99.2%[b], 98.7%[c]		99.0%[b], 98.3%[c]		99.9%[b], 99.0%[c]
3f		3g		3h		3i
99.9%[b], 99.6%[c]		99.9%[b], 97.7%[c]		99.4%[b], 98.7%[c]		99.2%[b], 97.3%[c]
3j		3k		3l		3m
99.6%[b], 99.6%[c]		99.7%[b], 98.3%[c]		99.2%[b], 98.1%[c]		99.0%[b], 99.3%[c]

^[a]Conditions: catalyst (50 mg), amine substrate (0.1 mmol), alkyne substrate (0.1 mmol), 808 nm laser (0.5 W cm⁻²), 24 h.

^[b]Conversion of the catalytic hydroamination reaction with different substrates, determined by GC analysis.

^[c]Selectivity of the catalytic hydroamination reaction with different substrates, determined by ¹H-NMR analysis.

Discussion

In summary, a series of type I porous liquids is readily fabricated through the ionic complexation strategy with an attainable quaternization reaction between the PEGXH chain bearing a distal carboxylic acid and a porous amine cage. The successful transformation of the hydrophobic amine cage into a hydrophilic ammonium one indeed minimizes the interpenetration of the hydrophobic terminal of PEGX⁻ chain into the Cage⁺ cavity to achieve the permanent porosity. Rational design for the porous liquids is comprehensively investigated to reveal the underlying factors, including the molar ratio of the PEGX⁻ counterpart to Cage⁺ in determining the essential liquidity of the resulting porous liquids, as demonstrated by elaborate rheology analysis. Unexpectedly, the porous liquids could be easily converted to an exceptionally stable radical state (lasting for at least one year in air) upon UV light stimulation due to the spontaneous electron transfer and delocalization effect. This imparts a fascinating photothermal conversion property, which in turn enhances catalytic performance for the solvent-free hydroamination reaction after encapsulating ultrasmall Au clusters as catalytic sites. With facile and scalable synthesis, along with precise control over precursor stoichiometry, we anticipate that this approach could provide valuable insights into the design and synthesis of porous liquids tailored for task-specific applications, pushing them forward as competitive or even more attractive alternatives to traditional materials in various technologies and industries.

Methods

Synthesis of CC3, rCC3, and Cl-Cage

The CC3 cage, reduced CC3 cage (rCC3), and cationized CC3 cage (Cl-Cage) were synthesized according to the literature³⁵. To be specific, a solution of 514.9 mg (4.5 mmol) (R,R)-1,2-diaminocyclohexane in 10 mL of DCM was added to a solution of 486.9 mg (3.0 mmol) 1,3,5-triformylbenzene in 10 mL of DCM solution. The mixed solution was ultrasonic for 1 min after the addition of 10 μL of trifluoroacetic acid, and then left to stand for one week. The resulting crystalline product was filtered, washed with methanol, and vacuum-dried at 80 °C for 24 h to afford CC3 as a white powder (708.4 mg, yield: 84%). Subsequently, 500.0 mg (0.45 mmol) of CC3 powder was dissolved in 25 mL of a methanol/DCM solution (v/v = 1:1). Then, 500.0 mg (13.2 mmol) of NaBH₄ was directly added as reductant, and the mixture was kept stirring for 15 h. After that, 1 mL of H₂O was added to quench the unreacted NaBH₄, and the solution was stirred for another 9 h. After removing the solvent under vacuum, the residual solid was washed with water, vacuum-dried at 40 °C for 24 h to afford rCC3 as a white powder (468.0 mg, yield: 92%). Finally, 114.2 mg (0.1 mmol) of rCC3 was dispersed in 5 mL of H₂O, followed by the addition of 120 μL of concentrated hydrochloric acid. The mixture was kept stirring for 24 h, directly vacuum-dried at 80 °C for 24 h to afford Cl-Cage as a white powder (156.8 mg, yield: 99%). The obtained samples were characterized for their chemical structures to ensure purity by ¹H-NMR measurements.

Synthesis of PEGX-Cage

The PEGX-Cage was synthesized by directly acidifying rCC3 with a carboxylic acid-monoterminated PEG chain. Specifically, ~1.2 mmol of PEGXH (828 mg for PEGLH with an average Mn~690 g mol⁻¹, 864 mg for PEGOH with an average Mn~720 g mol⁻¹, and 516 mg for PEGBH with an average Mn~430 g mol⁻¹) was dissolved in 10 mL of deionized water, then 114.2 mg (0.1 mmol) of rCC3 cage was added into the solution with a stoichiometric ratio of 12:1 (PEGXH/carboxylic acid group to rCC3). The resulting mixture was kept stirring for 24 h at room temperature, gradually dissolving the rCC3 cage to form a transparent solution. Finally, the solvent was removed by exhaustive freeze-drying at a pressure of <100 Pa for 48 h to afford the viscous liquid-like PEGX-Cage with 100% yield (943.1 mg for PEGL-Cage, 980.3 mg for PEGO-Cage, and 628.9 mg for PEGB-Cage, respectively). ¹H and ¹³C-NMR and mass spectrometry were utilized to verify the successful formation of PEGX-Cage porous liquids, Supplementary Figs. 57–65.

Scalable synthesis of PEGL-Cage

The scalable synthesis of PEGL-Cage was carried out as follows. 910.8 g (1.32 mol) of PEGLH was dissolved in 3 L of deionized water, and then 125.6 g (0.11 mol) of rCC3 cage was added to the solution. The mixture was stirred for 48 h to ensure the complete quaternization reaction. The PEGL-Cage was obtained by exhaustive freeze-drying at a pressure of <100 Pa for 48 h with around 1037.2 g in quantity (100% yield).

Synthesis of PEGL-Cage with different PEGL⁻ ratios

The synthesis of PEGL-Cage samples with different PEGL⁻ ratios was similar to the normal one, with the added amount of PEGLH varying from 828 mg (~1.2 mmol, 12:1 ratio, to produce 943.1 mg product) to 621 mg (~0.9 mmol, 9:1 ratio, to produce 736.7 mg product), 414 mg (~0.6 mmol, 6:1 ratio, to produce 527.1 mg product), 276 mg (~0.4 mmol, 4:1 ratio, to produce 390.9 mg product), and 207 mg (~0.3 mmol, 3:1 ratio, to produce 320.1 mg product). The obtained samples were all viscous liquids, except PEGL-Cage with a 3:1 ratio, which existed in a solid gel state at room temperature. ¹H-NMR and FT-IR spectroscopy were applied to verify the successful formation of PEGL-Cage porous liquids with different PEGLH to rCC3 molar ratios, Supplementary Figs. 18, 19, 57, 66–68.

Synthesis of non-porous PEGLNa/rCC3 mixture

828 mg (~1.2 mmol) of PEGLH was dissolved in 5 mL of deionized water, and then a NaOH aqueous solution (1 M) of 1.2 mL was added dropwise. The mixture was stirred for 10 min to ensure the quaternization reaction, after which 114.2 mg (0.1 mmol) of rCC3 cage dissolved in 5 mL of methanol was added. The resulting transparent solution was kept stirring for another 24 h, followed by a rotary-evaporation and exhaustive freeze-drying procedure at a pressure of <100 Pa for 48 h to afford the non-porous PEGLNa/rCC3 mixture as a white solid (at room temperature) with 100% yield (971.4 mg in quantity).

Synthesis of PEGL-DMA and PEGL-DMBA

The synthesis of small molecular complexation compounds of PEGL-DMA and PEGL-DMBA was similar to that of the PEGX-Cage synthesis. Typically, 690 mg (~1 mmol) of PEGLH was dissolved in 10 mL of deionized water, then 112.8 mg (1 mmol) of dimethylamine (DMA, 40 wt%) or 121.2 mg (1 mmol) of N-methylbenzylamine (DMBA) was added to the solution. The resulting mixture was kept stirring for 24 h at room temperature and freeze-dried at a pressure of <100 Pa for 48 h to afford the viscous gel-like PEGL-DMA and liquid-like PEGL-DMBA with 100% yield (734.4 mg for PEGL-DMA and 812.5 mg for PEGL-DMBA, respectively).

Photostimulation treatment

100 mg of the sample (PEGL-Cage, Au⊂PEGL-Cage, PEGL-DMA or PEGL-DMBA) was placed in a glass vessel and exposed to the 365 nm UV light irradiation (100 mW cm⁻²) for 1 h. The gradual color change was observed for PEGL-Cage (from light yellow to brown), Au⊂PEGL-Cage (from light yellow to brown), and PEGL-DMBA (from transparent to yellow), indicating the successful generation of PEGL-Cage^•, Au⊂PEGL-Cage^•, and PEGL-DMBA^• radical states. However, the color of PEGL-DMA remained constant, implying the absence of electron transfer-induced radical generation.

Synthesis of Au-encapsulated cage catalysts

The successful encapsulation of an Au cluster in the cage cavity of rCC3 was realized through the electrostatic complexation strategy reported previously by our group with certain modifications^73,74. In detail, 15.8 mg (0.01 mmol) of Cl-Cage was dissolved in 10 mL of deionized water, and then 0.5 mL of HAuCl₄ aqueous solution (1 mg mL⁻¹ Au in content) was added and vigorously vortexed for 5 min to induce the metal precursor (AuCl₄⁻ anion) inclusion through electrostatic complexation. After aging for 10 min, 1.0 mg (26 μmol) of NaBH₄ was added to the solution to initiate the reduction of the metal precursor under vigorous vortexing. The mixture was kept vortexing for 10 min to ensure the complete reduction. Afterwards, 180 μL of NaOH aqueous solution (1 M) was added to the above solution, which could feasibly react with Cl-Cage to generate precipitated rCC3 with encapsulated Au clusters inside the cage cavity. The precipitate was separated by centrifuge and washed with deionized water three times (5 mL × 3 times) to remove the impurity. The Au⊂rCC3 was obtained by exhaustive freeze-drying at a pressure of <100 Pa for 48 h with a quantity of 10.2 mg (around 86% yield). The Au⊂Cl-Cage and Au⊂PEGL-Cage were obtained by directly acidifying Au⊂rCC3 with hydrogen chloride acid (37 wt%) and carboxylic acid-monoterminated PEGLH at a stoichiometric ratio of 12:1, respectively. The synthetic procedures were the same as those of the PEGX-Cage synthesis.

Solvent-free hydroamination catalytic reaction

The hydroamination of phenylacetylene and aniline was selected as a model catalytic reaction to evaluate the performance of different catalysts (Au⊂PEGL-Cage^• and Au⊂Cl-Cage/PEGL-Cage^• mixture) under 808 nm laser irradiation (0.5 W cm⁻², local heating) and 80 °C oil bath (external heating) conditions, respectively. To be specific, 10.2 mg (0.1 mmol) of PA, 9.3 mg (0.1 mmol) of AN, and 7.7 mg (0.05 mmol) of biphenyl (as the internal standard for gas chromatography analysis) were placed in a 25 mL reaction tube to form a homogeneous solution. Then, liquid-like Au⊂PEGL-Cage^• (50 mg, containing 0.888 μmol of Au, corresponding to 0.888 mol% relative to the aniline substrate) was added into the tube, which was sealed with a rubber stopper. Afterward, the two identical reaction tubes were subjected to 808 nm laser irradiation (0.5 W cm⁻²) and an 80 °C oil bath, respectively, to initiate the hydroamination reaction. The aliquot reaction media was taken out at a certain time period, dissolved in ethyl acetate, and applied in the gas chromatography analysis to monitor the reaction process within 24 h. The reaction procedure for the Au⊂Cl-Cage/PEGL-Cage^• mixture catalyst was similar; however, the catalyst quantity was varied to keep the absolute Au content the same. In detail, Au⊂Cl-Cage (8.10 mg, containing 0.888 μmol of Au, corresponding to 0.888 mol% relative to the aniline substrate) and 50 mg of PEGL-Cage^• were used.

Other substrates with either electron-withdrawing or donating groups bonded to the aromatic ring were employed with the same molar ratios, including -Cl, -Br, -NO₂, and -OCH₃ functional groups. The catalytic conditions and procedures were the same as those of the hydroamination reaction between phenylacetylene and aniline. ¹H-NMR and ¹H-¹H 2D NOESY NMR measurements were employed for the catalytic hydroamination products to identify their chemical structure and verify their purity, Supplementary Figs. 43–56.

Author contributions

J.K.S. conceived the idea; L.T. and W.C. designed and carried out the experiments on material synthesis/characterization/property investigation/catalysis; K.Z. performed the rheological measurements and analyzed the data; J.H.Z. conducted the theoretical calculations and helped with the catalytic experiments; P.Z. and X.C. performed the positron annihilation measurements and analyzed the data; J.Y. provided resource support, technical guidance, and assisted in mechanism discussion; J.K.S. supervised the whole project, provided overall technical guidance, led the mechanism discussion, and secured the funding. All authors participated in the results discussion, writing, and revising of the manuscript. L.T., K.Z., J.H.Z., and W.C. contributed equally to this work.

Peer review

Peer review information

Nature Communications thanks Kaibin Chu, Zong-Jie Guan, and the other anonymous reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Data availability

All data generated or analyzed during this study are available within the Article, its Supplementary Information file, and the Source data file. All data are available from the corresponding author upon request. 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/s41467-025-63126-6.