The challenge the team set out to solve is rooted in the oil’s own chemistry. Zanthoxylum oil is traditionally extracted with heat, and the process leaves it loaded with reactive compounds, including the hydroxy-sanshool molecules that produce its characteristic tingling sensation. The oil is also rich in unsaturated fatty acids that readily react with oxygen, driving up peroxide values and degrading flavor during storage. On top of that, its hydrophobic nature and volatile aroma compounds make it prone to instability and flavor loss, limiting its use in industrial condiment applications.

Emulsification has long been a favored strategy for protecting sensitive lipids. In a Pickering emulsion, solid colloidal particles rather than conventional molecular surfactants anchor themselves at the oil-water interface, forming a physical armor around each droplet. Protein-polysaccharide complexes are particularly attractive stabilizers because their amphiphilicity can be tuned, they are biocompatible, and they can build robust interfacial networks. Soy protein isolate brings strong emulsifying and film-forming ability, while sodium alginate, a negatively charged polysaccharide, can bind to proteins through electrostatic attraction and hydrogen bonding under the right pH conditions.

But Zanthoxylum oil is not an ordinary oil phase, and the researchers were careful to point out why. Its high content of surface-active sanshools competes with the protein-polysaccharide particles for space at the oil-water interface, disrupting the dense, ordered particle film that gives Pickering emulsions their strength. Minor constituents such as unsaponifiable matter and small polar flavor molecules can further destabilize the system by altering the polarity of the aqueous phase and disturbing particle dispersion. Stabilizing this oil, in other words, demanded targeted solutions rather than a simple swap of the oil phase.

To build their stabilizer, the team hydrated soy protein isolate overnight, dissolved sodium alginate separately, and combined the two solutions in a one-to-one volume ratio with stirring at 40 degrees Celsius, then adjusted the mixture to neutral pH. Characterization revealed how the two biopolymers cooperate. Water contact angle measurements showed that soy protein alone, at roughly 102 degrees, is somewhat too hydrophobic, while alginate, at about 54 degrees, is far too hydrophilic to anchor at an interface on its own. The composite landed at 67.92 degrees, close to the ideal wettability range for Pickering stabilizers. Infrared spectroscopy and fluorescence measurements confirmed hydrogen bonding and electrostatic attraction between the partners, with shifts in hydroxyl, amide, and carboxylate bands and a red shift in the protein’s tryptophan fluorescence indicating partial unfolding and a more polar microenvironment.

When the composite particles were used to emulsify Zanthoxylum oil at a 50 percent oil fraction, the result was a milky, oil-in-water emulsion with small, round, uniformly sized droplets averaging 5.82 micrometers and a strongly negative zeta potential of minus 43.77 millivolts, a sign of robust electrostatic repulsion between droplets. Confocal laser scanning microscopy showed oil droplets encircled by a red fluorescent protein shell, with additional particles dispersed in the continuous phase, hinting at a three-dimensional network beyond the interface. Rheological tests backed this up: the emulsion shear-thinned like a fluid but its storage modulus consistently exceeded its loss modulus, the hallmark of a gel-like, solid-supported structure.

The emulsion also proved remarkably tough under stress. It survived centrifugation at 8000 rpm with only a trivial increase in droplet size and no visible separation, and it remained intact through two months of storage at 25 degrees Celsius, although droplets gradually coalesced after the first month. It tolerated salt concentrations up to 500 millimoles per liter without phase separation, with destabilization driven by compression of the electrical double layer rather than charge neutralization. Its sweet spot was pH 6 to 8; acidic conditions near the protein’s isoelectric point or alginate’s pKa broke the system down, and a single freeze-thaw cycle caused oil-water separation, a limitation the authors flag for cold-chain applications.

The real test came in accelerated oxidation, with samples held at 60 degrees Celsius for 18 days in the classic Schaal oven protocol. By day 18, the peroxide value of the bulk oil had climbed to 27.89 millimoles per kilogram, nearly double that of the emulsified oil, and the emulsion consistently showed lower oxidation rates throughout. The protective effect extended to flavor. Hydroxy-alpha-sanshool, which makes up more than 90 percent of the numbing compounds, degraded most rapidly over time, but by day 18 the emulsion retained 52 percent of total sanshools versus 46 percent in the bulk oil. Headspace gas chromatography-mass spectrometry identified 28 volatile compounds, and four key aroma molecules told the most striking story: d-limonene retention in the emulsion was nearly 50 percent, far above the roughly 30 percent seen in neat oil, and myrcene and linalool were similarly better preserved. Linalyl acetate was an exception, apparently hydrolyzing at its ester bond in the aqueous phase, a problem the authors suggest could be addressed by tuning interfacial pH or increasing oil phase viscosity.

The team proposes a two-part protection mechanism. At the interface, the adsorbed SPI/SA particles form a dense physical barrier that impedes oxygen permeation and suppresses volatile release. In the bulk, unadsorbed particles knit a gel-like network that slows molecular diffusion in both phases, hindering the mass transfer that fuels oxidation, while hydrophobic partitioning keeps aroma compounds locked inside the oil droplets. The authors are candid about caveats: the initial peroxide value of the emulsion was elevated by the shear and heat of homogenization, headspace partitioning differences between oil and emulsion matrices may inflate apparent retention figures, and the system has yet to be validated in real food products or sensory panels. Future work will target inert gas flushing during preparation, natural antioxidant co-addition, and industrial scale-up.

Even with those limitations, the study marks a meaningful step toward greener preservation of spice oils. Both soy protein isolate and sodium alginate are inexpensive, abundant, and accepted food ingredients, giving the approach a clear advantage in cost and scalability over more exotic encapsulation materials such as chitin nanowhiskers. If the freeze-thaw weakness can be engineered away and the chemistry holds up in real condiment formulations, the humble pairing of soybean protein and seaweed-derived alginate could become the standard shield for the numbing, aromatic soul of Sichuan cuisine.

Subject of Research: Pickering emulsion stabilization of Zanthoxylum oil using soy protein isolate and sodium alginate composite particles to retard oxidation and preserve flavor

Article Title: SPI/SA composite particles as Pickering stabilizers to retard oxidation and preserve flavor of Zanthoxylum oil

Article References: Huang, J., Wang, X., Wu, C., Zhu, X., Hou, X., Shen, G., Cui, Q., Yu, J., Zhang, J., Hu, W., Chen, S., & Zhang, Z. (2026). SPI/SA composite particles as Pickering stabilizers to retard oxidation and preserve flavor of Zanthoxylum oil. Food Chemistry: X, Article 104608. https://doi.org/10.1016/j.fochx.2026.104608

Image Credits: AI Generated

DOI: 10.1016/j.fochx.2026.104608

Keywords: Zanthoxylum oil, Pickering emulsion, soy protein isolate, sodium alginate, lipid oxidation, sanshools, flavor preservation, food-grade stabilizers, encapsulation, volatile aroma compounds, Sichuan pepper, food chemistry

The research team, led by Min Kyung Park and Han Sub Kwak, assembled seven commercial milk samples: four produced in Korea and three imported sterilized milks from Germany, Poland, and Australia. The domestic samples included two refrigerated ultra-high-temperature pasteurized milks with a government-regulated fourteen-day shelf life and two room-temperature sterilized milks with a six-week shelf life. The imported products, all UHT-treated with one-year shelf lives, were purchased through online marketplaces, reflecting how most Korean consumers now encounter them. Every sample was stored at four degrees Celsius and served within its expiration date, ensuring that the comparison captured genuine product differences rather than handling artifacts.

To dissect the chemistry, the team turned to headspace solid-phase microextraction coupled with gas chromatography and mass spectrometry, a technique that essentially sniffs milk at the molecular level. Three milliliters of each sample were sealed in amber vials with an internal standard, incubated at forty degrees Celsius, and probed with a divinylbenzene/carboxen/polydimethylsiloxane fiber that absorbs volatile compounds from the headspace. Separation occurred on a DB-WAXUI capillary column with helium as the carrier gas, and compounds were identified by matching retention times and mass spectra against authentic standards and the Wiley library. In total, forty-two volatile aroma compounds were quantified across the seven milks, providing a chemical fingerprint for each product.

The chemical results were striking. Imported milk contained significantly higher total levels of volatile aroma compounds than domestic milk, with the Australian and Polish samples standing out in particular. Ketones and benzene derivatives dominated the profiles, and acetone alone accounted for roughly half to seventy percent of the total volatiles, a dominance consistent with earlier international studies. In the imported milks, fatty acid-derived methyl ketones such as propan-2-one, butan-2-one, pentan-2-one, heptan-2-one, and nonan-2-one were predominant; these compounds, formed through lipid oxidation, are known key contributors to cheese and butter notes. Domestic milks, by contrast, leaned more heavily on benzene derivatives and carried a generally quieter aromatic profile.

Multivariate statistics sharpened the picture. Principal component analysis explained sixty percent of the total variance and cleanly separated domestic from imported samples along the first component, while partial least squares discriminant analysis identified nine volatile compounds as the molecular signatures driving the divide. Among them was methylsulfonylmethane, a sulfur-containing compound with a vegetable-like odor. Sulfur volatiles matter enormously in flavor science because their sensory thresholds are extremely low, meaning even trace amounts register strongly in the nose. The imported group was thus characterized by distinctive, pungent notes of cheese, fruit, and sulfur, while the domestic group showed a milder profile associated with sweet and aromatic compounds such as 1,2-xylene and phenylmethanol.

Why would milk from different countries smell so different? The researchers point to the entire chain from pasture to processing. Cows grazing on fresh grass produce milk richer in terpenes and carotenoids, lending grassy and creamy notes, whereas animals fed grain or total mixed rations yield milk with different fatty acid patterns and less distinct aromas. Pasture biodiversity, feed quality, and rearing climate all leave chemical traces in the glass. Processing adds another layer: intense UHT heating can generate caramel-like or cooked flavors through Maillard reactions. Korea’s limited land area and high population density mean dairy cows are typically raised in barns on formulated fodder, producing milk with a deliberately mild, less complex aroma, while pasture-based systems common in exporting countries produce richer, more intense flavor profiles.

Chemistry alone, however, cannot predict what people will like, so the team recruited 120 consumers from Jeonju-si and Wanju-gun under an approved institutional review board protocol, with 114 completing both sessions. On day one, participants tasted all seven milks blind, coded with random three-digit numbers and served slightly chilled at eight degrees, rating overall liking plus color, flavor, aroma, and mouthfeel on nine-point scales. They also profiled each sample using the rate-all-that-apply method across twenty-four attributes and recorded emotional responses with the EsSense25 questionnaire. On day two, the identical procedure was repeated, except this time the actual one-liter packages and prices were displayed, allowing the researchers to measure exactly how brand, country of origin, and cost information reshape perception.

The consumer results told a clear story. In the blind test, domestic refrigerated milks earned the highest liking scores, followed by domestic sterilized milks, with all three imported samples rated significantly lower. When packages were revealed, liking for domestic milk generally rose, and one familiar brand showed a statistically significant jump from 6.87 to 7.25, while ratings for imported milk slipped further. Familiarity followed the same pattern, increasing significantly for domestic products and dropping for imports once their origin was visible. Sensory profiling confirmed that blind tasters already described imported milk as yellowish, cheesy, buttery, fermented, and gamey, and these impressions intensified when shoppers knew what they were drinking. Partial least squares regression tied the negative attributes directly to the methyl ketones and sulfur compounds identified chemically, showing that the same molecules driving the imported aroma profile were dragging down overall liking.

Perhaps the most surprising finding concerned price. Polish milk in the study was nearly forty-six percent cheaper than its Korean counterpart, and imported products averaged roughly twenty-one percent lower prices overall, yet this economic advantage barely moved the needle. Only a handful of participants mentioned cost in open-ended feedback, and acceptance ratings for imports barely changed between blind and informed conditions. The researchers conclude that intrinsic factors, flavor and aroma above all, dominate milk choice, likely because milk remains a small, frequent purchase rather than a major household expense. For Korean consumers, decades of exposure have cultivated a taste for mild domestic milk, and the unfamiliar intensity of imported dairy aromas, sometimes described locally as a gamey off-note, outweighs any discount. As imported sterilized milk continues to gain shelf space in an increasingly expensive domestic market, the study suggests that winning over Korean palates will require more than competitive pricing; it may require either reformulating expectations or meeting consumers in flavored matrices, such as lattes, where those bold dairy notes can hide.

Subject of Research: Volatile aroma compound profiles and consumer perception of domestic versus imported milk in Korea

Article Title: Comparison of volatile aroma compounds and consumer perception between domestic and imported milk in Korea

Article References: Park, M. K., Park, S., Yoon, G., Lee, C.-L., & Kwak, H. S. (2026). Comparison of volatile aroma compounds and consumer perception between domestic and imported milk in Korea. Food Science of Animal Resources, 46(1), Article 52. https://doi.org/10.1007/s44463-025-00018-9

Image Credits: AI Generated

DOI: 10.1007/s44463-025-00018-9

Keywords: milk, volatile aroma compounds, GC-MS, consumer perception, country of origin, UHT milk, sensory evaluation, South Korea, food science, flavor chemistry, consumer ethnocentrism, dairy