The researchers built their study around a well-characterized rabbit model of aqueous-deficient dry eye. They injected concanavalin A, an immune-stimulating plant protein, directly into the lacrimal glands of New Zealand white rabbits under ultrasound guidance, damaging the tear-producing glands and triggering the hallmark features of the disease. Within five days, tear break-up time collapsed by roughly 81 percent, Schirmer tear test values fell by about 51 percent, and tear osmolarity rose significantly, confirming that the animals had developed a genuine dry eye phenotype. Notably, the rabbits’ corneas also became markedly less sensitive to touch, requiring much stronger stimulation to provoke a protective blink.

Measuring that sensitivity required a device with a long pedigree. The team used a Cochet-Bonnet esthesiometer, an instrument descended from a design first described in 1894, which touches the central cornea with a fine nylon filament. By shortening the filament in five-millimeter steps, the examiner increases the force delivered until the animal blinks and retracts its eye; the filament length at which this occurs provides an index of corneal touch threshold. Dry eye rabbits responded only when the filament was shortened to about 3.8 centimeters, compared with 5.7 centimeters in healthy controls, mirroring the reduced corneal sensitivity documented in human patients with the disease.

Against this baseline, the effect of phosphosulindac was striking. A single 25-microliter eye drop of the compound, formulated at 0.2 percent in a Carbopol-based hydrogel, restored corneal sensitivity to normal levels within five minutes, the first time point tested. The normalization persisted for roughly 26 to 27 hours before gradually dissipating, an unusually stable response that raises the possibility of once-daily dosing. Intriguingly, a higher concentration of 0.6 percent produced a shorter effect lasting only about two and a half hours, hinting at a dose-response relationship that is far from linear.

The choice of formulation proved critical. Cyclodextrin-based solutions, in which the hydrophobic phosphosulindac molecule is shuttled inside the cup-shaped interior of cyclic glucose oligomers, and an oil-in-water emulsion also restored sensitivity in dry eyes, but their effects lasted under an hour. Polymer nanoparticles, by contrast, were ineffective. The team also uncovered a sharp pharmacological threshold: in solution, 0.05 and 0.1 percent phosphosulindac did nothing, while 0.2 percent was fully active, and an eightfold higher dose of 1.6 percent added no further benefit. In cyclodextrin formulations, activity peaked at pH 6.2, consistent with pH-dependent shifts in the stability of the drug-cyclodextrin inclusion complex.

What distinguishes phosphosulindac is the company it failed to keep. None of six structurally related compounds, including sulindac itself, aspirin derivatives and an ibuprofen derivative, reproduced the effect. Neither did cyclosporine and lifitegrast, the two pillar immunomodulatory drugs approved for dry eye, nor a commercial artificial tear. Ketorolac and bromfenac, two non-steroidal anti-inflammatory eye drops, actually further depressed corneal sensitivity in the dry eye animals, as did the anesthetics proparacaine and benoxinate, which abolished responses almost immediately. Phosphosulindac, uniquely among everything tested, pushed sensitivity in the opposite direction.

In healthy eyes, the story inverted. Applied to normal rabbit corneas, phosphosulindac suppressed sensitivity, producing a peak reduction of nearly 56 percent at 50 minutes that resolved by about 140 minutes, a slower and longer-lasting effect than lidocaine, which numbed more powerfully but faded within 30 minutes. The authors speculate that in a normal cornea, densely populated with sensory endings, the drug may depolarize or overwhelm mechanosensory and polymodal nociceptors beyond the threshold needed to propagate a blink reflex, whereas in a dry eye, where nerve endings have been lost, the same action sensitizes the surviving receptors enough to restore measurable responses.

Several features of the data argue that the effect is direct and non-anesthetic. It appeared within minutes, was fully reversible within hours, and outlasted the measurable residence of the drug in the cornea reported in earlier pharmacokinetic work, pointing to downstream changes in nerve function rather than simple tissue numbing. Crucially, phosphosulindac is not itself a classical NSAID and does not suppress prostaglandin production in the cornea; unlike ketorolac and bromfenac, which have been linked to the devastating complication of corneal melt, it inhibits matrix metalloproteinases instead. The team also confirmed the findings in pigmented Dutch-belted rabbits, ruling out melanin sequestration as a confounder.

The therapeutic implications are twofold. For dry eye disease, where nearly all symptoms are mediated through the nerves of the ocular surface, a drug that both treats the underlying inflammation, as the team’s earlier rabbit work demonstrated by normalizing tear break-up time, osmolarity and lactoferrin, and rapidly restores sensory function could address both signs and symptoms in a way current medications, whose benefits often take weeks to appear, do not. For normal eyes, the numbing effect suggests a potential role in managing acute ocular pain, for example after cataract surgery, though the authors stress that any anesthetic-like use would demand careful safety evaluation.

Important caveats remain. The work was conducted in rabbits with an induced, aqueous-deficient form of the disease, whereas most human dry eye is evaporative in origin, and whether the striking dichotomy between normal and dry corneas translates to patients is unknown. Confirming the proposed mechanism, a direct action on mechanosensory and polymodal corneal nociceptors, will require electrophysiological, immunohistochemical and molecular studies. Still, with dry eye affecting a substantial share of the adult population and its neuropathic pain increasingly recognized as under-treated, a small molecule that can flip corneal sensitivity back on within minutes, safely and reversibly, is a result that both neuroscientists and ophthalmologists will be watching closely.

Subject of Research: Restoration of corneal sensitivity in dry eye disease by topical phosphosulindac in a rabbit model

Article Title: Phosphosulindac (OXT-328) restores the suppressed corneal sensitivity in rabbits with dry eye disease: therapeutic implications

Article References: Huang, W., Wen, Z., Tourmouzis, K., Huang, L., Goldstein, S. M., Natke, E., Honkanen, R., & Rigas, B. (2026). Phosphosulindac (OXT-328) restores the suppressed corneal sensitivity in rabbits with dry eye disease: therapeutic implications. Frontiers in Drug Delivery, 6, Article 1778440. https://doi.org/10.3389/fddev.2026.1778440

Image Credits: AI Generated

DOI: 10.3389/fddev.2026.1778440

Keywords: dry eye disease, phosphosulindac, OXT-328, corneal sensitivity, corneal nerves, rabbit model, ophthalmic drug delivery, cyclodextrin, hydrogel, NSAIDs, esthesiometry, ocular pain

Deep eutectic solvents, or DESs, are mixtures of two or more substances that, when combined in particular ratios, form a liquid with a melting point far lower than any of the individual components. They are often described as close cousins of ionic liquids, but with a crucial advantage: their ingredients can be cheap, abundant, biodegradable, and even food-grade. Because their polarity, hydrogen-bonding ability, and hydrophobicity can be tuned simply by swapping the constituent molecules, DESs have become attractive candidates for selectively dissolving specific natural products. The research team, led by Huixin You and Dongting Huang of Fujian University of Traditional Chinese Medicine together with colleagues including Wei Xu of Fujian Health College, exploited this tunability in a two-phase liquid–liquid extraction scheme designed specifically for ginkgo chemistry.

The first challenge the team faced was choosing the right solvent pair from an enormous design space. Rather than laboriously testing candidates one by one at the bench, the researchers turned to COSMOtherm, a computational tool that predicts how molecules interact with liquids by calculating activity coefficients from fundamental quantum chemistry. Screening eighty candidate deep eutectic solvents in silico, they identified promising hydrophilic systems for flavonoid extraction and hydrophobic ones for ginkgolic acid removal. This computational pre-selection dramatically narrowed the field before a single experiment was run, a strategy that saves time, reduces chemical waste, and illustrates how modern solvent design increasingly begins on the computer rather than in the fume hood.

Among the candidates validated experimentally, a hydrophilic DES composed of methyl-β-cyclodextrin and 1,4-butylene glycol in a mass ratio of 1:5 emerged as the standout performer for recovering ginkgo flavonoids. Methyl-β-cyclodextrin is a ring-shaped sugar molecule whose hollow, hydrophobic interior cavity can encapsulate flat, aromatic guests such as flavonoids, while its outer surface remains water-friendly. Pairing this molecular host with butylene glycol creates a solvent that can coax flavonoids out of plant tissue and hold them in solution. Under optimised conditions, the team achieved a maximum extraction yield of 10.06 milligrams of flavonoids per gram of material, a figure that demonstrates the practical viability of this cyclodextrin-based solvent for real plant extracts.

With the flavonoids recovered, the researchers then confronted the second half of the problem: ginkgolic acids. These long-chain phenolic acids are structurally related to compounds found in poison ivy and are responsible for allergic reactions in sensitive individuals, which is why pharmacopoeial standards limit their content in ginkgo products. To remove them, the team applied a hydrophobic deep eutectic solvent made from oleic acid and lauryl alcohol in a 5:1 ratio. Because this solvent forms its own phase and does not mix with the hydrophilic extraction medium, the two liquids could be brought into contact, allowing the ginkgolic acids to partition preferentially into the oily hydrophobic phase while the valuable flavonoids stayed behind.

The selectivity achieved in this second step is the study’s headline result. After optimising the liquid–liquid extraction conditions, the oleic acid–lauryl alcohol system removed 86.28 percent of five types of ginkgolic acids from the flavonoid-rich extract, while sacrificing only 6.46 percent of the flavonoids. That imbalance—large-scale removal of the undesirable compounds with minimal collateral loss of the desirable ones—is precisely what industrial downstream processing demands. It suggests that a carefully matched pair of deep eutectic solvents could in principle replace multi-step chromatographic or organic-solvent-based purification in the production of safer ginkgo ingredients.

Curious about why the cyclodextrin solvent worked so well, the researchers probed the underlying mechanism with two complementary techniques. First, they ran hybrid density functional theory calculations, a quantum-chemical method that allowed them to measure the interaction energy between the solvent components and the flavonoid molecules, revealing the strength and nature of the binding that anchors flavonoids in the DES. Second, scanning electron microscopy provided visual evidence of how plant material interacts with the solvent system at the microscopic scale. Together, these approaches support the idea that inclusion complexation between methyl-β-cyclodextrin’s cavity and the flavonoid scaffold underpins the enhanced extraction efficiency, turning an empirical observation into a mechanistically grounded explanation.

The team went further, deploying molecular dynamics simulations to track how thirteen different ginkgo compounds move through both the hydrophilic and hydrophobic solvent systems. By quantifying diffusion coefficients in the M-β-CD/BDO and oleic acid–lauryl alcohol media, the simulations offered a dynamic picture of transport behaviour, complementing the static energetic analysis from the DFT calculations. This combination of computational thermodynamics, quantum chemistry, microscopy, and molecular simulation represents a thorough multiscale interrogation of a separation process, and it points toward a future where extraction chemistries can be predicted and optimised computationally before scaling up.

The implications extend well beyond ginkgo. Herbal medicine, functional foods, and nutraceutical industries constantly wrestle with the dilemma of extracting trace beneficial compounds from complex plant matrices while excluding unwanted or toxic co-extractants. Conventional processes often consume large volumes of petrochemical solvents, generate problematic waste, and demand energy-intensive evaporation steps. Deep eutectic solvents built from substances as benign as cyclodextrins, glycols, fatty acids, and alcohols promise greener alternatives, and their ability to be tailored for specific tasks—as this study’s paired hydrophilic and hydrophobic systems demonstrate—makes them versatile tools for selective molecular separation across the natural products landscape.

The work, supported by the National Natural Science Foundation of China and the Fujian Provincial Department of Science and Technology, remains at the laboratory stage, and translating a bench-scale process into industrial manufacturing will require further work on solvent recovery, cost, and regulatory acceptance. Even so, the study offers a compelling template: screen computationally, validate selectively, explain mechanistically. If that recipe holds for other medicinal plants and other problematic contaminants, the era of sustainable, precision extraction may arrive not with dramatic new machinery, but with elegantly designed liquids that know exactly which molecules to keep and which to discard.

Subject of Research: Deep eutectic solvent-based liquid-liquid extraction for selective recovery of Ginkgo biloba flavonoids and removal of ginkgolic acids

Article Title: Selective recovery of Ginkgo biloba flavonoids and simultaneous removal of ginkgolic acids via deep eutectic solvent‑based liquid‑liquid extraction

Article References: You, H., Huang, D., Han, H., Shen, H., Liu, Y., Li, H., & Xu, W. (2026). Selective recovery of Ginkgo biloba flavonoids and simultaneous removal of ginkgolic acids via deep eutectic solvent‑based liquid‑liquid extraction. npj Science of Food. https://doi.org/10.1038/s41538-026-01176-w

Image Credits: AI Generated

DOI: 10.1038/s41538-026-01176-w

Keywords: deep eutectic solvents, Ginkgo biloba, flavonoids, ginkgolic acids, liquid-liquid extraction, COSMOtherm, cyclodextrin, green chemistry, density functional theory, molecular dynamics, food science, herbal medicine

The design philosophy starts with an observation about biology. Human skin is not a uniform material; its sensory structures are distributed asymmetrically through its thickness, which is part of why a fingertip can distinguish pressure from shear and bending from stretching with such precision. The research team replicated this asymmetry artificially using a gravity-driven self-assembly strategy. MXene nanosheets—atomically thin two-dimensional carbides prized for their metallic conductivity—were modified with tannic acid and then dispersed in a polyurethane matrix. As the mixture settled, gravity arranged the nanosheets into a gradient through the film’s thickness, producing what materials scientists call a Janus structure: two faces with different compositions and therefore different responses to mechanical stimuli.

The resulting composite achieved a combination of properties that flexible electronics researchers usually have to trade off against one another. The film could stretch to 114 percent of its original length before fracturing, while maintaining a tensile strength of 7.56 megapascals and an electrical conductivity of 9.2 × 10⁻³ siemens per centimeter. In practical terms, the material is tough and stretchy enough to survive repeated deformation on a moving joint, yet conductive enough that small mechanical distortions produce measurable changes in its electrical response. The asymmetric distribution of MXene through the thickness is what makes those responses selective: because the nanosheet density varies along the film’s cross-section, bending in one direction compresses a different portion of the conductive network than bending in the opposite direction, allowing a single sensor to discriminate between bending direction, applied pressure, and tensile strain.

Perhaps the most chemically interesting element of the design is the mechanism that lets the material heal itself. The researchers introduced a host-guest supramolecular interaction between cyclodextrin, a ring-shaped sugar molecule that acts as the host, and adamantane, a rigid cage-shaped hydrocarbon that acts as the guest. Cyclodextrin’s interior cavity is just the right size and chemical environment to capture adamantane, and the pair locks together reversibly in water-rich environments. By building these host-guest pairs into the polyurethane as physical cross-links, the team created a network in which the connections between polymer chains can break and reform dynamically rather than permanently. When the film is cut or damaged, bringing the surfaces back into contact allows the reversible cross-links to re-engage, stitching the material back together without heat, catalysts, or external agents.

This dynamic cross-linking network delivered a self-healing efficiency of 77.8 percent, meaning the material recovered more than three-quarters of its original mechanical and electrical performance after damage. The authors emphasize that this property is not merely a laboratory curiosity: it is what allows the sensor to maintain structural integrity and stable electrical response under the constant, dynamic deformation of real-world use. A wearable device attached to a wrist or finger is flexed thousands of times per day, and microscopic damage accumulates quickly in conventional elastomers. A material that continuously repairs its own conductive network can keep delivering reliable signals long after a static composite would have degraded.

Where the study moves from materials chemistry to applied intelligence is in its fusion with deep learning. A single flexible sensor produces a complicated, sometimes ambiguous electrical signal when attached to a moving body part—the same gesture performed at slightly different speeds or angles produces waveforms that differ in ways that would defeat simple threshold-based readout. The researchers trained machine learning algorithms on the sensor’s output streams and found that the system could reliably classify the signals into discrete categories. In one demonstration, the sensor was used as an input device for digital recognition: a user writing or gesturing digits produced characteristic signal signatures, and the trained algorithm identified which numeral had been traced with high adaptability across repeated trials.

The second, and more visually striking, demonstration involved table tennis. Return posture in table tennis is a fast, fine-grained motor behavior: the angle of the wrist, the orientation of the forearm, and the timing of the swing all determine whether the ball comes back as a loop, a chop, a drive, or a block. The team attached their flexible sensors to a player and recorded the electrical signals generated during return strokes, then used the integrated deep learning pipeline to perform dynamic pose analysis and ball return determination. The system could distinguish among return postures during live scoring conditions, verifying that the sensor-algorithm combination was feasible for capturing fine motion and recognizing complex, multi-joint movement behaviors—not just slow, deliberate gestures.

The significance of the work lies in how three separate design layers reinforce one another. The biomimetic Janus architecture provides multi-modal sensitivity, so one sensor element can distinguish several kinds of mechanical stimulus without additional hardware. The supramolecular host-guest cross-linking provides durability and self-repair, so the sensitivity survives prolonged use. And the machine learning layer provides interpretive power, converting noisy analog waveforms into discrete, actionable classifications. Each layer addresses a known failure mode of wearable sensors: single-mode devices that cannot disambiguate stimuli, brittle composites that fail under repeated strain, and raw signal outputs that require human experts to interpret. The authors describe the combination as a new theoretical basis and implementation path for high-performance flexible sensors built on biomimetic structural design, supramolecular dynamic regulation, and intelligent algorithm fusion.

The broader context is the rapid growth of flexible and wearable electronics for health monitoring, human-machine interfaces, and sports science. MXene-based composites have become a leading platform for such devices because of their exceptional conductivity and processability, but issues of oxidation, mechanical brittleness in dense films, and signal ambiguity under complex deformation have limited real-world deployment. Strategies like tannic acid modification, which can both stabilize MXene nanosheets and promote interfacial bonding with polymer matrices, and gravity-driven gradient assembly, which requires no expensive patterning equipment, point toward manufacturing routes that could scale. Meanwhile, the demonstration that a deep learning model can extract meaningful behavioral categories—down to the classification of an athlete’s stroke—from a minimal sensor array suggests a future in which lightweight wearable patches, rather than camera systems or instrumented gloves, provide motion analytics for training, rehabilitation, and gesture-based computing.

The study was supported by the Zhejiang Provincial Natural Science Foundation of China and the Zhejiang Sci-Tech University Fund, and it is published open access, making the full technical details available to researchers worldwide. For now, the image of a table tennis player’s wrist wrapped in a self-healing electronic skin that knows the difference between a loop and a chop stands as a vivid illustration of where the field is heading: materials that mimic biology in structure, repair themselves like living tissue, and lean on artificial intelligence to make sense of what they feel.

Subject of Research: A biomimetic Janus-structured MXene-polyurethane flexible sensor with host-guest self-healing cross-links integrated with deep learning for motion recognition

Article Title: Host-guest interaction viscoelastic polyurethane composite MXene flexible sensor: Integrated deep learning for digital recognition and table tennis return posture monitoring

Article References: Ao, Q., Wang, T., Liu, Y., Jiang, L., & Zhang, X. (2026). Host-guest interaction viscoelastic polyurethane composite MXene flexible sensor: Integrated deep learning for digital recognition and table tennis return posture monitoring. Advanced Composites and Hybrid Materials. https://doi.org/10.1007/s42114-026-02054-5

Image Credits: AI Generated

DOI: 10.1007/s42114-026-02054-5

Keywords: flexible sensor, MXene, polyurethane, host-guest interaction, cyclodextrin, adamantane, self-healing materials, Janus structure, deep learning, wearable electronics, motion capture, table tennis

The stubbornness of these emulsions has a molecular explanation. Naturally occurring polar components of crude oil, most importantly asphaltenes and resins, adsorb at the oil-water interface and build rigid or viscoelastic films around dispersed droplets. Asphaltenes are high-molecular-weight, aromatic, polar molecules containing nitrogen, oxygen and sulfur heteroatoms; their amphiphilic character lets them accumulate at the interface, lower interfacial tension and form elastic layers that resist rupture. Resins, smaller and more soluble, interact with asphaltene aggregates and can either peptize them, improving solubility, or promote smaller, rapidly adsorbing interfacially active complexes that paradoxically stabilize emulsions further. Hydrogen bonding, pi-pi stacking, acid-base interactions and polar functional groups all reinforce these films. Crucially, recent analysis emphasizes that interfacial rheology, not just equilibrium interfacial tension, governs emulsion persistence: asphaltene-rich layers show high dilatational and shear elasticity that slows film drainage and blocks coalescence even when measured interfacial tension looks acceptably low.

A new review published in the journal Results in Engineering by Roland Nagy and Rebeka Bejczi of the petroleum and chemical engineering community brings these threads together in a critical comparison of conventional and next-generation separation additives. The authors argue that while the literature is rich in mechanism-oriented or technology-oriented surveys, far fewer reviews explicitly weigh emerging, sustainability-oriented additive classes, including ionic liquids, beta-cyclodextrin-functionalized polymers, magnetic nanohybrids and bio-based systems, against each other in terms of mechanism, reported performance, dosage, recyclability, environmental limitations and industrial feasibility. Their conclusion is sobering but constructive: no single universal demulsifier or flocculant exists, and the most promising industrial strategy is almost always a tailored combination of chemicals and physical processes rather than one miracle molecule.

The review first draws a sharp functional line between the two main additive families. Flocculants, which include inorganic coagulants such as aluminium sulfate, ferric chloride and polyaluminium chloride alongside synthetic polymers like polyacrylamide and polydiallyldimethylammonium chloride, are chiefly deployed in oil-in-water systems such as produced water and oily wastewater. They work by charge neutralization, polymer bridging, patching and hydrophobic association, aggregating fine droplets and suspended solids so that sedimentation, flotation and filtration become viable. Demulsifiers, by contrast, target water-in-oil crude emulsions. Typically nonionic polyethers, alkoxylated resins and ethylene oxide-propylene oxide block copolymers, they diffuse to the interface, displace or weaken the natural asphaltene-resin film and allow droplets to coalesce into separable volumes, cutting the basic sediment and water content before crude leaves the field. In complex field fluids containing fine solids and mixed emulsion structures, the two approaches can be sequenced: a demulsifier weakens the film, then a flocculant sweeps up the destabilized debris.

Among the emerging materials, beta-cyclodextrin-functionalized flocculants stand out for their elegant molecular design. Cyclodextrin is a toroidal sugar with a hydrophilic exterior and a hydrophobic internal cavity, letting it encapsulate hydrophobic oil components while grafted cationic polymer chains neutralize negatively charged droplets and long graft chains perform adsorption bridging. The review highlights a recent report of a graft-functionalized beta-cyclodextrin flocculant that achieved 95.9 percent oil removal from emulsified oily wastewater at a remarkably low dose of just 2.5 milligrams per liter. However, the authors caution that charge density, hydrophobicity and graft length all interact with salinity and temperature: high ionic strength screens electrostatic interactions, while heat reshapes polymer conformation, so laboratory wins under low-salt model conditions must be validated across realistic salinity and temperature windows before field deployment.

Ionic liquids, salts that remain liquid near room temperature, are the most intensively studied new demulsifier class. Their appeal lies in low volatility, thermal stability and almost unlimited structural tunability: imidazolium, pyridinium, ammonium, phosphonium and Gemini-type cations can be paired with bromide, tetrafluoroborate, hexafluorophosphate, acetate or sulfonate anions, and alkyl chain length can be adjusted to tune oil-phase affinity. Selecting studies report high demulsification efficiency at doses around 1500 parts per million over several hours of treatment. The physics is a delicate balancing act: a sufficiently hydrophobic alkyl chain enhances penetration of asphaltene-rich films, but excessive oil solubility drains the effective interfacial concentration, while excessive water solubility limits access to water-in-oil interfaces. The review also flags the caveats. Industrial adoption remains constrained by cost, potential toxicity, questionable biodegradability and regeneration requirements, and the authors insist that future studies must go beyond bottle tests to include toxicity screening, recyclability assessment, environmental fate analysis and techno-economic evaluation.

Magnetic nanodemulsifiers add a third dimension: recoverability. Magnetite nanoparticles coated with silica, polymers, cyclodextrins, ionic liquids or carbon-based shells can adsorb at the interface, disrupt stabilizing films and then be retrieved with an external magnet for reuse. One flagship example cited is a nanomagnetic cyclodextrin system decorated with an ionic liquid, which delivered 92 percent demulsification efficiency in a water-in-oil crude emulsion, although at a high dose of 5000 parts per million over 24 hours. The trade-offs are familiar: nanoparticle losses during recovery, fouling, regeneration inefficiency, synthesis cost at scale and unresolved questions about ecotoxicity and the environmental fate of released nanoparticles. Bio-based demulsifiers, including biosurfactants, lignin derivatives, tannin materials, polysaccharides and cyclodextrin formulations, complete the sustainability portfolio, but the authors warn that renewable origin alone guarantees nothing; aquatic toxicity, biodegradability under OECD-type protocols, bioaccumulation potential, solvent demand and life-cycle impacts must all be verified quantitatively.

The review also confronts the hardest separation targets: Pickering emulsions, in which solid particles such as clays, silica, corrosion products, mineral fines and asphaltene aggregates adsorb at the interface and physically armor each droplet. These particle-stabilized systems shrug off conventional molecular demulsifiers, so treatment demands particle detachment, wettability modification, or hybrid assistance from ultrasound, electrocoalescence or membranes. Intriguingly, the authors connect this problem to capillary physics in porous media, noting that insights from spontaneous imbibition and nanoparticle-surfactant-brine studies, where interfacial tension and wettability control oil displacement, can inform the design of additives for particle-stabilized emulsions. Chemical treatment is likewise most powerful when hybridized with physical methods: heating lowers viscosity but burns energy, electrostatic coalescence enlarges droplets but demands suitable conductivity, membranes polish fine droplets but foul, and a practical sequence might run from chemical pretreatment through controlled mixing and electrocoalescence to flotation and membrane polishing, with antifouling strategies such as hydrophilic or zwitterionic coatings protecting the final stage.

Perhaps the review’s most valuable contribution is its blunt assessment of the laboratory-to-field gap. Real produced waters span pH values from 4.3 to 10, salinities from 1000 to as much as 400,000 milligrams per liter, oil-and-grease contents of 2 to 565 milligrams per liter, suspended solids up to 1000 milligrams per liter, and temperatures from roughly 3 to 80 degrees Celsius, often laced with residual production chemicals that compete with demulsifiers at the interface. High bottle-test efficiency, the authors argue, is merely an initial screening result, not proof of industrial applicability. They propose a standardized qualification workflow incorporating bottle tests at varying dosage and temperature, standardized water-and-sediment determination, dynamic interfacial tension and interfacial rheology, zeta potential, droplet-size distribution, turbidity and robustness testing across salinity, pH, shear and solids loading, together with design-of-experiments matrices that include asphaltene-to-resin ratio and aging time.

Looking forward, the authors see artificial intelligence and molecular simulation reshaping the field, with machine learning, molecular dynamics and density functional theory screening ionic liquids, polymer architectures and cyclodextrin systems before any synthesis is attempted. Smart stimulus-responsive additives, activated by pH, temperature, magnetic fields or carbon dioxide, promise controlled action, easier recovery and lower chemical consumption, while circular-economy thinking demands that recyclability be demonstrated through mass balance, toxicity data and life-cycle assessment rather than asserted from renewable feedstocks. The bottom line for an industry under pressure to cut both costs and environmental footprint: asphaltene-rich crudes need strong film-disrupting demulsifiers, solids-laden streams need particle detachment and flotation, high-salinity or offshore operations value robustness and speed over peak efficiency, and the winning technology will rarely be a single chemical but an engineered sequence of chemistry and physics matched to the dominant stabilization mechanism in each unique crude oil-water system.

Subject of Research: Advanced flocculants and demulsifiers for separating crude oil-water emulsions

Article Title: Advanced flocculants and demulsifiers for crude oil–water systems: recent developments, mechanisms and future perspectives

Article References: Nagy, R., & Bejczi, R. (2026). Advanced flocculants and demulsifiers for crude oil–water systems: recent developments, mechanisms and future perspectives. Results in Engineering, 32, Article 112816. https://doi.org/10.1016/j.rineng.2026.112816

Image Credits: AI Generated

DOI: 10.1016/j.rineng.2026.112816

Keywords: crude oil, emulsions, demulsifiers, flocculants, asphaltenes, ionic liquids, cyclodextrin, magnetic nanoparticles, Pickering emulsions, produced water, oil-water separation, interfacial rheology