Chemistry, applications, and future prospects of structured liquids†

Interestingly, employing the art of interfacial complexation in the Pickering emulsions provides the opportunity to stabilize the emulsion through the jamming of nanoparticles at the interface of two immiscible liquids. Additionally, structured emulsions facilitate the tailoring of viscoelastic properties and microstructures, enabling the creation of structured emulgels suitable for use in additive manufacturing.^38,39,60 In this regard, attractive Pickering emulsion gels (APEGs) were obtained via electrostatic interactions between the nanoparticles stabilizing the water droplets and a telechelic polymer within the oil phase.³⁹ In this approach, the water droplets are covered via negatively charged nanoparticles, i.e., carboxylic acid functionalized nanoparticles, providing active sites for electrostatic interaction with the positively charged telechelic polymeric ligands. Similar to typical NPS assemblies, this interaction is also based on the ammonium–carboxylated electrostatic interactions, in which telechelic polymeric ligands with two –NH₂ ends interact with two negatively charged nanoparticles in two neighboring aqueous droplets, bridging these droplets and facilitating the formation of APEGs (Fig. 2). In such a system, the type of nanoparticles, oil phase, and telechelic polysemic ligands are customizable, providing a pathway to engineering the composition of APEGs on demand. Importantly, such a jammed system induces a strong viscoelastic and shear-thinning character, enabling the direct ink writing of the APEGs to yield customized 3D structures in the air or even under water. Of note, the rheological properties of APEGs can also be tuned by the molecular weight of the telechelic polymeric ligand, pH, and temperature, providing several opportunities for designing functional materials with controllable composition and rheological characteristics.

	Fig. 2 (a) Formation of Pickering emulsion gel through the electrostatic interaction between the carboxylic acid functionalized nanoparticles in the aqueous phase and the α,ω-diamino functionalized polymer as the colloidal surfactant in the oil phase; the displayed confocal microscopy image showcases the formation of Pickering emulsion gel in which the oil phase is stained with Nile Red. (b) Interfacial assembly and formation of Pickering emulsion gel through the electrostatic interaction between the negatively charged nanoparticles and polymeric surfactant at the oil/water interface, starting from tailing and intra-droplet looping to inter-droplet bridging and looping. (a) and (b) Reproduced with permission.39 Copyright 2018, Wiley-VCH.

In another study, Zhang et al.⁶⁰ developed viscoelastic Pickering emulsion gel through NPSs jamming at the interface. The formed Pickering emulsion gel consisted of negatively charged nanoparticles dispersed within the aqueous phase, i.e., cellulose nanofiber (CNF), graphene oxide (GO), and nanoclay (LAPONITE®), and didodecyldimethylammonium bromide (DDAB) as the ligand dispersed within the oil phase.⁶⁰ The electrostatic interaction between the nanoparticles and ligands at the interface formed stable Pickering emulgels that stabilized the oil droplets within the continuous phase. In this system, the jamming of nanoparticles such as GO restricted the droplets and led to an increase in the viscosity and stiffness of the emulsion gel, leading to ideal printability. This also provides the possibility of homogeneous hybridization of non-water soluble and water-soluble nanoparticles at each respective oil and water phase, allowing for the customized composition and flexibility of design (Fig. 3a). It was evident that the GO stacking at the interface through jamming could improve the viscoelastic properties of the dispersions.^66,67 Upon an increase in the concentration of GO to more than 1 wt%, short distances among the GO flakes promote their interactions, which are mainly based on hydrogen bonding, limiting their movement and mobility. This could also be accompanied by the steric hindrance of large GO flakes that restrict their movement upon jamming at the interface.^67,68 The jamming of the nanoparticles substantially reduces the spacing between the droplets, limiting their ability to deform and move. This restriction enhances the viscosity and stiffness of the emulgels, thereby improving their printability.

	Fig. 3 (a) From the formation of Pickering emulsion gel to 3D printing; (step 1) emulsification of the aqueous phase containing nanoparticles and the oil phase containing ligand via homogenizer leading to Pickering emulsion gel formation; the jammed emulsion system provides the possibility to tune the viscoelastic character of the ink toward direct ink writing. (Step 2) 3D printing of the jammed emulsion gel, controlling the spatial arrangement of 3D constructs. (Step 3) crosslinking and reduction, and (step 4) drying to yield a 3D structured construct with the ability to add water soluble and non-soluble nanoparticles in each water and oil phase, respectively. (a) Reproduced with permission.60 Copyright 2021, Nature Publishing Group. (b) Formation of emulsion gel through interfacial complexation of MXene/GO and octadecyl amine as the ligand, enabling direct ink writing and 3D structural design. The jamming process allows tuning the rheological and viscoelastic characteristics of the ink toward 3D printing. (b) Reproduced with permission.38 Copyright 2022, American Chemical Society.

In contrast to Pickering emulgels, which are designed using classical surfactants without interfacial jamming, NPSs formed through electrostatic interactions between nanoparticles in the aqueous phase and ligands in the oil phase can stabilize emulsion systems more effectively. This enhanced stability arises from the significantly higher binding energy of NPSs to the interface compared to classical surfactants, thereby preventing droplet coalescence in emulsion systems. In this regard, Li et al.³⁸ employed the art of interfacial complexation to yield 3D printable GO/MXene emulsion gel inks at low nanoparticle concentrations. The system is composed of low concentrations of GO (2.8 mg cm⁻³) and MXene (4.1 mg cm⁻³) in the aqueous phase, electrostatically interacting with the octadecyl amine (ODA) as the ligand in the oil phase (toluene), forming sheet-like NPS assemblies at the interface (Fig. 3b). In this system, the anchored ligand interactions among droplets dictate the rheological characteristic of the system, allowing a gel-like behavior with required viscoelastic properties toward direct ink writing at low 2D nanomaterials concentration. This happens upon reduction in the interfacial tension by electrostatic interactions between the nanoparticles and ligands at the interface, their subsequent jamming, and minimizing the interfacial area of the emulsion droplets.^69,70 This locks in the shape of the droplets within the continuous aqueous phase and forms stable emulsion gel with ideal viscoelastic behavior toward additive manufacturing. It is worth noting that the interfacial assembly of 2D GO and MXene and the viscoelastic characteristics of the resulting emulsion gel are spurred by a substantial reduction in Helmholtz free energy, which is contingent upon the geometry and shape of these nanoparticles.⁴¹ This suggests that a broader array of 2D nanomaterials, such as MoS₂ or boron nitride, could potentially be employed to create emulsion gel systems in the future.

Importantly, from the technical perspective, the formation of all-liquid molded constructs is quite challenging due to several crucial criteria: (1) the binding energy that preserves the NPS assemblies at the interface must be high enough to retain their integrity, (2) the NPS assemblies formation must be faster than any relaxation of fluids, avoiding their deformation into equilibrium droplets, (3) the disordered interfacially driven elastic film generated from the jamming of NPS assemblies must be strong enough to suppress the instabilities during the removal of structured liquids from the mold and maintain the integrity of the liquid construct in a non-equilibrium form.^1,9 These criteria were satisfied through the prompt and strong binding of the cellulose nanocrystals (CNCs) with amine-terminated polystyrene (PS-NH₂) at the oil/water interface, forming robust interfacially driven assemblies of cellulose nanocrystal surfactants (CNCSs).⁵⁸ The turning point of CNCSs is their rapid and strong binding to the oil/water interface that could overcome the instabilities and retain the structural integrity of the liquid structures, paving the way for precisely designed 3D-structured all-liquid constructs.⁹

The all-liquid molding process starts from the placement of the CNC suspension into a predesigned mold, prewetted with an oil solution containing an amine-terminated polymeric ligand (Fig. 4a). Upon immersing the mold in the oil solution containing amine-terminated ligands, CNCSs rapidly form at the exposed interfaces, and the sculpted liquid constructs will be extracted from the mold because of the oil's higher density than the aqueous phase. Upon the extraction of the sculpted liquid construct from the mold, a decrease in the interfacial area will be initiated upon jamming of CNCSs assemblies at the interface, arresting the non-equilibrium shape of the molded liquid structure indefinitely.

	Fig. 4 (a) The process of all-liquid molding, starting from (I) prewetting the mold with oil/ligand solution and placement of the CNC suspension in the mold. (II) Immersion of the mold into the oil/ligand solution and (III) its subsequent density-driven extraction from the mold. (b) Surface coverage of the CNCSs assemblies altered via the variation of pH. (c) Effect of pH on the shape fidelity of the molded liquid constructs, and (d) formation of molded liquid structured alphabets via the liquid molding strategy. (a)–(d) Reproduced with permission.9 Copyright 2018, Wiley-VCH.

The electrostatic interaction of CNC and PS-NH₂ ligand is pH-driven and occurs as a result of the complexation of the deprotonated sulfate group (–OSO₃⁻, pK_a of ≈1.9) of CNC and protonated amine group (–NH₃⁺, pK_a of ≈9.0) of PS-NH₂, leading to disordered CNCSs assemblies at the interface and a dramatic decline in the interfacial tension. Accordingly, altering the pH of the medium can significantly affect the surface coverage of CNCSs assemblies, with higher surface coverage and structural stability at pH ∼3 (Fig. 4b and c). This enables reconfigurable all-liquid molded soft materials with ideal structural integrity and fidelity, e.g., complex geometries of alphabets with inherent mechanical robustness of NPS assemblies (Fig. 4d). Furthermore, the shape of the sculpted liquid objects can be altered via external stimuli by applying an electric field, external force, shear force, and varying the medium's pH,⁹ holding a great promise for developing smart and stimuli-responsive soft materials.

Liquid streaming's potential was recently showcased through the creation of structured liquid threads, also known as liquid worms.⁴⁰ In this approach, an aqueous suspension of graphene oxide (GO) was injected into a nonpolar domain containing hexane and aminopropylsobutyl polyhedral oligomeric silsesquioxane (POSS-NH₂). The fast and pH dependant electrostatic interaction between the deprotonated carboxylic acid functional groups of GO (–COO⁻, pK_a ≈ 4.2) and the protonated amine group of (POSS-NH₂) (–NH₃⁺, pK_a ≈ 9.0) was reported to stabilize the falling stream of GO in a nonpolar domain through suppressing the Plateau–Rayleigh instabilities (Fig. 5a–c).⁴⁰

	Fig. 5 (a) Schematic illustration of electrostatic interaction between the deprotonated carboxylic acid functional groups of GO and the protonated amine group of POSS-NH2, forming filamentous liquid threads by suppressing the instabilities. (b) (I) Break-up of the streamed liquid into discrete droplets to overcome the instabilities and (II) formation of filamentous liquid threads in a non-equilibrium form, stabilized through the formation of NPSs. (c) The process of “liquid streaming” starts from the injection of an aqueous ink containing custom composition/concentration of nanoparticles into a non-polar oil domain containing ligands, leading to (Phase I) interfacial complexation and formation of liquid threads, generating 3D structured multi-scale porous constructs upon (Phase II) ultra low-temperature freezing (−85 °C) and (Phase III) lyophilization, yielding core–shell filamentous threads composed of a porous core surrounded by a thin interfacially driven skin. (a)–(c) Reproduced with permission.40 Copyright 2018, Wiley-VCH.

It is important to highlight that to efficiently produce these tubular all-liquid structures, the speed of NPSs formation, assembly, and jamming needs to surpass the relaxation rate of fluids. Additionally, their binding energy must be sufficiently high to offer mechanical stability to the tubular liquid structures. Once these conditions are met, a nanometer-thin interfacially driven skin forms around the liquid thread. This skin consists of disordered assemblies of NPSs at the interface, reducing interfacial tension and ensuring the integrity of the non-equilibrium form of the liquid structure.⁴⁰ This development holds great promise for generating multifunctional liquid assemblies, enabling the creation of stimuli-responsive soft materials such as bicontinuous separation systems, all-liquid electronics, and biphasic reactors.^10,11,73,74

More interestingly, hybrid suspensions of GO with other nanomaterials, such as magnetic GO flakes (GO-Fe₃O₄) and 2,2,6,6-tetramethylpiperidine-1-oxyl (TEMPO)-oxidized CNFs, were also used to develop liquid threads through the liquid streaming approach.⁴⁰ This demonstrates the potential of GO as a universal carrier for liquid streaming, where other nanomaterials can be added to the aqueous phase. Accordingly, GO facilitates the interfacial assembly, while other nanomaterials in the system impart magnetic responsiveness or additional functionalities that can be leveraged for the final applications of these constructs. This potential of GO as a universal carrier for stabilizing hybrid systems in liquid streaming stems from its dual hydrophilic/hydrophobic nature.^19,40 It possesses hydrophilic functional groups such as carboxylic acid, hydroxyl, and epoxide functional groups, along with a hydrophobic basal plane. Consequently, GO acts as a surfactant, easily migrating to the interface and interacting with the ligand to form NPSs that subsequently stabilize the liquid streams.¹⁹ This minimizes interfacial tension, enhances the integrity of the liquid structures, and allows for the utilization of the inherent characteristics of the nanomaterials in the final assembly to create advanced functional materials.

In another study, researchers explored the potential of co-assembling GO and MXene (specifically, Ti₃C₂T_x) at the interface to create conductive tubular liquid worm-shaped threads (Fig. 6a).¹⁹ Ti₃C₂T_x on its own cannot generate liquid streams; instead, it breaks up into discrete droplets when its aqueous dispersion is injected into the hexane-POSS-NH₂ domain. It is worth noting that Ti₃C₂T_x flakes are negatively charged and possess various negative moieties like –O, –F, and –OH, which interact with the protonated amine groups of POSS-NH₂, leading to overlapping assemblies that jam at the interface.⁷⁵ However, these assemblies are not robust enough to maintain the non-equilibrium shape of liquid threads and end up breaking into discrete droplets. In contrast, when GO and Ti₃C₂T_x are co-assembled at the interface, GO rapidly and robustly assembles at the interface, forming a strong hybrid assembly with MXene and POSS-NH₂. This hybrid assembly can withstand applied compressive loads and maintain the integrity of the construct. This development is particularly practical for producing electrically conductive multi-scale porous aerogels through freezing and lyophilization of the formed liquid structures, offering potent electromagnetic interference (EMI) shielding capabilities.

	Fig. 6 (a) Hybrid GO/MXene liquid streams stabilized into a non-equilibrium filamentous form upon suppressing the instabilities and imposed compressive loads; in this case, the GO promptly assembled at the interface and preserved the integrity of the liquid stream in the presence of MXene. (a) Reproduced with permission.19 Copyright 2023, Wiley-VCH. (b) Formation of filamentous Janus liquid constructs with distinct composition/concentration of nanomaterials at opposite sides. (c) Janus structures with non-interfering magnetic and conductive domains were fabricated by merging streams of Ti3C2Tx/GO and GO/mGO. FESEM and EDX analysis of (d) Ti3C2Tx/GO:mGO/GO (conductive/magnetic worms) and (e) Ti3C2Tx/GO:GO Janus constructs (conductive/nonresponsive). (b)–(e) Reproduced with permission.3 Copyright 2023, Nature Publishing Group.

On the other hand, harnessing the potential of Plateau–Rayleigh instabilities in pure aqueous streams of MXenes allows for the creation of structured liquid beads (Fig. S2a–d, ESI†).¹⁹ When introducing the Ti₃C₂T_x stream into the hexane-POSS-NH₂ domain, the stream eventually breaks down into liquid droplets, forming MXene liquid beads after a certain distance. This is because the formed MXene NPS assemblies cannot adequately suppress the instabilities and the compressive loads imposed on the filaments.^76–78 However, the developed liquid beads were structured with a solid skin of NPS around the bulk of the constructs (Fig. S1a, ESI†). As a result, these liquid beads remained completely stable after streaming and did not merge even under constant shaking or aging. Thanks to the distinctive stability of liquid beads, the constructs developed were used as a liquid template to create conductive aerogel beads. These beads are characterized by their ultra-low density and are mechanically strengthened by a solid skin formed through liquid streaming and NPSs assemblies (Fig. S1e–g, ESI†).

In another study, Janus filamentous liquids were developed based on the concept of liquid streaming (Fig. 6b–e).³ These Janus structures, characterized by an anisotropic distribution of custom nanomaterials at opposite faces, are formed through the integration of two liquid streams within the hexane/ligand domain. The resulting liquid construct consists of two distinct sides, each housing a controllable composition and concentration of nanoparticles. This dual-sided composition allows for the utilization of the inherent characteristics of nanoparticles on each side, enabling on-demand responsiveness at independent sides of the liquid streams. To achieve this, Janus liquids with conductive, non-conductive, or magnetic sides were developed by merging streams of pure GO (non-responsive), GO/Ti₃C₂T_x (conductive), and GO/magnetic GO (magnetic). This method provides an opportunity to program the functionality of filamentous structured liquids, spanning from micro to macro scales, to produce smart soft materials with the desired responsiveness.

Stabilized liquid streams have also been used in liquid-in-liquid printing to design complex all-liquid 3D geometries with on-demand chemical or structural features.^6,41 To this end, 3D liquid-in-liquid printing of an aqueous suspension of carboxylated nanoparticles into a highly viscous silicone oil containing amphiphilic amine-terminated polymeric ligands, i.e., amine-terminated PDMS was introduced (Fig. 7a).⁶ Interestingly, these printed structures are responsive to their external environment, i.e., pH, possess a tunable lifetime, and exhibit a compartmentalized response to chemical cues. As a result, such a construct can be used in applications ranging from liquid electronics and biphasic reaction vessels to cell or active material encapsulation platforms.

	Fig. 7 (a) Liquid-in-liquid printing of aqueous phases containing carboxylated nanoparticles into highly viscous silicon oil. This method generates liquid constructs with a pre-defined 3D architecture upon NPS formation and creation of an elastic film at the interface. (a) Reproduced with permission.6 Copyright 2018, Wiley-VCH. (b) Schematic illustration of an all-fluidic device fabricated through interfacial complexation of nanoclay and polymeric surfactant, holding the integrity of the liquid constructs and allowing the flow of fluids through the channels. (c) The disconnection of microchannels in all-liquid microfluidic devices is possible by simply cutting the bridged connections. (b) and (c) Reproduced with permission.10 Copyright 2019, Nature Publishing Group. (d) Liquid-in-liquid printed constructs with tunable electrical conductivity, mechanical robustness, and morphology; the 3D printed threads form upon extruding the PEDOT:PSS-based aqueous suspensions into an oil phase enriched with amine-functionalized PDMS. (e) Schematic illustration of PSS-stabilized hydrogel threads within ligand-enriched silicone oil after UV curing of the acrylatemonomer. (d) and (e) Reproduced with permission.57 Copyright 2023, Nature Publishing Group.

Furthermore, reconfigurable all-liquid printed constructs have also been used to develop all-liquid microfluidic devices. In this regard, an assembly of NPSs, consisting of nanoclay and amine-functionalized polydimethylsiloxane (PDMS), was used to create 3D all-liquid fluidic devices.¹⁰ This device features spatially programmed functions and can be infinitely reconfigured (Fig. 7b). Employing 3D liquid-in-liquid printing, micropatterned 2D arrays were printed as channels for fluid flow. These channels are stabilized via a semi-permeable and anionic NPS membrane formed at the interface of liquids, allowing the on-demand chemical exchange in the system. Besides, the anionic membrane of the device can be further functionalized with small cationic molecules, nanocrystal catalysts, and enzymes for a wide range of applications. Interestingly, the device could undergo multi-step chemical transformations in the induced microchannels, allowing selective mass transportation across the assemblies at the liquid–liquid interface for task-oriented separation. These microfluidic devices can be automated using detectors, pumps, and appropriate control systems, demonstrating potential capabilities for learning and recognizing chemical logic (Fig. 7c).

In another attempt, conductive structures of poly (3,4-ethylenedioxythiophene):polystyrene sulfonate (PEDOT:PSS) were fabricated through 3D liquid-in-liquid printing, where an aqueous suspension of PEDOT:PSS was injected into an oil phase containing PDMS-NH₂ as a ligand. In this process, the colloidal PEDOT:PSS particles, consisting of a hydrophobic PEDOT core with a hydrophilic shell, electrostatically interact with the protonated amine group of PDMS-NH₃⁺ via the deprotonated sulfonate group R–SO₃⁻ of PSS upon injection into the ligand-enriched oil phase. This results in the formation of an elastic skin at the interface, which stabilizes the threads and maintains their non-equilibrium filamentous shape (Fig. 7d and e). To improve the mechanical strength of these conductive liquid threads, poly(ethylene glycol) diacrylate (PEG-DA) was added to the aqueous phase and crosslinked by being irradiated under 395 nm UV light. This approach enables the creation of mechanically robust and electrically conductive liquid structures suitable for biomedical applications, as well as customizable electro-microfluidic devices designed for implanting biochips for near-field communication within living organisms.⁵⁷

The role of hydrogen bonds in developing structured liquids has been recently demonstrated via interfacial complexation and jamming of amine-functionalized polyhedral oligomeric silsesquioxanes (POSS-nA) at the oil/water interface.¹⁰⁵ In this work, the assembly and jamming of POSS-nA molecules at the interface created a robust film that stabilized liquid droplets into non-equilibrium shapes (Fig. 15a). The results also highlighted the role of pH, POSS-nA concentration, and functionalization degree in modulating interfacial tension, revealing how these factors influence the formation of interfacial supramolecular films. At acidic conditions (pH 4), POSS-8A exhibits strong electrostatic repulsion due to its protonated amine groups, which prevents the particles from aggregating at the interface (Fig. 15b). This phenomenon is attributed to the discrete macroanionic nature of POSS-8A, carrying positive charges balanced by counter-anions, which leads to repulsive interactions among POSS units, hindering their close packing and assembly compression. As the pH increases, the study notes a transition in the behavior of POSS-8A. Beyond the pK_a value of the amine groups (around pH 9), the protonation of amine groups decreases, allowing for increased hydration and stabilization of POSS-8A at the interface through intermolecular hydrogen bonding (Fig. 15c and d). This results in a significant reduction of the equilibrium interfacial tension, indicating a stronger assembly and interfacial activity of amine-functionalized POSS at higher pH levels. The interfacial tension decreases further as the pH value increases, underscoring the importance of pH in modulating the assembly and stability of POSS molecules. Furthermore, through dynamic interfacial tension (IFT) measurements of aged aqueous droplets of POSS upon contact with a pure toluene solution, the influence of different levels of amine functionalization was examined. A comparison between POSS-4A, POSS-6A, and POSS-8A reveals that a higher degree of functionalization results in lower interfacial tensions and more stable assemblies.

	Fig. 15 (a) Interfacial assembly and Jamming of POSS-8A molecules at liquid–liquid interface. (b)–(d) The assembly of POSS-8A is pH-dependent. (a)–(d) Reproduced with permission.105 Copyright 2019, Wiley-VCH. (e and f) The assembly and jamming of P[5]AA at the oil/water interface is pH-dependent. (g) By adjusting the pH, the assembly of P[5]AA at the liquid–liquid interface can cycle between a jammed and an unjammed state. (e)–(g) Reproduced with permission.106 Copyright 2021, Royal Society of Chemistry.

In another study, Varshney et al.¹⁰⁶ developed reconfigurable structured liquids based on interfacial assembly and jamming of pillar[5]arene carboxylic acid (P[5]AA). The mechanism of this assembly can be explained based on hydrogen bonding between supramolecular species that can undergo successive jamming and unjamming states upon changing the pH (Fig. 15e–g). Interestingly, P[5]AA functions as a macrocyclic anion at elevated pH levels, where electrostatic repulsion between anions reduces interface stability, reflecting minimal interfacial activity. Conversely, at acidic pH levels of 2–4, P[5]AA assumes its protonated form (–COOH), enabling intermolecular hydrogen bonding and supramolecular assembly, which significantly enhances interface stability and lowers IFT to about 16 mN m⁻¹.

Fig. 15g showcases the transition from a jammed to an unjammed state in an aqueous droplet containing P[5]AA. Initially conducted at pH 10, a 20 mL aqueous solution of P[5]AA was introduced into 20 mL of toluene, forming a pendant droplet without any observable wrinkles during volume compression. However, the introduction of 2 mL (1 M) of trifluoroacetic acid (TFA) into the oil phase induced the formation of wrinkles upon volume reduction, attributable to hydrogen bonding interactions at acidic pH. Remarkably, the wrinkles disappeared, and the droplet reverted to its equilibrium shape upon the addition of an excess of triethylamine (4 mL; 1 M TEA), which elevated the solution's pH and disrupted the hydrogen bonding. The cycle of hydrogen bond-mediated assembly/jamming and disassembly/unjamming of P[5]AA aggregates was visually represented and quantified through surface coverage measurements after a 30-minute aging period of the droplet, following alternating additions of acid and base. Furthermore, pH-responsive molecular diffusion within a pendant droplet stabilized by P[5]AA aggregates was investigated, showcasing controlled release mechanisms dependent on the droplet's state. This was evidenced by the release patterns of a dye and the visual transition between locked and unlocked states of pendant droplets containing Rh6G dye. Consequently, this experiment highlights the potential of supramolecular assemblies to function as intelligent, responsive systems through deliberate manipulation of environmental conditions. This feature suggests these structured systems could serve as intelligent delivery systems, a topic that is thoroughly explored in Section 5 of this review article.

In another study, Hata and colleagues demonstrated that non-ionic crystalline surfactants, specifically n-octyl cello-oligosaccharide (Cell-C8), can assemble and jam at liquid–liquid interface, as shown in Fig. 16a.¹⁰⁷ The self-assembly of Cell-C8 is driven by intermolecular hydrogen bonding and van der Waals forces, showcasing remarkable mechanical and chemical stability under various conditions, including different solvents, high ionic strengths, and extreme pH levels.

	Fig. 16 (a) The assembly and jamming of Cell-C8 at the liquid–liquid interface can stabilize the interface by forming a solid-like monolayer of supermolecules at the interface. (a) Reproduced with permission.107 Copyright 2021, Elsevier. (b) A schematic illustration showing how the buckling phenomenon occurs due to the interfacial jamming of CEL-C6 at the n-undecane/water interface. (c) A sequence of images indicating changes in the shape of a pendant droplet filled with Cell-C6 as it is gradually siphoned off and then refilled. (b) and (c) Reproduced with permission.108 Copyright 2024, Elsevier.

Similarly, the interfacial jamming capabilities of nanocelluloses without the addition of any ligand to the system were studied.¹⁰⁸ For this aim, rectangular sheet-like synthetic nanocelluloses composed of cello-oligosaccharides were synthesized and functionalized with various moieties, including ethyl (CEL-C2), butyl (CEL-C4), and hexyl groups (CEL-C6), as well as without alkyl groups (CEL-OH). Despite their nonionic nature and the presence of alkyl groups, these synthetic nanocelluloses were water-dispersible. This was attributed to their flat surfaces being coated with terminal saccharide residues, ensuring effective hydration.

Interestingly, among all samples, CEL-C6 with longer alkyl chains was the only one capable of adsorbing and jamming at the n-undecane/water interface through hydrogen bonding and van der Waals forces (Fig. 16b). In this regard, upon contacting a pendant drop containing an aqueous phase of CEL-C6 in a nonpolar domain, visible wrinkles formed at the interface, showcasing the solid-like behavior of the interfacially driven skin (Fig. 16c). This behavior can be attributed to the unique structural properties of the hexylated nanocelluloses. First, due to its sheet-like structure, CEL-C6 has an exceptional surface area that pushes these materials to the interface to reduce the energy differences between two immiscible liquids. Second, CEL-C6 has a dual hydrophilic/hydrophobic nature that makes it capable of interacting with both immiscible phases at the liquid–liquid interface, further reducing the interfacial tension. Consequently, these findings hold promise for the advancement of all-liquid devices utilizing nanocellulose-based materials.

While π−π stacking and electrostatic forces might not match the strength of hydrogen bonding, their combination with each other or with other non-covalent forces has proven remarkably effective in fabricating supramolecular materials. This synergistic approach leverages the unique contributions of each interaction type to create more stable and functional supramolecular structures. In this context, a recent study explored three molecules capable of forming structured liquids through π−π stacking and/or electrostatic forces, aiming to demonstrate how these intermolecular interactions contribute to stabilizing the liquid–liquid interface.¹¹⁵ The selected molecules were 5,10,15,20-tetrakis(4-sulfonatophenyl)porphyrin (TPPS), the manganese(III) chloride derivative of TPPS (MnTPPS), and mellitic acid (MA), as depicted in Fig. 17a–c. However, as anticipated, these molecules showed different performances in stabilizing the interface and reducing the interfacial tension between two liquids (Fig. 17d–f). This variation can be elucidated based on their molecular structures and the resulting interactions at the oil/water interface.

	Fig. 17 (a) H6TPPS, (b) MnH4TPPS, and (c) MA chemical structures. Interfacial tension measurements and contraction relaxation tests were conducted on aqueous pendant drops containing (d) H6TPPS, (e) MnH4TPPS, and (f) MA in an external nonpolar domain of PS-NH2/toluene. Schematic illustration of interfacial assembly and jamming of systems containing a nonpolar domain of PS-NH2/toluene and an aqueous solution of (g) H6TPPS, (h) MnH4TPPS, and (i) MA against. (a)–(i) Reproduced with permission.115 Copyright 2019, Wiley-VCH.

For example, when an aqueous phase containing H₆TPPS meets the oil phase containing amine-terminated hydrophobic polymers (PS-NH₂), supramolecular polymer surfactants (SPSs) start forming at the interface. This process, powered by intermolecular electrostatic attractions and π–π stacking, leads to a decrease in interfacial tension and enables the liquids to maintain non-equilibrium shapes. In contrast, a similar molecule, MnH₄TPPS, acts more like a conventional surfactant by reducing interfacial tension without shaping liquids. This difference is partly attributed to one of the porphyrin surfaces in MnH₄TPPS being covered by a chloride ion, diminishing intermolecular interactions. Moreover, the size of the molecule plays a pivotal role in SPS formation. For instance, MA, with its weaker π–π stacking and a single benzene ring, lacks the necessary size to facilitate adequate intermolecular interactions for SPS formation (Fig. 17g–i).

In another investigation, the interfacial assembly and jamming of supramolecular polymer surfactants (SPSs) were examined through the color-changing properties of H₆TPPS, which vary based on its aggregation state at the interface.¹³ As detailed in Fig. 18, an aqueous solution of H₆TPPS is injected into an oil solution containing aminoethylaminopropyl isobutyl polyhedral oligomeric silsesquioxane (POSS-NHC₂H₄NH₂) dissolved in silicone oil. Upon contact, POSS-NHC₂H₄NH₂, acting as a potent surfactant, forms a monolayer at the interface to lower interfacial tension. H₆TPPS, once dissolved in water and transformed into H₄TPPS²⁻ along with H⁺ ions, migrates to the interface, where it electrostatically interacts with POSS-NHC₂H₄NH₂ to create in-situ SPS (H₄TPPS²⁻POSS-NH₂⁺C₂H₄NH₃⁺). This interaction prompts the formation of J-aggregates of H₄TPPS²⁻, which display a green color caused by edge-to-edge interactions, wherein the cationic nitrogen atoms neutralize the negatively charged SO₃⁻ groups. This green state marks the initial phase of pattern formation, as shown in Fig. 18b–d. Subsequently, as H₄TPPS²⁻ accumulates at the interface, it releases more protons to form H₂TPPS⁴⁻, which exhibits a light red color and interacts electrostatically with additional POSS-NH₂⁺C₂H₄NH₃⁺, transitioning into a different SPS structure (H-aggregates of H₂TPPS⁴⁻(POSS-NH₂⁺C₂H₄NH₃⁺)₂). This dynamic color change from J-aggregates to H-aggregates offers a visual mechanism to track the evolving structure and assembly of SPSs at the liquid–liquid interface, enabling real-time monitoring of the electrostatic interfacial dynamics.

	Fig. 18 (a) H4TPPS2− transitions to H2TPPS4− through interactions with POSS-NHC2H4NH2. (b)–(d) An overview of the 3D printing process and underling mechanisms that can visualize the assembly of J- and H-aggregates with different colours. (c) A schematic illustration of the interfacial assembly of SPSs at the liquid–liquid interface (H2TPPS4−(POSS-NH2+C2H4NH3+)2), showcasing a red-colored interface due to the formation of H-aggregates of H2TPPS4−. (d) A representation of the interfacial assembly of SPSs (H4TPPS2−POSS-NH2+C2H4NH3+), illustrating a green-colored interface as a result of J-aggregates of H4TPPS2− formation. (a)–(d) Reproduced with permission.13 Copyright 2021, Wiley-VCH.

Zheng et al.¹⁰⁹ reported a novel method for the host–guest interfacial assembly of supramolecular dendritic microcapsules within microfluidic media. Interestingly, they utilized a combination of hydrophilic and hydrophobic copolymers. The process involves the controlled co-assembly of these copolymers at the chloroform/water interface within a microfluidic droplet, which is then followed by host–guest complexation with cucurbit[8]uril (CB[8]). This results in the reversible crosslinking of the functionalized copolymers, leading to the formation of a polymeric skin at the interface, encapsulating the droplets, and forming monodisperse, supramolecular polymer-only microcapsules. Fig. 19a and b detail the mechanism and outcomes of this process, illustrating the encapsulation of water-soluble molecules (specifically Congo Red (CR)) by a dendritic copolymer (H1), and their transfer into an organic solvent. This is achieved through a microfluidic device where an aqueous phase of CB[8] and water-soluble poly(HEMA-co-StMV-co-FOA) (P1) overlaps with a stream of chloroform containing the dendritic copolymer H1. The result is the formation of chloroform-in-water microdroplets, where the interfacial assembly leads to the formation of a microcapsule skin, as evidenced by bright field and fluorescence microscopy (Fig. 19c and d). The successful encapsulation and structuring are further validated through fluorescence images, which confirm the entrapment of CR within the microcapsule wall. However, in the absence of CB[8], P1 and H1 failed to assemble at the interface, resulting in the formation of dense polymer beads of H1 upon solvent evaporation. This innovative approach not only showcases the feasibility of fabricating microcapsules via interfacial supramolecular host–guest polymerization but also highlights the efficiency of dendritic polymers in encapsulating small molecules. The utilization of microfluidic devices for this process underscores the precision and control achievable in the spatial organization and chemical functionalities of the resultant microcapsules, promising a wide range of applications from drug delivery to material encapsulation.^122,123

	Fig. 19 (a) A visual representation demonstrating the encapsulation of water-soluble CR by the dendritic copolymer, H1, followed by its transfer into an organic solvent. (b) A diagram illustrating the process of microdroplet formation through a microfluidic device, which includes an aqueous phase with CB[8] and P1 overlapping with a stream of chloroform. This chloroform contains the dendritic copolymer H1 with encapsulated CR, leading to the formation of chloroform-in-water microdroplets. (c) Optical and fluorescence microscopy images reveal the formation of uniform microdroplets, showing the interfacial assembly consisting of an inner H1 layer (with entrapped CR) and an outer P1 layer (labeled with fluorescein). (d) Bright-field microscopy images capture the creation of the microcapsule shell as the chloroform within the microdroplet evaporates, leading to a collapsed capsule-like structure. (a)–(d) Reproduced with permission.109 Copyright 2014, Nature Publishing Group. (e) Diagrams illustrating the process of microdroplet creation through microfluidic devices. (f) Images from laser scanning confocal microscopy showing empty microcapsules with component 2b assembled at the interface. Confocal microscopy images displaying microcapsules containing (g) fluorescein isothiocyanate-labeled dextran (FITC-dextran) and (h) encapsulated Escherichia coli cells. (i) Optical images of the microcapsules. (j) Schematic demonstration of the dehydration process applied to the microcapsules. (e)–(j) Reproduced with permission.111 Copyright 2012, American Association for the Advancement of Science.

Classical emulsification techniques often produce colloidal microcapsules with a wide size distribution. To narrow this distribution, researchers have increasingly turned to microfluidics for microcapsule production. In this regard, a one-step fabrication method for creating supramolecular microcapsules using microfluidic droplets was proposed to simplify and scale the production of microcapsules without sacrificing functionality or encapsulation efficiency (Fig. 19e).¹¹¹ Utilizing a host–guest system comprising polymer–gold (Au) nanoparticle composites and CB[8] ternary complexes, this approach allows for the generation of porous, dynamically stable microcapsules. In detail, microcapsules form as the oil phase strips away from a water phase containing CB[8], methyl viologen (MV)-functionalized Au nanoparticles, and naphthol-functionalized polymers. The Au nanoparticles migrate to the oil/water interface, where they facilitate host–guest interactions with the complementary polymers. This leads to the microcapsule's membrane formation, which can be used as cargo (Fig. 19f–h). Also, these uniformly sized microcapsules can be easily collected after the evaporation of the water droplets and retain their spherical shape during rehydration (Fig. 19i and j). The core content and shell composition of the microcapsules can be precisely controlled, marking a significant step in various applications, including drug delivery, diagnostics, and electronic displays.

Equally interesting, an approach to creating two-dimensional supramolecular organic frameworks (2D SOFs) via a host–guest interaction mechanism at a liquid–liquid interface was advanced recently.¹¹⁰ In this work, large-area, free-standing monolayers with excellent internal order were achieved by integrating a tris(methoxynaphthyl)-substituted truxene (Np-Trx) donor and a naphthalene diimide (MV-NDI) acceptor with CB[8] (Fig. 20a). To better understand the fabrication procedure, Fig. 20b–e illustrate the stepwise process of self-assembly and deposition for creating a 2D SOF. Initially, an aqueous solution with preassembled components is prepared, followed by the addition of a solvent containing the donor molecule. Over time, the two-phase system facilitates the assembly at the liquid–liquid interface. Finally, the solutions are drained to isolate the assembled structure. These schematics demonstrate the intricate methodology employed to achieve the highly ordered, monolayer 2D SOF, showcasing the interaction between the components throughout the assembly process.

	Fig. 20 (a) Schematic illustration of the assembled system containing Np-Trx, MV-NDI, and CB[8]. (b)–(e) Step-by-step assembly of these structured systems at the toluene/water interface. (b) This process begins with an aqueous solution containing preassembled MV-NDI and CB[8]. (c) The two-phase system is then formed by the addition of an Np-Trx/toluene phase to the aqueous phase. (d) The self-assembled structure is then formed at the interface after a 4-hour assembly period. (e) The final, isolated assembled structure is obtained after draining the solutions through a valve located at the bottom of the trough. (a)–(e) Reproduced with permission.110 Copyright 2015, American Chemical Society.

	Fig. 21 Photo-responsive structured liquids. (a) Photo-responsive structured liquids based on α-CD-modified Au nanoparticles and the oil-soluble Azo-terminated polymeric ligands. Under visible light, the jamming of NPSs at the oil/water interface forms stable wrinkles during droplet contraction, while UV irradiation at 365 nm causes wrinkles to vanish due to the unjammed state of NPSs. (a) Reproduced with permission.128 Copyright 2020, American Chemical Society. Optical image and schematic illustration of photo-responsive microcapsules based on Azo-PLLA/α-CD-modified Au NPSs (b) Under visible light and (c) after 12 min UV exposure. The capsules were loaded with RhB, and after the unjamming process, the release of RhB was visible. (b) and (c) Reproduced with permission.74 Copyright 2021, Wiley-VCH. (d) Photo-responsive host–guest interfacial assembly of Alg-β-CD and AzoC12 in the presence of cosurfactants, i.e., PVA and Span 60. TEM images of Alg-β-CD/AzoC12 assemblies (e) before UV-irradiation, (f) under UV-irradiation, and (g) after visible light-irradiation. (d)–(g) Reproduced with permission.129 Copyright 2023, Elsevier.

The photo-responsiveness of Azo-PLLA and Azo-PS-based NPSs and their assemblies was assessed through expansion/contraction tests of pendant drops. As depicted in Fig. 21a, under visible light irradiation, NPSs in both systems effectively jammed the oil/water interface, evident in the formation of stable wrinkles at the liquid–liquid interface upon droplet contraction. However, when exposed to UV light at a wavelength of 365 nm, the wrinkles gradually disappeared, and the droplet underwent relaxation, indicating the unjamming of the NPS assembly at the liquid–liquid interface. Here, before UV irradiation of Azo-PLLA-based samples, the jammed NPSs were firmly held at the interface due to interfacial host–guest interactions between trans-Azo-PLLA and αCD, along with hydrogen bonding between Azo-PLLA and water/αCD. However, under UV light, trans-Azo-PLLA underwent photoisomerization to form cis-Azo-PLLA, which could not interact with α-CD, resulting in an unjammed state at the liquid–liquid interface.

A similar study used the host–guest interactions between Azo-PLLA and α-CD-modified Au nanoparticles at the liquid–liquid interface to make structured photo-responsive microcapsules.⁷⁴ Notably, under UV irradiation, the Azo-PLLA/α-CD-modified Au NPSs could become unjammed, leading to the subsequent release of the cargo contained within the microcapsules into an external oil phase. As a proof of concept, the authors loaded these microcapsules with Rhodamine B (RhB) and quantitatively examined its release using fluorescence imaging (Fig. 21b and c). The results illustrated a time-dependent dissociation of RhB that could be further controlled/programmed by alternately exposing the samples to UV and visible light irradiation.

In another study, a photo-responsive host–guest interfacial assembly of alginate-modified β-cyclodextrin (Alg-β-CD) and azobenzene-functionalized alkyl chains (AzoC12) in the presence of cosurfactants, namely polyvinyl alcohol (PVA) and Span 60 was introduced.¹²⁹ In this work, Pickering emulsions were prepared by combining an aqueous phase containing PVA and Alg-β-CD with an oil phase containing AzoC12 oil and Span 60. The resulting soft supra-colloidal aggregates were then obtained through emulsion-solvent evaporation (Fig. 21d).

The supramolecular Alg-β-CD/AzoC12 polymer systems demonstrated exceptionally high interfacial activity, achieving an interfacial tension of ∼0.7 mN m⁻¹. This strong assembly is due to the robust interaction between β-CD and Azo at the oil/water interface. However, this elevated interfacial activity could be suppressed by UV irradiation. Before irradiation, the Azo moieties of trans-AzoC12 demonstrated a strong interaction with the hydrophobic moieties of Alg-β-CD through host–guest interactions. However, upon UV irradiation, trans-AzoC12 underwent photoisomerization to form cis-AzoC12, resulting in the cessation of host–guest interactions, as evidenced by interfacial tension measurements and TEM investigations (Fig. 21e–g).

Photo-responsive assemblies at the liquid–liquid interface have also been used for controllable heterogeneous catalysis. In this regard, Li et al.¹³⁰ introduced a novel light-responsive Pickering emulsion, leveraging Pd-supported silica nanoparticles, a water-soluble azobenzene-based ionic liquid surfactant (ILS) – specifically [C4AzoC2DMEA]Br, [C4AzoC4DMEA]Br, [C4AzoC6DMEA]Br, or [C4AzoC8DMEA]Br – along with n-octane and water. The experimental results showed that the initially stable Pickering emulsions experienced destabilization, leading to a full phase separation between water and oil when subjected to UV irradiation with stirring. However, subsequent irradiation with visible light after homogenization effectively restored the stable emulsions. This observed phase behavior is attributed to the adsorption of the azobenzene ionic liquid surfactant on the surface of nanoparticles and the light-responsive activity of the ionic liquid surfactant. The developed Pickering emulsion serves as a versatile microreactor, facilitating catalytic reactions, product separation, and the recycling of both emulsifiers and catalysts.

In this regard, thermo-responsive core–shell nanoparticles with a starch core and thermo-responsive polymer brush shell were introduced to stabilize Pickering hexane/water emulsions.¹³⁴ To synthesize thermo-responsive nanoparticles, the starch nanoparticles were first seeded with bromine (Br-SNP) and then grafted by PNIPAM through surface-initiated single electron transfer living radical polymerization (SI-SET-LRP) to get SNP-g-PNIPAM. The interfacial tension measurements showcased that pure starch nanoparticles seeded with bromine have no interfacial activity. However, after PNIPAM grafting, the resulting core–shell nanostructures significantly reduced the IFT between hexane and water and stabilized the Pickering emulsions. This Pickering system was stable at room temperature; however, it became completely unstable at temperatures higher than LCST due to the formation of nanoparticle clusters. Interestingly, the system was cycled to re-form stable Pickering emulsions by reducing the temperature below LCST, when the nanoparticle aggregates disappeared and SNP-g-PNIPAM regained their interfacial activity.

In another study, temperature-sensitive Janus silica nanoparticles featuring one silanol and one PNIPAM side were developed to stabilize a thermally responsive oil/water Pickering emulsion.¹³⁵ The efficacy and thermal responsiveness of these Janus silica nanoparticles in stabilizing tetradecane/water and crude oil/water systems were assessed compared to symmetric PNIPAM-modified silica nanoparticles. Strikingly, Janus nanoparticles exhibited a higher transition temperature (∼50 °C) at which a stable emulsion deteriorated, attributed to their superior hydrophilicity arising from the presence of silanol on one side. Beyond this temperature, the disruption of hydrogen bonds and the dehydration and collapse of PNIPAM chains result in the de-emulsification of the system.

More interestingly, Yu et al.¹³⁶ synthesized dual-responsive colloidal supramolecular microcapsules with both thermo- and photo-responsive properties by employing a droplet-based microfluidic approach. The microcapsules were formed by combining Methyl viologen-bearing colloidal particles (MCP) grafted with thermal-responsive PNIPAM, along with cucurbit[8]uril, and the photo-responsive Azo-terminated PVA (Azo-PVA). When the system's temperature exceeds the LCST of PNIPAM, the microcapsule's shell collapses, which creates larger inter-particle gaps and facilitates the release of the encapsulated cargo. Simultaneously, the incorporation of AP imparts photo-responsiveness to the microcapsules. Upon exposure to UV light, the photoisomerization of Azo units induces the detachment of supramolecular complexes from the interface, resulting in cargo release.

It is worth noting that the abovementioned studies were all designed based on the LCST of thermally responsive polymers, where the resulting emulsions remain stable at room temperature but undergo de-emulsification at elevated temperatures beyond the LCST. However, Ranka et al.¹³⁷ reported a system that operates in the reverse direction, with the developed thermally responsive emulsions remaining stable at high temperatures and degrading at room temperature. In their study, a stable emulsion system was created using silica nanoparticles stabilized by schizophrenic di-block copolymers comprising poly(N-isopropyl methacrylamide) (PNIPMAM) as the LCST block (42 °C) and poly(sulfobetaine methacrylamide) (PSBMA) as the upper critical solution temperature (UCST) block (16 °C). At room temperature, both blocks exhibited hydrophilic and water-soluble characteristics, causing the decorated nanoparticles to have a high affinity for water molecules, display no surface activity, and fully disperse into the water phase. However, at an elevated temperature of 65 °C, the N-alkyl methacrylamide on the LCST block underwent dehydration, competing with the hydration of the polyzwitterionic block on the UCST block. This dynamic interplay resulted in the formation of interfacially active nanoparticles capable of stabilizing an oil-in-water emulsion at higher temperatures, as illustrated in Fig. 22a. Such a system holds potential for various applications, including enhanced oil recovery, where an emulsion system can be formed at elevated temperatures in oil reservoirs to effectively absorb oil. However, upon reaching the surface and experiencing reduced temperatures, the de-emulsification process occurs, facilitating oil–water separation without incurring additional costs for the emulsifier/oil/water separation process.

	Fig. 22 Thermo- and electric-field- responsive structured liquids. (a) Thermo-responsive Pickering emulsions based on silica nanoparticles stabilized by schizophrenic di-block copolymers comprising PNIPMAM as the LCST block and PSBMA as the UCST block. (a) Reproduced with permission.137 Copyright 2017, American Chemical Society. (b) Controlling the assembly of positively charged TTMA-Au nanoparticles between two electrolyte solutions by applying a voltage across the interface. The electrolyte solutions are an aqueous suspension of NaCl and dichloroethane containing BTPPATPFB. The nanoparticles tended to remain in the bulk phase without applying voltage. However, with a slightly positive electric potential difference, the nanoparticles assemble at the liquid–liquid interface. (c) and (d) GISAXS analysis confirms the assembly of nanoparticles at the liquid–liquid interface upon applying different electric potentials at the interface. (b)–(d) Reproduced with permission.138 Copyright 2014, American Chemical Society. (e) Using an applied electric field to deform structured liquids consisting of carboxylated polystyrene nanoparticles/PDMS-NH2 NPSs. Upon applying an electric field, the structured droplets deformed and adopted an ellipsoid shape with a larger surface area. (f) This ellipsoid shape remained stable over time. (e) and (f) Reproduced with permission.41 Copyright 2013, American Association for the Advancement of Science.

In this regard, strategies such as applying an external electric field or introducing deep potential walls through electric screening have proven effective in suppressing the electrostatic repulsion forces between charged particles at the interface of two immiscible electrolyte solutions.¹²⁴ Su et al.¹³⁹ explored this theory, having designed a liquid–liquid system with mercaptosuccinic acid (MOA)-stabilized Au nanoparticles at the water/1,2-dichloroethane (DCE) interface. Their work demonstrated that the concentration of nanoparticles at the liquid–liquid interface could be reversibly controlled by applying a potential bias. Further strengthening this theory, Janjua et al.¹⁴⁰ conducted research on the assembly of micro-to-nanoscale particles with diverse shapes, e.g., rods, cubes, and spheres, at both liquid–liquid and air–liquid interfaces. They effectively tuned the assembly by changing the strength and frequency of the applied electric field. Together, these studies highlight the intricate interplay between capillary and electrostatic forces, offering insights into controlling particle behavior at liquid interfaces.

Bera et al.¹³⁸ also demonstrated that the interaction among nanoparticles at the liquid–liquid interface can be adjusted by applying a voltage across the interface. Their experimental system included 2 nm Au nanoparticles coated with positively charged trimethylammonium-terminated ligands (TTMA-Au nanoparticles) and two electrolyte solutions: water containing sodium chloride (NaCl) and dichloroethane containing bis(triphenylphosphoranylidene) ammonium tetrakis pentafluorophenyl borate (BTPPATPFB) (Fig. 22b). In the first step, they performed interfacial tension measurements under different voltages. When the electric potential difference (Δϕ^w–o = ϕ(w) − ϕ(o)) between the aqueous (w) and organic (o) solutions remained negative, the calculated interfacial tension was similar to that of a pure water/organic system. This suggests that the nanoparticles tended to remain in the bulk phase, unable to reach the liquid–liquid interface to minimize interfacial energy. However, when a slightly positive Δϕ^w–o was applied, for instance, +30 mV, the calculated interfacial tension markedly decreased due to nanoparticle assembly at the liquid–liquid interface. Such a voltage-dependent assembly of TTMA-Au nanoparticles was further validated using the grazing-incidence small-angle X-ray scattering (GISAXS) technique, as illustrated in Fig. 22c and d. The findings revealed that, regardless of the applied electric potential, the particles consistently formed a hexagonal arrangement at the 2D interface. However, the distance between nanoparticles and the in-plane correlation length depend on the applied voltage and can be effectively controlled.

More recently, Russell and co-workers utilized an external electric field to manipulate the configuration of structured liquids formed by assemblies of NPSs at the liquid–liquid interface.⁴¹ Within their system, an aqueous droplet containing carboxylated polystyrene nanoparticles was suspended in a bath containing silicon oil and PDMS-NH₂. Within this system, PDMS molecules diffused into the interface, converting the previously negative charge of the oil/water interface to a positive one by forming PDMS-NH₃⁺ layers. This shift lowered the interfacial energy, thereby promoting the migration of functionalized polystyrene nanoparticles towards the interface. The carboxylic functional groups of the polystyrene nanoparticles were then deprotonated and electrostatically bonded to protonated amine groups of PDMS-NH₃⁺. This process created ligand/nanoparticle assemblies at the interface, which, upon jamming, stabilized the water droplets and formed a solid-like skin around the droplet.

Upon applying an external electric field with a strength of 4.6 kV cm⁻¹, the water droplets deformed, adopting an ellipsoid shape with a larger surface area (see Fig. 22e). With an increased surface area, more NPSs formed at the interface, expanding the solid skin around the droplet to cover the entire structure. Upon the removal of the electric field, the water droplet could not return to its original shape, as the newly expanded skin pinned the droplet to its ellipsoid shape. Interestingly, this ellipsoid shape remained stable over time, up to one month, and the droplet could not revert to its original form (Fig. 22f). Additionally, it should be noted that in the absence of PDMS-NH₂ as a ligand, the deformed droplet returned to its original spherical shape after the removal of the external electric field, underscoring the role of NPS assemblies and jamming in stabilizing liquids in highly non-equilibrium configurations.

Russell and his colleagues were pioneers in the creation of ferromagnetic structured liquids by utilizing the concept of NPS assembly and jamming at a liquid–liquid interface.¹⁴³ In their study, a suspension of carboxylated Fe₃O₄ in water was immersed in an inorganic phase containing POSS-NH₂ and toluene, resulting in the formation of structured ferromagnetic liquids. During immersion, the deprotonated carboxylic groups of magnetic nanoparticles and protonated amine groups of POSS-NH₂ electrostatically interacted, forming NPSs at the liquid–liquid interface. As the concentration of NPSs at the interface increased, i.e., achieved by increasing the magnetic nanoparticles' concentration in the bulk aqueous phase or reducing the pH, the NPSs effectively jammed at the interface, transforming the ferrofluid droplet into a structured ferromagnetic liquid droplet.

The transition from paramagnetic to ferromagnetic behavior was investigated through the analysis of magnetic hysteresis loops for identical ferrofluid droplets with and without POSS-NH₂ (Fig. 23a). The magnetization, represented by the magnetic moment per unit volume (M_s), remained consistent between both samples, given the comparable total number of magnetic nanoparticles. However, the sample exhibiting interfacial jamming of magnetic NPSs displayed finite magnetic remanence (indicated by M_r, the residual magnetization in a material after the external magnetic field is removed) and coercivity (represented by H_c, the magnetic field strength required to demagnetize a material), which are key attributes of ferromagnetic materials. To further elaborate on this phenomenon, the study explored the influence of magnetic nanoparticle concentration in the bulk phase, pH variations, and droplet size. The results revealed that upon jamming, the resulting structured liquids exhibited ferromagnetic traits, a behavior further amplified by enhancing the surface coverage of magnetic NPSs at the liquid–liquid interface.

	Fig. 23 Magnetic-field responsive structured liquids. (a) The transition from paramagnetic to ferromagnetic behavior for the structured system containing carboxylated Fe3O4/POSS-NH2 NPSs. (b) The developed ferromagnetic liquids can respond to a gradient magnetic field produced by an aluminum solenoid. (c) These ferromagnetic liquid cylinders also showcase N–N, S–S, and N–S dipole interactions similar to solid magnets. (a)–(c) Reproduced with permission.143 Copyright 2019, American Association for the Advancement of Science. (d) A schematic illustration of a ferromagnetic liquid containing OA-coated Fe3O4 nanoparticles/amine-functionalized silica nanoparticles exposed to an external magnetic field. (e) This magneto-responsive structured liquid can be used as a carrier to trap, transport, and release other oil-based droplets immersed in the water phase. (d) and (e) Reproduced with permission.5 Copyright 2022, American Chemical Society. (f) The magnetic responsiveness and relaxation of surfactant assemblies consisting of Fe3O4 and a zwitterionic ligand, i.e., 2-(3-butyl-1H-imidazol-3-ium-1-yl)acetate in DMF, and a complementary Lewis basic PDMS-NH2 in PDMS oil. (f) Reproduced with permission.144 Copyright 2018, American Association for the Advancement of Science.

To further elaborate on this idea, the study mentioned above also focused on developing ferromagnetic liquid cylinders with a 2:1 aspect ratio through microfluidic injection of the aqueous phase containing functionalized magnetic nanoparticles into a toluene/POSS-NH₂ bath. Intriguingly, when subjected to a gradient magnetic field generated by the aluminum solenoid, the produced liquid cylinder demonstrated displacement within an all-liquid medium (Fig. 23b). Additionally, as illustrated in Fig. 23c, these ferromagnetic liquid cylinders exhibited behavior akin to solid magnets, showcasing N–N, S–S, and N–S dipole interactions. Notably, magnetized liquid cylinders that were initially separated by 1 mm were observed to attract each other through N–S dipole interactions, a phenomenon that holds potential in fields where the flowability of structured liquids and magnetic-like behavior resembling traditional solid magnets are crucial – such as in liquid robots, energy-dissipative assemblies, and programmable liquid constructs.

In another study, researchers used a structured liquid composed of oleic acid (OA)-coated Fe₃O₄ nanoparticles dispersed in mineral oil, combined with an aqueous suspension of amine-functionalized silica nanoparticles. This innovative mixture was utilized to develop a novel ferromagnetic liquid (FML).⁵ In the absence of a magnetic field, OA-coated Fe₃O₄ nanoparticles assemble at the interface; here, OA undergoes deprotonation and acts as a ligand, electrostatically interacting with the protonated amine group of silica nanoparticles to form a solid-like interface at the liquid–liquid contact line. Upon applying an external magnetic field, the interface deforms, and additional assemblies of silica/magnetic nanoparticles occur at the interface, expanding the solid skin. This expanded skin retains the structured shape even after the removal of the magnetic field, and the deformed structure cannot return to its original shape (Fig. 23d).

Interestingly, this category of magneto-responsive structured liquids was utilized as a carrier to trap, transport, and release other oil-based droplets immersed in the water phase (Fig. 23e). It is worth noting that these droplets are inherently miscible with mineral oil, which is the main constituent of the developed magnetic structured liquids. However, due to the hydrophilic nature of capped groups of silica nanoparticles, which are densely packed at the interface of these structured liquids, the oil cargo cannot mix with the magnetic liquid constructs. Therefore, this class of magneto-responsive structures can be considered a novel category of all-liquid robots that can be used to transfer inorganic oil-based contaminants in water-based mediums.

Zhang et al.,¹⁴⁴ on the other hand, developed magnetic-magneto all-oil reconfigurable structured liquids. Their system consists of a cationic/Lewis acidic dispersion of Fe₃O₄ in N,N-dimethylformamide (DMF), and a complementary Lewis basic PDMS-NH₂ polymer (5% w/w) as a solution in PDMS oil. The Lewis acid–Lewis base pairing of the PDMS-NH₂ and cationic Fe₃O₄ was reported to produce mesoscopic polymer surfactant at the liquid–liquid interface and significantly reduce the interfacial tension between two immiscible phases (Fig. S8a, ESI†). The assembly of these polymer surfactants at the interface was irreversible, as evident in the contraction test of a pendant drop. In this case, upon contracting a droplet of DMF/Fe₃O₄ in a bath of PDMS-NH₂/oil, obvious wrinkles were formed at the interface that remained stable over time. However, by the addition of a competitive zwitterionic ligand, i.e., 2-(3-butyl-1H-imidazol-3-ium-1-yl)acetate, to DMF, the system was able to return to its liquid state, as evidenced by the dissipation of wrinkles upon the contraction test of a pendant drop (Fig. S8b and c, ESI†). The authors reasoned that the transition from solid-like to liquid-like behavior at the interface is due to the reversible binding of the zwitterionic ligand to cationic Fe₃O₄, which was competitive with the binding of PDMS-NH₂ (Fig. S8c, ESI†).

The magnetic responsiveness and relaxation of surfactant assemblies upon applying an external magnetic field were also investigated in the same study. In this regard, placing the pendant droplet between two magnets of opposite polarity resulted in the displacement of the assembled structure in space, causing the encapsulated fluid to expand along the axis of the magnetic field. As a result, the interface was compressed along the axis normal to the field, and the interfacial tension was increased. For instance, when the concentration of the zwitterionic ligand was 2.5 mg mL⁻¹, the interfacial tension was increased from 4 to 5.6 mN m⁻¹ upon applying the magnetic field. However, the interfacial characteristics of the droplet were returned to its original state when the external magnetic field was removed (Fig. 23f).

Russell and his colleagues were pioneers in developing pH-responsive structured liquids that capitalize on the reversible transition between protonated and deprotonated states of nanoparticles/ligands at a liquid–liquid interface.⁸ Their system comprised an aqueous dispersion of carboxylated polystyrene nanoparticles within a nonpolar domain containing toluene/PDMS-NH₂. At a pH of approximately 5 or higher, the carboxylic functional groups of nanoparticles undergo deprotonation, engaging in electrostatic interaction with protonated amine groups of the ligands to establish stable NPSs at the liquid–liquid interface. However, reducing the pH to near or below the effective pK_a of carboxylic functional groups prevents their deprotonation. This impedes the electrostatic interaction responsible for NPS formation, leading to the ejection of nanoparticles from the interface. The pH responsiveness of NPSs is intriguingly demonstrated by applying an external electric field (Fig. 24a and b). Under this field and at high pH, droplets containing carboxylated nanoparticles deform into an ellipsoid with an aspect ratio of 1.78, as discussed in the previous section on the electric-field responsive structures. Upon removing the external field, the aqueous phase retains its ellipsoidal shape and does not revert to its spherical form. However, at lower pH values than the pK_a of carboxylic functional groups, the droplet can be reconfigured to its spherical shape after removing the electric field. In this lower pH range, NPSs are terminated, and nanoparticles are pushed away from the interface, allowing the droplet to return to its original spherical shape with the lowest possible surface area.

	Fig. 24 pH-responsive structured liquids. (a) and (b) The pH responsiveness of NPSs, i.e., PS-NH2/CNC-OSO3H, is demonstrated by applying an external electric field. (a) and (b) Reproduced with permission.8 Copyright 2016, Wiley-VCH. (c) Various states of charge-reversible (CH3)2N–(CH2)10COONa surfactants result in precise control over stability, viscosity, and droplet size in n-octane-in-water emulsions. (c) Reproduced with permission.146 Copyright 2021, Wiley-VCH.

The impact of pH on the formation and strength of NPSs was also emphasized in a liquid modeling project.⁹ In this investigation, NPS formation relies on the electrostatic interaction between protonated amine-terminated polystyrene (PS-NH₃⁺) and deprotonated sulfate-functionalized cellulose nanocrystals (CNC-OSO₃⁻) at the liquid–liquid interface. At a pH of around 3, the assembly and jamming of NPSs are robust enough to yield all-liquid objects that replicate the shape and details of the mold within an external oil phase. However, with an increase in pH, the surface coverage of CNC at the interface diminishes, leading to deformation and a loss of fidelity in the molded structures.

In another study, pH-responsive oil-in-water Pickering emulsions were presented based on negatively charged SiO₂ nanoparticles and a zwitterionic carboxyl betaine surfactant as a stabilizer (Fig. S9a–c, ESI†).¹⁴⁷ Under acidic conditions, the ligand assumes a cationic state, adsorbing onto SiO₂ nanoparticles to stabilize the emulsions. However, with an increase in pH beyond 8.5, the ligands undergo a conversion to the zwitterionic state and subsequently desorb from the surfaces of the nanoparticles, leading to demulsification, as shown in Fig. S9d (ESI†).

Developing charge-reversible surfactants provides another avenue for fabricating pH-responsive emulsions. In a recent study, such a surfactant, namely (CH₃)₂N–(CH₂)₁₀COONa, was synthesized and coupled with silica nanoparticles to stabilize a smart pH-responsive n-octane-in-water emulsion.¹⁴⁶ Three scenarios for this system were considered based on pH (Fig. 24c). At a high pH of 9.3, the surfactant adopts an anionic carboxylate state. This anionic state, coupled with negatively charged silica nanoparticles, co-stabilizes low-viscous oil-in-water emulsions. Conversely, under a low pH condition of 4.1, the surfactant transforms into a cationic form, creating amine salt that can render the previously negatively charged silica nanoparticles hydrophobic and stabilize a viscous oil-in-water Pickering emulsion. Furthermore, at a neutral pH (7.5), the surfactant undergoes conversion to a zwitterionic form. In this state, it weakly hydrophobized the silica particles, stabilizing oil-in-water Pickering emulsions characterized by large droplet size. Moreover, demulsification can be swiftly triggered by pH adjustments. As a result, this system exhibits the capability for intelligent and precise control of stability, viscosity, and droplet size in emulsions.

In another study, Luo et al.¹⁴⁸ devised Janus zwitterionic nanoplates capable of transitioning between a locked and unlocked state at oil/water interfaces by adjusting surface charges through pH changes. To achieve this, they initially functionalized negatively charged kaolinite nanoparticles with a cationic polymer, specifically poly[2-(dimethylamino)ethyl methacrylate] (PDMAEMA), followed by grafting with hydrophobic poly(lauryl methacrylate) (PLMA). These chemical moieties promote intermolecular attraction among nanoparticles in the locked state through the interaction of oppositely charged groups. Conversely, a robust electrostatic repulsion occurs when PDMAEMA polymers are fully protonated at low pH, representing the unlocked state.

Sun et al.¹⁵¹ were pioneers in the development of redox-responsive structured liquids, focusing on the interfacial interaction between an electron-deficient acceptor dissolved in water and an electron-rich donor dissolved in toluene at a liquid–liquid interface. In their approach, a charge transfer interaction involving methyl viologen (MV²⁺ dissolved in water) and naphthalene-terminated poly-L-lactic acid (NT-PLLA dissolved in toluene) was employed to create an interfacial skin at the liquid–liquid interface (see Fig. S10a, ESI†). During the compression and relaxation test of a pendant drop, visible wrinkles formed at the interface, indicating the jammed state of these assemblies. However, the introduction of sodium dithionite (Na₂S₂O₄) as a reductant led to MV²⁺ accepting an electron and transforming into a radical cation in an electron-rich state, denoted as MV⁺˙. In this reduced state, MV⁺˙ is no longer an electron acceptor, preventing interaction with NT-PLLA and resulting in a relaxed and unjammed state at the liquid–liquid interface (see Fig. S10b, ESI†). Nevertheless, the reduced MV⁺˙ can be oxidized and converted back to MV²⁺ by using sodium hypochlorite (NaClO) as the oxidant. As a result, the interfacial interactions between MV²⁺ and NT-PLLA can be revived, facilitating the reassembly and jamming of MV²⁺/NT-PLLA at the liquid–liquid interface (Fig. S10a and b, ESI†).

Shortly after, the same research group developed redox-responsive all-liquid constructs based on ferrocene-terminated PLLA (Fc-PLLA) and β-cyclodextrin (β-CD) functionalized Au nanoparticles.¹⁵² Normally, uncharged Fc is bound into the cavity of β-CD and forms NPSs through host–guest interaction at the toluene/water interface (Fig. 25a). However, by oxidizing Fc into the ferrocene cation (Fc⁺), host–guest interaction is disrupted, and NPSs are disassembled at the interface. As a result, the structured liquids lose their structure and adopt a more liquid-like behavior. This jamming/unjamming transition can be repeatedly cycled with the sequential addition of NaClO as the oxidant and Na₂S₂O₄ as the reductant. Interestingly, this redox-responsive host–guest interaction at the liquid–liquid interface was also combined with the concept of liquid-in-liquid printing to fabricate NPS-stabilized aqueous tubules in toluene (Fig. 25b). As shown in Fig. 25c and d, these aqueous tubules can be loaded with a cargo, here Rhodamine B, that can be released upon oxidizing Fc in a reconfigurable manner.

	Fig. 25 Redox-responsive structured liquids. (a) Redox-responsive all-liquid constructs based on Fc-PLLA and β-CD-functionalized Au nanoparticles. (b) This redox-responsive host–guest interaction at the liquid–liquid interface was also integrated with the liquid-in-liquid printing approach to make aqueous tubules stabilized by NPSs within toluene. (c) These aqueous tubules have the capacity to encapsulate a cargo, such as Rhodamine B. (d) This cargo can be selectively released through the oxidation of Fc in a reconfigurable manner. (a)–(d) Reproduced with permission.152 Copyright 2021, American Chemical Society. (e) Reconfigurable structured liquids based on the interfacial assembly and jamming of CNC-based NPS. This system is responsive to various stimuli, including redox potential, pH, and voltage, offering multiple routes to control the assembly of NPSs at the interface and cycle them from a jammed to an unjammed state. (e) Reproduced with permission.153 Copyright 2023, Wiley-VCH.

More recently, Yang et al.¹⁵³ have introduced reconfigurable structured liquids through the interfacial assembly and jamming of CNC-based NPS. Their approach hinged on electrostatic interactions between CNC–OSO₃⁻ and ferrocenium-terminated polystyrene (PS-Fc⁺) at the oil–water interface (see Fig. 25e). Importantly, Fc⁺ is responsive to various stimuli, including redox potential, pH, and voltage, offering multiple routes to control the assembly of NPS at the interface and cycle them from a jammed to an unjammed state. The pH-dependent interfacial activity of PS-Fc mirrors that observed with amine-functionalized ligands discussed in previous sections. In acidic conditions, PS-Fc undergoes protonation, facilitating its assembly at the interface and reducing the interfacial tension. However, increasing the pH through the addition of NaOH induces deprotonation, which significantly suppresses the interfacial activity of PS-Fc. This leads to the subsequent disruption of NPSs at the liquid–liquid interface (Fig. S10c, ESI†). On the other hand, the redox responsiveness of CNC–OSO₃⁻/PS-Fc⁺ NPS was analyzed by reducing the Fc⁺ group to the Fc group using hydrazine hydrate (N₂H₄) as the main reduction agent. Furthermore, the easily achievable and repeatable electrochemical stimulus was also employed to reconfigure these structures, presenting a novel way to control and recycle structured emulsions (see Fig. S10d, ESI†).

	Fig. 26 (a) The electrostatic interaction between deprotonated ligands (positively charged) and negatively charged MXene nanosheets stabilizes conductive liquid threads, and (b) and (c) fabricates complex geometries that can be used for all liquid electronics. (a)–(c) Reproduced with permission.76 Copyright 2019, American Chemical Society. (d)–(f) The electrostatic interaction of negatively charged PS-COOH nanoparticles with positively charged PAH stabilizes the aqueous stream of Dextran/PEG into an aqueous bath of PEG. The printed tubules' walls consist of tiny colloidosomes. (g) Water-in-water printing is versatile, and various nanoparticles, including GO and Fe3O4 with magnetic characteristics and TiO2 with photocatalytic activity, were integrated into this printing approach. The resulting magnetoresponsive liquid constructs were used to absorb and catalytically remove rhodamine B, an environmental pollutant. In this scenario, the fabricated liquid robots can move to a specific location using an external magnetic field, absorb the pollutants, and, upon exposure to UV light, degrade the pollutants due to the catalytic activity of TiO2. (d)–(g) Reproduced with permission.18 Copyright 2022, American Chemical Society.

Later on, capitalizing on 3D liquid-in-liquid printing of MXene-based aqueous suspension in a nonpolar domain containing butylamine and toluene, Zettl and co-workers demonstrated the first reported all-liquid on-and-off switches and sensors.¹⁵ In their work, they demonstrated that a stabilized aqueous stream of MXene can be used to connect to electrodes and conduct electricity, similar to what we see for normal metal-based connectors. However, the liquid connectors can easily be cleaved to turn off the circuit and then get back into the initial state by only adding a few droplets of MXene aqueous suspension to repair the MXene tubular wires (Fig. S11a, ESI†). It is worth noting that although the idea of all-liquid circuits seems impractical and in its infant stage, such a system provides an opportunity to reconfigure electronic systems indefinitely and even provides a pathway to recycle electronics easily at the end of their life.

In another interesting study, Yan et al.¹⁶ explored the interfacial assembly of polyelectrolytes to develop structured-liquid batteries. The interfacial assembly was achieved in a biphasic system containing two aqueous phases: one containing dextran and poly(sodium 4-styrenesulfonate) (PSS-Na) and the other containing PEG and poly (diallyldimethylammonium chloride) (PDADMA-Cl). As active materials, cationic methyl viologen was added to the dextran phase to form the anolyte, while the catholyte was designed by the addition of ferrocyanide to the PEG phase (Fig. S11b, ESI†). Upon reduction, methyl viologen forms a radical cation, whereas, upon oxidation, ferrocyanide changes its oxidation state but remains anionic. This dual behavior enables the developed batteries to store energy and release it on demand effectively.

To regulate the formation of ion pairs and slow down the rate of crossovers within the cell, NH₄Cl was added to the system. This adjustment allowed the liquid batteries to maintain functionality over hundreds of cycles. Crucially, the power output of these designed batteries was comparable to that found in biological circuits and sensors, highlighting the significant potential of this interfacially driven energy storage system for future applications.

The use of structured liquids for absorption, separation, and catalytic reduction of environmental pollutants is still in its infancy, with many possibilities yet to be explored. In this context, recent advancements, such as the development of aqueous liquid robots designed to capture and catalytically reduce environmental pollutants, highlight the potential of this emerging field.¹⁸ These liquid robots typically consist of a biphasic system containing two aqueous phases, with one phase referred to as the aqueous ink (10 wt% 10000 g mol⁻¹ Dextran and 5 wt% 8000 g mol⁻¹ PEG) being printed inside an external bath (25 wt% 8000 g mol⁻¹ PEG).

To stabilize the liquid threads of the ink in the external bath, nanoparticles or polymers with complementary functional groups were added to each phase, resulting in interfacial complexation that stabilized these non-equilibrium shapes. Specifically, the addition of negatively charged carboxyl-functionalized polystyrene (PS-COOH) nanoparticles to the ink and positively charged poly(allylamine hydrochloride) (PAH) to the bath phase was reported to stabilize these nanoequilibrium shapes and enable all-aqueous liquid-in-liquid printing. In this context, the tubules' walls consist of tiny colloidosomes, resembling blood vessels lined with endothelial cells (Fig. 26d–f). These colloidosomes are formed through the stabilization of interfacial phase-separated small droplets using oppositely charged nanoparticles and polyelectrolytes.

Notably, this method is versatile, as various nanoparticles with different functionalities were employed to stabilize these all-aqueous systems. Moreover, the system can serve as a carrier for various materials to perform diverse reactions in situ within the printed structures. For example, the interfacial complexation of Fe₃O₄ nanoparticles and GO nanoflakes with abundant carboxylic functional groups demonstrated the fabrication of magnetoresponsive all-aqueous constructs capable of gripping, transporting, and releasing objects in an external aqueous medium. Furthermore, the addition of TiO₂ to the ink phase of these magnetoresponsive liquid constructs led to the development of a programmable liquid robot capable of absorbing and catalytically removing rhodamine B, an environmental pollutant. In this scenario, the fabricated liquid robots can move to a specific location using an external magnetic field, absorb the pollutants, and, upon exposure to UV light, degrade the pollutants due to the catalytic activity of TiO₂ (Fig. 26g).

The all-aqueous liquid robots discussed here offer an effective and environmentally friendly approach to water purification. Additionally, it can initiate target reactions as needed in limited spaces, such as biomedical applications like targeted photothermal therapy for tumor cells within confined areas in vivo.

	Fig. 27 (a)–(c) Various cationic species can be selectively anchored to the walls of all-liquid microfluidic channels composed of amine-functionalized PDMS/nanoclay NPSs. (d) The oxidation of TMB in the presence of H2O2 yielded a yellow-colored diimine product. (e) The oxidative coupling of phenol and 4-AAP produced a red-colored quinone–imine conjugate that could also be visually tracked in the liquid channels. (a)–(e) Reproduced with permission.10 Copyright 2019, Nature Publishing Group.

Capitalizing on this strategy, all-liquid microfluidic devices containing cationic HRP were also utilized for a variety of reactions.¹⁰ For instance, when a solution containing 3,3′,5,5′-tetramethylbenzidine (TMB) and hydrogen peroxide (H₂O₂) was passed through the fabricated channels, the TMB molecules oxidized and transformed into a yellow-colored diimine product, easily traceable in the microchannels (Fig. 27d). In another instance, a solution of phenol, 4-aminoantipyrine (4-AAP), and H₂O₂ was pumped through the microfluidic device, leading to the oxidative coupling of phenol and 4-AAP, producing a red-colored quinone–imine conjugate that could also be visually tracked in the liquid channels (Fig. 27e). Considering the success of these reactions, all-liquid microfluidic devices with exceptional reconfigurability can be envisioned for future applications in designing chemical logic and learning platforms.

In another interesting study, Liu et al.²¹ developed all-liquid microfluidic devices with multilayer walls for enzyme immobilization (Fig. 28a). In their study, aqueous tubes of CNCs were first stabilized and printed inside a nonpolar bath containing POSS-NH₂. The pH of the aqueous suspension was set at 3 to ensure strong electrostatic interaction between the deprotonated CNC sulfate ester groups (–OSO₃⁻) and protonated oligomeramine groups (NH₃⁺) at the oil/water interface. This electrostatic interaction forms CNC-based NPS at the liquid–liquid interface, which, upon jamming, stabilizes the liquid threads in silicone oil, thus creating microfluidic devices.

	Fig. 28 (a) Fabrication of all-liquid microfluidic devices with multilayer walls for enzyme immobilization. (b) and (c) Examination of LRSC-HRP enzyme immobilization on microchannel walls using laser scanning confocal microscopy. (d) Immobilizing HRP on the multilayer enabled the use of the multilayer fluidic system as an enzymatic reactor, where OPD was oxidized by H2O2 to form brown DAP. It is worth noting that HRP acted as a catalyst in this reactor, influencing the kinetics and efficiency of the reaction. (e) and (f) Increasing the number of deposited layers in the microchannels improved the catalytic performance of HRP significantly. (a)–(f) Reproduced with permission.21 Copyright 2022, Wiley-VCH.

Upon preparing CNC-based liquid channels, a continuous flow system was designed by connecting each side of a channel to a syringe pump. In such a system, fluid can be injected from one side of the channel and collected from the other side. Then, chitosan (CH) and hyaluronic acid (HA) layers were alternately deposited on the inner side of the channel to fabricate a microfluidic device with a multilayer wall. The deposition process involved first washing the channel with water (pH ∼ 3) and then injecting a suspension of CH into the fabricated channel. In this case, the electrostatic interaction of the cationic CH and the anionic CNC layer was reported to deposit a layer of CH on top of the CNC interfacially driven layer. After the deposition of the CH layer, the microfluidic system was rinsed with water (pH ∼ 7), and then a suspension of HA was injected into the channels to form an HA layer on top of the CH layer. This deposition process was repeated several times to generate microfluidic channels with walls containing alternating CH/HA layers.

The developed CH/HA multilayer provides an excellent platform for immobilizing the HRP enzyme. Prior to enzyme injection, the microchannels were washed with 2-(N-morpholino)ethanesulfonic acid (MES) buffer solution. Subsequently, a MES buffer solution containing HRP is injected into microchannels to optimize enzyme adsorption. Following this, the microchannel is washed with pure MES buffer solution to eliminate unattached HRP. To assess the immobilized HRP on the CH/HA multilayer, lissamine rhodamine B sulfonyl chloride-labeled HRP (LRSC-HRP) was synthesized, and its attachment to the inner wall of microchannels was examined by laser scanning confocal microscopy, as shown in Fig. 28b and c.

Upon immobilizing HRP on the multilayer, the multilayer fluidic system was used as an enzymatic reactor, where o-phenylenediamine (OPD) was oxidized by H₂O₂ to form brown 2,3-diaminophenazine (DAP). In this reactor, HRP acted as a catalyst, influencing the kinetics and efficiency of the reaction (Fig. 28d). Interestingly, increasing the number of deposited layers in the microchannels significantly enhanced the catalytic performance of HRP (Fig. 28e and f). Overall, this strategy for designing all-liquid microfluidic devices is simple and widely applicable, addressing key impediments faced by all-liquid microreactors and opening new applications in the field of chemical biotechnology.

The aforementioned all-liquid microfluidic devices rely on a nonpolar bath, e.g., silicone oil, which limits their application in certain scenarios. To tackle this challenge, Xie et al.¹⁵⁴ developed all-aqueous microfluidic systems based on a stabilized water/water interface. In their study, an aqueous solution of dextran with poly(sodium 4-styrenesulfonate) (PSS) was printed in an aqueous bath of PEG containing poly(diallyldimethylammonium chloride) (PDADMAC). It is important to note that the dextran solution and PEG phase are immiscible and do not diffuse into each other. In this scenario, the interfacial complexation of oppositely charged PSS and PDADMAC was reported to generate a solid-like skin that stabilizes the aqueous suspension of dextran and suppresses Plateau–Rayleigh instabilities.

The developed all-aqueous printed channels were used as microreactors for chemical synthesis and separation. For instance, continuous compartmentalized reactions can be performed in these aqueous devices. In this scenario, a suspension of dextran containing urea, HRP enzyme, and H₂O₂ was directed through the printed microchannels, while urease and 2,20-azino-bis(3-ethylbenzothiazoline-6-sulphonic acid) diammonium salt (ABTS) were added to the external PEG bath (Fig. S12a, ESI†). In the first reaction, urease catalyzes the hydrolysis of urea into carbon dioxide and ammonia. To validate this reaction, the pH values of both aqueous phases were monitored. After continuous urea hydrolysis for 1 hour, the pH in the PEG phase increased from 4 to 8.5, while the pH in the dextran phase showed a smaller increase due to the diffusion of ammonia (Fig. S12b and c, ESI†). In a second reaction, ABTS was oxidized into ABTS⁺ using H₂O₂ as the oxidizing agent and HRP as the catalyst. Here, the 415 nm peak of ABTS⁺ in the ultraviolet-visible (UV-Vis) spectrum was only detected in the dextran solution, not in the PEG phase, confirming the compartmentalization of reactions (Fig. S12d, ESI†). Overall, the ability of these microfluidic devices to simultaneously perform compartmentalized and separate reactions represents a significant innovation not yet seen in other fluidic devices.

In addition to microfluidic devices, structured liquid droplets and bead systems also offer novel opportunities for designing all-liquid microreactors.¹⁵⁵ Yang et al.²⁰ explored this concept by developing multi-responsive structured liquid droplets that can be directed to specific locations with external stimuli, assemble on demand, and perform chemical reactions. These structured droplets were formed through electrostatic interactions between Au nanoparticles decorated with 11-mercaptoundecanoic acid (MUA) featuring a –COOH termination and oleic acid (OA) functionalized Fe₃O₄ or PbS nanoparticles. As explained before, the electrostatically driven assembly and jamming of these nanoparticles at the interface generated a solid-like skin that stabilized these structures in non-equilibrium shapes (Fig. 29a).

	Fig. 29 The developed structured droplets are stabilized through the electrostatic interaction of different nanoparticles with complementary functional groups at the liquid–liquid interface. (b) The developed structured droplets were light-responsive. Under laser irradiation, the droplets moved towards the beam and reversibly assembled into closely packed structures. (c) Two structured droplets containing high-viscosity solutions of Congo red or HCl were moved toward each other by an external magnetic field and then welded together by heating the contact point using a laser source. (d) Two droplets carrying CoCl2 and 2-methylimidazole were brought into contact by laser light and welded by an electric field to form ZIF-67. (a)–(d) Reproduced with permission.20 Copyright 2018, Nature Publishing Group.

In their experiments, the droplets were prepared using a water/ethylene glycol mixture, while the external bath consisted of a dispersion of Fe₃O₄ or PbS nanoparticles in 1,2-dichlorobenzene (DCB) or a mixture of toluene and DCB with a density of 1.05 g mL⁻¹. In these solvents, aqueous droplets were buoyant and localized slightly below the surface rather than rising and spreading to it. Furthermore, these droplets responded to an external magnetic field and formed ordered assemblies in high-field regions. However, due to the presence of a solid-like skin encasing the droplets, they remained distinct from each other without mixing or exchanging contents for at least several days. This characteristic ensures stability and containment in applications requiring controlled interactions.

More interestingly, these structured droplets were also light-responsive and could be moved by applying an external laser source, specifically using a 660 nm diode laser. Under laser irradiation, the fabricated droplets moved towards the beam and formed densely packed configurations, as illustrated in Fig. 29b. Switching off the laser led to droplet disassembly within approximately 20 seconds. This cycle of assembly and disassembly could be repeated numerous times without observable changes in droplet behavior.

On the other hand, external stimuli such as electric fields and laser heating have been used to weld these droplets together, creating nonspherical shapes and allowing the contents of different droplets to mix. These techniques, combined with external magnetic and optical fields, facilitated the development of reconfigurable and programmable all-liquid microreactors. In these systems, various reactants can be loaded into separate droplets and mixed on demand to carry out specific reactions, enhancing the versatility and functionality of microfluidic platforms. Two experiments were performed to showcase this potential and highlight the importance of rapid mixing in all-liquid microreactors. In the first scenario, two droplets containing high-viscosity solutions of Congo red or HCl were moved toward each other and fixed using an external magnet. Upon irradiation with an external laser source and heating of the droplets, the contents of these droplets rapidly mixed together over a few minutes, as shown in Fig. 29c. In the other experiment, two droplets carrying CoCl₂ and 2-methylimidazole were brought into contact by laser light and fused by an electric field, i.e., short pulses of an electric field produced by a Zerostat gun. Upon diffusion and mixing of the droplets' contents, zeolitic imidazole framework 67 (ZIF-67) was formed, depicted by the blue color in Fig. 29d.

In another intriguing study, Shi and colleagues demonstrated a multi-responsive and reconfigurable droplet network with significant potential for developing all-liquid microreactors.¹⁷ In their system, an aqueous solution containing methyl viologen functionalized sodium carboxymethylcellulose (CMC-MV²⁺) and CB[8] interacted with an oil phase consisting of azobenzene-terminated poly-L-polylactic acid (Azo-PLLA) and toluene. At the oil/water interface, the host CB[8] could bond with other guest molecules, MV²⁺ and Azo, stabilizing an aqueous droplet in the nonpolar phase. Interestingly, the CB[8] surfactants generated at the liquid–liquid interface have cavities inside their structures, allowing the exchange of materials between two different droplets that come into contact with each other (Fig. 30a). As a result, compared to previous all-liquid microreactor systems, the droplets stabilized by CB[8] surfactants are more versatile. A simple manual compression of these droplets towards each other can initiate a reaction without the need for external stimuli, such as an electric field or heating with a laser, as described in the previous study.

	Fig. 30 (a) The all-liquid microdroplets form through the interfacial complexation of CB[8], CMC-MV2+, and Azo-PLLA at the oil/water interface. Upon a simple compression of these droplets together, their cargo can easily mix together and initiate a reaction. (b) and (c) Formation of Fe(SCN)3 with a deep red color upon contacting two droplets containing FeCl3 and KSCN, respectively. (d) and (e) Synthesis of ZIF67 in all-liquid microreactors. (f) and (g) Cascade reactions were performed by contacting three droplets with different cargos, i.e., droplet A containing glucose, droplet B containing HRP enzyme and GOD, and droplet C containing OPDA. (a)–(g) Reproduced with permission.17 Copyright 2024, Nature Publishing Group.

The adaptability of this droplet network as a new generation of all-liquid microreactors has been demonstrated through various reactions, where droplets carrying different reactants are simply pushed toward each other to initiate the reaction. In the first example, two droplets containing ferric chloride (FeCl₃) and potassium thiocyanate (KSCN) were brought together, resulting in the formation of ferric thiocyanate (Fe(SCN)₃) with a deep red color (Fig. 30b and c). In another example, this droplet system was used to synthesize ZIF-67 in-situ by connecting two droplets containing CoCl₂ and 2-methylimidazole (2-MI), respectively (Fig. 30d and e). Furthermore, cascade reactions were also performed using this all-liquid microreactor system. For this purpose, a droplet containing glucose was contacted by another droplet containing HRP enzyme and glucose oxidase (GOD). When glucose and GOD react together, H₂O₂ forms in the system. The presence of H₂O₂ was visualized by adding another droplet containing o-phenylenediamine (OPDA) to the system. Due to the presence of H₂O₂, OPDA is oxidized to 2,3-diaminophenazine (DAP), and a yellow-green substance is produced in the droplets (as shown in Fig. 30f and g). Overall, such an all-liquid and reconfigurable microreactor system, which requires only a simple compression of droplets together to initiate a reaction, may find several applications in fields such as chemical synthesis, biotechnology, and drug discovery, offering a versatile platform for efficient and controlled reactions in a compact and flexible setup.

When an electromagnetic wave encounters a conductive porous structure, a part of the wave is reflected due to the impedance mismatch between the electrically conductive domain and free space. This reflected power is quantified by the reflection index (R), which ranges from 0 to 1. It is important to note that the reflectance index primarily depends on the electrical conductivity of the material, and reducing this index is a key focus in the development of effective EMI shielding.^169,170

In addition to reflection, a portion of the incident wave is absorbed by the porous constructs through interactions with electrically conductive materials and active electric/magnetic dipoles. This absorbed portion is measured by the absorption index (A), and the goal is to maximize this factor to create an absorption-dominant EMI shield, ideally achieving an “A” value of at least 0.5. Such high absorption levels are particularly achievable in porous materials due to a mechanism often referred to as “multiple internal reflections” or “internal scattering.” This process involves multiple back-and-forth reflections within the shield, facilitated by interfaces with mismatching impedance in the porous material, which significantly enhances its absorption capabilities. Furthermore, a portion of the electromagnetic wave may pass through the structure, resulting in a transmitted wave. This transmission is quantified by the transmission index (T), and minimizing this index is a critical objective to enhance the shielding effectiveness of the structure.^169,170

It is important to recognize that current studies on the EMI shielding capabilities of porous materials evaluate performance based on the R, A, and T indexes to identify the primary shielding mechanisms. Additionally, these studies often utilize another parameter known as total shielding effectiveness (SE_T), measured in decibels (dB). The SE_T is crucial as it quantifies the shield's ability to protect against electromagnetic waves and is defined as the logarithmic ratio of the incident power to the transmitted power of the electromagnetic waves. An aerogel EMI shield is considered to exhibit exceptional EMI shielding characteristics if it has a high shielding effectiveness (SE_T > 30 or higher) and an absorption-dominant shielding mechanism (A > 0.5 or higher). The total density and thickness of the developed shields also play critical roles in evaluating shield performance. Reducing these factors can decrease the total cost of EMI shields and increase their adaptability to various shielding scenarios, enhancing their practicality.

With these parameters in mind, we can see why liquid-templated aerogels are an appealing option for designing the next generation of EMI shielding materials. These aerogels can be engineered to achieve high absorption indices and excellent shielding effectiveness while maintaining low density and customizable thickness, making them versatile and cost-effective solutions for diverse EMI shielding applications.^169,171,172

Shi et al.⁶⁹ were among the first researchers to utilize liquid-templated aerogels of MXenes for EMI shielding. In their study, they reported the co-assembly of Ti₃C₂T_x and POSS-NH₂ at the toluene/water interface, which stabilized oil-in-water Pickering emulsions. These emulsion templates were then concentrated through simple centrifugation and used as a template to fabricate lightweight, hydrophobic, isotropic MXene aerogels with outstanding mechanical properties (refer to Fig. 32a). These aerogels served as a framework for developing conductive epoxy nanocomposites; the MXene aerogels were impregnated with epoxy precursors and allowed to cure. The resulting polymer nanocomposites, with a very low MXene filler content (0.40 vol% of Ti₃C₂T_x), exhibited SE_T of 34.5 dB, significantly surpassing conventional polymer nanocomposites of MXene. However, the EMI shielding mechanism of these aerogels primarily relied on reflecting electromagnetic waves due to their relatively high electrical conductivity of 195.3 S m⁻¹, leading to a substantial impedance mismatch at the interface of these structures.

	Fig. 32 (a) Schematic showing the fabrication of MXene-based structured Pickering emulsions that were used for developing aerogel-based EMI shields through the liquid templating approach. (a) Reproduced with permission.69 Copyright 2019, Wiley-VCH. (b) Underlying mechanisms that define the concept of “electromagnetic traps.” (I) showcases the developed “electromagnetic traps.” (II) shows that when electromagnetic waves pass through the worm-like aerogels and reach the conductive tape, they are reflected back toward the aerogels. (III) showcases the concept of internal scattering within and between filaments in worm-like aerogels. (IV) Losses due to interfacial and dipole polarization and (V) conductive losses caused by electron migration and hopping. In (IV) and (V), the yellow rods signify the CNF, while the blue and red dots indicate positive and negative charges, respectively. (b) Reproduced with permission.56 Copyright 2024, The Royal Society of Chemistry.

More recently, we utilized worm-like aerogels of Ti₃C₂T_x/GO, reduced GO, reduced magnetic GO/GO, and carbonized CNF/GO to develop a new class of EMI shields.^3,19,40,56 Our strategy focused on two key factors: (1) keeping the electrical conductivity of these aerogels relatively low to allow incident electromagnetic waves to enter the structures with minimal reflectance. (2) Leveraging the dual micro-to-macro-scale porosities of worm-like aerogels, unique to these structures, to enhance internal scattering and absorption capabilities. This approach enabled us to achieve high SE_T values (>30 dB) and absorption-dominant characteristics (A > 0.5) for some of these samples. For instance, using worm-like templates of Ti₃C₂T_x/GO, ultra-lightweight aerogels with a density of 3–7 mg cm⁻³ were obtained. When these aerogels served as the backbone for epoxy nanocomposites, unique polymer nanocomposites with extremely low filler content (0.3–0.7 wt%) and exceptional EMI shielding characteristics (SE_T of 31.7–51.4 dB and A ≈ 52–54% were achieved, indicating dominant absorption behavior) for the samples.¹⁹ This highlights the potential of liquid-templated aerogels for large-scale and cost-effective fabrication of modern EMI shields for military and commercial purposes.

In another study, worm-like aerogels of GO/CNF were used to develop “electromagnetic traps” with enhanced absorption dominant EMI shielding characteristics.⁵⁶ In this system, conductive carbonized worm-like aerogels of GO/CNF were placed at the top of highly conductive metal tapes to create EMI shields with remarkably high SE_T values of 53–89 dB and absorbance exceeding 0.7. Several considerations were taken into account when designing electromagnetic traps (refer to Fig. 32b). Firstly, the conductivity of the worm-like aerogels was maintained in the range of 10–40 S m⁻¹ to ensure electromagnetic waves could penetrate the structures with minimal reflectance. Upon entering the shield, the waves' energy was absorbed due to interactions with electric/magnetic dipoles and materials with finite electrical conductivity. The presence of micro-to-macro-scale porosities further enhanced the dissipation of electromagnetic waves through internal scattering mechanisms, as previously explained. Moreover, as the wave path traversed through the aerogel and reached the conductive tape, it reflected back to the aerogel, dissipating and absorbing its remaining energy. This study clearly demonstrated the potential of worm-like aerogels with dual-scale porosities in designing absorption-dominant EMI shields with exceptional performance.

Janus aerogels of Ti₃C₂T_x/GO:mGO/GO, with non-interfering magnetic/conductive domains, are also utilized for developing absorption-dominant aerogel traps, exhibiting impressive performance with a SE_T of approximately 51 dB, SE_R of around 0.4 dB, and A = 0.91 (one of the highest reported absorbance values in the field of EMI shielding).³ The system design and underlying mechanism are akin to what was reported for “electromagnetic traps” in a previous study, where the designed Janus aerogels were positioned atop a highly conductive copper tape. However, the inclusion of non-interfering magnetic/conductive domains in these aerogels further reduces reflectance, as the presence of magnetic stripes adjacent to conductive strips at the interface significantly further reduces the impedance mismatch. Moreover, the interface between magnetic and conductive domains within the aerogels contributes to internal scattering and enhances the absorption capability of these aerogels. These advantages, combined with the absence of thermal and chemical reduction requirements, position these Janus aerogels at the forefront of the EMI shielding field.

It is important to acknowledge that while the developed Janus aerogels have a detection range limited to approximately 60% of compression strain, they may not be the most sensitive option available in the field of strain and pressure sensing. However, these aerogels offer significant advantages, notably their straightforward fabrication process. Unlike other sensors, the Janus aerogels do not require chemical or thermal reduction, nor do they need pre-treatment techniques such as freeze-casting to develop their structures. Additionally, the Janus platform is particularly appealing for creating functional, flexible aerogels. This method allows for customization where one side of the aerogel is tailored for electrical conductivity, while the other side maintains mechanical flexibility. This dual functionality makes Janus aerogels a compelling choice for applications requiring a balance of electrical performance and physical adaptability, enhancing their utility in diverse fields, including wearable technology and smart materials.³

In this context, a study by Shi et al.⁶⁹ is noteworthy, where the interfacial assembly of Ti₃C₂T_x and POSS-NH₂ at the toluene/water interface was utilized for fabricating structured Pickering emulsions, which were then transformed into hydrophobic aerogels with high porosity (Fig. 33a and b). In this intriguing study, the hydrophobic characteristics of the aerogels stem from the hydrophobic nature of POSS molecules that uniformly distribute throughout the aerogel matrix without requiring additional surface modification or conventional chemical or thermal treatments typically used in designing carbon-based oil absorbers.

	Fig. 33 The developed MXene-based aerogel showcased (a) ultra-low density and (b) unique hydrophobic characteristics. (c) Photographs showcasing the oil absorption capability of developed MXene-based aerogels (the yellow phase is a mixture of n-hexene and azobenzene). (a)–(c) Reproduced with permission.69 Copyright 2019, Wiley-VCH. (d) The capability of the carbonized worm-like aerogels of GO to absorb various non-polar solvents. (e) Upon compression, the carbonized worm-like aerogels readily release the absorbed oil. (f) The prepared worm-like aerogels can effortlessly float on the water surface and swiftly separate oil from water in around 5 seconds. (d)–(f) Reproduced with permission.40 Copyright 2019, Wiley-VCH.

The surface wetting behavior of liquid-templated MXene-based aerogels was investigated by analyzing the water contact angle. While the pure MXene film (evaluated as a reference) displayed hydrophilicity with a contact angle of 42°, attributed to the presence of oxygen and fluorine terminal groups on the MXene surface, the liquid-templated MXene aerogels exhibited a substantial increase in the water contact angle to 123°, indicating the hydrophobic nature of this structure. This change can be attributed to the presence of hydrophobic POSS-NH₂ on the MXene surface. Additionally, these aerogels demonstrated exceptional capability for absorbing organic solvents such as toluene, n-hexane, and silicone oil. For instance, toluene absorption of up to 9000 wt% was reported for these aerogels (see Fig. 33c).

An intriguing example can be found in worm-like aerogels of reduced GO, which present an optimal solution for oil absorption in aqueous media.⁴⁰ In this study, worm-like aerogels of GO underwent thermal reduction at 800 °C to carbonize the porous structure and render the aerogels hydrophobic. The abundant microscale porosities within the filamentous aerogels, coupled with macroscale porosities between the filaments, contributed to the remarkable oil absorption capacity that surpassed all previous studies on aerogel-based oil absorbers (Fig. 33d). For instance, the resulting aerogels demonstrated an absorption capability of approximately 46000 wt% for chloroform, 34000 wt% for diesel oil, 26000 wt% for pump oil, and 20000 wt% for motor oil. Furthermore, these worm-like aerogels can release organic solvents after compression (Fig. 33e). Moreover, they can effortlessly float on the surface of water, achieving rapid separation of oil from water in less than 5 seconds (Fig. 33f). These characteristics render these aerogels particularly attractive for addressing oil spill cleanups.

Within pendent drop tensiometry, the configuration of the liquid–liquid interface is dictated by gravitational and surface forces, as well as the boundaries that confine the droplet. Taking these factors into account, a pendent drop at equilibrium adheres to the Young–Laplace equation, establishing a connection between the Laplace pressure across the interface, the curvature of the interface, and the interfacial tension, as shown in the following equation.^14,195

	(1)

The adsorption of nanoscale assemblies, including nanoparticles and supramolecular clusters at the liquid–liquid interface, is known to reduce the measured interfacial tension between two immiscible liquids dramatically.^14,24,62 The primary contribution to this reduction comes from those nanoscale assemblies positioning themselves between the two immiscible liquids and screening them in a way that the required energy to form the interfaces reduces significantly. However, this phenomenon cannot be straightforwardly represented through the Young–Laplace equation, as this equation assumes the existence of a single isotropic surface tension, which is not valid for the case of nano-to-macro scale particle assemblies.^197,198 In general, eqn (1) loses its accuracy when dealing with solid interfaces or other surfaces with anisotropic stress. This is because we need to consider not just surface energy but also the mechanical characteristics of the material arranged at the interface. Hence, for pendent drop tensiometry, it becomes crucial to integrate a pressure sensor on the droplet side. This allows us to account for the intricate properties of interfaces that involve nanoscale assemblies.¹⁴

6.2.1.1. Determining the binding energy of nanoparticles at the liquid–liquid interface through interfacial tension measurements. Theoretically, in the absence of particle–particle interactions, the decrease in interfacial energy between two immiscible liquids upon introducing particles into one of the liquids is correlated with the binding energy per area for each particle at the liquid–liquid interface, as depicted in the following equation.^14,199

	(2)

Here, ΔE represents the energy required for a single particle to bind to the interface, γ₀ and γ denote the interfacial tension between two immiscible liquids before and after the introduction of particles into one of the liquids (with γ − γ₀ representing the reduction in interfacial tension), and N_s stands for the number of particles adsorbed onto the interface.

In the context of spherical nanoparticles, Dinsmore and colleagues rearranged eqn (2) to compute ΔE as follows:¹⁹⁹

	(3)

where is the area fraction of particles at the interface and R is the radius of the nanoparticles. To calculate ΔE accurately, measurements of γ₀ and γ at a known value of η are essential, and pendent drop tensiometry is employed for this purpose. The referenced study presumed that nanoparticles are densely packed at the interface with η = 0.91. In practice, the value of η can also be determined using in-situ techniques such as SAXS or GISAXS. These methods allow for a relatively precise determination of the average center-to-center distance between neighboring nanoparticles.¹⁴

The energy needed to detach a spherical particle from the interface to the adjacent bulk fluid, i.e., the binding energy, can also be expressed as a function of the particle's wetting characteristics. This perspective assumes a simplified scenario where no energy barriers are hindering the particles from reaching the fluid–fluid interface.^14,200

ΔE = −γ0πR2(1 − |cosθ|)2	(4)

Here, the angle θ, the angle θ, representing the contact angle at the interface, is measured relative to the aqueous phase.

Capitalizing on this equation, Zhang et al.¹⁹⁸ combined SEM analysis and pendant drop tensiometry to establish an estimated correlation between the area fraction of particles at the interface η and the measured fluid–liquid surface tension, as shown in the following equation.

γ = γ0[1 − η(1 − |cosθ|)2]	(5)

Arguably, the most crucial finding in the aforementioned study is that eqn (5) remains valid even at near-complete packing levels at the interface. This suggests that the particle–particle interaction has an insignificant impact on the reduction of interfacial tension, assuming the particles are highly attracted to the interface due to their wetting properties and face no barriers preventing their adsorption onto the interface.¹⁹⁸

The binding energy of an individual particle to the interface can also be determined through dynamic surface tension measurements at a short timescale – specifically, by utilizing the initial 10 seconds of the recorded surface tension data, as expressed in the following equation:²⁰¹

	(6)

In this context, γ represents the interfacial tension at time t, N_A is the Avogadro constant, C₀ is the initial bulk concentration of particles, and D is the diffusion coefficient dictating the adsorption rate at the oil/water interface. Notably, in this scenario, D is equivalent to the diffusion coefficient derived from the Stokes–Einstein relation.

Beyond assessing binding energy, dynamic surface tension measurements offer a means to evaluate the adsorption barrier of particles in comparison to a scenario with an unrestricted bare interface (t → 0). This is achieved by analyzing the long timescale data (t → ∞) from interfacial tension measurements. This highlights the effectiveness of dynamic surface tension measurement as a powerful tool for concurrently evaluating two essential parameters: the binding energy and adsorption barrier of particles.^14,201

The equations mentioned above solely consider the wetting properties of nanoparticles at the liquid–liquid interface, overlooking particle–particle interaction at the interface. Hence, a more comprehensive theoretical framework has been proposed to assess both these parameters in reducing interfacial tension (Fig. 37a and b), as outlined in the following equation:^202,203

γ = γ0 − πSA(t) − πNP(t)	(7)

where γ denotes the interfacial tension at time t, π_SA(t) stands for the surface pressure arising from particle–particle interaction, and π_NP(t) represents the surface pressure resulting from the wetting of particles at the interface. These parameters are computed using the following equations, providing a realistic interpretation of interfacial tension reduction caused by the particles’ interactions at the liquid–liquid interface.²⁰²

πNP = |ΔE| × Γ	(8)

πSA = −KTΓc→∞[ln(1 − ϑ) − 0.5kϑ2]	(9)

where Γ represents the surface excess of nanoparticles at the interface, given by , signifies the fractional coverage of the surface-active materials at the interface, and Γ_c→∞ denotes the maximum coverage at the interface. Additionally, the parameter K characterizes the interaction potential between particles in units of kT, typically on the order of ±1 for systems involving nanoparticles. A positive value for K indicates repulsive interactions, leading to an intuitive rise in surface pressure, whereas negative K implies attractive interactions and a reduction in surface pressure. The special case of K = 0 corresponds to noninteracting particles.^14,202,203

	Fig. 37 Macroscopic evaluation of nanoscale assemblies at liquid–liquid interfaces. (a) Segregating nanoparticles into the liquid–liquid interface and (b) nanoparticle–nanoparticle interaction at the interface as parameters that affect interfacial tension. (a) and (b) Reproduced with permission.14 Copyright 2019, Wiley-VCH. During droplet contraction, its volume diminishes, resulting in a proportional reduction in interfacial area and the imposition of compressive forces on the interface. These compressive forces may cause (c) the desorption of active materials, (d) the formation of surface wrinkles at the liquid–liquid interface that subsequently undergoes relaxation and disappearance, and (e) the creation of permanent wrinkles that signify the irreversible bonding of active materials at the interface. (c) Reproduced with permission.204 Copyright 2012, American Chemical Society. (d) Reproduced with permission.205 Copyright 2014, American Chemical Society. (c)–(e) Reproduced with permission.14 Copyright 2019, Wiley-VCH. (e) and (f) Ferromagnetic micro-buttons with amphiphilic surface chemistry were designed for a magnetic button rheometer. (g) A fluorescence microscopy image was captured from the shear rheology apparatus that features magnetic buttons. (e)–(g) Reproduced with permission.206 Copyright 2011, Nature Publishing Group. (h) The elastic characteristics of the interface can be evaluated based on the pendant drop's shape and wrinkling patterns. (h) Reproduced with permission.207 Copyright 2013, American Chemical Society.

The interaction of nanoscale assemblies at a liquid–liquid interface under the influence of compressive forces is intricately linked to the equilibrium between the compression rate, referred to as the strain rate , and the characteristic timescale of particle desorption represented by τ_des. If ≫ τ_des⁻¹, no wrinkles appear at the interface during contraction, and the shape of the pendant drop conforms to the Young–Laplace equation (Fig. 37c).²⁰⁴ In this scenario, the adsorption of surface-active species into the interface is reversible, and compression leads to the desorption of species, ultimately causing the relaxation of surface tension to an equilibrium value independent of area. On the other hand, when dealing with large nanoparticles having a radius of 5–10 nm, and in some cases with polymers or proteins, we may encounter a scenario where ∼ τ_des⁻¹. This is characterized by the formation of wrinkles at the interface, which, over time, undergo disappearance due to relaxation (Fig. 37d).²⁰⁵

Ultimately, when surface-active species irreversibly bond to the interface ( ≪ τ_des⁻¹), interfacial wrinkles persist indefinitely (Fig. 37e).¹⁴ In this context, the particles typically maintain a monolayer when compressed. However, if compressed to the point of solidification, they do not form a multilayer. Instead, the particle assembly buckles out of the plane, creating permanent wrinkles, which reveal a strong interfacial membrane at the liquid–liquid boundary. Recently, we applied such a scenario to elucidate the irreversible bonding of various nanomaterials, including pure or hybrid forms of GO, magnetic GO, MXene, and CNF, at a liquid–liquid interface.^3,19,40,208 This involved the formation of NPSs through the interaction of these nanomaterials at the interface with an interfacially active ligand, i.e., POSS, effectively pinning the nanomaterials to the interface. Upon contraction of the pendant drop in the aforementioned systems, observable wrinkles emerged at the liquid–liquid interface. These wrinkles persisted over time for both large and small droplets, demonstrating the irreversible adsorption of nanoparticles onto the interface that improves the mechanical robustness of the interface.

The additional surface stress in eqn (10) can be explained using the Boussinesq–Scriven model, particularly when dealing with a liquid-like and viscous interface, denoted as τ_v.^{209,211–213}

The additional surface stress can also be calculated for solid-like elastic interfaces, i.e., denoted as τ_e based on the following equation:^209,214

	(12)

In practical applications, the models outlined in eqn (11) and (12) require material properties (η_s, κ_s, κ_s, K_s, G_s) or more intricate material functions. These parameters are measured under precisely defined kinematic conditions and serve as inputs for the models. Thorough explanations of these models, including their physical foundations and advanced methods for characterizing interfacial rheology, can be found in the comprehensive review by Jaensson and Vermant.²⁰⁹